# NONFERROUS NANOMATERIALS & COMPOSITES FOR ENERGY STORAGE AND CONVERSION

EDITED BY : Jiexi Wang, Qiaobao Zhang and Kaili Zhang PUBLISHED IN : Frontiers in Chemistry

#### Frontiers Copyright Statement

© Copyright 2007-2019 Frontiers Media SA. All rights reserved. All content included on this site, such as text, graphics, logos, button icons, images, video/audio clips, downloads, data compilations and software, is the property of or is licensed to Frontiers Media SA ("Frontiers") or its licensees and/or subcontractors. The copyright in the text of individual articles is the property of their respective authors, subject to a license granted to Frontiers.

The compilation of articles constituting this e-book, wherever published, as well as the compilation of all other content on this site, is the exclusive property of Frontiers. For the conditions for downloading and copying of e-books from Frontiers' website, please see the Terms for Website Use. If purchasing Frontiers e-books from other websites or sources, the conditions of the website concerned apply.

Images and graphics not forming part of user-contributed materials may not be downloaded or copied without permission.

Individual articles may be downloaded and reproduced in accordance with the principles of the CC-BY licence subject to any copyright or other notices. They may not be re-sold as an e-book.

As author or other contributor you grant a CC-BY licence to others to reproduce your articles, including any graphics and third-party materials supplied by you, in accordance with the Conditions for Website Use and subject to any copyright notices which you include in connection with your articles and materials.

All copyright, and all rights therein, are protected by national and international copyright laws.

The above represents a summary only. For the full conditions see the Conditions for Authors and the Conditions for Website Use. ISSN 1664-8714 ISBN 978-2-88945-977-3 DOI 10.3389/978-2-88945-977-3

### About Frontiers

Frontiers is more than just an open-access publisher of scholarly articles: it is a pioneering approach to the world of academia, radically improving the way scholarly research is managed. The grand vision of Frontiers is a world where all people have an equal opportunity to seek, share and generate knowledge. Frontiers provides immediate and permanent online open access to all its publications, but this alone is not enough to realize our grand goals.

### Frontiers Journal Series

The Frontiers Journal Series is a multi-tier and interdisciplinary set of open-access, online journals, promising a paradigm shift from the current review, selection and dissemination processes in academic publishing. All Frontiers journals are driven by researchers for researchers; therefore, they constitute a service to the scholarly community. At the same time, the Frontiers Journal Series operates on a revolutionary invention, the tiered publishing system, initially addressing specific communities of scholars, and gradually climbing up to broader public understanding, thus serving the interests of the lay society, too.

### Dedication to Quality

Each Frontiers article is a landmark of the highest quality, thanks to genuinely collaborative interactions between authors and review editors, who include some of the world's best academicians. Research must be certified by peers before entering a stream of knowledge that may eventually reach the public - and shape society; therefore, Frontiers only applies the most rigorous and unbiased reviews.

Frontiers revolutionizes research publishing by freely delivering the most outstanding research, evaluated with no bias from both the academic and social point of view. By applying the most advanced information technologies, Frontiers is catapulting scholarly publishing into a new generation.

### What are Frontiers Research Topics?

Frontiers Research Topics are very popular trademarks of the Frontiers Journals Series: they are collections of at least ten articles, all centered on a particular subject. With their unique mix of varied contributions from Original Research to Review Articles, Frontiers Research Topics unify the most influential researchers, the latest key findings and historical advances in a hot research area! Find out more on how to host your own Frontiers Research Topic or contribute to one as an author by contacting the Frontiers Editorial Office: researchtopics@frontiersin.org

# NONFERROUS NANOMATERIALS & COMPOSITES FOR ENERGY STORAGE AND CONVERSION

Topic Editors: Jiexi Wang, Central South University, China Qiaobao Zhang, Xiamen University, China Kaili Zhang, City University of Hong Kong, Hong Kong

Citation: Wang, J., Zhang, Q., Zhang, K., eds. (2019). Nonferrous Nanomaterials & Composites for Energy Storage and Conversion. Lausanne: Frontiers Media. doi: 10.3389/978-2-88945-977-3

# Table of Contents

*07 Mesoporous NH4NiPO4·H2 O for High-Performance Flexible All-Solid-State Asymmetric Supercapacitors*

Yong Liu, Xiaoliang Zhai, Keke Yang, Fei Wang, Huijie Wei, Wanhong Zhang, Fengzhang Ren and Huan Pang


Ying Wang, Shengxiang Wu, Chao Wang, Yijing Wang and Xiaopeng Han


Chunli Guo, Minshuai Yin, Chun Wu, Jie Li, Changhui Sun, Chuankun Jia, Taotao Li, Lifeng Hou and Yinghui Wei

*64* In-situ *Grown SnS2 Nanosheets on rGO as an Advanced Anode Material for Lithium and Sodium Ion Batteries*

Hezhang Chen, Bao Zhang, Jiafeng Zhang, Wanjing Yu, Junchao Zheng, Zhiying Ding, Hui Li, Lei Ming, D. A. Mifounde Bengono, Shunan Chen and Hui Tong

*73 Interfaces Between Cathode and Electrolyte in Solid State Lithium Batteries: Challenges and Perspectives*

Kaihui Nie, Yanshuai Hong, Jiliang Qiu, Qinghao Li, Xiqian Yu, Hong Li and Liquan Chen

*92 Controllable Synthesis of Na3V2(PO4) <sup>3</sup>/C Nanofibers as Cathode Material for Sodium-Ion Batteries by Electrostatic Spinning*

Ling Wu, Yueying Hao, Shaonan Shi, Xiaoping Zhang, Huacheng Li, Yulei Sui, Liu Yang and Shengkui Zhong

*101 Fabrication of Cobalt-Nickel-Zinc Ternary Oxide Nanosheet and Applications for Supercapacitor Electrode*

Chun Wu, Lei Chen, Xuechun Lou, Mei Ding and Chuankun Jia

*111 A Porous and Conductive Graphite Nanonetwork Forming on the Surface of KCu7S4 for Energy Storage*

Wei-Xia Shen, Jun-Min Xu, Shu-Ge Dai and Zhuang-Fei Zhang

	- Liubin Song, Fuli Tang, Zhongliang Xiao, Zhong Cao and Huali Zhu

Hongwei Mi, Xiaodan Yang, Jun Hu, Qianling Zhang and Jianhong Liu

*191 Nitrogen-Doped Multi-Scale Porous Carbon for High Voltage Aqueous Supercapacitors*

Xichuan Liu, Rui Mi, Lei Yuan, Fan Yang, Zhibing Fu, Chaoyang Wang and Yongjian Tang


Zi-Min Jiang, Ting-Ting Xu, Cong-Cong Yan, Cai-Yun Ma and Shu-Ge Dai


Jinhuan Yao, Jing Yan, Yu Huang, Yanwei Li, Shunhua Xiao and Jianrong Xiao

*247 Boosting Lithium-Ion Storage Capability in CuO Nanosheets via Synergistic Engineering of Defects and Pores*

Zhao Deng, Zhiyuan Ma, Yanhui Li, Yu Li, Lihua Chen, Xiaoyu Yang, Hong-En Wang and Bao-Lian Su

	- Ziqi Wang, Jiaojiao Liang, Kai Fan, Xiaodi Liu, Caiyun Wang and Jianmin Ma

Yong Liu, Huijie Wei, Chao Wang, Fei Wang, Haichao Wang, Wanhong Zhang, Xianfu Wang, Chenglin Yan, Bok H. Kim and Fengzhang Ren

*295 Cryptomelane-Type KMn8O16 as Potential Cathode Material — for Aqueous Zinc Ion Battery*

Jiajie Cui, Xianwen Wu, Sinian Yang, Chuanchang Li, Fang Tang, Jian Chen, Ying Chen, Yanhong Xiang, Xianming Wu and Zeqiang He


Sicen Yu, Yi Wan, Chaoqun Shang, Zhenyu Wang, Liangjun Zhou, Jianli Zou, Hua Cheng and Zhouguang Lu

*342 High-Performance Lithium-Sulfur Batteries With an IPA/AC Modified Separator*

Yafang Guo, Aihua Jiang, Zengren Tao, Zhiyun Yang, Yaping Zeng and Jianrong Xiao

*351 Synthesis and Electrochemical Performance of Molybdenum Disulfide-Reduced Graphene Oxide-Polyaniline Ternary Composites for Supercapacitors* Li-Zhong Bai, Yan-Hui Wang, Shuai-Shuai Cheng, Fang Li, Zhi-Yi Zhang and Ya-Qing Liu *358 Capacity Increase Investigation of Cu2 Se Electrode by Using Electrochemical Impedance Spectroscopy* Xiuwan Li, Zhixin Zhang, Chaoqun Liu and Zhiyang Lin *365 Electrospun Single Crystalline Fork-Like K2V8O21 as High-Performance Cathode Materials for Lithium-Ion Batteries* Pengfei Hao, Ting Zhu, Qiong Su, Jiande Lin, Rong Cui, Xinxin Cao, Yaping Wang and Anqiang Pan *374 Cross-Linked Nanohybrid Polymer Electrolytes With POSS Cross-Linker for Solid-State Lithium Ion Batteries* Jinfang Zhang, Xiaofeng Li, Ying Li, Huiqi Wang, Cheng Ma, Yanzhong Wang, Shengliang Hu and Weifeng Wei *384 Reduced Graphene Oxide Decorated Na3V2(PO4 ) <sup>3</sup> Microspheres as Cathode Material With Advanced Sodium Storage Performance* Hezhang Chen, Yingde Huang, Gaoqiang Mao, Hui Tong, Wanjing Yu, Junchao Zheng and Zhiying Ding *392 Comparative Investigation of 0.5Li2MnO3·0.5LiNi0.5Co0.2Mn0.3O2 Cathode Materials Synthesized by Using Different Lithium Sources* Peng-Bo Wang, Ming-Zeng Luo, Jun-Chao Zheng, Zhen-Jiang He, Hui Tong and Wan-jing Yu *401 Porous Hollow Superlattice NiMn2O4/NiCo2O4 Mesocrystals as a Highly Reversible Anode Material for Lithium-Ion Batteries* Lingjun Li, Qi Yao, Jiequn Liu, Kaibo Ye, Boyu Liu, Zengsheng Liu, Huiping Yang, Zhaoyong Chen, Junfei Duan and Bao Zhang *412 SiC Nanofibers as Long-Life Lithium-Ion Battery Anode Materials* Xuejiao Sun, Changzhen Shao, Feng Zhang, Yi Li, Qi-Hui Wu and Yonggang Yang *419 N/S Co-doped Carbon Derived From Cotton as High Performance Anode Materials for Lithium Ion Batteries* Jiawen Xiong, Qichang Pan, Fenghua Zheng, Xunhui Xiong, Chenghao Yang, Dongli Hu and Chunlai Huang *428 High-Level Heteroatom Doped Two-Dimensional Carbon Architectures for Highly Efficient Lithium-Ion Storage* Zhijie Wang, Yanyan Wang, Wenhui Wang, Xiaoliang Yu, Wei Lv, Bin Xiang and Yan-Bing He *438 Effect of Nb and F Co-doping on Li1.2Mn0.54Ni0.13Co0.13O2 Cathode Material for High-Performance Lithium-Ion Batteries* Lei Ming, Bao Zhang, Yang Cao, Jia-Feng Zhang, Chun-Hui Wang, Xiao-Wei Wang and Hui Li *450 High-Power-Density, High-Energy-Density Fluorinated Graphene for Primary Lithium Batteries*

Guiming Zhong, Huixin Chen, Xingkang Huang, Hongjun Yue and Canzhong Lu

# Mesoporous NH4NiPO4·H2O for High-Performance Flexible All-Solid-State Asymmetric Supercapacitors

Yong Liu1,2 \* † , Xiaoliang Zhai 1†, Keke Yang1†, Fei Wang<sup>1</sup> , Huijie Wei <sup>1</sup> , Wanhong Zhang<sup>1</sup> \*, Fengzhang Ren<sup>1</sup> and Huan Pang<sup>3</sup>

#### Edited by:

*Qiaobao Zhang, Xiamen University, China*

#### Reviewed by:

*Zhen-Dong Huang, Nanjing University of Posts and Telecommunications, China Yuxin Zhang, Chongqing University, China Hongkang Wang, Xi'an Jiaotong University, China*

#### \*Correspondence:

*Yong Liu liuyong209@haust.edu.cn Wanhong Zhang zhangwh@haust.edu.cn*

*†These authors have contributed equally to this work*

#### Specialty section:

*This article was submitted to Physical Chemistry and Chemical Physics, a section of the journal Frontiers in Chemistry*

> Received: *31 August 2018* Accepted: *15 February 2019* Published: *07 March 2019*

#### Citation:

*Liu Y, Zhai X, Yang K, Wang F, Wei H, Zhang W, Ren F and Pang H (2019) Mesoporous NH*4*NiPO*4·*H*2*O for High-Performance Flexible All-Solid-State Asymmetric Supercapacitors. Front. Chem. 7:118. doi: 10.3389/fchem.2019.00118* *<sup>1</sup> Collaborative Innovation Center of Nonferrous Metals of Henan Province, Henan Key Laboratory of High-Temperature Structural and Functional Materials, School of Materials Science and Engineering, Henan University of Science and Technology, Luoyang, China, <sup>2</sup> Henan Key Laboratory of Non-Ferrous Materials Science & Processing Technology, Henan University of Science and Technology, Luoyang, China, <sup>3</sup> School of Chemistry and Chemical Engineering, Yangzhou University, Yangzhou, China*

Nowadays, wearable energy storage devices have been growing rapidly, but flexible systems with both excellent cycling stability and decent flexibility are still challenging. In this work, a flexible all-solid-state NH4NiPO4·H2O//graphene supercapacitor with remarkable performance was successfully assembled. When cycled at a current density of 5 mA cm−<sup>2</sup> , the device delivered 121 mF cm−<sup>2</sup> , and showed good cycling stability after 3,000 cycles. Moreover, the all-solid-state NH4NiPO4·H2O//graphene supercapacitor also exhibit high mechanical flexibility with well-maintained specific capacitance, even under bending to arbitrary angles (up to 180◦ ) and different weights (up to 50 g).

Keywords: NH4NiPO4 ·H2O, flexible supercapacitor, asymmetrical, all-solid-state, electrochemical performances

### INTRODUCTION

Recently, for the reason of environmental and energy issues, research on energy storage have become one of hot spots all over the world (Wang et al., 2017, 2019; Zheng et al., 2017; Huang et al., 2018a,b,c; Liu et al., 2018; Wang F. et al., 2018; Wang H. et al., 2018; Wang H. K. et al., 2018; Zhang et al., 2018a; Zhao et al., 2018). Among the energy storage systems, supercapacitors have sparked increasing attention due to their high power density and long cycle life (Conway, 1999; Aricò et al., 2005; Miller and Simon, 2008; Yuan et al., 2012; Bin Jiang et al., 2018; Zhu et al., 2018). One type of supercapacitors, pseudocapacitors, which include fast Faradic reactions on the electrodes, could deliver greater specific capacitance than electrochemical double-layer capacitors, which could make the device have higher energy density (Dai et al., 2018; Gao et al., 2018; Zhang et al., 2018b; Zhao et al., 2018a; Zheng et al., 2018). For the electrode materials of pseudocapacitors, electrochemical active materials are often used, such as transition-metal oxides [e.g., RuO<sup>2</sup> (Zhai et al., 2018), NiO (Zuo et al., 2016), MnO<sup>2</sup> (Yang et al., 2016), Co3O<sup>4</sup> (Zhang et al., 2016a)], and conducting polymers (Xie and Wang, 2016). Nevertheless, the high price of Ruthenium makes it hardly be utilized as electrode materials for pseudocapacitors. In this case, it is crucial to synthesize an electrode material with low cost and high performance.

**7**

On the other hand, portable devices generally require small size, light weight, which the traditional capacitors could not achieve, and all of these limit the development of this area (Huang et al., 2018). As a newly developed energy-storage device, the flexible all-solid-state supercapacitors are small and light compared to the conventional capacitors (Lv et al., 2018). And flexible all-solid-state supercapacitor could deliver much higher energy density than conventional capacitors (Gao et al., 2014a; Wei et al., 2015; Yousaf et al., 2016; Zhang et al., 2016b; Wu et al., 2018). Furthermore, with two electrodes made of different materials, these asymmetric supercapacitors could show better performance in energy density. Together with their high power density, flexible asymmetric all-solid-state supercapacitor are promising for the wearable energy storage systems (Zhang et al., 2016b; Wu et al., 2018).

In the past few years, ammonium/transition metal phosphate NH4MPO4·H2O (M = Co2+, Ni2+) have been studied as electrodes in the field of supercapacitors (Pang et al., 2012; Zhao et al., 2013; Wang et al., 2014a). For instance, Wang and his colleagues utilized a facile hydrothermal method to synthesize layered NH4CoPO4·H2O microbundles which consist of 1D layered microrods (Wang et al., 2014a). The layered microbundle electrode showed good high-rate capability as well as excellent cycling stability. In the previous work, we have successful fabricated mesoporous NH4NiPO4·H2O nanostructures using one-pot hydrothermal method (Zhao et al., 2013). In this work, we assembled them into flexible all-solid-state asymmetric supercapacitors and studied their electrochemical performances. The specific capacitance of the device can reach 121 mF cm−<sup>2</sup> , and shows good long-term cycling stability. And this device exhibit excellent mechanical flexibility under bending to arbitrary angles (up to 180◦ ) and different weights (even 50 g).

### MATERIALS AND METHODS

### Synthesis of Mesoporous NH4NiPO4·H2O Nanostructures

NH4NiPO4·H2O nanostructures were synthesized by reacting 0.40 g Ni(NO3)<sup>2</sup> and 0.40 g (NH4)3PO<sup>4</sup> at 200 ◦ C for 45 h under hydrothermal condition in 20.0 mL ethylene glycol, and the autoclave was then cooled to room temperature as described elsewhere (Zhao et al., 2013). The green and yellow precipitates were obtained and filtered. After being washed with distilled water and ethanol repeatly, the final product was obtained after being dried in air for 24 h.

### Characterizations

The crystal structures of the samples were analyzed by X-ray diffraction (XRD) (Rigaku-Ultima III with Cu Kα radiation, λ = 1.5418 Å). The microstructures of as-prepared samples were revealed using a field-emission scanning electron microscope (FESEM; JEOL JSM-6701F, 5.0 kV), transmission electron microscopy (TEM) and high-resolution transmission electron microscopy (HRTEM) (JEM-2100, 200 kV). Nitrogen adsorption–desorption isotherms were measured on a Gemini VII 2390 Analyzer at 77 K, and the specific surface area was calculated by the Brunauer-Emmett-Teller (BET) method.

### Fabrication of Flexible All-Solid-State NH4NiPO4·H2O//Graphene Supercapacitors

The PET substrates were first deposited with a layer of Pt film (about 3–5 nm thick) and then coated with the slurry containing the active materials (NH4NiPO4·H2O or graphene via a procedure similar to that in the three-electrode system and were used as the working electrode after drying). Meanwhile, 1.52 g PVA was mixed into 10 ml deionized water to form a mixture. After the mixture is clarified at the constant temperature of 75◦C,the prepared 5 ml 3 mol/L KOH are slowly dripped into the mixture with continuous stirring. Then the gel-like electrolyte was obtained. Then, two pieces of such electrodes were immersed in the gel solution for 5–10 min to coat a layer of gel electrolyte. After the excess water was vaporized, two pieces of such electrodes containing electrolyte were pressed together with sandwiched structures. Finally, the stacked all-solid-state NH4NiPO4·H2O//graphene asymmetric supercapacitors were fabricated.

### Electrochemical Measurements

Electrochemical study on all-solid-state NH4NiPO4·H2O//graphene asymmetric supercapacitor was carried out using an electrochemical working-station (CHI 660D, Shanghai Chenhua). The electrochemical performance measurements were conducted in a conventional two-electrode system with graphene electrode as counter and reference electrode. Cyclic voltammetry (CV) and galvanostatic chargedischarge methods were used to investigate capacitive properties of all-solid-state NH4NiPO4·H2O//graphene asymmetric supercapacitor with a potential window between 0 and 1.4 V. And electrochemical impedance spectroscopy (ESI) measurements were carried out by using PARSTAT2273 at

0.4 V over the frequency range of 100 kHz to 10 mHz with an amplitude of 5 mV.

### RESULTS AND DISCUSSIONS

As shown in **Figure 1b**, all peaks correspond well to those of NH4NiPO4·H2O (JCPD No. 50-0425, **Figure 1a**), indicating the good crystallinity of the samples. This is consistent with the observations reported elsewhere (Zhao et al., 2013). **Figure S1b** shows the typical N<sup>2</sup> adsorption-desorption isotherms of mesoporous NH4NiPO4·H2O, and from the calculation, the specific surface area of the sample was ∼418 m<sup>2</sup> g −1 and poresizes were within the range about 2.0–18.5 nm (**Figure S1a**). The presence of mesoporous provides channels for the ion

transport, and high specific surface area could facilitate the contact of electrolyte and electrode, which are beneficial to the electrochemical properties of electrode.

The morphologies of the as-prepared NH4NiPO4·H2O samples were examined by FESEM and TEM. As shown in **Figures 2a,b**, the samples are in uniform nano-almond structures, and even when the scale bar is 1.5µm (**Figure 2a**), showing the high uniformity of nanostructures. The sizes of single nano-almond are in the range of 300∼350 nm. Furthermore, the uniform shape and size were further proved by TEM, which are shown in **Figures 2c,d**. **Figures 2e,f** show the HRTEM and selected area electron diffraction (SAED) pattern of as-prepared NH4NiPO4·H2O samples. The d-spacing of lattice fringes in **Figure 2e** is ∼0.278 nm, which is corresponding to the (121) lattice spacing of NH4NiPO4·H2O. The SAED patterns in **Figure 2f** confirm the polycrystalline nature of the samples, which show NH4NiPO4·H2O phase. As shown in **Figure 2e**, the measured diameters of pores are ∼2.0 nm, and the porous structure may facilitate electrolyte access, resulting in fast ion intercalation and extraction.

In this work, flexible all-solid-state hybrid supercapacitors were assembled using as-prepared NH4NiPO4·H2O and graphene as positive and negative electrode, respectively. The CV and galvanostatic charge and discharge tests were carried out to test the electrochemical properties of the samples. As the **Figure 3A** shows, the charging voltage of the device is 0 to 1.4V. When the scan rates range from 5 to 50 mV s−<sup>1</sup> . The curves show a quasi-rectangular geometry, which shows that the sample not only has the characteristic of pseudo capacitance, but also has the characteristics of electric double layer capacitance at these rates (Gao et al., 2014b). Furthermore, when rate is as high as 50 mV s −1 , the shape of CV curve could still preserved, indicating that the hybrid supercapacitor has very good rate capability (Dai et al., 2018). When the hybrid supercapacitor was charged and discharged in the current density of 0.2, 0.5, 1.2, 2.0, 3.0, 5.0 mA cm−<sup>2</sup> , as the **Figure 3B** shows, these curves are approximately in triangular shape, which means the supercapacitor have excellent reversibility and capacitance at each current density. And the capacitances is calculated from galvanostatic charge-discharge curves by the following Formula:

$$C\_{\text{spec}} = (I \times \Delta t) / (\Delta V \times S) \tag{1}$$

Where I is the current density, t is the discharge time, V is the potential range (V = 1.4 V) and S is the area of the supercapacitors (Roldán et al., 2015). After calculation, we plotted the specific capacitance of the supercapacitor. As shown in **Figure 3C**. When the current density is 0.5 mA cm−<sup>2</sup> , its areal specific capacitances could achieve 180 mF cm−<sup>2</sup> . Remarkably, even at as high as 5 mA cm−<sup>2</sup> , this value can still reach 121 mF cm−<sup>2</sup> . The capacity retention rate is about 88.8%, after 3,000 cycles with the current density of 5 mA cm−<sup>2</sup> (**Figure 3D**). This capacitance decay may be attributed to some irreversible reactions between the electrodes and electrolyte (Wang et al., 2014b). Noticeably, even after 3,000 cycles, the

nanostructured morphology of the electrode material was wellsustained (**Figure S2**). A comparison of the electrochemical performance of the supercapacitors with other hybrid solid state devices are shown in **Table S1**.

To evaluate the potential of the all-solid-state hybrid supercapacitor for flexible energy storage under real conditions, the CV curves of the device at 5 mV s−<sup>1</sup> were collected under normal and bent conditions. As shown in **Figure 4a**, when the hybrid supercapacitor was bent to 30◦ , 90◦ , 180◦ , the curves change slightly, suggesting the good capacitance stability of this flexible supercapacitor (Qin et al., 2018; Wang W. et al., 2018). **Figures 4b–e** show the all-solid-state hybrid supercapacitor under different weights and corresponding CV curves with 0–1.4 V range at a scan rate of 10 mV s−<sup>1</sup> . Similar to the device with different bending angles, the CV curves of the device under different weight (5, 20, and 50 g) change slightly, and the corresponding specific capacitance of the device is well-maintained. All the above results show that this hybrid supercapacitor has excellent mechanical flexibility (Qin et al., 2018; Wang W. et al., 2018).

We test the electrochemical impedance spectra (EIS) of the supercapacitor before and after 3,000 cycles at a current density of 5.0 mA cm−<sup>2</sup> . An equivalent circuit was given in the inset of **Figure 4f**, which is similar to the circuit employed for the working electrode of a supercapacitor. The EIS data can be fitted by a bulk solution resistance R<sup>s</sup> , a chargetransfer resistance Rct and a pseudocapacitive element C<sup>p</sup> from the redox process of electrode materials, and a CPE to account for the double-layer capacitance (Pang et al., 2013). As shown in **Figure 4f**, the intrinsic resistance R<sup>s</sup> of the device before and after 3,000 cycles are around 27.2 and 38.1 , respectively. And the Rct after 3,000 cycles is around 218 ohms, which is higher than the 176 ohms of the initial Rct. The increase of charge transfer resistance may be due to the irreversible reaction between the electrodes and the electrolyte, which is consistent with the decrease in the capacitance after cycling (**Figure 3D**).

### CONCLUSION

In summary, a flexible all-solid-state NH4NiPO4·H2O//graphene device was successfully assembled, which showed great performance. When cycled for 3,000 cycle at the current density of 5.0 mA cm−<sup>2</sup> , the hybrid supercapacitor shows 88.8% in capacitance retention rate. The device also showed excellent flexibility, especially when bent to various degrees and under different weights. The as-prepared flexible all-solid-state device could be integrated in to large scale flexible systems that require an energy storage unit. And further study will be focused on improving the device performance.

5 g; (c) 20 g; (d) 50 g, and corresponding (e) CV curves within 0–1.4 V range at scan rate of 10 mV s−<sup>1</sup> when under different weights. (f) The Nyquist plots of the hybrid supercapacitor before and after 3,000 cycles at a current density of 5.0 mA cm−<sup>2</sup> , inset shows the EIS equivalent circuitry.

### AUTHOR CONTRIBUTIONS

YL, WZ, and HP conceived and designed the experiments. YL, XZ, FW, and KY performed the experiments. XZ, HW, and FR analyzed the data. YL and XZ wrote the paper. HP and WZ revised the paper, which could be found in the list of corrections we have submitted.

### FUNDING

This work is supported by the Program for New Century Excellent Talents of the University in China (grant no. NCET-13-0645) and Natural Science Foundation of China (21671170), the Program for Changjiang Scholars and Innovative Research Team in University (IRT\_16R21), Program for Innovative Research Team (in Science and Technology) in University of Henan Province (14IRTSTHN004),

### REFERENCES


Open Fund of National Joint Engineering Research Center for abrasion control and molding of metal materials (HKDNM201807, HKDNM201802), Henan International Science and Technology Cooperation Project of China (134300510051), Scientific Research Starting Foundation for Ph.D. of Henan University of Science and Technology (13480065), the Scientific and Technological Project of Henan Province (182102210297), the Student Research Training Plan of Henan University of Science and Technology (2018029), Science Foundation for Youths of Henan University of Science and Technology (2013QN006).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2019.00118/full#supplementary-material


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Liu, Zhai, Yang, Wang, Wei, Zhang, Ren and Pang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Facile Synthesis of Manganese Cobalt Oxide/Nickel Cobalt Oxide Composites for High-Performance Supercapacitors

Wang Chen Huo1†, Xiao Li Liu1†, Yun Song Yuan<sup>2</sup> \*, Nan Li <sup>3</sup> , Tian Lan<sup>3</sup> , Xiao Ying Liu<sup>4</sup> and Yu Xin Zhang<sup>1</sup> \*

*<sup>1</sup> State Key Laboratory of Mechanical Transmissions, College of Materials Science and Engineering, Chongqing University, Chongqing, China, <sup>2</sup> College of Urban Construction and Environmental Engineering, Chongqing University, Chongqing, China, <sup>3</sup> Aerospace Institute of Advanced Materials & Processing Technology, Beijing, China, <sup>4</sup> Engineering Research Center for Waste Oil Recovery Technology and Equipment, Ministry of Education, College of Environment and Resources,*

*Chongqing Technology and Business University, Chongqing, China*

#### Edited by:

*Qiaobao Zhang, Xiamen University, China*

#### Reviewed by:

*Shuge Dai, Zhengzhou University, China Jianglan Shui, Beihang University, China Kexin Yao, King Abdullah University of Science and Technology, Saudi Arabia Shuangxi Xing, Northeast Normal University, China Limin Jin, Shenzhen Graduate School of Harbin Institute of Technology, China*

#### \*Correspondence:

*Yun Song Yuan yuan\_ys@cqu.edu.cn Yu Xin Zhang zhangyuxin@cqu.edu.cn*

*†These authors have contributed equally to this work*

#### Specialty section:

*This article was submitted to Nanoscience, a section of the journal Frontiers in Chemistry*

Received: *15 September 2018* Accepted: *19 December 2018* Published: *17 January 2019*

#### Citation:

*Huo WC, Liu XL, Yuan YS, Li N, Lan T, Liu XY and Zhang YX (2019) Facile Synthesis of Manganese Cobalt Oxide/Nickel Cobalt Oxide Composites for High-Performance Supercapacitors. Front. Chem. 6:661. doi: 10.3389/fchem.2018.00661* Transition metal oxides (TMOs) with spinel structures have a promising potential as the electrode materials for supercapacitors application owning to its outstanding theoretical capacity, good redox activity, and eco-friendly feature. In this work, MnCo2O4.5@NiCo2O<sup>4</sup> nanowire composites for supercapacitors has been successfully fabricated by using a mild hydrothermal approach without any surfactant. The morphology and physicochemical properties of the prepared products can be well-controlled by adjusting experimental parameters of preparation. The double spinel composite exhibits a high specific capacitance of 325 F g−<sup>1</sup> (146 C g−<sup>1</sup> ) and 70.5% capacitance retention after 3,000 cycling tests at 1 A g−<sup>1</sup> .

Keywords: transition metal oxides, spinel structure, composites, supercapacitor, MnCo2O4.5@NiCo2O4 nanowire

### INTRODUCTION

The rising concerns about environmental crisis and increasing demand of renewable energy sources have attracted extensive attention for developing a secure, high-performance and sustainable storage technology (Lukatskaya et al., 2016; Zhang et al., 2018a). Recently, various energy storage technologies have emerged, such as lithium ion batteries (Zheng et al., 2018), Li-ion sulfur batteries (Sun et al., 2018) and supercapacitors (Qu et al., 2017) etc. Supercapacitors, also named as electrochemical capacitors (ECs), will be one of the most desirable power devices of next generation, owing to their high power density (Simon and Gogotsi, 2008), rapid charging-discharging rate (Miller and Simon, 2008), and excellent cycle stability (Dai et al., 2018). Various materials have been investigated as electrodes for ECs, such as carbon materials (Zhao et al., 2010; Niu et al., 2013; Zhang et al., 2018b), transition metal oxides (TMOs) (Liu et al., 2012; Nam et al., 2012) and conductive polymers (Lim et al., 2015). Generally, TMOs are superior in specific capacitance and stability. Their specific capacitance is 10∼100 times of carbon materials, and they have better stability than conductive polymers. Among the TMOs, the spinel structures, especially spinel ternary TMOs, including the MnCo2O4.5 (Li et al., 2014), NiCo2O<sup>4</sup> (Wang C. et al., 2016), and ZnCo2O<sup>4</sup> (Wu et al., 2015) etc., have been extensively investigated and exhibited an excellent electrochemical performance as electrode materials for ECs, due to its extremely high theoretical specific capacitance, good redox activity, and eco-friendliness (Qiu et al., 2018; Parveen et al., 2019), thus implying that these could be the most potential electrode materials for next-generation ECs.

Over the past few years, numerous researchers have been devoted to develop new strategies to enhance the electrochemical performance of NiCo2O<sup>4</sup> nanomaterials for ECs electrodes (Sun et al., 2016). For example, Wang C. et al. (2016) reported that the NiCo2O<sup>4</sup> nanoneedle was directly anchored on the Ni foam and carbon fabrics, respectively, via a facile hydrothermal method following with a calcination process in air. This strategy would enhance the combination of electrode material and substrate for improving the electrical conductivity and facilitating electrochemical performance. Sun et al. (2016) found that the porous NiCo2O<sup>4</sup> nanograss supported on Ni foam shown a surprisingly high specific capacitance of 807.7 F g−<sup>1</sup> at 1 mA cm−<sup>2</sup> (0.38 A g−<sup>1</sup> ) after suffering the hydrogenation process for 3 h, which is ascribed to the formation of oxygen vacancies in disordered surface layers during the hydrogenation process and enhance the electrical conductivity. And there are many other strategies to improve the agglomeration of nanomaterials and facilitating contacting of electrolyte and electrode to promote the performance, such as NiCo2O4@3DNF framework (Parveen et al., 2019) and Co9S8@NiCo2O<sup>4</sup> nanobrushes (Liu et al., 2019) etc. However, the price of Ni is eight times than the Mn (Information comes from the SMM Information & Technology Co, Ltd.), the high cost severely restricted its commercial applications. Hence, utilizing Mn to substitute Ni or constructing the Mn-Co-O@Ni-Co-O double spinel composites, should be a feasible and effective method for reducing the cost and promoting its application. Unfortunately, the Mn substituted spinel structure always exhibited the poor specific capacitance (Li et al., 2014, 2015; Hao et al., 2015; Wang K. et al., 2016). Thus, fabricating the Mn-Co-O@Ni-Co-O double spinel composites would be an alternative preferable strategy, which is not only decreasing the commercial applications cost, but also improving the electrochemical performance of double spinel composites for ECs via utilizing the synergistic effect of the core and shell.

Herein, the core-shell MnCo2O4.5@NiCo2O<sup>4</sup> double spinel structures were successfully synthesized by a facile hydrothermal route without any surfactant, where the core, hierarchical MnCo2O4.5 nanowires with a diameter of 300∼500 nm was provided by Li et al. (2015). The prepared MnCo2O4.5@NiCo2O<sup>4</sup> hybrids exhibit a promising electrochemical performance via comparing with literature results of selected samples which have similar components and structures (**Table 1**) (Wu et al., 2011; Kuang et al., 2014; Li et al., 2014, 2015; Hao et al., 2015; Sun et al., 2016; Wang C. et al., 2016; Wang K. et al., 2016). This strategy not only diminishes the dosage of Ni, but also elevates the electrochemical performance of the double spinel composites for ECs, which are favor of the commercial applications for next generation energy storage devices.

### EXPERIMENTAL SECTION

### Materials Synthesis

MnCo2O4.5 nanowires provided by Li's group were adopted as the reaction substrates in the experimental. The synthesis procedure of MnCo2O4.5 has been reported in detail in Li's work (Li et al., 2015). **Figure 1** shows the schematic illustration of the MnCo2O4.5@NiCo2O<sup>4</sup> nanowires, and the composites were fabricated as follows: 0.001 mol Ni(NO3)2·6H2O, 0.002 mol Co(NO3)2·6H2O, 0.002 mol NH4F and 0.005 mol urea were dissolved in 35 mL deionized water at room temperature, then 20 mg MnCo2O4.5 were added into the mixture, supplemented by ultrasonic treatment until the MnCo2O4.5 powder were dispersed uniformly. Then the suspension was put in the Teflonlined stainless autoclave, sealed and heated at 120◦C for 6 and 12 h, respectively. After cooling down to room temperature, collected precipitates were washed several times using ethanol and deionized water to remove the attached reaction products and/or residual reactants, then put into the vacuum oven at 60◦C to dry. Finally, the metal hydroxides were calcined at 350◦C for 2 h to convert into metal oxides most likely as described by the following equations (Xu and Wang, 2011):

$$\text{Ni}^{2+} + 2\text{Co}^{2+} + 6\text{OH}^- \rightarrow \text{NiCo}\_2\text{(OH)}\_6 \tag{1}$$

$$\text{NiCo}\_2\text{(OH)}\_6 + \frac{1}{2}\text{O}\_2 \rightarrow \text{NiCo}\_2\text{O}\_4 + 3\text{H}\_2\text{O} \tag{2}$$

### Characterization

X-ray photoelectron spectroscopy (Kratos XSAM800, XPS), the powder X-ray diffraction (D/max 2500, Cu Kα, XRD) and thermosgravimetric analyzer-differential scanning calorimeter (NETZSCH STA 449C, TGA-DSC) were used to characterize the crystallinity and components of prepared materials. Morphological structure was analyzed by scanning electron microscopy (Zeiss Auriga FIB/SEM). The detailed structures of the materials were collected by transmission electron microscopy (FEI TECNAI G2 F20, TEM).

### Electrochemical Measurement

In this work, the three-electrode configuration with the potentiostat/galvanostat model (CHI660E electrochemical workstation) was employed to detect the electrochemical performance of the prepared electrode materials, where nickel foam (1 × 1 cm) supported MnCo2O4.5@NiCo2O<sup>4</sup> composites worked as the working electrode, saturated calomel electrode (SEC) served as the reference electrode and platinum plate acted as the counter electrode in 3 M KOH electrolyte. The materials for the preparation of working electrode are active materials, acetylene black and polyvinylidene difluoride (PVDF) (mass ratio, 7:2:1). After calculation, the net mass of MnCo2O4.5@NiCo2O<sup>4</sup> loaded on each working electrode is 1.5 mg.

The cyclic voltammetry (CV) and charge-discharge (GCD) measurements were tested on a CHI660E electrochemical workstation. The CV curves were carried out at disparate scan rates between 0 and 0.5 V (vs. SCE), while the charge-discharge curves (0∼0.45 V vs. SCE) were monitored at different current densities. The electrochemical impedance spectroscopy (EIS) plots were obtained over the frequency from 10−<sup>2</sup> to 10<sup>5</sup> Hz with the 5 mV amplitude. The normalized specific capacitance (**C**, F g −1 ) and (**Q**, C g−<sup>1</sup> ) was calculated from charge-discharge curves and following the equation (Dai et al., 2017):

$$C = I \Delta t / \Delta V m \tag{3}$$

$$Q = I \Delta \text{t/m} \tag{4}$$


TABLE 1 | Comparison of specific capacitances of selected literature results obtained from materials with similar components and this work.

*All values are derived from characterizations in three-electrode systems.*

where I is the operated current (A), 1t the discharge time (s), m the mass of active electrode materials (g), and 1V the potential window of discharge (V).

### RESULTS AND DISCUSSION

The X-ray diffraction peaks of the prepared products were well-indexed to the cubic phase of MnCo2O4.5(JCPDS 32- 0297) (**Figure 2A**) (Hao et al., 2015). No additional peak for other phase of MnCo2O4.5 was observed and the sharp and narrow diffraction peaks forecast that the MnCo2O4.5 microstructures have high crystallinity. The diffraction peaks of the MnCo2O4.5@NiCo2O<sup>4</sup> have minor left shift (0.1◦ ) compared with MnCo2O4.5 (**Figure 2B**), which is ascribed the crystallinity of NiCo2O<sup>4</sup> (JCPDS 20-0781) is very similar to MnCo2O4.5 (Sun et al., 2016; Wang C. et al., 2016), but the lattice constant is tiny bigger than MnCo2O4.5. NiCo2O<sup>4</sup> and MnCo2O4.5 are both spinel structured crystals and have a high degree of lattice matching (∼0.3% mis-match). The lattice parameters of MnCo2O4.5 are 8.08<sup>∗</sup> 8.08<sup>∗</sup> 8.08<90.0<sup>∗</sup> 90.0<sup>∗</sup> 90.0>, and NiCo2O<sup>4</sup> 8.11<sup>∗</sup> 8.11<sup>∗</sup> 8.11<90.0<sup>∗</sup> 90.0<sup>∗</sup> 90.0> and the **Figures 3A,B** show the crystal structure of MnCo2O4.5 and NiCo2O4, respectively. Moreover, in the XRD results the peak intensities of the (311) and (440) planes are higher than others, therefore NiCo2O<sup>4</sup> is most likely to grow along the (311) and (440) planes and completely coherent with MnCo2O4.5 at the (311) and (440) planes. **Figures 3C,D** shows the possible epitaxial growth patterns of MnCo2O4.5@NiCo2O<sup>4</sup> as mentioned above.

Thermogravimetric (TGA) and differential scanning calorimetry (DSC) were performed at a 10◦C min−<sup>1</sup> heating rate to investigate the thermal properties (**Figure 2C**). The TGA-DSC curve of the Ni-Co hydroxide on MnCo2O4.5 is shown in **Figure 2C**. The evaporation of the physically adsorbed water resulted in the weight loss (about 3%) till 280◦C. And the followed 17% weight loss between 280 and 400◦C is most likely to be caused by the crystalliferous water loss and the decomposition of Ni-Co hydroxide. There is no obvious weight loss after 400◦C. Based on these findings, 350◦C is chosen and believed to be sufficient for calcination treatment of hydroxides.

SEM was adopted to investigate the morphologies of asprepared composites. As shown in **Figure S1**, the MnCo2O4.5 nanowires exhibit fibrous morphologies with smooth surface. The length MnCo2O4.5 nanowires are several micrometers and the average diameter is about 300∼500 nm. **Figures 4A,B** shows that NiCo2O<sup>4</sup> nanowires with smaller diameter are successfully and uniformly grown on the surface of MnCo2O4.5 nanowires. When the reaction time increases from 6 to 12 h, dense NiCo2O<sup>4</sup> nanowires are observed (**Figure 4B**) with more nanowires locating within the unit area of MnCo2O4.5 surface. The two insets in **Figures 4A,B** are the images of corresponding samples at higher magnifications. The EDS mapping demonstrates the uniform distribution of Mn, Ni,

FIGURE 2 | (A) XRD patterns of MnCo2O4.5 nanowires and MnCo2O4.5@NiCo2O4 composites. (B) The enlarged XRD patterns at the 2θ of 30–40. (C) TGA-DSC curves of Ni-Co hydroxide@MnCo2O4.5 composites.

Co, and O (**Figures 4C,D**), suggesting successful fabrication of well-distributed MnCo2O4.5@NiCo2O<sup>4</sup> nanowires and forming the core-shell structure. Furthermore, the detailed structural information of the coated NiCo2O4.5 shell nanowires was collected by TEM at different magnifications (**Figures 5A,B**). The primary NiCo2O<sup>4</sup> nanowires have a diameter of 50∼80 nm. High-resolution TEM (HRTEM, **Figure 5B**) image of the NiCo2O<sup>4</sup> nanowires reveal well-resolved lattice fringe having an inter-planar spacing of 0.20 and 0.47 nm (**b-1**), which are wellconsistent with the distance of the (400) and (111) plane of

NiCo2O4, respectively. And the (**b-2**) shows inter-planar spacing of 0.24 and 0.47 nm, which represent the (311) and (111) facet of NiCo2O4. Therefore, these results about the crystal structure and facets illustrate that the (01-1) facets of NiCo2O<sup>4</sup> are exposed (**Figures 5C,D**).

XPS measurement was employed to reveal the composition and surface chemical states of the MnCo2O4.5@NiCo2O4, and the detailed results are shown in **Figure 6**. It can be clearly seen from survey spectrum (**Figure 6A**) that the MnCo2O4.5@NiCo2O<sup>4</sup> consists of O, Ni, Co, and Mn elements, being consistent with

EDS results (**Figure 3D**). The high resolution XPS spectrum of Ni 2p (**Figure 6B**) reveals that two obvious shakeup satellites (indicated as "Sat.") are close to two spin orbit doublets at 855.4 and 872.8 eV, which represents the Ni 2p3/<sup>2</sup> and Ni 2p1/<sup>2</sup> splitting in Ni2<sup>+</sup> chemical state, respectively (Hareesh et al., 2016; Sun et al., 2016; Wang et al., 2017; Zhang et al., 2018b). In the Co 2p XPS spectrum, the spin orbit splitting to the Co 2p1/<sup>2</sup> (795.0 eV) and Co 2p3/<sup>2</sup> (780.1 eV) reaches 14.9 eV (**Figure 6C**), indicating the co-existence of Co2<sup>+</sup> and Co3<sup>+</sup> in MnCo2O4.5@NiCo2O<sup>4</sup> (Wu et al., 2015; Liu et al., 2017; Wang et al., 2017). Likewise, the peaks of Mn element locating at 641.9 eV and 653.6 eV in **Figure 6D** consistent with Mn 2p3/<sup>2</sup> and Mn 2p1/2, respectively (Hui et al., 2016; Liu et al., 2017; Zhao et al., 2018; Zhu et al., 2018).

The electrochemical performance of the samples was analyzed by CV and GCD measurements. **Figures 7A,B** illustrates the CV and GCD curves of the three different electrodes at the same scan rate (100 mV S−<sup>1</sup> ) and current density (1 A g−<sup>1</sup> ), respectively. MnCo2O4.5@NiCo2O4-12h displays better rate performance with lower distortion compared with MnCo2O4.5@NiCo2O4-6h (**Figure 7B**). This is possibly due to the mass increasing of NiCo2O<sup>4</sup> and the lower mass transport resistance within smaller interstitial regimes between the electrolyte and electrode materials. In **Figure 7C**, the CV curves of MnCo2O4.5@NiCo2O4-12h exhibit a quasireversible oxidative and reductive shape, indicating an ideal capacitance characteristic and a good reversibility of the architectures with good electronic conductivity and low internal resistance (Zhu et al., 2015; Hui et al., 2016; Sun et al., 2016; Wang et al., 2017). With the increment of scan rates, slight distortion of curves is observed, implying the favorable rate ability. The galvanostatic charge-discharge results of MnCo2O4.5@NiCo2O4-12h are presented in **Figure 7D**. Similarly, pseudo-capacitive behaviors are observed and the shape of GCD curves is not triangular ones which are believed to be pure electrical double layer capacitors (EDLCs).

The symmetrical shapes of the charge side and discharge side indicate the good reversibility of the synthesized composites. The MnCo2O4.5@NiCo2O4-12h composites (**Figure 7D**) have a higher specific capacitance of 325 F g−<sup>1</sup> (146 C g−<sup>1</sup> ) than that of MnCo2O4.5@NiCo2O4-6h [162.5 F g −1 (73 C g−<sup>1</sup> ), **Figure 7B**] at the same discharge current density (1 A g−<sup>1</sup> ), which is owning to its higher mass ratio of active NiCo2O4. Compared with pristine MnCo2O4.5, the electrochemical performance of synthesized core-shell composites is greatly improved based on the following two effects: MnCo2O4.5@NiCo2O4-12h has larger accessible surface area from loose NiCo2O<sup>4</sup> shell layer which is built with isolated free-standing nanowires with smaller diameters; and shorter ion diffusion distance from smaller free-standing NiCo2O<sup>4</sup> nanowires and interstitial regime between nanowires as well.

**Figure 7E** depicts the electrical impedance spectroscopy (EIS) of the composites before and after 3,000 cycling tests. The inset is magnified spectrum at low impedance region. The impedance spectrum is a semicircle parts at the high frequency and a linear curve at the low frequency. The intercept of the curves on the real axis reveals the total value of the ohmic resistance of electrolyte added with the resistance of the active materials. After cycling test, the radius of the semicircle increased obviously while first intercept and slope of the linear part change slightly. The apparent increasing of semicircle radius means more difficult for charge transfer of active species inside the electrode materials. The minor increasing of first intercept and decreasing of slope indicate higher contacting impedance and mass transport resistance after cycling tests. All the changes are derived from the potential structural damage of electrode materials during cycling tests. However, the overall capacitance retention is still up to 70.5% after 3,000 cycling tests (**Figure 7F**), which means this damage over cycling is at low level. This good cycling ability of MnCo2O4.5@NiCo2O<sup>4</sup> is attributed to its unique one-dimensional structure. The one-dimensional structure has much free spaces between the nanowires, being able to accommodate the volume change of the materials during cycling and maintain the activity of the materials at high level.

### CONCLUSIONS

The MnCo2O4.5@NiCo2O<sup>4</sup> composites with core-shell structure has been successfully synthesized with a facile and green hydrothermal method for high performance supercapacitors. The nanostructures, morphology and electrochemical performance of the composites are fully demonstrated. The MnCo2O4.5@NiCo2O<sup>4</sup> composites exhibit remarkable capacitive performance with maximum capacitances of 325 F g−<sup>1</sup> (146 C g −1 ) and retention 70.5% of the initial capacitance after 3,000 cycles at 1 A g−<sup>1</sup> . This excellent electrochemical performance of MnCo2O4.5@NiCo2O<sup>4</sup> composites are attributed to the unique core-shell double spinel structures with shell built of isolated free-standing nanowires. The free spaces between nanowires are able to provide channels for mass transfer of active species and accommodate the volume change during cycling.

### REFERENCES


### AUTHOR CONTRIBUTIONS

YXZ and YSY designed the research. WCH, XLL, and XYL performed the experiments. NL and TL carried out the partial experimental characterization. All authors incorporated in the discussion of experimental results.

### ACKNOWLEDGMENTS

We gratefully thank the financial support of the Fundamental Research Funds for the Central Universities (2018CDYJSY0055 and 106112017CDJXSYY0001), the National Natural Science Foundation of China (Grant no. 21576034), the key scientific and technological projects of Chongqing Municipal Education Commission (KJZD-K201800801), Joint Funds of the National Natural Science Foundation of China-Guangdong (Grant no. U1801254), the project funded by Chongqing Special Postdoctoral Science Foundation (XmT2018043), and Technological projects of Chongqing Municipal Education Commission (KJZDK201800801). We also thank the Electron Microscopy Center of Chongqing University for materials characterizations.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2018.00661/full#supplementary-material

rate supercapacitors. CrystEngComm 16, 2335–2339. doi: 10.1039/c3ce4 2581a


framework. Electrochim. Acta 293, 84–96. doi: 10.1016/j.electacta.2018.0 8.157


Nano Res. 4, 1013–1098. doi: 10.1007/s12274-011-0 160-7


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Huo, Liu, Yuan, Li, Lan, Liu and Zhang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Morphology Controllable Synthesis of NiO/NiFe2O<sup>4</sup> Hetero-Structures for Ultrafast Lithium-Ion Battery

Ying Wang<sup>1</sup> , Shengxiang Wu<sup>1</sup> , Chao Wang<sup>1</sup> \*, Yijing Wang<sup>2</sup> \* and Xiaopeng Han2,3 \*

<sup>1</sup> School of Chemistry & Materials Science, Jiangsu Key Laboratory of Green Synthetic Chemistry for Functional Materials, Jiangsu Normal University, Xuzhou, China, <sup>2</sup> Key Laboratory of Advanced Energy Materials Chemistry (MOE), College of Chemistry, Nankai University, Tianjin, China, <sup>3</sup> School of Materials Science and Engineering, Tianjin Key Laboratory of Composite and Functional Materials, Tianjin University, Tianjin, China

### Edited by:

Jiexi Wang, Central South University, China

### Reviewed by:

Hongshuai Hou, Central South University, China Shengjie Peng, Nanjing University of Aeronautics and Astronautics, China Jianmin Ma, Hunan University, China Vito Di Noto, Università degli Studi di Padova, Italy

#### \*Correspondence:

Chao Wang wangc@jsnu.edu.cn Yijing Wang wangyj@nankai.edu.cn Xiaopeng Han xphan@tju.edu.cn

#### Specialty section:

This article was submitted to Physical Chemistry and Chemical Physics, a section of the journal Frontiers in Chemistry

Received: 26 September 2018 Accepted: 17 December 2018 Published: 10 January 2019

#### Citation:

Wang Y, Wu S, Wang C, Wang Y and Han X (2019) Morphology Controllable Synthesis of NiO/NiFe2O4 Hetero-Structures for Ultrafast Lithium-Ion Battery. Front. Chem. 6:654. doi: 10.3389/fchem.2018.00654 Rational design of high performance anode material with outstanding rate capability and cycling stability is of great importance for lithium ion batteries (LIBs). Herein, a series of NiO/NiFe2O<sup>4</sup> hetero-structures with adjustable porosity, particle size, and shell/internal structure have been synthesized via a controllable annealing process. The optimized NiO/NiFe2O<sup>4</sup> (S-NFO) is hierarchical hollow nanocube that is composed of ∼5 nm subunits and high porosity. When being applied as anode for LIBs, the S-NFO exhibits high rate capability and excellent cycle stability, which remains high capacity of 1,052 mAh g−<sup>1</sup> after 300 cycles at 5.0 A g−<sup>1</sup> and even 344 mAh g−<sup>1</sup> after 2,000 cycles at 20 A g−<sup>1</sup> . Such impressive electrochemical performance of S-NFO is mainly due to three reasons. One is high porosity of its hierarchical hollow shell, which not only promotes the penetration of electrolyte, but also accommodates the volume change during cycling. Another is the small particle size of its subunits, which can effectively shorten the electron/ion diffusion distance and provide more active sites for Li<sup>+</sup> storage. Besides, the hetero-interfaces between NiO and NiFe2O<sup>4</sup> also contribute toitsfast charge transport.

Keywords: lithium ion battery, NiO/NiFe2O4 , morphology control, hetero-structure, electrochemical performance

### INTRODUCTION

Lithium ion batteries (LIBs), with the advantages of high energy density and environmental benignity, have become the most widely used energy storage systems for portable electronic equipment. The growing needs for high-performance electronic devices and electric vehicles, however, constantly demand LIBs for further innovation, in terms of higher energy/power density, longer lifetime, greater rate capability, and lower cost. It is therefore crucial to develop high-performance electrode materials (Poizot et al., 2000; Goriparti et al., 2014; Peng et al., 2015; Wang et al., 2015a,b; Hou et al., 2016; Pagot et al., 2017; Wei et al., 2018; Xu et al., 2018; Zhang et al., 2018; Huang et al., in press; Liu et al., in press), or to develop other energy storage devices, such as sodium ion battery, metal-O<sup>2</sup> battery (Han et al., 2014, 2017, 2018), and Li-S battery (Kim et al., 2016; Shen et al., 2018). As for the anode materials in LIBs, NiFe2O<sup>4</sup> has drawn extensive attention, due to its low price, earth abundant, and high theoretical capacity (915 mAh g−<sup>1</sup> ) (Park et al., 2015; Jin et al., 2017; Lina et al., 2017). During the Li<sup>+</sup> insertion/extraction, however, NiFe2O<sup>4</sup> suffers from unsatisfied capacity retention and poor rate capability that causes by the server volume change and limited electrochemical kinetics. Although numerous strategies have been proposed to improve the electrochemical performance of NiFe2O<sup>4</sup> (Gao et al., 2017; Jiang et al., 2017; Li D. et al., 2017; Zhou et al., 2017), it is still a big challenge to deliver decent capacities at high current density (above 10 A g−<sup>1</sup> ) or well-capacity retention over 2000 cycles.

Smart design of hetero-structures is an emerging strategy to improving the electrochemical kinetics of transition metal oxides (TMOs) (Zheng X. et al., 2018). For example, Hou et al. reported the appealing electrochemical performance of ZnFe2O4/ZnO, which outperformed the single ZnFe2O<sup>4</sup> (Hou L. et al., 2015). Ting et al. observed the fast charge transfer in the CoO/Cu2O hetero-structure, where Li<sup>+</sup> diffusion kinetics and electronic conductivity were promoted (Tingting et al., 2017). Considering the high theoretical capacity (718 mAh g−<sup>1</sup> ) and non-pollution of NiO (Shi et al., 2018; Yin et al., 2018), hybridizing NiO with NiFe2O<sup>4</sup> host great promising for high performance LIBs. However, only limited studies have been paid to the Li storage performance of NiO/NiFe2O<sup>4</sup> hybrid (Du et al., 2016).

Tailoring the nano-architecture of electrode is another important approach to enhance the electrochemical performance. Previous literatures had evidenced that morphology of electrode materials can effectively affect its electrochemical performance (Li et al., 2013; Hou H. et al., 2015; Liang et al., 2018). Particularly, hierarchical hollow electrode have received great attention (Chen et al., 2015; Zheng Z. et al., 2018). Nowadays, TMOs in different morphologies, such as nanotube (Gang et al., 2014; Huang et al., 2014), porous plates (Hui et al., 2016; Wang et al., 2016), and hollow octahedron, have been realized. However, the relationship between morphology of electrode and its electrochemical performance is still under investigation.

Taking all the discussion above into consideration, fabricating NiO/NiFe2O<sup>4</sup> hetero-structure, and further tailoring its morphology seem to have great potential for ultrafast LIBs. Ni3[Fe(CN)6]2, a typical Prussian blue analog, was selected as precursor to synthesize morphology controllable NiO/NiFe2O<sup>4</sup> hetero-structures, since it contains both Fe and Ni element at the same. By taking advantage of the unique reactivity and thermal stability of Ni3[Fe(CN)6]2, herein, we obtain a series of NiO/NiFe2O<sup>4</sup> with adjustable porosity, particle size, and shell/internal structure via a simple calcination procedure. The obtained samples are porous filled-nanocubes (P-NFO), hierarchical hollow-nanocubes with ∼5 nm subunits (S-NFO), and hierarchical hollownanocubes with ∼12 nm subunits (L-NFO). The optimized S-NFO exhibits excellent Li<sup>+</sup> storage performance, with high capacity of ∼1,052 mAh g−<sup>1</sup> after 300 cycles at 5.0 A g−<sup>1</sup> and even ∼344 mAh g−<sup>1</sup> after 2,000 cycles at 20 A g−<sup>1</sup> . The superior electrochemical performance of S-NFO is attributed to its unique structural advantages, which combined the hetero-structure, high porosity, and small particle size. Further investigation reveals that the electrochemical reaction of S-NFO is dominated by capacitive behavior. The simple and large-scalable synthesis of these NiO/NiFe2O<sup>4</sup> hetero-structures may shed light on design for other high performance electrode materials.

### EXPERIMENTAL

### Synthesis of NiO/NiFe2O<sup>4</sup> Hetero-Structures

NiO/NiFe2O<sup>4</sup> hetero-structures were obtained as follows. First, Ni3[Fe(CN)6]<sup>2</sup> (NiFe-PBA) was synthesized according to previous literatures (Xuan et al., 2018) with minor modifications. Second, the NiFe-PBA was transferred into a tube furnace, heated to the target temperature (350, 450, or 550◦C) at a heating ramp of 0.5◦C min−<sup>1</sup> , and held for 6 h in air atmosphere. The resultant samples were marked as P-NFO, S-NFO, or L-NFO according to their respective morphology.

### Characterization

X-ray diffraction (XRD, Bruker, D8-Advance), field-emission scanning electron microscopy (FESEM, JEOL, SU8010), field-emission transmission electron microscope (FETEM), highresolution TEM (HR-TEM, Tecnai G2 F20 S-TWIN), and energy dispersive X-ray (EDX) analysis (taken with X-ray spectroscopy attached to the Tecnai G2 F20 S-TWIN), were employed to characterize the phase and morphological structures. Thermogravimetric analysis (TGA, TA-Q50) was performed to analyze the thermal stability. N<sup>2</sup> adsorption/desorption isotherms (Quantachrome, Autosorb-IQ2-VP) was conducted to obtain the physical surface area and pore distribution. X-ray photoelectron spectroscopy (XPS, Thermo ESCALAB 250XI) were employed to detect the elemental composition and surface oxidation states of synthesized samples.

### Electrochemical Measurements

Electrochemical performance of as-prepared samples were carried out by assembling standard 2,032 coin cells in an argonfilled glove box with the oxygen and water content below 0.1 ppm. Active materials, super conductive carbon black, and poly(vinyldifluoride) (PVDF, Sigma Aldrich) were mixed in a weight ratio of 7:2:1, dispersed in N-methyl-2-Pyrrolidinone (NMP), then milled for 30 min to form a slurry. The slurry was cast onto copper foil using a doctor blade and vacuum dried at 120◦C overnight. 1 M LiPF<sup>6</sup> (Sigma Aldrich) in ethylene carbonate (EC, Sigma Aldrich), diethyl carbonate (DEC, Alfa Aesar), and fluorinated ethylene carbonate (FEC, Sigma Aldrich) (volume ratio 6:3:1) was used as the electrolyte. Polypropylene (PP, MTI Cooperation) was used as the separator. For halfcells, a lithium disc (MTI Corporation) was used as the counter electrode. Galvanostatic charge-discharge tests were carried out at room temperature on a battery testing system (LAND Wuhan, China) in a potential range of 0.01–3.00 V (vs. Li/Li+). Cyclic voltammetry (CV) tests and electrochemical impedance spectroscopy (EIS) measurements were performed on a CHI-660E electrochemical work station. For the full-cells, the cathodes were assembled by mixing LiCoO<sup>2</sup> (Sigma Aldrich) with carbon black, and PVDF in a weight ratio of 8:1:1. The electrolytes and separator in full-cells were same as those in the half-cells. Electrochemical performance of full-cells were tested in a voltage window between 1.0 and 3.9 V. The weight ratio of positive materials to negative materials was designed as 4:1. Specific

capacity of all the cells was calculated based on only the mass of active materials in anode.

### RESULTS AND DISCUSSION

Crystallographic structure and purity of the NiFe-PBA precursors were studied by XRD (**Figure S1a**). All the diffraction peaks match well with fcc Ni3[Fe(CN)6]<sup>2</sup> (JCPDS No. 86-0501), suggesting the high purity of NiFe-PBA. The SEM image (**Figure S1b**) shows that NiFe-PBA are uniform nano-cubes with an average size of ∼180 nm and smooth surfaces. The synthesis process of NiFe-PBA was referred to previous literatures (Xuan et al., 2018): the formation of these NiFe-PBA cube is caused by coordinate process between Ni2<sup>+</sup> and [Fe(CN)6] <sup>3</sup>−, and

FIGURE 1 | (a) Schematic illustration for morphology evaluation of NiO/NiFe2O4 hetero-structures. (b–d) SEM images, (e–g) TEM pictures, and (h–j) N2 isothermals of the P-NFO, S-NFO, and L-NFO, respectively. Insets in (b–d) and (e–g) are the enlarged SEM and TEM images of an individual P-NFO, S-NFO, and L-NFO, respectively.

dominated by a kinetically controlled process to obtain uniform cubes (Hu et al., 2013). TGA curve (**Figure S2**) reveals thermal stability of the NiFe-PBA, which starts to decompose at 252 ◦C in air. Therefore, the NiFe-PBA was heated in air at 350, 450, or 550 ◦C to obtain the morphology control of NiO/NiFe2O<sup>4</sup> hetero-structures.

The NiO/NiFe2O<sup>4</sup> hetero-structures that collected after controllable annealing at different temperatures show distinct nano-architectures (**Figure 1a**). The sample that annealed at 350◦C well retains the cubic shape of the Ni-Fe PBA with a diameter size of ∼130 nm. The magnified SEM image (inset **Figure 1b**) reveals the integrate surface of the P-NFO sample. When elevated the calcination temperature to 450 ◦C, the size of cubes shrinks further to ∼100 nm and hierarchical surfaces are formed, which consist of about 5 nm nanoparticles (inset of **Figure 1c**). As for the L-NFO, although hierarchical surfaces are still observed, the subunits grow to a larger size of around 12 nm (inset **Figure 1d**). TEM images (**Figures 1e–g**) display more morphological details about the samples. The P-NFO shows a highly porous structure with filled internal. Internal cavities are observed in the S-NFO and L-NFO, confirming the formation of hierarchical hollow-nanocubes. Interestingly, besides the particle size of subunit, the internal cavity also grows larger in the L-NFO compared to that of S-NFO sample. The formation of these distinct NiO/NiFe2O<sup>4</sup> samples is dominated by two major factors: thermally induced oxidation process of the NiFe-PBA, and further growth of the formed NiO/NiFe2O<sup>4</sup> composite (Zhang et al., 2013). In the former, metallic elements in NiFe-PBA

FIGURE 2 | (a) XRD patterns of P-NFO, S-NFO, and L-NFO. Chemical and structural investigation of S-NFO: (b) XPS spectrum, (c–e) high-resolution XPS spectrum for Ni 2p, Fe 2p, and O 1s, (f) EDX elemental mapping, and (g) HR-TEM images.

react with oxygen to form homogenous NiO/NiFe2O<sup>4</sup> heterostructure. As the annealing temperature raises to higher value, the rapid mass-transport from core to shell causes the formation of hollow interior in resultant samples (S-NFO and L-NFO) (Guo et al., 2015). Meanwhile, the latter results a further growth of NiO/NiFe2O<sup>4</sup> subunits, that following an Oswald ripening process. Therefore, compared with S-NFO, the L-NFO shows a larger interior hollow and bigger subunit size.

N<sup>2</sup> adsorption-desorption isothermals (**Figures 1h–j**) of these samples show a similar type form, exhibiting typical IV with H1 hysteresis loops, indicative of their mesoporous structure. The specific surface area is 45.2, 32.1, and 31.3 m<sup>2</sup> g −1 for P-NFO, S-NFO, and L-NFO sample, respectively. When increasing the annealing temperature, both the particle size of subunits and interior cavity grow to larger values, therefore, the smallest surface area is observed in the L-NFO sample, which is in agreement with the observation in **Figure 1**. The narrow and small hysteresis loops also suggest co-existence of macropores in all samples. Notably, the combination of mesopores and macropores can facilitate better penetration of electrolyte, which is an important factor for enhancing the electrochemical performance (Jiang et al., 2018). The pore volume of P-NFO, S-NFO, and L-NFO are 0.204, 0.210, and 0.188 cm<sup>3</sup> g −1 , respectively. High pore volume of the P-NFO and S-NFO indicates that the good porous feature of NiFe-PBA is well-reserved in both samples. Therefore, NiO/NiFe2O<sup>4</sup> hetero-structures with different porosity, particle size, and shell/internal structures are successfully synthesized via a simple controllable annealing process.

**Figure 2** presents more structural information about the resultant samples. XRD patterns of the P-NFO, S-NFO, and L-NFO (**Figure 2a**) match well with crystallographic structure of cubic NiO (JCPDS No. 01-1239) and NiFe2O<sup>4</sup> (JCPDS No. 01-074-2081), confirming their hybrid composition. The XRD pattern of S-NFO was selected for GSAS Rietveld refinement to determine the phase content of NiFe2O<sup>4</sup> and NiO, as shown in **Figure S3**. According to the refinement result, the phase content of NiFe2O<sup>4</sup> and NiO is determined to be 63 and 37%. The broad diffraction peak of the XRD pattern indicated a smaller particle size. Specifically, the particle size of the P-NFO, S-NFO, and L-NFO is calculated to be 7.67, 5.73, and 13.14 nm, respectively, according to the Scherrer equation based on the peak at 43.4 degree. XPS spectra of the S-NFO are shown in **Figures 2b–e**. XPS survey reveals the existence of only Ni, Fe, and O elements, suggesting high purity of the S-NFO sample. In the Ni 2p spectrum, the main peak at 855.7 eV (Ni 2p3/2) and the satellite peak at 860.7 eV are the typical Ni2<sup>+</sup> bond in NiO and NiFe2O<sup>4</sup> (Song et al., 2018). In **Figure 2d**, the dominate peaks at 710.0 eV, together with the satellite peak are the typical Fe 2p3/<sup>2</sup> signal of NiFe2O<sup>4</sup> (Gao et al., 2017). For the O1s spectrum, peaks locate at 528.9 and 530.0 eV are assigned to the typical metal-oxygen bonds (Jin et al., 2017). The O1s peak at 532.1 eV is most likely associated with defects and under-coordinated lattice oxygen (Qiu et al., 2015; Yuan et al., 2015). Elemental mapping of an isolated S-NFO nanocube is used to investigate its chemical composition (**Figure 2f**). The hollow structure of S-NFO is further confirmed by the color contrast between shell (darker) and hollow interior (lighter). Clearly, the S-NFO consists of Ni, Fe, and O elements, and all the elements are homogeneous distributed, indicating homodistribution of NiO/NiFe2O<sup>4</sup> hetero-structures. Taking the Fe mapping for example, the homogenous distribution of Fe in S-NFO is confirmed by the full cover of green dots through the selected cube. Comparing with the intensive part at the edge, these green dots are less intensive in the interior, proving its hollow structural feature. The uniform elemental distribution was also observed in the P-NFO and L-NFO (**Figures S4a–f**). HR-TEM image in **Figure 2g** reveals that the S-NFO exhibits a hierarchical structure that is composed of small nanoparticles (∼5 nm) as its subunits. The lattice fringes of 0.21 and 0.25 nm can be well indexed to the (200) and (311) lattice planes of cubic NiO and NiFe2O4, revealing the intimate junction between NiO and NiFe2O4. The weight ratio of Ni to Fe in S-NFO was determined to 32: 31 by EDS spectrum (**Figure S5**). Therefore, the S-NFO consists of 65 wt.% NiFe2O<sup>4</sup> and 35 wt.% NiO, similar with the XRD refinement result.

Electrochemical measurements of the NiO/NiFe2O<sup>4</sup> heterostructures were carried out to clarify the relationship between porosity, particle size and electrochemical performance. The charge-discharge curves of P-NFO, S-NFO, and L-NFO at the first, second, fifth, ninth, and tenth cycle are given in **Figures 3A–C**. In the first discharge process, all samples exhibit similar profiles with an obvious plateau at ∼0.75 V, which shifts to ∼1.0 V and remains stable in the subsequent cycles. The upward shifted voltage platform is caused by structural reorganization, new phase formation, and a polarization change of electrodes materials (Zou et al., 2014). The initial discharge capacity of P-NFO, S-NFO, and L-NFO is 1314, 1805, and 813 mAh g−<sup>1</sup> , respectively. The charge profiles are relatively smooth with two bumps at ∼1.3 and 2.3 V, corresponding to the oxidation of Ni<sup>0</sup> and Fe<sup>0</sup> (Park et al., 2015; Du et al., 2016). The initial charge capacity is 986, 1330, and 701 mAh g −1 for the P-NFO, S-NFO, and L-NFO, corresponding to an initial coulombic efficiency (ICE) of 75, 74, and 86%, respectively. The irreversible capacity loss in the first cycle may be attributed to formation of solid electrolyte interface (SEI) film, and irreversible side reactions of the NiO/NiFe2O<sup>4</sup> composites (Park et al., 2015).The higher ICE value of the L-NFO is due to its smaller surface area, which reduces the formation of the SEI film. In the 10th cycle, the P-NFO, S-NFO, and L-NFO show a stable discharge/charge capacity of 1092/1077, 1539/1501, 814/802 mAh g−<sup>1</sup> , respectively. The high capacity of S-NFO [theoretical capacity is 846 mAh g−<sup>1</sup> , 718 × 35% (NiO)+915×65% (NiFe2O4)] is caused by the extra contribution of its nano-subunits, and other factors such as kinetic limitation and/or intrinsic nature of materials (Hou L. et al., 2015; Yuan et al., 2015). Superior electrochemical performance of the S-NFO indicates that the smaller subunits and high porosity play important roles in affecting the Li<sup>+</sup> insertion/extraction process (see below for detailed discussions).

**Figure 3D** displays the typical cyclic voltammetry (CV) curves of S-NFO at a scan rate of 0.5 mV s−<sup>1</sup> , which is in accordance with the above charge-discharge profiles. A dominate reduction peak at 0.4 V is observed in the first cathodic sweep, which is assigned to the reduction of NiO and NiFe2O<sup>4</sup> according to Equation (1, 2), and the formation of the SEI film (Guo et al., 2015; Chen et al., 2018). The weak and broad peak at 1.1 V might be due to the insertion of Li<sup>+</sup> into electrode. In the following cycles, these two peaks shift to higher potential of 0.7 and 1.3 V, indicating irreversible capacity loss. Two peaks at 1.7 and 2.3 V in the anodic sweep are observed, which could be ascribed to the oxidation of Ni<sup>0</sup> to Ni2+, and Fe<sup>0</sup> to Fe3+, respectively Equations (3,4) (Guo et al., 2015; Chen et al., 2018). The highly overlapped CV curves in the subsequent cycles suggest a good reversibility of the electrochemical reactions.

$$2\text{NiO} + 2Li^{+} + 2e^{-} \rightarrow \text{Ni} + Li\_{2}O \tag{1}$$

$$2\text{ NiFe}\_2\text{O}\_4 + 8Li^+ + 8e^- \rightarrow Ni + 2Fe + 4Li\_2\text{O} \tag{2}$$

$$\text{Ni} + \text{Li}\_2\text{O} \rightarrow \text{NiO} + 2\text{Li}^+ + 2\text{e}^- \tag{3}$$

$$2Fe + 3Li\_2O \to Fe\_2O\_3 + 6Li^+ + 6e^- \tag{4}$$

Significantly improved rate performance is observed in the S-NFO in comparison with P-NFO and L-NFO benchmarks (**Figure 3E**). As the current density progressively increased from 1.0 to 10 A g−<sup>1</sup> , the S-NFO delivers high reversible capacities of 1292, 1130, 1004, 856, and 776 mAh g−<sup>1</sup> at 1.0, 3.0, 5.0, 8.0, and 10 A g−<sup>1</sup> , respectively. Even at ultrahigh current densities of 15 and 20 A g−<sup>1</sup> , high capacities of 641 and 522 mAh g−<sup>1</sup> are still observed. More importantly, as the current density drops to 1.0 A g−<sup>1</sup> , discharge capacity of the S-NFO gradually recovers to 1189 mAh g−<sup>1</sup> as well. Apparently, the S-NFO shows high capacity retention of 92 % after 130 cycles at various current densities (from 1.0 to 20 A g−<sup>1</sup> ). Such excellent rate performance outperforms not only P-NFO and L-NFO, but also compete with other NiFe2O<sup>4</sup> or NiO electrodes (Park et al., 2015; Yu et al., 2015; Li C. et al., 2017) and some other TMOs/MTMOs hybrids (Zou et al., 2014; Hou L. et al., 2015; Yuan et al., 2015). Detailed rate capability comparison of this S-NFO with other NiFe2O<sup>4</sup> and NiO based electrodes is illustrated in **Figure 3F**.

The distinct electrochemical performance of P-NFO, S-NFO, and L-NFO hetero-structures may be caused by their different porosity, particle size, and internal structure (**Figure 4**). For full utilization of active material, a better contact between electrolyte and electrode as well as a shorter path for Li<sup>+</sup> transportation is needed. The hollow and hierarchical architecture of S-NFO simultaneously facilitates better electrolyte penetration and Li<sup>+</sup> transportation. The small subunits of S-NFO not only effectively shorten the diffusion path of Li<sup>+</sup> to improve the kinetics, but also provide more active sites for Li<sup>+</sup> storage. Although the L-NFO also possesses a hierarchical hollow structure, its subunits are larger than the S-NFO, corresponding to a longer diffusion distance of Li+, and resulting in poor kinetics and low utilization of electrode, especially at high current densities. After Li<sup>+</sup> insertion/extraction cycles, the small cavity in S-NFO may disappear due to the nanosize effect of conversion mechanism and the inevitable volume expansion. Fortunately, the initial small hollow in S-NFO can provide extra room for the inward volume expansion, resulting in a hierarchical cube with well dispersed subunits, which is helpful to stabilize the electrode structure. On the contrary, the lager hollow in L-NFO make the subunit near inner shell lack of confine from the interaction with each other, resulting in a continuous expanding during cycling and finally leading to a structure collapse (Cao et al., 2016). As for the P-NFO, similar volume expansion also occurs, but the filled internal in P-NFO cannot buffer such inward expansion, which causes also structural collapse after many repeated cycles.

Further characterizations about the cycle stability of S-NFO at high current densities are carried out (**Figures 5a**,**b**). At 5 A g −1 , the S-NFO shows high capacity of 1,052 mAh g−<sup>1</sup> after 300 cycles, with the CE stabilizes at almost 100 %, suggesting a good structural stability. More impressively, even at ultra-high current density of 20 A g−<sup>1</sup> , a high reversible capacity of 344 mAh g−<sup>1</sup> is still delivered after 2,000 cycles, corresponding to only 0.018 % capacity decay per cycle. The rapid capacity decay during the initial 200 cycles and the continuous capacity growth in the following cycles are normally observed for conversion type anodes, which may be due to the reversible formation of a polymeric gel-like film and activation process (Jing et al., 2014). Structural stability of the S-NFO electrode is also investigated. As demonstrated in **Figures 5c,d**, the nanocube morphology is still well embedded in the electrode after 2,000 cycles and

Wang et al. Lithium-Ion Battery

no obvious agglomeration or pulverization is observed after such long cycles. The corresponding post-cycled CV plots are presented in **Figure S6**, consistent with **Figure 3D**, implying the high electrochemical reversibility of S-NFO electrode. Therefore, it is reasonable to deduce that the small particle size and interior cavity of S-NFO can provide more room to buffer the volume changes during cycling, which benefits the structural stability and thus leads to better cycling performance.

Effect of particle size on electrochemical kinetics of electrode is investigated by Nyquist plots (**Figure 6A**). The depressed semicircle in high frequency is related to charge transfer resistance of electrodes (Rct), and the inclined straight line in low frequency is associated with the mass transfer property. The Rct value in S-NFO and L-NFO is calculated based on an equivalent circuit as shown in **Figure S7,** whereas R<sup>s</sup> , CPE, and W<sup>o</sup> corresponds to the resistance of electrolyte, the constant phase elements, and the Warburg impedance, respectively. Clearly, the Rct value of S-NFO electrode (192 ) is smaller than that of L-NFO (295 ), indicating faster kinetics of charge transfer. Therefore, small particle size can promote faster ion/electron transportation is confirmed, which is beneficial for good utilization of electrode at high currents. **Figure 6B** compares the Nyquist plot of the S-NFO after different cycles. The decreased Rct in the initial ten cycles may be ascribed to the formation of a conductive intermediate (Hu et al., 2015). No obvious change is found between the 10th and 2000th cycles **(Figure S8)**, indicating the well preservation of kinetic superiority after the long-term cycles. The larger Rct of L-NFO is in consistent with its structure, which features longer diffusion distance, as illustrated in **Figure 4**. Moreover, there is a profound difference among the phase angel of S-NFO, and L-NFO. Generally, the more vertical the phase angle is, the more capacitive behavior there is. Therefore, it is reasonable to speculate the existence of capacitive behavior in the S-NFO electrode.

The CV curves of S-NFO electrode at different sweep rates (0.5 ∼ 1.2 mV s−<sup>1</sup> , **Figure 6C**) are used to determine its capacitive behavior. As shown in **Figure 6D**, about 86% faction of the total charge roots from the capacitive process according to the calculation method previously reported (Wang et al., 2016). Such high percentage of capacitance behavior contributes to the excellent rate performance and cycle stability of S-NFO. The extraordinary electrochemical performance of this S-NFO demonstrates its huge potential for fast Li ion storage.

To verify the practical application of this S-NFO, full-cells are assembled by using the S-NFO as anode and commercial LiCoO<sup>2</sup> as cathode, as demonstrated in **Figure 7A**. An LED array with JSNU logo that is consisted of 33 red LEDs in parallel was powered by an S-NFO based full-cell. As shown in **Figure 7B**, the whole LED array could be easily lighted up, demonstrating the viability and practical applications of this S-NFO electrode material. Charge-discharge curves of the full-cells at 0.1 A g−<sup>1</sup> in a voltage range of 1.0–3.9 V were presented in **Figure 7C**. In the first cycle, a charge/discharge capacity of 961/737 mAh g−<sup>1</sup> is delivered, corresponding to an ICE value of 76.7%. About 97.2% capacity was retained in the second cycle. Rate performance of the full-cells was tested at various current densities, as shown in **Figure 7D**. When increasing the current density from 0.1 to 3.0 A g −1 , a high reversible capacity of 649, 572, and 500 mAh g−<sup>1</sup> was delivered at 0.5, 1.0 and 3.0 A g−<sup>1</sup> , respectively. As the current density drops back to 0.1 A g−<sup>1</sup> , 98.1% capacity was recovered, implying the high reversibility of full-cells. Cycling performance

of the full cells was carried out at 0.5 A g−<sup>1</sup> after activating at 0.1 A g−<sup>1</sup> for three cycles (**Figure S9**), which shows a reversible capacity of 300 mAh g−<sup>1</sup> at the end of 50 cycles.

### CONCLUSION

In summary, by combining the advantages of hetero-structure and nano-architecture, the electrochemical performance of NiFe2O4/NiO hybrid material is significantly improved. A series of NiO/NiFe2O<sup>4</sup> hetero-structures with different porosity, particle size, and shell/internal structures has been synthesized and investigated as anode materials for ultra-fast LIBs. Experimental results highlights the interactions between structural geometry and electrochemical behavior, demonstrating that hetero-structure phase, small particles size, and high porosity offer obvious advantages in improving the Li<sup>+</sup> storage property. By optimizing these factors, faster electrochemical kinetics and long-term cycling stability have been achieved in the S-NFO electrode. Impressive rate performance and cycle stability at high current densities are observed in the S-NFO sample, which retains 1,052 and 344 mAh g−<sup>1</sup> after 300 and 2,000 cycles at 5.0 and 20 A g−<sup>1</sup> , respectively. The post-cycled SEM images demonstrate the well-reserved structural durability of this unique electrode. The results here may shed light on

### REFERENCES


further design and fabrication of other high performance micronanostructured materials for energy storage and conversion technologies.

### AUTHOR CONTRIBUTIONS

YingW conducted the experiments and helped writing the manuscript. SW helped operating experiments and data analysis. YijingW, CW, and XH supervised this research work. All authors read and approved the final manuscript.

### FUNDING

This work was supported by the National Natural Science Foundation of China (No. 21805117), Jiangsu Province Science Foundation for Youths (BK20181014), and The Natural Science Foundation of the Jiangsu Higher Education Institutions of China (No. 18KJB150015).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2018.00654/full#supplementary-material


frameworks for superior lithium ion battery anodes. J. Mater. Chem. A 2, 8048–8053. doi: 10.1039/C4TA00200H


high-performance anode for lithium ion batteries. Chem. Eng. J. 347, 563–573. doi: 10.1016/j.cej.2018.04.119


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Wang, Wu, Wang, Wang and Han. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Improved Electrochemical Performance of Surface Coated LiNi0.80Co0.15Al0.05O<sup>2</sup> With Polypyrrole

Zhaoyong Chen<sup>1</sup> \*, Kaifeng Cao<sup>1</sup> , Huali Zhu2,3, Xiaolong Gong<sup>1</sup> , Qiming Liu<sup>1</sup> , Junfei Duan<sup>1</sup> and Lingjun Li <sup>1</sup>

*<sup>1</sup> College of Materials Science and Engineering, Changsha University of Science and Technology, Changsha, China, <sup>2</sup> College of Physics and Electronic Science, Changsha University of Science and Technology, Changsha, China, <sup>3</sup> Department of Chemistry, University of New Hampshire, Durham, NH, United States*

### Edited by:

*Qiaobao Zhang, Xiamen University, China*

#### Reviewed by:

*Xianwen Wu, Jishou University, China Shan Ji, Jiaxing University, China Huanan Duan, Shanghai Jiao Tong University, China Manickam Minakshi, Murdoch University, Australia*

#### \*Correspondence:

*Zhaoyong Chen chenzhaoyongcioc@126.com*

#### Specialty section:

*This article was submitted to Physical Chemistry and Chemical Physics, a section of the journal Frontiers in Chemistry*

Received: *31 August 2018* Accepted: *12 December 2018* Published: *09 January 2019*

#### Citation:

*Chen Z, Cao K, Zhu H, Gong X, Liu Q, Duan J and Li L (2019) Improved Electrochemical Performance of Surface Coated LiNi0.80Co0.15Al0.05O2 With Polypyrrole. Front. Chem. 6:648. doi: 10.3389/fchem.2018.00648* Nickel-rich ternary layered oxide (LiNi0.80Co0.15Al0.05O2, LNCA) cathodes are favored in many fields such as electric vehicles due to its high specific capacity, low cost, and stable structure. However, LNCA cathode material still has the disadvantages of low initial coulombic efficiency, rate capability and poor cycle performance, which greatly restricts its commercial application. To overcome this barrier, a polypyrrole (PPy) layer with high electrical conductivity is designed to coat on the surface of LNCA cathode material. PPy coating layer on the surface of LNCA successfully is realized by means of liquid-phase chemical oxidation polymerization method, and which has been verified by the scanning electron microscopy (SEM), transmission electron microscope (TEM) and fourier transform infrared spectroscopy (FTIR). PPy-coated LNCA (PL-2) exhibits satisfactory electrochemical performances including high reversible capacity and excellent rate capability. Furthermore, the capability is superior to pristine LNCA. So, it provides a new structure of conductive polymer modified cathode materials with good property through a mild modification method.

Keywords: nickel-rich layered oxide, polypyrrole coating, cathode materials, conductive polymer, lithium-ion batteries

### INTRODUCTION

Rechargeable lithium ion batteries (LIBs) (Zheng et al., 2018; Zhang et al., 2018a,b) have been applied successfully to electric vehicles (EVs) (Xu et al., 2015) and hybrid electric vehicles (HEVs) (Goodenough and Park, 2013; Choi and Aurbach, 2016; Chen et al., 2017), which brings a great convenience and reduces exhaust emissions. However, the unsatisfied performance of cathode materials in LIBs, limits its extensive application in EVs and HEVs in the future, including low energy density, poor cycling performance et al. (Thackeray et al., 2012; Goodenough and Kim, 2014) Therefore, it is expected that the cycling performance and rate capability would be greatly improved.

LiNi0.80Co0.15Al0.05O<sup>2</sup> (LNCA) produced by substituting Co and Al for Ni in LiNiO<sup>2</sup> has been considered as a promising cathode material because of it has higher capacity and lower price than LiCoO<sup>2</sup> (Delmas et al., 1993; Levi et al., 1999; Cao et al., 2004). Nonetheless, the LNCA undergoes drastic capacity fading and poor cyclability due to the increase in interfacial resistance on the surface of cathode (Amine et al., 2001; Itou and Ukyo, 2005),

**35**

micro-cracks generated in the particles (Watanabe et al., 2014a,b), and formation of NiO-like rock-salt domains (Abraham et al., 2002, 2003; Hayashi et al., 2014). Currently, an effective method to enhance the electrochemical properties of LNCA is to modify its surface with some stable coating layer like oxides(Cho and Cho, 2010; Cho et al., 2013; Lai et al., 2016; Dai et al., 2016a,b; Yan et al., 2017), phosphates(Kim and Cho, 2007; Lee et al., 2011; Huang et al., 2014), fluoride (Kim et al., 2008; Lee et al., 2013; Zhang et al., 2014; Liu et al., 2016), and other cathode material (Liu et al., 2012; Du et al., 2013; Lu et al., 2014; Zhao et al., 2017), etc. These coating layers can inhibit effectively side reactions between electrode and electrolyte. However, the coating materials impede the diffusion of lithium ions and electronic transmission in a degree.

Recently, conducting polymers have been investigated as additives to improve the electrode performance (Chen et al., 2013; Wu et al., 2013; Ramkumar and Minakshi, 2015, 2016a; Ramkumar and Sundaram, 2016b) because the polymer can provide a high electrical conductivity for the active materials. Nonetheless, polypyrrole (PPy) modified LNCA cathode material via liquid-phase chemical oxidation polymerization method has been not reported yet. In order to improve the conductivity and initial coulombic efficiency of LNCA, we have demonstrated that an ultrathin ppy layer can enhance the electrical conductivity of LNCA via a liquid-phase chemical oxidation polymerization method using ammonium persulphate as oxidant, the ppy layer is highly conductive and contributes to the fast electron transportation on the surface of LNCA particles. So, the ppy coated LNCA reaches a high electrochemical capacity of 206.6 mAh g−<sup>1</sup> at 0.1 C with coulombic efficiency of 91% and the optimized sample maintains 195.1 mAh g−<sup>1</sup> with 88.9% capacity retention after 100 cycles at a current density of 1 C with the voltage of 2.75–4.3 V vs. Li/Li<sup>+</sup> at room temperature, because the PPy reduces the interfacial impedance of the cathode material and provides high electrochemical reaction kinetics, leading to a superior structural stability during lithiation and delithiation cycling.

## EXPERIMENTAL

### Material Synthesis

LiNi0.80Co0.15Al0.05O<sup>2</sup> (LNCA) powder was prepared by mixing Ni0.80Co0.15Al0.05(OH)<sup>2</sup> and LiOH·H2O (99%, Sinopharm Chemical Reagent) at molar ratio of 1:1.05 and firing at 450◦C for 5 h, subsequently calcining at 750◦C for 12 h in O<sup>2</sup> flow. After sieving, powders with an average particle size of 10µm were used for further experiments.

PPy-coated LiNi0.80Co0.15Al0.05O<sup>2</sup> was prepared by wet chemical method in **Figure 1** Desired amounts of pyrrole monomer (AR, 99%) and sodium dodecyl benzene sulfonate (SDBS) as doping agent and 2 g LiNi0.80Co0.15Al0.05O<sup>2</sup> powder were ultrasonically dispersed in 100 ml deionized water. The aqueous solution of ammonium persulfate (AR, 99%) was slowly added as an oxidizing agent, and the concentration of ammonium persulfate was 0.06 M. The molar ratio of pyrrole monomer to oxidant and sodium dodecyl benzene sulfonate was 1:1, 1:1.5, respectively. The mixture was kept in ice-bath for 12 h under magnetically stirring and polymerized on the surface of the LNCA. The PPy coated samples were filtered, washed with deionized water and ethanol for several times, and finally dried under vacuum at 80◦C for overnight. Finally, the PPy coated samples (PL-1, PL-2, and PL-5 corresponding to 1 wt, 2 wt, and 5 wt% PPy) were obtained.

### Material Characterization

coating of the ultrathin PPy layer.

Crystal structures of the prepared LiNi0.80Co0.15Al0.05O<sup>2</sup> (LNCA) and PPy coated LNCA were characterized by X-ray powder diffraction (XRD) with Cu Kα radiation at 40 kV and 40 mA in the 2θ range from 10◦ to 90◦ . The morphology was characterized by field emission scanning electron microscopy (FESEM, FEI QUANTA 250). Transmission electron microscopy (TEM) and high-resolution transmission electron microscopy (HRTEM) were carried out on a JEOL JEM-2100 instrument.

### Electrode Preparation and Electrochemical Measurements

For electrochemical tests, the cathode materials were tested using a coin type cell (2025). The working electrode was prepared by pressing a powder mixture of the sample (PPy-NCA or pristine LNCA), Super P conductive carbon black and polyvinylidene difluoride (PVDF) in N-methyl-2-pyrrolidone (NMP) with a weight ratio of 80:10:10 and coated onto an aluminum foil and dried at 120◦C in vacuum. Two electrode button cells (half cells) were assembled in an argon filled glove box using fresh lithium foil as the counter electrode, The electrolyte was 1 M LiPF<sup>6</sup> dissolved in ethyl carbonate (EC) and dimethyl

carbonate (DMC) (1:1 in volume) and the separator was Cellgard 2400 membrane. The charged / discharged measurements on a LAND battery-testing instrument within 2.75–4.3 V vs. Li/Li<sup>+</sup> at 25◦C. For cycling tests, the cells were cycled at the rate of 1 C (1 C rate is defined as 180 mAh g−<sup>1</sup> ). The rate capability was measured at rates of 0.1 C, 0.2 C, 0.5 C, 1 C, 3 C, and 5 C. A cyclic voltammogram (CV) test was conducted between 2.75 and 4.3 V at a scanning rate of 0.1 mV s−<sup>1</sup> using a CHI 660E electrochemical work station. The electrochemical impedance spectroscopy (EIS) of the cells was conducted using the CHI 660E electrochemical measurement system in a frequency range from 1 mHz to 100 kHz.

The electrodes after many cycles were prepared by disassembling coin cells fully discharged to 2.75 V and rinsed with dimethyl carbonate (DMC) several times in the glovebox, in order to investigate that the microstructure and morphology of cathode materials after long-term cycles.

### RESULTS AND DISCUSSION

### Structural Characterization

**Figure 2a** shows the XRD pattern of pristine LNCA and PL-2 samples (PL-1 and PL-5 is shown **Figure S1**). The diffraction peaks of both samples are well indexed in terms of R-3m structure of hexagonal α-NaFeO2, which match well with those of the standard XRD patterns for (Li0.99 Ni0.01) (Ni0.798 Co0.202) O<sup>2</sup> (PDF#: 87-1562). It should be noted that the pair reflections (006)/(102) and (018)/(110) are well-split for all samples, demonstrating the good hexagonal ordering and layered characteristics (Ohzuku et al., 1993). The intensity ratio of the (003) and (104) lines in the XRD patterns is more than 1.4 for all samples, indicating the suppressed cation mixing at a degree (Yang et al., 2014).

The Fourier transform infrared spectroscopy (FTIR) spectra of pristine LNCA, PPy, and PPy-coated LNCA are shown in **Figure 2b**. It is found that the PPy-coated LNCA has typical absorption peaks of PPy. The band at 1,541 cm−<sup>1</sup> is due to aromatic C=C in PPy. C=N shows peaks around 1,193 cm−<sup>1</sup> . Obviously, the PPy-coated LNCA sample has a stronger absorption peak than pristine LNCA at the position of C=C and C=N spectra. The region of the C–H in-plane deformation vibration situated at 1,043 cm−<sup>1</sup> are observed at the position of PPy-LNCA spectra(Blinova et al., 2007). The results demonstrate that the PPy is successfully coated onto the surface of LNCA particles.

The surface morphology of the pristine LNCA and PL-2 samples are shown in **Figure 2**. Apparently, the spherical morphology of the pristine LNCA is well-retained, and the estimated average particles diameter of the LNCA (**Figure 2c**) is around 8µm. Furthermore, a fuzzy layer was obviously observed in the surface of LNCA in **Figure 2d**, which may be PPy films grown by liquid-phase chemical oxidation polymerization method. And it is in accordance with the above-mentioned FTIR results.

FIGURE 4 | (a) Initial charge/discharge profiles at 0.1 C in the voltage range of 2.75–4.3 V, (b) rate performance, (c,d) initial charge/discharge profiles at various rates of pristine LNCA and PL-2.

The TEM (**Figures 3a,d**) and HRTEM (**Figures 3b,e**) images combined with the corresponding selected-area fast Fourier transform (FFT) pattern (**Figures 3c,f**) further confirm that the spherical morphology of the LNCA is composed of a well crystallized layered structure. As can be seen from **Figures 3a,c**, pristine LNCA particles exhibit a smooth surface and show a highly ordered lattice.

Bright and dark zones of a few nanometers wide are based on the inset of **Figure 3b**. The electron diffraction spots obtained after Fast Fourier Transform (FFT) were calculated and the crystal surface spacing was 0.236 nm, which was deduced to be the inter-planar spacing of the (006) plane of the layered structure. HRTEM images in **Figure 3e** shows that a nanoscale PPy film layer with a thickness of 10∼12 nm is clearly seen on the surface of LNCA particles. It should be noted that the PPy layer is highly conductive, contributing a lot to the improvement of rate capacity of the LNCA cathode material. As depicted in **Figure 3f**, corresponding selectedarea fast Fourier transform (FFT) pattern (in region 4 of **Figure 3e**) further confirm that the bulk of PL-2 still maintain the ordered layered structure. According to the results of SEM, TEM and HRTEM, it is concluded that a PPy nano layer is successfully coated on the surface of LNCA cathode materials.

### Electrochemical Performance

The electrochemical performance of pristine LNCA cathode materials and PL-2 are evaluated and shown in **Figure 4**. The electrochemical performance of PL-1, PL-5 samples are presented in the ESI (**Figures S2–S4**). The initial charge/discharge profiles at 0.1 C rate are shown in **Figure 4a**. The pristine LNCA cathode materials deliver a discharge capacity of 189.5 mAh g−<sup>1</sup> with an initial coulombic efficiency of 85.9%. However, for the PL-2 cathode materials, the discharge capacity approaches 206.6 mAh g−<sup>1</sup> , with an initial coulombic efficiency of 91%. The rate capability combined with corresponding charge/discharge profiles at various rates (**Figures 4c,d**) shows a superiority of PL-2, achieving higher rate capabilities than pristine LNCA. The voltage and discharge capacity of the cells gradually decrease with an increase of discharge current density due to the effect of ohmic polarization and interphase resistances caused by the side reaction on the interface between LNCA cathode and electrolytes. However, the higher discharge capacities of PL-2 are 163.5 mAh g −1 at 5 C, more than 146.2 mAh g−<sup>1</sup> of pristine LNCA in

FIGURE 5 | (a) Cycling performance at 1.0 C rate in the voltage range of 2.75–4.3 V, (b,d) initial discharge profiles evolution during cycling of pristine LNCA and PL-2. (c) Comparison of the electrochemical impedance spectra (inset is a magnified view of the high frequency region and equivalent circuit diagram ) of the pristine and PL-2 charged to 4.3 V vs. Li/Li+.

TABLE 1 | The Li<sup>+</sup> ion migration resistance (Rsei) and the charge transfer resistance (Rct) of different samples.


**Figure 4b**. These test results indicate that PPy coating greatly reduces the electrode resistance of the cathode material and provides high electrochemical reaction kinetics, which ensured less discharge potential drop during high-rate cycling.

**Figure 5** shows the cycling performance and electrochemical impedance spectra of pristine LNCA and PL-2 samples. As observed, the discharge capacities of both samples decrease with the increasing of the cycle number. As can be seen from **Figure 5a**, the capacity retention of the PL-2 sample is 88.9% after the 100 th cycle at 1 C rate between 2.75 and 4.3 V, while the prestine LNCA material exhibits 71.6% capacity retention after 100 cycles. For the PL-2 sample, corresponding to the initial discharge profiles evolution during cycling of both samples in **Figures 5b,d**, the cycle performance is much improved. It is definitely demonstrated that the PL-2 show better structure stability than pristine LNCA. The poor cycling performances of the pristine LNCA can be attributed to its vigorous surface reactivity with electrolyte, where the harmful interfacial side reactions suppress lithium ion diffusion.

Electrochemical impedance spectroscopy (EIS) measurements are used to elucidate the beneficial effect of the PPy coating layer for LNCA material. As can be seen from **Figure 5c**, the depressed semicircle in the high frequency region is related to the surface film impedance of the active cathode material (Wu et al., 2017). A magnified view of the high frequency region curve and equivalent circuit diagram are shown in the inset. The Li<sup>+</sup> ionmigration resistance (Rsei) and the charge transfer resistance (Rct) of different samples is listed in **Table 1**. Obviously, it is easy to conclude that PL-2 material has smaller electrochemical impedance compared to the pristine sample before cycling. The mid frequency region responds to the charge transfer impedance. It indicates that PPy coating reduces the charge transfer resistance of the material due to the high electrical conductivity of the PL-2 sample, which facilitates transfer of electrons from the conductive agent and current collector to the

FIGURE 7 | HRTEM images of: (a) the LNCA material after 100th cycle; (b,c) PL-2 sample after the 100th cycle; The index marked by subscript R and F is related to the rhombohedral (R-3m) phase and cubic (Fd-3m) phase. XRD pattern and SEM images of pristine LNCA and PL-2 cathode material during the charge and discharge cycles. XRD pattern of: (d) the pristine LNCA and PL-2 cathode material after 100th cycle; SEM images of: (e) the LNCA material after 100th cycle; (f) PL-2 sample after the 100th cycle.

cathode material and the electrode materials are transfer to each other.

To investigate the degree of the electrochemical reversibility, the cyclic voltammetry (CV) curves of LNCA and PL-2 between 2.75 and 4.3 V vs. Li/Li<sup>+</sup> at a scan rate of 0.1 mV s−<sup>1</sup> are depicted in **Figure 6**. Obviously, three pairs of reduction and oxidation peaks are observed in all the both samples. Meanwhile, according to the nickel-rich material phase transition process, caused by phase transitions of hexagonal phase to monoclinic phase (H1 to M), monoclinic phase to hexagonal phase (M to H2) and hexagonal phase to hexagonal phase (H2 to H3) during the charge-discharge process(Li et al., 1993; Xie et al., 2015). The first cycle are significantly different from those for the following cycles, because of the irreversible electrochemical reaction during the first discharge cycle. However, no obvious alteration is observed between the second cycle and the third cycle, suggesting the good reversibility of lithium insertion and extraction reactions. As can be seen from the **Figures 6a,b**, the peak position of PL-2 material shifted is smaller than LNCA. This illustrates that the PPy materials possess lower activation energy and higher electrical conductivity for the active materials. Therefore, PPy coated layer is beneficial to reducing the electrochemical polarization as well as improving the electrochemical performance.

In order to study the structural evolution of bare LNCA and PL-2 samples upon cycling, HRTEM and FFT imaging is performed. **Figure 7a** present the HRTEM images of the bare LNCA sample after the 100 th cycle, while **Figures 7b,c** present the HRTEM images of the PL-2 sample after the 100 th cycle, As can be seen from **Figure 7a**, the distortion region is observed over distances of ca. 10 nm from the surface of the LNCA material after 100 th cycle. The FFT in the inset of **Figure 7a** exhibits a spinel-like phase (space group Fd-3m). The formation of spinellike phase in the surface region of the cycled LNCA sample is attributed to the migration of TM ions to the Li sites without breaking down the lattice.

**Figures 7b,c** show the HRTEM and corresponding FFT results of the PL-2 sample. We have found that the surface region of the cycled PL-2 sample still remains in the rhombohedral phase. This is also consistent with the cycling performances and electrochemical reversibility (**Figures 5a, 6**). This observation likely indicates that the phase transition from layered to spinel can be suppressed by this PPy coated layer.

Furthermore, the XRD pattern and SEM images of pristine LNCA and PL-2 cathode material during the charge and discharge cycles are presented in the **Figures 7d–f**. According to the XRD pattern, the peaks of the both samples are basically similar. It is demonstrated that the cycled PL-2 sample still remains as the rhombohedral phase. According to SEM images of pristine LNCA and PL-2 cathode material after the charge and discharge cycles, the morphology of the pristine LNCA particles has obvious collapse, resulting in decrease of capacity in process of charge and discharge. However, the PL-2 cathode material still remains the spherical morphology after cycles. This likely indicates that the PPy maintains the structural stability of cathode material during the cycles.

### CONCLUSION

In summary, we have successfully synthesized a PPy-coated LNCA cathode material via Liquid-phase chemical oxidation

### REFERENCES

Abraham, D. P., Twesten, R. D., Balasubramanian, M., Kropf, J., Fischer, D., Mcbreen, J., et al. (2003). Microscopy and spectroscopy of lithium nickel oxide-based particles used in high power lithiumion cells. J.Electrochem. Soc. 150, A1450–A1456. doi: 10.1149/1.16 13291

polymerization method. According to the results of XRD, HRTEM, and FTIR, PPy was uniformly dispersed on the surface of the LNCA cathode material without changing its structure. Furthermore, the PPy coated LNCA cathode material shows excellent rate capability and capacity retention at 1, 3, and 5 C rates, indicating a good kinetics characteristic and the effectiveness of surface coating for protecting cathode materials from the side reaction between electrode and electrolyte. Meanwhile, the PPy-coated LNCA showed an improved coulombic efficiency compared with the pristine one. Therefore, as a highly Li-ions and electrons conductive material, the PPy coating shows great potential in surface modification and is expected to improve performance for other types cathode materials of LIBs.

## AUTHOR CONTRIBUTIONS

ZC and KC conceived and designed the study; KC and ZC prepared all materials. KC and HZ conducted SEM experiments. XG and QL conducted XRD, FTIR experiments. ZC and KC analyzed the data. KC wrote the manuscript and LL, JD commented on it. ZC supervised the implementation of the project. All authors read and approved the final manuscript.

### ACKNOWLEDGMENTS

We thank the financial support from Project funded by the National Natural Science Foundation of China (Grant No. 51874048), the National Science Foundation for Young Scientists of China (Grant No. 51604042 and Grant No. 21601020). This project was supported by the Research Foundation of Education Bureau of Hunan Province (Grant No. 16A001), and Science and Technology Plan Changsha (Grant No. kq1701076).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2018.00648/full#supplementary-material

XRD pattern and electrochemical performance of samples in different PPy amounts (1wt%, 2wt%, 5wt%). XRD patterns of the bare LNCA, PPy coated LNCA material: PL-1, PL-2, and PL-5 samples. The initial charge/discharge profiles at 0.1 C rate of pristine LNCA cathode materials and PPy coated LNCA material: PL-1, PL-2, and PL-5 samples. The rate performance profiles and cycling performance profiles at 1 C rate of pristine LNCA cathode materials and PPy coated LNCA material: PL-1, PL-2, and PL-5 samples.


batteries. J. Power Sources 97–98, 684–687. doi: 10.1016/S0378-7753(01)0 0701-7


cycling performance at 55 ◦C. J. Power Sources 196, 7742–7746. doi: 10.1016/j.jpowsour.2011.04.007


carbonate treatment. Appl. Surf. Sci. 403, 426–434. doi: 10.1016/j.apsusc.2017. 01.089

Zheng, Z., Zao, Y., Zhang, Q., Cheng, Y., Chen, H., Zhang, K., et al. (2018). Robust Erythrocyte-like Fe2O3@carbon with Yolkshell Structures as High-performance anode for lithium ion batteries. Chem. Eng. J. 347, 563–573. doi: 10.1016/j.cej.2018. 04.119

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Chen, Cao, Zhu, Gong, Liu, Duan and Li. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# High Performance and Structural Stability of K and Cl Co-Doped LiNi0.5Co0.2Mn0.3O<sup>2</sup> Cathode Materials in 4.6 Voltage

#### Zhaoyong Chen<sup>1</sup> \*, Xiaolong Gong<sup>1</sup> , Huali Zhu2,3, Kaifeng Cao<sup>1</sup> , Qiming Liu<sup>1</sup> , Jun Liu<sup>1</sup> , Lingjun Li <sup>1</sup> and Junfei Duan<sup>1</sup>

#### Edited by:

*Qiaobao Zhang, Xiamen University, China*

### Reviewed by:

*Xunhui Xiong, South China University of Technology, China Chaopeng Fu, Shanghai Jiao Tong University, China Yunjian Liu, Jiangsu University, China Qiulong Wei, University of California, Los Angeles, United States*

\*Correspondence:

*Zhaoyong Chen chenzhaoyongcioc@126.com*

#### Specialty section:

*This article was submitted to Physical Chemistry and Chemical Physics, a section of the journal Frontiers in Chemistry*

Received: *27 September 2018* Accepted: *11 December 2018* Published: *08 January 2019*

#### Citation:

*Chen Z, Gong X, Zhu H, Cao K, Liu Q, Liu J, Li L and Duan J (2019) High Performance and Structural Stability of K and Cl Co-Doped LiNi0.5Co0.2Mn0.3O2 Cathode Materials in 4.6 Voltage. Front. Chem. 6:643. doi: 10.3389/fchem.2018.00643* *<sup>1</sup> College of Materials Science and Engineering, Changsha University of Science and Technology, Changsha, China, <sup>2</sup> College of Physics and Electronic Science, Changsha University of Science and Technology, Changsha, China, <sup>3</sup> Department of Chemistry, University of New Hampshire, Durham, NH, United States*

The high energy density lithium ion batteries are being pursued because of their extensive application in electric vehicles with a large mileage and storage energy station with a long life. So, increasing the charge voltage becomes a strategy to improve the energy density. But it brings some harmful to the structural stability. In order to find the equilibrium between capacity and structure stability, the K and Cl co-doped LiNi0.5Co0.2Mn0.3O<sup>2</sup> (NCM) cathode materials are designed based on defect theory, and prepared by solid state reaction. The structure is investigated by means of X-ray diffraction (XRD), rietveld refinements, scanning electron microscope (SEM), XPS, EDS mapping and transmission electron microscope (TEM). Electrochemical properties are measured through electrochemical impedance spectroscopy (EIS), cyclic voltammogram curves (CV), charge/discharge tests. The results of XRD, EDS mapping, and XPS show that K and Cl are successfully incorporated into the lattice of NCM cathode materials. Rietveld refinements along with TEM analysis manifest K and Cl co-doping can effectively reduce cation mixing and make the layered structure more complete. After 100 cycles at 1 C, the K and Cl co-doped NCM retains a more integrated layered structure compared to the pristine NCM. It indicates the co-doping can effectively strengthen the layer structure and suppress the phase transition to some degree during repeated charge and discharge process. Through CV curves, it can be found that K and Cl co-doping can weaken the electrode polarization and improve the electrochemical performance. Electrochemical tests show that the discharge capacity of Li0.99K0.01(Ni0.5Co0.3Mn0.2)O1.99Cl0.01 (KCl-NCM) are far higher than NCM at 5 C, and capacity retention reaches 78.1% after 100 cycles at 1 C. EIS measurement indicates that doping K and Cl contributes to the better lithium ion diffusion and the lower charge transfer resistance.

Keywords: lithium ion batteries, LiNi0.5Co0.2Mn0.3O2 , co-doping, cation mixing, phase transition

## INTRODUCTION

Nowadays, the vigorous development of lithium-ion batteries (LiBs) (Chen et al., 2017; Zhang et al., 2018a) has accelerated the production of energy storage devices (Zhang et al., 2018b; Zheng et al., 2018), electric vehicles (EVs), and hybrid electric vehicles (HEVs) (Terada et al., 2001; Goodenough and Park, 2013; Xiong et al., 2013, 2014b; Xu et al., 2015b; Choi and Aurbach, 2016; Liu et al., 2018b; Su et al., 2018). However, unsuitable performance limits the application of LiBs cathode materials, such as low energy density of LiCO<sup>2</sup> and LiFeO4, and lithiumrich layered oxide (LRLO) cathode materials with low coulombic efficiency and voltage attenuation. Under these circumstances, researchers turn their attention to cathode materials with high energy density and low prices, therefore, lithium transition metal oxides (LiNixCoyMn1−x−yO2) due to its high capacity, low price (Chen et al., 2003; Shin et al., 2005; Li et al., 2009; Martha et al., 2009; Sun et al., 2009; Kim, 2013; Yue et al., 2013b; Xiong et al., 2014a) and its properties adjusted by the relative ratio of different TM ions (Kim et al., 2016) according to the requirement are diffusely researched, in particular, LiNi0.5Co0.2Mn0.3O<sup>2</sup> (NCM) cathode materials has been attracting much more attentions.

It is all well-known that the Ni element plays a vital role in providing capacity for NCM. Unfortunately, the presence of Ni element also causes Ni to escape from the 3b sites into the 3a sites of the lithium layer during the preparation and charging because the radius of Ni2<sup>+</sup> and Li<sup>+</sup> is similar. And these defects are intensified during high-voltage cycling because of the increasing number of Li vacant sites. This Ni migration trigger cation mixing and phase transformation from layered (R-3m) to spinel (Fd-3m) and rock salt (Fm-3m) phase at some micro areas (Kojima et al., 2011; Boulineau et al., 2013; Jung et al., 2014; Lin et al., 2014), which results in structural degradation, poor cycle stability and slow lithium ion diffusion coefficient of NCM cathode material.

In the past few decades, extensive studies have been confirmed that ion substitution such as Na<sup>+</sup> (Chen et al., 2013; Hua et al., 2014), Mg2<sup>+</sup> (Luo et al., 2016), Fe3<sup>+</sup> (Liu et al., 2006), Ti4<sup>+</sup> (Seungtaek et al., 2005), V5<sup>+</sup> (Zhu et al., 2014), F<sup>−</sup> (Shin et al., 2006; Yue et al., 2013a) and so on is considered as an efficacious strategy to decrease the cation mixing degree, ameliorate the microstructure in stability and improve rate performance. Among them, Na<sup>+</sup> substitution is regard as a typically dopant to ameliorate the performance of NCM. Li1.1−xNaxNi0.2Co0.3Mn0.4O<sup>2</sup> (Park et al., 2006) are prepared by sol-gel method with better rate performance, and lower cation mixing are exhibited when x was 0.05 and 0.1. But he cycle stability and structural stability of the material have not been apparently improved. In addition, many researchers further improve the stability of the material during cycling and enhance the electrochemical performance of the material by anionic doping. For instance, G-H. Kim et al. (Kim et al., 2005) synthesized LiNi1/3Co1/3Mn1/3O2−zF<sup>z</sup> by partially replacing O with F, and improves structural stability of materials. However, it did not solve the cation mixing and improve the rate performance.

As far as we know most of these attainable studies are limited to a single replacement and do not synchronously improve the cycle stability, lithium ion diffusion coefficient and cation mixing. Therefore, in this study, aiming to improving the structure stability and rate performance under 4.6 V, we designed K and Cl co-doped Li0.99K0.01Ni0.5Co0.2Mn0.3O1.99Cl0.01 (KCl-NCM) cathode material and prepared it using solid-state reaction. Because of the tangible that the radius of K<sup>+</sup> (r<sup>+</sup> <sup>K</sup> = 1.33 Å) is much larger than that of Li<sup>+</sup> (r<sup>+</sup> Li = 0.76 Å), we partially replace Li with K into the structure of NCM to reduce the mixing of the cations and improve the lithium ion diffusion coefficient. Simultaneously, we also partially replace O with Cl into the crystal structure because of the covalent radii and the electronegativity of Cl much than O (Singh et al., 2017), moreover, Cl doping is associated with the reinforcement of MnO<sup>6</sup> octahedral in the framework by the strong ionic Mn-Cl, Ni-Cl, and Co-Cl bonds (Kim et al., 2014), which makes the structure more stable and improves cyclic performance. Through the co-doping, cycle performance and rate performance of NCM are markedly improved. Moreover, the content of Ni occupies Li sites (2.77%) for the KCl-NCM is lower than NCM (3.3%) identified by Rietveld refinements, which effectively reduces the cation mixing.

### EXPERIMENTAL

### Preparation of the Samples

Li0.99K0.01Ni0.5Co0.3Mn0.2O1.99Cl0.01 (KCl-NCM) layered cathode materials were prepared via solid-state reaction using stoichiometric of KCl, commercial transition-metal hydroxide precursors Ni0.5Co0.2Mn0.3(OH)<sup>2</sup> and LiOH·H2O as raw materials, wherein the ratio of Li to the transition metal is 1:1, K and Cl were added to the mass fraction of 1%. The raw materials were mixed at an agate mortar, and grind time was 1 h to make it fully mixed, then which was heated at 480◦C for 2 h and calcined at 880◦C for 12 h at a heating rate of 5◦C min−<sup>1</sup> in air. Finally, the sample was cooled slowly in the furnace to room temperature. Meanwhile, synthesis conditions of LiNi0.5Co0.3Mn0.2O<sup>2</sup> are consistent with KCl-NCM except that a certain stoichiometric ratio of KCl is added, which is regard as reference sample.

### Materials Characterization

X-ray diffraction (XRD, Rigaku D/Max 200PC, Japan) analysis was carried out on a Rigaku/Max-RAX powder diffractometer with Cu Kα-radiation. The scanning speed is 5◦ min−<sup>1</sup> and scanning range is 10◦ < 2θ < 90◦ . The morphologies and microstructures of all samples were determined by scanning electron microscopy (SEM, Nova NanoSEM-230), and energy dispersive X-ray spectroscopy (EDS) is carried out on OXFORD7426 as the attachment of SEM, with the acceleration voltage of 20 kV. Transmission electron micrographs (TEM) were recorded by a JEOL JEM-2010 transmission electron microscope.

### Electrochemical Measurement

The positive electrode (about 4.30 mg cm−<sup>2</sup> ) consists of 80 wt.% as-prepared composites, 15 wt.% acetylene black and 5

FIGURE 1 | (A) XRD patterns of NCM and KCl-NCM samples, (B) and (C) are partial enlarged views of (A); Refinements patterns for samples: (D) NCM (E) KCl-NCM; well-ordered R-3m structure of samples: (F) NCM (G) KCl-NCM; partially cation mixed phase with TM ions in Li slab at highly charged state: (H) NCM (I) KCl-NCM.

wt.% polyvinylidene fluoride (PVDF) as a binder, and metal Al foil is used as collector. Celgard 2,400 is used as separator which is soaked in 1.0 mol L−<sup>1</sup> LiPF6/EC+DMC (EC:DMC = 1:1 in volume ratio) electrolyte. Lithium metal foil is used as the counter electrode during electrochemical measurements. All the cells are assembled in an argon-filled glove box. The charge/discharge test is carried out by using a Land BT2001A automatic battery test system in the voltage range of 2.7∼4.6 V, and the density of current is measured by 1 C (1 C means 150 mAh g−<sup>1</sup> ). The electrochemical impedance is measured in the frequency range from 10−<sup>3</sup> to 10<sup>5</sup> Hz on a CHI660B electrochemical working station (Chenhua, Shanghai, China), and the perturbation amplitude is controlled at ±5 mV.

### RESULTS AND DISCUSSION

### Structural Characterization

**Figure 1A** displays the XRD patterns of NCM and KCl-NCM. From XRD patterns, we can observe that all the samples are indexed to a R-3m structure of hexagonal, and no other impurities is detected. From the **Figures 1B,C**, we can clearly observe that the peaks of (006)/(102) and (108)/(110) are separated, indicating that the material have a good layered phase structure (Lee et al., 2013; Zhu et al., 2014; Xu et al., 2015a). The lattice constants c/a and R(I003/I104) of all samples are shown in **Table 1**. When K and Cl are co-doped into the NCM crystals, the lattice constants increase obviously, indicating that K and Cl are successfully incorporated into the crystal lattice. It was reported that the R value of the samples is >1.2, and also increases after doping, which indicates the cation mixing is reduced to a certain degree. It will be beneficial to the improvement of the electrochemical properties of the material.

TABLE 1 | Lattice constants of NCM and KCl-NCM samples.


TABLE 2 | The results of Rietveld refinements for NCM and KCl-NCM samples.


To further explain the role of K substitution for Li in the Li layers, rietveld refinements is used to further analyze the XRD pattern of the samples (Li et al., 2012b). It is assumed that Li, TM, and O occupy the 3a, 3b, and 6c sites, respectively (Chen et al., 2013). In this work, we assume that K completely occupies the Li site, which leads to the highest reliability factors. And the pictures of Rietveld refinements are shown in **Figures 1D,E**. **Table 2** is occupancies of atoms for all samples. Obviously, it can be seen that the Ni/Li mixing degree is decreased prominently by K substitution. Furthermore, compared with NCM (3.3%), the Ni content in the Li layer (2.77%) of KCl-NCM is lower. The result

100 cycles.

can be attributed to the incorporation of K<sup>+</sup> into the Li layer, which would generate a big driving force to separate Li<sup>+</sup> ions from the transition metal layer and thus avoid the Li/Ni disorder of the KCl-NCM. Hence, the substitution of Li<sup>+</sup> by K<sup>+</sup> leads to a more ordered layered structure, a larger Li layer distance, and a lower cation mixing degree in KCl-NCM. In order to make the results of Rietveld refinements and XRD more specific, we simulate the cation disorder with R-3m structure for NCM and KCl-NCM. **Figures 1F,H** present a perfect R-3m structure of Li-oxygen-TM-oxygen-Li, which clearly separates TM sites (3b) and lithium sites (3a). But Ni ions are easy to enter into the Li layers because the similar to the ionic radius of Ni2<sup>+</sup> and Li<sup>+</sup> during the highly charged state, as shown **Figure 1G**. **Figure 1I** shows TM ions in Li slab at highly charged state for KCl-NCM, since the K<sup>+</sup> radius is much larger than the radius of Ni2+, which reduces the number of Ni2<sup>+</sup> migration to the Li site. As a result, K<sup>+</sup> doping can bring down the cation mixing to some extent, and it is also consistent with the results of the Rietveld refinements.

The SEM images of NCM, KCl-NCM and the corresponding EDS mappings are illustrated in **Figure 2**. A uniform nearspherical microstructure of about 5 microns can be observed, which are agglomerated by uniform size of a particle. The corresponding EDS mappings of KCl-NCM display all elements including K and Cl are uniformly distributed, which reveals K and Cl are successfully incorporated into the NCM.

To further determine the signal of K and Cl, XPS is performed. **Figure 3** shows the XPS patterns of transition metal elements Ni, Co, Mn, K, Cl and O in LiNi0.5Co0.3Mn0.2O<sup>2</sup> samples before and after KCl doping, as shown, the electron binding energies of Ni2p, Co2p, and Mn2p in LiNi0.5Co0.3Mn0.2O<sup>2</sup> samples obtained by doping with KCl have not change significantly, which are 855.3 eV, 780.4 eV and 642.8 eV, respectively, the observed binding energies for Ni 2p3/2, Co 2p3/<sup>2</sup> and Mn2p3/<sup>2</sup> of oxidation state coincide well. The binding kinetics peaks of K and Cl are shown in samples doped with KCl, indicating that the dopant elements are present in the sample.

To provide the detailed information and investigate local structural changes of the samples, high-resolution transmission electron microscopy (HRTEM) and fast fourier transformation (FFT) are conducted on NCM and KCl-NCM. Various regions in the sample are examined to avoid any confusion. **Figures 4a,b** exhibit a good layered structure and no any trace of a secondary phase regardless of near the surface or the inner region before electrochemical testing, which reveals that K and Cl co-doping have not destroy the layered structure of NCM. Moreover, from the insets in **Figures 4a,b**, we can clearly see that the interplanar spacing of the sample doped with K<sup>+</sup> and Cl<sup>−</sup> is larger than NCM sample, indicating that the doping of K<sup>+</sup> enlarges the spacing of Li layers, which is consistent with the result that the c value of the KCl-NCM sample is larger than the c value of the NCM sample in the XRD. As a result, it will also contribute to improve the rate performance. However, the local structure has change dramatically after cycling 100 times at 4.6 V for NCM (**Figure 4c**). The additional crystal planes can be indexed as (400)<sup>S</sup> and (531)<sup>S</sup> in **Figure 4c** compared with **Figure 4a**, corresponding to a spinel structure. It indicates that NCM undergoes a transition from hexagonal phase to spinel phase in cyclic testing. In general, Ni ions occupying Li sites will lead to Li deficiency, and it can give rise to phase transformation at some micro areas. And it triggers the collapse of the layered structure. In contrast, we find that the structure of K and Cl co-coped sample (**Figure 4d**) is distinctly different from that of the NCM sample after 100 cycles at 4.6 V. A welllayered structure is still maintained after high-voltage cycling, corresponding to the (104)<sup>R</sup> of the FFT images. This enhanced structural stability is attributed to the K substitution, which reduces the mixing of Li and Ni, suppressing it from the severe structural degradation induced during charge and discharge process. As a result, this suppression of phase transition intensely ameliorates the deterioration of electrochemical performance of Ni-rich cathode materials during high-voltage cycling (Yang and Xia, 2015).

### Electrochemical Performance

**Figure 5** describes electrochemical performance of NCM and KCl-NCM. **Figure 5A** illustrates atypical initial charge-discharge curve of the NCM. The initial discharge capacity for the NCM and KCl-NCM is 203.9 and 210.3 mAh/g. In contrast, it is obvious that the coulombic efficiency and initial discharge capacity of KCl-NCM sample is superior to those of NCM. The rate capacity of NCM and KCl-NCM is evaluated in **Figures 5B,C**, the discharge capacity of NCM samples drops dramatically with the current density increasing, and the discharge capacities of NCM are from 203.9 mAh g−<sup>1</sup> at 0.1C to 152.74 and 116.0 mAh g−<sup>1</sup> at 3 C and 5 C, which are only 74.9 and 56.9% of the discharge capacity at 0.1 C. However, the discharge capacities of the sample doped with K and Cl at 3 C and 5 C is, respectively, 175 and 162.5 mAh/g, corresponding to 83.7 and 77.7% of its capacity of 209.1 mAh/g at 0.1 C. Apparently, the rate performance of K and Cl substituted sample is remarkably enhanced compare with NCM, which may be due to the fact that K replaces the Li site and increases the diffusion channel of lithium ions because the radius of K<sup>+</sup> (r<sup>+</sup> <sup>K</sup> <sup>=</sup> 1.33 Å) is higher than that of Li<sup>+</sup> (r<sup>+</sup> Li = 0.76 Å), in addition, according to the literature (Singh et al., 2017), the doping of Cl plays a role in the improvement of the rate performance because the radius of Cl is larger than the radius of O. **Figure 5D** demonstrates the cycle performance of two samples at 1 C rate. The remaining discharge capacity for NCM after 100 cycles is 124.8 mAh/g, and the capacity retention is 73.2%. With regard to KCl-NCM, the discharge capacity is 155.54 mAh/g after 100 cycles, and the capacity retention is improved to 83.0%. The cycle performance of sample co-doped with K and Cl is significantly improved. The possible reason is the fact that K substitution reduces the mixing of Li and Ni. On the other hand, Cl substitution can reduce the reactivity of the cathode toward electrolyte oxidation and associate with the reinforcement of MnO<sup>6</sup> octahedral in the framework by the strong ionic Mn-Cl,

TABLE 3 | The values of R<sup>s</sup> + Rct and DLi <sup>+</sup> for NCM and KCl-NCM.


Ni-Cl, and Co-Cl bonds (Kim et al., 2014). Therefore, K and Cl substitution synergistically improved the rate performances and the structure stability during cycling.

To further understand the effect of K and Cl doping on the lithium ion transport of NCM cathode materials, the electrochemical impedance spectroscopy (EIS) and corresponding relationships between Z′ re and ω −1/2 conducted are shown in **Figure 6**. The diffusion coefficient of lithium ion (DLi+) can be calculated via the equation as described in references (Li et al., 2012a; Mai et al., 2013; Zheng et al., 2014, 2019; Choi et al., 2015; Liu et al., 2018a). From the **Figure 6** and **Table 3**, we can see that the impedance of NCM and KCl-NCM samples are 134.8 and 46.4 , and it is clear that the doping K and Cl reduces the electrode resistance of the sample. Compared to the undoped sample, the diffusion coefficients of lithium ions doped with K and Cl increases from 2.62 × 10−<sup>10</sup> to 2.37 × 10−<sup>9</sup> cm<sup>2</sup> s −1 . Generally, the DLi<sup>+</sup> is known as an intrinsic property for a given positive electrode, which depends only on the structure of active material in the charge state. It has been proven that the activity energy for the Li-ion transport in solid could be reduced effectively for the reason of increasing Li layer distance and reducing cation mixing (Hua et al., 2014). So, the doped samples can offer a large amount of lithium ion in the intercalation and deintercalation reaction at large charge and discharge current. Therefore, KCl-NCM have a faster Li diffusion probably due to the larger Li layer spacing and the lower Li/Ni disorder. The decrease of the impedance and the increase of the diffusion coefficient of the lithium ion show that the KCl-NCM

#### TABLE 4 | The results of cyclic voltammogram for NCM and NCM-KCl.


reduce the polarization of the electrode, and improves the cycle performance, which is consistent with the electrochemical test result.

**Figure 7** presents the cyclic voltammogram of two samples. As can be seen from **Figure 7**, these CVs demonstrate quite reversible electrochemical behavior with well resolved oxidation/reduction peaks related to the Liextraction/insertion accompanied with the Ni2+/Ni4<sup>+</sup> and Co3+/Co4<sup>+</sup> oxidation/reduction, respectively. From the **Table 4**, the oxidation peaks for NCM and NCM-KCl of the first cycle centerat 3.8873 V and 3.8598 V, corresponding to the reduction peaks centerat 3.6827 V and 3.7131 V, respectively, it is obviously that the difference value between the oxidation peaks and reduction peaks for the KCl-NCM (0.1461 V) is smaller compared to NCM (0.2046 V), and the same pattern is presented in the second and third cycle. It is well-known that the bigger the potential difference between lithium ions intercalating and deintercalating, the stronger the electrode polarization is. This smaller difference between oxidation and reduction peaks positions indicates the better reversibility of Li<sup>+</sup> ions during intercalating/deintercalating in the KCl-NCM materials, which is consistent with the result of initial charge-discharge curves for the NCM and KCl-NCM. Meanwhile, it ensures reduced capacity fade during cycling. Therefore, K and Cl co-doped can weaken the electrode polarization and improve the electrochemical performance.

### CONCLUSION

In a word, we have researched out an effectual method to improve the structural stability and electrochemical performance

### REFERENCES


of the Ni-rich layered oxide cathode during high-voltage cycling. By XRD and TEM analysis, it is found that the dopant materials have a higher cation ordering degree and complete layered structure. Rietveld refinements prove K and Cl substitutes can effectively reduce cation mixing. Through electrochemical performance analysis, KCl-NCM has a better comprehensive performance compared to NCM. The initial capacity is improved, at the same time the rate performance has also been greatly improved because of reducing the electrode impedance and improving lithium ion diffusion coefficient. Especially, doping K and Cl into the layered structure of NCM could effectually inhibit the phase transition to some degree during high-voltage cycling, leading that layered structure of KCl-NCM remains more complete than NCM after 100 cycles.

### AUTHOR CONTRIBUTIONS

ZC and XG conceived the idea. XG and ZC prepared all materials and wrote the manuscript. HZ, KC, and XG analyzed the data. QL and JL conducted XRD, SEM, and TEM experiments. JD and LL played active roles in providing constructive suggestions.

### ACKNOWLEDGMENTS

We thank the financial support from Project funded by the National Natural Science Foundation of China (Grant No. 51604042, 51874048 and 21601020), the Research Foundation of Education Bureau of Hunan Province (Grant No. 16A001), and Science and Technology PlanChangsha (Grant No. kq17 01076).


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Chen, Gong, Zhu, Cao, Liu, Liu, Li and Duan. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Highly Stable Gully-Network Co3O<sup>4</sup> Nanowire Arrays as Battery-Type Electrode for Outstanding Supercapacitor Performance

Chunli Guo<sup>1</sup> \* † , Minshuai Yin<sup>1</sup> , Chun Wu<sup>2</sup> , Jie Li <sup>1</sup> , Changhui Sun<sup>3</sup> , Chuankun Jia2,4 \* † , Taotao Li <sup>1</sup> , Lifeng Hou<sup>1</sup> and Yinghui Wei 1,5

*<sup>1</sup> College of Materials Science and Engineering, Taiyuan University of Technology, Taiyuan, China, <sup>2</sup> College of Materials Science and Engineering, Changsha University of Science & Technology, Changsha, China, <sup>3</sup> School of Chemistry and Chemical Engineering, Qilu Normal University, Jinan, China, <sup>4</sup> Key Laboratory of Advanced Energy Materials Chemistry (Ministry of Education), Nankai University, Tianjin, China, <sup>5</sup> College of Materials Science and Engineering, Taiyuan University of Science and Technology, Taiyuan, China*

### Edited by:

*Qiaobao Zhang, Xiamen University, China*

#### Reviewed by:

*Shenglin Xiong Xiong, Shandong University, China Yuxin Zhang, Chongqing University, China Junmin Xu, Zhengzhou University, China Wei Luo, Donghua University, China*

#### \*Correspondence:

*Chunli Guo guochunli@tyut.edu.cn Chuankun Jia jack2012ding@gmail.com*

*†These authors have contributed equally to this work*

#### Specialty section:

*This article was submitted to Nanoscience, a section of the journal Frontiers in Chemistry*

Received: *17 September 2018* Accepted: *07 December 2018* Published: *21 December 2018*

### Citation:

*Guo C, Yin M, Wu C, Li J, Sun C, Jia C, Li T, Hou L and Wei Y (2018) Highly Stable Gully-Network Co3O4 Nanowire Arrays as Battery-Type Electrode for Outstanding Supercapacitor Performance. Front. Chem. 6:636. doi: 10.3389/fchem.2018.00636* 3D transition metal oxides, especially constructed from the interconnected nanowires directly grown on conductive current collectors, are considered to be the most promising electrode material candidates for advanced supercapacitors because 3D network could simultaneously enhance the mechanical and electrochemical performance. The work about design, fabrication, and characterization of 3D gully-network Co3O<sup>4</sup> nanowire arrays directly grown on Ni foam using a facile hydrothermal procedure followed by calcination treatment will be introduced. When evaluated as a binder-free battery-type electrode for supercapacitor, a high specific capacity of 582.8 C g−<sup>1</sup> at a current density of 1 A g−<sup>1</sup> , a desirable rate capability with capacity retention about 84.8% at 20 A g−<sup>1</sup> , and an outstanding cycle performance of 93.1% capacity retention after 25,000 cycles can be achieved. More remarkably, an energy density of 33.8 W h kg−<sup>1</sup> at a power density of 224 W kg−<sup>1</sup> and wonderful cycling stability with 74% capacity retention after 10,000 cycles can be delivered based on the hybrid-supercapacitor with the as-prepared Co3O<sup>4</sup> nanowire arrays as a positive electrode and active carbon as negative electrode. All the unexceptionable supercapacitive behaviors illustrates that our unique 3D gully-network structure Co3O<sup>4</sup> nanowire arrays hold a great promise for constructing high-performance energy storage devices.

Keywords: Co3O4 nanowire arrays, 3D gully-network structure, hybrid-supercapacitor, stable cycle performance, battery-type electrode

### INTRODUCTION

With the rapid depletion of fossil fuels and deterioration of the environment, an urgent demand has emerged for the development of sustainable and clean energy storage devices in terms of high power density, energy density, and good safety performance (Wang et al., 2012b; Ren et al., 2015; Zhang et al., 2018; Zeng et al., 2019). Because of the admirable supercapacitive behaviors, ecological features, and safety characteristics, supercapacitors have been extensively considered as the major devices for energy storage adhibition thanks to the great potential as power sources for various applications, from electric vehicles to smart grids (Raj et al., 2015; Yan et al., 2017; Wu et al., 2018). As documented in the previous investigations that the supercapacitive behaviors of these devices heavily rely on the properties of electrode materials (Chang et al., 2016; Cui et al., 2019). Therefore,

**54**

the vital point to acquire wonderful behaviors of these devices is to hunt for proper materials and suitably design the electrode structure (Xu et al., 2017).

Because of the suitability for large-scale fabrication, high specific capacity, and rich redox reactions involving different ions, the transition metal oxides have been intensively studied as the most potential electrode materials in supercapacitors, which are divided into a capacitive electrode materials (MnO<sup>2</sup> and RuO2) having a quasi-rectangular cyclic voltammetry (CV) curve and a battery-type electrode materials (cobalt and nickel based compounds) having a strong separated redox peaks by the shape of the CV curve (Huang et al., 2014; Shao et al., 2015; Wang et al., 2015; Dai et al., 2017, 2018; Jiang et al., 2017; Wu et al., 2019). In particular, cobalt oxide, as a typical battery-type electrode material, has attracted a great deal of attention due to its low cost, natural abundance, high surface area structural characteristics, and high theoretical capacity (Yuan et al., 2012; Kuang et al., 2015; Jiang et al., 2016). Wang et al. successfully prepared Co3O<sup>4</sup> nanostructures with a specific capacity of 354.6 C g−<sup>1</sup> at 0.5 A g −1 , and the cycle performance in particular is about 97% after 2,000 cycles with a current density of 1 A g−<sup>1</sup> (Wang et al., 2016). Well-crystalline porous cobalt oxide (Co3O4) nanorods (NRs)synthesized by hydrothermal method by Shin's group (Jang et al., 2017). A specific capacity about 316.4 C g−<sup>1</sup> at 10 mV s−<sup>1</sup> can be achieved based on porous Co3O<sup>4</sup> NRs-300◦ C electrode. In addition, about 76% capacity retention of the porous Co3O<sup>4</sup> NRs electrode can be maintained after 5,000 cycles. 3D-nanonet hollow structured Co3O<sup>4</sup> electrode for battery-type supercapitor, prepared by a facile, low cost, and eco-friendly route under ambient temperature and pressure, can exhibit 410, 377.28, 346.68, and 328 C g−<sup>1</sup> under various scan rates about 5, 10, 20, and 30 mV s−<sup>1</sup> , respectively (Wang et al., 2014). The asprepared electrode also reveals good stability and keeps 90.2% of its initial capacity after 1,000 cycles at 5 A g−<sup>1</sup> . However, there is a big distance in the obtained Co3O<sup>4</sup> electrode reaching to the theoretical specific capacity and a better cycling stability owing to the intrinsic highly resistive nature or the agglomeration of the Co3O<sup>4</sup> nanomaterials during the electrode fabrication process (Rajeshkhanna et al., 2016; Wang et al., 2016; Kong et al., 2017).

One effective way proved to avoid the disadvantages caused by the binder addition and enhance the electrochemical performances of the Co3O<sup>4</sup> electrodes is to grow the active electrode materials directly on conductive substrates such as Ni foam (Mao et al., 2016). Zhang et al. confirmed that Co3O<sup>4</sup> nanowire array grown directly on Ni foam, fabricated by a simple hydrothermal method together with a post-annealing process, exhibited a specific capacity of 580 C g−<sup>1</sup> at 1 A g−<sup>1</sup> and a cycle performance of 90.6% after 5,000 cycles at 8 A g−<sup>1</sup> (Zhang et al., 2012). Gao et al. also reported that 3D star-shaped Co3O<sup>4</sup> nanowire growing on Ni foam prepared by a simple hydrothermal method displayed the specific capacity of 700 C g −1 at 1 A g−<sup>1</sup> . However, the specific capacity retention rate is 92.26% after only 1,000 cycles at a current density of 12.5 A g−<sup>1</sup> (Gao et al., 2016). Qing et al. investigated that Co3O<sup>4</sup> nanoflower structure on Ni foam showed a high specific capacity of 514.8 C g −1 at 1 A g−<sup>1</sup> , but its cycle performance was only 78.2% at 3 A g−<sup>1</sup> after 1,000 cycles (Qing et al., 2011). As a result, the integrated electrodes exhibited the improvement of the specific capacity, but their poor cycling stability was far from solved, which partly resulted from the large volume changes of Co3O<sup>4</sup> electrodes during the repeated cycle process (Yao et al., 2011), leading to structural instability, active material shedding off, and the capacity decay.

To address the above issues and achieve the excellent supercapacitive electrode materials, herein, we report our findings about the design, fabrication, and characterization of 3D gully-network structure Co3O<sup>4</sup> nanowire arrays directly grown on Ni foam through a facile and scalable hydrothermal procedure followed by annealing treatment. It has been demonstrated that the direct growth of 3D gully-network structure on Ni foam could ensure that every nanowire possesses good mechanical adhesion and participates in electrochemical reactions, so a stable 3D mechanical network benefits most to its excellent cycling ability. More importantly, good electrical contact with the current collector, which would result in low contact resistance and wonderful supercapacitive behaviors in terms of high specific capacity, admirable rate capability as well as outstanding cycle stability.

### EXPERIMENTAL

### Preparation of 3D Gully-Network Co3O<sup>4</sup> Nanowires on Ni Foam

All the purchased chemicals were analytical grade and used without further purification. The Ni foam was cleaned ultrasonically for 15 min in 3M HCl solution, absolute ethanol, and deionized water to remove the surface oxide layer and was dried in a vacuum oven for 2 h. 0.4851 g Co(NO3)<sup>2</sup> and 0.0912 g urea were dissolved in 40 mL deionized water. Next, 0.0834 g cetyltrimethylammonium bromide (CTAB) and 0.0133 g sodium dodecyl benzene sulfonate (SDBS) were added to the above mixed solution and were stirred magnetically for 15 min, respectively. Then, both the mixed solution and the readyprepared Ni foam were transferred into a 60 mL Teflon-lined stainless steel autoclave, sealed, and placed in an electric oven at 140◦C for 4 h, and then allowed to cool to room temperature naturally. The precursors and precipitates were collected and washed three times with absolute ethanol and deionized water, respectively, then dried at 60◦C overnight. Finally, the prepared precursor was annealed at 300◦C in air for 2 h with a heating rate of 1◦C min−<sup>1</sup> to obtain 3D gully-network Co3O<sup>4</sup> nanowires (NWAs) on Ni foam. The mass loading of Co3O<sup>4</sup> NWAs on Ni foam was about 2.1 mg cm−<sup>2</sup> .

### Preparation of the Hybrid-Supercapacitor

The hybrid-supercapacitor was assembled by the Co3O<sup>4</sup> NWAs on Ni foam as the positive electrode, the activated carbon (AC) as the negative electrode, polypropylene separator as the separator, 6 M KOH as the electrolyte. Specifically, the negative electrode was prepared as follows: First, AC and polytetrafluoroethylene (PTFE) (mass fraction: 5%) were mixed in a ratio of 9:1 and ultrasonically diluted with anhydrous ethanol. The Ni foam (3 × 3 cm) was soaked in the mixed solution for 3 min. Finally, the AC was uniformly distributed on Ni foam after drying in air at 60◦C for 6 h, and the mass loading of AC electrode was 2.4 mg cm−<sup>2</sup> . Positive and negative electrodes in hybrid-supercapacitor must meet charge balance (q<sup>+</sup> = q <sup>−</sup>). The charge contained in each electrode is calculated using the following equation [Equation (1)]:

$$\text{q } = \text{Q}\_{\text{ua}} \times \text{ m}\_{\text{ua}} \times \text{ S} \tag{1}$$

Where Qua (C g−<sup>1</sup> ) is the specific capacity per unit area of the electrode, mua (g) is the mass per area (cm<sup>2</sup> ) of the active material and S (cm<sup>2</sup> ) is the area of the electrode.

In order to achieve q<sup>+</sup> = q <sup>−</sup>, the area ratio satisfies equation (2):

$$\frac{\text{S}\_{+}}{\text{S}\_{-}} = \frac{\text{Q}\_{\text{ua}-} \times \text{m}\_{\text{ua}-}}{\text{Q}\_{\text{ua}+} \times \text{m}\_{\text{ua}+}} \tag{2}$$

According to the specific capacity of both the positive and the negative electrode, the best positive and negative electrode area ratio was found to be S−/S<sup>+</sup> = 3. Finally assembled into hybridsupercapacitor.

### Materials Characterization

Scanning electron microscope (SEM) analyses were performed using a SU8010 SEM, operated at 10 kV. Transmission electron microscope (TEM) images and selected area electron diffraction (SAED) patterns were collected on JEOL-2100F at 200 kV. Xray diffraction (XRD) using Cu-Ka radiation (Ka = 1.5418 Å). Raman scattering spectra were taken on a SPEX-1403 Raman spectrometer (Ar ion laser, 514.5 nm). X-ray photoelectron spectra (XPS) were recorded on an Axis Ultra, Kratos (UK) spectrometer, using a standard Al Kα X-ray source (150 W) and an analyzer pass energy of 30 eV. Samples were mounted using double-sided adhesive tape and binding energies are referenced to the C (1 s) binding energy of adventitious carbon contamination taken to be 284.8 eV.

### Electrochemical Measurements

The CV, galvanostatic charge/discharge (GCD), and electrochemical impedance spectroscopy (EIS) techniques were used to evaluate the electrochemical performance of Co3O<sup>4</sup> NWAs and hybrid-supercapacitor on a CHI660E electrochemical workstation at room temperature. Co3O<sup>4</sup> NWAs were tested in a three-electrode system where Hg/HgO electrode is the reference electrode, Pt electrode (3 × 3 cm) is the counter electrode and the preparation electrode material is the working electrode. However, hybrid-supercapacitors were tested in a two-electrode system with Co3O<sup>4</sup> NWAs as the positive electrode and AC as the negative electrode. All electrochemical tests were performed in 6M KOH. We calculated the specific capacity (Qsc) (C g−<sup>1</sup> ), energy density (E) (Wh Kg−<sup>1</sup> ), and power density (P) (W Kg−<sup>1</sup> ) based on the discharge time in the GCD curve and calculate it according to Equations (3–5).

$$\mathbf{Q}\_{\rm sc} = \frac{\rm It}{\rm m} \tag{3}$$

$$\begin{aligned} \mathbf{E} &= \frac{1}{2} \mathbf{Q}\_{\text{sc}} \mathbf{V} \\ \mathbf{P} &= \frac{\mathbf{E}}{\mathbf{t}} \end{aligned} \tag{5}$$

Where I (A) is the discharge current, t (s) is the discharge time, V (V) is the voltage window, and m (g) is the mass of the electrode material.

### RESULT AND DISCUSSION

**Figure 1** was the SEM images of a precursor nanowire array on Ni foam. **Figure 1a** presents that the precursor nanowire array uniformly grows on the Ni foam surface. According to the missing part, the thickness of the nanowire array is about 15µm. It is worth noting that the top of the nanowire is bent instead of perpendicular to the surface of the Ni foam. It is further observed (**Figure 1a**) that the tops of the nanowires were clustered together to form a ridge-like structure, and the connecting ridges eventually formed an irregular gully-network structure. **Figure 1b** confirms the top of each precursor nanowire gathering and fixing on the ridge to form a stable structure. **Figure 1c** shows that the precursor nanowires cross each other with a diameter of 50–100 nm, further improving the structure stability of the precursor nanowire array. **Figure 1d** displays the cross-sectional of the precursor nanowires. The cross-sectional presents a hexagon. It is possible to add anionic and cationic surfactants in a certain proportion during the preparation to make the solution more polar, which limits the lateral growth of nanowires (Stellner et al., 1986; Wu et al., 2013; Zhang et al., 2016). The precursor is converted into Co3O<sup>4</sup> by the heat treatment. During the heat treatment, the precursor thermally decomposes to produce H2O and CO<sup>2</sup> [reaction (6)], which is derived from the slow hydrolysis of urea [reaction (7)–(9)]. It has been reported that the slow hydrolysis of urea facilitates the growth of nanowires and has better crystallinity along the

FIGURE 1 | (a–c) SEM images of precursor nanowire arrays on Ni foam at different magnifications, (d) the cross-sectional SEM image of the precursor nanowires.

longitudinal axis. Synthesized ultra-long nanowires benefit from the addition of urea (Wang et al., 2012a). The hydrolyzed CO2<sup>−</sup> 3 and OH<sup>−</sup> combine with Co2<sup>+</sup> to form the precipitate Co(CO3)0.5(OH)·0.11H2O (reaction (10): (Fan et al., 2017).

3Co (CO3)0.5 (OH) · 0.11H2O <sup>1</sup>→Co3O<sup>4</sup> (6)

$$\uparrow + 1.83 \text{H}\_2\text{O} \uparrow + 1.5 \text{CO}\_2 \uparrow$$

$$\text{H}\_2\text{NCONH}\_2 + \text{H}\_2\text{O} \rightarrow \text{ 2NH}\_3\uparrow + \text{CO}\_2\uparrow\tag{7}$$

CO<sup>2</sup> + H2O → CO2<sup>−</sup> <sup>3</sup> <sup>+</sup> 2H<sup>+</sup> (8)

$$\text{NH}\_3 + \text{H}\_2\text{O} \rightarrow \text{NH}\_4^+ + \text{OH}^-\tag{9}$$

$$\text{Co}^{2+} + 0.5\text{CO}\_3^{2-} + \text{OH}^- + 0.11\text{H}\_2\text{O} \rightarrow \text{Co}(\text{CO}\_3)\_{0.5}$$

$$(\text{OH}) \cdot 0.11\text{H}\_2\text{O} \downarrow$$

The SEM images of Co3O<sup>4</sup> NWAs on Ni foam obtained after annealing are shown in **Figures 2b–f.** Comparing **Figures 2a,b**, it can be seen that the gully-network structure does not change significantly before and after annealing. Because the gullynetwork structure makes the nanowires cross each other so that the nanowires can be fixed. Therefore, it has a good spatial buffer to adapt to the volume change in the annealing process. Further magnification of **Figure 2c** shows the sides of the gully-network structure, where the nanowires gather along the gully orderly and form a hemp rope structure at the top, so that the nanowire array forms a tight whole. As shown in **Figure 2d**, the Co3O<sup>4</sup> nanowires diameter is 40–70 nm, which is 10–30 nm smaller than the precursor nanowire, owing to the loss of CO<sup>2</sup> and H2O during annealing treatment (Guan et al., 2017).

Co3O<sup>4</sup> nanowires further characterized by TEM equipped with HRTEM and SAED as shown in **Figures 2e,f**. The surface of the Co3O<sup>4</sup> nanowires (**Figure 2e**) is uneven and highly porous, providing more specific surface area and active sites. The SAED plot (inset in **Figure 2e**) displays the single crystal diffraction spots of Co3O<sup>4</sup> nanowires, and diffraction spots of (220) (311) (511) (400) and (440) are observed. The long axis direction of the nanowires is perpendicular to the (220) crystal plane and the nanowires grow along the (110) crystallographic axis. It is worth noting that the crystal growth direction is different from that in other articles along the (111) direction (Xia et al., 2012). HRTEM image of the middle part of the Co3O<sup>4</sup> nanowire (**Figure 2e**) is shown in **Figure 2f**. The lattice spacing of the vertical lattice fringes is 0.202 and 0.143 nm corresponds to the (004) and (440) crystal planes of Co3O<sup>4</sup> (JCPDS card no.42- 1467). In addition, there are two crystal planes with the same lattice spacing of 0.233 nm on the HRTEM image, corresponding to the (222) and (2¯22) crystal planes of Co ¯ <sup>3</sup>O<sup>4</sup> with the crystal plane angle of 109.5◦ . The FTT corresponding to HRTEM is shown in the inset in **Figure 2f**, which exhibits the (222) and (2¯22) crystal planes compared to SAED. Obviously noticed that ¯ there is no complete and clear lattice fringes in the dotted circle. The dashed circle may be a surface defect because of the loss of CO<sup>2</sup> and H2O during the annealing treatment, which has high electrochemical activity and improves electrochemical performance.

FIGURE 2 | (a) SEM images of precursor nanowire arrays on Ni foam, (b–d) SEM images of Co3O4 NWAs on Ni foam at different magnifications, (e) TEM images (inset: SAED pattern) of Co3O4 nanowires, (f) HRTEM image of Co3O4 nanowires and corresponding FFT images (inset).

XRD, Raman spectrum, and XPS are used to analysis the crystal structure of Co3O<sup>4</sup> NWAs. The XRD pattern of the precursor in **Figure S1A** is consistent with Co(CO3)0.5(OH)·0.11H2O (JCPDS card no. 48-0083) (Xiong et al., 2012). After annealing, the XRD pattern is shown in **Figure 3A**. Except for the three diffraction peaks of the Ni foam (JCPDS card no. 04-0850) (Yu et al., 2015), the remaining diffraction peaks exactly corresponds to Co3O<sup>4</sup> (JCPDS card no. 42-1467) (Che et al., 2014). **Figure S1B** indicates the Raman spectra with Raman peaks appearing at 476, 516, 605, 675 cm−<sup>1</sup> corresponding to Eg, F 1 2g , F 2 2g , and A1g peaks of Co3O<sup>4</sup> (Kim et al., 2011). To study the chemical bonding state of Co3O<sup>4</sup> NWAs, XPS measurement is utilized. Three elements of Co, O, and C in the XPS spectrum (**Figure 3B**) can be noted and no other elements are found, indicating that there are no impurities in the synthesis of Co3O<sup>4</sup> NWAs. **Figure 3C** shows the XPS spectrum of Co2p with two sets of peaks, Co 2p3/<sup>2</sup> (781.2 eV/779.7 eV) and Co 2p1/<sup>2</sup> (796.7 eV/795.0 eV), which are consistent with Co2+/Co3<sup>+</sup> in Co3O<sup>4</sup> NWAs, respectively (Hu et al., 2016; Zhang et al., 2019). Two satellite peaks of sat.Co 2p3/<sup>2</sup> (786.6 eV) and sat.Co 2p1/<sup>2</sup> (803.6 eV) are also observed (Li et al., 2014). The O 1 s spectrum (**Figure 3D**) consists of 529.7, 531.3, and 533.3 eV. These three peaks represent the O element in Co3O4, the O element in the water adsorbed on the surface, and the oxygen vacancies formed by gas escape during annealing, respectively (Xiong et al., 2009; Qu et al., 2017). Co3O<sup>4</sup> NWAs is further demonstrated the successful synthesis by XPS.

Electrochemical performance evaluations of Co3O<sup>4</sup> NWAs with gully-network structure are conducted by CV, GCD, and EIS techniques. As shown in **Figure 4A**, the CV curve of Co3O<sup>4</sup> NWAs on Ni foam obtained in potential range of 0–0.6 V at the scan rate from 2 to 100 mv s−<sup>1</sup> shows a pair of reversible redox peaks at 0.36 and 0.26 V, displaying a typical faraday behavior (Guan et al., 2017). It is known that the redox reaction occurs between Co3O<sup>4</sup> and OH<sup>−</sup> in the electrolyte. No OH<sup>−</sup> in the electrolyte, no redox reaction occurs. To confirm the above conclusion, the electrochemical tests of Co3O<sup>4</sup> NWAs under alkaline and neutral electrolyte conditions are shown in **Figure S2**. The results from **Figure S2a** show that the area of the CV curve in the alkaline electrolyte is significantly larger than that in the neutral electrolyte. It indicates that although the Co3O<sup>4</sup> NWAs possesses a large specific surface area, the provided double layer capacitance is very limited, so the faraday behavior is dominant. Meanwhile, the reversible redox reactions occurred are as follows:

$$\text{Ca}\_3\text{O}\_4 + \text{OH}^- + \text{H}\_2\text{O} \leftrightarrow \text{3CoOH} + \text{e}^- \tag{11}$$

$$\text{CoOOH} + \text{OH}^- \leftrightarrow \text{CoO}\_2 + \text{H}\_2\text{O} + \text{e}^- \tag{12}$$

It is noteworthy that the anode and cathode peaks move slightly toward the high and low potentials, respectively, as the scan rate increases. The above phenomenon can be ascribed to the internal ion diffusion resistances enhancing with the increase of the scan rates. However, there is no significant shift in the redox peak, indicating that the electrode exhibits good reaction kinetics (Wang et al., 2012b; Raj et al., 2015). In addition, the nanowire array has a porous structure, which facilitates rapid reversible oxidation-reduction reactions. **Figure 4B** displays the GCD curves of Co3O<sup>4</sup> NWAs at different current densities under 0–0.5 V, which presents the excellent symmetry of the chargedischarge curves and no significant voltage drop can be observed, indicating the small electrode resistances of the electrodes. **Figure 4C** shows the specific capacity values at different current densities calculated from the discharge time in **Figure 4B**. It is clearly observed that a high specific capacity is 582.8 C g−<sup>1</sup> at a current density of 1 A g−<sup>1</sup> can be achieved. The specific capacity still maintains 84.8% with the value of 494 C g−<sup>1</sup> when the current density increases to 20 A g−<sup>1</sup> , which reveals the excellent rate performance of the Co3O<sup>4</sup> NWAs electrode. Furthermore, the long-term cycling stability of the Co3O<sup>4</sup> NWAs electrode at 15 A g−<sup>1</sup> is presented in **Figure 4D**. As the cycle progressed, the capacity showed a trend of first increase and then decrease behaviors. The initial increase can be attributed to the activation process of the Co3O<sup>4</sup> NWAs electrode. After that, the slow decay started due to the inactivation of the structure. More remarkably, the capacity still maintains 93.1% after 25,000 cycles, indicating excellent long-term cycling stability. Compared with the related literatures, the 3D gully-network structure Co3O<sup>4</sup> NWAs has the superior electrochemical performances, especially the cyclic stability (**Table 1**). The inset in **Figure 4D** shows the final GCD curve during the cycles, which is basically consistent with that before the cycles, demonstrating wonderful reversibility of the Co3O<sup>4</sup> NWAs electrode. The SEM images before and after 25,000 cycles are exhibited in **Figure 5**, from which the microstructure of the Co3O<sup>4</sup> NWAs remained essentially intact, indicating the outstanding stability of the gully-network structure. With the unique structure, the electrode would possess excellent adaptability to the volume expansion and contraction during the cycle, thus resulting in the brilliant cycling performance.

In order to determine the changes about the resistance of the Co3O<sup>4</sup> NWA electrode, the EIS analysis are applied before


and after 25,000 cycles (**Figure 6**). The Nyquist plot is divided into the high-frequency region of the curve portion and the low-frequency region of the straight-line portion. The highfrequency region is controlled by the electrode reaction kinetics to represent the charge transfer resistance (Rct), and the lowfrequency region can be stand for the diffusion resistance (W) between the electrolyte and the electrode material. In addition, the intersection point between the impedance curve and the horizontal axis is the intrinsic resistance (Rs) of the electrode (Xiong et al., 2012; Yang et al., 2013; Cai et al., 2014; Xu et al., 2016). The inset of **Figure 6** shows that the intrinsic resistance of Co3O<sup>4</sup> NWAs decreases after 25,000 cycles (from 0.72 to 0.66 ). It may be that a portion of the Co3O<sup>4</sup> NWAs breaks away from the Ni foam during the cycles, resulting in the intrinsic resistance decrease. The Rct (from 1 to 1.66 ) and W (from 0.06 to 0.21 ) before and after the cycle were observed to increase. This phenomenon is probably due to the breakage of fewer nanowires during the cycles, leading to the destruction of the path with the smallest electrical resistance transported between the Co3O<sup>4</sup> NWAs and Ni foam and thereby increasing the charge transfer resistance and diffusion resistance. In general, the impedances are approximately the same before and after 25,000 cycles, which further proves that the gullynetwork structure of Co3O<sup>4</sup> NWAs possess excellent cycling stability.

To explore the practical application of Co3O<sup>4</sup> NWAs electrodes, the Co3O<sup>4</sup> NWAs//AC hybrid-supercapacitor based on the Co3O<sup>4</sup> NWAs positive electrode and AC negative electrode is successfully assembled. **Figure 7A** shows the CV curves of the Co3O<sup>4</sup> and AC electrode materials at a scan rate of 20 mv s−<sup>1</sup> in a three-electrode system. The voltage window of the AC negative electrode ranges from −1 to 0 V, and the curve exhibits a typical double-layer behavior with a quasi-rectangular shape. Meanwhile, the positive voltage window of the Co3O<sup>4</sup> NWAs electrode is 0–0.6 V, and the obvious redox peaks suggest the faraday behavior. The CV curves of the voltage window ranging from 0–0.8 V to 0–1.6 V for the Co3O<sup>4</sup> NWAs//AC hybrid-supercapacitor device at a scan rate of 20 mv s−<sup>1</sup> are

shown in **Figure 7B**. The area of the CV curve becomes larger as the voltage window increases and no obvious redox peaks appear, indicating a good capacitive behavior. A significant polarization occurs when the voltage window increases to 1.6 V, so the voltage window choose to be 0–1.5 V. **Figure 7C** displays the CV curves at different scan rates under 0–1.5 V, presenting that a squarelike curve remains similar at high scan rates. The GCD curves at different current densities are shown in **Figure 7D**, which displays good supercapacitive behavior and agrees with the CV curve results. A specific capacity of 162 C g−<sup>1</sup> at 0.2 A g−<sup>1</sup> of the device can be achieved and still remains 78 C g−<sup>1</sup> at a high current density of 10 A g−<sup>1</sup> . **Figure 7E** is Ragone plots of the Co3O<sup>4</sup> NWAs//AC devices based on the GCD discharge time, which shows that an energy density of 33.8 W h kg−<sup>1</sup> at a power density of 224 W kg−<sup>1</sup> is obtained. Even at a high power density of 12,000 W kg−<sup>1</sup> , an energy density of 16.25 W h kg−<sup>1</sup> can also be kept. The energy and power densities of the Co3O<sup>4</sup> NWAs//AC

FIGURE 7 | (A) The CV curve of AC electrode and Co3O<sup>4</sup> NWAs electrode at a scan rate of 20 mV s−<sup>1</sup> , (B) CV curves of Co3O4 NWAs//AC device at different potential ranges, (C) CV curves of Co3O4 NWAs//AC device at different scan rates, (D) GCD curves of Co3O4 NWAs//AC device at different current densities, (E) Ragone plot of the Co3O<sup>4</sup> NWAs//AC device, (F) Cycling performance of the Co3O<sup>4</sup> NWAs//AC device at a current density of 2 A g−<sup>1</sup> (inset is the photo of the lighted LED by the double Co3O4 NWAs//AC device).

hybrid-supercapacitor are much higher than those of some other electrode materials, such as Co3O4//AC (24.9 Wh Kg−<sup>1</sup> at 225 W Kg−<sup>1</sup> ) (Zhang et al., 2013), α-Co(OH)2/Co3O4//AC (19.2 Wh Kg−<sup>1</sup> at 145 W Kg−<sup>1</sup> ) (Jing et al., 2014), Co3O4@MnO2//MEGO (17.7 Wh Kg−<sup>1</sup> at 750 W Kg−<sup>1</sup> ) (Huang et al., 2014). To evaluate the practical application of the device, the cycle performance of the hybrid-supercapacitor device is measured at a current density of 2 A g−<sup>1</sup> (**Figure 7F**). It can be observed that the hybrid-supercapacitor devices assembled with Co3O<sup>4</sup> NWAs also exhibits excellent cycle behaviors, maintaining a specific capacity of 74% after 10,000 cycles. The gully-network structure of Co3O<sup>4</sup> NWAs materials ensures the stability of the structure and reduces the loss of capacity during the cycles. Moreover, the Co3O<sup>4</sup> NWAs//AC hybrid-supercapacitor devices successfully light four LED lamps (inset in **Figure 7F**), illustrating that the as-prepared Co3O<sup>4</sup> NWAs//AC devices can be wildly utilized for practical applications.

The admirable electrochemical behaviors of the Ni foam supported Co3O<sup>4</sup> battery-type electrode materials for hybridsupercapacitor can be ascribed to the following aspect: Firstly, the highly stable gully-network of Co3O<sup>4</sup> NWAs effectively alleviates the volume expansion and contraction during the long cycle measurements and prevents the agglomeration of active materials, generating the outstanding cycle stability. Meanwhile, its porous structure enables the electroactive material to be completely exposed in the electrolyte solution, which facilitates the diffusion of ions and electrons, and allows strong storage of ions, resulting in excellent cycle stability. Secondly, the highly porous interconnected nanowire arrays consisted of numerous highly crystalline nanoparticles can greatly facilitate electrolyte transport and increase the surface area, thus promoting the electrolyte to easily diffuse into the inner region of the electrode. Finally, the direct growth of Co3O<sup>4</sup> NWAs on Ni foam can significantly improve electrical conductivity, creating an expressway for electric charge transfer and resulting in desirable rate capability.

### CONCLUSION

3D gully-network structure Co3O<sup>4</sup> NWAs directly grown on Ni foam have been successfully prepared via a facile and scalable hydrothermal procedure followed by calcination treatment. The unique porous gully-network structure of the Co3O<sup>4</sup> NWAs electrode can offer large surface area, provide open channels for efficient ion transport as well as adaptability to the volume expansion and contraction during electrochemical reactions, which greatly improve the specific capacity, rate capability, and cycling stability of the Co3O<sup>4</sup> NWAs electrode. When evaluated as a binder-free electrode for battery-type supercapacitor, unexceptionable faraday behaviors in terms of high specific capacity (582.8 C g−<sup>1</sup> at a current density of 1 A g −1 ), desirable rate capability (with capacity retention about 84.8% at 20 A g−<sup>1</sup> ), and outstanding cycle performance (93.1% capacity retention after 25,000 cycles) can be achieved. More remarkably, an energy density of 33.8 W h kg−<sup>1</sup> at a power density of 224 W kg−<sup>1</sup> and wonderful cycling stability with 74% capacity retention after 10,000 cycles can be acquired based on the hybrid-supercapacitor with the as-prepared Co3O<sup>4</sup> NWAs positive electrode. All the excellent supercapacitive behaviors illustrates that our unique gully-network structure Co3O<sup>4</sup> nanowire arrays hold a great promise for constructing high-performance energy storage devices.

### AUTHOR CONTRIBUTIONS

CG design and performed the experiments. CG, MY, CW, JL, and CS prepared the samples and analyzed the data. CG, CJ, TL, LH, and YW participated in interpreting and analyzing the data. All authors read and approved the final manuscript.

### ACKNOWLEDGMENTS

The authors acknowledge the National Science Foundation of China (Grants 21201129, 51208333, 51374151) and the National

### REFERENCES


Natural Science Foundation of Shanxi Province (2013011012-3) for providing funding support to the current work. CJ is grateful for support of the 100 Talented Team of Hunan Province and the 111 project (B12015) at the Key Laboratory of Advanced Energy Materials Chemistry (Ministry of Education), Nankai University.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2018.00636/full#supplementary-material


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Guo, Yin, Wu, Li, Sun, Jia, Li, Hou and Wei. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# In-situ Grown SnS<sup>2</sup> Nanosheets on rGO as an Advanced Anode Material for Lithium and Sodium Ion Batteries

Hezhang Chen<sup>1</sup> , Bao Zhang<sup>1</sup> , Jiafeng Zhang<sup>1</sup> , Wanjing Yu<sup>1</sup> , Junchao Zheng<sup>1</sup> , Zhiying Ding<sup>2</sup> , Hui Li <sup>1</sup> , Lei Ming<sup>1</sup> , D. A. Mifounde Bengono<sup>1</sup> , Shunan Chen<sup>1</sup> and Hui Tong<sup>1</sup> \*

*<sup>1</sup> School of Metallurgy and Environment, Central South University, Changsha, China, <sup>2</sup> School of Chemistry and Chemical Engineering, Central South University, Changsha, China*

### Edited by:

*Qiaobao Zhang, Xiamen University, China*

#### Reviewed by:

*Manickam Minakshi, Murdoch University, Australia Guiming Zhong, Fujian Institute of Research on the Structure of Matter (CAS), China Yongchang Liu, University of Science and Technology Beijing, China Xianwen Wu, Jishou University, China Qiulong Wei, UCLA Department of Environmental Science and Engineering, United States*

#### \*Correspondence:

*Hui Tong huitong@csu.edu.cn*

### Specialty section:

*This article was submitted to Physical Chemistry and Chemical Physics, a section of the journal Frontiers in Chemistry*

Received: *14 March 2018* Accepted: *03 December 2018* Published: *18 December 2018*

#### Citation:

*Chen H, Zhang B, Zhang J, Yu W, Zheng J, Ding Z, Li H, Ming L, Bengono DAM, Chen S and Tong H (2018) In-situ Grown SnS*2 *Nanosheets on rGO as an Advanced Anode Material for Lithium and Sodium Ion Batteries. Front. Chem. 6:629. doi: 10.3389/fchem.2018.00629* SnS<sup>2</sup> nanosheets/reduced graphene oxide (rGO) composite was prepared by reflux condensation and hydrothermal methods. In this composite, SnS<sup>2</sup> nanosheets *in-situ* grew on the surface of rGO nanosheets. The SnS2/rGO composite as anode material was investigated both in lithium ion battery (LIB) and sodium ion battery (SIB) systems. The capacity of SnS2/rGO electrode in LIB achieved 514 mAh g−<sup>1</sup> at 1.2 A g−<sup>1</sup> after 300 cycles. Moreover, the SnS2/rGO electrode in SIB delivered a discharge capacity of 645 mAh g−<sup>1</sup> at 0.05 A g−<sup>1</sup> ; after 100 cycles at 0.25 A g−<sup>1</sup> , the capacity retention still keep 81.2% relative to the capacity of the 6th cycle. Due to the introduction of rGO in the composite, the charge-transfer resistance became much smaller. Compared with SnS2/C electrode, SnS2/rGO electrode had higher discharge capacity and much better cycling performance.

Keywords: SnS2 , reduced graphene oxide, thin nanosheets, anode material, lithium ion batteries, sodium ion batteries

### INTRODUCTION

Lithium ion batteries (LIBs) are being widely used in the electric vehicles and energy storage fields (Wu et al., 2017; Chen et al., 2018b,c; Cui et al., 2018; Zhang et al., 2018; Zheng et al., 2018). However, the commercial graphite anode is far to meet the requirements of the high-performance LIBs due to its low theoretical capacity (Ryu et al., 2016; Li et al., 2017b; Yan et al., 2017; Wang et al., 2018). Furthermore, lithium resource is limited in nature. It is necessary to develop low cost and high storage performance sodium ion batteries (SIBs) to satisfy the energy demand (Li et al., 2017a; Zhu et al., 2017; Chen et al., 2018a). Sodium is rich on the earth, and possesses similar physical and chemical properties as lithium (Qu et al., 2014; Jiang et al., 2015; Wang et al., 2015b; Zhang et al., 2015a, 2017a; Fang et al., 2016; Xu et al., 2016c; Leng et al., 2017). However, the commercial graphite material is not suitable for SIBs (Qu et al., 2014; Fang et al., 2016; Xu et al., 2016c; Zhang et al., 2016a). Therefore, it is urgent to develop new anode materials with low cost and good electrochemical properties for SIBs.

Currently, many researches have been carried out in transition metal oxides (Zhang et al., 2016b), metals (Lin et al., 2013) and carbonaceous materials (Xu et al., 2016a). These anode materials were used in both LIBs and SIBs, but suffer from the disadvantages of poor sodium storage, poor cycling property, etc. The layered structure materials, such as SnS2, have been considered as promising anode materials for SIBs (Chao et al., 2016), due to high theoretical capacity, electrochemical stability, low cost and environmental friendliness. SnS<sup>2</sup> has a layered structure with a large interlayer distance which can promote the intercalation and deintercalation

**64**

of lithium and sodium ions. SnS2/C prepared by using polyacrylonitrile as carbon source delivered the capacity of 570 mAh g−<sup>1</sup> after 100 cycles at discharge current density of 0.05 A g −1 (Wang et al., 2015b). However, SnS<sup>2</sup> electrode still suffers the poor electrochemical performance, such as large cycling capacity loss for its low electronic conductivity. To overcome these problems, graphene was introduced into the anode materials. The graphene can improve the electric conductivity and decrease the volume change during the intercalation and deintercalation process of the lithium and sodium ions. So, the electrochemical performance of the anode material could be improved by reduced graphene oxide (rGO) introduction. Du et al. reported that, after 50 cycles, Co3S4-PNS/graphene sheet electrode still show the discharge capacity of 329 mAh g−<sup>1</sup> at 0.5 A g−<sup>1</sup> (Du et al., 2015). SnSe<sup>2</sup> nanoplate/graphene composite was synthesized by hydrothermal method and showed a better storage performance than that of SnSe<sup>2</sup> nanoplates without graphene (Choi et al., 2011). Zhang et al. reported that FeSe2/sulfurdoped rGO sheets displayed the discharge capacities of 383.3 and 277.5 mAh g−<sup>1</sup> at high current densities of 2.0 and 5.0 A g−<sup>1</sup> (Zhang et al., 2016a). Nanosheet materials could be also prepared by solution combustion synthesis method (Ramkumar and Minakshi, 2015; Ramkumar and Sundaram, 2016).

In this work, a novel method was developed to synthesize SnS2/rGO composite anode material for LIBs and SIBs. In this composite, SnS<sup>2</sup> grew on the surface of rGO nanosheets, which possessed good electronic conductivity. So, the SnS2/rGO electrode exhibited outstanding lithium and sodium storage properties and cycling performance. The preparation mechanism, as well as the physical and electrochemical properties of SnS2/rGO composite was carefully discussed.

### EXPERIMENTAL

Graphene oxide (GO) was dispersed in deionized water, with the concentration of 1.5 mg mL−<sup>1</sup> . 20 mL ethylene glycol and 30 mL GO solution were introduced into a round-bottom flask. The mixed liquid was ultrasonically treated for 0.5 h. 0.5438 g SnCl4·5H2O dissolved in 10 mL EG was then added into the mixed liquid. After magnetically stirred for 0.5 h, the suspension was heated to 120◦C and treated by reflux condensation method for 2 h, and then cooled down to the room temperature. 0.6014 g thioacetamide and the suspension were added to a 100 mL Teflon-lined stainless steel autoclave and then magnetically stirred for 20 min. Then, the autoclave was transferred into an oven and kept at 160◦C for 12 h. After cooling down naturally to room temperature, the precipitate was centrifuged, and washed by deionized water and absolute alcohol for several times, and then dried at a 80◦C vacuum oven for a whole night. The mixture was sintered at 500◦C for 4 h under Ar atmosphere. Finally, SnS2/rGO composite with 15 wt% rGO was obtained. The SnS2/rGO composites with 10 and 20 wt% rGO were also prepared through the same way by adding different contents of GO. For comparison, the SnS2/C composite was synthesized by the same method without using GO, replaced by 1 g glucose as the carbon source added into the suspension together with thioacetamide.

The crystalline phase of SnS2/rGO composite was analyzed by X-ray diffraction (XRD, Rigaku D/Max 200PC, Japan) using Cu Kα radiation. The scanning rate was 5◦ per minute, and the range of scanning diffraction angle (2θ) was from 10◦ to 80◦ . The Raman spectra were conducted by Raman spectroscopy (Lab RAM Aramis, Jobin Yvon, France). The oxidation states of Sn, S, and C elements in the samples were studied by X-ray photoelectron spectroscopy (XPS, PHI5700, USA). The morphologies of the samples were observed by scanning electron microscopy (SEM, FEI, Nova NanoSEM-230, USA), and high resolution transmission electron microscopy (TEM, FEI, Tecnai G2 F20 S-Twin, USA), working at 200 kV. The element contents of the samples were studied by energy dispersive Xray spectroscopy (EDS). The carbon and sulfur analyzer (CS744, Leco, USA) was applied to quantify the amount of carbon and sulfur in the composite.

The SnS2/rGO anode material was evaluated using 2025-type coin cells prepared in a pure argon-filled glove box. The product slurry was prepared by mixing SnS2/rGO powder, carbon black and polyvinylidene fluoride by 8:1:1 in weight, and dispersed in N-methyl pyrrolidinone. The result slurry was pasted onto a Cu foil. After dried, the foil with SnS2/rGO material was cut to a 14 mm diameter disk. The electrolyte for SIBs was 1 M solution of NaClO<sup>4</sup> dissolved in ethylene carbon (EC)/dimethyl carbonate (DMC) (weight ratio 1:1) and 5 wt% fluoroethylene carbonate. The electrolyte for LIBs was 1 M solution of LiPF<sup>6</sup> in EC-DMC (weight ratio 1:1). The counter electrode for SIBs was a sodium foil and a Whatman GF/D as the separator. And lithium foil counter electrode and Celgard 2500 separator were used in LIBs. Electrochemical tests were conducted on LAND battery cycler by using an automatic galvanostatic chargedischarge unit (LAND CT2001A, China), with the potentials of 0.01–2.50 V vs. Na/Na<sup>+</sup> electrode and 0.01-1.80 V vs. Li/Li<sup>+</sup>

electrode. The cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS) measurements were recorded by an electrochemical workstation (CHI660D, CH Instruments, USA). The EIS spectra were recorded by applying an AC voltage of 5 mV amplitude in the frequency of 10−2−10<sup>5</sup> Hz. The scanning rate of CV was 0.1 mV s−<sup>1</sup> .

### RESULTS AND DISCUSSION

SnS2/rGO composite was prepared by reflux condensation and hydrothermal methods. The XRD pattern of SnS2/rGO composite was showed in **Figure 1**. The pattern shows that all the peaks are indexed to SnS<sup>2</sup> (JCPDS#23-0677) with 2T-type layered structure. There were no obvious impurity peaks in the XRD pattern, especially at 10.9◦ which is the location of characteristic peak of GO. It means that there is no impurity in SnS<sup>2</sup> sample and GO was reduced to rGO through heat treatment.

The SEM images of SnS2/rGO composites with different contents of rGO were shown in **Figure 2**. The designed contents of rGO in the composites were 10 wt% (**Figure 2A**), 15 wt% (**Figure 2B**) and 20 wt% (**Figure 2C**). In **Figure 2A**, it can be observed that SnS<sup>2</sup> not only grew on rGO nanosheets but also self-assembled as nanoplates. When the rGO content increased to 15 wt%, there were no SnS<sup>2</sup> nanoplates observed in **Figure 2B** and the thickness of SnS2/rGO sheets became thicker than that of rGO sheets (not shown). In **Figure 2C**, the morphology of the sample seems no obvious difference with that of the sample in **Figure 2B**. The rGO in the samples can be regarded as a template for the growth of SnS2. SnS<sup>2</sup> self-assembled as nanoplates when the content of rGO was not enough. And the SnS<sup>2</sup> nanoplates disappeared as the rGO content increased in the composite. The rGO content in the composite (**Figure 2C**) is excessive, which reduce the volume energy density and raise product cost. So, the optimum content of rGO was 15 wt% among the three samples. The measured content of S element in the composite (**Figure 2B**) was 29.53 wt%. So, the content of rGO in the composite was calculated as 15.8 wt%, which is close to the designed value of 15 wt%. The properties of SnS2/rGO (15 wt%) composite were further investigated.

The distributions of Sn, S, and C elements of SnS2/rGO composite were studied by EDS. The EDS mappings further show that SnS<sup>2</sup> was distributed homogeneously on rGO nanosheets. The microstructure of SnS2/rGO nanosheets was conducted by TEM, as shown in **Figure 3**. In **Figure 3A**, the TEM image shows that SnS<sup>2</sup> nanosheets were coated by the plicate graphene nanosheets. The inset figure in **Figure 3A** shows that SnO<sup>2</sup> nanoparticles grew on the GO nanosheets before the hydrothermal synthesis of SnS2/rGO. The SnO<sup>2</sup> nanoparticles were about 5–10 nm and coated on the surface of GO nanosheets tightly. Thus, SnS<sup>2</sup> nanosheets could in-situ grow on the surface of GO nanosheets. The side view of the SnS<sup>2</sup> nanosheets is shown in **Figure 3B**, and it is seen that the thickness of the nanosheets is about 10 nm. The parallel fringe spacing in **Figure 3C** was 0.586 nm and the lattice distance in **Figure 3D** was 0.319 nm, corresponding to (001) plane and (100) plane of SnS<sup>2</sup> with 2Ttype layered structure, respectively. There is no obvious crack or collapse in SEM and TEM images, which is beneficial for good cycle stability of the composite.

The XPS analysis was applied to investigate the electronic states of Sn, S, and C elements in SnS2/rGO composite, as shown in **Figure 4**. The Sn 3d, C 1s and S 2p peaks can be observed clearly in **Figure 4A**. In **Figure 4B**, the C 1s peak was resolved into four parts. The peaks centered at 284.6, 285.6, 286.3, and 287.3 eV correspond to C-C, C-O, C = O and O-C = O type

(E) Sn, (F) S, and (G) C.

bonds, respectively (Qu et al., 2014; Zhang et al., 2014; Fang et al., 2016). In **Figure 4C**, the high-resolution Sn 3d spectrum exhibited two signals at 487.2 and 495.6 eV for Sn 3d3/2 and Sn 3d5/2, respectively, corresponding to Sn4+. In **Figure 4D**, the presence of SnS<sup>2</sup> can be confirmed by S 2p peak at 163.4 and 162.3 eV (Zhang et al., 2015b).

The electrochemical performances of the samples were investigated by coin cells. **Figure 5** shows the lithium storage, rate and cycling properties. The CV test of SnS2/rGO electrode was conducted in the voltage of 0.01–2.6 V. The alloying and dealloying reactions happened between Li and SnS2, as described in Equations (1) and (2) (Zhai et al., 2011). In the first cycle of cathodic scans, LixSnS<sup>2</sup> formed at 1.8 V by the lithium intercalation into SnS<sup>2</sup> layers. Two broader peaks were present at about 1.3 and 0.12 V. The peak at about 1.3 V corresponds to the decomposition of the SnS<sup>2</sup> into metal Sn and Li2S. The peak at about 0.12 V is ascribed to the formation of LixSn by lithium and metal Sn (Zhai et al., 2011; Zhang et al., 2014). The peaks in 1.8-2.6 V correspond to the formation of SnS2. In the anodic scans, the peaks appeared at around 0.5 V are mainly related to the dealloying of LixSn. **Figures 5B,C** show the typical charge and discharge profiles of SnS2/rGO and SnS2/C electrodes at 600 mA g−<sup>1</sup> , respectively. The charge and discharge voltage was between 0.01 and 1.8 V. The SnS2/C electrode exhibited initial discharge and charge capacity 1574 and 790.1 mAh g−<sup>1</sup> , with a coulombic efficiency of 50.2%. However, SnS2/rGO electrode delivered higher discharge and charge capacity of 1616.6 and 899.6 mAh g−<sup>1</sup> , with higher initial coulombic efficiency of 55.6% compared with that of SnS2/C electrode (50.2%).

$$\text{SnS}\_2 + 4\text{ Li} \rightarrow \text{Sn} + 2\text{Li}\_2\text{S} \tag{1}$$

$$\text{Sn} + 4.4\,\text{Li} \to \text{Li}\_{4.4}\text{Sn} \tag{2}$$

Rate capability is highly crucial for anode materials in LIBs, and the rate performances of the samples are shown in **Figure 5D**. SnS2/rGO electrode exhibited outstanding rate performance compared with that of SnS2/C electrode. SnS2/rGO electrode showed high discharge capacities of 776, 715, 635.6, 595.2, 517.5 and 447.1 mAh g−<sup>1</sup> at 0.2, 0.5, 1, 2, 5 and 8 C, respectively. After cycling at different current densities, SnS2/rGO electrode delivered 675 mAh g−<sup>1</sup> at 0.2 C. However, SnS2/C composite as a contrastive electrode showed worse rate performance, achieving the discharge capacities of 718.2, 609.7, 546.7, 467.7, 297 and 116 mAh g−<sup>1</sup> at 0.2, 0.5, 1, 2, 5 and 8 C, respectively. **Figure 5E** shows the cycling properties of SnS2/C and SnS2/rGO electrodes at 1.2 A g−<sup>1</sup> . Obviously, for SnS2/rGO electrode, a higher discharge capacity of 514 mAh g−<sup>1</sup> can be obtained after 300 cycles. Compared with SnS2/rGO electrode, the SnS2/C electrode showed a worse cycle performance. The cycle performance of SnS2/rGO electrode in this work and other literatures were

summarized in **Table 1**. It is found that SnS2/rGO electrode in this work exhibited an excellent cycle performance.

To further understand the reason for the improvement of electrochemical properties of SnS2/rGO electrode, the EIS of which was measured. **Figure 5F** displays the Nyquist plots of SnS2/rGO and SnS2/C electrodes. The curves exhibited similar shapes, including a semicircle and a straight line in the high frequency region and the low frequency region, respectively. The inset figure of **Figure 5F** is an equivalent circuit model, which is the fitting result of the Nyquist plots. R<sup>1</sup> is the ohmic resistance of the electrolyte and electrode; Rct is the charge transfer resistance, the value of which is the sum of R<sup>2</sup> and R3; Z<sup>ω</sup> is the Warburg impedance; CPE represents the double layer capacitance and passivation film capacitance (Zhang et al., 2017b; Tong et al., 2018). R<sup>1</sup> of SnS2/rGO and SnS2/C electrodes was 6.4 and 6.2 , respectively; Rct of SnS2/rGO was 79.7 , which is much smaller than that of the SnS2/C electrode (105.2 ). The decrease of Rct in SnS2/rGO electrode implies that the charge transfer process was successfully facilitated by rGO nanosheet introduction (Xu et al., 2016b).

The sodium energy storage performance was also investigated and shown in **Figure 6**. The CV for the first three cycles of SnS2/rGO electrode was shown in **Figure 6A**, which was tested between 0.01 and 3.00 V. The electrochemical reactions described in Equations (3) and (4) (Zhang et al., 2015c). The peak at about 1.6 V corresponds to the intercalation of sodium into SnS<sup>2</sup> layers and the formation of NaxSnS<sup>2</sup> during the first cathodic scan (Qu et al., 2014; Wang et al., 2015b; Chao et al., 2016). The peak at about 0.45 V is ascribed to the conversion and alloying reactions. And the peak was shifted to about 0.6 V in the second and third cycles. It is due to the formation of solid electrolyte interface (SEI) film. In the anodic scans, there were three peaks at about 0.3, 0.75, and 1.3 V, which correspond to the formation of metal Sn, NaxSnS2, and SnS2, respectively. The second and the third cycles were almost the same, which suggests that SnS2/rGO electrode possesses a good reversibility during the sodiation and desodiation process.

$$\text{SnS}\_2 + 4\text{ Na} \rightarrow \text{Sn} + 2\text{Na}\_2\text{S} \tag{3}$$

$$\text{Sn} + \text{3.75 Na} \rightarrow \text{Na}\_{3.75}\text{Sn(Na}\_{15}\text{Sn}\_{4})\tag{4}$$

The charge and discharge curves are shown in **Figure 6B**. The charge and discharge tests were carried out at the current density of 50 mA g−<sup>1</sup> . In the initial discharge process, a short plateau

EIS results fitting.

appeared about 1.7 V, which corresponds to the formation of NaxSnS<sup>2</sup> by sodiation into SnS<sup>2</sup> layers. A tilted plateau from 0.8 to 0.6 V corresponds to the sodiation of more sodium ions into NaxSnS2, and the formation of metal Sn and Na2S. During this process, SEI film was formed at the same time (Luo et al., 2012). In the plateau below 0.6 V, a reaction occurred between metal Sn and sodium ions to form NaxSn (in theory the x is less than 3.75). The discharge and charge capacities of the first cycle were 960.0 and 641.8 mAh g−<sup>1</sup> , respectively. The irreversible capacity loss of the first cycle was 318.2 mAh g−<sup>1</sup> , which is usually caused by the formation of SEI film. The charge and discharge curves of the second cycle were also shown in **Figure 6B**. The charge and discharge capacities in the second cycle were 645 and 677.1 mAh g−<sup>1</sup> , respectively. The charge curves in the first and second cycles almost overlapped, suggesting that the SnS2/rGO anode has a good electrochemical reversibility. The details of rate performance are shown in **Figure 6C**. SnS2/rGO electrode showed the discharge capacities of 645, 586, 536, 462, TABLE 1 | Rate performances of SnS2/rGO electrodes in this work and the other literatures.


387 and 320 mAh g−<sup>1</sup> at 0.05, 0.1, 0.25, 0.5, 1.0 and 2.0 A g −1 , respectively. The cycle performance was also investigated, as shown in **Figure 6D**. The electrode in the first four cycles was discharged and charged at 0.05 A g−<sup>1</sup> , and then at 0.25 A g −1 . The capacity was 405 mAh g−<sup>1</sup> after 100 cycles, and the capacity loss was 18.8%. Meanwhile, the coulombic efficiency was stable at about 99% for a long time after the first few cycles.

### CONCLUSION

In summary, SnS2/rGO nanosheet composite was synthesized by reflux condensation and hydrothermal methods. SnS2/rGO composite as an anode material showed excellent electrochemical properties for both LIBs and SIBs. The excellent sodium storage performance of the SnS2/rGO composite could be attributed to the following reasons. The nanosheet structure of SnS<sup>2</sup> can shorten the diffusion path of lithium and sodium ions. Furthermore, rGO can enhance the electronic conductivity of the composite. Therefore, the SnS2/rGO composite could be considered as a promising anode material for LIBs and SIBs.

### REFERENCES


### AUTHOR CONTRIBUTIONS

HC carried out the experiment and wrote the manuscript. HL, LM, DB, and SC participated in the experiment. BZ, JZ, WY, JZ, and ZD contributed to the discussion. HT supervised the experiment and proofread the manuscript.

### ACKNOWLEDGMENTS

This work was supported by National Natural Science Foundation of China (Grant No. 51502350, 51702367 and 51772334), China Postdoctoral Science Foundation (Grant No. 2016M592447), and The International Postdoctoral Exchange Fellowship Program (Grant No. 155212).


cathode materials for sodium ion batteries. J. Alloys Compd. 728, 976–983. doi: 10.1016/j.jallcom.2017.09.020


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Chen, Zhang, Zhang, Yu, Zheng, Ding, Li, Ming, Bengono, Chen and Tong. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

## Interfaces Between Cathode and Electrolyte in Solid State Lithium Batteries: Challenges and Perspectives

Kaihui Nie1,2, Yanshuai Hong1,2, Jiliang Qiu1,2, Qinghao Li 1,2 \*, Xiqian Yu1,2 \*, Hong Li 1,2 and Liquan Chen1,2

*<sup>1</sup> Renewable Energy Laboratory, Institute of Physics, Chinese Academy of Sciences, Beijing, China, <sup>2</sup> University of Chinese Academy of Sciences, Beijing, China*

Solid state lithium batteries are widely accepted as promising candidates for next generation of various energy storage devices with the probability to realize improved energy density and superior safety performances. However, the interface between electrode and solid electrolyte remain a key issue that hinders practical development of solid state lithium batteries. In this review, we specifically focus on the interface between solid electrolytes and prevailing cathodes. The basic principles of interface layer formation are summarized and three kinds of interface layers can be categorized. For typical solid state lithium batteries, a most common and daunting challenge is to achieve and sustain intimate solid-solid contact. Meanwhile, different specific issues occur on various types of solid electrolytes, depending on the intrinsic properties of adjacent solid components. Our discussion mostly involves following electrolytes, including solid polymer electrolyte, inorganic solid oxide and sulfide electrolytes as well as composite electrolytes. The effective strategies to overcome the interface instabilities are also summarized. In order to clarify interfacial behaviors fundamentally, advanced characterization techniques with time, and atomic-scale resolution are required to gain more insights from different perspectives. And recent progresses achieved from advanced characterization are also reviewed here. We highlight that the cooperative characterization of diverse advanced characterization techniques is necessary to gain the final clarification of interface behavior, and stress that the combination of diverse interfacial modification strategies is required to build up decent cathode-electrolyte interface for superior solid state lithium batteries.

Keywords: cathode, solid electrolyte, solid state lithium battery, cathode-solid electrolyte interface, advanced characterization

### INTRODUCTION

The daily increasing energy consumption demands advanced batteries with higher energy density and superior safety performance, particularly for large-scale applications like electric vehicles and grid storage (Tarascon and Armand, 2001). In solid state lithium batteries, conventional liquid electrolyte based on flammable carbonate components is replaced by solid electrolyte. Thereby, the safety concern related to thermal runaway and electrolyte combustion is likely to be much

#### Edited by:

*Jiexi Wang, Central South University, China*

#### Reviewed by:

*Lifang Jiao, Nankai University, China Wei Luo, Tongji University, China*

#### \*Correspondence:

*Qinghao Li lqh@iphy.ac.cn Xiqian Yu xyu@iphy.ac.cn*

#### Specialty section:

*This article was submitted to Physical Chemistry and Chemical Physics, a section of the journal Frontiers in Chemistry*

Received: *31 August 2018* Accepted: *29 November 2018* Published: *12 December 2018*

#### Citation:

*Nie K, Hong Y, Qiu J, Li Q, Yu X, Li H and Chen L (2018) Interfaces Between Cathode and Electrolyte in Solid State Lithium Batteries: Challenges and Perspectives. Front. Chem. 6:616. doi: 10.3389/fchem.2018.00616*

**73**

mitigated (Zhang et al., 2013). Owing to the mechanical properties of solid electrolyte, solid state lithium batteries could resist lithium dendrite in a great degree and the cycle life could be extended longer than lithium batteries based on liquid electrolyte. The wide electrochemical stability window of solid electrolyte may further enable the application of Li metal as anode and cathodes with even higher oxidization potential. With larger lithium chemical potential difference between anode and cathode, the energy storage can be much improved correspondingly. Owing to these glittering properties, solid state lithium batteries have attracted much research attention in recent years and become promising candidates for next generation energy storage devices with the expectations of improved safety performance, longer cycle life, and higher energy density (Bates et al., 2000; Duan et al., 2017). Depending on whether the battery contains liquid electrolyte or not, solid state lithium batteries can be divided into all solid state lithium batteries and hybrid solid liquid electrolyte lithium batteries (Cao et al., 2018).

The development of practically accessible solid state lithium batteries is hindered by two major bottle-necks. The first one is the low ionic conductivity of solid electrolyte, which is severalorders-lower than that of liquid electrolyte at room temperature (RT). Continuous research efforts have been devoted to designing superior solid electrolyte in the past decades, and much progress has been achieved so far. Up to now, the RT ionic conductivities of some systems have approached or even surpassed that of liquid electrolytes. RT conductivities of NASICON-type oxides (Aono et al., 1990; Fergus, 2012) and lithium garnets (Murugan et al., 2007) have reached ∼1 mS cm−<sup>1</sup> . Kato et al. further increased the number to ∼25 mS cm−<sup>1</sup> in Li9.54Si1.74P1.44S11.7Cl0.3 (Kato et al., 2016). Decent solid electrolytes are now available with intrinsically high Li<sup>+</sup> conductivity, Lithium ion transference number tLi<sup>+</sup> ∼1, and particularly no desolvation step compared to organic liquid electrolytes, tLi<sup>+</sup> is around 0.2 0.5 (Xu, 2004; Zugmann et al., 2011). Corresponding solid state lithium batteries are expected to exhibit large capacities and high power densities for future applications (Kato et al., 2016).

Despite the rapid development of solid electrolyte itself, the even more serious hinderance for solid state lithium batteries is the high interfacial resistance caused by poor contact and interfacial reactions (Zhang et al., 2016). Without liquid fluidity, it's challenging to obtain intimate contact between solid electrolyte and electrode. The periodic electrode expanding and shrinking during cycle further deteriorates the mechanical particle-to-particle contact. As a consequence, high polarization, and low utilization of active materials are conventional in solid state lithium batteries. Meanwhile, the high voltage instability of solid electrolyte is another noteworthy concern for solid state lithium batteries. Solid electrolytes are expected to provide wider electrochemical stability window compared with liquid electrolyte. Literatures also reported wide electrochemical window up to 5.0 V or even 6.0 V in inert electrode system. However, some computation results and experimental results confirmed that the window is not as high as reported before (Han et al., 2016). Especially, for SPE the prevailing experimental reports of which are mostly based on LiFePO<sup>4</sup> cathode cycled within 3.8 V. Considering the catalytic behavior of transition metal oxides, the practical stability of solid electrolyte at high voltage still needs further investigation, and verification. Moreover, the electric potential profile across the electrodeelectrolyte interface is still a problem unanswered, which has significant influences on interface reaction and battery performance. Many investigations have been carried out on the abrupt change of electric potential across cathode-electrolyte interface (Liang et al., 2018). And it has been pointed out that the interfacial side reactions may be accelerated dramatically due to the specific local electric potential. However, an intensive and systematic understanding is still lacking on the potential profile distribution across the interface and its corresponding influence on the interface behavior.

It can be inferred from above that the key to realize solid state lithium batteries with competitive performance mostly relies on the construction of a stable and intimate interface, where different strategies have been developed. Direct co-sintering of electrode and electrolyte may be an effective and simple method to achieve good interfacial contact. However, the high temperature facilitates ion interdiffusion across interface, leading to side reactions between the electrode and solid electrolyte. Insitu synthesis of solid electrolyte or cathode is another promising choice, but necessary sintering procedure also encounters the problem of ion interdiffusion. Due to the interfacial passivation layer formation, the dynamics performance of solid state lithium batteries may be deteriorated. Diverse strategies have been proposed to build up proper artificial interlayer, including cathode coating, interface softening, buffer layer introducing, and etc. These strategies can effectively improve the physical contact, diminish interfacial side reactions, and mitigate the space charge layer (SCL) in sulfide solid electrolyte, but corresponding solid state lithium batteries are still far from practical applications. Till now much efforts have been devoted to interface modification and progresses have been obtained, but interface property is still a major obstacle on the way to practical solid state lithium batteries.

Interface research has become a challenging but hot topic in solid state batteries (Gao et al., 2018) (Lu et al., 2018; Xu et al., 2018). The interface between lithium anode and solid electrolyte has been extensively investigated. Note that the cathode-solid electrolyte interface serves as a hinge to obtain batteries with improved safety, longer cycle life, and higher energy density. So, the interface between cathode and solid electrolyte is equally important to the interface at anode side. Here in this review, we put a special focus on the fundamental issues about cathode-solid electrolyte interfaces in solid state lithium batteries based on diverse cathode-electrolyte materials. We hope to summarize the previous understandings and recent advances on the interface research. Furthermore, we hope to shed light on the possible approach to the final understanding of interface phenomenon with advanced characterization techniques. In chapter 2, we present a brief overview on basic principle of battery operation and scientific issues relevant to interface layer. In chapter 3, the interfacial problems between cathodes and four kinds of prevailing solid electrolytes are specifically discussed, corresponding optimization methods are also introduced. In chapter 4, advanced characterization techniques used for the investigation of solid-solid interface behavior are consolidated, corresponding advances and achievements are summarized. Finally, we give a comprehensive conclusion about the cathodesolid electrolyte issues and perspectives for building favorable interfaces.

### BASIC PRINCIPLE AND ISSUES AT THE CATHODE-SOLID ELECTROLYTE INTERFACE

Solid state lithium batteries have three major components cathode, anode, and solid electrolyte. The cathode material herein refers to the same lithium-containing compound as the lithium ion battery. During charging, Li<sup>+</sup> are extracted from the cathode and migrate to anode via solid electrolyte, while electrons transfer from the cathode to anode through external circuit. In this process, oxidation and reduction reactions take place at the cathode and anode sides, respectively. During discharging, Li<sup>+</sup> and electrons migrate toward the reverse direction, accompanied with cathode reduction, and anode oxidation. The following reaction steps are involved at electrodeelectrolyte interface in solid state lithium batteries: (i) Li<sup>+</sup> diffusion in the electrolyte, (ii) Li<sup>+</sup> hop into the first lattice site of the electrode and oxidation/reduction reaction happened at the same time i.e., the charge transfer process, (iii) Li<sup>+</sup> diffusion in the electrode, and (iv) Surface reaction, etc. A stable and intimate interface is necessary to ensure the above reaction steps proceed smoothly.

Interface instability may derive from chemical or electrochemical problems, a most fundamental origin is the abrupt electrochemical potential change at electrodeelectrolyte interface. As illustrated in **Figure 1A**, the lowest unoccupied molecular orbital (LUMO) and highest occupied molecular orbital (HOMO) of electrolyte determines the electrochemical stability window of solid state lithium batteries. The electrochemical potential of anode and cathode is marked as µ<sup>A</sup> and µC, which need to match with the electrochemical window of electrolyte to achieve thermodynamic stability (Goodenough, 2013). **Figure 1B** shows the ionic and electronic structures of electrode and electrolyte before and after contact, where electrolyte exhibits higher Li<sup>+</sup> chemical potential. Li<sup>+</sup> will migrate from solid electrolyte to oxide cathode to achieve thermodynamic equilibrium. This will lead to the alignment of Li<sup>+</sup> electrochemical potentials and a space charge layer (SCL) formation with an inner electric field after contacted. While band bending and alignment of Fermi level will happen due to the formation of a heterojunction. As a result, original position of energy levels, inner electric field formation, band bending as well as the energy levels change during charging/discharging determine barriers for charge carriers transfer. From the barriers for charge carriers transfer, whether electrons/holes could transfer at the interface i.e., the oxidation/reduction of the electrolyte, could be concluded. (Hausbrand et al., 2014) It was pointed out that side reactions at interface may be further accelerated due to the large polarization from electric potential (ϕ) drop (Ohta et al., 2006; Zhou et al., 2016). In such cases, solid electrolyte decomposition and intermediate transition layer formation may take place at interface. In addition, conventional high temperature processing may further induce interfacial interdiffusion of TM (transition metal) elements and favor the formation of specific transition region.

Based on the intrinsic properties of different kinds of solid electrolytes and cathode materials, there are mostly three types of electrode-electrolyte interfaces in solid state lithium batteries, as shown in **Figure 1C** (Zhu et al., 2016). Type 1 is a stable interface scenario with no electrolyte decomposition or chemical side reactions. This is the ideal interface, which seldomly appears in practical systems. Type 2 represents interface which is electronic insulating but provides Li<sup>+</sup> migration channels. Within this scheme, further interfacial side reactions can be suppressed and battery operation can be maintained. The LiCoO2/LiPON interface may be a proper example for this case (Zhu et al., 2016). Type 3 is an undesirable but most common interface with mixed ionic and electronic conductivity. In this scheme, continuous side reactions occur, and battery fade happens, as in LiCoO2/LGPS interface. Depending on the intrinsic property of electrode and electrolyte, different types of interfaces will be built up, but only type 1 and 2 are accessible for practical applications. By introducing proper buffer layers between cathode and electrolyte, a stable artificial layer can be constructed and convert interface from type 3 to type 2. Considering the significance of building proper Li<sup>+</sup> conducting layer and balancing interfacial potential drop, we will present detailed discussion in following chapters according to the characteristics of specific solid electrolytes.

Apart from the chemical stability of the interface, mechanical behavior also has a significant impact on battery performance. In conventional lithium ion batteries based on liquid electrolyte, cathode particles can be totally immersed in liquid electrolyte and passivation layer called solid electrolyte interphase (SEI) may form. Good contact between electrode and liquid electrolyte could therefore be maintained throughout battery cycle, **Figure 2A** (Liu et al., 2006; Takamatsu et al., 2012). However, it is challenging to maintain intimate electrode-electrolyte interface in solid state lithium batteries, especially over many cycles (Goodenough, 2013). The deficient contact in solid state lithium batteries may well-lead to low utilization of active particles, large polarization and even contact loss during cycle.

Due to the distinguished mechanical properties, there is distinct difference in contact behavior among various types of electrolytes. Solid electrolytes can be generally classified into SPE and solid inorganic electrolyte, the latter can be further classified into solid oxide and solid sulfide electrolyte. Polymer electrolyte has moderate contact with cathode due to the elasticity and deformability of organic polymers. Nevertheless, vacant cavities will still generate due to interface reactions and cathode pulverization during cycling (**Figure 2B**) (Nakayama et al., 2010). The effective contact area between cathode and polymer electrolyte will consequently reduce with battery cycle. Due to reasonable mechanical ductility, deformable sulfide particles could also change its shape to match with cathode particles. Hence, the poor contact between electrode and sulfide electrolyte can be much improved by mechanical pressing

types of the solid electrolyte/solid electrode interfaces.

(**Figure 2C**) (Sakuda et al., 2013; Ito et al., 2017). While contact loss will also happen upon cycling along with the shrinkage and expansion of cathode particles (Koerver et al., 2017). Solid oxide electrolytes have the worst point-contact with cathode due to the rigid ceramic nature (**Figure 2D**) (Ohta et al., 2013, 2014). The insufficient mechanical contact facilitates cathode particles completely isolated from solid electrolyte, i e., the "dead" area. Due to the lack of percolation paths, neither electrons nor Li<sup>+</sup> can be transferred from/into the dead areas. The "dead" areas not only lead to direct capacity fading, but also induce locally strong non-uniform current and strain distribution (Zhang et al., 2018). The poor solid-solid contact typically brings about large polarization and low capacity. To improve the interface contact, various strategies have been adopted, such as in-situ synthesis of solid electrolyte, interface buffer layer, cathode coating, gel system etc. Based on different properties of various electrolytes, specific strategies will be adopted, and introduced specifically in Chapter 3.

### CHALLENGES AND SOLUTIONS ON INTERFACES BETWEEN CATHODE AND DIVERSE SOLID ELECTROLYTES

### Interface Between Cathode and Solid Polymer Electrolyte

After Wright's discovery of alkali metal ions conductivity in poly(ethylene oxide) (PEO) in 1973 (Fenton et al., 1973), Armand firstly proposed PEO with lithium salt as solid electrolyte for solid state lithium batteries (Armand, 1983). PEO-based SPE is widely accepted as a most promising candidate for solid state lithium batteries owning to its advantages such as easy fabrication, low cost and excellent compatibility with lithium salt. In SPE, Li<sup>+</sup> can migrate in the free volume of polymer host assisted by the motion of the polymer chains when temperature is above T<sup>g</sup> (glass transition temperature) (Bruce, 1995). In PEO based SPE, Li<sup>+</sup> were coordinated by ether oxygen and transport with the breaking/forming of Li-O bonds (Bruce, 1995; Xu, 2004). However, PEO-based SPE is not stable above

layer will form. (B) Cathode particles are distributed in a Li-PEO binder with good contact while voids will generate upon cycling because of the interfacial reactions or pulverization of cathode particles. (C) Sulfide particles have favorable mechanical properties as ductility and deformability, which could change its shape to match with the rigid solid electrode. (D) Solid oxide electrolyte: Poor point-contact will form due to the rigid ceramic nature. Interface layer will form in all the aforementioned system if decomposition reactions or interdiffusion occurred at the interface.

4.0 V, which confines the pairing cathode operating within lowvoltage range. In most literature reports, the prevailing choice is LiFePO<sup>4</sup> (Croce et al., 2001). Considering the catalytic effect of transition metal oxides, PEO decomposition may well-be triggered at the interface region. And improving the antioxidant properties of SPE to high-voltage range is essential to realize high energy density solid state lithium batteries based on PEO (Fan et al., 2002). There are mostly three types of SPE available (i) dry solid-state polymer, (ii) gel/plasticized polymer electrolyte, and (iii) polymer composites (Manuel Stephan, 2006). Diverse optimization strategies have been utilized for different SPE systems, as discussed below.

"Dry" solid-state polymer refers to PEO, PPO, PAN etc. and their derivatives containing Li salt, corresponding optimization has been focused on polymer architecture and Li salt selection. On the electrochemical instability of PEO-based SPE above 4.0 V (Croce et al., 2001), Nakayama et al. (2010) proposed a model that two sequential factors affect cyclic degradation. The first one is the local current enhancement induced by cathode pulverization, which can be attributed to the solid-solid contact between electrolyte and electrode. The second is continuous and uneven decomposition of TFSI [N (CF3SO2) − 2 ] due to local polarization. These results indicate that the mechanical property of SPE and the Li salt selection are both essential. Ma et al. proposed a novel SPE composed of PEO and extra-stable lithium salt- (Li[(CF3SO2)(n-C4F9SO2)-N], LiTNFSI). This SPE exhibits a homogeneous and compact morphology and high electrochemical stability at ∼4.0 V vs. Li+/Li (**Figure 3A**). Longterm cycling stability and sufficient thermal stability (>350◦C) was also obtained in this novel SPE (Ma et al., 2016b).

Improving the antioxidative capability of PEO is another critical aspect to promote high voltage interface stability. Copolymerization, branching, and crosslinking are common polymer modification methods, which also favor designing more antioxidative polymers (Tong et al., 2014; Porcarelli et al., 2016; Wang et al., 2016a). UV-induced (co)polymerization can promote effective interlinking between polyethylene oxide (PEO) chains plasticized by tetraglyme. Hereby, Porcarelli et al. (2016) obtained SPE with wide electrochemical stability window (>5 V vs. Li/Li+) from LSV test. **Figure 3B** illustrates the synthesis process and photograph of as-prepared SPE, together with cross-sectional FESEM images of optimized cathode-electrolyte interface. It is clear that electrolyte creates conformal coating by following the contours of active particles, which leads to improved active materials utilization. Similar optimized SPE have also been achieved by a PEO and liquid-crystalline copolymers with small molecular liquid crystals as fillers. High ionic conductivity, lithium ion transference number combined with wide electro-chemical stability window of the copolymer facilitated a good electrochemical performance (Tong et al., 2014).

Except for developing PEO derivatives, exploiting other antioxidative polymer electrolyte has also attracted much attention (Zhang et al., 2015; Chai et al., 2017). Chai et al. (2017) prepared a kind of novel poly (vinylene carbonate) (PVCA) based SPE which possessed both interfacial compatibility with Li anode and high-voltage LiCoO<sup>2</sup> cathode (4.3 V vs. Li/Li+). From in-situ polymerization of PVCA, polymer electrolyte can be even incorporated into the porous cathodes and the effective contact area can be much increased as a result. Owing to the good contact and compatibility, the battery exhibited high discharge capacity and excellent cycling performance.

The second category SPE is called "gel polymer electrolyte" or "plasticized polymer electrolyte" which contains both liquid and solid components. Thus, gel polymers possess both cohesive properties of solids and the diffusive property of liquids (Manuel Stephan, 2006) which makes hybrid solid liquid electrolyte lithium batteries have unique advantages (Huang et al., 1996). As reported, by modifying PEO electrolyte with plasticizing liquid dimethyl sulphoxide (DMSO) (<5%), electrochemical stability window can be extended above 4.1 V, exceeding Fermi levels of several high voltage cathodes. The long terms cycling stability was also improved obviously (Zewde et al., 2018). By phase inversion technique, Deng et al. prepared a microporous polymer electrolyte (MPE) based on poly (vinylidene fluoride) (PVDF)/PEO star polymer, which exhibits wide electrochemical stability window of ∼5V (Deng et al., 2015).

Solid composite electrolyte, as SPEs subset, is another competitive candidate among kinds of SPEs. Solid composite electrolyte combined the virtues of both polymer and ceramic, exhibiting excellent mechanical stability, high ionic conductivity, wide electrochemical stability window, intimate contact performance, and etc. Relevant researches will be specifically introduced in Chapter 3.4.

Apart from solid electrolyte modification, cathode surface modification is another effective way to mitigate the interface degradation. Note that a principle factor that restricts PEO application at high voltage is the strong oxidation/catalytic property of TM oxides cathode. Consequently, surface modification on cathode material becomes another way to enable PEO operation at high voltage. Yang et al. synthesized continuous and compact LATP coating layer on LiCoO<sup>2</sup> through solution-procession and low temperature treatment. solid state lithium batteries assembled with PEO-based SPE and LATP modified LiCoO<sup>2</sup> shows high capacity retention (93.2% after 50 cycles) at 4.2 V, which suggests that surface coating can effectively suppress PEO oxidation at high voltage (Yang et al., 2018). By cathode coating, PVCA-coated LiCoO<sup>2</sup> also showed much enhanced cycling stability of PEO based SPE at 4.45V (Ma et al., 2017).

### Interface Between Cathode and Solid Oxide Electrolyte

Oxide-based solid electrolytes exhibit good chemical stability against air and compatibility with high voltage cathodes. Typical solid oxide electrolytes include garnet-type Li7La3Zr2O<sup>12</sup> (LLZO), NASICON-type LiTi2(PO4)3, LiSICON-type Li14Zn (GeO4)4, and Perovskite-type La0.5Li0.5−δTiO<sup>3</sup> (LLTO). Solid oxide electrolyte is a most competitive choice for solid state lithium batteries (Chen et al., 1980; Delmas et al., 1988; Inaguma et al., 1993). However, there are two major challenges for solid oxide electrolytes. The first one is the generally low ionic conductivity, which is lower than sulfide electrolytes. Despite the phenomenally low intrinsic bulk conductivity, recent investigations point to the high interface polarization that restrains battery dynamics. The second challenge is the rigid ceramic nature, which causes poor point-contact at electrodeelectrolyte interface, as discussed above. Solid oxide electrolytes have a key advantage of intrinsic wide electrochemical window. For garnet-type electrolytes, the experimental value can be even wide as ∼0–6 V (Li et al., 2015; Thangadurai et al., 2015). Among all solid oxide electrolytes, garnet-type electrolyte is an attractive candidate due to its high RT ionic conductivity (∼1 mS cm−<sup>1</sup> ), practically wide electrochemical window and chemical stability against Li etc. In the following discussion, we mostly take garnet LLZO as a typical example to discuss the interfacial problems solid oxide electrolyte faced with, other systems are briefly mentioned at the end.

Since interfacial resistance from poor contact (**Figure 2D**) is proven to be the main reason for the high internal resistance of solid state lithium batteries (Park et al., 2016; Han et al., 2017), quite a few approaches have been applied to reduce interfacial resistance, including co-sintering (Wakayama et al., 2016), in-situ synthesized electrolyte layer (Yoshima et al., 2016; Kazyak et al., 2017), interface buffer layers (Kato et al., 2014; Park et al., 2016), interface softening (Seino et al., 2011; Sakuda et al., 2012; Liu et al., 2016), surface coating (Han et al., 2017), and amorphous cathode (Matsuyama et al., 2016; Nagao et al., 2017) etc.

Electrode-electrolyte co-sintering, cathode layer in-situ synthesizing, and thin film deposition are proven effective in promoting surface contact, however necessary high-temperature handling (>500◦C) will lead to elements interdiffusion, electrolyte decomposition and deteriorated performance (Wakayama et al., 2016). Tremendous efforts have been devoted to lower the sintering temperature of solid oxide electrolytes to mitigate the interdiffusion problem, while very finite progress has been achieved so far. With a combination of ab initio calculations, thermal analysis, and X-ray-diffraction, Ceder' group elucidated the decomposition reactions between highvoltage spinel cathode (Li2NiMn3O8, Li2FeMn3O8, LiCoMnO4) and solid oxide electrolyte (LLZO, LATP) (Miara et al., 2016). XRD revealed that spinel cathode and LLZO were not compatible with each other at 600◦C, and the decomposition products can be predicted from calculated phase diagrams. In 2016, Park et al. (2016) studied the three-dimensional elemental distribution at LiCoO2/LLZO interface by TOF-SIMS. As illustrated in **Figure 4A**, Co diffuses into LLZO, and Zr/La diffuses into LiCoO2. While Al was leached out of LLZO, and diffuses into LiCoO2, cubic LLZO at the interface transformed to tetragonal phase. It was further proved that interface modification with Li3BO<sup>3</sup> can reduce chemical cross-contamination and improve physical bonding.

Introducing interfacial buffer layer, such as Nb, LiNbO2, BaTiO<sup>3</sup> was found to be an effective way to mitigate interface interdiffusion (Kato et al., 2014). By radio frequency (RF) magnetron sputtering, Kato et al. (2014) introduced a thin Nb layer (10 nm) on LLZO and then LiCoO<sup>2</sup> was deposited on the Nb-modified LLZO by PLD at 600◦C. In-situ synthesis of LiCoO<sup>2</sup> by PLD guaranteed an intimate contact between cathode and solid electrolyte, while introducing Nb layer improved the interface performance by forming Li-Nb-O amorphous region. The Li+- conductivity of the amorphous Li-Nb-O region is high as 1 × 10−<sup>6</sup> S cm−<sup>1</sup> , which will facilitate Li<sup>+</sup> transport at interface (Glass et al., 1978). As a result, the mutual diffusion thickness is 40 nm, which is much smaller than the 100 nm Li<sup>+</sup> insulating La2CoO<sup>4</sup> region without Nb modification (Kim et al., 2011). Kazyak et al. presented a thermal ALD (atomic layer deposition) process which can significantly lower the formation temperature of the cubic phase to 555◦C. The schematic processes and SEM images of LLZO products are illustrated in **Figure 4B**. Low melting compounds were also employed for good interfacial

[Reprinted with permission from (Ma et al., 2016b). Copyright (2016) American Chemical Society]. (B) Interconnected PEO chains with hypothesized branched clusters of tetraglyme oligomers (top left) and the real aspect of a freshly prepared ISPE (top right); Cross-sectional FESEM images showing the optimum interface achieved after UV curing (down). [Reprinted with permission from (Porcarelli et al., 2016). Copyright (2016) Nature].

contact in solid state lithium batteries. Liu et al. used Li3BO<sup>3</sup> (melting point ca. 700◦C, Li<sup>+</sup> conductivity ca. 2 × 10−<sup>6</sup> S cm−<sup>1</sup> ) as bonding aid for LCO/LLZO interface. Corresponding interface resistance reduced dramatically and electrochemical performance improved significantly. Yoshima et al. (2016) designed a gel to soften interface and achieved intimate contact at interface. LLZO coated with polyacrylonitrile (PAN)-based gel was prepared as electrolyte sheet, which reduced internal resistance of the whole battery. The assembled solid state lithium batteries exhibited good rate capability and cycling stability between −40 and 80◦C. Very recently, Han et al. reported an all-ceramic cathode-electrolyte by thermally soldering LiCoO<sup>2</sup> and LLZO together with Li2.3−xC0.7+xB0.3−xO<sup>3</sup> solid electrolyte interphase can be spontaneously coated on both LLZO and LiCoO<sup>2</sup> (Han et al., 2018). The simultaneous improvements in interfacial contact, (electro) chemical stability, ionic conductivity, and mechanical property of the all-ceramic cathode-electrolyte enabled an all solid state Li/LLZO/LiCoO<sup>2</sup> battery with extremely high electrochemical performance.

Owing to the rigid ceramic nature, most solid oxide electrolytes face similar interfacial challenges when paired with solid cathode. The aforementioned interface modifying strategies can also be applied to diverse solid oxides electrolytes, except that typical solid oxide electrolyte are further hindered by other factors. Perovskite-type LLTO was firstly prepared by Inaguma et al. (1993) which exhibits low ionic conductivity across grain boundary (around 10−<sup>5</sup> S cm−<sup>1</sup> ) and poor stability with anode (instable below 1.8 V vs. Li+/Li). As a result, most works on LLTO focus on improving ionic conductivity and chemical stability vs. Li anode (Chen and Amine, 2001; Kotobuki et al., 2011; Huang et al., 2016). Li1+xAlxGe2−x(PO4)<sup>3</sup> (LAGP) and Li1+xAlxTi2−x(PO4)<sup>3</sup> (LATP) are two common NASICON-type solid oxide electrolyte. Because of Ti4<sup>+</sup> reduction, LATP suffers redox reaction at 2.5 V vs. Li+/Li. Although LATP shows high ionic conductivity (Delmas et al., 1988), its incompatible with low potential anodes, especially Li, confines its application in solid state lithium batteries. In these solid oxide electrolytes related research, interface softening, and in-situ synthesizing have also been carried out (Kim et al., 2017; Zhang et al., 2017b) and corresponding investigations are still in progress.

### Interface Between Cathode and Solid Sulfide Electrolyte

Solid sulfide electrolytes are the derivatives of solid oxide electrolytes by substituting oxygen with sulfur. Since the electronegativity of S is less than O, Li<sup>+</sup> binding energy is smaller and Li<sup>+</sup> can move more freely. Among all solid electrolytes, solid sulfide electrolyte exhibits the highest Li<sup>+</sup> conductivity.

Another attractive feature of solid sulfide electrolyte is their mechanical property. These materials exhibit plastic deformation under mechanical pressure, and this softness makes it possible to prepare densely packed interface (Koerver et al., 2017). In recent years, the research focus of solid sulfide electrolyte is Li2S-P2S<sup>5</sup> based systems, which exhibit superior Li<sup>+</sup> conductivity, electrochemical stability, and mechanical properties. According to the composition difference, Li2S-P2S<sup>5</sup> system can be divided into binary solid sulfide electrolyte (composed of Li2S and P2S5, such as Li3PS4, Li7P3S11) and ternary solid sulfide electrolyte (composed of Li2S, P2S5, MS2, M = Si, Ge, Sn, such as Li10GeP2S12). According crystallinity difference, the two kinds of SSEs can be further divided into glass, glass ceramic, and ceramic form solid electrolyte, which exhibit different performance in terms of ionic conductivity, chemical stability, and contact with solid electrode.

Owing to Li<sup>+</sup> chemical potential difference between oxide cathode and solid sulfide electrolyte, Li<sup>+</sup> may migrate from electrolyte to cathode, resulting in SCL at both sides. Due to the mixed ionic and electronic conductivity in oxide cathode, Li<sup>+</sup> gradient concentration can be compensated at cathode side. However, SCL will remain at electrolyte side due to the single ionic conductivity of electrolyte. The resulting SCL can wellimpede Li<sup>+</sup> transport and induce high polarization. SCL was firstly proposed by Wagner (1972) and extensively investigated on conduction type and conductivity change of composite materials, polycrystalline and heterojunctions (Liang, 1973; Maier, 1995; Bhattacharyya and Maier, 2004). With theoretically calculation, Haruyama et al. (2014) elucidated the characteristics of SCL between LiCoO<sup>2</sup> and β-Li3PS4, the effect of LiNbO<sup>3</sup> buffer layer interposition was also clarified. DFT calculation further revealed Li+-preferred adsorption at oxygen bridge sites, e.g., CoO6, and on Li layer, which may be the origin of deformed interface and SCL. Li chemical potentials based on vacancy formation energy indicate that the subsurface Li in sulfide electrolyte may transfer under electric field, suggesting that SCL grows immediately at the beginning of charging (**Figure 5**). Since the attractive sites on LiCoO<sup>2</sup> surface disappear with insulating LiNbO<sup>3</sup> layers attachment, the SCL at this interface is significantly suppressed. This result consistently explained SCL at atomic-scale and clearly indicated the effect of buffer layers. To eliminate SCL at the oxide cathode/SSE interface, oxide layer with high Li<sup>+</sup> conductivity, and chemical stability (mostly LiNbO3) is always introduced at interface and combined with other modification techniques (Kitaura et al., 2011; Haruyama et al., 2017; Koerver et al., 2017).

Even though solid sulfide electrolyte has moderate physical deformability, electrochemically driven mechanical failure also contributes to interfacial resistance increase and capacity fading. Koerver et al. evaluated the interfacial behavior for solid state lithium battery using nickel-rich NCM-811 cathode and β-Li3PS<sup>4</sup> solid electrolyte (Koerver et al., 2017). Results suggest that the majority of passivating layer is developed during the first charge and present slow growth upon further cycling. It was further found that electrode-electrolyte contact lose occurs in first charging due to electrochemical contraction. The mechanical failure even deteriorates in the following cycles (**Figure 6A**) and lead to high polarization and capacity decay. In order to achieve and sustain intimate interface contact, different methods, and strategies have been developed and investigated. Sticking the supercooled liquid state of electrolyte on active material particles combined with a hot press was used to achieve an intimate electrode-electrolyte interface (Kitaura et al., 2011). In contrast to the interface formed by RT pressing, hot pressing at around T<sup>g</sup> may well-obtain intimate contact along with an interfacial layer between LiCoO<sup>2</sup> and 80Li2S·20P2S<sup>5</sup> glass electrolyte. LiNbO<sup>3</sup> coating layer can be further introduced to suppress the reaction of LiCoO<sup>2</sup> with the 80Li2S·20P2S<sup>5</sup> glass solid electrolyte. Yao et al. (2016) reported a general interfacial architecture, i.e., Li7P3S<sup>11</sup> electrolyte particles (around 10 nm) anchored on cobalt sulfide nanosheets, by insitu liquid-phase approach. The STEM-EDS elemental mapping of an individual nanocomposite in **Figure 6B** confirms that the cobalt sulfide–Li7P3S<sup>11</sup> nanocomposites are homogeneously distributed throughout the nanosheets and have an intimate contact. The obtained intimate contact contributed to an excellent rate capability and cycling stability. Similar intimate contacts could be achieved by sulfide electrolyte coating onto active materials to form a favorable interface (Ito et al., 2017). By Mixing LiCoO<sup>2</sup> particles with different grain sizes during the electrolyte coating process, higher packing density pellets with less voids were obtained both before and after cycling which ensured fine networks of ionically conductive pathways. Moreover, Oh et al. (2016) discovered continuous LGPS decomposition at LGPS/acetylene black (AB) interface above 4.5 V. The decomposition layer could also isolate the delithiated LixNi0.5Mn1.5O<sup>4</sup> (x∼0) from Li<sup>+</sup> and/or electron conduction channels in cathode composite, resulting in contact loss, and severe capacity fading upon cycling. The research demonstrates that suitable conductive additive and sulfide solid electrolyte are crucial to overcome the poor cycle performance of high-voltage solid state lithium batteries. Yoon et al. (2018) also investigated the interface between Li10GeP2S<sup>12</sup> and diverse carbon conductive agents in solid state lithium batteries and confirmed the solid electrolyte decomposition and surface degradation during cycle.

### Solid Composite Electrolyte Design and Interface Optimization

Solid composite electrolyte is a subset of polymer electrolytes by dispersing electrochemically inert fillers, such as Al2O<sup>3</sup> and TiO<sup>2</sup> nanoparticles or inorganic solid electrolyte into polymer electrolyte. (Weston and Steele, 1982; Croce et al., 1998, 1999) These composite electrolytes have excellent mechanical stability (due to the ceramic fillers into polymer network) and high ionic conductivity (promoted by the high surface area of the dispersed fillers). Due to the absence of liquid components and interfacial stabilizing action from dispersed fillers, composite electrolyte offers wide electrochemical stability window (Croce et al., 1999). With advantages such as high ionic conductivity, wide electrochemical stability window, favorable interface mechanical properties, composite electrolytes have attracted extensive attention.

The inorganic fillers in solid composite electrolytes could be oxides without Li<sup>+</sup> conducting ability, such as Al2O3, TiO2, SiO2, etc. (Lin et al., 2016; Pal and Ghosh, 2018) and other solid electrolytes, such as LLZO, LAGP, LGPS, etc. (Zhao et al., 2016; Chen et al., 2017, 2018a,b; Zhai et al., 2017). In 2016, Lin et al. introduced a novel in-situ synthesis of a SiO<sup>2</sup> filler inside PEO polymer. Much stronger chemical/mechanical interactions between SiO<sup>2</sup> nanospheres and PEO chains can be obtained, which significantly suppresses PEO crystallization and facilitates polymer segmental motion for Li<sup>+</sup> conducting (Lin et al., 2016). Two possible interaction mechanisms are shown in **Figure 7A**, including chemical bonding between PEO chains and hydroxyl groups on MUSiO<sup>2</sup> and mechanical wrapping of PEO chains during MUSiO<sup>2</sup> spheres growth. At the same time, electrochemical stability window can be largely extended up to 5.5 V, much wider than ex-situ CPE and ceramic-free CPE (**Figure 7A**). The improvement of electrochemical stability indicates that the adsorption effect on anion is much stronger in in-situ CPE, which suppresses anodic decomposition at high potential (Park et al., 2003).

Garnet type LLZO electrolyte, has been widely studied as kind of filler in PEO matrix (Zheng et al., 2016; Chen et al., 2017, 2018a; Zhang et al., 2017a). Recently, Chen et al. (2018a) proposed a synergistic-composite approach to fabricate flexible solid state lithium batteries using PEO-based composite cathode layers (filled with LiFePO<sup>4</sup> particles) and composite electrolyte layers (filled with Al-LLZTO particles) which exhibits a wide electrochemical stability window ∼6 V, much wider than pure PEO. The all-composite approach is favorable for improving both mesoscopic and microscopic interfaces (**Figure 7B**) inside solid state lithium batteries and may provide a new toolbox for solid state lithium batteries design and fabrication. The interface

between composite cathode and composite electrolyte layers may keep its structural integrity albeit the large volumetric change during cycling (Chen et al., 2017). Since the synergeticcomposite electrolyte combines the virtues of two components, compositing stands a chance in building favorable interfaces, and further realizing high energy density solid state lithium batteries. Chen et al. (2018a) prepared composite ceramic/polymer solid electrolyte containing garnet Li6.4La3Zr1.4Ta0.6O<sup>12</sup> (LLZTO), PEO, and Lithium bis(trifluoromethanesulfonyl)imide (LiTFSI). The composite solid electrolyte possesses high self-standing and flexibility, which exhibits electrochemical stability window up to 5.0 V vs. Li/Li+. The assembled solid-state LiFePO4|Li batteries with electrolytes from "ceramic-in-polymer" to "polymer-in-ceramic" exhibit excellent cycling stability and wide electrochemical stability window (more than 5.0V vs. Li+/Li). The "ceramic-in-polymer" electrolyte exhibits a greater flexibility (**Figure 7C**) and lower cost, while "polymerin-ceramic" electrolyte shows higher mechanical strength

FIGURE 7 | (A) Schematic figures showing the procedure of *in situ* hydrolysis and interaction mechanisms among PEO chains and MUSiO2 (up) and the electrochemical stability windows curves of three kinds of solid electrolyte. [Reprinted with permission from (Lin et al., 2016). Copyright (2016) American Chemical Society]. (B) The sketch map SEM image of the interface between Al-LLZTO/PEO composite cathode containing 15 wt% polymer and the composite electrolyte. The curves refer to first three galvanostatic charge and discharge curves. [Reprinted with permission from (Chen et al., 2018a). Copyright (2018) American Chemical Society]. (C) Schematic illustration for PEO-LLZTO solid composite electrolyte: (a) "ceramic-in-polymer"; (b) "intermediate"; (c) "polymer-in-ceramic"; the typical surface morphologies and flexibility of composite electrolyte (1-x) wt%[PEO8-LiTFSI]-x wt% LLZTO: (d,g) 10 wt%; (e,h) 50 wt%; (f,i) 80 wt%; the liner sweep voltammograms for different compositional solid composite electrolytes at 55◦C with a scan rate of 1 mV s−<sup>1</sup> . [Reprinted with permission from (Chen et al., 2017). Copyright (2017) American Chemical Society].

and safety but brittler for bend cracks formation. Hence, by varying the composition of composite electrolyte, different properties will be obtained for special applications. Apart from LLZO, other oxide solid electrolyte, such as LATP and LAGP are also widely combined with PEO in order for better composite solid electrolyte (Wang et al., 2017b; Zhai et al., 2017).

Compared with TM oxide particles, solid sulfide electrolyte, such as LGPS and Li3PS4, exhibits fast ionic conductivity (Han et al., 2011; Haruyama et al., 2014; Kato et al., 2016). As a result, solid sulfide electrolyte incorporate into PEO matrix can provide excellent Li<sup>+</sup> conducting channel. Zhao et al. fabricated SPE membranes comprised of LGPS and PEO matrix. The optimal composite membrane exhibits high ionic conductivity ∼1.21 × 10−<sup>3</sup> S cm−<sup>1</sup> at 80◦C and wide electrochemical window of 0–5.7 V (Zhao et al., 2016). Instead of simply mixing ceramic particles with polymers, the same group prepared PEO/Li3PS<sup>4</sup> hybrid polymer electrolyte via new in-situ approach (Chen et al., 2018b). The optimal electrolyte of PEO-2% vol Li3PS<sup>4</sup> presents the highest Li<sup>+</sup> conductivity and widest electrochemical window. In anodic process, in-situ prepared electrolyte shows no anodic current until 5.1 V, while corresponding voltage for mechanicalmixed electrolyte and PEO are 4.9 V and 4.6 V, respectively, as reported in their article. The differences in ionic conductivity and stability may originate from more homogeneous dispersion of fillers in PEO by in-situ preparation than by mechanical-mixing.

According to the mechanical properties including flexibility, deformability, and strength of the aforementioned four kinds of electrolytes, the different interface performance and modifications at cathode side are summarized in **Table 1**.

### ADVANCED SOLID-SOLID INTERFACE CHARACTERIZATION TECHNIQUES

As discussed above, the interface behaviors play an important part in determining the final solid state lithium battery parameters and performances, including internal resistance, kinetic response, and cycle stability etc. However, the buried solid-solid interfaces in solid state lithium batteries are extremely difficult to investigate directly, and present knowledge on interfacial reactions and interfacial kinetics is still deficient. As a result, it is increasingly important and urgent to develop novel characterization techniques for more detailed understanding into the interface behavior (Hu et al., 2006). Note that the solid electrolyte-cathode interface involves several aspects correlated with each other, including lattice structure, electronic band structure, and chemical/electrochemical stability. The dynamic Li shuttle back and force across the interface further makes the interface behavior more complicated within time and voltage domain. Hence, advanced characterization techniques with in-situ and atomic-scale resolution are strongly necessary to gain more insights into the complex interface processes (Zheng et al., 2014; Lin et al., 2017). So far, diverse advanced characterizations have been utilized by different research groups worldwide and significant information have been obtained. These research works provide valuable perspective of interface performance and have profound guiding significance for designing more favorable interfaces in superior solid state lithium batteries.

Since a significant reason for interfacial instability is the abrupt change of electric potential across the cathode-electrolyte interface, dynamic observation of the potential profiles would help identify sources of typically large interfacial resistance. Ogumi' group contributed a lot in studying the potential distribution and interface stability mechanisms in solid state lithium batteries (Yamamoto et al., 2010, 2012; Okumura et al., 2011). With this objective, EH (quantitative electron holography) combined with EELS (electron energy loss spectroscopy) was used to directly observe the potential distribution at the LiCoO2/Li1+x+yAyTi2−ySixP3−xO<sup>12</sup> interface (Yamamoto et al., 2010). Results showed the Li<sup>+</sup> and electron typical distribution of the measured potential near the cathode-electrolyte interface during charging and the origin of the shift of the electronic band structures. This research identified the sources of reaction resistance and kinetic factors in solid state lithium battery. EH also clearly show how the metallic lithium is formed inside the solid electrolyte during the initial charging process of the solid state lithium battery (Yamamoto et al., 2012). Results showed that the smooth potential distribution at the electrode/solid electrolyte interface leads to the low interfacial resistance.

With the unique sensitivity and operability in SCL detection, AFM was performed to better understand the potential distribution at the cross section of particles (Liang et al., 2018). By introducing LATP coating at cathode surface, the LiNi0.6Co0.2Mn0.2O2/ poly(ether-acrylate) (ipn-PEA) interface realized a mitigated polarization and excellent kinetic performance. The significantly improved interface dynamics is attributed to a weakened SCL or a gradual slope of potential formed at the interface, verified by AFM interfacial potential measurements, which relieves polarization, alleviates side reactions, and enhances cycling stability and dynamic properties (**Figure 8**).

Apart from interfacial potential research techniques, advanced characterizations adopted to investigate structural and chemical stability of the interface also promote the mechanism understanding of the interface behavior. In/ex situ characterization techniques including spectroscopy, microscopy, and diffractometry are widely used to monitor the structural evolution and chemical reactions at the interface.

Spectroscopy including XAS (X-ray absorption spectroscopy), XPS (X-ray photoelectron spectroscopy), NMR (Nuclear Magnetic Resonance Imaging), etc. in an in-situ mode with high spatial resolution play important role in the solid-solid interface understanding. Okumura et al. (2011) developed a depth-resolved XAS to directly observe the chemical state and local structure at the LiCoO2/LATP interface with/without NbO<sup>2</sup> modification layer. XAS results revealed that the introduction of NbO<sup>2</sup> layer is effective for restricting the large Co–O bond change at the interface during delithiation. As a result, the charge transfer process is smoother owing to a relieved interface stress. Wenzel et al. developed in-situ XPS by using the internal argon ion gun of the instrument which was adopted to sputter a metallic target (Wenzel et al., 2015). The chemical stability of the LLTO/metallic lithium interface was investigated. The same in-situ XPS method was adopted by the same group to investigate the interface between lithium metal and Li10GeP2S12. XPS recorded the decomposition products which revealed the formation of lithium sulfide, lithium phosphide, and germanium-lithium alloy/germanium metal (**Figures 9A,B**) (Wenzel et al., 2016). In situ NMR is also widely used to study the lithium distribution in solid lithium batteries (Bhattacharyya et al., 2010; Nakayama et al., 2010; Wang et al., 2014; Romanenko et al., 2016; Chien et al., 2018). Very recently, three-dimensional <sup>7</sup>Li magnetic resonance imaging (MRI) is employed to examine Li distribution homogeneity in solid electrolyte Li10GeP2S<sup>12</sup> within symmetric Li/Li10GeP2S12/ Li batteries by Chien et al. (2018) (**Figure 9C**). The three-dimensional Li distribution revealed that the significant Li loss at interfaces is mitigated via PEO coating (**Figure 9D**) (Chien et al., 2018). This study demonstrates a powerful tool for non-invasively monitoring the


TABLE 1 | Interfacial challenges exist in cathode-solid electrolyte systems according to the different characteristics of the four types of solid electrolytes and the corresponding solutions, recent advances and limitations still exist.

Li distribution at the interfaces and in the bulk of solid state lithium batteries as well as a convenient strategy for improving interfacial stability. As mentioned in chapter 3, TOF-SIMS can also be used to study the interface element distribution around the interface in solid state lithium batteries (Park et al., 2016).

Microscopy [e.g., TEM (transmission electron microscope), STEM (scanning transmission electron microscopy), etc.] techniques are powerful tools to investigate the structural and chemical stability of solid electrode/solid electrolyte interface. Wang et al. (2016b) used in situ STEM-EELS with high spatial resolution observed the interfacial phenomena of LiCoO2/LiPON with a nanoscale resolution (**Figure 9E**). An unexpected structurally disordered interfacial layer was discovered without cycling. The interfacial layer accumulates lithium and evolves to rock salt CoO after cycling along with Li2O and Li2O<sup>2</sup> formation (**Figure 9F**) (Wang et al., 2016b). Rapid capacity decay or even cathode inactivation will happen along with the thickening of this layer. In situ STEM was also introduced to study the interface stability of lithium metal/solid electrolyte (Ma et al., 2016a). Gong et al. (2017) designed an in situ atomic-scale TEM observation of electrochemical delithiation induced structure evolution of LiCoO<sup>2</sup> cathode in solid state lithium batteries, which provides atomic-scale structure information for designing better solid state lithium batteries.

In situ diffractometry including XRD (X-ray diffraction) and ND (neutron diffraction) were also widely used to monitor the structural change upon cycling in solid state lithium batteries (Shin et al., 2014; Wang et al., 2017a; Hu et al., 2018). Wang et al. developed an in situ ND technique to monitor the Li distribution

and transport in garnet-based solid-state cells during cycling (**Figures 9G,H**) (Wang et al., 2017a). When Li is deposited outside the reversible layer, it becomes "dead lithium". A 3D mixed electron–ion conductive framework is preferred as a Li metal host to increase the contact area, shorten the Li diffusion distance, and overcome the anticipated volume change.

### CONCLUSION AND PERSPECTIVES

This review provides a brief survey of recent research and development with respect to cathode–solid electrolyte interfaces in solid-state lithium batteries. We summarized the basic electrochemistry and principle at cathode-solid electrolyte interface, fundamental factors inducing interface challenges, and research progresses on building better interfaces. The interface issues in solid organic electrolytes, solid inorganic electrolytes, and solid composite electrolytes were reviewed and corresponding solutions are summarized on the basis of intrinsic characteristics of different solid electrolyte.

The interface degradation in solid state lithium battery may stem from the chemical/electrochemical stability and mechanical stability. Poor chemical/electrochemical stability between cathode and electrolyte may cause electrolytes decomposition or elements interdiffusion and transition region formation. The fundamental mechanism of interfacial chemical instability lies in the distribution of ionic electrochemical potential and inner electric potential gradient. Electronic and ionic conductivity of the transition region fundamentally determine whether a stable interface will form or not. An ionic conducting and electronic insulating transition layer will prevent further oxidation/reduction of electrolytes and SCL growth. While a mixed ionic and electronic conducting layer will result in continuous reactions and transition region growth.

To improve the chemical stability at cathode/SPE interface, different strategies can be adopted for different types of solid electrolytes. Modification of PEO matrix and lithium salts, adding proper plasticizer, cathode coating, and compositing with inorganic fillers are favorable for SPE and have made great progress. Solid oxide electrolytes are fairly stable with cathode compared with other solid electrolytes. While interdiffusion will take place along with high temperature dealing during in-situ synthesis, co-sintering, and deposition. Proper modification layer is needed to guarantee both chemical stability and intimate contact. Solid sulfide electrolytes, which suffer from SCL problem and decomposition reactions, should introduce other surface modifying method on the basis of introducing a proper buffer layer to eliminate SCL.

In addition to the chemical instability, it's difficult for inorganic solid electrolytes especially solid oxide electrolyte to maintain intimate contact with cathode due to the rigid ceramic nature. To achieve good mechanical contact, strategies such as surface coating, co-sintering, in-situ synthesis of electrolyte layer, interface softening, interface buffer layers were employed. By interface modification, the above issues are mitigated to a

(Wenzel et al., 2016). Copyright (2016) American Chemical Society]. (C) Pictures and schematic of a cylindrical cell for MRI. (D) Li density profiles at different depths of electrochemically cycled LGPS pellets. [Reprinted with permission from (Chien et al., 2018). Copyright (2018) American Chemical Society]. (E) Schematic of *in situ* TEM biasing of nanobattery. (F) STEM image and EELS characterization. (a–c) HAADF image of the nanobattery stack along with Li K-edge concentration mapping of (a) pristine, (b) *ex situ*, and (c) *in situ* samples with scale bar represents 200 nm. [Reprinted with permission from Wang et al. (2016b). Copyright (2016) American Chemical Society]. (G) Schematic of the NDP system. (H) 2D projection of the NDP spectra collected at 5 min intervals during cycling. [Reprinted with permission from (Wang et al., 2017a). Copyright (2017) American Chemical Society].

large degree, while a novel solid electrolyte with high ionic conductivity, chemical stability, compatibility with both cathode, and anode is still a long way to go.

In order to gain more insights into interface behavior, advanced characterization techniques are necessary, particularly with time, and atomic-scale resolution. Recent investigations and progress obtained from diverse advanced characterization techniques are summarized here. Note that interface phenomena involve multiscale and multidimensional properties change, the combination of diverse technologies may help provide comprehensive understanding. Electron holography and internal potential from AFM help uncover the electric potential profiles across interface. Spectroscopy including XAS, XPS, and NMR directly probes the chemical state and local structure. Microscopy including TEM and STEM provide atomic arrangement at interface structure. While XRD and ND can well-monitor structure evolution upon battery cycle. MRI and TOF-SIMS may further build up 3-dimensional distribution of particular elements. Despite valuable information obtained from these tools, here we highlight that the wider cooperation between diverse techniques will provide stronger support to the final clarification of interface related phenomena.

According to detailed analysis of interface issues and solutions for different systems, we further conclude that to realize a favorable interface, a hybrid solution should be employed to achieve both mechanical and chemical/electrochemical stability. It's hard for one kind of solid electrolyte to combine both excellent elasticity to achieve intimate contact and favorable strength to resist lithium dendrite formation. On the one hand, an elastic and transformable electrolyte which could shrink and expand in pace with cathode is essential for solid state lithium batteries, such as SPE and special solid sulfide electrolyte. Note that even deformable sulfide electrolyte could not keep contact with electrode particles upon cycling, therefore other strategies such as in-situ synthesis and surface coating may serve as proper ways to modify solid sulfide electrolytes. In order to obtain pristine intimate contact before cycling, strategies such as flashburning, interface sintering, deposition method, and in situ polymerization are necessary. Meanwhile, to relieve the strain and stress resulted from the shrinkage and expansion of cathode material, nano-sized solid electrolyte and electrode may serve

### REFERENCES


as a good choice. On the other hand, deficiency in strength and toughness of sulfides and polymers calls for toughening or compositing with other solid electrolytes.

Chemical/electrochemical stability are equally important, theoretical calculations, and some experimental results both revealed that the actual stability windows of solid electrolytes are not wide as expected. A stable interface layer with high Li<sup>+</sup> conductivity and low electrical conductivity is expected. Licompounds like LiF, Li2S, Li2O, Li3N, and LiNbO<sup>3</sup> are favorable interface components, while electronic conducting constituents such as metal sulfides (e.g., CoS) and Li-Metal alloys (e.g., Li-Ge alloy) should be avoided. To combine both mechanical and chemical/electrochemical demands of solid electrolyte, developing SPE with high strength and wide electrochemical window is necessary. Compositing is a promising method to utilize synergy effects among various electrolytes, but mitigating the disadvantages of each component need further study. Gel softening interface is also a promising way to achieve intimate contact at electrode-electrolyte interface. But flammable liquid components with narrow electrochemical stability window should be avoided.

According to various requirements, creating an ideal cathodesolid electrolyte interface requires a combination of various factors and methods. By building favorable electrode-electrolyte interface, solid-state batteries with higher safety performance, longer cycle life, and higher energy density are predictable.

### AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

### FUNDING

This work was supported by funding from the National Key R&D Program of China (Grant 2017YFB0102004), National Natural Science Foundation of China (Grant No. 51822211), China Postdoctoral Science Foundation, Beijing Municipal Science & Technology Commission (Grant Z171100000917021), One Hundred Talent Project of the Chinese Academy of Sciences and Thousand Talents Program for Young Scientists.

lithium microstructures in lithium batteries. Nat. Mater. 9, 504–510. doi: 10.1038/nmat2764


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Nie, Hong, Qiu, Li, Yu, Li and Chen. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Controllable Synthesis of Na3V2(PO4)3/C Nanofibers as Cathode Material for Sodium-Ion Batteries by Electrostatic Spinning

Ling Wu<sup>1</sup> , Yueying Hao<sup>1</sup> , Shaonan Shi <sup>1</sup> , Xiaoping Zhang<sup>1</sup> , Huacheng Li <sup>2</sup> , Yulei Sui <sup>1</sup> \*, Liu Yang<sup>1</sup> and Shengkui Zhong1,2 \*

*<sup>1</sup> School of Iron and Steel, Soochow University, Suzhou, China, <sup>2</sup> Citic Dameng Mining Industries Limited, Chongzuo, China*

### Edited by:

*Jiexi Wang, Central South University, China*

#### Reviewed by:

*Kai Jiang, Huazhong University of Science and Technology, China Lingjun Li, Changsha University of Science and Technology, China Shuqiang Jiao, University of Science and Technology Beijing, China*

\*Correspondence:

*Yulei Sui suiyulei@suda.edu.cn Shengkui Zhong zhongshengkui@suda.edu.cn*

#### Specialty section:

*This article was submitted to Physical Chemistry and Chemical Physics, a section of the journal Frontiers in Chemistry*

Received: *10 October 2018* Accepted: *29 November 2018* Published: *10 December 2018*

#### Citation:

*Wu L, Hao Y, Shi S, Zhang X, Li H, Sui Y, Yang L and Zhong S (2018) Controllable Synthesis of Na*3*V*2*(PO*4*)*3*/C Nanofibers as Cathode Material for Sodium-Ion Batteries by Electrostatic Spinning. Front. Chem. 6:617. doi: 10.3389/fchem.2018.00617* Na3V2(PO4)3/C nanofibers are prepared by a pre-reduction assisted electrospinning method. In order to maintain the perfect fibrous architecture of the Na3V2(PO4)3/C samples after calcining, a series of heat treatment parameters are studied in detail. It is found that the heat treatment process shows important influence on the morphology and electrochemical performance of Na3V2(PO4)3/C composite nanofibers. Under the calcining conditions of 800◦C for 10 h with a heating rate of 2.5◦C min−<sup>1</sup> , the well-crystallized uniform Na3V2(PO4)3/C nanofibers with excellent electrochemical performances are successfully obtained. The initial discharge specific capacities of the nanofibers at 0.05, 1, and 10C are 114.0, 106.0, and 77.9 mAh g−<sup>1</sup> , respectively. The capacity retention still remains 97.0% after 100 cycles at 0.05C. This smooth, uniform, and continuous Na3V2(PO4)3/C composite nanofibers prepared by simple electrospinning method, is expected to be a superior cathode material for sodium-ion batteries.

Keywords: sodium-ion batteries, cathode materials, Na3V2 (PO4 )3 , electrospinning, nanofibers

### INTRODUCTION

As an alternative strategy to lithium-ion batteries (LIBs), sodium-ion batteries (SIBs) recently have been paid increasing attention due to the low cost and the abundant reserves of sodium resource in the earth (Chen et al., 2018; Wu et al., 2018a,b; Ge et al., 2019). Nevertheless, comparing with Li-ion, Na-ion has larger ionic radius (1.02 Å), and heavier atomic weight, which is not conducive to the ion diffusion during the inserting/extracting processes. Thus, those compounds which have open framework are more suitable for the transmission of Na-ions. Among the cathode materials of SIBs, Na3V2(PO4)<sup>3</sup> is concerned widely due to its NASICON structure. The theoretical specific capacity of Na3V2(PO4)<sup>3</sup> is 117 mAh g−<sup>1</sup> (de-intercalating 2 Na+), and its average operating voltage is as high as 3.4 V. Most importantly, the NASICON structure can provides faster channels for the insertion/extraction of Na-ions, making Na3V2(PO4)<sup>3</sup> has relatively higher ionic conductivity than many other polyanionic compounds (Nanjundaswamy et al., 1996; Song et al., 2014a,b). However, the absence of electronic delocalization through direct -M-O-Mlinks results in a poor intrinsic electronic conductivity, and to some degree, the Na-ion diffusion is still limited by its large radius and heavy weight. Thus, the electronic/ionic conductivities of Na3V2(PO4)<sup>3</sup> further need to be improved (Jamesh and Prakash, 2008; Palomares et al., 2013; Kim et al., 2018). In order to increase the conductivity of cathode material, conductive agent coating (Prosini et al., 2001; Si et al., 2014; Ali et al., 2016; Chu and Yue, 2016), and particle size refining (Klee et al., 2016b; Wei et al., 2017; Zhang et al., 2017; Zheng et al., 2017) are usually used for modification. Compared with the solid-state method (Jian et al., 2012; Klee et al., 2016a), sol-gel method (Lim et al., 2012; Böckenfeld and Balducci, 2014), or solution evaporation method (Zheng et al., 2017), electrospinning is an effective way to achieve both of these goals. The Na3V2(PO4)<sup>3</sup> nanofibers prepared by electrospinning can weave a threedimensional (3D) conductive network which is beneficial for the fast transmission of electrons and sodium-ions (Chen et al., 2012, 2015; Zhong et al., 2016). Therefore, nano-sized Na3V2(PO4)<sup>3</sup> fibers wrapped with amorphous carbon are expected to exhibit better electrochemical performances.

Recently years, one-dimensional Na3V2(PO4)3/C composites have been synthesized by electrospinning method (Kajiyama et al., 2014; Liu et al., 2014; Li et al., 2015a,b). Nanofibers synthesized by Kajiyama et al. (2014) showed large size differences after heat treatment, giving rise to an unsatisfactory cycle stability. (Liu et al., 2014); Li et al. (2015a) reports that Na3V2(PO4)3/C nanofibers obtained at 800◦C tends to have a more uniform nanorod shape, but the rate performance of the nanofibers still needs to be enhanced. Li et al. (2015a) synthesized the Na3V2(PO4)3/C with budding willow branches shape by electrospinning method as well, yielding a discharge capacity of 116.2 mAh g−<sup>1</sup> at 0.05 C rate, but its rate performance was not investigated further in the paper.

Lately, the smooth and uniform nano-sized Na3V2(PO4)3/C composite fibers are successfully prepared by our group through a pre-reduction assisted electrospinning method (Wu et al., 2018c). The synthesized Na3V2(PO4)3/C nanofibers present significantly improved electrochemical performances. It is generally believed that the particle size, morphology, and structure can greatly affect the material electrochemical properties. While heat treatment conditions (such as temperature, duration, and heating rate) have direct effects on the morphology, structure, and particle size of nanofiber materials (Chen et al., 2015; Xu et al., 2015; Jing et al., 2016). However, the effect of heat treatment parameters on the properties of the Na3V2(PO4)3/C wires after electrospinning has not been studied yet. Therefore, a series of electrospun Na3V2(PO4)3/C nanofibers under different heating conditions were studied in detail in this paper. As the heat

treatment parameters are adjusted, the morphologies of the Na3V2(PO4)3/C nanofibers represent a series of significant and regular changes. The effect of morphology on its electrochemical performance is further discussed in particular below.

### EXPERIMENTAL

### Synthesis of Materials

Na3V2(PO4)3/C nanofibers were synthesized by the following procedures. (1) 0.12 M solution (30 mL) was prepared using H2C2O4·2H2O and deionized water and (2) NH4VO<sup>3</sup> and NaH2PO<sup>4</sup> (V: Na = 2:3, molar ratio) were added into the above solution, and a green solution (Solution I) was got after mixing at 70◦C for 2 h. (3) Polyvinylpyrrolidone (PVP K90, MW = 1,300,000) was added into the deionized water. After stirring for 2 h at room temperature, 1.5 g mL−<sup>1</sup> PVP solution (Solution II, 30 mL) was obtained. (4) Solution I was mixed with solution II under stirring, and then the final spinning solution (Solution III) was obtained after 2 h. (5) Solution III was pumped into an injector with a stainless steel needle pipe (inside diameter: 0.6 mm). In the spinning process, the distance, and voltage between the collector (Al-foil) and needle tip is 25 cm and 12 kV, respectively. The injection speed of spinning is 0.05 mL min−<sup>1</sup> . Then the nanofibers of precursor were collected and dried at 120◦C in oven for 12 h. (6) The nanofibers of precursor were calcined at 750–900◦C in Ar atmosphere for 6–12 h with the different heating rate of 1–5◦C min−<sup>1</sup> . And a series of Na3V2(PO4)3/C composite nanofibers were obtained after cooling to ambient temperature.

### Characterization

The phase and crystal structure of Na3V2(PO4)3/C samples were characterized by X-ray diffraction (XRD, Rigaku ultima VI). The morphology of Na3V2(PO4)3/C nanofibers was observed by scanning electron microscopy (SEM, Hitachi-SU5000). The amount of residual carbon of samples was measured by a C-S analyzer (Eltar, Germany).

### Electrochemical Measurements

The positive electrode plate was prepared with the Na3V2(PO4)3/C samples, acetylene black and PVDF (8:1:1, weight ratio) by using N-methylpyrrolidone (NMP) solvent with an Al-foil as current collector. The electrode loading density is about 2.5 mg cm−<sup>2</sup> . The button batteries (CR2025) were assembled in an Ar-filled glove box. A glass fiber membrane (Whatman, GF/A) and a metallic Na-foil were used as the separator and negative electrode, respectively. The NaClO<sup>4</sup> (1 M) solution in propylene carbonate (PC) and fluoroethylene carbonate (FEC) (1:0.05 in volume) was used as the electrolyte. The electrochemical performances of cells were tested on a LAND BT2013A battery tester at ambient temperature. The cells were charged/discharged at 0.05–10 C rates (1C = 118 mAh g −1 ) between the potentials (vs. Na/Na+) of 2.5 and 4.2 V. The electrochemical impedance spectroscopy (EIS) was measured by a CHI 660D workstation with the amplitude of 5 mV and the frequency range of 0.01–100 kHz.

TABLE 1 | The lattice parameters of Na3V2(PO4)3/C synthesized at different temperatures.


## RESULTS AND DISCUSSION

temperatures.

The main challenge of obtaining excellent Na3V2(PO4)3/C nanofibers is the structure stability during the heat treatment process, which requires to prevent the damage to the carbon layer and the fiber morphology. (**Figures 1a–d)** shows the SEM images of the Na3V2(PO4)3/C nanofibers prepared at 750– 900◦C. As shown, the samples calcined at 750 and 800◦C exhibit continuous fibrous morphology. Furthermore, the diameter of nanofibers synthesized at 800◦C is finer and more mean. When the heating temperature rises to 850◦C, the nanofibers are broken. The surface of the sample prepared at 900◦C can be unable to withstand the thermal stress so that the nanowires are completely disconnected and tend to grow into larger particles. Thus, the synthesis temperature exhibits a significant effect on the morphology of Na3V2(PO4)3/C nanofibers. And if the calcining temperature is not higher than 800◦C, the filamentous morphology of nanowires is more likely to be preserved. In addition, the TEM images in (**Figures 1e–g**) prove that the surface of nanofibers is uniformly and smoothly coated by amorphous carbon layer with a thickness of several nanometers.

**Figure 2** represents the XRD patterns of the samples synthesized at various temperatures. As shown, the diffraction

times.

peaks of all the samples are sharp and well-defined. All the samples can be fully indexed to Na3V2(PO4)<sup>3</sup> phase (space group of R-3C) (Zatovsky, 2010), and no impurity phases are detected.

The C-S analysis shows that the residual carbon contents of the samples calcined at 750–900◦C are 12.1, 11.7, 10.4, and 9.8%, respectively. The residual carbon content decreases as the calcining temperature increases. However, there are no obvious diffraction peaks of carbon can be observed, indicating the carbon is amorphous. The lattice constants of samples are listed in **Table 1**. As shown, both a and c increase with the synthesis temperature, indicating that the calcining process strengthened the crystallization of the samples.

The charge-discharge curves at 0.05–10 C rates of Na3V2(PO4)3/C samples synthesized at different temperatures are shown in **Figure 3**. As seen, all the samples show an obvious charge and discharge platform near 3.4 V. The discharge specific capacities at various rates increases first and then decreases with the heat temperature rising. When the calcining temperature reaches 800◦C, the material possesses the optimal comprehensive performances with the highest specific capacities and best rate capability. It exhibits a first discharge specific capacity of 114.0 mAh g−<sup>1</sup> at 0.05 C rate, close to the theoretical capacity, and still maintains a capacity of 77.9 mAh g−<sup>1</sup> at 10 C.

Although all samples are well-crystallized according to the XRD results, there are large differences in electrochemical properties among these samples. The reasons can be ascribed to the various morphologies of samples obtained at different

temperatures. In particular, the core-shell Na3V2(PO4)3/C nanofibers synthesized at 800◦C exhibit continuous, smooth, and uniform filamentous architecture (**Figures 1e–g**). The Na3V2(PO4)<sup>3</sup> fibers are wrapped with a carbon layer with a suitable thickness. In contrast, the nanofibers obtained at 750◦C are non-uniform and the average diameter is thicker, and samples at the region of 850–900◦C evenly cannot maintain fibrous structure. Thus, there are significant interconnections between the morphological characteristics of Na3V2(PO4)3/C nanowires and their electrochemical properties. In order to obtain better comprehensive electrochemical performances for the materials, the core-shell fibers should be continuous, smooth, and uniform and coated by proper carbon layer. Therefore, adjusting the electrospinning preparation conditions (adding oxalic acid and optimizing the heat treatment condition) can lead to excellent electrochemical properties of the material.

Further, the samples are synthesized at a heating rate of 2.5◦C min−<sup>1</sup> at 800◦C for 6–12 h, respectively. **Figure 4** presents the SEM images of the as-synthesized samples. After 6 and 8 h heat treatment, the originally smooth nanowires precursor become finer branches with crystalline particles on the surface. As the calcining time is prolonged, the branch-like crystalline particles are re-melted into the nanowires. The surface of the fibrous material becomes smooth and its diameter increases. The XRD patterns of the Na3V2(PO4)3/C materials obtained under the above conditions are shown in **Figure 5**. The crystallized Na3V2(PO4)<sup>3</sup> without obvious diffraction peak of carbon can be obtained. The presence of Na3V2(PO4)<sup>3</sup> grains on the surface of the nanowires responds to a higher intensity diffraction peak in the XRD patterns. As the calcining time is prolonged, the intensity of the diffraction peak slightly decreases. When

the calcining time reaches 12 h, the Na4P2O<sup>7</sup> (JCPDS 10-1087) heterogeneous phase appears. The carbon contents of the samples calcined for 6, 8, 10, and 12 h are 12.7, 12.2, 11.9, and 11.5%, respectively.

The charge and discharge curves of the samples with heat treatment time of 6–12 h are shown in **Figure 6**. It is obvious that the sample synthesized for 10 h shows the excellent properties. The sample synthesized for 8 h with grains on the surface has a first discharge specific capacity of 113.4 mAh g−<sup>1</sup> at 0.05 C,

rates.

but only 13.0 mAh g−<sup>1</sup> at 10 C. The capacity decay is more pronounced in the charge and discharge test of the sample synthesized for 6 h. These indicate that the electrochemical performances of Na3V2(PO4)3/C grains are much worse than the intact Na3V2(PO4)3/C nanofibers, which is consistent with the above results. Compared with the granular samples, the continuous, smooth, and uniform nanofibers evenly coated by a carbon layer can build up a perfect network with high electronic conductivity, which will greatly enhance the rate performances of Na3V2(PO4)3, especially at high current rates. An additional small platform around 3.9 V appears in the charging curve of the sample synthesized for 12 h, corresponding to the appearance of the heterogeneous phase in the XRD pattern. Thus, as the calcining time extended, the nanofibers gradually become more smooth, and continuous, and exhibiting a better rate performance. However, if the calcining time is too long, heterogeneous phase appears in the synthesized samples would lead to a decreased electrochemical property.

Based on the above results, another batch of samples are synthesized at 800◦C for 10 h with different heating rates. The SEM images of the samples prepared at different heating rates are shown in **Figure 7** At the heating rate of 1◦C min−<sup>1</sup> , the filament structure of the precursor is completely destroyed due to the excessive warm-up time, which is similar to the sample morphology at 900◦C in **Figure 1**. In order to keep the nanometer filament of the material, the heating rate is increased to 2.5 and 5 ◦C min−<sup>1</sup> . Obviously, the faster the heating rate is, the smaller the diameter of the wire gets.

The XRD patterns of the Na3V2(PO4)3/C materials obtained under the above conditions are shown in **Figure 8**. It shows that the slower of the calcining speed, the stronger of the samples' diffraction peaks. In the XRD pattern of 5◦C min−<sup>1</sup> sample, the intensity of diffraction peak is lower, and the half width is larger which may be the result of insufficient diffusion of elements during the process of heat treatment under a fast heating rate.

As shown in **Figure 9**, the electrochemical performance of the sample heated at 1◦C min−<sup>1</sup> is even worse than that of the 900◦C sample because of the most loss of filamentous morphology. The initial discharge capacity of the sample heated at the rate of 5◦C min−<sup>1</sup> is only 108.8 mAh g−<sup>1</sup> at 0.05 C, and its capacity decay is obvious at low current density. This may be due to the unsatisfied crystallization of the Na3V2(PO4)3.

Through the above analysis, the optimal synthesis conditions are as follows: calcine at 800◦C for 10 h with a heating rate of 2.5◦C min−<sup>1</sup> . For the Na3V2(PO4)3/C sample synthesized at the

### REFERENCES


optimal synthesis conditions, the rate-cycle curves and cycling perormance at 0.05 C rate are shown in **Figure 10**. The optimal sample exhibits excellent rate performance and cycle stability. From **Figure 10A**, it can be noticed that as the current rate reverses back to 0.05 C after cycling at different C rates, the specific capacity can recover to nearly initial values, indicating the good reversibility and structural stability of the Na3V2(PO4)3/C sample. And when 100 times cycle at 0.05 C (**Figure 10B**), the capacity retention reaches up to 97.0%.

### CONCLUSIONS

A series of Na3V2(PO4)3/C nanofibers are synthesized through a pre-reduction assisted electrospinning method. Under the different synthesis conditions, the morphology, and electrochemical performance of the Na3V2(PO4)3/C nanofibers are obviously different. As the temperature increases, the nanofibers become thinner, but gradually lose the filamentous morphology. With the elongation of the calcining time, shootlike crystalline particles on the fiber surface can be re-melted into nanowires. The heating rate also play a critical role in the structure and morphology of the Na3V2(PO4)3/C samples. As a result, the structure, morphology, and electrochemical performance of Na3V2(PO4)3/C composite nanofibers can be controlled by adjusting the heat treatment parameters. Under the optimum synthesis conditions of 800◦C, 10 h and heating rate of 2.5◦C min−<sup>1</sup> , the obtained Na3V2(PO4)3/C composite nanofibers present excellent electrochemical performances.

### AUTHOR CONTRIBUTIONS

LW, YH, and SS did the main experiment and write the manuscript. XZ and HL envolved the discussion of the experiment and revised the manuscript. LY assisted the material synthesis. YS and SZ made the research plan. SZ and LW also provided the financial support.

### FUNDING

We gratefully acknowledge the financial support from National Natural Science Foundation of China (51774207, 51574170, and 51774210), Qing Lan Project of Jiangsu Province (2017), and Major Projects for Science & Technology Development of Guangxi Province, China (AA16380043).

hybrid structures for sodium-ion batteries. J. Power Sour. 406, 110–117. doi: 10.1016/j.jpowsour.2018.10.058


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Wu, Hao, Shi, Zhang, Li, Sui, Yang and Zhong. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Fabrication of Cobalt-Nickel-Zinc Ternary Oxide Nanosheet and Applications for Supercapacitor Electrode

#### Chun Wu<sup>1</sup> \*, Lei Chen<sup>1</sup> , Xuechun Lou<sup>1</sup> , Mei Ding<sup>1</sup> and Chuankun Jia1,2 \*

*<sup>1</sup> College of Materials Science and Engineering, Changsha University of Science and Technology, Changsha, China, <sup>2</sup> Key Laboratory of Advanced Energy Materials Chemistry, Ministry of Education, Nankai University, Tianjin, China*

Mesoporous cobalt-nickel-zinc ternary oxide (CNZO) nanosheets grown on the nickel foam are prepared by a simple hydrothermal treatment and subsequent calcination process. The physical characterizations show that the as-obtained CNZO nanosheets possess the mesoporous structure and a high specific surface of 75.4 m<sup>2</sup> g <sup>−</sup><sup>1</sup> has been achieved. When directly applied for the binder-free supercapacitor electrode for the first time, the nickel foam supported mesoporous CNZO nanosheet electrode exhibits an ultrahigh specific capacity about 1172.2 C g−<sup>1</sup> at 1 A g−<sup>1</sup> . More significantly, an asymmetric supercapacitor based on the as-obtained CNZO positive electrode and an activated carbon negative electrode shows a high energy density of 84.2 Wh kg−<sup>1</sup> at a power density of 374.8 W kg−<sup>1</sup> , with excellent cycle stability (keeps 78.8% capacitance retention and 100% coulombic efficiency after 2,500 cycles). The excellent supercapacitive properties suggest that the nickel foam supported CNZO nanosheet electrodes are promising for application as high-performance supercapacitor.

### Edited by:

*Qiaobao Zhang, Xiamen University, China*

#### Reviewed by:

*Cao Guan, National University of Singapore, Singapore Xinhui Xia, Zhejiang University, China*

#### \*Correspondence:

*Chun Wu wuchundd@163.com Chuankun Jia jack2012ding@gmail.com*

#### Specialty section:

*This article was submitted to Nanoscience, a section of the journal Frontiers in Chemistry*

Received: *15 September 2018* Accepted: *15 November 2018* Published: *29 November 2018*

#### Citation:

*Wu C, Chen L, Lou X, Ding M and Jia C (2018) Fabrication of Cobalt-Nickel-Zinc Ternary Oxide Nanosheet and Applications for Supercapacitor Electrode. Front. Chem. 6:597. doi: 10.3389/fchem.2018.00597* Keywords: Co-Ni-Zn ternary oxide, electrode materials, nanosheet, supercapacitor, energy storage

## INTRODUCTION

Supercapacitors have been wildly investigated in recent years because of their advanced properties including long cycle stability, fast charge/discharge rate, high power density and environment-friendly and is becoming one of the promising energy storage system (Jia et al., 2015; Ding et al., 2018; Wang et al., 2019; Zhang et al., 2018b). Meanwhile, they also play an essential role to solve the serious environmental and energy crisis. Among different types of the supercapacitors, the faradaic capacitors with energy storage mechanisms based on the redox reactions have attracted considerable attention, which exhibit much higher specific capacitance than that of electrical double-layer capacitors (EDLCs) (Zeng et al., 2019). Accordingly, develop high-performance transition metal oxides and/or conducting polymers as faradaic capacitor electrode material with abundance in resources, low price and facile and scalable preparation process seems to be particularly important.

Recently, the transition metal oxides, such as MnO2, NiO, Co3O4, et al. have been extensively explored due to the advantages including easy large-scale fabrication and excellent flexibility in morphology and structures (Cheng et al., 2014; Wu et al., 2018; Cui et al., 2019). Particularly, mixed transition metal oxides with two or three metal ions are emerging as promising electrode materials thanks to several advantages compared to the simple transition metal oxides, such as enhanced

**101**

electronic conductivity (because of multiple oxidation states based on different metal ions for reversible faradaic reactions), high electrochemical activities, and relatively low activation energy for electron transfer between cations. However, several drawbacks of low intrinsic conductivity, big volume change and slow ion diffusion rates during the electrochemical measurement process, limit the applications of the transition metal oxides for high-performance supercapacitor.

Studies illustrate that the electrode materials directly grown on the conductive substrates such as carbon cloth and nickel foam and applied as the binder-free electrodes can significantly improve the electrochemical performance of the supercapacitors. Specifically, the well-designed threedimensional (3D) hierarchical architectures grown on the substrates with advantages of large specific surface areas resulting in reactive sites and better permeabilities of the electrolyte, have been employed for practical applications in supercapacitors, lithium ion batteries, and other energy storage devices (Li et al., 2015; Zhao et al., 2017; Zhang et al., 2018a). A supercapacitor electrode of 3D self-supported Co3O4@CoMoO<sup>4</sup> core-shell architectures on nickel foam has been prepared by Wang et al. The Co3O<sup>4</sup> nanocones grown vertically on the nickel foam are served as the core and CoMoO<sup>4</sup> nanosheets developed on the surface of the nanocones are acted as the shell. High specific capacitance, good rate capability, and cycling stability can be achieved (Wang et al., 2016). Via a facile strategy, Zhang et al. synthesized hierarchical Co3O4@NiCo2O<sup>4</sup> nanowire array nanoforests. The smart combination of Co3O<sup>4</sup> and NiCo2O<sup>4</sup> nanostructures in this hybrid architectures presents a promising synergistic supercapacitive effect with greatly improved properties. The electrode exhibits a high areal capacitance of 2.04 F cm−<sup>2</sup> at 5 mV s−<sup>1</sup> and 0.79 F cm−<sup>2</sup> even at 30 mA cm−<sup>2</sup> . Significantly, when the current turned back to 10 mA cm−<sup>2</sup> after long cycles with high current density, a areal capacitance of 1.18 F cm−<sup>2</sup> can be recovered, which can remain for another 1,500 cycles with excellent stability (Zhang et al., 2013). Another study about the supercapacitor electrode of 3D NiCo2O4@CoxNi1−x(OH)<sup>2</sup> core-shell nanosheet arrays have been rationally designed and facilely synthesized on the nickel foam via an electro-deposited method. Electrochemical measurements show that a large areal capacitance about 5.71 F cm−<sup>2</sup> at 5.5 mA cm−<sup>2</sup> can be achieved when the mass loading of the as-prepared NiCo2O4@CoxNi1−x(OH)<sup>2</sup> electrode material is about 5.5 mg cm−<sup>2</sup> . Furthermore, it also shows an excellent rate capability (Xu et al., 2014). The efficient electron and ion transportation, large surface areas resulting in easy electrolyte access to electrode and good strain accommodation makes 3D substrate supported hierarchical architectures attracting much attentions (Mohana Reddy et al., 2012; Gu et al., 2019; Wu et al., 2019). It can not only achieve high contact between substrate and electrode active materials resulting in great improvement of ion transportation, but also provide channels for the transportation of the charge. The above properties are beneficial to maximize the utilization of electrochemically active material. Therefore, simple processes for design and fabrication of mixed transition metal oxides with desired properties are needed to be exploited urgently.

Herein, a facile and environmental-friendly approach to synthesize 3D mesoporous CNZO nanosheet grown on nickel foam has been reported and the physical and electrochemical properties as binder-free electrode material for supercapacitors have been characterized. The mixed ternary metal oxides are supposed to offer a synergistic effect contributed by zinc, nickel, and cobalt ions of the mixed metal oxides on redox reactions, relating to the reasons that the electrical conductivity and capacitive performance can be enhanced by Zn, which possess good electrical conductivity; The active site density and conductivity can be improved by Ni, which shows high capacity. Meanwhile, Co can provide enhanced electronic conductivity. In addition, multi-phase metal oxides would be generated by incorporating various metal ions (Liu et al., 2013). All these would be benefit for the electrochemical performance improvement of the electrode.

The physical characterizations show that the as-obtained CNZO nanosheets possess the mesoporous structure and a high specific surface of 75.4 m<sup>2</sup> g <sup>−</sup><sup>1</sup> has been achieved. When directly applied for the binder-free supercapacitor electrode for the first time, the nickel foam supported mesoporous CNZO nanosheet electrode exhibits an ultrahigh specific capacity about 1172.2 C g−<sup>1</sup> at 1 A g−<sup>1</sup> . More significantly, the binderfree CNZO electrode reveals low internal resistance (Rs) and interfacial charge transfer resistance (Rct) values. The excellent supercapacitive properties suggest that the nickel foam supported CNZO nanosheet electrodes are promising for application as high-performance supercapacitor.

### EXPERIMENTAL

### Synthesis of CNZO Nanosheet Material

The nickel foam supported CNZO nanosheet material is prepared by a simple hydrothermal treatement and a followed calcination process. The nickel foam with 3.5 × 6 cm size was washed and dried for the reaction. 12 mmol urea (CO(NH2)2), 4 mmol ammonium fluoride (NH4F), 3 mmol nickel sulfate hexahydrate (NiSO4·7H2O), 6 mmol cobalt sulfate heptahydrate (CoSO4·7H2O), and 3 mmol zinc sulfate heptahydrate (ZnSO4·6H2O) were mixed with 80 mL DI water. The mixed solution and the treated nickel foam were kept at 130◦C for 5 h after transferred into a 100 mL Teflon-lined stainless steel autoclave. The sample was kept in an ultrasound bath and washed with distilled water for several times and dried at 60◦ C for 12 h. Finally, put the as-obtained samples into the furnace and treated in air atmosphere at 350◦C for 2 h to achieve the nickel foam supported CNZO nanosheet material. The mass of the CNZO materials on the nickel foam is about 1.6 mg cm−<sup>2</sup> .

### Characterization of Material

Field emission scanning electron microscopy (FESEM Hitachi S4800) was used to gain the microstructure and morphology information of the as-prepared CNZO sample. By using X-ray diffraction (Rigaku SmartLab XRD), the crystalline structure of the CNZO sample can be achieved. Furthermore, to investigate the

microstructure of the CNZO sample, a Quantachromeautosorb automated gas sorption system was employed and adsorption/desorption isotherm of nitrogen was tested.

### Electrochemical Measurements

In the electrolyte of 6 M KOH solution, a three-electrode system with CNZO electrode acting as working electrode, nickel foam serving as counter electrode and Hg/HgO applying as reference electrode was employed to study the electrochemical performances of CNZO electrode. All of the supercapacitive characterizations, including cyclic voltammetry measurements (CV), galvanostatic charge/discharge measurement as well as electrochemical impedance spectroscopy (EIS), were conducted on a electrochemical workstation system (CHI660). The cycle stability of the supercapacitor is measurement under the Neware battery testing system.

### RESULTS AND DISCUSSION

### Structure Analysis

**Figure 1** illustrates the preparation process of the mesoporous CNZO nanosheet electrode materials. First, by hydrothermal treatment under 130◦C for 5 h, the raw mixed solutions react and the precursor of mesoporous CNZO nanosheet materials have been synthesized. Then, via a calcination process in air atmosphere under 350◦C for 2 h, the asobtained CNZO nanosheet precusor transforms into the final mesoporous CNZO nanosheet electrode material. The growth of the mesoporous CNZO nanosheet on the current collector would result in direct contact between the asprepared active material and the substrate, thus lead to the short electrons and ions diffusion path and achieve very low contact resistance, which are beneficial for the electrochemical performance improvement of the as-obtained CNZO electrode.

The observation of microstructure and morphology for the as-prepared mesoporous CNZO nanosheet material has been investigated by FESEM. It can be notably noticed from **Figures 2a,b** of the low magnification SEM image that numerous nanosheets homogeneously grow on the surface of the nickel foam to form a 3D nanostructure. The diameter of the nanosheet is about 20–40µm, and the thickness is about 0.4µm measured from the high magnification SEM image of **Figures 2c,d**. More interestingly, it also can be seen that numerous burr-like structures on the surface of the nanosheets, which would lead to larger contact area between electrode material and electrolyte compared to the pure nanosheet and result in excellent electrochemical behaviors. The microstructure of the as-prepared electrode material is further observed by TEM and present as follows. It can be noted that the porous structure is existed in the edge part of the nanosheet, which constructed from numerous nanoparticle subunits (see **Figure 2f**). Highly resolved lattice fringes with a measured interplanar spacing of 0.462 nm is also shown, which corresponds to the (111) plane of the spinel Co-based metal oxide phase (Xiao et al., 2014).

Via the elemental mapping analysis under SEM observation, the elemental distribution of the as-prepared CNZO nanosheets can be observed. **Figure 3** displays the corresponding element mapping results, which apparently presents the existence of the Co, Ni, Zn, and O and their homogeneous distribution, indicating the successful formation of ternary metal oxide CNZO electrode material.

After being scraped off from the substrate to eliminate effects originated from the nickel foam, the crystal structure of CNZO nanosheet material power is characterized by XRD analysis. All the diffraction peaks of the CNZO sample in **Figure 4** can be indexed to the the spinel Co3O<sup>4</sup> (PDF No. 74-2120) with slightly shift. The reason can be ascribed to the differences of the metal ionic radii of Ni, Co, and Zn. However, the substitution of Ni and Zn element does not significantly affect the crystal structure of

spinel Co3O<sup>4</sup> (Li et al., 2014). Besides, no additional peaks for other phases can be observed.

The investigation of the as-prepared mesoporous CNZO nanosheet microstructures is conducted via the nitrogen adsorption measurement. **Figure 5** exhibits the pore size distribution curves and nitrogen adsorption-desorption isotherms. Evidently, the curve of the nitrogen adsorptiondesorption isotherm for the as-prepared CNZO material is close to that of IUPAC type-IV with distinct hysteresis loops (Rojas et al., 2002), suggesting the mesoporous structure of the as-prepared sample. The pore size distribution of the as-prepared CNZO material is presented in the inset curve, it can be seen that the pore size is centered at about 2.5 nm, which further confirms the presence of the mesopores in this CNZO nanosheet material. The BET specific surface area of the as-prepared mesoporous CNZO nanosheet material is measured to be about 75.4 m<sup>2</sup> g −1 . With large specific surface area, the nickel foam supported CNZO nanosheet material would bring about more reactive sites and result in better penetration of the electrolyte into the whole electrode materials during the electrochemical measurement process, all these would be beneficial for the excellent supercapacitive properties.

X-ray photoelectron spectroscopy is utilized to characterized the elemental composition and oxidation state of the electrode material. **Figure 6** shows the high-resolution of Co, Ni, Zn, and O spectra. As depicted in **Figure 6a**, two distinguished doublets located at a high-energy (Co 2p1/2) and low-energy band (Co 2p3/2) can be observed in the high-resolution of Co 2p spectrum. The spin-orbit splitting value of Co 2p1/2 and Co 2p3/2 is over 15 eV, indicating the presence of Co3+and Co2<sup>+</sup> (Xiong et al., 2015). Meanwhile, the high-resolution of Ni 2p spectrum in **Figure 6b** can be fitted with two shakeup satellites and two spinorbit doublets, which shows the characteristic of Ni3<sup>+</sup> and Ni2<sup>+</sup> (Xiong et al., 2015). Besides, as for the high-resolution of Zn 2p spectrum in **Figure 6c**, a major peak at 1020.6 eV can be seen,

which can be attributed to the Zn 2p3/2 of Zn(II). Furthermore, two peaks at 531.2 and 529.3eV of the O 1s can be noted from **Figure 6d**, corresponding to the oxygen species contained in the CNZO material. The above results demonstrates that the Co3+, Co2+, Ni3+, Ni2+, and Zn2<sup>+</sup> are existed in the CNZO material.

### Electrochemical Behaviors of the CNZO Electrode

CV measurement under the potential window from 0 to 0.55 V is employed to investigated the supercapacitive behaviors of the CNZO electrode. **Figure 7a** present the CV curves of the NZO, CZO, and CNZO electrodes at 20 mV s−<sup>1</sup> , which shows that the CNZO electrode exhibits the largest increment in the CV integrated area, suggesting the largely enhanced electrochemical reaction activity of the CNZO electrode material. **Figure 7b** displays the charge/discharge curves of the NZO, CZO, and CNZO electrodes at 1 A g−<sup>1</sup> , it can be seen that the CNZO electrode exhibits the longest discharge time and thus delivers the highest specific capacitance. The CV curves of the CNZO electrodes at the scan rates from 2 to 50 mV s−<sup>1</sup> (see **Figure 7c**). It can be evidently seen that the shapes of all the CV curves exhibit typical faradaic behaviors. A pair of redox peaks can be clearly observed in the CV curves, indicating that the electrochemical behaviors of CNZO electrode generate from their faradaic reactions. The energy storage mechanism of the asprepared CNZO electrode can be attributed to the Faradic redox reactions assigned to the M-O/M-O-OH (M stands for both Co and Ni ions) (Wang et al., 2011). Furthermore, with the increase of the scan rates, the redox peak positions shift progressively, this can be ascribed to the existence of polarization (Wu et al., 2012). Moreover, the specific capacity can be calculated based on the CV curves according to the following equation (1) (Brousse et al., 2015):

$$C = \frac{Q}{m} = \int \frac{i dt}{m} \tag{1}$$

Here i represents a sampled current value (A), dt stands for a sampling time span (s), and m is the mass of the CNZO materials (g). The maximum specific capacity of the CNZO electrode is calculated to be about 1185.1 C g−<sup>1</sup> at the scan rates of 2 mV s−<sup>1</sup> .

The supercapacitive behaviors of the CNZO electrode can be further studied by the charge/discharge measurement and the charge/discharge curves under 1–5 A g−<sup>1</sup> are exhibited

in **Figure 7d**. Obviously, all curves are symmetric in shape, indicating the excellent electrochemical behaviors of the CNZO electrode. Moreover, the specific capacity can be calculated based on the charge/discharge curves according to the following equation (2) (Brousse et al., 2015):

$$\mathbf{C} = i\_m^\* \Delta t \tag{2}$$

Where, i<sup>m</sup> (A g −1 ) is the current density and 1t is the discharge time (s). The specific capacity of the CNZO electrode is calculated to be about 1172.2 C g−<sup>1</sup> at 1 A g−<sup>1</sup> . And the specific capacities under various current densities are exhibited in **Figure 7e**, excellent rate capability with 95.4% capacity retention can be achieved even at a high current density of 5 A g−<sup>1</sup> . The high specific capacity and wonderful rate capability suggests that the CNZO electrode can be a very promising materials for the applications of supercapacitor.

To further study the supercapacitive performance of the as-prepared CNZO electrode, the EIS measurement with amplitude of 5 mV ranging from 10<sup>5</sup> to 10−<sup>2</sup> Hz is operated (see in **Figure 7f**). The curve exhibits a straight line in the low frequency region and a small semicircle in the high frequency region. The straight line can be attributed to the presence of the Warburg impedance, which displays the result of the frequency dependence of ion diffusion to the electrode interface (Zhang et al., 2012). In the high frequency region from the inset curve, the first intersection

point shows the internal resistance (Rs) and the diameter of the semicircle stands for the interfacial charge transfer resistance (Rct) (Qiu et al., 2015). As can be noted that the R<sup>s</sup> and Rct values of CNZO electrode are very small, demonstrating the excellent electrochemical performance of the CNZO electrode.

The cycle stability of the electrode material is an essential parameter for the supercapacitor application, and **Figure 7g** presents the cycle behavior of the CNZO electrode under the current density of 5 A g−<sup>1</sup> . It can be obviously noted that after 2,000 cycles, the capacity retention is about 81.6% and the coulonbic efficiency remains 100%, suggesting the excellent cycle performance of the CNZO electrode.

To further evaluate the practical supercapacitors application of the CNZO electrode material, an asymmetric supercapacitor (CNZO//AC ASC device) based on the CNZO positive electrode and the active carbon (AC) negative electrode has been fabricated. The electrochemical performance of the AC electrode is reported in the previous work (Wu et al., 2015). The supercapacitive behaviors of the CNZO//AC ASC device is conducted during the potential window range from 0 to 1.5 V in 6 M KOH solution. **Figure 8a** displays the typical CV curves of the CNZO//AC ASC device under the scan rates of 1–10 mV s−<sup>1</sup> . A pair of well-defined redox peaks can be seen from the CV curves between the potential window at 0.8 and 1.5 V, indicating battery-type behavior of the as-fabricated CNZO//AC ASC device. Meanwhile, the corresponding charge/discharge curves under the current densities of 0.5–5 A g−<sup>1</sup> are shown in **Figure 8b**, it can be seen that the discharge curves are slightly non-linear during the potential window, which is in agreement with the above CV measurements. The maximum specific capacitance of the CNZO//AC ASC device is calculated to be 269.5 F g−<sup>1</sup> . As we all known the the energy and power density are the vital parameters for the supercapacitor application, which can be calculated based on the charge/discharge measurements of the CNZO//AC ASC device according to the following equations (Wu et al., 2015):

$$E = \frac{1}{2}CV^2\tag{3}$$

$$P = \frac{E}{\Delta t} \tag{4}$$

where E represents the energy density of the device (Wh kg−<sup>1</sup> ), C is the specific capacitance of the device (F g−<sup>1</sup> ), V shows the potential window (V), P is power density of the device (W kg−<sup>1</sup> ), 1t stands for the discharge time (s). A Ragone plot for the as-fabricated ASC device is presented in **Figure 8c**, which shows an high energy density of 84.2 Wh kg−<sup>1</sup> at a power density of 374.8 W kg−<sup>1</sup> and energy density of 41.6 Wh kg−<sup>1</sup> even at a power density of 3.8 kW kg−<sup>1</sup> . The electrochemical behaviors are better than those of the previous reports, such as Co3O4//AC (Zhang et al., 2014), NiCo2O4//AC (Zhang et al., 2015), ZnCo2O4//AC (Guan et al., 2014), NiWO4//AC (Niu et al., 2013), and CoMoO4//AC (Yu et al., 2014). Furthermore, the cycle performance of the CNZO//AC ASC device under the current density of 5 A g−<sup>1</sup> is shown in **Figure 8d**, which presents that the capacitance retention keeps 78.8% and the coulombic efficiency remains 100% after 2,500 cycles, demonstrating the excellent cycle stability of the as-fabricated CNZO//AC ASC device.

The wonderful electrochemical behavior of the CNZO electrode can be ascribed to the following aspects: First, the synergistic effect among different metal element that Ni ion can make a contribution of the high capacity, Co and Zn ions may provide enhanced electronic conductivity, may result in improved capacitive property of the CNZO electrode. Second, the nickel foam supported CNZO nanosheet could eliminate use of polymer binders, thus result in a very low "dead volume" and lead to high interfacial contact between the substrate and the active materials, these would improve the ion transportation and achieve excellent supercapacitive behaviors. Third, the mesoporous CNZO nanosheet material with highly porous structures and large specific surface area may significantly bring about more reactive sites and result in better penetration of the electrolyte into the whole electrode materials during the electrochemical measurement process, all these would be beneficial for the excellent supercapacitive properties.

### REFERENCES


### CONCLUSIONS

The nickel foam supported mesoporous CNZO nanosheet electrode material has been successfully prepared by a simple hydrothermal treatment and subsequent calcination process. The physical characterizations show that the as-obtained CNZO nanosheets possess the mesoporous structure and a high specific surface of 75.4 m<sup>2</sup> g <sup>−</sup><sup>1</sup> has been achieved. When directly applied for the binder-free supercapacitor electrode for the first time, the nickel foam supported mesoporous CNZO nanosheet electrode exhibits an ultrahigh specific capacity about 1172.2 C g−<sup>1</sup> at 1 A g −1 . More significantly, an asymmetric supercapacitor based on the CNZO nanosheet positive electrode and an activated carbon negative electrode shows an high energy density of 84.2 Wh kg−<sup>1</sup> at a power density of 374.8 W kg−<sup>1</sup> , with excellent cycle stability (keeps 78.8% capacitance retention and 100% coulombic efficiency after 2,500 cycles). With the excellent supercapacitive properties, the nickel foam supported CNZO nanosheet electrode could be employed for the high-performance supercapacitor electrodes and other energy storage system.

### AUTHOR CONTRIBUTIONS

CW designed and performed the experiments. CW, LC, and XL prepared the samples and analyzed the data. CW, MD and CJ participated in interpreting and analyzing the data. All authors read and approved the final manuscript.

### ACKNOWLEDGMENTS

CJ is grateful for support of the 100 Talented Team of Hunan Province and the 111 project (B12015) at the Key Laboratory of Advanced Energy Materials Chemistry (Ministry of Education), Nankai University.


forests for high-performance supercapacitors. Nano Energy 11, 687–696.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Wu, Chen, Lou, Ding and Jia. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# A Porous and Conductive Graphite Nanonetwork Forming on the Surface of KCu7S<sup>4</sup> for Energy Storage

Wei-Xia Shen, Jun-Min Xu, Shu-Ge Dai\* and Zhuang-Fei Zhang\*

*Key Laboratory of Material Physics of Ministry of Education, Zhengzhou University, Zhengzhou, China*

A flexible all-solid-state supercapacitor is fabricated by building a layer of porous and conductive nanonetwork on the surface of KCu7S<sup>4</sup> nanowires supported on the carbon fiber fabric, where the porous and conductive nanonetwork is assembled by graphite nanoparticles. This porous graphite layer plays a key role in providing ion diffusion channels to access the KCu7S<sup>4</sup> through the pores for electrochemical reactions and forming electron transport pathways from the graphite network to the electronic collector of the carbon fiber fabric. This flexible supercapacitor exhibits excellent electrochemical performance with high specific capacitance of 408 F g−<sup>1</sup> at a current density of 0.5 A g−<sup>1</sup> and high energy density of 36 Wh kg−<sup>1</sup> at a power density of 201 W kg−<sup>1</sup> . Moreover, it is cost-effective, easy to scale up and environmentally friendly with high flexibility. Our investigation demonstrates that such a porous and conductive nanonetwork could be used to improve the charge storage efficiency for a wide range of electrode materials.

Keywords: flexible, porous, graphite nanonetwork, KCu7S4 nanowires, supercapacitor

### INTRODUCTION

Nowadays, it is a great challenge to develop supercapacitors (SCs) with flexibility, lightweight and high electrochemical performance. In general, the quality of the SCs strongly depends on the design of an appropriate configuration and the innovation of electrode materials (Niu et al., 2013). For the traditional electrodes in SCs, carbonaceous materials (activated carbon, graphite, carbon nanotubes, and graphene) can offer very high power density and excellent cycling ability (Niu et al., 2017; Du et al., 2018; Liu et al., 2018). However, the energy density of carbon-based materials is still too low to meet the requirement for SCs in practical applications (Lu et al., 2014; Guan et al., 2015; Wang et al., 2016; Xia et al., 2017; Dai et al., 2018). Compared with carbon-based SCs, transition-metal oxides/sulfides have attracted particular attention since they could offer much higher energy density by Faradaic reactions (Augustyn et al., 2014; Simon et al., 2014; Dai et al., 2017; Qu et al., 2017; Xu et al., 2017; Zhang et al., 2018a). However, they usually suffer from low electrical conductivity, poor rate performance and limited cycling stability (Liu et al., 2010; Xia et al., 2014; Dai et al., 2016a; Jiang et al., 2018). To overcome the accumulation of produced charges on the surface of pseudo-capacitor material which could not successfully reach electron collector, the design of hybrid structure electrodes is an efficient way for SCs with excellent electrochemical performance (Chang et al., 2012; Qu et al., 2018a,b; Zhang et al., 2018b; Zheng et al., 2018a).

Recently, transition-metal oxides are emerging as promising electrode materials for energy storage devices, such as RuO2, MnO2, NiO, Fe2O3, WO3, V2O<sup>5</sup> (Xue et al., 2011; Dang et al., 2018; Zheng et al., 2018b). Among them, manganese oxides have been widely studied as electrode

### Edited by:

*Qiaobao Zhang, Xiamen University, China*

### Reviewed by:

*Aibing Chen, Hebei University of Science and Technology, China Haibin Sun, Shandong University of Technology, China Shanglong Peng, Lanzhou University, China*

#### \*Correspondence:

*Shu-Ge Dai shugedai@zzu.edu.cn Zhuang-Fei Zhang zhangzf@zzu.edu.cn*

#### Specialty section:

*This article was submitted to Nanoscience, a section of the journal Frontiers in Chemistry*

Received: *27 September 2018* Accepted: *29 October 2018* Published: *21 November 2018*

#### Citation:

*Shen W-X, Xu J-M, Dai S-G and Zhang Z-F (2018) A Porous and Conductive Graphite Nanonetwork Forming on the Surface of KCu7S4 for Energy Storage. Front. Chem. 6:555. doi: 10.3389/fchem.2018.00555* materials for SCs due to their high theoretical capacitance, lowcost, environmentally friendliness and natural abundance. The α-MnO<sup>2</sup> is constructed from double chains of octahedral [MnO6] structure with 2 × 2 and 1 × 1 tunnels, which is beneficial for Li<sup>+</sup> transportation (Park et al., 2007; Reddy et al., 2009). However, its actual capability is often much lower than the theoretical value owing to its low electronic conductivity. Besides, it also displays poor capacity retention and large volume change during Li<sup>+</sup> insertion/extraction (Wang et al., 2014a). Similar to the crystal structure of α-MnO2, the KCu7S<sup>4</sup> has one-dimensional double tunnels along c axis, which is composed of a three-dimensional Cu-S framework that contains pseudo-one-dimensional channels in which K ions reside in the channels (Hwu et al., 1998; Dai et al., 2013, 2014a). Compared with a-MnO2, the KCu7S<sup>4</sup> exhibits greater conductivity and capacity retention, which is one of the most promising electrode materials for energy storage (Dai et al., 2013, 2014a; Guo et al., 2016). Moreover, the KCu7S<sup>4</sup> has significant advantages, such as large surface area, low-cost, easy synthesis, and environmentally friendliness. To improve the performance, many researchers have focused on the surface modification of the micro/nano electrode materials, such as Au nanoparticles coated WO3−<sup>x</sup> NWs (Lu et al., 2012), graphene quantum dots coated VO<sup>2</sup> arrays (Chao et al., 2015), CNTs decorated MoO<sup>3</sup> (Yang et al., 2014a). It is an effective way to enhance the electrical conductivity of the electrode materials, which improves the ion diffusion kinetics and electron transport by coating of nanostructured conductive layer. Herein, we design a porous and conductive nanonetwork by coating graphite nanoparticles on the surface of KCu7S<sup>4</sup> nanowires, which not only ensures the multichannel diffusion of electrolyte ions insert the KCu7S<sup>4</sup> material, but also improves the electron transportation. It is no doubt that this porous and conductive nanonetwork structure will attract more attention in the design of the electrodes for SCs.

Currently, the fabricated electrodes based on KCu7S<sup>4</sup> materials are too rigid and bulky, which could not meet the practical requirements for flexible and wearable electronic devices (Dai et al., 2014b, 2015). Therefore, the exploration of flexible, lightweight, or even wearable SCs based on the KCu7S<sup>4</sup> materials will be interesting work. Recently, carbon fiber fabric (CFF) attracts many people's interest because of its unique characteristics, such as low corrosion resistance, low thermal expansion coefficient and excellent flexibility. Moreover, all-solid-state supercapacitors based on CFF can be easily bent or twisted, which could meet the requirements for flexible and wearable electronic devices (Yuan et al., 2012).

In this work, we report a highly flexible all-solid-state SC based on a layer of porous and conductive graphite nanonetwork coated on the surface of KCu7S<sup>4</sup> nanowires, which is supported on a carbon fiber fabric (GN/KCu7S4/CFF). The GN/KCu7S4/CFF SC exhibits great electrochemical performance with the highest specific capacitance of 408 F g−<sup>1</sup> and the highest energy density of 36 Wh kg−<sup>1</sup> at a power density of 201 W kg−<sup>1</sup> . The enhanced capacity attributed to the porous and conductive nanonetwork on the surface of the KCu7S<sup>4</sup> nanowires, which provides rich ion diffusion channels to access the KCu7S4, and shortens the electron transmission paths through the graphite network to

the electronic collector of CFF. This work demonstrates that the porous and highly conductive graphite nanonetwork could be used to improve the charge storage for a wide range of electrode materials, revealing a promising application in the flexible energy-storage devices.

### EXPERIMENTAL SECTION

## Preparation of GN/KCu7S4/CFF Electrode

Carbon fiber fabric (Shanghai Lishuo Composite Material Technology Company) and the graphite ink (from Hero, Shanghai Ink Factory in China) were used as purchased. First, 1 mmol of CuCl2·2H2O, 2.5 mmol of S, and 53 mmol of KOH were dissolved in deionized water (10 mL) in the Teflon containers, followed by addition of 2 mL of hydrazine monohydrate. Then the mixed solution was retained at 150◦C for 12 h. After cooled down to room temperature, the product was rinsed with ultrapure water, and dried under vacuum at 60◦C overnight. The GN/KCu7S4/CFF was made as follows: 100 mg of as-prepared KCu7S<sup>4</sup> nanowires was first dispersed in ultrapure water (10 mL). Then the graphite ink was dropped into the KCu7S<sup>4</sup> solution (the ratio of ink to water is 1:10) under magnetic stirring for 24 h at 95◦C. Finally, the mixture was filtered on the CFF to obtain the GN/KCu7S4/CFF, where free nanoparticles were removed through the pores of the CFF. The product was put into oven for 2 h at 60◦C for drying.

### Fabrication of All-Solid-State Supercapacitor

The separator (Whatman 8µm filter paper) covered with a layer of PVA-LiCl gel as a solid electrolyte on both sides and, sandwiched between the two pieces of the GN/KCu7S4/CFF electrodes to form a two electrode device. The detailed fabrication process of the electrode was reported in our previous work (Javed et al., 2015). Here, the mass loading on the carbon fiber fabric is about 2 mg cm−<sup>2</sup> and the working area of each electrode is 4 cm × 1.5 cm.

### Characterization and the Electrochemical Measurements

The morphology, chemical composition, and the structure of the products were observed by X-ray diffraction (XRD) analysis (XRD, PA National X′ Pert Pro with Cu Kα radiation). The microstructure and morphology of NC nanomaterials were characterized using field emission scanning electron microscopy (Zeiss, sigma300) and high-resolution transmission electron microscopy (HRTEM, JEOL, JEM-2100) with energy dispersive X-ray spectrometry (EDS). The nitrogen adsorptiondesorption isotherm measurement of the sample was performed using a ASAP2420-4MP. The specific surface area was obtained by the Brunauer-Emmett-Teller (BET) method. The electrochemical measurement was conducted with an

electrochemical workstation (CHI 760D). X-ray photoelectron spectrometer (XPS) analysis was performed on an ESCA Lab MKII using Mg Ka as the exciting source.

### RESULTS AND DISCUSSION

**Figure 1** shows the schematic diagram of preparing of GN/KCu7S<sup>4</sup> nanowires and the fabrication of flexible GN/KCu7S4/CFF SC, respectively. The X-ray diffraction of KCu7S<sup>4</sup> and GN/KCu7S<sup>4</sup> nanowires indicate that the samples are well crystallized (**Figure 2A**). All the diffraction peaks can be unambiguously assigned to tetragonal KCu7S<sup>4</sup> structure. To understand the porosity and surface area of as-prepared samples, N<sup>2</sup> adsorption-desorption isotherms of KCu7S<sup>4</sup> and GN/KCu7S<sup>4</sup> conducted at 77.350 K were investigated and are displayed in **Figure 2B**. Through BET analysis, the surface areas of KCu7S<sup>4</sup> and GN/KCu7S<sup>4</sup> samples were identified as 1 m<sup>2</sup> g −1 and 18.6 m<sup>2</sup> g −1 , respectively. To identify the chemical states of Cu element in the samples, the XPS survey spectrum of the KCu7S<sup>4</sup> nanowires and high-resolution XPS spectrum of Cu 2p were also conducted (**Figures 2C,D**). It consists of two binding energy of Cu 2p3/<sup>2</sup> and Cu 2p1/<sup>2</sup> peaks at 932.3 and 952.2 eV, respectively, which are in agreement with the previous reports (Colleen and McShane, 2012; Wang et al., 2013). Scanning electron microscopy (SEM) images of as-prepared KCu7S<sup>4</sup> and GN/KCu7S<sup>4</sup> samples are shown in **Figure 3**, **Figure S1**. The KCu7S<sup>4</sup> nanowires have a diameter of 200–500 nm and length up to 110µm. The enlarged image (**Figure 3b**) of GN/KCu7S<sup>4</sup> nanowires clearly indicates that the KCu7S<sup>4</sup> nanowires were coated with graphite nanoparticles with high homogeneity. For further confirmation, the EDS of a single GN/KCu7S<sup>4</sup> nanowire is presented in **Figure 3c**, revealing the main compositions of C, K, Cu, and S. This good composite nanostructure was also further confirmed by transmission electron microscopy (TEM) analysis, as shown in **Figures 3d,e**. In order to explore the composition of the graphite ink and GN/KCu7S4, we also carried out a Raman test and the results are presented in **Figure 3f**, **Figure S2**. The G and D peaks are clearly observed at 1355 cm−<sup>1</sup> (attributed to the disordered carbonaceous component) and 1585 cm−<sup>1</sup> (attributed to the ordered graphitic component), respectively, which exhibits that the active component in graphite ink is mainly graphitic carbon (Cai et al., 2012; Dai et al., 2014c). The peak at 472 cm−<sup>1</sup> corresponds to the KCu7S<sup>4</sup> (**Figure S2**). Moreover, the TEM-EDX elemental mapping of the GN/KCu7S<sup>4</sup> reveals a relatively uniform distribution of K, Cu, S, and C elements over the nanowire, which indicates the KCu7S<sup>4</sup> nanowires were well wrapped by the graphite nanoparticles. Owing to the strong adhesion of the graphite nanoparticles bounded together to form a porous nanonetwork structure on the surface of the KCu7S<sup>4</sup> nanowires, the nanonetwork can provide efficient ion diffusion multichannels to access the KCu7S<sup>4</sup> and shorten the electron transport pathways to

the electronic collector of CFF (Fu et al., 2012; Dai et al., 2016b).

The electrochemical performance of the supercapacitors based on the KCu7S4/CFF and GN/KCu7S4/CFF are characterized by using cyclic voltammetry (CV), galvanostatic charge-discharge (GCD) cycling and electrochemical impedance spectroscopy (EIS), respectively. **Figure 4A** shows the CV curves of the bare CFF, KCu7S4, and /GN/KCu7S4/CFF based SCs at a constant scan rate of 100 m V/s. It is note that the GN/KCu7S4/CFF SC shows a higher capacitance behavior as compared with others. **Figure 4B** exhibits the CV curves of the GN/KCu7S4/CFF SC at different scan rates in potential windows from 0 to 0.8 V. All the CV curves exhibit an approximate shape with slight variations, even at a scan rate of 100 m V/s, revealing the good capacitive behavior of the GN/KCu7S4/CFF electrodes. The CV curves of KCu7S<sup>4</sup> SC at different scan rates were also collected and is shown in **Figure S3A**. The galvanostatic charge-discharge curves of the GN/KCu7S4/CFF SC at various current densities in potential windows from 0 to 0.8 V (**Figure 4C**) exhibit good linear and almost symmetrical voltage-time profiles with small IR drops, indicating high output power of the GN/KCu7S4/CFF SC. The corresponding galvanostatic charge-discharge curves of the KCu7S4/CFF SC at various current densities are shown in **Figure S3B**. The specific capacitances of KCu7S4/CFF and GN/KCu7S4/CFF SCs were calculated by the mass loading of KCu7S<sup>4</sup> and GN/KCu7S<sup>4</sup> NWs on the CFFs, respectively, and the results are shown in **Figure 4D**. The maximum specific capacitance of 408 F g−<sup>1</sup> at a current density of 0.5 A g−<sup>1</sup> for the GN/KCu7S4/CFF SC was calculated, which is two times higher than that of KCu7S4/CFF SC (167 F g−<sup>1</sup> ). The enhanced electrochemical performance of the GN/KCu7S4/CFF electrodes benefits from the following facts. First, the nanonetwork assembled by the graphite nanoparticles on the surface of KCu7S<sup>4</sup> nanowires improves the conductivity of the KCu7S<sup>4</sup> nanowires, which greatly increases the electron transmission rate. Secondly, these nanoparticles aggregated together to form a porous structure on the surface of the KCu7S<sup>4</sup> nanowires, which provides rich channels for ions to access to electroactive sites for fast and reversible redox reactions (Guan et al., 2015). The specific capacitance of the GN/KCu7S4/CFF SC in this work is higher than that of the previously reported for the hybrid SCs, such as 80.8 F g−<sup>1</sup> at 0.5 A g−<sup>1</sup> for the GNS/αMWCNT@PDAA SC (Sun et al., 2015), 56 F g−<sup>1</sup> at 0.58A g−<sup>1</sup> for the MSCS-O SC (Kim et al., 2015), 156 F g−<sup>1</sup> at 0.5A g−<sup>1</sup> for the PGpaper SC (Shu et al., 2015), and 189 F g−<sup>1</sup> at 0.5A g−<sup>1</sup> for the FeMnO3/RGO SC (Li et al., 2014). These results indicate that the electrochemical performance of the KCu7S<sup>4</sup> nanowires is improved by the successful coating of the graphite nanoparticles and this method can also be applied for other metal sulfides.

The EIS is measured in the frequency from 100 kHz to 1 Hz, and the Nyquist impedance plots of the KCu7S4/CFF and GN/KCu7S4/CFF SCs are shown in **Figure S4A**. In the high frequency range, the intercepts of the Nyquist curves on the real axis are about 2.43 and 2.18 for the

KCu7S4/CFF and GN/KCu7S4/CFF SCs, respectively, indicating better conductivity after coating the graphite nanoparticles. A smaller arc is observed for the GN/KCu7S4/CFF SC, which demonstrates an enhanced ion accessibility of the GN/KCu7S<sup>4</sup> nanowires compared with that of KCu7S<sup>4</sup> nanowires, due to the highly porous network structure. The Nyquist plots show almost a vertical line in the low frequency, indicating an excellent capacitive behavior of SC. To obtain more detailed information, the dependence of the phase angle on the frequency for the KCu7S4/CFF and GN/KCu7S4/CFF SCs are shown in **Figure S4B**. The relaxation time τ0(τ0=1/f<sup>0</sup> ) evaluated from the frequency at 45◦ impedance phase angle is 0.09 s for the GN/KCu7S4/CFF, which is shorter than that of the KCu7S4/CFF (0.14 s), revealing larger power response of the GN/KCu7S4/CFF SC (Liu et al., 2015).

Energy density (E) and power density (P) are two important parameters for evaluating the electrochemical performance of SCs (Lu et al., 2012). The energy density viruses the average power density is calculated from the charge-discharge curves (**Figure 5a**), which are estimated according to the following equations (Dai et al., 2014c).

$$E = \frac{CV^2}{2M} \tag{1}$$

$$P = \frac{E}{t} \tag{2}$$

where C, M, V, and t are the total capacitance of the device, effective mass of the electrode, voltage and the discharge time, respectively. The highest energy density of the GN/KCu7S4/CFF SC is 36 Wh kg−<sup>1</sup> at a power density of 201 W kg−<sup>1</sup> , which is higher than that of KCu7S4/CFF SC with the energy density of 14 Wh kg−<sup>1</sup> at a power density of 190 W kg−<sup>1</sup> . The maximum energy density of the GN/KCu7S4/CFF SC is higher than those previously reported, such as 6.3 Wh kg−<sup>1</sup> for the WL-MnO<sup>2</sup> SC (Yang et al., 2014b), 17 Wh kg−<sup>1</sup> for the MnFe2O4/graphene/polyaniline SC (Sankar and Selvan, 2015), 12.3 Wh kg−<sup>1</sup> for the MnO2@KCu7S<sup>4</sup> hybrid SC (Wang et al., 2014c), 22 Wh kg−<sup>1</sup> for the CoOH//VN SC (Wang et al., 2014b), and 1.46 Wh kg−<sup>1</sup> for the Al-doped α-MnO<sup>2</sup> SC (Hu et al., 2015).

For efficient energy storage devices, flexible, lightweight, and portable electronic devices are desired in practical applications. **Figure 5b** displays the high flexibility of asprepared GN/KCu7S4/CFF SC, and it can be folded and twisted without destroying its physical structure. Moreover, the CV curves of the GN/KCu7S4/CFF SC hardly change under different bending angles, indicating its good flexibility. For practical applications, it is necessary to connect SCs in series and/or in parallel to increase the operating voltage and/or current in some situations (Yuan et al., 2013). **Figure 5c** shows three GN/KCu7S4/CFF SCs connected in series can light 12 commercial light-emitting diodes (LEDs) for about 5 min after charging at 12 A g−<sup>1</sup> for 50 s (for detailed information, see **Supporting Information**). The excellent properties of the flexible GN/KCu7S4/CFF SC reveal a potential application in superior storage devices. In addition, the GN/KCu7S4/CFF SC exhibits a long-term cycling stability between 0 and 0.8 V at a current density of 2 A g−<sup>1</sup> and keeps 90% of its initial capacitance after 5,000 cycles (**Figure 5d**), revealing its good cycling life.

### CONCLUSION

In summary, we have successfully designed a porous and highly conductive nanonetwork structure electrode by coating graphite nanoparticles on the surface of the KCu7S<sup>4</sup> nanowires. Such a porous nanonetwork not only facilitates the diffusion of the electrolyte ions into the pseudocapacitive material, but also improved the electron transmission, which greatly enhance the charge storage efficiency. Moreover, a highly flexible all-solidstate hybrid SC based on the GN/KCu7S<sup>4</sup> nanowires is fabricated, which shows excellent electrochemical properties, including the high specific capacitance (408 F g−<sup>1</sup> ), high energy density (36 Wh kg−<sup>1</sup> ), and good cyclic stability. All the results indicate that such porous and highly conductive nanonetwork forming on nanostructured pseudocapacitive materials could improve the charge storage efficiency of supercapacitors.

### REFERENCES


### AUTHOR CONTRIBUTIONS

W-XS carried out the material preparation, electrochemical test, and analyzed the XRD, SEM, TEM, and Raman analysis. S-GD wrote the paper and J-MX discussed the results and revised the manuscript. Z-FZ attained the main financial support for the research and supervised all the experiments.

### ACKNOWLEDGMENTS

This work is supported by the National Natural Science Foundation of China (Grant No. 21805247, Grant No. 11704340), the China Postdoctoral Science Foundation (Grant No. 2018M630831), the Youth Teacher Start Fund of Zhengzhou University (Grant No. 11704340, Grant No. 32210813).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2018.00555/full#supplementary-material


electrochemistry. Chem. Mater. 10, 6–9. doi: 10.1021/cm97 05395


electrode for supercapacitors. J. Power Sources 275, 399–407. doi: 10.1016/j.jpowsour.2014.10.183


nanosheets@N-doped carbon nanotube array electrode for highperformance asymmetric supercapacitors. Energy Storage Mater. doi: 10.1016/j.ensm.2018. 06.026


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Shen, Xu, Dai and Zhang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Ion Selectivity and Stability Enhancement of SPEEK/Lignin Membrane for Vanadium Redox Flow Battery: The Degree of Sulfonation Effect

Jiaye Ye1,2†, Xuechun Lou1†, Chun Wu<sup>1</sup> , Sujuan Wu<sup>2</sup> , Mei Ding1,3 \*, Lidong Sun2,3 \* and Chuankun Jia1,3

*China, <sup>3</sup> National Engineering Laboratory of Highway Maintenance Technology, Changsha University of Science &*

*<sup>1</sup> College of Materials Science and Engineering, Changsha University of Science and Technology, Changsha, China, <sup>2</sup> State Key Laboratory of Mechanical Transmission, School of Materials Science and Engineering, Chongqing University, Chongqing,*

### Edited by:

*Qiaobao Zhang, Xiamen University, China*

#### Reviewed by:

*Du Yuan, Nanyang Technological University, Singapore Wenyue Li, Texas Tech University, United States Fan Xinzhuang, Hong Kong University of Science and Technology, Hong Kong*

#### \*Correspondence:

*Mei Ding dingmei2017@163.com Lidong Sun lidong.sun@cqu.edu.cn*

*†These authors have contributed equally to this work*

#### Specialty section:

*This article was submitted to Nanoscience, a section of the journal Frontiers in Chemistry*

Received: *31 August 2018* Accepted: *23 October 2018* Published: *12 November 2018*

#### Citation:

*Ye J, Lou X, Wu C, Wu S, Ding M, Sun L and Jia C (2018) Ion Selectivity and Stability Enhancement of SPEEK/Lignin Membrane for Vanadium Redox Flow Battery: The Degree of Sulfonation Effect. Front. Chem. 6:549. doi: 10.3389/fchem.2018.00549* A membrane of high ion selectivity, high stability, and low cost is desirable for vanadium redox flow battery (VRB). In this study, a composite membrane is formed by blending the sulfonated poly (ether ether ketone) with lignin (SPEEK/lignin), and optimized by tailoring the degree of sulfonation. The incorporation of lignin into the SPEEK matrix provides more proton transport pathway and meanwhile adjusts the water channel to repulse vanadium ions. The VRB cells assembled with the composite membranes exhibit high coulombic efficiency (∼99.27%) and impressive energy efficiency (∼82.75%). The cells maintain a discharge capacity of ∼95% after 100 cycles and ∼85% after 200 cycles at 120 mA cm−<sup>2</sup> , much higher than the commercial Nafion 212. The SPEEK/lignin composite membranes are promising for application in VRB system.

Keywords: vanadium flow battery, Sulfonated poly(ether ether ketone), SPEEK/lignin composite membrane, ion selectivity, degree of sulfonation

### INTRODUCTION

*Technology, Changsha, Hunan, China*

The vanadium redox flow battery (VRB) has attracted tremendous interest as a large-scale energy storage technique, for environment protection and sustainable development, in light of its long cycle life, fast response, flexible design, and great reliability via a cost-effective and eco-friendly means (Zhang et al., 2014a; Jia et al., 2015b; Yang et al., 2015; Ye et al., 2016; Lu et al., 2017; Wu et al., 2018, 2019). Proton exchange membrane is a key component in the flow battery, which performs as a separator to isolate the positive and negative electrolyte compartments, and meanwhile to conduct protons (Jia et al., 2012; Yu et al., 2016). An ideal membrane is expected to exhibit high proton conductivity, good chemical and mechanical stability, accurate ion selectivity, and low-cost fabrication approach (Jia et al., 2010; Ding et al., 2018; Yuan et al., 2018). To date, the commercial Nafion membranes have been widely used in VRBs, because of its good proton conductivity, remarkable chemical and mechanical stability (Li et al., 2014a; Dai et al., 2017). However, the high crossover rate of vanadium ions hampers its further application in VRBs (Zhang et al., 2018). Accordingly, several groups have been devoted to enhancing the performance of Nafion membranes by different methods, such as changing the casting solvent and annealing temperature

**120**

(Dai et al., 2017), altering pretreatment process (Jiang et al., 2016), employing surface modification (Teng et al., 2015), forming composite structure with organic materials, inorganic materials or both (Zeng et al., 2008; Mai et al., 2011; Teng et al., 2013, 2014). The modified membranes usually lower the permeability of vanadium ions. Nevertheless, the extremely high cost of the Nafion membranes is a critical barrier for VRB commercialization (Yuan et al., 2017). Therefore, it is appealing to explore alternative systems of high ion selectivity, good stability and low cost toward practical application.

The sulfonated hydrocarbon polymers and their derivatives are promising candidates as the substitutional membranes (Wang et al., 2013). The sulfonated poly (ether ether ketone; SPEEK) is of particular interest, in view of its low vanadium ion permeability, simple preparation, high chemical, and mechanical stability (Winardi et al., 2014; Jia et al., 2015a). More importantly, the SPEEK membrane is cost-effective, only accounting for about several tenths of the commercial Nafion (DuPont). To enhance the proton conductivity, organic or inorganic materials with abundant hydrophilic groups are usually introduced to form composite membranes. Moreover, the interaction between the additives and the SPEEK matrix also ensures a good chemical and mechanical stability under harsh condition during the VRB operation. Jia et al. (2015a) prepared a composite membrane by blending the SPEEK with functionalized carbon nanotubes. The membrane shows not only high coulombic efficiency (CE), voltage efficiency (VE), and energy efficiency (EE), but also good mechanical stability and low capacity loss, compared with the pristine SPEEK and Nafion 212 membrane. Other SPEEK-based composite membranes also exhibited excellent performances for VRB application, such as SPEEK/SPES [sulfonated poly (ether sulfone)] (Ling et al., 2012), SPEEK/GO (graphene oxide; Kong et al., 2016; Park and Kim, 2016), SPEEK/QPEI [quaternized poly(ether imide)] membranes (Liu et al., 2014, 2015).

Considering the cost of the additives, the lignin has recently been focused in our group, which is a byproduct in paper industry and bio-fuel producing process (Gong, 2016). The lignin possesses abundant hydroxyl groups and thus improves the wettability of the polymer matrix and promotes the proton conductivity of the blend membrane. **Figure 1** shows a general structure of lignin, in which the phenol propane unit is linked by alkyl–aryl, alkyl–alkyl and aryl–aryl ether bonds (Tolba et al., 2010; Ge et al., 2014; Zhang et al., 2014b; Zhu et al., 2016; Atifi et al., 2017; Rahman et al., 2018). The lignin interlaces with the SPEEK substrate and reduces the size of water channels. This gives rise to enhanced proton conductivity and meanwhile suppressed ion permeability. On the other hand, the degree of sulfonation (DS) in the SPEEK matrix, i.e., the amount of - SO3H groups, also affects the proton conductivity. In general, the conductivity increases with the DS. However, a high DS seriously influence the chemical and mechanical stability of the membrane (Xi et al., 2015). In this study, the DS effect on the ion selectivity and stability was systematically studied. An optimized degree of sulfonation was proposed toward the application in VRBs.

### EXPERIMENTAL

### Materials and Membranes Preparation

The lignin powders (Sigma-Aldrich) were soaked in 2 M hydrochloric acid solution (mass to volume ratio, 1 g per 20 mL) and stirred for 2 h. After that, sodium hydroxide solution was used to neutralize the above mixture, followed by filtration and freeze drying. The DS of SPEEK was controlled to about 41, 50, 59, and 72%, which were measured by titration method according to our previous work (Jia et al., 2015a). For the preparation of SPEEK/lignin composite membranes, 1.6 g of SPEEK was dissolved in 50 mL dimethyl sulfoxide (DMSO), and stirred at 60◦C until being dissolving completely. After cooling down to room temperature, 240 mg pretreatment lignin powder was added and vigorously stirred overnight. The mixture was casted on a home-made glass plate and dried in oven at 100◦C for solvent evaporation. After cooling, the membrane was peeled off from the glass and soaked in water immediately. Based on the mass ratio of lignin to SPEEK, the composite membranes were named as SPEEK41/L15, SPEEK50/L15, SPEEK59/L15, and SPEEK72/L15. Commercial Nafion 212 membrane (DuPont) was used as the reference. All of the reagents were used as received.

### Characterization

Fourier transform infrared spectroscopy (FT-IR, Thermo Fisher Nicolet iS10) was investigated in the range of 400–4,000 cm−<sup>1</sup> . The microstructure of the as-prepared membranes was examined with field-emission scanning electron microscopy (FESEM, Zeiss Auriga FIB/SEM). Water uptake (WU) was measured as follow: (1) dry membrane was soaked into distill water for 24 h; (2) the membrane was taken out and cleaned with filter paper immediately; (3) the cleaned membrane was weight by Mettler-Toledo analytical balance (ME204E). Swelling ratio (SR) was obtained by measuring the length variation of the membrane before and after immersing in deionized water for 24 h. The WU and SR of these membranes can be calculated through the following equations (Ling et al., 2012).

$$\text{WU} = \frac{W\_{\text{wet}} - W\_{\text{dry}}}{W\_{\text{dry}}} \tag{1}$$

$$\text{SR} = \frac{L\_{\text{wet}} - L\_{dry}}{L\_{dry}} \tag{2}$$

where the Wdry and Wwet are the weight of the membrane before and after soaking, respectively; the Ldry and Lwet are the length of the membrane before and after soaking, respectively.

The permeability of VO2<sup>+</sup> ion across the membranes was investigated as follow: (1) isolating two reservoirs that were filled 70 mL of 1.5 M VOSO<sup>4</sup> in 3.0 M H2SO<sup>4</sup> solution and 70 mL of 1.5 M MgSO<sup>4</sup> in 3.0 M H2SO4, respectively, with a membrane of 2.01 cm<sup>2</sup> active area; (2) stirring continuously and measuring the concentration of VO2<sup>+</sup> in MgSO<sup>4</sup> compartment at 24 h intervals by TU-1900 UV-vis spectrometer; (3) sampling with replacement to keep the solution volume stable. A typical experimental setup

FIGURE 1 | The general structure of lignin and schematic illustration showing the water channel in a composite membrane for vanadium redox flow battery.

was shown in **Figure 2**. The permeability value of the membrane can be calculated using Equation (3) (Jia et al., 2015a):

$$\mathbf{V}\frac{d\mathbf{C}(t)}{dt} = \mathbf{A}\frac{P}{L}(\mathbf{C} - \mathbf{C}(t))\tag{3}$$

where V, A, P, L, C, and C(t) are the volume of the VOSO<sup>4</sup> solution, the effective area of the membrane, the permeability of the vanadium ions, the thickness of the membrane, the initial concentration of VO2<sup>+</sup> in the VOSO<sup>4</sup> compartment, and the vanadium concentration in the MgSO<sup>4</sup> compartment at the moment t, respectively.

The area resistance (R) of membranes was investigated by a resistance tester (DME-20, DM, China). The electrolytes in both compartments were 1.5 M VOSO<sup>4</sup> and 3 M H2SO4, and the conductivity (σ) of the membrane can be calculated as follow (Ling et al., 2012):

$$
\sigma = \frac{L}{AR} \tag{4}
$$

where R represents the resistance difference between the cell with and without membrane, L is the thickness of the membrane, and A is the active area of membrane (13.5 cm<sup>2</sup> ).

### Cell Testing

The VRB single cell consisted of a composite membrane (13.5 cm<sup>2</sup> ) sandwiched between two carbon felt electrodes (13.5 cm<sup>2</sup> ), and two graphite polar plates (current collectors). The 1.5 M VO<sup>+</sup> 2 in 3.0 M H2SO<sup>4</sup> and 1.5 M V3<sup>+</sup> in 3.0 M H2SO<sup>4</sup> solutions were used as catholyte and anolyte, respectively. The cell performance was measured by Arbin battery testing system (BT-I, Arbin, USA) including open circuit voltage decay (OCV), long cycle charge-discharge, and rate performance. The OCV (75% state of charge) was terminated when the voltage of the testing cell declined below 0.85 V. The potential range was between 0.7 and 1.75 V at room temperature.

## RESULTS AND DISCUSSION

### Membrane Characterization

**Figure 3** shows the FT-IR spectra of lignin, pure SPEEK, and the composite membranes. The broad band at 3,400 cm−<sup>1</sup> is ascribed to the hydrogen bond and OH vibration. For lignin, the peaks at 2,938 and 2,849 cm−<sup>1</sup> are assigned to the C-H stretch in methyl and methylene group (-CH2-), and those at 1,594 and 1,511 cm−<sup>1</sup> are attributed to the aromatic rings of phenyl propane skeleton (characteristic bands of lignin; Faix, 1991). For SPEEK membrane alone, the following fingerprint absorption peaks are present: 3,428 cm−<sup>1</sup> (O-H stretching of -SO3H groups), 1,076 cm−<sup>1</sup> (symmetric stretching of O=S=O), and 706 cm−<sup>1</sup> (S-O stretching; Li et al., 2014b; Ma et al., 2018). For composite membranes, the intensity of the peaks for -SO<sup>−</sup> 3 (1,076, 1,020, and 706 cm−<sup>1</sup> ) increases with the DS, because of the increased amount of the -SO3H. More importantly, all of the composite membranes exhibit the characteristic peaks of lignin, indicating the incorporation of lignin into the SPEEK matrix. Compared with the pure SPEEK, the peaks of O-H stretching (3,248 cm−<sup>1</sup> ) shift toward the low frequency and the peak intensity decreases in the composite membranes. This suggests the hydrogen bond interaction between the -OH groups of lignin and the -SO3H groups of SPEEK.

**Figures 4A,B** reveals that the lignin powders are homogeneously dispersed in the SPEEK solution without any precipitates, even after 240 h. This is a prerequisite for forming a uniform composite membrane. **Figures 4C,D** displays the resulting SPEEK and SPEEK50/L15 membranes which exhibit uniform and dense surface. This would hamper the crossing of vanadium ions and enhance the cycle stability of the membrane. Moreover, the lignin particles are homogeneously embedded into the polymer matrix and can improve the wetting property of the composite membrane. The 3D structure of lignin also provides more pathways for protons transport and reduces the water channels to repulse vanadium ions. Therefore, the lignin is introduced into the SPEEK matrix, which can enhance the proton conductivity while inhibit the vanadium ions permeation.

The proton transport generally proceeds via the vehicle and Grotthuss mechanism in the membrane. As such, the water uptake is a critical property. It has been demonstrated that the proton conductivity of the proton exchange membrane enhances with the WU amount. However, a high water uptake usually results in low mechanical stability. **Table 1** shows that the WU of SPEEK/lignin membranes increases with the DS, as more hydrophilic -SO3H groups enhance the wetting property of the membrane. Similarly, the SR displays the same tendency as the WU.

The permeability of the resulting membranes is shown in **Figure 5**. The VO2<sup>+</sup> permeability of SPEEK/lignin membranes increases with the degree of sulfonation. This is attributed to the high DS that generally imparts abundant -SO<sup>−</sup> 3 groups in the polymer matrix, which drastically improve the proton conductivity (see **Table 1**) and also accelerate the crossover of vanadium ions through the membrane. The ion selectivity (Zhang et al., 2014a; Ji et al., 2017), namely the ratio of proton conductivity to ion permeability, is widely used to describe the balance between the two processes. **Figure 5B** shows that the ion selectivity of the composite membranes is much higher than that of the Nafion 212, regardless of the DS in SPEEK, because of the suppressed permeability of VO2<sup>+</sup> ion. This clearly demonstrates the advances of the composite system. In particular, the SPEEK59/L15 membrane exhibits the best performance (61.96 × 10<sup>4</sup> S min cm−<sup>3</sup> ), nearly five-fold increment as compared to the Nafion 212 (12.78 × 10<sup>4</sup> S min cm−<sup>3</sup> ). The incorporation of lignin into the SPEEK substrate provides more proton transport pathway and meanwhile adjusts the water channels to repulse vanadium ions, thus giving rise to high ion selectivity.

### Cell Performances

To systematically study the performance of the composite membranes, the OCV, long cycle process, and rate performance of VRB single cells were carried out under different conditions. The OCV is a critical parameter to verify the vanadium ions cross rate in the membranes, as the vanadium ions crossing the membrane results in self-discharge and therefore the cell voltage declines accordingly. **Figure 6A** reveals that the OCV

FIGURE 4 | Photographs of SPEEK solution and SPEEK/lignin mixture in DMSO: (A) initial and (B) after 240 h; SEM images of (C) pure SPEEK and (D) SPEEK50/L15 membranes.


curves decrease rapidly with the enhanced DS in the composite membrane. It is obvious that the voltage decay (above 0.85 V) of cells assembled with the composite membranes is much slower than that of Nafion 212 (288.6 vs. 14.05 h for SPEEK41/L15 vs. Nafion). This indicates that the SPEEK/lignin composite membrane efficiently suppresses the permeation of vanadium ions, in good agreement with the results in **Figure 5**.

**Figure 6B** displays the typical charge-discharge curves of the VRB cells using different membranes under current density of 120 mA cm−<sup>2</sup> . As a high area resistance (see **Table 1**) usually

FIGURE 6 | (A) Open circuit voltage decay of different membranes, (B) the charge-discharge performance of VRB cells employing different membranes at 120 mA cm−<sup>2</sup> .

leads to high ohmic polarization, the average charge voltage of cells with SPEEK/lignin membranes is slightly higher than that of the Nafion 212, with the average discharge voltage being on the contrary. However, the SPEEK59/L15 membrane exhibits the best discharge capacity, as a result of high ion selectivity, consistent with the above discussion.

**Figure 7** shows the cycling performance of cells assembled with SPEEK59/L15 and Nafion 212 membranes from 50 to 320 mA cm−<sup>2</sup> . The Coulombic efficiency of the cells with SPEEK59/L15 (up to 99.56%) is higher than that of Nafion 212 over the whole rate range. This originates from the reduced vanadium ion permeability with the composite membranes. It is noted that the CE of both cells increases with the current density. This is mainly due to the shortened charge-discharge time under high current density, which suppresses the crossover of vanadium ions through the membrane. **Figure 7B** reveals that the discharge capacity of SPEEK59/L15 is higher than that of the Nafion 212 under current density ≤250 mA cm−<sup>2</sup> . This can be attributed to the overpotential and ohmic polarization under high current. Therefore, the better performance of SPEEK59/L15 is ascribed to the synergistic effect from the lignin additives and SPEEK matrix, which provides more proton transport pathway and meanwhile suppresses the crossover of the vanadium ions.

**Figure 8** displays the stability performance of the corresponding VRB cells operated at 120 mA cm−<sup>2</sup> . The Coulombic efficiency decreases with the degree of sulfonation (**Figure 8A**). This agrees well with the varying trend of VO2<sup>+</sup> permeability (**Figure 5A**), as a high DS improves the proton conductivity but accelerates the crossover of vanadium ions through the membrane. As such, the voltage efficiency increases with the DS (**Figure 8B**). Accordingly, the cells with SPEEK59/L15 membranes exhibit an impressive energy efficiency (EE = CE × VE, up to 82.75%), comparable to that of the Nafion 212. This is attributed to the best ion selectivity for the SPEEK59/L15 membrane. It is noteworthy that the cells with SPEEK72/L15 fail after only 29th cycles, as a high DS deteriorates the mechanical and chemical stability of the membrane.

The capacity retention of VRB cells is a key factor to measure the ion imbalance in the operation process. **Figure 8D** discloses that the discharge capacity retention of the cells assembled with Nafion 212 maintains just 55.96% after 100 cycles under 120 mA cm−<sup>2</sup> . In contrast, under the same conditions, the cells with SPEEK59/L15 keep about 94.80% after 100 cycles, and more than 85% after 200 cycles. This demonstrates that the SPEEK59/L15 membrane substantially suppresses the crossover of vanadium ions and thus enhances the stability and prolongs the cycle life. The cells with the SPEEK59/L15 membranes show the best capacity retention and outperform the other ones. The SPEEK59/L15 membranes synthesized by an eco-friendly and cost-effective approach exhibits high ion selectivity and excellent stability, making it a promising candidate for efficient VRB system.

### CONCLUSIONS

The SPEEK/lignin composite membranes were optimized by controlling the degree of sulfonation toward the VRB application. The VRB cells with SPEEK59/L15 membranes exhibit an impressive energy efficiency up to 82.75%, low vanadium ion permeability, high ion selectivity, and high capacity retention (94.80% after 100 cycles and over 85% after 200 cycles). The good performance is assigned to the synergistic effect from the lignin additives and SPEEK matrix, which improves the proton conductivity and suppresses the crossover of the vanadium ions. The eco-friendly and cost-effective composite membranes make it a competent candidate for VRB energy storage technique.

## AUTHOR CONTRIBUTIONS

MD, LS, and CJ conceived the idea and supervised the work. JY and XL synthesized the membranes and performed most of the experiments. JY, XL, CW, and SW prepared the samples and analyzed the data. JY, XL, MD, LS, and CJ participated in analyzing the data and wrote the report. All authors read and approved the final manuscript.

### ACKNOWLEDGMENTS

We acknowledge the support from the 100 Talented Team of Hunan Province.

## FUNDING

The project was supported by Open Fund of National Engineering Laboratory of Highway Maintenance Technology, Changsha University of Science & Technology (kfj170105, kfj170107). LS was grateful for the financial support from the National Natural Science Foundation of China (No. 51871037, 51501024), the Fundamental Research Funds for the Central Universities (No. 2018CDQYCL0027), and the Chongqing Entrepreneurship and Innovation Program for the Returned Overseas Chinese Scholars (No. cx2017035).

### REFERENCES


biological devices, and energy applications. Chem. Rev. 116, 9305–9374. doi: 10.1021/acs.chemrev.6b00225

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Ye, Lou, Wu, Wu, Ding, Sun and Jia. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Energy Storage and Thermostability of Li3VO4-Coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> as Cathode Materials for Lithium Ion Batteries

#### Liubin Song<sup>1</sup> , Fuli Tang<sup>1</sup> , Zhongliang Xiao<sup>1</sup> \*, Zhong Cao<sup>1</sup> and Huali Zhu<sup>2</sup>

*<sup>1</sup> Hunan Provincial Key Laboratory of Materials Protection for Electric Power and Transportation, School of Chemistry and Biological Engineering, Changsha University of Science and Technology, Changsha, China, <sup>2</sup> School of Materials Science and Engineering, Changsha University of Science and Technology, Changsha, China*

### Edited by:

*Jiexi Wang, Central South University, China*

### Reviewed by:

*Xiqian Yu, Institute of Physics (CAS), China Feixiang Wu, Max Planck Institute Stuttgart, Germany Chenghao Yang, South China University of Technology, China*

\*Correspondence:

*Zhongliang Xiao csustslb@csust.edu.cn*

#### Specialty section:

*This article was submitted to Physical Chemistry and Chemical Physics, a section of the journal Frontiers in Chemistry*

Received: *14 August 2018* Accepted: *19 October 2018* Published: *08 November 2018*

#### Citation:

*Song L, Tang F, Xiao Z, Cao Z and Zhu H (2018) Energy Storage and Thermostability of Li3VO4-Coated LiNi0.8Co0.1Mn0.1O2 as Cathode Materials for Lithium Ion Batteries. Front. Chem. 6:546. doi: 10.3389/fchem.2018.00546* The electrochemical performances and thermostability of LiNi0.8Co0.1Mn0.1O<sup>2</sup> is affected by temperature. High ambient temperature or irregular heat distribution accelerates the decline of LiNi0.8Co0.1Mn0.1O<sup>2</sup> performance, shortens cathode material life. In this work, the energy storage and thermostability of the Li3VO4-coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material were studied for the first time by electrochemical calorimetry methode at different temperatures and rates. Results show that Li3VO4-coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material has excellent rate and cycle performance. The thermal electrochemical experiments further show that the thermal stability of Li3VO4-coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material in charge-discharge energy storage and conversion system is better than LiNi0.8Co0.1Mn0.1O<sup>2</sup> at 30, 40, and 50◦C. The enhanced performance can be attributed to the fact that Li3VO<sup>4</sup> coating promotes the transmission of lithium ions and protects the active material from electrolyte corrosion at different temperature, as well as reduces side reaction, electrode polarization and heat generation of cathode materials. The Li3VO4-coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material has excellent energy storage properties and thermostability, which are beneficial to the development of electronic equipment.

Keywords: LiNi0.8Co0.1Mn0.1O2 , cathode material, electrochemical calorimetry, thermo-electrochemistry performance, thermostability, energy storage

### INTRODUCTION

As the problem of energy storage and security becomes more and more serious, the research on new energy materials becomes more and more urgent. New energy materials play a vital role in the sustainable development of human society. As the main or auxiliary power source of new energy vehicles, lithium-ion batteries cathode materials have become an indispensable part of the development of new energy vehicles (Anseán et al., 2013; Pan et al., 2018). However, the hightemperature safety of high-capacity and high-power lithium-ion batteries for automobiles has attracted considerable attention with the widespread application of lithium-ion batteries in electric vehicles (Konishi et al., 2013). Considering the increase in energy density of electric passenger power batteries, LiNi0.8Co0.1Mn0.1O<sup>2</sup> has a high specific capacity and is expected to become the mainstream of power energy batteries. However, high nickel in the LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material causes several difficulties, such as thermal instability, low diffusion coefficient of lithium ions and severe capacity attenuation (Li et al., 2006; Xiong et al., 2013; Nitta et al., 2015). Currently, the commonly used means to solve the above problems are surface modification (Cho and Cho, 2010; Xiong et al., 2012; Lu et al., 2014; Zheng et al., 2015), ion doping (Luo et al., 2010; Du et al., 2015; Chen et al., 2017), core shell and gradient structure construction (Wen et al., 2015; Liao et al., 2016), and electrolyte optimisation of electrolyte (Plichta and Behl, 2000). Surface modification is the most common means used in Li3VO<sup>4</sup> modification its ion migration skeleton structure can promote lithium ion transport. Surface modification has a broad application prospect in lithium-ion battery cathode materials. Huang et al. (2014); Zhang et al. (2017), and Wang et al. (2015) used vanadate compounds to modify the surface of LiNi0.8Co0.1Mn0.1O2. They showed that the modified LiNi0.8Co0.1Mn0.1O<sup>2</sup> formed on the surface of cathode material with a stable interfacial film, lithium ion diffusion and improved electronic transmission and the reduced charge transfer resistance remarkably improved interfacial electrochemistry reaction activity. Although the electrochemical performance of lithium-ion batteries by Li3VO<sup>4</sup> modification has remarkably improved, many safety problems are still observed on lithium-ion batteries for vehicle power supply, especially the high temperature of batteries, which limit their wide application in electronic equipment and electric vehicles (Anseán et al., 2016; Liu et al., 2016). Therefore, the thermo-electrochemistry and high-temperature energy storage properties of LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode materials modified by Li3VO<sup>4</sup> should be investigated.

The thermo-electrochemical method is a combination of electrochemical, thermodynamic, physical, and chemical methods, which is used to analyse the battery charge and discharge performance and corresponding heat production at different temperatures (Song et al., 2013). This method can not only study the electrochemical performance of the battery under different charging and discharging states, but also obtain the data of current, heat flow and voltage change over time through the LAND battery test system and isothermal calorimeter, and obtain the thermo-electrochemical parameters, providing new theoretical basis for solving the battery safety problems. Currently, researchers have extensively investigated the electrochemical performance of surface-modified lithiumion batteries. However, the research on thermo-electrochemical and high-temperature performance should be improved. In recent years, the thermal stability, reaction kinetics and apparent activation energy of LiNi0.8Co0.1Mn0.1O<sup>2</sup> and other cathode materials have been analyzed by differential thermal analysis, thermogravimetric analysis, and differential scanning calorimetry. However, these analyses failed to reflect the electrochemical properties at the same time (Lee et al., 2009; Fu et al., 2014; Peng and Jiang, 2016). Traditionally, the performance of electrode materials is evaluated by using charge and discharge capacity, cycle performance and rate performance, which cannot reflect their heating characteristics and high-temperature energy storage performance. Local and global studies have been conducted to couple the battery charge and discharge test device and calorimeter to investigate the temperature change or heat generation of batteries during charge and discharge cycles at different current densities (Ye et al., 2012; Xiao et al., 2014; Mcturk et al., 2018; Song et al., 2018). Eddahech et al. (2013) used electrochemical-calorimetry to evaluate the thermal effects of high-capacity commercial nickel– cobalt–manganese–lithium batteries and verified the feasibility of electrochemical-calorimetry in the thermal electrochemical study of lithium-ion batteries. Du et al. (2017) analyzed the LiFePO<sup>4</sup> battery evolving law of irreversible heat production and its different rates and granular component of thermal analysis to provide effective guidance for the battery thermal management system design. Krause et al. (2012), Vallverdu et al. (2016), and Balasundaram et al. (2016) assessed the thermal behavior and electrochemical properties of lithium-ion batteries under different conditions by using a calorimeter combined with a battery charging and discharging test device. To date, considerable studies have been conducted on the thermoelectric properties of pure-phase cathode materials by electrochemical-calorimetry, whereas few studies have been performed on materials modified by coating. However, the high-temperature discharge and heat release have become serious with the rapid application of lithium-ion batteries, and the internal heat problem of high discharge rate batteries should be addressed. Therefore, the thermo-electrochemistry energy storage and thermostability of modified LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode materials should be investigated.

In this study, LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material was coated with 3 wt.% Li3VO4, and the thermo-electrochemistry energy storage and thermostability of LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode materials before and after modification were investigated using electrochemical-calorimetry combination technology. The combined thermodynamic and electrochemical methods in analyzing the LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material before and after modification is conducive for the optimisation of the thermal design and material development of the battery system and improvement of the safety performance of the battery. These conditions are of immense scientific significance to the research on high-performance LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode materials.

### EXPERIMENTAL

### Materials Preparation

The Ni0.8Co0.1Mn0.1(OH)<sup>2</sup> precursor was mixed with LiOH·H2O at a molar ratio of 1:1.05, completely ground in an agate mortar and was calcined at a high temperature in an oxygen atmosphere to obtain the lithium-ion battery LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material. Then, a mechanical fusion method was used to add LiOH·H2O and V2O<sup>5</sup> at a molar ratio of 3:1 to the ethanol solution for mixing and grinding. After the LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material was added for grinding, and the completely ground material was calcined at 700◦C for 8 h. The Li3VO4-coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material was obtained.

### Battery Assembly

The weights of positive electrode material, acetylene black and polyvinylidene fluoride were 0.3200, 0.0400 and 0.0400 g based on the mass ratio of 8:1:1. Then, in N-methyl-2-pyrrolidone solvent grinding for 20 min, the uniformity of viscous liquid was regularly coated on the aluminum foil and was placed in 80◦C oven drying after 4 h to make a positive plate. Button batteries (type 2025) were assembled in a glove box filled with argon gas (99.999%) by using an anode shell, a cathode shell, a cathode piece, an anode piece (Li), a nickel net, a membrane (Celgard 2300), and an electrolyte [1 mol· L <sup>−</sup><sup>1</sup> LiPF<sup>6</sup> (DMC+EMC+EC (volume ratio 1:1:1))]. Electrochemical cyclic voltammetry (CV) test were conducted on the CHI760E electrochemical system.

### Material Analysis and Characterization

The X-ray powder diffraction (XRD, Rint-2000, Rigaku) was analyzed to the phase structure of the pristine and Li3VO4 coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> samples. A scanning electron microscope (SEM, JEOL, JSM-5612LV) was used to analyse the microstructure of materials. The microstructure of the sample and surface coating of the material were observed by using a transmission electron microscope (HRTEM, Titan G2 60-300 with image corrector, 200 kV).

## Performance Test of Electrochemical-Calorimetry Combination

The thermo-electrochemical properties of LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode materials before and after modification were analyzed by using LAND test system and TAM air calorimeter. The battery was placed in an ampoule that contains methyl silicone oil. The positive and negative electrodes of the battery were connected to a LAND test system by long copper wires and were placed in a calorimeter at 30, 40, and 50◦C. The LAND test system has a charging and discharging voltage range of 2.5–4.3 V and a charge and discharge multiplier of 0.1, 0.5, 1, and 2◦C in evaluating the battery's rate performance and cycle performance at different temperatures. The TAM air calorimeter is an eight-channel milliwatt-scale thermal conductivity isothermal calorimeter. The calorimeter is calibrated, the calibration constant is obtained, the experimental data are calibrated and the heat flow generated by the battery during charging and discharging is recorded.

### RESULTS AND DISCUSSION

### Morphology and Interface Analysis

In order to study the effect of Li3VO4-coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode materials, XRD was used to analyze the structure of LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode materials before and after Li3VO<sup>4</sup> coating, as shown in **Figure 1**. It can be seen from figure that the XRD pattern of the sample diffraction peak is consistent with the diffraction peak of R3m space group and has the structure of α-NaFeO2. The diffraction peak of Li3VO4-coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material is quite sharp, indicating that the material has good crystallinity. The peak splitting of (108) and (110) is obvious, indicating that a better layered structure has been formed. According to the diffraction data, the I(003)/I(104) values of cathode materials were 1.265 (pristine), and 1.399 (coated). In cathode materials, I(003)/I(104)>1.2, indicating that the cation mixing degree of active materials is small. In conclusion, the Li3VO4-coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material will not change the

layered phase or layered framework and can effectively inhibit the mixing degree of cation and improve the properties of the material.

The SEM images of all samples are illustrated in **Figures 2a–d**, which is mainly used to analyse and evaluate the surface morphology of the sample. As shown in **Figures 2a–d**, LiNi0.8Co0.1Mn0.1O<sup>2</sup> and Li3VO4-coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode materials have spheroid structure, and their spheroid size is approximately 8µm. Li3VO<sup>4</sup> can be filled tightly on LiNi0.8Co0.1Mn0.1O<sup>2</sup> particles because spheroid particles have good fluidity. **Figures 2c,d** show that the shape and size of the LiNi0.8Co0.1Mn0.1O<sup>2</sup> material after coating slightly change, and the Li3VO4-coated surface becomes rough, which is the sample close between particles and particles. This condition may be due to the melted Li3VO<sup>4</sup> gathered in the modification process of heat treatment. In addition, we conducted TEM analysis of LiNi0.8Co0.1Mn0.1O<sup>2</sup> and Li3VO4-coated LiNi0.8Co0.1Mn0.1O2, as shown in **Figures 2e,f**. As can be seen from the **Figure 2f**, there is a clear Li3VO<sup>4</sup> layer on the surface of the coated sample, and the thickness of the coating is about 1∼2 nm, which indicates that Li3VO<sup>4</sup> is successfully coated on the surface of LiNi0.8Co0.1Mn0.1O2. The schematic illustration of the synthesis of Li3VO4-coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material is shown in **Figure 2g**. A mechanical fusion method is applied in the LiNi0.8Co0.1Mn0.1O<sup>2</sup> ethanol soluble solution mixed with LiOH·H2O and V2O<sup>5</sup> grinding, and the Li3VO4 coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material is obtained under 700◦C calcination. To study the degree of material coating, we further analyse the Li3VO4-coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> HAADF-STEM of cathode materials and the mapping diagram of Ni, Co, Mn, and V elements, as shown in **Figure 3**. The element mapping signal is obtained by scanning the HAADFrectangle in the STEM. As shown in **Figure 3**, Ni, Co, Mn, and V elements are uniformly distributed in the material, which indicate that Li3VO<sup>4</sup> is uniformly coated on the surface of LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material. In Li3VO4-coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material, Li3VO<sup>4</sup> acts as a fast ion

FIGURE 3 | HAADF-STEM images of Li3VO4-coated LiNi0.8Co0.1Mn0.1O2 cathode materials and Mapping of Nickel, Cobalt, Manganese, and Vanadium elements.

materials.

conductor layer that promotes lithium ion transport, reduces the electrolyte between the active material and side effects and improves the thermal stability and electrochemical performance of materials.

### Electrochemical Properties

**Figure 4** shows the cyclic voltammetry curve of LiNi0.8Co0.1Mn0.1O<sup>2</sup> and Li3VO4-coated LiNi0.8Co0.1Mn0.1O2. It can be seen from **Figure 4** that all the peaks are similar and no significant new redox peak appears after Li3VO<sup>4</sup> coating, indicating that the modification by Li3VO<sup>4</sup> does not change the main structure of LiNi0.8Co0.1Mn0.1O2. The potential difference of the redox peak of Li3VO4-coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> is 0.166 V and the pure LiNi0.8Co0.1Mn0.1O<sup>2</sup> sample is 0.272 V, which indicated Li3VO4-coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> has smaller potential difference. The larger the potential difference between the deimmobilization and embedding of lithium ions, the greater the electrode polarization effect. It indicating that the Li3VO<sup>4</sup> surface layer inhibits the direct contact between the active material and the electrolyte, enhances the lithium ion diffusion between the electrode/electrolyte interface, and improves the electrochemical performance of the battery. At the same time, the peak area of the cyclic voltammetry curve was integrated. It was found that after Li3VO4-coated LiNi0.8Co0.1Mn0.1O2, the peak area of the first cycle were closer to the second and third cycles, indicating that Li3VO4-coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material has better reversibility. Compared with pure LiNi0.8Co0.1Mn0.1O2, Li3VO4-coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material has better lithium diffusion kinetics.

**Figure 5** depicts the initial charge/discharge capacities curve of the pristine and Li3VO4-coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material at 0.1◦C. All the samples show initial sloping region and smooth charge/discharge curves with similar capacities. The initial discharge capacities of pristine samples were 200.6 (30◦C), 201.6 (40◦C) and 207.0 mAh g−<sup>1</sup> (50◦C). The initial discharge capacities of Li3VO4-coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode materials were 193.2 (30◦C), 195.9 (40◦C), and 203.1 mAh g−<sup>1</sup> (50◦C). The higher the ambient temperature, the more adequate the initial reaction of the battery, so the initial discharge capacity increases slightly as the temperature increases. **Figure 6** shows the rate performance of LiNi0.8Co0.1Mn0.1O<sup>2</sup> and Li3VO4 coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode materials under different rates (0.1, 0.5, 1, and 2◦C) at 30, 40, and 50◦C. As shown in the figure, the discharge capacity decreases with increased current density in the charging and discharging processes. The lithiumion battery shows a high initial capacity when the temperature increases from 30 to 50◦C. However, the cycle stability of the battery decreases and capacity decay rate increases. It can be seen from the figure that LiNi0.8Co0.1Mn0.1O<sup>2</sup> has the best rate performance at 30◦C, and Li3VO4-coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> has the best rate performance at 40◦C, indicating that Li3VO4 coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material is more suitable at high temperatures. The increase in electrode polarization during charging is the main cause for the rapid decline of battery capacity at high temperature. The discharge specific capacity is similar to the specific discharge capacity at 1◦C for the first time when the material is charged and discharged after returning to 1◦C at 0.1, 0.5, 1, and 2◦C, which indicates that the Li3VO4-coated cathode materials does not affect the reversibility of LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material. In addition, the temperature increases, and the discharge capacity of Li3VO4 coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode materials is remarkably better than that of pure LiNi0.8Co0.1Mn0.1O<sup>2</sup> with increased current density. This condition is because the reaction inside the battery accelerates and the side effects increase with increased temperature. However, in Li3VO4-coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material, Li3VO<sup>4</sup> promotes lithium ion transport, reduces the high-temperature electrolyte between the active material and side effects and improves the stability of materials under high-temperature conditions. Therefore, the Li3VO4 coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material has the best rate performance.

**Figure 7** shows the cycle performance of LiNi0.8Co0.1Mn0.1O<sup>2</sup> and Li3VO4-coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material under different temperatures (30, 40, and 50◦C) at 1◦C for 50 cycle times. As shown in figure, the Li3VO4-coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material exhibit an excellent cycle stability after three cycles of activation at 0.1◦C. The sharp reduction in the discharge capacities after three cycles is caused by the changes in the current density and temperature. The specific discharge capacities of LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode materials at 30, 40, and 50◦C at 1◦C are 197.1, 176.8, and 172.8 mAh·g −1 , respectively. The capacity retention ratios are 93.2, 91.4, and 80.4% after 50 charge/discharge cycles. For the Li3VO4-coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material, the specific discharge capacities are 194.2 (30◦C), 175.5 (40◦C), and 168.1 (50◦C) mAh·g −1 after 50 times of charge/discharge cycle capacity retention of 96.7, 95.9, and 91.4%, respectively. With the increase of temperature, the decay rate of LiNi0.8Co0.1Mn0.1O<sup>2</sup> is large, while the Li3VO4-coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material is relatively stable. This condition is because the lithium ion diffusion of pure LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material is relatively slow and cannot maintain with the electron transfer rate. Thus, electrode electrochemical polarization occurs, which causes capacity loss and performance degradation. In addition, passive film formation is accelerated in the cycle process by accelerating the oxidation electrolyte decomposition of LiNi0.8Co0.1Mn0.1O<sup>2</sup> between the cathode materials and electrolyte and the adverse event that occurs between the binder and electrolyte. The thick organic passivation film on the surface of cathode material particles leads to increased interface impedance of anode/solution, which decreases the battery's high-temperature circulating capacity. After the modification of LiNi0.8Co0.1Mn0.1O<sup>2</sup> by Li3VO<sup>4</sup> coating, the diffusion rate of lithium ions is accelerated and the interface impedance of the positive/solution is reduced, which is consistent with the analysis results of rate performance. Therefore, the capacity retention rate of LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material coated with Li3VO<sup>4</sup> is remarkably improved with excellent cycling performance.

### Thermoelectric Chemical Properties

In order to analyze Li3VO4-coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material further, we used electrochemical calorimetry to analyze the thermo-electrochemical properties. **Figures 8**–**10** show the

heat-time and voltage-time curves of LiNi0.8Co0.1Mn0.1O<sup>2</sup> and the Li3VO4-coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material at different temperatures (30, 40, and 50◦C) and different rates (0.1, 1, and 2◦C). As shown in the heat flow time curve in **Figures 8**–**10**, the heat flow curve of the cathode material in the charging and discharging processes shows many exothermic peaks with the change of time under the low rate of 0.1◦C. However, the heat flow curve shows obvious exothermic peak and no impurity peak under 1 and 2◦C. This condition is because the battery can be approximated as a reversible process during charge and discharge and the reversible heat and irreversible heat are relatively close in at a low rate. In this case, the exothermic peak under 1 and 2◦C is generated by the superposition of electrode polarization and battery chemical reaction. However, the heat generated by electrode polarization is dominant. In general, the polarization heat of the battery gradually increases with increased magnification. Thus, the polarization heat exceeds the chemical reaction heat of the battery, and the peak of reaction heat can be

completely covered. Therefore, the exothermic peak under high rate is generated by the superposition of electrode polarization and battery chemical reaction, and the heat generated by electrode polarization is dominant. To determine the thermal electrochemical properties of cathode materials before and after coating, the electric quantity and heat generated during the charging and discharging processes of the battery are obtained by integrating the current-time and heat flow-time curves. The formula is expressed as follows:

$$\mathbf{Q}\_{\rm ch} = \int\_{0}^{t\_{\rm l}} \mathbf{i(t)} dt,\tag{1}$$

$$\mathbf{Q}\_{\text{disch}} = \int\_{0}^{t\_{\text{l}}} \mathbf{i}(\mathbf{t}) d\mathbf{t},\tag{2}$$

$$\mathbf{Q}\_{\text{total}} = \mathbf{Q}\_{\text{ch}} + \mathbf{Q}\_{\text{disch}},\tag{3}$$

$$\mathbf{q}\_{\rm ch} = \int\_{0}^{t\_1} h(t)dt,\tag{4}$$

$$\mathbf{q}\_{\text{disch}} = \int\_0^{t\_2} h(t)dt,\tag{5}$$

$$\mathbf{q}\_{\text{total}} = \mathbf{q}\_{\text{ch}} + \mathbf{q}\_{\text{disch}},\tag{6}$$

$$
\Delta q = q\_{ch} - q\_{disch}, \tag{7}
$$

where Qtotal represents the amount of electricity consumed during charge (Qch) and discharge (Qdisch) of the battery, expressed in C, i(t) represents the change of current with time, expressed in mA, t is the time of the entire charge and discharge processes of the battery, expressed in s, qtotal is the charge (qch), and discharge (qdisch) of the battery heat production with time, 1q is the difference between the amount of charge generated by the battery and the amount of discharge, expressed in mJ, and

FIGURE 9 | The change curves of heat flow and voltage of LiNi0.8Co0.1Mn0.1O2 (A–C) and the Li3VO4-coated LiNi0.8Co0.1Mn0.1O2 (D–F) cathode materials with time under 40◦C at different rates.

h(t) represents the change value of heat flow with time, expressed in mW. On the basis of the total amount of electricity consumed by the battery during charging and discharging, the total number of moles of reaction can be calculated. The formula is expressed as follows:

$$\ln = \frac{Q}{F}.\tag{8}$$

where n is the total number of moles of the reaction, F is Faraday's constant and its value is 96 485◦C·mol−<sup>1</sup> . The total enthalpy change (1rHm) of chemical reaction can be calculated based on the relationship between the total heat produced by the battery during charging and discharging and the total number of moles of reaction, which is expressed in kJ·mol−<sup>1</sup> . Entropy (1rSm) is a measure of the disorder degree of the reaction system, which is expressed in J·mol−<sup>1</sup> ·K −1 . The formula is expressed as follows:

$$
\Delta\_I H\_m = \frac{q}{n},
\tag{9}
$$

$$
\Delta\_r \mathcal{S}\_m = -\frac{\Delta q}{2\pi T} = -\frac{q\_{ch} - q\_{disch}}{2\pi T}.\tag{10}
$$

As shown in **Table 1**, the heat yield of lithium ion battery cathode material increases when the temperature increases from 30 to 50◦C. This condition is because the temperature inside the battery increases, and the redox reaction speed increases with increased temperature. At this time, the side reaction accompanying the battery is intense, and the reaction heat of the battery side is increased, which increases the total heat of

TABLE 1 | Thermodynamic parameters for LiNi0.8Co0.1Mn0.1O2 and Li3VO4-coated LiNi0.8Co0.1Mn0.1O2 cathode materials.


*"–" stands for exothermic.*

FIGURE 11 | Curve of enthalpy change with rate for LiNi0.8Co0.1Mn0.1O2 (A) and Li3VO4-coated LiNi0.8Co0.1Mn0.1O2 (B) cathode materials at different temperatures.

the battery. The heat production of the material before and after Li3VO<sup>4</sup> coating is similar at 30 and 40◦C. However, the heat yield of Li3VO4-coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material under 50◦C at 1 and 2◦C is less than LiNi0.8Co0.1Mn0.1O2. This condition is due to the high-temperature coating layer that protects the material from electrolyte erosion and reduces the side effects between the electrolyte and cathode material, which reduces the battery heat yield and improves the thermal stability and security of materials.

**Figure 11** shows the change curve of enthalpy of LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode materials with current rate at different temperatures before and after Li3VO<sup>4</sup> coating. At the same temperature, the enthalpy increases with increased rate. At the same current rate, the enthalpy increases with increased temperature. The greater the change in enthalpy is, the worse safety condition of the battery will be. The difference between the Li3VO4-coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> and pure LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material is 12.19 kJ·mol−<sup>1</sup> when the temperature is 50◦C and the charge–discharge rate is 2◦C. The Li3VO4-coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material has a low enthalpy change, which indicates that the thermal stability of the material is high and secure at high temperatures.

**Figure 12** shows the entropy change with temperature for LiNi0.8Co0.1Mn0.1O<sup>2</sup> (**Figure 12A**) and Li3VO4-coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> (**Figure 12B**) cathode materials at different rates. In the reaction system, the entropy increases with increased temperature and rate. The second law of thermodynamics states that any change or chemical reaction in an isolated system is constantly in the direction of an increase in entropy. Under this principle, the principle of entropy increase is applicable when a battery system and its surrounding environment are considered as a new isolated system. The battery reaction rate increases with increased temperature at 0.1, 1, and 2◦C. However, the entropy value of Li3VO4-coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material is less than that of pure LiNi0.8Co0.1Mn0.1O2, which indicates that the Li3VO4-coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material is more orderly in the charge–discharge reaction process and the side reaction between the active material and electrolysis is less with higher structure stability.

### CONCLUSIONS

The energy storage and thermostability properties of the LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode materials before and after Li3VO<sup>4</sup> coating at different temperatures and rate were investigated

### REFERENCES


by electrochemical calorimetry for the first time. The results showed that the specific discharge capacities of Li3VO4-coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material at 1◦C are 194.2 (30◦C), 175.5 (40◦C), and 168.1 (50◦C) mAh g−<sup>1</sup> and the capacity retention rates are 96.7, 95.9, and 91.4%, with excellent rate and cycle performance after 50 charge/discharge cycles. Thermoelectrochemical experiments indicated that the current rate increase, heat quantity, enthalpy and entropy increase, and the specific discharge capacity decreases in the charge and discharge processes of lithium-ion battery cathode materials at the same temperature. The temperature increase, heat quantity, enthalpy and entropy increase and the specific discharge capacity increases first and rapidly decays at the same current rate. In addition, the heat production, enthalpy change, and entropy change of the Li3VO4-coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode material in the charge–discharge reaction system are lower than that of pure LiNi0.8Co0.1Mn0.1O<sup>2</sup> at 30, 40, and 50◦C. This finding shows that Li3VO<sup>4</sup> coating remarkably improves the thermal stability of the material. This condition is because the Li3VO<sup>4</sup> coating layer as a fast ion conductor promotes lithium ion transport, protects the active material from electrolyte erosion, reduces the side effects and electrode polarization and improves the thermal stability of the Li3VO4-coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode materials, which have excellent electrochemical performance and thermal stability. In conclusion, Li3VO4-coated LiNi0.8Co0.1Mn0.1O<sup>2</sup> cathode materials is proven to be an effective method in improving the energy storage and thermostability properties of cathode materials. This experiment will provide a new direction and basis for the study on high-temperature and thermoelectrochemical properties of lithium ion batteries.

### AUTHOR CONTRIBUTIONS

LS and FT conceived the idea. FT and LS prepared all materials. LS, FT, and ZX conducted electrochemical-calorimetry experiments. LS, FT, and ZX analyzed the data. FT wrote the manuscript. LS, ZC, and HZ commented on it. LS, ZX, ZC, and HZ supervised the implementation of the project.

### FUNDING

This work was financially supported by the National Natural Science Foundation of China (Nos. 21501015, 51604042, 31527803, and 21545010), the Hunan Provincial Natural Science Foundation of China (2018JJ2428).


capacity and long cycle life. J. Mater. Chem. A 6, 592–598. doi: 10.1039/C7TA0 8346G


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Song, Tang, Xiao, Cao and Zhu. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Li2O-Reinforced Solid Electrolyte Interphase on Three-Dimensional Sponges for Dendrite-Free Lithium Deposition

Chao Shen<sup>1</sup> \*, Huibo Yan<sup>1</sup> , Jinlei Gu<sup>1</sup> , Yuliang Gao<sup>1</sup> , Jingjing Yang<sup>2</sup> \* and Keyu Xie<sup>1</sup>

*<sup>1</sup> State Key Laboratory of Solidification Processing, Center for Nano Energy Materials, School of Materials Science and Engineering, Northwestern Polytechnical University and Shaanxi Joint Laboratory of Graphene (NPU), Xi'an, China, <sup>2</sup> School of Materials and Chemical Engineering, Xi'an Technological University, Xi'an, China*

Lithium (Li) metal, with ultra-high theoretical capacity and low electrochemical potential, is the ultimate anode for next-generation Li metal batteries. However, the undesirable Li dendrite growth usually results in severe safety hazards and low Coulombic efficiency. In this work, we design a three-dimensional CuO@Cu submicron wire sponge current collector with high mechanical strength SEI layer dominated by Li2O during electrochemical reaction process. The 3D CuO@Cu current collector realizes an enhanced CE of above 91% for an ultrahigh current of 10 mA cm−<sup>2</sup> after 100 cycles, and yields decent cycle stability at 5 C for the full cell. The exceptional performances of CuO@Cu submicron wire sponge current collector hold promise for further development of the next-generation metal-based batteries.

Keywords: CuO@Cu, submicron wire sponge, current collector, lithium anode, dendrite-free

## INTRODUCTION

Research has been focused on practical applications of Li metal anode since 1970s, due to its theoretical capacity (3,860 mAh g−<sup>1</sup> ) and low electrochemical potential (−3.040 V vs. SHE; Li et al., 2014; Mukherjee et al., 2014; Cheng et al., 2017; Shen et al., 2018). Unfortunately, the commercialization of Li metal anode has been retarded for several decades by the problem of Li dendrites which cause poor Coulombic efficiency and mass capacity loss (Amine et al., 2014). More seriously, the sharp Li dendrites will pierce through the separator, generating internal shortcircuit, bringing about severe safety hazards (Bouchet, 2014; Lu et al., 2014). It is well accepted that the growth of Li dendrites is mainly attributable to two reasons (Bouchet, 2014; Xu et al., 2014; Aryanfar et al., 2015; Cheng et al., 2017). (1) The solid electrolyte interphase (SEI) layer forms on the Li metal anode with insufficient mechanical strength. Li reacts instantly in contact with liquid electrolytes and rapidly forms an SEI film. This passivation layer of SEI prevents further loss of Li and electrolyte caused by their continued reaction (Lu et al., 2015; Shin et al., 2015). However, natural SEI layer does not have enough mechanical strength to withstand large volume change during Li charge/discharge processes (Liang et al., 2015). (2) Inhomogeneous distribution of Li<sup>+</sup> on anode. Li<sup>+</sup> accumulates and deposits on the "hot spots" because of the roughness of current collector, eventually forms numerous Li dendrites (Lin et al., 2016; Shen et al., 2018).

Scientists proceed from above two reasons to address the Li dendrites problems, and have achieved solid progress on suppressing dendrite formation and growth (Camacho-Forero et al., 2015; Lu et al., 2015; Shin et al., 2015; Zu et al., 2016; Jin et al., 2017; Liu et al., 2017). Cui summarized traits indispensable for an ideal SEI layer (Liu et al., 2017a). (1) Homogeneity in all

#### Edited by:

*Qiaobao Zhang, Xiamen University, China*

#### Reviewed by:

*Wei Tang, Xi'an Jiaotong University, China Yunjian Liu, Jiangsu University, China Chunyi ZHI, City University of Hong Kong, Hong Kong*

#### \*Correspondence:

*Chao Shen shenchao@nwpu.edu.cn Jingjing Yang yangjingjing@xatu.edu.cn*

#### Specialty section:

*This article was submitted to Physical Chemistry and Chemical Physics, a section of the journal Frontiers in Chemistry*

Received: *31 August 2018* Accepted: *09 October 2018* Published: *06 November 2018*

#### Citation:

*Shen C, Yan H, Gu J, Gao Y, Yang J and Xie K (2018) Li*2*O-Reinforced Solid Electrolyte Interphase on Three-Dimensional Sponges for Dendrite-Free Lithium Deposition. Front. Chem. 6:517. doi: 10.3389/fchem.2018.00517*

**140**

aspects. (2) High modulus and compact structure. (3) Flexibility to accommodate the ineluctable interface fluctuation during battery cycling. (4) High ionic conductivity. Natural SEI can hardly meet all the requisites above, hence necessitates the rational design of SEI. Various electrolyte additives and artificial SEI films have been employed to reinforce the SEI layer and suppress the formation of Li dendrites. Additives currently proven effectual mainly include vinylene carbonate (Ota et al., 2004; Stark et al., 2011), fluoroethylene carbonate (Liu et al., 2015), LiNO<sup>3</sup> and lithium polysulfide (Li et al., 2015), lithium fluoride (Choudhury and Archer, 2016), ionic liquid (Schweikert et al., 2013), metal ions (Ding et al., 2013), as well as trace amount of water and gases (Christensen et al., 2012; Qian et al., 2015). The concept of artificial protective film has been deeply explored in previous studies, and various artificial films have been applied on Li foil surfaces, such as lithium polyacetylene (Sakamoto et al., 2001), tetraethoxysilane (Umeda et al., 2011), lithium phosphorus oxynitride (Dudney, 2000), Cu3N nanoparticles compounded styrene butadiene rubber (Liu et al., 2017a), a hollow carbon nanospheres layer (Zheng et al., 2014), a boron nitride layer (Luo et al., 2015), a modified poly (dimethylsiloxane) film (Zhu et al., 2017), and a Li3PO<sup>4</sup> layer (Li et al., 2016). However, stable cycling cannot be guaranteed due to the consumption of additives in long-term cycle, and the artificial protective film can increase the impedance and reduce specific energy density.

Therefore, it is necessary to obtain a SEI layer with high mechanical strength to suppress Li dendrites effectively. In general, the SEI films dominated by inorganic crystalline components such as Li2CO<sup>3</sup> (Fujieda et al., 1994; Shang et al., 2012), Li2O (Billone et al., 1986; Zhang et al., 2016), and LiF (Combes et al., 1951; Takehara, 1997) are strong in mechanical strength while those dominated by organic components such as lithium alkyl carbonates (ROCO2Li) are found to be porous and fragile with low shear modulus under 1 GPa (Stone et al., 2012; Karkera and Prakash, 2018). Theoretical predictions have shown that a solid film with elasticity modulus of 1 GPa should be sufficient in suppressing dendrites (Monroe and Newman, 2005; Stone et al., 2012). M.C. Billone etc. reported that Li2O has a high elasticity modulus of 108 GPa (Billone et al., 1986), much higher than the threshold. Furthermore, Zen-ichiro Takehara proved that the Li2O containing SEI layer is adjacent to the Li metal anode (Park et al., 2017).

However, a dentrite-free morphology requires not only a reinforcement SEI but also homogenized electric field which is crucial for the uniform deposition of Li. Other approaches exploring 3D conductive carbon-based and metal-based current collectors to achieve uniform Li<sup>+</sup> deposition and adapt to volumetric change during Li plating/stripping. Scientists have gained ground on the conversion of Cu foil into a 3D host current collector such as 3D porous Cu (Yun et al., 2016), Cu nanoclusters structure (Zhang et al., 2016) and aligned CuO nanosheets on a planar Cu foil (Zhang et al., 2018). The carbon-based 3D current collectors include nitrogen-doped graphene (Zhang et al., 2017), nanoparticles anchored on carbon nanofibers (Yang et al., 2017), and hollow carbon spheres (Shen et al., 2018). Recently, Wei et al. found that the tortuous pores of the porous media can drastically reduce the local flux of Li<sup>+</sup> moving toward the anode and effectively extend the physical path of dendrite growth (Li et al., 2018). These studies reveal that 3D current collectors and inter-layer can homogenize the Li metal deposition, therefore suppress the formation of Li dendrites.

Herein, we design a three-dimensional (3D) porous CuO@Cu submicron wire sponge to inhibit the formation of Li dendrites. The 3D CuO@Cu submicron wire sponges own unique porous microstructure, which can homogenize the distribution of charges and inhibit the dendrite growth. Furthermore, Li2O gradually forms adjacent to the surface of Cu during the electrochemical reaction process of 2Li+CuO→ Li2O+Cu. As a result, the 3D CuO@Cu collector within a SEI film dominated by Li2O is not only helpful for the enhancement of Li<sup>+</sup> diffusion kinetics, but also beneficial for suppressing the Li dendrite due to the high shear modulus and rather strong mechanical strength.

### MATERIALS AND METHODS

### Synthesis of the Porous 3D CuO@Cu Submicron Wire Sponges

The precursor materials were composed of sodium hydroxide (NaOH, 40 mL 15 M), copper sulfate (CuSO4, 200 µL 1 M), ethylenediamine (EDA, 300 µL 99 wt%), and hydrazine (50 µL 35 wt%), and the precursor suspension was initially dispersed by ultrasonication and added to a plastic vessel. The sealed plastic vessel was heated at 70◦C in a water bath for 12 h to form a continuous Cu submicron wire hydrogel. After that, the as-synthesized Cu submicron wire hydrogel was washed with hydrazine solution (5wt%) several times to remove NaOH. Then, the as-synthesized Cu submicron wire hydrogel was treated for 5 min in constant temperature and humidity test chamber (60◦C, 80%). Finally, the CuO@Cu hydrogel was frozen and dried into sponges to retain the original gel volume.

### Characterizations

Field emission scanning electron microscopy (FESEM) measurements were carried out with Nova NanoSEM 450 equipped with an EDX spectroscopy attachment. X-ray diffraction (XRD) was recorded from 20 to 85◦ on a Bruker D8 advance diffractometer with CuKα radiation (λ = 1.5406 Å). The contact angle was measured by an Optical Contact Angle & interface tension meter (SL200KB, Kino, USA) at room temperature in air, and a 3.0 µL droplet of the ether-based electrolyte was used in the experiment.

### Electrochemical Measurements

The 3D porous CuO@Cu submicron wire sponge was first pressed and punched out into circular disks with a diameter of 12 mm as 3D porous current collectors for Li metal anodes. For repeated Li deposition/stripping testing, CR2032 coin cells were assembled using a 3D porous CuO@Cu submicron wire sponge or a planar Cu foil as the working electrode, a Li foil as the counter electrode, and a Celgard microporous polypropylene film as the separator. The Li deposition capacity is fixed at 1.0 mAh cm−<sup>2</sup> and the cut-off potential for the stripping process is configured to be 1.0 V. The electrolyte was 1 M lithium bis(trifluoromethane sulfonyl)imide (LiTFSI) in cosolvent of 1,3-dioxolane (DOL) and 1,2-dimethoxyethane (DME; 1:1 in volume) with 2% LiNO3. For the symmetrical cell test, 1 mAh cm−<sup>2</sup> of Li was first plated onto the current collectors at a current density of 2 mA cm−<sup>2</sup> , then the cells were cycled at a current density of 0.5 mA cm−<sup>2</sup> for 0.5 h in each half cycle. For the LiFePO<sup>4</sup> full cells, the LiFePO<sup>4</sup> electrodes were prepared by mixing LiFePO4, polyvinylidene fluoride, and carbon black in the ratio of 8:1:1 with N-methyl-2-pyrrolidone as the solvent. The areal mass loading of the LiFePO<sup>4</sup> electrodes was about 4.2 mg cm−<sup>2</sup> . The electrolyte is consisted of 1.0 M LiPF<sup>6</sup> in ethylene carbonate (EC)/dimethyl carbonate (DMC) (1:1 in volume). The 3D porous CuO@Cu submicron wire sponges or planar Cu foil as a current collector was first assembled into a half cell using a Li foil as the counter electrode. After depositing 5 mAh cm−<sup>2</sup> of Li metal onto the current collector, the cell was disassembled in an Ar-filled glove box, then the deposited Li current collector as anode was further reassembled into a full cell against LiFePO<sup>4</sup> cathode. The LiFePO<sup>4</sup> full cells were galvanostatically cycled between 2.4 and 4.3 V at 1 C. All the cells were tested using a CT2001A cell test instrument (LAND Electronic Co, BT2013A, China) or an 88-channel battery tester (Arbin Instruments, BT2000, USA). The cyclic voltammetry (CV) curves were measured with Solartron. For the Li||3D CuO@Cu CV curves, the voltage sweep rate was 0.1 mV s−<sup>1</sup> between 0.01 and 3 V vs. Li/Li+. For full-cell CV curves, the voltage sweep rate was 0.1 mV s−<sup>1</sup> between 2.4 and 4.3 V vs. Li/Li+.

### RESULTS AND DISCUSSION

The morphologies of the 3D porous Cu submicron wire sponge are shown in **Figures 1a,b**. It can be seen that as-synthesized Cu sponge consists of plenty of long intertwined submicron wires (**Figure 1a**). These wires, with length more than 100µm, have an average diameter of about 450 nm (**Figure 1b**) and the surface is very smooth. As shown in **Figure 1f**, the XRD was then employed to study the phase structure of the Cu submicron wires sponge. The patterns for Cu wires sponge is consistent with the JCPDS date (PDF#04-0836), indicating that Cu sponges have a face-centered cubic structure (Gao et al., 2001; Yu et al., 2005). In addition, there are three diffraction peaks at around 43.2◦ , 50.5◦ , and 74.1◦ correspond to the (111), (200), and (220) planes of the copper. Different from surface topography of 3D Cu wires, the high-magnification FESEM of

the as-synthesized CuO@Cu submicron wire sponge shows that the fiber surface is rough (**Figure 1c**), and the corresponding elemental mapping further confirms the distribution of Cu and O (**Figures 1d,e**). As shown in **Figure 1i**, the XRD of CuO@Cu has two more obvious different diffraction peaks at around 35.5◦ and 38.9◦ compare with Cu submicron wire sponge, which are highly consistent with the (110) and (002) lattice plane attribute to CuO (PDF#41-0254), confirming the presence of CuO on the surface of Cu submicron wire (Liu et al., 2006; Yu et al., 2013; Zhang et al., 2016). As shown in **Figure 1g**, the 3D porous CuO@Cu submicron wire sponge exhibits decent mechanical and processing properties, it can be easily folded and blended without fracture. To evaluate the wettability between the electrolyte and 3D porous CuO@Cu submicron wire sponge, the contact angles of LiTFSI-based electrolyte on the planar Cu foil and 3D porous CuO@Cu submicron wire sponge were measured (**Figure 1h**). The contact angle of electrolyte droplet on the planar Cu foil is 38◦ (**Figure 1h**, left), while it is nearly 0◦ on the 3D porous CuO@Cu submicron wire sponge (**Figure 1h**, right), indicating a better wettability between the CuO@Cu submicron wire sponge and the electrolyte.

The electrochemical performances of the cells, with planar Cu foil and 3D porous CuO@Cu submicron wire sponge as current collectors, confirm that 3D porous CuO@Cu submicron wire sponge can effectively inhibit Li dendrites and exhibits better electrochemical performance. All the cells were first cycled from 0 to 1 V at 50 µA to remove surface contamination and stabilize the SEI film (Xu et al., 2014; Li et al., 2016). Coulombic efficiencies are shown in **Figure 2A**. At current density of 2, 5, and 10 mA cm−<sup>2</sup> , the Coulombic efficiencies of the planar Cu foil remain in a relative stable level (80–90%) within 30 cycles, and gradually decreased or fluctuated in the subsequent cycles, as a result of the SEI films are sabotaged by Li dendrites (Gao et al., 2001; Yu et al., 2005). In contrast, the Coulombic efficiencies of the 3D porous CuO@Cu submicron wire sponge current collector remain as high as 98% at current density of 2 mA cm−<sup>2</sup> , and 96% at current density of 5 mA cm−<sup>2</sup> after 100 cycles. Even at an ultrahigh current density of 10 mA cm−<sup>2</sup> , the Coulombic efficiency still remains at 91% after 100 cycles. Cycling stabilities have been further investigated by symmetric cell test, which is a common technique to evaluate the characteristics of the interface in electrochemical devices (Zhu et al., 2017). The voltage profiles of metallic Li deposition/stripping in symmetric cells with planar Cu foil or 3D porous CuO@Cu submicron wire sponge are shown in **Figure 2B**, and the 3D CuO@Cu submicron wire sponge shows much more stable cycling than its planar Cu foil counterpart with severe fluctuations due to the polarization caused by the repeatedly breaking and repairing the SEI film (Monroe and Newman, 2005). In other words, a stable SEI film, as well as long-term cycling stability, can be realized in the symmetric cell test with 3D porous CuO@Cu submicron wire sponge. The prominent electrochemical properties of 3D CuO@Cu submicron wire sponge can be further confirmed by the electrochemical impedance spectroscopy (EIS) analysis conducted on after initialization process (**Figure 2C**) and the 100th cycles at current density of 5 mA cm−<sup>2</sup> (**Figure 2D**). The diameter of semicircle at high frequency range is an indicator of the SEI film resistance. The SEI film resistance of the 3D Cu submicron wire sponge current collector is always lower than that of the planar current collector, indicating that the porous structure of the 3D current collector is beneficial for the kinetics of electrochemical reactions of electrodes (Monroe and Newman, 2005). It is worth noting that, after initialization process and 100th cycles of Li deposition/dissolution, the SEI film resistance reduces. The drop in resistance is associated with residual lithium on the current collector that increase the interfacial area between the electrolyte and lithium metal, which results in the reduction of resistance of Li/electrolyte interface.

As shown in **Figure 3**, after 100 cycles, the morphologies of Li deposition on planar Cu foil current collector and 3D porous CuO@Cu submicron wire sponge current collector are totally different. There are a great number of fiber-like Li dendrites with length of more than 10µm and width of 3µm on the surface of planar Cu foil current collector (**Figures 3a,b**). These Li dendrites could short-circuit the cell and cause safety hazard. It can be observed from the cross-sectional FESEM images that the original compact Li metal has become porous after cycles (**Figures 3b,f**), some of them may become electrically isolated and eventually form so-called "dead" Li, resulting in low Coulombic efficiency and rapid capacity loss (Zhang et al., 2017). However, compared to planar Cu foil covered with lots of mossy and dendritic Li after repeated deposition and stripping cycles, Li deposition is compact on the surface of CuO@Cu submicron wires (**Figures 3d,h**). According to the **Figure 3e** and **Figure S1A**, the Li adhered on the current collector is 200µm thick after 100 cycles. In other words, the thickness of the Li-planar Cu foil anode is increased by 200µm. However, the thickness of 3D porous CuO@Cu current collector hardly changes after 100 cycles (**Figure 3g** and **Figure S1B**). Thus, the 3D porous structure CuO@Cu current collector can adapt to volume changes during charging and discharging process. The submicron wire structure could homogenize the electric field distribution, as a result, uniform Li deposition is expected to cover the submicron wire surface. Meanwhile, Li2O enhances the mechanical strength of the SEI layer and is helpful for the enhancement of Li<sup>+</sup> diffusion kinetics. During the charge/discharge process, Li<sup>+</sup> gains electrons and eventually deposits between the SEI layer and the Cu surface. As a result, formation of Li dendrites is inhibited effectively.

Reduction sweep CV curves in **Figure 4a** corresponding to the conversion of CuO + 2Li<sup>+</sup> + 2e−→ Cu + Li2O, which is identified by the redox peaks in CV (Gao et al., 2004; Zhang et al., 2016). During the first reduciton process, all Cu2<sup>+</sup> is reduced to Cu<sup>+</sup> at 1 V and all Cu<sup>+</sup> is reduced to CuO at 0.1 V. In order to confirm whether or not CuO was transformed, we tested the XRD of the CuO@Cu submicron wire sponge before and after reduction. **Figure 4b** presents the XRD of the electrode before and after the electrochemical reduction of the CuO. The disappearance of the CuO phase justifies the reduction of CuO. The weak signal of Li2O after the reduction should be attributed to the formation of Li2O during reduciton process. FESEM images of CuO@Cu submicron wire sponge in **Figures 4c,d**

cm−<sup>2</sup> , respectively. The deposition capacity of Li is fixed at 1 mAh cm−<sup>2</sup> . (B) Voltage profiles of Li metal deposition/stripping at current density of 0.5 mA cm−<sup>2</sup> in symmetric cells with the planar Cu foils or the 3D CuO@Cu sponge as the current collectors. (C,D) The EIS curves of after initialization process and after 100 cycles at current density of 5 mA cm−<sup>2</sup> .

indicate that CuO on the rough CuO@Cu surface is reduced to Cu and Li2O, gradually forming a smooth layer of SEI@Cu at 1 V. When the potential drops to 0.1 V, a thick Li2O-reinforced SEI can be clearly seen due to a serious amount of Li2O deposition. Wang reported that in-situ filled with Li2O on SEI layer formed by the CuO + 2Li<sup>+</sup> + 2e−→ Cu + Li2O chemical reaction has a rather strong mechanical strength, which dendrite Li can hardly penetrate (Zhang et al., 2016). Guruprakash Karkera reported an in situ formed shielding layer composed of Li2O by 2Li+ + 2e<sup>−</sup> + ½ O<sup>2</sup> → Li2O which keeps the Li anode intact during vigorous cell conditions, providing faster Li-ion diffusion kinetics and stable cycling performance (Karkera and Prakash, 2018). Therefore, we propose that Li<sup>+</sup> passes through the Li2O layer and adhere to the Cu surface, forming a Li layer (**Figure 3i**).

To demonstrate the potential practical application of the 3D porous CuO@Cu submicron wire sponge current collector, full cells were built with the LiFePO<sup>4</sup> as cathodes and Li-3D CuO@Cu sponge as anodes. The CV curves of Li-planar Cu foil || LiFePO<sup>4</sup> and Li-3D CuO@Cu sponge || LiFePO<sup>4</sup> are shown in **Figure S2**. It can be seen that there is only one peak pair, consisting of one anodic peak and one cathodic peak, which corresponds to the two-phase charge/discharge reaction of the Fe3+/Fe2<sup>+</sup> (Yuan et al., 2011). Except for the first cycle, the CV curves of the cyles are almost coincident, indicating the preferable stability of the Li-3D CuO@Cu sponge and Li-planar Cu foil electrodes. For Li-3D

CuO@Cu sponge anode, the voltage separation gets smaller than Li-planar Cu foil anode, indicating 3D CuO@Cu sponge anode is helpful for the enhancement Li<sup>+</sup> diffusion kinetics and a low voltage hysteresis (Liu et al., 2017b; Su et al., 2018). As displayed in **Figure 5a**, the rate capacity at low current density is identical between the Li-3D CuO@Cu sponge anode and Li-planar Cu foil anode. However, at a high current density of 5 C, rate capacity of full cell with the Li-3D CuO@Cu sponge anode is as high as 80.3 mAh g−<sup>1</sup> , while that of the Cu foil holds only 54.8 mAh g −1 (**Figure 5a**). A remarkable 47% increase in specific capacity indicates the superiority of Li-3D CuO@Cu sponge anode. As shown in **Figure 5b**, the Li-3D CuO@Cu sponge anode realizes more stable cycling performance for full cells at 5C. After 120 cycles, the reversible capacity of the 3D current collector remains 87.8 mAh g−<sup>1</sup> , while the Li-planar Cu foil anode exhibits a sudden capacity attenuation after 95 cycles and the capacity drops to less than 40 mAh g−<sup>1</sup> . The sudden decay of Li-planar Cu foil's capacity is due to formation of dendritic Li. While Li dendrite in Li-3D CuO@Cu sponge anode is effectively inhibited, so the capacity remains high and stable. The galvanostatic charge and discharge profiles of the full cells at 5C are plotted in **Figure S3** for the 5th, 50th and 100th. The 3D CuO@Cu sponge anode realizes not high but stable discharge and charge capacities (Zheng et al., 2019). At the same time, 3D CuO@Cu sponge anode has a lower voltage hysteresis in the charge/discharge process indicating the significant kinetic advantage of the Li2Oreinforced SEI (Song et al., 2018). The FESEM images of the Li-planar Cu foil anode and Li-3D CuO@Cu sponge before and after cycles are shown in **Figures 5c–f**. The original smooth surface of the Cu foil (**Figure 5c**) covered with mossy Li after 120 cycles (**Figure 5d**). For Li-3D CuO@Cu sponge anode, there is no difference between the morphologies before and after cycling (**Figures 5e,f**), indicates that the Li-3D CuO@Cu sponge anode has an efficacious SEI layer with sufficient mechanical strength that guarantees a homogeneous Li depositing process, therefore renders a superior cycle stability.

FIGURE 4 | (a) The CV curves of Li||3D porous CuO@Cu sponge. (b) XRD of the CuO@Cu electrode before and after the electrochemical reduction. (c,d) FESEM images of the 3D porous CuO@Cu submicron wire sponge during the first reduction sweep at 1V and 0.1V, respectively.

sponge). (c,d) SEM images of the Li-planar Cu foil anode before and after cycling. (e,f) SEM images of the Li-3D CuO@Cu sponge anode before and after cycling.

In summary, we introduced a simple but effective strategy to suppress Li dendrite growth by using CuO@Cu submicron wire sponge as current collector. The 3D porous structure of the CuO@Cu submicron wire sponges, with SEI layer dominated by Li2O with strong mechanical strength, is conducive to homogenizing electric field distribution, therefore renders dendrite-free Li deposition. The Coulombic efficiency of the 3D porous CuO@Cu submicron wire sponge current collector remains 98% at current density of 2 mA cm**–**<sup>2</sup> , and 96% at current density of 5 mA cm**–**<sup>2</sup> after 100 cycles. Even at an ultrahigh current density of 10 mA cm**–**<sup>2</sup> , the Coulombic efficiency still remains at 91% after 100 cycles. At a high current density of 5C, rate capacity of full cell with the Li-3D CuO@Cu sponge anode is as high as 80.3 mAh g**–**<sup>1</sup> , and after 120 cycles, the reversible capacity remains 87.8 mAh g−<sup>1</sup> . Compared with the planar Cu foils, CuO@Cu submicron wire sponge current collector displays superior electrochemical cycling performance with higher and more stable CE and longer service life. We believe this work can offer valuable guidance as well as deep understanding in design of novel materials or structures to suppress Li dendrite growth for further development of next-generation Li metal batteries, such as Li-S or Li-air batteries.

### REFERENCES


### AUTHOR CONTRIBUTIONS

CS developed the concept and designed the experiments. JY conducted the experiments. HY and JG built the cells and carried out the performance characterizations. YG and KX co-supervised the research. YG revised the work critically for important intellectual content. CS and HY co-wrote the manuscript. All authors discussed the results and commented on the manuscript.

### ACKNOWLEDGMENTS

The authors acknowledge the financial support from the National Key R&D Program of China (2018YFB0104204), National Natural Science Foundation of China (51804259, 51402236, and 51674202), the Fundamental Research Funds for the Central Universities (3102018jgc004), and the Key R&D Program of Shaanxi (2017ZDCXL-GY-08-03).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2018.00517/full#supplementary-material


polymer electrolytes for rechargeable lithium metal batteries. J. Electrochem. Soc. 159, A222–A227. doi: 10.1149/2.030203jes


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Shen, Yan, Gu, Gao, Yang and Xie. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Tin Nanoparticles Encapsulated Carbon Nanoboxes as High-Performance Anode for Lithium-Ion Batteries

Ziming Yang1†, Hong-Hui Wu2†, Zhiming Zheng<sup>1</sup> , Yong Cheng<sup>1</sup> , Pei Li <sup>1</sup> , Qiaobao Zhang<sup>1</sup> \* and Ming-Sheng Wang<sup>1</sup> \*

*<sup>1</sup> Department of Materials Science and Engineering, College of Materials and Pen-Tung Sah Institute of Micro-Nano Science and Technology, Xiamen University, Xiamen, China, <sup>2</sup> Department of Chemistry, University of Nebraska-Lincoln, Lincoln, NE, United States*

### Edited by:

*Jiexi Wang, Central South University, China*

### Reviewed by:

*Renzong Hu, South China University of Technology, China Huiqiao Li, Huazhong University of Science and Technology, China Chuanxin He, Shenzhen University, China*

#### \*Correspondence:

*Qiaobao Zhang zhangqiaobao@xmu.edu.cn Ming-Sheng Wang mswang@xmu.edu.cn*

*†These authors have contributed equally to this work*

#### Specialty section:

*This article was submitted to Nanoscience, a section of the journal Frontiers in Chemistry*

Received: *21 September 2018* Accepted: *12 October 2018* Published: *31 October 2018*

#### Citation:

*Yang Z, Wu H-H, Zheng Z, Cheng Y, Li P, Zhang Q and Wang M-S (2018) Tin Nanoparticles Encapsulated Carbon Nanoboxes as High-Performance Anode for Lithium-Ion Batteries. Front. Chem. 6:533. doi: 10.3389/fchem.2018.00533* One of the crucial challenges for applying Sn as an anode of lithium-ion batteries (LIBs) is the dramatic volume change during lithiation/delithiation process, which causes a rapid capacity fading and then deteriorated battery performance. To address this issue, herein, we report the design and fabrication of Sn encapsulated carbon nanoboxes (denoted as Sn@C) with yolk@shell architectures. In this design, the carbon shell can facilitate the good transport kinetics whereas the hollow space between Sn and carbon shell can accommodate the volume variation during repeated charge/discharge process. Accordingly, this composite electrode exhibits a high reversible capacity of 675 mAh g−<sup>1</sup> at a current density of 0.8 A g−<sup>1</sup> after 500 cycles and preserves as high as 366 mAh g−<sup>1</sup> at a higher current density of 3 A g−<sup>1</sup> even after 930 cycles. The enhanced electrochemical performance can be ascribed to the crystal size reduction of Sn cores and the formation of polymeric gel-like layer outside the electrode surface after long-term cycles, resulting in improved capacity and enhanced rate performance.

Keywords: lithium-ion battery, anode material, yolk-shell structure, Sn@C nanoboxes, electrochemical performance

### INTRODUCTION

Over the past few decades, rechargeable secondary lithium-ion batteries (LIBs) have become one of the most popular energy sources for electric vehicles, various portable devices, and grid-scale storage systems because of their high capacity, terrific safety, and steady cycling performance (Bruce et al., 2012; Mahmood et al., 2016; Li et al., 2017b; Liu et al., 2018). However, with the increasing demand of longer usage time and higher capacity about the batteries, graphite, as the widely used commercial anode materials, gradually cannot meet the market needs for its relatively low theoretical capacity of 372 mAh g−<sup>1</sup> (Hu et al., 2012a; Lim et al., 2012; Leng et al., 2016; Zhang et al., 2018b). Therefore, extensive researches of alloy-based materials (Si, Ge, Sn, etc.) have been studied in LIBs for the higher capacity of these anode materials (Obrovac and Chevrier, 2014; Li et al., 2017a; Wang et al., 2017, 2018; Zhang et al., 2018a). Metallic Sn was reported with high theoretical capacity (997 mAh g−<sup>1</sup> ), good electroconductibility and non-toxicity (Hassoun et al., 2007; Hu et al., 2008, 2011, 2012b). However, the giant volume change of 260% during lithiation/delithiation process would result in the pulverization of the active materials after long cycles, and thus the fast capacity fading (Rhodes et al., 2012).

To solve these problems, reducing the size of Sn particles seems to be a good solution because the nano-particles could endure higher stress and effectively prevent the active materials from pulverization (Leng et al., 2015). Nevertheless, nanoparticles prefer to aggregate during the charge and discharge process. A promising strategy to mitigate this drawback is the smart hybridization of Sn with carbon-based materials (graphene, carbon nanotube, etc.) to Sn@carbon composites (Zhang et al., 2014; Huang et al., 2015; Cheng et al., 2016; Qin et al., 2016), wherein the carbon materials could significantly prevent aggregation, buffer the stress concentration resulted from volume expansion and enhance the electroconductibility. Among them, the yolk@shell structure of Sn@C composites is a preferable choice because the hollow space between the core and carbon shell could accommodate more volume variation free from breaking the carbon shell, which could improve the performance of active materials compared with the Sn@C core@shell composites (Zhang et al., 2014, 2017; Qin et al., 2016).

Inspired by the previous work, herein, we synthesize Sn@C nanoboxes with yolk-shell nanostructure comprising a spherelike Sn core within a hollow cavity surrounded by carbon nanobox via a one-pot spray pyrolysis process followed by hydrogen-thermal reduction. By virtue of structural advantages, the as-prepared electrode exhibits an outstanding reversible capacity of 675 mAh g−<sup>1</sup> at a current density of 0.8 A g−<sup>1</sup> after 500 cycles. The crystal size reduction of Sn cores and the formation of polymeric gel-like layer outside the electrode surface during cycling could explain the increase of the reversible capacity during long cycles. The current work shed light on the improvement of anode materials for next-generation highperformance LIBs.

### EXPERIMENTAL

## Synthesis of ZnSnO<sup>3</sup> Nanocubes

The ZnSnO<sup>3</sup> nanocubes are fabricated via a hydrothermal precipitation method according to the literature (Zhang et al., 2017). In a typical process, 2.875 g (10 mmol) of zinc sulfate heptahydrate (ZnSnO4·7H2O) is added into 100 mL deionized water and stirred till completely dissolved, and then 20 mL of sodium stannate solution (NaSnO3·3H2O, 2.667 g, 10 mmol) is added into above zinc sulfate heptahydrate solution. The solution turns into milky immediately and the mixed solution is stirred at 70◦C for 4 h. After the reaction, the precipitates are collected by centrifugation and washed with deionized water and alcohol 3 times and dried at 60◦C for one night.

### Synthesis of Sn@C Nanoboxes

According to the literature (Zhang et al., 2017), 0.5 g ZnSnO<sup>3</sup> nanoparticles, 0.46 cetyltrimethylammonium bromide, and 14.08 mL H2O are added into a beaker before 0.5 h ultrasonic treatment and 1 h stirring, then 0.7 g resorcinol, 56.0 mL absolute ethanol and 0.2 mL NH3·H2O are added successively and stirred for 0.5 h at 35◦C. Finally, 0.1 mL formaldehyde is added by dropwise. After 6 h stirring and polymerization by aging over one night, the obtained ZnSnO3@resorcinol formaldehyde (RF) nanoboxes are collected by centrifugation and washed with deionized water and alcohol 3 times respectively. The yolk-shell Sn@C powder is obtained by heated the ZnSnO3@RF at 600◦C for 5 h with a rate of 2◦C min−<sup>1</sup> under H<sup>2</sup> (5%)/Ar (95%) atmosphere.

## Synthesis of Sn NPs

1.0 g ZnSnO<sup>3</sup> nanoparticles are heated at 600◦C for 5 h with a rate of 2◦C min−<sup>1</sup> under H<sup>2</sup> (5%)/Ar (95%) atmosphere.

### Characterizations

The crystal structure is recorded by a powder X-raying diffraction (Ri gaku Ultima IV). The structure and morphology of the samples are investigated via an FEI Talos-F200s transmission electron microscope (TEM) and a Zeiss SUPRA 55 scanning electron microscope (SEM). Thermogravimetric analysis is investigated by an SDT Q600 Simultaneous TGA/DSC instrument with a heating rate of 10◦C min−<sup>1</sup> in the air.

### Electrochemical Measurements

The electrochemical measurements are carried out by employing CR2025 coin cells and the working electrode is synthesized via mixing the active materials, conductive acetylene black and sodium carboxymethylcellulose with a weight ratio of 8:1:1. The slurry is coated on a Cu foil and dried at 80◦C inside a vacuum oven for 12 h. The Cu foil is first cut into circular disks with a diameter of 14 mm, and the masses of the asobtained Cu foil circular disk (m1) and the slurry coated on the Cu foil circular disk after dried (m2) are determined using a microbalance (Mettler Toledo XS3DU) with an accuracy of 1 µg. The active mass of mass loading of the electrode is then calculated as 0.8<sup>∗</sup> (m<sup>2</sup> – m1) and about 0.8 mg cm−<sup>2</sup> . The measured specific capacities of the electrodes are based on the total active mass loading of Sn@C. Li foil is taken as a counter electrode and 1 M LiPF<sup>6</sup> is mixed with ethyl carbonate, dimethyl carbonate, and diethyl carbonate (EC/DMC/DC = 1:1:1, volume ratio) are used as the electrolyte. The cells are assembled in the Ar-glovebox under Ar atmosphere with both oxygen and moisture below 0.1 ppm. Galvanostatic charge-discharge cycles are tested in the CT2001A LAND battery tester with potential windows of 0.01– 3.00 V. And cycle voltammetry measurements are conducted via a CHI660E electrochemical workstation under a scanning rate of 0.1 mV s−<sup>1</sup> . Electrochemical impedance spectroscopy (EIS) is measured in the frequency range from 0.01 to 100 kHz at open circuit potential with an amplitude of 5 mV.

### RESULTS AND DISCUSSION

As illustrated in **Figure 1**, the resorcinol-formaldehyde (RF) is coated outside of the cubic ZnSnO<sup>3</sup> nanoboxes, and then the ZnSnO3@RF is heated at 600◦C under the H<sup>2</sup> (5%)/Ar atmosphere for 300 min (Zhang et al., 2017). During this process, resorcinol formaldehyde is gradually turned into an amorphous carbon shell while SnO<sup>2</sup> and ZnO are reduced to metallic Sn and Zn. Due to the low boiling point (907◦C), Zn gradually evaporates from the material, which finally generates the yolkshell Sn@C composite (Zhang et al., 2014). Compared with other reported methods (Zhang et al., 2008; Ni et al., 2013; Wang et al., 2013a, 2015), this one-pot spray pyrolysis method could product yolk-shell Sn@C composite more efficient and controllable.

The X-ray diffraction patterns of the ZnSnO<sup>3</sup> and Sn@C are shown in **Figure 2A**. Most peaks of the precursor could be allocated to the ZnSnO<sup>3</sup> phase (JCPDS NO: 11-0274). After hydrogen-thermal reduction, the final product shows sharp peaks which are indexed to the Sn Phase (JCPDS No: 04-0673). Obviously, there are no obvious diffraction peaks from Zn, SnO2, or ZnO, which suggests that most Zn and O had been removed from the precursor. The thermogravimetric curve of the material after hydrogen-thermal reduction (**Figure 2B**) shows that the loss of weight below 200◦C is ascribed to the evaporation the residual moisture. In addition, the distinct loss from 450 to 560◦C is attributed to the carbon oxidization whereas the increase from 200 to 450◦C is caused by the gradual oxidization of Sn to SnO2. The slight increase after 450◦C may be attributed the further oxidization of some inner Sn that is not oxidized completely. Assuming the final materials is SnO2, the carbon content of the Sn@C nanoboxes is about 21% from the following equation:

$$\text{Sn (wt\%)} = 100 \times \frac{\text{molecular weight of Sn}}{\text{molecular weight of SnO2}}$$

$$\times \frac{\text{final weight of SnO2}}{\text{initial weight of Sn} \oplus \text{C}}$$

Scanning electron microscope (SEM) and TEM are employed to observe the morphology and structure of the active materials. As presented in **Figures 3a,b**, the size of cubic ZnSnO<sup>3</sup> ranges from 180 to 200 nm. The element mapping and energy dispersive spectrometer of ZnSnO<sup>3</sup> are shown in **Figures S3, S4**, respectively. The cubic ZnSnO3@RF is fabricated successfully by coating the RF with a uniform thickness of ∼20 nm (**Figure 3c**). After hydrogen-thermal reduction treatment, the outside RF transforms into amorphous carbon while the inside ZnSnO<sup>3</sup> gradually decomposed, leading to a majority of Zn volatilize from the carbon shell for the relatively low boiling point (907◦C). However, Sn is conserved because of the relatively high boiling point (2,260◦C). The remaining Sn transforms into the spherical structure to form the yolk-shell Sn@C structure when it is cooled to room temperature. According to **Figures 3d,e,g**, the TEM and SEM images clearly show the yolk-shell structure where the material in core position is Sn nanoparticles (∼100 nm) and the outside shell is carbon (∼30 nm). The well-designed yolk-shell structure could completely contain the Sn nanoparticles in each carbon shell and prevent the aggregation of nanoparticles. The selected-area electron diffraction (SAED) patterns shown in the inset of **Figure 3d** demonstrate the highly crystalline of Sn cores and amorphous nature of carbon shell. A small amount of Sn outflows from the carbon shell, which may be caused by the ultrasmall holes in the carbon shell. As shown in **Figure 3f**, the highresolution TEM image displays a series of parallel fringes with a space around 0.295 nm, which can be indexed to the (200) plane of crystalline Sn (JCPDS no. 04-0673). Furthermore, the STEM

FIGURE 3 | (a) SEM, (b) TEM image of cubic ZnSnO3, (c) SEM of ZnSnO3@RF, (d) TEM, inset is its corresponding SAED patterns taken from the rectangular area, (e) SEM, and (f) high-resolution TEM image of Sn@C, (g) its corresponding STEM image, (h,i) corresponding elemental mapping of (h) Carbon and (i) Sn, respectively. image (**Figure 3g**) and element mapping image in **Figures 3h,i** shows that the Sn nanoparticles locate in the carbon shell without any Zn signal, which suggests that most Zn is removed from the shell during the thermal treatment. Besides, the void space in the yolk-shell structure is about 60%, which is designed to accommodate the volume variation during the Li<sup>+</sup> insertion/deinsertion reaction. As a comparison, the XRD and SEM images of Sn NPs are shown in **Figures S1, S2**.

Cyclic voltammograms of the Sn@C electrode (**Figure 4A**) are tested to understand the electrochemical reactions during the charge/discharge processes. As can be seen, during the first cathodic scan, the three small peaks at 0.25, 0.5, and 0.55 V can be attributed to the alloying reaction between lithium and tin, forming of LixSn alloys (Dai et al., 2016). During the first anodic scan, four oxidation peaks between 0.40 and 0.80 V are assigned to the delithiation process of LixSn alloys (Li et al., 2013; Liu J. et al., 2015). Besides, the broad peak at 1.25 V is caused by the Li<sup>+</sup> extraction from carbon. In addition, there is a peak at 2.85 V in the first anodic scan, which disappears in the following cycles, and may be due to solid electrolyte interface (SEI) layer decomposition induced by nanoscaled Sn particles (Dai et al., 2016). The difference between the first and following cycles are caused by the electrolyte decomposition and formation of SEI films (Luo et al., 2012; Xu et al., 2012; Liu Y. et al., 2015). **Figure 4B** exhibits the charge and discharge profiles of the yolk-shell Sn@C nanoboxes at a current density of 0.8 A g−<sup>1</sup> over a voltage range from 0.01 and 3.0 V. Compared with the first cycle of charge curve, the plateau at 0.48 V of the following cycles gradually disappears, besides, the charge and discharge curves become sloping, indicating a linear timedependent change of potential at a constant current. Combined with the CV curves, the slope curves could be caused by the diffusion-controlled reaction and the capacitive capacity. The disappearance of potential plateaus in the following cycles and the linearization of the charge-discharge curve indicate the extra pseudocapacitive contribution (Augustyn et al., 2013; Li et al., 2015; Xu et al., 2016). The charge and discharge profiles of the Sn NPs at 0.8 A g−<sup>1</sup> over a voltage between 0.01 and 3.0 V are given in **Figure S5**. To further explore the electrochemical properties of the active materials, long cycling test (**Figure 4C**) is carried out. The Sn@C yolk-shell materials deliver initial discharge and charge capacities of 821.5 and 500.3 mAh g−<sup>1</sup> at

0.8 A g−<sup>1</sup> in a voltage range from 0.01 to 3 V, corresponding to the first coulombic efficiency of 60.9% (**Figure 4C**). The largely irreversible capacity loss at the first cycle is related to the formation of SEIs consuming active Li, and can be compensated by prelithiation through either chemical or electrochemical methods or by using stabilized Li metal powder (Zhang et al., 2018a; Zheng et al., 2018a). The reversible capacity quickly decreases to 272.3 mAh g−<sup>1</sup> at 37th cycle, which could mainly be attributed to the structural degradation and reorganization (Sun et al., 2014). Interestingly, the coulombic efficiency gradually increases to 98.5% from the 2nd cycle to 37th cycle, indicating that a relatively stable SEI layer forms during the initial cycles and renders the active surface substantially inert to further electrolyte decomposition, despite the extreme volume changes experienced by the underlying material during discharge/charge (Zhang et al., 2018a). The capacity after 37th cycle gradually increases to 674.6 mAh g−<sup>1</sup> after 500 cycles, suggesting an excellent cycling performance. Such an improved performance is mainly due to the following features: the Sn@C yolk-shell structure could steadily encapsulate the Sn nanoparticles and the hollow space between the carbon shells and the Sn cores could effectively buffer expansion caused by the giant volume expansion. Such an increased capacity has also been observed in the other Sn/C composites, SnO2/C composites and other metal oxide composites (Wang et al., 2010, 2012, 2013b; Guo et al., 2013; Xu et al., 2013; Liu Y. et al., 2015). It is general for various metal (Cheng et al., 2016) or metal oxides (Sun et al., 2013; Sn, Wang et al., 2013b, Mn, Yu et al., 2006; Liu et al., 2014; Xiao and Cao, 2015; Lian et al., 2017, Co, Laruelle et al., 2002, Fe, Zheng et al., 2018b) electrodes to exhibit such a capacity rise. Recent studiesshow that the capacity rise may be caused by the reversible formation and decomposition of an organic polymeric/gel-like layer from the electrolyte decomposition. And the polymeric/gel layer outside the active materials could enhance the mechanical cohesion and provide extra Li<sup>+</sup> insertion sites at the interface of the active materials, which is called "pseudo-capacitance-type behavior" (Wang et al., 2012). Moreover, the gradually increased intensity of the peaks at about 0.6 V (**Figure 4B**) may be due to the reversible decomposition of the organic polymeric/gellike layer. It has been confirmed that the layer forms at low voltage and decomposed at high voltage. On the contrary, the Sn NPs shows an initial reversible discharge capacity of 923.3 mAh g <sup>−</sup><sup>1</sup> under the same conditions and then the capacity gradually decreases to 71.4 mAh g−<sup>1</sup> after 100 cycles. Besides, the cycling performance of Sn@C and Sn NPs at 0.2 A g−<sup>1</sup> are presented in **Figures S6, S7**. **Figure 4D** demonstrates the rate performance of the Sn@C composite. The reversible capacities at the current density of 0.1, 0.2, 0.5, 1, and 2 A g−<sup>1</sup> are 499.5, 465.8, 420.0, 389.8, and 320.3 mAh g−<sup>1</sup> , respectively, which are much higher than those of the Sn NPs electrode. Besides, when the current density is recovered to 0.1 A g−<sup>1</sup> , a capacity of 509.0 mAh g−<sup>1</sup> is obtained and continued to increase after high-rate cycles, indicating that the yolk-shell structure could keep the integrity of the Sn nanoparticles and enable them to work normally at different current density. **Figures 4E,F** displays the impedance spectra of the Sn@C and Sn NPs, respectively. The semicircle in the high frequency is caused by the electrolyte resistance and charge transfer resistance (Liu et al., 2011). Obviously, the semicircle of Sn@C is much bigger than that of Sn NPs at 1st cycle, suggesting that the amorphous carbon shell hinders the Li<sup>+</sup> transform at first. In the following cycles, the semicircle of Sn@C gradually decreases but that of the Sn NPs increases, and the semicircle of Sn@C is much smaller than that of Sn NPs at 500th cycle, which indicates that the yolk-shell structure could protect the active materials and decrease the impedance in long-term cycles.

To further understand the underlying mechanism of the reversible capacity increase for yolk-shell Sn@C composites, TEM and SEM are conducted to capture the Sn@C structure after long cycles at 0.8 A g−<sup>1</sup> . The structure and morphology of Sn@C electrodes after 40 and 500 cycles are observed. As shown in **Figures 5a,b**, after 40 cycles, most Sn cores swell, which is attributed to the volume change after Li<sup>+</sup> insertion and extraction process. Additionally, the void space between the carbon shell and Sn core decreases obviously. However, the carbon shell is the same as the pristine state and the Sn nanoparticles are still encapsulated within the shell. Nevertheless, the morphology change is relatively obvious after 500 cycles. According to **Figures 5c,d**, most of the Sn cores transform into smaller nanoparticles with the diameter range 5–10 nm or even smaller, which is due to the gradually accumulated press/strain caused during the lithiation/delithiation process. Actually, those ultra-small Sn nanoparticles tend to accumulate to a bigger one because the ultra-small nanoparticles have extremely high specific surface area and high surface energy, which could reverse the transform process and cause the rapid capacity decay (Qin et al., 2016). As shown in **Figure 5d**, a polymericgel-like layer is also formed outside of the Sn@C composite. Such a structure coating outside the Sn@C yolkshell anode could enhance the mechanical cohesion and offer more lithium interfacial storage sites, which is called "pseudocapacitance-type behavior" (Purbia and Paria, 2015; Cheng et al., 2016). Moreover, the carbon shell is able to lock the nanoparticle in each carbon nanobox, and the rearrangement of Sn nanoparticles could reduce the stress and strain of the electrode and thus reserve the integrated yolk-shell structure (Xu et al., 2012; Leenheer et al., 2016). Such a phenomenon is observed in MnO (Leenheer et al., 2016), which also exhibits the reversible capacity increase of 800 mAh g−<sup>1</sup> at 5 A g−<sup>1</sup> from 200 to 2,000 cycles. Cao etc. explained that more active materials in carbon sheets may crush into small size nanograins upon cycling, which could increase the specific surface to some extent, thus leading to the enhancement of the interfacial Li<sup>+</sup> storage ability and the improved capacity (Yu et al., 2006; Sun et al., 2013; Liu et al., 2014; Xiao and Cao, 2015; Lian et al., 2017).

To investigate the improved chemical performance after long cycles, the rate performance, and cycling properties at high current density are measured. As shown in **Figure 6A**, after 500 cycles at 0.8 A g−<sup>1</sup> , rate performance of Sn@C is tested at 0.2, 0.5, 1.0, 2.0, and 4.0 A g−<sup>1</sup> and the capacities are 1036.9, 956.0, 839.2, 701.8, 395.8 mAh g−<sup>1</sup> respectively, which are almost more twice than the capacities of the pristine electrode. From the CV curve (**Figure 4A**), the electrode also shows relatively smooth curves

of charge and discharge after long cycles, which is caused by the combination of capacitive contribution and diffusion-controlled reaction. According to the charge-discharge curves (**Figure 4B**), the potential plateaus becomes shorter and almost disappears and the curve becomes nearly straight line after long cycles, explaining the increase of the extra pseudocapacity (Purbia and Paria, 2015). Besides, the electrode is also tested at 3 A g−<sup>1</sup> after 500 cycles at 0.8 A g−<sup>1</sup> , the reversible capacity (**Figure 6B**) maintains 366.2 mAh g−<sup>1</sup> after 930 cycles and the coulombic efficient is almost 100%. Compared to previous reports, this yolk-shell Sn@C composite exhibits improved reversible capacity of 694.4 mAh g−<sup>1</sup> between 16th to 500th cycles at a current density of 0.2 A g−<sup>1</sup> , which have been not observed before (The electrochemical performances of other Sn/C composites are shown in the **Table S1**). Besides, at a current density of 0.8 A g−<sup>1</sup> , the obvious increase between 37th to 500th cycles approach 402.1 mAh g−<sup>1</sup> . Compared with the Sn NPs without carbon shell, the electrochemical performance of Sn@C is considerably improved. Moreover, this concept can also be employed to other anode materials.

The excellent electrochemical performance is contributed by the following aspects: firstly, the uniform Sn nanoparticles are completely encapsulated in the amorphous carbon shell, which prevents the aggregation of the nanoparticles, and the hollow space between Sn core and carbon shell could significantly accommodate the volume variation to protect the carbon shell from pulverization; secondly, the process of forming ultra-small Sn nanoparticles during lithiation/delithiation to maintain the yolk-shell structure; thirdly, the ultra-small Sn nanoparticles of 10–20 nm after long-term cycles have a larger specific surface area which could provide more Li<sup>+</sup> insertion location; finally, the polymeric gel-like layer may provide the extrinsic pseudocapacitance after long-term cycles. In conclusion, the transformation of morphology and structure of the yolkshell Sn@C nanoboxes is mainly responsible for the gradual rise of the reversible capacity and the long-term cycling ability.

### CONCLUSION

In summary, we report a facile method to synthesize novel Sn@C nanoboxes into yolk-shell structure with enhanced reversible capacity. The as-prepared Sn@C nanoboxes electrode exhibit good electrochemical performance that can respectively deliver a reversible capacity of 674.6 mAh g−<sup>1</sup> at 0.8 A g−<sup>1</sup> and 1032.2 mAh g−<sup>1</sup> at 0.2 A g−<sup>1</sup> after 500 cycles. There are two main reasons for this distinguished performance: firstly, the Sn nanoparticles are encapsulated completely in the carbon shell and the void space can significantly accommodate the volume expansion. Besides, the SEI film forms outside the carbon

### REFERENCES


shell could maintain steadily during charge/discharge processes. Secondly, the ultra-small Sn nanoparticles gradually crush from the original Sn particles, which can effectively relieve the stress concentration on the electrode during the long cycles. Overall, the proposed concept in this work could be applied to prepare other metal/carbon composites with desired performance in next-generation LIBs electrode materials.

### AUTHOR CONTRIBUTIONS

QZ and M-SW designed this project. ZY carried out the material preparation and electrochemical test; ZY, ZZ, YC, and PL carried out and analyzed the XRD, SEM, and TEM analysis; ZY and H-HW wrote the paper. All authors discussed the results and revised the manuscript.

### ACKNOWLEDGMENTS

This work is financially supported by the Fundamental Research Funds for the Central Universities (Xiamen University: 20720170042) and the National Natural Science Foundation of China (Grants No. 21703185).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2018.00533/full#supplementary-material


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Yang, Wu, Zheng, Cheng, Li, Zhang and Wang. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Plane Double-Layer Structure of AC@S Cathode Improves Electrochemical Performance for Lithium-Sulfur Battery

#### Zengren Tao<sup>1</sup> , Zhiyun Yang<sup>1</sup> , Yafang Guo<sup>1</sup> , Yaping Zeng<sup>1</sup> \* and Jianrong Xiao1,2 \*

*<sup>1</sup> College of Science, Guilin University of Technology, Guilin, China, <sup>2</sup> Guangxi Key Laboratory of Electrochemical and Magnetochemical Functional Materials, Guilin University of Technology, Guilin, China*

#### Edited by:

*Qiaobao Zhang, Xiamen University, China*

### Reviewed by:

*Biao Gao, Wuhan University of Science and Technology, China Baihua Qu, Xiamen University, China Shengkui Zhong, Soochow University, China*

#### \*Correspondence:

*Yaping Zeng yapingz@126.com Jianrong Xiao xjr@glut.edu.cn*

#### Specialty section:

*This article was submitted to Physical Chemistry and Chemical Physics, a section of the journal Frontiers in Chemistry*

Received: *17 July 2018* Accepted: *07 September 2018* Published: *29 October 2018*

#### Citation:

*Tao Z, Yang Z, Guo Y, Zeng Y and Xiao J (2018) Plane Double-Layer Structure of AC@S Cathode Improves Electrochemical Performance for Lithium-Sulfur Battery. Front. Chem. 6:447. doi: 10.3389/fchem.2018.00447* Due to the high theoretical specific capacity of lithium-sulfur batteries, it is considered the most promising electrochemical energy storage device for the next generation. However, the development of lithium-sulfur battery has been restricted by its low cycle efficiency and low capacity. We present a Plane double-layer structure of AC@S cathode to improve the electrochemical performance of lithium-sulfur batteries. The battery with this cathode showed good electrochemical performance. The initial discharge capacity of the battery with the structure of AC@S cathode could reach 1,166 mAhg−<sup>1</sup> at 0.1 C. After 200 cycles, it still remains a reversible capacity of 793 mAh g−<sup>1</sup> with a low fading rate of 0.16% per cycle. Furthermore, the batteries could hold a discharge capacity of 620 mAh g−<sup>1</sup> after 200 cycles at a typical 0.5 C rate. The improvement of electrochemical performance is attributed to that the polysulfide produced during charge/discharge can be better concentrated in the cathode by the planar double-layer structure, thus reducing the loss of sulfur.

Keywords: plane double-layer structure, active material, diffusion, polysulfide adsorption, lithium-sulfur battery

## INTRODUCTION

With the development of the global electric automobile and the requirement of users for the endurance mileage, the demand for high-energy density batteries has increased unprecedented (Manthiram et al., 2015). Lithium-sulfur batteries have a high theoretical specific capacity (1,675 mAh g−<sup>1</sup> ), abundant reserves of sulfur and low production cost, it has been considered as the most promising electrochemical energy storage device (Song et al., 2016). As a result, many researchers have been attracted to improve electrochemical performance of Lithium - sulfur batteries.

However, there still are some problems restricting the commercialization of lithium-sulfur batteries:(1) Poor conductivity of sulfur (5 × 10−<sup>30</sup> S cm−<sup>1</sup> ) and discharge products lithium sulfides (3.6 × 10−<sup>7</sup> S cm−<sup>1</sup> ), poor reversibility of the discharge product lithium sulfides, easy to lose electrochemical activity, which would results in the loss of active-material. (2) During discharge, sulfur is first reduced to long chain polysulfide ions and dissolved into organic electrolytes. The dissolved long chain polysulfide ions S2<sup>−</sup> n (n ≥ 4) migrate through the separator to the negative electrode and are reduced to short chain polysulfide ions. Among them, some shortchain polysulfide ions remigrate back to the positive electrode, resulting in a "shuttle effect." The stronger "shuttle effect," the more overcharge obviously of the battery. Another part of the short-chain polysulfide is further reduced to insoluble substance Li2S2/Li2S on the anode of lithium, results in a slow discharge/charge progress and a low practical capacity (Lin et al., 2017). In the course of repeated shuttles, continuous loss of active substances S, leads to the continuous attenuation of the battery capacity and the deterioration of the cycle performance. (3) The density of elemental sulfur (2.03 g cm−<sup>3</sup> ) and Li2S (1.67 g cm−<sup>3</sup> ) are different, obvious volume expansion will occur during the cycling process, resulting in the destruction of the sulfur cathode (Rauh et al., 1979; Peled et al., 1989; Cheon et al., 2004; Mikhaylik and Akridge, 2004; Barchasz et al., 2012; Manthiram et al., 2015; Urbonaite et al., 2015; Song et al., 2016).

In order to solve these problems, people have made a lot of meaningful efforts, for example, designing of cathode material (Xiao et al., 2015a; Zhao et al., 2017), the modification of separator (Guo et al., 2017a,b, 2018), the protection of negative electrode and the improvement of electrolyte system (Yan et al., 2013; Chen et al., 2017). In these researches, the exploration of sulfur cathode materials is particularly outstanding. Among cathode materials, the most representative are sulfur/carbon composites (Schuster et al., 2012; Ma et al., 2014), sulfur/conductive polymer composites (Xiao et al., 2012; Zhang et al., 2012) and sulfur/oxide composites (Ma et al., 2015; Yuan et al., 2017). However, these cathode materials are only synthesized by various physical and chemical methods, seldom works are on cathode structural design of lithium-sulfur batteries. Therefore, researchers designed the structure for cathode material, and proposed various hierarchical structure cathode materials, for example:Coaxial carbon nanotube structure (Zhang et al., 2016, 2018; Li X. et al., 2017; Wang et al., 2017), spherical nanodelamination structure (Huang et al., 2017; Ni et al., 2017; Cheng et al., 2018), and plane hierarchical structure (Chung et al., 2016; He et al., 2016; Huang et al., 2017; Li G. et al., 2017; Ni et al., 2017; Zhao et al., 2018). The preparation process of the planar hierarchical structure is relatively simple, and the prepared battery has good electrochemical performance. The hierarchical structure exhibits a lot of excellent properties, firstly, it can protect the active-material between the interlayer from escaping, thus reducing the loss of the active material and improves the cycle stability (Zhang et al., 2016; Ni et al., 2017). Secondly, the hierarchical structure can provide an efficient conductive network for the active substances thus improve the conductivity, then improving the Coulomb efficiency (He et al., 2016; Li G. et al., 2017). Thirdly, we can change the thickness of each layer, for example, increase the thickness of the active material layer, reduce the thickness of the barrier layer, and therefore increase the capacity of the active material (Chung et al., 2016). Therefore, the electrochemical performance of lithium-sulfur battery cathode materials can be improved by using hierarchical structure cathode materials.

Activated carbon has porous structure and good adsorption performance, sulfur can be filled into the mesh of activated carbon and can be adsorbed strong, so the sulfur loss can be reduced by using activated carbon as the framework material of sulfur (Ji et al., 2009; Schuster et al., 2012; Lee et al., 2014; Zhang J. et al., 2014; Zhang S. et al., 2014; Xiao et al., 2015b). In this study, the sulfur and carbon planar double layer structure cathode materials were prepared by using activate carbon (AC) and sulfur in different proportions. We propose a simple preparation method to fabricated cathode materials. The test results show good electrochemical performance: The initial discharge specific capacity of a lithium-sulfur battery with two-layer structure of AC@S cathode could reach 1,166 mAh g−<sup>1</sup> at 0.1 C. After 200 cycles, it can still deliver a reversible capacity of as high as 793 mAh g−<sup>1</sup> with a low fading rate of 0.16% per cycle, and a capacity-retention rate of 68% after 200 cycles. The reason for the hierarchical cathodes display good cyclability is explored by analyzing their activation process and the excellent polysulfide retention brought about by the hierarchical electrode structure.

### EXPERIMENTAL SECTION

### Preparation of AC/S Active Material

First, S and AC (In different proportions 9:1, 1:9, 8:2, 2:8, 7:3, 3:7) were first placed in an agate mortar and ground for 1 h before being transferred to a polytetrafluoroethylene reaction vessel. Then, to exclude the residual air so that S would not be oxidizing at a high temperature, we ensured that the reaction vessel was static and open in an argon-filled glove box for 0.5 h. After that, the reaction vessel removed from the glove box and heated at 155◦C for 12 h. At this temperature, the melted sulfur can easily penetrate into the pores of AC. Finally, the AC/S composite was obtained after cooling down to room temperature.

## Preparation of Hierarchical Cathodes

The working electrodes were prepared from various ratios a mixture of the S/AC composite, Super P, and polyvinylidene fluoride (PVDF) binder in N-methyl-2- pyrrolidinone solution (NMP) with a mass ratio of 8:1:1. The mixed active-material paste was first dropped onto the middle of the aluminumfoil current collectors. The following five samples of doublelayer structure cathode cells were prepared: The first layer was 8: 2, the second layer was 2:8 and marked AC/S-1; The first layer was 9: 1, the second layer was 1:9 and marked AC/S-2; The first layer was 7: 3, the second layer was 3:7 and labeled AC/S-3; The first layer was S, the second layer was AC and marked AC/S-4; The sulfur monolayer labeled AC/S-5. The first layer is about 14µm, and the second is about 13µm. These samples were dried in a vacuum oven at 60◦C for 12 h before cutting. Finally, the cathode was punched into a disk measuring 14 mm in diameter for assembling.

## Cell Assembly

The CR-2025-type button cells is using to tests all of the electrochemical for the sulfur cathode, which were assembled in an argon-filled glove box using various ratios a mixture of the S/AC composite plane double layer structure anode/separators and Li metal as the counter electrode. The electrolyte used in this study was 1 M Li TFSI/DME + DOL (1:1, v/v) containing LiNO3 (1 wt%). The hierarchical cathodes and lithium anodes were connected to, respectively, an aluminum tab and a nickel tab. Pouch cells were sealed with an aluminum soft packaging film. The assembled coin and pouch cells were allowed to rest for 30 min at 25◦C before electrochemical measurements.

### Material Characterization

Field emission scanning electron microscopy (SEM, HITACHIS-4800) was used to characterize the morphology at before and after 200 cycles of the AC/S-1 cathode. The distribution of the elements on the surface of the AC/S-1 cathode was identified with an energy dispersive spectrometer (EDS). X-ray diffraction (XRD, X'Pert PRO) was used to characterize the crystallinity of AC/S-X (X = 1, 2, 3, 4, 5).

### Electrochemical Measurement

The CHI750E electrochemical workstation was used to measure the cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS). Discharge and charge profiles and cyclability data were evaluated under galvanostatic conditions between 1.5 and 2.8 V with a programmable battery cycler. CV measurements were performed at a scan rate of 0.01 mV s−<sup>1</sup> in the voltage range

of 1.5–3.0 V. The EIS tests were carried out at the frequency range of 0.01–100 kHz with a perturbation amplitude of 5 mV.

### RESULTS AND DISCUSSION

### Configuration and Morphology

The configuration of the hierarchical cathode is shown in **Figure 1**. In the plane-double layer structure, the electrochemically stable carbon film is covered onto a layer of active-material coating as a carbon-film shield to form the hierarchical cathode. Polysulfides Li2S<sup>n</sup> (4≤n≤8) of AC/S-5 monolayer cathode can be easily escaped from the cathode without protection, resulting in a great loss of S. Although the AC/S-4 cathode is protected by carbon layer, the polysulfide can easily escape from the edge of cathode and cause sulfur loss. However, due to the existence of AC in both layers of the cathode of AC/S-1 plane double-layer structure, because sulfur coated by activated carbon of each layer, the polysulfide produced during

discharge can be adsorbed by AC and migrated to the upper layer, which can provide enough space for polysulfide, thus, the loss of sulfur can be minimized.

From **Figure 2**. The single mass sulfur peak is strong and sharp (AC/S-5), and there are obvious diffraction peaks in the whole scanning range, indicating that the material has stronger crystal structure, and the main diffraction peak is located at 2θ = 23.083◦ and 2θ = 27.769◦ , which belongs to the skew square type diffraction peak and is a typical S8 structure (Yuan et al., 2009; Xiao et al., 2015b). The activated carbon had a "steamed bread peak" at 2θ = 20∼30◦ , which showed an amorphous state. The four kinds of cathode materials showed typical "steamed bread peak" with activated carbon, and there were diffraction peaks of crystal sulfur, but the intensity of the crystal peak of sulfur decreased obviously. The results show that both crystallinity and amorphous states of sublimated sulfur were found in the composites. It can be seen that the diffraction peaks of AC/S-1 and AC/S-3 cathode are obvious because the high content of S in the surface layer (the content of S of AC/S-1 is 20% and the content of S of AC/S-3 is 30%), and most of the sulfur goes into the nano-pores of activated carbon, but a small amount of sulfur is deposited on the surface of carbon material. However, the diffraction peaks of AC/S-3 and AC/S-4 cathode materials are weakened, which is due to the low content of S in the surface layer (the content of S of AC/S-3 is 10% and the content of S of AC/S-3 almost 0), so basically all the S enter into the nano-micropores of activated carbon, and enhance the amorphous structure of the materials.

The SEM diagram of the positive pole of AC/S-1 before and after 200 cycles is shown in **Figure 3**. Before cycling, the surface of the AC/S-1 cathode is covered with activated carbon particles of varying sizes bonded together by PVDF, and there are abundant pores between the particles, these pores can not only store lithium-sulfur battery electrolyte, but also provide space for intermediate products and inhibit the shuttle effect during reaction.

The SEM contrast diagram of AC/S-1 cathode after 200 cycles at 0.1 C is given in **Figures 3c,d**. It is clear from the diagram that the surface of the carbon particles at the positive pole is relatively smooth before cycle. The surface of the carbon particles deposited has a large amount of scalelike pimples after cycle, these uniformly distributed pimples are intermediate products Li2Sn, which is intercepts by the AC/S-1 cathode during the charging and discharging of the battery (Guo et al., 2017a,b). The EDS surface scan of the four elements (S, C, O, F) of AC/S-1 cathode before and after

cycle in **Figure 4a.** The distribution of element C is uneven, and there is obvious agglomeration, but the signal of three elements O, F, S is very weak. After cycles, the signal of element C was strengthened in **Figure 4b**, and the phenomenon of agglomeration disappeared, while the signals of three elements of O, F, C were obviously enhanced and uniformly dispersed. These

differences can be attributed to the adhesion of polysulfide to active carbon hierarchical cathode material and the accumulation of electrolyte.

The contrast diagram of SEM cross section of AC/S-1 cathode before and after cycle are shown in **Figures 5a,b**. It is obvious that the delamination can be seen in the hierarchical structure cathode before the cycle, but it difficult to distinguish after 200 cycles at 0.1 C. This is because of the polysulfide continuous diffusion to the surface layer, and the active substance uniform diffusion throughout the cathode during the battery cycle.

The cross section of the layered cathode by EDS are presented in **Figures 6a,b**. It can be seen from the diagram that the content of S in the first and second layers is obviously different before the cycle. It is evident that sulfur-containing species start to diffuse out from the active-material layers and are absorbed by AC of upper layer, as a result of the stabilized polysulfide migration (Chung et al., 2016). After 200 cycles, the diffusing polysulfidesare stabilized within the hierarchical electrode and the elemental sulfur signals are, therefore, uniformly distributed in the conductive matrix.

### Electrochemical Stability

In order to determine what proportion of AC and S for prepare the planar double - layer structure cathode play the best role, and greatly improve electrochemical performance of lithiumsulfur battery. We have measured the cycle performance and the C-rate performance of the five kinds of AC/S-X(X = 1, 2, 3, 4, 5) batteries, at different discharge current rates. Excellent electrochemical utilization and stability allow the hierarchical cathodes to attain high discharge capacity and long-term cyclability for 200 cycles at various cycling rates as shown in **Figure 7**. Coulomb efficiency of AC/S-1 greater than 90% (**Figure 7A**), it indicate that the structure cathode battery has higher reversible capacity. As can be seen from the **Figure 7B**, the initial discharge capacities of AC/S-X(X = 1, 2, 3, 4)at 0.1 C reached 1,166,1,057,939,721 mAh g−<sup>1</sup> ,AC/S-X(X = 1, 2, 3)cells at after 200 cycles, the reversible discharge capacities were 793,686,556 mAh g−<sup>1</sup> ,AC/S-4 cells at after 100 cycles is 612 mAh g−<sup>1</sup> respectively. Compared with the original sulfur monolayer structure cell (AC/S-5), the cycle performance obviously improved, especially when the first layer AC: S = 2: 8 and the second layer AC: S = 8:2, the performance tends to be optimal. This is because when the AC content of first layer reaches 20%, it can provide enough pores to adsorb S, when the AC content of second layer reaches 80%, it can effectively prevent the polysulfide from escaping in the cathode, and as far as possible to reduce the loss of S, alleviates the occurrence of the shuttle effect. Therefore, the first discharge specific capacity of planar double-layer structure battery with AC/S-1 ratio can reach to 893 mAh g−<sup>1</sup> at 0.5 C, after 200 cycles the battery still maintain a good capacity(about 620 mAh g−<sup>1</sup> ), as shown in the **Figure 7D**.

**Figure 8A** shows the C-rate performance of the battery, which is another important aspect to evaluating the electrochemical performance of lithium sulfur battery. The AC/S-X batteries was tested at varying rates (0.1 C→0.2 C→0.3 C→0.5 C→0.1 C). It can be seen from the graph, that the initial discharge capacity of the cell with the AC/S-1 hierarchical cathode reached 1139 mAh g −1 . Although the capacity of the battery declined rapidly in the first six cycles, the capacity of the battery began to stabilize with the increase of the current ratio. Capacity of the battery slow descent from 990 mAh g−<sup>1</sup> at 0.1 C to 922, 799, and 540 mAh g −1 at 0.2, 0.3, and 0.5 C respectively. More importantly, when the current rate is back to 0.1 C, the capacity of the AC/S-1 planar double-layer structure battery is obviously higher than that of the other four sample batteries, it is shown that the planar double layer structure of this AC/S ratio has a more effective limiting

FIGURE 10 | Electrochemical impedance spectrum (EIS) of the cells with AC/S-X cathode (A) before cycling and (B) after 200 cycles at 0.1 C, (C) Equivalent circuit used for fitting the impedance spectra.

effect on the diffusion of polysulfide. From the charge/discharge profiles at various rates shown in **Figure 8B**, the AC/S-1 planar double-layer structure characteristic can be clearly identified at 0.1 C, has a longer charge/discharge curve, indicative of low polarization, sulfur can be fully react to form Li2S2/Li2S, thus enhances the utilization of active material (Zhang et al., 2017).

Typical cyclic voltammograms (CVs) at a scan rate of 0.1 mV s−<sup>1</sup> and the potential range of 1.5–3 V are presented in **Figure 9A**. There are two cathodic peaks, relating to the formation of high-order S<sup>n</sup> (4 ≤ n ≤ 8) and insoluble Li2S2/Li2S. Therefore, in the subsequent anodic scan, two oxidation peaks are observed, corresponding to the oxidation of insoluble Li2S2/Li2S to soluble polysulfides and polysulfides to sulfur (Lee et al., 2002; Zhang, 2013; Zhu et al., 2016; Zhang et al., 2017). While comparing with the other batteries, the AC/S-1 planar doublelayer structure battery possesses larger and stable current density of redox peaks in CVs, explaining a low polarization, good reversibility, and excellent cycle stability. Thus, the cathode of the planar double-layer structure can improve the electrochemical performance of the battery; in particular, the cathode of the AC/S-1 ratio has the greatest influence on the improvement of the electrochemical performance. Here are the charge-discharge curves of AC/S-1 batteries at the 2nd, 10th, 20th, 30th, and 40th times, at 0.1 C are presented in **Figure 9B**. The upperdischarge plateau at 2.3 V indicates the reduction of sulfur leads to the formation of soluble polysulfide. The fast sulfur(solid) polysulfides(liquid) reduction reactions involve the formation, dissolution, and diffusion of polysulfides (Huang et al., 2015; Li G. et al., 2017).

To get a further understanding with the contribution of the designed plane hierarchical structure on the performance, electrochemical impedance spectra (EIS) is shown in **Figure 10**. The electrochemical workstation was used to test the electrochemicl impedance with the frequency range of 0.01 Hz-100 kHz. The R1 in the illustrations denotes the resistance of the electrolyte,and R<sup>2</sup> denotes the charge transfer resistance of the battery, W<sup>1</sup> indicates that the Walburg diffusion impedance, CPE<sup>1</sup> is a constant phase original. Before and after the cycle, the electrochemical impedance curves of the sample cells are composed of the medium and high frequency band semicircle (corresponding to the charge transfer resistance) and low frequency band sloping line (corresponding to the

### REFERENCES


Warburg impedance). In addition, the ohmic resistance of the lithium-sulfur battery is derived from the intercept between the coordinate axis and the curve in the high frequency range (Kolosnitsyn et al., 2011; Deng et al., 2013; Zhang et al., 2015). By comparing the diameters of semicircle we know that the charge transfer resistance of the battery decreases before and after the cycle. Furthermore, due to the active-material redispersing process in AC/S-1 planar-double layer structure, enable the active-material to be fully utilized. The charge transfer resistance of the battery is the smallest after cycle.

## CONCLUSIONS

In summary, the planar double-layer cathode can enhance the electrochemical stability of the lithium-sulfur battery. The capacity attenuation rate of the battery is 0.16% and long term cyclability (more than 200 times) still keeps high capacity. The planar double-layer cathode used as the charge/discharge platform has the potential to increase the sulfur content and applied to the battery. The migration processes of polysulfides were discussed by observation of the cross-section microstructure and elemental analysis. We attributed the more stable electrochemical performances to the special planar doublelayer cathode, when active material moved to the upper layer, the structure would restrain the dissolve of polysulfides by the physical adsorption ability of activated carbon. Therefore, the utilization of the planar double-layer structure as sulfur cathodes presents a potential opportunity to ameliorate the long-term cycle stability of the corresponding Li-S batteries.

### AUTHOR CONTRIBUTIONS

ZT prepared all materials and performed electrochemical characterizations. ZY conducted SEM and XRD experiments. YG and ZT analyzed the data. ZT and YZ wrote the manuscript. JX supervised the implementation of the project.

### ACKNOWLEDGMENTS

This work was supported by Guangxi Key Laboratory of Electrochemical and Magnetochemical Functional Materials Open Foundation (No. EMFM20182203).


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Tao, Yang, Guo, Zeng and Xiao. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# High Performance Composite Polymer Electrolytes Doped With Spherical-Like and Honeycomb Structural Li0.1Ca0.9TiO<sup>3</sup> Particles

Wei Xiao<sup>1</sup> \*, Zhiyan Wang<sup>1</sup> , Chang Miao<sup>1</sup> , Ping Mei <sup>1</sup> , Yan Zhang<sup>1</sup> , Xuemin Yan<sup>1</sup> , Minglei Tian<sup>1</sup> , Yu Jiang<sup>1</sup> and Jingjing Liu<sup>2</sup> \*

*<sup>1</sup> College of Chemistry and Environmental Engineering, Yangtze University, Jingzhou, China, <sup>2</sup> Environmental Monitoring Department, Changsha Environmental Protection College, Changsha, China*

### Edited by:

*Jiexi Wang, Central South University, China*

## Reviewed by:

*Shaohua Luo, Northeastern University, United States Hongwei Mi, Shenzhen University, China*

#### \*Correspondence: *Wei Xiao*

*xwylyq2006@126.com Jingjing Liu ljjsx2007@163.com*

#### Specialty section:

*This article was submitted to Physical Chemistry and Chemical Physics, a section of the journal Frontiers in Chemistry*

> Received: *26 August 2018* Accepted: *11 October 2018* Published: *25 October 2018*

#### Citation:

*Xiao W, Wang Z, Miao C, Mei P, Zhang Y, Yan X, Tian M, Jiang Y and Liu J (2018) High Performance Composite Polymer Electrolytes Doped With Spherical-Like and Honeycomb Structural Li*0.1*Ca*0.9*TiO*3 *Particles. Front. Chem. 6:525. doi: 10.3389/fchem.2018.00525* The spherical-like and honeycomb structural Li0.1Ca0.9TiO<sup>3</sup> particles are prepared by spray drying combined with following calcination confirmed by X-ray diffraction (XRD) and scanning electron microscopy (SEM) with energy dispersive X-ray spectrometer (EDS). The poly (vinylidene fluoride-*co*-hexafluoropropylene) (P(VDF-HFP))-based composite polymer electrolytes (CPEs) modified with the particles are fabricated by phase inversion and activation processes. The characterization results show that the as-prepared CPE membranes possess the smoothest surface and most abundant micropores with the lowest crystallinity with adding the particles into the polymer matrix, which results in high ionic conductivity (3.947 mS cm−<sup>1</sup> ) and lithium ion transference number (0.4962) at ambient temperature. The interfacial resistance can be quickly stabilized at 508 Ω after 5 days storage and the electrochemical working window is up to 5.2 V. Moreover, the mechanical strength of the membranes gains significant improvement without lowering the ionic conductivity. Furthermore, the assembled coin cell can also deliver high discharge specific capacity and preserve steady cycle performance at different current densities. Those outstanding properties may be ascribed to the distinctive structure of the tailored spherical-like and honeycomb structural Li0.1Ca0.9TiO<sup>3</sup> particles, which can guarantee the desirable CPEs as a new promising candidate for the polymer electrolyte.

Keywords: Li0.1Ca0.9TiO3 , composite polymer electrolyte, spherical-like, honeycomb, lithium ion battery

### INTRODUCTION

At present, more and more attentions have been invested into the polymer electrolytes in the field of the lithium ion battery, which can not only realize the flexible assembly of the packed battery, but also effectively prevent the explosion and spontaneous combustion incidents of the assembled battery (Croce et al., 1998; Song et al., 1999; Scrosati, 2000; Wang et al., 2014; Wu et al., 2017b; He et al., 2018a). However, the solid polymer electrolytes (SPEs) are excluded from the practical use because of their quite low ionic conductivity of the assembled lithium ion battery (Murata et al., 2000; Xie et al., 2012; Kim et al., 2015; Zhang H. et al., 2017; Wu et al., 2018). The ionic conductivity at room temperature can gain noticeable improvement when the polymer electrolyte membranes adsorb some liquid electrolytes to form the novel gel polymer electrolytes (GPEs), which are mainly composed of the polymer matrix and the entrapped liquid electrolytes (Zhang et al., 2007; Hu et al., 2015; Zhang M. Y. et al., 2017; Wang et al., 2018c). However, the poor mechanical strength of the swollen GPEs cannot well maintain the separation between the positive and negative active materials during the long-term Li<sup>+</sup> insertion-extraction behaviors (Rao et al., 2012; Liu et al., 2017; Luo et al., 2017). Therefore, some countermeasures have been taken to circumvent the obstacles. Adding inert inorganic nano-particles, such as TiO<sup>2</sup> (Wu et al., 2011; Chen et al., 2015; Zhang et al., 2015; He et al., 2018b), Al2O<sup>3</sup> (Lee et al., 2014; Liang et al., 2015; Wu et al., 2017a), and SiO<sup>2</sup> (Park et al., 2010; Li et al., 2013; Huang et al., 2017; Shim et al., 2017), into the polymer matrix to fabricate the composite polymer electrolytes (CPEs) has been ever proved to be an available and appealing approach to balance the contradiction between the ionic conductivity and mechanical strength at room temperature (Costa et al., 2013; Wang et al., 2018b). However, the poor dispersibility of the added nano-inorganic particles into the polymer substrate is also the barrier for the as-fabricated CPEs. Perovskite structural ABO<sup>3</sup> particles doped with metal ions can produce more vacancies, which can facilitate the ions transfer to enhance the ionic conductivity as well as improving the mechanical strength (Kay and Bailey, 1957; Mather et al., 2007). In the work, spherical-like and honeycomb perovskite structural Li0.1Ca0.9TiO<sup>3</sup> particles are successfully synthesized by spray drying combined with sintering, and the desirable CPEs are prepared by traditional phase inversion method. The corresponding electrochemical and battery performance of the CPEs are also carefully studied.

### EXPERIMENTAL

### Preparation of the Li0.1Ca0.9TiO<sup>3</sup> Powders

The spherical-like and honeycomb structural Li0.1Ca0.9TiO<sup>3</sup> particles are prepared by spray drying combined with following calcination. At first, stoichiometric calcium acetate and lithium acetate are successively added into deionized water to form solution A, and the mixture of tetrabutyl titanate, acetylacetone, and absolute ethyl alcohol with the volume ratio of 4:1:4 is dropwise added into the solution A to obtain homogeneous solution B with continuous agitation at 40◦C for 6 h. Secondly, the solution B is pumped into the spray drier machine (SD-2500) by peristaltic pump at 1,500 mL h−<sup>1</sup> and atomized at 180◦C with atomizing pressure of 0.2 MPa to product the precursor powders with conditions similar to the scheme (Yang et al., 2014; Su et al., 2018). The target product Li0.1Ca0.9TiO<sup>3</sup> particles can be gained after the precursor powders are further calcined in air at 700◦C for 6 h. All chemicals are purchased from Macklin Biochemical Co., Ltd. with analytical grade and used without any further purification.

### Fabrication of the CPEs

The fabrication processes of the CPEs can be concisely summarized as fabrication and activation of the polymer electrolyte membranes. At first, a certain content of Li0.1Ca0.9TiO3, urea and poly (vinylidene fluoride-cohexafluoropropylene) P(VDF-HFP) (Atofina, Kynar Flex, 12 wt.% HFP) are successively fed into the N, N-dimethylformamide (DMF) solvent to yield uniform coating solution under vigorous agitation at room temperature, in which Li0.1Ca0.9TiO3, urea and P(VDF-HFP) powders are used as dopant, pore-forming agent and polymer matrix, respectively, and the weight ratio is maintained at about 5:1:100. After that, the homogeneous casting solution is cast onto a pre-cleaned glass substrate with a doctor blade to prepare the wet membranes. Then, the wet membranes are immersed into deionized water for 12 h at room temperature to thoroughly fulfill phase transfer and the free-standing CPE membranes can be obtained by being dried for 24 h at 60◦C to remove the residual solvent and then cut into a disk with a diameter about 16 mm for use (Wang et al., 2018b; Xiao et al., 2018). Secondly, the desirable CPEs can be acquired after immersing the as-prepared CPE membranes into 1.0 M LiPF6-EMC/EC/DMC (1:1:1, v/v/v) provided by Dongguan Shanshan Battery Materials Co., Ltd. for 0.5 h, which are labeled as CPE-LCT for convenience. The polymer electrolytes doped with commercial CaTiO<sup>3</sup> and without inorganic particles are also fabricated using the same method for comparison and named as CPE-CT and CPE-0, respectively.

### Properties Characterization

Scanning electron microscopy (SEM, JSM6301F) equipped with an energy dispersive X-ray spectrometer (EDS) and Xray diffraction (XRD, Rint-2000) are used to investigate the micro-structure, chemical composition, and physical phase constitution of the as-prepared Li0.1Ca0.9TiO<sup>3</sup> particles. The surface morphology and crystallinity of the as-fabricated CPE membranes are identified by SEM and XRD, respectively. The porosity (P) and liquid uptake rate (A) of the as-prepared CPE membranes are calculated according to the our previous work in the following Equations (1, 2), respectively, in which ρ<sup>a</sup> and ρ<sup>b</sup> are the density of n-butanol and the dry CPE membrane, m<sup>a</sup> (w1) and m<sup>b</sup> (w0) are the mass (weight) of the membranes with and without n-butanol, respectively (Xiao et al., 2016). The mechanical tensile strength (T) of the as-prepared CPE membranes is performed on an electronic universal testing machine (INSTRON-5500) at a crosshead speed of 100 mm min−<sup>1</sup> at room temperature, which refers to the literature (Wang et al., 2018a). To reveal the electrochemical performance, the as-fabricated CPEs are packed in the simulated battery in the following three types. The first type is that the CPE is assembled between two stainless steel (SS) electrodes, i.e., SS/CPE/SS, which is employed to test the ionic conductivity at various temperatures (293–363 K) by electrochemical impedance spectroscopy (EIS) on the CHI660E electrochemical workstation with frequency range 10<sup>5</sup> -1 Hz. The ionic conductivity can be calculated from the bulk resistance (R) according to Equation (3), where l is the thickness of the CPE, S is the effective contact area, R is obtained from the Nyquist impedance plots (Xiao et al., 2016). The second one is that the CPE is assembled between a lithium disk electrode and a SS electrode, i.e., Li/CPE/SS, which is utilized to evaluate the electrochemical working window of the CPE by the EIS from 2.50 to 6.65 V with a scanning rate of 5 mV s−<sup>1</sup> . The last one is that the CPE is packed between two lithium disk electrodes, i.e., Li/CPE/Li, which is used to monitor the interfacial resistance changes with different storage times at 30◦C. The lithium ion transference number introduced by the literature (Bruce and Vincent, 1987; Xiao et al., 2016) of the symmetrical Li/CPE/Li cells is measured by the combination of Dc polarization and EIS techniques. The corresponding value can be calculated according to the reference in the following Equation (4), in which I<sup>0</sup> and Iss are the initial and the steady current, R<sup>0</sup> and Rss are the initial interfacial and the steady-state resistance, respectively, and 1V is the potential difference (Xiao et al., 2016).

$$P\% = \frac{m\_d \Big/\_{\rho\_a}}{\binom{m\_a \Big/\_{\rho\_a}}{\rho\_a} + \binom{m\_b \Big/\_{\rho\_b}}{\rho\_b}} \times 100\% \tag{1}$$

$$A\% = \frac{\mathcal{w}\_1 - \mathcal{w}\_0}{\mathcal{w}\_0} \times 100\% \tag{2}$$

$$\sigma = \bigvee\_{\{\mathbf{R} \cdot \mathbf{S}\}} \tag{3}$$

$$T\_{Li^{+}} = \frac{I\_{\rm ss}(\Delta V - R\_{\rm ss}I\_{0})}{I\_{0}(\Delta V - R\_{0}I\_{\rm ss})} \tag{4}$$

The charge-discharge cycling performance of the assembled Li/CPEs/LiCoO<sup>2</sup> coin cells are performed on a Land Battery Test System (Wuhan Land Electronic Corporation, China) at different current densities (0.1, 0.2, 0.5, 1.0, and 2.0 C) with cutoff voltages of 2.75–4.25 V at room temperature, in which the working electrode is fabricated by casting the mixed slurry of 80 wt.% LiCoO<sup>2</sup> as active material, 10 wt.% acetylene black as conductive agent and 10 wt.% polyvinylidene fluoride as binder onto aluminum foils with mass loading of 2–3 mg cm−<sup>2</sup> , and the thickness of the as-fabricated CPEs is about 120µm (Xiao et al., 2016, 2018; Wang et al., 2018b).

### RESULTS AND DISCUSSION

**Figure 1** displays the micro-structure, chemical composition, and physical phase constitution of the as-prepared Li0.1Ca0.9TiO<sup>3</sup>

particles. It can be clearly seen from **Figures 1A,B** that the SEM images present homo-dispersed spherical-like particles with the average diameter of 250 nm, which can be confirmed by the enlarged SEM image of the area 1 demonstrated in **Figure 1C**. It is worth noting that the Li0.1Ca0.9TiO<sup>3</sup> particles show coarse and damaged surface in **Figure 1C**. It can be observed from the enlarged SEM image of the area 2 demonstrated in **Figure 1D** that the coarse and damaged surface of the as-prepared particles has abundant honeycomb structures. Therefore, we can speculate that the inside of the as-prepared Li0.1Ca0.9TiO<sup>3</sup> particles may possess well-developed interpenetrating porous network structures. Those results indicate that the as-prepared Li0.1Ca0.9TiO<sup>3</sup> particles can not only absorb and retain more liquid electrolytes, but also provide more extra passages for lithium ions transfer, which can markedly improve the battery performance. **Figure 1E** demonstrates the EDS results of the selected area 2 in **Figure 1C**. It can be obviously found that the as-synthesized nano-inorganic fillers are composed of Ca, Ti, and O elements and the percentage of Ca and Ti atoms is about 0.902:1, which is in accord with the predetermined stoichiometric ratio. The Li element cannot be detected in the EDS plots because of its less content and lighter weight (Sabiha et al., 2017). As demonstrated in **Figure 1F**, the XRD patterns of the as-synthesized Li0.1Ca0.9TiO<sup>3</sup> particles are well indexed to the perovskite structural CaTiO<sup>3</sup> (JCPDS#82-0228), which would benefit the lithium ions transfer. Moreover, few impurities corresponding to Li2TiO<sup>3</sup> (JCPDS#51-0050) can be detected due to the excessive added lithium salt. Those results suggest that the spherical-like and honeycomb structural Li0.1Ca0.9TiO<sup>3</sup> particles are successfully synthesized by spray drying combined with following calcination.

The SEM images of the polymer electrolyte membranes doped with different kinds of inorganic particles are displayed in **Figure 2**. The surface of the as-fabricated CPE membranes presents significant differences with adding the inorganic particles into the polymer substrate, where the surface gets smoother and the pore-size distribution becomes more uniform compared with the pure polymer electrolyte membrane. Although the added particles are obviously discovered in the CPE-CT and CPE-LCT membranes, few aggregations appear on the surface of the CPE-CT membrane displayed in **Figure 2B**, which can be ascribed to the high specific surface energy of the added commercial nano-CaTiO<sup>3</sup> powders. The as-synthesized Li0.1Ca0.9TiO<sup>3</sup> particles can be uniformly dispersed into the polymer matrix demonstrated in **Figure 2C** because of their unique spherical-like and honeycomb structure, which can be confirmed by the partial enlarged SEM image of the selected area 1 displayed in **Figure 2D**. The added Li0.1Ca0.9TiO<sup>3</sup> particles well

TABLE 1 | Porosity, uptake rate, tensile strength, ionic conductivity, activation energy, and lithium ion transference number of different CPE membranes at room temperature.


dispersed into the polymer electrolyte membranes with abundant micro-pore structures can not only enhance the mechanical properties by providing more physical crosslinking sites, but also improve the ionic conductivity by retaining more liquid electrolytes at the same time, which can be proved by the results listed in **Table 1**. As listed in **Table 1,** the porosity, uptake rate and tensile strength of the as-prepared CPE membranes gain remarkable improvements with doping the inorganic fillers into the polymer substrate, in which the values of the CPE-LCT membrane can reach to the highest value of about 92.73, 168.5%, and 27.32 Mpa, respectively. Clearly, these results can be mainly to the unique spherical-like and honeycomb structural of the added Li0.1Ca0.9TiO<sup>3</sup> particles, in which the external honeycomb structure can provide more space to store liquid electrolytes and transfer lithium ions and the inert spherical-like structure can maintain the mechanical properties of the membranes.

**Figure 3** demonstrates the XRD patterns of the as-prepared Li0.1Ca0.9TiO<sup>3</sup> powders and different CPE membranes. It can be observed from **Figure 3** that the XRD patterns of the doped CPE membranes present distinct differences compared to the pristine P(VDF-HFP) one, in which the characteristic diffraction peaks of the polymer matrix P(VDF-HFP) at around 18.3 and 26.6◦ disappear and the one at 20.0◦ gets weak. Those results suggest that adding inorganic particles into the polymer matrix can impair the crystallinity of the polymer electrolyte membranes and release more amorphous areas for lithium ions transfer, which are the same as our previous results (Wang et al., 2018c). What is noteworthy is that the CPE-LCT membrane demonstrated in **Figure 3** presents the weakest intensity, and we have reasons to speculate that the CPE-LCT may have high ionic conductivity at room temperature, which can be partly attributed to the Lewis acid-base interactions between the doped particles and the polymer chains, and partly to the increasing amorphous areas

(Xiao et al., 2014a). Moreover, the new peak can be observed at 32.6◦ in the doped CPE membranes, which can be ascribed to the characteristic diffraction peak of the as-synthesized perovskite Li0.1Ca0.9TiO<sup>3</sup> inorganic particles that are well indexed to the JCPDS#82-0228.

To further investigate the ionic conductivity of the as-prepared CPEs, we have tested the EIS pots of the assembled SS/CPE/SS battery at different temperatures and the corresponding curves on reciprocal temperature dependence of ionic conductivity are demonstrated in **Figure 4A**. It can be distinctly discovered that the curve of the SS/CPE-LCT/SS battery is always above the other ones, which indicates the CPE doped with Li0.1Ca0.9TiO<sup>3</sup> powders may possess the highest ionic conductivity. The ionic conductivity at room temperature of the various CPEs is demonstrated in **Table 1**, in which the value of the CPE-0, CPE-CT, and CPE-LCT is about 1.954, 2.818, and 3.947 mS cm−<sup>1</sup> , respectively. As displayed in **Figure 4A**, the plots of the reciprocal temperature vs. ionic conductivity can well fitted by the Arrhenius equation, which indicates that the lithium ions transfer in the CPEs system is mainly attributed to the free diffusions rather than the hopping movements elaborated in our previous work (Xiao et al., 2014b, 2016, 2018; Wang et al., 2018a). The lithium ions can easily migrate from the inside of the spherical-like and honeycomb structural Li0.1Ca0.9TiO<sup>3</sup> particles to the interface between the electrodes and the as-prepared CPEs to participate into the intercalation and deintercalation processes in the CPEs system. The corresponding threshold energy of the lithium ions transfer for the different CPEs is calculated and listed in **Table 1**. As demonstrated in **Table 1,** the CPE-LCT has the lowest activation energy of about 4.019 KJ mol−<sup>1</sup> . The highest ionic conductivity with the lowest activation energy at room temperature of the as-prepared CPE-LCT can be explained as the following reasons. Firstly, the appropriate size and uniform distribution of the micropores in the as-fabricated CPE-LCT membrane can adsorb and hold more liquid electrolytes, which can significantly promote the ionic conductivity of the system. Secondly, the Lewis acid-base interactions between the added inorganic particles and the polymer matrix can release more amorphous areas for lithium ions transfer (Xiao et al., 2018). Moreover, the lithium ions can migrate in the extra passageways provided by the abundant and interpenetrated micropores in the as-prepared honeycomb structural Li0.1Ca0.9TiO<sup>3</sup> particles. The as-synthesized Li0.1Ca0.9TiO<sup>3</sup> particles can not only dramatically enhance the ionic conductivity, but also improve the mechanical strength of the corresponding doped CPE membranes at the same time, which are principally ascribed to the excellent mechanical performance of the unique spherical-like and honeycomb structure. As displayed in **Figure 4B**, the mechanical strength is sharply enhanced from 18.84 MPa of the CPE-0 membrane, to 26.87 MPa of the CPE-CT one, and to 27.32 MPa of the CPE-LCT one. Moreover, the elongations of the CPE-CT and CPE-LCT membranes are also apparently greater than the one of the CPE-0 membrane, in which the elongation at the critical point is about 19.7 and 21.4% for the CPE-CT and CPE-LCT membranes, respectively, and the value is only 13.1% for the pure one. Those results can be ascribed to the more crosslinking sites provided by the doped particles, which can well

improve the flexibility of the as-prepared polymer electrolyte membranes to enhance the battery performance.

**Figure 5** displays the LSV plots of the different CPE membranes. The electrochemical working window of the assembled SS/CPEs/Li cell with inorganic particles is much wider than the one of the cell with the CPE-0, in which the value of the steady electrochemical working window is about 4.3, 5.0, and 5.2 V for the CPE-0, CPE-CT, and CPE-LCT membrane, respectively. The dramatically enhanced working voltage of the doped CPE membranes can be resulted from the gelatinization of the entrapped liquid electrolytes and the added inert nano-particles. The gelation enhances the stability of the CPE membranes due to the interactions between the entrapped liquid electrolytes and the polymer matrix, which can restrict the free movement and the activity of the liquid electrolytes in the asprepared CPEs (Wang et al., 2018a). The results can be ulteriorly reinforced by adding the as-synthesized Li0.1Ca0.9TiO<sup>3</sup> particles into the polymer matrix because of their intrinsic chemical inertness and unique spherical-like and honeycomb structure. Based on the analysis mentioned above, it is creditable to conclude that the CPEs doped with the as-prepared spherical-like and honeycomb structural Li0.1Ca0.9TiO<sup>3</sup> particles can provide high lithium ion transference number. Therefore, we design experiments to calculate the lithium ion transference number. The Dc polarization plots of the assembled Li/CPE-LCT/Li battery are demonstrated in **Figure 6** and the corresponding results calculated to Equation (4) are displayed in **Table 1**. It can be observed from **Table 1** that the lithium ion transference number increases from 0.1623 of the CPE-0, to 0.2764 and 0.4962 of the CPE-CT and CPE-LCT, respectively, which can perfectly verify the above suppositions. Those results can be mainly ascribed to the added spherical-like and honeycomb structural Li0.1Ca0.9TiO<sup>3</sup> particles, which can not only provide more passageways for ion transfer, but also release more lithium ions to increase the carrier concentration.

FIGURE 7 | Nyquist plots of the assembled Li/CPEs/Li simulated cells with different CPEs for various storage times at 30◦C (A, CPE-0; B, CPE-CT; C, CPE-LCT), where the insert shows the corresponding equivalent circuit.

Interface stability is a key parameter for the practical polymer electrolytes in the lithium ion battery because it can seriously exert an effect on the initial charge-discharge and rate performance of the as-assembled battery (Zhang M. Y. et al., 2017). Therefore, the EIS variations of the Li/CPEs/Li simulated cells with the CPEs for different storage times at 30◦C are investigated and the corresponding plots are displayed in **Figure 7**. It can be easily found from **Figure 7** that the shape of the curves of the assembled Li/CPEs/Li simulated cells with three kinds of polymer electrolytes show significant differences, in which the plots of the Li/CPE-CT/Li and Li/CPE-LCT/Li simulated batteries are composed of a semicircle at high and medium frequency range originated from the bulk resistance (R<sup>b</sup> ) of polymer electrolyte and a compressed arc at low frequency range derived from the interfacial resistance (Ri), while the one of the Li/CPE-0/Li simulated cell simply contains a semicircle in the whole tested frequency range. Those differences can be mainly ascribed to the different interfacial performance of the asassembled simulated cells with different CPEs. To further reveal the mechanism of the interfacial performance, the EIS plots are fitted by the corresponding equivalent circuit displayed in the insert for the each plot, in which we decompose interfacial resistance (Ri) into interfacial reaction resistance (Rct) and the resistance of charge transfer (R<sup>f</sup> ) in the electronic doublelayer for convenience (Xiao et al., 2012). Apparently, the fitting curves are well-matched with the experimental data for the different CPEs shown in **Figure 7**. The R<sup>b</sup> value in each plot keeps constant and the Rct value increases firstly and then reaches a stable value, in which the Rct values of the CPE-CT and CPE-LCT can stabilize at around 943 and 508 Ω after being monitored for 5 days at room temperature, while the Rct value of the CPE-0 keeps growing. The results suggest that the battery assembled with the CPEs may present excellent interfacial performance, especially with the CPE-LCT, which can be mainly ascribed to the added spherical-like and honeycomb structural Li0.1Ca0.9TiO<sup>3</sup> particles that can markedly improve the compatibility between the electrodes and the as-prepared CPEs by entrapping any impurities such as water and trace organic solvent to inhibit the destructive reaction on the electrodes (Xiao et al., 2016).

Initial charge-discharge curves of the assembled Li/CPE-LCT/LiCoO<sup>2</sup> cell at different current densities (0.1, 0.2, 0.5, 1.0, and 2.0 C) are shown in **Figure 8A**. There is no difference between the cells assembled with the as-prepared CPE membranes and the commercial polyolefin membrane in terms of the shape of the curves, which suggests that the chargedischarge mechanism of the battery cannot be affected when the as-fabricated CPEs are assembled into the coin cell. As found in **Figure 8A**, the discharge specific capacity slowly declines from 145.7 mAh g−<sup>1</sup> at 0.1 C, to 139.5 and 132.4 mAh g−<sup>1</sup> at 0.2 and 0.5 C, respectively, and maintains at 116.8 mAh g−<sup>1</sup> even at 2.0 C, which indicate that the battery assembled with the CPE-LCT presents excellent charge-discharge performance. **Figure 8B** demonstrates the cycle and coulombic efficiency curves of the

### REFERENCES


Li/CPE-LCT/LiCoO<sup>2</sup> coin cell at different current densities (0.1, 0.2, 0.5, 1.0, and 2.0 C). It can be obviously observed that the coulombic efficiency of the assembled battery is ∼100% except for the first two cycles, which can be mainly ascribed to the formation of steady solid electrolyte interphase films on the surface of the electrodes (Reddy et al., 2013). The discharge capacity gently fades with increasing the current density during the repetitive cycles, in which the battery can deliver about 125.7 mAh g−<sup>1</sup> discharge specific capacity at 1.0 C after 100 cycles with 86.32% capacity retention ratio. Moreover, the discharge specific capacity of the battery can rebound back to 136.4 mAh g <sup>−</sup><sup>1</sup> when the current density drops down to 0.1 C again after 190 cycles. Those results suggest that the Li/CPE-LCT/LiCoO<sup>2</sup> coin cell shows outstanding rate cycle performance with approaching 100% coulombic efficiency.

### CONCLUSIONS

The CPEs doped with the spherical-like and honeycomb structural Li0.1Ca0.9TiO<sup>3</sup> particles synthesized by spray drying combined with following calcination are successfully fabricated by the classical phase inversion processes. The investigation results show that the CPEs present excellent physicochemical properties, such as improved ionic conductivity and lithium ion transference number, enhanced interfacial stability, and mechanical strength and so on. The assembled Li/CPE-LCT/LiCoO<sup>2</sup> coin cell can also deliver high initial discharge specific capacity and steady rate cycle performance. Those excellent results indicate that the as-prepared CPEs modified with the spherical-like and honeycomb structural Li0.1Ca0.9TiO<sup>3</sup> particles can be developed as a new kind of practical polymer electrolyte.

### AUTHOR CONTRIBUTIONS

WX, PM, and JL contributed conception and design of the study. CM organized the database. XY and MT performed the statistical analysis. ZW wrote the first draft of the manuscript. XY, YZ, YJ, and MT wrote sections of the manuscript.

### FUNDING

This work was financially supported by the National Natural Science Foundation of China (No. 51404038, 51874046, and 51503020), the Educational Commission of Hubei Province Foundation of China (No. B2016040), and the Yangtze Youth Talents Fund (No. 2016cqr05).

fluoride-co-hexafluoropropylene) and the incorporation of TiO2- (2-hydroxyethyl methacrylate). J. Power Sources 273, 1127–1135. doi: 10.1016/j.jpowsour.2014.10.026

Costa, C. M., Silva, M. M., and Lanceros-Mendez, S. (2013). Battery separators based on vinylidene fluoride (VDF) polymers and copolymers for lithium ion battery applications. RSC Adv. 3, 11404–11417. doi: 10.1039/c3ra 40732b


high performance lithium-ion batteries. J. Mater. Chem. 22, 5560–5567. doi: 10.1039/c2jm15955d


Zhang, S., Cao, J., Shang, Y., Wang, L., He, X., Li, J., et al. (2015). Nanocomposite polymer membrane derived from nano TiO2-PMMA and glass fiber nonwoven: high thermal endurance and cycle stability in lithium ion battery applications. J. Mater. Chem. A 3, 17697–17703. doi: 10.1039/c5ta02781k

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Xiao, Wang, Miao, Mei, Zhang, Yan, Tian, Jiang and Liu. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Carbothermal Synthesis of Nitrogen-Doped Graphene Composites for Energy Conversion and Storage Devices

Hongwei Mi\*, Xiaodan Yang, Jun Hu, Qianling Zhang and Jianhong Liu\*

*College of Chemistry and Environmental Engineering, Shenzhen University, Shenzhen, China*

#### Edited by:

*Qiaobao Zhang, Xiamen University, China*

#### Reviewed by:

*Xianwen Wu, Jishou University, China Hong Chen, Stockholm University, Sweden Hong Guo, Yunnan University, China Ting-Feng Yi, Northeast University, China Chen Weihua, Zhengzhou University, China*

#### \*Correspondence:

*Hongwei Mi milia807@szu.edu.cn Jianhong Liu liujh@szu.edu.cn*

#### Specialty section:

*This article was submitted to Physical Chemistry and Chemical Physics, a section of the journal Frontiers in Chemistry*

> Received: *31 July 2018* Accepted: *01 October 2018* Published: *22 October 2018*

#### Citation:

*Mi H, Yang X, Hu J, Zhang Q and Liu J (2018) Carbothermal Synthesis of Nitrogen-Doped Graphene Composites for Energy Conversion and Storage Devices. Front. Chem. 6:501. doi: 10.3389/fchem.2018.00501*

Metal oxides and carbonaceous composites are both promising materials for electrochemical energy conversion and storage devices, such as secondary rechargeable batteries, fuel cells and electrochemical capacitors. In this study, Fe3O<sup>4</sup> nanoparticles wrapped in nitrogen-doped (N-doped) graphene nanosheets (Fe3O4@G) were fabricated by a facile one-step carbothermal reduction method derived from Fe2O<sup>3</sup> and liquid-polyacrylonitrile (LPAN). The unique two-dimensional structure of N-doped graphene nanosheets, can not only accommodate the volume changes during lithium intercalation/extraction processes and suppress the particles aggregation but also act as an electronically conductive matrix to improve the electrochemical performance of Fe3O<sup>4</sup> anode, especially the rate capability. What's more, by etching Fe3O4@G to remove the iron-based oxide template, porous N-doped graphene composites (NGCs) were prepared and presented abundant pore structure with high specific surface area, delivering a specific capacitance of 172 F·g <sup>−</sup><sup>1</sup> at 0.5 A·g −1 . In this way, Fe2O<sup>3</sup> was both template and activator to adjust the pore size of graphene. And the effect of specific surface area and pore size tuned by the Fe2O<sup>3</sup> activator were also revealed.

Keywords: liquid-polyacrylonitrile (LPAN), carbothermal reduction, template activated method, supercapacitors, lithium-ion battery

### INTRODUCTION

Due to increasing energy and environmental demands, the utilization of energy storage devices have become a pressing essential need in both civil and military applications (Dunn et al., 2011; Etacheri et al., 2011; Chu and Majumdar, 2012; Li et al., in press). As materials play a leading role in the research of energy storage devices, metal oxides are considered as promising materials for electrochemical energy storage and conversion devices, such as secondary rechargeable batteries(Chen et al., 2017; Cui et al., 2018; Yi et al., 2018; Zhao et al., 2018; Zheng et al., 2018), fuel cells and electrochemical capacitors (Jiang et al., 2012; Wang et al., 2012, 2016; Wu et al., 2012; Nithya and Arul, 2016). Among various metal oxides, Fe3O<sup>4</sup> is extensively studied as an alternative electrode material for LIBs, with advantages of low cost, natural abundance, high electronic conductivity and high capacity (926 mAh·g −1 ; Huang et al., 2017; Liu et al., 2017; Wang et al., 2018; Yan et al., 2018). However, its practical application is hindered, because of huge volume change during cycle processes which resulted in severe capacity losses as well as electrode pulverization (Zhu et al., 2011a; Wu et al., 2013).

In order to maintain the electrode integrity, some strategies including coating with carbonaceous materials (He C. et al., 2013) and fabricating nanostructure (Behera, 2011; Lim et al., 2012; Zeng et al., 2014) have been widely reported. Nevertheless, to realize these improvements, many in situ synthetic methods, such as sol-gel polymerization (Jung et al., 2013), solvothermal or hydrothermal method (Yuan et al., 2011; Zhu et al., 2011a), electrospinning (Wang et al., 2008) and chemical vapor deposition (Zhu et al., 2013) have been utilized, but they are short for large-scale application. What' more, carbonaceous materials especially graphene, which received worldwide attention owing to its outstanding properties, shows superior performances in high-performance lithium-ion batteries due to their good conductivity and large surface areas (Behera, 2011; Yan et al., 2014). In this regard, it is an effective approach for coating graphene on Fe3O<sup>4</sup> to improve conductivity and relieve the volume change during cycles at the same time.

Besides batteries, carbonaceous materials also draw great attention as the electrode of the electrochemical double layer capacitors (EDLC) (Zhang and Zhao, 2009). Generally, to obtain high-performance EDLC electrode materials, there are usually several factors to consider. First of all, the specific surface area can greatly determine the capacitance of carbon (Zhao et al., 2017). In this respect, the fabrication of hollow (Han et al., 2014; Xu et al., 2015) or porous structure is an effective way to obtain the carbonaceous materials with large specific surface area. For example, Zhao et al. (Zhao et al., 2017) reported a novel 3D hierarchical carbon-based nanocages prepared by in-situ Cu template method, which can relieve the inevitable π-π aggregation and restacking of graphene sheets. Apart from the specific surface area, pore distribution has a vital influence on the capacity. On the one hand, by increasing the proportion of micropore in the material, the specific capacity of the material increased significantly. KOH activation is a popular method to prepare microporous carbons to achieve higher capacitance (Zhu et al., 2011b; Zheng et al., 2015). On the other hand, when the micropore volume increases to a certain extent, the resistance of ions transport to the porous carbon channel increases, resulting in poor capacitance at high current density. Zheng et al. (2015) reported that mesopore could connect multiple micropores, and speed up the electrolyte ions transferring from the electrode surface to the materials, resulting in a greater extent microporous energy storage ability into full play. Porous carbon with abundant pore structure (micropore, mesopore, and macropore), perform excellent rate capacity. Nevertheless, porous carbon is usually prepared by a complex process with KOH activation and template method (Xing et al., 2009).

Herein, we developed a facile carbothermal reduction method to fabricate Fe3O4@N-doped graphene composites (Fe3O4@G) as anode for Li-ion battery. Derived from Fe2O<sup>3</sup> and liquid polyacrylonitrile (LPAN), the Fe3O4@G can not only present enhanced conductivity but also accommodate the volume expansion of Fe3O4. In addition, after etching by hydrochloric acid (HCl) to remove the metal oxides template, porous N-doped graphene composites (NGCs) were obtained. Furthermore, the controllable preparation of porous graphene materials by template activated method was established. This approach takes some advantages. Firstly, Fe2O<sup>3</sup> was not only raw material to transform to Fe3O4, but also a template and an activating agent to adjust the pore size of graphene in which we call it a template activated method. Secondly, as previously reported (Mi et al., 2014; Zhuo et al., 2014), the LPAN used in this paper is the reductant and graphene precursor, which shows two-dimensional structure of N-doped graphene nanosheets after a carbothermal process in flowing argon gas. Lastly, the carbothermal method is simple to operate, and is also an approach for large-scale production of composites for energy storage.

### EXPERIMENTAL

### Preparation of Fe3O4@G Composites for Li-ion Battery

Fe2O<sup>3</sup> (5 g, Shanghai Lingfeng Chemical Reagent Co., Ltd., China) and LPAN (2 g) were mixed and stirred in ethanol for 4 h. Then the mixture was preoxidated in air at 220◦C for 3 h and carbonized in an argon atmosphere at a series of temperature (500, 600, 700, 800, and 900◦C) respectively for 4 h to prepared Fe3O4@G and N-doped graphene nanosheets (G). The as-prepared samples were named as Fe3O4@G-X, which X represented the carbonization temperature. As comparison, pure LPAN was also preoxidated in air at 220◦C for 3 h and annealed in an argon atmosphere at 600◦C. The product was named as G.

### Preparation of NGCs for Supercapacitors

The as-prepared samples (Fe3O4@G-600, Fe3O4@G-700, Fe3O4@G-800 and Fe3O4@G-900) were treated with HCl solution (4 mol·L −1 ) for 48 h and repeatedly washed by deionized water. Finally, the products were dried in a vacuum oven at 90◦C for 3 h. The final product was referred to as NGC600, NGC700, NGC800, or NGC900, corresponding to the carbonization temperature of 600, 700, 800, or 900◦C. As comparison, pure LPAN was also cured in air at 220◦C for 3 h and carbonized in an argon atmosphere at 700◦C. The product was named as G700.

### MATERIALS CHARACTERIZATION

The morphology and structure of the samples were characterized by field emission scanning electron microscope (FESEM, JSM-7800F & TEAM Octane Plus, 15 kV) and a Tecnai G2 transmission electron microscope (TEM, FEI, USA). The crystalline structures were obtained by a D8 advance X-ray diffraction spectrometer (XRD, Bruker, Germany) using Cu Kα radiation. X-ray photoelectron spectroscopy (XPS) was carried out on the ESCAlab220iXL electron spectrometer from VG scientific using 300-W Al Kα radiation. Raman spectra were performed at the room temperature (inVia Reflex, Renishaw, UK). The specific surface area and pore size distributions of the samples were measured by BELSORP-MAX with N<sup>2</sup> as absorbate at 77 K. All the samples were degassed at 150◦C for 3 h before measurement. The specific surface area was obtained by the BET equation and the pore size distribution was estimated from the desorption branch of N<sup>2</sup> isotherms by the BJH method. The thermogravimetric analysis (TG-DTA, Netzsch, Germany) was used to calculate the mass fraction of graphene.

## ELECTROCHEMICAL MEASUREMENTS FOR LI-ION BATTERY

Mixtures, which consisted of 80 wt% active materials (Fe2O3, G, Fe3O4@Gs), 10 wt% carbon black (CB) and 10 wt% polyvinylidene fluoride (PVDF) dispersed in N-methyl pyrrolidinone (NMP), were pasted on copper foil. Then the coated Cu foil was dried at 100◦C for 12 h and then cut into pieces with a diameter (ϕ) of 14 mm. The loading amount of active material was ∼0.8 mg·cm−<sup>2</sup> . The electrochemical performances of the samples were tested using 2,032 coin-type cells, Celgard 2400 separator, 1 mol·L <sup>−</sup><sup>1</sup> LiPF6/EC:EMC:DMC (1:1:1 by volume) electrolyte, and Li-foil as the counter electrode in an Ar-filled glove box (MBRAUN, Germany) with oxygen and moisture contents of <0.1 ppm. Galvanostatic charge-discharge measurements were performed on a LAND-CT2001A battery test system (China) in a voltage range of 0.01–3.0 V (vs. Li+/Li) at various current densities. Cyclic voltammetry (CV) was evaluated at 0.1 mV·s <sup>−</sup><sup>1</sup> on a Solartron analytic 1470E cell test system in the range of 0.01–3.0 V. Electrochemical impedance spectroscopy (EIS) was conducted on a Solartron Impedance analyzer 1260A at an AC voltage of 10 mV amplitude from 100 kHz to 0.01 Hz.

### ELECTROCHEMICAL MEASUREMENTS FOR SUPERCAPACITORS

Electrochemical performances were estimated by symmetric electrode-type coin cells. The fabrication of working electrodes was described as follow: the mixture of active materials, carbon black additive and PTFE emulsion (with a mass ratio of 85:10:5) were added in ethanol solvent. After a full stirring, the as-prepared slurry was coated on the nickel foam (ϕ14 mm) and dried at 80◦C for 6 h in a vacuum oven. The loading amount of active material was 0.5–0.8 mg·cm−<sup>2</sup> . Finally, the working electrodes can be obtained by further pressing at a pressure of 8 MPa. Subsequently, two electrodes with similar loading mass were selected as electrodes, and separated by a cellulose membrane filled with 6 M KOH electrolyte. Cyclic voltammetry (CV), chronopotentiometry (CP), and electrochemical impedance spectroscopy (EIS) were performed by electrochemical workstation (CHI 670 C) at room temperature. Cycle performance was tested by the LAND CT2001A instrument. The EIS was conducted using a sinusoidal signal of 5 mV over the frequency range from 100 kHz to 0.01 Hz.

The corresponding specific capacitance is calculated by the following equation: (Xie et al., 2012; Wang et al., 2016).

$$\text{Cs} = \frac{2I\Delta t}{mU}$$

where I was the current, 1t was the discharge time, U was the potential range, and m was the average mass of the samples on both electrodes.

### RESULTS AND DISCUSSION

As shown in **Figure 1**, Fe3O4@G was fabricated by a facile onestep carbothermal reduction method. Firstly, Fe2O<sup>3</sup> powders were mixed with LPAN and stirred for 4 h in absolute ethyl alcohol solvent. Subsequently, the mixture was cured in air and carbonized for 4 h under argon flow. During carbonization,

LPAN transformed to graphene, and Fe2O<sup>3</sup> was reduced to Fe3O4. With the temperature increased, iron-based oxides (FeOx) further reacted with graphene, etching the graphene to increase the specific area as well as pore size. After treated with HCl solution, the FeO<sup>x</sup> was removed, and NGC could be obtained.

### CARBOTHERMAL REDUCTION METHOD TO PREPARE FE3O4@G

The TG analysis and XRD were operated to reveal the reaction changes with various temperatures. TG curves shown in **Figure S1A** were measured from 30 to 1,000◦C under the nitrogen flow. For the pure Fe2O<sup>3</sup> sample, the TG curve was almost horizontal, indicating that thermal decomposition reactions didn't take place inside the Fe2O<sup>3</sup> under 1,000◦C except for the release of a little water. In the TG curve of pure LPAN, thermal decomposition reaction inside LPAN took place all the time because of the inherent nature of the organics. Before estimating by TG, Fe2O3/LPAN precursor was heated at 220◦C in air for 3 h, to make the crosslinking reaction occurred inside the LPAN. There was a big mass break from 500 to 700◦C in the mixtures. And the weight of samples didn't change until the temperature reaches 750◦C, which indicated the reduction reactions inside the Fe2O3/LPAN precursors completed. As the Fe3O4@G was the intermediate of the reduction reactions, so the following carbothermal temperature will range from 500 to 700◦C. LPAN is the oligomer of chained acrylonitrile and converts to graphene after carbonization (Mi et al., 2014; Zhuo

FIGURE 2 | SEM images of (A) the initial state for purchased Fe2O3 particles, (B) Fe3O4@G-500, (C) Fe3O4@G-600, and (D) Fe3O4@G-700. (E–G) TEM images and (G inset) SAED patterns of Fe3O4@G-600.

et al., 2014). As shown in **Figure S1B**, the main diffraction peaks of the Fe3O4@G can be indexed to magnetite-based on their good agreement with JCPDS Card No. 88-0866, indicating the reduction reaction occurred from the Fe2O3/LPAN precursor (Zeng et al., 2014; Li et al., 2017). During carbonization, LPAN transforms into graphene, and Fe2O<sup>3</sup> is mainly reduced to Fe3O4. With the temperature increased, Fe3O<sup>4</sup> further reacted with carbon to form FeO and Fe, which we can also call activation (He X. et al., 2013).

According to the XRD results, when the temperature was below 600◦C, Fe2O<sup>3</sup> was mainly reduced to Fe3O4, which can be described as Equations (1,2). But if the temperature is 700◦C, the other reduced phases including FeO (41.8◦ ) (Equation 3) and Fe (44.8◦ ) (Equation 4) were produced and the content of Fe3O<sup>4</sup> will

FIGURE 3 | CV curves of (A) Fe2O<sup>3</sup> and (B) Fe3O4@G-600 electrodes from the first cycle to the third cycle at a scan rate of 0.1 mV·s −1 in the potential range of 0.01 and 3.0 V (vs. Li+/Li). (C) Cycle performances at 0.1 A·g <sup>−</sup><sup>1</sup> of the Fe2O3, G, Fe3O4@G-500, Fe3O4@G-600, and Fe3O4@G-700 electrodes. (D) The first to fifth galvanostatic charge-discharge profiles and (E) rate capabilities of Fe3O4@G-600 electrode. (F) Nyquist plots of the Fe2O3, G, Fe3O4@G-500, Fe3O4@G-600, and Fe3O4@G-700 electrodes.

decrease accordingly. The equations of carbothermal reaction and activated reaction were given as followed (He X. et al., 2013). As the FeO phase and Fe phase show relatively inactive to Li+, the specific capacity of the prepared composites will decay quickly with the increased content of these new phases. Consequently, the carbothermal temperature should not exceed 700◦C to avoid generating the inactive products. Herein, Fe3O4@G prepared at the temperature of 500, 600, and 700◦C are used as anode for Li-ion battery.

$$\text{2C} + \text{O}\_2 \longrightarrow \text{2CO} \tag{1}$$

$$\rm CO + 3Fe\_2O\_3 \longrightarrow 2Fe\_3O\_4 + CO\_2 \tag{2}$$

$$\text{Fe}\_3\text{O}\_4 + \text{CO} \longrightarrow \text{3FeO} + \text{CO}\_2 \tag{3}$$

$$\text{FeO} + \text{CO} \longrightarrow \text{Fe} + \text{CO}\_2 \tag{4}$$

SEM and TEM images further revealed the influence of temperature on morphology. As shown in **Figure 2**, the Fe3O<sup>4</sup> nanoparticles and graphene layers could be observed. As the carbonization temperature played an important impact on the structure of the Fe3O4@G, the particles of Fe3O4@G-500 and Fe3O4@G-700 aggregated together, while it was obvious that the diaphanous graphene nanosheets coated on particles well in Fe3O4@G-600 (**Figures 2A–F**). According to the TGA curves in the **Figure S2**, the carbon contents of Fe3O4@G-500,

Fe3O4@G600 and Fe3O4@G-700 were 23.29, 22.37, and 22.43%, respectively. High-resolution TEM and SAED images confirmed the crystallization of Fe3O<sup>4</sup> nanoparticles (Wang et al., 2013), and lattice fringes with a spacing of 0.47 nm can be seen from the HRTEM image, corresponding to the (111) planes of Fe3O<sup>4</sup> (**Figure 2G**; Li et al., 2017).

CV curves of pure Fe2O<sup>3</sup> and Fe3O4@G-600 were shown in **Figures 3A,B**. In **Figure 3A**, pure Fe2O<sup>3</sup> electrode exhibited a clear cathodic peak at about 0.55 V in the first curve, for the reduction of iron from Fe3<sup>+</sup> to Fe<sup>0</sup> . In the second and third cycles, the reduction peak at 0.55 V disappeared and the anodic peak at 1.75 V become weak, ascribed to an irreversible phase transformation in the initial cycle (Hassan et al., 2011; Du et al., 2012). As shown in **Figure 3B**, in the first cycle, the cathodic peak at 0.48 V was attributed to the reversible reduction of Fe3O<sup>4</sup> to Fe<sup>0</sup> and the irreversible side reactions including the formation of SEI and decomposition of the electrolyte, and the anodic peak at 1.78 V in the reverse anodic scan corresponded to the reversible oxidation of Fe to Fe3O<sup>4</sup> (He C. et al., 2013). However, in the second cycle, the cathodic peak became weak and moved to 0.94 V, which revealed the occurrence of some irreversible reactions and formation of SEI film. It was noted that the third CV curve almost overlapped with the second, exposed a stable reversibility of the composite electrode.

In order to compare the cycle performances of the Fe2O3, G and Fe3O4@Gs, the cycle performance at 0.1 A·g <sup>−</sup><sup>1</sup> was investigated as shown in **Figure 3C**. Compared with pure G (377.3 mAh·g −1 ) and Fe2O<sup>3</sup> (162.7 mAh·g −1 ), the Fe3O4@G-600 delivered a capacity of 453.6 mAh·g −1 after 50 cycles. Fe3O4@G-600 delivered a high specific capacity of 1,023 mAh·g −1 at the first discharge process with a reversible specific capacity of 726.8 mAh·g −1 , where the initial coulombic efficiency was around 71%. The relatively low initial coulombic efficiency resulted from the irreversible capacity loss for the formation of SEI and decomposition of the electrolyte (Wu et al., 2014).

**Figure 3D** showed the first to fifth galvanostatic chargedischarge profiles of Fe3O4@G-600 electrode at 0.1 A·g −1 . The first discharge voltage curve exhibited one plateau at 0.76 V corresponding to the transformation as described in Equations (5,6) (Wang et al., 2010; Jin et al., 2013). At the same time, charge voltage plateau at about 1.6 V in the initial cycle was attributed to the reversible reactions between Fe3O<sup>4</sup> and Li+, The electrochemical reversible reaction mechanism during the charge/discharge processes can be described as Equation (7) (Xia et al., 2013). In comparison, the bare Fe2O<sup>3</sup> electrode displayed poor electrochemical properties because of the poor conductivity and pulverization in lithium intercalation/extraction processes (**Figure S3**). The electrochemical reversible reaction mechanism of Fe2O<sup>3</sup> during the charge/discharge processes can be given by Equation (8) (Hassan et al., 2011).

	- Fe3O<sup>4</sup> + 8Li<sup>+</sup> + 8e<sup>−</sup> ←→ 3Fe<sup>0</sup> + 4Li2O (7)
	- Fe2O<sup>3</sup> + 6Li<sup>+</sup> + 6e<sup>−</sup> ←→ 2Fe<sup>0</sup> + 3Li2O (8)

Rate capabilities of the Fe3O4@G-600 electrodes were shown in **Figure 3E**, where the discharge specific capacity is 453.6, 436, 431.2, 424, 355.6 mAh·g −1 at 0.1, 0.5, 1, 2, and 5 A·g −1 after 50 cycles. In order to confirm the improving electrochemical performance, the EIS was estimated (**Figure 3F**). The typical Nyquist plots contain a high-frequency semicircle followed by a linear tail in the low-frequency region. And the semicircle corresponded to the charge-transfer resistance (Rct). It is obvious that the semicircle for Fe3O4@G-600 was smallest among Fe3O4@Gs, suggesting Fe3O4@G-600 presented the smallest charge transfer resistance (Rct) (Wu et al., 2017). EIS measurements were fitted and the parameter results were listed in **Table S1**. Surface morphology of the Fe3O4@G-600 electrode also revealed the remarkable cycle stability after 5 and 50 cycles (**Figure S4**).

According to the results above, Fe3O4@G-600 presented remarkable performance. Owing to the N-doped graphene coating, it can not only accommodate the volume change but also inhibit the aggregation of Fe3O<sup>4</sup> particles. Meanwhile, the unique two-dimensional graphene nanosheets can promote rapidly electron transport and maintain the structural integrity during the electrochemical lithium insertion/extraction reaction so as to enhance the rate capability of the prepared composite electrode. Comparison of the electrochemical performance of ferric oxide anodes for lithium-ion batteries reported recently was shown in **Table S3**.

### TEMPLATE ACTIVATED METHOD TO PREPARE NGC

According to the result of XRD in **Figure S2B**, activation occurs only when the temperature was beyond 600◦C. Although the FeO phase and Fe phase showed relatively inactive to Li<sup>+</sup> for Li-ion batteries, recently Hu's group reported (Zhao et al., 2017) that metal template is favor to forming carbon materials with high conductivity, compared with metal oxides template, and higher carbonization temperature is also favorable to forming higher conductivity carbonaceous materials. In this regard, Fe3O4@G-600, Fe3O4@G-700, Fe3O4@G-800, and Fe3O4@G-900 were used to prepare a series of porous N-doped graphene composites (NGCs) by etching the FeOx.

NGCs were estimated to explore the effect of pyrosis temperature on crystallinity and defect degree by X-ray diffraction and Raman spectrum. As shown in **Figure 4A**,



corresponded to the D and G bands of graphene,

(**Figure 4B**). The broad peak and

XRD pattern showed two peaks at 2θ = 26.5 and 44.4◦ of graphene, and the intensity of the peaks increase with a rise of carbonization temperature, which indicated a higher temperature was favorable for the higher crystallinity and graphitization. There was no other peak, indicating that ferric oxide templates were removed completely by HCl. The peaks at 1,360 and

respectively. There was a broad peak of 2D band of carbon

lower intensity of the 2D band indicated the existence of several graphene layers. The ID/I<sup>G</sup> values which represent the defect quantity were calculated to be 1.036, 0.923, 0.913, and 1.006, According to the IUPAC classification, the isotherm curves of NGCs belonged to the type IV. And at the P<sup>0</sup> = 0.4–0.9, a clear hysteresis loop of H4 type was presented, implying the presence of mesopore. However, there was no hysteresis loop on the curve of G700, indicating that there was no mesopore in the G700. The specific surface area was obtained by the Brunauer– Emmett–Teller (BET) equation and the pore size distribution was estimated from the desorption branch of N<sup>2</sup> isotherms by the Barrett–Joyner–Halenda (BJH) method (Zhao et al., 2015). And the calculation results were summarized in **Table 1**.

From **Figures 4C,D** and the calculation results in **Table 1**, the specific surface area of NGCs were much larger than G700, indicating that the addition of template agent can increase the specific surface area. Obviously, it easy to find that the BET specific surface areas firstly increased from 256.87 m<sup>2</sup> ·g −1 (NGC600) to 505.41 m<sup>2</sup> ·g −1 (NGC700), and then decreased

1,590 cm−<sup>1</sup>

at around 2,600–3,100 cm−<sup>1</sup>

to 372.02 m<sup>2</sup> ·g −1 (NGC800) and 220.69 m<sup>2</sup> ·g −1 (NGC900). NGC600 presented a small specific area, for Fe2O<sup>3</sup> mainly transform to Fe3O<sup>4</sup> at 600◦C (**Figure 4C**). The reaction between FeO<sup>x</sup> and carbon materials enhanced with the annealing temperature increasing (from 600 to 700◦C), thinning of the channel and resulting in the increase of surface area. Then the specific area of NGC800 and NGC900 decreased because the quantities of tunnels diminished resulting from the further activation of Fe2O3. Schematic diagram of the activation of porous carbon by FeO<sup>x</sup> was presented in **Figure 4E**.

The morphology and structure of NGC700 were estimated by SEM and TEM. As shown in **Figure 5A**, the surface of G700 was smooth without tunnels. On the contrary, NGC700 presented a cellular structure, consisting of abundant pores, which offered a fast path to ion transportation (**Figures 5B,C**). As shown in **Figures 5D–F**, the hollow structure could be observed as well as thin graphene layers on the edge. TEM in higher magnification images showed that NGC700 presented graphene lattice with an inter layer spacing of 0.36 nm.

X-ray photoelectron spectroscopy was conducted to further understand the composition of elements and chemical states of the surface (**Figures 5G–J**). **Figure 5G** presented three peaks of C (284.5 eV), N (399 eV), and O elements (531 eV) (Mi et al., 2014, 2016; Zhuo et al., 2014). The curves of C, N, and O elements were fitted to understand the existing state of C and N elements in composites. Further analysis of the C1s (**Figure 5H**) can reveal three peaks: sp2C-C (284.6 eV), N-sp2C (285.4 eV), and N-sp3C (288.4 eV). Most of the carbon atoms form π-π conjugated system, for the main peak at 284.6 eV corresponding to sp<sup>2</sup> carbon. N 1s (**Figure 5I**) revealed three peaks: pyridinic N (398.5 eV), pyrrolic N (399.6 eV), and graphitic N (400.5 eV; Zhao et al., 2015). Among these three forms, pyridinic N was viewed as the most suitable for facilitating the electronic conductivity and the charge transfer. The N-doped content was calculated to be 9.52% (**Table S2**). Ascribed to the O and N-doped in the carbon, the defects increased and the electrochemical activity improved. The polar C-N and C-O on the surface of carbon also increased the wettability of the composites and the contact area of the electrolyte (Tong et al., 2016; Wang et al., 2016). The functional groups on the carbon surface can improve the pseudocapacitance of the materials and the total specific capacitance (Sui et al., 2015; Zhao et al., 2017).

To understand the relationship between the structure and the EDLC capacitance performance of the NGCs, cyclic voltammetry (CV), chronopotentiometry (CP) and electrochemical impedance spectroscopy (EIS) were measured

in a two-electrode system by coin-type cells. In **Figures 6A,C**, the CV curve presented a rectangular-like shape at 100 mV·s −1 for all samples, which indicated an EDLC capacitive. Chronopotentiometry at the current density of 0.5 A·g −1 for samples was in **Figure 6B** and the specific capacitances of the samples were calculated to be 83, 173, 101, and 122 F·g −1 , respectively, which were in accordance with the results of the specific surface area above (**Table 1**). The further electrochemical performance of NGC700 was estimated in **Figures 6C,D**.

Chronopotentiometry at the current density from 0.5 to 10 A·g −1 for other NGCs were also estimated, and the calculation correlation of specific capacitances with scan rates for the composites were shown in **Figure 7A**. When the current density increases, it was more difficult for ions to diffusion and transfer in the micropore. But the mesopore and macropore were favorable for ion diffusion and transfer at high current density (Zhang and Zhao, 2009). Therefore, with the increase of current density, the specific capacity of NGCs all decreased. Ascribed to suitable mesopore distribution, NGC700 showed well capacitance retaining at high current density, of which the capacitance decreased from 173, 111, 106, 94, and 84 F·g −1 at the current density of 0.5, 1, 2, 5, and 10 A·g −1 , respectively. Generally, the capacitance of materials showed correlation with not only specific surface area but also the pore distribution, conductivity, wettability, etc. The improvement of specific capacitance from NGC600 to NGC700 could be ascribed to its larger specific surface area and conductivity as discussed above. The decrease in capacitance from NGC700 to NGC800 could blame for the same reason. It was worth noting that NGC900 presented higher specific capacitance but lower specific surface area than NGC800. Because high temperature was favor for composites forming high conductivity and more defects. It was consistent with the test results of Raman and BET. After 4,000 cycles, the specific capacitance of NGC700 kept at a stable level and delivered a capacity of 150 F·g −1 at the current density of 0.5 A·g −1 , indicating excellent cycle performance (**Figure 7B**).

Electrochemical impedance spectroscopy further verified our speculation above. As shown in the Nyquist plot, there was a partial semicircle at high frequency and a vertical straight at the low frequency in each plot. In the high-frequency area, the partial semicircle corresponding to electrode resistances were 2.202, 1.579, 0.730, and 0.350 (**Table 2**). It was not difficult to find a downward tendency of the electrode resistance as the carbonization temperature increases, which could be ascribed to that the higher temperature was favorable for the higher crystallinity of graphene, so as to improve the conductivity. At the low frequency, NGC700 showed a nearly tilted line, indicating the electrode possessed of the best ionic conductivity and capacitive performance, which was ascribed to its large specific surface area and suitable pore size distribution. Obviously, NGC700 shows

TABLE 2 | Kinetic parameters of NGC600, NGC 700, NGC 800, and NGC 900 electrodes.


much greater performance than others in the electrochemical characterization. This improvement could be ascribed to the proper pore size distribution of NGC700.

### CONCLUSIONS

In summary, the prepared Fe3O4@G from Fe2O3/LPAN precursors showed optimal cycle stability and rate capability among these synthetic conditions, which delivered a capacity of 355.6 mAh·g −1 at the high current density of 5 A·g −1 after 50 cycles. What's more, NGCs were prepared by removing the FeO<sup>x</sup> template form Fe3O4@G, which presented high specific surface area and abundant pore structure. Ascribed to the activated template FeOx, it was reduced with graphene during carbonization, resulting in not only an increase in mesoporous and micropores but an increase in pore size. As a result, NGC700 performed higher specific capacitance and electrochemical stability, delivering a specific capacitance of 172 F·g −1 at 0.5 A·g −1 current density after 4,000 cycles for supercapacitor. Prepared Fe3O4@G by one step carbothermal

### REFERENCES


reduction method and NGC by activated template method in one approach, should be very worthy for consideration. Significantly, the raw materials Fe2O<sup>3</sup> is abundant in the earth, so the low cost will guarantee the prospect of the products promising for the next generation LIB and other application in energy storage.

### AUTHOR CONTRIBUTIONS

HM and JL contributed conception and design of the study. HM, XY and JH did the experiments, organized the database and performed the statistical analysis. HM and XY wrote the first draft of manuscript. HM and QZ revised the manuscript. All authors contributed to manuscript revision, read and approved the submitted version.

### ACKNOWLEDGMENTS

This work was financially supported by the National Natural Science Foundation of China (Nos. 21601126, 21571131), the Natural Science Foundation of Guangdong (No. 2014A030311028), Shenzhen Science and Technology Project Program (Nos. JCYJ201708171000919133, JCYJ20170818092720054, XCL201110060), the Natural Science Foundation of SZU (Nos. 2017031).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2018.00501/full#supplementary-material


for enhanced adsorption and energy storage. Nat. Commun. 6:7221. doi: 10.1038/ncomms8221


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Mi, Yang, Hu, Zhang and Liu. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Nitrogen-Doped Multi-Scale Porous Carbon for High Voltage Aqueous Supercapacitors

Xichuan Liu1,2, Rui Mi <sup>2</sup> , Lei Yuan<sup>2</sup> , Fan Yang<sup>2</sup> , Zhibing Fu<sup>2</sup> , Chaoyang Wang<sup>2</sup> \* and Yongjian Tang1,2 \*

*<sup>1</sup> Shanghai EBIT Lab, Key Laboratory of Nuclear Physics and Ion-beam Application, Department of Nuclear Science and Technology, Institute of Modern Physics, Fudan University, Shanghai, China, <sup>2</sup> Science and Technology on Plasma Physics Laboratory, Research Centre of Laser Fusion, China Academy of Engineering Physics, Mianyang, China*

#### Edited by:

*Qiaobao Zhang, Xiamen University, China*

### Reviewed by:

*Chundong Wang, Huazhong University of Science and Technology, China Juchen Guo, University of California, Riverside, United States Shuge Dai, Zhengzhou University, China*

#### \*Correspondence:

*Chaoyang Wang wangchy807@caep.cn Yongjian Tang tangyongjian2000@sina.com*

#### Specialty section:

*This article was submitted to Nanoscience, a section of the journal Frontiers in Chemistry*

Received: *27 July 2018* Accepted: *20 September 2018* Published: *17 October 2018*

#### Citation:

*Liu X, Mi R, Yuan L, Yang F, Fu Z, Wang C and Tang Y (2018) Nitrogen-Doped Multi-Scale Porous Carbon for High Voltage Aqueous Supercapacitors. Front. Chem. 6:475. doi: 10.3389/fchem.2018.00475* Recently, "Water-in-salt" electrolyte has been reported to extend the working voltage of aqueous supercapacitor. However, this electrolyte needs the electrode materials possess some good features such as proper pore structure, high electron and ion conductivity. Herein, we fabricated the nitrogen-doped multi-scale porous carbon (NMC) by the simple enriching melamine-resorcinol-formaldehyde xerogels method with integrating triblock copolymer for micro-pores formation. All the results confirmed that our NMC is provided with a very high specific surface area (3,170 m<sup>2</sup> g −1 ) and its monoliths are composed of multi-scale porous structure. By employing the nanostructured NMC as electrode materials, we have investigated the capability for high-voltage aqueous supercapacitor applications. The superconcentrated "Water-in-salt" electrolyte expand stability operating potential window of aqueous symmetric supercapacitor up to 2.4 V with a high energy density of 33 Wh kg−<sup>1</sup> at power density of 0.3 kW kg−<sup>1</sup> . Our studies indicate that the NMC is potential materials for high performance over wider voltage range.

Keywords: supercapacitors, aqueous electrolyte, water in salt, porous carbon, high voltage, energy density

## INTRODUCTION

For the fast consumption of fossil fuels leads to global severe environmental issues and energy crisis, exploitation of new energy sources become an urgent issue for humanity. For decades, many works have been devoted to develop new technologies to use new energy sources from the ambient or renewable sources like wind, solar, tide, electromagnetic fields, mechanical movement and so on, and converted to electrical energy in an energy storage device like batteries (Zhao et al., 2017; Zhang et al., 2018a). However, due to the intermittent nature of these energy sources, batteries will be charged repeatedly which resulted in rapid decay of their cycle life. In this case, supercapacitors (SCs) with favorable features of long cycling stability, fast charging/discharging ability and high power density are generally more suitable than batteries (Jia et al., 2018; Liu et al., 2018b; Qu et al., 2018). Specifically, since the SCs have high specific power characteristic, it also have been widely used in a wide variety of applications such as portable electronics, electric or hybrid electric vehicles, aircraft and smart grids.

Nevertheless, the low energy densities of SCs restrict its widespread applications. According to the Equation (1), the energy density (E) is related to the capacitance (C) and operating voltage (V). For increasing the energy stored in SCs, previous works have been widely focused on the improvement of capacitance which takes advantage of various topical subjects like the selection, the

**191**

construction, and the modification of electrode materials (Zhong et al., 2015; Dai et al., 2017; Liu et al., 2018a). So far, few researches focus on the crucial factors correlating to the operating voltage, even though it is more efficient to increase the energy density and power density (P), according to the Equation (2) (where R is the internal resistance) by expanding the operating voltage, since energy density and power density are directly proportional to the square of voltage.

$$E = \mathbf{C}V^2/2\tag{1}$$

$$P = V^2 / 4\text{R} \tag{2}$$

The SCs are usually use three types electrolyte (Zhong et al., 2015): aqueous, organic and ionic liquid (Kühnel and Balducci, 2014). Using organic or ionic liquid electrolyte can efficiently expand the potential window, which because of the organic or ionic liquid electrolyte has a good electrochemical stability with higher decomposition voltage (2.5–4.5 V) than aqueous. However, a series of undesired features severely limit the wide application of organic electrolyte. For example, the SCs with organic and ionic liquid electrolytes often suffer from low capacitances and power densities due to their large-size ion and low ionic conductivity nature. In addition, the organic and ionic liquid electrolytes are not only noxious and flammable result in environmental and security issues, but also require rigorous manufacturing procedures. On the contrary, aqueous electrolytes get more attention due to its inherently safety, low-cost, and easy operation characters. For the smaller-size ion and faster ionic conductivity enable aqueous SCs with larger capacitances and higher power densities (Zao et al., 2016; Hwang et al., 2017; Zeng et al., 2018a). So, it is urgent to study aqueous SCs with both high energy and power density fulfilling the application of the SCs.

The most challenge to obtain high-voltage aqueous SCs is expanding the electrochemical stability window of water (1.23 V), previous studies have devoted to asymmetry structure or neutral aqueous electrolytes, and the highest potential window even beyond 2 V (Yang et al., 2017; Zuo et al., 2017; Fu et al., 2018). Very recently, Yu et al. (2017) summarized the new insight into the high voltage of aqueous SCs. Many crucial tactics of expanding the operating voltage have been studied. Specifically, Xu et al. (Suo et al., 2015) reported an intriguing breakthrough that a superconcentrated lithium bis(trifluoromethane sulfonyl)imide (LiTFSI) aqueous solution named "water-in-salt" electrolyte displays a obviously high electrochemical stability up to 3 V in lithium-ion batteries applications. Obviously, this "water-in-salt" electrolyte also can be used in high voltage aqueous SCs (Gambou-Bosca and Bélanger, 2016; Díez et al., 2017; Reber et al., 2017). For instance, Hasegawa, et al. (Zhao et al., 2016) fabricates symmetric SCs using 5 M LiTFSI aqueous solution achieved a maximum stable operating voltage of 2.4 V. Nevertheless, although using the LiTFSI aqueous solution expand the operating voltage of SCs, it is also suffered the sacrifice of capacitances, which result in the limited increasing of energy density. Therefore, it is needed to choose one proper electrode materials for matching the LiTFSI molecule to maximization the capacitances as well as expanding the operating voltage. In this case, multi-scale carbonaceous materials with high good thermal and chemical stability, good porous network, and satisfactory electrical conductivity have been widely studied, and it is very suitable for SCs. (Huang et al., 2012; Fang et al., 2013; Hasegawa et al., 2016). Moreover, N-doped carbon materials have many attractive functional properties. It also gives more active sites for electrochemical reactions in double layer capacitors (Geng et al., 2011; Ci et al., 2012; Zhong et al., 2013; Zhang et al., 2018b). Furthermore, for the excellent performance of Ndoped carbon, it has been studied in long-term performance in SCs (Wen et al., 2012; Zhu et al., 2016; Zeng et al., 2018b).

In this work, Nitrogen-doped multi-scale carbon (NMC) was fabricated by simple sol-gel reaction with additional CO<sup>2</sup> activation. The structure of this material is comprised of a multiscaled pore with nano-porous carbon in a network of micron-size percolated hollow-duct. In particular, the prepared NMC was utilized as electrode materials to explore high-voltage aqueous supercapacitors.

### EXPERIMENTAL

### Materials Synthesis

NMC were fabricated by sol-gel method from a solution containing resorcinol (R), formaldehyde (F) and melamine (M), followed by aging, solvent exchange, drying and pyrolysis. In a typical process, firstly, 1.6 g of triblock copolymer Pluronic F68 (PEO76-PPO29-PEO76) was dissolved in a small amount of deionized water and ethanol at 60◦C, followed by adding 7.7 g of melamine and 18 mL of formaldehyde solution (37 wt %), with vigorous stirring until melamine was completely dissolved. Then, 6.5 g of resorcinol and 8.7 mL of formaldehyde solution were added into the above mixture solution, and stirring until resorcinol was completely dissolved. After that, 6 mL of NaOH solution (0.02 M) was added into the above mixture using as catalyst. At last, amount of deinoized water was added to meet the volume at 50 mL. The mixture was stirred at 60◦C for 3 min. Hereafter, the mixture was placed in a sealed container and kept at 60◦C for 72 h to finish the gel process. The M-R-F hydrogels containing PEO76-PPO29-PEO<sup>76</sup> were immersed in a solution of trifluoroacetic acid and ethanol (3:97 in volume) at room temperature for 72 h. Afterwards, the residual solvent was substitute for ethanol for 6 times per 24 h to remove water. Subsequently, the hydrogels were dried at 60◦C and carbonized with argon gas flow rate of 100 mL min−<sup>1</sup> at 800◦C for 4 h. And this nitrogen-containing porous carbon without CO<sup>2</sup> activation was denoted as NC. In the CO<sup>2</sup> activation process, the NC was heated in a tubular furnace at 950◦Cfor 8 h under a stable CO<sup>2</sup> flow (150 mL min−<sup>1</sup> ) and then the NMC was obtained. For comparison, the commercialized active carbon YP-50 (AC) was purchased from Kuraray chemical co. (Japanese).

### Characterization of the Samples

The morphology of the NC and NMC was observed by scanning electron microscopy (SEM) and high-resolution transmission electron microscopy (HRTEM). The crystallographical information and phase of the samples were investigated by X-ray powder diffraction (XRD) and Raman spectroscopy. N<sup>2</sup> adsorption/desorption isotherms were tested by an AUTOSORB-IQ surface area analyzer (Quantachrome Instrument Corporation) at 77 K. The chemical composition of the NMC was conducted with X-ray photoelectron spectroscopy (XPS).

### Electrochemical Measurements

For electrochemical experiments, the working electrodes were fabricated with 80 wt% active materials, 10 wt% acetylene black and 10 wt% polytetrafluoroethylene (PTFE) in ethanol to form a mixture solution. And the mixture was pressed onto stainless steel network at 20 MPa with a diameter of 16 mm and a mass of around 2 mg. The electrode was tested in two- and three-electrode cells. The 20 mol kg−<sup>1</sup> (m) LiTFSI aqueous solution was used as electrolytes. Two-electrode capacitor was tested in a CR2032-type coin cell with comparable mass of active materials, while a piece of sulfonated polypropylene membrane was employed as separator. In a three electrode cell, a pair of electrodes was used as working and counter electrode respectively, and the saturated calomel electrode (SCE) was used as reference electrode. The galvanostatic charge/discharge tests (GCD) and cycling performance were tested at LAND instrument (CT2001, China). Cyclic voltammetry (CV) was tested in the same range by a CHI 760 electrochemical workstation. Electrochemical impedance spectroscopy (EIS) measurements were performed from 100 kHz to 10 mHz.

### RESULTS AND DISCUSSION

### Morphologies and Crystallographical Information

Studies on the R-F sol-gel reaction (Al-Muhtaseb and Ritter, 2003), it is known that the resorcinol is react with water to form hudroxymethyl derivatives (-CH2OH), and a condensation reaction of the hudroxymethyl derivatives with F to form methylene (-CH2-) and methylene ether (-CH2OCH2-) bridged compounds. At the same ambient aqueous condition, a sol-gel reaction between M and F were also observed (Raymundo-Pinero et al., 2002). In this work, we simultaneously put the M, R, F, and PEO76-PPO29-PEO<sup>76</sup> in one react system, condensation reaction happens among the hydroxymethyl groups between M, R, and F to form small M-R-F clusters, which act as nucleation sites and incessantly increase to form a larger colloids. In this procedure, PEO76-PPO29-PEO<sup>76</sup> micelles are coinstantaneous embedded within the growing colloids, which can stabilize the M–R–F three dimensional structure. **Figure 1A** shows the morphological and structural analysis of NC. It shows that the NC with a three dimensional structure comprises interconnected carbon spheres, and contain pores about one micron in size derived from the decomposition of M-F and PEO76-PPO29-PEO<sup>76</sup> (Xu et al., 2012). The NC shows a foam-like microstructure with many internal interconnected channels, and this result is well agreed with other reports (Goldmints et al., 1997; Gutiérrez et al., 2009; Zhou et al., 2013). After activation at 950◦C for 10 h by CO<sup>2</sup> (**Figure 1B**), the NMC also present a foam-like structure except that the carbon skeleton becomes smaller and the size of primary carbon spheres decreases (Lin and Ritter, 2000). This result is owing to CO<sup>2</sup> etching effect, so the activation process lead to carbon loss. In addition, one would expect that amount of nano-pores are produced in the carbon skeleton by the CO<sup>2</sup> etching effect, which could be further measured by N<sup>2</sup> adsorption-desorption test in detail.

It is known that the CO<sup>2</sup> activation can produce nanopores in carbon materials (Lin and Ritter, 2000; Chang et al., 2013) and cause different structure formation in nature, however, these features are hardly detected by SEM images (**Figure 1B**). So we further adopt HRTEM imaging, XRD patterns and Raman spectra to examine these changes of the prepared carbon materials. **Figures 1C,D** shows the HRTEM imaging of NC and NMC. It reveals the structure of NC and NMC is basically amorphous in nature. Specifically, some partly graphitized carbon structure can be observed from HRTEM imagines, and these structures have been reduced after CO<sup>2</sup> activation. This phenomenon is coinciding with the XRD test results shown in **Figure 1E**. It shows two broad characteristic diffraction peaks at 2θ of 23.5◦ (002) and 43.6◦ (101), respectively, which can be regarded as a partly graphitized carbon. On the other hand, XRD pattern of the activated NMC displays a broader peak than the NC due to the decrease of the partly graphitized carbon structure by the CO<sup>2</sup> etching. Moreover, Raman spectra of NC and NMC (**Figure 1F**) show two typically broad peaks of D band (around 1350 cm−<sup>1</sup> ) and G band (around 1600 cm−<sup>1</sup> ). Generally, The D band is related to disordered features of graphitic carbon, while the G band is the typical characteristics of graphitic crystallites (Ji and Zhang, 2009). This result is corresponding to previous reports (Ferrari, 2007; Kicinski et al., 2013; Yi et al., 2017). The intensity ratio of D band and G band (ID/IG) is the parameter usually used to analyze defective structures in carbonaceous materials. Calculated from the **Figure 1F**, the ID/I<sup>G</sup> value of NMC and NC is 0.848 and 0.803, respectively, the increased value of the NMC indicating that during the CO<sup>2</sup> activation, the overall structure comprises graphite grains with an increasing amount of structural defects formed. The result well coincide with the HRTEM and XRD results which reveal the existence of amorphous carbon structure of NMC and NC.

### Textural Properties

One expected that our resulted NMC is a multi-scaled carbon material including macropores (>50 nm), mesopores (2–50 nm), and micropores (<2 nm). The nitrogen adsorption/desorption isotherms can examine in detail these pore properties shown in **Figure 2A**. And the related pore size distributions profiles determined by the density functional theory (DFT) program are shown in **Figure 2B**. Corresponding pore properties are displayed in **Table 1**. Obviously, both NMC and AC have a higher absorbed amount than the NC, which reveals that the NMC and AC possess higher specific surface area and pore volume than NC (**Table 1**). In depth, all these isotherms of NC, NMC and AC possess adsorbed amount at low relative pressure which exhibit Type I characteristics behavior designating the existence of micropores. And the distinct increasing of adsorbed

amount for the isotherms can be found, which reveals the increasing amounts of micropores. Another adsorption occurred at middle relative pressure display Type IV characteristics demonstrating the existence of mesopores (Long et al., 2008; Wang et al., 2008). Although, their adsorption curves are mostly consistent with their desorption curves, but the differences also exists at mid-/high-relative pressure. These kinds of isotherms are Type IV curves and Type H4 hysteresis loops, suggesting the existence of slit-shaped pores. This result can be further confirmed by the pore size distributions displayed in **Figure 2B**. The additional CO<sup>2</sup> activation process bring abundant 2 nm-pores of the NMC, and it possess a high BET specific surface area of 3,170 m<sup>2</sup> g −1 and a total pore volume of 1.880 cm<sup>3</sup> g −1 shown in **Table 1**. In summary, NMC possessing high surface area and multi-scale porous structure was successfully prepared. More mesopores exist in NMC (**Figure 2B**) are beneficial to electrochemical performance in the high-voltage aqueous supercapacitors (Hasegawa et al., 2016).

### XPS Study

XPS was employed to evaluate the surface chemistry in NC and NMC. As shown in **Figure S1**, three elements (C, N, O) exist in both NC and NMC. The total nitrogen heteroatom doping content was 4.9 at.% in NC and decreased to 3.2 at.% in NMC. The decrease of nitrogen content is owing to its higher reaction activities than that of carbon during high temperature activation process (Liu et al., 2012). The highresolution spectrum of C1s and N1s are shown in **Figure 3**. The C1s spectrum of NC and NMC can be fitted to five peaks show in **Figures 3A,C**, respectively. The peak located around 284.4, 285.3, 286.5, and 288, 289.3 eV are attribute to the C–C or C=C band (C-1), sp<sup>3</sup> -like defects (C-2), C–N or C–O species (C-3), C=O band (C-4), and π-π <sup>∗</sup> band (C-5), respectively

TABLE 1 | Corresponding pore parameters of the NC, AC, and NMC.


(Hernández-Fernández et al., 2010; Lim et al., 2012). The analysis results of C1s spectrum could verify the amorphous structure of the NC and NMC Analysis high-resolution spectrum of N1s spectra (**Figures 3B,D**), four peaks around 398, 400.5, 401.6, and 402.8 eV, which can be assigned to pyridinic-N (N-6), pyrrolic-N (N-5), quaternary-N (N-Q), and pyridine-N-oxide (N-X), respectively (Raymundo-Pinero et al., 2002; Long et al., 2008; Braghiroli et al., 2012; Horikawa et al., 2012). **Figure 4** shows the different types of nitrogen atoms in a carbon matrix. And the corresponding contents of nitrogen in NC and NMC are shown in **Table 2**. Obviously, compared with the NMC and NC, the nitrogen chemical state is not change even after CO<sup>2</sup> activation process. These N-containing functional groups should make both NC and NMC more electrochemically active, indicating it has excellent capacitance properties (Geng et al., 2011; Lim et al., 2012). To sum up, this change of surface chemical compositions is good for electrochemical properties of NMC.

### Electrochemical Results

The multi-scale porous features of the electrode is expected to facilitate the rapid diffusion of electrolyte ion within the electrode (Qin et al., 2016). To investigate the kinetic stability of NMC in high voltage aqueous SCs, the electrochemical performances were evaluated by using two- and three-electrode cells in 20 m LiTFSI aqueous solution. **Figure 5A** displays the typical CV curves of the NMC symmetric SCs at the same scan rate of 5 mV s−<sup>1</sup> in a different operating voltage window from 1 to 2.4 V with stepwise shifting voltage of 0.2 V. It is obvious that LiTFSI-based aqueous electrolytes possess a high stability comparing with organic electrolyte (Zhao et al., 2016), revealing the high adaptability in high voltage cell. Furthermore, the CV profiles of the NMC electrode are nearly rectangle in shape and no obvious redox peaks are detected, which is a very common feature of electrochemical double layer (EDL) capacitor (the schematic of symmetric two-electrode configuration shown in **Figure S3**). Moreover, it is well known that the charge which is stored within the capacitor may be determined by integrating the CV. The increased area under the curve with increasing scan potential range is clearly observed, indicating an increasing capacitance and high voltage capability of the electrode. This result is consistent with the GCD measurements shown in **Figure 5B.** The GCD curves of the NMC performed at current density of 0.1 A g−<sup>1</sup> in a different operating voltage window from 1 to 2.4 V with stepwise shifting voltage of 0.2 V. It can be observed all the GCD curves display a similar linear change of the voltage. With the increasing charge/discharge voltage, the nearly symmetric relationship between the potential vs. time was also observed, suggesting the desired fast charging, and discharging property of the NMC. Slight internal resistance (IR) drops of NMC electrode are observed for any of the curves, which indicate high conductivity of our electrode materials. Calculated from the GCD curves (**Figure 5B**), the single electrode specific capacitance (Cs) increase from 120 to 160 F g−<sup>1</sup> while the operating voltage increase from 1 to 2.4 V. With the working voltage increased, more sufficient surface area can be used to form EDL in the NMC electrode, which mainly attributes to higher Cs. When the symmetric capacitor operate at a high voltage of 2.4 V in 20 m LiTFSI electrolyte, the real potential of the positive and negative electrodes was separately determined by incorporating a SCE reference electrode (**Figure 5C**). The long term cycling stability applying high voltage of 2.4 V is the critical factor to appraise the practical applications of electrodes. In order to explore this, the cycle stability of NMC was further investigated by repeating the GCD test at a current density of 1 A g−<sup>1</sup> for 10,000 cycles in 20 m LiTFSI electrolyte as shown in **Figure 5D**. It was observed that the initial capacitance of NMC electrode is ∼150 F g−<sup>1</sup> , and gradually decrease to 120 F g−<sup>1</sup> during the first 1,000 cycles, but nearly no obvious capacitance decrease during the next thousands cycles. Impressively, even after 10,000 continuous charge/discharge cycles, the NMC electrode retains

about 80% of the initial capacitance and exhibits excel lent cycle stability. In contrast, the specific capacitance of AC electrode decreases rapidly from 118 to 86 F g−<sup>1</sup> during the first 300 cycles, and then gradually decreases to 72 F g−<sup>1</sup> (only 61% of the initial capacitance is retained) after 10,000 cycles. **Figure 5E** shows the CV profiles of NMC symmetric SCs at a scan rate of 5, 10, 20, 50, and 100 mV s−<sup>1</sup> , respectively. The rectangularshape of CV profile is moderately distorted with the increasing scanning rate, which is attributed to the difficult diffusion of ions from electrolyte to the porous structure at high scan rate. In addition, the capacitance decrease with the increasing scan rate, which is consistent with the GCD tests at different current density (**Figure 5F**). The calculated C<sup>s</sup> of the NMC in 2.4 V is 167, 160, 155, 146, 124, and 112 F g−<sup>1</sup> at current densities of 0.1, 0.2, 0.5, 1, 2, and 5 A g−<sup>1</sup> , respectively, demonstrating that the specific capacitance decrease with increasing current density. Furthermore, the retain of capacitance is about 74% when the current density increase from 0.1 to 5 A g−<sup>1</sup> in 2.4 V.

To further discuss the kinetic stability of the NMC under high voltage. The CV and GCD tests for NMC, AC and AC from 1 to 2.4 V were investigated respectively, as shown in **Figure 6**

and **Figure S2**. The observed integrated area increases with NC < AC < NMC, indicating that the proper multi-scale pore of NMC is beneficial for the specific capacitance. The Cs of NC, AC, and NMC is 58, 94, and 120 F g−<sup>1</sup> at the operating potential of 1 V, respectively. Whereas, when the operating potential reaches up to 2.4 V, the Cs increases to 73, 114, and 160 F g −1 , respectively. The increased Cs is due to the achievement of optimal synergistic effects of multi-scale porous structure and proper nitrogen contents (Zhang et al., 2016). The results obtained here are also in consistent with the SEM morphologies and N<sup>2</sup> adsorption-desorption test result, suggesting the NM C morphology provides good channels fascinating fast charge


intercalation/deintercalation process. As shown in **Figure 6C**, Nyquist plots for all these three samples consist of a semicircle at high frequency value followed by a slant at low frequency value. The semicircle is ascribed to the charges transfer processes between the electrode and the electrolyte. It is clear that the NC and NMC have the smaller radius of semicircle meaning the smaller charge transfer impedance, which because of the presence of nitrogen provides good electronic conductivity for NC and NMC (Liu et al., 2015). In addition, the slant form Warburg impedance indicates the electrolyte ion diffusion into the electrode. And the AC electrode shows a higher Warburg angel meaning lower ion diffusion which is attributed to the AC have a large number of micropores but lack of microchannels for the fast ion diffusion (Díez et al., 2017). **Figure 6D**

FIGURE 5 | (A) CV profiles (5 mV s−<sup>1</sup> ) and (B) GCD curves (0.1 A g−<sup>1</sup> ) of the NMC performed in a symmetric SCs with stepwise shifting of the maximum voltage of 0.2 V. (C) Potential changes of the positive (E+) and negative (E–) NMC electrodes. (D) Cycle performance of the NMC and AC symmetric SCs at 1 A g−<sup>1</sup> with a maximum voltage of 2.4 V. (E) CV profiles of the NMC SCs at different scan rates. (F) GCD curve and capacitance of NMC SCs with different current density. (All above results are tested in 20 m LiTFSI).

shows the relationship between energy density and power density of the different electrode materials performed in symmetric capacitor. The specific energy and power were derived from GCD tests at voltage of 2.4 V evaluated using the Equations (1) and (2). It is clear that the symmetric capacitor based on NMC electrodes materials delivered the highest overall energy density of 33 Wh kg−<sup>1</sup> at 0.3 kW kg−<sup>1</sup> . This result is higher than that of previous homologous works (Gambou-Bosca and Bélanger, 2016; Hasegawa et al., 2016; Díez et al., 2017). Moreover, it is worth noting that the NMC can provide 18 Wh kg−<sup>1</sup> of specific energy when the power density reaches up to 12 kW kg−<sup>1</sup> . Compared with previous literatures (Hasegawa et al., 2016; Díez et al., 2017; Reber et al., 2017), our works exhibit better performance (**Figure 6D**) in both energy density and power density.

### CONCLUSIONS

N-doped multi-scale porous carbon has been synthesized by a simple sol-gel method and successfully applied in high voltage aqueous electrolyte up to 2.4 V. Morphological and textural characterizations show that the NMC exhibit three dimensional porous channels with a high specific surface area of 3,170 m<sup>2</sup> g −1 , large pore volume of 1.880 cm<sup>3</sup>

g −1 , multi-scale pores and suitable contents of nitrogen functional groups. Electrochemical performances of NMC was observed for a symmetric electrodes system offering good capacitive performance, possessing energy density of 33 Wh kg−<sup>1</sup> at 0.3 kW kg−<sup>1</sup> in 20 m LiTFSI electrolyte and stable cycle performance (about 73%) over 10,000 charge-discharge cycles in high voltage of 2.4 V. The high performance characteristics of NMC contribute to the synergistic effect benefiting ion diffusion, transport and adsorption, and charge accumulation. This work demonstrated this nitrogen-doped multi-scale porous carbon is a promising electrode material for high-voltage aqueous electrolyte applications.

### AUTHOR CONTRIBUTIONS

YT developed the concept. XL and CW designed the experiments. XL, RM, and LY conducted the preparation of materials. XL, FY, and RM built the cells and carried out the performance characterizations. ZF and CW supervised the research. XL and FY co-wrote the manuscript. All authors discussed the results and commented on the manuscript.

### FUNDING

This study was financially supported by the Research Program of Sichuan province (Grant No. 2016GZ0235).

### ACKNOWLEDGMENTS

We are grateful to Xi Yang, Minglong Zhong, Qi Yang, Yong Zeng, and Jia Li (Science and Technology on plasma physics

### REFERENCES


Laboratory, Research Centre of Laser Fusion, China Academy of Engineering physics, Mianyang, China) for their valuable discussions and assistance in measurements.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2018.00475/full#supplementary-material


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Liu, Mi, Yuan, Yang, Fu, Wang and Tang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Sintering Temperature Induced Evolution of Microstructures and Enhanced Electrochemical Performances: Sol-Gel Derived LiFe(MoO4)<sup>2</sup> Microcrystals as a Promising Anode Material for Lithium-Ion Batteries

#### Edited by:

*Qiaobao Zhang, Xiamen University, China*

#### Reviewed by:

*Xianwen Wu, Jishou University, China Huan Pang, Yangzhou University, China Hongen Wang, Wuhan University of Technology, China*

#### \*Correspondence:

*Yan Zhao zhaoyan@sicnu.edu.cn Daojiang Gao daojianggao@sicnu.edu.cn; daojianggao@126.com*

#### Specialty section:

*This article was submitted to Physical Chemistry and Chemical Physics, a section of the journal Frontiers in Chemistry*

Received: *30 August 2018* Accepted: *27 September 2018* Published: *16 October 2018*

### Citation:

*Wang L, He Y, Mu Y, Liu M, Chen Y, Zhao Y, Lai X, Bi J and Gao D (2018) Sintering Temperature Induced Evolution of Microstructures and Enhanced Electrochemical Performances: Sol-Gel Derived LiFe(MoO4)*2 *Microcrystals as a Promising Anode Material for Lithium-Ion Batteries. Front. Chem. 6:492. doi: 10.3389/fchem.2018.00492*

Li Wang<sup>1</sup> , Yuanchuan He<sup>1</sup> , Yanlin Mu<sup>1</sup> , Mengjiao Liu<sup>1</sup> , Yuanfu Chen<sup>2</sup> , Yan Zhao1,2 \*, Xin Lai <sup>1</sup> , Jian Bi <sup>1</sup> and Daojiang Gao<sup>1</sup> \*

*<sup>1</sup> College of Chemistry and Materials Science, Sichuan Normal University, Chengdu, China, <sup>2</sup> School of Electronic Science and Engineering, University of Electronic Science and Technology of China, Chengdu, China*

A facile sol-gel process was used for synthesis of LiFe(MoO4)<sup>2</sup> microcrystals. The effects of sintering temperature on the microstructures and electrochemical performances of the as-synthesized samples were systematically investigated through XRD, SEM and electrochemical performance characterization. When sintered at 650◦C, the obtained LiFe(MoO4)<sup>2</sup> microcrystals show regular shape and uniform size distribution with mean size of 1–2µm. At the lower temperature (600◦C), the obtained LiFe(MoO4)<sup>2</sup> microcrystals possess relative inferior crystallinity, irregular morphology and vague grain boundary. At the higher temperatures (680 and 700◦C), the obtained LiFe(MoO4)<sup>2</sup> microcrystals are larger and thicker particles. The electrochemical results demonstrate that the optimized LiFe(MoO4)<sup>2</sup> microcrystals (650◦C) can deliver a high discharge specific capacity of 925 mAh g−<sup>1</sup> even at a current rate of 1 C (1,050 mA g−<sup>1</sup> ) after 500 cycles. Our work can provide a good guidance for the controllable synthesis of other transition metal NASICON-type electrode materials.

Keywords: LiFe(MoO4 )2 microcrystals, anode material, sol-gel process, sintering temperature, electrochemical performance

## INTRODUCTION

With the increase of environmental pollution and the rapid depletion of fossil fuels, a significant worldwide interest has been driven into the exploitation of clean and renewable energy devices (Gu et al., 2015; Mohanty et al., 2017; Mu et al., 2017; Wang T. Y. et al., 2017; Luo et al., 2018). As one of the most promising energy storage devices, lithium-ion batteries (LIBs) have been widely applied in many fields, such as portable electronic devices and electronic vehicles (Wu et al., 2016; Cai et al., 2017; Zhang Q. B. et al., 2018), which are attributed to their excellent features including high energy density, long lifespan, no memory effect and environmental benignity (Hassoun et al., 2014; Jiang et al., 2018; Zheng Z. M. et al., 2018). Nowadays, graphite is definitely as the common used anode Wang et al. Sintering Temperature Induced LiFe(MoO4)2 Anode

material in the commercial LIBs. However, its relatively low theoretical capacity of 372 mAh g−<sup>1</sup> falls short to meet the evergrowing requirement and hinders its further application (Wang et al., 2010, 2015; Zhao and Byon, 2013; Xiong et al., 2015; Hu et al., 2016; Wang H. E. et al., 2017; Li et al., 2018). Therefore, it is urgent to search alternative anode materials with high capacity and good cycling stability.

During the past decades, numerous efforts have been devoted to develop novel anode materials, which can be divided into two types: alloy anodes (Si, Ge and Sn) and conversion anodes (transition metal oxides, transition metal sulfides, transition metal phosphides and transition metal nitride and so on) according to their lithium storage mechanism (Lu et al., 2018). Among the above-mentioned anode materials, molybdenumcontaining metal oxides have given rise to considerable attention due to multiple oxidation states, high capacity and high energy density (Sharma et al., 2004; Tao et al., 2011; Zhang L. et al., 2017).

Generally speaking, transition metal molybdates can be classified into single metal molybdates and binary metal molybdates (Zhang L. et al., 2017). Single metal molybdates, such as CoMoO<sup>4</sup> (Yang et al., 2015), MnMoO<sup>4</sup> (Guan et al., 2016), and NiMoO<sup>4</sup> (Park et al., 2018) etc., can deliver the capacity of ∼1,000 mAh g−<sup>1</sup> when adopted as anode material. Therefore, these molybdates have been received extensive attention lately. Whereas, for the binary metal molybdates, there are relative seldom relevant available reports although they have multiple oxidation states and can offer much higher capacity. As a kind of typical binary metal molybdates, LiFe(MoO4)2, which belongs to a novel NASICON-type material with a structure of triclinic symmetry (space P-1), is constituted of the deformational LiO<sup>6</sup> octahedron and separated FeO<sup>6</sup> octahedron, and these octahedrons are connected by the MoO<sup>4</sup> tetrahedron (Devi and Varadaraju, 2012; Chen et al., 2014). During the discharge process, 15 mol electrons can be transfered along with the reduction of Fe3<sup>+</sup> to Fe<sup>0</sup> and Mo6<sup>+</sup> to Mo<sup>0</sup> , leading to the theoretical capacity as high to 1,050 mAh g−<sup>1</sup> . That is to say, LiFe(MoO4)<sup>2</sup> is a very promising anode material, which may be beneficial to largescale energy storage applications in the future. However, how to accurately control the microstructures is the key role in the synthesis of LiFe(MoO4)<sup>2</sup> electrode material, which restricts its further development. Up to date, the synthesis of LiFe(MoO4)<sup>2</sup> via solid state method (Chen et al., 2014) and sol-gel method (Devi and Varadaraju, 2012) have been reported, and the lithium storage mechanism for LiFe(MoO4)<sup>2</sup> anode is also expatiated. Unfortunately, the controllable of the microstructures is not deeply discussed. It is well-known that the sol-gel processing conditions have a significant influence on the microstructures, especially for the sintering temperature. Particularly, the sintering temperature plays an important role and has remarkable influences on the microstructures (including the crystallinity, morphology and grain size) and properties of material (Bahiraei et al., 2014; Xia et al., 2015; Dubey et al., 2017). Hence, it can be concluded that the electrode materials obtained at different sintering temperatures have various microstructures and electrochemical performances. As far as we know, there is no available report about the influences of sintering temperature on the microstructures and electrochemical performances of LiFe(MoO4)<sup>2</sup> electrode materials up to now.

Based on the urgent and neglected aspects in the controllable synthesis for the NASICON-type binary metal molydbates, the present work is aimed at systematical investigation the influences of sintering temperature on the microstructures for sol-gel derived LiFe(MoO4)<sup>2</sup> microcrystals. The obtained LiFe(MoO4)<sup>2</sup> microcrystals at different temperatures have been carried out a series of electrochemical performances tests, and the LiFe(MoO4)<sup>2</sup> microcrystals with superior electrochemical performance can be obtained just through precisely controlling the sintering temperature. Our work can provide a good guidance for the precise synthesis of other transition metal NASICON-type electrode materials.

### EXPERIMENTAL

### Synthesis

All the reagents were of analytical grade and used without further purification. LiFe(MoO4)<sup>2</sup> microcrystals were synthesized via a facile sol-gel method. A typical synthesis of LiFe(MoO4)<sup>2</sup> microcrystals is depicted as follows: 1.7273 g of MoO<sup>3</sup> and 0.3960 g of CH3COOLi were dissolved in 20 mL dilute NH3·H2O aqueous to form solution A, whereas 2.4240 g of Fe(NO3)3·9H2O was dissolved in 20 mL dilute HNO<sup>3</sup> solution to form solution B, then the solution A and solution B were mixed together. Subsequently, 2.5217 g of citric acid was added into the above mixed solution under continuous stirring at 80◦C to form gel. The obtained gel was further dried at 120◦C in a vacuum oven over night. This dried gel was pre-sintering at 300◦C for 3 h, and then the yellow precursors were calcined at 600, 650, 680, and 700◦C for 6 h to obtain the final products, respectively. The obtained samples were denoted as LFM-600, LFM-650, LFM-680, and LFM-700, respectively. The synthetic procedures for LiFe(MoO4)<sup>2</sup> microcrystals were illustrated in **Figure 1**.

### Characterization

The phase structure of the samples was characterized by X-ray diffraction (XRD, Rigaku miniflex) with Cu Kα radiation (λ = 0.15406 nm). The morphology was observed by scanning electron microscope (SEM, Quanta 250, FEI). X-ray photoelectron spectroscope (XPS) measurements were obtained on a Thermo Scientific Escalab 250Xi.

### Electrochemical Performance Measurements

The electrochemical performances of as-prepared products were evaluated using 2025-type coin cells, which were assembled in an argon-filled glove box with H2O and O<sup>2</sup> contents of <0.5 ppm. The working electrode was prepared by mixing of 70 wt% LiFe(MoO4)<sup>2</sup> powders, 20 wt% Super P Li (conducting additive) and 10 wt% polyvinylidene fluoride (PVDF, as binder) in the N-methyl pyrrolidone (NMP). The obtained slurry was coated onto copper foil and followed by drying at 100◦C for 12 h in a vacuum oven. Lithium foil served as the counter electrode and 1 mol L−<sup>1</sup> LiPF<sup>6</sup> dissolved in ethylene carbonate/dimethyl carbonate (EC:DMC = 1:1, V/V) was used as electrolyte. Galvanostatic charge/discharge tests were carried out on a battery test system (LAND, CT2001A, China) between 0.01 and 3.0 V. The cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS) measurements were performed on a CHI660E electrochemical workstation. The CV curves were investigated on a scanning rate of 0.1 mV s−<sup>1</sup> , while the EIS were measured at the frequency ranged from 0.01 Hz to 10 kHz.

## RESULTS AND DISCUSSION

**Figure 2** depicts the XRD patterns of LiFe(MoO4)<sup>2</sup> microcrystals at different sintering temperatures. As can be seen, all the diffraction peaks of the four samples match well with the standard card of LiFe(MoO4)<sup>2</sup> (JCPDS No. 72-0753), indicated that the pure phase products can be obtained at different sintering temperatures and the gained products belong to triclinic structure (space P-1). Careful observation, with the increase of sintering temperature, the relative intensity for (002) peak and (020) peak of the obtained LiFe(MoO4)<sup>2</sup> samples exhibit slight discrepancy. When the sintering temperature is 600◦C, the strongest diffraction peak of LFM-600 is (020) peak. While the sintering temperature increases to 650◦C, the (002) peak and the (020) peak of LFM-650 have almost the same intensity. With further increasing sintering temperature, the strongest diffraction peaks of LFM-680 and LFM-700 turn into (002) peak. These phenomena indicate that there is an obvious orientation growth in the final LiFe(MoO4)<sup>2</sup> microcrystals with the increasing sintering temperature, implying that they may possess various morphologies.

The morphology and size of LiFe(MoO4)<sup>2</sup> microcrystals under various sintering temperatures are shown in **Figure 3**. When the sintering temperature is 600◦C, the obtained products reveal the irregular morphology and vague grain boundary, indicating that the as-obtained products possess relative poor crystallinity. When the temperature increases to 650◦C, the obvious grain boundary can be observed, suggesting that the crystallinity of the products improved evidently. Careful observation, LFM-650 exhibits regular shape and uniform grain (mean size of 1–2µm). This structure may favor the effective penetration of electrolyte. When the temperatures further increase to 680 and 700◦C, the grain size of the products become larger and larger (mean size of 3–6µm). The SEM results verify that the four samples possess various orientation growth direction, indicating that the sintering temperature really has a significant influence on the microstructures of the obtained LiFe(MoO4)<sup>2</sup> microcrystals.

To better investigate the chemical composition and valance state of LiFe(MoO4)<sup>2</sup> microcrystals, LFM-650 is selected to perform XPS characterization, as presented in **Figure 4**. The survey spectrum (**Figure 4A**) shows the presence of Li, Fe, Mo, and O, along with thimbleful C from the reference electrode. It can be observed that the binding energy at 56.5 eV corresponds to the Li 1s, which is characteristic of Li<sup>+</sup> (**Figure 4B**; Dedryvere et al., 2010). The two peaks with binding energies at 712.1 and 725.3 eV can be ascribed to Fe 2p3/<sup>2</sup> and 2p1/2, representative of the existence of Fe3<sup>+</sup> (**Figure 4C**; Zhang Y. Y. et al., 2017). In the high-resolution spectrum of Mo 3d (**Figure 4D**), two peaks with binding energies at 232.7 and 235.9 eV correspond to Mo 3d5/<sup>2</sup> and Mo 3d3/2, which are assigned to the characteristic of Mo6<sup>+</sup> (Yao et al., 2014). The O 1s XPS spectrum can be divided into two peaks, the binding energies at 530.6 and 532.3 eV are attributed to the lattice oxygen and chemisorbed oxygen, respectively (Gieu et al., 2017). The XPS results manifest that the chemical composition of the obtained LiFe(MoO4)<sup>2</sup> microcrystals including Li+, Fe3+, Mo6+, and O2−, respectively.

**Figure 5** exhibits the charge-discharge curves of LFM-600, LFM-650, LFM-680, and LFM-700 between 0.01 and 3.0 V at a current rate of 1 C (1,050 mA g−<sup>1</sup> ) for the 1st, 2nd, and 500th cycles, respectively. As can be seen, the initial discharge specific capacity of LFM-650 reaches up to 1,923 mAh g−<sup>1</sup> , which is higher than those of LFM-600, LFM-680, and LFM-700, respectively. Although LFM-650 exhibits higher capacity, it also suffers from the low initial coulombic efficiency, which should be assigned to irreversible structural transformation of LiFe(MoO4)<sup>2</sup> microcrystals and the formation of the solid electrolyte interface (SEI) film. It is worth noting that the discharge specific capacity of LFM-650 still can achieve 925 mAh g−<sup>1</sup> after 500 cycles. Why the sample LFM-650 show such a high capacity? The answers can be gained by the XRD and SEM results. Comparison with the other samples, LFM-650 possesses optimal microstructure including good crystallinity, better uniformity and suitable grain size, which can enlarge the contact area between the active material and the electrolyte, bring much more active sites, profit the lithium ions and the electrons transportation, leading to superior electrochemical performance.

In order to evaluate the electrochemical performance of the four samples, various electrochemical measurements are carried

out. **Figure 6A** presents cyclic voltammetry profiles of LFM-600, LFM-650, LFM-680, and LFM-700 for the 1st cycle in the voltage range from 0.01 to 3.0 V at a scan rate of 0.1 mV s−<sup>1</sup> , respectively. It can be observed that the four samples possess similar curves, which means that the redox process of the four samples during the charge-discharge process is identical. Careful observation, in the cathodic scan curve, the reduction peaks located at about 2.6 and 1.7 V are ascribed to the reduction of Fe3<sup>+</sup> to Fe2<sup>+</sup> and Mo6<sup>+</sup> to Mo4+, respectively (Chen et al., 2014). The following reduction peaks at low voltage have a bit difference, but they also can be attributed to the reduction of Fe2<sup>+</sup> to Fe<sup>0</sup> , Mo4<sup>+</sup> to Mo<sup>0</sup> and the formation of SEI film (Zhang et al., 2015). The reason that the four samples exhibit these differences can be speculated as follows: (i) Since the LFM-600 owns inferior crystallinity, only two obvious reduction peaks can be detected. In fact, there is a weak broad reduction peak locating at about 0.2 V. (ii) As for LFM-650, it possesses optimal microstructure including good crystallinity and suitable grain size, hence three legible reduction peaks can be easily observed. (iii) Both LFM-680 and LFM-700 show the relative larger grain size (diameter and height), which is adverse to the insertion/extraction of lithium ions and prolong the transport pathway of lithium ions, resulting in merely two reduction peaks appearance. In the anodic scan curve, the broad peak around 1.75 V is ascribed to the oxidation of Mo<sup>0</sup> to Mo6<sup>+</sup> and Fe<sup>0</sup> to Fe3+. Interestingly, the above oxidation peak unexpectedly merges to a broad peak, hence the oxidation of Mo and Fe metals could not be distinguished. Moreover, the second and

third scan curves for LFM-600, LFM-650, LFM-680, and LFM-700 are displayed in **Figure S1**. Clearly, the two curves are a bit different from the first curve, which can be assigned to the irreversible destruction of the LiFe(MoO4)<sup>2</sup> structure and the formation of the SEI film (Gong et al., 2013; Wang et al., 2014). Further observation, the both curves of each sample exhibit good overlapping, indicating that the four samples have good reversibility, respectively.

**Figure 6B** displays the cycling performances of the four samples. Apparently, in the initial 40 cycles, the specific capacities begin to fade severely due to the formation of SEI film and the irreversible structural transformation (Xu et al., 2016; Wei et al., 2018). With the increase of cycling times, the specific capacities of the four samples gradually raise until to the stable value. These phenomena usually occur in transition metal oxide anode materials (Zhang et al., 2015; Guan et al., 2016), which can be attributed to the reversible growth of a polymeric gel-like film origination from kinetic activation process, and this can promote interfacial lithium storage. Therefore, the capacities gradually increase with extended cycling (Liu et al., 2018). It can be seen that LFM-650 can deliver a high discharge specific capacity of 925 mAh g−<sup>1</sup> and own higher retention rate (88%, calculated based on the theoretical capacity of 1,050 mAh g−<sup>1</sup> ) even at a current rate of 1 C after 500 cycles, which is far beyond LFM-600 (325 mAh g−<sup>1</sup> ), LFM-680 (343 mAh g−<sup>1</sup> ), and LFM-700 (337 mAh g−<sup>1</sup> ), respectively. The higher capacity for LFM-650 can be ascribed to the good crystallinity and suitable grain size, which is beneficial for lithium ions transmission and repeating insertion/extraction. Comparatively speaking, LFM-600 has poor crystallinity, resulting in unsustainable longtime charge-discharge. Meanwhile, LFM-680 and LFM-700 possess larger grain size and this microstructure is disadvantageous to the insertion/extraction of lithium ions.

**Figure 6C** illustrates the rate capabilities of the four samples. Obviously, the specific capacities of LiFe(MoO4)<sup>2</sup> microcrystals gradually fade with the increase of the current rate (from 0.2 to 5 C). Comparison with the four samples, the specific capacities of LFM-600, LFM-680, and LFM-700 are far below that of LFM-650 at different current rates. It can be observed clearly that the specific capacities of the above three samples fade severely, especially at high rates. Moreover, when the current rate comes back to 0.2 C, the specific capacities of LFM-650 can also reach 666 mAh g−<sup>1</sup> , which is much higher than those of other samples (LFM-600, LFM-680, and LFM-700), exhibiting superior rate capability. Fundamentally speaking, the distinction of the rate capability for LiFe(MoO4)<sup>2</sup> microcrystals also may be related to the crystallinity and particle size.

To further understand why LFM-650 exhibits such superior electrochemical performance, electrochemical impedance spectroscopy (EIS) after the first discharge-charge cycle is

equivalent circuit.

carried out, as shown in **Figure 6D**. It is clearly observed that all the Nyquist plots of the four samples contain two parts, including two semicircles and slope line. The two semicircles at high and medium frequency region stand for the resistance of charge transfer in the electrolyte/electrode surface (Rct) and the resistance for the formation of SEI film in the electrode surface (R<sup>f</sup> ), respectively. The slope line at low frequency reveals the Warburg impedance, which represents the diffusion of lithium ions in the bulk material (Sun et al., 2014; Li et al., 2015). The EIS data are fitted based on the equivalent circuit model (Wu et al., 2017), corresponding to the ohmic resistance (Rs), the SEI film resistance (R<sup>f</sup> ), dielectric relaxation capacitance (CPE<sup>f</sup> ), the charge transfer resistance (Rct), double-layer capacitance (CPEct) and Warburg impedance (W1), as presented in the inset of **Figure 6D**. The fitting results are tabulated in **Table 1**. It is obvious that the Rs , R<sup>f</sup> , and Rct for LFM-650 are much smaller compared with those of LFM-600, LFM-680, and LFM-700, demonstrating that LFM-650 possesses a more stable surface film and a faster charge transfer process, leading to the enhancement of lithium storage performance. More importantly, the notable increase in Rct for LFM-600, LFM-680, and LFM-700 indicated that these electrode materials own higher kinetic barrier for lithium

TABLE 1 | Fitted impedance parameters of LFM-600, LFM-650, LFM-680, and LFM-700.


ions insertion/extraction, thus resulting in poor electrochemical performance (Zheng J. M. et al., 2018), just like what is displayed in **Figures 6B,C**.

### CONCLUSION

In summary, LiFe(MoO4)<sup>2</sup> microcrystals with pure triclinic structure have been successfully synthesized by a simple sol-gel method. The influence of sintering temperature on the microstructures and electrochemical performances was investigated in detail. The sample LFM-650 exhibits enhanced cycling stability and rate capability contrasted to the samples LFM-600, LFM-680, and LFM-700, which can deliver a high discharge specific capacity of 925 mAh g−<sup>1</sup> even at a current rate of 1 C after 500 cycles. The superior lithium storage performance was attributed to the good crystallinity and uniformity as well as suitable grain size.

### AUTHOR CONTRIBUTIONS

LW designed and conducted the experiments. YH, YM, and ML helped the characterization and data analysis. LW wrote the paper. YC, YZ, XL, JB, and DG revised the paper.

### REFERENCES


### ACKNOWLEDGMENTS

This work was supported by the Scientific Research Fund of Sichuan Provincial Education Department of Sichuan province (Grant No. 17ZA0325), and Open Foundation of Key Laboratory of Sichuan Province Higher Education Systems (SWWT2016-3).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2018.00492/full#supplementary-material


as anodes for long-life lithium-ion batteries. ACS Appl. Mater. Interfaces 6, 20414–20422. doi: 10.1021/am505983m


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Wang, He, Mu, Liu, Chen, Zhao, Lai, Bi and Gao. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Combustion Characteristics of Physically Mixed 40 nm Aluminum/Copper Oxide Nanothermites Using Laser Ignition

Florin Saceleanu<sup>1</sup> , Mahmoud Idir <sup>2</sup> , Nabiha Chaumeix <sup>2</sup> and John Z. Wen<sup>1</sup> \*

<sup>1</sup> Mechanical and Mechatronics Engineering, University of Waterloo, Waterloo, ON, Canada, <sup>2</sup> Institut de Combustion, Aérothermique, Réactivité et Environnement, Centre National de la Recherche Scientifique, Orleans, France

This paper reports on the ignition and flame propagation characteristics of aluminum/copper oxide (Al/CuO) nanothermite at different packing density, manufactured from 40 nm commercial Al and CuO nanopowders. A 3.5 W continuous wave laser was used to ignite the samples in argon at atmospheric pressure, and a high speed camera captured the flame propagation. The high speed images revealed that the fast laser heating creates significant material ablation, followed by heat transfer along the heated surface. The bulk ignition occurs near the edge of the top surface, followed by the self-sustained burning. Lightly pressed powders (90% porosity) ignited in ∼0.1 ms and the burning front propagated at around 200 m/s, while the dense pellets (40–60% porosity) ignited in ∼1 ms and the burning front propagated at around 10 m/s. These results indicate that the reaction mechanism changes from mass convection to heat diffusion with increasing the packing density. The ignition and burn speeds of these Al/CuO nanothermites at different equivalence ratios (ERs), along with SEM images of pre- and post-combustion, illustrate that the homogeneity of the mixture is a critical parameter for optimizing the performance. The Al rich mixtures show significantly lower ignition delays and higher burn speeds.

Keywords: nanothermite reaction, nanoparticle, laser ignition, heterogeneous combustion, flame propagation

### INTRODUCTION

Nanothermites, or metastable intermolecular composites (MIC), are highly energetic solid state materials with tunable combustion properties. A typical example is the Aluminum/Copper Oxide (Al/CuO) nanothermite, which is a good candidate for applications that involve propellants and explosives due to its fast ignition and gas release characteristics. The overall reaction is given by

$$\text{--}2Al\_{(s)} + \text{3CuO}\_{(s)} \rightarrow \text{ Al}\_2\text{O}\_{3(l)} + \text{3Cu}\_{(l-\text{g})}$$

The heat release is 4,075 J/g and the adiabatic flame temperature is 2,843◦K (Fischer and Grubelich, 1998). Aluminum and copper oxide are mixed at the nano scale in order to enhance the reaction rates. Detailed description of the methods of preparation and characterization of the Al based nanothermites and their applications can be found in literatures (Rossi, 2015; Lafontaine and Comet, 2016).

#### Edited by:

Kaili Zhang, City University of Hong Kong, Hong Kong

#### Reviewed by:

Xiang Zhou, Nanjing University of Science and Technology, China Hailin Li, West Virginia University, United States

> \*Correspondence: John Z. Wen john.wen@uwaterloo.ca

#### Specialty section:

This article was submitted to Physical Chemistry and Chemical Physics, a section of the journal Frontiers in Chemistry

Received: 03 August 2018 Accepted: 14 September 2018 Published: 09 October 2018

#### Citation:

Saceleanu F, Idir M, Chaumeix N and Wen JZ (2018) Combustion Characteristics of Physically Mixed 40 nm Aluminum/Copper Oxide Nanothermites Using Laser Ignition. Front. Chem. 6:465. doi: 10.3389/fchem.2018.00465

The fast oxidation of aluminum in a thermite reaction can occur in (1) condensed phase, where the oxygen ion is transported across the Al/CuO interface, and (2) gas phase, where the CuO first dissociates, and then O<sup>2</sup> gas reacts with Al in a similar way that Al oxidizes/burns in an oxygen rich atmosphere. Once the Al/CuO is ignited, the combustion flame self-propagates via mass and heat transfer processes such as thermal conduction and mass convection. Constant volume cell experiments illustrated that the pressure signal leads the optical emission signal, since CuO can fully decompose at temperatures below the adiabatic flame temperature (Sullivan and Zachariah, 2010). Also, electrically heated nanothermite coatings showed that CuO decomposes prior to ignition but only at heating rates higher than 2,000◦K/s (Williams et al., 2013). Moreover, lightly packed Al/CuO powders was shown to have three modes of combustion with increasing the surrounding inert gas pressure, where the combustion velocity decreased from 1,000 to 2 m/s (Weismiller et al., 2009). Convection was shown to be dominant at low (near atmospheric) pressures and thermal conduction to be dominant at high pressures. The observations above suggest a gas–solid reaction mechanism. However, it was found that the burning rates of high density Al/CuO pellets are uniform with the nitrogen pressure (Egorshev et al., 2013). This observation suggests a condensed phase reaction mechanism. At the nano scale, in-situ TEM images of the Al/CuO nanothermite reaction revealed a new condensed state mechanism, noted as reactive sintering (Sullivan et al., 2012). In this mechanism, reaction initiates at the Al/CuO interface, and the heat produced melts the adjacent products. The process is sustained due to capillary forces that deliver unreacted material to the reaction site. It was shown that reactive sintering occurs on a timescale of 10 µs, which is much faster than the burning time of ∼1 ms. It was also shown that the nanostructure of the reactants is not preserved locally during the reaction. Similar TEM experiments showed that CuO releases oxygen faster than ignition (Egan et al., 2014). However, the condensed state reactions between Al and reduced CuO occurred on a timescale of 1 µs, which is much faster than the heterogeneous Al/oxygen reaction. Furthermore, SEM images illustrated large micron size spherical products after the Al/CuO reaction, and it was concluded that most of the reaction must have proceeded through a condensed phase mechanism (Jacob et al., 2015).

An exponential decrease of the flames speeds (from 100 to 1 m/s) with increasing the pellet density [20–70% of the theoretical maximum density (TMD)] was observed in the Al/MoO<sup>3</sup> nanothermites (Sanders et al., 2007; Pantoya et al., 2009). This phenomenon was explained by a change in the combustion mechanisms from thermal conduction in low porosity pellets to mass convection in high porosity pellets. At high bulk densities, numerical modeling of the flame speed of Al/MoO<sup>3</sup> based on conductive heat flow agrees with the experimental data (Kim, 2014). However, at low bulk densities, burn tube experiments in Asay et al. (2004) indicated that convection is the dominant transport mechanism, and the reduced particle size was elementary for enhancing convection (Prentice et al., 2006). Also, simple scaling arguments in Egan and Zachariah (2015) suggest that burning speeds on the order of 100 m/s can only be sustained by the transport of the condensed phases to the unreacted zone, since energy transfer via heat conduction and product gas convection/condensation is not enough to ignite the adjacent reactants.

The correlation between ignition and thermal analysis at low heating rates experiments illustrated four reaction steps in the Al/CuO combustion. It was suggested that the decomposition of CuO is the first step due to low activation energies, and that ignition is reached when 4% of the reaction enthalpy is released (Umbrajkar et al., 2006). Furthermore, ignitions of dense nanothermite on electrically heated filaments showed that ignition temperatures increase while pressurization rates decrease with the heating rate (Williams et al., 2013). The authors postulate that low temperature reactions prior to ignition alter the transport properties of the alumina layer and have important contribution on the ignition mechanism. In Zhou et al. (2017) and Yu et al. (2018), new methods to prepare core-shell Al/CuO thin film nanorods are described, and these nanothermites display several low-temperature reactions before the main exothermic reaction. These low temperature reactions are less resolved as the heating rate increases. Furthermore, the Al/CuO nanorods have the highest energy release at the Al rich conditions, and their onset temperature is independent of the stoichiometry. Other electrically heated filaments experiments showed that ignition delay is proportional to the oxide shell thickness, which supports the diffusion controlled mechanism of reaction (Chowdhury et al., 2010). The ignition temperature of Al/CuO was also correlated to the oxygen release temperature from CuO (Jian et al., 2013), but the activation energy for oxygen release decreased with the heating rate (Jian et al., 2014), which indicates that oxygen transport is limited.

The objective of this paper is to observe and analyze the ignition and burning speeds of Al/CuO nanothermites that are consolidated at different densities using physically mixed 40 nm nanopowders. The effect of nanoparticle ablation on the ignition delay and the effect of stoichiometry on the burning speeds are discussed. Similar experiments in literature focus mostly on the ignition delay and flame velocity. However, in this paper high speed and resolution images of the ignition and combustion processes illustrate specific macroscopic features of the Al/CuO mixtures prior and post ignition. A new method is proposed for calculating the speed of the burning front from the reacted zone to the unreacted zone. The heat and mass transfer processes that occur at the macroscopic scale determine the nanothermite performance both kinetically, in terms of the ignition delay and flame speeds, and thermodynamically, in terms of the combustion efficiency. It is found that the reactivity of the nanothermite mixture is strongly affected by the homogeneity of the reactants, and the fuel rich mixtures show reduced ignition delay, faster burning speeds, and higher pressurization rates compared to the stoichiometric mixtures.

### EXPERIMENTAL

### Materials and Setup

Aluminum and copper oxide nano powders with APS (Aerodynamic Particle Size) diameters of 40 nm were purchased from US Research Nanomaterials Inc. The powders were added to a glass vial according to the stoichiometric or fuel rich ratios [equivalence ratio (ER) of 1.5 and 2], assuming that the aluminum nanoparticles are 65% active (based on the oxidation limits in a thermogravimetric analyzer). Hexane (10 mL) was added to the Al/CuO mixture (1,000 mg) and ultra-sonicated for 20 min to reduce agglomeration. The suspension was dried in a fume hood overnight on an evaporating plate, and then placed on a heating plate for 30 min to remove adsorbed species. The reactant mixture was consolidated at various densities using a pellet cast (6 mm diameter) and a hydraulic press, or packed lightly in an acrylic tube (7 mm inner diameter and 15 mm length). The final sample mass was 200 ± 20 mg.

A diagram of the laser ignition and high speed imaging setup is shown in **Figure 1**. The pellet was held near the center of a cylindrical vessel (inner diameter of 100 mm), which was fitted with quartz windows on the ends. A continuous wave argon laser (3.5 W, 100 ms pulse duration) was used to ignite the pellet by heating its top surface, using a focusing lens to increase the power density from 225 W/cm<sup>2</sup> to 40 kW/cm<sup>2</sup> . The signal from a photodiode (1 ns response time) was used to trigger a Phantom high speed camera, which was set to record at 200,000 fps and 500 ns exposure time with extreme dynamic range (EDR). A piezoelectric pressure transducer (1 µs response time) was installed on the vessel wall. Two oscilloscopes were used in order to capture high time resolution signals, and the full duration signals at lower time resolution. All tests were carried in argon at 1 atm.

### Methods

The ignition delay was measured using the high speed images, as the time difference between the initial laser light and the formation of the ignition front. This was also validated using the delay measured by the photodiode. The average burning speed within the nanothermite microstructure was estimated using the filtered high speed images, assuming a 2D planar burning front. Details of this procedure are described in section Average Burning Speed and Reaction Mechanism.

The normalized time-averaged photodiode and pressure signals were calculated according to

Y<sup>n</sup> = R t2 t1 Ydt (t2−t1) m, where Y is the photodiode or pressure

transducer signal, t<sup>1</sup> and t<sup>2</sup> are the initial and final times (according to full width at 20% maximum), and m is the sample mass.

Additionally, SEM (Scanning Electron Microscopy) images of the mixed Al/CuO reactants and the combustion products were taken, and EDX (Energy-dispersive X-ray spectroscopy) analysis of the products was performed.

### RESULTS AND DISCUSSION

### Reactant Characterization

The as-received Al nanopowder is spherical, with an APS diameter and a specific surface area (SSA) of 40 nm and 30– 50 m<sup>2</sup> /g, respectively. Furthermore, thermogravimetric analysis at a heating rate of 10◦C/min up to 1,200◦C indicates that the active Al content is 65% by mass. Under these conditions, the Al oxidizes fully to Al2O3. Theoretically, a 40 nm Al particle with an active content of 65% has a SSA of 53.33 m<sup>2</sup> /g. The higher SSA can be attributed to the particle size distribution and soft agglomeraton/aggregation of the nanoparticles. The as-received CuO nanopowder is nearly spherical, with an APS diameter and a SSA of 40 nm and 20 m<sup>2</sup> /g, respectively. Theoretically, a 40 nm CuO particle has a SSA of 23.44 m<sup>2</sup> /g. The SSA of the CuO nanopowder is much closer to the theoretical SSA compared to the Al nanopowder.

**Figure 2** shows SEM images, at a magnifications of 50 k, of the as-prepared Al/CuO nanothemites with ER of 1 and 1.5, and their particle size distribution (based on the diameters of 300 particles at higher magnifications, 100 to 110 k). Refer to section Materials and Setup for the method of preparation. It can be seen that the stoichiometric mixture forms larger agglomerates, whereas the fuel rich mixture has larger particles. Since hexane is a nonpolar and hydrophobic liquid, the bulk density of the mixture may play a major role on the homogeneity of the suspension. The stoichiometric mixture has a higher bulk density compared to the fuel rich mixture due to the denser CuO. The most probable particle sizes are around 67 and 72 nm for the ER of 1 and 1.5, respectively. These values are larger than the APS diameters of the Al and CuO nanoparticles due to sintering during the sonication process.

The reference TMD for mixed Al/CuO nanopowders are calculated using the weight average of the Al, Al2O3, and CuO densities. The TMD for ERs of 1, 1.5, and 2 are 4.94, 4.61, and 4.37 g/cm<sup>3</sup> , respectively. The bulk density of the pellets and the lightly packed powders is relative to these TMD-values.

### High Speed Imaging of the Ignition and Flame Propagation

**Figure 3** illustrates the high speed frames of the ignition and flame propagation in (a) an Al/CuO pellet with ER = 2 and density = 37.3 %TMD, and (b) Al/CuO powder in acrylic tube with ER = 2 and density = 13.8 %TMD. The 5 × 5 edge hipass filter in the Phantom PCC software is used to clearly show the location of the burning zones.

The camera frames in **Figure 3a** show that the laser creates significant nanopaticle ablation on the pellet surface. The heat produced by the laser propagates on the top and side surfaces due to the low thermal conductivities of the Al and CuO nanoparticles, while the nanoparticle ablation continues. The time frames indicate that heat propagates faster on the side surfaces, since the heat conductivity is enhanced axially due to the compaction. The bulk ignition occurs after 1.38 ms, around the top edge of the pellet. The edges shown in **Figure 3a** are preferential locations for ignition due to the elliptical shape of the laser beam. Then, the reaction self-propagates into the unburned mixture until the pellet is disintegrated at 1.85 ms. The flame front is well defined but non-planar due to the heterogeneous reaction sites. The low porosity of the pellet impedes the combustion gases to propagate into the unburned zone; hence the heat diffusion is a major component of the heat transfer that maintains the self-propagated sustainable reaction.

The photos in **Figure 3b** show that the laser also creates significant nanopaticle ablation on the surface of the lightly pressed powders. Similarly to the pellet, the heat propagates first on the top surface. The bulk ignition of the nanothermite powder starts after 0.175 ms, and the reaction propagates downwards into the unburned mixture. The reaction front is faster and disordered due to the high porous microstructure. The reaction zone reaches the end of the unburned mixture after 0.250 ms. It is expected that the large pores reduce the resistance to mass diffusion, and gaseous and molten products penetrate through the unburned zone to break the adjacent agglomerates and effectively reduce the diffusion lengths for the reaction; hence the mass convection is a major process in the self-propagated reaction.

### Ignition and Ignition Delay

The bulk ignition of the Al/CuO pellet was consistently initiated near the edges of the top surface, away from the laser heating spot. This phenomenon is caused by the combined effects of nanoparticle ablation and net heat transfer on the surface. Ignition temperature is not reached in the laser vicinity since the heated nanoparticles are continuously removed from the surface. Theoretically, the energy balance on the laser heated surface can be simplified to

$$
\rho c\_p \frac{\partial T}{\partial t} = \alpha \dot{Q}\_{laser} - \left(\dot{Q}\_{mix} + \dot{Q}\_{Ar}\right) + \dot{Q}\_{chem}
$$

here ρ is the packing density, c<sup>p</sup> is the effective specific heat capacity, α is the absorption coefficient, Q˙ laser is the constant laser irradiance, Q˙ mix is the rate of heat transfer due to heat conduction and convection within the mixture, Q˙ Ar is the rate of heat convection and radiation to the argon, and Q˙ chem is the rate of heat release by the nanothermite reaction where its value depends on the local temperature relative to Tign, the ignition temperature

$$\begin{aligned} \dot{Q}\_{chem} &= 0 \, for \, T < T\_{\text{ign}}\\ \dot{Q}\_{chem} &> 0 \, for \, T > T\_{\text{ign}} \end{aligned}$$

The ignition temperature is reached by minimizing the Q˙ mix + Q˙ Ar term. It can be assumed that the heat convection to the argon is much lower compared to the heat transfer within the mixture, which is dominated by the heat conduction since gas generation is very low before ignition. Hence the optimal conditions for the bulk ignition (thermal runaway) is around the edge of the laser heated surface, where the net heat transfer losses and nanoparticle ablation are minimized.

**Figure 4** shows that the ignition delays increases with the packing density and the fuel rich mixtures have reduced ignition delays. As explained above, the ignition delay is controlled by the effective thermal conductivity coefficient (i.e., the porosity weighted average of the heat conductivities for the Al/CuO solid phase and the argon fluid phase). It has been shown that thermal conductivity increases linearly and absorption coefficient is uniform with density in consolidated aluminum pellets (40– 75%TMD) (Stacy et al., 2014), and the same trend is expected to exist for Al/CuO. Therefore ignition is delayed in the denser pellets due to the larger heat diffusion on the surface and within the pellet. On the other hand, for a fixed density, the fuel rich mixture with ER = 1.5 has the shortest ignition delay. This suggests that the ignition delay is controlled by both the

thermal properties of the mixture and the Al/CuO interfacial homogeneity, which controls the decomposition of CuO that provides the O<sup>2</sup> required for the ignition. If the mixture is too fuel rich (ER = 2), the additional Al enhances the overall thermal conductivity, which delays the ignition.

It should be noted that the alumina passivation layer has a significant effect on the thermal properties of the aluminum nanoparticles used in these experiments. The thermal conductivities Al2O<sup>3</sup> and CuO are equivalent, whereas the thermal conductivity of Al is an order of magnitude larger. Furthermore, the thermal conductivity of these components decreases with temperature (Kaye and Laby, 2018). The effective thermal conductivity is higher in the Al rich mixtures, and the laser heating propagates faster on the surface to reach the optimal conditions for ignition.

### Average Burning Speed and Reaction Mechanism

In order to visualize the propagation of the burning front, the location of the bulk ignition is assigned to position (0, 0) and t = 0 on the 2D graph shown in **Figure 5**. Then, four points are selected on the flame front at different time frames and fitted with

a line of best fit. The velocity of the flame front is calculated as

$$\nu = \frac{d\_i}{t\_i}$$

where d<sup>i</sup> is the orthogonal distance, and t<sup>i</sup> is the time from the ignition spot to the i th frame. The average value of these flame speeds is assumed to be the average burning speed characteristic to the pellet. It should be noted that bulk ignition does not occur at a single location (refer to **Figure 3a**); in this paper, the flame speeds are calculated using the reference ignition spot that provides that fastest propagation. The combustion front of the nanothermite has complex 3D burning features; however, in this paper it is assumed that the fastest 2D flame speeds define a specific Al/CuO mixture.

**Figure 6** shows the average burning speeds of the stoichiometric and the fuel rich Al/CuO mixtures. Two regimes are observed: the burning speed is on the order of 10 m/s in the consolidated pellets, and increases to the order of 100 m/s in the lightly pressed nanothermites. The tube confinement of the lightly pressed nanothermites plays a role on the flame speeds since the tube alters the rate of pressurization during combustion and the heat transfer to the argon; however, these effects are expected to be less significant than the porosity within the unreacted mixture.

A similar trend was observed in Al/WO<sup>3</sup> nanothermites (Prentice et al., 2006), and 80 nm Al nanoparticles/CuO nanorods (Apperson et al., 2007). It should be noted that the burning speeds obtained in these experiments are relatively low compared to literature data (Apperson et al., 2007) since the 40 nm Al particles have significantly more alumina mass, which reduces the thermal conductivity. The 80 nm Al particles have an active content of 80% (Gromov and Teipel, 2014), compared to 65% of the Al particles used in these experiments. The effect of the large alumina shell mass is more pronounced in the consolidated pellets due to the higher local concentration of the alumina. The flame propagation is reduced since the pressed alumina shells increase the diffusion resistance.

To further examine the effects of the bulk density on the burning velocity, the Andreev number, An, is examined. This has

Saceleanu et al. Combustion of Nanothermite Pellet

TABLE 1 | Andreev numbers for the Al/CuO mixtures at different packing densities.


been proposed recently to explain the transition from conductive to convective reactive flow in porous media (Weismiller et al., 2009). The An number is defined as the ratio of the convective heat transfer coefficient to the heat conduction coefficient,

$$An = \frac{\rho u d\_{\varepsilon} c\_{p}}{k\_{\text{g}}}$$

where ρ is the packing density of the nanothermite, u is the average burning velocity, d<sup>ε</sup> is the mean pore diameter, c<sup>p</sup> is the specific heat capacity of the nanothermite, and k<sup>g</sup> is the thermal conductivity of the gas. The mean pore diameter is found using Kozeny's equation for spherical particles (Skorokhod et al., 1988),

$$d\_{\varepsilon} = \frac{2}{3} \left( \frac{s}{1 - \varepsilon} \right) d\_{\mathbb{P}}$$

where ε is the porosity, and d<sup>p</sup> is the mean particle diameter. The specific heat under constant pressure is determined using the porosity averaged heat capacities of the solid and gas phases. The specific heat capacity and thermal conductivity values are examined at 2,000◦K (Murphy and Arundell, 1994; NIST, 2017), assuming that the adiabatic flame temperature of 2,800◦K is not reached. Also, it is assumed that the gas phase is a mixture of argon and oxygen, which is the major gaseous species during the reaction (Baijot et al., 2015).

**Table 1** outlines the Andreev number for the stoichiometric and fuel rich Al/CuO mixtures. Generally, An is between 1 and 10 in the slow burning regime, indicating that heat conduction is significant for the self-sustaining reaction. In the fast burning regime, An is between 100 and 1,000, which indicates that thermal and mass convection drive the flame propagation.

Overall, the fuel rich Al/CuO mixtures generated higher flame speeds than the stoichiometric mixtures. Similarly, maximum pressures and burn speeds in pressure cell experiments were obtained under fuel rich conditions (ER = 1.1) (Sanders et al., 2007), and this was assumed to be due to higher thermal conductivities (Jacob et al., 2017). Fuel rich Al/MO<sup>3</sup> also showed improved propagation velocities compared to stoichiometric mixtures, due to optimum gas and liquid Mo generation that enhances the convective heat transfer (Son et al., 2007). As noted in **Table 1**, the fast burning regime is reached in the fuel rich lightly packed mixtures, but not in the stoichiometric mixture. These observations indicate that the mixture homogeneity is a critical parameter for fast flame propagation, and the local reaction rates control the gas generation that drives the convective burning. In Apperson et al. (2007), the selfassembled Al/CuO composites produced higher combustion rates compared to the physically mixed powders due to the larger interfacial area.

To further study the nanothermite reaction, post-combustion SEM images of the consolidated pellets and lightly packed powder under stoichiometric and fuel rich conditions are shown in **Figure 7**. Generally, the products of the nano scale reactants are on the micro scale, and composed of spherical Cu particles and aggregates of Al2O<sup>3</sup> and AlxCuyO<sup>z</sup> intermetallics. Similar structures have been observed in (Jacob et al., 2015) and (Jacob et al., 2017).

Although the initial pellets densities of the stoichiometric and fuel rich mixture in **Figures 7a,b** are similar, the morphologies of the products are different. The reactant aggregates observed in the post-combustion of the stoichiometric pellet indicates a highly incomplete combustion. This is attributed to the larger reactant agglomerates, and inhomogeneous mixing, shown previously in **Figure 2a** as compared to **Figure 2b**. The products of the lightly pressed nanothermites in **Figures 7c,d** show that reactive sintering occurs in the highly porous mixtures. Despite the high porosity that promotes convection, the condensed phase reactions are much faster than the heterogeneous reactions. For example, the reaction time of Al and CuO in condensed phase is on the order of 1 µs (Egan et al., 2014), much faster than the burning time of Al nanoparticles in an oxygen atmosphere which is on the order of 100 µs (Bazyn et al., 2006). It should be noted that significant reactant aggregates exist in the products of the stoichiometric loose powders (**Figure 7c**), which is evidence of an incomplete combustion. Incomplete Al/CuO reactions have also been reported in Jacob et al. (2017) and Glavier et al. (2015). The large ignition delay and the low burning speeds of the stoichiometric mixtures are mainly caused by inhomogeneous mixing of the reactants that limits the local Al-CuO reactions in the condensed and heterogeneous phase. Consequently, this limits the gas generation that promotes the mass convection, and the combustion temperature that promotes the heat conduction.

### Reaction Performance of the Pellets

The photodiode and pressure signals were analyzed for the different pellet densities in order to assess their performance. The key factors depend on the application. For example, propulsion and igniter applications required fast pressurization rates, whereas welding applications require high reaction temperatures. It should be noted that the pressure transducer was not installed in the reaction zone; hence the pressure signals are only used for comparison purpose.

**Figure 8** shows the raw signals of the photodiode and pressure transducer during the combustion of a high density Al/CuO pellet with ER = 1.5. The reaction occurs before the gas

FIGURE 7 | Post-combustion SEM of high density pellets with (a) ER = 1 (59.5 %TMD) and (b) ER = 1.5 (58.5 %TMD); and low density powders with (c) ER = 1 (12.8 %TMD) and (d) ER = 1.5 (13.6 %TMD).

release, and the burn duration is faster than the pressurization time. It should be noted that some researchers observed the decomposition of CuO prior to the ignition; however, the pressure transducer in these experiments can only measure the overall pressurization near the vessel walls. **Figure 9** shows the normalized and time-averaged photodiode and pressure signals of the Al/CuO pellets. The normalized time-averaged photodiode signal term represents the specific energy release rate by the reaction. This term is independent of the density, which indicates that the degree of oxidation of Al is similar. The mixtures with ER of 1.5 have a larger normalized photodiode signal compared to the mixture with ER of 2 since the extra Al

in the richer mixture acts as a heat sink. The normalized timeaveraged pressure signal increases linearly with the pellet density. A similar linear increase with the density was also observed for the pressurization rate, which is estimated from the slope of the raw pressure curve. The higher pressure rates in the denser pellets are caused by the reduced volume for gas expansion, which is also predicted by a theoretical model assuming local thermodynamic equilibrium (Baijot et al., 2015). Furthermore, it is expected that the consolidated Al and CuO reactants have more reactive interfaces, which enhance the initial reaction rates in the condensed phase. In Glavier et al. (2015), the maximum pressure and pressurization rates of Al/CuO pellets also increase with the %TMD. Theoretically, much higher pressures are predicted since the actual gas phase chemistry is unknown. It should be noted that the normalized pressure signals of the stoichiometric Al/CuO pellets are low compared to the fuel rich mixtures, and independent of density. This is further indication that the local reaction rates are inhibited by the homogeneity of the mixture.

The burning durations of the Al/CuO nanothermites in these experiments, as measured by the duration of the photodiode signal, are an order of magnitude longer than the durations of flame propagation. This indicates that the degree of oxidation of the Al nanoparticles is limited by the heterogeneous Al– O<sup>2</sup> reactions. In Jacob et al. (2017), temporal temperature measurements in Al/CuO nanopowders were near the flame temperature of Al particles in air (∼4,000◦K). Thus, in order to improve the reactivity of these nanothermites, the Al-CuO interfaces should be optimized to ensure that most of the reaction

### REFERENCES


initiates in the condensed phase, and oxygen decomposition from the CuO can readily react with the Al.

### CONCLUSIONS

The experiments show that the reactivity of physically mixed Al/CuO nanothermites is highly sensitive to the homogeneity, ER and packing density of the mixture. The fuel rich mixtures burned much faster than stoichiometric mixtures due to formation of smaller agglomerates in the reactants. The ignition delay on the pellet surface is controlled by the nanoparticle ablation and the net heat transfer on the surface, such that the pellet edges are the preferential spots for the bulk ignition. The propagation speed of the burning front increases from an order of 10 m/s in the consolidated pellets (40–60%TMD) to an order of 100 m/s the in lightly packed powders (10%TMD). Enhanced flame speed is caused by a change in the controlling mechanism from heat conduction to mass convection with decreasing the packing density. The reactivity of the Al/CuO pellets increases generally with the packing density, whereas the normalized pressurization rate increases linearly with the density.

### DATA AVAILABILITY

The raw data supporting the conclusions of this manuscript will be made available by the authors, without undue reservation, to any qualified researcher.

### AUTHOR CONTRIBUTIONS

FS obtained the experimental results, analyzed the data, and wrote the manuscript. MI and NC set up the experiments and provided the lab facilities. JW revised the manuscript and supported this research.

### FUNDING

This research was supported by the Natural Sciences and Engineering Research Council of Canada (NSERC) Discovery program and Ontario's Early Researcher Awards program.

### ACKNOWLEDGMENTS

The authors would like to acknowledge the funding from the NSERC Discovery Program.


effect of impurities on flame propagation. Energ. Fuels 20, 2370–2376. doi: 10.1021/ef060210i


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Saceleanu, Idir, Chaumeix and Wen. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Urchin-Like Ni2/3Co1/3(CO3)1/2(OH)·0.11H2O for High-Performance Supercapacitors

### Zi-Min Jiang, Ting-Ting Xu\*, Cong-Cong Yan, Cai-Yun Ma and Shu-Ge Dai\*

*Key Laboratory of Material Physics of Ministry of Education, Zhengzhou University, Zhengzhou, China*

Here, we report our finding in the fabrication of novel porous urchin-like Ni2/3Co1/3(CO3)1/2(OH)·0. 11H2O (denoted as NC) nanomaterial composed of numerous nanoneedles through an one-step hydrothermal method, which deliveres a high specific capacity of 318 C g−<sup>1</sup> at a current density of 1 A g−<sup>1</sup> . Moreover, an architectural composite electrode consisting of the porous NC nanoneedles wrapped by reduced graphene oxide (rGO) nanosheets exhibits large specific capacity (431 C g−<sup>1</sup> at 1 A g−<sup>1</sup> ), high rate capability and long cycling life (94% capacity retention after 5,000 cycles at 20 A g−<sup>1</sup> ). The presence of rGO in the composite electrode greatly improves the electronic conductivity, providing efficient current collection for fast energy storage.

#### Edited by:

*Qiaobao Zhang, Xiamen University, China*

#### Reviewed by:

*Linfeng Hu, Fudan University, China Hongwei Mi, Shenzhen University, China*

### \*Correspondence:

*Ting-Ting Xu xutt@zzu.edu.cn Shu-Ge Dai shugedai@zzu.edu.cn*

#### Specialty section:

*This article was submitted to Nanoscience, a section of the journal Frontiers in Chemistry*

Received: *30 July 2018* Accepted: *31 August 2018* Published: *28 September 2018*

#### Citation:

*Jiang Z-M, Xu T-T, Yan C-C, Ma C-Y and Dai S-G (2018) Urchin-Like Ni*2/3*Co*1/3*(CO*3*)*1/2*(OH)*·*0.11H*2*O for High-Performance Supercapacitors. Front. Chem. 6:431. doi: 10.3389/fchem.2018.00431* Keywords: porous, NiCo hydroxides, graphene, composites, supercapacitors

### INTRODUCTION

Supercapacitors (SCs), as an electrochemical energy storage device with high power density, long cycle life and fast charging capability, are considered as one of the most promising energy storage devices (Liang et al., 2017). The most attractive feature of SCs is its fast charging capability with large current density than other energy storage devices (Aricò et al., 2005). The structures and properties of the electrode materials are critical to the performance of the energy storage devices (Shi et al., 2017). Therefore, it has greatly attracted the interest of the researches to develop new electrode materials with excellent specific capacitance and good cycle life (Zhang et al., 2018a; Zhao et al., 2018a).

The rate performance and cycle life of the electrode materials are playing a significant role in the practicality for supercapacitors (Dai et al., 2017). Transition metal compounds such as cobalt hydroxides (Jiang et al., 2018; Long et al., 2018; Yang et al., 2018a), nickel hydroxides (Zhang et al., 2010; Ji et al., 2013; Jiang et al., 2017) with high theoretical capacity have been considered as the most promising positive electrodes for high-performance SCs. Unfortunately, the drawbacks of transition metal materials, such as limited useful cycle life and the poor rate performance, still need further improvement. One of the solutions is to design a nanostructure with large specific surface area, which can enlarge the electrode-electrolyte contact area and improve the effective utilization of the electrode materials, thereby expanding the specific capacitance of the SCs (Liu et al., 2017). Although some novel nanostructures of transition metal compounds have been reported for SCs, such as nanoparticles (Wei et al., 2014), nanowires (Zhao et al., 2018b), nanoflakes (Cao et al., 2007; Pan et al., 2018). However, the rate performance and cycle life of those electrode materials are not satisfactory, the depth of the electrode reaction decreases with the increasing current. In particular, urchin-like nanostructure composed of numerous nanounits have attracted great interest as an unusual group with well-defined nanoneedles structure and shell structure in functional materials (Jin et al., 2018). Their unique structure possess larger surface area and higher pore volume, which are favorable for bringing about more active reaction sites in the electrolyte, the ultra-fine nano-needle structure can also effectively reduce the structural force during charging and discharging, and the electrode material has higher stability and safety.

Recently, some research results show that the composites of nickel and cobalt bimetallic electrode materials show a better cycle stability than the corresponding nickel or cobalt electrode materials because of their complementary advantages and synergistic effects (Chen et al., 2018a; Liu et al., 2018). Nonetheless, how to synthesize such electrode materials with urchin-like nanostructure in a simple and efficient way is still a challenge. In this work, we report our findings in the synthesis and characterization of the unique urchin-like nanostructure with large surface-tovolume ratio. In particular, the NC Co-doping crystal with Co(CO3)x(OH)<sup>y</sup> anions possessed one-dimensional chain-like crystal structure was successfully prepared, which delivered a specific capacity of 318 C g−<sup>1</sup> at a current density of 1 A g−<sup>1</sup> . Moreover, we successfully prepared the NC/rGO nanocomposite by introducing rGO, which exhibited enhanced electrochemical performance compared to that of the NC nanomaterial.

### EXPERIMENTAL SECTION

### Preparation of NC Nanomaterial

All chemical reagents in this work were used as received without further purification. The urchin-like NC nanomaterial was synthesized by one-step hydrothermal method. First, 1 mmol of NiCl2·6H2O (aladdin, 99%), 3 mmol of CH4N2O (aladdin, 99%) and 0.5 mol of CoCl2·6H2O were first dissolved in 10 mL of deionized water in the Teflon containers by stirring for 30 min to form homogeneous solution. Then the autoclave was sealed and maintained at a temperature of 100◦C for 12 h. After natural cooling to room temperature, the samples were taken out, rinsed with deionized water until pH is close to 7, and dried under vacuum at 70◦C overnight.

### Preparation of NC/rGO Nanomaterials

The NC/rGO composite was synthesized using a hydrothermal method similar to the synthesis of NC. 1 mmol of NiCl2·6H2O (aladdin, 99%), 3 mmol of CH4N2O (aladdin, 99%) and 0.5 mmol of CoCl2·6H2O were first dissolved in 10 mL of GO solution (2 mg mL−<sup>1</sup> ) in the Teflon container by stirring 30 min to form homogeneous solution. Then the autoclave was sealed and maintained at a temperature of 100◦C for 12 h. After natural cooling to room temperature, the samples were taken out, rinsed with deionized water until pH is close to 7, and dried under vacuum at 70◦C overnight.

### Material Characterization

The phase structures of the samples were characterized by X-ray diffraction (XRD) analysis (XRD, PA National X′ Pert Pro with Cu Kα radiation), The microstructure and morphology of NC nanomaterials were characterized using field emission scanning electron microscopy (Zeiss, sigma300) and highresolution transmission electron microscopy (HRTEM, JEOL, JEM-2100) with energy dispersive X-ray spectrometry (EDS). The nitrogen adsorption-desorption isotherm measurement of the sample was performed using a ASAP2420-4MP. The specific surface area was obtained by the Brunauer-Emmett-Teller (BET) method. The element atomic ratio analysis were characterized by inductively coupled plasma mass spectrometry (ICP MS) using a Agilent 7700.

### Electrochemical Measurement

The as-prepared NC and NC/rGO electrodes were used as the working electrodes with a platinum sheet electrode and a Ag/AgCl (prefilled with 4 M KCl aqueous solution saturated with AgCl) reference electrodes (Gogotsi, 2014), respectively. The electrochemical tests were performed using a CHI760 electrochemical workstation. Cyclic voltammetry (CV), galvanostatic charge-discharge (GCD) and electrochemical impedance spectroscopy (EIS) tests were carried out using a three-electrode configuration in 2 M KOH aqueous solution at room temperature. For the working electrode, a mixture containing of 80 wt% electroactive material, 10 wt% Super P Li (TIMCAL Graphite and Carbon) powders, and 10 wt% polytetrafluoroethylene (PTFE) binder was well mixed to prepare working electrodes, and then rolled with the assistance of ethanol to form a uniform film with a typical area mass loading of approximately 1.5 mg cm−<sup>2</sup> , and dried under vacuum at 80◦C for 12 h, then pressed it between two nickel foams. The cyclic voltammograms were acquired in a potential range of 0–0.5 V at different scan rates, the charge-discharge processes were performed by cycling the potential from 0 to 0.45 V at different current densities in a 2 M KOH aqueous electrolyte, and the EIS was performed in the frequency range from 0.01 Hz to 100 k Hz under open-circuit condition. The specfic capacity Q

(C g−<sup>1</sup> ) of the battery-type electrodes was calculated as follows (Mai et al., 2013):

(Wei et al., 2017):

$$\begin{aligned} CO(NH\_2)\_2 + H\_2O &= CO\_2 + 2NH\_3 \\ CO\_2 + H\_2O &= 2H^+ + CO\_3^{2-} \end{aligned} \quad \text{(1)}$$

$$\begin{aligned} CO\_2 + H\_2O &= 2H^+ + CO\_3^{2-} \\ NH\_3 + H\_2O &= NH\_4^+ + OH^- \end{aligned} \quad \text{(2)}$$

$$\begin{aligned} \text{CO}\_3\text{}^{2-} + 2\text{A}^{2+} + 0.22\text{H}\_2\text{O} + \text{OH}^- &= 2\text{A(CO}\_3\text{)}\_{0.5} \text{(OH)}\\ \bullet 0.11\text{H}\_2\text{O} &\qquad \text{(4)} \end{aligned}$$

In the chemical reaction system, firstly, CO(NH2)<sup>2</sup> combined with H2O to produce CO<sup>2</sup> and NH<sup>3</sup> (equation 1). Then, CO<sup>2</sup> and NH<sup>3</sup> combined with H2O immediately to produce CO2<sup>−</sup> 3 , OH<sup>−</sup> and NH4<sup>+</sup> (equations 2 and 3), respectively. Finally, CO2<sup>−</sup> 3 and OH<sup>−</sup> combined with A2<sup>+</sup> (Ni2<sup>+</sup> or Co2+) to form the sample (equation 4).

SEM was performed to examine the morphology and structure of the as-synthesized samples. **Figure 2a** shows the lowmagnification SEM image of as-synthesized NC nanomaterial, which appears to be thorn balls constructed of numerous nanoneedles, and similar to the urchin-like microspheres. In **Figure 2b**, the high magnification SEM image reveals that the diameters of those nanoneedles are around 10–20 nm. The morphology evolution mechanism of urchin-like NC may be as follows: firstly, the ultrashort nanoneedles in situ grow on the surface of the solid core as the self-generated templet without crack and finish solid core growth. Owing to the increasing of the amount of NH<sup>3</sup> and CO2<sup>−</sup> 3 , NH<sup>3</sup> holds the metal ions moving to the top of the ultrashort nanoneedles with high surface energy and generates precipitate, which

$$Q = i\_m \Delta t \tag{1}$$

where i<sup>m</sup> = I/m (A g−<sup>1</sup> ) is the current density, I is the current, m is the mass of the active electrode material and 1t is the discharge time.

### RESULTS AND DISCUSSION

It is well known that the polymorphs of nickel hydroxides depend on the hydrothermal reaction conditions such as the ratio of precursors, reaction temperature and time (Li et al., 2016). By tuning the initial mole ratio of reactants, we synthesized the urchin-like NC nanomaterials by a simple one-step solvent-thermal method. The XRD pattern of the NC (black curve) sample is illustrated in **Figure 1**. All diffraction peaks of the XRD pattern can be indexed to orthorhombic Co(CO3)1/2(OH)·0.11H2O (JCPDS NO. 48-0083). The strong diffraction peaks centered at 2θ = 17.2◦ , 26.4◦ , 33.8◦ , 39.4◦ , and 46.9◦ , attributed to (0 2 0), (2 2 0), (2 2 1), (2 3 1), and (3 4 0), respectively. And the XRD pattern of NC/rGO (red curve) sample is extremely similar to the NC pattern, and no other obvious peaks are detected. The reaction mechanism may be as follows

TABLE 1 | Results of ICP-MS for NC-1 and NC-2 and NC-3.

SOLUTION CONCENTRATION (MG L−<sup>1</sup> )


leads to the preferred growth of product at the top of short nanoneedles and makes the short nanoneedles slowly into the long nanoneedles. The SEM images of the NC/rGO samples are shown in **Figures 2c,d**, and the morphology of the composite is nanosheets. These NC nanoneedles are distributed uniformly in the rGO nanosheets, and numerous nanosheets over-lap together to form a irregular three-dimensional architecture. Notably, this unique nanostructure with large surface is favorable for efficient and fast transport of the electrolyte to the surface of the active materials, thus improving the effective utilization of the active material. The high magnification TEM images (**Figures 3a,b**) show the clear morphology of the nanoneedles in the NC and NC/rGO materials. These nanoneedles with a diameter of 10∼20 nm (**Figure 3a**) build an unique urchin-like nanostructure, which greatly shorts the electron transport channel, and improves the conductivity of the electrodes. In **Figure 3b**, the NC nanoneedles are attached to the rGO nanosheets, forming a composite nanostructure. However, the urchin-like suffered from the deficiency of those shells, the internal active material could not make full contact with the electrolyte and then affect the performance of the whole electrode. Fortunately, the addition of rGO nanosheet overcomes this shortcoming, these nanoneedles distributed uniformly on the graphene nanosheets, which provides an effective pathway for charge transpor, weakens the polarization phenomenon at high current, and further improves the effective utilization of the active material and the rate performance of the electrodes. **Figures 3c,d** shows the bright field TEM image and mapping images of the NC/rGO composite, which clearly reveals that the NC nanoneedles are distributed homogeneously in the rGO nanosheets. The molar ratio of Ni:Co is determined to be 2:1 based on the result analysis, which is similar to the results of ICP-MS (∼70.03:36.89, **Table 1**), and close to the mole ratio of adding nickel and cobalt

sources, proving that urchin-like NC nanomaterial possesses a molecular formula of NC. To investigate the porosity and surface area of as-prepared samples, N<sup>2</sup> adsorption-desorption isotherms of NC and NC/rGO conducted at 77.350 K were investigated and are displayed in **Figure 4A**. Through Brunauer-Emmett-Teller (BET) analysis, the surface areas of NC and NC/rGO composite were identified as 62.55 m<sup>2</sup> g −1 and 138.27 m<sup>2</sup> g −1 , respectively. In addition, the pore size distribution curves were also investigated via using Barrett-Joyner-Halenda (BJH) method (**Figure 4B**). For NC, no obvious peaks arise in the entire range of pore size distribution curve. For NC/rGO, an obvious peak at around 40∼50 nm in the pore size distribution curve was observed, which was attributed to the rGO sheets. These nanosheets provide a lot of attachment points for the growth of nanoneedles through the hydrothermal treatment. Higher surface area offers abundant reaction sites and enhances the specific capacity of the NC/rGO composite electrodes.

The electrochemical performance of the NC microspheres and NC/rGO nanocomposite were first evaluated with a three-electrode configuration in 2 M KOH aqueous electrolyte, the results are shown in **Figures 5a–d** and **Figure S1**. For comparison, the electrochemical performance of NC and NC/rGO electrodes in the potential range between 0 and 0.5 V (versusAg/AgCl) at a scan rate of 10 m V s−<sup>1</sup> are demonstrated in **Figure 5a**. Evidently, the NC/rGO electrode exhibits a higher current density and larger increment in the CV integrated area than NC electrode, which indicates an obvious enhanced of electrochemical reaction activity and charge storage, suggesting the improvement of conductivity and electrochemical performance. **Figure 5b** displays the CV curves of the NC/rGO composite electrode at different scan rates; all the CV curves exhibit a pair of redox peak, which indicates that the energy storage in our electrodes appears to be battery-type materials. The corresponding CV curves of NC microspheres was also collected and shown in **Figure S2**. It is clear that all of the plateaus at around 0.05–0.15 V, which correspond to the redox processes. Compared with the CV curves of NC, all of the CV curves of the NC/rGO composite electrode turn into smooth, indicating that the rGO greatly improved the conductivity of the composite electrodes.

The specific capacities of the NC and NC/rGO electrodes were calculated based on the GCD curves, and the corresponding GCD curves of NC and NC/rGO electrodes are shown in **Figure S2** and **Figure 5c,** respectively. As shown in **Figure 5d**, the NC electrode delivered a specific capacity of 318 C g−<sup>1</sup> at a current density of 1 A g−<sup>1</sup> , and retained ∼71% (228 C g−<sup>1</sup> ) of the capacity when the current density was increased from 1 to 20 A g−<sup>1</sup> . In comparison, the NC/rGO electrode exhibted a specific capacity of 431 C g−<sup>1</sup> at a current density of 1 A g−<sup>1</sup> , and retained ∼74% (320 C g−<sup>1</sup> ) of capacity, which is higher than that of the NC electrode. The obvious discharge platform further proves the Faraday energy storage mode, demonstrating that the composite electrodes possess battery-like effect in the charge and discharge process (Yang et al., 2018b). Besides, the specific capacity of the NC/rGO were higher than other transition metal compounds reported in literature, such as Co0.45Ni0.55O/RGO nanocomposites (411 C g−<sup>1</sup> at a current density of 1 A g−<sup>1</sup> )(Xiao and Yang, 2012), NiCo2-RGO nanocomposites (417 C g−<sup>1</sup> at a current density of 1 A g−<sup>1</sup> )(Wang et al., 2011), NiCo2O<sup>4</sup> nanocomposites (296 F g−<sup>1</sup> at a current density of 1 A g−<sup>1</sup> )(Xiao and Yang, 2011). The good electrochemical performance of NC/rGO composite nanomaterial may be attributed to its unique morphology and crystal structure. First, the self-contained OH<sup>−</sup> and the one-dimensional chain-like crystal structure unit increase the electrode material ionic conductivity. Second, the porous composite nanostructure has large specific surface area of 138.27 m<sup>2</sup> g −1 , which greatly increases the effective area between active material and electrolyte, and facilitates more effective exposure of surface active sites for fast Faradic redox reaction. Third, the rGO nanosheets provides a great specific area for the NC nanoneedles growth, constructing a threedimensional conducting network, which ensures the excellent rate performance. Cycling life is an important property for the SCs. The NC/rGO composite electrode could still retain ∼94% of the initial capacity over 5,000 cycles at a high current density of 20 A g−<sup>1</sup> , which is higher than the NC electrode

(∼88% of the initial capacity), revealing its good cycling stability (**Figure 5e)**. Furthermore, the cycling stability of the NC/rGO in 2 M aqueous KOH aqueous solution using a typical threeelectrode cell were superior to many transition metal compounds electrode materials reported in literature, such as Co(OH)<sup>2</sup> nanosheets (89.1% capacitance retention after 5,000 cycles at 20 A g−<sup>1</sup> ) (Chen et al., 2018b), Co-doped α-Ni(OH)2/RGO nanosheet (87.9% capacitance retention after 1,000 cycles at 10 A g−<sup>1</sup> ) (Zhang et al., 2018b), Ni(OAc)2/Co(NO3)<sup>2</sup> (91.5% capacitance retention after 1,000 cycles at 5 A g−<sup>1</sup> ) (Wei et al., 2018), illustrating the excellent capacity and longterm electrochemical stability of the NC/rGO composite electrode.

### CONCLUSION

In summary, a novel porous NC nanomaterial with urchinlike architecture has been successfully prepared by onestep hydrothermal process. When tested in a three-electrode configuration in 2 M KOH aqueous electrolyte, it is found that the NC nanomaterial exhibits a high specific capacity of 318 C g −1 at 1 A g−<sup>1</sup> . Moreover, a porous NC/rGO composite electrode was also successfully prepared, which exhibited good specific capacity (431 C g−<sup>1</sup> at 1 A g−<sup>1</sup> ), great rate capability (∼74% capacity retention) and long cycling life (94% capacity retention over 5,000 cycles), demonstrating its promising potential for the high-performance supercapacitors.

### AUTHOR CONTRIBUTIONS

Z-MJ carried out the material preparation and electrochemical test; T-TX and S-GD carried out and analyzed the XRD, SEM, and TEM analysis; Z-MJ wrote the paper and all authors discussed the results; Z-MJ, T-TX, C-CY, C-YM and S-GD revised the manuscript; T-TX attained the main financial support for the research and supervised all the experiments.

### ACKNOWLEDGMENTS

This work was supported by the National Natural Science Foundation of China (Grant No. 11504331), the China Postdoctoral Science Foundation (Grant No. 2015M582196,

### REFERENCES


Grant No. 2015M582193, Grant No. 2015M582194, Grant No. 2018M630831), the Outstanding Young Talent Research Fund of Zhengzhou University (Grant No. 1521317005, Grant No. 1521317001), the National Natural Science Foundation of China (Grant No. 21805247), the National Natural Science Foundation of China (Grant No. 21805247), the China Postdoctoral Science Foundation (Grant No. 2018M630831).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2018.00431/full#supplementary-material


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Jiang, Xu, Yan, Ma and Dai. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Nitrogen Doped Carbon Nanosheets Encapsulated in situ Generated Sulfur Enable High Capacity and Superior Rate Cathode for Li-S Batteries

Zhijun Guo<sup>1</sup> , Xiaoyu Feng<sup>1</sup> \*, Xingxing Li <sup>1</sup> , Xuming Zhang<sup>1</sup> , Xiang Peng<sup>2</sup> , Hao Song<sup>1</sup> , Jijiang Fu<sup>1</sup> , Kang Ding<sup>1</sup> , Xian Huang<sup>1</sup> and Biao Gao<sup>1</sup> \*

*<sup>1</sup> The State Key Laboratory of Refractories and Metallurgy and Institute of Advanced Materials and Nanotechnology, Wuhan University of Science and Technology, Wuhan, China, <sup>2</sup> School of Materials Science and Engineering, Wuhan Institute of Technology, Wuhan, China*

### Edited by:

*Qiaobao Zhang, Xiamen University, China*

#### Reviewed by:

*Ruitao Lv, Tsinghua University, China Xiwen Wang, Hunan University, China*

#### \*Correspondence:

*Xiaoyu Feng fengxiaoyu@wust.edu.cn Biao Gao gaobiao@wust.edu.cn*

#### Specialty section:

*This article was submitted to Nanoscience, a section of the journal Frontiers in Chemistry*

Received: *31 July 2018* Accepted: *30 August 2018* Published: *25 September 2018*

#### Citation:

*Guo Z, Feng X, Li X, Zhang X, Peng X, Song H, Fu J, Ding K, Huang X and Gao B (2018) Nitrogen Doped Carbon Nanosheets Encapsulated in situ Generated Sulfur Enable High Capacity and Superior Rate Cathode for Li-S Batteries. Front. Chem. 6:429. doi: 10.3389/fchem.2018.00429* Lithium-sulfur batteries (LSBs), with large specific capacity (1,675 mAh g−<sup>1</sup> ), are regarded as the most likely alternative to the traditional Lithium-ion batteries. However, the intrinsical insulation and dramatic volume change of sulfur, as well as serious shuttle effect of polysulfides hinder their practical implementation. Herein, we develop three-dimensional micron flowers assembled by nitrogen doped carbon (NC) nanosheets with sulfur encapsulated (S@NC-NSs) as a promising cathode for Li-S to overcome the forementioned obstacles. The *in situ* generated S layer adheres to the inner surface of the hollow and micro-porous NC shell with fruitful O/N containing groups endowing both efficient physical trapping and chemical anchoring of polysulfides. Meanwhile, such a novel carbon shell helps to bear dramatic volume change and provides a fast way for electron transfer during cycling. Consequently, the S@NC-NSs demonstrate a high capacity (1,238 mAh g−<sup>1</sup> at 0.2 C; 1.0 C = 1,675 mA g−<sup>1</sup> ), superior rate performance with a capacity retention of 57.8% when the current density increases 25 times from 0.2 to 5.0 C, as well as outstanding cycling performance with an ultralow capacity fading of only 0.064% after 200 cycles at a high current density of 5.0 C.

Keywords: in situ generated sulfur, nitrogen doped carbon, cathode, chemical anchoring, physical trapping, lithium-sulfur batteries

### INTRODUCTION

Lithium-sulfur batteries (LSBs) own many advantages including high energy density (2,600 Wh kg−<sup>1</sup> , five times larger than lithium ion bateries), natural abundance, and environmental friendliness (Rehman et al., 2016; Xu et al., 2017; Qu et al., 2018; Zhang et al., 2018). Thus, LSBs have become one of the most promising candidates for large-scale energy storage and long endurance electric vehicles applications (Seh et al., 2016; Chen et al., 2017; Ye et al., 2018). However, the practical implement of the LSBs is impeded by several challenges, such as electronically and ionically insulating of S, large volume change (∼80%) during cycling, as well as shuttle effect induced by the high solubility and mobility of the lithium polysulfides (LiPSs) intermediates. To tackle with the above issues, one of the effective routes is encapsulating S into conductive carbon hosts [active carbon (Moreno et al., 2015; Li F. et al., 2017), carbon spheres (Qu et al., 2013; Zhou et al., 2017), graphene (Tang et al., 2016; Du et al., 2017), carbon nanotubes (Li M. et al., 2017) and nanofibers (Liu et al., 2018)] with high porosity or/and high specific surface area to improve electron/ion conductivity, tolerate volumetric expansion and physically trap LiPSs. Nevertheless, the weak interaction between nonpolar carbon and polar LiPSs inevitably results in effusion and irreversible loss of LiPSs from the cathodes and thus rapid capacity fading during cycling. Polar metal oxides, metal sulfides, metal carbides, and metal nitrides have recently been explored as efficient host materials for S cathodes in LSBs and the S/metal oxide composites have been demonstrated to achieve improved cycle stability via chemical anchoring between the metal oxide and LiPSs in the cathodes. However, limited active sites of these polar hosts with low surface area lead to low immobilization efficiency for LiPSs (Shi et al., 2018). Recently, heteroatom doping (e.g., N, S, P) has been developed to enable carbon hosts to be polarity and retain the large surface area simultaneously, which can not only trap the LiPSs physically, but also improve conversion kinetics from LiPSs into solid Li2S (Wang X et al., 2014; Pang et al., 2015; Zhou et al., 2015a; Wang et al., 2016, 2019; Li C. et al., 2018). Moreover, two-dimensional (2D) nanocarbon materials have drawn special attention as S hosts due to shorter ion/electron transport distance and more active sites (Chabu et al., 2017; Song et al., 2018; Sun et al., 2018; Wang Q et al., 2018).

In this work, we develop a hierarchical three-dimensional (3D) nano-flowers host consisting of hollow 2D nitrogen doped carbon nanosheets (NC-NSs) with micro-porosity shell as the S host. The S is encapsulated in the 2D NC-NSs (S@NC-NSs) with a thickness of 25–40 nm via in-situ oxidization of NC coated ZnS NSs (ZnS@NC-NSs) by Fe3+. Such unique hierarchical S@NC-NSs assembled by nanosheets endow several advantages as the S host for Li-S: Firstly, the NC shell with typical microsized pores can not only physically confine the LiPSs but also chemically anchor LiPSs via rich N/O containing functional groups; Secondly, the in situ generated S is totally encapsulated in the NC shell, resulting high utilization; Thirdly, S@NC-NSs have large inner space and the elastic carbon shell can accommodate the volume expansion of the S during lithiation; Lastly, 3D hierarchical nano-flowers structure assembled by 2D NC-NSs provides a fast conductive network, and integral structure supporters. Ultimately, the S@NC-NSs deliver a high specific capacity of 1,238 mAh g−<sup>1</sup> at a current density of 0.2 C (1.0 C = 1,675 mA g−<sup>1</sup> ), superior rate performance with a capacity retention of 57.8% when the current density increases 25 times from 0.2 to 5.0 C as well as a remarkable cycling performance with a capacity loss of 0.064% after 200 cycles at a high current density of 5.0 C.

### EXPERIMENTAL

The ZnS NSs were fabricated through a hydrothermal method. Briefly, 0.446 g of Zn(NO3)2·6H2O and 0.114 g of CS(NH2)<sup>2</sup> were added into an aqueous solution containing 20 mL of diethylenetriamine (DETA) and 20 mL of deionized water (DW) under magnetic stirring. Then, the mixed solution was transferred into a 50 mL Teflon-lined autoclave and placed in an oven at 180◦C for 12 h. Afterward, the ZnS NSs were obtained by filtration with absolute alcohol and DW for several times. Then, 200 mg of ZnS NSs were dispersed in the 400 mL of DW under magnetic stirring followed by adding 484 mg of Tris-HCl buffer and 400 mg of dopamine (DA) to coat ZnS with polydopamine (PDA) via chemical polymerization for 1.0 h under stirring in air. After washing with DW and filtrating for several times, the dried products were further carbonized at 700◦C for 2 h to produce ZnS@NC-NSs in N2. Then, the ZnS@NC-NSs were dispersed into aqueous ferric chloride (FeCl3) solution and stirred over 12 h to convert the ZnS core into S because of the strong oxidizing ability of Fe3+. Finally, the reactant was washed with 1 M HCl solution and DW to obtain S@NC-NSs after vacuum drying over night at 60◦C.

### Materials Characterizations

The morphology, structure and composition of ZnS NSs, ZnS@NC-NSs and S@NC-NSs were characterized by fieldemission scanning electron microscopy (SEM, FEI Nova 450 Nano), transmission electron microscopy (TEM, JEM-2100 UHR STEM/EDS), X-ray photoelectron spectroscopy (ESCALAB 250Xi), and X-ray diffraction [XRD, Philips X' Pert Pro (Cu Kα radiation, λ = 1.5418 Å)]. Micromeritics ASAP 2020 analyzer was applied to measure pore size distribution and the N<sup>2</sup> adsorption-desorption behavior of the NC shell. The S content in the S@NC-NSs was obtained by Naichi Corporation STA449C from room temperature to 600◦C with a heating rate of 10◦C min−<sup>1</sup> in Ar. The concentration of LiPSs was determined by Ultraviolet-visible spectrophotometer (CARY 300).

### Electrochemical Tests

The S@NC-NSs were mixed with acetylene black and polyvinylidene fluoride (PVDF) with a weight ratio of 7:2:1 to form a homogeneous slurry in N-Methyl-2-pyrrolidinone (NMP). Then, the slurry was uniformly coated on Al foil (15µm) and vacuum dried at 60◦C for 12 h. The coin-type cells were assembled with S@NC-NSs as cathode and Li metal foil as anode. The electrolyte consisted of 1 M lithium bis (trifluoromethane) sulfonimide (LiTFSI) in 1,3 dioxolane/1,2 dimethoxyethane (DOL/DME) (1:1, v/v) containing 0.2 M LiNO3. The electrochemical impedance spectroscopy (EIS) in the range of 100 kHz and 10 mHz and the cyclic voltammetry (CV) with a scan rate of 0.1 mV s−<sup>1</sup> from 1.7 to 2.8 V were conducted on an electrochemical work station (CHI760E). The galvanostatic charging-discharging (GCD) tests were carried out on Neware battery testing system (CT-4008) with different current densities of 0.2, 0.5, 1.0, 1.5, 2.0, and 5.0 C.

### RESULTS AND DISCUSSIONS

The synthesis procedure of S@NC-NSs is illustrated in **Figure 1A**. The ZnS NSs with a diameter of about 3–5µm (**Figure 1B**), consisting of 2D nanosheets, were fabricated via a hydrothermal route. The XRD pattern of the as-obtained product (**Figure S1**) can be indexed into the wurtzite phase

ZnS (JCPDS Card No. 36-1450). The high resolution TEM (HR-TEM) image in **Figure S2A** indicates that the thickness of the ZnS NSs is 15–20 nm and discloses a lattice spacing of 0.31 nm corresponding to the (002) plane of wurtzite phase ZnS (Yao et al., 2005). Subsequently, a thin layer of amorphous C is uniformly coated on the surface of the ZnS NSs after the polymerization and annealing strategy. The ZnS@NC-NSs still maintain the flower-like morphology as can be seen from **Figure 1C**. During the chemical polymerization, PDA can adhere to the surface of the ZnS NSs, and further converted into the NC via annealing in N2. The HR-TEM image of the annealed product shows a typical core-shell structure with a crystal core and amorphous shell of 10–15 nm (**Figure S2B**). All XRD peaks of the carbon coated sample can also be assigned to the wurtzite phase ZnS (**Figure S3A**). And the XRD pattern (**Figure S3B**) and Raman spectra (**Figure S3C**) of the NC shell suggest that the obtained carbon shell is amorphous and has many defects. To further obtain the in situ formed S, the ZnS core of the ZnS@NC-NSs was oxidized to S by FeCl<sup>3</sup> solution according to the reaction of ZnS (s) + 2Fe3<sup>+</sup> (aq.) = Zn2<sup>+</sup> (aq.) + S (s) + 2Fe2<sup>+</sup> (aq.) (Ding et al., 2015; Ma et al., 2017). After oxidization, all diffraction patterns of the sample are corresponded to the orthorhombic S (JCPDS Card No. 08-0247) and no additional ZnS peaks can be observed (**Figure S4**). Meanwhile, the SEM image in **Figure 1D** indicates that S@NC-NSs can well preserve the morphology of flowers assembled by nanosheets, which is inherited from that of ZnS and ZnS@NC. Such a microflower structure is further confirmed in **Figure 2A**. The elemental mappings displayed in **Figure 2B** indicate that C, N, and S are uniformly distributed in the micro-sized flowers. The TEM image of the S@NC-NSs (**Figure 2C**) discloses that the nanosheets are hollow double-shell structures with a thickness of 25–40 nm, which provide enough void for the volume expansion of sulfur upon the discharge process. EDS line scan of single S@NC-NSs demonstrates that S mainly exists in the inner wall of the hollow NC, which suggests that the in situ formed S is totally encapsulated and adhered to the inner surface of NC shell, as shown in **Figure 2D**. The mass loading of S in the S@NC-NSs can be optimized by varying the polymerization time. That is, the S@NC-NSs obtained at different polymerization time of 0.5, 1.0, and 2.0 h own different S mass loading of 65, 46.5 and 40.5% (**Figure S5A**), respectively. Among them, the sample polymerized 1.0 h shows the highest capacity and best cycling stability (**Figure S5B**), which is characterized in following study.

The X-ray photoelectron spectroscopy (XPS) is used to further analyze the chemical state of the S@NC-NSs. As shown in **Figure 3A**, the S@NC-NSs contain C, N, O, and S and no observable Zn signal is presented. The C 1s spectrum (**Figure 3B**) can be devolved into three peaks centered at 284.6, 285.6, and 289.2 eV, corresponding to C–C/C=C (Li et al., 2016), C–S/C–N (Wang Z et al., 2014), and O–C=O (Yang et al., 2016), respectively. It is obvious that there are functional groups containing N/O in NC shell. The N 1s (**Figure 3C**) could be ascribed to three chemical states: pyridinic N (398.1 eV; Sun et al., 2012), pyrrolic N (400.1 eV; Peng et al., 2016), and quaternary N (400.8 eV; Cai et al., 2017), confirming the effective nitrogen doping of PDA derived carbon. The N doped carbon is beneficial for improving conductivity and wettability (Qiu et al., 2014; Chen et al., 2015). Moreover, in **Figure 3D**, 163.7 and 164.9 eV are corresponding to S–S/S–C bonds (Zhou et al., 2015b). These functional groups are reported to improve the chemical adsorption ability for LiPSs (See et al., 2014; Song et al., 2014; Chen et al., 2018).

In order to measure the pore structures of the NC shell, the S@NC-NSs were washed in CS<sup>2</sup> solution to remove the inner S and obtain the pure NC. The pore-sized distribution

and the nitrogen adsorption-desorption isotherms of the pure NC are carried out and displayed in **Figure S6**. The NC shell with a BET surface area of 354.6 m<sup>2</sup> g −1 is mainly consisted of micropores, which is benefit for physically trapping LiPSs. To further evaluate the LiPSs trapping ability of the NC shell, the absorption experiment in orange Li2S<sup>6</sup> solution was conducted. After adding 25 mg of NC shell into 5 mM Li2S<sup>6</sup> solution (4 mL), the orange LiPSs solution turns colorless after 5.0 h (**Figure 4A**). Moreover, according to the UV–Vis spectra, the characteristic adsorption peaks at 310 (S2<sup>−</sup> 6 /S2<sup>−</sup> 4 ) and 410 nm (S2<sup>−</sup> 4 ; (Xiao et al., 2015; Li X. et al., 2017)) have been eliminated after adding the NC shell (**Figure 4B**). Such results imply the strong adsorption capability of NC shell for LiPSs. This excellent adsorption performance is attributed to the physical trapping of the micropores below 2 nm (**Figure S6**) and chemical immobilization effect of the NC with fruitful O and N containing functional groups (Borchardt et al., 2016; Kang et al., 2016).

To illustrate the remarkable Li storage performance of S@NC-NSs, the coin cell type LSBs were fabricated. And the commercial S was used as the control sample. The cyclic voltammogram (CV) curves of S@NC-NSs and commercial S cathode between 1.7 and 2.8 V at a scan rate of 0.1 mV s−<sup>1</sup> are presented in **Figure 5A**. In the cathodic process, the two reduction peaks between 1.98 and 2.30 V can be found in both samples, which are assigned to the formation of soluble long-chain Li2S<sup>n</sup> (4 ≤ n ≤ 8) intermediates and solid lithium sulfide (Li2S<sup>2</sup> or Li2S). In the subsequent anodic process, the CV curve of S@NC-NSs shows two individual peaks at 2.37 and 2.43 V on virtue of the reactions from lithium sulfide to Li2S<sup>n</sup> (4 ≤ n ≤ 8) and S, while the commercial S only shows a broaden peak at 2.42 V because of the large polarization. Compared to the commercial S electrode, it is clearly seen that the oxidation and reduction peaks of the S@NC-NSs electrode are sharper and shift toward the quasi-equilibrium potential, revealing lower polarization and higher reaction kinetics. **Figure 5B** presents charge-discharge curves of the S@NC-NSs electrode at 0.2, 0.5, 1.0, 1.5, 2.0, and 5.0 C. Two typical plateaus of the S@NC-NSs cathode can be observed at 0.2 C in discharge process delivering a high capacity of 1,238 mAh g−<sup>1</sup> . Even at a high current density of 5.0 C, a remarkable plateau at 1.9 V still exists due to the rapid kinetics of the S@NC-NSs electrode. The capacity retention of the S@NC-NSs cathode is 57.8% when the current density increases 25 times from 0.2 to 5.0 C, which is much better than that of the commercial S electrode, as shown in **Figure 5C**. The capacity and rate performance of S@NC-NSs are competitive among recently reported S/C composite electrodes (Xia et al., 2017; Zhu et al., 2017; Wang X et al., 2018; Yao et al., 2018). Furthermore, once the rate is reverted back to 0.5 C, the specific capacity of 920 mAh g−<sup>1</sup> is obtained, revealing its excellent reversibility and

rate performance. **Figure 5D** displays the cyclability of S@NC-NSs and commercial sulfur cathodes at the current density of 0.2 C. Although the S@NC-NSs cathode has an obvious capacity decay at first three cycles, afterwards, it delivers a high and stable capacity of 1,025 mAh g−<sup>1</sup> over 50 cycles, which is far exceeding that of pure S. Even at a high current density of 5.0 C (**Figure 5E**), the S@NC-NSs cathode achieves a high capacity of 600 mAh g−<sup>1</sup> over 200 cycles with a low decay rate of 0.064%. Furthermore, the Nyquist plots of commercial S and S@NC-NSs cathodes after 200 cycles at 5.0 C between 100 kHz and 0.01 Hz are shown in **Figure S7**. The equivalent impedance of the S@NC-NSs cathode is much smaller than the commercial S cathode, indicating a faster electrons/ions transportation because of special structures of S@NC-NSs assembled by micro-sized flowers with high conductivity. The outstanding cycling stability of S@NC-NSs cathodes can be attributed to dual functions of NC shell with both physical trapping and chemical immobilization of LiPSs.

### CONCLUSION

In summary, a dual-shell and hollow S@NC-NSs assembled by micro-sized flowers are fabricated for LSBs via in situ oxidization process. The in situ generated S is encapsulated and adheres to the inner wall of the NC shell with abundant micropores and fruitful N or O containing function groups,

which offers both physical trapping and chemical tethering to eliminate the shuttle effect of LiPSs. Moreover, the conjoint hollow NC-NSs also provide high conductive channels for electron transport and enough space for volumetric expansion of S. The S mass loading of S@NC-NSs can be easily adjusted via tuning the thickness of the carbon shell. As the Li-S battery cathode, the S@NC-NSs achieve a high capacity of 1,238 mAh g−<sup>1</sup> at 0.2 C and outstanding rate performance with capacity retention of the 57.8% when the current density increased 25 times from 0.2 to 5.0 C. Importantly, S@NC-NSs demonstrate the excellent stability with a high capacity of 600 mAh g−<sup>1</sup> and an ultraslow capacity decay rate of 0.064% after 200 cycles at a high current density of 5.0 C. With high conductivity, efficient physical and chemical immobilization as well as adequate inner space, the NC encapsulated in situ formed S cathode with outstanding electrochemical performance can bode well for promising application in LSBs.

### AUTHOR CONTRIBUTIONS

ZG carried out the experiment and wrote the paper. BG and XF supervised this research. XL, HS, KD, and XH gave a lot of help for analyzing data. XZ, XP and JF helped writing.

### ACKNOWLEDGMENTS

This work was financially supported by National Natural Science Foundation of China (No. 51572100 and 51504171) and Major project of Technology Innovation of Hubei Province (2018AAA011).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2018.00429/full#supplementary-material

## REFERENCES


discharge products on amino-functionalized reduced graphene oxide. Nat. Commun. 5:5002. doi: 10.1038/ncomms6002


high performance lithium-ion batteries. Energy Environ. Sci. 11, 669–681. doi: 10.1039/C8EE00239H


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Guo, Feng, Li, Zhang, Peng, Song, Fu, Ding, Huang and Gao. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Preparation of ZnFe2O4/α-Fe2O<sup>3</sup> Nanocomposites From Sulfuric Acid Leaching Liquor of Jarosite Residue and Their Application in Lithium-Ion Batteries

Jinhuan Yao, Jing Yan, Yu Huang, Yanwei Li\*, Shunhua Xiao and Jianrong Xiao\*

*Guangxi Key Laboratory of Electrochemical and Magneto-Chemical Functional Materials, College of Chemistry and Bioengineering, Guilin University of Technology, Guilin, China*

#### Edited by:

*Qiaobao Zhang, Xiamen University, China*

#### Reviewed by:

*Ting-Feng Yi, Northeast University at Qinhuangdao Campus, China Xiao-Dong Zhu, Harbin Institute of Technology, China Lingjun Li, Changsha University of Science and Technology, China*

#### \*Correspondence:

*Yanwei Li lywhit@126.com Jianrong Xiao xjr@glut.edu.cn*

#### Specialty section:

*This article was submitted to Physical Chemistry and Chemical Physics, a section of the journal Frontiers in Chemistry*

Received: *10 July 2018* Accepted: *05 September 2018* Published: *25 September 2018*

#### Citation:

*Yao J, Yan J, Huang Y, Li Y, Xiao S and Xiao J (2018) Preparation of ZnFe2O4/*α*-Fe2O3 Nanocomposites From Sulfuric Acid Leaching Liquor of Jarosite Residue and Their Application in Lithium-Ion Batteries. Front. Chem. 6:442. doi: 10.3389/fchem.2018.00442* Recycling Zn and Fe from jarosite residue to produce high value-added products is of great importance to the healthy and sustainable development of zinc industry. In this work, we reported the preparation of ZnFe2O4/α-Fe2O<sup>3</sup> nanocomposites from the leaching liquor of jarosite residue by a facile chemical coprecipitation method followed by heat treatment at 800◦C in air. The microstructure of the as-prepared ZnFe2O4/α-Fe2O<sup>3</sup> nanocomposites were characterized by X-ray diffraction (XRD), Mössbauer spectroscopy, scanning transmission electron microscope (STEM), and X-ray photoelectron spectrum (XPS). The results demonstrated that the ZnFe2O4/α-Fe2O<sup>3</sup> composites are composed of interconnected ZnFe2O<sup>4</sup> and α-Fe2O<sup>3</sup> nanocrystals with sizes in the range of 20–40 nm. When evaluated as anode material for Li-ion batteries, the ZnFe2O4/α-Fe2O<sup>3</sup> nanocomposites exhibits high lithium storage activity, superior cyclic stability, and good high rate capability. Cyclic voltammetry analysis reveals that surface pseudocapacitive lithium storage has a significant contribution to the total stored charge of the ZnFe2O4/α-Fe2O3, which accounts for the enhanced lithium storage performance during cycling. The synthesis of ZnFe2O4/α-Fe2O<sup>3</sup> nanocomposites from the leaching liquor of jarosite residue and its successful application in lithium-ion batteries open up new avenues in the fields of healthy and sustainable development of industries.

Keywords: lithium-ion batteries, ZnFe2O4 /α-Fe2O<sup>3</sup> composites, anode materials, jarosite residue, chemical coprecipitation method

### INTRODUCTION

As one of the most promising energy storage systems, rechargeable lithium-ion batteries (LIBs) have been widely applied in portable electronic devices, electric vehicles, and smart grids (Dunn et al., 2011; Scrosati et al., 2011; Larcher and Tarascon, 2014). To meet the ever-increasing requirements of high power density and high energy density, developing new electrode materials with high specific capacity and high rate capability is very crucial for manufacturing the next-generation LIBs (Tang et al., 2015; Yi et al., 2016; Zhu and Yi, 2016; Massé et al., 2017; Yao et al., 2018a). The commercial graphite anode material for LIBs suffers from low theoretical capacity (about 372 mAh g−<sup>1</sup> ), poor rate performance, and serious safety issues,

**236**

and therefore cannot fulfill demands for LIBs with high energy density, high power density, and good safety in operation (Goriparti et al., 2014; Yi et al., 2015; Zhang et al., 2018). Over recent years, transition metal oxides (TMOs) have attracted extensive attention for their high theoretical specific capacities (600–1,200 mAh g−<sup>1</sup> ) and good safety as anode materials for LIBs (Reddy et al., 2013; Nitta and Yushin, 2014; Yuan et al., 2014; Pan et al., 2015b; Zhu et al., 2015). Among the various TMOs, Fe-based oxides have been widely studied due to their abundance, low cost, safety, wide availability, and environmental friendliness (Zhang et al., 2014; Pan et al., 2015a; Xu et al., 2015; Yao et al., 2018c; Zheng Z. M. et al., 2018). In particular, ZnFe2O<sup>4</sup> and Fe2O<sup>3</sup> stand out from the Fe-based oxides because of their high theoretical capacities (1,072 mAh g−<sup>1</sup> for ZnFe2O<sup>4</sup> and 1,007 mA h g−<sup>1</sup> for Fe2O3). However, the poor power capability, and fast capacity fading owning to the huge volume changes during discharge/charge cycles for ZnFe2O<sup>4</sup> and Fe2O3, severely hinders their practical applications in LIBs (Reddy et al., 2013; Zhang S. L. et al., 2017; Yao et al., 2018b). Constructing appropriate nanostructures has been demonstrated to be an effective way to improve the electrochemical performance of electrode active materials (Zhang et al., 2013; Yu et al., 2018). Various nanostructured ZnFe2O<sup>4</sup> and Fe2O<sup>3</sup> materials have been fabricated using diverse methods, such as sol-gel method (Thankachan et al., 2015), hydrothermal synthesis (Lin and Pan, 2015; Li L. L. et al., 2017; Zheng Z. M. et al., 2018), solvothermal synthesis (Lu et al., 2013; Yang et al., 2017), high energy ball-milling, electrospinning method (Wang C. L. et al., 2017), and reactive pulsed laser deposition method (NuLi et al., 2004). Although improved electrochemical performances have been achieved, most of the reported methods are generally timeconsuming and involving complicated steps, high cost due to high energy consumption and expensive raw materials, and difficult to scale up, which greatly restrict the development and practical application of nanostructured ZnFe2O<sup>4</sup> and Fe2O<sup>3</sup> materials in LIBs. Therefore, it is still a big challenge but urgent demand to pursue a facile, efficient, and inexpensive method for massive production of nanostructured Fe-based oxide electrode materials with high performance for nextgeneration LIBs. Recycling waste materials can ease the resource crisis, reduce environmental pollution, and create new economic values, which has a great practical significance for the healthy and sustainable improvement of the human living environment. Huge quantities of jarosite residue was produced from traditional zinc hydrometallurgy process in the world. Most of the jarosite residue was stored up, which takes massive land and brings forth great risk of environmental pollution (Ju et al., 2011). The leaching liquor (containing Zn and Fe) of jarosite residue can be directly used to prepare high value-added Fe-based oxides functional materials, which has not been reported until now.

Herein, we reported the use of leaching liquor of jarosite residue to prepare ZnFe2O4/α-Fe2O<sup>3</sup> nanocomposites by a chemical coprecipitation method followed by heat treatment at 800◦C in air. This method offers the merits of low-cost, facile, and scalable production of nanostructured ZnFe2O4/α-Fe2O<sup>3</sup> composites. When studied as anode materials for LIBs, the as-prepared porous ZnFe2O4/α-Fe2O<sup>3</sup> nanocomposites exhibit high reversible capacity, excellent cycling stability and good high rate capability, which can be attributed to the synergetic effect between ZnFe2O<sup>4</sup> and α-Fe2O<sup>3</sup> nanoparticles and the significant pseudocapacitive behavior during discharging and charging processes.

### EXPERIMENTAL

### Preparation of ZnFe2O4/α-Fe2O<sup>3</sup> Nanocomposites

All chemicals employed in this work were of analytical reagent grade and used without further purification. The leaching liquor contains 44 mM Zn2+, 121 mM Fe2+, and a small amount of Cu2+, Ca2+, As3+, and In3+. In a typical procedure, 0.464 g ZnSO4·7H2O was dissolves in 100 mL leaching liquor so that the molar ratio of Zn2<sup>+</sup> and Fe3<sup>+</sup> in the solution is 1:2. After ultrasonic treatment for 10 min, an aqueous solution of NH4OH (200 mL, ∼2.67 M) was added dropwise into the above mixed solution under constant vigorous stirring at room temperature. After continuous stirring for 3 h, the deposit was kept in the mother solution at room temperature for 12 h and allowed to settle, then washed with deionized water, filtered, and dried in an oven at 80◦C overnight to get the precursor. Finally, the asprepared precursor was calcinated at 800◦C for 2 h in air to obtain the ZnFe2O4/α-Fe2O<sup>3</sup> sample.

### Physical Characterization

The crystal structure of the as-prepared sample was identified with a powder X-ray diffractometer (XRD, Dutch PANalytica X'Pert<sup>3</sup> powder) with Cu Kα radiation (λ = 1.5406 Å). The X-ray tube voltage and current were set at 40 kV and 40 mA, respectively. Field-emission scanning electron microscope (FESEM, Hitachi SU500) and transmission electron microscope (HRTEM, JEM-2100F) were used to analyze the microstructure of the as-prepared sample. <sup>57</sup>Fe Mössbauer spectrum was recorded by using an MFD-500A Mossbauer spectrometer (Topologic Systems, Japan). To determine the surface composition and chemical states of the sample, X-ray photoelectron spectroscopy (XPS) analysis was carried out using an ESCALAB 250 spectrometer (Perkin-Elmer), with an Al Kα source (1486.6 eV) operated at 15 kV and 150 W, at a base pressure of 2 × 10−<sup>9</sup> Torr.

### Electrochemical Measurements

The electrochemical performance of the ZnFe2O4/α-Fe2O<sup>3</sup> nanocomposites was evaluated in CR2032-type coin cells assembled in an argon-filled glovebox. The test electrodes were fabricated by mixing active materials (ZnFe2O4/α-Fe2O<sup>3</sup> nanocomposites), Super-P carbon black, and binder (polyvinylidene difluoride, PVDF) with a weight ratio of 6:3:1 in Nmethyl-2-pyrrolidone (NMP) solvent. The obtained slurry was uniformly coated onto a copper foil and then dried under at 80◦C for 12 h in a vacuum oven. The mass loading of the working electrode is ∼1.0 mg cm−<sup>1</sup> . Metallic lithium foils were used as both the counter and reference electrodes. A Celgard 2400 microporous polypropylene membrane was used as the separator. The electrolyte was 1M LiPF<sup>6</sup> solution, composing of ethylene carbonate (EC), diethyl-carbonate (DEC), and dimethyl carbonate (DMC) (1:1:1, in volume). Cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS) profiles were carried out on a CHI760E electrochemical workstation. The CVs were recorded within the potential range of 0.01–3.0 V (vs. Li/Li+) at various of scanning rates. The EIS spectra were measured in the frequency ranging from 100 mHz to 100 kHz by applying an AC amplitude of 5 mV at a fully charged state. The galvanostatic discharge/charge measurements were performed on a NEWARE battery testing system in the voltage range of 0.01–3.0 V (vs. Li/Li+). All the electrochemical tests were performed at room temperature.

## RESULTS AND DISCUSSION

### Structure Characterization

**Figure 1** shows XRD pattern of the as-prepared ZnFe2O4/α-Fe2O<sup>3</sup> nanocomposites. The diffraction peaks of the nanocomposites can be well indexed to either rhombohedra α-Fe2O<sup>3</sup> (JCPDS 87-1164) or spinel ZnFe2O<sup>4</sup> (JCPDS 82-1049).

No diffraction peaks from impurities are detected, suggesting the nanocomposites are composed of spinel ZnFe2O<sup>4</sup> phase and α-Fe2O<sup>3</sup> phase. The average crystal sizes of the ZnFe2O<sup>4</sup> phase and α-Fe2O<sup>3</sup> phase in the nanocomposites estimated using Scherrer's formula are about 30 and 44 nm, respectively. <sup>57</sup>Fe Mössbauer spectroscopy was used to determine the phase composition and metal cation redistribution in the ZnFe2O4/α-Fe2O<sup>3</sup> nanocomposites and the result is presented in **Figure 2b**. The dots in **Figure 2b** are experimental spectrum and the continuous curves are fitting lines. **Table 1** summarizes the Mössbauer refined parameters from the fitting of the spectrum. The Mössbauer spectrum of the ZnFe2O4/α-Fe2O<sup>3</sup> nanocomposites are fitted with one sextet and two doublets. The sextet can be assigned to α-Fe2O<sup>3</sup> with iron content 63% in the composites (Pailhé et al., 2008; Lazarevic et al., ´ 2014), and the doublets can be due to super paramagnetic ZnFe2O<sup>4</sup> with iron content 37% in the composites (Yao et al., 2012; Amir et al., 2018). The doublet with a lower quadrupole splitting can


be assigned to the Fe3<sup>+</sup> at tetrahedral sites in ZnFe2O<sup>4</sup> (Amir et al., 2018). The site occupation in ZnFe2O<sup>4</sup> can be represented by (Zn1−λFeλ)tet[ZnλFe2−λ]octO<sup>4</sup> where λ is the inversion parameter (0 ≤ λ ≤ 1). The calculated inversion parameter of the ZnFe2O<sup>4</sup> phase is 0.22. Thus, the chemical formula of the ZnFe2O<sup>4</sup> phase in the nanocomposites can be expressed as (Zn0.78Fe0.22) [Zn0.22Fe1.78]O4.

**Figure 2a** presents the SEM image of the as-prepared ZnFe2O4/α-Fe2O<sup>3</sup> nanocomposites. It can be seen that the sample is composed of agglomerated nanosized primary particles. More detail structure information about the ZnFe2O4/α-Fe2O<sup>3</sup> nanocomposites was obtained from STEM analyses. The low- and high-magnification STEM images (**Figures 2b,c**) show that the ZnFe2O4/α-Fe2O<sup>3</sup> nanocomposites are actually composed of interconnected nanocrystals with size in the range of 20∼50 nm, which is consistent with the result from the XRD analysis. The ZnFe2O<sup>4</sup> nanocrystals and α-Fe2O<sup>3</sup> nanocrystals could be well distinguished in the highresolution STEM image (**Figure 2c**). The selected area electron diffraction (SAED, **Figure 2d**) pattern of the ZnFe2O4/α-Fe2O<sup>3</sup> nanocomposites suggests its polycrystalline character and the distinct diffraction spots could be assigned sequentially to α-Fe2O<sup>3</sup> (012), ZnFe2O<sup>4</sup> (311), and α-Fe2O<sup>3</sup> (116) planes from the center, which also appear in the XRD pattern (**Figure 1A**). The primary nanocrystals of the ZnFe2O4/α-Fe2O<sup>3</sup> nanocomposites offer large surface area and short diffusion pathways for fast Li<sup>+</sup> diffusion; the interconnected primary nanocrystals provide

continuous electronic transfer channels in the ZnFe2O4/α-Fe2O<sup>3</sup> composites; the large amount of void spaces among the interconnected nanocrystals would benefit the penetration of electrolyte in electrode, and accommodate the strain induced by the volume change upon discharge/charge cycling. Moreover, the interconnected ZnFe2O4/α-Fe2O<sup>3</sup> heterojunctions could provide an enhanced inner electric field at the interface between ZnFe2O<sup>4</sup> and α-Fe2O<sup>3</sup> nanocrystals, which will greatly accelerate the charge-transfer kinetics during electrochemical reactions (Qiao et al., 2013; Zhao et al., 2014; Zheng et al., 2016).

The composition and element valence of the as-prepared ZnFe2O4/α-Fe2O<sup>3</sup> nanocomposites were characterized by XPS and the result are displayed in **Figure 3**. The survey XPS spectrum (**Figure 3A**) shows the presence of elements Zn, Fe, and O, as well as C, which comes from the adsorbed CO<sup>2</sup> and/or hydrocarbon contaminations.

The Fe 2p core peak spectrum shows two main peaks at 711.1 and 725.1 eV, which can be assigned to Fe3/<sup>2</sup> and Fe1/2, respectively (Guo et al., 2014). The presence of two satellite peaks at 719.5 and 733.4 eV is characteristic of the Fe3<sup>+</sup> in the ZnFe2O4/α-Fe2O<sup>3</sup> nanocomposites (Zhao et al., 2014). The Zn 2p core peak spectrum is composed of two intense peaks at 1021.4 and 1044.5 eV, which can be ascribed to Zn 2p3/<sup>2</sup> and Zn 2p1/<sup>2</sup> of Zn2+, respectively (Lu et al., 2017). The O 1s core peak spectrum consists of two peaks at 530.1 and 531.3 eV, which can be attributed to the lattice oxygen in the nanocomposites and surface-adsorbed hydrocarbon (Zhang L. H. et al., 2017).

### Electrochemical Characterization

**Figure 4A** displays the typical CV profiles of ZnFe2O4/α-Fe2O<sup>3</sup> nanocomposites electrode for the first four cycles between 0.01 and 3.0 V at a scan rate of 0.1 mV/s. During the first cycle, two intense reduction peaks can be observed at about 0.70 and 0.81 V, which can be associated with the inital reduction of ZnFe2O<sup>4</sup> to metallic Fe<sup>0</sup> /Zn<sup>0</sup> and the complete reduction of α-Fe2O<sup>3</sup> to metallic Fe<sup>0</sup> , as well as the formation of amorphous Li2O matrix and solid electrolyte interface (SEI) film (Zhang Y. M. et al., 2017); the broad oxidation peak at about 1.63 V can be attributed to the oxidation of the metallic Zn<sup>0</sup> and Fe<sup>0</sup> to to Zn2<sup>+</sup> and Fe3+, respectively (Yao et al., 2018c). In the subsequent cycles, the two reduction peaks merges together and

shifts to 0.97 V, which can be due to the drastic lithium-driven structural and/or textural modifications on the electrode during the first lithiation process; the oxidation peak slightly shifts to 1.69 V, which is can be ascribed to structure rearrangement after the first lithiation/delithiation process (Zhang Y. M. et al., 2017; Zhou et al., 2017). The well-overlapped CV profiles of the third and the fourth cycle suggest the good electrochemical reaction reversibility of the nanocomposites electrode. **Figure 4B** gives the cycling performance of the ZnFe2O4/α-Fe2O<sup>3</sup> nanocomposites electrode at a current density of 500 mA g−<sup>1</sup> . The initial discharge and charge capacities are 1,522 and 1,027 mAh g−<sup>1</sup> , with a coulombic efficiency of 67.5%. This large irreversible capacity loss during the first cycles is mainly caused by the irreversible reaction and formation of the SEI layer on the surface of electroactive materials (Reddy et al., 2013). The electrode exhibits a rapid capacity decay in the first 20 cycles and relatively slow capacity decay from the 20 to 80th cycle, which can be ascribed to the mechanical degradation and unstable SEI formation upon discharge/charge cycling (Sun et al., 2014). After 80 cycles, the discharge capacity increases gradually and gets 1,206 mAh g−<sup>1</sup> in the 400th cycle. The rise of the discharge capacity may result from the synergistic combination of the refinement of nanoparticles and the formation of organic polymeric/gel-like layer by electrolyte decomposition with the increase of the

September 2018 | Volume 6 | Article 442


number of cycles (Hassan et al., 2011; Rai et al., 2014; Wang M. Y. et al., 2017). **Figure 4C** presents the typical discharge/charge curves of the nanocomposites electrode in different cycles at a current density of 500 mA g−<sup>1</sup> . After the first discharge/charge cycle, the long discharge plateau change into a slope. The change of the voltage plateaus and the large capacity loss after the first cycle match well with the CV profiles (**Figure 4A**). Moreover, the voltage hysteresis between discharge and charge profiles enlarges in the first 100 cycles, and then reduces gradually, implying that the electrode state and polarization degree vary with discharge/charge cycles. **Figure 4D** displays the Nyquist plots of the nanocomposites electrodes recorded before cycling and after selected discharge/charge cycles. Each plot consists of one depressed semicircle or two semicircles in the high to moderate frequency region and an inclined line in the low frequency region. The depressed semicircles can be assigned to the resistance of Li<sup>+</sup> migrating through the SEI film and the charge-transfer resistance at electrode/electrolyte interface (denoted as Rsf+ct), and the inclined line represents the Warburg impedance related to Li<sup>+</sup> diffusion process into the bulk of the electrode (Wang et al., 2015; Li et al., 2016; Kong et al., 2017; Lee et al., 2017). Before cycling, the Warburg straight line is almost vertical to the real-axis (capacitive behavior), suggesting that there is almost no detectable Li<sup>+</sup> intercalation in the electrode for fresh cell. With the number of cycles increasing from 1 to 100, the Warburg straight line gradually decreases to an angle of ∼45◦ to the real-axis, showing the characteristic of Li<sup>+</sup> diffusion in the electrode; however, further increasing the number of cycles from 100 to 400, the Warburg straight line gets more and more steep, presenting the characteristic of pseudocapacitive behavior, which may derive from the reversible formation of organic polymeric/gel-like layer (Laruelle et al., 2002). The Nyquist plots were fitted by the equivalent circuits shown as inset in **Figure 4E**. In the circuits, the Re, Rsf, and Rct are electrolyte resistance, SEI layer resistance, and charge transfer resistance, respectively. CPE1 and CPE2 are the corresponding capacitances of Rsf and Rct. W is the Warburg impedance. The calculated Rsf+ct values of the electrodes before and after different discharge/charge circles is presented in **Figure 4E**. Before cycling, the Rsf+ct value of the electrode is 40 . During the initial 100 cycles, the Rsf+ct value increases from 51 to 195 ; further increasing the number of cycles increase from 100 cycles to 300 and 400 cycles, the Rsf+ct value decreases from 195 to 80 and 77 , respectively. The EIS results reveals that the decrease in capacity during the initial 80 cycles can be due to the slow Li<sup>+</sup> diffusion process and large resistances from SEI film and charge-transfer process (high polarization); the increasing capacity in the following cycles from 80 to 400 cycles ca be explained by the enhanced Li<sup>+</sup> diffusion process and reduced resistances from SEI film and charge-transfer process (low polarization).

The rate performance of the of ZnFe2O4/α-Fe2O<sup>3</sup> nanocomposites electrode was tested after the electrode is activated at 500 mA g−<sup>1</sup> for 250 cycles, and the result are shown in **Figure 5**. With the increase of current density, the discharge capacity decreases gradually. Even at the very high current density of 5,000 mA g−<sup>1</sup> , the specific capacity (535 mAh g−<sup>1</sup> ) still obviously higher than the theoretical capacity (372 mAh g−<sup>1</sup> ) of graphite. When the current density reverses back to 500 mA g −1 , the discharge capacity almost recovers to the original values, implying the good capacity retention performance of the electrode.

**Figure 6a** gives long-term cycling performance of the ZnFe2O4/α-Fe2O<sup>3</sup> nanocomposites electrode at current density of 1,000 mA g−<sup>1</sup> for 900 cycles. It can be seen that the electrode exhibits excellent long-term cycling performance with a high capacity of about 1,000 mAh g−<sup>1</sup> after 900 cycles at the constant current density of 1,000 mA g−<sup>1</sup> . Compared with previous work, the as-prepared ZnFe2O4/α-Fe2O<sup>3</sup> nanocomposites deliver better electrochemical performance in terms of reversible discharge capacities and cycling stability (**Table 2**). **Figures 6b–e** presents the SEM images of the ZnFe2O4/α-Fe2O<sup>3</sup> nanocomposites electrode before and after 900 cycles. To identify ZnFe2O4/α-Fe2O<sup>3</sup> nanocomposites more clearly in the electrode, backscattered electron images are also provided. It can be seen that the as-prepared electrode shows the presence of ZnFe2O4/α-Fe2O<sup>3</sup> nanocomposites constructed by interconnected primary nanocrystals before cycling (**Figures 6b,c**). After 900 cycles, the electrodes almost maintains the original morphology (**Figures 6b,c**); the ZnFe2O4/α-Fe2O<sup>3</sup> nanocomposites can be clearly observed in the electrode and the nanocrystals are well interconnected with each other, implying the ZnFe2O4/α-Fe2O<sup>3</sup> nanocomposites are flexible to alleviate volume expansion upon repeated discharge/charge cycles.

To better understand the superior lithium storage performance of the ZnFe2O4/α-Fe2O<sup>3</sup> nanocomposites, we analyze the charge storage mechanism by the sweep voltammetry method proposed by Dunn et al. (Qu et al., 2017). In this method, the contributions from capacitive effect and diffusioncontrolled Li<sup>+</sup> process can be quantified by the following equations:

$$i(V) = k\_1 \nu + k\_2 \nu^{1/2} \tag{1}$$

$$i(V)/\nu^{1/2} = k\_1 \nu^2 + k\_2 \tag{2}$$

where I(V) and ν represent the total current response at a given potential V and scan rate for the CV measurements; k1ν and k2ν 1/2 represent the current due to surface capacitive effects and current due to diffusion-controlled reaction process, respectively. By determining k<sup>1</sup> and k2, the currents arising from capacitive effect and diffusion-controlled Li<sup>+</sup> process can be distinguished. **Figure 7A** gives the series of CV curves of ZnFe2O4/α-Fe2O<sup>3</sup> nanocomposites electrode recorded at different scan rate. **Figures 7B–D** illustrate the typical voltage profiles for the capacitive current (blue shaded region) in comparison with the total current obtained at the scan rate of 0.1, 1.0, and 2.0 mV s−<sup>1</sup> , respectively. Obviously, capacitive charge storage contributes a significant proportion to the total capacity, in particular in the low potential region (0.8–0.01 V) during delithiation process. With the increase of scan rate, the portion from capacitive capacity increases dramatically. As shown in **Figure 7F**, the capacitive capacity makes up about 29% of the total capacity at the scan rate of 0.1 mV s−<sup>1</sup> , whereas this value increases to 56% and 64% at the scan rate of 1 and 2 mV s −1 , respectively. Similar results have also been reported for the nano-sized NiO and Ni(OH)<sup>2</sup> anode materials (Li Y. W. et al., 2017b; Zheng Y. Y. et al., 2018). This significant surface or near surface charge storage due to capacitive behavior benefits the high rate capability and cycling stability of electrode active materials (Rauda et al., 2013; Augustyn et al., 2014; Li Y. W. et al., 2017a).

The superior lithium storage performance of the ZnFe2O4/α-Fe2O<sup>3</sup> nanocomposites can be ascribed to following several aspects: (1) the primary nanocrystals facilitate the transport of both Li<sup>+</sup> and electrons because of the short diffusion distance, which enhances the kinetic performance; (2) the numerous void spaces among the interconnected primary nanoparticles and among the nanoparticles can accommodate the strain induced by the volume change during discharge/charge cycles, and therefore improve the cycling performance; (3) the unique ZnFe2O4/α-Fe2O<sup>3</sup> heterojunctions provides an enhanced inner electric field at the interface between ZnFe2O<sup>4</sup> and α-Fe2O<sup>3</sup> nanocrystals, which may efficiently accelerate the charge-transfer kinetics during electrochemical reactions and boost the rate capability; (4) the significant pseudocapacitive behavior during discharge/charge process is also an important reason for the outstanding high rate capability and long-term cycling stability.

### CONCLUSIONS

Hybrid ZnFe2O4/α-Fe2O<sup>3</sup> nanocomposites have been successfully fabricated with the leaching liquor of jarosite residue as raw material by a facile chemical coprecipitation method followed by heat treatment in air. The ZnFe2O4/α-Fe2O<sup>3</sup>

nanocomposites are composed of interconnected ZnFe2O<sup>4</sup> and α-Fe2O<sup>3</sup> nanocrystals with sizes in the range of 20–40 nm. Due to the unique heterojunction nanostructure, the ZnFe2O4/α-Fe2O<sup>3</sup> nanocomposites exhibits high lithium storage activity, superior cyclic stability, and good high rate capability when evaluated as anode materials for lithium-ion batteries. The reversible capacity of 1,000 mAh g−<sup>1</sup> is achieved over 900 cycles at the constant current density of 1,000 mA g−<sup>1</sup> ; even at the high current density of 5,000 mA g−<sup>1</sup> , the specific discharge capacity of 535 mAh g −1 can be obtained, which is still significantly higher than the theoretical capacity (372 mAh g−<sup>1</sup> ) of graphite. Charge storage mechanism analysis demonstrates that surface pseudocapacitive lithium storage has a significant contribution to the total stored charge of the ZnFe2O4/α-Fe2O<sup>3</sup> nanocomposites, which accounts for the enhanced lithium storage performance during cycling. This work provides a facile, efficient, and low-cost method for the synthesis of high-performance Fe-based oxides anode materials by utilizing the leaching liquor of jarosite residue as raw material, which can make use of the industrial jarosite residue as resource, reduce the environment pollution, create

### REFERENCES


high value-added products, and will achieve both good social and economic benefits.

### AUTHOR CONTRIBUTIONS

JinhY and YL conceived the idea. JingY prepared all materials and performed electrochemical characterizations. YH conducted SEM experiments. YL, JingY, JinhY, JX, and SX analyzed the data. JinhY and JingY wrote the manuscript. YL and JX commented on it. JinhY supervised the implementation of the project.

### ACKNOWLEDGMENTS

The authors thank the financial supports from the National Natural Science Foundation of China (No. 51464009 and 51664012), Guangxi Natural Science Foundation of China (2017GXNSFAA198117 and 2015GXNSFGA139006), and Guangxi Key Laboratory of Electrochemical and Magnetochemical Functional Materials (EMFM20181102/ EMFM20181117).


high-performance anode for lithium ion batteries. Chem. Eng. J. 347, 563–573. doi: 10.1016/j.cej.2018.04.119


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Yao, Yan, Huang, Li, Xiao and Xiao. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Boosting Lithium-Ion Storage Capability in CuO Nanosheets via Synergistic Engineering of Defects and Pores

Zhao Deng<sup>1</sup> , Zhiyuan Ma<sup>1</sup> , Yanhui Li <sup>1</sup> , Yu Li <sup>1</sup> , Lihua Chen<sup>1</sup> , Xiaoyu Yang<sup>1</sup> , Hong-En Wang<sup>1</sup> \* and Bao-Lian Su1,2

*<sup>1</sup> State Key Laboratory of Advanced Technology for Materials Synthesis and Processing, Wuhan University of Technology, Wuhan, China, <sup>2</sup> Laboratory of Inorganic Materials Chemistry, University of Namur, Namur, Belgium*

CuO is a promising anode material for lithium-ion batteries due to its high theoretical capacity, low cost, and non-toxicity. However, its practical application has been plagued by low conductivity and poor cyclability. Herein, we report the facile synthesis of porous defective CuO nanosheets by a simple wet-chemical route paired with controlled annealing. The sample obtained after mild heat treatment (300◦C) exhibits an improved crystallinity with low dislocation density and preserved porous structure, manifesting superior Li-ion storage capability with high capacity (∼500 mAh/g at 0.2 C), excellent rate (175 mAh/g at 2 C), and cyclability (258 mAh/g after 500 cycles at 0.5 C). The enhanced electrochemical performance can be ascribed to the synergy of porous nanosheet morphology and improved crystallinity: (1) porous morphology endows the material a large contact interface for electrolyte impregnation, enriched active sites for Li-ion uptake/release, more room for accommodation of repeated volume variation during lithiation/de-lithiation. (2) the improved crystallinity with reduced edge dislocations can boost the electrical conduction, reducing polarization during charge/discharge. The proposed strategy based on synergic pore and defect engineering can pave the way for development of advanced metal oxides-based electrodes for (beyond) Li-ion batteries.

#### Keywords: copper oxides, porous nanosheets, crystal engineering, anode, lithium ion batteries

### INTRODUCTION

Lithium-ion batteries (LIBs) have been widely used to power various portable electronic devices since their commercialization (Masse et al., 2017). Now their performances are further improved for targeted use in (hybrid) electric vehicles and smart electric grids. Such emerging applications pose strict requirements on their energy/power density, cost, calendar life and safety, which can hardly be fully satisfied by current anode (mainly graphite) and cathode [LiCoO<sup>2</sup> (Chai et al., 2017), LiMn2O<sup>4</sup> (Xi et al., 2012), and LiFePO<sup>4</sup> (Lu et al., 2011) etc.]. New anode and cathode materials have been testified to provide higher capacity and better cyclability as well as safety (Cai et al., 2016; Wu et al., 2016, 2017; Tan et al., 2017; Wang et al., 2017; Zhang Q. B. et al., 2018; Zheng et al., 2018). Meantime, new electrochemical energy storage devices, including sodium-ion batteries (Wu et al., 2016; Hu et al., 2017; Zhu et al., 2018), lithium-sulfur batteries (Li et al., 2018), and lithium-air batteries (Liu et al., 2016a,b; Wu et al., 2017) have also been proposed for large-scale

#### Edited by:

*Qiaobao Zhang, Xiamen University, China*

### Reviewed by:

*Hongkang Wang, Xi'an Jiaotong University, China Qiulong Wei, University of California, Los Angeles, United States Zhouguang Lu, Southern University of Science and Technology, China*

\*Correspondence:

*Hong-En Wang hongenwang@whut.edu.cn*

#### Specialty section:

*This article was submitted to Nanoscience, a section of the journal Frontiers in Chemistry*

Received: *22 July 2018* Accepted: *30 August 2018* Published: *24 September 2018*

#### Citation:

*Deng Z, Ma Z, Li Y, Li Y, Chen L, Yang X, Wang H-E and Su B-L (2018) Boosting Lithium-Ion Storage Capability in CuO Nanosheets via Synergistic Engineering of Defects and Pores. Front. Chem. 6:428. doi: 10.3389/fchem.2018.00428* applications in electric grid and storage of renewable energies. Among various new anode candidates for LIBs, copper oxide (CuO) has been considered to be promising because of its high theoretical capacity (670 mAh g−<sup>1</sup> ), non-toxicity, low cost, and environmental friendliness (Zhang et al., 2014). However, the practical application of CuO has been plagued by its low conductivity and large volume change during continuous Li-ion insertion/extraction, giving rise to poor rate capacity and cycling stability.

To address these issues, various nanostructured CuO, such as nanoparticles (Hu et al., 2017), nanowires/nanorods/nanotubes (Yin D. M. et al., 2017; Yuan et al., 2017; Sun et al., 2018; Wang et al., 2018), porous and hollow nanostructures (Lyu et al., 2017; Peng et al., 2017; Tan et al., 2017; Yin H. et al., 2017; Zhang J. et al., 2018; Zhou et al., 2018), has been designed and synthesized to enlarge the electrode/electrolyte interface and shorten Li-ion diffusion paths. Particularly, porous ultrathin CuO nanosheets manifest combined merits of large exposed surface and antiaggregation character of nanoparticles, enabling superior Li-ion storage capability. Additionally, coating or hybridization of CuO with conductive agents [e.g., carbon nanotubes (Cui et al., 2018), graphene (Chen et al., 2017; Li et al., 2017), amorphous carbon (Peng et al., 2017; Tan et al., 2017; Yin H. et al., 2017; Yuan et al., 2017] has been used to boost the electron transport within CuO-based composite electrode. However, the introduction of secondary additives may also bring some disadvantages. (1) It can complicate the whole synthetic procedure and thus increase processing cost. (2) The existence of non-uniform distribution of CuO and conductive component may impedance the mass transfer of electrolyte and Li-ion transport within the composite, leading to increased polarization. (3) The introduction of carbon may raise safety concerns due to the formation of Li-dendrites on carbon surface at high-rate charge/discharge and lower the volumetric energy density of the LIBs.

Besides surface conductive coating or compositing, crystal engineering is another promise way to boost the conductivity of CuO by manipulation of its crystallinity and crystallographic defects. For example, it has been reported that creating oxygen vacancy defects in TiO<sup>2</sup> can significantly boost its sodium-ion storage property (Zhao et al., 2018). Generally, CuO synthesized by wet-chemical routes can possess relatively high surface area by avoiding undesired sintering at high temperatures. However, they may also contain a large number of structural disorder (e.g., dislocation), surface functional groups and foreign ligands, lowering electrical conduction. Rational engineering the defect content while retaining relatively high exposed surface can significantly promote the electron transport and Li-ion insertion/extraction, thus enabling enhanced electrochemical properties.

Inspired by these, in this paper we report the rational design and facile synthesis of porous ultrathin CuO nanosheets on a large scale by a mild solution process. We further engineer the crystallinity and defect contents in the resultant CuO product by a simple yet effective post-annealing process. Our results reveal the solution-derived CuO subject to mild annealing at 300◦C exhibits an improved crystallinity with reduced dislocations while still maintains a high accessible surface due to the well-preserved porous nanosheet structure, which is beneficial for fast and reversible Li-ion storage even at high rates.

### EXPERIMENTAL

### Materials Synthesis

All chemicals were analytically grade and used as received without further purifications. In a typical synthesis, 10.34 g Cu(NO3)2·3H2O was dissolved in 500 mL deionized water with magnetic stirring at room temperature. After stirring for 10 min, 3.45 g NaOH was added into Cu(NO3)<sup>2</sup> solution. After reaction for 30 min, the resulting precipitate was heated at 80◦C for 12 h. The resulting black-brown product was collected by filtration, washed with deionized water and absolute ethanol several times and finally dried at 120◦C for 12 h in air.

The samples were further annealed at 300, 400, and 500◦C for 1 h, respectively with a heating rate of 1◦C /s. The corresponding products were denoted as CuO-300, CuO-400, and CuO-500, respectively for clarity.

### Characterization

X-ray diffraction (XRD) patterns were recorded on Bruker D8 diffractometer with a Cu Kα radiation (λ = 0.1542 nm). The morphology of the samples was observed using a Hitachi S-4800 scanning electron microscope (SEM). Transmission electron microscopy (TEM) and high-resolution TEM micrographs were operated on a JEM-2100F transmission electron microscope with an accelerating voltage of 200 kV. Brunauer-Emmett-Teller (BET) surface areas of the samples were derived using N<sup>2</sup> adsorption tests on a Micromeritics ASAP 2020. Before adsorption, the samples were outgassed at 120◦C overnight in a vacuum line. Pore size distribution was calculated from the desorption branch of the isotherms using the Barrett-Joyner-Halenda (BJH) method.

### Electrochemical Measurements

The working electrodes were prepared by mixing active materials, carbon black (super-P) and PVDF binder with a weight ratio of 80:10:10 in NMP to form a slurry. The resultant slurry was uniformly coated on Cu foils and then dried at 70◦C in vacuum for 12 h. 1 M solution of LiPF<sup>6</sup> dissolved in ethylene carbon (EC)/dimethyl carbonate (DMC) (1: 1 w/w) was used as electrolyte. CR2025 coin cells were assembled with lithium foils as counter and reference electrodes in an argon-filled glovebox with water and O<sup>2</sup> contents below 1 ppm. Galvanostatic chargedischarge (GCD) tests were recorded on a multichannel battery testing system (LAND CT2001A) within a potential range of 0.01∼3.0 V (vs. Li+/Li). Cyclic voltammetry (CV) measurements were carried out using a CHI 760D electrochemical workstation at a scan rate of 0.2 mV/s. Electrochemical impedance spectra (EIS) tests were recorded on an Autolab workstation. Before EIS tests, the Li-half cells were activated at 0.2 C for 5 cycles (1◦C = 670 mA/g). All electrochemical tests were carried out at room temperature.

### RESULTS AND DISCUSSION

**Figure 1** depicts the synthetic process of the CuO nanostructures. First, a large number of Cu(OH)<sup>2</sup> nuclei were produced by the precipitation reaction between Cu2<sup>+</sup> and OH<sup>−</sup> due to the low solubility of Cu(OH)2. The resultant Cu(OH)<sup>2</sup> nuclei clustered together to form some large secondary sheet-like structure during subsequent aging. After heating at 120◦C, the Cu(OH)<sup>2</sup> nanosheets were converted into porous CuO nanosheets following a quasi-topotactic reaction via dehydration. Such structure/phase transition led to formation of rich structural defects due to the incomplete crystallization at low temperature. Next, an additional annealing step was taken to further optimize the crystallinity of the CuO product. The samples prepared by solution route (heated at 120◦C), solution reaction followed by annealing at 300, 400, and 500◦C were denoted as CuO-120, CuO-300, CuO-400, and CuO-500, respectively for clarity. The relationship of structural aspects and electrochemcial property of the CuO samples were also established.

The crystal structures of the CuO products were firstly studied by XRD as shown in **Figure 2**. All the diffraction peaks of the three samples can be readily indexed to monoclinic

CuO (JCPDS card No. 48-1548) (Li et al., 2010; Liu et al., 2012; Zhu et al., 2017). The lack of diffraction peaks for other Cu-based compounds hints the high purity of the assynthesized CuO products. In addition, the full-width at halfmaximum (FWHM) values of the XRD reflections become narrower along with annealing at high temperatures, suggesting the enhanced crystallinity with increased crystallite sizes. As shown in the right panels of **Figure 2**, the average crystallite sizes of the CuO-120, CuO-300, and CuO-500 samples can be respectively estimated to be ca. 18, 22, and 31 nm by Scherrer equation based on the FWHM values of (11-1) planes.

The surface electronic states of the samples were studied by XPS measurements. **Figure 3** shows the high-resolution Cu 2p XPS spectrum of CuO-300 product. The two bands at 933 eV (2p3/2) and 952.9 eV (2p1/2) along with shake-up satellite peaks at higher binding energies reveal the d<sup>9</sup> electronic state of Cu2<sup>+</sup> ions in CuO.

The morphology of the CuO samples was first observed with SEM as depicted in **Figure 4**. Clearly, the CuO-120 product is composed of many monodisperse, leaf-like nanosheets with lateral sizes of 300–500 nm and thickness of 40–50 nm. These nanosheets are constructed by plenty of interconnected nanoparticles, exhibiting a hierarchical structure with very rough surface (**Figure 4A**). After annealing at 300◦C, the CuO-300 sample (**Figure 4B**) retains the sheet-like morphology, albeit with the formation of some mesopores on the surface. Note that such CuO nanosheets can be obtained in large quantities as shown in **Figure S1**. Further increasing annealing temperature to 400◦C apparently reduces the number density of the mesopores on the nanosheets of the CuO-400 product (**Figure 4C**). Meantime, the surface of the CuO nanosheets becomes smooth possibly due to the partial fusion of adjacent nanoparticles. In contrast, the CuO-500 sample (**Figure 4D**) obtained after annealing at 500◦C contains irregular particles without pores. These results suggest that upon annealing at high temperatures the primary nanoparticles can locally

migrate or change their position, leading to gradual fusion of adjacent nanoparticles with reduced accessible surface and total surface energy of the whole system. High annealing temperature (e.g., ≥500◦C) also tends to destroy the original 2D sheet-like morphology together with severe loss of pore structure.

The microstructure of the as-obtained CuO samples was further investigated by TEM and HRTEM characterizations as shown in **Figure 5**. The porous leaf-like nanosheets morphology can be clearly observed for the CuO-120, CuO-300, and CuO-400 samples, albeit with the significantly decreased pores in the samples from heat-treatment with increased annealing temperatures (**Figures 5A,C,E** and their insets). In addition, the clear lattice images in **Figures 5B,D,F** depict the nanosheets in the three samples are all crystallized. The lattice fringes with interplanar spacing of ∼2.31 and ∼2.52 Å can be readily indexed to the (200) and (−111) planes of monoclinic CuO phase. Moreover, the corresponding fast Fourier transform (FFT) patterns in the insets of **Figures 5B,D,F** confirms the single-crystalline nature of the as-obtained CuO nanosheets. However, careful inspection discloses that the diffraction spots in inset of **Figure 5B** has a slightly weak brightness than that of inset in **Figure 5D**, suggesting a difference in crystallinity. Nonetheless, the diffraction spots in inset of **Figure 5D** show somewhat deformation, indicating the possible presence of lattice distortion. In-depth analyses were next performed to probe more details of the possible effects of annealing temperatures on the crystallinity and structural defects in the CuO products. **Figure S2** shows the inverse FFT images of the (−111) planes in CuO-120, CuO-300, and CuO-400 samples. Clearly, the CuO-120 product contains plentiful structural distortions (marked by dotted circles) and dislocations (marked by "T"). In CuO-300, the number of these structural defects is obviously reduced. Meantime, the pores are retained and most of them are located on the surface as revealed by HRTEM (**Figure S3**). Similarly, the number of structural defects further decreases in CuO-400. Therefore, it is anticipated that the electrical transport of the CuO product can be significantly boosted by simply annealing process through removing surface groups and eliminating structural disorders.

However, the annealing step may also lead to a considerable change of the surface area of the resultant CuO product which is closely related with the pore structure. For this purpose, N<sup>2</sup> adsorption experiments were performed in the following. From **Figure 5A**, the isotherms of the CuO samples display type-III type curves, indicating the presence of slit-like mesopores formed by interparticle stacking. Based on Brunauer-Emmett-Teller (BET) method, the specific surface areas are calculated to be 24.8, 21.8, 15, and 6.1 m<sup>2</sup> /g for CuO-120, CuO-300, CuO-400, and CuO-500 samples, respectively. Clearly, the CuO-300 has a comparable specific surface area to that of CuO-120 due to the well-preserved pore structure. In contrast, the surface area of CuO-500 has been sharply reduced due to the loss of pore and sintering (fusion) of nanoparticle building-blocks. In addition, the pore size distribution of the

CuO nanosheets by the Barrett-Joyner-Halenda (BJH) method suggests that all the four samples have a narrow pore size distribution ranging from 1.8 to 4 nm with a maximum pore diameter of ∼2.4 nm (**Figure 6B**), which is roughly in accordance with the SEM and TEM observation. Obviously, the mild synthetic approach and the resultant unique nanosheet morphology of the CuO products contributed to the relatively high specific surface area and pore volume, holding great promise for a series of energy-related applications such as LIBs.

The electrochemical performances of the as-prepared CuO samples were evaluated in CR2025 Li-half cells. **Figure 7A** depicts the cyclic voltammetry (CV) curves of the CuO-400 electrode within the potential range of 0.01∼3 V at a scan

rate of 0.2 mV s−<sup>1</sup> . During the first cathodic sweep, a large current response at 1.0 V is noted, which can be mainly due to the formation of solid-electrolyte interface (SEI) layer. In the following cycles, there are three cathodic peaks located at 2.3, 1.1, and 0.6 V, respectively during discharge. The first broad peak at 2.3 V corresponds to the formation of CuII 1−xCu<sup>I</sup> <sup>x</sup>O1−x/<sup>2</sup> (0≤x≤0.4) solid solution (Yuan et al., 2017). The second peak at 1.1 V signals the formation of Cu2O phase, while the third cathodic peak at 0.6 V correlates to the decomposition of Cu2O into metallic Cu and Li2O. In the subsequent charge process, three anodic peaks are noted at 1.6, 2.5, and 2.7 V, respectively, which suggests a multistep electrochemical reaction involving the decomposition of SEI film, re-oxidation of Cu to Cu2O, and finally to CuO. In addition, the 2nd and 3rd CV curves roughly overlap, indicating the high electrode reversibility. The first discharge-charge profiles of CuO-300 are shown in **Figure S4**. The CuO-300 electrode displays an initial discharge and charge capacity of 712 and 607 mAh/g, respectively with a Coulombic efficiency of 85%. **Figure 7B** presents the second and third galvanostatic charge-discharge (GCD) curves of the CuO-300 electrode at 0.2◦C. Similarly, three potential plateaus can be noted during the discharging and recharging process, which

agrees with that of the CV sweep result. The CuO-300 electrode exhibits a high discharge capacity of 592 mAh/g and charge capacity of 579 mAh/g with a high Coulombic efficiency (CE) of 97.8% in the 2nd cycle, which improves to 99% in the 3rd cycle. **Figure 7C** compares the rate capability of CuO-120 and CuO-300 electrodes at varied current rates of 0.2–2◦C. Both two electrodes demonstrate high and comparable Li-ion storage capacities of ∼527, 458, and 337 mAh/g at 0.2, 0.5, 1 ◦C, respectively. However, CuO-300 electrode delivers a higher capacity of 175 mAh/g, higher than that of CuO-120 electrode (80 mAh/g) at 2 C. Note that the rate performance of the CuO-300 is also better than the CuO nanosheets coupled with CNTs (252 mAh/g at 0.75 C, 165.8 mAh/g at 1.5 C) (Yuan et al., 2018). The better rate capacity of CuO-300 electrode at higher current rate can be mainly ascribed to its enhanced electronic conductivity with improved crystallinity and reduced defects. The cycling performance of the four electrodes at 0.5◦C is shown in **Figure 7D**. Clearly, CuO-300 electrode manifests the highest capacity and best cyclability among all the four electrodes. It delivers an initial capacity of 450 mAh/g, which retains 259 mAh/g over 500 cycles with a capacity retention of 58%. This result indicates that the increase of crystallinity is an effective way to improve the cycle stability. In addition, pore structure plays another key role in affecting the Li-storage capability. Note that the cycling properties of CuO-400 and CuO-500 electrodes are poor, even worse than that of CuO-120 electrode. This can be mainly attributed to the sharp reduction of pore structures and specific surface area of the CuO samples subject to heat-treatment at higher temperatures of 400 and 500◦C as revealed by the BET results (**Figure 6**). The electrochemical performance of the CuO-300 sample is also superior to some reported nanostructures, such as pillow-shaped porous CuO with ∼320 mAh/g after 50 cycles at 0.1 C (Wan et al., 2011).

Electrochemical impedance spectroscopy (EIS) measurement was further performed to probe the kinetic process of the electrode reaction. **Figure 8** shows the Nyquist plots of the four electrodes after five cycles at 0.2◦C. All the EIS spectra contain two semicircles and a sloping line. The two semicircles at high and high-to-medium frequencies represent the resistances of the SEI film (RSEI) and charge transfer (Rct) at electrode/electrolyte interface, while the sloping line at low frequency domain depicts the Li<sup>+</sup> diffusion in the solid state, respectively. Evidently, the CuO-300 electrode shows the lowest Rct value reflected by the smallest diameter of the second semicircle among all the four electrodes, suggesting the lowest polarization and fastest reaction kinetics of the CuO-300 electrode.

### CONCLUSION

Highly porous single-crystalline CuO nanosheets have been successfully fabricated on a large scale by a facile solution approach followed by heat-treatment at different temperatures. The interplay between electrochemical performance and the structural characteristics (crystallinity, crystallite size, and pore architecture) of the CuO materials has been established. The results indicate that the CuO-300 material delivers the best electrochemical properties in terms of superior rate and stable cycling performance. The superior electrochemical property of the CuO material can be mainly due to the improved crystallinity with boosted electronic conductivity and retained porous structure with highly accessible surface and reduced Li<sup>+</sup> and electron diffusion path lengths. Our work based on synergetic engineering of structural defects and pore structures may pave the way for development of highperformance electrode materials for next-generation lithium-ion batteries.

### AUTHOR CONTRIBUTIONS

ZD and H-EW conceived and designed the experiments. ZD, ZM, and YaL carried out the experiments and characterizations. YuL, LC, XY and B-LS contributed to data analysis and scientific discussion.

### ACKNOWLEDGMENTS

This work is supported by the Hubei Provincial Natural Science Foundation (2016CFB337). YuL and H-EW acknowledges the Hubei Provincial Government for the Chutian Scholar Program.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2018.00428/full#supplementary-material.

### REFERENCES


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Deng, Ma, Li, Li, Chen, Yang, Wang and Su. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Porous NaTi2(PO4)<sup>3</sup> Nanocubes Anchored on Porous Carbon Nanosheets for High Performance Sodium-Ion Batteries

Ziqi Wang<sup>1</sup> , Jiaojiao Liang<sup>1</sup> , Kai Fan<sup>1</sup> , Xiaodi Liu1,2 \*, Caiyun Wang<sup>3</sup> \* and Jianmin Ma1,4 \*

*<sup>1</sup> School of Physics and Electronics, Hunan University, Changsha, China, <sup>2</sup> College of Chemistry and Pharmaceutical Engineering, Nanyang Normal University, Nanyang, China, <sup>3</sup> ARC Centre of Excellence for Electromaterials Science, Intelligent Polymer Research Institute, AIIM Facility, University of Wollongong, North Wollongong, NSW, Australia, <sup>4</sup> Institute of Advanced Electrochemical Energy, Xi'an University of Technology, Xi'an, China*

### Edited by:

*Qiaobao Zhang, Xiamen University, China*

#### Reviewed by: *Chao Lai,*

*Jiangsu Normal University, China Guiming Zhong, Fujian Institute of Research on the Structure of Matter (CAS), China Anmin Cao, Institute of Chemistry (CAS), China*

#### \*Correspondence:

*Xiaodi Liu liuxiaodiny@126.com Caiyun Wang caiyun@uow.edu.au Jianmin Ma nanoelechem@hnu.edu.cn*

#### Specialty section:

*This article was submitted to Nanoscience, a section of the journal Frontiers in Chemistry*

Received: *05 June 2018* Accepted: *16 August 2018* Published: *19 September 2018*

#### Citation:

*Wang Z, Liang J, Fan K, Liu X, Wang C and Ma J (2018) Porous NaTi*2*(PO*4*)*3 *Nanocubes Anchored on Porous Carbon Nanosheets for High Performance Sodium-Ion Batteries. Front. Chem. 6:396. doi: 10.3389/fchem.2018.00396* NaTi2(PO4)<sup>3</sup> has attracted great interest as anode material for sodium ion batteries owing to its open three-dimensional framework structure and limited volume changes during the charge and discharge process. However, the poor intrinsic electronic conductivity of NaTi2(PO4)<sup>3</sup> needs to be improved for high rate capability. In this work, porous NaTi2(PO4)<sup>3</sup> nanocubes anchored on porous carbon nanosheets (NaTi2(PO4)3/C) are designed and developed. This material exhibits a large discharge capacity and good rate capacity including a first discharge capacity of 485 mAh g−<sup>1</sup> at a current density of 0.1 A g−<sup>1</sup> , and 98 mAh g−<sup>1</sup> retained at a high rate of 4 A g−<sup>1</sup> even after 2,000 cycles. These results suggest that NaTi2(PO4)3/C is a promising anode material for sodium-ion batteries.

Keywords: NaTi2 (PO4 )3 , nanocubes, carbon nanosheets, anode, sodium-ion batteries

## INTRODUCTION

Sodium-ion batteries (SIBs), as an alternative energy storage system for lithium-ion batteries (LIBs), have attracted increasing attention due to their low cost and the abundant resource of sodium (Gao et al., 2017; Cui et al., 2018; Liang et al., 2018c). The electrochemical performance of SIBs is closely related to the properties of electrode materials, especially anode materials (Chen et al., 2018; Fan et al., 2018; Hu A. J. et al., 2018; Liang et al., 2018b,d; Wan et al., 2018; Wei et al., 2018). Recently, Na super ion conductor (NASICON) type NaTi2(PO4)<sup>3</sup> has been considered as one of promising anode materials for SIBs owing to its "zero-stress" three-dimensional (3D) framework, high Na<sup>+</sup> conductivity, and good thermal stability (Kabbour et al., 2011; Wu et al., 2013; Sun et al., 2016; Ye et al., 2017).

However, the poor intrinsic electrical conductivity of NTP leads to poor rate capability (Pang et al., 2014b; Roh et al., 2017). To improve the Na<sup>+</sup> ions insertion-extraction kinetics, two common approaches used include synthesis of various nanostructures and fabrication of carbon composites. Morphology control of NaTi2(PO4)<sup>3</sup> has been applied to realizeexcellent electrochemical performance. Different nanostructures such as hollow nanocubes, nanoparticles, and hierarchical microspheres have been demonstrated (Wu et al., 2015; Fang et al., 2016; Ye et al., 2017). Among them, porous structures have gained great attention owing to the afforded large surface areas and improved kinetics (Dirican et al., 2015; Zhang et al., 2017; Zhao et al., 2017; Zhou et al., 2017). Moreover, the electronic conductivity of anodes can be largely enhanced by hybridizing them with conductive materials. For example, the coating of carbon or graphene on NaTi2(PO4)<sup>3</sup> micro/nanostructures can effectively improve their properties and higher quality of the conductive materials could result in better electrochemical performance. Nevertheless, the contents of carbon or graphene in the previously reported composites were only 3.4–6.8 wt% (Pang et al., 2014b; Fang et al., 2016; Geng et al., 2017; Hu Q. et al., 2018; Liang et al., 2018a). Thus, to obtain better electrochemical properties, the contents of conduction materials should be increased. It has been found that the embedding of anode materials in carbon/graphene matrixes can realize high content of conductive carbon materials for enhanced electrochemical properties (Fu et al., 2015; Guo et al., 2015; Choi et al., 2016; Sun et al., 2016). Motivated by the above potentials, we have prepared porous NaTi2(PO4)<sup>3</sup> nanocubes anchored on porous carbon nanosheets (NaTi2(PO4)3/C) through ultrasonic treatment. To the best of our knowledge, it is the first report that NaTi2(PO4)<sup>3</sup> nanocubes with porous structures have been embedded in a porous carbon matrix. This NaTi2(PO4)3/C material exhibited a high discharge capacity, good rate performance, and excellent long-time cycling stability.

## EXPERIMENTAL SECTION

## Synthesis of NaTi2(PO4)3/C

The synthesis of porous carbon nanosheets was firstly conducted from the uniform mixture of Zn(CH3COO)2·2H2O (5 g) and oleic acid (5 g) in an agate mortar for 30 min. Then the above mixture was transferred into a tubular furnace and calcined at 700◦C for 2 h with a ramping rate of 2◦C min−<sup>1</sup> in Ar atmosphere to form ZnO/C slices. The ZnO/C slices were washed using 6 mol L−<sup>1</sup> aqueous HCl solution to form porous carbon nanosheets. These carbon nanosheets were washed by deionized water and absolute ethanol, then dried in vacuum at 50◦C for 8 h. Similar process was used to prepare the MnO/graphene composite using oleic acid as carbon sources (Guo et al., 2015).

The synthesis of NaTi2(PO4)<sup>3</sup> nanocubes was conducted following the reported procedures (Wu et al., 2015). Briefly, sodium acetate (0.16 g) was added in a mixed solvent glacial acetic acid (0.7 mL), phosphoric acid (4 ml) and ethylene glycol (25 ml), followed by the addition of tetrabutyl titanate (1.36 g). The resultant mixture was heated at 180◦C for 12 h. Finally, the white precipitate NaTi2(PO4)<sup>3</sup> was obtained.

For synthesizing NaTi2(PO4)3/C, 0.16 g precursor NaTi2(PO4)3, 0.04 g porous carbon nanosheets, and 3.6 g cetyltrimethylammonium bromide (CTAB) were added into 30 mL absolute ethanol. After being stirred for 2 h and ultrasonically dispersed for 2 h, the precursor NaTi2(PO4)3/C was collected by centrifugation, washed with deionized water and anhydrous ethanol, and dried at 60◦C for 12 h. Subsequently, the precursor NaTi2(PO4)3/C was further calcined at 700◦C for 2 h with a ramping rate of 2◦C min−<sup>1</sup> in Ar atmosphere to form NaTi2(PO4)3/C. For comparison, porous NaTi2(PO4)<sup>3</sup> cubes were prepared after annealing without the porous carbon nanosheets.

### Characterizations

Rigaku D/max-2500 X-ray diffractometer (Cu Kα, λ = 1.54056 Å) was used to investigate the crystal structures. The morphology and nanostructure were observed by Hitachi S4800 scanning electron microscopy and JEOL 2010 transmission electron microscopy. The Brunauer-Emmett-Teller special (BET) surface area and pore size were tested at 77 K on a Nova 2000e volumetric adsorption analyzer. The thermogravimetric (TG) analysis was performed with a WCT-1D instrument over a range of 30–800◦C at a heating rate of 10◦C·min−<sup>1</sup> in air atmosphere.

### Electrochemical Measurements

Active materials, acetylene black and carboxymethylcellulose sodium with a weight ratio of 80:10:10 were uniformly mixed, and the obtained slurry was coated on Cu foil. Then, the electrodes were assembled into CR2025 coin cell in the glove box. The glass microfiber filter membrane (Whatman, grade GF/A) was used as the separator. Metallic sodium film was used as counter/reference electrodes. The electrolyte was 1 mol L <sup>−</sup><sup>1</sup> NaClO<sup>4</sup> dissolved in a mixture of ethylene carbonate and diethyl carbonate (1:1 vol%) with 5 wt% fluoroethylene carbonate. Galvanostatic tests were evaluated by Neware Battery Testing System. Cyclic voltammetry (CV) tests and impedance measurement were carried out on a CHI660C Electrochemical Workstation.

### RESULTS AND DISCUSSION

The phase of NaTi2(PO4)<sup>3</sup> and NaTi2(PO4)3/C were confirmed by XRD, as shown in **Figure 1A**. All diffraction peaks are in accordance with the standard pattern of NaTi2(PO4)<sup>3</sup> (JCPDS No. 84-2008). No peaks for impurities can be detected, suggesting the high purity of these two samples. Moreover, the diffraction peaks of carbon material is not clearly discerned due to the sharp and strong diffraction of NaTi2(PO4)3, implying the high crystalline nature. In addition, according to the TGA curve of NaTi2(PO4)<sup>3</sup> (**Figure 1B**), the relative weight fraction of carbon for NaTi2(PO4)3/C was determined to be ∼18.3%.

The morphology of porous carbon nanosheets, NaTi2(PO4)3, and NaTi2(PO4)3/C was characterized by SEM. **Figure 2A** shows the low-magnified SEM image of porous carbon nanosheets. It is clear that the sample is exclusively nanosheets with irregular morphologies. The high-resolution SEM image (**Figure 2B**) shows that the carbon nanosheets are curved and have an average thickness of ∼10 nm. SEM images of NaTi2(PO4)<sup>3</sup> (**Figures 2C,D**) show that the products have uniform cubic shapes and their sizes are in the range between 50 and 100 nm, similar with the results reported in literatures (Liang et al., 2018a).

It was observed that the NaTi2(PO4)<sup>3</sup> were uniformly anchored on the porous carbon nanosheets for NaTi2(PO4)3/C sample (**Figures 3A,B**). The detailed structural characteristics of NaTi2(PO4)3/C was further investigated by TEM. The TEM image in **Figure 3C** illustrates that NaTi2(PO4)<sup>3</sup> in a size range of 50 and 100 nm were scattered over the carbon nanosheets. This is consistent with the SEM results in **Figure 3B**. Notably, both nanocubes and carbon nanosheets have obvious porous structure, which is beneficial for the transport of

BET analysis was performed to study the pore size and specific surface area of NaTi2(PO4)3/C. The Nitrogen adsorptiondesorption isotherm of NaTi2(PO4)3/C (**Figure 4A**) reveals a type-IV isotherm with an obvious H1-type hysteretic loop in the range of 0.4–1.0 (P/P0), indicating that the products possess porous structures (Takashima et al., 2015). The BET analysis indicates that the specific surface area of NaTi2(PO4)3/C was ca. 103.1 m<sup>2</sup> g −1 . Moreover, as shown in **Figure 4B**, the sample possessed a broad pore-size distribution and the poresize distribution maximum was centered at 15.4 nm. The large surface area and porous structure of NaTi2(PO4)3/C is beneficial to improve the sodium-ion storage properties (Wang H. et al., 2016; Wang G. et al., 2018).

The electrochemical properties of NaTi2(PO4)3/C were studied as anode material for SIBs. The cyclic voltammogram (CV) of NaTi2(PO4)3/C at a scan rate of 0.1 mV s−<sup>1</sup> was analyzed

FIGURE 2 | (A) SEM image and (B) high-resolution SEM image of porous carbon nanosheets; (C) SEM image and (D) high-magnification SEM image of NaTi2(PO4)3.

to investigate their redox kinetic properties. In **Figure 5A**, at the 1st cycle, a pair of redox peaks at 1.97/2.29 V can be attributed to conversion reaction of Ti4+/Ti3<sup>+</sup> (Pang et al., 2014a; Fang et al., 2016; Ye et al., 2017). Moreover, another pair of cathodic/anodic peaks located at 0.27/0.57 V can be attributed to the redox reaction between Ti3<sup>+</sup> and Ti2<sup>+</sup> (Senguttuvan et al., 2013; Wang D. et al., 2016). That is, Ti4<sup>+</sup> in the reduction process was firstly reduced to Ti3<sup>+</sup> (NaTi2(PO4)<sup>3</sup> + 2Na<sup>+</sup> + 2e<sup>−</sup>

→ Na3Ti2(PO4)3) and then formed into Ti2<sup>+</sup> (Na3Ti2(PO4)3+ Na<sup>+</sup> + e <sup>−</sup> → Na4Ti2(PO4)3). In the following cycles, the anodic peaks shift to higher potentials (1.97 vs. 2.09 V; 0.27 vs. 0.30 V), which was probably caused by the stress/strain change, similar to other NASICON-type anodic materials (Li et al., 2014). More importantly, in the following 2nd, 3rd, and 5th cycles, two pairs of reduction/oxidation peaks almost remained unchanged, indicating the excellent reversibility.

**Figure 5B** showed the galvanostatic discharge-charge curves of NaTi2(PO4)3/C electrode in the voltage window between 0.01 and 3.0 V. The initial discharge capacity was 485 mAh g−<sup>1</sup> , which was higher than the theoretical capacity (133 mAh g−<sup>1</sup> ). However, the initial charge capacity was 227 mAh g−<sup>1</sup> with an unsatisfied Coulombic efficiency of 46.8%. Such a large capacity loss is mostly ascribed to the formation of solid electrolyte interface (SEI) layers for the existence of carbon substrates, as well as the decomposition of electrolyte (Hasegawa et al., 2016; Wang D. et al., 2016). On the contrary, the first discharge capacity of the NaTi2(PO4)<sup>3</sup> electrode was only 229 mAh g−<sup>1</sup> (**Figure S1**, Supporting Information). In the subsequent cycles, the NaTi2(PO4)3/C electrode possessed good cycle stability and excellent reversibility for Na<sup>+</sup> ion insertion and extraction. For example, at the 5th and 10th cycles, the discharge capacity retained to be 221 and 203 mAh g−<sup>1</sup> with the coulombic efficiency of 94 and 96%, respectively. **Figure 5C** displayed the cycling behavior of the NaTi2(PO4)3/C and NaTi2(PO4)<sup>3</sup> electrodes at a current density of 0.1 A g−<sup>1</sup> . It can be seen thatafter 100 cycle NaTi2(PO4)3/C still delivered a discharge capacity of 172 mAh g−<sup>1</sup> , which was much larger than that of NaTi2(PO4)<sup>3</sup> (20 mAh g−<sup>1</sup> ). Accordingly, the NaTi2(PO4)3/C electrode exhibited a capacity retention of 69% (relative to the 2nd cycle), higher than that of NaTi2(PO4)<sup>3</sup> (30%). In addition, the long-term cycling performance for the NaTi2(PO4)3/C electrode at a relatively high rate of 4 A g−<sup>1</sup> was further studied. In **Figure 5E**, it can be clearly found that the Coulombic efficiency could

.

exceed 98% since the 10th cycle, and the electrode can still maintain a discharge capacity of 98 mAh g−<sup>1</sup> even after 2,000 cycles. All these results indicate that NaTi2(PO4)3/C afforded improved electrochemical stability compared with that of NaTi2(PO4)3.

Furthermore, the rate capability of NaTi2(PO4)3/C electrode was also investigated by increasing rate from 0.02 to 4 A g−<sup>1</sup> and back to 0.2 A g−<sup>1</sup> . As illustrated in **Figure 5D**, the discharge capability of NaTi2(PO4)3/C was 280 mAh g −1 at 0.02 A g−<sup>1</sup> , and then it slowly decreased with the increasing current density. When the current density was reversed to 0.2 A g−<sup>1</sup> , a capacity of 164 mAh g−<sup>1</sup> could be restored. Obviously, NaTi2(PO4)3/C has excellent rate capacity.

Lastly, EIS measurements were carried out to further study the surface reaction activities of NaTi2(PO4)3/C and NaTi2(PO4)3. Before the EIS tests, the coin cells were cycled three times in the voltage range of 1.0–2.5 V, and the corresponding Nyquist plots are shown in **Figure 6**. It can be seen that each Nyquist plot exhibited a semicircle at high frequency region and a straight line at low frequency region. The surface charge-transfer resistance (Rct) of NaTi2(PO4)3/C was found to be smaller than that of NaTi2(PO4)3, suggesting that the diffusion of Na<sup>+</sup> in NaTi2(PO4)3/C is faster than NaTi2(PO4)<sup>3</sup> (Lu et al., 2014; Longoni et al., 2016). In addition, the Na<sup>+</sup> diffusion coefficient (D) can be calculated by the following equations (Ko et al., 2017):

$$\mathbf{D} = \mathbf{R}^2 \mathbf{T}^2 / 2\mathbf{A}^2 \mathbf{n}^4 \mathbf{F}^4 \mathbf{C}^4 \sigma^2 \tag{1}$$

$$Z' = \mathbf{R\_o} + \mathbf{R\_{ct}} + \sigma \boldsymbol{\alpha}^{-0.5} \tag{2}$$

in which R is the ideal gas constant, T is the ambient temperature, A is the surface area of the electrode, n is the number of electrons per molecule during intercalation, F is the Faraday constant, C is the concentration of Na<sup>+</sup> in the active material, σ is the Warburg coefficient, Z' is the real part of the impedance, ω is the angular frequency. The σ value can be calculated by the slope of the plot of Z' vs. ω <sup>−</sup>0.5 and presented in **Figure S2**. The σ value of NaTi2(PO4)3/C was 254 Ω s <sup>−</sup>0.5, much lower than that of NaTi2(PO4)<sup>3</sup> (1264 Ω s <sup>−</sup>0.5). Accordingly D of NaTi2(PO4)3/C was larger than that of NaTi2(PO4)3. Summarily, NaTi2(PO4)3/C can effectively restrain the increasing of chargetransfer resistance after multiple discharge and charge cycles, which can improve the rate capability and enhance the cyclic performance at high rate (Song et al., 2014; Roy and Srivastava, 2015).

According to the above results, NaTi2(PO4)3/C has high discharge capacity, good rate capacity, and excellent long-term cycling stability. In addition, compared to other previously reported NaTi2(PO4)3@C composites, the obtained NaTi2(PO4)3/C electrode exhibits excellent properties (**Table S1**, Supporting Information). The good properties of NaTi2(PO4)3/C could be ascribed to the following reasons: (i) The crystal structure of NASICON-type NaTi2(PO4)<sup>3</sup> is an open 3D framework of PO<sup>4</sup> tetrahedra corner-shared with TiO<sup>6</sup> octahedra, which can not only provide large spaces for Na<sup>+</sup> insertion but also supply open tunnels for Na<sup>+</sup> transport (Boilot et al., 1983; Pang et al., 2014b; Zhao et al., 2015). (ii)

The porous structure of nanostructured NaTi2(PO4)<sup>3</sup> and carbon matrix can decrease the diffusion length of Na<sup>+</sup> (Gibaud et al., 1996; Huang et al., 2015; Rui et al., 2016). (iii) The embedding of NaTi2(PO4)<sup>3</sup> nanocubes in carbon nanosheets can effectively inhibit the aggregation of the nanocubes, leading to the electrolyte easily penetrating to the active sites.

### CONCLUSION

In summary, the composition of NaTi2(PO4)<sup>3</sup> porous nanocubes and carbon porous nanosheet are successfully developed. The asobtained NaTi2(PO4)3/C electrodes have good electrochemical properties, including large energy density, excellent rate capacity, and good cycling performance, owing to their special structures and components. The results demonstrate that such NaTi2(PO4)3/C anode is a promising anode for SIBs.

### AUTHOR CONTRIBUTIONS

ZW, XL, CW, and JM design the whole experiment, and write the paper. JL and KF conduct some electrochemical analysis.

### FUNDING

This work is supported by the National Natural Science Foundation of China (No. 21501101), the China Postdoctoral Science Foundation (No. 2017M622564), the Program for Science and Technology Innovation Talents in Universities of Henan Province (No. 15HASTIT007), and the Natural Science Foundation of Hunan Province (2017JJ1008).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2018.00396/full#supplementary-material

### REFERENCES


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Wang, Liang, Fan, Liu, Wang and Ma. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Activated Amorphous Carbon With High-Porosity Derived From Camellia Pollen Grains as Anode Materials for Lithium/Sodium Ion Batteries

Kaiqi Xu<sup>1</sup> , Yunsha Li 1,2, Jiawen Xiong<sup>2</sup> , Xing Ou<sup>2</sup> , Wei Su<sup>1</sup> , Guobin Zhong<sup>1</sup> and Chenghao Yang<sup>2</sup> \*

*<sup>1</sup> Electric Power Research Institute of Guangdong Power Grid Co., Ltd., Guangzhou, China, <sup>2</sup> Guangzhou Key Laboratory for Surface Chemistry of Energy Materials, New Energy Research Institute, School of Environment and Energy, South China University of Technology, Guangzhou, China*

### Edited by:

*Qiaobao Zhang, Xiamen University, China*

#### Reviewed by: *Yan-Bing He,*

*Tsinghua University, China Xianwen Wu, Jishou University, China Hongshuai Hou, Central South University, China*

> \*Correspondence: *Chenghao Yang esyangc@scut.edu.cn*

#### Specialty section:

*This article was submitted to Physical Chemistry and Chemical Physics, a section of the journal Frontiers in Chemistry*

Received: *28 June 2018* Accepted: *30 July 2018* Published: *04 September 2018*

#### Citation:

*Xu K, Li Y, Xiong J, Ou X, Su W, Zhong G and Yang C (2018) Activated Amorphous Carbon With High-Porosity Derived From Camellia Pollen Grains as Anode Materials for Lithium/Sodium Ion Batteries. Front. Chem. 6:366. doi: 10.3389/fchem.2018.00366* Carbonaceous anode materials are commonly utilized in the energy storage systems, while their unsatisfied electrochemical performances hardly meet the increasing requirements for advanced anode materials. Here, activated amorphous carbon (AAC) is synthesized by carbonizing renewable camellia pollen grains with naturally hierarchical structure, which not only maintains abundant micro- and mesopores with surprising specific surface area (660 m<sup>2</sup> g −1 ), but also enlarges the interlayer spacing from 0.352 to 0.4 nm, effectively facilitating ions transport, intercalation, and adsorption. Benefiting from such unique characteristic, AAC exhibits 691.7 mAh g−<sup>1</sup> after 1200 cycles at 2 A g −1 , and achieves 459.7, 335.4, 288.7, 251.7, and 213.5 mAh g−<sup>1</sup> at 0.1, 0.5, 1, 2, 5 A g −1 in rate response for lithium-ion batteries (LIBs). Additionally, reversible capacities of 324.8, 321.6, 312.1, 298.9, 282.3, 272.4 mAh g−<sup>1</sup> at various rates of 0.1, 0.2, 0.5, 1, 2, 5 A g−<sup>1</sup> are preserved for sodium-ion batteries (SIBs). The results reveal that the AAC anode derived from camellia pollen grains can display excellent cyclic life and superior rate performances, endowing the infinite potential to extend its applications in LIBs and SIBs.

Keywords: activated amorphous carbon, anode, high porosity, lithium ion batteries, sodium ion batteries

### INTRODUCTION

The emergence and rapid development of portable electronic devices, electric vehicles and renewable energy industries urge preeminent large-scale energy storage and conversion systems, which exhibit superior energy density and specific capacity as well as phenomenal life-span with trustworthy safety (Hannan et al., 2017; Zhang et al., 2017). Among prevalent energy storage systems, lithium ion batteries (LIBs) and sodium ion batteries (SIBs) attract most attentions owing to their high working potential and extraordinary storage capacity (Shen et al., 2018). It is well-known that in order to promote practical applications of LIBs and SIBs, developing high-performing electrode materials are inevitable (Zheng et al., 2015; Luo et al., 2016). With increasing improvements in cathode materials, commercial graphite as the mainstream anode material gradually becomes dissatisfactory due to its limited theoretical capacity (372 mAh g−<sup>1</sup> for LIBs) (Hou et al., 2015; Zhang et al., 2016b). Besides, the highly regular distance between graphene layer (0.335 nm) can hardly accommodate the size of Na+, prohibiting the possibility to form stable Na-intercalation compounds (Irisarri et al., 2015; Yang et al., 2017). Though graphite hinders the improvement of LIBs and SIBs, carbonaceous materials still consumingly appeal to researchers, thanks to their significant advantages of low lithium/sodium ion insertion/desertion potential, highly structural stability, cost-efficiency, and sufficient resources (Hou et al., 2017a,b).

Amorphous carbon, compared to graphite, can reach higher specific capacity with enlarged interlayer distance and shortened ion transportation paths. Owing to environment amiability, sustainability and cost efficiency, ta large amount of diverse selectable biomass materials, such as protein (Li et al., 2013), peanut shell (Lv et al., 2015), bee pollen grain (Tang et al., 2016), rice husk (Zhang et al., 2016a), cotton (Li et al., 2016; Xiong et al., 2018; Yang et al., 2018), garlic skin (Zhang et al., 2018c), have been investigated in depth of their electrochemical capabilities. Generally, biomass-derived amorphous carbon can be synthesized by directly pyrolysis (Li et al., 2016) <sup>−</sup>(Wang et al., 2015) or hydrothermal treatment (Li et al., 2015). It is noted that most of the biomass-derived carbon can successful maintain the uniform sphere shape, inherent textile and hollow tubes. Despite of advantages of biomass-derived carbon, it is convinced that its electrochemical potential has not been ultimately exploited. To address this challenge, researchers have exerted their efforts on constructing inner microstructures and modifying surface properties of biomass-derived carbon by utilizing the activation methods (Gu et al., 2014; Wu et al., 2016a). Noticeably, chemical activation by mixing carbon materials and chemical reagents, KOH for instance, is less energy-consuming than physical activation resulting from the relatively low anneal temperature (Gao et al., 2016).

Pollen grains, distinctively, presenting uniformly discrete particle distribution and divergent particle shapes from various kinds of flowers, are usually employed in synthesis of various materials as templates (Chen et al., 2016). When regarded as anode for LIBs, cattail pollen carbon after physically activating exhibits 382 mAh g−<sup>1</sup> at 25◦C at rate of 37.2 mA g −1 , distinguishing itself as a competitive candidate in anode materials beyond graphite (Tang et al., 2016). Impractical test current density is reason of the slightly better capacity, which still need further improvements. Therefore, further exploration of beneficiated activations for fabricating pollen grain's microstructure is required.

In this work, taking the advantage of pollen grains, we put forward a homogeneous and tunable synthetic method for activated amorphous carbon (AAC) derived from camellia pollen grains. Pollen grains were activated through chemical activation using KOH as activating reagent (Su et al., 2017), followed by a two-stage carbonization process. The eventual product is identified as poly-porous amorphous carbon with enhanced specific surface area around 660 m<sup>2</sup> g −1 and enlarged layer distance from 0.35 to 0.4 nm, which contributes to advanced specific capacities and cyclic performances with excellent rate when acting as anode materials for LIBs and SIBs. The outstanding characteristics provide camellia-pollen-grainderived carbon with the opportunity to meet the increasing demand of advanced storage devices.

### EXPERIMENTAL

### Material Preparation

The activated amorphous carbon was synthesized by annealing camellia pollen grain. After rinsing with acetone and deionized water, 3 g raw pollen grains were mixed with 80 mL KOH solution (0.025 g mL−<sup>1</sup> ) and stirred for 6 h at 40◦C, following by drying process at 80◦C. The obtained pollen grains were then transferred into a tube furnace and underwent the two-stage heat treatment under N<sup>2</sup> atmosphere, which suggested maintaining temperature at 400◦C for 1 h at the first stage and 800◦C for 3 h at the second stage with the fixed heating rate of 5◦C/min for the entire calcination. Finally, the sample was etched by 3M HCL and thoroughly washed with deionized water and desiccated at 80◦C for 24 h. The as-prepared activated amorphous carbon was denoted as AAC.

For comparison, another 3 g raw pollen grains were directly annealed through two-stage heat treatment aforementioned to acquire pristine amorphous carbon, denoting as AC.

### Material Characterizations

Scanning electron microscope (SEM, FEIQuanta 200 FEG) and transmission electron microscope (TEM, Tecnai G2 F20 S-TWIN, Japan) were performed to unveil the morphology and microscopic structure of both samples. To gain the intrinsic features of the samples, X-ray diffraction (XRD, Bruker D8 Advance, Germany) (Cu, Kα, λ = 1.5405 Å), Raman spectrometer (JOBIN-Yvon HR800) and X-ray photoelectron spectroscopy (XPS, Thermo/ESCALAB 250XI) were conducted. N<sup>2</sup> adsorption/desorption isotherms for calculation of specific surface area (SSA) and accumulative pore volume based on Brunauer-Emmett-Teller and density functional theory methods was carried out on ASAP 2020 Micromeritics.

### Electrochemical Measurements

The working electrode was fabricated using copper foil covered with slurry comprised of 70 wt % of AAC or AC, 20 wt % of PVDF binder and 10 wt % carbon black. Subsequently the electrode is dried out with 0.5 mg cm−<sup>2</sup> average active material loading to assemble CR2032 coin cells together with PE/PP films or glass-fiber papers as separators, metal lithium or sodium foils as counter electrodes and corresponding electrolyte (1 M LiPF<sup>6</sup> or 1 M NaCF3SO<sup>3</sup> in ethylene carbonate, dimethyl carbonate and ethyl methyl carbonate=1:1:1(v: v: v) for lithium ion batteries and sodium ion batteries, respectfully). CV and EIS results were attributed to the employment of CHI660A electrochemical workstations while galvanostatic charge/discharge profiles were resulted from the measurements of LAND CT2001A batterytesting instruments.

### RESULTS AND DISCUSSIONS

The overall synthesis of AAC is shown in **Figure 1**. The KOH activation followed by carbonization and acid etching process are involved in this strategy to introduce poly-pores into the interior structures of camellia pollen grains (details in experimentally section). The microscopic morphologies of the

original pollen grains, AC and AAC are analyzed via SEM, TEM and HRTEM images (**Figure 2** and **Figure S1**). Camellia pollen grains are consisted of irregular spheres with rough surfaces, and the uniform diameter is about 30µm (**Figures S1A,B**). In comparison with AC (**Figures S1C,D**), **Figures 2A,B** visualizes that, in morphology, AAC possesses smooth surface, vast amount of mesopores and micropores, while primitive bulk shape is well-maintained, proving the effective function of KOH agent as impetus aiming at high porosity. Consistency can be found in corresponding TEM images (**Figure 2C**), in which explicit micro- and mesopores morphology. Moreover, it is noticeable in HRTEM (the inset figure of **Figure 2D**) that AAC is comprised of turbostratic nanodomains with ∼0.4 nm interlayer distance in average graphene interlayers. The evidence of dispersed diffraction ring in selected area electron diffraction pattern (SAED) can be observed, consistent with the HRTEM results, confirming the amorphous characteristic of AAC.

To gain a deeper insight of the composition and crystal structure of both samples, XRD patterns, and Raman spectra are investigated as displayed in **Figures 3A–C**. The broad peaks in XRD pattern referring to (002) facet are detected at 2θ of 21.8◦ and 25◦ for AAC and AC, respectively (**Figure 2A**), confirming the amorphous property of both samples. It is noted that the left shift of 2θ for AAC is ascribed to the enlarged interlayer. According to Bragg equation (Ou et al., 2016), interlayer distance between graphene layer can be calculated to be 0.409 and 0.352 nm for AAC and AC, respectively, both larger than 0.335 nm for graphite (Hou et al., 2017a). Such appealing expansion for AAC should be ascribed to KOH activation mechanism, which is explained by three steps as followed: (i) Firstly potassium-containing compounds commence etching carbon substrate to form the initial porous framework, which is called chemical activation. (ii) Then H2O and CO<sup>2</sup> escape to impel the increment of porosity for physical activation. (iii) Metallic K permeates into carbon matrices, achieving the larger graphene interlayer distance, resulting in the lattices expansion (Romanos et al., 2012; Wang and Kaskel, 2012). The enlarged interlayer distance has favorable influences on the electrochemical performance of AAC, which enables reversible intercalation for Li+/Na<sup>+</sup> ion and well maintain the structural integrity at the same time (Wen et al., 2014; Kim et al., 2017; Zou et al., 2017; Yang et al., 2018; Zhu et al., 2018). Subsequently, Dband representing defects induction at ∼1334 cm−<sup>1</sup> and G-band representing in-plane vibration at ∼1577 cm−<sup>1</sup> are located in Raman spectra (**Figure 3B**), whereby the ID/I<sup>G</sup> can be calculated as 0.84 for AAC and 1.11 for AC. Furthermore, based on the ID/I<sup>G</sup> ratio, the AAC is of higher graphitization degree, providing a crucial foundation for improving electrical conductivity (Lv et al., 2015; Zhang et al., 2018c).

In addition, N<sup>2</sup> adsorption/desorption isotherms measurement and differential pore volume distributions are tested for both sample in **Figure 3C**. The specific surface area (SSA) of AAC dramatically increases from 21.32 to 660.05 m<sup>2</sup> g −1 . While concurrently the accumulative pore volume of AAC scales up to 0.29 cm<sup>3</sup> g −1 , overwhelming larger than the AC of merely 0.01253 cm<sup>3</sup> g −1 . Unlike AC showing negligible few pores, AAC contains ample micropores and mesopores concentrating at ∼6 and 11 nm. The improved SSA and porosity, resulted from synergic effect between the chemical conversion of KOH and formation of H2O and CO2, offers not only extra active sites for Li+/Na<sup>+</sup> ion but also shortened ion transportation path with reduced inner resistance (Yu et al., 2016). To understand the existing chemical states of all elements in AAC, XPS measurement is characterized. Judging from XPS curve as displayed in **Figure 3D**, the peaks at 285.08 and 534.08 eV are attributed to C and O elements, yielding an occupancy of 94.56 and rest 5.44%, respectively. Individually, as displayed in **Figures 3E,F**, four predominant peaks are specified

as C-C (284.78 eV), C-OH (285.48 eV), C=O (286.88 eV), and C-OOH (290.08 eV) in deconvoluted C 1s spectrum. Besides, three apparent peaks in deconvoluted O 1s spectrum are assigned to the representative C=O (531.88 eV), C-O-H (533.28 eV), and C-O-C (534.98 eV) (Lin et al., 2016; Li et al., 2017a).

**Figure 4** illustrates the electrochemical performances of AAC and AC as anode materials for LIBs. Electrochemical behaviors of AAC and AC during the initial three cycles are examined through cyclic voltammetry (CV) test at a scan rate of 0.1 mV s−<sup>1</sup> within voltage range between 0.01 and 3.0 V, as presented at **Figure 4A** and **Figure S3A**, respectively. Like AC sample, three identifiable cathodic peaks for AAC, which center at range of 1.8–0.4 V in the first CV curves and vanish in the following cycles are ascribed to the formation of irreversible solid electrolyte interface (SEI) and

unanticipated side reactions due to the consumption of oxygenic functional groups (Lv et al., 2015; Li et al., 2017b; Huang et al., 2018) Meanwhile, a reversible cathodic peak appears at potential window of 0.5–0.01 V, in concert with the anodic peak at 0.2 V, is originated from the reversible lithiation/delithiation process. It is observed that the 2nd and 3rd cycle of CV curves are scarcely discriminated, suggesting that SEI films formed at the first cycles securely protect the structural integrity of AAC an AC, and enduring the rapid ion insert/extraction. Correspondently, the first galvanostatic charge/discharge portrait at 0.1 A g−<sup>1</sup> for AAC (**Figure S2A**) shows charge and discharge plateaus at around 1.6 and 1.2 V, respectively, which are generated from unavoidable SEI fabrication and electrolyte decomposition, leading to an irreversible capacity loss. Therefore, the charge and discharge capacities of AAC for the first cycle are 777.8 mAh g−<sup>1</sup> and 1803.8 mAh g−<sup>1</sup> , with initial coulombic efficiency (ICE) about 43.12%, which is exceeding the AC anode.

Evaluation of rate performance for AAC anode is conducted by gradient current densities, as depicted in **Figures 4B,C**. AAC anode achieves the charge capacities of 459.7, 335.4, 288.7, 251.7, and 213.5 mAh g−<sup>1</sup> at rate of 0.1, 0.5, 1, 2, and 5 A g−<sup>1</sup> , respectively, whereas AC delivers an overall inferior performance with unsatisfactory capacity decline at high current density of 5 A g −1 (**Figure S3B**). When the current density decreases from 5 to 0.1 A g−<sup>1</sup> by stages, the charge capacity of AAC are restored to nearly 100% compared with the pristine state, reflecting the outstanding rate tolerance and stable structural maintenance of AAC. In galvanostatic charge/discharge test (**Figures 4D,E**), the charge capacity of AAC attractively reaches 479.2 mAh g−<sup>1</sup> after 150 cycles at a current density of 0.5 A g−<sup>1</sup> with 96% retention. Furthermore, it can achieve reversible capacity of 691.7 mAh g−<sup>1</sup> after 1200 cycles at a high rate of 2 A g−<sup>1</sup> with an ascending tendency in the first few hundred cycles, which is stemmed from gradual electrochemical activation of AAC and is similar to most carbonaceous anode materials (Qie et al., 2012; Lv et al., 2015; Xiong et al., 2018; Yang et al., 2018; Zhang et al., 2018b). On the contrary, ordinary AC merely delivers 261.6 mAh g−<sup>1</sup> at 0.5 A g−<sup>1</sup> and 229.1 mAh g −1 at 2 A g−<sup>1</sup> under the same cycling condition. Conspicuous improvements in electrochemical performance for AAC anode can be explained by synergistic effort of highly porous structure and expanded interlayer of d<sup>002</sup> plane due to KOH activation. Specifically, the interface between electrode and electrolyte is modified by multiple pores for wide wettability, facilitating a much more efficient accesses for ions transportation through minimizing inherent impedance and shorten the diffusion paths. Besides, the larger interlayer distance favors the reversibility of ion intercalation and extraction. Furthermore, taking the advantages of this unique structure, additional active sites for Liions adsorption are provided, and integral structure to mitigate volume variation are guaranteed, making AAC into a favorable anode material for LIBs (Qin et al., 2014; Xiong et al., 2016, 2017).

For further evidences, CV curves acquired at different scan rate varied from 0.1 to 10.0 mV s−<sup>1</sup> with similar feature are shown in **Figure 5A**. Capacitive behavior of AAC is estimated via the following equation:

$$\text{The first-order coupling between the two-dimensional } \mathcal{N} \text{-matrices is the only possible } \mathcal{N} \text{-matrices with } \mathcal{N} = \{0, 1, 2, \dots, N\} \text{ and } \mathcal{N} = \{0, 1, 2, \dots, N\}.$$

or:

$$i(\mathbf{V}) = \mathbf{k}\_1 \boldsymbol{\nu} + \mathbf{k}\_2 \boldsymbol{\nu}^{1/2} \tag{1}$$

1/2

$$i(\mathbf{V})/\nu^{1/2} = \mathbf{k}\_1 \nu^{1/2} + \mathbf{k}\_2 \tag{2}$$

where i is the responsive current at various scan rates (v) at randomly chosen potential (V), and k (both k<sup>1</sup> and k2) are constants. k<sup>1</sup> can be attained by calculating the slope of the liner relationship between i(V)/ν 1/2 and ν 1/2 at every fixed potential,

evaluating the proportion of charge storage from capacitance and calculating capacity contributions from pseudocapacitance and diffusion (Chen et al., 2017; Wang et al., 2018; Zhang et al., 2018a). As scan rate increases, capacitive distribution increases in the same time and achieves 42.7% at a scan rate of 0.2 mV s −1 (**Figures 5B,C**). When tested at a scan rate of 10 mV s−<sup>1</sup> , the capacitive distribution can reach as high as 84.6%. The Li<sup>+</sup> ions are easy to make absorption or reactions in macromicropores, resulting in the pseudocapacitive behavior at edges or on the surface of amorphous carbon. The higher ratio of pseudocapacitance allows charges to transfer in an effective way. With rational design and modification of AAC anode, excellent rate capability can be accomplished by reallocating the proportion of capacitive contribution to a more dominant occupancy. In order to confirm the smaller resistance of AAC, the electrochemical impedance spectroscopies (EIS) for AAC and AC anodes are carried out in **Figure 5D**. Obviously, the chargetransfer resistance (Rct) of AAC located in high frequency is much smaller than AC, reducing from 319.5 to 89.07 , which results in the considerably improved electronic conductivity and reaction kinetics (Wu et al., 2016b, 2017). In accordance with capacity distribution, the effective rate response of AAC can be ascribed to the excellent reaction dynamics (Wang et al., 2018).

The increasing price for Li-containing cathodes incents researchers to seek for alternative energy storage system, among which SIB is suitable and promising. Therefore, with the instruction of the employment in LIBs, the AAC can be also applied for SIBs (**Figure 6**). **Figure 6A** and **Figure S3C** present the CV curves of AAC and AC performed at a sweep rate of 0.1 mV s−<sup>1</sup> within 0.01-3.0 V, respectively. Analyzing the CV curve of AAC anode (**Figure 6A**), it can observe a broad hump with two protuberant peaks at 1.0 and 0.4 V, which are disappeared in next two cycles. These irreversible peaks should be assigned to the formation of protective SEI film and undesirable reactions of oxygen-containing functional groups (Zhu et al., 2018), and this analogous phenomenon is also demonstrated in CV profiles of AC anode (**Figure S3D**) While a redox peak at 0.03 and 0.06 V is observed, referring to Na+ ion insertion/extraction. The overlapped CV profiles after the first cycle indicate a fully covered SEI film on AAC, suggesting the highly reversible redox reaction and stable structure. As shown in **Figure S2B**, AAC anode can exhibit an initial charge capacity of 328.7 mAh g−<sup>1</sup> with the coulombic efficiency about 70%, while it maintains almost 100% columbic efficiency in the succeeding charge/discharge cycles. It is worth noticing that the initial coulombic efficiency of AAC is surprisingly higher than AC (**Figure S2B**), which can be reasonably elucidated to the activation of AAC. Based on adsorption-insertion sodium storage mechanism of amorphous carbon (Hou et al., 2017a; Li et al., 2017c), increasing amount of active sites (pores and defects) can attract more Na<sup>+</sup> ion and the enlarged interlayer spacing is found suitable for Na<sup>+</sup> intercalation/extraction between interlayers.

The comparison of AAC and AC anodes is measured sequentially at different current rates from 0.1 to 5 A g−<sup>1</sup> (**Figures 6C,D**). The reversible capacities of AAC anode at every rate are much higher than AC anode, reaching to 324.8, 321.6, 312.1, 298.9, 282.3, and 272.4 mAh g−<sup>1</sup> at 0.1, 0.2, 0.5, 1.0, 2.0, and 5.0 A g−<sup>1</sup> , respectively. Surprisingly, it can recover to 343.5 mAh g−<sup>1</sup> when current density rolls back to 0.5 A g−<sup>1</sup> , indicating the superior rate performance. Besides, the long-term galvanostatic charge/discharge property of AAC and AC anodes are presented in **Figures 6E,G**. AAC anode obtains a charge capacity of 361.7 mAh g−<sup>1</sup> at 0.5 A g−<sup>1</sup> after 200 cycles and maintains 189.1 mAh g−<sup>1</sup> at high rate of 2 A g−<sup>1</sup> beyond 1000 cycles, surpassing the AC anode, which possesses awfully low charge capacity of only 113.1 mAh g−<sup>1</sup> and 41.1 mAh g−<sup>1</sup> at 0.5 A g−<sup>1</sup> and 2 A g−<sup>1</sup> , respectively, with continuous degradation. The above enhanced cycling performance is originated from the three-dimension hierarchical porous structure equipped with well-fabricated SEI film as well as intrinsic shortened transport routes and facile intercalation for Na<sup>+</sup> (Ou et al., 2017).

EIS measurements (**Figure 6B**) validate the excellent cyclic and rate capability of AAC by simulating the equivalent circuits. **Figure 6B** displays that AAC exhibits satisfactory low impedance Rct of about 26.2 , which is much smaller than that of AC (160.3 ), proving the low charge transfer resistance and fluent reaction process in AAC. Furthermore, the calculation of pseudocapacitive contributions at various sweep rates are also carried out in **Figure 6F** and **Figure S4**. Capacitive behavior of AAC is accountable for 78.9% of total capacity at sweep rate of 0.2 mV s−<sup>1</sup> , signifying that the critical capacitive behavior induced by larger specific surface area ensures the outstanding cyclic and rate performance of AAC. Compared with other published carbon anodes (**Tables S1** and **S2**), it confirms that AAC exhibits attracting electrochemical performance for both lithium and sodium storage.

### CONCLUSION

In general, the activated amorphous carbon (AAC) with porous structure derived from renewable camellia pollen grains is synthesized by a facile approach of homogeneous KOH activation followed by carbonization. AAC exhibits the enhancement in specific surface area (660.04 m<sup>2</sup> g −1 ), the enrichment in micro/mesopores (0.29 cm<sup>3</sup> g −1 for accumulated pore volume) and the enlargement in interlayer distance (from 0.352 to 0.409 nm) as well. Furthermore, the modified microstructure can facilitate the infiltration for electrolyte

### REFERENCES


into the interior of AAC electrode. On the other hand, it accesses the efficient transportation of Li+/Na<sup>+</sup> ions by lowering the charge transfer resistance and shorten the diffusion paths. It is noted that the intercalation and adsorption of Li+/Na<sup>+</sup> ion are coincidentally promoted, leading to the outstanding electrochemical performances. As a result, AAC anode surprisingly achieves a reversible charge capacity of 691.7 mAh g−<sup>1</sup> after 1200 cycles at 2.0 A g−<sup>1</sup> and 213.5 mAh g−<sup>1</sup> at ultrahigh rate of 5.0 A g−<sup>1</sup> for LIBs. Moreover, AAC delivers reversible capacities of 324.8, 321.6, 312.1, 298.9, 282.3, and 272.4 mAh g−<sup>1</sup> at 0.1, 0.2, 0.5, 1.0, 2.0, and 5.0 A g−<sup>1</sup> in gradient rate tests for SIBs. It has been discussed in plenty of articles that, concentration of activating reagent (KOH in this article) and activation duration are mutually responsible for the eventualities, which means that with rational adjustments can we delicately tailor the microstructure of pollen-grainderived amorphous carbon. Alone with the particularity of pollen grains as a renewable and cost-effective biomass, the activated amorphous carbon with such intriguing structure is likely to become the future promising anode materials for lithium and sodium ion batteries.

### AUTHOR CONTRIBUTIONS

KX and JX conducted the experiments, CY is the supervisor of this research work. YL and XO helped writing. KX, YL, and XO performed the characterization and data analysis. All authors involved the analysis of experimental data and manuscript preparation.

### ACKNOWLEDGMENTS

We gratefully acknowledge the financial support from the Guangdong Power Grid Co., Ltd (Grant No. GDKJXM20161890).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2018.00366/full#supplementary-material


for high performance lithium/sodium storage. Adv. Funct. Mater. 28:1706294. doi: 10.1002/adfm.201706294


spheres as low-cost and high-capacity anodes for lithium-ion batteries. Nano Energy 25, 120–127. doi: 10.1016/j.nanoen.2016.04.043


**Conflict of Interest Statement:** KX, YL, WS, and GZ were employed by company Electric Power Research Institute of Guangdong Power Grid Co., Ltd.

The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Xu, Li, Xiong, Ou, Su, Zhong and Yang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Size-Tunable Natural Mineral-Molybdenite for Lithium-Ion Batteries Toward: Enhanced Storage Capacity and Quicken Ions Transferring

Feng Jiang<sup>1</sup> , Sijie Li 2,3, Peng Ge2,3, Honghu Tang<sup>2</sup> , Sultan A. Khoso<sup>1</sup> , Chenyang Zhang<sup>1</sup> , Yue Yang<sup>1</sup> , Hongshuai Hou2,3, Yuehua Hu<sup>1</sup> , Wei Sun<sup>1</sup> \* and Xiaobo Ji 2,3 \*

*<sup>1</sup> School of Minerals Processing and Bioengineering, Central South University, Changsha, China, <sup>2</sup> College of Chemistry and Chemical Engineering, Central South University, Changsha, China, <sup>3</sup> State Key Laboratory of Powder Metallurgy, Central South University, Changsha, China*

### Edited by:

*Qiaobao Zhang, Xiamen University, China*

#### Reviewed by:

*Huixin Chen, Xiamen Institute of Rare Earth Materials, China Yunhua Xu, Tianjin University, China Xianwen Wu, Jishou University, China*

#### \*Correspondence:

*Wei Sun sunmenghu@csu.edu.cn Xiaobo Ji xji@csu.edu.cn*

#### Specialty section:

*This article was submitted to Physical Chemistry and Chemical Physics, a section of the journal Frontiers in Chemistry*

> Received: *06 July 2018* Accepted: *13 August 2018* Published: *28 August 2018*

#### Citation:

*Jiang F, Li S, Ge P, Tang H, Khoso SA, Zhang C, Yang Y, Hou H, Hu Y, Sun W and Ji X (2018) Size-Tunable Natural Mineral-Molybdenite for Lithium-Ion Batteries Toward: Enhanced Storage Capacity and Quicken Ions Transferring. Front. Chem. 6:389. doi: 10.3389/fchem.2018.00389* Restricted by the dissatisfied capacity of traditional materials, lithium-ion batteries (LIBs) still suffer from the low energy-density. The pursuing of natural electrode resources with high lithium-storage capability has triggered a plenty of activities. Through the hydro-refining process of raw molybdenite ore, containing crushing–grinding, flotation, exfoliation, and gradient centrifugation, 2D molybdenum disulfide (MoS2) with high purity is massively obtained. The effective tailoring process further induce various sizes (5, 2, 1 and 90 nm) of sheets, accompanying with the increasing of active sites and defects. Utilized as LIB anodes, size-tuning could serve crucial roles on the electrochemical properties. Among them, MoS2-1µm delivers an initial charge capacity of 904 mAh g −1 , reaching up to 1,337 mAh g−<sup>1</sup> over 125 loops at 0.1 A g−<sup>1</sup> . Even at 5.0 A g−<sup>1</sup> , a considerable capacity of 682 mAh g−<sup>1</sup> is remained. Detailedly analyzing kinetic origins reveals that size-controlling would bring about lowered charge transfer resistance and quicken ions diffusion. The work is anticipated to shed light on the effect of different MoS<sup>2</sup> sheet sizes on Li-capacity ability and provides a promising strategy for the commercial-scale production of natural mineral as high-capacity anodes.

Keywords: natural molybdenite ore, molybdenum disulfide, size effect, lithium-ion battery, electrochemical performance

## INTRODUCTION

Lithium-ion batteries (LIBs) are renewable energy storage devices commonly used in consumer electronics, high-power tools, and electric vehicles because of their excellent capacities, such as high energy density, long cycle life, low self-discharge, no memory effect (Li et al., 2017; Yang et al., 2017; Zhang et al., 2018; Zheng et al., 2018). Graphite is the current commercial anode material due to its flat potential profile and great structure stability during cycling. However, six carbon atoms are required to accommodate one Li ion, and the theoretical specific capacity (372 mAh g −1 ) of graphite is insufficient to meet the increasing requirements of the ever-growing market of high-performance batteries(Shim and Striebel, 2003; Yoshio et al., 2003, 2004).

Two-dimensional (2D) metal dichalcogenides (MDCs) as an alternative material for graphite has received considerable attention(Chhowalla et al., 2013; Yang et al., 2015; Zhang et al., 2015; Ge et al., 2018a,b). Among them, molybdenum disulfide is a typical graphene analog, in which two adjacent S-Mo-S layers are linked by weak van der Waals forces. Given its novel mechanical, optical, electrical, and electrochemical properties, MoS<sup>2</sup> has been widely studied for different applications in lubricants(Xiao et al., 2017; Wu et al., 2018), photocatalytic degradation catalysts(Li et al., 2014; Su et al., 2016; Liu et al., 2018), sensors(Liu et al., 2014; Wang and Ni, 2014), electrocatalytic hydrogen generation(Gao et al., 2015a,b; Zhu et al., 2015; Geng et al., 2016), field-effect transistors(Dankert et al., 2014; Roy et al., 2014), supercapacitors(Ma et al., 2013; Acerce et al., 2015), and electrode material for batteries(Liang et al., 2011; Yang et al., 2015; Hai et al., 2018). Compared with graphite, MoS<sup>2</sup> has a wider lattice spacing (∼0.65 nm), which is conducive to rapid insertion and extraction of alkali metal ions. After insertion, LixMoS<sup>2</sup> can further react with Li<sup>+</sup> ions to form Li2S and Mo atom, and the theoretical specific capacity of MoS<sup>2</sup> in LIBs is 670 mAh g−<sup>1</sup> , which is much higher than that of graphite (Stephenson et al., 2014). Meanwhile, a number of studies reported that the capacity of MoS<sup>2</sup> can reach >1,000 mAh g−<sup>1</sup> , which arises from Mo atoms accommodating a large amount of Li ions over prolonged discharging process (Wang et al., 2018).

Most previous studies synthesized MoS<sup>2</sup> by chemical methods to obtain nanosheets with desired size and thickness. Hydrothermal, chemical vapor deposition, and hot injection are typical approaches that use molybdenum salts as precursors (Altavilla et al., 2011; Wang et al., 2014). Although the aforementioned chemical synthetic methods can be used for the large-scale preparation of MoS<sup>2</sup> nanosheets, their industrial applications are limited by their rigid reaction conditions and environmentally pernicious reactants (Yang et al., 2016, 2018; Zhang et al., 2016). MoS<sup>2</sup> is abundant in the form of molybdenite in nature and is generally extracted and processed into molybdenum metal and compounds through beneficiation, smelting and chemical synthesis. Thus, fabrication of MoS<sup>2</sup> materials directly from natural molybdenite ore can eliminate many intermedia complex processes and reduce synthetic contaminants. In addition, the appropriate size of MoS<sup>2</sup> for LIBs remains unknown. In view of the fact that size exerts a noteworthy influence on the electrochemical properties of many materials (Kim et al., 2005; Liu et al., 2005; Drezen et al., 2007; Wagemaker et al., 2007; Kiani et al., 2010; Jiang et al., 2017), understanding the effects of different sizes of MoS<sup>2</sup> on battery performance and electrochemical properties is important to application of MoS<sup>2</sup> in LIBs.

Herein, a hydro-refining technology combining crushinggrinding, flotation, mechanical exfoliation, and classification processes was developed to prepare a series of size-controlled MoS<sup>2</sup> sheets directly from natural raw molybdenite ore. This method is simple, eco-friendly, and high-yielding. When used the as-prepared MoS<sup>2</sup> sheets as LIB anodes, size displays an important effect on electrochemical properties. Among them, the MoS2-1µm electrode demonstrated excellent electrochemical properties with lower charge transfer resistance and rapider Li ions diffusion, delivering a higher specific capacity and initial coulombic efficiency. These results suggest the proper MoS<sup>2</sup> sheet size for LIBs and indicate the present approach is promising for industrial-scale production of natural molybdenite as highcapacity anodes.

### MATERIALS AND METHODS

### Materials

Natural raw ore (rock size: 5–10 cm, MoS<sup>2</sup> content: 1–2%) was received from China Molybdenum Co., Ltd. Raw ore was crushed to small stones (particle size ∼2 mm) and then ball-milled with water at a concentration of 66.6% to reduce the granularity. Ballmilled production, which is also called pulp (particle size: 75% <74µm), was transferred to flotation cell, and water was added to adjust the concentration to 33%. In brief, 333 mg/L sodium silicate as depressant, 35 mg/L kerosene as molybdenite collector, and 15 mg/L terpineol as foaming agent were added sequentially to the pulp during agitation. Then, the pulp was aerated, and flotation froth was generated above the pulp and collected as the rough molybdenite concentrate (MoS<sup>2</sup> content: 2–5%), which then was reground to a fineness of 85% <37µm by stirred mill. Finally, the reground rough concentrate was flotation cleaned eight times to improve the molybdenite concentrate grade. In the first cleaning operation, 2 g/L sodium sulfide was added to the pulp as the other sulfide minerals' depressant. Then, the obtained concentrate froth was transferred to the next cleaning operation, in which the sodium sulfide dosage was half of that used in the previous step. The final concentrate froth from the eighth cleaning operation was filtered and dried to achieve molybdenite concentrate (MoS<sup>2</sup> content: ∼92%).

Differently sized MoS<sup>2</sup> sheets were prepared through an intense shearing process. Molybdenite concentrate (10 g), polyvinylpyrrolidone-K30 (0.25 g, PVP-K30), and deionized water (500 mL) were placed in a stainless steel homogenizer. The homogenizer was run at 12,000 rpm for 5 h to exfoliate the molybdenite content and acquire a MoS<sup>2</sup> suspension. The homogeneous dispersion was gradient centrifuged at 1,000, 3,000, 5,500, and 10,000 rpm, and the precipitates were collected and rinsed by deionized water several times to remove the residual PVP. Afterward, the as-prepared differently sized MoS<sup>2</sup> sheets were dried at 60◦C in a vacuum oven for 24 h.

### Material Characterization

The crystal structure of the as-prepared materials was identified by X-ray diffraction (XRD, Bruker D8 diffractometer with monochromatic Cu Kα radiation and wavelength of 1.5406 Å). The composition of the samples was characterized by X-ray fluorescence (XRF). The particle size distribution was measured by laser diffraction (Malvern Mastersizer 2000). The morphology was analyzed by field emission scanning electron microscopy (FEI Quanta 200, Japan) and atomic force microscopy (AFM, Bruker Multimode V, Germany).

### Electrochemical Characterization

The active materials, carboxymethyl cellulose, and conductive additive (Super P, carbon black) were mixed in a weight ratio of 75:15:15 by using deionized water as the solvent. Then, the steady slurry was evenly painted on a copper foil. After drying at 80◦C in a vacuum oven for 12 h, the copper foil was cut into wafer electrodes. The mass of the active material in each electrode was approximately 1.0 mg cm−<sup>2</sup> . The CR2016 cointype cells were assembled in an argon-filled glovebox (MBRAUN, Germany) by using as-prepared electrodes as the anode, metallic lithium disk as the counter electrode, and LiClO<sup>4</sup> (1 M) in ethylene carbonate and dimethyl carbonate (1:1, v/v) as the electrolyte. The capacities of Li-ion half cells were measured at different current densities in the voltage range of 0.01–3 V vs. Li+/Li by using an Arbin battery testing system (BT2000). Cyclic voltammetry (CV) was performed by CHI660D electrochemical station (Shanghai Chenhua, China) in the voltage range of 0.01– 3 V vs. Li+/Li. Electrochemical impedance spectroscopy (EIS) was performed at the frequency range of 0.01 Hz to 100 kHz, and the excitation amplitude applied to the cells was 5 mV. All of the electrochemical tests were conducted at a temperature of 25◦C.

### RESULTS AND DISCUSSION

A schematic showing the hydro-refining process of preparing a series of size-controlled MoS<sup>2</sup> sheets directly from natural raw ore is illustrated in **Figure 1**. Initially, the particle size of natural raw ore is reduced by crushing and ball milling. Using flotation, molybdenite in the form of concentrate froth is separated from other nontarget minerals, and the recovery rate of molybdenite is ∼85%. The obtained molybdenite concentrate is further downsized by a homogenizer, which has a strong shearing force to exfoliate bulky molybdenite (i.e., MoS2). Finally, the MoS<sup>2</sup> suspension is size-classified via high-speed gradient centrifugation. This method is low cost, environmental friendly, high-yielding, and is very promising for the large-scale preparation of MoS<sup>2</sup> sheets with various sizes.

The chemical composition of the natural raw ore and molybdenite concentrate is presented in **Table 1**. In natural raw ore, the dominant elements are O and Si, while the Mo content is only 0.82%, thus a facile and low-cost flotation process is indispensable to obtain pure molybdenite concentrate (Jiangang et al., 2012; Liu et al., 2012a). After flotation, the Mo content can reach to 55%, representing the high purity of the molybdenite concentrate. The slight oxidation of the natural molybdenite surface is due to the exposure to oxidative environment. The crystal structures and phases of the molybdenite concentrate and the differently sized MoS<sup>2</sup> are investigated by XRD (**Figure 2A**). All of these samples exhibit similar XRD patterns, which match well with the 2H MoS<sup>2</sup> phase (JCPDS no. 37-1492) (Ding et al., 2012; Xie et al., 2015; Sun et al., 2017). No extra peaks appear in the pattern, indicating their high purity, which agrees well with the XRF results. The peak at approximately 14.4◦ is the characteristic peak of (002) facet. Decreasing peak intensity and broadening peak width of (002) facet signify the thickness reduction of MoS<sup>2</sup> sheets (Wang et al., 2013d). Using the results from the XRD patterns, we calculate the grain parameters of each sample by the Scherrer equation:

$$D = \mathbf{K}\lambda / \beta \cos \theta \tag{1}$$

Where, D is the grain size, K is the Scherrer constant (0.89), λ is the diffraction light (X-ray) wavelength (0.15406 nm), β is the full width at half maximum, and θ is the Bragg angle. As shown in **Table 2**, the MoS2-90 nm sample has the smallest grain size among them. Moreover, the volume average diameters of the samples are tested with a laser diffraction-based particle size analyzer. As shown in **Figure 2B**, the volume average diameters of the molybdenite concentrate and differently sized MoS<sup>2</sup> are 25.964, 5.346, 1.978, 1.023, and 0.092µm, respectively.

The morphological of the samples are conducted by SEM and shown in **Figure 3**. **Figure 3A1** shows the morphology of the molybdenite concentrate where molybdenite particles exhibit various textures (flaky, blocky, and irregular shapes), and their size is mainly tens of microns, which can be attributed to the complex factors in natural mineralization. In addition, several small pieces of debris are found on the surface of large molybdenite particles with a size distribution from a few microns to submicron. From the higher-magnification observations, stacked compacted 2D layer structure is found distinctly in **Figures 3A2,A3**. By contrast, MoS2-5µm, MoS2- 2µm, and MoS2-1µm show a lamellar morphology. As shown in **Figures 3B1–B3**, several thick sheets with size of ∼5µm are distributed in the MoS2-5µm sample, which thickness is around 300 nm. Meanwhile, stratified structure and uneven edges are detected, accompanying with an increasing of active sites and defects. In the SEM images of MoS2-2µm and MoS2-1µm, small sheets with average sizes of ∼1µm and ∼500 nm can be observed. The curved sheets shown in the higher-magnification images of **Figures 3C3,D3** indicate the thinness and flexibility of the MoS<sup>2</sup> sheets, which significantly ease the volume expansion during the charge and discharge cycles and enhance the stability of the batteries. **Figures 3E1,E2** show the compact agglomeration of nano-MoS<sup>2</sup> sheets in the MoS2-90 nm sample, revealing the strong tendency of MoS<sup>2</sup> nanosheets to aggregate because of their high surface area and energy. This agglomeration dramatically decreases the active sites of the material and hinder Li<sup>+</sup> diffusion, which led to a low capacity. **Figure 3F** displays the compositions of molybdenite concentrate by energy disperse spectroscopy (EDS) analysis. No evident incidental element appears, and the atomic ratio of S to Mo is approximately 2, which further demonstrate the high purity of the molybdenite concentrate obtained from natural raw ore.

For exploring the crystalline characteristics of MoS<sup>2</sup> sample, TEM and HRTEM tests with various magnifications are performed. As shown in **Figures S1A,B**, thin sheets are detected in MoS2-1µm, accompanying with clear 2D layer structure. **Figure S1C** shows the HRTEM image of MoS2-1µm, revealing the abundant defects existing in MoS<sup>2</sup> sheets. Stripes spaced 0.273 nm apart in the insetmap are in good accordance with the (100) facet of MoS2, as well as the single-crystal SAED pattern of MoS2-1µm shows the typical hexagonal spot pattern (**Figure S1D**). AFM tests are further carried out to obtain detailed information about the morphologies of MoS2-5µm, MoS2-2µm, MoS2-1µm, and MoS2-90 nm. As shown in **Figures S2A2–C2**, the thickness of the MoS<sup>2</sup> sheets in the MoS2-5µm, MoS2- 2µm, and MoS2-1µm samples gradually decrease from ∼330 to ∼170 nm and then to ∼100 nm. The same trend is observed

FIGURE 1 | Schematic plan of hydro-refining process of natural raw ore.

TABLE 1 | Chemical composition of natural raw ore and molybdenite concentrate.


for the sheet diameter (**Figures S2A1–C1**). Stratified structures and rough edges can also be observed in the 3D plots (**Figures S2A3–C3**), indicating more active sites can be exposed for Li ions. The image of the MoS2-90 nm sample shown in **Figure S2D1** displays three irregular particles with a thickness of ∼230 nm and a diameter of ∼1µm. Similar to the SEM results, the AFM findings indicate that these uncommon particles are the agglomeration of nano-MoS<sup>2</sup> sheets. When observing at a small height scale, two pieces of thin films are visible with a thickness of ∼0.65 nm, indicating single-layer MoS<sup>2</sup> films distributing in the MoS2-90 nm sample.

The electrochemical properties of the as-prepared samples are measured by galvanostatic charge–discharge test at various current densities. **Figure 4A** shows the initial charge and discharge curves of molybdenite concentrate and MoS<sup>2</sup> samples at 100 mA g−<sup>1</sup> , where two potential plateaus at approximately 1.1 and 0.6 V vs. Li/Li<sup>+</sup> in the first discharge (lithiation) of the electrodes are observed. The first plateau at 1.1 V could be attributed to the intercalation of Li<sup>+</sup> into MoS<sup>2</sup> interlayers (MoS<sup>2</sup> + xLi<sup>+</sup> + xe <sup>−</sup> → LixMoS2), and the low plateau at 0.6 V is due to the conversion reaction of LixMoS<sup>2</sup> to Mo metal and Li2S (LixMoS<sup>2</sup> + (4 – x)Li<sup>+</sup> + (4 – x)e<sup>−</sup> → Mo + 2Li2S). Only one significant potential plateau at approximately 2.3 V appeared in the first charge (delithiation) process, and it corresponds to the delithiation of Li2S (Li2S – 2e<sup>−</sup> → 2Li<sup>+</sup> + S). This result demonstrates that the conversion reaction is irreversible (Xiao et al., 2010; Stephenson et al., 2014). The electrochemical behavior is further analyzed by CV (**Figure 4B**). In the first cathodic sweep, two peaks appear at approximately 0.93 and 0.23 V, which are attributed to the insertion and conversion reactions, respectively. Meanwhile, these two peaks weaken in subsequent cathodic cycles. Instead, a sharp reduction peak arises at approximately 1.84 V, which matches well with the behavior in Li-S battery and corresponds to the reaction of S to Li2S (Ji and Nazar, 2010; Elazari et al., 2011). In the anodic sweep, one shallow peak at 1.69 V and one sharp peak at 2.33 V are observed. The first oxidation peak is due to the delithiation of residual LixMoS2, and the latter peak represents

TABLE 2 | Crystal parameters of molybdenite concentrate and MoS2 samples.


the conversion of Li2S to S (Song et al., 2013; Stephenson et al., 2014).

As shown in **Figure 4A**, the initial discharge specific capacities of molybdenite concentrate, MoS2-5µm, MoS2-2µm, MoS2- 1µm, and MoS2-90 nm are 688, 779, 868, 1134, and 1004 mAh g −1 at 100 mA g−<sup>1</sup> , respectively, while the initial charge capacities are 589, 653, 555, 904, and 611 mAh g−<sup>1</sup> . Among them, the MoS2-1µm has a higher capacity owing to its compared richer active sites. **Figure 4C** shows the cycling performance of the molybdenite concentrate and MoS<sup>2</sup> samples at 100 mA g−<sup>1</sup> . The molybdenite concentrate exhibits an unsatisfied stability, which capacity gradually decreases to 217 mAh g−<sup>1</sup> after 125 cycles, showing a low capacity retention of 37%. MoS2-5µm displays a specific capacity of ∼600 mAh g−<sup>1</sup> before 50 cycles with no evident fading, while fades quickly to 355 mAh g−<sup>1</sup> . The poor cycling stabilities of the molybdenite concentrate and MoS2- 5µm can be ascribed to the large volume expansion of bulk MoS<sup>2</sup> during repeated charge/discharge processes, causing the harmful shedding of active materials. Meanwhile, MoS2-2µm, MoS2-1µm, and MoS2-90 nm show excellent stability without any capacity decay. As shown, the capacities of MoS2-2µm, MoS2-1µm, and MoS2-90 nm increase with the cycling going on, reaching up to 1013, 1337, and 881 mAh g−<sup>1</sup> after 125 cycles. The data reported here are higher than most of the reported works (**Table 3**). The promotion in capacity may be attributed to the increased Mo atoms created by the irreversible redox reaction during repeated charge/discharge processes, bringing about better conductivity. Meanwhile, Mo atoms accommodate a large amount of Li ions over prolonged discharging process, increasing the electrode's Li-capacity. The significant differences between these prepared samples indicate that decreasing the particle size of MoS<sup>2</sup> can significantly improve the cycling stability and capacity of batteries due to the stronger and more flexible structure and more active spots. However, MoS2-90 nm displays a lower capacity than MoS2-1µm may due to the particle agglomeration, accompanying with the reduction in active spots.

**Figure 4D** shows the coulombic efficiencies of the MoS<sup>2</sup> samples at 100 mA g−<sup>1</sup> . The initial coulombic efficiencies of MoS2-5µm, MoS2-2µm, MoS2-1µm, and MoS2-90 m are 83.9, 63.9, 79.7, and 60.9%, respectively, which rapidly increase to

>97% after five cycles. The significant difference in initial coulombic efficiency between MoS<sup>2</sup> samples can be explained through the electrochemical behavior during the first lithiation process. Unlike the conversion reaction, Li ion intercalation is a reversible reaction. Thus, a high ratio of intercalation capacity can result in a high initial coulomb efficiency. As shown in **Figure 4E**, the intercalation capacity ratios of MoS2-5µm, MoS2-2µm, MoS2-1µm, and MoS2-90 m are calculated to be 23.06, 9.06, 23.05, and 8.26%, respectively, which correspond well to the initial coulombic efficiencies. Moreover, the initial coulombic efficiency is an important parameter that determines the industrial application feasibility of electrode materials. Individual MoS2-5µm and MoS2-1µm have much higher initial coulombic efficiencies, suggesting that they are more conducive to the application of full batteries than MoS2-2µm and MoS2- 90 nm.

The galvanostatic charge and discharge profiles of four MoS<sup>2</sup> electrodes at 100 mA g−<sup>1</sup> are shown in **Figures 4F–I**. **Figure 4H** shows that, different from the initial discharge curve, a new potential plateau emerges at 2.0 V vs. Li/Li+, and the two aforementioned potential plateaus at 1.1 and 0.6 V disappear in the second discharge profile. This appearance indicates that the dominant reaction of the discharge process turns into S lithiation (S + 2Li<sup>+</sup> + 2e<sup>−</sup> → Li2S) (Chang et al., 2013; Zhu et al., 2014), which is in good accordance with the aforementioned CV results. **Figure 4F,G,I** show the charge and discharge curves of the three other electrodes, which are similar to that of the MoS2-1µm electrode.

.

**Figure 5A** shows the rate performances of MoS2-5µm, MoS2- 2µm, MoS2-1µm, and MoS2-90nm. Apparently, the capacity of MoS2-1µm is much higher than those of MoS2-5µm, MoS2- 2µm, and MoS2-90nm. The charge capacities of the MoS2-1µm anode at 0.5, 1.0, 2.0, and 5.0 A g−<sup>1</sup> are 931, 900, 857, and 682 mAh g−<sup>1</sup> , respectively. When the current density reverts to 0.1 A g−<sup>1</sup> , the capacity recovers to a high value of 1,239 mAh g−<sup>1</sup> , indicating the strong tolerance of the electrode for the rapid charge–discharge process and the remarkable capacity recoverability of the MoS2-1µm electrode. Meanwhile, the charge capacities of MoS2-5µm are 516, 464, and 342 mAh g−<sup>1</sup> at 0.5, 1.0, 2.0, and 5.0 A g−<sup>1</sup> , respectively, and then reverts to 597 mAh g−<sup>1</sup> at 0.1 A g−<sup>1</sup> , which is close to the initial capacity. However, along with increasing loops, the capacity declines following a similar pattern to the previous result. The charge capacities of MoS2-2µm and MoS2-90 nm are 355 and 217 mAh g−<sup>1</sup> at 1.0 A g−<sup>1</sup> and 146 and 76 mAh g−<sup>1</sup> at 5.0 A g−<sup>1</sup> , respectively, which are unsatisfactory. **Figures 5B–E** display the comparation of the charge and discharge curves

of MoS2-5µm, MoS2-2µm, MoS2-1µm, and MoS2-90 nm at various current densities. As shown in **Figure 5E**, the MoS2- 1µm electrode keeps a similar charge and discharge curves even at a high current density, as well as considerable capacity retention, further revealing its excellent rate performance. While for MoS2-5µm, MoS2-2µm, and MoS2-90 nm, it is difficult for them to maintain the original charge and discharge behavior at high current densities, leading to a sharp declining in capacity (**Figures 5B,C,F**).

To confirm the difference in electrochemical performance of the differentially expressed MoS2, EIS tests are performed to analyze the electronic conductivity and ion diffusion rate of the samples. **Figure 6A** shows the Nyquist plots at fully uncharged-undischarged state, accompanied by fitted equivalent circuit. The semicircular loop at the high-middle frequencies is related to the resistance of solid electrolyte interface and charge transfer resistance (Rct), while the slope line at low frequencies represents the Warburg impedance, which is connected to Li ion diffusion of the electrode materials(Wang et al., 2013a). TABLE 3 | Composition of this work and other previous reported results.


The smaller semicircle of MoS2-1µm compared with MoS2- 5µm, MoS2-2µm, and MoS2-90 nm indicates a lower Rct. Thus, MoS2-1µm is more conducive to charge transfer compared with the other samples (Jiang et al., 2017). **Figure 6D** shows the relationship between Z<sup>r</sup> and negative square root of angular frequency (ω −1/2 ) in the low-frequency region at fully uncharged–undischarged state. Using the slope of the fitted line (Warburg coefficient), the Li ion diffusion coefficient can be calculated according to the following equation (Wu et al., 2016, 2017; Li et al., 2018):

$$D\_{Li^{+}} = \mathfrak{0.5R}^{2}T^{2}/A^{2}n^{4}F^{4}C^{2}\sigma^{2} \tag{2}$$

where DLi+ is the Li ion diffusion coefficient, R is the gas constant (8.314 J mol−<sup>1</sup> K −1 ), T is the absolute temperature (298 K), A is the area of the electrode (1.53 cm<sup>2</sup> ), n is the transfer electrons (for Li+, n = 1), F is the Faraday constant (96,485 C mol−<sup>1</sup> ), C is the Li ion lattice concentration (0.001 mol cm−<sup>2</sup> ), and σ is the Warburg coefficient. As shown in **Figure 6G**, at fully uncharged– undischarged state, the DLi+ values of MoS2-5µm, MoS2-2µm, MoS2-1µm, and MoS2-90 nm are 7 × 10−15, 4.68 × 10−15, 7.56 × 10−13, and 3.21 × 10−<sup>15</sup> cm<sup>2</sup> s −1 , respectively. Apparently, the Li ion diffusion coefficient of MoS2-1µm is two orders of magnitude larger than those of the three other samples, which can reflect the higher initial capacity of MoS2-1µm (**Figure 4C**).

**Figures 6B,E** show the Nyquist plots at initially discharged to 1.1 V vs. Li/Li<sup>+</sup> state where Li ion intercalation occurs and the corresponding relationship between Z<sup>r</sup> and ω −1/2 . The semicircles of MoS2-1µm and MoS2-5µm are smaller than those of MoS2-2µm, and MoS2-90 nm, respectively. Thus, Rct is lower and charge transfer is much easier for MoS2-1µm and MoS2-5µm than for MoS2-2µm and MoS2-90 nm. The calculated DLi+ values of MoS2-5µm, MoS2-2µm, MoS2-1µm, and MoS2-90 nm at initially discharged to 1.1 V vs. Li/Li<sup>+</sup> state are 2.46 × 10−13, 2.19 × 10−15, 4.80 × 10−13, and 5.43 × 10−<sup>14</sup> cm<sup>2</sup> s −1 , respectively (**Figure 6H**). On the basis of the results of Rct and DLi+, the intensities of the Li ion intercalation can be ranked as MoS2-1µm > MoS2-5µm > MoS2-90 nm > MoS2-2µm. As shown in **Figure 6C**, the semicircles of Nyquist plots, which at initially discharged to 0.6 V vs. Li/Li<sup>+</sup> state

where conversion reaction occurs, gradually enlarge from MoS2- 1µm to MoS2-5µm. Simultaneously, the calculated DLi+ values for MoS2-5µm, MoS2-2µm, MoS2-1µm, and MoS2-90 nm are 7.66 × 10−14, 1.9 × 10−13, 2.41 × 10−12, and 6.47 × 10−<sup>13</sup> cm<sup>2</sup> s −1 , respectively (**Figure 6I**). These results indicate the significantly stronger conversion reactions of MoS2-1µm than other samples.

CV tests are conducted to further investigate the electrochemical kinetics of the as-prepared samples. **Figures 7A,B,D,E** show the CV curves of the MoS<sup>2</sup> samples at different scanning rates, where the four MoS<sup>2</sup> samples display similar CV behaviors. The dominant oxidation and reduction peaks appear at approximately 2.48 and 1.80 V vs. Li/Li+, respectively. Moreover, the peak at 2.48 V splits into two parts, which agrees well with the gradient conversion from element S<sup>8</sup> to polysulfides and then to Li2S (Xiao et al., 2011). As the scan rate increasing, the peak current elevates, and the oxidation peak potential shifts positively while the reduction peak potential toward negatively. As shown in **Figure 7G**, the peak intensities clearly show the following trend: MoS2-1µm > MoS2-5µm > MoS2-90 nm > MoS2-2µm, indicating the largest capacity of MoS2-1µm electrode (Chou et al., 2011). **Figures 7C,F** show the relationship between the peak current and square root of scan rate (v1/<sup>2</sup> ), which can be expressed by the following equation (Wang et al., 2013b; Sun et al., 2017):

$$i\_p = 2.69 \times 10^5 n^{\frac{3}{2}} A D^{\frac{1}{2}} v^{\frac{1}{2}} C\_0 \tag{3}$$

where i<sup>p</sup> is the peak current, v is the scan rate, n is the transfer electrons (for Li<sup>+</sup> , n = 1), A is the area of the electrode (1.53 cm<sup>2</sup> ), D is the Li ion diffusion coefficient, and 1C<sup>0</sup> is the change in Li<sup>+</sup> concentration in the electrochemical reaction. Ion diffusion is a rate-determining step in the electrode. Thus, when scanning at a slow rate (<1 mV s−<sup>1</sup> ), the peak current (ip) varied linearly with the square root of scan rate (v 1/2 ). Hence, the slope can be utilized to characterize the Li ion diffusion coefficient (D). The results suggest that the fitting line slope of MoS2-1µm is higher than that of the other samples (**Figure 7H**), revealing that MoS2-1µm has better Li ion diffusion rate than the other samples, which is in good accordance with the EIS test results.

## CONCLUSIONS

Herein, 2D MoS<sup>2</sup> sheets were successfully prepared from abundant natural raw molybdenite ore by a low-cost, environmental-friendly and high-yielding hydro-refining technology, containing crushing–grinding, flotation, physical exfoliation, and gradient centrifugation. Furthermore, the efficient tailoring and classification processes realized a series of size-controlled (5µm, 2µm, 1µm, 90nm) MoS<sup>2</sup> sheets to improve Li-capacity and stability. When used as LIB anodes, size displayed significant effects on electrochemical performance. The MoS2-1µm electrode demonstrated a higher initial charge capacity of 904 mAh g−<sup>1</sup> , further increasing to 1,337 mAh g−<sup>1</sup> over 125 cycles at 0.1 A g−<sup>1</sup> . The excellent rate performance of the MoS2-1µm electrode showed considerable capacities of 857 and 682 mAh g−<sup>1</sup> at 2.0 and 5.0 A g−<sup>1</sup> , respectively. Owing to extraordinary morphology brought from tailoring craft, the as-prepared sheets offering rich active sites and defects for interacting with Li ions. Meanwhile, flexible structure could relieve volume expansion, significantly promoting the cycling stability. What's more, in-depth electrochemical kinetic analysis disclosed that the MoS2-1µm electrode shows a lower charge transfer resistance and higher Li ion diffusion coefficient at various states, resulted from the successful size-tuning process. This work presents the remarkable effect of different MoS<sup>2</sup> sheet sizes on Li-storage performance and provides a promising strategy for the large-scale production of MoS2-based LIB anodes from natural molybdenite mineral.

### AUTHOR CONTRIBUTIONS

FJ conducted the experiments. WS and XJ are the supervisor of this research work. SL, PG, and SK helped writing. HT and HH helped operating experiments. FJ, SL, HT, CZ, YY and YH performed the characterization and data analysis. All authors involved the analysis of experimental data and manuscript preparation.

### FUNDING

This research was financially supported by the National 111 Project (B14034), collaborative Innovation Center for Clean and Efficient Utilization of Strategic Metal Mineral Resources, Found of State Key Laboratory of Mineral Processing

### REFERENCES


(BGRIMM-KJSKL-2017-13), National Natural Science Foundation of China (51374247, 51704330, 51622406, 21673298 and 21473258), National Key Research and Development Program of China (2017YFB0102000), Scientific Research Starting Foundation of Central South University (202045006).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2018.00389/full#supplementary-material


Liu, G. Y., Lu, Y. P., Zhong, H., Cao, Z. F., and Xu, Z. H. (2012a). A novel approach for preferential flotation recovery of molybdenite from a porphyry copper– molybdenum ore. Miner. Eng. 36–38, 37–44. doi: 10.1016/j.mineng.2012.02.008

Liu, H., Su, D., Zhou, R., Sun, B., Wang, G., and Qiao, S. Z. (2012b). Highly ordered mesoporous MoS<sup>2</sup> with expanded spacing of the (002) crystal plane for ultrafast lithium ion storage. Adv. Energy Mat. 2, 970–975. doi: 10.1002/aenm.201200087


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Jiang, Li, Ge, Tang, Khoso, Zhang, Yang, Hou, Hu, Sun and Ji. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Nitrogen-Doped Carbon Coated WS<sup>2</sup> Nanosheets as Anode for High-Performance Sodium-Ion Batteries

Yong Liu<sup>1</sup> \* † , Huijie Wei 1†, Chao Wang1,2†, Fei Wang<sup>1</sup> , Haichao Wang<sup>1</sup> , Wanhong Zhang<sup>1</sup> , Xianfu Wang<sup>2</sup> \*, Chenglin Yan<sup>2</sup> , Bok H. Kim1,3 and Fengzhang Ren<sup>1</sup> \*

#### Edited by:

*Qiaobao Zhang, Xiamen University, China*

### Reviewed by:

*Baofeng Wang, Shanghai University of Electric Power, China Zhouguang Lu, Southern University of Science and Technology, China Huan Pang, Yangzhou University, China*

#### \*Correspondence:

*Yong Liu liuyong209@haust.edu.cn Xianfu Wang wangxianfu@suda.edu.cn Fengzhang Ren renfz@haust.edu.cn*

*†These authors have contributed equally to this work*

#### Specialty section:

*This article was submitted to Physical Chemistry and Chemical Physics, a section of the journal Frontiers in Chemistry*

> Received: *03 May 2018* Accepted: *04 June 2018* Published: *23 August 2018*

#### Citation:

*Liu Y, Wei H, Wang C, Wang F, Wang H, Zhang W, Wang X, Yan C, Kim BH and Ren F (2018) Nitrogen-Doped Carbon Coated WS2 Nanosheets as Anode for High-Performance Sodium-Ion Batteries. Front. Chem. 6:236. doi: 10.3389/fchem.2018.00236* *<sup>1</sup> The Key Laboratory of Henan Province on Nonferrous Metallic Materials Science and Fabrication Technology, Collaborative Innovation Center of Nonferrous Metals of Henan Province, School of Materials Science and Engineering, Henan University of Science and Technology, Luoyang, China, <sup>2</sup> Jiangsu Provincial Key Laboratory for Advanced Carbon Materials and Wearable Energy Technologies, Collaborative Innovation Center of Suzhou Nano Science and Technology, College of Physics, Optoelectronics and Energy, Soochow Institute for Energy and Materials Innovations, Soochow University, Suzhou, China, <sup>3</sup> Division of Advanced Materials Engineering, Hydrogen and Fuel Cell Research Center, Chonbuk National University, Jeonbuk, South Korea*

Due to the cost-effectiveness of sodium source, sodium-ion batteries (SIBs) have attracted considerable attention. However, SIBs still have some challenges in competing with lithium-ion batteries for practical applications. Particularly, the high rate capability and cycling stability are posing big problems for SIBs. Here, nitrogen-doped carbon-coated WS<sup>2</sup> nanosheets (WS2/NC) were successfully synthesized by a high-temperature solution method, followed by carbonization of polypyrrole. When used as anode electrodes for SIBs, WS2/NC composite exhibited high-rate capacity at 386 and 238.1 mAh g−<sup>1</sup> at 50 and 2,000 mA g−<sup>1</sup> , respectively. Furthermore, even after 400 cycle, the composite electrode could still deliver a capacity of ∼180.1 mAh g−<sup>1</sup> at 1,000 mA g−<sup>1</sup> , corresponding to a capacity loss of 0.09% per cycle. The excellent electrochemical performance could be attributed to the synergistic effect of the highly conductive nature of the nitrogen-doped carbon-coating and WS<sup>2</sup> nanosheets. Results showed that the WS2/NC nanosheets are promising electrode materials for SIBs application.

Keywords: tungsten disulfide, N-doped carbon, nanosheets, sodium ion batteries, electrochemical performances

### INTRODUCTION

Nowadays, lithium ion batteries (LIBs) have become the most widely used energy storage devices for many applications ranging from high performance portable electronics and electrical vehicles to sustainable energy smart grids. The advantages of LIBs include high energy density, long life span, and so on (Armand and Tarascon, 2008; Yang et al., 2011; Lu et al., 2017; Wu et al., 2017; Xu et al., 2017; Geng et al., 2018; Wang et al., 2018; Zhang et al., 2018). However, these large-scale applications may be gradually hindered due to insufficient lithium resource and its uneven distribution in the Earth's crust (Hou et al., 2017a; Fu et al., 2018). As a representative of the promising battery systems, sodium-ion batteries (SIBs) have attracted considerable attention as an alternative to LIBs (Hou et al., 2017b; Wei et al., 2017; Zheng et al., 2017; Tang K. et al., 2018). The interest in SIBs comes from the superiority of the sodium element, including its abundance in nature, low price, and

**286**

negative redox potential (−2.71 V vs. SHE) (Palomares et al., 2012; Li et al., 2013; Slater et al., 2013). However, the practical applications of sodium ion batteries have been hindered by lacking of applicable electrode materials to accommodate sodium ions, which are bigger than Li<sup>+</sup> in radius (Wang X. et al., 2016). Graphite is well-known for being not suitable to host sodium ions since sodium seldom forms stable intercalation compounds with graphite (Komaba et al., 2011; Tian et al., 2017).

Metal sulfides with layered structures, such as MoS<sup>2</sup> (Xie et al., 2015), WS<sup>2</sup> (Wang B. et al., 2016; Wang Y. et al., 2016), SnS (Xiong et al., 2017), VS<sup>2</sup> (Zhou et al., 2017), and SnS<sup>2</sup> (Tu et al., 2017), have been investigated as potential anode materials for SIBs (Xie et al., 2015). The layered structure of these types of materials allows sodium ions to intercalate reversibly. However, the further application of two-dimensional metal sulfides is hampered by their intrinsic limitations (Xie et al., 2015). First of all, their intrinsic low electronic conductivity will prevent the fast electrochemically Na<sup>+</sup> storage. Secondly, these thermally unstable nanomaterials are inclined to restack, due to the high surface energy and interlayer van der Waals attractions (Wang et al., 2014). Furthermore, the remarkable volume expansion during Na <sup>+</sup> intercalation and deintercalation could lead to poor contact between current collector and active materials and the failure of the electrode, resulting in inferior cycling performances (Xie et al., 2015; Luo et al., 2018).

As one of the promising two-dimensional layered metal sulfides as anode materials for SIBs, WS<sup>2</sup> has attracted considerable attention in recent years. However, the reversible capacity of the bare WS<sup>2</sup> is low; the rate performance and long cycling stability of the bare WS<sup>2</sup> must be improved owing to its poor conductivity and serious aggregation during the insertion/extraction process of sodium-ions into/from WS<sup>2</sup> layers (Chen et al., 2014). Hence, several efforts were made to improve their electrochemical performances. For example, tungsten disulfides were composited with ordered mesoporous carbon (CMK-3) (Pang et al., 2017), carbon nanotube-reduced graphene oxide (CNT-rGO) (Wang B. et al., 2016), and other carbon materials (Li et al., 2016) as anode materials for SIBs to buffer volume change, and excellent electrochemical performances were obtained. In addition, nitrogen-doped conductive carbon/WS<sup>2</sup> nanocomposites (WS2- NC) were fabricated by doping N element into conductive carbon with WS<sup>2</sup> nanosheets. The composite electrodes show reversible capacity of ∼360 mAh g−<sup>1</sup> at 100 mA g−<sup>1</sup> , presenting much better electrochemical performances than pristine WS<sup>2</sup> and WS2/conductive carbon (Wang X. et al., 2016).

In this study, we have successfully prepared nitrogen-doped carbon-coated WS<sup>2</sup> nanosheets (WS2/NC) with high content of pyridinic and pyrrolic nitrogen species, and explored their sodium-ion storage performance. The pyridinic and pyrrolic nitrogen species could generate more defects and expose more edge sites in the plane of carbon skeletons, which is beneficial to Na<sup>+</sup> diffusion. On the other hand, nitrogen-doping could greatly increase the electronic conductivity of the carbon layer coated on the WS<sup>2</sup> electrode, which delivers faster charge transfer during the Na<sup>+</sup> intercalation/deintercalation processes. Due to the synergistic effect, the WS2/NC composite exhibited superior long-term cycling performance and rate capability than the pure WS<sup>2</sup> electrode, which may show some potential applications in high performance anode for SIBs.

## MATERIALS AND METHODS

## Synthesis of WS<sup>2</sup> Nanosheets

WS<sup>2</sup> nanosheets were synthesized by a high-temperature solution method as described elsewhere (Cheng et al., 2014). Typically, WCl<sup>6</sup> (2 mmol) was mixed with a mixture consisting of 1 octadecene (15 mL) and oleylamine (30 mL) in a flask (100 mL) at ambient temperature. To eliminate water and oxygen, the temperature of the above mixture was firstly increased to 140◦C under vigorous stirring in Ar gas atmosphere for about 30 min. Then, the solution was quickly heated to ∼300◦C and maintained for 30 min under Ar gas protection. Later, a sulfur solution, which consists of sulfur powder (4 mmol) and oleylamine (10 mL), was added into the flask in 10 min at 300◦C, and kept for 1 hour. After the mixture was cooled down, WS<sup>2</sup> nanosheets were obtained by adding absolute ethanol (ca. 40 mL), collected by centrifugation, and washed repetitively with ethanol. After drying for 30 hours by lyophilization to remove the organic residue, the as-made product was annealed in argon at 500◦C for 2 h, and WS<sup>2</sup> nanosheets were obtained later.

## Synthesis of WS2/NC

One hundred milligrams of as-prepared WS<sup>2</sup> nanosheets were first ultrasonically dispersed in 100 mL distilled water to form a suspension, and the obtained suspension was mixed with 0.1 mL pyrrole monomer and 0.01 g FeCl2. After addition of 0.5 mL of H2O<sup>2</sup> to the mixture, the mixture was stirred for 6 h. After centrifugation, the product was washed repetitively with distilled water, and vacuum-dried for 12 h. Then, WS2/NC nanosheets were obtained, after annealing at 500◦C for 2 h under Ar.

### Characterization

The phases of the as-prepared samples were characterized by X-ray diffraction (XRD, Bruker D8 ADVANCE, Cu kα source). The morphology, microstructure, and the nanometer-range energy dispersive X-ray spectroscopy (EDS) were investigated by scanning electron microscopy (SEM, FEI Quanta 200 FEG) and transmission electron microscopy (TEM, JEM-2100F). Raman spectra were recorded on a Raman spectrometer (LabRAM HR Evolution), using an excitation laser wavelength of 532 nm. Surface elemental analysis was performed on an X-ray photoelectron spectroscopy (XPS, Kratos Axis Ultra Dld, Japan).

### Electrochemical Measurements

The electrochemical tests were carried out using CR2025-type coin cells. To prepare working electrodes, active materials, super P carbon black, and polyvinylidene fluoride (PVDF) in the weight ratio of 7:2:1 were mixed and dispersed in N-methyl pyrrolidone (NMP) to form slurry. The slurry was uniformly coated on the copper foil and then dried at 80◦C overnight in a vacuum oven. The areal loading of active materials on the copper foil was ≈1.0 mg cm−<sup>2</sup> . Coin cells were assembled in an argon-filled glove box with contents of H2O and O<sup>2</sup> below 1 ppm. Sodium foil

was used as counter and reference electrode, glass fiber filters (Whatman, 1820-047) were used as separators. The electrolyte was 1 M NaClO<sup>4</sup> in 1:1 v/v EC/PC. Galvanostatic charge and discharge testing at different specific currents were carried out within 0.01–3.0 V (vs. Na/Na+) on a battery testing system (LAND CT 2001A, Wuhan, China) at room temperature. Cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS) measurement were conducted on a CHI 660E (Chenhua Shanghai, China) electrochemical workstation.

### RESULTS AND DISCUSSION

The phases of the as-prepared WS<sup>2</sup> and the WS2/NC were analyzed by X-ray diffraction, as shown in **Figure 1**. The peaks at 33.8◦ , 59.8◦ , and 69.9◦ correspond well to the (101), (008), and (108) planes of the hexagonal structure of WS<sup>2</sup> (JCPDS No. 84-1398), respectively (Von Lim et al., 2017), indicating successful conversion of the tungsten precursor to layered WS<sup>2</sup> without any discernible impurities (Cheng et al., 2014). The asprepared WS2/NC sample exhibits the similar XRD pattern as the as-prepared WS2, indicating the phase of the WS<sup>2</sup> remains unchanged after the N-doped carbon coating.

The microstructures of the pristine WS<sup>2</sup> and WS2/NC were investigated by SEM and TEM. As displayed in the **Figures 2A,B**, the WS<sup>2</sup> and WS2/NC nanosheets had typical size of 50-200 nm, demonstrating that as-prepared WS<sup>2</sup> and WS2/NC are nano-size. The SEM images clearly show the flower shape of the WS<sup>2</sup> nanosheets, with uniform shapes and flat surfaces. The nanometer size can not only shorten the diffusion distance of Na<sup>+</sup> in charge and discharge process, but also increase the contact between electrode and electrolyte, providing the possibility to improve capacity. From **Figure 2B**, it can be clearly observed that the morphology of WS2/NC is a representative nanosheet. WS2/NC nanometer flake has a large surface area and the thickness of atomic scales, which makes WS2/NC possess a larger specific surface area, could provide sufficient contact area and could be conducive to the insertion/deinsertion of Na+. Further, to investigate the influences of carbon coating to the morphology of WS<sup>2</sup> nanosheets, we take TEM analysis of WS2/NC. As shown in **Figure 2C**, carbon-derived (pyrrole-carbon) from polypyrrole (PPy) is uniformly coated on WS<sup>2</sup> nanosheets in nanometer size, and the boundary between WS<sup>2</sup> and pyrrole-carbon can be clearly distinguished. The carbon-coated structure can greatly improve the electronic conductivity, reduce the resistance of interface, and enhance the diffusion ability of sodium ion. Additionally, the structure can buffer the volume change during the repeated sodiation/desodiation processes and maintain the structural stability of WS<sup>2</sup> nanosheets, which is beneficial for the rate capability and cycle stability (Li et al., 2017).

In the HRTEM image (**Figure 2D**) of WS2/NC, all the independent nanoplates show clear streaks, which is ascribed to the large interlayer spacing along the c-axis of WS<sup>2</sup> nanoflakes. As shown in **Figure 2D**, the lattice fringes with interlayer spacing of 0.3089 nm could be ascribed to (004) planes of hexagonal WS2, which is well-adapted to the XRD pattern (JCPDS No. 84-1398). The large spacing between the planes of WS2/NC can not only provide a fast path in the process of Na<sup>+</sup> diffusion but also offer abundant space to store sodium ion, which permits the supply of a higher reversible capacity for the batteries. To further ensure the elements of WS2/NC, elemental mapping analysis of the asprepared materials were carried out. **Figures 2E–I** demonstrated that the W, S, C, and N elements exist in the WS2/NC. The superimposted image of elemental mapping of tungsten to sulfur (**Figures 2F,G**), and that of carbon to nitrogen (**Figures 2H,I**) are observed. The results indicate the formation of nitrogen-doped carbon supported on WS<sup>2</sup> nanosheets, and this is consistent with the HRTEM images.

Raman spectra were carried out to further verify the structure of the WS2/NC sample. As shown in **Figure 3A**, two characteristic peaks for WS2/NC are located at 350 and 413 cm−<sup>1</sup> , corresponding to the in-plane vibrational mode (E<sup>1</sup> 2g) and the out-of-plane vibrational mode (A1g) of layered WS2, respectively (Zeng et al., 2016). In addition, two peaks at 1,350 and 1,585 cm−<sup>1</sup> can also be observed, which are corresponding to the D and G bands of carbon, and owed to the disordered sp<sup>2</sup> -hybridization of the graphitic carbon structure, and the in-plane vibrational mode of the sp<sup>2</sup> -bonded carbon atoms, respectively (Ferrari et al., 2006). The results are consistent with the morphological studies, indicating that the PPy coated on WS<sup>2</sup> was partially graphitized after the carbonization process at high temperature.

To make a careful analysis of N-doped carbon coated WS<sup>2</sup> nanosheets, X-ray photoelectron spectroscopy (XPS) studies were performed. In the C 1 s spectrum of WS2/NC (**Figure 3B**), the sharp peak at about 284.6 eV is associated with sp<sup>2</sup> carbon with C=C bonds. The other two peaks at around 286 and 288.5 eV correspond to C-N bonds, demonstrating that nitrogen has been successfully incorporated into the carbon layer on WS2/NC surface, which is consistent with the Raman results. It is known that nitrogen-doping can efficiently increase the electronic conductivity of the coated carbon layer on the electrode materials, which would lessen the Ohmic polarization and is beneficial for the fast electrochemically ion-storage.

N-doping types in the carbon layer was also investigated by XPS as depicted in **Figure 3C**. The XPS spectrum of N 1 s can be deconvoluted into three main peaks at about 398.7, 401.1, and 403.4 eV, which can be ascribed to pyridinic nitrogen (N1), pyrrolic nitrogen (N2), and graphitic nitrogen (N3), respectively. The graphitic nitrogen is formed by substituting a carbon atom

with N atom that only occurred on the edges or inside of the carbon layer, which cannot damage the carbon skeletons (Wang et al., 2012). However, the pyridinic nitrogen is often formed through substituting a carbon atom by N on edges or defect sites in the plane, and the pyrrolic nitrogen species commonly expose planar edges or defect sites, meaning that pyridinic and pyrrolic nitrogen species will cause some defects and more edges in the carbon layer, which could enhance the Na<sup>+</sup> diffusion velocity, and thus improve the performance of Na<sup>+</sup> storage (Gao et al., 2015, 2017; Shen et al., 2015). By integrating the area that the fitted curve covered, the relative atomic content of pyridinic and pyrrolic nitrogen species on the surface of the carbon layer was calculated to be above 70%. Such high content of pyridinic and pyrrolic nitrogen species is expected to greatly improve the Na<sup>+</sup> storage performance of the composite WS2/NC electrode. Analysis of the S and W XPS spectra was also performed to study the composition of the WS<sup>2</sup> in the composite electrode. As shown in **Figure 3D**, two strong peaks at about 162 and 163.3 eV can be observed, which are associated with the S 2p5/<sup>2</sup> and S 2p3/<sup>2</sup> in the WS2, respectively. For the XPS spectrum of W represented in **Figure 3E**, two peaks shown at 32.3 and 34.5 eV corresponding to W 4f7/<sup>2</sup> and W 4f5/<sup>2</sup> orbitals, respectively, are in good agreement with the binding energies of W4<sup>+</sup> in WS<sup>2</sup> (Tang Y. et al., 2018). Besides, the intensity of peak at 37.8 eV could be assigned to W4<sup>+</sup> 5p3/2. The results demonstrate that no oxidation of W occurred during the preparation process because no W6<sup>+</sup> signal was found from the XPS results. From the XPS analysis, we can conclude that WS2/NC composite with high content of pyridinic and pyrrolic nitrogen species (high defects and edges) in the carbon layer was successfully obtained, which is expect to be a good electrode for Na<sup>+</sup> storage when used as the anode for Na-ion batteries.

Sodium-ion storage performance of the composite electrode was investigated in coin cells using Na metal as the counter electrode, and 1 M NaClO<sup>4</sup> in EC/PC as electrolyte. The sodiation and desodiation of WS2/NC composite anode was firstly characterized using cyclic voltammetry (CV) in the range of voltage is 0.01∼3.0 V vs. Na+/Na with a scan rate of 0.5 mV s −1 . As shown in **Figure 4A**, there are two obviously reductive peaks at 0.08 and 0.36 V, corresponding to the formation of the solid electrolyte interface (SEI) layer and the insertion of Na<sup>+</sup> in the first discharge process, during which, the Na<sup>+</sup> was firstly inserted into the crystal of WS<sup>2</sup> and formed NaxWS<sup>2</sup> without phase transformation. At the low potential of 0.08 V, metal tungsten and Na2S were formed by the conversion reaction of WS<sup>2</sup> crystal and sodium ions. During the charge process, the reversible process can occur, and the peak at 1.88 V is associated with the oxidation of NaxWS<sup>2</sup> into tungsten disulfide in the process of desodiation. In the second cycle, the peaks of 0.47 and 1.88 V in the charge process and the peaks of 0.38 and 0.74 V in the discharge process ascribe to the formation of different chemical compound, respectively. Along with the increase of cycle number, the CV curves turn gradually steady. The two curves of the fourth and the fifth cycle basically overlap, demonstrating the reversible intercalation/deintercalation of sodium ion into/from WS<sup>2</sup> is more and more stable with the charge/discharge processes. **Figure 4B** displays the chargedischarge profiles of the WS2/NC composite electrode at current density of 200 mA/g. During the first discharge process, the composite electrode delivered a relatively high specific capacity of about 600 mAh/g, which is attributed from the Na-ion insertion into the WS<sup>2</sup> and the formation of SEI film. During the following charge process, the electrode showed a reversible capacity of about 375 mAh/g, with the initial coulombic efficiency of 62.5%, meaning irreversible process occurred in the first charge/discharge cycle. It is worth noting that the composite electrode demonstrated similar charge and discharge capacities during the following the cycling, revealing the relatively high stability of the WS2/NC electrode for Na<sup>+</sup> storage.

Rate performances of the pure WS<sup>2</sup> nanosheets and WS2/NC electrodes were tested to demonstrate the enhanced Na<sup>+</sup> storage performance of the composite electrode. As shown in **Figure 4C**, the pristine WS<sup>2</sup> electrode delivered specific capacities of about 358, 320, 287.3, 241.8, 200.3, and 146.7 mAh g−<sup>1</sup> at current densities of 0.05, 0.1, 0.2, 0.5, 1.0, and 2.0 A g−<sup>1</sup> , respectively. When the current densities were set back to 1.0 and 0.5 A g−<sup>1</sup> , the pristine WS<sup>2</sup> electrode showed specific capacities of about 192.5 and 236.1 mAh g−<sup>1</sup> with relatively low reversibility. Interestingly, after N-doped carbon coating, the WS2/NC displayed enhanced specific capacities at various current densities, namely 386, 355.1, 326.4, 301.4, 274, and 238.1 mAh g−<sup>1</sup> at current densities of 0.05, 0.1, 0.2, 0.5, 1.0, and 2.0 A/g, respectively. Also, the composite electrode showed higher reversible capacities of about 271 and 305.5 mAh g−<sup>1</sup> when the current density went back to 1.0 and 0.5 A g−<sup>1</sup> , respectively. The improved rate performance of WS2/NC electrode clearly demonstrates its fast charge transfer during the electrochemical reaction, which can be ascribed to the enhanced electronic conductivity and Na<sup>+</sup> diffusion velocity resulted from the N-doped carbon coating. Long-term stability of the electrodes was further studied to illustrate the enhanced electrochemical performance of WS2/NC for Na<sup>+</sup> storage. As displayed in **Figure 4D**, the composite electrode showed an initial discharge capacity of 320.1 mAh g−<sup>1</sup> and a reversible specific capacity of 290 mAh g−<sup>1</sup> at current density of 1.0 A/g. For the pure WS<sup>2</sup> electrode, the initial capacity at 1.0 A/g is the same with that of WS2/NC electrode, but the reversible capacity is much lower (only 240.9 mAh g−<sup>1</sup> ). After 400 charge-discharge cycles, the composite electrode delivered a capacity of about 180.1 mAh g−<sup>1</sup> , corresponding a loss of 0.09% per cycle, which is much lower than that of the pure WS<sup>2</sup> with a loss of 0.14% per cycle. The long-term stability comparison further demonstrates the polished Na<sup>+</sup> storage performance after N-doped carbon coating of WS<sup>2</sup> electrode.

To uncover the reason for the improved electrochemical performance of the WS2/NC composite electrode for Na<sup>+</sup> storage, AC impedance measurements were performed at the initial state and after cycling, as shown in **Figure 5**. R<sup>e</sup> is the internal resistance of the as-assembled sodium-ion battery, R<sup>f</sup> and CPE<sup>1</sup> are associated with the resistance and constant phase element of the SEI film corresponding to the highfrequency semicircle, Rct and CPE<sup>2</sup> are corresponding to the charge-transfer resistance and constant phase element of the electrode/electrolyte interface (the semicircle in the mediumfrequency region), and Z<sup>W</sup> represents the Warburg impedance corresponding to the sodium-diffusion process (the inclined line area). The WS2/NC composite electrode, at the initial stage, showed a Rct value of about 400 , which is much lower than that of the pristine WS<sup>2</sup> electrode (about 900 ), revealing the faster charge transfer in the composite electrode. Furthermore, after charge-discharge cycling, the Rct values were reduced to 200 and 520 for WS2/NC and pure WS<sup>2</sup> electrodes, respectively, due to the gradual activation and stability of the electrode. Furthermore, the slope of the line in the low-frequency region for WS2/NC electrode is much higher than that of the pure WS2 electrode, demonstrating the faster sodium-ion diffusion in the composite electrode during the electrochemically Na+ intercalation/deintercalation process. This fact confirms that the incorporation of N-doped carbon can preserve the high conductivity of the WS2/NC composite electrode and greatly enhance rapid electron transport during the electrochemical sodium insertion/extraction reaction, resulting in significant enhancement in the electrochemical performances (Chang and Chen, 2011).

### CONCLUSIONS

In summary, WS2/NC nanosheets were synthesized and investigated as a high-performance anode for SIBs. It was found that the N-doped carbon layer in the composite electrode contains high content of pyridinic and pyrrolic nitrogen species,

FIGURE 4 | (A) CV curves of 1st to 5th cycles of the WS2/NC measured at voltage window of 0.01–3.00 V. (B) Galvanostatic charge/discharge profiles of the 1st to 5th cycles of WS2/NC measured at 200 mA g−<sup>1</sup> . (C) Galvanostatic rate capabilities of WS2/NC and pure WS2 electrodes measured at different current densities. (D) Cycling performance of WS2/NC and pure WS<sup>2</sup> electrodes measured at 1 A g−<sup>1</sup> .

which could generate more defect and expose more edge sites in the plane of carbon skeletons, resulting in faster Na<sup>+</sup> diffusion velocity. Besides, nitrogen doping can efficiently increase the electronic conductivity of the coated carbon layer on the electrode materials, which would decrease the Ohmic polarization and be beneficial for the fast electrochemically sodium ion-storage. As a result, when compared to the pristine WS<sup>2</sup> electrode, the composite WS2/NC electrode showed long-term stability with a loss of 0.09% per cycle and enhanced rate performances. This novel kind of WS2/NC composites with high reversible capacity, excellent cyclic stability, and high-rate capability would find wide applications as promising anode materials for SIBs.

### AUTHOR CONTRIBUTIONS

YL, XW, FR, BK, and CY conceived and designed the experiments; HuW, CW, FW, and HaW performed the experiments; HuW, CW, FW, and WZ analyzed the data; HuW and CW wrote the paper; YL, XW, FR and BK revised the paper.

### ACKNOWLEDGMENTS

This work was supported by the National Natural Science Foundation of China (21603157), Natural Science Foundation

### REFERENCES


of Jiangsu Province (no. BK20150311), College Natural Science Foundation of Jiangsu Province (no. 16KJB430025), and Postdoctoral Science Foundation of China (2016T90488, 2015M580459), Henan International Cooperation Project in Science and Technology (134300510051), the Plan for Scientific Innovation Talent of the Henan Province (144200510009 and 144100510015), the Program for Changjiang Scholars and Innovative Research Team in University (IRT\_16R21), the Program for Science and Technology Innovation Talents in Universities of Henan Province (17HASTIT026) and the Program for Science and Technology Innovation Team of Henan University of Science and Technology (2015XTD006), the Scientific and Technological Project of Henan Province (182102210297), Scientific Research Starting Foundation for Ph.D. of Henan University of Science and Technology (13480065), Science Foundation for Youths of Henan University of Science and Technology (2013QN006), and High-end Foreign Experts Recruitment Program (GDW2017410125).


graphene as an active and durable anode for sodium-ion batteries. Energy Environ. Sci. 10, 1757–1763. doi: 10.1039/C7EE01628J


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Liu, Wei, Wang, Wang, Wang, Zhang, Wang, Yan, Kim and Ren. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Cryptomelane-Type KMn8O<sup>16</sup> as Potential Cathode Material — for Aqueous Zinc Ion Battery

Jiajie Cui <sup>1</sup> , Xianwen Wu1,2 \*, Sinian Yang<sup>1</sup> , Chuanchang Li <sup>3</sup> , Fang Tang<sup>1</sup> , Jian Chen<sup>3</sup> , Ying Chen<sup>1</sup> , Yanhong Xiang<sup>2</sup> , Xianming Wu1,2 and Zeqiang He<sup>2</sup>

*<sup>1</sup> School of Chemistry and Chemical Engineering, Jishou University, Jishou, China, <sup>2</sup> The Collaborative Innovation Center of Manganese-Zinc-Vanadium Industrial Technology, Jishou University, Jishou, China, <sup>3</sup> School of Energy and Power Engineering, Changsha University of Science and Technology, Changsha, China*

Aqueous battery has been gained much more interest for large-scale energy storage fields due to its excellent safety, high power density and low cost. Cryptomelane-type KMn8O<sup>16</sup> confirmed by X-ray diffraction (XRD) was successfully synthesized by a modified hydrothermal method, followed by annealed at 400◦C for 3 h. The morphology and microstructure of as-prepared KMn8O<sup>16</sup> investigated by field-emission scanning electron microscopy (FE-SEM) with the energy spectrum analysis (EDS) and transmission electron microscopy (TEM) demonstrate that one-dimensional nano rods with the length of about 500 nm constitute the microspheres with the diameter about 0.5∼2µm. The cyclic voltammetry measurement displays that the abundant intercalation of zinc ions on the cathode takes place during the initial discharge process, indicating that cryptomelane-type KMn8O<sup>16</sup> can be used as the potential cathode material for aqueous zinc ion batteries. The electrode shows a good cycling performance with a reversible capacity of up to 77.0 mAh/g even after 100 cycles and a small self-discharge phenomenon.

Keywords: intercalated potassium compound, aqueous rechargeable battery, cathode material, energy storage and conversion, self-discharge

## INTRODUCTION

Although the lithium-ion batteries (LIBs) as one of the most promising energy storage devices have gained a great improvement in energy density and life span, and correspondingly dominated in the fields of portable mobile devices, electric vehicles (EVs) and hybrid electric vehicles (HEVs), the nonnegligible safety issues resulted from the flammable organic electrolytes seem to restrict their large-scale applications (Liu et al., 2014; Wang et al., 2015b; Su et al., 2018; Zhang et al., 2018). On the contrary, the aqueous rechargeable batteries can overcome these disadvantages mentioned above, which have attracted extensive attentions since the aqueous batteries of VO2/LiMn2O<sup>4</sup> was firstly proposed by Dahn's group in 1994 (Li et al., 1994), especially the aqueous rechargeable batteries based on zinc anode, considering its multivalent characteristic, low cost, abundance and environmental benignity of zinc. However, up to now, only a few cathode materials have been developed as the intercalation hosts for metal ions. Rechargeable hybrid aqueous battery (ReHAB) system based on LiMn2O<sup>4</sup> as the cathode and zinc as the anode was reported firstly by Chen's

#### Edited by:

*Qiaobao Zhang, Xiamen University, China*

#### Reviewed by:

*Qiulong Wei, University of California, Los Angeles, United States Yunjian Liu, Jiangsu University, China*

#### \*Correspondence:

*Xianwen Wu wxwcsu2011@163.com*

#### Specialty section:

*This article was submitted to Physical Chemistry and Chemical Physics, a section of the journal Frontiers in Chemistry*

> Received: *16 June 2018* Accepted: *25 July 2018* Published: *17 August 2018*

#### Citation:

*Cui J, Wu X, Yang S, Li C, Tang F, Chen J, Chen Y, Xiang Y, Wu X and He Z (2018) Cryptomelane-Type KMn8O16 as Potential Cathode Material — for Aqueous Zinc Ion Battery. Front. Chem. 6:352. doi: 10.3389/fchem.2018.00352* group, of which the capacity retention is up to 90.0% even after 1,000 charge/discharge cycles (Yan et al., 2012). After that, the similar systems such as Zn/LiMn2O<sup>4</sup> (Lu et al., 2016; Zhu et al., 2016; Sun et al., 2017), Zn/LiMnPO<sup>4</sup> (Minakshi et al., 2006), Zn/LiMn0.8Fe0.2PO<sup>4</sup> (Zhao et al., 2016), Zn/LiCo1/3Mn1/3Ni1/3PO<sup>4</sup> (Kandhasamy et al., 2012), Zn/LiFePO<sup>4</sup> (Zhang et al., 2013), Zn/LiCo1/3Mn1/3Ni1/3O<sup>2</sup> (Wang et al., 2015a) were reported. Nevertheless, the processing cost and the limited lithium resources result in a tremendous challenge for application. Thus, it is urgent for us to explore new non-lithium intercalation compounds as the cathode so as to match with zinc anode.

Among them, tunnel-type manganese oxides have been mostly investigated, including α-, β-, γ-, and δ-types MnO<sup>2</sup> (Xu et al., 2012, 2014; Alfaruqi et al., 2015a,b,c; Pan et al., 2016; Han et al., 2017; Zhang et al., 2017). Although considerable initial discharge capacity up to 200 mAh/g at low C-rate can be delivered, they suffer from the poor rate performance and a rapid capacity fading owing to the repeated phase transitions and the dissolution of Mn2<sup>+</sup> owing to Mn3<sup>+</sup> disproportionation upon cycling. Moreover, the reaction mechanism of MnO<sup>2</sup> remains controversial (Lee et al., 2015). A family of prussian blue analogs (abbreviated as PBAs) such as zinc hexacyanoferrate (ZnHCF) are also the attractive cathode materials based on zinc anode, which allow the rapid metal ion diffusion due to their cubic open-framework structures (Zhang et al., 2015a,b; Liu et al., 2016). However, these cathodes delivered the limited capacities (about 50 mAh/g) and suffered oxygen evolution under the high voltage.

One dimensional (1D) tunnel structured cryptomelane type manganese dioxides, Mn8O<sup>16</sup> (α-MnO2) as the cathode materials have been received extensive concerns, as they can reversibly host various cations including Li<sup>+</sup> and K<sup>+</sup> and so on, of which KxMn8O<sup>16</sup> was previously reported in the fields of catalysis and lithium ion battery (Poyraz et al., 2017). Herein, this study aims to establish the aqueous hybrid battery based on cheap intercalated potassium compound KMn8O<sup>16</sup> as the cathode material and zinc as the anode. We try to characterize the structure and its morphology and demonstrate its charge/discharge mechanism and excellent electrochemical performances.

### EXPERIMENTAL

Cryptomelane-type KMn8O<sup>16</sup> was synthesized by a modified hydrothermal method (Poyraz et al., 2017). The typical preparation procedure is as follows, 6 mmol of MnSO4·H2O and 12 mmol of (NH4)2SO<sup>4</sup> and 12 mmol K2SO<sup>4</sup> were dissolved in 50 mL of 6 mmol (NH4)2S2O<sup>8</sup> solution. After being magnetic stirred for 30 min, the solution was then transferred to a Teflonlined stainless steel autoclave and was maintained at 130◦C for 24 h. After it was naturally cooled to room temperature, the resulting product was washed with distilled water and anhydrous alcohol for several times, collected by centrifuge and followed by dried at 60◦C overnight. Finally, it was annealed at 400◦C for 3 h to obtain KMn8O16.

The phase and structure of as-prepared samples were identified by powder X-ray diffraction (XRD, D8 Discover, Bruker) employing Cu Kα (λ = 0.15406 nm) radiation from 10◦ to 65◦ . Morphology observation for KMn8O<sup>16</sup> was conducted on a field-emission scanning microscopy (FE-SEM, Leo-1530, Zeiss) with the energy spectrum analysis (EDS) (accelerating voltage of 20 kV). The morphology and microstructure of the material was measured by transmission electron microscopy (TEM, Tecnai G12, 200 kV). The surface element analysis of the as-synthesized product was used by X-ray photoelectron spectroscopy (XPS, K-Alpha 1063), and then the spectra obtained was fitted with XPS peak software (version 4.1).

The working electrode was prepared by casting the slurries of 80 wt% KMn8O16, 10 wt% polyvinylidene fluoride (PVDF) and 10 wt % acetylene black on graphite foil. After being blended in N-methyl pyrrolidinone, the mixed slurry was spread uniformly on graphite foil and dried in a vacuum for 4 h at 60◦C. Disks of 14 mm diameter were cut (the typical active material loading was about 2∼3 mg/cm<sup>2</sup> ) and soaked in hybrid electrolyte solution under vacuum for 15 min. AGM (Absorbed Glass Mat, NSG Corporation) was used as the separator. Then the electrochemical performances were evaluated using CR2032 coin-type battery based on zinc foil as the anode. The galvanostatic charge-discharge was performed by way of a battery tester (LAND in China) in the potential range of 0.8∼1.9 V at room temperature, and the cyclic voltammetry curve was tested on an electrochemical workstation (CHI 660E).

### RESULTS AND DISCUSSION

The crystalline structure of as-obtained products after heat treated at 400◦C in air for 3 h was analyzed by XRD. It can be clearly observed from **Figure 1** that all the diffraction peaks of the sample at 2θ =12.745◦ , 18.059◦ , 25.651◦ , 28.737◦ , 37.62◦ , 42.029◦ , 49.896◦ , 56.184◦ , and 60.24◦Can be readily indexed to (110), (200), (220), (310), (211), (301), (411), (600), and (521), which are in well agreement with the pure tetragonal

cryptomelane structures of KMn8O<sup>16</sup> (JCPDS card No. 29-1020; space group:I4/m(87), a = b = 9.815 Å, c = 2.847 Å, and α = β = γ = 90◦ ), and the average crystal sizes of KMn8O<sup>16</sup> determined by using the Scherrer formula (L = 0.89λ/βcosθ) at 2θ = 18.1◦ is 17 nm. Meanwhile, EDS elemental mapping in **Figure 2** reflects the uniform distribution of K, Mn and O in the material, and there are no other impurity elements, indicating that K<sup>+</sup> ions can be embedded into the interlayer space of

MnO<sup>2</sup> structure. These results demonstrate that KMn8O<sup>16</sup> can be successfully prepared.

Furthermore, FE-SEM and TEM are used to demonstrate the morphology and microstructure of the as-prepared products in **Figure 3**. The results in **Figure 3A** show that the secondary particles of KMn8O<sup>16</sup> exhibit the microspheres with the diameter about 0.5∼2µm. Meanwhile, it is carefully noted that many one-dimensional nano rods with the length of about 500 nm constitute the microspheres in **Figure 3B**. Further the details of the structural characteristic are indicated by TEM. The primary particle is made up of nano rods interconnected each other in **Figure 3C**, and the lattice fringes with the interplanar acing of 0.2384, 0.3105, 0.4879, and 0.6675 nm from the HRTEM images in **Figure 3D** are assigned to the (211), (310), (200), and (110) planes of KMn8O16, respectively. No other diffraction peaks of impurities have been detected. Meanwhile, the selected area electron diffraction (SAED) pattern in **Figure 3E** corresponds to the characteristic diffraction rings of (110), (200), (310), (400), and (301) planes of tetragonal KMn8O16, all of which agree well with the XRD results.

The oxidation states and the composition of the samples were examined by XPS, and the spectra obtained were fitted with the XPS peak software. As shown in **Figure 4A**, the full scanning spectrum of tetragonal KMn8O<sup>16</sup> shows that the presence of K, Mn, and O elements, which is in accordance with the EDS results. The K/Mn atomic ratios on the surface of

FIGURE 4 | XPS spectra of the KMn8O16 sample: (A) survey spectrum; (B) K 2p spectrum; (C) Mn 2p spectrum; (D) Mn 3s; (E) O 1s spectrum.

sample is 0.135, which is very close to the bulk K/Mn atomic ratios measured by ICP, demonstrating that potassium is well dispersed in the sample. Thus, the chemical formula of the product is KMn8O16. In **Figure 4B** it can be noted that there are two separate peaks with the binding energies of 292.03 and 294.83 eV respectively, attributed to K2p1/<sup>2</sup> and K2p3/2, and the binding energy difference between these two peaks of 2.8 eV, which is consistent with the previous reports. In addition, the high resolution spectra of Mn 2p in the as-prepared sample in **Figure 4C** show the binding energies (BEs) of the Mn 2p1/<sup>2</sup> and Mn 2p3/<sup>2</sup> peaks located at 654.18 and 642.48 eV, and the Mn2p3/<sup>2</sup> binding energy of KMn8O<sup>16</sup> are between that of the Mn2O<sup>3</sup> and MnO<sup>2</sup> powder standard, indicating that the existence of multiple Mn valences ions, which can be decomposed into two peaks of Mn3<sup>+</sup> at 643.92 eV and Mn4<sup>+</sup> at 642.33 eV. It is wellknown that the separation of peak energies (1E) of Mn 3s can estimate the average oxidation state (AOS) of Mn, which were calculated from the 1E of Mn 3s peaks {AOS = 8.956–1.126 × 1E(3s)}. Thus, the calculated AOS of Mn is about 3.6 based on the Mn 3s splitting energy (4.73 eV) in **Figure 4D**. In addition, the high resolution O 1s peak is deconvoluted to three sub-peaks in **Figure 4E**, demonstrating three kinds of oxygen atoms in the sample, of which the fitting peak at 529.9 eV represents the typical Mn-O-Mn lattice oxygen, the fitting peak at 532.0 eV is assigned to Mn-OH surface hydroxyls or defect-oxide, while the peak at 533.4 eV is associated with the oxygen in the OH group adsorbed water.

In order to evaluate the charge/discharge behavior in 1 mol/L ZnSO<sup>4</sup> and 0.3 mol/L K2SO<sup>4</sup> without or with 0.05 mol/L MnSO4, the cyclic voltammetry (CV) curves in the range of 0.8∼1.9 V were conducted at a sweep rate of 1 mV/s. As is shown in **Figure 5A**, there is a small oxidation peak at 1.54 V and two reduction peaks, a very smaller reduction peak situated at 1.42 V and a larger reduction peak located at 1.23 V, respectively, indicating the strange initial charge/discharge process in 1 mol/L ZnSO<sup>4</sup> and 0.3 mol/L K2SO4. That's to say, the initial coulombic

efficiency exceeds 100%. However, apparently different from that shown in **Figure 5A**, after adding 0.05 mol/L MnSO<sup>4</sup> into electrolyte in **Figure 5B**, the initial oxidation peak area is increased, the reduction peak area at 1.26 V tends to decrease, and the other of reduction peak at about 1.0 V disappear during the cathodic sweeping process, indicating the much better cyclic reversibility than that in **Figure 5A**. Notably, the position and CV profile in the subsequent sweeping curves are close to that in the first one. One cathodic peak at about 1.26 V nearly overlap, and the peak area of another cathodic peak at about 1.38 V seems to increase, demonstrating that the discharge capacity increases gradually with the cycle number increasing.

In fact, the schematic illustration of charge/discharge mechanism for Zn/KMn8O<sup>16</sup> hybrid aqueous battery in **Figure 6** is a little different from those in previous reports by our groups (Wu et al., 2015, 2017). K+, Zn2<sup>+</sup> and Mn2<sup>+</sup> co-exist in the electrolyte. Based on the investigation on Na3V2(PO4)<sup>3</sup> for aqueous zinc ion battery in Huang's group and the similarity of

FIGURE 8 | (A) The float charge procedure, (B) the float charge current density and (C) the charge/discharge curve of Zn/Zn/KMn8O16 battery before and after float charge.

K <sup>+</sup> and Na<sup>+</sup> (Li et al., 2016), it can be inferred that potassium ions are de-intercalated from KMn8O<sup>16</sup> and dissolved into the electrolyte quickly during the initial charge process. In the anode side, zinc ions in the electrolyte accept two electrons and deposit on the current collector. However, the reversible process is different from the previous process during the discharge process. It means that the intercalation of little potassium ions on the cathode will happen during the initial discharge process, accompanied mainly by the abundant intercalation of zinc ions. The main reason is that the radius of a zinc ion 0.6 Å is much smaller than that of potassium ions 1.33 Å, and the zinc ion can be intercalated easily into the MnO2-host structure. Thus, their electrode and total reactions on the positive side can be simply demonstrated as follows.

During the initial charge process, the equation of K<sup>+</sup> deintercalation can be stated as follows.

$$\rm{KMn\_8O\_{16}} \Leftrightarrow \rm{xK^+} + K\_{1-x}Mn\_8O\_{16} + \rm{xe^-} \tag{1}$$

While during the reversible intercalation of metal ion, the equation on the positive side can be described as follows.

$$0.5\varkappa Zn^{2+} + K\_{1-x}Mn\_8O\_{16} + \varkappa e^- \Leftrightarrow Zn\_{0.5x}K\_{1-x}Mn\_8O\_{16} \quad \text{(2)}$$

To better clarify the excellent cycling and rate stability of KMn8O<sup>16</sup> material, the hybrid aqueous battery system was established based on KMn8O<sup>16</sup> as the cathode and zinc as the anode, and the galvanostatic charge/discharge curves of the electrode were measured in **Figure 7**. Interestingly, all of the phenomenon can be confirmed from the cycling and charge/discharge curves at 1C-rate. The initial coulombic efficiency is as high as 290.5% whether adding 0.05 mol/L MnSO<sup>4</sup> into 1 mol/L ZnSO<sup>4</sup> and 0.3 mol/L or not in **Figures 7B,C**, which is consistent with the CV results. However, the cycling performance of battery with 0.05 mol/L MnSO<sup>4</sup> is much better than that without MnSO<sup>4</sup> in **Figure 7A**, and the discharge capacity of the former is still up to 77.0 mAh/g even after 100 cycles, which is much higher than that 41.7 mAh/g of the latter. The main reason is that the addition of MnSO<sup>4</sup> has inhibited Mn dissolution, decreased the polarization overpotential and facilitated zinc dissolution (Wu et al., 2017), which has improved the cycleability of the electrode.

A series of side reactions about electrodes usually happen (Wu et al., 2016). Here the float charge and self-discharge were evaluated. The battery was cycled for three times at 50 mA/g firstly, and then charged to 1.9 V, standing for 24 h or going on

### REFERENCES


charging at constant voltage at room temperature in **Figures 8A**, **9A**. Lastly it was cycled for three times. As can be seen from **Figure 8B**, the float charge current density decreased gradually with the time increasing, and quickly approaches to 0 mA/g. Even after float charge in **Figure 8C**, there is not capacity fading upon discharge process. On the contrary, the charge capacity increases to 105.2 mAh/g after float charge. During the selfdischarge curve in **Figure 9B**, the voltage decreased to 1.5299 V. However, there is no much capacity fading in **Figure 9C** upon discharge process. That's to say, the discharge capacity can nearly return to the original level, indicating the smaller self-discharge phenomenon.

### CONCLUSIONS

In conclusion, KMn8O<sup>16</sup> microspheres were successfully synthesized by a modified hydrothermal method, which were characterized by XRD, FE-TEM, EDS, TEM, and XPS. The material of cryptomelane-type structure was constituted by one-dimensional nano rods. When used as cathode material for aqueous battery, it shows an excellent cycling performance with a reversible capacity of up to 77.0 mAh/g even after 100 cycles and the small self-discharge phenomenon. CV test indicates that it can be used as the cathode material for aqueous zinc ion batteries in potential large scale energy storage field. Therefore, the results presented here will provide an alternative for aqueous batteries with high safety, low cost and high power density.

### AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

### ACKNOWLEDGMENTS

This research was financially supported by the Key Laboratory of Efficient & Clean Energy Utilization of the Education Department in Hunan Province (2017NGQ003), the Key Planned Science and Technology Project of Xiangxi Tujia & Miao Autonomous Prefecture (No. 2018GX2001), the Natural Science Foundation of Hunan Province (No. 2018JJ3415) and the National Natural Science Foundation of China (No.51704124, No.51762017, No.51662010, No.51364009, No. 51262008, No.51472107 and No.51672104), which were greatly appreciated.


for aqueous rechargeable battery. Electrochim. Acta 60, 170–176. doi: 10.1016/j.electacta.2011.11.028


Zn/Na0.44MnO<sup>2</sup> based on hybrid electrolyte. J. Power Sources 336, 35–39. doi: 10.1016/j.jpowsour.2016.10.053


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Cui, Wu, Yang, Li, Tang, Chen, Chen, Xiang, Wu and He. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Fabrication of WO3·2H2O/BC Hybrids by the Radiation Method for Enhanced Performance Supercapacitors

Fan Yang1†, Jinzhi Jia1,2†, Rui Mi <sup>1</sup> , Xichuan Liu<sup>1</sup> , Zhibing Fu<sup>1</sup> , Chaoyang Wang<sup>1</sup> , Xudong Liu1,3 \* and Yongjian Tang<sup>1</sup> \*

*<sup>1</sup> Science and Technology on Plasma Physics Laboratory, Research Centre of Laser Fusion, China Academy of Engineering Physics, Mianyang, China, <sup>2</sup> School of Materials Science and Engineering, Southwest University of Science and Technology, Mianyang, China, <sup>3</sup> College of Materials Science and Engineering, Chongqing University, Chongqing, China*

Edited by:

*Qiaobao Zhang, Xiamen University, China*

#### Reviewed by:

*Shuge Dai, Zhengzhou University, China Zhao Yan, Jiangsu University, China*

#### \*Correspondence:

*Xudong Liu 8sliuxudong@caep.cn Yongjian Tang tangyongjian2000@sina.com*

*†These authors have contributed equally to this work*

#### Specialty section:

*This article was submitted to Nanoscience, a section of the journal Frontiers in Chemistry*

Received: *06 May 2018* Accepted: *25 June 2018* Published: *13 August 2018*

#### Citation:

*Yang F, Jia J, Mi R, Liu X, Fu Z, Wang C, Liu X and Tang Y (2018) Fabrication of WO3*·*2H2O/BC Hybrids by the Radiation Method for Enhanced Performance Supercapacitors. Front. Chem. 6:290. doi: 10.3389/fchem.2018.00290* In this study, we described a facile process for the fabrication of tungsten oxide dihydrate/bamboo charcoal hybrids (WO3·2H2O/BC) by the γ-irradiation method. The structural, morphological, and electrochemical properties of WO3·2H2O/BC hybrids were investigated using X-ray diffraction (XRD), field emission scanning electron microscopy (FESEM), transmission electron microscopy (TEM), cyclic voltammetry (CV), galvanostatic charge/discharge (GCD), and electrochemical impedance spectroscopy (EIS) techniques. The combination of BC (electrical double layer charge) and WO3·2H2O (pseudocapacitance) created a combined effect, which enhanced the specific capacitance and superior cyclic stability of the WO3·2H2O/BC hybrid electrode. The WO3·2H2O/BC hybrids showed the higher specific capacitance (391 F g−<sup>1</sup> at 0.5 A g <sup>−</sup><sup>1</sup> over the voltage range from −1 to 0 V), compared with BC (108 F g−<sup>1</sup> ) in 6 M KOH solution. Furthermore, the hybrid electrode showed superior long-term performance with 82% capacitance retention even after 10,000 cycles. The experimental results demonstrated that the high performance of WO3·2H2O/BC hybrids could be a potential electrode material for supercapacitors.

Keywords: γ-irradiation method, WO3 ·2H2O/BC hybrids, higher specific capacitance, cyclic stability, supercapacitors

### INTRODUCTION

For the rapid increase of global energy demand and the depletion risk of fossil fuels, developing alternative sustainable, affordable, efficient, and clean energy has become very important (Zhao Y. et al., 2016, 2017; Xu et al., 2017; Yi et al., 2017; Zhao B. et al., 2017). Among energy storage devices, supercapacitors (or ultracapacitors) are promising, owing to their safe operation, super-high service life, and great power density (Wang K. et al., 2014; Dai et al., 2017; Zhao B. et al., 2017; Zhao et al., 2018). They have wide application areas such as electric vehicles, pulse power systems and portable devices (Wang et al., 2018). Depending on the charge storage mechanism, supercapacitors are generally classified into electrical double layer charge (EDLC) and pesudoprocess charge storage. The former stores charges electrostatically in double layers, whereas the latter stores charges on the surface of the electrode active materials as faradaic redox reactions (Zhang and Park, 2017). In general, carbon-based materials are EDLC type (Pang et al., 2016), whereas transition metal oxides and conducting polymers are pseudocapacitor-type materials (Zhang et al., 2012; Yao C. et al., 2017; Yao S. et al., 2017).

Yang et al. WO3.2H2O/BC Hybrids for Supercapacitors

The charge storage of pesodocapacitors is much higher compared to those of EDLCs. More recently, transition metal oxides (i.e., RuO2, V2O5, NiO, MnO2, SnO2, and WO3) have been widely investigated in the applications for supercapacitors, and their charge storage originates from fast superficial redox reactions (Zhu and He, 2012; Zhang et al., 2015; Qiu et al., 2016; Zeng et al., 2017; Zhang and Park, 2017; Liu et al., 2018a). Among many of the reported transition metal oxides, RuO<sup>2</sup> has been considered as the proper material with excellent capacitive performance. However, RuO<sup>2</sup> is expensive and rare, constraining the wide practical applications in electrode materials (Cai et al., 2014). WO<sup>3</sup> is a promising electrode material owing to its various morphologies, high theoretical specific capacitance, environmental friendliness, and low cost (Qiu et al., 2016). Nevertheless, its low electrical conductivity (10−5–10−<sup>6</sup> S cm−<sup>1</sup> ) has limited the wide applications. If we can improve the conductivity of WO3, higher specific capacitances could be achieved as expected. Therefore, many researchers have focused on incorporating WO<sup>3</sup> with highly conductive carbon materials to establish a hybrid-type material that combines advantages of each component (Lu et al., 2012; Xiao et al., 2012; Reddy et al., 2015; Sun et al., 2015; Wang et al., 2015; Yuksel et al., 2016). Wang et al. fabricated the WO3/carbon aerogel hybrids with outstanding long-term stability, and the specific capacitance was ∼700 F g−<sup>1</sup> (at a scan rate of 25 mV s−<sup>1</sup> in 0.5 M H2SO<sup>4</sup> over a voltage window of −0.3 to 0.5 V; Wang Y. H. et al., 2014). Huang et al. also designed graphene-WO<sup>3</sup> hybrids with enhanced supercapacitor capacitance (Xing et al., 2016). Wang et al. fabricated the graphene nanosheetstungsten oxides with high supercapacitor performance (Wang et al., 2015). Ma et al. fabricated a hybrid based on graphene and WO<sup>3</sup> via the hydrothermal method, which possessed high specific capacitance and superior rate capability (Ma et al., 2015).

Among the conductive materials, activated carbon-based materials are the most promising candidates for supercapacitor applications, due to their unique characteristics of large surface areas, and high electrochemical stability and conductivity (Yang et al., 2014; Li and Wu, 2015; Wang et al., 2016; Boyjoo et al., 2017; Dai et al., 2018). In carbon materials, bamboo charcoal draws research attention for its extraordinarily porous microstructure, cost efficiency, and high absorptive capacity (Wang et al., 2012, 2013, 2016; Zhang et al., 2013; Yang et al., 2014; Li and Wu, 2015; Yu et al., 2015). Li et al. studied the water bamboo-derived porous carbon with a maximum specific capacitance of 268 F g−<sup>1</sup> at a current density of 1 A g −1 in 6 M KOH electrolyte and good capacity retention of 97.28% even over 5,000 cycles at a current density of 10 A g −1 (Li and Wu, 2015). Yang et al. also synthesized BC by KOH activation, and the specific capacity retention was more than 91% after 3,000 cycles (Yang et al., 2014). Impressively, BC has long-lasting life, whereas WO<sup>3</sup> has high theoretical specific capacity. For this respect, combining each advantage of BC and WO3·2H2O to improve the performance might bring novel and excellent properties. However, so far, to the best of our knowledge, nearly no works have been done in this aspect.

For decades, researchers have reported many ways to fabricate WO3/carbon materials, such as hydrothermal method, impregnation methods, and sol-gel method. Compared with these methods, as we previously reported, the radiation method can improve the contact between the doped materials and pristine carbon, and this method has been successfully applied in H<sup>2</sup> storage (Zhong et al., 2015, 2016). Obviously, good adhesion may improve the stability of hybrids and optimize the conduction of electrons, which will enhance the capacitive properties of the hybrids. But so far, no studies have been conducted on the preparation of metal oxides and carbon hybrid materials for supercapacitor applications by the irradiation method. To widen application of this method and further improve the capacitive performance, it is important to develop this method.

In this work, novel WO3·2H2O/BC hybrids were designed and fabricated by a facile γ-irradiation strategy. Morphologies and microstructures of the samples were investigated by XRD, SEM, TEM, and XPS, whereas CV, GCD, and EIS were carried out to study capacitive properties. The electrochemical results demonstrated that WO3·2H2O/BC hybrids delivered a high specific capacity (391 F g−<sup>1</sup> at 0.5 A g−<sup>1</sup> ) and superior long-term stability (82% retention even after 10,000 cycles). The combination of bamboo (EDLC) and WO3·2H2O (pseudocapacitance) provided short ion diffusion path and fast electron transport, leading to a great supercapacitor.

### EXPERIMENTAL DETAILS

## Synthesis of WO3·2H2O/BC Hybrids

All the reagents and solvents were analytical grade and used without further purification. In a typical process, WO3·2H2O/BC hybrids were prepared as follows. WCl<sup>6</sup> (3 mg) was added into isopropyl alcohol (20 mL) under stirring for 20 min in a glass vial at room temperature. Then, the BC monoliths (0.2 g) were slowly impregnated with 10 ml of WCl<sup>6</sup> solution. After 30 min of continuous stirring, 2-propanol was added with proper amount to scavenge H<sup>∗</sup> and OH<sup>∗</sup> radicals, which were generated during irradiation. The mixture was irradiated at room temperature with a <sup>60</sup>Co γ-ray source at a dose rate of 200 Gy min−<sup>1</sup> , and the total dose was 500 kGy. The product was collected by centrifugation and rinsed several times with DI water and ethanol, and then dried at 60◦C for 12 h. **Figure 1** shows the schematic diagram for the synthesis of WO3·2H2O/BC hybrids and their supercapacitor performance. For comparison, bamboo charcoal was prepared by carbonization of natural bamboo with the KOH-modified method, as described elsewhere (Yang et al., 2014; Li and Wu, 2015).

### Material Characterization

The crystalline phase of WO3·2H2O/BC hybrids was examined by X-ray powder diffraction (XRD) employing monochromatized CuKα incident radiation. The morphology and microstructure were analyzed by using a field emission scanning electron microscope (FESEM, Nova 600i) with an attached energy dispersive X-ray spectroscopy (EDS)

analysis and transmission electron microscopy (TEM). X-ray photoelectron spectroscopy (XPS) was studied for detecting chemical composition and oxidation states of WO3·2H2O/BC hybrids.

### Electrochemical Measurements

Electrochemical tests were carried out at room temperature in a conventional three-electrode configuration with 6 M KOH as an electrolyte using an electrochemical workstation (CHI 660E), WO3·2H2O/BC as a working electrode, platinum foil as a counter electrode, and Hg/HgO as a reference electrode. In electrochemical tests, WO3·2H2O/BC (80%) was mixed with acetylene black (10%) and polyvinylidene fluoride (PVDF, 10%) in N-methyl-2-pyrrolidone (NMP) to form slurry. Then the slurry was coated on glassy carbon to fabricate the working electrode. The potential range for CV tests is from −1 to 0 V, and the measurement range for EIS tests is between 0.1 Hz and 100 kHz with an AC amplitude of 5 mV.

### RESULTS AND DISCUSSION

## Characterization of WO3·2H2O/BC Hybrids

Crystal structures of the as-prepared BC and WO3·2H2O/BC hybrids were first studied by XRD. As shown in **Figure 2**, two broad peaks near 23 and 43◦ correspond to (002) and (100), respectively, which can be identified for the amorphous

forms of BC. For WO3·2H2O/BC hybrids, two broad peaks of BC also appeared, and all the peaks exclusively assigned to the characteristic structure of WO3·2H2O (JCPDS-00-18-1420). Furthermore, no other impurity phase peak can be detected, and the existence of strong and sharp peaks also indicates that WO3·2H2O/BC hybrids have high crystallinity. The structure of WO3·2H2O has been reported to be attractive as electrode materials (Ma et al., 2015; Li et al., 2016; Mitchell et al., 2017). Crystalline WO<sup>3</sup> is much more stable than amorphous WO<sup>3</sup> due to the denser structure and lower dissolution rate in electrolytes, which is a very important point in terms of practical applications (Liu et al., 2017). The diffraction pattern of WO3·2H2O/BC hybrids is the combination of the peaks from

FIGURE 3 | Typical SEM images of (a) BC and (b) WO3·2H2O/BC hybrids. (c) The EDS mapping results and (d) the corresponding spectrum for the WO3·2H2O/BC hybrids. (e) TEM and (f) HRTEM images of the WO3·2H2O/BC hybrids.

BC and WO3·2H2O, demonstrating the successful composite. However, for the amorphous forms of BC, no distinct peaks of BC can be observed in WO3·2H2O/BC hybrids. Nevertheless, the presence of BC in the WO3·2H2O/BC hybrids can be confirmed by the results of SEM, TEM, and XPS.

The surface morphologies of BC and WO3·2H2O/BC hybrids were characterized by SEM and TEM. As shown in **Figure 3a**, we can clearly see that the BC has smooth surface and irregular forms. **Figure 3b** shows the morphology of WO3·2H2O/BC hybrids, and the skeleton of BC can be clearly seen with a random distribution of WO3·2H2O. EDS mapping was further used to demonstrate the formation of WO3·2H2O/BC hybrids, and is shown in **Figure 3c**. Obviously, the C, W, and O elements exist in WO3·2H2O/BC hybrids, and this result is in accord with the EDS spectrum shown in **Figure 3d**. As shown in **Figures 3e,f**, TEM and high-resolution TEM (HRTEM) micrographs further indicate microscopic structures of WO3·2H2O/BC hybrids. The TEM image of WO3·2H2O/BC hybrids clearly reveals that WO3·2H2O is successfully connected to BC. The HRTEM image of WO3·2H2O/BC (**Figure 3f**) shows that the spacing between adjacent lattice planes is 0.37 nm, corresponding to the (001) plane of WO3·2H2O, indicating that WO3·2H2O grows preferentially along (001) (Mitchell et al., 2017). Therefore, WO3·2H2O/BC hybrids can be successfully synthesized via a simple γ-irradiation method, in which WO3·2H2O is sufficiently connected to BC. And the robust contact between WO<sup>3</sup> and BC can be maintained by ultrasound for 30 min without WO<sup>3</sup>

shed (see **Figure 3e**). This great interfacial contact between WO3·2H2O and BC may be possibly favorable for the electronic transport process, thus resulting in the enhanced supercapacitor performance (Cai et al., 2014; Chu et al., 2017; Liu et al., 2018a,b). To further check the surface chemical composition of WO3·2H2O/BC hybrids, XPS measurements were carried out.

The detailed composition and surface valence state of WO3·2H2O/BC hybrids was probed by XPS measurements. **Figure 4A** shows the survey scan spectrum of the WO3·2H2O/BC hybrids, and only W, O, and C elements exist in the hybrids without evidence of any other impurity atoms. **Figure 4B** shows the XPS spectrum of the W 4f doublet peak in the high-resolution scan. The peaks located at 37.87 and 35.73 eV are attributable to the W 4f5/<sup>2</sup> and W 4f7/2, respectively. The observed energy position of the doublet is in accord with the previous report for the W6<sup>+</sup> oxidation state (Shinde et al., 2016; Xu et al., 2016; Liu et al., 2017; Wu and Yao, 2017). The splitting between W 4f7/<sup>2</sup> and W 4f5/<sup>2</sup> is 2.14 eV, demonstrating a typical state of W6<sup>+</sup> in WO3·2H2O/BC hybrids, which is well analogous to the XRD study. The XPS spectrum of C 1s from the WO3·2H2O/BC hybrids (see **Figure 4C**) is also decomposed into three peaks at 284.57 eV (C–C), 285.68 eV (C–O), and 287.81 eV (C = O), suggesting that the bonding between carbon atoms of BC and oxygen atoms of WO3·2H2O improves the conductivity and accelerates charge transport for the hybrids (Cai et al., 2014;

Nayak et al., 2017). And the deconvolution peak of the O 1s spectrum can be resolved into two components of 530.98 and 533.02 eV (see **Figure 4D**). The low binding energy component at 530.98 eV is attributed to the O2<sup>−</sup> bond with wolfram, and the latter peak is assigned to OH<sup>−</sup> (Xing et al., 2016; Nayak et al., 2017).

### Electrochemical Performance of WO3·2H2O/BC Hybrids

The electrochemical performance of WO3·2H2O/BC hybrids was first investigated by the CV test. The CV curves of the BC and WO3·2H2O/BC electrodes are shown in **Figure 5A**. Compared with those of BC, the CV curves of WO3·2H2O/BC changed from the rectangular shape to the "dolphin-like" shape, and nearly no obvious redox peaks detected are characteristic among various WO<sup>3</sup> (Reddy et al., 2015; Thind et al., 2016; He et al., 2017; Nayak et al., 2017). Moreover, the stored charge of the hybrids can be calculated by the enclosed area of the CV curve. The observed integrated area of the WO3·2H2O/BC electrode at the same current density is much larger than that of the BC electrode, suggesting the contribution of WO3·2H2O incorporation to the enhanced specific capacitance of WO3·2H2O/BC, and the combined effect of WO3·2H2O and BC is significant (Cai et al., 2014; Liu et al., 2018a). The results obtained here are also consistent with the SEM and TEM morphologies, suggesting that the WO3·2H2O/BC hybrid morphology provides good contact fascinating the fast charge intercalation/deintercalation process, and the enhanced electronic conductivity after carbon incorporation contributes to the superior electrochemical performance. **Figure 5B** shows the CV curves of WO3·2H2O/BC in 6 M KOH at different scan rates (2, 5, 10, 20, and 50 mV s−<sup>1</sup> ) over a potential window of −1 to 0 V. The corresponding CV curves of BC are provided in **Figure S1a**. The increased area under the curve with scan rate is clearly observed, indicating an

excellent capacitance behavior and high-rate capability of the electrode.

Charge-discharge measurements were conducted under galvanostatic conditions at different applied current densities. The GCD plots of the BC and WO3·2H2O/BC electrodes at a current density of 1 A g−<sup>1</sup> are presented in **Figure 5C**. According to the galvanostatic discharge curves, the specific capacitance (C<sup>s</sup> , F g−<sup>1</sup> ) of the electrode is calculated according to the following equation:

$$C\_s(F\,g^{-1}) = \frac{I\Delta t}{m\Delta V}$$

where I (mA) represents the applied current, 1t (s) the discharge time, 1V (V) the potential window, and m (mg) the weight of the active material. In addition, the discharge time of WO3·2H2O/BC is much larger than that of BC, showing higher capacitance. And this result is also consistent with the CV tests. GCD curves of WO3·2H2O/BC and BC (**Figure S1b**) electrodes recorded at 0.5, 1, 2, 5, and 10 A g−<sup>1</sup> are shown in **Figure 5D**. With the increasing charging and discharging currents, a highly linear and nearly symmetric relationship between the potential and time was also observed, suggesting the desired fast charging and discharging property of the materials. No obvious internal resistance (IR) drop of the BC (**Figure S1b**) and WO3·2H2O/BC electrodes was observed for any of the curves, which indicates high conductivity of the materials. The higher C<sup>s</sup> of WO3·2H2O/BC is due to the combined effects between WO3·2H2O and BC (Chu et al., 2017), and the superior electrochemical performance may be mainly attributed to the enhanced electronic conductivity after carbon incorporation. As shown in **Figure 5E**, the calculated C<sup>s</sup> of

FIGURE 7 | Anchored WO3.2H2O crystals with conductive BC; both electron and electrolyte can access WO3.2H2O surfaces, and an enhanced pseudocapacitive process can be formed on the surface of WO3.2H2O/BC.

WO3·2H2O/BC are 391, 270, 227.5, 203, and 177 F g−<sup>1</sup> at current densities of 0.5, 1, 2, 5, and 10 A g−<sup>1</sup> , respectively, demonstrating that the C<sup>s</sup> decreases with increasing current density (Shinde et al., 2017). The coulombic efficiency of WO3.2H2O/BC hybrids is ∼100%, exhibiting a good reversibility during the charge/discharge process. Furthermore, ∼45% of the capacitance was retained when the current density increased from 0.5 to 10 A g −1 , higher than the reported graphene nanosheets-tungsten oxides composites (34% capacitance retention from 0.1 to 5 A g −1 ), owing to the sluggish intercalation/deintercalation process of WO3.2H2O/BC at a high scan rate (Cai et al., 2014). These results are comparable to those of WO3-carbon-based electrodes in earlier reports (**Table S1**). To sum up, the enhancement in the electrochemical performance for WO3·2H2O/BC hybrids is mainly explained as follows: (i) the great interfacial contact between WO3·2H2O and BC provides short ion diffusion paths and the rapid electronic transports; (ii) the superior electrochemical performance may be mainly attributed to the enhanced electronic conductivity after carbon incorporation.

The EIS study was conducted to elucidate electrical conductivity and ion transfer features of BC and WO3·2H2O/BC electrodes. Electrochemical impedance characteristics of the electrode were investigated in a frequency range from 100 kHz to 0.1 Hz with an AC amplitude of 5 mV in 6 M KOH electrolyte. **Figure 5F** displays the typical Nyquist plots of these BC and WO3·2H2O/BC samples and the inset figure shows the magnified plots. For all the Nyquist plots, a semicircle can be seen in the high-frequency region, whereas in the low-frequency region, a straight line can be found. According to the previous reports (Sun et al., 2015; Chu et al., 2017; Yao C. et al., 2017), a straight line should relate to the ion diffusion into the active electrode (Zw), whereas the semicircle can be assigned to the charge transfer resistance (Rct) owing to the faradic and non-faradic reactions at the electrode/electrolyte interface. After doped with WO3·2H2O, the WO3·2H2O/BC electrode exhibits larger Rct and Zw than BC, mainly due to the poor electric conductivity of WO3.2H2O (Shinde et al., 2017). For comparison of first cycle and after 10,000 cycles, the Rct of WO3·2H2O/BC hybrids after the cycle is slightly larger than the value before the cycle, suggesting that WO3·2H2O/BC hybrids have good stability (Cai et al., 2014).

The cycling stability is the key factor to evaluate the practical applications of electrodes. To explore this, the cycle stability of WO3·2H2O/BC hybrids was further investigated by repeating the GCD test between −1 and 0 V at 4 A g−<sup>1</sup> for 10,000 cycles (**Figure 6**). Impressively, after 10,000 cycles, the initial Cs (220.8 F g−<sup>1</sup> ) of the WO3·2H2O/BC electrode slightly declined to 180.8 F g−<sup>1</sup> , and approximately 82% of the initial capacitance was retained. The inset shows the 1st, 4,000th and 10,000th GCD curves, respectively, indicating that the GCD profiles retain superior linearity and symmetry even after 10,000 cycles. Compared with other studies, to the best of our knowledge, in our work, the WO3·2H2O/BC hybrid electrode shows comparable cycling stability (**Table S1**); moreover the BC used in our work is much cheaper. Therefore, the combined effect of WO3·2H2O and BC accelerates the sufficient ion diffusion. This result confirms that the prepared WO3·2H2O/BC hybrids were highly stable as a novel supercapacitor electrode.

**Figure 7** shows the storage mechanism of WO3·2H2O/BC hybrids. Briefly, the mechanism of WO3·2H2O/BC hybrids can be described as follows. First, the high conductivity of BC facilitates the rapid transfer of electrons. Secondly, after the γirradiation process, the good contact of BC and WO3·2H2O guarantees a low internal resistance between BC and WO3·2H2O, and also facilitates the transmission of low-loss electrons to the anchored WO3·2H2O. Thirdly, the structure of WO3·2H2O facilitates ion adsorption and insertion in the electrolyte, so an enhanced pseudocapacitor process can be formed on the surface of WO3·2H2O. In addition, the good contact of BC and WO<sup>3</sup> also makes it difficult for WO3·2H2O to agglomerate during the charge/discharge process, and also ensures the long-term stability of WO3·2H2O/BC hybrids. All of these characteristics of WO3·2H2O/BC hybrids contribute to the high specific capacitance and good cycling stability.

### CONCLUSIONS

In summary, we successfully synthesized WO3·2H2O/BC hybrids via a facile γ-irradiation method. The electrochemical capacitive behaviors of the WO3·2H2O/BC hybrids and BC have been demonstrated in 6 M KOH electrolyte between −1 and 0 V. In comparison with the BC electrode, the WO3·2H2O/BC hybrid electrode showed a higher C<sup>s</sup> (391 F g−<sup>1</sup> at 0.5 A g−<sup>1</sup> ) and superior cycling performance (approximately 82% retention even after 10,000 cycles). The excellent performance achieved by the WO3·2H2O/BC hybrid electrode is owing to the combined effect of BC with good conductivity and WO3·2H2O with superior pseudocapacitive behavior. This γ-irradiation method would also pave a new way of designing other conducting semiconductors as promising electrode materials with enhanced performance for energy storage device applications.

### ETHICS STATEMENT

On behalf of, and having obtained permission from all the authors, I declare that: (a) the material has not been published in whole or in part elsewhere; (b) the paper is not currently being considered for publication elsewhere; (c) all authors have been personally and actively involved in substantive work leading to the report, and will hold themselves jointly and individually responsible for its content; I testify to the accuracy of the above on behalf of all the authors.

## AUTHOR CONTRIBUTIONS

YT developed the concept. XDL designed the experiments. FY and JJ conducted the experiments. RM and XCL built the cells and carried out the performance characterizations. ZF and CW supervised the research. FY and JJ co-wrote the manuscript. All authors discussed the results and commented on the manuscript.

### FUNDING

This study was financially supported by the Research Program of Sichuan province (grant no. 2016GZ0235) and the Development Foundation of China Academy of Engineering Physics (grant no. 2015B0302003) and the national key scientific instrument and equipment development project of China (grant no. 2014YQ090709).

### ACKNOWLEDGMENTS

We acknowledge Q. Yang, Y. Zeng, and J. Li (Science and Technology on plasma physics Laboratory, Research Centre

### REFERENCES


of Laser Fusion, China Academy of Engineering physics, Mianyang, China) for the TEM, SEM, and XRD measurements, respectively.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2018.00290/full#supplementary-material

nanowires nanocomposite. ACS Sustainable Chem. Eng. 5, 10128–10138. doi: 10.1021/acssuschemeng.7b02135


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Yang, Jia, Mi, Liu, Fu, Wang, Liu and Tang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Tuning the Catalytic Activity of Ir@Pt Nanoparticles Through Controlling Ir Core Size on Cathode Performance for PEM Fuel Cell Application

Hao-Bo Zheng<sup>1</sup> , Lu An<sup>1</sup> , Yuying Zheng<sup>1</sup> , Chong Qu<sup>2</sup> , Yanxiong Fang<sup>1</sup> , Quanbing Liu<sup>1</sup> \* and Dai Dang<sup>1</sup> \*

*<sup>1</sup> School of Chemical Engineering and Light Industry, Guangdong University of Technology, Guangzhou, China, <sup>2</sup> Department of Materials Science and Engineering College of Engineering, Peking University, Beijing, China*

### Edited by:

*Qiaobao Zhang, Xiamen University, China*

### Reviewed by:

*Haibin Sun, Shandong University of Technology, China Tao Wei, University of Jinan, China*

#### \*Correspondence:

*Quanbing Liu liuqb@gdut.edu.cn Dai Dang dangdai@gdut.edu.cn*

#### Specialty section:

*This article was submitted to Nanoscience, a section of the journal Frontiers in Chemistry*

Received: *31 May 2018* Accepted: *29 June 2018* Published: *26 July 2018*

#### Citation:

*Zheng H-B, An L, Zheng Y, Qu C, Fang Y, Liu Q and Dang D (2018) Tuning the Catalytic Activity of Ir@Pt Nanoparticles Through Controlling Ir Core Size on Cathode Performance for PEM Fuel Cell Application. Front. Chem. 6:299. doi: 10.3389/fchem.2018.00299*

Pulse electrochemically synthesis of a series of core-shell structured Ir@Pt/C catalysts in cathode catalysts layer are achieved to fabricate membrane electrode assemblies (MEA) with cathode ultra-low Pt loading. The single cell performance of the MEAs in a H2/air PEMFC greatly rely on the sizes of the Ir core nanoparticle, and the optimum activity occurs with Ir core size of 4.1 nm. The cathode MEA with core-shell structured catalysts with optimal Ir core size exhibited excellent performance in a H2/air single fuel cell, comparable to that of a commercial Pt/C MEA (Johnson Matthey 40% Pt), even though the Pt loading in Ir@Pt was only 40% that of the commercial Pt cathode (0.04 vs. 0.1 mg cm−<sup>2</sup> ). The catalysts were characterized by X-ray diffraction, X-ray photoelectron spectroscopy (XPS) and scanning transmission electron microscopy. Based on the characterization results, especially from XPS, we suggest that the effect of Ir core particle size on MEA performance may arise from the interactions between the Pt shell and the Ir core. The XPS results showed that the Ir@Pt/C-300 catalyst has the highest Pt<sup>0</sup> fraction among the four tested samples. This work demonstrates the alternative to enhance the cathode performance in single cell of Pt-based core-shell structured catalysts by varying size of the core metal under the Pt shell.

Keywords: core size effect, core-shell structure, low Pt loading, membrane electrode assembly, fuel cell

### INTRODUCTION

With the ever-growing concerns of global environment and increasing consumption of fossil fuels, the development of new energy conversion and storage system is of great significance across the world (Stamenkovic et al., 2016; Dai et al., 2017; Zhang et al., 2017, 2018; Zhao et al., 2018). Proton exchange membrane fuel cells (PEMFCs), regarded as the most encouraging clean power sources for future automobile, have been receiving unprecedented attentions due to their high energy efficiency, zero emission, and remarkable environmental acceptability (Debe, 2012; Dang et al., 2018). Nonetheless, there are some major issues that still slow down the pace of commercialization of PEMFCs (Shao et al., 2016; Tian et al., 2016). For example, the high loading of platinum and high cost of the Pt catalyst with sluggish oxygen reduction kinetics at cathode (Chen et al., 2015). However, the vital factor that affects the fuel cell performance is the catalyst layer within the membrane electrode assembly (MEA), which consists of Pt/carbon black and Nafion <sup>R</sup>

ionomer mixture (Kim et al., 2013). In general, the conventional MEA (prepared by catalyst coated membrane method, CCM) requires high Pt content (20–60%) to satisfy the chemical reactions needs. Whereas, a large fraction of Pt catalysts is not utilized by this approach (CCM) because Pt active nanoparticles are either losing ion contact with solid electrolyte or unable to access to the electronic path with carbon (Gasteiger et al., 2005). Accordingly, reducing the Pt loading within the electrode as well as without sacrificing on the cell performance is in harsh demand for PEMFC market, which not only requires the development of novel catalysts, but also strongly recommends a satisfactory nanostructured catalyst layer in the MEA.

Core-shell structured catalysts were introduced and developed for decades to find an answer. The core structure effect, including the shape, particle size, porosity, and composition were thoroughly investigated, and has shown significant impact on the oxygen reduction reaction (ORR) activity (Gan et al., 2012; Yang et al., 2013; Chen et al., 2014; Lu et al., 2014; Wittkopf et al., 2014; Zhang et al., 2014; Takimoto et al., 2017).

In recent years, Ir@Pt/C catalysts have been made to use in acid media for methanol oxidation and oxygen reduction reaction with the satisfactory results (Strickler et al., 2017). However, to the best of our knowledge, there is still no research about the core size effect in terms of Ir@Pt/C series catalysts tested in the single cell. Meanwhile, the Ir@Pt/C catalysts operated in a real PEM fuel cell cathode electrode environment has seldom been reported. Inspired by the continued achievements, we demonstrate a facile synthesis route for the core-shell structured catalyst within cathode catalysts layer, which is realized by pulse electrochemical deposition (PED) method. Intriguingly, we discovered that the Ir cores with different sizes may had some pronounced effects on the ORR performance. The Ir@Pt/C MEA prepared in the present work with the ultra-low Pt loading of 0.04 mg cm−<sup>2</sup> at cathode is outperformed than that of the commercial JM Pt/C with Pt loading of 0.1 mg cm−<sup>2</sup> .

### EXPERIMENTAL

### Ir/C Catalyst Preparation

Ir/C, the carbon-supported Ir core (Ir/C; 20 wt.% of Ir loading), was prepared by an impregnation-reduction method previously reported by our group (Dang et al., 2014). Briefly, IrCl<sup>3</sup> and pretreated XC-72R carbon black were both added in ethanol to form mixture. The mixture was magnetically stirred at 70◦C for 8 h to eliminate ethanol. The black powder was then placed in a ceramic boat and heated in a tubular furnace under flowing <sup>H</sup><sup>2</sup> for 2 h at 180, 240, 300, 400, and 500◦C, respectively. Then the heat-treated catalysts were labeled as Ir/C-180, Ir/C-240, Ir/C-300, Ir/C-400, and Ir/C-500.

### Synthesis of Ir@Pt/C MEA

The two-stage strategy was adopted to obtain cathode Ir@Pt/C electrode. Firstly, home-made Ir/C catalyst was mixed with 5 wt.% Nafion <sup>R</sup> ionomer solution (DuPont, USA) and isopropanol, then sonicated for 30 min to achieve homogeneous dispersion. The mixture was then directly sprayed onto one side of the membrane (Nafion <sup>R</sup> 212, Dupont) possessing an area of 5 cm<sup>2</sup> . The weight ratio of the Ir/C catalyst to dry Nafion <sup>R</sup> was 2.5:1. The Ir loading was 0.039 mg cm−<sup>2</sup> . Secondly, A laboratory device was used to prepare the Ir@Pt/C based MEA. The Ir/C based MEA was placed in a fixed area of 5 cm<sup>2</sup> , with the Ir/C serving as the working electrode, platinum wire and an Ag/AgCl electrode (3 M KCl) as the counter and reference electrode, respectively. Peak current densities were set to 30 mA cm−<sup>2</sup> , 0.3 ms of the time on and 0.15 ms of time off for PED process. The Pt loadings of the Ir@Pt/C cathodes were 0.04 mg cm−<sup>2</sup> for each of them, which were detected by atomic absorption spectroscopy (AAS).

The JM Pt/C MEA were prepared by the catalyst coated membrane method previously reported by our group using JM Pt/C (Johnson Matthey, 40%Pt) catalyst for both anode and cathode.

### Fuel Cell Measurements

The MEA was assembled by putting gas diffusion layers, which were prepared by spraying a carbon-Teflon <sup>R</sup> mixture on the pretreated carbon paper, on the anode and cathode side.

TABLE 1 | Crystallite sizes and lattice parameters from XRD measurements and binding energies from XPS measurements of Ir/C series catalysts.


The MEA was tested using a Fuel Cell Testing System (Arbin Instruments, USA). The cell temperature was maintained at 70◦C. H<sup>2</sup> and air, as the fuel and oxidant gas respectively, were fully humidified (100% humidification, hydrogen and air both were set at 70◦C) before feeding with a flow rate of 120 sccm for H<sup>2</sup> and 800 sccm for air. The back pressure for both the anode and cathode were 30 psi.

### Characterizations of the MEAs

The morphology of the Ir@Pt/C catalyst, which was washed off from the cathode MEA by ethanol and was observed using a high-resolution transmission electron microscope (JEOL JEM-2010HR, Japan) operated at 300 kV. High-angle annular dark field (HAADF) images and energy dispersive spectrometer (EDS) elemental line scan analysis were obtained using scanning transmission electron microscopy (STEM) mode on an aberration-corrected FEI Titan G<sup>2</sup> 60–300 field emission transmission electron microscope (FEI), operated at 300 kV (a max∼100 mrad). The nanoparticle crystal structure was determined by X-ray diffraction (TD-3500, Tongda, China) using filtered Cu Kα radiation at 40 kV and 30 mA. The 2θ region between 20◦ and 80◦ was measured at a scan rate of 4◦ min−<sup>1</sup> . X-ray photoelectron spectroscopy (XPS) on a PerkinElmer PHI1600 system (PerkinElmer, USA) using a single Mg Kα X-ray source operating at 300 W and 15 kV. The XPS spectra (BEs) were calibrated using the C 1 s peak of graphite at 284.6 eV as the reference.

corresponding EDS line scan profiles (c).

### RESULTS AND DISCUSSION

**Figure 1** shows the XRD patterns of Ir/C catalysts annealed at various temperatures. For all the samples, the diffraction peaks at 2θ of 40.6◦ , 47.31◦ , and 69.1◦ can be indexed, respectively, to the (111), (200), and (220) planes of face-centered cubic (fcc) iridium. The expected sharper and more intense diffraction peaks observed as the temperature rose from 180 to 500◦C were caused by an increase in the extent of crystallization and particle agglomeration at higher temperatures. The particle sizes, which were calculated from the XRD data by Jade software, were 1.8, 2.9, 4.1, 5.8, and 7.1 nm for Ir/C-180, Ir/C-240, Ir/C-300, Ir/C-400, and Ir/C-500, respectively. The (111) diffraction peak was used to calculate the lattice parameters, which are listed in **Table 1**. It can be seen that the parameters were almost the same for each sample, indicating that the lattice was not affected by the annealing temperature when it increased from 180 to 500◦C. However, the 4f binding energy of Ir in the samples clearly shifted with the annealing temperature or particle size (see **Figure 2** and **Table 1**). As the annealing temperature/particle size increased, the binding energy of Ir 4f decreased, which is quite consistent with previous reports from other groups (Kuznetsova et al., 2012).

**Figure 3** presents the TEM images of Ir@Pt/C-300 catalysts to get information about structure and morphology changes after the Pt deposition. Typically, the catalysts for the TEM analysis TABLE 2 | Binding energies (BE) of Pt 4f 7/2 and Ir 4f 7/2 for Pt/C, Ir@Pt/C-180, Ir@Pt/C-300, and Ir@Pt/C-500, with the fraction of Pt<sup>0</sup> and Ir<sup>0</sup> content in each catalyst.


were peeled off from the MEA's surface by ethanol. The Ir@Pt-300 nanoparticles shown in **Figure 3a** were homogenous dispersed on the carbon support with a narrow size distribution. The average particle size of Ir@Pt-300 nanoparticles was calculated to be 5.4 nm.

In order to further confirm the core-shell structure of Ir@Pt/C-300 catalysts, the high-angle annular dark-field (HAADF) TEM and EDS line scan on a single nanoparticle were performed as shown in **Figures 3b,c**. It can be seen that Pt signal was highly populated at the edge of particle while Ir signal was concentrated at the center, which clearly manifested the core-shell structure of our Ir@Pt/C-300 catalyst.

XPS results revealed that the particle size of the Ir core nanoparticles significantly affected the electronic structure of

the catalysts. After Pt deposition on the different Ir cores, the core-level binding energies of Pt 4f differed from those in monometallic Pt, as shown in **Figure 4** and summarized in **Table 2**. All the Pt 4f signals of the Ir@Pt/C samples exhibited a positive shift compared to the Pt/C catalysts. However, the Pt 4f binding energy of Ir@Pt/C-300 (71.82 eV) was higher than that of Ir@Pt/C-180 (71.73 eV) and Ir@Pt/C-500 (71.76 eV), suggesting a considerable tuning of the Pt surface electronic structure of Ir@Pt/C-300. In addition, as **Figure 4** shows, the Ir@Pt/C-300 catalyst exhibited a higher proportion of Pt<sup>0</sup> (76.7%) than the rest of the catalysts: Ir@Pt/C-180 (56.7%), Ir@Pt/C-500 (69.5%), and commercial Pt/C (54.5%). It has been reported that the Pt<sup>0</sup> fraction is a crucial factor associated with ORR catalysis; the relatively high proportion of metallic Pt in Ir@Pt/C-300 implies that it had weaker oxophilicity than the other catalysts, resulting in high ORR performance (Dang et al., 2015). On the other hand, the Ir 4f binding energy of the Ir@Pt/C series MEAs tells a different story. After Pt deposition, the Ir 4f 7/2 values for Ir@Pt/C-180 and Ir@Pt/C-300 shifted negatively compared to Ir/C-180 and Ir/C-300, but the value for Ir@Pt/C-500 shifted positively, indicating an interaction between the Pt shell and the Ir core, and that the interaction varied with the particle size of the Ir core NPs. The slight decrease in Ir 4f binding energy for Ir@Pt/C-300 suggests electron transfer occurred between Pt and Ir, which may have changed the d-band centers of the surface metals and thus improved their surface adsorption and desorption behavior in the ORR process.

In order to investigate the surface structure and to explore the electrochemical surface area (ECSA) of different catalysts, cyclic voltammograms was implemented in the cell system. As shown in Figure S1, CV plots of different cathode MEAs were measured in a single cell. It can be observed that JM Pt/C MEA prepared by CCM method, has the largest ECSA (86.1 m<sup>2</sup> g −1 ) since there may be too much contact between Pt atoms and Nafion <sup>R</sup> in the catalyst layer. As for Ir@Pt/C MEA realized by pulse electrochemical deposition approach, the ECSAs were smaller than that of JM Pt/C MEA. It should be mentioned that every Pt atom would be reduced on the surface of Ir nanoparticle at triple-phase area and establish an appropriate contact to the nearby Nafion <sup>R</sup> ionomer rather than fully wrapped in it. Besides, it can be observed that when the Ir core size increases from 1.8 to 7.1 nm, the intensity of the H<sup>2</sup> desorption peak decreases and thus the calculated ESCAs of this series Ir@Pt/C MEA became smaller with the same Pt amounts. Generally, increasing the Ir core size would lead to the decrease of the surface area of Ir nanoparticle, and therefore it could lower the ECSA of Ir@Pt/C catalysts. Furthermore, the positive shift of Pt reduction peak of the Ir@Pt/C MEA at around 0.8 V was also discovered. This may be a sign of Pt coverage on the Ir core and implies that Pt oxides are less likely to form on the Ir core and thus tune the catalytic activity of the core-shell structured catalysts.

Through decorating of platinum atoms on the Ir nanoparticles, a series of Ir@Pt/C MEA were successfully prepared by PED method with ultra-low Pt loadings down to 0.04 mgPt cm−<sup>2</sup> at cathode. The anode with Pt loading of 0.1 mgPt cm−<sup>2</sup> was prepared using JM 40% Pt/C catalysts. **Figure 5A** shows the polarization curves obtained by Pt mass with different Ir@Pt/C MEAs and JM Pt/C MEA. As shown in the **Figure 5A**, the Ir@Pt/C-300 MEA exhibited a best cell performance and Pt utilization efficiency in the whole polarization region whereas JM Pt/C MEA resulted in the poorest performance. With the increasing of Ir core size (from 1.8 to 4.1 nm), the cell performances have been substantially improved. It is interesting that the cell performance then slightly deceased when further elevating the heating temperature from 300 to 500◦ .

In **Figure 5B**, the maximum power density of Ir@Pt/C MEAs and JM Pt/C MEA were calculated and benchmarked by Pt loading. It reveals that there is a solid improvement in Pt utilization with Ir@Pt/C-300 at the cathode compared to that of the JM Pt/C MEA cathode. The **Figure 5B** also exhibited the correlation between maximum power densities and Ir core size. The maxmum mass activity increases and reaches its peak value at Ir core size of 4.1 nm, and then drops slightly when the Ir core size further increases. This indicates that different Ir core sizes leads to distinct cathode catalytic behavior of Ir@Pt/C catalysts, which significantly affect the ORR activity through tuning the surface electronic structure between Ir and Pt.

Figure S2 showed a preliminary durability test of fuel cell with Ir@Pt/C-300 MEA for 100 h under constant discharge process at 800 mA cm−<sup>2</sup> . There was no obvious fluctuation of operation voltage during the test, which revealed good initial stability of our Ir@Pt/C-300 MEA.

### CONCLUSION

A series of Ir@Pt/C MEAs were successfully prepared through construction of core-shell structured catalysts in the cathode catalyst layer with a facile pulse electrodeposition synthesis route. By optimizing the Ir core size dependent on the thermal treatment, the cell performance of different MEA were investigated and balanced to the acceptable choice. The asprepared Ir@Pt/C series MEA that contains the ultra-low Pt loading (0.04 mg cm−<sup>2</sup> at cathode) showed the competitive single cell performance and Pt utilization efficiency compared with those of JM Pt/C MEA (0.1 mg cm−<sup>2</sup> at cathode). It is suggested that the optimum Ir core size with 4.1 nm shows appropriate crystallinity which will tune the surface electronic structure between Ir and Pt and finally lead to the high cell performance.

### REFERENCES


### AUTHOR CONTRIBUTIONS

DD and QL conceived the project. H-BZ carried out the experiment and collected the data. LA performed the electrochemical tests. YZ, CQ, and YF made contribution to the analysis and discussion of the results. DD drafted the manuscript, and all authors contributed to the final version of the manuscript.

### ACKNOWLEDGMENTS

This work was supported by the National Natural Science Foundation of China (No.21606050), Pearl River Science and Technology New Star Project (No. 201806010039), and Guangdong University Characteristics Innovation Project (No. 2017KTSCX055).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2018.00299/full#supplementary-material

transfer in fuel cells using inverse opal structure. Nat. Commun. 4:2473. doi: 10.1038/ncomms3473


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Zheng, An, Zheng, Qu, Fang, Liu and Dang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# SPEEK Membrane of Ultrahigh Stability Enhanced by Functionalized Carbon Nanotubes for Vanadium Redox Flow Battery

Mei Ding<sup>1</sup> , Xiao Ling<sup>2</sup> , Du Yuan<sup>3</sup> , Yuanhang Cheng<sup>2</sup> , Chun Wu<sup>1</sup> , Zi-Sheng Chao<sup>1</sup> , Lidong Sun4,5 \*, Chuanwei Yan<sup>2</sup> and Chuankun Jia1,2,5 \*

*<sup>1</sup> College of Materials Science and Engineering, Changsha University of Science and Technology, Changsha, China, <sup>2</sup> Institute of Metal Research, Chinese Academy of Sciences, Shenyang, China, <sup>3</sup> Department of Materials Science and Engineering, Faculty of Engineering, National University of Singapore, Singapore, Singapore, <sup>4</sup> State Key Laboratory of Mechanical Transmission, School of Materials Science and Engineering, Chongqing University, Chongqing, China, <sup>5</sup> Key Laboratory of Advanced Energy Materials Chemistry (Ministry of Education), Nankai University, Tianjin, China*

Edited by:

*Qiaobao Zhang, Xiamen University, China*

### Reviewed by:

*Gen Chen, Central South University, China Fan Yang, Purdue University, United States*

#### \*Correspondence:

*Lidong Sun lidong.sun@cqu.edu.cn Chuankun Jia jack2012ding@gmail.com*

#### Specialty section:

*This article was submitted to Physical Chemistry and Chemical Physics, a section of the journal Frontiers in Chemistry*

> Received: *31 May 2018* Accepted: *25 June 2018* Published: *26 July 2018*

#### Citation:

*Ding M, Ling X, Yuan D, Cheng Y, Wu C, Chao Z-S, Sun L, Yan C and Jia C (2018) SPEEK Membrane of Ultrahigh Stability Enhanced by Functionalized Carbon Nanotubes for Vanadium Redox Flow Battery. Front. Chem. 6:286. doi: 10.3389/fchem.2018.00286* Proton exchange membrane is the key factor of vanadium redox flow battery (VRB) as their stability largely determine the lifetime of the VRB. In this study, a SPEEK/MWCNTs-OH composite membrane with ultrahigh stability is constructed by blending sulfonated poly(ether ether ketone) (SPEEK) with multi-walled carbon nanotubes toward VRB application. The carbon nanotubes disperse homogeneously in the SPEEK matrix with the assistance of hydroxyl group. The blended membrane exhibits 94.2 and 73.0% capacity retention after 100 and 500 cycles, respectively in a VRB single cell with coulombic efficiency of over 99.4% at 60 mA cm−<sup>2</sup> indicating outstanding capability of reducing the permeability of vanadium ions and enhancing the transport of protons. The ultrahigh stability and low cost of the composite membrane make it a competent candidate for the next generation larger-scale vanadium redox flow battery.

Keywords: redox flow battery, vanadium redox flow battery, ion exchange membrane, sulfonated poly(ether ether ketone), carbon nanotube

### INTRODUCTION

Vanadium redox flow battery (VRB) is one of the most promising large-scale energy storage technologies with the advantages of decoupled energy storage and power output, flexible design, long lifetime, and so on (Li et al., 2011; Skyllas-Kazaocs et al., 2011; Weber et al., 2011; Yang et al., 2011; Wang et al., 2013). Nevertheless, the commercialization of the VRB is still hindered by low stability and high cost of the membrane (Jia et al., 2014; Noorden, 2014; Zhang, 2014), which is the key component to separate the catholyte and anolyte and to transport protons (Jiang et al., 2016; Yu and Xi, 2016). An ideal membrane is expected to bear high chemical stability, good mechanical strength, high proton conductivity, and low vanadium ions crossover between anolyte and catholyte to suppress the imbalance of electrolyte. To date, the Nafion membranes are widely used in the VRB systems because of their high proton conductivity and excellent chemical stability (Reed et al., 2015; Teng et al., 2015). However, the high cost and high vanadium ion permeability deteriorate the battery performance, and thus are unsuitable for large-scale applications and future commercialization (Xi et al., 2007; Luo et al., 2013; Kima et al., 2014; Lin et al., 2015). Recently the sulfonated poly (ether ether ketone) (SPEEK) membrane is considered as a substitution to the Nafion, in view of its low cost (Ling et al., 2012; Liu et al., 2014) and high proton conductivity (Jia et al., 2012; Wei et al., 2012). Nonetheless, the SPEEK membrane suffers from noticeable vanadium crossover in light of the high degree of sulfonation, giving rise to low mechanical/chemical stability and hence short cycling lifetime of VRB. Many reports have reported that blending the high sulfonated SPEEK membrane with materials lowers its crossover of vanadium ions (Wang et al., 2012; Dai et al., 2014a; Li et al., 2014). However, the mechanical and chemical stabilities of these blended SPEEK membranes are still unable to meet the long-term cycling requirements of VRB. On the other hand, blended SPEEK membrane with graphene and carbon nanotubes exhibits better stability compared to the membrane blended with other materials (Dai et al., 2014b; Jia et al., 2015a). The single VRB cell with SPEEK/graphene composite membrane reported by Dai et al. exhibited a capacity retention of 56.3% upon 300 cycles (Dai et al., 2014b). Our previous work reported a reinforced mechanical strength of the SPEEK/SCCT composite membrane for VRB (Jia et al., 2015a). However, until now, to the best of our knowledge, ultra-long lifetime (>1,000 cycles) in SPEEK based membranes has been a critical but unaddressed issue for future commercialization of VRBs.

Herein, we report a novel SPEEK membrane enhanced by hydroxylated multi-walled carbon nanotubes (SPEEK/MWCNTs-OH). The composite membrane exhibits high mechanical strength and low vanadium ions permeability, owing to the reinforcement effect of SPEEK and MWCNTs-OH. The proton conductivity is promoted, since the OH groups on MWCNTs form additional channels for proton transport. The VRB single cells based on the SPEEK/MWCNTs-OH membrane show an ultralong cycling lifetime (94.2% capacity retention after 100 cycles, 73.0% capacity retention after 500 cycles) and ultrahigh Coulombic efficiency (>99.4%). The current study contributes to the design of advanced membranes for efficient VRB toward large-scale energy storage applications.

### EXPERIMENTAL SECTION

Poly(ether ether ketone) (Victrex, PEEK 450 PF), Dimethylsulfoxide (DMSO) and sulfuric acid (98 wt.%) were purchased from Sinopharm Chemical Reagent Co. Ltd. The MWCNTs-OH (length: 0.5–2 mm, diameter: <8 nm, purity: >95%, OH content: 5.58 wt.%) were purchased from Nanjing XFNANO Materials Tech Co. Ltd. All chemicals were used as purchased without further purification.

The sulfonation process of PEEK was described in our previous work (Jia et al., 2010). The degree of sulfonation for the SPEEK was determined by titration, about 50.2% in this study. In order to enhance the mechanical property of the SPEEK matrix while keep its low price, low electron conductivity, a low MWCNTs-OH concentration (0.7 wt%) was chosed in this work. The SPEEK/MWCNTs-OH membrane was prepared sequentially by the steps as follows: (1) 1.7 g SPEEK was dissolved in 40 mL DMSO, resulting a SPEEK solution; (2) 12 mg MWCNTs-OH was added into the above SPEEK solution under magnetic stirring; (3) the resulting solution was stirred for more than 5 h and subsequently sonicated for 10 min at room temperature; (4) the SPEEK/MWCNTs-OH solution was cast onto a home-made glass mold and heated at 100◦C for 15 h in an oven until all of DMSO being evaporated; (5) the resultant membrane was cooled down to room temperature and peeled off from the glass mold. The thus-obtained SPEEK/MWCNTs-OH membrane (thickness is 90µm) was stored in deionized water before its characterization and application in VRB.

The surface morphology of the membrane was examined by scanning electron microscopy (SEM, ZEISS). The membrane was sputtered with gold prior to SEM observation. The mechanical property of the membrane was measured using a CMT 6502 tension tester (Shenzhen Instron Corporation China). The tensile strength of membrane was calculated using the equation described in literature as follows (Ling et al., 2012):

$$\text{Mechanical strength} = \text{S}\_{\text{m}}/(\text{W} \times \text{L})$$

Where S<sup>m</sup> is the maximum strength of membrane, W is the width of different membrane samples and L is thickness of different membranes.

The permeability of VO2<sup>+</sup> ions across the membrane was measured according to our previous method (Jia et al., 2010, 2015a). Briefly, a redox flow single cell was filled up with two kinds of electrolytes in respective reservoirs, i.e., 70 mL of 1.5 M VOSO<sup>4</sup> in 2.0 M H2SO<sup>4</sup> solution and 70 mL of 1.5 M MgSO<sup>4</sup> in 2.0 M H2SO4. Solution of 1 mL was sampled from the MgSO<sup>4</sup> compartment at certain time interval and was measured with the UV-vis spectrometer. The solutions in both reservoirs were continuously stirred to avoid concentration polarization. The permeability was calculated using the following equation:

$$\mathbf{V\_B \frac{dC\_B\left(t\right)}{dt} = A\frac{P}{L}\left(C\_A - C\_B\left(t\right)\right)}$$

where V<sup>B</sup> is the volume of MgSO<sup>4</sup> in the reservoir, CB(t) is the concentration of VO<sup>+</sup> 2 ions in the MgSO<sup>4</sup> compartment as a function of time t, C<sup>A</sup> is the VO<sup>+</sup> 2 concentration in the VOSO<sup>4</sup> compartment, A is the active area of the membrane (28 cm<sup>2</sup> ), L is the thickness of the membrane, and P is the permeability of V (IV) ions.

A single cell with 1.5 M VOSO<sup>4</sup> and 2 M H2SO<sup>4</sup> as the catholyte (80 mL) and anolyte (80 mL), respectively, was employed to determine the area resistance (R) of the membranes by DME-20 Battery Internal Resistance Tester. The area resistance and conductivity (σ) of the membranes were computed as:

$$\mathbb{R} = (\mathbb{R}\_1 - \mathbb{R}\_2) \times \mathbb{A}$$

and σ = L/R

where R<sup>1</sup> and R<sup>2</sup> represent the resistances of the single cell with and without SPEEK/MWCNTs-OH, respectively.

Galvanostatic intermittent titration technique (GITT) measurement was conducted in a VRB single cell to further study the area resistance and the permeation of vanadium ions through the membranes, as detailed in reference (Jia et al., 2015b). In this test, 25 mL 1.5 M V2+/V <sup>3</sup><sup>+</sup> in 2.0 M H2SO<sup>4</sup> and 25 mL 1.5 M VO2+/VO<sup>+</sup> 2 in 2.0 M H2SO<sup>4</sup> solutions were cycled through anodic and cathodic reservoirs, respectively. The GITT measurements were carried out ranging from 0.7 to 1.65 V. The cell was charged and discharged at a current density of 40 mA cm−<sup>2</sup> for 4 min, followed by an open circuit relaxation for 4 min.

The open circuit voltage (OCV) of the VRB single cell was monitored at room temperature after charged to a charge state of 75%.

The VRB single cell was assembled with two carbon felt electrodes, two conductive plastic plate current collectors, membrane and steel endplate. The active area is 28 cm<sup>2</sup> . 25 mL 1.5 M V2+/V <sup>3</sup><sup>+</sup> in 2.0 M H2SO<sup>4</sup> and 25 mL 1.5 M VO2+/VO<sup>+</sup> 2 in 2.0 M H2SO<sup>4</sup> were used as anolyte and catholyte, respectively. The VRB single cell was charged and discharged at a constant current density of 50 mA cm−<sup>2</sup> with the voltage ranging from 0.7 to 1.65 V. After around 380 cycles, the compartments were refilled with fresh electrolytes and the carbon felt electrodes were replaced with new pieces. The rate performance was studied at a current density ranging from 40 to 120 mA cm−<sup>2</sup> and then back to 80 mA cm−<sup>2</sup> for cycling.

### RESULTS AND DISCUSSION

The mechanical property of the membranes is of first concern in view of the long-term stability and service for a VRB system. It has been determined that the mechanical strength increases upon blending with the MWCNTs-OH from 40.1 to 61.8 MPa. This can be attributed to the reinforcement from the carbon nanotubes, which are of high mechanical strength and behaves as strengthening component in the composite. The hydroxyl groups that graft on the surface play an important role in forming the composite structure, making the nanotubes bonded strongly to the SPEEK matrix, as illustrated in **Figure 1**.

The permeability of SPEEK membranes is highly suppressed by blending with the carbon nanotubes, which is even better than commercial Nafion 212. As can be seen from **Figure 2A** that the concentration of VO2<sup>+</sup> ions across the SPEEK/MWCNTs-OH membrane is much lower than the Nafion 212, about 10-fold less at 120 h. The permeability of VO2<sup>+</sup> ions is thus computed to be about 1.93 × 10−<sup>7</sup> cm<sup>2</sup> min−<sup>1</sup> for the SPEEK/MWCNTs-OH membranes, over 4 times lower than that of Nafion 212 (8.23 × 10−<sup>7</sup> cm<sup>2</sup> min−<sup>1</sup> ). The permeation of vanadium ionic species (cathode, VO<sup>+</sup> 2 /VO2+; anode, V3+/V2+) across the membrane leads to a severe self-discharge of the battery and in turn decay of the open circuit voltage (OCV). **Figure 2B** compares the OCV decay of the VRB cells with different membranes, where the cells were initially charged to 75% state of capacity. Both of the decay curves gradually decrease until a drastic drop appears. The retention time over 1.3 V in SPEEK/MWCNTs-OH membrane is about 100 h longer than that of Nafion 212. This further confirms the superior ability of the SPEEK/MWCNTs-OH membrane in suppressing vanadium ion crossover, consistent with the permeability test. In general, the crossover becomes more serious when the cells are under intermittent charging-discharging process. **Figure 2C** displays a longer retention time of the discharge potential for SPEEK/MWCNTs-OH membrane as compared to that of Nafion 212, in good agreement with the above discussion. Consequently, the composite membrane of SPEEK/MWCNTs-OH suppresses the permeability of vanadium ions. This can be attributed to the additives of carbon nanotubes that block the diffusion of bulky vanadium ions.

The charge-discharge curves of VRB cells with SPEEK/MWCNTs-OH and Nafion 212 membrane are compared in **Figure 3A**. It clearly shows that the discharge capacity of cells with SPEEK/CNTs-OH membrane is larger than that with Nafion 212, thereby giving rise to a higher Columbic efficiency (CE). This is mainly due to the lower permeability and thus lower self-discharge. Meanwhile, a relatively higher and lower average charge and discharge voltage for cells with SPEEK/MWCNTs-OH membranes are observed respectively, because of the large IR drop resulted from the high area resistance (1.04 vs. 0.6 cm<sup>2</sup> for SPEEK/MWCNTs-OH and Nafion 212 membranes Jia et al., 2012, respectively). The corresponding IR drop is around 70 and 55 mV, in accordance with the GITT results (see **Figure 2C**). The discharge capacity of the VRB single cell with SPEEK/MWCNTs-OH membrane is larger than 0.6 A h (>60% of the initial discharge capacity at 40 mA cm−<sup>2</sup> ) even at a current density of 120 mA cm−<sup>2</sup> , indicating a good rate performance (**Figure 3B**).

High Columbic efficiency (CE) in the VRB cell indicates low capacity loss in a charge-discharge cycle, as the CE is the ratio of discharge to charge capacity (He et al., 2015; Jia et al., 2015a). The capacity loss mainly arises from permeability of vanadium ions, which is in turn closely related to the membrane structure. The SPEEK/MWCNTs-OH membrane shows no significant phase separation while presents much rougher surface as compared to the original SPEEK membrane, as displayed in **Figures 4a,b**. It reveals that small clustering of carbon nanotubes are embedded uniformly in the SPEEK matrix, therefore increasing the surface roughness and decreasing the pathway for ion diffusion. Besides, the carbon nanotubes behave as extra barriers to vanadium ions.

retention (DR, blue).

Therefore, the composite membrane suppresses the permeability of vanadium ions. More interesting, the proton conductivity is also enhanced with the composite structure (about 10.1 mS cm−<sup>1</sup> , 1.5-fold higher than that of SPEEK membrane). Detailed examination discloses worm-like structures (>500 nm in length) on the surface of composite membrane, as shown in **Figure 4c**. This can be assigned to the traces of carbon nanotubes (500– 2,000 nm in length used in this study) that bond to the SPEEK matrix. Generally, the carbon nanotubes play the following roles in the SPEEK polymer matrix (see the inset in **Figure 4c**): (1) the nanotubes are aligned to the SPEEK backbones, forming protective layers to reduce the corrosion of aromatic backbones in a VRB cell; (2) the nanotubes increase the tortuosity of ionic channels developed by sulfonic acid groups (S-channel); (3) the hydroxyl groups at the nanotube surface form new ionic channels (C-channel) and hence facilitate proton transport (He et al., 2015); (4) the carbon nanotubes bridge between the clusters and further contribute to the proton transport (Tunuguntla et al., 2016). Therefore, with the combination of reduced permeability of vanadium ions and enhanced conductivity of protons, the composite membrane of SPEEK/MWCNTs-OH provides an effective solution to the problem of capacity fade and short lifetime.

It is a consensus that the bright and dark domains in an AFM image represent hard hydrophobic and soft hydrophilic areas, respectively (James et al., 2000; Affoune et al., 2004). As shown in **Figure 5a**, the dark hydrophilic regions of SPEEK membrane form some long channels and are separated by bright hydrophobic regions. In contrast, for SPEEK/MWCNTs-OH membrane in **Figure 5b**, the hydrophilic regions are separated by carbon nanotubes and yield many small isolated dots. This is due to the carbon nanotubes that form C-channels, bridge Schannels, and therefore facilitate proton transport. This further echoes the mechanism described in **Figure 4**.

The long-term stability of a VRB system is of paramount importance in its practical application for large scale energy storage. However, it is difficult to verify the stability of membrane by cycling in a VRB cell, since the stability of carbon felt electrode under strong acid and VO<sup>+</sup> 2 condition (e.g., the formation of V2O<sup>5</sup> precipitates in catholyte) also leads to the VRB operation failure (Jia et al., 2010; Yuan et al., 2016). In this regard, the electrodes and electrolyte were designed to replace with refresh ones after the first charge-discharge test (i.e., the VRB cell running for 440 h). It reveals that red V2O<sup>5</sup> precipitates formed at the edge of cathodic electrode after the test (see Figure S1 in the supporting information). Meanwhile, the XPS results show that the amount of oxygen-containing groups on carbon felt increases, such as –COOH, OH (see Figure S2 in the supporting information), resulting in an increased electrode resistance and thus a higher IR drop (see Figure S3 in the supporting information). In contrast, no increment in resistance and IR drop is observed when replacing the electrodes and electrolyte but keeping the SPEEK/MWCNTs-OH membrane (see **Figure 3B**). As such, the replacement was implemented in the subsequent stability test.

**Figure 6** shows the cycling performance of the VRB with SPEEK/MWCNTs-OH membranes, which were initiated at a current density of 50 mA cm−<sup>2</sup> . After 380 cycles (i.e., 450 h), the catholyte and anolyte were refreshed while the electrodes were replaced. The cycling was continued at the following current densities sequentially: 40, 60, 80, 100, and 120 mA cm−<sup>2</sup> and then back to 80 mA cm−<sup>2</sup> . The first cycle in the charge-discharge curves at each current density is presented in **Figure 3B**. **Figure 6** shows that the coulombic efficiency of the cell employing the SPEEK/MWCNTs-OH membrane is over 99% during the entire cycling test (1,550 cycles, 1,030 h). The energy efficiency (EE) is about 80% at the first cycling test at 50 mA cm−<sup>2</sup> , while about 75% at subsequent cycling test (2nd test in VRB at 80 mA cm−<sup>2</sup> ). The discharge retention is over 94.2% after 100 cycles and above 50% after 950 cycles. Thereafter, the retention drops substantially, because of decreased active area for chemical reactions caused by increased corrosion of carbon electrodes and deposition of VO<sup>+</sup> 2 at the cathode. Therefore, the VRB single cell with SPEEK/MWCNTs-OH membranes exhibit an ultrahigh CE and long lifetime.

The FTIR spectra were measured from the SPEEK/MWCNTs-OH membrane before and after the VRB single cell test. The spectra are normalized to intensity at 877.6 cm−<sup>1</sup> and therefore the difference between the normalized spectra was obtained by subtracting the spectrum of SPEEK/CNT-OH membrane after single cell test from the fresh SPEEK/CNT-OH membrane. The peaks at 926, 1,023, and 1,080 cm−<sup>1</sup> are assigned to the specific S-O vibration [v(S-O)] (Swier et al., 2005), the asymmetric vibration [vas(SO3H)] and the symmetric vibration [vs(SO3H)] (Xing et al., 2004), respectively. The value of the difference spectrum ISPEEK/MWCNTs−OH – ISPEEK at the three peaks arising from -SO3H are almost zero (Figure S4), indicating almost the same amount of -SO3H in both membranes. This is consistent with the fact that small amount of carbon nanotubes was added to the SPEEK matrix without affecting the equivalent weight (Jalani and Datta, 2005). Peaks from - SO3H appear in the difference spectrum Ibefore − Iafter of the SPEEK/MWCNTs-OH membrane indicating the degradation of the SPEEK/MWCNTs-OH membrane resulting from falling of the -SO3H off the surface (**Figure 7**). Nevertheless, the degradation is not significant as the difference between the SPEEK/MWCNTs-OH membrane before and after single cell test is within 5%. We obtained the intensity of the three peaks from -SO3H in both the difference spectrum Ibefore − Iafter and the spectrum from membrane before test by integrating the corresponding regions as remarked in **Figure 7** (Table S1), the ratio of the difference spectrum to the spectrum from membrane before test are ∼ 5%. After a long cycling time in the single VRB cell, SPEEK/MWCNTs-OH membranes show distinctively low degradation indicating a remarkably enhanced chemical stability of the SPEEK membrane. The stability enhancement of the membrane is indeed results from the addition of carbon nanotubes.

### CONCLUSION

A composite membrane of SPEEK polymer blended with hydroxylated multi-walled carbon nanotubes (MWCNTs-OH) is developed toward VRB application. The membrane of SPEEK/MWCNTs-OH exhibits high mechanical strength of 61.8 MPa, suppressed permeability of vanadium ions, enhanced proton conductivity. This is attributed to the carbon nanotubes of high strength that bond strongly to the SPEEK matrix, which provides barriers for vanadium ion diffusion while additional channels for proton transport. The superior ionic selectivity, excellent proton conductivity and stability of the composite

REFERENCES


membranes ensure ultrahigh stability of over 1000 cycles in VRB cells. Such a composite membrane of SPEEK/MWCNTs-OH is of low cost, high performance and long-term stability, making it promising for the next generation large-scale VRB applications.

### AUTHOR CONTRIBUTIONS

MD design and performed the experiments. MD, XL, DY, YC, and CW prepared the samples and analyzed the data. MD, Z-SC, LS, CY, and CJ participated in interpreting and analyzing the data. All authors read and approved the final manuscript.

### ACKNOWLEDGMENTS

The authors acknowledge the support from the 100 Talented Team of Hunan Province, the financial support from the National Natural Science Foundation of China (No. 51501024), the Chongqing Research Program of Basic Research and Frontier Technology (No. cstc2015jcyjA90004), and the Fundamental Research Funds for the Central Universities (No. 2018CDQYCL0027), and the 111 project (B12015) at the Key Laboratory of Advanced Energy Materials Chemistry (Ministry of Education), Nankai University.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2018.00286/full#supplementary-material


performance vanadium redox flow battery. J Power Sources 257, 221–229. doi: 10.1016/j.jpowsour.2014.01.127


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Ding, Ling, Yuan, Cheng, Wu, Chao, Sun, Yan and Jia. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Effect of Environmental Temperature on the Content of Impurity Li3V2(PO4)3/C in LiVPO4F/C Cathode for Lithium-ion Batteries

Taotao Zeng<sup>1</sup> , Changling Fan1,2 \*, Zheng Wen<sup>1</sup> , Qiyuan Li <sup>1</sup> , Zeyan Zhou<sup>1</sup> , Shaochang Han<sup>1</sup> and Jinshui Liu1,2 \*

<sup>1</sup> College of Materials Science and Engineering, Hunan University, Changsha, China, <sup>2</sup> Hunan Province Key Laboratory for Advanced Carbon Materials and Application Technology, Hunan University, Changsha, China

#### Edited by:

Jiexi Wang, Central South University, China

#### Reviewed by:

Xianwen Wu, Jishou University, China Jie Shu, Ningbo University, China Guochun Yan, Collège de France, France Haoran Jiang, Hong Kong University of Science and Technology, Hong Kong

#### \*Correspondence:

Changling Fan fancl@hnu.edu.cn Jinshui Liu jsliu@hnu.edu.cn

#### Specialty section:

This article was submitted to Physical Chemistry and Chemical Physics, a section of the journal Frontiers in Chemistry

> Received: 26 April 2018 Accepted: 22 June 2018 Published: 24 July 2018

### Citation:

Zeng T, Fan C, Wen Z, Li Q, Zhou Z, Han S and Liu J (2018) Effect of Environmental Temperature on the Content of Impurity Li3V2(PO4)3/C in LiVPO4F/C Cathode for Lithium-ion Batteries. Front. Chem. 6:283. doi: 10.3389/fchem.2018.00283 Previous studies have shown that the impurity Li3V2(PO4)<sup>3</sup> in LiVPO4F will adversely affect its electrochemical performance. In this work, we show that the crystalline composition of LiVPO4F/C is mainly influenced by the environmental temperature. The content of Li3V2(PO4)<sup>3</sup> formed in LiVPO4F/C is 0, 11.84 and 18.75% at environmental temperatures of 10, 20, and 30◦C, respectively. For the sample LVPF-30C, the SEM pattern shows a kind of alveolate microstructure and the result of selected area electron diffraction shows two sets of patterns. The LiVPO4F/C cathode without impurity phase Li3V2(PO4)<sup>3</sup> was prepared at 10◦C. The selected area electron diffraction result proves that the lattice pattern of LiVPO4F is a regular parallelogram. Electrochemical tests show that only one flat plateau around 4.2 V appears in the charge/discharge curve, and the reversible capacity is 140.4 mAh·g <sup>−</sup><sup>1</sup> at 0.1 C, and 116.3 mAh·g <sup>−</sup><sup>1</sup> at 5 C. From these analyses, it is reasonable to speculate that synthesizing LiVPO4F/C at a low environmental temperature is a practical strategy to obtain pure crystalline phase and good electrochemical performance.

Keywords: lithium-ion batteries, lithium vanadium fluorophosphates, environmental temperature, alveolate structure, electrochemical performance

## INTRODUCTION

The rechargeable lithium-ion battery has been widely studied because of its applications in electric vehicles, mobile phones, and energy storage devices (Huang et al., 2009; Konarov et al., 2017). LiFePO<sup>4</sup> delivers superior thermal stability and excellent cyclic performance, but a low working potential decreases its energy density (Yamada et al., 2003; Kim et al., 2015; Eftekhari, 2017; Wu et al., 2017).

A novel cathode lithium vanadium fluorophosphate (LiVPO4F) material has been reported (Gover et al., 2006). The working potential (4.2 V) of LiVPO4F is much higher than that of LiFePO<sup>4</sup> and LiCoO<sup>2</sup> (Ma et al., 2013a; Hu et al., 2014; Wu et al., 2016). Moreover, the thermal stability of LiVPO4F is better than that of LiFePO<sup>4</sup> and LiCoO<sup>2</sup> (Wang et al., 2014; Xu et al., 2015). If the shortcoming of electronic conductivity is solved, LiVPO4F will be an outstanding cathode material (Reddy et al., 2010; Ma et al., 2013b; Satish et al., 2016). Some improvements have been adjusted to LiVPO4F cathode, such as cation doped, carbon coated and various synthesized routes (Wang et al., 2013a; Liu et al., 2016; Wu et al., 2018). Recently, adopting facile and controllable methods to prepare LiVPO4F is the key areas of research. LiVPO4F was reported by two-step carbothermal reduction in some references. However, this method suffers from high energy consumption and a large content of carbon, because the intermediate VPO<sup>4</sup> is prepared separately at 700–800◦C (Ma et al., 2014; Liu et al., 2015; Wang et al., 2016).

Thus, a novel one-step method in which the synthesis of VPO<sup>4</sup> is omitted and carbon content is restricted to a very low level is of great research interest. Although the electrochemical performance of LiVPO4F prepared is improved, the plateaus belonging to impurity Li3V2(PO4)<sup>3</sup> come into being (Liu et al., 2012; Wang et al., 2013b; Xiao et al., 2013). Therefore, the formation of Li3V2(PO4)<sup>3</sup> is observed even though we use a synthesis method that employs a novel chemical reduction route. The content of Li3V2(PO4)<sup>3</sup> should be carefully controlled because it may adversely affect the performance of the LiVPO4F cathode.

In this work, we discovered that the formation of impurity Li3V2(PO4)<sup>3</sup> is directly related to the environmental temperature. The formation mechanism was investigated through further analysis of the structure and synthesis procedures.

### EXPERIMENTAL

### Materials Synthesis

LiVPO4F/C was synthesized by using a novel chemical reduction method. The chemical reagent used was of analytical reagent grade. 0.03 mol H2C2O<sup>4</sup> dissolved in deionized water was used as a chelating agent and reducing agent. 0.01 mol V2O<sup>5</sup> was added slowly under vigorous magnetic stirring at 60◦C. LiF and NH4H2PO<sup>4</sup> at the molar ratio of 1:1 to vanadium were introduced in after 10 min. A PVDF carbon source of 1.4943 g was dispersed in 30 ml water in a solution of hexadecyl trimethyl ammonium bromide under ultrasonic assistance at 50◦C. Subsequently, the PVDF suspension was added to the reaction system. Finally, the suspension was dried overnight in vacuum at 85◦C. The precursor was presintered at 400◦C for 5 h and sintered at 800◦C for 4 h in a tubular furnace with flowing high-purity argon.

### Characterization

The crystal structure of the material was examined by Xray diffraction (XRD, Rigaku D/MAX 2500). The morphology and elemental content were investigated with scanning electron microscopy (SEM, Navo NanoSEM230) and energy disperse spectroscopy (EDS). Nanoscale morphology and selected area electron diffraction (SAED) were performed by using highresolution transmission electron microscopy (HRTEM, JEOL-3010).

### Electrochemical Test

The electrochemical performance of LiVPO4F/C electrodes was evaluated using an Arbin BT2000 battery test system. The cathode film was fabricated by mixing LiVPO4F/C (80 wt.%), acetylene black (15 wt.%), and PVDF (5 wt.%) in the solvent Nmethyl pyrrolidone, and the slurry was coated on an aluminum collector. The electrodes were dried in a vacuum oven at 120◦C for 12 h and 2016 coin-type cells were assembled in a glove box (S1220/750). The electrolyte was 1.3 mol·L <sup>−</sup><sup>1</sup> LiPF<sup>6</sup> in a mixing solvent of ethylene carbonate, dimethyl carbonate, and ethyl methyl carbonate (1:1:1). A lithium foil and a polypropylene separator (Celgard 2400) were used as counter electrode and separator, respectively.

### RESULTS AND DISCUSSION

The electrochemical performance of triclinic LiVPO4F/C is partially determined by the content of impurity Li3V2(PO4)3/C in it. Our study revealed that LiVPO4F prepared at a high environmental temperature delivers poor performance. To investigate the reason for this, we synthesized LiVPO4F/C at different environmental temperatures (30, 20, and 10◦C), and named the respective samples as LVPF-30C, LVPF-20C, and LVPF-10C.

The XRD patterns of the samples are shown in **Figure 1A**. The main diffraction peaks correspond to a triclinic system with the space group of P-1, and can be indexed as the standard pattern of LiVPO4F (Barker et al., 2003; Huang et al., 2009). The absence of peaks corresponding to crystalline carbon proves that carbon is amorphous. No impurity peaks in LVPF-10C, which delivers the strongest peaks among the samples, was found. The refined cell parameters a, b, and c were 5.174, 5.308, and 7.509 Å, and the cell volume was 174.18 Å<sup>3</sup> . These results compare well with the classic results reported by Barker (Barker et al., 2005). However, the peaks at 20.69◦ , 23.53◦ , and 24.48◦ belonging to the impurity Li3V2(PO4)<sup>3</sup> (symbol # in **Figure 1A**) occur in the curves of LVPF-30C and LVPF-20C (Zhu et al., 2008). The percentages of Li3V2(PO4)<sup>3</sup> in LiVPO4F were estimated by refining the XRD patterns in **Figure 1B**. The content of Li3V2(PO4)<sup>3</sup> increased gradually from 0% (10◦C) to 11.84% (20◦C) and 18.75% (30◦C). Hence, our preliminary presumption is that low environmental temperature plays an important role in the preparation of pure LiVPO4F.

In **Figure 1C**, four flat plateaus (A, B, C, and D) appear in the discharge curves of LVPF-30C. The predominant plateau A around 4.2 V is attributed to LiVPO4F/C, and is in accordance with Barker's work (Barker et al., 2003), and B, C, and D are assigned to Li3V2(PO4)3/C. The specific capacities at 0.1 C and 5 C are 138.6 and 101.1 mAh·g −1 . However, when temperature drops to 20◦C (corresponding to LVPF-20C, **Figure 1D**), the plateaus of Li3V2(PO4)3/C are shorter than before, establishing the decreasing content of impurity. The specific capacity increases obviously, especially at 5 C (118.3 mAh·g −1 ). Further, only a plateau A at 4.2 V without other plateaus of impurity Li3V2(PO4)3/C is observed in LVPF-10C (**Figure 1E**). It is important to note that Li3V2(PO4)3/C disappear entirely. The specific capacities at 0.1 C 1 C and 5 C are 140.4 mAh·g −1 , 130.6 mAh·g −1 and 116.3 mAh·g −1 , which are very close to that of LVPF-20C in **Figure 1F**. LVPF-30C delivers the worst performance at a high current density. LVPF-20C with 11.84%

impurity Li3V2(PO4)3/C possesses the optimum capacity at a high current density. The reason is that Li3V2(PO4)3/C is a fast ion conductor and allows a fast transfer of lithium ions in the cathode. Nevertheless, an excess of the impurity Li3V2(PO4)3/C in LiVPO4F/C adversely affects the rate and the cycling capability.

In **Figure 2A**, the alveolate structure can be easily observed in LVPF-30C. The surface of most particles is broken. This structure is observed in the HRTEM image. The SAED pattern is made up of two sets of lattices with different characteristics (inset of **Figure 2B**). These parallelogram lattices are attributed to triclinic LiVPO4F (bottom) and monoclinic Li3V2(PO4)<sup>3</sup> (top). In **Figure 2C**, the EDS image in the alveolate field proves the existence of Li3V2(PO4)<sup>3</sup> distinctly because the content of fluorine is much lower than that of vanadium. **Figures 2D,E** show that vanadium is uniformly distributed on the surface of particles and a small quantity of fluorine is detected. This proves that impurity Li3V2(PO4)<sup>3</sup> without fluorine is formed in the alveolate zone.

There is no alveolate structure on the flawless surface of LVPF-10C (**Figure 2F**) and the lattice fringes can be clearly observed (**Figure 2G**). The pattern of SAED in the square frame is a typical parallelogram, and is similar to the bottom lattice in **Figure 2B**. This pattern is attributed to the typical crystalline form of LiVPO4F with the triclinic system. Thus, LVPF-10C possesses a good crystalline morphology with a thin layer covering on the surface of the crystalline LiVPO4F. Its lattice pattern is a series

of concentric circles, which is the characteristic of amorphous carbon (Song et al., 2008). The atomic contents of vanadium and fluorine are 8.61 and 8.30%, respectively, and match well with the atomic ratio of LiVPO4F in **Figure 2H**. Thus, we conclude that low temperature (10◦C) helps to prepare pure phase LiVPO4F.

The formation mechanism of the alveolate structure is investigated in **Figure 3**. On one hand, the excessive oxalic acid hydrolyzes in deionized water and produces hydrogen ions in aqueous solution. Ammonium dihydrogen phosphate generates ammonium ions in the hydrolysis reaction. A fluoride compound is formed when a hydrogen ion and an ammonium ion are combined with a fluoride ion released by LiF. Therefore, HF and NH4F are formed in the reaction (Zhou et al., 2009). It is well known that fluoride compounds are unstable and easily evaporate. From the viewpoint of reaction kinetics, the volatilization rate of fluoride will increase at least 6 to 8 times Zeng et al. Regulation of Li3V2(PO4)3/C in LiVPO4F/C

at 30◦C compared to 10◦C in reaction and drying. Therefore, the content of fluorine in the precursor at 30◦C is evidently lower than that at 10◦C. It can be inferred that the impurity Li3V2(PO4)<sup>3</sup> is formed in this condition. On the other hand, the temperature of the tubular furnace drops slowly at 30◦C. The cooling rate of LVPF-30C is lower than that of LVPF-10C. The longer cooling time of LVPF-30C accelerates the evaporation of fluoride, especially at 600–800◦C. Therefore, this ensures that the fluorine content in LVPF-30C is much less than the value determined. The impurity Li3V2(PO4)<sup>3</sup> is formed, which is in accordance with the above analysis of its structure and morphology.

Hence, the volatilization of fluoride should be inhibited in the preparation processes of LiVPO4F. Based on all of the evidence we have presented above, we legitimately conclude that a lower environmental temperature is more helpful to synthesize a LiVPO4F/C cathode with a low content of impurity and excellent electrochemical performance.

### CONCLUSIONS

A sample of LVPF-10C, which was prepared at an environmental temperature of 10◦C, exhibited a regular parallelogram space

### REFERENCES


pattern that is attributed to the pure triclinic form of LiVPO4F. High environmental temperature accelerates the volatilization of fluoride in the drying and sintering process and decreases the fluorine content. Then, a large quantity of Li3V2(PO4)<sup>3</sup> reduces the plateaus in the discharge curves and deteriorates the rate of performance in LVPF-30C. Therefore, our work is devoted to give a direction to improve the synthetic process and advise what we need to do in the future.

### AUTHOR CONTRIBUTIONS

TZ wrote the paper and designed the main part of the experiment. CF was the main advisor. ZW and QL carried out material preparation and the electrochemical test. ZZ discussed and refined the paper. TZ, CF, ZW, QL, and ZZ proposed the research. CF, SH, and JL obtained the main financial support for the research and supervised all the experiments.

### ACKNOWLEDGMENTS

This work was supported by the National Natural Science Foundation of China [51472082, 51672079, 51372079] and science and technology project of Changsha City [k1508010–11].


ion batteries. J. Power Sources 251, 325–330. doi: 10.1016/j.jpowsour.2013. 11.095


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Zeng, Fan, Wen, Li, Zhou, Han and Liu. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Ultrafine NaTi2(PO4)<sup>3</sup> Nanoparticles Encapsulated in N-CNFs as Ultra-Stable Electrode for Sodium Storage

Sicen Yu, Yi Wan, Chaoqun Shang, Zhenyu Wang, Liangjun Zhou, Jianli Zou, Hua Cheng\* and Zhouguang Lu\*

*Department of Materials Science and Engineering, Southern University of Science and Technology, Shenzhen, China*

### Edited by:

*Qiaobao Zhang, Xiamen University, China*

#### Reviewed by:

志聪 施*, Guangdong University of Technology, China Huixin Chen, Xiamen Institute of Rare Earth Materials, China Jianmin Ma, Hunan University, China*

#### \*Correspondence:

*Hua Cheng chengh@sustc.edu.cn Zhouguang Lu luzg@sustc.edu.cn*

### Specialty section:

*This article was submitted to Nanoscience, a section of the journal Frontiers in Chemistry*

Received: *15 May 2018* Accepted: *15 June 2018* Published: *06 July 2018*

#### Citation:

*Yu S, Wan Y, Shang C, Wang Z, Zhou L, Zou J, Cheng H and Lu Z (2018) Ultrafine NaTi2(PO4)3 Nanoparticles Encapsulated in N-CNFs as Ultra-Stable Electrode for Sodium Storage. Front. Chem. 6:270. doi: 10.3389/fchem.2018.00270* We present a feasible method for the preparation of one-dimensional N-doping carbon nanofibers encapsulated NaTi2(PO4)<sup>3</sup> (NTP-NCNFs) through electrospinning accompanied by calcination. The poor electrical conductivity of NTP is significantly improved and the as-prepared NTP-NCNFs exhibit stable and ultrafast sodium-storage capability. The NTP-NCNFs maintains a stable specific capacity of 121 mAh g−<sup>1</sup> at 10 C after 2,000 cycles, which only drop to 105 mAh g−<sup>1</sup> after 20,000 cycles. Furthermore, the NTP-NCNFs show excellent rate performance from 0.2 to 20 C, whose recovery efficiency still reaches 99.43%. The superior electrochemical property is mainly attributed to the large specific surface area, high porosity, N-doping carbon coating, and one-dimensional structure of NTP-NCNFs.

Keywords: NTP, N-doping carbon nanofibers, cycling performance, electrospinning, sodium-storage

### INTRODUCTION

Because of the high energy density, stable cycling performance, and environmental benignity, Liion batteries have been widely applied in portable devices and electric vehicles (Duncan et al., 2016; Kim et al., 2016; Zhang et al., 2018; Zheng et al., 2018). However, the limited lithium mineral reserves restrict the wide application of LIBs in grid-scale energy storage system. As cost-effective alternatives to Li-ion batteries, sodium ion batteries have been investigated for next-generation energy storage system, benefiting from sodium abundance (Wu et al., 2016; Wang et al., 2017, 2018; Xu et al., 2017). SIBs basically have similar battery components and electrical storage mechanisms as LIBs. However, the poor electrochemical kinetics and large volume change caused by large size of Na ions leads to severe capacity loss and cycling degradation. Therefore, the challenge of looking for sodium storage materials with good stability and high-rate capacity still remains.

Recently, NTP as one of sodium super-ionic conductor (NASICON) has been demonstrated as potential long life-time and high-rate electrode material for SIBs (Guin and Tietz, 2015; Wu et al., 2015; Hu et al., 2018; Liang L. et al., 2018). The strong P-O covalent bond in the phosphates offers remarkable structural and thermal stability and the open three-dimensional (3D) framework in NTP allows for fast sodium ions transfer(Li et al., 2012; Yang et al., 2015; Wang et al., 2016). However, pristine NTP with low intrinsic electronic conductivity displays poor electrochemical performance. In order to address this issue, some strategies including nano-sizing the particle, coating a conductive layer on the surface, and mixing with high conductive materials have been proposed (Fang et al., 2016; Ha-Kyung et al., 2016; Liang et al., 2018b). Although significant enhancement has been achieved, the satisfactory electrochemical performance with high rate capability and stability of NTP is still of great urgent.

Electrospinning is a fascinating way to prepare CNFs (Li et al., 2017; Zhu et al., 2017; Liang et al., 2018a). Meanwhile, due to the presence of carbon and nitrogen source from the starting polymer, electrospinning can be adapted to a feasible preparation for N-doping carbon matrixes encapsulated NTP nanoparticles to realize high performance for sodium storage. Herein, NTP nanoparticles are embeded into conductive N-doping carbon nanofibers (denoted as NTP-NCNFs). 1D nanofibers provide fast charge transfer pathway, ensuring the NTP-NCNFs with superb rate performance. The NCNFs matrix also contributes to the ultra-long cycling stability. At 10 C rate, the NTP-NCNFs maintained a specific capacity of 105 mAh g−<sup>1</sup> after 20,000 cycles.

### EXPERIMENTAL SECTION

### Preparation of NTP-NCNFs

NaH2PO4·2H2O (0.211 g) and polyacrylonitrile (PAN) (0.8 g) were dissolved into N, N-dimethylformamide (DMF) (10 ml) to obtain a homogeneous solution. Then, titanium tetraisopropanolate [(CH3CH3CHO)3Ti] (0.5 ml) was dropped into above solution and stirred overnight. And then, the solution was injected into the syringe with a 21 G needle, which the flow rate was 10 uL min−<sup>1</sup> . Al foil was employed as the collector with distance to the needle of 15 cm and voltage of 15 kV. The as-electrospun fibers were carbonized in a tube furnace at 700◦C (denoted as 700NTP-NCNFs), 800◦C (denoted as 800NTP-NCNFs), 900◦C (TiN) for 2 h under inert atmosphere. As a fair comparison, the NaTi2(PO4)<sup>3</sup> powder was synthesized via mixing the same ratio of NaH2PO4·2H2O and [(CH3CH3CHO)3Ti] annealed at 800◦C under argon atmosphere.

### Materials Characterization

The crystal structure of the fibers was characterized using powder X-ray diffraction (XRD) on a Rigaku D/Max-2400 Xray diffractometer with Cu-Kα radiation (λ = 1.54056 Å). The specific surface area and the pore size distribution of as-prepared N-doping carbon coating NTP nanofibers (700 and 800◦C) were evaluated by the Brunauer-Emmet-Teller (BET) at 77 K using a NOVA 1200e Surface Area. Raman spectra of samples were acquired with a Lab RAM HR 800 Raman microscope with an excitation laser beam (λ = 532 nm). SEM images were obtained on a scanning electron microscope (Hitachi, S4800) attached with an energy-dispersive X-ray spectroscopy (EDS) facility. TEM and HR-TEM images were recorded on a JEOL JEM-2010 (JEOL Ltd, Tokyo, Japan) at 200 kV. Ex-situ XPS records were performed on a VG scientific ESCALAB 250 spectrometer.

### Electrochemical Analysis

The electrochemical properties of the N-doping carbon coating NTP nanofibers was tested by assembling 2016 coin-type cells with Na as the counter electrode. NTP-NCNFs, Super P, and polyvinylidene fluoride (PVDF) binders dissolved in Nmethylpyrrolidone (NMP) were mixed into slurry with a weight ratio of 8:1:1, which was coated on a Cu foil and dried in a vacuum oven at 110◦C for 6 h, further dividing into wafers with 12 mm diameter. The active material is about 0.64 mg cm−<sup>2</sup> . The separator was a glass fiber filter. The electrolyte was 1 M NaClO4/ethylene carbonate (EC) and propylene carbonate (PC) with volume ratio of 1:1. The specific capacity is based on the whole mass of NTP-NCNFs. Electrochemical capacity measurements of the NTP-NCNFs were tested on the Neware battery test system by applying galvanostatic charge-discharge. Cyclic voltammetry (CV) was performed on a BioLogic-VMP3 electrochemical workstation at the same voltage window with a sweep rate of 0.1 mV s−<sup>1</sup> . Electrochemical impedance spectroscopy (EIS) was recorded at an AC voltage of 5 mV amplitude in the frequency range from 1.0 to 100 mHz at room temperature.

### RESULTS AND DISCUSSION

**Figure 1** depicts the synthesis of NTP-NCNFs. It should be noted that (CH3CH3CHO)3Ti was added into the mixture solution under vigorous stirring followed by electrospinning. **Figures 2a,b**

mapping images: (d) C, (e) O, (f) P, (g) Ti, and (h) Na.

show the continuous nanofibers with diameter of ∼ 500 nm. The as-prepared nanofibers will be subjected to a calcination process under argon atmosphere at 800◦C, during which the PAN will convert into N-doped carbon materials while keeping its 1D morphology (Wang et al., 2013). The result of thermogravimetric analysis shows that the mass ratio between NaTi2(PO4)<sup>3</sup> and N-doped carbon materials is about 1:1 (Figure S1). Meanwhile, the NaH2PO<sup>4</sup> and (CH3CH3CHO)3Ti form NTP nanoparticles via a solid reaction process (Ribero et al., 2016; Wang et al., 2016). As shown in **Figures 2c,d**, the surface of the fiber become rough and evenly decorated by nanoparticles after calcination. However, after pyrolysis at 900◦C in argon atmosphere, the as-prepared sample has changed to TiN (JCPDF#65-0715) as demonstrated by the XRD analysis (Figure S2). This might be caused by the reduction of NTP at higher pyrolysis temperature by ammonia, which was formed by the decomposition of PAN (Liu et al., 2013). As-prepared TiN still keeps as nanofibers, shown in Figure S3. Though it is beneficial to the electrical conductivity of NTP-NCNFs, the further pyrolysis of carbon layer leads to a bad influence on its electrochemical performance, which was exhibited in Figure S4. Therefore, the further research mainly focuses on 700/800NTP-NCNFs.

High resolution TEM (HRTEM) measurements further confirmed the formation of NTP. **Figure 3a** shows that NTP nanoparticles evenly distributed over the carbon fiber. HRTEM (**Figure 3b**) reveals the lattice fringes of 0.21 nm, matching with the spacing of (119) plane of NTP. The elemental mapping using EDS coupled with HAADF/STEM was utilized to investigate the compositional distributions of C, O, P, Ti, and Na in 800NTP-NCNFs (**Figures 3c–h**), which clearly shows that these elements are uniformly distributed in 800NTP-NCNFs. The uniform structure of N-doping carbon matrix should improve the electrical conductivity of NTP.

**Figure 4A** shows the XRD patterns of NTP-NCNFs, in which all the peaks can be indexed to the standard NTP peaks (JCPDF#33-1296). To clarify the degrees of defect structure and graphitization, Raman spectroscopy is carried out. As shown in **Figure 4B**, the D (1,350 cm−<sup>1</sup> ) and G (1,600 cm−<sup>1</sup> ) bands are significantly observed, which represents disordered and graphitization, respectively. The ID/I<sup>G</sup> value of 800NTP-NCNFs is 1.05, which is evidently lower than that of 700NTP-NCNFs

(1.14) (Wang et al., 2015). The significantly decreased value of ID/I<sup>G</sup> indicates an enhanced degree of graphitization, which is likely beneficial to fast electron transport of NTP-NCNFs.

N<sup>2</sup> adsorption and desorption isotherms (**Figure 5A**) of both 700NTP-NCNFs and 800NTP-NCNFs can be identified as type IV isotherms, with pronounced hysteresis loops, implying the mesopores in NTP-NCNFs. The BET surface area of 800NTP-NCNFs is 153.86 m<sup>2</sup> g −1 , which is much higher than that of 700NTP-NCNFs (∼10 m<sup>2</sup> g −1 ). The high BET surface area of 800NTP-NCNFs is likely due to the higher carbonization degree and the release of gas generated at higher pyrolysis temperature. The pore size distributions are shown in **Figure 5B**, which suggest the formation of micro-, meso-, and macro-pores. The larger specific surface area and hierarchical pore structures can effectively increase the mass transport and the contact between NTP and electrolyte, which would be beneficial to the electrochemical performance of the 800NTP-NCNFs. Therefore, we believe that 800NTP-NCNFs will have better electrochemical performance than 700NTP-NCNFs (Figure S5).

We first use cyclic voltammetry (CV) to investigate the sodium storage mechanism of 800NTP-NCNFs (**Figure 6A**). During the initial cathodic scan, the peaks at 0.81 and 0.51 V are higher than those of subsequent cycles, which are caused by the formation of solid electrolyte interface layer (SEI). Remarkably, nearly all the shape and the position of the peaks are overlap after the second CV curve, suggesting excellent electrochemical reversibility and stability of 800NTP-NCNFs for sodium storage.

The discharge-charge profiles of 800NTP-NCNFs are recorded at various rates (**Figure 6C**). The plateaus at ∼2.1

and ∼0.32 V during discharge and plateaus at ∼0.5 and ∼2.2 V during charge are in agreement with the CV results. Here, the NCNFs contribute to Na<sup>+</sup> storage and the overall specific capacity is based on the whole mass of 800NTP-NCNFs (Stevens and Dahna, 2000; Dahbi et al., 2014).

**Figure 6D** shows the rate performance of 800NTP-NCNFs at different current densities. The reversible specific capacity was 176 mAh g−<sup>1</sup> at 0.1 C. As the rate increased from 0.2 to 0.5, 1, 2, 5, 10, and 20 C, the corresponding specific capacity was 163, 149, 138, 127, 110, 95, and 71 mAh g−<sup>1</sup> , with a capacity retention ration of 92.61, 84.66, 78.41, 72.16, 62.50, 53.98, and 40.34% as reference to that at 0.1 C, respectively. As the rate returned to 0.1 C, the specific capacity of 800NTP-NCNFs recovered to 175 mAh g−<sup>1</sup> , which demonstrated the excellent reversibility of 800NTP-NCNFs during Na<sup>+</sup> insertion and extraction. As illustrated in **Figure 6E**, the 800NTP-NCNFs deliver ultra-stable cycling performance and rate capability for 20,000 cycles at 10 C. The 800NTP-NCNFs maintain a stable specific capacity of 121 mAh g−<sup>1</sup> after 2,000 cycles, which is about 80% of initial capacity. After 20,000 cycles, the 800NTP-NCNFs even deliver a considerable capacity of 105 mAh g−<sup>1</sup> . The astonishing electrochemical property of 800NTP-NCNFs is attributed to several features: (i) NASICON-structured NTP ensures sufficient ion transport; (ii) N-doping carbon coating on the surface of NTP improves the electronic conductivity of NTP, which provides fast electronic transport; (iii) 1D structure with large surface area ensures rapid ion transport by enhancing electrode and electrolyte contact; (iv) the hierarchical pore distribution is favorable for the electrolyte penetration and accommodation of volume change.

The 800NTP-NCNFs displayed superior cycling stability with high specific capacity to that of pristine NTP powder (**Figure 6B**), which further demonstrated that 800NTP-NCNFs intrinsically improved the electronic conductivity of NTP. To investigate the improvement of 800NTP-NCNFs, EIS was performed in **Figure 7**. The charge-transfer resistance (Rct) for pristine NTP powder electrode is 1100 , and this high resistance reflects the low intrinsic electrical conductivity of NTP. The Rct of 800NTP-NCNFs is significantly lower than that of pristine NTP powder electrode (around 150 ), which implies that NTP encapsulated in N-doping carbon fiber structure successfully enhance the electrical conductivity and, in turn, the electrochemical performance of NTP (**Figure 6B**).

### CONCLUSION

In summary, we presented a feasible method to prepare NTP-NCNFs which exhibits excellent rate capability and stable cycling performance for sodium storage. The NTP-NCNFs could deliver a specific capacity of 105 mAh g−<sup>1</sup> under 10 C even after 20000 cycles. The nanosized NTP shorten the solid-state ion diffusion length and accelerate surface electrochemical reaction in the electrode. The 1D N-doped carbon coating enhances electronic conductivity of NTP and ensures the fast electron transfer. Moreover, the 3D woven network is beneficial to the penetration of electrolyte and accommodates the volume change of NTP during cycling.

### ASSOCIATED CONTENT

### Supporting Information

TGA curve of 800NTP-NCNFs, XRD patterns of TiN, SEM and TEM images of TiN, and electrochemical performance of TiN and 700 NTP-NCNFs including CV curves for the initial 4 cycles and the cycling performance at a current density of 200 mA g−<sup>1</sup> (1 C) and corresponding coulombic efficiency.

### AUTHOR CONTRIBUTIONS

ZL and CS designed the research. CS, SY, YW, ZW, and LZ performed the experiments. JZ, HC, ZL, and CS incorporated in the interpretation of experimental results. All authors participated in the general discussion.

### ACKNOWLEDGMENTS

This work is finically supported by the National Natural Science Foundation of China (No. 21671096, and No. 21603094), Natural Science Foundation of Guangdong Province (No. 2016A030310376), the Guangdong Special Support for the Science and Technology Leading Young Scientist (No. 2016TQ03C919), and the Basic Research Project of the Science and Technology Innovation Commission of Shenzhen (No. JCYJ20170412153139454 and No. JCYJ201708171102 51498).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2018.00270/full#supplementary-material

### REFERENCES


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Yu, Wan, Shang, Wang, Zhou, Zou, Cheng and Lu. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# High-Performance Lithium-Sulfur Batteries With an IPA/AC Modified Separator

Yafang Guo1,2, Aihua Jiang<sup>1</sup> , Zengren Tao<sup>1</sup> , Zhiyun Yang<sup>1</sup> , Yaping Zeng<sup>1</sup> \* and Jianrong Xiao1,2 \*

*<sup>1</sup> College of Science, Guilin University of Technology, Guilin, China, <sup>2</sup> Guangxi Key Laboratory of Electrochemical and Magnetochemical Functional Materials, Guilin University of Technology, Guilin, China*

To inhibit the polysulfide-diffusion in lithium sulfur (Li-S) batteries and improve the electrochemical properties, the commercial polypropylene (PP) was decorated by an active carbon (AC) coating with lots of electronegative oxygenic functional group of –OH. Owing to the strong adsorption of AC and the electrostatic repulsion between the –OH and negatively charged polysulfide ions, the Li-S batteries demonstrated a high initial discharge capacity of 1,656 mAh g−<sup>1</sup> (approximately 99% utilization of sulfur) and the capacity can still remain at 830 mAh g−<sup>1</sup> after 100 cycles at 0.2 C. Moreover, when the rate was increased to 1 C, the batteries could also possess a discharge capacity of 1,143 mAh g−<sup>1</sup> . The encouraging cycling stability make clear that this facile approach can successfully restrain the shuttle effect of polysulfides and make further progress to the practical application of Li-S batteries.

Keywords: separator, active carbon, isopropyl alcohol, polysulfide adsorption, lithium-sulfur battery

### INTRODUCTION

In order to meet the ever increasing demand for high-capacity, long cycle life and stable rechargeable batteries, more and more electrochemical workers are starting to pay attention to lithium sulfur batteries, which possess a high theoretical capacity (1,675 mAh g−<sup>1</sup> ) and high specific energy (2,600 Wh kg−<sup>1</sup> ) (Zhang et al., 2015; Zhou et al., 2015). Compared with the conventional Liion battery, Li-S cell displays more advantages such as cost-effective, rich reserve and environmentfriendly (Zu and Manthiram, 2013; Wang et al., 2014, 2016a,b; Gong et al., 2016; Zhang et al., 2018). Nevertheless, some intrinsic properties still hindered the massive implementation of Li-S cells: (1) poor electric and ionic conductivity of S<sup>8</sup> and its final reaction products (Li2S2/Li2S), (2) severe diffusion of the polysulfide intermediates (Li2Sx, 4≤ × ≤8), (3) low electrochemical utilization of the active materials (Cai et al., 2015; Lai et al., 2015; Wang et al., 2015; Zhu et al., 2016).

Tremendous efforts have been devoted to solve these scientific issues in Li-S cells by holding sulfur in various composites with special structures or exploiting new electrolytes. Although significant progress has been made in the utilization of elemental sulfur and the cyclic stability, the synthetic methods are usually relatively complex, which not only need a variety of additives but also have a higher requirements for the manufacturing processes (Xiong et al., 2012; Huang et al., 2013; Wang et al., 2013; Hu et al., 2014; Zhang et al., 2014; Deng et al., 2015; Lee et al., 2015; Liu et al., 2016; Lu et al., 2016; Nersisyan et al., 2016; Yang et al., 2016). Alternatively, modifying the commercial separators have been proved to be a facile and commendable strategy to improve

#### Edited by:

*Qiaobao Zhang, Xiamen University, China*

#### Reviewed by:

*Baihua Qu, Xiamen University, China Xiwen Wang, Hunan University, China Liuqing Yang, National University of Singapore, Singapore*

#### \*Correspondence:

*Yaping Zeng yapingz@126.com Jianrong Xiao xjr@glut.edu.cn*

#### Specialty section:

*This article was submitted to Physical Chemistry and Chemical Physics, a section of the journal Frontiers in Chemistry*

> Received: *14 April 2018* Accepted: *28 May 2018* Published: *14 June 2018*

#### Citation:

*Guo Y, Jiang A, Tao Z, Yang Z, Zeng Y and Xiao J (2018) High-Performance Lithium-Sulfur Batteries With an IPA/AC Modified Separator. Front. Chem. 6:222. doi: 10.3389/fchem.2018.00222*

**342**

the electrochemical performance through effective regulation of polysulfide shuttle (Chung and Manthiram, 2014a,b; Li G. C. et al., 2015; Balach et al., 2016; Conder et al., 2016; Fan et al., 2016). Especially the introduction of functional groups on the surface of separator is gradually studied. For instance, Yu X et al. modified the separator with carboxyl functional group through a sequence of hydroxylating, grafting and hydrolyzing processes to bring about a negatively charged environment in Li-S cells (Yu et al., 2016). In order to constrain the diffusion of electronegative polysulfides, Li Z et al. introduced oxygenic functional groups (-OH, -COOH) onto the surface of separator by O<sup>2</sup> plasma treatment (Li Z. et al., 2015). Similarly, a method of one-step plasma-induced graft co-polymerisation was used to develop negatively charged –SO<sup>−</sup> 3 onto the microporous membrane and this separator showed a good ability to inhibit the shuttle effect (Conder et al., 2015).

In this study, we present a facile approach to achieve a high-performance active carbon coated separator with hydroxyl groups, which can perform excellent physical adsorption and electrostatic exclusion at the same time, bringing about a strong inhibition of soluble electronegative polysulfides. In comparison to the batteries assembled using pristine PP separator, the Li-S batteries with modified separator exhibit significantly enhanced cyclic stability and rate capability.

### EXPERIMENTAL SECTION

### Materials Preparation

First, 1.0 g active carbon was added to 30 mL isopropyl alcohol (IPA) and magnetic stirred for 24 h to permeate IPA into the pores of active carbon. Then the prepared solution was dried at 60◦C for 6 h to obtain IPA/AC composite material.

A slurry method was used to coat the PP (Celgard 2400) separator with the IPA/AC composite. A mixture of IPA/AC composite and polyvinylidene fluoride (PVDF) (8:1, by mass) was placed in N-methy-2-pyrolidone (NMP) to form slurry, which was subsequently coated on the cathode side of the pristine PP separator. The IPA/AC modified separator was then dried in vacuum oven at 60◦C for 4 h. In addition, AC modified separator was prepared in the same way for comparison.

### Material Characterization

The morphology was characterized by a field emission scanning electron microscopy (SEM, HTTAHIS-4800). Energy dispersive spectrometer (EDS) was employed to identify the distribution of the elements on the surface of the IPA/AC-coated separators. The chemical state of the carbon and oxygen in samples were tested with X-ray photoelectron spectroscopy (XPS, ESCA LAB 250Xi).

### Battery Assembly and Electrochemical Measurement

A solid solution method was used to fabricate the active composite materials with a mixture of S<sup>8</sup> and AC (7:3, by mass). Sulfur cathode was made of as-prepared S/AC composite, acetylene black and PVDF dissolved in NMP with a mass ratio of 7:2:1. The obtained homogeneous slurry was coated onto aluminum foil with a doctor blade, followed by drying in a vacuum oven at 60◦C for 12 h. The active substance sulfur loading was about 3.17 mg cm−<sup>2</sup> .

CR-2025-type button cells were assembled in an argon-filled glove box with pristine separators, AC-coated separators and IPA/AC modified separators for comparison. Lithium metal was used as the counter electrode. The electrolyte consisted of 1.0 wt% LiNO<sup>3</sup> and 1.0 M LiTFSI in a mixed solvent of DME and DOL at the volume ratio of 1:1.

Cyclic voltammetry (CV) were measured by a CHI750E electrochemical workstation at a scan rate of 0.1 mV s−<sup>1</sup> within the voltage range of 1.5–3.0 V. Electrochemical impedance spectroscopy (EIS) of the cells was carried in the frequency range of 10 mHz−100 kHz with a perturbation amplitude of 5 mV. In addition, galvanostatic charge-discharge tests and rate capability were conducted to evaluate the cycle stability of Li-S cells on the basic of S<sup>8</sup> at different current densities under LAND test instrument.

## RESULTS AND DISCUSSION

### Characterization of IPA/AC Modified Separators and Pristine Separators

The shuttling effect inhibition principle in Li-S cell is showed in **Scheme 1**. In the battery with pristine separator, the polysulfide ions of Sn2<sup>−</sup> (4≤n≤8) can freely shuttle back and forth between the two poles. But the shuttle effect can be effectively suppressed in the battery with IPA/AC modified separator. From **Scheme 1B** we can see the IPA/AC coating is on one side of the bare separator, facing the sulfur cathode and act as a surface barrier. This barrier contains porous active carbon with strong adsorption and electronegtive oxygenic functional group of –OH, which can simultaneously take advantage of physical adsorption and electrostatic repulsion to prevent the diffusion of polysulfide ions to the lithium anode.

To investigate the functional groups that exist in IPA/AC composites, we performed the FTIR characterizations on the samples. **Figure 1** gives the FTIR spectrum of AC and IPA/AC. The wide peak in the FTIR spectrum of IPA/AC at 1,150 cm−<sup>1</sup> is the characteristic peak of –OH, by comparing the intensity and width of the peaks, we can confirm that hydroxyl groups are successfully introduced into the IPA/AC composites.

The content of C1s and O1s in active carbon materials and IPA/AC composites were measured by X-ray photoelectron spectroscopy (XPS). **Figures 2A,B** show the intensity of C1s and O1s in IPA/AC (9 and 3.8, respectively) are obviously higher than those in AC (8 and 3, respectively), indicating that IPA/AC contains more C1s and O1s. High-resolution O1s XPS spectra of IPA/AC is shown in **Figure 2D**, three peaks can be easily identified at the binding energy 531.71, 533.08, and 533.17 eV, corresponding to O = C-O, C-OH and O = C-O groups, respectively (López et al., 1991; Stevens et al., 2014), which are similar to the spectrum of AC in **Figure 2C**. To determine how much C-OH was introduced, we calculated the percentage of its peak area. The result shows the C-OH in IPA/AC is 40%, higher than that of AC (30%). This consequence reveals that

hydroxyl group was successfully introduced in to the active carbon particles.

SEM was used to examine the morphology of the original separator and IPA/AC modified separator, as given in **Figures 3a,b**. A smooth surface with uniformly substantial slit pores structure is presented for routine separator, which promotes ion conduction but restricts the transportation of electron (Gong et al., 2009). In comparison with the bare separator, the surface of IPA/AC modified separator is covered with micrometer active carbon particles. These particles have a large specific surface area and superior conductivity, which can not only provide rich attachment points for polysulfide ions but also contribute to reducing the internal resistance of Li-S battery.

**Figure 3c** reveals the cross-section of IPA/AC modified separator. From the image we know the IPA/AC coating is

about 12.5µm and in good contact with the Celgard separator. **Figure 3d** shows the electrolyte affinity test of the routine separator and IPA/AC modified separator. As seen, the droplet is not dispersed on the routine separator, whereas the IPA/AC modified separator wetted a large area. It can be predicted that the electrochemical performance of Li-S battery with IPA/AC modified separator would be significantly improved, during to an increased rate of ion transmission.

To ascertain the utility of IPA/AC coating, the morphological changes before and after cycling were observed, as summarized in **Figures 4a,b**. Before cycling, the surface of IPA/AC particles is relatively smooth (**Figure 4a**), while after 100 cycles the surface turns rough with clumps of different sizes, which indicates the effective physical adsorption and electrostatic repulsion of free dissolved polysulfides in cathode region (**Figure 4b**). For further supporting this conclusion, the contrast of elemental mappings before and after 100 cycles at 0.2C is conducted by energy dispersive spectrometer (EDS). As given in **Figures 4c,d**, only very weak elemental signals of O, F and S could be detected on the IPA/AC coating before cycling (**Figure 4c**), and these faint element signals may come from the impurities in active carbon materials. After cycles these signals became obvious and distributed evenly, which can be attributed to the fully infiltration of the electrolyte and the mass attachment of the polysulfides as well (**Figure 4d**).

### Electrochemical Performance of Batteries With IPA/AC Modified Separator

The function of IPA/AC modified separator on electrochemical performance is investigated based on CR-2025-type button cells. Cyclic performance at different discharge current rate of Li-S batteries using IPA/AC modified separator, AC-coated separator and original separator for comparison are presented in **Figures 5A–D**. As anticipated, the cell with IPA/AC modified separator reveals a significant enhancement at each current rate.

Wetting behavior of electrolyte on the pristine separator and IPA/AC modified separator.

The initial discharge capacities of the cells with IPA/AC modified separators are 1,656, 1,246, 1,190 and 1,143 mAh g−<sup>1</sup> at 0.2C, 0.3C, 0.5C, and 1C, respectively, which are much higher than those of the other two cells at these current rate, as given in **Table 1**. In particular, after 50 cycles at high current rate of 1C, the capacity of Li-S cell with IPA/AC modified separator can maintain at 584 mAh g−<sup>1</sup> , and the Coulombic efficiency is above 98%. Therefore, we can conclude that the reaction intermediates are largely trapped within the IPA/AC coating, which effectively reduces the irreversible loss of active substances.

**Figure 5E** presents the rate performance of three Li-S batteries at a vary current rate of 0.2C→0.3C→0.5C→1C→0.5C→0.3C→0.2C. From this chart we can see the cell with IPA/AC modified separator delivered the highest initial discharge capacity of 1,650 mAh g−<sup>1</sup> at 0.2C, demonstrating the high utilization of sulfur which can be ascribed to the easy penetration of electrolyte and the significant blocking effect of the IPA/AC coating. When increased the current rate to 1C, the capacity of this battery is as high as 927 mAh g−<sup>1</sup> , while only 781 and 427 mAh g−<sup>1</sup> could be obtained from the cells with AC-coated separator and bare separator, respectively. In addition, after 35 cycles, the capacity of IPA/AC sample still retained at 1066 mAh g−<sup>1</sup> (approximately 65% of the initial reversible capacity), attesting to the efficient electrostatic repulsion between the –OH and negatively charged polysulfide ions, leading to a excellent rate performance of Li-S cell.

The initial discharge profiles of the Li-S batteries using IPA/AC modified separator, AC-coated separator and pristine separator at 0.2 C are exhibited in **Figure 6A**. It is found that each profiles consists of two typical discharge potential plateaus corresponding to the reduction from elemental sulfur to long-chain polysulfides at high voltages and from long-chain polysulfides to short-chain Li2S2/Li2S (Jianrong et al., 2014; Guo et al., 2017). However, there are distinctly differences in the height and length of the plateau. Apparently, the IPA/AC separator battery possesses the highest and longest voltage

platform, revealing the considerable utilization of active materials along with a thorough chemical reaction. **Figure 6B** gives the initial cyclic voltammetry curves (CV) of the three batteries at a scanning rate of 0.1 mV s−<sup>1</sup> . One anodic peak and two cathodic peaks can be discerned from these CV curves. And the positions of these two cathodic are consistent with the discharge potential plateaus of discharge profiles (Xiao et al., 2015). Moreover, one thing we should pay attention is that in IPA/AC battery the position of cathodic peaks are higher than those of the other two batteries, while the position of anodic peak is lower, indicating that the IPA/AC coating can not only reduce the charge voltage but also make a great improvement on the discharge depth, displaying the admirable transport kinetic of ions and electrons.

The enhanced electrochemical performance is further verified by the electrochemical impedance spectrum (EIS) measurement within a frequency range of 10 mHz−100 kHz and the equivalent


TABLE 1 | The cycling performance of Li-S cells with IPA/AC modified separators, AC-coated separators and pristine separators at different current rates (mAh g−<sup>1</sup> ).

FIGURE 6 | (A)Initial discharge profiles of Li-S cells with pristine separator, AC-coated separator and IPA/AC modified separator at 0.2 C. (B) Cyclic voltammetry of Li-S cells with pristine separator, AC-coated separator and IPA/AC modified separator at a 0.1 mV s−<sup>1</sup> scanning rate.

circuit is acquired by Z-view software. In the equivalent circuit, R<sup>1</sup> denotes the resistance of the electrolyte, R<sup>2</sup> is the charge transfer resistance, CPE<sup>1</sup> represents the constant-phase elements, and W<sup>1</sup> is the Warburg diffusion impedance (**Figure 7** inset) (Hou et al., 2017). From **Figure 7**, a semicircle can be saw at high and medium frequency, representing the charge transfer resistance (Rct) (Li G. C. et al., 2015). From the diameter of the semicircle we know the cells with AC-coated separator and pristine separator have a larger Rct than IPA/AC sample both before and after cycles, demonstrating the reduction of the charge transfer resistance by the special function of the IPA/AC coating,

which can act as a surface collector to reserve enough electrolyte and accelerate the diffusion of lithium ion.

In order to visual observe the retention of polysulfide species by the introduced IPA/AC separator, we conducted the polysulfide diffusion test for the three separator samples, as shown in **Figure 8**. The polysulfide solution in glass tubes was made by adding 7.77 mg S<sup>8</sup> and 2.23 mg Li2S into 5 ml DME:DOL (1:1,v:v), and solution in beakers was 4 ml DME:DOL (1:1,v :v). As expected, the pristine separator does not suppress the diffusion of polysulfides, thus the color of DME:DOL solution already changed to yellow after 5 min of rest. In contrast, the

after 100 cycles at 0.2 C.

IPA/AC separator largely suppressed the diffusion of polysulfide species, therefore even after 30 min of rest only a little change in color was observed, indicating that the porous active carbon coating with hydroxyl groups has good retention capability of polysulfide, which is attributed to the physical adsorption of porous carbon structure and the electrostatic repulsion between the hydroxyl and negatively charged polysulfide ions.

### CONCLUSION

In conclusion, the IPA/AC modified separator successfully integrates the strong physical adsorption of active carbon and the electrostatic repulsion between the hydroxyl and negatively charged polysulfide ions to obstruct the shuttle effect in the Li-S battery. With this special coating, the battery can present a apparent improvement on the storage of electrolyte, ion conduction and the utilization of active substances, leading

### REFERENCES


to a stable cycle ability and excellent rate performance. In addition, the modification method only acquires active carbon and isopropyl alcohol, which is environment-friendly and easy to operate, indicating that the IPA/AC modified separator provides a great potential in the commercial production of lithium sulfur battery.

### AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

### ACKNOWLEDGMENTS

This work was supported by Guangxi Key Laboratory of Electrochemical and Magnetochemical Functional Materials Open Foundation (No. EMFM20182203).


of lithium–sulfur batteries. J. Mater. Sci. Mater. Electr. 28, 17453–17460. doi: 10.1007/s10854-017-7679-7


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The reviewer, BQ, and handling Editor declared their shared affiliation.

Copyright © 2018 Guo, Jiang, Tao, Yang, Zeng and Xiao. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

## Synthesis and Electrochemical Performance of Molybdenum Disulfide-Reduced Graphene Oxide-Polyaniline Ternary Composites for Supercapacitors

Li-Zhong Bai\*, Yan-Hui Wang, Shuai-Shuai Cheng, Fang Li, Zhi-Yi Zhang and Ya-Qing Liu\*

Shanxi Province Key Laboratory of Functional Nanocomposite Materials, North University of China, Taiyuan, China

Edited by: Qiaobao Zhang, Xiamen University, China

### Reviewed by:

Shuge Dai, Zhengzhou University, China Linfeng Hu, Fudan University, China

#### \*Correspondence:

Li-Zhong Bai lzbai@nuc.edu.cn Ya-Qing Liu lyq@nuc.edu.cn

#### Specialty section:

This article was submitted to Nanoscience, a section of the journal Frontiers in Chemistry

Received: 08 May 2018 Accepted: 25 May 2018 Published: 12 June 2018

#### Citation:

Bai L-Z, Wang Y-H, Cheng S-S, Li F, Zhang Z-Y and Liu Y-Q (2018) Synthesis and Electrochemical Performance of Molybdenum Disulfide-Reduced Graphene Oxide-Polyaniline Ternary Composites for Supercapacitors. Front. Chem. 6:218. doi: 10.3389/fchem.2018.00218 Molybdenum disulfide/reduced graphene oxide/polyaniline ternary composites (MoS2/rGO/PANI) were designed and synthesized by a facile two-step approach including hydrothermal and in situ polymerization process. The MoS2/rGO/PANI composites presented an interconnected 3D network architecture, in which PANI uniformly coated the outer surface of the MoS2/rGO binary composite. The MoS2/rGO/PANI composites with a weight percent of 80% (MGP-80) exhibits the best specific capacitance (570 F g−<sup>1</sup> at 1 A g−<sup>1</sup> ) and cycling stabilities (78.6% retained capacitance after 500 cycles at 1 A g−<sup>1</sup> ). The excellent electrochemical capacitive performance is attributed to its 3D network structure and the synergistic effects among the three components that make the composites obtain both pseudocapacitance and double-layer capacitance.

Keywords: molybdenum disulfide, reduced graphene oxide, polyaniline, ternary composites, supercapacitors

### INTRODUCTION

To meet the burgeoning need of light-weight and portable electronic devices, efficient and environmentally-friendly electrochemical energy storage systems are urgently developed (Liu et al., 2017). Among the various energy storage systems, supercapacitors have drawn tremendous research attention due to their low cost, environmental friendliness, fast charging and discharging rate, excellent power density, high cycling stability, and long cycle life (Dunn et al., 2011; Yu et al., 2013; Wu et al., 2017; Qu et al., 2018). According to the charge-storage mechanism, supercapacitors are classified into two categories: electrochemical double-layer capacitors (EDLCs) and pseudocapacitors (Zhang et al., 2016). In general, pseudocapacitors of transition metal oxides and conducting polymers possess much higher specific capacitance than EDLCs of carbon materials, but their cycling stability is inferior (Li et al., 2014). Naturally, to draw on each other's strengths, a binary or ternary hybrid material composed carbon materials, a transition metal oxide and conducting polymer is more effective for high specific capacitance and long life time (Chen et al., 2014).

Graphene, a two-dimensional monolayer of sp<sup>2</sup> carbon atoms, is considered as an extremely promising candidate for future advanced applications in supercapacitors due to its excellent electrical conductivity, high surface area, good mechanical flexibility, and chemical stability (Zhang et al., 2012). Meanwhile, it is an ideal substrate for the growth and anchoring of nanomaterials, such as metal oxide, metal sulfide and conducting polymers, to exploring hybrid composite for improved electrochemical properties (Zhao et al., 2018). Among various hybrid materials, graphene/MoS2, graphene/PANI, MoS2/PANI binary nanostructure are found to be promising electrode materials for supercapacitors (Ataca et al., 2012; Huang et al., 2013; da Silveira Firmiano et al., 2014; Thangappan et al., 2016). For example, Thangappan et al. reported a facile one step preparation of a molybdenum disulfide (MoS2) nanosheet-graphene (MoS2/G) composite with the in situ reduction of graphene oxide, which exhibited a high specific capacitance of 270 and 90 F g−<sup>1</sup> at 0.1 and 1.0 A g−<sup>1</sup> , respectively. In addition, its specific capacitance can still remain 89.6% after 1,000 cycles at 0.6 Ag−<sup>1</sup> (Thangappan et al., 2016). However, three major defects found in the binary composite including the low practical capacitance, poor cycle stability and poor rate performance, hinder its wide application in supercapacitors (David et al., 2014).

Recently, a ternary composite of MoS2/graphene wrapped with Fe3O4, polypyrrole and carbon nanotubes has been tried out in the fields of lithium ion batteries, electromagnetic wave absorption and electrocatalyst (Khan et al., 2016; Xie et al., 2016; Li et al., 2017). However, there are few reports on the synthesis of MoS2/rGO/PANI ternary nanostructure for supercapacitors applications.

In this work, we prepared molybdenum disulfide/reduced graphene oxide/polyaniline (MoS2/rGO/PANI) ternary composites by a facile two-step method. In the first step, the MoS<sup>2</sup> nanosheets are uniformly grown on the surface of the GO nanosheets through a hydrothermal process to produce a MoS2/rGO binary composite. In the second step, the MoS2/rGO/PANI ternary composites were synthesized by in situ polymerization of aniline on the out face of the MoS2/rGO binary composite. In the MoS2/rGO/PANI ternary composites, the weight percent of PANI can effectively improve their electrochemical performance when they serve as the electrode materials in supercapacitors. The results indicate that the MoS2/rGO/PANI ternary composites deliver a very high specific capacity and excellent cyclic stability compared with the MoS2/rGO binary composite.

### EXPERIMENTAL

## Synthesis of a MoS2/rGO Binary Composite

Graphene oxide (GO) was synthesized from natural graphite flakes by a modified Hummers method (Zhang et al., 2015). The MoS2/rGO binary composite was prepared by a facile hydrothermal method. Typically, GO (0.8 g), (NH4)6Mo7O24·4H2O (1.236 g), thiourea (4 g) and HCl solution (0.2 ml) were dispersed in deionized water (40 ml) and sonicated for 1 h to form a uniform suspension. The above mixed dispersion was transferred into a Teflon-line stainless steel autoclave (50 ml) and annealed at 220◦C for 18 h. After cooling down, the black precipitate was collected by centrifugation, washed with deionized water and ethanol for several times, and dried at 80◦C for 24 h.

### Synthesis of MoS2/rGO/PANI Ternary Composites

The MoS2/rGO/PANI ternary composites were synthesized by in situ polymerization in the presence of MoS2/rGO and aniline. Typically, a certain amount of MoS2/rGO and dodecyl benzenesulfonic acid (4.065 g) were dispersed into 50 ml deionized water with ultrasonic radiation. Aniline (1.16 g) and deionized water (75 ml) were poured into the above suspension and sonicated for 1 h. A solution of APS (0.5 M, 25 ml) was dropwise added to the above mixed dispersion and continually stirred at 0◦C for 5 h and then at 20◦C for 2 h. After that, the blackish green product was filtered and washed with acetone, and then dried in a vacuum oven at 60◦C for 8 h. The resultant MoS2/rGO/PANI ternary composites were denoted as MGP-X, where X (X = 50, 60, 70 and 80) represents the weight percentage loading of PANI in the ternary composites.

### Material Characterization

The morphology and microstructure of the samples were characterized by a S-4800 scanning electron microscope (SEM) and a JEOL 2010 field-emission transmission electron microscope (TEM). The crystalline structures of the samples were performed on a Rigaku D/Max-2500 X-ray diffractometer with Cu Kα radiation (λ = 0.1542 nm) in the 2θ range from 5◦ to 90◦ . Fourier transform infrared (FT-IR) spectra were recorded on a Bruker Optics TENSOR 27 spectrometer using KBr pellets in the wave-number range of 400–4,000 cm−<sup>1</sup> .

### Electrochemical Measurements

The electrochemical performance of the samples used as electrode materials for supercapacitors were measured in a threeelectrode system. The working electrodes were fabricated by mixing 80 wt% active materials, 10 wt% carbon black, and 10 wt% polytetrafluoroethylene (PTFE) solution. The mixture was pasted on stainless steel network (1 × 1 cm<sup>2</sup> ) and dried at 80◦C for 12 h in a vacuum oven. The mass of active materials loaded on the working electrodes were 4–5 mg. A platinum foil and a saturated calomel electrode (SCE) were used as the counter electrode and reference electrode, respectively, and 1 M H2SO<sup>4</sup> aqueous solution was used as the electrolyte. Cyclic voltammetry (CV) tests were obtained at different scan rates (1, 2, 5, 10, 20, 50, and 100 mV s−<sup>1</sup> ) within a potential window of 0–1.0 V vs. SCE. Electrochemical impedance spectroscopy (EIS) measurements were carried out in the frequency range from 0.01 Hz to 100 KHz with 5 mV AC voltage amplitude at open circuit potential. Galvanostatic charge-discharge (GCD) investigations were performed at various current density (1, 2 3, 4, and 5 A g−<sup>1</sup> ) in a potential range of 0–1.0 V vs. SCE. The capacities of the samples were calculated based on the mass of the active materials.

### RESULTS AND DISCUSSION

### Structure and Morphology of MoS2/rGO/PANI Ternary Composites

**Figure 1** shows XRD patterns of the PANI, MoS2/rGO binary composite, and MoS2/rGO/PANI ternary composites, respectively. The diffraction peaks of the PANI located at 2θ = 20.3◦ and 25.3◦ , which can be assigned to (020) and (200) crystal planes of the emeraldine PANI salt, respectively (Tong et al., 2014). The MoS2/rGO binary composite exhibits the diffraction peaks centered at 2θ = 14.1◦ , 33.1◦ , 39.6◦ , and 58.9◦ , which can be ascribable to the (002), (100), (103) and (110) crystal planes of 2H-phase MoS<sup>2</sup> (JCPDS no. 37-1492), respectively. The diffraction peak of the rGO located at 2θ = 26.5◦ cannot be detected in the MoS2/rGO binary composite, which indicates that the restacking of graphene layers was inhibited by MoS<sup>2</sup> nanosheets (Wang et al., 2014; Dai et al., 2017). In the diffraction spectrum of the MoS2/rGO/PANI ternary composites, there are some characteristic diffraction peaks of both MoS2/rGO and PANI, revealing that the PANI is successfully attached onto the surface of the MoS2/rGO binary composite. Moreover, the intensities of the diffraction peaks from PANI gradually increase with the elevated weight percent of PANI. All the results indicate that these three components are fully compounded together.

**Figure 2** shows FT-IR spectra of the PANI, MoS2/rGO binary composite, and MoS2/rGO/PANI ternary composites, respectively. As shown in **Figure 2**, the spectrum of PANI shows strong absorption peaks at 1,120, 1,250, 1,310, 1,489, and 1,550 cm−<sup>1</sup> due to the C = N stretching, C–N stretching of the second amine, the aromatic C = C stretching vibration of the benzenoid and quinonoid rings, respectively (Cong et al., 2013). In the FT-IR spectra of the MoS2/rGO binary composite, there is no obvious peak arisen from the vibration of oxygen containing functional groups on GO, which is attributed to the reduction of graphene oxide after hydrothermal treatment. The weak peak at about 500 cm−<sup>1</sup> is assigned to MoS<sup>2</sup> vibration. Furthermore, the FT-IR spectra of MGP-50 show the obvious existence of all the PANI and MoS2/rGO characteristic peaks. With the increasing weight percent of PANI, the intensities of the main characteristic PANI peaks in the MoS2/rGO/PANI ternary composites all show an

increase. This indicated that the PANI was successfully coated on the surface of the MoS2/rGO binary composite, which is helpful to improve the dispersibility of MoS2/rGO/PANI ternary composites in the electrolyte.

The morphologies of the MoS2/rGO binary composite and MoS2/rGO/PANI ternary composites are shown in **Figure 3**. As

MoS2/rGO/PANI ternary composites.

FIGURE 3 | SEM and TEM images of the MoS2/rGO binary composite (a,b), and SEM images of the MoS2/rGO/PANI ternary composites [MRP-50 (c), MRP-60 (d), MRP-70 (e), and MRP-80 (f)].

shown in **Figures 3a,b**, the MoS<sup>2</sup> nanosheets were well-scattered on the surface of rGO nanosheets to form hetero-layered architecture. The MoS2/rGO binary composite as an attractive substrate can supply a large number of active sites for the growth of PANI. **Figures 3c–f** show the morphology of the MoS2/rGO/PANI ternary composites with different weight percents of PANI. As shown in **Figure 3c**, the MGP-50 ternary composite shows that a well-defined and interconnected 3D network architecture. The PANI in a planar shape were polymerized and attached onto the surface of the MoS2/rGO binary composite due to the electrostatic interaction between negatively charged MoS2/rGO and positively charged PANI (Luo et al., 2015). With the increased weight percent of PANI, PANI coated MoS2/rGO assemble to generate a compact and laminated morphology so that the layer structure of MoS2/rGO is not clearly seen due to the low content (**Figures 3d–f**). Such particular structure of the MoS2/rGO/PANI ternary composites could increase the dispersion of PANI and improve

the interfaces of PANI with electrolyte, which might be beneficial for the improvement of electrochemical performance of the MoS2/rGO/PANI ternary composite as electrode materials.

### Electrochemical Performance of MoS2/rGO/PANI Ternary Composites

**Figure 4** shows the CV curves of the MoS2/rGO binary composite and MoS2/rGO/PANI ternary composites at scan rates of 2 and 5 mV s−<sup>1</sup> . In the case of the MoS2/rGO/PANI ternary composites, it can be found that CV loop exhibit larger areas under redox curve than the MoS2/rGO binary composite, which indicates its higher specific capacitance. From the CV curves for the MoS2/rGO/PANI ternary composites at different PANI content, it was noted that the MGP-80 electrode shows the largest rectangular curve corresponding to the highest capacitance among the four samples. The CV curves for the MGP80 at a low scan rate of 2 mV s−<sup>1</sup> show severely distorted rectangular shapes as well as two conspicuous pairs of small, broad oxidation and reduction peaks, resulting from the cocontribution of electrical double-layer capacitance generated from MoS2/rGO and pseudocapacitance arising from PANI (Wang et al., 2015). The redox peaks become less obvious, and the potential of the oxidation peak moves to higher potential while the potential of the reduction peak shifts to lower at a high scan rate of 50 mV s−<sup>1</sup> . It should be noted that MoS2/rGO/PANI ternary composites have fast ionic transport for charge-discharge operations, which results from synergetic effect of MoS2/rGO and PANI.

The supercapacitance behavior of the MoS2/rGO/PANI ternary composites was further investigated using GCD measurements, as shown in **Figure 5**. **Figure 5A** displays the GDC curves of the MoS2/rGO,MoS2/rGO/PANI with different amounts of PANI at a current density of 1 Ag−<sup>1</sup> . It can be seen that the shapes of the GDC curves of ternary composites are similar to that of binary composite, which indicates that the ternary composite possesses the co-contribution of electrical double-layer capacitance and pseudo-capacitance. The ternary composites with different amounts of PANI exhibit a higher discharge time and lower IR drop than binary composite, which indicates that the introduction of PANI can enhance the capacitance and conductivity (Wang et al., 2013). The discharge time of the ternary composites increased with an increase amounts of PANI. The MGP-80 ternary composite exhibits the maximum discharge time. Furthermore, the GCD curves of the MRP-80 at different current densities from 1 to 5 A g−<sup>1</sup> are shown in **Figure 5B**. It could be seen that with the increase of current density, the discharge time of the composite decreased due to partial accessibility for electrolyte ions within the active material at high currents. **Figure 5C** reveals the specific capacitances of the ternary composites at the current densities of 1, 2, 3, 4, and 5 Ag−<sup>1</sup> . As can be observed, the capacitances of the MGP-80 were 570, 400, 303, 220, and 212 Fg−<sup>1</sup> , respectively. **Figure 5D** shows the cycle stability of the ternary composite at a current density of 1 A g−<sup>1</sup> . After 500 charging and discharging cycles, the specific capacitances of MPG-50, MPG-60, MPG-70

FIGURE 6 | Nyquist plots of the MoS2/rGO binary composite and MoS2/rGO/PANI ternary composites (the inset is an enlarged view of the Nyquist curves).

and MPG-80 retained 64.3, 67.7, 71.5, and 78.6% of the value of the first cycle, respectively. The good cycling stability of MPG-80 is ascribed to the fact that the synergistic effect between the MoS2/rGO and PANI could relieve the volumetric shrinkage or swelling of PANI during the charge/discharge process (Dai et al., 2013).

EIS was carried out to describe the electrochemical process of the electrode/electrolyte interface. **Figure 6** shows the Nyquist plots of MoS2/rGO binary composite and MoS2/rGO/PANI ternary composites. In high-frequency regions, the intercept of the semicircle with the X-axis represents the equivalent series resistance (Rs) of the electrode materials, while the diameter of the semicircle corresponds to the charge transfer resistance (Rct) (Sk et al., 2014). From the Nyquist plots, the R<sup>s</sup> of MoS2/rGO binary composite and MoS2/rGO/PANI ternary composites (MGP-50, MGP-60, MGP-70 and MGP-80) were 0.28, 0.90, 0.93, 0.97, and 1.03 , respectively, indicating that the introduction of PANI could decrease the electrical conductivity of binary composite. However, Rct displays a opposite trend, probablely because that the hierarchical structures of MoS2/rGO/PANI ternary composites show the fast charge transfer rates and reflect a preferable electrochemical performance. In low frequency regions, the nearly parallel to imaginary axis of the lines show that MoS2/rGO/PANI ternary composites have an ideal capacitive behavior. These results indicated that the ternary composite could improved charge storage and transportation within the electrode, which should be considered as a better electrode material (Liu et al., 2014).

### CONCLUSIONS

MoS2/rGO/PANI ternary composites with different amounts of PANI were successfully prepared by a two-stage synthesis combing a hydrothermal method with in situ chemical oxidative polymerization, and their supercapacitor performance were further investigated using electrochemical measures. The introduction of PANI in the MoS2/rGO binary composite not only hindered the agglomeration of MoS2/rGO, but also resulted in a synergistic effect among these three components. MoS2/rGO/PANI ternary composite with 80% PANI (MGP-80) showed the highest specific capacitance of 570 F g−<sup>1</sup> at of 1 A g−<sup>1</sup> which was comparatively larger than MoS2/rGO binary composite (200 F g−<sup>1</sup> ). Furthermore, it was noticed that the specific capacitance of the MGP-80 composite retained more than 78.6% after 500 cycles at 1 A g−<sup>1</sup> . All the evidences suggest that the MoS2/rGO/PANI ternary composite is a high-performance electrode material for next-generation supercapacitors.

### REFERENCES


### AUTHOR CONTRIBUTIONS

The work cannot be completed without kind cooperation of all authors. L-ZB and FL: Carried out the material preparation and electrochemical test; Y-HW and S-SC: Carried out and analyzed the FT-IR, XRD, and SEM analysis; L-ZB: Wrote the paper and all authors discussed the results and revised the manuscript; Z-YZ and Y-QL: Attained the main financial support for the research and supervised all the experiments.

### FUNDING

This work was financially supported by the Natural Science Foundation of Shanxi Province (No. 201701D221088) and the Basic Scientific Research Foundation of North University of China (No. XJJ2016010).


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Bai, Wang, Cheng, Li, Zhang and Liu. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Capacity Increase Investigation of Cu2Se Electrode by Using Electrochemical Impedance Spectroscopy

### Xiuwan Li\*, Zhixin Zhang, Chaoqun Liu and Zhiyang Lin

*Fujian Provincial Key Laboratory of Light Propagation and Transformation, College of Information Science and Engineering, Huaqiao University, Xiamen, China*

Cu2Se nanoflake arrays supported by Cu foams are synthesized by a facile hydrothermal method in this study. The Cu2Se materials are directly used as an anode for lithium ion batteries, which show superior cycle performance with significant capacity increase. Combining with previous reports and scanning electron microscope images after cycling, the capacity increase caused by the reversible growth of a polymeric film is discussed. Electrochemical impedance spectroscopy is used to test the reversible growth of the polymeric film. By analyzing the three-dimensional Nyquist plots at different potentials during the discharge/charge process, detailed electrochemical reaction information can be obtained, which can further verify the reversible formation of a polymeric film at low potential.

### Edited by:

*Qiaobao Zhang, Xiamen University, China*

#### Reviewed by:

*Xiukang Yang, Xiangtan University, China Hongshuai Hou, Central South University, China Huixin Chen, Xiamen Institute of Rare Earth Materials, China*

\*Correspondence:

*Xiuwan Li lixiuwan@hqu.edu.cn*

#### Specialty section:

*This article was submitted to Physical Chemistry and Chemical Physics, a section of the journal Frontiers in Chemistry*

> Received: *20 April 2018* Accepted: *28 May 2018* Published: *12 June 2018*

#### Citation:

*Li X, Zhang Z, Liu C and Lin Z (2018) Capacity Increase Investigation of Cu*2*Se Electrode by Using Electrochemical Impedance Spectroscopy. Front. Chem. 6:221. doi: 10.3389/fchem.2018.00221* Keywords: Cu2Se, capacity increase, electrochemical impedance spectroscopy, polymeric film, lithium ion battery

### INTRODUCTION

With the rapid development of electric vehicles and hybrid electric vehicles, the trend of lithium ion batteries (LIBs) toward achieving higher energy density and greater output power is determined by the electrode materials (Wang et al., 2011; Zhou et al., 2012; Fu et al., 2013). The quality of materials is mainly judged from several aspects, including cycle performance, rate performance, energy density, etc.

For superior cycle performance, theoretically, the cycle curve should remain stable. In practice, this is difficult to achieve. For most electrode materials in LIBs, the capacities decay after repeated cycles. However, in recent years, there are some reports regarding LIB electrode materials which show the opposite trend, i.e., an increase in capacity after repeated cycles (Li et al., 2013; Ao et al., 2017; Cui et al., 2017; Huang et al., 2017; Yuan et al., 2017; Zhang et al., 2018; Zheng et al., 2018). These materials are primarily transition metal oxides and sulfides, and discussions regarding capacity increase focus on the activation of the material and the reversible formation of a gel-film caused by electrolyte decomposition (AbdelHamid et al., 2017; Deng et al., 2017; Li Z. et al., 2017; Tang et al., 2017; Zhu et al., 2017). For example, Ao et al. synthesized a novel honeycomb-like composite composed of carbon-encapsulated SnO<sup>2</sup> nanospheres for lithium ion and sodium ion batteries (Ao et al., 2017). This composite shows a capacity increase at a rate of 500 mA g−<sup>1</sup> in LIBs, and the enhancement was assigned to the activation process of the SnO2-based electrodes. Yuan et al. fabricated SnO2/polypyrrole hollow spheres by a liquid-phase deposition method using colloidal carbon spheres as a template, and the composite electrode showed a significant increase from 404 to 899 mAh g−<sup>1</sup> (Yuan et al., 2017).

**358**

This phenomenon was explained by the continuously reversible formation of a polymeric gel-like film. Similarly, Abdel Hamid et al. prepared iron oxide/rGO and SnO2/rGO nanosheets for use in LIBs, and these electrodes also show capacity increase (AbdelHamid et al., 2017). From these examples, we can see that reversible polymeric gel-like films greatly increase the electrode capacity, and sometimes the increased capacity exceeds the theoretical capacity of the material.

In this paper, Cu2Se nanoflake arrays supported by Cu foams were prepared by a simple hydrothermal method. Cu2Se is an important and meaningful material in the field of energy storage and conversion, such as solar cell, thermoelectric, oxygen reduction reaction, sodium, and lithium ion battery et al. (Xue et al., 2006; Liu et al., 2013; Nguyen et al., 2013; Li et al., 2014; Ge et al., 2018c) For LIBs, the Cu2Se electrode shows interesting cycle performance: the capacity after the 20th cycle is 213.4 mAh g −1 , and the capacity of the 200th cycle is 736.7 mAh g−<sup>1</sup> , which is about 3.5 times the previous capacity. The capacity increase was discussed and electrochemical impedance spectroscopy (EIS) was used to understand the physical mechanisms governing the observed capacity increase.

## EXPERIMENT

## Synthesis of Cu2Se Electrode

Typical synthesis of Cu2Se supported by Cu foam is as follows. CH4N2Se (Selenourea, 30 mg) and 40 mL DI water were mixed under stirring. After stirring, it was transferred into a Teflonlined stainless steel autoclave with 50 mL capacity. A piece of cleaned Cu foam (110 PPI pore size and 1.6 mm thick) with an area of 4.0 × 2.0 cm<sup>2</sup> was put into the autoclave as both a reactant and substrate. The autoclave was maintained in an oven at 140◦C for 12 h. After the autoclave was allowed to cool to room temperature, the Cu foam piece with the synthesized Cu2Se was fetched out and rinsed with deionized water and ethanol several times, and then dried at 60◦C in vacuum.

On average, Cu2Se mass loading is ∼2 mg cm−<sup>2</sup> . By carefully weighing the mass of Cu foam before and after reaction, the active mass of Cu2Se (mCuSe) was derived from mCuSe = 1m × 240.2/64.13, where 1m is the mass difference of the Cu foam before and after hydrothermal synthesis.

### Structural Characterization

The structure and morphology of the Cu2Se materials were characterized by X-ray powder diffraction (XRD, Rigaku D/Max-2400 with Cu Kα radiation) and field-emission scanning electron microscopy (SEM, Hitachi, S-4800).

### Electrochemical Characterization

Electrochemical characterization was carried out using a CR-2032-type coin cell, which was assembled in a high-purity glove box filled with argon (H2O < 0.1 ppm, O<sup>2</sup> < 0.1 ppm, Mikrouna Co., Ltd.). Cu2Se was used as the working electrode and Li foil was used as the counter and reference electrode. Celgard 2320 was used as the separator membrane. The electrolyte was 1 M lithium hexafluorophosphate (LiPF6) dissolved in ethylene carbonate: dimethyl carbonate: ethyl methyl carbonate in a 1:1:1 volume ratio. Galvanostatic discharge/charge cycling, cyclic voltammetry, and electrochemical impedance spectroscopy (EIS) were carried out at room temperature using a multichannel battery tester (Neware, BTS-610) and an electrochemical workstation (CHI, 660E), respectively. For EIS test, the frequency range is from 1 to 100,000 Hz, including the middle- and high-frequencies. The test voltage is from 0.02 to 3 V, and the step is 0.1 V. Thirty sets of tests from 0.02 to 3 V are without interruption.

### RESULTS AND DISCUSSION

Scanning electron microscope (SEM) images of Cu2Se supported by Cu foams after hydrothermal synthesis are shown in **Figures 1a–c**. A porous nanoflake array structure is found in the surface of the Cu foams. These nanoflakes, which are approximately 100 nm in thickness, are arranged vertically on the substrate and interconnect with each other, forming a 3D porous structure, as shown in **Figure 1c**. This structure can benefit the transport of lithium ions and electrons thanks to their porous and 3D interconnected characteristics (Zhang et al., 2015). More importantly, the pores in a porous array structure can increase the contact area between material and electrolyte, as well as relieve volume expansion during the discharge/charge process. The as-prepared electrode is characterized by XRD, as shown in **Figure 1d**. There are three typical diffraction peaks located at 43.3, 50.4, and 74.1◦ , corresponding to the (111), (200), and (220) faces in Cu, respectively (JCPDS No. 85-1326). Five secondary diffraction peaks can be observed at 13.0, 26.2, 26.5, 39.8, and 44.0◦ , which match well with (030), (060), (221), (090), and (012) planes of Cu2Se, respectively (JCPDS No. 27-1131). These peaks indicate that the material formed during hydrothermal synthesis is Cu2Se.

Electrochemical tests were conducted on coin-type cells to evaluate the electrochemical performance of the Cu2Se as anodes in LIBs. Figure S1A shows the initial three discharge/charge voltage profiles at 100 mA g−<sup>1</sup> . In the first cycle, there are three main potential plateaus in the discharge curve, located at 1.7, 1.6, and 1.2 V. Correspondingly, the charge curve shows two main potential plateaus at 1.7 and 2.2 V. For the second and third cycles, the number of main discharge and charge potential plateaus is consistent with first cycle. The phenomenon of complex potential plateaus has been reported in the Cu2Se, Cu2−xSe, CuxS, and CuS electrodes (Xue et al., 2006; Ni et al., 2012; Chen et al., 2014; Zhou et al., 2014; Li H. et al., 2017). Correspondingly, the CV curves of the Cu2Se electrode are tested at a scan rate of 0.1 mV s−<sup>1</sup> and are shown in Figure S1B. Compared to the three main potential plateaus in the first discharge process, there are only two peaks in the first cathodic scan, which can be attributed to the closeness of the two potential plateaus. Meanwhile, two peaks appear at 1.94 V and 2.23 V during the anodic scan. For the 2nd and 3rd cycles, three reduction peaks near 2.1, 1.7, and 1.5 V and two oxidation peaks at about 1.8 and 2.2 V are observed, which is consistent with the results from the 2nd and 3rd discharge/charge voltage profiles.

**Figure 2A** shows the cycling performance of Cu2Se nanoflake electrode at a current density of 100 mA g−<sup>1</sup> . The first and second discharge capacities are 397.3 and 326.8 mAh g−<sup>1</sup> , respectively. During the first 20 cycles, the capacity decreases to 213.4 mAh g −1 . However, after the 20th cycle, the capacity increases quickly. Compared to the 20th discharge capacity, the capacity of 200th cycle is 736.7 mAh g−<sup>1</sup> , which is about 3.5 times the previous capacity. This capacity far exceeds its theoretical capacity (260 mAh g−<sup>1</sup> ). The phenomenon of increased capacity has been found in many metal oxide, sulfide, and selenide materials, and a lot of useful discussions have been reported (Ge et al., 2018a). There are two general explanations: One is the activation of materials during cycling; the other is the formation of a gel film caused by decomposition of the electrolyte. However, for this electrode, the main reason for the capacity increase cannot be the activation of the material as the capacity increase from 213.4 to 736.7 mAh g−<sup>1</sup> (i.e., from 0.795 to 2.385 mAh) is much higher than its theoretical value. In order to determine the real reason for the capacity increase, contrast voltage profiles curves of the 2nd and 200th discharge/charge are gathered and are shown in **Figure 2B**. We find that the voltage profiles of the 2nd cycle show plateau features, which corresponds to electrochemical reactions. After the capacity increase, the voltage profiles of the 200th cycle show significant changes and the plateau feature disappears. Figure S2 shows the discharge voltage profiles during the cycles. During the initial stages, we can find the plateaus for Cu2Se at 1.5 and 0.75 V. With repeated cycling, the plateaus slowly disappear, and the capacity quickly increases. In order to further study the plateau changes, the CV curves after 200 cycles are tested at a scan rate of 0.5 mV s−<sup>1</sup> between 0.02 and 3.0 V. The CV results are found to be changed compared to those from 3 initial cycles (as shown in Figure S1B). During the cathodic scan, three obvious peaks can be seen at 2.3, 1.4, and 0.6 V. These two peaks at 2.3 and 1.4 V are related to the peaks at 2.1 and 1.7 V of the initial CV curves. However, the largest peak at 0.6 V is a new peak, which cannot be found in the initial CV curves. We can also find a new peak at 2.6 V during the anodic scan. The appearance of two new peaks in the CV curves is caused by the reversible formation of a polymeric gel-film, which has reported to occur in other materials (Han et al., 2014; AbdelHamid et al., 2017; Deng et al., 2017; Huang et al., 2017).

As early as 2002, the Tarascon group has systematically studied the problem of capacity increase in metal oxides (Laruelle et al., 2002; Grugeon et al., 2003). Tarascon found that the reversible formation/dissolution of a conducting-type polymeric film at low potential in an alkyl carbonate solution led to the increased capacity and opined that highly reactive pristine metallic nanograins promote the growth of the polymeric film. To prove this point, Cu2Se was imaged using SEM after 200 cycles, as shown in **Figure 3**. It can be seen that a polymeric film layer has grown in most areas of the material surface. According to the SEM characteristics, these polymeric films possess a certain degree of electrical conductivity. In some exposed areas, we can find a large number of nanoparticles with about 10 nm particle

size (**Figure 3b**). This nanoparticle morphology is different from that of the previous nanoflake array, which confirms Tarascon's view. More importantly, compared to the nanoflake structure, a nanoparticle structure has a larger specific surface area. This can provide more locations for growing a reversible polymeric film (RPF), resulting in a greater capacity increase. EDS results show the elemental composition of the sample after 200 cycles. In addition to the copper and selenium elements that came from the original active material, we also found carbon, oxygen, fluorine, and phosphorus elements on the surface of the material. These elements come from the electrolyte and form reversible polymeric and solid electrolyte interface (SEI) films that remain on the surface of the material.

To further verify the reversible growth of RPF, we performed electrochemical impedance spectroscopy (EIS) after 200 cycles and after 3 CV cycles in the Cu2Se samples. EIS is an effective electrochemical test technique, which can be used to perform non-destructive testing of the reaction process inside the electrode in real-time (Fattah-alhosseini and Imantalab, 2015; Ge et al., 2018b; Joshi et al., 2018; Peng et al., 2018). We examine the growth of RPF by testing medium- and high-frequency EIS at different potentials (0.02–3 V vs. Li/Li+). **Figure 4A** shows the 3D Nyquist plot at different potentials after 200 cycles. It can be clearly seen that the Nyquist curves are different and continuous at different potentials (0.02–3 V vs. Li/Li+), illustrating the difference and continuity of the electrode reaction at different potentials. In the low potential area, the second semicircle significantly increased, proving the growth of the RPF of the electrode material in the low potential area. To the opposite end, the two semicircles at a high potential area of about 2.5–3.0 V are all small, and this phenomenon may be caused by the decomposition of the RPF. This conclusion is consistent with the CV results after 200 cycles (as shown in Figure S2B). Correspondingly, we also tested the EIS after 3 CV cycles at different potentials, and the 3D Nyquist plot is shown in **Figure 4B**. The Nyquist curves after 3 CV cycles at different potentials are obviously smaller than the curves after 200 cycles, proving that the internal resistance after 3 CV cycles is smaller than the resistance after 200 cycles. This may occur due to decomposition of the electrolyte caused by the formation of RPF after 200 cycles. As can be seen from **Figure 4B**, the position of the big semicircle is about 1.5–2.0 V, corresponding to the electrode reaction potential. This is because a strong electrochemical reaction occurs at the electrode reaction potential and a large amount of charge is transferred from the ions to the electrons, which results in a large charge transfer resistance to form a larger semicircle (Jeon et al., 2014; Rodrigues et al., 2016; Yang et al., 2017).

To further analyze the EIS results, the obtained impedance data were analyzed by fitting the equivalent electrical circuit (as shown in Figure S3). R0, RSEI, and Rct are the electrolyte resistance, SEI film resistance, and charge transfer resistance, respectively. The fitted impedance parameters after 200 cycles and after 3 CV cycles are listed in **Figures 4C,D**, respectively. For ease of understanding, the respective CV curves were added in the right-Y scale, and detailed numerical comparison of the impedance is listed in Table S1. From **Figure 4D**, we can see that the resistances of R0, RSEI, and Rct in the sample after

3 CV cycles at different potentials are relatively stable (these resistance values are <15 ohms), indicating that the active material/SEI film/electrolyte film has a stable structure under different charging and discharging potentials. More importantly, the stable material structure proves that there is no formation of RPF during the discharge/charge process, which is consistent with the CV curve in **Figure 4D**. Compared to the sample tested after 3 CV cycles, the resistances after 200 cycles at different potentials are more complicated. First, the electrolyte resistances after 200 cycles are about 9.2 ohms, which is higher than the sample after 3 CV cycles (6.0 ohms) and is caused by electrolyte reduction due to the decomposition of the electrolyte after 200 cycles. Second, the SEI film resistances after 200 cycles at different potentials are between 10 and 40 ohms, which are also much higher than that of the sample after 3 CV cycles (as shown in Table S1). The high SEI film resistance may be caused by the irreversible accumulation of a polymeric film after 200 discharge/charge cycles, as can be seen from **Figure 3**. Finally, the charge transfer resistances after 200 cycles vary greatly with the potential. The highest resistance can reach up to 59.81 ohms at 0.4 V, and the lowest resistance is about 5.74 ohms at 3.0 V. Charge transfer resistance is related to the electrochemical reaction, and in the discharge/charge process of Cu2Se after 200 cycles, there are two electrochemical reactions. One is the redox reaction of Cu2Se, and the other is the synthesis and decomposition reaction of RPF (Yuan et al., 2009). According

to relevant reports and previous analysis, the synthesis reaction of RPF for Cu2Se electrode occurs in the low potential area (Grugeon et al., 2003; Ponrouch et al., 2012; Li et al., 2013). In this potential area, the RPF synthesis provides additional capacity. Therefore, the charge transfer reaction occurs in the RPF between the electrode and the SEI film instead of within the electrode. With the continuous formation of the RPF, the thicknesses of the SEI film and the RPF increase, and the charge transfer resistance sharply increases. At the same time, the SEI film resistance also increases accordingly. In the high potential area, the corresponding electrochemical reaction is the decomposition of the RPF. As the film decomposes, the charge transfer resistance decreases, and the SEI film resistance also decreases, which is consistent with the CV curve in **Figure 4C**. As shown in Table S1, one can see that, whether we consider SEI film resistance or charge transfer resistance, they have the smallest difference in the high potential area. These EIS results show that the RPF grows in the low potential region and decomposes in the high potential region.

### CONCLUSION

In this paper, we prepared Cu2Se nanoflake arrays through a facile hydrothermal method, and the synthesized Cu2Se material was used as an anode for lithium ion batteries, which shows superior cycle performance with high capacity. The

FIGURE 4 | (A,B) 3D Nyquist plots at different potentials after 200 cycles and 3 CV cycles; (C,D) show the corresponding fitting resistances at different potentials with the CV curves after 200 cycles and 3 CV cycles.

capacity increase was discussed and electrochemical impedance spectroscopy was used to test the reversible growth of a polymeric film. By analyzing the Nyquist plots at different potentials during discharge/charge cycles, we can obtain detailed electrochemical reaction information and further verify that the reversible formation of a polymeric film at low potential leads to the observed capacity increase.

### AUTHOR CONTRIBUTIONS

XL is responsible for experimental design, ZL is responsible for data analysis, ZZ and CL are responsible for instrument operation, and XL and ZL are responsible for writing article.

### REFERENCES


### FUNDING

This project was financially supported by the Fujian Provincial Natural Science Foundation of China (No. 2017J05008), the Special Foundation for theoretical physics Research Program of China (No. 11747092) and the Natural Science Foundation of China (No. 11504116).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2018.00221/full#supplementary-material

as anode materials for high performance lithium- and sodium-ion battery. J. Power Sources 359, 340–348. doi: 10.1016/j.jpowsour.2017.05.064


high-performance lithium ion battery anode. J. Power Sources 362, 20–26. doi: 10.1016/j.jpowsour.2017.07.024


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Li, Zhang, Liu and Lin. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Electrospun Single Crystalline Fork-Like K2V8O<sup>21</sup> as High-Performance Cathode Materials for Lithium-Ion Batteries

Pengfei Hao, Ting Zhu, Qiong Su, Jiande Lin, Rong Cui, Xinxin Cao, Yaping Wang and Anqiang Pan\*

*Department of Materials Physics, School of Materials Science and Engineering, Central South University, Changsha, China*

### Edited by:

*Kaili Zhang, City University of Hong Kong, Hong Kong*

### Reviewed by:

*Chaopeng Fu, Shanghai Jiao Tong University, China Ling Wu, Soochow University, China Chenghao Yang, South China University of Technology, China Ali Eftekhari, Queen's University Belfast, United Kingdom*

\*Correspondence:

*Anqiang Pan pananqiang@csu.edu.cn*

### Specialty section:

*This article was submitted to Physical Chemistry and Chemical Physics, a section of the journal Frontiers in Chemistry*

> Received: *06 April 2018* Accepted: *14 May 2018* Published: *01 June 2018*

#### Citation:

*Hao P, Zhu T, Su Q, Lin J, Cui R, Cao X, Wang Y and Pan A (2018) Electrospun Single Crystalline Fork-Like K2V8O21 as High-Performance Cathode Materials for Lithium-Ion Batteries. Front. Chem. 6:195. doi: 10.3389/fchem.2018.00195* Single crystalline fork-like potassium vanadate (K2V8O21) has been successfully prepared by electrospinning method with a subsequent annealing process. The as-obtained K2V8O<sup>21</sup> forks show a unique layer-by-layer stacked structure. When used as cathode materials for lithium-ion batteries, the as-prepared fork-like materials exhibit high specific discharge capacity and excellent cyclic stability. High specific discharge capacities of 200.2 and 131.5 mA h g−<sup>1</sup> can be delivered at the current densities of 50 and 500 mA g −1 , respectively. Furthermore, the K2V8O<sup>21</sup> electrode exhibits excellent long-term cycling stability which maintains a capacity of 108.3 mA h g−<sup>1</sup> after 300 cycles at 500 mA g <sup>−</sup><sup>1</sup> with a fading rate of only 0.043% per cycle. The results demonstrate their potential applications in next-generation high-performance lithium-ion batteries.

Keywords: potassium vanadate, electrospinning, fork-like nanostructure, cathode materials, lithium-ion batteries

### INTRODUCTION

Rechargeable lithium-ion batteries (LIBs) are one of the most important energy storage devices in microchips, cell phones, electric vehicles (EVs), and hybrid electric vehicles (HEVs) (Whittingham, 2004; Armand and Tarascon, 2008; Zheng et al., 2015; Ou et al., 2017; Pan et al., 2017a,b; Yang et al., 2017). Layer-structured LiCoO<sup>2</sup> has been extensively studied and largely used as commercial cathode materials due to their high working potential, high energy density and good cycling performance. However, it delivers a relatively low capacity of ∼130 m Ah g−<sup>1</sup> , which is only half of its theoretical capacity, thus restricting its further expansion in LIBs (Goodenough and Kim, 2010; Liang et al., 2014). In addition, the scarcity of Co resources has further pushed up the cost of LIBs manufacturing. Therefore, it would be interesting to find alternative electrode materials with lower cost, larger specific capacity and better safety.

As important cathode candidates in LIBs, vanadium oxides and vanadate have attracted extensive interests owing to their abundant reservation, high specific capacity, and high Li<sup>+</sup> diffusion efficiency (Maingot et al., 1993; Torardi and Miao, 2002; Ng et al., 2007; Liu et al., 2009; Mai et al., 2010a,b; Wee et al., 2010; Liu and Yang, 2011; Pan et al., 2011a; Rui et al., 2011; Varadaraajan et al., 2011; Wang et al., 2011; Liang et al., 2013b; Jian et al., 2014; Zhou et al., 2014; Meng et al., 2016; An et al., 2017). For example, vanadium pentoxide (V2O5) has a high theoretical capacity of 442 mA h g−<sup>1</sup> , but its commercialization is still limited by poor cyclic stability (Owens et al., 1999; Cao et al., 2005; Wang et al., 2006). Many studies have demonstrated that the incorporation of metal cations (such as Li+, Na+, K+, Ag+, Zn2+, Cu2+, etc.) (Ma et al., 2008; Liu and Tang, 2009; Cheng and Chen, 2011; Liang et al., 2013a, 2014; Bach et al., 2014; Yang et al., 2016) into V2O<sup>5</sup> interlayers is favorable to improve the electronic conductivity (Zhou et al., 2014), simultaneously creating more Li<sup>+</sup> intercalation channels (Khoo et al., 2010) and improve the structural stability (Pan et al., 2011b). It was reported that introduction of K <sup>+</sup> into vanadium oxygen layers is effective to deliver high specific capacity and excellent reversibility by the pillar effect and larger interlamellar space (Baddour-Hadjean et al., 2011, 2014; Xu et al., 2012, 2015; Fang et al., 2015, 2016; Meng et al., 2016). In addition, K2V8O<sup>21</sup> is a more stable crystal phase compared to V2O<sup>5</sup> or KVO<sup>3</sup> and it has a higher theoretical capacity of 261 mA h g−<sup>1</sup> . Manev et al. reported the usage of K2V8O<sup>21</sup> as a cathode material for the first time, which deliver an initial discharge capacity of about 190 mA h g −1 , but with a poor capacity retention (Manev et al., 1993). The electrochemical performance of K2V8O<sup>21</sup> can be improved by incorporation water into the layered structure by hydrothermal treatment (Aleksandrova et al., 2006, 2009). By doping inactive metal ions such as Nb5<sup>+</sup> and Ti4<sup>+</sup> into the crystal structure by mechanical ball milling, the capacity of K2V8O<sup>21</sup> (Ni et al., 2015) can also be enhanced. However, the cycling performance of K2V8O<sup>21</sup> is still unsatisfactory, in particular for the long-term cycling stability.

One-dimensional (1D) nanostructures such as nanowires (Mai et al., 2010b), nanorods (Gu et al., 2015), nanofibers (Wee et al., 2010) could offer shorter Li-ion diffusion pathways, higher specific surface area and faster electron transfer along longitudinal direction (Li et al., 2016). Besides, 1D nanostructures with layer-by-layer flakes have been proved favorable to increase the Li<sup>+</sup> diffusion rate and enlarge the contact areas between electrode and electrolyte (Liang et al., 2013a). layer-by-layer structured K0.25V2O<sup>5</sup> electrode exhibits a high initial discharge specific capacity of 256 mA h g−<sup>1</sup> and superior long-term cycling stability (Fang et al., 2015).

In this work, we reported the synthesis of single crystal forklike K2V8O<sup>21</sup> (KVO) by a facile electrospinning method with a subsequent calcination process. Fork-like K2V8O<sup>21</sup> material with layer-by-layer stacking structure has been fabricated and is used as a cathode material in LIBs. The K2V8O<sup>21</sup> forks exhibit high specific capacity and superior long-term cycling stability.

### EXPERIMENTAL

### Materials and Synthesis

All reagents and solvents were of analytical grade and used as received without further purification. K2V8O<sup>21</sup> was synthesized by single-nozzle electrospinning technique with subsequent annealing. Vanadium pentoxide (V2O5, ≥ 99.0 %), oxalic acid (H2C2O4·2H2O, ≥ 98.0 %), potassium nitrate (KNO3, ≥ 99.0 %) and polyvinylpyrrolidone (PVP, Mw ≈ 1, 300, 000) were used as starting materials. In a typical synthesis, 0.1 g of V2O<sup>5</sup> and H2C2O4·2H2O in a molar ratio of 1:3 were dissolved in 5 mL of de-ionized water (DI) under vigorous stirring at 80◦C for about 30 min to form a clear dark blue solution. Then, 0.0278 g of KNO<sup>3</sup> was added into the above vanadium oxalate solution under magnetic stirring for 10 min before the addition of 5 mL of N,N-Dimethylformamide (DMF) and 1 g of PVP to form a homogeneous viscous dark blue precursor solution. Subsequently, the precursor solution was loaded into a 10 mL plastic syringe with a 21-gauge stainless steel nozzle. The solution was subjected to electrospinning at a DC voltage of 12 kV under a flow rate of 0.04 mm min−<sup>1</sup> . The electrospun nanofibers were collected on aluminum foil with a distance of 15 cm between the nozzle and aluminum collector. Finally, the obtained precursor nanofibers were annealed in a muffle furnace at 500◦C in air for 2 h to yield the fork-like K2V8O21. The temperature ramping rate was set of 1◦C min−<sup>1</sup> .

### Material Characterization

The crystal phase of the as-prepared K2V8O<sup>21</sup> were characterized by X-ray diffraction (XRD, Rigaku D/Max 2500) with nonmonochromated Cu Kα radiation (λ = 1.54178 Å). The microscopic morphology of the products was investigated by scanning electron microscopy (SEM, FEI Nova Nano SEM 230) and transmission electron microscopy (TEM, JEOL JEM-2100 F). Thermogravimetric analysis (TGA) was carried out on a NETZSCH STA 449C analyzer in air from room temperature to 700◦C with a heating rate of 10◦C min−<sup>1</sup> .

### Electrochemical Measurements

Electrochemical measurement was performed using standard CR2016 type coin cells. The fork-like K2V8O21, acetylene black and polyvinylidene fluoride (PVDF) binder in a weight ratio of 70:20:10 were added into N-methyl-2-pyrrolidone (NMP) solution to make the slurry, which was coated on alumina foil and dried at 100◦C overnight under vacuum to obtain the electrodes. The mass loading of the K2V8O<sup>21</sup> cathode material for coin cell testing was about 1 mg cm−<sup>2</sup> . All coin cells were assembled in a glove box (Mbraun, Germany) filled with ultra-high pure argon gas. Metallic lithium foils and polypropylene membrane were used as counter electrode and separator, respectively. And 1 M LiPF<sup>6</sup> solution in ethylene carbonate/dimethyl carbonate (EC/DMC; 1:1, v/v) was used as the electrolyte. The cyclic voltammetry (CV) measurements of Li/K2V8O<sup>21</sup> coin cell was conducted using an electrochemical workstation (CHI660C, China) at a scan rate of 0.1 mV s−<sup>1</sup> in the voltage of 1.5–4 V. The galvanostatic charge/discharge (GCD) performances of the K2V8O<sup>21</sup> electrodes were conducted at room temperature on a Land Battery Tester (Land CT2001A, China) in a voltage range of 1.5–4.0 V vs. Li/Li+.

### RESULTS AND DISCUSSION

**Figure 1** shows the synthesis process of fork-like K2V8O<sup>21</sup> through the single-nozzle electrospinning technique and a subsequent annealing. Firstly, KNO3, V2O5, H2C2O4·2H2O, PVP, DMF and DI were mixed to form a homogeneous viscous precursor, which was electrospun into nanofibers by electrospinning process. The followed annealing treatment can convert the precursor nanofibers to the fork-like crystalline K2V8O<sup>21</sup> nanostructures in air at 500◦C.

**Figure 2** shows the structure and morphology of the electrospun precursor and its annealed products of fork-like

K2V8O21. **Figure 2a** shows the SEM image of the precursor nanofibers. The K2V8O<sup>21</sup> precursor nanofibers has an average diameter of 100 nm and a smooth and uniform surface. The formation of fiber-like structure can be attributed to the polymers serving as templates during the electrospinning process. Typically, the subsequent annealing could remove the backbones of polymers, and control the morphologies of final products such as nanofibers (Wee et al., 2010), nanowires (Mai et al., 2010b), and nanotubes (Zhao et al., 2015). 350◦C was chosen for sintering the precursor nanofibers, and it can be seen from **Figure 2b** that the fibrous structure was not retained. At this temperature, PVP was decomposed and then converted to carbonaceous species (Liang et al., 2007; Teh et al., 2013). Meanwhile, the KVO nanoparticles that encapsulated in the nanofibers were gradually grown into flat flakes. Because of PVP enclosing the precursor with a fibrous structure and restricting the growth of KVO nanoparticles, KVO small flakes were formed into a hierarchical nanobelt structure. As shown in **Figure 2c**, the final fork-like K2V8O<sup>21</sup> that combined nanorods and nanoflakes formed when the heating temperature was elevated to 500◦C. During the annealing process, the transformation of KVO nanoflakes into KVO forks occurred, which was caused by the re-crystallization of K2V8O21. Large space can be clearly seen among the K2V8O<sup>21</sup> forks, which may be favorable to the diffusion of lithium ion. **Figure 2d** shows a single K2V8O<sup>21</sup> fork at high magnification, which reveals the layered structures of KVO forks, indicating a feature of layer-by-layer structure. K2V8O<sup>21</sup> compound was also prepared by sol-gel route for comparison. The width of sol-gel K2V8O<sup>21</sup> compound (SG-KVO) is larger than that of the electrospun fork-like K2V8O<sup>21</sup>

as shown in Figure S1 (Supplementary Material). Furthermore, the electrospun fork-like K2V8O<sup>21</sup> (ES-KVO) show two rod-like tips which are different from the sol-gel K2V8O21, which could be induced by the combustion of PVP and re-crystallization

nanofibers in air from room temperature to 700◦C. The temperature ramping rate was 10◦C min−<sup>1</sup> .

of K2V8O<sup>21</sup> clusters. **Figure 2f** shows the element mapping of fork-like K2V8O<sup>21</sup> in **Figure 2e**, in which the red, blue, yellow and green spot represent the element of carbon, potassium, vanadium and oxygen elements, respectively. It can be seen that these elements are present in the sample and homogeneously dispersed on the fork-like nanostructures, further confirming that C element is well distributed in the fork-like K2V8O21.

Thermogravimetric analysis (TGA) was performed from 25 to 700◦C to investigate the thermal decomposition of KVO at a heating rate of 10◦C min−<sup>1</sup> (**Figure 3**). The as-obtained KVO nanofibers lost mass weight owing to the decomposition of metallic precursor and PVP during the annealing process, leading to the significant surface rupture and the formation of K2V8O<sup>21</sup> simultaneously. The mass loss within the range of 25–200◦C was attributed to the loss of volatile components, such as residual solvent (DMF, H2O) and adsorbed moisture (Ko et al., 2015). At 332.2◦C, there was a prominent DSC exothermic peak with a dramatic mass-loss. This exothermic peak may be due to the decomposition of vanadium oxalate, potassium nitrate and the degradation of PVP, which has both intra—and intermolecular transfer reactions according to the degradation mechanisms (Azhari and Dish, 1998). The following exothermic peak, which was accompanied with a significant weight loss at 478.3◦C was probably due to the formation of K2V8O<sup>21</sup> compound and the oxidation of carbon and carbon monoxide that originated from PVP molecules (Wang et al., 2008). It can be concluded that at

500◦C the chemical reaction was completed and the K2V8O<sup>21</sup> was obtained. From 500 to 570◦C, there was a mass loss about 4.33% and stabilized at about 15% at temperatures above 570◦C, which implied that some C elements were remained when the fork-like KVO was obtained.

The fork-like K2V8O<sup>21</sup> prepared via electrospinning route was examined by X-ray diffraction and the result is showed in **Figure 4a**. The as-obtained product shows obvious XRD diffraction peaks, which illustrate the good crystallinity of the product. As for the XRD pattern, all the diffraction peaks show good agreement with K2V8O<sup>21</sup> phase (JCPDS Card No. 24-0906). No obvious peaks from other phases can be found, demonstrating that the as-prepared K2V8O<sup>21</sup> is of high purity. It is worth noting that the diffraction pattern of K2V8O<sup>21</sup> has not been fully studied so far since the structure of this material was found.

The structure of the as-prepared fork-like K2V8O<sup>21</sup> was further examined by transmission electron microscopy (TEM). The TEM image in **Figure 4b** shows the morphology of the endpoint of the obtained K2V8O21. It can be confirmed that the fork-like nanostructure is consistent with the SEM observation (**Figure 2c**). As is shown in **Figure 4c**, a detailed examination of a KVO folk reveals the broken tips of the K2V8O<sup>21</sup> folk and the endpoint of K2V8O<sup>21</sup> are transparent under the electron beam, implying an ultrathin property of the nanorods. Moreover, it is distinct that the nanorods exhibit layer-by-layer stacked structures, which is beneficial to the lithium intercalation and de-intercalation during the charge-discharge process (Pan et al., 2011b). The selected area electron diffraction (SAED) pattern is displayed in **Figure 4d**, which indicates that the as-prepared K2V8O<sup>21</sup> nanorods are single-crystalline which is consistent with a previous report (An et al., 2010). It is generally accepted that single-crystalline nanostructured materials hold no significant grain boundaries with few defects, which facilitates the Li-ion diffusion during electrochemical reactions because Li ions need not to go across grain boundary and defects (Liang et al., 2007; Chen, 2013).

X-ray photoelectron spectroscopy (XPS) was further conducted to prove the chemical components and element valences of fork-like K2V8O21. As shown in **Figure 5A**, full survey spectrum reveals the existence of potassium, vanadium, oxygen and carbon elements in the as-prepared product. The peaks at 530.2, 517.2, 292.6, and 284.8 eV can be ascribed to O 1s, V 2p3/2, K 2p3/2, and C 1s, respectively. Correspondingly, the peaks at 292.6 and 295.4 eV which are 2.8 eV apart in the highresolution spectrum of K 2p (**Figure 5B**) could be ascribed to K 2p3/<sup>2</sup> and K 2p1/<sup>2</sup> of K<sup>+</sup> (Li et al., 2001). As shown in **Figure 5C**, the characteristic doublet of potassium vanadate V 2p is found at 517.6 and 524.7 eV, revealing that the fork-like K2V8O<sup>21</sup> only contains V5<sup>+</sup> valence state and no V4<sup>+</sup> can be detected (Silversmit et al., 2004). In the broad spectrum (**Figure 5A**), some C element was observed and in the high-resolution spectrum of C 1s (**Figure 5D**), the peaks centered at 284.8, 286.4 and 288.5 eV

were due to the C-C, C-O and C = O bonds, respectively. The existence of C element derived from PVP was partially retained after annealing.

The electrochemical properties of the electrospun fork-like K2V8O<sup>21</sup> as cathode are evaluated by assembling CR2016 type coin half lithium-ion batteries. The cyclic voltammetry (CV) curves of K2V8O<sup>21</sup> electrode is recorded in a voltage window of 1.5–4.0 V at a scan rate of 0.1 mV s−<sup>1</sup> . As shown in **Figure 6A**, two main reduction peaks located at 2.49 V and 2.82 V are ascribed to the intercalation of the lithium ions into K2V8O<sup>21</sup> electrode in several steps. Moreover, a broad peak at about 1.75 V is also detected, which suggest further intercalation of lithium ions into the electrode material. During the anodic scan, two main oxidation peaks near 2.70 V and 3.05 V are well detected, which can be attributed to the de-intercalation of the lithium ions (Manev et al., 1993). According to the three consecutive cycles, the electrode materials has the structural reversibility due to the redox peak positions do not change much, although the peak current intensity decreases gradually. The intercalation/de-intercalation behavior of lithium in the cycling process of K2V8O<sup>21</sup> electrode can be expressed as equation (1):

$$\rm{K\_2V\_8O\_{21} + xLi^+ + xe^- \rightleftharpoons Li\_xK\_2V\_8O\_{21}} \tag{1}$$

Based on the previous reports on the voltage window and CV curves (**Figure 6A**), 1.5–4 V was chose as voltage window (Ni et al., 2015). **Figure 6B** showsthe typical charge-discharge voltage profiles of K2V8O<sup>21</sup> electrode and enumerates the 1st, 5th, 10th, 30th, and 50th cycles at a constant current density of 100 mA g −1 . There are two voltage plateaus in both charge and discharge curves as shown in **Figure 6B**. In discharge process, there are two discharge plateaus at around 2.5 and 2.8 V, respectively, which are consistent with the previous CV curves (**Figure 6A**) that is due to the lithium-ion intercalation process. These charge/discharge profiles in different cycles have shown similar shapes, which demonstrate the good structural stability of this material. It is worth noting that, although the capacity fades upon cycling, the overpotentials on both charging and discharging are reduced, this can compensate the loss in the energy density upon cycling. So the energy efficiency of the 5th, 10th, 15th, 30th, and 50th cycles at a constant current density of 100 mA g−<sup>1</sup> was calculated according to the following equation (2) (Eftekhari, 2017):

$$\text{Energy efficiency} = \frac{\text{energy density (discharge)}}{\text{energy density (charge)}} \times 100\% \qquad \text{(2)}$$

At various cycles of 5th, 10th, 15th, 30th, and 50th, the electrode exhibits energy efficiencies of 69.19, 71.15, 74.52, 77.55, and 78.02%, respectively. The improvement of Li<sup>+</sup> diffusion may cause the energy efficiency increase during the cycling process, which is beneficial to use K2V8O<sup>21</sup> as a cathode material. **Figure 6C** displays the cycling performance and coulombic efficiency of the electrospun K2V8O<sup>21</sup> and sol-gel K2V8O<sup>21</sup> at 50 mA g−<sup>1</sup> , respectively. An initial specific discharge capacity of 124.6 mA h g−<sup>1</sup> is obtained from ES-KVO electrode and then the capacity increased gradually to reach a maximum value of 200.2 mA h g−<sup>1</sup> at the 12th cycle. The activation process and the wetting of active electrode at the beginning may give rise to a capacity increase. After 100 cycles, the ES-KVO cathode still retains a discharge capacity of 169.2 mA h g−<sup>1</sup> with an average capacity fading rate of 0.037% per cycle based on the discharge capacity of 5th cycle. Furthermore, the coulombic efficiency can be maintained above 95%. For comparison, the SG-KVO electrode can only deliver a maximum specific discharge capacity of 183.6 mA h g−<sup>1</sup> and fade rapidly to <100 mA h g −1 . It can be demonstrated from the above results that the as-obtained electrospun K2V8O<sup>21</sup> electrode has shown a better cycling stability.

The long-term cycling performance of K2V8O<sup>21</sup> electrode is also examined. The long-term cycling performance and coulombic efficiency of electrospun K2V8O<sup>21</sup> and sol-gel K2V8O<sup>21</sup> electrode at a same current density of 500 mA g−<sup>1</sup> are shown in **Figure 6D**. Similar to the cycling performances performed at 50 mA g−<sup>1</sup> , the capacity of ES-KVO electrode increased in the initial several cycles at 500 mA g−<sup>1</sup> . This may be caused by the activation at the beginning and the improvement of lithium ion accessibility in the electrode materials during the cycling process. Although the initial discharge capacity of the cell is 120.8 mA h g−<sup>1</sup> , then the specific capacity increases slowly to reach a maximum value of 131.5 mA h g−<sup>1</sup> at 72th cycle. After 300 cycles, a discharge capacity of 108.3 mA h g−<sup>1</sup> still can be retained with a fading rate of only 0.043 % per cycle based on the capacity of 4th cycle. Besides, a high coulombic efficiency ∼98% can be reached throughout the cycling, which shows a good reversibility of this K2V8O<sup>21</sup> electrode at a high current rate. The initial coulombic efficiency of the K2V8O<sup>21</sup> cathode was 114.07 %, which may be caused by the partial removal of K<sup>+</sup> ions from the electrode materials during the charge process at initial cycle. An initial discharge capacity of SG-KVO electrode about 118.7 mA h g−<sup>1</sup> is obtained but then dropping rapidly for subsequent cycles. After 300 cycles, the sol-gel electrode can only deliver a low discharge capacity of 25.2 mA h g−<sup>1</sup> , showing a poor long-term cycling stability. The fork-like K2V8O<sup>21</sup> prepared by electrospinning has displayed superior specific capacity and long-term cycling stability compared to the SG-KVO, implying that it is a promising candidate for lithium-ion batteries as a cathode material. Moreover, Table S1 (Supplementary Material) summaries many reported electrochemical performance of metal vanadium oxides. Based on the comparison result, the reported fork-like K2V8O<sup>21</sup> in this work shows excellent cycle performance when it was used as a cathode material for LIBs.

As discussed above, the electrospun fork-like K2V8O<sup>21</sup> has demonstrated superior electrochemical performance, which may

### REFERENCES


be attributed to the following reasons: (1) the morphology of layer-by-layer stacked fork-like nanostructure is propitious to lithium-ion diffusion and has enlarged the contact area between electrode and electrolyte; (2) the large space between the fork-like K2V8O<sup>21</sup> could promote the diffusion of lithium-ion and buffer the volume change during the electrochemical reactions; (3) the single crystalline property of K2V8O<sup>21</sup> can offer convenient pathways for lithium ion diffusion to increase the efficiency of lithiation and delithiation.

### CONCLUSIONS

In summary, we have successfully prepared a single crystalline fork-like K2V8O<sup>21</sup> via facile electrospinning method followed by an annealing process in air at 500◦C for 2 h. The single crystalline fork-like K2V8O<sup>21</sup> has shown layer-bylayer stacked nanostructure with large space and conductive carbon through heat treatment of KVO precursor. Due to this advantageous feature, the fork-like K2V8O<sup>21</sup> demonstrates excellent electrochemical performances including high specific discharge capacity of 200.2 mA h g−<sup>1</sup> at a current density of 50 mA g−<sup>1</sup> with a good capacity retention (96.52%) after 100 cycles. Moreover, the electrodes demonstrate superior long-term cycling stability up to 300 cycles at a current density of 500 mA g −1 . The results from our work have demonstrated that the as-electrospun K2V8O<sup>21</sup> is a promising cathode candidate for next-generation high-performance LIBs.

### AUTHOR CONTRIBUTIONS

PH did the main experiment and write the manuscript. TZ involved in the discussion of the experiment and revised the manuscript. QS and JL did the SEM experiment. RC assisted the material synthesis. XC, YW, and AP made the research plan. AP also provided the financial support.

### ACKNOWLEDGMENTS

This work was supported by Natural Science Foundation of Hunan Province (2018JJ1036), China, and Innovation-driven Program of Central South University (2017CX001).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2018.00195/full#supplementary-material


Armand, M., and Tarascon, J.-M. (2008). Building better batteries. Nature 451, 652–657. doi: 10.1038/451652a


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Hao, Zhu, Su, Lin, Cui, Cao, Wang and Pan. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Cross-Linked Nanohybrid Polymer Electrolytes With POSS Cross-Linker for Solid-State Lithium Ion Batteries

Jinfang Zhang<sup>1</sup> \*, Xiaofeng Li <sup>1</sup> , Ying Li <sup>1</sup> , Huiqi Wang<sup>1</sup> , Cheng Ma<sup>2</sup> , Yanzhong Wang<sup>1</sup> , Shengliang Hu<sup>1</sup> and Weifeng Wei <sup>2</sup> \*

*<sup>1</sup> School of Materials Science and Engineering, North University of China, Taiyuan, China, <sup>2</sup> State Key Laboratory of Powder Metallurgy, Central South University, Changsha, China*

#### Edited by:

*Qiaobao Zhang, Xiamen University, China*

#### Reviewed by:

*Baihua Qu, Xiamen University, China Wei Luo, Tongji University, China Jun Jin, Shanghai Institute of Ceramics (CAS), China*

#### \*Correspondence:

*Jinfang Zhang zhangjf@nuc.edu.cn Weifeng Wei weifengwei@csu.edu.cn*

#### Specialty section:

*This article was submitted to Physical Chemistry and Chemical Physics, a section of the journal Frontiers in Chemistry*

> Received: *29 March 2018* Accepted: *07 May 2018* Published: *25 May 2018*

#### Citation:

*Zhang J, Li X, Li Y, Wang H, Ma C, Wang Y, Hu S and Wei W (2018) Cross-Linked Nanohybrid Polymer Electrolytes With POSS Cross-Linker for Solid-State Lithium Ion Batteries. Front. Chem. 6:186. doi: 10.3389/fchem.2018.00186* A new class of freestanding cross-linked hybrid polymer electrolytes (HPEs) with POSS as the cross-linker was prepared by a one-step free radical polymerization reaction. Octavinyl octasilsesquioxane (OV-POSS) with eight functional corner groups was used to provide cross-linking sites for the connection of polymer segments and the required mechanical strength to separate the cathode and anode. The unique cross-linked structure offers additional free volume for the motion of EO chains and provides fast and continuously interconnected ion-conducting channels along the nanoparticles/polymer matrix interface. The HPE exhibits the highest ionic conductivity of 1.39 × 10−<sup>3</sup> S cm−<sup>1</sup> , as well as excellent interfacial compatibility with the Li electrode at 80◦C. In particular, LiFePO4/Li cells based on the HPE deliver good rate capability and long-term cycling performance with an initial discharge capacity of 152.1 mAh g−<sup>1</sup> and a capacity retention ratio of 88% after 150 cycles with a current density of 0.5 C at 80◦C, demonstrating great potential application in high-performance LIBs at elevated temperatures.

Keywords: solid-state lithium ion batteries, cross-linked, polymer electrolyte, poly (ethylene oxide), POSS

## INTRODUCTION

Currently, lithium ion batteries (LIBs) have attracted intensive attention by virtue of the high output voltage, low self-discharge rate, large energy density, no memory effect, and environmental friendly nature (Dunn et al., 2011; Qu et al., 2014; Xu, 2014; Zhang et al., 2018). As an important component of LIBs, the electrolyte affects the specific energy, charge/discharge performance, cycle life, safety performance, and production cost of the battery (Zhou et al., 2014; Li et al., 2017; Fan et al., 2018). The leakage and explosive nature of traditional organic electrolyte solvents for commercial LIBs could cause excessive charge/discharge, short circuit and overheat, resulting in the fire or explosion safety concerns (Quartarone and Mustarelli, 2011; Wang et al., 2017; Qu et al., 2018). Polymer electrolytes (PEs) show great advantages in broadening the working temperature range, extending the service life, improving safety performance and flexibility of multifunctional structure and shape design (Ramesh and Ling, 2010; Samulski, 2011; Young et al., 2014). However, compared with liquid electrolyte, PEs exhibit low ionic conductivity at room temperature (<10−<sup>6</sup> S cm−<sup>1</sup> ) and poor interfacial compatibility with electrode materials, resulting in the electrochemical cycle degradation and successive capacity fading (Nakayama et al., 2010; Khurana et al., 2014). Therefore, high performance electrolyte materials have become an active research area for lithium-ion battery applications.

Nanoparticle-containing hybrid polymer electrolytes (HPEs) by grafting/blending nanofillers, such as ceramic particles or surface-modified nanoparticles, have been considered as promising candidates to improve the ionic conductivity, since nanoparticles could provide new pathways for lithium ion migration (Ahn et al., 2003; Khurana et al., 2014; Shim et al., 2014). Furthermore, nanoparticles could act as a protective layer against interfacial side reactions to enhance the interfacial compatibility between electrode and electrolyte materials (Shim et al., 2015; Ma et al., 2016). Among them, polyhedral oligomeric silsesquioxane (POSS), consisting of inorganic framework and organic functional groups, has attracted significant interest as an effective nanofiller to improve the ion conductivity owing to its unique multiple chain-ended structure and highly ordered nanoscale organic/inorganic hybrid structure (Kuo and Chang, 2011; Wang et al., 2011). Several studies have shown that POSSbased HPEs exhibit high ionic conductivity and thermal stability and improved dimensional stability and interfacial compatibility, which demonstrates the potential application in LIBs (Kim et al., 2012, 2013; Kwon et al., 2014; Shim et al., 2014; Wei et al., 2014; Villaluenga et al., 2015). For example, Kwon et al. prepared organic/inorganic hybrid semi-interpenetrating network (semi-IPN) polymer electrolytes (HIPEs) based on poly(ethylene oxideco-ethylene carbonate) (PEOEC) and POSS for all-solid-state lithium battery applications and the study showed that the HIPEs exhibit improvement in ionic conductivity along with enhanced dimensional stability due to the presence of the rigid and bulky POSS moiety (Kwon et al., 2014). Kim et al. have reported a series of organic/inorganic block and random hybrid polymer electrolytes containing POSS and PEG moieties, which exhibited enhanced ionic conductivity as well as dimensional stability due to the nanophase separation forming the ion-conducting channels (Kim et al., 2012).

Different from most of the reported polymer electrolytes with POSS as a nanofiller, here a new class of freestanding cross-linked HPE membranes with POSS as the cross-linker was prepared by a one-step free radical polymerization reaction based on our previous work (Zhang et al., 2016b). Octavinyl octasilsesquioxane (OV-POSS) with eight functional corner groups was used to provide cross-linker sites for the connection of polymer segment and the needed mechanical strength to separate the cathode and anode. Compared to PEGMEM/SMA polymer electrolyte, such cross-linked HPEs of PEGMEM/SMA-g-POSS with POSS as the cross-linker exhibit high ionic conductivity, stable interfacial compatibility and improved cycling stability and rate performance in solid-state LIBs.

### EXPERIMENTAL SECTION

### Materials

Poly(ethylene glycol) methyl ether methacrylate (PEGMEM, M<sup>n</sup> = 936 g mol−<sup>1</sup> ), stearyl methacrylate (SMA, M<sup>n</sup> = 338 g mol−<sup>1</sup> ), octavinyl octasilsesquioxane (OV-POSS, M<sup>n</sup> = 633 g mol−<sup>1</sup> ) and lithium bis(trifluoromethane sulfonamide) (LiTFSI) were purchased from Sigma-Aldrich. LiTFSI was stored in a desiccator and dried under vacuum condition at 120◦C for 12 h before used. Tetrahydrofuran (THF), 2,2-azobisisobutyronitrile (AIBN), ethyl acetate and petroleum ether, purchased from Sinopharm Chemical Reagent Co., Ltd., were used as received.

### Synthesis of PEGMEM/SMA-g-POSS Copolymer

Organic/inorganic hybrid copolymers of PEGMEM/SMA-g-POSS were synthesized via free radical polymerization, as shown in **Scheme 1**. PEGMEM (2.5704 g, 85 wt%), SMA (0.4536 mL, 15 wt%) and AIBN (0.5 wt% of total mass) were dissolved in 8 mL of ethyl acetate to obtain solution A. Various amounts of OV-POSS (0.0775 g, 0.1592 g, 0.2452 g, 0.336 g) were dissolved in 8 mL of THF to form solution B. Solutions A and B were mixed in a 50 mL three-neck flask equipped with a condenser and the mixture was heated to 70◦C in an oil bath under constant stirring and N<sup>2</sup> atmosphere. After 10 h, the resultant mixture was dissolved in THF and precipitated in petroleum ether for three times. The final product was dried at 60◦C for 12 h in vacuum oven and then the copolymers of PEGMEM/SMA-g-POSS were obtained.

### Prepare of the Hybrid Polymer Electrolytes (HPEs)

PEGMEM/SMA-g-POSS copolymers and LiTFSI with a desired [EO]/[Li] molar ratio were dissolved in THF and then poured into the Teflon mold with groove. The mixed solution was transferred to the fume hood and stood for 12 h at room temperature. The HPEs were obtained after the mixture was dried in vacuum oven at 100◦C for 12 h. And then the HPEs with crosslinked branching structure with POSS as the cross-linker were obtained, as shown in **Scheme 1**, and stored in glove box for further used. The thickness of polymer electrolyte membrane is around 200µm.

### Materials Characterizations and Electrochemical Measurements

<sup>1</sup>H NMR analysis (Avance III 400 MHz Digital NMR spectrometer) was used to characterize the structure of HCPs. Gel permeation chromatography (GPC) test (Waters 515–2410 instrument) was used to measure the molecular weight of HCP. Thermogravimetric analysis (TG) (TA-SDTQ600, 10◦C min−<sup>1</sup> under N<sup>2</sup> flow) was performed to assess the thermal stability. Field emission scanning electron microscopy (FE-SEM) analysis (Nova Nano SEM230) was carried out to investigate the microstructure of electrolytes.

Symmetric stainless steel/HPE/stainless steel cells were assembled to measure the ionic conductivity of HPEs using a PARSTAT 4000 system (Ametek Advanced Measurement Technology INC.) over the frequency range of 0.1 Hz−100 kHz from 25 to 80◦C with a perturbation voltage of 10 mV. Linear sweep voltammetry (LSV) was taken on stainless steel/HPE/Li cells to estimate the electrochemical stability of HPEs at a scan rate of 5 mV s−<sup>1</sup> . The symmetric Li/HPE/Li cells were fabricated to evaluate the interfacial stability between the electrolyte and lithium electrode. LiFePO4/HPE/Li coin cells were also assembled to investigate the electrochemical performance of HPEs in LIBs. Energy-type LiFePO4 was purchased from a reliable commercial source. The mixture of LiFePO<sup>4</sup> (80 wt%),

carbon black (10 wt%), and PVDF (10 wt%) was dispersed in Nmethyl-2pyrrolidone (NMP) to form a slurry. Subsequently, the resultant slurry was coated onto aluminum foil, and then dried under vacuum at 110◦C for 12 h to remove the residual NMP. The diameter of LiFePO<sup>4</sup> cathode, electrolyte membrane and Li metal electrode in the coin cell is 12, 18, and 16 mm, respectively. The galvanostatic charge/discharge tests were carried out in a battery testing system (LANHE CT2001A, Wuhan LAND electronics Co., PR China) between 2.5 and 3.7 V at different rates.

## RESULTS AND DISCUSSION

### Synthesis and Structural Characterization of Hybrid Copolymers (HCPs)

<sup>1</sup>H NMR spectrum of the copolymers with the POSS molar ratio of 5% (HCP-5) is shown in **Figure 1**. All the resonance peaks in **Figure 1** are assigned to each proton of the HCP-5. The signals of g at 3.41–4.32 ppm are assigned to the characteristic proton of CH2-CH2-O (EO) units in PEGMEM segments (Shim et al., 2014), while the resonance peak of e at 1.23 ppm is assigned to the characteristic proton of CH<sup>2</sup> units in SMA segments (Zhang et al., 2016b). The protons from the reacted vinyl protons of POSS could be observed at 1.92 ppm (a). However, all the eight vinyls of POSS do not fully participate in the polymerization reaction since the multiple resonance signals of i and j at 5.88–6.07 ppm are assigned to the unreacted vinyl protons of POSS (Xu et al., 2005). The <sup>1</sup>H NMR spectrum indicates that the HCPs combine the structural characteristic of PEGMEM, SMA and POSS. The POSS mole contents of HCPs can be estimated on the basis of <sup>1</sup>H NMR spectrum, as shown in following equation (Equation 1):

$$\begin{aligned} \text{POSS mol\%} &= \left[ (I\_{i+j}/3 + I\_a/2)/8 \right] / [(I\_{i+j}/3 + I\_a/2)/8 \\ &+ I\_{\text{\%}}/34\_{+}I\_{\text{\%}}/76] \end{aligned} \tag{1}$$

where Ii+<sup>j</sup> represents the integral area of the resonance peaks at 5.88–6.07 ppm from the unreacted vinyl protons of POSS, I<sup>a</sup> represents the integral area of the resonance peaks at 1.92 ppm from the reacted vinyl protons of POSS, I<sup>e</sup> represents the integral area of the resonance peak of protons from CH<sup>2</sup> units in SMA segments at 1.23 ppm and I<sup>g</sup> represents the integral area of the resonance peak from the protons of CH2-CH2-O (EO) units in PEGMEM segments at 3.41–4.32 ppm. The calculated values of various POSS mole contents are summarized in **Table 1**.

Gel permeation chromatography (GPC) was used to measure the molecular weight of HCPs and the results are summarized in **Table 1**. The molecular weights (Mw) of are in the range of 13.2– 14.8 × 10−<sup>3</sup> g mol−<sup>1</sup> and values of polydispersity index (PDI)

TABLE 1 | Synthesis results of HCPs with different POSS contents.


*<sup>a</sup>Feed ratio. <sup>b</sup>Estimated based on <sup>1</sup>H NMR spectroscopy. <sup>c</sup>Determined by GPC results.*

are ranging from 1.23 to 1.30, suggesting that the distribution of molecular weights for these HCPs was quite narrow.

### Morphology of HPEs

**Figure 2** shows the SEM image and EDX mapping results of HPEs. Red represents C element mainly from PEGMEM chains, green represents F element derived from LiTFSI, and blue represents Si segments derived from OV-POSS. It can be found that the HPEs exhibit a smooth surface and the C, F, and Si elements dispersed in HPEs evenly, demonstrating that the HPEs were prepared successfully.

### Thermal and Electrochemical Stabilities

Thermal and electrochemical stabilities are two important parameters used to estimate the safety of the electrolyte. As shown in the inset of **Figure 3A**, a colorless, transparent and freestanding electrolyte membrane was obtained. **Figure 3A** compares the TG results of the copolymer containing 5% POSS (HCP-5) and without POSS (PEGMEM/SMA) from 30 to 600◦C. The degradation temperature of HCP-5 is 298◦C while that of PEGMEM/SMA copolymer is 239◦C, indicating that the addition of POSS moieties could improve the thermal stability of the copolymers (Xu et al., 2005). HCP-5 has a char yield of 6.6% at 600◦C, which is higher than that of PEGMEM/SMA copolymer (2.3%) due to the combustion residues of POSS. The electrochemical window of the electrolyte based on HCP-5 is evaluated using LSV at 25, 80, and 120◦C, as shown in **Figure 3B**. The oxidative decomposition potentials of 5.62, 5.34, and 5.27 V vs. Li/Li<sup>+</sup> can be observed at 25, 80, and 120◦C, respectively, suggesting that the electrolyte is less stable at high temperatures. Even so, such electrolyte still exhibits a high stable electrochemical window (>5.0 V vs. Li/Li+) at high temperatures, demonstrating great potential in high-potential LIB applications.

### Conductivity of HPEs

Symmetric stainless steel/HPE/stainless steel cells were assembled to measure the ionic conductivity of HPEs using a PARSTAT 4000 system. The ionic conductivity was obtained by the following equation (Equation 2):

$$
\sigma = \mathcal{L}/\text{SR} \tag{2}
$$

where L and S were the thickness and area of the HPEs respectively, and R was the bulk resistance of the HPEs obtained from the Nyquist plot. The ionic conductivities of HPEs with various POSS contents (2.5 wt.% POSS, 5 wt.% POSS, 7.5 wt.% POSS, 10 wt.% POSS) at different LiTFSI concentrations were investigated, as shown in **Figure 4**. For all the samples, the ionic conductivity first increases with increasing the LiTFSI content, and then decreases with further increasing the lithium salt concentration. The maximum values of the conductivity were

observed for HPEs with a [EO]/[Li] ratio of 10/1. Both the mobility of EO segments and the number of charge carriers have

hybrid polymer electrolyte containing 5% POSS; (B) LSV curve of the electrolyte based on HCP-5 at 25, 80, and 120◦C.

effect on the ionic conductivity of the electrolyte (Dissanayake et al., 2003; Panday et al., 2009). The number of charger carriers plays an active role at low LiTFSI content, so the ionic conductivity increases with increasing the LiTFSI content since the addition of LiTFSI could produce more charge carriers (Zhang et al., 2016a). However, when the concentration of LiTFSI exceeds a certain threshold, the number of effective charge carriers begins to decrease with increasing LiTFSI content since it may be prone to forming ion pairs or ion aggregates. Meanwhile, the increase of intermolecular interactions between the lithium cations and the EO chains could also lead to a

decline in the mobility of EO chains (Singh and Bhat, 2003). As observed in other studies (Singh and Bhat, 2003; Ma et al., 2016; Zhang et al., 2016a,b), this phenomenon involves a tradeoff between the number of charge carriers and the mobility of EO segments.

**Figure 5A** shows the temperature-dependent ionic conductivity of HPEs with various POSS contents at [EO]/[Li] of 10/1. It is noted that the ionic conductivity increases with the addition of POSS, up to a maximum when the POSS content is 5 wt.%, and then decreases with further increase of the POSS content. The HPE containing 5 wt.% POSS exhibits the highest ionic conductivity of 3.94 × 10−<sup>5</sup> S cm−<sup>1</sup> at 25◦C and 1.39 × 10−<sup>3</sup> S cm−<sup>1</sup> at 80◦C, much higher than that of the PEGMEM/SMA polymer electrolyte (**Table 2**) and other PEO based solid polymer electrolytes (Table S1). As discussed in other studies (Croce et al., 1998; Zhu et al., 2014), the addition of hybrid particles to polymer matrix with an optimized concentration results in the highest conductivity. The introduction of the POSS hybrid particles significantly improves the ionic conductivity by disrupting the order of EO segments and increasing the free volume for Li<sup>+</sup> migration at low POSS content. Simultaneously, additional segmental motion TABLE 2 | Ionic conductivity at 25 and 80◦C and other parameters for the HPEs with various POSS contents at [EO]/[Li] of 10/1 derived from the VTF fit.


*<sup>a</sup>From our previous work (Zhang et al., 2016b).*

of EO chains grafted on POSS particles provide new pathways along the nanoparticles/polymer matrix interface for lithium ion transport, as illustrated in **Figure 6B**. However, when the POSS particles content exceeds 5 wt.%, the ionic conductivity begins to decrease since the bulky POSS groups cannot provide channels for Li<sup>+</sup> migration and the excess POSS nanoparticles will occupy the free volume, resulting in the decline of EO chain mobility.

Temperature-dependent ionic conductivity of HPEs obeys the Vogel–Tamman–Fulcher (VTF) equation (Equation 3):

$$
\sigma\_- = \mathcal{A} T^{-1/2} \exp \frac{-E\_a}{k(T - T\_0)} \tag{3}
$$

Where A is the pre-exponential factor, reflecting the number of charge carriers (Ratner et al., 2000); T is the testing temperature; E<sup>a</sup> is the activation energy; k is Boltzmann constant (8.314 × 10−<sup>3</sup> kJ mol−<sup>1</sup> K −1 ) and stands for is the equilibrium glass-transition temperature (Panday et al., 2009).

**Table 2** shows the ionic conductivity at 25 and 80◦C and other parameters for the HPEs with various POSS contents at [EO]/[Li] of 10/1 derived from the VTF fit. The highest value of A (3.64 S cm−<sup>1</sup> K 1/2 ) and conductivity (3.94 × 10−<sup>5</sup> S cm−<sup>1</sup> at 25◦C and 1.39 × 10−<sup>3</sup> S cm−<sup>1</sup> at 80◦C) are observed for the HPE containing 5 wt.% POSS, indicating that the introduction of POSS could facilitate the transmission of charge carriers by increasing free volume for the motion of EO chains. However, the ionic conductivity decreases with the further increase of POSS, since the bulky POSS segments act as a barrier to restrain the migration of charge carriers. The value of E<sup>a</sup> in **Table 2** is between 4.01 and 11.01 kJ mol−<sup>1</sup> , which is in agreement with the value for solid polymer electrolyte (Lin et al., 2013; Wu et al., 2014; Daigle et al., 2015). The lowest E<sup>a</sup> (4.01 KJ mol−<sup>1</sup> ) and T<sup>0</sup> (213 K) also could be observed for the HPE containing 5 % POSS. **Figure 5B** shows the comparison between the experimental data and VTF fitting results for HPE with 5 % POSS and PEGMEM/SMA copolymer and it could be observed that the experimental result of conductivity is definitely consistent with the fitting equation. Consequently, the HPE with 5% POSS content at [EO]/[Li] of 10/1 exhibits the highest conductivity and was used for the following electrochemical tests.

The improvement in ionic conductivity for the HPE with 5% POSS content can be attributed to the reduced crystallinity of the EO groups. **Figure 6A** shows the typical DSC analysis of the hybrid polymer electrolyte containing 5% POSS and without POSS (0% POSS). Various parameters of melting temperature (Tm), melting enthalpy (1Hm) and degree of crystallinity (Xc) obtained from DSC are summarized in Table S2. Notably, the hybrid polymer electrolyte containing 5% POSS possesses the lower degree of crystallinity, which could be attributed to the strong interactions between polymer segments and POSS nanoparticles. In contrast to PEGMEM/SMA polymer electrolyte, PEGMEM/SMA-g-POSS copolymer electrolyte exhibits unique cross-linked structure and the steric hindrance effect among POSS groups offers additional free volume for the motion of EO chains. Previous research in our laboratory has showed that the addition of POSS increase the amorphous fraction of the copolymer (Zhang et al., 2016a), and the unique cross-linked structure also facilitates the reduction of the crystallinity of EO segments (Ma et al., 2016). Meanwhile, the branches of SMA and PEGEME segments grafted on the surface of the POSS particles allow for additional segmental motion and provide low energy, fast and continuously interconnected ion-conducting channels along the nanoparticles/polymer matrix interface for Li<sup>+</sup> transport, resulting in higher ionic transmission efficiency, as schematically illustrated in **Figure 6B**.

### Interfacial Stability of HPE/Li Interface

Interfacial stability between polymer electrolyte and electrode materials is the key factor to affect the performance of lithium ion batteries. Polarization testing with a constant current density of 0.1 mA cm−<sup>2</sup> for a Li/Li symmetric cell was carried out to investigate the interfacial stability at 25 and 80◦C. As shown in **Figure 7**, the polarization voltage at 0.1 mA cm−<sup>2</sup> is ∼0.698 and 0.046 V at 25 and 80◦C, respectively. When the current density increases to 0.2 mA cm−<sup>2</sup> , polarization voltage presents ∼1.288 and 0.144 V at 25 and 80◦C, respectively. No obvious fluctuations could be observed during the whole cycling process, demonstrating that such HPE exhibits stable interfacial compatibility with lithium metal electrode and shows great potential for application in high performance LIBs.

mechanism designed for the cross-linked HPEs.

### Battery Performance

LiFePO4/Li coin cells using HPEs with 5 wt.% POSS content were assembled for evaluating the battery performance. **Figure 8A** shows the discharge curves for LiFePO4/Li coin cell under various C-rates with a charge current density of 0.1 C at 80◦C. Discharge capacities of 154 mAh g−<sup>1</sup> were achieved at 0.1 C and ∼140 mAh g−<sup>1</sup> could be observed at 0.2, 0.5, 1, and 2 C, and the flat potential plateaus could also be observed up to 2 C rate. Even at higher discharg rates of 5 and 10 C, the discharge capacities of 99 and 61 mAh g−<sup>1</sup> could still be obtained, respectively. **Figure 8B** presents a comparison of charge and discharge performance under different circulatory rates. The capacity decreases when the current density increases from 0.2 C to 5 C, and then recovers to its original value when the current density back to 0.2 C. **Figure 8C** presents the long-term cycling performance of LiFePO4/Li battery assembled using HPEs with 5 wt.% POSS content with a current density of 0.5 C at 80◦C. Note that the cell delivers an initial discharge capacity of 152.1 mAh g−<sup>1</sup> with a capacity retention ratio of 88% after 150 cycles and the coulombic efficiency of ∼99% could be observed in the charge/discharge process, which is higher than other PEO-based solid polymer electrolyte (Table S1). The results indicate that the cells display great potential application in high-performance LIBs at elevated temperatures.

### CONCLUSION

A new class of freestanding cross-linked HPEs with POSS as the cross-linker was prepared by a one-step free radical polymerization reaction. PEGMEM/SMA-g-POSS HPEs exhibit unique cross-linked structure and provide low energy, fast and continuously interconnected ion-conducting channels along the nanoparticles/polymer matrix interface. Compared to PEGMEM/SMA polymer electrolyte, such cross-linked HPEs exhibit enhanced thermal and electrochemical stabilities, improved ionic conductivity and interfacial compatibility. Moreover, LiFePO4/Li cells assembled using HPEs with 5 wt.% POSS content delivers not only good rate capability but also enhanced long-term cycling performance at 80◦C, displaying

great potential application in high-performance LIBs at elevated temperatures.

## AUTHOR CONTRIBUTIONS

JZ and CM carried out the experimental work and the data collection and interpretation. XL participated in the design and coordination of experimental work, and acquisition of data. YL and HW participated in the study design, data collection, analysis of data and preparation of the manuscript. YW, SH, and WW carried out the study design, the analysis and interpretation of data and drafted the manuscript. All authors read and approved the final manuscript.

### REFERENCES


### ACKNOWLEDGMENTS

The authors would like to acknowledge financial support from Shanxi Province Science Foundation for Youths (No. 201701D22108, No. 201701D221085), National Natural Science Foundation of China (21703209) and North University of China Fund for Scientific Innovation Team.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2018.00186/full#supplementary-material

based on comb-like copolymers. J. Power Sources 279, 372–383. doi: 10.1016/j.jpowsour.2014.12.061


their application to solid-state electrolytes for high-temperature lithium-ion batteries. Polym. Chem. 5, 3432–3442. doi: 10.1039/C4PY00123K


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The reviewer, BQ, and handling Editor declared their shared affiliation.

Copyright © 2018 Zhang, Li, Li, Wang, Ma, Wang, Hu and Wei. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Reduced Graphene Oxide Decorated Na3V2(PO4)<sup>3</sup> Microspheres as Cathode Material With Advanced Sodium Storage Performance

Hezhang Chen<sup>1</sup> , Yingde Huang<sup>1</sup> , Gaoqiang Mao<sup>1</sup> , Hui Tong<sup>1</sup> \*, Wanjing Yu<sup>1</sup> , Junchao Zheng<sup>1</sup> and Zhiying Ding<sup>2</sup>

*<sup>1</sup> School of Metallurgy and Environment, Central South University, Changsha, China, <sup>2</sup> School of Chemistry and Chemical Engineering, Central South University, Changsha, China*

Reduced graphene oxide (rGO) sheet decorated Na3V2(PO4)<sup>3</sup> (NVP) microspheres

were successfully synthesized by spray-drying method. The NVP microspheres were embedded by rGO sheets, and the surface of the particles were coated by rGO sheets and amorphous carbon. Thus, the carbon conductive network consisted of rGO sheets and amorphous carbon generated in the cathode material. NVP microspheres decorated with different content of rGO (about 0, 4, 8, and 12 wt%) were investigated in this study. The electrochemical performance of NVP exhibited a significant enhancement after rGO introduction. The electrode containing about 8 wt% rGO (NVP/G8) showed the best rate and cycle performance. NVP/G8 electrode exhibited the discharge capacity of 64.0 mAh g <sup>−</sup><sup>1</sup> at 70◦C, and achieved high capacity retention of 95.5% after cycling at 10◦C for 100 cycles. The polarization of the electrode was inhibited by the introduction of rGO sheets. Meanwhile, compared with the pristine NVP electrode, NVP/G8 electrode exhibited small resistance and high diffusion coefficient of sodium ions.

\*Correspondence: *Hui Tong huitong@csu.edu.cn*

*Technology, China*

*Technology, Hong Kong*

Edited by: *Kaili Zhang,*

*Hong Kong* Reviewed by: *Haoran Jiang,*

*Lingjun Li,*

*City University of Hong Kong,*

*Hong Kong University of Science and*

*Changsha University of Science and*

#### Specialty section:

*This article was submitted to Physical Chemistry and Chemical Physics, a section of the journal Frontiers in Chemistry*

> Received: *29 March 2018* Accepted: *30 April 2018* Published: *23 May 2018*

#### Citation:

*Chen H, Huang Y, Mao G, Tong H, Yu W, Zheng J and Ding Z (2018) Reduced Graphene Oxide Decorated Na*3*V*2*(PO*4*)*3 *Microspheres as Cathode Material With Advanced Sodium Storage Performance. Front. Chem. 6:174. doi: 10.3389/fchem.2018.00174* Keywords: Na3V2 (PO4 )3 , microspheres, reduced graphene oxide, amorphous carbon, sodium ion batteries

### INTRODUCTION

With the exhaustion of the traditional resources and the increasing environmental problems, researchers expect to develop new clean energy and related storage systems (Xiao et al., 2015; Yang et al., 2016; Luo et al., 2017; Yang Z. H. et al., 2017; Zheng et al., 2017). Lithium ion batteries (LIBs) have being used as the energy storage systems in new energy vehicles. However, the lithium resource cannot meet the requirement of the increasing market. Sodium ion batteries (SIBs) attract the attention of the researchers. Sodium possesses similar chemical properties as lithium. Furthermore, sodium resource is abundant on earth and the price is cheap. Thus, SIBs are considered as the most promising competitive alternative to LIBs (Jian et al., 2014; Wang H. et al., 2015; Chao et al., 2016; Xu et al., 2016). Developing SIBs with excellent electrochemical performance becomes necessary and challenging.

The cathode materials are the key restraining factor of the electrochemical performance in LIBs and SIBs. The electrochemical performances of the batteries, such as energy density, cycle and rate performance, mainly depend on the properties of cathode materials. In SIBs, the diffusion of sodium ions is more difficult compared with that of lithium ions due to the relative larger sodium ion radius (Li et al., 2015). In this case, large volume change during the sodium ion insertion/extraction process leads to structure collapse of the cathode materials, causing poor rate and cycle performance. Various cathode materials, such as Na0.44MnO<sup>2</sup> (Sauvage et al., 2007), Na4Fe(CN)<sup>6</sup> (Qian et al., 2012), Na2CoP2O<sup>7</sup> (Barpanda et al., 2013), and NaFeF<sup>3</sup> (Yamada et al., 2011), had been investigated for SIBs to improve the cycle performance and rate capability (Fang et al., 2015; Wang J. et al., 2015; He et al., 2016; Zhang Y. et al., 2016; Zhang Z. et al., 2016). In the reported cathode materials, Na3V2(PO4)<sup>3</sup> (NVP) is considered as one of the most promising cathode materials for SIBs (Zhang et al., 2017a; Chen et al., 2018). NVP has a NASICON framework and exhibits high energy density, high operating voltage and excellent thermal stability. Unfortunately, the poor electronic conductivity of the phosphate salt electrode materials also leads to bad rate capacity and cycle performance (Liang et al., 2017). Many methods are applied to improve the electrochemical performance of NVP cathodes, such as cation doping, particle size reduction and conductive materials coating (Qu et al., 2014; Li et al., 2015; Xu et al., 2016; Liang et al., 2017). Among these methods, carbon coating and particle size reduction are the most effective methods. The sodium ion diffusion path can be shortened by diminishing particle size, and the volume change during the charge and discharge process was also reduced (Lin et al., 2013; Qu et al., 2014; Xiao et al., 2014). The electronic conductivity of NVP material can be enhanced effectively by carbon coating (Li et al., 2014). Graphene belongs to carbon conductive materials, which owns superior electronic conductivity and high surface area (Zhang et al., 2014; Li et al., 2016; Yang C. et al., 2017). The electronic conductivity of cathode materials can be obviously enhanced through grapheme coating. Graphene is also used as a template to diminish the particle size (Fang et al., 2016; Xu et al., 2016). Herein, we report a novel NVP microsphere cathode material. Reduced graphene oxide (rGO) decorated NVP (NVP/rGO) microsphere materials were synthesized through spray drying method. In this structure, the NVP microspheres were coated and embedded by rGO sheets. The rGO sheets and amorphous carbon formed carbon conductive network, which improve the electronic conductivity. Consequently, the electrochemical performances of NVP/rGO were improved significantly. To explore the as-prepared NVP/rGO materials, the affecting factors on electrochemical properties, such as morphology, crystal structure and resistance, were systematically investigated.

### MATERIALS AND METHODS

In this study, rGO sheets were prepared by Hummers methods from nature flake graphite. NVP/rGO microspheres were prepared by spray-drying method. **Figure 1** illustrates the synthetic process of NVP/rGO microspheres. In a typical experiment, 15 mmol NaHCO3, 10 mmol NH4VO3, 15 mmol NH4H2PO4, and 30 mmol citric acid were dissolved together in the deionized water by stirring. Then, the solution was heated to 80◦C and kept for about 30 min. After that, the solution was cooled down to room temperature, and it was mixed with the suspension of graphene oxide (GO) by ultrasound and stirring. The obtained slurry was spray dried at inlet and outlet temperature kept around at 250 and 150◦C to get precursor powder. Then, the precursor was calcined at 800◦C for almost 10 h in the inert atmosphere. The reduction reaction from GO to rGO occurred simultaneously, and finally the NVP/rGO materials were synthesized. NVP/rGO samples with different content (about 4, 8, and 12 wt%) of rGO was labeled as NVP/G4, NVP/G8, and NVP/G12, respectively. For comparison, NVP without rGO (labeled as NVP/C) was prepared through the same way.

The crystal phases of the prepared materials were studied by X-ray diffraction (XRD, Rigaku D/Max 200PC, Japan) using Cu Kα radiation, with the scanning range of 10–60◦ . Thermal gravimetric analysis (TGA) of the samples was obtained by a SDT Q600TGDTA apparatus from room temperature to 650◦C in air, with the heating rate of 5◦C min−<sup>1</sup> . The Raman spectra were investigated by Raman spectroscopy (HORIBA JOBIN YVON, France). The chemical valence states of the carbon were studied by X-ray photoelectron spectroscopy (XPS, ESCALAB 250Xi, Thermo Fisher Scientific Co., Ltd, USA). The sample morphologies were observed by scanning electron microscopy (SEM, Nova NanoSEM-230, USA), and the microstructures were observed by high-resolution transmission electron microscopy (HRTEM, FEI Tecnai G2 F20 S-Twin, USA), working at 200 kV. The prepared NVP samples were evaluated using coin cells fabricated in a glovebox, with sodium foil as the counter electrode and a glass microfiber (Whatman GF/D) as the separator. The sample powder, carbon black and polyvinylidene fluoride (mass ratio 8:1:1) were mixed in N-methylpyrrolidinone to get the slurry. The aluminum foil pasted with the slurry was dried in the oven, 120◦C for 4 h. The electrolyte was 1 M NaClO<sup>4</sup> dissolved in ethylene carbon and dimethyl carbonate (1:1), and 5 vol% fluoroethylene carbonate. Electrochemical measurements were performed by a constant current/constant voltage method, with the potential of 2.0–3.8 V (vs. Na+/Na). Electrochemical impedance spectroscopy (EIS) tests were carried out with an electrochemical workstation (CHI 660D, CH Instruments, China), with the frequency range of 10−<sup>2</sup> -10<sup>5</sup> Hz. Cyclic voltammetry (CV) measurements were also studied with this workstation, and the scan rate of which was 0.1 mV s−<sup>1</sup> .

### RESULTS AND DISCUSSION

NVP/rGO samples were prepared by spray-drying method. To investigate the effect of rGO sheets on the structure, the microspheres were characterized by XRD. The XRD patterns of NVP/rGO samples were shown in **Figure 2A**. All diffraction peaks are indexed to the R3c space group of the rhombohedral NASICON structure without any impurities. It indicates that the crystal structure of NVP was not changed by rGO embedding. The carbon contents of NVP/rGO samples were measured with carbon-sulfur analyzer, which were 1.00 (NVP/C), 4.53 (NVP/G4), 7.11 (NVP/G8), and 10.11 (NVP/G12) wt%, respectively. The carbon of NVP/C derived from the decomposition of citric acid (Chen et al., 2018). The content of the NVP/G8 was also confirmed by TGA test (**Figure 2B**), and

carbon content was 7.02 wt%, which is close to the result of carbon-sulfur analyzer.

Raman spectroscopy was used to investigate the characteristics of the carbon materials. **Figure 2C** displays the Raman spectra of NVP/rGO samples. The D (disorder-induced phonon mode) and G (graphite band) bands were observed around 1,300 and 1,500 cm−<sup>1</sup> for all samples, respectively. The relative intensity ratio (ID/IG) indicates the degree of structural disorder of the carbon materials. The ID/I<sup>G</sup> ratio values of NVP/C, NVP/G4, NVP/G8, and NVP/G12 samples were 1.02, 0.95, 0.93, and 0.90, respectively. It is found that the ID/I<sup>G</sup> ratio values decreased with the increase of rGO sheet content in the samples, and the ID/I<sup>G</sup> ratio value of NVP/G8 material was the lowest. The lower ID/I<sup>G</sup> ratio value suggests the graphitization degree of the material is higher, and the electric conductivity of the electrode materials is also higher (Xu et al., 2016). **Figure 2D** shows the C1s XPS analysis spectrum of NVP/G8 material. The fitted curve was resolved into three peaks (blue, red, and black), and the binding energies of 284.5, 285.7, and 286.5 ev corresponds to C-C, C-O, and C=O functional groups, respectively (Fang et al., 2016).

**Figure 3** shows the SEM images of NVP/rGO materials. It is seen that all the NVP/rGO particles of the samples were spherical, and the size distribution was from 0.2 to 1µm. It is found that there are some secondary particles on the surface of the microsphere in **Figure 3a**, however, the secondary particles are hardly seen in **Figures 3b–d**, as the rough surface of the microspheres were covered with rGO sheets. The inner structure of NVP/rGO microspheres was investigated by TEM and HRTEM. **Figures 4a,b** shows the low

FIGURE 4 | (a,b) TEM and (c) HRTEM images of NVP/G8 sample.

magnification TEM image of NVP/G8 microsphere. It is found that the surface of the small particles was covered with rGO sheets, and which were also embedded in the microsphere. The secondary particle size of NVP/G8 is smaller than 100 nm, as the microsphere was separated by rGO sheets and the particle growth was controlled. The HRTEM image (**Figure 4c**) shows that the crystal lattice spacing of NVP was about 0.285 nm, corresponding to (211) plane of NVP. The functional groups of GO, such as -COOH and -OH, reacted with Na<sup>+</sup> and PO3<sup>−</sup> 4 to make the electron transport easily. In the TEM images, it is further confirmed that rGO sheets distributed uniformly in the microspheres. The carbon conductive network consisted of rGO sheets and amorphous carbon creates good interfacial contact among NVP particles, which is benefit for accelerating the electron transport in the electrode material. The electrochemical performances of all the samples were also investigated, as shown in **Figure 5**. **Figure 5A** shows the charge and discharge curves of NVP electrodes with different content of rGO sheets, the charge and discharge rates were 0.2 and 1 C (1 C = 110 mA g−<sup>1</sup> ), respectively. The discharge capacity of NVP/C, NVP/G4, NVP/G8, and NVP/G12 were 105.6, 102.1, 100.7, and 96.7 mAh g−<sup>1</sup> , respectively. The discharge capacity decreased with the increase of the rGO content, as rGO did not involve the electrochemical reaction. The inset figure in **Figure 5A** shows the rate performances of all the samples. All the NVP/rGO samples showed better rate performances than

NVP/C sample. And, the NVP/G8 sample had the best rate performance. It suggests that the rate performance is enhanced due to the introduction of rGO sheets. For investigating the effect of rGO sheets on intercalation/deintercalation behavior of sodium ions, the NVP/C and NVP/G8 electrodes were evaluated by CV, as displayed in **Figure 5B**. The CV curves of the samples were similar. The redox couple located at about 3.15 and 3.50 V, which are attributed to the V4+/V3<sup>+</sup> redox couple (Liang et al., 2017). The potential differences of the redox peaks of NVP/C and NVP/G8 electrodes were 0.25 and 0.19 V, respectively. The lower difference between the oxidation and the reduction peaks suggests the smaller degree of the electrode polarization. Furthermore, the redox peaks of NVP/G8 electrode became sharper. It indicates that the introduction of rGO sheets can decrease the polarization. The galvanostatic charge and discharge process of the NVP/C and NVP/G8 electrodes was

studied at different rates and shown in **Figures 5C,D**. It is seen that the NVP/G8 electrode delivered capacities of 102.2, 100.7, 98.1, 90.8, 85.8, 82.3, 78.5, and 72.0 mAh g−<sup>1</sup> at 0.2, 1, 5, 20, 30, 40, 50, and 60◦C, respectively; even at 70◦C, the electrode still had discharge capacity of 64 mAh g−<sup>1</sup> . As a contrast, NVP/C electrode delivered capacities of 112.3, 105.6, 98.8, 92.6, 87.0, 80.0, 73.1, and 66.8 mAh g−<sup>1</sup> at 0.2, 1, 5, 10, 15, 20, 25, and 30◦C, respectively. Although NVP/C electrode delivered high capacity at low rate, it did not work at high rate (above 30◦C). It is concluded that the potential polarization of NVP/G8 is smaller than the NVP/C, resulting in the improvement of discharge capacities. **Figure 5E** shows the rate properties of NVP/C and NVP/G8 electrodes. The initial discharge capacity of NVP/G8 was lower than that of NVP/C at low rate, as rGO has no contribution to the capacity. When the current density was higher than 5◦C, NVP/G8 electrode showed higher discharge capacity than NVP/C electrode. With the increase in discharge current density, the capacity of NVP/C decreased quickly. In contrast, the capacity of NVP/G8 decreased slowly. Even the rate up to 70◦C, the capacity retention still kept at 62.6% (relative to the capacity at 0.2◦C). When the rate decreased to 0.2 C, the discharge capacity of NVP/G8 electrode almost increased to the value of initial capacity. The improvement of rate performance implies that carbon network can enhance the electronic conductivity of the NVP materials. To further investigate the electrochemical performance of NVP samples, the cycle performances of NVP/C and NVP/G8 were evaluated at 10◦C (**Figure 5F**). The initial discharge capacities of NVP/C and NVP/G8 at 10◦C were 92 and 95.5 mAh g−<sup>1</sup> , respectively; after 100 cycles, which decreased to 71.6 and 91.0 mAh g−<sup>1</sup> , with the capacity retentions of 77.8 and 95.5%, respectively. It is demonstrated that the cycle performance was obviously improved by the addition of rGO sheets.

For understanding the effect of rGO introduction on electrode reaction kinetics, the NVP/C and NVP/G8 electrodes were investigated by EIS. **Figure 6A** illustrates Nyquist plots of the two electrodes. Each plot consists of a semicircle and a straight line at high-middle and low frequency area, respectively. The intercept at high frequency area corresponds to the ohmic resistance of the electrode and electrolyte. The semicircle corresponds

TABLE 1 | Impedance parameters of NVP/C and NVP/G8 electrodes obtained from equivalent circuit fitting.


to charge-transfer resistance, while the straight line related to sodium ion diffusion in the NVP particles. The equivalent circuit model is shown in the inset figure of **Figure 6A**. In this model, R<sup>s</sup> corresponds to the ohmic resistance, Rct corresponds to the charge-transfer resistance, CPE and W represents the doublelayer capacitance and Warburg impedance with regard to sodium ion diffusion in the NVP particles. The impedance parameters are listed in **Table 1**. It is found that R<sup>s</sup> and Rct of NVP/G8 electrode were much smaller than those of NVP/C electrode. It suggests that the charge-transfer speed through the electrode and electrolyte interface is increased by rGO introduction. Furthermore, the diffusion coefficient of sodium ions (DNa+) was calculated by the equations as follows (Zhang et al., 2015a,b, 2017b):

$$\mathbf{D}\_{\mathrm{Na^{+}}} = \frac{\mathbf{R}^{2}\mathbf{T}^{2}}{2\mathbf{A}^{2}\mathbf{n}^{4}\mathbf{F}^{4}\mathbf{C}^{2}\sigma^{2}} \tag{1}$$

$$Z\_{\rm real} = \mathcal{R}\_s + \mathcal{R}\_{\rm ct} + \sigma \alpha^{-1/2} \tag{2}$$

where R represents the gas constant, T is the absolute temperature, A is the surface area of the cathode, n signifies the number of transferred electrons, F is the Faraday constant, C is the concentration of sodium ions, σ is the Warburg coefficient, and ω is the angular frequency in the low frequency region. The linear fitting of Z ′ versus ω −1/2 is shown in **Figure 6B**, in which the value of σ is the slope. As presented in **Table 1**, DNa<sup>+</sup> of the NVP/G8 and NVP/C electrode was 1.24 × 10−<sup>13</sup> and 8.72 × 10−<sup>14</sup> cm<sup>2</sup> s −1 , respectively. The NVP/G8 electrode showed higher value of sodium ion diffusion coefficient, as the sodium ion diffusion path was reduced in the small particles and the contact area between NVP and the electrolyte increased. The R<sup>s</sup> and Rct of NVP/G8 electrode both became smaller, as the electron conductivity was enhanced by rGO sheets. Therefore, lower R<sup>s</sup> and Rct, and higher DNa<sup>+</sup> are benefit for improving the electrochemical performance.

### CONCLUSION

In summary, NVP microspheres decorated with different content of rGO sheets were synthesized by spray-drying method. GO plays as a binder during the spray-drying process. The carbon network creates good interfacial contacts among NVP particles and accelerates the electron transport. By introduction of rGO sheets into NVP microspheres, the electrochemical properties of NVP/rGO materials were all obviously enhanced, especially at high rates. NVP/G8 electrode exhibited the capacities of 102.2 mAh g−<sup>1</sup> at 0.2◦C, and 64 mAh g−<sup>1</sup> at 70◦C. Even cycled at 10◦C for 100 cycles, the capacity retention of NVP/G8 electrode remained 95.5%. The significant improvement in electrochemical performance of NVP sample is attributed

### REFERENCES


to the lower resistance and higher DNa<sup>+</sup> compared with those of the pristine sample. Consequently, NVP microspheres decorated with rGO sheets may be a promising cathode material to achieve superior electrochemical performance for energy storage.

### AUTHOR CONTRIBUTIONS

HC carried out the experiment and wrote the manuscript; YH and GM participated in the material preparation; HT supervised all the experiments and proofread the manuscript; WY, JZ, and ZD contributed to the discussion.

### ACKNOWLEDGMENTS

This work was supported by National Natural Science Foundation of China (Grant No. 51502350 and 51772334), China Postdoctoral Science Foundation (Grant No. 2016M592447), and The International Postdoctoral Exchange Fellowship Program (Grant No. 155212).


for sodium-ion batteries. Energy Environ. Sci. 10, 107–113. doi: 10.1039/C6EE 03173K


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Chen, Huang, Mao, Tong, Yu, Zheng and Ding. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Comparative Investigation of 0.5Li2MnO3·0.5LiNi0.5Co0.2Mn0.3O<sup>2</sup> Cathode Materials Synthesized by Using Different Lithium Sources

Peng-Bo Wang1†, Ming-Zeng Luo1,2†, Jun-Chao Zheng<sup>1</sup> \*, Zhen-Jiang He<sup>1</sup> , Hui Tong<sup>1</sup> and Wan-jing Yu<sup>1</sup>

*<sup>1</sup> School of Metallurgy and Environment, Central South University, Changsha, China, <sup>2</sup> College of Chemistry and Chemical Engineering of Xiamen University, Xiamen, China*

Edited by: *Qiaobao Zhang, Xiamen University, China*

### Reviewed by:

*Bo Chen, Nanyang Technological University, Singapore Huixin Chen, Xiamen Institute of Rare Earth Materials, China Xifei Li, Xi'an University of Technology, China*

\*Correspondence:

*Jun-Chao Zheng jczheng@csu.edu.cn*

*†These authors have contributed equally to this work and co-first authors.*

#### Specialty section:

*This article was submitted to Physical Chemistry and Chemical Physics, a section of the journal Frontiers in Chemistry*

> Received: *13 March 2018* Accepted: *20 April 2018* Published: *15 May 2018*

#### Citation:

*Wang P-B, Luo M-Z, Zheng J-C, He Z-J, Tong H and Yu W (2018) Comparative Investigation of 0.5Li2MnO*3·*0.5LiNi0.5Co0.2Mn0.3O<sup>2</sup> Cathode Materials Synthesized by Using Different Lithium Sources Front. Chem. 6:159. doi: 10.3389/fchem.2018.00159* Lithium-rich manganese-based cathode materials has been attracted enormous interests as one of the most promising candidates of cathode materials for next-generation lithium ion batteries because of its high theoretic capacity and low cost. In this study, 0.5Li2MnO3·0.5LiNi0.5Co0.2Mn0.3O<sup>2</sup> materials are synthesized through a solid-state reaction by using different lithium sources, and the synthesis process and the reaction mechanism are investigated in detail. The morphology, structure, and electrochemical performances of the material synthesized by using LiOH·H2O, Li2CO3, and CH3COOLi·2H2O have been analyzed by using Thermo gravimetric analysis (TGA), X-ray diffraction (XRD), Scanning electron microscope (SEM), Transmission electron microscope (TEM), X-ray photoelectron spectroscopy (XPS), and electrochemical measurements. The 0.5Li2MnO3·0.5LiNi0.5Co0.2Mn0.3O<sup>2</sup> material prepared by using LiOH·H2O displays uniform morphology with nano particle and stable layer structure so that it suppresses the first cycle irreversible reaction and structure transfer, and it delivers the best electrochemical performance. The results indicate that LiOH·H2O is the best choice for the synthesis of the 0.5Li2MnO3·0.5LiNi0.5Co0.2Mn0.3O<sup>2</sup> material.

Keywords: lithium-ion battery, lithium-rich manganese-based cathode materials, lithium sources, solid reaction, electrochemical performances

### INTRODUCTION

Rechargeable Li-ion batteries (LIBs) play a dominant role in energy storage devices of portable electronic devices and electric vehicles (EVs) for its excellent safety, long cycle life, and high energy density (Armand and Tarascon, 2008; Goodenough and Park, 2013; Zhang Q. et al., 2015, 2016; Li et al., 2017a,b; Zhang et al., 2018). However, the specific energy density of LIBs still cannot meet the demand of EVs owing to lower energy density of cathode material (Whittingham, 2014). At present, classic cathode materials, such as LiFePO<sup>4</sup> (Zheng et al., 2008, 2017), LiMn2O<sup>4</sup> (Kim et al., 2008), LiCoO<sup>2</sup> (Kang et al., 1999), and LiNiaMnbCo1−a−bO<sup>2</sup> (Li et al., 2017) etc., offer a reversible discharge capacity less than 200mAh g−<sup>1</sup> . Recently, lithium-rich manganese-based cathode materials has been attracted enormous interests as a promising cathode material for nextgeneration LIBs because of high discharge capacity (more than 250mAh g−<sup>1</sup> ) and low cost (Ohzuku et al., 2011).

Lithium-rich manganese-based cathode materials contain double component: one phase of Li2MnO<sup>3</sup> with C2/m space group and the other phase of LiMO<sup>2</sup> with R-3m space group. Because Mn4<sup>+</sup> in Li2MnO<sup>3</sup> phase cannot be oxidized any more, it possesses electrochemically inert. However, Li<sup>+</sup> and O<sup>2</sup> can be extracted from the TM (transition metal) layer and the lattice, respectively, which indicates that the Li2MnO<sup>3</sup> phase is activated by initial charging process and forms an irreversible loss of Li2O (Yabuuchi et al., 2011). It is demonstrated that the high capacity of the materials originates from the oxygen escape from Li2MnO<sup>3</sup> phase at high voltage (Yabuuchi et al., 2011). In addition, there are so many fatal disadvantages in lithium-rich manganesebased cathode materials, such as severe voltage fading during cycling (Zheng J. et al., 2015; Zhang T. et al., 2016) poor rate performance (Fan et al., 2015; Rozier and Tarascon, 2015; Zhang K. et al., 2015), large initial irreversible capacity, and low initial coulombic efficiency (Bai et al., 2015). Presently, many methods like comprising doping (Dianat et al., 2013; Wang et al., 2013; Li et al., 2014; Zhang H. et al., 2014), coating (Shi et al., 2012, 2013; Gu et al., 2013; Zhang et al., 2013; Zhou et al., 2017; Liu et al., 2018), nano crystallization (Wang et al., 2010), and morphology control (Yang et al., 2013; Remith and Kalaiselvi, 2014) have been proposed to promote the electrochemical performance of the materials.

Generally speaking, the compositions of this kind of material are varied. According to the chemical formula of lithium-rich manganese-based cathode materials, it can be simply written as xLi2MnO3·(1-x)LiMO2(M = Ni, Co, Mn, Ti, Fe, etc.; Ohzuku et al., 2011; Yabuuchi et al., 2011; Bai et al., 2015; Fan et al., 2015; Rozier and Tarascon, 2015; Zheng J. et al., 2015; Zhang T. et al., 2016). Common and representative components of lithium-rich manganesebased materials are 0.5Li2MnO3·0.5LiNi0.5Mn0.5O<sup>2</sup> and 0.5Li2MnO3·0.5LiNi1/3Co1/3Mn1/3O<sup>2</sup> (Shi et al., 2012, 2013; Dianat et al., 2013; Gu et al., 2013; Wang et al., 2013; Yang et al., 2013; Zhang et al., 2013; Zhang H. et al., 2014; Li et al., 2014; Zhou et al., 2017).

To the best of our knowledge, the presence and content of nickel that improves the cathode capacity (Sun et al., 2005). Zheng Z. et al. (2015) discussed that the roles and the functions of nickel in electrochemical cycling of lithiumrich Mn-based cathode materials. Yang et al. (2017) reported that the Ni substitution at 2c sites not only enhances oxygen stability and delays oxygen loss from the lattice but also suppresses the cation mixing that induces the undesired phase transition. Gao et al. (2017) demonstrated that Li-rich material Li1.2(Ni0.25Co0.25Mn0.5)0.8O<sup>2</sup> was prepared by a novel coreshell structure, in which Ni element acts as stabilizing ions to inhibit the Jahn-Teller effect of active Mn3+. Based on the above considerations, lithium-rich manganese-based material 0.5Li2MnO3·0.5LiNi0.5Co0.2Mn0.3O<sup>2</sup> is designed for the first time.

In fact, the cathode materials with different morphology and electrochemical performance can be achieved by using different lithium sources (Zhang B. et al., 2014; Cao et al., 2017). In this work, we; try to synthesize 0.5Li2MnO3·0.5LiNi0.5Co0.2Mn0.3O<sup>2</sup> materials with nanosize particles by using different lithium sources (LiOH·H2O, Li2CO<sup>3</sup> and CH3COOLi·2H2O). The solid state reaction mechanism is investigated and the effects of lithium sources on the morphology, structure and electrochemical performance are clarified.

## EXPERIMENTAL

### Material Preparation

The 0.5Li2MnO3·0.5LiNi0.5Co0.2Mn0.3O<sup>2</sup> cathode material was prepared via solid-state reaction. Analytical grade chemicals NiC4H6O4·4H2O (AR, 99.9%), CoC4H6O4·4H2O (AR, 99.5%), MnC4H6O4·H2O (AR, 99%), and different lithium sources, LiOH·H2O, Li2CO3, and CH3COOLi·2H2O (excess 3.33% molar ratio, AR 95%, AR 98%, AR 99%, respectively), with a stoichiometric amount were mixed thoroughly and ball milled (200 rpm) for 1 h. With an amount of ethanol added, the materials were ball milled (200 rpm) continually for 3 h and dried in an oven at 80◦C for 12 h to obtain a uniform mixed precursor. Then the dried precursor was ball milled for 0.5 h and sintered at 900◦C for 10 h in air atmosphere to prepare the targeted 0.5Li2MnO3·0.5LiNi0.5Co0.2Mn0.3O<sup>2</sup> material. The rate of heating was retained at 5◦C min−<sup>1</sup> . The 0.5Li2MnO3·0.5LiNi0.5Co0.2Mn0.3O<sup>2</sup> compounds synthesized by using LiOH·H2O, Li2CO3, and CH3COOLi·2H2O as lithium sources are marked as Sample A, Sample B, and Sample C, respectively.

### Sample Characterization

The crystalline structure of 0.5Li2MnO3·0.5LiNi0.5Co0.2Mn0.3O<sup>2</sup> was tested by X-ray diffraction (XRD, Rigaku D/maxb) with Cu Kα radiation(λ = 1.54056Å) in the range of 10◦ -80◦ with the speed of 5◦ min−<sup>1</sup> . The morphology was investigated with scanning electron microscopy (SEM, Philips, FEI Quanta 200 FEG) and transmission electron microscopy (TEM, TECNAI G2 F20, FEI). The sample was examined by Thermo gravimetric/Differential Scanning calorimeter (TG/DSC, SDT Q600) under the air from ambient temperature to 1,000◦C at 10◦C min−<sup>1</sup> . X-ray photoelectron spectroscopy (XPS, VG Multilab 2000) was used to observe the chemical valence of the TMs (transition elements Ni, Co, Mn, O) of the sample.

### Electrochemical Measurements

Electrochemical measurements of 0.5Li2MnO3·0.5LiNi0.5Co0.2Mn0.3O<sup>2</sup> were tested by CR2025 coin-type cells. The positive electrode was operated as slurry by 80% active material (0.5Li2MnO3·0.5LiNi0.5Co0.2Mn0.3O2), 10% acetylene black, 10% polyvinylidene fluoride (PVDF), and N-methylpyrrolidone (NMP) solvent. Then the electrode slurry was cast on aluminum foil and dried at 120◦C for 12 h under vacuum atmosphere. Typical active material areal loadings were about 1.2 mg cm−<sup>2</sup> . The cells were assembled in a filled argon glove box. Lithium metal was used as the anode and the separator was a Celgard 2500. The electrolyte utilized was a 1 M LiPF<sup>6</sup> solution by mixtures of ethylene carbonate (EC) and dimethyl carbonate (DMC) with a volume ratio of 1:1. The Galvanostatic charge-discharge measurements were carried out using NEWARE CT-3008 battery testing system (Shenzhen, China) within the voltage range of 2.0–4.8 V at room temperature. Cyclic voltammetry (CV) measurements were conducted using CHI660D Electrochemical Workstation (Shanghai Chen Hua) at 0.1 mV s−<sup>1</sup> between 2.0 and 4.8 V. Electrochemical impedance spectroscopy (EIS) of the cell was carried out using CHI660D Electrochemical Workstation (Shanghai Chen Hua) in the frequency range of 0.1 Hz−10 kHz, and the AC voltage was applied as 5 mV.

### RESULTS AND DISCUSSIONS

In order to ascertain the optimum temperature for heat treatment and explore the effects of different lithium sources, TG-DSC analyses are done for the precursor in the air. As presented in **Figure 1**, there are three main stages for weight losses in the TG plots and several endothermic and exothermic peaks in the DSC plot. The temperature range from ambient to about 200◦C, the weight loss is the release of hydration water from precursor (Deng et al., 2010). In the region from 200 to 500◦C, there is a sharp exothermic peak (**Figures 1A,B**) or two exothermic peaks (**Figure 1C**) accompanied by abrupt weight loss observed in DTG/DSC curves, it should be related to the volatilization of crystallized water from MC4H6O4·4H2O(M = Ni, Co, Mn) and the decomposition of precursor. The weight loss of precursor mainly comes from the escape of water and carbon dioxide during the reaction. As the temperature increases from 500 to 1,000◦C, weight loss (**Figures 1A**,**C**) almost can't be observed in DTG curves. However, a little exothermic peak and a small amount weight loss is observed in DSC curves (**Figure 1B**) when the temperature reaches 720◦C, which corresponding to the melting temperature of lithium carbonate. We think that it attributes to carbon dioxide emission from lithium carbonate (Li2CO3).

The mass loss of sample A, B, and C is 51.44, 54.93, and 62.63%, respectively. The weight loss is from the decomposition of precursor and the release of water and carbon dioxide under heating (Cao et al., 2017). The greater the weight losses, the greater the quantity of gas releases (Cao et al., 2017). During the material formation under heat treatment, a large number of gas releases will destroy the primary particle morphology and promote it growth, resulting in aggregation. Because the mass loss of sample A is the least, the average size of particles and uniformity of the sample A should be much better than others. It will be proved in subsequent SEM images (Cao et al., 2017).

The XRD patterns for sample A synthesized at different temperatures are shown in Figure S1. It can be seen that the crystallinity of the samples increases accompanied by temperature increment from 600 to 1,000◦C. As the temperature increases to 800◦C, there appears a couple of peaks during 20– 23◦ attributed to the super lattice diffraction of the monoclinic Li2MnO<sup>3</sup> phase (Gao et al., 2017). With the temperature increasing to 900◦C, the diffraction peaks of sample A are well indexed to a hexagonal α-NaFeO<sup>2</sup> structure (Seteni et al., 2017). Sample A displays the (006)/(012) and (018)/(110) peaks with a fine splitting, it indicates that it possesses highly ordered good crystallinity layered structure. When the temperature reaches 1,000◦C, the peak at 36.5◦ corresponding to LiMn2O<sup>4</sup> with spinel structure. It indicates that LiMn2O<sup>4</sup> structure can be formed at higher temperature (Zhang B. et al., 2014). Based on the above

analysis, the optimum sintering temperature is chosen as 900◦C in this work.

The XRD patterns of the sample A, B and C sintered at 900◦C are shown in **Figure 2**. The peaks of the samples can be indexed to the hexagonal α-NaFeO<sup>2</sup> phase (R-3m) and the little peaks from 20 to 23◦ (**Figure 2B**) are attributed to Li2MnO<sup>3</sup> phase (C2/m). Li2MnO<sup>3</sup> phase can be described as the ordering of lithium ions and transition metal ions in the layer for transition metal and the forming of LiMn<sup>6</sup> arrangement (Gao et al., 2017). Besides, the two pairs of the (006)/(012) and (018)/(110) peaks

are well separation, illustrating that the samples possesses good crystallinity and fine layered structure. Nevertheless, there are a group of minor peaks at 36.5◦ and 44◦ of sample C corresponding to LiMn2O4(Fd-3m) of formed under the high temperature. Lattice parameters of 0.5Li2MnO3·0.5LiNi0.5Mn0.3Co0.2O<sup>2</sup> are listed in **Table 1**. The c/a ratios of samples are more than 4.9, it indicates that the material possesses layered characteristics (Zhang M. et al., 2016). The I(003)/I(104) ratios of materials are much larger than 1.2, which indicates that the samples have low cationic mixing (Deng et al., 2010).

The SEM images of 0.5Li2MnO3·0.5LiNi0.5Co0.2Mn0.3O<sup>2</sup> layered materials obtained before and after heat treatment are shown in Figure S2. The images of all samples possess morphology of similar aggregation, but sample A and B (Figures S2B,E) synthesized by LiOH·H2O and Li2CO<sup>3</sup> have higher homogeneity. The sample C (Figures S2H,I) forms an irregular aggregation of primary particles. As shown in Figures S2B,E, sample A synthesized by LiOH·H2O shows much smaller primary particle sizes compared with other samples. The smaller particle size can shorten the diffusion distance of lithium ions and improve the electrochemical performance of the materials (Cao et al., 2017).

To confirm SEM results, TEM analysis and the Fast Fourier Transform (FFT) are performed, as shown in **Figure 3**. Sample A composed of 200–300 nm primary particles and the distance of the lattice fringes of particles is calculated to be 0.204 nm, matching well with the d(104) planes, which attributed to layer structure R-3m (Yang et al., 2016). **Figures 3C,D** show that sample B composed of 300–400 nm primary particles and the lattice spacing are 0.273 and 0.368 nm, corresponding to the planes d(111) and d(−111) of Li2MnO<sup>3</sup> phase (C2/m) (Luo et al., 2014). As illustrated in **Figures 3E,F**, Sample C has severe aggregation although it possesses small primary particles from some regions owing to the destruction of released gas. The distance of the lattice fringes of this material is calculated to be

TABLE 1 | Lattice parameters of 0.5Li2MnO3·0.5LiNi0.5Mn0.3Co0.2O<sup>2</sup> synthesized by different lithium sources.


0.47 nm, corresponding to the d(003) planes of layer structure (Yang et al., 2016).

In order to determine the chemical valence of major elements, the X-ray photoelectron spectra of sample A, B, and C is analyzed and shown in **Figure 4**. In contrast with the samples, the O 1s, Ni 2p, Co 2p, and Mn 2p peaks have no obvious chemical shift. In **Figure 4A**, the peak of O 1s located at 529.54 eV can be indexed to the O2<sup>−</sup> in the lattice of the samples. The Ni 2p3/<sup>2</sup> XPS spectra with binding energy of 854.98 eV, corresponding to the Ni 2p3/<sup>2</sup> peaks of Ni2<sup>+</sup> and Ni3<sup>+</sup> located at 854.0 ± 0.2 and 856.0 eV, respectively (Zhang M. et al., 2016). As shown in **Figure 4C**, the Co 2p3/<sup>2</sup> binding energy peaks of the samples centers at 780.20 eV, which is agreed with the binding energy of Co3<sup>+</sup> in LiCoO<sup>2</sup> (Seteni et al., 2017). The Mn 2p3/<sup>2</sup> peaks is 641.94 eV, which is consistent with the value of Mn4<sup>+</sup> (Lou et al., 2017). So the chemical valences of O, Ni, Co, and Mn are−2, +2/+3, +3, and +4, respectively, which indicate that lithium sources cause no effect on valence states.

**Figure 5A** shows the initial charge/discharge curves of the sample A, B, and C at C/10 within the voltage window of 2.0 and

4.8 V. During initial charge, two plateaus from 3.8 to 4.4 V and from 4.4 to 4.6 V are observed for all samples. The plateau from 3.8 to 4.4 V can be accorded with the oxidation of Ni2+/Ni3<sup>+</sup> to Ni4<sup>+</sup> and Co3<sup>+</sup> to Co4+, which is in good agreement with the reversible extraction of Li<sup>+</sup> from LiMO<sup>2</sup> (M = Ni, Co, Mn) phase (Seteni et al., 2017). Furthermore, the later plateau is attributed to the Li<sup>+</sup> and O2<sup>−</sup> irreversible extraction as Li2O from the inert Li2MnO<sup>3</sup> phase, resulting in high irreversible capacity (Yang et al., 2016). When the voltage is from 3.8 to 4.4 V during the first charging cycle, the charging mechanism is written as (Lou et al., 2017),

than the others. **Figure 5B** shows rate performance of the Sample A, B, and C at different rates. It can be seen that the discharge capacities of all samples decrease as the rates increased due to the poor conductivity of the material and inert of Li2MnO<sup>3</sup> (Seteni et al., 2017). However, the sample A shows much higher rate property because of smaller primary particle, which can shorten the Li<sup>+</sup> ions diffusion pathway. The midpoint voltage decay of the samples during cycling at rate of C/10 is shown in **Figures 5C,D**. The discharge capacities of sample A, B, and C reaches 231.8, 157.7, and 121.0 mAhg−<sup>1</sup> after 30 cycles at C/10 rate, with capacity retentions of 88.4, 63.8, and 88.6%,

$$0.5\text{Li}\_2\text{MnO}\_3 \cdot 0.5\text{LiNi}\_{0.5}\text{Co}\_{0.2}\text{Mn}\_{0.3}\text{O}\_2 \xrightarrow{charge} 0.5\text{Li}\_2\text{MnO}\_3 \cdot 0.5\text{Ni}\_{0.5}\text{Co}\_{0.2}\text{Mn}\_{0.3}\text{O}\_2 + 0.5\text{Li}^+ + 0.5\text{e}^-$$

When the voltage is from 4.4 to 4.6 V, the charging mechanism is as follows (Lou et al., 2017),

respectively. **Figure 5D** indicates that sample A holds the most stable voltage from 3.71 to 3.43 V during cycling. However, a

$$0.5 \text{Li}\_2 \text{MnO}\_3 \cdot 0.5 \text{Ni}\_{0.5} \text{Co}\_{0.2} \text{Mn}\_{0.3} \text{O}\_2 \xrightarrow{\text{charge}} 0.5 \text{MnO}\_2 \cdot 0.5 \text{Ni}\_{0.5} \text{Co}\_{0.2} \text{Mn}\_{0.3} \text{O}\_2 + 0.5 \text{Li}\_2 \text{O}\_3$$

The initial charge/discharge capacity and the coulombic efficiency of the sample A, B, and C are listed in **Table 2**. Obviously, the discharge capacity of the sample A is much higher sudden drop appears for the midpoint voltage of Sample B after 20 cycles. The results indicate sample A keeps the most stable voltage during cycling.

Cyclic voltammetry curves of sample A, B, and C for the initial three cycles during the voltage range of 2.0–4.8 V at the scan rate of 0.1 mV s−<sup>1</sup> are shown in **Figure 6**. The CV curves are similar. There are two oxidation peaks and two reduction peaks in the initial cycle. The oxidation peak at about 4.0 V is the oxidation of Ni2+/Ni3<sup>+</sup> to Ni4<sup>+</sup> and Co3<sup>+</sup> to Co4<sup>+</sup> with the reversible extraction of Li<sup>+</sup> from LiMO<sup>2</sup> (M = Ni, Co, Mn) phase (Xiao et al., 2017). The oxidation peak at around 4.6 V corresponds to the Li<sup>+</sup> and O2<sup>−</sup> irreversible extraction from the Li2MnO<sup>3</sup> phase. In the following reduction process, the peak at about 3.7 V is the reduction of Ni4<sup>+</sup> to Ni2+/Ni3<sup>+</sup> and Co4<sup>+</sup> to Co3+, and the peak at around 3.2 V related to Li<sup>+</sup> insertion into layered MnO<sup>2</sup> (Seteni et al., 2017). Besides, the peaks of sample A and B (**Figures 6A,B**) are higher than sample C, indicated that sample A and B have more steady structure. It is also worth mentioning that, there are two peaks at about 2.9 V and 2.5 V for the sample C (**Figure 6C**), corresponding to oxidation peak and reduction peak, respectively, related to a small amount of spinel structure (Cao et al., 2017).

Electrochemical impedance spectroscopy (EIS) can be used to investigate the electrode kinetic process of samples A, B, and C. As shown in **Figure 7**, The EIS plots consist of a semicircle arc and a straight line. The semicircle arc at high frequency region corresponds to the charge transfer process, and the straight line at low frequency region is the lithium diffusion process. The plots are fitted using the electric equivalent circuit model, as shown in **Figure 7**. The parameters of the



equivalent circuit are listed in **Table 3**. In the equivalent circuit, R<sup>s</sup> and Rct represent the solution resistance and charge-transfer resistance, respectively (He et al., 2015). CPE is related to capacitance of the surface layer (Toprakci et al., 2013). Z<sup>W</sup> represents the Warburg impedance (Xiao et al., 2017) (Z' is the real impedance and Z" is the imaginary impedance). It is observed that the R<sup>s</sup> and Rct of sample A (5.39 , 87.93 ) are smaller than those of sample B (7.968 , 72.08 ) and sample C (27.31 , 120 ). The results indicate that the solution resistance and charge-transfer resistance of sample A is the smallest, which mainly because sample A prepared by using LiOH·H2O owns the smaller primary sizes (Cao et al., 2017). Hence, sample A exhibits the best electrochemical properties.

TABLE 3 | EIS fitting values of the samples A, B, and C.

equivalent circuit model.


## CONCLUSIONS

In summary, 0.5Li2MnO3·0.5LiNi0.5Co0.2Mn0.3O<sup>2</sup> materials have been successfully synthesized by using three kinds of lithium sources, LiOH·H2O, Li2CO3, and CH3COOLi·2H2O, respectively. The effects of morphology, structure, electrochemical performances of the 0.5Li2MnO3·0.5LiNi0.5Co0.2Mn0.3O<sup>2</sup> material prepared by using different lithium sources have been investigated. 0.5Li2MnO3·0.5LiNi0.5Co0.2Mn0.3O<sup>2</sup> material prepared by using LiOH·H2O shows the most uniform morphology with the particle diameters of about 200–300 nm and stable layer structure. It delivers the best electrochemical performances with the initial discharge capacity reaching 255.6 mAh g−<sup>1</sup> at C/10, and the capacity retention is 88.4% after 30 cycles at C/10. LiOH·H2O is the best choice for the synthesis of 0.5Li2MnO3·0.5LiNi0.5Co0.2Mn0.3O<sup>2</sup> material compared with Li2CO<sup>3</sup> and CH3COOLi·2H2O.

### REFERENCES


### AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

### ACKNOWLEDGMENTS

This study was supported by National Natural Science Foundation of China (Grant No. 51774333, 51604081) and the Innovation-Driven Project of Central South University (NO.2016CX021).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2018.00159/full#supplementary-material

stability for Lithium-ion batteries. ACS Appl. Mater. Interfaces 6, 10330–10341. doi: 10.1021/am5017649


lithium-ion battery under high cutoff voltage. J. Alloys Compd. 673, 237–248. doi: 10.1016/j.jallcom.2016.03.003


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The handling Editor declared a shared affiliation, though no other collaboration, with one of the authors, M-ZL.

Copyright © 2018 Wang, Luo, Zheng, He, Tong and Yu. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Porous Hollow Superlattice NiMn2O4/NiCo2O<sup>4</sup> Mesocrystals as a Highly Reversible Anode Material for Lithium-Ion Batteries

Lingjun Li 1,2 \*, Qi Yao<sup>1</sup> , Jiequn Liu<sup>3</sup> , Kaibo Ye<sup>1</sup> , Boyu Liu<sup>1</sup> , Zengsheng Liu<sup>1</sup> , Huiping Yang<sup>1</sup> , Zhaoyong Chen<sup>1</sup> , Junfei Duan<sup>1</sup> and Bao Zhang<sup>4</sup> \*

*<sup>1</sup> School of Materials Science and Engineering, Changsha University of Science and Technology, Changsha, China, <sup>2</sup> Hunan Provincial Key Laboratory of Efficient and Clean Energy Utilization, Changsha University of Science and Technology, Changsha, China, <sup>3</sup> School of Iron and Steel, Soochow University, Suzhou, China, <sup>4</sup> School of Metallurgy and Environment, Central South University, Changsha, China*

### Edited by:

*Qiaobao Zhang, Xiamen University, China*

### Reviewed by:

*Xiao-Dong Zhu, Harbin Institute of Technology, China Baihua Qu, Xiamen University, China Cao Guan, National University of Singapore, Singapore*

#### \*Correspondence:

*Lingjun Li lingjun.li@csust.edu.cn Bao Zhang csuzb@vip.163.com*

#### Specialty section:

*This article was submitted to Physical Chemistry and Chemical Physics, a section of the journal Frontiers in Chemistry*

> Received: *13 March 2018* Accepted: *18 April 2018* Published: *15 May 2018*

#### Citation:

*Li L, Yao Q, Liu J, Ye K, Liu B, Liu Z, Yang H, Chen Z, Duan J and Zhang B (2018) Porous Hollow Superlattice NiMn*2*O*4*/NiCo*2*O*4 *Mesocrystals as a Highly Reversible Anode Material for Lithium-Ion Batteries. Front. Chem. 6:153. doi: 10.3389/fchem.2018.00153* As a promising high-capacity anode material for Li-ion batteries, NiMn2O<sup>4</sup> always suffers from the poor intrinsic conductivity and the architectural collapse originating from the volume expansion during cycle. Herein, a combined structure and architecture modulation is proposed to tackle concurrently the two handicaps, via a facile and well-controlled solvothermal approach to synthesize NiMn2O4/NiCo2O<sup>4</sup> mesocrystals with superlattice structure and hollow multi-porous architecture. It is demonstrated that the obtained NiCo1.5Mn0.5O<sup>4</sup> sample is made up of a new mixed-phase NiMn2O4/NiCo2O<sup>4</sup> compound system, with a high charge capacity of 532.2 mAh g−<sup>1</sup> with 90.4% capacity retention after 100 cycles at a current density of 1 A g−<sup>1</sup> . The enhanced electrochemical performance can be attributed to the synergistic effects of the superlattice structure and the hollow multi-porous architecture of the NiMn2O4/NiCo2O<sup>4</sup> compound. The superlattice structure can improve ionic conductivity to enhance charge transport kinetics of the bulk material, while the hollow multi-porous architecture can provide enough void spaces to alleviate the architectural change during cycling, and shorten the lithium ions diffusion and electron-transportation distances.

Keywords: lithium-ion battery, transition metal oxide, superlattice structure, hollow multi-porous architecture, electrochemical kinetics

## INTRODUCTION

Nowadays, Graphite-based materials are the main anode material for current commercial lithiumion battery (LIB) (Kang et al., 2006; Scrosati et al., 2011; Xie et al., 2012; Goodenough and Park, 2013; Yuan et al., 2014; Yan et al., 2017). However, the low theoretical specific capacity of 372 mAh g −1 and poor rate capability limit its application in next generation high energy density LIB (Mai et al., 2011; Pan et al., 2014; Liu et al., 2015; Ma Z. et al., 2015; Wang J. X. et al., 2016; Wu et al., 2017; Su et al., 2018). In this regard, a variety of simple metal oxides, such as tin oxide, manganese oxide, iron oxide and cobalt oxide with high energy density, have drawn much attention as the desired candidates for anode materials to satisfy the high energy storage requirements (Mai et al., 2011; Xie et al., 2012; Li Q. et al., 2013; Pan et al., 2014; Yuan et al., 2014; Liu et al., 2015; Ma Z. et al., 2015; Wang J. X. et al., 2016; Yan et al., 2017; Su et al., 2018). In addition, complex oxides, such as NiMn2O4, ZnCo2O4, CoMn2O4, ZnFe2O4, CuCo2O4, and ZnMn2O4, have shown better electronic conductivity and higher reversible capacity than those of single-component metal oxides, which could be ascribed to the multiple aliovalent cations and corresponding more versatile redox reactions (Zhou et al., 2012; Li J. F. et al., 2013a,b,c; Chen et al., 2014c; Shen et al., 2014, 2015; Yuan et al., 2014; Zhang et al., 2014, 2015, 2016; Kang et al., 2015; Leng et al., 2015; Ma Y. et al., 2015; Ma Z. et al., 2015; Wang Y. K. et al., 2016; Wu L. J. et al., 2016; Yan et al., 2017). Among them, as a typical spinel structure, NiMn2O<sup>4</sup> has been extensively studied and demonstrated as a promising next generation anode material for LIBs (Kang et al., 2015; Ma Z. et al., 2015; Shen et al., 2015). However, the poor intrinsic electronic conductivity of NiMn2O<sup>4</sup> still impairs the electron transport kinetics during the process of the redox reaction, and NiMn2O<sup>4</sup> shows rapid capacity fading because of the slow ion-diffusion rates and large volume changes during the cycling process (Kang et al., 2015; Ma Y. et al., 2015; Ma Z. et al., 2015; Shen et al., 2015; Zhang et al., 2015).

To solve the aforementioned problems and enhance the electrochemical performance of transition metal oxides, one efficient method is to synthesize hollow multi-porous architecture of transition metal oxides (Zhou et al., 2012; Choi and Kang, 2013; Li J. F. et al., 2013a,b,c; Li Q. et al., 2013; Zhu et al., 2013a,b; Chen et al., 2014c; Luo S. et al., 2014; Pan et al., 2014; Wang L. Y. et al., 2014; Zhang et al., 2014, 2015, 2018; Gao et al., 2015; Kang et al., 2015; Leng et al., 2015; Ma Z. et al., 2015; Shen et al., 2015; Su et al., 2015; Wang Y. K. et al., 2016; Wu et al., 2018). Compared with bulk counterparts, the hollow multi-porous structures can provide enough void space to buffer architectural changes and alleviate the formation of the aggregate particles during the process of redox reactions (Li J. F. et al., 2013c; Li Q. et al., 2013; Zhang et al., 2014, 2015, 2018; Kang et al., 2015; Leng et al., 2015; Ma Z. et al., 2015; Shen et al., 2015; Wang Y. K. et al., 2016). In addition, the hollow multi-porous character is conducive to the electrolyte penetration, offering a short path for the lithium ions diffusion and electron-transportation (Li J. F. et al., 2013c; Li Q. et al., 2013; Gao et al., 2015; Zhang et al., 2015, 2018; Wang Y. K. et al., 2016).

Most recently, cation substitution has been certified to be an effective route to improve the electrical conductivity and charge transfer ability of the anode materials (Mai et al., 2011; Qiu et al., 2011; Li et al., 2012, 2015, 2016; Li Q. et al., 2013; Reddy et al., 2013; Suo et al., 2013; Chen et al., 2014a,b; Liu et al., 2014; Sun et al., 2014; Wang J. X. et al., 2014; Xu et al., 2014; Choi et al., 2015; He et al., 2015, 2017; Ma Z. et al., 2015; Mo et al., 2015; Jin et al., 2016; Wu F. X. et al., 2016; Wu L. J. et al., 2016). Ma Y. et al. (2015) found that the optimal level of iron doping in spinel NiMn2O<sup>4</sup> can improve the electrical conductivity and structural robustness and exhibit exciting lifespan. Tu et al. (Mai et al., 2011) found that "Co-doped NiO nanoflake arrays samples showed excellent rate capability and good capacity retention." Furthermore, it is noted that the content of Co could also affect the structure and morphology of the transition metal oxides. Qian et al. (Li J. F. et al., 2013b) found that the distribution and morphology of MnCo2O<sup>4</sup> particles are very different from those of CoMn2O<sup>4</sup> particles. Typically, "the resultant pores are much larger in CoMn2O<sup>4</sup> than that of MnCo2O4, possibly due to the longer transfer distance of ions" (Li J. F. et al., 2013b). These results imply that Co substitution is critical to the electrical conductivity and controlled porous architecture of transition metal oxides. However, to the best of our knowledge, the studies on Co substituted NiMn2O<sup>4</sup> mesocrystals with controlled phase structure and morphology are scarcely reported.

Herein, we first report a solvothermal method for the successful preparation of well-distributed hollow multi-porous superlattice NiMn2O4/NiCo2O<sup>4</sup> mesocrystals. As expected, Co element is uniformly distributed among the NiMn2O<sup>4</sup> mesocrystals, and that Co substitution induces a great evolution of the structure and morphology. The structure is changed from the single-phase NiMn2O<sup>4</sup> to a new mixed-phase NiMn2O4/NiCo2O<sup>4</sup> compound system. At the same time, the morphological transformation of the NiMn2O<sup>4</sup> sample to the Co-substituted samples take place in the significant conversion from irregular particles to well-distributed hollow multi-porous mesocrystals. The optimized NiCo1.5Mn0.5O<sup>4</sup> sample maintains 532.2 mAh g−<sup>1</sup> with 90.4% capacity retention after 100 cycles at a current density of 1 A g−<sup>1</sup> with the voltage of 0.01– 3.00 V vs. Li/Li<sup>+</sup> at room temperature, because the superlattice NiMn2O4/NiCo2O<sup>4</sup> structure can enhance the charge transport kinetics, and the uniform hollow multi-porous architecture improve the structural stability and facilitate the electrolyte penetration during lithiation and delithiation cycling.

### EXPERIMENTAL

## Materials Synthesis

NiMn2−xCoxO<sup>4</sup> (x = 0, 0.5, 1, 1.5, 2) oxide materials are synthesized by solvothermal reaction. The stoichiometric ratio of Ni(CH3COO)2·4H2O, Mn(CH3COO)2·2H2O, Co(CH3COO)2·2H2O, and hexamethylenetetramine (HMT) are dissolved in a mixed solvent of 13 mL water and 67 mL triethylene glycol (TEG) to obtain a homogeneous and transparent solution. Then the solution is put into a Teflon-lined stainless steel autoclave. After the autoclave is maintained at 200◦C for 16 h, it is cooled to room temperature naturally. The precipitate is collected after filtration, washing, and drying to obtain the precursor. The as-prepared precursor is first calcinated at 250◦C in air for 3 h and then at 900◦C in air for 3 h.

NiMn2−xCoxO<sup>4</sup> (x = 0, 0.5, 1, 1.5, 2) oxide materials, are labeled as NM (x = 0), Co-1 (x = 0.5), Co-2 (x = 1), Co-3 (x = 1.5), and Co-4 (x = 2), respectively, and these corresponding precursors are labeled as NMP, CoP-1, CoP-2, CoP-3, and CoP-4, respectively.

### Materials Characterizations

X-ray diffraction (XRD, Rigaku D/Max 200PC, Japan) is employed to characterize the phases of all samples. The scanning range of diffraction angle (2θ) is 10◦∼90◦ and the scanning rate is 5◦ min−<sup>1</sup> . The morphologies of all samples are examined by scanning electron microscopy (SEM, Nova NanoSEM-230). The phase structures of the NM and Co-3 are analyzed by high resolution transmission electron microscopy (HRTEM, FEI Tecnai G2 F20 S-Twin working at 200 kV). The energy dispersive X-ray spectroscopy (EDS) is used for the elemental characterization of all samples by OXFORD 7426 as the attachment of SEM, with the acceleration voltage of 20 kV. A specific surface area (SSA) analysis was used to measure material's BET SSA (SSAA, 3H2000, BSD, China).

### Electrochemical Measurement

The electrochemical tests are performed in CR2025 coin-type cells. Typical working electrode are prepared by a slurry coating process and the loadings are between 1.1 and 1.2 mg cm−<sup>2</sup> , and an electrode diameter of 10 mm is used. The working electrode consists of as-synthesized anode material, acetylene black and polyvinylidene difluoride (PVDF) binder with a weight ratio of 5:3:2. A lithium metal foil is used as the counter electrode. 1 M LiPF<sup>6</sup> dissolved in the mixtures of dimethyl carbonate (DMC), Ethyl Methyl Carbonate (EMC) and ethylene carbonate (EC) with a volume ratio of 1:1:1 is used as electrolyte. The assembly of the cells is carried out in a dry an argon-filled MIKROUNA Universal 24401750 glovebox. Electrochemical performance of the materials are tested using an automatic galvanostatic charge-discharge unit, Land CT2001 battery cycler, in a cutoff voltage range within 0.01–3.0 V vs. Li/Li<sup>+</sup> at room temperature. The cyclic voltammetry (CV) are operated at a scan rate of 0.1 mV s−<sup>1</sup> in a cutoff voltage range within 0.01– 3.0 V at room temperature with a CHI660D electrochemical analyzer. Finally, the electrochemical impedance spectroscopy (EIS) measurements are conducted by a CHI660D impedance analyzer, using 2-electrode cells. All cells are initially discharged and charged at a current density of 0.05 A g−<sup>1</sup> , and then discharged and charged different times of 1st and 100th at a current density of 1 A g−<sup>1</sup> with the voltage of 0.01 – 3.00 V vs. Li/Li<sup>+</sup> at room temperature. The amplitude voltage is 5 mV and the frequency range is 0.01–100,000 Hz.

### RESULTS AND DISCUSSION

**Figure 1a** displays the X-ray diffraction patterns of the NMP, CoP-1, CoP-2, CoP-3, and CoP-4 precursors. Compared with standard XRD patterns of NiCO<sup>3</sup> (JCPDS no. 12-0771), MnCO<sup>3</sup> (JCPDS no. 85-1109), and CoCO<sup>3</sup> (JCPDS no. 01-1020), all the diffraction lines could be indexed on the basis of the hexagonal phase with the space group of R-3c (Qiu et al., 2011; Sun et al., 2014; Choi et al., 2015; Kang et al., 2015; Wu F. X. et al., 2016). With the increase of cobalt concentration, the manganese concentration decreases, and the diffraction peaks shift right to the positions at large angles, indicating that the substituted Co incorporates into the crystal lattice, resulting in the decrease of the lattice constant. Furthermore, from the diffraction lines of CoP-2 and CoP-3 precursors, it is noticed that a second phase appears in the pattern of the carbonate precursors at 2θ of about 32◦ and become obviously with the increase of the Co concentration, which can confirm that high cobalt concentrations prefer to form two phases rather than one phase (Ni1/3Mn2/3−xCoxCO3) in the carbonate precursors through the solvothermal process.

**Figure 1b** shows the X-ray diffraction patterns of the NM, Co-1, Co-2, Co-3, and Co-4 samples. The diffraction peaks of the NM sample in the XRD patterns can be assigned to well-crystallized cubic spinel NiMn2O<sup>4</sup> (JCPDS no. 71-0852). Compared to that of the NM, the diffraction peaks of the Co-1 and Co-2 shift right to the positions at high difraction angles, as the cobalt concentration increases, because the lattice constant is decreased by substituting Co3<sup>+</sup> (0.068 nm) into Mn3<sup>+</sup> (0.072 nm) site (Suo et al., 2013; Mo et al., 2015; He et al., 2017). Furthermore, from the diffraction line of Co-3, the distinct peaks of (311), (400), and (440) at 2θ of about 36, 43, and 64◦ are all completely separated, respectively, indicating a second cubic spinel phase of NiCo2O4, which is in accordance with the precursors results. These results suggest that the Co-3 is consistently made up of two phases, which allows hybridizing the phase of NiMn2O<sup>4</sup> with the phase of NiCo2O<sup>4</sup> into the architecture. In addition, the diffraction lines of Co-4 is assigned to CoO (JCPDS no. 70-2855) and NiCo2O<sup>4</sup> (JCPDS no. 73-1702), which might be ascribed that the single phase NiCo2O<sup>4</sup> cannot be synthesized through our heat-treat process. Chen et al. (Mo et al., 2015) reported that NiCo2O<sup>4</sup> is synthesized at 400◦C with a temperature ramp of 4◦C min−<sup>1</sup> and kept for 5 h under ambient air. However, in our study, Co-4 was first calcinated at 250◦C in air for 3 h and then at 900◦C in air for 3 h, which is consistency with the other samples.

To further confirm the effect of Co substitution on the microstructure of NiMn2O4, HRTEM and corresponding FFT (fast Fourier transform) images of NM and Co-3 samples are exhibited in **Figures 1c,d**. The lattice fringes and FFT patterns of both samples reveal that the NiMn2O<sup>4</sup> structure is successfully formed as expected. However, compared to that of NM (c-I and c-II), the corresponding FFTs of Co-3 (d-I and d-II) exhibit another two interplanar distances of 1.56 and 2.45 Å, which corresponds to the d-spacing of the (511) and (311) planes of NiCo2O<sup>4</sup> (JCPDS no. 73-1702), respectively. The NiMn2O<sup>4</sup> and NiCo2O<sup>4</sup> phases are mixing with one another at the atomic level in Co-3 particle, which suggests a two-phase composite nature of the sample derived from small fields of view. These results can determine that the Co-3 is a new mixed-phase NiMn2O4/NiCo2O<sup>4</sup> compound system with the presence of a superstructure. In addition, it is reported that the chemical composition of Ni and Co elements in NiCo2O<sup>4</sup> contains Ni2+/Ni3<sup>+</sup> and Co2+/Co3+, and the chemical composition of Ni and Mn elements in NiMn2O<sup>4</sup> contains Ni2+/Ni3<sup>+</sup> and Mn2+/Mn3<sup>+</sup> (Kang et al., 2015; Ma Y. et al., 2015; Mo et al., 2015; Shen et al., 2015; Li et al., 2016). Therefore, we hold this assumption that the chemical composition of Ni Co and Mn elements in NiCo1.5Mn0.5O<sup>4</sup> contains Ni2+/Ni3+, Co2+/Co3+, and Mn2+/Mn3+.

The SEM images of all carbonate precursors, corresponding EDS mappings of Ni, Co, Mn, and O for CoP-3 carbonate precursors and the EDX contents of Ni, Co and Mn element in all carbonate precursors are illustrated in **Figures 2a–j**. As can be seen in **Figures 2a,b**, the NMP and CoP-1 are composed by microflakes, microspheres and spindle-like particles. With the increase of Co concentration (**Figures 2c–e**), the microflakes disappear, and the microspheres are transformed into welldistributed spindle-like particles. EDS mapping result confirms that all elements, including Co, are uniformly distributed. EDX analysis exhibits that the ratio of Ni, Co and Mn element in

FIGURE 1 | (a) X-ray diffraction patterns of NMP, CoP-1, CoP-2, CoP-3, and CoP-4, (b) X-ray diffraction patterns of NM, Co-1, Co-2, Co-3, and Co-4, HRTEM images of (c) NM and (d) Co-3 powders, and corresponding FFT (Fast Fourier transform) images.

NMP, CoP-1, CoP-2, CoP-3, and CoP-4 is about 1:0:2, 1:0.5:1.5, 1:1:1, 1:1.5:0.5, 1:2:0, respectively, which is in accordance with our expectation. Therefore, it could be concluded that the stoichiometrical precursors are successfully synthesized via hydrothermal process, and that Co substitution is conducive to the good morphological consistency and distribution of the NMP particles.

The SEM images of all samples are illustrated in **Figures 2k–o** and Figure S1. Particles possessing porous character are observed for all the samples, which could be ascribed to the decomposition of carbonate precursor during heating process. It is also noted that the particles morphology for all the samples are inherited from those of the precursors. For example, particles from the NM sample show irregular morphology and random distribution. However, for the Co-1 and Co-2, the particles exhibit hollow spherical morphology and good distribution. As the increase of Co concentration, particles from the Co-3 exhibit spindlelike morphology and hollow multi-porous architecture. EDS mapping of Co-3 (Figure S1) shows a uniform distribution of Ni, Co, Mn, and O throughout the particle. These results confirm that Co incorporates into the Co-3, and Co substitution changes the morphology of both primary and aggregate particles, which might be benefit for lithium and electron diffusion during lithiation/delithiation reaction.

Schematic diagram for the synthesis process of the porous hollow superlattice NiMn2O4/NiCo2O<sup>4</sup> mesocrystals are illustrated in **Figure 2p**. The acetate molecules and hexamethylenetetramine dissolve in the solvent then gather together to form small droplets. During solvothermal treatment, the hexamethylenetetramine decomposes, and creates CO<sup>2</sup> bubbles. By capturing the metal ions on bubble surface, an oval-like microsphere precursor is obtained successfully. During calcination process, metal ions move into crystal structure of the superlattice NiMn2O4/NiCo2O<sup>4</sup> mesocrystals, meanwhile, CO<sup>2</sup> release in the carbonate precursors, leaving pores on the obtained hollow complex oxides (Luo D. et al., 2014; Luo et al., 2015; Ma Z. et al., 2015; Shen et al., 2015).

To understand the oxidation/reduction and phase transformation processes in electrode reactions of the synthesized samples, the CV curves for the initial three cycles of NM and Co-3 samples are presented in **Figures 3A,B**, which are tested at a scan rate of 0.1 mV s−<sup>1</sup> in the voltage range of 0.01–3.00 V vs. Li/Li<sup>+</sup> at room temperature, respectively. The CV curves of both samples for the first cycle are significantly different from those for the following cycles, because of the irreversible electrochemical reaction during the first discharge cycle. However, no obvious alteration is observed between the second cycle and the third cycle, suggesting the good reversibility of lithium insertion and extraction reactions. In the first cycle, as can be seen in **Figure 3A**, there is a sharp reduction peak at 0.9 V and a broad reduction peak between 0.3 and 0.8 V. The sharp one associates with the initial reduction of Mn3+–Mn2+, and the broad one can be attributed to the reduction of NiMn2O<sup>4</sup> to metallic Mn and Ni and the formation of Li2O (Li J. F. et al., 2013b; Kang et al., 2015; Ma Y. et al., 2015; Ma Z. et al., 2015). With the increase of Co concentration (Figure S2), it is clear that the sharp reduction peak at 0.9 V is disappeared, and the reduction peaks which is attributed to the reduction of the oxides to metal and the formation of Li2O are narrower than that of NiMn2O4. In comparison, as shown in **Figure 3B**, a new reduction peak appear at 1.0 V, which can be corresponded to the reduction of Co3+–Co2+, and the broad reduction peaks

of Co-3 are completely separated. Yuan et al. (Ma Z. et al., 2015) and Lee et al. (Kang et al., 2015) reported that "the sharp peak at 0.5 V should correspond to the reduction of NiMn2O<sup>4</sup> to metallic Mn and Ni," while Yang et al. (Zhang et al., 2016) and Wang et al. (Leng et al., 2015) reported that "the peak at 0.8 V should correspond to the reduction of NiCo2O<sup>4</sup> to metallic Co and Ni," and both peaks "relate to the formation of Li2O and solid-electrolyte interface (SEI)" (Li J. F. et al., 2013c; Zhang et al., 2014, 2018; Gao et al., 2015). These results further confirm that Co-3 is a kind of mixed-phase NiMn2O4/NiCo2O<sup>4</sup> composite material, which is in accordance with the XRD results and TEM results as aforesaid. For the Co-3 sample, in the anodic sweep, a wide oxidation peak (∼2.0 V) can be attributed to the oxidation of metals to oxides accompanying the decomposition of Li2O. Based on the above CV analysis, in conjunction with the previously reported storage mechanism for NiMn2O<sup>4</sup> and NiCo2O<sup>4</sup> (Li J. F. et al., 2013c; Zhang et al., 2014, 2016, 2018; Gao et al., 2015; Kang et al., 2015; Leng et al., 2015; Ma Y. et al., 2015;

Schematic diagram (p) for the synthesis process of the porous hollow superlattice NiMn2O4/NiCo2O4 mesocrystals.

Ma Z. et al., 2015; Su et al., 2015), the electrochemical reactions for the electrodes can be summarized as follows:

$$\text{NiMnO}\_{4} + 8\text{Li}^{+} + 8\text{e}^{-} \rightarrow \text{Ni} + 2\text{Mn} + 4\text{Li}\_{2}\text{O} \quad \text{(1)}$$

$$\text{Ni} + 2\text{Mn} + 3\text{Li}\_{2}\text{O} \leftrightarrow \text{NiO} + 2\text{MnO} + 6\text{Li}^{+} + 6\text{e}^{-} \text{(2)}$$

$$\text{NiCo}\_{2}\text{O}\_{4} + 8\text{Li}^{+} + 8\text{e}^{-} \rightarrow \text{Ni} + 2\text{Co} + 4\text{Li}\_{2}\text{O} \quad \text{(3)}$$

$$\text{Ni} + 2\text{Co} + 3\text{Li}\_{2}\text{O} \leftrightarrow \text{NiO} + 2\text{CoO} + 6\text{Li}^{+} + 6\text{e}^{-} \text{(4)}$$

$$\text{NiMn}\_{2-x}\text{Co}\_{x}\text{O}\_{4} + 8\text{Li}^{+} + 8\text{e}^{-} \rightarrow \text{Ni}$$

$$\qquad + (2-x)\text{Mn} + x\text{Co} + 4\text{Li}\_{2}\text{O} \qquad \text{(5)}$$

$$\text{Ni} + (2-x)\text{Mn} + x\text{Co} + 3\text{Li}\_{2}\text{O} \leftrightarrow \text{NiO}$$

$$\qquad + (2-x)\text{MnO} + x\text{CoO} + 6\text{Li}^{+} + 6\text{e}^{-} \qquad \text{(6)}$$

**Figure 3C** depicts the initial charge-discharge curves of the NM, Co-1, Co-2, Co-3, and Co-4 samples. The electrochemical tests are carried out at a current density of 0.05 A g−<sup>1</sup> with the voltage of 0.01–3.00 V vs. Li/Li<sup>+</sup> at room temperature. As can be seen in **Figure 3C**, there is a representative plateaus of the profile for

NM around 0.7 V in the first discharge process. By comparison, a representative profile of Co-3 which is composed of two plateaus around 0.9 V and 0.7 V in the first discharge process can be observed clearly, which relates to the insertion of lithium into the crystal structure of NiCo2O<sup>4</sup> and NiMn2O4, respectively, accompanied by the formation of Li2O (Wang J. X. et al., 2014; He et al., 2015, 2017; Mo et al., 2015; Li et al., 2016). The initial charge capacity for the NM, Co-1, Co-2, Co-3, and Co-4 is 513.5, 580.8, 636.9, 665.4, and 595.7 mAh g−<sup>1</sup> , respectively. It is obvious that the initial charge capacities of the Co-substituted samples are superior than that of the NM, particularly, the Co-3 sample exhibit the optimal initial charge capacities of all samples.

The rate capability of all samples are given in **Figure 3D**. The cells are charged/discharged at a current density of 0.05 A−1.6 A g−<sup>1</sup> , and then charged/discharged at a current density of 0.1 A g−<sup>1</sup> with the voltage of 0.01–3.00 V vs. Li/Li<sup>+</sup> at room temperature. It is noted that the Co-substituted samples show improved rate capability compared to the pristine counterparts at high rates. Specifically, the Co-3 demonstrates the best rate performance and delivers the highest reversible capacities of all samples. Remarkably, it can maintain the charge capacities of 611.4, 591.1, 554.6, 499.1, 467.6, 435.6, and 606.8 mAh g−<sup>1</sup> after 5 cycles at the current density of 0.1, 0.2, 0.5, 1, 1.2, 1.6, and 0.1 A g−<sup>1</sup> , respectively. Even at a current density as high as 1.6 A g −1 , the Co-3 can still deliver a stable capacity of about 435.6 mAh g−<sup>1</sup> after 5 cycles, which is still higher than the theoretical capacity of the graphite (372 mAh g−<sup>1</sup> ), indicating the excellent rate capability.

The comparison of the cycling stability between the NM and the Co-substituted samples are presented in **Figure 3E**. The cells were charged/discharged 2 times at a current density of 0.05 A g−<sup>1</sup> , and then charged/discharged 100 times at a current density of 1 A g−<sup>1</sup> with the voltage of 0.01–3.00 V vs. Li/Li<sup>+</sup> at room temperature. The charge capacity retention reaches 61.3, 79.7, 73.8, 90.4, and 70.2% after 100 cycles for NM, Co-1, Co-2, Co-3, and Co-4, respectively. In particular, as can be seen in **Figure 3F**, the Co-3 maintains 532.2 mAh g−<sup>1</sup> with 67.8% capacity retention after 500 cycles at a current density of 1 A g−<sup>1</sup> with the voltage of 0.01–3.00 V vs. Li/Li<sup>+</sup> at room temperature. It means that the Co-3 can still deliver a stable capacity of about 360.8 mAh g−<sup>1</sup> after 500 cycles, which is still near the theoretical capacity of the graphite (372 mAh g−<sup>1</sup> ), demonstrating the outstanding cycling stability. In conclusion, the porous Co-substituted samples in this work exhibit a superior electrochemical performance compared with the NM, specially, the Co-3 exhibits the optimal electrochemical performance of all samples.

To further understand the effects of the porous architecture on the electrochemical properties of Co-substituted samples, the Co-1 is selected as a reference, which has the similar structure but different morphological architecture, compared to the NM. The SEM images of NM and Co-1 electrodes before and after 100 cycles are shown in **Figure 4**. The cells are charged to 3.00 V at high delithiation state. As can be seen in **Figure 4**, after 100 cycles, both samples (**Figures 4b,d**) show fuzzy surface, which may be due to the formation of Li2O and the electrolyte decomposition (Wang J. X. et al., 2014; Mo et al., 2015; Li et al., 2016; He et al., 2017). Compared to the fresh electrode (**Figure 4a**), the irregular morphological NiMn2O<sup>4</sup> particles suffer serious damage and expand into aggregate particles after 100 cycles (**Figure 4b**). The aggregation can be ascribed to the volume expansion on

FIGURE 4 | SEM images of (a,b) NM electrodes, (c,d) Co-1 electrodes, (a,c) before cycling, (b,d) after 100 cycles at a current density of 1A g−<sup>1</sup> with the voltage of 0.01–3.00 V vs. Li/Li+.

FIGURE 5 | N2 adsorption and desorption isotherm: (A) NM, (B) Co-1, and (C) Co-3. Nyquist plots of as-prepared electrodes, (D) after the 1st cycle and (E) after the 100th cycle, (F) the relationships between Z're and ω <sup>−</sup>1/<sup>2</sup> after the 1st cycle for the NM, Co-1 and Co-3 electrodes.

account of the irreversible electrochemical process from the high lithiation state (Pan et al., 2014; Zhang et al., 2015). In contrast, for Co-1, as shown in **Figures 4c,d**, there is no obvious alteration in the hollow multi-porous particle morphology after 100 cycles. These results demonstrate that the hollow multiporous architecture can provide enough void spaces to alleviate the architectural change, and offer a stable structure for the intercalation and de-intercalation cycling (Zhang et al., 2015). In addition, Nitrogen adsorption/desorption isotherms are shown in **Figures 5A–C**, the sorption isotherms are of type III isotherms, they have a distinct H3-type hysteresis loop in the range of P/P<sup>0</sup> = 0.2–0.99, which does not close until the saturation pressure is reached. Obviously, the BET specific surface area (SSA) of the Co-1 is 6.909 m<sup>2</sup> g −1 , which is larger than that of the NM sample (5.199 m<sup>2</sup> g −1 ). The larger SSA can be derived from the hollow multi-porous architecture, leading to a shorter Li ions diffusion distance during the charge-discharge processes.

Furthermore, nyquist plots of the as-prepared electrodes and corresponding EIS and lithium ion diffusion coefficient (DLi+) results are exhibited in **Tables 1**, **2** and **Figures 5D–F**. The semicircle at high and intermediate frequency region relates the SEI film impedance (Rsei) and the charge transfer impedance (Rct), respectively. The inclined straight line at the low frequency region is attributed to Warburg impedance (Rw), which is associated with Li<sup>+</sup> diffusion into the bulk (Qiu et al., 2011; Li et al., 2012; Chen et al., 2014a,b; Sun et al., 2014; Wang L. Y. et al., 2014; Xu et al., 2014; Choi et al., 2015; Wu F. X. et al., 2016; Wu et al., 2018). The lithium ion diffusion coefficient was calculated via a widely accepted method (Reddy et al., 2013; Li et al., 2015). With the increase of cycle number, it is obvious that the values of R for all samples rise. After 100 cycles, the resistance of all samples increase, and both samples (**Figures 4b,d**) show fuzzy surface. These results indicate that the SEI formed on the electrode. In addition, as shown in **Figures 4c,d**, there is no obvious alteration in the hollow multiporous particle morphology after 100 cycles, which suggests that the uniform hollow multi-porous architecture can alleviate the pore size/function change originating from the volume expansion and the SEI formation during cycling. Specially, the



TABLE 2 | The values of specific surface area and DLi+ for the NM and Co-3 samples.


value of R for NM after 1 cycle and 100 cycles is 218.5 and 719.3 , respectively, the dramatical increase of the resistance could be ascribed to the formation and expansion of aggregate particles (as seen in **Figure 4**), which is certainly harmful to electrolyte penetration and the lithium ions and electron transport. In comparison, the Co-substituted samples do not have significant changes, which should be attributed to that the hollow multi-porous architecture can provide enough void spaces to prevent particle aggregation, resulting in enhanced the charge transport kinetics during discharging/charging processes (Zhang et al., 2015). It is clear from the above discussions that the excellent rate capability and cycle stability of Cosubstituted samples are attributed to the porous architecture by following factors: (1) provides enough void spaces to suppress the architectural collapse during cycle, (2) reduces the lithium ions diffusion and electron-transportation distances; (3) facilitates the electrolyte penetration resulting in enhanced the charge transport kinetics.

To further evaluate the effects of the superlattice structure on electrochemical properties of the NiMn2O4/NiCo2O4, the electrochemical kinetics of the Co-3 and Co-1 samples have been studied. The reason why choose Co-3 and Co-1 is partly due to the Co-3 exhibits superior electrochemical properties and superlattice structure, and partly due to Co-1 owns similar porous morphology with the Co-3, while is made up with a single phase. The BET results reveal that the SSA of the Co-1 and Co-3 samples is 6.909 and 7.050 m<sup>2</sup> g −1 , respectively. The similar SSA means the similar Li ions diffusion distance during the charge-discharge process. Therefore, it is supposed that the electrochemical kinetic difference between the Co-1 and Co-3 mainly lies on structure. As shown in **Table 1**, the Co-3 exhibits the lowest initial R and the most stable resistance during cycling among all samples. Furthermore, it can be seen from **Table 2**, the DLi+ of the Co-3 sample is 1.19 × 10−<sup>11</sup> cm<sup>2</sup> S −1 , which is much higher than that of the Co-1 sample (3.49 × 10−<sup>12</sup> cm<sup>2</sup> S −1 ). Since the morphology and the SSA of the Co-1 is similar to those of Co-3, the superior electron and Li ion transport diffusive capacity of the Co-3 should be related to the superlattice NiMn2O4/NiCo2O<sup>4</sup> structure. A similar phenomenon has also been reported by Chen et al (Suo et al., 2013). They pointed out that the composite phases can supply high ionic conductivity. These results demonstrate that the superlattice NiMn2O4/NiCo2O<sup>4</sup> structure can facilitate the ionic conductivity to enhance charge transport kinetics of the bulk material.

The schematic illustrations of the porous hollow superlattice NiMn2O4/NiCo2O<sup>4</sup> compound and the mechanism of electrochemical process are exhibited in **Scheme 1**. Many reports (Kang et al., 2015; Ma Y. et al., 2015; Ma Z. et al., 2015; Zhang et al., 2015, 2016) indicate that the aggregation from the irreversible volume expansion and the low charge transport kinetics, including the poor intrinsic electronic conductivity and the slow ion-diffusion rates, is responsible for the fading capacity of NiMn2O4. In this work, the porous hollow architecture can provide enough void spaces to suppress architectural collapse originating from the volume expansion during cycling, and offer a short stable path for the lithium

ions diffusion and electron-transportation. Moreover, the superlattice NiMn2O4/NiCo2O<sup>4</sup> structure can facilitate the electronic and ionic conductivity to enhance charge transport kinetics of the bulk material. The synergistic effects of the superlattice NiMn2O4/NiCo2O<sup>4</sup> structure and the uniform hollow multi-porous architecture are guaranteed for the excellent electrochemical performance.

### CONCLUSIONS

In summary, the Co substituted NiMn2O<sup>4</sup> mesocrystals with controlled phase structure and morphology are successfully synthesized via a simple solvothermal method. The optimal sample with well-distributed hollow multi-porous superlattice NiMn2O4/NiCo2O<sup>4</sup> mesocrystals exhibits excellent rate ability and enhanced cycling performance, which maintains 532.2 mAh g−<sup>1</sup> with 90.4% capacity retention after 100 cycles, and can still deliver a stable capacity of about 360.8 mAh g −1 after 500 cycles, at a current density of 1 A g−<sup>1</sup> . The excellent electrochemical performance can be attributed to the synergistic effect of the superlattice NiMn2O4/NiCo2O<sup>4</sup> structure and the uniform hollow multi-porous architecture. The superlattice structure can enhance the Li ion diffusion coefficient to enhance the charge transport kinetics. While, the uniform hollow multi-porous architecture can alleviate the architectural change originating from the volume expansion during cycling and facilitate electrolyte penetration. These combined effects

### REFERENCES

Chen, Y. J., Zhu, J., Qu, B. H., Lu, B., and Xu, Z. (2014a). Graphene improving lithium-ion battery performance by construction of NiCo2O4/graphene hybrid nanosheet arrays. Nano Energy 3, 88–94. doi: 10.1016/j.nanoen.2013.10.008

make the NiCo1.5Mn0.5O<sup>4</sup> a promising anode material for the practical application in EV/HEV and energy storage systems.

### AUTHOR CONTRIBUTIONS

LL and QY conceived the idea; QY and LL prepared all materials; QY, HY, and JL conducted SEM experiments; KY, BL, and ZL conducted XRD experiments; LL, QY, and BZ analyzed the data; QY wrote the manuscript and BZ, ZC, and JD commented on it; LL supervised the implementation of the project.

### ACKNOWLEDGMENTS

This work was financially supported by the National Natural Science Foundation of China (No. 51774051 and No. 51404156), the Changsha City Fund for Distinguished and Innovative Young Scholars (No. KQ1707014), Scientific Research Fund of Hunan Provincial Education Department (No. 17B002), the Key Laboratory of Efficient and Clean Energy Utilization, The Education Department of Hunan Province (No. 2016 NGQ005).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2018.00153/full#supplementary-material

Chen, Y. J., Zhuo, M., Deng, J. W., Xu, Z., Li, Q. H., and Wang T. H., (2014b). Reduced graphene oxide networks as an effective buffer matrix to improve the electrode performance of porous NiCo2O<sup>4</sup> nanoplates for lithium-ion batteries. J. Mater. Chem. A 2, 4449–4456. doi: 10.1039/c3ta1 4624c


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The reviewer, BQ, and handling Editor declared their shared affiliation.

Copyright © 2018 Li, Yao, Liu, Ye, Liu, Liu, Yang, Chen, Duan and Zhang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# SiC Nanofibers as Long-Life Lithium-Ion Battery Anode Materials

Xuejiao Sun1†, Changzhen Shao2†, Feng Zhang<sup>2</sup> , Yi Li <sup>2</sup> \*, Qi-Hui Wu<sup>1</sup> \* and Yonggang Yang<sup>2</sup>

*<sup>1</sup> Department of Materials Chemistry, School of Chemical Engineering and Materials Science, Quanzhou Normal University, Quanzhou, China, <sup>2</sup> Jiangsu Key Laboratory of Advanced Functional Polymer Designand Application, Department of Polymer Science and Engineering, College of Chemistry, Chemical Engineering and Materials Science, Soochow University, Suzhou, China*

The development of high energy lithium-ion batteries (LIBs) has spurred the designing and production of novel anode materials to substitute currently commercial using graphitic materials. Herein, twisted SiC nanofibers toward LIBs anode materials, containing 92.5 wt% cubic β-SiC and 7.5 wt% amorphous C, were successfully synthesized from resin-silica composites. The electrochemical measurements showed that the SiC-based electrode delivered a stable reversible capacity of 254.5 mAh g−<sup>1</sup> after 250 cycles at a current density of 0.1 A g−<sup>1</sup> . It is interesting that a high discharge capacity of 540.1 mAh g <sup>−</sup><sup>1</sup> was achieved after 500 cycles at an even higher current density of 0.3 A g−<sup>1</sup> , which is higher than the theoretical capacity of graphite. The results imply that SiC nanomaterials are potential anode candidate for LIBs with high stability due to their high structure stability as supported with the transmission electron microscopy images.

Keywords: nanofibers, silicon carbide, anode, lithium-ion batteries, long-life

### INTRODUCTION

In recent years, due to the wide applications of lithium ion batteries (LIBs) in the portable electronic devices and especially in electric vehicles (EVs), higher criterion on their energy capacity and density as well as cycleability would be established. To further enhance the electrochemical performances of LIBs, it is necessary to find alternative electrode materials to replace the currently commercial ones e.g. graphitic carbons (An et al., 2017a,b; Zhang et al., 2018). SiC, a high bandgap semiconductor, is generally thought to be electrochemically inactive to Li, and often acts as backbone or buffer matrix to enhance the strength of electronic composite materials (Timmons et al., 2007; Jeon and Lee, 2014; Wang C. et al., 2015; Wang W. et al., 2015). But it is interesting that some researchers have studied the cycling performance of SiC-based nanocomposites as anode materials for LIBs. For example, nanoscopic carbon-coated SiC has been synthesized using a simple CVD method (Sri Devi Kumari et al., 2013), which delivered a reversible lithium storage capacity of about 1,200 mAh g−<sup>1</sup> after 200 cycles, showing excellent cyclability. Zhang and Xu (2014) prepared a nanocrystalline SiC thin film electrode grown on the stainless steel, which showed a stable discharge capacity of 309 mAh g−<sup>1</sup> over 60 cycles. Hu et al. (2016) produced SiC nanowires on the graphite paper, which delivered 397 mAh g−<sup>1</sup> after 100 cycles. These results indicated that when a non-conducting SiC material reduces to nano size, it can also serve as anode materials for LIBs. As explained by Li et al. (2016), the capacity of SiC nanomaterials could be activated upon an intimate contact of SiC with graphite, which facilitates the electronic and ionic transport while suppressing the oxidation of SiC. Moreover, during the electrochemical studies of the SiC nanomaterials, the gradual enhancement of capacities was found, which has not been well studied.

Edited by:

*Qiaobao Zhang, Xiamen University, China*

Reviewed by:

*Shaohua Guo, Nanjing University, China Biao Gao, Wuhan University of Science and Technology, China Huiqiao Li, Huazhong University of Science and Technology, China*

\*Correspondence:

*Yi Li liyi@suda.edu.cn Qi-Hui Wu qhwu@qztc.edu.cn*

*†These authors have contributed equally to this work.*

#### Specialty section:

*This article was submitted to Physical Chemistry and Chemical Physics, a section of the journal Frontiers in Chemistry*

> Received: *12 March 2018* Accepted: *23 April 2018* Published: *14 May 2018*

#### Citation:

*Sun X, Shao C, Zhang F, Li Y, Wu Q-H and Yang Y (2018) SiC Nanofibers as Long-Life Lithium-Ion Battery Anode Materials. Front. Chem. 6:166. doi: 10.3389/fchem.2018.00166* Sun et al. SiC Anode Materials

Up to date, the reports on SiC as anode material for LIBs are few. As a potential material for long-life LIBs, the lithium ions diffusion/reaction mechanism in/with SiC nanostructures needs to be further investigated, which would benefit to later similar studies.

SiC nanostructures can be synthesized by converting well defined SiO<sup>2</sup> nanoarchitectures through carbothermal reduction at a high temperature over 1,400◦C, by which the SiC products could preserve the structure regularity of the starting SiO<sup>2</sup> nanomaterials (Miller et al., 1979; Martin et al., 1998). Nanosized SiC materials can also be fabricated by pyrolysis of Sicontaining polymers such as polybissilsesquioxanes (Wang C. et al., 2015; Zhang et al., 2015). In our previous work, we reported SiC/C composite nanotubes (69 wt% SiC and 31 wt% C) prepared via pyrolysis of resorcinol-formaldehyde (RF) resinsilica composite nanotubes at 1,400◦C under argon atmosphere. The RF resin-silica composite was synthesized using a sol-gel transcription method taking a chiral gelator as template (Shao et al., 2017b). Herein, SiC nanofibers (with 7.5 wt% residual C) were fabricated using RF resin-silica composite nanofibers as the starting material and then applied as electrode materials for LIBs. The electrochemical data showed that the SiC-based electrode possessed superior lithium ion storage capability and cycling stability. Further, the electrochemical reaction mechanism of lithium and SiC nanofibers was proposed.

### EXPERIMENTAL

### General Methods

Field emission scanning electron microscopy (FE-SEM, 4800 instrument) and transmission electron microscopy (TEM, FEI TecnaiG220) images were taken at 3.0 and 200 kV to observe the samples' morphologies and nanostructures, respectively. A Jobin Yvon Horiba HR 800 LabRAM confocal microprobe Raman system was applied to collect the Raman spectrum with Ar laser excitation at 514.5 nm and a power of 10 mW. Wide angle Xray diffraction (WAXRD) patterns were obtained based on an X'Pert-Pro MPD X-ray diffractometer with Cu Kα radiation (1.542Å). Specific surface area and pore-size distribution were determined according to N<sup>2</sup> adsorption isotherm measured from a Micromeritics Tristar II 3020 instrument via the Brunauer-Emmett-Teller (BET) and Barrett-Joyner-Halon (BJH) methods.

### Synthetic Procedure of RF Resin-Silica Composite and SiC Nanofibers

Gelator D-1, whose molecular structure is shown in Figure S1, was synthesized according to the literature (Li et al., 2013; Zhang and Xu, 2014). A typical synthesis route of the resin-silica composite was as following: D-1 (200 mg, 0.31 mmol) and resorcinol (180 mg, 1.63 mmol) were dissolved in deionized water (35 mL) and ethanol (5.0 mL) mixed solution at 40◦C under vigorous stirring. After that, concentrated ammonia aqueous solution (0.6 mL), formaldehyde (0.23 mL), and tetraethylorthosilicate (TEOS, 0.8 mL, 3.6 mmol) were added into the solution in sequence. The reaction mixture was then stirred at 40◦C for 10 h followed by heating to 80◦C and standing in static for 24 h. After filtration and washing with water, the as-prepared sample was brown-reddish powders and denoted as **S1**. Thereafter, sample **S1** was annealed at 350◦C for 2.0 h and then at 1,400◦C for 4.0 h with a heating rate of 2.0◦C· min−<sup>1</sup> in Ar atmosphere to gain the as-synthesized SiC product, denoted as **S2**. After the sample was cooled down to room temperature naturally, it was immersed into10 wt% HF aqueous solution for 24 h at room temperature to completely remove the resident SiO2, and finally washed with deionized water for three times.

### The Electrochemical Measurements

Electrochemical tests were performed using CR2016 coin-type cells assembled in an Ar-filled glove box. For preparation of the working electrode, **S2** (80 mg), acetylene black (AB, 10 mg) and binder polyvinylidene fluoride (10 mg, dissolved in Nmethylpyrrolidone) were mixed together. The resultant slurry was then uniformly coated on a Cu foil current collector and dried overnight under vacuum. Electrochemical cells were assembled using **S2** electrode as the cathode, metallic lithium foil as the anode, Celgard 2325 porous film as the separator, and 1.0 M LiPF<sup>6</sup> solution (dissolved in a mixed solvent of ethylene carbonate and diethyl carbonate, 1:1 by volume) as the electrolyte. The cells were charged and discharged between 3.0 and 0.01 V at room temperature.

### RESULTS AND DISCUSSION

In this work, the RF resin-silica composite **(S1)** was used as the source to fabricate aimed product-SiC as mentioned in the experimental section. The template and RF resin in **S1** was firstly converted to C at lower temperature (∼600◦C) when heated in Ar. When the temperature rose up to 1,400◦C, the silica in **S1** reacts with the C to produce SiC. The obtained sample **S2** was firstly characterized using WAXRD and Raman spectroscopy. As shown in **Figure 1**, in the WAXRD pattern (**Figure 1A**), peaks displayed at 2θ = 35.6, 41.4, 60.1, and 71.9◦ are indexed to (111), (200), (220), and (311) planes of cubic β-SiC phase (JCPDS: 29- 1129). At the same time, a broad peak centered at 2θ = 22.9◦ is ascribed to the (200) plane of graphite. There are not any diffraction lines that could be assigned to SiO2, which suggests that the SiO<sup>2</sup> compound has been safely removed. In the Raman spectrum (**Figure 1B**), the D and G bands appear at 1,331 and 1,581 cm−<sup>1</sup> , respectively. The value of IG/I<sup>D</sup> is 0.95, indicating that the carbon in sample **S2** is predominantly amorphous. Due to the relatively low intensities of SiC Raman shifts compared to those of C, it is difficult to define the SiC Raman peaks in **Figure 1B**. TGA analysis, shown in **Figure 2**, discloses that ∼7.5 wt% of free carbon presents in the sample **S2**. These results show that **S1** has been successfully converted to crystalline SiC with small amount of residual carbon.

FE-SEM and TEM images of the samples **S1** and **S2** are shown in Figure S2 and **Figure 3**. Sample **S1** are right-handed helical nanofibers with width, wall thickness and helical pitch of 80–150, 15–30, and 800–1,200 nm, respectively. The lengths of the nanofibers are several microns as observed in Figure S2. After pyrolysis, the sample **S2** exhibits twisted fiber-like morphology, as shown in **Figures 3a,b**. The average length and diameter of the nanofibers are several microns and 30–50 nm.

In the HRTEM image (**Figure 3c**), the interplanar distance is 0.25 nm, which is consistent with the lattice fringes of cubic β-SiC (111) face. The selected area electron diffraction (SAED) pattern (**Figure 3d**) implies that the obtained SiC is highly crystalline. These observations are consistent with the results of WAXRD analysis in **Figure 1**. The N<sup>2</sup> adsorptiondesorption isotherms and the BJH pore size distribution plots calculated from the adsorption branch for the sample **S2** are shown in **Figure 4**. It shows type-IV isotherms with H3 hysteresis loop, which are the characteristic of microporous and mesoporous materials. These micro- and mesopores are produced mainly due to the breakup of C-C bonds and formation of crystalline SiC during the pyrolysis process as well as the dissolution of SiO<sup>2</sup> compound during the HF solution treatment. The BET specific surface area and BJH adsorption pore volume are 326.4 m<sup>2</sup> ·g −1 and 0.86 cm<sup>3</sup> ·g −1 , respectively.

Cycling performance of the **S2** electrode was evaluated, as shown in **Figure 5A**. The first discharge and charge capacities are 309.3 and 221.9 mAh g−<sup>1</sup> , respectively, giving an initial Coulombic efficiency of 71.7%, however, which quickly increases and keeps at above 97% in the following cycles. The discharge capacity quickly decreases to 195.4 mAh g−<sup>1</sup> in the 8th cycle, thereafter stabilizes at about 170 mAh g−<sup>1</sup> and rises slowly to 205.4 mAh g−<sup>1</sup> in the 250th cycle. Apparently, the relatively low capacity of the **S2** electrode is primarily attributed to its low carbon content and the electrochemically inertial nature of SiC. Moreover, the cycling performance of the **S2** electrode tested at a higher current density of 0.3 A g−<sup>1</sup> is also shown in **Figure 5B**. In the initial 40 cycles, the discharge capacity decreases from 247.9 to 101.8 mAh g−<sup>1</sup> . After that, it starts to increase gradually, and reversible capacities of 254.5 mAh g−<sup>1</sup> at the 250th cycle and 540.1 mAh g−<sup>1</sup> at the 500th cycle are obtained. It seems that high current density will accelerate the activation of SiC electrode. Clearly, the SiC electrode exhibits higher capacity than the commercial graphite electrode (∼370 mAh g−<sup>1</sup> ) and also better cycling stability than the common Si electrode (Zhang et al., 2014).

It's interesting to observe that the capacity increases after long cycles at the current density of 0.3 A g−<sup>1</sup> , as displayed in **Figure 5B**. We firstly reported similar capacity ascending curve for the SiC/C composite (containing 17 wt% carbon, at the current density of 0.1 A g−<sup>1</sup> ), which was attributed to a gradually activation of electrode in the lithiation/delithiation process similar to some metal oxides (Cheng et al., 2013; Wang C. et al., 2015; Sun et al., 2016; Li J. et al., 2017; Shao et al., 2017a). Zhang and Xu also reported the capacity increasing phenomenon for the nanocrystalline SiC from the second cycle. They proposed the reduction of Si4<sup>+</sup> to Si<sup>0</sup> and the formation of Li-C and Li-Si alloys in the electrode (Zhang and Xu, 2014). This is more rather correct explanation, because a high current density will

from 0.1 A g−<sup>1</sup>

to 5 A g−<sup>1</sup>

; (D) Nyquist plot for S2 electrode in the frequency range from 100 KHz to 10 MHz before and after cycling.

fasten the reaction of Li with SiC nanomaterials to form Li-Si and Li-C alloys. Consequently, after tens times of cycling, the SiC electrode will slowly convert to Si/C nanostructure composite, because the re-formation of SiC from Si/C composite needs at rather high temperature. In this case, previously electrochemical inert SiC electrode gradually transfer to electrochemically active Si/C composite electrode, which will interpret why the electrode capacity keeps growing at 0.3 A g−<sup>1</sup> current density. In order to prove our hypothesis, the **S2** electrode was cycled at 5 A g−<sup>1</sup> as shown in Figure S3. Its capacity decreases to about 76 mAh g −1 at the 16th cycle, and then gradually increases to 128 mAh g −1 at the 40th cycle. It is found that the capacity enhancement is much earlier and faster and the electrodes cycled at 0.1 and 0.3 A g−<sup>1</sup> . It is very interesting to find that after SiC partially transfer to Si/C composite, it still maintains original nanofiber structure. This may explain why the SiC electrode could possess a rather good cycling stability. FE-SEM images of the **S2** electrode after cycling test were then taken and showed in **Figure 6**. The nanofibers keep original morphology, and no changes in lengths or diameters are observed. Figure S4 shows the XRD patent of the electrode after cycling at 5 A g−<sup>1</sup> for 40 times. It could be seen that the bump for C species slightly increases and the diffractions for the SiC decreases, which indicates the change of component in the SiC/C composite. Out of our expectation, the signals for Si were not observed in the patent, which may because that the Si decomposed from SiC is not crystalline but amorphous.

Unlike Si-based electrodes, which suffer from huge volume changes during the lithiation/delithiation process, resulting in the pulverization of Si nanostructure and consequently shortening the cycling properties of the batteries. SiC nanostructures have

at 0.1 A g−<sup>1</sup> ; (b) 500 cycles at 0.3 A g−<sup>1</sup> .

rather high hardness, the partial decomposition of SiC into to Si and C composites would not alter the base structure of SiC, meanwhile, the decomposed Si nanoclusters could well distribute into the C matrix. This may explain why after longtime cycling the Si/C composite conversed from SiC/C still


\**1C* = *0.37 A g*−<sup>1</sup> *;* \*\**the data wasn't marked in Timmons et al. (2007).*

preserves nanofiber structure. The Raman spectra of **S2** electrode after cycling have been measured and are showed in Figure S5. The values of IG/I<sup>D</sup> are 0.86 at the current density of 0.1 A g−<sup>1</sup> after 250 cycles and 0.94 at the current density of 0.3 A g−<sup>1</sup> after 500 cycles, respectively. Compared with the sample before cycling (the IG/I<sup>D</sup> value is 0.95), the graphitization change is small. It is reasonable to suggest that the decomposed C from SiC is also amorphous, because under the electrochemical reactions it is not easy to form crystalline carbon nanostructure during lithiation/delithiation.

Compared with our previously reported results, the C content in the SiC/C composites exhibited great effect on the capacity, as listed in **Table 1**. At low current density (0.1 A g−<sup>1</sup> ), pure carbon shows the biggest capacity value, and the small amount of SiC will lower the capacity because of its electrochemical inertness. When SiC is the major moiety, the capacity decreases with the content of C after long cycles. However, at higher current density (over 0.3 A g−<sup>1</sup> ), the SiC/C composites show higher capacity than pure carbon after long cycles, because high current density will accelerate the activation of SiC electrode. The SiC transforms into Si/C composite keeping original SiC nanostructure (Han et al., 2011). The theoretical capacity of Si is much higher than that of C (Ren et al., 2013, 2014; Sun et al., 2015; Li H. et al., 2017), therefore, SiC/C composites would show both higher capacity than C materials and superior stability than Si materials. The relative lower capacity values obtained in current work is due to the lower original C content, which may slowly increase with the cycling number.

The rating capability of the **S2** electrode was evaluated at various current densities from 0.1 to 5 A g−<sup>1</sup> , as shown in **Figure 5C**. Reversible capacities of 165.1, 111.7, 76.2, 62.9, 51.5, 110.1, and 160.1 mAh g−<sup>1</sup> are obtained, respectively, indicating excellent rating capability. To study the conductivity of the **S2** electrode, the electrochemical impedance spectra (EIS) were also collected and shown in **Figure 5D**, which illustrates an impedance of 180 before cycling, 120 after 100 cycles, and 50 after 500 cycles. The rapidly decreased impedance implies that the SiC electrode after cycling also possesses a good electrical conductivity and a rapid charge-transfer reaction for Li ion insertion and extraction. This result is consistent with our above proposal that SiC would gradually decompose into Si and C during electrochemical Li insertion/extraction.

### CONCLUSIONS

Novel SiC nanofibers derived from resorcinol-formaldehyde resin/silica composites were designed and synthesized successfully. When used as anode materials for LIBs, they exhibited superior cycling stability, good rating capability and low impedance. A special capacity increasing phenomenon was observed after long cycles, which was ascribed to the partial decomposition of SiC nanostructure into Si/C composites during the lithiation/delithiation process. In addition, the effect of C content in the SiC/C composite was compared and discussed. The results in this work disclose that the nano-sized SiC materials are promising anode candidate for long-life LIBs due to their high nanostructure stability.

### AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

### ACKNOWLEDGMENTS

This work was supported by the National Natural Science Foundation of China (Nos. 21574095, 51473106), the Priority Academic Program Development of Jiangsu High Education Institutions (PAPD), the Natural Science Foundation of Fujian Province, China (Grant No. 2016J01069).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2018.00166/full#supplementary-material

### REFERENCES


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Sun, Shao, Zhang, Li, Wu and Yang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# N/S Co-doped Carbon Derived From Cotton as High Performance Anode Materials for Lithium Ion Batteries

Jiawen Xiong<sup>1</sup> , Qichang Pan<sup>1</sup> , Fenghua Zheng<sup>1</sup> , Xunhui Xiong<sup>1</sup> , Chenghao Yang1,2 \*, Dongli Hu<sup>3</sup> and Chunlai Huang<sup>3</sup>

*<sup>1</sup> Guangzhou Key Laboratory for Surface Chemistry of Energy Materials, New Energy Research Institute, School of Environment and Energy, South China University of Technology, Guangzhou, China, <sup>2</sup> Guangdong Engineering and Technology Research Center for Surface Chemistry of Energy Materials, New Energy Research Institute, School of Environment and Energy, South China University of Technology, Guangzhou, China, <sup>3</sup> Jiangsu Key Lab of Silicon Based Electronic Materials, Jiangsu GCL Silicon Material Technology Development Co., Ltd, Xuzhou, China*

### Edited by:

*Qiaobao Zhang, Xiamen University, China*

### Reviewed by:

*Jinhu Yang, Tongji University, China Yunxiao Wang, Institute for Superconducting and Electronic Materials, Australia Hui Xia, Nanjing University of Science and Technology, China*

\*Correspondence:

*Chenghao Yang esyangc@scut.edu.cn*

### Specialty section:

*This article was submitted to Physical Chemistry and Chemical Physics, a section of the journal Frontiers in Chemistry*

> Received: *30 January 2018* Accepted: *08 March 2018* Published: *26 April 2018*

#### Citation:

*Xiong J, Pan Q, Zheng F, Xiong X, Yang C, Hu D and Huang C (2018) N/S Co-doped Carbon Derived From Cotton as High Performance Anode Materials for Lithium Ion Batteries. Front. Chem. 6:78. doi: 10.3389/fchem.2018.00078* Highly porous carbon with large surface areas is prepared using cotton as carbon sources which derived from discard cotton balls. Subsequently, the sulfur-nitrogen co-doped carbon was obtained by heat treatment the carbon in presence of thiourea and evaluated as Lithium-ion batteries anode. Benefiting from the S, N co-doping, the obtained S, N co-doped carbon exhibits excellent electrochemical performance. As a result, the as-prepared S, N co-doped carbon can deliver a high reversible capacity of 1,101.1 mA h g−<sup>1</sup> after 150 cycles at 0.2 A g−<sup>1</sup> , and a high capacity of 531.2 mA h g <sup>−</sup><sup>1</sup> can be observed even after 5,000 cycles at 10.0 A g−<sup>1</sup> . Moreover, excellently rate capability also can be observed, a high capacity of 689 mA h g−<sup>1</sup> can be obtained at 5.0 A g−<sup>1</sup> . This superior lithium storage performance of S, N co-doped carbon make it as a promising low-cost and sustainable anode for high performance lithium ion batteries.

Keywords: lithium-ion batteries, anode materials, sustainable, cotton, N/S co-doped carbon

### INTRODUCTION

In the past decade, lithium-ion batteries (LIBs) have been widely used as power sources for computing, communications, consumer batteries (3C battery), and electric vehicle. And LIBs with a wide range of applications which due to its have high working voltage, high energy density, and long service life (Goodenough and Kim, 2010; Goodenough and Park, 2013; Zheng et al., 2015). Currently, graphite is mostly used as anode material for commercial LIBs due to its good electronic conductivity, low cost and outstanding cycling stability. However, graphite can not meet the increased energy and power density of high performance LIBs due to the low specific capacity and poor rate performance (Huang et al., 2016; Wang et al., 2017; Pan et al., 2018). On the other hand, the low lithiation/de-lithiation potentials (< 0.3 V vs. Li+/Li) which resulting in seriously security issue (Guo et al., 2011). Therefore, it is necessary to develop novel anode materials to replace graphite anode for high-performance LIBs (Pan et al., 2017a).

Recently, amorphous carbon and hard carbon are attracted attentions as promising anode to replace graphite anode for LIBs, which due to these carbonaceous materials exhibit higher specific capacity and offer higher lithiation/de-lithiation potential (Wang et al., 2009; Casas and Li, 2012; Tang et al., 2012). Moreover, these carbonaceous materials with partially graphitic carbons which can accommodate Li<sup>+</sup> in the disordered interlayers as well as in the micropores, and can exhibit excellent cycling stability and rate capability (Wu et al., 2003). On the other

**419**

hand, other carbonaceous materials such as carbon nanotube, graphene and fullerenes were also developed as anode for LIBs, and exhibit excellent electrochemical performance (Etacheri et al., 2015; Wang et al., 2015). However, in order to synthesize these carbonaceous materials which rely on hydrocarbon precursors, resulting in expensive cost and commercially nonviable. Therefore, it is necessary to explore a scalable and inexpensive precursor as carbon sources for these carbonaceous materials as anode materials for LIBs.

Nowadays, multitudinous biomass raw materials have been extensively used as precursor for carbonaceous materials and application in LIBs. So many biomass raw materials attracted attention such as peanut (Ding et al., 2015), ramie (Jiang et al., 2016), sisal (Yu et al., 2015), bamboo (Jiang et al., 2014), green tea leaves (Han et al., 2014), peat moss (Ding et al., 2013), rice husk (Wang et al., 2013), banana peel (Lotfabad et al., 2014), and so on. However, cotton attracted more attention and have been considered as the most promising compared with the other biomass materials due to its abundant and low cost. On the other hand, in China, cotton is planted around 550 Million tons per year and giant cotton-products are abandoned which from clothes, medical alcohol cotton, and so on. Moreover, lots of abandoned cotton-products will bring lots of problems, such as environmental pollution, safe question, and so on.

Herein, we addressed the above mentioned issues by prepared high performance carbon-based anode materials using cotton as precursor, as shown in **Figure 1**. Highly porous carbon with large surface areas were prepared from cotton via s sample method. And the N/S-coped carbon were further obtained using thiourea as nitrogen and sulfur sources. When evaluated as anode materials for LIBs, these carbon materials exhibit outstanding rate capability and long-term cycling stability.

### EXPERIMENTAL

### Material Preparation

1.5 g cotton was dipped in homogeneous Mg(NO3)<sup>2</sup> solution (8 mol L−<sup>1</sup> , 20 ml), then dried in the oven. After that, the obtained cotton were annealed at 800◦C under N<sup>2</sup> atmosphere for 3 h with a heating rate of 5◦C min−<sup>1</sup> . After cooling, the obtained cotton carbon (denoted as CC) were washed with 1 M HCl and distilled water several times, respectively, then dried in the oven.

To obtained the N, S co-doped carbon, the obtained CC were immersed in 100 ml thiourea solution with ratio of 10:1. After drying, the obtained powders were calcined at 800◦C under N<sup>2</sup> atmosphere for 3 h. The N, S co-doped cotton carbon (denoted as NS-CC) powders were obtained after cooling.

### Material Characterizations

The XRD patterns of all samples were conducted on the Bruker D8 Advance (Germany) (Cu, Kα, λ = 1.5405 Å). Raman spectra were obtained on a JOBIN-Yvon HR800 Raman spectrometer. Morphology of the all samples were studied by SEM (FEI Quanta 200 FEG) and TEM (Tecnai G2 F20 S-TWIN, Japan). BET method and non-linear density functional theory (NLDFT) (ASAP 2020 Micromeritics) were applied to test the specific surface area and the pore size distribution.

### Electrochemical Measurements

The samples were executed in CR2025 coin cells. The mingled ratio of sample CC, sample NS-CC and carbon black and poly (vinylidenedifluoride) were 7:2:1 (the loaded active electrode materials are about 0.5 mg cm−<sup>2</sup> ). The mixture was coating on copper foil to prepare electrodes. Metal lithium boil as counter electrode and 1 M LiPF<sup>6</sup> dissolved in ethylene carbonate (EC) and dimethyl carbonate (DMC) (1:1, v/v) as electrolyte. CV and EIS were conducted on a CHI660A electrochemical workstation. Galvanostatic charge/discharge and cycling performance were executed at 25◦C based on the active electrode material corresponding specific capacity.

### RESULTS AND DISCUSSIONS

The morphology of the all samples was characterized by SEM firstly. **Figures 2A,C** exhibit the SEM images of cotton carbon, which shows micron size bulk materials and composed of nanosheets. **Figures 2B–D** shows the SEM images of the cotton carbon after N and S co-doped, which exhibits similar morphology to cotton carbon. Moreover, cotton carbon and S, N co-doped cotton carbon with a large number of pores in the nanosheets according to the HRSEM (**Figures 2C,D**). On the other hand, EDS element mapping of S, N co-doped cotton carbon were investigated, which indicated that C, N, O, and S elements exist S, N co-doped cotton carbon. Furthermore, S and N elements evenly distributed (**Figures 2G–J**) in the carbon matrix. Therefore, the results indicated that the N and S elements were successfully doped into cotton carbon after heat treatment the carbon in presence of thiourea. The microstructure of CC and NS-CC was further studied by TEM, and the results are shown in **Figures 2E,F**. It clearly illustrates that the samples are amorphous carbon with nano/meso porous structure (Chen et al., 2014). Furthermore, typical selected area electron diffraction further proved that the carbon were amorphous, which corresponding to TEM results (Zhu and Akiyama, 2016).

The XRD patterns of the CC and NSCC are showed in **Figure 3A.** There are no obvious peaks for magnesium compounds in CC sample, which indicated that the Mg(NO3)<sup>2</sup> or MgO are removed completely by washed with dilute hydrochloric acid. And two broad peaks at around 23 and 43◦ can be observed both at CC and NS-CC samples, which can be attributed to (002) and (100) graphitic planes, respectively (Hou et al., 2015; Chen et al., 2017). The Raman spectra of the CC and NS-CC are shown in **Figure 3B**, two peaks at around 1,361 and 1,596 cm−<sup>1</sup> were obtained, which corresponding to the D band and G band for carbon materials, respectively (Li et al., 2015; Gao et al., 2017). Furthermore, the D band arises from edges, defects, and disordered carbon, whereas the G band is ascribed to sp<sup>2</sup> hybridized carbon (Pan et al., 2017b). Therefore, a high ID/I<sup>G</sup> band intensity ratio indicates the generation of large amounts of defects. The ID/I<sup>G</sup> ratio for NS-CC is higher than that of CC, which indicated that more vacancies and defects generated by doping N and S atoms into the carbon material. More importantly, more vacancies and defects are beneficial for the transmission of Li-ion and offer more active site for Li storage,

which resulting in improved electrochemical performance (Qie et al., 2015; Lu et al., 2017).

The XPS measurement was conducted to confirmed that the presence of N and S elements in NS-CC. As shown in Figures S1A,B, the survey spectrum for NS-CC and CC exhibit two predominant peak at around 284.8 and 532 eV can be observed, which can be assigned to C and O. It certified for the NS-CC that N and S atoms were successfully doped corresponding with EDS element mapping. In Figures S1C,D, the C 1s XPS spectrum for both CC and NS-CC can be deconvoluted into five peaks, which corresponding to C = C, C-C/C = N, C-O, C = O and π-π ∗ (Liu et al., 2017), respectively. The high resolution N 1s spectrum as shown in **Figure 3E**, the N 1s speak can be fitted by three component peaks at around 401.9, 400.5, and 398.3 eV, which can be ascribed to graphitic N, pyrrolic N and pyridinic N, respectively (Ou et al., 2015). As for high resolution of S 2p spectrum (**Figure 3F**), there were five peaks attributed to -C-S-C- bond and -C-SOX-C- bond. Therefore, these results indicated that the N and S has been successfully incorporated into the carbon structure of NS-CC. And the content of N and S in NS-CC were confirmed for 3.0 and 1.4%, respectively. The heteroatoms N and S can effectively enlarge the interlayer space because of their lager radius than C atom, resulting in forming the defects and providing more active sites for Li-ions on the carbon materials (Xu et al., 2015, 2016; Xiong et al., 2016). Moreover, pyridinic N and quaternary N are favorable for Li<sup>+</sup> and electrons and the doped S in the carbon materials can participate in the redox reactions contribute to the reversible capacity (Ma et al., 2018).

The nitrogen adsorption/desorption isotherms and the pore size distribution of CC and NS-CC are shown in **Figures 3C,D.** As shown in **Figure 3C** both of the two samples showed type IV isotherms (Islam et al., 2017). The BET specific surface area of CC and NS-CC are 1235.35 and 1326.20 m<sup>2</sup> g −1 , respectively. The BET specific surface areas of NS-CC increased compare than the CC after N and S atoms doping. In addition, the highly porous structure of the two samples were further evaluated by Barrett-Joyner-Halenda (BJH) calculations (**Figure 3D**). It can be seen that the two samples exhibit a broad pore size distribution, and the pore size of the two samples were centered at around 2 and 5 nm, respectively. Therefore, NS-CC exhibits larger specific surface area which can provide more active sites for lithium ion storage. Moreover, the highly porous structure of the two samples can greatly shorten the diffusion distance of both electrons and ions, which resulting in improved rate performance (Hao et al., 2014).

The electrochemical performance of CC and NS-CC was first measured by cyclic voltammetry (CV) with voltage range of 3.0– 0.01 V. **Figure 4A** exhibits the CV curves of the NS-CC sample. During the first discharge cycle, three cathodic peaks at around 1.4 and 0.65 V can be obtained and disappeared in the subsequent cycles, which corresponding to formation of a solid-electrolyte interphase (SEI) film (Wang et al., 2011; Jiang et al., 2013) as well as some irreversible and side reactions associated with the decomposition of electrolyte (Yoshio et al., 2000). Moreover, the CV profiles almost overlapped after the initial scanning cycle, which indicates that the structural stability of the NS-CC electrode during the subsequently cycling. On the other hand, the CV curves of CC electrode (Figure S2A) are similar to the NS-CC electrode.

The charge and discharge profiles of CC and NS-CC electrode at 0.1 A g−<sup>1</sup> with cutoff voltage window of 0.01–3.0 V were

studied. As shown in **Figure 4B** and Figure S2B, two plateaus at around 1.4 and 0.65 V can be observed during the first discharge process, which corresponding to the forming of SEI layer and decomposition of the electrolyte, in good agreement with the above CV results. The first discharge capacity of CC and NS-CC are as high as 3,275.1 and 4,179.1 mA h g−<sup>1</sup> , while the initial reversible capacity are only 1,512.3 and 1,823.3 mA h g −1 , corresponding to low initial Coulombic efficiency of 46.2 and 43.6%, respectively. The large irreversible capacity loss for the CC and NS-CC electrode can be ascribed to formation of a SEI layer on the relatively large specific surface area (Jiang et al., 2013). Furthermore, a mass of reduction of oxygen functionalities on the carbon materials surface (Bhattacharjya et al., 2014) and reduction of electrolyte components on the active electrode of the CC and NS-CC electrode (Hu et al., 2007) also contributed to the irreversible capacity loss.

The cycling performance of the CC and NS-CC electrode were evaluated at 0.2 A g−<sup>1</sup> (**Figure 4D**). NS-CC electrode exhibits excellent cycling stability and a high reversible capacity of 1101.1 mA h g−<sup>1</sup> can be observed after 150 cycles, but for CC electrode, a lower reversible capacity of 637.1 mA h g−<sup>1</sup> can be obtained. Rate performance is very important for LIBs, especially application in electric vehicles. Therefore, the rate performance of the samples were evaluated at various current densities from 0.1 to 5.0 A g−<sup>1</sup> . As seen in **Figure 4C**, NS-CC electrode can deliver reversible capacities of 1,443, 1,035, 954, 884, 802, and 689 mA h g−<sup>1</sup> at 0.1, 0.2, 0.5, 1.0, 2.0, and 5.0 A g−<sup>1</sup> , respectively. In contrast, the reversible capacities are 1,020, 655, 541, 478, 427, and 370 mA h g−<sup>1</sup> at 0.1, 0.2, 0.5, 1.0, 2.0, and 5.0 A g−<sup>1</sup> for CC electrode. Obviously, the NS-CC electrode exhibits excellent rate capability. Therefore, the long-term cycling at high current density was also tested for NS-CC electrode, the results as shown

(E) N spectrum (F) S2p spectrum electrons of NS-CC.

in **Figure 4E**. NS-CC electrode can deliver a high reversible capacity of 531.2 mA h g−<sup>1</sup> can be obtained even after 5,000 cycles at 10.0 A g−<sup>1</sup> . However, CC electrode only delver a lower reversible of 283 mA h g−<sup>1</sup> after 5,000 cycles at the same current density. Therefore, NS-CC electrode exhibits excellent rate performance and long-term cycling stability, which shows better electrochemical performance than previous reported carbonbased materials, as shown in Table S1. As a result, The NS-CC electrode delivered amazing electrochemical performance especially with ultrahigh specific capacity and rate capability which can be given rise to the following reasons: (1) The carbon materials interlayer spacing are expanded by N and S successfully co-doping which benefit Li-ions diffusion. (2) The marked large surface of carbon materials offer plentiful micropores and mesopores structure shorten the diffusion distance, sufficient contact between electrolyte and electrode and active sites for lithium ion storage. (3) The ample pyridinic N, pyrrolic N and -S-C-S- covalent bonds built adequate active sites to improve surface capacity contribution (Xia et al., 2017).

The electrochemical impedance spectroscopy for CC and NS-CC electrode were further investigated to understand the significantly improved electrochemical performance. As shown in Figure S3, the Nyquist plots of CC and NS-CC electrode have shown the typical characteristics of one semicircle and a sloping straight line (Liu et al., 2016). The diameter of the semicircle is reduced in the plots of the NS-CC electrode compared with that of the CC electrode, indicating the decreased charge-transfer resistance at the electrode/electrolyte interface after doping of N, S atoms into the carbon. On the other hand, the charge transfer resistance presents a decreasing trend along with the cycles for both CC and NS-CC electrode, which due to formation of stable SEI film and the process of activation after cycling (Pan et al., 2016).

In order to further understand the high-rate performance, the capacitive behavior of the NS-CC and CC electrode were investigated and their kinetics were also analyzed with CV measurements. **Figure 5A** and Figure S4A show the CV curves of NS-CC electrode at various scan rates ranging from 0.1 to 10 mV s−<sup>1</sup> . All of the curves display a similar shape, two cathodic peaks and one anodic peak are evidently on each curve. The peak current is not proportional to the square root of the sweep rate (v), indicating that the charge/discharge process is comprised of

faradic and non-faradic processes. According to the equation of the relationship of i and v:

$$i = a\nu^b \tag{1}$$

or

$$
\log(i) = b \times \log(\nu) + \log(a)' \tag{1}
$$

Here, a and b are constants. The process is an ionic diffusion controlled behavior when b-value is equal to 0.5, while is Li<sup>+</sup> capacitive behavior when b-value is equal to 1.0. **Figure 5B** presents log(i)-log(v) plots for NS-CC electrode on the CV curves at peak 1, 2 and 3 potentials and the b-value are 0.84, 0.75 and 0.77, respectively. And for CC the b-value are 0.75, 0.71, and 0.97, respectively (Figure S4B). Therefore, It can be seen that all these values of b indicate fast kinetics resulting from the pseudocapacitive effect. Moreover, To quantify the pseudocapacitive contribution, we can divide the current response i at a fixed potential V into pseudocapacitive (k1v) and diffusion-controlled contributions (k2v 0.5) by following equation (Muller et al., 2015).

$$i(V) = k\_1 \nu + K\_2 \nu^{1/2}$$

By calculating both k1 and k2 constants, the overall contribution of pseudocapacitor at various scan rates can be obtained. The detail pseudocapacitive contribution of NS-CC and CC electrode at 10 mV s−<sup>1</sup> as shown in **Figure 5C** and Figure S4C, in which 78.9% as capacitive (red region). Therefore, all contribution ratios of the capacitive capacity at scan rates of 0.1, 0.2, 0.5, 1, 2, and 5 mV s−<sup>1</sup> were also obtained. **Figure 5D** and Figure S4D shows contributions of the pseudocapacitive behaviors at various scan rates. The proportion of capacitive contribution for NS-CC electrode are 29.8, 31.7, 36.9, 43.5, 52.1, 64.7, and 78.9% at 0.1, 0.2, 0.5, 1, 2, 5, and 10.0 mV s−<sup>1</sup> , respectively. Furthermore, the contribution for CC electrode are 27.1, 28.2, 32.3, 41.3, 50.1, 62.6, and 75.6% at 0.1, 0.2, 0.5, 1.0, 2.0, 5.0, and10.0 mv s−<sup>1</sup> . As a

result, the capacitive contribution for CC is smaller than the NS-CC electrode, which can attribute to the NS-CC electrode with larger specific area after N and S atoms co-doping, resulting in enhanced pseudocapacitive contribution (Augustyn et al., 2014; Chao et al., 2016).

## CONCLUSION

In summary, highly porous carbon were prepared by using cotton as precursor with a sample method. Subsequently, the sulfurnitrogen co-doped carbon were obtained via heat treatment the carbon in presence of thiourea, which can induce the defects and the expanded interlayer of the carbon. Therefore, the expanded interlayer and defects can reduce the diffusion distance of Li ions as well as offer more active sites for lithium storage. As a result, S, N co-doped carbon exhibits excellent electrochemical performance when evaluated as anode materials for Lithium-ion materials. The as-prepared S, N co-doped carbon can deliver a high reversible capacity of 546.4 mA h g−<sup>1</sup> even after 5,000 cycles at 10 A g−<sup>1</sup> . Moreover, excellently rate capability also can be observed, a high capacity of 600 mA h g−<sup>1</sup> can be obtained at 5.0 A g−<sup>1</sup> . This superior lithium storage performance of S, N codoped carbon make it as a promising low-cost and sustainable anode material for lithium ion batteries.

## AUTHOR CONTRIBUTIONS

JX conducted the experiments CY is the supervisor of this research work. JX and QP helped writing. JX, FZ, XX, DH, and CH performed the characterization and data analysis. All authors involved the analysis of experimental data and manuscript preparation.

### ACKNOWLEDGMENTS

We gratefully acknowledge the financial support from the Science and Technology Planning Project of Guangdong Province, China (No. 2017B090916002), Guangdong Natural Science Funds for Distinguished Young Scholar (2016A030306010), Guangdong Innovative and Entrepreneurial Research Team Program (2014ZT05N200) and Fundamental Research Funds for Central Universities, China (2017 ZX010).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2018.00078/full#supplementary-material

### REFERENCES


as anode material for high-rate and long life Li-ion batteries. J. Mater. Chem. A 5, 4576–4582. doi: 10.1039/C6TA10932B


**Conflict of Interest Statement:** DH and CH were employed by company Jiangsu Key Lab of Silicon Based Electronic Materials.

The other authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Xiong, Pan, Zheng, Xiong, Yang, Hu and Huang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# High-Level Heteroatom Doped Two-Dimensional Carbon Architectures for Highly Efficient Lithium-Ion Storage

Zhijie Wang1,2†, Yanyan Wang2†, Wenhui Wang3†, Xiaoliang Yu<sup>4</sup> , Wei Lv <sup>2</sup> , Bin Xiang<sup>1</sup> \* and Yan-Bing He<sup>2</sup> \*

*<sup>1</sup> CAS Key Lab of Materials for Energy Conversion, Department of Materials Science and Engineering, Synergetic Innovation Center of Quantum Information Quantum Physics, University of Science and Technology of China, Hefei, China, <sup>2</sup> Engineering Laboratory for the Next Generation Power and Energy Storage Batteries, Engineering Laboratory for Functionalized Carbon Materials, Graduate School at Shenzhen, Tsinghua University, Shenzhen, China, <sup>3</sup> China Key Laboratory of Optoelectronic Devices and Systems of Ministry of Education and Guangdong Province, College of Optoelectronic Engineering, Shenzhen University, Shenzhen, China, <sup>4</sup> Center for Green Research on Energy and Environment Materials, National Institute for Materials Science, Tsukaba, Japan*

#### Edited by:

*Qiaobao Zhang, Xiamen University, China*

### Reviewed by:

*Huan Pang, Yangzhou University, China Chenghao Yang, South China University of Technology, China Bote Zhao, Georgia Institute of Technology, United States*

#### \*Correspondence:

*Bin Xiang binxiang@ustc.edu.cn Yan-Bing He he.yanbing@sz.tsinghua.edu.cn*

*†These authors have contributed equally to this work.*

#### Specialty section:

*This article was submitted to Physical Chemistry and Chemical Physics, a section of the journal Frontiers in Chemistry*

> Received: *16 February 2018* Accepted: *20 March 2018* Published: *05 April 2018*

#### Citation:

*Wang Z, Wang Y, Wang W, Yu X, Lv W, Xiang B and He Y-B (2018) High-Level Heteroatom Doped Two-Dimensional Carbon Architectures for Highly Efficient Lithium-Ion Storage. Front. Chem. 6:97. doi: 10.3389/fchem.2018.00097* In this work, high-level heteroatom doped two-dimensional hierarchical carbon architectures (H-2D-HCA) are developed for highly efficient Li-ion storage applications. The achieved H-2D-HCA possesses a hierarchical 2D morphology consisting of tiny carbon nanosheets vertically grown on carbon nanoplates and containing a hierarchical porosity with multiscale pore size. More importantly, the H-2D-HCA shows abundant heteroatom functionality, with sulfur (S) doping of 0.9% and nitrogen (N) doping of as high as 15.5%, in which the electrochemically active N accounts for 84% of total N heteroatoms. In addition, the H-2D-HCA also has an expanded interlayer distance of 0.368 nm. When used as lithium-ion battery anodes, it shows excellent Li-ion storage performance. Even at a high current density of 5 A g−<sup>1</sup> , it still delivers a high discharge capacity of 329 mA h g−<sup>1</sup> after 1,000 cycles. First principle calculations verifies that such unique microstructure characteristics and high-level heteroatom doping nature can enhance Li adsorption stability, electronic conductivity and Li diffusion mobility of carbon nanomaterials. Therefore, the H-2D-HCA could be promising candidates for next-generation LIB anodes.

Keywords: 2D carbon nanomaterials, hierarchical structure, high-level heteroatom doping, Li-ion batteries, high-rate capability

### INTRODUCTION

Lithium ion batteries (LIBs) have been regarded the most important power sources for portable electronic devices and promising candidates to power future electric vehicles (Armand and Tarascon, 2008; Geng et al., 2018). In order to meet the increasing demand for energy density and fast discharge-charge abilities, it is urgent to develop LIB electrode materials with higher specific capacities, better rate capabilities and excellent cycle stabilities (Arico et al., 2005; Zhang et al., 2018). Graphite has served as the most popular anode materials for its low price, appropriate working voltage platform and high Columbic efficiency (Yazami and Touzain, 1983; Kaskhedikar and Maier, 2009; Lu et al., 2017).

**428**

Unfortunately, it suffers from limited Li storage capacity (372 mA h g−<sup>1</sup> , according to the intercalation mechanism with the formation of LiC<sup>6</sup> composites) and poor rate performance, which cannot satisfy the practical application requirements (Yazami and Touzain, 1983; Lu et al., 2017). Therefore, developing advanced alternative materials to replace graphite has attracted great research interest in recent years (Kaskhedikar and Maier, 2009; Zhang Q. et al., 2016; Wu et al., 2017; Deng et al., 2018).

Current researches have already proved that carbon nanomaterials delivered better LIB performance than graphite, since abundant Li-ion storage sites and rapid ion diffusion channels can be provided (Zhou et al., 2003; Dai et al., 2012; Zhang Q. et al., 2016; Wang et al., 2018). In addition, successful structural modification could further enhance the electrochemical performance of nanocarbons (Wu et al., 2003; Landi et al., 2009). Heteroatom doping plays an important role in the modification because it can adjust the physical and chemical properties of carbon nanomaterials (Wu et al., 2011). For instance, both experimental and theoretical results demonstrated that nitrogen (N) doping can positively affect the electric conductivity and electrochemical activity of nanocarbons (Ma et al., 2012; Zheng et al., 2014). Especially pyrrolic N (N-5) and pyridinic N (N-6) are able to create active sites for Li-ion adsorption in the carbon framework. Hence, increasing the doping concentration of N-5 and N-6 is beneficial to the LIB performance of nanocarbon electrodes (Wang et al., 2011; Ma et al., 2012; Mao et al., 2012; Zheng et al., 2014). Besides, sulfur (S) heteroatoms can enlarge the interlayer distance of carbons because of the larger covalent radius (102 pm) compared with that of C (77 pm) (Qie et al., 2015; Xu et al., 2016). The enlarged interlayer distance facilitates the insertion-extraction of electrolyte ions, and thus is able to improve rate capabilities (Qie et al., 2015; Xu et al., 2016; Liang et al., 2018). It is worth noting that, Li-ion storage performance of carbon nanomaterials can be promoted not only by improving heteroatom doping concentration, but also by the synergistic effects between different kinds of dopants (Ai et al., 2014). Therefore, highlevel N, S co-doping can be an effective strategy to achieve high-performance nanocarbon anodes.

In the previous work, we constructed two-dimensional (2D) hierarchical carbon architectures (2D-HCA) with N, S codoping nature for superior LIB anodes (Wang et al., 2016). Here, we further developed high-level heteroatom doped 2D hierarchical carbon architectures (H-2D-HCA) with enhanced Li-ion storage performance by increasing the N concentration in 2D-HCA. The obtained H-2D-HCA contains a much higher heteroatom concentration of 16.4%, with 15.5% of N and 0.9% of S. Interestingly, the electrochemically active N, i.e., N-5 and N-6, accounts for the majority of total N atoms (47 and 37%, respectively). Furthermore, it has an expanded interlayer space of 0.368 nm compared with that of graphite, which can enhance the Li ion diffusion speed. Benefiting from such high doping level and the unique microstructure, the H-2D-HCA can be used for highly efficient Li-ion storage. To be specific, even at a high current density of 5 A g−<sup>1</sup> , it still delivered a high specific capacity of 329 mA h g−<sup>1</sup> after 1,000 cycles.

## EXPERIMENTAL METHODS

### Preparation of H-2D-HCA

Mg-Al layered double hydroxides (Mg-Al LDH) and Mg-Al layered double oxides (Mg-Al LDO) were prepared following our previous work (Wang et al., 2016). Typically, Mg(NO3)2·6H2O (12.82 g, 99.9%, Macklin), Al(NO3)2·9H2O (9.38 g, 99.9%, Macklin) and urea (90.09 g, 99.9%, Macklin) were dissolved into 500 ml water in a round-bottom flask. Then the solution was heated 100◦C and kept for 12 h under magnetic stirring. After that, the temperature was decreased to 94◦C and kept for another 12 h without stirring. The produced white powder was collected with vacuum filtration and freeze dried, then Mg-Al LDH was obtained. As-prepared Mg-Al LDH was calcined at 500◦C for 3 h in a muffle furnace and ground to get Mg-Al LDO. To adsorption OII, 150 ml water was filled into a three-mouth-flask. After driving the dissolved air with Ar gas flow, 0.6 g Orange II (>85%, Macklin) was dispersed into the DI water. Then, 1 g Mg-Al LDO was added to the solution to adsorb the dye for 48 h. Magnetic stirring and Ar gas flow were maintained during the overall adsorption process. Afterwards, the solution was vacuum filtrated and washed with water for several times, then the obtained powder (RLDH/OII) was freeze dried and ground. The obtained RLDH/OII was mixed with 0.6 g melamine (99%, Macklin), and the mixture was dispersed into 50 mL methanol in a beaker. After that, the solution was heated to 50◦C under magnetic stirring. The temperature was kept until the methanol was totally evaporated, and then the RLDH-OII-melamine mixture (RLDH-OII-M) was collected. The carbonization of RLDH-OII-M was performed at 800◦C for 2 h under Ar atmosphere with a heating rate of 2◦C/min. Then the obtained black powder was washed with HCl (6M) and NaOH (2M) respectively at 70◦C for several hours. After freeze drying process, the target product H-2D-HCA was obtained.

2D-HCA was synthesized as reference samples followed the above procedures except melamine mixing process. OII and OIImelamine mixture (mass ratio: 1:1) were carbonized and washed with the above-mentioned conditions to get the reference sample C-OII and high-level doped C-OII (H-C-OII), respectively.

## Microstructure Characterization

The morphology observation and elemental mapping analyzing were conducted by using a field electron microscopy (FESEM, ZEISS SUPRA55). The microstructures of the four samples were examined by a field high resolution transmission electron microscopy (HRTEM, FEI Tecnai G<sup>2</sup> F30). A Rigaku D/max 2500/PC diffractometer (Cu-Kα radiation with λ = 1.5418 Å) was used to measure the powder X-ray diffraction (XRD) patterns of samples. A HORIBA Labram HR Evolution Raman spectrometer was used to measure the Raman spectra of the samples (the excitation wavelength of the leaser is 532 nm). The XPS analyses were carried out on a PHI 5000 VersaProbe II spectrometer using monochromatic Al K(alpha) X-ray source. The N2 adsorption/desorption isotherms were measured by using a Micromeritics ASAP 2020 automated adsorption apparatus at 77 K. The specific surface areas was determined based on Brunauer-Emmett-Teller (BET) equation, and the pore-size

distribution was calculated by utilizing density functional theory (DFT).

### Electrochemical Tests

Li-ion storage performance of these samples was tested with halfbattery methods by using coin-type cells (CR2032). To prepare the working electrodes, a slurry consisted a mixture of 70 wt % of H-2D-HCA (or other three carbon samples), 20 wt % of poly (vinylidene fluoride) (PVDF) and 10 wt % of acetylene black in N-methyl pyrrolidone was coated on a copper foil. After being dried at 80◦C for 12 h, the electrode was punched into disks with a diameter of 12 mm. The mass loading of active materials was ∼ 1.2 mg cm−<sup>2</sup> . 1.0 M LiPF<sup>6</sup> dissolved in ethylene carbonate/diethyl carbonate (EC/DEC, with a volume ratio of 1:1) was used as the electrolyte. Lithium foil was used as both the counter electrode and reference electrodes. Battery assembling was carried out in an Ar-filled glove box with both moisture and oxygen concentrations below 1 ppm. Cyclic voltammetry was measured using an electrochemical workstation (CHI 660E) in the potential window of 0.005–3 V vs. Li/Li<sup>+</sup> with a scanning rate is 0.2 mV s−<sup>1</sup> . The cycling performance and rate capabilities were tested at in the potential window of 0.005–3 V vs. Li/Li<sup>+</sup> using a LAND battery tester (CT2001A).

## RESULTS AND DISCUSSION

H-2D-HCA was synthesized based on a template-assistant method that we reported before (Wang et al., 2016). Mg-Al layered double oxides (Mg-Al LDO, LDO for short) were used as the templates. Orange II (OII for short, an N- and Scontaining organic dye) was used as both carbon precursors and heteroatom sources, and melamine were used as extra N sources. The preparation approach is illustrated in **Figure 1** Firstly, Mg-Al layered double hydroxides (Mg-Al LDH, LDH for short) were calcined to derive LDO. Afterwards, the LDO was dispersed into OII solution to adsorb this organic dye. During this process, the LDO was rehydrated to LDH (RLDH). More importantly, a morphology change has occurred simultaneously. This should be attributed to the morphology change occurred in the rehydration process of LDO (Wang et al., 2016). As a result, the obtained RLDH-OII composite had a 2D hierarchical structure, rather than a 2D smooth plating structure like LDH or LDO. After that, the RLDH-OII was mixed with melamine in methanol under magnetic stirring, and a RLH-OII-melamine mixture (RLDH-OII-M) was collected. Then, the RLDH-OII-M was heated at 800◦C to carbonize the organic, meanwhile, RLDH was calcined to LDO again. Finally, the obtained LDO-carbon (LDO-C) mixture was washed with NaOH and HCl to remove the LDO templates, and H-2D-HCA was achieved at last. Detailed procedures can be seen in the Experimental Methods section.

Both LDH and LDO have a hexagonal plating morphology with a smooth surface (Supplementary Figure 1). However, after OII adsorption and melamine mixing process, RLDH-OII and RLDH-OII-M show a hierarchical structure of some small-size sheets decorating on the surface of the hexagonal plates (Supplementary Figure 2). This should be attributed to the morphology change during the rehydration process of LDO (Wang et al., 2016). After carbonization and template removal process, the obtained H-2D-HCA has successfully maintained this kind of hierarchical structure, in which small carbon nanosheets growing on the surface of large-size hexagonal carbon nanoplates (**Figure 2A**). The diameter of H-2D-HCA is ∼2.2µm (**Figure 2B**) and the average thickness is ∼200 nm (Supplementary Figure 3). The small carbon nanosheets growing on the surface of H-2D-HCA have a size of ∼100 nm and a thickness of ∼10 nm (Supplementary Figure 3). The adjacent carbon nanosheets have formed many half-open pores with diameters of 10 of nanometers (Supplementary Figures 3, 4). These pores can act as reservoirs to storage electrolyte and guarantee good contact of the nanocarbon electrodes with electrolyte. 2D-HCA has similar hierarchical structure and the size with H-2D-HCA (Supplementary Figure 5), which was also reported in the previous work (Wang et al., 2016). This kind of hierarchical structure can alleviate aggregation issues of carbon nanomaterials, which caused by π-π interaction between carbon layers, and thus decrease the electrochemically active surface loss (Zhao et al., 2014; Yu et al., 2015). For comparison, H-C-OII and C-OII have irregular structure with size of about tens of micrometers (Supplementary Figure 5). These structures do not benefit electrolyte access to carbon electrodes and hinder the diffusion of Li ions.

The microstructure of H-2D-HCA was investigated by high-resolution transmission electron microscopy (HRTEM). Interestingly, H-2D-HCA has an amorphous nature and graphitic micro-crystallites with an average interlayer distance of 0.368 nm (**Figure 2C**), which is larger than that of graphite (Ou et al., 2017; Yang et al., 2018).The broaden peaks in the X-ray diffraction (XRD) pattern also confirms the amorphous nature of H-2D-HCA (Supplementary Figure 6). The peak located at 24.2◦ can be assigned to (002) diffraction peak of graphite (Xu F. et al., 2015; Zheng et al., 2015; Zhao et al., 2016; Yang et al., 2017). This peak corresponds to an interlayer distance of 0.367 nm (calculated based on the Bragg's law), which agrees well with the HRTEM results. The enlarged interlayer distance can promote Liion diffusion and help to improve the high rate performance (Qie et al., 2015; Xu et al., 2016; Liang et al., 2018). 2D-HCA has a similar amorphous structure with few graphitic layers, while H-C-OII and C-OII only show amorphous porous structure without graphite micro-crystallites could be observed (Supplementary Figure 7).

**Figure 2D** shows SEM elemental mapping. It revealed that H-2D-HCA contained not only C elements, but also abundant N and S heteroatoms. The doped N and S distribute uniformly throughout the carbon framework.

X-ray photoelectron spectroscopy (XPS) was utilized to evaluate the surface chemistry in H-2D-HCA. N and S heteroatom concentrations can be determined to be 15.5 and 0.9%, respectively (Supplementary Figure 8). The total N and S heteroatom doping level of 16.4% in H-2D-HCA, which is much higher than that of most N, S dual-doped carbon nanomaterials (Supplementary Table 1) (Ai et al., 2014; Sun et al., 2015; Xu G. et al., 2015; Zhou et al., 2015; Zhuang et al., 2015; Shan et al., 2016; Zhang et al., 2016a). The high-resolution spectrum of C1s can be fitted to four peaks (**Figure 3A**). The peak located

at 284.4 eV can be attributed to C-C/C=C of carbon; and the peak located at 285.1 eV reveal the presence of C-N and C-S (Xu G. et al., 2015; Zhuang et al., 2015; Zhang et al., 2016a). The peaks located at 286.2 and 288 eV can be assigned to C-O and C=O, respectively. The high-resolution N1s spectrum can be deconvoluted into three different peaks located at 398.3, 300.8, and 401.1 eV, corresponding to pyridinic-N (N-6), pyrrolic-N (N-5) and graphitic-N (N-Q), respectively (**Figure 3B**; Wang et al., 2011; Ma et al., 2012; Mao et al., 2012; Zheng et al., 2014; Li et al., 2017). N-6, N-5, and N-Q accounted for 37, 47, and 16% of total N atoms, respectively (**Table 1**). The electrochemically active N species, N-6 and N-5, occupied a large proportion of total N atoms (84%). The fine split peaks in high-resolution of S2p spectrum indicated the presence of C-S (163.5 eV), C=S (164.8 eV) and SO<sup>X</sup> group (167.7 eV) (**Figure 3C**; Zhang X. et al., 2016). As demonstrated by other work, S heteroatom can enlarge the interlayer distance of carbon nanomaterials because of the larger covalent radius (Qie et al., 2015; Xu et al., 2016).

The extended interlayer space can promote insertion-extraction speed of Li ions in carbon nanomaterials and thus was expected to improve the fast charge-discharge properties. The elemental composition of H-2D-HCA, 2D-HCA, H-C-OII, and C-OII are summarized in **Table 1** (the data of H-C-OII and C-OII were summarized based on XPS results shown in Supplementary Figures 9, 10, and the data of 2D-HCA were calculated from the Reference of Wang et al., 2016). Obviously, H-2D-HCA has the highest doping level and electrochemically active N proportion. Besides, because of the higher N concentration, H-2D-HCA has a much higher heteroatom doping level than 2D-HCA, and H-C-OII also has a much higher heteroatom doping level than C-OII (**Table 1**). This should be attributed to the use of melamine as extra N source during the synthesizing approach (Sheng et al., 2011).

The N<sup>2</sup> adsorption/desorption isotherms of H-2D-HCA exhibit a combination of type I and type II characteristics, with a distinct hysteresis lop at relative pressure P/P0 ranging from 0.42 to 1 (**Figure 3D**). The specific surface area (SSA) can be calculated to be 535 m<sup>2</sup> g −1 , which is lower than that of 2D-HCA (Wang et al., 2016). The lower SSA could be attributed to melamine decomposition during carbonization process, and this phenomenon was also reported by other previous work (Zheng et al., 2011). Pore-size distribution (PSD) curves reveal that the H-2D-HCA had both micropores which peaks located at 0.52, 1.23, and 1.71 nm, and mesopores which peaks located at 3.97 nm (**Figure 3E**). The presences of mesopores with size ranging from 6 to 30 nm can also be seen in the enlarged PSD curve (inset of **Figure 3E)**. The co-existence of micropores and mesopores in H-2D-HCA suggests its hierarchical porous


TABLE 1 | Surface physiochemical properties of various samples by XPS tests.

structure (Wang et al., 2008; Hu et al., 2015). The hierarchical porous structure benefits LIB performance of nanocarbon electrodes, in which the micropores offer abundant active sites for Li-ion storage and the mesopores can promote the rapid diffusion of Li ions (Zhao et al., 2012). By comparison, the pores in H-C-OII (Supplementary Figure 11) and C-OII (Wang et al., 2016) are mainly micropores (size less than 2 nm), with the much less presence of mesopores.

**Figure 3F** shows the Raman spectrum of H-2D-HCA. Two remarkable peaks of D band located at 1,341 cm−<sup>1</sup> and G band at 1,581 cm−<sup>1</sup> can be observed. As is known, G band corresponds to the zone center E2g mode related to phonon vibrations in sp2 carbon materials, and D band related to structural defects (such as edge, heteroatoms, etc.) and partially disordered structures of the sp2 domains framework (Ferrari et al., 2006). The high ID/I<sup>G</sup> ratio of 1.15 for H-2D-HCA and 1.11 for H-C-OII (Supplementary Figure 12) indicated that they have low graphitization degree and abundant defects (Ferrari et al., 2006). The defects mainly include the heteroatoms doped in the carbon framework. In contrast, because of the lower heteroatom doping level, 2D-HCA and C-OII have lower ID/I<sup>G</sup> ratio of 0.96 and 0.98, respectively (Supplementary Figure 12).

Li-ion storage properties of H-2D-HCA were evaluated using half-battery test methods. Cyclic voltammetry (CV) curves suggest that H-2D-HCA has similar Li-ion storage behaviors with other carbon nanomaterials. As shown in **Figure 4A**, in the first cycle, the pronounced irreversible anodic peak at around 0.5 V relates to the electrochemical decomposition of electrolyte and the formation of solid electrolyte interface (SEI) on the huge surface (Sun et al., 2015; Xu G. et al., 2015; Deng et al., 2016; Zhang et al., 2016b). The followed anodic peak near 0 V corresponds to the electrochemical intercalation of Li ions into graphitic structures. The cathodic peaks at 0.2 and 1.2 V can be assigned to Li ions extraction from graphitic layers and defect sites, respectively (Zhang et al., 2016b). The curves in the following cycles overlap together, suggesting the stable formation of SEI layers. When applied as anodes for LIBs, H-2D-HCA exhibits high specific capacity and excellent rate performance. As shown in **Figure 4B**, H-2D-HCA delivers an initial discharge capacity of 1,861 mA h g−<sup>1</sup> at current density of 200 mA g−<sup>1</sup> . When the current density increases to 500, 1,000, 2,000, 5,000 mA g −1 , the discharge capacities of H-2D-HCA were 756, 636, 504, and 348 mA h g−<sup>1</sup> , respectively (measured from the middle cycle in each current density). After that, with the current density returns to 2,000, 1,000, 500, and 200 mA g−<sup>1</sup> , the discharge capacity recovers to the initial capacity values with only little fade.

Long-term cyclability at high current density has been tested to further examine Li-ion storage performance of H-2D-HCA. For comparison, 2D-HCA, H-C-OII and C-OII have been tested at the same condition. First, all the samples were firstly activated at a low current density of 200 mA g−<sup>1</sup> for two cycles, and then the current density was directly increased to 5 A g−<sup>1</sup> . The voltage profiles of these samples in the first two cycles are shown in Supplementary Figure 13. H-2D-HCA delivered an initial discharge capacity and charge capacity of 1,924 and 1,824 mA h g−<sup>1</sup> , together with an initial columbic efficiency (ICE) of 56.4% (**Figure 4C**). Because of the large surface area and porous structure, the H-2D-HCA delivered much higher specific capacity than graphite. The irreversible capacity loss in the first cycle is due to the SEI formation and the irreversible insertion of Li ions into micropores. H-2D-HCA has the highest ICE in these four samples (Supplementary Table 2). However, it was still far from satisfaction for practical application. Reducing surface area, coating with dense carbon layer, optimizing pore structure and pre-lithiation may have positive effects on improving the ICE. In addition, it is noted that the ICE of H-2D-HCA is higher than that of 2D-HCA, and the ICE of H-C-OII is higher than that of C-OII, suggesting that increasing N doping level may improve the ICE of nanocarbon anodes. When the current density increased to 5 A g−<sup>1</sup> , H-2D-HCA delivers a discharge capacity of 660 mA h g−<sup>1</sup> and a CE of 77.9%. The CE rises to a value of higher than 97% in the followed cycles (**Figure 4C**). Then the discharge capacity of H-2D-HCA drops to 450 mA h g−<sup>1</sup> in 25 cycles, and increases afterwards, reaches 550 mA h g−<sup>1</sup> in the 196th cycles. After that, the discharge capacity remains stable only with slight decay. Even after 1,000 cycles, H-2D-HCA can still deliver a high discharge capacity of 329 mA h g−<sup>1</sup> , and a high CE of 99.2% (**Figure 4C**). In contrast, 2D-HCA delivered a first discharge capacity of 713 mA h g−<sup>1</sup> and a corresponding CE of 65% at 5 A g−<sup>1</sup> . The higher initial capacity of 2D-HCA can be attributed to the larger SSA. The following discharge capacity of 2D-HCA gradually decreases with cycling. After 1,000 cycles, the discharge capacity fades to 222 mA h g−<sup>1</sup> , which is much lower than that of H-2D-HCA (**Figure 4C**). As for H-C-OII and C-OII, they discharge capacity respectively are 88 and 78 mA g −1 after 1,000 cycles, only about a quarter of that of H-2D-HCA (**Figure 4C**). The corresponding CE of 2D-HCA, H-C-OII, and C-OII were shown in Supplementary Figure 14, which were

worse than that of H-2D-HCA. In addition, the electrochemical performance of H-2D-HCA is better than most of N, S co-doped carbon nanomaterials (summarized in Supplementary Table 1).

In view of above results, benefiting from the unique microstructure, H-2D-HCA delivers highly efficient Li-ion storage performance. We performed first principle calculations to further understand the effect of high-level heteroatom doping on the electrochemical performance of carbon nanomaterials. The influences of dopant types and interlayer distance on Li adsorption energy, electrical conductivity and Li ion diffusion barriers were studied in this work.

To evaluate the stability of Li ions adsorbed on the carbon systems, the adsorption energy was calculated with the followed equation:

$$E\_{abs} = E\_2 - E\_1 - E\_{Li},\tag{1}$$

Where Eabs is the adsorption energy, E<sup>2</sup> is the total energy of the geometry optimized structure with the absorbed Li atoms, E<sup>1</sup> is the energy of different carbon systems, and ELi is the energy of single Li atom in bulk form (Zhou et al., 2004). Different carbon systems with an expanded interlayer distance of 0.368 nm were considered and their Eabs have been calculated: pure carbon (P-C), N-6 doped carbon (N-6/C), N-5 doped carbon (N-5/C), N-Q doped carbon (N-Q/C), and N, S co-doped carbon (N/S/C). For comparison, Eabs of abovementioned carbon systems with an interlayer distance of 0.34 nm were also calculated. In the 0.368 nm spacing case, Eabs for the P/C, N-5/C, N-6/C, N-Q/C and N/S/C are −0.08, −2.74, −3.28, 0.58, and −1.68 eV, respectively (the detailed structures are presented in Supplementary Figure 15). Whereas in the 0.34 nm spacing case, the corresponding Eabs are 0.18, −2.61, −3.25, 0.25 and −0.94 eV, respectively (**Figure 5A**). As is known, the more negative Eabs value means more stable Li absorption (Ma et al., 2012). Obviously, the carbon system with 0.368 nm interlayer distance have enhanced Li absorption stability. In addition, N-6 and N-5 doped carbon systems have lower Eabs. This indicates that increasing doping concentration of N-5 and N-6 can obviously improve Li-ion storage ability of carbon nanomaterials. Furthermore, the carbon systems with co-doping of N and S are also have lower Eabs compared with pure carbon, indicating more stable Li adsorption.

Another factor that affects Li-ion storage performance is the electrical conductivity of nanocarbon electrodes. In general, materials conductivity is determined by their density of states (DOS) at the Fermi level. We calculated the DOS of carbon systems of pure one and N, S co-doped one (structures can be seen in Supplementary Figures 15, 16). Comparing with the pure carbon system, the N, S co-doped carbon system has higher DOS value (3.0 eV) than the pure carbon (0.63 eV) at Fermi

level (**Figure 5B**), demonstrating its better electric conductivity. The diffusion of kinetics of Li ions also plays a dominant part in the electrochemical performance of nanocarbon anodes. The diffusion kinetics are examined by investigating Li ion diffusion barriers (activation energy). With the climbing-image nudged elastic band (CI-NEB) method, we studied a specific diffusion path, in which Li ions migrate between two adjacent hollow sites (**Figure 5C**). Convergence tests show that five intermediate images are adequate to accurately describe the activation energy barriers. The calculated activation energy barriers for 0.34 and 0.368 nm are 90.1 and 71.2 m eV, respectively (**Figure 5D**). This suggests that Li ions can diffusion much easier in carbon nanomaterials with extended interlayer distance. Theoretical calculations indicate that the expanded interlayer distance and the doped high-level N and S heteroatoms in the carbon framework can effectively enhance the Li absorption stability, Li diffusion mobility and electronic conductivity, which facilitates the transportation of Li ions and electrons, and thus improve Li-ion storage performance of H-2D-HCA.

### CONCLUSIONS

In summary, by using template-assistant methods, we have developed high-level heteroatom doped two-dimensional carbon architectures (H-2D-HCA) for highly efficient Li-ion storage. The hierarchical structure can alleviate the electrochemical active surface loss, and the porous structure provided rapid diffusion channels for Li ions. More importantly, the highly concentrated heteroatoms of 0.9% sulfur and 15.5% nitrogen are able to created abundant electrochemical active sties to storage Li ions. Also the increased N doping concentration improved the ICE of carbon nanomaterials. Furthermore, the expanded interlayer distance can promote insertion-extraction speed of Li ions. First principle calculations confirmed the enhanced Li absorption stability as well as the electronic conductivity by co-doping of N and S heteroatoms and the accumulated diffusion mobility of Li ions owing to expanded interlayer space. Benefiting from these unique microstructure characteristics and high-level heteroatom doping nature, H-2D-HCA exhibited enhanced Li-ion storage performance. Even at a high current density of 5 A g−<sup>1</sup> , it can exhibit a high discharge capacity of 329 mA h g−<sup>1</sup> after 1,000 cycles. Because of such superior electrochemical performance, H-2D-HCA can be promising electrode candidates for fast charge-discharge Li-ion batteries.

### AUTHOR CONTRIBUTIONS

ZW and YW did the materials preparation and characterization, as well as electrochemical tests. WW did the first principle calculations. BX and Y-BH supervised this work. All the authors discussed the results and wrote the manuscript.

### FUNDING

This work was supported by the National Key Basic Research Program of China (2014CB932400), the National Natural Science Foundation of China (51672156 and 51232005), the Guangdong special support program (2015TQ01N401), Guangdong Province Technical Plan Project (2017B010119001 and 2017B090907005), Production-study-research cooperation project of Dongguan City (2015509119213), Shenzhen Technical Plan Project

### REFERENCES


GJHS20170314165324888).

(KQJSCX20160226191136, JCYJ20170412170706047, JCYJ20170307153806471 JCYJ20150529164918734, and

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2018.00097/full#supplementary-material

study on a live sodiation/desodiation process. Adv. Fun. Mater. 27:1606242. doi: 10.1002/adfm.201606242


high-performance sodium ion battery anode. Mater. Today Energy 8, 37–44. doi: 10.1016/j.mtener.2018.02.001


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Wang, Wang, Wang, Yu, Lv, Xiang and He. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Effect of Nb and F Co-doping on Li1.2Mn0.54Ni0.13Co0.13O<sup>2</sup> Cathode Material for High-Performance Lithium-Ion Batteries

Lei Ming<sup>1</sup> , Bao Zhang<sup>1</sup> , Yang Cao1,2, Jia-Feng Zhang<sup>1</sup> \*, Chun-Hui Wang<sup>1</sup> , Xiao-Wei Wang<sup>1</sup> and Hui Li <sup>1</sup>

*<sup>1</sup> School of Metallurgy and Environment, Central South University, Changsha, China, <sup>2</sup> Medical Engineering Center, Xiangya Hospital of Central South University, Changsha, China*

#### Edited by:

*Qiaobao Zhang, Xiamen University, China*

### Reviewed by:

*Wang Renheng, Nanyang Technological University, Singapore Lingjun Li, Changsha University of Science and Technology, China Keyu Xie, Northwestern Polytechnical University, China*

### \*Correspondence:

*Jia-Feng Zhang yjyzjf@csu.edu.cn*

#### Specialty section:

*This article was submitted to Physical Chemistry and Chemical Physics, a section of the journal Frontiers in Chemistry*

> Received: *30 January 2018* Accepted: *08 March 2018* Published: *05 April 2018*

#### Citation:

*Ming L, Zhang B, Cao Y, Zhang J-F, Wang C-H, Wang X-W and Li H (2018) Effect of Nb and F Co-doping on Li1.2Mn0.54Ni0.13Co0.13O2 Cathode Material for High-Performance Lithium-Ion Batteries. Front. Chem. 6:76. doi: 10.3389/fchem.2018.00076* The Li1.2Mn0.54−xNbxCo0.13Ni0.13O2−6xF6x (*x* = 0, 0.01, 0.03, 0.05) is prepared by traditional solid-phase method, and the Nb and F ions are successfully doped into Mn and O sites of layered materials Li1.2Mn0.54Co0.13Ni0.13O2, respectively. The incorporating Nb ion in Mn site can effectively restrain the migration of transition metal ions during long-term cycling, and keep the stability of the crystal structure. The Li1.2Mn0.54−xNbxCo0.13Ni0.13O2−6xF6x shows suppressed voltage fade and higher capacity retention of 98.1% after 200 cycles at rate of 1 C. The replacement of O2<sup>−</sup> by the strongly electronegative F<sup>−</sup> is beneficial for suppressed the structure change of Li2MnO<sup>3</sup> from the eliminating of oxygen in initial charge process. Therefore, the initial coulombic efficiency of doped Li1.2Mn0.54−xNbxCo0.13Ni0.13O2−6xF6x gets improved, which is higher than that of pure Li1.2Mn0.54Co0.13Ni0.13O2. In addition, the Nb and F co-doping can effectively enhance the transfer of lithium-ion and electrons, and thus improving rate performance.

Keywords: Li1.2Mn0.54Ni0.13Co0.13O2 , Nb and F co-doping, cathode material, coulombic efficiency, electrochemical property

### INTRODUCTION

Lithium-ion batteries (LIBs) have been broadly used in the portable electronics, and regarded as the most promising energy storages for hybrid electric vehicles (HEVs) and electric vehicles (EVs) (Tarascon and Armand, 2001; Armand and Tarascon, 2008; Chiang, 2010). It is generally believed that the cathode materials are the primary factors for the improvements of the lithium-ion batteries. However, conventional cathode materials, such as LiCoO2, LiMn2O4, and LiFePO<sup>4</sup> show low specific capacity and unsatisfied energy density, which will limit further practical application in the energy storage system (Ding et al., 2011; Luo et al., 2012; Zheng et al., 2015a). Among the developed cathode materials, lithium-rich layered material attracts great attentions of scientists, due to their high capacity of above 250 mAh g−<sup>1</sup> , high operating voltage and low cost compared with other cathode materials (Thackeray et al., 2005; Park et al., 2007). It is a pity that lithium-rich layered materials have some intrinsic drawbacks of huge initially irreversible capacity, poor rate capability, the continous capacity and voltage decay during long-term cycling, which block their practical applications (Ellis et al., 2010). The poor rate capability is due to the poor electrical conductivity of Li2MnO<sup>3</sup> component in lithium-rich materials, while the low initial coulombic efficiency is related to the elimination of O2<sup>−</sup> make the change of structure of Li2MnO<sup>3</sup> in charge process (Johnson et al., 2007; He et al., 2012). Additionally, the capacity and voltage decay is caused by the structure transformation and formation of passivation layer during cycling (Lu and Dahn, 2002; Armstrong et al., 2006).

To solve above problems, many methods are proposed, such as surface coating, ion doping, and particle size reducing. Generally, surface coating could effectively suppress the side reaction between lithium-rich layered material and electrolyte and elimination of the oxygen vacancies, thus improving initial coulombic efficiency and cycling stability (Li et al., 2014, 2015, 2016). The reducing particles size could shorten the pathway to enhance the rate capability (Zheng et al., 2017). However, above approaches cannot effectively suppressed the voltage decay during cycling. Bulk cationic doping with Al, Mg, Cr, Zr, and Ru could effectively suppress the migration of transition metal (TM) during cycling, and mitigate the capacity and voltage decay (Kim et al., 2006; Jiao et al., 2007; Luo and Dahn, 2011; Sathiya et al., 2013; Xu et al., 2014). In addition, O2<sup>−</sup> site is replaced by anions, such as Cl<sup>−</sup> and F−, which could be beneficial for suppressing the structure change of Li2MnO<sup>3</sup> from the eliminating of oxygen in initial charge process, and thus Li<sup>+</sup> could return to the material lattice in subsequent charge-discharge process (Kang and Amine, 2005; Park et al., 2005). Therefore, the bulk doping also increase the electronic conductivity, and improve the rate performance of material bulk.

Therefore, we described the incorporation of Nb5<sup>+</sup> and F<sup>−</sup> into the Mn site and O site of Li1.2Mn0.54Ni0.13Co0.13O<sup>2</sup> (LMNCO), respectively. The Nb5<sup>+</sup> and F<sup>−</sup> co-doping suppress the TM migration during cycling and alleviate Li<sup>+</sup> loss during the elimination of O2<sup>−</sup> in the charging process. In this paper, the doped Li-rich layered oxide (Li1.2Mn0.54−xNbxCo0.13Ni0.13O2−6xF6x) is synthesized by high temperature solid phase method, the effects of Nb5<sup>+</sup> and F<sup>−</sup> co-doping on the initial coulomb efficiency, rate performance, cycle performance and work voltage are discussed in detail.

### EXPERIMENTAL

### Materials Synthesis

The fabrication process of Li1.2Mn0.54−xNbxCo0.13Ni0.13O2−6xF6x (x = 0, 0.01, 0.03, 0.05) by high temperature solid phase method is schemed in **Figure 1**. Typically, the stoichiometric Li(CH3COO)·2H2O, Mn(CH3COO)2·4H2O, Ni(CH3COO)2·4H2O, Co(CH3COO)2·4H2O, Nb2O5, LiF and citric acid were mixed with 50 wt% of deionized water by ball-milling for 8 h (all chemicals of 99% purity). The mole ratio of Li(CH3COO)·2H2O, Mn(CH3COO)2·4H2O, Ni(CH3COO)2·4H2O, Co(CH3COO)2·4H2O, Nb2O5, LiF are 1.2: 0.54-x: 0.13: 0.13: x/2: 6x (x = 0, 0.01, 0.03, 0.05), respectively. Then the mixtures were dried at 80◦C for 12 h, and ground into fine particles. Finally, the mixture powders were calcined at 550◦C for 5 h, follow at 850◦C for 15 h in air to get a set of Li-rich layered oxide materials Li1.2Mn0.54−xNbxCo0.13Ni0.13O2−6xF6x. The Li1.2Mn0.54−xNbxCo0.13Ni0.13O2−6xF6x materials with x = 0, 0.01, 0.03, 0.05 are shorted as LMNCO-NF0, LMNCO-NF1, LMNCO-NF3, and LMNCO-NF5, respectively.

### Characterization

All materials were characterized through an X-ray diffraction (XRD: Rigaku, D/max 2500v/pc) with Cu Kα radiation. The scanning electron microscopy (SEM: Philips, FEI Quanta 200 FEG) and transmission electron microscopy (TEM: JEM-2010, JEOL) were applied to observed the microstructure and the structure of all materials. The elemental chemical states of all

materials were analyzed by X-ray photoelectron spectroscopy (XPS, PHI 5000VersaProbe).

### Electrochemical Evaluation

Electrochemical performance of all Li-rich layered oxide Li1.2Mn0.54−xNbxCo0.13Ni0.13O2−6xF6x were tested using CR2032 coin cell. The electrode preparation process was consisted of three steps as follow. Firstly, 80% active material (LMNCO), 10% acetylene black, and 10% polyvinylidene fluoride (PVDF) binder were mixed with NMP solvent. Secondly, as prepared viscous cathode slurry was cast on aluminum foil. Thirdly, the foil was dried at 90◦C under vacuum for 12 h. Then it was punched into 12 mm diameter disks with the loading of active cathode mass in the range of 3–4 mg cm−<sup>2</sup> . The coin cells were assembled in an argon-filled dry box. The lithium metal and the Celgard 2500 were used as anode material and the separator, respectively. 1 M LiPF<sup>6</sup> in ethylene carbonate/diethyl carbonate (V/V = 1:1) was used as electrolyte. The galvanostatic chargedischarge measurements were carried out on LAND CT2001A battery testing system (Wuhan, China). Cyclic voltammetry (CV) measurements were performed by IM6 electrochemical testing station at scan rates of 0.1 mV s−<sup>1</sup> between 2.0 and 4.8 V. Electrochemical impedance spectroscopy (EIS) was conducted by IM6 electrochemical testing station between 100 kHz and 0.01 Hz by applying perturbation AC voltage signal of 5 mV.

### RESULTS AND DISCUSSION

**Figure 2** shows the XRD patterns of Li1.2Mn0.54−xNbxCo0.13Ni0.13O2−6xF6x materials (x = 0, 0.01, 0.03, and 0.05). As seen Figure S1A, the XRD pattern of LMNCO-NF0 material belongs to the layered α-NaFeO<sup>2</sup> structure with space group R3m (Figure S1A). There is a weak diffraction peak around 20–25◦ in the XRD pattern of the LMNCO-NF0, corresponding to the short-range cation ordering of Li<sup>+</sup> and Mn4<sup>+</sup> in the transition metal layers, as illustrated for Li2MnO<sup>3</sup> structure in Figure S1B (Jarvis et al., 2011). The adjacent peaks of (006)/(012) and (108)/(110) show obvious separation, indicating the perfect layer structure of LMNCO-NF0 (Gong et al., 2004). Meanwhile, the intensity ratio of I(003)/I(104) is the indication of mixing degree for transition-metal ions in the lithium layer (Zheng et al., 2015b). For LMNCO-NF0, the I(003)/I(104) value reach 1.6, suggesting low mixing degree of transition-metal ions in the lithium layer. After doping, the XRD patterns of all samples are similar to that of LMNCO-NF0, and adjacent peaks of (006)/(012), (108)/(110) and I(003)/I(104) value remain significantly unchange. The XRD patterns of LMNCO-NF1 and LMNCO-NF3 show highly pure phase, implying that Nb and F ions are successfully doped into the crystal lattice. But when the doping amount increases, the impurity phases of Li3NbO4, Nb2O5, and LiF are observed in the XRD pattern of LMNCO-NF5, due to the solid solubility of Nb and F elements in the Li1.2Mn0.54Co0.13Ni0.13O<sup>2</sup> material is beyond the limitation. In addition, for all doped materials, the diffraction peaks of doped samples slightly shift to lower 2θ compared to that of the LMNCO-NF0, indicating that Nb and F doping can enlarge the interlayer spacing. Furthermore, the lattice parameters of all samples are calculated by Rietveld refinement, and the results are listed in **Table 1**. It is clearly seen that the values of lattice parameter a and c get higher along with increasing amount of Nb and F ion doping. While the increasing ratio of c/a for doped sample represents low mixing degree for transition-metal ions in the lithium layer. This phenomenon also suggests that cell volume is enlarged after doping with Nb5<sup>+</sup> and F−, which is beneficial for the diffusion of the Li<sup>+</sup> ions (Jafta et al., 2012). In addition, the average crystallite size of the LMNCO-NF0, LMNCO-NF1, LMNCO-NF3, and LMNCO-NF5 are calculated by using Scherrer equation [βcos(θ) = kλ/D], where β is full-width at half-maximum (FWHM) of the XRD peak and k is a constant (0.9) as given in Table S1. This suggests that the addition of Nb2O<sup>5</sup> and LiF could suppressed the growth of crystallite size.

In order to analyze the effect of Nb and F co-doping on chemical composition of LMNCO, the stoichiometric amounts

of metal element in all samples have been determined by ICP analysis, and the result are listed in Table S2. As seen in Table S2, the molar ratio of Li: Ni: Co: Mn in pristine LMNCO is 1.213: 0.133: 0.132: 0.543, which is close to the theoretical ratio of 1.2: 0.133: 0.133: 0.54. The molar ratio of Li: Ni: Co: Mn of all doped samples (LMNCO-NF1, LMNCO-NF3, and LMNCO-NF5) does not vary compare to pristine LMNCO, suggesting that the Nb and F co-doping not affect the chemical composition of the samples.

The SEM images of Li1.2Mn0.54−xNbxCo0.13Ni0.13O2−6xF6x materials (x = 0, 0.01, 0.03, and 0.05) are shown in Figure S2. The particle sizes of all doped samples (LMNCO-NF1, LMNCO-NF3, and LMNCO-NF5) are slightly smaller than that of LMNCO-NF0. Particularly, the particles sizes of doped samples decrease with the increasing amount of Nb5<sup>+</sup> and F<sup>−</sup> elements, suggesting

TABLE 1 | The crystallographic parameters of LMNCO-NF0, LMNCO-NF1, LMNCO-NF3, and LMNCO-NF5, respectively.


Nb2O<sup>5</sup> and LiF can exhibit space steric effect, thus effectively suppressing the growth of particles. The tap density of pure LMNCO and all doped samples (LMNCO-NF1, LMNCO-NF3, and LMNCO-NF5) are listed in Table S3. The tap density the pure LMNCO and all doped samples (LMNCO-NF1, LMNCO-NF3, and LMNCO-NF5) are 1.51, 1.50, 1.52, and 1.49, respectively. The tap density of all doped samples (LMNCO-NF1, LMNCO-NF3, and LMNCO-NF5) not vary compared to pure LMNCO. In addition, EDS elements mapping test has been performed on the LMNCO-NF3. As seen Figure S2E, the element mappings clearly demonstrate that Ni, Co, Mn, Nb, and F elements are all homogeneously distributed in LMNCO-NF3 structure, which confirm that Nb and F are doped into the bulk material.

**Figure 3** show TEM, HETEM and selected area electron diffraction (SAED) patterns of the LMNCO-NF0, LMNCO-NF1, LMNCO-NF3, and LMNCO-NF5. The interplanar spacing of the lattice fringes (003) gradually expand along with the increase content of Nb and F co-doping, which is consistent with the XRD calculated results. This phenomenon is ascribed to that the Nb5<sup>+</sup> ions doped into the Mn4<sup>+</sup> sites and the F<sup>−</sup> ions occupy the pack oxygen sites. The (SAED) patterns (**Figures 3A3-D<sup>3</sup>** reveal that all the samples belong to the hexagonal symmetry of the local structure.

In order to investigate the effect of Nb and F ion codoping on the oxidation states of some elements (Ni, Mn, Co,

Nb, F, O) for all samples, **Figure 4** shows the XPS spectra of Li1.2Mn0.54−xNbxCo0.13Ni0.13O2−6xF6x materials (x = 0, 0.01, 0.03, and 0.05), respectively. For the doped samples, the Ni, Mn, Co binding energy peak all shift to higher binding energy compare to that of LMNCO-NF0, resulting from the density of electron clouds reduce around LMNCO. The Nb5<sup>+</sup> ion doping into Mn4<sup>+</sup> site will reduce the electron clouds. The electronegative of F<sup>−</sup> is stronger than that of O2−, and the electron clouds of all transition metal elements tend to bond with F <sup>−</sup>, suggesting that the F<sup>−</sup> is successfully doping into the site of pack oxygen. The F 1s and Nb 3d binding energy peaks are not detected in XPS spectrum of LMNCO-NF0. However, after doped Nb and F elements, it is obviously observed the binding energy peak of F 1s and Nb 3d for the LMNCO-NF1, LMNCO-NF3, and LMNCO-NF5. The peak intensities of F1s and Nb 3d increase along with the increment amount of the Nb and F doping. The above analysis suggests that Nb and F successfully doped into the Mn site and O site of Li1.2Mn0.54Co0.13Ni0.13O2.

The initial charge/discharge curves of LMNCO-NF0, LMNCO-NF1, LMNCO-NF3, and LMNCO-NF5 between 2.0 V and 4.8 V at 0.1 C. are showed in the **Figure 5**. All samples exhibit two plateaus. One plateau below 4.5 V is related to lithium extraction form the layered LiMO2, and the other plateau above 4.5 V corresponds to the lithium-ion extraction from the Li2MnO<sup>3</sup> component and accompanied by the extraction of oxygen (Johnson et al., 2004). The charge/discharge capacities of LMNCO-NF0, LMNCO-NF1, LMNCO-NF3, and LMNCO-NF5 are 354.8/231.8, 325.1/254.1, 337.3/269.3, and 318.9/239.9, respectively. Therefore, for LMNCO-NF1, LMNCO-NF3, and LMNCO-NF5, the initial coulombic efficiency reaches 78.2, 79.84, and 75.2%, respectively, which are higher than that of LMNCO-NF0(65.3). Furthermore, as can be seen in Table S6, the coulombic efficiency of LMNCO-NF3 is better than these previously reported articles (Wang et al., 2013; Chao et al., 2014; Jin et al., 2014; He et al., 2015; Yin et al., 2015). It is confirmed that the covalency of the metaloxygen bond and electronegativity of the dopant ions have significant influence on the degree of oxygen loss from the lattice, easily mitigating the structure change of the Li2MnO<sup>3</sup> during charging process (Wang and Manthiram, 2013). Therefore, the initial coulombic efficiency is improved by Nb and F effective co-doping.

**Figure 6** shows the charge/discharge curves of all samples at different rates of 0.1, 0.5, 1, and 5C. It is seen that the rate performance is enhanced by moderate amount of Nb and F co-doping. The discharge capacities of LMNCO-NF1 are 254.8, 245.2, 213.5, 162 mAh g−<sup>1</sup> at 0.1 C, 0.5 C, 1 C, and 5 C, while the discharge capacities of LMNCO-NF3 are 269.8, 257.3, 235.3, and 173.3 mAh g−<sup>1</sup> respectively, which are higher than that of LMNCO-NF0 (231.2, 201.9, 149.9, and 70.4 at 0.1C, 0.5C, 1C,

and 5C, respectively). The improved rate performance is related to the inequitable valent doping of Nb and F element in Mn and O sites, respectively, which can increase the oxygen vacancies in material surface, enhancing the electronic conductivity of host material eventually. However, the discharge capacities of LMNCO-NF5 are 240.6, 206.7, 158, and 94.8 mAh g−<sup>1</sup> at 0.1C, 0.5C, 1C, and 5C, respectively. The LMNCO-NF5 show relatively poor rate performance and lower initial coulombic efficiency, owing to that the excessive Nb and F co-doping can form a thick Li3NbO4/Nb2O5/LiF layer, and lengthen the lithium ion diffusion path.

**Figure 7** shows the cycling performance and selected chargedischarge cures of LMNCO-NF0, LMNCO-NF1, LMNCO-NF3, and LMNCO-NF5, respectively. The discharge capacity of LMNCO-NF0 is only 115.6 mAh g−<sup>1</sup> with capacity retention of only 76.5% after 200 cycles at 1 C. In comparison, the discharge capacity of LMNCO-NF1, LMNCO-NF3, and LMNCO-NF5 are 191.8, 221.5, and 140.9 mAh g−<sup>1</sup> with capacity retention of 88.3, 94.2, and 89.3%, respectively, which are higher than that of LMNCO-NF0. The fading capacity of LMNCO-NF0 is attributed to the side reaction between organic electrolyte and electrode to form inactive surface layers, and the unfavorable structure change during cycling process. As surface corrosion of the sample can trigger the dissolved Mn, resulting in the capacity fade, Nb and F co-doping can effectively suppress Mn dissolution (Table S4), stabilize the surface structure. The improved cycling performance is ascribed to Nb and F co-doping. Specifically, the binding between O and Nb (Table S5), which is stronger than the Mn-O bond, and the stronger binding of Nb-O can effectively suppress the loss of oxygen from the lattice (Deng and Manthiram, 2011).

Meanwhile the strongly electronegative of F<sup>−</sup> ions can keep the structure stability of Li2MnO3. Partial substitution of O2<sup>−</sup> anions by F<sup>−</sup> is also proposed as a way to stabilize the layered structures with the formation of strong M-F bonds (Li et al., 2015).

The inhibitory effect of structural transformation of Nb and F co-doped could be seen from the charge-discharge curves. As seem **Figure 7B**, the corresponding discharge profile of LMNCO-NF0 at 1C exhibited obvious voltage fade after cycling process. The average voltage only is 2.83 V with the average voltage

of 77.5% at first cycle after 200 cycles. During the long-term cycling process, the migration of transition metal ions will result in the transformation of layered structure to spinel structure, accompanied by the continous voltage fade as the consequence (**Figure 8E**). While, after Nb and F co-doping, the voltage decay are effectively suppressed. The average voltages of LMNCO-NF1, LMNCO-NF3, and LMNCO-NF5 can reach 3.12 V, 3.23 V, and 3.01 V after 200 cycles, respectively, The Nb5<sup>+</sup> and F<sup>−</sup> doped into the bulk material which could suppress the migration of TM ions to Li layer, and thus suppress the layered to spinel phase transformation (**Figure 8F**).

In order to further confirm the effect of Nb and F co-doping on the electrochemical performance, the cyclic voltammetry (CV) measurements from 2.0 to 4.8 V at a scan rate of 0.1 mVs−<sup>1</sup> were carried out, as shown in Figure S3. For all samples, there are two anodic peaks at 4.08 and 4.62 V during the initial cycle. The anodic peak at 4.08 V belongs to lithium deintercalation from LMO structure accompanied by the oxidation of Ni2+/Ni4<sup>+</sup> and Co3+/Co4+, and the other peak at 4.62 V corresponds to lithium deintercalation from Li2MnO<sup>3</sup> structure, which is associated with the Mn4<sup>+</sup> activation process (Yu and Zhou, 2012). However, the anodic peak at high potential of 4.62 V is ascribed to the irreversible reaction about removal of Li2O from the Li2MnO<sup>3</sup> component, which will disappear in the following cycles (Ohzuku et al., 2011). During the reduction process, the cathodic peak at 3.75 V corresponds to the redcution of Ni2+/Ni4<sup>+</sup> and Co3+/Co4+, and the activation of the Li2MnO<sup>3</sup> component. there is the peak at about 3.20 V in following cathodic reaction (Jin et al., 2014). In the subsequent cycles, the peak at 4.6 and 3.9 V of all materials disappears, and the peak at 3.7 V emerges. In addition, for LMNCO-NF0, the reduction peak below 3.0 V is observed after 100 cycles, and the intensity gradually increase with the increase of the cycle, suggesting the formation of the spinel phase during the cycling. However, the reduction peak below 3.0 V for LMNCO-NF1, LMNCO-NF3, and LMNCO-NF5 are hardly observed after 100 cycles, and the reduction peak below 3.0 V appeared when the number of cycles up to 200 cycles. This result indication that the phase transformation from layered into spinel structure is effectively suppressed after Nb and F co-doping.

The results of HRTEM investigations of LMNCO-NF0, LMNCO-NF1, LMNCO-NF3, and LMNCO-NF5 after 100 and 200 cycles are displayed in **Figure 8**. As seen **Figure 8A**, the interplanar spacing of LMNCO-NF0 after 100 and 200 cycles show (003) planes of layer phase, and the (112) planes of the spinel phase. Moreover, LMNCO-NF0 was found to the formation of local amorphous and spinel phase domains after 100 cycles, and the local amorphous areas increase after 200 cycles. In contrast local amorphous domains after 100 cycles do not appear. While local amorphous domains and spinel phase of all doped materials after 200 cycles were observed. Meanwhile the local amorphous areas of all doped materials after 200 cycles are less than that of LMNCO-NF0. Furthermore, the (220) and (440) planes of the spinel phase not be observed after 100 cycles, but after 200 cycles emerges. The Nb and F co-doping can effectively mitigate the migration of TM

TABLE 2 | Impedance parameters derived using equivalent circuit model (Figure S4B) for LMNCO-NF0, LMNCO-NF1, LMNCO-NF3, and LMNCO-NF5 electrodes before cycling (fully discharged).


ions, and suppressed the voltage fade during high voltage cycling.

To gain insight into the effect of Nb and F co-doping on kinetic behavior of lithium ion diffusion, the EIS was carried out. Figure S4 and **Figure 9** show Nyquist plots of all samples before and after cycling, respectively. All the Nyquist plots are consist of the depressed semicircles and a slope. The high frequency intercept at the real axis belongs to the ohmic resistance (R) of interaction between electrolyte and electrode, and the semicircle in the middle frequency region corresponds to charge-transfer resistance (Rct). The slope in the low frequency region is related to the diffusion of lithium ion in the bulk material (Koga et al., 2013; La Mantia et al., 2013; Wang et al., 2015, 2017). The values of Rct and R of all materials before cycling are simulated by the equivalent circuit and the result were listed in **Table 2**. In addition, the lithium-ion diffusion coefficients of all samples are calculated by the following equations (Fey et al., 2001; Levi and Aurbach, 2004; Lin et al., 2013) and also listed in **Table 2**.

$$\mathbf{D}\_{\mathrm{Li}}{}^{+} = \frac{\mathbb{R}^{2}\mathbb{T}^{2}}{2\mathbf{n}^{4}\mathbb{F}^{4}\mathbb{C}^{2}\_{\mathrm{Li}^{2}}\sigma^{2}} \tag{1}$$

$$\boldsymbol{Z}^{'} = \boldsymbol{R}\_{\Omega} + \boldsymbol{R}\_{\text{ct}} \boldsymbol{\sigma} \boldsymbol{\alpha}^{-1/2} \tag{2}$$

Here, T is 298 K, R is gas constant (8.314 J K−<sup>1</sup> mol−<sup>1</sup> ), A is the surface area of the electrode, F is the Faraday constant (96,485 C mol−<sup>1</sup> ), n is the number of electrons involved in reaction, C is the concentration of lithium ion. Where ω is the angular frequency in the low frequency region and σ is the Warburg coefficient. The graph of Z′ against ω −1/2 in the low frequency region is a straight line with the slope of σ.

As seen **Table 2**, the Rct of LMNCO-NF1, LMNCO-NF3, and LMNCO-NF5 are much lower than that of LMNCO-NF0, suggesting that the Nb and F co-doping will increase electronic conductivity of bulk material and improve the kinetics of lithiumion diffusion, due to Li slab space is enlarged. Therefore, all doped materials have higher capacity and excellent rate performance.

Meanwhile, further explain the effect of the Nb and F doped on the cycling performance were performed by EIS after the 1st, 100th, and 200th cycles. For all samples, the value of Rct is simulated by the equivalent circuit and the lithium-ion diffusion coefficients are calculated by the equation (1 and 2), while the results are listed in the Table S7. The charge transfer resistance (Rct) of LMNCO-NF0 increases continuously, while lithium ion diffusion coefficient (DLi+) decreases rapidly and remains only half of the pristine value after 200 cycles, resulting from the severe structure change during long-term cycling. While the charge transfer resistance (Rct) and lithium ion diffusion coefficient (DLi+) of all doped materials exhibit a little variation and keep acceptable values after 200 cycles, indicating the excellent reaction kinetic. These suggest that Nb and F co-doping could keep the capacity stability which could be ascribed to that doping of Nb and F ions into the bulk layered materials could suppress the change structure from a layered into a spinel structure during cycling.

### CONCLUSION

Nb and F co-doped Li1.2Mn0.54Ni0.13Co0.13O<sup>2</sup> have been fabricated by using traditional solid phase method. The Nb5<sup>+</sup> and F<sup>−</sup> ions are successfully doping into the Mn4<sup>+</sup> site and O2<sup>−</sup> sites, respectively, which is beneficial for suppress the loss of the oxygen and the mixed migration of transition metal ions. Therefore, Nb and F co-doping can enhance initial coulombic efficiency of 81.4% and rate performance with discharge capacity with 269.8, 257.3, 235.3, and 173.3mAh g−<sup>1</sup> at the discharge rates of 0.1, 0.5, 1 C, 5 C and cycling performance with discharge capacity of 221.5 mAh g−<sup>1</sup> after 200 cycles at 1 C, as well as suppress voltage fade of Li1.2Mn0.54Ni0.13Co0.13O<sup>2</sup> during cycling. It is convinced that the Li1.2Mn0.54Ni0.13Co0.13O<sup>2</sup> after Nb and F co-doping can satisfy the requirements of the electric vehicle and the renewable energy storage, and become advanced lithium ion cathode materials for application of Li-ion battery.

### AUTHOR CONTRIBUTIONS

LM: Designer of the scheme and main performer of the experiment. BZ: Main advisor. YC: Provide assistance in characterization. J-FZ: Main advisor and participant. C-HW: Participant of the experiment. X-WW: Provide assistance in experiment. All authors listed, have made substantial, direct and intellectual contribution to the work, and approved it for publication.

### ACKNOWLEDGMENTS

This study was supported by National Natural Science Foundation of China (Grant No. 51272290, 51472272, and 51502350).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2018.00076/full#supplementary-material

## REFERENCES


Li1.2Mn0.54Ni0.13Co0.13O<sup>2</sup> Li-rich layered oxides. J. Power Sources 346, 31–39. doi: 10.1016/j.jpowsour.2017.02.036


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Ming, Zhang, Cao, Zhang, Wang, Wang and Li. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# High-Power-Density, High-Energy-Density Fluorinated Graphene for Primary Lithium Batteries

Guiming Zhong1,2†, Huixin Chen1,2†, Xingkang Huang3†, Hongjun Yue1,2 \* and Canzhong Lu1,2

*<sup>1</sup> CAS Key Laboratory of Design and Assembly of Functional Nanostructures, and Fujian Provincial Key Laboratory of Nanomaterials, Fujian Institute of Research on the Structure of Matter, Chinese Academy of Sciences, Fuzhou, China, <sup>2</sup> Xiamen Institute of Rare Earth Materials, Haixi Institutes, Chinese Academy of Sciences, Xiamen, China, <sup>3</sup> Department of Mechanical Engineering, University of Wisconsin-Milwaukee, Milwaukee, WI, United States*

### Edited by:

*Jiexi Wang, Central South University, China*

#### Reviewed by:

*Hong Guo, Yunnan University, China Xiaobo Ji, Central South University, China*

> \*Correspondence: *Hongjun Yue*

*hjyue@fjirsm.ac.cn*

*† These authors have contributed equally to this work.*

#### Specialty section:

*This article was submitted to Physical Chemistry and Chemical Physics, a section of the journal Frontiers in Chemistry*

Received: *24 January 2018* Accepted: *22 February 2018* Published: *09 March 2018*

#### Citation:

*Zhong G, Chen H, Huang X, Yue H and Lu C (2018) High-Power-Density, High-Energy-Density Fluorinated Graphene for Primary Lithium Batteries. Front. Chem. 6:50. doi: 10.3389/fchem.2018.00050* Li/CF<sup>x</sup> is one of the highest-energy-density primary batteries; however, poor rate capability hinders its practical applications in high-power devices. Here we report a preparation of fluorinated graphene (GFx) with superior performance through a direct gas fluorination method. We find that the so-called "semi-ionic" C-F bond content in all C-F bonds presents a more critical impact on rate performance of the GF<sup>x</sup> in comparison with sp<sup>2</sup> C content in the GFx, morphology, structure, and specific surface area of the materials. The rate capability remains excellent before the semi-ionic C-F bond proportion in the GF<sup>x</sup> decreases. Thus, by optimizing semi-ionic C-F content in our GFx, we obtain the optimal x of 0.8, with which the GF0.8 exhibits a very high energy density of 1,073 Wh kg−<sup>1</sup> and an excellent power density of 21,460 W kg−<sup>1</sup> at a high current density of 10 A g−<sup>1</sup> . More importantly, our approach opens a new avenue to obtain fluorinated carbon with high energy densities without compromising high power densities.

Keywords: fluorinated graphene, carbon fluoride, primary lithium battery, nuclear magnetic resonance, high power density

### INTRODUCTION

Fluorinated carbon (CFx) possesses a very high theoretical energy density (2,180 Wh kg−<sup>1</sup> when x equals 1 for fluorinated graphite) as a cathode material for primary lithium batteries, thus has been strongly desired in many civil and military applications that require a long service-life, wide range of operating temperatures, as well as high energy densities and reliability. Fluorinated graphite has been widely investigated (Nakajima et al., 2002; Guérin et al., 2004; Zhang Q. et al., 2010). Besides graphite, other types of carbons such as carbon nanotubes, carbon nanofibers, C60, and mesoporous carbon, have also been applied for fluorination (Matsuo and Nakajima, 1996; Mickelson et al., 1998; Lam and Yazami, 2006; Yazami et al., 2007; Zhang W. et al., 2010; Fulvio et al., 2011; Guérin et al., 2012). Among these, a fluorinated mesoporous carbon (CF0.54) and a fluorinated coke displayed excellent performance; the fluorinated mesoporous carbon delivered a capacity of 515 mAh g−<sup>1</sup> with discharge plateau of 2.75 V at a current rate of 5C (Fulvio et al., 2011), while the fluorinated coke displayed a maximum power density of about 14,400 W kg−<sup>1</sup> with energy

density of 500 Wh kg−<sup>1</sup> (Lam and Yazami, 2006). However, the power densities of these materials are far from satisfaction because of the poor electronic conductivity of the CF<sup>x</sup> materials due to the strong covalent C-F bond.

Coating of highly conductive materials, such as carbon, polypyrrole, and polyaniline on the surface of carbon fluorides is helpful to improve the rate capability (Zhang Q. et al., 2010; Groult et al., 2011; Li et al., 2016); for example, a graphite fluoride coated with polyaniline delivered an energy density of about 1,200 Wh kg−<sup>1</sup> with power density higher than 10,000 W kg−<sup>1</sup> at current rate of 8C (Li et al., 2016). An amazing rate performance (48,800 W kg−<sup>1</sup> at 30C) was achieved by reducing the fluorine content on the surface of the carbon fluorides through hydrothermal method, which greatly improved the electronic conductivity (Dai et al., 2014). However, the hydrothermal reaction would be restricted for practical application because it could introduce a large amount of hydroxyl group, jeopardizing the calendar life of the Li/CF<sup>x</sup> battery.

Fluorinated graphene, as a two-dimensional (2D) material, can shorten the diffusion path of lithium ions, which is helpful for rapid transfer of lithium ions (Zhang S. S. et al., 2009; Feng et al., 2016), thus opening an alternative avenue to chase excellent rate capability. Until now the best fluorinated graphene reported in the literature only can work at 5C, gaining a capacity of 356 mAh g−<sup>1</sup> (GF0.47) (Damien et al., 2013; Meduri et al., 2013; Zhao et al., 2014; Feng et al., 2016). Therefore, uncovering the reasons hindering fluorinated graphene from achieving excellent rate capability is highly desired.

In this study, a fluorinated multilayered graphene (GFx) was prepared by a direct gas fluorination of RGO instead of graphene oxides (Damien et al., 2013; Meduri et al., 2013; Zhao et al., 2014; Feng et al., 2016), and was investigated using <sup>13</sup>C and <sup>19</sup>F NMR spectra, indicating that the controlled formation of the so-called "semi-ionic" C-F bond in the GF<sup>x</sup> is the most critical factor to achieve high power densities with high energy densities. With an optimal semi-ionic C-F bond ratio, our GF<sup>x</sup> showed extraordinary performance with a power density of 21,460 W kg−<sup>1</sup> and an energy density of 1,073 Wh kg−<sup>1</sup> when the x in GF<sup>x</sup> equals 0.8, superior to most of the previously reported fluorinated carbons (Mickelson et al., 1998; Lam and Yazami, 2006; Shulga et al., 2007; Yazami et al., 2007; Zhang W. et al., 2010; Fulvio et al., 2011; Groult et al., 2011; Guérin et al., 2012; Damien et al., 2013; Meduri et al., 2013; Sun et al., 2014; Zhao et al., 2014; Feng et al., 2016; Li et al., 2016; Wang et al., 2016). In addition, our approach applied to synthesize GF<sup>x</sup> is facile, and easy for scale-up, exhibiting very promising practical application.

## EXPERIMENTAL

### Preparation of Fluorinated Graphene

Fluorinated graphenes were prepared by a one-step gas-phase fluorination of RGO as described in previous work (Yue et al., 2013; Shao et al., 2016). Graphene oxides were prepared using Hummers method (Hummers and Offeman, 1958), which was subjected to thermal reduction for 10 h under a H<sup>2</sup> flow (5 vol. % in Ar) of 20 sccm at 1,000◦C. The resulting RGO was employed to prepared fluorinated graphene materials at 400, 430, and 460◦C in fluorinating gas atmosphere for 12 h, obtaining GF0.5, GF0.8, and GF1.1, respectively. The F/C atomic ratios in these samples were determined by quantitative <sup>13</sup>C NMR: F/C = (SCF + 2 × SCF2)/(SC+ SCF2+ SCF), where S is the integrated intensities of the 13C NMR peaks.

### Material Characterization

X-ray powder diffraction (XRD) technique was employed to characterize phases of as-prepared materials, using Cu K<sup>α</sup> radiation (1.54178 Å) on a Miniflex600 (Rigaku, Japan) instrument. XRD patterns were collected with a step of 0.0167◦ , and 20 s per step. <sup>13</sup>C and <sup>19</sup>F magic angle spinning (MAS) NMR experiments were performed on Bruker 600 MHz AVANCE III spectrometer using Hahn-echo pulse under the spinning frequencies of 12 and 60 kHz, respectively. Recycle delays of 60 and 20 s were applied for complete relaxation of excited magnetization for the acquisition of quantitative <sup>13</sup>C and <sup>19</sup>F NMR spectra. The chemical shifts of <sup>13</sup>C and <sup>19</sup>F were referenced to diamantine (38.6 ppm) and LiF (−204 ppm). X-ray photoelectron spectroscopy (XPS) of the samples was measured by an ESCALAB 250Xi spectrometer (Thermo Fisher). SEM images were performed on scanning electron microscopy (SEM) (ZEISS). The transmission electron microscopy (TEM) and high-resolution TEM (HRTEM) analysis were performed on Tecnai F20 (FEI, US), operating at 200 kV. Nitrogen adsorption/desorption isotherms and Brunauer-Emmett-Teller (BET) surface area were performed on a Quantachrome instrument Autosorb-iQ.

### Electrochemical Test

The cathode was prepared by mixing 80 wt.% fluorinated graphene, 10 wt.% acetylene black, and 10 wt.% poly (vinylidene fluoride) (PVDF). Aluminum disks were employed as current collectors and the active materials on the Al disks are between 1.5 and 2 mg cm−<sup>2</sup> . A lithium metal disk was used as a counter electrode, and electrolytes were 1 M LiPF<sup>6</sup> dissolved in ethylene carbonate/dimethyl carbonate (1:1 volume ratio). Discharge tests were performed at various currents with a cutoff voltage of 1.5 V by a LAND CT2001A battery test system at 25◦C.

## RESULTS AND DISCUSSION

### Electrochemical Performance

It is well known that the fluorine content in fluorinated carbon significantly affects the electrochemical performance. A low fluorine content results in a good power density but relatively low energy density. To achieve excellent power densities with high energy densities, we prepared fluorinated graphene (GFx) with various F content and investigated the factors influencing the electrochemical performance. The F content in the GF<sup>x</sup> was controlled by varying the temperatures during the fluorination; for example, GF0.5, GF0.8, and GF1.1 were obtained at 400, 430, and 460◦C, respectively.

When the x in GF<sup>x</sup> equals 0.5, as shown in **Figure 1A**, the GF0.5 material delivered a capacity of 615 mAh g−<sup>1</sup> (98.7% of its theoretical capacity) with a discharge plateau at ∼2.9 V at a low current density of 20 mA g−<sup>1</sup> (∼1/32C, 1C = 623 mA g−<sup>1</sup>

for GF0.5). When the current densities increased to 1 A g−<sup>1</sup> , the capacity of the GF0.5 decreased slightly to around 580 mAh/g. Even if the current densities of 5 and 10 A g−<sup>1</sup> were applied, the GF0.5 was still able to deliver 76.9 and 62.4%, respectively, of its theoretical capacity, exhibiting very excellent rate capability. To the best of our knowledge, this is the best rate capability among the untreated fluorinated carbon materials reported in the literatures (Giraudet et al., 2006; Yazami et al., 2007; Zhang W. et al., 2009; Fulvio et al., 2011; Dubois et al., 2012).

With the impressive rate capability, the specific capacity of the GF0.5 is yet to be satisfied due to its low theoretical capacity (623 mAh g−<sup>1</sup> ). In contrast, the theoretical capacity of the GF1.1 is up to 896 mAh g−<sup>1</sup> . As shown in **Figure 1B**, the GF1.1 depicts a capacity of 779 mAh g−<sup>1</sup> , which is 1.34 times of that of the GF0.5 at 20 mA g−<sup>1</sup> . However, at high current densities, the voltage plateaus of the GF1.1 decreased much more significantly compared with those of the GF0.5, suggesting much higher polarization for the GF1.1 material. For example, at 5 A g−<sup>1</sup> the voltage plateau of the GF1.1 dropped to 1.77 V with a capacity of 511 mAh g−<sup>1</sup> that is only 57% of its theoretical capacity. Furthermore, the GF1.1 could not work at 10 A g−<sup>1</sup> at all.

The relatively poor rate capability of the GF1.1 is associated with its poor electrical conductivity, which needs to be addressed. Therefore, the way to gain high power densities while retaining high energy densities is to prepare CF<sup>x</sup> with high F content without compromising the electrical conductivity. To achieve this goal, instead of with complicated surface treatments (Groult et al., 2011; Reddy et al., 2013; Dai et al., 2014), we took the full advantage of high electrical conductivity of the semi-ionic CF bond in CF<sup>x</sup> (discussed in detail below). In other words, we made great effort to improve the F content within the limit without losing the semi-ionic bonds in the CFx. With an optimal F content, the as-designed GF0.8 showed high capacity with excellent rate performance (**Figure 1C**). The specific capacity of the GF0.8 was 770 mAh g−<sup>1</sup> , 97.8% of its theoretical capacity, at 20 mA g−<sup>1</sup> (about 1/39C, 1C = 788.2 mA g−<sup>1</sup> for GF0.8). When the current densities enhanced to 10 A g−<sup>1</sup> (∼12C), the GF0.8 exhibited an extraordinary power density of 21,460 W kg−<sup>1</sup> with a capacity of 511 mAh g−<sup>1</sup> , corresponding to 64.9% of the theoretical capacity and an energy density of 1,073 Wh kg−<sup>1</sup> .

To understand the electrochemical performance of GF<sup>x</sup> electrode, we compared the kinetic properties of GF0.5, GF0.8, and GF1.1 by electrochemical impedance spectroscopy (EIS) measurement. To exclude the effect of conductive carbon, we prepared the electrodes with the GF<sup>x</sup> materials without addition of carbon black. In the absence of conductive carbon, however, the impedances of the fresh electrodes were too extremely high to obtain accurate results for comparison (Figure S1A) at opencircuit potentials. Therefore, 1% of the theoretical capacity was discharged to reasonably compare the impedances of the CF<sup>x</sup>

electrodes. As illustrated in Figure S1B, the charge-transfer resistance (Rct) of the GF1.1 is 2,500 , much higher than that of the GF0.5 (∼1,400 ) and the GF0.8 (∼1,600 ).

Ragone plots were employed to depict the advanced electrochemical performance of the GF0.8 (**Figure 1D**), which is superior to most of the fluorinated carbon and other primary batteries in the literature (Table S1) (Giraudet et al., 2006; Lam and Yazami, 2006; Yazami et al., 2007; Zhang W. et al., 2009; Fulvio et al., 2011; Meduri et al., 2011; Dubois et al., 2012; Guérin et al., 2012; Wang et al., 2012; Adcock et al., 2013; Reddy et al., 2013; Dai et al., 2014, 2017; Sideris et al., 2014; Li and Feng, 2015; Liang et al., 2015; Zhang et al., 2015; Li et al., 2016; Zhu et al., 2017). For example, Lam and Yazami prepared fluorinated coke materials, which showed a maximum power density of about 14,400 W kg−<sup>1</sup> with energy density of 500 Wh kg−<sup>1</sup> (Lam and Yazami, 2006). Sulfur material typically display greater energy density but much lower power density due to awfully low electrical conductivity (∼10−<sup>30</sup> S cm−<sup>1</sup> ), for which a large amount of carbon has to been applied during electrode preparation (Manthiram et al., 2015; Pang et al., 2015).

## Bonding Characteristics of GF<sup>x</sup> Materials

As mentioned above, we achieve high energy densities with excellent power densities by the full use of the advantage of semiionic C-F bonds in the CFx. Ionic, semi-ionic, and covalent C-F bonds are the three types of C-F bonds in the CFx. Ionic C-F bonds are typically only formed when x in CF<sup>x</sup> is very small (e.g., <0.05 for graphite; Amine and Nakajima, 1993; Nansé et al., 1997; Giraudet et al., 2006), which is not useful in CFx/Li batteries because of the very low capacity. Covalent C-F bonds possess the characteristics of the sp<sup>3</sup> C with the F-C-C angles larger than 90◦ and the neighboring C-C bond length of ∼0.153 nm (Sato et al., 2004; Figure S2A). High covalent C-F ratio may ruin the conductive network of the conjugated double bonds, exhibiting insulating property (electrical conductivity lower than 10−<sup>15</sup> S cm−<sup>1</sup> ; Sato et al., 2004).

In contrast, semi-ionic C-F bonds are essentially covalent, with which, however, the conjugated C-C bonds are preserved between carbon atoms unbounding to fluorine with the F-C-C angle of 90◦ and the neighboring C-C bond length of ∼0.14 nm (Sato et al., 2004; Figure S2B). Unlike what happens in the presence of covalent C-F bonds, with semi-ionic C-F bonds and a certain proportion of sp<sup>2</sup> C, CF<sup>x</sup> may have high electronic conductivity, for example, the electronic conductivity of CxF with x around two ranged between 5 × 10−<sup>8</sup> and 1 × 10−<sup>7</sup> S cm−<sup>1</sup> (Sato et al., 2004). In a word, the semi-ionic C-F bonds do not significantly degrade the electrical conductivity of the CF<sup>x</sup> (Mallouk and Bartlett, 1983; Amine and Nakajima, 1993; Sato et al., 2004; Zhang W. et al., 2010), which can be employed to gain high power densities with high energy densities.

### Solid-State <sup>19</sup>F and <sup>13</sup>C NMR Spectra

<sup>19</sup>F NMR spectra were employed to distinguish covalent and semi-ionic C-F bonds. **Figure 2A** shows that the <sup>19</sup>F MAS NMR spectra acquired at a spinning frequency of 60 kHz, in which the <sup>19</sup>F resonances consist of three parts. Resonance peaks located between −80 and −135 ppm belong to the signal of CF<sup>2</sup> groups. Resonance peaks located between −140 and −180 ppm were assigned to the signal of the semi-ionic CF group, while the peaks located between −185 and −189 ppm were assigned to the signal of the covalent CF group (Panich et al., 1997; Krawietz and Haw, 1998; Giraudet et al., 2007; Leifer et al., 2010; Ahmad et al., 2013).

The semi-ionic C-F bond ratios in the GF<sup>x</sup> were determined by fitting of <sup>19</sup>F NMR spectra (Figure S3). As shown in **Figure 2C**, the semi-ionic bond ratio in GF0.5 is 38.5%, which slightly decreases to 37.7% when the x in GF<sup>x</sup> increases to 0.8. Beyond that point, the semi-ionic bond ratio in the GF<sup>x</sup> dramatically dropped (e.g., 16.1% in GF1.1), which is consistent with the facts that the rate capability of the GF0.8 is similar with that of the GF0.5, but much better than that of the GF1.1. For example, at 5 A g−<sup>1</sup> , the energy densities of the GF0.5, GF0.8, and GF1.1 are 10,856, 11,352, and 898 Wh kg−<sup>1</sup> , respectively (**Figure 1**).

Another possible reason for poor electrical conductivity related to the low semi-ionic C-F bond ratio is the high F content resulting in low content of the sp<sup>2</sup> hybridized C. **Figure 2B** exhibits the <sup>13</sup>C NMR spectra, in which the resonance peaks at around 130, 111, and 87 ppm are associated with the sp<sup>2</sup> C, CF2, and CF, respectively. Their contents were calculated by fitting the peaks (Figure S4). The sp<sup>2</sup> C contents in the GF0.5, GF0.8, and GF1.1 are calculated to be 49.8, 22.8, and 3.4%, respectively (**Figure 2D**). When the sp<sup>2</sup> C content goes down to too low value (e.g., 3.4% in the GF1.1), the electrical conductivity is compromised (Yue et al., 2013), resulting in a poor rate capability.

Based on the <sup>19</sup>F and <sup>13</sup>C NMR analysis, we can conclude that with the increasing F content in GFx, the sp<sup>2</sup> C ratio decreases while the semi-ionic C-F bond ratio remains unchanged until the critical x of 0.8, beyond which the electron-transfer ability of sp<sup>2</sup> C is compromised. This matches very well the electrochemical performance of GF0.5, GF0.8, and GF1.1, namely, the GF0.5 and GF1.1 depicted high power densities and high capacities, respectively, but the GF0.8 exhibited the optimal electrochemical performance (21,460 W kg−<sup>1</sup> and 1,073 kWh kg−<sup>1</sup> ).

### XPS Spectra

Semi-ionic C-F bond ratios in CF<sup>x</sup> also have been analyzed using X-ray photoelectron spectroscopy (XPS) spectra (Doniach and Sunjic, 1970; Tressaud et al., 1996; Nansé et al., 1997; Leiro et al., 2003; Park et al., 2008). We therefore conducted XPS analysis for our GF<sup>x</sup> materials for comparison. As shown in **Figure 3A**, the carbon in the RGO was mainly composed of C=C sp<sup>2</sup> bonds (284.4 eV) and minor bonds C-O-C (∼286.5 eV) and O-C=O (∼288.6 eV). In the GF0.5 (**Figure 3B**), the binding energies at 288.0 and 288.8 eV were assigned to the semi-ionic and the covalent CF bonds, respectively (Tressaud et al., 1996; Nansé et al., 1997; Wang et al., 2014).

In contrast, the semi-ionic C-F bonds were barely detected by XPS C1s spectra for the GF0.8 and GF1.1 (**Figures 3C,D**), which are consistent with the results from the XPS F1s spectra (Figure S5); no semi-ionic C-F bond was detected in the GF0.8 and GF1.1. In other words, the C-F bond on the surface of the GF0.8 and GF1.1 materials were mainly covalent. Apparently, these results are inconsistent with those from NMR, where the semi-ionic C-F ratios in GF0.8 and GF1.1 are much higher. This is because that

the surface of the GF<sup>x</sup> is subjected to more attack than the bulk during fluorination, resulting in higher fluorinating levels on the GF<sup>x</sup> surface than in the bulk. The semi-ionic C-F in GF0.8 and GF1.1can be detected by NMR but not by XPS because that is for surface analysis while the former obtains bulk information. As a conclusion, NMR is more suitable for analyzing semi-ionic C-F bond ratio in CF<sup>x</sup> than XPS (**Figures 3E,F**).

Besides the semi-ionic C-F bond ratio, other factors that may influence the electrochemical performance of the GF<sup>x</sup> were also investigated, including structure, morphology, and surface area.

### XRD Patterns and Morphology Features XRD Patterns and TEM Images

XRD patterns of fluorinated graphene materials were shown in Figure S6. Two peaks centered at around 25.4 and 43.2◦ were observed for the RGO, which correspond with the 002 and 100 reflections of graphitic carbon, respectively. After fluorination, the 002 reflection decreased significantly while a new peak at ∼15◦ developed for the three fluorinated samples, which may be assigned to the 001 plane of fluorinated phase (Hamwi, 1996; Meduri et al., 2013) or the expanded 002 plane (Zhang et al., 2015). The layer thickness of the RGO increased from ∼0.4 to ∼0.6 nm after fluorination as indicated by HRTEM (Figures S7–S9), which is in accordance with results from XRD patterns (Figure S6). Note also that the 100 reflections shifted to lower angle with increasing fluorination levels, indicating an increasing C-C in-plane length, which is consistent with the trend of fluorinated graphite (Guérin et al., 2004). Theoretically, increased interplanar distances in the GF<sup>x</sup> may facilitate the intercalation of lithium ions. However, considering the fact that GF1.1 shows a poor rate capability than the GF0.8, the increasing of d-spacing does not offer enough help the GF1.1 to gain excellent power density.

### SEM Images

**Figure 4** showed the SEM images of the RGO and GF<sup>x</sup> materials, indicating their secondary particle sizes larger than 10 microns due to the aggregation of GO upon drying. Interestingly, compared with pristine RGO, the fluorinated graphene materials exhibit clear lamellar architectures that were marked by the yellow arrows in **Figures 4d,f,h**; especially in GF0.8 and GF1.1, some fluorinated graphene layers seem to be peeled off. Although fluorination introduces F into the interlayers of C, the d-spacing of 002 facet only expands from ∼0.33 to ∼0.6 nm. Therefore, we believe the splitting of graphene layers is not due to the F expansion, but to the attack by the fluorinating gas during the fluorination at high temperatures. As a matter of fact, the GF1.1

was stripped more severely than the GF0.8 and much more than the GF0.5, which agrees well with their preparation temperatures. The splitting of graphene layers might facilitate the diffusion of solvated lithium ions compared with other types of fluorinated carbon materials (Meduri et al., 2013), thereby enhancing the rate capability of GFx. However, the power density of the GF1.1 is poorer than that of the GF0.8, which is inconsistent with the lager expanded spaces. Therefore, the effect of morphology change on the rate performance is inferior to the semi-ionic C-F bond ratio in the GFx.

### BET Results

During fluorination, the surface area may change, thereby contributing to the improved rate performance. Therefore, the specific surface areas of the pristine RGO and the fluorinated graphene materials were analyzed using Brunauer– Emmett–Teller (BET) method. **Figure 5A** exhibits the nitrogen adsorption-desorption isotherm; accordingly, the BET specific surface areas for the RGO, GF0.5, GF0.8, and GF1.1 were measured to be 70.2, 152.4, 215.8, 238.6 m<sup>2</sup> g −1 , respectively. The relatively small specific area of the RGO is due to the relatively thick

graphene, which is confirmed by HRTEM (Figure S7). The thickness of the RGO is up to 20 layers from the HRTEM observation.

After fluorination, the specific area increased, which is consistent with the SEM observation, in which the fluorinated graphenes exhibit more lamellar architectures, compared with the pristine RGO (**Figure 5**). The pore sizes of the as-prepared fluorinated graphenes were analyzed based on a quenched solid density functional theory (QSDFT) kernel applied to the adsorption branch using a slit/cylindrical pore model (Ravikovitch and Neimark, 2006; Gor et al., 2012). As shown in **Figure 5B**, the RGO possesses mesopores (∼4.2 nm) without any micropores. In contrast, after fluorination, micropores were developed at 1.1 nm for the GF0.5 and an additional micropore size was observed at 0.6 nm for the GF0.8 and GF1.1. Although the increased pores and surface areas might facilitate lithium ion transfer, the higher surface the as-obtained fluorinated graphenes did not result in a better rate capability. Therefore, the surface area of the fluorographenes is not the determining factor affecting the power densities of the fluorinated graphenes.

### CONCLUSIONS

Fluorinated graphenes were prepared using one-step gas fluorination of RGO at elevated temperatures. The impacting factors, including semi-ionic C-F ratio, sp<sup>2</sup> C content, structure, morphology, and specific surface area are investigated to gain fluorinated graphenes with high power densities and high energy densities. The semi-ionic C-F ratio in the fluorinated graphene shows the most critical influence on achievement of high rate performance. Thus, by manipulating the semi-ionic C-F proportion in the fluorinated graphene by temperature control, we obtain the optimal x of 0.8 in GFx; the GF0.8 exhibited a

### REFERENCES


high energy density of 1,073 Wh kg−<sup>1</sup> and an excellent density of 21,460 W kg−<sup>1</sup> at a high current density of 10 A g−<sup>1</sup> (about 12C rate). Compared with those using additional steps (such as C coating and hydrothermal treatment) to improve the rate performance of as-obtained CFx, we offer a one-step approach to obtain high energy densities without compromising power densities for preparation of fluorinated carbon, showing very promising practical application.

## AUTHOR CONTRIBUTIONS

The work cannot be completed without kind cooperation of all authors. GZ: Acquired and analyzed the NMR and XPS data; HC: Carried out the material preparation and electrochemical test; XH and HC: Carried out and analyzed the SEM, TEM, and BET analysis; GZ and XH: Wrote the paper and all authors discussed the results and revised the manuscript; HY, GZ, HC, and XH: Proposed the research; HY and CL: Attained the main financial support for the research and supervised all the experiments.

## ACKNOWLEDGMENTS

The study was supported by the National Natural Science Foundation of China (21503232 and 21603231); Fujian Natural Science Foundation of China (2060203); and Xiamen Science and Technology Planning Project of China (3502Z20161246 and 3502Z20172030).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2018.00050/full#supplementary-material


ultrafast discharge and improved electrochemical performances. J. Electrochem. Soc. 164, A1–A7. doi: 10.1149/2.0451614jes


fullerene and derivatives to graphite. Phys. Chem. Chem. Phys. 12, 1388–1398. doi: 10.1039/B914853A


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The reviewer XJ, and handling Editor declared their shared affiliation.

Copyright © 2018 Zhong, Chen, Huang, Yue and Lu. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.