# CONVERSION MATERIALS FOR RECHARGEABLE BATTERIES, 2nd Edition

EDITED BY : Feixiang Wu, Shengkui Zhong, Junchao Zheng and Danni Lei PUBLISHED IN : Frontiers in Chemistry and Frontiers in Materials

### Frontiers eBook Copyright Statement

The copyright in the text of individual articles in this eBook is the property of their respective authors or their respective institutions or funders. The copyright in graphics and images within each article may be subject to copyright of other parties. In both cases this is subject to a license granted to Frontiers. The compilation of articles constituting this eBook is the property of Frontiers.

Each article within this eBook, and the eBook itself, are published under the most recent version of the Creative Commons CC-BY licence. The version current at the date of publication of this eBook is CC-BY 4.0. If the CC-BY licence is updated, the licence granted by Frontiers is automatically updated to the new version.

When exercising any right under the CC-BY licence, Frontiers must be attributed as the original publisher of the article or eBook, as applicable.

Authors have the responsibility of ensuring that any graphics or other materials which are the property of others may be included in the CC-BY licence, but this should be checked before relying on the CC-BY licence to reproduce those materials. Any copyright notices relating to those materials must be complied with.

Copyright and source acknowledgement notices may not be removed and must be displayed in any copy, derivative work or partial copy which includes the elements in question.

All copyright, and all rights therein, are protected by national and international copyright laws. The above represents a summary only. For further information please read Frontiers' Conditions for Website Use and Copyright Statement, and the applicable CC-BY licence.

ISSN 1664-8714 ISBN 978-2-88966-196-1 DOI 10.3389/978-2-88966-196-1

### 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

# CONVERSION MATERIALS FOR RECHARGEABLE BATTERIES, 2nd Edition

Topic Editors: Feixiang Wu, Central South University, China Shengkui Zhong, Soochow University, China Junchao Zheng, Central South University, China Danni Lei, Sun Yat-sen University, China

Publisher's note: In this 2nd edition, the following article has been updated: Wang X, Jiang Q, Zhang Y, Yuan N and Tang J (2020) High Efficient and Environment Friendly Plasma-Enhanced Synthesis of Al2O3-Coated LiNi1/3Co1/3Mn1/3O2 With Excellent Electrochemical Performance. *Front. Chem.* 8:72. doi: 10.3389/fchem.2020.00072

Citation: Wu, F., Zhong, S., Zheng, J., Lei, D., eds. (2020). Conversion Materials for Rechargeable Batteries, 2nd Edition. Lausanne: Frontiers Media SA. doi: 10.3389/978-2-88966-196-1

# Table of Contents


Yan Li, Han Zhang, Zhe Xiao and Renheng Wang


Jinlei Gu, Chao Shen, Zhao Fang, Juan Yu, Yong Zheng, Zhanyuan Tian, Le Shao, Xin Li and Keyu Xie


Hao Guo, Yong Hu, Xiaoping Zhang, Rongliang Zhang, Dong Hou, Yulei Sui and Ling Wu

*104 Graphene-Modified 3D Copper Foam Current Collector for Dendrite-Free Lithium Deposition*

Juan Yu, Yangyang Dang, Maohui Bai, Jiaxin Peng, Dongdong Zheng, Junkai Zhao, Linbo Li and Zhao Fang


Xinzhi Wang, Qianqian Jiang, Yichi Zhang, Nannan Yuan and Jianguo Tang

# Heterostructured SnO2-SnS2@C Embedded in Nitrogen-Doped Graphene as a Robust Anode Material for Lithium-Ion Batteries

Hui Li <sup>1</sup> , Bao Zhang<sup>1</sup> , Xu Wang<sup>1</sup> , Jie Zhang<sup>1</sup> , Tianhui An<sup>1</sup> , Zhiying Ding<sup>2</sup> , Wanjing Yu<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*

Tin-based anode materials with high capacity attract wide attention of researchers and become a strong competitor for the next generation of lithium-ion battery anode materials. However, the poor electrical conductivity and severe volume expansion retard the commercialization of tin-based anode materials. Here, SnO2-SnS2@C nanoparticles with heterostructure embedded in a carbon matrix of nitrogen-doped graphene (SnO2-SnS2@C/NG) is ingeniously designed in this work. The composite was synthesized by a two-step method. Firstly, the SnO2@C/rGO with a nanolayer structure was synthesized by hydrothermal method as the precursor, and then the SnO2-SnS2@C/NG composite was obtained by further vulcanizing the above precursor. It should be noted that a carbon matrix with nitrogen-doped graphene can inhibit the volume expansion of SnO2-SnS<sup>2</sup> nanoparticles and promote the transport of lithium ions during continuous cycling. Benefiting from the synergistic effect between nanoparticles and carbon matrix with nitrogen-doped graphene, the heterostructured SnO2-SnS2@C/NG further fundamentally confer improved structural stability and reaction kinetics for lithium storage. As expected, the SnO2-SnS2@C/NG composite exhibited high reversible capacity (1201.2 mA h g−<sup>1</sup> at the current rate of 0.1 A g−<sup>1</sup> ), superior rate capability and exceptional long-life stability (944.3 mAh g−<sup>1</sup> after 950 cycles at the current rate of 1.0 A g−<sup>1</sup> ). The results demonstrate that the SnO2-SnS2@C/NG composite is a highly competitive anode material for LIBs.

Keywords: SnO2 -SnS2 , heterostructure, nitrogen-doped graphene, nanoparticle, anode

# INTRODUCTION

In recent years, with the widespread use of portable electronic products and electric vehicles, lithium-ion battery technology needs to be rapidly upgraded (Armand and Tarascon, 2008; Wang et al., 2017a,b, 2019; Chen J. et al., 2018; Zheng et al., 2018; Tong et al., 2019). At present, the main problems faced by lithium-ion batteries (LIBs) are as follows: energy density, cycle life, safety and cost. These problems are mainly related to the anode and cathode of the battery. As far as anode materials are concerned, commercial graphite on the market today is unable to meet the increasing energy density requirements due to its lower capacity (about 372 mAh g−<sup>1</sup> ) (Chen et al., 2018b; Xiao et al., 2018). In the past few decades, researchers developed many high-capacity and structurally

### Edited by:

*Danni Lei, Sun Yat-sen University, China*

### Reviewed by:

*Renheng Wang, Shenzhen University, China Xunhui Xiong, South China University of Technology, China*

> \*Correspondence: *Hui Tong huitong@csu.edu.cn*

### Specialty section:

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

Received: *19 March 2019* Accepted: *25 April 2019* Published: *14 May 2019*

### Citation:

*Li H, Zhang B, Wang X, Zhang J, An T, Ding Z, Yu W and Tong H (2019) Heterostructured SnO*2*-SnS*2*@C Embedded in Nitrogen-Doped Graphene as a Robust Anode Material for Lithium-Ion Batteries. Front. Chem. 7:339. doi: 10.3389/fchem.2019.00339*

**5**

stable anode materials, among which tin oxide anode materials with low discharge potential and natural abundance, received extensive attention and become a strong competitor for the next generation of lithium-ion battery anode materials (Chou et al., 2009; Sahoo and Ramaprabhu, 2018; Woo et al., 2018; Ye et al., 2019). Besides, owing to the conversion reaction (SnO<sup>2</sup> + 4Li<sup>+</sup> + 4e<sup>−</sup> → Sn + 2Li2O) and alloying reaction (Sn + 4.4Li<sup>+</sup> + 4.4e<sup>−</sup> ↔ Li4.4Sn), SnO<sup>2</sup> anodes possess a high theoretical capacity (1,494 mAh g−<sup>1</sup> ) for lithium storage (Hu et al., 2017; Shao et al., 2017; Wang et al., 2018).

However, two main problems that delay the commercialization process of SnO<sup>2</sup> anodes are poor conductivity and severe volume expansion. Researchers have done a lot of research work to solve these problems, and the most effective strategy at present is to combine with carbon materials (Du et al., 2014; Chen et al., 2018d; Hou et al., 2018; Li et al., 2018), which is indeed a great strategy. The introduction of carbon can limit the size of the material to obtain nanomaterials (Wang et al., 2013; Chen et al., 2018c). Nanoparticles can effectively reduce the absolute volume change of each particle, greatly improving the structural stability of the material. Furthermore, the charge-diffusion path of ions and electrons are greatly shortened and a large number of electrochemically active sites existed in nanoparticles (Ying and Han, 2017). On the other hand, the introduction of carbon can greatly improve the electrical conductivity of the material and enhance the structural stability of the material. Compared with other carbon materials, graphene has a two-dimensional layer structure with a single atomic thickness, remarkable structural flexibility, excellent electrical conductivity, high mechanical strength, and a large specific surface area. Therefore, graphene has obvious advantages in the application of electrode materials, and is an ideal substrate for dispersing and limiting active substances (Chen et al., 2018a; Li et al., 2019).

Recently, some researchers propose that the charge transfer kinetics can be enhanced by constructing a reasonable heterogeneous structure with built-in driving force (Jiang et al., 2017; Li et al., 2018; Ren et al., 2018). The synergistic effect of ion/electron transport originating from solidsolid heterojunctions is particularly important for electrode performance of essentially enhanced kinetics. In this context, superior electrochemical performance was demonstrated for SnS/SnO<sup>2</sup> (Li et al., 2018), MoS2/SnS<sup>2</sup> (Jiang et al., 2017), MnMoO4/CoMoO<sup>4</sup> (Chen H. et al., 2014), and Fe2O3/Mn2O<sup>3</sup> (Ren et al., 2018). SnS<sup>2</sup> is a CdI2 type layered structure material with a large interlayer spacing (0.5899 nm) and a narrow band gap (Xu et al., 2015). Moreover, since the bond energy of Sn-S is weaker than the bond energy of Sn-O, this will promote the conversion of SnS<sup>2</sup> to Sn (Zhang S. et al., 2017; Wu et al., 2019). Therefore, the conversion reaction kinetics can be greatly improved by constructing a heterostructure between SnO<sup>2</sup> and SnS2. Whereas, the development of synergistic ultra-fine heterogeneous interface lithium storage materials still faces enormous challenges.

Herein, we propose a facile and scalable fabrication approach of heterostructured SnO2-SnS2@carbon/nitrogen-doped reduced graphene oxide (SnO2-SnS2@C/NG) composite. In this hierarchical structure, the as-prepared SnO2-SnS<sup>2</sup> nanoparticles were intimately embedded in carbon matrix with nitrogen-doped graphene, constructing an intimate cross-link conductive framework. Apart from the heterojunction structure of SnO2-SnS2, the carbon matrix with nitrogen-doped graphene accommodates the volume variation and facilitates the transport of lithium ions during continuous lithation/delithation cycling. The SnO2-SnS2@C/NG composite with hierarchical structure can fundamentally possess enhanced kinetics and reaction reversibility, thereby ensuring the stable cycling life.

# EXPERIMENTAL SECTION Preparation of SnO2-SnS2@C/NG Composite

All chemical reagents in this experiment are of analytical grade. The preparation method of SnO2@C/reduced graphene oxide (SnO2@C/rGO) refers to our previous report (Li et al., 2019). Firstly, 30 mL of graphene oxide (GO) solution (0.5 mg mL−<sup>1</sup> ) was added to the beaker, and then 0.25 g of SnCl<sup>2</sup> and 0.15 g of sodium alginate were thoroughly dispersed in the above GO solution, and slowly added to 30 mL of deionized water to form uniform solution. Secondly, the resulting solution was then hydrothermally treated at 180◦C for 12 h. After that, the obtained gray powder was calcined at 450◦C for 1 h under Ar atmosphere to obtain SnO2@C/rGO composite. Finally, 0.2 g of the obtained precursor and 0.4 g of thioacetamide were added to 50 ml of deionized water, and hydrothermally treated at 200◦C for 15 h to obtain the desired SnO2-SnS2@C/NG composite. SnO2- SnS2@C was prepared using the above method with 0.3 g of sodium alginate, and without using GO solution. SnO2@C was prepared using the above method with 0.3 g of sodium alginate, and without using GO solution and no further vulcanization.

### Characterization

The structure of SnO2-SnS<sup>2</sup> composite was determined by X-ray diffraction (XRD, Bruker D8) with Cu Kα radiation. The degree of the disorder of carbon was detected by Raman spectroscopy (HORIBA, Jobin-Yvon Lab RAM Aramis). The form of various elements in the SnO2-SnS2@C/NG was characterized by X-ray photoelectron spectroscopy (XPS, K-Alpha). The macroscopic morphology of the composite was observed by scanning electron microscopy (SEM, Zeiss Gemini DSM 982) and the microscopic morphology of the composite was observed by transmission electron microscopy (TEM, FEI Tecnai, G2 F20 STWIN). Thermogravimetric analyses (TGA, NETZSCH STA 409PG/PC) of the composites were performed in air with a heating rate of 5 ◦C min−<sup>1</sup> .

The electrochemical properties of the composites were evaluated using CR2025 cells. The composite material, conductive carbon (SP) and polyvinylidene fluoride (weight ratio 7:2:1) were dispersed in N-methylpyrrolidinone, and uniformly mixed to prepare a slurry. The slurry was uniformly coated on a copper foil current collector and dried at 120◦C overnight, and the dried electrode sheets were controlled to have a loading density of 1.1–1.5 mg cm−<sup>2</sup> . The CR2025 batteries were assembled in a closed glove box, in which lithium tablets were used as the counter electrodes, and 1 M LiPF<sup>6</sup> in ethylene carbonate and dimethyl carbonate (1:1, vol) was used as the electrolyte. The galvanostatic discharge/charging performances were tested by the battery test system (CT2001A, LAND) and the measured potential range was from 0.01 to 3.0 V. Cyclic voltammetry curves (CV, 0.01–3.0 V) and electrochemical impedance spectroscopy (EIS, 0.01–10<sup>5</sup> Hz) were recorded by electrochemical workstation measurements (CHI1000C, CH Instruments).

### RESULTS AND DISCUSSION

The representative fabrication process of heterostructured SnO2- SnS2@C/NG is presented in **Scheme 1**. Firstly, the SnO2@C/rGO composite was synthesized by hydrothermal method, using sodium alginate and GO to produce carbon matrix (C/rGO). Then, SnO<sup>2</sup> was partially vulcanized and grapheme was doped by nitrogen under hydrothermal reaction to finally obtain SnO2- SnS2@C/NG composite with a heterostructure structure, in which the thioacetamide works as a vulcanizing agent and a nitrogen dopant.

X-ray diffraction (XRD) patterns were collected to reveal the phase structures of the as-prepared SnO2-SnS2@C/NG and SnO2-SnS2@C composites. As exhibited in **Figure 1A**, the sharp diffraction peaks for both samples were consistent with SnO<sup>2</sup> (PDF No#21-1250) and SnS<sup>2</sup> (PDF#22-0951), respectively. The strong diffraction peaks represent a high degree of crystallinity of the composite. The results indicate that the composites consisted of SnO<sup>2</sup> and SnS<sup>2</sup> mixtures. Additionally, no obvious peaks indexed into carbon were observed, suggesting the amorphous state of carbon and graphene. SnO2@C sample (**Figure S1A**) was consistent with SnO<sup>2</sup> (PDF No#88-0287). To demonstrate the existence of graphene in composites, Raman spectroscopy at the range of 800–3.200 cm−<sup>1</sup> were collected in **Figure 1B**. It shows two obvious peaks at 1,337 and 1,583 cm−<sup>1</sup> , assigning to the disorder of carbon materials for D-band and characteristic of sp<sup>2</sup> hybridized carbon for G-band (Liu et al., 2010; Zhang et al., 2016). Compared with the SnO2-SnS2@C and SnO2@C (**Figure S1B**) (ID/I<sup>G</sup> = 0.72 and 0.71), the ID/I<sup>G</sup> ratio in SnO2-SnS2@C/NG (1.20) was much higher, suggesting that the existence of graphene, which is beneficial for the rapid electron transport. In addition, it can be found in the spectra that the SnO2-SnS2@C/NG composite had two distinct peaks at 2670.3 and 2916.1 cm−<sup>1</sup> due to the 2D and S3 bands, respectively (Liu et al., 2010). Notably, the symmetric 2D band indicates the presence of a single layer of graphene in the SnO2-SnS2@C/NG composite (Shah et al., 2017). To determine the carbon and SnS<sup>2</sup> content, TGA was estimated in air, as displayed in **Figure 1C**. As can be seen from the figure, the weight loss can be divided into three stages, which are divided into the following: the first stage is room temperature to 250◦C, the second stage is 250 to 450◦C, and the third stage is 450 to 600◦C. The weight loss in the first stage is due to the release of adsorbed water. The weight loss in the second stage corresponds to the conversion of SnS<sup>2</sup> to SnO<sup>2</sup> (SnS2+3O<sup>2</sup> = SnO2+SO2↑) (Lu et al., 2018), meanwhile, SnO2- SnS2@C and SnO2-SnS2@C/NG composites lost weights of 8.04 and 11.05 wt%, respectively. It can be calculated that the contents of the SnS<sup>2</sup> in SnO2-SnS2@C and SnO2-SnS2@C/NG composites material are 45.75 and 62.88 wt%, respectively, and the content of SnO<sup>2</sup> produced from those are 37.71 and 51.83 wt%, respectively. The weight loss in the third stage is ascribed to the oxidation of carbon, resulting in weight losses of 5.85 and 6.83 wt% for the SnO2-SnS2@C and SnO2-SnS2@C/NG composites, respectively. Until the last stage, the remaining products were all SnO2, and the SnO2-SnS2@C and SnO2-SnS2@C/NG composites retain 82.98 and 79.39 wt% (This contains two parts of SnO2: one part is the original SnO<sup>2</sup> in the composites and the other part is the SnO<sup>2</sup> converted from SnS2), respectively. Furthermore, it is inferred the specific composition of each component in the composite, as listed in **Table 1**. Surface chemical elements of SnO2-SnS2@C and SnO2-SnS2@C/NG composites were characterized by XPS analyses (**Figure 1D**). The S, O, Sn, and C elements were found in both composites in the full spectrum of XPS. Compared with SnO2-SnS2@C, it can be observed that SnO2-SnS2@C/NG had a

TABLE 1 | Contents of each component in the composites.


weak intensity peak of N 1s at 400 cm−<sup>1</sup> , suggesting that the N element has been doped into the graphene lattice.

To obtain more information about the chemical state of the elements in the prepared SnO2-SnS2@C/NG composite, Highresolution XPS spectra are depicted in **Figure 2**. As shown in **Figure 2A**, the high-resolution Sn 3d XPS spectrum of the SnO2-SnS2@C/NG composite can be fitted to the intensity peaks at 487.2 and 495.6 eV, respectively, confirmed that tin in the composite exists in the form of Sn4<sup>+</sup> (Lu et al., 2018). The C 1s spectrum is applied to further analyze the carbon form in the SnO2-SnS2@C/NG composite. As observed in **Figure 2B**, the high-resolution C 1s XPS spectrum of SnO2- SnS2@C/NG can be fitted to three peaks located at 284.7, 285.5, and 286.9 eV for C=C, C-N, and C-O, respectively (Li et al., 2018). The relative weak intensity of the C-O peak indicates that the oxide in the graphene is partially reduced. Besides, the presence of C-N peak further demonstrates nitrogen doping into the graphene lattice, which opens the band gap, adjusts the conductivity type, changes the electronic structure, and increases the free carrier density in the graphene, thereby improving the conductivity of SnO2-SnS2@C/NG electrodes. By fitting the high-resolution N 1s spectrum (**Figure 2C**), it can be found that N 1s spectrum can be divided into three peaks located in at 400.0, 401.8, and 402.7 eV, corresponding to the pyridinic, pyrrolic, and graphitic nitrogen atoms, respectively (Wu et al., 2018). **Figure 2D** shows the O 1s spectrum of the SnO2-SnS2@C/NG composite, which can be fitted into four types of oxygen-containing bond units located at 530.9, 531.2, 532.5, and 534.9 eV, respectively. The peaks at 530.9, 531.2 and 532.5 eV correspond to O-Sn4+, H-O-H, and C-O-C bonds, respectively. Notably, the peak at 534.9 eV corresponds to the S-Sn4+-O bond, which is a strong evidence of the presence of a heterojunction in the composite (Zhang et al., 2018). This confirmed the presence of a chemical bond between SnO<sup>2</sup> and SnS<sup>2</sup> particles.

The morphology and detailed structural feature of the prepared composite were observed by SEM, TEM and HRTEM. As can be observed in **Figure 3a**, the morphology of the SnO2- SnS2@C/NG composite shows an orderly lamination structure, forming a fluffy gap between the layers, which is very beneficial to the infiltration of the electrolyte. The ultrafine SnO2-SnS<sup>2</sup> nanoparticles embedded into the crumpled carbon matrix, thereby forming tremendous voids in the composite. Due to the uniform distribution of nanosized particles, which are hardly observed from the SEM image (**Figure 3b**). In the absence of graphene-involved combination, the particles of the SnO2- SnS2@C composite were agglomerated into microspheres of 200–300µm diameter (**Figure S2A**). TEM is also applied to characterize the detailed structural feature of SnO2-SnS2@C/NG composite. As presented in **Figure 3c**, it can be clearly observed that the nanoparticles are perfectly wrapped by a single layer of graphene sheet. The graphene sheet in the figure also has void spaces and wrinkles, which can significantly suppress

the volume expansion of the nanoparticles and enhance the conductivity of the composites. Under the observation of TEM, it is found that the microspheres of SnO2-SnS2@C composite (**Figure S2B**) are composed of primary particles of about 100–500 nm, and the surface of the primary particles is covered with an amorphous carbon. **Figure 3d** clearly depicts the nanoparticles with the diameter of 5–10 nm and high degree of crystallization, which are uniformly embedded in graphene with a curved lattice fringe of 0.424 nm. The particle in nanometer size can significantly shorten the lithium diffusion path and greatly improve the rate performance of the composite. Meanwhile, **Figure 3d** also displays the obvious several small nanocrystals with lattice spacings of 0.26, 0.34 and 0.32 nm, corresponding to (101) and (110) planes of SnO<sup>2</sup> and (100) plane of SnS2, respectively (Shi and Lu, 2014; Chen et al., 2015; Shan et al., 2015; Yin et al., 2017). More importantly, as denoted in **Figure 3d**, the heterojunctions are formed in the overlapped junction regions. In **Figure S2C**, the heterojunction phenomenon of SnO2-SnS2@C can also be found. The formation of heterojunctions composed of (101) plane of SnO2, (100) plane of SnS2, and (110) plane of SnO2, which strongly confirms the existence of heterogeneous SnO2-SnS2. Such a hierarchical structure is imperative for circumventing the selfaggregation of Sn-based nanoparticles, as well as the Sn and/or LixSn nanoparticles produced during cycling. Additionally, EDS mapping illustrates the homogeneous distribution of C, N, O, S, and Sn elements on the SnO2-SnS2@C/NG composite, as presented in **Figure 4**, which is consistent with the TEM analysis. The characterization of various physical and chemical means suggests that SnO2-SnS2@C/NG composite is a well-designed hierarchical structure, in which heterojunctions composed of SnO<sup>2</sup> and SnS<sup>2</sup> ultrafine nanoparticles are encapsulated in nitrogen-doped graphene sheets.

In order to explore the mechanism reaction of the composite during the charging and discharging process, the CV curve in the first three cycles of SnO2-SnS2@C/NG electrode was studied, as shown in **Figure 5A**. During the first scan of the cathodic process, the peaks at 1.87 and 1.70 V are attributed to the Li deintercalation in the SnS<sup>2</sup> layer, and there was no phase decomposition in this process, so the corresponding peaks were not found in the subsequent cycle (Yin et al., 2017). Decomposition of SnS<sup>2</sup> to metallic tin and Li2S results in a peak at 1.29 V. The conversion of SnO<sup>2</sup> and SnO to Sn produced a cathode peak at around 0.92 V and formed a synchronous products of Li2O and SEI film (Chen et al., 2015). The strong cathode peak at 0.18 V is attributed to LixSn produced by the alloying reaction of elemental tin and lithium ions, and peak at the same potential was also observed in subsequent cycles, indicating that the alloying reaction here is reversible. In the anodic scan process, the strong peak at 0.5 V is caused by the de-alloying of LixSn, corresponding to the 0.18 V peak in the cathodic scan. The peak at 1.86 V is attributed to the oxidation and sulfurization of Sn to SnO<sup>2</sup> and SnS2, respectively. The peak at 2.35 V is related to partial decomposition of Li2S, which also produces the first small irreversible capacity (Lu et al., 2018). It is worth noting that due to phase decomposition and structural collapse, the cathodic peak of 1.29 V in the first cycle shifted to 1.36 V in the subsequent cycle (Lu et al., 2018). Meanwhile, **Figure 5B** shows the first three discharge/charge curves of the SnO2-SnS2@C/NG composite, with a voltage window of 0.01–3.0 V and a current rate of 0.1 A g−<sup>1</sup> , which is consistent with the CV results. The first discharge/charge capacity achieved 1727.2 and 1201.2 mAh g−<sup>1</sup> , respectively, with initial coulombic efficiency of 69.54%. The excess capacity may come from the formation of SEI film on the surface of active materials, lithium-insertion reaction in acetylene black and interfacial storage, which is similar to the phenomena of other Sn-based anodes (Chen Y. et al., 2014; Fu et al., 2018; Li et al., 2019). The low initial coulombic efficiency is mainly due to the formation of SEI films, which is usually observed for the nanosized anode materials (Liu et al., 2014; Yang et al., 2018), whereas the coulombic efficiency approached 99.9% in the following cycles. In contrast, the initial charge capacity for SnO2-SnS2@C reached 812.4 mAh g−<sup>1</sup> under the

performances for SnO2-SnS2@C and SnO2-SnS2@C/NG composites at current rate of 0.5 A g−<sup>1</sup> , and (E) long cycling property for SnO2-SnS2@C/NG composite at current rate of 1 A g−<sup>1</sup> .

same condition (**Figure S3**), which is relatively lower than that of SnO2-SnS2@C/NG.

As depicted in **Figure 5C**, the rate capabilities of the composite electrodes were performed at various current rates ranging from 0.1 to 2 A g−<sup>1</sup> . The SnO2-SnS2@C electrodes delivered stable capacities of 815.7, 628.6, 544.5, 478.6, and 395.8 mAh g −1 at the rates of 0.1, 0.2, 0.5, 1, and 2 A g−<sup>1</sup> , respectively. When current rate returned to 0.1 A g−<sup>1</sup> , it recovered to 475.7 mAh g−<sup>1</sup> . In contrast, the rate performance of the SnO2@C electrode was relatively poorer, as its capacity dropped sharply when the current density increased until the cell failed. The possible reason is that the heterojunction in the SnO2- SnS2@C composite can significantly improve the conductivity, resulting in an increase in rate performance. Apparently, the rate performance of SnO2-SnS2@C/NG composite was the best. The SnO2-SnS2@C/NG electrodes delivered stable capacities of 1206.7, 936.3, 777.3, 666.3, and 559.0 mAh g−<sup>1</sup> at the rates of 0.1, 0.2, 0.5, 1, and 2 A g−<sup>1</sup> , respectively. When the current rate returns to 0.1 A g−<sup>1</sup> , it still can recover to 724.1 mAh g −1 . The SnO2-SnS2@C/NG composite also exhibited the best cycle performance. The capacity of SnO2-SnS2@C/NG composite electrode at current rate of 0.5 A g−<sup>1</sup> reached 717.6 mAh g−<sup>1</sup> after 200 cycles and gradually increased to 1039.4 mAh g−<sup>1</sup> after 400 cycles (**Figure 5D**). In addition, the charge and discharge curves for the 400th cycle of the SnO2-SnS2@C/NG composite are shown in the **Figure S4**. The capacity ratio in the voltage of 2.0–3.0 V was about 42%, which may be related to the reversible formation of the polymer gel-like film. However, the SnO2-SnS2@C electrode exhibited a low capacity of 208.0 mA h g−<sup>1</sup> at 0.5 A g−<sup>1</sup> after 200 cycles. Remarkably, at the high current rate of 1 A g−<sup>1</sup> (**Figure 5E**), the SnO2-SnS2@C/NG electrode still performed well after 950 cycles, exhibiting a strong reversible capacity of 944.3 mAh g−<sup>1</sup> , with the stabilized coulombic efficiency of around 99.5% throughout the whole cycling, confirming the merits of SnO2-SnS2@C/NG composite during the long-term cycling. In contrast with previous reported anodes materials, the SnO2-SnS2@C/NG composite exhibits outstanding cycle stability at higher current rates and

exhibits outstanding electrochemical performance (as shown in **Table S1**). It is noteworthy that the capacity had a slightly decrease during the first 50 cycles, then gradually increased during the subsequent cycles, and finally maintaining a stable performance. This phenomenon is attributed to the electrode activation and reversible formation of the polymer gel-like film, followed by the continuous reconstruction of SEI film on the surface of nanomaterials (Wang et al., 2013; Zhou et al., 2013; Liu et al., 2014).

In order to gain a deep understanding of the excellent electrochemical properties and to evaluate the lithium storage kinetics and charge transfer capacity of the composites, the SnO2- SnS2@C and SnO2-SnS2@C/NG electrodes were investigated by EIS. As displayed in **Figure 6A**, all Nyquist plots consist of a depressed semicircle at high frequency and a straight drift at low-medium frequency, which can be perfect fitted with an equivalent circuit as displayed in the insert of **Figure 6A**. Compared with charge-transfer resistance (Rct) of SnO2- SnS2@C (323.6 ), the Rct of SnO2-SnS2@C/NG (47.37 ) is much smaller. Furthermore, the lithium-ion diffusion coefficient (DLi+) can be further calculated according to the following equations (Yi et al., 2017; Zhang B. et al., 2017; Tong et al., 2018) (R, gas constant; T, absolute temperature; A, surface area; n, average reacted electrons; F, Faraday constant; C, Li<sup>+</sup> concentration, respectively):

$$\mathbf{D}\_{\rm Li}^{+} = \mathbf{R}^{2} \mathbf{T}^{2} / 2 \mathbf{A}^{2} \mathbf{n}^{4} \mathbf{F}^{4} \mathbf{C}^{2} \sigma^{2} \tag{1}$$

$$Z' = \mathbf{R}\_s + \mathbf{R}\_{\rm ct} + \sigma \,\omega^{1/2} \tag{2}$$

in which, σ can be calculated by the plot of Z' vs. ω −0.5 (angular frequency) as depicted in **Figure 6B**. Therefore, the calculated values of D<sup>+</sup> Li listed in **Table S2** imply the rapid diffusion of lithium ions and excellent reaction kinetics of the SnO2-SnS2@C/NG electrode.

In order to further explore the excellent rate performance of the SnO2-SnS2@C/NG electrode, CV analyses at various scan rates from 0.2 to 1 mV s−<sup>1</sup> were carried out (**Figure 6C**). Considering that SnO2-SnS2@C/NG is superior than SnO2- SnS2@C in high-rate performance and fast lithium ion transport, the pseudocapacitance contribution of SnO2- SnS2@C/NG may be more than that of SnO2-SnS2@C during charge/discharge process. In this case, the ratios of capacitive contribution can be further quantitatively quantified by separating the current response (i) at a fixed potential (V) into capacitive contribution (k1v) and

diffusion-controlled contribution (k2v 1/2 ), according to the following equation:

$$i = k\_1 \nu + k\_2 \nu^{1/2} = a \nu^b \tag{3}$$

in which, a and b present adjustable values. It must be noted that there is no proportional relationship between the measured current (i) and the square root of the scan rate (v) at a certain potential, suggesting the redox process can hardly be assigned to entire ion-diffusion control. The value of b is usually between 0.5 and 1, where 0.5 represents the diffusion control process and 1.0 illustrates the ideal pseudocapacitive behavior according to previous mechanism (Zhao et al., 2017). As displayed in **Figure 6D**, by fitting log(i) and log(v) at various redox potentials, several b values >0.8 can be obtained therefrom, demonstrating a dominated pseudocapacitive-controlled contribution process for the SnO2- SnS2@C/NG electrode. Furthermore, the contribution of total pseudocapacitive at a specific scan rate can be accurately calculated by the following equation ((Zhao et al., 2017); Zhang et al., 2018):

$$i(V)/\nu^{1/2} = k\_1 \nu^{1/2} + k\_2 \tag{4}$$

By carefully comparing the capacity contributions at different scan rates (**Figure 6E**), it is not difficult to conclude that the percentage of pseudocapacitive contribution increases gradually as the scan rate increases, and the lithiation/delithiation reaction of the SnO2-SnS2@C/NG composite gradually becomes capacitive control, and the pseudocapacitivecontrolled contribution ratios of the SnO2-SnS2@C/NG composite are always higher than those of the SnO2-SnS2@C composite. As highlighted in **Figure 6F**, when the scan rate is 1 mV s−<sup>1</sup> , nearly 58.02% of the total current (red region) is attributed to the pseudocapacitive contribution for the SnO2-SnS2@C/NG electrode, which is the maximum at all scan rates.

The distinct pseudocapacitive behavior and enhanced reaction kinetic proved by EIS results are originated from the synergistic effect of SnO2-SnS<sup>2</sup> heterojunction incorporated into the highly conductive N-doped graphene. The ultrafine SnO2- SnS<sup>2</sup> nanoparticles can provide shorten transport distance of lithium ions, especially the formation of heterojunctions at the interface of SnO2-SnS<sup>2</sup> nanoparticles effectively enhance the electronic transmission, which greatly promote the rate capability. Furthermore, nitrogen-doped graphene also plays a key role in lithium ion and electronic conductivity, which can build highly-conductive framework among the dispersive nanoparticles with integrity architecture, efficiently facilitating the lithium transport path and thus resulting in the excellent electrochemical performance.

The nitrogen-doped graphene nanosheets interconnect the nanoparticles and act as cushions for its volume expansion during long-term lithiation/delithiation process, further enhancing the structural stability and improving the life-cycle of entire electrode, as schematically illustrated in **Figure 7a**. The correlations between cycling properties and structural features for the as-prepared SnO2-SnS2@C/NG and SnO2- SnS2@C composites were characterized by TEM tests after 400th cycle at the current rate of 0.5 A g−<sup>1</sup> . As presented in **Figure 7b**, the flake morphology of SnO2-SnS2@C/NG was well-maintained after long-term cycling. The repeated volume expansion is effectively restricted by the robust architecture of the carbon matrix, in which the amorphous carbon locks the SnO2-SnS<sup>2</sup> nanoparticles on the nitrogen-doped graphene sheets, inhibiting its severe volume variation. In comparison, SnO2-SnS2@C composite without incorporation of graphene suffer from the particle pulverization and agglomeration after continuous cycling (**Figure 7c**), and the nanoparticles are easily detached from the current collector, which leads to the unstable structure and severe electrode degradation with bad cycling performance. Furthermore, the repeated formation of thick SEI films overlying on the surface SnO2-SnS<sup>2</sup> particles has adverse effects on the reversibility of conversion reaction, directly impeding the lithium ion and electron transfer and resulting in the continuous capacity decrease. All results indicate that heterostructured SnO2-SnS2@C/NG composite demonstrates enhanced structural stability and superior electrochemical performance.

## CONCLUSIONS

In this work, the well-designed SnO2-SnS2@C/NG composite was ingeniously fabricated by a facile approach, in which the heterojunction of SnO2-SnS<sup>2</sup> was created and embedded into the carbon matrix with highly conductive nitrogendoped graphene nanosheets. As expected, SnO2-SnS2@C/NG composite exhibited a high reversible capacity (1201.2 mA h g −1 at current rate of 0.1 A g−<sup>1</sup> ), superior rate capability and long-life stability (1039.4 mAh g−<sup>1</sup> after 400 cycles at current rate of 0.5 A g−<sup>1</sup> , and 944.3 mAh g−<sup>1</sup> after 950 cycles at current rate of 1.0 A g−<sup>1</sup> ). The excellent electrochemical properties may be ascribed to the following reasons. The lithium ion diffusion path is shortened in SnO2-SnS<sup>2</sup> nanoparticles with ultrafine size, and reversibility and reaction kinetics are enhanced through the introduction of heterostructure. Meanwhile, the nitrogen-doped graphene sheets improve the electronic conductivity and structural stability, which can effectively accommodate the volume variation and maintain the steady formation of SEI films without particle pulverization. This strategy contributes a new insight for preparing the high-performance electrode materials for LIBs.

### 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

HL carried out the experiment and wrote the manuscript. XW, JZ, and TA participated in the experiment. BZ, ZD and WY 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, 51772334, and 51702367), China Postdoctoral Science Foundation (Grant No. 2016M592447), The International Postdoctoral

### REFERENCES


Exchange Fellowship Program (Grant No. 155212), and Hunan Provincial Innovation Foundation for Postgraduate (CX2017B060).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2019.00339/full#supplementary-material

and sodium-ion batteries. ACS Appl. Mater. Interfaces 9, 27697–27706. doi: 10.1021/acsami.7b06572


**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 Li, Zhang, Wang, Zhang, An, Ding, Yu 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.

# Spray-Drying Synthesis of LiFeBO3/C Hollow Spheres With Improved Electrochemical and Storage Performances for Li-Ion Batteries

Yulei Sui <sup>1</sup> , Wei Chen<sup>1</sup> , Shibao Tang<sup>1</sup> , Ling Wu<sup>1</sup> \*, Binjue Wang<sup>1</sup> , Huacheng Li <sup>2</sup> , Wei Li <sup>3</sup> and Shengkui Zhong<sup>1</sup> \*

*<sup>1</sup> School of Iron and Steel, Soochow University, Suzhou, China, <sup>2</sup> Citic Dameng Mining Industries Limited, Chongzuo, China, <sup>3</sup> Guangxi Key Laboratory of Electrochemical and Magnetochemical Functional Materials, Guilin University of Technology, Guili, China*

### Edited by:

*Juchen Guo, University of California, Riverside, United States*

### Reviewed by:

*Qiaobao Zhang, Xiamen University, China Biao Gao, Wuhan University of Science and Technology, China*

### \*Correspondence:

*Ling Wu lwu@suda.edu.cn Shengkui Zhong zhongshengkui@suda.edu.cn*

### Specialty section:

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

Received: *29 March 2019* Accepted: *09 May 2019* Published: *28 May 2019*

### Citation:

*Sui Y, Chen W, Tang S, Wu L, Wang B, Li H, Li W and Zhong S (2019) Spray-Drying Synthesis of LiFeBO3/C Hollow Spheres With Improved Electrochemical and Storage Performances for Li-Ion Batteries. Front. Chem. 7:379. doi: 10.3389/fchem.2019.00379* LiFeBO3/C cathode material with hollow sphere architecture is successfully synthesized by a spray-drying method. SEM and TEM results demonstrate that the micro-sized LiFeBO3/C hollow spheres consist of LiFeBO3@C particles and the average size of LiFeBO3@C particles is around 50–100 nm. The thickness of the amorphous carbon layer which is coated on the surface of LiFeBO<sup>3</sup> nanoparticles is about 2.5 nm. LiFeBO3@C particles are connected by carbon layers and formed conductive network in the LiFeBO3/C hollow spheres, leading to improved electrical conductivity. Meanwhile, the hollow structure boosts the Li<sup>+</sup> diffusion and the carbon layers of LiFeBO3@C particles protect LiFeBO<sup>3</sup> from moisture corrosion. Consequently, synthesized LiFeBO3/C sample exhibits good electrochemical properties and storage performance.

Keywords: Li-ion batteries, LiFeBO3 , hollow sphere, cathode materials, spray drying

# INTRODUCTION

With the development of electric vehicles in the twenty-first century, Li-ion batteries (LIBs) play an increasingly important role in modern society (Nayak et al., 2018; Wu et al., 2019) and various cathode materials are developed (Ma et al., 2015; Zheng et al., 2018a,b; Yang et al., 2019). Recently, borate-based materials (LiMBO3, M = Fe, Mn, and Co) have received wide attention in the field of LIBs (Lin et al., 2012; Tao et al., 2013). Compared with the commonly used phosphate-based materials (LiMPO4, M = Fe, Mn, and Co), borate groups (LiMBO3, M = Fe, Mn, and Co) with a low specific weight show higher capacities. For instance, the theoretical capacity of LiMnBO<sup>3</sup> is 222 mAh g−<sup>1</sup> , whereas the phosphate counterpart, LiMnPO4, is limited to 170 mAh g−<sup>1</sup> (Moskon et al., 2016; Zhang et al., 2017). Moreover, the volume changes of borate-based materials during charging-discharging process are less than those of phosphate-based materials so that they usually show good structural stability (Loftager et al., 2017). Among various borate-based cathode materials, LiFeBO<sup>3</sup> with the theoretical capacity of 220 mAh g−<sup>1</sup> presents moderate working voltage, relatively high intrinsic conductivity, and good structural stability (Michalski et al., 2017). Therefore, LiFeBO<sup>3</sup> is considered as a potential cathode material for LIBs. However, there is only one-dimensional (1D) pathway for Li<sup>+</sup> diffusing in the LiFeBO<sup>3</sup> crystal, which leading to low electronic and ionic conductivity (Dong et al., 2008; Kalantarian et al., 2013). Furthermore, LiFeBO<sup>3</sup> is highly sensitive to moisture so that surface poisoning by H2O molecules of air would seriously disrupt its electrochemical properties (Gu et al., 2017). These problems make it challenging to achieve LiFeBO3-based materials with high reversible specific capacity and good storage performance. In order to improve the electrochemical performance of LiFeBO3 based materials, various strategies have been introduced, such as metal-ions doping, conductive carbon coating, and nanoarchitecturing (Aravindan and Umadevi, 2012; Afyon et al., 2013; Le Roux et al., 2015; Sin et al., 2015). Among these methods, carbon coating is a promising and effective way to improve

TABLE 1 | Results of structural analysis obtained from X-ray Rietveld refinement of LiFeBO3/C.


the properties of LiFeBO3, which attributes to the improved conductivity and the prevention of moisture corrosion. For example, Li et al. reported that the carbon coated LiFeBO<sup>3</sup> nanoparticles exhibit the improved first discharge capacities of 190.4 and 106.6 mAh g−<sup>1</sup> at 0.1 and 1◦C rate, respectively (Li et al., 2015). The LiFeBO3/C prepared by Zhang et al. also indicated carbon coating is an effective way to improve the properties of LiFeBO3. However, the rate capability and storage performance of LiFeBO<sup>3</sup> are still not satisfied (Zhang et al., 2014). Therefore, achieving homogeneous carbon coated LiFeBO<sup>3</sup> cathode materials with high electrochemical properties and good storage performance is still a challenge right now.

Herein, LiFeBO3/C with hollow porous sphere architecture is successfully synthesized by spray-drying method at the temperature as low as 450◦C (Li et al., 2015). Amorphous carbon layer which was produced by polyethylene glycol 6,000 (PEG-6000) decomposition is coated on the surface of LiFeBO<sup>3</sup> nanoparticles. The LiFeBO3@C particles are connected together and formed LiFeBO3/C hollow porous spheres. The obtained LiFeBO3/C sample with distinctive architecture shows the improved electrochemical and storage performances as cathode for LIBs.

### EXPERIMENTAL

### Synthesis of LiFeBO3/C and LiFeBO<sup>3</sup>

Firstly, LiNO3, 1.38g LiNO3, 8.08g Fe(NO3)3·9H2O, and 1.24g H3BO<sup>3</sup> were weighed and dissolved gradually in deionized water under stirring, and the mixed solution was named solution A. Meanwhile, 1.22 g PEG-6000 was dropped and dissolved in deionized water under stirring to form solution B. Then the solution B was added into the solution A and stirred carefully. A homogeneous sol was obtained when the mixed solution has been stirred at 80◦C for 1 h. The precursor of LiFeBO3/C was synthesized using the obtained sol as raw material via a spray drying process. The inlet and outlet air temperatures are 200 and

100◦C during the spray drying process with the air pressure of 0.25 MPa. The as-obtained precursor was calcined at 350◦C (3 h), followed by crystallized at 450◦C for 10 h in an argon atmosphere, and the final LiFeBO3/C was obtained. For contrast, the LiFeBO<sup>3</sup> sample was also prepared by the similar synthetic process but without PEG-6000. All the chemical regents employed in this work were of analytic grade.

### Characterization

The structure of LiFeBO3/C and LiFeBO<sup>3</sup> samples was investigated by XRD (Rigaku/Ultima-IV) and Raman microspectroscopy. And the range of XRD and Raman spectroscopic analysis is 2θ = 10 ∼ 80◦ and 600–2,000 cm−<sup>1</sup> , respectively. Morphology and microstructure of LiFeBO3/C and LiFeBO<sup>3</sup> samples were investigated by SEM (JSM/6380LV) and TEM (TecnaiG220). C-S analysis (Eltar) was used to determine the carbon content of samples.

### Electrochemical Measurements

The electrochemical measurements were performed in cointype cells (CR2025) with lithium metal as the negative electrode. The positive electrodes were prepared by mixing asprepared LiFeBO3/C or LiFeBO<sup>3</sup> powder, acetylene black (99.9%, Sinopharm Chemial Reagent Co., Ltd.) and polyvinylidene fluoride (PVDF, 99.9%, Sinopharm Chemial Reagent Co., Ltd. the weight ratio is 80:10:10) in N-methylpyrolline onto an Al foil and dried at 120◦C for 4 h in vacuum oven. Then the cointype cells were assembled in a glove box filled with high purity argon. 1M LiPF<sup>6</sup> solution in a mixture of ethylene carbonate and dimethyl carbonate with 1:1 volumetric ratio was used as electrolyte (Sinopharm Chemial Reagent Co., Ltd.). The cells

FIGURE 5 | (A) Cyclic voltammetry profiles of LiFeBO3/C and LiFeBO<sup>3</sup> at the scan rate of 0.1 mV s−<sup>1</sup> ; (B) Electrochemical impedance spectra of as-prepared LiFeBO3/C and LiFeBO3.

were tested in the voltage range of 1.5–4.5 V under various charge/discharge current densities from 5 to 100 mA g−<sup>1</sup> at room temperature. The electrochemical impedance spectroscopy (EIS) and the cyclic voltammetry (CV) was both measured by a CHI 660D workstation.

### RESULTS AND DISCUSSION

XRD patterns of the synthesized LiFeBO3/C and LiFeBO<sup>3</sup> samples are shown in **Figure 1A**. Obviously, the LiFeBO<sup>3</sup> with monoclinic structure is well-crystallized. XRD patterns of two samples are similar, indicating the combination of LiFeBO<sup>3</sup> and carbon has little influence on the crystal structure. Although the amount of carbon in the LiFeBO3/C composites is 4.9 wt. % (proved by C-S analysis), the diffraction peak of carbon is not detected from the XRD patterns, implying the carbon in the LiFeBO3/C composite is amorphous.

In order to further clarify the crystal structure and lattice parameters of the synthesized LiFeBO3/C, XRD data was refined by Rietveld method (Rietveld, 1967) using Maud software (Lutterotti, 2011), and the Rietveld refinement XRD pattern and atoms positions and occupancy for Li and Fe atoms in LiFeBO3/C investigated. As can be seen from **Figure 1B** and **Table 1**, the sharp peaks of the sample can be identified as monoclinic LiFeBO<sup>3</sup> with space group of C2/c and the structure is crystallized well. Meanwhile, the observed and calculated patterns match well, and the reliability factors are good. In accordance with the refinement results, the lattice parameters of LiFeBO<sup>3</sup> are a = 5.1662 Å, b = 8.9141 Å, c = 10.1700 Å, β = 91.25◦ , agreeing well with the reported literature (Zhang et al., 2014).

**Figure 2** represents the Raman spectroscopy of the asprepared LiFeBO3/C sample. As can be seen from **Figure 2**, the broad peak observed at 1,354 cm−<sup>1</sup> correspond to the D (disordered) band of carbon while the broad peak located at 1,590 cm−<sup>1</sup> agreeing well with G (graphitized) band of sp<sup>2</sup> type carbon (Liu et al., 2015; Zhu et al., 2017). The strong D-peak further reveals that the carbon is amorphous, which is in agreement with the XRD result.

SEM and TEM were employed to analyze morphology and micro-structure of the synthesized LiFeBO<sup>3</sup> and LiFeBO3/C. As shown in **Figure 3a**, the LiFeBO<sup>3</sup> sample displays a spherical shape with a size distribution in the range of 1–4 µm and the surface is compact. While, the LiFeBO3/C sample exhibits hollow spherical shape with a rough and porous surface layer (**Figures 3b,c**). The comparison of the darker outer shell and the lighter central regions indicates that the LiFeBO3/C spheres present porous and hollow internal structure (**Figure 3d**), which


is consistent with the SEM results. As shown in **Figure 3c**, it is obvious that the spherical LiFeBO3/C are actually consist of numerous primary nanoparticles. **Figure 3e** shows that nanosized primary LiFeBO<sup>3</sup> particles are well wrapped by thick carbon layer. This carbon-coated layer is benefit to prevent LiFeBO<sup>3</sup> from moisture corrosion. As we can see from **Figure 3f**, the lattice fringe with an interior planar distance of 0.4472 nm is correspond well to (1 1 0) crystal planes of LiFeBO3. The coating layer with the thickness of 2.5 nm could be assigned to the amorphous carbon, so that LiFeBO3@C shows a core-shell structure. In addition, the carbon-coated layer and conductive carbon network can improve the conductivity of LiFeBO3-based cathode material, which is also benefit to the electrochemical performances.

In the potential (vs. Li/Li+) range of 1.5–4.5 V, the cells were charged/discharged at 10 mA g−<sup>1</sup> , and the curves of the assynthesized LiFeBO3/C and LiFeBO<sup>3</sup> samples are exhibited in **Figures 4A,B**, respectively. It is noted that the initial charge curves of both samples show obviously higher voltage plateau than the subsequent cycles, which mainly due to the high polarization before cycling. This phenomenon is very similar to cases of LiMnBO<sup>3</sup> and LiCoBO<sup>3</sup> (Afyon et al., 2014; Tang et al., 2015). The charge plateau (∼3.2 V) and the discharge plateau (∼2.8 V) in the voltage profiles correspond to the plateaus of LiFeBO3. There is a reduced thermodynamic potential for the degraded LiFeBO<sup>3</sup> phase so that a short discharge plateau ∼1.8 V appeared in the discharge profiles (Bo et al., 2012).

The cycling performances of the as-prepared LiFeBO3/C and LiFeBO<sup>3</sup> samples at 10 mA g−<sup>1</sup> are shown in **Figure 4C**. The initial discharge specific capacities of LiFeBO3/C and LiFeBO<sup>3</sup> are 212.4 and 189.5 mAh g−<sup>1</sup> , respectively. The specific capacity of LiFeBO3/C is as high as 184.1 mAh g−<sup>1</sup> while the specific capacity of LiFeBO<sup>3</sup> is only 132.9 mAh g−<sup>1</sup> after 50 cycles, indicating the LiFeBO3/C sample has a better cyclic stability. **Figure 4D** displays the rate capability of the as-prepared LiFeBO3/C and LiFeBO<sup>3</sup> samples under different current densities. The initial capacity of the as-prepared LiFeBO3/C is 219.2 mAh g−<sup>1</sup> at 5 mA g−<sup>1</sup> , which is approximately 98.7% of the theoretical capacity. When the current density is increased up to 40 and 100 mA g−<sup>1</sup> , the LiFeBO3/C sample holds a stable reversible capacity of 187.3 and 175.8 mAh g−<sup>1</sup> , respectively. However, the discharge capacity of LiFeBO<sup>3</sup> at 40 and 100 mA g −1 are only 144.5 and 115.5 mAh g−<sup>1</sup> , respectively. If the current density returns back to 5 mA g−<sup>1</sup> after 75 cycles test, and 98.6% of the initial discharge specific capacity (216.3 mAh g−<sup>1</sup> ) can be recovered for LiFeBO3/C but only 75.1% was recovered for LiFeBO<sup>3</sup> electrode (149.6 mAh g−<sup>1</sup> ), implying the LiFeBO3/C sample has a better capacity recovery ability than LiFeBO3. The improved electrochemical performances can be attributed to the special hollow sphere structure and network of amorphous carbon layer. The conductivity is improved by the network of amorphous carbon layer and the volume expansion during the charge and discharge progresses is againsted due to the special hollow sphere architecture provides a flexible structure, so that the electrochemical performance of the synthesized cathode is improved (Li et al., 2013).

**Figure 5A** shows the cyclic voltammetry (CV) of the as-prepared LiFeBO3/C and LiFeBO<sup>3</sup> samples. The oxidation/reduction peaks located at ∼2.95/2.35 V correspond to the phase transition between LiFeBO<sup>3</sup> and FeBO3. It is clearly observed that LiFeBO3/C sample shows sharper peaks and larger areas than LiFeBO3, implying the synthesized LiFeBO3/C has a better reversibility. In addition, a couple of oxidation/reduction peaks are located at ∼2.09/1.85 V, and these patterns can be ascribed to the degradation of LiFeBO<sup>3</sup> (Afyon et al., 2014).

In order to analyze the conductivity of the as-prepared LiFeBO3/C and LiFeBO<sup>3</sup> samples, the EIS tests (**Figure 5B**) are measured in our study. Then the EIS data is fitted with Z-view software using an equivalent circuit, and the related parameters are listed in **Table 2**. R<sup>s</sup> , Zw, and Rct in the equivalent circuit (insert **Figure 5B**) represent the resistance of the electrolyte, Warburg impedance and charge transfer resistance, respectively. The Rct of LiFeBO3/C (238.6 ) is lower than that of LiFeBO<sup>3</sup> (354.1 ), implying the charge transfer speed during the charge and discharge processes of electrode is significantly improved. The exchange current density (j0) of synthesized LiFeBO3/C and LiFeBO<sup>3</sup> is calculated by the following equation, respectively.

$$j\_0 = \frac{RT}{nFR\_{ct}}$$

Where T and n refer to the temperature and the electrons number, respectively. The exchange current density (j0) of LiFeBO3/C (10.78 × 10−<sup>4</sup> mA cm−<sup>2</sup> ) is higher than that of LiFeBO<sup>3</sup> (7.26 × 10−<sup>5</sup> mA cm−<sup>2</sup> ), which indicates that LiFeBO3/C has a better electrode reaction reversibility than LiFeBO3.

To evaluate the storage stability, LiFeBO3/C and LiFeBO<sup>3</sup> samples were stored in air exposure for 6 months. XRD patterns of the LiFeBO3/C and LiFeBO<sup>3</sup> samples before and after storage

are shown in **Figure 6**. Gratifyingly, the stored LiFeBO3/C sample also remains its original crystal structure, verifying high storage stability of LiFeBO3/C, as shown in **Figure 6A**. However, the XRD pattern of LiFeBO<sup>3</sup> changed after storing 6 months. As can be seen from **Figure 6B**, the intensity of (1 1 2) peak tends to increase while the intensity of (1 1 2) peak tends to decrease, and the two peaks prone to overlap each other. This phenomenon can be attributed to phase changing of LiFeBO<sup>3</sup> from monoclinic to orthorhombic structure with air corrosion (Bo et al., 2012; Chen et al., 2015).

After 6 months of storage, the cycling performances of LiFeBO3/C and LiFeBO<sup>3</sup> samples are investigated, and the results are shown in **Figure 7**. The capacity of the stored LiFeBO3/C is as high as 198.9 mAh g−<sup>1</sup> and there is 76.9% capacity retention after 50 cycles. However, the stored LiFeBO<sup>3</sup> delivers a capacity of 191.7 mAh g−<sup>1</sup> and only 56.1% capacity retention after 50 cycles. Obviously, the stored LiFeBO3/C shows higher stability and cycling performance than the stored LiFeBO3. In summary, carbon coating and structural modification improve the conductivity of LiFeBO3, shorten the Li<sup>+</sup> diffusion/conduction path and protect LiFeBO<sup>3</sup> from moisture corrosion, thus leading to good electrochemical performance.

### CONCLUSIONS

LiFeBO3/C hollow sphere is successfully synthesized by a facile spray-drying method. Primary LiFeBO3@C particles are connected by carbon layers and formed LiFeBO3/C hollow spheres with improved electrical conductivity. Meanwhile, the hollow porous structure boosts the Li<sup>+</sup> diffusion and the carbon layers of LiFeBO3@C particles protect LiFeBO<sup>3</sup> from moisture corrosion. Therefore, the synthesized LiFeBO3/C sample shows good electrochemical performances and improved storage performance. This work is conducive to obtaining promising and high-performance cathode materials for LIBs.

### REFERENCES


### 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

YS, ST, and WC did the main experiment and write the manuscript. LW and BW envolved the discussion of the experiment and revised the manuscript. HL and WL assisted the material synthesis. LW and SZ made the research plan. SZ and LW also provided the financial support.

### FUNDING

This study was supported by the National Natural Science Foundation of China (51574168, 51574170, and 51774207), Guangxi Key Laboratory of Electrochemical and Magnetochemical Functional Materials (EMFM20182202), and Major Projects for Science & Technology Development of Guangxi Province, China (AA16380043).

for lithium-ion batteries. J. Nanosci. Nanotechno. 15, 7186–7190. doi: 10.1166/jnn.2015.10555


rate capability and long cycling stability for lithium-ion batteries. Nano Energy 53, 613–621. doi: 10.1016/j.nanoen.2018.09.014


**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 Sui, Chen, Tang, Wu, Wang, Li, Li 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 Li4Ti5O12@CN Composite With Enhanced Rate Properties

### Hui Xiao<sup>1</sup> , Xiaobing Huang<sup>2</sup> \*, Yurong Ren<sup>1</sup> \*, Xiang Ding<sup>2</sup> and Shibiao Zhou<sup>2</sup>

*<sup>1</sup> School of Materials Science and Engineering, Jiangsu Collaborative Innovation Center of Photovolatic Science and Engineering, Changzhou University, Changzhou, China, <sup>2</sup> Hunan Province Cooperative Innovation Center for The Construction and Development of Dongting Lake Ecological Economic Zone, College of Chemistry and Materials Engineering, Hunan University of Arts and Science, Changde, China*

Folic acid is first time applied as a carbon-nitrogen precursor to fabricate Li4Ti5O12@CN composites via ball milling Nano-TiO2, Li2CO<sup>3</sup> and folic acid with ethanol as solvent, and then followed by heating treatment in argon. XRD, SEM, TEM, XPS, charge-discharge test and EIS are used to evaluate the influence of N-doped carbon coating on its structure, morphologies and electrochemical property. It is demonstrated that the N-doped carbon coated Li4Ti5O<sup>12</sup> composite exhibits superior high-rate performance compared with pure Li4Ti5O12. It possesses a high discharge capacity of 174, 165 mAh g <sup>−</sup><sup>1</sup> at 0.5 and 10 C, respectively. Additionally, an initial specific capacity of 96.2% is obtained after 200 cycles at 10 C. The remarkable performance might be put down to the N-doped carbon layer providing efficiently electron conductive network and nanosized decreasing lithium ion diffusion path.

### Edited by:

*Junchao Zheng, Central South University, China*

### Reviewed by:

*Zhian Zhang, Central South University, China Ling Wu, Soochow University, China*

### \*Correspondence:

*Xiaobing Huang hxb220170@126.com Yurong Ren ryrchem@163.com*

### Specialty section:

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

Received: *20 March 2019* Accepted: *27 May 2019* Published: *14 June 2019*

### Citation:

*Xiao H, Huang X, Ren Y, Ding X and Zhou S (2019) Fabrication of Li4Ti5O12@CN Composite With Enhanced Rate Properties. Front. Chem. 7:432. doi: 10.3389/fchem.2019.00432*

Keywords: Li-ion batteries, anode material, folic acid, Li4Ti5O12, N-doped carbon coating

# INTRODUCTION

In the past few years, carbon materials have been regarded as the most commercially valued lithium battery anode material (Ma et al., 2016; Long et al., 2017; Li et al., 2019; Wu et al., 2019). Unfortunately, its low lithiation potential (∼0.2 V vs. Li+/Li) will result in the activation of dendritic lithium, thus further creating safety problems. Moreover, insufficient diffusion coefficient of lithium ion and large volume change will result in poor rate performance and cycle stability (Han et al., 2018; Yi et al., 2018). Therefore, it is essential to find alternative anodes with excellent electrochemical properties and outstanding safety characteristics. Among the reported alternative candidates, Spinel lithium titanium (Li4Ti5O12) is suggested as a potential anode material used in lithium-ion batteries (LIBs) due to the following reasons: (i) The higher flat discharge and charge plateau (1.55 V vs. Li+/Li) can prevent lithium metal dendrites from evolving during the electrochemical reaction process (Wang S. et al., 2017). (ii) Small volume expansion provides superior cycle stability and reversibility among charge and discharge process (Tian et al., 2017; Wang Q. et al., 2017). Nevertheless, its small lithium ion diffusion coefficients (10−<sup>9</sup> ∼10−<sup>13</sup> cm<sup>2</sup> S −1 ) and weak electrical conductivity (∼10−<sup>13</sup> S cm−<sup>1</sup> ) (Jiang et al., 2017), resulting in serious electrode polarization and poor capability at high current density, has been considered as the main bottleneck of its commercial application for high-power LIBs. In order to overcover this problem, several strategies have been suggested by researchers, including synthesis of Li4Ti5O<sup>12</sup> with porous structure (Lu et al., 2017), coating the surface of Li4Ti5O<sup>12</sup> particles with conductive materials (Tang et al., 2017), construction of nanoscale particle size (Chiu et al., 2017), and introduction of metal and nonmetal ion into Li4Ti5O<sup>12</sup> (Chen et al., 2017; Cheng et al., 2017; Liang et al., 2017).

Thin carbon coating on the surface of Li4Ti5O<sup>12</sup> particles has been considered as an effective method to enhance its electrochemical properties in all reported literatures, since the carbon coating can both improve the surface electron conductivity and inhibit the growth of primary particles in the heat treatment process, this further leads to faster lithium ion diffusion. Very recently, the introduction of nitrogen-doped carbon to modify the electrode materials has been received more and more attention. It is well-known that N atoms can provide additional electrons to further increase the conductivity of the coated carbon layer (Jiang et al., 2017; Xu et al., 2017).In addition, N-doped is favorable to reduce the barrier of energy of lithium-ion penetration and enhance reaction sites (Xiong et al., 2018). We have just proposed folic acid as a new carbon nitrogen precursor to prepare Na3V2(PO4)3@CN composite material, it possesses excellent rate performance and excellent cycling performance when used as cathode electrode of sodium ion batteries. In this work, a similar strategy was suggested to prepared Li4Ti5O12@CN composite. It is well-expected that the as-prepared Li4Ti5O12@CN sample will possess excellent electrochemical characters.

## EXPERIMENTAL

### Material Preparation

Li4Ti5O12@CN composite was synthesized via a solid-state process, and the detail process was described as following. Firstly, 2.2281 g of Li2CO<sup>3</sup> (99.5%),5.9985 g of TiO<sup>2</sup> (99.9%) and 1 g of folic acid were mixed by ball milling in ethyl alcohol for 6 h. Secondly, the solvent was evaporated by drying the readyprepared mixture at 80◦C for 2 h. Thirdly, the synthesized precursor was heated under argon atmosphere at 750◦C for 8 h. For comparison, pure Li4Ti5O<sup>12</sup> was also synthesized via the similar method without using folic acid as the starting material.

### Characterization of Materials

X-ray diffraction instrument was used to investigate the structure and composition of both as-prepared Li4Ti5O<sup>12</sup> samples. The surface chemical states of Li4Ti5O12@CN were identified by XPS. SEM was used for observing morphologies of the two ready-prepared Li4Ti5O<sup>12</sup> samples. N-doped carbon layer was further investigated by TEM. Elemental analyzer was carried out to investigate the content of carbon and nitrogen for Li4Ti5O12@CN composite. Four-point probe method was used to investigate the electronic conductivities of both Li4Ti5O<sup>12</sup> samples.

### Electrochemical Measurements

The fabricated working electrodes consisted of as-synthesized Li4Ti5O<sup>12</sup> sample, LA-132 and Super-P in a weight ratio of 85:5:10. The mixture was uniformly casted onto the aluminum foil and then dried in vacuum. CR2032 coin type cells were prepared in glove box with filled argon by composing of lithium piece as the counter electrode, 1 mol/L LiPF<sup>6</sup> in EC/DEC/DMC (1:1:1 in volume) as the electrolyte. Celgard 2400 as the separator,. LAND CT2001 system were used to investigate the chargedischarge experiments between the potential range of 1–3V. Electrochemical impedance spectra (EIS) were investigated by using CHI600E electrochemical station in a frequency range of 10−<sup>2</sup> − 10<sup>5</sup> Hz.

### RESULTS AND DISCUSSION

Crystal structure of both as-obtained Li4Ti5O<sup>12</sup> samples was characterized via XRD, and the results are plotted in **Figure 1**. It clearly verifies that the dominating diffraction peaks of both assynthesized Li4Ti5O<sup>12</sup> samples are in consistence with the base peaks of spinel Li4Ti5O<sup>12</sup> (PDF No.49-0207), demonstrating that the nitrogen-doped carbon layer coating process has little effect on the formation of spinel Li4Ti5O<sup>12</sup> (Li et al., 2013a; Chang et al., 2014; Wang P. et al., 2017). Perhaps due to the amorphous morphology of carbon, the diffraction peak relative to carbon was not observed (Xu et al., 2017; Liu et al., 2018). The detail of the enlarged peak corresponding to Li4Ti5O<sup>12</sup> (111) plane was described in **Figure 1B**. As clearly found that the central position of this peak shifts to larger angles after N-doped carbon coating, suggesting that nitrogen atoms possibly doped into the Li4Ti5O<sup>12</sup> lattice and form a new thin layer of TiNx, similar phenomenon was also reported by Li et al. (Zhang et al., 2013). In addition, the intensity of the peaks in Li4Ti5O12@CN composite is lower than that of pure Li4Ti5O12, indicating that the Ndoped carbon coating on the surface of Li4Ti5O<sup>12</sup> will prevent the growth of particles. In order to investigate the content of carbon and nitrogen in Li4Ti5O12@CN composite, elemental analysis measurement is carried out. The content of carbon and nitrogen for Li4Ti5O12@CN composite is 1.46 and 0.24%, respectively.

The obtained results of surface chemical state of Li4Ti5O12@CN composite evaluated by X-ray photoelectron spectroscopy (XPS) are given in **Figure 2**. Li1s, Ti2p, O1s, C1s, and N1s peaks are observed from the XPS spectra of Li4Ti5O12@CN composite (**Figure 2A**). The high-resolution N1s of Li4Ti5O12@CN composite is demonstrated in **Figure 2B**. As illustrated in **Figure 2B**, pyridine (N1), pyrrole (N2) and graphitic (N3) correspond to peaks centered at 398.2, 399.8, and 401.4eV, respectively. (Li H. et al., 2014; Long et al., 2015; Wang et al., 2015). The above results clearly showed that nitrogen-doping is successfully introduced in the carbon layer by using folic acid as a carbon-nitrogen precursor, which might result in produce the flaws in the symmetric offset spread of aromatic rings carbon, and thus further increase the diffusion of Li<sup>+</sup> in the interface (Li et al., 2013a; Wang et al., 2015). Additionally, a peak at about 397 eV being attributed to the interaction energy of TiN is observed in **Figure 2B**, indicating that titanium nitride (TiN) phase is created during the sintering process (Wan et al., 2012). As well-accepted, the existence of TiN with a metallic conductivity will improve electronic conductivity (Li H. et al., 2014). The high-resolution Ti2p of Li4Ti5O12@CN composite is depicted in **Figure 2C**. Clearly, two peaks appeared at approximately 464.2 and 458.5eV are

FIGURE 1 | (A) X-ray diffraction profiles of Li4Ti5O12 and Li4Ti5O12@CN composite, (B) Enlarged (111) peak of Li4Ti5O12 and Li4Ti5O12@CN composite.

observed, which represents to peaks of Ti 2p1/2 and Ti 2p3/2 of Ti4<sup>+</sup> in the sample (Li et al., 2014). In addition, two extra peaks located at about 459 and 464.7eV are detected, corresponding to the peaks of Ti 2p1/2 and Ti 2p3/2 of Ti3<sup>+</sup> in the sample, respectively, which suggests that Ti3<sup>+</sup> sites were introduced in the Li4Ti5O12@CN composite due to the reduced ability of N-doped carbon. Similar phenomena were also demonstrated in previous reports (Wan et al., 2012; Ming et al., 2014). It further verified that the titanium nitride (TiN) phase was formed.

The different morphology and particle size between Li4Ti5O<sup>12</sup> and Li4Ti5O12@CN composites were investigated by SEM characterization. The results are shown in **Figure 3**. As clearly demonstrated in **Figure 3**, Li4Ti5O12@CN composite possesses a much smaller particle size than that of Li4Ti5O12. It is wellaccepted that in situ N-doped carbon coating will well-prohibit the Li4Ti5O<sup>12</sup> particles from growing into larger crystals (Wang C. et al., 2014). The SEM results agree well with the XRD results. To further verify this prediction, the specific surface area of both as-obtained Li4Ti5O<sup>12</sup> samples was investigated. Li4Ti5O12@CN composite and pure Li4Ti5O<sup>12</sup> possess a specific surface area of 12.75 and 7.08 m<sup>2</sup> g −1 , respectively. Generally, much smaller particles of Li4Ti5O12@CN composite gives the larger specific surface area. As well-accepted that fabrication of material with much smaller particle size and larger specific surface area will decrease the lithium ion diffusion pathways and thus enhance the kinetics of lithiation/delithiation (Long et al., 2015). The carbon-nitrogen layer of Li4Ti5O12@CN composite was further studied by TEM characterization and results are demonstrated in **Figure 4**. It was found that a carbon-nitrogen layer with a thickness of 2 to 5 nm is formed on the exterior of Li4Ti5O12. The electronic conductivities of the as-synthesized Li4Ti5O<sup>12</sup> samples were confirmed by four-point probe method. Pure Li4Ti5O<sup>12</sup> and Li4Ti5O12@CN composite have the electronic conductivities of 7.67 × 10−<sup>5</sup> and 1.06 × 10−<sup>2</sup> S cm−<sup>1</sup> , respectively.

The first charge-discharge cycle for as-synthesized Li4Ti5O<sup>12</sup> and Li4Ti5O12@CN composite at a rate of 0.5 C between a voltage range of 1–3V are described in **Figure 5**. Both asprepared Li4Ti5O<sup>12</sup> samples possess a voltage plateau at ∼1.55V, suggesting a two-phase reaction is carried out on the basis of the redox pair of Ti3+/Ti4<sup>+</sup> (Wang et al., 2012; Wang B. et al., 2014). However, the potential separation between the charge and discharge curves of Li4Ti5O12@CN composite material is much smaller than that of Li4Ti5O12, indicating that the Li4Ti5O12@CN electrode has less polarization and better reaction kinetics, which

demonstrates that the enhancement of electrical conductivity of Li4Ti5O<sup>12</sup> after introducing N-doped carbon coating (Li et al., 2013a; Zhang et al., 2014).

The rate capabilities of Li4Ti5O12@CN composite and Li4Ti5O<sup>12</sup> are shown in **Figure 6**. Pure Li4Ti5O<sup>12</sup> obtained a discharge capacity of 158 mAh g−<sup>1</sup> at a low rate of 0.5 C, and the capacity decreases remarkably as the rate increased from 0.5 to 1, 2, 5, and 10 C, respectively. Especially, its discharge capacity is only 79 mAh g−<sup>1</sup> at 10 C. The poor rate properties of pure Li4Ti5O<sup>12</sup> could be due to its poor conductivity. While much-improved discharge capacity at each rate for Li4Ti5O12@CN composite in comparison with pure Li4Ti5O12. At 0.5, 5, and 10 C, its discharge capacity were 174, 168, and 165 mAh g−<sup>1</sup> , respectively. The superior rate properties

of Li4Ti5O12@CN composite could be due to three reasons: (i) As demonstrated in SEM results (Seen in **Figure 4**), the smaller particle size for Li4Ti5O12@CN composite is favorable for the faster Li<sup>+</sup> diffusion and further enhancement of kinetic coefficient of lithium ion embedded into Li4Ti5O<sup>12</sup> structure (Long et al., 2015). (ii) The electronic conductivity of Li4Ti5O<sup>12</sup> is deemed to be enhanced by N-doped carbon coating (Xu et al., 2017), and the defects in the carbon layer caused by N-doping facilitate Li<sup>+</sup> migration in the interface (Li et al., 2013b). (iii) The electronic conductivity properties of Li4Ti5O<sup>12</sup> will be further increased with the existence of TiN<sup>x</sup> in the composite.

The cyclic property of Li4Ti5O12@CN composite at a rate of 10 C is depicted in **Figure 7**. As clearly seen, the initial discharge capacity of Li4Ti5O12@CN composite is 165 mAh g−<sup>1</sup> , and 96.2% of its capacity is obtained after 200 cycles, suggesting that the as-synthesized Li4Ti5O12@CN composite possesses good cycle stability.

To further investigate the effect of N-doped carbon coating on electrode behavior, electrochemical impedance spectroscopy measurements were performed on Li4Ti5O<sup>12</sup> and Li4Ti5O12@CN electrodes, and the results are plotted in **Figure 8**. Obviously, each electrode exhibits a similar profile with a straight line at the low frequency region and a depressed semicircle at the high-middle frequency range, being correlated with Liion diffusion and charges transfer resistance (Rct) in the electrode/electrolyte interface, respectively (Zheng et al., 2018; An et al., 2019; He et al., 2019; Lv et al., 2019; Xiao et al., 2019; Zhou et al., 2019). As obtained from **Figure 8**, Li4Ti5O12@CN electrode exhibits much smaller charge-transfer resistance of 52 than that of 194 for pure Li4Ti5O<sup>12</sup> electrode, suggesting the improved electronic conductivity of Li4Ti5O12@CN composite in comparison with pure Li4Ti5O<sup>12</sup> caused by the highly electronic conductive coating with Ndoped carbon as well as the presence of TiN<sup>x</sup> phase (Zhou et al., 2017). According to the following equation, the Li-ion

FIGURE 8 | Nyquist plots of Li4Ti5O12 and Li4Ti5O12@CN composite.

diffusion coefficient of both as-prepared Li4Ti5O<sup>12</sup> samples can be obtained:

$$\mathbf{D} = \mathbf{R}^2 \mathbf{I}^2 / 2\mathbf{A}^2 \mathbf{n}^4 \mathbf{F}^4 \mathbf{C}^2 \sigma^2 \tag{1}$$

$$\mathbf{Z\_{re}} = \mathbf{R\_D} + \mathbf{R\_L} + \sigma \alpha^{-1/2} \tag{2}$$

The value of σ for both samples could be obtained from the lines described in **Figure 9**. Based on the above equations and the results from **Figure 9**, the obtained lithium ion diffusion coefficient for Li4Ti5O<sup>12</sup> and Li4Ti5O12@CN composite is 6.58 × 10−<sup>11</sup> and 2.89 × 10−<sup>9</sup> cm<sup>2</sup> s −1 , respectively. There is no doubt that the Li4Ti5O12@CN electrode has a larger Li-ion diffusion coefficient due to the smaller particle size and the carbon layer defect of the Li4Ti5O12@CN composite, indicating that a valid enhancement of the diffusion kinetics of Li4Ti5O<sup>12</sup> after the introduction of the N-doped carbon coating. Based on the above results and discussion, coating with N-doped carbon can dramatically improve the lithium ion and conductive property. Consequently, Li4Ti5O12@CN composite has much improved electrochemical performance in compassion with pure Li4Ti5O12.

### CONCLUSIONS

In this study, N-doped carbon-coated Li4Ti5O<sup>12</sup> is prepared with folic acid as a carbon-nitrogen source. Li4Ti5O12@CN composite has the outstanding reversible capacity, high rate capability in compassion with pure Li4Ti5O12. The discharge capacity of the Li4Ti5O12@CN composite at 10 C was 165 mAh g−<sup>1</sup> , and the initial specific capacity remained at 96.2% after 200 cycles. The superior properties of Li4Ti5O12@CN composite could be owing to the improved electronic conductivity caused by the N-doped carbon layer and the TiN phase as well as enhanced Li-ion diffusion coefficient rising from the smaller particle size and the defects in the carbon layer.

### REFERENCES


### 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

XH and YR contributed conception and design of the study. HX carried out experiments and wrote the manuscript. XD performed analyzed experimental results. SZ revised the manuscript.

### FUNDING

This project was financially supported by the National Nature Science Foundation of China (Nos. 21771062, 21576030 and U1607127), the Research Fund of Hunan Provincial Education Department (18A366), the Opening Project of Material Corrosion and Protection Key Laboratory of Sichuan Province of China (No. 2018CL15).


performance in lithium-ion battery applications. Electrochim. Acta 116, 224–229. doi: 10.1016/j.electacta.2013.11.038


**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 Xiao, Huang, Ren, Ding and Zhou. 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.

# The Effects of Reversibility of H2-H3 Phase Transition on Ni-Rich Layered Oxide Cathode for High-Energy Lithium-Ion Batteries

Jie Chen1†, Huiping Yang1†, Tianhao Li <sup>1</sup> , Chaoyang Liu<sup>1</sup> , Hui Tong<sup>2</sup> , Jiaxin Chen<sup>1</sup> , Zengsheng Liu<sup>1</sup> , Lingfeng Xia<sup>1</sup> , Zhaoyong Chen<sup>1</sup> , Junfei Duan<sup>1</sup> and Lingjun Li 1,3 \*

*<sup>1</sup> School of Materials Science and Engineering, Changsha University of Science and Technology, Changsha, China, <sup>2</sup> School of Metallurgy and Environment, Central South University, Changsha, China, <sup>3</sup> Hunan Provincial Key Laboratory of Modeling and Monitoring on the Near-Earth Electromagnetic Environments, Changsha University of Science and Technology, Changsha, China*

### Edited by:

*Shengkui Zhong, Soochow University, China*

### Reviewed by:

*Yunjian Liu, Jiangsu University, China Haisheng Fang, Kunming University of Science and Technology, China*

> \*Correspondence: *Lingjun Li lingjun.li@csust.edu.cn*

*†These authors have contributed equally to this work as co-first authors*

### Specialty section:

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

Received: *13 June 2019* Accepted: *01 July 2019* Published: *16 July 2019*

### Citation:

*Chen J, Yang H, Li T, Liu C, Tong H, Chen J, Liu Z, Xia L, Chen Z, Duan J and Li L (2019) The Effects of Reversibility of H2-H3 Phase Transition on Ni-Rich Layered Oxide Cathode for High-Energy Lithium-Ion Batteries. Front. Chem. 7:500. doi: 10.3389/fchem.2019.00500* Although LiNi0.8Co0.1Mn0.1O<sup>2</sup> is attracting increasing attention on account of its high specific capacity, the moderate cycle lifetime still hinders its large-scale commercialization applications. Herein, the Ti-doped LiNi0.8Co0.1Mn0.1O<sup>2</sup> compounds are successfully synthesized. The Li(Ni0.8Co0.1Mn0.1)0.99Ti0.01O<sup>2</sup> sample exhibits the best electrochemical performance. Under the voltage range of 2.7–4.3 V, it maintains a reversible capacity of 151.01 mAh·g <sup>−</sup><sup>1</sup> with the capacity retention of 83.98% after 200 cycles at 1 C. Electrochemical impedance spectroscopy (EIS) and differential capacity profiles during prolonged cycling demonstrate that the Ti doping could enhance both the abilities of electronic transition and Li ion diffusion. More importantly, Ti doping can also improve the reversibility of the H2-H3 phase transitions during charge-discharge cycles, thus improving the electrochemical performance of Ni-rich cathodes.

Keywords: lithium ion batteries, cathode materials, LiNi0.8Co0.1Mn0.1O2 , Ti-doped, phase transitions

# INTRODUCTION

With the rapid development of renewable energy of wind and solar power, the large-scale energy storage system has become more and more important (Manthiram et al., 2016; Chen et al., 2018; Wu et al., 2018; Wu C. et al., 2019; Xia et al., 2019). Among them, lithium ion batteries have been powering our daily life from mobile phones to electric vehicles (EVs), due to their high energy density, long cycle life and environmental friendly (Chen et al., 2014; Li et al., 2018; Wu L. et al., 2019; Ye et al., 2019). With the rapid development of high-performance silicon-carbon composite anodes (Su et al., 2018; Xiao et al., 2018), cathode materials become the technological bottleneck for obtaining high performance lithium-ion batteries (Myung et al., 2016; Chen Z. et al., 2017; Xu et al., 2017; Zheng et al., 2019). Among the various types of layered cathodes, LiNi0.8Co0.1Mn0.1O<sup>2</sup> (NCM) is attracting an increasing amount of attention on account of its high specific capacity and low cost, compared to traditional LiCoO<sup>2</sup> material (Zhang et al., 2012; Li et al., 2016a; Chen M. et al., 2017; Yang et al., 2019).

However, the long-term capacity retention of NCM is not satisfactory for large-scale application, which can be ascribed to the undesired side reactions between the highly reactive species Ni4<sup>+</sup> and liquid electrolyte at the interface of the cathodes, triggering the irreversible phase transition from initial layered R-3m phase to spinel Fd-3m phase and further to the rock-salt phase (Nam et al., 2013; Lin et al., 2014a,b). Worse yet, the NiO passivation layer will increase the impedance, hindering the electrochemical kinetics, and resulting in the deterioration of electrochemical performance (Meng et al., 2017; Ryu et al., 2018a).

During the extraction of Lithium-ion, Ni-rich materials undergoes a series of phase transitions: The original layered structure (H1) transforms to the monoclinic phase (M), the second hexagonal phase (H2), and the third hexagonal phase (H3) (He et al., 2018; Gao et al., 2019; Wu et al., 2019a). It has been reported that the H2-H3 transition will cause detrimental lattice shrinkage along the c-direction, resulting in the volume change and the local stress accumulation, and further leading to the microcracks generation and propagation in secondary particles (Lee et al., 2014; Sun and Manthiram, 2017; Yoon et al., 2018). Ryu et al. (2018b) reported the cross-sectional SEM images of the charged LiNi0.90Co0.05Mn0.05O<sup>2</sup> particles, numerous cracks emanating from the particle core and some cracks traverse across the entire particle and nearly fracture the secondary particle. The produced cracks create fresh surface where phase transitions and corrosion as well as side reactions occur, thereby further accelerating the structural degradation of the cathodes.

Such severe structural collapse affected by mechanical strain associated with the poor irreversibility of the H2-H3 phase transition during cycling. Cations substitutions have been regarded as promising way to overcome these challenges and enhance the structural stability of Ni-rich materials (Xia et al., 2018; Li et al., 2019; Susai et al., 2019; Weigel et al., 2019). Wu et al. (2019b) demonstrated that by doping Ti4<sup>+</sup> in LiNi0.9Co0.1O<sup>2</sup> materials, the improved reversibility of the H2- H3 phase transitions and the lossless H3 phase suppress the generation of microcracks and structural degradations. It is also reported that, by doping with boron, the LiNi0.9Co0.05Mn0.05O<sup>2</sup> shows no visible cracks, which is consistent with its good reversibility of H2-H3, attesting to the beneficial effect of boron doping and enables deflection of the internal strain posed by the phase transition (Park et al., 2018).

In this work, the structural and electrochemical performances of Ti-doped LiNi0.8Co0.1Mn0.1O<sup>2</sup> (0, 0.5, 1, 2%) have been systematically studied. The Ti doping can keep the layered structure materials with one phase. Among them, the Li(Ni0.8Co0.1Mn0.1)0.99Ti0.01O<sup>2</sup> sample exhibits the best electrochemical performance. Within a voltage window of 2.7**–**4.3 V, the Li(Ni0.8Co0.1Mn0.1)0.99Ti0.01O<sup>2</sup> sample maintains a reversible capacity of 151.01 mAh·g <sup>−</sup><sup>1</sup> with 83.98% capacity retention after 200 cycles, corresponding 0.08% decay per cycle. The improved electrochemical performance can be ascribed to the enhanced abilities of electronic transition and Li ion diffusion. More importantly, the reversibility of H2-H3 phase transitions and the lossless H3 phase during prolonged cycling can also improve the electrochemical performance of Ni-rich cathodes.

### EXPERIMENTAL SECTION

### Materials Synthesis

The pristine LiNi0.8Co0.1Mn0.1O<sup>2</sup> (denoted as NCM) spherical material was synthesized by solid-state method. Stoichiometric ratio of LiOH·H2O and Ni0.8Co0.1Mn0.1(OH)<sup>2</sup> precursor were mixed and ground in an agate mortar. Then, under oxygen atmosphere, the mixtures were precalcined at 480◦C for 5 h and calcined at 830◦C for 12 h with a heating rate of 5◦C·min−<sup>1</sup> . Next, the sample was cooled to room temperature naturally in a tube furnace.

The 0.5, 1, 2 mol% Ti-doped NCM (denoted as Ti-0.5, Ti-1, Ti-2) samples were prepared by the following steps. Firstly, the stoichiometric amount of C16H36O4Ti was dissolved into 60 mL absolute ethanol, then stirring continuously in water bath at the temperature of 60◦C. Secondly, the LiOH·H2O were added to the solution and kept stirring for about 30 min, subsequently the Ni0.8Co0.1Mn0.1(OH)<sup>2</sup> precursor was added. Afterward, the mixed solution was dried overnight at 120◦C. Finally, the Tidoped samples were obtained by calcining under the same condition as NCM.

### Materials Characterization

The crystalline structures of all the samples were determined by X-ray diffraction (XRD, Bruker D8 advance) using Cu Kα radiation in the 2θ range of 10–90◦ with a scan rate of 5◦ ·min−<sup>1</sup> and operated at 40 kV and 40 mA. Scanning electron microscopy (SEM, Nova NanoSEM-230) was performed to observe the particle morphologies of samples. The microstructures of the NCM and Ti-1 samples were examined by high-resolution transmission electron microscopy (HRTEM, FEI Talso 200s).

### Electrochemical Measurements

The electrochemical properties were measured using a CR2025 coin-type half cell. The working electrodes were prepared by mixing the active materials, acetylene black, and polyvinylidenedifluoride (PVDF) with a weight ratio of 8:1:1 in N-methyl-2-pyrrolidone (NMP). The resulting slurry was cast uniformly onto aluminum foil, followed by drying at 120◦C in a vacuum oven for 4 h and pouched. The typical mass loading of positive electrode was about 2.3 mg·cm−<sup>2</sup> with an electrode diameter of 12 mm. 1 M LiPF<sup>6</sup> dissolved in a solution of dimethyl carbonate (DMC), ethylene carbonate (EC), and ethyl methyl carbonate (EMC) (volume ratio = 1:1:1) was used as the electrolyte. All the cells were assembled in an Ar-filled glovebox where the moisture and oxygen content is under 0.1 ppm by using a lithium metal anode and separator (Celgard 2500 polyethylene). Then, the cells were silenced before electrochemically tests. Electrochemical tests were performed on NEWARE BTS7.6 battery test system between 2.7 and 4.3 V vs. Li/Li<sup>+</sup> at room temperature (25◦C). The cycling performance was initially charged and discharged at 0.1 C (18 mA·g −1 ) with 2 cycles, then charged and discharged at 1 C in the subsequent cycles. The cyclic voltammetry (CV) was measured at a scan

TABLE 1 | Lattice parameters of all samples obtained by XRD analysis.


rate of 0.1 mV·s <sup>−</sup><sup>1</sup> within 2.7–4.5 V through an electrochemical workstation (Solartron 1470E). Electrochemical impedance spectroscopy (EIS) tests were conducted in the frequency range of 10−<sup>3</sup> -10<sup>5</sup> Hz.

# RESULTS AND DISCUSSION

The structure characteristics of the NCM and Ti-doped NCM samples are analyzed by XRD measurements as shown in **Figure 1A**. All the diffraction peaks of all samples can be well indexed to the hexagonal α-NaFeO<sup>2</sup> layered structure (PDF#70- 4314) with R-3m space group, and no secondary phase is observed, indicating that Ti4<sup>+</sup> is successfully incorporated into the bulk structure. Furthermore, the clear splitting of (006)/(102) and (108)/(110) couples for all samples manifest a well-developed layered structure. The partial magnified patterns in **Figure 1B** displays that the (003) diffraction peaks shift to a lower angle with increasing Ti-doped content, demonstrating lattice expansion caused by the successfully dopant of Ti4+. The result can also be proved by the increase of lattice parameter c and the cell volume as shown in **Table 1**.

The morphologies of the pristine NCM and Ti-1 samples are shown in **Figures 2a,d**. The pristine and modified samples images exhibit very similar morphology. It is found that the secondary particles are spherical with the size is ∼12µm in diameter, which are composed of densely packed primary particles. Apparently, the result indicates that Ti doping cannot change the particle morphology and particle size. To further investigate the effect of Ti doping on the microstructure of NCM sample, the HRTEM images of NCM (**Figures 2b,c**) and Ti-1 (**Figures 2e,f**) are performed. As shown, the rectangle region is selected as representative to observe the atomic-scale crystal structure, and the fast fourier transform (FFT) of the selected region are shown on the right. The FFT image of the NCM demonstrates that layered phase (space group R-3m) is successfully formed. For the Ti-1 sample, the diffraction spots of the corresponding FFT can be assigned to the (104) lattice plane, indicating that the microstructure cannot be changed by 1% Ti doping.

Electrochemical properties of the pristine and Ti-doped NCM samples are tested in lithium-ion half-cells at room temperature. **Figure 3A** shows the initial charge/discharge profiles of all

FIGURE 3 | Electrochemical performance of all electrodes in the voltage range of 2.7–4.3 V at room temperature: (A) Initial charge-discharge profiles at 0.1 C. (B) Rate capabilities at various current densities. (C) Cycling performance at a current rate of 1 C.

electrodes between 2.7 and 4.3 V at the rate of 0.1 C. It can be observed that all the discharge profiles are similar to each other. The initial discharge specific capacity is decreased slightly from 202.24 to 180.95 mAh·g <sup>−</sup><sup>1</sup> with the increasing of Ti amounts. However, the Ti-1 electrode shows the higher Coulombic efficiency of 83.20% than NCM electrode showing Coulombic efficiency of 80.82%, most probably owing to that suitable Ti doping can reduce the irreversible capacity.

The rate performances of the pristine and Ti-doped NCM electrodes at the current density ranging from 0.1 to 10 C are presented in **Figure 3B**. Comparing to the NCM electrode, the Ti-0.5 and Ti-1 electrodes show improved rate capability. Particularly, the Ti-1 electrode maintains a reversible capacity of 145.71 mAh·g −1 at high current rates of 10 C, while the NCM cathodes only deliver a specific capacity of 105.52 mAh·g −1 . The remarkable rate property is mainly related to the slight enlargement along c axis by Ti doping, which benefits the fast Li ion diffusion. In addition, the cycling performances of all electrodes are compared at 1 C rate in the range of 2.7–4.3 V. As can be seen in **Figure 3C**, the NCM electrode shows fast capacity degradation from 179.03 to 136.88 mAh·g −1 after 200 cycles, corresponding to 76.45% capacity retention with 0.118% decay per cycle. In comparison, the Ti-doped electrodes possess the higher capacity retention than pristine electrode, specifically, the Ti-1 electrode maintains 151.01 mAh·g <sup>−</sup><sup>1</sup> with 83.98% capacity retention after 200 cycles, corresponding 0.08% decay per cycle.

**Figure 4** shows the charge-discharge curves of NCM, Ti-0.5, Ti-1, Ti-2 electrodes at different cycle number, respectively. It is distinctly observed that the reversible capacity and discharge voltage plateau of the NCM electrode declined rapidly with increasing cycle, while the Ti-1 electrode shows stable chargedischarge curves and discharge capacity, indicating that suitable Ti doping can effectively improve the cycle stability of NCM and alleviate the voltage reduction during cycling.

To further understand the different charge-discharge behavior of the NCM and Ti-doped electrodes, the differential capacity (dQ/dV) curves are shown in **Figure 5**. All curves display three couples of redox peaks, which are ascribed to the phase transition of hexagonal (H1) to monoclinic (M), monoclinic (M) to hexagonal (H2), and hexagonal (H2) to hexagonal (H3) during the delithiation/lithiation processes. As shown in **Figure 5A**, the oxidation peaks for the pristine NCM shift to high potential

significantly, while the reduction peaks shift to low potential during cycling, and the corresponding peaks areas are greatly decreased. As shows, the poor reversibility of H2-H3 phase transition and gradually loss of H3 phase associated with poor structural stability of NCM, should take responsibility for the faster capacity drop of the NCM sample. In contrast, the voltage hysteresis for the Ti-doped electrodes is substantially suppressed. For Ti-1, the negligible change of overlap of dQ/dV curves within 200 cycles suggests the lower polarization and better H2-H3 reversibility of the modified electrodes. In conclusion, the Tidoped electrodes exhibit improved electrochemical properties, which can be attributed to the better structural stability due to the enhanced reversibility of H2-H3 phase transition.

To understand the relationship between electrochemical degradation and resistance parameters, EIS measurements are conducted for all samples in a charged state of 4.3 V after different cycle numbers. **Figures 6A–C** show the Nyquist plots of the electrodes and corresponding equivalent circuit at the fresh cells, after 5 and 200 cycles, respectively. The curves consist of a semicircle and a sloped line in **Figure 6A**. The semicircle is regarded as the charge transfer resistance (Rct) and the followed sloped line relates to the Warburg impedance (Zw). Moreover, the intercept with real axis (Z′ ) is assigned to the solution resistance (Rs). Differently, **Figures 6B,C** display an increased semicircle in the high-frequency region, which can be ascribed to the film resistance (R<sup>f</sup> ) due to the solid electrolyte interface (SEI) film formed on the electrode during high-voltage cycling. The fitting results of the electrodes based on the equivalent circuit are summarized in **Table 2**. The Rct of the NCM electrode in the fresh cell is 96.5 , and the resistance increases dramatically after 200 cycles (1372 ). In comparison, the Ti-doped electrodes exhibit more stable values of Rct during cycling, implying better structural stability. According to the relationships between Z′ re and ω −1/2 in the low-frequency region (**Figure 6D**), the lithium ion diffusion coefficient (DLi+) is calculated and the results are listed in **Table 2** (Li et al., 2015, 2016b; Liu et al., 2019; Zhu et al., 2019). Notably, the Ti-1 electrode shows highest DLi<sup>+</sup> values than that of the NCM electrode after 200 cycles, contributing to the best rate property among all electrodes. The EIS results demonstrate that the Ti-doped samples exhibit better

FIGURE 6 | Nyquist plots of (A) as-prepared electrodes before cycling, (B) all electrodes at the charge state of 4.3V after 5th, (C) all electrodes at the charge state of 4.3V after 200th (insets are the equivalent circuit models), (D) the relationships between Z′ re and ω <sup>−</sup>1/<sup>2</sup> based on the 200th cycles.

TABLE 2 | The values of R<sup>f</sup> , Rct, and D<sup>+</sup> Li for as-prepared electrodes and cycled electrode.


kinetic behavior than those of the NCM, which can be related to the outstanding structural stability by the modification of Ti doping.

To study the redox behavior associated with phase transition during battery operation, Cyclic voltammetry (CV) is employed at a scan rate of 0.1 mV·s <sup>−</sup><sup>1</sup> between 2.7 and 4.5 V. **Figures 7A–D** illustrate the first three CV curves of all samples, respectively. For these CV curves, three pairs of oxidation/reduction peaks are observed, which can be assigned to the phase transitions from H1 to M, H2 and H3. It is worth noting that the H2-H3 phase transition can induce sharp lattice contraction along c axis, result in the anisotropic changes of cell volume (Yang and Xia, 2016; Xu et al., 2019). Apparently, the peak intensity of the phase transition of H2-H3 gradually become weaker with increasing Ti concentration, implying that the Ti doping could suppress the detrimental phase transition during charging/discharging cycles. In other words, the structural stability of the NCM is improved by Ti doping.

# CONCLUSION

To address the problem related to the structural degradations of NCM, Ti is intentionally introduced to enhance the structural stability and electrochemical performance of LiNi0.8Co0.1Mn0.1O2. The Ti-doped LiNi0.8Co0.1Mn0.1O<sup>2</sup> (0, 0.5, 1, 2%) composites are successfully synthesized via one-step calcination method. Among them, the Li(Ni0.8Co0.1Mn0.1)0.99Ti0.01O<sup>2</sup> sample exhibits the best electrochemical performance, it maintains a reversible capacity of 151.01 mAh·g −1 after 200 cycles at 1 C with

83.98% capacity retention. The superior electrochemical performance can be ascribed to two aspects: (1) the enhanced reversibility of H2-H3 phase transitions and the lossless H3 phase during prolonged cycling; (2) the lower electrochemical impedance and the improved Li-ion diffusion ability.

### DATA AVAILABILITY

All datasets generated for this study are included in the manuscript/supplementary files.

### AUTHOR CONTRIBUTIONS

LL, JieC, and HY: conceived the idea. JieC, HY, and LL: prepared all materials. JieC, HY, and TL: conducted SEM experiments. CL,

### REFERENCES

Chen, J., Li, L., Wu, L., Yao, Q., Yang, H., Liu, Z., et al. (2018). Enhanced cycle stability of Na0.9Ni0.45Mn0.55O<sup>2</sup> through tailoring O3/P2 JiaC, ZL, and LX: conducted XRD experiments. LL, JieC, and HY: analyzed the data. JieC and HY: wrote the manuscript. HT, ZC, and JD: commented on the manuscript. LL: supervised the implementation of the project.

### FUNDING

This work was supported by the National Natural Science Foundation of China (Grant Nos. 51774051 and 51502350), this work was also supported by the Changsha City Fund for Distinguished and Innovative Young Scholars (Grant No. KQ1707014), the Hunan Provincial Natural Science Foundation of China (No. 2018JJ2428), the Scientific Research Fund of Hunan Provincial Education Department (No. 17B002), and the Key Laboratory of Modeling and Monitoring on the Near-Earth Electromagnetic Environments (No. 20170109).

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

Chen, M., Zhao, E., Chen, D., Wu, M., Han, S., Huang, Q., et al. (2017). Decreasing Li/Ni disorder and improving the electrochemical performances of Ni-Rich LiNi0.8Co0.1Mn0.1O<sup>2</sup> by Ca doping. Inorg. Chem. 56, 8355–8362. doi: 10.1021/acs.inorgchem.7b01035


**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, Yang, Li, Liu, Tong, Chen, Liu, Xia, Chen, 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.

# Concentrated LiODFB Electrolyte for Lithium Metal Batteries

Juan Yu<sup>1</sup> , Na Gao<sup>1</sup> , Jiaxin Peng<sup>1</sup> , Nani Ma<sup>1</sup> , Xiaoyan Liu<sup>2</sup> , Chao Shen<sup>2</sup> , Keyu Xie<sup>2</sup> and Zhao Fang<sup>1</sup> \*

<sup>1</sup> School of Metallurgical Engineering, Xi'an University of Architecture and Technology, Xi'an, China, <sup>2</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

Nowadays, lithium (Li) metal batteries arouse widespread concerns due to its ultrahigh specific capacity (3,860 mAh g−<sup>1</sup> ). However, the growth of Li dendrites has always limited their industrial development. In this paper, the use of concentrated electrolyte with lithium difluoro(oxalate)borate (LiODFB) salt in 1, 2-dimethoxyethane (DME) enables the good cycling of a Li metal anode at high Coulombic efficiency (up to 98.1%) without dendrite growth. Furthermore, a Li/Li cell can be cycled at 1 mA cm−<sup>2</sup> for over 3,000 h. Besides, compared to conventional LiPF6-carbonate electrolyte, Li/LiFePO<sup>4</sup> cells with 4 M LiODFB-DME exhibit superior electrochemical performances, especially at high temperature (65◦C). These outstanding performances can be certified to the increased availability of Li<sup>+</sup> concentration and the merits of LiODFB salt. We believe that the concentrated LiODFB electrolyte is help to enable practical applications for Li metal anode in rechargeable batteries.

### Edited by:

Junchao Zheng, Central South University, China

### Reviewed by:

Haisheng Fang, Kunming University of Science and Technology, China Chaoyi Chen, Guizhou University, China Ying Hong Yang, Northeastern University, China

> \*Correspondence: Zhao Fang fangzhao@xauat.edu.cn

### Specialty section:

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

Received: 21 April 2019 Accepted: 26 June 2019 Published: 18 July 2019

### Citation:

Yu J, Gao N, Peng J, Ma N, Liu X, Shen C, Xie K and Fang Z (2019) Concentrated LiODFB Electrolyte for Lithium Metal Batteries. Front. Chem. 7:494. doi: 10.3389/fchem.2019.00494 Keywords: Li-metal batteries, concentrated electrolyte, LiODFB, high temperature, dendrites free

### INTRODUCTION

In the past several decades, Li-ion batteries have played successful role in the consumable electronic device market (Etacheri et al., 2011; Goodenough and Kim, 2014). However, the limited specific capacity of the graphite anode limits its wider applications. The theoretical capacity of graphite is only 372 mAh g−<sup>1</sup> , so it is difficult to achieve the urgent need for high-energy density batteries to adapt to the electrical miniaturization trend (Placke et al., 2017). Based on this reason, the Li metal batteries (LMBs) have attracted wide attention because of its high theoretical specific capacity of 3,860 mAh g−<sup>1</sup> , small density of 0.534 g cm−<sup>3</sup> and the relatively negative electrochemical potential (−3.040 V vs. Li/Li+) (Zhamu et al., 2012). However, the shortcomings of LMBs are also a headache, such as unsatisfied Coulombic efficiency (CE) and dendritic Li growth. In addition, poor cycle performance caused by Li dendrites seriously affected the commercial development of LMBs.

Significant efforts have been made in the past to solve these problems, including prepare the current collector (Yang et al., 2015; Liang et al., 2016; Zhang et al., 2016; Shi et al., 2017), ex-situ protective coating of Li anode (Thompson et al., 2011; Kozen et al., 2015), introduce the battery intermediate protective layer (Liang et al., 2015; Wu et al., 2015; Cheng et al., 2016; Xie et al., 2016, 2017), and electrolyte modification. Electrolyte, which occupies 15% of the weight and about 32% of the volume of the entire battery, is a key part of a battery (Cekic-Laskovic et al., 2017). Therefore, the study of electrolyte on the development of Li metal secondary battery is of great significance. Meanwhile, compared to other modification methods, electrolyte modification is more convenient and cost-effective. Till now, electrolyte research progress more or less has been made to address the dendrite problem. Nevertheless, the compatibility of these electrolytes with cathode has usually been overlooked. For example, some film-forming electrolyte additives, such as vinylene carbonate (VC) and fluoroethylene carbonate (FEC) are helpful to change Li dendrite formation nature (Aurbach et al., 2003; Jung et al., 2013; Webb et al., 2014; Qian et al., 2016; Pritzl et al., 2017; Xu et al., 2017), but the cell's internal resistance is greatly increased, particularly after long-term cycling, and thus limits their wide applications.

Recently, the concept of high concentrated electrolyte gets into the eyes of people. Suo et al. reported a new class of nonaqueous liquid "Solvent-in-Salt" electrolytes and applied them in Li-S batteries. It is demonstrated that the use of "Solventin-Salt" electrolyte inhibits the dissolution of polysulphide and protects metallic Li anodes against the formation of Li dendrites, effectively (Suo et al., 2013). A significant breakthrough has been achieved by Qian et al. (2015). They found that, with 4 M lithium bis(fluorosulfonyl)imide (LiFSI) in DME as the electrolyte, a Li/Li cell can be cycled at for more than 6,000 cycles. Meanwhile, a Cu/Li cell can be cycled for more than 1,000 cycles with an average CE of 98.4%, without dendrite growth. Nevertheless, the compatibility of the electrolyte with the cathode materials has not been investigated.

Generally, LiFSI has inherent characteristics of corrosion of aluminum (Al) foil and other metal parts in the battery (Abouimrane et al., 2009; Li et al., 2011). To solve this problem, Park et al. found that lithium borate salts are the ideal additives as corrosion inhibitors in LiFSI electrolytes. The inhibition ability of Al is revealed to be in the following order: lithium oxalyldifluoroborate (LiODFB) > lithium tetrafluoroborate (LiBF4) > lithium hexafluorophosphate (LiPF6) > lithium bis(oxalato)borate (LiBOB) (Park et al., 2015). Noticed that LiODFB is considered to combine the half structures of LiBOB and LiBF4, and thus, it combines the advantages of LiBOB and LiBF<sup>4</sup> (Zhang, 2006). The main advantages of this salt is given by low viscosity, high ionic conductivity, good film-forming, high temperature performance, good compatibility with the positive electrode, passivation of Li foil and so on (Liu et al., 2007; Zugmann et al., 2011; Wu et al., 2012; Zhou et al., 2012).

However, it should be noted that previous studies (Zhang et al., 2010; Li et al., 2015; Zhou et al., 2016; Bian et al., 2017) only focused on the effect of LiODFB as an additive or auxiliary salt on battery performances. So far, high concentration of LiODFB single salt in ether solvent for LMBs has not been systematically reported before (Zhang et al., 2015; Poyraz et al., 2019; Yamada et al., 2019). Thus, in this work, we first employed the concentrated electrolyte based on LiODFB in LMBs. Due to its increased availability of Li<sup>+</sup> concentration, a highly uniform and stable solid electrolyte interface (SEI) film was formed on Li metal anode without any dendrite. Meanwhile, the compatibility of concentrated LiODFB electrolyte and LiFePO<sup>4</sup> cathode was first investigated by the electrochemical performance of Li/LiFePO<sup>4</sup> cell at room temperature and high temperature. We found that 4 M LiODFB-DME significantly improve electrochemical performance of LiFePO<sup>4</sup> cathode, compared to LiPF<sup>6</sup> electrolyte. In particular, at high temperature, the improvement is much greater. In a sense, our work shows concentrated LiODFB electrolyte may have great potential for LMBs.

# EXPERIMENTAL SECTION

# Electrolytes and Electrode Preparation

The LiPF<sup>6</sup> dissolved in ethylene carbonate (EC), ethyl methyl carbonate (EMC), dimethyl carbonate (DMC) with a volume ratio of 1:1:1 was formed 1.0 M LiPF<sup>6</sup> electrolyte and used as the blank electrolyte. LiODFB was purchased from Suzhou Fluolyte Co., Ltd., China. LiODFB was dissolved in dimethyl ether (DME) and formed 4 M LiODFB-DME in an Ar-filled glove box (H2O < 0.1 ppm, O<sup>2</sup> < 0.1 ppm). The electrolytes were stirred to maintain homogeneity, then sealed and stored in the glove box.

The LiFePO<sup>4</sup> cathodes were prepared by mixing 80 wt.% LiFePO<sup>4</sup> (Aladdin Co., Ltd. China), 10 wt.% conductive carbon, and 10 wt.% polyvinylidene (PVDF, Solef) binder in Nmethylpyrrolidone (Kelong, Chengdu). The mixed slurry was coated onto Al foil and then dried overnight at 110◦C under vacuum. Disk-shaped electrodes with a diameter of 13 mm were then punched from the foil (the active material loading was 1.52 mg cm−<sup>2</sup> ). CR2016 coin-type cells were assembled in an Arfilled glove box with an LiFePO<sup>4</sup> cathode working electrode, a Li foil counter electrode, and a polypropylene (Celgard 2400) separator; 45 mL of electrolyte was then injected into the cells. Finally, the assembled coin cells were sealed for further tests. Li/Li and Li/Cu cells with different electrolyte solutions and the above-mentioned separator were assembled for further tests.

### Electrochemical Measurements

The assembled Li/LiFePO<sup>4</sup> coin cells were left to stand for 10 h before galvanostatic precycling for 200 cycles at 1 ◦C (1◦C = 170 mAh g−<sup>1</sup> ) between 2.5 and 4.2 V at 25◦C using the Land battery test system (Wuhan LANHE Electronics, China). Rate-performance tests were conducted by changing the rate (0.2, 0.5, 1.0, 2.0, 5.0, and 0.2◦C) every 10 cycles. Cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS) measurements were carried out using a SI-1287 Electrochemical System (Transmission Precision Measurement Company, UK); the CV tests were performed between 2.5 and 4.2 V at a scan rate of 0.1 mV s−<sup>1</sup> and the EIS tests were performed over 100 kHz to 10 mHz range with an amplitude of 10 mV before and after rate tests. Then the assembled cells were galvanostatic precycled at 65◦C.

The current density for the Li metal plating/stripping was set to 1.0 mA cm−<sup>2</sup> using a Land battery testing station at room temperature. And the deposition time was 1.5 h. The effective area of the Cu foil for Li deposition was 2.11 cm<sup>2</sup> . The current density was 1.0 mA cm−<sup>2</sup> from 0.01 to 1.00 V. Li/Li symmetric cells were assembled with Li metal used as the working and counter electrodes. The batteries were tested in a Tenney JR environmental chamber to ensure the stable temperature during long-term cycling process.

### Morphological Characterizations

Scanning electron microscopy (SEM) (FE-SEM, LEO 1530) was employed to observe the surface topography of the LiFePO<sup>4</sup>

FIGURE 2 | Electrochemical performances of Li/Cu cell (A,B) and Li/Li cells (C,D) with two electrolytes at a current density of 1.0 mA cm−<sup>2</sup> . (A,C) the 1 M LiPF6 electrolyte; (B,D) the 4 M LiODFB-DME.

electrode and Cu foil at different cycling stages at an accelerating voltage of 5 kV. The XPS is tested to further study the composition of SEI film on the Cu foil by the Thermo Electron model K-Alpha surface analysis system. The electrodes retrieved from the cells were thoroughly rinsed in high purity dimethyl carbonate (DMC) solvent for three times and dried under vacuum before transferring to the observation chamber.

### EXPERIMENTAL RESULTS

## Li Metal Deposition Morphology

The morphologies of Li deposition in both electrolytes are evaluated by using coin-type Cu/Li cells. The surface morphologies of Li electrodes cycled after 1.5 h in different electrolytes is shown in **Figure 1**. In conventional carbonate based electrolyte, significant dendrite morphology can be observed (**Figure 1a**). It can easily pierce most conventional separators, causing the direct contact of the positive and negative electrodes to form a short circuit, and lead to serious security risks. On the contrary, Li deposit from the 4 M LiODFB-DME electrolyte exhibits smooth solid particle morphology (**Figure 1b**), which dramatically inhibits the growth of Li dendrite. Based on the Li metal deposition morphologies, it is reasonable to conclude that the 4 M LiODFB-DME has good negative electrode stability. The SEI film formed in the high concentration electrolyte has high ionic conductivity, which makes a uniform and stable SEI film formed. The uniform and stable SEI film can suppress the growth of Li dendrite.

# Li Metal Plating/Stripping Cycling Stability

The impedance evolution of the coin-type Cu/Li cells after 0.5 h and 1.5 h are depicted in **Figure S1**. In the Nyquist plots, the intercept with the abscissa represents the ohmic resistance, particularly the electrolyte resistance in the cell. A significant difference in the impedance between the two electrolytes is observed. After 0.5 and 1.5 h of deposition process, the commercial electrolyte exhibits relatively higher electrode impedance **Figure S1A**. At the same time, 4 M LiODFB-DME electrolyte shows small impedance **Figure S1B**.

FIGURE 5 | (A) cycle performance of the Li/LiFePO<sup>4</sup> cells at 65◦C with 4 M LiODFB-DME and 1 M LiPF6, (B) rate performance of Li/LiFePO<sup>4</sup> cells at 65◦C with 4 M LiODFB-DME and 1 M LiPF6.

The most possible reason for the small impedance is attributed to the formation of the stable SEI film, which is consistent with the results of a dense surface morphology in **Figure 1b**. Polarization measurements of the Li/Li cells were conducted to understand the effect of different electrolytes on stabling the interface between the Li metal and electrolyte. During this experiment, a constant deposition/dissolution current density of 1.0 mA cm−<sup>2</sup> was passed through the Li/Li cells for 3,000 h at room temperature. The 1 M LiPF<sup>6</sup> electrolyte shows the great over-voltage and an ascending tendency with time (**Figure 2**), which is consistent with previous report (Webb et al., 2014). Surprisingly, the 4 M LiODFB-DME electrolyte appears excellent Li deposition/dissolution performance with the stable and low over-voltage. Polarization is lower than 1 M LiPF<sup>6</sup> electrolyte. Besides, influenced by high viscosity, on the one hand, it possibly increases the pressure from the electrolyte to push back growing dendrites, resulting in a more uniform deposition on the surface of the anode. On the other hand, high viscosity limits anion convection near deposition area, which is also helpful to deposit uniformly. All of the above results confirm that the high concentration of LiODFB electrolyte in ether solvent can ensure the cycle stability in the negative.

# DISCUSSION

### Discussion for the Possible Function Mechanism

**Figure 3A** compares the C1s spectra of the delithiated graphite electrodes extracted from the cell after several cycles. The dominant peak at 284.6 eV originates mainly from the sp<sup>2</sup> hybridized graphite, but also includes contributions from conductive carbon added into the electrode composites. The broad feature at 286.6 eV is assigned mainly to the C atoms of the C-O-C groups in the poly(ethylene oxide), PEO, which is formed upon solvent polymerization. The formation of PEO is also confirmed by the peak at 533.6 eV in the O1s XPS spectra in **Figure 3B**. The rather broad peaks at 288.6 eV in **Figure 4A** are assigned to carbon in C=O groups in organic Li alkyl carbonates (ROCO2Li) and Li carbonates (Li2CO3), respectively. The corresponding peak of the oxygen atoms in these groups is detected in the O1s spectra (**Figure 3C**) at 531.8 eV. **Figure 3C**, the F1s peak at 687.5 eV are assigned with PVDF, while the peak at 684.2 eV corresponding to LiF is much stronger, agreeing with the inorganic inner layer of SEI. However, there is almost no difference in contrast to the case of the commercial electrolyte in the C1s, O1s, and F1s spectra. The difference appears at F1s spectra. The B1s peaks at 193.8 eV and 192.6 eV (**Figure 3D**) indicates the presence of LiODFB and material with B–O, respectively. And no such peak in commercial electrolyte.

This result indicates that the B-O bond is involved in the formation of the film, making the SEI film is much denser (Xu et al., 2016). It is consistent with the results of a dense surface morphology in **Figure 1b**.

# The Electrochemical Performances of LiFePO<sup>4</sup> Batteries

The assembled Li/LiFePO<sup>4</sup> coin cells are used to compare the cycling and rate stability using 4 M LiODFB-DME electrolyte and 1 M LiPF<sup>6</sup> electrolyte at room temperature in this part.

**Figure 4A** shows the first charge/discharge profiles of the cells in different electrolytes at the 0.2◦C current rate (1◦C = 170 mAh g−<sup>1</sup> ). The cells are initially activated at 0.2◦C for three cycles and then cycled at 0.5◦C. The cells display similar voltage curves, but the cell using the 4 M LiODFB-DME electrolyte exhibits slightly smaller discharge capacities of 129 mAh g−<sup>1</sup> , while the initial discharge capacity of 1 M LiPF<sup>6</sup> is 133 mAh g −1 . The result is directly related to the lower ionic conductivity of 4 M LiODFB-DME electrolyte compared to LiPF6-based electrolyte. The long-term cycling performances of Li/LiFePO<sup>4</sup> cells are depicted in **Figure 4C**. There is no obvious difference after 200 cycles, and 4 M LiODFB-DME electrolyte shows little superiority compared to commercial LiPF6-based electrolyte. However, LiODFB electrolyte shows excellent CE of 99% over 200 cycles; while the cell with the LiPF<sup>6</sup> is only 89% after 200 cycles (**Figure 4B**). It shows that LiODFB salt greatly improves the cycling stability, but the delivered capacity might have been hindered by the electrolyte's undesirable ionic conductivity. **Figure 4D** presents the rate capability of Li/LiFePO<sup>4</sup> cell in 1 M LiPF<sup>6</sup> and 4 M LiODFB-DME electrolytes. Discharge capacity of cells with 4 M LiODFB-DME are 154.8 mAh g−<sup>1</sup> at 0.2◦C, 143.3 mAh g−<sup>1</sup> at 0.5◦C, 134.4 mAh g−<sup>1</sup> at 1.0◦C, 123.1 mAh g −1 at 2.0◦C, and 99.5 mAh g−<sup>1</sup> at 5.0◦C. As comparison, discharge capacity of cells with 1 M LiPF<sup>6</sup> are 143.9 mAh g−<sup>1</sup> at 0.2◦C, 136.7 mAh g−<sup>1</sup> at 0.5◦C, 128.3 mAh g−<sup>1</sup> at 1.0◦C, 118.9 mAh g−<sup>1</sup> at 2.0◦C, and 95.8 mAh g−<sup>1</sup> at 5.0◦C. Voltage profiles of both cells are also very similar **Figure S2**. It seems that discharged capacity at different cycling rates of cells with 4 M LiODFB-DME electrolyte are slightly higher than cells with 1 M LiPF<sup>6</sup> electrolyte. The outstanding thermal stability and water stability LiODFB leads to very weak side reactions and the dissolution of iron is suppressed. These are the reasons of better cycling performance of cells in 4 M LiODFB-DME electrolyte. All in all, the electrochemical performances of the two electrolytes capacity reach a considerable level at room temperature. These also can be confirmed on the impedance test results in **Figure S2**.

Furthermore, the cycling performances of Li/LiFePO<sup>4</sup> cells at 65◦C are assessed in both electrolytes (**Figure 5A**). In **Figure 5A**, when the cell is working in LiPF<sup>6</sup> electrolyte, the capacity decay rapidly after cycles. We can discover it only has capacity retention of 79.3% after 100 cycles, which is consistent with the result reported by ZHANG et al. As a comparison, excellent cycle stability is observed when using 4 M LiODFB-DME electrolyte, with high capacity retention of 92.5% after 100 cycles. The rate capability of Li/LiFePO<sup>4</sup> at 65◦C in different electrolytes is obtained in **Figure 5B**. Much larger discharge capacities are showed at higher rates for the 4 M LiODFB-DME. For example, the LiPF<sup>6</sup> electrolyte shows discharge capacities of 101.1 and 62.5 mAh g−<sup>1</sup> at the 1.0 and 2.0◦C rates, respectively, even the capacities of cell is almost close to zero at 5.0◦C; While these values are increased to 148.1, 142.8, 127.7 mAh g−<sup>1</sup> at the 1.0, 2.0, and 5.0◦C rates for the LiODFB electrolyte. This demonstrates that LiODFB not only improve the cycling stability but also improve the rate capability in high temperature. It seems that the better thermal stability of LiODFB salts plays an important role in improving the high-temperature resilience of LiFePO<sup>4</sup> electrode.

We used TGA to study the thermal stability of fully charged electrodes in electrolytes (**Figure 6**). Evaporation of the adsorbed organic solvent appears as their weight loss before 150◦C in 4 M LiODFB-DME electrolyte. As the heating progressed, a significant weight loss begins to appear at about 250◦C, which originates from the decomposition of the SEI film. In contrast, the large-scale weight decay starts at 50◦C in 1 M LiPF6, and the mass attenuation has reached a minimum at 150◦C, proving that the SEI film has completely broken before 150◦C. It shows that 4 M LiODFB-DME electrolyte produces a more stable SEI film and establishes a more gentle interface between the electrode and the electrolyte. So the fully delithiated electrode is protected from directly contacting with the electrolyte. Therefore, the LiODFB electrolyte improves the thermal stability of the highly oxidized charged electrode.

# SEM Micrographs of LiFePO<sup>4</sup> Electrodes

SEM analysis is performed to illustrate electrode images before and after 100 cycles in both electrolytes at 65◦C. The LiFePO<sup>4</sup> electrode after long-term cycling in 4 M LiODFB-DME electrolyte (**Figure 7b**) shows a similar morphology to the pristine LiFePO<sup>4</sup> electrode (**Figure 7a**). In comparison, the electrode is severely eroded in 1 M LiPF<sup>6</sup> electrolyte after 100 cycles at 65◦C in **Figure 7c**, and we can found that the LiFePO<sup>4</sup> electrode surface in 1 M LiPF<sup>6</sup> electrolyte displays some cracks among the particles. It is result from the poor thermal stability of the 1 M LiPF<sup>6</sup> electrolyte. The decomposition of LiPF<sup>6</sup> in high temperature causes serious capacity decay rapidly, resulting in very poor electrochemical performance, as shown in **Figure 5A**. The higher the current density, this effect is more serious in **Figure 5B**. These SEM observations confirm that 4 M LiODFB-DME electrolyte has excellent compatibility with LiFePO<sup>4</sup> cathodes during battery cycling at high temperature **Figure S3** shows the EDS layered image of LiFePO<sup>4</sup> electrode after 100 cycles at high magnification. The surface of the LiFePO<sup>4</sup> electrode consists of five main elements: O, P, F, and Fe. The Fe and P elements are mainly derived from LiFePO<sup>4</sup> electrode and the O, F, and B elements are derived from LiODFB lithium salt. It is proved that the B element from the LiODFB-based electrolyte participates the formation of the surface film of LiFePO<sup>4</sup> cathode. And the participation of the B elements makes the surface film of LiFePO<sup>4</sup> cathode smooth and compact. All in all, the LiODFB lithium salt protects the cathode. In addition, the SEM images of the morphologies of Al foils is shown in **Figure S4**. The surface of the Al foils had a large difference. The concentrated electrolytes suppress the Al corrosion more effectively.

## CONCLUSIONS

In this work, we first developed a concentrated electrolyte based on LiODFB in ether solvents for LMBs. Based on 4 M LiODFB-DME electrolyte, Li was deposited on the copper foil and its surface showed smooth solid particle morphology without any dendrites. For the Li/LiFePO<sup>4</sup> cells, it is demonstrated that 4 M LiODFB-DME electrolyte exhibit good electrochemical performance, especially at the elevated temperature. Li/LiFePO<sup>4</sup>

# REFERENCES


cells using 4 M LiODFB-DME electrolyte show a good capacity retention (92.5%) at elevated temperature (65◦C), which is much higher than that of 1 M LiPF<sup>6</sup> electrolyte (79.2%). This research offers the possibility of rapid industrialization of LMBs.

# 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

ZF and JY conceived and designed the experiments. NG, JP, XL, and NM performed the experiments. NG, CS, and XL analyzed the data. NG and NM wrote the manuscript. KX designed the scheme. All authors reviewed the manuscript and approved the final version.

# ACKNOWLEDGMENTS

We acknowledge the financial support of this work by the Project (51574191, 51304151) supported by the National Natural Science Foundation of China; Project (18JK0474) supported by Shaanxi Provincial Education Department, China; Project (2018JM5135) supported by the Natural Science Basic Research Plan in Shaanxi Province, China.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2019.00494/full#supplementary-material


nanofiber-guided uniform lithium deposition for battery electrodes. Nano Lett.


high power lithium-sulfur batteries. ACS Appl. Mater. Interfaces 9, 4605–4613. doi: 10.1021/acsami.6b14039


**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 Yu, Gao, Peng, Ma, Liu, Shen, Xie and Fang. 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.

# Controlled Synthesis of NixCoyS4/rGO Composites for Constructing High-Performance Asymmetric Supercapacitor

Mingxia Dong, Zhixing Wang, Jiexi Wang, Huajun Guo, Xinhai Li and Guochun Yan\*

*School of Metallurgy and Environment, Central South University, Changsha, China*

Nickel-cobalt sulfides (Ni*x*Co*y*S4) are promising supercapacitor materials due to their high capacitance, while the sluggish kinetics in terms of charge transfer limits their energy density. To achieve both high energy and power density for the Ni*x*Co*y*S4–based supercapacitor, sulfur-doped reduced graphene oxide (rGO) is incorporated into Ni*x*Co*y*S<sup>4</sup> by an *in situ* growth—ion exchange strategy to synthesize Ni*x*Co*y*S4/rGO composites. Profited from the synergetic effect between rGO and Ni*x*Co*y*S4, the Ni1.64Co2.40S4/rGO electrode delivers high specific capacitance of 1,089 F g−<sup>1</sup> at 1 A g−<sup>1</sup> , and remains 92.6% of its original capacitance at 20 A g−<sup>1</sup> (1,008 F g−<sup>1</sup> ). Asymmetric supercapacitors assembled with active carbon (AC) and Ni1.64Co2.40S4/rGO (Ni1.64Co2.40S4/rGO//AC) offer both high specific capacitance (265.5 F g−<sup>1</sup> at 1 A g−<sup>1</sup> ) and superior rate capability at 50 A g−<sup>1</sup> (recovering 63.6% of the capacitance determined at 1 A g−<sup>1</sup> ). In addition, the assembled device exhibits a high capacitance retention of 92.6% after 10,000 cycles at 10 A g−<sup>1</sup> , which implies an excellent cyclic stability. Ragone plot reveals that the energy density of Ni1.64Co2.40S4/rGO//AC asymmetric supercapacitor do not vanish as it delivers 30.4 Wh kg−<sup>1</sup> at 10 kW kg−<sup>1</sup> , demonstrating its promising application.

Keywords: nickel-cobalt sulfides, in-situ growth, high energy density, asymmetric supercapacitor, high cyclic stability

# INTRODUCTION

With gradual consumption of fuel energy and the growing environmental problems, is the need to develop renewable energies and corresponding energy storage and conversion devices is urgent (Zhong et al., 2012; Liu et al., 2018a; Zhang Q. et al., 2018). Lithium ion battery (Li et al., 2015; Liu et al., 2016; Zhong et al., 2017; Wang J. et al., 2018), sodium-ion battery (Chen J. et al., 2018; Yan et al., 2018), Lithium-Sulfur battery (Tan et al., 2018; Wu F. et al., 2018), and supercapacitor (Gilshteyn et al., 2017; Saha et al., 2018) are representative electrochemical energy storage devices. A supercapacitor (SC) is one of the most promising energy storage devices due to their desirable characteristics, such as high power-density, super-long cycle life and high safety performance, and it plays an important role in both automobile and electronic industry (Simon et al., 2014; Eftekhari et al., 2017; Zhang L. et al., 2018). Classified by different energy storage mechanisms, electrical double layer capacitors (EDLCs), and pseudocapacitors (PCs) show differentiated advantages in the specific field. Although EDLCs deliver fast electron transfer, their

### Edited by:

*Shengkui Zhong, Soochow University, China*

### Reviewed by:

*Biao Gao, Wuhan University of Science and Technology, China Chaopeng Fu, Shanghai Jiao Tong University, China*

\*Correspondence:

*Guochun Yan happyygc@csu.edu.cn*

### Specialty section:

*This article was submitted to Energy Materials, a section of the journal Frontiers in Materials*

> Received: *29 May 2019* Accepted: *08 July 2019* Published: *24 July 2019*

### Citation:

*Dong M, Wang Z, Wang J, Guo H, Li X and Yan G (2019) Controlled Synthesis of NixCoyS*4*/rGO Composites for Constructing High-Performance Asymmetric Supercapacitor. Front. Mater. 6:176. doi: 10.3389/fmats.2019.00176* bottleneck lies in the low specific capacitance problem. On the contrary, PCs deliver satisfying specific capacitance, but poor charge transfer ability limits their application. Neither of them can be used to build a capacitor that possesses both high energy density and power density. Developing a hybrid capacitor combining with the advantages of EDLCs and PCs is an effective strategy to remedy the deficiency.

To fulfill the above design, developing advanced materials with both high capacitance and rate capability are the key factors. Normally, the composite formed with proper composition of two materials results in complementary advantages. As known to all, reduced graphene oxide (rGO) is one of the most promising electrode materials for EDLCs due to its large specific surface area, exceptionally high electronic conductivity and high theoretical specific capacitance (Booth et al., 2008; Lee et al., 2008; Ke and Wang, 2016). When rGO is introduced into the composite properly, it enhances not only the electronic conductivity of composite, but also the structural stability of the electrode (Ke and Wang, 2016; Deng et al., 2017). Therefore, taking reduced graphene oxide as a framework is an effective way to reinforce the mechanical strengths of the composites. Nickel-cobalt sulfide is one of the most promising one among various pseudocapacitor electrode materials, such as metal oxides, metal sulfides, and conductive polymers (Jinlong et al., 2017; Wen et al., 2017; Zhao et al., 2017; Liang et al., 2018). Although, there is some doubt about the definition of pseudocapacitance (Brousse, 2015; Jiang and Liu, 2019), many studies reported that transition metal sulfides were classified as one kind of pseudocapacitive materials (Xiong et al., 2015; Wang et al., 2019). However, some studies classify transition metal sulfides as battery-type electrode materials (Mohamed et al., 2018; Ma et al., 2019). Putting this controversial issue aside, it has been proven that nickelcobalt sulfides (NixCoyS4) exhibit high electronic conductivity and flexibility due to the lower optical band gap of Ni–Co and electronegativity of sulfur (Chen et al., 2013; Zhu et al., 2015). Great efforts have been devoted to the fabrication of composites, such as NiCo2S4/rGO (Annamalai et al., 2017; Nan et al., 2018; Sarkar et al., 2018; Wang Y. et al., 2018), CoNi2S4/rGO (Gao et al., 2018), NiS/rGO (Qu et al., 2018), C-NiCo2S<sup>4</sup> (Yuan et al., 2017; Mohamed et al., 2018), CNTs-NiCo2S<sup>4</sup> (Li D. et al., 2016), and so on. However, the influence of Ni/Co ratio of nickel-cobalt sulfide materials on the characteristics and the pseudocapacitance performances are not systematically studied.

In this work, we prepare a series of NixCoyS4/rGO composites with various Ni/Co ratios in the raw materials (Ni/Co = 2.0, 1.0, 0.5, and 0.25), in which the bimetallic Ni-Co hydroxy-carbonates in-situ grow onto the rGO framework during the synthesis process. We demonstrate that asymmetric supercapacitor (Ni1.64Co2.40S4/rGO//AC) assembled with Ni1.64Co2.40S4/rGO composite (positive electrode) and active carbon (negative electrode) exhibits excellent electrochemical performance with outstanding capacitance retention of 92.6% after 10,000 cycles at high current density of 10 A g−<sup>1</sup> and high energy density of 30.4 Wh kg−<sup>1</sup> at the power density of 10 kW kg−<sup>1</sup> . Our results offer new insights into developing advanced electrode materials for supercapacitors with high energy and power density.

# EXPERIMENTAL

The graphene oxide (GO) was purchased from Suzhou Tanfeng technology co LTD, and all of other chemical reagents were provided by Sinopharm Chemical Reagent Corp in an analytical grade that were used as-received without further purification.

# Synthesis of NixCoyS<sup>4</sup>

In a typical synthesis of precursors, 0.025 g GO was dissolved into 100 mL deionized water to form homogeneous brown solution by ultrasonic dispersion. A certain mass of NiCl2·6H2O, CoCl2·6H2O (molar ratios of Ni: Co are 2:1, 1:1; 1:2, and 1:4) and 1.6365 g urea were dissolved into the GO solution under stirring. After stirring for 30 min, the obtained suspension was transferred into a 200 mL Teflon-lined autoclave. The hydrothermal reaction was carried out at 120◦C for 10 h so that the bimetallic Ni-Co hydroxy-carbonates in-situ grow onto the GO framework. The products were washed with deionized water and collected through filtration, and then which was lyophilized. Based on the Ni/Co ratio in the raw materials, the as-obtained precursors were labeled as NC2010, NC1515, NC1020, and NC0624, respectively.

For synthesizing the NixCoyS4/rGO by sulfurizing the aforementioned precursors, the representative processes are described as followed. First, 0.2 g precursor was dispersed into 100 mL deionized water and the mixture was stirred for 30 min. Subsequently, 1.7156 g Na2S·9H2O was added into the suspension and stirred for 30 min. Then, the mixture was transferred into a 200 mL Teflon-lined autoclave, which was heated at 180◦C for 8 h. After cooling naturally to room temperature, the black product was collected in the same way as former procedure. To further illuminate the content of Ni, Co and S in the NixCoyS4/rGO composites, the Inductively Coupled Plasma-Optical Emission Spectrometry (ICP-OES) results and the C-S analysis results of NixCoyS4/rGO composites are shown in **Table S1**. The as-prepared products were labeled as Ni2.15Co1.37S4/rGO, Ni1.64Co2.40S4/rGO, Ni1.02Co2.98S4/rGO, Ni0.60Co3.60S4/rGO on the basis of ICP and C-S analysis results.

### Materials Characterization

Structure, morphology, and microstructure of the precursors and the corresponding NixCoyS4/rGO composites were characterized by X-ray diffraction using Cu Kα (λ = 1.5406 Å) radiation (XRD, Bruker D8 AdvanceRint-2000), scanning electron microscopy (SEM, Sirion 200), and transmission electron microscopy (TEM, Tecnai G12, 200 kV). The Surface elemental valence of the sample was characterized by X-ray photoelectron spectroscopy (XPS, PHI5600, PerkineElmer).

### Electrochemical Measurement

Circular Ni foams were cleaned in 1.0 mol L−<sup>1</sup> KOH solution with ultrasonic cleaning for 30 min to remove impurities on the surface, and washed with deionized water three times. The preparation of electrodes was followed in the same procedure as described in previous work (Dong et al., 2018), and the diameter of Ni foam for positive and negative electrode is 12 and 16 mm, respectively.

Electrochemical tests were performed in three-electrode systems with 6 mol L−<sup>1</sup> KOH as electrolyte, the working electrode was prepared by mixing NixCoyS4/rGO composites, conductive additive (Super P), and binder (60 wt% PTFE emulsion) in a ratio of 8:1:1, Hg/HgO electrode as the reference electrode and Pt electrode as the counter electrode, respectively. By contrast, electrochemical tests carried out in the asymmetric supercapacitor were based on the abovementioned working electrode (positive electrode) and active carbon electrode (negative electrode), in which the active carbon electrode was prepared by mixing active carbon, conductive additive (Super P) and binder (60 wt% PTFE emulsion) in a ratio of 85:10:5. Cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS) were tested by a CHI 660D electrochemical workstation (Chenhua Instruments, Shanghai). Galvanostatic charge-discharge (GCD) tests were carried out with the LAND electrochemical test system. Additional experimental details are described in the **Supplementary Material** of the work.

### RESULT AND DISCUSSION

The crystallographic characteristics of the as-obtained precursors and NixCoyS4/rGO composites are characterized by XRD. As shown in **Figure S2**, for all samples, most of the diffraction peaks can be indexed to NixCoy(CO3)0.5(OH) (x+y = 1) that is derived from Co(CO3)0.5(OH)·0.11H2O (JCPDS Card No.48-0083). No impurity peaks were observed, which illustrates the substitution of Co by Ni during the growth of crystals with the increasing of nickel contents in the precursors (Xiao and Yang, 2011; Wan et al., 2013). With the decrease of Ni/Co ratio, the intensities of the diffraction peaks increase gradually, suggesting that the Co element plays a role in improving the crystallinity of precursor. For the XRD patterns of NixCoyS4/rGO composites (**Figure 1**), all of the XRD patterns of all NixCoyS4/rGO composites show weak diffraction peaks. The diffraction peaks at the 2θ of 31.4◦ , 38.2◦ , 50.1◦ , and 55.1◦ are ascribed to (311), (400), (511), and (440) planes of the cubic NiCo2S<sup>4</sup> (JCPDS Card No.20-0782) or CoNi2S<sup>4</sup> (JCPDS Card No.24-0334). The Ni1.64Co2.40S4/rGO composite owns the strongest diffraction peaks, reflecting the highest crystallinity among all samples.

SEM and TEM technique were used to characterize the morphology of the precursors and NixCoyS4/rGO composites as shown in **Figures S3, S4** and **Figure 2**. **Figures S3a–d** show SEM images and EDS results of NC2010, NC1515, NC1020, and NC0624, respectively. For all precursors, a nano-needle type of Ni-Co compounds were embedded onto the surface of rGO network. With the decrease of Ni/Co ratio, the size of nano-needles gradually increases as shown in **Figure S3**. The NC0624 possess maximum microscopic size with a longitudinal length larger than 2µm, indicating that Co plays a crucial role in the vertical growth of crystal. EDS results reveals that the ratio of Ni/Co content decrease from 1.5 to 0.2, which is in well-accordance with our experimental design. Furthermore, the TEM and high-resolution transmission electron microscope (HRTEM) images of NC1515 are carried out to further illustrate the morphology and microstructure of the precursor. As shown

in **Figure S4**, the Ni-Co compounds with solid nano-needle morphology evenly distributed on the surface of rGO with transparent morphology, and the diameter of the nanoneedle is about 20 nm. The interplanar crystal spacing of 0.88 nm in the HRTEM image (**Figure S4d**) can be referred to the (100) plane of NixCoy(CO3)0.5(OH).

**Figures 2a–d** show SEM images of Ni2.15Co1.37S4/rGO, Ni1.64Co2.40S4/rGO, Ni1.02Co2.98S4/rGO, and Ni0.60Co3.60 S4/rGO composites, respectively. It is obvious that the sulfides inherit the morphological features of corresponding precursors with the NixCoyS<sup>4</sup> decorated on the surface of rGO. It is worth noting that the TEM images of Ni1.64Co2.40S4/rGO (see **Figure S4** and **Figures 2e–h**) confirms that the solid nanoneedles transfer to nano-tube during ion exchange process from the precursor to sulfides. Specifically, the hollow structures with a rough surface possess an outer diameter of about 35 nm, which is larger than the diameter of the precursor (20 nm). The TEM results of precursors and Ni1.64Co2.40S4/rGO composite suggest that the Ni2<sup>+</sup> ions and Co2<sup>+</sup> ions diffused outwards from the precursor and combined with sulfur ions to form sulfides. The EDS results of NixCoyS4/rGO composites in **Figure S1** reveals that Ni, Co, and S elements exist on the surface of the composites. In addition, it is worth noting that the S-doped graphene is also formed during the ion-exchange process, which is evidenced by a set of comparison experiments. Specifically, the GO was reduced in Na2S·9H2O solution under the same condition as preparing NixCoyS4/rGO composites, which can be confirmed by the C1s, and O1s XPS spectra in **Figure S5**. Moreover, the characteristic of sulfur peak in the EDS spectra (**Figure S6**) and C=S, C-S-C bonds in S2p spectra of aforementioned rGO sample confirm the existence of sulfur in rGO, which is beneficial to improve the pseudo-capacitance because of the heteroatom doping of S (Liu et al., 2017).

To further probe the formation of nickel-cobalt sulfides (NixCoyS4) and the S-doped rGO in the sulfide composite, XPS

FIGURE 2 | The morphology of Ni*x*Co*y*S4/rGO composites. (a–d) SEM images of Ni*x*Co*y*S4/rGO composites; Ni2.15Co1.37S4/rGO (a), Ni1.64Co2.40S4/rGO (b), Ni1.02Co2.98S4/rGO (c), and Ni0.60Co3.60S4/rGO (d). (e–h) TEM and HRTEM images of Ni1.64Co2.40S4/rGO; TEM images of Ni1.64Co2.40S4/rGO (e,f); HRTEM images of Ni1.64Co2.40S4/rGO (g,h).

(X-ray photoelectron spectroscopy) is utilized to analyze the elemental composition and valence state of the material. **Figure 3** shows survey, Ni 2p, Co 2p, S 2p, C 1s, and O 1s spectrum of the Ni1.64Co2.40S4/rGO composite, and **Table S2** lists the peak fitting parameters of all spectrums. As shown in **Figure 3A**, it can be confirmed that Ni, Co, S, C, and O elements exist on the surface of Ni1.64Co2.40S4/rGO composite. In high resolution spectrums of Ni 2p and Co 2p, the spectrums can be classified into 2p 3/2 and 2p 1/2 sections, and the corresponding satellite peaks are labeled as "Sat" in the plot. Specifically, the peaks at 853.5 and 870.8 eV in Ni 2p correspond to Ni2+, and the peaks at 857.7 and 875.5 eV can be assigned to Ni3+, confirming the coexistence of both Ni2<sup>+</sup> and Ni3<sup>+</sup> in the composite. Similarly, the peaks located at 779.0 eV, 794.1 eV in Co 2p corresponds to the Co3+, and the peaks at 782.2 eV, 798.8 eV of Co 2p spectrums associates to the Co2+, suggesting the coexistence of Co3<sup>+</sup> and Co2<sup>+</sup> in Ni1.64Co2.40S4/rGO composite as well (Fan et al., 2017; Qin et al., 2017). For the S 2p spectrum, the peak located at 161.8 eV for S 2p 3/2 is attributed to S2<sup>−</sup> or C-S-C bonds on the surface (Chen X. et al., 2014), while the peak at 162.9 eV for S 2p1/2 can be ascribed to S-M (metal) bonds. In addition, the peak at a high binding energy of 169.6 eV (satellite) can be attributed to the highly oxidized state sulfur species (S4+) (Ma et al., 2016). The added Na2S·9H2O act as both of S2<sup>−</sup> sources and reductant during the high temperature ion-exchange reaction, in which part of S2<sup>−</sup> loses electrons and been oxidized to the higher-valence, and the GO been reduced to rGO simultaneously. Furthermore, the C 1s spectrum exhibits a major peak at 284.7 eV attributed to C-S-C bonds (Wenfang et al., 2015). Additionally, a peak located at 285.9 eV in the C 1s spectra corresponds to the C=C bond (aromatic-linked carbon), while the weak peak at 289.2 eV can be ascribed to the carboxyl groups (C=O) and sp<sup>2</sup> hybridized carbon from rGO, the low strength of the peak at 289.2 eV also indicates that the graphene has been reduced (Qin et al., 2017). The S 2p and C 1s XPS results reveal the existence of S-M (M=Ni or Co) and C-S-C bonds, respectively, which verifies the doping of S into rGO and the formation of transition metal sulfides in the composite material. Based on the XRD, SEM, TEM, and XPS results as analyzed in **Figures 1–3**, we proofed that the NixCoyS4/rGO composites can be synthesized by such in situ growth—ion exchange strategy.

The electrochemical performances of NixCoyS4/rGO composites are measured by cyclic voltammetry at various scan rates in a three-electrode cell as shown in **Figure 4**. For all composites, anodic and cathodic peaks with good symmetry indicate the good reversibility of the materials. The positions of anodic and cathodic peaks shift to the more positive and more negative directions with the increase of scan rates from 1 mV s−<sup>1</sup> to 100 mV s−<sup>1</sup> , which implies an increase of the polarization. It is known to all that the relationship between scan rate and current density is an effective method to distinguish capacitive contribution by the i = aν b formula. Specifically, when the b value is 1/2, the electrode shows a total diffusion-limited behavior (i = aν 1/2 ); and when the b value is 1, the electrode shows a capacitive process (Lim et al., 2015). **Figures 4E–H** show the b values of Ni2.15Co1.37S4/rGO, Ni1.64Co2.40S4/rGO, Ni1.02Co2.98S4/rGO and Ni0.60Co3.60S4/rGO as deduced from the slope of log (i)—log (v) curves, which are 0.55, 0.78, 0.86, and 0.89, respectively. The b values of the electrodes increase with the decrease of Ni/Co ratio, which implies that the Ni2.15Co1.37S4/rGO composite is mainly diffusion-controlled kinetics behavior, and gradually transforms to the capacitive behavior for the Ni0.60Co3.60S4/rGO composites. As the switch of kinetics behavior from diffusion-controlled to capacitive behavior, it will accelerate the charge transfer process so as to improve the rate capability of the NixCoyS4/rGO composites.

The CV results of Ni2.15Co1.37S4/rGO, Ni1.64Co2.40S4/rGO, Ni1.02Co2.98S4/rGO, and Ni0.60Co3.60S4/rGO at 1 mV s−<sup>1</sup> are presented in **Figure 5A**. Obviously, the Ni2.15Co1.37S4/rGO material has the largest integral area, indicating the highest specific capacitance as compared to other samples. The specific capacitance of the material decreases with the decrease of Ni/Co ratio in NixCoyS4/rGO composites, which is related to the microstructure transformation of the material. Specifically, Ni1.64Co2.40S4/rGO consists of tiny one-dimensional nanoneedles that convince to the sufficient contact between the active material and the electrolyte during the electrode process. In addition, the response current of the CV curves is a combination of capacitor and diffusion-controlled behavior. The respective contribution between diffusion-controlled and capacitive capacitance can be calculated by the following equation (Lim et al., 2015):

$$\mathbf{i}(\mathbf{V}) = \mathbf{k}\_1 \boldsymbol{\upsilon} + \mathbf{k}\_2 \boldsymbol{\upsilon}^{1/2} \tag{1}$$

The capacitance proportion of diffusion-controlled process and capacitor-like processes are expressed by k1ν and k2ν 1/2 , respectively. The capacitive-contribution of Ni1.64Co2.40S4/rGO is 79.83% at 1 mV s−<sup>1</sup> as shown in **Figure S7**, which confirms that the specific capacitance is mainly contributed by the capacitive-like process. Besides, the difference between the oxidation potential and reduction potential decrease obviously (the potential differences of NixCoyS4/rGO composites are 0.132 V (Ni2.15Co1.37S4/rGO), 0.104 V (Ni1.64Co2.40S4/rGO), 0.089 V (Ni1.02Co2.98S4/rGO), and 0.093 V (Ni0.60Co3.60S4/rGO) respectively), indicating an gradually improved reversibility of the composites, which is beneficial to improve the rate performance of the electrodes. **Figures 5B,C** shows the rate capability of the composites as the current density range from 1 A g−<sup>1</sup> to 20 A g−<sup>1</sup> . It is noteworthy that the rate performance is improved distinctively with the decrease of Ni/Co molar ratio in composites. Specifically, when the current density increase to 20 A g−<sup>1</sup> , the Ni1.64Co2.40S4/rGO, Ni1.02Co2.98S4/rGO, and Ni0.60Co3.60S4/rGO composites can recover 92.6, 99.5, and 99.6% of their capacitance determined at 1 A g−<sup>1</sup> , respectively (**Table S3**). In contrast, the Ni2.15Co1.37S4/rGO composite can only retain 68.9% of its capacitance determined at 1 A g−<sup>1</sup> under the same condition (**Table S3**). Our detailed electronic conductivity results (**Figure 5D**) evidence that higher Co content in NixCoyS4/rGO composites is beneficial to improve the electronic conductivity of the composites, which is the main reason for the improvement of rate performance. Considering the merits of capacitance and rate capability of the composites, Ni1.64Co2.40S4/rGO exhibits the best comprehensive performance in the application of supercapacitor, which delivers the capacitance of 1,089 F g−<sup>1</sup> at the current density of 1 A g−<sup>1</sup> ,

and retains 1,008 F g−<sup>1</sup> at the current density of 20 A g−<sup>1</sup> . The GCD curves of Ni2.15Co1.37S4/rGO, Ni1.64Co2.40S4/rGO, Ni1.02Co2.98S4/rGO, and Ni0.60Co3.60S4/rGO at the current density of 1 A g−<sup>1</sup> show nearly symmetric GCD curves, implying the high coulomb efficiency of the charge-discharge process (**Figure 5E** and **Figure S8**). Meanwhile, the time corresponding to the discharge platform gradually shortens with the decrease of Ni/Co ratio, indicating that the proportion of pseudocapacitance originated from Faraday reaction reduces. It is in well-accordance with the result of CV results as shown in **Figure 5A**. Moreover, the GCD curves of Ni1.64Co2.40S4/rGO at various current densities in **Figure 5F** suggests that the electrode have excellent rate performance.

To demonstrate the practical application of Ni1.64Co2.40S4/rGO composite for energy storage, an asymmetric supercapacitor marked as Ni1.64Co2.40S4/rGO//AC was assembled, in which Ni1.64Co2.40S4/rGO acts as the positive material and active carbon (AC) acts as the negative material. As shown in **Figure 6A**, the shape of the CV curve shows slight differences at various scan rate, indicating that the Ni1.64Co2.40S4/rGO//AC asymmetric supercapacitor has good charge-discharge characteristics. The specific capacitances of Ni1.64Co2.40S4/rGO//AC deduced from the integral area of CV curves at various current density are calculated and shown in **Figure 6B**. High capacitance (270.7 F g−<sup>1</sup> at 1 mV s−<sup>1</sup> ) and high capacitance retention (68.5% at 100 mV s−<sup>1</sup> ) confirm the superior rate performance of the device, which is attributed to the proper composition and hollow structure of Ni1.64Co2.40S4. The relationship between scan rate and current density of Ni1.64Co2.40S4/rGO//AC during the charge-discharge process

(A); rate performance of Ni*x*Co*y*S4/rGO composites at various current densities from 1 A g−<sup>1</sup> to 20 A g−<sup>1</sup> (B); capacitance retention of Ni*x*Co*y*S4/rGO composites with the increase of current density (C); electronic conductivity of Ni*x*Co*y*S4/rGO composites (D); GCD curves of different composites at the current densities of 1 A g−<sup>1</sup> (E); GCD curves of Ni1.64Co2.40S4/rGO at various current densities from 1 A g −1 to 20 A g−<sup>1</sup> (F).

is also calculated as shown in **Figure 6C**. The b value in the charge and discharge process are −0.70 and 0.77, respectively, which is quite consistent with the previous results as measured in the three-electrode cell. The specific capacitance and the corresponding GCD curves of the Ni1.64Co2.40S4/rGO//AC asymmetric supercapacitor at various current densities are presented in **Figures 6D,E**, the specific capacitances are 265.5, 209.4, 195.6, 191.3, 180.0, and 168.8 F g−<sup>1</sup> at the current density of 1, 10, 20, 30, 40, and 50 A g−<sup>1</sup> respectively. The high capacitance retention of 63.6% at 50 A g−<sup>1</sup> as compared to 1 A g−<sup>1</sup> further verifies the excellent rate performance of the asymmetric supercapacitor. Furthermore, the long-term

GCD curves of Ni1.64Co2.40S4/rGO//AC at various current densities (D); average specific capacitances in dependence of cycle number at various current densities from 1 A g−<sup>1</sup> to 20 A g−<sup>1</sup> (E); cycle performance of Ni1.64Co2.40S4/rGO//AC at the current density of 10 A g−<sup>1</sup> (F); columbic efficiency of Ni1.64Co2.40S4/rGO//AC at the last 100 cycles (G); GCD curves of Ni1.64Co2.40S4/rGO//AC at the last 10 cycles of 10000 cycles (H).

cycling measurements of the Ni1.64Co2.40S4/rGO//AC at 10A g−<sup>1</sup> (as shown in **Figure 6F**) evidence the good cycling stability as it can retain 92.6 % of its initial capacitance after 10,000 times, demonstrating the high cyclic stability in terms of electrochemical and mechanical properties of the Ni1.64Co2.40S4/rGO composites. Eventually, the high columbic efficiency approaching 100% and the symmetric GCD curves of the last few loops (**Figures 6G,H**) confirm the excellent electrochemical performance of the Ni1.64Co2.40S4/rGO//AC asymmetric supercapacitor.

For positioning the level of as-obtained Ni1.64Co2.40S4/rGO composite, we carefully investigated and compared the energy density and power density of supercapacitors using similar materials from the peer-reviewed papers. Ragone plot of these asymmetric supercapacitors derived from GCD tests are summarized and plotted in **Figure 7** and **Table S4**. Specifically, the Ni1.64Co2.40S4/rGO//AC asymmetric supercapacitor can reach the maximum energy density of 47.2 Wh kg−<sup>1</sup> at the power density of 200 W kg−<sup>1</sup> , and retains an energy density of 30 Wh kg−<sup>1</sup> even at high power density of 10 kW kg−<sup>1</sup> . Impressively, the Ni1.64Co2.40S4/rGO//AC asymmetric supercapacitor presented in this work is comparable or even superior to those of previous reported transition metal sulfidesbased asymmetric supercapacitors and other supercapacitors (listed in **Table S5**), such as NCS/rGO/CNT-3//AC asymmetric supercapacitors (21.06 Wh kg−<sup>1</sup> at the power density of 729 W kg−<sup>1</sup> and 14.43 Wh kg−<sup>1</sup> at the power density of 3646 W kg−<sup>1</sup> ) (Chiu and Chen, 2018), rGO/CoNiSx/N–C//AC asymmetric supercapacitors (32.9 Wh kg−<sup>1</sup> at the power density of 229.2 W kg−<sup>1</sup> and 20.9 Wh kg−<sup>1</sup> at the power density of 1833.3 W kg−<sup>1</sup> )

(Chen Q. et al., 2018), NiCo2S4@CNFs//AC@pCNFs asymmetric supercapacitors (32.1 Wh kg−<sup>1</sup> at the power density of 67.6 W kg−<sup>1</sup> and 5.3 Wh kg−<sup>1</sup> at the power density of 1,270 W kg−<sup>1</sup> ) (Liu et al., 2018b), GCM//H-GCS FASC (24.6 Wh kg−<sup>1</sup> at the power density of 459.3 W kg−<sup>1</sup> and 17.7 Wh kg−<sup>1</sup> at the power density of 9005.3 W kg−<sup>1</sup> ) (Wu S. et al., 2018), NSC-900/MnO2//NSC-900 asymmetric supercapacitors (39.15 Wh kg−<sup>1</sup> at the power density of 901 W kg−<sup>1</sup> and 28.8 Wh kg−<sup>1</sup> at the power density of 8,907 W kg−<sup>1</sup> ) (Miao et al., 2018), NiCo2S4@NiO//AC (30.38 Wh kg−<sup>1</sup> at the power density of 288 W kg−<sup>1</sup> ) (Huang et al., 2016), NiCo2S4/RGO//AC(24.4 Wh kg−<sup>1</sup> at the power density of 750 W kg−<sup>1</sup> ) (Li Z. et al., 2016), NiCo2S<sup>4</sup> nanotube arrays-Ni foam//RGO (31.5 Wh kg−<sup>1</sup> at the power density of 156.6 W kg−<sup>1</sup> ) (Chen H. et al., 2014), Co9S<sup>8</sup> nanoflake//AC (31.4 Wh kg−<sup>1</sup> at the power density of 200 W kg−<sup>1</sup> and 26.3 Wh kg−<sup>1</sup> at the power density of 4,000 W kg−<sup>1</sup> ) (Rakhi et al., 2014).

# CONCLUSION

We have reported an in situ growth—ion exchange method to synthesize NixCoyS4/rGO composites with high energy and power density. We concluded from the composition, electronic conductivity, and electrochemical measurements that proper Ni/Co molar ratio in NixCoyS4/rGO composites is vital to synthesize the material with high capacitance and rate capability.

### REFERENCES


Here the as-prepared Ni1.64Co2.40S4/rGO composite shows the best overall electrochemical performance among the obtained samples as it delivers a specific capacitance of 1,089 F g−<sup>1</sup> at 1 A g−<sup>1</sup> , and it maintains 1,008 F g−<sup>1</sup> at high current density of 20 A g−<sup>1</sup> . The derived Ni1.64Co2.40S4/rGO//AC asymmetric supercapacitor offers high energy density of 47.2 Wh kg−<sup>1</sup> (200 W kg−<sup>1</sup> ), together with a superior cyclic stability (maintaining 92.6 % of capacitance retention after 10,000 cycles at 10 A g−<sup>1</sup> ). Through this study, the synthesized and characterized Ni1.64Co2.40S4/rGO composite provides a promising material for fabricating supercapacitors with both high energy and power density.

# DATA AVAILABILITY

All datasets generated for this study are included in the manuscript/**Supplementary Files**.

# AUTHOR CONTRIBUTIONS

The design and conduction of the experiment, the acquisition of data, the writing and modification of the paper are completed by MD. ZW and JW helped perfecting the experimental ideas and technical route. GY devoted many efforts to revise the manuscript. HG and XL gave advices during the discussion of the experimental results.

# FUNDING

This research was supported by the National Science Foundation of China (51674295, 51804344), the Fundamental Research Funds for the Central University of Central South University (2018zzts017), the Hunan Provincial Innovation Foundation for Postgraduate (CX2018B088).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmats. 2019.00176/full#supplementary-material

Supporting information for publication is available. XRD patterns, SEM images, TEM images of precursors; XPS, SEM, and EDS results of GO and rGO-Na2S·9H2O; GCD curves of Ni2.15Co1.37S4/rGO, Ni1.02Co2.98S4/rGO, and Ni0.60Co3.60S4/rGO; C-S and ICP results of NixCoyS4/rGO composites.


hybrid structures for sodium-ion batteries. J. Power Sources 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 © 2019 Dong, Wang, Wang, Guo, Li and Yan. 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.

# Flexible Li[Li0.2Ni0.13Co0.13Mn0.54]O2/Carbon Nanotubes/Nanofibrillated Celluloses Composite Electrode for High-Performance Lithium-Ion Battery

### Yan Li, Han Zhang, Zhe Xiao and Renheng Wang\*

*International Collaborative Laboratory of 2D Materials for Optoelectronics Science and Technology of Ministry of Education, Institute of Microscale Optoelectronics, Shenzhen University, Shenzhen, China*

### Edited by:

*Feixiang Wu, Central South University, China*

### Reviewed by:

*Xianwen Wu, Jishou University, China Xunhui Xiong, South China University of Technology, China*

> \*Correspondence: *Renheng Wang wangrh@szu.edu.cn*

### Specialty section:

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

Received: *14 May 2019* Accepted: *22 July 2019* Published: *06 August 2019*

### Citation:

*Li Y, Zhang H, Xiao Z and Wang R (2019) Flexible Li[Li*0.2*Ni*0.13*Co*0.13*Mn*0.54*]O*2*/Carbon Nanotubes/Nanofibrillated Celluloses Composite Electrode for High-Performance Lithium-Ion Battery. Front. Chem. 7:555. doi: 10.3389/fchem.2019.00555*

Rapidly-growing demand for wearable and flexible devices is boosting the development of flexible lithium ion batteries (LIBs). The exploitation of flexible electrodes with high mechanical properties and superior electrochemical performances has been a key challenge for the rapid practical application of flexible LIBs. Herein, a flexible composite electrode was prepared from the mixed solutions of Li[Li0.2Ni0.13Co0.13Mn0.54]O<sup>2</sup> (LLOs), carbon nanotubes(CNTs), and nanofibrillated celluloses (NFCs) via a vacuum filtration method. The resulting LLOs/CNTs/NFCs electrode delivered an initial discharge capacity of 253 mAh g−<sup>1</sup> at 0.1 C in the voltage range from 2.0 to 4.6 V, and retained a reversible capacity of 178 mAh g−<sup>1</sup> with 83% capacity retention after 100 cycles at 1 C. The LLOs/CNTs/NFCs electrode exhibited excellent flexibility along with repeated bending in the bending test. The LLOs/CNTs/NFCs electrode after bending test remained a discharge capacity of 149 mAh g−<sup>1</sup> after 100 cycles at 1 C, and the corresponding capacity retentions was 76%. The excellent electrochemical performance and high flexibility can be ascribed to the framework formed by CNTs with high conductivity and NFCs with good mechanical properties. The results imply that the as-fabricated electrode can be a promising candidate for the flexible LIBs.

Keywords: lithium ion battery, flexible electrode, Li[Li0.2Ni0.13Co0.13Mn0.54]O2 , carbon nanotubes, nanofibrillated celluloses

### INTRODUCTION

Along with the rise and development of the wearable and flexible electronic products, the concept of flexible devices emerges at the right moment and engulfes the entire world (Manthiram et al., 2008; Kim and Cho, 2010; Agnès et al., 2014). One of the biggest obstacles to the application of flexible electronic products is the development of flexible electrochemical energy storage devices (Ju et al., 2014; Li et al., 2014). LIBs have unique advantages when applied to electronic devices due to its light weight, high specific power, high energy density, and excellent electrochemical performance (Ding et al., 2011; Dunn et al., 2011; Strong et al., 2012; Yan et al., 2013; Yuan et al., 2014; Nitta et al., 2015; Li et al., 2017a,b; Yang et al., 2019). However, the separation of electrode materials and collector fluid affects the electrochemical properties when fold the traditional rigid LIBs, and even lead to short circuit and other serious safety problems (Hu and Sun, 2014; Zhou et al., 2014; He et al., 2016). Therefore, it is of great significance for the development of flexible electronics to endow LIBs with flexibility and develop flexible electrode with excellent electrochemical performance. A well-known strategy to prepare flexible electrode is adopting flexible organic substrates such as polymer and textile instead of traditional copper/aluminum foil as current collector for carrying the active materials (Wang et al., 2006, 2008; Chen et al., 2007; Gao et al., 2015; Huang et al., 2016; Zhen et al., 2016). However, these non-electroactive substrates with low conductivity will greatly affect the quick charge-discharge process of batteries. Therefore, flexible electrodes with conductive matrix such as graphite paper and carbon nanotube film have become a hot research topic, while the weak binding force between active materials and current collector will inevitably result in serious capacity fading and limit further applications (Chou et al., 2008; Xin et al., 2011; Hu et al., 2013). Recently, many researchers had proposed to build the 3D thin-film electrodes which compounding the traditional active electrode material with the conductive carbon-based material and flexible matrix. The carbon-based materials acted as the basic element of the conductive network, and the flexible matrix served on the supporting framework of the whole electrode. This method can fully combine the advantages of carbon-based materials and traditional active electrode materials (Su et al., 2010; Tai et al., 2012; Lian et al., 2014; Yu et al., 2014; Wang et al., 2017b).

Lithium-rich layered oxides (LLOs) have been extensively investigated owing to their high reversible capacity, high working voltage, high energy density, lost cost, and amity to environment (Wang et al., 2009; He et al., 2013; Hu et al., 2018). With a reversible capacity of 250 mAh g−<sup>1</sup> , LLOs were attractive enough to high-energy LIBs and have demonstrated enormous potential in terms of electric vehicles and smart grid applications. As one of the most abundant renewable polymers, nanofibrillated cellulose (NFC) has several advantages such as nanoscale diameter, high homogeneity, high tensile strength and high Young modulus. The one dimension nano-structure is beneficial to bind active electrode particles in the nanometer scale, and allows for the formation of thinner films with higher space efficient (Ho et al., 2011). NFC with excellent mechanical properties can be utilized as flexible matrix and binder for composite electrodes, and has great prospects in flexible energy storage devices. For example, a flexible NFC/LiFePO4/Super-P paper electrode was prepared by filtration process. NFCs acted both as electrode binder material and the separator material. The resulting thin electrode was very strong which can withstand of strength up to 5.6 Mpa. The flexible electrode delivered a reversible capacity of 146 mAh g−<sup>1</sup> and an energy density of 188 mWh g−<sup>1</sup> of full paper battery at 0.1 C (Leijonmarck et al., 2013). Recently, Wang et al. developed a flexible LiFePO4/graphene/NFC electrode with the mechanical support of NFC, and the composite electrode shows high flexibility as well as excellent electrochemical properties (Wang et al., 2018). Carbon-based materials with high conductivity were highly desirable for high-performance LIBs. Many studies have proved that CNTs can improve the electronic conductivity of electrode and offer the buffer spaces for the volume expansion of active material during the cycling process (Chen et al., 2007; Fu et al., 2013; Kang et al., 2016). Thanks to the graphitic structure and good electrical conductivity, CNTs has the capability of forming conductive network structure and is very suitable for preparing self-supporting film electrodes (Bai et al., 2015; Wan et al., 2015). For instance, a binder-free LiMn2O4/CNT electrode with good flexibility were fabricated by in-situ hydrothermal growth for flexible LIBs, and the flexible electrode showed good cycling and rate performance (Jia et al., 2011).

Herein, a flexible LLOs/CNTs/NFCs composite electrode was fabricated via a facile vacuum filtration route. The resulting LLOs/CNTs/NFCs composite electrode exhibited good electrochemical performance as well as high flexibility, which

can be ascribed to the three-dimensional conductive framework formed by NFCs and CNTs.

## EXPERIMENTAL

### Synthesis and Characterization

LLOs were prepared from the Mn0.54Ni0.13Co0.13(CO3)0.8 precursors (Ningbo Institute of Industrial Technology, CAS.) and lithium carbonate via solid state reaction. Commercially available CNTs (wt % > 95%, Shanghai Macklin Biochemical Co., Ltd.) and NFCs (solid content = 5%, width = 5–20 nm, length ≈ 400µm, Ningbo Rouchuang Nanotechnology Co., Ltd.) were used without further processing. As shown in **Figure 1**, the flexible LLOs/CNTs/NFCs electrode was fabricated via a facile vacuum filtration route. Proper amount of LLOs, CNTs and NFCs with a mass ratio of 8:1:1 were dispersed in 100 ml ethanol and then stirred for 20 h, the resulting LLOs/CNTs/NFCs suspension was filtered using a PTFE membrane (0.22 lm pore size, Tianjin Jinteng Experiment Equipment Co., Ltd.). At last, the LLOs/CNTs/NFCs electrode was prepared from the filter cake after drying and cutting. The diameter of the obtained electrode was 10 mm and the weight of the active materials was about 8 mg.

For comparison, the traditional LLOs electrode with aluminum foil as current collector was also prepared according to the procedures reported earlier (Li et al., 2018).

Morphologies of the samples were investigated with scanning electron microscopy (SEM, Gemini300). The structure of the LLOs/CNTs/NFCs electrode were determined by X-ray diffraction (XRD, Bruker D8) equipped with Cu Kα radiation ranging from 10◦ to 80◦ at a scan rate of 10◦ ·min−<sup>1</sup> . Raman spectroscopy (FSRS, Renishaw in Via Reflex) was carried out to analysis of the electrode using a laser wavelength of 532 nm.

## Electrochemical Measurement

The as-prepared LLOs/CNTs/NFCs electrode and the traditional LLOs electrode acted as the working electrode for the electrochemical measurements of LIBs. Li metal foil, microporous polypropylene film and the liquid organic electrolytes [1 M LiPF<sup>6</sup> in EC/DMC/EMC (1:1:1 by volume)] were used as the counter electrodes, separator and electrolyte, respectively. The charge/discharge cycling was carried out in the potential range of 2.0–4.6 V using a CR2032 coin-type cell. The LLOs/CNTs/NFCs electrodes were bended repeatedly with 5 mm curvature radius for different times. The electrochemical performances of the LLOs/CNTs/NFCs electrodes after cyclical bending were tested to investigate the application prospects for the flexible electrode. Electrochemical impedance spectroscopy (EIS) was conducted using a CHI660e electrochemical workstation in the frequency range of 10−<sup>2</sup> -10<sup>5</sup> Hz.

### RESULTS AND DISCUSSION

**Figure 2A** shows the XRD patterns of the flexible LLOs/CNTs/NFCs electrode. The sample was consistent with the well-crystallized LLOs which adopts a hexagonal α-NaFeO<sup>2</sup> type structure. It's worth noting that the characteristic peak (marked by the red box) can be assigned to the (002) peak of NFCs (22.5◦ ) and CNTs (Sassi and Chanzy, 1995; Tessonnier et al., 2009). FSRS in **Figure 2B** shows two distinct Raman peaks at 1350 and 1588 cm−<sup>1</sup> , which corresponded to the D-band and G-band of CNTs, respectively (Dresselhaus et al., 2007). The other evident peak centered at 1,095 cm−<sup>1</sup> was assigned to the asymmetric stretching vibration mode of C–O in the NFCs (Leijonmarck et al., 2013). These results revealed that the as-prepared electrode was composed of LLOs and small amount of NFCs and CNTs, and the crystalline structure of LLOs were maintained well in the composite electrode. It can be expected that the introduction of NFCs and CNTs could enhance the overall flexibility and conductivity of the composite electrode.

SEM images of the samples and photographs of the LLOs/CNTs/NFCs electrode were presented in **Figure 3**. **Figure 3a** shows the morphology of LLOs particles, the particles have a micro-grade spherical shape with a diameter ranging from 5 to 20µm, and the microspheres were formed by numerous primary particles. SEM image in **Figure 3c** indicates that CNTs remained a regular nanotube shape with a diameter in the range of 10–20 nm. **Figures 3c,d** shows the SEM images of the LLOs/CNTs/NFCs electrode in different magnification. The LLOs particles, CNTs and NFCs could be clearly observed in **Figure 3e**. It was evident that the micro-sized LLOs particles

LLOs/CNTs/NFCs electrode before and being bending.

were tightly compacted and agglutinated by the frame of NFCs and CNTs, and thus forming the flexible composite electrode. The cross-section images in **Figures 3f,g** indicate that the thickness of the flexible electrode was about 200µm, and the LLOs particles, CNTs and NFCs were uniform mixed and distributed in the flexible electrode. The photos of the LLOs/CNTs/NFCs electrode with a diameter of 10 mm illustrate its structural integrity and high flexibility, as seen in **Figure 3h**. The surface of the LLOs/CNTs/NFCs electrode was verified to be smooth and level. The structural integrity the electrode can be attributed to the entanglement of NFCs and CNTs. The photo of the electrode being bending illustrates that the as-prepared LLOs/CNTs/NFCs electrode demonstrated good flexibility.

The electrochemical performances of the traditional LLOs electrode and LLOs/CNTs/NFCs electrode were presented in **Figure 4**. **Figure 4A** shows the first cycle curves of the two electrodes at 0.1C rate, the traditional LLOs electrode and the LLOs/CNTs/NFCs electrode exhibite discharge capacities of 253 and 259 mAh g−<sup>1</sup> , respectively. They show a similar initial coulombic efficiency of about 76%. The cycle performances of the two electrodes were presented in **Figure 4B**. After 100 cycles at 1 C rate, the discharge capacities of the traditional LLOs electrode and the LLOs/CNTs/NFCs electrode remain 165 and 178 mAh g−<sup>1</sup> , respectively. The corresponding capacity retentions were 77 and 83%, respectively. Also notice that both the two samples exhibite nearly 100% coulombic efficiency. As depicted in **Figure 4C**, the LLOs/CNTs/NFCs electrode shows better rate capability than that of the traditional LLOs electrode. When the current density was 2 C, the discharge capacity of the traditional LLOs electrode and the LLOs/CNTs/NFCs electrode were 157 and 168 mAh g−<sup>1</sup> , respectively. The results indicate that the flexible LLOs/CNTs/NFCs electrode exhibites better reversibility and superior rate capability than the traditional LLOs electrode under the same condition. **Figure 4D** shows the EIS plots, the fitting results and the corresponding equivalent circuit of the two samples. The curves consist of a semicircle and a straight line. The semicircle in the high and medium frequency can be explained as the charge transfer resistance (Rct) and double layer capacitance (Cdl), and the straight line in the low frequency was a representation of the Warburg impedance Z<sup>w</sup> (Wu et al., 2019). By comparing the two curves, it was found that the traditional LLOs electrode has higher interfacial impedance compared with the LLOs/CNTs/NFCs electrode. The lower impedance of the flexible LLOs/CNTs/NFCs electrode can

Frontiers in Chemistry | www.frontiersin.org

be attributed to fast charge transfer and high electrochemical activity provided by the continuous conductive network. This also demonstrates the reasons for the better cycle and rate performances of the LLOs/CNTs/NFCs electrode.

The SEM images of the LLOs/CNTs/NFCs electrode after 100 cycles at 1 C were illustrated in **Figure 5**. The lowmagnification image in **Figure 5a** reveals that the integrity of the electrode was maintained well and the LLOs particles keeped the original spherical shape. It was clear from **Figure 5b** that the dense and uniform SEI film was covered to the surface of LLOs microspheres (Wang et al., 2017a). The result indicates the dominant structure stability of the LLOs/CNTs/NFCs electrode, and further proves the foregoing reasons for the good electrochemical performances of the flexible LLOs/CNTs/ NFCs electrode.

In order to investigate the practical prospect of the flexible LLOs/CNTs/NFCs electrode, electrochemical performances of the flexible electrode were measured after bending. As exhibited in **Figure 6A**, the LLOs/CNTs/NFCs electrode was bent with a 5 mm curvature radius for 0–50 times. For the LLOs/CNTs/NFCs electrode, the discharge capacity and coulombic efficiency in the first cycle at 0.1 C decreased with the increase of bending times. The incipient discharge capacity of the flexible electrode after 50 times bending was 231 mAh g−<sup>1</sup> , and the capacity retention rate was 91% after bending for 50 times. At the same time, the initial coulombic efficiency reduces from 76 to 72% after 50 times bending. The cycle performance and coulombic efficiency of the LLOs/CNTs/NFCs electrode after bending for 50 times was depicted in **Figure 6B**. After 100 cycles at 1 C, the LLOs/CNTs/NFCs electrode delivered a discharge capacity of 149 mAh g−<sup>1</sup> , the corresponding capacity retentions was 76%, and the coulombic efficiency keeped nearly 100%. These tests manifest that the as-prepared LLOs/CNTs/NFCs electrode had high structure stability and flexibility, and the electrochemical properties declined slight after repeated bending. In the LLOs/CNTs/NFCs electrode, NFCs with extraordinary mechanical properties is beneficial to bind active electrode particles and form a flexible freestanding electrode (Nakagaito and Yano, 2004; Aulin et al., 2012). CNTs with superior mechanical properties and electrical conductivity can greatly enhance the mechanical strength and electrochemical activity of the active materials (Hu et al., 2012; Peng et al., 2014). Therefore, the improvement of mechanical and electrochemical performances of the LLOs/CNTs/NFCs electrode can be attributed to the continuous conductive network constructed by NFCs and CNT.

### CONCLUSIONS

In summary, the flexible LLOs/CNTs/NFCs electrode was constructed via the vacuum filtration method. The resulting LLOs/CNTs/NFCs electrode exhibited enhanced initial discharge capacity, superior cycle stability and improved rate performance in contrast to the traditional LLOs electrode. The LLOs/CNTs/NFCs electrode exerted high flexibility during the repeated bending processes, and the electrochemical performance performance degraded less after bending. The good electrochemical performances and flexibility can be attributed to the conductive framework formed by NFCs with good mechanical properties and CNTs with high conductivity. This study provides a potential strategy to fabricated flexible electrode for LIBs with good mechanical properties and electrochemical performances.

### DATA AVAILABILITY

All datasets generated for this study are included in the manuscript/supplementary files.

### AUTHOR CONTRIBUTIONS

YL and RW designed and engineered the samples. YL performed the experiments. YL, HZ, ZX, and RW performed the data analysis. YL and RW wrote the paper. All authors contributed to the theoretical analysis and the general discussion.

### FUNDING

The work was supported by the National Science Foundation for Young Scientists of China (Grant No. 51804199), the Science and Technology Innovation Commission of Shenzhen

### REFERENCES


(Grant No.20180123), and the National Natural Science Fund (Grant No. 61435010).

### ACKNOWLEDGMENTS

Authors also acknowledge the supports from Instrumental Analysis Center of Shenzhen University (Xili Campus) and Shenzhen Institute of Information Technology.


for Lithium-Sulfur Batteries. Adv. Funct. Mater. 24, 2772–2781. doi: 10.1002/adfm.201303296


anode for advanced lithium ion batteries. Appl. Surf. Sci. 423, 375–382. doi: 10.1016/j.apsusc.2017.06.129


**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 Li, Zhang, Xiao and Wang. 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.

# Synthesis and Electrochemical Performance of ZnSe Electrospinning Nanofibers as an Anode Material for Lithium Ion and Sodium Ion Batteries

Peng Zhou<sup>1</sup> , Mingyu Zhang<sup>1</sup> \*, Liping Wang<sup>2</sup> , Qizhong Huang<sup>1</sup> , Zhean Su<sup>1</sup> , Liewu Li <sup>1</sup> , Xiaodong Wang<sup>1</sup> , Yuhao Li <sup>1</sup> , Chen Zeng<sup>1</sup> and Zhenghao Guo<sup>1</sup>

*<sup>1</sup> State Key Laboratory of Powder Metallurgy, Central South University, Changsha, China, <sup>2</sup> Department of Biological and Environmental Engineering, Changsha University, Changsha, China*

ZnSe nitrogen-doped carbon composite nanofibers (ZnSe@N-CNFs) were derived as anode materials from selenization of electrospinning nanofibers. Electron microscopy shows that ZnSe nanoparticles are distributed in electrospinning nanofibers after selenization. Electrochemistry tests were carried out and the results show the one-dimensional carbon composite nanofibers reveal a great structural stability and electrochemistry performance by the enhanced synergistic effect with ZnSe. Even at a current density of 2 A g−<sup>1</sup> , the as-prepared electrodes can still reach up to 701.7 mA h g−<sup>1</sup> after 600 cycles in lithium-ion batteries and 368.9 mA h g−<sup>1</sup> after 200 cycles in sodium-ion batteries, respectively. ZnSe@N-CNFs with long cycle life and high capacity at high current density implies its promising future for the next generation application of energy storage.

Keywords: anode material, Li-ion and Na-ion batteries, electrospinning nanofibers, ZnSe, synthesis, electrochemical performance

# INTRODUCTION

Over the latest 20 years, lithium-ion batteries (LIBs) have experienced great development to meet the demand of portable electronic devices and hybrid electric vehicles (Mai et al., 2010; Ji et al., 2012; Li W. et al., 2016; Wu F. et al., 2017; Tian et al., 2018). However, the theoretical capacity (372 mA h g −1 ) of the conventional graphite anode in LIBs can't meet the increasing expectations (Li L. et al., 2016; Zhang Y.-C. et al., 2016; Lee et al., 2018). At the same time, Sodium-ion batteries (SIBs), as one of the most competitive alternatives of LIBs, are drawing much attention due to the low cost and high abundance of sodium in the crust (Li et al., 2017; Zhang et al., 2017; Wu et al., 2018). Unfortunately, the larger ionic diameter of Na<sup>+</sup> (0.106 nm) compared with Li<sup>+</sup> (0.076 nm) results in the kinetic limitation and larger volume expansion of anode materials, which further leads to reduced capacity in SIBs (Xu et al., 2015; Deng et al., 2018). Therefore, it is a topmost priority to develop high-performance anode materials for LIBs and SIBs (Nitta et al., 2015; Du et al., 2018; Fan and Xie, 2019). Recently, metallic selenides (SnSe2, FeSe2, CoSe2, etc.) have attracted much attention as anode materials thanks to their high energy density and excellent rate performance (Ko et al., 2016; Park et al., 2016; Zhang L. et al., 2016; Cui et al., 2018). Among them, ZnSe is considered as one of the most promising anode materials because of its impressive performance in both LIBs and SIBs (Cao et al., 2018). Particularly, when ZnSe was used

### Edited by:

*Danni Lei, Sun Yat-Sen University, China*

### Reviewed by:

*Baihua Qu, Xiamen University, China Ming Zhang, Hunan University, China*

### \*Correspondence: *Mingyu Zhang*

*zhangmingyu@csu.edu.cn*

### Specialty section:

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

Received: *22 May 2019* Accepted: *26 July 2019* Published: *14 August 2019*

### Citation:

*Zhou P, Zhang M, Wang L, Huang Q, Su Z, Li L, Wang X, Li Y, Zeng C and Guo Z (2019) Synthesis and Electrochemical Performance of ZnSe Electrospinning Nanofibers as an Anode Material for Lithium Ion and Sodium Ion Batteries. Front. Chem. 7:569. doi: 10.3389/fchem.2019.00569*

**69**

as anode materials in LIBs, the Zn reduction by ZnSe could react with Li<sup>+</sup> to form LiZn and provide additional capacity (Kwon and Park, 2014; liu et al., 2018). However, the pulverization and amorphization of ZnSe during charge and discharge result in poor cycling stability (Fu et al., 2015).

To overcome the problems mentioned above, constructing ZnSe/carbon hybrid material is considered as one of the most effective way to improve electrochemical performance of electrodes. ZnSe shows great synergistic effect with carbon, which highly improves the capacity of anode in energy storage (Zhang et al., 2015). For example, Chen et al. (2017) reported that ZnSe ND@N-PC by using zeolitic imidazolate framework (ZIF-8) and delivered an outstanding capability to LIBs of 1,134 mA·h·g −1 at 0.6 A·g −1 after 500 cycles. Tang et al. (2018) synthesized a ZnSe microsphere/multiwalled carbon nanotube composite used as SIBs anode materials, which exhibited a high specific capacity of 382 mA h g−<sup>1</sup> at 0.5 A g−<sup>1</sup> after 180 cycles. Cao et al. dispersed ZnSe nanoparticles in reduced graphene oxides to synthesize ZnSe-rGO nanocomposite as an anode material for both LIBs and SIBs. The capacity of ZnSe-rGO in LIBs is 530 mA h g−<sup>1</sup> at 0.5 A g−<sup>1</sup> after 100 cycles and that of SIBs is 259.5 mA h g−<sup>1</sup> at 0.1 A g−<sup>1</sup> after 50 cycles (Cao et al., 2018). But the reported ZnSe/carbon electrodes are mainly microspheres or irregular nanoparticles (Xu Y. et al., 2016). The composite mode and microstructure of ZnSe and carbon are inefficient and then require further design. The specific capacity and stability of the ZnSe/carbon hybrid anode, especially at high current density, also need to be improved (Wang et al., 2017).

In this work, we have successfully synthesized ZnSe nitrogendoped carbon composite nanofibers (ZnSe@N-CNFs) for both LIBs and SIBs via electrospinning and a simple selenization treatment. This unique one-dimensional (1D) nanostructure has a shorter ion diffusion path and higher electronic conductivity. In the meanwhile, the synergistic effect of encapsulated ZnSe nanoparticles in nitrogen-doped carbon nanofibers can effectively suppress the pulverization and amorphization. As a result, the as-prepared ZnSe@N-CNFs electrodes exhibit an excellent electrochemical performance as anode material for both LIBs and SIBs. The specific capacity of ZnSe@N-CNFs reach to 1,226.1 and 455.0 mA h g−<sup>1</sup> in LIBs and SIBs, respectively. Meanwhile, the ZnSe@N-CNFs also shows outstanding specific capacity and stability (701.7 mA h g−<sup>1</sup> after 600 cycles in LIBs and 365.6 mA h g−<sup>1</sup> after 200 cycles in SIBs) at high current density of 2 A g−<sup>1</sup> . It is expected that the ZnSe@N-CNFs with such great electrochemical performance have promising applications as anodes for both LIBs and SIBs, and would be a direction for design of the other anode materials.

### EXPERIMENTAL SECTION

### Preparation of ZnSe@N-CNFs

0.5268 g of Zinc acetate dihydrate (C4H6O4Zn·2H2O, AR, Sinopharm) was dissolved in 3 ml N, N-dimethylformamide (DMF, AR, Sinopharm). 0.4534 g polyacrylonitrile (PAN, Mw 150000, Macklin) was dissolved in 3 ml DMF under magnetic stirring for 30 min at 60◦C. The two solutions were mixed and stirred for another 12 h. Then the mixture was transferred into

a syringe to electrospin with the distance 20 cm and voltage 13 kV, respectively. The obtained nanofibers were mixed with 1.3 g Selenium powder (99.9%, Aladdin) and calcined at 650◦C for 2 h (a ramp rate of 10◦C min−<sup>1</sup> ) in a tube furnace under vacuum to produce the ZnSe@N-CNFs.

In the meantime, carbon nanofibers (CNF) and ZnSe@N-C were prepared as comparison samples. The carbon nanofibers (CNFs) were synthesized in the same steps without adding Zinc acetate dihydrate. The ZnSe@N-C was synthesized by precursor solution directly dried at 80◦C overnight and calcined at the same way with Se powder.

### Materials Characterization

The crystal structure and the composition of the samples were investigated by X-ray diffraction (XRD, Rigaku Dmax/2550VB + 18 kW) and energy dispersive X-ray spectroscopy (EDX, FEI Nova Nano SEM230). The morphology and microstructure of samples was observed through scanning electron microscopy (SEM, FEI Nova Nano SEM230) and transmission electron microscopy (TEM, JEOL JEM-2010). The thermogravimetric analysis (TGA) was tested in air at a ramp rate of 10◦C min−<sup>1</sup> . The Raman spectrum test was conducted on a Renishaw in Via 2000. The specific surface areas and the pore size distribution were measured by the Brunauer-Emmett-Teller (BET) test and Barrett-Joyner-Halenda (BJH) method, respectively. The surface chemical composition of the sample was tested by the X-ray photoelectron spectroscopy (XPS, Thermo Scientific ESCALAB 250XI).

### Electrochemical Characterization

A slurry made of ZnSe@N-CNFs (80 wt.%), acetylene black (10 wt.%), and carboxymethyl cellulose sodium (10 wt.%) was dissolved in deionized water and ethanol (3:2), spread onto Cu foil and dried at 80◦C for 12 h to prepare the anodes. The LIBs were assembled into 2,032 coin-type cells in an Argonfilled glove box with lithium metal as the reference electrode,

LiPF6 (1M) in ethylene carbonate (EC) and dimethyl carbonate (DMC) at a 1:1 volume ratio as electrolyte and polypropylene film (Celgard 2400) as separator. The SIBs were assembled with sodium metal as the reference electrode, NaCF3SO3 (1M) in diethyleneglycol dimethylether (DEGDME) used as electrolyte and glass microfiber (Whatman GF/D) as separator.

All electrochemical tests were carried out at 27◦C. The cyclic voltammetry (CV) was measured with voltage window of 0.01– 3.0 V by using CHI760E electrochemical workstation. The rate capability and cycle life were tested with LAND CT2001A battery test system. The electrochemical impedance spectroscopy (EIS) tests were carried out with the frequency range of 100 kHz−0.01 Hz by CHI760E electrochemical workstation.

### RESULTS AND DISCUSSION

As shown in XRD pattern of **Figure 1** and **Figure S1**, all the peaks of ZnSe@N-CNFs and ZnSe@N-C can be fully indexed to ZnSe (JCPDS 37-1463), indicate the effective formation of ZnSe as we design (Tang et al., 2018). The wide peak shown at 30◦ of CNFs is attributed to the amorphous carbon (Lallave et al., 2007). The carbon peaks aren't obvious in the diffraction pattern of ZnSe@N-CNFs and ZnSe@N-C as the highly crystalline ZnSe reflections suppressed those of carbon.

**Figures 2a,b** and **Figures S2, S3** indicate the SEM image of assynthesized samples. It can clearly reveal that the ZnSe@N-CNFs and CNFs are continuously fiber structure without aggregated particles. In contrast, ZnSe@N-C is irregular particles. To further reveal the microstructure of ZnSe@N-CNFs, TEM, and HRTEM were used and the images were shown in **Figures 2c,d**. It is clearly observed in **Figure 2c** that the morphology of ZnSe@N-CNFs is composed of composite nanofibers of about 200 nm diameter with the ZnSe particles dispersed in the fibers or deposited on their surface. The phenomenon of deposited ZnSe particles on the surface is attributed to the crystal growth during calcined (Ning et al., 2017). Clear lattice fringes can be observed from the HRTEM image in **Figure 2d** of 0.32 nm corresponding to the (111) plane of ZnSe crystal (JCPDS 37-1463). In addition, the EDX analysis is shown in **Figure S4**. The atomic ratio of Zn and Se in ZnSe@N-CNFs is about 1:1, matching the stoichiometric ratio of ZnSe compounds. The nitrogen is mainly from PAN, which would increase the conductivity and the number of active sites (Cho et al., 2016).

The TGA of ZnSe@N-CNFs was tested from room temperature to 800◦C. As shown in **Figure 3A**, the huge weight loss around 300–600◦C correspond to the volatilization of SeO<sup>2</sup> and CO<sup>2</sup> (Cui et al., 2017). The mass percentages of ZnSe in ZnSe@N-CNFs is calculated to be 49.45%. In addition, **Figure 3B** gives the Raman spectra of samples. The two wide peaks can be observed at around 1,340 and 1,586 cm−<sup>1</sup> , which are corresponding with disordered carbon atoms (D band) and graphitic carbon atoms (G band) of carbon, respectively (Wu Q. et al., 2017). The ID/I<sup>G</sup> ratio of the ZnSe@N-CNFs was calculated to be 1.16. The peaks observed at around 500 cm−<sup>1</sup> in ZnSe@N-CNFs and ZnSe@N-C spectra correspond to 2LO modes of ZnSe (Tang et al., 2018). To further compare

the pore size of ZnSe@N-CNFs and ZnSe@N-C, the specific surface was determined by nitrogen adsorption/desorption and the pore size distribution curve calculate by BJH method as shown in **Figure S5**. The specific surface area of ZnSe@N-CNFs is 30.2 m<sup>2</sup> g −1 , higher than 8.1 m<sup>2</sup> g <sup>−</sup><sup>1</sup> of ZnSe@N-C. The larger specific surface area is attributed to the unique 1D nanostructure. The pore diameter of ZnSe@N-CNFs and ZnSe@N-C are focused on around 3.5–3.8 nm. In addition, the number of pores in ZnSe@N-CNFs is much larger than ZnSe@N-C as shown in **Figure S5**. Although ZnSe@N-CNFs isn't designed as a porous material, it still shows an excellent specific surface area and suitable mesopore size, which can provide more ion storage active sites and reducing ion migration path (Li et al., 2019).

To investigate the surface physicochemical properties and the chemical composition of ZnSe@N-CNFs, the XPS was measured and the survey spectra of ZnSe@N-CNFs is shown in **Figure S6**. The peaks corresponding to Zn 2p, Se 3d, C 1s, and N 1s can be clearly observed, which further suggests the presence of Zn, Se, C, and N elements in ZnSe@N-CNFs. As the high resolution of Zn 2p XPS spectrum shown in **Figure 3C**, two peaks are located at 1,021.5 and 1,044.6 eV, respectively, with an energy difference of 23.1 eV between them, which confirms that the zinc exists as Zn2<sup>+</sup> form (Ning et al., 2017). The fitted peaks shown in Se 3d spectrum (**Figure 3D**) at 54.2 and 56.1 eV are corresponding to the Se 3d5/2 and Se 3d3/2 spin orbit, respectively, which indicate the Se mainly exists as Se2−. The peak of SeO<sup>x</sup> locked at 59.0 eV is caused by surface oxidation (Cui et al., 2018). **Figure 3D** indicate that Se in ZnSe@N-CNFs mainly exists as ZnSe and a small amount of Se has been oxidized to SeO<sup>x</sup> on the surface. The three fit peaks of high resolution C1s spectrum (**Figure 3E**) located at binding energy of 284.6, 285.3, and 286.5 eV are related to the C-C bonds, C-N bonds, and C = N bonds, respectively (Liao et al., 2016). Furthermore, the type of nitrogen can be obtained by analyzing the N 1s high-resolution spectrum in the **Figure 3F**. The fit peaks positioned at 398.3, 400.1, and 400.9 eV can be fit well with pyridinic N, pyrrolic peak N, and graphitic peak N, respectively (Wang et al., 2016).

In view of the special compositional advantages and appealing micro-structures of continuously composite nanofibers ZnSe@N-CNFs as we design as discussed above, these properties should be beneficial for both LIBs and SIBs. The Li-storage and Na-storage properties of samples were further measured to prove the potential of ZnSe@N-CNFs as anodes material.

**Figure 4** shows the electrochemical performance of ZnSe@N-CNFs as anodes in LIBs. The CV measurements were conducted at 0.2 mV s−<sup>1</sup> . As shown in **Figure 4A**, the peak at 0.6 V in the initial cathodic process is attributed to the reduction from ZnSe to Zn and the formation of a solid electrolyte interface (SEI) layer (Lu et al., 2017). During the follow scan, the curves lapped well, indicating an excellent cycle reversibility. The sharp cathodic peak at around 0.8 V is ascribed to the reduction reaction of Zn2<sup>+</sup> to Zn<sup>0</sup> . After that, a series of small peaks observed at around 0.5 V are corresponding to the multi-step of forming LiZn alloy (Fu et al., 2016). In addition, the pair of small peaks at 1.6 and 2.3 V are ascribed to the formation and decomposition of Li2Se, respectively (Xu Y. et al., 2016). The reaction of ZnSe@N-CNFs anode in LIBs could be described as the following chemical equations (Kwon and Park, 2014):

$$\text{ZnSe} + 2\text{Li}^+ + 2\text{e}^- \begin{array}{c} \leftrightarrow \text{ Zn} \end{array} + \text{Li}\_2\text{Se} \tag{1}$$

$$\text{Zn} + \text{Li}^+ + 2\text{e}^- \leftrightarrow \text{LiZn} \tag{2}$$

**Figure 4B** shows the discharge and charge profiles of ZnSe@N-CNFs anode at 0.1 A g−<sup>1</sup> at first four cycles. The voltage window of LIBs is 0.005–3.0 V (vs. Li/Li+). For the initial discharge curve, the extended charge plateau at around 0.8 V can be observed, which is due to the decomposition of ZnSe and the formation of the SEI layer. The initial discharge capacity and coulombic efficiency of ZnSe@N-CNFs are 984.7 mA h g −1 and 81.3%, respectively. For the next three cycles, the charge/discharge profiles show a similar shape, which the charge/discharge plateau is consistent with the CV curves. The coulombic efficiency improves remarkably from the second cycle, and reaching to 97.6 % for the third cycle and 99.1% for the 4th cycle.

**Figure 4C** depicts the rate performance of ZnSe@N-CNFs, ZnSe@N-C, and CNFs anode under various current densities from 0.1 to 5 A g−<sup>1</sup> . The ZnSe@N-CNFs anode shows an excellent rate performance of 743.0, 730.3, 699.4, 662.8, 645.8, 636.2, 602.6, and 570.9 mA h g−<sup>1</sup> at 0.1, 0.2, 0.5, 1, 1.5, 2, 3, 5 A g−<sup>1</sup> , and recovers to 822.7 mA h g−<sup>1</sup> when the current density reduced back to 0.2 A g−<sup>1</sup> , respectively. In contrast, the capacity of ZnSe@N-C and CNFs anode only has got the values of 43.0 and 26.4 mA h g−<sup>1</sup> at 5 A g−<sup>1</sup> , and recovers to 379.8 and 228.9 mA h g−<sup>1</sup> when the current density back to 0.2 A g−<sup>1</sup> . Because of the simple carbonization and without any additional processing, the CNFs exhibits worse Li<sup>+</sup> storage performance especially at high current density. Furthermore, when the current density is changed to 0.1A g−<sup>1</sup> , the capacity of ZnSe@N-CNFs gradually increases to 1,226.1 mA h g−<sup>1</sup> after 200 cycles, much higher than ZnSe@N-C and CNFs. The high capacities of ZnSe@N-CNFs is attributed to the special 1D structure. The carbon can effectively prevent the agglomeration of ZnSe particles and lead to more interface of ZnSe and carbon which can provide additional pseudocapacitive capacities In addition, the gradual amorphization process of metallic selenides

electrodes during the reversible reactions of anode can provide more interface of ZnSe particle and carbon, which enhance the pseudocapacitive capacities (Gu et al., 2015). More importantly, as shown in **Figure 4D**, the ZnSe@N-CNFs electrode exhibits excellent discharge capacity of 701.7 mA h g−<sup>1</sup> after 600 cycles at 2 A g−<sup>1</sup> , and the coulombic efficiency retained over 96% after the first cycle. However, the ZnSe@N-C and CNFs only achieved 155.5 and 99.4 mA h g−<sup>1</sup> at the same current density after 600 cycles.

**Figure 5** displays the electrochemical performance of ZnSe@N-CNFs, ZnSe@N-C, and CNFs as anodes in SIBs. The CV curves of ZnSe@N-CNFs anode in SIBs are shown in **Figure 5A** at 0.2 mV s−<sup>1</sup> . The obvious peak observed at around 0.4 V in the initial cathodic process is associated with the formation of the SEI layer and insertion of sodium-ion, which is similar to lithium storage (Ge et al., 2015). After that, the following CV curves are well-overlapped, which means the wonderful cycle reversibility of ZnSe@N-CNFs electrodes. The pair of peaks at 0.7 and 1.1 V relate to the transformation of Zn2<sup>+</sup> and Zn<sup>0</sup> . And the pair of peaks at 1.5 and 2.5 V relates to synthesis and decomposition of Na2Se, respectively (Tang et al., 2018). In summary, the reaction of ZnSe@N-CNFs anode in SIBs could be described as following chemical equations (Cao et al., 2018):

$$\text{ZnSe} + 2\text{Na}^+ + 2\text{e}^- \Leftrightarrow \text{Zn} + \text{Na}\_2\text{Se} \tag{3}$$

The cycling performances of ZnSe@N-CNFs at 2 Ag−<sup>1</sup> with voltage window of 0.005–3 V (vs. Na/Na+) were shown in **Figure S7**. The reversible capacity declined rapidly after 30 cycles. According to previous reports, by increasing the cutoff voltage can reduce the irreversible reactions with carbon and decomposition of electrolytes (Cui et al., 2018). Besides, as **Figure 5A** shows, there is no obvious reaction peak at low voltage. Therefore, 0.1–3 V (vs. Na/Na+) was chosen as the voltage window during discharge and charge test for SIBs. The voltage profiles of ZnSe@N-CNFs with 0.1 Ag−<sup>1</sup> for the first four cycles are shown in **Figure 5B**. The first discharge and charge capacities are 441.5 and 362.3 mA h g−<sup>1</sup> , respectively, resulting in a coulombic efficiency of 82.1%. The huge capacity loss is attributed to SEI layers (Park and Kang, 2016). The coulombic efficiency increases rapidly to 100.3% for the second cycle. After the first cycle, the curves coincide well, indicate that the formed SEI layers is very stable and the great cycle reversibility of ZnSe@N-CNFs electrodes in SIBs (Liu et al., 2018).

The rate capability of as-synthesized samples was evaluated at various currents in the range of 0.1–2 A g−<sup>1</sup> , and the results are given in **Figure 5C**. The capacity of ZnSe@N-CNFs is retained as 358.2, 357.9, 346.1, 334.8, 326.7, and 323.3 mA h g−<sup>1</sup> at current densities of 0.1, 0.2, 0.5, 1, 1.5, 2 A g−<sup>1</sup> , respectively. After the current densities back to 0.1 A g−<sup>1</sup> , the capacity is raised to 455.0 mA h g−<sup>1</sup> . Moreover, as shown in **Figure 5D**, the ZnSe@N-CNFs deliver an initial discharge capacity of 446.2 mA h g−<sup>1</sup> with initial columbic efficiency of 72.9% at 2 A g−<sup>1</sup> . And thereafter from the second cycle on, the columbic efficiency is over 99% and the reversible capacity is 365.6 mA h g−<sup>1</sup> after 200 cycles. Both the rate capability and cycle performance are much better than ZnSe@N-C and CNFs. Both high capacity and increasing capacity can be attributed to the pseudocapacitive effect and its growth during cycling, the same as LIBs.

FIGURE 6 | (A) CV curves of the ZnSe@N-CNFs electrode for LIBs at different scan rates; (B) corresponding log(i) vs. log(v) plots for LIBs at each redox peak (peak current: i, scan rate: v) of the ZnSe@N-CNFs electrode; (C) capacitive and diffusion controlled contributions to charge storage of ZnSe@N-CNFs electrode for LIBs at 0.5 mV·s −1 ; (D) normalized contribution ratio of capacitive and diffusion-controlled capacities of ZnSe@N-CNFs electrode for LIBs at different scan rates; (E) CV curves of the ZnSe@N-CNFs electrode for SIBs at different scan rates; (F) corresponding log(i) vs. log(v) plots for SIBs at each redox peak (peak current: i, scan rate: v) of the ZnSe@N-CNFs electrode; (G) capacitive and diffusion controlled contributions to charge storage of ZnSe@N-CNFs electrode for SIBs at 0.5 mV·s −1 ; (H) normalized contribution ratio of capacitive and diffusion-controlled capacities of ZnSe@N-CNFs electrode for SIBs at different scan rates.

No matter in LIBs or SIBs, the ZnSe@N-CNFs all shows satisfactory performance with high capacity and strong cycling stability. To further investigate the reason for the outstanding performance and evaluate the pseudocapacitive behavior of ZnSe@N-CNFs composite electrodes, the CV tests at different scan rates of 0.2–2 mV s−<sup>1</sup> in LIBs and SIBs were conducted, as shown in **Figures 6A,E**. The relation between peak current (i) and scan rate (v) can describe by following two equations (Xu Y. et al., 2016):

$$i = av^b \tag{4}$$

$$
\log(i) = b \log(\nu) + \text{loga} \tag{5}
$$

When the value of the slope b is 0.5, the ion storage process can be considered as diffusion controlled. When the value of slope b is 1, the ion storage process can be considered as capacitive controlled (Tang et al., 2018). The value of b of ZnSe@N-CNFs anode in LIBs and SIBs is calculated and showing in **Figure 6B,F**, respectively. The value of b in each peak is between 0.5 to 1, which can be considered that the ion storage process of ZnSe@N-CNFs anode is controlled by both capacitive and diffusion (Zhou et al., 2019).

According to the relational equation of i = av for capacitive processes and i = av <sup>1</sup>/<sup>2</sup> for diffusion processes, by introducing the adjustable constant parameters as k1and k2, the current i(v) response at fix voltage can be separated to two parts of capacitive and diffusion currents by following equation (Xu D. et al., 2016):

$$i\_{(\nu)} = k\_1 \nu + k\_2 \nu^{1/2} \tag{6}$$

As **Figures 6C,G** shows, the contribution from capacitive capacity of ZnSe@N-CNFs anode in LIBs and SIBs at 0.5 mV·s −1 are calculated to ∼74 and 67%, respectively. With the scan rate increase, the contribution of capacitive gradually increases. As the **Figure 6D** shows, when the scan rate increased to 2 mV·s −1 , the capacitive contributions of ZnSe@N-CNFs anode in LIBs reach to as high as 80%. And as for SIBs shown in **Figure 6H**, the capacitive contribution of ZnSe@N-CNFs anode also reaches to as high as 71% at 2 mV·s −1 .The capacitive contribution for SIBs is lower than that in LIBs, which is attributed to the larger diameter of Na+. The large capacitive contribution of ZnSe@N-CNFs indicates that the unique 1D structure can effectively provide the additional capacity, which explains the high capacity in both LIBs and SIBs (Chen et al., 2016). Moreover, as shown in **Figure S8**, the capacitive contribution increased in both LIBs and SIBs, which proved that the capacity increases along with the cyclic test mainly caused by the capacitive contribution.

The morphology of ZnSe@N-CNFs anodes after cycles have been observed and shown in **Figure 7**. Even after long cycles at large current densities, the ZnSe@N-CNFs in both LIBs and SIBs still remain it's 1D structure without additional agglomeration, while the excellent microstructural stability may explain the great cycles stability of ZnSe@N-CNFs. The ZnSe particles disappeared in **Figure 7** comparing with **Figure 2c** after the long cycles is due to the transformation to amorphous phase and, which can provide more interface for the capacitive behavior.

In addition, the EIS analysis was carried out for LIBs and SIBs of ZnSe@N-CNFs to evaluate the conductivity of anode. **Figures S9, S10** show the Nyquist plots and fitting curve of ZnSe@N-CNFs anode in LIBs and SIBs, respectively (Cui et al., 2018). The R<sup>e</sup> is the electrolyte resistance, the R<sup>f</sup> is the SEI layer resistance and Rct corresponds to the charge transfer resistance. As the fitting results are shown in **Table S1**, the Re, R<sup>f</sup> and Rct of the ZnSe@N-CNFs anode in LIBs is 4.74, 1.86, and 8.82 Ω, respectively. It also can be observed that the ZnSe@N-CNFs anode shows great performance of R<sup>e</sup> (9.55 Ω), R<sup>f</sup> (1.95 Ω), and Rct (10.37 Ω) in SIBs. The impedance parameters in SIBs is slightly larger than in LIBs, indicating the higher kinetic resistance of SIBs. The satisfactory conductivity performance of ZnSe@N-CNFs anode suggests that ZnSe@N-CNFs can provide a short way for electrons and ions transfer, leading to an outstanding electrochemical performance (Miao et al., 2015; Zhao et al., 2017).

### REFERENCES


### CONCLUSION

In summary, the ZnSe@N-CNFs anode was successfully fabricated from 1D electrospinning nanofibers with excellent electrochemical performance in both LIBs and SIBs even at higher current density. The ZnSe@N-CNFs anode delivered a high-capacity of 1,214.0 mA h g−<sup>1</sup> and 447.5 mA h g−<sup>1</sup> in LIBs and SIBs, respectively. Furthermore, even the current density was set to 2 A g−<sup>1</sup> , the ZnSe@N-CNFs electrode delivered still maintained at 701.7 mA h g−<sup>1</sup> after 600 cycles in LIBs and 368.9 mA h g−<sup>1</sup> after 200 cycles in SIBs, respectively. The remarkable performance is attributed to the high capacitive contribution and stable conductive structure, so that the design scope of the metal selenide electrodes could be further expanded.

### DATA AVAILABILITY

All datasets generated for this study are included in the manuscript/**Supplementary Files**.

### AUTHOR CONTRIBUTIONS

MZ supervised the materials synthesis, tests, and manuscript preparation of PZ. All other authors attended part of the work and provided some beneficial advises and discussions on this work.

### FUNDING

The authors gratefully acknowledge support from the Natural Science Foundation of Hunan Province, China (Grant No. 2018JJ2513), and the Key Project of Research Foundation of Education Bureau of Hunan Province, China (Grant No. 18A378).

### ACKNOWLEDGMENTS

The authors also gratefully thank Professor Kejian He for his linguistic assistance during the preparation of this manuscript.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2019.00569/full#supplementary-material


towards long-lifetime anode materials for sodium-ion batteries. J. Mater. Chem. A 6, 7088–7098. doi: 10.1039/c8ta01168k


**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 Zhou, Zhang, Wang, Huang, Su, Li, Wang, Li, Zeng and Guo. 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.

# Toward High-Performance Li Metal Anode via Difunctional Protecting Layer

Jinlei Gu<sup>1</sup> , Chao Shen<sup>1</sup> \*, Zhao Fang<sup>2</sup> , Juan Yu<sup>2</sup> , Yong Zheng<sup>3</sup> , Zhanyuan Tian<sup>3</sup> , Le Shao<sup>3</sup> , Xin Li <sup>1</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 Metallurgical Engineering, Xi'an University of Architecture and Technology, Xi'an, China, <sup>3</sup> Shaanxi Coal and Chemical Technology Institute Co., Ltd, Xi'an, China*

### Edited by:

*Feixiang Wu, Central South University, China*

### Reviewed by:

*Xunhui Xiong, South China University of Technology, China Jia Feng Zhang, Central South University, China Xianwen Wu, Jishou University, China*

### \*Correspondence:

*Chao Shen shenchao@nwpu.edu.cn Xin Li 1814938717@qq.com Keyu Xie kyxie@nwpu.edu.cn*

### Specialty section:

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

Received: *12 June 2019* Accepted: *29 July 2019* Published: *20 August 2019*

### Citation:

*Gu J, Shen C, Fang Z, Yu J, Zheng Y, Tian Z, Shao L, Li X and Xie K (2019) Toward High-Performance Li Metal Anode via Difunctional Protecting Layer. Front. Chem. 7:572. doi: 10.3389/fchem.2019.00572* Li-metal batteries are the preferred candidates for the next-generation energy storage, due to the lowest electrode potential and high capacity of Li anode. However, the dangerous Li dendrites and serious interface reaction hinder its practical application. In this work, we construct a difunctional protecting layer on the surface of the Li anode (the AgNO3-modified Li anode, AMLA) for Li-S batteries. This stable protecting layer can hinder the corrosion reaction with intermediate polysulfides (Li2Sx, 4 ≤ x ≤ 8) and suppress the Li dendrites by regulating Li metal nucleation and depositing Li under the layer uniformly. The AMLA can cycle more than 50 h at 5 mA cm−<sup>2</sup> with the steady overpotential of lower than 0.2 V and show high capacity of 666.7 mAh g−<sup>1</sup> even after 500 cycles at 0.8375 mA cm−<sup>2</sup> in Li-S cell. This work makes great contribution to the protection of the Li anode and further promotes the practical application.

Keywords: AgNO3 , difunctional protecting layer, Li anode, Li-S battery, Li dendrites

# INTRODUCTION

Li metal is the most promising anode material for the next-generation batteries (Li-metal batteries), such as Li-S batteries (Su et al., 2018) and Li-air batteries (Cao et al., 2019), due to the lowest electrode potential (−3.04 V, compare to the hydrogen electrode) and the high capacity (3,860 mAh g −1 , which is 10 times than the commercial graphite) (Liu et al., 2016; Yan K. et al., 2016; Liang et al., 2017; Zhang K. et al., 2017; Bai et al., 2018; Cheng et al., 2018, 2019; Li et al., 2018; Terlicka et al., 2019). However, the intense chemical activity of Li metal leads to sever interface reactions between Li and electrolyte, which results in low Coulomb efficiency and increasing interface impedance (Zheng et al., 2014; Shen et al., 2018). Besides, the inhomogeneous Li deposition leads to forming the dangerous Li dendrites on the Li anode (Tao et al., 2017). During the battery cycling, the Solid Electrolyte Interface (the SEI film) to rupture and re-repair continuously since the uncontrollable Li dendrites. It will trigger the serious side reaction to consume the electrolyte and Li (Hou et al., 2019). More severely, the Li dendrites will pierce through the separator, and cause serious safety hazard (the battery short-circuited, producing a large amount of joule heat and triggering an explosion; Peng et al., 2016; Yang et al., 2017; Zhang et al., 2018; Hou et al., 2019). Thus, the Li anode cannot be commercialized without addressing the above problems (Cheng et al., 2018).

Recently, a variety of approaches are employed to hinder the interface reaction and suppress the Li dendrites to promote the practicality of the Li anode. Firstly, the electrolyte additives are added into electrolyte to increase the stability and dense of the SEI film to suppress the serious interface

**79**

reaction (Zhang, 2012; Jing et al., 2015; Xie et al., 2016; Shiga et al., 2017; An et al., 2019; Chen et al., 2019; Wang et al., in press). However, the SEI film still has limited structure strength, the growth of Li dendrites will destroy it during the cycles. In addition, researchers suppress the Li dendrites through controlling nucleation sites of Li (Zhang et al., 2016, 2018; Jin et al., 2017; Pei et al., 2017; Zhang R. et al., 2017; Hou et al., 2019; Li et al., 2019) and forming a protective layer (Hiratani et al., 1988; Choi et al., 2004; Kozen et al., 2015; Liu et al., 2016, 2018; Peng et al., 2016; Liang et al., 2017; Xie et al., 2017; Bai et al., 2018; Li et al., 2018; Wang et al., 2019) on the Li anode. For example, Liang et al. form a Li-rich composite alloy/LiCl layer on the Li anode (Liang et al., 2017). The Li<sup>+</sup> goes through the protecting layer fast to let Li be deposited under it to suppress the Li dendrites because of the high ionic conduction of the alloy layer. Nevertheless, the uneven deposition of Li is not improved, which means that Li dendrites still are formed. Moreover, Yang et al. struct a 3D host material with ubiquitous and uniform nanoseeds (the ultrafine Ag nanoparticles) to regulate the Li nucleation and deposition homogeneously (Yang et al., 2017). However, this method increases the interface area between electrolyte and Li anode to aggravate the interface reaction and cannot protect the SEI film as the large volume change during the deposition and dissolution process of the Li. Thurs, single method cannot protect the Li anode in the Li-metal batteries well. It is very important to form a protecting layer on the Li anode, which can lead the homogeneous deposition of Li under it and remain the stable SEI film to hinder the interface reaction.

Herein, we form such difunctional protecting layer on the Li anode as the AgNO3-modified Li anode (AMLA) through a simple way, dropping the solution of AgNO<sup>3</sup> on the Li anode. In order to study the effect of this layer, we apply it to the Li-S batteries. This difunctional protecting layer is consisted of the LiAg alloy and the modified SEI film by LiNO3. On the one hand, the layer can suppress the Li dendrites to keep the integrity of the SEI film. The LiAg alloy can regulate Li nucleation and let Li depositing under the protecting layer uniformly to suppress the Li dendrites. On the other hand, the modified SEI film by LiNO<sup>3</sup> can protect the Li anode from the electrolyte to suppress the corrosive reaction between Li and intermediate polysulfides (Jing et al., 2015; Li et al., 2015; Yan C. et al., 2016). The difunctional protecting layer can improve the Li anode through suppressing the Li dendrites and hindering the corrosion reaction in the Li-S batteries.

### MATERIALS AND METHODS

### Synthesis of the Protecting Layer on the Li Anode

The AgNO<sup>3</sup> solution used polar solvent tetrahydrofuran (THF) as the solvent to disperse the AgNO3. Put 50 mg AgNO<sup>3</sup> (AR, ≥99.8%) in 10 ml THF (≥99.8%, HPLC) solvent in the glove box filled with argon (H2O < 0.1 ppm, O < 0.1 ppm). After heating and stirring for 24 h, the solution was ultrasonic disperse until uniform clarification solution is obtained. The template for making the solution is 30◦C and the speed of stirring is 500 rpm min−<sup>1</sup> . Drop 20 µL of prepared solution onto the Li anode through using the pipette. Heat for 3 min to ensure no THF solvent residual with the template of 40◦C.

### Characterizations

X-ray diffraction (XRD) was characterized by STOE PANalytical Empyrean. The experimental conditions are as follows: the radiation is the CuKα radiation and the λ is 1.5406 Å, the scanning Angle is 20–90◦ , the current and voltage are 40 mA and 40 kV, the scanning time is 8 min (the scanning step length is 3◦ , the dwell time is 20 s). X-ray photoelectron spectroscopy (XPS) characterizations were characterized by an ESCALAB250xi XPS system. Energy Dispersion Spectrum (EDS) and Scanning electron microscope (SEM) studies were carried out with FESEM, FEI Tecnai G2 F30. All of the samples were washed by the 1,2 dimethoxyethane (DME) for three times and tested after the DME evaporated.

### Electrochemical Measurements

The CR2016-type coin cells were used to study the electrochemical performance, which assembled with one 0.5 mm metallic gasket in the glove box. Celgard type 2400 polypropylene film was used as separator. The electrolyte in the experiments was that 1 M Li bis(trifluoromethane sulfonyl)imide (LiTFSI) is dissolved in the solvent consisting of 1,2-dimethoxyethane (DME) and 1,3-dioxolane (DOL), where the ratio was 1:1 by volume. There were two kinds of cells (the symmetrical cell and Li-S cell) used to study the electrochemical performance of the protecting layer. The symmetric cells used the Li anode or the AMLA on each side, as shown in the **Figure 4a**. The electrochemical performance of symmetrical cells was measured by symmetrical cycle test and electrochemical impedance spectroscopy (EIS) test. As the symmetrical cycle test, the symmetrical cells were cycled with the Li deposit amount of 0.5 mAh cm−<sup>2</sup> at 1, 2, and 5 mA cm−<sup>2</sup> to make the change of voltage—time curve on a LAND battery system. The frequency range of the EIS test was from 0.01 to 1,000,000 Hz and the amplitude was 10 mV, which was carried out on Solartron electrochemical workstation (1260 + 1287, England).

For the Li-S cell, the S cathode was formation: firstly, forming the slurry through mixing the S, binder (polyvinylidene fluoride, PVDF) and conductive carbon black (super-P acetylene black), at a weight ratio of 8:1:1, in the solution of N-methyl-2-pyrrolidone (NMP); then, casting the slurry on carbon-coated aluminum foil at 200µm with the doctor blade as S cathodes; in the last, vacuum drying the cathode at 60◦C for 10 h and rushing into 12 mm diameter wafer, the plate load per unit area is about 1.0 mg cm−<sup>2</sup> . The Li-S cell was composed of the S cathode and Li anode or the AMLA.

Galvanostatic charge-discharge tests were carried to study the electrochemical performance of the cell, and the current density was 0.5 C (1 C = 1,675 mA g−<sup>1</sup> ) with the voltage windows of 1.7–2.8 V.

FIGURE 1 | The illustration of the synthesis of the AMLA.

# RESULTS AND DISCUSSION

The AMLA is formed by a very simple way, dropping the THF solution of AgNO<sup>3</sup> on the Li anode, as illustrated in **Figure 1**. After the redundant THF evaporates, a thin mixing layer of LiNO<sup>3</sup> and Ag is formed on the Li anode and we can see the color of the Li metal changes to yellow. The mixing layer is made in situ spontaneous reduction of AgNO<sup>3</sup> by Li, due to that the redox potential of Ag/Ag<sup>+</sup> couples (0.8 V, compare to the hydrogen electrode) is much higher than the Li/Li+. And this reaction can

be represented by the following equation:

$$\text{AgNO}\_3 + \text{Li} \rightarrow \text{LiNO}\_3 + \text{Ag} \tag{1}$$

The formation of LiNO<sup>3</sup> and Ag can be confirmed through employing the XRD. The characteristic diffraction peaks of Ag (PDF#01-1167) at 44.50◦ , 77.63◦ , and 81.57◦ corresponding to the (200), (311), and (222) facets appear on the AMLA (**Figure 2b**), comparing the only characteristic diffraction peaks of Li in the **Figure 2a** (Taillades and Sarradin, 2004; Offor et al., 2015). In addition, there are another four characteristic diffraction peaks of LiNO3, which is at 24.82◦ , 32.10◦ , 35.28◦ , and 42.45◦ (**Figure 2b**), corresponding to the (012), (104), (006), and (113) facets in the PDF#01-1225. In addition, the morphologies of the Li anode (**Figure 2c**) and AMLA (**Figure 2d**) are characterized by the SEM. The surface of the Li anode is very smooth, while the AMLA is rough with a lot of Ag nanoparticles. The size of these nanoparticles is smaller than 100 nm and they are homogeneous distribution as confirmed by the EDS of the AMLA (**Figure 2e**). All of these results confirm that we form the mixing uniform layer of LiNO<sup>3</sup> and Ag on the Li anode as the Equation (1).

The electrochemical performances of the cells with the AMLA show much better due to the mixing layer of LiNO<sup>3</sup> and Ag, comparing to the symmetrical cells with the Li anode (**Figure 3**). First of all, the symmetrical cells with the AMLA own more stable cycle performance and longer lifetime. The symmetrical cells are cycled at 1, 2, and 5 mA cm−<sup>2</sup> with the Li deposit amount of 0.5 mAh cm−<sup>2</sup> . The voltage-time curves of the Li anode (back) and AMLA (red) are showed in the **Figures 3a–c**. The symmetrical cell with the Li anode is only stably cycled for 200 h with overpotential of 0.1 V, when the current density is 1 mA cm−<sup>2</sup> . Subsequently, the overpotential raise up with the decomposition of the electrolyte and the thicker SEI film during the cycling (Yang et al., 2017). After cycling for 300 h, the overpotential of the cell becomes higher than 0.5 V. In addition, the cell with the Li anode appears violent voltage fluctuation after 76 h at low current density of 2 mA cm−<sup>2</sup> and 35 h at high current density of 5 mA cm−<sup>2</sup> (**Figures 3b,c**), which is caused by the reaction between Li and electrolyte (Liu et al., 2018). And there is a performance that as the voltage first decreases and then increases just like a neck shape in the voltage-time curve of the symmetrical cell with the Li anode at 5 mA cm−<sup>2</sup> . This special curve results from the generation and growth of dendrites (Liu et al., 2016). However, the symmetrical cell with the AMLA shows much more stable cycling at different current density. As we can see, the cell with the AMLA can be cycled more than 50 h with steady overpotential, even at 5 mA cm−<sup>2</sup> , which is lower than 0.2 V (**Figure 3c**). Furthermore, the cell with the AMLA shows the lower resistance after 30 times' cycles (**Figures 3d,e**). The resistance is consisted of

schematic of change on Li anode during charge/discharge progress. (d,e) The schematic of change on AMLA during charge/discharge progress. (f,g) SEM images of cross-section and top-view of Li anode after cycles. (h,i) SEM images of cross-section (h) and top-view (i) of AMLA after cycles. (j) SEM images of cross-section AMLA with the deposit amount of 1 mAh cm−<sup>2</sup> as the current density of 1 mA cm−<sup>2</sup> . The XRD curves of AMLA (k).

the ohmic resistance (Rb), interfacial resistance of the electrode (Ri), and diffusion impedance of lithium ions in solids (Zw) (Wu et al., 2019; Zhou et al., 2019). In the symmetrical cells, the most important resistance is the Ri, because the R<sup>b</sup> (4 ) and the Z<sup>w</sup> (the slop of the sloping line at the low frequency region are the same) are the same in different electrodes. As for the AMLA, there are two semicircles in the high frequency region of the EIS curve, due to the mixing layer of LiNO<sup>3</sup> and Ag: the first one represents the lithium ion transfer impedance (Rp) and the second one represents the charge transfer impedance (Rct) (Wu et al., 2019). Therefore, the R<sup>i</sup> of the AMLA is consisted of R<sup>p</sup> and Rct. However, there is only one semicircle for the Li anode, because of the two synchronous processes of lithium ion transfer and charge transfer. So, the R<sup>i</sup> is the same as the Rct in the Li anode. The R<sup>i</sup> of the AMLA is negatively changed at different current density as 23 at 1 mA cm−<sup>2</sup> , 26 at 2 mA cm−<sup>2</sup> , and 34 at 5 mA cm−<sup>2</sup> (**Figure 3d**). Whereas, the R<sup>i</sup> (Rct) of the Li anode is twice higher than the AMLA and increased with the higher current density (45 at 1 mA cm−<sup>2</sup> , 60 at 2 mA cm−<sup>2</sup> , and 70 at 5 mA cm−<sup>2</sup> , **Figure 3e**). The hysteresis in the voltage profile is even lower than 0.2 V for the AMLA at current density of 5 mA cm−<sup>2</sup> , while it increased higher than 0.35 V for the Li anode (**Figures 3f,g**). The mixing uniform layer of LiNO<sup>3</sup> and Ag can improve the electrochemistry performance in the symmetrical cells, which is formed in the AMLA before cycling.

Except for the electrochemistry performance, the mixing uniform layer of LiNO<sup>3</sup> and Ag changes the surface of the Li anode after cycling (**Figures 4f–i**). For the Li anode, there are a lot of drastic Li dendrites on the surface in the cross-section and top view SEM images (**Figures 4f,g**). These Li dendrites are wire shape with length of 5–10µm and diameter of 1– 2µm. It is indicating that the nucleation and deposition of Li are inhomogeneous (**Figures 4b,c**). These Li dendrites will increase the interface of the Li anode contacting with electrolyte to accelerate the side reaction as violent voltage fluctuation during the cycling (**Figures 3a–c**). And, it will consume large amounts of electrolyte and deteriorate the cycle performance. Furthermore, these Li dendrites will pierce the separator and cause safety hazard. In contrast, the morphologies of the Li anode are much smooth without any Li dendrites (**Figures 4d,e**).

In order to understand the mechanism of performance of the AMLA, we use the XRD to study the surface of the Li anode and AMLA anode after the cycling (**Figure 4k**). Once disassembling the symmetrical cell, the first thing we find that the color of the AMLA anode becomes black from yellow. It means that there are somethings formed on the anode during the cycling. In the XRD spectrum of it, there are some characteristic diffraction peaks of another three main composites except Li, corresponding to the LiAg alloy, Li3N, and LiNxOy, which can be represented:

$$\text{Ag} + \text{Li}^+ + \text{e}^- \rightarrow \text{LiAg} \tag{2}$$

$$\rm NO\_3^- + Li^+ + e^- \rightarrow LiN\_xO\_Y + Li\_3N \tag{3}$$

Equation (2) is the reaction between Li and Ag to form the LiAg alloy on the electrode during cycling. Equation (3) is the decomposition of LiNO<sup>3</sup> (Jing et al., 2015), which forms beyond the anode in the electrolyte. Apart from the XRD, we also use the XPS to study the surface of the Li anode and AMLA anode (**Figure 5**). The full XPS spectra (**Figure 5A**) shows: (i) the ration of Li element in the Li anode (2.6%) is lower than in the AMLA anode (10.5%); (ii) the ration of F element in the Li anode (46.0%) is higher than in the AMLA (40.8%); (iii) the peaks of Ag element come on in the AMLA anode. Without AgNO3, the severe interface reaction between the Li anode and electrolyte forms much by-products with F, C, O, and so on, as the higher ratio of F on the surface. In addition, these byproducts thicker the SEI film, which will make less Li be detected, as the lower ratio of Li. Moreover, Ag 3d spectrum shows two peaks at binding energy of 367.84 eV for Ag 3d5/<sup>2</sup> and 373.86 eV for Ag 3d3/2, which means the LiAg alloy is formed (Zhang et al., 2018), shown in the **Figure 5C**. N 1 s (**Figure 5B**) can be assigned to Li3N (398.94 eV) and LiNxO<sup>y</sup> (407.11 eV), while no LiNxO<sup>y</sup> can be detected in the Li anode (**Figure 5D**) (Yan C. et al., 2016; Zhang et al., 2018). These results are the same as the XRD result in **Figure 4k**, which also further confirm the reaction as the Equations (2) and (3). Due to adding AgNO<sup>3</sup> on the surface of the Li anode, we construct outstanding protecting layers with LiAg, LiN3, and LiNxOy. According to the result of the XRD and XPS spectrum, we propose that there is a difunctional protecting layer on the surface of the Li anode to suppress Li dendrites and protect it from being decomposed by electrolyte. This difunctional protecting layer contains the LiAg alloy and the modified SEI by LiNO3. And, it can be explained as the schematic of change on the AMLA anode during charge/discharge progress

(**Figures 4d,e**). First of all, the LiAg alloy can suppress Li dendrites and maintain the integrity of the SEI film. The metal Ag has a great solubility in Li (9 at.% @ 145.5◦C), which means that the metal Ag is lithiophilia. Moreover, the LiAg alloy is formed before the pure Li phase (Yan K. et al., 2016). Thus, the LiAg alloy particles are great nucleation sites of Li, because of the extreme lithiophilia. In addition, the metal Ag has the best conductivity in all of metal as the resistivity of 1.586 × 10−<sup>8</sup> ·m at 25◦C, which makes the LiAg alloy particles have better conduction than the pure Li (Zhang et al., 2018). To sum up, the distribution of the current and the nucleation of Li are uniform as the homogeneously distribution of LiAg nanoparticles. They result in even Li deposition. In addition, the Li diffusion in the phase LiAg alloy (>10−<sup>8</sup> cm<sup>2</sup> s −1 ; Ma et al., 2014) is far higher than in the metal Li (<10−<sup>10</sup> cm<sup>2</sup> s −1 ; Hiratani et al., 1988), which makes the Li deposit under the LiAg alloy. In a conclusion, the LiAg alloy will form firstly, and then the Li will deposit uniformly under the LiAg alloy (Yan K. et al., 2016), as shown in the **Figure 4j**. The thickness of deposited Li is about 5–10µm. Hence, the LiAg alloy can suppress the Li dendrites and protect the integrity of SEI with the low expansion volumetric, which makes no violent voltage fluctuation during the cycles of the cell (Terlicka et al., 2019). Secondly, the modified SEI by LiNO<sup>3</sup> can protect the Li from electrolyte. The LiNO<sup>3</sup> made in the AMLA anode will react with electrolyte to form the Li3N and LiNxO<sup>y</sup> (Shi et al., 2018). In addition, the great ionic conductivity of Li3N will make the Li<sup>+</sup> diffuse through the SEI easily (Ma et al., 2014). Above, these two reacted productions will form dense and stable SEI with great ionic conductivity (Shen et al., 2018). Therefore, this SEI protects the Li from the electrolyte, which will prevent the reaction between the electrolyte and Li, thereby reducing electrolyte consumption (Shi et al., 2018). It also hinders the side reaction between the intermediate Li polysulfides (Li2Sx, 4 ≤ x ≤ 8) and the Li anode to decrease the loss of active materials, when the AMLA anode is used in the Li-S battery (**Figure 6a**; Yan C. et al., 2016; Zhang K. et al., 2017).

The electrochemical performances of the Li-S cells are showed in the **Figure 6b**. These cells with different electrodes (the Li anode and AMLA) are tested by galvanostatic at 0.8375 mA cm−<sup>2</sup> (0.5 C, 1 C = 1,675 mA g−<sup>1</sup> ). As shown, the discharge capacity of Li-S cell with the AMLA anode remains as high as 666.7 mAh g−<sup>1</sup> and the capacity retention after 500 cycles is 60.8%. However, the cell with the Li anode shows a capacity of 310.8 mAh g−<sup>1</sup> with a capacity retention of only 29.1%, which is just half of the AMLA. What's more, the morphologies of the AMLA are still smooth without any Li dendrites the same as it in symmetric cells (**Figures 6c–f**). In a conclusion, the protecting layer formed by the AgNO<sup>3</sup> can protect the Li electrode by suppressing the Li dendrites and prolong the cycle lifetime by hindering the corrosion reaction in the Li-S cell. This effect can be furtherly studied by the XPS in the **Figures 6g,h**. S 2p spectrum of the Li anode shows two main peaks at binding energy of 163.49 eV for Li2S and 164.48 eV for Li2S2, which are the reaction products of Li with intermediate Li polysulfides (**Figure 6h**). These two composites decrease the active materials in the S cathode. However, the rations of Li2S<sup>2</sup> and Li2S are much lower than SO2<sup>−</sup> 3 (168.11 eV), R-SO2-R′ (169.12 eV), and SO2<sup>−</sup> 4 (170.37 eV), which are the compositions of the SEI film in the S 2p spectrum of the AMLA (**Figure 6g**). It means that there are negative Li2S<sup>2</sup> and Li2S and the corrosion reaction is suppressed. The difunctional protecting layer formed in the AMLA is favorable to suppress the Li dendrites, and protect the Li from the corrosion reaction by the intermediate Li polysulfides at the same time.

### CONCLUSIONS

In general, we introduce an effective and simple strategy to improve the Li anode by forming the difunctional protecting layer on the Li anode. The difunctional protecting layer is manufactured through dropping the solution of AgNO<sup>3</sup> on the Li anode directly. They suppress the Li dendrites by depositing Li<sup>+</sup> under them uniformly and hinder the Li anode from the electrolyte to suppress the corrosion reaction. Comparing to the Li anode, the AMLA shows superior electrochemical performance with stable overpotential in the symmetric cell and

### REFERENCES


longer cycling lifetime in Li-S battery. The cell with the AMLA can be cycled more than 50 h at 5 mA cm−<sup>2</sup> with the steady overpotential of lower than 0.2 V. In the Li-S battery, after 500 cycles, the AMLA can still remain the high discharge specific capacity of 666.7 mAh g−<sup>1</sup> as capacity retention rate of 60.8 % at 0.8375 mA cm−<sup>2</sup> (0.5 C, 1 C = 1,675 mA g−<sup>1</sup> ). We believe this simple approach can improve the Li anode to offer great guidance for further application of next-generation batteries, such as Li-S batteries.

### DATA AVAILABILITY

All datasets generated for this study are included in the manuscript and/or the supplementary material.

### AUTHOR CONTRIBUTIONS

CS developed the concept. XL designed the experiments. JG conducted the experiments. ZF and JY built the cells. YZ, ZT, and LS carried out the performance characterizations. CS and KX co-supervised the research. JG and ZF co-wrote the manuscript. All authors discussed the results and commented on the manuscript.

### FUNDING

This work was supported by the National Key R&D Program of China (2018YFB0104200).


**Conflict of Interest Statement:** YZ, ZT, and LS were employed by company Shaanxi Coal and Chemical Technology Institute 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 © 2019 Gu, Shen, Fang, Yu, Zheng, Tian, Shao, Li 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.

# Electrochemical Analysis for Enhancing Interface Layer of Spinel LiNi0.5Mn1.5O<sup>4</sup> Using p-Toluenesulfonyl Isocyanate as Electrolyte Additive

Zhe Xiao† , Renheng Wang\* † , Yan Li, Yiling Sun, Shuting Fan, Keyu Xiong, Han Zhang and Zhengfang Qian\*

*College of Physics and Optoelectronic Engineering, Shenzhen University, Shenzhen, China*

### Edited by:

*Junchao Zheng, Central South University, China*

### Reviewed by:

*Zhiqiang Zhu, Nanyang Technological University, Singapore Bin Huang, Guilin University of Technology, China Zhenjiang He, Central South University, China*

### \*Correspondence:

*Renheng Wang wangrh@szu.edu.cn Zhengfang Qian zqian2013@qq.com*

*†These authors have contributed equally to this work*

### Specialty section:

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

Received: *13 June 2019* Accepted: *09 August 2019* Published: *27 August 2019*

### Citation:

*Xiao Z, Wang R, Li Y, Sun Y, Fan S, Xiong K, Zhang H and Qian Z (2019) Electrochemical Analysis for Enhancing Interface Layer of Spinel LiNi0.5Mn1.5O4 Using p-Toluenesulfonyl Isocyanate as Electrolyte Additive. Front. Chem. 7:591. doi: 10.3389/fchem.2019.00591* LiNi0.5Mn1.5O<sup>4</sup> (LNMO) is a potential cathode material for lithium-ion batteries with outstanding energy density and high voltage plateau (>4.7 V). However, the interfacial side reaction between LNMO and the liquid electrolyte seriously causes capacity fading during cycling at the high voltage. Here, p-toluenesulfonyl isocyanate (PTSI) is used as the electrolyte additive to overcome the above problem of LNMO. The results show that the specific capacity of LNMO/Li cell with 0.5 wt.% PTSI at the first cycle is effectively enhanced by 36.0 mAh/g and has better cycling performance than that without PTSI at 4.98 V. Also, a stable solid electrolyte interface (SEI) film derived from PTSI is generated on the electrode surface, which could alleviate the strike of hydrofluoric acid (HF) caused by electrolyte decomposition. These results are explained by the molecular structure of PTSI, which contains SO3. The S=O groups can delocalize the nitrogen nucleus to block the reactivity of PF5.

Keywords: lithium ion battery, LiNi0.5Mn1.5O4 , p-toluenesulfonyl isocyanate, solid electrolyte interface, electrolyte additive

# INTRODUCTION

Over the past few years, the high energy and power density capability of lithium-ion batteries (LIBs) have been interested extremely, due to potential applications in electric vehicles (EVs), hybrid electric vehicles (HEVs), and plug-in hybrid electric vehicles (PHEVs) (Taracson and Armand, 2001; Armand and Tarascon, 2008; Ji et al., 2011; Kim et al., 2012). In order to improve the energy density and power density of batteries, a large number of Li compounds (e.g., olivine-type materials, silicates, Mn-rich, and Ni-rich layered materials) have been studied by researchers all over the world (Chen et al., 2014; Zhang et al., 2014; He et al., 2015a,b; Panchal et al., 2017; Chan et al., 2018; Li et al., 2018; Qiu et al., 2018). Spinel LiNi0.5Mn1.5O<sup>4</sup> (LNMO) is a promising material to replace layered LiCoO<sup>2</sup> as a cathode for high power density LIBs (Carlier et al., 2003; Su et al., 2017; Sun et al., 2018). LNMO has an high charge-discharge platform (>4.7 V) and outstanding cycling stability (Wang F. et al., 2017). Unfortunately, the high charging voltage (∼4.7 V) is higher than the stable voltage of LiPF6-based electrolyte, resulting in rapid oxidation decomposition of the electrolyte and unnecessary secondary reactions at the LNMO/electrolyte interface (Li et al., 2013; Deng et al., 2017; Ma et al., 2019). Furthermore, hydrofluoric acid (HF) derives from hydrolysis

**88**

of LiPF6-based electrolyte, which can dissolve Mn3<sup>+</sup> from LNMO (Xiao et al., 2017). The Mn3<sup>+</sup> dissolution into the electrolyte causes a cracking solid electrolyte interface (SEI) and reduces rapidly specific capacity, so the LNMO cell exhibits poor cycling peculiarity (Liu et al., 2017; Mou et al., 2018).

One way is to make electrolyte additives form a stable SEI film on the cathode, which inhibits LNMO electrode interface erosion and electrolyte decomposition, scavenging type to capture HF (Haregewoin et al., 2016; Wang et al., in press). As the strong acid produced from LiPF<sup>6</sup> is considered the initiator which induces the cleavage and polymerization of cyclic carbonate under high voltage conditions, many researchers are trying to add some oxidation-resistant solvents, for example, sulfones (Hilbig et al., 2017; Su et al., 2017), nitriles (Abu-Lebdeh and Davidson, 2009) and fluoro solvents (Kim et al., 2017). However, when adding the oxidation-resistant solvents there are new problems, including a decrease of conductivity, an increase of viscosity and poor compatibility. Hence, a number of suitable functional additives, such as (pentafluorophenyl) diphenylphosphine (PFPDPP) (Bolloju et al., 2019), dimethyl phenylphosphonite (DMPP) (Mai et al., 2015), tris (trimethylsilyl) phosphite (TMSP) (Wang et al., 2016), triethyl borate (TEB) (Chen et al., 2017), lithium bisoxalatodifluorophosphate (LiBODFP) (Yang et al., 2019), and so on, have been developed to perform better of LIBs under high voltage. As previously reported (Wang R. H. et al., 2015), p-toluenesulfonyl isocyanate (PTSI) has excellent physical and chemical properties because of SO<sup>3</sup> and S=O groups. What's more, the lowest unoccupied molecular orbital (LUMO) (−0.2469 Ha) occupied by PTSI is lower than that by vinylene carbonate (VC, LUMO = −0.2274 Ha) (Xu, 2004; Wu et al., 2012).

In this work, PTSI will be used as an additive for LiPF6-based electrolyte. The main direction of the experiment is to study the SEI film generated between electrolyte and LNMO electrode surface. We hope the PTSI can form a stable SEI film, suppress the corrosion of LNMO electrode by HF, reduce the formation of other products, and improve the circulation ability of LNMO battery at high voltage.

# EXPERIMENTAL

### Preparation of the Electrolyte

The basic electrolyte (Jiangxi Youli New Materials Co., Ltd., China) was a 1 M (M = mol/L) ethylene carbonate (EC)/ethyl carbonate (EMC)/diethyl carbonate (DEC) LiPF6-based electrolyte in 1:1:1 configuration. The desired concentration 0.5 wt.% of PTSI additive was achieved by dissolving the corresponding amount of PTSI in the base electrolyte and stirring for 5 min in an argon-filled glovebox, and the oxygen and water content were <1 ppm. The supernatant was measured using a Karl Fischer 831 Coulometer (Metrohm) for H2O and Karl Fischer 798 GPT Titrino (Metrohm) for HF, respectively.

### Electrochemical Characterization

LNMO electrodes were prepared from 80 wt.% LNMO powder, 10 wt.% carbon black, and 10 wt.% poly vinylidene fluoride (PVDF). N-methyl pyrrolidinone (NMP) was then added and ground evenly. Next, spread the mixture slurry evenly on the thin aluminum foil and vacuum dry at 120◦C for 12 h. The 10 mm diameter electrode disc was then perforated from the coated foil. The LNMO/Li of 2,032 coins were assembled in argon filled ball cases with 2,400 Celgard dividers.

Electrochemical impedance spectroscopies (EIS) of LNMO/Li cells after 1 cycle, 2 cycles, and 3 cycles at 4.98 V were recorded by an electrochemical workstation (CHI660E, Chenhua, Shanghai), and the open-circuit voltages of the cells were set as the initial potential. The frequency range of LNMO/Li cells was 10,000∼0.01 Hz. Cyclic voltammetry (CV) was detected by an electrochemical workstation with five cycles at a sweep rate of 0.1 mV s−<sup>1</sup> .

### Surface Detections of the LNMO Electrode

LNMO/Li cells after cycles were disassembled. First, the LNMO electrode was washed three times with high-purity DMC, and then transferred to a vacuum drying box at 45◦C and placed in it for 4 h. The microstructure and morphology of LNMO electrode were recorded by scanning electron microscope (SEM). The surface morphology of LNMO electrode was detected though the transmission electron microscopy (TEM). X-ray photoelectron spectroscopy (XPS) was used to analyze the composition of chemical elements on the surface of LNMO electrode.

### RESULTS AND DISCUSSION

### Cycling Performance Analyses

It can be clearly seen that the discharge specific capacity of LNMO/Li cell with 0.5 wt.% PTSI was higher than that without additives at the first cycle from in **Figure 1**. From in **Figures 1A,B**, the discharge specific capacity of LNMO/Li cell with additional PTSI reached 143.8 mAh/g, while the discharge specific capacity of the cell without additive is only 107.6 mAh/g. It can be concluded that PTSI additive could improve the initial discharge specific capacity of LNMO/Li cell. What's more, the specific capacity of LNMO/Li cell with 0.5 wt.% PTSI added after 40 cycles was much higher than that of LNMO/Li cell without additive. This also reflected that PTSI can indeed improve the specific capacity of LNMO/Li cell and provide a new scheme for improving the energy density of LIBs. In addition, the coulomb efficiency of the battery was constantly improved maintained a high level, as shown in **Figure 1C**. The coulomb efficiency of LNMO/Li cell without additives is 96% at the 40th cycle, while LNMO/Li cell with 0.5 wt.% PTSI added reached 99%. This indicated that the PISI additive can improve the coulomb efficiency during the charging/discharging cycle of LNMO/Li cell.

### Impedance Analysis

In order to explore the interface impedance of SEI film between electrolyte and LNMO electrode, EIS of LNMO/Li cells with 0.5 wt.% PTSI and with no additive were recorded, as shown in **Figure 2**. In impedance spectroscopy, the semicircular high frequency region represents the migration of lithium ions through the interface at the surface of the LNMO electrode, and the center frequency range of the semicircle corresponds to the charge transfer process (Zhao et al., 2018). The results

show that the interfacial impedance of LNMO electrode with 0.5 wt.% PTSI additive is significantly lower than that of LNMO electrode with no additive, suggesting that the surface of the LNMO electrode with 0.5 wt.% PTSI is improved. What is more, the two semicircles in the impedance spectra with 0.5 wt.% PTSI are significantly reduced compared with those with no additive. As the number of cycles increases, the impedance change of LNMO electrode with the addition of PTSI is much smaller than that of the LNMO electrode with no additive. In the lithiation/delithiation process, the surface layer impedance reduction and charge transfer will reduce ohmic polarization and activation polarization, which also confirms the above superior cyclic performance of the LNMO/Li cells with 0.5 wt.% PTSI.

### CV Measurements

In order to better understand the effect of the additive PTSI on the LNMO cell, the battery was subjected to the CV measurements at a sweep rate of 0.1 mV s−<sup>1</sup> at 25◦C, and the results are shown in **Figure 3**. It can be seen from the figure that there is a major redox peak at around 4.7 V, which corresponds to the redox process of Ni2<sup>+</sup> and Ni4+(Talyosef et al., 2005). According to the comparison of **Figures 3A,B**, the peak current of the Li/LNMO cell to which 0.5 wt.% of PTSI was added is significantly larger than the peak current of the Li/LNMO cell without additives. In addition, the potential difference between the two peaks in the CV diagram of the LNMO cell to which PTSI was added is small, and as the number of cycles increases, the coincidence degree of the CV measurements pattern of the LNMO cells to which the PTSI was added is higher, indicating that the addition of PTSI makes the circulation of the LNMO cell more stable.

### SEM and TEM Analyses

To further study the effect of additive PTSI on the performance of LNMO electrode, SEM tests were carried out on the fresh electrode and the cycled LNMO electrode, as shown in **Figure 4**. The surface of the fresh electrode is very smooth and clean, without sediments, while the cycled LNMO electrodes show very different surface morphology. Compared to the fresh LNMO electrode, the surface of LNMO electrode with no additive is not smooth, which is coated with thick materials. Therefore, it increases the surface area, leading to an interface reaction that affects the transport of Li<sup>+</sup> through the electrode. In contrast, the surface of the cycled LNMO electrode with 0.5 wt.% PTSI shows smooth and flat, forming thin materials on the surface. The relatively low viscosity of the additive PTSI improves permeability of electrolytes to LNMO electrode (Wang R. et al., 2015). Meanwhile, the reduction potential of PTSI was higher than carbonate solvents, which hinders the solvent decomposition in LiPF6-based electrolyte.

In addition, in order to better describe the effect of additive PTSI on LNMO electrode, the corresponding TEM images of LNMO electrode after 40 cycles were obtained in **Figure 5**. It can be clearly seen that the LNMO electrode with 0.5 wt.% PTSI has a relatively clear layer boundary, which is the SEI film (∼3 nm) generated on the surface of the electrode. The film is dense and uniform, which can effectively protect the LNMO electrode. However, a very uneven and thick SEI film (∼12 nm) is generated on the LNMO electrode with no additive, which will slow down the transfer of Li<sup>+</sup> between electrolyte and electrode. It markedly indicates that the decomposition of electrolyte, as well as the electrode erosion from the electrolyte with no additive, is more serious.

### XPS Analysis

In order to verify the specific elements of surface layer about fresh electrode, non-additive electrode and the electrode with 0.5 wt.% PTSI after 40 cycles were detected by XPS in **Figure 6**. The C 1s spectra have four main peaks: C-C bond from the carbon black (284.1 eV), C-H bond roots in lithium alkyl carbonates (R-CH2OCO2-Li) and PVDF (286.0 eV), C=O bond belongs to lithium alkyl carbonates (R-CH2OCO2-Li) and polycarbonates (287.6 eV), and Li2CO<sup>3</sup> (290.1 eV) (Funabiki et al., 1997; Levi et al., 2000; Dedryvere et al., 2010; An et al., 2016; Wang R. H. et al., 2015). It can be seen that the strength of Li2CO<sup>3</sup> on the surface of LNMO electrode with 0.5 wt.% PTSI is significantly weaker than that on the surface of LNMO electrode with no additive, indicating the inhibitory effect of PTSI on electrolyte decomposition.

The O 1s spectrum displays five different peaks, including C-O peak (532.8 eV), Mn-O bond (529.7 eV), Li2CO<sup>3</sup> (531.8 eV), C=O bond (532.4 eV), and C-O-C bond in lithium alkyl carbonates (R-CH2OCO2-Li) (533.8 eV) (Dedryvère et al., 2005; Bae et al., 2014; Wang et al., 2016). After PTSI was added, the C=O bond energy intensity increased, indicating that the polarization of EC and DEC solvents is effectively inhibited. In addition, Li2CO<sup>3</sup> peak with no additive is stronger. That is, there are many inorganic decomposition products on the surface of the LNMO electrode with no additive.

In the F 1s spectrum, there is a significant difference between the two electrodes after cycling in different electrolytes. The

peak strength of LiF (684.5 eV) and PVDF (687.7 eV) with 0.5 wt.% PTSI was significantly lower than that with no additive (Park et al., 2010; Zhou et al., 2011), indicating that there are few inorganic products on the electrode surface. When LiF content in the SEI film increasing, it will cause erosion to the electrode and inhibit the transport of Li ions. Hence, the impedance of LNMO electrode surface will increase accordingly. So, the SEI film generated by PTSI enhances the electrical charge transfer channel between LNMO electrode and electrolyte.

In the Mn 2p spectrum, there are three main characteristic peaks, which belong to Mn3<sup>+</sup> (641.7 eV), Mn4<sup>+</sup> (642.9 eV), and Mn 2p1/<sup>2</sup> (653.6 eV) (Treuil et al., 1999). It is found that the Mn3<sup>+</sup> peak strength of no additive electrolyte is lower than that of PTSI additive electrolyte, suggesting that HF causes the erosion of LNMO electrode surface.

In conclusion, SEM, TEM, and XPS indicate that the SEI film of LNMO electrode with 0.5 wt.% PTSI is thinner than that of LNMO electrode with no additive. The optimization of SEI film can greatly promote the transport of Li<sup>+</sup> to a large extent and inhibit the oxidation decomposition of electrolyte, which can prevent the product from damaging the electrode.

# The Proposed Mechanism for LNMO/Electrolyte Interface Film

According to the above analysis, a schematic diagram of SEI film formation on the surface of LNMO electrode is obtained, as shown in **Figure 7**. Compared with the alkyl carbonic (Li2CO<sup>3</sup> and ROCO2Li) generated by EC, the decomposition products (Li2SO3, Li2S, and ROSO2Li) formed by the additive PTSI are relatively more stable, which inhibits HF to corrode the surface of LNMO electrode and reduces interface impedance (Wang R. et al., 2017). Therefore, the diffusion of Li<sup>+</sup> across the surface of LNMO was enhanced.

PF<sup>5</sup> acted as a catalyst for the oxidation and corrosion of electrolyte, and could guide the reaction path of electrolyte to HF and H2O (Sloop et al., 2003). What's more, PF<sup>5</sup> decomposed into EC and DEC. The open-loop reaction of EC was catalyzed by PF5, which leads to the polymerization of the reaction to produce polyethylene carbonate (PEC) and polyethylene oxide (PEO) similar products. As PF<sup>5</sup> lack electrons, PTSI contains many electrons, including the S=O group, which caused the nitrogen nucleus to be delocalized and the weak base to be sited as inhibiting PF<sup>5</sup> reactivity (Wu et al., 2012; Wang R. H. et al., 2015). The HF generated and LiF formed from LiPF<sup>6</sup> will be inhibited. The SEI film formed on the surface of the LNMO electrode can reduce the interfacial resistance between LNMO and electrolyte.

What's more, PTSI played an important role in the development of SEI film, which successfully prevented HF from passing through the modified film to corrode the LNMO electrode. By reducing the reaction between PF<sup>5</sup> and electrolyte, the content of LiF in the SEI film was reduced and the formation of HF was inhibited. The results show that PTSI can significantly inhibit the degree of oxidative decomposition of carbonate solvent during LNMO/Li cell cycle. It is concluded that PTSI is used as electrolyte additive for LNMO electrode at a high range voltage of 3.0–4.98 V.

# CONCLUSIONS

In this work, we report a electrolyte based on 1 M LiPF<sup>6</sup> EC/EMC/DEC (1:1 by wt.%) with 0.5 wt.% PTSI for LNMO/Li. Electrochemical tests, EIS, CV, SEM, TEM, and XPS display that the decomposition of carbonate solvent has been inhibited and a dense SEI film on the electrode surface is formed. The electrolyte using PTSI as a non-aqueous electrolyte additive has good electrochemical stability at high voltages 4.98 V. The SEI film generated from PTSI is a stable protective layer, which inhibits HF erosion and reduces the interface resistance. As a result, LNMO/Li cells show excellent cycling performance.

## DATA AVAILABILITY

All datasets generated for this study are included in the manuscript/supplementary files.

# AUTHOR CONTRIBUTIONS

ZX, RW, and ZQ designed and engineered the samples. ZX and RW performed the experiments. ZX, RW, YL, YS, SF, KX, HZ, and ZQ performed the data analysis. ZX, RW, and ZQ wrote the paper. All authors contributed to the theoretical analysis and the general discussion.

### FUNDING

The work was supported by the National Science Foundation for Young Scientists of China (Grant No. 51804199), the Science and Technology Innovation Commission of Shenzhen (Grant No. 20180123), and the National Natural Science Fund (Grant No. 61435010).

### ACKNOWLEDGMENTS

Authors acknowledge the supports from Instrumental Analysis Center of Shenzhen University (Xili Campus) and

### REFERENCES


Shenzhen Institute of Information Technology. The authors also wish to acknowledge the assistance on SEM and TEM observations received from the Electron Microscope Center of the Shenzhen University.


**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 Xiao, Wang, Li, Sun, Fan, Xiong, Zhang and Qian. 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 One-Step Hydrothermal Synthesis of Na3V2(PO4)2F3@C/CNTs Tetragonal Micro-Particles as High Performance Cathode Material for Na-Ion Batteries

Hao Guo1,2, Yong Hu<sup>1</sup> , Xiaoping Zhang<sup>1</sup> , Rongliang Zhang<sup>2</sup> , Dong Hou<sup>1</sup> \*, Yulei Sui <sup>1</sup> and Ling Wu<sup>1</sup> \*

*<sup>1</sup> School of Iron and Steel, Soochow University, Suzhou, China, <sup>2</sup> Zhangjiagang Campus School of Metallurgical and Material Engineering, Jiangsu University of Science and Technology, Zhenjiang, China*

### Edited by:

*Junchao Zheng, Central South University, China*

### Reviewed by:

*Yunjian Liu, Jiangsu University, China Qiaobao Zhang, Xiamen University, China Lingjun Li, Changsha University of Science and Technology, China*

### \*Correspondence:

*Dong Hou houdong0702@suda.edu.cn Ling Wu lwu@suda.edu.cn*

### Specialty section:

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

Received: *04 August 2019* Accepted: *07 October 2019* Published: *18 October 2019*

### Citation:

*Guo H, Hu Y, Zhang X, Zhang R, Hou D, Sui Y and Wu L (2019) Facile One-Step Hydrothermal Synthesis of Na3V2(PO4)2F*3*@C/CNTs Tetragonal Micro-Particles as High Performance Cathode Material for Na-Ion Batteries. Front. Chem. 7:689. doi: 10.3389/fchem.2019.00689* In this paper, we report a facile one-step hydrothermal method to synthesize tetragonal Na3V2(PO4)2F3@C particles which are connected by carbon nanotubes (CNTs) networks, using water as hydrothermal solvents. In this strategy, the reduction and crystallization of materials are carried out in the hydrothermal process (180◦C, 12 h), no additional heat treatment is required. The well-crystallized Na3V2(PO4)2F<sup>3</sup> tetragonal grains (5–10µm) are coated with amorphous nano-carbon and connected by highly conductive CNTs. The addition of CNTs can not only improve the conductivity of materials but also effectively inhibit the Na3V2(PO4)2F<sup>3</sup> grains over growth. The Na3V2(PO4)2F3@C/CNTs composite possesses very flat charge/discharge platforms of 3.6 and 4.1 V. The sample exhibits an initial discharge specific capacity of 120.2 and 74.3 mAh g−<sup>1</sup> at 0.1 and 10 C rate, respectively, and shows excellent cyclical stability. The composite owns excellent electrochemical performances owing to the three-dimensional highly conductive network which is co-constructed by the CNTs and nano-carbon coating layer.

Keywords: Na-ion batteries, cathode materials, Na3V2(PO4)2F3, carbon nanotubes, hydrothermal

# INTRODUCTION

Recently years, sodium-ion batteries (SIBs) have been receiving great attention due to the low cost and abundant sodium resource. At present, the cathode materials of SIBs mainly include the transition metal oxides (NaxMeO2, Me = Fe, Mn, Co, Ni, and V, etc.) (Delmas et al., 1981; Doeff et al., 1994; Guignard et al., 2013; Vassilaras et al., 2013; Chen et al., 2017, 2018), polyanionic compounds (Na3V2(PO4)3, Na3V2(PO4)2F3, and Na2MnPO4F, etc.) (Xie et al., 2017; Wu et al., 2018a,b; Zheng et al., 2018; Ge et al., 2019; Leng et al., 2019) and Prussian blue compounds (AxMA[MB(CN)6]·yH2O) (Qian et al., 2012; Lee et al., 2013; Wang et al., 2013; Yue et al., 2014). Among these materials, Na3V2(PO4)2F<sup>3</sup> (NVPF) is widely concerned owing to its NASICON structure (Na superionic conductor) and excellent electrochemical performances. Based on the strong induction effect of PO3<sup>−</sup> 4 and the strong electronegativity of F−, NVPF shows high operating voltages (3.6 and 4.1 V), high energy density (∼507 Wh kg−<sup>1</sup> ) and excellent thermal stability (Bianchini et al., 2015; Song et al., 2015). Most importantly, the particular NASICON

**96**

structure can provide three-dimensional (3D) channels for the fast transmission of Na+, making NVPF exhibit relatively higher ionic conductivity than many other SIBs cathode materials.

Generally, NVPF materials are mainly synthesized by solidstate reaction method (Gover et al., 2006; Shakoor et al., 2012; Park et al., 2014), spray-drying method (Eshraghi et al., 2017; Shen et al., 2018), sol-gel method (Jiang et al., 2009; Pineda-Aguilar et al., 2017), and hydrothermal method (Serras et al., 2014; Cai et al., 2018; Guo et al., 2018), etc. In order to obtain a well-crystallized NVPF, most of the above mentioned methods usually require high temperature calcination processes (480–850◦C). Nevertheless, high temperature will result in that the NVPF particles become irregular, inhomogeneous or even agglomerate together to form larger clusters, which is unfavorable for the Na-ions transmission. Also, the F element may be lost during the calcination process, introducing some impurity phases such as Na3V2(PO4)<sup>3</sup> (Eshraghi et al., 2017). Furthermore, the excess calcination process will greatly increase the energy consumption. Although several solvothermal routes (Xu et al., 2014; Qi et al., 2015; Zhao et al., 2015; Zhu et al., 2017) have been reported to synthesize NVPF without heat treatment, the required organic solvents [ethanol, acetone, DMF, and acid-base coupling extractant (PN) etc.] have high vapor pressures which are very dangerous. Thus, the vessel used for the solvothermal process must withstand great high pressures, leading to an extra production cost. And the organic raw materials [such as Vanadium (III) acetylacetonate] used in the solvothermal process are also very expensive. In addition, the organic solvents are usually toxic and harmful to the environment. Thus, the solvothermal method is difficult for the practical application.

Therefore, a green and cheap synthetic route for NVPF without additional calcination process should be developed urgently. Here, we propose a water based hydrothermal route to prepare well-crystallized NVPF, without any extra heat treatment process. In this route, all the raw materials (H3PO4, NH4VO3, NaF, and oxalic acid) are common and cheap, and NVPF@C tetragons can be one-step synthesized at 180◦C with deionized water as solvent. Particularly, here oxalic acid is used as a chelating agent, reducing agent, and carbon source at the same time. In addition, we further add CNTs in the process to obtain NVPF@C/CNTs composite material. The addition of CNTs can not only enhance the conductivity of materials but also restrict the grains over growth. Most importantly, the embedded and wrapped CNTs combined with a uniform nano-carbon layer for NVPF construct a highly conductive 3D network for the transmission of electron and Na+. As a result, the micron scale (5–10µm) NVPF@C/CNTs composite synthesized in this study shows excellent rate capability and cycling performance, which can compete with the nanoscale NVPF materials reported in many other reports (Li et al., 2018a,b). It is well-known that the micron-sized cathode materials are preferred rather than nanosized ones for the commercial batteries (such as LiCoO<sup>2</sup> and LiNixCoyMnzO2), as the bigger particles need less binder, solvent and conductive additive to prepare an efficient electrode (Choi and Aurbach, 2016). The micron-sized particles also show higher tap density than the nanosized ones. In a word, this is a green, low cost and efficient water based hydrothermal method for the preparation of NVPF cathode material, and the micron sized NVPF@C/CNTs composite produced by this method is a promising cathode material for SIBs.

### EXPERIMENTAL

### Materials Synthesis

The synthesis route of Na3V2(PO4)2F3@C/CNTs cathode material is shown in **Figure 1**. Firstly, 0.01 mol H3PO<sup>4</sup> (A.R., Aladdin) and 0.05 g CNTs (commercially available carbon nanotubes) were added into 30 ml deionized water, followed by sonication for 15 min. Then, 0.015 mol C2H2O<sup>4</sup> (A.R., Aladdin), 0.01 mol NH4VO<sup>3</sup> (A.R., Aladdin) and 0.015 mol NaF (A.R., Aladdin) were sequentially added, and continuously stirred for 30 min. Thereafter, the mixture was transferred to a Teflon cup of 50 mL inner volume, placed in a stainless steel autoclave and kept in an oven at 180◦C for 12 h. After cooling to room temperature, the sample in the kettle was taken out, centrifuged and dried to get Na3V2(PO4)2F3@C/CNTs composite material (denoted as NVPF@C/CNTs). The Na3V2(PO4)2F3@C (denoted

as NVPF@C) cathode material was prepared without the addition of CNTs by using the same synthesis route.

# Characterization

XRD was carried out on a Rigaku Ultima IV diffractometer using Cu Kα radiation (40 kV); step scan 0.02, 2θ range 10–90◦ , step time of 2◦ min−<sup>1</sup> . IR spectrum was tested on a VERTEX70 spectrometer with a transmission mode in the range of 600–2,000 cm−<sup>1</sup> . XPS was performed using a Kratos X-ray photoelectron spectrometer (Axis Ultra-DLD) with Mg Kα radiation (hν = 1283.3 eV), and all the spectra were calibrated with the C 1s peak at 284.8 eV. The carbon content was measured by C–S analysis (Eltar, Germany). SEM (Hitachi SU-5000) was performed to study the morphology of materials. HRTEM and EDX analyses were carried out with a FEI Tecnai G2 F20 microscope at 200 kV.

## Battery Fabrication and Electrochemical Tests

The positive electrode was prepared with the Na3V2(PO4)2F<sup>3</sup> samples, acetylene black and PVDF in the weight ratio of 8:1:1. The N-methylpyrrolidone and Al-foil were used as solvent and current collector, respectively. The metallic sodium-foil and glass fiber membrane (Whatman GF/A) were used as the negative electrode and separator, respectively. The NaClO<sup>4</sup> (1 M) solution in propylene carbonate (PC) and fluoroethylene carbonate (FEC) (95% PC and 5% FEC in volume) was used as the electrolyte. The CR2025 cells were assembled in an Ar-filled dry glove box (H2O and O<sup>2</sup> < 0.1 ppm). The cells were tested at different C-rates (0.1–10 C, 1C = 120 mA g−<sup>1</sup> ) between 2.5 and 4.3 V at ambient temperature. The cyclic voltammetry (CV) was carried out with a CHI-660D electrochemical workstation; potential range 2.5– 4.3 V (vs. Na+/Na), scan rate 0.1 mV s−<sup>1</sup> .

# RESULTS AND DISCUSSION

**Figure 2A** shows the XRD patterns of NVPF@C and NVPF@C/CNTs samples. From **Figure 2A**, it is found that there is no obvious difference between the NVPF@C and NVPF@C/CNTs samples, and all diffraction peaks can be fully indexed as the tetragonal structure of Na3V2(PO4)2F<sup>3</sup> (PDF#89-8485) with space group P42/mnm, and no other impurity phases are detected. In addition, the element contents of both samples are also measured. It is found that the molar ratios of Na: V: P: F of NVPF@C and NVPF@C/CNTs are 2.99: 2: 2.01: 3.01 and 3.01: 2: 1.99: 2.98, respectively, which are close to their theoretical values. The above results indicate that the well-crystallized pure phase Na3V2(PO4)2F<sup>3</sup> can be successfully synthesized through water-based hydrothermal method, and the addition of CNTs does not affect the crystal structure of Na3V2(PO4)2F3. For the sake of contrast, the XRD pattern of CNTs is also shown in **Figure 2B**. However, those diffraction peaks corresponding to CNTs are not clearly observed


TABLE 1 | Lattice parameters and crystallite size of NVPF@C and NVPF@C/CNTs samples obtained from XRD Rietveld refinement.

in NVPF@C/CNTs. Therefore, in order to further clarify the structure of the samples, the XRD data of NVPF@C and NVPF@C/CNTs samples are refined by Rietveld method, and the refinement results are shown in **Figures 2C,D**, respectively. As shown, each fitting curve is well-matched with the measured data, and the refinement errors R<sup>w</sup> and Rexp are acceptable, implying the results are reliable. The refinement results in **Figure 2D** demonstrate the presence of CNTs and indicate that the phase content of CNTs is about 2.31 ± 0.14 wt%. To verify the accuracy of CNTs content, the samples were subjected to C-S analysis. C-S test results reveal that the carbon content of NVPF@C and NVPF@C/CNTs is 3.04 wt% and 5.21 wt%, respectively. Compared with NVPF@C, the increased carbon content (2.17 wt%) of NVPF@C/CNTs should be ascribed to CNTs. This value (2.17 wt%) is close to the XRD refinement result. The above results reveal that CNTs are successfully compounded with NVPF@C. **Table 1** shows the lattice parameters and crystallite size of the two samples. As shown, the lattice parameters of both samples are similar, and all the values are consistent with those reported in the previous references (Gover et al., 2006; Shakoor et al., 2012; Cai et al., 2018). However, the crystallite size (D value) of NVPF@C/CNTs is smaller than that of NVPF@C. The results demonstrate that although the addition of CNTs does not affect the crystal structure of NVPF, it can restrict the growth of crystal in some degree, which is of great benefit to obtaining smaller NVPF crystals.

Infrared absorption spectroscopy (FTIR) analysis of the NVPF@C and NVPF@C/CNTs samples are shown in **Figure 3**.

It is found that the peaks of both samples appear in the same band, but the absorption peaks of NVPF@C/CNTs are stronger. The band corresponding to the symmetric stretching vibration of PO3<sup>−</sup> 4 anion is located at 673 cm−<sup>1</sup> , and the broadband at 1,050 cm−<sup>1</sup> is ascribed to the asymmetric stretching of PO3<sup>−</sup> 4 anion. In addition, the band at 944 cm−<sup>1</sup> manifests the presence of V–F (Qi et al., 2015; Li et al., 2018b). Moreover, the vibration from V3+— O2<sup>−</sup> bonds in isolated VO4F<sup>2</sup> octahedra, is detected at 914 cm−<sup>1</sup> . While, the typical bands of V5<sup>+</sup> in the VO4F<sup>2</sup> octahedron are not detected at 760 cm−<sup>1</sup> and 950 cm−<sup>1</sup> (Song et al., 2014; Pineda-Aguilar et al., 2017; Cai et al., 2018), meaning that the V5<sup>+</sup> ions in NH4VO<sup>3</sup> have been reduced to V3+.

In order to further determine the valence state of Vn<sup>+</sup> in the samples, XPS spectra of V 2p are shown in **Figure 4**. It can be observed that the peaks corresponding to V 2p3/<sup>2</sup> and V 2p1/<sup>2</sup> are present at binding energies of 515.7 and 522.9 eV, which are similar to the binding energies of V3<sup>+</sup> in the previous literature (Ni et al., 2017; Li et al., 2018a). Based on the above analysis, V5<sup>+</sup> is successfully reduced to V3<sup>+</sup> by oxalate acid during the hydrothermal process, and well-crystallized NASICON-type Na3V2(PO4)2F<sup>3</sup> can be obtained by the one-step hydrothermal method. Most significantly, no additional heating process is required.

**Figures 5a–d** shows the SEM images of NVPF@C and NVPF@C/CNTs. As shown, both samples exhibit the tetragonal architecture, but the sample NVPF@C/CNTs shows more regular tetragonal architecture and smaller particle size than NVPF@C. The particle size of NVPF@C and NVPF@C/CNTs is about in the ranges of 30–50 and 5–10µm, respectively. It is obvious that the NVPF@C grains are too large for the fast insertion/extraction of Na-ions. While the grain size of NVPF@C/CNTs is suitable and very close to those of commercial Li-ion battery materials, such as LiCoO<sup>2</sup> and LiNi0.33Co0.33Mn0.33O2. These changes in the shape and size of NVPF particles are mainly ascribed to the addition of CNTs, which is not only beneficial to restrict the NVPF particles over growth, but also can weave a 3D conductive network between the micro-sized tetragonal grains. This 3D network structure can be expected to greatly improve the conductivity of NVPF cathode materials. The tap densities of NVPF@C and NVPF@C/CNTs are 1.12 and 1.03 g cm−<sup>3</sup> , respectively, which are higher than those of nanoscale powders.

TEM images of the NVPF@C and NVPF@C/CNTs samples are shown in **Figures 5e,h**, respectively. As can be seen, both samples show tetragon shape, but NVPF@C/CNTs sample exhibits smaller grain size than NVPF@C. Furthermore, NVPF crystals in NVPF@C/CNTs sample are intertwined with CNTs. From the HRTEM images (**Figures 5f,i**), it can be clearly seen that the surface of the NVPF crystal for both samples are uniformly coated with an amorphous nano-carbon layer (about 5 nm thick). The carbon layer is mainly derived from the carbonization of oxalic acid. The regular lattice fringes

FIGURE 5 | SEM images of NVPF@C (a,b) and NVPF@C/CNTs (c,d). TEM and HRTEM images of NVPF@C (e,f) and NVPF@C/CNTs (h,i). TEM image of NVPF@C/CNTs (g) and corresponding EDS line-scanning curves of Na, V, P, O, F, and C elements (j).

corresponding to the (2 0 0) and (2 1 1) crystal planes of NVPF can be observed in NVPF@C and NVPF@C/CNTs samples, respectively. In addition, the lattice fringes of CNTs [d(002) = 3.395 Å] can be clearly found in NVPF@C/CNTs, and some lattice fringes of CNTs embed in the lattice fringes of NVPF crystals. The above results confirm that some CNTs are embed into the NVPF crystals and the others wrap around the NVPF@C particles. As a result, the embedded and wrapped CNTs combined with the uniform nano-carbon layer construct a 3D conductive network for the transmission of electron and Na+. Moreover, the illustrations in the upper left and lower right areas in **Figure 5i** are the fast Fourier transform (FFT) of the two selections. The upper left area corresponds to the diffraction spots of the NVPF, and the lower right area corresponds to the diffraction spots of the CNTs. The FFT images further prove the coexistence of NVPF and CNTs. For further study, we perform a EDS linear scanning analysis for NVPF@C/CNTs sample along the line marked in **Figure 5g**, and the results are shown in **Figure 5j**. As shown, the Na, V, P, O, and F elements are mainly distributed in the NVPF particles, and the carbon signal in the particles confirms the presence of the carbon layer on the surface of NVPF

particles. Significantly, the location of the CNTs shows a very high carbon signal.

**Figures 6A,B** shows the charge and discharge curves of the samples at different C-rates. It is obvious that the NVPF@C/CNTs sample shows more flat voltage plateaus, smaller polarization and higher average working voltage than NVPF@C at various C-rates during charge and discharge processes. Furthermore, the initial discharge specific capacity of NVPF@C/CNTs (120.2 mAh g−<sup>1</sup> ) is higher than that of NVPF@C (105.8 mAh g−<sup>1</sup> ) at 0.1C rate. With the increasing of C-rate (from 1C to 10C), the discharge capacities of NVPF@C/CNTs also exhibit less decay than NVPF@C. For example, the ratio of 10C/0.1C discharge capacity for NVPF@C sample is only 37.6%, while the ratio for NVPF@C/CNTs is as high as 61.8%. **Figure 6C** shows the cycle performance of the two samples at various C-rates. It is distinct that NVPF@C/CNTs shows better rate capability and cycling performance than NVPF@C. Especially, the NVPF@C/CNTs sample exhibits a capacity retention rate of 99.5% after 20 cycles at 0.1 C rate, and shows almost no capacity fading when cycling at 1C, 2C, 5C, and 10C, respectively. After cycling at various C-rates for 80 cycles, the discharge

FIGURE 6 | The first charge/discharge curves of NVPF@C (A) and NVPF@C/CNTs (B) at various C-rates. Cycling performance of NVPF@C and NVPF@C/CNTs at various C-rates (C). CV curves of NVPF@C/CNTs at a scan rate of 0.1 mV s−<sup>1</sup> (D).

specific capacity of NVPF@C/CNTs at 0.1C (the 81st cycle) can still holds 99.4% of its initial value of the first cycle (120.2 mAh g−<sup>1</sup> ), which indicates that the structure of NVPF is very stable. Moreover, the coulomb efficiency of each cycle is almost 100% at various C rates. **Figure 6D** presents the CV profiles of NVPF@C/CNTs sample at a sweep rate of 0.1 mV s−<sup>1</sup> . As shown, the sample exhibits the oxidation and reduction peaks of 3.66/3.57 and 4.05/3.98 V, respectively. It should be noted that the 1st, 2nd, and 3rd CV curves overlap closely, which further proves the stable structure of NVPF. The excellent electrochemical performances of NVPF@C/CNTs should be owing to: (1) the well-crystallized NVPF can offer a stable crystal structure; (2) the addition of CNTs can effectively restrict the NVPF grains over growth, which is conducive to the migrating of Na-ions during the charge/discharge processes; (3) the uniform nanocarbon layer, embedded and wrapped CNTs together construct a highly conductive 3D network for the electron transmission.

# CONCLUSIONS

Well-crystallized NVPF@C/CNTs cathode material with tetragonal architecture is successfully synthesized by a onestep water based hydrothermal method. In this method, the reduction and crystallization of materials are simultaneously accomplished in the hydrothermal process, no additional heat treatment is needed. The obtained NVPF crystals are coated with a uniform nano-carbon layer, and CNTs are embed into and wrapped around the NVPF grains. Both the coating layer

# REFERENCES


and CNTs weave a highly conductive 3D network for the migration of electron and Na-ion, which leads to the excellent rate capability and cycling performance of the micron-sized (5–10µm) NVPF@C/CNTs composite. Significantly, the raw materials used in this method are very ordinary, cheap and eco-friendly, the solvent is water and the synthesis temperature is only 180◦C. Therefore, we believe that this is a green, low cost, efficient, and facile method for the synthesis of NVPF-based cathode materials.

# DATA AVAILABILITY STATEMENT

All datasets generated for this study are included in the article/supplementary files.

# AUTHOR CONTRIBUTIONS

HG and YH did the main experiment and wrote the manuscript. XZ, RZ, and YS envolved the discussion of the experiment and revised the manuscript. LW and DH made the research plan. LW also provided the financial support.

# FUNDING

This study was financially supported by the National Natural Science Foundation of China (51574170, 51774210, and 51774207).

thanks to well-dispersed CNT network within NVPF particles. Electrochim. Acta 228, 319–324. doi: 10.1016/j.electacta.2017.01.026


exchange: NaVPO4F or Na3V2(PO4)2F3. Adv. Energy Mater. 8:1801064. doi: 10.1002/aenm.201801064


**Conflict of Interest:** 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 Guo, Hu, Zhang, Zhang, Hou, Sui and Wu. 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.

# Graphene-Modified 3D Copper Foam Current Collector for Dendrite-Free Lithium Deposition

Juan Yu1,2, Yangyang Dang<sup>1</sup> , Maohui Bai <sup>3</sup> , Jiaxin Peng<sup>1</sup> , Dongdong Zheng<sup>1</sup> , Junkai Zhao1,2, Linbo Li 1,2 and Zhao Fang1,2 \*

<sup>1</sup> School of Metallurgical Engineering, Xi'an University of Architecture and Technology, Xi'an, China, <sup>2</sup> Shaanxi Province Metallurgical Engineering and Technology Research Centre, Xi'an, China, <sup>3</sup> School of Metallurgy and Environment, Central South University, Changsha, China

Lithium (Li) metal is regarded as the ideal anode for rechargeable Li-metal batteries such as Li-S and Li-air batteries. A series of problems caused by Li dendrites, such as low Coulombic efficiency (CE) and a short circuit, have limited the application of Li-metal batteries. In this study, a graphene-modified three-dimensional (3D) Copper (Cu) current collector is addressed to enable dendrite-free Li deposition. After Cu foam is immersed into graphene oxide (GO) suspension, a spontaneous reduction of GO, induced by Cu, generates reduced graphene oxide on a 3D Cu (rGO@Cu) substrate. The rGO@Cu foam not only provides large surface area to accommodate Li deposition for lowering the local effective current density, but also forms a rGO protective layer to effectively control the growth of Li dendrites. As current collector, the rGO@Cu foam shows superior properties than commercial Cu foam and planar Cu foil in terms of cycling stability and CE. The rGO@Cu foam delivers a CE as high as 98.5% for over 350 cycles at the current density of 1 mA cm−<sup>2</sup> . Furthermore, the full cell using LiFePO<sup>4</sup> as cathode and Li metal as anode with rGO@Cu foam as current collector (LiFePO4/rGO@Cu-Li) is assembled to prove the admirable capacities and indicates commercialization of Li-metal batteries.

### Edited by:

Shengkui Zhong, Soochow University, China

### Reviewed by:

Yunjian Liu, Jiangsu University, China Xianwen Wu, Jishou University, China

\*Correspondence: Zhao Fang fangzhao@xauat.edu.cn

### Specialty section:

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

Received: 21 August 2019 Accepted: 21 October 2019 Published: 27 November 2019

### Citation:

Yu J, Dang Y, Bai M, Peng J, Zheng D, Zhao J, Li L and Fang Z (2019) Graphene-Modified 3D Copper Foam Current Collector for Dendrite-Free Lithium Deposition. Front. Chem. 7:748. doi: 10.3389/fchem.2019.00748 Keywords: Li metal anode, rGO@Cu foam, Li dendrite, current collector, LiFePO<sup>4</sup> cathode

# INTRODUCTION

Electrical energy storage has garnered great attention when considering the problems of pollution and global warming from the burning of fossil fuels and biomass (Goodenough and Kim, 2010). One of the most important candidates is the rechargeable Li battery (Ji et al., 2011; Lu et al., 2014b; Zhao et al., 2015), since it offers higher stored volume and gravimetric energy density (Goodenough and Kim, 2010; Manthiram et al., 2012; Narayanan et al., 2012). Li metal is considered to be a very promising anode material, because it holds quite a higher theoretical specific capacity (3,860 mAh g −1 ), lower density (0.53 g cm−<sup>3</sup> ), and a lowest redox potential −3.04 V vs. the standard hydrogen potential electrode) than other anode materials (Xu et al., 2014; Zhao et al., 2015). However, the safety hazards caused by the Li dendrite limit its real applications. Unlike the commercial "host" material, the Li metal anode undergoes the process of Li deposition/stripping during cycling, leading to uncontrollable dendritic Li via repeated charging and discharging (Ding et al., 2013; Yun et al., 2016; Li Q. et al., 2017; Ma et al., 2017). When Li is deposited on a planar substrate, the small dendrites are formed first, resulting in an electric field that is distributed unevenly, further promoting the Li<sup>+</sup> inhomogeneous concentrate (Yun et al., 2016; Li Q. et al., 2017). As the Li dendrites grow, it can render the breakdown/repair of the solid electrolyte interphase (SEI) repeatedly, owing to continuous interfacial reactions (Liu et al., 2017; Gu et al., 2018; Yan et al., 2018). In this way much of the electrolyte is consumed, severely reducing the Coulombic efficiency and capacity decay (Lee et al., 2015; Ye et al., 2017). At the same time, Li dendrites can penetrate the separator further causing an irreversibly infinite volume change and even catastrophic safety hazards (Song et al., 2011; Cheng et al., 2017; Tang et al., 2018).

As noted above, multifarious strategies have been proposed in order to make the Li metal electrodes viable by controlling Li<sup>+</sup> flux to facilitate uniform deposition of Li and enable formation of a stable SEI (Wu et al., 2017; Li et al., 2018). A variety of electrolyte additives have been employed to modify electrolyte composition and to stabilize the SEI film (Haregewoin et al., 2016). Different kinds of polymeric and ceramic electrolytes have been proven to suppress Li dendrite growth effectively (Kamaya et al., 2011; Kotobuki et al., 2011; Bouchet et al., 2013; Han et al., 2017; Li et al., 2019), however, a series of problems such as chemical instability and low ionic conductivity for application still exists when in contact with Li metal (Li et al., 2015). LiPF<sup>6</sup> was mixed with LiTFSI-LiBOB dual-salt/carbonatesolvent-based electrolytes which can improve the charging capability and cycling stability of Li-metal batteries notably (Zheng et al., 2017), and which cannot achieve stability with the constant consumption of additives in long-term cycles. Another method to suppress Li dendrite growth is planting an artificial SEI layer by manipulating the formation process (Gao et al., 2017) or inserting an interface protective layer (Lu et al., 2014a; Kozen et al., 2015; Lin et al., 2016; Bobnar et al., 2018; Shi et al., 2019). While, there is not enough mechanical strength of the artificial film to suit large volume expansion in extended periods (Yan et al., 2018). It is reported that a flexible, interconnected, hollow amorphous carbon nanosphere coated on Li metal is desired to accommodate the volumetric expansion of Li deposition without mechanical damage (Zheng et al., 2014). In addition, a composite Li metal anode with an ion-conducting mesoscale skeleton can improve electrochemical performance significantly (Liang et al., 2019). Very recently, the other approach to inhibit Li dendrites by modifying the current collector has been widely discussed. It was found that by creating the CuO nanostructure on the Cu surface as a current collector, exhibits an improved capacity for Li batteries (Liu et al., 2016; So et al., 2018; Zhang et al., 2018). Moreover, the 3D conductive framework as a current collector can suppress the growth of dendrites, because of its large specific surface and low local current density for the Li metal anode (Xie et al., 2016; Yun et al., 2016; Li Q. et al., 2017). However, the pore size of the 3D substrate is thought to have a great impact on electrical conductivity. 3D Cu foam that is too large in size is considered unsuitable for current collectors (Yang et al., 2015).

Based on the above-mentioned discussion, we propose a spontaneous formation of rGO@Cu foam via the reaction of 3D Cu foam with graphene oxide to substitute the commercial Cu as a current collector. To prove the excellent properties of the rGO@Cu foam current collector, it was directly used to assemble a half-cell as an electrode with Li metal as the counter electrode. By contrast, 3D Cu foam and 2D planar Cu foil were also directly used separately to assemble batteries with a Li metal sheet as a counter electrode. Compared to three kinds of cells, the rGO@Cu foam inhibits the growth of Li dendrites effectively as well as exhibiting superior properties to 3D Cu foam and planar Cu foil current collectors. The 3D porous structure can reduce the effective electrode current density and provides more space to accommodate Li metal deposition because of its large specific surface. Furthermore, the rGO layer can enhance electrical conductivity and structural stability (Li G. et al., 2017). After cycling, there are minimal dendrites in the 3D Cu foam substrate and the maximum dendrites in the planar Cu foil. Furthermore, the LiFePO4/rGO@Cu-Li full cell was fabricated to estimate the advantageous capabilities of the rGO@Cu foam current collector. The remarkable cycling stability and rate performance of the LiFePO4/rGO@Cu-Li full cells make it possible to utilize in commercial.

# EXPERIMENTAL SECTION

### Materials Preparations

The schematic diagram for the preparation of the rGO@Cu foam is shown in **Figure 1a**. To begin this process, GO was prepared from purified natural graphite according to the Hummers' method (Hu et al., 2013; Zaaba et al., 2017; Bai et al., 2018) (details refer to the **Supplementary Material**). The commercial Cu foam was immersed into the 1 mg/ml graphene oxide (GO) suspension (**Figure S1**) for 12 h until the black reduced graphene oxide layer covered the whole surface. Then, the Cu foam was removed from the GO suspension and transferred to a vacuum drying oven (DZF-2B) for 24 h, after which, the rGO@Cu foam was obtained.

# Material Characterizations

Morphologies of samples were conducted on a field-emission scanning electron microscope (FESEM, NOVA 450, USA). At the same time, Energy-dispersive (EDS) investigation was collected to analyze element distribution of the sample. X-ray photoelectron spectroscopy (XPS) was chosen to characterize the compound of samples. XPS analysis was conducted with the SDTQ600 Type thermo gravimetric analyzer and the XPS spectra were adopted using monochromatic Al Kα (1486.6eV) X-ray emission-spot size was 650 µm.

# Electrochemical Characterizations

The bare Cu foil, 3D Cu foam and rGO@Cu foam were punched out into discs with a diameter of 12 mm and were tested as different current collectors. To evaluate the properties of repeated Li deposition/stripping in planar Cu foil, commercial Cu foam, and rGO@Cu foam current collectors, CR2016 coin cells were directly assembled using the three kinds of current collectors, with bare Li metal as counter electrodes. The separator was a Celgard-2325 microporous polypropylene film. The electrolyte was a 1 M lithium bis (trifluoromethanesulphonyl) imide (LiTFSI) in 1,3-dioxolane (DOL) and 1,2- dimethoxyethane (DME) (volume ratio 1:1) with 2% lithium nitrate (LiNO3) without any additives. Before the symmetrical cell test, the current collectors were first initialized by cycling from 0 to 1 V at 50 µA for five cycles to remove

surface contaminations and to stabilize the interface. Then 1 mAh cm−<sup>2</sup> of Li was plated on the current collector at a current density of 1 mA cm−<sup>2</sup> and the cutoff potential for the discharge process was set to 1.0 V. To confirm the possibility of a practical application of the modified current collector, the LiFePO<sup>4</sup> full cell was built using LiFePO<sup>4</sup> as a cathode and Li metal as an anode with 2D planar Cu foil (LiFePO4/2D-Cu-Li), 3D Cu foam (LiFePO4/3D-Cu-Li), or a rGO@Cu foam current collector. 1 mAh cm−<sup>2</sup> of Li was first deposited onto the different Cu current collectors by a half-cell using Li foil as a counter electrode. The cell was then disassembled in an argon-filled glovebox and a new anode was reassembled into a full cell with LiFePO<sup>4</sup> as the cathode (for details refer to the **Supplementary Material**). The electrolyte was 1.0 M LiPF<sup>6</sup> in ethylene carbonate (EC)/dimethyl carbonate (DMC) (1:1 in volume). The LiFePO<sup>4</sup> full cells were cycled between 2.0 and 4.2 V at 1 C. All the cells were assembled in the argon-filled glovebox and were tested using the LANHE multi-channel battery testing system and PARSTAT 2273 Electrochemical System (Princeton Applied Research, USA).

# RESULTS AND DISCUSSION

The pictures of planar Cu foil, bare Cu foam current collector and rGO@Cu foam are shown in **Figure S2**. A bare Cu foam of reddish brown changed into black after soaking in the GO suspension, showing that the graphene is evenly covered on bare Cu foam. The top view SEM images of commercial Cu foam at different magnifications are shown in **Figures 1b–d** and rGO@Cu foam is shown in **Figures 1e–g**. An abundance of pores can be observed in the SEM images of both commercial Cu foam and rGO@Cu foam, confirming the 3D porous structure of samples. Meanwhile, it can be clearly seen that rGO is attached to the Cu skeleton. **Figure 2a** displays the top view SEM images of a modified Cu foam current collector. Compared with the bare Cu foam, there is a layered structure connected to the Cu skeleton (**Figure S3**). This layered structure is considered to be rGO because the GO is spontaneously reduced by Cu, and Cu is oxidized to form CuO. The SEM image of the surface of the layered structure is magnified on **Figure 2b** and the corresponding elemental mappings are shown in **Figures 2c,d**. It can be discovered that only C (**Figure 2c**) and O (**Figure 2d**), as major elements, are distributed evenly and independently on the layered structure. To further illustrate the composite of Cu foam with GO, XPS spectra of samples are collected in **Figures 2e–h**. **Figure 2e** exhibits the XPS survey spectra measured in the range of binding energies from 0 to 1,361 eV and the peaks of C 1s, O 1s, and Cu 2p can be clearly observed. As shown in **Figure 2f**, the intensities of the C-O (286.25 eV) and C=O (288.38 eV) functional groups are significantly decreased compared with the

spectra of GO (Stankovich et al., 2007; Luo et al., 2011; Kim et al., 2013). At the same time, the main peak at 284.75 eV indicates that most of the C-C functional group was formed (Stankovich et al., 2007). This result can be explained by the fact that GO is reduced to rGO. The high-resolution scan of Cu 2p is displayed in **Figure 2g**, a double peak with binding energies at 932.75 and 952.75 eV is attributed to Cu 2p3/2 and Cu 2p1/2 of Cu2<sup>+</sup> in CuO, respectively. Also, the gap between the Cu 2p3/2 and Cu 2p1/2 energy levels is 20.0 eV, corresponding to the split orbit for Cu2<sup>+</sup> (Soleimani and Moghaddami, 2017). These results can be considered to show that the Cu was oxidized to CuO. To further prove the existence of oxygen-containing functional groups and CuO, the O 1s core-level spectrum is shown in **Figure 2h**. There are mainly three peaks centered at 532.88, 531.82, and 530.7 eV, which is approximately consistent with the O-C, O=C, and O-Cu (Novakov and Prins, 1971; Mattevi et al., 2009). The peaks of O 1s of the sample are significantly increased compared with GO, indicating that most of the oxygen functional groups are reduced and the rGO is formed (Kim et al., 2013). In addition, the existence of O-Cu further proves the formation of CuO. As a result, the existence of rGO and CuO without any external conditions proves that the spontaneous reaction is real.

The electrochemical performance of the coin cells with 2D planar Cu foil, 3D commercial Cu foam, and rGO@Cu foam as an electrode with Li metal as a counter electrode is displayed in **Figure 3**. The coin cells with three types of current collectors were first cycled five times from 0 to 1 V at 50 µA to remove surface impurities and to help form a stable SEI film. **Figures 3a–d** exhibit the galvanostatic cycling voltage profiles on the 2D planar Cu foil, 3D commercial Cu foam, and rGO@Cu foam electrode for cycles at 10th, 50th, 150th, and 300th. After the initial 10 cycles, 2D planar Cu foil and 3D commercial Cu foam display a higher discharge voltage plateau and a lower charge voltage plateau than rGO@Cu foam (**Figure 3a**), indicating a much lower polarization of planar Cu foil and 3D Cu foam compared to rGO@Cu foam. This might be attributed to the fact that the surface of rGO@Cu foam was not activated completely by the electrolyte, causing the accumulation of Li<sup>+</sup> on the current collector. Notice that this phenomenon disappears after fifty cycles (**Figure 3b**). In addition, compared with the 50th (**Figure 3b**), 150th (**Figure 3c**), and 300th (**Figure 3d**) cycles, a significant capacity decay of planar Cu foil is observed, which may be due to the growth of Li dendrites further leading to the formation of dead Li and the repeated consumption of the electrolyte caused by unstable SEI. For commercial Cu foam, the capacity decay appears about 300 cycles, illustrating that there is less irreversible Li deposition on the 3D Cu and more stable SEI than planar Cu. By contrast, the galvanostatic charge/discharge profiles of the rGO@Cu foam shows outstanding stability even after 300 cycles, and rGO@Cu foam also exhibits a superior capacity retention. These results demonstrate that the deposition of Li on rGO@Cu foam is more uniform and the formation of SEI is more stable than planar Cu and commercial Cu foam. To evaluate the long-term cycling stabilities of the Li anode with the three types of current collectors, the symmetric cells were assembled and examined at a current density of 1 mA cm−<sup>2</sup> with a capacity of 1 mAh cm−<sup>2</sup> . As shown in **Figure S4**, after 84 h, an abrupt voltage drop is detected for the planar Cu foil current collector with fluctuating voltage in the later hours. This could be explained by an internal soft short-circuit with Li dendrite penetration (Lin et al., 2016). A distinct decrease in polarization is observed in commercial Cu foam, in the initial 80 h, may be due to the gradual stabilization of the SEI layer. By contrast, the 3D rGO@Cu current collector maintains a much lower and more stable hysteresis after 400 h, illustrating that its extended reduced graphene layer can reduce the local current density and inhibit the growth of Li dendrites. The cycling Coulombic efficiencies of 2D Cu foil, 3D commercial Cu foam, and rGO@Cu foam current collectors were examined at a current density of 1 mA cm−<sup>2</sup> . CE is computed by the ratio of the total amount of Li stripped away, vs. the deposited amount on the current collector in each cycle. As shown in **Figure 3e**, the CE of 2D Cu foil is stabilized at 97% during the 60 cycles, and then the CE is reduced gradually to <92% after 110 cycles. This may be ascribed to the irreversible Li deposition because of the formation of Li dendrites and dead Li. As for the commercial Cu foam current collector, the CE can reach more than 97% and be sustained during 225 cycles, showing that the Cu foam current collector with a special porous structure does help to accommodate more Li+, hence hindering the growth of Li dendrites and extending the cycle life, whereas the CE of the 3D Cu foam becomes unstable after only 260 cycles and drops to below 96%. The rGO@Cu foam as current collector was conducted to further develop the cycling performance in comparison to 2D planar Cu foil and commercial Cu foam. The Columbic efficiency of rGO@Cu foam maintains a stability as high as 98.5% for over 350 cycles, which demonstrates that the covered reduced graphene oxide on the 3D Cu foam has significant effects and improves the Columbic efficiency and cycling stability in contrast to commercial Cu foam. To further explain the effect of the reduced graphene oxide layer on the improvement of the Cu foam interfaces, EIS was elected to evaluate the interfacial resistance of the planar Cu foil, commercial Cu foam, and rGO@Cu foam current collectors. The Nyquist plots of the three types of current collectors is displayed in **Figure 3f**. Comparing the intercept and the diameter of the first semicircle in each sample, the rGO@Cu foam possesses smaller SEI film resistance (4 cm<sup>2</sup> ) than commercial Cu foam (5.9 cm<sup>2</sup> ) and planar Cu foil (6.5 cm<sup>2</sup> ), which could be attributed to the stable SEI film modified by the reduced graphene layer. In addition, the second semicircle reflects the charge transfer (interfacial) resistance Rct (Schipper et al., 2018; Wu et al., 2019). The rGO@Cu foam and commercial Cu foam have considerably smaller charge transfer resistance (24 and 30 cm<sup>2</sup> ) compared with planar Cu foil (184 cm<sup>2</sup> ). It suggests that a large amount of Li dendrites and dead Li are deposited on the surface of Cu foil. All the results confirm that the rGO@Cu foam current collector possesses superior properties such as higher electrochemical stability, better capacity retention, and Li dendrite suppression.

**Figure 4** illustrates the behaviors of Li deposition on the 2D planar Cu, 3D commercial Cu foam, and rGO@Cu foam current collectors. Li is deposited on the substrate and unavoidably form some protuberances on the surface of planar Cu foil (**Figure 4a**), leading to the electric field being distributed unevenly, further promoting Li accumulation at the protuberances, finally growing into Li dendrites or even forming "dead Li." On the other hand, the formation of Li dendrites destroys SEI and further leads to the repeated formation of SEI film. Furthermore, these problems could deteriorate during cycling. In a Cu foam current collector (**Figure 4b**), owing to its special 3D porous structure, more space could be provided to accommodate Li. The porous current collector exhibits lower interfacial resistance and local current density compared to planar Cu, therefore Li<sup>+</sup> will distribute homogeneously on the surface of the 3D Cu foam and few Li dendrites will be formed. However, the phenomenon of Li dendrites still worsens after cycling. For the rGO@Cu foam current collector (**Figure 4c**), Li dendrites are restrained effectively. The rGO@Cu foam current collector displays outstanding electric conductive and mechanical strength for Li deposition. Li favors deposition on the inner surface of

metal as a counter electrode at a current density of 1 mA cm−<sup>2</sup> with the deposition capacity at 1 mAh cm−<sup>2</sup> . (f) Nyquist plots of Cu foil, Cu foam, and rGO@Cu foam. The impedance is obtained after 300 cycles with a current density at 1 mA cm−<sup>2</sup> and a cycling capacity of 1 mAh cm−<sup>2</sup> .

the rGO@Cu foam current collector. This obviously strengthens the stability of SEI and reduces the consumption of Li. In order to clearly observe the morphology of Li dendrite growth, the SEM images of the surfaces of current collectors are shown in **Figures 4d–f** with high magnifications. In the planar Cu foil, many Li filaments (**Figure 4d**) with lengths of <3µm (**Figure S5**a) could be discovered after long-term cycling, leading to penetration of the separator and causing short circuiting of the cell. In addition, a loose structure is composed as a result of the formation of Li dendrites and dead Li. This loosely aggregated structure further gives rise to the electrical field being distributed unevenly and accelerates Li being deposited unequally. By contrast, for a 3D commercial Cu foam, there are only a few fibrous Li and some mossy Li formed on the

surface of the current collector (**Figure 4e**), indicating that the porous structure of the 3D Cu foam current collector revealed an excellent property to accommodate the expansion of the deposited Li. **Figure S5**b shows a magnified surface of 3D Cu foam after cycling, illustrating that it provides a more stable working electrode structure and interface than planar Cu foil. Moreover, compared with commercial Cu foam, the SEM images of rGO@Cu foam, as shown in **Figure 4f** and **Figure S5**c, clearly show that the surface of this composite electrode is remarkably smooth in general without obvious Li dendrites or mossy Li under the same conditions. This result may be explained by the fact that the rGO@Cu foam current collector can prevent the growth of Li dendrites and remove the potential of Limetal battery hazards. This flat surface could be due to the rGO@Cu foam combined with the benefits of the rGO and 3D porous structure. rGO works as a protective layer to inhibit the growth of Li dendrites and the 3D porous structure promotes uniform deposition of Li into the interspace of the rGO@Cu foam current collector, improving cycling stability and the lifespan of Li-metal batteries.

To further investigate the battery performance under practical applications, the LiFePO4/2D-Cu-Li, LiFePO4/3D-Cu-Li, and LiFePO4/rGO@Cu-Li full cells were assembled and galvanostatically cycled from 2.0 to 4.2 V at 1 C. **Figure 5a** exhibits the long-term cycling performance of the LiFePO4/2D-Cu-Li, LiFePO4/3D-Cu-Li, and LiFePO4/rGO@Cu-Li full cells at 1 C. For LiFePO4/2D-Cu-Li cells, the discharge capacity is about 132.2 mAh g−<sup>1</sup> in the first cycle and starts to decrease obviously after 15 cycles. The reversible capacity of LiFePO4/2D-Cu-Li cells is dropped to 57.6 mAh g−<sup>1</sup> with a CE of 95.7% at 200 cycles. The depletion of the electrolyte is due to the continuous decomposition at Li metal and the formation of "dead Li." While as for the LiFePO4/3D-Cu-Li cells, the reversible capacity degrades gradually from the beginning with 131.7 to 98.6 mAh g −1 and the CE is maintained at about 99% during 200 cycles. It can be seen that the cycling performance of LiFePO4/3D-Cu-Li cells is greatly improved compared with LiFePO4/2D-Cu-Li. By contrast, the LiFePO4/rGO@Cu-Li cells exhibit a higher initial capacity of 132.9 mAh g−<sup>1</sup> and better capacity retention with a discharge capacity of 126.2 mAh g−<sup>1</sup> after 200 cycles. Moreover, it shows a more stable CE of 99% for more than 200 cycles, illustrating that the LiFePO4/rGO@Cu-Li cells display outstanding capacity retention and cycling stability. **Figure 5b** shows the rate performance of these three kinds of LiFePO<sup>4</sup> full cells with 2D Cu, 3D Cu foam, and rGO@Cu foam current collectors under different current densities. The rate capability of cells was evaluated under cycling at rates varying from 0.2 C to 5 C with intervals of 10 cycles. It can be observed that the capacity almost goes back to the previous corresponding value when the rate is recovered to 0.5 C, and all the full cells display obvious capacity decay with current density increases. The LiFePO4/rGO@Cu-Li cells exhibit a capacity of 88 mAh g−<sup>1</sup> at 5 C, indicating its enhanced discharge capacity, especially at a high cycling rate. While under the same conditions,

the capacity of LiFePO4/2D-Cu-Li cells almost reach 67 mAh g −1 and LiFePO4/3D-Cu-Li cells are 70 mAh g−<sup>1</sup> , which is still less than the LiFePO4/rGO@Cu-Li cells. In addition, the LiFePO4/rGO@Cu-Li cells always remain at the highest capacity compared to LiFePO4/2D-Cu-Li and LiFePO4/3D-Cu-Li at different current densities. These results indicate that the LiFePO4/rGO@Cu-Li shows excellent rate performance and a higher capacity, demonstrating its great potential for practical application.

### REFERENCES


# CONCLUSIONS

In summary, in this study, we have reported a simple but effective strategy to suppress the growth of Li dendrites and to facilitate the uniform deposition of Li using a rGO@Cu foam as a current collector. After soaking in GO suspension, the commercial Cu foam is covered with a layered structure of rGO. The rGO@Cu foam exhibits superior performance because the rGO layer works as a protective film to inhibit the growth of Li dendrites. Furthermore, its special porous structure can provide more space to accommodate Li metal deposition and reduces the effective electrode current density. The half-cell using rGO@Cu foam as an electrode and Li metal as a counter electrode exhibits a high CE of above 98.5% after 350 cycles under the current density of 1 mA cm−<sup>2</sup> . As for the LiFePO4/rGO@Cu-Li full cells, the reversible capacity is maintained at 126.2 mAh g −1 and the CE can remain as high as 99% after 200 cycles at 1 C. The LiFePO4/rGO@Cu-Li full cells also show a stable rate performance compared to LiFePO4/2D-Cu-Li and LiFePO4/3D-Cu-Li cells. The superior properties of the rGO@Cu foam current collector are beneficial in achieving the effective suppression of Li dendrites for next-generation Li-metal battery applications.

# DATA AVAILABILITY STATEMENT

All datasets generated for this study are included in the article/**Supplementary Material**.

# AUTHOR CONTRIBUTIONS

All authors contributed to manuscript revision, read, and approved the submitted version.

## FUNDING

This study was funded by the National Natural Science Foundation of China (51974219, 51974222); Natural Science Basic Research Plan in Shaanxi Province (2019JQ-764, 2018JM5135); and Project (18JK0474) supported by Shaanxi Provincial Education Department, China.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2019.00748/full#supplementary-material

metallic lithium for application in the high energy batteries. Sci. Rep. 8:5819. doi: 10.1038/s41598-018-23991-2


Cu foil for lithium ion battery anode. Diam. Relat. Mater. 68, 102–108. doi: 10.1016/j.diamond.2016.06.013


**Conflict of Interest:** 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 Yu, Dang, Bai, Peng, Zheng, Zhao, Li and Fang. 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.

# Effect of Aging-Induced Dioxolane Polymerization on the Electrochemistry of Carbon-Coated Lithium Sulfide

Lucas Lodovico1,2, Alberto Varzi 1,2 and Stefano Passerini 1,2 \*

*<sup>1</sup> Helmholtz Institute Ulm (HIU), Ulm, Germany, <sup>2</sup> Karlsruhe Institute of Technology (KIT), Karlsruhe, Germany*

Lithium sulfide-based materials have been considered as potential positive electrodes for the next generation batteries. Lithium sulfide is the fully lithiated form of sulfur, i.e., they share the same high theoretical capacity. However, it has the benefit of already containing lithium, which allows making cells with lithium-free negative electrodes. Lithium sulfide, however, shares with sulfur the polysulfide dissolution drawback upon cycling. One possible solution to this problem is to envelop the active material particles with carbonaceous materials. In this work, we investigate the effect of a nitrogen-rich carbon coating on lithium sulfide particles. The effect of such coating on the surface properties and electrochemistry of lithium sulfide cathodes is investigated in details, in particular, regarding its interaction with fresh vs. aged electrolyte. The polymerization of dioxalane (DOL) due to aging is found to affect the electrochemistry of lithium sulfide and, interestingly, to improve the cycling performance.

### Edited by:

*Shengkui Zhong, Soochow University, China*

### Reviewed by:

*Donghai Wang, Pennsylvania State University (PSU), United States Yanwei Li, Guilin University of Technology, China*

> \*Correspondence: *Stefano Passerini stefano.passerini@kit.edu*

### Specialty section:

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

Received: *13 September 2019* Accepted: *11 December 2019* Published: *10 January 2020*

### Citation:

*Lodovico L, Varzi A and Passerini S (2020) Effect of Aging-Induced Dioxolane Polymerization on the Electrochemistry of Carbon-Coated Lithium Sulfide. Front. Chem. 7:893. doi: 10.3389/fchem.2019.00893* Keywords: lithium-sulfur battery, lithium sulfide, dioxalane, electrolyte aging, N-rich carbon-coating, ethylenediamine

### INTRODUCTION

Lithium-sulfur batteries (LSBs) have attracted much attention in recent years, with the prospect of replacing lithium-ion batteries (LIBs) as energy source for automotive and, especially, flight applications as soon as 2030 (International and Agency, 2018). The main reason for this is the high theoretical capacity of sulfur, around 1,675 mAh g−<sup>1</sup> , promising Li-S batteries with energy densities (based on the electrode materials only) as high as 2,600 Wh kg−<sup>1</sup> . Sulfur-based cathodes, however, still possess a multitude of problems that need to be solved. The main challenge posed by sulfur is the high solubility of the intermediates formed during cycling (Xu et al., 2015). These intermediates take the form of polysulfides with general formula Li2Sx, which are chain of sulfur atoms with x usually comprised between 3 and 8. At the end of discharge, solid discharge products (Li2S and Li2S2) are finally formed. The first and most obvious consequences of polysulfides dissolution are low Coulombic efficiencies and loss of active material during battery operation.

In order to avoid these effects, a commonly used strategy is to physically constrain the polysulfides in the positive electrode (cathode), often by enveloping the active material's particles with an inert coating, such as carbon (Jeong et al., 2013; Agostini et al., 2014; Chen et al., 2014; Nan et al., 2014; Hwa et al., 2015; Wu et al., 2016). Carbon coating of sulfur, however, is all but trivial, owing to its low melting point and tendency to easily sublimate at moderate temperature. In these terms, the use of lithium sulfide as cathode active material presents interesting advantages. Besides being already in the lithiated state, which allows it to be paired with lithium-free negative (anode) electrodes (Nanda et al., 2018), lithium sulfide is thermally stable up to rather high temperature allowing processing such as carbonization.

In this work, we investigate the effect of a nitrogenrich carbon coating on lithium sulfide particles. Specifically, ethylenediamine-embedded lithium sulfide, previously reported by our group (Lodovico et al., 2019), is pyrolized in order to obtain carbon-coated lithium sulfide, with the ethylenediamine molecules serving as the carbon source. The use of this kind of diamine lead to a N-doped carbon coating with C:N weight ratios around 2:1. The effect of such coating on the surface properties and electrochemistry of lithium sulfide cathodes is investigated in details, in particular, regarding its interaction with fresh vs. aged electrolyte. It was found that upon prolonged storage (12+ months) the dioxolane-based electrolyte tends to polymerize, which greatly affects the behavior of Li2S-based cathodes.

### METHODS

Chemical and electrode preparations, infrared measurements, and cell assembly were carried out in a glove box (MBraun) filled with argon (O<sup>2</sup> and H2O content below 0.1 ppm). The 1:1 volume ratio mixture of Dimethoxyethane (DME, Solvionic) and 1,3-Dioxolane (DOL, Solvionic) was used as the electrolyte solvent. Prior to mixing, the two solvents were individually dried using 3Å molecular sieves (Sigma Aldrich) to reach water contents below 20 ppm as determined via Karl-Fischer titration (Mettler-Toledo Titrator Compact C30). Appropriate amounts of lithium bis(trifluoromethanesulfonyl)imide (LiTFSI, Solvionic) and lithium nitrate (LiNO3, Alfa Aesar) were dissolved into the DME-DOL solvent mixture to achieve concentration of 1 and 0.25 mol L−<sup>1</sup> , respectively, using aluminum bottles. This electrolyte was used soon after its preparation (i.e., within 2 weeks; labeled as "fresh") or after storage (i.e., after more than 1 year, labeled "aged"). Besides the aging, both the electrolytes were stored in sealed aluminum bottle inside the glove box.

Carbon-coated lithium sulfide (Li2S-CC) was prepared using ethylenediamine-containing lithium sulfide (Li2S-En) as precursor. Li2S-En synthesis, carbon coating procedure, and electrode preparation is described in a previous work (Lodovico et al., 2019). The final material possesses a particle coating with a C:N weight ratio of about 2:1.

Li2S-CC was employed as active electrode materials. The electrodes were prepared by manually grind-mixing the active material (Li2S-CC) with the conductive carbon (Super C65; Imerys Graphite & Carbon) and the binder (polyvinylidene fluoride; PVdF 6020 from Solvay). The respective weight ratios were 53:37:10. The mixed powders were then dispersed with N-Methyl-2-pyrrolidone (NMP, Sigma Aldrich), and the resulting slurry cast over aluminum current collectors using the doctor blade method. Finally, the electrode tapes were dried at 60 ◦C under Ar.

For the determination of the electrolyte electrochemical stability window, carbon electrodes composed of Super C65 and PVdF were prepared in the 85:15 weight ratio. NMP was added to prepare the slurry, which was cast over the aluminum current collector as well.

Carbon electrodes were also prepared using nitrogenrich carbon derived from carbonization of 1-ethyl-3 methylimidazolium tricyanomethanide (EMImTCM, IoLiTec). In short, EMImTCM was placed in an alumina boat inside a horizontal tube furnace under argon flow. The temperature was raised to 500◦C at a rate of 5◦C min−<sup>1</sup> and kept for 5 h. The resulting carbonized sample (N-rich carbon) was composed of 62% carbon, 33% nitrogen, 2% hydrogen, and 3% oxygen (all values given as weight percentages). This composition is rather similar to that of the carbon coating in Li2S-CC. The N-rich carbon was used to prepare electrodes as well, in which Super C-65, N-rich carbon, and PVdF were mixed in a 70:15:15 weight ratio, dispersed in NMP, and cast over aluminum foil, and dried at 120◦C under vacuum for 48 h.

Electrochemical tests were performed using three electrode Ttype cells (Swagelok), with lithium metal as both the reference and counter electrodes. The cells were assembled using 100 µL of electrolyte soaked into glass fiber separators (Whatman GF/A). For the capacity retention experiments, the cells stored at 20◦C in climatic chamber (Binder) were galvanostatically cycled using a battery tester (Maccor S4000). The Li2S-CC based cells were activated by charge at a C/20 rate (1C = 1,165 mAh g−<sup>1</sup> ) up to 4 V vs. Li/Li+, followed by cycling at C/10 between 1.9 and 3.0 V vs. Li/Li+.

Electrochemical impedance spectroscopy (EIS) measurements were performed using a BioLogic VMP3 multi-channel potentiostat. For that, the cells were cycled using the same procedures as mentioned above. However, after each discharge, the cell was allowed to rest at open-circuit for 2 h prior to measuring impedance. Spectra were recorded between frequencies of 200 kHz and 100 mHz, using a 5 mV signal amplitude.

The anodic electrochemical stability of the electrolytes was studied by cyclic voltammetry (CV), where the potential was scanned from 3 to 4 V vs. Li/Li+. After cycling, each cell was disassembled inside a glove box, and the electrolyte recovered by centrifuging the glass fiber separators at 6,000 rpm for 10 min. The recovered electrolytes were analyzed via ATR-IR.

### RESULTS AND DISCUSSION

**Figure 1A** shows the capacity retention of Li2S-CC-based electrodes when using either the fresh (Li2S-CC\_F) or the aged (Li2S-CC\_A) electrolyte. It is evident that the electrolyte aging has a large impact on the cell performance. Surprisingly, the aged electrolyte provides higher discharge capacity as well as capacity retention upon 100 cycles. In order to better understand the reason behind this effect, the discharge curves of the cathode in each electrolyte were closely examined. The inset of **Figure 1A** shows the typical discharge curve of a Li2S-based electrode after the initial activation charge leading to fully oxidized S. The first region (I) shows a potential plateau (around 2.4 V vs. Li/Li+) followed by a slopy region, both associated with the reduction of molecular sulfur (S8) to medium chain polysulfides

FIGURE 1 | (A) Capacity retention for cells using carbon-coated Li2S either in fresh (Li2S-CC\_F) or aged electrolyte (Li2S-CC\_F). Inset: Typical discharge profile showing the two main discharge plateaus. Evolution of the capacity of (B) region I and (C) region II for selected cycles.

(Li2Sx, x = 4–6). Afterwards, a second region (II) characterized by a long plateau at lower potential (around 2.1 V vs. Li/Li+) appears, resulting from the reduction of lithium polysulfides to form the final solid discharge products (Li2Sx, 1 ≤ x ≤2) (Xu et al., 2015).

**Figures 1B,C** show the evolution of the capacity arising in regions I and II, respectively, for a few selected cycles (1st to 10th and 50th to 60th cycle). The individual capacity of each region is calculated as shown in the inset of **Figure 1A**. Namely, Region I is delimited between the beginning of discharge until the dip in potential before the low-potential plateau. Region II starts at the dip in potential until the end of discharge. As shown in **Figure 1B**, the capacity in the first region decrease sharply during the first 10 cycles to stabilize, however, after the 50th cycle. This indicates that, in both cells, less and less elemental sulfur (S8) is formed in each charge, until a minimum is reached. In **Figure 1C**, the evolution of the capacity associated with the second region shows a considerably more peculiar behavior. It is clearly observed that the initial lower capacity of the fresh electrolyte cell with respect to that employing the aged one mostly arose from region II, where the solid discharge products are formed. In fact, while the Li2S-CC\_A cell delivered a stable capacity upon cycling, the Li2S-CC\_F cell showed a lower and more rapidly decreasing capacity in this region. This indicates that understanding the effect of the N-rich carboncoating and the aged electrolyte on this voltage region is vital to a deeper comprehension of the lithium sulfide electrochemistry.

First of all, the electrolytes were studied by IR spectroscopy. **Figure 2** shows the infrared spectrum of a freshly prepared electrolyte (fresh), as well as that of one (with the same composition) after storage for at least 12 months (aged). Clearly, there are significant changes occurring during storage. The band at around 1,450 cm−<sup>1</sup> due to the –CH<sup>3</sup> deformation in DME (Bailey, 1985) has a constant intensity before and after aging, showing that DME is stable. On the contrary, the band at 960 cm−<sup>1</sup> due to the ring-breathing mode of DOL (Yang et al., 2005) shows an appreciable decrease upon storage. In fact, DOL is known to be relatively unstable (Okada et al., 1964), being prone to ring-opening reactions such as isomerization and polymerization (Okada et al., 1964). Isomerization of DOL leads to the formation of ethyl formate. However, no band corresponding to the vibration of the carbonyl group can be

detected in the aged electrolyte, thus ruling out this option. On the other hand, polymerization of DOL has been extensively studied and known to happen when a cationic initiator is present (Okada et al., 1964; Berman et al., 1969; Yang et al., 2005), leading to the formation of polydioxolane (polyDOL, **Figure 2B**), a polyether. The formation of polyDOL in the aged electrolyte is proven by the increase in the band at around 1,190 cm−<sup>1</sup> , which is caused by the vibrations of the C-O bonds in linear molecules (Yoshida and Matsuura, 1998). Being a cyclic compound, C-O-C vibrations in DOL involve the whole ring and are thus heavily shifted (Makarewicz and Ha, 2001). The exact cause of this polymerization, in this case, is still unknown. Although DOL is known to be reactive, polymerization only happens in the presence of an appropriate initiator, usually one that can promote chain growth through a cationic mechanism. Analysis of the pure solvent mixture (1:1 DME:DOL) without addition of salt and stored under the same conditions for the same period of time shows no change in its spectrum, such that the culprit is likely the salts added to form the electrolyte (LiNO<sup>3</sup> and LiTFSI). Li<sup>+</sup> is known to coordinate the oxygen atoms in ether molecules (Blint, 1995), placing a partial positive charge on it. This weak initiating ability could explain the long storage times necessary to observe the polymerization appreciably.

In summary, the polymerization of DOL occurring upon prolonged storage of the electrolyte appears to be beneficial in Li2S-CC\_A cell, which shows a higher capacity than the Li2S-CC\_F cell (**Figure 1A**) mainly due to a higher and more stable capacity delivered in region II (**Figure 1C**). Similar results have been already reported in literature (Li et al., 2016). The comparison of electrolytes based on DME:DOL, TEGDME:DOL, and poly(ethylene glycol) dimethyl ether (PEGDME):DOL, for which the chain length increases in the aforementioned order, showed an increase in capacity of up to 2.5 times for the PEGDME:DOL based electrolyte compared to the DME:DOL based one. In order to better clarify the reason for the Li2S-CC cell behavior, the EIS response of all electrodes in the discharged states upon cycling was collected.

The spectra of Li2S-CC\_F (**Figure 3A**) show a small high frequency (HF) semicircle evolving into a pseudo-inductive loop, followed by the beginning of a large, partially depressed, semicircle at low frequencies (LF). The pseudo-inductive loop has been related to intermediate species, i.e., polysulfides, adsorbed on the electrode's surface (Ding et al., 2013; Jeong et al., 2013; Lin et al., 2013; Xie et al., 2017). The depressed LF semicircle is typical of charge-transfer reactions, and corresponds to the reversible conversion of polysulfides to solid products. However, for Li2S-CC\_A (**Figure 3B**) some differences are clearly observed, the most relevant being the large HF semicircle before the pseudoinductive loop. This feature may be associated to the presence of a surface layer on the electrode. The formation of a passivation layer is not unexpected. In fact, during the activation step the cathode is charged up to 4 V vs. Li/Li+, while the electrolyte is only stable up to 3.6 V vs. Li/Li<sup>+</sup> (Yim et al., 2013). To test the effect of the N-rich carbon-coating on the electrode passivation, the anodic stability window of carbon electrodes (Li2S-free) was investigated.

**Figure 4** shows the anodic stability limit measurements of the electrolytes performed using carbon electrodes containing either the conductive carbon only (labeled as SC 65) or the mixture of conductive carbon and nitrogen-rich carbon (labeled as SC 65 + N-rich C), the latter mimicking carbon coating in Li2S-CC electrodes. As seen in **Figures 4A,B**, the two electrolytes showed oxidation peaks above 3.5–3.6 V vs. Li/Li+. For the fresh electrolyte, comparable peaks were observed with both kinds of carbon electrodes in the first cycle. The same occurred in the second cycle, where almost no oxidation current was recorded, indicating that the carbon electrodes underwent a passivation process upon the previous scan. However, the recorded peak current was substantially larger with the aged electrolyte,

especially for the electrode solely composed by SuperC 65 and the binder.

The electrolyte was then extracted from such cells and analyzed. The infrared spectra of the electrolytes after cycling (**Figure 4C**) show the disappearance of the DOL band, indicating its consumption to form the passivation layer on the carbon electrodes. The aged electrolyte (**Figure 4D**) contains less DOL than the fresh one even before cycling, due to the age-induced polymerization. However, there is only a little, if any, change in the spectra of the recovered aged electrolyte in comparison to the pristine aged electrolyte. Still, the cyclic voltammograms show pretty clearly that the oxidation current of the electrode using only Super C65 is much larger than that of the electrode containing also the N-rich carbon. This is a strong indication that the formation of a passivation layer still occurred, but at a much reduced extent on the electrode containing the N-rich carbon. Though the exact mechanism is not clear yet, it can be proposed that the N-rich carbon layer facilitates the anchoring of the PolyDOL from the aged electrolyte, resulting in the formation of a protective layer on the carbon-coated Li2S, which leads to the improved performance of such electrodes in the aged electrolyte.

### CONCLUSIONS

In summary, the effect of N-rich carbon-coating of Li2S-based cathode was studied. The presence of polyDOL in the aged electrolyte has an unexpected, but appreciable impact on the performance of the Li2S electrodes, improving the performance of the N-rich carbon-coated Li2S electrodes. The obtained results suggest that PolyDOL can form a stable, hydrophilic passivation layer over the N-rich carbon-coating, which improves the lithium sulfide formation during discharge leading to larger discharge capacities.

# DATA AVAILABILITY STATEMENT

The datasets generated for this study are available on request to the corresponding author.

# AUTHOR CONTRIBUTIONS

LL performed the experiments. LL and AV analyzed the experimental data. All authors wrote the manuscript and conceived the work.

# ACKNOWLEDGMENTS

LL would like to acknowledge the financial support from CNPq/Science without Borders program [205186/2014-9]. AV and SP acknowledge the financial support of the Helmholtz Association. The data presented here was partially used in LL's Doctorate dissertation (Lodovico, 2018).

# REFERENCES


**Conflict of Interest:** 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 © 2020 Lodovico, Varzi and Passerini. 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.

# Fluorine-Doped Carbon Coated LiFePO3.938F0.062 Composites as Cathode Materials for High-Performance Lithium-Ion Batteries

Zhixiong Yan1,2, Dequan Huang1,2, Xiaoping Fan1,2, Fenghua Zheng1,2 \*, Qichang Pan1,2 , Zhaoling Ma1,2, Hongqiang Wang1,2, Youguo Huang1,2 and Qingyu Li 1,2 \*

*<sup>1</sup> School of Chemical and Pharmaceutical Sciences, Guangxi Normal University, Guilin, China, <sup>2</sup> Guangxi Key Laboratory of Low Carbon Energy Materials, Guangxi Normal University, Guilin, China*

### Edited by:

*Feixiang Wu, Central South University, China*

### Reviewed by:

*Dianlong Wang, Harbin Institute of Technology, China Jin Tao Zhang, Nanyang Technological University, Singapore Wang Ding, Kunming University of Science and Technology, China Jia Feng Zhang, Central South University, China*

### \*Correspondence:

*Fenghua Zheng zhengfh870627@163.com Qingyu Li liqingyu62@126.com*

### Specialty section:

*This article was submitted to Energy Materials, a section of the journal Frontiers in Materials*

Received: *30 August 2019* Accepted: *17 December 2019* Published: *24 January 2020*

### Citation:

*Yan Z, Huang D, Fan X, Zheng F, Pan Q, Ma Z, Wang H, Huang Y and Li Q (2020) Fluorine-Doped Carbon Coated LiFePO3.938F0.062 Composites as Cathode Materials for High-Performance Lithium-Ion Batteries. Front. Mater. 6:341. doi: 10.3389/fmats.2019.00341* Fluorine-doped carbon coated olivine LiFePO3.938F0.062 composite (LFPF/CF) is synthesized by a simple solid-state reaction method, and the Tween40 and polyvinylidene fluoride (-(CH2-CF2)n-, PVDF) were used as carbon source and fluorine sources, respectively. Benefiting from the Tween40 (C22H42O6(C2H4O)n) is attributed to formation a homogeneous carbon layer on the surface of LiFePO<sup>4</sup> particles. And polyvinylidene fluoride could produces fluoride in the thermal decomposition process, which is doped into carbon and LiFePO<sup>4</sup> to form fluoride-doped carbon layer and LiFePO3.938F0.062, respectively. In this constructed architecture, the F-doped carbon layer acts as conductive network for LFP, which can enhance the electronic conductivity of overall electrode. Furthermore, the crystal lattice of LFP was enlarged by the F doping, which facilitates the Li<sup>+</sup> intercalation/deintercalation. On the other hand, a strong electronic coupling between F-doped carbon and LiFePO3.938F0.062 can effectively suppress the shedding of carbon layer during cycling process, which keep stabilized of the reaction interface, and thus enhance the cycling stability. As a result, LFPF/CF composite shows superior rate performance (164.8, 159.2, 148.6, 135.8, and 102.3 mAh g−<sup>1</sup> at 0.1, 0.5, 1, 5, and 10 C), and excellent cycling stability (high capacity retention of 95.6% after 500 cycles at high rate of 5 C).

Keywords: lithium ion batteries, cathode materials, LiFePO3.938F0.062, fluorine-doped carbon, electrochemical performance

# INTRODUCTION

Lithium ion batteries (LIBs) have been considered as one of the promising power sources in practical applications, for examples, electric vehicles (EVs), hybrid electric vehicles (HEVs) (Armand and Tarascon, 2008; Van Noorden, 2014; Whittingham, 2014). The energy storage mechanism of LIBs is the extraction/insertion of Li<sup>+</sup> in the electrode materials during charge/discharge process, and all of the Li<sup>+</sup> from cathode materials (Dunn et al., 2011). Therefore, cathode materials are very important for lithium ion battery (Zheng et al., 2015; Pan et al., 2017). Among commercial cathode materials, lithium iron phosphate (LiFePO4) has aroused great interest because of its high theoretical capacity (170 mAh g−<sup>1</sup> ), high safety, stability, low cost, and environment friendly (Park et al., 2007; Huang and Goodenough, 2008; Liu et al., 2009; Wu et al., 2011; Li et al., 2015, 2018; Zheng et al., 2015). However, LiFePO<sup>4</sup> shows poor rate performance, resulting in lower mass, and charge transport kinetics, which caused by its low electronic conductivity and Li-ion diffusivity at room temperature (Wang et al., 2005; Sun et al., 2011; Yang et al., 2012; Kuss et al., 2017).

Many strategies have been proposed to improve the rate performance, such as cationic doping and surface coating. Cationic doping is a effective method to expand lithium ion diffusion channels of cathode materials, e.g., cations include Al3+, Mg2+, Co2+, Zn2+, Al3+, and anions include F−, S2−, Cl<sup>−</sup> (Chung and Chiang, 2003; Ni et al., 2005; Hui and Zhou, 2006; Liu et al., 2006; Sun et al., 2010; Chiang et al., 2012; Gao et al., 2017; Gupta et al., 2017). On the other hand, surface coating have been confirmed is another strategy to improve rate performance which due to surface coating can improve the electronic conductivity of materials (Belharouak et al., 2005; Shin et al., 2006; Liu et al., 2008; Lu et al., 2009; Wagemaker et al., 2009; Wu et al., 2013; Lepage et al., 2014; Wang B. et al., 2016; Wang L. L. et al., 2016; Wang et al., 2019). Zheng et al. synthesized carbon coated LiFePO<sup>4</sup> by solid-state reaction method, which shows higher capacity of 125 mAh g−<sup>1</sup> at 5 C (Huang et al., 2013). Zhang et al. prepared Li4SiO4-coated LiFePO<sup>4</sup> by solgel method and microwave heating, delivers reversible capacity of 100 mAh g−<sup>1</sup> at 5 C (Zhang et al., 2012). Therefore, these results indicate that the cationic doping or surface coating is effective approach to improve rate performance of LiFePO4. However, the electrochemical potential of LiFePO<sup>4</sup> cannot be completely realized only by cationic doping or surface coating. Thus, an upgrade strategy combined with cationic doping and carbon coating is used to enhance the rate performance of LiFePO4. Nevertheless, the common strategy combined with cationic doping and the carbon coating is very complicated, which must via a two or three steps including the preliminary synthesis of lithium iron phosphate and subsequent surface coating and cationic doping (Shu et al., 2014; Li et al., 2015). In addition, the carbon was only modified on the surface of LiFePO4, which shows weak interaction between LiFePO<sup>4</sup> and the carbon layers, resulting in poor rate performance and cycling stability. Therefore, it is necessary to develop simple and effective strategies to modified LiFePO<sup>4</sup> by the combined with surface coating and cationic doping.

In this study, fluorine-doped carbon coated LiFePO3.938F0.062 was synthesized by facile one step solid phase method. And choice carbon source and fluorine sources is very important for the electrochemical performance. The Tween40 surfactant contain hydrophobic group and hydrophilic group. Therefore, Tween40 surfactant without impurity elements could uniformly mix with materials in aqueous solution and adsorb on the particle surface to form a carbon layer during heat treatment, which can coat tightly on the surface of the cathode material. While, PVDF only contains two elements of carbon and fluorine, and the fluorine content is 59%, which can contribute to form fluorine doping successfully in the heat process. The incorporation of fluorine (F) into carbon materials could tailor their electron-donor properties, which could further enhanced electronic conductivity of carbon layer and increase bonding force between carbon layer and LiFePO4. Meanwhile, fluorine (F) doped into LiFePO<sup>4</sup> could enlarge lithium ion diffusion channel, and thus facilitates the Li<sup>+</sup> intercalation/deintercalation. We focus on study the effect of the F-doped carbon coating and cationic doping on electrochemical performance. In addition, the physical, structural, and electrochemical properties of the F-doped carbon coat LiFePO4−xF<sup>x</sup> composites were systematically investigated. As a result, benefiting from combined the advantages of surface coating and cationic doping, the obtained LFPF/CF exhibits high reversible capacity, excellent cycling stability, and rate performance.

### EXPERIMENTAL

### Preparation of Materials

F-doped carbon coat LiFePO3.938F0.062 composites (LFPF/CF) were prepared by one step solid phase method, using Tween40 and PVDF as carbon sources and fluorine sources, respectively. The stoichiometric amount of FePO4, Li2CO3, Tween40, and PVDF were mixed by ball milling for 2 h and using deionized water as the liquid medium (The mass ratio of Tween40 and PVDF equal to 2:0.60). The obtained mixture was dried at 80◦C for 12 h and disintegrated by crusher to obtain the precursor powders. As-prepared precursors were calcined at 750◦C for 10 h to obtain F-doped carbon coat LiFePO3.938F0.062 composites (LFPF/CF). The same procedure was also used to synthesize carbon modified LiFePO<sup>4</sup> in the absence of the polyvinylidene fluoride (PVDF) and pure LiFePO<sup>4</sup> in the absence of the Tween40 and polyvinylidene fluoride (PVDF). Pure LiFePO4, carbon coated LiFePO4, and fluorine-doped carbon coated LiFePO3.938F0.062 are denoted as LFP, LFP/C, and LFPF/CF, respectively.

### Characterizations

An X-ray diffraction (XRD: Rigaku, D/max 2500v/pc, Cu Kα radiation) was used to analysis the structure of all samples. In order to observe the microstructure of the powder, the scanning electron microscopy (SEM: Philips, FEI Quanta 200 FEG) and transmission electron microscopy (TEM: JEM-2010, JEOL) were carried out. Thermo gravimetric analysis (TGA) was determined by thermal analyzer (SDTO600) in air flow with temperature rising of 5◦C min−<sup>1</sup> . X-ray photoelectron spectroscopy (XPS) measurements were carried out by a Phi X-Tool XPS system to determine the valence state of an element on the surface of all samples. Raman measurement of the composite was performed using a laser Raman spectrometer (Jobin Yvon, T6400).

### Electrochemical Evaluation

The electrochemical performances of all samples were evaluated using CR2032 coin cells. The active materials, acetylene black, and polyvinylidene fluoride (PVDF) at a weight ratio of 8:1:1 were mixed for 2 h in an N-methyl-2-pyrrolidene (NMP) solvent, and finally obtain the cathode slurry. The slurry were coated on aluminum foil and dried in vacuum at 120◦C for 24 h. Then the CR2032-type half coin cells was assembled in a Ar-filled glovebox (Etelux LAB2000). The components of the CR2032 type half coin cells are as prepared samples as cathode, lithium metal as the counter electrode, Celgards 2200 separator and A 1:1 volumetric ratio mixture of ethylene carbonate (EC) and dimethyl carbonate (DMC) containing 1 M LiPF<sup>6</sup> as the electrolyte. The LAND CT2001A battery testing system (Wuhan, China) was used to test electrochemical performance at voltage range of 2.3–4.3 V under different rates. The cyclic voltammetry (CV) measurements were carried out on IM6 electrochemical workstation at a scan rate of 0.1 mV s−<sup>1</sup> . Electrochemical impedance spectroscopy (EIS) of the cell was measured by using IM6 electrochemical workstation, with the 5 mV amplitude of the AC signal at the frequency range between 100 kHz and 0.01 Hz.

### RESULTS AND DISCUSSION

**Scheme 1** illustrates the detailed formation process of F-doped carbon coated LiFePO3.938F0.062 composites (LFPF/CF). First, the FePO4, Li2CO3, Tween40, and PVDF were mixed through a ball mill method. In this process, Tween40 could bonded with the raw materials, and adsorbed their surface, which could form obstructer film. While, PVDF exists between FePO<sup>4</sup> and Li2CO3. Subsequently, the obtained precursor was sintered at 750◦C in Ar atmosphere, and LFPF/CF composite was obtained. Tween40 form a homogenous carbon layer on the surface of particles, and fluoride from PVDF decomposition at high temperature process is doped into carbon layer and bulk materials to form F-doped carbon coat LiFePO3.938F0.062 composite. Hence, the optimal x value, F content of F-doped carbon layer and the carbon content were chosen in the follow-up investigation (**Table S5** and **Figures S5, S6**).

**Figure 1A** shows the XRD patterns of pure LFP, LFP/C, and LFPF/CF composite. The characteristic peaks of all samples belong to single phase of orthorhombic olivine-type structure LiFePO<sup>4</sup> with the Pnma space group (**Figure 1D**) (Choi and Kumta, 2007). There are no other characteristic peaks in the LFP/C and LFPF/CF sample, suggesting that carbon content is extremely low, and have little effect on the structure. In order to analysis the XRD data of all samples (**Figure 1B** and **Figures S1A,B**), the Rietveld refinement method was implemented, and the crystallographic data were summarized in **Table S1**. A slight decrease of the lattice parameter a, b, and c indicated that F has been successfully doped into the crystal lattice. Concerning fluorine ion, some references suggest that it is substituted for the oxygen rather than the whole PO3<sup>+</sup> 4 group. Meanwhile, PO3<sup>+</sup> 4 group has very high formation energy, so it can't be replaced by fluorine (Liao et al., 2007; Pan et al., 2011; Milovic et al., 2013). The lattice parameters evolution inferred that the average bond lengths of P-O in the PO3<sup>+</sup> 4 group decrease after F doped, which due to the ionic size of F<sup>−</sup> (1.33 Å) is smaller that of O2<sup>−</sup> (1.40 Å) and fluorine are highly electronegative (Shannon and Prewitt, 1969). Therefore, the contractions of the PO3<sup>+</sup> 4 group enlarged the lithium ion diffusion tunnel after F doped, which lead to the lithium ions deintercalate/intercalate more easily in the tunnel (**Figure 2**).

Raman spectroscopy was used to study the exist of carbon and the effect of F-doped on the structure of carbon layer. **Figure 1C** exhibits the Raman spectra of pure LFP, LFP/C, and LFPF/CF. All samples shows a relatively small band at about 948 cm−<sup>1</sup> , which is assigned to the symmetric PO<sup>4</sup> stretching vibration of the LiFePO4. The bands below 400 cm−<sup>1</sup> are assigned to vibration of Fe-O (Wu et al., 2013). In addition, two bands at 1,350 and 1,605 cm−<sup>1</sup> for both LFP/C and LFPF/CF composite are related to the D and G bands of carbon, respectively. The D and G bands are ascribed to the sp<sup>2</sup> graphite-like structure and the sp3 type carbon, respectively (Tian et al., 2015). Furthermore, LFPF/CF composite shows higher ID/I<sup>G</sup> (0.975) value than that of LFP/C (0.911), which are assigned to increase the defects in carbon layer after fluorine doped. Therefore, the F-doping carbon layer provides sufficient electronic pathways, and consequently enhances the kinetic process. In addition, The carbon content of LFP/C and LFPF/CF composites are analyzed by TGA, and the carbon content in the samples is about 1.83 and 2.02%, respectively (**Figure S2**).

In order to analyze the chemical states and compositions of pure LFP, LFP/C, and LFPF/CF, the X-ray photoelectron spectroscopy (XPS) was carried out, and the results are shown in **Figure 3**. The high-resolution of Fe 2p for pure LFP splits two peaks at 709.2 and 723.2 eV along with two satellite peaks at 713.6 and 727.6 eV, corresponding to Fe 2p3/<sup>2</sup> and Fe 2p1/2, which proved that iron is oxidation state of +2 (Zhou et al., 2016). For LFP/C, there is no different in the position of the Fe 2p spectrum compare to pure LFP. While, the main peaks of Fe 2p for LFPF/CF composite shows a lower binding energy shifted compared to that of pure LFP, which due to the density of electron clouds increase around LiFePO4. As illustrated in **Figure 4**, Fluorine doping can make the carbon atom in the electron-rich state, and electron cloud tends to LFP, forming a strong electronic coupling between carbon layer and bulk material. The C1s high resolution XPS spectrum is de-convoluted into three peaks at 284.6, 285.4, and 287.9 eV, which belong to the C-C (sp<sup>3</sup> -C), C=O (sp<sup>2</sup> -C) and C-F bonds, respectively (Wu et al., 2018). Moreover, there are two peaks at 686.2 and 688.2 eV in the fitting of F1s high revolution XPS spectrum, which correspond to F-Li and F-C, respectively (Feng et al., 2017). Therefore, above results indicated that F was doped into the LFP lattice and carbon, respectively.

The effect of F-doped on the microstructure was studied by SEM. **Figure 5** exhibits the SEM images of pure LFP, LFP/C, and LFPF/CF. As seen in **Figure 5A**, pure LFP shows agglomerated particles with a larger diameter of about 600–800 nm. However, for LFP/C and LFPF/CF, the size of particles are smaller than that of pure LFP, suggesting that carbon and F-doped carbon coated could effectively inhibit the growth of particles. **Figures 5D1–D<sup>4</sup>** shows the EDX elemental mapping images of LFPF/CF. As seen **Figures 5D1–D4**, Fe, C, and F elements shows uniform distribution in the LFPF/CF composite, which indicated that Fdoped carbon layer is homogeneously coated on the surface of the LFP particles. In addition, the tap densities of pure LFP, LFP/C, and LFPF/CF are 1.25, 1.22, and 1.20 g/cm<sup>3</sup> (**Table S2**), suggesting that tap density not vary after surface modified.

Meanwhile, the microstructures of pure LFP, LFP/C, and LFPF/CF composite were further investigated by the TEM and

HRTEM, and the results are shown in **Figure 6**. **Figure 6A<sup>1</sup>** shows that the pure LFP particles with size range of 600– 800 nm, which is in consistent with the SEM results. In addition, the HRTEM image shows that pure LFP exhibits a smooth surface without carbon layer (**Figure 6A2**). **Figures 6B1,B<sup>2</sup>** shows the TEM and HRTEM images of LFP/C. The particles

FIGURE 2 | Refined crystal structure of each material of (A) pure LFP and (B) Fluorine-doped carbon coated olivine LiFePO3.938F0.062.

size of LFP/C are smaller than that of pure LFP, whereas thin carbon layer with the thickness of 2–5 nm on the surface of LFP particles were observed. Moreover, as shown in **Figure 6C1**, LFPF/CF shows smaller particles size than that of pure LFP, and there are F-doped carbon between LFP particles. HRTEM image of the LFP@CF (**Figure 6C2**) shows the F-doped carbon layer with 2–5 nm were coated on the surface of LFP particles. Meanwhile, it is observed that the d-spacing of lattice fringes for LFP and LFP/C are almost 0.272 nm, corresponding to the (102) plane of the LFP. While, LFPF/CF composite show the d-spacing of lattice fringes of 0.274 nm, which is higher than that of LFP and LFP/C. Obviously, the interlayer spacing was enlarged with fluorine doping into LiFePO<sup>4</sup> lattice.

**Figure 7A** shows the initial charge/discharge curves of pure LFP, LFP/C, and LFPF/CF at 0.1 C. There are two plateaus for charge/discharge curves of pure LFP, LFP/C, and LFPF/CF, which is ascribed to Li<sup>+</sup> intercalation/deintercalation from LiFePO4, accompanied by the oxidation/reduction of Fe2+/Fe3<sup>+</sup> (Wang et al., 2018). Pure LFP delivers a discharge capacity of 131.8 mAh g <sup>−</sup><sup>1</sup> with a coulombic efficiency of 85.8%. However, the LFP/C exhibits slightly higher coulombic efficiency of 91.2%, which due to the carbon coated can increase the electronic conductivity. However, for LFPF/CF, the discharge capacity reach to 164.8 mAh g <sup>−</sup><sup>1</sup> with a high coulombic efficiency of 97.5%, which is higher than that of pure LFP and LFP/C. Meanwhile, LFPF/CF exhibits a lower charge voltage plateau and a higher discharge voltage plateau compared with the pure LFP and LFP/C, suggesting that LFPF/CF has lower polarization. Therefore, these results confirmed that the LFPF/CF with better ion transport ability and high electronic conductivity compare to pure LFP and LFP/C.

**Figure 7B** and **Figure S3** shows the rate performance of pure LFP, LFP/C, and LFPF/CF. As seen **Figure S3**, the discharge capacity of LFPF/CF are 164.8, 159.2, 148.6, 135.8, and 102.3 mAh g−<sup>1</sup> at 0.1, 0.5, 1, 2, 5, and 10 C, respectively, which shows excellent rate performance. However, the pure LFP and LFP/C delivers lower discharge capacity of 131.8, 121.7, 103.4, 83.2, and 41.6 mA h g−<sup>1</sup> and 147.8, 139.5, 123.3, 106.8, and 72.8 mA h g −1 at 0.1, 0.5, 1, 5, and 10 C. In order to study the common effect of the F doped into LFP lattice and F-doped carbon coated the kinetic process, the cyclic voltammograms of pure LFP, LFP/C, and LFPF/CF composite at different scanning rate were performed out, and the results are provided in **Figure 7C** and **Figure S4**. The potential intervals (1V) are increased between the anodic peak and cathode peak of all samples with the increased scanning rate, which can be attributed to the kinetic process limitations (Huang et al., 2014). However, the LFPF/CF (**Figure 7C**) shows lower potential interval (1V) than that of pure LFP (**Figure S4B**) and LFP/C (**Figure S4A**) at different sweeping rates. Meanwhile, the redox peak current value of all samples also increases with the increase of scanning rate. The LFPF/CF exhibits higher redox peaks current than that of pure LFP and LFPF/CF composite. Above result indicated that LFP/CF has excellent kinetic process, which are attributed to the introduction of F into the lattice could enlarged interlayer spacing and F-doped carbon coated could form an electronic conductive network.

**Figures 7D,E** shows the cycling performance of pure LFP, LFP/C, and LFP/CF at 0.5 C and 5 C, respectively. Pure LFP can deliver a discharge capacity of 124.7 mAh g−<sup>1</sup> with capacity retention of 82.1% after 100 cycles at 0.5 C. Nevertheless, low capacity retention of 43.5% were achieved after 500 cycles at 5C. The poor cycling stability is attributed to the kinetic process limitations, which lead to Li<sup>+</sup> are irreversibly intercalation/deintercalation from LiFePO<sup>4</sup> during charge/discharge process. However, LFP/C shows a capacity retentions of 92.7 and 82.3% after 100 cycle at 0.5 C and 500 cycles at 5 C, respectively, which is higher than that of pure LFP. Therefore, these results suggested that carbon coated can effective increase electronic conductivity, resulting in increased kinetic process, as well as suppress the capacity degradation to some extent during cycling process. On the other hand, the LFPF/CF can deliver outstanding cycling stability compare to that of LFP/C. The LFPF/CF deliver high specific capacity of 158.7 mAh g−<sup>1</sup> after 100 cycles, and a high capacity retention of 98.6% is maintained. More importantly, a high capacity retention of 91.4% can be achieved even over 500 cycles at 5 C. Such excellent cycling stability can be ascribed to the introduction of F into the lattice and F-doped carbon coated enhance lithium ion diffusion and electron transport, so that more lithium ions can reversibly de/intercalate into LiFePO<sup>4</sup> during long cycling at high rate. Meanwhile, a strong electronic coupling form the binding force between F-doped carbon layer and bulk material, which could keep the reaction interface stability of LFPF/CF.

Frontiers in Materials | www.frontiersin.org

To better explore the reasons for the improved cycling stability of LFPF/CF compare to pure LFP and LFP/C, the ICP and TEM analysis after cycling were carried out, and the results as shown in **Table 1** and **Figure 8**. As seen **Table 1**, the lithium content for pure LFP in charge/discharge state are 28.53 and 72.32% after 300 cycles, suggesting lithium ion are irreversibly from the LiFePO<sup>4</sup> lattice. Above results indicated that the irreversible deintercalation/intercalation of lithium ion increase as the increased cycles. **Figures 8A1,A<sup>2</sup>** exhibits the TEM images of pure LFP after 300 and 500 cycles, and the surface and structure of pure LFP have not change, suggesting that the deterioration of cycling stability which be attributed to poor the kinetic process that cause lithium ions be irreversibly intercalated/deintercalated from LiFePO<sup>4</sup> lattice at high rate. The lithium percentage in charge/discharge state for the LFP/C after cycling are lower/higher than that of pure LFP, suggesting that carbon coated could enhance the kinetic process. However, after 500 cycles, the lithium ion percentage after charge/discharge of LFP/C show a certain degree of increase/decrease (12.3%/85.38%). Through TEM observation, it was found that the expansion and contraction of LiFePO<sup>4</sup> led to the shedding of carbon layer during the charge/discharge process at high rate (**Figures 8B1,B2**), which was caused by the different expansion coefficients of the two materials. The shedding of carbon layer directly decreases the electrical conductivity of LFP/C. By contrast, the lithium ion percentage of the LFPF/CF not vary much with the increased cycles. While we observe that the F-doped carbon layer still coated on surface of LiFePO<sup>4</sup> after 300 and 500 cycles (**Figures 8C1,C2**). The difference between LFP/C and LFPF/CF is attributed to a strong electronic coupling form the binding force between F-doped carbon layer and bulk material, so as to prevent the carbon layer shedding from LiFePO<sup>4</sup> surface, and stabilized the reaction interface of LFPF/CF composite.

**Figure 9A** exhibits the Nyquist plots of pure LFP, LFP/C, and LFPF/CF before and after cycling. As seen **Figure 9A**, the

FIGURE 8 | TEM images of all samples of pure LFP: (A1) after 300 cycles, (A2) after 500 cycles, LFP/C: (B1) after 300 cycles, (B2) after 500 cycles and LFPF/CF: (C1) after 300 cycles, (C2) after 500 cycles at 5 C rate, respectively.

TABLE 1 | The ICP analysis of pure LFP, LFP/C, and LFPF/CF composite after cycling, respectively (fully charged/discharge).


impedance spectrum is consist of an intercept at high frequency, a depressed semicircle at medium frequency and a straight line in the low frequency region, which relates to the ohmic resistance (R<sup>Ω</sup> ), the charge transfer resistance (Rct) and the Warburg impedance (W), respectively. These impedance spectra were analyzed by the equivalent circuit model (**Figure 9E**), and corresponding parameters are shown in the **Tables S3, S4**. Meanwhile, the Li<sup>+</sup> ion diffusion coefficiency (D<sup>+</sup> Li) is calculated by the following equations (Li et al., 2010; Wang L. L. et al., 2016; Wang et al., 2017):

$$\mathrm{D}\_{\mathrm{Li}}^{+} = \frac{\mathrm{R}^{2}\mathrm{T}^{2}}{2\mathrm{A}^{2}\mathrm{n}^{4}\mathrm{F}^{4}\mathrm{C}\_{\mathrm{Li}}^{2}\sigma^{2}} \tag{1}$$

$$Z' = \mathcal{R}\_{\Omega} + \mathcal{R}\_{\mathfrak{C}} \sigma \,\omega^{-1/2} \tag{2}$$

where T, R, A, F, n, C, ω is constant temperature, gas constant, surface area of cathode, Faraday constant, number of involved electrons, Li<sup>+</sup> ions concentration, and angular frequency in the low frequency region, respectively. Meanwhile, σ is the Warburg coefficient, it is associated with Z′ against ω −1/2 .

The LFPF/CF composite shows Rct value of 178.6 Omega, which is lower than that of pure LFP (558.8 ) and LFP/C (339.2 ) before cycling. In addition, the Li<sup>+</sup> diffusion coefficiency of LFPF/CF composite is 2.72 × 10−<sup>12</sup> cm<sup>2</sup> s −1 , which is higher that of pure LFP (0.87 × 10−<sup>12</sup> cm<sup>2</sup> s −1 ) and LFP/C (1.43 × 10−<sup>12</sup> cm<sup>2</sup> s −1 ). The results suggest that the introduction of F-doped carbon coated could greatly improve the charge and Li<sup>+</sup> ion diffusion transport coefficiency.

**Figures 9B–D** show the Nyquist plots of pure LFP, LFP/C and LFPF/CF at 5 C after different cycles, respectively. For pure LFP, when the number of cycles goes from 1st to 500th, the Rct values increase from 629.8 and 1372.9 , and the D<sup>+</sup> Li value decreased from 0.77 × 10−<sup>12</sup> cm<sup>2</sup> s −1 to 0.35 × 10−<sup>12</sup> cm<sup>2</sup> s −1 . The LFP/C show the increased Rct values and the decreased D<sup>+</sup> Li value after 300 and 500 cycles, which are lower than that of pure LFP. While, for LFPF/CF, the Rct values is 254.7 after 300 cycles, and increased to 314.8 after 500 cycles which are lower than those of pure LFP and LFPF/CF composite. Meanwhile, the D<sup>+</sup> Li value of LFPF/CF composite not vary much as the increased cycle numbers. This case clearly shows that the F-doped carbon coated can significantly keep the stability of resistance and Li<sup>+</sup> diffusion coefficient during cycling, which is attributed to a strong electronic coupling form the binding force between F-doped carbon layer and bulk material to stabilize the reaction interface.

### CONCLUSIONS

In summary, the F-doped carbon coated LiFePO3.938F0.062 composite (LFPF/CF) were synthesized by a simple solid-state reaction method. The surface of LiFePO<sup>4</sup> particles is uniformly coated by F-doped carbon layer, and a strong electronic coupling was formed between LFP and F-doped carbon layers, which resulting in enhanced electronic conductivity and reaction interfacial stability. In addition, minor F fluorine doped into the LiFePO<sup>4</sup> lattice, which can enlarge the lithium ion diffusion

channel. As a result, the LFPF/CF shows excellent rate capability (164.8, 159.2, 148.6, 135.8, and 102.3 mAh g−<sup>1</sup> at 0.1, 0.5, 1, 2, 5, and 10 C, respectively) and high cycling stability (high capacity retention reach 91.4% after 500 cycles at 5 C). Therefore, we believe that as-prepared F-doped carbon coated LFPF/CF composite could be one of an advanced cathode materials for Li-ion battery technology.

# DATA AVAILABILITY STATEMENT

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation, to any qualified researcher.

# AUTHOR CONTRIBUTIONS

ZY conducted the experiments. QL is the supervisor of this research work. ZY and FZ helped writing. ZY, DH, XF, FZ, QP, and ZM performed the characterization and data analysis. All authors involved the analysis of experimental data and manuscript preparation.

# FUNDING

This research was supported by National Natural Science Foundation of China (51864007), Guangxi Key Research and Development Program of Science and Technology (GUIKE AB17195065 and AB17129011), and Guangxi Technology Base and Talent Subject (GUIKE AD18126001).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmats. 2019.00341/full#supplementary-material

# REFERENCES


coated with ZrO<sup>2</sup> nanolayer, Electrochem. Commun. 10, 165–169. doi: 10.1016/j.elecom.2007.11.016


**Conflict of Interest:** 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 © 2020 Yan, Huang, Fan, Zheng, Pan, Ma, Wang, Huang 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 Efficient and Environment Friendly Plasma-Enhanced Synthesis of Al2O3-Coated LiNi1/3Co1/3Mn1/3O<sup>2</sup> With Excellent Electrochemical Performance

Xinzhi Wang1†, Qianqian Jiang1,2 \* † , Yichi Zhang<sup>1</sup> , Nannan Yuan<sup>1</sup> and Jianguo Tang<sup>1</sup> \*

*1 Institute of Hybrid Materials, National Center of International Research for Hybrid Materials Technology, National Base of International Science and Technology Cooperation, College of Materials Science and Engineering, Qingdao University, Qingdao, China, <sup>2</sup> School of Chemical and Biomedical Engineering, Nanyang Technological University, Singapore, Singapore*

## Edited by:

*Junchao Zheng, Central South University, China*

### Reviewed by:

*Jin Tao Zhang, Nanyang Technological University, Singapore Lingjun Li, Changsha University of Science and Technology, China*

### \*Correspondence:

*Qianqian Jiang kaiqian2008@163.com Jianguo Tang jtang951@163.com*

*†These authors have contributed equally to this work*

### Specialty section:

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

Received: *27 June 2019* Accepted: *22 January 2020* Published: *20 March 2020*

### Citation:

*Wang X, Jiang Q, Zhang Y, Yuan N and Tang J (2020) High Efficient and Environment Friendly Plasma-Enhanced Synthesis of Al*2*O*3*-Coated LiNi1*/*3Co1*/*3Mn1*/*3O<sup>2</sup> With Excellent Electrochemical Performance. Front. Chem. 8:72. doi: 10.3389/fchem.2020.00072* PLA-1-Al2O3@LNCM synthesized using an efficient and facile plasma-enhanced method exhibits markedly improved capacity retention of 98.6% after 100 cycles, which is much larger than that of LNCM at 80% after 100 cycles. What is more, it also exhibits significantly enhanced cyclicity compared to that of 1-Al2O3@LNCM cathodes prepared using the normal solid state method, which further illustrates the efficiency and superiority of this plasma-enhanced method. More importantly, the rate performance of PLA-1-Al2O3@LNCM is improved because of the better electrolyte storage of the assembled hierarchical architecture of the Al2O<sup>3</sup> coating layer according to unimpeded Li+ diffusion from electrode to electrolyte. When cycling at 55◦C, the PLA-1-Al2O3@LNCM shows 93.6% capacity retention after 100 cycles, which is greatly enhanced due to the uniform Al2O<sup>3</sup> layer. Further, growth of polarization impedance during cycling can be effectively suppressed by the Al2O<sup>3</sup> layer, which can further confirm the effect the Al2O<sup>3</sup> layer coated on the surface of the LNCM. The enhanced cycling performance and thermal stability illustrates that this facile surface modification, using the plasma-enhanced method, can form an effective structured coating layer, which indicates its prospects as an application in the modification of other electrode materials.

Keywords: plasma-enhanced method, Al2O<sup>3</sup> layer, LNCM, elevated temperature performance, Li-ion batteries

# INTRODUCTION

With the development of the global energy demand and the urgency of protecting the environment, there is a growing need to develop clean and more efficient power sources like lithium ion batteries (LIBs) (Tang et al., 2013; Chen et al., 2016), which have become the most feasible power source for portable electronic devices, mainly due to their high gravimetric and volumetric energy densities (Xu et al., 2017; Lv et al., 2019) Considering that the specific capacity of LIBs heavily relies on cathode materials, efforts have been made to develop cathode materials with a high energy density and large power capability (Lee et al., 2014; Zhang et al., 2018).

Until now, in cathode materials, such as layered LiCoO<sup>2</sup> (Zhao et al., 2015), spinel LiMn2O<sup>4</sup> (Jiang et al., 2016b), LiNiO2, and olivine LiFePO<sup>4</sup> have been widely applied for commercial LIBs in electrical vehicles (EVs) or hybrid electric vehicles (HEVs). Among them, LiCoO<sup>2</sup> is the earliest commercialized cathode material which cannot meet the increasing practical demand due to its low practical capacity, relatively high cost, toxicity of cobalt, and unreliable safety. Researchers have therefore focused on finding alternative cathode materials to satisfy the required environmental, safety, and cost standards for LIBs (Ohzuku and Makimura, 2001; Deng et al., 2008; Makkus et al., 2010). In comparison with LiCoO2, layered LiNixCoyMnzO<sup>2</sup> (x + y + z = 1) is considered an excellent candidate as a cathode for commercial LIBs. In 2001, LiNi1/3Co1/3Mn1/3O<sup>2</sup> (LNCM) was successfully implemented, and has the same layered α-NaFeO<sup>2</sup> structure with overall hexagonal symmetry (space group R-3 m) as LiCoO<sup>2</sup> (Xia et al., 2014) but with a much higher capacity, better thermal stability, is more environmental friendly and costs less (Lee et al., 2004; Luo et al., 2006; Cho et al., 2011). Nevertheless, LNCM still has some disadvantages, such as an inferior cycle performance at high working voltage, relatively poor conductivity, and irreversible side reactions with electrolytes and cation a disorder between Li<sup>+</sup> and Ni2+, which leads to some technical challenges related to cyclability and rate capability, seriously affecting its large-scale applications (Yao et al., 2013; Cui and Xu, 2015; Zhu et al., 2015). Researchers worldwide have proposed various solutions to overcome these disadvantages. For example, surface coating is an easy and effective way to improve the electrochemical performances of cathodes. Various coating materials such as carbon and metal oxides (Al2O3, Li et al., 2014; Li2TiO3, Lu et al., 2013; ZrO2, Hu et al., 2009; TiO2, Li et al., 2006; AlF3, Myung et al., 2010; AlPO4, Wang et al., 2013, etc.) have been studied, which can increase the electronic conductivity and inhibit the side reactions between cathode and electrolyte (Jiang et al., 2018). Among the metal oxides mentioned above, Al2O<sup>3</sup> is regarded as a desirable surface coating material due to its excellent chemical and thermal stability, which is good for the electrochemical performance cathode material. However, the normal method used for coating Al2O<sup>3</sup> is solid-state (Kim et al., 2006) or solution methods (Qiu et al., 2014), which often result in no uniform elemental distribution and inferior electrochemical performance (Yan et al., 2016). In addition, the experimental conditions (pH, temperature, and mixing speed) should be carefully controlled to synthesize homogeneous hydroxide or carbonate precipitates, making this method complicated (Manthiram et al., 2016). Therefore, it becomes challenging to develop a facile and general approach for the preparation of coating on the LNCM with favorable structures and excellent electrochemical performance.

Herein, we first report a novel facile plasma-enhanced lowtemperature approach in LiNi1/3Co1/3Mn1/3O<sup>2</sup> coated with Al2O<sup>3</sup> with a superior hierarchical structure, as illustrated in **Scheme 1**. This method produces a simple, uniform coating, which can delay the cathode/electrolyte interfacial reactions and facilitate Li-ion diffusion in the batteries. The cell made of LiNi1/3Co1/3Mn1/3O<sup>2</sup> coated with Al2O<sup>3</sup> delivers a high specific reversible capacity of 213.8 (0.2◦C), 208.4 (0.5◦C), 199.6 (1◦C), and 189.0 (2◦C) mAhŁEg−<sup>1</sup> , respectively. More importantly, all the LiNi1/3Co1/3Mn1/3O<sup>2</sup> coated with Al2O<sup>3</sup>

composite electrodes show improved cycling performance and rate capability, especially in an elevated temperature, compared to the pristine LiNi1/3Co1/3Mn1/3O<sup>2</sup> electrode, which further illustrates the effect of the Al2O<sup>3</sup> layer as a medium of preventing the material from degradation during cycling.

# EXPERIMENTAL

# Synthesis of PLA-1-Al2O3@LNCM and 1-Al2O3@LNCM Cathode Electrodes

The LiNi1/3Co1/3Mn1/3O<sup>2</sup> coated with Al2O<sup>3</sup> layer was prepared by an efficient plasma-enhanced method. In order to make the uniform mixture, the LiNi1/3Co1/3Mn1/3O<sup>2</sup> (denoted as LNCM) powder was supplied by one Tech. Company. Aluminum isopropoxide (C9H21AlO3, purchased from Aladdin official online mall) was chosen as the source of aluminum, and the Al content was set at molar ratios of Al:(Ni + Co + Mn) = 0, 0.5, 1.0, and 1.5% by controlling the amount of C9H21AlO3. First, the mixture was transferred into the plasma-enhanced tube furnace. The oxygen gas was introduced into the reactor for about 20 min to remove the air in the tube. The temperature of the tube furnace was set to 350◦C with the heating rate at 10◦ /min. Then the pressure was controlled under 50 Pa. During the process, the applied RF power was set to 200 W. The plasma treatment time was controlled for 20 min. The samples of LiNi1/3Co1/3Mn1/3O2, coated with Al2O<sup>3</sup> synthesized by the plasma-enhanced method using C9H21AlO<sup>3</sup> as the Al source, was defined as PLA-X Al2O3@ LNCM (X represents the Al content: 0.5, 1.0, and 1.5%). For comparison, the conventional annealing method was also utilized to prepare LiNi1/3Co1/3Mn1/3O<sup>2</sup> coated with a Al2O<sup>3</sup> layer using C9H21AlO<sup>3</sup> for the Al source. The mixed powders with an Al content of 1% were sintered at 500◦C for 6 h. Finally, the Al2O3 coated LiNi1/3Co1/3Mn1/3O<sup>2</sup> was obtained after naturall cooling the product to room temperature (denoted as 1-Al2O3@LNCM).

### Materials Characterizations

The crystalline phases and morphologies of the samples were characterized by an X-ray diffractometer (D/MAX-2200V, Rigaku, Japan) using Cu Kα radiation (k = 0.15406 nm) and scanning electron microscopy (JEOL JSM-7800F, Japan) equipped with an energy dispersive spectrometer (EDS) for elemental analysis. X-ray diffraction studies were carried out using Cu-Kα radiation. The morphology, crystal lattice, and the micro-region structure of the sample coating layer were characterized by the HR-TEM (HR-TEM, JEOL JEM-2100Plus, Japan).

# Cell Fabrication and Electrochemical Measurements

Electrochemical experiments performed in R2032 coin type cells were assembled in an argon-filled glove box, using lithium metal as the counter electrode. The electrolyte consisted of a solution of 1 M LiPF<sup>6</sup> in ethylene carbonate (EC), dimethyl carbonate (DMC), and ethyl methyl carbonate (EMC) solution (1:1:1 by volume). The composite electrodes were made from the active materials powder (80 wt.%), acetylene black (10 wt.%), and polyvinylidene fluoride (PVDF) binder (10 wt.%), homogeneously mixed in Nmethyl pyrrolidinone (NMP) solvent. The slurry was then coated onto the aluminum foil current collector and dried under a vacuum at 100◦C for 12 h. The mixed slurry was pasted on Al foil and subsequently dried at 120◦C for 12 h under a vacuum and then roll-pressed. All cells were assembled in an argon filled gloved box. The galvanostatic charge and discharge cycle tests were carried out at 25 and 55◦C between 2.8 and 4.3 V using a battery tester at the current density of 0.2◦C (1◦C = 1,675 mAhŁEg−<sup>1</sup> ) (LAND CT2001A, China). The electrochemical impedance spectra (EIS) were conducted on UT85794 (made in the Netherland) in the range from 100 kHz to 10 mHz.

# RESULTS AND DISCUSSION

The X-ray diffraction patterns of the bare LNCM and PLA-1- Al2O3@LNCM are shown in **Figure 1**. All the peaks of the two samples in the XRD patterns can be indexed to the NaFeO<sup>2</sup> structure using space group R-3m without impurity phases, and matched well with those reported in the literature (Li et al., 2018). Furthermore, distinct splitting (006)/(102) and (108)/(110) peaks were observed in all the XRD patterns, indicating that a wellformed layered structure was obtained, illustrating that the thin Al2O<sup>3</sup> surface coating did not change the structure of the LNCM. No peak of Al2O<sup>3</sup> was observed due to the low quantity, which reveals that only Al2O<sup>3</sup> exists and no other impurity presents during the plasma-enhanced coating process. To confirm the differences between PLA-1-Al2O3@LNCM and LNCM, the lattice parameters calculated by Rietveld refinement programs are shown in **Table S1**. From this data, it can be seen that a slight difference is noted for PLA-1-Al2O3@LNCM, which has a slight increase in the a and c axes. In addition, there is no shift in the peaks, indicating that Al-doping does not take place during the plasma-enhanced process, further illustrating the efficiency of the plasma-enhanced method for the preparation of PLA-1-Al2O3@LNCM electrode materials.

The morphologies of LNCM and Al2O<sup>3</sup> coated LNCM are shown in **Figure 2**. The two samples fall under a similar size range of 0.5–0.8µm. It can however, clearly be seen that the surface of PLA-1-Al2O3@LNCM becomes rough and loses its smooth contours, due to the Al2O<sup>3</sup> grown on the surface of LNCM with layermorphology. When analyzed using EDS during SEM analysis, as-prepared Al2O<sup>3</sup> coated electrodes, clearly revealed that the molar ratios of Ni, Co, and Mn of the two samples are both close to 1:1:1, which agrees with the ideal layer structure LNCM. In addition, it exhibits an increase in Al and O content compared to that of pristine LNCM as shown in **Figure S1**, which can further confirm that the Al2O<sup>3</sup> has successfully grown on the surface of LNCM. At the same time, the SEM images of the LNCM with different Al2O<sup>3</sup> consent, have been invested as shown in **Figure S2**. We can see that when the consent of Al2O<sup>3</sup> is 1 wt.%, it shows a uniform morphology, which will be good for the electrochemical performance of the material. When the consent of Al2O<sup>3</sup> is less or more, it may not grow uniformly or may aggregate on the surface of LNCM, which will hinder the speed of the lithium ion insert/disinsert during cycles. SEM images of the PLA-1-Al2O3@LNCM and LNCM,

after cycling at 0.2◦C, are shown in **Figure S3**. It can be clearly seen that the SEM image of PLA-1-Al2O3@LNCM presents no obvious changes in the morphology, which further illustrates the stability of sample PLA-1-Al2O3@LNCM due to the uniform Al2O<sup>3</sup> layer.

In order to check the uniformity of the Al2O<sup>3</sup> coating layer on the surface of LNCM, EDS mapping was performed, as shown in **Figure 3**. The bright dots assigned to the elements O and Al are similar to the SEM of the PLA-1-Al2O3@LNCM sample, which indicates that Al2O<sup>3</sup> has uniformly dispersed on the surface of LNCM. This result is consistent with that of XRD patterns, SEM, and EDAX, which further confirms that the Al2O<sup>3</sup> layer has successfully grown on the surface of LNCM.

In order to further analyze the thickness and microstructure of Al2O<sup>3</sup> on the surface coating layer, TEM of LNCM and PLA-1-Al2O3@LNCM materials at different magnification are shown in **Figure 4**. Because the coating carbon layer is very thin, it is difficult to find any obvious differences in **Figures 4a,b**. However, the HR-TEM image in **Figure 4c** clearly shows that the host LNCM with interplanar distances of the parallel lattice value of 0.47 nm, indexed to the distance of the closely packed (003) plane of the R3m-layered structure (Guo et al., 2015), which illustrates that the Al atom has not changed the structure of LNCM and only grows on the surface of LNCM. In addition, it can be seen that the Al2O<sup>3</sup> coating layer with an amorphous phase is about 5 nm, and is clearly distinguishable from the crystalline LNCM, which is consistent with XRD and SEM analysis. TEM analysis demonstrates that the coating method is an effective way to coat the Al2O<sup>3</sup> layer on the surface of LNCM, which can reduce the contact area between the electrode and electrolyte, and further protects the cathode material from dissolving into the electrolyte.

The initial charge-discharge capacity of PLA-1- Al2O3@LNCM at the rate of 0.2◦C in the voltage of 2.8–4.3 V

at room temperature are compared with LNCM as shown in **Figure 5A**. It can be seen that the initial discharge capacity of PLA-1-Al2O3@LNCM is much higher than that of bare LNCM, probably due to the uniform Al2O<sup>3</sup> layer, which can improve the electrical conductivity of the material and even prohibit the Jahn-Teller distortion occurring at the surface under non equilibrium conditions (Jiang et al., 2016a). At the same time, LNCM coated with different Al2O<sup>3</sup> consent are also characterized in Li-ion batteries (as shown in **Figure S4A**). Among the three samples, PLA-1-Al2O3@LNCM shows the highest initial discharge performance, due to the suitable thickness of Al2O<sup>3</sup> formed only with the consent Al of 1%. In order to further make sure the efficient of this plasma-enhanced method, the initial discharge performance of Al2O3@LNCM synthesized form the normal solid state method are invested as shown in **Figure S5A**. The initial discharge of Al2O3@LNCM is 196.9 mAh·g −1 , which is lower than that of PLA-1-Al2O3@LNCM. This result illustrates this method not only can improve the efficient of this coated experiment, but also can improve the electrochemical performance of the material, indicating that the plasma-enhanced method can synthesized the coated LNCM with suitable Al2O<sup>3</sup> layer with lower temperature and less reaction time.

**Figure 5B** summarizes cycling performances of LNCM and coated LNCM cathodes cycled at 0.2◦C in a potential range of 2.8–4.3 V at room temperature. Capacity retention of PLA-1-Al2O3@LNCM upon cycling between 4.3 and 2.8 V is 98.46%, which is much higher than that of LNCM (74.66%). In other words, the discharge capacity of LNCM only maintained 74.66% of its initial capacity; while the discharge capacities of PLA-1-Al2O3@LNCM only decrease slightly from 213.8 to 210.5 mAhŁEg−<sup>1</sup> after 100 cycles, maintaining 98.46% of its initial capacity, which may be associated with cobalt dissolution and structural deterioration at the high operating voltage, resulting in increased interfacial resistance. At the same time, LNCM coated with different Al2O<sup>3</sup> consent are also characterized in Li-ion batteries as shown in **Figure S4B**. The capacity of The PLA-0.5-Al2O3@LNCM and PLA-1.5-Al2O3@LNCM samples suffers 11.51 and 10.54% capacity loss, respectively, while PLA-1-Al2O3@LNCM has the highest capacity retention of 98.46% after 100 cycles at 0.2◦C between 2.8 and 4.3 V. It can be seen that when the mass concentration of Al2O<sup>3</sup> is less or more than 1 wt.%, the capacity loss also undergoes a certain fading, which further illustrates that the suitable consent of Al content is 1 wt.%. For the PLA-1-Al2O3@LNCM sample with a proper coating amount, the hierarchical nanosheets of metal oxide can sequester HF in electrolytes containing LiPF6. The rich pores improve the ionic conductivity of the coating layer and the suitable thin film of Al2O<sup>3</sup> can strongly inhibit the electronic conductivity (Jung and Han, 2013). Additionally, to confirm the efficiency of this plasma enhanced method, the cycling performance of Al2O3@LNCM, synthesized form the normal solid state method,

is also invested, as shown in **Figure S5B**. Compared to PLA-1- Al2O3@LNCM, 1-Al2O3@LNCM, reaches a discharge capacity of only 181.8 mAhŁEg−<sup>1</sup> after 100 cycles, which only represents 92.38% of its initial discharge capacity (196.9 mAhŁEg−<sup>1</sup> ). However, PLA-1-Al2O3@LNCM shows much higher capacity retention (98.46%) after 100 cycles, which further illustrates that the plasma-enhanced method has higher efficiency than that of the normal solid state method. These results clearly indicate that the materials obtained by the plasma-enhanced strategy can be used as a promising cathode material in lithium-ion batteries.

To further investigate the effect of the Al2O3-coating on the electrochemical performance, the AC impedance curves of all the cycled samples were performed as shown in **Figure 6A**. The EIS plot consists of a depressed semi-circle in the highmedium frequency region that is attributed to the lithium-ion migration, through the SEI film and charge transfer reaction, and a straight line in the low frequency region attributed to the lithium-ion diffusion in the bulk electrode (Wang et al., 2014). A more detailed analysis of the impedance spectra was studied on the basis of the proposed equivalent circuit, where Rs represents the solution resistance and Rct represents charge transfer resistance during the electrochemical reactions. The inclined line in the low-frequency region represents the Warburg impedance (Zw) associated with Li-ion diffusion in the electrode active material (Liu et al., 2004). EIS plots in **Figure 6A** were fitted using the equivalent circuit model. During cycling, the formation of the solid-state interface layer may lead to raised impedance. From the fitted impedance parameters, it can be seen that the charge transfer resistance (Rct) of PLA-1-Al2O3@LNCM is much smaller (Rct ≈ 152.78 Ł) than that of the pristine LNCM (Rct ≈ 343.97 Ł), which indicates that the Al2O<sup>3</sup> layer can enhance Li+ transportation with enlarged interlayer. Moreover, it may be due to the removal of the unidentified phase in the precycled electrode or the influence of a native passivation layer on the electrodes (Lin et al., 2014). Therefore, the decrease of impedance illustrates the effect of the Al2O<sup>3</sup> layer on restraining the process during cycling, which further verifies the fact that the Al2O<sup>3</sup> layer is responsible for the good performance of PLA-1-Al2O3@LNCM. In addition, the impedance reduction exhibits an enhancement in the kinetics of the lithium-ion diffusion through the surface layer, and an obvious increase in rate capability as shown in **Figure 6B**. During rate capability testing, the effect of the coating layer on improving capacity retention is relatively apparent. Rate measurement is performed as the currents gradually increase to 2◦C, finally returning back to 0.2◦C, as shown in **Figure 6B**. The pristine LNCM delivers a capacity of 168.2, 142.9, 111.8, and 77.8 mAhŁEg−<sup>1</sup> at 0.2, 0.5, 1, and 2◦C, respectively. For PLA-1-Al2O3@LNCM, the corresponding capacity is 213.8, 208.4, 199.6, and 189.0 mAhŁEg−<sup>1</sup> , respectively, which shows an obvious improved rate performance than that of the pristine LNCM. Moreover, when the current returns back to 0.2◦C, the PLA-1-Al2O3@LNCM electrodes recover 99.58% of their original 0.2◦C discharge capacity, compared to 89.66% for the uncoated electrode due to the attack from the electrolyte, further concluding that coating LNCM with Al2O<sup>3</sup> can significantly improve its rate performance. The superior retention capability is due to the coating which improves the homogeneity of near-surface current density on the coated electrodes and uniform Al2O<sup>3</sup> layer for the improvement transfer of lithium ions between the electrode and electrolyte. Therefore, the excellent rate performance can be attributed to the uniform coating, which can cut off contact with the electrolyte and suppress the dissolution of active substances.

The cyclic performances of bare LNCM and PLA-1- Al2O3@LNCM at 0.2◦C in a potential range of 2.8–4.3 V at a high temperature (55◦C) are shown in **Figure 7**. The initial discharge capacity of bare LNCM is 165.2 mAhŁEg−<sup>1</sup> , whereas that of the PLA-1-Al2O3@LNCM is 209.1 mAhŁEg−<sup>1</sup> . There is a lot of difference between the LNCM and PLA-1- Al2O3@LNCM sample, indicating that the 5 nm Al2O<sup>3</sup> coating layer affects the Li+ diffusion between the LNCM particles and the electrolyte during the first cycle. More importantly, there are significant differences in cyclic performance between LNCM and PLA-1-Al2O3@LNCM as shown with the increasing cycling numbers. After 100 cycles, the discharge capacities of bare LNCM is only maintained at 86.6 mAhŁEg−<sup>1</sup> , suggesting that the material suffers stronger destruction as compared to the side reaction aggravated under a higher temperature. PLA-1-Al2O3@LNCM shows better capacity retention with 85.08% of the initial discharge, presumably to an inhibiting effect of

the Al2O<sup>3</sup> coating layer on the metal ion dissolution from the LNCM electrode to the electrolyte. In order to further confirm the effect of the plasma-enhanced method on the material, the cyclic performances of 1-Al2O3@LNCM are invested, as shown in **Figure S6**. The initial discharge capacity of PLA-1- Al2O3@LNCM is much higher than that of 1-Al2O3@LNCM, which indicates that the different synthesis methods have many differences on the first discharge cycle. Moreover, PLA-1- Al2O3@LNCM exhibited much better cyclic performance than that of 1-Al2O3@LNCM, which further confirms the superiority of the plasma-enhanced method in synthesizing the LNCM with the uniform Al2O<sup>3</sup> layer. The proper coating using a special synthesis method can also suppress the formation of the passive film of the solid state electrolyte acting as a highly effective lithium ion conductor, which consequently improves the rate capability even at an elevated temperature. In addition, we can clearly see that PLA-1-Al2O3@LNCM exhibits excellent electrochemical performances compared to many relevant references as shown in **Table S2**.

### CONCLUSION

A novel and effective Al2O<sup>3</sup> coating has been successfully achieved via a facile plasma-enhanced route method, investigating the effects of the Al2O<sup>3</sup> coating on the high temperature cyclic performance of LNCM. From the physical test, it can clearly be seen that the Al2O<sup>3</sup> uniformly grows on the surfaces of LNCM particles with a thickness of about 5 nm, causing no changes on the structure of LNCM. Futhermore,

### REFERENCES

Chen, Z., Wang, J., Chao, D., Tom, B., Bai, L., Chen, S., et al. (2016). Hierarchical porous LiNi1/3Co1/3Mn1/3O<sup>2</sup> nano-/micro spherical cathode material: minimized cation mixing and improved Li+mobility for enhanced electrochemical performance. Sci. Rep. 6:25771. doi: 10.1038/srep25771

PLA-1-Al2O3@LNCM exhibited a better cyclic performance and its discharge capacity was 213.8 mAhŁEg−<sup>1</sup> at 0.2◦C, and its capacity retention was 98.46% after 100 cycles, which is much higher than that of bare LNCM and 1-Al2O3@LNCM synthesized using the normal solid state method. What's more, PLA-1-Al2O3@LNCM shows better rate capability and cycling stability, which can be attributed to the uniform coating of Al2O3. When cycling at 55◦C, PLA-1-Al2O3@LNCM shows 85.08% capacity retention after 100 cycles, which is much higher than that of bare LNCM only, with a 52.42% capacity retention. The enhanced electrochemical performance further illustrates the efficiency and superiority of the Al2O<sup>3</sup> layer, which is achieved using the plasma-enhanced method, and which can protect cathode material thus alleviating the severe dissolution of metal ions from dissolving in the electrolyte and reduces decomposition of the electrolyte at the cathode. This facile and efficient surface modification with low-cost starting material and an effective coating layer structure can be adopted to enhance the cycling performance and thermal stability for other electrode materials, which can be used to synthesize other candidate cathode materials for lithium ion batteries.

### DATA AVAILABILITY STATEMENT

The datasets generated for this study are available on request to the corresponding author.

### AUTHOR CONTRIBUTIONS

QJ and XW wrote and modified the manuscript. YZ tested the performances of materials. NY synthesized the materials. JT modified the manuscript.

# ACKNOWLEDGMENTS

The work was supported by the National Natural Science Foundation of China (Grant Nos. 51473082, 51273096, and 51603109), the programme of Introducing Talents to the Universities (111 plan), the National One-Thousand Foreign Expert Program (Grant No. WQ20123400111), the Natural Science Foundation of Shandong province (Grant Nos. ZR2017BEM047, ZR2019BB002), and the postdoctoral science foundation of China (Grant No. 2017M610408).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2020.00072/full#supplementary-material


cathodes for molten carbonate fuel cells (MCFC). Cheminform 26:10012. doi: 10.1002/chin.199510012


**Conflict of Interest:** 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 © 2020 Wang, Jiang, Zhang, Yuan 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.