# ADVANCED CARBON CHEMISTRY FOR RECHARGEABLE BATTERIES

EDITED BY : Hongshuai Hou, Weijie Li, Tang Wei and Weihua Chen PUBLISHED IN : Frontiers in Chemistry

#### 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 978-2-88966-127-5 DOI 10.3389/978-2-88966-127-5

#### 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

# ADVANCED CARBON CHEMISTRY FOR RECHARGEABLE BATTERIES

Topic Editors: Hongshuai Hou, Central South University, China Weijie Li, University of Wollongong, Australia Tang Wei, Xi'an Jiaotong University, China Weihua Chen, Zhengzhou University, China

Citation: Hou, H., Li, W., Wei, T., Chen, W., eds. (2020). Advanced Carbon Chemistry for Rechargeable Batteries. Lausanne: Frontiers Media SA. doi: 10.3389/978-2-88966-127-5

# Table of Contents


Kaiqi Xu, Qicang Pan, Fenghua Zheng, Guobin Zhong, Chao Wang, Shijia Wu and Chenghao Yang


Yanyou Yin, Yu Zhang, Nannan Liu, Bing Sun and Naiqing Zhang

*84 Nitrogen-Doped Graphene via* In-situ *Alternating Voltage Electrochemical Exfoliation for Supercapacitor Application* Mingjun Jing, Tianjing Wu, Yazheng Zhou, Xilong Li and Yong Liu

# Editorial: Advanced Carbon Chemistry for Rechargeable Batteries

#### Hongshuai Hou<sup>1</sup> \*, Weijie Li <sup>2</sup> , Weihua Chen<sup>3</sup> and Wei Tang<sup>4</sup>

<sup>1</sup> Central South University, Changsha, China, <sup>2</sup> University of Wollongong, Wollongong, NSW, Australia, <sup>3</sup> Zhengzhou University, Zhengzhou, China, <sup>4</sup> Xi'an Jiaotong University, Xi'an, China

Keywords: carbon materials, batteries, electrochemistry, energy storage, energy chemistry

#### **Editorial on the Research Topic**

#### **Advanced Carbon Chemistry for Rechargeable Batteries**

The growing demand for green, alternative, and sustainable energy technology greatly expedites the development of energy storage/conversion devices like rechargeable batteries and supercapacitors. Carbon-based materials, which serve as a prevalent candidate in rechargeable batteries, have been widely explored due to their abundance, non-toxicity, stability, and durability. Furthermore, in other types of anode materials (including alloys, metal oxides, and metal sulfide), the introduction of carbon heterophase nanocomposite has been considered a prevalent strategy in enhancing the cycling performance due to an integrated structure stability during the charging/discharging process.

Amorphous carbon literally represents carbon with a low graphitization degree. The long-term disordered and short-term ordered structure is a typical feature for amorphous carbon in terms of micro texture. Specifically, amorphous carbon comprised of sp2 hybridized carbon which is rich in suspending bonds and other structural defects. Compared with graphite materials, amorphous carbon anodes exhibit a larger specific surface area, an abundance of porous structures and more active sites, which could bring about improved electrochemical properties.

When applied in rechargeable batteries, the microstructure and morphology, the graphitization degree, the specific surface area, the surface functionality, and porosity of carbon materials work synergistically to affect electrochemical performances. The unmodified bulk carbon used in anode material suffers from poor electrochemical performances due to the limited electron transport, sluggish ion migration, and undeveloped active sites. Consequently, in order to enhance the reversible capacity and electrochemical performance of carbon-based anode materials, the rational design and modification for advanced carbon-based materials has been proposed.

Heteroatom doping (B, N, P, S) has been regarded as a promising method to enhance the electrochemical properties of the carbon electrode. Heteroatom doping of large radius atoms (P, S) increases the interlayer spacing and induces structural distortion, providing more active sites for Li+, Na+, and K<sup>+</sup> storage. Heteroatom doping for small diameter atoms (B, N) may adjust the electron cloud density to facilitate the charge transfer and enhance the electrode-electrolyte interactions, and prove the wettability of the surface. In this topic collection, P/N-co-doped carbon nanosheets, N-doped graphene, and S-doped hard carbon@rGO composites for rechargeable batteries are reported, the chemical mechanism and process are discussed systematically. In addition, defects, pore structure, and micromorphology

#### Edited and reviewed by:

Nosang Vincent Myung, University of California, Riverside, United States

#### \*Correspondence:

Hongshuai Hou hs-hou@csu.edu.cn

#### Specialty section:

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

Received: 01 June 2020 Accepted: 29 June 2020 Published: 08 September 2020

#### Citation:

Hou H, Li W, Chen W and Tang W (2020) Editorial: Advanced Carbon Chemistry for Rechargeable Batteries. Front. Chem. 8:667. doi: 10.3389/fchem.2020.00667

**4**

of carbon materials on the electrochemical performances are discussed as well.

Carbon materials are often utilized as "additives" to improve the electrochemical performances of other electrode materials, such as metal oxides/sulfides/phosphides. The introduction of carbon materials can not only enhance the conductivity but also accommodate the volume change of electrode materials, which leads to the improved cycle stability and rate capability. In this topic collection, silicon, FeOOH, CuCo2S4, and Cu3P modified by functional carbon materials are reported.

We hope it will be helpful for readers to further understand the advanced carbon chemistry for rechargeable batteries.

## AUTHOR CONTRIBUTIONS

HH, WL, WC, and WT co-edit this Research Topic. All authors contributed to the article and approved the submitted version.

**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 Hou, Li, Chen 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.

# Hierarchical Nitrogen-Doped Porous Carbon Microspheres as Anode for High Performance Sodium Ion Batteries

Kaiqi Xu<sup>1</sup> , Qicang Pan<sup>2</sup> , Fenghua Zheng<sup>2</sup> , Guobin Zhong<sup>1</sup> , Chao Wang<sup>1</sup> , Shijia Wu<sup>1</sup> and Chenghao Yang<sup>2</sup> \*

*<sup>1</sup> Electric Power Research Institute of Guangdong Power Grid Co., Ltd., Guangzhou, China, <sup>2</sup> Guangzhou Key Laboratory for Surface Chemistry of Energy Materials, New Energy Research Institute, School of Environment and Energy, South China University of Technology, Guangzhou, China*

Sodium ion batteries (SIBs) have been considered as a promising alternative to lithium ion batteries (LIBs) for large scale energy storage in the future. However, the commercial graphite anode is not suitable for SIBs because of its low Na<sup>+</sup> ions storage capability and poor cycling stability. Recently, another alternative as anode for SIBs, amorphous carbon materials, have attracted tremendous attention because of their abundant resource, nontoxicity, and most importantly, stability. Here, N-doped hierarchical porous carbon microspheres (NHPCS) derived from Ni-MOF have been prepared and used as anode for SIBs. Benefiting from the open porous structure and expanded interlayer distance, the diffusion of Na<sup>+</sup> is greatly facilitated and the Na<sup>+</sup> storage capacity is significantly enhanced concurrently. The NHPCS exhibit high reversible capacity (291 mA h g−<sup>1</sup> at current of 200 mA g−<sup>1</sup> ), excellent rate performance (256 mA h g−<sup>1</sup> at high current of 1,000 mA g−<sup>1</sup> ), and outstanding cycling stability (204 mA h g−<sup>1</sup> after 200 cycles).

Edited by: *Hongshuai Hou, Central South University, China*

#### Reviewed by:

*Zhian Zhang, Central South University, China Yunhua Xu, Tianjin University, China*

#### \*Correspondence:

*Chenghao Yang esyangc@scut.edu.cn*

#### Specialty section:

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

Received: *06 September 2019* Accepted: *14 October 2019* Published: *31 October 2019*

#### Citation:

*Xu K, Pan Q, Zheng F, Zhong G, Wang C, Wu S and Yang C (2019) Hierarchical Nitrogen-Doped Porous Carbon Microspheres as Anode for High Performance Sodium Ion Batteries. Front. Chem. 7:733. doi: 10.3389/fchem.2019.00733* Keywords: Ni-MOF, hierarchical porosity, carbon microspheres, anode material, sodium ion batteries

## INTRODUCTION

Lithium ion batteries (LIBs) have been widely used as the power sources for portable electric devices, electric vehicles (EV), and hybrid electric vehicles (HEVs), because of their high energy density, high operate voltage and long services life (Zheng et al., 2015; Huang et al., 2016; Wang et al., 2017; Pan et al., 2018a). While, the applications LIBs in large-scale electrochemical energy system (EES) is restricted by the limited lithium resources in the earth's crust (Liu Y. Z. et al., 2018). Alternatively, sodium ion batteries (SIBs) working with the similar intercalation chemistry have attracted much attention, for the abundance of Na on the earth (Larcher and Tarascon, 2015; Li W. et al., 2015; Li et al., 2018). Nevertheless, the commercial graphite anode is not suitable for the fabrication of high performance SIBs, because of the poor diffusion kinetics of the much larger Na<sup>+</sup> ions compared to Li<sup>+</sup> ions (Luo et al., 2016; Hwang et al., 2017). Therefore, it is urgent to find an appropriate anode material for SIBs.

Recently, metallic alloy, metal sulfide and metal oxide have been explored as anode materials for SIBs (Liu et al., 2015; Ning et al., 2015; Pan et al., 2018b). Although exhibiting high capacity, these materials are subject to large volume change during the sodiation/desodiation process, and as a result, they are reported to suffer fast capacity loss and poor rate capability (Yabuuchi et al., 2014).

**6**

Carbonaceous materials, for example, expanded graphite, graphene, carbon nanotubes, and other amorphous carbon, were also studied as alternatives (Wen et al., 2014; Luo et al., 2015; Xu et al., 2015; Li et al., 2016a). These materials have attracted many attentions because of their abundant resource, stability, and low cost (Hou et al., 2017a).Among them, amorphous carbon, e.g., hard carbon and soft carbon, show high Na<sup>+</sup> accommodation capacity (Li et al., 2016b; Jian et al., 2017), but low initial coulombic efficiency, low capacity, and fast capacity degradation (Qian et al., 2018).

Heteroatom-doping with B, N, S, or P is an effective strategy to enhance the electrical conductivity and increase the specific surface area carbonaceous materials, hence enhance electron transport and provide more Na<sup>+</sup> storage active sites (Li D. et al., 2015; Yang et al., 2015). P-doped carbon nanosheets, S-doped hard carbon, N-doped carbon, and S, N-codoped carbon nanosheets have been demonstrated to effectively improve the electrochemical performance (Wang et al., 2016; Hou et al., 2017b; Yang et al., 2017; Hong et al., 2018). Hollow porous structures have promise to improve the cycling and rate performance of carbonaceous materials for SIBs, as this unique porous structure (1) can offer more Na<sup>+</sup> storage active sites, and therefore has the potential to give rise to higher capacity; (2) offer faster Na<sup>+</sup> transportation channels, which contributes to excellent rate capability; and (3) buffer the volume expansion during the repeated sodiation/desodiation process, leading to improved cycling performance (Li Z. et al., 2015; Zhang et al., 2015).

Herein, we developed a synergetic strategy to combine the above-mentioned merits of the heteroatom-doping and porous structures into one material by preparing Ni-MOF derived N-doped hierarchical porous carbon microspheres (NHPCS). This hierarchical porous structure are shown to offer more active sites for Na<sup>+</sup> storage, provide fast Na<sup>+</sup> diffusion channels, and alleviate the volume expansion during the sodiation/desodiation process. Meanwhile, the incorporation of N expanded interlayer distance and increased the surface area of the carbon microspheres, which enhances the cycling and rate performance. Benefiting from this unique design, NHPCS combines the advantages of both heteroatom-doping and porous structures, and are shown to exhibit high reversible capacity, outstanding rate performance and good cycling stability.

## EXPERIMENTAL SECTION

0.864 g nickel nitrate hexahydrate, 0.3 g 1, 3, 5-benzenetricarboxylic acid (H3BTC), and 3.0 g polyvinylpyrrolidone (PVP) were dissolved in a mixed solution (20 ml DMF, 20 ml absolute ethanol, and 20 ml deionized water, stirring for 30 min). The as-obtained mixture was transferred into 100 mL Teflonlined autoclave and heated at 150◦C for 10 h. Afterwards, the formed participate was centrifuged with ethanol three times and dried at 80◦C overnight. Subsequently, the obtained Ni-MOF was annealed with melamine (with a weight ratio of 1:1) at 800◦C for 1 h under Ar atmosphere. Finally, the obtained product was wash with 2 M HCl to remove Ni nanoparticles and the N-doped hierarchical porous carbon microspheres (NHPCS) were obtained. Hierarchical porous carbon microspheres (HPCS) were synthesized with the same procedure except that annealing process was conducted without melamine.

## MATERIAL CHARACTERIZATION

The crystalline phases of samples were characterized by XRD (Rigaku D/max 2500). The morphology of NHPCS/HPCS were observed by SEM (Hitachi, SU8010) and TEM (JEOL JEM-2100F). The XPS tests were carried out to determine the chemical composition and chemical state of elements of the sample.

To determine the specific area, the nitrogen absorption/desorption isotherm measurements were performed using ASAP2020 Surface Area and Porosity Analyzer. Raman measurements were performed using a laser Raman spectrometer (Jobin Yvon, Model T6400).

## ELECTROCHEMICAL MEASUREMENTS

To prepare electrode, the NHPCS/HPCS active material was mixed with PVDF and acetylene black with a mass ratio of 8:1:1. N-methyl-2-pyrrolidene (NMP) was added to the mixture and a uniform slurry was obtained after grinding. The slurry was then coated onto the Cu foil, and dried at 80◦C for 12 h. Subsequently, 2,032 coin cells were assembled using sodium metal foil as counter electrode, glass fiber as separator. The electrolyte was 1 M NaCF3SO<sup>3</sup> in DEGDME. The cycling and rate performance of NHPCS and HPCS was tested at Land BT2013A systems using CR2032 coin cell.

## RESULTS AND DISCUSSIONS

**Figure 1** shows the overall schematic illustration for the fabrication of N-doped hierarchical porous carbon microspheres (NHPCS). Firstly, Ni-MOF precursors were obtained via a simple hydrothermal method (see Experimental Section for detail). PVP plays an important role to form the Ni-MOF spheres during the hydrothermal process. The shape inducing effects of PVP could be attributed to the groups of hydrophobic vinyl and hydrophilic carbonyl that lead to the formation of polarized micelles. Subsequently, the obtained Ni-MOF precursors were calcinated with melamine under Ar atmosphere. This annealing process converts MOF to N-C microspheres while maintains the framework of Ni-MOF template, with the Ni nanoparticles uniformly distributing in the carbon microspheres. Finally, the NHPCS was obtained after removing the Ni nanoparticles by etching with HCl aqueous solution.

The morphology of the Ni-MOF, Ni/C, Ni/N-C microspheres was investigated by SEM (**Supporting Information**). The

SEM images in **Figures S1A,B** show that of Ni-MOF precursor exhibits a very uniform sphere structure with a diameter of around 1.5µm. This uniform sphere structure was successfully retained after the annealing process, as confirmed from the morphology of the as-formed Ni/C microspheres (**Figure S1C**) The existence of nickel metal was confirmed by XRD (**Figure S2A**; Lou et al., 2017). Ni/N-C microspheres was synthesized by annealing the Ni-MOF precursor with melamine. Notably, the uniform sphere structure was well maintained after the N-doping process, as shown in **Figure S1D**. XRD patterns were collected for HPCS and NHPCS (**Figure S2A**). A broad peak located at around 24.6◦ was observed for both samples, which attributed to the (002) diffraction of the disordered carbon structure (Xu et al., 2016; Niu et al., 2017). In the final product, the existence of small amount of Ni nanoparticles might greatly enhance its electronic conductivity.

The structural features of HPCS and NHPCS were examined by SEM and TEM. The SEM results strongly support that the morphology of HPCS and NHPCS are well consistent with the Ni-MOF precursor (**Figures 2A,B,D,E**), which confirms that the morphology of Ni-MOF precursor can be well maintained during the annealing process. TEM images of HPCS and NHPCS were displayed in **Figure S3**. There are abundant micropore interconnected with each other in HPCS and NHPCS. On the other hand, minor amount of Ni nanoparticles were retained in the obtained HPCS and NHPCS, which is in good accordance with the XRD results. The HRTEM images exhibit that the interlayer distance of HPCS and NHPCS are 0.34 and 0.35 nm, respectively (**Figures 2C,F**). This indicates that nitrogen doping can increase the interlayer spacing between the carbon layers (Yan et al., 2016). Moreover, the EDS elemental mapping of NHPCS shows that the N atoms are uniformly dispersed in the carbon microspheres, suggesting that the nitrogen is successfully doped into carbon spheres (**Figures 2H,I**).

**Figure 3A** displays the Raman spectra of HPCS and NHPCS. Two peaks are exhibited at around 1,338 and 1,582 cm−<sup>1</sup> , which can be ascribed to the D band and G band corresponding to defect-induced band and crystalline graphite band, respectively (Liu Y. et al., 2018). Therefore, the relative amount of disorder or defects in the carbon structure can be obtained according to the relative intensity ration of the D peak to the G peak (ID/IG). The ID/I<sup>G</sup> values of HPCS and NHPCS are 1.0 and 1.02, respectively, implying that nitrogen doping improves the disorder degree of carbon materials (Li et al., 2016c).

The XPS measurement was carried out to investigate the surface chemical composition of NHPCS. The survey spectra

HPCS and NHPCS.

indicate that C, O, Ni, and N elements exist in NHPCS, which can be further confirmed that the existence of small amount of Ni nanoparticles etching with HCl aqueous solution (**Figure S2B**). And the high-resolution N 1 s spectra as shown in **Figure 3B**, four peaks at 398.5, 399.2, 401.0, and 403.3 eV can be ascribed to the typical pyridinic, pyrrolic, graphitic, and quaternary nitrogen, respectively (Liu et al., 2016). The XPS results further confirm that the N-doped carbon was successfully obtained. Furthermore, nitrogen adsorption-desorption isotherms were performed to characterize the surface area and pore structure of HPCS and NHPCS. **Figures 3C,D** displays the nitrogen adsorptiondesorption isotherms for HPCS and NHPCS (inset of C and D are the corresponding pore size distribution curves), respectively. The two curves exhibit similar typical IV isortherm with H2-type hysteresis loops at the relative pressure (P/P0) range of 0.5– 1.0, suggesting the mesoporous features of the structures, and the specific surface area for HPCS and NHPCS are 200.21 and 223.25 m<sup>2</sup> g −1 , respectively. The pores of HPCS are mainly distributed at 3.97 and 7.81 nm, while the pores of NHPCS are mainly distributed at 3.72 and 21.15 nm according to the pore size distribution curves, which further suggesting the mesoporous features of the structures in the HPCS and NHPCS.

The electrochemical performance of HPCS and NHPCS were evaluated using 2,032 coin cells and 1 M NaCF3SO<sup>3</sup> in DEGDME used as the electrolyte. **Figure 4A** shows the CV curves of NHPCS electrode during the initial four cycles. In the first discharge process, two peaks at around 1.0 and 0.6 V can be observed, which are disappeared in next three cycles, corresponding to

the formation of SEI film and side reactions (Hong et al., 2014; Zhu et al., 2018). Subsequently, a strong redox peak at around 0.03 V can be observed, which can be ascribed to Na<sup>+</sup> insertion/extraction from NHPCS. And the CV curves are highly overlapped with each other after the first cycle, which indicating an outstanding reversible cycling stability.

**Figure 4B** displays the cycling performance of HPCS and NHPCS at 200 mA g−<sup>1</sup> , a high reversible capacity of 291.8 mA h g−<sup>1</sup> can be obtained after 100 cycles for NHPCS. And NHPCS also exhibits outstanding cycling stability during the charge/discharge process at 200 mA g−<sup>1</sup> . But for HPCS, lower reversible capacity of 245.9 mA h g−<sup>1</sup> was obtained after 100 cycles at the same current density. Furthermore, the coulombic efficiency of NHPCS approaches 100% during the cycling process, which is higher than HPCS. Subsequently, the charge and discharge profiles of NHPCS are provided in **Figure 4C**, a long slope below 1.0 V in the first sodiation process can be observed. For the charge process, the curve became steeper. The initial discharge and charge capacities of NHPCS are 387 and 291 mA h g−<sup>1</sup> , corresponding to an initial coulombic efficiency of 75.2%. And the charge and discharge profiles of HPCS are provided in **Figure S4A**, which show similar shapes to NHPCS. Furthermore, the initial discharge and charge capacities of HPCS are 297.2 and 239.3 mA h g−<sup>1</sup> , corresponding to an initial coulombic efficiency of 80.2 %. Power density is very important for SIBs in practical application for electric vehicles. Therefore, the rate performances of NHPCS and HPCS at different current density range from 200 to 1,000 mA g−<sup>1</sup> have been tested (**Figure 4D**). A high reversible capacity of 298, 285, 276, 265, and 256 mA h g−<sup>1</sup> at current density of 200, 400, 600, 800, and 1,000 mA g−<sup>1</sup> , respectively. And for HPCS, lower reversible capacity of 240, 231, 214, 192, and 172 mA h g−<sup>1</sup> at current density of 200, 400, 600, 800, and 1,000 mA g−<sup>1</sup> were obtained, respectively. These results indicate that high reversible capacity and cycling stability can be attributed to, which can facilitate Na<sup>+</sup> transport, offer more active sites for Na+, and improve the adsorption capacity of Na+. Furthermore, long-term cycling performance of NHPCS was further tested at 500 mA g−<sup>1</sup> , and the result is display in **Figure 4E**. NHPCS exhibits excellent cycling stability and high reversible capacity. 204 mA h g−<sup>1</sup> can be remained after 300 cycles.

To further analyze the reasons for the improved cycling and rate performance of NHPCS, the EIS measurements of NHPCS and HPCS were tested, and the corresponding Nyquist plots are shown in **Figure 5**. The two Nyquist plots exhibit similar shapes and consist of a semicircle in high-frequency region and a sloping line in low-frequency region (Pan et al., 2017; Zheng et al., 2018). Subsequently, charge transfer resistances of NHPCS and HPCS can be obtained according to the EIS curves and equivalent electric circuit (**Figure 5**). The Rct of NHPCS and HPCS are 3.98 and 5.37, respectively (as shown in **Table S1**). The lower charge transfer resistance of NHPCS was attributed to the nitrogen doped which can enhance electronic conductivity and facilitate the transport of Na<sup>+</sup> ions (Wang et al., 2013; Fu et al., 2014). Furthermore, the morphology change after cycling were also investigated to further analyze

the outstanding cycling stability of NHPCS. The SEM images of NC after 100 cycles as shown in **Figure S6**. The sphere structure was maintained well after 100 cycles, indicating that the NHPCS with a good structural stability during the cycling process.

NHPCS exhibits remarkable cycling performance and rate capability when tested as anode for SIBs, which can be owing to the unique hierarchical porous structure and N doping. The detailed reasons can be summarized as follows: (1) The unique hierarchical porous structure can offer more Na<sup>+</sup> storage active sites and buffer the volume change during the sodiation/desodiation process. (2) The hierarchical porous structure can facilitate Na<sup>+</sup> transport and provide a short diffusion distance for Na+. (3) The nitrogen doping can enhance electronic conductivity to facilitate the transport of Na<sup>+</sup> and the electrons.

#### CONCLUSIONS

In summary, N-doped hierarchical porous carbon microspheres (NHPCS) have been successfully fabricated from Ni-MOF with annealed exist of melamine. Benefiting from the nitrogen doped and hierarchical porous structure, NHPCS can offer more Na<sup>+</sup> storage active sites and provide fast Na<sup>+</sup> ions transfer channels. As a result, when evaluated as anode for SIBs. NHPCS can exhibit high reversible capacity of 291 mA h g−<sup>1</sup> at 200 mA g −1 . When the current density increase to 1,000 mA g−<sup>1</sup> , high reversible capacity of 256 mA h g−<sup>1</sup> was also can be obtained. Moreover, NHPCS possess a high reversible capacity of 204 mA h g−<sup>1</sup> after 300 cycles at 500 mA g−<sup>1</sup> . Therefore, NHPCS should be considered as promising anode materials for SIBs because of its outstanding rate capability and excellent cycling performance.

## DATA AVAILABILITY STATEMENT

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

## AUTHOR CONTRIBUTIONS

KX conducted the experiments. CY is the supervisor of this research work. KX and QP helped with writing. KX, QP, FZ, GZ, CW, and SW performed the characterization and data analysis. All authors involved the analysis of experimental data and manuscript preparation.

## REFERENCES


## ACKNOWLEDGMENTS

We gratefully acknowledge the financial support from the Guangdong Power Grid Co., Ltd. (Grant No. GDKJXM20161890).

## SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2019.00733/full#supplementary-material


**Conflict of Interest:** KX, GZ, CW, and SW are now employed by the company Electric Power Research Institute of Guangdong Power Grid Co., Ltd. The authors declare that this study received funding from Electric Power Research Institute of Guangdong Power Grid Co., Ltd. The funder had the following involvement with the study: KX, GZ, and CW were involved in the study design and data analysis. KX and SW were involved in the preparation of the manuscript and decision to publish.

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 Xu, Pan, Zheng, Zhong, Wang, Wu and Yang. 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.

# Optimal Quantity of Nano-Silicon for Electrospun Silicon/Carbon Fibers as High Capacity Anodes

Renheng Wang<sup>1</sup> \*, Yiling Sun<sup>1</sup> , Keyu Xiong<sup>1</sup> , Junchao Zheng<sup>2</sup> , Zhengfang Qian<sup>1</sup> \* and Zhenjiang He2,3 \*

*<sup>1</sup> College of Physics and Optoelectronic Engineering, Shenzhen University, Shenzhen, China, <sup>2</sup> School of Metallurgy and Environment, Central South University, Changsha, China, <sup>3</sup> College of Environmental Science and Engineering, Donghua University, Shanghai, China*

#### Edited by:

*Weihua Chen, Zhengzhou University, China*

#### Reviewed by:

*Chao Lai, Jiangsu Normal University, China Zhisheng Lv, Nanyang Technological University, Singapore*

#### \*Correspondence:

*Renheng Wang wangrh@szu.edu.cn Zhengfang Qian zq001@szu.edu.cn Zhenjiang He hzjcsu@csu.edu.cn*

#### Specialty section:

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

Received: *25 September 2019* Accepted: *02 December 2019* Published: *17 January 2020*

#### Citation:

*Wang R, Sun Y, Xiong K, Zheng J, Qian Z and He Z (2020) Optimal Quantity of Nano-Silicon for Electrospun Silicon/Carbon Fibers as High Capacity Anodes. Front. Chem. 7:867. doi: 10.3389/fchem.2019.00867* In this study, silicon/carbon composite nanofibers (Si@CNFs) were prepared as electrode materials for lithium-ion batteries via a simple electrospinning method and then subjected to heat treatment. The morphology and structure of these materials were characterized by X-ray diffraction (XRD), scanning electron microscopy (SEM), and transmission electron microscopy (TEM). The results show that the structure provides good electrical conductivity and affords sufficient space to accommodate volume expansion during charging/discharging. Furtherly, electrochemical performance tests show that the optimized Si@CNFs have an initial reversible capacity of 1,820 mAh g−<sup>1</sup> at a current density of 400 mA g−<sup>1</sup> and capacity retention of 80.7% after 100 cycles at a current density of 800 mA g−<sup>1</sup> . Interestingly, the optimized Si@CNFs have a superior capacity of 1,000 mAh g−<sup>1</sup> (400 mA g−<sup>1</sup> ) than others, which is attributed to the carbon substrate nanofiber being able to accommodate the volume expansion of Si. The SEI resistance generated by the Si@CNFs samples is smaller than that of the Si nanoparticles, which confirms that SEI film generated from the Si@CNFs is much thinner than that from the Si nanoparticles. In addition, the connected carbon substrate nanofiber can form a fiber network to enhance the electronic conductivity.

Keywords: composite anode, high capacity, silicon, electrospun, nanofiber

## INTRODUCTION

Lithium-ion batteries (LIBs) are recognized as the green energy of the new century due to their high specific energy, high operating voltage, long cycle life, environmental friendliness, and lack of a memory effect. They have been widely used in the energy storage systems of smartphones, portable audio-visual equipment, small aircraft, electric bicycles, and various small power tools. However, conventional graphite anodes and lithium metal oxide cathodes show relatively low specific energies (∼400 Wh kg−<sup>1</sup> theoretically and ∼200 Wh kg−<sup>1</sup> practically), which cannot meet the growing demand for large-capacity electrochemical energy storage.

The development of new materials with excellent performance and low cost is of great significance for improving battery performance and reducing battery cost. Therefore, research on LIBs should mainly focus on the three aspects of battery cost, battery capacity, and new electrode materials (Lv et al., 2018a,b, 2019; An et al., 2019; Li R. et al., 2019; Li Y. et al., 2019; Wu et al., 2019; Yuan et al., 2019). With the development of anode materials for LIBs, the defects and advantages of various new materials are highlighted. Among all anode materials, silicon (Si) is particularly promising due to its extremely high theoretical capacity (∼4,200 mAh g−<sup>1</sup> ), low discharge voltage, and natural abundance (Wen et al., 2013; Pan et al., 2015; Tang et al., 2015; Xie et al., 2015; Casimir et al., 2016; Wang and Yang, 2017; Yamaguchi et al., 2017; Zuo et al., 2017; Fang et al., 2019). However, Si as an anode material undergoes significant volume expansion during the Li<sup>+</sup> ion lithiation/delithiation process, leading to particle fracture and loss of capacity (Datta et al., 2011; Zhong et al., 2014; Huang et al., 2016; Jeong et al., 2016; Lin et al., 2016; Yang et al., 2016; Kim et al., 2017).

One generally accepted strategy to overcome this volume expansion is to synthesize nanoscale Si particles. Kim devised a simple, reproducible and cost-effective method for fabricating nanoporous Si flakes that demonstrate high electrochemical performance (Kim et al., 2017). Chen et al. reported an effective method for synthesizing Si nanotubes that exhibit significantly improved electrochemical properties (Wen et al., 2013). However, the above method of nanocrystallization only partially alleviates this problem and meanwhile improves the cycle performance to only a limited extent. Nowadays, silicon/carbon (Si/C) is being widely synthesized for use as an anode in LIBs because it combines the advantages of both carbon and Si and has been proposed to a very promising material to replace the commercially available graphite. Nathalie et al. have developed a two-stage LCVP to synthesized coreshell Si/C nanoparticles (Sourice et al., 2016). Sun reported a 3D Si/CNCS nanocomposite consisting of a CNC three-dimensional interconnected conductive network and a unique Si/CNC hollow sphere structure. The Si/CNCS nanocomposite can effectively alleviate volume expansion and improve electronic conductivity (Yue et al., 2016). Additionally, Si/C composite nanofibers (Si@CNFs) can be prepared for use as anodes in LIBs through a simple and low-cost method combining electrospinning and subsequent thermal treatment (Ji and Zhang, 2009; Hwang et al., 2012; Wu et al., 2014, 2016).

In this work, Si@CNFs materials were fabricated via a facile electrospinning technique. The Si@CNFs display an initial reversible capacity of 1,820 mAh g−<sup>1</sup> at a current density of 400 mA g−<sup>1</sup> and capacity retention of 80.7% after 100 cycles at a current density of 800 mA g−<sup>1</sup> and exhibit a superior capacity of 1,000 mAh g−<sup>1</sup> at a current density of 400 mA g −1 . The SEI resistance generated by the Si@CNFs samples is smaller than that of the Si nanoparticles, which confirms that SEI film generated from the Si@CNFs is much thinner than that from the Si nanoparticles. In addition, the connected carbon substrate nanofiber can form a fiber network that enhances the electronic conductivity. The crystal structure was measured by X-ray diffractometer (PAnalytical X'Pert ProMRD, Holland). The surface morphology was analyzed by field emission scanning electron microscopy (SEM, JEOL, JSM-5612LV). The microstructure was observed by transmission electron microscopy (TEM, JEOL-JEM-2100F). The cycling performances of Si nanoparticles and Si@CNFs are evaluated. The charge transfer resistances of Si nanoparticles and Si@CNFs anodes are analyzed by electrochemical impedance spectroscopy (EIS).

## EXPERIMENTAL METHODS

The preparation of Si@CNFs was carried out through the facile electrospinning method shown in **Figure 1a**. Specifically, first, 1 g polyacrylonitrile (0.15 M), purchased from Sigma Aldrich, was dissolved in 10 mL N, N-Dimethylformamide (DMF). Meanwhile, 0.10, 0.20, or 0.30 g of Si nanoparticles was mixed well with the above solution, followed by stirring and keeping the mixed solution under sonication for 10 min. This solution was added to the nanofibers for electrospinning using an electrospinning device. The operating parameters were as follows: an applied voltage of direct current (DC) 25 kV (Dongwen, P503-1ACDF), working distance of 15 cm between the collector (aluminum foil) and solution syringe needle, and pumping rate of 2 mL h−<sup>1</sup> . The prepared composite nanofibers were annealed in air at 260◦C for 2 h and in an argon atmosphere at 600◦C for 6 h to obtain Si@CNFs. The Si nanoparticles were labeled as S1. The Si@CNFs prepared using 0.10, 0.20, and 0.30 g Si nanoparticles correspond to the names S2, S3, and S4 in this article, respectively.

The electrochemical performances of all samples were tested using 2025-type coin half-cells. Celgard 2400 microporous polypropylene membrane was used as the separator, and the lithium metal sheet was used as the negative electrode. One mol L−<sup>1</sup> LiPF6/EC+EMC+DMC (by weight ratio, 1:1:1) was used as the electrolyte. Si nanoparticles, polyvinylidene fluoride (PVDF), and acetylene black were weighed according to the mass ratio of 8:1:1 and thoroughly mixed and ground, an appropriate amount of N-methylpyrrolidinone (NMP) was added to prepare an electrode slurry, and the slurry was uniformly coated on the copper foil with an applicator. After drying in a vacuum oven at 120◦C for 15 h, it was punched into a pellet with a mass of 2.0 mg. The Si@CNFs were used directly as an electrode film. The CR2025 button cell was assembled in a glove box filled with argon at a relative humidity of less than 5% and an oxygen pressure of less than 1.0 ppm. After 12 h of activation, the electrochemical performance tests were performed in a voltage range of 0.01– 2.0 V from 400 to 4,000 mA g−<sup>1</sup> .

#### RESULTS AND DISCUSSION

**Figure 1b** shows the typical XRD patterns of all samples. The pure carbon fiber shows a diffraction peak at 2θ =25.0◦ , belonging to the (002) plane (Li et al., 2014). However, the peak of the nanofiber is weak and broad, resulting from disordered carbon. The diffraction peaks of Si@CNFs are at 2θ values of 28.5◦ , 47.4◦ , 56.3◦ , 69.2◦ , 76.8◦ , and 88.3◦ , respectively, corresponding to the (111), (220), (311), (400), (331), and (422) peaks of Si crystals in the nanofibers (Chen et al., 2011; Xie et al., 2014). It is obvious that the fluctuations in the XRD patterns of S2, S3, and S4 at 2θ of 5◦ -40◦ are caused by the action of carbon nanofibers. In addition, under these synthesis conditions, the synthesis procedures have no impact on the crystal structure of silicon.

**Figures 1c–f** illustrates SEM images of Si nanoparticles and electrospun Si@CNFs. The size of Si nanoparticles is about 30– 60 nm, as shown in **Figure 1c**, as they easily form agglomerated

particles. However, the independent fiber skeleton of Si@CNFs has diameters of 300–500 nm, as displayed in **Figures 1d–f**. The net structure of the interconnected carbon network can be beneficial in improving the electrode surface and providing fast ion-conducting channels. The structure is expected to provide good electrical conductivity and afford sufficient space to accommodate volume expansion during charging/discharging. Besides, a compact fiber can avoid the formation of an unstable solid electrolyte interphase (SEI) layer on the surface of Si nanoparticles (Wu et al., 2016). The distributions and elemental mappings of Si in Si@CNFs were measured by energy dispersive spectrometer (EDS), as shown in **Figure 2**. The Si component is uniformly distributed in the three matrixes along both the radial and axial directions.

**Figure 3** displays the microstructures of Si nanoparticles (S1) and Si@CNFs (S3). The clear crystal grain edge and the complete crystallinity of the 0.3123 nm lattice spacing are clearly seen in **Figures 3a,b**. The lattice spacing is derived from Si (111), with different lattice stripes extending to the particle boundaries. As seen in **Figures 3c,d**, the structure of the carbon nanofibers looks like a fishnet. The Si in S3 is well wrapped up by carbon nanofibers, providing enough space to accommodate huge volume changes during charging and discharging (Zhou et al., 2016; Wang et al., 2017a,b). It can be concluded that electrospinning is an effective technique for synthesizing S3.

**Figure 4** and **Table 1** show the electrochemical performances of Si nanoparticle and Si@CNF half-cells at different current

FIGURE 2 | EDS images of sample S2 (a,b), sample S3 (c,d), and sample S4 (e,f).

densities. **Figures 4A–D** display the initial charge and discharge of four samples. It can be seen that the initial charge-discharge capacities of Si@CNF samples are higher than that of the Si nanoparticle sample at all current densities. When the current density is 400 mA g−<sup>1</sup> , the capacity of the Si nanoparticles is 3,555 mAh g−<sup>1</sup> , but the capacity declines sharply to 966 mAh

FIGURE 3 | HRTEM images of Si nanoparticles (S1) (a,b), Si@CNFs (S3) (c,d).

g <sup>−</sup><sup>1</sup> when the current density increases to 4,000 mA g−<sup>1</sup> , only 27.2% of the initial capacity. However, Si@CNFs (take sample S3 as an example) have better rate performances. The initial capacities are 1,820, 1,166, 1,050, and 1,000 mAh g−<sup>1</sup> at 400, 800, 2,000, and 4,000 mA g−<sup>1</sup> , respectively. It is clear from **Figure 4E** that Si nanoparticles still delivered the worst cycle performance at 800 mA g−<sup>1</sup> , only 5.8% of the initial capacity after 50 cycled charging/discharging processes. However, Si@CNFs remarkably improved the cycle stability vs. Si nanoparticles. After 100 cycles, the discharge capacities of sample S2, S3, and S4 represent 89.3, 80.7, and 50.8% of the initial capacity, respectively. Obviously, a reduction in the capacity of Si@CNFs will occur with a decrease in the silicon content, which is due to the higher specific capacity of silicon nanoparticles than of carbon nanofibers. However, the cycle stability of Si@CNFs dropped with an increase in silicon content. Therefore, ensuring that there is an appropriate amount of silicon nanoparticles wrapped in carbon nanofibers is a key point for improving the rate capacity and


Si@CNFs (S2, S3, S4) between 0.01 and 2.0 V at 25◦C.

cycling performance at the same time. Here, sample S3 displays the best comprehensive performance.

In order to achieve an in-depth understanding of the superior electrochemical performance of Si@CNFs, EIS measurements were used to evaluate the electrode kinetics factors of Si nanoparticles and Si@CNFs, as shown in **Figure 5**. The inset illustration is the equivalent circuit diagram fitted by ZView software (Wang et al., 2017a,b). From **Figure 5A**, it can be found that the EIS plots were composed of three parts: a semicircular arc at high frequency corresponding to SEI resistance (Rsf), a semicircle at intermediate frequency related to charge-transfer resistance (Rct), and a straight line at low frequency connected to the Warburg diffusion process of lithium ions into the electrode material (WO). R<sup>e</sup> represents solution resistance, and CPE is attributed to the double-layer capacitance and the passivation film capacitance. The corresponding resistance values are displayed in **Table 2**. The R<sup>e</sup> values of all samples are very similar and small. The Rsf and Rct values of the Si@CNFs samples are smaller than those of the Si nanoparticles, which confirms that the SEI film generated from the Si@CNFs was much thinner than that from the Si nanoparticles. This may prove that the Si@CNFs can effectively reduce the diffusion resistance and charge transfer resistance, thus making it easier for Li<sup>+</sup> to migrate/embed in the lattice. It is noted that Rct increases from 68.0 (S2) to 332.8 Ω (S1) when the Si content is increased. The decrease in Rct may result from the enhanced electron and ionic conductivity of carbon nanofibers. In detail, the existence of carbon nanofibers can reduce Rct, but the influence becomes weak when the silicon content reaches a certain value. Furthermore, the specific capacity of Si@CNFs decreased sharply when the silicon content was reduced. In addition, the exchange current densities (i 0 ) were calculated by the formula (i <sup>0</sup> = RT/nFRct), and the results are shown in **Table 2**. The S2 sample has the highest exchange current density (3.78 × 10−<sup>4</sup> mA cm−<sup>2</sup> ) in all samples, and S3 also has a higher exchange current density (2.49 × 10−<sup>4</sup> mA cm−<sup>2</sup> ) than S1 (7.72 × 10−<sup>5</sup> mA cm−<sup>2</sup> ) and S4 (1.78 × 10−<sup>4</sup> mA cm−<sup>2</sup> ). Thus, the EIS results indicate that the Si@CNFs possesses good charge transfer kinetics within the electrode.

TABLE 2 | Impedance parameters of Si nanoparticles and Si@CNFs.


**Figure 5B** shows the relationship between the real axis and the square root of the lower angular frequency (Xie et al., 2014; Wang et al., 2017b). The slopes of the straight lines indicate the values of the Warburg impedance coefficient (σw). It is noted that the lithium-ion diffusion process resistance of sample S2 is lower than those of the other samples, indicating that its net structure has the shortest diffusion length.

#### CONCLUSION

Si@CNFs materials have been synthesized by electrospinning technology, and the crystal structures of Si@CNFs are similar to those of Si nanoparticles. The Si nanoparticles in Si@CNFs are well wrapped up by carbon nanofibers. This network structure may provide good electrical conductivity, afford sufficient space to accommodate volume expansion during charging/discharging, enhance Li<sup>+</sup> diffusion, and reduce charge-discharge resistance. It is interesting that the optimized Si@CNFs hold a superior capacity of 1,000 mAh g−<sup>1</sup> (400 mA g−<sup>1</sup> ), which is attributed to the carbon substrate nanofiber being able to accommodate the volume expansion of Si. The SEI resistance generated by Si@CNFs is lower than that of others. However, too much carbon content will decrease the specific capacity of Si@CNFs. Sample S3 displays the best comprehensive performance and delivers a reversible capacity of 1,820 mAh g−<sup>1</sup> between 0.01 and 2.0 V at 400 mA g−<sup>1</sup> , corresponding to 80.7% of the 100-cycle initial discharge capacity at 800 mA g−<sup>1</sup> . We believe that Si@CNFs are a promising anode material for LIBs.

### DATA AVAILABILITY STATEMENT

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

#### AUTHOR CONTRIBUTIONS

RW and ZH designed and engineered the samples and performed the experiments. RW, ZH, and ZQ performed the data analysis. ZH, YS, ZQ, 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

#### REFERENCES


Science and Technology Innovation Commission of Shenzhen (Grant No. 20180123), the Natural Science Foundation of Guangdong Province (2019A1515012111), the Shenzhen Science and Technology Program (KQTD20180412181422399), the National Science Foundation of China (Grant No. 51604081), and the Chenguang Program supported by the Shanghai Education Development Foundation and Shanghai Municipal Education Commission (Grant No. 16CG40).

#### ACKNOWLEDGMENTS

The authors also acknowledge the assistance on SEM and TEM observations received from the Electron Microscope Center and Instrumental Analysis Center of Shenzhen University (Xili Campus).

composite electrode for high-performance lithium-ion battery. Front. Chem. 7:555. doi: 10.3389/fchem.2019.00555


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

# Defect Rich Hierarchical Porous Carbon for High Power Supercapacitors

Peng Cai <sup>1</sup> , Kangyu Zou<sup>1</sup> , Xinglan Deng<sup>1</sup> , Baowei Wang<sup>1</sup> , Guoqiang Zou<sup>1</sup> \*, Hongshuai Hou<sup>1</sup> and Xiaobo Ji 1,2

*<sup>1</sup> College of Chemistry and Chemical Engineering, Central South University, Changsha, China, <sup>2</sup> School of Materials Science and Engineering, Zhengzhou University, Zhengzhou, China*

Tuning hierarchical pore structure of carbon materials is an effective way to achieve high energy density under high power density of carbon-based supercapacitors. However, at present, most of methods for regulating pores of carbon materials are too complicated to be achieved. In this work, a durian shell derived porous carbon (DSPC) with abundant porous is prepared through chemical activation as a defect strategy. Hierarchical porous structure can largely enhance the transfer rate of electron/ion. Furthermore, DSPC with multiple porous structure exhibits excellent properties when utilized as electrode materials for electric double layer capacitors (EDLCs), delivering a specific capacitance of 321 F g−<sup>1</sup> at 0.5 A g−<sup>1</sup> in aqueous electrolyte. Remarkably, a high energy density of 27.7 Wh kg−<sup>1</sup> is obtained at 675 W kg−<sup>1</sup> in an organic two-electrode device. And large capacity can be remained even at high charge/discharge rate. Significantly, hierarchical porous structure allows efficient ion diffusion and charge transfer, resulting in a prominent cycling stability. This work is looking forward to providing a promising strategy to prepare hierarchical porous carbon-based materials for supercapacitors with ultrafast electron/ion transport.

#### Edited by:

*Cheng Zhong, Tianjin University, China*

#### Reviewed by:

*Jia Ding, Tianjin University, China Jin Huile, Wenzhou University, China*

> \*Correspondence: *Guoqiang Zou gq-zou@csu.edu.cn*

#### Specialty section:

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

Received: *04 December 2019* Accepted: *14 January 2020* Published: *04 February 2020*

#### Citation:

*Cai P, Zou K, Deng X, Wang B, Zou G, Hou H and Ji X (2020) Defect Rich Hierarchical Porous Carbon for High Power Supercapacitors. Front. Chem. 8:43. doi: 10.3389/fchem.2020.00043* Keywords: high power density, hierarchical porous carbon, supercapacitors, biomass carbon, EDLC

## INTRODUCTION

Supercapacitors (SCs) have been triggered intensive attention in energy storage due to their fast reaction dynamics, ultrahigh power density, and excellent cycling stability (Li et al., 2011; Xie et al., 2012; Hao et al., 2013; Wang and Xia, 2013; Béguin et al., 2014). Among various kinds of electrode materials for SCs, carbon materials are most promising because of their large availability, high conductivity, and cycling stability. Preparing carbon materials economically and efficiently, increasing the energy density, and enhancing the power density of carbon materials are important aspects of developing carbon materials for advanced supercapacitors (Chen X. et al., 2014; Han et al., 2014; Yang and Zhou, 2017). Since the capacity of carbon materials depends on the amounts of electrostatic charges stored at the interfaces of the electrodes and electrolytes, a lot of researches have been done to design the desired carbon materials for ions storage. Tailoring the hierarchical pore structure to match the appropriate electrolyte ions is an effective method to achieve large capacitance, efficient ion diffusion, and charge transfer (Zhu et al., 2011; Fang et al., 2012; Chen et al., 2016, 2017; Yu et al., 2016; Liu et al., 2017; Shi et al., 2017, 2019).

In recent years, some brilliant materials (Li et al., 2015, 2019a; Han et al., 2016; Cao et al., 2017; Lv et al., 2017a,b; Xu B. et al., 2017; Zhou and Hu, 2017; Zhou et al., 2018), including carbon nanotubes, grapheme, and their derivatives, have been widely explored and utilized because of their excellent thermal stability, remarkable tenacity, and high strength (Wen et al., 2016; Zhu et al., 2017; Liu et al., 2018). However, there are some obstacles in applying these materials as electrode materials for supercapacitors. The charge/discharge processes are directly affected by ionic kinetics, obviously under harsh charge/discharge conditions (current density higher than 10 A g−<sup>1</sup> ) (An et al., 2001; Ra et al., 2009; Gao et al., 2012; Hahm et al., 2012; Sun et al., 2013; Zhu et al., 2016a). For instance, a capacitance of 362.2 F g−<sup>1</sup> is obtained for the treated fullerene/graphite carbon electrodes at the low current density. However, when the current density increases to fifty times, the capacitance drops rapidly to half of the original value (Chen Z. et al., 2014; Zheng et al., 2015; Xu M. et al., 2017; Chen et al., 2018; Li et al., 2019b). It is not hard to note that increasing energy density under high power density of carbon electrodes is a great challenge. To the best of our knowledge, the poor performance in large current condition is associated with the sluggish ion migration in pores (Ye et al., 2019). It is difficult to keep good performances during charge/discharge processes without reasonable pore design. And electronic conductivity and ionic kinetics in electrodes can also be improved through regulating intrinsical structures of carbon materials (Raymundo-Piñero et al., 2011; Mun et al., 2013; Zhu et al., 2016b). In general, internal pores affect ion transport while external pores are related to ion diffusion from electrolyte. Particularly, the reduced electrode potential caused by low access to ion in pores is the main impediments under large current (Presser et al., 2011; Gao et al., 2013; Wang et al., 2015). In order to explain the importance of the pores in carbon electrode theoretically, the transmission line model is employed to imitate the charge reserve characteristic of different pore shapes, showing that porous electrodes are conducive to charge accumulation (Black and Andreas, 2010; Zhang et al., 2017). The results are proposed that electrodes with multiscale pores are beneficial to ion diffusion, which is the key to obtain large capacitance at ultra-high current density. Micropores larger than 0.7 nm are found to be used as electrolyte reservoirs to store ions (Péan et al., 2014). Under the effect of the electric field, due to the shortened length of the ion diffusion, the accumulated ions can diffuse into the nearby micropores in a shorter time than the ions in the electrolyte. And pores larger than 2 nm assist to decrease the ion confinement in the micropores, significantly enhancing the effective self-diffusion coefficient of the ions, which is advantageous for obtaining excellent rate performance (Forse et al., 2017). Therefore, a better understanding of the relationship between pore structure and capacitance performance is essential to fabricate hierarchical porous structure in carbon materials. It will accelerate ion kinetics to obtain carbon electrode materials with high ion transfer rate, which can achieve high energy storage under high current density. In order to obtain highperformance carbon electrode materials with hierarchical pore stricture, the strategy for fast ion transfer leaded by economical and efficient synthesis has been focused as main interests (Chen et al., 2013; Wang et al., 2017).

The main approaches to design hierarchical porous carbon materials are as following: (1) pores or defects generated on the surface or matrix through physical or chemical activation; (2) pores created by multidimensional carbon coupling/selfassembly; (3) adjustment of microstructures by means of templates; (4) pores congregated by the polymerization of nanoscale building blocks (Kang et al., 2015; Ni and Li, 2016; Yin et al., 2018; Kong et al., 2019). In these methods above, the size and shape of the pore array can be well-adjusted by changing the characteristics of template. However, the instability of nanoscale templates under high temperature or solvent treatment limits its extensive application (Fuertes et al., 2005; Lv et al., 2012). The process of multi-dimensional carbon coupling with the guest species and polymerization through building blocks is also tedious. Consequently, a wide range of biomass precursors combined with efficient activation methods are properly employed to prepare high-performance hierarchical porous carbon materials. Hierarchical pores in the surface would be beneficial to the transport of ions between adjacent layers, maximizing the reachable surface area. Due to the resulted multiple porous structure, the advantageous electronegativity also shows that carbon materials possess superior electrochemical performance in energy storage (Gao et al., 2019; Cai et al., 2020). Durian is widely distributed in the Southeast Asia region and around 90,000 metric tons of durian wastes were generated annually. Note that the advantages of durian shells are easy availability, cheap and high economic value. The reason of choosing durian shell as precursor to prepare carbon electrode can be attributed to main two reasons: (1) the hierarchical porous structure of the durian shell shows fast mass transfer channel, carbon material derived from durian shell precursor maintain the inherit structure features, possessing the fast ion/electron transport rate; (2) durian shell contain some heteroatoms, which can introduce defects during carbonization process, increasing the wettability of the surface of the carbon material, and promoting the contact of electrolyte area. In this work, a multi-stage porous carbon material (DSPC) is successfully obtained through the pre-carbonization and activation of the durian shell. By adjusting the activation temperature, the degree of porosity of the DSPC can be well-manipulated. The DSPC exhibits typical hierarchical porosity, reasonable aperture distribution, and high specific surface area. Surprisingly, symmetric supercapacitors based on DSPC electrodes deliver large capacity and fast ion transfer characteristics in aqueous electrolyte and organic electrolyte.

### EXPERIMENTAL SECTION

#### Preparation of DSPCs

Firstly, the durian shell is cleaned several times by distilled water to remove some impurities and put into an oven under 120oC and for 24 h. Secondly, the dried durian shell is directly transferred to a tubular furnace in Ar atmosphere. The carbonized temperature is set to 400◦C for 2 h with a heating rate of 3oC min−<sup>1</sup> . The obtained sample is marked as DSPC. Finally, the DSPC is manufactured with surface defects to obtain hierarchical channels. DSPC is mixed and grinded with KOH of 4:1 mass ratio. In order to explore the most appropriate the ratio of KOH to C, the corresponding electrochemical results are obtained in **Figure S1**. They are divided into five parts for subsequent activation and immediately transferred to a tubular furnace for generating surface defects. The temperature is 600, 700, 800, 900, and 1,000◦C for 1 h. All the samples after being calcined are added 6 M HCl until pH = 7. All samples are then washed to neutral and dried in an oven. The carbon materials activated under different temperature are denoted as DSPC-600, DSPC-700, DSPC-800, DSPC-900, and DSPC-1000, respectively.

#### Material Characterization

The structure of DSPCs are explored by X-ray diffraction (XRD) with a scan rate of 10◦C min−<sup>1</sup> by using Kα radiation of Cu (λ = 0.1542 nm). Degree of graphitization is measured through Raman spectroscopy (DXR, Thermo-Fisher Scientific). The morphology of all the samples are investigated by fieldemission scanning electron microscopy (FESEM, JSM-7600F, JEOL) and transmission electron microscopy (JEM-2010F at 200 kV, JEOL). The specific surface area is acquired from the N<sup>2</sup> adsorption-desorption isotherms and is obtained by the Brunauere Emmette Teller (BET) equation. The chemical bond of DSPCs are collected from X-ray photoelectron spectrometer (XPS) instrument.

#### Electrochemical Characterization

All the samples are measured on a three-electrode system using aqueous electrolyte and on two-electrode system using aqueous electrolyte and organic electrolyte for determining their capacitive performance.

For three-electrode aqueous electrolyte system, every carbon samples are grinded with carbon black and polytetrafluoroethylene (PTFE) with a weight ratio of 8:1:1 loaded on nickel foam. The electrode plates are then dried in a vacuum at 80◦C for 12 h. Pieces are pressed under 10 MPa for 1 min. The mass load of each electrode plate is 2.2 mg/cm<sup>2</sup> . Platinum electrode is selected as the counter electrode, and a saturated calomel electrode is selected as the reference electrode. The electrolyte is 6 M KOH.

For two-electrode aqueous electrolyte system, every carbon samples are grinded with carbon black and polytetrafluoroethylene (PTFE) with a weight ratio of 8:1:1 loaded on nickel foam. The electrode plates are then dried in a vacuum at 80◦C for 12 h. Pieces are pressed under 10 MPa for 1 min. The electrolyte is 6 M KOH. CR2016-type coin cells are assembled accordingly. Electrochemical test begins after several hours.

For two-electrode organic electrolyte system, each carbon samples are grinded with carbon black and polyvinylidene fluoride (PVDF) with a weight ratio of 8:1:1 loaded on Al foils. The electrode plates are then dried in a vacuum at 120◦C for 12 h. CR2016-type coin cells are assembled inside the Braunglovebox with argon atmosphere (<0.1 ppm of both O<sup>2</sup> and H2O). Galvanostatic charge/discharge (GCD) tests are carried out between 0 and 2.7 V by an Arbin BT2000 instrument. The cyclic voltammetry (CV) tests are carried out at different scan rates on a Solartron 1470 Multistat system. The electrochemical impedance spectroscopy (EIS) measurements are performed using an electrochemical workstation.

The gravimetric capacitance is calculated from the CV curves and discharge curves, according to the formula:

$$Cs = \frac{I\Delta t}{m\Delta V}$$

where Cs is the specific capacitance and 1V is the voltage drop. 1t is the discharging time. I/m is referred to the current density.

The power density and energy density of the devices are obtained thorough using the following equations:

$$\begin{aligned} P &= \frac{\triangle V \times I}{2 \times m} \\ E &= \frac{P \ast t}{3600} \end{aligned}$$

Here, 1V is referred to the discharge voltage, I is referred to the discharge current, t is referred to the discharge time and m is referred to the total mass of active materials, respectively.

## RESULTS AND DISCUSSION

SEM and TEM images of hierarchical porous carbon materials prepared from the durian shell are shown in **Figure 1** and **Figure S2**. It is obvious that the pores are hardly manufactured at 600◦C. When the temperature is increased to 700◦C, the pore structure in the carbon material is gradually generated, and an uneven structure is produced on the surface. As the temperature is further risen up to 800◦C, the pore structure formed in the carbon material are widely ranged and the structure is uniform. When the temperature is raised to 900◦C, the defect of the surface is degenerated and destroyed, and the pore structure of the material is broken. As the temperature is kept at 1,000◦C, the structure of hierarchical pores is collapsed more severely, and the surface defect structure is difficult to be observed. According to TEM images, an amorphous dominated porous structure and defect architecture are further revealed.

The XRD measurement of all samples are exhibited in **Figure 2A**. The broad peaks located at 23◦ and inconspicuous peaks centered at 43◦ are perceived, indicating that all the samples are in amorphous constructions maintained few regions of crystallinity (Hou et al., 2015, 2017; Ge et al., 2018; Zou et al., 2018; Wu et al., 2019). More detailer pore structures of the samples are collected by the nitrogen adsorption/desorption measurements in **Figures 2B,C**. The isotherms of all DSPCs exhibit a combination of I-type and IV-type isotherm characteristics at a relative pressure >0.4, suggesting that both micropores and mesopores exist. Notably, the specific surface area of DSPC-600, DSPC-700, DSPC-800, DSPC-900, and DSPC-1000 are 1006, 1217, 2535, 2602, and 899 m<sup>2</sup> g −1 , respectively. When the temperature is increased slightly, the surface area of micropores and mesoporous increase. As the increase of temperature, the proportion of micropores decreases while the proportion of mesopores increases according

to the higher platform and lower starting angles of the lowpressure areas of DSPC-900 in **Figure 2B**. The plots of pore size distributions are revealed the porous structure for all the samples. Furthermore, wide existence of nanopores in plots of pore size distributions illustrates hierarchical porous structure of DSPC-800. Risen from 800◦C to 1,000◦C, the distribution of micropores gradually decreases and the distribution of mesopores becomes larger. In general, as depicted in **Figure 2C**, the DSPC-800 exhibits a hierarchically aperture and porosity structure. The Raman spectra of the acquired samples are deconvoluted into two peaks including G peak and D peak, and the intensity ratio of G band vs. D band (IG/ID) can be used to describe the degree of graphitization. The G peak is induced by Sp2-hybrid carbon and D peak is related to the disordered graphite structure. As shown in **Figure 2D**, the values of IG/ID of all samples are 1.65, 2.52, 6.6, 2.06, and 1.59, respectively, demonstrating the high graphitization degree of DSPC-800. Note that Sp2 hybridizationdominant carbon is widely employed to improve ion transfer rate and efficient energy storage because of large graphitization, high electrical conductivity, and thermal stability. Benefiting from high value of IG/ID in DSPC-800 superior kinetics performance during chare/discharge process could be delivered. Furthermore, the surface chemical environment of DSPC-800 is determined in details by XPS analysis. Carbon content in the DSPC-800 is 91.02 at%, and the oxygen content of the DSPC-800 is 8.98 at %. As shown in **Figure 2E**, the C1s profiles is divided into four peaks of COOH (290 eV), C = O (286.7 eV), C-O (285–286.eV) and C = C/C-C (284.6 eV). The high-resolution O1s spectra can be deconvoluted into three peaks located at 531.0, 533, and 536 eV, which is corresponding to the C=O, C-O, and carbonyl (**Figure 2F**), respectively (Yang et al., 2019; Zhu et al., 2020).

The capacitive performance of DSPCs are firstly measured in 6.0 M KOH/H2O electrolyte by using a three-electrode system. The CV curves of all the samples exhibit quasi-rectangular appearance at the scan rate of 5–100 mV s−<sup>1</sup> , implying an ideal capacitive characteristic in **Figures 3a1–e1**. The DSPC-800 shows the largest area of CV curves among all the samples, suggesting outstanding charge storage feature. In addition, according to the

results in **Figure S1**, the ratio of KOH to C should be selected as 4: 1 rather than 3:1 or 5:1. As displayed in **Figure 3c2**, the GCD curve of DSPC-800 presents the longest discharge time among the five samples at the current density of 10 A g−<sup>1</sup> , suggesting the largest capacitance among all the samples. Based on the results of GCD curves, the specific capacitances of DSPC-600, DSPC-700, DSPC-800, DSPC-900, and DSPC-1000 are 239.5, 336.6, 320.5, 209.6, and 186 F g−<sup>1</sup> at a current density of 0.5 A g−<sup>1</sup> (**Figures 3a2,b2,d2,e2**), exhibiting that the specific capacitances of DSPCs are 194.1, 249.1, 253.1, 158.9, and 159 at 5 A g−<sup>1</sup> , respectively. Impressively, the specific capacitances of DSPCs are 130.4, 93.7, 154.75, 50, and 60 F g−<sup>1</sup> at a current density of 50 A g−<sup>1</sup> , separately. The capacitive performances of DSPCs are subsequently evaluated in 6.0 M KOH/H2O electrolyte using a two-electrode system. The CV curves of the device are listed in **Figure 3c3**, in which the quasi-rectangular appearance of DSPC-800 is kept even at the scan rate of 100 mV s−<sup>1</sup> , demonstrating an excellent rate performance. This result is further illustrated by the GCD curves at different current densities in **Figures 3a3–e3**. When the current density is 0.5 A g−<sup>1</sup> , the specific capacitances of DSPCs devices are determined as 56.1, 60.9, 82, 52, and 46.3 F g −1 , respectively, corresponding to the specific capacitances of 45.1, 42.7, 63.3, 33, and 32 F g−<sup>1</sup> at 5 A g−<sup>1</sup> , respectively. As the current density increases to 20 A g−<sup>1</sup> , specific capacitances of 30.2, 25.1, 32.1, 9.3, and 12 F g−<sup>1</sup> are achieved for DSPCs samples. The highest specific capacitance of DSPC-800 is attributed to hierarchical porous structure, which can bring numerous active sites for the charge storage in **Figures 3a4–e4**.

All the samples are further tested in organic electrolyte in two-electrode devices based on DSPCs. The device of DSPC-600 measured in organic electrolyte is shown an obvious polarization, which is attributed to sluggish ion kinetics behavior of the electrode (**Figure 4a1**). The low energy density of the device of DSPC-600 is 0.01 Wh kg−<sup>1</sup> , which is suggested andante charge transport and ion diffusion in large current density shown in **Figure 4a3**. The low capacity can be attributed to the less surface defects and low conductivity due to the low activation temperature. The other four devices are shown with a quasi-rectangular shape CV curve that can be interpreted as the presence of higher surface defects. The DSPC-800 device shows a highly symmetric shape, suggesting great electrochemical reversibility (**Figures 4b1–e1**). It is important to noted that the device of DSPC-800 exhibits smallest IR drop at 10 A g−<sup>1</sup> , implying that surface defect structure with relatively high hierarchical aperture and porosity can facilitate transfer rate of the electrolyte ions and decrease the ionic diffusivity resistance. This result is further illustrated by the GCD curves at different current densities (**Figures 4a3–e3**). The energy density of all the devices are 21.5, 24.3, 27.7, 22, and 10.4 Wh kg−<sup>1</sup> at a power density of 675 W kg−<sup>1</sup> , respectively. Furthermore, energy densities of 0.6, 20.7, 21.5, 16.4, and 8.4 Wh kg−<sup>1</sup> can be obtained for related samples at a power density of 6,750 W kg−<sup>1</sup> . Surprisingly, even at 67,500 W kg−<sup>1</sup> , the device of DSPC-800 can also deliver an energy density of 4.2 Wh kg−<sup>1</sup> . The highest energy density of the devices with DSPC-800 as electrode are attributed to multiple pores

that can supply numerous active sites for the charge storage in **Figure 4**. It is clear that hierarchical porous structure can effectively improve the electrochemical properties of materials in both aqueous electrolyte and organic electrolyte system, suggesting that the introduction of hierarchical porous structure is significant.

A number of electrochemical measurements including cycling stability and electrochemical impedance spectroscopy (EIS) are carried out to investigate the electrochemical performance of DSPCs. Low resistance and rate capability are determined by the efficiency of charge transport and ion diffusion in aqueous electrolyte system (Xu et al., 2014). The contact resistance between the electrode material and the electrolyte also affects resistance (Rs). Because the ion transfer characteristics of these synthesized materials are different, Rs is generally different (Jin et al., 2018a,b; Nawwar et al., 2019; Wang and Cui, 2019;

Yuan and Lu, 2019). The surface defect of DSPCs is critical for achieving such high specific capacitance. EIS analysis is revealed that DSPC-800 exhibits an ignorable combined series resistance, which is greater than other samples. The steep slope of the Nyquist plot in the low frequency domain indicates fast ion transport in the surface defect electrode. The lack of semicircle of DSPC-800 in the middle frequency domain further suggests that the charge transfer resistance (Rct) at the electrode/electrolyte interface is negligible (**Figure 5A**). In organic electrolyte system, DSPC-800 also exhibits a small resistance, which is similar to the results in aqueous electrolyte system in **Figure 5A**. The large resistance of DSPC-600 is correlated with previous results. Bode phase diagram also discloses some important information in **Figures 5B,E**. Firstly, the 0.1 Hz phase angle is approximately the ideal phase angle, that is −90◦ , and the ideal phase angle is closed to the ideal capacitive behavior. Secondly, the characteristic frequency (f0) is corresponded to a −45◦ phase angle. The characteristic time constant can be expressed as: τ<sup>0</sup> = 1/f0. The smaller τ<sup>0</sup> represents the faster frequency response and τ<sup>0</sup> is related to the rate characteristics. Bode phase diagram is revealed that surface defect is contributed to ion diffusion (**Figures 5B,E**). A high capacity residue of 88.71% can be acquired for DSPC-800 in aqueous electrolyte system at 20 A g−<sup>1</sup> after 10,000 cycles. The capacity retentions of other devices are 82.9, 84.75, 57.64, and 72.98% in aqueous electrolyte system at 20 A g−<sup>1</sup> (**Figure 5C**), respectively. DSPC-800 is also preserved good cycle performance in organic electrolyte

system after several thousand times (**Figure 5**). Note that carbon materials with hierarchical porous structure assembled devices can deliver remarkable performance and show practical applications (**Figures 5G–I**).

## CONCLUSION

Here, defect strategy is demonstrated as an effective method to construct hierarchical porous structure with fast ion transport property. The hierarchical porous structure provides larger surface area for the infiltration of electrolyte for ions transport into interior multiple pores. Consequently, hierarchical porous structure can optimize ion kinetics by affecting ion transport of internal pores and ion diffusion of external pores from electrolyte, especially at large current density. Due to the hierarchical porous structure tuned high ionic conductivity and optimized ionic kinetics of electrodes, the power performance is impressive, delivering excellent gravimetric capacitance both in aqueous and organic electrolyte. This work provides a bright and effective approach to fabricate the carbon materials with hierarchical porous structure. Surprisingly, it has been successfully applied in energy storage under large current density. Considering the advantages of low cost and abundance, hierarchical porous carbon is extremely promising for the development of advanced electrode materials in super capacitors.

## 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

PC and GZ designed the research, performed the experiments, and made contributions to the acquisition, analysis, and interpretation of data for the work. KZ, XD, and BW carried out the partial experimental characterization. HH and XJ revised this article and approved of the version to be published. All authors incorporated in the discussion of experimental results.

#### REFERENCES


#### FUNDING

This work was financially supported by the National Key Research and Development Program of China (2018YFC1901605), the National Postdoctoral Program for Innovative Talents (BX201600192), and the Hunan Provincial Science and Technology Plan (2017TP1001), Innovation Mover Program of Central South University (GCX20190893Y).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2020.00043/full#supplementary-material


alkaline aqueous batteries. Nano-Micro Lett. 12, 1–18. doi: 10.1007/s40820-019- 0355-0

Zou, G., Hou, H., Foster, W., Banks, C. E., Guo, T., Jiang, Y., et al. (2018). Advanced hierarchical vesicular carbon Co-doped with S, P, N for high-rate sodium storage. Adv. Sci. 5:1800241. doi: 10.1002/advs.201800241

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

# Boosting Specific Energy and Power of Carbon-Ionic Liquid Supercapacitors by Engineering Carbon Pore Structures

Dong Zhang1†, Hongquan Gao<sup>1</sup> \* † , Guomin Hua1†, Haitao Zhou<sup>1</sup> , Jianchun Wu<sup>1</sup> , Bowei Zhu<sup>1</sup> , Chao Liu<sup>1</sup> , Jianhong Yang<sup>1</sup> \* and De Chen1,2 \*

*<sup>1</sup> School of Materials Science and Engineering, Jiangsu University, Zhenjiang, China, <sup>2</sup> Department of Chemical Engineering, Norwegian University of Science and Technology, Trondheim, Norway*

#### Edited by:

*Hongshuai Hou, Central South University, China*

#### Reviewed by:

*Xing Ou, Central South University, China Yirong Zhu, Hunan University of Technology, China*

#### \*Correspondence:

*Hongquan Gao hongquangao@ujs.edu.cn Jianhong Yang jhyang@ujs.edu.cn De Chen de.chen@ntnu.no*

*†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: *06 December 2019* Accepted: *07 January 2020* Published: *18 February 2020*

#### Citation:

*Zhang D, Gao H, Hua G, Zhou H, Wu J, Zhu B, Liu C, Yang J and Chen D (2020) Boosting Specific Energy and Power of Carbon-Ionic Liquid Supercapacitors by Engineering Carbon Pore Structures. Front. Chem. 8:6. doi: 10.3389/fchem.2020.00006* Carbon-ionic liquid (C-IL) supercapacitors (SCs) promise to provide high capacitance and high operating voltage, and thus high specific energy. It is still highly demanding to enhance the capacitance in order to achieve high power and energy density. We synthesized a high-pore-volume and specific-surface-area activated carbon material with a slit mesoporous structure by two-step processes of carbonization and the activation from polypyrrole. The novel slit-pore-structured carbon materials provide a specific capacity of 310 F g−<sup>1</sup> at 0.5 A g−<sup>1</sup> for C-IL SCs, which is among one of the highest recorded specific capacitances. The slit mesoporous activated carbons have a maximum ion volume utilization of 74%, which effectively enhances ion storage, and a better interaction with ions in ionic liquid electrolyte, thus providing superior capacitance. We believe that this work provides a new strategy of engineering pore structure to enhance specific capacitance and rate performance of C-IL SCs.

Keywords: hierarchical porous carbon, slit-pore structure, ion-packing density, ionic liquid, supercapacitors

### INTRODUCTION

As the portable energy storage market grows, such as the grid energy storage of electric vehicles and renewable energy, it becomes urgent to develop energy storage devices possessing high specific power and high specific energy (Armand and Tarascon, 2008; Simon and Gogotsi, 2008; Fang et al., 2018; Yun et al., 2019a; Zou et al., 2019). A supercapacitor (SC), a safe and reliable energy storage device with fast charge–discharge capability and a long cycling life, is a competitive energy storage option to meet the increasing power demands (Winter and Brodd, 2004; Aricò et al., 2005; Miller and Simon, 2008; Miller, 2012; Hou et al., 2015; Li et al., 2019; Yun et al., 2019b). However, compared with various batteries (60–200 Wh kg−<sup>1</sup> ), the wide application of commercial activated carbon–based SCs has been limited due to their low specific energy (<5 Wh kg−<sup>1</sup> ; Aurbach et al., 2000; Burke, 2000; Jang et al., 2011; Zhou et al., 2018). Therefore, without sacrificing the specific power of the SCs itself, increasing its specific energy is the ultimate method to address the problem.

The specific energy density of an SC can be calculated by the equation E = CV<sup>2</sup> /2 (Miller, 2012). Therefore, the energy density can be increased by developing a high capacitance electrode material or enlarging its voltage window. In particular, it is more remarkable to increase the specific capacitance of SCs. Therefore, the preparation of a carbon-based SC with a high ion-accessible specific surface area (SSA) will improve its capacitance greatly. However, this is not the case

**32**

when the SSA of the material is very large (>2,000 m<sup>2</sup> g −1 ). For instance, the maximum SSA of activated carbon reported in the previous literature was limited at about 4,000 m<sup>2</sup> g −1 , but the material showed a relatively low specific capacitance (165 F g−<sup>1</sup> ) in aqueous electrolyte (To et al., 2015). It can be seen that the specific capacitance is also related to the ionaccessible holes. Chmiola et al. (2006) demonstrated that anions and cations in the organic electrolyte can enter the microporous region [pore size ≈ ion size (dion) of electrolyte] to obtain the highest specific capacitance. Recently, Wei et al. have prepared a specific microporous activated carbon electrode by a chemical activation method, and this kind of activated carbon exhibits an ultra-high specific capacitance of 200–300 F g−<sup>1</sup> in an organic electrolyte or an ionic liquid (IL) (Wei et al., 2011, 2012). Extremely narrow micropores have been proven to possess optimized high capacitance. However, as with the IL electrolyte with a high voltage window used, the power characteristics of extremely narrow micropores are still limited, especially under the condition of high specific power. With such a limitation mainly due to the large ionic size of IL, the ion transport resistance in the IL electrolyte is high, which causes a rapid decrease in capacitance, thereby affecting the rate performance of the SC (Kondrat et al., 2012; Merlet et al., 2012; Peng et al., 2014; Wen et al., 2016). Chen et al. have found that the pore size region of the favorable capacitors, except for the micropore region, is the mesoporous region (pore size ≈ 2 to 3.5 times ion size of IL; Wang et al., 2016b). Compared with the micropores, the mesopores are also more favorable for rapid ion migration and improving the rate performance of the ILbased SC. Due to the model-confined favorable ion packing, the rich mesopore nanocarbon materials synthesized in this study delivered high capacitance (290 F g−<sup>1</sup> ). Chen and coworkers also proposed a model to describe the confined ion packing in the cylindrical pores and pointed to the importance of the threedimensional (3D) structure of the pores, instead of only a 2D surface, in determining the capacitance of the carbon-IL SCs. The higher mesopore volume and reasonable mesoporous pore size distribution (PSD) not only increase the specific capacitance of the material but also reduce the resistance during electrolyte ion transport, resulting in superior rate capability. However, the capacitance and pore volume relationship needs experimental validation. In addition, based on the analysis of the 3D pore– IL interaction, not only the PSD and pore volume but also the pore morphology could play a privileged role for capacitance. However, the possible effect of the pore morphology on the capacitance has not been addressed so far.

Herein, we focus on the effect of the pore properties such as pore morphology, pore volume, and PSD on the capacitance and rate capability of carbon-IL SCs. A high-pore-volume and specific-surface-area activated carbon material with a slit-pore structure was synthesized by two-step process of carbonization and activation. A series of spherical porous carbon materials was synthesized, namely, carbon nanospheres (CNSs), by using a polypyrrole (PPy) as the carbon source. The PPy after carbonization was activated by KOH at different temperatures. By changing the activation temperature, the regulation of the PSD of the activated carbon material and the increase of the mesopore volume ratio are realized. Superior performance of CNSs-IL SCs is achieved by slit-pore morphology, large pore volume, and proper PSD.

## EXPERIMENT

## Synthesis of the CNSs

A solution of 1 M hydrochloric acid (HCl) (37%) solution containing 4.04 g of pyrrole (Aladdin 99%) was mixed at a volume ratio of 1:1 with another solution of 1 M HCl containing 13.74 g of ammonia peroxydisulfate (Aladdin 98%) oxidant. After the mixture was stirred at 500 rad min−<sup>1</sup> for 2 h, the PPy was filtered and washed with water and ethanol and then dried in a vacuum overnight at 60◦C. Then, the synthesized PPy was carbonized at 650◦C with a heating rate of 10◦C min−<sup>1</sup> for 2 h in a quartz tube furnace under an argon atmosphere. The carbonized PPy (CPPy) and potassium hydroxide (KOH) pellets (Aladdin 95%) were then ground in ethanol at a mass ratio of 1:4. The mixture was then activated between 800 and 900◦C. The heating rate was 5◦C min−<sup>1</sup> , and the maximum temperature was maintained for 1 h. Therefore, the PPys activated (referred to as APPy) at 800, 850, and 900◦C were named APPy-800, APPy-850, and APPy-900, respectively. The synthesized APPy samples were washed repeatedly with 1 M HCl solution, distilled water, and ethanol until pH was 7. Finally, the samples were dried overnight at 60◦C in air.

## Material Characterization

The morphology information and microstructures of the synthesized samples were determined by field-emission scanning electron microscopy (FE-SEM, JSM-7800F) and high-resolution transmission electron microscopy (HR-TEM, Tecnai G2 F30). Brunauer–Emmett–Teller (BET) SSA and PSD of the CNSs were studied by a Micromeritic BEL, BELSORP-MAX instrument. Xray photoelectron spectroscopy (XPS, Thermo ESCALAB 250XI) was carried out for analysis of the surface chemical composition. X-ray diffraction (XRD) analysis was conducted by a Bruker AXS D8 Discover diffractometer with the Cu K<sup>α</sup> radiation (λ = 0.1540598 nm). In addition, the carbon materials were investigated by Raman spectroscopy (Renishaw inVia) with a laser wavelength of 532 nm.

## Electrochemical Characterization

The electrode materials were prepared by mixing 80 wt.% APPy, 10 wt.% acetylene black, and 10 wt.% polytetrafluoroethylene (PTFE) in a small amount of ethanol. The obtained electrode materials of the SCs were coated over a nickel foam disk (Alfa Aesar) and pressed at 6 MPa. The mass of the coated active material was approximately 2 mg cm−<sup>2</sup> . Then the electrodes were dried at 120◦C for 12 h in a vacuum oven. Finally, the twoelectrode cells were assembled using CR2025 coin cells in an argon-filled glove box (<1 ppm O<sup>2</sup> and <1 ppm H2O). The two electrodes in the coin cells were separated by a thin microporous monolayer membrane (Celgard 3501) separator and filled with 15 ml of neat 1-ethyl-3-methylimidazolium tetrafluoroborate (EMIMBF4) (Sigma-Aldrich 99%) electrolyte.

The prepared CR2025 coin cells (SCs) were stably placed overnight at room temperature before electrochemical measurements were taken. Impedance spectra (EIS) of the electrodes recorded in a frequency range from 10 to 100 kHz with a voltage amplitude of 10 mV at the open-circuit potential and cyclic voltammetry (CV) of the electrodes were carried out on a Princeton VersaSTAT potentiostat analyzer. Galvanostatic charge–discharge (GCD) tests and cycling stability tests were performed on an MTI S eight-channel battery analyzer. The specific capacitance, specific energy, and specific power of the cells were calculated from the GCD measurements by using Equations S1–S3 (as seen in the **Supplementary Material**).

#### RESULTS AND DISCUSSION

We prepared PPy-derived porous carbon by a carbonization– activation process (**Figure 1a**). This preparation of porous carbon by a two-step process has a range of physical properties depending on the carbon source and the corresponding activation process. In the present work, the microstructure and morphology information of the synthesized PPy, CPPy, and APPy-850 was characterized using FE-SEM and HR-TEM. The image shows the agglomerates composed of 3D spherical PPys (**Figure S1A**) with diameters of 100–200 nm. After pyrolysis of the PPy at 650◦C for 2 h (**Figure S1B**), the diameters of the sphere remain unchanged. The surface of the sphere becomes rough after activation at 850◦C (**Figure 1b**), and uniformly distributed mesopores can be observed on the macroporous framework based on high-magnification SEM (**Figure 1c**) and HR-TEM (**Figure 1d**) observation. The image (**Figure 1e**) clearly shows that CPPy is etched with KOH during the activation process, thus obtaining a distribution of micro- and sub-mesopores. It is also shown that the electron diffraction photograph in the inset of **Figure 1d** confirms the graphitic domain structure (circled area in **Figure 1e**; Xu et al., 2019). The micro- and sub-mesopores are mainly caused by the delamination of the graphitic domain. After activation, these domains are delaminated, and slit-shaped pores are produced. When the degree of delamination is high, cylindrical or conical pores form. A large number of slit-shaped pores formed in APPy-850 are apparent in **Figure S1C** (circled area). Simultaneously, such a hierarchical pore structure allows the IL to be better stored and transferred inside the electrode.

The N<sup>2</sup> adsorption/desorption method was used to measure the pore characteristics of the APPys obtained at different activation temperatures. Although the SSA obtained from the BET method is not rigorous (Rouquerol et al., 2007), herein, the BET method is used to evaluate the relative surface area of different samples. The results indicate that the activation temperature has a significant effect on the SSA, the pore structure, and PSD (**Figures 2A,B**). The SSA of APPy-800, APPy-850, and APPy-900 are 2207.1, 3818.8, and 2620.5 m<sup>2</sup> g −1 , respectively.

Regarding the specific pore structure in the APPys, the hysteresis loop shapes have often been identified with specific mesopore structures. The hysteresis loop made by slit pores seems nearly horizontal and parallel over a wide range of p/po. And the hysteresis loop made by cylindrical pores is almost

vertical and nearly parallel over an appreciable range of gas uptake (Sing, 1985; Thommes et al., 2000, 2002). Illustrated hysteresis loops corresponding to cylindrical pores and slit pores are presented as an insert in **Figure 2A**. As presented in **Figure 2A**, the hysteresis loop between the N<sup>2</sup> adsorption and desorption isotherms indicates that the porous carbon materials mainly contain slit mesopores, rather than cylindrical pores. Furthermore, the mesopore content in the porous carbon materials can be evaluated by the size of the hysteresis loop (Sing, 1982). In general, a larger hysteresis loop indicates a high proportion of mesopores, and a small hysteresis loop implies a low proportion of mesopores. By comparison of the hysteresis loops of three porous carbon materials, APPy-800, APPy-850, APPy-900, it can observed that APPy-850 has the highest content of mesopores.

According to the non-local density functional theory (NL-DFT) based on the slit-pore model, the pore structure parameters of APPys are listed in **Table S1**. It can be observed that the pore volumes of APPy-800, APPy-850, and APPy-900 are 0.987, 2.098, and 1.487 cm<sup>3</sup> g −1 , respectively. The PSD of APPys are

presented in **Figure 2B**; it can be seen that pore volumes of APPy-850 and APPy-900 increase in the range of mesopore size and decrease in the range of micropore size in comparison to APPy-800. The narrow small mesopore size ranges from 2 to 5 nm. The ratios of mesopore volume to total pore volume in APPy-850 and APPy-900 samples are 72 and 68%, respectively. In addition, it can be observed that the activation process at high temperature (900◦C) decreases SSA and pore volume, which is detrimental to achieving high-performing porous carbon materials. Such degradation may be due to excessive temperature, which makes the ablation of C atoms in the KOH activation process more serious, leading to collapse of the pores (Xu et al., 2014).

The chemical compositions of APPy C1s, O1s, and N1s signals are analyzed by the XPS spectra, and the surface composition of carbons is shown in the XPS survey spectra (**Figure 3A** and **Table S2**). From the XPS survey, the samples of APPy-850 are mainly made up of 90.99 at.% carbon with 8.33 at.% oxygen and 0.68 at.% nitrogen, with no other heteroatoms being observable. The high-resolution C1s spectra of APPy (**Figure S2**) can be deconvoluted by five peaks, representing C-I (284.6 eV), a dominated component for graphitized carbon; C-II (286.0– 286.3 eV), carbon in phenolic, alcohol, ether, or C=N groups; C-III (287.3–287.6 eV), carbon in carbonyl or quinine groups; C-IV (288.8–289.1 eV), carbon in carboxyl or ester groups; and C-V (290.5–291.2 eV), carbon in carbonate groups and/or adsorbed CO and CO<sup>2</sup> (Zhou et al., 2007, 2019). The highresolution O1s core level spectra can be mainly deconvoluted by four peaks, representing O-I (531.4 eV), C=O quinonetype groups, O-II (532.5 eV), C-OH phenol groups and/or C-O-C ether groups, O-III (533.4 eV), non-carbonyl (ether-type) oxygen atoms in esters, and O-IV (534–535 eV), anhydrides groups and COOH carboxylic groups (**Figure S2**; Li et al., 2014). It is widely accepted that the surface redox reaction between oxygen functionalities on carbons and electrolyte ions can contribute to the pseudocapacitance through the quinone/hydroquinone redox pair (about 60% of total oxygen; Liu et al., 2016). It is confirmed that in non-aqueous electrolytes, the reversible oxidation/reduction of hydroquinone/quinone and carbonyl (C=O) groups may cause pseudocapacitance (Gupta and Linschitz, 1997). So a small peak of pseudocapacitance at ∼2 V can be observed on the discharge section of the CV curve of APPy-850-SC (**Figures 4B,E**). The nitrogen content in the three APPy samples is relatively low (<0.82 at.%). These traces of N atoms may be doped into carbon materials to enhance the wettability of the interface between the electrode and electrolyte (Li et al., 2007).

samples; the inset is a magnification of the high-frequency range data and fitting equivalent circuit. (D) Galvanostatic charge–discharge (GCD) curve of APPy-850-SC at current densities of 1, 2, 4, 6, 8, and 10 A g−<sup>1</sup> . (E) CV curve of APPy-850-SC at a scan rate of 20–200 mV s−<sup>1</sup> . (F) Cycling stability and coulombic efficiency of APPy-850-SC at 6 A g <sup>−</sup><sup>1</sup> for 10,000 cycles. The inset shows photographs of two blue LEDs.

The structural features of the porous carbon were further evaluated by using XRD. The XRD pattern (**Figure 3B**) shows two discernible broad peaks at 25–30◦ , which corresponds to the (002) plane, and 42–44◦ , corresponding to the (100) plane of graphite, respectively (Jiang et al., 2013). The broad (002) XRD peaks confirm the low degree of order in these carbons, which is understandable since they are both carbonized and activated at low temperatures. A large increase in the lowangle scatter from the porous carbon can be also noted, which is consistent with the presence of a high density of pores. These results are consistent with the observations from HR-TEM (**Figure 1e**), which indicate that APPy porous carbon consists predominantly of curved monoatomic layers of carbon, forming a large number of slit-shaped pores between the interlaced flakes (arrows in **Figure S1C**). The two characteristic G- and Dbands of the Raman spectrum of APPy materials are shown at ∼1,595 and ∼1,350 cm−<sup>1</sup> , respectively (**Figure 3C**; Tsai et al., 2013). The former corresponds to the graphitic order, while the latter corresponds to the degree of disorder/defectiveness in the structure. As the activation temperature increases, the ID/I<sup>G</sup> ratio is gradually increased from ∼1.11 in APPy-800, ∼1.23 in APPy-850 to ∼1.41 in APPy-900, indicating an increase of the degree of disorder/defectiveness in the structures. According to XRD diffraction and the Raman spectrum, it can be confirmed that the obtained porous carbon materials consist of stacking graphenelike layers. Intriguingly, as the activation temperature increases, the ID/I<sup>G</sup> ratio is gradually increased from ∼1.11 in APPy-800, ∼1.23 in APPy-850 to ∼1.41 in APPy-900, indicating an increase of the disorder degree in the structures. This is a little controversy to the general observation. Such a discrepancy can be attributed to the activation temperature. When the activation temperature is below 900◦C, the increase of activation temperature will lead to poor graphitization degree (Yan et al., 2020).

The GCD measurement results are in the range of 0.5–10 A g −1 , which confirmed the superior capacitance performance of APPy, as shown in **Figure 4A**. The APPy-850 shows a high gravimetric capacitance of 310 F g−<sup>1</sup> (0–4 V) at 0.5 A g−<sup>1</sup> and high capacitance retention of 83% up to 10 A g−<sup>1</sup> . The excellent performance of this sample is attributed to the large specific surface area and high pore volume, contributed mainly by the rich mesoporous structure. This structure facilitates the rapid transport of ions and thus contributes to high rate performance (Guo et al., 2013). The CV curves of the three APPy samples at the same scan rates (100 mV s−<sup>1</sup> ) are shown in the fully operated voltage ranging from 0 to 4 V (**Figure 4B**). Among them, the APPy-850 shows a larger CV area, indicating that it possesses a higher specific capacitance. The Nyquist plots to investigate the electrochemical behavior of the three APPy samples are shown in **Figure 4C**, and the inset is an EIS plot at high-frequency region and equivalent circuit. The impedance spectrum mainly consists of a semicircle in the high-frequency region and an approximate straight line in the low-frequency region, where the behavior is mainly capacitive. The 45◦ segment in the Nyquist plot may be due to the inhomogeneity of the internal pores, indicating that the transport of electrolyte ions during charge and discharge is controlled by the diffusion process (Manohar et al., 2008; He and Mansfeld, 2009). The semicircle reflects the resistance during charge transfer and is related to the porous structure of the electrode (Gamby et al., 2001). From the inset of **Figure 4C**, the amplitude of the semicircle of the APPy-850 is much smaller than that of the others, indicating that the APPy-850 shows the capacitive behavior at lower resistance values than others. That means the impedance of the electrode with the APPy-850 is less dependent on frequency, and thus, the pore ion accessibility of the APPy-850 is higher than that of the others, indicating that the activation temperature of 850◦C provides the best porosity structure for IL ion diffusion.

For the optimized APPy-850, the charge/discharge curve (**Figure 4D** and **Figure S3**) still shows a quasi-triangle, indicating excellent electrochemical reversibility and charge/discharge efficiency (Stoller et al., 2008). As shown in **Figure 4E** and **Figure S3**, the CV exhibits an approximately rectangular shape, confirming that the capacitance obtained is mainly due to the electric double layer capacitance contribution. Even at a high scanning rate of 200 mV s−<sup>1</sup> , its rectangular shape is maintained, indicating excellent current response capability. This good current performance is due to the structure in which the mesopores and the micropores are cross-linked with each other can store a large amount of electrolyte ions and effectively shorten the ion transport time (Chen et al., 2012; Wang et al., 2016a). Apart from high capacitive performance, long-term cyclic performance is another important factor for a superior energy storage device. The cells were investigated by GCD at a current density of 6 A g−<sup>1</sup> within the voltage range of 1.8–3.6 V (**Figure 4F**). After 10,000 cycles, it still maintains about 81% of the initial capacitance, indicating that APPy-850 porous carbon is a well-stabilized SC electrode material (Zhang et al., 2015). Since there is still sufficient free space between the mesoporous structures of these porous carbons, the volume expansion due to prolonged ion insertion/extraction is effectively buffered. Two devices also were connected to light emitting diodes (LEDs, 3.6 V), and the two blue LEDs were successfully lighted up (the inset of **Figure 4F**) for over 10 min after charging for 15 s.

According to the specific capacitance data measured by GCD in **Table S1**, **Figure 5A** reveals that the increase in the effective slit-pore volume of APPys between the slit-pore diameters of 1 and 10 nm can lead to the increase in the specific capacity. The specific capacitance increases linearly from 193 to 310 F g −1 , as the slit-pore volume increases from 0.61 to 1.727 cm<sup>3</sup> g −1 . This result shows that the specific capacity is novel linearly proportional to the slit-pore volume. Regarding the relationship between pore volume and specific capacity, a cylindrical pore model has recently been used to interpret such a relationship in terms of Equation 1 and Equation S4 (Wang et al., 2016a)

$$N\_{total} = \sum\_{D\_1}^{D\_l} [V(D\_i) \cdot f(\phi(D\_i/d\_{ion}), d\_{ion})] (i = 1, 2, 3 \dots n) \tag{1}$$

where Ntotal is the number of adsorbed ions, V is the specific pore volume, f is the ion-packing function, Φ is a function of ion volume utilization fraction (Φ = the volume of the adsorbed ions/the total volume of the pore), dion is effective ion size (0.97 nm), and D is the pore size. Due to the ion volume utilization fraction, Φ is a function of D/dion in the cylindrical pore model, so the increase in specific capacity is not a perfect linear dependence on the increase in cylindrical pore volume. However, the novel slit-pore structure carbon

FIGURE 5 | (A) Relationship between pore volume and specific capacitance, (B) Pore size distribution curve calculated using different models and charge ion-packing configurations. Note: CNS1 stands for carbon nanosponge (CNS), which carbonized at 650◦C, followed by activation at 900◦C for 60 min. The data of CNS1 are cited from Wang et al. (2016b).

Zhang et al. Carbon-Ionic Liquid Supercapacitors

materials in this work give rise to a linear dependence of specific capacity on slit-pore volume; it implies that the ion volume utilization fraction Φ is a constant rather than a function of D/dion. Such a difference in the ion volume utilization fraction Φ indicates that the ion-packing configuration in the novel slit-pore structure is different from that in cylindrical pore structure.

In the following, the difference of ion-packing configuration between a cylindrical pore structure and a slit-pore structure will be investigated. **Figure 5B** shows PSD curves of the two materials. The PSD curves of CNS1 are calculated using a cylindrical pore DFT model with the pore volume of 1.5 cm<sup>3</sup> g <sup>−</sup><sup>1</sup> between the pore diameters of 1 and 10 nm. The PSD curves of APPy-850-30 min are calculated using a slit-pore DFT model with the pore volume of 1.56 cm<sup>3</sup> g <sup>−</sup><sup>1</sup> between the pore diameters of 1 and 10 nm. The PSD in the region of 1–5 nm mainly contributing to the pore volume is considered in this work. When the pore size is (D/dion < 1), no ions can access the surface inside the pore. When the pore size can exactly adapt the integer number of ions, the densest-surface ion-packing configurations lead to the highest volume utilization. When the pore size within the range accommodates the two nearest integer number of ions, the volume utilization decreases with the increase of D/dion, as demonstrated by the cylindrical pore model (Hodak and Girifalco, 2003; Mughal et al., 2011, 2012). An example is illustrated as schemes A to B in **Figure 5B**. In the case of D/dion equals 1 and 2, the ion volume utilization fractions are 21.2 and 15%, respectively (calculated by φ of Equation 1). It is also interesting to note that the volume utilization fractions of other integer number of ions (3, 4, 5, 6, 7, 8, and 9) are not higher than the case where the ion-packing number is 1 in the range of 1–5 nm. The densest-surface ion-packing configuration of each layer in the surface of the slit pore has only a hexagonal close-packed (HCP) mode in scheme C of **Figure 5B**, and the layer of IL ions having an HCP mode will stack up layer by layer as the slit-pore size increases. The HCP structure of ions in the slit pore has the highest volume utilization (74%) (Szczurek et al., 2014). It is evident that a slit-pore structure is superior to the conventional cylindrical pore structure, and it can accommodate more electrolyte ions at an identical pore volume. Therefore, in this work, the slit-pore structures of the APPy materials increase the specific capacitance of the material in the same volume.

To further confirm the superior performance of our SC with an APPy-850–based electrode, the Ragone plot for APPy-850 material is compared with other electrode materials in **Figure 6**. Based on the active mass in the electrode, the APPy-850-SC possessed an excellent specific energy of 171.5 Wh kg−<sup>1</sup> at the specific power of 664 W kg−<sup>1</sup> . Even at a much higher specific power of 15 kW kg−<sup>1</sup> , the APPy-850-SC still maintains a high specific energy of 129.1 Wh kg−<sup>1</sup> . In **Table S3**, the performance of the current system is compared with other SCs based on the same IL electrolyte. The APPy-850-SC exhibits significantly higher specific capacity and remains quite competitive with other performances.

## CONCLUSIONS

Here we have shown a privilege role of the ion-packing density in determining the performance of carbon-IL SCs. We have shown that the specific capacity of the porous material is linearly proportional to the effective pore volume and is also closely related to the pore structure. We report that the 2D slit-pore structure with the HCP configuration of ions has the highest ion-packing density. The slit-pore structure has superior ion-packing density compared to cylinder-like pores. As a result, optimized synthetic APPy-850 hierarchical porous carbon with a high pore volume and a rich slit mesoporous structure achieved a specific capacity of 310 F g−<sup>1</sup> at 0.5 A g −1 for IL SCs, which is one of the highest recorded specific capacitances. At the same time, high specific energy and excellent cycling performance at high specific power are also shown. This work provides a new idea to increase the specific capacity and rate performance of carbon materials by increasing the volume fraction of the slit-type mesopores, which makes it potentially possible to develop SCs with higher specific energy and specific power.

## DATA AVAILABILITY STATEMENT

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

## AUTHOR CONTRIBUTIONS

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

## FUNDING

We acknowledge funding from the National Natural Science Foundation of China (grants 51774151 and 51702131) and Natural Science Foundation of Jiangsu Provence, China (grant SBK2017041705). This work was also supported by the Institute of Green Materials and Metallurgy (grant 5501670001) and Jiangsu University Foundation (grant 4111220019).

#### REFERENCES


#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2020.00006/full#supplementary-material


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

# Freestanding Needle Flower Structure CuCo2S<sup>4</sup> on Carbon Cloth for Flexible High Energy Supercapacitors With the Gel Electrolyte

Tian Xie, Jinxiao Xu, Jie Wang, Chuanli Ma, Linghao Su, Fengying Dong and Liangyu Gong\*

College of Chemistry and Pharmaceutical Sciences, Qingdao Agricultural University, Qingdao, China

A facile hydrothermal approach was adopted to the direct synthesis of bimetallic sulfide (CuCo2S4) on carbon cloth (CC) without binders for the supercapacitor's electrodes. A possible formation mechanism was proposed. The prepared bimetallic electrode exhibited a high specific capacitance (Csp) of 1,312 F·g <sup>−</sup><sup>1</sup> at 1 A·g −1 , and an excellent capacitance retention of 94% at 5 A·g <sup>−</sup><sup>1</sup> over 5,000 cycles. In addition, the asymmetric supercapacitor (CuCo2S4/CC//AC/CC) exhibited energy density (42.9 wh·kg−<sup>1</sup> at 0.8 kW·kg−<sup>1</sup> ) and outstanding cycle performance (80% initial capacity retention after 5,000 cycles at 10 A·g −1 ). It should be noted that the electrochemical performance of a supercapacitor device is quite stable at different bending angles. Two charged devices in series can light 28 red-colored LEDs (2.0 V) for 5 min. All of this serves to indicate the potentially high application value of CuCo2S4.

Keywords: flexible electrode, nanoneedles, CuCo2S4, pseudocapacitance, supercapacitors

## INTRODUCTION

Supercapacitors are extremely valuable for energy storage, owing to their rapid rechargeable ability and extended life cycle (Gao et al., 2015b; He and Chen, 2015). In recent years, they have been widely applied in many fields, such as power backup devices, hybrid electric vehicles, and portable and wearable electronic products (Dong et al., 2015). While flexible electronic products have experienced explosive growth, all electronic devices require flexible and long-life energy storage systems (Hu et al., 2010; Shen et al., 2014; Liu G. et al., 2018). Currently, the main commercial supercapacitors use carbon materials as electrode materials and thus suffer from low-energy density (Dong et al., 2016). Therefore, designing advanced nanostructures with high-energy density and outstanding performance at high-current densities is still given a large amount of attention and holds great appeal (Bao and Li, 2012; Wang et al., 2014; Huang et al., 2016; Xing et al., 2018).

Transition metal compounds, including oxides and sulfides, can provide superior specific capacitance (Csp) due to the rapid faraday redox reaction (Guo et al., 2014; Padmanathan and Selladurai, 2014; Pu et al., 2014; Gao et al., 2015b; Moosavifard et al., 2016). However, the maximum capacitance and energy density of these materials are severely limited by the lower electrical conductivity. Among them, transitional bimetallic sulfides can provide a richer redox reaction,

Edited by:

Tang Wei, Xi'an Jiaotong University, China

#### Reviewed by:

Bingbing Tian, Shenzhen University, China Chengxin Peng, University of Shanghai for Science and Technology, China

#### \*Correspondence:

Liangyu Gong lygong@163.com

#### Specialty section:

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

Received: 19 December 2019 Accepted: 20 January 2020 Published: 27 February 2020

#### Citation:

Xie T, Xu J, Wang J, Ma C, Su L, Dong F and Gong L (2020) Freestanding Needle Flower Structure CuCo2S<sup>4</sup> on Carbon Cloth for Flexible High Energy Supercapacitors With the Gel Electrolyte. Front. Chem. 8:62. doi: 10.3389/fchem.2020.00062

**41**

resulting in a generally superior electrochemical performance (Xiong et al., 2015; Sun et al., 2017; Wang J. et al., 2018; You et al., 2018; Dong et al., 2019). For example, Li et al. supported NiCo2S<sup>4</sup> nanorods on Ni foam, which displayed a 94.96% capacitive retention over 10,000 cycles (Li M. L. et al., 2018; Li X.-X et al., 2018). At the same time, it has been found that transition bimetallic sulfides also possess rich redox couples and enhance the electronic conductivity (Ai et al., 2016; Wang et al., 2017). In recent years, CuCo2S<sup>4</sup> as electrode material has been found to have excellent electronic conductivity, good electrochemical stability, and higher conductivity when compared to its oxide counterparts. In addition, the synergistic effects due to the interaction between diverse metal compounds leads to an enhanced electrochemical energy storage performance, as compared with monometallic sulfides (Shinde et al., 2019). For example, Tian et al. anchored CuCo2S<sup>4</sup> to graphene aerogel through solvent thermal reaction with 668 F·g −1 at 1A·g −1 . Asymmetrical supercapacitors based on CuCo2S4/GA exhibit a maximum of 22 Wh·kg−<sup>1</sup> (Tian et al., 2018). Guo et al. prepared Co2CuS<sup>4</sup> on graphene nanosheets, which exhibited Csp of 1,005 F·g −1 at 1A·g −1 (Guo et al., 2016). In addition, designing nanostructures with large surface areas is important for promoting ion transport in electrochemical processes (Wang et al., 2016; Shang et al., 2018).

In order to adapt to the trend of wearable and flexible supercapacitors, others found it desirable to prepare electrode materials on flexible substrates (Dingshan et al., 2015; He et al., 2018). According to previous reports, carbon fiber, nickel foam, and copper foam are widely applied as conductive substrate (Martti et al., 2009; Yang et al., 2014; Xiao et al., 2015; Sun et al., 2016). Among them, the carbon fiber cloth is considered to be an ideal conductive substrate due to its flexibility, electrical conductivity, and low cost. At the same time, directly growing nanostructured materials on carbon cloth avoids the use of binders, resulting in rapid electro transport (Jiang et al., 2011; Liu et al., 2011; Yuan et al., 2012; Zhang et al., 2012, 2016; Wang et al., 2013).

Here, we confirm that carbon cloth is a conductive substrate that can be used to grow a novel 3D needle-flowerlike CuCo2S<sup>4</sup> nanostructure. As a result, the flowerlike CuCo2S<sup>4</sup> nanostructure exhibits excellent performance (1,312 F·g −1 at 1 A·g −1 ). Beyond that, an asymmetric solid-state device was manufactured. As a result, it's provided a significant energy density of 42.9 Wh·kg−<sup>1</sup> with an initial capacitance retention rate of 80%–even after 5,000 cycles in the 1.6 V-wide working potential window, showing excellent cycling stability.

### MATERIALS AND METHODS

## Preparation of CuCo2S4/CC

All reagents were analytical-grade and used without any additional purification. The carbon cloth was purchased from Ce Tech WOS1009. First, a piece of carbon cloth was refluxed at 100◦C for 1 h with HNO<sup>3</sup> and subsequently rinsed with deionized water and ethanol. As is typical, CuCl2·6H2O (2 mmol), CoCl2·2H2O (4 mmol), urea (30 mmol), and NH4F (20 mmol) were dissolved into 60 mL DI water and stirred for 10 min. Then, the mixture solution and the treated CC were transferred into 100 mL autoclave. After the hydrothermal reaction (140◦C for 6 h), the precursors on carbon cloth were washed with deionized water and ethanol, and then dried at 60◦C for 12 h.

Next, thioacetamide (TAA, 5 mmol) was dissolved in 60 mL of absolute ethanol with stirring. Then, the solution and the as-obtained precursor were transferred together to the 100 ml autoclave (140◦C for 6 h). Finally, after being washed and dried at 60◦C overnight, the carbon cloth of uniform load CuCo2S<sup>4</sup> was obtained. The loading mass of CuCo2S<sup>4</sup> was about 2.3 mg·cm−<sup>2</sup> . For comparison, CuCo2S<sup>4</sup> was directly obtained without the placement of a carbon cloth and the CC without the loaded sample is named blank-CC.

#### Characterization

X-ray diffraction (XRD) patterns were recorded on a Bruker D8. The morphology of samples was studied by scanning electron microscopy (SEM) (JEOL-7500F, Japan) and transmission electron microscopy (TEM) (HITACHI-7650, Japan). Cycling voltammetry (CV), galvanostatic charge-discharge (GCD), and electrochemical impedance spectroscopy (EIS) were performed with a CHI760E (Shanghai CH, China) electrochemical workstation. Additionally, the cycle life was tested on a CT2001A LAND Battery Test System (LAND, Wuhan, China).

#### Electrochemical Measurements

The electrochemical measurements were carried out in a threeelectrode system by using 3 M KOH aqueous solution as the electrolyte, where the CuCo2S4/CC was directly used as the working electrode, Pt foil was used as the counter electrode, and a saturated calomel electrode (SCE.) was used as the reference electrode. The Csp of materials was calculated through the equation Csp = I × 1t / m × 1V, where I, ρt, m, and V are the discharge current (A), discharge time (s), loading mass (g), and voltage range.

The flexible solid-state asymmetric supercapacitor (FSASC) device was assembled with a CuCo2S4/CC cathode, active carbon (AC)/CC anode, and PVA/KOH gel electrolyte. The mass ratio between CuCo2S4/CC and AC/CC was m+/m<sup>−</sup> = 0.19, according to charge conservation (Yuan et al., 2018). The AC/CC electrode was produced by mixing AC 80 wt%, acetylene black 10 wt%, and PTFE 10 wt% and then coated onto carbon cloth. The PVA/KOH gel electrolyte was maintained by mixing KOH (3 g) and PVA (3 g) in 30 ml DI water at 80◦C for 1 h with stirring. The energy and power densities were measured by using equation E (Wh·kg −1 ) = Csp × 1V 2 / 2 and P (W·kg−<sup>1</sup> ) = E / 1t.

## RESULTS AND DISCUSSION

As illustrated in **Scheme 1**, the CuCo2S4/CC was successfully synthesized through a simple two-step process. First, the Cu-Co precursor was directly grown on the CC via a hydrothermal process. Subsequently, the Cu-Co precursor is converted into CuCo2S4/CC via a sulfurization strategy.

The chemical composition is recorded by XRD (**Figure 1**). The different peaks of 16, 26, 31.2, 37.9, 46.9, 49.9, and 54.8◦ are identical to the standard diffraction data of CuCo2S<sup>4</sup> (JCPDS No.42-1450) (You et al., 2018). And the broad peaks at 25 and 45 ◦ are derived from the CC substrate (Liu and Wu, 2018).

**Figures 2A–H** shows the topographical changes of CuCo2S<sup>4</sup> on carbon cloth. In **Figure 2A**, the CuCo2S<sup>4</sup> was uniformly adhered to the carbon fiber at a reaction time of 2 h. As seen in **Figure 2E**, the flowerlike structure, composed of acicular CuCo2S4, is supported on the carbon fiber. When the reaction time was extended to 4 h (**Figures 2B,F**), the acicular structure of CuCo2S<sup>4</sup> began to aggregate. When the reaction time was 6 h (**Figures 2C,G**), the acicular structure was further aggregated to form a flowerlike structure. The acicular structure provides more active sites, providing abundant pores, while the surface that is initially aggregated by the needle structure provides a larger reaction area. When the reaction time was extended to 8 h, the flowerlike structure supported on the carbon fiber is composed of agglomerated sheets, which reduces the reactive sites and electrochemical performance (**Figures 2D,H**). The morphological structure of the CoCu-pre and CuCo2S<sup>4</sup> is shown in **Figure S1**. In **Figure S1A**, the needle CuCo-pre constitutes a flowerlike structure and uniformly grows on the carbon fibers. **Figure S1B** clearly shows the flowerlike CuCo2S<sup>4</sup> structure on the carbon fiber; the exchange of anions during vulcanization does not destroy the overall morphology. In contrast, the CuCopre and CuCo2S<sup>4</sup> were obtained without adding a carbon cloth substrate (**Figures S1C,D**). It can be seen that both exhibit similar morphologies as those loaded on carbon cloth.

In addition, the microstructure of the CuCo2S4/CC shown by TEM (**Figure 3A**) further demonstrates that the acicular structure of CuCo2S<sup>4</sup> is attached to the carbon fibers. A HRTEM image shows clear lattice fringes with lattice spacing corresponding to the (113) and (004) planes (**Figure 3B**). The SAED in inset indicates the polycrystalline of the materials. Meanwhile, the TEM and SAED image of CuCo2S<sup>4</sup> shows that it is composed of many primary nano-needles and its polycrystalline (**Figures S1E,F**).

Through the above observations, the growth mechanism of flowerlike CuCo2S<sup>4</sup> on carbon cloth can be preliminarily obtained. First, in an alkaline environment, Cu2<sup>+</sup> and Co2<sup>+</sup> ions form a CuCo(OH)<sup>x</sup> precursor (Wan et al., 2015; Jayaraman et al., 2017). At this time, the solution becomes a supersaturated solution with a large amount of Cu-Co precursor particles. Due to the high surface energy and thermodynamic instability, the nanoparticles adhere to the carbon fibers to reduce the surface energy. Based on the hydrolysis rate control of NH4F, the precursors will be oriented to grow into the acicular structure and further assemble into a flowerlike structure (Liu S. et al., 2018). The second step is the vulcanization process. The S2<sup>−</sup> released by TAA undergoes anion exchange with OH<sup>−</sup> to form a CuCo2S<sup>4</sup> flowerlike structure.

The electrochemical properties of CuCo2S4/CC were measured. **Figure 4A** shows the CV curve of CuCo2S4/CC, CuCo2S<sup>4</sup> and blank-CC at 10 mV·s −1 . The curve area of CuCo2S4/CC is larger than that of CuCo2S<sup>4</sup> and blank-CC, which means that CuCo2S4/CC has a higher specific capacitance.

**Figure 4B** is a CuCo2S4/CC and CuCo2S<sup>4</sup> GCD curves at 1 A·g −1 . Similarly, the discharge time of the CuCo2S4/CC material is much higher than that of CuCo2S4. And, the capacitance contribution of blank-CC is negligible. **Figure 4C** shows typical CV curves for CuCo2S4/CC at 0–0.6 V. Each curve shows a significant redox peak, indicating that the Cu(Co)-S-O/Cu(Co)- S-OH-related faraday reaction was carried out (Jayaraman et al., 2017). As the scan rate increased, the redox peak appeared to shift due to the presence of polarization (Ratha et al., 2017). The possible electrochemical reaction equations are as follows (Kaverlavani et al., 2017; Zhu et al., 2017):

CuCo2S<sup>4</sup> <sup>+</sup> OH<sup>−</sup> <sup>+</sup> <sup>H</sup>2O↔CuSOH <sup>+</sup> 2CoSOH <sup>+</sup> <sup>e</sup> <sup>−</sup> (1)

$$\text{CuSOH} + \text{OH}^- \leftrightarrow \text{CuSO} + \text{H}\_2\text{O} + \text{e}^- \qquad \text{(2)}$$

$$\text{CoSOH} + \text{OH}^- \leftrightarrow \text{CoSO} + \text{H2O} + \text{e}^- \qquad \text{(3)}$$

In order to accurately compare the specific capacitances, the exact Csp was calculated from GCD measures. **Figure 4D** shows the GCD curves of CuCo2S4/CC from 1 to 10 A·g −1 . From the comparison of Csp graphs of CuCo2S4/CC and CuCo2S<sup>4</sup> (**Figure 4E**), the discharge-specific capacitance of CuCo2S4/CC is 1,312, 1,228, 1,116, 1,010, 864, and 820 F·g −1 at 1, 2, 3, 5, 8, 10 A·g −1 , which is much larger than CuCo2S4. The specific capacity values were compatible relative to those previously reported (**Table S1**). On the other hand, lower charge transfer resistance (Rct) and Warburg impedance (W) also means that CuCo2S4/CC can show superior electrochemical performance compared to CuCo2S<sup>4</sup> (**Figure S2**). It is noteworthy that the capacities decrease with an increase in current density, which may be due to the fact that the transport and/or diffusion of electrolyte ions are suppressed in high current density (Brousse et al., 2015). The CuCo2S4/CC electrode maintained an initial capacity of 94% after cycling 5,000 times at 5 A·g −1 , while the Coulomb efficiency remained almost 100%, indicating excellent cycling stability (**Figure 4F**).

To prove the application potential, the FSASC device was made of CuCo2S4/CC and AC/CC as the positive and negative electrodes, respectively. **Figure 5A** is the CV curves of CuCo2S4/CC and AC/CC at 20 mV·s −1 . The detailed electrochemical data of the AC is shown in **Figure S3**. The electrochemical properties of the assembled FSASC were evaluated. The CV and GCD measurements of the ASC device are shown in **Figures 5B,C**, respectively. At the same time, the charging time of the device is higher than the

FIGURE 4 | CV curves of CuCo2S4/CC, CuCo2S4, and blank-CC at 10 mV·s −1 (A); GCD curves of CuCo2S4/CC, CuCo2S4, and blank-CC at 1A·g −1 (B); CV curves of CuCo2S4/CC at 5–50 mV·s −1 (C); GCD curves of CuCo2S4/CC at 1–10A·g −1 (D); the rate capability of the as-prepared two electrodes (E); Cycling performance at 5A·g −1 (F).

discharge time in GCD curves, which may be due to the fact that transport of the electrolyte ion cannot be fully carried out at a low current density (**Figure 5C**). This phenomenon stems from the nature of the gel electrolyte and the inherent properties of the material (Qi et al., 2015; Wang P. et al., 2018). When the current density is 10 A·g −1 , a high Csp

of 106.3 F·g −1 can be delivered, which shows a retention of 88.1% (**Figure 5D**).

The cycle properties of the ASC device were evaluated by cycling 5,000 times at 10 A·g −1 (**Figure 6A**). The Csp decreases from 110 to 87.5 F·g −1 , which exhibits an excellent retention of 80%. Meanwhile, the coulombic efficiency is almost maintained at 100%, which indicates that the prepared device has excellent performance. **Figure 6B** shows a maximum energy density of 42.9 Wh·kg−<sup>1</sup> at 0.8 kW·kg−<sup>1</sup> . And it's still has 37.8 Wh·kg−<sup>1</sup> at 8 kW·kg−<sup>1</sup> . In addition, the assembled FSASC device exhibits outstanding flexibility—the CV curves at different angles proving that the bending has little effect on the device (**Figures 6C,D**).

To demonstrate the practical application potential of the equipment, 28 red LEDs assembled in parallel can be illuminated for more than 5 min (**Figure 7**), which indicates that CuCo2S<sup>4</sup> possess excellent energy storage potential.

#### CONCLUSIONS

In short, we directly grew needle flower structure CuCo2S<sup>4</sup> on carbon cloth by means of a hydrothermal reaction and vulcanization process; the prepared CuCo2S4/CC shows an excellent electrochemical performance (1,312 F·g −1 at 1 A·g −1 and 94% retention over 5,000 cycles). When used as an electrode material for flexible asymmetric supercapacitors, the device provides higher energy density (maximum of 42.9 Wh·kg−<sup>1</sup> ) and power density (maximum of 8 kW·kg−<sup>1</sup> ), long-term cycling stability (80% retention over 5,000 cycles), and superior flexibility. The needle flower structure promotes the transport of electrolyte ions, providing a higher specific capacitance. The rapid transport of ions/electrons on carbon fibers imparts high-rate characteristics to flexible solid-state supercapacitors. This indicates that CuCo2S4/CC is promising as a flexible supercapacitor electrode material.

#### REFERENCES


#### DATA AVAILABILITY STATEMENT

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

#### AUTHOR CONTRIBUTIONS

The manuscript was written through contributions of all authors. All authors have given their approval of the final version of the manuscript.

#### FUNDING

This work was financially supported by the Natural Science Foundation of Shandong Province (ZR2017LB031).

#### ACKNOWLEDGMENTS

The authors thank the Analytical and Testing Center of Qingdao Agricultural University for allowing them to use its facilities.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2020.00062/full#supplementary-material


for high-performance asymmetric supercapacitors. J. Mater. Chem. A 5, 14301–14309. doi: 10.1039/C7TA03943C


lithium-oxygen batteries. Chem. Commun. 53, 8711–8714. doi: 10.1039/C7CC 02621H


substrates as binder-free electrodes for high-performance supercapacitors. Energy Environ. Sci. 5:9453. doi: 10.1039/c2ee22572g


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

# Ion Liquid Modified GO Filler to Improve the Performance of Polymer Electrolytes for Li Metal Batteries

Zhongliang Hu\*, Xiaojing Zhang, Jilei Liu and Yirong Zhu\*

*Department of Inorganic Nonmetallic Material, College of Metallurgy and Material Engineering, Hunan University of Technology, Zhuzhou, China*

Polymer electrolytes for Li metal batteries (LMBs) should be modified to improve their ionic conductivity and stability against the lithium electrode. In this study, graphene oxide (GO) was modified by ion liquid (IL), and the IL modified GO (GO-IL) had been used as a filler for polyethylene oxide (PEO). The obtained solid polymer electrolyte (SPE) is of high ionic conductivity, low crystallinity and excellent stability against the lithium electrode. The PEO/GO-IL was characterized by various techniques, and its structure and performance were analyzed in detail. By addition of 1% GO-IL, the ionic conductivity of the PEO/GO-IL SPE reaches 1.8 × 10−<sup>5</sup> S cm−<sup>1</sup> at 25◦C, which is 10 times higher than PEO (1.7 × 10−<sup>6</sup> S cm−<sup>1</sup> ), and the current density for stable Li plating/stripping in PEO/GO-IL can be increased to 100µA cm−<sup>2</sup> at 60◦C. LiFePO4/Li cell (using PEO/GO-IL SPE) tests indicated that the initial discharge capacity can reach ∼145 mA h g−<sup>1</sup> and capacity retention can maintain 88% even after 100 cycles at a rate of 0.1C and at 60◦C. Our creative work could provide a useful method to develop SPEs with excellent performance, thus accelerating the commercial application of LMBs.

Keywords: graphene oxide, ionic liquid, polymer electrolyte, Li metal anode, Li metal batteries, rechargeable batteries

### INTRODUCTION

Li metal anodes have a high theoretical capacity (3,860 mA h g−<sup>1</sup> ) and extremely low redox potential (∼3.04 V vs. standard hydrogen electrodes), so Li metal batteries (LMBs) have been thought as promising secondary power sources for next generation of cutting-edge devices such as electric vehicles, autonomous aircrafts and smart grids (Zhang et al., 2016; Cheng et al., 2017; Wang et al., 2019; Cai et al., 2020; Wu et al., 2020; Zhu et al., 2020; Liao et al., in press). However, Li dendrite growth would happen during Li plating/stripping, probably resulting in short circuit and performance deterioration (Xu et al., 2014; Gao et al., 2019; Yan et al., 2020). In addition, liquid organic electrolytes commonly used in Li ion batteries are flammable and easy-leakage, and hence they would bring more severe safety problem when used in LMBs (Fu et al., 2016). Therefore, solid electrolytes have attracted increasing attention owing to their non-flammability, high safety, and good flexibility (Fan et al., 2018; Zhang et al., 2018).

At present mainly two kinds of solid electrolytes with quite different characteristics have been explored, which can be classified into inorganic electrolytes and polymer electrolytes (Fu et al., 2016). Although inorganic electrolytes have high ionic conductivities (∼10−<sup>3</sup> S cm−<sup>1</sup> ) at room temperature, they are stiff and fragile, resulting in their poor processability and huge interfacial resistance between them and Li metal anodes. On the contrary, solid polymer electrolytes (SPEs) containing lithium salts can be easily processed into a membrane owing to their

#### Edited by:

*Hongshuai Hou, Central South University, China*

#### Reviewed by:

*Jianmin Ma, Hunan University, China Yongchang Liu, University of Science and Technology Beijing, China*

#### \*Correspondence:

*Zhongliang Hu david10103@sina.com Yirong Zhu zhuyirong2004@163.com*

#### Specialty section:

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

Received: *09 January 2020* Accepted: *10 March 2020* Published: *31 March 2020*

#### Citation:

*Hu Z, Zhang X, Liu J and Zhu Y (2020) Ion Liquid Modified GO Filler to Improve the Performance of Polymer Electrolytes for Li Metal Batteries. Front. Chem. 8:232. doi: 10.3389/fchem.2020.00232*

**50**

properties of flexibility and stretchability. The commonly investigated polymers as SPEs include poly(ethylene oxide) (PEO) (Ito et al., 1987), polyacrylonitrile (PAN) (Slane and Salomon, 1995), poly(methyl methacrylate) (PMMA) (Appetecchi et al., 1995), and poly(vinylidene fluoride-cohexafluoropropene) (PVDF-HFP) (Boudin et al., 1999). Among them, PEO has been considered as the most appropriate polymer to prepare SPEs because PEO molecule has rich lone pair electrons in its O atoms, thereby endowing it with strong coordination ability (Mohanta et al., 2018). Unfortunately, the ionic conductivity of PEO is relatively low (∼10−<sup>6</sup> S cm−<sup>1</sup> at room temperature) (Zhang Q. et al., 2017), and obviously it can not reach the required value (>10−<sup>5</sup> S cm−<sup>1</sup> ) for SPEs of LMBs at ambient and moderate temperature.

Li ions can transport along with the PEO chains, so the ionic conductivity of PEO can be enhanced by various means of lowing its crystallization, including adding plasticizers (Zhang D. et al., 2017), mixing with other co-polymers (Puthirath et al., 2017) and introducing inorganic particles (Zhang J. et al., 2017). The investigation conducted by Cui's group demonstrated that the ion transport in polymers can be effectively ameliorated and their good physicochemical properties are well-maintained by incorporating inorganic fillers into polymers (Lin et al., 2015).

Recently, due to its unique structure and special characteristics, graphene oxide (GO) has been regarded as a promising active nanofiller to improve the performance of SPEs. GO has rich oxygen-containing groups (epoxy, hydroxyl, carboxyl, etc), and these groups can attract Li ions by acid-base interactions, consequently further promoting the dissociation of the lithium salt into free ions. Furthermore, GO can facilitate creating continuous ion channels within polymer electrolytes (Shim et al., 2014). The investigation conducted by Ardebili indicated that addition of 1% GO filler can induce 260% increase in tensile strength, and significantly improve thermomechanical stability of the polymer electrolyte (Yuan et al., 2014). However, GO tends to restack at high temperature owing to strong interlayer interaction, thus drastically expediting the capacity fading and greatly lowing the coulombic efficiencies (Yang et al., 2013).

On the other hand, recently the incorporating ionic liquids (ILs) into the polymer electrolytes becomes a popular approach to enhance their conductivity (Shin et al., 2005). Their beneficial properties such as non-volatility, non-flammability, high conductivity and high thermal stability can make them be excellent dopants for polymer electrolytes (Kar et al., 2019). We conceived that the filler of IL modified GO can wellimprove the performance of PEO. Firstly, GO can coordinate with oxygen atoms in PEO chains, thereby effectively reducing PEO's crystallization and weakening the interactions between PEO and ions as well as those among various ions (Lee et al., 2015). Secondly, the cations in ILs can also lower the interaction between Li<sup>+</sup> and PEO, thus promoting the migration of more lithium ions (Long et al., 2016). Finally, ILs can inhibit the aggregation of GO sheets by reacting IL-NH<sup>2</sup> with GO (Lin et al., 2014). The IL modified filler is expected to not only sharply enhance Li ion migration number but also greatly increase mechanical strength of SPEs.

In this work, 1-(3-aminopropyl)-3-methylimi-dazolium bromide (IL-NH2), and graphite oxide were used as precursors to synthesize IL-GO filler, which was adopted to modify PEO. The resultant PEO/GO-IL electrolyte possesses a high ionic conductivity at 25◦C (1.8 × 10−<sup>5</sup> S cm−<sup>1</sup> ), can effectively suppress the growth of Li dendrites on Li metal anode, and the LMBs with PEO/GO-IL electrolyte display good performance. The study could provide a methodlogy to solve the dilemma of LMBs by designing novel SPEs with excellent performance.

## EXPERIMENTAL

## Materials

Graphite powders (100 mesh, Nanjin XFNANo), LiFePO<sup>4</sup> cathode slurry (MTI Co.), Lithium bis(trifluoromethanesulfonyl) imide (LiTFSI, Alfa Aesar), Poly(ethylene oxide) (PEO, Mw=1000000, Sigma), 1-(3-aminopropyl)-3 methylimidazolium bromide (Lanzhou Institute of Chemical Physics), Potassium permanganate (KMnO4, Sinopharm), Sulphuric acid (H2SO4, concentrated, Sinopharm), Hydrogen peroxide (H2O2, Sinopharm), Potassium hydroxide (KOH, Sinopharm), Sodium nitrate (NaNO3, Sinopharm).

## Preparation of Graphite Oxide

Graphite oxide was prepared following the Hummers' method (Hummers and Offeman, 1958). Briefly, 4 g of graphite powder and 4 g of NaNO<sup>3</sup> were put into a flask, and then 200 mL of concentrated H2SO<sup>4</sup> was added into the flask at 0◦C under vigorous stirring. Atfer 24 g of KMnO<sup>4</sup> was gradually added, the mixture solution was first stirred at 0◦C for 2 h and then at 35◦C for 2 h. Subsequently 120 mL of distilled water was slowly dropped to cause an increase in temperature to 98◦C and this temperature was held for 25 min. After 200 mL of distilled water was again added into the mixture solution, 30 mL H2O<sup>2</sup> was added, ensuring that the residual permanganate and manganese dioxide were reduced to manganese sulfate. Finally, the reaction product was repeatedly washed with distilled water, centrifuged until the supernatant solution became neutral. The graphite oxide was collected, dried and preserved in a desiccator.

## Preparation of GO-IL Filler

IL-GO was synthesized by an epoxide ring-opening reaction between GO and IL-NH<sup>2</sup> (Yang et al., 2009; Lin et al., 2014). Firstly, 20 mg of IL-NH<sup>2</sup> and 20 mg of KOH were successively added into 20 mL of graphie oxide (0.5 mg/mL) solution, and the obtained turbid solution had been treated by ultrasonication until it became a homogeneous solution. Subsequently the mixture solution was transferred into a water bath to reflux at 80◦C for 24 h. The solid reaction product was centrifuged and washed with ethanol to remove the unreacted IL-NH2. Finally 10 g of the solid product and 50 mg of LiTFSI were successively added into 20 mL of distilled water, and the mixture solution reacted at 50◦C for 8 h under stirring. After centrifugation and washing with distilled water, the final GO-IL filler was obtained.

height profiles along the line shown in AFM image.

### Preparation of Electrode

The cathode slurry (the weight ratio of LiFePO4, Super P and PVDF is 8: 1: 1, NMP solvent) was evenly coated on an aluminum foil and then dried at room temperature to form a film. The loading mass of slurry in the film was ensured to be 2.5∼3.0 mg cm−<sup>2</sup> , and as-obtained LiFePO<sup>4</sup> (LFP) material was used as a cathode of LMB for later electrochemical tests.

#### Preparation of PEO/GO-IL SPE

The PEO/GO-IL SPE was prepared by solvent casting technique. 10 mg of the IL-GO filler was added into 10 g of absolute acetonitrile. After 2 h ultrasonication and subsequent 12 stirring, the mixture solution become homogeneous. Then, 800 mg of PEO and LiTFSI (the molar ratio of EO and Li is 20: 1) were added into the above solution and stirred for 12 h, resulting in a uniform and brown slurry. The above slurry was spread on a round dish and heated at 45◦C overnight to evaporate acetonitrile. Finally, the resultant membrane was peeled off from the dish and dried under vacuum at 45◦C for 12 h to obtain the PEO/GO-IL SPE.

### Materials Characterization and Electrochemical Measurements

The powder X-ray diffraction (XRD) patterns were collected using a Rigaku D/max-2500 with Cu Ka radiation. Atomic force microscope (AFM) images were obtained with a Digital Instruments nanoscope IIIa (Multimode,Veeco), operating in tapping mode. Fourier transform infrared (FTIR) spectra were recorded using a FTIR spectrometer (VERTEX 70v spectrometer, Bruker) in the wave number range of 400–4,000 cm−<sup>1</sup> . Scanning electronic microscopy (SEM) images were performed using a Hitachi SU8020 electron microscopy.

The cells were assembled in an argon-filled glove box. The half-cells were charged and discharged with a constant current between 2.5 V and 4.0 V. The charge/discharge cycling studies were performed on a battery test system (LAND CT2001, China). Electrochemical impedance spectra tests (EIS) were carried out in the frequency ranging from 0.1 to 100 KHz in an electrochemical workstation (Autolab). Linear sweep voltammetry (LSV) was conducted in an electrochemical workstation (Interface-1000E, Gamry) using Li/composite/stainless steel cells and the sweeping voltage ranged from 2.6 to 6.0 V with a scanning rate of 10 mV s−<sup>1</sup> .

### RESULTS AND DISCUSSIONS

The sketch of the synthesis processes of GO-IL filler is shown in **Figure 1**. The graphite oxide prepared by Hummers' approach can act as a good precursor for GO and it can easily form GO solution under ultrasonication when dispersed in alkaline solution. The GO reacted with IL-NH<sup>2</sup> by epoxide ringopening reaction between epoxy groups and -NH<sup>2</sup> groups in the precursors (Yang et al., 2009), and then reacted with LiTFSI by anion-exchange, forming the GO-IL filler.

**Figure 2A** shows the XRD pattern of graphite oxide, the precursor of GO. A peak at 12.2◦ can be clearly observed, which can be attributed to the (001) crystal plane of graphite oxide. The XRD characterization demonstrates that graphite oxide has a good crystalline structure. However, when it is dispersed in water or a base solution, it can easily dissociate

into single layer of graphene oxide, which was proved by AFM characterization. **Figure 2B** presents the AFM image and corresponding height profile of GO, and the thickness of GO is ∼1.05 nm, indicating that GO is of single layer (Zhu et al., 2010).

**Figure 3** displays the FTIR spectra of GO, IL-NH2, and IL-GO. For GO, the peak at 1,732 cm−<sup>1</sup> can be attributed to the stretching vibration of carboxyl groups of GO. The peak at ∼859 cm−<sup>1</sup> comes from epoxy groups, and the broad peak centered at 3,321 cm−<sup>1</sup> can be ascribed to the stretching vibration of hydroxyl groups of GO. For IL-NH2, The peak at ∼2,947 and 2,882 cm−<sup>1</sup> can be assigned to the stretching vibrations of methylene and methyl, respectively. The peak at 1,587 cm−<sup>1</sup> is due to imidazolium cations. For GO-IL, several new peaks at 2,926, 2,866, and 1,592 cm−<sup>1</sup> can be observed, inferring that GO has been successfully bonded with IL-NH2. In addition, the peak belonging to epoxy groups of GO disappears in GO-IL, further demonstrating GO has successfully reacted with IL-NH2.

The AFM image in **Figure 2** has fully proved that the GO is of single layer with a small size and the nanostructured GO-IL filler can be evenly distributed in PEO matrix. The SEM images of PEO and PEO/GO-IL are presented in **Figure 4**. Obviously, there exist very clear grain boundaries in PEO, and in some places the PEO crystals are evenly separated from each other by some pores, indicating that PEO has a good crystalline texture. In contrast, PEO/GO-IL does not exhibit any crystal boundaries (**Figure 4B**), and instead a smooth and uniform surface is presented, inferring that the crystallinity of PEO has been effectively lowed.

The XRD patterns of PEO and PEO/GO-IL are displayed in **Figure 4C**. Both samples have peaks at 2θ = 19.2 and 23.3◦ , which can be ascribed to the characteristic peaks of PEO. However, in contrast to PEO, PEO/GO-IL presents much broader and weaker peaks, indicating that the crystallinity of PEO in PEO/GO-IL has deteriorated greatly. Therefore, the XRD results further proved that the addition of GO/IL fillers can significantly low the crystallinity of PEO, in good agreement with SEM results.

The evolutions of the ionic conductivity with temperature were investigated and the results are presented in **Figure 4D**. Obviously, PEO/GO-IL (1.8 × 10−<sup>5</sup> S cm−<sup>1</sup> ) has a much higher ionic conductivity than PEO (1.7 × 10−<sup>6</sup> S cm−<sup>1</sup> ) at 25◦C, and it can be due to two reasons. Firstly, the addition of GO-IL decreased the crystallinity of PEO, which was beneficial to enhancing the ionic conductivity. Secondly, GO can dissociate lithium salt into free ions and facilitate creating continuous ion channels within polymer electrolytes (Shim et al., 2014). The structure variation of PEO in the electrolyte can be attributed to the addition of GO-IL, in which IL-NH<sup>2</sup> has reacted with GO by covalent bonding, thereby inhibiting the agglomeration of GO sheets. In this study, GO-IL can promptly increase the ionic conductivity at a low content. Unfortunately, when the content of GO-IL is >1%, the ionic conductivity will no longer improve and even decrease slightly with further increasing the GO-IL content. Hence the optimal content of GO-IL is 1%. The utilization of IL to modify GO via chemical bonding can make good use of the advantages of GO and IL, and it has be proved to be a promising way to enhance the performance of SPEs.

LSV has been performed to investigate the electrochemical stability of the electrolytes at 30◦C and the results are shown in **Figure 5A**. For PEO, the oxidation stage starts at 4.0 V. In contrast, for PEO/GO-IL, there is no oxidation current until 4.48 V. The wide electrochemical window of PEO/GO-IL is very related to its modified structure. In this study, the IL-NH<sup>2</sup> was grafted to GO sheets by covalent bonding, thus the agglomeration of GO can be effectively prevented and the role of GO in the composites can be well-performed. On the other hand, the addition of GO-IL filler further reduces the crystallinity of PEO. Suppression of lithium dendrite growth is crucial for the application of LMBs. Li/(PEO/GO-IL)/Li symmetrical cells were used to investigate the cycling performance and the tests were carried out under galvanostatic condition at 30◦C. As shown in **Figure 5B**, the cells with PEO/GO-IL electrolyte can be well-charged/discharged sequentially for more than 300 h with negligible cell voltage loss at 0.1 mA cm−<sup>2</sup> . In contrast, for PEO,

the voltage is unstable and decreases fast, and a short circuit would happened upon charging/discharging within ∼150 h.

Solid-state LFP/Li cells with PEO and PEO/GO-IL electrolytes were assembled to verify their electrochemical performances and the electrochemical experiments were performed under different rates at 60◦C. In cells using PEO/GO-IL electrolyte, polarization voltage increases slightly and the coulomb efficiency remains at 90% at 1C (**Figure 6A**), and the initial discharge capacity is 145 mA h g−<sup>1</sup> at 0.1C. In contrast, the charge/discharge curves become very rough in cells using PEO electrolyte, indicating that the internal lithium ion channel in PEO collapsed quickly when charged/discharged at 1C (**Figure 6B**). The great performance difference between PEO/GO-IL and PEO indicates that GO can act as a role to support the lithium ion channel in polymer.

The cycle performance of the cells with PEO/GO-IL electrolyte is evaluated at a rate of 0.1C at 60◦C (**Figure 6C**). The cell delivers the relatively smooth charge/discharge curves and high coulombic efficiency, inferring that electrode/electrolyte interface is very compatible during charge/discharge process. Moreover, the capacity retention is as high as 88% after 100 cycles (**Figure 6C**), manifesting considerable electrochemical stability of PEO/GO-IL electrolyte.

#### CONCLUSION

Novel PEO/GO-IL SPE has been developed by using GO-IL filler. The GO-IL filler is of high ionic conductivity, low crystallinity and excellent stability against the lithium anode electrode. By addition of 1% GO-IL filler, the crystallinity of PEO is significantly lowered and its performance is greatly improved. The ionic conductivity of PEO/GO-IL can reach 1.8 × 10−<sup>5</sup> S cm−<sup>1</sup> at 25◦C and the electrolyte can effectively suppress the Li dendrite growth against Li anode electrode. Moreover, the current density for stable Li plating/stripping in the PEO/GO-IL solid electrolyte can increase to 0.1 mA cm−<sup>2</sup> . The investigation in this study indicated that the performance of SPEs can be greatly improved by using the fillers with excellent properties and they can effectively solve the problems of the safety and performance deterioration of Li metal anode, thus accelerating the application of LMBs.

#### DATA AVAILABILITY STATEMENT

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

#### AUTHOR CONTRIBUTIONS

ZH and XZ conducted the most experiments. JL and YZ performed the characterization and data analysis. All

#### REFERENCES


authors involved the analysis of experimental data and manuscript preparation.

#### ACKNOWLEDGMENTS

The authors would like to acknowledge the financial support from the National Natural Science Foundation of China (Nos. 21576075 and 21376069).

of the counter anion. J. Appl. Polym. Sci. 135:46336. doi: 10.1002/app. 46336


Zhang, J., Zang, X., Wen, H., Dong, T., Chai, J., and Li, Y., et. al. (2017). Highvoltage and free-standing poly (propylene carbonate) /Li6.75La3Zr1.75Ta0.25O<sup>12</sup> composite solid electrolyte for wide temperature range and flexible solid lithium ion battery. J. Mater. Chem. A 5, 4940–4948. doi: 10.1039/C6TA10066J


lithium dendrite growth. Adv. Mater. 28, 2155–2162. doi: 10.1002/adma.2015 04117


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

# Sulfur-Doped and Bio-Resin-Derived Hard Carbon@rGO Composites as Sustainable Anodes for Lithium-Ion Batteries

Qinyuan Huang, Jinbo Hu\*, Shujing Wen, Xiang Zhang, Gonggang Liu, Shanshan Chang\* and Yuan Liu

*College of Materials Science and Engineering, Central South University of Forestry and Technology, Changsha, China*

Hard carbon derived from fossil products is widely used as anode material for lithium-ion batteries. However, there are still several main shortcomings such as high cost, and poor rate performance, which restrict its wide application. Then tremendous efforts have been devoted to developing biomaterials in the battery applications. Recently, especially agricultural and industrial by-products have attracted much attention due to the electric double-layer capacitors. Herein, we report the sulfur-doped hard carbon (SHC) materials from the tannin-furanic resins (TF-Resin) of the derived agricultural by-products, followed by enveloping rGO on its surface through the hexadecyl trimethyl ammonium bromide. SHC provides sites for the storage of lithium, while the rGO layers can offer a highly conductive matrix to achieve good contact between particles and promote the diffusion and transport of ions and electrons. As a result, the SHC@rGO shows excellent lithium storage performance with initial discharge capacity around 746 mAh g−<sup>1</sup> at a current density of 50 mA g−<sup>1</sup> , and shows superb stability keeping capacity retention of 91.9% after 200 cycles. Moreover, even at a high current density of 2,000 mAg−<sup>1</sup> , SHC@rGO still delivers a specific capacity of 188 mAg−<sup>1</sup> . These desired promising properties are active to the implement in the possible practical application.

Keywords: bio-resin-derived hard carbon, tannin–furanic resins, sulfur-doped, rGO, lithium-ion battery

## INTRODUCTION

Nowadays, lithium-ion batteries (LIBs) have been widely used with energy storage systems and portable digital devices because of their long cycle stability and high energy density (Etacheri et al., 2011; Wang et al., 2014). With a high demand for LIBs, the preparation of low-cost, environmentally friendly, and high-performance anode materials has been substantially researched in recent years (Liao et al., 2012). Various carbonaceous materials, such as graphitic carbon (Funabiki et al., 1999; Song et al., 2011), amorphous carbon (hard carbon) (Zhang et al., 2015, 2016, 2017) and soft carbon (Chae et al., 2014; Huang et al., 2018; Wang et al., 2018)), nanostructured carbon (graphene) (Raccichini et al., 2016; Ferguson et al., 2017), and carbon nanotubes (Liu et al., 2018; Yuan et al., 2018) have been widely investigated for their LIBs applications. Among all the anode materials for LIBs, graphitic material is the most commercially unitized because of its low cost, low potential (≈0.2 V vs. Li/Li+), and optimal electrical conductivity (Han et al., 2015). Nevertheless, graphite anode is far from meeting the demands for high energy/power density as a result of its limited capacity (372 mAh g−<sup>1</sup> ) and inferior rate performance (Ge et al., 2018).

#### Edited by:

*Weijie Li, University of Wollongong, Australia*

#### Reviewed by:

*Xing-Long Wu, Northeast Normal University, China Xianhong Rui, Guangdong University of Technology, China*

#### \*Correspondence:

*Jinbo Hu hjb1999@hotmail.com Shanshan Chang changelxy@hotmail.com*

#### Specialty section:

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

Received: *14 January 2020* Accepted: *13 March 2020* Published: *02 April 2020*

#### Citation:

*Huang Q, Hu J, Wen S, Zhang X, Liu G, Chang S and Liu Y (2020) Sulfur-Doped and Bio-Resin-Derived Hard Carbon@rGO Composites as Sustainable Anodes for Lithium-Ion Batteries. Front. Chem. 8:241. doi: 10.3389/fchem.2020.00241*

Alternatively, majority of researchers have taken notice of hard carbon because of its high specific capacity (740 mAh g −1 ) (Fey and Chen, 2001). As well-known, the precursors for preparing hard carbon are mainly petrochemical raw materials, such as a phenolic resin (Liu et al., 1996), high molecular polymers (Piotrowska et al., 2013), asphalt (Larcher et al., 1999; Mochida et al., 2001), etc. These raw materials are non-renewable substances, and the price is volatile because of fluctuations in international oil prices. Also, carbonization of these products often causes environmental issue such as the release of formaldehyde or other toxic carcinogens. Fortunately, biomass is a rich and renewable potentially green material. A large amount of biomass with little or no economic value can be used as a cheap and effective source of carbon precursors to produce materials that are environmentally and economically of high added value. Most natural resources have been reported including bamboo chopsticks (Jiang et al., 2014), wood (Zhang et al., 2019), soybean (Xu et al., 2015), wheat flour (Lim et al., 2017), corn cob (Liu et al., 2016), peanut shells (Ding et al., 2015), cherry stones (Arrebola et al., 2010), silk (Hou et al., 2015), coconut oil (Gaddam et al., 2016), and mangrove (Liu et al., 2010), were widely used as electrodes for energy storage applications. However, the lower conductivity of hard carbon than that of graphitizable carbons is another defect, which results in poor rate performance. In order to tackle these problems, great efforts have been made to adjust the surface structure of hard carbon, such as the construction of hybrid anodes (Guo et al., 2008), thermal carbon coatings (Ohzawa et al., 2005; Lee et al., 2007), and vacuum and oxidation treatments (Fujimoto et al., 2010; Liu et al., 2010). Although these surface modification methods may improve the coulombic efficiency of hard carbon, their rate performance and cyclability are far from satisfactory. Moreover, the further modification optimization of hard carbon materials is quite meaningful for practical application.

Corn cob and wood bark are abandoned agricultural byproducts and are generally burnt, which not only leads to air pollution but also wastes resources. To date, tannin– furanic resins have received great attention because of tannin and furfural alcohol, which are extracted from wood bark and corn cob (Tondi and Pizzi, 2009; Meikleham and Pizzi, 2010). Thus far, the synthesis of resins without formaldehyde has been an attractive focus (Tondi et al., 2009; Li et al., 2013). With the development of technology, possible applications of the natural tannin–furanic resin were recently presented (Tondi et al., 2016). It is active that the epoxy novolac resin as the hard carbon source and coated rGO for LIBs have shown the outstanding stability and rate capability due to via constructing the conductive network (Zhang et al., 2015). Herein, we report the synthesis of SHC materials derived from the biomass TF-resin with the carbonization method and enveloped with the rGO on its surface through the hexadecyl trimethyl ammonium bromide. SHC provides active sites for the storage of lithium, whereas the rGO layer can offer a highly conductive matrix to achieve contact between particles and promote the diffusion and transport of ions and electrons. As a result, the SHC@rGO shows excellent lithium storage performance with an initial discharge capacity of approximately 746 mAh g −1 at a current density of 50 mA g−<sup>1</sup> , and shows superb stability keeping a capacity retention of 91.9% after 200 cycles. Furthermore, even at a high current density of 2,000 mA g−<sup>1</sup> , SHC@rGO still delivers specific capacity of 188 mA g−<sup>1</sup> . This study compares the following petrochemical raw materials: (i) tannins and furfuryl alcohol raw materials, which have the advantages of environmental friendliness, low-cost, and renewability; (ii) the full biomass-derived hard carbon, which has optimal electrochemical properties; and (iii) it provides a reference for the application of the full biomass resin materials in energy storage materials.

#### EXPERIMENTAL SECTION

Tannin extract (60 g), furfuryl alcohol (180 g), 100 mL deionized water, and 8 mL p-toluene sulfonic acid (65 wt%), which acted as the hardener, were mixed at room temperature. The precursor was treated at 60◦C for 1 h in a rotary evaporator and then cured in the drying box at 60, 100, and 150◦C for 2, 1, and 24 h, respectively. The cured precursor was heated at 500◦C for 1 h with 3◦C min−<sup>1</sup> and then ground to powders after cooling under the argon atmosphere. Finally, the sample was heated at 1,000◦C under the argon flow for 1 h with a ramp rate of 10◦C min−<sup>1</sup> to prepare the SHC.

SHC (2 g) and GO (0.4 g) were sonicated for 0.5 h in 120 mL (20 wt%) ethanol to disperse GO sheets and SHC powders. The 0.1 g hexadecyl trimethyl ammonium bromide was dissolved in the resulting suspended liquid through magnetic stirring for 5 h at room temperature. The suspension was vacuum filtered, dried in air at 100◦C, and then heated to 1,000◦C under argon flow for 1 h at a heating rate of 10◦C min−<sup>1</sup> in a vacuum tube furnace to prepare the SHC@rGO.

## MATERIAL CHARACTERIZATION

The morphologies of the sample were investigated with a scanning electron microscope (SEM). The morphology and microstructure of the SHC@rGO sample were characterized by transmission electron microscopy (TEM F20) and scanning transmission electron microscopy equipped with an energy dispersive spectrometer (EDS). The X-ray photoelectron spectra (XPS) were obtained by a Thermo Fisher Scientific ESCALAB 250Xi spectrometer. X-ray diffraction (XRD) was recorded on an XD-2X X-ray diffractometer with Cu– Kα radiation (λ = 1.5406 A, 30 KV, 20 mA) at the scan rate of 8◦C min−<sup>1</sup> . Raman spectra were collected using a Raman spectrometer (Alpha300-R) with 532 nm. Nitrogen adsorption and desorption isotherms were determined through nitrogen physisorption on a Micro Active for an ASAP 2460 analyzer. Comprehensive thermal analysis (TG-DSC) was obtained using a SETARAM AETSYS-24 comprehensive thermal analysis instrument, starting from room temperature to 1,200◦C at a heating rate of 10◦C min−<sup>1</sup> under an N<sup>2</sup> gas atmosphere.

## ELECTROCHEMICAL TESTING

Electrochemical measurements were performed using a 2016 coin-type battery using lithium metal as the anode. The working electrode was composed of active materials (SHC and SHC@rGO), the conductivity material (super P), and the binder (PVDF) in a weight ratio of 8:1:1 dissolved in N-methyl pyrrolidinone. The slurry was then spread evenly on the copper foil and dried in a vacuum oven at 120◦C for 12 h. After drying, the electrode was cut into a disc with a diameter of 12 mm. The carrying mass of SHC@rGO on the anode is 0.55– 0.66 mg·cm−<sup>2</sup> . The mass loads of rGO in SHC@rGO is 0.11– 0.13 mg·cm−<sup>2</sup> . A solution of 1 M LiPF<sup>6</sup> in ethylene carbonate, diethyl carbonate, and dimethyl carbonate (1:1:1 by volume ratio) was used as an electrolyte, and a polypropylene film was used as a separator. The cycling and rate performance of SHC and SHC@rGO were tested on a land battery test system at 25◦C. Cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS) measurements were performed on an Ivium electrochemical workstation.

### RESULTS AND DISCUSSION

**Figure 1** shows the schematic illustration for the synthesis of SHC and SHC@rGO materials derived from the biomass TF-Resin. Firstly, TF-Resin precursors were obtained a simple solidify method (see Experimental Section for detail). Subsequently, the obtained TF-Resin precursors were calcinated under Ar atmosphere to prepare SHC. Finally, the SHC@rGO was obtained after enveloped with the rGO on its surface through the hexadecyl trimethyl ammonium bromide and assembled into a 2016 Coin battery.

**Figure 2** shows the SEM images of SHC and SHC@rGO. It can be seen that SHC sample presents irregular morphology and the surface is non-porosity and glossy (**Figures 2A,B**). The SHC@rGO sample consists of irregular HC particles and lamellar reduced graphene oxide (rGO) with a wrinkled surface (**Figures 1C,D**). which is help construct the conductive network and guaranteeing fast electron conduction.

The microstructure of SHC@rGO composites was further studied by TEM and high-resolution TEM (HRTEM). As presented in **Figure 3A**, the SHC is partially wrapped by corrugated graphene nanosheets. It can be known that winkled rGO is well bonded to SHC particles due to van der Waals force between graphene sheets and SHC, which is beneficial to transport electrons and lithium ions. HRTEM image of SHC@rGO sample taken on the edges of rGO is shown in **Figure 3B**. The rGO consists of several stacked layers and its thickness is <5 nm. Graphene coating can effectively promote the transmission rate of electrons and ions and maintain the integrity of the conductive network, which contribute to improved electrochemical performance. The EDS mapping of SHC is shown in **Figure 3C**. The EDS element mapping of C and S confirms that the S element is evenly distributed throughout the hard carbon.

To determine the doping level and bonding configuration of S, we performed the XPS analyzation of SHC@rGO. As presented in **Figure 4A**, the predominant peaks at 165.0, 228.0, 285.0, 402.0, and 532.0 eV are assigned to S 2p, S 2s, C 1s, N 1s, and O 1s, respectively. The S 2p spectra (**Figure 4B**) showed three

peaks at 164.1, 165.3, and 169.2 eV. The former two peaks are assigned to the thiophene-S groups (–C–S–C–), and the third peak corresponds to the –C–SOx groups (X = 2–4) (Choi et al., 2011; Kiani et al., 2019). The XPS results indicate that the S atom has been chemically bonded to the SHC.

The XRD patterns of the SHC and SHC@rGO samples are presented in **Figure 5A**. It can be seen that the XRD patterns of SHC and SHC@rGO are both composed of wide diffraction peaks at 22◦ -26◦ and 44◦ -45◦ 2θ corresponding to (002) and (100), respectively, which are typical of carbonaceous materials with an amorphous-like structures (Inagaki et al., 2001). The D<sup>002</sup> values of SHC and SHC@rGO are summarized in **Table S1**. SHC@rGO (0.351 nm) exhibits smaller average interlayer distance than SHC (0.380 nm), which is induced by rGO addition, because of the graphene oxide is reduced at high temperature to form a layered structure with decreasing interlayer spacing.

**Figure 5B** shows the Raman spectra of SHC and SHC@rGO. The characteristic peaks of the two samples are similar. The D-bands peak at about 1,340 cm−<sup>1</sup> and the G-bands peak at about 1,590 cm−<sup>1</sup> . Peak D-bands and G-bands represent sp<sup>3</sup> hybridization and sp<sup>2</sup> hybridization, respectively. The ratio of the integrated areas of the two peaks can manifest the degree of order of carbon materials. The ID/I<sup>G</sup> of the SHC and SHC@rGO is 1.13 and 1.08, respectively (Ferrari and Robertson, 2000). It was shown that the carbon atoms in sp<sup>2</sup> hybridization increased after graphene was added, the structure order and graphitization degree of the materials increased. This result was consistent with the results of XRD.

The nitrogen adsorption-desorption isotherm test (77 K) is an important tool of characterizing the specific surface area and pore volume of porous materials. **Figure 5C** shows the N<sup>2</sup> adsorptiondesorption curve and the pore size distribution based on the BJH (Barrett-Joiner-Halenda) model of SHC and SHC@rGO. It can be seen that the N<sup>2</sup> adsorption isotherm of SHC and SHC@rGO belong to the III type isotherm. The calculated pore size distributions from the adsorption branches using the density functional theory (DFT) (inset of **Figure 5C**) model for SHC and SHC@rGO, suggest the characteristics of mesoporosity and microporosity. The specific surface area SBET and pore volume data are listed in **Table S1**. It can be seen from the table that the BET specific surface area increases from 11.21 m<sup>2</sup> g <sup>−</sup><sup>1</sup> of SHC to 21.02 m<sup>2</sup> g <sup>−</sup><sup>1</sup> of SHC@rGO, the pores volume remains basically comparative. For these two samples, the mesopore volume accounts for more than 90% of the total pore volume.

**Figure 5D** shows the TG/DSC curves for GO and SHC@rGO. The TG/DSC of pure SHC curve is not shown because it has no heat loss and exotherm. It can be seen that GO has a distinct exothermic peak at 190◦C with significant thermal weight loss, which is caused by the volatilization of a large number of water molecules adsorbed on GO. The thermogravimetric curve of SHC@rGO showed a slight decrease due to the small proportion of GO in the SHC@rGO mixture. When the temperature increases from 254 to 388◦C, the DSC curve of GO shows a taro peak with a slight thermal weight loss, indicating a moderate exothermic reaction, which is ascribed to escape of the large number of H atoms in GO. When the temperature is at 781 and 1,000◦C, there is a significant endothermic peak with obvious thermal weight loss, the TG curve still slightly decreases, which is caused by the large amount of O atoms in GO continuously escaping at high temperature. It also shows that GO can be

reduced at high temperatures. When the pyrolysis temperature is 1,200◦C, the residual mass ratio of GO and SHC@rGO are 46.6 and 86.6%, respectively. Assuming that the mass of SHC does not change, the theoretical calculation shows that the residual mass ratio of SHC@rGO is 88.3%, which is basically consistent with the experimental results. It indicates that at high temperatures, most of the H and O atoms in the GO are removed.

The electrochemical profiles of SHC and SHC@rGO are characterized by coin half cells, which are composed of SHC or SHC@rGO as an anode, a lithium metal as counter electrodes, electrolyte, and a Celgard 2,400 separator. **Figure 6** shows the initial two CV curves of SHC and SHC@rGO between 0 and 3 V at a sweep rate of 0.1 mV s−<sup>1</sup> . It can be seen that cathode reduction peaks appear at 0.70–0.75 V in the first cycle and disappear in the subsequent cycle (**Figure 6B**). These reduction peaks suggest that the solid electrolyte interphase (SEI) layer being formed on the surface of the carbon materials during the first lithium intercalation (Buqa et al., 2006). Cathodic current peaks reach 0.21 V because of irreversible side reactions of lithium with absorbed species or surface functional groups, a weak reduction peak was found in two samples during the first discharge process but disappeared in the second cycle (**Figure 6B**), which contributes to part of capacity loss leading to large irreversible capacity (Zhang et al., 2014). In addition, the anodization peak potential of SHC@rGO was lower than that of pure SHC, indicating that when graphene is added to the hard carbon material, lithium-ions are more easily deintercalated in the hard carbon material during the charge/discharge process, which will be beneficial to improve its electrochemical performance. Therefore, the graphene in SHC@rGO not only provides electronic conductivity but also improves the reversible migration ability of lithium-ions in materials.

The initial galvanostatic discharge/charge curves of SHC and SHC@rGO at a current density of 50 mA g−<sup>1</sup> are shown in **Figure 7A**. The charging and discharging processes correspond to the lithium removal and insertion process, respectively. It can be seen that the shape of the first charge-discharge curves of SHC and SHC@rGO is analogical, indicating that the lithium intercalation/deintercalation mechanism of SHC has not been changed after modification by wrapped GO. It can be seen that the specific capacity of SHC and SHC@rGO for the initial discharge capacities are 605 and 746 mAh g−<sup>1</sup> , and the charge capacities are 321 and 486 mAh g−<sup>1</sup> , respectively. The first irreversible specific capacity of SHC and SHC@rGO are 284 and 260 mAhg−<sup>1</sup> , and the initial Coulomb efficiency (ICE) of SHC and SHC@rGO are 53.1% and 65.1%, respectively.

Unfortunately, the pure SHC electrodes show low ICE at the same current density, similar to previous reports about hard carbon material (Zhang et al., 2015). The battery of the low ICE is probably attributed to the following two major reasons. Primarily, the decomposition of electrolyte during the initial charge/discharge process has significantly displayed in the battery formulation of SHC derived the resin. Secondly, the formation of an irreversible solid electrolyte interface (SEI) on the surface of anode could also lead to a low ICE (Liu et al., 2016), consistent with the CV results. Nevertheless, it is active that the ICE can be improved from 53.1 to 65.1% owing to the presence of rGO in SHC@rGO. The ICE improvement of SHC@rGO was ascribed to the crucial rGO, which can construct the conductive network and promote the transport of Li<sup>+</sup> ions in the electrolyte. Therefore, the SHC would be progressively modified in the prospective battery.

The rate performance of SHC and SHC@rGO prepared at different current densities are shown in **Figure 7B**. The rate performance of SHC@rGO is superior to that of SHC under different current densities from 0.1 A g−<sup>1</sup> (410 mAh g−<sup>1</sup> ) to 2 A g−<sup>1</sup> (188 mAh g−<sup>1</sup> ). When the current rate decreased from 2 A g−<sup>1</sup> to 0.1 A g−<sup>1</sup> after the 25th rate cycle corresponding to discharge capacity of 444.7 mAh g−<sup>1</sup> . As for SHC, a much lower rate capacity was delivered at varied current densities. Both the capacity of SHC and SHC@rGO increased when the current density returned to 0.1 A g−<sup>1</sup> , this phenomenon can be confirmed by the electrolyte infiltration and electrode material activation during different current charge/discharge processes. **Table S2** lists TF-resin prepared in this work vs. other hard carbons derived from epoxy novolac resin (Zhang et al., 2014), pitch (Kim et al., 2016), and polyphenylene sulfide (Luo et al., 2018), which demonstrates that the SHC derived from TF-resin deliver better discharge capacity and rate performance.

Cycle stability of SHC and SHC@rGO was estimated at the current density of 400 mA g−<sup>1</sup> (**Figure 7C**). It can be inferred that from this cyclic performance, SHC and SHC@rGO delivered 134.1 and 271.8 mAh g−<sup>1</sup> capacity after 200 cycles with the Coulombic efficiency near 100%, keeping capacity retention of 77.7 and 91.9%, respectively. Therefore, the SHC@rGO hybrid anode shows obviously higher cycling stability and reversible capacity, which are S-doping hard carbon provides active sites for the storage of lithium, while the rGO layers offer a highly conductive matrix and high contact area can lead to effective contact with the electrolyte into the electrode and quickly transport ions into the deeper parts of the SHC@rGO particles and graphite layer. A comparison of hard carbon prepared in this work with other biomass-derived carbons is shown in **Table S3**, which demonstrates that the prepared hard carbon delivers better rate capabilities and cycling performance.

The EIS measurements of SHC and SHC@rGO were tested (**Figure 5D**). The plots consist of a semicircle in the highfrequency region and a straight line in the low-frequency region. the straight line and the semicircle corresponding to Warburg diffusion impedance (W) and charge transfer resistance (Rct), respectively (Pan et al., 2018). Based on equivalent electric circuit, the Rct of SHC and SHC@rGO are 65.34 and 50.49, respectively. The lower Rct of SHC@rGO was attributed to the

#### REFERENCES

Arrebola, J. C., Caballero, A., Hernán, L., Morales, J., Olivares-Marín, M, and Gómez-Serrano, V. (2010). Improving the performance of biomass-derived carbons in li-ion batteries by controlling the lithium presence of conductive rGO able to constructing the conductive network and facilitate the transport of Li<sup>+</sup> ions, resulting in a superior electrochemical performance.

#### CONCLUSION

In summary, biomass TF resin-derived SHC and SHC@rGO composite material was prepared as the anode material for lithium-ion battery. It is demonstrated that the rGO networks are enveloped on SHC constructing the desired microstructure. Compared with the biomass-derived pure SHC, the SHC@rGO composites show enhanced conductivity offered by rGO networks and large Li ion storage sites supported by SHC. As a consequence, SHC@rGO composite exhibits a superior electrochemical rate performance and reversible capacity. Based on this study, a new biomass derived hard carbon composites with superior electrochemical properties is reported, which is promising for low-cost and eco-friendly anode material of LIBs.

## DATA AVAILABILITY STATEMENT

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

#### AUTHOR CONTRIBUTIONS

JH, XZ, SC, and YL: conceived and designed the experiments. QH, SW, and JH: performed the experiment. SC, JH, and XZ: supervised the work. QH, JH, and YL: wrote the paper. JH, GL, and XZ: revised the paper. JH and YL: contributed equally. All authors reviewed and approved the final manuscript.

### FUNDING

The research was funded by the National Key Research and Development Program of China (2017YFB1303901; to JH) and the Scientific research project of Hunan Provincial Department of Education (19A505; to JH).

### ACKNOWLEDGMENTS

The authors thank Priof. Xiaobo JI from Central South University in China for the analysis.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2020.00241/full#supplementary-material

insertion process. J. Electrochem. Soc. 157, A791–A797. doi: 10.1149/1.34 25728

Buqa, H., Würsig, A., Vetter, J., Spahr, M. E., Krumeich, F., and Novák, P. (2006). SEI film formation on highly crystalline graphitic materials in lithium-ion batteries. J. Power Sour. 153, 385–390. doi: 10.1016/j.jpowsour.2005.05.036


using a molecular weight controlled pitch derived from petroleum residue. J. Ind. Eng. Chem. 41, 1–9. doi: 10.1016/j.jiec.2016.07.006


energy density. J. Mater. Sci. 52, 10418–10430. doi: 10.1007/s10853-017- 1206-3


**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 Huang, Hu, Wen, Zhang, Liu, Chang and Liu. 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.

# β-FeOOH Interlayer With Abundant Oxygen Vacancy Toward Boosting Catalytic Effect for Lithium Sulfur Batteries

Yingying Li <sup>1</sup> , Xifei Li 1,2,3 \*, Youchen Hao<sup>2</sup> , Alibek Kakimov <sup>2</sup> , Dejun Li <sup>1</sup> , Qian Sun<sup>4</sup> , Liang Kou<sup>5</sup> , Zhanyuan Tian<sup>5</sup> , Le Shao<sup>5</sup> , Cheng Zhang<sup>5</sup> , Jiujun Zhang2,6 and Xueliang Sun1,2,4

*<sup>1</sup> Tianjin International Joint Research Centre of Surface Technology for Energy Storage Materials, Energy & Materials Engineering Centre, College of Physics and Materials Science, Tianjin Normal University, Tianjin, China, <sup>2</sup> Xi'an Key Laboratory of New Energy Materials and Devices, Institute of Advanced Electrochemical Energy & School of Materials Science and Engineering, Xi'an University of Technology, Xi'an, China, <sup>3</sup> State Center for International Cooperation on Designer Low-Carbon & Environmental Materials (CDLCEM), Zhengzhou University, Zhengzhou, China, <sup>4</sup> Department of Mechanical and Materials Engineering, University of Western Ontario, London, ON, Canada, <sup>5</sup> Shaanxi Coal Chemical Industry Technology Research Institute Co., Ltd., Xi'an, China, <sup>6</sup> Department of Chemistry, College of Sciences/Institute for Sustainable Energy, Shanghai University, Shanghai, China*

#### Edited by:

*Hongshuai Hou, Central South University, China*

#### Reviewed by:

*Qiaobao Zhang, Xiamen University, China Xunhui Xiong, South China University of Technology, China*

> \*Correspondence: *Xifei Li*

*xfli2011@hotmail.com*

#### Specialty section:

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

Received: *25 February 2020* Accepted: *27 March 2020* Published: *23 April 2020*

#### Citation:

*Li Y, Li X, Hao Y, Kakimov A, Li D, Sun Q, Kou L, Tian Z, Shao L, Zhang C, Zhang J and Sun X (2020)* β*-FeOOH Interlayer With Abundant Oxygen Vacancy Toward Boosting Catalytic Effect for Lithium Sulfur Batteries. Front. Chem. 8:309. doi: 10.3389/fchem.2020.00309* Due to the shuttle effect and low conductivity of sulfur (S), it has been challenging to realize the application of lithium-sulfur (Li-S) batteries with high performance and long cyclability. In this study, a high catalytic active CNTs@FeOOH composite is introduced as a functional interlayer for Li-S batteries. Interestingly, the existence of oxygen vacancy in FeOOH functions electrocatalyst and promotes the catalytic conversion of intercepted lithium polysulfides (LiPS). As a result, the optimized CNTs@FeOOH interlayer contributed to a high reversible capacity of 556 mAh g−<sup>1</sup> at 3,200 mA g−<sup>1</sup> over 350 cycles. This study demonstrates that enhanced catalytic effect can accelerate conversion efficiency of polysulfides, which is beneficial of boosting high performance Li-S batteries.

Keywords: CNTs@FeOOH, oxygen vacancy, catalytic effect, polysulfides, Li-S batteries

## INTRODUCTION

Currently, Li-S batteries have received extensive attention due to their high theoretical capacity (1,675 mAh g−<sup>1</sup> ), high energy density (2,600 Wh kg−<sup>1</sup> ) (Ji and Nazar, 2010; Manthiram et al., 2013), low cost and environmental friendliness (Nazar et al., 2014; Yang et al., 2018). Considering the low conductivity of S (σ = 5.0 × 10−<sup>30</sup> S cm−<sup>1</sup> ) and the shuttle effect of LiPS, research on Li-S batteries is strongly delayed and thus hardly meets actual needs (Zhang, 2013; Rosenman et al., 2015).

To solve these problems, many pioneering works are using a porous carbon-based host, a functional interlayer, absorptable polar composites [such as CoS2/C (Yuan et al., 2016), MnO2/GO/CNT (Kong et al., 2017), S@TiO<sup>2</sup> (Wei Seh et al., 2013) etc.] and a catalytic effect on LiPS conversion. Among them, carbon-based hybrid materials with absorbility to LiPS have always shown attractive characteristics (Tang and Hou, 2018). For instance, Pang and Nazar obtained C3N<sup>4</sup> by pyrolysis of melamine, which has rich pyridine nitrogen adsorption sites. This can lead to the fact that the sulfur electrode with ultra-low long-term capacity fades out by 0.04% for a cycle over 1,500 cycles at a practical rate of 0.5C (Pang and Nazar, 2016). However, it should be noted that absorbed LiPS is easy to release, because these materials are soluble in the electrolyte. Thus, a more efficient strategy is urgently needed to meet this problem. Catalyst has been employed to accelerate the transformation of LiPS from liquid to solid, which corresponds to the transmission from long-chain Li2S<sup>6</sup> to short-chain Li2S. We strongly believe that this is an alternative option to modify Li-S batteries by introducing catalytic materials to catalyze the conversion of LiPS into insoluble products. Yang group reported the Fe3C/Fe-Nx@NPCN modified separator, due to the catalytic effect of Fe3C to LiPS, the modified batteries yielded a high capacity of 1,517 mAh g−<sup>1</sup> at 0.1C and displayed a capacity decay rate of 0.034% per cycle after 500 cycles at 1C (Yang et al., 2019). Bian et al. took multi-functional porous carbon nanofibers (g-C3N4@PCNF) as the sulfur host, in which g-C3N<sup>4</sup> contributed to rapid oxidation-remediation conversion of S species and slowed down LiPS yield. Consequently, the g-C3N4@PCNF/S cathode achieves good flexibility and excellent cycling retention, e.g., long cycling with decay power of only 0.056% per cycle for 500 cycles at 1.0 A g−<sup>1</sup> (Bian et al., 2019). In addition, Lee and coworkers have demonstrated that catalytic activity in anoxic sites is higher than in saturated sites as the oxygen vacancy can contribute to the transformation of electrons and the formation of S3<sup>−</sup> radicals (Lin et al., 2018). Therefore, we are inspired to develop more powerful carbon-based composites with better catalytic effect for LiPS.

As previously reported, FeOOH with abundant oxygen vacancies (Zhang et al., 2018) can be a promising candidate as the catalyst for LiPS. We propose using CNTs@FeOOH composite materials to enhance the stability of Li-S batteries. On the one hand, FeOOH is efficient to catalyze the transformation of polysulfides; on the other hand, CNTs provide a fast electron transport channel, which ensures the sustainability of the reaction under high current and reduces the occurrence of polarization. As a result, the obtained electrodes maintained a high reversible capacity of 556 mAh g−<sup>1</sup> at 2C for 350 cycles. This work plays a significant contribution to the development of Li-S batteries with high performance and long lifespan.

## EXPERIMENTAL SECTION

## Synthesis of CNTs@FeOOH Composite and Separator

As reported in our previous work (Hao et al., 2019), CNTs@FeOOH composites with various mass ratio can be obtained by controlling the content of iron source. For comparison, the pure FeOOH phase was synthesized under the same conditions without adding CNTs.

CNTs@FeOOH composites were homogenized into paste by N, N-2-methyl pyrrolidone, the slurry was evenly coated on the polypropylene separator (PP separator). The CNTs@FeOOHcoated separator was dried overnight in vacuum at 40◦C and then cut into circular disks (16 mm). CNTs and pure FeOOH phase were wrapped on the separator surface by the same method as the control experiment.

## Preparation of the Pure Sulfur Cathode

The commercial S powder, acetylene black and polyvinylidene fluoride binder (PVDF) with a mass ratio of 55:30:15 were mixed into N-methyl-2-pyrrolidone (NMP) solvent and stirred on an electromagnetic stirrer for 24 h to obtain the slurry. The slurry was then cast on Al foil and dried overnight in a vacuum oven at 60◦C. The load mass of S is about 0.8 mg cm−<sup>2</sup> . And the lithium foil was used as the anode electrode.

## Materials Characterization

X-ray diffraction (XRD) spectra were tested by Bruker AXS D8 Advance diffractometer with Cu/Kα radiation. Thermogravimetric analysis (TGA, Pyris Diamond6000 TG/DTA, PerkinElmer Co., America) was performed to confirm the FeOOH content in composites. The morphologies of asobtained samples were measured by a field-emission scanning electron microscope (SEM Hitach SU8010) and JEOLJEM-3000F transmission electron microscope (TEM). The composition of the elements on the composites surface was verified by X-ray photoelectron spectroscopy (XPS PHI5000 Versa Probe).

## Electrochemical Characterizations

Electrochemical performance was studied using CR2032 cointyped cells assembled in an argon filled glovebox. 1.0 M lithium bis-trifluoromethanesulfonylimide (LiTFSI) in 1,3 dioxolane (DOL) and 1,2-dimethoxyethane (DME) at a volume ratio of 1: 1 with 1 wt% LiNO<sup>3</sup> additive was utilized as electrolyte. Galvanostatic charge/discharge characteristics were tested on the LAND CT2001A battery tester. Princeton Applied Research Versa STAT4 was used to perform cyclic voltammetry at a scanning rate of 0.1 mV s−<sup>1</sup> . Electrochemical impedance spectroscopy (EIS) was performed on Princeton Applied Research Versa STAT4. The frequency range was 0.01 Hz−100 kHz amplitude of AC was 5.0 mV. All electrochemical tests were performed in the range of 1.7–2.8 V.

## RESULTS AND DISCUSSION

The morphologies of CNTs, bare FeOOH and CNTs@FeOOH composites were visualized via scanning electron microscope (SEM). As depicted in **Figures 1a–c**, an increasingly obvious stick-like FeOOH being grown on the surface of CNTs (denoted as CNTs@FeOOH-I, II, and III, respectively). Compared to pure CNTs and FeOOH in **Figure S1**, the formation of rod-shaped FeOOH may induced by the reaction conditions. In addition, the X-ray diffraction (XRD) patterns of all samples are compared in **Figure 1d**. The peaks of pure FeOOH are located at 2θ = 11.925◦ , 16.901◦ , 26.874◦ , 34.185◦ , 35.311◦ , 39.386◦ , 46.656◦ , 52.349◦ , 56.158◦ , 61.278◦ , 64.718◦ , and 68.117◦ are attributed to the (110), (200), (130), (400), (211), (301), (411), (600), (251), (002), (541), and (132) reflection planes of β-FeOOH (JCPDS 75-1594). The FeOOH peaks become more evident with increasing FeOOH content and no impurity phase is detected. Transmission electron microscope (TEM) images of CNTs@FeOOH-II, demonstrated that FeOOH particles are uniformly adhered on the surface of CNTs, which consistent well with the results of SEM (see **Figures S2a,b**). The selected area electron diffraction pattern of the CNTs@FeOOH-II nanomaterials shows diffraction rings characteristic of FeOOH and CNTs (see **Figure S2c**), while the image of high resolution TEM in **Figure S2d** indicates lattice fringes corresponding to (110), (200), (211), and (330) planes. The results are consistent with those of XRD spectrums. Besides, thermogravimetric analysis (TGA) was used to calculate the proportion of components in various composites (Zhang et al., 2017). As shown in **Figure 1e**, the FeOOH content is measured as 24.8, 45.7, and 71.5 wt% for the composites CNTs@FeOOH-I, II, and III, respectively.

X-ray photoelectron spectroscopy (XPS) was employed to investigate the composition of elements on the surface of CNTs@FeOOH nanocomposites and the chemical states of various bond elements. The result shown the presence of Fe, O, and C atoms at the CNTs@FeOOH sheet surface (**Figure S3a**). The binding energy peak observed in the high-resolution C 1s profile at 284.9 eV (**Figure S3b**) can be attributed to graphite carbon in carbon nanotubes. The peak at 285.4 eV belongs to epoxy and hydroxyl (Beamson et al., 1994). The other two peaks are caused by carbonyl (C=O, 288.7 eV) and the oxygenated carbons of carboxyl (O-C=O, 291.2 eV) (Gardella et al., 1986; Kokai, 1990). These peaks reveal the existence of oxygencontaining functional groups on the surface of CNTs (Zhang et al., 2017). Meanwhile, the finescanned Fe 2p XPS spectra of that sample was also shown in **Figure S3c**, and the Fe 2p3/2 and Fe 2p1/2 peaks located at 711.0 and 724.9 eV could be indexed to Fe3<sup>+</sup> and Fe2+, respectively, and satellite peaks at 719.4 and 733.9 eV correspond well with FeOOH (Tan et al., 1990). In addition, the O 1s peaks (**Figure 1f**) can be assigned to Fe-O-Fe (529.5 eV), Fe-O-H (530.1 eV), and H-O-H bonds (531.2 eV). It is worthwhile to point out that the peak at 531.2 eV is attributed to defect sites with low oxygen coordination (Zhang et al., 2019). According to previous reports, surface oxygen vacancies are involved in the LiPS transformation reaction, which significantly improves the kinetics of the reaction, thus contributing to the fast LiPS transformation at high rate.

According to the previous paper, the shuttle effect of polysulfides can be diminished with an additional intermediate layer (Fan et al., 2019). In this work, various composites of CNTs@FeOOH were coated on the surface of the PP separator as a functional interlayer to study their effect on Li-S batteries. **Figure S4a** shows a schematic representation of conventional PP-separator Li-S structures and advanced Li-S batteries with functional CNTs@FeOOH layers. The surface morphology of the modified interlayer is shown in **Figures 2a–c**, the preserved porous structure will facilitate electrolyte penetration and lithium ions (Li+) transfer. **Figure 2d** and **Figures S4b,c** show the crosssectional appearance, and the thickness of the interlayer is about 20 um.

Coin-typed cells with different separators were assembled to evaluate electrochemical performances. **Figure S4d** exhibits the cycling performance of all Li-S cells with PP separator, CNTs, pure FeOOH and three proportional composites interlayer at 0.2C (1C = 1,675 mA g−<sup>1</sup> ) between 1.7 and 2.8 V. After 100 cycles, CNTs@FeOOH-II revealed the best cyclability of 662.1 mAh g−<sup>1</sup> , which indicates a good synergistic effect between CNTs and FeOOH.

In order to explore the lithium diffusion properties and investigate the role of composite materials in Li-S batteries, we performed cyclic voltammetry (CV) measurements under various scanning rates ranging from 0.1 to 0.5 mV s−<sup>1</sup> between 1.7 and 2.8 V (vs. Li/Li+). As shown in **Figures 3A–D**, all curves show typical reduction/oxidation reaction of S cathode, with two

distinct cathode peaks and one anode peak. The cathode peak at about 2.3 V corresponds to the transformation of sulfur bonding with Li<sup>+</sup> into soluble long-chain polysulfide [Li2S<sup>x</sup> (x = 4–8)]. Furthermore, the cathode peak around 2.0 V corresponds to the transformation of long-chain polysulfide into insoluble Li2S or Li2S<sup>2</sup> (Chung et al., 2016). In subsequent anode scanning, the oxidation peak at ∼2.4 V corresponds to its reverse process. According to the relationship between CV scanning rate (ν 0.5) and peak current (Ip), the lithium diffusion performance can be estimated using the classical Randles Sevcik equation:

$$Ip = (2.69 \times 10^{5}) n^{1.5} \text{SD}\_{\text{Li}+}{}^{0.5} \text{C}\_{\text{Li}} \upsilon^{0.5} \tag{1}$$

where, Ip-peak current (A), n-number of electrons per type of reaction (n = 1), S-electrode area (S = 1.13 cm<sup>2</sup> ), DLi+-diffusion coefficient of lithium ion (cm<sup>2</sup> s −1 ), CLi+-initial concentration of lithium ion in the cathode (CLi<sup>+</sup> = 1 mol cm−<sup>3</sup> ), ν-potential scanning rate (V s−<sup>1</sup> ) (Tao et al., 2016). The n, S, and CLi<sup>+</sup> are constant in our battery system. The slope of the curve in **Figure S5a** is positively correlated with the corresponding Li<sup>+</sup> diffusion. The calculated results show that the modified CNTs@FeOOH-II had the greatest diffusion capacity of Li 4.40 × 10−12, better than intact (2.08 × 10−12). Typically, the PP separator has difficulty catching soluble LiPS, which tends to dissolve in electrolytes in large quantities. As a result increasing viscosity of electrolyte leads to slower diffusion of Li+. On the contrary, since CNTs@FeOOH-II material can accelerate the conversion of LiPS to Li2S<sup>2</sup> or Li2S, it is easier to increase the diffusion rate of Li<sup>+</sup> in modified cells.

The discharge process in Li-S batteries can be expressed as follows:

$$\text{S}\_8 + \text{2e}^- \rightsquigarrow \text{S}\_8^{2-} \tag{2}$$

$$\text{^3S}\_8 + 2\text{e}^- \leftrightarrow 4\text{S}\_6^{2-} \tag{3}$$

$$\text{^2S}\_6^{2-} + \text{2e}^- \leftrightarrow \text{^3S}\_4^{2-} \tag{4}$$

$$\rm{^1S\_4}^{2-} + 4\rm{Li^+} + 2e^- \leftrightarrow 2\rm{Li\_2S\_2} \tag{5}$$

$$\rm Li\_2S\_2 + 2Li^+ + 2e^- \leftrightarrow 2Li\_2S \tag{6}$$

The theoretical discharge capacity of Li-S battery at different stages is calculated by referring the number of electrons transferred. The details are given in **Table 1** and specific formulas are given below (Diao et al., 2013):

$$\mathbf{q} = \mathbf{n} \mathbf{F}/\mathbf{M} \tag{7}$$

Among them, q is the specific discharge capacity, mAh g−<sup>1</sup> ; n is the number of transfer electrons per mole mass, mol−<sup>1</sup> ; F is the amount of electricity owned by 1 M electrons, 26.8 Ah; M is the molar mass of elemental sulfur, 32 g mol−<sup>1</sup> (Diao et al., 2013). Here, the discharge capacity of S<sup>8</sup> → S 2− 4 is recorded as S1, and that of S<sup>8</sup> → Li2S as S2. Accordingly, S<sup>1</sup> : S<sup>2</sup> is approach to 1:3. **Figures 4A–D** shows the discharge curve of the PP separator and various barrier interlayers circulating for TABLE 1 | The relationship between DOD and the discharge specific capacity.


200 cycles at 2C. The calculated results show that the capacity ratio of CNTs@FeOOH-II (1:2.44) is closest to the theoretical value, higher than 1:2.26, 1:2.27, and 1:1.74 of PP separator, CNTs and FeOOH, respectively, which indicates the enhanced transformation ability of CNTs@FeOOH-II to polysulfide ions at high current density.

**Figure 4E** compares the cyclic characteristics of the batteries with PP separator, CNTs, pure phase FeOOH and CNTs@FeOOH-II separator in a voltage range of 1.7–2.8 V at 2C. The initial discharge capacities are 870.6, 1,093, 1326.4, and 1121.9 mAh g−<sup>1</sup> , respectively, show that CNTs@FeOOH-II interlayer can strengthen the utilization of S upon cycling. After 350 cycles, CNTs@FeOOH-II maintained a high reversible capacity of 556 mAh g−<sup>1</sup> . In detail, although CNTs have good electrical conductivity, the weak van der Waals interactions between polar LiPS and non-polar carbon materials results in slow release of S-active substances from carbon materials and obvious capacity decay during long cycle (Song et al., 2016). Simultaneously, it is remarkable that the highest reversible capacity of 1326.4 mAh g−<sup>1</sup> can be obtained for pure FeOOH group. This can be explained by the presence of oxygen vacancies in FeOOH, which makes it electrocatalytic and prompts the rapid conversion of long-chain LiPS to solid Li2S<sup>2</sup> and Li2S. Since 75% of the discharge capacity (1,254 mAh g−<sup>1</sup> ) occurs in this conversion process, an enhanced reaction kinetics is beneficial for increasing the reversible capacity (Lim et al., 2019). However, since previous studies have shown that the absorption of LiPS by insulating interlayer is considered to be a "death zone" without transferring electrons during cycling (Hao et al., 2017). The fading trend in the battery with FeOOH interlayer mainly result from its low conductivity (10−<sup>5</sup> S cm−<sup>1</sup> ). Compared with relevant studies (**Table S1**), the introduction of this functional interlayer delivered better performance improvement for Li-S batteries.

The existence of functional interlayer can be used both as a conductive top current collector and as a physical barrier to polysulfide diffusion and lithium (Li) metal, debase the corrosion of Li metal. In case of rate performance (**Figure S5b**), CNTs@FeOOH-II composites exhibit high discharge capacity of 1292.6, 957.5, 802.3, and 630.8 mAh g−<sup>1</sup> at various rates from 0.2 to 2C, which is more satisfactory than CNTs and FeOOH. In particular, when the rate was restored to 0.2C, the specific capacity of the battery returned to 972.5 mAh g−<sup>1</sup> . These results confirm that CNTs@FeOOH-II interlayer enhances the stability

mAh g−<sup>1</sup> ; (E) cycle performance at 3,200 mAh g−<sup>1</sup> .

of S electrode. The presence of FeOOH in CNTs@FeOOH-II composite can enhance the rapid transformation of polysulfide ions and the hysteretic conversion kinetics of Li-S batteries, thus improving the rate capability of Li-S batteries.

The electrocatalytic effects of the different interlayers on PP separator were studied by electrochemical impedance spectroscopy (EIS). In **Figure 5** shows the Nyquist plots when discharges up to 2.1 V after 10, 30, 50 cycles, respectively. Each plot consists of one oblique line in a low frequency region and one or two compressed semicircles in a medium and high frequency region. The corresponding equivalent circuit model is shown in **Figure 5A**. In the equivalent circuit, R<sup>s</sup> represents the ohmic resistance of the reaction system; R<sup>f</sup> is related to the resistance of the solid electrolyte interface (SEI), corresponding to a semicircle of the high frequency region; Rct represents the charge transfer resistance, corresponding to a semicircle of the mid-frequency region, the diameter of the semicircle is the size of Rct, the larger the diameter, the greater the impedance, the more unfavorable to the high performance; CPE-double layer electrode/electrolyte capacitance; W characterizes the Warburg diffusion impedance of the electrode, which corresponds to an oblique line in the low frequency band, it characterizes the diffusion rate of Li<sup>+</sup> in the

TABLE 2 | The comparison of Rct values of different interlayer and PP separator.


material, the larger the slope, the better the high performance (Hu et al., 2018).

The fitting values of Rct were exhibited in **Table 2**, the S electrode with CNTs@FeOOH-II interlayer endowed the lowest Rct value after several charge-discharge cycles. The results are in good accordance with the results of the electrochemical cycle test. The conductivity of CNTs@FeOOH-II was measured by four-point probe method, which is about 4.6 S cm−<sup>1</sup> . High conductivity of CNTs@FeOOH-II may accelerate electron transfer and reduce electrochemical polarization. This further indicates that electrocatalytic materials with high conductivity and faster electron and ion transfer rates can improve the electrochemical performance of Li-S batteries.

#### CONCLUSION

In conclusion, FeOOH combined with CNTs with excellent catalytic ability is applied for high performance Li-S batteries. Compared to the pristine samples, the modified battery exhibited a good performance of 556 mAh g−<sup>1</sup> at 3,200 mA g−<sup>1</sup> for 350 cycles. CNTs@FeOOH-II plays the following main roles: (i) oxygen vacancies in FeOOH promote the rapid transformation of polysulfide ions, thus enhancing the reaction kinetics; (ii) the presence of FeOOH can effectively adsorb soluble polysulfides, which sluggishs the further diffusion to the anode; (iii) CNTs@FeOOH-II with high electric conductivity can be used as a "vice-electrode" to accelerate electron transfer and thus improve the rate capability. Therefore, for a high performance of Li-S battery, it is necessary to consider the high conductivity, adsorption, and fast conversion of LiPS.

#### DATA AVAILABILITY STATEMENT

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

### AUTHOR CONTRIBUTIONS

YL, XL, and YH contributed conception and design of the study. YL organized the database, performed the statistical analysis, and wrote the first draft of the manuscript. All authors contributed to manuscript revision, read, and approved the submitted version.

#### FUNDING

This research was supported by the Natural Science Basic Research Plan in Shaanxi Province of China (2019JLP-04), the National Natural Science Foundation of China (51672189), Xi'an Science and Technology Project of China [201805037YD15CG21(20)], and Tianjin Science and Technology Project (18PTZWHZ00020).

## REFERENCES


## SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2020.00309/full#supplementary-material


**Conflict of Interest:** LK, ZT, LS, and CZ were employed by the company Shaanxi Coal Chemical Industry Technology Research 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 © 2020 Li, Li, Hao, Kakimov, Li, Sun, Kou, Tian, Shao, Zhang, Zhang and Sun. 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.

# Biomass-Derived P/N-Co-Doped Carbon Nanosheets Encapsulate Cu3P Nanoparticles as High-Performance Anode Materials for Sodium–Ion Batteries

Yanyou Yin<sup>1</sup> , Yu Zhang<sup>1</sup> , Nannan Liu<sup>1</sup> , Bing Sun<sup>2</sup> \* and Naiqing Zhang1,3 \*

*<sup>1</sup> State Key Laboratory of Urban Water Resource and Environment, Harbin Institute of Technology, Harbin, China, <sup>2</sup> Faculty of Science, Center for Clean Energy Technology, School of Mathematical and Physical Science, University of Technology Sydney, Sydney, NSW, Australia, <sup>3</sup> Harbin Institute of Technology, Academy of Fundamental and Interdisciplinary Sciences, Harbin, China*

#### Edited by:

*Weijie Li, University of Wollongong, Australia*

#### Reviewed by:

*Xiaodong Guo, Sichuan University, China Dongjiang Yang, Qingdao University, China*

#### \*Correspondence:

*Bing Sun bing.sun@uts.edu.au Naiqing Zhang znqmww@163.com*

#### Specialty section:

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

Received: *17 February 2020* Accepted: *30 March 2020* Published: *05 May 2020*

#### Citation:

*Yin Y, Zhang Y, Liu N, Sun B and Zhang N (2020) Biomass-Derived P/N-Co-Doped Carbon Nanosheets Encapsulate Cu*3*P Nanoparticles as High-Performance Anode Materials for Sodium–Ion Batteries. Front. Chem. 8:316. doi: 10.3389/fchem.2020.00316* Biomass-derived approaches have been accepted as a practical way for the design of transitional metal phosphides confined by carbon matrix (TMPs@C) as energy storage materials. Herein, we successfully synthesize P/N-co-doped carbon nanosheets encapsulating Cu3P nanoparticles (Cu3P@P/N-C) by a feasible aqueous reaction followed by a phosphorization procedure using sodium alginate as the biomass carbon source. Cu-alginate hydrogel balls can be squeezed into two-dimensional (2D) nanosheets through a freeze–drying process. Then, Cu3P@P/N-C was obtained after the phosphorization procedure. This rationally designed structure not only improved the kinetics of ion/electron transportation but also buffered the volume expansion of Cu3P nanoparticles during the continuous charge and discharge processes. In addition, the 2D P/N co-doped carbon nanosheets can also serve as a conductive matrix, which can enhance the electronic conductivity of the whole electrode as well as provide rapid channels for electron/ion diffusion. Thus, when applied as anode materials for sodium-ion batteries, it exhibited remarkable cycling stability and rate performance. Prominently, Cu3P@P/N-C demonstrated an outstanding reversible capacity of 209.3 mAh g−<sup>1</sup> at 1 A g <sup>−</sup><sup>1</sup> after 1,000 cycles. Besides, it still maintained a superior specific capacity of 118.2 mAh g−<sup>1</sup> after 2,000 cycles, even at a high current density of 5 A g−<sup>1</sup> .

Keywords: sodium-ion batteries, biomass, Cu3P, P/N-co-doped carbon, nanosheets

## INTRODUCTION

In recent years, lithium-ion batteries (LIBs) have been widely applied from portable electronic devices to electric vehicles (Zou et al., 2017, 2019; Qiu et al., 2019). However, the shortage of lithium sources limited the further development of LIBs (Wang et al., 2018a; Wu C. et al., 2018). Over the past few years, sodium-ion batteries (SIBs), owing to their low cost of production and earth-abundant sodium sources, show tremendous potential as a promising replacement of LIBs for large-scale energy storage applications (Kundu et al., 2015; Larcher and Tarascon, 2015; Zhang et al., 2016, 2018; Song et al., 2018; Xiao et al., 2018, 2020). However, when compared with LIBs, SIBs are still an immature technology that is confronted with a lot of challenges, such as low specific energy, poor cycleability, and low power density (Wessells et al., 2011; Lotfabad et al., 2014; Li and Zhou, 2018). Anode, as the main component of SIBs, has a great influence on the overall electrochemical performance. In recent years, red phosphorus has been regarded as one of promising SIB anode materials owing to its comparatively low redox potential (∼0.4 V vs. Na/Na+) and extremely high theoretical specific capacity (2,596 mAh g−<sup>1</sup> ) (Kim et al., 2013; Qian et al., 2013; Zhou et al., 2017; Hu et al., 2018; Wu Y. et al., 2018). However, the low electrical conductivity (∼10−<sup>14</sup> S cm−<sup>1</sup> ) and the huge volume expansion (∼490%) during the continuous Na<sup>+</sup> insertion/extraction process make red phosphorus suffer from inferior cycling stability and rate performance (Sun et al., 2014; Wang et al., 2018b). Fortunately, forming transition metal phosphides (TMPs) by combining red phosphorus with conductive transition metals has been proven to be an efficient way to enhance the electronic conductivity and reduce the volume change of phosphorus-based anode materials (Fullenwarth et al., 2014; Pramanik et al., 2015; Fan et al., 2016;

Wang X. et al., 2017; Zhang et al., 2017; Liu et al., 2018). Transitional metal phosphides (TMPs, M = Fe, Cu, Co, etc.) have drawn tremendous attention because of their high specific capacity and safe operating potential (Kim et al., 2013; Qian et al., 2013; Sun et al., 2014; Zhou et al., 2017; Hu et al., 2018; Wu Y. et al., 2018). Particularly, copper phosphide-based anode materials for SIBs have a low reduction potential (0.015–0.4 V vs. Na+/Na) and comparatively high specific capacity (Fan et al., 2016; Kong et al., 2018). However, similar to other conversiontype SIB anode materials, the volume expansion during the charge/discharge process has not been entirely overcome, and its diffusion kinetics is comparatively weak (Ge et al., 2017; Miao et al., 2017; Wang J. et al., 2017).

Fortunately, several effective strategies revealed promising potential ability in boosting the sodium storage performance of TMP anode materials. For example, constructing nanostructured materials, such as nanospheres and nanoparticles, can not only improve the reaction kinetics by shortening the diffusion distance of Na ions within the solid state but also relieve the mechanical strain generated by the large volume change during the conversion reaction (Xu et al., 2012; Ma et al., 2018). In addition, combining TMPs with conductive carbon matrices can also buffer the huge volume expansion and enhance the electronic conductivity of the electrode, thus resulting in better sodium storage performance (Qian et al., 2014; Li et al., 2017; Zhang et al., 2019). For instance, Cu3P/reduced graphene oxide nanocomposite synthesized by Tong et al. showed superior cycling stability and good rate capability (Liu et al., 2016). The carbon-confined Cu3P nanoparticles prepared by Zhou et al. exhibited superior cycling stability with a high capacity of 159 mAh g−<sup>1</sup> at 1.0 A g−<sup>1</sup> over 100 cycles (Kong et al., 2018). Although much progress has been achieved by reducing the particle size of Cu3P as well as introducing conductive carbon, it still remains a major challenge to develop a scalable and inexpensive method for practical application.

Biomass-derived carbon, profiting from its economy, environmental benignity, and sustainability, has attracted increasing attention (Moreno et al., 2014; Wu et al., 2017). Among them, sodium alginate has high availability and biodegradability, is non-toxic, and of low price (Comaposada et al., 2015). In particular, sodium alginate can cross-link with di- or trivalent ions (Marcos et al., 2016), thus making it a proper carbon source to synthesize TMP anode materials combined with conductive carbon matrices.

Herein, we present the preparation of Cu3P nanoparticles encapsulated in P/N-co-doped carbon nanosheets (Cu3P@P/N-C) through a feasible aqueous reaction followed by a phosphorization procedure. Cu3P nanoparticles are welldispersed and encapsulated in two-dimensional (2D) carbon nanosheets, which can not only buffer the large volume expansion but also prevent the agglomeration of Cu3P nanoparticles, thereby maintaining the integrity of the whole electrode. Furthermore, the 2D carbon nanosheet structure can shorten the Na<sup>+</sup> diffusion path, provide more active sites of Na+, as well as enhance the electronic conductivity of the entire electrode. Benefiting from these advantages mentioned above, Cu3P@P/N-C exhibited a long cycle life and outstanding rate performance when applied as anode for SIBs. Cu3P@P/N-C anode materials demonstrated a long cycle life (209.3 mAh g−<sup>1</sup> at 1 A g−<sup>1</sup> after 1,000 cycles) and excellent rate performance (118.2 mAh g−<sup>1</sup> even at a high current density of 5 A g−<sup>1</sup> after 2,000 cycles).

#### EXPERIMENTAL SECTION

#### Materials

All materials in the experiment were used without further purification. Cu(NO3)2·3H2O (ACS, 98.0–102.0%) was purchased from Aladdin. Sodium alginate (AR) was purchased from Aladdin. Red phosphorus (AR) was purchased from Kermel. Argon gases were supplied in cylinders by Qinghuaqiti with 99.999% purity.

## Preparation of Cu-Alginate Gel

Sodium alginate (1.0 g) was dispersed in 50 ml distilled water to form an aqueous solution. Cu(NO3)2·3H2O (2.0 g) was also dispersed in 50 ml distilled water to form an aqueous solution. Then, the sodium alginate aqueous solution was dropped slowly by a disposable plastic dropper into Cu(NO3)2·3H2O aqueous solution to form Cu-alginate hydrogel under magnetic stirring at room temperature. The obtained hydrogel was separated from the solution after 6 h and washed with deionized water several times. The as-prepared Cu-alginate hydrogel was frozen by liquid nitrogen and then dried through freeze–drying for 24 h to obtain Cu-alginate aerogel.

## Preparation of Cu3P@P/N-C

In a typical synthesis of Cu3P@P/N-C, Cu-aerogel was kept at 300◦C for 1 h at a heating rate of 2◦C min−<sup>1</sup> under air atmosphere. After cooling to room temperature, the CuO@C nanosheets aerogel was collected (**Figure S1**). One hundred milligrams of the obtained CuO@C nanosheet aerogel and 200 mg red phosphorus were put into two separate ceramic boats in a tube furnace and then were heated to 800◦C for 2 h under Ar atmosphere with a heating rate of 5◦C min−<sup>1</sup> to produce Cu3P@P/N-C. After cooling to room temperature, the sample was washed with deionized water three times by centrifugation and dried at 60◦C in vacuum oven for 12 h.

## Morphology and Structural Characterization

The morphology of the obtained samples was characterized by a field-emission scanning electron microscope (Hitachi Limited SU-8010) and transmission electron microscopy (JEOL-2100FS). The X-ray diffraction (XRD) pattern was determined by PANalytical X'Pert PRO (PANalytical X'Pert PRO, monochromated Cu Kα radiation 40 mA, 40 kV) to characterize the crystal structure. X-ray photoelectron spectroscopy (XPS) was performed with a Thermo Fisher Scientific K-Alpha (Fisher Scientific Ltd., Nepean, ON).

### Electrochemical Measurements

The anode slurry was prepared by mixing 70 wt% active materials, 20 wt% Super-P, and 10 wt% polyvinylidene fluoride (PVDF) by a high-speed electric agitator for 12 h. The slurry was pressed onto a cleaned copper foil by a doctor-balding method and dried in a vacuum oven at 80◦C for 12 h. The performance of the SIBs was tested using standard 2032-type coin cells in an argon-filled glove box. The separator was glass fiber (GF/D) from Whatman, and sodium foils were used as the counter and reference electrodes. The electrolyte was 1.0 M NaClO<sup>4</sup> in diethylene glycol dimethyl ether (Diglyme). The active material loading of the electrode was 0.5–0.8 mg cm−<sup>2</sup> . The cells were galvanostatically charged and discharged over a cutoff voltage window of 0.01–3.00 V at room temperature on a battery test system (Shenzhen Neware Electronic Co., China). Cyclic voltammetry behavior was studied by the CHI 650d electrochemical workstation at a scan rate of 0.1 mV s−<sup>1</sup> .

## RESULTS AND DISCUSSION

The carbon source we chose is sodium alginate. Sodium alginate is a natural polysaccharide extracted from brown seaweeds and some kinds of bacteria, which consists of a linear copolymer of (1–4)-linked β-D-mannuronic acid (M) and α-L-guluronic acid (G) in alternating blocks. Sodium alginate has high availability and biodegradability and low price. Its aqueous solution has a high viscosity and is non-toxic, which makes it widely useful as food thickeners, stabilizers, emulsifiers, etc. (Comaposada et al., 2015; Zou et al., 2018). In particular, sodium alginate can cross-link with di- or trivalent ions to form the uniform, transparent, water-insoluble, and thermo-irreversible gels at room temperature (Marcos et al., 2016). **Scheme 1** exhibits the typical synthesis of Cu3P@P/N-C. Sodium alginate aqueous solution was added dropwise into Cu(NO3)2·3H2O aqueous solution to form Cu-alginate hydrogel balls. During the freezing process by liquid nitrogen, subsequently, the growth of the ice crystals squeezed the Cu-alginate macromolecules into 2D nanosheets. Then, the sample was dried by freeze drying. In the end, after the phosphorization procedure, Cu3P@P/N-C was obtained.

As shown in **Figure 1a**, the crystal structure and composition of the as-prepared samples were first tested by XRD measurements. There are several diffraction peaks in the XRD pattern of Cu3P@P/N-C at 36.0, 39.1, 41.6, 45.1, 46.1, and 47.3◦ . These peaks can be indexed to the (112), (202), (211), (300), (113), and (212) lattice planes of Cu3P crystalline (PDF#71-2261), matching well with a formerly reported study (Wang R. et al., 2018). In addition, there is a broad peak at around 24.7◦ , corresponding to the (002) plane of amorphous carbon. There are no other crystalline phases observed, which suggests that the as-prepared Cu3P@P/N-C has high purity.

The morphologies of the Cu-alginate aerogel and Cu3P@P/N-C were investigated by scanning electron microscopy (SEM). **Figure S2** exhibits the typical 2D nanosheet morphology of Cu-alginate aerogel. And as shown in **Figures 1b–d**, the 2D nanosheet morphology remained well after the phosphorization procedure. **Figure 1e** displays the typical energy-dispersive X-ray (EDX) mappings of Cu3P@P/N-C, where P, N, Cu, and C elements were observed, suggesting the homogeneous distribution of these several elements. The distribution of P element was uniform, indicating that P not only came from Cu3P but also doped in the carbon nanosheets. The typical 2D nanosheet morphology was also confirmed by the transmission electron microscopy (TEM) image of Cu3P@P/N-C (**Figures 1f, g**). The Cu3P nanoparticles with sizes of around 30 nm distributed uniformly as well as encapsulated in P/N-codoped carbon nanosheets, which can not only provide rapid diffusion channels for the electron/ion but also prevent Cu3P nanoparticles from agglomeration. **Figure 1h** shows the highresolution transmission electron microscopy (HRTEM) image in which the lattice fringes can be observed distinctly. The distance of these lattice fringes is 0.20 nm, matching well with the Cu3P crystalline (300) lattice plane (Liu et al., 2016), verifying the existence of the Cu3P nanoparticles.

Furthermore, the X-ray photoelectron energy spectra (XPS) measurement was applied to study the detailed chemical states of Cu3P@P/N-C. As shown in **Figure 2A**, the signals of the C, N, O, P, and Cu elements were observed obviously without other impurities. The element O may come from the absorbed O species in the air. The N element came from the nitrate radical of Cu(NO3)2·3H2O, which was not completely removed during the washing process. The C 1s spectrum in **Figure 2B** could be attributed to three peaks at 284.6, 286.3, and 289.1 eV. The major peak at approximately 284.6 eV was attributed to graphitic carbon. The other two peaks at 286.3 and 289.1 eV were fitted by carbon bonding with phosphorus and nitrogen, respectively, which can manifest the co-doping of both P and N atoms into the carbon nanosheets. The P2p spectrum in **Figure 2C** indicates the P chemical states in Cu3P@P/N-C. The P2p spectrum could be fitted into four peaks at 129.9, 134.2, 135.1, and 135.6 eV. The peak at 129.9 eV was attributed to the P in Cu3P. The two peaks at 135.1 and 135.6 eV were ascribed, respectively, to P– O and P=O bonds. And the peak at 134.2 eV could be ascribed to the C–P bond, corresponding to the C 1s bonding peak at 286.3 eV. The N 1s spectrum (**Figure 2D**) can be fitted into three component peaks at 399.4, 401.8, and 402.3 eV, which can be assigned to pyridinic nitrogen, pyrrolic nitrogen, and graphitic nitrogen, respectively. This could further confirm that both P and N are doped into the as-prepared carbon nanosheets.

The electrochemical performance of Cu3P@P/N-C was measured in detail as anode materials for SIBs. First, its electrochemical reaction was studied by cyclic voltammetry (CV) measurements from 0.01 to 3 V at a scan rate of 0.1 mV s−<sup>1</sup> (**Figure 3A**). In the first cathodic process of the CV curves of Cu3P@P/N-C, there was a reduction peak at around 1.1 V, which can be associated with side reactions [the irreversible decomposition of the electrolyte and the formation of the solid electrolyte interface (SEI) layer on the electrode surface] (Zhu et al., 2019). Then, a strong reduction peak appeared between

FIGURE 1 | (a) X-ray diffraction (XRD) pattern of Cu3P@P/N-C. (b–d) SEM images, (e) energy-dispersive X-ray (EDX) mappings, (f,g) transmission electron microscopy (TEM) images, and (h) high-resolution TEM (HRTEM) image of Cu3P@P/N-C.

0.07 and 0.4 V, which could be ascribed to the reaction of Cu3P with sodium. In the anodic scan process, the reverse of this reaction formed a distinct oxidation peak at around 0.9 V, as the following equation with the release of sodium ions shows:

$$\text{Na}\_3\text{P} + \text{3Cu} \rightarrow \text{Cu}\_3\text{P} + \text{3Na}^+ \tag{1}$$

The CV profiles from the second to the fifth cycles matched well, indicating the highly reversible and stable cycling performance of Cu3P@P/N-C as SIB anode materials.

**Figure 3B** shows the galvanostatic cycling performance of the Cu3P@P/N-C electrode under 1.0 A g−<sup>1</sup> (0.01 and 3.0 V vs. Na+/Na). Pure Cu3P was also synthesized by the phosphorization of commercial Cu powder using red phosphorus as a P source for comparison (as shown in **Figures S3**–**S5**). Compared with pure Cu3P, the Cu3P@P/N-C electrode exhibited better electrochemical performance, maintaining an excellent reversible capacity of 209.3 mAh g−<sup>1</sup> at 1.0 A g−<sup>1</sup> after 1,000 cycles. The charge–discharge profiles of the initial three cycles of Cu3P@P/N-C are shown in **Figure S6**. The initial discharge and charge capacities were 1,029.6 and 562.3 mAh g−<sup>1</sup> ; thus, the initial coulombic efficiency was 54.6%. This capacity loss was mainly caused by the irreversible formation of the SEI layer. Additionally, to estimate the sodium storage performance of the Cu3P@P/N-C electrode at a high rate, a high current density at 5.0 A g−<sup>1</sup> was chosen. An outstanding cycling performance of 118.2 mAh g−<sup>1</sup> is maintained (**Figure 3E**), even after 2,000 cycles. As shown in **Table S1**, Cu3P@P/N-C exhibited superior cycling performance and excellent rate performance. In sharp contrast to Cu3P@P/N-C, pure Cu3P had rapid capacity loss, resulting in only 44.0 mAh g−<sup>1</sup> at 1.0 A g−<sup>1</sup> after 200 cycles and 30.6 mAh g −1 at 5.0 A g−<sup>1</sup> after 500 cycles. Thereby, the superior cycling performance of Cu3P@P/N-C reflects their structural stability as anode materials for SIBs. The SEM images and the EDX mappings of the selected area of Cu3P@P/N-C after 500 cycles are exhibited in **Figures S7**, **S8**. The typical nanosheet morphology of Cu3P@P/N-C can still be observed obviously, indicating its superior stability.

Electrochemical impedance spectroscopy (EIS) was investigated to further understand the cycling performance of Cu3P@P/N-C. As shown in **Figure 3D**, the semicircle in the high-frequency region corresponded to the charge transfer resistance of the electrode. The sloped line in the low-frequency region is attributed to the Warburg impendence. After the first cycle, the semicircle became much smaller, which could be explained by the formation of the SEI layer. The semicircles overlaid well after the second cycle, indicating the excellent cycling stability of Cu3P@P/N-C.

**Figure 3C** shows the rate performance of the Cu3P@P/N-C and pure Cu3P electrodes. The current densities were selected from 0.2 to 10 A g−<sup>1</sup> . The reversible specific capacities acquired were of 311.8, 250.3, 226.8, 207.2, 177.1, and 144.8 mAh g−<sup>1</sup> at the current densities of 0.2, 0.5, 1, 2, 5, and 10 A g−<sup>1</sup> , respectively. When the current density was set back to 0.2 A g −1 , the reversible specific capacity remained at 265.2 mAh g−<sup>1</sup> , suggesting the outstanding rate performance. On the contrary,

pure Cu3P showed reversible specific capacities of only 104.0, 59.1, 41.0, 33.4, 18.1, and 10.2 mAh g−<sup>1</sup> at the current densities of 0.2, 0.5, 1, 2, 5, and 10 A g−<sup>1</sup> , respectively, which were far inferior to those of Cu3P@P/N-C. The difference between these two materials could be ascribed to the introduction of P/Nco-doped carbon nanosheets, which could enhance the electron transfer during the discharge/charge process.

To further understand the electrochemical reaction process of the Cu3P@P/N-C electrode, we disassembled the cells for the ex situ XRD measurement at different states during the charge and discharge processes. As shown in **Figure 4**, the pristine Cu3P@P/N-C electrode only exhibited diffraction peaks of Cu3P. During the discharge process, the primary diffraction peak of Cu3P at around 45.1◦ corresponding to the (300) plane gradually receded and almost disappeared when completely discharged to 0.01 V. At the same time, a peak at about 43.5◦ came out, originating from the generation of Cu. In addition, a new peak came out at about 37◦ corresponding to the (103) plane of Na3P. During the charging process, the above composition change reversed. Such observations can identify the reversible sodiation/desodiation process of Cu3P conducting by conversion mechanism as in the following equation (Fan et al., 2016):

$$\text{Cu}\_3\text{P} + \text{ 3Na} \leftrightarrow \text{Na}\_3\text{P} + \text{ 3Cu} \tag{2}$$

This conversion mechanism was consistent with the CV measurements.

#### CONCLUSION

In summary, through a feasible aqueous reaction at room temperature followed by a phosphorization procedure, we have successfully fabricated biomass-derived P/N-co-doped

carbon nanosheets encapsulating Cu3P nanoparticles as highperformance anode materials for sodium-ion batteries. Given the help of the 2D P/N-co-doped carbon nanosheets, the Cu3P nanoparticles could be well-encapsulated, thus being prevented from agglomeration. When applied as anode materials, superior cycling stability and excellent rate performance were exhibited. Prominently, Cu3P@P/N-C exhibited an outstanding reversible capacity of 209.3 mAh g−<sup>1</sup> at 1 A g−<sup>1</sup> after 1,000 cycles. In particular, an excellent specific capacity of 118.2 mAh g−<sup>1</sup> could be maintained after 2,000 cycles, even at an ultrahigh current density of 5 A g−<sup>1</sup> . This remarkable electrochemical performance is mainly attributed to the rational design of the 2D P/N-co-doped carbon nanosheet structure, which could buffer the volume change of Cu3P nanoparticles as well as improve the electron/ion transport kinetics during the Na<sup>+</sup>

#### REFERENCES


insertion/extraction process. Additionally, the 2D P/N-co-doped carbon nanosheets also could serve as a conductive matrix, which could enhance the electronic conductivity of the electrode. This feasible and facile preparation approach could be further developed to the synthesis of a variety of 2D P/N-co-doped carbon nanosheets encapsulating TMP electrodes for energy storage devices.

## DATA AVAILABILITY STATEMENT

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

### AUTHOR CONTRIBUTIONS

YY and NZ contributed to the conception and design of the study. YY organized the database. YY, YZ, and NL performed the statistical analysis. YY wrote the manuscript. BS and NZ helped perform the analysis with constructive discussions. All authors contributed to manuscript revision, read, and approved the submitted version.

## FUNDING

This work was supported by the National Natural Science Foundation of China (no. 21646012), China Postdoctoral Science Foundation (nos. 2016M600253, 2017T100246), the State Key Laboratory of Urban Water Resource and Environment, Harbin Institute of Technology (no. 2019DX13), and the Fundamental Research Funds for the Central Universities (grant no. HIT. NSRIF. 201836). BS gratefully acknowledges financial support by the Australian Research Council (ARC) through the ARC Discovery Early Career Researcher Award (DE180100036).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2020.00316/full#supplementary-material


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

# Nitrogen-Doped Graphene via In-situ Alternating Voltage Electrochemical Exfoliation for Supercapacitor Application

Mingjun Jing1,2, Tianjing Wu1,2 \*, Yazheng Zhou<sup>2</sup> , Xilong Li <sup>2</sup> and Yong Liu<sup>1</sup> \*

*<sup>1</sup> State Key Laboratory of Powder Metallurgy, Central South University, Changsha, China, <sup>2</sup> Department of Chemistry, Xiangtan University, Xiangtan, China*

#### Edited by:

*Weihua Chen, Zhengzhou University, China*

#### Reviewed by:

*Liwei Mi, Zhongyuan University of Technology, China Xifei Li, Xi'an University of Technology, China Yufeng Zhao, Shanghai University, China*

#### \*Correspondence:

*Tianjing Wu twu@xtu.edu.cn Yong Liu yonliu@csu.edu.cn*

#### Specialty section:

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

Received: *07 February 2020* Accepted: *23 April 2020* Published: *04 June 2020*

#### Citation:

*Jing M, Wu T, Zhou Y, Li X and Liu Y (2020) Nitrogen-Doped Graphene via In-situ Alternating Voltage Electrochemical Exfoliation for Supercapacitor Application. Front. Chem. 8:428. doi: 10.3389/fchem.2020.00428* Doping heteroatom, an effective way to enhance the electrochemical performances of graphene, has received wide attention, especially related to nitrogen. Alternating voltage electrochemical exfoliation, as a low cost and green electrochemical approach, has been developed to construct *in-situ* N-doped graphene (N-Gh) material. The N-Gh presents a much higher capacity than that of pure graphene prepared via the same method, which might be attributed to the introduction of nitrogen, which has much more effects and a disordered structure. As-prepared N-Gh exhibits a low O/C ratio that is helpful in maintaining high electrical conductivity. And the effects and disorder structure are also conductive to reduce the overlaps of graphene layers. A symmetric supercapacitor assembled with N-Gh electrodes displays a satisfactory rate behavior and long cycling stability (92.3% retention after 5,000 cycles).

Keywords: N-doped graphene, alternating voltage, electrochemical exfoliation, electrochemical performances, supercapacitors

## INTRODUCTION

Graphene exhibits exceptional electronic conductive ability and carrier mobility due to its unique quantum Hall effect on a honeycomb sp<sup>2</sup> carbon lattice. Because of this, it became one of the most significant candidate materials for next-generation electronic and energy storage devices (Novoselov et al., 2004; Low et al., 2013; Gong et al., 2016). It is important to note that heteroatomdoped graphene might be better applied to supercapacitors through creating defects or embedding impurities. Among the various kinds of heteroatom-doped graphene materials, N atom is a general nominee because of its atomic size similarity to the carbon atom and unique valence electrons that generate a stable covalent bonds structure with adjacent C atoms (Low et al., 2013; Chaban and Prezhdo, 2015; Xu et al., 2018). Meanwhile, nitrogen atoms in N-doped graphene materials could become a redox active center, which might induce pseudocapacitance to increase the specific capacitance of materials (Luo et al., 2013; Yang et al., 2016). Hence, N-doped graphene or N-doped graphene-based composite materials are getting more and more attention.

In situ doping can be favorable for the formation of homogeneous doping (Qu et al., 2010; Yang et al., 2016). Some approaches have been developed to construct N-doped graphene. For example, N-graphene has been obtained via the chemical vapor deposition (CVD) method using a nitrogen-containing mixed gas (Bulusheva et al., 2017; Bu et al., 2018). Additionally, Ngraphene also can be formed through the segregation growth approach (Zhang et al., 2011). However, most of these methods usually require expensive devices, multistep transfer processes, or result in a low yield. At present, developing a green and low cost method to prepare mass production of N-doped graphene is still a major challenge.

There are several methods for graphene preparation, such as chemical vapor deposition (Suk et al., 2011), epitaxial growth (Yang W. et al., 2013), mechanical exfoliation (Yi and Shen, 2015), chemical exfoliation (Liu and Wang, 2011), electrochemical exfoliation (Yang et al., 2015; Bakunin et al., 2019), and so on. It is worth noting that electrochemical exfoliation has been deemed a useful technique in producing high-quality graphene on a large scale owing to it being environmentally friendly, low cost, and requiring only simple operations (Low et al., 2013; Ejigu et al., 2019). Two electrochemical types, cathodic and anodic methods, have been mainly performed in electrochemical exfoliation with graphite as a working electrode. On the one hand, cathodic exfoliation with the graphite material as a cathode usually takes place in organic solvents (Yang Y. et al., 2013; Taheri Najafabadi and Gyenge, 2015). This process typically needs some intercalates cations from the electrolyte, such as alkylammonium salts, ionic liquids, molten salts, and so on. On the other hand, anodic exfoliation is typically carried out in aqueous electrolytes with graphite as an anode (Parvez et al., 2014). The main issue of this method is the requirement of a high positive voltage (about a few tens of volts) in the electrochemical process, which might induce structural degradation and oxidation of the carbon lattice. Recently, a novel alternating voltage electrochemical exfoliation approach has been applied to prepare few-layer graphene flakes in aqueous electrolytes (Jing et al., 2015). Compared with the direct voltage exfoliation, the degree of oxidation of the carbon lattice can be reduced in the alternative redox process. And the two graphite electrodes are used as working electrodes during the alternating voltage process, which is conducive to improving the exfoliation efficiency.

Electrolyte solution is one of the key factors in all types of electrochemical exfoliation methods. Li salts as cathodic exfoliation electrolyte organic solution can release Li<sup>+</sup> ions that are reversibly intercalated into the inner spacing of graphite (Low et al., 2013). Aqueous H2SO<sup>4</sup> solution as anodic exfoliation electrolyte system can produce oxygen radicals (O·) and hydroxyl (OH·) to open boundaries, which is helpful in facilitating SO2<sup>−</sup> 4 intercalation, and then releasing SO<sup>2</sup> to expand the interlayer distance of graphite (Yang et al., 2015). Inorganic salts aqueous solutions (such as (NH4)2SO4) as anodic exfoliation electrolyte system shows a similar electrochemical mechanism in H2SO<sup>4</sup> solution, except for the existence of OH<sup>−</sup> ions at the edge sites and grain boundaries (Zabihi et al., 2019). Aqueous NaOH/H2O<sup>2</sup> solution has also been utilized during anodic electrochemical exfoliation, which could generate OH<sup>−</sup> and O2<sup>−</sup> 2 intercalation ions and appears to result in NaOH-induced electrochemical reduction of the oxygen functional groups of graphene (Rao et al., 2014). In addition, a range of reductive agents [such as sodium borohydride, (2,2,6,6-tetramethylpiperidin-1-yl]oxyl, ascorbic acid, and so on) as additives in electrolyte solution can improve the atomic ratio of C/O and control the exfoliation process (Rao et al., 2014). All these previous studies further indicate that the composition of electrolyte solution could mainly influence the functional groups, defects, atomic ratio of C/O, and yield of graphene. Based on the above analysis, in-situ nitrogen doping approaches might be achieved via adding nitrogen compounds into electrolyte solution during the electrochemical exfoliation of graphite. At present, few nitrogen compounds as additives (protic ionic liquid ethylammonium nitrate, ammonia, and natural biocompatible glycine) have been discussed to produce N-doped graphene in an anodic electrochemical exfoliation process (Usachov et al., 2011; Wang et al., 2012). But the related research is still poor, especially utilizing an alternating voltage electrochemical technique.

In this study, the alternating voltage electrochemical technique has been successfully applied to in-situ construct Ndoped graphene (N-Gh) on a large scale by adding ammonium chloride salt to NaOH aqueous solution. Compared with as-prepared pure graphene (Gh) utilizing the same process, the N-Gh sample presents a larger size and much more effects. And the electrochemical properties of N-Gh have been investigated in three-electrode and two-electrode systems. The N-Gh sample reveals a satisfactory rate behavior and long cycling stability.

## EXPERIMENTAL SECTION

## Synthesis of N-Doped Graphene Electrode Material

Alternating voltage electrochemical exfoliation was fabricated with a two-electrode system utilizing two graphite rods as working electrodes. N-doped graphene (N-Gh) was prepared in 3 M NaOH and 3 M NH4Cl mixed aqueous solution. Both graphite rods were exfoliated via 5.0 V alternating voltage (50 Hz, YK-BP81005 regulator transformer) for 5 h. Then the as-exfoliated substrate was separated, and further washed using distilled water until the pH value was close to 7. At last, the N-Gh sample was obtained via the freeze-dried method. Pure graphene (Gh) was also put in 3 M NaOH aqueous solution under the same preparation conditions.

## Materials Characterization

The phase character of materials was studied via the Xray diffractometer (XRD, Rigaku D/max 2550 VB+) from 10 to 80<sup>o</sup> at 5 ◦ min−<sup>1</sup> with Cu Kα radiation. The raman spectra of the as-prepared products were collected using a Raman spectrometer (HORIBA Labram HR Evolution). The morphology of the as-obtained materials were explored through scanning electron microscopy (SEM, JSM-6510LV) and transmission electron microscopy (TEM, JEM-2100F). Furthermore, the atomic arrangement was studied utilizing high-resolution transmission electron microscopy (HRTEM, JEM-2100F). Then, FT–IR spectrophotometer (AVTATAR, 370) was applied to test the surface functional groups

of materials using KBr as a reference. Thermogravimetric analysis (TGA, NETZSCH STA449F3) from 25 to 900◦ C was utilized to measure the thermostability of materials with a heating rate of 5◦ C min−<sup>1</sup> in air. Moreover, X-ray photoelectron Spectroscopy (XPS, ESCALab250) was tested to analyze the surface chemical composition of the asobtained samples with C1s photoelectron peak at 284.6 eV as the reference.

### Electrochemical Measurement

Active materials, super P, and polyvinylidene fluoride (PVDF) were mixed in N-methyl-2-pyrrolidone (NMP) with a mass ratio of 80:10:10 to form a coating slurry. Then, the asobtained slurry was pressed onto a round nickel foam current collector. Finally, the electrodes with a loading mass of about 1.5 mg cm−<sup>2</sup> were formed through drying at 50 ◦ C in a vacuum overnight and pressing under a 10 MPa pressure. A classic three-electrode electrochemical test system was utilized to investigate the electrochemical characteristics of the as-prepared materials. This test system was composed of a working electrode, platinum foil counter electrode, and Hg/HgO reference electrode. It is worth noting that the working electrodes should be soaked in 2 M KOH aqueous solution for 12 h before the electrochemical test. Moreover, the symmetric supercapacitor was equipped with two as-prepared working electrodes using 2 M KOH as an electrolyte solution and a glassy fibrous material as a separator. And the related calculation for the symmetric supercapacitor is based on the total mass of active material. Cyclic voltammetry (CV) curves were measured on MULTI AUTOLAB M204 (MAC90086) at various scanning rates. Electrochemical impedance measurements (EIS) were tested on a CHI 660B electrochemical working station with the frequency range between 100 and 0.01 Hz. Galvanostatic discharge/charge files were investigated at room temperature on Land CT2001A battery cycler.

## RESULTS AND DISCUSSION

## The Electrochemical Exfoliation via Alternating Voltage

The electrochemical processes of the as-prepared Gh and N-Gh materials via alternating voltage electrochemical exfoliation have been displayed in **Scheme 1**. On the basis of previous reports (Wang et al., 2014; Jing et al., 2015), the surface of graphite electrodes were alternately oxidized and reduced during the electrochemical process of alternating voltage. Meanwhile, some cations and anions in the electrolyte solution can intercalate the layers of graphite to accelerate stripping speed. In detail, some defects and oxygen-containing functional groups on the surface of the graphite electrode have been induced during the anodic process. Then, some oxidized graphite was reduced via a cathodic reaction. In NaOH solution, the Na<sup>+</sup> and OH<sup>−</sup> can intercalate into the graphite layers. Certain amounts of hydrogen gas can be produced during the electrochemical process, which would promote the exfoliation rate of graphite. For the N-Gh sample, the addition of NH<sup>+</sup> 4 and Cl<sup>−</sup> ions might be conducive to exfoliate the graphite electrode through much more intercalation. Moreover, the Cl<sup>−</sup> ions might be transformed into ClO<sup>−</sup> or Cl<sup>2</sup> during the electrochemical anodic process (Munuera et al., 2017). The related oxidation-reduction of NH<sup>+</sup> 4 also took place during alternate anodic and cathodic reactions, which could generate C-NH2, -C-NH-C, and C-N-C<sup>2</sup> functional groups. With the introduction of NH4Cl in the NaOH solution, the exfoliation

rate can be effectively enhanced via alternating the voltage electrochemical process.

## The Microstructure, Morphology, and Composition of Samples

The SEM has been utilized to analyze the morphologies of the as-prepared Gh and N-Gh, which is shown in **Figure 1**. As displayed in **Figures 1A,B**, the pure Gh sample presents thin flakes with various sizes (0.5–5 µm<sup>2</sup> ). In **Figures 1C,D**, the N-Gh displays porous thin sheets. It can also be seen that the size of N-Gh is much larger than that of pure Gh. Further, the characterization of the morphology has been measured using TEM, as shown in **Figures 2A,C**. These results reveal that the exfoliated Gh and N-Gh flakes typically feature some overlapping regions. Additionally, the HRTEM image of Gh (in **Figure 2B**) presents lattice spacing of 0.337 nm, corresponding to (002) plane of graphene (Yang et al., 2014). In **Figure 2C**, a few defect-free and disorder domains can also be found in the as-obtained pure Gh. While the N-Gh sample exhibits a much more disordered structure and obvious pore structure in **Figure 2D**. The N-Gh prepared by alternating voltage exfoliation presents with a larger size and more defects than those of pure graphene, which might be due to the fast stripping and N doping processes.

**Figure 3A** presents the XRD powder pattern of pure Gh and N-Gh. The sharp peak of pure Gh at 26.4◦ is indexed as (002) crystal plane of graphene (JCPDS Card no.41-1487). This sharp peak illustrates that the Gh maintains a high degree of crystallization and electrical conductivity (Xu et al., 2015). According to the results of contrasting the curves of pure Gh and N-Gh, the peak intensity of the N-Gh sample is obviously weaker, which is consistent with the HRTEM conclusion. The order/disorder structures and defects characterization of the asobtained materials have been further analyzed through Raman measurement, which is displayed in **Figure 3B**. The presence of G band at 1,577 cm−<sup>1</sup> is related to E2g symmetry phonon mode, corresponding to ordered in-plane sp<sup>2</sup> carbon atoms (Wang et al., 2017). The D band and D' shoulder band are at about 1,321 and 1,621 cm−<sup>1</sup> , respectively, which belong to the disorder in the carbon hexagons and edge carbons (Deng et al., 2011). Moreover, the I2D/I<sup>G</sup> and ID/I<sup>G</sup> values of samples can typically reflect the number of layers and the degree of disorder structure, respectively (Soin et al., 2017). The I2D/I<sup>G</sup> values of pure Gh and N-Gh are 61.9 and 62.7%, respectively, which could illustrate that the flakes of both samples have only a few layers. The ID/I<sup>G</sup> value of N-Gh is 0.71, which is higher than that of pure Gh (0.57). This result again reveals that much more defects and disorder structure appear during the exfoliation process for N-Gh samples.

Furthermore, the surface functional groups of the as-obtained pure Gh and N-Gh have been detected via FT-IR spectrum analysis, as is displayed in **Figure 3C**. The large peaks at 3,432 cm−<sup>1</sup> are related to O-H bending from H2O (Wang X. et al., 2018). The weak peaks from 1,639 to 1,579 cm−<sup>1</sup> can be indexed to sp<sup>2</sup> -hybridized C=C stretching in plane vibrations, which

reveals the presence of the π-conjugation structure (Wang X. et al., 2018; Lee et al., 2019). The weak peaks from 1,383 to 1,308 cm−<sup>1</sup> correspond to oxygen-containing functional groups (C-O, C-OH, C-O-C) which illustrate the existence of a few hydroxyl/phenolic/alkoxy groups on the surface of the exfoliated Gh and N-Gh samples (Lee et al., 2019). A small peak at 728 cm−<sup>1</sup> might correspond to C-N stretching (Islam et al., 2016), which indicates that N can be successfully doped in-situ during alternating voltage electrochemical exfoliation. In **Figure 3D,** the mass loss from 25 to 300◦ C of N-Gh might be mainly resulted from adsorbed water and coordinated water (Chen et al., 2020). Compared to the pure Gh, the TGA curve of N-Gh with the temperature from 300 to 700◦C displays a quick downward trend, which illustrates the existence of much more defects and disorder structure in the as-obtained N-Gh sample (Xu et al., 2015).

The XPS spectra of N-Gh are displayed in **Figure 4**. As shown in **Figure 4A**, the C, O, and N elements all lie in the asprepared N-Gh sample. The N atom content is 4.5% and the O/C

atom ratio in this sample is 0.09. Further, a dominant peak at 284.6 eV is shown in the high-resolution XPS spectrum of C 1s (**Figure 4B**), which is assigned to graphitic C=C species, and the other two weak peaks at 285.5 eV and 286.4 eV correspond to sp<sup>3</sup> carbons (C-OH/C-NH2) and oxygen carbons (C-O-C/C=O), respectively (Wang et al., 2014). The O 1s spectrum presents two peaks at 532.1 and 533.7 eV in **Figure 4C**, which are related to O=C and O-C, respectively (Hou et al., 2015; Bulusheva et al., 2017). A few oxygen-containing functional groups in the N-Gh sample could be helpful to relieve overlapping. The N 1s spectrum (in **Figure 4D**) has been fitted into four peaks at 399.3 eV, 400.5 eV, 401.7 eV, and 402.9 eV, which are assigned to pyridinic nitrogen (19.8%), pyrrolic nitrogen (54.2%), graphite nitrogen (11.9%), and C-NH<sup>2</sup> (14.1%), respectively (Lee et al., 2014; Hong et al., 2019). These results again confirm that N has been successfully doped in N-Gh, which might contribute to the improvement of electrochemical performances.

## The Electrochemical Properties of Samples Based on a Three-Electrode System

The electrochemical performances of pure Gh and N-Gh samples have been firstly explored using CV tests in 2 M KOH electrolyte solution with the voltage range from −1.0 to 0 V based on a threeelectrode system, which is shown in **Figure 5A**. It can be clearly seen that pure Gh and N-Gh samples display rectangular CV curves, suggesting obvious electric double-layer storage behaviors (Munuera et al., 2017). The curve area of N-Gh is much larger than that of the pure Gh sample, which indicates the doping of N can obviously improve the specific capacity. Furthermore, the CV measurements of N-Gh based on increased scan rates from 5 to 100 mV s−<sup>1</sup> have been studied, as is revealed in **Figure 5B**. The intensities of CV files increased with the increased scan rates, yet the shapes of curves remained broadly stable (Zhu et al., 2018; Tang et al., 2019). This result reveals that N-Gh might present good electrochemical reversibility.

Then, the charge-discharge files of the as-prepared N-Gh electrode in 2 M KOH electrolyte solution at various current densities are displayed in **Figure 6A**. Based on the specific capacitance formula (C<sup>s</sup> = It/m1V, F g−<sup>1</sup> ) (Wang Y. et al., 2018; Wei et al., 2019), the specific capacitances of N-Gh electrode at the current densities of 1, 2, 5, 10, 15, and 20 A g−<sup>1</sup> are 143.6, 129.1, 114.2, 103.2, 96.5, and 91.5 F g−<sup>1</sup> , respectively, with high coulombic efficiency around 100%, which is displayed in **Figure 6B**. Compared with the specific capacitance at 1 A g−<sup>1</sup> , the capacity retention rate is up to 63.7% even at 20 A g−<sup>1</sup> . This high rate behavior might be related to the existence defects and doping N in the N-Gh sample. Moreover, the EIS curves of pure Gh and N-Gh electrodes in 2 M KOH aqueous solution at their open voltages are shown in **Figure 6C**. Both samples present very small semicircles in the high frequency and high slope in

the low frequency region, suggesting the pure Gh and N-Gh electrodes display high electrical conductivity and ion diffusion. The cycling stability of N-Gh electrodes have been investigated, which is displayed in **Figure 6D**. After 3,000 charge/discharge cycles in KOH aqueous solution, the specific capacity retention of N-Gh is up to 93.1% at a current density of 2 A g−<sup>1</sup> . All these improved electrochemical performances could be derived from the structure of N-Gh sample with N doping, which has few effects and a disordered structure.

## The Electrochemical Performances of N-Gh Based on a Two-Electrode System

To better study the electrochemical capacity performances of the as-obtained N-Gh, the symmetric supercapacitor has been fabricated with two N-Gh electrodes. **Figure 7A** displays the CV curves of N-Gh//N-Gh symmetric supercapacitor under the potential voltage from 0 to 1 V at different scan rates. It can be clearly seen that the shape of files stays the same as the scan rate increases (from 5 to 100 mV s−<sup>1</sup> ), indicating the N-Gh//N-Gh symmetric supercapacitor might display good electrochemical reversibility. The charge/discharge files of N-Gh//N-Gh symmetric supercapacitor at the current densities between 0.5 and 10 A g−<sup>1</sup> are revealed in **Figure 7B**. The charge curves and discharge curves are almost symmetrical, revealing high columbic efficiency. Additionally, the specific capacities of the N-Gh//N-Gh symmetric supercapacitor have been calculated on the discharge files on the basis of the total mass of negative and positive electrode slices, as shown in **Figure 7C**. The specific capacities of the symmetric supercapacitor are 34.2, 31.2, 29.4, 26.7, and 24.2 F g−<sup>1</sup> at current densities of 0.5, 1, 2, 5, and 10 A g−<sup>1</sup> , respectively. The energy density of this symmetric supercapacitor is 4.76 Wh kg−<sup>1</sup> at a power density of 500 W kg−<sup>1</sup> , and the retention ratio is up to 68.7% at the power density of 10,000 W kg−<sup>1</sup> . Compared with some other N doped carbonbased materials (in Figure 7C) (Chang et al., 2013; Kang et al., 2013; Balaji et al., 2018; Du et al., 2018), the energy density and rate behavior of N-Gh//N-Gh is satisfactory. The cycling stability has been also investigated in **Figure 7D**, presenting with 92.3% retention of N-Gh//N-Gh initial specific capacity after 5000 cycles. The perfect rate behavior and cycling stability of the N-Gh//N-Gh symmetric supercapacitor further illustrates the asprepared N-Gh might be a promising material for various kinds of composites in supercapacitors.

## CONCLUSION

In summary, the N-Gh sample has been prepared through an in-situ alternating voltage electrochemical exfoliation technique with the introduction of NH4Cl in to NaOH aqueous solution. The N chemical states in the N-Gh sample mainly present pyrrolic nitrogen. Compared with the as-obtained pure Gh sample, the N-Gh shows a larger size, much more effects, and a disordered structure. Additionally, the related electrochemical behaviors have been investigated in a three-electrode aqueous solution system, indicating that N-Gh displays a much higher specific capacity than that of pure Gh. Moreover, it also displays a good cycling stability and high rate behavior with 63.7% of the capacity retention rate even at a current density of 20 A g −1 . All these good electrochemical characteristics of N-Gh could be ascribed to the doping N, the existence of effects, and disorder structure, which is conducive to producing faradaic pseudocapacitance and reducing overlapping layers of graphene. The results of the symmetric supercapacitor fabricated with two N-Gh electrodes further illustrate the satisfactory cycling stability with 92.3% retention of N-Gh//N-Gh initial specific capacity after 5,000 cycles. These insights illustrate that the N-Gh sample prepared via an in-situ alternating voltage approach could have promising applications to construct composites for enhanced supercapacitors.

## DATA AVAILABILITY STATEMENT

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

## REFERENCES


## AUTHOR CONTRIBUTIONS

MJ and TW designed and engineered the samples, performed the experiments, and wrote the paper. All authors contributed to performing the data analysis and general discussion.

### FUNDING

This work was supported by the National Natural Science Foundation of China (No. 21701044, 51625404), Postdoctoral Science Foundation Natural Science of China (No. 2019M662800), and Research Foundation of Education Bureau of Hunan Province, China (Grant no. 17A086).

## ACKNOWLEDGMENTS

The authors also acknowledge the assistance on XPS measurements received from the Instrumental Analysis Center of Central South University.


doi: 10.1016/j.carbon.2012.11.030

Appl. Mater. Interfaces 5, 11184–11193. doi: 10.1021/am403427h Munuera, J. M., Paredes, J. I., Enterría, M., Pagán, A., Villar-Rodil, S., Pereira, M. F. R., et al. (2017). Electrochemical exfoliation of graphite in aqueous sodium halide electrolytes toward low oxygen content graphene for energy and environmental applications. ACS Appl. Mater. Interfaces 9, 24085–24099. doi: 10.1021/acsami.7b04802

graphene flakes and their potential applications. Carbon N. Y. 54, 1–21.


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