Flour food waste derived activated carbon for high-performance supercapacitors

Abstract

It is of great importance to develop electrode materials at a low cost for constructing high-performance electrochemical energy storage systems. Herein, we employed flour food waste residue as a raw material to prepare porous carbon based supercapacitor electrodes through the carbonization and subsequent KOH activation process. The as-prepared biomass-derived activated carbon exhibits an interconnected porous network with numberous open micropores and appropriate mesopores. Meanwhile, its moderate total pore volume results in a high mass density of 0.86 g cm⁻³. When used as supercapacitor electrodes in an aqueous solution of KOH (6 mol L⁻¹), the porous carbon material shows a high specific capacitance (278 F g⁻¹, 241 F cm⁻³) and good cycling stability with 95% capacitance retention over 3000 cycles at a current density of 2 A g⁻¹. Furthermore, even at an ultrahigh current density of 100 A g⁻¹ (charge/discharge completed within 3 s), it still maintains a high specific capacitance of 142 F g⁻¹ (122 F cm⁻³).

---

Introduction

With the growing demand for hybrid electric vehicles and portable devices, it is urgent to construct an energy storage system with a high energy density and a power density. Supercapacitors have received considerable attention and been considered as promising energy-storage devices by virtue of their advantages in power density and cycling life.^1–4 According to the charge storage mechanisms, supercapacitors can be classified into two types, namely, electric double layer capacitors (EDLCs) and pseudocapacitors.^5,6 Pseudocapacitors store charges via faradic redox reaction, while EDLCs store energy in a non-faradic process through the electrostatic charge adsorption at the electrode/electrolyte interfaces.^7,8 Therefore, pore structure and surface area of the electrode materials are crucial factors for determining the performance of EDLCs.⁹ In the past decades, various carbon materials, like activated carbon,¹⁰ templated carbon,¹¹ carbon nanotubes¹² and graphene¹³ have been widely used in electrode materials for EDLCs. Among these, activated carbon is one of the most promising candidates and have been commercially available due to its large SSA, scalable synthesis procedure and especially low costs.^14,15 In recent years, recycling biomass waste like willow catkins,^16,17 pomelo peel,¹⁸ wheat bran,¹⁹ nutshell,²⁰ loofah sponge,²¹ potato residue²² to produce activated carbon-based supercapacitor electrodes have been quite attractive, since it helps to achieve green energy storage while protecting the environment. After physical or chemical activation, the biomass-derived activated carbons shows quite high SSA values.¹⁸ However, as supercapacitor electrode materials, their electrochemical performances including specific capacitance and high-rate capacity retention are not satisfying due to their unoptimized pore structures. The micropores are quite deep and not easily available to the electrolyte ions, resulting in low utilization rates of the surface area. Meanwhile, the shortage of mesoporous ion diffusion channels leads to poor high-rate performance.²³ Besides, the activated carbon materials with a high mesoporosity and macroporosity result in over high pore volume^20,23 and low mass density (≈0.5 g cm⁻³) of electrode material. Consequently, they often showed weak volumetric capacitance performance, which is a pivotal parameter for energy storage devices in practical applications.²⁴ Therefore, developing biomass-derived activated carbon with optimized pore structure is of great significance and quite challenging.

It is well known that wheat is one of the most important crops around the world. As the staple down-stream product of wheat, flour can be made into many kinds of main food, such as breads, cookies, Chinese steamed buns. At the same time, quantities of flour food waste were produced. This causes severe food waste and meanwhile environmental pollution problem. Thus, it is of great importance to take good utilization of such waste residue. As is well known, the leftover waste from flour food mainly contains starch and protein. Therefore, it is practical to utilize wheat based food waste as carbon source for preparing high performance activated carbon.

In this work, a biomass-derived nanosheet-like porous carbon was prepared by carbonization of flour food waste residue and subsequent KOH activation process. The mass ratios of carbon/KOH were adjusted to optimize the pore structure related to capacitive performance. The optimized activated carbon exhibits an interconnected porous network with a large quantity of open micropores and appropriate mesopores. More importantly, its moderate total pore volume results in a high mass density of 0.86 g cm⁻³. When used as supercapacitor electrodes in an aqueous solution of KOH (6 mol L⁻¹), the porous carbon material exhibits a high specific capacitance (278 F g⁻¹, 241 F cm⁻³), good cycling stability (95% capacitance retention after 3000 cycles), as well as a superior high-rate performance. Even at an ultrahigh current density of 100 A g⁻¹ (charge/discharge completed within 3 s), it still maintains quite high specific capacitance (142 F g⁻¹, 122 F cm⁻³) with capacity retention rate of 51%. Moreover, for practical applications, a symmetric supercapacitor was also assembled and the device exhibits a high volumetric performance of 7.1 W h L⁻¹ at 25 W L⁻¹ and 2.2 W h L⁻¹ at 5.8 kW L⁻¹. This work provides a good example of recycling trash for developing high-performance electrode materials used for electrochemical energy storage.

Results and discussion

The schematic preparation process of flour porous carbon material is shown in Fig. S1.† It was synthesized by the carbonization of flour food waste residue (waste Chinese steamed buns residue in this work) and subsequent KOH activation. To optimize the experimental conditions, we prepared a series of samples named as CKmn, where mn represent the mass ratio of flour-based carbon (FBC) and KOH (C/K ratio). Preparation procedure and measurement details are provided in ESI.†

Structure characterization

The morphologies, microstructure and compositions of the as-prepared CK samples were investigated in detail. Fig. 1 shows the corresponding SEM and TEM images. It is seen from the SEM images that, as the mass ratio of KOH increases, the etching effect on the carbon framework by the oxidation of the carbon and the intercalation of potassium compounds^25,26 becomes more apparent. The CK21 produced with the lowest KOH ratio shows irregularly shaped particle (Fig. 1a), while CK11 (Fig. 1c) and CK12 (Fig. 1e) are in nanosheet-like shape. The carbon sheet of CK12 is much thinner than that of CK11. Moreover, the CK14 exhibits corrugated graphene-like structure (Fig. 1g). These nanosheet-like structures of CK11, CK12, and CK14 can shorten the diffusion distance of electrolyte ions, thus are helpful for the diffusion of electrolyte ions into micropores.¹⁰ Fig. 1b, d, f, and h show the TEM images of the different samples. The as-prepared CK samples exhibit nanosheet-like shape. The carbon sheet become thinner as the KOH ratio increased, agreeing well with the SEM images. No obvious lattice fringe was observed in the HRTEM images, indicating the amorphous structure of the carbon sheet. This is further confirmed by X-ray diffraction (XRD) and Raman spectrum measurements. XRD patterns of FBC and CK samples was shown in Fig. 2a. The broad peak of CK samples at 43.8° can be attributed to the (100) reflection of the disordered layers of the amorphous carbon. Raman spectra of CK samples in Fig. 2b show two bands concerted at 1330 cm⁻¹ and 1582 cm⁻¹, corresponding to the D (defect)- and G (graphitic)-bands of amorphous carbon. The intensity ratio of D/G is approximate 1, suggesting the amorphous structure of the samples, which is consistent with the results of TEM and XRD.

Fig. 1 SEM and TEM images of sample CK21 (a and b), CK11 (c and d), CK12 (e and f) and CK14 (g and h).

Fig. 2 XRD pattern (a) and Raman spectra (b) of FBC and activated samples; N₂ adsorption/desorption isotherms (c) and the corresponding pore size distributions (d) of activated samples.

To investigate the pore structures of the activated samples, N₂ adsorption/desorption was conducted at 77 K and the results were displayed in Fig. 2c and d. Fig. 2c presents the N₂ adsorption/desorption isotherms of CK samples. All CK samples exhibits a hybrid sorption isotherm of type I and type IV with a H4 hysteresis loop. The sharp increase of N₂ adsorption at a low relative pressure (<0.01) is due to the abundant micropores, and the obvious type H4 hysteresis loop between 0.5 and 0.9 of the P/P₀ value demonstrates the existence of mesopores.^27,28 As the KOH ratio increases, the activated sample exhibits a higher amount of nitrogen adsorption, revealing pores developed significantly. The corresponding pore size distribution (PSD) curves were shown in Fig. 2d. A bimodal distribution was observes, corresponding with mesopores (at about 3–5 nm) and micropores (at about 1 nm).

To further investigate the formation process of the pore structure, FBC, FBC800 and a series of de-nitrogen (DN) samples were prepared for comparison. Flour-based carbon (FBC) was prepared by the carbonization of the waste residue at 500 °C and FBC800 was prepared by further carbonization of FBC at 800 °C for 1 h. The DN samples were prepared with the de-nitrogen flour as raw material for the steamed buns and the same treatment process (see details in ESI†). SEM images and elemental mapping results of FBC were shown in Fig. S2.† The FBC sample contains abundant micrometer-sized voids. The elemental mapping images indicate that oxygen and nitrogen are homogeneously distributed in the carbon. Furthermore, X-ray photoelectron spectroscopy (XPS) measurements were also performed to obtain more composition information of the samples. The survey spectra of FBC in Fig. 3a confirm the coexistence of C, O and N, and the fitting result demonstrates that FBC primarily contain carbon (81.92 at%), oxygen (13.59 at%), and nitrogen (4.5 at%). The high-resolution C 1s spectrum can be resolved into four individual peaks (Fig. 3b), respectively referring to C–C (284.6 eV), C–N (285.2 eV), C–O (286.3 eV) and C=O (288.9 eV).²⁹ Similarly, the O 1s spectrum (Fig. 3c) can be resolved in three peaks centered at 531.7, 533.4 and 534.9 eV, corresponding to C–O, C=O and –O–, respectively.³⁰ The N 1s high-resolution spectrum (Fig. 3d) can be assigned to three types including the pyridinic N at 398.3 eV, pyrrolic N at 398.8 eV, and graphitic N at 400.6 eV.³¹ It is worth pointing out that the peak area of the graphitic nitrogen (82% of the whole N 1s peak area) is much larger than that of the two other types (18% add up). Physicochemical properties of the samples were listed in Tables 1 and S1.† It can be seen that the nitrogen content of the FBC800 is about 4.97 at%. This is close to that of FBC (4.5 at%) but much higher than the CK samples (<1 at%), indicating that the obvious decrease of nitrogen content is attributed to the etching effect caused by the KOH activation rather than the high temperature treatment. Moreover, DN–CK11 sample exhibits an obvious lower nitrogen uptake than CK11 with no hysteresis loop (Fig. 4a). The PSD (inset in Fig. 4a) shows that DN–CK11 contains no mesopores and much less micropores than CK11. More disparities can be seen from the calculated results listed in Table 1. The values of micropore surface area and pore volume of CK11 are much higher than those of DN–CK11. There is nearly no mesopore in the DN–CK11. Hence, these results demonstrate that nitrogen is a key factor in the formation process of pore, especially for mesopores. According to the XPS results (Fig. 3d), nitrogen doped in the carbon framework can be assigned to three types, namely, pyridinic, pyrrolic, and graphitic³² (Fig. 4b). The graphitic nitrogen is the main component for the FBC samples. Therefore, it may be the abundant in-planed graphitic nitrogen (Fig. 4b) that become active sites, which were etched by KOH in preference to carbon atoms and then resulting the forming of pores at the locations of the nitrogen atoms.

Fig. 3 XPS survey spectra of FBC (a) and the high-resolution XPS spectra of C 1s (b), O 1s (c) and N 1s (d).

Table 1 Physicochemical properties of the samples

Samples	Elemental analysis			SBET^a (m^2 g^−1)	Smicro^b (m^2 g^−1)	Vtotal^c (cm^3 g^−1)	Vmicro^d (cm^3 g^−1)	Vmeso^e (cm^3 g^−1)	ρ^f (g cm^−3)
	C%	O%	N%
a Specific surface area calculated by BET method.b Specific surface area of micropores.c Total pore volume.d Micropore volume.e Mesopore volume.f Mass density of the samples.
FBC	81.92	13.59	4.5	<1	—	—	—	—	—
CK21	88.97	10.21	0.81	969	903	0.409	0.357	0.052	1.10
CK11	88.64	10.69	0.67	1516	1416	0.663	0.565	0.098	0.86
CK12	86.37	12.99	0.64	1813	1623	0.865	0.685	0.180	0.73
CK14	87.97	11.63	0.39	1834	1669	0.928	0.761	0.167	0.70
DN–CK11	89.63	10.37	—	1047	1036	0.437	0.427	0.01	1.07

---

Fig. 4 N₂ adsorption/desorption isotherms and the corresponding pore size distributions of CK11 and DN–CK11 (a); and different types of nitrogen doping configuration (b).

Detailed porosity characteristics of the CK samples and the results calculated by t-plot method were listed in Table 1. We can see that all samples activated by KOH exhibit a high surface area contributed by abundant micropores and appropriate mesopores. As the ratio of KOH vs. carbon was increased from 0.5 to 4, the surface area of the activated samples increased from 969 to 1834 m² g⁻¹. Moreover, it is noted that the activated CK samples exhibit a moderate total pore volume from 0.409 to 0.928 cm³ g⁻¹, which is much lower than the previous works.^18,23 Such a dense structure with a high density (0.7 to 1.1 g cm⁻³) make CK samples have promising advantages in enhancing the volumetric capacitance performance as electrode materials.

Surface functional groups play a significant role as supercapacitor electrode materials owing to its positive effects from the wettability and pseudocapacitance. Fig. S3a† presents the XPS survey spectra of CK samples and the semi-quantitative analysis results were summarized in Table 1. All the activated samples exhibit obvious O 1s peaks with a high content of over 10%. The high-resolution C 1s and O 1s spectrum shown in Fig. S3b and c† indicate that oxygen mainly exists in the configurations of C–O, C=O and –O–, corresponding with phenol type groups, quinone groups and ether groups. Such a high oxygen content with abundant surface oxygen functionalities can effectively enhance the wettability of the electrode in the aqueous system.⁸ These oxygen may also participate in the faradaic reactions and contribute some pseudocapacitance.³³

Electrochemical characterization

The electrochemical properties of the as-prepared CK samples for supercapacitor electrode were evaluated through cyclic voltammetry (CV), galvanostatic charge/discharge (GC) and electrochemical impedance spectroscopy (EIS) ranging from −1 to 0 V in 6 M KOH aqueous solution. Fig. 5a presents the CV curves of CK samples recorded at a scan rate of 5 mV s⁻¹. One can see all the activated samples exhibit quasi-rectangular shaped CV curves with an inconspicuous broad peak at about −0.8 V, implying the double layer capacitive behavior during the fast charge/discharge with some pseudocapacitance caused by redox reactions of surface oxygen containing functional group. Moreover, the CV curves of CK11 performed at various scan rates (Fig. S4†) shows that the curves maintained its rectangular shape well even at a high scan rate of 100 mV s⁻¹, demonstrating the good rate performance of the samples as supercapacitor electrodes.

Fig. 5 CV curves at a scan rate of 5 mV s⁻¹ (a), GC curves at a current density of 0.2 A g⁻¹ (b), Nyquist plots (c), gravimetric (d) and volumetric capacitances (e) at different current density from 0.1 A g⁻¹ to 100 A g⁻¹ of different samples. (f) The cycling performance at 2 A g⁻¹ and the GC lines of the first and the 3000th cycle (inset) of the CK11.

To evaluate the capacitive performance of the as-prepared activated samples, GC test was performed and the results were shown in Fig. 5b. All GC curves exhibit isosceles triangle shapes with slight distortion at a current density of 0.2 A g⁻¹, confirming the good capacitive behavior of EDLC and some pseudocapacitance. The pseudocapacitance is caused by the redox reactions of oxygen containing functional group and it can improve the specific capacitance apparently. The gravimetric capacitance of the CK samples were calculated by the discharge curves and the results were listed in Table 2. As the KOH/carbon ratio increased from 0.5 to 4, the corresponding gravimetric capacitance of the samples increased from 220 to 321 F g⁻¹. Such an improvement in the capacitance can be ascribed to the enlarged micropores area (Table 1), which is the key factor in EDLC behavior for carbon materials by providing abundant electrochemical active sites to accumulate charges. It is noted that CK14 samples contained a larger surface area of micropores (1669 m² g⁻¹) than CK12 (1623 m² g⁻¹) but they showed approximate gravimetric capacitance. This can be attributed to a high oxygen content of CK12, which provided more pseudocapacitance and thus enhanced the specific capacitance.³⁴ In practical applications, volumetric capacitance is also a pivotal parameter for supercapacitors, and we thus calculate the volumetric capacitance of the activated samples and summarize the results in Table 2. All the activated samples exhibit a high C_v value of over 220 F cm⁻³, which is much higher than those of the commercial porous carbon (60 F cm⁻³).³⁵ More importantly, due to the sharp decrease of mass density when the KOH/carbon ratio was increased, CK21 and CK11 samples demonstrate higher volumetric capacitance than CK12 and CK14, and CK11 exhibit a remarkable volumetric capacitance of 241 F cm⁻³.

Table 2 Electrochemical performances of different samples

	Cg^a (F g^−1)	Cv^b (F cm^−3)	C20^c (F cm^−3)	C100^d (F cm^−3)	Rs^e (ohm)	Rct^f (ohm)
a Gravimetric capacitance calculated from GC curves at a current density of 0.1 A g^−1.b Volumetric capacitance calculated as Cv = ρCg.c Volumetric capacitance at a current density of 20 A g^−1.d Volumetric capacitance at a current density of 100 A g^−1.e The equivalent internal resistance.f The charge transfer resistance.
CK21	220	239	137	82	0.396	0.126
CK11	278	241	165	122	0.311	0.227
CK12	321	234	145	110	0.283	0.239
CK14	321	223	136	104	0.319	0.131

---

The Nyquist plots of EIS tests were shown in Fig. 5c. A segment of circle at high frequency region and a nearly vertical curve at low frequency region suggest that the as-prepared CK-based electrodes exhibit an easy diffusion of electrolyte and an ideal capacitive behavior, respectively. More specifically, the intersection of the curve and the real axis signals the value of the equivalent internal resistance (R_s), which is the equivalent of the sum of the intrinsic resistance of the electrode materials, the interfacial contact resistance between active material and current collector, and the electrolyte resistance.³⁶ The diameter of the semicircle impedance loop represents the charge transfer resistance. It can be seen from Table 2 that all the CK samples show a quite low R_s of less than 0.40 ohm and a R_ct less than 0.24 ohm, demonstrating the high electronic conductivity and weak pseudocapacitive behavior of as-prepared activated samples. As a key factor of supercapacitors, rate performance involved with the specific capacitance measured at current density from 0.1 to 100 A g⁻¹ of the as-prepared activated samples was summarized in Fig. 5d. As the current density increased, the specific capacitance of the samples obviously decreased due to the incomplete diffusion of the electrolyte ion into the pores. All the activated samples exhibit a high capacitance retention. The samples prepared with a high KOH/carbon ratio show a higher capacitance as result of their larger specific surface area. The CK12 and CK14 samples exhibit a specific capacitance of 321 F g⁻¹ at 0.1 A g⁻¹, however, as the current density was increased to 20 and 100 A g⁻¹, the specific capacitances of CK12 and CK14 reached 199 and 194 F g⁻¹, and 150 and 148 F g⁻¹ at 100 A g⁻¹, respectively. The better capacitance performance of CK12 than CK14 is attributed to the larger mesoporosity that provide more channels for the smooth ion transport.³⁷ The volumetric capacitance performance of the as-prepared samples at different current density were also calculated and compared in Fig. 5e. The obvious gap between CK11 sample and the other samples is due to its both high SSA and mass density. The CK11 shows a high volumetric capacitance of 241 F cm⁻³ at 0.1 A g⁻¹, 165 F cm⁻³ with a retention of 68.4% at 20 A g⁻¹, and 122 F cm⁻³ with a retention of 51% even at a high current density of 100 A g⁻¹.

Considering the values of C_g, C_v, and rate performance, the CK11 sample demonstrates to be the best one among the as-prepared activated carbon materials. Note that, the capacitance performance of the CK11 based electrode is better than the most of the previous porous carbon (Table S2†). The cycling performance of the CK11 based electrode was evaluated by GC tests at 2 A g⁻¹ (Fig. 5f). It can be observed that the specific capacitance has a retention of 95% after 3000 cycles. The GC curves of the first cycle is almost similar to that of the 3000th cycle (inset in Fig. 5f), indicating the highly reversible charge/discharge process and the good stability of the CK11 based electrode.

To explore the performance of the as-prepared activated carbon material in supercapacitors more practically, a symmetric two electrodes aqueous supercapacitor was also fabricated using a 2032 stainless steel coin cell. Fig. 6 presents the Ragone plot of the CK11-based symmetric supercapacitor. We can see that the device exhibits a high volumetric energy density of 7.1 W h L⁻¹ at 25 W L⁻¹ and even at a high power density of 5.8 kW L⁻¹, it still maintains 2.2 W h L⁻¹.

Fig. 6 Ragone plot of the CK11-based symmetric supercapacitor.

Conclusions

In summary, nanosheet-like porous carbon prepared from flour food waste residue exhibits an optimal pore structure and excellent electrochemical capacitance. When used as supercapacitors electrodes, it shows high gravimetric (278 F g⁻¹) and volumetric capacitance (241 F cm⁻³). Meanwhile, it shows excellent rate performance with 68.4% retention at 20 A g⁻¹ and with 51% retention even at a high current density of 100 A g⁻¹ (charge/discharge completed within 3 s), as well as good cycling stability with 95% capacitance retention after 3000 cycles. Moreover, a symmetric supercapacitor was assembled more practically, and the device exhibits a superior volumetric performance of 7.1 W h L⁻¹ at 25 W L⁻¹ and 2.2 W h L⁻¹ at 5.8 kW L⁻¹. Thus, the as-prepared cost-effective activated carbon from trash recycling shows great prospect for applications in high-performance supercapacitor.

Acknowledgements

This work was financially supported by the National Natural Science Foundation of China (Grant No. 51672151, 51232005) and 973 program of China (No. 2014CB932401).

Notes and references

1. W. Deng, Y. Zhang, L. Yang, Y. Tan, M. Ma and Q. Xie, RSC Adv., 2015, 5, 13046–13051.
2. C. Leng and K. Sun, RSC Adv., 2016, 6, 57075–57083.
3. E. Haque, M. M. Islam, E. Pourazadi, M. Hassan, S. N. Faisal, A. K. Roy, K. Konstantinov, A. T. Harris, A. I. Minett and V. G. Gomes, RSC Adv., 2015, 5, 30679–30686.
4. M. Sevilla and A. B. Fuertes, ACS Nano, 2014, 8, 5069–5078.
5. F. Yang, M. Zhao, Q. Sun and Y. Qiao, RSC Adv., 2015, 5, 9843–9847.
6. J.-J. Li, M.-C. Liu, L.-B. Kong, D. Wang, Y.-M. Hu, W. Han and L. Kang, RSC Adv., 2015, 5, 41721–41728.
7. Y. Zhang, H. Feng, X. Wu, L. Wang, A. Zhang, T. Xia, H. Dong, X. Li and L. Zhang, Int. J. Hydrogen Energy, 2009, 34, 4889–4899.
8. M. Wu, P. Ai, M. Tan, B. Jiang, Y. Li, J. Zheng, W. Wu, Z. Li, Q. Zhang and X. He, Chem. Eng. J., 2014, 245, 166–172.
9. H. Jiang, P. S. Lee and C. Li, Energy Environ. Sci., 2013, 6, 41–53.
10. C. Long, X. Chen, L. Jiang, L. Zhi and Z. Fan, Nano Energy, 2015, 12, 141–151.
11. Y. Liang, L. Cai, L. Chen, X. Lin, R. Fu, M. Zhang and D. Wu, Nanoscale, 2015, 7, 3971–3975.
12. X. Cui, R. Lv, R. U. R. Sagar, C. Liu and Z. Zhang, Electrochim. Acta, 2015, 169, 342–350.
13. G. Xin, Y. Wang, X. Liu, J. Zhang, Y. Wang, J. Huang and J. Zang, Electrochim. Acta, 2015, 167, 254–261.
14. F. Gao, G. Shao, J. Qu, S. Lv, Y. Li and M. Wu, Electrochim. Acta, 2015, 155, 201–208.
15. X. Zhang, Y. Jiao, L. Sun, L. Wang, A. Wu, H. Yan, M. Meng, C. Tian, B. Jiang and H. Fu, Nanoscale, 2016, 8, 2418–2427.
16. K. Wang, R. Yan, N. Zhao, X. Tian, X. Li, S. Lei, Y. Song, Q. Guo and L. Liu, Mater. Lett., 2016, 174, 249–252.
17. K. Wang, N. Zhao, S. Lei, R. Yan, X. Tian, J. Wang, Y. Song, D. Xu, Q. Guo and L. Liu, Electrochim. Acta, 2015, 166, 1–11.
18. Q. Liang, L. Ye, Z.-H. Huang, Q. Xu, Y. Bai, F. Kang and Q.-H. Yang, Nanoscale, 2014, 6, 13831–13837.
19. D. Wang, Y. Min and Y. Yu, J. Solid State Electrochem., 2015, 19, 577–584.
20. J. Xu, Q. Gao, Y. Zhang, Y. Tan, W. Tian, L. Zhu and L. Jiang, Sci. Rep., 2014, 4, 5545.
21. Y. Luan, L. Wang, S. Guo, B. Jiang, D. Zhao, H. Yan, C. Tian and H. Fu, RSC Adv., 2015, 5, 42430–42437.
22. G. Ma, Q. Yang, K. Sun, H. Peng, F. Ran, X. Zhao and Z. Lei, Bioresour. Technol., 2015, 197, 137–142.
23. X. Li, W. Xing, S. Zhuo, J. Zhou, F. Li, S.-Z. Qiao and G.-Q. Lu, Bioresour. Technol., 2011, 102, 1118–1123.
24. G. P. Hao, A. H. Lu, W. Dong, Z. Y. Jin, X. Q. Zhang, J. T. Zhang and W. C. Li, Adv. Energy Mater., 2013, 3, 1421–1427.
25. M. Sevilla and R. Mokaya, Energy Environ. Sci., 2014, 7, 1250–1280.
26. M. Wu, Q. Zha, J. Qiu, Y. Guo, H. Shang and A. Yuan, Carbon, 2004, 42, 205–210.
27. X. Yang, H. Huang, G. Zhang, X. Li, D. Wu and R. Fu, Mater. Chem. Phys., 2015, 149–150, 657–662.
28. Y. Liang, W. Mai, J. Huang, Z. Huang, R. Fu, M. Zhang, D. Wu and K. Matyjaszewski, Chem. Commun., 2016, 52, 2489–2492.
29. L. Qie, W. Chen, H. Xu, X. Xiong, Y. Jiang, F. Zou, X. Hu, Y. Xin, Z. Zhang and Y. Huang, Energy Environ. Sci., 2013, 6, 2497–2504.
30. D. Hulicova-Jurcakova, M. Seredych, G. Q. Lu and T. J. Bandosz, Adv. Funct. Mater., 2009, 19, 438–447.
31. L. Hao, B. Luo, X. Li, M. Jin, Y. Fang, Z. Tang, Y. Jia, M. Liang, A. Thomas and J. Yang, Energy Environ. Sci., 2012, 5, 9747–9751.
32. Z.-H. Sheng, L. Shao, J.-J. Chen, W.-J. Bao, F.-B. Wang and X.-H. Xia, ACS nano, 2011, 5, 4350–4358.
33. V. Ruiz, C. Blanco, E. Raymundo-Piñero, V. Khomenko, F. Béguin and R. Santamaría, Electrochim. Acta, 2007, 52, 4969–4973.
34. T.-X. Shang, R.-Q. Ren, Y.-M. Zhu and X.-J. Jin, Electrochim. Acta, 2015, 163, 32–40.
35. E. Frackowiak and F. Béguin, Carbon, 2001, 39, 937–950.
36. C. Zhan, Q. Xu, X. Yu, Q. Liang, Y. Bai, Z.-H. Huang and F. Kang, RSC Advances, 2016, 6, 41473–41476.
37. X. Yu, C. Zhan, R. Lv, Y. Bai, Y. Lin, Z.-H. Huang, W. Shen, X. Qiu and F. Kang, Nano Energy, 2015, 15, 43–53.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra18056f

---