Cu-modified biomass-derived activated carbons for high performance supercapacitors

HU Jia-rong, ZHOU Jia-wei, JIA Yu-xin, LI Shuang*

（School of Chemical Engineering, Northwest University, Xi’an 710069, China）

Abstract： Porous carbons are widely used in supercapacitors, owing to their long cycle life and natural abundance. However,most of these electrode materials give a low capacitance, which leads to low energy density. Cu-doped biomass-derived activated carbons (Cu-ACs) were synthesized using a simple, low-cost carbonization and KOH activation method. The copper nanoparticles had mixed valence states (CuO, Cu2O, Cu0) and were uniformly dispersed on the surface of the AC. Due to the fast electron/ion transfer paths provided by the pore structure, and an accelerated redox reaction between the three Cu species, the Cu-ACs achieved an excellent capacitive performance. In a three-electrode system, the Cu-AC sample prepared by KOH activation with a KOH/ （Cu+char）mass ratio of 2 had a high specific capacitance of 360 F g−1 at 0.5 A g−1, 1.21 times that of AC (163 F g−1). When it was fabricated into a symmetric capacitor, the device had a good electrochemical performance with a specific capacitance of 143.44 F g−1 at 0.5 A g−1 and a good cyclic stability with an 81.8% capacitance retention after 6 000 cycles.

Key words: Biomass；Copper；Hierarchical porous carbon；Electrode；Supercapacitors

1 Introduction

Supercapacitor (SC) has been widely used in electric vehicles, laptops and other fields owing to its fast charging-discharging process, wide operating temperature range, low maintenance cost, high power density and long life. And an growing demand for advanced energy storage/conversion systems has brought them a lot of attention[1–3].

Typically, SC has 2 categories based on energy storage mechanism: double layer capacitors (EDLCs)and pseudo-capacitors. EDLCs are based on the accumulation of electrostatic charge at electrode/electrolyte interface, and it strongly depends on the effective surface area of the electrode. Carbon-based electrode materials such as active carbons[4], carbon nanotubes[5], carbon microspheres[6], graphene[7]and reduced graphene oxide[8]have been widely used in EDLCs due to their excellent cycling stability and rate capability. And their electrochemical properties, flexibility and processing have been studied[9]. Cost of these materials is expensive, even if their uses by other workers previously show relatively high specific capacitance. Low cost and high-performance carbonbased electrode materials are necessary for SC applications. Renewable biomass-derived carbons have been widely reported as adsorbents and electrode materials,such as poplar catkin[10], pomelo peel[11], kapok flower[12]. Moreover, some resultant porous carbons retain some oxygen/nitrogen groups from their precursors. Previous research works have demonstrated that optimizing the pore structure of carbon materials can sufficiently improve the electrode/electrolyte interface for charge storage[13]. Hierarchical porous carbon materials are expected to be high-performance SC electrode materials because their large pores can store electrolytes as ion buffer devices, the mesopores can rapidly distribute electrolyte ions, and the micropores can improve their specific capacitances[14,15]. Wang et al. fabricated a highly flexible PB/rGO films with a porous architecture, which showed a high specific capacitance of 286 F g−1at 0.3 A g−1along with superior rate and cycling performance[16]. Wu et al. prepared a N,S dual-doped porous carbon with a large specific surface area and hierarchical structure,which had a high specific capacitance (360 F g−1at 1.0 A g−1), superior cycling durability and excellent rate characteristic[17]. Wu et al. obtained a N-enriched hierarchical porous carbon derived from tobacco stem,which delivered a capacitance of 252.9 F g−1at 0.5 A g−1in 6 mol L−1KOH when it was used as a SC electrode material[18].

In general, the pore structure is strongly dependent on the activation method. Physical and chemical activation method are regarded as two widely used methods[19]. Through physical activation, the carbon precursors are carbonized followed by reactions of gas activators with carbon to develop pores[20]. Chemical activation using KOH, ZnCl2, etc[21,22]as activating agents generates pores with narrow distribution and large specific surface area, thus providing excellent capacitive performance[23]. Among the commonly used activating agents, KOH activation has the advantages in obtaining porous carbons with large specific surface area. For example, ultra-high surface area was derived from lotus leaves by KOH activation (up to 3 560 m2g−1)[24].

The charge storage mechanism of pseudo-capacitance is based on the reversible faraday redox reactions on electrode surface, the transition metal oxides are mainly selected as potential electrode materials for pseudo-capacitors such as RuO2[25]and MnO2[26]. They generally have much larger specific capacitances than EDLC active materials. However, RuO2is expensive and toxic, thus seriously limiting its application[27].MnO2electrode is limited by its intrinsic poor conductivity, structural stability and lack of mechanical flexibility[28].

Copper oxide has the advantages of chemical stability, wide source and environmental compatibility[29]. CuO owns a high theoretical high specific capacitance[30], but it suffers from low conductivity[31]and the destruction of the structure[32]during the electrochemical redox process. The carrier mobility of Cu2O is higher than that of CuO. It was reported that CuO and Cu2O nanostructures of different sizes and shapes can be used to meet the requirement of energy storage[33,34]. The electrical conductivity of carbon can be increased by addition of Cu[35], and the hybridization of CuxO and biomass-derived carbon results in both high EDLCs and battery-type capacity.The Cu@C composite derived from Cu-MOF showed a specific capacitance of 260.5 F g−1at 0.5 A g−1, and achieved an energy density of 18.38 Wh kg−1at a power density of 350 W kg−1as a symmetric supercapacitor[36]. A Cu-C nanocomposite as an electrode had a specific capacitance could up to 318 F g−1at 1 A g−1,and remained at nearly 100% after 10 000 cycles[37].Moreover, a copper/carbon (Cu/C) composite delivered a specific capacitance of 316 F g−1at 0.5 A g−1in 0.1 mol L−1H2SO4aqueous electrolyte[38]. These results indicated that Cu@C could be utilized as electrode material with good electrochemical performance. However, tedious preparation of carbon support makes it difficult to achieve large-scale preparation and limits its wide practical applications.

In this work, Cu/CuxO modified porous carbons(Cu-AC) were prepared using walnut shell as a precursor, and one of samples had a large specific surface area of 661 m2g−1and hierarchical porous structure. When it was used as a SC electrode material, it showed high specific capacitance of 360 F g−1at 0.5 A g−1in a 3 electrode system with 6 mol L−1KOH,and a good cyclic stability of 81.8% after 6 000 cycles in a two electrode system, indicating its good electrochemical performance and promising application in energy storage.

2 Materials and methods

2.1 Materials

Walnut shell was collected from market and ground into powder less than 20 μm, potassium hydroxide (KOH) and copper sulfate anhydrous (CuSO4)were provided by Sinopharm Chemical Reagent Co.(China). All tests were performed in deionized water.

2.2 Preparation

A mixture of walnut shell (2 g) and CuSO4(0.159 g) was added and stirred in 200 mL deionized water at 25 °C for 12 h, and then dried in a rotary evaporator at 80 °C. Subsequently, the obtained powder was carbonized at 600 °C for 0.5 h in N2atmosphere with a heat rate of 5 °C min−1to obtain a sample denoted as Cu@C. Then, Cu@C was mixed with KOH at different ratios (mass ratios of 1∶1 and 1∶2) and activated at 800 °C for 2 h under N2atmosphere. Finally, the activated samples were washed by 1 mol L−1HCl and distilled water to eliminate any impurities, and marked as Cu-AC-x (x was the mass ratio of KOH to Cu@C). For comparison, the walnut shell was carbonized at 600 °C for 0.5 h and activated at 800 °C for 2 h with KOH at a mass ratio of 1∶1,the resultant sample was named as AC-1.

2.3 Structure characterization

Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) were conducted on a Carl Zeiss (ZEISS SIGMA) and a JEM-2100F(JEOL Ltd.), respectively. Powder X-ray diffraction(XRD) and Raman spectra were performed on a Rigaku D/max 2200/Pc and a Renishaw InVia Reflex(laser excitation at 532 nm), respectively. The surface area and pore size distribution were obtained by a JWBK122W. The elemental composition was characterized by X-ray photo-electron spectroscopy (XPS,Kratos AXIS NOVA) and an elemental analyzer (EA,vario EL Ⅲ).

2.4 Electrochemical measurement

For a three-electrode system (3E) cell, an Ag/AgCl and a platinum foil was served as the reference electrode and counter electrode, respectively.And the working electrode was fabricated after the mixture of the active material, polytetrafluoroethylene (PTFE) and acetylene black was coated on a nickel foam (1×1 cm2) with a mass proportion of 8∶1∶1,and the loading of active material for each working electrode was about 4.0 mg cm−2. Finally, the working electrode was pressed into a sheet at 10 MPa, and dried at 60 °C for 12 h. In a two-electrode system (2E)cell, the symmetric device was fabricated by two electrodes of identical active material loading. The electrolyte used in both system was 6.0 mol L−1KOH aqueous solution.

And the electrochemical performance was characterized by a CHI660E electrochemical workstation(Chenhua, China). Cyclic voltammetry (CV) and Galvanostatic charge/discharge (GCD) measurement were conducted at various scan rates (5.0−100 mV s−1) and different current densities (0.5−10 A g−1), respectively. The electrochemical impedance spectra (EIS)were obtained at the open circuit voltage with an amplitude of 5.0 mV in the frequency range from 100 kHz to 1 Hz. The cycle stability with 6 000 cycles was tested at 10 A g−1.

The gravimetric specific capacitanceCGCD(F g−1)in 3E was obtained by GCD curves according to the following equation:

WhereI, ∆t,mtotaland ∆Vare the discharge current(A), discharge time (s), total mass of active substance(g) and voltage window, respectively.

The energy density (Ecell) and power density(Pcell) could be calculated as the equation (3) and (4):

3 Results and discussion

The SEM images of samples are shown in Fig. 1. AC-1 and Cu-AC-x with pleated structures were composed of a large number of nanosheets. All samples exhibited an interconnected structure with numerous pores due to the KOH activation. During activation process, KOH and its derivants (K, K2O,K2CO3, etc) etched the carbon skeleton to form a pore network structure accompanied by formation of CO and CO2, which further promoted the development of pores[39]. Furthermore, many particles of different sizes were uniformly distributed on the surfaces of Cu-AC-x as shown in Fig. 1 (c-f). According to the EDS results in Fig. S1 and Fig. S2, it could be seen that the particles were Cu and CuxO, suggesting Cu/CuxO modified porous carbons were successfully synthesized by carbonization and activation. And the quantity of Cu/CuxO particles on Cu-AC-2 was more than that of Cu-AC-1 while the particle sizes of the former were relatively smaller than that of Cu-AC-1.

The nitrogen adsorption/desorption isotherms and the pore size distributions of AC-1, Cu-AC-1 and Cu-AC-2 nanocomposites are displayed in Fig. 2. The N2uptake atp/p0< 0.01 suggested the presence of micropores and the H4 hysteresis loop atp/p0> 0.4 proved the existence of mesopores in all samples(Fig. 2a). Pore size distributions of AC-1, Cu-AC-x based on the DFT method (Fig. 2b) further indicated a large number of micropores and a certain amount of mesopores in these samples. Combined with the structural parameters of the samples in Table 1, it could be seen that AC-1 had the highest specific surface area and pore volume of 1 028 m2g−1and 0.19 cm3g−1, respectively. Both the specific surface areas and pore volumes of Cu-AC-1 and Cu-AC-2 were smaller than AC-1, which might be due to the pore blockage by Cu/CuxO nanoparticles. Besides, the mesopore surface area of Cu-AC-2 (153 m2g−1) was higher than that of Cu-AC-1 (143 m2g−1). The mesoporous structure was beneficial to electrochemical performance in SCs[15]. Such a micro-mesoporous structure was important to improve the electrical double layer capacitance, which provided more space for energy storage and high energy density[40]. The hierarchical porous structures could provide abundant charge storage sites, offering electrochemically active materials for SCs[40,41].

Table 1 Pore volumes and specific areas of AC-1 and Cu-AC-x.

XRD diffraction patterns of AC-1 and Cu-AC-x nanocomposites are shown in Fig. 3a. A broad and wide diffraction peak at about 22.4° was confirmed to be the (002) plane of the graphite carbon. The characteristic peaks at 43.29°, 50.4° and 73.13° were attributed to the (111), (200) and (220) planes of Cu (JCPDS No. 04-0836), respectively. The weak peak at 36.4° was attributed to the (111) plane of the Cu2O(JCPDS No. 75-1531)[42]. Therefore, the crystal phase of Cu0and Cu2O was evident in the Cu-AC-2 nanocomposite after KOH activation. Meanwhile, Raman spectroscopy was used to obtain graphitization degree and defect information of samples. The Raman spectra of AC-1 and Cu-AC-x nanocomposites are presented in Fig. 3b. The bands at 146, 217 and 636 cm−1indicated the existence of Cu2O[43]. The two dominant peaks at about 1 341 and 1 593 cm−1corresponded to the D-band and G-band of graphite, respectively.Dband was attribute to the disordered graphite structure that destroyed the symmetry whileGband corresponded to the sp2-hybridized carbons associated with hexagonal carbon atoms[44]. TheID/IGratio of AC-1(2.21) was lower than that of Cu-AC-1 (2.58) and Cu-AC-2 (2.87), indicating the existence of more disordered and defective structure in Cu-AC-2 nanocomposite. The increase of defects could generate more active sites to enhance the specific capacitance of a SC[45].

XPS was adopted to illustrate the chemical states of the three samples. XPS spectra of AC-1 and Cu-AC-x composites are shown in Fig. 4a. It was found that these samples contained C, O, N, Cu. Fig. 4b shows their C1s XPS spectra, the peaks at around 284.8, 286.2 and 288.5 eV corresponded to C―C/C＝C band, C―O band and C＝O band, respectively[46,47]. 3 peaks of O 1s (Fig. 4c) appeared at 531.5 (O-I), 532 (O-II), 533.4 eV (O-Ⅲ), corresponding to oxygen in metal oxides[48], C＝O[49]and C―O[50], respectively. Fig. 4d displays N 1s spectra of AC-1 and and Cu-AC-x, the three peaks were assigned to pyridine-N-oxide (N―X, 403 eV), quaternary-N (N―Q, 400.6 eV) and pyrrolic-N (N-5,399.9 eV)[15]. N-5 could contribute to the enhancement of the pseudo-capacitance[51], and N―Q in carbon skeleton might help to improve electrical conductivity[15]. Thus, the presence of N-containing functional groups was conducive to the improvement of electrochemical performance. Fig. 4e and Fig. 4f display Cu 2p spectra of Cu-AC-1 and Cu-AC-2, respectively. The appearance of Cu 2p3/2at 932.5 eV and Cu 2p1/2at 952.4 eV further confirmed the presence of Cu0. Cu2O was identified by the peaks of Cu 2p3/2at 933.3 eV and Cu 2p1/2at 953.7 eV. The peaks of Cu 2p3/2at 935.0 eV and Cu 2p1/2at 955.0 eV, as well as the satellite peaks of 941.9, 944.4 and 962.8 eV verified the existence of CuO[52,53]. A quantitative analysis results of Cu0, Cu2O and CuO according to XPS results are listed in Table S1. Generally, bivalent copper has a higher theoretical specific capacitance, and its theoretical specific capacity is about 2 412 F g−1(670 mAh g−1), univalent copper shows a larger carrier mobility and Cu0has better conductivity, which may endow Cu-AC-x an admirable electrochemical property when they are used as electrode materials of SCs. Considering the impurities in samples had a certain influence on the electrochemical performance,walnut shell, AC-1 and Cu-AC-x were investigated by ICP, the results are shown in Table S2. Clearly, walnut shell contained quite a few metal species such as K, Ca and Al, and K showed the highest content of 0.26 wt%. The content of these metals in AC-1 and Cu-AC-x were significantly reduced after carbonization, activation, pickling and washing. Besides, Cu content in Cu-AC-2 was 18.97 wt%, which was higher than that of Cu-AC-1 (11.05 wt%).

CV, GCD, and EIS were applied to examine the electrochemical properties of the prepared electrodes.As shown in Fig. 5a, the CV curves of AC-1 did not deviate from the ideal rectangular shape, demonstrating the electric double layer charging mechanism of AC-1. The Cu-AC-x electrodes containing sharp oxidation and reduction peaks revealed a typical batterytype capacitive charge storage behavior[54]. All CV curves of Cu-AC-x consisted of two strong oxide peaks between −0.2 and −0.5 V associated with the transformation of Cu0to Cu+and Cu+to Cu2+. Similarly, two reduction peaks were ascribed to the conversion of Cu2+to Cu+and Cu+to Cu0. CV curves suggested the energy storage mechanism of Cu-AC-x were an integration of EDLC and battery-type capacitance.Compared with Cu-AC-1, the CV curve area of Cu-AC-2 increased because the Cu/CuxO nanoparticle sizes were smaller and more uniformly distributed for the latter (Fig. 1), which generated more reactive sites,facilitating the completion of the redox reaction in the CV process. The mechanism of battery-type capacitance energy storage in 6 mol L−1KOH solution has been represented by the equations (5) and (6)[55].

The CV curves of Cu-AC-2 at various scan rates are shown in Fig. 5b. Clearly, as the scan rate increasing, the paired redox peaks were gradually weakened.At lower scan rates, the ions in the electrolyte could diffuse to the accessible pores, contributing to an enhanced redox process and better insertion/extraction of the ions[54]. At higher scan rates, the ions did not have enough time to access all the reaction sites, an intensity reduction of the redox peak was observed.The gradually shifting redox peaks were related to the increasing of charge diffusion polarization in the electrode material[54]. This phenomenon inferred that the Cu/CuxO nanoparticles in the Cu-AC-2 electrode might undergo a sufficient valence state transition in the potential window from −0.5 to −0.2 V. In addition,the CV curves of the Cu-AC-2 remained a similar curve shape even at 100 mV s−1, suggesting the excellent capacitive behavior and fast ion transmission speed at the electrode interface, which might be related to the hierarchical porous structure and short diffusion paths of AC.

To further evaluate the super-capacitive characteristics, GCD plots of AC-1 and Cu-AC-x composite electrodes at a 0.5 A g−1are displayed in Fig. 5c. The GCD profile of AC-1 was approximately an isosceles triangle, which was caused by the characteristics of the electric double layer. In GCD curves of Cu-AC-1 and Cu-AC-2, there were two obvious changes in the slope of the potential, which indicated that the redox reaction occured between electrolyte ions and Cu/CuxO nanoparticles. The first potential peak A1 appeared during the discharging process, which could be ascribed to the oxidation of Cu to Cu2O (Eq. (6)).The second potential peak A2 indicated that Cu2O was further oxidized to CuO (Eq. (5)). According to GCD measurements, the specific capacitances of AC-1, Cu-AC-1 and Cu-AC-2 were 163, 306 and 360 F g−1at 0.5 A g−1, respectively. The increased specific capacitance could be ascribed to the presence of Cu/CuxO.The performance difference of Cu-AC-1 and Cu-AC-2 might be caused by reactive Cu/CuxO nanoparticles.Small and evenly distributed Cu/CuxO nanoparticles in Cu-AC-2 (Fig. 1 (e, f)) could give reactive sites,leading to an adequately redox reactions during the GCD test. More importantly, the content of Cu/CuxO nanoparticles in Cu-AC-2 was higher than that of Cu-AC-1 according to ICP results, which contributed more active sites for electrochemical performance.

The obtained results were compared with the reported copper oxide and carbon composite materials as shown in Table 2. The GCD profiles of Cu-AC-x electrodes at different current densities shown in Fig. 5d were not linear in nature, demonstrating the involvement of the faradaic reaction mechanism between the electrode and electrolyte. The specific capacitance of Cu-AC-2 (Fig. 5e) decreased from 360 to 144 F g−1as the current density increasing from 0.5 to 10 A g−1. With increasing the current density,the diffusion depth of electrolyte ions in the pores of electrode material decreased, resulting in the decrease of active sites. Thus Cu-AC-2 showed relatively low rate performance. Table S3 shows the coulombic efficiency of Cu-AC-2 at different current densities. Obviously, it had the lowest coulomb efficiency of 80.2% at 0.5 A g−1, indicating some irreversibility.And this was consistent with the obvious asymmetrical triangle of GCD curves in Fig. 5d.

Table 2 A comparison of electrochemical performance of Cu-AC-2 electrode and other copper-based carbon electrodes.

EIS was obtained in the frequency range from 1 Hz to 100 kHz as shown in Fig. 6. In the low-frequency region, the fast ion diffusion of electrolyte ions in the pores could be proved by the straight line slope, which represented a good capacitive behavior[60]. Compared with AC-1, the linear part of Cu-AC-2 was more vertical, indicating an ideal capacitor. The diameter of the semicircle in the highmiddle frequency implied the lower charge transfer resistance at electrode/electrolyte interface[61]. To illustrate the characteristic of EIS, an equivalent circuit is provided in Fig. 6.ZW,Rctand CPE1 are Warburg impedance, resistance including adsorption/desorption of electroactive substances and a constant phase element with non ideal capacitive behavior, respectively. The x-intercept stands for the equivalent series resistance (Rs), displaying a low resistance of 0.74,0.74 and 0.70 Ω for AC-1, Cu-AC-1 and Cu-AC-1, respectively. And the value ofRsandRct(0.40, 0.23 and 0.17 Ω) showed that Cu-AC-2 had the most excellent conductivity and the lowest charge transfer resistance.

To further evaluate the applicability of Cu-AC-2 for SCs, the electrochemical performance of a symmetrical SC device assembled with two identical Cu-AC-2 electrodes in 6 mol L−1KOH electrolyte was studied (Fig. 7). Fig. 7a is the CV curves of the symmetric SC at scanning rates of 5-100 mV s−1. With increasing the scan rate, the shape of CV profile was not deformed, indicating a good capacitive behavior.Compared with the three-electrode system, the voltage plateau of GCD curves in Fig. 7b was not obvious in 2E system, and the calculated specific capacitance was also lower than that of three-electrode(143.44 F g−1at 0.5 A g−1). This phenomenon was mainly caused by electrode polarization for the electrode material. The 2E system had high Coulombic efficiencies (around 90%) under various current densities and maintained excellent reversible redox capacity during charging-discharging process. Fig. 7c is a Ragone plot of the symmetrical device at different current densities. At 0.5 A g–1, its power density was 250.0 W kg–1and the energy density was 4.98 Wh kg–1. 6 000 charge-discharge cycle tests were performed at 10 A g–1and are shown in Fig. 7d. It could be seen that the specific capacitance remained 81.8%of the initial value, demonstrating an excellent cycle stability of the symmetrical device. The excellent electrochemical performance could be ascribed to the hierarchical porous structure of Cu-AC-2.

4 Conclusion

The Cu-AC-2 nanocomposite was prepared through a simple and low-cost procedure. The Cu/CuxO on AC effectively improved the electrochemical performance of Cu-AC-2. The hierarchical porous structure of Cu-AC-2 provided a rich active area for rapid reversible redox reactions and transfer pathways of ions at the electrode/electrolyte interface.When Cu-AC-2 was used as an electrode material, it showed good electrochemical properties including high rate capability, high specific capacitance and good cycle stability. The capacity of Cu-AC-2 electrode was 360 F g−1at 0.5 A g−1. Besides, the symmetrical device displayed a specific capacitance of 143.44 F g−1at 0.5 A g−1and an excellent cycle stability of 81.8% capacitance retention after 6 000 cycles.Therefore, Cu-AC-2 showed the potential to become an efficient electrode material for the new generation of SCs with high power and high reliability. This study procided a new way to exploit high energy capacitive materials in a wide spectrum of amorphous materials and may promote the practical implementation of high-rate and large-capacity energy storage.

Acknowledgements

This work was financially supported by the National Natural Science Foundation of China(21878244) and the Foundation of State Key Laboratory of Coal Conversion (J1920904).