CoFe₂O₄ nanoparticles as a bifunctional agent on activated porous carbon for battery-type asymmetrical supercapacitor

Material and agents

Eucommia ulmoides Oliver (EUO)-wood was harvested and collected from the experimental plot in the Northwest Agriculture and Forestry University, Yangling District, Shaanxi Province, China^[15]. After removing the bark and leaf, the EUO-wood was dried and smashed to obtain a wood meal (40-60 meshes). All chemicals (NaOH, HCl, FeCl₃·6H₂O, CoCl₂·6H₂O, and FeSO₄·7H₂O) in analytical pure (≥ 99.0%) were acquired from Aladdin Biochemical Pharmaceutical Co., Ltd (Shanghai, China).

Preparation of AC derived from EUO-wood

AC derived from the EUO-wood was synthesized via one-step activation using the same technology in previous work^[15]. The EUO-wood (10.00 g) was impregnated by 25 wt% phosphoric acid for 9 h with the solid to liquid ratio of 1:4. Then, the 12 h-drying desiccative sample was transferred to a tubular furnace to prepare AC under certain conditions of pyrolysis temperature (400 °C), heating rate (5 °C/min), holding period (6 h), and N₂ flow rate (600 mL/min). After that, the AC sample was rinsed with deionized water to be neutral for further utilization.

Preparation of modified AC by solvothermal method

The preparation of modified activated carbons (MACs) was conducted by a solvothermal method under different conditions of the heating temperature (120, 140, and 160 °C) and the holding time (18, 24, and 30 h) [Figure 1A]. The modification solution was prepared by dissolving FeCl₃·6H₂O (3.917 g), FeSO₄·7H₂O (2.000 g) and CoCl₂·6H₂O (5.235 g) in deoxidized and deionized water (100 mL). The AC precursor (5.000 g) was mixed with the modification solution using the magnetic stirrer at 90 °C for 1 h, and then NaOH powder (4.0640 g) was blended in the above suspension for 2 h. The obtained precipitate was transferred in the stainless reaction kettle lined with a tetrafluoroethylene tank (100 mL volume), which was subsequently heated in the muffle furnace at the determined temperature for 24 h. Finally, the solid products were washed with deionized water and ethyl alcohol to be neutral, and the samples were named 120MAC24, 140MAC24 and 160MAC24, respectively. Additionally, a part of the above precipitate was also heated in the same heating way at 140 °C for 18 and 30 h, which were labeled as 140MAC18 and 140MAC30, respectively.

Figure 1. (A) The schematic illustration of magnetic carbon synthesized via the hydrothermal method; (B) N₂ adsorption-desorption isotherms; (C) pore diameter distribution; (D) XRD; and (E) Roman spectrum of AC and MACs. XRD: X-ray diffractometer; AC: activated carbon; MACs: modified activated carbons.

Characterization of AC and MACs

The porous structures of AC and MACs were characterized by scanning electron microscopy and energy dispersive X-ray Detector (SEM-EDX) (TM3000, Hitachi, Japan)^[29] and transmission electron microscope (TEM) (Talos F200X S/TEM)^[15]. The Brunauer-Emmett-Teller (BET) surface area, Barrett-Joyner-Halenda (BJH) pore size distribution, and pore volume were measured following the multipoint N₂ adsorption method using Tristar II 3020 (Micromeritics, Shanghai, China)^[30]. The phase structure and the surface elemental compositions were identified by an X-ray diffractometer (XRD) with a Cu Kα radiation (λ = 1.5418) in 2θ range from 10° to 80° (D8 ADVANCE A25, Bruker, German) and X-ray photoelectron spectroscopy (XPS) (Thermo Fisher K-Alpha, England), respectively^[18]. XPS spectra were analyzed using Avantage Thermo software (Version 2.9.9.9). The Fourier transform infrared spectroscopy (FTIR) (Vertex70, Bruker, German) with a wavenumber range from 4,000 to 400 cm^-1 at a resolution of 4 cm^-1 was analyzed to characterize the functional groups. The Raman spectra of samples were recorded at ambient temperature on a Raman microscope (Horiba JobinYvon, Longjumeau, France) equipped with a confocal microscope (Olympus BX51, Tokyo, Japan) and a motorized x and y stage with an argon-ion laser at an excitation wavelength of 532 nm^[31].

Measurements of electrochemical and supercapacitive performance

AC or MACs were the active material in the electrode, which contained 80 wt% of AC or MACs, 10 wt% carbon black (SUPER-P, TIMCAL) and 10 wt% of polytetrafluoroethylene (PTFE, ALDRICH). The electrochemical performance measurement was conducted in the three-electrode system. The mixed films covered on stainless steel mesh were used as working electrodes in which the active material weight reached 3.2 mg. The platinum foil was employed as the counter electrode, the Hg/HgO electrode was taken as the reference electrode, and 6 M KOH was the electrolyte. Then, the supercapacitive properties were evaluated in the 2032 button cell system. In detail, the mixed films with the active material weight of 2 mg/cm² were covered in copper foil. The AC film was taken as the cathode, and the MAC film was used as the anode. Meanwhile, 6 M KOH was the electrolyte, and glass fiber was the separator.

All the cyclic voltammetry (CV) experiments were conducted on an electrochemical working station (CHI660E, Shanghai, China). The potential window ranged from -1.0 to +0.2 V in the alkaline electrolyte (6 M KOH). The electrical impedance spectroscopy (EIS) technique was performed on an electrochemical workstation (Zahner Im6ex), and the frequency ranged from 100 kHz to 10 mHz with an alternating current (ac) signal amplitude of 5 mV at open circuit potential. Galvanostatic charge-discharge (GCD) experiments were conducted using a Land Battery test System (Land, China) and the charging-discharging current densities ranged from 0.2 to 5 A/g. The specific capacitance (Cs, F/g), energy density (E, Wh/kg), and power density (P, W/kg) were calculated by

where I was the current (A), m was the weight of activated material (g), Δt was the charge and discharge time, Q was the fitting area of CV curves, and ΔV was the voltage window. The CV curve of the electrodes was obtained at different scanning rates (5, 10, 20, 50 mV/s), and the potential window ranges from -1.0 to 0 V. The EIS data were collected at a frequency between 100 kHz and 10 mHz with an ac signal amplitude of 5 mV.

The economic assessment of synthesizing AC and MAC

The economic assessment of the carbon products by different preparation processes was conducted based on unit of specific capacitance divided by cost of synthesis and an operation period^[32]. The cost involved expenditures for the electricity and material consumption but did not include the cost of equipment, building, labor, and overhead production^[29]. In detail, the parameters of synthesizing different products were listed in Supplementary Tables 1 and 2.

The structure and morphological characterization of AC and MACs

N₂ adsorption-desorption isotherms of the AC and MACs increased sharply at low and high relative pressures, indicating the major microporous and macroporous structures in these materials, respectively [Figure 1B]^[33]. The hysteresis loop occurring under medium relative pressure conditions appeared on the curves due to the existence of mesopores^[34]. Among all solvothermal products, the isotherm of 140MAC24 appeared to have the most distinct shape, implying the greatest mesopore distribution, which was confirmed by Figure 1C. The porous properties of all samples showed that the SSA of AC (1,503.70 m²/g) was higher than that of MACs, which may be attributed to the blockage caused by metal particles [Table 1]. Among MACs, 140MAC24 has the highest SSA of 1,377.69 m²/g, which was superior to that of 120MAC24 (1,133.07 m²/g) and 160MAC24 (1,355.56 m²/g). This phenomenon implied that proper enhancement of the solvothermal reaction temperature is conducive to the development of the micropore structure and excessively high reaction temperature could cause environmental pressure, resulting in pore collapse and damage to porosity. Meanwhile, by comparing the group of MACs synthesized under the same solvothermal temperature with different heating times, the SSA values of 140MAC18 and 160MAC18 were higher than those of 140MAC30 and 160MAC30 but lower than those of 140MAC24 and 160MAC18. The results demonstrated that increasing heating time could exert a helpful influence on pore development. However, prolonging reaction time could provide more energy for creating channel shells, resulting in the micropore fracture and the macropore formation. This could explain the better priority of 160MAC24 than 120MAC24 as the larger heating and cooling heating time was conducive to developing the pore structure^[35]. Among all MAC samples, the outstanding porosity of 140MAC24 could undoubtedly promote the diffusion and transference of the electrolyte ions to the interior of the electrode. Meanwhile, the weak transfer resistance could facilitate the charge storage^[36].

The graphitization extent of the AC and MACs was evaluated using Raman spectroscopy [Figure 1D]. The presence of two distinct peaks at 1,342 and 1,581 cm^-1 may be attributed to the D- and G-band, respectively. These bands were indicative of the disordered carbon and ordered graphite crystalline structures, as supported by previous research^[37]. The presence of a wide peak seen at around 2,750 cm^-1 was ascribed to the 2D-band, which signified the existence of layered carbon structures^[38]. The value of I_D/I_G, calculated as the ratio of the intensities of both bands, could indicate the extent of graphitization and the conductivity of the material^[31]. The I_D/I_G value of MAC was significantly higher than that of AC due to the existence of Co and Fe compound NPs. Among all MAC samples, 140MAC24 exhibited the highest graphitization degree (I_D/I_G = 0.9727), which could be consistent with the superiority of the priority mentioned above. Moreover, this phenomenon implied the great defects of the porous structure and the disordered structure in 140MAC24, which was conducive to providing more contact area for the electrolyte ion transport, increasing the material wettability, and further enhancing the charge storage^[37]. In addition, the I_D/I_G values of MACs prepared under different conditions presented the similar tends to the above BET results.

The XRD spectra of AC and MACs were characterized by the diffusion peaks similar to those of the carbon structure and CoFe₂O₄ (PDF#46-1543) [Figure 1E]^[27], respectively. As can be seen from the spectra, AC exhibits typical characteristic peaks [C (002) diffraction peaks] centered at 24° with a low peak height and wide peak width, which belongs to a typical amorphous carbon structure^[38]. The XRD pattern of CoFe₂O₄ presents characteristic peaks at 2θ = 18.46°, 30.29°, 35.55°, 43.40°, 53.76°, 57.22° and 62.79° on the spectrum of solvothermal products, which corresponded separately to the (111), (220), (311), (400), (422), (511), and (440) planes^[39]. Emphatically, the synthesis conditions of solvothermal temperature and heating time exerted a nonnegligible effect on the surface energy of the CoFe₂O₄ crystal nucleus^[15]. The spectrum of 140MAC24 was observed with the sharpened and narrow diffraction peaks of CoFe₂O₄, implying the highest development degree of crystalline^[23]. Meanwhile, the weak and broad characteristic peaks on the 140MAC30 spectrum may be caused by the continuous variation of lattice spacing during the excessive reaction phase^[15]. Additionally, the cobalt ferrite located on the MAC surface may tend to exist in the various grain sizes with the different crystallinity^[15].

Figure 2A showed the FTIR spectra of AC and MACs in a wavelength range from 4,000 to 400 cm^-1. The broad band at 3,430 cm^-1 corresponded to the stretching oscillation of the -OH group, which could improve the material surface wettability^[29]. The presence of C=O (1,603 cm^-1) and C=C (1,535 cm^-1) could generate the -OH group^[40]. Among all samples, the peak strength of -OH on the 140MAC24 spectrum was the highest, implying the low transmission resistance of electrolytes and eminent electrochemical performance in supercapacitors^[41]. Additionally, the rise of the solvothermal temperature may facilitate the depolymerization of lignocellulose, but too high temperature and heating time could cause the transformation of aliphatic groups to aromatic structure. Furthermore, the diffraction band at about 599 cm^-1 was related to stretching vibrations of Fe-O, suggesting the successful loading of iron elements^[29].

Figure 2. The FTIR spectrum and XPS spectra of AC and MACs. (A) The FTIR spectrum of AC and MACs; (B) The full-scan survey spectra of MACs; (C) The C 1s spectra of 140MAC24; (D) The O 1s spectra of 140MAC24; (E) The Fe 2p_3/2 spectra of 140MAC24; (F) The Co 2p_3/2 spectra of 140MAC24. FTIR: Fourier transform infrared spectroscopy; XPS: X-ray photoelectron spectroscopy; AC: activated carbon; MACs: modified activated carbons.

In addition, XPS spectra were performed to detect the element composition with a focus on 140MAC24. Figure 2B showed the differences of the element information using high-solution spectra. The presence of Fe 2p and Co 2p peaks distinctly suggests that CoFe₂O₄ was successfully synthesized on the MAC surface. The highest relative contents (4.44% and 2.57%) were measured at 140MAC24 by semi-quantitative analysis [Supplementary Table 3 and Supplementary Figure 1]. Undoubtedly, the O/C atomic ratio (0.300) also reveals that 140MAC24 had the more active sites and abundance of oxygen-containing functional groups (OFGs) than other samples^[42]. This result implied the eminent surface wettability and highly efficient electron transfer, consistent with the FTIR analysis mentioned above. In depth, C-O (285.6 eV), C=O (288.5 eV) and O-H peaks (532.3 eV) were fitted by deconvolving from the spectra of C 1s and O 1s, which were conducive to increased electrolyte penetration and decreased transmission resistance [Figure 2C and D]^[43]. The peak at 530.6 eV corresponded to M-O-M (M=Co or Fe). The high peak strength of 140MAC24 might suggest the superior crystallin degree of CoFe₂O₄ NPs, which agreed well with the XRD analysis [Supplementary Table 4]. After fitting by Gaussian method, Fe (III) peaks (712.6 and 719.1 eV), Fe (II) peaks (714.4 and 726.8 eV), and its satellite peak (714.4 eV) were observed at the Fe 2p spectrum [Figure 2E and Supplementary Figure 2]. The high-resolution spectrum of Co 2p, presenting two spin-orbit doublet characteristic peaks named Co 2p_3/2 and Co 2p_1/2, was deconvoluted into peaks of Co (III) (787.9 eV), Co (II) (784.8 and 797.2 eV) and its satellite peaks (787.4 and 804.1 eV) [Figure 2F]^[15]. The coexistence of Co and Fe with several valent also demonstrated the successful preparation of CoFe₂O₄ NPs on the MAC surface.

There may be an important effect of solvothermal conditions on developing pore and CoFe₂O₄ NPs, and 140MAC24 was regarded as the focus and investigated emphatically. The microstructures of ACs and MACs were observed by TEM and SEM [Figure 3 and Supplementary Figure 3]. AC obtained a rougher surface and an obviously layered carbon structure, indicating the favorable porosity^[44]. Meanwhile, this structure benefited from the intense solvothermal reaction. The MAC samples exhibited the corrugated graphitic layer, resulting in the carbon structure rearrangements^[45]. Meanwhile, the self-aggregation of the CoFe₂O₄ NPs appeared in the complexes of solvothermal reaction, which may contribute to the electrolyte transfer. According to the growth model of directionally attached crystals^[15], the high-pressure condition of the solvothermal preparation technology could increase the surface energy of the crystal nucleus^[41]. This theory could explain the larger size of CoFe₂O₄ NPs observed in the MAC sample prepared at higher solvothermal temperature. The corresponding EDX mapping of Fe and Co elements also could reveal the well-location of CoFe₂O₄ NPs. The TEM images could verify the SEM analysis results that the agglomeration degree of CoFe₂O₄ NPs evolved more intensively in the products synthesized at higher solvothermal temperature and longer heating time [Figure 3]. In addition, the pores of 140MAC24 with different extension directions indicated that ions and charges could diffuse in the carbon material through several channels^[41]. Meanwhile, a distinct distribution of CoFe₂O₄ NPs was on the hole wall, which could provide imaginary capacitance in electrochemical applications^[4]. Therefore, the capacitive performance of all samples was subsequently evaluated as the active substance of supercapacitor electrodes and tested by electrochemical measurement.

Figure 3. The SEM images of AC and MACs (×2000), EDX mapping images and TEM images of 140MAC24. SEM: Scanning electron microscopy; AC: activated carbon; MACs: modified activated carbons; EDX: energy dispersive X-ray detector; TEM: transmission electron microscope; 140MAC24: modified activated carbon heated at 140 °C for 24 h.

The electrochemical performance of AC and MACs

The charging and discharging behaviors of the AC and MACs were evaluated in a three-electrode configuration with Hg/HgO as the reference electrode (the potential window ranging from -1.0 to 0.2 V), 6 M KOH solution as the electrolyte, and Pt as the counter electrode. The measurement experiment of electrochemical properties was conducted using CV and GCD. Among AC and MACs, 140MAC24 exhibited a broader integral area [Figure 4A] and longer charging-discharging interval [Figure 4B] under the same scanning rate and current density conditions. This phenomenon verified that the solvothermal loading-magnetic process could promote the electrochemical properties of carbon electrodes. Figure 4A displays the charging-discharging curves of samples obtained using the CV method at a scanning rate of 5 mV/s. The curve of AC presented a rectangular shape due to an electric double-layer capacitor (EDLC)^[46]. In contrast, the curves of MACs revealed the obvious redox peaks due to Faradaic reaction from Fe and Co ions, implying the generation of the pseudo capacitance^[47]. Furthermore, the integral area of MACs was obviously larger than that of AC, and 140MAC24 possessed the greatest capacitance (521.88 F/g) calculated by formula (2). The result was consistent with the above XPS findings, implying the contribution of OFGs and CoFe₂O₄ on enhancing the electrochemical characteristic^[48]. In addition, capacitance of materials escalated with the increasing solvothermal temperature and reaction time, but excessive synthesis temperature or duration could cause the collapse of the porous structure, weakening the electrochemical property [Table 2]. In Figure 4C, as the scanning rate ranged from 5 to 100 mV/s, the redox reaction peaks of CV curves gradually dampened, which was owing to the slowing ion diffusion and intercalation-deintercalation process along with the incomplete redox reaction of CoFe₂O₄ NPs at a rapid scan rate^[34]. Meanwhile, the surrounding region on the CV curve grew with a homologous shape as the scanning rate increased. The anodic and cathodic peaks severally move towards the positive and negative potentials, suggesting the abundant active sites on the electron surface followed with fast and reversible transportation of ionic and electronic^[49].

Figure 4. (A) The CV curves of AC and MACs at 0.005 V/s; (B) CV curves of 140MCA24 with different sweep rates; (C) GCD curves of AC and MACs at 0.2 A/g; (D) GCD curves of 140MAC24 with different current density; (E) EIS spectra for AC and MAC electrodes; and (F) long cycling performance of 140MAC24 electrodes. CV: Cyclic voltammetry; AC: activated carbon; MACs: modified activated carbons; 140MCA24: modified activated carbon heated at 140 °C for 24 h; GCD: galvanostatic charge-discharge; EIS: electrical impedance spectroscopy.

GCD could reliably evaluate the electrochemical capacitance performance of active substances. The GCD curves of materials at a current density of 0.2 A/g exhibited different inclination phenomena, indicating that the redox reactions occurred during the charging-discharging process^[31]. It was noteworthy that a distinct potential drop was ascribed to internal resistance (IR) in the initial stage of the GCD curves, directly reflecting the electrical conductivity of materials [Table 2]^[50]. Among all products, 140MAC24 possessed the minimum IR potential drop (0.049 V), attributed to its favorable mesoporous ratio (72.91%) and monolithic conductivity agreed with the above-mentioned Roman spectroscopy and XPS analysis. With the current density ranging from 0.2 to 1 A/g, the time span between charging and discharging gradually reduced, and the corresponding capacitance of 140MAC24 ranged from 571.36 to 229.36 F/g. Furthermore, there was a noteworthy tendency of the IR potential drop as the current density amplified [Figure 4D], attributed to fewer electrolyte ion diffusion resistance in mesopores under the high current density conditions^[49].

To reveal the potential mechanism of ion transfer, charge diffusion and capacitance generation, the EIS measurements were tested at open circuit potential in the frequency range of 10⁵-10^-2 Hz^[51]. As illustrated in Figure 4E, there were the typical impedance curves with a quasi-semicircle at the high frequency region and a sloped line at the low frequency region^[52]. Furthermore, the X-axis intercept of the plots in high frequencies (>10 Hz) often reflected the intrinsic ohmic resistance (Rs), and the diameter of the semicircle indicated the charge transfer resistance (Rct) while the angle of straight-line slope (value below 0°) at low frequencies implied the ions or electrolyte diffusion resistance [Table 2]^[53]. In general, mesopore structure could supply the expedited ion transfer channel and eminent charge diffusion kinetic for the surface OFGs and the redox reaction of loaded CoFe₂O₄ NPs, which could culminate in less resistance of material^[54]. Undoubtedly, 140MAC24-containing electrodes exhibited the best electrical conductivity, the least IR (Rs = 1.11 Ω and Rct = 1.27 Ω), and the fastest electron transfer (the angle = -71°), indicating a capacitive behavior of EDLC coupled with partial pseudo capacitance^[30].

The cyclical stability of electrodes could also play a vital role in the charging-discharging process, and the repeating GCD test for 10,000 cycles was conducted at a current density of 1 A/g. It was observed that 140MAC24 could display an excellent retention of 88.05% after 10,000 cycles [Figure 4F]. In detail, the capacitance of 140MAC24 electrodes initially reduced until about the 5,000th cycle. This result may be due to the irreversible decrease of effective electrochemical active area^[21]. Subsequently, a rising trend of property appeared, which could contribute to the gradual penetration of electrode solution. Therefore, 140MAC24 possesses great potential for supercapacitor electrodes with well-built stability performance.

In the practical application of devices for storing energy, the preparation cost was still a vital factor in estimating the actual performance of electrode material. Here, the basic economic assessment of AC and 140MAC24 was carried out to investigate the feasibility. Supplementary Table 5 showed the material and electricity costs, which were then added up and divided by the unit of specific capacitance. In detail, during the charging-discharging process, compared to AC, the charging-discharging of electrodes based on 140MAC24 could obviously save expenditure, which suggested the superior economic feasibility. This result was mainly due to the greater electrochemical properties of 140MAC24 than that of AC. Overall, preparing magnetic carbon material through a solvothermal heating method exhibited the splendid economic effectiveness for energy storage and could have the potential to significantly appeal to the capital market.

The electrochemical function mechanism of CoFe₂O₄

As can be seen in Figure 4B, the obvious asymmetry features on the charging-discharging curves of 140MAC24 during the charging and discharging process may be caused by the irreversible reaction of CoFe₂O₄^[27]. To investigate the electrochemical function mechanism of CoFe₂O₄, several three-electrode systems were employed using 140MAC24 as a working electrode, and the charge and discharge under the different potential windows for 1,000 cycles were conducted. The electrode was discharged at the current density of 0.2 A/g until the potential of -1.0 V, and subsequently charged to the desired potential, which was labeled as samples with charging process. The characteristics of electrodes were further assessed using XRD, Raman, and FTIR techniques. In particular, the CoFe₂O₄ diffraction peaks predominated the XRD spectrum of the 140MAC24 electrode at the initial stage of -1.0 V [Figure 5A]. Then, the diffraction peaks of Fe₃O₄, Fe₂O₃, and FeO appeared on the spectrum during the charging phases, and their intensities gradually increased with the expanded potential span until 1 V. In this case, iron compounds could function as accelerators for charging by changing into the oxide form (M^IIO or M^IIIO) and generating faradic capacitance. However, there were few differences between the spectra of the charging potential of 0 and 0.2 V. It indicated a slight degree of redox reaction and little amount of ion transfer^[37], which corresponded to the limited integral area of CV curves ranging from 0 to 0.2 V. In contrast, only the discharging region between 0.2 and 0 V exhibited obvious transformation of CoO diffraction peaks of the XRD spectrum, indicating the promoting effect of cobalt element during discharging. This phenomenon could explain why the CV curve section of discharge process included an arc of redox peak from 0.2 to 0 V and a line from 0 to -1.0 V. Additionally, the difference between diffraction peaks at the initial discharging and final charging phases could provide the evidence to the irreversible reaction of CoFe₂O₄ NPs.

Figure 5. The (A) XRD and (B) Raman spectrums of MAC electrode after working at different voltage windows. XRD: X-ray diffractometer; MAC: modified activated carbon.

According to the XRD results, CoFe₂O₄ NPs could transform into the oxide and hydroxide form of metal. This phenomenon may increase the disorder degree of electrodes, which could also be assessed using Raman methods. On the spectrum of the 140MAC24 electrode at a charging potential of -1.0 V, the value of I_D/I_G (0.9869) was higher than that of 140MAC24 (0.9727) [Figure 5B]. This result may be attributed to the mix with carbon black. As the charging potential enhanced from -1.0 to 0 V, CoFe₂O₄ NPs reacted more strongly, which brought out the higher disorder degree. Meanwhile, the change tendencies of I_D/I_G value in the charging process from 0 to 0.2 V, discharging process from 0.2 to 0 V, and discharging process from 0.2 to -1.0 V were also coincident with the results of XRD analysis. Besides, the generation of CoOOH may be responsible for the upward tendency of -OH group abundance on FTIR spectrum of the 140MAC24 electrode [Supplementary Figure 4]. Meanwhile, the penetration of electrolyte may lead to a slight increase of -OH group between the charging potential of 0 and 0.2 V^[38]. The strength of the -OH group and Fe–O bond decreased under the potential condition from 0.2 to 0 V during the discharge stage, which may be ascribed to the oxidation reaction of the oxide and hydroxide^[55]. Above all, CoFe₂O₄ NPs on the 140MAC24 surface could generate faradic capacitance by converting to the oxide and hydroxide form of metal elements in the charging stage of -1.0 to 0 V^[36], and the reaction products could conduct the redox reaction in the discharging stage of 0.2 to 0 V^[49]. Therefore, CoFe₂O₄ NPs underwent different degrees of redox reactions in asymmetric voltage intervals during the charge-discharge stages, resulting in the irreversible charge-discharge reactions and asymmetric charge-discharge curve features. On the whole, CoFe₂O₄ NPs showed an excellent ability to generate pseudocapacitance during the charging phase, indicating the great feasibility of using as anode material.

The supercapacitive properties of MAC

For evaluating the application of electrode-based 140MAC24, the supercapacitors were assembled in the form of coin batteries (CB). Generally, the operating potential window is essential for the performance of energy storing devices^[18]. Therefore, the CV measurement with various operating voltage windows (0-1.8 V) was conducted to obtain the proper condition, and the results were shown in Figure 6A. Specifically, a favorable association was between the amount of electric charge calculated by the integral area of CV region and the potential interval up to 1.8 V. The CV plots were measured at several scanning rates of 5-100 mV/s in the optimum working voltage window of 1.8 V. The CV curves showed distinct redox peaks, and the peak location gradually moved towards low potential as the scanning rate ranged from 5 to 100 mV/s [Figure 6B]. Meanwhile, the specific capacitance increased to the maximum of 171.31 F/g at the scanning rate of 5 mV/s, suggesting the outstanding reversibility of redox reaction and fast electron diffusion kinetic of faradic process^[41].

Figure 6. The supercapacitive properties of AC/MAC in the CB system. (A) The CV curves of CB measured at different voltage windows; (B) the CV curves of CB at different scan rates (5-100 mV/s); (C) the GCD curves of CB at different current densities; (D) the Ragone plot CB compared with other studies; (E) the cycling stability of CB; and (F) the electrochemical impedance spectroscopy of CB^{[10,12,16,35]}. AC: Activated carbon; MAC: modified activated carbon; CB: coin battery; CV: cyclic voltammetry; GCD: galvanostatic charge-discharge.

The charging-discharging feature was analyzed at the various current densities, with the curves represented in Figure 6C. The peaks of the redox reaction were linked to the modest change in the symmetrical shape of the curves. This characteristic indicated the stable structure performance of electrodes in the charging-discharging process. Moreover, an obvious potential drop appeared on the discharging stage under the different current densities, which was up to 1.8 V at the current density of 0.2 A/g. The supercapacitive property of fabricated devices had been evaluated using formula (3), which displayed the highest value of 156.60 F/g. The specific capacitance was lower than that of a single electrode, and the potential drop in a button battery was around double that of a three-electrode system. This effect could result from the resistance of the separator and electrode, which limited the ion and charge transfer^[19]. Then, the energy density was estimated and compared with the studies about energy storing devices [Figure 6D]. The higher values of energy densities calculated at the current density of 0.2 A/g were up to 73.36 Wh/kg, higher than those supercapacitors assembled with pure or modified carbon^[45,55-60]. Meanwhile, the assembled supercapacitor conducted the CGD test for 10,000 cycles, and the capacitance could maintain a retention rate of 82.49% with a gradual decrease trend [Figure 6E].

The EIS spectra of a coin system were conducted to investigate the physical properties such as transfer kinetics of electrolytes and intrinsic resistance of supercapacitors. The results were represented in the form of a Nyquist plot [Figure 6F], which resembled the equivalent circuit in the three-electrode configuration. In detail, the charge transfer resistance of assembled devices (1.5 Ω) was similar to that in the single electrode system, which demonstrated the outstanding kinetic of ion transfer^[34].