Facile Synthesis of Templated Activated Carbon from Cellulose Nanofibers and MgO Nanoparticles via Integrated Carbonization-activation Method as an Eco-friendly Supercapacitor

Abstract

A green, rapid, and facile synthesis method of activated carbon was developed and optimized using 50 nm MgO nanoparticle (NP)-based pore templating in tandem with an integrated drying and carbonization-activation heat treatment. TEMPO-oxidized cellulose nanofibers (TOCN) were mixed with MgO NPs and KOH, freeze-dried, heat treated first at 350 °C then at 800 °C to get templated activated carbon. Selective pore formation by MgO and KOH resulting in the development of mesopores and micropores, respectively was confirmed. NP-based templating is confirmed to be viable even with continuous heat treatment, and does not affect any transformations during each heat treatment phase. Through optimization for the highest surface area and optimal hierarchical porosity, templated activated carbon had a high specific capacitance of 269 F g⁻¹ at 0.5 A g⁻¹ with 95.92 % cycle stability after 2000 charge-discharge cycles at 10 A g⁻¹.

1. Introduction

Due to the shift towards sustainable production of carbon materials, biomass-based activated carbon (AC) for supercapacitor electrode applications that use less energy-intensive synthesis methods as well as utilize sustainably sourced raw resources have garnered attention in recent years.¹ AC synthesis starts with carbon materials as its precursor that is mixed with an activating agent such as KOH,^2,3 or CO₂,⁴ to be heated at higher temperatures.⁵ When starting from biomass, the synthesis process becomes long and energy-intensive because converting biomass to activated carbon would then take 2 heat treatment steps, thus making biomass-based AC production less environmentally friendly. An alternative and more eco-friendly method is performing both carbonization and activation within one heat treatment cycle.^6,7 AC synthesized via one-step carbonization-activation has been reported to have nearly identical or had higher surface area than that of AC synthesized through the conventional two-step method.^8,9

Whether AC is synthesized via one-step or two-step, the addition of porogens to induce the desired pore structure is still required. Micropore formation could easily be integrated in production through the addition of the aforementioned activating agents. However, meso-macroporosity is often dependent on the inherent structure of its organic precursor^10,11 such as the existing plant fiber,^10,12 thus requiring techniques that intentionally induce mesopore formation. Other studies have reported phase separation on polyacrylonitrile (PAN)^13,14 to create hierarchical structures that are retained after carbonization. The method however is not applicable to all types of carbon precursors, preventing scalability and production flexibility. Another method that can be more generally utilized for biomass is nanoparticle templating, wherein a removable particle is embedded in the carbon precursor before heat treatment. When the template is removed, pores that match the size of the template would be formed. Using polymer nanoparticles (NPs)¹⁵ or surfactant micelles¹⁶ have been reported, and the organic template undergoes thermal degradation during carbonization. However, polymer NPs require an additional interface polymer^15,17 that binds the template and the matrix to ensure proper templating. Meanwhile, metal oxides (MOx) such as SiO₂,¹⁸ zeolites,¹⁹ ZnO,²⁰ and MgO^21,22 that range from nanospheres to nanorods, are more mechanically and thermally stable than polymer NPs, and do not need additional interface stabilizers. MOx NPs could then easily be removed by washing the carbon material in acid such as HCl.²³

When synthesizing templated biomass-derived AC, the carbon precursor needs to have good affinity to the template to ensure that the template stays in the matrix until templating occurs.¹⁵ However, untreated biomass such as wood^24,25 and animal chitin shell polymers^26,27 have tightly packed fibers or polymer chains, preventing proper template embedding.²⁸ Other materials have been used, ranging from monosaccharides such as glucose²⁹ or fructose,³⁰ to raw biomass such as garlic peel,³¹ pomelo peel,¹⁰ or bamboo,¹² but in our previous study, we have reported that using nanofibers as a precursor could physically entangle⁸ to form networks that trap and stabilize NPs,¹⁷ making nanofibers work well with templating. Nanofibers ranging from 5–50 nm have been synthesized through TEMPO oxidation^32,33 results in better interaction between the matrix and the template,¹⁷ resulting in higher template loading and better template-pore fidelity.

Thus, in this work, an optimized, green, and facile synthesis method for templated activated carbon (TAC) from TEMPO-oxidated cellulose nanofibers (TOCN) with dispersed 50 nm MgO NPs and KOH via an integrated MgO drying and one-step carbonization-activation heat treatment method is presented (Scheme 1). The effects of mesopore and micropore-inducing agents MgO and KOH, and activation temperature, were investigated and optimized. Immediately introducing MgO and KOH in the TOCN, and performing carbonization and activation in direct sequence as one prolonged heat treatment phase resulted in a fast and facile method in synthesizing TACs with high surface area. MgO and KOH pore formation do not affect one another, showing selective development in AC hierarchical porosity. Cyclic voltammetry (CV) and galvanostatic charge-discharge (GCD) measurements show that TACs have high specific conductivity, with potential use in supercapacitor applications.

Scheme 1.

Synthesis of TACs via continuous, one-step carbonization activation of cellulose nanofibers.

2. Experimental

2.1 Reagents

The list of reagents used are as follows: cellulose powder (Nacalai Tesque), TEMPO (Tokyo Chemical Industry), sodium bromide (Sigma-Aldrich), 8.5–13.5 % sodium hypochlorite solution (Nacalai Tesque), 0.5 mol/L NaOH solution (Sigma-Aldrich), 99.9 % 0.05 µm magnesium oxide (Wako Pure Chemical Corporation), potassium hydroxide pellets (Wako Pure Chemical Corporation), hydrochloric acid (Wako Pure Chemical Corporation) 0.8 % Nafion solution prepared from 5 % Nafion solution (Wako Pure Chemical Corporation) diluted with isopropanol, carbon black (Strem Chemicals. Inc.), and sulfuric acid (Wako Pure Chemical Corporation). All reagents were used as received.

2.2 Synthesis of TEMPO-oxidized cellulose nanofibers (TOCN)

The synthesis of TOCN was based on our previous study.¹⁷ Cellulose (4.0 g), TEMPO, (0.075 g), and sodium bromide (0.768 g) were dispersed in 300 mL water and stirred for 2 h until homogeneous. Sodium hypochlorite (18 mL) was added to the mixture dropwise while maintaining a maximum pH of 10.5. The reaction was allowed to continue overnight at ambient temperature while maintaining a constant pH of 10.5 by using a Hiranuma Auto-titrator COM-1600 with 0.5 mol L⁻¹ NaOH as the titrant. When the reaction was complete as indicated by a color change of the cellulose from yellow to white, the cellulose pulp was isolated from the liquid via centrifugation at 4000 rpm for 5 min. To purify the cellulose, the pulp was redispersed in water, and again centrifuged at 4000 rpm for 5 min. The process of centrifugation and redispersion was repeated four times. The purified pulp was then diluted to 1 wt% and fibrillated using a Waring Laboratory Blender 7010HS until cellulose nanofibers were fully dispersed in the water to form the TOCN dispersion.

2.3 Synthesis of templated activated carbon (TAC)

In a typical synthesis, 10 g of TOCN dispersion (corresponding to 100 mg dry cellulose), MgO with average size 52.6 ± 5.9 nm (Fig. S1), KOH, and water for a total volume of 15 mL were placed in a 50 mL conical tube, and shaken until the KOH was fully dissolved, and the MgO was homogeneously dispersed. The mixture was freeze-dried for 48 h, yielding cellulose monoliths (CM). CMs were then loaded into an electric furnace under Ar atmosphere with a constant Ar flow rate of 300 mL min⁻¹. The heat treatment was performed as follows: using a heating rate of 5 °C min⁻¹, the temperature was raised to 350 °C and held for 1 h, followed by a heating rate of 10 °C min⁻¹ to raise the temperature to 800 °C held for 1.5 h. The samples were then cooled slowly overnight. The activated carbon with MgO were washed with 6 mol L⁻¹ hydrochloric acid to remove excess KOH as well as dissolve the MgO template, forming TACs. The TACs were washed with water until neutral pH, and dried in an oven at 85 °C overnight. The list of experiments and their corresponding conditions are listed in Table 1.

Table 1. List of experiments performed and the TAC label code for the corresponding conditions with constant 100 mg TOCN content.

Code	Mass of KOH/ mg	Mass of MgO/ mg	Activation temperature/ °C
Studies on the effect of MgO and KOH on pore formation
KxM30	—	300	800
K5Mx	50	—	800
Variation of KOH amount
K2	20	100	800
K5 (= M10)	50	100	800
K7	70	100	800
K10	100	100	800
Variation of MgO amount
M5	50	50	800
M10 (= K5)	50	100	800
M30 (= C800)	50	300	800
M50	50	500	800
Variation of Activation Temperature
C600	50	300	600
C700	50	300	700
C800 (= M30)	50	300	800

2.4 Characterization

Scanning Electron Microscopy (SEM) micrographs were taken using a Hitachi SU3500 microscope. Thermogravimetric analysis (TGA) was conducted using a SII TG/DTA7200 instrument. Brunauer-Emmett-Teller (BET) surface area analysis was performed using a Quantachrome NOVA 4200e Surface Area & Pore Size Analyzer. Barrett-Joyner-Halenda (BJH) and Density Functional Theory (DFT) pore size analysis methods were also performed through the same software used for the BET. All samples were vacuum dried at 90 °C for 24 h prior to analysis. X-ray Photoelectron Spectroscopy (XPS) spectra were taken using a JEOL JPS-9010MC XPS instrument with monochromatic AlKα-radiation. CasaXPS Version 2.3.15 was used in analyzing the C1s narrow XPS profile. Raman spectrographs were taken using a Jasco NRS-3100 Laser Raman Spectrophotometer (532.190 nm).

2.5 Electrochemical studies

A three-electrode system was used to measure the electrochemical specific capacitance of TACs. The system consisted of an Ag|AgCl reference electrode (RE-1B, BAS) with an internal solution of 3 mol L⁻¹ NaCl, a Pt wire counter electrode, and a glassy carbon electrode (diameter = 3 mm, BAS), with TAC-based carbon as the working electrode, and a 1 mol L⁻¹ sulfuric acid solution as the electrolyte. The working electrode was prepared by first grinding 200 mg TAC into a powder (d ≤ 45 µm), then dispersing the TAC powder, and 5 wt% carbon black in a 0.8 % Nafion and isopropanol solution to obtain a 30 mg mL⁻¹ casting solution. 1.5 µL of the solution was cast on the glassy carbon part of the electrode, and then air-dried for 24 h to give a total TAC loading mass of 45 µg and an average disc thickness of 12.7 ± 4.2 µm. Prior to measurement, the working electrode was immersed in the electrolyte solution and placed under vacuum at 5 min intervals five times to ensure deep penetration of the electrolyte and the removal of bubbles on the electrode surface. The entire system was purged with Ar and kept at a constant temperature of 25 °C throughout the measurement. All measurements were performed using an ALS/CH Model 600 electrochemical analyzer, with the program ALS/CH 7002e used for data collection.

CV was performed at varying scan rates (50, 100, 200, 300, 500 mV s⁻¹) with a potential window of −0.2–0.8 V and a rate of 10 cycles per scan. GCD measurements were performed at varying current densities i_m (0.5, 1, 2, 3, 4, 5 A g⁻¹), with a potential window of −0.2–0.8 V. Because all samples exhibited near-linear charge-discharge behavior, specific capacitance C_sp was calculated from the GCD spectra using the equation below (Eq. 1),   

\begin{equation} C_{\text{sp}} = I\Delta t/\Delta Vm = i_{\text{m}}\Delta t \end{equation} (1)

where I is the current (A), Δt is the time for one charge cycle (s), ΔV is the voltage range (V), and m is the mass of the active material (g). The equation can be further simplified to Δt multiplied by i_m, because the potential window is 1.0 V.

3. Results and Discussion

3.1 Characterization

Even though only one continuous heat treatment is performed on the CMs, a number of transformations occur to form TAC. Using TGA (Fig. 1), it was determined that TOCN undergoes the dehydration at 219 °C,¹⁷ then subsequent degradation at 316 °C and 363 °C.^34,35 MgO dispersed in TOCN turned into Mg(OH)₂, which then underwent dehydration at 209 °C (Fig. S2a) and 363 °C to form back into MgO.³⁶ Mass loss from further pyrolysis occurs at 448 °C, and lastly 715 °C mass loss from KOH activation through gaseous evolution of CO₂ and K₂O³⁷ also shown in the thermal decomposition TGA of KOH at Fig. S2b.

Figure 1.

TGA curve of TAC-K5.

To determine the effect of MgO and KOH on the pore formation mechanism of TAC, samples with no MgO (TAC-K5Mx) and no KOH (TAC-KxM30) were synthesized. From the SEM micrographs, TAC-K5Mx showed macropores (Fig. 2a) naturally forming from carbonized TOCN, but not mesopores. Meanwhile, the addition of MgO with TAC-KxM30 (Fig. 2b) resulted in the formation of mesopores, but no macropores were present in TAC-K5Mx. The BET curves of both samples are in agreement with the SEM (Fig. 2c) as shown by the significant difference in the absorbed N₂ at the 0.0–0.2 p/p_o region, and the lack of a hysteresis loop for TAC-K5Mx. The DFT distribution (Fig. 2d) also highlights that TAC-KxM30 does not have any significant degree of microporosity. The small degree of microporosity observed for TAC-KxM30, evidenced by BJH micropore volume (V_micro) at 0.18 cm³ g⁻¹ (Table S1) is likely the effect of carbonizing TAC-KxM30 at 800 °C, resulting in some micropore formation.

Figure 2.

SEM micrographs of a) TAC-K5Mx and b) TAC-K5M30 and the corresponding c) BET curves and d) DFT pore size distribution.

3.2 Effect of KOH and MgO

After confirming that the pore formation mechanism of KOH and MgO are mutually exclusive, the amount of each pore inducing agent was optimized. The amount of KOH was adjusted to 20–100 mg, which was determined by predicting the eventual KOH : carbon impregnation ratio.^6,38 Studies that perform carbonization and activation separately have commonly reported using 1 : 1 to 4 : 1 KOH : carbon.^6,39 Because the average CM mass yield from 100 mg TOCN is 20 ± 4 mg, the amount of KOH was adjusted to 20, 50, 70, and 100 mg corresponding to 1 : 1, 2.5 : 1, 3.5 : 1, and 5 : 1 KOH : carbon.

As shown in the SEM micrographs (Figs. 3a–3d), the effect of KOH on both micro- and mesopore structure is dependent on the amount of KOH. TAC-K2 and TAC-K5 show a similar morphology as that of TAC-K5Mx (Figs. 3a and 3b), thus KOH does not contribute to meso-macropore formation. When the KOH amount is increased to 70–100 mg (TAC-K7 and TAC-K10, respectively), the surface morphology exhibits a fine porous structure (Figs. 3c and 3d) which was likely formed as excessive amounts of KOH led to formation of KOH crystals that then served as another templating material.⁴⁰ This is evidenced by the increase in the meso-macropore volume (V_meso-macro) from 70.0 % of TAC-K5 to 73.5 % and 73.9 % of TAC-K7 and TAC-K10, respectively (Table 2). However, excessive template loading leads to pore and network collapse,¹⁷ resulting in lower surface area (S_BET) from 1611 m² g⁻¹ (TAC-K5) to 1407 m² g⁻¹ (TAC-K10). Signs of pore uniformity were also seen in the BET curves (Fig. 3e), as shown by TAC-K2 and TAC-K5 exhibiting Type IV H1/3 hysteresis loop. TAC-K7 and TAC-K10 meanwhile exhibit a Type IV H3 hysteresis loop, indicating less pore uniformity. TAC-K5 also has a S_BET of 1611 m² g⁻¹ (Table 2). To prevent uncontrolled secondary templating while still having the highest S_BET, 50 mg KOH was selected as the optimal KOH amount.

Figure 3.

SEM micrographs of a) TAC-K2, b) TAC-K5, c) TAC-K7, and d) TAC-K10, and their respective e) BET curves. f) Raman spectra of TAC-K5.

Table 2. S_BET, BJH pore volume analysis, and XPS elemental composition of TAC with varying amounts of KOH.

Sample	SBET/	BJH Vtotal/	BJH Vmicro/		BJH Vmeso-macro/		XPS
	m2 g−1	cm3 g−1	cm3 g−1	%	cm3 g−1	%	C at %	O at %	O/C
TAC-K2	1102	0.97	0.29	29.6	0.68	70.4	86.35	13.65	0.158
TAC-K5	1611	1.88	0.56	30.0	1.32	70.0	88.07	11.93	0.135
TAC-K7	1560	1.46	0.39	26.5	1.08	73.5	86.24	13.76	0.160
TAC-K10	1407	1.85	0.48	26.1	1.36	73.9	86.69	13.31	0.154

Varying the amount of KOH from the tested range does not seem to have a clear effect on the ratios of the carbon phases observed via Raman spectroscopy (Table S2). Peak fitting of the Raman spectra for TAC-K5 (Fig. 3f) however, shows that TAC shows a relatively high degree of amorphousness as observed from the P-band (1190 cm⁻¹) and Am-band (1520 cm⁻¹). All TAC samples had similar values for I_D/I_G, as well as I_P/I_G and I_Am/I_G, indicating that both graphitic and amorphous sections of TAC were unaffected.

The same observation as made using XPS, where the relative C % and O % did not vary (Table 2) despite higher amounts of KOH. Analysis of both C1s and O1s narrow spectra (Table S3) likewise show similar ratios for all deconvoluted peaks.

Increasing the amount of MgO results in higher degree of mesoporosity. SEM micrographs of TAC-M5 to TAC-M50 (Figs. 4a–4c) show an increasing number of pores made by MgO especially with honeycomb-like porosity for TAC-M30 (Fig. 4b). The DFT pore size distributions (Fig. 4d) confirm the increase in mesoporosity as a result of increased MgO loading. The mesopores range from the size of 5–30 nm, and all peaks consistently grow in proportion to the amount of MgO. However, the pore distribution curve peaks of TAC-M30 and TAC-M50 having similar intensities indicate that the degree of mesoporosity does not further increase when over-templating occurs. The BET curves (Fig. 4e) likewise show a similar increase in the size of the hysteresis loop at the 0.5–1.0 p/p_o region. The 0–0.2 p/p_o region of Fig. 4d however does not change, again showing that MgO does not affect microporosity.

Figure 4.

SEM micrographs of a) TAC-M5, b) TAC-M30, and c) TAC-M50, and their respective d) BET curves and e) DFT pore size plots. Note that TAC-M10 is the same as TAC-K5.

The optimal amount of MgO that could be loaded into TOCN is observed to be at 300 mg, with the highest S_BET observed at 1711 m² g⁻¹ (Table 3). V_total and V_meso-macro increases proportionally with MgO amount, but template overloading occurred at 500 mg as shown by the disappearance of a clear pore structure (Fig. 4c) and decrease in S_BET for TAC-M50.

Table 3. S_BET, BJH pore volume analysis, and XPS elemental composition of TAC with varying amounts of MgO.

Sample	SBET/	BJH Vtotal/	BJH Vmicro/		BJH Vmeso-macro/		XPS
	m2 g−1	cm3 g−1	cm3 g−1	%	cm3 g−1	%	C at %	O at %	O/C
TAC-M5	1626	1.15	0.44	37.9	0.72	62.1	83.46	16.54	0.198
TAC-M10	1611	1.88	0.56	30.0	1.32	70.0	88.07	11.93	0.135
TAC-M30	1711	2.80	0.60	21.4	2.20	78.6	84.99	15.01	0.177
TAC-M50	1574	2.92	0.58	19.9	2.34	80.1	87.34	12.66	0.145

Similar to that of KOH, the amount of MgO did not have a clear effect on the I_D/I_G ratio of the TAC samples (Table S4), while there is an increase in the amount of amorphous carbon from TAC-M5 to TAC-M30 evident from the I_P/I_G and I_Am/I_G peaks. The values for all Raman bands begin to plateau with TAC-M50, indicating that above 300 mg, MgO starts to affect only the surface area. Likewise, the amount of MgO does not have a clear effect on the elemental composition of TAC, as shown in the XPS elemental composition C % and O % (Table 3), but the increased mesoporosity resulted in the slight increase of C=O bond formation as seen in the C1s narrow spectra rising from 17.7 % of TAC-M5 to 31.2 % of TAC-M30 (Table S5).

3.3 Effect of activation temperature

After optimizing the amount of KOH and MgO, the effect of activation temperature on TAC-M30 was investigated. The mesopore structure of TAC treated from 600–800 °C remained the same when observed with SEM (Figs. 5a and 5b, Fig. 4b). S_BET however significantly increased from 731 m² g⁻¹ of TAC-C600 to 1711 m² g⁻¹ of TAC-C800 (Fig. 5c, Table 4). Likewise, V_total, V_micro, and V_meso-macro all increased for TAC-C800, while the relative ratio of V_micro to V_meso-macro stayed the same. Therefore, the overall porosity of TAC is affected by the final hold temperature of the heat treatment. DFT pore distribution also shows the same trend with all peaks increasing with higher temperature (Fig. 5d). The increase in microporosity (Fig. 5c) can be explained by the temperature-dependent reactions of KOH that result in surface etching. Below 700 °C KOH forms K₂O and K₂CO₃,¹⁰ but reactions where K₂CO₃ react with more carbon to form CO or CO₂ only occur above 700 °C, thus etching the surface to form the micropores.⁵ Activation further progresses with higher temperature, shown by TAC-C800 having the highest surface area.

Figure 5.

SEM micrographs of a) TAC-C600 and b) TAC-C700. TAC-C600 to TAC-C800 characterization through c) BET, d) DFT pore size distribution, and e) Raman spectra. Note that TAC-C800 is the same as TAC-M30.

Table 4. S_BET, BJH pore volume analysis, and XPS elemental composition of TAC activated at varying temperatures.

Sample	SBET/	BJH Vtotal/	BJH Vmicro/		BJH Vmeso-macro/		XPS
	m2 g−1	cm3 g−1	cm3 g−1	%	cm3 g−1	%	C at %	O at %	O/C
TAC-C600	731	0.957	0.24	25.2	0.72	74.8	83.46	16.54	0.198
TAC-C700	1196	1.653	0.31	19.0	1.34	81.0	88.07	11.93	0.135
TAC-C800	1711	2.798	0.60	21.4	2.20	78.6	84.99	15.01	0.177

Higher activation temperature also resulted in the increase in I_D/I_G ratio (Table S6, Fig. 5e) from the increasing D-band. This could either indicate that more edge sites of the graphitic basal plane are being exposed,⁴¹ or that there is an increasing number of defects on the basal plane from increased KOH etching. The relative phases of amorphous carbon associated with the I_P/I_G and I_Am/I_G are similar for TAC-C700 and TAC-C800.

The elemental composition of the TAC did not change significantly despite the elevated activation temperature (Table 4, Table S7). Slight decrease in O content, possibly from the evolution of CO and CO₂ during activation, was observed. Higher ratios of C=O and lower C-O percent (Table S7) for TAC-C700 and TAC-C800 compared to TAC-C600 indicate oxidation as the primary route for O release at high temperature.

3.4 Electrochemical studies

The performance of TACs activated at different temperatures as potential supercapacitors were studied. From the CV curves in Figs. 6a–6c, electrochemical capacitance increased with increasing temperature like from the following factors: 1) increase in S_BET, 2) better-developed hierarchical porosity, and 3) exposure of edge sites during activation. First, the higher S_BET and pore volume allowed for better exposure and penetration of the electrolyte ions. Due to TAC-C600 and TAC-C700 having less developed hierarchical porosity compared to that of TAC-C800, the specific capacitance of both samples was considerably lower at 174 F g⁻¹ and 233 F g⁻¹ at 0.5 A g⁻¹, respectively. TAC-C800 has the highest S_BET among all TAC samples, and has both optimized KOH and MgO amounts, resulting in a well-developed pore structure, resulting in a very high capacitance of 269 F g⁻¹ at 0.5 A g⁻¹. The increased number of edge sites from higher activation temperature, indicated by Raman spectroscopy, could also explain the trend in increasing specific capacitance. A study by Randin et al.⁴² reported that edge site capacitance is higher (50–70 µF cm⁻²) compared to the basal plane (16 µF cm⁻²). The difference is apparent between TAC-C600 and TAC-C700, with I_D/I_G increasing from 0.997 to 1.184, respectively. Compared to that of TAC-C600, TAC-C700 also had more rectangular voltammogram shape, indicating stable charge-discharge behavior desired in supercapacitor materials. The difference in the I_D/I_G between TAC-C700 and TAC-C800 is smaller, and both samples exhibited good charge-discharge behavior, thus the difference in specific capacitance between TAC-C700 and TAC-C800 could be better explained by the increase in surface area and pore volume instead (Table 4). Both samples exhibited stable CV curves from 50–500 mV s⁻¹ (Figs. 6b and 6c). Despite the different charge-discharge behavior observed in CV for TAC-C600, all samples had stable GCD curves (Fig. 6d, Fig. S3) at 0.5–5 A g⁻¹. TAC-C600 however had the least stable charge-discharge behavior. All TAC had no ion impurities such as residual K⁺ or Mg²⁺ ions that could influence the measured specific capacitance, as confirmed from the lack of peaks on both the XPS wide and narrow spectra (Fig. S4).

Figure 6.

CV curves of a) TAC-C600, b) TAC-C700, and c) TAC-C800 at scan rates from 50–500 mV s⁻¹. d) specific capacitance of TAC-C600, TAC-C700, and TAC-C800 at various current densities. e) GCD plots of TAC-C800 at various current densities. f) cycle stability plot of TAC-C800 measured at 10 A g⁻¹.

Measuring the specific capacitance of all samples at increasing current density (Fig. 6e) showed that TAC-C600 decreased to 117 F g⁻¹ at 5 A g⁻¹, which corresponded to a capacitance retention of 67.2 %. TAC-C700 had a capacitance drop to 144 F g⁻¹ and a capacitance retention of 61.8 %. TAC-C800 on the other hand had a much higher capacitance retention of 80.3 %, dropping only to 216 F g⁻¹ at 5 A g⁻¹. TAC-C800 cycle stability was further evaluated by repeating 2000 charge-discharge cycles at 10 A g⁻¹ (Fig. 6f), and showed that TAC-C800 still had a cycle stability of 95.92 %. The first and 2000th cycles were also nearly identical, indicating no decrease in the charge-discharge stability.

To confirm the synergistic effect of pore formation by MgO and KOH in TAC, the electrochemical properties of TAC-K5x and TAC-KxM30 were also evaluated to compare with TAC-C800 (Fig. S5). Due to not having a well-developed micropore structure, TAC-KxM30 did not exhibit high specific capacitance as evidenced by the small area of the CV curves. There was also uneven charge-discharge behavior, which is not optimal for the operational electrical output demanded from supercapacitor electrode materials. The same observation could be said about the GCD curves (Fig. S5b), and the calculated C_sp of TAC-KxM30 is only 92.8 F g⁻¹ at 0.5 A g⁻¹. Meanwhile, TAC-K5Mx had much more stable charge-discharge behavior across all scan rates (Fig. S5c), but due to the lack of the mesopores both S_BET and electrolyte penetration are lower compared to TAC-K5M30. This reflected in the GCD curves, and the calculated C_sp at 0.5 A g⁻¹ was 185.4 F g⁻¹.

Because the heat treatment performed was similar to a one-step carbonization-activation method, TAC-C800 was compared with carbon materials from other literature that reported performing one-step carbonization-activation (Table 5). A previous study done by Jiang et al.⁴³ utilized nicotinic acid as the carbon precursor to induce N-doping onto the AC, and also used Mg(OH)₂ NPs as the hard template instead of MgO. They reported a slightly higher specific capacitance of 282 F g⁻¹ at 1 A g⁻¹, but the carbon precursor is expensive and not environmentally-friendly. Furthermore, compared to MgO, Mg(OH)₂ had poorer size uniformity, which resulted in less uniform pore structure. Cellulose acetate nanofibers were used by Wang et al.⁸ as the carbon source, but without a mesopore-inducing agent, the carbon material had lower surface area, less developed hierarchical pore structure, and lower specific capacitance. A similar result was reported when the carbon source was switched to a phenol-formaldehyde resin,⁴⁴ thus showing that selectively inducing mesoporosity through templating in AC production as a whole is a good way to improve AC electrochemical specific capacitance. A study by Lin et al.⁴⁵ on using nanosilica (SBA-15) as template and ZnCl₂ as the activating agent for AC reported a maximum specific capacitance similar to this research. It has been reported that ZnCl₂ is a better activation agent, but environmental problems⁵ related to ZnCl₂ waste disposal remains to be a challenge, making KOH one of the more practical and eco-friendly chemical activating agent with similar AC performance. Other templates and activating agents have also been explored,⁴⁶ but these methods are less environmentally friendly.⁵

Table 5. Comparison of observed specific capacitance for TAC-C800 with carbon materials in other literature.

Carbon precursor	Activation method	Electrolyte	Specific capacitance /F g−1	Reference
TOCN with 50 nm MgO NP template	KOH, 800 °C	1 mol L−1 H2SO4	269 F g−1 at 0.5 A g−1	This work
Studies with one-step carbonization-activation method
Nicotinic acid with nano Mg(OH)2	KOH, 800 °C	6 mol L−1 KOH	282 F g−1 at 1 A g−1	Jiang et al.43
Cellulose acetate nanofibers	NaOH, 800 °C	6 mol L−1 KOH	229.4 F g−1 at 0.2 A g−1	Wang et al.8
Phenol-formaldehyde resin	KOH, 750 °C	6 mol L−1 KOH	234 F g−1 at 0.1 A g−1	Zheng et al.44
Poly(furfuryl alcohol) with SBA-15 template	ZnCl2, 800 °C	6 mol L−1 KOH	270 F g−1 at 1 A g−1	Lin et al.45
Cornstalk-based cellulose with LiCl and ZnCl2 template	K2C2O4 and CaCO3, 800 °C	1 mol L−1 H2SO4	375 F g−1 at 0.5 A g−1	Li et al.46
Cellulose-based carbon
Hydrothermally treated cellulose and sludge	KOH, 750 °C	6 mol L−1 KOH	286.68 F g−1 at 1 A g−1	Xu et al.47
N, P-doped pomelo peel	None, 900 °C	2 mol L−1 KOH	240 F g−1 at 0.5 A g−1	Wang et al.10

Comparing TAC-C800 with other cellulose-based carbon shows potential avenues for further development of TAC. A study by Xu et al.⁴⁷ using N-doped rice husk-derived cellulose and sludge, as well as N, P-doped pomelo peel-based cellulose by Wang et al.¹⁰ as an environmentally friendly carbon material shows the benefit of employing both hierarchical pore structure design as well as heteroatom doping to further improve the performance of green AC.

4. Conclusions

This study presented a facile and eco-friendly method of utilizing TEMPO-oxidized cellulose nanofibers to stabilize MgO NPs as the carbon precursor for KOH-activated AC via an integrated MgO drying, carbonization, and activation heat treatment method. It was shown that KOH and MgO contribute to the formation of a hierarchical porous structure through different mechanisms that do not affect one another. Activation temperature affected the overall pore formation process. Due having excellent surface area, pore volume, and graphene basal plane edge site exposure, TAC-C800 exhibited high specific capacitance of 269 F g⁻¹ at 0.5 A g⁻¹ with good capacitance retention of 80.3 % when current density was increased to 5 A g⁻¹. Cycle stability of 95.92 % was observed even after undergoing 2000 charge-discharge cycles at 10 A g⁻¹, therefore making TAC-C800 a viable material for supercapacitor applications.

Acknowledgments

The researchers would like to thank Taro Uematsu, Susumu Kuwabata, Yo Shimura, Shinji Tamura, Nobuhito Imanaka, Shohei Maruyama, Tomoko Fukuhara, Akihide Sugawara, and Judit Rebeka Molnar for providing assistance in conducting this research. This research was financially supported by the JSPS KAKENHI Research Fund (20H02797), the JST-Mirai Program (JPMJMI8E3), and the Nippon Glass Sheet Foundation.

Data Availability Statement

The data that support the findings of this study are openly available under the terms of the designated Creative Commons License in J-STAGE Data at https://doi.org/10.50892/data.electrochemistry.20024309.

CRediT Authorship Contribution Statement

Mark Adam Ferry: Conceptualization (Lead), Data curation (Lead), Formal analysis (Lead), Investigation (Lead), Methodology (Lead), Project administration (Lead), Validation (Lead), Visualization (Lead), Writing – original draft (Lead), Writing – review & editing (Supporting)

Jun Maruyama: Conceptualization (Supporting), Formal analysis (Supporting), Funding acquisition (Supporting), Methodology (Supporting), Supervision (Lead), Writing – review & editing (Lead)

Taka-Aki Asoh: Methodology (Supporting), Supervision (Supporting)

Hiroshi Uyama: Funding acquisition (Lead), Supervision (Supporting), Writing – review & editing (Supporting)

Declaration of Conflicts of Interest

The authors declare that there are no competing interests that influenced the work reported in this paper.

Funding

Japan Society for the Promotion of Science London: 20H02797

JST-Mirai Program: JPMJMI8E3

Nippon Sheet Glass Foundation for Materials Science and Engineering

Footnotes

J. Maruyama and H. Uyama: ECSJ Active Members

References

• 1)  P. González-García, Renewable Sustainable Energy Rev., 82, 1393 (2018).
• 2)  B. Chang, B. Yang, Y. Guo, Y. Wang, and X. Dong, RSC Adv., 5, 2088 (2015).
• 3)  B. Ding, S. Huang, K. Pang, Y. Duan, and J. Zhang, ACS Sustain. Chem. Eng., 6, 177 (2018).
• 4)  A. Dobashi, J. Maruyama, Y. Shen, M. Nandi, and H. Uyama, Carbohydr. Polym., 200, 381 (2018).
• 5)  Z. Heidarinejad, M. H. Dehghani, M. Heidari, G. Javedan, I. Ali, and M. Sillanpää, Environ. Chem. Lett., 18, 393 (2020).
• 6)  G. Lin, R. Ma, Y. Zhou, Q. Liu, X. Dong, and J. Wang, Electrochim. Acta, 261, 49 (2018).
• 7)  Y. Jiang, Z. He, X. Cui, Z. Liu, J. Wan, Y. Liu, and F. Ma, New J. Chem., 46, 6078 (2022).
• 8)  Y. Wang, J. Cui, Q. Qu, W. Ma, F. Li, W. Du, K. Liu, Q. Zhang, S. He, and C. Huang, Microporous Mesoporous Mater., 329, 111545 (2022).
• 9)  O. Oginni, K. Singh, G. Oporto, B. Dawson-Andoh, L. McDonald, and E. Sabolsky, Bioresour. Technol. Rep., 7, 100266 (2019).
• 10)  Z. Wang, Y. Tan, Y. Yang, X. Zhao, Y. Liu, L. Niu, B. Tichnell, L. Kong, L. Kang, Z. Liu, and F. Ran, J. Power Sources, 378, 499 (2018).
• 11)  C. Zhang, W. Song, Q. Ma, L. Xie, X. Zhang, and H. Guo, Energy Fuels, 30, 4181 (2016).
• 12)  C. Zequine, C. K. Ranaweera, Z. Wang, S. Singh, P. Tripathi, O. N. Srivastava, B. K. Gupta, K. Ramasamy, P. K. Kahol, P. R. Dvornic, and R. K. Gupta, Sci. Rep., 6, 31704 (2016).
• 13)  K. Okada, M. Nandi, J. Maruyama, T. Oka, T. Tsujimoto, K. Kondoh, and H. Uyama, Chem. Commun., 47, 7422 (2011).
• 14)  J. Zhou, Z. Sun, M. Chen, J. Wang, W. Qiao, D. Long, and L. Ling, Adv. Funct. Mater., 26, 5368 (2016).
• 15)  S. Kubo, R. J. White, K. Tauer, and M.-M. Titirici, Chem. Mater., 25, 4781 (2013).
• 16)  M. Peer, M. Lusardi, and K. F. Jensen, Chem. Mater., 29, 1496 (2017).
• 17)  M. A. Ferry, J. Maruyama, T.-A. Asoh, and H. Uyama, RSC Adv., 11, 2202 (2021).
• 18)  H. Li, Y. Zhao, S. Liu, P. Li, D. Yuan, and C. He, Microporous Mesoporous Mater., 297, 109960 (2020).
• 19)  H. Noh, S. Choi, H. G. Kim, M. Choi, and H.-T. Kim, ChemElectroChem, 6, 558 (2019).
• 20)  H. Wang, S. Yu, and B. Xu, Chem. Commun., 52, 11512 (2016).
• 21)  E. Unur, S. Brutti, S. Panero, and B. Scrosati, Microporous Mesoporous Mater., 174, 25 (2013).
• 22)  T. Morishita, Y. Soneda, T. Tsumura, and M. Inagaki, Carbon, 44, 2360 (2006).
• 23)  S. Li, X. Song, X. Wang, C. Xu, Y. Cao, Z. Xiao, C. Qi, M. Wu, Z. Yang, L. Fu, X. Ma, and J. Gao, Carbon, 160, 176 (2020).
• 24)  A. J. Fletcher and K. M. Thomas, Langmuir, 15, 6908 (1999).
• 25)  K. Kaare, E. Yu, A. Volperts, G. Dobele, A. Zhurinsh, A. Dyck, G. Niaura, L. Tamasauskaite-Tamasiunaite, E. Norkus, M. Andrulevičius, M. Danilson, and I. Kruusenberg, ACS Omega, 5, 23578 (2020).
• 26)  L. Gao, L. Xiong, D. Xu, J. Cai, L. Huang, J. Zhou, and L. Zhang, ACS Appl. Mater. Interfaces, 10, 28918 (2018).
• 27)  M. J. Ahmed, B. H. Hameed, and E. H. Hummadi, Carbohydr. Polym., 247, 116690 (2020).
• 28)  C. Zhu and T. Akiyama, Green Chem., 18, 2106 (2016).
• 29)  X. Du, S. Ma, W. Zhao, Z. Zheng, T. Qi, and Y. Wang, J. Nanosci. Nanotechnol., 17, 3835 (2017).
• 30)  W. Zhao, M. Zhang, P. Pan, D. Song, S. Huang, J. Wei, X. Li, W. Qi, K. Zhang, J. Zhao, and Z. Yang, Surf. Coat. Tech., 320, 595 (2017).
• 31)  S. Li, Q. Chen, Y. Gong, H. Wang, D. Li, and Y. Zhang, ACS Omega, 5, 29913 (2020).
• 32)  A. Isogai, T. Saito, and H. Fukuzumi, Nanoscale, 3, 71 (2011).
• 33)  Z. Tang, W. Li, X. Lin, H. Xiao, Q. Miao, L. Huang, L. Chen, and H. Wu, Polymers (Basel), 9, 421 (2017).
• 34)  P. Köll, G. Borchers, and J. O. Metzger, J. Anal. Appl. Pyrolysis, 19, 119 (1991).
• 35)  S. D. Stefanidis, K. G. Kalogiannis, E. F. Iliopoulou, C. M. Michailof, P. A. Pilavachi, and A. A. Lappas, J. Anal. Appl. Pyrolysis, 105, 143 (2014).
• 36)  L. Yan, J. Zhuang, X. Sun, Z. Deng, and Y. Li, Mater. Chem. Phys., 76, 119 (2002).
• 37)  J. Wang and S. Kaskel, J. Mater. Chem., 22, 23710 (2012).
• 38)  L. Gao, H. Lu, H. Lin, X. Sun, J. Xu, D. Liu, and Y. Li, Chem. Res. Chin. Univ., 30, 441 (2014).
• 39)  Y. Wu, J.-P. Cao, X.-Y. Zhao, Z.-Q. Hao, Q.-Q. Zhuang, J.-S. Zhu, X.-Y. Wang, and X.-Y. Wei, Electrochim. Acta, 252, 397 (2017).
• 40)  M. Li, C. Liu, H. Cao, H. Zhao, Y. Zhang, and Z. Fan, J. Mater. Chem. A, 2, 14844 (2014).
• 41)  J. Maruyama, S. Maruyama, T. Fukuhara, and K. Hanafusa, J. Phys. Chem. C, 121, 24425 (2017).
• 42)  J.-P. Randin and E. Yeager, J. Electroanal. Chem. Interfacial Electrochem., 36, 257 (1972).
• 43)  Y. Jiang, Y. Wang, J. Cui, J. Liu, Y. Zhang, and Y. Wu, J. Solid State Electrochem., 23, 2355 (2019).
• 44)  Z. Zheng and Q. Gao, J. Power Sources, 196, 1615 (2011).
• 45)  G. Lin, F. Wang, Y. Wang, H. Xuan, R. Yao, Z. Hong, and X. Dong, J. Electroanal. Chem., 758, 39 (2015).
• 46)  J. Li, L. Wei, Q. Jiang, C. Liu, L. Zhong, and X. Wang, Ind. Crops Prod., 154, 112666 (2020).
• 47)  Z. X. Xu, X. Q. Deng, S. Zhang, Y. F. Shen, Y. Q. Shan, Z. M. Zhang, R. Luque, P.-G. Duan, and X. Hu, Green Chem., 22, 3885 (2020).