Softwood kraft lignin produced from southern pine (BioChoice™) was provided by Domtar Inc. All other reagents were ACS grade and used as is. Total hydroxyl group content of lignin was ∼5 mmol/gr as determined by ³¹P NMR (Liu et al., 2020).

Lignin was dissolved at a 30 wt% concentration in 1 M sodium hydroxide (NaOH) and stirred overnight at room temperature for complete dissolution. Epichlorohydrin at different molar ratios (epichlorohydrin: lignin: 0.6:1, 1:1, 1:2, 1:3, corresponding to 44:56, 32:68, 19:81, 13:87 w/w %/% ratio) was added into lignin solution and stirred for 15 min. The mixtures were transferred into glass vials, sealed, and placed in an oven preheated at 75°C for 24 h. After the crosslinking reaction and gel formation, lignin gels were carefully removed from glass vials, rinsed, and solvent-exchanged by keeping the gels in distilled water and absolute ethanol sequentially for 3 days each, by exchanging with fresh solvent two times a day. Lignin gels in ethanol (alcogels) were supercritical-dried with liquid carbon dioxide using Tousimis Autosamdri 815B Critical Point Dryer. To produce carbon aerogels (CA), lignin aerogels were heat-treated in a tube furnace from room temperature to 900°C with a heating rate of 5°C/min and at 900°C for 1 h under nitrogen gas flow. For production of activated carbon aerogels (A-CA), CA samples were immersed into 6 M potassium hydroxide solution (3:1 KOH: lignin w/w), dried in oven at 105°C, and heated in a tube furnace at 800°C for 1 h under a nitrogen gas flow. After heat processing, A-CA samples were neutralized with 0.1 N hydrochloric acid (HCl), thoroughly washed with distilled water to remove residual impurities, and dried in oven at 105°C.

The morphology of lignin-based carbon aerogels was analyzed by field emission scanning electron microscopy (FE-SEM, Hitachi 4,700) at accelerating voltage of 5 kV. The samples were mounted onto metal holders with carbon tape, and sputter-coated with gold. To evaluate the crystal structure of carbon aerogels, X-ray diffraction patterns were collected in the range of from 2θ = 5° to 2θ = 60° at a CuKα radiation wavelength of λ = 1.5406° from a generator operating at 40 kV and 40 mA. The pore volume, surface area and pore size distributions of lignin-based carbon aerogels were determined from nitrogen (N₂, 77 K) adsorption-desorption and carbon dioxide (CO₂, 273 K) adsorption isotherms using Micromeritics 3Flex physisorption analyzer. Samples were degassed under vacuum at 350°C for 6 h prior to gas sorption experiments. The total pore volume (V _t ) was obtained from the nitrogen adsorption isotherm at p/p₀ ∼0.99 and the specific surface area (S _BET ) was determined using the Brunauer-Emmett-Teller (BET) method. The micropore volume (V _micro ) was calculated using t-plot analysis. The pore size distributions and average pore diameter is determined from the N₂ desorption branch using Barrett, Joyner, and Halenda (BJH) method. In addition, micropore volume and pore size distributions of ultramicropores (pore size < 1 nm) were analyzed from the CO₂ adsorption isotherm (V _micro-CO2 ) using Dubinin- Radushkevich equation and DFT models. The bulk density ( ρ ,   g/cm³) was calculated by dividing the mass of the sample to its volume for all aerogels after supercritical drying and carbonization. Fourier transform infrared (FTIR) spectra of lignin aerogels and carbon aerogels were recorded from 600 to 4,000 cm⁻¹ at a resolution of 4 cm⁻¹ using a Bruker Invenio spectrometer.

The electrochemical experiments were conducted in a two-electrode configuration with stainless steel current collectors, a polyvinylidene fluoride (PVDF) filter paper as the separator, and 1.0 M H₂SO₄ as the aqueous electrolyte at room temperature. Carbon electrodes were prepared from a slurry of finely ground lignin carbon aerogel, carbon black, and the PVDF binder with the mass ratio of 8:1:1 dispersed in N-methyl pyrrolidone (NMP) solution. An equal amount of slurry was coated on metal collectors (area:1 cm²), dried in an oven at 105°C for at least 24 h, and immersed in electrolyte prior to the electrode assembly. Electrochemical tests including cyclic voltammetry (CV), galvanostatic charge-discharge (GCD), and electrochemical impedance (EIS) were carried out using the Gamry Interface 1010 E potentiostat workstation with the operation potential from 0–0.8 V. CV tests were conducted with different scan rate from 5 to 200 mV/s. The GCD tests were investigated at a current density of 0.1, 0.5, and 1 A/g, respectively. EIS tests were conducted in the frequency range of 10^–1–10⁶ Hz to understand the resistance of the electrode.

The specific capacitance (C_s) of the single electrode were calculated from the CV curve by Eq. 1. C S = 2 × ∫ I V d V m × v × Δ V (1) where I, V, m, v, and ∆V are the current, potential, electrode mass, scan rate, and operating voltage window, respectively.

The specific capacitance (C_s) for the single electrode was also calculated from the discharge curve by Eq. 2. C S = 4   × I × Δ t m × Δ V (2) where I, ∆t, m, ∆V are the discharge current, the discharge time, the total mass of the electrode materials in both working electrodes, and the operating voltage range, respectively.

The energy density (E; Wh/kg) and power density (P; W/kg) were calculated from Eqs. 3, 4 E = C S × Δ V 2 8 × 3.6 (3) P = 3600 × E Δ t (4) where C _S, ∆V, ∆t is the specific capacitance from the GCD curve, the operating voltage range, and discharge time, respectively.

In this study, biomass-derived carbon aerogels were produced from a commercially available lignin isolated by the kraft pulp processing of softwood species. Kraft lignin has various functional groups such as aromatic/aliphatic hydroxyl, carboxylic acid and carbonyl groups that are reactive towards chemical derivatization and copolymerization (Glasser, 2019). By taking advantage of lignin’s aromatic nature and chemical functionality, cross-linked network gels were synthesized and converted into aerogel materials by subsequent sol-gel processing, supercritical drying and pyrolysis stages (Figure 1). To synthesize lignin gels, lignin was first solubilized under aqueous alkaline conditions, which makes the aromatic hydroxyl groups charged (phenoxy ion) and active, and reacted with an industrially used epoxy compound, epichlorohydrin. The suggested reaction mechanism between lignin and epichlorohydrin (Figure 1) includes an ring-opening of oxirane ring followed by epoxidation of hydroxyl groups of lignin, and reaction of an epoxidized lignin with another lignin anion, resulting in the crosslinking of lignin macromolecules (Saidane et al., 2010; Nypelö et al., 2015). FT-IR spectra of lignin and aerogels with different lignin contents confirmed the structural changes of lignin after crosslinking reaction (Figure 2). The peaks characteristic to lignin’s hydroxyl (O-H) stretching at 3,400 cm⁻¹ and aromatic skeletal vibrations at 1,595 cm⁻¹ and 1,510 cm⁻¹ shifted in the spectra of all aerogels after the reaction. For the region corresponding to C-H stretching of methyl and methylene groups, the intensity of 2,935 cm⁻¹ peak increased and a new peak appeared at 2,877 cm⁻¹. In addition, the peak at 1,365 cm⁻¹ that belong to bending of phenolic hydroxyl groups of lignin disappeared and intensity of strong peaks at 1,125 cm⁻¹ and 1,085 cm⁻¹ increased, confirming the crosslinking reaction and formation of ether (C-O-C) linkages (Passauer et al., 2012; Nypelö et al., 2015). The peak belonging to carbonyl stretching of unconjugated groups of lignin shifted from ∼1710 cm⁻¹ to ∼1730 cm⁻¹ only for L-56 and L-68 aerogels, which also suggested epoxidation of carboxyl acid groups of lignin. For aerogels with higher crosslinker content (44 and 32%), the peak that belongs to the oxirane ring appeared at 750 cm⁻¹ while it disappeared for the aerogels with reduced amount of crosslinker (19 and 13%). This observation is in agreement with previous studies that the higher crosslinker-to-lignin ratio favoured epoxidation of lignin over crosslinking while reducing the amount of crosslinker resulted in more crosslinked structure (Saidane et al., 2010; Nypelö et al., 2015).

After formation of the crosslinked network structure, lignin gels (hydrogels) were kept in water for several days to remove unreacted components and converted into “alcogels” in ethanol by repetitive solvent exchange steps. This step is critical as ethanol is miscible with liquid carbon dioxide, which reduces the surface tension and interfacial capillary forces within the pores, decreases the collapse of pore structure and excessive pore shrinkage upon removal of solvent during supercritical drying (Grishechko et al., 2013a; Baldino et al., 2020). The bulk density and porosity of aerogels and carbon aerogels prepared with different lignin-to-crosslinker ratios are shown in Table1. The porosity was calculated based on the ratio of bulk density of aerogels and skeletal density of lignin (1.4 g/cc) and amorphous carbon (2.0 g/cc) (Szczurek et al., 2011; Grishechko et al., 2013b). The bulk density of aerogels was in the range of 0.20 g/cm³ to 0.74 g/cm³, increased after carbonization, and ranged from 0.48 g/cm³ to 0.83 g/cm³ for carbon aerogels due to the combination of volumetric shrinkage and mass loss during pyrolysis. Increasing the amount of lignin relative to epichlorohydrin increased the bulk density of aerogels while reducing the porosity, which varied from 47% to 86%-- these data were similar to reported values for biobased aerogels prepared from lignin-phenol-formaldehyde and condensed tannin (Szczurek et al., 2011; Grishechko et al., 2013a, 2013b). These results suggested that the number of crosslinks reduced with the decreasing lignin-to-crosslinker ratio, resulting in weaker internal pore structure and higher shrinkage during drying process, therefore yielding higher bulk density. Note that porous carbon materials with higher bulk density enable higher packing density of active component in electrodes, which is desirable for high performance energy storage applications such as supercapacitors with outstanding volumetric capacitance (Tao et al., 2013; Wang Q. et al., 2016; Jin et al., 2018). In addition, due to high carbon content (60–65%) of lignin, the overall carbon yield of aerogels increased from 35 wt% to 47 wt% as the lignin content increased from 56 wt% to 87 wt%, which is an important criteria for the production of low-cost and high-yield electrodes from renewable carbon.

X-ray analysis of carbon aerogels showed that all samples had similar diffraction patterns characteristic to amorphous disordered carbon (Figure 3). The broad peaks at 2θ = 23° and 2θ = 43° corresponds to the d₀₀₂ and d₁₀₁ interlayer spacings of graphite, respectively (Xu et al., 2018). Also, the weak peak observed at about 2θ = 12° might be due to the presence of residual oxygenated groups (Li J. et al., 2019) due to partial graphitization of lignin at 900°C as revealed by elemental analysis and FTIR results (data not shown). However, it is likely that this weak peak might be related to the porous structure of carbon aerogels and the presence of nanopores which are less than 1 nm (Meynen et al., 2009; Zhang et al., 2017). The d₀₀₂ interlayer spacing of carbon aerogels were slightly decreased from 4.01 Å to 3.88 Å with increasing lignin content, indicating that the CA-L87 carbon aerogel had higher degree of graphitization and improved stacking structure compared to the CA-L56 (Xu et al., 2018).

The surface morphology and porous structure of carbon aerogels prepared with different lignin-to-crosslinker ratios were examined by SEM (Figure 4). Depending on the composition of aerogels, the structure of pores and the shape of carbon particles changed significantly. The samples with lower lignin-to-crosslinker ratios (CA-L87 and CA-L81) showed a three-dimensional porous network structure composed of interconnected particles (Figures 4A–D), which is typical for carbon aerogels with mesopores (2–50 nm) (Szczurek et al., 2011; Li F. et al., 2019). On the other hand, the samples CA-L68 and CA-L56 had no visible mesoporous network structure but composed of either spherical carbon particles with sizes ranging from 1–3 µm or micron-size pores (Figures 4E–G). Interestingly, higher magnification SEM images revealed that CA-L68 and CA-L56 samples had secondary carbon particles with less than 50 nm and additional surface features suggesting the presence of micropores (Figures 4F–H). This difference in the surface morphology might be explained by the changes in the chemical structure of lignin as revealed by FTIR analysis, which showed the presence of epoxidized lignin derivative in aerogels especially when the crosslinker ratio was higher. Prior to the formation of a fully crosslinked network, lignin’s hydroxyl groups are first derivatized with epoxy functionality, which increases the free volume of lignin and therefore reduces its glass transition temperature and increasing its ability to flow at elevated temperatures. Thus, it is conjectured the formation of larger size primary carbon particles and the absence of observable mesoporous network in CA-L68 and CA-L56 samples might be due to the thermal flow and fusion of epoxidized lignin particles during carbonization process. These results suggested that a thermal-stabilization step of aerogels prior to carbonization, especially for the aerogels with higher content of epoxidized lignin derivatives, would prevent thermal fusion of particles and therefore maintain the mesoporous structure (Nypelö et al., 2015).

To better understand the porous structure of lignin carbon aerogels, nitrogen (N₂) adsorption-desorption isotherms were collected and the BET surface area, pore volume, average pore diameter and pore size distributions of all samples were calculated (Figures 5A,C; Table2). CA-L87 and CA-L81 samples had significantly higher BET surface areas (154 and 397 m²/g) and total pore volumes (0.26 and 0.21 g/cm³) compared to those of CA-L68 and CA-L56 samples, which is in accordance with the SEM observations. In addition, the hysteresis loop between adsorption and desorption branches and the pore size distribution curves obtained by the BJH method confirmed the presence of mesopores (2–50 nm), especially with the CA-L87 sample (Figure 5C). Interestingly, CA-L81 samples showed significantly higher micropore volume and surface area, as calculated by t-plot method, whereas CA-L87 had higher external surface area for meso- and macropores. On the contrary of the SEM images at higher magnifications, suggesting microporosity, N₂ adsorption results showed insignificant amount of micropores especially for of CA-L68 and CA-L56 samples. N₂ adsorption is the most used technique for the characterization of porous carbon materials, however, it has drawbacks for characterization of smaller micropores (<1 nm) due to limited diffusion rate of nitrogen at cryogenic temperatures (77K) (Thommes and Cychosz, 2014; Ambroz et al., 2018). Alternatively, CO₂ adsorption at 273K has become a common practice to characterize microporous carbons because of its of higher saturation pressure and ease of access to smaller micropores with sizes down to 0.35 nm (Thommes and Cychosz, 2014; Ko et al., 2021). The micropore volumes were calculated using the Dubinin-Radushkevich equation and the distribution of pores for the ultra-micropore region was obtained by the DFT models (Figure 5E; Table2). The results revealed that all samples had significant amount of pore volume in the range of 0.27–0.29 g/cm³ for the pore sizes of about 0.5 and 0.8 nm.

To further improve the pore volume and surface area of carbon aerogels, KOH chemical activation process was used, which is a frequently used method for porous carbon materials in energy applications (Romanos et al., 2011; Wang and Kaskel, 2012). KOH reacts with carbon through redox reactions at elevated temperatures above 400°C, resulting in etching of carbon framework and development of porosity through formation of H₂O and CO₂. In addition, metallic potassium compounds formed during activation expands the carbon lattices, creating higher microporosity and surface area upon removal of metallic K by washing after activation process (Romanos et al., 2011; Wang and Kaskel, 2012). N₂ adsorption isotherms and pore size distribution of activated samples are shown in Figure 5B and Figure 5D and the structural properties of activated samples (A-CA) calculated from the N₂ and CO₂ adsorption isotherms were summarized in Table2. After surface activation, the BET surface area, total pore volume, and micropore volumes of all carbon aerogel samples improved considerably with values reaching up to 1,609 m²/g and 0.98 cm³/g with 0.68 cm³/g, respectively. All samples showed substantial amount of microporosity with average pore diameters in the range of 2.7–3.4 nm. Moreover, the micropore volumes for the ultra-micropore region increased for all the samples and ranged from 0.36 to 0.44 cm³/g after activation (Figure 5F).

Electrochemical performance of lignin-based carbon aerogels was evaluated through CV and GCD tests in the symmetric two-electrode system. The properties of supercapacitor electrodes fabricated from CA-L87 sample were compared with electrodes assembled from a commercial carbon aerogel (CCA) and activated carbon (A-40), which had BET surface areas of 845 and 1,071 m²/g, respectively (Table2). Based on the XRD and N₂ adsorption results, CA-L87 carbon aerogel was selected for comparison due its higher bulk density, degree of graphitization, total pore volume, and mesoporosity as compared to other samples. The CV curves of the CA-L87, CCA, and A-40 at the 20 mV/s scan rate are shown in Figure 6A. The A-40 had a better quasi-rectangular shape, while CA-L87 exhibited a larger area under the CV curve indicating higher specific capacitance (C_s), which were 84.0, 66.6, and 58.7 F/g for CA-L87, CCA, and A-40, respectively. Figure 6B shows the CV curves of the CA-L87 sample at different scan rates. The test demonstrated that the shape of the CV curves gradually shifted and tilted with the increase of the scan rate, while the CA-L87 could maintain the quasi-rectangular shape and electrochemical stability even at a higher scan rate of up to 200 mV/s. The C_s of the CA-L87 at 5, 10, 20, 50, 100, and 200 mV/s scan rates were 232.3, 152.7, 88.7, 58.7, 45.5 and 33.4 F/g, respectively. The shape of the charge-discharge curves showed a blunt and slanted quasi-rectangular shape, and the slope of the curves became more pronounced as the scan rate increased. This phenomenon is because of the equivalent series resistance of the electrodes which contain nonconductive binder and other resistive elements within the supercapacitor cells (Mathis et al., 2019). At higher scan rates, the electrolyte ions have a shorter ion migration time to penetrate the pores of the samples, and they are mainly accumulated only on the outer layer surface of electrode (Norouzi et al., 2021). Therefore, the specific capacitance of the CA-L87 decreased with the increase of scan rate. Although the C_s of CA-L87 could reach above 200 F/g at 5 mV/s, it had a 14.4% capacitive retention rate at a 200 mV/s scan rate. In contrast, the CCA and A-40 had lower C_s of 73.9 and 62.5 F/g at 5 mV/s, while they had higher capacitive retention rates of 79.6 and 78.9% at 200 mV/s scan rate, indicating better fast charge-discharge capability.

The GCD curves of CA-L87, CCA, and A-40 at 0.5 A/g current density showed a desired electric double-layer capacitor (EDLC) behavior of linear triangular shapes (Figure 6C). The GCD curve of CA-L87 at the different current densities demonstrated that the CA-L87 sample could maintain the symmetric linear triangular shape at 1 A/g current density with a faster charge-discharge rate (Figure 6D). The C_s for the CA-L87 sample obtained from the slope of the discharge curves were 263.3, 165.3, and 122.3 F/g at the current density of 0.1, 0.5, and 1  A/g, respectively. The C_s of CA-L87 had a 46.4% capacitive retention when the current density increased from 0.1 to 1 A/g.

Impedance (EIS) tests were also conducted to understand the electrochemical kinetics of the supercapacitor cells. Figure 6E shows the Nyquist plots of CA-L87, CCA, and A-40, all of which exhibited typical supercapacitor behavior with a semi-circle at the high-frequency range and a vertical line at the low-frequency range (Liu et al., 2015; Song et al., 2015). The x-intercept of the Nyquist curve represents the equivalent series resistance (R_es) of the supercapacitor cell, and the diameter of the semi-circle means the charge-transfer resistance (R_ct) from electrolytes moving through the electrodes (Yoo et al., 2014). CA-L87 had the lowest R_es (∼0.2 Ω) compared to A-40, and CCA samples which also showed low R_es values of < 1 Ω. The CA-L87 and A-40 had smaller R_ct than CCA, demonstrated higher ion migration rates at the interface between electrolyte and electrode (Wang H. et al., 2016). All the samples had vertical tails at the low-frequency range, which showed good ion accessibility for the electrolyte to penetrate the internal pores. Overall, these observations suggested a lower resistivity of CA-L87 sample, which can be correlated with the N₂ and CO₂ gas adsorption and XRD results showing the existence of both meso- and micropores and higher degree of graphitization. A long cycling life with good stability is critical for the supercapacitor electrode materials (Cao et al., 2021). The capacitive retention test of the CA-L87 sample after 5,000 charge-discharge cycles demonstrated that CA-L87 could still maintain ∼90% of its capacitance (Figure 6F). The GCD curves had similar behavior after 5,000 cycles, indicating outstanding cycle stability of the CA-L87 and comparable capacitive retention rates with other porous carbon-materials (Chen et al., 2017; Geng et al., 2020; Zhang et al., 2022).

After activation with KOH, the electrochemical performance of the carbon aerogels improved significantly. Figure 7A shows that all A-CA samples had symmetric and quasi-rectangular CV curves at a 20 mV/s scan rate, indicating a superior EDLC behavior compared to non-activated CA samples. The C_s of A-CA-L56, A-CA-L68, A-CA-L81 and A-CA-L87 samples obtained from the CV curves at 20 mV/s scan rate were 94.9, 78.2, 108.9, and 130.0 F/g, respectively. For comparison, the C_s of CA-L87 sample improved about 46% after activation at the same CV scan rate (20 mV/s).

The specific capacitance of carbonous materials and the ion transfer and charge storage ability of EDLCs are related not only to the surface area but also to the porous size distribution. In particular, having hierarchical porous structures with a combination of ultra-, micro-, meso-, and macropores is critical to the superior electrochemical performance of carbonous materials (Lin et al., 2019; Thomas et al., 2021). As evidenced by N₂ and CO₂ gas adsorption results, both surface area and total pore volume, especially for the ultramicropores and micropores, of carbon aerogels increased significantly after activation process. As a result, it is suggested that A-CA samples could maintain higher specific capacitance values even at higher charge-discharge and scan rates compared to non-activated samples.

Figure 7B shows the CV curves of A-CA-L87 at different scan rates. The CV curves could maintain the symmetric quasi-rectangular shape at higher scan rates, and the C_s values calculated for 5, 10, 20, 50, 100, and 200 mV/s were 147.5, 138.6, 130.0, 110.4, 89.4, and 64.7 F/g, respectively. A-CA-L87 sample had a capacity retention rate of 43.9% as the scan rate increased from 5 mV/s to 200 mV/s, which was about three-fold improved performance compared the non-activated CA-L87 sample. As presented in Figure 7C, the GCD curves at 0.5 A/g current density indicated a characteristic EDLC linear triangular shape for all A-CA samples. The C_s values calculated with 0.5 A/g discharging current density were 126.4, 96.5, 128.4, and 144.7 F/g for A-CA-L56, A-CA-L68, A-CA-L81 and A-CA-L87 samples, respectively. Figure 7D shows the GCD curves of A-CA-L87 at different current densities, at which the C_s values were 151.3 F/g at 0.1 A/g, 144.7 F/g at 0.5 A/g, and 134.2 F/g at 1  A/g. The Cs of A-CA-L87 had a ∼88.7% capacitive retention when the current density increased from 0.1 to 1 A/g, which showed a two-fold increase in the performance compared to the non-activated CA-L87 carbon aerogel.

Nyquist plots with a semi-circle at the high-frequency range and a vertical line at the low-frequency range revealed that all A-CA samples had promising EDLC characteristics (Figure 7E). The R_es of the A-CA samples were in the range of 0.2–0.4 Ω, and the R_ct of the testing samples were in the range of 1.5–2.5 Ω. Interestingly, A-CA-L87 and A-CA-L81 samples, which are activated carbon aerogels with higher lignin contents, had smaller semi-circles with a more vertical tail line, indicating lower R_es and R_ct with better ion accessibility. The Ragone plot of A-CA-L87 and A-CA-L81 activated carbon aerogels in comparison to a commercial activated carbon (A40) and a RF-based carbon aerogel (CCA) is shown in Figure 7F. A-CA-L87 and A-CA-L81 had an outstanding energy and power densities, 3.2 Wh/kg at 209.1 W/kg and 2.9 Wh/kg at 210.8 W/kg, respectively, which were better than CCA (1.5 Wh/kg at 209.6 W/kg) and A-40 (1.7 Wh/kg at 197.3 W/kg) at a similar power density range. These results demonstrated that ACA-L87 and ACA-L81 activated carbon aerogels are promising candidates as supercapacitor electrodes.