Cellulose fibers (bleached kraft softwood pulp, CFs) were provided by Zhejiang Huafeng Paper Technology Co., Ltd. Commercial cellulase (CTec3, Novozymes, Sweden) with a nominal activity of 700 EGU/g was used in this study. Aniline (ANI) (≥ 99.5%) was purchased from McLean Biochemical Technology Co., Ltd. (Shanghai, China) and distilled at -0.1 MPa with a rotary evaporator for later use. Ammonium persulfate (APS, ≥ 98%), sulfosalicylic acid (SSA, ≥ 95%), citric acid (CA, ≥ 99.5%) and sodium citrate (≥ 99.0%) were purchased from Aladdin (Shanghai) Co., Ltd. All aqueous solutions were prepared with distilled water.

Prior to enzymatic treatment, the bleached kraft softwood pulp was refined using a Paper and Fibre Research Institute (PFI) mill. Briefly, 30.0 g of oven-dry pulp was adjusted to a consistency of 10 wt% and refined for 2,500 revolutions at a beating pressure of 3.33 N/mm at room temperature (25 °C). Commercial CTec3 cellulase with a nominal activity of 700 EGU/g was diluted with deionized water to obtain a working cellulase solution with an activity of 1.75 EGU/g. Subsequently, 0.5 g (oven-dry basis) of the PFI-beaten CFs was dispersed in 50 mL of citrate-sodium citrate buffer (pH 5.0) pre-incubated at 50 °C, followed by the addition of 0.5 g of the diluted cellulase working solution. The total cellulase activity was therefore 0.875 EGU, corresponding to a cellulase dosage of 1.75 EGU/g oven-dry cellulose. Bio-etching was conducted for 0, 5, 10, and 20 min, and the resulting samples were denoted as Samples 1-4, respectively. The enzymatic reaction was terminated using anhydrous ethanol. The samples were then filtered and thoroughly washed with 2.0 L of deionized water until the filtrate became clear and colorless. The recovered cellulose fibers were dried to constant mass, and the cellulose yield and mass loss were determined gravimetrically based on the oven-dry masses before and after enzymatic treatment, as described in Supplementary Table 1. Three independently prepared samples were analyzed for each condition, and the results are reported as mean values.

A 0.5 g portion of wet CFs [or modified cellulose fibers (m-CFs)] with a moisture content of 83.64% and an oven-dry cellulose mass of 0.0818 g was added to a three-necked flask, followed by the addition of 50 mL of 0.3 M sulfosalicylic acid and 0.5 mL of aniline; subsequently, 50 mL of 0.1 M ammonium persulfate was added dropwise. In-situ polymerization was performed in an ice water bath for 3 h. The mixture was transferred into a mesh bag and rinsed with deionized water for 5 min to remove soluble salts, unreacted species, and loosely bound oligomers. The resulting precipitate was collected via vacuum filtration, then processed into paper-based electrodes by extrusion drying with a flat hot press and dried to constant mass before gravimetric analysis. Composites prepared with and without bio-etching were denoted as PANI/m-CFs and PANI/CFs respectively, and detailed material parameters are provided in Supplementary Table 2.

Electrochemical performance was characterized using an AutoLab PGSTAT 204 electrochemical workstation (Wantong, Switzerland). The paper-based electrode was cut into a 1 cm² square working electrode, fixed with a polytetrafluoroethylene fixture, and immersed in 1 M H₂SO₄ electrolyte. A platinum plate and an Ag/AgCl electrode were used as the counter and reference electrodes, respectively. Cyclic voltammetry (CV) and galvanostatic charge-discharge (GCD) measurements were conducted from -0.1 to 0.9 V versus Ag/AgCl. Electrochemical impedance spectroscopy (EIS) measurements were performed at open-circuit potential over 10⁵ to 0.1 Hz with a 5 mV sinusoidal perturbation. Impedance spectra were fitted in ZView using a circuit containing series resistance, charge-transfer resistance, a constant phase element, and a Warburg diffusion element. The fitted parameters and fitting quality are provided in the Supplementary Materials.

(3) $$ C_{\mathrm{a}}=\frac{I \times \Delta_{t}}{\left(\Delta V-I R_{\text {drop }}\right) \times A} $$

where C_a is the areal capacitance (mF/cm²), I is the discharge current (mA), Δt is the discharge time (s), ΔV is the working potential window (V), IR_drop is the IR voltage drop, and A is the area of the working electrode (cm²).

(4) $$ C_{\mathrm{m}}=\frac{I \times \Delta_{t}}{\left(\Delta V-I R_{\text {drop }}\right) \times M} $$

where C_m is the mass-specific capacitance (F/g), I is the discharge current (mA), Δt is the discharge time (s), ΔV is the working potential window (V), IR_drop is the IR voltage drop, and M is the dry PANI:SSA active-phase mass in the working electrode (mg), as specified in Supplementary Table 2. With I expressed in mA and M in mg, Equation (4) yields C_m in mF/mg, which is numerically equivalent to F/g.

(5) $$ E=\frac{C_{\mathrm{a}} \times\left(\Delta V-I R_{\text {drop }}\right)^{2}}{2 \times 3.6} $$

(6) $$ \quad P=\frac{E \times 3600}{\Delta t} $$

The areal energy density E (μWh/cm²) and power density P (μW/cm²) of the symmetric two-electrode supercapacitor were calculated using Equations (5) and (6), respectively. The device capacitance was normalized to the area of one electrode (1.00 cm²). The effective discharge voltage excludes the initial IR_drop, and Δt is the discharge time over the same voltage interval.

The relationship between the peak current (i) and scan rate (v) was fitted to obtain the b value and assess the relative importance of diffusion-controlled and surface-controlled contributions at the selected redox peaks.

(7) $$ \quad i=\mathrm{a} v^{b} $$

(8) $$ \log (i)=\log ({ a })+b \log (v) $$

(9) $$ \quad i=k_{1} v+k_{2} v^{1 / 2} $$

(10) $$ \quad i / v^{1 / 2}=k_{1} v^{1 / 2}+k_{2} $$

where i is the CV current (A), v is the sweep rate (V/s), a is a constant, and k₁v and k₂v^1/2 correspond to the surface-controlled and diffusion-controlled current contributions, respectively.

The bio-etching process of cellulose fibers is schematically illustrated in Figure 1A. Controlled bio-etching modifies the fiber interface in two complementary ways. First, it increases surface roughness and accessibility while maintaining the continuous fiber framework, potentially providing a larger interfacial area and additional physical anchoring and nucleation sites for subsequent PANI growth. Second, the shift toward a more negative zeta potential indicates a change in the surface charge state that is favorable for interaction with protonated aniline species. Figure 1B schematically illustrates the possible hydrogen-bonding and electrostatic interactions at the cellulose-PANI interface. This interpretation is supported by the surface-morphology, surface-charge, active-phase-loading, and elemental-distribution results discussed below.

The SEM images in Figure 2 show a clear dependence of fiber morphology on bio-etching time. The pristine CFs exhibit a smooth surface [Figure 2A and B]. After 5 min of bio-etching, shallow grooves and fine corrugations appear [Figure 2C and D], indicating that enzymatic attack begins at the outer fiber wall. At 10 min of bio-etching, a well-developed wrinkled microstructure forms across the fiber surface, while the continuous cellulose framework remains preserved [Figure 2E and F]. This balance between surface roughening and structural integrity may provide more accessible anchoring sites for subsequent PANI nucleation and growth. In contrast, extending the bio-etching time to 20 min results in pronounced pore formation and localized collapse of the fiber wall [Figure 2G and H], indicating excessive hydrolysis of the cellulose framework. Importantly, the cellulose yield remained above 98% after 20 min of enzymatic treatment, indicating that the bio-etching process mainly induced surface modification rather than significant bulk cellulose degradation. The resulting loss of surface continuity and structural support may reduce the effective area available for stable PANI anchoring and hinder the formation of a continuous conductive coating, providing a structural explanation for the performance deterioration associated with over-etching. Higher-magnification images further confirm the surface roughening and groove formation induced by bio-etching [Supplementary Figure 1].

The difference caused by bio-etching becomes more evident after aniline polymerization. On pristine CFs, PANI appears as discontinuous aggregates, indicating limited deposition on the pristine fiber surface [Figure 3A and B]. In contrast, PANI/m-CFs exhibit a dense and uniform coating along the fiber surface [Figure 3C and D]. Energy-dispersive X-ray spectroscopy (EDS) maps of PANI/CFs [Figure 3E-H] and PANI/m-CFs [Figure 3I-L] show the spatial distributions of C, N, O, and S. The more continuous N and S distributions in PANI/m-CFs are consistent with a more uniform coating. These observations further establish the relationship between interfacial modification and PANI deposition. Compared with the relatively smooth pristine CFs, the roughened and more accessible bio-etched fiber surface may provide more available sites for PANI nucleation and growth^[18,19]. Accordingly, PANI/m-CFs exhibit a denser and more continuous coating, consistent with similar fibrous cellulose/PANI architectures reported previously^[20]. The more homogeneous N and S distributions in the EDS maps are also consistent with increased PANI incorporation.

With cellulase dosage and temperature held constant, the effect of bio-etching duration on the properties of CFs was systematically evaluated. As depicted in Figure 4A and Supplementary Table 2, the PANI:SSA active-phase areal loading increases from 1.08 mg/cm² for untreated fibers to 1.77 and 2.93 mg/cm² after 5 and 10 min, respectively, before decreasing to 2.04 mg/cm² after 20 min. This optimal duration also yielded the lowest resistivity of 1.13 Ω·m and a maximum conductivity of 0.88 S/m [Figure 4B]. Prolonging the bio-etching time to 20 min causes a decline in PANI-SSA active-phase loading, consistent with the structural damage observed after excessive bio-etching. Specifically, over-etching triggers localized collapse of the fiber microstructural framework, which may reduce the number of effective anchoring sites [Figure 2G and H]. Moreover, structural damage may impair the mechanical support provided by CFs, impeding the construction of a stable core-shell conductive network.

The surface charge characteristics, as evaluated by zeta potential measurements, exhibit a pronounced dependence on bio-etching duration. As shown in Figure 4C and D, bio-etching shifts the zeta potential toward more negative values, reaching -29.1 mV after 10 min. The 20 min sample remains strongly negative; therefore, the loading maximum at 10 min cannot be explained by surface charge alone. The corresponding zeta-potential distribution curves of the other samples are provided in Supplementary Figure 2. The enhanced negative surface charge may be associated with bio-etching-induced microfibrillation, which increases the exposed surface area and promotes the exposure of negatively charged functional groups on the fiber surface. Greater surface accessibility may facilitate subsequent PANI:SSA deposition and contribute to the formation of a denser coating [Figure 1B]. To further quantify the variation in accessible carboxyl group content, conductometric titration was performed [Supplementary Table 3]. The titratable carboxyl group content increased from 0.0446 mmol/g for pristine CFs to 0.0563 mmol/g after 10 min of bio-etching, indicating increased accessibility of titratable carboxyl groups.

AFM images provide quantitative evidence for the surface changes induced by bio-etching^[21]. The three-dimensional topographies [Figure 5A and B] show that bio-etching produces a more uneven surface with greater height variations. The corresponding height maps and cross-sectional profiles are shown in Figure 5C and D for pristine CFs and Figure 5E and F for m-CFs, respectively. The roughness parameters increase from R_a = 57.5 nm and R_q = 77 nm for pristine CFs to R_a = 110 nm and R_q = 140 nm for bio-etched CFs. The cross-sectional profiles taken from representative upper, middle, and lower regions show the same trend: untreated fibers have small height fluctuations, whereas m-CFs display larger peak-to-valley differences. These data indicate that bio-etching reconstructs the fiber surface rather than merely changing its chemical composition. The roughened surface is important for the later polymerization step^[22]. The increased roughness may provide additional accessible sites for PANI deposition, consistent with the more continuous coating observed by SEM.

FTIR spectra of CFs and m-CFs [Figure 6A] show characteristic cellulose bands at approximately 3,300 and 2,880 cm^-1, corresponding to O-H and C-H stretching vibrations, respectively. The characteristic cellulose bands remain present after bio-etching, consistent with retention of the main cellulose functional groups^[23]. The bands at 1,106, 1,050, and 1,020 cm^-1 are associated with characteristic C-O and C-O-C vibrations of cellulose. Following PANI deposition, characteristic bands appear at 1,569 and 1,487 cm^-1, corresponding to the quinoid and benzenoid ring stretching vibrations of PANI, respectively^[24]. The band at approximately 1,300 cm^-1 is assigned to C-N stretching vibrations of PANI. Relative to PANI/CFs, PANI/m-CFs exhibited stronger PANI-associated absorptions, suggesting increased polymer deposition following bio-etching, in agreement with the TGA and XPS results.

The X-ray diffraction (XRD) patterns in Figure 6B show that the cellulose I crystalline structure is retained after bio-etching. Pristine CFs exhibit diffraction peaks at 2θ = 16.3° and 22.35°, assigned to the (101) and (002) planes, respectively. After bio-etching, these peaks remain visible with no obvious change in their positions. The crystallinity index calculated using the Segal method was 80.12% for CFs and 80.71% for m-CFs [Supplementary Table 4], consistent with retention of the cellulose I structure. For the PANI-loaded composites, a broad diffraction feature appears at approximately 2θ = 25°, accompanied by a decrease in the intensity of the cellulose reflections, indicating the incorporation of PANI into the cellulose-based composite. The XRD results provide structural information on the coexistence of cellulose and PANI.

TGA was used to compare thermal behavior and composition [Figure 6C]. The mass loss below 100 °C is mainly associated with physically adsorbed water. Bio-etched CFs exhibited a greater mass loss near 230 °C than pristine CFs, implying an increased amount of bound water after treatment. For PANI-containing composites, thermal degradation begins at approximately 250 °C, which is attributed to PANI decomposition and is consistent with the DTG profiles [Supplementary Figure 3]. The pronounced mass loss observed between 280 and 300 °C is related to the degradation of cellulose chains^[25]. The residual mass of PANI/m-CFs (36.25%) exceeded that of PANI/CFs (35.86%), indicating a higher PANI incorporation in the bio-etched composite, in agreement with the FTIR and XPS analyses.

XPS analysis confirms the compositional differences induced by bio-etching [Supplementary Figure 4]. The nitrogen content increases from 5.02% for PANI/CFs to 8.10% for PANI/m-CFs. Because nitrogen originates from PANI, the higher N concentration indicates higher polymer incorporation on the bio-etched fibers. The high-resolution C 1s spectra [Figure 6D] show a decrease in the O-C-O signal at 288.1 eV, attributed to bio-etching-mediated cleavage of β-1,4-glycosidic bonds and formation of new chain ends^[26].

In the PANI-loaded samples, the peaks at 284.8 eV (C-C), 287.1 eV (C-O/C-N), and 288.4 eV (C=N⁺) become stronger in PANI/m-CFs [Figure 6E], indicating increased exposure of reactive groups and higher PANI deposition. The N 1s spectra [Figure 6F] show three distinct peaks assigned to quinoid imine (398.7 eV, -N=), benzenoid amine (399.3 eV, -NH-), and protonated amine (401.5 eV)^[27], which indicates that PANI mainly exists in the emeraldine state. Notably, the stronger imine and protonated amine peaks in PANI/m-CFs suggest that bio-etching facilitated effective PANI protonation and carrier availability^[28].

The electrochemical behavior of PANI/CFs and PANI/m-CFs was investigated in a three-electrode cell. The paper-based composites were employed as working electrodes, while a platinum plate and Ag/AgCl electrode served as the counter and reference electrodes, respectively [Figure 7A]. The charge-storage process involves Faradaic reactions of PANI together with ion transport within the porous fiber framework. At 10 mV/s, PANI/m-CFs exhibits a larger CV response with approximately symmetric redox features than PANI/CFs, indicating enhanced pseudocapacitance [Figure 7B]^[29]. The GCD profiles at 1 mA/cm² exhibit a longer discharge duration and higher capacitance for PANI/m-CFs [Figure 7C]^[30]. This behavior is attributed to increased PANI incorporation and a three-dimensional conductive network generated by bio-etching, which promotes electrolyte accessibility and active-material utilization. EIS was employed to investigate interfacial charge-transfer characteristics [Supplementary Table 5 and Figure 7D]^[31]. The Nyquist plots reveal similar series resistance (R_s) values (2.39 Ω) for both electrodes, whereas PANI/m-CFs display lower charge-transfer resistance (R_ct) (1.56 Ω) and Warburg resistance (R_W) (1.23 Ω) values than PANI/CFs (2.73 and 3.18 Ω, respectively) [Figure 7E and F]. These results suggest that the porous, fold-rich surface generated by bio-etching facilitates charge transfer and ion transport within the electrode structure^[32].

At 1 mA/cm², the areal capacitance increased from 1,254 mF/cm² for PANI/CFs to 2,098 mF/cm² for PANI/m-CFs [Figure 7G]. The PANI/m-CFs electrode also retains 78.9% of its initial capacitance after 3,000 cycles, compared with 73.6% for PANI/CFs [Figure 7H]. The improved capacitance retention may be associated with strengthened interfacial interactions between PANI and the bio-etched cellulose fibers, which help maintain the continuity and structural integrity of the conductive network during repeated charge-discharge cycling^[29]. As shown in Figure 7I, the PANI/m-CFs electrode exhibits a competitive areal capacitance compared with previously reported PANI-based electrodes^[33-37]. Because these values were obtained with different electrolytes, test configurations, and current densities or scan rates, the corresponding testing conditions are summarized in Supplementary Table 6 for reference.

The electrochemical response of PANI/m-CFs can be understood from the combined roles of the PANI coating and the bio-etched fiber network. The PANI layer contributes redox-active sites and electron-conduction paths, while the roughened cellulose network provides electrolyte-accessible channels [Figure 8A]. The CV curves maintain similar shapes as the scan rate increases, consistent with charge storage over the tested range [Figure 8B]^[38,39]. The GCD curves are close to triangular, which is consistent with mixed electric double-layer capacitance and PANI pseudocapacitance [Figure 8C]^[40]. At 1 mA/cm², the areal capacitance reaches 2,098 mF/cm², and the coulombic efficiency is also improved [Figure 8D]^[41]. The electrode also maintains a high mass-specific capacitance [Figure 8E]. Figure 8F presents the electrode-level areal energy and power densities of PANI/m-CFs derived from the three-electrode GCD tests. These values are associated with the high PANI loading and the porous fiber framework produced by bio-etching.

The kinetic behavior was further analyzed using the relationship between peak current and scan rate^[42]. As shown in Figure 8G, the fitted b values for the oxidation and reduction peaks were 0.53 and 0.54, respectively. The values close to 0.5 indicate that the selected redox peak currents have a predominantly diffusion-dependent scan-rate response. To quantify the contributions, the current response was separated into k₁v and k₂v^1/2 terms according to Equations (9) and (10), and i/v^1/2 was fitted as a function of v^1/2[43]. As shown in Figure 8H, both contributions coexist throughout the potential window at 50 mV/s, consistent with contributions from both diffusion-controlled and surface-controlled processes in PANI:SSA. As shown in Figure 8I, the surface-controlled contribution increases from 20% to 57% as the scan rate increases. This trend can be attributed to the reduced time available for ion diffusion into the bulk electrode as the scan rate increases, causing the charge-storage process to increasingly rely on rapid surface-accessible reactions. These results reveal that the charge storage mechanism involves both diffusion-controlled Faradaic reactions and surface-controlled processes.

A symmetric supercapacitor was assembled using two identical PANI/m-CFs electrodes, and tested over a voltage window of 0 to 1.0 V in 1 M H₂SO₄^[44]. The CV curves retain similar profiles as the scan rate increases from 10 to 50 mV/s, with a corresponding increase in current response, consistent with charge storage within 0 to 1.0 V [Figure 9A]. The GCD profiles of the symmetric device exhibit stable and reversible charge-discharge behavior at current densities ranging from 1 to 10 mA/cm² [Figure 9B]. The Nyquist plot and equivalent circuit fit characterize the device impedance [Figure 9C]. The device achieves an areal capacitance of 1,265 mF/cm² at 1 mA/cm² and maintains 217 mF/cm² at 10 mA/cm² [Figure 9D], corresponding to 17.2% retention as the current density increases tenfold. Based on the GCD measurements, the device delivers a maximum areal energy density of 111.6 μWh/cm² at a power density of 398.5 μW/cm². Even at a high power density of 2,716.9 μW/cm², an energy density of 8.9 μWh/cm² is retained [Figure 9E], highlighting the potential of the assembled symmetric supercapacitor for practical energy-storage applications.

Mechanical deformation tolerance was further assessed by static bending^[45]. CV curves recorded at 0°, 40°, 80°, and 120° nearly overlap, with only minor differences in current response [Figure 9F]. The GCD profiles at the same bending angles also show comparable charge-discharge times and curve shapes [Supplementary Figure 5]. These results indicate that the electrode maintains a similar electrochemical response at the tested static bending angles, consistent with preservation of conductive pathways within the cellulose-fiber/PANI network. The bending measurements were performed on the electrode and do not establish the mechanical durability of the assembled symmetric device or its resistance to repeated bending.