Leaf-like hematite-decorated flexible carbon-textile for enhancing mass transfer at triphasic interfaces in photoanodes

Abstract

Photoelectrodes play a critical role in photoelectrochemical (PEC) reactions. However, the sluggish mass/electron transfer kinetics at the triphasic interface and inherent structural rigid features significantly limit their practical and scalable applications. In this work, we used the confinement effect of nanofibers to uniformly nucleate and grow leaf-like α-Fe₂O₃ nanoarrays on the surface of flexible, porous carbon textile-based photoelectrodes by a mild hydrothermal method. This strategy significantly enhances the bubble desorption while maintaining a high density of electrochemically active sites for electrolyte infiltration. This new PEC photoanode structure exhibits a current density of 0.4 mA cm⁻² under visible-light irradiation, which is 8 times higher than that of α-Fe₂O₃ arrays on traditional F-doped tin oxide (FTO) glass. The α-Fe₂O₃@oxidized carbon cloth also demonstrates excellent oxygen evolution reaction (OER) activity in the PEC system, with an overpotential of 193 mV at 10 mA cm⁻², a low Tafel slope of 42 mV dec⁻¹, and an oxygen production rate of 1.76 mmol g⁻¹ h⁻¹. Moreover, the flexible, free-standing PEC photoanode can withstand extreme working conditions such as folding and twisting, and can be designed into various shapes to expose a larger active surface. This work demonstrates a new photoanode strategy that solves the problems of slow triphasic interface mass transfer and rigidity, and provides great prospects for portable and wearable PEC devices.

---

1. Introduction

The utilization of solar energy through photoelectrochemical (PEC) cells has been identified as a potential solution to address future energy demands, including solar-driven water splitting and environmentally sustainable pollutant degradation.^1–7 The oxygen evolution reaction (OER) at the anode is considered the efficiency-determining key process for water splitting, as it is a multistep, four-electron transfer half-reaction.⁸ To meet the needs of PEC water splitting, the photoanode must process excellent photocorrosion resistance, an appropriate bandgap for solar energy harvesting, and low surface charge recombination.⁹ Semiconductor materials are regarded as optimal PEC photoelectrode candidates due to their capacity to readily adjust the band edge position and band gap width. Among various photoanode materials, hematite (α-Fe₂O₃) is of particular interest due to its suitable bandgap of approximately 2.1 eV, which enables a high theoretical current density at 1.23 V (vs. RHE) for water splitting.^10,11 Conventional rigid electrodes, including FTO and metallic foams, pose significant challenges in achieving a porous 3D network structure, deformability and shapeable nature simultaneously.^12–15 The rationale behind the utilization of carbonized natural fiber cloth as a new catalyst support is threefold: its low cost, high conductivity, porous three-dimensional network, and mechanical flexibility.^16–19 In general, an increase in graphitization leads to elevated electrical conductivity in carbon materials. However, during the carbonization process, it is challenging to preserve a flexible porous structure, a high surface area for loading catalysts and reactions, and to regulate the hydrophilicity of the triphasic interface simultaneously.

Generally, almost all PEC processes occur at the gas–liquid–solid triphasic interface. To sustain these chemical reactions, the reactants must be efficiently transported to the solid surface, while the gaseous products, such as oxygen, must be released from the solid phase.^20–22 However, most oxide-based photoanode materials exhibit an oxygen-philic surface that tightly adsorbs oxygen. This results in slow bubble release and blocking of active sites by gas bubbles. Continuous oxygen production consumes valuable reaction sites and significantly hinders the efficiency of the reaction. The overall activity of the system is further influenced by mass diffusion, active site exposure, and electron transfer. These are determined by the efficiency of gas–liquid–solid contact at the triphasic interface.²³ It has been reported that effective bubble detachment from the photoelectrode surface is essential to minimize mass transfer limitations and improve overall OER efficiency.²⁴

Recent advances in surface morphology and wettability modulation have demonstrated promising pathways to address this challenge. The regulation of nanostructures can not only maximize the electrochemically active surface area, but also establish channels for rapid gas transport, thereby reducing the bubble pinning effect. However, due to the diversity of physical and chemical environments on the surface of rough fibers, it is still challenging to uniformly grow high-curvature, hydrophobic tip nanostructures on carbonized natural fiber cloth. To increase the compatibility of the interface, the carbon cloth is usually subjected to hydrophilic and oxygen-containing functionalization pretreatment before crystal growth. Such a pretreatment strategy can uniformly introduce oxygen-containing functional groups on the surface of carbon fibers.^25,26

This study demonstrates a new strategy to introduce oxygen-containing functional groups on the surface of hydrophobic carbon materials by mild hydrothermal treatment. This process transforms the surface into a more hydrophilic carbon oxide cloth, thereby enabling the uniform formation of PEC semiconductor sites on each carbon fiber of the conductive carbon cloth. These sites contribute to the reactions occurring on the surface area and also regulate the behavior of the triphasic interface at the microscopic level. This enhances the overall performance and efficiency of the PEC process. The finite element analysis (FEA) simulation and the in situ bubble observation demonstrate the rapid absorption of liquids and the rapid desorption of bubbles. Based on these unique characteristics, the all-in-one PEC photoanode with high conductivity, flexibility, catalytic activity and triphasic interface behavior control achieved a photo-current of 0.4 mA cm⁻²; the current density is 8-times higher than that of α-Fe₂O₃@FTO, and the oxygen production rate of α-Fe₂O₃@oxidized carbon cloth-64 is 1.76 mmol g⁻¹ h⁻¹. The flexible carbon cloth electrode exhibits remarkable resilience to various forms of deformation, including bending, torsion, stretching, and folding, thereby ensuring reliable function even under significant strain. The innovative design of these electrodes enables their adaptation to a wide range of complex curved shapes, enhancing the stability of the catalyst during the reaction process. This study shows a new design method of a self-supporting structure, with the aim of facilitating mass transfer at triphasic interfaces. This strategy holds considerable promise for the development of wearable energy supply systems in the future.

2. Experimental section

2.1 Materials

Polyvinylpyrrolidone (PVP, M_w ≈ 1.3 × 10⁶) and Fe(III) acetylacetonate (Fe(acac)₃, AR) were obtained from Alfa Aesar Chemical Co., Ltd. Acetone (C₃H₆O, AR), acetic acid (C₂H₄O₂, AR), sodium hydroxide (NaOH, AR), ethanol (C₂H₆O, AR) and anhydrous sodium sulfate (Na₂SO₄, AR) were obtained from Sinopharm Chemical Reagent Co., Ltd. The cotton cloth was obtained from Zhongheng Dayao Textile Technology Co., Ltd and was trimmed to 30 × 40 cm² in size. Nafion (5 wt%, AR) was obtained from Sigma-Aldrich. All chemicals were used as received without further purification. The water used in all experiments was filtered through a Millipore filtration system with a resistivity of 18 MΩ cm.

2.2 Fabrication of the oxidized carbon cloth

The cotton cloth was soaked in 15 g L⁻¹ NaOH solution and boiled for 1 h to remove the slurry and impurities on the surface, then washed in ultrapure water until neutral, tested with a pH meter (PHSJ-SF, Shanghai Leici Sensor Technology Co., Ltd), and dried in an oven at 60 °C. The pretreated cloth was further cut into 4 × 2 cm² long strips and placed separately in a porcelain boat for heat-treatment under a N₂ atmosphere in a tube furnace at 600, 700 and 800 °C for 1 h. The treated carbon cloths were named carbon cloth-600, carbon cloth-700 and carbon cloth-800, respectively. The heating rate was 2 °C min⁻¹, while the cooling rate was 5 °C min⁻¹. At the end of calcination, the carbon cloth was cleaned by ultrasonic treatment with ethanol and acetone, and then dried in an oven at 60 °C; a photograph of the carbon cloth after washing can be found in Fig. S1.† Carbon cloth-800 was placed under an internal air atmosphere at 100, 150, 200, 250, 300, 350, and 400 °C for 2 h. The calcination heating rate was 5 °C min⁻¹, with natural cooling. The fabricated oxidized carbon cloths were named oxidized carbon cloth-100, oxidized carbon cloth-150, oxidized carbon cloth-200, oxidized carbon cloth-250, oxidized carbon cloth-300, oxidized carbon cloth-350, and oxidized carbon cloth-400, respectively. After calcination, the oxidized carbon cloth was cleaned with ethanol and acetone by ultrasonic treatment, and then the pretreated oxidized carbon cloth was put in the oven for 60 °C drying.

2.3 Fabrication of Fe(acac)₃/PVP nanofibers

Fe(acac)₃/PVP nanofibers were obtained by electrospinning the precursor containing 0.6 g of PVP, 4.5 mL of ethanol, 5.3 mL of acetic acid and 1.2 g of Fe(acac)₃ with a flow rate of 0.3 mL h⁻¹ at 17.5 kV. The relative humidity was maintained below 40%.

2.4 Fabrication of the α-Fe₂O₃@oxidized carbon cloth photoanode

16, 32 and 64 mg of Fe(acac)₃/PVP nanofibers were placed in a 25 mL PTFE skeleton, respectively, then 6.5 mL of acetic acid was added to the skeleton with 13.5 mL of ultrapure water and mixed evenly by sonication. The oxidized carbon cloth-300 (3.5 × 1.5 cm²) was soaked in the Fe(acac)₃/PVP nanofiber solution at the above different concentrations and the materials were named α-Fe₂O₃@oxidized carbon cloth-16, α-Fe₂O₃@oxidized carbon cloth-32, and α-Fe₂O₃@oxidized carbon cloth-64, and then they were sealed in a hydrothermal reactor and placed in an oven at 180 °C for 8 h. After the reaction, the α-Fe₂O₃@oxidized carbon cloth was repeatedly rinsed with ultrapure water, acetone, and ethanol to remove residual PVP.

2.5 Characterization

Scanning electron microscopy (SEM) images were obtained using a FEI field-emission microscope (Nova Nano SEM 230). Transmission electron microscopy (TEM) images were collected using a transmission electron microscope (Tecnai G2 T20, FEI). The square resistance of all samples was obtained using an RTS-9 type four-probe tester The crystal structure information was obtained with an X-ray diffractometer (Bruker, D8 advance using Cu-Kα radiation, λ = 1.5406 Å). Raman spectra were recorded on a Jobin Yvon LabRAM HR800 micro-Raman spectrometer with a 532 nm laser. Thermogravimetric analysis (TGA) was conducted on an SDT Q600 at a heating rate of 20 °C min⁻¹ under air flow. The contact angle was collected by the contact angle instrument (OCA 15 plus, Dataphysics). Fourier transform infrared (FTIR) spectra were collected using a Nicolet 5700 spectrometer.

2.6 Photoelectrochemical measurements

Photoelectrochemical performances of the photoanode were measured at a three-electrode applied constant voltage at 0.6 V (vs. Ag/AgCl) in 0.2 M Na₂SO₄. The α-Fe₂O₃@oxidized carbon cloth, the platinum wire, and the silver–silver chloride (Ag/AgCl) electrode were applied as the working, counter and reference electrodes, respectively. The α-Fe₂O₃@oxidized carbon cloth photoelectrodes were illuminated using a 300 W xenon lamp (Beijing Au-Light, CEL-HXF300-T3), the actual optical density was measured using a densitometer. Upon irradiation, the responding photocurrent was recorded using an electrochemical workstation (Gamry Reference 3000). The electrochemical impedance spectra (EIS) were measured in the frequency range of 0.1 Hz to 1 MHz at the open-circuit voltage. 0.5 M H₂SO₄ electrolyte was used for the OER performance test, the carbon rod and Ag/AgCl were used as the counter electrode and the reference electrode. The stable polarization curve was recorded after at least 50 cycles of cyclic voltammetry (CV) at a scan rate of 2 mV s⁻¹. The recorded potential (E_Ag/AgCl) was calibrated to the reversible hydrogen electrode (RHE) using the equation: E_RHE = E_Ag/AgCl + 0.197 + 0.0591 × pH.

2.7 Simulations of the dynamic wettability droplet adsorption

These results were simulated using COMSOL Multiphysics 6.0. The droplet drops were simulated using Laminar flow, phase-field interface and transient state analysis of the horizontal set in COMSOL multiphysics 6.0. The gravity was considered in the calculation. The top surface was set as the open border.

3. Results and discussion

3.1 Structure and morphology of the fabricated photoanode

As illustrated in Scheme 1a, the design for enhancing PEC performance entails the following: the reaction site in the triphasic interfaces is exposed to the maximum extent, and the reaction performance is enhanced by regulating the desorption of the product oxygen bubble.²⁷

Scheme 1 (a) Mechanism diagram for the enhanced photoanode performance of the α-Fe₂O₃@oxidized carbon cloth photoanode. (b) The left shows the strategy of mildly introducing oxygen-containing functional groups, and the right shows the traditional method.

At the micro level (Scheme 1b), most of the Fe₂O₃ will nucleate homogeneously in the solution when the traditional method is used to grow Fe₂O₃, and it is difficult to grow Fe₂O₃ grains on the hydrophobic carbon surface in an array manner. Therefore, the development of hydrophobic carbon surfaces often necessitates complex chemical modifications, such as the introduction of hydrophilic functional groups, to enhance their compatibility with hydrophilic oxides. This static modification operation is complex, and it is difficult to achieve a wide range of uniform coverage.^28,29

An innovative strategy was adopted, entailing the gradual transformation of a more hydrophobic carbon surface to hydrophilicity through mild hydrothermal treatment. The carbon cloth after oxidation treatment has a higher specific surface energy and more easily adsorbs Fe³⁺. This approach unveils a new method for the homogeneous heterogeneous nucleation of hydrophilic oxides on hydrophobic carbon substrates. The strategy meticulously crafts an interface environment that can be dynamically regulated during the hydrothermal treatment process. As the hydrothermal reaction progresses, a substantial number of oxygen-containing functional groups, including hydroxyl, carboxyl, and carbonyl, are introduced onto the surface of carbon materials in a gentle and gradual manner. These functional groups can chemically adsorb Fe³⁺ to form chemical bonds, thus providing an attachment point for the growth of α-Fe₂O₃. Some of the functional groups are negatively charged after dissociation in water, while Fe³⁺ is positively charged. The attraction between the two is driven by electrostatic interaction, which further enhances the adsorption and aggregation of Fe³⁺ on the surface of carbon cloth. The addition of CH₃COOH during the hydrothermal process regulated the pH of the crystal growth environment (pH = 2), which in turn changed the ionic strength in the solution. The increase of solution viscosity will reduce the size of the crystal nucleus. From the reported work, we can infer that Fe³⁺ can chelate three CH₃COO⁻ groups or simply coordinate with six OH⁻ groups to form two octahedral structures with different stability.³⁰ This difference can change the kinetics of nucleus formation and crystal growth, resulting in anisotropy of nucleus growth, thus forming leaf-like Fe₂O₃. This interesting heterostructure exhibits a leaf-like array morphology with a single leaf size of about 60 nm × 80 nm. The staggered arrangement is not shielded from each other and is completely exposed to sunlight. Therefore, the multiple scattering of light near the α-Fe₂O₃ leaf array can improve the light capture efficiency.²³

In addition, polyvinylpyrrolidone (PVP) in Fe(acac)₃/PVP nanofibers is an amphiphilic polymer with a special structure, which contains pyrrolidone groups and C–C chains. The hydrophilic pyrrolidone group can form hydrogen bonds with –OH and –COOH groups on the surface of the carbon cloth. The C–C chain has a certain extent of hydrophobicity, which can be combined with the hydrophobic part of the carbon cloth surface through hydrophobic interaction. The van der Waals force between the C–C chains can also enhance the overall binding force. The presence of PVP can help delay the agglomeration of the obtained particles. Through the interaction between the hydroxyl/carboxyl group of PVP and the hydroxyl group on the oxide, the strong space force between the two can stabilize the generated nanoparticles and form monodisperse crystals. Concurrently, the elevated temperatures and pressures within the environment expedite the uniform diffusion of ions within the solution, thereby effectively mitigating the risk of excessive local nucleation. Under these microscopic effects, a sufficiently dispersed array of Fe₂O₃ is formed on the surface of the carbon cloth.

Fig. S1† illustrates optical pictures of the cotton cloth and carbon cloth, which clearly show that the color changes from white to black. As shown in Fig. 1a, the carbonization and oxidation process remain in the initial porous network of the interwoven oxidized carbon cloth structure within the cotton cloth.^31,32 As shown in Fig. S2(a),† the surface of the unoxidized carbon cloth fiber is smooth, and the oxidation process increases the surface wrinkles of the carbon cloth (Fig. 1b). The formation mechanism of surface wrinkles during carbon cloth oxidation can be attributed to two primary factors: firstly, the chemical structure change induced by the oxidation reaction, and secondly, local thermal expansion and stress accumulation. During the oxidation process, oxygen-containing functional groups are formed on the surface of carbon fibers, thereby altering the chemical properties and physical structure of the surface. This, in turn, leads to fracture and the formation of chemical bonds, as well as an increase in surface energy, consequently resulting in surface distortion. Concurrently, the local thermal expansion coefficient disparity and stress accumulation further promote the formation of wrinkles. Compared with carbon cloth loaded α-Fe₂O₃, the surface roughness of oxidized carbon cloth can induce more α-Fe₂O₃ (Fig. 1c and S2(b)†). The surface wrinkles make the carbon oxide cloth have a large specific surface area and porous structure, which provides more growth space for α-Fe₂O₃. When α-Fe₂O₃ grows on the surface of carbon oxide cloth, it will be driven by surface energy, and its surface energy will gradually decrease. A stable interface is formed between α-Fe₂O₃ and the carbon cloth. This interface interaction includes not only chemical bonding, but also physical forces such as van der Waals force. These interaction forces can enhance the bonding strength between the particles and the matrix and prevent the particles from falling off during the growth process, thus ensuring that α-Fe₂O₃ can grow smoothly on the carbon oxide cloth and form a solid composite structure.

Fig. 1 (a) SEM image of the carbon cloth surface. SEM surface topography of (b) oxidized carbon cloth-300 and (c) α-Fe₂O₃ loading on the surface of oxidized carbon cloth. (d) HRTEM image of α-Fe₂O₃@oxidized carbon cloth and (e) the inverse-FFT pattern. The red and yellow boxes of (d) correspond to (e). (f) EDS mapping image of α-Fe₂O₃@oxidized carbon cloth, respectively. Strength profile along the red line 1–2 in the illustration.

In order to further verify the effect of the carbonization process on the structure of the material, we further analyzed the defects of the material and the changes of functional groups by Raman spectroscopy and FTIR. These analyses provide a more comprehensive perspective for understanding the structural evolution during oxidative modification.

As shown in Fig. 1d and e, two well-defined lattice fringes with d-spacings of 0.251 and 0.268 nm were identified and assigned to the lattice planes of (110) and (104) of α-Fe₂O₃, respectively, which was consistent with previous reports.^33,34 Fe(acac)₃/PVP nanofibers obtained by electrospinning are shown in Fig. S3.† Phase composition of the composites was analyzed by X-Ray diffraction (XRD) (Fig. 2f), the results show high crystallinity of the main characteristic surfaces such as (012), (104), (110), (024), and (116). The energy dispersive spectrometer (EDS) mapping image (Fig. 1f) shows that α-Fe₂O₃ can be firmly combined with the oxidized carbon cloth substrate. In this case, a tightly integrated interface enhances the directional electron transport ability.³⁵

Fig. 2 (a) XRD patterns, (b) Raman spectrum and (c) I_D/I_G histogram of carbon cloth with different carbonization temperatures. (d) The thermogravimetric curve of carbon cloth-800 under an air atmosphere. (e) FTIR spectra of carbon cloth-800, oxidized carbon cloth-300, oxidized carbon cloth-350. (f) XRD spectrum of α-Fe₂O₃@oxidized carbon cloth. (g) XPS spectra of O 1s of oxidized carbon cloth and α-Fe₂O₃@oxidized carbon cloth-64.

Fig. S4(a–d)† show the SEM images of the carbon oxide cloth which underwent hydrothermal reaction for 2 h, 4 h, 6 h and 8 h, respectively. We can clearly see that the loaded particles gradually become smaller and denser. At the beginning, Fe(acac)₃/PVP nanofibers were mostly spherical after dissolution in the hydrothermal solution. With the increase of hydrothermal time, oxygen-containing functional groups were gradually introduced on the surface of the carbon cloth. The anchoring effect of Fe³⁺ makes Fe₂O₃ grow uniformly on the carbon cloth. At the same time, the EDS energy spectrum records the distribution of each element of α-Fe₂O₃@oxidized carbon cloth-64 at the different amplified times (Fig. S4(e and f)†). The uniformity of Fe₂O₃ growth on the carbon cloth can be seen.

Fig. 2a depicts the XRD patterns of carbon cloth at different carbonization temperatures. As evidenced by the broad peaks of the (002) and (100) planes, 2θ is located at approximately 23.15° and 43.5°, respectively. As the carbonization temperature increases, the relative intensities of peaks become stronger, and the full width-half maxima values get narrower. These demonstrate that the degree of graphitization is increased. A significant change in crystallinity was not observed as the temperature increased from 600 to 800 °C. In order to better study the effect of the carbonization temperature on the structural characteristics of carbon cloth, the degree of graphitization of the carbon cloth was measured by Raman analysis. Fig. 2b shows two characteristic peaks at 1334 cm⁻¹ and 1587 cm⁻¹ of three different carbonization temperatures, attributed to the D bands (sp³ hybridized carbon) and G bands (sp² hybridized graphitic carbon).^36,37 The intensity ratio of D and G bands (I_D/I_G) is related to the structural defects and disordered states of carbon materials. In Fig. 2c, I_D/I_G increased from 0.82 to 0.97, the increase of I_D/I_G value indicates an increase in the internal defects. This further indicates that enhancing the carbonization temperature can increase the structural defects and improve the graphitization degree of carbon cloth.

The carbon cloth has both flexibility and electrical conductivity comparable to FTO, but its fiber surface will become relatively smooth and hydrophobic. Thermal oxidation treatment of carbon cloth was carried out to improve the roughness and liquid absorption rate of the fiber surface without losing the conductivity. Thermal decomposition test was conducted under an air atmosphere to determine the appropriate thermal oxidation temperature interval. As shown in Fig. 2d, carbon cloth began to lose weight sharply around 400 °C. Therefore, moderate oxidation could be achieved below 400 °C. The thermal oxidation temperature interval was finally found to be below 400 °C. The temperature at around 400 °C is the most active temperature range for the oxidation of carbon-based materials during the oxidation process, and it will not lead to excessive oxidation and serious damage to the surface structure. We further tested carbon cloth-800, oxidized carbon cloth-300, and oxidized carbon cloth-350 to determine the internal functional group changes during oxidative modification. The FTIR spectra demonstrated in Fig. 2e confirm the presence of O–H (∼3416 cm⁻¹) and C=O (∼1624 cm⁻¹) groups. The presence of the C=O group observed in oxidized carbon cloth showed the oxidation of carbon cloth with the oxidation treatment. With the oxidation treatment, the content of groups increased, and the oxidized carbon cloth changed from hydrophobic to hydrophilic. It provided the basis for the active sites to fully react with the reactants.

Fig. 2g shows the X-ray photoelectron spectra (XPS) of oxidized carbon cloth and α-Fe₂O₃@oxidized carbon cloth-64. Before hydrothermal treatment, the O 1s spectrum mainly contains two peaks, corresponding to the C=O bond (approximately 531.5 eV) and C–O bond (approximately 533.5 eV). The presence of these peaks indicates that a certain amount of oxygen-containing functional groups exist on the sample surface after oxidation at 300 °C. After hydrothermal loading of α-Fe₂O₃, the O 1s XPS spectrum shows that in addition to the original C=O and C–O peaks, a new peak located at approximately 529.5 eV appears, which corresponds to the Fe–O bond. This peak indicates that α-Fe₂O₃ is successfully loaded on the sample surface, and Fe³⁺ forms chemical bonds with the oxygen-containing functional groups on the sample surface through Fe–O bonds. The formation of Fe–O bonds also confirms the anchoring effect of Fe³⁺. The anchoring of Fe³⁺ can enhance the interaction between α-Fe₂O₃ and the carbon material, thereby improving the stability and activity of the catalyst. The anchoring effect is of great significance for improving the performance of the catalyst, as it can prevent the aggregation and detachment of α-Fe₂O₃ during the reaction, thus maintaining the structural integrity of the catalyst and the accessibility of active sites.

Fig. S5(a)† shows the C 1s XPS spectra comparing the samples before and after hydrothermal treatment. The peak at approximately 284.8 eV corresponds to the C–C bond, while the peaks at approximately 286.5 eV and 288.5 eV correspond to the C–O bond and C=O bond, respectively. By comparing the two spectra, it is observed that the intensities of the C–O and C=O peaks significantly increase after hydrothermal treatment, indicating that oxygen-containing functional groups are successfully introduced during the hydrothermal process. Fig. S5(b)† displays the Fe 2p XPS spectrum, reflecting the chemical state of Fe on the surface of the sample after the hydrothermal loading of Fe₂O₃. The peaks at approximately 710–712 eV and 724–726 eV correspond to Fe 2p3/2 and Fe 2p1/2, respectively. Spectral analysis shows that Fe in the sample mainly exists in the Fe³⁺ oxidation state. In addition, an ICP-MS test was carried out and the Fe element loading per milligram of carbon cloth was calculated to draw a histogram (Fig. S6†). The content of Fe in Fe₂O₃@carbon oxide cloth-64 was the highest, which was about 0.0014 mg mg⁻¹. This indicates that Fe₂O₃ was successfully loaded on carbon cloth by hydrothermal treatment.

3.2 Flexibility, electrical conductivity, and hydrophilic surface structure performance

The oxidized carbon cloth exhibits good electrical conductivity and hydrophilicity. Fig. 3a shows the flexibility of carbon cloth. As an electrode material alternative to FTO, oxidized carbon cloth not only has electrical conductivity but also has the feature of flexibility and bending. An important indicator to evaluate the electrode material is the conductivity. When the carbonization temperature is 600 to 800 °C, the square resistance is 120.9 ± 2.31 to 32.9 ± 2.49 Ω sq⁻¹, meeting the needs of electrode materials (Fig. S7†). This change may be related to the rearrangement of carbon chains, the increase of crystallinity and the decrease of structural defects during carbonization. As shown in Fig. S8,† as the thermal oxidation temperature increases, the sheet resistance of the carbon oxide cloth remains around 33 Ω sq⁻¹. To assess the electrical stability of the flexible PEC anode, carbon cloth-800 was tested for electrical conductivity under bending deformation (Fig. S9†). The electrical stability of the flexible PEC anode was evaluated by testing its conductivity under bending deformation. Carbon cloth-800, positioned between the vernier caliper, showed stable resistance even as the bending degree changed, demonstrating its good flexibility and potential for use in flexible electrodes.

Fig. 3 Schematic diagram of (a) fabrication procedure of flexible carbon cloth. (b) FEA simulation of water absorption of carbon oxide cloth and carbon cloth. (c) The schematic illustration and dynamic contact angles of the oxygen bubble on the surfaces of (i) the α-Fe₂O₃@oxidized carbon cloth and (ii) the oxidized carbon cloth in water.

To better understand the impact of surface modification on liquid adsorption, FEA simulations were performed to simulate the dynamic droplet adsorption process on carbon cloth surfaces. As shown in Fig. 3b, the blue spheres represent droplets, and the red background represents air. FEA simulation suggested that the oxidized carbon cloth surface adsorbs liquid within 32 ms more than the carbon cloth surface. This result shows that the increased surface functional groups can change the surface nature from hydrophobic to hydrophilic (Fig. S10(a)†). The enhancement of liquid absorption indicates that carbon materials can provide more reaction sites in the catalytic reaction (Fig. S10(b)†). Fig. S11† shows the water absorption experiment of oxidized carbon cloth (Fig. S11(a)†) and carbon cloth (Fig. S11(b)†). The experimental results are consistent with the simulation results. This is conducive to the adsorption and penetration of the electrolyte inside the carbon fiber, thereby improving the electrochemical performance. Considering that more oxygen bubble adsorption at the wetting interface can seriously occupy active sites, as shown in Fig. 3c, the α-Fe₂O₃@oxidized carbon cloth had a larger bubble contact angle (136°) than the oxidized carbon cloth (122°), and the bubbles escaped rapidly from the surface for only 120 ms. The larger bubble contact angle (136°) of α-Fe₂O₃@oxidized carbon cloth indicates that the heterostructure exhibits stronger aerophobic properties. This leads to weaker adhesion forces between the surface and oxygen gas bubbles, resulting in faster bubble release and improved oxygen evolution reaction (OER) performance. The oxygen production rate of 1.76 mmol g⁻¹ h⁻¹ for α-Fe₂O₃@oxidized carbon cloth-64 not only confirms the effectiveness of this aerophobic interface but also highlights the superior catalytic performance of the modified photoanode.

3.3 Evaluation of the photoanode performance

The α-Fe₂O₃@oxidized carbon cloth was used as the photoanode with intriguing hydrophilic/aerophobic features. The performance of the α-Fe₂O₃@oxidized carbon cloth photoanode was evaluated by measuring the photocurrent density–potential (J–V) curve in a three-electrode system (Fig. 4a and b). As shown in Fig. 4a, when the light starts to illuminate the electrode, the current density increases rapidly and stabilizes. The α-Fe₂O₃@oxidized carbon cloth-64 exhibited the maximum photocurrent, which is 2.5-fold higher than that of α-Fe₂O₃@oxidized carbon cloth-16. This significant enhancement in photocurrent density is attributed to the synergistic effect of the increased surface area of oxidized carbon cloth and the enhanced charge separation efficiency facilitated by the α-Fe₂O₃ layer. The transient photocurrent measurements of photoanodes were further performed to evaluate the charge recombination behavior in the photoanode. The effective coupling of α-Fe₂O₃@oxidized carbon cloth represented a superior photocurrent density of 0.4 mA cm⁻², and α-Fe₂O₃@oxidized carbon cloth-64 exhibited a high current density, which is 8-times higher than that of α-Fe₂O₃@FTO.³⁰ Compared with other anode research work, the α-Fe₂O₃@oxidized carbon cloth composite optical anode shows superior photoelectric performance (Fig. 4b).

Fig. 4 (a) The transient photocurrent density of α-Fe₂O₃@oxidized carbon cloth-16, 32, and 64 composite materials. (b) The photocurrent density changes of α-Fe₂O₃@oxidized carbon cloth. (c) EIS of α-Fe₂O₃@oxidized carbon cloth. (d) LSV response of oxidized carbon cloth and α-Fe₂O₃@oxidized carbon cloth-16, 32, and 64 composite materials under light illumination. (e) Tafel plots derived from the polarization curves in (d). (f) Comparison of catalytic activity of α-Fe₂O₃@oxidized carbon cloth-64 with other work.

The charge transfer process within the α-Fe₂O₃@oxidized carbon cloth photoanode was conducted by EIS. The electrode material shows a typical impedance curve, and the semicircular diameter of the high-frequency area decreases with the increasing load, implying a sharp decrease in the charge transfer resistance (Fig. 4c). By fitting part of the EIS data, an equivalent circuit consists of an oxidized carbon cloth substrate (CPE1) and an α-Fe₂O₃ catalyst (CPE2), and the charge transfer resistance (CTR) consists of two parts, namely the transfer resistance (CTR1) between the substrate and the transfer resistance (CTR2) between the catalyst and the electrolyte.^38,39 When the loading increases, CTR2 drops sharply to 54.1 Ω (Table S1†), because the porous 3D network oxidized carbon cloth with a larger specific surface area provides effective places for the loading of α-Fe₂O₃ and can weaken the crystal accumulation phenomenon to avoid the transmission of electrons between weakly conductive crystals. Moreover, the α-Fe₂O₃ branches provided a direct pathway for electron transfer, thus accelerating the photoanode catalytic reaction.

The porous carbon fabric-modified leaf-like α-Fe₂O₃ array photoelectrode demonstrated in this work relies on its high conductivity, flexibility, and catalytic activity to quickly release the bubbles which are generated on the electrode surface, thereby ensuring a high density of unoccupied active sites on the electrode surface. The porous structure, large specific surface area and efficient interfacial charge transfer of the α-Fe₂O₃@oxidized carbon cloth greatly improve the performance of the photoanode, and meet the needs of PEC water splitting. The electrochemical water oxidation performance of all samples was further evaluated in a three-electrode system in 0.5 M H₂SO₄. Fig. 4d shows the OER polarization curve of reverse scanning linear sweep voltammetry (LSV) at a scan rate of 5 mV s⁻¹. In all samples, the α-Fe₂O₃@oxidized carbon cloth-64 exhibits excellent OER activity, with an overpotential of 193 mV at a current density of 10 mA cm⁻², which is much lower than that of the oxidized carbon cloth (393 mV), α-Fe₂O₃@oxidized carbon cloth-32 (238 mV) and α-Fe₂O₃@oxidized carbon cloth-16 (286 mV). The corresponding Tafel slope of α-Fe₂O₃@oxidized carbon cloth-64 is 42 mV dec⁻¹, which is significantly smaller than those of α-Fe₂O₃@oxidized carbon cloth-32 (82 mV dec⁻¹) and α-Fe₂O₃@oxidized carbon cloth-16 (96 mV dec⁻¹), indicating the fast OER kinetics (Fig. 4e). We tested the electrochemical stability of samples with different loadings. We applied a voltage of 0.6 V. At the beginning, the current of α-Fe₂O₃@carbon oxide cloth-64 reached 2.8 mA, and the other two samples had a current basically at about 2.3 mA. In order to reflect the stability more intuitively, we selected the working current value per 10 000 s as the comparison point. The proportion of the reduction difference between each point and the previous point was calculated and plotted in Fig. S12.† The image shows that the stability of α-Fe₂O₃@carbon oxide cloth-64 is the best of the three. In general, the current reduction gradually stabilized after working for 20 000 s. Table S2† briefly introduces the performance of the catalysts used for PEC water splitting in this work and in recent years.^40–52 It is not difficult to understand that the active sites increase with the increase of catalyst loading, and the PEC performance is also better. However, the load of α-Fe₂O₃ should not be increased blindly, otherwise it is difficult for the electrolyte to fully come in contact with the photoanode. The functional groups on the surface of the carbon cloth make it have good hydrophilicity, and the electrolyte can fully come in contact with the photoanode. The addition of α-Fe₂O₃ makes the photoanode more hydrophobic on the basis of re-hydrophilicity. As soon as the oxygen bubble is generated, it is quickly detached from the surface of the photoanode, thereby exposing more active sites. By managing the generation–desorption of bubbles at the triphasic interfaces, we obtained photoelectrodes with excellent electrochemical activity. The oxygen content was measured by gas chromatography, and we found that the oxygen production rate of α-Fe₂O₃@oxidized carbon cloth-64 is 1.76 mmol g⁻¹ h⁻¹. Fig. 4f shows a comparison of the oxygen production catalytic performance of α-Fe₂O₃@oxidized carbon cloth-64 with other work.^53–62

Compared with the traditional compact photoanodes such as particle-filled ones, this stable heterostructure with leaf-like α-Fe₂O₃ loaded on oxidized carbon cloth possessed various structural superiorities: (i) top-down arrangement of α-Fe₂O₃ arrays improved the utilization of visible light through multiple light-scattering. The leaf-like α-Fe₂O₃ arrays endow the rapid desorption of oxygen bubbles, thus achieving regulation of reaction sites. (ii) Oxidized carbon cloth with an excellent hygroscopic surface enlarged the exposure of active sites and promoted mass/electron transport. (iii) Oxidized carbon cloth as a conductive substrate, avoiding the use of adhesives while having excellent electrical conductivity.

4. Conclusion

In summary, a photoanode with excellent electrochemical activity was obtained by regulating bubble management at the triphasic interface. In terms of the material structure, a dynamic strategy of mild hydrothermal treatment was adopted to gradually introduce controlled oxygen-containing functional groups on the surface of carbon materials for anchoring Fe³⁺ sites. Thus, a leaf-like α-Fe₂O₃ array was designed on the carbon oxide cloth. This approach provides a new method for the homogeneous and heterogeneous nucleation of hydrophilic oxides on hydrophobic carbon substrates. The leaf-like α-Fe₂O₃ nanoarrays on oxidized carbon cloth exhibit unique hydrophilicity/aerophobicity features, which facilitate the water affinity and release of generated oxygen bubbles at the active sites. The rationality of the design was verified by FEA simulation and in situ bubble generation. The oxygen production rate of α-Fe₂O₃@oxidized carbon cloth-64 is 1.76 mmol g⁻¹ h⁻¹, which also confirms the effectiveness of bubble management. The α-Fe₂O₃ arrays on oxidized carbon cloth can improve the utilization of visible light through multiple light-scattering. In addition, the flexible nature allows the photoanode to adapt well to harsh working environments such as twisting and folding. Based on these characteristics, the α-Fe₂O₃@oxidized carbon cloth shows good structural-stability and exceptional performances towards oxygenation. The α-Fe₂O₃@oxidized carbon cloth delivers a superior photocurrent, which is 8-times higher than that of α-Fe₂O₃@FTO. This study demonstrated the efficacy of carbonized natural fiber cloth as a substrate material, exhibiting high conductivity, a high surface area, and wettability. The study indicated that this material has the potential to replace traditional rigid substrates in the field of photoanode preparation. The findings provide a new approach for the rational design and optimization of advanced photoanode materials for wearable energy supply.

Author contributions

Zhou Zhou: conceptualization, methodology, investigation, formal analysis, and writing, original draft. Mengmeng Zhu: conceptualization, formal analysis, and writing. Chengkun Song: conceptualization, formal analysis, and investigation. Mingyu Tang: software. Shujing Li: writing. Xiangyu Meng: funding acquisition, conceptualization, supervision, and formal analysis. Yueming Sun: supervision. Yunqian Dai: funding acquisition, conceptualization, supervision, formal analysis, and writing, review and editing.

Data availability

The data supporting the findings of this study are available within the article and its ESI.† Additional data that support the findings of this study are available from the corresponding author upon reasonable request.

Conflicts of interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

This work was financially supported by the National Key Research and Development Program of China (no. 2022YFA1505700), the National Natural Science Foundation of China (no. 22475044), the Equipment Pre-research Fund of Ministry of Education of China (no. 8091B022212), the Qinglan Talent Project in Jiangsu Province, the Fundamental Research Funds for the Central Universities (5007042405), the Key Laboratory of High Performance Fibers & Products Ministry of Education Donghua University, the Jiangsu Funding Program for Excellent Postdoctoral Talent (JB23001), the Natural Science Foundation of Jiangsu Province (BK20241355), and the China Postdoctoral Science Foundation (2024M750411 and 2024T170137).

References

1. K. Ren, J. Zhou, Z. Wu, Q. Sun and L. Qi, Small, 2023, 20, 2304835.
2. J. Wang, H. Cheng, Y. Cui, Y. Yang, H. He, Y. Cai, Z. Wang, L. Wang and Y. Hu, ACS Nano, 2022, 16, 21248–21258.
3. C. Li, M. Huang, Y. Zhong, L. Zhang, Y. Xiao and H. Zhu, Chem. Mater., 2018, 31, 171–178.
4. H. Zhou, D. Zhang, H. Xie, Y. Liu, C. Meng, P. Zhang, F. Fan, R. Li and C. Li, Adv. Mater., 2023, 35, 2300914.
5. P. Kumari, N. Bahadur, X. A. Conlan, X. Zeng, L. Kong, L. A. O'Dell, A. Sadek, A. Merenda and L. F. Dumée, Chem. Eng. J., 2023, 452, 139374.
6. H. Rajput, E. E. Kwon, S. A. Younis, S. Weon, T. H. Jeon, W. Choi and K. H. Kim, Chem. Eng. J., 2021, 412, 128612.
7. J. J. Liu, S. N. Sun, J. Liu, Y. Kuang, J. W. Shi, L. Z. Dong, N. Li, J. N. Lu, J. M. Lin, S. L. Li and Y. Q. Lan, J. Am. Chem. Soc., 2023, 145, 6112–6122.
8. J. Deng, Y. Su, D. Liu, P. Yang, B. Liu and C. Liu, Chem. Rev., 2019, 119, 9221–9259.
9. G. Dong, L. Yan and Y. Bi, J. Mater. Chem. A, 2023, 11, 3888–3903.
10. Y. Zhang, Y. Huang, S. S. Zhu, Y. Y. Liu, X. Zhang, J. J. Wang and A. Braun, Small, 2021, 17, 2100320.
11. S. Y. Yuan, L. W. Jiang, J. S. Hu, H. Liu and J. J. Wang, Nano Lett., 2023, 23, 2354–2361.
12. X. Wang, L. Wang, W. Xiu, X. Yang, Y. Yang and X. Li, ACS Appl. Nano Mater., 2024, 7, 7327–7336.
13. H. Wang, Y. Li, D. Deng, M. Li, C. Zhang and L. Luo, ACS Appl. Nano Mater., 2021, 4, 9821–9830.
14. Z. Tu, X. Liu, D. Xiong, J. Wang, S. Gong, C. Xu, D. Wu and Z. Chen, Chem. Eng. J., 2023, 475, 146253.
15. S. Banerjee, Y. Myung, S. Raman and P. Banerjee, ACS Sustainable Chem. Eng., 2015, 3, 3197–3204.
16. J. Liang, S. Luo, D. Pan, P. Xu, F. Zhan and J. Li, Chem. Eng. J., 2023, 464, 142646.
17. Á. Torrinha and S. Morais, Trends Anal. Chem., 2021, 142, 116324.
18. L. Li, S. Peng, N. Wang, M. Srinivasan, S. G. Mhaisalkar, Q. Yan and S. Ramakrishna, Small, 2015, 11, 2429–2436.
19. L. Zhang, Y. D. Hu and J. H. Zheng, J. Mater. Chem. A, 2017, 5, 18664–18673.
20. X. Meng, C. Zhu, X. Wang, Z. Liu, M. Zhu, K. Yin, R. Long, L. Gu, X. Shao, L. Sun, Y. Sun, Y. Dai and Y. Xiong, Nat. Commun., 2023, 14, 2643.
21. M. A. Marwat, M. Humayun, M. W. Afridi, H. Zhang, M. R. A. Karim, M. Ashtar, M. Usman, S. Waqar, H. Ullah, C. Wang and W. Luo, ACS Appl. Energy Mater., 2021, 4, 12007–12031.
22. J. Sun, H. Xue, J. Sun, N. Guo, T. Song, Y. R. Hao and Q. Wang, Chin. Chem. Lett., 2024, 35, 109002.
23. X. Sheng, Z. Liu, R. Zeng, L. Chen, X. Feng and L. Jiang, J. Am. Chem. Soc., 2017, 139, 12402–12405.
24. S. Yan, Y. Li, X. Yang, X. Jia, J. Xu and H. Song, Adv. Mater., 2023, 36, 2307967.
25. Y. Duan, L. Wei, C. Cai, J. Mi, F. Cao, J. Chen, X. Li, X. Pang, B. Li and L. Wang, Nano Res., 2024, 17, 4764–4772.
26. Y. Huang, Q. Yu, M. Li, S. Jin, J. Fan, L. Zhao and Z. Yao, Chem. Eng. J., 2021, 418, 129474.
27. F. Feng, C. Li, J. Jian, X. Qiao, H. Wang and L. Jia, Chem. Eng. J., 2019, 368, 959–967.
28. G. Zhu, M. L. Sushko, J. S. Loring, B. A. Legg, M. Song, J. A. Soltis, X. Huang, K. M. Rosso and J. J. De Yoreo, Nature, 2021, 590, 416–422.
29. Y. Xia, K. D. Gilroy, H. C. Peng and X. Xia, Angew. Chem., Int. Ed., 2017, 56, 60–95.
30. W. Fu, K. Liu, X. Zou, W. Xu, J. Zhao, M. Zhu, S. Ramakrishna, Y. Sun and Y. Dai, ACS Sustainable Chem. Eng., 2019, 7, 12750–12759.
31. J. Dong, X. Ouyang, B. Huo, D. M. Deng, X. X. Yan and L. Q. Luo, ACS Appl. Energy Mater., 2024, 7, 6564–6573.
32. H. Liu, M. Osenberg, L. Ni, A. Hilger, L. Chen, D. Zhou, K. Dong, T. Arlt, X. Yao, X. Wang, I. Manke and F. Sun, J. Energy Chem., 2021, 61, 61–70.
33. Y. Su, W. Yu, L. Liao, X. Xiong, H. Chen, L. Hu, T. Lei, J. Zhao, D. Chen and W. Mai, Adv. Funct. Mater., 2023, 33, 2303976.
34. L. Jia, P. Bogdanoff, A. Ramírez, U. Bloeck, D. Stellmach and S. Fiechter, Adv. Mater. Interfaces, 2015, 3, 1500434.
35. A. H. Asif, N. Rafique, R. A. K. Hirani, L. Shi, S. Zhang, S. Wang and H. Sun, Chem. Eng. J., 2023, 451, 138630.
36. W. Zhang, R. Guo, J. Sun, L. Dang, Z. Liu, Z. Lei and Q. Sun, J. Colloid Interface Sci., 2019, 553, 705–712.
37. C. Kim, J. Lim, S. Kim, Y. Park and W. Choi, Chem. Eng. J., 2022, 429, 132108.
38. X. Z. Fan, Q. Q. Pang, S. S. Yi, X. Du, S. Zhang, Z. Y. Liu and X. Z. Yue, Appl. Catal., B, 2021, 292, 120152.
39. X. Meng, Q. Zhan, Y. Wu, M. Zhu, K. Liu, N. Wang, K. Yin, Y. Sun, S. Dong and Y. Dai, Nano Res., 2022, 16, 4178–4187.
40. L. Li, G. Zhang, C. Zhou, F. Lv, Y. Tan, Y. Han, H. Luo, D. Wang, Y. Liu, C. Shang, L. Zeng, Q. Huang, R. Zeng, N. Ye, M. Luo and S. Guo, Nat. Commun., 2024, 15, 4974.
41. A. Li, S. Kong, K. Adachi, H. Ooka, K. Fushimi, Q. Jiang, H. Ofuchi, S. Hamamoto, M. Oura, K. Higashi, T. Kaneko, T. Uruga, N. Kawamura, D. Hashizume and R. Nakamura, Science, 2024, 384, 666–670.
42. X. Wang, Z. Li, H. Jang, C. Chen, S. Liu, L. Wang, M. G. Kim, J. Cho, Q. Qin and X. Liu, Adv. Energy Mater., 2025, 15, 2403388.
43. B. Deng, G. Yu, W. Zhao, Y. Long, C. Yang, P. Du, X. He, Z. Zhang, K. Huang, X. Li and H. Wu, Energy Environ. Sci., 2023, 16, 5210–5219.
44. Y. Chen, J. Mao, H. Zhou, L. Xing, S. Qiao, J. Yuan, B. Mei, Z. Wei, S. Zhao, Y. Tang and C. Liu, Adv. Funct. Mater., 2023, 34, 2311664.
45. W. Chen, L. Li and Z. Lin, Chem. Eng. J., 2024, 492, 152338.
46. S. Mondal, S. Sarkar, M. Riyaz, M. Kar, A. C. Fortuin, S. Vashishth, R. Das, M. Eswaramoorthy, D. Kramer and S. C. Peter, ACS Energy Lett., 2024, 9, 3440–3447.
47. Y. Q. Zhang, M. Liu, L. T. Zhang, N. Lu, X. Wang, Z. G. Li, X. H. Zhang, N. Li and X. H. Bu, Adv. Funct. Mater., 2025, 35, 2412406.
48. A. Etxebarria, M. L. Luna, A. Martini, U. Hejral, M. Ruscher, C. Zhan, A. Herzog, A. Jamshaid, D. Kordus, A. Bergmann, H. Kuhlenbeck and B. R. Cuenya, ACS Catal., 2024, 14, 14219–14232.
49. Z. Li, R. Wu, D. Duan, X. Liu, R. Li, J. Wang, H. Chen, S. W. Chen, Y. Wu, H. Wang, S. Jiang, X. Zhang and Z. Lu, Joule, 2024, 8, 2920–2937.
50. H. Yang, J. Dong, X. Gao, C. Ni, Z. Li, Y. Liu, J. Li, X. He, W. Tan, L. Feng and L. Tian, Small, 2025, 21, 2410399.
51. L. Tian, H. Chen, X. Lu, D. Liu, W. Cheng, Y. Liu, J. Li and Z. Li, J. Colloid Interface Sci., 2022, 628, 663–672.
52. H. Yang, H. Wang, H. Cheng, X. Xu, J. Li, X. He, L. Tian and Z. Li, Chem. – Eur. J., 2024, 30, e202402972.
53. R. A. Borse, Y. X. Tan, J. Lin, E. Zhou, Y. Hui, D. Yuan and Y. Wang, Angew. Chem., Int. Ed., 2024, 63, e202318136.
54. S. Xie, R. Liu, N. Liu, H. Xu, X. Chen, X. Wang and D. Jiang, Angew. Chem., Int. Ed., 2025, 64, e202416771.
55. W. Wang, B. Li, J. Shi, K. Zhu, Y. Zhang, X. Liu, C. Li, F. Hu, X. Xi and S. Kawi, Appl. Catal., B, 2025, 362, 124766.
56. S. Liu, V. D. Cabanes, D. Fan, L. Peng, Y. Fang, M. Antonietti and G. Maurin, ACS Catal., 2024, 14, 2562–2571.
57. S. Xu, S. Chen, Y. Li, Q. Gao, X. Luo, M. Li, L. Ren, P. Wang, L. Liu, J. Wang, X. Chen, Q. Chen and Y. Zhu, Small, 2024, 20, 2400344.
58. E. Zhao, J. Shan, P. Yin, W. Wang, K. Du, C. C. Yang, J. Guo, J. Mao, Z. Peng, C. H. Wang and T. Ling, ACS Catal., 2024, 14, 14711–14720.
59. Z. Li, X. Xu, X. Lu, C. He, J. Huang, W. Sun and L. Tian, J. Colloid Interface Sci., 2022, 615, 273–281.
60. X. Yin, X. Li, H. Liu, W. Gu, W. Zou, L. Zhu, Z. Fu and Y. Lu, Nano Energy, 2018, 49, 489–497.
61. J. Zhang, G. Zhang and J. Zhang, Appl. Surf. Sci., 2019, 475, 256–263.
62. Q. Xu, J. Han, F. Tian, X. Zhao, J. Rong, J. Zhang, P. She, J. S. Qin and H. Rao, J. Am. Chem. Soc., 2025, 147, 10587–10597.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d5nr00898k

‡ These authors contributed equally to this work.

---