English

• Developing low-cost, high-efficiency water splitting electrolyzers is essential for large-scale renewable energy storage, especially hydrogen fuel [1-3]. The oxygen evolution reaction (OER), a key half-reaction in water splitting, is slow and requires efficient electrocatalysts to overcome its kinetic barriers [4,5]. While noble metal catalysts like IrO₂ and RuO₂ show excellent OER activity, their high cost and limited availability hinder large-scale use. Therefore, replacing these metals with affordable, commercially available alternatives for anodic OER catalysis is crucial for sustainable energy generation.

  Metal-free catalysts, alongside earth-abundant metal oxides and (oxy)hydroxides [6-10], have emerged as promising alternatives for OER, offering lower costs, high conductivity, and moderate catalytic performance [11-19]. Carbon-based materials like doped graphitic carbon, graphene, and CNTs show catalytic activity comparable to or surpassing metal-based catalysts [20,21]. However, the identification of true metal-free catalysts remains contentious, as trace transition metals, often introduced adventitiously at undetectable concentrations, have been shown to significantly enhance catalytic performance [22-26]. While trace metals in the as-synthesized catalysts have typically been excluded from the analysis [12,14,27,28], their potential involvement during the electrocatalytic process has received little attention. This issue mirrors the challenges faced in metal-free oxygen reduction reaction (ORR) electrocatalysis, where trace metals have led to misleading conclusions about the active sites and materials [23,29].

  Hybridizing metal (oxyhydr)oxides with carbon-based materials can enhance catalytic performance, often outperforming individual components [30-41]. This synergy likely extends beyond surface area and charge transport improvements, but its exact nature remains poorly understood. For example, Dai et al. suggested that M–O–C bonds between NiFe LDH and oxidized CNTs enhance catalytic activity [31], while Bao demonstrated that transition metals could modify carbon’s electronic structure for direct OER catalysis [42]. Sun proposed that ketone and carboxylate groups on oxidized MWCNTs aid proton transfer during OER, though the evidence is limited [30].

  Herein, we demonstrate that anodically activated carbon cloth exhibits high OER efficiency, with catalytic activity primarily driven by trace metal oxyhydroxides FeNiO_xH_y rather than oxygen-containing groups, as previously suggested for metal-free OER catalysts [12,14,27]. Sequential anodic analysis reveals site-specific deposition of FeNiO_xH_y nanoparticles on the carbon substrate. Microstructural characterization using high-resolution transmission electron microscopy (HRTEM) and X-ray absorption spectroscopy (XAS) shows that FeNiO_xH_y resembles γ-FeOOH but with a low-coordination, disordered structure. We also identify a strong interfacial synergistic effect between FeNiO_xH_y and the carbon substrate. The correlation of FeNiO_xH_y turnover frequency (TOF) with various oxygen-containing groups reveals a synergistic effect between FeNiO_xH_y and C=O groups. These findings highlight the need to consider adventitious trace metal effects in evaluating metal-free OER catalysts and provide new insights into the synergistic interactions in carbon-based hybrid catalysts.

  Carbon cloth (CC) is commonly used as a substrate for electrocatalysts. During testing of bare CC, we observed exceptional catalytic activity for the oxygen evolution reaction (OER) after activation via cyclic voltammetry (CV) or chronopotentiometric measurements in alkaline electrolytes (NaOH or KOH) (Fig. 1 and Fig. S1a in Supporting information). The anodically activated carbon cloth (ACC) often outperformed catalysts loaded onto CC, sometimes overshadowing the target catalyst analysis. This phenomenon presents an interesting opportunity for investigation, either to mitigate interference or to provide new insights into catalyst and electrode design.

  Figure 1

  

  Figure 1.  Electrochemical characterization of anodically activated carbon cloths and identification of the true catalytic active species of ACC-8M. (a) Anodic activation of carbon cloths in NaOH electrolytes of varying concentrations (8, 4.2, and 1 mol/L). The current density was set to 1 mA/cm². Insets show SEM images of pristine carbon cloths and ACC-8M. (b) Linear sweep voltammetry (LSV) of anodically activated carbon cloths (ACC-8M, ACC-4.2M, and ACC-1M) in 1 mol/L KOH. Scan rate: 1 mV/s. (c) CV scans of ACC-8M before and after acid washing. The scan rate is 50 mV/s. (d) EDX spectra of ACC-8M. The inset shows the corresponding elemental mapping image.

  DownLoad: Full-Size Img PowerPoint

  For a fair comparison, we used a chronopotentiometric activation process with a moderate anodic current density of 1 mA/cm² and NaOH electrolyte at concentrations of 1, 4.2, and 8 mol/L. The activated CC samples are denoted ACC-1M, ACC-4.2M, and ACC-8M. Chronopotentiometric curves showed a significant potential drop in the first 2 h, stabilizing thereafter (Fig. 1a). OER catalytic performance, evaluated in 1 mol/L KOH, showed a substantial negative shift in polarization curves for ACCs (activated for 2 h) compared to non-activated CC, indicating significant OER catalytic activity (Fig. 1b). The catalytic activity of ACCs depended on the electrolyte concentration, with higher concentrations resulting in increased catalytic activity (ACC-8M > ACC-4.2M > ACC-1M). ACCs activated in 16 mol/L NaOH, nearing its saturated solubility, exhibited catalytic activity comparable to ACC-8M (Fig. S1b in Supporting information).

  To assess catalytic performance, we measured the overpotential at 10 mA/cm² (η_{@10 mA/cm²}) and the current density at 325 mV (j_{@325 mV}). The overpotentials at 10 mA/cm² were 338 mV for ACC-8M, 379 mV for ACC-4.2M, and 395 mV for ACC-1M (Fig. 1b). The current densities at 325 mV were 3.22, 0.57 and 0.45 mA/cm², respectively (Fig. 1b and Fig. S2a in Supporting information). Notably, ACC-8M outperformed most metal-free OER catalysts, showing comparable performance to some transition metal oxide catalysts (Table S1 in Supporting information). ACC-8M, ACC-4.2M, and ACC-1M exhibit Tafel slopes in the range of 32–43 mV/dec (Fig. S2b in Supporting information), which agree well with previously reported values. This indicates a rate-determining step involving either a chemical reaction following a two-electron transfer (30 mV/dec) or a proton-coupled electron transfer reaction after the first electron transfer (40 mV/dec).

  The durability of ACC-8M was evaluated with chronoamperometric measurements, showing stable current density at 7.5 mA/cm² for 24 h at an overpotential of 330 mV (Fig. S3a in Supporting information). The Faradaic efficiency remained consistent over a 5.5-h electrolysis (Fig. S3b in Supporting information), ruling out the electrochemical oxidation of carbon fibers to CO or CO₂. Scanning electron microscopy (SEM) images suggest that electrochemical carbon corrosion occurred during the anodic activation process, as evidenced by the roughened surface morphology after activation (insets in Fig. 1a).

  A similar anodic activation process has been reported for metal-free OER catalysts, including oxidized carbon materials [12,14,27]. Studies have suggested that epoxides adjacent to ketone groups are the true active sites, as higher catalytic activity correlates with an increased formation of ketone groups in situ [12,14,27]. This hypothesis appears to be supported by our data, where a strong linear correlation was observed between catalytic activity (j_{@325 mV}) and the number of ketone groups for ACC-8M activated over different periods (0.5, 1, 2, and 12 h, Figs. S3c and S4 in Supporting information). When ACC-8M was treated with HNO₃ or HCl, it lost almost all catalytic activity, ruling out oxygen-containing groups as the main contributors to reactivity (Fig. 1c). Given that oxygen-containing groups typically survive or even form under acid treatment [27], this acid-induced activity decay rules out oxygen-containing groups as the primary contributors to the catalytic reactivity of ACC-8M. This conclusion is further supported by the poor performance of chemically oxidized CCs, such as those treated with oxygen plasma, Hummers’ method, or HNO₃. Interestingly, all chemically oxidized carbon substrates can still be activated by the anodic process (Fig. S5 in Supporting information), suggesting that unknown active species, rather than oxygen-containing groups, are formed on ACC-8M. The acid-induced activity decay further indicates that the real active species are either soluble or degradable in acid.

  XPS detected no transition metals (Co, Ni, Fe, or Mn) (Fig. S6 in Supporting information), but CV scans revealed small redox peaks, suggesting trace metal oxyhydroxides on ACC-8M (Fig. 1c). These peaks disappeared after acid treatment, indicating that the trace metals, likely Fe and Ni, are responsible for the catalytic activity (Fig. 1c) [26]. ICP-MS confirmed trace amounts of Fe (0.12 ± 0.04 µg/cm²) and Ni (0.07 ± 0.02 µg/cm²) on ACC-8M, with negligible Co and Mn. STEM-EDS mapping confirmed FeNiO_xH_y on the surface, with a higher Fe peak attributed to residual Fe in CC fibers (Fig. 1d). ICP-MS analysis of non-activated CC showed negligible Fe residue after acid washing, with a mass loading of 0.037 µg/cm²—less than a quarter of the Fe loading observed on ACC-8M. This suggests that the Fe residue does not contribute to the catalytic activity and does not affect our analysis. No other OER-active elements (Ni, Co, or Mn) were detected in non-activated CC, supporting the idea that the metal oxyhydroxides on ACC-8M were anodically deposited from the electrolyte.

  Further experiments involved anodic activation of carbon cloths in Fe-free and Ni-free NaOH (8 mol/L) (Fig. S3d in Supporting information). The catalytic activities of ACC-8M(Fe-free) and ACC-8M(Ni-free) were both enhanced compared to that of non-activated CC (Fig. S3d). This confirms that the source of catalytic activity is trace metal oxyhydroxides anodically deposited onto the carbon cloths. However, the apparent activities of these variants were lower than that of ACC-8M (Fig. S3d), consistent with the previously reported synergistic effect between Ni and Fe.

  In our case, the concentration of trace metals in both the electrolytes and the carbon cloth (CC) is extremely low, posing a significant challenge to elucidating the anodic deposition process. To address this issue, we conducted sequential anodic activation of four pieces of CC in the same 8 mol/L NaOH electrolyte. This stepwise anodic deposition led to a gradual decrease in the concentration of trace metals, which allowed us to investigate the concentration dependence of the anodic electrodeposition process (Fig. 2). After each activation step, we measured the Fe concentration in the electrolytes, the catalytic performance, and the metal loading of the ACC-8M. As expected, the Fe concentration in the electrolyte decreased progressively with each round of anodic activation, dropping from 41 µg/L to 17 µg/L over four activation cycles (Fig. 2a). This reduction in trace metal concentration led to a higher potential during the initial stages of anodic deposition and a longer time required to stabilize the potential (Fig. 2b). These changes in potential and deposition time are consistent with the anodic deposition of trace metals from the electrolyte, further supporting the role of metal oxyhydroxides in the catalytic activity. The anodic deposition curves were fitted using the following function (Eq. 1):

  $ y=y_0+\operatorname{Aexp}(x / \tau) $

  (1)

  Figure 2

  

  Figure 2.  Dependence of anodic deposition on trace metal concentration in 8 mol/L NaOH. (a) Fe concentration in 8 mol/L NaOH (left y-axis) and the anodic deposition time constant (τ, right y-axis) during the sequential four-round anodic activation. Fe exhibits the most distinguishable ICP-MS signals (corresponding to the highest concentration) compared to other metals, and thus is used to represent the trace metal concentration. (b) Chronopotentiometric measurements during the sequential four-round anodic activation process. The curves are fitted with Eq. 1. The fitted data are shown as solid lines. (c) Metal loadings of ACC-8M during the sequential four-round anodic activation. (d) Polarization curves of ACC-8M after each round of the sequential anodic activation. Electrochemical measurements were conducted in 1 mol/L KOH.

  DownLoad: Full-Size Img PowerPoint

  where y₀ is a constant, A is the prefactor and τ is the anodic deposition time constant (Fig. 2b). As the Fe concentration in the electrolyte decreased, the anodic deposition constant increased, indicating a correlation between the trace metal concentration and the deposition rate (Figs. 2a and b).

  Despite variations in the time constant, all the samples eventually stabilized at a similar potential during the anodic activation process (Fig. 2b). As a result, the metal oxyhydroxide mass loading in each activation process was similar (0.07–0.16 µg/cm² for Fe and 0.05–0.09 µg/cm² for Ni, Fig. 2c), and the catalytic performance for OER was almost identical across the samples (Fig. 2d). This further confirmed that the true catalytic active species are trace metal oxyhydroxides, which are anodically deposited from the electrolytes. Moreover, the observation that the metal concentration in the electrolyte affects the anodic deposition rate but does not alter the total amount of trace metal oxyhydroxides or catalytic activity suggests that the deposition occurs on specific sites of the CC via a random collision process.

  Despite the low loading of trace metal oxyhydroxides, large-area TEM imaging revealed a few FeNiO_xH_y nanoparticles with low crystallinity, as indicated by distorted lattice fringes (Fig. 3a). These nanoparticles, sized between 2 nm and 5 nm, were observed to adhere to the surface of amorphous carbon fibers, consistent with the anodic deposition process. To investigate the atomic and electronic structures of FeNiO_xH_y nanoparticles, X-ray absorption spectroscopy (XAS) was employed, with reference spectra from typical oxides, hydroxides, and oxyhydroxides (Figs. 3b and c). The Fe and Ni K-edge X-ray absorption near-edge spectra (XANES) of the FeNiO_xH_y nanoparticles closely resemble those of their corresponding oxyhydroxides, namely γ-FeOOH and γ-NiOOH, suggesting a mixed oxyhydroxide composition (Figs. 3b and c). The average metal oxidation states, derived from the main absorption edge (using the first derivative), indicate that the FeNiO_xH_y in ACC-8M consists of Fe cations in a near +3 oxidation state and Ni cations in a + 2.95 oxidation state.

  Figure 3

  

  Figure 3.  Microstructural analysis of trace metal oxyhydroxides on ACC-8M. (a) HRTEM image showing trace metal oxyhydroxides nanoparticles, marked by dashed circles. (b) XANES Fe K-edge spectra, with reference samples of γ-FeOOH, FeO, Fe₃O₄, and Fe₂O₃ included. (c) XANES of Ni K-edge spectra, with reference samples of NiOOH, Ni(OH)₂, and NiO included. (d) EXAFS Fe K-edge R-space spectra.

  DownLoad: Full-Size Img PowerPoint

  Extended X-ray absorption fine structure (EXAFS) k³χ(R) spectra were analyzed to probe the local atomic coordination. Previous studies have shown that FeNiO_xH_y often adopts the crystal structure of metal oxyhydroxides. Both γ-NiOOH and γ-FeOOH were used as structural models for XAS fitting, where partial Ni cations are substituted by Fe cations in the former, and vice versa in the latter. For simplicity, we focused on the Fe K-edge EXAFS spectra, given the satisfactory data quality (Fig. 3d). The R-space spectra indicate that the coordination environment of FeNiO_xH_y closely resembles that of FeOOH, with the first peak at approximately 1.5 Å corresponding to the octahedral Fe-O shell, and the second peak at 2.6 Å corresponding to the Fe-Fe (Ni) path. We fitted the R-space spectra with a γ-FeOOH model, where Fe cations were partially substituted by Ni cations in the 6-coordinate octahedral (O_h) site (Fig. 3d and Table S2 in Supporting information). The coordination number (CN) of Fe-O in FeNiO_xH_y was smaller than that in γ-FeOOH, suggesting a lower coordination environment for the Fe ions. The total CN for Fe-Fe (Ni) was also smaller than in γ-FeOOH, which is consistent with the observed poor crystallinity. This finding was further supported by the significantly larger Debye-Waller (DW) factor, which reflects the average distance of atoms from their ideal positions within the crystal. Notably, the DW factor for the Fe-Ni path was even larger than for the Fe-O path, indicating a high degree of disorder in the Ni ions, which are significantly displaced from their ideal positions. Overall, the EXAFS analysis reveals that the atomic structure of FeNiO_xH_y on ACC-8M resembles that of γ-FeOOH but features a low-coordination environment and long-range disorder.

  FeNiO_xH_y nanoparticles often interact electronically with their substrates, leading to enhanced catalytic performances. This improvement is attributed to changes in the electronic structure, modulation of interfaces, or alterations in the local microenvironment. To explore the specific interactions in ACCs, we analyzed the relationship between the intrinsic catalytic activity of ACCs and the concentration of the electrolyte used for anodic activation. ICP-MS measurements revealed a slight increase in metal loading as electrolyte concentrations were elevated (Fig. S7a in Supporting information). However, this small increase in loading does not fully explain the significant improvement in OER catalytic activity of ACC-8M compared to ACC-4.2M and ACC-1M (Fig. 1b).

  To clarify this, we calculated the turnover frequency (TOF), assuming Fe as the active site for OER catalysis (denoted as TOF_Fe, Fig. S7b in Supporting information), as Fe is generally regarded as the active site for OER in similar systems [24,25]. It is important to note that the true active sites for our FeNiO_xH_y catalysts remain elusive and are beyond the scope of this work. Nevertheless, TOF_Fe provides a reliable estimate of the intrinsic catalytic activity, particularly for our catalysts, which are well-dispersed, small in size, exhibit low metal loading, and have a close molar ratio of Fe:Ni. At an overpotential of 300 mV, ACC-8M exhibited a TOF_Fe of 1.70 s^-1, which is 6–14 times higher than that of ACC-4.2M and ACC-1M. To eliminate the impact of composition variations between samples, we also calculated the total turnover frequency TOF_total, assuming all metal cations (Fe and Ni) contribute to OER catalysis (Fig. S7b). Again, ACC-8M showed a significantly higher TOF_total than both ACC-4.2M and ACC-1M. This large difference in activity suggests a specific interaction between the FeNiO_xH_y nanoparticles and the carbon substrates in ACC-8M.

  To investigate the synergistic effect, we analyzed how the catalytic activity of ACCs depends on anodic activation time (Fig. 4a). As expected, the activity increased with longer activation times across all NaOH concentrations, along with higher metal loading (Fig. S7c in Supporting information). At each activation time, activity was higher with increased electrolyte concentration: ACC-1M < ACC-4.2M < ACC-8M (Fig. 4a). However, metal loading alone does not fully explain the catalytic activity difference between ACC-8M and the others. To isolate mass loading contributions, we compared the intrinsic catalytic activity TOF_Fe at 325 mV (Fig. 4b). ACC-4.2M and ACC-1M had nearly identical TOF_Fe values, while ACC-8M had much higher TOF_Fe, with the highest value at 2.73 s^-1, over 13 times higher than ACC-1M and ACC-4.2M (~0.2 s^-1). ACC-8M also showed a TOF_Total of 1.22 s^-1, more than three times higher than NiFe LDH and NiFeO_xH_y [26,43,44], supporting the synergistic effect.

  Figure 4

  

  Figure 4.  Catalytic activity of ACCs anodically activated in NaOH with varying concentrations and the correlation between catalytic activity and oxygen-containing groups on ACC-8M. (a) j_{@300 mV} of ACC-1M, ACC-4.2M, and ACC-8M after anodic activation for different periods (0.5, 1, 2, and 12 h). (b) TOF_{@325 mV} of ACC-1M, ACC-4.2M, and ACC-8M after anodic activation for different periods (0.5, 1, 2, and 12 h). For clarity, logarithmic x-axes are used in (a) and (b). (c, d) Content of oxygen-containing groups (left y-axis) and TOF_{@325 mV} (right y-axis) after anodic activation for different periods (0.5, 1, 2, and 12 h). A logarithmic scale is used for the x-axis for clarity. Dependence of TOF_{@325 mV} on the content of oxygen-containing groups: (e) C—O, (f) C=O, and (g) O—C=O on ACC-8M. Carbonate content is shown in Fig. S11 (Supporting information). Data for ACC-4.2M (red circle) and ACC-1M (black triangle) are also included.

  DownLoad: Full-Size Img PowerPoint

  There is no clear correlation between TOF_Fe and surface area (double-layer capacitance, Fig. S8 in Supporting information) [45], suggesting the presence of a substrate effect [46-48]. Given the extended electrochemical oxidation of carbon substrates during the anodic activation process, we hypothesize that the FeNiO_xH_y nanoparticles interact with oxygen-containing groups progressively implanted on the carbon substrates, contributing to their enhanced catalytic activity. To investigate the interaction between FeNiO_xH_y and oxygen-containing groups, we quantified the amounts of four oxygen-containing groups (C=O, C—O, O—C=O, and carboxyl) throughout the activation process on ACC-8M (Figs. 4c and d). The contents of these groups were plotted against anodic activation times, and their variation trends were compared with the corresponding turnover frequency (TOF_Fe). Interestingly, only C=O groups showed a variation pattern similar to TOF_Fe, indicating a strong coupling effect between C=O groups and the nearby FeNiO_xH_y. To further illustrate this, we analyzed the correlation between TOF_Fe and the content of each oxygen-containing group. TOF_Fe showed a clear correlation with the amount of ketone groups (Fig. 4f), whereas no such relationship was found with the other three oxygen-containing groups (C—O, O—C=O, and carboxyl) (Figs. 4e and g, and Fig. S9 in Supporting information). This provides strong evidence for the interfacial synergistic effect between FeNiO_xH_y nanoparticles and C=O groups on ACC-8M.

  Moreover, we observed that TOF_Fe increased when the content of C=O groups exceeded 10.25% (Fig. 4f). This suggests that the higher concentration of C=O groups promotes more direct interactions between FeNiO_xH_y and the carbon substrate. For comparison, we included data from ACC-4.2M and ACC-1M in the correlation analysis (Figs. 4e-g and Fig. S9). While these samples showed a less pronounced correlation with C—O, O—C=O, and carboxyl groups, they still fit well with the correlation to C=O groups, further supporting the specific interaction between FeNiO_xH_y nanoparticles and C=O groups in ACC-8M. Notably, the C=O content in both ACC-4.2M and ACC-1M remained below 10.25%, suggesting that the maximum electrochemical implantation of C=O groups occurs around this threshold. Any further increase in C=O content requires catalysts such as FeNiO_xH_y, which promote the formation of additional C=O groups during the oxygen evolution reaction. These newly formed C=O groups are strongly coupled with FeNiO_xH_y nanoparticles, further enhancing the catalytic activity. Additionally, the role of highly concentrated electrolytes in this process remains crucial, though the underlying mechanism is not yet fully understood. Our analysis provides new insights into the strong correlation between current density and C=O content (Fig. S3c in Supporting information), challenging previous claims about the catalytic activity of oxygen-containing groups in so-called metal-free catalysts [12,14,27].

  A similar synergistic effect between transition metal oxyhydroxides and Au electrodes has been suggested but lacks quantitative evidence [46-48]. For metal (oxyhydr)oxide/carbon hybrids, while several studies report a strong synergistic effect, understanding of this interaction remains limited. Dai et al. detected a large carboxyl peak via XANES and attributed it to M−O−C (M = Ni, Fe) bonds on oxidized carbon nanotubes (CNTs). However, this peak could also result from carbonates between NiFe LDH layers. Carbon materials in metal (oxyhydr)oxide/carbon hybrids are often mildly oxidized, acquiring oxygen-containing groups that likely enhance catalytic activity, though this has received little attention [30-36,42]. Our proposed mechanism provides insights into the strong synergistic effect seen in carbon-based hybrid OER catalysts.

  Zhao reported metal-free OER catalysts based on mildly oxidized CNTs and suggested that epoxides adjacent to ketone groups are active sites [12]. Although residual metal in CNTs was ruled out, the potential role of trace metals introduced during electrochemical activation was overlooked. Electrochemical deposition of metal oxyhydroxides likely occurred in their study, leading to the formation of hybrid catalysts. This phenomenon may also apply to metal-free OER catalysts doped with elements like N, P, S, and B, especially when electrochemical activation is necessary for catalysis. Further investigation is needed to distinguish hybrid catalysts from truly metal-free ones and consider the possibility of trace metals encapsulated by carbon materials, which may be exposed under OER conditions [48].

  In conclusion, we highlight the significant impact of trace metals from electrolytes on the OER catalytic activity of anodically activated carbon-based materials, challenging previous claims about the role of oxygen-containing groups in metal-free catalysts. FeNiO_xH_y nanoparticles, with a low-coordination and disordered structure, are deposited onto carbon substrates as the true active sites for efficient OER catalysis. Carbon cloth activated in 8 mol/L NaOH shows comparable performance to metal-based OER catalysts, requiring only 338 mV to achieve 10 mA/cm². A strong synergistic effect between FeNiO_xH_y nanoparticles and carbon substrate is identified, where ketone groups on the surface enhance catalytic activity. FeNiO_xH_y nanoparticles may also promote C=O group formation, further boosting performance. This study underscores the need to distinguish true metal-free catalysts from hybrid catalysts and re-evaluate prior metal-free OER systems. Our findings offer valuable insights for optimizing electrochemical analysis and designing efficient carbon-based OER catalysts.

  Declaration of competing 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.

  CRediT authorship contribution statement

  Qianqing Xu: Writing – original draft, Formal analysis, Data curation, Conceptualization. Qu Jiang: Writing – review & editing, Conceptualization. Haoyue Zhang: Methodology. Fang Song: Writing – review & editing, Funding acquisition, Conceptualization.

  Acknowledgments

  This work was financially supported by the National Natural Science Foundation of China (No. 22479097), the Shanghai Science and Technology Committee (Nos. 23ZR1433000 and 22511100400), the National High-Level Talent Program for Young Scholars, the Start-up Fund (F. Song) from Shanghai Jiao Tong University. We also acknowledge the SJTU Instrument Analysis Centre for the measurements.

  Supplementary materials

  Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2025.111417.

  
  
• 1. [1]

     S.J. Davis, N.S. Lewis, M. Shaner, et al., Science 360 (2018) eaas9793. doi: 10.1126/science.aas9793

  2. [2]

     X. Yang, C.P. Nielsen, S. Song, M.B. McElroy, Nat. Energy 7 (2022) 955–965. doi: 10.1038/s41560-022-01114-6

  3. [3]

     J. Lei, Z. Wang, Y. Zhang, et al., Carbon Neutrality 3 (2024) 25. doi: 10.1007/s43979-024-00101-y

  4. [4]

     J. Song, C. Wei, Z.F. Huang, et al., Chem. Soc. Rev. 49 (2020) 2196–2214. doi: 10.1039/c9cs00607a

  5. [5]

     S.Z. Oener, A. Bergmann, B.R. Cuenya, Nat. Synth. 2 (2023) 817–827. doi: 10.1038/s44160-023-00376-6

  6. [6]

     Y. Jiao, Y. Zheng, M.T. Jaroniec, S.Z. Qiao, Chem. Soc. Rev. 44 (2015) 2060–2086. doi: 10.1039/C4CS00470A

  7. [7]

     C.G. Morales-Guio, L. Liardet, X. Hu, J. Am. Chem. Soc. 138 (2016) 8946–8957. doi: 10.1021/jacs.6b05196

  8. [8]

     N.T. Suen, S.F. Hung, Q. Quan, et al., Chem. Soc. Rev. 46 (2017) 337–365. doi: 10.1039/C6CS00328A

  9. [9]

     Y. Zheng, A. Serban, H. Zhang, et al., ACS Energy Lett. 8 (2023) 5018–5024. doi: 10.1021/acsenergylett.3c01866

  10. [10]

      Q. Zhang, Y. Hu, H. Wu, et al., ACS Nano 17 (2023) 1485–1494. doi: 10.1021/acsnano.2c10247

  11. [11]

      J.J. Duan, S. Chen, M. Jaroniec, S.Z. Qiao, ACS Catal. 5 (2015) 5207–5234. doi: 10.1021/acscatal.5b00991

  12. [12]

      X.Y. Lu, W.L. Yim, B.H.R. Suryanto, C. Zhao, J. Am. Chem. Soc. 137 (2015) 2901–2907. doi: 10.1021/ja509879r

  13. [13]

      X. Liu, L.M. Dai, Nat. Rev. Mater. 1 (2016) 16064. doi: 10.1038/natrevmats.2016.64

  14. [14]

      L.Q. Li, H.B. Yang, J.W. Miao, et al., ACS Energy Lett. 2 (2017) 294–300. doi: 10.1021/acsenergylett.6b00681

  15. [15]

      B. Wang, X. Cui, J. Huang, et al., Chin. Chem. Lett. 29 (2018) 1757–1767. doi: 10.1016/j.cclet.2018.11.021

  16. [16]

      F. Liu, R. Shi, Z. Wang, et al., Angew. Chem. Int. Ed. 58 (2019) 11791–11795. doi: 10.1002/anie.201906416

  17. [17]

      S. Sahani, K. Malika Tripathi, T.Il Lee, et al., Energy Convers. Manag. 252 (2022) 115133. doi: 10.1016/j.enconman.2021.115133

  18. [18]

      Z. Zhang, Y. Lei, W. Huang, Chin. Chem. Lett. 33 (2022) 3623–3631. doi: 10.1016/j.cclet.2021.11.041

  19. [19]

      X. Ren, H. Liu, J. Wang, J. Yu, Chin. Chem. Lett. 35 (2024) 109282. doi: 10.1016/j.cclet.2023.109282

  20. [20]

      L. Lin, Z. Yu, X. Wang, Angew. Chem. Int. Ed. 58 (2019) 6164–6175. doi: 10.1002/anie.201809897

  21. [21]

      H. Lin, J. Wu, F. Zhou, et al., J. Environ. Sci. 124 (2023) 570–590. doi: 10.1016/j.jes.2021.11.017

  22. [22]

      D.A. Corrigan, J. Electrochem. Soc. 134 (1987) 377. doi: 10.1149/1.2100463

  23. [23]

      A. Ambrosi, S.Y. Chee, B. Khezri, et al., Angew. Chem. Int. Ed. 51 (2012) 500–503. doi: 10.1002/anie.201106917

  24. [24]

      L. Trotochaud, S.L. Young, J.K. Ranney, S.W. Boettcher, J. Am. Chem. Soc. 136 (2014) 6744–6753. doi: 10.1021/ja502379c

  25. [25]

      J. Masa, A.Q. Zhao, W. Xia, et al., Electrochim. Acta 128 (2014) 271–278. doi: 10.1016/j.electacta.2013.11.026

  26. [26]

      I. Roger, M.D. Symes, J. Am. Chem. Soc. 137 (2015) 13980–13988. doi: 10.1021/jacs.5b08139

  27. [27]

      N.Y. Cheng, Q. Liu, J.Q. Tian, et al., Chem. Commun. 51 (2015) 1616–1619. doi: 10.1039/C4CC07120D

  28. [28]

      Y. Zhao, R. Nakamura, K. Kamiya, et al., Nat. Commun. 4 (2013) 2390. doi: 10.1038/ncomms3390

  29. [29]

      M. Gong, Y. Li, H. Wang, et al., J. Am. Chem. Soc. 135 (2013) 8452–8455. doi: 10.1021/ja4027715

  30. [30]

      H.L. Wang, H.J. Dai, Chem. Soc. Rev. 42 (2013) 3088. doi: 10.1039/c2cs35307e

  31. [31]

      X. Long, J.K. Li, S. Xiao, et al., Angew. Chem. Int. Ed. 53 (2014) 7584–7588. doi: 10.1002/anie.201402822

  32. [32]

      D. Tang, J. Liu, X.Y. Wu, et al., ACS Appl. Mater. Interfaces 6 (2014) 7918–7925. doi: 10.1021/am501256x

  33. [33]

      T.Y. Ma, S. Dai, M. Jaroniec, S.Z. Qiao, J. Am. Chem. Soc. 136 (2014) 13925–13931. doi: 10.1021/ja5082553

  34. [34]

      W. Ma, R. Ma, C. Wang, et al., ACS Nano 9 (2015) 1977–1984. doi: 10.1021/nn5069836

  35. [35]

      L. Wang, H. Chen, Q.T. Daniel, et al., Adv. Energy Mater. 6 (2016) 1600516. doi: 10.1002/aenm.201600516

  36. [36]

      X.J. Cui, P.J. Ren, D.H. Deng, et al., Energy Environ. Sci. 9 (2016) 123–129. doi: 10.1039/C5EE03316K

  37. [37]

      Q. Mou, X. Wang, Z. Xu, et al., Chin. Chem. Lett. 33 (2022) 562–566. doi: 10.1016/j.cclet.2021.08.028

  38. [38]

      L. Lv, B. Tang, Q. Ji, et al., Chin. Chem. Lett. 34 (2023) 107524. doi: 10.1016/j.cclet.2022.05.038

  39. [39]

      C. Wang, K. Zhang, B. Yang, Chin. Chem. Lett. 36 (2025) 110538. doi: 10.1016/j.cclet.2024.110538

  40. [40]

      J. Zhang, J. Liu, J. Ran, et al., Chin. Chem. Lett. 36 (2025) 110403. doi: 10.1016/j.cclet.2024.110403

  41. [41]

      L. Wang, Y. Yan, R. Li, et al., Chin. Chem. Lett. 35 (2024) 110011. doi: 10.1016/j.cclet.2024.110011

  42. [42]

      D. Friebel, M.W. Louie, M. Bajdich, et al., J. Am. Chem. Soc. 137 (2015) 1305–1313. doi: 10.1021/ja511559d

  43. [43]

      F. Song, X. Hu, Nat. Commun. 5 (2014) 4477–4485. doi: 10.1038/ncomms5477

  44. [44]

      A.S. Batchellor, S.W. Boettcher, ACS Catal. 5 (2015) 6680–6689. doi: 10.1021/acscatal.5b01551

  45. [45]

      B.S. Yeo, A.T. Bell, J. Am. Chem. Soc. 133 (2011) 5587–5593. doi: 10.1021/ja200559j

  46. [46]

      Y. Gorlin, C.J. Chung, J.D. Benck, et al., J. Am. Chem. Soc. 136 (2014) 4920–4926. doi: 10.1021/ja407581w

  47. [47]

      R. Frydendal, M. Busch, N.B. Halck, et al., ChemCatChem 7 (2015) 149–154. doi: 10.1002/cctc.201402756

  48. [48]

      Y. Yi, J. Tornow, E. Willinger, et al., ChemElectroChem 2 (2015) 1929–1937. doi: 10.1002/celc.201500268

• 

  Figure 1  Electrochemical characterization of anodically activated carbon cloths and identification of the true catalytic active species of ACC-8M. (a) Anodic activation of carbon cloths in NaOH electrolytes of varying concentrations (8, 4.2, and 1 mol/L). The current density was set to 1 mA/cm². Insets show SEM images of pristine carbon cloths and ACC-8M. (b) Linear sweep voltammetry (LSV) of anodically activated carbon cloths (ACC-8M, ACC-4.2M, and ACC-1M) in 1 mol/L KOH. Scan rate: 1 mV/s. (c) CV scans of ACC-8M before and after acid washing. The scan rate is 50 mV/s. (d) EDX spectra of ACC-8M. The inset shows the corresponding elemental mapping image.

  下载: 全尺寸图片 幻灯片
  

  Figure 2  Dependence of anodic deposition on trace metal concentration in 8 mol/L NaOH. (a) Fe concentration in 8 mol/L NaOH (left y-axis) and the anodic deposition time constant (τ, right y-axis) during the sequential four-round anodic activation. Fe exhibits the most distinguishable ICP-MS signals (corresponding to the highest concentration) compared to other metals, and thus is used to represent the trace metal concentration. (b) Chronopotentiometric measurements during the sequential four-round anodic activation process. The curves are fitted with Eq. 1. The fitted data are shown as solid lines. (c) Metal loadings of ACC-8M during the sequential four-round anodic activation. (d) Polarization curves of ACC-8M after each round of the sequential anodic activation. Electrochemical measurements were conducted in 1 mol/L KOH.

  下载: 全尺寸图片 幻灯片
  

  Figure 3  Microstructural analysis of trace metal oxyhydroxides on ACC-8M. (a) HRTEM image showing trace metal oxyhydroxides nanoparticles, marked by dashed circles. (b) XANES Fe K-edge spectra, with reference samples of γ-FeOOH, FeO, Fe₃O₄, and Fe₂O₃ included. (c) XANES of Ni K-edge spectra, with reference samples of NiOOH, Ni(OH)₂, and NiO included. (d) EXAFS Fe K-edge R-space spectra.

  下载: 全尺寸图片 幻灯片
  

  Figure 4  Catalytic activity of ACCs anodically activated in NaOH with varying concentrations and the correlation between catalytic activity and oxygen-containing groups on ACC-8M. (a) j_{@300 mV} of ACC-1M, ACC-4.2M, and ACC-8M after anodic activation for different periods (0.5, 1, 2, and 12 h). (b) TOF_{@325 mV} of ACC-1M, ACC-4.2M, and ACC-8M after anodic activation for different periods (0.5, 1, 2, and 12 h). For clarity, logarithmic x-axes are used in (a) and (b). (c, d) Content of oxygen-containing groups (left y-axis) and TOF_{@325 mV} (right y-axis) after anodic activation for different periods (0.5, 1, 2, and 12 h). A logarithmic scale is used for the x-axis for clarity. Dependence of TOF_{@325 mV} on the content of oxygen-containing groups: (e) C—O, (f) C=O, and (g) O—C=O on ACC-8M. Carbonate content is shown in Fig. S11 (Supporting information). Data for ACC-4.2M (red circle) and ACC-1M (black triangle) are also included.

  下载: 全尺寸图片 幻灯片
•