Activating lattice oxygen in NiFe-based (oxy)hydroxide for water electrolysis

Transition metal oxides or (oxy)hydroxides have been intensively investigated as promising electrocatalysts for energy and environmental applications. Oxygen in the lattice was reported recently to actively participate in surface reactions. Herein, we report a sacrificial template-directed approach to synthesize Mo-doped NiFe (oxy)hydroxide with modulated oxygen activity as an enhanced electrocatalyst towards oxygen evolution reaction (OER). The obtained MoNiFe (oxy)hydroxide displays a high mass activity of 1910 A/gmetal at the overpotential of 300 mV. The combination of density functional theory calculations and advanced spectroscopy techniques suggests that the Mo dopant upshifts the O 2p band and weakens the metal-oxygen bond of NiFe (oxy)hydroxide, facilitating oxygen vacancy formation and shifting the reaction pathway for OER. Our results provide critical insights into the role of lattice oxygen in determining the activity of (oxy)hydroxides and demonstrate tuning oxygen activity as a promising approach for constructing highly active electrocatalysts.

In summary, albeit the system itself and the reported performance are of high potential interest to the field, the conclusions drawn are not sufficiently supported by the presented experimental and computational results to justify publication in Nature Communication. The authors are encouraged to complete the missing investigations and resubmit.
results on the 18 O-labeled NiFe and MoNiFe (oxy)hydroxide showed the signals of m/z = 32, m/z = 34, and m/z = 36 (Figure A1), suggesting the presence of 16 O2, 16 O 18 O and 18 O2 in the gas production. These results imply that both NiFe and MoNiFe (oxy)hydroxide follow the LOM mechanism. The mass spectrometric cyclic voltammograms (MSCVs) which plot the real-time gas product contents as a function of applied potential can provide direct comparison about the participation of lattice oxygen in OER process. The 18 O-labeled MoNiFe (oxy)hydroxide is with noticeably higher contents of 18 O 16 O and 18 O2 in the reaction product than the 18 O-labeled NiFe (oxy)hydroxide ( Figure A2), implying that the lattice oxygen of MoNiFe (oxy)hydroxide participated more actively into the OER reaction than that of NiFe (oxy)hydroxide.  signals are relative to the 16 O2 signal. All data were taken from the first cycle.
In addition to the DEMS measurement, we carried out additional Raman spectroscopy measurement to confirm the lattice oxygen activation during OER. First, the NiFe and MoNiFe (oxy)hydroxide were completely activated in KOH solution with H2 16 O. Then, the (oxy)hydroxide catalysts were subjected to a constant potential of 1.65 V (vs. RHE) for 30 min in KOH solution with H2 18 O. Raman spectra of obtained samples with 18 O-labeling are shown in Figure A3. The Raman peaks of NiFe and MoNiFe (oxy)hydroxide with 18 O-labeling shifted towards lower wavenumber because of the impact oxygen mass on the vibration mode (Angewandte Chemie International Edition, 2020, 59(21): 8072-8077;Angewandte Chemie, 2019, 131(30): 10401-10405.).
The MoNiFe (oxy)hydroxide showed a more obvious Raman shift to lower wavenumber than NiFe (oxy)hydroxide, suggesting a more oxygen in the lattice got exchanged between the lattice oxygen and electrolytes during OER on the MoNiFe (oxy)hydroxide.
The DEMS and Raman spectra results on the 18 O-labeled samples consistently suggest that Mo doping in NiFe (oxy)hydroxide promoted the lattice oxygen to exchange with electrolyte during OER, which is in accord with the conclusion we got in our previous manuscript. The newly added 18 O-labeling experiments and related discussion were added to the revised manuscript.  Fig. S18), suggesting the presence of 16 O2, 16 O 18 O and 18 O2 in the gas production [31][32][33] . This result implies that both NiFe and MoNiFe (oxy)hydroxide follow the LOM mechanism 14,30 . The mass spectrometric cyclic voltammograms (MSCVs) which plot the real-time gas product contents as a function of applied potential can provide direct comparison about the participation of lattice oxygen in OER process. The 18 O-labeled MoNiFe (oxy)hydroxide is with noticeably higher contents of 16 O 18 O and 18 O2 in the reaction product than the 18 O-labeled NiFe (oxy)hydroxide (Fig. 3a-d and Supplementary Fig. S19-S20), implying the lattice oxygen of MoNiFe (oxy)hydroxide participated more actively into the OER reaction than that of NiFe (oxy)hydroxide.
In addition to the DEMS measurement, the Raman spectra were also used to confirm the participation of lattice oxygen in OER. The Raman peaks of NiFe and MoNiFe (oxy)hydroxide with 18 O-labeling shifted towards lower wavenumber (Fig. 3ef) because of the impact oxygen mass on the vibration mode 2,34   The reference recommended by the reviewer were cited in the revised manuscript.
The RDS of MoNiFe is the deprotonation of *OOH (Figure 3b), which cannot be substantiated by the different performance of MoNiFe in D2O and H2O solution ( Figure   3e). To scientifically determine the RDS, the kinetic isotope effect (KIE) of MoNiFe and NiFe should be calculated (Angew. Chem. Int. Ed. 2021, 60, 3095). In other hand, it may be more scientific that isotopic experiments are performed in dissolved KOD with D2O rather than KOH with D2O.

Response:
We thank the reviewer for the valuable suggestion. As suggested by the reviewer, we redid the isotopic experiment in dissolved NaOD with D2O, and calculated the kinetic isotope effect (KIE) of MoNiFe and NiFe (oxy)hydroxide.

The linear sweep voltammetry (LSV) curves for NiFe and MoNiFe
(oxy)hydroxide measured in 1 M NaOH (dissolved in H2O) and NaOD (dissolved in D2O) solution were shown in Figure A4a and b. To show the kinetic isotope effect (KIE) for NiFe and MoNiFe (oxy)hydroxide clearly, the ratio of current density obtained in NaOH and in NaOD at the given potential is plotted in Figure A4c. MoNiFe (oxy)hydroxide exhibited a noticeably larger KIE value in comparison to NiFe (oxy)hydroxide, suggesting a severe degradation of OER activity in NaOD. This result suggested that proton transfer had a greater impact on the OER process on MoNiFe (oxy)hydroxide than that on NiFe (oxy)hydroxide. The deuterium isotopic experiments performed in NaOH/NaOD with a different concentration of 0.5 M provided consistent results ( Figure A5). The large isotopic effect of MoNiFe (oxy)hydroxide suggests that the proton transfer is involved in the potential determining step (PDS). This conclusion is in accord with the DFT calculation results, which show that the PDS step of OER on MoNiFe (oxy)hydroxide is the deprotonation of *OOH.    Fig. S23). The large isotopic effect of MoNiFe (oxy)hydroxide suggests that the proton transfer is involved in the potential determining step (PDS). This conclusion is in accord with the DFT calculation results, which show that the PDS step of OER on MoNiFe (oxy)hydroxide is the deprotonation of *OOH (Fig. 4b)." The Figure   The reference recommended by the reviewer and the related reference were added to the revised manuscript.
Page 14, line 18 in manuscript, "[39] Bai, L., Lee, S., Hu, X. Spectroscopic and electrokinetic evidence for a bifunctional mechanism of the oxygen evolution reaction. Angew. Chem. Int. Ed. Engl. 60, 3095-3103 (2021). [40] Tse,E. C. M.,Hoang,T. T. H.,Varnell,J. A.,Gewirth,A. A. Observation of an inverse kinetic isotope effect in oxygen evolution electrochemistry. ACS Catal. 6, 5706-5714 (2016)." were cited in the revised manuscript as reference [39] and [40] in "To show the kinetic isotope effect (KIE) for NiFe and MoNiFe (oxy)hydroxide clearly, the ratio of current density obtained in NaOH and in NaOD at the given potential 39, 40 is plotted in Fig. 4f." (3) The AEM pathway of NiFe and MoNiFe oxyhydroxide should also be calculated to fully prove the favorable kinetics of LOM pathway (Nat. Commun. 2020, 11, 4066). It should be noted that the in-situ Raman spectra have confirmed that the lattice oxygen of NiFe LDH wasn't involved in OER (Angew. Chem. Int. Ed. 2020, 59, 8072). In other words, the AEM pathway of NiFe may be more thermodynamically favorable than LOM of NiFe. Therefore, it is improper to only evaluate the energy barriers by using LOM process. Besides, the optimized DFT models of OER process should be provided for confirming the rationality of schematic illustration of the proposed OER pathway ( Figure 3a).

Response:
We thank the reviewer for the valuable suggestion. As suggested by the reviewer, we added the optimized DFT models and the calculation details of OER process following AEM mechanism in the revised manuscript.
1) Optimized DFT model of OER using for both AEM and LOM process The slab model of NiFe (oxy)hydroxide used for both AEM and LOM reaction pathway is terminated by the (001) surface ( Figure A6). For the AEM pathway, the metal site should be exposed to the reactants. Thus, two vacuum spaces were inserted along (001) and (010). To eliminate the interaction between periodic slabs, the thickness of vacuum spaces in both models was more than 10 Å. In addition, part of hydrogen atoms was removed because of the oxidation atmosphere. To find the stable configuration of Mo doping, we have built three slab models with different Mo sites ( Figure A7). The relative stability of Mo replacement was determined by calculating the formation energy (∆ ), which was computed as:  Table A1 show good agreement with references (The Journal of Physical Chemistry B, 2004, 108(46): 17886-17892.; Journal of the American chemical Society, 2013, 135(36): 13521-13530.).  Society, 2013, 135(36): 13521-13530.).
For the AEM pathway in an alkaline electrolyte, the four-electron reactions are: *OOH + OH - * + O2 (g) + H2O (l) + e -, where "*" represents the adsorption sites, which are generally the exposed metal sites.
The configurations of AEM pathway are shown in Figure A8. The free energy changes of each step can be calculated as: where U is the potential with respect to the normal hydrogen electrode (NHE).
2) Determining reaction energy barriers in AEM pathway In the AEM pathway, the OER reaction involves four subsequent proton-electron transfer steps, including *OH species adsorption, *O radical formation, *OOH transformation, and O2 desorption ( Figure A10a). To identify the active site in AEM pathway, both Ni site and Fe site have been considered (Figure A10b-c). The Fe sites were found to be the active sites with lower barrier than that on Ni sites. As shown in Figure A10c, the deprotonation of *OH in AEM pathway serves as potential determining step (PDS) for both NiFe and MoNiFe (oxy)hydroxide, with a barrier of 1.05 eV and 0.76 eV, respectively. (3) Comparison of AEM and LOM reaction pathway In the LOM pathway, the (oxy)hydroxides first go through the deprotonation process to form oxyhydroxide (step 1) ( Figure A11a). The exposed lattice oxygen then receives OH-via nucleophilic attack to form *OOH (step 2). After the deprotonation of *OOH (step 3), gaseous O2 releases from the lattice, and an oxygen vacancy is generated on the surface (step 4). The resulting oxygen vacancy sites are refilled by OHand the surface is recovered (step 5). The calculated Gibbs energy diagrams of OER on NiFe and MoNiFe (oxy)hydroxide are displayed in Figure A11b. For the NiFe (oxy)hydroxide, the desorption of O2, which was accompanied by the formation of oxygen vacancy, was found to be the PDS with a high energy barrier of 0.75 eV. In contrast, the barrier of oxygen vacancy formation became much smaller after Mo  Fig. S21a) and LOM pathway (Fig.4a) of OER were considered. In the AEM pathway, the Fe sites were found to be the active sites with lower barriers than Ni sites (Supplementary Fig. S21b-c). The Figure A6 has been added in Supplementary Information as Supplementary Fig.   S36.
The Figure A7 has been added in Supplementary Information as Supplementary Fig.   S37.
The Figure A8 has been added in Supplementary Information as Supplementary Fig.   S38.
The Figure A9 has been added in Supplementary Information as Supplementary Fig.   S39.
The Figure A10 has been added in Supplementary Information as Supplementary Fig.   S21.
The For the AEM pathway in an alkaline electrolyte, the four-electron reactions are: where "*" represents the adsorption sites, which are generally the exposed metal sites.
The configurations of AEM pathway are shown in Supplementary Fig. S38. The free energy changes of each step can be calculated as: where U is the potential with respect to the normal hydrogen electrode (NHE).
The calculated overpotential (η) was then determined by: The LOM pathway includes five steps, which are: where "*" represents the vacancy sites. Ol denote the lattice oxygen atoms.
The configurations of LOM pathway are shown in Supplementary Fig. S39. The energy barriers of LOM pathway were calculated by: The overpotential of LOM is calculated by:

"
The reference recommended by the reviewer and the related reference were added to the revised manuscript.  Figure A12). In addition, it is noting that despite the density of states of metal 3dorbital, especially for Ni 3d-orbital, seems to upshift closed to Fermi level, leading to an increase in the anti-bonding states below the Fermi level ( Figure A13). Such an effect weakens the metal-oxygen bonds, which is consistent with the COHP calculations.  result, as anodic potential is applied, the electron removal from oxygen sites was strongly facilitated. (Fig. 4d) 11,14 ." To "Such an enlarged U value gives rise to the downshift of LHB (Fig. 5d). As a result, as anodic potential is applied, the electron removal from oxygen sites was strongly facilitated 11,14 . It is noting that the LHB center located beneath the O 2p band center.

Therefore, the downshift of LHB center and upshift of O 2p band center for MoNiFe
(oxy)hydroxide leads to smaller overlap of metal 3d-orbital and oxygen 2p-orbital, which results in the weaker metal-oxygen bond. In addition, the density of states of metal 3d-orbital, especially for Ni 3d-orbital, upshift closed to Fermi level, leading to an increase in the anti-bonding states below the Fermi level (Fig. 5c). Such effect weakens the metal-oxygen bonds, which is consistent with the COHP calculations ( Fig.   5b)." The Fig. 5 in manuscript was changed to:

Response:
We thank the reviewer for the valuable suggestion. The LHB was determined by the 3d-orbital distribution below EF in DOS diagrams, while the UHB was determined by the unoccupied 3d-orbitals distribution above EF. The centers of LHB and UHB were calculated by: where ε and n(ε) are the energy level and number of states at this energy level, respectively.
As shown in Figure A14, the specific positions of LHB and UHB were calculated to be -4.36 eV and 2.01 eV for NiFe (oxy)hydroxide, and -4.67 eV and 2.90 eV for MoNiFe (oxy)hydroxide, respectively. "2) Determining the LHB and UHB band center: The LHB was determined by the 3d-orbital distribution below EF in DOS diagrams, while the UHB was determined by the unoccupied 3d-orbitals distribution above EF.
The center of LHB and UHB were calculated by: and where ε and n(ε) are the energy level and number of states at this energy level, respectively." (6) The crystal structures of catalysts have been characterized by HRTEM and SAED.
However, no X-ray diffraction (XRD) patterns of catalysts were provided in this manuscript. The authors should do the XRD measurement for the further structure characterizations.

Response:
We thank the reviewer for the valuable suggestion. We indeed carried out X-ray diffraction (XRD) measurement on the MoNiFe (oxy)hydroxide samples which were loaded on carbon cloths, as shown in Figure A15. However, because of the low loading mass, we can only see the peaks of the carbon cloths substrate, and no noticeable signals of (oxy)hydroxide can be observed. Therefore, we relied on the TEM measurement to determine the crystal structure of our catalysts. To further confirm the form of MoNiFe (oxy)hydroxide, we added extra aberration-corrected high-angle annular dark-field scanning transmission electron microscope (HAADF-STEM) measurement and the results are shown in Figure A16. The bright points in the HAADF-STEM image represent the Mo atoms in MoNiFe (oxy)hydroxide due to the higher atomic mass of Mo than Ni and Fe atoms, which confirms the presence of Mo doping in MoNiFe (oxy)hydroxide. However, the specific doping site cannot be observed due to the low crystallinity of MoNiFe (oxy)hydroxide and its susceptibility to be damaged under electron beam irradiation. Figure A15. The X-ray diffraction pattern of bare carbon cloths and MoNiFe (oxy)hydroxide. (oxy)hydroxide using X-ray diffraction measurement (Supplementary Fig. S5). We relied on the transmission electron microscopy (TEM) measurement to confirm the formation of (oxy)hydroxide phase." The Figure A15 has been added in Supplementary  (7) With regard to XPS data, only Ni 2p spectra of catalysts were analyzed. The Fe 2p, Mo 3d and O 1s spectra should also be measured for the elucidation of chemical states.

Response:
We thank the reviewer for the valuable suggestion. As suggested by the reviewer, we added the Fe 2p, Mo 3d, and O 1s XPS spectra and corresponding discussion to the revised manuscript.
As shown in Figure A17, the Fe 2p XPS spectra consist of two peaks located at ~710.7 eV and ~723.7 eV, which can be attributed to the spin-orbital splitting of Fe 3+ (Energy & Environmental Science, 2020, 13(1): 86-95.). After Mo doping, the Fe 2p spectra shift to a higher energy level, suggesting the higher valence state of Fe in MoNiFe (oxy)hydroxide, which is consistent with the Fe L-edge XAS results ( Fig. 6c in manuscript).  As shown in the Figure A19a- Figure   A19c, the MoNiFe (oxy)hydroxide shows higher defective O content than NiFe (oxy)hydroxide, suggesting that the MoNiFe (oxy)hydroxide might have more unsaturated oxygen sites, which is consistent with the higher oxygen activity as revealed by DFT calculations. The Figure A19 has been added in Supplementary   (9) In this manuscript, the Mo dopant is responsible for the activity improvement.
Therefore, the performance differences among samples with various Fe contents may arise from the amount variation of Mo dopant. The authors should precisely determine the metal contents of different samples by ICP spectroscopy.

Response:
We thank the reviewer for the valuable suggestion. The metal contents of different samples were determined by ICP measurement. As shown in Figure A21,  Fig. S7)." The Figure A21 has been added in the Supplementary Information       The chemical composition of MoNiFe (oxy)hydroxide after CP measurement was identified by X-ray photoelectron spectroscopy (XPS) (Figure A26) Supplementary Fig. S14) Fig. S15). The spacing between two adjacent lattice planes was quantified to be 0.21 nm (Supplementary Fig. S15b), which is assigned to the (105) plane of oxyhydroxide. Such value is slightly larger than that of the pristine MoNiFe  (105) and (110) plane for Ni-based oxyhydroxide (PDF-#06-0075) (Supplementary Fig. S15c) Fig. S16) and TEM-EDS ( Supplementary   Fig. S15d) mapping, the distribution of Mo, Ni, Fe elements in MoNiFe is uniform after CP measurement, and are the same as the pristine one.
The chemical composition of MoNiFe (oxy)hydroxide after CP measurement was identified by X-ray photoelectron spectroscopy (XPS) (Supplementary Fig. S17 The authors focused on Mo-doped NiFe (oxy)hydroxides, which is a strong candidate of a catalyst for the oxygen evolution reaction (OER). In their study, the Mo-doped NiFe (oxy)hydroxides were synthesized using the sacrificial template-directed approach. It is suggested that Mo-doped NiFe (oxy)hydroxides can be followed the lattice oxygen mechanism (LOM), and the authors clarified the importance of oxygen activity in the material lattice structure as an electrocatalyst for OER. They applied the various analyses to reveal the role of oxygen in atomic level and electric states by using both experimental and theoretical methods. The ideas are interesting and the findings in this work can be applied for a guideline for future catalyst design; however, I cannot provide my final recommendation, because some key information is missing. Therefore, please address the issues below and I need to read a new version to make my final recommendation.

Response:
We thank the reviewer for the valuable comments and suggestions. All the concerns raised by the reviewer have been addressed in detail as follows. 2) On Fig. 4b, the meaning of "TM" is necessary in figure caption Response: We very much appreciate the reviewer's careful reading of our manuscript. "TM" in Fig. 4b is referred to transition metal. The caption of Fig. 4 in the manuscript has been revised as follows.    (Fig. 6b,c) (Fig. 5b)." The reference recommended by the reviewer and other related literatures have been added to the revised manuscript: 4) The authors should show the raw data of sXAS and the information of their background data, which is subtracted from the raw data, in Supporting Information. The background data is important for the reliability of the data, because the results in Fig.   5a-c have the clear difference only in peak intensity.

Response:
We thank the reviewer for the valuable suggestion. The raw data of sXAS spectra and the corresponding background were provided in Figure A28-    IrO2 samples, as shown in Figure A31a. The MoNiFe (oxy)hydroxide delivered an overpotential of 242 mV at the current density of 10 mA/cm 2 , which was much lower than RuO2 (277 mV) and IrO2 (363 mV). To reach a current density of 100 mA/cm 2 , the MoNiFe (oxy)hydroxide required only an overpotential of 290 mV, while RuO2 and IrO2 needed 385 mV and 466 mV, respectively. To assess the intrinsic activity of the catalysts, the mass activity was obtained by normalizing the CV curves by loading mass ( Figure A31b). The MoNiFe (oxy)hydroxide delivered a mass activity of 1910 A/g at the overpotential of 300 mV, which is much higher than that of 112 A/g and 5.56 A/g for RuO2 and IrO2, respectively ( Figure A31c).  Fig. S12, note 1)." Page 25, line 1 in Supplementary Information To assess the intrinsic activity of the catalysts, the mass activity was obtained by normalizing the CV curves by loading mass (Supplementary Fig. S12b). The MoNiFe (oxy)hydroxide delivered a mass activity of 1910 A/g at the overpotential of 300 mV, which is much higher than that of 112 A/g and 5.56 A/g for RuO2 and IrO2, respectively ( Supplementary Fig. S12c)." The manuscript presents combined experimental and computational results on Modoped NiFe oxy-hydroxide with modulated oxygen activity for alkaline oxygen evolution reaction (OER). The system and the findings are of potential interest to the field in general and to readers of Nature Commun. specifically, but the information provided in several key areas is insufficient to render sufficient support for the conclusions drawn. Therefore, the manuscript does not warrant publication in the present form, as discussed below.

Response:
We thank the reviewer for the valuable comments and suggestions. All the concerns raised by the reviewer have been addressed in detail as follows.
1. The paper fails to discuss, let alone cite, key papers on the specific topic of lattice experiments? These results should be provided to support the proposed mechanism and discussed against relevant literature.

Response:
We thank the reviewer for the valuable comments and suggestions. All the concerns raised by the reviewer have been addressed in detail as follows.

1) Adding of necessary references
The reference mentioned by the reviewer and other key papers related published recently on the topic of lattice oxygen have been added to the revised manuscript:     Fig. S18) (Fig. 3a-d and Supplementary Fig. S19-S20) What reference is used to calculate the formation energy of the oxygen vacancy (O2 gas or an H2O-reference), etc? This data is needed to assess and discuss the findings.

Response:
We thank the reviewer for the valuable suggestions. The details of the DFT calculations have been provided.
The slab models of NiFe (oxy)hydroxide used for both AEM and LOM reaction pathway were terminated by the (001) surface ( Figure A35). For the AEM pathway, the metal site should be exposed to the reactants. Thus, two vacuum spaces were inserted along (001) and (010). To eliminate the interaction between periodic slabs, the thickness of vacuum spaces in both models was more than 10 Å. In addition, part of hydrogen atoms was removed because of the oxidation atmosphere. To find the stable configuration of Mo doping, we have built three slab models with different Mo sites ( Figure A36). The relative stability of Mo replacement was determined by calculating the formation energy (∆ ), which was computed as: where 2 , , and are Gibbs free energies of O2, surface with oxygen vacancy, and the clean surface, respectively. Since DFT calculations are inaccurate at describing the oxygen molecules, the Gibbs free energy of O2 was calculated by: where 2 and 2 are Gibbs free energies of H2O and H2, respectively. The Gibbs free energy change of H2O  H2 + O2 is 4.92 eV.
The Figure A35 has been added in Supplementary Information as Supplementary Fig.   S36.
The Figure A36 has been added in Supplementary Information as Supplementary Fig.   S37. Supplementary Information, added: "1) Optimized DFT model of OER using for both AEM and LOM process:

Page 33, line 14 in
The slabs model of NiFe (oxy)hydroxide used in AEM and LOM pathway were terminated by the (001) surface (Supplementary Fig. S36). For the AEM pathway, the metal site should be exposed to the reactants. Thus, two vacuum spaces were inserted along (001) and (010). To eliminate the interaction between periodic slabs, the thickness of vacuum spaces in both models was more than 10 Å. In addition, part of hydrogen atoms was removed because of the oxidation atmosphere.
To find the stable configuration of Mo doping, we have built three slab models with different Mo sites (Supplementary Fig. S37) (Supplementary Fig. S37d)." Page 35, line 1 in Supplementary Information, added: "3) Oxygen vacancy formation energy calculation: The formation energy of oxygen vacancy ( ) was calculated with respect to the Gibbs free energy of O2 at 298.15 K and 1.0 bar.  This information and discussion are also missing.

Response:
We thank the reviewer for the insightful question.
First, we found that Mo doping strongly impacts the intrinsic OER activity of the NiFe (oxy)hydroxide. To rule out the impact of the morphology of catalysts on the OER activity, we assessed the intrinsic activity using mass activity, which was obtained by normalizing the CV curves by loading mass (Figure A37). All the Mo doping samples exhibited higher mass activity than the counterpart without Mo doping, suggesting that the Mo doping can effectively enhance the intrinsic OER activity of (oxy)hydroxide.
Particularly, the MoNiFe (oxy)hydroxide exhibited the highest mass activity among all samples, with a current density of 1910 A/g at the overpotential of 300 mV. Such ultrahigh mass activity of MoNiFe-27% (oxy)hydroxide is about 60 times higher than that of NiFe-27% (oxy)hydroxide ( Figure A38). We agree with the reviewer the structure of the catalyst can critically impact its performance, and we believe the ultrathin nature of MoNiFe (oxy)hydroxide is one reason for its high stability. Chen et al. (Advanced Materials, 2019, 31(41): 1903909.) reported that the slow diffusion of proton acceptors within interlayer in NiFe hydroxide could lead to a local acidic environment, which results in a local etching process. The authors believed that such an etching process degrades the performance of multilayer NiFe hydroxide. In this work, the MoNiFe (oxy)hydroxide we obtained was ultrathin with an atomic thickness of 0.8 nm (mono-layer, denoted as 1L) or 1.5 nm (doublelayer, denoted as 2L) as revealed by the atomic force microscopy (AFM) (Figure A39).
Such ultra-thin nature of our MoNiFe (oxy)hydroxide can effectively prevent such local etching, and therefore is beneficial for the catalyst to remain stable during operation in alkaline solution.  To reveal information about the Mo doping sites, we relied on DFT calculations to identify the possible site of Mo doping. We have built three slab models with different Mo sites (Figure A41). The relative stability of Mo replacement was determined by calculating the formation energy (∆ ), which was computed as:  Fig. S4) Fig. S37) (Supplementary Fig. S37d)." Page 9, line 16 in manuscript, added: "It is reported that the OER stability of (oxy)hydroxide is strongly dependent on its structural charcteristics 28,29 ." The reference recommended by the reviewer has been added to the revised manuscript: Page 9, line 16 in manuscript, "[29] Tripkovic, V., Hansen,H. A.,Vegge,T. From 3D to 2D Co and Ni oxyhydroxide catalysts: Elucidation of the active site and influence of doping on the oxygen evolution activity. ACS Catal. 7, 8558-8571 (2017)." was cited in the revised manuscript as reference [29] in "It is reported that the OER stability of (oxy)hydroxide is strongly dependent on its morphology structure 28, 29 ." 5. In summary, albeit the system itself and the reported performance are of high potential interest to the field, the conclusions drawn are not sufficiently supported by the presented experimental and computational results to justify publication in Nature Communication. The authors are encouraged to complete the missing investigations and resubmit.

Response:
We truly appreciate the reviewer for his/her valuable comments and suggestions, which enormously improved the quality and clarity of this manuscript. All of the comments and suggestions from the reviewer have been taken into account in the revised manuscript by adding additional experiments, including the 18 O isotopelabeling experiments, in-situ differential electrochemical mass spectrometry (DEMS) measurements, HRTEM, and so on, as well as related discussions. We do hope the reviewer find these changes satisfactory and are willing to further improve the manuscript if needed. and discussed.

Response:
We thank the reviewer for the valuable comments and suggestions. We  As suggested by the reviewer, we changed the experiment procedure to be similar as that in the references (Angew. Chem. Int. Ed. 2019, 58, 10295;Angew. Chem. Int. Ed. 2021, 60, 3095-3103). labelled MoNiFe (oxy)hydroxide within 1 min of treatment, which is much faster than that for the NiFe (oxy)hydroxide (20 min). This result suggests that, while both samples follow the LOM mechanism, the MoNiFe (oxy)hydroxide exhibited much higher rate of oxygen exchange between lattice oxygen and electrolyte. This result is in consistence with the DEMS results ( Fig. 3 in the manuscript). It is needed to note the reason we carried quasi in-situ Raman spectra measurement instead of continuously measuring Raman spectra during applying potential is due to the influence of oxygen bubbles generated during OER on the Raman measurement. In the revised manuscript we replace the previous results with the newly added quasi in-situ Raman measurement results.
Page 10, line 18 in manuscript, change from: "In addition to the DEMS measurement, the Raman spectra were also used to confirm the participation of lattice oxygen in OER.
The Raman peaks of NiFe and MoNiFe (oxy)hydroxide with 18 O-labeling shifted towards lower wavenumber (Fig. 3e-f) (Fig. 3e,f)." (2) In this revised manuscript, the deuterium isotopic experiments and DFT calculations indicate that the potential determining step (PDS) of MoNiFe (oxy)hydroxide is the deprotonation of *OOH. Since the deprotonation of *OOH refers to a concerted protonelectron transfer (CPET) process in this manuscript, the PDS of MoNiFe (oxy)hydroxide includes the CPET. While, the previous reports indicate that the strong pH-dependence of OER activities for LOM-based catalysts arises from the potentialdetermining chemical deprotonation step (Nat. Energy 2019, 4, 329;Nat. Commun. 2021, 12, 3992). The authors should explain it.

Response:
We thank the reviewer for the valuable comments. We totally agree with the reviewer that the pH-dependent activity on the RHE scale means that OER includes a non-concerted proton-electron transfer (nCPET) process. For the catalysts with the LOM mechanism, the OER reaction normally involves the nCPET process. In some literature (Nature Communications, 2020, 11(1): 2002Nature Communications 2021, 12, 3992), such as the ones pointed out by the reviewer, the proton and electron transfer process occurred in two separate reaction steps, in which the PDS is the step that involved proton transfer. For instance, Huang et al. reported that the PDS of OER on NaxMn3O7 is the chemical deprotonation step, while the electron transfer was accompanied by the O2 desorption step (Nature Communications 2021, 12, 3992).
Nevertheless, there are also many previous literature which reported that although the electron and proton transfer both occurred in the same step, these two processes proceed sequentially instead of simultaneously (Catalysis Today, 2016, 262: 2-10.;Advanced Materials, 2018, 30(32): 1802912.;EcoMat, 2020, 2(2): e12021.).   Figure A5b). We found that the reaction barrier for the LOM-1 pathway is significantly lower than that for the LOM-2 pathway, as shown in Figure A5c. This result suggests that the LOM-1 pathway is more favorable for our cases. Since we observed a strong pH dependence for the MoNiFe (oxy)hydroxide, we believe that although the proton and electron transfer both occurred in the PDS, they are actually transferred sequentially, i.e., the electron and proton transfer processes are decoupled as shown in Figure A6. DFT method is known to be problematic dealing with charged systems, and it is challenging to assign charge to an atom during calculations. Therefore, it is difficult to verify the sequence of proton transfer and electron transfer in our PDS step. Figure A6. Illustration of decoupled proton-electron transfer of the potential determining step in MoNiFe (oxy)hydroxide.
In the revised manuscript, we emphasised that although the proton and electron transfer both occurred during the PDS step, i.e., the deprotonation of *OOH, the pH dependence results suggest that they actually occurred sequentially, i.e. the electron and proton transfer are decoupled.
Page 14, line 10 in manuscript, changed from: "The higher ρ RHE for MoNiFe (oxy)hydroxide implied a stronger pH-dependent OER activity, which may be due to the higher degree of decoupled proton-electron transfer during the PDS step, i.e., the deprotonation of *OOH. (Supplementary Fig. S22) 8,9,38 ." To "The higher ρ RHE for MoNiFe (oxy)hydroxide implied a stronger pH-dependent OER activity, which may be due to the higher degree of decoupled proton-electron transfer during the PDS step, i.e., the deprotonation of *OOH. (Supplementary note 3,   Supplementary Fig. S22-23) 8,9,38 ." Page 30, line 5 in Supplementary Information, added: "Supplementary note 3 The pH-dependent activity on the RHE scale means that OER includes a nonconcerted proton-electron transfer (nCPET) process. For the catalysts with the LOM mechanism, the OER reaction normally involves the nCPET process. In some literature, the proton and electron transfer process occurred in two separate reaction steps, in which the PDS is the step that involved proton transfer 29,30 . For instance,Huang et al. 29 reported that the PDS of OER on NaxMn3O7 is the chemical deprotonation step, while the electron transfer was accompanied by the O2 desorption step. Nevertheless, there are also many previous literature which reported that although the electron and proton transfer both occurred in the same step, these two processes proceed sequentially instead of simultaneously [31][32][33] . In our work, we indeed considered two possible pathways (LOM-1 and LOM-2) in our DFT calculations. While both proton transfer and electron transfer occur on the deprotonation of *OOH step in the LOM-1 pathway (Supplementary Fig. S22a), the proton transfer occurs on the deprotonation of *OOH step and the electron transfer occurs on the O2 desorption step in the LOM-2 pathway (Supplementary Fig. S22b).
We found that the reaction barrier for the LOM-1 pathway is significantly lower than that for the LOM-2 pathway, as shown in Supplementary Fig. S22c. This result suggests that the LOM-1 pathway is more favorable for our cases.
Since we observed a strong PH dependence for the MoNiFe (oxy)hydroxide, we believe that although the proton and electron transfer both occurred in the PDS, they are actually transferred sequentially, i.e., the electron and proton transfer process are decoupled as shown in Supplementary Fig. S23. DFT method is known to be problematic dealing with charged systems, and it is challenging to assign charge to an atom during calculations. Therefore, it is difficult to verify the sequence of proton transfer and electron transfer in our PDS step." The Figure A5 has been added in Supplementary Information as Supplementary Fig.   S22.
The reference recommended by the reviewer and the related reference have been cited in the revised Supplementary  (3) The DOS and COHP calculations prove that metal-oxygen bonds of MoNiFe (oxy)hydroxide are weaker than those of NiFe (oxy)hydroxide. However, the soft Xray absorption spectroscopies, the XPS spectra and the in-situ Raman spectra ( Figure   6) indicate the higher valence state of Ni and Fe in MoNiFe (oxy)hydroxide, which may lead to the improvement of covalency (Energy Environ. Sci. 2021, 14, 4647-4671). In other words, the theoretical calculations aren't consistent with the experimental analysis. In fact, most of previous works suggests the significance of high covalency for triggering lattice oxygen activation. The authors should explain such contradictive conclusions.

Response:
We thank the reviewer for the valuable comments. The covalency of metaloxygen bond was reported to be determined by the overlap of oxygen 2p orbitals and metal 3d orbitals in DOS, which can be quantified by the distance between the centers of the metal d-band and oxygen p-band ( 3 − 2 ) (Nature Catalysis, 2020, 3 (7): 554-563.). In our work, we found that the MoNiFe (oxy)hydroxide showed a higher overlap between Ni 3d-orbital and O 2p-orbital. As shown in Figure A7, the specific positions of O 2p (Ni 3d)-band center were calculated to be -1.58 eV (-2.72 eV) and - -O 2p band is enhanced after Mo doping, such overlap occurred more on the antibonding states below Fermi level as highlighted in dash circles in Figure A7. As a consequence, a weaker Ni-O bond was observed in the MoNiFe (oxy)hydroxide, which can facilitate oxygen vacancy formation. Consistently, we observed a lower oxygen vacancy formation energy in the MoNiFe (oxy)hydroxide than that in NiFe (oxy)hydroxide ( Figure A8). We believe that the lower oxygen vacancy formation energy, which is related to a higher oxygen lattice activity is the reason for the low reaction barrier for the LOM pathway for MoNiFe (oxy)hydroxide.  We strongly agree with the reviewer that higher covalency of metal-oxygen bond is favorable to trigger lattice oxygen activation as reported in many previous literatures.
In fact, in addition to the covalency of metal-oxygen bond (represented by the relative energy alignment between metal 3d-and oxygen 2p-bands, 3 − 2 ), oxygen activity (represented by the absolute energy level of the O 2p-band, 2 ) also exhibited a profound influence on the OER mechanism. As shown in Figure A9 Figure A10) and demonstrated that oxygen 2p-band center of spinel oxides is required to be high enough to guarantee the lattice oxygen to escape from the lattice. The reason why catalyst with high oxygen 2p band position is likely to follow the LOM mechanism ( Figure A9 and A10 Figure A7) and lower oxygen vacancy formation energy ( Figure A8). The higher lattice oxygen activity of MoNiFe (oxy)hydroxide leads to a promoted LOM pathway. To "It is reported that O 2p-band center is required to be high enough to guarantee the lattice oxygen to escape from the lattice 43 (Fig. 5c). Such an effect weakens the metal-oxygen bonds, which is consistent with the COHP calculations (Fig.   5b)." To "In addition, the density of states of metal 3d-orbital, especially for Ni 3d-orbital,  Fig. 5c and resulted in a weaker Ni-O bond, which is consistent with the COHP calculations (Fig. 5b)." The reference recommended by the reviewer and the related reference have been cited in the revised manuscript:  Sci. 4, 3966-3970 (2011)." was cited in the revised manuscript as reference [42] in "As a consequence, oxygen with high O 2p band position exhibited facilitated oxygen vacancy formation process and thus promoted the LOM mechanism 42 ." (4) The O 1s XPS spectra reveal the evident presence of oxygen defects ( Figure A19).
Since the oxygen defects play an important role on the reaction mechanisms, the oxygen defects should be considered when performing DFT calculations.

Response:
We thank the reviewer for the valuable suggestion. As suggested by the reviewer, we performed additional DFT calculations for OER reaction on NiFe and MoNiFe (oxy)hydroxide with oxygen defects.
The Gibbs free energy diagrams of OER in the AEM pathway on NiFe and MoNiFe (oxy)hydroxide with oxygen vacancy are shown in Figure A11. The corresponding configurations of reaction intermediate are shown in Figure A12. We found that the Fe sites serve as active sites in the presence of oxygen vacancy for both NiFe and MoNiFe (oxy)hydroxide, which is similar to the case without oxygen vacancy.
As shown in Figure A11c, the deprotonation of *OH in the AEM pathway serves as PDS for both NiFe and MoNiFe (oxy)hydroxide, with a barrier of 0.87 eV and 0.90 eV, respectively. The Gibbs free energy diagrams of OER in the LOM pathway on NiFe and MoNiFe with oxygen vacancy are shown in Figure A13. The corresponding configurations of reaction intermediate are shown in Figure A14. In the LOM pathway, the formation of gaseous O2 and the deprotonation of *OOH act as PDSs for NiFe and MoNiFe (oxy)hydroxide, with a barrier of 0.75 eV and 0.42 eV, respectively ( Figure   A13b), which is the same as the case without oxygen vacancy. These DFT results show that, after introducing oxygen vacancy on the surface, the reaction barrier of the LOM pathway is still lower than that in the AEM pathway. Therefore, the LOM pathway is still dominant for both NiFe and MoNiFe (oxy)hydroxide when surface defects were considered, and the MoNiFe (oxy)hydroxide exhibited a lower reaction barrier.   NiFe and MoNiFe (oxy)hydroxide, which is similar to the case without oxygen vacancy.
As shown in Supplementary Fig. S27c Supplementary Fig. S29b), which is the same as the case without oxygen vacancy. These DFT results show that, after introducing oxygen vacancy on the surface, the reaction barrier in the LOM pathway is still lower than that in the AEM pathway. Therefore, the LOM pathway is still dominant for both NiFe and MoNiFe (oxy)hydroxide when surface defects were considered, and the MoNiFe (oxy)hydroxide exhibited a lower reaction barrier." The Figure A11 has been added in Supplementary Information as Supplementary Fig.   S27.
The Figure A12 has been added in Supplementary Information as Supplementary Fig.   S28.
The Figure A13 has been added in Supplementary Information as Supplementary Fig.   S29.
The Figure A14 has been added in Supplementary Information as Supplementary Fig.   S30.
- I think the authors clearly answered the questions that I pointed out.
The authors show an inventive system and high catalytic performance in the present manuscript, and it will receive interest from the community of the field. The mechanism of the high catalytic performance was explained rationally, both from experimental analysis and theoretical calculations.
I think it is suitable to publish to Nature Communications as is.

Response:
We thank the reviewer for acknowledging the acceptance of our work. The authors have done an impressive job in addressing the comments and concerns outlined by the reviewers in the extensively revised manuscript. The manuscript has improved significantly and now warrants publication in Nature Communication.

Response:
We thank the reviewer for acknowledging the acceptance of our work.
Before publication, the authors should consider the effect of correcting the known systematic errors in the use of the PBE exchange correlation functional for OER, as outlined in Christensen, et al. 10.1021/acs.jpcc.6b09141 (Table A1, etc.).

Response:
We thank the reviewer for the valuable suggestion. We agree with the reviewer that systematic errors exist in the PBE functional. As mentioned in the reference (The Journal of Physical Chemistry C, 2016, 120 (43) ). Therefore, we corrected the systematic error in both LOM and AEM mechanisms considered in this work. Table A1, the systematic errors caused by PBE functional will not affect our conclusions in this work. Specifically, the correction of PBE functional will lower the total energy of OO* and OOH* by 0.2 eV. Therefore, for those sites where the peroxide species are not involved in the PDS steps, the overpotential value remains unchanged. On the other hand, the general trends still hold if peroxide species are involved in the PDS. For example, the overpotential of OER on the surface of NiFe (oxy)hydroxide increases by 0.2 eV after the correction. In contrast, the overpotential on MoNiFe surfaces remains the same because the energies of OO* and OOH* are shifted with the same magnitude.