Restructuring highly electron-deficient metal-metal oxides for boosting stability in acidic oxygen evolution reaction

The poor catalyst stability in acidic oxidation evolution reaction (OER) has been a long-time issue. Herein, we introduce electron-deficient metal on semiconducting metal oxides-consisting of Ir (Rh, Au, Ru)-MoO3 embedded by graphitic carbon layers (IMO) using an electrospinning method. We systematically investigate IMO’s structure, electron transfer behaviors, and OER catalytic performance by combining experimental and theoretical studies. Remarkably, IMO with an electron-deficient metal surface (Irx+; x > 4) exhibit a low overpotential of only ~156 mV at 10 mA cm−2 and excellent durability in acidic media due to the high oxidation state of metal on MoO3. Furthermore, the proton dissociation pathway is suggested via surface oxygen serving as proton acceptors. This study suggests high stability with high catalytic performance in these materials by creating electron-deficient surfaces and provides a general, unique strategy for guiding the design of other metal-semiconductor nanocatalysts.

In this manuscript X. Liu et al. fabricated and characterized an Ir based IMO electrocatalyst for the OER. They found that these catalysts present higher OER activity at low over-potential as well as higher activities under reaction conditions. The manuscript is well written, and the results are well supported. However there are some issues that the authors should consider in order to improve the quality of this manuscript: 1)Can the author include some reference for a better comparison with the acquired spectra in this research (i.e. Ir0 and IrO2 (IV))? Ideally one should consider to collect both XPS and NEXAFS. 3)The EELS spectra are of bad quality. The authors should consider to plot another EELS spectra with better quality or remove these spectra from the text. The information provided by the X-ray spectroscopy should be enough.
4)The XPS spectra fitting is not optimal. It is necessary to re-fit the whole set of spectra, indicate the parameters used (peak position, peak shape and fwhm), and justify why these peaks where used to fit the spectra. In this point the reference spectra can help significantly. In addition it will be nice to show the C1s and VB. Furthermore, it should be indicated how the spectra were calibrated. Figure 5e should be plotted without the XPS spectra for clarity. 6)I cannot see Figure S14 in the SI.

5)
Reviewer #3 (Remarks to the Author): The paper of Liu et al. presents synthesis and characterization of metal/MoO3 catalysts. The materials were prepared using electrospinning synthetic method and analyzed by microscopic and spectroscopic methods, while their oxygen evolution activity and durability were studied in an RDE set-up. The DFT calculations on different IR sites were also provided in this work. However, this paper has to be reviewed before publishing, specifically the following points has to be addressed: 1)energy calibration in XPS and XANES analysis was not mentioned in the experimental section, while it is very important while discussing the spectra shifts. Are the observed shifts in XPS related to charging (MoO3 as a semiconductor may lead to this)? In case of XANES, the Ir edge (metal)is at 11.21520 keV, that does not correspond to the metallic Ir provided in the manuscript 2) Analysis of XPS spectra. The comparison of Ir and Mo core-levels cannot be done when the fitting parameters are not consistent (the intensity ratio are not fixed, the relative binding energy etc). Please correct this.
3) The authors claim the formation of metallic Ir in case of IMO sample where there is no spectroscopic evidence, it looks like it is an oxide, however the oxidation state is lower than IV. Moreover, the analysis of Ir4f XPS spectra shows higher binding energy in comparison to IrO2, that is related to Ir (III) oxidation state as it was discussed in many publications previously (exception to a common situation higher BE = higher oxidation state). 4) The analysis of the electrochemical data shows higher activity of the synthesized materials, specifically for IMO. The Tafel slope of 48 mV dec-1 in case of IMO in comparison to other materials with 70-80 mV dec-1 should correspond to a change in the rds, however nothing is discussed in the manuscript. 5) The durability analysis of IOMO (Figure 5e) shows an interesting change at 20h, what is happening to the sample ? Was it analyzed ? This should be discussed in the manuscript and compared with IMO behavior. 6) Based on the DFT data, the authors proposed the OER mechanism on the synthesized materials, however it is not well discussed and not clear to a reader -has to be improved.

Manuscript ID: NCOMMS-21-23624-T
Title: Restructuring highly electron-deficient metal-metal oxides for boosting stability in acidic oxygen evolution reaction We are grateful to the editor, editorial staff, and reviewers for their critical comments and valuable suggestions. The manuscript has been strictly modified and improved after addressing all the suggestions as listed below: (The explanations to the comments from reviewers are shown in blue color with yellow highlight).

Reviewer #1
Reviewer #1: The manuscript entitled "Restructuring highly electron-deficient metal-metal oxides for boosting stability in acidic oxygen evolution reaction" has reported highly electron-deficient metal on semiconducting metal oxides as a high-performance catalyst for OER catalysis. The conception that highly electron-deficient surface of metal can contribute to the outstanding OER performance is novel and interesting. Meanwhile, the whole paper is well written and the characterizations can basically support the conclusions, thus, I could recommend it to be published on Nature Communications. However, some major problems still need to be addressed.
Response: We are grateful for the time and effort. Reviewer 1 has reviewed our manuscript, which are constructive and valuable to improve our manuscript.
(1-1-1) Why the valance state of surface atom of Ir NPs can exceed that of Ir atom in IrO2 whose Ir atoms are even surrounded by O atoms? In addition, if a highly electron-deficient metal can support excellent OER performance, why not select the IrO3 that has Ir atoms bearing 6+? Answer 1-1-1) Yes, thanks for the constructive comment. As we mention in our manuscript, we are sure that the valance state of surface atoms of Ir NPs on IMO nanocomposite can exceed the Ir atoms in IrO2 by two factors: (i) surface oxygen of Ir; (ii) the electron-withdrawing material of MoO3. The detailed explanation with simple model (Figrue R1) is as below: Figure R1. The simple demonstrating two factors for making the highly electron-deficient surface of Ir in IMO. Note that the Ir binding with the surface oxygen has the high valence state, evidenced from the argon-etching XPS spectra (detailed information was shown below).   In addition, revealing the Ir atom bound to the oxygen species in IMO shows a high valance state. To prove our arguments, we added new XPS data after argon (Ar) etching as below. To uncover which part of Ir NPs in IMO shows the valance state higher than 4 + , the high-resolution XPS measurement was conducted before and after Ar etching experiment (Energy Environ. Sci., 2020, 13, 5152-5164). As we expected, the Ir metallic property of Ir NPs in IMO was revealed after argon etching ( Figure S11), which is consistent with the commercial Ir metal. The oxygen adhering to the Ir surface was removed by the Ar etching, and finally Ir metal properties were obtained. This is consistent with XRD pattern and EXAFS spectra. We newly added  In addition, we newly added a sentence in our revised manuscript as "To prove whether the interfacial Ir atoms binding to MoO3 or the surface Ir atoms binding to adsorbed oxygen species have higher valence states, the HR-XPS measurement about Ir 4f in IMO is performed with an argon ion etching treated sample 41 , demonstrating zero-valence state of Ir metal in IMO, which is consistent with commercial Ir metal ( Figure S11) and XRD pattern (Figure 2a). This result confirms that the surface Ir in IMO has a high valence state." (1-1-2) "In addition, if a highly electron-deficient metal can support excellent OER performance, why not select the IrO3 that has Ir atoms bearing 6+?" Answer 1-1-2) Yes, thanks for the constructive comment. We believe that a highly electrondeficient metal can support excellent OER performance, based on our experimental result and previous literature (ACS Energy Lett. 2021, 6, 4, 1588-1595;). However, we did not select IrO3, based on the below two reasons:  Specifically, IrO3 may corrode into electrolyte as IrO4ion, as shown in Figure R3. (1-2) Which atoms (or part) of Ir NPs in IMO show valance state higher than 4+, the interfacial Ir atoms binding to Mo? Or the Ir atoms binding to adsorbed oxygen species? and why? This issue is critical for active sites in OER and need to be explained in the work.
Answer 1-2) Thanks for the constructive comment. The part of Ir atoms binding to adsorbed oxygen species in IMO shows a high valance state. To prove our arguments, we added new XPS data after argon (Ar) etching as below. To uncover which part of Ir NPs in IMO shows the valance state higher than 4 + , the high-resolution XPS measurement was conducted before and after Ar etching experiment (Energy Environ. Sci., 2020, 13, 5152-5164). As we expected, the Ir metallic property of Ir NPs in IMO was revealed after argon etching ( Figure S11), which is consistent with the commercial Ir metal. The oxygen attached to the Ir surface was removed by the Ar etching and finally gave Ir metallic property, which is consistent with XRD and EXAFS spectra.
We newly added Fig. S11. In addition, we newly added a sentence in our revised manuscript as "To prove whether the interfacial Ir atoms binding to MoO3 or the surface Ir atoms binding to adsorbed oxygen species have higher valence states, the HR-XPS measurement about Ir 4f in IMO is performed with an argon ion etching treated sample 41 , demonstrating zero-valence state of Ir metal in IMO, which is consistent with commercial Ir metal ( Figure S11) and XRD pattern ( Figure 2a). This result confirms that the surface Ir in IMO has a high valence state."  We newly added one sentence in our revised manuscript as below: Note that Ir L3-edge of IMO is also positively shifted compared with that of Ir metal foil due to the high electron-deficient surface of IMO ( Figure 4a).
(1-4) For practical usage, a durability test under higher current density (e. g. >50 mA/cm2) is recommended to be provided for comparation in the work.
Answer 1-4) Thanks for the constructive comment. By following the review's comments, we newly conducted the durability test under a high current density of 100 mA/cm 2 for IMO. As expected, IMO showed good durability in the high current, which is promising for practical usage. We newly added Fig. S18 in the revised manuscript as below: Further, the chronopotentiometry test at a high current density of 100 mA cm −2 for 48 h demonstrates the exceptional stability of IMO, which is promising for practical usage ( Figure S18).
(1-5) In the part of choosing active sites for OER, the OH* should be a more proper adsorption species for testing rather than O2, because the OH* is the first intermediate formed in OER.
Answer 1-5) Thanks for the constructive comment. We totally agree with the review's comments.
We newly modified by choosing active sites using the HO* rather than O2 in revised Figures 6a and 6b as below. Answer 1-6) Thanks for the constructive suggestion. We totally agree with the reviewer's comment to cite the valuable literatures. By following the comments, we cited those articles in three parts, as shown below.
(1) Hydrogen (H2) fuel, as a clean energy carrier, is promising to provide an environmentally benign solution for global energy needs (Advanced Functional Materials, 2020, 1908708;Nature Communications, 2021, 12, 2351. (2) Xin Wang and coworkers proposed a lattice oxygen oxidation mechanism pathway using metal oxyhydroxides when two adjacent oxidized oxygen atoms can hybridize their oxygen holes without sacrificing metal-oxygen hybridization (Nature Energy, 2019, 4, 329).

Reviewer #2
Reviewer #2: In this manuscript X. Liu et al. fabricated and characterized an Ir based IMO electrocatalyst for the OER. They found that these catalysts present higher OER activity at low over-potential as well as higher activities under reaction conditions. The manuscript is well written, and the results are well supported. However, there are some issues that the authors should consider in order to improve the quality of this manuscript: Response: We are grateful for the time and effort. Reviewer 2 has spent reviewing our manuscript.
The review comments are constructive and valuable to improve our manuscript. As we expected, the Ir L3-edge was also positively shifted compared with the Ir metal, which is consistent with the results of Mo K-edge in IMO. We newly added one sentence in revised our manuscript as "Note that Ir L3-edge of IMO is also positively shifted compared with that of Ir metal foil due to high electron-deficient surface of IMO (Figure 4a)" For the XPS, we newly added the references of Ir foil and IrO2 as shown below: Figure S10. HR-XPS for the Commercial Ir, Commercial IrO2, and IMO. As a result, the IMO still has a high energy shift in comparison with commercial IrO2, which is consistent with the results of IOMO. We newly added one sentence in our revised manuscript as "We confirm that the Ir NPs of the IMO has an electron-deficient surface, as evidenced by a higher Ir surface valence state of IMO than that of IOMO and commercial IrO2 (Figures 3a and S10)." Answer 2-2) Yes, we thank the reviewer for raising the issue. We totally agree with the review's comments. By following reviews comments, we swapped images from the Fig. S5 of SI to Fig.2 of the main text and removed Fig S5 from the SI. Thanks a lot for your constructive comments.
We newly revised in the revised manuscript.   S0968-4328(97)00033-4). Thus, it is a challenge to obtain the highquality signal of Ir in this current stage. So, by following the review's comments, we removed the EELS spectra in the revised manuscript. We strongly believe that the XPS and XAFS spectra are enough to support our statements. Thanks again for your constructive comments.
(2-4) The XPS spectra fitting is not optimal. It is necessary to re-fit the whole set of spectra, indicate the parameters used (peak position, peak shape and fwhm), and justify why these peaks where used to fit the spectra. In this point the reference spectra can help significantly. In addition it will be nice to show the C1s and VB. Furthermore, it should be indicated how the spectra were calibrated.
Answer 2-4) Yes, we thank the reviewer for raising the issue.
Firstly, we re-fitted the whole set of spectra of XPS. To clearly demonstrate the peak position, peak shape, and FWHM, we newly made the table S1, as shown below. To justify these peaks that we used for fitting, we newly added the reference spectra of commercial Ir metal and IrO2. Additionally, the intensity ratio was fixed using standard ratios of 3d3/2: 3d5/2 = 2 : 3 and 4f5/2 : 4f7/2 = 3 : 4, which can be helpful to justify the peaks used to fit in those spectra. For energy calibration, all XPS spectra were calibrated using C 1s at 284.6 eV, which also was added in Fig. S7 in the experimental section of the revised manuscript. In fact, all spectra of XPS were measured in the Cooperative Center for Research Facilities (CCRF) of SKKU and the expert operator always has used C 1s as the standard process for spectra calibrations, which is a general process for the technology of XPS calibration. We strongly believe that the C 1s is valid for the energy calibration, demonstrating the C 1s position for all spectra in Figure S7, which indicated that the energy of spectra was properly corrected. Figure S7. The survey spectrum with energy calibration using C 1s at 284.6 eV. a, Commercial Ir.
(2-5) Figure 5e should be plotted without the XPS spectra for clarity.
Answer 5) Yes, Thanks for the constructive comment. We agree with the review's comments and we deleted the XPS spectra in Fig. 5 in revised manuscript and also the deleted XPS spectra of Fig. 5 was moved as Figure S19 in the revised Supporting information.  (2-6) I cannot see Figure S14 in the SI.
Answer 2-6) Yes, we thank the reviewer for raising the issue. We are so sorry for our mistake.
When the pdf file was converted from the word file, Fig. S14 was not generated. We carefully checked all figures properly at this time when we converted the word file to a pdf file. In this new version of our manuscript, Figure S14 was re-named as Figure S19, as shown below: catalysts. The materials were prepared using electrospinning synthetic method and analyzed by microscopic and spectroscopic methods, while their oxygen evolution activity and durability were studied in an RDE set-up. The DFT calculations on different IR sites were also provided in this work.
However, this paper has to be reviewed before publishing, specifically the following points has to be addressed: Response: We are grateful for the time and effort. Reviewer 3 has spent reviewing our manuscript and agree published after addressing the comments. The review comments are constructive and valuable to improve our manuscript. For energy calibration of XPS, all XPS spectra were calibrated using C 1s at 284.6 eV, which also was added in the experimental section of the revised manuscript. In fact, all spectra of XPS were measured in the Cooperative Center for Research Facilities (CCRF) of SKKU and the expert operator always uses C 1s as the standard process for spectra calibrations, which is a general process for the technology of XPS calibration. We strongly believe that the C 1s is valid for the energy calibration, demonstrating the C 1s position for all spectra in Figure S7, which indicates that the energy of spectra was properly corrected. We newly added the detailed information as "X-ray photoelectron spectroscopic (XPS) measurements were performed on a VG Microtech ESCA 2000 using a monochromic Al X-ray source (97.9 W, 93.9 eV); Note that all XPS spectra were calibrated using C 1s at 284.6 eV." For energy calibration of XANES, the energy was calibrated by setting the first inflection points of reference metal foils to the absorption edge energy as 11.215 keV for Ir-L3 edge and 20 keV for Mo K-edge.
We newly added the detailed information as "Ir L3-edge and Mo K-edge XAFS experiments were conducted in an ambient condition at 10C beamline of Pohang Light Sources-II (PLS-II).
The incident beam was monochromatized using a Si (111) double crystal monochromator and detuned to ~70 % of its maximum intensity for reducing higher-order harmonics. For energy calibration, the first inflection points of reference metal foils were set to the corresponding absorption edge energy (11.2152 keV for Ir-L3 edge and 19.995 keV for Mo K-edge). After normalization, the EXAFS signal was k 3 -weighted to magnify high-energy oscillations and Fourier-transformed in the k-range from 2.5 to 11.5 Å -1 ." For the question of (3-1-2), Are the observed shifts in XPS related to charging (MoO3 as a semiconductor may lead to this)?
Answer 3-1-2) Yes, we agree with the reviews' comments. As we mentioned in our original manuscript, we observed Ir 4f shifts in XPS for (i) surface oxygen of Ir and (ii) the electronwithdrawing material of MoO3. For the second factor, the charges (electrons) of Ir NPs are transferred to MoO3, evidenced by the charge density difference simulation (Figures 3c and 3d) and the Mo 5+ from the XPS (Figure 3b). For the question of (3-1-3), In case of XANES, the Ir edge (metal) is at 11.21520 keV, that does not correspond to the metallic Ir provided in the manuscript.
Answer 3-1-3) Thank the reviewer for raising the issue. We are sorry for our mistakes when we plotted the Figures. We corrected the plot that is corresponded to 11.2152 keV for Ir-L3 edge and revised all the Figures (Figs. 4a and 4b) based on the energy calibration mentioned above, as shown below: Answer 3-2) We thank you for the constructive comment. We re-fitted the whole set of spectra of XPS. To clearly demonstrate the peak position, peak shape, and FWHM, we newly made the table S1, as shown below. To justify these peaks that we used for fitting, we added the reference spectra of commercial Ir metal and IrO2. Additionally, the intensity ratio was fixed using standard ratios of 3d3/2: 3d5/2 = 2 : 3 and 4f5/2 : 4f7/2 = 3 : 4, which can be helpful to justify the peaks used to fit in those spectra. Moreover, the analysis of Ir4f XPS spectra shows higher binding energy in comparison to IrO2, that is related to Ir (III) oxidation state as it was discussed in many publications previously (exception to a common situation higher BE = higher oxidation state).

Answer 3-3)
We thank the reviewer for giving constructive comments. We agree with the review's comments. We newly added Table S1 in the revised Supporting Information. As the review mentioned, our measurement of newly added Table S1 showed that Ir (III) oxidation state is higher than the IrO2, as below:  2a), but not matched with Ir(III) (Figure R4).   (Figure S11), which is consistent with the commercial Ir metal.
The oxygen attached to the Ir surface was removed by the Ar etching and finally gave Ir metallic property, which is consistent with XRD and EXAFS spectra. We newly added Figure S11. In addition, we newly added a sentence in our revised manuscript as "To prove whether the interfacial Ir atoms binding to MoO3 or the surface Ir atoms binding to adsorbed oxygen species have higher valence states, the HR-XPS measurement about Ir 4f in IMO is performed with an argon ion etching treated sample 41 , demonstrating zero-valence state of Ir metal in IMO, which is consistent with commercial Ir metal ( Figure S11) and XRD pattern (Figure 2a). This result confirms that the surface Ir in IMO has a high valence state." Based on the three reasons above, we believe that Ir in the IMO sample has a high valence state due to the high electron-deficient state of Ir metal, which is not directly related to the IrCl3.
3-4. The analysis of the electrochemical data shows higher activity of the synthesized materials, specifically for IMO. The Tafel slope of 48 mV dec-1 in case of IMO in comparison to other materials with 70-80 mV dec-1 should correspond to a change in the rds, however nothing is discussed in the manuscript.
Answer 3-4) We thank the reviewer for giving constructive comments. By following the review's comments, we newly added more explanations in the part of Tafel slope, as shown below: By following the previous literature (Sci Rep 5, 13801 (2015.), according to Tafel slope value, the Tafel slope of 120 mV dec -1 was observed when the surface species formed in the step just before the rate-determining step was predominant. In the other cases, the Tafel slope was lower than 120 mV dec -1 . When the surface adsorbed species produced in the early stage of the OER remained predominant, the Tafel slope decreased. In particular, a low theoretical Tafel slope of ~ (2013)).
The IMO showed the low Tafel slope of 48 mV dec −1 due to a large number of oxygen species.
The absorbed oxygen species not only gave the high valence state surface of Ir in IMO but also decreased the Tafel slope, which could accelerate the process of OER and increase the OER efficiency. In addition, based on the DFT simulation, when the oxygen species adopted the surface of Ir metal, the proton dissociating pathway was more preferred because it could further decrease the energy barrier and break the scaling relationship between HOO* and HO*.
Thus, by following the review's constructive suggestion, we newly added the discussion in our revised manuscript as "The Tafel slope of IMO is 48 mV dec −1 , which is remarkably lower than the state-of-the-art RuO2 at 75 mV dec −1 , consistent with the previous reports 1, 2 and Ir at 71 mV dec −1 , consistent with the previous reports 2 . All measured values of Tafel slope is lower than 120 mV dec −1 and we can conclude surface species formed in the step just before the ratedetermining step is not predominant. 3 Due to the high coverage of active species at empty sites that decrease the value of Tafel slope 3, 4, 5 , the low Tafel slope of IMO can be attributed to a large number of oxygen species on surface Ir in IMO. The absorbed oxygen species not only provide the high valence state surface of Ir in IMO but also reduce the value of Tafel slope, which accelerates the process of OER and increases the OER efficiency." "In addition, to break the scaling relation between HOO* and HO* 6, 7 , the proton dissociation pathway (PDP) is suggested for the models IMO (O-8). As a result, PDP indicates the lowest energy barrier when the proton is transferred to the neighbor surface oxygen compared to the relative other OER pathway, and the corresponding configurations are illustrated (Figure 6c), which is consistent with the experimental results of low overpotential and Tafel slope. These results significantly demonstrate that the surface oxygen participates in the OER reaction as a proton acceptor, powerfully uncovering the origin of IMO's excellent catalytic OER performance, which opens up a new avenue for designing highly efficient catalysts. Answer 3-5) We thank the reviewer giving the constructive comments. By following the reviews guidance, we measured the XPS after the stability test for the samples of IMO and IOMO, as shown in new Figure S19. The fitting parameters were newly shown in Table S4.
Based on the XPS result, we can clearly see that the surface structure of IOMO was corroded since the lattice oxygen ratio sharply decreased. Even the Ir 4f of used IOMO (4f7/2 : 62.51 eV; 4f5/2 : 65.38 eV) showed the low valence state compared with that of used IMO (4f7/2 : 62.64 eV; 4f5/2 : 65.64 eV), the surface structure of IOMO was destroyed since the its lattice oxygen sharply decreased after OER test (Table S4). This result showed the IrO2 in IOMO was unstable in comparison with the high valence state of Ir in IMO, which is the origin of the high activity and stability for IMO. We newly added this part in the revised manuscript as follows. "To gain insight into the mechanism, we measure the XPS spectra after the OER experiment. The energy state of the Ir, Mo, and O underwent a minor high-energy shift after the OER process, indicating that these elements lost the electron during the OER ( Figure S19 and Table S4). Note that the surface structure of the IMO is still preserved well after the OER reaction since the synergic effect of the high surface state of Ir with the help of the Mo 5+ can withstand resistance in an oxidation state ( Figure S19a).
However, the surface structure of IOMO can be corroded during the OER process because the ratio of lattice oxygen (OL) decreases compared with that in fresh of IOMO, (Figure S19a), which is the origin of the instability for IOMO in comparison with IMO."