Reconstructed Ir‒O‒Mo species with strong Brønsted acidity for acidic water oxidation

Surface reconstruction generates real active species in electrochemical conditions; rational regulating reconstruction in a targeted manner is the key for constructing highly active catalyst. Herein, we use the high-valence Mo modulated orthorhombic Pr3Ir1−xMoxO7 as model to activate lattice oxygen and cations, achieving directional and accelerated surface reconstruction to produce self-terminated Ir‒Obri‒Mo (Obri represents the bridge oxygen) active species that is highly active for acidic water oxidation. The doped Mo not only contributes to accelerated surface reconstruction due to optimized Ir‒O covalency and more prone dissolution of Pr, but also affords the improved durability resulted from Mo-buffered charge compensation, thereby preventing fierce Ir dissolution and excessive lattice oxygen loss. As such, Ir‒Obri‒Mo species could be directionally generated, in which the strong Brønsted acidity of Obri induced by remaining Mo assists with the facilitated deprotonation of oxo intermediates, following bridging-oxygen-assisted deprotonation pathway. Consequently, the optimal catalyst exhibits the best activity with an overpotential of 259 mV to reach 10 mA cmgeo−2, 50 mV lower than undoped counterpart, and shows improved stability for over 200 h. This work provides a strategy of directional surface reconstruction to constructing strong Brønsted acid sites in IrOx species, demonstrating the perspective of targeted electrocatalyst fabrication under in situ realistic reaction conditions.

6. The reference citation in the following sections is insufficient.
On line 179-187, the discussions on Figure 2e.
On line 213-218, "Two successive oxidation peaks…" and "…a deprotonation process of active oxygen intermediates regulated by Mo.".
In the part of discussing about Mo-buffered charge compensation during reconstruction, there is a lack of extensive reference here. 7. Please label the Raman peaks in Fig. 6e so it is easier to understand the structure evolution. 8. In figure 5b and 5c, the peak for Ir<V is located at the lower energy compared to that for IrIII, indicating that the valence is below +3. Please provide more details on this analysis. 9. In figure 2d, please double check the scale in the x axis.
Reviewer #3 (Remarks to the Author): The manuscript "Reconstructed Ir-Obri-Mo species with strong Brønsted acidity for acidic water oxidation" by Chen et al describes the synthesis and investigation of Pr3Ir1−xMoxO7 and the resulting Ir-Obri-Mo active species by different methods, including DFT, XPS, Raman and electrochemical measurements. Currently, green hydrogen plays a crucial role in the global energy discussion, and with this comes also a special importance of iridium based OER catalysts. The here presented results show new insights and solution in how to improve the stability and activity of the iridium based catalysts. It is shown how the addition of new elements into Iridium oxide (namely Pr and Mo) can improve the performance of the catalyst, and also that the conversion during the OER has to be considered. The manuscript gives deep insights into the appearing processes and the properties of the resulting species. The graphs are presented in a very nice and clear manner and the discussion in the text is easy to understand and to follow. Overall, it is around the most easiest to understand and most clear structured paper I have read within the last few months.
The combination of the theoretical calculations and the electrochemical testing, as well as the characterization of the catalyst before and after the catalyst testing, makes this a nice manuscript with high interest to the research community.
I therefore suggest this manuscript to be accepted for publications after addressing the following minor points: -It is not totally clear to me how the authors choose Pr3IrO7 as a system to do DFT calculations on. Therefore, some sentences should be added at the beginning to make the line of thought of the authors more clear.
-Likewise, a short discussion and outlook should be added at the end to put this research work into the wider perspective Best regards, Steffen Czioska 1 We thank the reviewers for the detailed and constructive comments in improving the quality of this manuscript. Provided below is our detailed point-to-point response to each question. The changes in the manuscript have been highlighted and listed below.

Reviewer #1
The authors present a combined experiment/theory study of the structural evolution and catalytic performance of Mo-doped Pr3IrO7. Their results show that Mo-doping leads to a surface restructuring involving Pr leaching. The resulting surface exhibits highly active species formed by Ir and Mo bridged by an oxygen atom. The authors argue that this surface enables a bridging-oxygen-assisteddeprotonation pathway with a reduced overpotential.
While the experiments are extensive and seem -as far as a theoretician can judge -to be carefully conducted, the supporting DFT calculations leave to desire. Also, the discussion of results is not always sufficiently exhaustive to be accessible to a non-specialist reader, which, in my eyes, is a requirement for publication in a general-audience journal such as Nature Communications. Therefore, I cannot recommend publication of this manuscript.
Following is a detailed list of points of criticism.

Comment 1:
The motivation in the introduction presently does not make complete sense. For example, it is not clear from the cited literature why a multi-valent dopant such as Mo can prevent Ir dissolution. Also, why is Mo chosen here and would other transition metals work as well or maybe even better?

Response 1:
We thank the reviewer for the comment. Firstly, preferential OER mechanisms depend heavily on the electronic structure of electrocatalysts. Studies have suggested that electrocatalysts with high metaloxygen covalency are expected to follow lattice oxygen-mediated mechanism (LOM) because of the strong oxygen character of the band near the Fermi level (Nat.  (2021)). Indeed, the leached elements help with regulating reconstruction process, but they fail to work within the reconstructed layers. As a result, acid-stable doping element will be a priority. According to the acid-stable periodic 2). Page 2, line 62-65, one sentence is revised as follows.
Considering the crystal structure of Ln3IrO7, the multi-valent nature of Mo, and the modulation of Brønsted acidity by Mo and the acidic-stability of molybdenum oxide, 4, 5, 21 it is expected that substitution of Ir in Ln3IrO7 with Mo can realize controllable surface reconstruction towards desired Ir-O (Ⅱ-δ)− -Mo active species with reduced Ir consumption and optimized electronic configuration.

Comment 2:
The language will need major revision. A large part of the manuscript is really difficult to understand and, as such, will presently not appeal to a wide readership.

Response and Modification 2:
We thank the reviewer for the comment. We are sorry for the obscure expressions in the manuscript.
As suggested by the reviewer, to make the manuscript appeal to general-audience, we have revised the language and simplified the expressions.

Comment 3:
Computational methods: i) the authors use plain PBE, which gets redox energetics of Mo oxides very wrong. The materials project recommends using a Hubbard U correction of 4.38 eV on Mo. The authors need to carefully evaluate what is the effect of U on their results and, in case a large effect is observed, recalculate the main results with a more appropriate setup than plain PBE. ii) the valence configuration of the PAW potentials needs to be given to allow the reproduction of the results. iii) no information on the computational setup of the surface calculations is given.

Response 3:
We thank the reviewer for the valuable and perceptive comment.
i) As suggested, we have recalculated the Mo-PIO and Mo-IrO2-Ov systems by PBE + U to evaluate the effect of U on Mo. In general, the trends of vacancy formation energy, reaction free energy difference and electronic distribution have kept consistent with the previous PBE calculations, so the same conclusions can be drawn. In detail, there are some differences in the energy values after a Hubbard U correction of 4.38 eV on Mo atoms. Firstly, all the formation energies (Ef) of different Mo-substitution sites have increased, but the Ir-site substitution still exhibits the most favorable energy (0.25 eV) (Table S1). Secondly, the geometric properties of Mo-PIO have a little change ( Figure S2).   Table S3). Fifthly, the vacancy formation energies in Table S4, Table S5 and Table S6 have  iii) We have added the information computational setup of the surface calculations. In addition, the surface and bulk calculations share similar methods, except for the cut-off energy and the Brillouin zone integrations, which have been highlighted in Computational Methods in the revised manuscript.
As seen, the formation energy is energy favorable for Ir-site substitution (0.25 eV) than that for Pr-site substitutions (7.94 eV for Pr(1), 9.57 eV for Pr(2)), demonstrating that Mo atoms are preferred to occupy six-coordinated Ir sites ( Figure S1 and Tables S1, Supporting Information).
As shown in Table S5 (Table S6).      Table S1 is revised as follows.    Table S6 is revised as follows. Charge density differences are always given with respect to something. Therefore, it is very unclear what figure 1c represents and how it should be interpreted.

Response 4:
We thank the reviewer for the comment. We are sorry for the negligence of detail information about the charge density differences. Detailed description about the charge density differences has been added. And the interpretation for Figure 1c has been highlighted in the revised manuscript.
Charge density difference also verifies the charge redistribution upon Mo substitution ( Figure 1c).
2). Page 3, line 93-94, one sentence is added as follows. 9 The difference was implemented by subtracting the partial charge density of the Ir, Mo, Pr, and O atoms from the whole charge density of the (Mo-) Pr3IrO7.
An obvious charge accumulation (red shadows) is observed around Ir sites accompanied with charge depletion (green shadows) around Mo sites. The Bader charge analysis further quantifies that the Ir atoms of Mo-PIO show the lower (0.16 |e|) charge depletion than that of PIO, indicating the electrons are transferred from Mo to Ir (Table S2).

Comment 5:
The authors use bulk calculations in presence of a Mo dopant and/or oxygen vacancies to explain stability and bonding. This is a huge approximation, as defect formation at surfaces and in the bulk can deviate significantly. Since the authors anyway study surfaces for the OER cycles, the same sampling of defects should be performed at the surface.

Response and Modification 5:
We thank the reviewer for the valuable comment. In previous studies, the formation energy of oxygen It is true that the chemical reaction occurs on catalytic surface. Therefore, we further conducted DFT calculations based on the most thermodynamically stable IrO2 (110) surface for elucidating the role of Mo for OER in reconstructed layers ( Figure 7). As seen, the introduction of Mo has strengthened the Brønsted acidity of Obri in Ir-Obri-Mo species, thus optimizing the proton transfer process in OER following BOAD pathway. The potential determining step is the elemental step of OOtop* + OHbri* → * + O2 + (H + + e -) with a lower overpotential of 0.43 V than that of Ir-Obri-Ir species (0.94 V).

Comment 6:
The authors invoke shifts of the band centers to explain changes in bonding. While I can follow for the O 2p bands, the discussion related to the Ir 5d band is not clear to me.

Response 6
We thank the reviewer for the comment. The difference between O 2p-band center and Ir 5d-band center denotes the charge-transfer energy, which we have given in Table S3 and Figure 1d.   (Table S3).
It can be inferred that the decrease of O 2p band center is mainly caused by O(1) and O(2), and the enlarged Ir-O covalency, which facilitates the reactivity of lattice oxygen, mainly origins from the increase of Ir 5d band center ( Figure 1d, Table S3). Table S3 is highlighted as follows. The schematic in Fig. 5e is impossible to understand. Where is the surface and what does the reconstruction look like? The same is true for many of the figures, like figure 7a or S12, where the viewing direction is unclear.

11
We thank the reviewer for the comment. We have checked and given the viewing direction of Figure   5e. The schematics in Figure 5e represent the evolution of the near surface region. The bulk is not shown after reconstructions. The annotations above each schematic briefly describe the reconstruction.
The viewing direction in Figure 7a and Figure S16 has been given in the revised manuscript.

Comment 8:
In Fig. 8 the PDS should be highlighted to ease the discussion. 13

Response 8:
Thanks for the nice reminder from the reviewer. We have labeled the PDS in Figure 7d in the revised manuscript.  3). Supporting Information, Figure S17 is revised as follows. O (red). The green octahedra represent IrO6 octahedra. PDS is labeled by the green star.

Comment 9:
When discussing the different OER pathways, the authors assume that chemical steps associated with the BOAD pathway are readily overcome. Is this really justified for steps of more than 1 eV? What temperatures would be required to enable these processes on a reasonable time scale?

Response 9:
We thank the reviewer for the comment.  Figure S21. It reveals that the reaction rate of 1 site -1 s -1 corresponds to an ΔGTS of 0.75 eV at room temperature (300 K). As a consequence, the activation free energy of 0.75 eV can be considered as a  2). Supporting Information, Figure S18 is added as follows. 3). Supporting Information, Supporting Note 6 is added as follows.

Supporting Note 6
In the transition state theory (TST), it is generally acknowledged that the TS locates at a point of no return between reactant state and product. And the transition of reactant to TS is the rate determining step (RDS). 21 Based on this, the rate constant can be deduced as:  Figure S21. 21 It reveals that the reaction rate of 1 site -1 s -1 corresponds to an ΔGTS of 0.75 eV at room temperature (300 K). As a consequence, the activation free energy of 0.75 eV can be considered as a limit barrier for the chemical reaction steps at room temperature. Then we calculated the free energy barrier (ΔG) of the chemical step: Otop* +H2O → OOHtop* + OHbri*. The corresponding ΔGTS is 0.64 eV, which below 0.75 eV, indicating the kinetic barrier of chemical steps associated with the BOAD pathway can be readily overcome at room temperature.

Reviewer #2
In this work, the authors have demonstrated Pr3Ir1−xMoxO7 with high-valence Mo modulated to be a good catalyst toward acidic OER. The accelerated surface reconstruction with Mo doping also leads to buffered charge compensation, which prevents the fierce Ir dissolution and excessive lattice oxygen loss. The results are interesting and supported by comprehensive theoretical predictions and solid experimental evidences. However, there are still some issues to be addressed before being publishing in Nature Communications.
Please find the comments below.

Comment 1:
The KIE analysis in Fig.6d is not rigorous at the current state. First, only the H2O was replaced by the D2O, however, the HClO4 was used in both electrolyte; the H is not completely substituted by D.
Secondly, the potential scale was normalized on the RHE scale, rather the overpotential scale. Actually, the pD/pH, and the equilibrium potential for D2O/O2 and H2O/O2 are all different. Please check the literature and do a more rational analysis. 18

Response 1:
Thanks for the rigorous and valuable comment from the reviewer. Actually, we tried to buy deuterated perchloric acid (DClO4) at first but did not make it. Since DClO4 is not commonly used, many chemical reagent suppliers claimed to be out of stock or would not offer its production. We calculated the mole fraction of 1 H in all hydrogen to be only 0.0913% (Supporting Note 4), which has negligible effect on the experiment. So HClO4 was used for KIE analysis in the submitted manuscript.
We have been aware of the different equilibrium potential for D2O/O2 (1.262 V) and H2O/O2 (1.229 V) and converted the potential to overpotential based on different equilibrium potential correctly.
Although DClO4 was not available, we got deuterated sulfuric acid (D2SO4). To eliminate the influence

Supporting Note 4: Calculation of the mole fraction of 1 H in all hydrogen
We calculated the mole fraction of 1 H in all hydrogen to elucidate that 1 H in nondeuterated HClO4 will not affect the results of the experiment significantly.
Take 1 L 0.1 M HClO4 (in D2O) as an example. The deuterium isotopic experiments conducted in 0.5 M H2SO4 (in H2O) and 0.5M D2SO4 (in D2O) also provided consistent results ( Figure S14). Figure S14 is added as follows.

Response 2:
Thanks for the insightful comment from the reviewer. We have synthesized IrO2 and Mo doped IrO2  Figure S19 is added as follows.  Figure S20 is added as follows.

5). Supporting Information, Supporting
Note 7 is added as follows.

Supporting Note 7
As shown in Figure S19a, powder XRD patterns of as synthesized IrO2 and Mo-IrO2 can be well indexed to rutile IrO2 (PDF # 97-005-6009). HRTEM of IrO2 ( Figure S19b) and Mo-IrO2 ( Figure S19c) confirm the good crystallization. The interplanar spacing of about 0.259nm in Figure S19b  Mo can be successfully introduced into IrO2 system and indeed has a positive effect during acidic OER. 6). Supporting Information, Supporting Note 8 is added as follows.

Supporting Note 8
As shown in Figure S20a, powder XRD patterns of as synthesized RuO2 and Mo-RuO2 can be well indexed to rutile RuO2 (PDF # 97-005-6007). HRTEM of RuO2 ( Figure S20b) and Mo-RuO2 ( Figure   S20c) confirm the good crystallization. The interplanar spacing of about 0.256 nm in Figure S20b  higher OER activity but importantly, the excessive oxidation at higher potentials of Mo-RuO2 has been obviously impeded, which is more important for practical applications.

Comment 3:
In this work, OER performance was assessed in 0.1M HClO4 electrolyte, which is different from those in literature (usually pH ≈ 0). It would be helpful to provide the OER performance tested in 1 M HClO4 or 0.5 M H2SO4, and then compared to other acidic OER catalysts.

Response 3:
We thank the reviewer for this comment. We have conducted OER performance tests in 0.5 M H2SO4 and the comparisons with other reported Ir-based electrocatalysts are provided based on geometric and mass activities.
And OER performance tests conducted in 0.5 M H2SO4 further reveal that 0.2Mo-PIO-post is among the most active electrocatalysts for OER in acid ( Figure S9, Table S12).
2). Supporting Information, Figure S9 is added as follows.  Table S12 is added as follows.

Comment 4:
The stability measurement is conducted at the geometric current density of 5 mA cm -2 , which is quite  Table S16).
2). Supporting Information, Figure S12 is added as follows.  Table S16 is added as follows. where nO2 and nIr (dissolved) refer to the total amount of evolved oxygen (calculated from total charge) during the chronopotentiometry test and the amount of dissolved Ir extracted from ICP-MS results, respectively.

3). Supporting Information,
1. Geiger S, et al. The stability number as a metric for electrocatalyst stability benchmarking. Nat.

Comment 5:
In introduction, authors claimed that "the mass content of iridium in Ln3IrO7 compounds (26.4% in Pr3IrO7) is obviously lower than that in IrO2 (85.7%), perovskite and pyrochlore structures (58.6% in SrIrO3 and 49.4% in Pr2Ir2O7). Consequently, if properly tuned, directional surface reconstruction with enhanced stability and further decreased iridium consumption of Ln3IrO7 will be anticipated." It is not rigorous, since in the PEM electrolytic water system, we are more concerned about the loading amount of precious metals on the membrane electrode. It is more convincing to compare the mass activity (normalized to Ir mass) of the as-prepared catalysts herein to other Ir-based catalysts.

Response 5:
Thanks for the valuable comment from the reviewer. We have revised this sentence and compared the mass activity of catalysts herein with other Ir-based electrocatalysts in acidic media.
Consequently, if properly tuned, directional surface reconstruction with improved stability and mass activity (normalized to Ir mass) of Ln3IrO7 will be anticipated.
As shown in Figure S8, 0.2Mo-PIO-post gives a mass activity of 415 A·gIr −1 at 1.52 V, about 4.3 and 88.3 times higher than that of PIO-post (97 A·gIr −1 ) and commercial IrO2 (4.7 A·gIr −1 ). The iridium mass activity was also compared with reported Ir-based catalysts in acidic media (Table S11), demonstrating the high mass activity of 0.2Mo-PIO-post among reported Ir-based catalysts. Figure S8 is added as follows.  Table S11 is added as follows. The reference citation in the following sections is insufficient.
On line 213-218, "Two successive oxidation peaks…" and "…a deprotonation process of active oxygen intermediates regulated by Mo.".
In the part of discussing about Mo-buffered charge compensation during reconstruction, there is a lack of extensive reference here.

Response 6:
We thank the reviewer for the comment. We have added references in corresponding sections in the revised manuscript.

Comment 8:
In figure 5b and 5c, the peak for Ir <V is located at the lower energy compared to that for Ir III , indicating that the valence is below +3. Please provide more details on this analysis.

Response 8:
We thank the reviewer for the comment. As supported by reported theory and experimental evidence,  . 131 72 (1984)). As a result, although the peak of Ir <V species is located at the lower energy compared to that of Ir III species, the average oxygen state is above +3. We have provided more detailed analysis in the revised manuscript. 2). Supporting Information, Supporting Note 3 is added as follows.

Supporting Note 3
As supported by reported theory and experimental evidence, the main line peaks of Ir III species appear at around 62.3 eV and 65.3 eV for Ir 4f7/2 and Ir 4f5/2, respectively, which show positive shift of binding energies compared with those for Ir Ⅳ species (around 61.7 eV and 64.7 eV for Ir 4f7/2 and Ir 4f5/2, respectively). As a result, although the peak of Ir <V species is located at the lower energy compared to that of Ir III species, the average oxygen state is above +3. After 2 electrochemical cycles, the surface lattice oxygen oxidation (O2 release) and subsequent occupation of oxygen vacancies with water molecules or hydroxyl contribute to the reduction of metal oxidation state. As shown in HRTEM and corresponding SAED images (Figure 4a, b right panel), sparse particles appear on the surface but with well-ordered bulk crystalline, indicating an ongoing surface reconstruction. So, the surface is not fully covered with hydroxide species IrOx.

Comment 9:
In figure 2d, please double check the scale in the x axis.

Response 9:
We thank the reviewer for the critical comment. We have checked and corrected the x axis scale in

Reviewer #3
The manuscript "Reconstructed Ir-Obri-Mo species with strong Brønsted acidity for acidic water oxidation" by Chen et al describes the synthesis and investigation of Pr3Ir1−xMoxO7 and the resulting Ir-Obri-Mo active species by different methods, including DFT, XPS, Raman and electrochemical measurements. Currently, green hydrogen plays a crucial role in the global energy discussion, and with this comes also a special importance of iridium based OER catalysts. The here presented results show new insights and solution in how to improve the stability and activity of the iridium based catalysts. It is shown how the addition of new elements into Iridium oxide (namely Pr and Mo) can improve the performance of the catalyst, and also that the conversion during the OER has to be considered. The manuscript gives deep insights into the appearing processes and the properties of the resulting species.
The graphs are presented in a very nice and clear manner and the discussion in the text is easy to understand and to follow. Overall, it is around the easiest to understand and most clear structured paper I have read within the last few months.

37
The combination of the theoretical calculations and the electrochemical testing, as well as the characterization of the catalyst before and after the catalyst testing, makes this a nice manuscript with high interest to the research community.
I therefore suggest this manuscript to be accepted for publications after addressing the following minor points:

Comment 1:
It is not totally clear to me how the authors choose Pr3IrO7 as a system to do DFT calculations on.
Therefore, some sentences should be added at the beginning to make the line of thought of the authors clearer.

Response 1:
We thank the reviewer for the comment. We have added some sentences at the beginning to elucidate the line of the thought to choose Pr3IrO7 as a system to do DFT calculations on in the revised manuscript. Based on the flexible crystal structure and characteristic band structure of Pr3IrO7, high-valent Mo is doped into Pr3IrO7 to accelerate surface reconstruction and obtain active reconstructed layers ( Figure   1).

3). Supporting Information, two sentences are added in Computational Methods.
Pr3rO7 crystal with Cmcm space group was selected, and the optimal lattice parameter is a = 7.55 Å, b = 11.07 Å, c = 7.63 Å. For Mo-doped Pr3rO7, the Ir site is substituted with one Mo atom, and the corresponding lattice parameter have a little variation with a=7.56 Å, b=11.08 Å, c=-7.67 Å.

Comment 2:
Likewise, a short discussion and outlook should be added at the end to put this research work into the wider perspective.