In-situ spectroscopic observation of dynamic-coupling oxygen on atomically dispersed iridium electrocatalyst for acidic water oxidation

Uncovering the dynamics of active sites in the working conditions is crucial to realizing increased activity, enhanced stability and reduced cost of oxygen evolution reaction (OER) electrocatalysts in proton exchange membrane electrolytes. Herein, we identify at the atomic level potential-driven dynamic-coupling oxygen on atomically dispersed hetero-nitrogen-configured Ir sites (AD-HN-Ir) in the OER working conditions to successfully provide the atomically dispersed Ir electrocatalyst with ultrahigh electrochemical acidic OER activity. Using in-situ synchrotron radiation infrared and X-ray absorption spectroscopies, we directly observe that one oxygen atom is formed at the Ir active site with an O-hetero-Ir-N4 structure as a more electrophilic active centre in the experiment, which effectively promotes the generation of key *OOH intermediates under working potentials; this process is favourable for the dissociation of H2O over Ir active sites and resistance to over-oxidation and dissolution of the active sites. The optimal AD-HN-Ir electrocatalyst delivers a large mass activity of 2860 A gmetal−1 and a large turnover frequency of 5110 h−1 at a low overpotential of 216 mV (10 mA cm−2), 480–510 times larger than those of the commercial IrO2. More importantly, the AD-HN-Ir electrocatalyst shows no evident deactivation after continuous 100 h OER operation in an acidic medium.

new information it adds. Also in the main text, the word "dynamic" seems to have no important meaning and most sentences would contain the same information when it was deleted.
2) Please add the source of the Ir atoms to the main text. It was hard to find this important information in the 3) EXAFS analysis a) Please add the details of the Fourier transform and used window function, otherwise the figure is not reproducible. b) Please include the data in k-space, which contains more information c) P7. What is meant by "oscillation frequency"? The EXAFS oscillations in Fig. 2e? Only one period is shown and it matches that of IrO2 and not that of a material with Ir-N bonds d) How were the fits performed? In k-space in which range or in R-space or Fourier-filtered kspace (Q-space in Artemis)? This is important to reproduce the work and judge its meaningfulness e) What structural model was used and what were the parameters to obtain the phase functions? f) The red line in Fig. 2f has peaks at higher R. Can they be fit with Ir-Ir interactions (thus indicating no single atom Ir)? why should the peaks in the FT around 3 A not be significant? They look well above the noise level and fall into the range expected for Ir-Ir distances. g) Is a better background subtraction possible for the data in Fig. 5b? Can it be excluded that the difference in the analyzed peak height is due to the distortion that causes the peak at 1 A? The kspace data would help experts to identify issues with the data and judge if the analysis of the peak at 1.6 A is sound. 4) Activity a) Please test the activity and stability also at higher current densities such as 500 mA/cm2, 1000 mA/cm2 and 2000 mA/cm2 if possible. PEM electrolyzers are operated under harsher conditions and I am not convinced stability at 10 mA/cm2 is relevant for this application. b) Please repeat at least the activity measurements to get statistics. c) It is good to perform measurements in a two-electrode cell but the one used by the authors is by no means close enough to a PEM electrolyzer system make statements about "practical electrolyzers" d) P15. 100VmA is not a universally suitable potential window for Cdl determination by cyclic voltammetry and the range of sweep speeds is rather narrow (https://iopscience.iop.org/article/10.1088/2515-7655/abee33). This can introduce large systematic errors in the determination of specific activity (in Fig. S11).
5) At what current density are the Tafel slopes evaluated? Please state explicitly. There seem to be at least two Tafel slopes in the plot of each trace and the traces are not very linear in this plot 6) It was not quite clear to me during the first read why the catalyst electrode is compared to PANI-Ir. Please revise the text with a statement that the catalyst Ir-NC is based on PANI-Ir but has an additional step in the preparation procedure. 7) P10. How can Rct be used to represent the H2O/OOH ion adsorption resistance? Rct relates to charge transfer across an interface. Does the statement make use of the SRIR results discussed later? The connection between Rct and "adsorption kinetics" is likewise unclear. Please revise. In the first half of P10, there is no data yet that OOH is created at all. 8) P14. Conclusion: "Hence, this discovery of active sites evolution under working conditions can provide a coordination-engineered strategy for designing advanced acidic-OER electrocatalysts." What is meant? How exactly are the reported findings useful for other electrocatalysts? What should other scientists look for in good OER catalysts in acid? Finally, what is meant by "active sites evolution"? Species must adsorb at the active site? Does this refer to the O adsorption on the other side? In the latter case, does it desorb again at lower potential or stay there? The insight that the active state of electrocatalysts differs from the as-made material is not novel and was reported numerous times.
7) The panel labels a,b,c,… seem to be missing for most of the figures. 8) P17. How was the data smoothed? What method? What was the difference between neighboring steps? 2 cm-1? 100 points could smooth the spectra too much. Please also show the unprocessed data. 9) Please have the manuscript checked by a native English speaker. It is largely understandable but some sentences do not read well (e.g. The N K-edge in Figure 2b, the peaks of excitation of C-N and hetero-Ir-N were obtained.)

Response to Reviewers' Comments
We are grateful to the reviewers for having given us important and valuable comments on the manuscript NCOMMS-21-20746 entitled: The manuscript by Su et al. is addressing an interesting topic related to the electrocatalytic oxygen evolution reaction (OER). The choice of methods is appropriate, and the authors performed controlled experiments such as operando X-ray absorption fine structure and infrared spectroscopy under OER working conditions for investigating the catalytic dynamic process. The results are likely to be of wide interest to catalytic reaction mechanisms and catalyst design. Overall, I consider that both the significance and the novelty of this work could satisfy the high standards of Nature Communications. Therefore, I recommend the publication of this manuscript after the following issues are clarified:

Question:
The design principle of the catalysts should be further clarified. In the experimental step, why did the authors select nitric acid rather than hydrochloric acid or sulfuric acid during the electrodeposition process?
Reply: This is a very good question. We are sorry that the principles for selecting electrolyte in electrodeposition were not clearly described in the previous manuscript.
Nitric acid was chosen as the electrolyte mainly for the exclusion of impurities and it would be discussed in more detail below.
In this work, in order to obtain atomically dispersed Ir atoms with covalent coupling in a hetero-nitrogen-configured 3D carbon substrate, the most key step is first electrodeposition of polyaniline layer onto a 3D carbon substrate and then the surface was functionalized with the amino group for strongly bonding to Ir complex ions.
Firstly, in the electrodeposition process, the most important point is to avoid introducing heteroatoms in the electrodeposition process of polyaniline. The Cl (HCl) and S (H2SO4) are easily introduced into the polyaniline skeleton to form the doping of Cl and S elements in the later high temperature annealing. Therefore, nitric acid, which contains no miscellaneous elements, is the best choice for acidic electrolytes in electrodeposited polyaniline. Furthermore, the NO3is easily decomposed at high temperature without introducing any miscellaneous elements for catalysts. Above all, considering the 3 avoidance of impurity doping of anions, we chose HNO3 as electrolyte in the electrodeposition process.
Accordingly, in line 21, page 16 of the revised manuscript, the following text has been added: "Considering the avoidance of impurity doping of anions, we chose HNO3 as electrolyte in the electrodeposition process." 2. Question: Temperature is an important factor in the synthesis process of the samples in this work. The temperature of carbonization will affect the structure of carbon and then affect the performance of the sample. Did the author try other temperatures?
Reply: Thank you for your careful consideration. Specifically, the polyaniline electrode is heated to 800 °C for 3 h to obtain the AD-HN-Ir electrocatalyst. At 800 °C, polyaniline would produce pyrolysis and be gradually carbonized to porous highly conductive hetero-nitrogen-configured 3D carbon materials. At the same time, the Ir atoms were chemically coupled with the N sites to form the hetero-Ir-N4 active centers on a hetero-nitrogen-configured 3D carbon substrate.
Noticeably, the graphitization of the substrate has a great influence on the properties, and the graphitization trend of the carbon substrate will be more obvious with the increase of temperature. However, the disadvantages of higher temperature are the collapse of substrate morphology and the agglomeration of single atoms on the surface, thus reducing the catalytic activity (Adv. Energy Mater. 8, 1702476 (2018)).
Considering that the pyrolysis temperature will affect the carbonization degree of polymer and the distribution of Ir, the samples were also performed in different pyrolysis temperatures (700 °C and 900 °C). As shown in the Figure N1, the assynthesized Ir-700 samples (pyrolysis temperature of 700 °C) has clear dendritic structure similar to polyaniline and delivers an overpotential of 285 mV at 10 mA cm -2 , which reveals low electron conduction and material diffusion attributed to the inadequate polyaniline carbonization at a lower temperature. Furthermore, Figure N1d clearly shows that the small Ir nanoparticles were distributed on the 3D carbon substrate and a higher overpotential of 252 mV at 10 mA cm -2 was observed for the as-4 synthesized Ir-900 sample (pyrolysis temperature of 900 °C), suggesting Ir atoms aggregated into nanoparticles at a higher pyrolysis temperature. Above all, considering the conductivity of the carbon substrate and the phase composition of Ir sites, the pyrolysis temperature of 800 °C was selected for AD-HN-Ir electrocatalyst. It can insure a stable hetero-Ir-N4 moieties spatially confined in the hetero-nitrogenconfigured 3D carbon substrate for an efficient 4-electron OER process.

Questions:
The author mentioned that an oxygen is dynamically coupled at the Ir

Reply:
We are greatly grateful to the reviewer for your nice question and it is useful for improving the quality of this work. The atomically dispersed Ir active sites coupled onto hetero-nitrogen-configured 3D carbon substrate can form stable and highly active Ohetero-Ir-N4 structure, obviously accelerating the 4-electron kinetics process and promoting the formation of key OER reaction intermediates * OOH for efficient acidic-OER activity.
Firstly, a new absorption band at 784 cm -1 was observed by in-situ SRIR measurements for AD-HN-Ir electrocatalyst at potential of 1.25 V, which can be assigned to the emergence of Ir-O during OER process. Simultaneously, the in-situ XAFS results reveal that the intensity of white-line peak of AD-HN-Ir electrocatalyst increases gradually with the increase of voltage, accompanied by a slight positive-shift, indicating more electrons move from Ir to nearby atoms and oxygen-species.
Interestingly, the peak intensity increased significantly at the potential of 1.25 V and the fitting result clearly shows an additional first shell of Ir-O coordination, suggesting a local structural self-optimization and adsorption of reactive species under OER working conditions, which optimizes the electronic structure of the active site and speeds up the production of the reaction intermediate. Importantly, a new absorption band in the vibration frequency of 1055 cm -1 was observed for AD-HN-Ir electrocatalyst when the potential > 1. 35  to form stable and highly active O-hetero-Ir-N4 structure, obviously accelerating the 4electron kinetics process and promoting the formation of key OER reaction intermediates OOH for efficient acidic-OER activity.
Accordingly, in line 20, page 15 of the revised manuscript, the following text has been added: "Above all, the in-situ SRIR and XAFS results jointly reveal that Ir active sites coupled onto the hetero-nitrogen-configured 3D carbon substrate by a strong coupling substrate effect can dynamically couple one oxygen atom to form a stable and highly active O-hetero-Ir-N4 structure during the reaction process, obviously accelerating the 4-electron kinetics process and promoting the formation of key OER reaction intermediates OOH for efficient acidic-OER activity (Figure 5d)." 4. Question: Environment for electrochemical testing has a certain effect on the performance of the catalyst. The potential temperature effects should be considered and the related discussion on this issue should be included.
Reply: Thank you for your nice questions. We really praise the reviewer for your expert knowledge in the field of electrochemistry. According to your suggestion, the electrochemical measurements were performed under different temperatures to seek the relationship between working temperature and activity. Seen from Figure N2, it can be drawn that the AD-HN-Ir electrocatalyst delivers faster kinetics and better activity at a higher temperature.
The temperature of the electrochemical cell is an important parameter that significantly affects the electrochemical performance tests (J. Am. Chem. Soc. 140, 2926−2932(2018). Generally, the performance tests of the electrochemical cell were carried out at room temperature. In order to further explore the temperature effect, the electrochemical cell was put on the temperature-control heater with temperature controlled at 25, 50 and 80 °C for temperature-dependent studies. As shown in the Figure N2, the OER activity are quite temperature-dependent, and the overpotentials of AD-HN-Ir electrocatalyst decrease from 216 to 198 mV at a current density of 10 mA cm -2 with the increase of temperature to 80 °C, which is consistent with what has been 7 observed that a higher temperature provides faster OER kinetics and better OER activity (J. Power Sources 167, 235−242 (2007)). Based on the above results, the AD-HN-Ir electrocatalyst shows temperature-dependent OER activity with faster ORR kinetics at higher temperatures. Accordingly, Figure N2 has been added in the Supplementary Supplementary Fig. 11)."

Question:
The presentation of the experimental results in the manuscript should be further improved. For example, some of the potentials listed throughout the manuscript need to be listed/provided relative to a reference electrode.
In this manuscript, single-atomic iridium based ultralow-iridium electrocatalyst was synthesized via "electric-driven amino-induced" strategy, which delivers a low overpotential of 216 mV to achieve a current density of 10 mA cm -2 . With the operando synchrotron radiation X-ray absorption fine structure (XAFS) spectroscopy analysis, the author revealed that oxygen was dynamically pre-adsorbed on the Ir site in the form of O-hetero-Ir-N4 moiety under low driven-potential to accelerate the transfer of electrons from the metal sites to neighboring atoms. The work is scientifically sound and well presented. However, some questions remain unresolved regarding the origin of the observed stability of the catalyst. The stability of catalysts is a more important issue than the catalytic activity for the practical application. A suitable explanation for the observed stability of the catalyst would open a great avenue for further developments of single atom catalysts. With that part augmented, this work would be a welcome addition to Nature communications.

Reply:
We are greatly grateful to the reviewer for your nice question and it is useful for improving the quality of this work. According to your suggestion, the stability of acidic-OER for AD-HN-Ir electrocatalyst will be analyzed in detail from three aspects: preparation, structure and performance.
The operation stability is an important index to evaluate the performance of electrocatalyst, especially for acidic OER. It is known that for highly reactive iridium Adv. Mater. 30, 1804333 (2018)). To improve the durability of the electrocatalyst under acidic OER operation condition, the key is to construct stable coordination structure of Ir active sites to inhibit the peroxidation and the aggregation and dissolution of the surface active phase under high potentials.

Preparation:
In this work, the AD-HN-Ir electrocatalyst with atomically dispersed Ir active sites coupling in the hetero-nitrogen-configured 3D carbon substrate was synthesized via a controllable "electric-driven amino-induced" strategy. The polyaniline layer was first electrodeposited onto a 3D carbon substrate, and then functionalized with amino groups by mild heat treatment in ammonia solution. It is noted that the surface reductive benzenoid-amine groups and -NH2-derived uncoordinated N sites act as the anchoring sites for Ir atoms during ions exchange and pyrolysis processes, and the atomically dispersed Ir active sites with stable hetero-Ir-N4 moieties were strongly coupled on the 3D carbon substrate. This stable configuration has the potential to inhibit the peroxidation and aggregation of Ir active sites at high potentials.

Structure:
The in-situ XAFS and SRIR techniques reveal that one oxygen atom was dynamically coupled on the Ir active site in the form of O-hetero-Ir-N4 moiety under low driven-potential, and then H2O adsorption under a low driven-potential and crucial * OOH intermediate production were achieved. Simultaneously, the in-situ XAFS results under working conditions doubtlessly proved that the Ir active sites were only weakly oxidized to a highly active oxidation state and remains atomically dispersed after long-time operation. Most importantly, one oxygen atom was dynamically formed on the Ir active site with stable O-hetero-Ir-N4, which endows higher coordination numbers and higher 5d electron occupied states at the Ir active sites to inhibit the dissolution of Ir sites for high activity and durability.

Performance:
The stability tests at potentials of 1.53 and 1.75 V were performed.
The activity of acidic OER did not decrease significantly after continuous operation for 50 h under a low current density of 10 mA cm -2 ( Figure 3f) and a high current density of 100 mA cm -2 ( Figure N3). Meanwhile, the morphology characterization showed that no collapse of the carbon substrate and no obvious particles were observed. Above all, the AD-HN-Ir electrocatalyst with atomically dispersed hetero-Ir-N4 moieties delivers 11 high durability due to the optimization of electronic and coordination structure by the dynamic oxygen formed over the Ir sites. as an electrocatalysts for the oxygen evolution reaction in acid. The material and its activity are characterized by SEM, TEM, STEM_EDS, XAS at the C-K, N-K, Ir-L3 edges, XPS, synchrotron IR spectroscopy, cyclic voltammetry and impedance spectroscopy. Ir is a known active site for the OER in acid. I see the main achievements of the authors in reducing its loading while maintain high activity and stability, which is on the roadmap of many international initiatives. The authors also convincingly explain why their catalyst is more active than simply putting Ir on PANI.
The main claims are C1 -Large mass activity C2 -High stability C3 -The good performance is due to pre-adsorption and facile OOH production Reply: Thank you for your positive evaluation on our work. The specific responses to your constructive questions are as follows.
C1. I checked this carefully and I agree that the produced electrodes are among the highest reported for the OER in acid. I do not find that data/references in Supplemental Table 2 are sufficient to support the claim.
Please cite and include the state-of-the-art catalysts from recent reviews, e.g. https://onlinelibrary.wiley.com/doi/full/10.1002/adma.201806296 https://pubs.rsc.org/en/content/articlelanding/2020/nr/d0nr02410d#!divAbstract A previous report in Nat. Commun should be discussed and compared in detail: https://www.nature.com/articles/s41467-019-12886-z What are the major steps forward in this work? (and why is this not included in the Suppl. Table?) Reply: We thank the reviewer for the constructive suggestion that is quite useful for improving the quality of this work. According to your suggestion, more detailed 13 research results and related reference literature are rearranged and presented in Table   N1. The AD-HN-Ir electrocatalyst showed optimal electrocatalytic activity according to the mass activity results at a lower overpotential of 216 mV.
Mass activity is an important parameter for the evaluation of the intrinsic activity of electrocatalysts. As seen from Table N1, the mass activity of the AD-HN-Ir electrocatalyst is up to 2860 A gmetal -1 at the overpotential of 216 mV, which is remarkably larger than those of the reported electrocatalysts and the commercial IrO2.
It is noted that a recent reported Ru-N-C catalyst also displays a high mass activity of 3571 A gmetal -1 at an overpotential of 267 mV (Nat. Commun. 10, 4849 (2019)). For better comparison, the mass activity of the AD-HN-Ir electrocatalyst at the overpotential of 260 mV can reach 10900 A gmetal -1 , which is obviously better than those activities of other reported electrocatalysts.  (2019)) have been included in the Table 2 in the revised Supplementary Information.
How reproducible are the activity results? How many electrodes were prepared and tested? Please add error bars to support a consistently large mass activity.
Reply: That is a good question. In the performance test, the electrode needs to undergo continuous CV test first, and then the realistic performance test is carried out after the electrode is stable to ensure the reliability of performance results. In addition, samples and electrodes were prepared repeatedly, and the results presented in the manuscript were all obtained after at least 10 tests of different batches of electrodes, which can ensure the repeatability of performance results. According you suggestion, the error bars have been added in the Figure N4 to support their consistently large mass activities. C2 -The electrodes have high stability at 10 mA/cm2. However, this is not a realistic condition for a PEM electrolyzer. I appreciate that at least the activity was given at 100 mA/cm2, which also is too low to be useful for an electrolyzer.

Reply:
We thank the reviewer for this constructive suggestion. According to your suggestion, the stability at 100 mA cm -2 was performed with the test results shown in Figure N5. The AD-HN-Ir electrocatalyst reaches a large current density of 100 mA cm -2 at a small overpotential of 292 mV. Intriguingly, at the high current density of 100 mA cm -2 , the AD-HN-Ir electrocatalyst still exhibits robust stability with a prominent retention of ~80% of the initial current density after 50 h acidic OER operations.
Importantly, the SEM image shows that no significant structural collapse was observed in the AD-HN-Ir electrocatalyst. Reply: Thank for your nice question. To reach the high current density of 500 mA cm -2 for the AD-HN-Ir electrocatalyst, the high potential of 1.75 V was applied on the electrode. The AD-HN-Ir electrocatalyst retains ~50% of the initial current density after 10 h acidic OER operations ( Figure N5). The OER performance in acid condition is a great challenge for carbon-based materials because of its easy oxidation at high anodic voltage (~1.75 V) in an acidic environment (ChemElectroChem 5, 583-588 (2018); Angew. Chem. 130, 16749-16753 (2018)). Therefore, the well-designed Ir sites were spatially confined in the 3D conductive carbon substrate to display high activity and robust stability during acidic OER process only below the potential of 1.75 V. It is difficult to achieve a high activity and stability at a higher current density of 1000 and 2000 mA cm -2 because of oxidation corrosion of carbon substrate under high oxidation potential.  Fig. 21c and 21d)." And in reverse line 3 of page 10 of the revised manuscript, the following text has been added: "It is difficult to achieve a high activity and stability at higher current densities of 1000  (2018)). According your suggestion, the dissolved content of metal Ir, the calculated S-number and lifetime are shown in Figure   N7. Significantly higher S-number and longer lifetime of the AD-HN-Ir electrocatalyst were observed at the current density of 100 mA cm -2 . It further proves that the AD-HN-Ir electrocatalyst delivers a good activity and stability at the current density of 100 mA cm -2 related to 500 mA cm -2 .
As is known, many reported Ir based oxide electrocatalysts have been dissolved and surface structure reconstruction in acidic solution under the voltage driven, thus achieve a high catalytic activity (Science 353, 1011-1014 (2016), Nat. Commun. 7, 12363 (2016)). The S-number is closely related to the amount of dissolved iridium and the evolved oxygen, which can be used to evaluate the stability of the active site of Ir-based oxide electrocatalysts. In this work, atomically dispersed Ir active sites coupled in 3D carbon substrate with active hetero-Ir-N4 moieties, which can efficiently avoid the obvious changes of the Ir oxidation state and significant dissolution of Ir active sites.
The advanced in-situ XAFS and SRIR techniques have revealed that the real active 19 structure is the O-hetero-Ir-N4 moiety during OER process. However, a high oxidation potential applied on the electrode causes the oxidation of 3D carbon substrate, and then the dissolution of Ir active sites can be observed. Therefore, the S-number cannot directly reveal the relationship between the structure of Ir active site and the performance of atomically dispersed Ir electrocatalyst system, but it can also be used as a parameter to evaluate the stability of the catalyst. As shown in Figure N7a, the low amount of dissolved iridium at low potential of 1.53 V (100 mA cm -2 ) is due to the stable O-hetero-Ir-N4 structure. Furthermore, the calculated S-number and lifetime of AD-HN-Ir under lower potential (1.53 V @ 100 mA cm -2 ) are significantly higher than those of higher potential (1.75 V @ 500 mA cm -2 ), suggesting the decay of the stability attributed to the corrosion of the carbon substrate under a high potential. Please also address the following points: 1) Title a) The title states "single atom", yet Fig. 1c is too small to judge and Fig. S5 shows larger spots (near the 5 of the scale bar). How many spots were analyzed? Is the selection of Fig. S5b representative? I find the expression "atomically dispersed Ir" much better than claiming all single atoms, which may not be supported by EXAFS analysis, see below.
Reply: This question is very important to improve the quality of this work. According to your suggestion, the more appropriate magnification of HAADF-TEM image was provided and shown in the Figure N8. It demonstrates that atomically dispersed Ir atoms were uniformly anchored to the hetero-nitrogen-configured 3D carbon substrate without significant agglomeration. In the Figure N9, more than 30 points were analyzed in HAADF-TEM image. The bright spots corresponded to Ir atoms are uniformly dispersed along with similar dimensions. As shown in the Figure N9b, the bright spots are highly dispersed across the substrate and intensity profiles in different selection were further analyzed, suggesting the atomically dispersed Ir atoms chemically coupled onto hetero-nitrogen-configured 3D carbon substrate with a particle size of ~0.21-0.23 nm.   Fig. 5)." Moreover, according to your suggestion, to more accurately describe the form of AD-HN-Ir electrocatalyst, the expression "single atom Ir" in the previous manuscript has been replaced by the "atomically dispersed Ir" in the revised manuscript.
b) The study is in situ but not operando. Operando requires a simultaneous product measurement (e.g. https://www.sciencedirect.com/science/article/pii/S0920586104008065?via=ihub ). That could be the current for electrocatalysis but no data from in situ experiments is shown. Please either add this data (which must be convincing) or use "in situ".

Reply:
We thank the reviewer for the constructive suggestion that is quite useful for improving the accuracy of this work. According to your suggestion, the "operando" in the previous manuscript has been replaced by "in-situ" in the revised manuscript. c) I am not sure what is meant by "dynamic-coupling oxygen". Oxygen species adsorb and desorb during the catalytic cycle. This is expected. In order to oxidize hydroxide/water, one must adsorb it and then release O2. I recommend to remove the confusing expression or discuss in detail what new information it adds. Also in the main text, the word "dynamic" seems to have no important meaning and most sentences would contain the same information when it was deleted.
Reply: Thanks for your nice question. We are sorry that the meaning and concept of "dynamic-coupling oxygen" was not clearly expressed in the previous manuscript. In this work, we design a type of atomically dispersed Ir active sites anchored onto the 3D carbon substrate with the hetero-Ir-N4 moieties. Interestingly, by combining in-situ XAFS and S-FTIR technologies, we observed in experiment that one oxygen atom was 24 dynamically coupled to the Ir active site with the form of the O-hetero-Ir-N4 structure under working conditions. After removal the working potential, the active structure returns to the hetero-Ir-N4 structure proved by in-situ SRIR results ( Figure N10). Therefore, we use "dynamic-coupling oxygen" to highlight the potential driven oxygen atom coupled to the Ir active site during the reaction process. Furthermore, according to your suggestion, we have checked the whole manuscript carefully and removed some superfluous expressions of "dynamic" without compromising the meaning of the statement in the revised manuscript.  Subsequently, k 2 -weighted χ(k) data in the k-space ranging from 3.0-11.8 Å -1 were Fourier transformed to real space using a Hanning windows (dk = 1.0 Å -1 ) to separate the EXAFS contributions from different coordination shells." b) Please include the data in k-space, which contains more information Reply: It is a nice question. According to your suggestion, the L-edge EXAFS k 2 (k) functions of AD-HN-Ir electrocatalyst and reference samples were added in Supplementary Fig. 8 of the revised Supporting Information. As shown in the Figure   N10, it can be inferred that the high-valence Ir active site is formed in the first-shell of Ir-N coordination, based on the position of the Ir L3 absorption edge and the first-shell coordination peak at ~1.6 Å. Figure N11b shows the difference of the EXAFS 26 oscillation frequency between the AD-HN-Ir electrocatalyst and the reference samples.
This phenomenon indicates that the atomically dispersed Ir active sites with a Ir-N coordination configuration of N atoms in the first-shell are anchored to a heteronitrogen-configured 3D carbon substrate. f) The red line in Fig. 2f has peaks at higher R. Can they be fit with Ir-Ir interactions (thus indicating no single atom Ir)? why should the peaks in the FT around 3 A not be significant? They look well above the noise level and fall into the range expected for Ir-Ir distances.

Reply:
We thank the reviewer for the insightful suggestion that is quite useful for improving the consistency of this work. In order to identify the source of the high R peak, we tried several k ranges for FT and multi-shells fitting, and it reveals that the peak at 2.74 Å may come from the contribution of a small amount of the first Ir-Ir coordination shell (less than 10%).
Firstly, in order to realize the best background removal for EXAFS data analysis, we selected the different Rbkg values of 0.9, 1.0 and 1.1 Å to remove low frequency noise.
As shown in Figure N12b, the best background removal was at the Rbkg=1.1 Å, and the low frequency noise was removed fully but the high frequency noises are almost the same for the different Rbkg values. Furthermore, it can be found that the larger noises appear in the k region of 8 and 12.5 Å -1 ( Figure N12a). Subsequently, the four different k ranges were selected for the fast Fourier transform, and the results were shown in Figure N12c. It can be observed that the 2.74 Å FT peak intensity of the 3.0-10.2 Å -1 region is reduced by 100% in comparison with that of 3.0-11.8 Å -1 region, suggesting the increase of the high-frequency noise in the high k region of 3.0-11.8 Å -1 because of the low Ir content for fluorescence XAFS measurement. However, the large k region of 3.0-11.8 Å -1 was necessary for more independent free points (Nipt=2 k·R/π=2(11.8-3.0)(2.5-1.0)/π=8) for the better data curve fitting. Finally, the FT spectrum in Figure   N12d shows that the peaks at 2.59 and 2.85 Å can be attributed to the first Ir-Ir shell (Ir foil) and the second Ir-O-Ir shell (IrO2), respectively. Noted that the FT peak at 2.74 Å for the AD-HN-Ir electrocatalyst is obviously different from that of the Ir-Ir (Ir-O-Ir) coordination shell of Ir foil (IrO2), but the existence of Ir-Ir bond cannot be directly excluded. Therefore, the EXAFS spectrum for the AD-HN-Ir electrocatalyst under 1.45 V was fitted by considering of two subshells of first-shell of Ir-N/O bond and the first shell of Ir-Ir bonds. As shown in Figure N12e and N12f, the coordination number of the first shell of Ir-Ir coordination is 0.9, suggesting that Ir-Ir coordination is less than 10%.

31
Above results significantly prove that 90% Ir atoms were atomically dispersed on 3D carbon-nitrogen substrate with hetero-Ir-N4 moieties. Therefore, the FT curves after background removal proceeded at the Rbkg=1.1 Å for the AD-HN-Ir electrocatalyst were refitted only considering first-shell of Ir-N/O coordination, because a small amount (10%) of Ir-Ir bonds does not affect the conclusion of this manuscript.  Accordingly, the Figure N12 and Table N2 have been added in the revised Supplementary Information as Supplementary Fig. 28 and Supplementary Table 5. In line 11, page 7 of the revised manuscript, the following text has been added: "The peaks at higher R can be attributed to higher frequency noise and a small amount (10%) of Ir-Ir bonds because of the low Ir content for fluorescence XAFS measurement." g) Is a better background subtraction possible for the data in Fig. 5b? Can it be excluded that the difference in the analyzed peak height is due to the distortion that causes the peak at 1 A? The k-space data would help experts to identify issues with the data and judge if the analysis of the peak at 1.6 A is sound.

Reply:
We are greatly grateful to the reviewer for your nice question and careful inspection. In order to obtain a reliable EXAFS results, one of important matters is to perform background removal and data standardization for the EXAFS data analysis.
The background subtraction is to fit the smooth part with a polynomial by the least square method. The background function obtained by the fitting is easy to deduct the background part of slow change from μ(E). As we all known, the values of Rbkg and k-Sample Path N R (Å) σ (10 -3 Å 2 ) E 0 (eV) R-factor weight in the background subtraction will slightly affect the FT results. The best background removal was at the Rbkg=1.1 Å (Figure N12b), and the low frequency noise was removed fully but the high frequency noises are almost the same for the different Rbkg values. Furthermore, the difference of L-edge EXAFS k 2 (k) functions and the FT curves for the AD-HN-Ir electrocatalyst under different applied-potentials further prove dynamic evolution of Ir active sites during OER process ( Figure N13). improving the quality of this work. According to your suggestion, we performed the activity and stability of AD-HN-Ir electrocatalyst at higher densities such as 100 and 500 mA cm -2 . As shown in the Figure N14, the AD-HN-Ir electrocatalyst reaches a large current density of 100 mA cm -2 at a small overpotential of 292 mV. Intriguingly, at the high current density of 100 mA cm -2 , the AD-HN-Ir electrocatalyst still exhibits robust stability with a prominent retention of ~80% of the initial current density after continuous 50 h acidic OER operations. However, to reach the high current density of   c) It is good to perform measurements in a two-electrode cell but the one used by the authors is by no means close enough to a PEM electrolyzer system make statements about "practical electrolyzers"

AD-HN-Ir
Reply：We thank the reviewer for the constructive suggestion that is quite useful for improving the quality of this work. According to your suggestion, a PEM electrolyser system was selected to assess the practical application capability of the AD-HN-Ir To capture the industrial potential of the AD-HN-Ir electrocatalyst, we also carried out the stability of the AD-HN-Ir electrocatalyst at high current densities by a PEM electrolyser system. As shown in Figure N16, the 3D carbon electrode is bonded to the surface of the proton exchange membrane, while commercial Pt/C is sprayed on the other side of the proton exchange membrane as a cathode to form the membrane electrode ( Figure N16a). The PEM electrolyser system was tested under simulated industrial conditions (80 °C). It can be drawn from Figure Supplementary Fig. 23) Fig. S11).
Reply: Thank you for your nice questions. We really compliment the reviewer for your expert knowledge in the field of electrochemistry. There is a close relationship between double layer capacitance (Cdl) and potential window and sweep speeds of cyclic voltammetry (J. Phys. Energy 3, 034013 (2021)). According to your suggestion, the more suitable potential window (300 mV) and the wider range of sweep speeds (0.01-1.0 V/s) were selected to perform cyclic voltammetry (CV) tests. As shown in the Figure N16, the Cdl can be calculated as 5.2 mF form the cyclic voltammetry tests.
As is known, in order to obtain a reliable Cdl by the CV tests, it is necessary to select measuring windows larger than 100 mV not only to ensure reaching a steady charging current, but also to ensure that iC values extracted from the anodic and cathodic scans are similar (J. Phys. Energy 3, 034013 (2021)). In this work, we performed CV tests at a wide redox process-free window of 300 mV to endure stead charging current and similar anodic and cathodic scans. Furthermore, to achieve a small deviation and sufficient potential points, it is recommended to use a scan rate range as wide as possible to reach suitable current values. On the one hand, it needs to avoid too slow scanning speed to make the experiment unnecessarily time-consuming; on the other hand, it also needs to avoid too large scan rates that may lead to large deviations originating from potentiostat limitations. Therefore, in this work, the suitable sweep speeds of 0.01, 0.05, 0.1, 0.5 and 1.0 V/s were selected to perform CV tests. It can be seen from Figure N15, the Cdl of AD-HN-Ir electrocatalyst can be calculated as 5.2 mF by the CV tests, slightly larger than those of Ir-NC and IrO2 (4.1 and 3.7 mF, Figure N17 and N18). And the electrochemically active surface area (ECSA) of AD-HN-Ir electrocatalyst was obtained by roughness factor, and the specific activity of AD-HN-Ir electrocatalyst still surpasses those of Ir-NC and IrO2 when normalizing the current density to per ECSA ( Figure N19).  Reply: It is a nice question. As is known, the Tafel slope is an important parameter to evaluate the reaction kinetics of electrocatalysts (J. Electrochem. Soc. 160, 142 (2013)), which can be used to assess the electron transport capacity. In the previous manuscript, in order to obtain electron conductivity and electron transfer kinetics of the electrocatalysts, Tafel slopes were evaluated under density of 1-10 mA cm -2 . However, the calculated Tafel slope is not linear for the highly active catalyst at low current where the solid-liquid interface is controlled by the mixture of telephone transmission and substance diffusion. More accurately, the overpotentials interval with the current density of 10-100 mA cm -2 was re-selected to evaluate the Tafel slopes as shown in Figure N19.
Exactly, the trace of Tafel slopes usually remains linear to describe the kinetics of the electron transfer. It is noted that the reverse reaction cannot be ignored under low overpotential and will cause the trace to deviate from the linearity. The calculated Tafel slope is not linear for the highly active electrocatalyst at low current density (1~10 mA cm -2 ). Therefore, overpotentials interval with the current density of ~10~100 mA cm -2 was selected to evaluate the Tafel slopes. Tafel slopes in Figure N20  Accordingly, the Figure 3c in the previous manuscript has been replaced by the Figure N20 in the revised manuscript. In line 10, page 9 of the revised manuscript, the following text has been added: "Moreover, Figure 3c displays a smaller Tafel slope of 39 mV dec -1 , suggesting a faster OER kinetics and electron transfer occurred over Ir active sites." 6) It was not quite clear to me during the first read why the catalyst electrode is compared to PANI-Ir. Please revise the text with a statement that the catalyst Ir-NC is based on PANI-Ir but has an additional step in the preparation procedure.
Reply: This is a good question. As the reviewer said, the general process of synthesis for PAni-Ir is similar to that of the preparation of AD-HN-Ir electrocatalyst. However, it is noted that PAni was not treated with concentrated ammonia water for substrate surface functionalization, and the final electrocatalyst did not form hetero-nitrogen coordination. Therefore, it can be called Ir-NC eletrocatalyst as a reference sample.
According your suggestion, the "PAni-Ir-800C" in the previous manuscript has been replaced by "Ir-NC" in the revised manuscript. 7) P10. How can Rct be used to represent the H2O/OOH ion adsorption resistance? Rct relates to charge transfer across an interface. Does the statement make use of the SRIR results discussed later? The connection between Rct and "adsorption kinetics" is likewise unclear. Please revise. In the first half of P10, there is no data yet that OOH is created at all.

Reply:
Thanks for your nice question. The Rct is related to the kinetics of the interfacial charge transfer reaction. In accordance with the work of Harrington and Conway (Electrochem. Acta., 32, 1703(1987), Rct cannot be interpreted simply as the charge transfer resistances of the electro-adsorption and electro-desorption steps, but are attributed to the properties of more steps including kinetics of adsorption of reactive species in the overall reaction (J. Electrochem. Soc., 160, H142 (2013)). Recently, insitu EIS measurements were performed to track the evolution of the adsorbed OH * intermediates during OER by the Rct fitting (J. Am. Chem. Soc. 142, 12087 (2020)). In this work, we also performed in-situ EIS measurements to access to the kinetics of adsorption for oxygen-containing reactive species. The kinetics of reactive species adsorption on the electrode surface would significantly affect the kinetics of interfacial electron transfer, leading to significant changes in the fitted Rct. So the Rct can be used to represent the oxygen-containing reactive species ion adsorption kinetics.
According to your suggestion, to improve the logic of the description, the "H2O/ * OOH" in the previous manuscript have been replaced by "oxygen-containing reactive species" in the revised manuscript. The later in-situ SRIR results reveal that key * OOH intermediates were generated over Ir active sites, which further verify the results of in-situ EIS measurements.

Reply:
We are greatly grateful to the reviewer for your nice question and careful inspection. I agree with you that the description in the previous manuscript was not rigorous enough. For the sake of accuracy, the sentence of "Hence, this discovery of active sites evolution under working conditions can provide a coordination-engineered strategy for designing advanced acidic-OER electrocatalysts." in the previous manuscript has been removed in the revised manuscript.
As for the evolution of the active sites, in this work, the in-situ XAFS and SRIR reveal one oxygen atom coupled on the Ir active site with O-hetero-Ir-N4 under low potential of 1.25 V. Most importantly, the intensity of vibration absorption showed a positive correlation with the further increase of applied-potentials and remain unchanged when the potential exceeds 1.35 V. This revealed that the active O-hetero-Ir-N4 structure formed by oxygen coupling was the real active site, and this stable active structure can be maintained during the oxidation reaction to accelerate the OER reaction kinetics. Interestingly, when the applied voltage is removed for a period of time, the SRIR results show that the original absorption vibration peak of Ir-O disappears ( Figure   N21). This result suggests that the original dynamically-coupled O during the reaction has been desorbed when the OER is stopped. Above results reveal that the coupling of one oxygen atom on the Ir active site is a dynamic process with the change of voltage.  Accordingly, the Figure N21 has been added in the revised Supplementary Information as Supplementary Fig. 27. In line 14, page 13 of the revised manuscript, the following text has been added: "The dynamically-coupled O disappears after reaction by the SRIR results ( Supplementary Fig. 27), suggesting that the coupling of one oxygen atom on the Ir active site is a dynamic process with the change of potentials." 9) The panel labels a,b,c,… seem to be missing for most of the figures.

Reply:
We are greatly grateful to the reviewer for your nice question and careful inspection. We have examined the entire manuscript carefully and the missing panel labels have been added in the revised manuscript.
10) P17. How was the data smoothed? What method? What was the difference between neighboring steps? 2 cm-1? 100 points could smooth the spectra too much. Please also show the unprocessed data.

Reply:
We thank the reviewer for the constructive suggestion that is quite useful for improving the consistency of this work. The in-situ SRIR data were processed by OPUS 47 software. The curve of an electrolyte without voltage is used as background, and new peaks that appear in the curves after the background subtraction represent the absorption of new species vibration. To obtain a spectral curve with a high SNR, the curves of in-situ SRIR were smoothed through 10 points, meaning that 10 consecutive points are averaged to improve the SNR of the curves. Moreover, each high-resolution infrared absorption spectrum with resolution of 2 cm −1 was obtained by averaging 514 scans. It suggests that the difference between neighboring steps was ~2 cm -1 . To further clarify the differences before and after data smoothing, the original spectra without smoothing are presented in Figure N22. The SNR of the smoothed curve is improved to a certain extent, and the new peaks and change trend of the curves are consistent with the original curves, which means that the smoothed curves are reliable to prove the appearance of key new species under working conditions. Figure N22. In-situ SRIR measurements (a) with smoothing and (b) without smoothing in the range of 1300-600 cm -1 under various potentials for AD-HN-Ir electrocatalyst during the OER process.