A bio-inspired coordination polymer as outstanding water oxidation catalyst via second coordination sphere engineering

First-row transition metal-based catalysts have been developed for the oxygen evolution reaction (OER) during the past years, however, such catalysts typically operate at overpotentials (η) significantly above thermodynamic requirements. Here, we report an iron/nickel terephthalate coordination polymer on nickel form (NiFeCP/NF) as catalyst for OER, in which both coordinated and uncoordinated carboxylates were maintained after electrolysis. NiFeCP/NF exhibits outstanding electro-catalytic OER activity with a low overpotential of 188 mV at 10 mA cm−2 in 1.0 KOH, with a small Tafel slope and excellent stability. The pH-independent OER activity of NiFeCP/NF on the reversible hydrogen electrode scale suggests that a concerted proton-coupled electron transfer (c-PET) process is the rate-determining step (RDS) during water oxidation. Deuterium kinetic isotope effects, proton inventory studies and atom-proton-transfer measurements indicate that the uncoordinated carboxylates are serving as the proton transfer relays, with a similar function as amino acid residues in photosystem II (PSII), accelerating the proton-transfer rate.

1) The desire for an OER catalyst that operates with less than 200 mV overpotential at 10 mA/cm2 seems to be an arbitrary distinction. Why do the authors feel that 200 mV overpotential is the important cut off for an active system? Second, the authors' catalyst operates at 10 mA/cm2 at 189 mV. Many would argue that 187 mV is essentially the same as 200 mV…it is only 13 mV lower overpotential! The authors don't clarify in the main text if this is an average result or the "best" result, and they provide no standard deviations or any way for the readers to ascertain whether the 13 mV difference is statistically meaningful. Note that in the conclusions the authors report 188 mV overpotential, creating further ambiguity as to the standard deviation in the activity measurements. In addition, there are numerous NiFe-based systems on Ni foam that show comparable activity to the NiFeCP system (see for instance: Nat. Commun., 2015, 6, 6616.;J. Mater. Chem. A, 2016, 4, 13499-13508.;Chem. Mater., 2016, 28, 6934-6941;Small, 2018, 14, 1802204;J. Colloid Interface Sci., 2018, 523, 121-132;Yao et al., J. Power Sources, 2019, 424, 42-51; among dozens of others). I think the "outstanding electrocatalytic water oxidation activity" of the NiFeCP is the least important aspect of the paper, and should be put into better context by comparing to known reported systems. In addition, if one is to argue outstanding activity then one should make comparisons to other reported systems based also on mass activity or activity per surface area (see Chem Mater, 2017, 29, 120-140).
2) A smaller KIE for NiFeCP compared to NiFeLDH does not necessarily prove the existence of the secondary coordination sphere effect. Perhaps an important control experiment would be to conduct OER experiments with NiFeLDH and add terephthalic acid to the electrolyte and see whether the KIE decreases or remains constant (and also whether the activity increases or remains constant). If the KIE decreases and activity increases at high concentrations of terephthalic acid (where the masstransport controlled local concentration would be similar to that in the NiFeCP material), this would support the idea that free carboxylic acid groups can increase catalytic activity by stabilizing reactive intermediates.
3) Electrochemical proton inventory studies may give additional information about the number of hydrogenic sites involved in the rate determining step (Inorg. Chem., 2017, 56, 11254-11265;Nat. Commun., 10, 2019, 1683. For example, according to Scheme 2 NiFeLDH should have one hydrogenic site (O-H in water) with an isotope fractionation factor ~0.33 (based on the KIE = 3).
However, in the NiFeCP sample there should be two hydrogenic sites (O-H in water and O-H in terephthalic acid), and thus should show a non-linear dependence on deuterium oxide concentration. 4) There is a peak at 1.1-1.32 V vs RHE in Figure 1a that is pH-dependent. What is this peak? The authors should discuss this peak and its pH-dependent shifts.
5) The authors do not provide convincing evidence that the MOF structure is maintained during OER. The XRD shows degradation to an amorphous structure post OER and the FTIR shows de-coordination of terephthalate from the metal centers. Both of these results are consistent with many other precursor films reported in the literature that degrade to amorphous materials upon oxidation. Given that the structure during and post OER is amorphous and unknown, the proposed proton-transfer processes shown in Scheme 2 may be misleading because they invoke a well-ordered structure. 6) Given that the authors support their catalyst on a high-surface area Ni foam, I would strongly caution the authors from interpreting their NiO and Ni-OH from XPS as belonging to the NiFeCP material. There is no indication that these Ni species are not from the Ni foam, and therefore attempting to derive information of the catalyst structure from these peaks or their ratios (e.g. decoordination of carboxylate groups) may be an oversimplification of the system-although the NiFeCP is "scratched" from the Ni foam surface, there is no indication that some of the Ni foam itself was not scratched off as well. The authors may want to conduct an experiment using a planar, non-Ni based substrate (e.g. glassy carbon) to better quantify Ni oxidation states and ratios via XPS.
7) The authors should conduct electrolyses in D2O to confirm the Faradaic Efficiency for O2 production is the same as in the case of H2O. Figure 5, the authors switch between units of mA/cm2 and A/cm2. Also, 5a, b, and d should probably be on the same y-axis scale, and 5c, e, and f should be on the same scale for easier comparison unless there is a specific reason otherwise.

8) In
9) The authors report the optimal Ni-Fe ratio for the NiFeCP/NF sample and use this in the OER studies. The authors should consider including the data for the other Ni-Fe ratios in the SI. 10) In Fig. S3, the y-axis label is not consistent with the figure caption. The label reads as oxygen concentration, but is described as Faradaic efficiency in the caption. 11) In the Raman spectra (Fig. 4b), the peaks for uncoordinated carboxylate groups at 1631 and 1451 wavenumbers don't seem to be present in the pristine and post NiFeCP samples, but in the main text the authors state that they are observed before and after the OER. The authors should clarify their statement to be consistent with Sun et al report "A Bio-inspired Coordination Polymer, NiFeCP/NF, as Outstanding Water Oxidation Catalyst via Second Coordination Sphere Engineering". This catalytic system exhibits a low overpotential of 187 mV at 10 mA cm^-2 in 1.0 KOH, as well as a small Tafel slope and excellent stability. Interestingly, this excellent activity is related to the second sphere uncoordinated carboxylates which is serving as the proton transfer relays during the catalytic process. The design of this work is novel and give a simple but effective way to construct new material catalyst for water splitting. In addition, the second sphere uncoordinated carboxylates play the role as similar to the amino acid in PSII, this work also gives a new way to mimic the function of amino acid in the PSII. The catalyst used in this work was well characterized and the electrochemical performance was also well investigated. According to the comparison of different reported WOCs, this bio-inspired Coordination Polymer, NiFeCP/NF,gives an impressive catalytic performance. This is an elegant work in this field and I suggest this work to be accepted for publication on Nature Comm. The paper of Li et al. reports on an iron/nickel terephthalate coordination polymer deposited on nickel foam (NiFeCP/NF) to catalyze the anodic oxygen evolution reaction in alkaline media. Proposed in this work catalyst requires an overpotential of 188 mV at 10 mAcm-2 and exhibits a Tafel slope of 29 mV dec-1 and, thus, shows better performance compared to the chosen benchmark material NiFe LDH. However, NiFe-based catalysts with even lower Tafel slopes were reported in literature (e.g. catalyst with Tafel slope of 15 mV dec-1 in J. Phys. Chem. C 2008, 112, 3655-3666). Overall approach of incorporation of a catalyst in a polymer matrix can be interesting, but there are concerns regarding the stability of such polymer on the long run under the OER conditions. Polymers are typically unstable in the alkaline media, and their stability is drastically decreasing at high anodic potentials because the functional groups providing conductivity are prone to be attacked by the OH radicals. This challenge is very difficult to overcome and this is hindering development of membrane for alkaline water electrolysis in particular. There is no evidence in the current manuscript that the polymer matrix is stable during long term operation. Moreover, the HAADF-STEM images presented in Figure 2 suggest decrease in Carbon content after the electrolysis which indicates the degradation of the polymer. The data on electrochemical stability of the polymer itself is crucial, still it is not provided in the manuscript. The main question is what would be the gain in incorporating the catalyst into potentially unstable polymer matrix if already reported in the literature NiFe-based catalysts have comparable or even better performance? I doubt that without this important information the manuscript can influence the thinking in the field of electrocatalysis, as one would expect from the paper published in Nature Communications. Another important point that is missing in the manuscript is comparison of the electrochemically active surface area before the anodic scan and after it. From the XRD data presented in Figure S5, it becomes clear that the OER leads to amorphization of the catalytic surface, which will affect the ECSA. Considering that polarization curves presented in Fig. 2 are taken till relatively high current densities the amorphization is probably ongoing while the polarization curve is being recorded and related to it increase of ECSA may significantly affect the measured reactivity.
-The XPS fitting model does not really match the experimental curve and some of important features are not considered in Figure 3.It is well known that anodic polarization of Ni-Fe based materials leads to formation of oxyhydroxides. However, the peak corresponding to the OH groups is missing in the XPS fitting model of O 1s level presented in Figure 3b. This should be corrected, since hydroxy species are included in the spectra of Ni and Fe in Figure 3. Also the fitting model doesn't include contribution of several important features, e.g. fitted line doesn't include small shoulder at ca 529 eV in Figure 3b after OER. The additional component at 583 eV is missing in the fitting model of Ni spectrum.
Minor questions -what was the collection efficiency in GC measurements to estimate the efficiency of the OER? Ni and Fe are both stable towards dissolution under anodic polarization in the alkaline medium. What electrochemical process is responsible for the rest 4% of the current? Polymer oxidation/degradation?

Reviewer's Comments:
The authors report a NiFe coordination polymer material (NiFeCP) which when adsorbed on Ni foam electrodes shows promising activity and selectivity for the OER. The material oxidizes water at 187/188 mV overpotential at 10 mA/cm2 current density and shows constant activity over 17 h of polarization. The authors characterized the material beforeand-after OER tests with various techniques including XRD, HAADF-STEM, FTIR, TEM-EDS, and XPS. The authors compare this material to a NiFeLDH/NF material without a coordination polymer, and the NiFeCP/NF shows improved activity for OER. The authors show that NiFeLDH/NF has a KIE ~ 3 but NiFeCP/NF has a KIE ~2.1. Both KIE suggest a rate-determining proton transfer step, but the smaller KIE for NiFeCP/NF was used to suggest there is a secondary-coordination sphere effect from uncoordinated terephthalate groups in the polymer that help OER at the NiFeCP/NF material as shown in Scheme 2. This is a comprehensive study, and the inclusion of KIE studies is well considered. The activity of NiFeCP/NF for OER, while impressive, is not the main novelty of this paperthere are several materials deposited on Ni foam that show impressive comparable activity. Instead, the most interesting part of the manuscript was the postulated secondary coordination sphere effects that could be contributing to the enhanced activity of NiFeCP compared to NiFeLDH. Based on this alone, this manuscript has sufficient novelty and impact for publication in Nature Communications. However, the authors should address several comments below before publication.
Response: First, we sincerely thank this referee for your time, effort, and insights on our manuscript. In particular, the constructive comments and suggestions below are very helpful for us to revise and improve our manuscript. We have studied your comments carefully and performed some additional experiments accordingly to address your concerns.
Question 1: The desire for an OER catalyst that operates with less than 200 mV overpotential at 10 mA/cm 2 seems to be an arbitrary distinction. Why do the authors feel that 200 mV overpotential is the important cut off for an active system? Second, the authors' catalyst operates at 10 mA/cm 2 at 189 mV. Many would argue that 187 mV is essentially the same as 200 mV. It is only 13 mV lower overpotential! The authors don't clarify in the main text if this is an average result or the "best" result, and they provide no standard deviations or any way for the readers to ascertain whether the 13 mV difference is statistically meaningful. Note that in the conclusions the authors report 188 mV overpotential, creating further ambiguity as to the standard deviation in the activity measurements. In addition, there are numerous NiFe-based systems on Ni foam that show comparable activity to the NiFeCP system (see for instance: Nat. Commun., 2015, 6, 6616.; J. Mater. Chem. A, 2016, 4, 13499-13508.;Chem. Mater., 2016, 28, 6934-6941;Small, 2018, 14, 1802204;J. Colloid Interface Sci., 2018, 523, 121-132;Yao et al., J. Power Sources, 2019, 424, 42-51; among dozens of others). I think the "outstanding electrocatalytic water oxidation activity" of the NiFeCP is the least important aspect of the paper, and should be put into better context by comparing to known reported systems. In addition, if one is to argue outstanding activity then one should make comparisons to other reported systems based also on mass activity or activity per surface area (see Chem Mater, 2017, 29, 120-140).

Response:
We fully agree with this referee that "the most interesting part of the manuscript was the postulated secondary coordination sphere effects that could be contributing to the enhanced activity of NiFeCP compared to NiFeLDH"; and "outstanding electrocatalytic water oxidation activity of the NiFeCP is the least important aspect of the paper." Comparatively, our description in the original manuscript was suspicious for overselling the activity of NiFeCP/NF for OER. About the "200 mV overpotential is the important cut off", we have changed the description to make the statements more clear, all changes in the main text are marked in color. We have added more discussions in "Introduction" section to explain why a comprehensive understanding of the underlying mechanism of proton-coupled interfacial electron transfer process in the rate-determining step is vitally important. We have also clarified in the main text about the "best result" of our catalyst.

Question 2:
A smaller KIE for NiFeCP compared to NiFeLDH does not necessarily prove the existence of the secondary coordination sphere effect. Perhaps an important control experiment would be to conduct OER experiments with NiFeLDH and add terephthalic acid to the electrolyte and see whether the KIE decreases or remains constant (and also whether the activity increases or remains constant). If the KIE decreases and activity increases at high concentrations of terephthalic acid (where the mass-transport controlled local concentration would be similar to that in the NiFeCP material), this would support the idea that free carboxylic acid groups can increase catalytic activity by stabilizing reactive intermediates.  There is a peak at 1.1-1.32 V vs RHE in Figure 1a that is pH-dependent. What is this peak? The authors should discuss this peak and its pH-dependent shifts.

Response:
Response: The position of the redox peak in Figure 1a is indeed pH-dependent. As  The authors do not provide convincing evidence that the MOF structure is maintained during OER. The XRD shows degradation to an amorphous structure post OER and the FTIR shows de-coordination of terephthalate from the metal centers. Both of these results are consistent with many other precursor films reported in the literature that degrade to amorphous materials upon oxidation. Given that the structure during and post OER is amorphous and unknown, the proposed proton-transfer processes shown in Scheme 2 may be misleading because they invoke a well-ordered structure.

Response:
We fully agree with this referee. The schematic structure in Scheme 2 has been revised into non-ordered structure. We hope such modified illustration could avoid any misleading.

Question 6:
Given that the authors support their catalyst on a high-surface area Ni foam, I would strongly caution the authors from interpreting their NiO and Ni-OH from XPS as belonging to the NiFeCP material. There is no indication that these Ni species are not from the Ni foam, and therefore attempting to derive information of the catalyst structure from these peaks or their ratios (e.g. de-coordination of carboxylate groups) may be an oversimplification of the system-although the NiFeCP is "scratched" from the Ni foam surface, there is no indication that some of the Ni foam itself was not scratched off as well.
The authors may want to conduct an experiment using a planar, non-Ni based substrate (e.g. glassy carbon) to better quantify Ni oxidation states and ratios via XPS.
Response: According to the suggestions of this referee, NiFeCP on glassy carbon substrate (NiFeCP@GC) has been prepared, and the Ni oxidation states were measured by XPS. Without the influences of Ni foam, the ratio for the relative intensity of metalhydroxyl species and metal-oxygen bonds was found obviously increased after OER. The ratio of the integrated area associated with the Ni-OH/NiO peaks increased from 6:10 to 7.2:10 after electrolysis for the NiFeCP@GC sample, which is similar to the NiFeCP@NF sample. Figure S12. High-resolution XPS spectra of (a) C 1s, (b) O 1s, (c) Fe 2p, and (d) Ni 2p for particles detached by sonication from the as prepared NiFeCP/GC and NiFeCP/GC after 5 h OER test.
Question 7: The authors should conduct electrolyses in D2O to confirm the Faradaic Efficiency for O2 production is the same as in the case of H2O.

Response:
The Faradaic Efficiency for O2 production in D2O has been measured by electrolysis, the result is the same as in the case of H2O, see Figure S14. Question 8: In Figure 5, the authors switch between units of mA/cm2 and A/cm2. Also, 5a, b, and d should probably be on the same y-axis scale, and 5c, e, and f should be on the same scale for easier comparison unless there is a specific reason otherwise.

Response: Units and scales of figures have been unified in the manuscript.
Question 9: The authors report the optimal Ni-Fe ratio for the NiFeCP/NF sample and use this in the OER studies. The authors should consider including the data for the other Ni-Fe ratios in the SI.
Response: Data for the other Ni-Fe ratios have been added in SI as shown in Figure S2. Question 10: In Fig. S3, the y-axis label is not consistent with the figure caption. The label reads as oxygen concentration, but is described as Faradaic efficiency in the caption.
Response: This error has been corrected in the revised manuscript.

Sun et al report "A Bio-inspired Coordination Polymer, NiFeCP/NF, as Outstanding Water
Oxidation Catalyst via Second Coordination Sphere Engineering". This catalytic system exhibits a low overpotential of 187 mV at 10 mA cm -2 in 1.0 KOH, as well as a small Tafel slope and excellent stability. Interestingly, this excellent activity is related to the second sphere uncoordinated carboxylates which is serving as the proton transfer relays during the catalytic process. The design of this work is novel and give a simple but effective way to construct new material catalyst for water splitting. In addition, the second sphere uncoordinated carboxylates play the role as similar to the amino acid in PSII, this work also gives a new way to mimic the function of amino acid in the PSII.
The catalyst used in this work was well characterized and the electrochemical performance was also well investigated. According to the comparison of different reported WOCs, this bio-inspired Coordination Polymer, NiFeCP/NF, gives an impressive catalytic performance. This is an elegant work in this field and I suggest this work to be accepted for publication on Nature Comm.

Response to comments:
First of all, we sincerely appreciate your very positive comments on our work. Your constructive suggestions are very helpful for us to improve the quality of this work and we have revise our manuscript accordingly.  Summarizing the pH-independence OER activities, KIEs, proton inventory studies and atom proton transfer measurements, solid evidences have been provided now to prove that the uncoordinated carboxylate can serve as proton relay in NiFeCP.

Question 2:
Is it possible to tune the catalytic activity by tuning the pKa of uncoordinated carboxylate? Biophys. Acta Bioenerg. 2007Bioenerg. , 1767ChemCatChem 2010, 2, 724.). It suggests that, in artificial water oxidation catalysts, 'smart' removal of protons from the catalytic site may be also an issue when aiming at fast and efficient water oxidation. We believe that it is possible to tune the catalytic activity by tuning the pKa of additional uncoordinated ligands. This is an ongoing work and final results and discussions will be published in the future. Polymers are typically unstable in the alkaline media, and their stability is drastically decreasing at high anodic potentials because the functional groups providing conductivity are prone to be attacked by the OH radicals. This challenge is very difficult to overcome and this is hindering development of membrane for alkaline water electrolysis in particular.

Response
There is no evidence in the current manuscript that the polymer matrix is stable during long term operation. Moreover, the HAADF-STEM images presented in Figure  respectively. Because the proton has a much larger mass than the electron, proton transfer is considered to be much slower and which will control the reaction rate of a PCET reaction (J. Am. Chem. Soc. 2011, 133, 13224;Chem. Rev. 2010, 110, 6939).
Understanding of the underlying mechanism of proton-coupled interfacial electron transfer process in the rate-determining step for the current strategies of catalyst design is obviously lacking of this important part. Our work is exactly focusing on this puzzle, therefore, we believe that this work is important and suitable for Nature Communications.
The most important part of this work is the postulated secondary coordination sphere effects (uncoordinated carboxylates) that could be contributing to accelerate proton transfer in the rate-determining step and enhance the activity of OER catalysts.
We completely agree with you that some NiFe-based catalysts may display even more outstanding activities, but not every catalyst can be used as the reference to get important kinetic information. In our work, we claimed that NiFeCP shows better performance compared to the benchmark material NiFe LDH, because of the much lower overpotential and the much higher ECSA normalized current density, not just due to the We are very sorry to make you confused due to the unclear description in the original manuscript. We have made a clearer description in the revised supportting information as following: "Before water oxidation measurements, the as prepared NiFeCP/NF electrodes were activated at 50 mA cm −2 current density in 1.0 M KOH for 10 mins to remove the excessive terephthalates in NiFeCP introduced by the fast electrochemical deposition process. The activated NiFeCP/NF electrodes were rinsed with water, dried in air, and then water oxidation activities were measured the in fresh electrolytes." About "the HAADF-STEM images presented in Figure 2 suggest decrease in Carbon content" you mentioned. First, as long as carbon content in the NiFeCP particle could be observed after OER on HAADF-STEM images, together with FTIR, Raman and XPS, which can prove that carboxylates still existed in NiFeCP. Second, in our supporting information, we have described the measurement of TEM, samples "were added dropwise onto a carbon-coated copper grid". Because of that the carbon-coated copper grid is the substrate for TEM measurement, which will affect the quantitative analysis, therefore, we didn't discuss the quantity changes of carbon content from HAADF-STEM.

Question 2
Another important point that is missing in the manuscript is comparison of the electrochemically active surface area before the anodic scan and after it. From the XRD data presented in Figure S5, it becomes clear that the OER leads to amorphization of the catalytic surface, which will affect the ECSA. Considering that polarization curves presented in Fig. 2 are taken till relatively high current densities the amorphization is probably ongoing while the polarization curve is being recorded and related to it increase of ECSA may significantly affect the measured reactivity.
Response: According to you suggestions, the ECSA after OER, has been measured, as shown in Figures S3 and S5. The related discussions have been added to the revised manuscript. Figure S3 and S5. Cyclic voltammetry (CV) curves of (a) NiFeCP/NF after OER and (c) activated NiFeCP/NF in 1 M KOH with different scan rates at selected potential range; (b) and (d) the corresponding capacitance Δj (|jcharge−jdischarge|) versus the scan rates.

Question 3
The XPS fitting model does not really match the experimental curve and some of important features are not considered in Figure 3.It is well known that anodic polarization of Ni-Fe based materials leads to formation of oxyhydroxides. However, the peak corresponding to the OH groups is missing in the XPS fitting model of O 1s level presented in Figure 3b. This should be corrected, since hydroxy species are included in the spectra of Ni and Fe in Figure 3. Also the fitting model doesn't include contribution of

Question 4
What was the collection efficiency in GC measurements to estimate the efficiency of the OER?
Ni and Fe are both stable towards dissolution under anodic polarization in the alkaline medium.
What electrochemical process is responsible for the rest 4% of the current? Polymer oxidation/degradation?

Response:
We have re-measured the Faradaic Efficiency of NiFeCP in a small volume electrolysis cell, the collection efficiencies are 98.4±0.6%. The Faradaic Efficiency of NiFeLDH was also measured, a Faradaic Efficiency of 97.8±1.7% has been obtained.