Unraveling the mechanism for paired electrocatalysis of organics with water as a feedstock

Paired electroreduction and electrooxidation of organics with water as a feedstock to produce value-added chemicals is meaningful. A comprehensive understanding of reaction mechanism is critical for the catalyst design and relative area development. Here, we have systematically studied the mechanism of the paired electroreduction and electrooxidation of organics on Fe-Mo-based phosphide heterojunctions. It is shown that active H* species for organic electroreduction originate from water. As for organic electrooxidation, among various oxygen species (OH*, OOH*, and O*), OH* free radicals derived from the first step of water dissociation are identified as active species. Furthermore, explicit reaction pathways and their paired advantages are proposed based on theoretical calculations. The paired electrolyzer powered by a solar cell shows a low voltage of 1.594 V at 100 mA cm−2, faradaic efficiency of ≥99%, and remarkable cycle stability. This work provides a guide for sustainable synthesis of various value-added chemicals via paired electrocatalysis.


Reviewer #1
In the paper, Yang et al. reported Fe-Mo-based phosphide heterojunctions for both 4-NBA reduction and HMF oxidation. The catalysts were well-characterized and reaction pathways were well-investigated. However, these reactions on non-noble catalysts and their reaction pathways have been well-reported in the literature, such as, ACS Sustainable Chemistry & Engineering 9.5 (2021): 1970-1993, CCS Chem. 7, 507-515 (2020, Angew. Chem. Int. Ed. 58, 9155-9159 (2019). Considering the novelty of this work, it is hard to be accepted by Nature Communications. Other comments are shown below. Response: We greatly appreciate the reviewer for his/or her tireless review of our paper and thank the reviewer for the recognition on the characterization of our catalysts and the study of the reaction pathways. The novelty and significance of our work are further elucidated as follows: About three literature as reviewer mentioned. We have read carefully and two of them were cited in our original manuscript. First two works (ACS Sustainable Chemistry & Engineering 9.5 (2021): 1970-1993CCS Chem. 7, 507-515 (2020)) focus on the study of the 5-HMF EOR or 4-NBA EER half-reactions.While a feature of our work is the high-efficiency overall paired 5-HMF EOR and 4-NBA EER, which differentiates with above two. The third one (Angew. Chem. Int. Ed. 58, 9155-9159 (2019)) preliminarily investigated the paired electrocatalysis with NiBx as both anode and cathode. However, the same material often exhibits difference in activity for two electrodes, resulting in a compromising performance for assembling overall electrolyzer. Particularly, the catalytic reaction mechanisms are not elucidated in these works. To expedite the advance of this nascent realm, pursuing more efficient electrocatalysts and clarifying the catalytic mechanism is highly important and meaningful.
In our work, to achieve synchronous and efficient paired electrocatalysis of 5-HMF EOR and 4-NBA EER, we have reasonably designed the matched electrocatalysts, where FeP-MoP supported on Fe foam (FF) for cathodic 4-NBA EER by learning from the biological [Fe-Mo]-nitrogenase and a similar material containing Ni (FeP-NiMoP2/FNF) for catalyzing anodic 5-HMF EOR. The similar components of two electrodes are convenient for the industrial catalyst integration. Meanwhile, the introduction of Ni accelerates the reaction kinetics of anodic 5-HMF EOR, making the reaction kinetics of 4-NBA EER and 5-HMF EOR more matchable. The matchable design in these aspects contributes greatly to the impressive performance of our paired electrolyzer with low voltage expense, high faradaic efficiency and remarkable cycle stability. It reflects our rational and intellectual thinking in catalyst design for satisfying the highly efficient paired electrolysis of organic chemistry.
More importantly, we have systematically investigated the reaction mechanisms of the paired system, including identifying the active species and uncovering the origin of kinetic matchablity for 4-NBA EER and 5-HMF EOR, which has rarely been touched in previous reports. We corroborated water as sole H and O source rather than organics or dissolved oxygen, and clarified the specific water activation and dissociation and hydrogen/oxygen participation patterns. For anodic electrooxidation. the OH* species stemmed from the first step of water decomposition are determined as the active species. The first step of water decomposition requires the smallest energy compared with the subsequent three steps that produce OOH*, O* and O2, thus avoiding the maximum energy barriers of water splitting. Meanwhile, for cathodic electroreduction, active H* species also come from the first step of water decomposition. Both the H and O species from the first step of water decomposition allow to the maximum matching of subsequent paired cathodic 4-NBA EER and anodic 5-HMF EOR with lowest energy consumption. Besides, water as sole H and O source for the paired system is a sustainable and recyclable route. Because the valuable target products can be continuously and efficiently produced by the addition of the reactants at two-electrode chambers without producing side products. Otherwise, if H or O species come from organics, such as H provided by hydroxyl groups of orgaincs or dehydrogenation of anodic oxidation and O originated from deoxygenation of cathodic reduction, other reaction pathways occur and side reactions happen, resulting in reducing the selectivity and faradaic efficiency of target products. Or if dissolved oxygen from air as O source is also not sustainable due to the small amount. Above reaction mechanistic studies are greatly significant, which can not only guide for the development of this field and catalyst design, but also greatly extend to development and application of other fields with water as a feedstock.
Overall, the rational designability of our paired electrocatalysis and the deep mechanistic studies make that our work is novel and meaningful and worthy to be published in Nature Communications.

Q1.
Line 11, the authors mentioned "active H* species for organic electroreduction originate from water." Why did they want to highlight this in the abstract? Is there any other possible H source in the aqueous solution? Response: We appreciate the reviewer for the comment. Actually, besides the H source for water, organics can probably provide the H source as described above, such as the hydroxyl groups of 4-NBA (J. Am. Chem. Soc. 138, 10128-10131 (2016))) or aldehyde dehydrogenation of anodic 5-HMF (Nat Catal. 5, 66-73 (2022)). By studying reaction mechanisms, we have verified that active H* species for organic electroreduction come from water. And only water as H (or O) source for the paired system is a sustainable and recyclable route. The relative discussion has been added in our revised manuscript (Paper 17-18, Line 423-432).
Q2. Line 28-29, "both hydrogen and oxygen evolution reactions (HER and OER) are slow kinetics". We usually agree on that only OER is kinetically sluggish, which is major barrier to water electrolysis. Response: We thank the reviewer for the comment. It may be that we did not describe it accurately. Comparatively, OER is more sluggish than HER owing to the complex four proton-coupled electron transfer as reviewer said. Following this comment, we have rewritten this content as follow (Page 2, Line 28-31): "However, both hydrogen and oxygen evolution reactions (HER and OER) are inefficient due to the high activation barriers, and particularly the sluggish kinetic of complex OER process hinders the water electrolysis efficiency." Q3. Line 102-103, how did the authors test the nanosheet thickness by TEM? Why not just using the cross-section SEM to measure the thickness? Response: We appreciate the reviewer for the constructive comment. Following this suggestion, we have carried out the cross-section SEM characterization to give the information of the nanosheet thickness. As shown, the FeP-MoP nanosheets with~50 nm in thickness are clearly observed (inset of Fig. 1b). The relative content has been added in our revised manuscript (Paper 4, Line 106-109).

Q4.
Line 136-137, should it be called magnetic FeNi3 or alloy FeNi3? It is not the metallic, right? Response: Following the reviewer's suggestion, we have revised it as alloy FeNi3 in our manuscript (Paper 6, Line 135).
Q6. Line 174-176, "where FeP-MoP/FF exhibits a better 4-NBA ERR activity than FeP-NiMoP2/FNF ( Supplementary Fig. 9)." What could be the reason to explain FeP-MoP/FF exhibits higher activity? Is it because of the higher surface area or any other possibilities? Did the authors test electrochemical active surface area (ECSA)? Response: We thank the reviewer for the comment. Following this suggestion, the ECSA of FeP-MoP/FF and FeP-NiMoP2/FNF were calculated ( Supplementary Fig.  9c). FeP-MoP/FF shows the higher ECSA than FeP-NiMoP2/FNF, indicating that the higher ECSA can contribute to part of good activity of FeP-MoP/FF but not all. More importantly, Fe site is identified as the active center for 4-NBA ERR. Electron accumulation on Fe active site facilitates to attract more protons to reduce -NO2 into -NH2. However, when introducing Ni, it traps electrons from Fe, verified by XPS result, resulting into the decrease of electrons on Fe site, which is not conducive for the electroreduction of 4-NBA. This is the intrinsic origin that the 4-NBA ERR activity of FeP-NiMoP2/FNF is inferior to that of FeP-MoP/FF. Furthermore, the adsorption energy of 4-NBA (E4-NBA) of FeP-MoP and FeP-NiMoP2 was also calculated and compared (Supplementary Table 8 Q7. Figure 3b-c, Figure 3h, and line 197-199. The authors tested the LSV and ECSA among three catalysts. It seems that the higher activity of FeP-MoP/FF than FeP/FF and MoP/FF is mainly originated from the higher surface area of FeP-MoP/FF? What is the ECSA-normalized reaction activity (or current density)? Response: We thank the reviewer for the constructive comments. Following this comment, we have supplied the ECSA-normalized activity of FeP-MoP/FF, FeP/FF, and MoP/FF catalysts ( Supplementary Fig. 13). As shown, FeP-MoP/FF delivers higher ECSA-normalized activity compared to the FeP/FF and MoP/FF, further suggesting the excellent intrinsic 4-NBA ERR activity of FeP-MoP heterojunction. Actually, apart from the high surface area, other merits of FeP-MoP/FF, such as highly active heterointerface and rapid electronic conductivity also contribute the enhanced activity. The excellent activity of FeP-MoP/FF is the result of comprehensive effects. The relative content has been added in our revised manuscript (Paper 9, Line 207-209, SI: Paper 14, Line 115-120). Response: We thank the reviewer for the comment. Yes, the potential of 0.004 V is versus the RHE potential. In our electrochemical test, the Hg/HgO electrode was used as reference electrode. Then the potentials were quoted with respect to the reversible hydrogen electrode through ERHE = EHgO/Hg + 0.059 × pH + 0.098 V. The reason to select the potential of 0.004 V vs. RHE is to avoid the HER interference (Since HER initiates after this value and it overlaps with 4-NBA ERR). The positive potential applied for electroreduction is commonly reported in the literature (e.g. ACS Catal. 11, 13510-13518 (2021)). Above content has been added in our revised manuscript (Paper 8, Line 178-180).
Response: According to this constructive suggestion, the ECSA-normalized activity of catalysts for 5-HMF EOR are provided ( Supplementary Fig. 22). Similarly, FeP-NiMoP2/FNF also delivers higher activity compared to the FeP-MoP/FF, NiMoP2/FNF and FeP/FF, further suggesting the superb intrinsic 5-HMF EOR activity of FeP-NiMoP2/FNF heterojunction. The relative content has been added in our revised manuscript (Paper 10, Line 251-252 and Paper 11, Line 261-262 SI: Paper 23, Line 204-208). indicates that Ni has more stronger capturing electron ability than Fe, which not only captures electrons from Mo, but also captures electrons from Fe. Therefore, the XPS and DFT results are consistent. This is also the reason that we design the introduce of Ni in Fe-Mo-based composite, which can improve the reaction kinetics and activity of 5-HMF EOR. The relative discussion has been added in our revised manuscript (Paper 7, Line 171-175).

Q11.
Line 367-379, the authors used isotopic labelling and confirmed the proton and oxygen source are both from water. Are there any other possibilities of proton/oxygen sources? Based on previous literatures and the high conversion and selectivity in this work, the H/O source would not be coming from the contaminations, it is obviously from H2O to provide H/O source.
Response: Thank the reviewer very much for the comment. For cathodic reaction, besides the H source from water, H source may come from the hydroxyl groups of 4-NBA and aldehyde dehydrogenation of anodic 5-HMF as responded in Q1. With respect to anodic reaction, oxygen source may come from the deoxygenation of 4-NBA cathodic reduction, including the nitro deoxygenation and hydroxyl deoxygenation (ACS Catal. 9, 8068-8072 (2019);Energy Environ. Sci. 13, 917-927 (2020)) or dissolved oxygen originated from air (Green Chem. 14, 143-147 (2012)) in addition to that from water. Just due to the H/O source from water for the paired electrocataylsis of organics, no other side products producing, so such high conversion and selectivity of catalysts were achieved. Thus the systematical and deep investigation of the catalytic reaction mechanisms, including confirming the proton and oxygen source by using various means, such as isotopic labelling, is very important and meaningful. The relative content has been added in our revised manuscript (Paper 17-18, Line 423-432).
Q12. HMF is not stable in strong alkaline solutions, e.g., 1 M KOH, ACS Catal. 2018, 8, 2, 1197-1206. Did the authors observe the degradation of HMF or Cannizzaro reaction in the electrolyte? Response: We appreciate the reviewer for the valuable comment. The Cannizzaro reaction in the alkaline solution is an endothermic reaction. The reaction rate is very slow at low or normal temperature and it takes a long time. The same is for degradation reaction. We further performed some additional experiments to verify this point ( Supplementary Fig. 18). As shown, there is no side products and the HMF concentration change in 1.0 M KOH solution (pH 14) containing 10 mM HMF at the temperature of 298K (test conditions in our experiment) for a long time (24 h) with no adding catalysts, suggesting that HMF is stable in strong alkaline solution without the degradation of HMF or Cannizzaro reaction. when adding our efficient catalyst, the time of converting HMF (10 m M) is very short (<2 h), so none of these reactions occur in our experiment. When increasing the concentration of HMF to 100 mM, the time of converting HMF (100 m M) is relative short (<5 h) by using our catalyst, along with high conversion and selectivity (≥99% and ≥98.0%). During this period still no the degradation of HMF or Cannizzaro reaction was observed. Additionally, considering industrial scale-up application, a large volumes of more higher concentrated HMF solutions (>0.5M) are usually used and HMF remains in solution for extended periods of time. There may occur the the degradation of HMF or Cannizzaro reaction in strong alkaline solutions (pH 14) as reviewer said and the literature reported (ACS Catal. 8, 2, 1197-1206(2018). So the pH value of 14 may not be a viable condition for industrial processes. We also notice this point. In our work, weak alkaline or even neutral systems (pH= 11.63, 10.02, 8.31 and 6.86) were explored. FeP-NiMoP2/FNF shows excellent activities in a wide pH range for HMF EOR, implying the good universality of our catalyst. Our further studies will pay more attention to the exploration of this reaction in weak alkaline or neutral systems. The relative content has been added in our revised manuscript (Paper 10, Line 235-236 SI: Paper 19, Line 168-172).

Q4.
All the potentials normalized to RHE may be inaccurate because the pH value often vary during the water-involving organic reaction. Generally, all the electrochemical data vs the actual reference electrode should be provided in the text (see the ref. Nature Communications 2021, 12, 3881). So, the data vs Hg/HgO are suggested in the manuscript, or the potential vs Hg/HgO electrode should be add to Fig. 3a, the fist electrochemical figure.
Response: We thank the reviewer for the valuable suggestion. The potential vs Hg/HgO electrode was added on top axis in Fig. 3a and other electrochemical figures and the highly related work (Nat. Commun. 12, 3881 (2021)) had been added as Ref.
43 in our manuscript.
. Q5. In Fig. 5a, how is the voltage of the two-electrode electrolytic system converted into the relative RHE potential? Here I think it is the cell voltage. "Potential (V vs. RHE) " should be correct to be "Cell voltage (V vs. counter electrode)", "E (V vs. counter electrode)" or "E (V)". Response: We sincerely apologize for this error. It is the "E (V)", which has been revised in our manuscript. Thanks again.

Reviewer #3
Yang and coworkers presented a family of electrocatalysts that reduce/oxidize organic molecules with very high faradaic efficiencies, compared to the background hydrogen and oxygen evolution reactions (HER/OER). FeP-MoP supported on Fe foam (FF) and a similar material containing Ni (FeP-NiMoP2/FNF) were the best performers for reduction of 4-NBA and oxidation of 5-HMF respectively. The combined oxidation/reduction of such organic molecules demonstrate an appealing path to be further explored. Before publication, however, there are blind spots that need clarification. Response: We genuinely appreciate the reviewer for very positive evaluation of our work and thank the reviewer for constructive comments, which help to improve the quality of our article. The questions raised by the reviewer have been fully addressed.

Q1.
The main one I currently observe is related to the models in Suppl Fig 26, specifically FeP-NiMoP2 (e). It seems that the authors produced it from that of FeP-MoP (d). The specific details about how the authors produced both of them should be dully explained. For instance, it seems that they commensurated FeP and MoP in a 3:2 ratio (Figure 1d), but that is not said anywhere (eg, lines 107-108 and 120 do not say that). Same with NiMoP2 ( Figure 1h). Response: We thank the reviewer for the valuable comment. Actually, the FeP-NiMoP2 model (Suppl Fig. 26e) was built based on single FeP (Suppl Fig. 26a) and NiMoP2 (Suppl Fig. 26c) phases, instead of that of FeP-MoP. Additionally, the crystal planes of theoretical modeling for NiMoP2, FeP and MoP are based on the exposed planes of their TEM results, where FeP and MoP or NiMoP2 were matched well when the FeP and MoP or NiMoP2 were commensurated in a 3:2 ratio. Therefore, the 3×2 FeP supercell, 2×2 MoP and 2×2 NiMoP2 were used as crystal models for facet cleavage and construction of heterojunctions and a four-layer of slab supercell was chosen as the surface slab supercell to decrease the complexity of calculation. Following this constructive suggestions, we have added the specific description about the model establishment of FeP-NiMoP2 and FeP-MoP in the manuscript (Page 22, Line 543-554).

Q2.
Can these surfaces reconstruct upon cleavage? NiMoP2 (001) and MoP(100). Response: We thank the reviewer for the excellent comment. Generally, when cutting the crystal, it is metastable (due to high energy) relative to the bulk phase, surface reconstructions (or defects) are produced to reduce the energy. Due to the complexities of surface reconstructions, including numerous configurations, fully investigating this effect is difficult for both experimentally and theoretically. The premise that surface reconstruction affects the heterostructure is the heterojunction synthesized by a two-step method, which one phase is synthesized firstly, and then another phase is synthesized on the basis of the previous phase and the surfaces of both two phases will reconstruct to form a heterojunction. Our FeP-MoP and FeP-NiMoP2 heterojunctions are synthesized by one-pot method. The Fe-Mo oxide or Fe-NiMo oxide precursors are fabricated by synchronously adding the metal sources, in which the Mo, Ni and Fe metals in the formed precursors are fully coordinated and have the lowest free energy. By further converting into phosphide heterojunctions, it does not involve the issue of surface reconstructions of two surfaces.

Q3.
Regarding the elemental map of FeP-NiMoP2/FNF. It seems that P does not densify in the "core" part that is rich in Mo and Ni. This may suggest to the reader that the core structure is more of a NiMox alloy rather than a NiMoP2 structure. Response: We thank the reviewer for the comment. Based on XRD (Fig. 2b), there is the existence of NiMoP2 phase, but no the phase of NiMox alloy in the heterojunction, indicating the structure of the composite is NiMo phosphide rather than NiMox alloy. The relative low density of P can be due to low amount of P than Mo and Ni, which is verified by XPS results (Supplementary Table 7). In addition, generally, in the elemental mapping images, the elements with larger atomic numbers, such as Mo, Ni, looked brighter than C, P, and Si that have smaller atomic numbers (Nanoscale, 12, 16586-16595 (2020)). This may be an other reason that P looks like low dense than Mo and Ni.