Bimetallic nickel-molybdenum/tungsten nanoalloys for high-efficiency hydrogen oxidation catalysis in alkaline electrolytes

Hydroxide exchange membrane fuel cells offer possibility of adopting platinum-group-metal-free catalysts to negotiate sluggish oxygen reduction reaction. Unfortunately, the ultrafast hydrogen oxidation reaction (HOR) on platinum decreases at least two orders of magnitude by switching the electrolytes from acid to base, causing high platinum-group-metal loadings. Here we show that a nickel-molybdenum nanoalloy with tetragonal MoNi4 phase can catalyze the HOR efficiently in alkaline electrolytes. The catalyst exhibits a high apparent exchange current density of 3.41 milliamperes per square centimeter and operates very stable, which is 1.4 times higher than that of state-of-the-art Pt/C catalyst. With this catalyst, we further demonstrate the capability to tolerate carbon monoxide poisoning. Marked HOR activity was also observed on similarly designed WNi4 catalyst. We attribute this remarkable HOR reactivity to an alloy effect that enables optimum adsorption of hydrogen on nickel and hydroxyl on molybdenum (tungsten), which synergistically promotes the Volmer reaction.


The major issue of Ni as a HOR catalyst in alkaline is that Ni is full passivated within the HOR potential region and thus has no affinity toward H 2 , which is reflected by the negligible limiting HOR current density of Ni as shown in
. Therefore, the major merit of the MNi 4 lies in the high limiting current density of ~3 mA/cm 2 at 2,500 rpm. The authors shall explain why the Ni remained unpassiviated, or the current is delivered by Mo(W) (I don't think so)? and shall at least hypothesize what are the active sites. The XAS results actually provide some hints. As seen in Figure  2a and 2b, the Ni in MoNi 4 and WNi 4 remains largely at reduced phase ex situ exposed to air. On the other hand, the surface of Ni nanoparticles is passivated ex situ. Response: We appreciate the reviewer for his/her thoughtful comments and questions. The important contribution of this work comes from our findings that MoNi 4 and WNi 4 nanoalloys can catalyze the alkaline HOR comparable to Pt/C catalyst, whereas almost all previous PGM-free catalysts are readily deactivated owing to surface oxidation. We attribute the outstanding activity and stability (limiting current density of ~3 mA cm -2 at 2500 rpm up to 0.2 V versus RHE) to the following reasons: First, the alloy effect. Our single Ni sample undergoes expected surface oxidation during alkaline HOR, thus showing little HOR reactivity. When alloying with Mo and W to form MoNi 4 and WNi 4 alloys, remarkable HOR performances achieved. Prior research has demonstrated that alloyed structures commonly possess superior resistance to passivation as compared to single metals. For examples, Ni-Mo and Ni-W alloys have been observed to show much better passivation resistance than that of microcrystalline Ni in alkaline conditions (J. Appl. Electrochem. 2004, 34, 1085-1091Corros. Sci. 2011, 53, 1066-1071, because alloys usually have stronger N−M bonds than both the N−N and M−M bonds (Chem. Rev. 2016, 116, 10414-10472), which thus enable superior stability. Second, the optimized H ad adsorption efficiently consumes OH ad on the catalyst surface. Our computational results uncover that the HBEs are properly weakened on our MoNi 4 and WNi 4 alloys as compared to single Ni. As a consequence, hydrogen can dissociate and desorb rapidly from the Ni sites of MoNi 4 or WNi 4 alloys, while such adsorption on single Ni metal is too strong. It is well known that alkaline HOR on the catalyst surface involves the Volmer step that H ad and OH ad react to generate H 2 O (Sci. Adv. 2016, 2, e1501602). Thus, we can image that the optimized H ad adsorption on the surfaces of MoNi 4 or WNi 4 alloys would react and consume the absorbed OH ad efficiently, which significantly prevents the surface passivation of alloys under alkaline condition. Third, we performed additional XPS studies of our MoNi 4 or WNi 4 alloys and compared them to that of single Ni. As shown in Figure R1, the Ni 2p XPS shows two peaks at 852.6 eV and 855.7 eV, which can be assigned to Ni 0 and Ni 2+ , respectively (Nat. Commun. 2017, 8, 15437). Our XPS analyses clearly reveal that single Ni undergoes more severe surface oxidation as compared to the MoNi 4 or WNi 4 alloys, consistent with XAS observations. Furthermore, in our revised manuscript, the UPS and CO-stripping experiments, in conjunction with DFT calculations, reveal the alloy effect that allows for optimum adsorption of hydrogen on Ni sites and hydroxyl on Mo(W) sites, which substantially boost the rate-limiting Volmer step, leading to the observed HOR performance of the MoNi 4 and WNi 4 catalysts.
Overall, we surmise that the alloy effect not only offers robust metallic bonds, but also weakens the HBE that accelerates the removal of surface OH ad species through the Volmer step. These together leads to superior surface resistance to passivation of our designed alloyed catalysts and their outstanding reactivity in alkaline environment.  Figure 5c and Supplementary Fig. 38). On the basis of these results, we conclude that Ni sites on the MoNi 4 and WNi 4 catalysts are appropriate active sites for the adsorption of H ad species.
Recent experimental and computational studies further uncover that hydroxyl bonding energy (OHBE) should also be the activity descriptor for HOR catalysis. Importantly, the sluggish HOR on Pt catalyst in alkaline electrolytes was attributed to its weak OH ad binding. In this manuscript, our CO-stripping experiments clearly exhibit the enhanced OHBE on MoNi 4 and WNi 4 alloys as compared to Pt/C catalyst (Fig. 5b in the MS). Besides, the DFT results show that Mo and W sites have stronger OH absorption than the Ni sites (see Figure 5c and Supplementary Fig. 39). These results thus indicate that OH ad species are prefer to bind on Mo and W sites.
In light of above analyses, we reasonably come to a bi-functional mechanism (Angew. Chem. Int. Ed. 2017, 56, 15594) for our MoNi 4 and WNi 4 catalysts: Ni sites on the alloyed catalysts offer the optimum hydrogen adsorption, while Mo and W sites on the alloys enable suitable adsorption of OH species ( Figure R2a). As a result, the optimized HBE enabled by Ni and the optimized OHBE enabled by Mo or W substantially boost the rate-limiting Volmer step (Heyrovsky-Volmer pathway), leading to the marked HOR performance of our MoNi 4 and WNi 4 catalysts.
To further verify this mechanism, we calculated the free energy for Volmer step of the studied catalysts. The much lower energy barriers ( Figure R2b) of MoNi 4 and WNi 4 catalysts further evidence the promoted Volmer step thus the improved HOR energetics.  Response: We thank the reviewer for the thoughtful comments and questions. We want to address your doubts and questions from the following parts: First, by combing the FT curves and EXAFS wavelet transform at Mo K edges, we have demonstrated the formation of Mo-Ni alloy, where the predominant peaks located at 2.2 Å is ascribed to scattering from neighboring Ni atoms present at the shorter distance than Mo-Mo distance (see Figure 2d). Following your comments, we further fitted the EXAFS data at the Mo edge, as shown in Figure R4 below. Our fitting results show that the obtained first-shell Mo-Ni coordination number is about 6.9, which is smaller than the coordination number of 12 for the standard MoNi 4 crystal (Note: the coordination number for MoNi 4 crystal is 12 but not 4). The obtained smaller coordination number could be resulted from the nanoscale size and rich surface steps of our porous MoNi 4 nanoalloy. This observation is analogous to a recent report (Science 2020, 367, 777-781), where the NiMo alloys similarly have weak intensity for the peak at 2.2 Å. Figure R4. The raw and fitting Fourier transform of k 3 -weighted EXAFS spectra of Mo K-edge for MoNi 4 .
Second, we have to note that Mo(W) in Ni alloy are difficult to distinguish by the Ni K edge XAS.
Such dilemma has also been demonstrated previously (Science 2020, 367, 777-781 and Sci. Adv. 3, e1603068), which means that it is hard to differentiate the Ni-Ni and Ni-Mo(W) bonds from Ni K edge XAS. Owing to this reason, we used Ni-Ni(Mo) or Ni-Ni(W) in our original manuscript. Directly estimating Ni-Ni(Mo) bond distance from the XAS spectra is inaccurate. For example, a recent work (Science 2020, 367, 777-781) on MoNi alloy (atomic ratio=1: 3.49; close to MoNi 4 ) displays a similar position for Ni-Ni(Mo) peaks between NiMo alloy and Ni foil from the Ni K edge results (see Figure R5). Nevertheless, an accurate estimate of bond distance can be realized by the EXAFS fitting. Our EXAFS fitting results show a bigger Ni-Ni(Mo/W) bond distance of alloys than that of Ni metal and Ni foil (see Supplementary Table 2). Last, we highlight that the XRD pattern of our MoNi 4 alloy can be perfectly assigned to the tetragonal MoNi 4 (JCPDS 65-5480) phase (slightly shifting to lower degree as compared to pure Ni). More importantly, our ICP-AES results confirm a Mo:Ni atomic ratio of ~1:4 for 10 independent measurements ( Figure R6). Additionally, the Mo-Ni alloy phase diagrams ( Figure R7) reveal that new MoNi 4 phase forms when the Ni:(Mo+Ni) ratio is 0.8 (the ratio is the same with our MoNi 4 alloy) and the temperature is below 1100 K (see the Red Star in Figure R7).  Together, our above results and analyses could reasonably verify tetragonal MoNi 4 phase we have obtained. We further note that MoNi 4 alloy phase has also be synthesized and reported previously, such as Nat. Commun. 2017, 8, 15437 andAdv. Mater. 2017, 29, 1703311. 4. The lack of surface characterization makes it hard to understand the exceptional HOR activity. I understand it is extremely difficult to monitor the surface composition and valence state of surface metals in situ during HOR, but ex situ surface characterization such as XPS is recommended to determine the oxidation state of Ni, Mo, and W. A comparison of the oxidation state of the surface Ni in Ni 4 Mo and in Ni counterpart would be useful.

2017,
Response: We thank the reviewer for the good comments and suggestion. The inevitable exposure of the catalysts in air causes surface oxidation, which largely hampers the examination on the original surfaces of catalysts. As the reviewer mentioned, in-situ studies will eliminate the surface oxidation issue, but current techniques can not allow us (also for other research groups) to monitor the dynamic surface process during the rotating disk electrode testing. Hence, we try the best to avoid the air exposure of our catalysts and studied their surface chemistry before and after HOR process by XPS characterizations.
As demonstrated in Figure R9a, we find that the MoNi 4 and WNi 4 alloys still have strong Ni 0 signals even after the long-term stability test. The Ni 2+ signals come from their surface oxidation when exposing in air and in alkali. By contrast, Ni 0 signals almost disappeared for single Ni catalyst, with only Ni 2+ signals left. These results unambiguously reveal that our designed alloyed catalysts bear more robust surface structures that protect HOR active sites than single Ni metal catalyst when performing HOR in alkaline environments. Moreover, the Mo 3d and W 4f XPS spectra ( Figure   R9b and c)  We have added these new data in the revised SI (Supplementary Figure 12, 29) and provided some discussion properly over there and the revised MS.

I don't think it is possible to accurately evaluate the ECSA of MoNi 4 for the HOR.
Response: We thank the reviewer for this insightful comments.
We know that the best way to evaluate the intrinsic electrochemical activity of a catalyst is to calculate its specific activity based on its electrochemically active surface area (ECSA). However, the ECSA values are difficult to obtain for many non-noble metals: they cannot be calculated using the classic hydrogen under-potential deposition (UPD) like commonly done for Pt because no obvious hydrogen adsorption occurs prior to H 2 evolution. Alternatively, for non-noble catalysts, we have to turn to other methods to evaluate the ECSAs.
Commonly, the ECSA can also be quantified by the redox reaction of surface metals, which relies on the interaction between the surface metal atoms and oxygenated species. In the original manuscript, we evaluated the ECSAs of Ni-based catalysts from the OH desorption region using a charge density of 514 µC cm -2 Ni for one monolayer of OH adsorption. This method has recently been widely  Sci. 2014, 7, 1719-1724), etc. Therefore, we believe that the method we used to evaluate the ECSA values of our studied catalysts is reasonable, which could offer a fair comparison of our developed catalysts with that of other HOR catalysts reported previously. a, CV curves of MoNi 4 alloy collected at various scan rates ranging from 10 to 50 mV s -1 . b, The corresponding linear fitting of scan rates versus difference between the anodic and cathodic current at -0.15 V vs. Ag/AgCl. Motivated by your comments, we double checked the ECSA values of our catalysts from the double-layer capacitance according to equation ECSA = C dl /C s (where C s is the specific capacitance of the catalyst or the capacitance of smooth planar surface per unit area) that suggested by McCrory, Peters and Jaramillo (J. Am. Chem. Soc. 2013, 135, 16977). For the estimate of surface area, we adopted the general specific capacitance of C s = 11 µF cm -2 based on typical reported value (Angew. Chem. Int. Ed. 2019, 58, 10644-10649). As shown in Figure R10, our measurements give an ECSA value of 10.64 cm 2 for the MoNi 4 alloy loaded on the glassy carbon electrode (0.196 cm 2 ), matching well with our previous result of 10.32 cm 2 . Thus, for our PGM-free catalysts, we could gain relatively accurate ECSA values for the comparison of their intrinsic HOR activities.
We have added the new data in the revised SI (Supplementary Figure 18) and provided some details over there.

This is an interesting work that reported nickel-molybdenum (MoNi 4 ) and Nickel-tungsten (WNi 4 ) bimetallic catalysts for hydrogen oxidation reaction (HOR) catalysis in alkaline electrolytes. The authors reported that the two nanoalloy catalysts showed very high intrinsic activities towards HOR activity compared to most of previously reported non-previous metal catalysts. The experimental results are important and sufficient characterizations, tests and computations were provided. It can be published in Nature Communications after considering the following questions.
Response: We greatly appreciate the reviewer's high praise and support on the publication of this work.

1) The authors should decouple the effects of composition, structure and morphology. Since a microwave synthesis method was used to prepare high surface area of MoNi 4 and WNi 4 samples, What would the HOR activity be if they are just non-porous nanoparticles (not the nanosheet structure)? A blank experiment on HOR activity test on MoNi 4 /WNi 4 nanoparticles should be given. In addition, what is the composition effect? It seems that the authors have optimized the Mo-Ni composition, however, what is the reason behind the optimal composition? Is it because of change of the ratio of surface Ni, of the phase structure, or of chemical state of Ni, or some other reasons?
Response: We appreciate the reviewer for the thoughtful comments and questions. To synthesize the desired nanoalloys, we first prepared the sheet-like Mo(W)-doped Ni(OH) 2 precursors, which were then annealed in H 2 /Ar atmosphere to generate alloys. The high-temperature treatment leads to MoNi 4 porous sheets (composed of nanoparticles actually; see Supplementary Fig. 3) and WNi 4 nanoparticles (Supplementary Fig. 4). Following your suggestion, we further prepared non-porous MoNi 4 nanoparticles (refer to ACS Catal. 2013, 3, 166-169; see Figure R11 below) and compared their HOR activity with that of MoNi 4 porous sheets. We see that the non-porous MoNi 4 nanoparticles also bear certain HOR activity in alkaline electrolytes but the reactivity is inferior to our designed alloyed catalyst. These results suggest that the MoNi 4 phase is indeed HOR active, and the porous sheets composed of nanoparticles could provide much more active sites, thus giving superior HOR performances. Figure R11. Physical characterization and HOR activities of non-porous MoNi 4 nanoparticles. a, SEM image. b, XRD pattern. The inserted lines are indexed to MoNi 4 (JCPDS 65-5480). c, HOR polarization curve.
In our original manuscript, we performed a series of control experiments that determine the MoNi 4 and WNi 4 phases are optimum for HOR catalysis in alkali (see Supplementary Fig. 22 and 23).
Deviating this composition leads to inferior HOR activity. Our UPS and CO-stripping experiments, in conjunction with DFT calculations, uncover that MoNi 4 and WNi 4 structures simultaneously enable the optimized HBE and OHBE on their surfaces, which can substantially boost the Volmer step, leading to the marked HOR performance of the MoNi 4 and WNi 4 catalysts. Therefore, our studies indicate that altering the ration of Ni, oxidation state of Ni, or the phase structure will cause decreased HOR properties.
We have added the new data in the revised SI (Supplementary Figure 26) and provided some details over there.

2) The authors provided the HOR durability tests, and showed both catalysts are better than Pt/C, what is the main mechanism behind it? the authors should show some characterizations after HOR tests.
Response: We thank the reviewer for the useful question and suggestion.
Well, in our original manuscript, we have provided SEM, TEM, HRTEM and STEM-EDX analyses of the MoNi 4 and WNi 4 alloys after HOR tests (see Supplementary Fig. 30 and 32), which show that the morphology, composition and structure of the alloyed catalysts are well maintained. We here further collected the cycled alloyed catalysts and carried out XRD characterizations (Figure R12 below). Our results show that their crystal phases are also kept well after testing. These results clearly demonstrate the structural robustness of our alloyed catalysts that catalyze HOR in alkaline electrolytes. By contrast, we observed that commercial Pt/C catalyst shows gradual current drop during the stability assessment. To probe the intrinsic reason, we performed post-TEM analysis of the cycled Pt/C catalysts. As shown in Figure R13, we find clear particle size agglomeration as reaction time prolongs. After reacting from 0 h to 20 h, the Pt particle size grows obviously and forms agglomerated clusters, which could be the intrinsic reason that leads to inferior stability of the Pt/C catalyst. We have added these new data in the revised SI (Supplementary Figure 28, 31) and provided some discussion properly over there.

3) On the DFT computation, it is convincing that both HBE and OHBE play roles on HOR in alkaline electrolyte. Can the authors used DFT to give some insights into the CO-tolerant capability of the Mo(W)-Ni nanoalloy catalysts?
Response: We thank the reviewer for the positive comment and good suggestion. Following your suggestion, we performed DFT calculations on the studied catalysts to probe the CO adsorption ability, which could offer insights on their CO-tolerant capability. We created the models with adsorption of OH intermediates to represent the catalysts in alkaline environment (see Figure   R14a-c). Then, CO molecules were absorbed on the surfaces of these catalysts to form CO* intermediates (see Supplementary Fig. 33 for details). Our calculated adsorption energies for CO* intermediates are -1.68 eV, -1.45 eV and -1.44 eV for Pt(111), MoNi 4 (211), and WNi 4 (211) surfaces ( Figure R14d), respectively. These results show that CO* adsorbed on Pt surface is too strong, which covers the Pt active sites, leading to HOR deactivation. By contrast, CO adsorption on MoNi 4 and WNi 4 catalysts is much more appropriate, showing the CO-tolerant capability. We have added these new data in the revised SI (Supplementary Figure 33) and provided some discussion over there. Figure.16 should include the HOR activity on state-of-the-art Pt/C catalysts. Make sure that all the HOR activities on various non-previous metal catalysts were tested under same conditions. Response: We thank the reviewer for the useful suggestion. We accepted your suggestion and added the intrinsic activity of Pt/C catalyst for comparison ( Figure R15). We obtained the electrochemical active surface area of Pt/C by a commonly used CO-stripping route (see Figure 5b in the MS), assuming a specific charge of 420 µC cm -2 Pt for stripping CO monolayer on Pt. Figure R15 shows the ECSA-normalized exchange current density of various catalysts that gained at the same condition. We find that the ECSA-normalized exchange current densities are 0.065 and 0.068 mA cm -2 for MoNi 4 and WNi 4 catalysts, which are larger than that of 0.043 mA cm -2 for Pt/C catalyst, suggesting the superior intrinsic HOR activity of our designed alloyed catalysts. Figure R15. Comparison of the intrinsic HOR activities-ECSA normalized exchange current density-of MoNi 4 , WNi 4 , freshly-synthesized Ni, Pt/C and most reported PGM-free HOR catalysts measured in alkaline electrolyte (pH 13).

4) The authors showed the geometric activity of HOR on MoNi 4 is greater than a commercial Pt/C catalyst, however, they did not show the true intrinsic activity comparison between MoNi 4 and Pt/C. Supplementary
This new Figure was updated as Supplementary Fig. 19 in our revised SI.
5) It will be more exciting to see the new bimetallic catalysts tested in a realistic fuel cell, and show they can serve as a promising non-precious anode for realistic AEMFCs. Response: We thank the reviewer for the nice suggestion. Following your suggestion, we collaborated with colleagues and carried out the AEMFC tests. The AEMFC was achieved by using the quaternary ammonia poly(Nmethyl-piperidine-co-p-terphenyl) membrane and ionomer which need be transferred from Clto OHconductivity in KOH electrolytes before use. As shown in Figure R16, the cells show maximum power densities of 45 mW cm -2 for MoNi 4 anode and 37 mW cm -2 for WNi 4 anode with a high open circuit voltage (OCV) of above 1.0 V, whereas the cell with the Ni anode shows an extremely low OCV of 0.2 V and gives no current curves. Depite the AEMFC performance need be further improved by optimizing numerous parameters, yet we can see that the alloyed catalysts performed obviously superior to single Ni catalyst.
Currently, the COVID-19 pandemic has led to the laboratory shutdown, we are unable to systematically modify the experimental parameters for these new PGM-free catalysts although much improvement is expected. As a matter of fact, optimizing the MoNi 4 and WNi 4 catalysts and exploring their use in AEMFCs would be our future works, which need comprehensive and systematic research and we hope to report related results later separately. We thus have not included this very preliminary results in the present work. Figure R16. Polarization curve (no iR-corrections) of the AEMFC with MoNi 4 and WNi 4 anodes, respectively. The alloy loading in anode is 2 mg cm -2 , whereas the Pt loading in cathode is 0.4 mg cm -2 . The back pressure of gases is 1 atm at both sides of the cell. The operating temperature is 80 °C. Pure H 2 and O 2 are fully humidified and fed into the cell at a flow rate of 500 mL min -1 .

Reviewer #3 (Remarks to the Author):
This paper demonstrated that the nickel-molybdenum nanoalloy with tetragonal MoNi 4 phase can catalyze the HOR efficiently in alkaline electrolytes. This catalyst is tolerate to carbon monoxide poisoning. DFT calculations were performed to evaluate the HBE and OHBE. The authors rational the good efficiency in the combination of nickel and molybdenum for optimized adsorption of intermediates. Another analysis tool used is the d-band center shift trend. Some points for minor improvement: Response: We greatly appreciate the reviewer for the positive feedbacks on the contents presented in our manuscript.

Please clarify why no XC corrections were included for transition metals.
Response: We thank the reviewer for the thoughtful suggestion.
Commonly, HOR is thought to proceed via two reaction mechanism: Tafel-Volmer mechanism or Heyrovsky-Volmer mechanism (Sci. Adv. 2016, 2, e1501602). According to previous work (Adv. Mater. 2019, 31, 1808066), one can probe the reaction mechanism from the symmetry of Tafel plots and Tafel slopes of the catalysts. We find that the Tafel plot of MoNi 4 shows asymmetric |j k |~E relation, and its Tafel slope is ~48.6 mV dec -1 , reasonably close to 39 mV dec -1 (Figure R17), suggesting that the HOR on the MoNi 4 catalyst is conducted via a Heyrovsky-Volmer pathway with a Volmer rate-determining step (Adv. Mater. 2019, 31, 1808066). On the basis of above results, we conclude that the Heyrovsky-Volmer mechanism works for alkaline HOR on our alloy catalysts.

Note:
The Tafel slope is obtained in the overpotential region from 50 mV to 100 mV to avoid the incorrect fitting in too low overpotential region based on the definition of Tafel equation (J. Electrochem. Soc. 2010, 157, B1529-B1536).