From NiMoO4 to γ-NiOOH: Detecting the Active Catalyst Phase by Time Resolved in Situ and Operando Raman Spectroscopy

Water electrolysis powered by renewable energies is a promising technology to produce sustainable fossil free fuels. The development and evaluation of effective catalysts are here imperative; however, due to the inclusion of elements with different redox properties and reactivity, these materials undergo dynamical changes and phase transformations during the reaction conditions. NiMoO4 is currently investigated among other metal oxides as a promising noble metal free catalyst for the oxygen evolution reaction. Here we show that at applied bias, NiMoO4·H2O transforms into γ-NiOOH. Time resolved operando Raman spectroscopy is utilized to follow the potential dependent phase transformation and is collaborated with elemental analysis of the electrolyte, confirming that molybdenum leaches out from the as-synthesized NiMoO4·H2O. Molybdenum leaching increases the surface coverage of exposed nickel sites, and this in combination with the formation of γ-NiOOH enlarges the amount of active sites of the catalyst, leading to high current densities. Additionally, we discovered different NiMoO4 nanostructures, nanoflowers, and nanorods, for which the relative ratio can be influenced by the heating ramp during the synthesis. With selective molybdenum etching we were able to assign the varying X-ray diffraction (XRD) pattern as well as Raman vibrations unambiguously to the two nanostructures, which were revealed to exhibit different stabilities in alkaline media by time-resolved in situ and operando Raman spectroscopy. We advocate that a similar approach can beneficially be applied to many other catalysts, unveiling their structural integrity, characterize the dynamic surface reformulation, and resolve any ambiguities in interpretations of the active catalyst phase.


Synthesis
Nickel molybdate hydrate was synthesized in a common hydrothermal synthesis, inspired by Wang et al. and Zhang et al. 1,2 . 2.4 mmol (0.698 g) Ni(NO3)2·6H2O was dissolved in 30 mL deionized water and stirred with a magnetic for approx. 20 minutes at 500 rpm. 2.4 mmol (0.582 g) Na2MoO4·2H2O was also dissolved in 30 mL deionized water and slowly added after full dissolving to the prior solution under heavily stirring (500 rpm). The solution was stirred for 20 - 45 minutes. The pH value was measured as 6.8. In parallel the nickel foam (5 cm × 3 cm) was cleaned in an ultrasonic bath in EtOH to remove grease and other organic solvable dirt, followed by rinsing with deionized water and subsequent ultrasonic cleaning in 3 M HCl for 15 min each, to remove impurities and surface oxides. 3 M HCl solution was prepared by diluting 50 mL of 37 % (12 M) hydrochloride acid in 150 mL deionized water.
The foam was then rinsed with deionized water and EtOH and blown dry under nitrogen stream. The cleaned Ni foam and the 0.04 M Ni(NO3)2 and 0.04 M Na2MoO4 solution was transferred into a Teflon lined stainless steel autoclave. The foam was slightly bend and placed against the wall. Previous synthesizes have shown, that close to the wall the most precipitation is taking place. The autoclave was sealed and directly heated up to 150 °C for 6 hours with 2 °C min −1 heating ramp. After the holding time it was allowed to cool down naturally. The yellow greenish NiMoO4·nH2O on nickel foam was ultrasonicated for 10 minutes in EtOH or 2-Propanol, to remove loose particles, followed by rinsing with the same solvent and dried over night at air. The sample is denoted NiMoO4@Nif-2. For comparison, two more nickel molybdate hydrate on nickel foam was prepared by the same procedure, but with heating ramps of 0.5 °C min −1 and 5 °C min −1 , from here on called NiMoO4@Nif-0.5 and NiMoO4@Nif-5, respectively. We would like to clarify, that in all cases no annealing step is conducted and the hydrate structure of NiMoO4 is still present, even if not indicated by the abbreviations.

Characterization
Ex situ Raman spectroscopy was conducted in the described setup under air with a 20 × magnification lens was used with a laser intensity of 5 %, which was adjusted with neutral density filters. Time resolved in situ and operando Raman spectra were collected in the same system with an L-shaped 10 × magnification lens, 50 % power intensity and 5 s acquisition time. The sample was in a quartz glass filled with 1.0 M KOH. There was no quiet time in between the measurements, however, since the program itself needed time for storing the data, a deviation of 2 - 3 s accumulated over 180 s. For operando measurements this means the very first and the very last acquisition are in situ. In the in situ at 0.95 V vs Reversible Hydrogen electrode (RHE) and all operando measurements the bias was applied after collecting the first spectra, to see if the bias caused any effects in the first seconds. For the in situ at 0.95 V vs RHE and the operando measurements a platinum counter electrode and a saturated SI 4 Ag/AgCl reference electrode was placed in the same cell . To facilitate the mounting and to assure a   constant geometric adjustment of the electrode setup while moving the optical table for focusing, the   quartz cuvette and the wires and electrodes were fixed with a Teflon cap and a 3D printed cuvette   holder on the optical table of the spectroscope. The power intensity on the sample was measured for the most common acquisition conditions with a PM 160 optical power meter. The 20 × magnification lens with a laser intensity of 5 % showed a laser power of 5.53 mW on the sample. The samples measured with the L-shaped lens with 10 × magnification, used for in situ and operando Raman spectroscopy experienced a laser power between 22.5 mW (measured before the quartz glass) and 18.4 mW (behind the quartz glass with electrolyte). However, since we could not determine the area the laser beam is exciting, we cannot quantify the power density on the samples. Electrochemical measurements were done in the system described before. The potential of the reference electrode was checked prior to use. The reference electrode was prepared according to literature. 3 The three-compartment cell was stored overnight in 3 M HCl to remove impurities, followed by subsequent intensive rinsing with DI and acetone. As electrolyte a 1.0 M KOH was prepared from dissolving 14.03 g of KOH pellets in 250 mL ultrapure water. We confirmed pH 14 with pH paper. For each electrochemical measurement fresh electrolyte was used. The double layer capacitance (Cdl) of the WE was detected by cyclic voltammetry (CV) between 0.9 V -1.0 V vs RHE with scan rates of 150, 120, 100, 80, 60, 40, 20, 10 and 5 mV s −1 for the bare nickel foam and NiMoO4@Nif.
Since the Ni III /Ni II reduction is sluggish, the reduction current interfered with the capacitive current for the ECSA measurement after catalysis, leading to a drift of the CVs for measurement. In order to decrease the effect of this interference, only the scan rates from 150 mV s −1 to 40 mV s −1 were used, since they showed the most stable behavior (SI Figure 19). All scans were not ohmic drop (iR) compensated. For each scan rate we took two cycles and used the second one for determining the difference in current density of forward and backward scan at 0.95 V vs RHE. This difference was  The energy dispersive X-ray (EDX) spectra of the mapping and the singe spot IDs shows a roughly equal amount of nickel and molybdenum in the nanorod structures, partly even with a slightly higher amount of molybdenum.

SI 8
Scanning electron microscopy (SEM) images of NiMoO4@Nif-0.5 shows a majorly nanorod structures with some dispersed nanoflowers, primarily in clusters in regions without nickel foam. The map spectrum shows an overall similar amount of nickel and molybdenum with a slightly larger atomic percentage of the former. The amount of oxygen is a bit more than four times larger. For the spectra 4 -7 the overall amount ratio of nickel to molybdenum is higher for the flowerlike structure, whereas the amount of molybdenum increases at spots with nanorods.
SI 11 The SEM investigation of NiMoO4@Nif-5 has shown of nanorods, but also the highest amount of nanoflowers among all samples. Those nanoflowers seem to preferably cluster in distance to the nickel foam.
SI 12 In general, the EDX spectra of NiMoO4@Nif-5 again shows the reported trend with having a higher amount of nickel compared to molybdenum. This is especially visible in the point ID spectra of the nanoflower in (c) and (e) compared with the spectrum with the nanorod in (d). However, in all spectra the oxygen percentage is lower than in the other samples and don't reach four times the amount of nickel. Due to charging effect and drift of the image, no trustworthy EDX analysis could be conducted for spectrum 5 and 6.
The elemental concentration and the Ni:Mo ratio is displayed and calculated in the table below.   The XPS spectrum of NiMoO4@Nif-5 also shows an increased nickel content. As for the others, also the high-resolution analysis of O 1s can be divided in a hydroxide part at 531.2 eV and metal oxide part at 529.5 eV. Nickel 2p3/2 was detected at 855.1 eV and 2p1/2 at 872.6 eV. As already mentioned, Mo 3d5/2 and 3d3/2 appears at 231.1 eV and 234.3 eV, respectively.     Table 4: Surface coverage by charge analysis of the CVs of NiMoO4@Nif-0.5 in SI Figure 17 (b).     Figure 19 (b)). From the retained trend in a clearly increased Cdl, we can conclude that the double layer capacitance markedly increased due to the CV compared to the as synthesized electrode.

SI
However, an exact value for the Cdl for each condition would be uncertain. When detecting the Cdl by using different scan rates, the CVs of lower scan rates show more drift because more charge is consumed during the acquisition time in the simultaneously reduction of Ni III to Ni II . Excluding low scan rates minimizes this deviation on the Cdl. Including the scan CVs taken with a scan rate of 40 mV s −1 and 60 mV s −1 results in an overestimated Cdl, however, since the difference to Cdl detected excluding them is only minor, we not consider this as crucial.
SI 24 The Nyquist plot shows an increase of the ohmic drop between before and after 500 cycles. The resistance for NiMoO4@Nif-2 from 2.52 Ω to 2.95 Ω, for NiMoO4@Nif-0.5 increased from 2.47 Ω to 2.72 Ω and for NiMoO4@Nif-5 from 2.62 Ω to 3.10 Ω. The XPS spectra of NiMoO4@Nif-2 are discussed in the main publication.  Intensity (a.u.)

Binding Energy (eV) O 1s
Me As for the other samples, NiMoO4@Nif-5 after electrochemistry shows the presence of potassium and a drastic decrease of molybdenum in the material. The nickel 2p3/2 and 2p1/2 peak appeared at 855.2 eV and 872.9 eV. The oxygen peak could be divided in three fittings. One for water at 531.6 eV, a second for hydroxide at 530.5 eV and one for lattice metal oxide at 528.8 eV. Based on those fittings the composition was calculated as 7.4 at.% from H2O, 11.8 at.% metal oxide and 80.8 at.% hydroxide. As for the others, the electrochemical process increased the amount of hydroxide drastically.
According to the work of Biesinger et al. 4 , the presence of γ-NiOOH can be detected by a small peak in the Ni LMM Auger spectra at around 832.8 eV kinetic energy. As described in the time resolved operando Raman spectroscopy section, we identified γ-NiOOH as the active species. Since the XPS was not done directly after the electrochemical measurements, we believe γ-NiOOH has transformed over time in another nickel-(oxy)hydroxide and therefore we cannot detect the characteristic small peak for γ-NiOOH in our spectra.
SI 33 SI