In Situ Raman Study of Amorphous and Crystalline Ni-Co Alloys for the Alkaline Oxygen Evolution Reaction

Amorphous Ni 79.2-x Co x Nb 12.5 Y 8.3 (x = 0, 5 at.%) alloys were synthesized using cryogenic mechanical alloying and evaluated as catalystfortheoxygenevolutionreaction(OER)inalkalinemediausingcyclicvoltammetryandTafelmeasurements.Electrochemical testingshowedthattheamorphousalloyspossessedlowerTafelvaluesfortheOERcomparedtocrystallineNiandNi 95 Co 5 . Anodic cycling of amorphous Ni 74.2 Co 5 Nb 12.5 Y 8.3 resulted in a lower onset potential for the OER and decreased Tafel values while no changes were observed for amorphous Ni 79.2 Nb 12.5 Y 8.3 and crystalline Ni 95 Co 5 . Pairing of in situ confocal Raman spectroscopy with anodic cycling showed that the amorphous Ni 74.2 Co 5 Nb 12.5 Y 8.3 alloy formed reversible hydrous Co surface species instead of irreversible CoO 2 typically seen on crystalline NiCo alloys in KOH. The formation of hydrous Co surface species upon cycling was alsoaccompaniedbyincreasedformationof β -NiOOHleadingtoenhancedcatalyticperformanceofamorphousNi 74.2 Co 5 Nb 12.5 Y 8.3 alloy over the amorphous Ni 79.2 Nb 12.5 Y 8.3 and the crystalline counterparts. © The Author(s) 2018. Published by ECS. This is an open access article distributed

Traditional alkaline water electrolysis has been successful industrially because of the use of low cost, stable, nickel-based electrocatalyst materials. The major drawback to these systems is that they use aqueous KOH as the electrolyte which is not desirable from a corrosion and packaging perspective. The use of novel anion exchange membranes (AEM) in alkaline water electrolysers is now of particular interest since it eliminates the aqueous KOH electrolyte and can provide enhanced form factors and much higher current densities that rival PEM-based electrolyzers. [1][2][3][4] Conventional Ni electrocatalysts for the oxygen evolution reaction (OER) still display large overpotentials and slow reaction kinetics. [5][6][7][8] The OER reaction remains the main hindrance for increasing total cell efficiency in water electrolysis. [5][6][7][8] The addition of Co to Nibased alloys has been noted to increase the activity by stabilizing the β-NiOOH phase over γ-NiOOH, but this often leads to an increase in the oxygen overpotential. 9,10 The issue of increased overpotential can be mitigated with the use of amorphous structures to reduce the oxygen overpotential with the addition of Co. 11 This previous research has been limited to planar amorphous metal oxides, which are not ideal for catalysts in AEM water electrolysis. [11][12][13][14][15][16][17] Effective AEM catalysts should be designed on similar principles as catalysts used in proton exchange membrane (PEM) water electrolysers. An ideal catalyst should have: a reduced particle size and porous morphology to increase the amount of electrochemically active sites, enhanced gas disengagement, and low charge transfer resistance to reduce the overpotential. 5 These features are often lacking in amorphous alloys fabricated by planar flow casting which results in smooth surfaces with a low electrochemically active surface area. A potentially attractive alternative production method is mechanical alloying.
Mechanical alloying is a solid-state process that can be used to produce amorphous alloys with high electrochemically active surface area while also facilitating new alloy chemistries by overcoming fabrication difficulties associated with solidification processes. 18,19 The amorphization process can be further enhanced by using cryogenic temperatures during the mechanical alloying process. 19 Amorphous Ni-Nb-Y alloys are of particular interest for this work as prospective electrocatalysts in AEM water electrolysis. Mattern et al. demonstrated that compositions derived from targeting deep eutectics in the binary Ni-Nb and Ni-Y systems resulted in the formation of two separate amorphous phases during planar flow casting. 20 This phase separation was attributed to the large positive heat of mixing between Nb-Y with nearly identical negative heats of mixing of Nb-Ni and Y-Ni. The amorphous heterogeneous structure is attractive for electrocatalysis as it has the potential to improve kinetics and reduce overpotential through synergistic catalytic effects or can produce nanoporosity if one of the phases is selectively removed. 21 The goal of this work is to utilize mechanical alloying as a means of producing novel amorphous (Ni,Co)-Nb-Y alloys to improve the OER kinetics. The amorphization process during mechanical alloying tends to favor high melting point intermetallics due to the ability to produce a high degree of defects without structural relaxation. 18 Compositions were developed using high melting point intermetallics reported in the binary phase diagram between Ni-Nb and Ni-Y (Ni 3 Nb and Ni 5 Y) while assuming a substitutional solid solubility of Ni and Co at low Co concentrations. Cryogenic mechanical alloying lead to a novel set of amorphous Ni 79.2-x Co x Nb 12.5 Y 8.3 alloys being produced. This work investigates the catalytic activity of amorphous Ni 79.2-x Co x Nb 12.5 Y 8.3 (x = 0, 5 at.%) for the OER compared to crystalline Ni and Ni 95 Co 5 .

Experimental
Amorphous Ni 79.2-x Co x Nb 12.5 Y 8.3 (x = 0, 5 at.%) alloys were synthesized using mechanical alloying. For mechanical alloying, elemental powders of nickel (99.9 wt%, -100 mesh), cobalt (99.8 wt%, -100 mesh), niobium (99.99 wt%, -325 mesh) and yttrium (99.9 wt%, -40 mesh) were used as starting material. Powders were packed under an argon atmosphere (99.99% purity, <10 ppm O 2 ) in stainless steel vials with two 7 mm diameter stainless steel balls. The ball to powder ratio (BPR) was set at 10:1 for all alloys. A Retsch CryoMill was used to mill the powders at a frequency of 30 Hz for 3, 6, and 12 hr. For cryogenic milling, liquid nitrogen was passed around the vials to reduce the temperature to -196 • C before the milling commenced. Cryogenic milling was performed using a cycle of 10 min on and 5 min off while liquid nitrogen continued to be passed around the vials to ensure they remained near cryogenic temperatures. The resulting particle geometry was investigated using scanning electron microscopy (SEM, Hitachi SU3500). Nominal compositions of powders were verified using inductively coupled plasma-atomic emission spectroscopy (ICP-AES, Agilent 720). X-ray diffraction (XRD) with CuK α radiation (λ = 1.54 Å) was used to analyze the microstructures of the powders with a step size of 0.05 • every 3 s from 20-80 • 2θ (Rigaku Miniflex 600). Transmission electron microscopy (TEM) was performed at 300 kV to image the powder and perform selected area diffraction (SAD) to confirm the amorphous microstructure if no crystalline features were observed in XRD (Hitachi HF 3300).
Cyclic voltammetry (CV), steady-state potentiostatic, and surface area measurements were performed using a Bio-Logic VSP-300 multichannel potentiostat/galvanostat. Measurements were performed using a static three-electrode set up in a Teflon cell with a Pt mesh as the counter electrode and a Hg/HgO reference electrode filled with 1 M KOH solution (+825 mV vs. RHE). Static measurements were performed to mirror in situ Raman spectroscopy testing. Working electrodes were prepared by depositing catalysts inks onto a 3 mm diameter glassy carbon electrode polished using 0.05 μm colloidal silica (CH-Instruments). Catalysts inks were prepared by combining mechanically alloyed powder (4 mg) with Nafion solution (8 μL) and isopropanol (200 μL). The ink was sonicated for 10 min and 15 μL was deposited on a glassy carbon electrode rotating at 150 rpm to ensure uniform coating thickness. Experiments were carried out at 30 • C in a 1 M KOH solution deaerated with argon gas. Electrolytes were prepared using AnalaR grade KOH pellets and Type I, 18.2 M water. The electrolyte was then pre-electrolysed at -1.7 V Hg/HgO for 24 hrs. prior to testing. Cyclic voltammetry was performed using a scan rate of 50 mV/s between 0.1 and 0.7 V Hg/HgO . Steady-state potentiostatic polarization was performed between 0.3 and 1 V Hg/HgO while stepping the potential 20 mV and holding for 10 min. Surface area measurements were performed using cyclic voltammetry in the non-faradaic regions in accordance with the method outlined by McCrory et al. 22 In situ confocal Raman spectroscopy was employed using a HORIBA iHR320 confocal Raman spectrometer and microscope to monitor the formation and reduction of surface species during potential cycling. Electrochemical measurements were performed using a Bio-Logic SAS SP-50 potentiostat at room temperature in a customdesigned Teflon cell containing 0.1 M KOH solution (pre-electrolysed for 24 hrs. at -1.7 V Hg/HgO ) and deaerated with argon gas (99.99%). A schematic cross section view of the custom Raman cell can be seen in Fig. 1. The reduced electrolyte concentration was required because of the quartz window in the in situ Raman spectra cell. A reversible hydrogen electrode (RHE) was used as a reference electrode (-782 mV vs. Hg/HgO). A glassy carbon plate with a geometric surface area of 3 cm 2 was used as the counter electrode. The working electrodes were prepared by depositing the catalysts ink onto a 9.5 mm diameter glassy carbon disk. Raman spectra were obtained for cyclic voltammetry cycles 5, 10, 20, 50, 75, and 100 using a scan rate of 0.4 mV/s. Other cycles where Raman spectra were not obtained used a scan rate of 50 mV/s to mimic ex situ measurements. The Raman spectra were acquired at 50 mV intervals with each measurements lasting 125 s. Each Raman spectrum is then attributed to an average value over a 50 mV window when cycling. For steady-state potentiostatic measurements, Raman spectra were obtained for each 20 mV step once the current stabilized.

Results and Discussion
Fabrication and structural characterization. -Fig. 2 shows the XRD spectra for Ni 74.2 Co 5 Nb 12.5 Y 8.3 after being cryogenically milled for 3, 6, and 12 hr. The use of cryogenic temperatures was necessary for the amorphization reaction to occur in this alloy system as performing milling at room temperature induced little amorphization even after 12 hr. The alloy milled for 6 hrs. shows a predominantly amorphous structure as evident by the broad peak observed in XRD. Residual trace amounts of Y 2 O 3 after 6 hrs. of milling are also observed. Although the yttrium is stored in an argon filled glove box, yttrium has a high affinity for oxygen leading to some oxidation during milling. Prolonged milling for 12 hrs. was noted to cause instability in the amorphous phase allowing some recrystallization to occur.
The microstructure of Ni 74.2 Co 5 Nb 12.5 Y 8.3 alloy after 6 hrs. of cryogenic milling was further investigated through imaging and electron diffraction in TEM (Fig. 3). The edges of particles were used for imaging and electron diffraction due to the still relatively large particle size for TEM. Even with thickness variations in the particle, there are no observed crystalline features in the bright field (BF) image. This observation is further validated through SAD patterns, as shown in Fig. 3. The presence of a singular broad ring indicated that the microstructure is amorphous and free of crystalline features. Previous work has shown similar milling behavior was observed for Ni 79.2 Nb 12.5 Y 8.3 with XRD and TEM studies indicating 6 hrs. of cryogenic milling was able to yield an amorphous structure. 23 Crystalline Ni 95 Co 5 was also prepared through mechanical alloying at room temperature to compare the effect of Co addition in both crystalline and amorphous alloys. The compositions of mechanically alloyed amorphous Ni 79.2-x Co x Nb 12.5 Y 8.3 (x = 0, 5 at.%) and crystalline Ni 95 Co 5 were measured using ICP-OES and compared to the nominal composition, shown in Table I. The ICP analysis was shown to be in excellent agreement with the nominal composition, henceforth in the manuscript the nominal composition was used to identify the powders. The excellent correlation also demonstrated that the methodology used to produce the alloys was robust and alloys can be fabricated within ± 1 atomic percent. Through ICP-OES no significant impurities were detected within these powders (i.e. Fe <0.1 at. % was observed). Fig. 4 shows SEM images comparing morphology and particle size of crystalline Ni 95 Co 5 and amorphous Ni 79.2-x Co x Nb 12.5 Y 8.3 (x = 0, 5 at.%) before and after electrochemical testing. Crystalline Ni 95 Co 5 is shown to have much larger particle sizes (43 ± 29 μm) which are more plate-like compared to the amorphous powders (9 ± 5 μm for Ni 79.2 Nb 12.5 Y 8.3 and 6 ± 3 μm for Ni 74.2 Co 5 Nb 12.5 Y 8.3 ). These differences are attributed to amorphous alloys being more brittle and milled under cryogenic temperatures while Ni 95 Co 5 was milled at room temperature. No noticeable changes in powder morphology were observed after electrochemical testing for all three alloys. Similarly, XRD showed no discernable differences in structure of the powders after electrochemical testing indicating the amorphous powders are stable electrocatalysts.
OER activity.-The OER electrocatalytic performance of amorphous Ni 79.2-x Co x Nb 12.5 Y 8.3 (x = 0, 5 at.%) powder (6 hrs.) was investigated by performing Tafel measurements using steady-state polarization after anodic cycling. These alloys were chosen for electrochemical studies based on XRD and TEM analysis that confirmed a fully amorphous microstructure. Fig. 5 shows a comparison of Tafel slopes between these alloys and crystalline Ni and Ni 95 Co 5 after being anodically cycled 50 times. The Tafel values, along with the OER onset potential (E onset ) and current density at 600 mV Hg/HgO (j 600 mV ), are shown in Table II. These results demonstrate that the addition of Co to Ni is beneficial for enhancing the OER kinetics in both crystalline and amorphous alloys by reducing the Tafel slope and increasing the current density at a given voltage.
It is also noted in Table II that the addition of Co reduced the onset potential for the OER (E onset ) for the amorphous alloy but increased it for the crystalline alloy. These results were consistent with previous reports which suggested that this difference in behavior for E onset is due to crystalline NiCo alloys forming non-reversible crystalline Co-based oxides on the surface while amorphous Ni-Co alloys form reversible hydrous Co oxy/hydroxides. 9,13,14 Prolonged anodic cycling of crystalline Ni 95 Co 5 and amorphous Ni 79.2-x Co x Nb 12.5 Y 8.3 (x = 0, 5 at.%) further accentuated these differences. Fig. 6 shows CV curves for alloys cycled between 0.1 and 0.7 V Hg/HgO 220 times using a scan rate of 50 mV/s. These alloys exhibited the formation (A 1 ) and reduction (C 1 ) peaks of NiOOH surface species. The crystalline Ni 95 Co 5 (Fig. 6a) and amorphous Ni 79.2 Nb 12.5 Y 8.3 (Fig. 6b) cathodic peak exhibited splitting into two peaks C 1 and C 1 ', which has been previously noted to be the reduction of β-NiOOH and γ-NiOOH, respectively. 14,24 The presence of cathodic peak splitting indicated  β-NiOOH is not stabilized and was converted to γ-NiOOH which lead to the lack of increased performance in these alloys with cycling. Continued cycling of the crystalline alloy exhibited an anodic peak growing around 0.62 V Hg/HgO (A 2 ). This peak was associated with the formation of CoO 2 and the absence of a corresponding cathodic peak suggested the oxide was not being fully reduced during cycling. 14 The observation of CoO 2 peak appeared to be linked with the formation of γ-NiOOH which suggest that as Co is oxidized to CoO 2 it is no longer able to aid in the stabilization of β-NiOOH. In contrast, amorphous Ni 74.2 Co 5 Nb 12.5 Y 8.3 (Fig. 6c) did not show a peak at 0.62 V Hg/HgO . The absence of this peak in the amorphous alloy implied that any Co surface species being formed were subsequently reducible and did not accumulate as a separate phase on the surface, which aided in the stabilization/doping of β-NiOOH. Fig. 6 shows the current at OER (0.7 V Hg/HgO ) for the crystalline alloy did not increase with potential cycling. The steady current indicated the alloy is not becoming more active with progressive cycling even though the currents associated with the anodic and cathodic peaks continued to increase. Conversely, the amorphous Ni 74.2 Co 5 Nb 12.5 Y 8.3 appeared to become more active toward the OER when cycled, in good agreement with other amorphous Ni-based systems shown to produce hydrous oxy/hydroxide surface species. 13 These results suggested that the formation of CoO 2 on the surface in the crystalline alloy is ultimately not beneficial for the OER activity. To further confirm this hypothesis, steady-state polarization was conducted after cycling 220 times. Values for the Tafel slope, E onset , and j 600 mV are tabulated in Table III. These results validated the observations seen in CV measurements and show that the speciation of Co is vital for the activation process. Comparing the results from Table III to those  seen in Table II for 50 cycles showed no major differences in activity for crystalline Ni 95 Co 5 and amorphous Ni 79.2 Nb 12.5 Y 8.3 . The conversion to γ-NiOOH, and CoO 2 surface species for crystalline Ni 95 Co 5 dominated the OER given that the NiOOH peaks (A 1 and C 1 ) continued to increase in current during cycling. The cycling characteristics of the amorphous Ni 74.2 Co 5 Nb 12.5 Y 8.3 did not show any presence of A 2 or C 1 splitting. The absence of these peaks resulted in the reduction in

Material
Tafel Slope (mV/cd) E onset (mV) j @ 600 mV Hg/HgO × 10 −3 (mA/cm 2 ) both Tafel slope (47 to 40 mV/cd) and E onset (518 vs. 509 mV) values with j 600 mV increasing more than twofold (92.2 to 209 mA/cm 2 ) when going from 50 cycles to 220 cycles. From Tables II and III it is clear that for a fixed overpotential of 600 mV Hg/HgO : (i) the amorphous Co containing alloy displays a higher current density than the corresponding crystalline alloy, (ii) this effect is enhanced by further cycling (4.5X at 50 cycles, 10.1X at 220 cycles, and (iii) the presence of Co is essential to the catalytic activity toward the OER as the amorphous Co free alloy is not as active, hence both structure and chemistry are critically important.
Another marked difference between amorphous and crystalline CV curves was the amorphous alloys have a slightly more positive A 1 peak position which remained constant upon cycling whereas the A 1 peak in the crystalline alloy shifted to more positive potentials upon cycling. It is suggested that this difference is a result of amorphous alloys forming hydrous (Ni,Co)OOH while crystalline Ni-Co alloys form NiOOH and CoO 2 . 12,25 Further XPS work is ongoing to confirm the binding shell chemistry.
In situ confocal Raman spectroscopy characterization.-In situ confocal Raman spectroscopy was employed to investigate the assumption that the crystalline alloy formed CoO 2 and the amorphous alloys formed hydrous (Ni,Co)OOH during potential cycling. The pairing of Raman spectroscopy with CV can provide useful information as to what surface species are being produced and reduced during cycling and the form in which they exist in situ. Fig. 7 demonstrates the ability for Raman spectroscopy to monitor the formation of NiOOH upon the anodic scan (a) and the reduction of NiOOH during the reverse scan (b). The two peaks observed around 460 cm −1 and 530 cm −1 are attributed to δ (Ni III -O) and ν (Ni III -O) vibrations respectively. 26,27 The broad feature occurring around 1000 cm −1 is the formation of NiOO -. 26 Raman spectra were also generated for Ni powder at cycles 5, 10, 20, 50, 75, and 100 to provide a baseline of peak positions throughout the cycling process as shown in Fig. 8a). With increased cycling, pure Ni was observed to deactivate after cycle 20 and the formation of oxygen on the surface interfered with the Raman signal. Similar experiments were also performed for Co powder (Fig. 8b) which showed peaks at 453, 539, 600, and 670 cm −1 . The peaks at 453 (E g ), 600 (F 2g ), and 670 (A 1g ) cm −1 have been previously reported to correspond with bare Co with the peak at 539 cm −1 (F 2g ) being attributed to the presence of CoO(OH). [27][28][29][30] Given that the CoO(OH) Raman peak position overlaps with the NiOOH peaks, and only a small concentration of the alloys is Co (5 at.%), this peak cannot be easily used to uniquely identify the presence of CoO(OH) over NiOOH. Instead Yeo and Bell 28 reported that anodic polarization of CoO x from 0 to 0.7 V Hg/HgO in 0.1 M KOH red-shifts a singular Raman peak from 609 cm −1 to 579 cm −1 as the oxidation state of Co goes from Co(II) to Co(IV). This red-shift is due to the applied redox potential and has been previously seen in the oxidation of CoO x in a O 2 /Ar plasma as the mole fraction of O 2 increased. 27 As such, the presence of a peak in the Raman spectra within the range of 609 to 579 cm −1 would suggest Co is present as an oxide rather than as CoO(OH). Fig. 9 shows a comparison of Raman spectra produced during cycle 100 in the cathodic direction between (a) crystalline Ni 95 Co 5 , (b) amorphous Ni 79.2 Nb 12.5 Y 8.3 , and (c) amorphous Ni 74.2 Co 5 Nb 12.5 Y 8.3 . The Raman intensities were normalized to the intensity of crystalline peaks for ease with identification and comparison. Initially, amorphous alloys showed at least 10x lower intensities compared to crystalline Ni 95 Co 5 that suggests the surface species of amorphous alloys are more disordered and hydrous in nature compared to those on the crystalline alloy. The most obvious peaks observed are the δ (Ni III -O) and ν (Ni III -O) Ni peaks. Amorphous Ni 79.2 Nb 12.5 Y 8.3 also shows a noticeable peak around 610 cm −1 which can be attributed to the ν s (Y-O) vibration mode for Y(OH) 3 peak. 31 The presence of Y  surface species was not observed in amorphous Ni 74.2 Co 5 Nb 12.5 Y 8.3 . The absence of Y Raman peaks in amorphous Ni 74.2 Co 5 Nb 12.5 Y 8.3 suggested that Co diffuses to the surface during anodic cycling producing a predominant Ni and Co rich surface. These observations indicate Ni and Co surface species in amorphous Ni 74.2 Co 5 Nb 12.5 Y 8.3 are largely responsible for the catalytic activity and correlate with previous XPS depth characterization reports on amorphous Ni-Co alloys. 13,14 Due to the low concentration of Co, the CoO x peak is not easily discernable. The CoO x peak position and intensity becomes more apparent when using the Raman spectra from the Ni powder to deconvolute the Ni peaks which is summarized in Table IV. The presence of a CoO x is highlighted using bold text. For amorphous Ni 79.2 Nb 12.5 Y 8.3 the Y(OH) 3 peak is italicized. As noted in Fig. 6a), the CV curve does not show a reduction peak for CoO 2 , yet the Raman spectra for the crystalline alloy consistently indicated the presence of Co oxides on the surface. By cycle 10 a peak around 590 cm −1 continued to be present for all remaining forward and reverse scans with very little change in Raman peak position. The lack of a change in position verifies that the surface oxide for the crystalline alloy was not being reduced during cycling as changes in Co speciation between Co(II), Co(III), and Co(IV) would have resulted in peak shifts. 27 Interestingly the presence of Co oxide was also observed in Fig. 9 for the amorphous Ni 74.2 Co 5 Nb 12.5 Y 8.3 at higher potentials even though it was not detected during CV measurements. From 0.5 to 0.6 V Hg/HgO the appearance of a peak around 590 cm −1 was observed but subsequently was reduced in the following reverse scan. This phe-nomenon was observed for each cycle for which Raman spectra were acquired. The formation of Co oxides was noted to correspond with the presence of β-NiOOH over γ-NiOOH. Previous in situ Raman spectroscopy work in 0.1 M KOH has shown the ratio of intensities between the δ (Ni III -O) and ν (Ni III -O) Ni peaks (I 460 /I 530 ) can provide insight into which phase is predonminant. 26,32 These studies have shown that higher ratios are associated with γ-NiOOH due to the disordered structure restricting molecular stretching vibrations (ν) and favoring bending ones (δ). 26,32 Lower intensity ratios refer to β-NiOOH and yield better electrochemical performance. 26,32 Table III shows that during potential cycling into the regime of the OER, the I 460 /I 530 ratio decreased for the amorphous and crystalline alloys with amorphous Ni 74.2 Co 5 Nb 12.5 Y 8.3 consistently showing the lowest values. Fig. 10 shows a comparison between these crystalline Ni 95 Co 5 and amorphous Ni 74.2 Co 5 Nb 12.5 Y 8.3 at 0.6 V Hg/HgO during cycle 100. It can be seen that when overlaying the normalized data that the ν (Ni III -O) peak (530 cm-1 ) increased for the amorphous Ni 74.2 Co 5 Nb 12.5 Y 8.3 indicating the formation of β-NiOOH. These results are consistent with potential cycling and Tafel measurements indicating amorphous materials are more OER active than their crystalline counterparts. Table III also shows that the lowest peak intensity ratios (I 460 /I 530 ) were also associated with the presence of a Co oxide peak for Ni 74.2 Co 5 Nb 12.5 Y 8.3 .
Raman spectra were captured during Tafel measurements using potentiostatic polarization to further investigate the effect of Co on Ni surfaces species during OER. Fig. 11 displays the peak intensity ratio These results showed that even with increasing potential, β-NiOOH is favored and continued to be stable. It was also noted that the formation of CoO x was first observed at 500 mV Hg/HgO which corresponds well with the onset of the OER for amorphous Ni 74.2 Co 5 Nb 12.5 Y 8.3 . Based on this observation it is believed that Co is doped into β-NiOOH whereby Co and Ni exist in the 3+ state and upon OER Co is oxidized to Co 4+ instead of Ni. This feature allows amorphous Ni 74.2 Co 5 Nb 12.5 Y 8.3 to cycle electrons between oxidation states sim- ilar to redox reactions. As a result, β-NiOOH, which corresponds to Ni 3+ , is stabilized and is not promoted to a higher oxidation state in the form of γ-NiOOH or NiO 2 .

Conclusions
In this study, Ni 74.2 Co 5 Nb 12.5 Y 8.3 powder was produced through mechanical alloying at cryogenic temperatures. Through XRD and TEM analysis it was shown that 6 hrs. of cryogenic milling were required to produce amorphous Ni 79.2-x Co x Nb 12.5 Y 8.3 alloys. Subsequent electrochemical testing resulted in amorphous Ni 74.2 Co 5 Nb 12.5 Y 8. 3 showing a reduced onset potential for the OER, lower Tafel values and higher current densities at fixed overpotentials compared to crystalline Ni and Ni 95 Co 5 , and amorphous Ni 79.2 Nb 12.5 Y 8.3 . Prolonged cycling led to even greater reductions in E onset and Tafel values for the amorphous Ni 74.2 Co 5 Nb 12.5 Y 8.3 while no changes were observed for crystalline Ni 95 Co 5 and amorphous Ni 79.2 Nb 12.5 Y 8.3 due to the formation of γ-NiOOH. The combination of electrochemical testing Figure 11. Peak intensity ratio of Ni δ and ν (I 460 /I 530 ) vs. voltage during potentiostatic polarization to Tafel regime after anodically cycling 100 times for amorphous Ni 74. 2  and in situ confocal Raman spectroscopy revealed that during anodic cycling amorphous Ni 74.2 Co 5 Nb 12.5 Y 8.3 produced a Ni-Co oxyhydroxide film on the surface which was reversible. Pairing of Tafel measurements and in situ Raman spectroscopy revealed the formation of CoO x at the onset of OER in crystalline Ni 95 Co 5 . This phenomena, along with the enhanced reversibility of hydrous surface species, result in the superior cycling and catalytic performance of Ni 74.2 Co 5 Nb 12.5 Y 8.3 .