An Oxalate Method for Measuring the Surface Area of Nickel Electrodes

This work establishes a new method to measure the electrochemically active surface area (AECSA) of nickel electrodes, in situ. The addition of 0.08 mol L−1 of an oxalate salt to an alkaline electrolyte solution shifts the half-wave potential of the Ni(II)/Ni(III) redox pair by about −80 mV. Further, these peaks are very narrow. The sharpest peak in this work has a full width at half maximum (FWHM) of just 11 mV. This unusual sharpness is attributed to the layered structure of nickel hydroxides and the adsorption of oxalate from the solution on the (001) surface. This is supported by attenuated total reflectance infrared (ATR-IR) peaks measured at 1265 cm−1, 1655 cm−1, and 1713 cm−1, which correspond to mononuclear bidentate oxalate bonded to nickel. At sufficiently fast potential scan rates (≥150 mV s−1), the adsorbed oxalate limits growth of the surface hydroxide to a single layer. During the reverse scan, the surface NiOOH/Ni(OH)2 reduction peak is well-separated from other electrochemical processes and may be used to accurately and precisely measure the AECSA. The error of this method is estimated at < 10%. © 2014 NRC Canada. Published by ECS. This is an open access article distributed under the terms of the Creative Commons Attribution 4.0 License (CC BY, http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse of the work in any medium, provided the original work is properly cited. [DOI: 10.1149/2.0711412jes] All rights reserved.

Nickel-based electrode materials find extensive use in a variety of applications in science and engineering such as electrochemical sensors, 1 and secondary battery electrodes. 2 Nickel is used in electro-catalysis for applications including electrolytic hydrogen production, [3][4][5][6] oxidation of organic compounds, 7 methanogenic microbial electrolysis, 8 the electrochemical treatment of wastewater, 9 and electrosynthesis. 10,11 High surface area nickel electrodes are readily prepared by electrodeposition. 12 Porous, three-dimensional felts and foams also find particular importance in synthesis and energy storage. 10,13 Furthermore, nickel-based alloys are common electrocatalyst materials (e.g., Ni x Al y , 14 Ni x Zn y , 14,15 and Ni x Mo y ). 5,16,17 The evaluation of an electro-catalyst's activity requires that the electrochemically active surface area (A ECSA ) is known. Unfortunately, there is no established method to measure the A ECSA of a metallic nickel electrode. Surface area measurements in electrochemical systems are typically performed with one of the methods outlined by Trasatti and Petrii, who describe in detail eleven in situ and four ex situ techniques. 18 However, the application of these methods can pose experimental challenges, especially for non-noble-metal electrodes such as nickel. Several methods have been used previously to estimate the surface area of nickel electrodes, including the voltammetric formation of a surface oxide/hydroxide layer, 19,20 the measurement of the electrochemical double layer capacitance (C dl ) by electrochemical impedance spectroscopy (EIS), 21,22 and by measuring the faradaic desorption of a surface intermediate. 23 However, all of these methods involve initial assumptions or estimations that limit the confidence of the calculated A ECSA .
Machado and Avaca assumed that the anodic peak at ∼0.3 V vs. the reversible hydrogen electrode (RHE) during a forward voltammetric sweep on metallic nickel corresponds to the formation of a monolayer of α-Ni(OH) 2 . 20 It has since been shown that the primary surface component formed at this potential is indeed α-Ni(OH) 2 . 24 However, the exact thickness of this surface layer has never been determined. Further, non-stoichiometric NiO x also forms in this potential region and the voltammetric behavior as a whole is heavily dependent on the electrode history and pretreatment. 24 Therefore, this method yields only an approximate surface area with limited accuracy and precision.
In contrast, the C dl of an electrode may be measured with high precision. However, the exact value of this capacitance is potentialdependent. 22,25 Furthermore, the C dl of a planar nickel electrode is unknown and therefore must be estimated. Estimated reference values range from 20-25 μF cm −2 , for metallic electrodes in general, 21,22,25 and 40-60 μF cm −2 for oxide covered electrodes. 26,27 Therefore, the accuracy of surface areas calculated from the double-layer capacitance is questionable, with large associated uncertainties. For example, Bockris and Otagawa measured the surface area of perovskite electrodes and estimated the measurement accuracy at ± 100%. 26 Another method used to measure A ECSA values on nickel electrodes involves the measurement of adsorbed hydroxide species formed in the Ni(III) potential region. 27  The total transient charge, Q decay , is plotted against the overpotential from the anodic charging step, η charge . At sufficiently high η charge , Q decay reaches a plateau because the fractional surface coverage of the intermediate approaches unity (θ → 1). Therefore, the limiting Q decay corresponds to desorption of a monolayer of the adsorbed species. By estimating the charge for a smooth electrode, one obtains an estimated maximum Q decay of 420 μC cm −2 . The authors clearly state that this reference value is merely a reasonable "universal accepted standard," which may be used to obtain approximate surface area estimates. 23,27 In addition to the limited accuracy of this reference value, it is also imprecise between measurements because there are two possible oxyhydroxides of nickel, βand γ-NiOOH, with different unit cell parameters and symmetries. 28 The amount of each at the electrode surface is dependent on the electrode preparation and treatment. 2,29 Furthermore, it has been shown that the formation and subsequent reduction of a nickel oxyhydroxide surface layer, necessary for this A ECSA measurement method, can change the surface roughness of metallic nickel. 24 Therefore, this method is unsuitable for metallic electrodes and catalysts. However, it may be used to measure the surface area of oxidized nickel surfaces, such as battery anodes and electro-catalysts for oxygen evolution. Therefore, the current transient decay method is perhaps the best approach developed to date for surface area measurements of nickel electrodes.
The purpose of this work is to develop a new method to measure the surface area of nickel electrodes, in situ, with improved accuracy and precision. This method may provide significant advantages for H788 Journal of The Electrochemical Society, 161 (12) H787-H795 (2014) the development of high surface area nickel electrodes for practical applications, such as electrosynthesis, energy storage, etc. A novel approach is presented that utilizes the addition of solution oxalate and the measurement of a surface-limited process with highly unusual voltammetric characteristics.

Experimental
Electrode preparation.-Voltammetric experiments utilized electrodes prepared from a metallic Ni rod (Alfa Aesar, ≥ 99%, 3.2 mm diameter) embedded in epoxy resin so that only one flat, circular face was exposed to the electrolyte solution. Electrodes that were used for subsequent characterization by Fourier transform infrared (FT-IR) spectroscopy were prepared from metallic Ni foil (Alfa Aesar, 99.5%, 0.787 mm thick) electrodes (1 cm × 2 cm). Electrodes were mechanically polished using 320 grit SiC paper followed by 9 μm polycrystalline diamond, 3 μm polycrystalline diamond (Buehler MetaDi Supreme) and 0.05 μm Al 2 O 3 (Buehler MasterPrep) suspensions, unless otherwise noted. Foil electrodes (for FT-IR) were partially covered with Teflon tape and the lower, exposed portion was fully immersed into the electrolyte solution during electrochemical measurements. Electrodes were rinsed and sonicated in high purity water (Millipore Milli-Q, 18.2 M cm) prior to their use.
Electrolyte solution preparation.-Electrolyte solutions were prepared immediately prior to electrochemical experiments (<4 h) using KOH pellets (Fisher Scientific, ACS grade), oxalic acid dihydrate (Fisher Scientific, ACS grade), and high purity water.
Electrochemical measurements.-Electrochemical measurements were conducted in a standard three-compartment glass cell. The working compartment was air-tight and purged with Ar gas (grade 5.0) or with electrolytically generated H 2 gas (Parker Hannifin H2PEM-165L, 99.9999%). A Pt foil (≥ 3 cm 2 ) served as the counter electrode. A saturated calomel electrode (SCE) served as the reference electrode. This reference was chosen for its stable measurements and fast response, which was important for high scan-rate voltammetry experiments. For most experiments in this work, the sample chamber was purged with H 2 to allow regular measurements of the SCE potential against the reversible hydrogen electrode (RHE) to correct for junction potential effects and drifts. Typical values from these potential measurements were used to estimate the SCE potential for the experiments that utilised Ar as the purge gas. The RE was separated from the working compartment by a glass frit.
Voltammetric measurements were performed with a Princeton Applied Research 273a potentiostat controlled with CorrWare software (v. 3.3c, Scribner Associates, Inc.).
Attenuated total reflectance (ATR) fourier transform infrared (FT-IR) spectroscopy.-The FT-IR measurements were performed at room temperature (22 • C) using a Bruker IF66/s spectrometer equipped with an ATR accessory and a deuterated triglycine sulfate (DTGS) detector. The optical chamber was purged with nitrogen to minimize the interference from water and carbon dioxide. Spectra were collected with OPUS software (v.5.0, Bruker Optik GmbH). Measurements were taken from several points on each sample and averaged.

Results and Discussion
The purpose of this work is to establish a new method to measure the A ECSA of nickel electrodes with high accuracy and precision. The method presented involves the addition of an oxalate salt to an alkaline electrolyte solution and the formation of an adsorbed oxalate layer. However, the method presented in this work is significantly different from those reviewed by Trasatti and Petrii. 18 These literature methods involved measuring the Coulombic charge from the underpotential deposition (UPD) of a metal ad-layer [30][31][32][33] or from the electrochemical adsorption or desorption of an adsorbed surface layer. [34][35][36][37] In the new method presented here, the formation of an oxalate ad-layer is utilized to stabilize the surface nickel hydroxide and limit its growth to  Table I). 38 Table I). 38 In these metal complexes, the oxalate molecule adopts a planar (D 2h ) configuration. However, the free anion in solution adopts a twisted conformation with a dihedral angle close to 90 • (D 2d , Figure 1b). The energy difference between the two conformers is very small, ∼17-25 kJ mol −1 . 39,40 As a ligand, oxalate adopts a variety of twisted conformations (D 2 ) such that the O-O distance adjusts to best match the cationic size. That is, the oxalate anion is planar or near-planar when bonded to small cations, whereas it is twisted when bonded to larger cations. 40 In addition to being a versatile chelating agent, oxalate is known to adsorb on a wide variety of metal oxide, hydroxide and oxyhydroxide surfaces (see Section 3.5 for several examples). As with the solution complexes, oxalate adjusts its dihedral angle, and therefore the O-O spacing, to match the size and spacing of the substrate (e.g., ferrihydrite 41 and anatase). 42 This makes oxalate a very versatile adsorbate compound.
In alkaline media, nickel electrodes typically form a nickel hydroxide surface layer. 24 Therefore, the geometric match between oxalate and nickel hydroxide was then considered (Figure 1c). It was found that the O-O spacing in oxalate compounds and nickel hydroxides are similar (Table II), especially when considering the variability of the O-O spacing in oxalate complexes. It is therefore possible that oxalate may adsorb onto a Ni(OH) 2 (001) surface by substituting surface hydroxide groups and forming five-membered rings. Thus, it was decided that oxalate is a good candidate species for adsorption onto nickel electrodes in alkaline media.
General effect of solution oxalate on the alkaline voltammetry of nickel.-Having established oxalate as a likely candidate for adsorption onto an oxidized nickel electrode, the effect of the addition of an oxalate salt to an alkaline electrolyte on the voltammetry of nickel was measured ( Figure 2). It can be qualitatively assessed that the solution oxalate has only a minor effect on the forward scan up to about 1.35 V RHE . The slight differences in the anodic current density in this potential range may be attributed to either the solution oxalate or simply to experimental errors. It is noted that the voltammetric behavior in this potential region is greatly affected by very small differences in electrode pretreatment and history. 22,24 Therefore, if the oxalate salt does affect the formation of a Ni(II) surface layer in this potential range, its role is presumably a subtle one, such as a distinct chemical process that occurs after the surface layer's electrochemical formation.
Above 1.35 V, the effect of solution oxalate on the voltammetry is marked. The anodic peak at 1.48 V, which corresponds to the oxidation of surface Ni(OH) 2 to NiOOH, is shifted to the left by about 80 mV, its width is decreased, and the peak current density is more than quadrupled. Beyond this unusually sharp peak, there is a slightly increased current density, relative to the scan measured in the absence of solution oxalate, and the onset of the oxygen evolution reaction is observed. The increase may correspond to an asymmetric tail of the Ni(II)/Ni(III) oxidation peak or to slightly enhanced oxygen evolution in the presence of oxalate. In the reverse potential scan, the NiOOH to Ni(OH) 2 reduction peak is likewise shifted to the left by about 80 mV in the presence of solution oxalate. Further, the peak width is greatly reduced, and the peak cathodic current density is increased by about a factor of seven. Continuing in the negative direction, there are no further clear effects of solution oxalate until the hydrogen evolution reaction onsets, near the RHE. Below this potential, there are several possible reactions, including the reduction of any surface oxide/hydroxide layers, the hydrogen evolution reaction, and absorption of hydrogen into the electrode. 24 The effect of solution oxalate on the voltammetry in this potential region is outside the scope of the present work, but it may involve these, and other, electrode processes.
Validity of the voltammetric measurements.-It should be noted that the use of a saturated calomel reference electrode (SCE) in a chloride-free, alkaline electrolyte solution is generally a poor choice. However, when these experiments were initially measured with a mercuric oxide reference electrode (Hg/HgO) the response was too slow to accurately measure the sharp voltammetric peaks shown in Figure 2. Thus, the SCE was employed, which provided considerably improved measurement rates. To minimize contamination, the reference was introduced immediately before measurements and the solution was changed frequently to prevent contamination with chloride anions. The use of the SCE also introduced a junction potential, which was measured by de-aerating the solutions with hydrogen gas and measuring the potential of the reversible hydrogen electrode (RHE). This additionally accounted for the junction potential effect from the addition of the oxalate salt, which was typically ∼10 mV. In turn, it was then considered whether the introduction of hydrogen into the electrochemical system affected the electrochemistry of the system. This was qualitatively assessed by performing identical cyclic voltammetry measurements using either hydrogen or argon as the purge-gas. It is seen in Figure 3 that there are only slight differences between these two measurements. Note that slight potential shifts arise because the exact potential of the SCE reference was unknown in the solution deaerated with argon. Therefore, it is concluded that the characteristics of the anodic peak at 1.4 V in Figure 3 are attributable to oxalate, and are unaffected by the use of the SCE or hydrogen.
The effect of the solution oxalate concentration on the voltammetry is not shown, but as the concentration decreases below 0.08 mol L −1 , the position of the Ni(II)/Ni(III) peak shifts to the right and the width of the voltammetric peak increases. As the oxalate concentration approaches zero, the peak characteristics are similar to the oxalate-free conditions. Therefore, the peak positions and widths shown in Figure 2 may be considered extreme cases. That is, these two peaks illustrate the high and low oxalate concentration limits. In concentrations exceeding 0.08 mol L −1 , there was no observed effect on the peak's position or width. Finally, it was observed that the most reproducible results were obtained by preparing the electrolyte solution immediately before, i.e., within a few hours of, voltammetric measurements. Voltammetric evidence of the adsorption of oxalate.-The effects of several organic molecule additives to alkaline electrolytes on the voltammetry of nickel have been studied previously. 43 Test compounds have included various alcohols, amines and carboxylate species. For the most part, the presence of these species does not have a significant effect on the voltammetry below the formation of a surface NiOOH layer at 1.4-1.5 V. Insofar as the voltammetry is unaffected below 1.35 V, these literature results are similar to the present observations for the addition of solution oxalate. The effect of oxalate anions on hydrogen evolution was not considered At more positive potentials, most of these organic solution species oxidize at the electrode surface, at or slightly below the diffusion-limited rate. 43 This is in contrast with the anodic peak in Figure 2, which is assigned to a surface process. Below the RHE, the adsorption of some organic compounds, including alcohols and amines, 44 can affect the kinetics of the hydrogen evolution reaction. In the potential range used here (E ≥ −0.3 V RHE ), there is a slight change in hydrogen evolution current density when oxalate is added to the electrolyte ( Figure 2). However, the effect is very small and may simply correspond to experimental error, which tends to be high when dealing with Ni metal electrodes. 22,24 The case may be considered in which electrode processes up to the Ni(II)/Ni(III) oxidation peak are unaffected by the presence of solution oxalate. Hence, the usual electrochemical behavior will be briefly discussed. During a forward potential scan up to the anodic peak at 1.4 V, the nickel electrode surface is covered with an oxide/hydroxide bilayer of limited thickness. At high scan rates, the surface layer is composed of α-Ni(OH) 2 and the underlying layer is composed of non-stoichiometric NiO x . At lower scan rates, the surface hydroxide is typically a mixture of α-Ni(OH) 2 and β-Ni(OH) 2 , where the former phase is a layered, hydrated polymorph of the latter, which is typically the more compact and crystalline form. 24,45 It may be postulated that solution oxalate does not directly participate in the surface electrochemistry, but rather it chemically adsorbs to the surface hydroxide as it forms. Therefore the oxidation peak at ∼1.4 V does not correspond to the adsorption or desorption of solution oxalate, but rather to the oxidation of the oxalate-stabilized surface hydroxide. This hypothesis is supported by considerable evidence of solution oxalate adsorbing to various metal oxides, hydroxides and oxyhydroxides (e.g., gibbsite, Al(OH) 3 ). 46 In order to test this idea, an electrode was scanned to 1.35 V in an alkaline, oxalate-containing solution, at which point the electrode was removed and rinsed thoroughly for 30 s with high purity water. The electrode was then placed in an oxalate-free 0.1 M KOH solution and the potential scan was then resumed at the same scan rate. As shown in Figure 4, the oxidation   Table III. peak at 1.4 V remained sharp, despite the absence of solution oxalate. This experiment demonstrates that the position and the sharpness of the Ni(II)/Ni(III) anodic peak do not correspond to a reaction involving solution oxalate, but rather oxalate adsorbed on the electrode surface. It is also observed that during the reverse scan, the reduction peak at 1.33 V also remains sharp in the absence of solution oxalate. This suggests that the oxalate remains adsorbed on the electrode surface, which rules out the possibility of an oxidative stripping process.
FT-IR evidence of the adsorption of oxalate.-To conclusively test the oxalate adsorption hypothesis, nickel foil electrodes were scanned to 1.35 V in alkaline solutions, with and without solution oxalate. Then, IR reflectance measurements were conducted to determine the nature of the surface layer. It has been suggested that the electrochemically formed α-Ni(OH) 2 surface component has a limiting thickness of two layers. 19 The adsorbed oxalate layer may therefore have a similar thickness of about two layers or less. Conventional IR reflectance measurements are incapable of detecting such thin surface components. To overcome this limitation, attenuated total reflectance (ATR) FT-IR was employed. The results of these measurements are shown in Figure 5 and the peak assignments are listed in Table III.
The IR peaks were assigned by consulting previous studies and by first considering the spectra collected from the sample prepared without any solution oxalate (Figure 5b). Previous ATR-IR investigations on several aluminum and iron oxide/hydroxide surfaces show that adsorbed carbonate (CO 3 2− (ads) ) has two strong peaks in this frequency region. These peaks are attributed to the ν 3 asymmetric stretching mode (E ), which is split into two bands in metal carbonates and in adsorbed carbonate species. 47,48 Further, the ν 3 peak position is  [47][48][49] This is consistent with the observed peaks at 1415 cm −1 and 1455 cm −1 . There are also peaks that are assigned to a surface α-Ni(OH) 2 species at 1370 cm −1 and 1600 cm −1 . These modes arise from O-H bending modes of lattice hydroxide and associated water molecules, respectively. Another lattice hydroxide mode, previously reported at 1480-1490 cm −1 , is not observed, but is likely just obscured by the adsorbed carbonate modes. Further, the asymmetry of the peak observed at 1600 cm −1 may be attributed to convolution with a previously reported peak at ∼1630 cm −1 . 50 Finally, although a NiO x component is expected at the electrode surface, NiO x does not have any IR absorption bands in this frequency range. 51 These peak positions and assignments are listed in Table III. Next, the IR spectra collected from nickel electrodes prepared in the presence of solution oxalate were examined and the α-Ni(OH) 2(surf) and CO 3 2− (ads) peaks were assigned (Table III). Note that the absorbance of the adsorbed carbonate modes is considerably less from these samples than it is from those prepared in the absence of oxalate. The remaining IR peaks at 1265 cm −1 , 1655 cm −1 and 1713 cm −1 were then assumed to correspond to an adsorbed oxalate species.
Previously reported IR peak positions of oxalate adsorbed several metal oxide/hydroxide/ oxyhydroxide surfaces are summarized in Table IV. Note that some of these peaks show a pH-dependency. For this work, the oxalate is expected to be fully deprotonated (pH sol = 13, pK a1 = 1.25, pK a2 = 3.81) 52 and so, where applicable, the high pH limit was taken while composing Table IV Table IV. The spectrum shown in Figure 5c clearly shows two of these modes, at 1265 cm −1 and 1713 cm −1 . A third peak is observed at ∼1655 cm −1 , although it is convoluted with the O-H bending modes of the α-Ni(OH) 2(surf) species. The fourth expected mode is not observed, but this may be attributed to a convolution with the adsorbed carbonate modes.
In summary, the voltammetric and ATR-IR measurements presented in this section conclusively demonstrate that during a forward voltammetric scan in an alkaline electrolyte that contains an oxalate salt, oxalate anions are indeed adsorbed on the electrochemically formed nickel hydroxide surface.
Effect of potential cycling on surface layer composition.-The reversibility of the Ni(II)/Ni(III) oxidation and reduction processes was tested by cycling through the anodic and cathodic peaks ( Figure 6). It is known that cycling over this potential range in alkaline solution normally results in the gradual conversion of the reduced electrode mass from α-Ni(OH) 2 to β-Ni(OH) 2 and increases the thickness of the surface hydroxide. 53 In the context of the present work the specification of the surface layer as either the αor β-phase may be an oversimplification. Nevertheless, the concept of a transformation from an initial structure to a more compact, crystalline one may be applied to interpret the results in Figure 6. The peak and half-wave potentials ( Figure  6b) from the first 100 cycles indicate that as the surface hydroxide layer transforms, the electrode potential for the more crystalline material is slightly higher than for the initially formed surface layer. Even after 100 cycles, the effect of solution oxalate is clear and the peaks are >40 mV negative to the oxalate-free positions in Figure 2. The peak widths (Figure 6c) are consistent with the transition from one pair of peaks to another. That is, they support the idea of an initially formed hydroxide that becomes more compact and crystalline upon cycling. Therefore, the experimentally observed peak widths may actually arise from the superposition of multiple voltammetric peaks, which explains the rise and fall of the peak widths in Figure 6c. It is noted that both the peak position and width of the cathodic peak level off more rapidly than those of the anodic peak.
Effect of voltammetric scan rate.-The effect of the potential scan rate on the voltammetry is qualitatively shown in Figure 7. The effect of the scan rate on the positions of the voltammetric peaks is shown in Figure 7c. At scan rates greater than 8 mV s −1 , the anodic peak potential increases at approximately 7.5 × 10 −2 mV [mV s −1 ] −1 (i.e., 0.075 s). In contrast, the cathodic peak position is approximately constant at scan rates greater than about 30 mV s −1 . The effect of the scan rate on the full-widths at half maximum is shown in Figure 7d. Although there appears to be considerable scatter in the data shown in Figure 7d, it should be noted that the uncertainty of the measurement (about ±2 mV) appears as a large scatter relative to the very small full widths at half maximum (FWHMs) measured in this study. The FWHMs follow a similar trend for both the anodic and cathodic peaks, following a linear    [2] Parameters are presented as the mean ± standard error of the regression (N = 47, R 2 = 0.949). Note that the FWHMs of the anodic peak for scan rates below 5 mV s -1 were omitted for this calculation because the widths of the anodic peak increases slightly at such slow scan rates. That being said, even at 1 mV s −1 , the FWHM remains quite sharp at 16 mV.
The hypothesis presented in this work thus far is that the electrochemical formation of a nickel hydroxide surface layer occurs normally, i.e., the same as it would in the absence of oxalate. Solution oxalate is subsequently adsorbed onto the surface. The initial surface hydroxide has been presented as a disordered, exfoliated structure analogous to α-Ni(OH) 2 . However, at slower scan rates, the surface typically consists of a mixture of αand β-Ni(OH) 2 components. The existence of a mixed character surface layer at slower scan rates could explain the slight anodic peak broadening observed for scan rates below 5 mV s −1 . Further, the results of the cycling experiment discussed in the previous section and shown in Figure 6b show that the peak positions that correspond to a better ordered, β-like surface layer are at slightly higher potentials than for the α-like surface hydroxide. This is consistent with the shift to more positive peak positions at very slow scan rates (Figure 7c). Therefore at slower scan rates, the electrochemically formed surface layer has a greater structural order, whereas at sufficiently fast scan rates, the surface layer may be considered to have exclusively α-like character. It is unclear, however, whether β-Ni(OH) 2 could actually form in the presence of oxalate anions, since the lattice spacing is greater than for the α-phase and exceeds the O-O distance of solution oxalate (Figure 1).
The effect of the scan rate on the anodic charge under the Ni(II)/Ni(III) peak was difficult to investigate because of the large anodic background, attributed to the thickening of the underlying NiO x layer and the onset of the OER above 1.23 V RHE . Therefore, the cathodic charge calculated during the reverse scan was instead chosen for investigation. The reverse peak was integrated using a linear background, although the background current was very small during the reverse scan for all scan rates studied in this work. Figure 7e shows that the cathodic charge decreases with increasing scan rate until it reaches a minimum value between 100-150 mV s −1 . This minimum value was calculated from the measurements with scan rates ≥150 mV s −1 and is 336 ± 17 μC cm −2 (mean ± standard deviation, N = 10). The relative standard deviation, 5.2%, is quite small and may be largely attributed to the mechanical polishing regime used between measurements.
The larger Coulombic charges calculated at slower scan rates indicate that a thicker surface layer is formed at ν < 100 mV s −1 . The y-intercept is estimated between 1150-1200 μC cm −2 , which corresponds to a surface hydroxide that is 3-4 layers thick. This estimate utilizes the theoretical value of 195 μC cm −2 and the experimentally measured electrode roughness of 1.78 that are calculated and discussed in section 3.8. Since the exact thickness is unknown, the low scan rate limit is unsuitable for surface area estimations.
The effect of slower scan rates was further examined by first scanning electrodes at a very slow scan rate, 2 mV s −1 , from −0.3 V to +1.35 V RHE . Second, the scan was continued at 2, 20, or 200 mV s −1 (Figure 8). The results demonstrate that the peak and half-wave potentials are determined by the scan rate before the Ni(II)/Ni(III) oxidation peak, as well as the scan rate during the peak itself. The Coulombic charge, however, is predominantly determined by the scan rate before the peak, rather than the scan rate during the peak itself. This is consistent with the interpretation that, at slower scan rates, a thicker layer with mixed α/β-Ni(OH) 2 character forms, whereas at sufficiently fast scan rates a monolayer of α-Ni(OH) 2 is formed. However, this interpretation of the slow scan rate results remains unsatisfactory. In particular, the results in Figure 7d appear contradictory with those in Figure  7e. Whereas the FWHM is quite small at very low scan rates, which usually corresponds to a single layer, the Coulombic charge suggests that the thickness exceeds a monolayer. How then can a multi-layer structure give rise to such a sharp peak? To answer this, one may consider the unusual nature of α-Ni(OH) 2 , especially its tendency to form unique layered structures. There are numerous literature reports of α-type nickel hydroxides intercalated with organic anions, including cetyltrimethylammonium bromide, 54 dodecyl sulfate, 9,54-61 dodecylbenzene sulfonate, 59 p-aminobenzoate, 62 and Tween 80. 54 Several of these intercalation materials maintain the lattice vibrational modes of the α-Ni(OH) 2 parent structure. Furthermore, the Ni-Ni interatomic spacing is also very close to that of the parent materials. However, the presence of intercalated anions generally increases the interlayer spacing of the Ni(OH) 2 sheets, which minimizes the chemical interactions between layers. Therefore, the thicker surface layer that forms at slow scan rates may be better described as a supramolecular assembly of many highly ordered monolayers, loosely held together by water and oxalate. This interlayer region is expected to have good ionic conductivity, which would facilitate a rapid electrochemical transformation. This proposed multi-layer structure of oxalate-intercalated α-Ni(OH) 2 may well have a slightly different redox potential than the rapidly formed monolayer on the electrode surface. Furthermore, it would indeed have a greater Coulombic charge because of the increased number of Ni atoms. Finally, the high structural order of the individual Ni(OH) 2 sheets could still give rise to very sharp voltammetric features, due to rapid transformation from Ni(II) to Ni(III).
A new method of surface area measurements on nickel electrodes.-The voltammetric peaks in this work are exceptionally narrow, despite being measured on mechanically polished polycrystalline nickel. Previously reported peaks of similar sharpness involve surface layers formed on single crystal electrodes and are treated as rapid phase transitions at the surface. 63 For example, the FWHM of Pb underpotential deposition onto Pt (111) in the presence of HI can be as little as 5 mV. 64 In another example, voltammetric peak widths reported from the reductive desorption of several alkanethiols on Au(111) are ∼20 mV. 65 The sharpness of these peaks may be attributed to the formation or deformation of highly ordered surface layers on a single crystal face. This is commonly represented by the Frumkin adsorption isotherm, which introduces a lateral interaction parameter, f, to the Langmuir isotherm. Sharp voltammetric peaks occur when these lateral interactions are attractive in nature (f < 0). 66,67 For sufficiently strong lateral interactions, voltammetric sweeps may induce a rapid phase transformation at the surface. 63,67 In general, electron transfer involving more than a monolayer or occurring on multiple crystal faces does not give rise to a single sharp peak. Moreover, such sharp peaks generally correspond to a quick transition from very low (θ ≈ 0) to very high (θ ≈ 1) surface coverage, rather than electron transfer involving only part of the surface layer. 66 These results are consistent with the oxidation and reduction on a particular surface orientation of Ni(OH) 2 , which may be logical upon examination of the material's structure. Both αand β-phases are composed of hexagonal sheets of Ni(OH) 2 . If oxalate adsorbs onto the surface hydroxide as it forms, it would likely do so on the (001) face by substitution of lattice hydroxide sites. This may prevent the growth of adjacent layers, limiting growth of the Ni(OH) 2 sheets to within the plane defined by the crystallographic a-and b-axes. Thus, the resulting structure may be very generally represented as Ni(OH) 2-x (C 2 O 4 ) x(ads) , where 0 ≤ x ≤ 2. This stoichiometry merely states the limiting cases of hydroxide substitution, whereas further investigation is necessary to determine the true value of x.
From the cycling experiment discussed in Section 3.3, and the scan rate experiments discussed in Section 3.4, it is known that at sufficiently high scan rates, the surface layer is analogous to α-Ni(OH) 2 and from the layered structure, one may assume that the surface area of this layer corresponds to the (001) orientation. It is reasonable to assume that the oxidation and reduction peaks correspond to the Ni(II)/Ni(III) redox pair and are therefore one-electron processes. One may therefore use the limiting charge shown in Figure 7e and the unit cell parameters of α-Ni(OH) 2 to estimate the surface area of the polished electrode. From the hexagonal structure of the Ni(OH) 2 sheet, the area for each Ni atom is: Where d is the Ni-Ni spacing [3.08 Å for α-Ni(OH) 2 and a = 3.13 Å for β-Ni(OH) 2 ]. 28,68 Therefore, the calculated area is 8.22 Å 2 atom −1 , or 195 μC cm −2 for a planar electrode. Thus, the roughness factor of the mechanically polished electrode in this work is 1.72 ± 0.09 (mean ± standard deviation), which is a reasonable value. Note that the Ni-Ni spacing of β-Ni(OH) 2 is only 0.05 Å greater than in the α-Ni(OH) 2 structure, which corresponds to a 3% difference in the estimated roughness factor (1.78). Furthermore, the Ni-Ni spacing of intercalated α-derivative materials is usually very close, within a few percent, to the spacing of the α-Ni(OH) 2 parent material. Therefore, the combined measurement error is estimated to be less than 10%. Clearly, further investigation is required to conclusively determine what occurs at slow scan rates. However, the low scan rate behavior does not affect the interpretation of the high scan rate results. That is, when ν ≥ 150 mV s −1 , solution oxalate limits the growth of the surface layer to a single layer, which can be utilized to estimate the value of A ECSA . Therefore the addition of oxalate to an alkaline solution may be used to measure the electrochemically active surface area, in situ, with good accuracy and precision.
There remain several potential obstacles that require further experimentation before this method may be applied to real electrode systems. The selectivity of this method to nickel is unknown, which may affect measurements of mixed metal (Ni x Zn y , etc.) and supported Ni electrodes. Further, it is unknown whether this method will be affected by prolonged hydrogen evolution, which is known to incorporate hydrogen atoms into the electrode material and, thus, alter the surface state. 24 It is also uncertain how the experimental parameters, such as potential scan rate and oxalate concentration, would need to be altered for measuring very high surface area electrodes (Ni foams, electrodeposited coatings, etc.). Alternatively, future work may utilize the method presented in this work on low surface area nickel electrodes to obtain reference values for the double-layer capacitance, C dl , which may be measured on high surface area materials by EIS or potential-step methods. Once reference values are established, it will be possible to calculate the A ECSA from the measured C dl values.

Conclusions
The purpose of this article is to develop a new method to measure the electrochemically active surface area of nickel electrodes, in situ. The method involves the addition of an oxalate salt to an alkaline electrolyte solution and the formation of an adsorbed oxalate layer at the electrode surface. The oxalate anion was selected because it is a versatile chelating ligand that is known to adsorb on numerous metal oxides, hydroxides and oxyhydroxides. Further, the O-O spacing is a good match with the lattice spacing of the nickel hydroxides that form on metallic nickel during a forward voltammetric sweep. The formation of an oxalate ad-layer is utilized to stabilize the surface nickel hydroxide and to limit its growth to a single layer. The faradaic oxidation and reduction of this stabilized Ni(OH) 2 and NiOOH surface layer is measured and the corresponding Coulombic charge is then used to calculate A ECSA .
First, the effects an oxalate salt added to the alkaline electrolyte solution on the voltammetry of nickel are explored. The addition of 0.08 mol L −1 oxalate shifts the half-wave potential of the Ni(II)/Ni(III) redox pair by about −80 mV. Further, the presence of solution oxalate decreases the corresponding voltammetric peak widths considerably. The narrowest peak observed in this work has a FWHM of just 11 mV, despite the use of mechanically polished polycrystalline nickel. Next, it is demonstrated that solution oxalate adsorbs onto the oxidized nickel electrode surface. An electrode was scanned in an oxalatecontaining solution up to just before the sharp oxidation peak at ∼1.4 V, removed, and rinsed with water. The electrode was then placed in an oxalate-free solution and the voltammetric sweep was continued. Despite the absence of solution oxalate, the oxidation and reduction peaks remain shifted and very narrow, which indicates that oxalate is adsorbed on the electrode surface. The presence of this adsorbed oxalate is also confirmed by ATR-IR absorption peaks at 1265 cm −1 , 1655 cm −1 , and 1713 cm −1 , which correspond to mononuclear bidentate oxalate ligands. Next, the effect of the voltammetric scan rate on the ad-layer formation was tested. At slower scan rates (<150 mV s −1 ), the composition of the surface hydroxide is unknown, but may be an oxalate-intercalated α-Ni(OH) 2 material or it may have mixed α/β-Ni(OH) 2 character. This material is estimated to be between 3-4 layers thick. At higher scan rates (≥150 mV s −1 ), a monolayer of Ni(OH) 2-x (C 2 O 4 ) x(ads) forms.
Finally, the new method for measuring A ECSA is presented with a discussion of the accuracy and precision of this new technique. When scan rates of 150 mV s −1 or greater are employed, a monolayer of Ni(OH) 2-x (C 2 O 4 ) x(ads) forms. Although the Ni(II)/Ni(III) oxidation peak overlaps with the OER at higher scan rates, the cathodic peak is very well separated from other electrode processes. Therefore, this peak is easily integrated with high precision using a simple linear background. Since this oxidation process apparently corresponds to a one-electron process for a single layer of an α-Ni(OH) 2 -like surface layer, the unit cell parameters may be used to correlate experimentally measured Coulombic charge with an electrochemically active surface layer. By assuming a Coulombic charge of 195 μC cm −2 , the roughness factor of the electrodes used in this work was calculated to be 1.78 ± 0.09 (mean ± standard deviation).
In conclusion, this work has successfully established a method to measure the surface area of a Ni electrode by simply adding oxalate to the alkaline electrolyte. The error of the measurements is estimated at less than 10%. Furthermore, this error is primarily attributed to variations that arose from the mechanical polishing step between measurements, rather than imprecision of the measurement method.