ATR-SEIRAS Method to Measure Interfacial pH during Electrocatalytic Nitrate Reduction on Cu

An ATR crystal served as the support for the thin film Cu working electrode (Fig. S1). The 60° germanium (Ge) ATR crystal (013-3132, Pike Technologies) was polished with diamond suspensions of 15, 3, and 1 μm (MF-2051, MF-2059, and MF-2054, BASi) using TexMet C pads (401102, Buehler) for the 15 μm polish and MicroCloth pads (407212, Buehler) for the 3 and 1 μm polishes. The crystal was sonicated in water in between each polishing step. Nanopure water (resistivity: 18.2 MΩ · cm, Millipore Sigma) was used for all experiments and measurements. The crystal surface was cleaned with water and isopropyl alcohol using lint-free wipes and blow-dried with nitrogen (N₂). The crystal was cleaned with oxygen (O₂) plasma for 3 min on high power (Pico, Diener Electronic). A thin film (<100 nm) of Cu was deposited on the crystal surface by contacting the surface with a plating solution (40 mM cupric sulfate (CuSO₄·5H₂O) + 0.2 M ethylenediaminetetraacetic acid (EDTA) + 0.2 M glyoxylic acid + 1 mM 2,2'-bipyridine, adjusted to pH 13 with sodium hydroxide (NaOH)) at 70 °C for 4 min. ²⁰ The deposition was performed in a custom-designed cell ²¹ (Fig. S1) in a water bath to maintain a constant temperature for both the crystal and the plating solution. To confirm a successful deposition, the resistance across the surface of the thin film was measured by a multimeter and verified to be consistent (1–2 Ω for all synthesized films in this study).

ATR–SEIRAS experiments were performed on a Thermo Scientific Nicolet iS50 FTIR Spectrometer with a Pike Technologies VeeMax III attachment in a custom, single-chamber electrochemical cell (Fig. S1). ²¹ The counter electrode was a graphite rod (40766, Alfa Aesar), and the reference electrode was a leakless Ag/AgCl electrode (3.4 M potassium chloride (KCl), ET072-1, eDAQ). All currents were converted to current density by dividing by the geometric surface area of the working electrode (2 cm²). All potentials were converted to the reversible hydrogen electrode (RHE) scale using the bulk pH unless otherwise noted. Each thin film Cu electrode was electrochemically conditioned by three CV scans from −0.40 to 0.50 V_RHE at 50 mv s⁻¹ for three cycles in 0.1 M sodium perchlorate (NaClO₄) (Fig. S2).

SEIRAS spectra for calibration and constant current experiments were an average of 32 spectra taken at a resolution of 4 cm⁻¹ for a total duration of 22 s. SEIRAS spectra for pulsed current experiments were an average of 4 spectra taken at a resolution of 16 cm⁻¹ for a total duration of 4 s. The pulsed current method on the potentiostat and the ATR–SEIRAS time series were started at the same time and the timestamp of each spectrum was correlated with the time recorded by the potentiostat to ensure the methods were synchronized. A background spectrum of water was taken at the start of each set of experiments. The negative logarithm of the ratio between the single-beam sample spectrum (R) and the single-beam background spectrum (R₀) gives the absorbance (A) spectrum of the sample in absorbance units (a.u.): $A=-\mathrm{log}(\tfrac{R}{{R}_{0}})$.

The bulk pH was measured with a pH meter (FiveEasy, Mettler Toledo) before and after each experiment. At the conclusion of each experiment, chronoamperometry (CA) was performed at 0.1 V_RHE and a SEIRAS spectrum was collected after 1 min to check the stability of the film.

Interfacial pH calibration data were gathered in 0.5 M phosphate buffer solutions during CA at 0.1 V_RHE after 1 min. The pH was adjusted in increments of approximately 0.2 pH units by adding 0.1 M NaOH or 0.1 M phosphoric acid (H₃PO₄). All experiments, including calibration, were performed at room temperature, because K_eq changes as a function of temperature.

The four species that comprise the phosphate buffer system are H₃PO₄, dihydrogen phosphate (H₂PO₄ ⁻), hydrogen phosphate (HPO${}_{4}^{2-}$), and the phosphate ion (PO${}_{4}^{3-}$). The phosphate buffer equilibrium reactions are

$$\begin{eqnarray}&&{{\rm{H}}}_{3}{\mathrm{PO}}_{4}\rightleftharpoons {{\rm{H}}}^{+}+{{\rm{H}}}_{2}{\mathrm{PO}}_{4}^{-}\end{eqnarray} \tag{ 1 }$$

$$\begin{eqnarray}&&{{\rm{H}}}_{2}{\mathrm{PO}}_{4}^{-}\rightleftharpoons {{\rm{H}}}^{+}+{\mathrm{HPO}}_{4}^{2-}\end{eqnarray} \tag{ 2 }$$

$$\begin{eqnarray}&&{\mathrm{HPO}}_{4}^{2-}\rightleftharpoons {{\rm{H}}}^{+}+{\mathrm{PO}}_{4}^{3-}\end{eqnarray} \tag{ 3 }$$

The relationship between pH and the calibration ratios is derived from the equilibrium Eqs. (Table S1). For example, rearranging the equation for K₁, we find that the pH is proportional to the log of the calibration ratio (derivation shown in the Supplemental Material).

$$\begin{eqnarray}&&\mathrm{pH}=-\mathrm{log}\left(\displaystyle \frac{[{{\rm{H}}}_{3}{\mathrm{PO}}_{4}]}{[{{\rm{H}}}_{2}{\mathrm{PO}}_{4}^{-}]}\right)-\mathrm{log}\left({{\rm{K}}}_{1}\right)\end{eqnarray} \tag{ 4 }$$

Three calibration regions were defined: acidic (pH 1–3), neutral (pH 5–9), and basic (pH 9–13). Representative ATR–SEIRAS spectra in each of the three pH regions are shown in Fig. 1. In each region, the calibration curve is based on the ratio of the peak areas of the two phosphate species with the highest concentration in that region (Fig. 2). The relationship between the pH and the calibration ratio is shown in Eq. 4. The calibration ratios, average peak positions, and standard deviation (SD) of peak positions are shown in Table I.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Table I. Calibration ratios for the three calibration regions: acidic (pH 1–3), neutral (pH 5–9), and basic (pH 9–13). Reported peak positions are the average values from the three calibration data replicates. The relationship between the pH and the calibration ratio is shown in Eq. 4.

Calibration Ratios
Region	pH range	Calibration ratio	Peak position/cm−1	Standard deviation/cm−1
Acidic	pH 1–3	$\displaystyle \frac{{\nu }_{s}(\mathrm{PO}){{\rm{H}}}_{3}{\mathrm{PO}}_{4}}{{\nu }_{a}(\mathrm{POH}){{\rm{H}}}_{2}{\mathrm{PO}}_{4}^{-}}$	1008	18.4
			947	2.7
Neutral	pH 5–9	$\displaystyle \frac{{\nu }_{a}(\mathrm{PO}){{\rm{H}}}_{2}{\mathrm{PO}}_{4}^{-}}{{\nu }_{s}(\mathrm{PO}){\mathrm{HPO}}_{4}^{2-}}$	1135	9.6
			990	1.2
Basic	pH 9–13	$\displaystyle \frac{{\nu }_{a}(\mathrm{PO}){\mathrm{HPO}}_{4}^{2-}}{{\nu }_{a}(\mathrm{PO}){\mathrm{PO}}_{4}^{3-}}$	1082	5.2
			1018	6.0

Peaks were identified from literature and confirmed through peak area trends as the pH was adjusted. For example, in the acidic region, the asymmetrical ν_a(POH) vibrations of ${{\rm{H}}}_{2}{\mathrm{PO}}_{4}^{-}$ peak at 947 cm⁻¹¹⁷ increased in area as the pH increased, and the symmetrical ν_s(PO) vibrations of H₃PO₄ peak at 1008 cm^{−1 22} decreased in area with increasing pH (Fig. S3), as expected for phosphate speciation in this pH region. The peak evolution in the acidic, neutral, and basic calibration regions can be seen in Figs. S3, S4, and S5, respectively.

In the acidic pH range (pH 1–3), the calibration ratio was the peak area from symmetrical ν_s(PO) vibrations of H₃PO₄ (1008 cm⁻¹, SD 18.4) ²² to the peak area from asymmetrical ν_a(POH) vibrations of ${{\rm{H}}}_{2}{\mathrm{PO}}_{4}^{-}$ (947 cm⁻¹, SD 2.7). ¹⁷ These two peaks were selected for the calibration ratio because the area trends were consistent with the concentration trends of the individual species (Fig. S3). In contrast, the area trends of the peaks at 1070 and 1155 cm⁻¹ revealed contributions from both H₃PO₄ and ${{\rm{H}}}_{2}{\mathrm{PO}}_{4}^{-}$ that could not be deconvoluted.

In the neutral pH range (pH 5–9), the calibration ratio was the peak area from asymmetrical ν_a(PO) vibrations of ${{\rm{H}}}_{2}{\mathrm{PO}}_{4}^{-}$ (1135 cm⁻¹, SD 9.6) to the peak area from symmetrical ν_s(PO) vibrations of HPO${}_{4}^{2-}$ (990 cm⁻¹, SD 1.2). ¹⁷ The peak at 940 cm⁻¹ was not selected due to low sensitivity (Fig. S4) and the peak at 1080 cm⁻¹ was not used because the area trends revealed contributions from both ${{\rm{H}}}_{2}{\mathrm{PO}}_{4}^{-}$ and HPO${}_{4}^{2-}$ that could not be deconvoluted.

In the basic pH range (pH 9–13), the calibration ratio was the peak area from asymmetrical ν_a(PO) vibrations of HPO${}_{4}^{2-}$ (1082 cm⁻¹, SD 5.2) to the peak area from asymmetrical ν_a(PO) vibrations of PO${}_{4}^{3-}$ (1018 cm⁻¹, SD 6.0). ¹⁷ The peak at 990 cm⁻¹ was not selected due to low sensitivity (Fig. S4).

Due to the similar appearance of the four peaks in the acidic and neutral regions (Fig. 1), a quantitative method using the position of the second peak was established for peak assignment. If the second peak position was ≥998 cm⁻¹ then the peak was assigned to ν_s(PO) H₃PO₄ and the acidic calibration curve was used. If the second peak position was <998 cm⁻¹ then the peak was assigned to ν_s(PO) HPO${}_{4}^{2-}$ and the neutral calibration curve was used. The cutoff value was set to be 2 SD above the average position for ν_s(PO) HPO${}_{4}^{2-}$ from the constant current data (986 cm⁻¹, SD 6.1). The ν_s(PO) HPO${}_{4}^{2-}$ peak was selected because there were more data points than the ν_s(PO) H₃PO₄ peak (n${}_{{\nu }_{{\rm{s}}}(\mathrm{PO}){\mathrm{HPO}}_{4}^{2-}}$ = 319, ${{\rm{n}}}_{{\nu }_{{\rm{s}}}(\mathrm{PO}){{\rm{H}}}_{3}{\mathrm{PO}}_{4}}=26$), and the constant current data had higher variance than the pulsed current data.

Excellent agreement is seen between the normalized peak areas and the relative fraction of each phosphate species (Fig. 2) calculated from the equilibrium constants (K_eq) (Table S1). ²³ The normalized peak areas for all data replicates are shown in Fig. S6, demonstrating the same trend. There is a seamless transition at pH 9 from the neutral to basic region. Due to the low sensitivity of the ν_s(PO) H₃PO₄ peak (Fig. S3), there is a gap from pH 3–5 at the transition from the acidic to neutral region, and the normalized peak area does not follow the calculated trend line quite as well as the other peaks.

Figure 3 shows all calibration data plotted as pH against the log of the calibration ratio, as derived from the equilibrium equation for phosphate speciation (Table S1, Eq. 4). Three complete sets of calibration data were collected for each calibration region (Table S2). The average calibration curves—used to calculate the interfacial pH during electrochemical experiments—are the average of the best fit linear calibration curves for each of the three data replicates in a given calibration region. The slope, intercept, and R² value of each calibration curve are shown in Table S3.

Zoom In Zoom Out Reset image size

The accuracy for the calibration varies with the pH, as shown by the shaded 95% confidence interval (CI) in Fig. 3 (Eq. S1). The margin of error (MOE) expressed as a percentage (Eq. S2) and as a function of pH is shown in Fig. S7. The region with the lowest accuracy is around pH 1, where the 95% CI is the largest. For example, at pH 1.25 the 95% CI is pH 1.01–1.48, a variability of 38%. The 95% CI for the acidic region improves as the pH increases, with a 17% variability at pH 1.75 and a 1% variability at pH 2.5. In the neutral region, the 95% CI is more evenly distributed, with a 30% variability at pH 5.5 decreasing to a 12% variability at pH 9. The basic region has the most consistent 95% CI, with 6%–8% variability across pH 9–13.

This pH-dependent variability demonstrates the importance of using repeat data sets to establish an average calibration curve and report the margin of error for the technique. This statistical analysis defines the accuracy of calculations based on the interfacial pH, such as the overpotential at the electrode interface. It also sets the limits when comparing with other studies and computational models, because we cannot distinguish between results with values within the 95% CI. The variation is due to subtle changes in the thin film morphology, which can occur during the deposition process, electrochemical conditioning, and electrochemical experiments. Because the thin film morphology and material can strongly influence the sensitivity of the SEIRAS spectra, there will also be some variability in the accuracy of this method between researchers. Thus, it is important to adhere to a well-defined experimental procedure for thin film deposition and consider the individual reported accuracy when comparing results between research groups. In the acidic region, the higher margin of error and the gap from pH 3–5 may be improved by changing the material from Cu to Ag or Au, ¹³ decreasing the thickness of the film, and increasing the surface roughness. However, these methods of enhancing the SEIRAS signal must also be balanced with film stability, which is increasingly important at higher current densities. Raising the temperature of the electrolyte would alter the phosphate buffer K_eq values which could also improve the acidic calibration.

${\mathrm{NO}}_{3}^{-}$ reduction was used as an example to demonstrate the capability of this technique to measure the interfacial pH during an aqueous electrochemical reaction. The expected reactions on Cu are ${\mathrm{NO}}_{3}^{-}$ reduction to nitrite (${\mathrm{NO}}_{2}^{-}$) (Eq. 5), ${\mathrm{NO}}_{3}^{-}$ reduction to NH₃ (Eq. 6), and hydrogen (H₂) evolution (Eq. 7). The corresponding reactions in acidic media are shown in Eqs. S3–S5.

$$\begin{eqnarray}&&{\mathrm{NO}}_{3}^{-}+{{\rm{H}}}_{2}{\rm{O}}+2{{\rm{e}}}^{-}\to {\mathrm{NO}}_{2}^{-}+2{\mathrm{OH}}^{-}\end{eqnarray} \tag{ 5 }$$

$$\begin{eqnarray}&&{\mathrm{NO}}_{3}^{-}+6{{\rm{H}}}_{2}{\rm{O}}+8{{\rm{e}}}^{-}\to {\mathrm{NH}}_{3}+9{\mathrm{OH}}^{-}\end{eqnarray} \tag{ 6 }$$

$$\begin{eqnarray}&&2{{\rm{H}}}_{2}{\rm{O}}+2{{\rm{e}}}^{-}\to {{\rm{H}}}_{2}+2{\mathrm{OH}}^{-}\end{eqnarray} \tag{ 7 }$$

${\mathrm{NO}}_{3}^{-}$ reduction to ${\mathrm{NO}}_{2}^{-}$ and HER each have a generation rate of 1 OH⁻ per electron, and ${\mathrm{NO}}_{3}^{-}$ reduction to NH₃ has a generation rate of 1.13 OH⁻ per electron. Cu has been experimentally ^4,24–32 and theoretically ³³ shown to be selective toward NH₃ production, although the selectivity may flip to favor ${\mathrm{NO}}_{2}^{-}$ depending on the applied potential, surface structure, and electrolyte. At the same current density, an electrocatalyst with a higher selectivity toward NH₃ (1.13 OH⁻ per electron) will experience a greater pH polarization than an electrocatalyst with higher selectivity toward ${\mathrm{NO}}_{2}^{-}$ and H₂ (1 OH⁻ per electron).

The time evolution of the interfacial pH during ${\mathrm{NO}}_{3}^{-}$ reduction under constant current conditions was investigated in 0.5 M sodium phosphate buffer electrolyte with a bulk pH of 1.84 (0.25 M H₃PO₄ + 0.25 M sodium dihydrogen phosphate (NaH₂PO₄)) with 10 mM sodium nitrate (NaNO₃) (Fig. 4). ATR–SEIRAS measurements were collected every 22 s for 10 min at four different constant current densities (−3, −5, −6, and −8 mA cm⁻²) and each condition was repeated three times (Table S2). A subset of corresponding spectra are shown in Fig. S10. Corresponding voltages are shown in Table S4. In all constant current experiments the bulk pH was unchanged.

Zoom In Zoom Out Reset image size

As shown in Fig. 4, a gradual increase in the interfacial pH was observed at −3 mA cm⁻², the smallest current density investigated. The interfacial pH increased for approximately 3 min before reaching a plateau of pH 5.7. At −5 mA cm⁻², the interfacial pH increased more quickly over the span of 2 min, reaching a plateau of pH 6.5. Similar behavior was observed for both −6 and −8 mA cm⁻². During the time when the interfacial pH is initially increasing, the potential sharply increases in magnitude before reaching a region of slow growth (Fig. S9). This dynamic period is due to charging of the double layer and changes in the concentration profile in the diffusion layer. ³⁴

The plateau behavior of the interfacial pH demonstrates that the rate of OH⁻ produced by the cathodic reactions is matched by phosphate buffering, maintaining the interfacial pH near pKa₂ = 7.20 (Eq. 2, Table S1). This plateau phenomenon at pKa₂ was also observed by Yang et al. during CO₂ reduction in 0.2, 0.5, and 1.0 M phosphate buffer at a Cu electrode. ¹⁷ While the buffer is necessary for this type of interfacial pH measurement, differences between electrochemical operating conditions will be suppressed near the buffer pKas. Researchers employing this technique may strategically select the buffer bulk pH and concentration based on the pH range of interest with respect to the pKas of the buffer species.

In the data set shown in Fig. 4 (rep 3), there was a statistically significant difference in the final pH (average of the last three data points, ca. 1 min) between −3 mA cm⁻² and the larger current densities. When this interfacial pH data is plotted against charge passed (Fig. S11), the interfacial pH at −3 mA cm⁻² during the initial ramp period is still lower than the interfacial pH at the higher current densities at the same charge passed, indicating that the lower current density condition may slightly suppress the pH polarization at the interface. However, when all three data sets are averaged (Fig. S12), there is no statistically significant difference in the final pH between these four current densities. This result indicates that, if there is a slight difference in the interfacial pH between −3 to −8 mA cm⁻², it is within the margin of error for this technique. At all four current densities, the plateau pH represents a significant increase in the interfacial pH of 5 pH points on average from the bulk pH (pH 1.84–7.15), even in a heavily buffered electrolyte.

These results demonstrate the importance of a technique that can quantify the pH so close to the electrode–electrolyte interface, often revealing a greater pH polarization than is measured by other methods. This pH polarization perturbs the reaction thermodynamics and kinetics, and under-estimating the extent of the pH gradient can lead to incorrect mechanistic interpretations and an overestimation of the catalytic overpotential due to polarization loss. ^2,8,17 A more precise calculation of the overpotential at the electrode surface enables a more accurate comparison between different studies. For example, scanning electrochemical microscopy (SECM, a type of SPM) was used to measure the local pH during ${\mathrm{NO}}_{3}^{-}$ reduction at a Cu electrode by Santos et al. ³⁵ The IrOx-coated Au microelectrode tip had a radius of 10 μm and the closest achievable measurement to the surface was 10 μm. In a 0.1 M sodium sulfate (Na₂SO₄) solution with 0.5 mM ${\mathrm{NO}}_{3}^{-}$ and a bulk pH of 2 at −1.0 V_Ag/AgCl (O(-1 mA cm⁻²)), the local pH at a distance of 10 μm was found to be pH 5.8. Considering this value is lower than the results presented here in a 0.5 M buffered solution at a similar current density (average pH 6.6 at −3 mA cm⁻², Fig. S12), we can predict that the interfacial pH in 0.1 M Na₂SO₄ was likely higher than pH 5.8 at the electrode–electrolyte interface. The SECM measurement at 10 μm from the electrode surface underestimates the actual interfacial pH, which we can more accurately quantify through ATR–SEIRAS because the measurement distance is three orders of magnitude closer to the electrode–electrolyte interface: just 10 nm rather than 10 μm.

A more accurate measurement of the interfacial pH under reaction conditions is important because the proton concentration can impact both the selectivity and kinetics of electrochemical reactions. In another report of ${\mathrm{NO}}_{3}^{-}$ reduction on Cu, Pérez-Gallent et al. found that acidic media favored the formation of NH₃ and nitric oxide (NO), while alkaline media favored ${\mathrm{NO}}_{2}^{-}$ and hydroxylamine (NH₂OH). ⁴ Wang et al. performed a computational study of ${\mathrm{NO}}_{3}^{-}$ reduction on Cu and confirmed higher selectivity toward NH₃ in acidic conditions. ⁶ If NH₃ is the desired product of ${\mathrm{NO}}_{3}^{-}$ reduction, then an acidic reaction environment is preferred.

Potential strategies to maintain such acidic conditions include buffered electrolytes, mass transport control, and pulsed electrolysis. Clearly, even highly concentrated buffers cannot maintain the desired acidic conditions at the electrode–electrolyte interface, as demonstrated in this study using a 0.5 M buffer and the report by Yang et al. measuring the interfacial pH during CO₂ reduction in 0.1–1.0 M buffers. ¹⁷ Guo et al. studied the effect of mass transport on ${\mathrm{NO}}_{3}^{-}$ reduction at a Ti electrode by varying the flow rate. ⁷ As the flow rate increased, the diffusion layer thickness decreased, enhancing the nitrate diffusion driving force. These changes increased nitrate reduction activity, producing more ${\mathrm{NO}}_{2}^{-}$, NH₃, and OH⁻ (Eqs. 5 and 6). While the smaller diffusion layer thickness would also enhance the OH⁻ diffusion driving force away from the electrode surface, a continuum model simulation found that the interfacial pH actually increased with increasing flow rate. This pH trend demonstrates that improving the mass transport increases the production of OH⁻ more than it enhances the dissipation of OH⁻ under the experimental conditions. In the next Section we explore the potential application of pulsed electrolysis to control the interfacial pH.

The time evolution of the interfacial pH during ${\mathrm{NO}}_{3}^{-}$ reduction under pulsed current conditions was investigated in 0.5 M sodium phosphate buffer electrolyte with a bulk pH of 1.84 (0.25 M H₃PO₄ + 0.25 M NaH₂PO₄) with 10 mM NaNO₃ (Fig. 5). Experiments were conducted for 10 min at four different pulsed current conditions (-5 mA cm⁻² for 8 s and −5, −6, and −8 mA cm⁻² for 4 s), and each condition was repeated three times (Table S2). A subset of corresponding spectra are shown in Fig. S13. Corresponding voltages are listed in Table S5. During pulsed current experiments, the applied current was a symmetric square wave alternating between current on and current off (zero current, open circuit). For example, at the −5 mA cm⁻², 4 s pulse condition, −5 mA cm⁻² was applied for 4 s followed by open circuit for 4 s, and repeated over the course of 10 min. An example of the pulsed current and corresponding voltage is shown in Fig. S14. ATR–SEIRAS spectra were collected every 4 s, one for each pulse under the 4 s conditions and two for each pulse under the 8 s conditions. In all pulsed current experiments the bulk pH was unchanged.

Zoom In Zoom Out Reset image size

As shown in Figs. 5a–5d, when the current was off (open circuit) the interfacial pH consistently returned to a value close to that of the bulk pH for all conditions, regardless of the cathodic current magnitude. When the current was on, each condition showed an initial ramp period where the pH steadily increased before reaching a plateau pH for the remainder of the 10 min experiment. The duration of these ramp periods (0.3–1.5 min) were very similar to those observed in the constant current experiments (Fig. 4). Across all pulsed current data replicates, there was a statistically significant trend of the pH ramp time decreasing with increasing pulsed current density (Fig. 5e and Fig. S15). The pH ramp times from the constant current experiments were not statistically analyzed because the longer ATR–SEIRAS measurement time (22 s for constant current versus 4 s for pulsed current) did not have sufficient time resolution to capture the switch from the acidic to neutral region.

The average interfacial pH of the final minute is shown for each data replicate and the average of all three data replicates in Fig. S16, demonstrating no statistically significant difference in final pH between the different pulsed conditions. We selected the final minute as a standard of comparison between data sets and methods because it was the furthest from the dynamic pH ramp period observed at the start of experiments. In Fig. 5f, the average final pH for the constant current and pulsed current conditions is compared. While the average value of the final pH for the 4 s pulsed current at −5, −6, and −8 mA cm⁻² was lower than the average value of the final pH for the corresponding constant currents, it was not statistically significant. Thus, while the pulsed current method may slightly suppress the increase in interfacial pH, it is within the margin of error for this technique. There is a transient period when the current is switched from off to on when the interfacial pH must be increasing from the near-bulk value to an average pH of 6.1, but this change occurs faster than the 4 s measurement time.

Pulsed electrolysis has been explored as a method to suppress the concentration polarization that occurs in the diffusion layer during electrocatalysis by promoting ion diffusion that periodically renews the electrode–electrolyte interface, ^9,12,36,37 including during ${\mathrm{NO}}_{3}^{-}$ reduction. ^7,38–40 Guo et al. found that pulsed electrolysis (10 s symmetric pulse) at a Ti cathode during ${\mathrm{NO}}_{3}^{-}$ reduction in a non-buffered electrolyte showed an improvement in NH₃ selectivity but also an increase in H₂ production when compared to constant potential experiments under the same conditions, and had similar product selectivity to a 0.5 M phosphate buffer with a bulk pH of 10.5. ⁷ The authors concluded that the pulsed electrolysis in the non-buffered electrolyte resulted in an interfacial pH similar to that of the 0.5 M phosphate buffer and lower than the interfacial pH under a constant applied potential. While this study finds no significant difference in the interfacial pH between the constant current and pulsed current conditions in 0.5 M phosphate buffer, it is possible that the pH polarization in a non-buffered solution is so much greater than in a buffered solution that the pulsed electrolysis is effective at maintaining a lower interfacial pH. Alternatively, the transient lower interfacial pH conditions achieved each time the current is turned back on may be enough to impact the product selectivity. Finally, it may be that resetting the interfacial concentration of ${\mathrm{NO}}_{3}^{-}$ and other electrolyte species accounts for the selectivity changes. Huang et al. similarly found an enhancement in nitrate-to-ammonia selectivity at a carbon-supported ruthenium (Ru) and indium (In) intermetallic compound (RuIn₃) during pulsed electrolysis (4 s cathodic, 0.5 s anodic) in a non-buffered electrolyte and used Raman spectroscopy to demonstrate the restoration of the ${\mathrm{NO}}_{3}^{-}$ concentration at the interface during the anodic pulse. ⁴⁰ The open question of how the pulsed electrolysis is impacting product selectivity in these two studies highlights a challenge with the ATR–SEIRAS technique in that it must be performed in a buffered electrolyte; we cannot confirm the interfacial pH under the non-buffered condition.

Despite this limitation, with the ATR–SEIRAS technique we are able to confirm that the electrolyte renewal does indeed reach the first 10 nm of the electrode–electrolyte interface (Fig. 5). Xin et al. used a fluorescent probe technique to show a restoration of the initial local pH conditions at 1 μm from the electrode–electrolyte interface during chromate (${\mathrm{HCrO}}_{4}^{-}$) reduction to the chromic ion (${\mathrm{Cr}}_{3}^{+}$) (2.3 OH⁻ per e⁻). ¹² The fluorescent probe method enabled spatially resolved visualization of the diffusion layer (40–80 μm) with a resolution of 200 nm. In the unbuffered electrolyte at 1 μm of the surface, the local pH increased from pH 3 to pH 5 when the voltage was pulsed on and returned to pH 3 when the voltage was pulsed off (5 s and 2.5 s symmetric pulse patterns, no current density reported). Considering our buffered system that produces 1–1.13 OH⁻ per electron shows an even higher pH increase (average pH 6.1, Fig. 5f), it is likely that the fluorescent probe technique is not capturing the full pH polarization at the interface. This finding demonstrates the value of the ATR–SEIRAS method to better understand the environment at the electrode–electrolyte interface.

While pulsed electrolysis appears to be effective at restoring the interfacial electrolyte conditions while the current is off, the concentration polarization quickly returns when the current is applied. Thus, pulsed electrolysis may beneficially impact product selectivity by periodically increasing reactant concentrations and decreasing product concentrations at the interface, but it is not effective at maintaining a higher H⁺ concentration, at least at a 4 s symmetric pulse rate in 0.5 M phosphate buffer. Xin et al. saw a decrease in the pH polarization (pH 4.5 instead of 5) within 1–2 μm of the surface when the pulse time was decreased to 1 s. ¹² However, the local pH under these conditions increased over the course of 30 s, similar to the pH ramp observed in this study (Fig. 5e). Had the authors continued the measurement beyond 30 s, it is possible that the local pH at the 1 s pulse time may have reached the same level as the 2.5 and 5 s pulse times. While decreasing the pulse time may be a viable strategy to controlling the interfacial pH, there are limits. The pulse duration must be longer than the charge and discharge times of the double layer, otherwise the current will not reach the desired Faradaic current when pulsed on and it will not reach zero current when pulsed off. ^9,37 Future work to decrease the measurement time for the ATR–SEIRAS technique will enable further exploration into the possibility of using pulsed electrolysis to maintain a desired interfacial pH.