Plasmonically enhanced electrochemistry boosted by nonaqueous solvent

Noble metal nanostructures have unique optical properties owing to their strong interaction with light.^1,2 As a result, noble metal nanoparticles are utilized in the fields of photovoltaics,^3–7 heterogeneous catalysis,^8–12 photodetectors,^13–17 and sensors.^18–22 The interaction of the time varying electric field of light with noble metals with at least one length dimensions shorter than the wavelength of light results in a collective oscillation of the free electrons of the metal, a phenomenon known as a localized surface plasmon resonance (LSPR). Excitation of LSPRs at the resonant electron oscillation frequency concentrates light at the surface of the metal, leading to local electromagnetic fields orders of magnitude larger than the excitation source.^23–25 These large and localized electromagnetic fields are the basis for many surface-enhanced spectroscopies.^26–30 Following excitation, LSPRs decay by emission of light at the plasmon frequency into the far-field or by nonradiative channels.^31,32 In the latter decay pathway, the photoexcited electrons dephase and are dampened by electronic transitions in the metal (e.g., inter- and intraband transitions), scatter from electrons and phonons, and eventually release energy in the form of heat to the medium that surrounds the plasmonic material. Nonradiative plasmon decay thus leads to the formation of hot charge carriers (i.e., electrons and holes in excited, nonequilibrium states) and localized heating, both of which have been utilized to drive chemical transformations.^10,32–36 Although localized heating is modest in many cases, temperature rises can impact surface chemistry and reaction rates.^37–40 However, the generation of hot charge carriers from plasmon excitation has been shown to have profound effects on the activity and selectivity of chemical reactions.^9,41–45

Hot electrons and holes generated by plasmon excitation can be transiently or irreversibly transferred to chemical adsorbates. Transient transfer provides a route to vibrationally activate adsorbates for chemical transformations.^46–49 Irreversible charge transfer leads to chemical transformations driven by redox reactions.⁵⁰ If hot electrons and holes are irreversibly transferred and their transfer rates are asymmetric, plasmonic materials can accumulate one type of charge carrier, leading to a modulation in the Fermi energy of the material. A change in the Fermi energy of a plasmonic material by light excitation and asymmetric charge transfer results in a photopotential on the material.⁸ This photopotential can be tuned by the properties of the plasmonic material (e.g., electronic structure, size, and morphology), the attributes of excitation light (e.g., intensity and frequency), and the rates of charge transfer.^23,31,51,52 Photoinduced voltages can also be established and controlled by other mechanisms, such as optical rectification,^53–55 photothermoelectric effects,^56–58 and plasmoelectric effects.^59,60

Chemical electron and hole scavengers have been effectively used to control asymmetric charge transfer and build up photopotentials on plasmonically excited materials.^{10,33,61–67} An emerging alternative is to employ electrochemistry, whereby an applied electric bias allows precise tuning of the Fermi energy of plasmonic materials and quenching of hot charge carriers.^{43–45,51,68–73} For example, Wilson and co-workers showed that plasmonically generated photopotentials enhanced the electrochemical hydrogen evolution reaction (HER) in an acidic medium by reducing the overpotential for the reaction and increasing interfacial electron transfer.⁶⁹ Kuo and co-workers plasmonically excited Ag nanocubes and Ag nanooctahedra to enhance the electrochemical HER in an acidic medium with ethanol added as a sacrificial agent.⁷² They demonstrated that both Ag nanoparticle morphologies enhanced the HER under light irradiation and that nanooctahedra with (111) facets produced higher current densities than nanocubes with (100) facets. A study by Mo and co-workers showed that core–shell nanostructures of Au–Ag are more efficient than commercial Pt catalysts for electrochemical hydrogen production under visible light excitation.⁷³ Beyond the HER, plasmonically excited electrodes have been employed to improve electrochemical oxygen reduction,⁵¹ ammonia oxidation,⁷⁰ and carbon dioxide reduction,⁷¹ among others. These examples highlight the promise of synergizing plasmonics and electrochemical transformations in aqueous media.

One limitation of the use of H₂O as a solvent in electrochemical transformations is the low solubility of many organic reagents and nonpolar gases. With the growing interest in using electrochemical methods for energy conversion, synthesizing chemical feedstocks, and pharmaceuticals,^74–79 there is an opportunity to expand the scope of plasmon-enhanced electrochemistry to nonaqueous solvents. Furthermore, nonaqueous solvents offer a wider range of solvent polarities and dielectric functions of media surrounding plasmonic materials, which may offer additional control over harvesting plasmonically generated charge carriers to improve the activity and selectivity of chemical transformations. Nonaqueous solvents are commonly added in relatively low concentrations to aqueous solutions to serve as sacrificial hole or electron scavengers in plasmonic photocatalysis.^80–82 For example, Bian and co-workers used a 2-propanol/H₂O mixture to evolve hydrogen from Au/TiO₂ superstructures, where 2-propanol served as a hole scavenger and electrons generated from plasmon excitation were injected into TiO₂ and eventually extracted to evolve hydrogen.⁸⁰ In another study, Kim and co-workers used different liquid organic additives in aqueous solutions to study how different hole scavengers affect the reduction rate of a model reaction driven by plasmon excitation.⁸² However, there is a knowledge gap in understanding how plasmon excitations affect chemical transformations in systems where the primary component of the solvent is not H₂O. To the best of our knowledge, plasmonically enhanced chemistry or electrochemistry in nonaqueous solvents has not been investigated.

In this Communication, we electrochemically reduce H₂O to H₂ (hydrogen evolution reaction, HER) in acetonitrile on plasmonically excited Au electrodes to demonstrate how plasmon excitation affects electrocatalysis in a nonaqueous medium. Using linear sweep voltammetry, we analyze the difference in the external electric potential required to generate defined current densities under plasmon excitation and dark conditions. Our results show that plasmon excitation develops a photopotential on the Au electrode, which is dependent on solvent polarity. We specifically observe that an increase in the concentration of H₂O in acetonitrile results in a decrease in photopotential. Under plasmon excitation conditions and at a H₂O concentration of 0.5M, a photopotential of ∼175 mV is established on the plasmonic electrode, a 3.5× increase compared to a H₂O concentration an order of magnitude higher. We propose a mechanism where an increase in solvent polarity (i.e., H₂O concentration) leads to a stabilization of the photocharged electrodes and a resultant decrease in the photopotential that can be harnessed and used as a supplement to an applied electric potential. Our results demonstrate a novel strategy to tune plasmonic photopotentials and broaden the scope of plasmonic chemistry to include chemical transformations in nonaqueous solvents.

Potassium chloride (KCl, lot No. SLBP3785V), tetrabutylammonium hexafluorophosphate (TBAPF₆, 98%, lot No. BCCD7712), and acetonitrile (HPLC gradient grade, 99.9%) were purchased from Sigma-Aldrich. TBAPF₆ was used as received and stored in a desiccator when not in use. Sulfuric acid (H₂SO₄, ACS grade, 95%–98%, lot No. 2020081109) was procured from VWR Chemicals BDH. Silver nitrate (AgNO₃, ACS grade, lot No. 20D1056402) was purchased from VWR Life Science. Ar gas was purchased from Welders Supply Co. H₂O used in all experiments was produced using an ultrapure H₂O system with an integrated UV lamp (Sartorius Arium), yielding a resistivity of 18.2 MΩ cm. Au, Ag/AgCl, and Ag/Ag⁺ electrodes were purchased from CH Instruments, Inc. The Pt wire was purchased from Alfa Aesar. 0.3 µm (α-phase Type DX, lot No. 210727) and 0.05 µm (γ-phase Type DX, lot No. 210628) alumina powders were purchased from Electron Microscopy Sciences.

First, a Au disk electrode (D = 0.2 cm) was polished using an aqueous slurry of 0.3 µm alumina powder for 1 min followed by rinsing with H₂O. Next, the Au disk electrode was polished using an aqueous slurry of 0.05 µm alumina powder for 1 min. Polishing was performed in a figure-8 motion on a cloth polishing pad in the respective aqueous alumina slurries. The Au disk electrode was then sonicated in H₂O for 2 min and dried using a N₂ gas stream. To electrochemically roughen the polished Au electrode, the polished Au electrode, Ag/AgCl (3M KCl) reference electrode, and Pt mesh counter electrode were placed in a 0.1M KCl solution (aq.). A pulsed-potential waveform generated using a 660E CH Instruments potentiostat was applied to electrochemically oxidize and reduce Au, transforming the smooth surface into a nanoscale roughened surface. The potential was switched between −0.3 V (held for 3 s) and +1.2 V (held for 1.2 s) for 50 cycles. A representative current transient is shown in Fig. S1. The surface of an electrochemically roughened Au disk electrode was characterized by scanning electron microscopy (SEM, Thermo Fisher, Apreo 2). Representative images are shown in Fig. S2. SEM images were collected using an accelerating voltage of 2 kV, a spot size of 6, a beam current of 25 pA, and an Everhart–Thornley detector at various magnifications. The electrochemical roughening procedure produced Au structural features as small as ∼25–50 nm. Larger Au particles in the ∼150–500 nm range are also distributed across the electrode, often with rough surfaces containing sub-100 nm features. Au disk electrodes were re-polished and electrochemically roughened for each independent experimental trial.

Cyclic voltammetry was performed to determine the electrochemically active surface area (ECSA) of electrochemically roughened and smooth, polished Au electrodes. An electrochemically roughened or smooth Au electrode was immersed in 0.05M H₂SO₄ (aq.) along with the Ag/AgCl (3M KCl) reference electrode and Pt mesh counter electrode, and the potential was cyclically scanned between 0 and +1.5 V at 10 mV/s (Fig. S3). The ECSA was determined by dividing the charge corresponding to the Au oxide reduction peak (∼0.9 V) by the conversion factor of 450 µC/cm².⁸³ The ECSAs for three separately roughened Au electrodes and three smooth Au electrodes are summarized in Table S1. The ECSA of the electrochemically roughened Au electrodes is ∼3× larger than that of the smooth, polished Au electrodes. Current densities were determined by dividing measured currents by the average ECSA of these trials for the respective electrodes.

A single-compartment, optically transparent, three-electrode electrochemical cell was used to carry out all electrochemical measurements. Nominally anhydrous 0.1M TBAPF₆ in acetonitrile was used as the electrolyte. H₂O was sequentially added to the electrolyte to reach a final H₂O concentration of 0.5M, 1.0M, 2.0M, or 5.0M. The electrolyte solution was purged with Ar gas for 20 min to remove oxygen from the solvent prior to electrochemical measurements. An Ar gas blanket was maintained over the electrolytic solution during the electrochemical measurements to prevent oxygen from dissolving back into acetonitrile and to eliminate competition from the oxygen reduction reaction. Without an Ar-saturated electrolyte and Ar blanket, an aerated acetonitrile solution produces a large cathodic current beginning in the potential range of ∼−0.6 to −0.8 V vs Ag/Ag⁺, corresponding to the oxygen reduction reaction (Fig. S4). By contrast, when an Ar environment is maintained, the cathodic peak corresponding to oxygen reduction disappears, indicating little to no oxygen reduction (Fig. S4). An electrochemically roughened or smooth Au disk electrode was used as the working electrode along with the Pt wire counter electrode and nonaqueous Ag/Ag⁺ (10 mM AgNO₃ and 0.1M TBAPF₆ in acetonitrile) reference electrode for electrochemical measurements investigating plasmonically enhanced electrochemistry. A 660E CH Instruments potentiostat was used to carry out all electrochemical measurements. Linear sweep voltammetry (LSV) was performed by sweeping the potential at a scan rate of 5 mV/s from −0.2 to −2.4 V. Electrochemical impedance spectroscopy (EIS) was carried out at a dc potential of −2.3 V. An ac potential of 10 mV was superimposed on the dc potential, and impedance was measured by scanning the frequency from 215.8 MHz to 10 kHz. Chronoamperometry (CA) measurements were carried out by applying an electric potential of −2.4 V for 6 min.

The photographs of the experimental setup for plasmonically enhanced electrochemical HER are shown in Fig. S5. To excite the plasmon resonances of the electrochemically roughened Au electrode, circularly polarized 532 nm laser light (Oxxius LCX-532, 77 mW) or circularly polarized 642 nm laser light (Spectra-Physics Excelsior 640 CRDH, 77 mW) was focused onto the working electrode. The intensity of light at the electrode surface was 2.4 W/cm² for both excitation wavelengths. LSV and EIS measurements were performed first under dark conditions (no laser illumination) followed by under plasmon excitation conditions (laser illumination). For a single experimental trial, LSV and EIS were performed in dark and plasmon excitation conditions in an electrolytic solution with different H₂O concentrations. LSV and EIS measurements were obtained after sequential additions of H₂O to the electrolyte. During CA measurements, the electrode was intermittently irradiated by a 532 nm laser in ∼1 min time intervals by manual modulation of the laser shutter. A freshly polished and electrochemically roughened Au electrode was used for each independent experimental trial.

A thermocouple with a digital readout (Extech Instruments, model EA11A, K-type thermocouple) was used to measure the temperature near the electrochemically roughened Au electrode surface in dark and plasmon excitation conditions. A photograph of the setup used for temperature measurements is shown in Fig. S6. The probe of the thermocouple was placed in an electrolytic solution ∼0.05 mm from the electrode surface without blocking laser irradiation. Temperature measurements were made in dark (T₁) and plasmon excitation conditions (T₂) in electrolytic solutions with different H₂O concentrations (Table S2). The electrode temperature was allowed to stabilize for 5 min before recording it in each condition. Temperature values reported in Table S2 are an average of two independent trials with the standard deviation in parentheses.

A photograph of the experimental setup for temperature-dependent electrochemical HER measurements is shown in Fig. S7. The temperature of the bulk electrolytic solution was controlled by immersing the electrochemical cell in a circulating H₂O bath. The temperature of the electrolytic solution was monitored with a thermometer submerged in the electrolytic solution. LSV and EIS measurements were carried out in dark conditions at temperatures that correspond to the temperatures measured near the electrode surface in dark and plasmon excitation conditions in electrolytic solutions with different H₂O concentrations.

The CH Instruments software was used to construct a Randles equivalent circuit (Fig. S8) to fit raw EIS data (Figs. S9–S14). From the fits, solution resistances (R_s) and charge transfer resistances (R_ct) were extracted (Tables S3 and S4). All LSV data were corrected for iR drop using the average solution resistance from three independent trials (Fig. S15). LSVs corrected for iR drop were analyzed by tabulating the potential (E) required to produce a defined current density (15% of the maximum cathodic current under dark conditions in a given concentration of H₂O in the electrolytic solution, j_15%,dark) and the current density (j) at −2.3 V. ΔE was determined by subtracting the potential at j_15%,dark in dark conditions from the potential at j_15%,dark in plasmon excitation conditions (i.e., E_light − E_dark). ΔE from the temperature control LSV data was determined by subtracting the potential at j_15%,dark in dark, ambient (T₁) conditions from the potential at j_15%,dark in dark, higher temperature (T₂) conditions (i.e., E_T2 − E_T1). All measurements were performed in triplicates, and data are reported as the average over three independent trials with error bars that represent the standard deviation of the three independent trials.

Acetonitrile was chosen as a nonaqueous solvent due to its large electrochemical potential window and miscibility with our model redox probe H₂O. Changing the concentration of H₂O in acetonitrile not only changes the chemical potential of the redox reaction but also tunes the effective polarity of the solvent. We first characterized the reduction of H₂O to H₂ in acetonitrile at an electrochemically roughened Au electrode in dark conditions using LSV (Fig. S16A). At applied potentials more negative than ∼−2.0 V vs Ag/Ag⁺, an increase in cathodic current density was observed, indicating the electrochemical HER. The HER was measured with H₂O concentrations of 0.5, 1, 2, and 5M. As the concentration of H₂O in acetonitrile increases, the onset potential for H₂ evolution shifts to more positive potentials and the current density increases at high overpotentials, leading to the formation of H₂ bubbles at the electrode surface. Next, we measured the HER in acetonitrile at an electrochemically roughened Au electrode irradiated with visible light using LSV (Fig. S16B). The same trend was observed with electrodes irradiated with light and electrodes in dark conditions: the onset potential for H₂ evolution shifted to lower energies, and larger current densities were measured as the concentration of H₂O in acetonitrile increased.

Figures 1(a) and 1(b) (also see Figs. S17–S19) show the HER in acetonitrile with 0.5 and 5M H₂O occurring at electrochemically roughened Au electrodes in dark conditions and irradiated with visible light as measured by LSV. At a concentration of 0.5M H₂O, the electrochemically roughened Au electrode irradiated by light required ∼175 mV less applied electric potential to drive the HER compared to the electrochemically roughened Au electrode in dark conditions. By contrast, at a concentration of 5M H₂O, the Au electrode irradiated by light reduced the applied electric potential required to drive the reaction by ∼50 mV. The latter potential shift is similar to what has been previously reported in aqueous electrochemical reactions enhanced by plasmon excitation.^{69,70,84–86} To the best of our knowledge, this is the first report of an electrochemical reaction in a nonaqueous solvent enhanced by visible light excitation of a plasmonic electrode. Strikingly, an order of magnitude reduction of the concentration of H₂O in acetonitrile results in a 3.5× increase in the potential difference at j_15%,dark in visible light excitation conditions compared to dark conditions (Fig. 1 and Table S5). Plots of ln (j) vs the applied electric potential in Figs. 1(c) and 1(d) show linear trends in the potential region where the HER begins. At a given concentration of H₂O, these slopes are statistically the same under dark and visible light excitation conditions, which indicates that light irradiation does not appreciably affect the reaction rate and, thus, the reaction mechanism.

Figure 2(a) quantifies the difference in reaction overpotential between electrochemically roughened Au electrodes in dark and 532 nm light excitation conditions as a function of the H₂O concentration. The largest shift in overpotential is observed at the lowest H₂O concentration, and the overpotential rapidly falls off as the concentration of H₂O increases. Figure 2(b) shows that the absolute value of the current density increase under light excitation is larger with larger concentrations of H₂O as expected due to a higher analyte concentration. However, the enhancement in current density [Fig. 2(c)] is the largest with the lowest H₂O concentration and drops off as the H₂O concentration increases, following the trend observed for the change in overpotential [Fig. 2(a)].

The enhancement in current density by visible light excitation is reversible at all H₂O concentrations as shown by chronoamperometry measurements at an applied potential of −2.4 V under chopped illumination (Fig. 3 and Fig. S20). The current density rapidly increases under laser irradiation (on) and rapidly decreases when the laser is shuttered (off). Following the initial rapid increase in current density under laser irradiation, the current density continues to gradually increase. This is likely due to the electrode releasing heat into the electrolyte, which increases mass transport. A steady-state current density is not reached in our experiment because the temperature of the system was not allowed sufficient time to equilibrate. After illumination, the current density in the dark decreases in two steps. The first step is large and rapid due to the absence of light excitation. The second step is gradual and is likely due to heat dissipation, which lowers the electrode and bulk electrolyte temperatures. A decrease in temperature reduces the diffusion coefficient of H₂O in acetonitrile and slows mass transport.

Hot electrons and holes are generated in Au nanomaterials via interband and intraband electronic transitions, which can be driven by nonradiative plasmon decay or direct excitation of Au by light. Hot charge carriers can subsequently be extracted to drive redox reactions or further decay, resulting in the transfer of heat to the surrounding medium; each pathway can affect electrochemical reactions. To determine whether light-excitation-induced changes to the HER observed in the current study are a result of nonradiative plasmon decay or direct electronic excitations, the HER in acetonitrile was measured by LSV using smooth, polished Au electrodes (electrodes with limited nanoscale features to support localized plasmon resonances) with 0.5M or 5M H₂O added to the solvent. 22 and 28 mV positive shifts in the HER overpotential and the corresponding increase in current density were observed when a smooth Au electrode was irradiated with 532 nm laser light compared to dark conditions with 0.5M H₂O and 5M H₂O in the electrolyte, respectively (Figs. S21–S23 and Table S6). Because the excitation light energy of 2.33 eV is at the threshold of interband transitions in Au,^87–89 the shift in the HER overpotential using the smooth electrodes may be attributed to the direct excitation of interband transitions, leading to the generation of hot electrons and holes. A fraction of the hot holes is quenched by the externally applied electric potential, leading to an accumulation of hot electrons on the smooth Au electrode, increasing the Fermi level of the electrode and reducing the overpotential for the HER.⁶⁹ The ∼25 mV shift in overpotential for the HER on smooth Au electrodes is smaller than the up to 175 mV shift in overpotential measured on electrochemically roughened Au electrodes. It should be noted that the smooth Au electrodes are mechanically polished with 0.3 and 0.05 µm alumina particles and therefore are not atomically smooth. Micro- and nanoscale scratches and defects are present, which may impart some plasmonic character. Thus, we conclude that the direct excitation of Au interband transitions by light affects the HER, but interband transitions via plasmon decay must be considered to account for the total measured shift in overpotential and enhancement in current density.

To further substantiate our assertion that interband transitions from direct excitation and plasmon decay enhance the electrochemical HER, we measured the HER using electrochemically roughened (Figs. S24–S26 and Table S7) and smooth Au electrodes (Figs. S27–S29, Table S8) irradiated by 642 nm laser light. Given the broad size distribution of the nanoscale features of the electrochemically roughened Au electrodes (Fig. S2), both 532 and 642 nm light will excite plasmon resonances of the electrode. However, 642 nm (∼1.9 eV) light is below the threshold for interband transitions in Au and can therefore only excite intraband transitions. Figure 4 compares the change in overpotential and enhancement in current density when an electrochemically roughened Au electrode is irradiated with 532 or 642 nm light compared to dark conditions. Under 642 nm light irradiation, the shift in the overpotential and enhancement in current density is lower than that observed under 532 nm light irradiation at all H₂O concentrations. As the H₂O concentration increases, the change in overpotential and enhanced current density monotonically decreases for both 532 and 642 nm light excitation conditions. On a smooth Au electrode, 642 nm laser irradiation produces a negligible shift in the overpotential and a negligible enhancement in current density for the HER with 0.5M or 5M H₂O added to the electrolytic solution (Figs. S27–S29 and Table S8). The data acquired from 642 nm light irradiation support that nonradiative plasmon decay contributes to the buildup of a photopotential on the electrochemically roughened Au electrodes. The larger overpotential shifts observed with 532 nm light irradiation compared to 642 nm light excitation are due to interband transitions, either excited directly or indirectly through plasmon decay. Interband transitions have a longer lifetime than intraband transitions,^90,91 enabling more efficient hole extraction by the external electric potential and a larger buildup of a photopotential on Au.

The excitation and subsequent decay of plasmon resonances lead to the formation of hot charge carriers and a release of energy in the form of heat to the surrounding environment.⁹² Both hot charge carriers^8,69,70 and heat^34,37,38 from plasmon excitation have been observed to drive chemical reactions. To determine if hot electrons and/or heat are responsible for the changes in the HER measured by LSV under plasmon excitation conditions, we performed control experiments in dark conditions at different bulk electrolyte temperatures. First, we determined the temperature near the surface of the electrode under plasmon excitation and dark conditions for different concentrations of H₂O in acetonitrile (Table S2). At all H₂O concentrations, the temperature rise near the electrode (∼0.05 mm between the thermocouple probe and electrode surface) was ∼2 °C under plasmon excitation conditions. Next, we measured the HER in acetonitrile using electrochemically roughened Au electrodes in dark conditions while maintaining the bulk electrolyte temperature at the temperatures that correspond to dark and plasmon excitation conditions [Figs. 5(a) and 4(b) and Figs. S30–S32]. As expected, a 2 °C change in the bulk electrolyte temperature showed no significant changes in the reaction rate [Figs. 5(c) and 5(d)], overpotential [Fig. 6(a)], absolute current density [Fig. 6(b)], or enhanced current density [Fig. 6(c) and Table S9] for the HER. A limitation of these control experiments is that the thermocouple probe measures bulk temperature and does not necessarily measure the actual surface temperature of the electrode.

There are several reports in the literature that show a local temperature rise of tens of degrees when plasmonic particles are irradiated with light intensities on the order of kW/cm² to MW/cm² depending on the plasmonic material, energy of light, morphology of the particle, and surrounding medium.^93–95 However, the light intensity of 2.4 W/cm² used in our experiments is orders of magnitude lower than those typically reported, to raise the temperature tens of degrees. Thus, our measurement of a 2 °C change in the bulk electrolyte temperature is likely close to the temperature change in the Au electrode surface. To provide further evidence that photothermal heating does not appreciably contribute to the shifts in overpotential or enhancement in current density using plasmonically excited, electrochemically roughened Au electrodes, we measured the HER in acetonitrile/H₂O mixtures with the bulk electrolyte maintained at 20.0 and 35.0 °C (Figs. S33–S35 and Table S10). At the elevated temperature of 35.0 °C, overpotential shifts of 10 and 20 mV were observed along with enhancements in current densities of 1.262 and 1.156 with 0.5 and 5M H₂O added to acetonitrile, respectively. Even in this extreme case, photothermal heating alone cannot account for the overpotential and current density changes observed from plasmonically excited Au electrodes. Moreover, it must be noted that the changes in the HER under plasmon excitation conditions change with the concentration of H₂O in acetonitrile. The intensity of the excitation light is fixed in our experiments, and the change in temperature near the electrode is nearly constant across all concentrations of H₂O. Together, these observations confirm that photothermal heating has a negligible role in the enhancement of the HER in our system and that the overpotential shifts and current density enhancements originate from hot charge carriers at steady-state conditions.

The positive shift in the overpotential of an electrochemical reduction reaction under plasmonically excited electrodes observed in our experiments is consistent with previous observations in aqueous media.^{69,70,84–86} An applied electric potential sets the Fermi level of the working electrode. Following the plasmon excitation of our nanoscale roughened Au electrode, hot electrons and holes are generated by electronic transitions in the metal. The applied electric potential quenches a fraction of the hot holes, leading to an accumulation of hot electrons on the metal. This establishes a photopotential that supplements an applied electric potential.^8,31,70 When the metal surface is photocharged in this manner, the Fermi level of the electrode increases, which reduces the external electrical energy needed to overcome the activation barrier (η) for the HER. The difference in activation energy between our electrodes in dark (η_dark) and plasmon excitation conditions (η_hv) is given by ΔE, which corresponds to the photopotential developed by light irradiation. When the applied electric potential is higher than the onset potential (i.e., activation energy) for the HER, ΔE increases the total applied potential, resulting in an enhancement of the current density. The magnitude of ΔE determines the enhancement in current density. As shown in Fig. 7, the measured ΔE values at different H₂O concentrations in our experiments were found to be linearly correlated with the natural logarithm of the current density enhancement, consistent with a Butler–Volmer relationship.

A primary finding of this study is that the photopotential on our plasmonically excited Au electrodes in acetonitrile is greater than the photopotentials reported in aqueous electrochemical systems and the photopotential depends on the H₂O concentration in acetonitrile. The steady-state photopotential depends on the properties of the plasmonic material (e.g., electronic structure, absorption cross section, size and morphology) and the photogeneration rate of hot electron–hole pairs (i.e., intensity of excitation light).^23,31,51 In our experiments, these are fixed quantities. Thus, the maximum photopotential of our plasmonically excited electrodes is constant across all our experimental conditions. At a fixed ratio of H₂O to acetonitrile, the reorganization energy is expected to be similar between light and plasmon excitation conditions because the refractive index of the medium is conserved and only a small temperature change was measured. This led us to hypothesize that the change in overpotential and current density measured between plasmon excitation and dark conditions at different concentrations of H₂O was due a change in the fraction of photopotential that can be harvested to assist in the electrochemical HER.

To explain why the measured photopotential is dependent on the H₂O concentration, we propose a mechanism that takes into consideration the interaction of solvent molecules (both acetonitrile and H₂O) with the photocharged plasmonic electrode. Under the conditions of a cathodically polarized and photocharged electrode, solvent molecules orient themselves around the charged electrode to stabilize it. Because H₂O has a stronger dipole moment than acetonitrile, when added to acetonitrile, the polar H₂O molecules will preferentially orient themselves on the electrode surface due to the presence of a large electric field at the surface (Fig. 8). The presence of H₂O stabilizes the photocharged electrode more than neat acetonitrile, which lowers the Fermi level of the plasmonically excited electrode, effectively increasing the external electric potential required to overcome the activation barrier and reducing ΔE [Fig. 8(a)]. As the concentration of H₂O in acetonitrile increases, the photocharged electrode is further stabilized because more H₂O molecules are present in the inner Helmholtz plane of the electrode surface [Fig. 8(b)]. An increased stabilization of the photocharged electrode results in a further decrease in the Fermi level under light excitation, an increase in the external electrical energy required to activate H₂O in the HER, and a decrease in the photopotential (ΔE). Therefore, decreasing the solvent polarity (e.g., increasing the ratio of acetonitrile to H₂O) allows a larger steady-state photopotential to develop on the plasmonically excited electrode, providing a greater enhancement to the reaction.

In summary, we demonstrated that in the nonaqueous solvent acetonitrile, the energy required from an external electric bias for electrochemical HER can be lowered by ∼175 mV under plasmon excitation conditions. As the amount of H₂O in acetonitrile increases, the photocharged electrode is increasingly stabilized, which reduces the usable photopotential. Our experiments demonstrate that mixtures of acetonitrile and H₂O can be used as a handle to tune the photopotentials developed on plasmonically excited materials. To generalize these results, further investigations are required across a polarity range of nonaqueous solvents and solvent mixtures. Moreover, complementary spectroscopic studies are vital to further our mechanistic understanding of how microenvironments can influence chemical reactivity in plasmon-driven and plasmon-assisted chemistry. These efforts are currently ongoing in our laboratory. The results of the current study in the acetonitrile–H₂O system highlight the opportunity to enhance additional reactions important in energy conversion and storage, electrosynthesis, and organic transformations in nonaqueous solvents, in particular due to the increase in the solubility of nonpolar gases and molecular reagents.

See the supplementary material for experimental design details, voltammetry and electrochemical impedance data acquired for individual trials, and a summary of all analyses.

This work was supported by start-up funds from the University of Louisville. The authors thank Dr. Lee Thompson for helpful discussions during the preparation of the manuscript.

The authors have no conflicts to disclose.

All authors read and approved the final manuscript.

The data that support the findings of this study are available from the corresponding author upon reasonable request.