Spontaneous Electric Fields Play a Key Role in Thermochemical Catalysis at Metal−Liquid Interfaces

Abstract

Large oriented electric fields spontaneously arise at all solid–liquid interfaces via the exchange of ions and/or electrons with the solution. Although intrinsic electric fields are known to play an important role in molecular and biological catalysis, the role of spontaneous polarization in heterogeneous thermocatalysis remains unclear because the catalysts employed are typically disconnected from an external circuit, which makes it difficult to monitor or control the degree of electrical polarization of the surface. Here, we address this knowledge gap by developing general methods for wirelessly monitoring and controlling spontaneous electrical polarization at conductive catalysts dispersed in liquid media. By combining electrochemical and spectroscopic measurements, we demonstrate that proton and electron transfer from solution controllably, spontaneously, and wirelessly polarize Pt surfaces during thermochemical catalysis. We employ liquid-phase ethylene hydrogenation on a Pt/C catalyst as a thermochemical probe reaction and observe that the rate of this nonpolar hydrogenation reaction is significantly influenced by spontaneous electric fields generated by both interfacial proton transfer in water and interfacial electron transfer from organometallic redox buffers in a polar aprotic ortho-difluorobenzene solvent. Across these vastly disparate reaction media, we observe quantitatively similar scaling of ethylene hydrogenation rates with the Pt open-circuit electrochemical potential (E_OCP). These results isolate the role of interfacial electrostatic effects from medium-specific chemical interactions and establish that spontaneous interfacial electric fields play a critical role in liquid-phase heterogeneous catalysis. Consequently, E_OCP—a generally overlooked parameter in heterogeneous catalysis—warrants consideration in mechanistic studies of thermochemical reactions at solid–liquid interfaces, alongside chemical factors such as temperature, reactant activities, and catalyst structure. Indeed, this work establishes the experimental and conceptual foundation for harnessing electric fields to both elucidate surface chemistry and manipulate preparative thermochemical catalysis.

Short abstract

Spontaneous charge transfer reactions enable wireless control of electric fields at catalytic interfaces, revealing that polarization impacts thermochemical catalysis—even for nonpolar reactions.

Introduction

Electric fields are known to play an important role in many catalytic transformations throughout chemistry¹⁻²⁰ and biology.^21,22 For example, in molecular catalysis, charged moieties in the secondary sphere of a coordination compound can impart oriented internal electric fields,^11,13−15 which have been used to accelerate CO₂ activation (Figure 1, left).¹¹ Similarly, in enzymes, amino acid residues sustain enormous oriented electric fields within the active site that can drive biological catalysis (Figure 1, center).^21,22 In principle, oriented electric fields could also be used to control thermochemical catalysis at solid–liquid interfaces (Figure 1, right), which is playing an increasingly important role in facilitating the transition to a low-carbon energy and chemical economy.²³⁻³⁰ Indeed, electrical charging or polarization arises spontaneously (i.e., without any external driving force) at all solid–liquid interfaces, creating large, intrinsically oriented electric fields at the interface.³¹ Yet, this spontaneous electrical polarization is rarely considered or invoked in thermochemical catalysis, primarily for two reasons. First, it is difficult to monitor electrical polarization at interfaces that are not connected to macroscopic wires. Such “distributed” interfaces, which include solid catalysts (e.g., supported metal nanoparticles) distributed throughout packed-bed or stirred-tank reactors, underpin virtually all of heterogeneous catalysis. Second, the heterogeneous catalysis community has historically focused on reactions at solid–gas interfaces, where electrical polarization is usually presumed to be negligible (though it is not necessarily so^4,32,33). Yet, a general understanding of how spontaneous electrical polarization influences thermochemical reactivity could provide a powerful new handle—complementary to existing chemical principles—for understanding and controlling heterogeneous catalysis.

Figure 1.

Spontaneous electric fields (E) in catalysis.

Electrical polarization of solid–liquid interfaces is most commonly studied within the context of electrochemistry. In this context, electrical polarization is controlled by passing charge through an external circuit driven by a power source (e.g., a potentiostat or battery). This driven polarization forms the basis for interconverting electrical and chemical energy in a wide array of electrochemical devices and is known at a molecular level to induce ion migration,³¹ order solvents,^31,34−38 polarize adsorbates,³⁹⁻⁴² and alter the free energy and kinetics of adsorption.^31,43

However, electrical polarization also arises spontaneously at all solid–liquid interfaces, even those entirely disconnected from any wires or electrical circuits. This phenomenon exists because differences in the chemical composition of the two phases can drive the spontaneous transfer of charged species (i.e., ions and/or electrons) across the interface. Specifically, the driving force for an interfacial charge transfer process is given by the difference in electrochemical potentials of the species involved.^31,44 By definition, the electrochemical potential, μ̅_j^α, of species j in phase α, measures the total work required to differentially add 1 mole of species j to phase α. Therefore, μ̅_j depends on both chemical and electrostatic contributions and can be parsed as follows: μ̅_j^α = μ_j + zFϕ^α, where μ_j^α is the chemical potential of j in an uncharged reference phase, z is the charge number of species j, F is Faraday’s constant, and ϕ^α is the electrostatic potential (specifically, the Galvani potential) of phase α.³¹ Upon contacting the catalyst surface with solution, any difference between the chemical potentials of the charged species at the surface and in solution will drive spontaneous, incremental transfer of that charged species until an electrostatic potential difference, Δϕ^{surface–solution} = ϕ^surface – ϕ^solution, is generated that counteracts the chemical potential difference;³¹ in this manner, electrochemical equilibrium is established. Such equilibration processes result in spontaneous electrical polarization of the solid–liquid interface, with attendant interfacial electric fields spanning molecular length-scales. Despite the long-standing appreciation that solid–liquid interfaces undergo spontaneous polarization,³¹ this fundamental phenomenon has been largely ignored in understanding heterogeneous catalysis at solid–liquid interfaces.

Spontaneous polarization effects are expected to be especially pronounced for heterogeneous catalytic reactions occurring in aqueous media, because water is particularly effective at solvating ions and protons. For example, during Pt-catalyzed aqueous hydrogenation reactions, the adsorbed H* intermediates involved in product formation are also in equilibrium with protons in solution (Figure 2a). Indeed, under such conditions, H* undergoes reversible deprotonation to produce a solution-phase proton, leaving behind a residual electron equivalent in the Pt (and vice versa; Figure 2b). Accordingly, the electrochemical potential of H* on the surface must equal the combined electrochemical potentials of H⁺ in bulk solution and electrons in the Pt:

where μ̅H+solution = μH+ + Fϕsolution and μ̅e–Pt = μe– – FϕPt;31 the electron chemical potential, μe–Pt, depends on the chemical composition of the Pt, and thus relates to the Pt work function. We note that the foregoing equation is simply the Nernst condition of equilibrium and reflects both chemical and electrostatic factors.31 In order to parse the two contributions, we may substitute in the expressions for μ̅H+ and μ̅e–Pt in order to arrive at the following expression for the equilibrium condition:

Consequently, in the limit that μ̅_H^surface is constant—for example, when H* is quasi-equilibrated with H₂ gas held at constant pressure—and the chemical potential of electrons is invariant, increasing the solution pH (i.e., lowering μ_H⁺) is expected to make Δϕ^{Pt–solution} more negative in order to maintain equilibrium. In other words, a more alkaline medium will drive deprotonation of H* species (which are replenished and held at constant activity via H₂ dissociative chemisorption), leaving behind electrons in the metal that negatively charge the surface and sustain a pH-dependent interfacial electric field (Figure 2c). At H₂-saturated Pt–water interfaces, pH-dependent electrostatic potential drops have been quantified using non-faradaic interfacial probe reactions.⁴⁵ These electric fields have also been invoked to explain pH-dependent rates in electrocatalysis.³⁶

Figure 2.

(a) Spontaneous interfacial reactions inherent to H₂-saturated Pt–water interfaces. (b) Electrochemical equilibrium for spontaneous interfacial proton transfer. μ̅_H is the electrochemical potential of surface H*, and μ_e− and μ_H+ are, respectively, the chemical potential of electrons in Pt and protons in solution. ϕ is the electrostatic potential (specifically, the Galvani potential). (c) Spontaneous proton transfer at Pt–water interfaces is expected to give rise to pH-dependent interfacial electric fields. We employ ethylene hydrogenation as a test reaction to probe the impact of spontaneous polarization on catalysis.

Despite the foregoing understanding, the role of polarization in thermochemical heterogeneous catalysis at metal–water interfaces remains ambiguous. While studies have invoked that spontaneous electric fields may influence catalysis,⁴⁶⁻⁴⁹ chemical effects convolute putative electrostatic contributions in these systems. For example, during the hydrogenation of maleic acid at a wired Pt–water interface, changes in rate with H₂ pressure were attributed to spontaneous electrical polarization, without accounting for changes in chemical driving force (i.e., the H₂ reaction order).⁴⁶ Likewise, pH-dependent CO oxidation^47,48 and phenol hydrogenation⁴⁹ rates were indirectly attributed to spontaneous polarization on the basis of shifts in the CO infrared spectrum and molecular dynamics simulations, respectively, but could not rule out contributions from chemical pH effects. Thus, it remains unclear whether spontaneous electrical polarization indeed plays a role in liquid-phase thermochemical heterogeneous catalysis.

To address this knowledge gap, we herein develop general methods for monitoring and controlling spontaneous electrical polarization during catalysis at distributed metal–liquid interfaces. By combining electrochemical and spectroscopic measurements, we demonstrate that different classes of reversible redox partners deployed in solution controllably, spontaneously, and wirelessly polarize distributed Pt surfaces during thermochemical catalysis. We employ ethylene hydrogenation as a probe reaction (Figure 2c) in order to minimize the influence of acid–base reactions and find that the rate of this nonpolar hydrogenation reaction is significantly influenced by spontaneous electric fields generated by both interfacial proton transfer in water and interfacial electron transfer in a polar aprotic ortho-difluorobenzene solvent. These results isolate the role of interfacial electrostatic effects from medium-specific chemical interactions and establish that spontaneous interfacial electric fields play a critical role in liquid-phase heterogeneous catalysis.

Results and Discussion

Infrared Stark Spectroscopy to Probe Spontaneous Polarization at Distributed Pt–Water Interfaces

Investigating the influence of spontaneous electrical polarization on thermochemical catalysis requires, first, a method for tracking the magnitude of interfacial polarization at distributed metal–liquid interfaces across reaction conditions. Unfortunately, it is not possible to directly quantitate electric fields or Δϕ at metal–liquid interfaces by electrochemical methods. This is because electrochemical methods measure electrochemical potentials (i.e., μ̅), which inherently combine chemical and electrostatic factors.^31,50 Therefore, although macroscopic electrodes allow for convenient electrochemical sensing, it cannot be assumed a priori that changes in the electrochemical potential, E, relative to a constant reference couple, translate into changes in Δϕ across a metal–liquid interface. Furthermore, the inability to make an electrical contact to a distributed interface necessitates a nonelectrochemical probe of the interfacial electric field.

In order to track changes in the degree of spontaneous polarization at a distributed Pt–liquid interface (i.e., Pt/C immersed in buffered aqueous electrolytes), we employ adsorbed CO as an interface-specific vibrational reporter in attenuated total reflectance (ATR)-infrared (IR) spectroscopy. The CO stretching frequency, ν(CO), is sensitive to interfacial polarization;^40,51−54 this phenomenon, termed a Stark shift, has been ascribed to surface charge affecting back-bonding into CO π* orbitals.^40,55 Since CO adsorbs strongly to Pt surfaces, these Stark shifts sample changes in the degree of polarization at a distributed Pt–liquid interface as a function of changes in the electrolyte composition. In order to correlate these Stark shifts to the electrochemical potential of the medium, we separately monitored the open-circuit potentials (E_OCP) of a macroscopic Pt sensing electrode under identical conditions. To ensure stable polarization via H₂/H⁺ equilibration, the system was simultaneously saturated with H₂ (in addition to CO) across all conditions.

At pH 5, we observe an IR absorption peak at 2046 cm^–1, which corresponds to atop-bound CO, ν(CO_A) (Figure 3a, red).^{40,47,51,52,54,56−58} The pronounced bipolar shape of the atop CO feature has been commonly observed for infrared reflectance measurements of CO adsorbed on carbon-supported metal nanoparticles and has been ascribed to optical effects.^57,59−62 As the pH is increased to 7 (Figure 3a, black) and 9 (Figure 3a, blue), ν(CO_A) redshifts to 2041 and 2035 cm^–1, respectively. Importantly, since the bipolar peak shape is preserved across the pH range investigated, this optical artifact would minimally convolute the extracted trends in peak position. We note that, depending on pH, we also observe a broad bridge-bound CO feature (Figure S2), consistent with prior studies;^{40,47,54,56,57,62} however, due to its breadth and convolution from water vapor, we refrain from analyzing this feature. However, the data clearly indicate a systematic redshift in ν(CO_A) as the solution pH is increased.

Figure 3.

(a) pH-dependence of ATR-IR spectra, taken at saturation, of atop-bound CO at a distributed Pt–water interface (i.e., Pt/C immersed in buffered aqueous electrolytes). (b) Peak maximum for atop-bound CO vs open-circuit potentials (E_OCP) of a Pt/C electrode. Conditions: 7 kPa CO, 94 kPa H₂, 60 wt % Pt/C, electrolyte = 0.5 M NaClO₄ in Britton–Robinson universal buffer (0.04 M acetic acid, 0.04 M phosphoric acid, 0.04 M boric acid in water); pH adjusted with NaOH. Performed at room temperature (294 ± 1 K).

In order to determine whether E_OCP of a macroscopic sensing electrode provides a faithful proxy for the degree of polarization at the distributed Pt/C particles, we dropcast the same Pt/C catalyst onto carbon paper (referred to hereafter as Pt/C electrodes). Under conditions identical to the ATR measurements, we observe a linear pH-dependence of E_OCP of the Pt/C electrodes in the presence of CO and H₂ (Figure S3, black squares); however, this scaling deviates from what is expected from the Nernst equation (Figure S3, gray circles). Because E_OCP is the potential that exactly balances anodic and cathodic processes, the non-Nernstian behavior observed in the presence of CO may be readily explained by the differential impact of site-blocking CO on the kinetics of H₂ oxidation and H⁺ reduction half-reactions. Nonetheless, a plot of ν(CO_A) vs E_OCP is linear (Figure 3b) and provides an inferred electrochemical Stark tuning slope of approximately ∼60 cm^–1 V^–1 for the adsorbed CO species across the distributed Pt–water interface. Although we refrain from quantitative interpretation of this slope because of the inherent complexities of the system (see supplementary information, SI, section 2.1 for a detailed discussion), this value is in line with the relatively wide (i.e., 25–62 cm^–1 V^–1) range of Stark tuning values reported for CO on Pt under driven (i.e., wired) polarization.^{40,51−54,62}

We note that other possible chemical contributions to changes in ν(CO_A) are inconsistent with the observed trend. In particular, the decreasing proton activity with increasing pH would be expected to attenuate H-bonding interactions (e.g., a CO–H⁺ adduct), but as previously discussed this effect would be expected to blueshift the CO stretching frequency.⁴⁷ This is at odds with the observed red shift. Additionally, although OH^– coadsorption might influence the CO band, the observed changes in E_OCP nearly offset changes in OH^– activity and OH^– adsorption is disfavored at the reducing potentials recorded here in the presence of CO and H₂.⁶³ Finally, although both ν(CO_A) and Stark tuning slopes are known to depend on CO coverage,⁵² electrochemical CO stripping experiments performed on the Pt/C catalyst under conditions mimicking the ATR measurements (see SI section 1.2.3 for full details) demonstrate that the CO coverage is roughly constant at a value of ∼1 across the pH range spanned by our IR experiments (Figures S4–S5 and Table S2; see SI section 2.2 for a detailed discussion). This suggests that the observed shift in ν(CO_A) with pH likely does not arise from pH- or buffer speciation-induced changes to the equilibrium CO coverage.

Based on the foregoing analysis, we ascribe the pH-dependence of ν(CO_A) to a progressive increase in the magnitude of spontaneous electrical polarization of the distributed Pt–water interface with increasing solution alkalinity. Additionally, these studies imply that, for this system, E_OCP as measured by a macroscopic sensing electrode provides a good estimate of changes in the magnitude of spontaneous polarization (i.e., changes to Δϕ) across a distributed Pt–water interface, consistent with prior work.⁴⁵

Tracking Spontaneous Polarization during Ethylene Hydrogenation at Distributed Pt–Water Interfaces

The above observations indicate that the degree of spontaneous polarization of a Pt–water interface can be tuned simply by varying the pH. Thus, we envisioned that the pH-dependence of a model heterogeneous thermochemical reaction could provide insight into the role of interfacial electric fields in catalysis. For these studies, we elected to examine the pH/potential-dependence of Pt/C catalyzed ethylene hydrogenation in aqueous media. This explicitly nonpolar transformation is expected to provide a baseline measure of electrostatic contributions to catalysis that would only be amplified for reactions of more polar substrates.

Experiments were carried out using a well-mixed slurry reactor equipped with a Pt sensing wire and reference electrode, which provided an operando measurement of Pt E_OCP during catalysis (Figure 4); we note that similar approaches have been used before.^100,101 A Pt wire was chosen for this purpose in order to minimize background catalytic activity. Moreover, because ethylene hydrogenation is structure-insensitive across the entire range spanning fully dispersed Pt/SiO₂ (i.e., all Pt atoms accessible to titration by H₂) to a Pt wire,⁶⁴ any impact of olefin surface chemistry on electrochemical pinning reactions would be expected to be similar on both Pt/C and the Pt sensing wire. During the reaction, ethylene (5 kPa) and hydrogen (96 kPa) gases were continuously sparged through the aqueous electrolyte containing suspended Pt/C catalyst particles (Figure 4). The pH was adjusted with HClO₄, NaOH, or a dilute borate buffer, and NaClO₄ was added as a supporting electrolyte to maintain 0.1 M total ionic strength. Importantly, under these conditions, the Pt sensing wire contributed negligibly to measured hydrogenation rates (Figure S6). Thus, measured rates reflect the catalytic activity of the dispersed Pt/C particles.

Figure 4.

Diagram of the well-mixed slurry reactor used to monitor spontaneous electrical polarization during liquid-phase ethylene hydrogenation catalysis. E_OCP refers to the open-circuit potential of the Pt wire vs the Ag/AgCl reference electrode.

Care was taken to ensure that measured rates reflect intrinsic reaction kinetics. Specifically, ethylene hydrogenation rates were measured at differential ethylene conversion and corrected for first-order deactivation (Figure S7). We report turnover rates, defined herein as the rate of ethane production per surface Pt atom as estimated from Pt loading and particle size distributions (see SI section 1.3 and Figure S8). Importantly, absolute hydrogenation rates scaled proportionally to catalyst loading (Figure S9), indicating negligible gas-to-liquid interphase mass transfer limitations.⁶⁵ Moreover, turnover rates were unaffected by Pt loading on the carbon support (Table S3), demonstrating that internal diffusion limitations are negligible within the catalyst pores.^66,67 Taken together, the latter two observations establish that our rate measurements are free from mass transfer artifacts and thus reflect intrinsic liquid-phase ethylene hydrogenation kinetics.

Aqueous ethylene hydrogenation rates decrease systematically with increasing pH. At pH = 2.0, we observe a hydrogenation rate of 3.0 ± 0.2 s^–1 (Figure 5a), which is similar to the gas-phase rate (i.e., 3.0 s^–1) at 273 K and similar C₂H₄ and H₂ partial pressures, as inferred from previously measured rates and reaction orders.⁶⁴ However, upon increasing the pH to 5.1, 9.0, and 12.9, the rate systematically decreases to 2.4 ± 0.1, 1.6 ± 0.3, and 1.0 ± 0.01 s^–1 respectively (Figure 5a). These changes constitute an exponential decrease in rate with increasing pH (Figure 5a). This trend of decreasing hydrogenation rates with increasing pH has also been observed for phenol hydrogenation.⁴⁹ In contrast to the hydrogenation of polar, protic phenol, which could be subject to Brønsted acid–base chemistry or promotion, the explicitly nonpolar nature of ethylene hydrogenation argues against a simple acid–base rationalization for the pH-dependence observed here. Moreover, the log–linearity of rate vs pH across 11 orders of magnitude in proton activity argues against a simple site-blocking model involving hydroxide ions. Finally, we note that the consistency in trend between unbuffered (pH 2.0, 5.1, 12.9) and borate-buffered (pH 9.0) electrolytes suggests that the borate buffer species do not significantly influence catalysis; this is consistent with the innocence of borate buffers in proton-coupled electron transfer at a Au–water interface.⁶⁸ Together, the data establish that aqueous, Pt/C-catalyzed ethylene hydrogenation rates exponentially decrease with increasing pH, in a manner inconsistent with simple acid–base or hydroxide site-blocking explanations.

Figure 5.

Ethylene hydrogenation in aqueous electrolytes on a Pt/C catalyst. (a) pH-dependence of aqueous ethylene hydrogenation turnover rates (note logarithmic y-axis). (b) pH-dependence of the steady-state open-circuit potential (E_OCP) of a Pt wire during aqueous ethylene hydrogenation. (c) Pt–water interfaces are spontaneously polarized by proton transfer during and in parallel to ethylene hydrogenation catalysis. (d) E_OCP-dependence of aqueous ethylene hydrogenation turnover rates (note logarithmic y-axis). Conditions: 273 K, 5 kPa C₂H₄, 96 kPa H₂, 0.24 wt % Pt/C, solvent = 0.09–0.1 M NaClO₄ in water, 0.1 M total ionic strength. pH adjusted as follows: pH 2.0 ± 0.1 = 0.01 M HClO₄, pH 5.1 ± 0.2 = nothing added, pH 9.0 ± 0.02 = 10 mM borate buffer, pH 12.9 ± 0.1 = 0.01 M NaOH. Uncertainties in rates and pH values reflect standard deviations of independent trials. Note: At 273 K, pK_w = 10^–15, so pH ranges from 0–15 and neutral water is pH = 7.5.

The Pt wire sensing electrode provides a real-time estimate of changes in the polarization of the distributed Pt/C catalysts during the reaction and across conditions (see above). For unbuffered electrolytes (pH 2.0, 5.1, 12.9), Pt E_OCP values drift less than 8 mV during the initial stages of reaction (∼25 min) before reaching a steady-state value that varied by less than ∼2 mV over the remainder of the experiment (Figure S10a, b, d). For the borate-buffered electrolyte (pH 9.0), Pt E_OCP cathodically drifts ∼30 mV during the initial 80 min of reaction (which may, in part, stem from gradual changes to boron speciation and/or trace boric acid evaporation⁶⁹), before stabilizing within a range of ∼2 mV (Figure S10c). The steady-state E_OCP values for all four conditions were found to be linearly dependent on pH with a slope of −53 ± 3 mV per pH unit versus the pH-independent (i.e., constant) standard hydrogen reference electrode (Figure 5b). Notably, this value is within error of the expected scaling of −54 mV per pH unit from the Nernst equation for H₂/H⁺ interconversion at 273 K. This observation implies that, while ethylene hydrogenation is taking place, the Pt surface is simultaneously catalyzing the rapid interconversion of H⁺ and H₂ (Figure 5c). This is consistent with prior work on the Pt-catalyzed hydrogenation of an unsaturated diol⁴⁵ and implies that the olefin hydrogenation minimally perturbs hydrogen evolution/oxidation catalysis. This in turn could reflect the fact that liquid water has been observed to mitigate formation of site-blocking ethylidyne during ethylene hydrogenation on Pt.⁷⁰ These E_OCP data establish that, during aqueous ethylene hydrogenation catalysis, varying the solution pH directly alters the degree of spontaneous polarization at the Pt–water interface.

Importantly, this electrochemical link between pH and E_OCP implies that the magnitude of spontaneous interfacial polarization is inextricably tied to an increase in solution alkalinity. Thus, it is possible that the observed pH-dependence of ethylene hydrogenation is actually a manifestation of a polarization-dependence of the reaction, which is induced by pH changes. Indeed, the ethylene hydrogenation rates scale exponentially with Pt E_OCP (Figure 5d; see below for detailed discussion of the slope), suggesting that E_OCP and pH are equally viable descriptors of ethylene hydrogenation rates in water.

Spontaneous Polarization of a Distributed Pt–Liquid Interface via Electron Transfer

Given that changes in pH and E_OCP are inextricably linked in aqueous electrolytes, we sought to deconvolute pH and electrostatic effects by employing an orthogonal class of charge transfer reactions—outer-sphere electron transfer (OSET)—to spontaneously polarize a Pt–liquid interface in an aprotic electrolyte.

In these experiments, we deployed metallocene-based redox buffers to drive polarization via OSET in an ortho-difluorobenzene (o-DFB) solvent. o-DFB was chosen not only because it is relatively noncoordinating and effective at solvating electrolytes,⁷¹ but also because it is relatively inert with respect to hydrogenation. The redox buffers employed were decamethylferrocene (DMFc), decamethylmanganocene (DMMc), cobaltocene (Cc), and decamethylcobaltocene (DMCc), along with their respective one-electron oxidized hexafluorophosphate salts (Figure 6a). Whereas changes in proton activity served to tune the degree of polarization in water, here the degree of polarization is tuned by virtue of the distinct standard reduction potentials of the metallocene couples employed. Indeed, the family of metallocenes employed here display standard reduction potentials that span 1.4 V at 273 K (Figure S11, Table S4), providing a considerably larger range than the 0.81 V (i.e., [0.054 V per pH unit] × [15 pH units]) spontaneous polarization window accessible via pH variation in water at 273 K.

Figure 6.

Ethylene hydrogenation in ortho-difluorobenzene (o-DFB) electrolytes on a Pt/C catalyst. (a) Metallocene-based outer-sphere electron transfer reactions used to spontaneously polarize Pt during ethylene hydrogenation catalysis in o-DFB. (b) Steady-state Pt open-circuit potentials (Pt E_OCP) vs glassy carbon open-circuit potentials (GC E_OCP) during ethylene hydrogenation catalysis. (c) Pt–(o-DFB) interfaces are spontaneously polarized via outer-sphere electron transfer during and in parallel to ethylene hydrogenation catalysis. (d) Pt E_OCP-dependence of ethylene hydrogenation turnover rates in o-DFB electrolytes (note logarithmic y-axis). Conditions: 273 K, 5 kPa C₂H₄, 96 kPa H₂, 0.21 wt % Pt/C, electrolyte = 0.2 M tetrabutylammonium hexafluorophosphate in o-DFB. Metallocene concentration = 2 or 6 mM; metallocenium hexafluorophosphate concentration = 1 mM. DMFc = decamethylferrocene (M = Fe, R = CH₃). Error bars reflect standard deviations from independent trials.

Rate measurements in o-DFB were performed largely analogously to our aqueous rate measurements. Specifically, solutions contained 0.2 M tetrabutylammonium hexafluorophosphate (TBAPF₆) as the supporting electrolyte, 1 mM of the metallocenium hexafluorophosphate salt, and either 2 mM (for Cc, DMMc, and DMFc) or 6 mM (for DMCc) metallocene. Activities of dissolved ethylene and H₂ were kept the same as in the aqueous measurements by using the same partial pressures of each gas (i.e., 5 and 96 kPa, respectively). These nonaqueous measurements are highly sensitive to trace quantities of water and O₂ and were performed with careful solvent drying and rigorously air-free techniques (see SI section 1.6 for full experimental details.) The electrochemical cell depicted in Figure 4 was augmented with a glassy carbon (GC) electrode (Figure S12). Because glassy carbon does not activate H₂, it only engages in outer-sphere electron exchange with the metallocene redox buffer. Consequently, the glassy carbon open-circuit potential (GC E_OCP) reports on the metallocene redox buffer ratio in solution, as given by the Nernst equation, without convolution from any hydrogen-mediated polarization contributions. Thus, GC E_OCP here serves a role analogous to that of pH in the aqueous measurements.

Metallocene redox buffers in solution effectively “pin” the electrochemical potential of electrons in Pt during ethylene hydrogenation catalysis. Specifically, with DMFc, DMMc, and Cc redox buffers, Pt and GC E_OCP values are equal throughout the reaction duration, and drift less than 4 mV with time-on-stream (Figures S13 – S15). With DMCc, Pt and GC E_OCP values are also nearly identical with time-on-stream, but drift anodically by ∼60 mV during the first hour on-stream before stabilizing (Figure S16). This transient in GC E_OCP, specifically, implies an increase in the ratio of DMCc⁺ to DMCc; however, this leads to only a modest deviation in E_OCP from the Nernstian potential expected from the initial metallocene concentrations (Figure S17). Regardless, for all four metallocene redox buffers, steady-state Pt E_OCP values are pinned to within ∼3 mV of GC E_OCP values (Figure 6b). These results establish that Pt is essentially perfectly polarized during ethylene hydrogenation catalysis to be in electronic equilibrium with the solution-phase activities of reduced and oxidized metallocene equivalents (Figure 6c).

Furthermore, ATR-IR experiments indicate that metallocene-driven changes to Pt E_OCP correspond to changes in the degree of spontaneous polarization at the Pt–(o-DFB) interface. Specifically, the peak maximum of ν(CO_A) redshifts from 2056 cm^–1 with a DMFc redox buffer to 2026 cm^–1 with a more reducing DMMc redox buffer (Figure S18, Table S5). This two-point comparison corresponds to a Stark shift of approximately ∼60 cm^–1 V^–1. The consistency between our aqueous and nonaqueous Stark shifts agrees with the similarity across solvents reported for CO Stark tuning on a polycrystalline Pt disk with driven polarization.⁷² Although Stark tuning experiments have not, to our knowledge, been reported in o-DFB, our Stark shift exceeds by ∼3-fold values reported in the same study for similar polar aprotic electrolytes containing tetraalkylammonium salts.⁷² Akin to our aqueous results, this difference may be attributable to Pt particle size or CO coverage effects, or convolution from the large negative-going feature of the bipolar peak (see SI section 2.1 for further discussion). While these potential sources of convolution preclude a quantitative analysis of the Stark tuning slopes, the data strongly indicate that the metallocene OSET method employed here effectively spontaneously polarizes the distributed Pt/C interface. In this regard, we stress that the similarity in Pt and GC E_OCP values during catalysis does not imply that Δϕ is similar at the Pt–(o-DFB) and GC–(o-DFB) interfaces; indeed, Δϕ at the respective interfaces would deviate in accordance with the difference between the electron chemical potentials (i.e., work functions) for Pt and GC.

During catalysis in o-DFB, ethylene hydrogenation rates systematically decrease with metallocene-driven, negative spontaneous polarization. In the presence of a DMFc redox buffer (steady-state Pt E_OCP = −0.03 V vs DMFc^+/0), a rate of 1.4 ± 0.02 s^–1 was observed (Figure 6d). However, upon negative spontaneous polarization via DMMc (Pt E_OCP = −0.49 V vs DMFc^+/0), Cc (Pt E_OCP = −0.82 V vs DMFc^+/0), and DMCc (Pt E_OCP = −1.34 V vs DMFc^+/0) redox buffers, rates systematically decrease to 0.70 ± 0.04, 0.32 ± 0.02, and 0.066 ± 0.01 s^–1 (Figure 6d). This trend constitutes an exponential decrease in rate with decreasing Pt E_OCP, consistent with our aqueous rate measurements.

We note that the observed trend in rate cannot be ascribed to site-blocking due to metallocene adsorption on the Pt surface. Indeed, the metallocene buffer concentration does not appear to influence observed rates, as we observe an ethylene hydrogenation rate of 0.66 s^–1 in the presence of a 2.5-fold higher DMMc buffer concentration. This invariance in rate with DMMc buffer concentration (i.e., 0.70 ± 0.04 vs 0.66 s^–1), coupled with (1) the fact that the metallocenes share similar chemical structures and (2) the observation that the rate in the presence of the DMFc redox buffer is within ∼50% of the previously reported gas-phase rate,⁶⁴ suggests that metallocenes do not significantly block catalytic Pt sites in this system. We note, however, that this is not necessarily true for all classes of redox buffers; whether or not a redox buffer might interfere with catalysis would in general depend not only on redox buffer class but also on factors such as the solvent, catalyst, and catalytic reaction mechanism.

In view of the negligible site-blocking of metallocene redox buffers, the observation that ethylene hydrogenation is strongly and similarly sensitive to polarization across two distinct solvents (water vs o-DFB) and polarization mechanisms (proton transfer vs electron transfer) leads us to conclude that, even in water, E_OCP, rather than pH, is the controlling variable.

Comparison of Spontaneous Polarization-Dependence of Ethylene Hydrogenation in Aqueous and Nonaqueous Media

The foregoing data establish that, even though ethylene hydrogenation is a redox neutral, thermochemical reaction, it is nonetheless highly sensitive to the electrochemical potential of the interface. The data further indicate that “pinning” the electrochemical potential of the Pt surface via either proton transfer in water or outer-sphere electron transfer in o-DFB leads to a log–linear scaling between rate and catalyst potential (Figures 5d and 6d) across a wide range of 0.58 V in water and 1.31 V in o-DFB. These data strongly support the notion that the electrochemical potential (i.e., E_OCP) of the catalyst, a generally overlooked parameter in thermochemical catalysis, dramatically influences reaction rates at solid–liquid interfaces.

The slopes of these semilog rate vs E_OCP plots provide quantitative insight into the magnitude of interfacial polarization effects on thermochemical catalysis. In particular, these semilog plots provide what we define as an apparent electrochemical activation slope, S_AEA,

which provides a quantitative measure of the susceptibility of a reaction rate to changes in the electrochemical potential (i.e., E_OCP) of the catalyst. A similar quantity has been previously defined for the electrochemical promotion of catalysis at solid–gas interfaces.^73,74 Notably, we observe that S_AEA = 1.9 V^–1 for Pt-catalyzed ethylene hydrogenation in aqueous media (Figure 5d), which is similar to the value of 2.3 V^–1 observed in o-DFB electrolytes with metallocene-based spontaneous polarization (Figure 6d). Although ethylene hydrogenation is a redox neutral reaction, these S_AEA values may nonetheless be converted into an equivalent Tafel slope, which is commonly used to analyze the potential-dependence of a charge transfer reaction:

This analysis returns equivalent Tafel slopes of 1200 and 1000 mV decade^–1 for water and o-DFB electrolytes, respectively. These values are an order-of-magnitude larger than those typically observed for faradaic charge transfer reactions, which typically span from 30 to 200 mV depending on the mechanism of catalyzed charge transfer.⁷⁵ The substantially larger effective Tafel slopes observed here for thermocatalytic ethylene hydrogenation indicate that far greater electrical polarization is required to effect the same change in ethylene thermochemical hydrogenation rate as for a faradaic reaction. This in turn implies that the kinetically relevant elementary steps of liquid-phase ethylene hydrogenation depend far more on the properties of neutral molecules than on charged species (i.e., electrons and/or ions). While ethylene hydrogenation is far less potential-dependent than an electrochemical (i.e., charge transfer) reaction, we stress that in contrast to an electrochemical reaction, changes in E do not alter the overall reaction free energy of thermochemical ethylene hydrogenation. Indeed, since no current is flowing, the changes in reaction rate observed here arise without any additional energy flux to the system.

Although a detailed mechanistic analysis of the origin of these polarization effects is beyond the scope of this initial report, the quantitative similarity in S_AEA across vastly different reaction media and spontaneous polarization mechanisms implies that electrostatic effects are largely independent of the specific chemical constituents in the reaction medium. We note that a “local short-circuit” model,^101,76−79 wherein balancing H₂ oxidation and ethylene reduction half-reactions take place at the same interface, cannot explain these results because of the lack of proton donors in the nonaqueous measurements. Instead, although changes in E_OCP do not alter the overall thermodynamic reaction driving force for ethylene hydrogenation, the interfacial electric field can alter the affinity of the surface for polar solvent molecules^31,36 and augment the population and energies of surface intermediates and transition states, particularly those with dipolar character;^{4,31,33,39,49,80} these effects, in essence, reflect electrostatic nonidealities inherent to the thermodynamics of condensed-matter interfaces.

We note that this electrostatic phenomenology appears, at least for this system, to dwarf chemical solvation effects on catalytic rates. Indeed, ethylene hydrogenation rates measured under liquid-phase conditions at modest polarization (this study) are within ∼50% of previously reported gas-phase rates,⁶⁴ implying that chemical solvation of surface species does not play a significant role under these conditions.⁸¹ However, in contrast to the minimal chemical solvation effect, electrostatic polarization appears to have a significant impact on rate—for example, up to a 21-fold change in o-DFB—with increasing polarization attenuating rates relative to the gas-phase value.

Given the strong polarization dependence we observe for what is, ostensibly, one of the most nonpolar surface reactions, these studies suggest that spontaneous electrical polarization may play a general role in defining the rates of a diverse array of thermochemical transformations at metal–liquid interfaces.

Conclusion

The studies described herein establish that spontaneous electrical polarization, a property inherent to all solid–liquid interfaces, plays an important role in governing reaction rates in thermochemical heterogeneous catalysis. By combining electrochemical measurements with infrared spectroscopy of CO on Pt—an interface-specific probe sensitive to polarization—we establish that distributed Pt–liquid interfaces are spontaneously electrically polarized via equilibrated proton transfer in water and equilibrated electron transfer from organometallic redox buffers in an aprotic, organic o-DFB solvent. Despite the disparate mechanisms of polarization and the disparate reaction media, we find that both modes of polarization influence the catalytic rate of an explicitly nonpolar probe reaction—liquid-phase ethylene hydrogenation on Pt/C—with similar scale factors. This points toward a general electrostatic phenomenology that may be inherent to all catalysis at metal–liquid interfaces across diverse reaction media.

The foregoing understanding has important implications for mechanistic inquiry and the design of efficient catalytic processes. In particular, our findings highlight that the electrochemical potential of the catalyst, E_OCP—a generally overlooked parameter in heterogeneous catalysis—warrants consideration in mechanistic studies of reactions at solid–liquid interfaces. Moreover, our work establishes simple methods for leveraging electrostatics to complement existing chemical approaches to elucidating surface chemistry and designing catalytic interfaces. More broadly, given that spontaneous polarization is operative at distributed interfaces, our approach offers a straightforward and scalable strategy to impose oriented electric fields to manipulate preparative chemical transformations, a long-standing challenge in chemistry.^18,19 Indeed, this work highlights the value of merging concepts from the historically disparate communities of heterogeneous catalysis and electrochemistry in order to advance the science of catalysis in condensed-phase media.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acscentsci.1c00293.

• Full experimental details, additional ATR-IR data, CO stripping data, additional rate data, additional E_OCP data, Pt particle size distributions, cyclic voltammograms (PDF)

The authors declare no competing financial interest.