Fundamental insight into electrochemical oxidation of methane towards methanol on transition metal oxides

Significance

The electrochemical conversion of CH₄ to liquid fuel such as CH₃OH offers a sustainable and environmentally benign route to utilize shale gas. This process requires controlled oxidation by optimal splitting and formation of C–H and C–O bonds to produce desired products at ambient conditions, which has been a long-standing challenge in electrocatalysis of hydrocarbons. Here, we combine electrochemical experiments with quantum chemical calculations to develop activity descriptors for identification of efficient electrocatalysts that can partially oxidize CH₄ to CH₃OH. This work provides much deeper insights into potential-driven breakage/formation of C–H and C–O bonds, which will greatly advance the scientific understanding of hydrocarbon electrochemistry.

Abstract

Electrochemical oxidation of CH₄ is known to be inefficient in aqueous electrolytes. The lower activity of methane oxidation reaction (MOR) is primarily attributed to the dominant oxygen evolution reaction (OER) and the higher barrier for CH₄ activation on transition metal oxides (TMOs). However, a satisfactory explanation for the origins of such lower activity of MOR on TMOs, along with the enabling strategies to partially oxidize CH₄ to CH₃OH, have not been developed yet. We report here the activation of CH₄ is governed by a previously unrecognized consequence of electrostatic (or Madelung) potential of metal atom in TMOs. The measured binding energies of CH₄ on 12 different TMOs scale linearly with the Madelung potentials of the metal in the TMOs. The MOR active TMOs are the ones with higher CH₄ binding energy and lower Madelung potential. Out of 12 TMOs studied here, only TiO₂, IrO₂, PbO₂, and PtO₂ are active for MOR, where the stable active site is the O on top of the metal in TMOs. The reaction pathway for MOR proceeds primarily through *CH_x intermediates at lower potentials and through *CH₃OH intermediates at higher potentials. The key MOR intermediate *CH₃OH is identified on TiO₂ under operando conditions at higher potential using transient open-circuit potential measurement. To minimize the overoxidation of *CH₃OH, a bimetallic Cu₂O₃ on TiO₂ catalysts is developed, in which Cu reduces the barrier for the reaction of *CH₃ and *OH and facilitates the desorption of *CH₃OH. The highest faradaic efficiency of 6% is obtained using Cu-Ti bimetallic TMO.

Electrochemical oxidation of methane (CH₄) at ambient conditions offers a sustainable route for efficient utilization of abundant natural resources, such as shale gas and biogas. However, the lower activity and selectivity of current electrocatalysts pose hurdles for the large-scale deployment of electrochemical technologies for efficient utilization of CH₄ (1). Currently, the majority (∼66%) of CH₄-rich resources are burned to produce electricity or to provide heating for the residential and commercial buildings, which contributes ∼1 gigaton of CO₂ emissions annually. CH₄ is also utilized to produce oxygenated chemicals such as CH₃OH using industrial processes like steam reforming followed by gas-phase conversion or direct thermocatalytic conversion (2). While the thermocatalytic routes often require high temperature and pressure and suffer from catalyst poisoning (3–5), the electrochemical technologies offer environmentally benign and sustainable routes for storing electrical energy by converting CH₄ and H₂O to CH₃OH and for generating electrical energy using a direct CH₄ fuel cell that primarily generates CO₂ (4, 6, 7). However, the primary challenge in such electrochemical processes is the first step of CH₄ activation on electrocatalysts, which is difficult at ambient conditions owing to its high C–H bond energy of 439 kJ/mol (8), high symmetry with tetrahedral molecular geometry, low polarizability of 2.448 ų (9), the low solubility of 1.272 mM in the water at standard temperature and pressure (10), and competitive oxygen evolution reaction (OER) (11, 12). A complete understanding of the mechanism of electrochemical activation followed by oxidation of CH₄ and its competition with OER on transition metal oxides (TMOs) has not been developed yet. Consequently, there is a need for the experimental and theoretical evaluation of the activity and selectivity of methane oxidation reaction (MOR) on various TMOs to elucidate the origins of higher activation energy and competitive kinetics of MOR and OER and to identify strategies for selective synthesis of CH₃OH under ambient conditions.

There are very few experimental reports available on the kinetics of electrochemical MOR at ambient conditions (13). The majority of the MOR work focuses on the high-temperature electrocatalysis in galvanic cell configuration, such as solid oxide fuel cells (SOFCs), in which the primary objective is to harvest electrical energy by fully oxidizing CH₄ to CO₂ (14–18). The conventional SOFCs operating at a temperature range of 300 to 700 °C (17, 19) using Ni-based composite anode (20–22) have also been studied for partial oxidation of CH₄ to hydrocarbons (e.g., CO, C₃H₄, C₂H₆, and CH₃OH), but their operating efficiencies drop rapidly because of carbon deposition (coking) (6, 19, 23). The low-temperature electrolytic systems operating at temperatures <120 °C using Pt, Platinized-Pt, or Pt/Au catalysts have also been reported for partial oxidation of CH₄ to CH₃OH, but the faradaic efficiencies (FE) are too low for a practical purpose (11, 24–26). A comprehensive table of the current density, FE, and reaction products for MOR on various electrocatalysts along with electrolyte composition, temperature, and operating potential are provided in SI Appendix, Table S1. So far, only precious metals like Pt, Pd, Ru, and Au have been explored for MOR either under harsh chemical conditions or at very high overpotentials (11, 27–29). Further experimental evaluation of activity, selectivity, and mechanism of MOR over other transition metal catalysts is required to identify efficient MOR catalyst and uncover fundamental limits of CH₄ activation on these catalytic systems. It is also desirable to develop a theoretical understanding of the competitive mechanism of MOR and OER on these TMOs and identify the noncompetitive routes for partial oxidation of CH₄.

Theoretical investigations of CH₄ adsorption via C–H bond activation followed by electrochemical oxidation are necessary to identify the activity descriptors and fundamental scaling relationships for MOR on TMOs (30). CH₄, being a stable, nonpolar molecule, is known to bind weakly on the transition metals through a dissociative mechanism (31), whereas all the CH_x (x = 1, 2, and 3) intermediates bind more strongly. Psofogiannakis and coworkers (32) showed that CH₄ oxidation to CO is preferable through the dissociative chemisorption of CH₄ on Pt with the sequence${}^{\ast }C{H}_3{\to }^{\ast }C{H}_2{\to }^{\ast }CH{\to }^{\ast }CHOH{\to }^{\ast }CHO{\to }^{\ast }CO$. A similar study showed, depending on the surface structure, the oxygen-assisted dehydrogenation drives the selectivity of the reaction to either CH_xO_y or CO_x products (33). The presence of oxygen on transition metals in the form of metal oxides has been determined to play a key role in C–H bond activation and the oxidation of CH₄ (34–36). Under the anodic overpotentials, the dissolved CH₄ dissociates to form *O–CH₃ (CH₃ bonded to metal oxide) and *O–H. Because of H–C–H bond symmetry in CH₄ and CH₃, the energy needed to break (or activate) the C–H bond is equal to the energy required to transfer H from CH₄ to *O (metal oxide). Therefore, the suggested activity descriptor for MOR is E_OH–E_O, whereas the activity descriptor for OER based on the rate-limiting step (*OH → *O) is E_O–E_OH (37–40). The participation of *O species in both MOR and OER is the primary cause for the competitive kinetics determining the selectivity of CH₄ oxidation on TMOs.

Regardless of the catalyst structure and composition, there has been an agreement in acknowledging that MOR proceeds through a dissociative mechanism. However, in the oxidation regime of the electrocatalytic MOR, the metals exist in the form of their oxides, and one descriptor (either E_H or E_OH–E_O) may not be sufficient to identify efficient catalyst, as the MOR activity can also depend on the binding energy of the CH_x to the metal oxide surface, and the availability of metal oxide sites to bind with CH_x and consequently to suppress OER (40, 41). Rossmeisl and coworkers (40) elucidate this further by studying the limitations of CH₃OH production on different TMO (110) surfaces. They show that activation energy for MOR is correlated with the reactivity of an adsorbed oxygen atom from the electrolyte. A catalyst is providing more reactive oxygen results in low energy barriers for OER, which leads to the suppression of MOR. Here, we hypothesize that the higher MOR activity requires higher values of E_OH–E_O, the higher surface coverage of *O and a noncompetitive binding site of *O. Hence, studying the activity descriptors in combination with the estimation of the binding energy of the CH_x species, the surface coverage of oxygen ad-atoms, and the availability of competitive versus noncompetitive sites for CH_x adsorption on TMOs are expected to provide mechanistic insights to identify active electrocatalysts.

In this article, we recognize the experimental challenges associated with C–H bond activation and mass transfer limitations of CH₄ in aqueous electrolytes as well as theoretical shortcomings because of a lack of comprehensive information about the MOR mechanism over TMOs. To address these challenges, we conduct controlled electrochemical experiments to study the activity and selectivity of CH₄ oxidation over 12 TMOs, Sc₂O₃, TiO₂, ZrO₂, Fe₂O₃, Co₃O₄, IrO₂, NiO, PtO₂, Cu₂O₃, ZnO, SnO₂, and PbO₂, in near-neutral and alkaline electrolyte using rotating-disk electrode cell, followed by theoretical evaluation of efficient MOR catalysts using density functional theory (DFT). To elucidate the mechanism of MOR, we implement experimental studies to measure the binding energy of CH₄ using reactant–impulse chronoamperometry (RIC), detect key reaction intermediate CH₃OH using transient open-circuit potential (t-OCP) measurement, and measure reactions product using chromatography.

The objectives of this study are to relate the activity of MOR to the binding energy of CH₄, identify the noncompetitive active sites for MOR-active catalysts, and determine reaction pathways and strategies to direct partial oxidation of CH₄ to CH₃OH. The remainder of the article is organized as follows. Methods explains the DFT calculations to estimate the adsorption energies and activation barriers of MOR and OER on TMO surfaces. Experimental Materials and Methods describes the setup of the electrochemical cell, preparation of the electrodes, the binding energy estimation, product distribution, and t-OCP experiments. Results and Discussion presents FE for the active TMOs, optimal binding energy range for MOR estimated by experimental methods, reaction profiles of MOR and OER on TMO surfaces showing relative energies of the adsorbed intermediates by computational methods, a discussion on favorable reaction pathways for MOR by combining both experimental and computational methods, and a strategy to produce CH₃OH using CuTi bimetallic catalyst. Conclusions and Perspectives presents conclusions and prospects in choosing/synthesizing efficient electrocatalysts for CH₄ oxidation.

Methods

DFT Calculations to Estimate Binding Energies and Free Energy Profile.

Estimated energies of adsorption and the free energy profile on four metal oxide surfaces, IrO₂, PbO₂, TiO₂, and SnO₂, have been calculated with DFT Vienna Ab initio Simulation Package (VASP) 5.3.5 (42). Calculations were performed using the Perdew-Burke-Ernzerhof (PBE) exchange-correlation functional with Grimme’s D3 corrections (43, 44). Here, the metal oxide (110) surfaces were modeled using a four-layer (4 × 4) periodic slab, and the successive slabs were separated by at least 25 Å of vacuum. Adsorption was allowed only on one side of the slab. Partial occupancies for each orbital were populated using the Fermi smearing method with a width of 0.1 eV. All energies reported were sampled at a 4 × 4 × 1 Monkhorst−Pack k-points and a 350 eV planewave cutoff. The convergence of total energy for the planewave cutoff and the k-point set are summarized in SI Appendix, section S.13. The computational hydrogen electrode model was used to calculate the energy of reaction intermediates, and an applied potential of 2.11 V versus reversible hydrogen electrode (RHE) is applied in all energy calculations (45). Solvation effects were modeled by reporting all energies with the corrections gained from the implicit solvation model, VASPsol (46).

Experimental Materials and Methods.

Electrochemical product distribution.

Transition metal plates of Sc, Ti, Fe, Co, Ni, Cu, Zn, Zr, Ir, Pt, Sn, and Pb (>99.9% purity, ACI Alloys) of 1-mm thickness were cut into disks of 8-mm diameter and polished using alumina suspensions followed by sonication in deionized water and drying under Ar flow. The Cu–Ti bimetallic catalyst was prepared by electrodepositing Cu on Ti disk from 0.1 M Cu (NO₃)₂ (pH = 2) at −2 V versus Ag/AgCl for 45 min. The disk was finally weighed to determine ∼10% loading of Cu on Ti. Electrochemical experiments were performed in a custom, three-dimensionally printed, two-compartment H-Cell separated by an anion exchange membrane (Excellion, SnowPure technologies) with the transition metal disks as the working electrodes, Ag/AgCl microreference electrode (Innovative Instruments), Pt strip as the counter electrode, and ∼6 mL of a well-stirred 0.1 M phosphate buffer solution (pH 7.0) or 0.1 M KOH (pH 13.0). Linear sweep voltammetry was performed under oxidation potentials to develop an oxide layer, which was followed by chronoamperometry in a CH₄-sparged H-Cell for ∼2 h at potentials in the range of 1.5 to 2.4 V versus RHE. At the interval of 20 min, the gas products evolved were detected by passing the outlet from the H-Cell to an SRI GC 8610C MG#5 to quantify gaseous products. A detailed experimental setup and the experimental procedure can be seen in SI Appendix, section S.2.

Characterization of electrodes.

The metal oxide electrodes for MOR were characterized by scanning electron microscopy (SEM), energy dispersive X-ray spectroscopy (EDS), and X-ray photoelectron spectroscopy (XPS) before and after electrocatalysis. SEM and EDS were performed using Hitachi S4800 Field Emission SEM. XPS measurement was conducted using Thermo Scientific ESCALAB 250XI microprobe with an Al Kα source. Details of characterization can be found in SI Appendix, section S.5.

Reactant-impulse chronoamperometry for estimation of CH₄ binding energy.

The dissolved CH₄ in the electrolyte must bind or at least interact with the active sites on the electrode before the activation of the C–H bond can occur (47). It has been postulated that the activation of the C–H bond involves the dissociation of CH₄ to form *O–CH₃ and *O–H (40). The energy of such dissociative adsorption of CH₄ is difficult to measure. However, the binding energy of CH₄ in the preactivation step (* + CH₄ → *CH₄) can be measured and correlated with the CH₄ activation energy. Here, we perform an RIC in a rotating disk electrode (RDE) cell by switching the Ar-saturated electrolyte with the CH₄-saturated electrolyte at a potential lower than the onset potential for MOR (e.g., 1.52 V versus RHE). The Ar-saturated and CH₄-saturated electrolytes were prepared by presaturating respective gases in a 0.1 M potassium phosphate buffer at pH 7. The dynamic variation in the OER current density was measured by sequentially varying electrolyte feed in the following manner: first feeding Ar-saturated electrolyte for 10 to 15 min, followed by CH₄-saturated electrolyte for 10 to 15 min, and then back to Ar-saturated electrolyte. In general, the CH₄-saturated electrolyte was found to suppress the OER current density, which could occur by either direct binding of CH₄ to the free active sites (*) or binding of CH₄ to the O atom of OER intermediates (*OH, *O, *OOH, and *O₂). Therefore, the fractional coverage of *CH₄ can be estimated by calculating the change in the OER current density as described below (see SI Appendix for details):

$$\theta =\frac{I_{OER,\text{Ar}}-I_{OER,{\text{CH}}_4}}{I_{OER,\text{Ar}}},$$ [1]

where $\theta$ is the fractional coverage of the *CH₄ on the electrode surface, $I_{OER,\text{Ar}}$ is the OER current density in Ar-saturated electrolyte, and $I_{OER,{\text{CH}}_4}$ is the OER current density in CH₄-saturated electrolyte. The equilibrium constant $K$ for the CH₄ adsorption/desorption

$$C{H}_4{+}^{*}{\rightleftharpoons }^{*}C{H}_4$$

can be obtained from the Everett isotherm equation for a dissolved CH₄ in contact with a TMO surface (48).

$$K=\frac{\theta \left(1-x^{*}\right)}{x^{*}\left(1-\theta \right)},$$ [2]

where $x^{\ast }$ is the mole fraction of dissolved CH₄ in the electrolyte. The binding energy can be estimated from the Gibbs free energy relation

$$\mathrm{\Delta }G=RT\ln \left(K\right),$$ [3]

where $R$ is the universal gas constant $\left(8.314\hspace{0.17em}\text{J}\hspace{0.17em}{\text{mol}}^{-1}{\text{K}}^{-1}\right)$and $T\left(298\text{.}15\hspace{0.17em}\text{K}\right)$ is the temperature at which the experiments were performed.

t-OCP measurement for CH₃OH detection.

The OCP identifies a mixed potential (or corrosion potential) when the multiple redox reactions—OER and MOR—are in equilibrium at net zero currents. The variation in OCP can be related to the variation in the activity of reactants and products in the double layer. Here, we use t-OCP measurement as a means to identify key reaction intermediates of MOR. The first step in t-OCP measurement is to perform chronopotentiometry to obtain a steady potential and FE at the desired current, which ensures fractional coverages of reaction intermediates also attain a steady state. The next step is to shift the applied current to zero and record t-OCP as the reaction intermediates go back from dynamic to static equilibrium. The t-OCP graph is sensitive to reaction pathways and coverages of stable reaction intermediates.

The t-OCP values for OER in Ar-saturated 0.1 M phosphate buffer were measured for initial currents of 0.25 mA and 0.5 mA in a custom-built H-cell with TiO₂ on RDE. The steady-state values of t-OCP were compared for 0.25 mA and 0.5 mA to verify the current-independent behavior of steady-state OCP for OER. Next, the t-OCP values for MOR were measured for an initial current of 0.25 mA (at 1.8 V versus RHE) and 0.5 mA (at 2.1 V versus RHE) on TiO₂ in a CH₄-saturated 0.1 M phosphate buffer solution. Since CH₃OH is one of the possible intermediates of MOR (32, 36), the variation in the fractional coverage of *CH₃OH will have a direct effect on the steady-state values of t-OCP. To identify the formation of *CH₃OH on TiO₂, a calibration curve was prepared by measuring the change in the steady-state values of t-OCP for increasing CH₃OH concentrations from 0 to 10 mM, which was added manually to CH₄-saturated 0.1 M phosphate buffer solution. This calibration curve was used to identify *CH₃OH produced at a higher current—0.5 mA. The detailed experimental setup and the experimental procedure can be seen in SI Appendix, section S.2.

Results and Discussion

Product Distribution and Faradaic Efficiency of CH₄ Oxidation in Near-Neutral and Alkaline Electrolytes.

Fig. 1 shows the product distribution obtained from the oxidation of CH₄-saturated electrolyte using the 12 stable TMO electrodes in neutral (pH 7.2) and alkaline (pH 13) conditions. The missing first-row transition metals such as V, Cr, and Mn in Fig. 1 were not stable in a wide pH range. The linear sweep voltammograms of all catalysts are shown in SI Appendix, section S.6. The product distributions under acidic conditions were also not investigated, as most TMOs dissolve in the oxidation regime, which can be seen from their Pourbaix diagrams in SI Appendix, section S.4. While O₂ is the dominant product on the majority of TMOs, CO₂ is produced only on TiO₂, IrO₂, and PbO₂ in phosphate buffer solution. Since phosphate ions can specifically adsorb and poison Pt surfaces (49, 50), PtO₂ did not show any activity toward MOR in phosphate buffer. However, a significant selectivity (∼10% FE) to MOR is observed on PtO₂ in 1 M KCl electrolyte. Comparing Fig. 1 A and B, it can be observed that these four active MOR catalysts show a higher FE in the neutral medium. This is because there is a higher concentration of OH⁻ in alkaline medium (0.1 M OH⁻) that specifically adsorbs to the positive electrode and oxidizes to produce O₂, which reduces the availability of the free site for MOR.

Fig. 1.

Product distribution and FE of OER and MOR over stable TMOs in (A) 0.1 M KOH (pH = 13) and (B) 0.1 M potassium phosphate buffer (pH = 7). PtO₂ was found to be MOR inactive in 0.1 M potassium phosphate buffer. The shaded stacked bar for PtO₂ in B shows the FE and the product distribution using 1 M KCl (pH = 7).

Fig. 2A shows the variation of FE of MOR producing CO₂ with increasing the applied potential of the active TMOs in the neutral pH phosphate buffer electrolyte. The onset potentials for MOR are found to be 1.65, 1.9, and 2.1 V versus RHE for TiO₂, IrO₂, and PbO₂, respectively. These onset potentials are the lowest applied potentials for which the produced CO₂ was detected in GC. Fig. 2A also shows that TiO₂ has the highest FE of 13.5% at 2.31 V versus RHE. The decrease in FE of MOR at higher applied potential can be due to mass transfer limitations of dissolved CH₄. Fig. 2B shows the partial current densities of MOR producing CO₂ at different applied potentials. The Tafel slopes of MOR increase in the following order: PbO₂, 190 mV/dec; TiO₂, 232 mV/dec; and IrO₂, 274 mV/dec. The exchange current density also follows a similar trend: PbO₂, 6.5 × 10⁻¹³ mA/cm⁻²; TiO₂, 9.2 × 10⁻¹¹ mA/cm⁻²; and IrO₂, 6.6 × 10⁻⁹ mA/cm⁻². IrO₂ has the highest intrinsic rate of MOR but suffers greatly from the higher Tafel slope. The larger Tafel slopes of these three TMOs are indicative of passive oxide layer formation, which increases the resistance for electron transfer (51).

Fig. 2.

(A) Faradaic efficiency and (B) partial current density of MOR producing CO₂ on TiO₂, IrO₂, and PbO₂ at different applied potentials in neutral pH phosphate buffer electrolyte.

Measurement of CH₄ Binding Energy on TMOs and Activity Descriptors for MOR.

To further understand the origin of the MOR activity on IrO₂, TiO₂, PbO₂, and PtO₂, the binding energy of CH₄ on different TMOs are measured using RIC as described in Experimental Materials and Methods. Several experimental and theoretical works have reported that CH₄ adsorbs physically on the metal oxides, followed by the activation of the C–H bond (52–55). Here, we implement the RIC technique to measure the binding energy of CH₄ adsorption on TMOs in the potential range below the onset potential of CH₄ activation in MOR. In brief, the decrease in OER current density was measured by exchanging Ar-saturated electrolyte with CH₄-saturated electrolyte. The adsorption of CH₄ on TMOs decreases the active sites for OER and thereby decreases the OER current density, which can be used to estimate fractional coverage and binding energy of *CH₄ on TMOs. Fig. 3A shows the schematic of this reversible process of CH₄ binding to the TMO and its influence on OER intermediate coverages. Fig. 3B shows the effect of reversible adsorption/desorption on the OER current density on TiO₂ by cyclic feeding of Ar-saturated electrolyte followed by CH₄-saturated electrolyte at a fixed potential of 1.52 V versus RHE (which is less than onset potential of MOR). The relative change in the OER current density provides the fractional change in the coverage of *CH₄ (see Eq. 2), which can be used in the Everett isotherm (Eq. 3 with Eq. 4) to obtain the binding energy of $*{\mathrm{CH}}_4\left(E_{*{\mathrm{CH}}_4}\right)$. Fig. 3C shows the binding energy of *CH₄ on 12 TMOs in near-neutral pH electrolyte at 1.52 V versus RHE. The measured binding energy of *CH₄ decreases with increasing the group number or decreasing the period number in the periodic table. A similar trend of adsorption energies of CH₄ on transition metals has been reported elsewhere (56, 57). It can be seen in Fig. 3C that the binding energies of CH₄ on MOR active catalysts—TiO₂, IrO₂, and PbO₂—are similar and in the range $0.24\pm 0.01\hspace{0.17em}\mathrm{eV}$. PtO₂ is also a MOR active catalyst in neutral pH KCl electrolyte with the binding energy close to the optimal range. Also, the free energy change for the first step of MOR $\left(E_{\ast C{H}_4}-E_{\ast C{H}_3}\right)$ for TiO₂, IrO₂, and PbO₂ is similar (Fig. 3C). Therefore, the binding energy of CH₄ can be considered as one of the activity descriptors for MOR, which governs only the surface coverage of *CH₄. The calculated surface coverages of *CH₄ on different TMOs are shown in SI Appendix, Fig. S3. The TMOs with the binding energy of CH₄ less than 0.23 eV are not active for MOR because of the lower surface coverages of *CH₄. However, there are a few catalysts such as Sc₂O₃, ZrO₂, and SnO₂ that have binding energies greater than 0.23 eV and are not active for MOR. This is due to competitive active sites for OER and MOR present on these catalysts (see Competitive Pathways for OER and MOR on TMOs).

Fig. 3.

(A) Schematic representation of OER intermediates on (110) surface of TMOs in Ar-saturated and CH₄-saturated electrolytes. As the Ar-saturated electrolyte is swapped with CH₄-saturated electrolyte, the CH₄ adsorbs on the TMO and reduces the number of active sites for OER. Switching the CH₄-saturated electrolyte back with Ar-saturated electrolyte allows CH₄ desorption and an increase in the number of active sites for OER. Light gray spheres are metal atoms, dark gray spheres are the oxygen atoms from the oxide layer, blue spheres represent oxygen from the electrolyte, red represents carbon, and white represents hydrogen atoms. (B) Reversible change in the OER current density with adsorption and desorption of CH₄ at a fixed applied potential of 1.52 V versus RHE. (C) Estimated binding energies of *CH₄ obtained using Everett isotherm for different TMOs at 1.52 V versus RHE. The MOR active TMOs—TiO₂, IrO₂, PbO₂, and PtO₂—have similar binding energy in the range 0.22 to 0.25 eV and similar activation energy in the range 0.09 to 0.12 eV. Black squares show the first-row TMOs, red circles show the second-row TMOs, and the blue triangles show the third-row TMOs. (D) Scaling relationship between measured binding energy of *CH₄ and the Madelung potential of metal in TMOs. The MOR active catalysts—TiO₂, IrO₂, PbO₂, and PtO₂—have a higher binding energy of *CH₄ and lower Madelung potential.

The mechanism of CH₄ adsorption involves the transformation of tetrahedral $\left(T_d\right)$ symmetry to the H–C–H bond-angle–distorted $\left(D_{2d}\right)$ structure followed by C–H bond elongation and adoption to $\left(C_s\right)$ conformation (47). These symmetry transformations can also be explained using the molecular orbital theory. The $T_d$ symmetry of CH₄ (with bond angle 109.5°) has two occupied orbitals—$a_1$ and $t_2$—and two unoccupied antibonding orbitals—$a_1^{\ast }$ and $t_2^{\ast }$. The interaction of CH₄ with TMO causes the threefold degenerate $t_2$ set to split into the destabilized $b_2$ state because of reduced overlap between 2p and 1s orbitals and the stabilized twofold degenerate $e$ state because of an increase in C–H bonding character. Similarly, the unoccupied $t_2^{\ast }$ orbital splits into a stabilized $b_2^{\ast }$ state and destabilized a twofold degenerate $e^{{}^{*}}$ state because of the antibonding character of C–H bonds. The destabilization of $t_2$ and $t_2^{\ast }$ orbitals results in the distortion of adsorbed CH₄ to attain $D_{2d}$ conformation with an increased H–C–H bond angle of 120°. The formation of the distorted structure $\left(D_{2d}\right)$ is then followed by the elongation of one of the C–H bonds to attain $C_s$conformation. The degree of C–H bond elongation governs the C–H bond activation on different TMOs. The schematics of molecular transformations during adsorption $\left(T_d\to D_{2d}\right)$ of CH₄ followed by activation $\left(D_{2d}\to C_s\right)$ are shown in SI Appendix, Fig. S12.

The interactions between CH₄ and TMO are mostly electrostatic in the nonfaradaic region of MOR. The coulombic interaction between TMO and CH₄ helps in the stabilization of $b_2$state of $t_2$ orbital and thereby the adsorption of CH₄. Therefore, the electrostatic (or Madelung) potential of the metal in the TMO should have a direct effect on the measured binding energy of CH₄. Fig. 3D shows a nearly linear trend between the measured binding energy and the Madelung potential of metal in TMOs. The Madelung potentials for TMOs were obtained from ref. 58. The MOR active catalysts have lower Madelung potentials (< −40 V) as compared to the other inactive catalysts. It can be inferred from Fig. 3D that the MOR active catalysts should have higher methane binding (>0.23 eV) energy and lower Madelung potential (< −40 V). All the MOR active catalysts—TiO₂, PbO₂, IrO₂, and PtO₂—satisfy this activity criteria. A few outliers here are ZrO₂ and SnO₂, which satisfy the activity criteria but are not active toward MOR. One of the reasons could be the presence of shared active sites for competitive OER and MOR on ZrO₂ and SnO₂ (see details in Competitive Pathways for OER and MOR on TMOs ). Also, the poor electrical conductivity of ZrO₂ could be the reason for the lower activity of OER and inactivity of MOR.

Competitive Pathways for OER and MOR on TMOs.

The MOR activity descriptors shown in Fig. 3D indicate six catalyst candidates that could be active for MOR, namely, ZrO₂, TiO₂, IrO₂, PbO₂, PtO₂, and SnO₂. Experimental evaluation of 12 TMOs in Fig. 1 shows TiO₂, IrO₂, PbO₂ and PtO₂ are the only four active catalysts for MOR. The other two catalysts—SnO₂ and ZrO₂—are not active for MOR. To further understand the activity and selectivity of MOR on the four candidate catalysts, TiO₂, IrO₂, PbO₂, and SnO₂, the OER and MOR pathways and their energy profiles were calculated using DFT. Experimental conditions dictate that these TMO surfaces are oxidized, but there is little information about the nature and distribution of the oxidation states on these catalysts’ surfaces. Also, it is well known that the defects on these surfaces can affect reaction pathways (45). Therefore, we evaluate OER pathways on two extreme surface conditions, fully reduced and fully oxidized states, and then study MOR only on the fully oxidized surface. The following steps for OER were considered, in which ${}^{\ast }$ indicates the bare TMO surface:

$${}^{*}{\text{H}}_2\text{O}\hspace{0.17em}\to {\hspace{0.17em}}^{*}\text{OH}+\hspace{0.17em}\frac{1}{2}{\text{H}}_{2\left(\text{g}\right)}$$

$${}^{*}\text{OH}\hspace{0.17em}\to {\hspace{0.17em}}^{*}\text{O}+\hspace{0.17em}\frac{1}{2}{\text{H}}_{2\left(\text{g}\right)}$$

$${}^{*}\text{O}+\hspace{0.17em}{\text{H}}_2{\text{O}}_{\left(\text{g}\right)}\to {\hspace{0.17em}}^{\text{*}}\hspace{0.17em}+\hspace{0.17em}{\text{O}}_{2\left(\text{g}\right)}+{\text{H}}_{2\left(\text{g}\right)}.$$

Fig. 4 A and B show that IrO₂ readily catalyzes OER on both the reduced and oxidized surfaces, which agrees well with previous theoretical and experimental findings (59). However, the OER pathways on PbO₂, SnO₂, and TiO₂ surfaces are more complicated and strongly dependent on their oxidation states. The OER pathways on TiO₂ and SnO₂ proceed preferentially on the reduced surface at the metal center binding site, which suggests that the coordinatively unsaturated metal centers are the active sites for OER. The OER pathway on PbO₂ proceeds on the oxidized surface at the bridging oxygen binding site. Also, PbO₂ is the only surface on which it is energetically favorable for the bridging oxygen to dissociate from the surface, as seen in SI Appendix, Fig. S25. The DFT-predicted OER activity on the sampled surfaces increases in the order PbO₂ (bridging O) < SnO₂ (on-top M) < TiO₂ (on-top M) < IrO₂ (on-top M), which aligns very well with the increasing order of experimental current densities for PbO₂ (0.2 mA/cm²), SnO₂ (0.3 mA/cm²), TiO₂ (2.5 mA/cm²), and IrO₂ (3.2 mA/cm²) at 2.1 V versus RHE shown in SI Appendix, Fig. S11. This indicates that the chosen computational methods can qualitatively describe the relative activities of these catalytic systems.

Fig. 4.

(A) Reaction profile for OER at the metal with oxygen binding site for the fully reduced surface (B) Reaction profile for OER at the bridging oxygen binding site for the fully oxidized surface. The relative energies of reaction are shown as a function of reaction coordinate, labeled by *X, where X indicates the adsorbed intermediate. To the right of the plot are pictorial representations of the adsorbed intermediate. Colors denote Sn (silver), Pb (gray), O (red), and H (pink). (C) Reaction profile for MOR at the metal with oxygen binding site for the fully oxidized surface. The relative energies of the reaction are shown as a function of reaction coordinate, labeled by *X, where X indicates the adsorbed intermediate along with the pictorial representations of the adsorbed intermediate. Colors denote Ir (gold), O (red), C (brown), and H (pink). All electronic energies are reported using PBE+D3 at an applied potential of 2.11 V versus RHE.

To study MOR on these four TMO surfaces, we considered the following steps:

$$\begin{gathered} {}^{*}+{\text{CH}}_{4\left(\text{g}\right)}{\to }^{*}{\text{CH}}_3+\frac{1}{2}{\text{H}}_{2\left(\text{g}\right)}{ \\ }^{*}{\text{CH}}_3{\to }^{*}{\text{CH}}_2+\frac{1}{2}{\text{H}}_{2\left(\text{g}\right)} \end{gathered}$$

$${}^{*}{\text{CH}}_2{\to }^{*}\text{CH}+\frac{1}{2}{\text{H}}_{2\left(\text{g}\right)}$$

$${}^{*}\text{CH}+{\text{H}}_2{\text{O}}_{\left(\text{g}\right)}{\to }^{*}\text{CO}+\frac{3}{2}{\text{H}}_{2\left(\text{g}\right)}$$

$${}^{*}\text{CO}+{\text{H}}_2{\text{O}}_{\left(\text{g}\right)}{\to }^{*}+{\text{CO}}_{2\left(\text{g}\right)}+{\text{H}}_{2\left(\text{g}\right)},$$

where *CH to *CO is a composite step consisting of *CH → *HCOH → *HCO → *CO.

Fig. 4C shows the first step in the catalytic cycle for MOR is comparable in energy to that of OER, which also validates E_OH–E_O as the common activity descriptors for OER and MOR. The reaction steps here show MOR occurring on a metal center that is occupied by an oxygen atom (M–O, see SI Appendix, Table S2). This binding site for catalytic mechanisms has been discussed in zeolites and metal–organic frameworks, but to the best of our knowledge, this mechanism has had little discussion in metal oxides (30). SI Appendix, Figs. S15–S17 show the MOR pathways on alternative binding sites. The first step for MOR in Fig. 4C is slightly higher in energy as compared to OER in Fig. 4 A and B, which is aligned with the experimental observation of lower FEs of MOR on these TMOs. Based on the energies of the reaction pathways on different types of binding sites on TMOs, the M–O binding site is where MOR occurs favorably for all experimentally active systems. M–O is also a recognized intermediate for the OER catalytic cycle; therefore, this binding site is readily available for catalysis on these surfaces. However, the coverage of the M–O site can vary on these catalysts, as the formation of the M–O site is more favorable thermodynamically on TiO₂ and IrO₂ as compared to SnO₂ and PbO₂ (Fig. 4A). Although the MOR is more energetically downhill on PbO₂ as seen in Fig. 4C, the lower coverage of M–O sites on this catalyst is a reason for its lower MOR activity as compared to TiO₂ (Fig. 2). The reaction steps also indicate that MOR proceeds readily on TiO₂, PbO₂, and IrO₂ but not SnO₂, as the first step in the catalytic cycle is positive for this surface. These DFT results justify the observed MOR activity on TiO₂, PbO₂, and IrO₂ catalysts. To gain a further understanding of the reaction kinetics, the reaction barriers need to be calculated, which is planned for future studies.

Identification of *CH₃OH Intermediate of MOR and Strategy to Harvest CH₃OH.

The reaction pathway for MOR proceeds primarily through *CH_x (x = 1, 2, 3, and 4) intermediates at lower potentials, as shown in Fig. 4C. However, a few reports have indicated the possibility of forming oxygenated intermediates such as *CH₃OH, *HCHO, and *HCOOH during MOR on TMOs (40, 60, 61). Fig. 5A shows alternate pathways for MOR going through oxygenated intermediates starting from *CH₃OH → *HCHO → *HCOOH → *CO₂ and a pathway for OER in an aqueous electrolyte on TMOs. The identification of oxygenated intermediates using spectroscopic methods can be very challenging at higher applied potential and current density (62, 63). An alternate method to identify reaction intermediate is by direct measurement of the OCP. OCP is similar to the Nernst potential for a single redox reaction, in which the variation in the local concentration of reactants and products has a direct effect on the OCP value. The stable OCP values were measured on TiO₂ for the initial current of 0.25 mA (1.8 V) with increasing concentrations of CH₃OH—0, 2, 3, 5, and 10 mM—added to CH₄-saturated phosphate buffer electrolyte. Fig. 5B shows the increase in stable OCP with increasing the logarithm of CH₃OH concentration, which is similar to the Nernst relation. The corresponding t-OCP values for different concentrations of CH₃OH are shown in the SI Appendix. Fig. 5B is used as a calibration curve to relate the change in the OCP of TiO₂ to the formation of *CH₃OH. The t-OCP is then measured on TiO₂ for initial currents of 0.25 mA (1.8 V) and 0.5 mA (2.6 V) in CH₄-saturated electrolyte (with no added CH₃OH). The increase in stable OCP values from 176 mV for 0.25 mA to 269 mV for 0.5 mA suggests an increase in *CH₃OH formation, as the concentrations of other species—CH₄, H⁺, OH⁻, H₂O, and O₂—are constant. Although *CH₃OH was identified on TiO₂, the concentration of free CH₃OH in the electrolyte was below the measurement limit that can be due to the overoxidation of *CH₃OH to *HCHO, *HCOOH, and *CO₂. SI Appendix, Fig. S13 also confirms CH₃OH oxidizes on TiO₂ at potentials >1.9 V versus RHE. To further confirm the formation of these oxygenated intermediates, similar t-OCP measurements were conducted on TiO₂ using varying concentration of HCOOH (SI Appendix, section S.11). The stable OCP increased with the increase in the logarithm of HCOOH concentration, similar to the Nernst relation. This indicates both *CH₃OH and *HCOOH could be present on TiO₂ at higher applied potentials. Hence, it is likely that MOR proceeds through two different mechanisms that are potential dependent. At lower applied potentials, MOR progresses through sequential CH_x dissociation, while at higher potentials, MOR proceeds through the formation of *CH₃OH, and in both cases, CO₂ is the major MOR product formed over TiO₂.

Fig. 5.

(A) Possible reaction pathways for MOR and OER in an aqueous electrolyte where M represents TMO. The species in green are the possible MOR products. (B) The shift in the stable OCP with increasing concentrations of CH₃OH in phosphate buffer electrolyte for an initial current of 0.25 mA at 1.8 V. The linear relation is similar to the Nernst equation. The increase in stable OCP from 176 mV for 0.25 mA to 269 mV for 0.5 mA of initial current in CH₄-saturated phosphate buffer electrolyte without added CH₃OH indicates the formation of *CH₃OH on TiO₂. (C) FE and partial current density of CH₃OH on Cu₂O₃–TiO₂ bimetallic catalyst in CH₄-saturated, 0.1 M potassium phosphate buffer.

Under a positive bias >1.4 V, TiO₂ could overcome the barrier for *CH₄ dissociation to yield lower fragments of *CH_x of higher binding energy (39), such that the TMO surface is mostly populated with *CH₃, leading toward the formation of CO₂. To produce CH₃OH, the *CH₃-populated TMOs must be brought in contact with another metal that can provide *OH and prevent overoxidation of CH₃OH. Lin et al. showed that the energy barrier for the reaction of *CH₃ with *OH is lowest for Cu—0.17 eV—among other transition metals (64). Cu not only has a lower barrier for CH₃OH formation (64) but also is resistant to further oxidation of CH₃OH (SI Appendix, Fig. S13). Fig. 5C shows the FE and partial current density of CH₃OH on Cu₂O₃–TiO₂ bimetallic TMOs. The maximum FE of 6% and current density of 0.35 mA/cm² are obtained for CH₃OH on Cu–Ti bimetallic TMOs.

Conclusions and Perspectives

In this article, we develop a fundamental understanding of electrochemical CH₄ oxidation on TMOs and report the binding energy of CH₄ on TMOs under relevant conditions, activity descriptors of MOR, identification of active sites for MOR, detection of *CH₃OH intermediate and its oxidation product, and strategies to harvest CH₃OH by minimizing its overoxidation.

The electrochemical oxidation of CH₄ on TMOs proceeds with the physical adsorption of CH₄ followed by the activation of C–H bonds forming either CH_x intermediates or oxygenated intermediates to yield products such as CO₂, CO, CH₃OH, or HCHO. The physical adsorption of CH₄ involves the transformation of the tetrahedral $\left(T_d\right)$ symmetry of CH₄ to a bond-angle–distorted $\left(D_{2d}\right)$ structure with an H–C–H bond angle of ∼120°. The physical adsorption is governed by the electrostatic interaction of CH₄ with the metal of TMO. To understand the physical adsorption process on TMOs, the binding energy of CH₄ is measured using pulsed-reactant chronoamperometry on 12 stable TMOs in neutral pH electrolyte. The measured binding energy of CH₄ scales linearly with the electrostatic (or Madelung) potential of metal in TMOs. We hypothesize that the MOR active catalysts should have higher CH₄ binding energy (>0.23 eV) and lower Madelung potential (<−40 V). Out of 12 TMOs studied here, TiO₂, IrO₂, PbO₂, PtO₂, SnO₂, and ZrO₂ satisfy the MOR activity criteria of higher binding energy and lower Madelung potentials. However, the experimental measurements show only TiO₂, IrO₂, PbO₂, and PtO₂ are the active catalysts for MOR. The inactivity of ZrO₂ and SnO₂ toward MOR is due to the poor electrical conductivity and lower population of MOR active sites, respectively.

The competitive pathways of OER and MOR are studied on the promising MOR catalysts—TiO₂, IrO₂, PbO₂, and SnO₂—using DFT to identify the active sites and the mechanism of the oxidation reactions. While the OER pathway on SnO₂, TiO₂, and IrO₂ is favorable on the undercoordinated metal site, this reaction on PbO₂ occurs on the bridging O site. The DFT-predicted OER activity on the sampled surfaces increases in the order PbO₂ (bridging O) < SnO₂ (on-top M) < TiO₂ (on-top M) < IrO₂ (on-top M), which aligns very well with the increasing order of experimental current densities. The M–O intermediate formed on these TMOs is also the most favorable active site for MOR. The energy profiles of MOR on M–O sites are energetically favorable on these TMOs, except SnO_2, for which the first dehydrogenation step is positive. The lower OER current and hence lower M–O active site coverages on SnO₂, along with the higher energy barrier for MOR, are the reasons for its inactivity toward MOR.

We also postulate that the reaction pathway for MOR proceeds primarily through *CH_x (x = 1, 2, 3, and 4) intermediates at lower potentials and through *CH_xO_y intermediates at higher potentials. Here, we develop a technique of t-OCP measurement to identify key reaction intermediates under operating conditions. The steady-state OCP varies linearly with the logarithm of CH₃OH and HCOOH (oxidation product of CH₃OH) concentration in the electrolyte, which is similar to the Nernst potential. The increase in the stable OCP value from 176 mV for 0.25 mA to 269 mV for 0.5 mA confirms the formation of *CH₃OH and its oxidation products at higher potentials. However, the overoxidation of *CH₃OH prevents the accumulation of CH₃OH in measurable quantity in the electrolyte. To produce CH₃OH, we show the *CH₃-populated TiO₂ can be mixed with Cu₂O₃ that can provide *OH and prevent overoxidation of CH₃OH. The preliminary result shows that 10% Cu₂O₃ on TiO₂ can increase the desorption of *CH₃OH and hence its FE to 6%. This work provides strategies that can be implemented to produce liquid fuels from CH₄ using bimetallic TMOs.

Data Availability

All study data are included in the article and/or SI Appendix.