How palladium inhibits CO poisoning during electrocatalytic formic acid oxidation and carbon dioxide reduction

The study addresses how palladium-based electrocatalysts avoid CO poisoning during formic acid oxidation (FAO) and carbon dioxide reduction reaction (CO2RR), two reactions central to direct formic acid fuel cells and reversible CO2/formate electrochemistry. On Pt, FAO proceeds via a dual-pathway mechanism: a desirable direct path to CO2 via a reactive intermediate (formate-related) and an indirect path that forms adsorbed CO, which poisons the surface. Pd-based catalysts show high FAO activity with minimal CO poisoning and are also effective for CO2RR to formate at low overpotentials. The key questions are whether FAO and CO2RR on Pd share similar intermediates/mechanisms and how Pd inhibits CO poisoning in both directions, with implications for reversible, unitized regenerative systems using the CO2/formate couple.

Prior work established Pt’s dual-path FAO mechanism with CO as a poisoning intermediate, limiting Pt’s practical utility. Pd-based materials (including Pd monolayers on Pt and Pd–Pt nanoparticles) exhibit high FAO activity and remarkable resistance to CO poisoning. Pd catalysts are also among the best for CO2RR to formate at low overpotentials. Epitaxial Pd monolayers on Pt single crystals provide model systems mimicking Pd single-crystal behavior while avoiding Pd bulk hydrogen absorption complexities. Previous studies suggest adsorbed formate is a stable, bidentate spectator species rather than directly oxidized to CO2, and that hydride-like surface H may be the active species for CO2RR to formate on Pd. These insights motivate a comparative mechanistic study of PdMLPt(111) vs Pt(111).

• Electrochemistry: Cyclic voltammetry (CV) performed in 0.1 M HClO4 with and without 50 mM HCOOH at room temperature using a three-electrode cell (Pt wire counter, RHE reference). Pt(111) and Pt(100) single crystals (bead-type; 2.27 and 3.46 mm diameters) for CV and 10 mm disks for online HPLC measurements; hanging meniscus configuration; Autolab PGSTAT302N and Bio-Logic SP-300 potentiostats. High-scan-rate CV up to 50 V s−1 to resolve reversible adsorption processes (e.g., formate) from kinetically controlled FAO currents; conventional scans at 50 mV s−1. Rotating electrode at 1600 rpm for cycling experiments.
• CO stripping: CO adsorption by holding at 0.1 V_RHE in CO-saturated electrolyte for 30 s, purge with Ar for 15 min, then oxidative stripping (10 mV s−1) to quantify CO coverage and stripping potentials.
• CO2RR product analysis: Electrolyte purged with CO2; potential scanned at 1 mV s−1. Online HPLC sampling via a capillary positioned ~10 μm from the electrode at 60 μL min−1, collecting 60 μL per sample (averaging over ~60 mV potential window). Subsequent positive-going scans used to strip and quantify adsorbed CO formed during CO2RR.
• Electrode preparation: Pd monolayers (PdML) deposited on freshly prepared Pt(111) and Pt(100) using cyclic voltammetry between 0.07–0.85 V_RHE at 50 mV s−1, monitoring characteristic Pd adatom peaks (e.g., 0.23 V_RHE for Pt(111)). STM literature indicates formation of monoatomic Pd islands and full monolayers without holes; ordered sulfate adlayers observed. For PdML on Pt(100), NO adsorption/desorption was used as electrochemical annealing to obtain full PdML coverage.
• DFT calculations: VASP with PAW and PBE functional. Slabs: PdMLPt(111) (Pt(111) covered by one Pd ML), Pt(111), Pd(111); 3×3 and 2×2 surface cells for coverages 0.11–0.33 and 0.25 ML; 5, 6, and 4 atomic layers respectively (bottom layers fixed at bulk lattice constants; Pt/PdMLPt 3.98 Å, Pd 3.93 Å). K-point meshes: up to 6×6×1; plane-wave cutoff 450 eV; Methfessel-Paxton smearing 0.2 eV for surfaces; Gaussian 0.001 eV for gas-phase molecules. Vacuum ~15 Å with dipole corrections. Adsorbates studied: *H, *OCHO (formate), *COOH, *CO, including explicit solvation models for *COOH (2 H2O) and *OCHO (1 H2O) where noted. Free energies computed using CHE for proton–electron transfers, with gas-phase energy corrections for CO2 and CO. Bader charge analysis and work function calculations performed to assess adsorbate charge and surface electronic properties.

• FAO voltammetry:
  • Pt(111): Onset ~0.35 V_RHE; peak current ~2.2 mA cm−2 (positive scan). Deactivation observed when cycling to 0.6 V_RHE (no CO stripping), with FAO current decreasing ~4× after 12 cycles due to CO accumulation. CO stripping peak at ~0.72 V_RHE.
  • PdMLPt(111): Onset ~0.20 V_RHE; peak current ~11.0 mA cm−2 at 0.38 V_RHE (∼4× Pt current at 0.20 V lower potential). Minimal hysteresis between forward/reverse scans from 0.05–0.40 V_RHE, indicating absence of CO poisoning. CO cannot be oxidatively stripped until 0.80–0.90 V_RHE, showing slower CO oxidation kinetics vs Pt(111).
  • CO coverage from stripping: Pt(111) ~0.69 ML; PdMLPt(111) ~0.75 ML, implying CO binds at least as strongly on PdMLPt(111), so the lack of poisoning during FAO is not due to weaker CO binding.
• Formate adsorption isotherms (fast CV, 50 V s−1):
  • Pt(111): Formate adsorption feature 0.38–0.65 V_RHE; maximum coverage ~0.25 ML at 0.65 V_RHE.
  • PdMLPt(111): Formate adsorption begins ~0.20 V_RHE, competitive with H desorption, and saturates between 0.33–0.45 ML by 0.40 V_RHE (0.33 ML minimum assuming co-adsorption; 0.45 ML maximum assuming electrosorption valency −1 and no co-adsorbed OH/ClO4−). Stronger formate adsorption linked to lower work function and higher anion affinity.
• CO2RR products and poisoning:
  • PdMLPt(111): Formic acid production starts at −0.29 V_RHE, peaking around −0.60 V_RHE. Adsorbed CO formation begins more negative than −0.475 V_RHE and saturates by ~−0.70 V_RHE; formate production decreases as CO coverage increases.
  • Pt(111): No measurable formic acid; adsorbed CO forms at potentials more negative than −0.25 V_RHE.
• DFT thermodynamics and co-adsorption effects:
  • Adsorption trends (1/9 ML, with solvation): On PdMLPt(111)/Pd(111): |*OCHO| > |*COOH| > |*H| > |*CO| (formate and COOH stronger than H, CO weakest). On Pt(111): |*OCHO| > |*H| > |*COOH| > |*CO|; notably *COOH more stable on Pt(111) than on PdMLPt(111).
  • Formate binds much more strongly on PdMLPt(111) than on Pt(111) at all coverages; above ~0.33 ML, bidentate adsorption becomes unfavorable.
  • Co-adsorbed formate weakens *CO adsorption markedly on PdMLPt(111): at 0.33 ML formate, *CO becomes ~0.20 eV less favorable than *H, suppressing CO formation; on Pt(111), the difference at similar coverage is only ~0.05 eV, and lower attainable formate coverage (~0.25 ML) leaves sites for *COOH and *CO formation.
  • Limiting potentials for *COOH-sol formation (relative to CO2, H+, e−): Pt(111) 0.49 V_RHE, Pd(111) 0.15 V_RHE, PdMLPt(111) 0.23 V_RHE, indicating *COOH forms at less negative potentials on Pt, consistent with earlier CO poisoning.
  • Work functions: PdMLPt(111) 5.14 eV < Pd(111) 5.29 eV < Pt(111) 5.74 eV; lower work function correlates with higher anion/formate affinity. H* carries more negative partial charge on PdMLPt(111)/Pd(111) than on Pt(111), supporting a hydride-like character implicated in CO2 → formate via nucleophilic attack.
• Mechanistic insights:
  • FAO on PdMLPt(111): High formate coverage (~1/3 ML) blocks ensemble sites needed for CO formation and weakens *CO/*COOH adsorption, preventing CO poisoning while allowing rapid direct pathway oxidation (likely via C–H-down formate interacting with free Pd sites).
  • CO2RR: On PdMLPt(111), hydride-like H* likely mediates formate formation at low overpotentials; at more negative potentials *COOH becomes favorable and CO poisoning develops. On Pt(111), stronger *COOH adsorption leads to early CO formation and no formate production.

The findings demonstrate that Pd monolayer-modified Pt(111) suppresses CO poisoning during FAO through strong formate adsorption leading to high surface coverage. This coverage both geometrically blocks the requisite multi-atom ensembles for CO formation from *COOH and electronically weakens *CO and *COOH adsorption, shifting stability toward *H and favoring non-CO pathways. Conversely, Pt(111) attains lower formate coverage, leaving accessible ensembles and maintaining sufficient *COOH stability, leading to rapid CO accumulation and deactivation. During CO2RR, the poisoning behavior is governed by COOH thermodynamics: its higher stability on Pt(111) leads to earlier CO formation than on PdMLPt(111). The hydride-like nature of H on PdMLPt(111) supports a nucleophilic mechanism for formate formation, offering a molecularly reversible picture relative to FAO (C–H cleavage vs formation). Thus, while the catalytic interconversion of formate/CO2 can be mechanistically reversible on Pd-based catalysts, the onset of poisoning differs between oxidation and reduction due to distinct controlling intermediates and coverages.

Using Pd monolayers on Pt(111) as model Pd electrocatalysts, the study reveals that high-coverage formate adsorption on PdMLPt(111) (≈1/3 ML) is central to inhibiting CO poisoning during formic acid oxidation by blocking ensemble sites and weakening *CO/*COOH adsorption. PdMLPt(111) exhibits higher FAO activity and lower onset potential than Pt(111), with negligible hysteresis indicative of the absence of CO poisoning. In CO2 reduction, PdMLPt(111) produces formate at low overpotentials (onset −0.29 V_RHE) and only forms CO at more negative potentials (≤ −0.475 V_RHE), whereas Pt(111) forms CO at much less negative potentials and does not generate formate. DFT supports these observations, attributing delayed poisoning on PdMLPt(111) to weaker COOH adsorption and highlighting the hydride-like character of H on Pd surfaces as key for formate formation. Overall, the mechanisms for FAO and CO2RR are consistent with a reversible interconversion via C–H bond making/breaking on Pd, while the poisoning mechanisms are distinct and potential-dependent. These insights provide design principles for CO-poisoning-resistant, reversible catalysts for the CO2/formate couple, relevant to unitized regenerative systems. Future work could directly probe intermediates under operando conditions and explore facet- and composition-dependent formate coverages and hydride character across Pd-based materials.

• The exact molecular mechanism for formate formation during CO2 reduction is hypothesized based on thermodynamics and charge analysis; there is no direct experimental identification of the reactive intermediate under reaction conditions.
• Hydrogen absorption/subsurface hydride effects on bulk Pd(111) were not modeled; conclusions are drawn from Pd monolayer systems where absorption is absent.
• On PdMLPt(100), the very high FAO activity prevented determining the saturation formate coverage via fast voltammetry, limiting validation of the protection mechanism on this facet.
• Generalization to all Pd facets and nanoparticle systems is inferred but not directly verified in this study.