A Electrochemical Framework for Predicting Catalytic Selectivity in CO2 Reduction

Thesis Abstract

Electrochemical CO2 reduction presents a complex interplay between catalyst surface properties, reaction intermediates, and mass transport, which collectively govern product selectivity and energy efficiency. This study addresses the pervasive gap in predictive capability for catalytic selectivity by developing an integrative electrochemical framework that links instrumental measurements, microkinetic modeling, and data-driven inference to forecast product distributions under diverse operating conditions. The aim is to formulate a theory-driven framework capable of predicting selectivity trends for CO2-to-CO, formate, and hydrocarbons across transition metal and oxide catalysts, while accommodating electrolyte effects and mass-transport limitations. Specific objectives include (i) characterizing catalyst surface descriptors (binding energies, roughness, and defect density) and their correlation with experimentally observed selectivity; (ii) constructing a multiscale model that couples density functional theory-derived adsorption intermediates with microkinetic simulations and finite-element mass-transport analysis; (iii) integrating experimental polarization curves, in situ spectroscopic signatures, and product distributions into a unified predictive algorithm; (iv) validating the framework against a curated dataset of 50 benchmark catalysts and 200 reaction conditions, and (v) assessing the framework’s extrapolative power for novel catalyst compositions. Methodologically, the study adopts a mixed-methods design grounded in theory-driven modeling and empirical validation. The population comprises heterogeneous electrocatalysts including copper, silver, gold, tin, and zinc-based systems, with carbon-supported and oxide-supported variants, tested under varying CO2-saturated aqueous electrolytes (0.1–1.0 M KHCO3) and current densities (-10 to -300 mA cm-2). A stratified sampling approach yields a dataset of 50 catalysts, each characterized by surface descriptors obtained from X-ray photoelectron spectroscopy (XPS), scanning electrochemical microscopy (SECM), and transmission electron microscopy (TEM). Data collection employs chronopotentiometry and chronoamperometry for polarization behavior, in situ attenuated total reflectance infrared spectroscopy (ATR-IR) and Raman spectroscopy for intermediate tracking, and gas chromatography–mass spectrometry (GC-MS) for liquid and gaseous products, producing a comprehensive matrix of current density, potential, selectivity, and catalyst descriptors. Instrumental measurements are complemented by ex situ X-ray absorption near-edge structure (XANES) analyses to assess oxidation states and local coordination environments. Analytical methods integrate first-principles insights with data-driven inference. Density functional theory (DFT) calculations provide adsorption energies and transition-state barriers for key CO2 reduction intermediates on representative facets, informing a microkinetic model. This model is solved to yield rate expressions for competing pathways, while finite-element simulations capture diffusion-layer and electrolyte transport effects. A Bayesian hierarchical framework fuses simulation outputs with experimental data, updating posterior probabilities for descriptor–selectivity relationships and enabling uncertainty quantification. Regression analyses (multivariate and ridge) identify robust correlations between surface descriptors and product selectivity, and ANOVA tests compare performance across catalyst classes and electrolyte conditions. The overarching model specification is a coupled electrochemical-mass-transport-microkinetic framework with probabilistic inference to accommodate measurement variability and model inadequacy. Expected findings include (i) identification of robust descriptor sets—such as atop-site binding energy of COOH*, surface roughness metrics, and defect density—that reliably predict selectivity shifts between CO, formate, and hydrocarbons; (ii) demonstration that mass transport constraints and bicarbonate buffer composition modulate selectivity via local pH and intermediate concentration gradients; (iii) validation that the integrated framework reproduces experimentally observed trends across 50 catalysts with ±15% predictive error in product fractions under tested conditions; and (iv) quantification of predictive confidence intervals for novel catalysts, enabling directed synthesis. The study contributes to knowledge by providing a transferable, theory-informed predictive framework that bridges quantum-chemical insights, microkinetic dynamics, and reactor-scale transport phenomena, thus advancing rational catalyst design for CO2 electroreduction. The conclusions are anticipated to support recommendations for catalyst design prioritizing descriptor tunability and reactor conditions that maximize CO2-to-value products while suppressing undesired hydrogen evolution. Recommendations include expanding the descriptor space to incorporate dynamic reconstruction effects under operation and applying the framework to tandem and hybrid catalytic systems to broaden predictive applicability in industrially relevant electrolytes.