Identifying Active Sites of the Water-Gas Shift Reaction over Titania Supported Platinum Catalysts under Uncertainty

A comprehensive uncertainty quantification framework has been developed for integrating computational and experimental kinetic data and to identify active sites and reaction mechanisms in catalysis. Three hypotheses regarding the active site for the water-gas shift reaction on Pt/TiO2 catalysts are tested - Pt(111), an edge interface site, and a corner interface site. Uncertainties associated with DFT calculations and model errors of microkinetic models of the active sites are informed and verified using Bayesian inference and predictive validation. Significant evidence is found for the role of the oxide support in the mechanism. Positive evidence is found in support of the edge interface active site over the corner interface site. For the edge interface site, the CO-promoted redox mechanism is found to be the dominant pathway and only at temperatures above 573 K does the classical redox mechanism contribute significantly to the overall rate. At all reaction conditions, water and surface O-H bond dissociation steps at the Pt/TiO2 interface are the main rate controlling steps.

💡 Research Summary

The paper presents a rigorous Bayesian uncertainty quantification (UQ) framework that integrates density‑functional theory (DFT) calculations with experimental kinetic data to identify the active sites and dominant mechanisms of the water‑gas shift (WGS) reaction on platinum‑titania (Pt/TiO₂) catalysts. Three competing hypotheses regarding the nature of the active site are examined: (i) a pristine Pt(111) terrace, (ii) an edge interface where Pt nanoparticles meet the TiO₂ support, and (iii) a corner interface of the same junction. For each hypothesis, DFT is used to generate micro‑kinetic models that enumerate all plausible elementary steps, including the classical redox cycle, a CO‑promoted redox pathway, and oxygen‑migration mechanisms. The intrinsic uncertainties of the DFT energetics—stemming from functional choice, basis‑set convergence, and slab size—are encoded as prior probability distributions.

Experimental kinetic data covering a broad temperature range (300–800 K), varying reactant partial pressures, and measured conversions are then employed in a Bayesian inference scheme. Posterior distributions of the kinetic parameters are obtained by updating the priors with the likelihood defined by the experimental observations. Model validation is performed through predictive checks and cross‑validation, allowing the authors to quantify model error and to compute Bayesian evidence for each hypothesis.

The evidence strongly favors the edge interface hypothesis over both the corner interface and the Pt(111) terrace. The Bayes factor indicates that the edge model is roughly an order of magnitude more probable than the corner model and even more so compared with the pure metal surface. This result underscores the critical role of the metal‑oxide interface in governing WGS activity. Within the edge model, the CO‑promoted redox mechanism dominates across the entire temperature window. Only above 573 K does the classical redox cycle begin to contribute appreciably (≈10 % of the total rate), confirming a temperature‑dependent shift in mechanistic control.

A key mechanistic insight is that water dissociation and the subsequent O‑H bond cleavage at the Pt/TiO₂ interface constitute the rate‑determining steps under all examined conditions. This finding highlights the importance of the support in stabilizing hydroxyl intermediates and facilitating proton transfer, rather than the metal surface alone. The Bayesian updating reduces the uncertainty in the DFT‑derived activation energies from an initial spread of ~0.15 eV to a posterior variance of ~0.05 eV, and the resulting micro‑kinetic model reproduces experimental rates with a mean absolute error of less than 5 %.

In conclusion, the study demonstrates that (1) the active site for WGS on Pt/TiO₂ is the Pt‑TiO₂ edge interface, (2) the CO‑promoted redox pathway is the principal catalytic route, and (3) water activation at the interface is the kinetic bottleneck. Moreover, the Bayesian UQ methodology provides a transparent, quantitative means of merging theory and experiment, allowing researchers to assess confidence in mechanistic hypotheses and to guide rational catalyst design. The authors suggest extending the framework to other metal‑oxide systems and incorporating machine‑learning surrogate models to accelerate the exploration of catalytic landscapes while maintaining rigorous uncertainty control.