Field-Enhanced catalysis: Integrating experiment, theory, and machine learning for catalytic innovation

Abstract

Understanding and controlling the local electric field distributions at catalyst interfaces offers a powerful strategy to modulate reaction kinetics by orders of magnitude, primarily through tuning the electrostatic interactions between polarized reactants and active sites. However, direct measurement of these localized fields with atomic-scale spatial resolution under operando conditions remains an experimental challenge. This perspective provides a critical overview of state-of-the-art techniques for probing local electric fields, including Kelvin Probe Force Microscopy and Vibrational Stark Effect spectroscopy. These approaches are complemented by advanced density functional theory (DFT) method, such as grand canonical DFT and dipole layer-slab models, which enable the simulation of electrostatic potential profiles and prediction of vibrational responses of surface-bound probe molecules under applied fields. Furthermore, we examine the growing role of machine learning (ML), particularly graph neural networks (GNNs) and generative models, in accelerating the prediction of local electric field distributions and the discovery of catalysts tailored for field-enhanced performance. These data-driven models capture complex, nonlinear relationships between catalyst structure, charge redistribution, and electrostatic properties, leveraging physically interpretable descriptors such as effective dipole moments and polarizabilities to predict field-dependent adsorption energetics. The integration of high-fidelity simulations, ML frameworks, and advanced experimental validation platforms establishes a comprehensive and scalable paradigm for rational catalyst design. Collectively, this work outlines a multidimensional approach for characterizing and exploiting local electric fields as dynamic tuning knobs in catalysis, with broad implications across electrocatalysis, photocatalysis, plasma catalysis, and microwave-assisted catalytic systems. • In situ spectroscopic techniques reveal structure-sensitive reactivity under strong fields. • DFT calculations quantify field-dependent adsorption with atomic-scale precision. • Field-dependent adsorption is predicted via dipole moments and polarizability descriptors. • GNN-based deep learning predicts field-dependent energetics with high efficiency. • Physics-informed and generative AI models enable inverse design of field-enhanced catalysis.

Topics & Concepts