Conventional catalyst research primarily focuses on optimising the static topology of surface active sites. However, this approach often overlooks the dynamic, time-dependent processes that dictate optimal pattern formation and the creation of transient intermediate structures essential for enhancing reaction rates. Drawing inspiration from nature, where enzymes function as "active matter" by coupling chemical reactions with autonomous motion, this work introduces a novel theoretical framework for catalyst optimisation.
By incorporating the time dimension through the Maximum Entropy Production (MEP) principle, we move beyond the concept of an unchanging energy landscape. We propose a non-equilibrium, nonlocal reaction-cross-diffusion (RXD) formulation to model the catalyst surface as an excitable medium. This approach captures the dynamic interactions between the catalyst and target molecules in both space and time, identifying "anti-chemotactic" behaviours where catalysts effectively perform a biased random walk to maximise reaction encounters.
The RXD framework provides a robust mathematical basis for bridging the gap between atomic-scale simulations and industrial-scale applications. It offers a rule-based foundation for future deep learning models and multiphysics upscaling (THMCE), enabling the design of next-generation carbon catalysts that are not only structurally optimised but dynamically tuned for a carbon-constrained world.