Bridging Nanostructure Growth and Gas-Sensing Kinetics in Metal-Functionalized Graphene Using Machine-Learned Molecular Dynamics

Graphene functionalized with catalytic transition metals offers high-performance chemiresistive gas sensing by coupling graphene’s exceptional electronic transport with the metal’s catalytic activity; yet the atomistic relationships connecting synthesis parameters, morphological outcomes, and sensor gas–surface reaction kinetics remain elusive. We developed an equivariant machine-learning interatomic potential with DFT accuracy to enable high-fidelity molecular dynamics (MD) simulations, from metal nanostructure growth on graphene to device-level sensing kinetics. Demonstrating our approach, we specifically investigate Pt-functionalized graphene for H 2 detection. MD simulations validated by TEM show that Pt deposition begins with dispersed nuclei coalescing into polycrystalline nanoclusters, while both MD and Raman spectroscopy reveal predominantly noncovalent metal–graphene interactions that induce moderate local strain and doping, while preserving graphene’s structural integrity. MD simulations confirm H 2 dissociative chemisorption and recombinative desorption primarily on Pt nanoclusters, with negligible spillover or chemical interaction with pristine graphene. However, H adsorption on Pt attenuates the Pt–graphene interfacial binding, providing an indirect electronic pathway for gas sensing. Adsorption/desorption kinetics reveal that an intermediate loading of metal nanostructures minimizes the detection limit; lower loadings facilitate faster response and recovery kinetics and enhance signal transduction, whereas higher loadings increase interfacial binding and graphene doping. The developed machine-learned MD framework accurately models metallic nanostructure growth on graphene, elucidates the gas-sensing mechanism, and correlates figures of merit─detection limit, sensitivity, and response/recovery times─extracted from gas–surface kinetics with metal nanostructure morphology, establishing a multiscale predictive pipeline from synthesis conditions to gas-sensor kinetics.