Data-Driven Molecular Dynamics and TEM Analysis of Crystal Growth and Hydrogen Sensing in Pt-Functionalized Graphene Chemiresistive Sensors

Graphene functionalized with catalytic transition metals offers high-performance gas sensing by coupling graphene's exceptional electronic transport properties with the metal's catalytic activity, yet the atomistic relationships connecting synthesis parameters, morphological outcomes, and sensor performance remain elusive. We developed an equivariant machine-learning interatomic potential (MLIP) with near-DFT accuracy to perform molecular dynamics (MD) simulations of Pt crystal growth on graphene and subsequent H$_2$-sensing. MD simulations validated by TEM show that Pt deposition begins with dispersed nuclei coalescing into polycrystalline nanoclusters with predominantly FCC interiors, while both MD and Raman spectroscopy reveal a predominantly non-covalent Pt-graphene interaction that induces moderate local strain and charge transfer yet preserves graphene's structural integrity. Reactive MD simulations confirm H$_2$ dissociative chemisorption exclusively on Pt, with negligible spillover onto pristine graphene. However, H adsorption on Pt attenuates the Pt-graphene interfacial binding, providing an indirect electronic pathway for H$_2$ sensing. Transient and steady-state kinetic analyses demonstrate that an intermediate Pt loading minimizes the limit of detection; lower loadings facilitate faster response and recovery kinetics and enhance signal transduction, whereas higher loadings increase the doping level of graphene. DFT charge analysis indicates that under-coordinated Pt clusters induce $n$-type doping in graphene, whereas continuous Pt films induce $p$-type doping, with both effects attenuated upon H adsorption. The developed machine-learned MD framework enables quantum-mechanically accurate modeling of metal crystal growth on graphene, elucidates the underlying H$_2$-sensing mechanism, and correlates several key sensing figures of merit with metal loading and morphology.