Photothermal Reaction Mechanisms in B-C-H Systems Using Quantum Dynamics Simulations Towards ML Based Simulations

We investigate the thermal decomposition and pyrolysis mechanisms of a boron–carbon–hydrogen cluster (B₁₀C₈H₂₆) using quantum molecular dynamics (QMD) simulations. The system, consisting of a boron cage linked to an eight-carbon chain, is initially equilibrated at a density of 0.466 g cm⁻³ in a canonical (NVT) ensemble and gradually heated from 300 K to 900 K to allow structural relaxation and intramolecular rearrangement. Subsequently, an isothermal–isobaric (NPT) ensemble with a Parrinello–Rahman barostat is employed to increase the density to 1.0 g cm⁻³, enabling structural compaction while maintaining chemical connectivity. Pyrolysis is initiated by rapid heating to 2500–3000 K, leading to extensive bond cleavage and the evolution of volatile species. The primary decomposition products are hydrogen (H₂), methane (CH₄), ethene (C₂H₄), and propane (C₃H₈), with hydrogen being the dominant species.

Our QMD results provide atomistic insight into the early stages of boron-carbide precursor decomposition and the electronic processes governing hydrogen release. Building on this dataset, we plan to develop large-scale machine-learning-based molecular dynamics (ML-MD) simulations to extend the accessible length and time scales of B–C–H pyrolysis. The QMD trajectories will serve as training data for equivariant neural-network potentials capable of reproducing quantum accuracy at classical computational cost. Such ML-accelerated simulations will enable exploration of cluster aggregation, long-range bond-exchange mechanisms, and ceramic-phase evolution, thereby bridging quantum chemistry and mesoscale modeling in boron-rich materials.