New insights into fragmentation chemistry of dihydropyran (DHP): a multiscale computational study from neutral dissociation to electron-impact mass spectrometry.

In this work, we present a multiscale computational investigation of the fragmentation chemistry of 3,4-dihydro-2H-pyran (DHP) by combining complementary quantum chemical calculations, ab initio molecular dynamics, potential energy surface analysis, QCxMS simulations, and machine-learning methods. First, the intrinsic fragmentation pathways of neutral DHP are investigated. Electronic structure calculations characterize the molecular properties, while molecular dynamics simulations reveal the dominant fragmentation channels and their dependence on internal excitation energy. The underlying reaction mechanisms are further elucidated by automated exploration of the potential energy surface identifying key intermediates, transition states, and activation barriers. In the second part, electron-ionization (EI) mass spectra are predicted using QCxMS and three machine-learning models (NEIMS, RASSP, and a developed by us HYBRID GNN-ResNet architecture) and compared with available experimental data. The results highlight the complementary roles of physics-based and data-driven approaches in describing different aspects of DHP fragmentation. The neutral-state molecular dynamics and potential-energy-surface calculations provide mechanistic insight into intrinsic dissociation pathways, whereas QCxMS and machine-learning models enable direct comparison with the experimental EI mass spectrum. Together, these approaches provide a multiscale description of DHP fragmentation while also defining the limits of correspondence between neutral-state dynamics and electron-impact fragmentation.