A machine-learning-driven dataset of 144,111 polycyclic aromatic hydrocarbon infrared spectra for interpreting aromatic infrared bands

Abstract

The interpretation of interstellar aromatic infrared bands (AIBs) relies on reference spectral libraries, but the chemical space covered by existing databases remains incomplete relative to the vast structural diversity of polycyclic aromatic hydrocarbons (PAHs) expected in the interstellar medium. We present a large-scale, high-throughput PAH infrared (IR) spectral database that complements existing reference libraries by providing significantly more continuous coverage of the PAH chemical space for small to medium-sized PAHs. We used a machine-learning double-harmonic (ML-DH) framework that integrates neural network force fields with electronic dipole models. This framework achieves quantum-chemical accuracy at a fraction of the computational cost. It has been validated against independent density functional theory calculations and benchmark spectral databases. The resulting dataset comprises 144,111 IR spectra for 48,037 closed-shell, even-carbon benzenoid PAH structures across neutral, cationic, and anionic charge states, complete with per-spectrum uncertainty estimates. The spectral library, optimized geometries, and model code are publicly available. They can be used to complement existing libraries for interpreting and modeling JWST AIB observations and those from future IR missions.