A data-driven modeling study on the accurate identification of Doppler-free saturated absorption spectra in diatomic...

The accurate identification of electronic transitions in high resolution molecular spectra is central to elucidating the complex energy level structures of excited states and characterizing intramolecular interactions in diatomic molecules. Here, we introduce a data-driven model that integrates the Dunham model with grid search iterative optimization and stepwise predictive matching methods to simultaneously characterize multiple coupled electronic excited states. Using Doppler-free saturated absorption spectra of molecular tellurium (130Te2) as a benchmark, our strategies enable the identification of 3701 lines across B0u+ ← X0g+, A0u+ ← X0g+, B1u- ← X1g-, and B1u+ ← X1g+ transitions with an accuracy on the order of MHz (∼10-4 cm-1) and extend the reliable assignment of rotational quantum numbers (J) up to ∼200. The diatomic constants and potential energy curves for both A0u+ and B0u+ of 130Te2 were updated, and the energy-level perturbations between B0u+ and B1u+ were also exhibited. The present data-driven framework establishes a universal paradigm for high resolution spectroscopic analysis of homonuclear diatomic molecules, offering a robust approach for analyzing highly congested spectra and enabling reliable spectral assignment.