Benchmarking electronic-structure methods for the description of dark transitions in carbonyls at and beyond the Franck-Condon point

TL;DR

This paper benchmarks various electronic-structure methods for accurately describing dark electronic transitions in carbonyl compounds, emphasizing their performance at and beyond the Franck-Condon point and their impact on photolysis predictions.

Contribution

It provides a comprehensive comparison of electronic-structure methods for dark transitions in carbonyls, including beyond Franck-Condon effects and photolysis implications.

Findings

ADC(2) and CC2 perform well at equilibrium geometry.

Methods vary significantly in predicting transition intensities beyond Franck-Condon point.

Photolysis half-life predictions are sensitive to the choice of electronic-structure method.

Abstract

Herein, we propose a comprehensive benchmark of electronic-structure methods to describe dark transitions, that is, transitions to excited electronic states characterized by a near-zero oscillator strength. This type of electronic state is particularly important for the photochemistry of molecules containing carbonyl groups, such as atmospheric volatile organic compounds (VOCs). The oscillator strength characterizing a dark transition can change dramatically by a slight alteration of the molecular geometry around its ground-state equilibrium, the so-called non-Condon effects. Hence, testing the performance of electronic-structure methods for dark transitions requires considering molecules at their Franck-Condon point (i.e., equilibrium geometry), but also beyond the Franck-Condon point. Our benchmark focuses on various electronic-structure methods - LR-TDDFT(/TDA), ADC(2), CC2, EOM-CCSD, CC2/3, XMS-CASPT2 - with CC3/aug-cc-pVTZ serving as a theoretical best estimate. These techniques are tested against a set of 16 carbonyl-containing VOCs at their equilibrium geometry. We then assess the performance of these methods to describe the dark transition of acetaldehyde beyond its Franck-Condon point by (i) distorting the molecule towards its S$_1$ minimum energy structure and (ii) sampling an approximate ground-state quantum distribution for the molecule and calculating photoabsorption cross-sections within the nuclear ensemble approach. Based on the calculated cross-sections, we calculate the photolysis half-life as depicted by the different electronic-structure methods - highlighting the impact of the different electronic-structure methods on predicted experimental photolysis observables.