OS100: A Benchmark Set of 100 Digitized UV-Visible Spectra and Derived Experimental Oscillator Strengths

Abstract

Excited-state quantum chemical calculations typically report excitation energies and oscillator strengths, ƒ, for each electronic transition. On the other hand, UV-visible spectrophotometric experiments report energy-dependent molar extinction/attenuation coefficients, ε(v), that determine the absorption band line shapes. ε(v) and ƒ are related, but this relation is complicated by various broadening and solvation effects. We fit and integrated experimental UV-visible spectra to obtain ƒexp values for absorption bands and estimated the uncertainty in the fitting. We derived 164 ƒexp values from 100 organic molecules ranging in size from 6-34 atoms. The corresponding computed oscillator strengths (ƒcomp) were obtained with time-dependent density functional theory and a polarizable continuum solvent model. By expressing experimental and computed absorption strengths using a common quantity, we directly compared ƒcomp and ƒexp. While ƒcomp and ƒexp are well correlated (linear regression R2=0. 921), ƒcomp in most cases significantly overestimates ƒexp (regression slope=1.34). The agreement between absolute ƒcomp and ƒexp values was substantially improved by accounting for a solvent refractive index factor, as suggested in some derivations in the literature. The 100 digitized UV-visible spectra are included as plain text files in the supporting information to aid in benchmarking computational or machine-learning approaches that aim to simulate realistic UV-visible absorption spectra.

Absorption Intensities of Organic Molecules from Electronic Structure Calculations versus Experiments: The Effect of Solvation, Method, Basis Set, and Transition Moment Gauge

“…Two comments regarding equation ( 12) must be made. First, it is common to include the (real) refractive index of the medium in the denominator of (12) to account for the effect of the solvent on the electric field "felt" by the solute. 18 15 This correction is expected to make f values obtained in different solvents directly comparable.…”

“…Nonetheless, f values obtained according to equation (12), as a function of the integrated intensity if nothing else, quantify the probability of an electronic transition in a way suitable to compare to theoretical probability computations.…”

“…62,[75][76][77] The calculations above were all performed using Gaussian 16 version C.01. 78 In addition, using the PCM/6-31+G* optimized geometries reported by Tarleton et al, 12 we carried out additional calculations using the 6-31+G* basis set with Q-Chem 5.3. 79 Specifically, we ran the calculation in vacuo and also using two additional solvation models; the conductorlike PCM (CPCM) model and COSMO.…”

“…Recently, we generated a collection of f values from experimental UV-Vis absorption spectra of 100 organic molecules in solution. 12 A total of 164 OSs were obtained by integrating the attenuation coefficient ϵ(ν) over the limit of well-defined bands in the spectra. Transitions were categorized as either very high (VH), high (H), medium (M), low (L), or very low (VL) confidence on the basis of the reproducibility and quality of the fitting.…”

“…While errors in experimental OSs are difficult to quantify, we expect that errors in the condensed phase should be smaller than errors in the gas phase. 12,13 Here, we employ this benchmark set to compare OSs computed using several TD-DFT functionals and wave function methods. Aspects that affect the comparison between theory and experiments, such as experimental deviations from the Beer-Lambert Law, solvent effects, and the f value dependence on the energy of the electric transition, are discussed.…”