Interpretable Machine Learning of Two‐Photon Absorption

Abstract

Molecules with strong two‐photon absorption (TPA) are important in many advanced applications such as upconverted laser and photodynamic therapy, but their design is hampered by the high cost of experimental screening and accurate quantum chemical (QC) calculations. Here a systematic study is performed by collecting an experimental TPA database with ≈900 molecules, analyzing with interpretable machine learning (ML) the key molecular features explaining TPA magnitudes, and building a fast ML model for predictions. The ML model has prediction errors of similar magnitude compared to experimental and affordable QC methods errors and has the potential for high‐throughput screening as additionally validated with the new experimental measurements. ML feature analysis is generally consistent with common beliefs which is quantified and rectified. The most important feature is conjugation length followed by features reflecting the effects of donor and acceptor substitution and coplanarity.

Absolute Stimulated Raman Cross Sections of Molecules

“…52 A most recent report on machinelearning-assisted prediction of TPA cross sections summarized the cross section over 900 molecules, most of which rest in the range of 0−5000 GM. 53 Table S3 summarizes the reported σ TPA .…”

“…ISD and squaraine dyes are designed to exhibit large value close to 10 000–76 000 GM, partially due to electronic resonant enhancement . A most recent report on machine-learning-assisted prediction of TPA cross sections summarized the cross section over 900 molecules, most of which rest in the range of 0–5000 GM Table S3 summarizes the reported σ TPA .…”

Uncovering the influence of the modifier redox potential on CO2 reduction through combined data-driven machine learning and hypothesis-driven experimentation

“…The contribution of the functional group descriptors to DFE-C 2+ was evaluated using Shapley values. 18,19 These values were calculated for each descriptor across various models and data points using the SHAP python package. Shapley values, borrowed from game theory, measure the impact of a given descriptor on a target.…”

MLatom 3: A Platform for Machine Learning-Enhanced Computational Chemistry Simulations and Workflows

“…MLatom 2 could perform training of the ML models, evaluate their accuracy, and then use the models for geometry optimization and frequency calculations. Special workflows were also implemented, such as acceleration of the absorption UV/vis spectra calculations with ML and prediction of two-photon absorption spectra . In addition, MLatom 2 could be used to perform simulations with the general-purpose AI-enhanced QM method AIQM1 and universal machine learning potentials of the ANI family ,− with the accurate scheme developed for calculating heats of formation with uncertainty quantification with these methods.…”