From orbital analysis to active learning: an integrated strategy for the accelerated design of TADF emitters

TL;DR

This paper develops an integrated computational strategy combining orbital analysis and active learning to efficiently design TADF emitters, achieving high accuracy with reduced data and demonstrating applications in bioimaging, photocatalysis, and photodetection.

Contribution

It introduces a semi-empirical protocol and active learning approach for predicting TADF properties, significantly reducing data requirements and enabling practical applications.

Findings

Support Vector Regression model achieves MAE = 0.024 eV and R2 = 0.96.

Active learning reduces data needs by about 25%.

Charge-transfer descriptors are key predictors for TADF properties.

Abstract

Thermally Activated Delayed Fluorescence (TADF) emitters must satisfy two competing requirements: small singlet-triplet energy gaps for thermal upconversion and sufficient spin-orbit coupling for fast reverse intersystem crossing. Predicting these properties accurately demands expensive calculations. We address this using a validated semi-empirical protocol (GFN2-xTB geometries, sTDA/sTD-DFT-xTB excited states) on 747 molecules, combined with charge-transfer descriptors from Natural Transition Orbital analysis. The hole-electron spatial overlap She emerges as a key predictor, accounting for 21% of feature importance for the triplet state alone. Our best model (Support Vector Regression) reaches MAE = 0.024 eV and R2 = 0.96 for $\Delta E_{ST}$. Active learning reduces the data needed to reach target accuracy by approximately 25% compared to random sampling. Three application domains are explored: NIR-emitting probes for bioimaging, photocatalytic sensitizers, and fast-response materials for photodetection.