Computation of Nonradiative Rates for High-Throughput Virtual Screening: Application to the Discovery of Potential TADF Molecules

Abstract

High-throughput virtual screening of molecules with suitable electronic properties rarely includes the computation of photophysical rate constants as target quantities. In this work, we develop a fully automated screening workflow capable of computing reverse intersystem crossing (RISC) rates for a large set of systems. The workflow is specifically designed to identify promising thermally activated delayed fluorescence (TADF) emitters and is applied to a medium-sized data set of 70 molecules. These candidates were selected from a database of 150,000 conjugated organic molecules not specifically tailored for TADF, based on desirable excited-state properties at the ground-state geometry (vertical singlet–triplet energy gap and oscillator strength). RISC rates were computed using a fully quantum-mechanical approach based on Fermi’s golden rule, which explicitly accounts for all nuclear degrees of freedom. The workflow autonomously performs the initial screening steps by excluding molecules that fail the required geometry optimization and frequency calculations and by diagnosing cases that cannot be reliably treated within the proposed framework, thereby ensuring robustness and reliability. Our results show that including RISC rate calculations drastically reduces the number of viable TADF candidates. This reduction arises from the combined effect of the key parameters governing the rate, mainly the adiabatic singlet–triplet energy gap and the reorganization energy.