Automatized Platform for Bandgap Optimization in Diarylethene Derivatives via Structural Substitutions

This study presents an automated framework for discovering Diarylethene derivatives with specific HOMO-LUMO bandgaps by integrating first-principles calculations with evolutionary optimization algorithms. We develop a protocol for performing automatic, parallel Density Functional Theory calculations using Turbomole, enabled by the Atomic Simulation Environment Python wrapper. This integration allows for fully autonomous evolutionary searches, removing the need for human intervention. To evaluate our platform, we conduct searches within the chemical space of Diarylethene using two optimization strategies: A “local” strategy, which explores local neighborhoods, and a “glocal” strategy, which combines local exploration with memory of past achievements.