Molecular photoswitches based on E/Z isomerization are central to the development of light-responsive materials, molecular machines, and dynamic-covalent systems. While azobenzene derivatives dominate this field, their imine analogues have historically been overlooked due to poor thermal stability, incomplete switching, and reliance on UV excitation. Recent work has demonstrated that aryliminopyrazoles constitute a promising class of imine photoswitches, combining quantitative visible-light switching with long-lived metastable states and high synthetic accessibility. These advances motivate a systematic expansion beyond the limited set of experimentally explored motifs. Here, we present a computational framework for the large-scale discovery of imine-based photoswitches. Building on established structure–property relationships and a validated quantum chemical protocol capable of reliably predicting thermodynamic and photophysical properties, we construct a hypothetical design space comprising over 25 million imines. This space is generated via combinatorial coupling of chemically realistic amine and aldehyde fragments, ensuring synthetic plausibility while enabling extensive chemical diversity. To efficiently navigate this vast space, we employ Bayesian optimization coupled to our quantum chemical workflow. This approach enables the targeted identification of candidate photoswitches that optimize multiple objectives, including high thermal stability of the metastable Z-isomer and selective activation with visible light. Furthermore, we outline strategies for experimental validation and iterative model refinement, enabling feedback between computation and synthesis.
Lasse Kreimendahl