Chemputing the synthesis and optimisation of arene fluorination in an automated platform

Positron emission tomography (PET) relies on the rapid and efficient incorporation of fluorine-18 (18F) into molecular imaging agents, yet selecting appropriate radiofluorination conditions for new substrates remains ad hoc. Whilst decades of development have generated a wealth of experimentally validated reactions, this knowledge remains dispersed throughout literature and is difficult to exploit systematically for reaction planning. Here the chemputation-enabled arene radiofluorination, that is optimization and automated execution, captures both reaction transformations and the associated experimental conditions, providing a foundation for reasoning. Three complementary levels of analysis are implemented to interrogate the knowledge base, progressing from molecular similarity and electronic descriptors, through reaction abstraction based on bond–electron matrices, to AI-assisted contextual reasoning. The recommended literature precedents provide experimentally informed starting points that are subsequently refined using an optimization algorithm, enabling efficient exploration of multidimensional reaction-condition space beyond direct literature transfer and the optimized procedure is executed on a robotic platform using the Chemical Description Language (χDL), establishing a continuous digital workflow from literature knowledge to reproducible chemical synthesis. By integrating curated reaction knowledge, computational reasoning, experimental optimization, and digital execution, this work provides a general framework for data-driven radiofluorination that reduces experimental trial-and-error while improving reproducibility, safety, and scalability in PET tracer development.