Computational prediction of substrate scope of a homogeneous catalyst: the case of metal-free C-H borylation by a frustrated Lewis pair catalyst.

Predicting substrate scope in catalytic reactions from first principles requires evaluating not only the lowest-energy productive pathway, but also competing processes that can suppress turnover. Here we present FRUST, an automated computational workflow for substrate-scope prediction, and apply it to metal-free C-H borylation mediated by the boron-based frustrated Lewis pair catalyst developed by Fontaine and co-workers (Science 2015, 349, 513-516). FRUST generates intermediates and transition-state guesses for each unique aromatic C-H position, performs conformational sampling, and estimates barriers using constrained xTB geometries refined by DFT single-point calculations. Benchmarking against literature substrates shows that the workflow captures the dominant mechanistic trends of the FLP catalytic cycle and identifies TS1 as the principal barrier-defining step across both benchmark and screening sets. Application to 69 aromatic substrates identified a small number of low-barrier candidates, including phenols, N-aryl amides, and selected aprotic arenes. However, targeted DFT validation showed that even the most promising aprotic candidates lie near or beyond the empirical reactivity window, suggesting that the current FLP catalyst is unlikely to exhibit broad substrate scope beyond known activated heteroarenes. Analysis of protic substrates further revealed important off-cycle limitations: O-H borylation of phenols is predicted to deactivate neighbouring C-H sites, while N-aryl amides are likely to undergo N-H borylation without subsequent productive C-H borylation. These results demonstrate that efficient computational screening can retain mechanistic resolution, but also highlight the need to include side-reaction prediction when assessing catalyst scope and designing next-generation metal-free borylation catalysts.