Decoding Regioselectivity in Cu(I)-Catalyzed Borylation of Al-kynes: Insights from Machine Learning and Artificial Intelligence

Vinyl boronates are highly valuable intermediates in chemical synthesis, extensively used in C‒C bond-forming reactions such as catalytic cross-coupling. Transition metal-catalyzed hydroboration of alkynes has emerged as a key method for synthesizing these building blocks. While classical approaches rely on noble metals like rhodium and iridium, copper-catalyzed hydroboration offers a sustainable and cost-effective alternative. This strategy utilizes bench-stable reagents under mild conditions, delivering highly stereoselective trans-vinylboronates. However, predicting regioselectivity re-mains a challenge due to the complex interplay of ligand structures, alkyne substitution patterns, and reaction conditions. To address this, we employed a combination of experimental data, high-throughput computational calculations, and ma-chine learning (ML) to develop predictive models for regioselectivity. Ligand and catalyst descriptors were derived from DFT calculations and literature databases, forming a robust dataset used to train ML algorithms. Further optimization proved effective in guiding experimental efforts by identifying promising ligands and improving hydroboration yields. This workflow integrates experimental and computational tools to achieve a stereocontrolled synthesis of substituted alkenyl boronates from alkynes. As a case study, we demonstrate the successful application of ML-guided optimization, reducing copper catalyst loading while improving yields and regioselectivity.