Electrosynthesis has become an increasingly popular platform in modern organic chemistry, possessing distinctive features and reaction parameters like applied current/potential, electrodes, electrolyte systems, and cell design. While these unique features give chemists more opportunities to control reactivity and selectivity, they also increase the dimensionalities of a reaction and complicate the interactions between variables, making the optimization more challenging. Herein, we present a machine learning (ML) workflow that leverages physical organic descriptor-based yield prediction and orthogonal experimental design to strike a delicate balance between the need for sampling diversity and the pursuit of yield improvements, thereby efficiently identifying ideal conditions for enantioselective palladaelectro-catalyzed annulation from extensive synthetic space. This work shows the potential of synergizing organic electrochemistry and a data-driven approach to tackle multidimensional chemical optimization problems.

电合成已成为现代有机化学中越来越流行的平台，具有独特的特征和反应参数，如施加电流/电位、电极、电解质系统和电池设计。虽然这些独特的功能为化学家提供了更多控制反应性和选择性的机会，但它们也增加了反应的维度并使变量之间的相互作用变得复杂，使优化更具挑战性。在此，我们提出了一种机器学习（ML）工作流程，利用基于物理有机描述符的产量预测和正交实验设计，在采样多样性的需求和追求产量提高之间取得微妙的平衡，从而有效地确定对映选择性钯电的理想条件-从广泛的合成空间催化成环。这项工作展示了有机电化学和数据驱动方法协同解决多维化学优化问题的潜力。