Dataset for "Closed-Loop Discovery of H Spillover-Driven CO2 Reduction in Acidic Electrolyte by Phosphine-Modified Cu"

Electrochemical CO2 reduction (CO2RR) in acidic media provides high carbon efficiency but suffers from strong competition with hydrogen evolution. Using a closed-loop discovery platform that integrates autonomous experimentation with active learning, we systematically explored phosphine-modified Cu catalysts and identified 1,1,1-tris(diphenylphosphinomethyl)ethane–Cu with 79% C2+ Faradaic efficiency and a highest partial current density of 319 mA cm−2. Beyond catalyst optimization, isotope-labeling and depth-profiling studies revealed a striking mechanistic feature: C2 products originate predominantly from gas-phase CO2 reacting inside the dry porous Cu layer, where *H species, generated at the water/Cu interface, spill over to hydrogenate accumulated *CO. This contrasts with the prevailing view that multi-carbon products mainly form at the wetted water–Cu–gas triphasic interface. By decoupling *H generation from CO2 supply, the optimized catalyst suppresses hydrogen evolution and promotes C–C coupling under acidic conditions. Our study thus demonstrates how closed-loop catalyst discovery not only delivers high-performing systems but also leads to uncovering overlooked mechanistic pathways in electrochemical CO2 conversion.