The active sites for CO₂ electroreduction (CO₂R) to multi-carbon (C₂₊) products over oxide-derived copper (OD-Cu) catalysts are under long-term intense debate. This paper describes the atomic structure motifs for product-specific active sites on OD-Cu catalysts in CO₂R. Herein, we describe realistic OD-Cu surface models by simulating the oxide-derived process via the molecular dynamic simulation with neural network (NN) potential. After the analysis of over 150 surface sites through NN potential based high-throughput testing, coupled with density functional theory calculations, three square-like sites for C–C coupling are identified. Among them, Σ3 grain boundary like planar-square sites and convex-square sites are responsible for ethylene production while step-square sites, i.e. n(111) × (100), favor alcohols generation, due to the geometric effect for stabilizing acetaldehyde intermediates and destabilizing Cu–O interactions, which are quantitatively demonstrated by combined theoretical and experimental results. This finding provides fundamental insights into the origin of activity and selectivity over Cu-based catalysts and illustrates the value of our research framework in identifying active sites for complex heterogeneous catalysts.

长期以来，关于在氧化物衍生的铜（OD-Cu）催化剂上将CO ₂电还原（CO ₂ R）转化为多碳（C ₂₊）产物的活性位点一直在争论不休。本文介绍了CO _2中OD-Cu催化剂上产物特定活性位的原子结构基序R.在此，我们通过具有神经网络（NN）电位的分子动力学模拟通过模拟氧化物衍生的过程来描述实际的OD-Cu表面模型。通过基于NN电位的高通量测试分析了150多个表面位点，再结合密度泛函理论计算，确定了三个正方形的C–C耦合位点。其中，Σ3晶界像平面正方形位和凸正方形位都负责乙烯的生产，而阶梯正方形位点即n（111）×（100），由于稳定​​乙醛中间体和破坏Cu-O相互作用的几何效应，有利于醇的生成，这是理论和实验结果相结合的定量结果。这一发现提供了对基于铜的催化剂的活性和选择性的起源的基本见解，并说明了我们的研究框架在识别复杂多相催化剂的活性位点方面的价值。