Department of Chemistry, College of Sciences, Shanghai University, Shanghai 200444, China.
Materials Genome Institute, Shanghai University, Shanghai 200444, China.
J Phys Chem Lett. 2021 Aug 12;12(31):7423-7430. doi: 10.1021/acs.jpclett.1c01939. Epub 2021 Jul 31.
Predicting the formability of perovskite structure for hybrid organic-inorganic perovskites (HOIPs) is a prominent challenge in the search for the required materials from a huge search space. Here, we propose an interpretable strategy combining machine learning with a shapley additive explanations (SHAP) approach to accelerate the discovery of potential HOIPs. According to the prediction of the best classification model, top-198 nontoxic candidates with a probability of formability () of >0.99 are screened from 18560 virtual samples. The SHAP analysis reveals that the radius and lattice constant of the B site ( and LC) are positively related to formability, while the ionic radius of the A site (), the tolerant factor (), and the first ionization energy of the B site () have negative relations. The significant finding is that stricter ranges of (0.84-1.12) and improved tolerant factor τ (critical value of 6.20) do exist for HOIPs, which are different from inorganic perovskites, providing a simple and fast assessment in the design of materials with an HOIP structure.
预测钙钛矿结构的混合有机-无机钙钛矿（HOIPs）的可成形性是从巨大的搜索空间中寻找所需材料的一个突出挑战。在这里，我们提出了一种结合机器学习和 Shapley 加法解释（SHAP）方法的可解释策略，以加速潜在 HOIPs 的发现。根据最佳分类模型的预测，从 18560 个虚拟样本中筛选出 198 种非毒性候选物，其成形概率（）>0.99。SHAP 分析表明，B 位（和 LC）的半径和晶格常数与可成形性呈正相关，而 A 位（）的离子半径、容忍因子（）和 B 位的第一电离能（）则呈负相关。一个重要的发现是，HOIPs 存在更严格的范围（0.84-1.12）和改进的容忍因子τ（6.20 的临界值），这与无机钙钛矿不同，为 HOIP 结构材料的设计提供了一种简单快速的评估方法。
