Npj Comput. Mater.: 数据驱动—助益复杂材料全局优化和科学分析

有机-无机杂化卤化物钙钛矿材料具有卓越的光电性能，并在太阳能电池，光电探测器，发光二极管、闪烁体和光电池中得到了一定的应用。然而，与硅或石墨烯不同，卤化物钙钛矿在日常环境条件下非常不稳定。例如，卤化物钙钛矿材料在氧气和水蒸气等大气分子的存在下会迅速降解，从而削弱器件和设备性能。了解和优化其水环境中光电化学性质对于开发高效和可持续的能量转换和存储技术(包括太阳能电池、水分解系统、光催化剂和环境传感器)至关重要。考虑到分子和表面工程卤化物钙钛矿的高度复杂和多维虚拟设计空间，需要进行更全面和系统的研究，以获得最佳的分子/钙钛矿复合体系，该体系在水溶液等恶劣条件下具有良好的光电性能和水稳定性。

来自中国南京信息工程大学的张磊教授及其团队成员，基于机器学习和材料计算，科学地评估了卤化物钙钛矿的稳定性。在这项研究中，他们系统地研究了在不同表面上功能的多分子修饰的CH3NH3PbI3膜的水性光电化学稳定性，发现了一种有效的多分子钙钛矿材料体系“calcein + PbBr2+ DMSO +CH3NH3PbI3”，具有出色的液态水环境光电稳定性（相同水环境下光生电流是CH3NH3PbI3薄膜输出的103倍)和92.5%的水稳定性。随后，通过遗传算法和极端随机树的Shapley分析来检测分子修饰钙钛矿材料的实验水性光电化学性质，以提供机器解释和解耦分子贡献，强调这些兼容分子的协同效应及其亲水/亲脂性对目标输出的重要性。DFT计算表明，该多分子全局优化体系存在大量氢键和阴离子··π表面相互作用以稳定其界面结构。除了预测高水稳定性的钙钛矿体系外，该研究团队还优化了光电化学、机器学习和DFT模型总体数据驱动工作流程用于评估卤化物钙钛矿稳定性。

该文近期发表于npjComputational Materials| (2024) 10:114，英文标题与摘要如下，点击https://www.nature.com/articles/s41524-024-01297-4可以自由获取论文PDF。

Figure 1 Post-hoc DFT calculation (atomic and electronic structures)

Figure 2 Fabrication details, classification method and molecular variables.

Figure 3 Feature analysis via SHAP.

Figure 4 Overall workflow of this study.

Data-driven optimization and machine learning analysis of compatible molecules for halide perovskite material

Shaojun Wang, Yiru Huang, Wenguang Hu & Lei Zhang

Optoelectronic stability of halide perovskite material in hostile conditions such as water is rather limited, preventing them from further industrial deployment. Here, we optimize and perform machine learning analysis on CH3NH3PbI3materials with additives, solvents and post-treatment molecules using combined experimental and data-driven methods. A champion system consisting of a compatible tertiary molecular combination ‘calcein+PbBr2+ DMSO’ active at diverse surfaces is identified, delivering a large aqueous photoelectrochemical (PEC) photocurrent of 10-5A/cm2and an improved aqueous stability of 92.5%. Subsequently, machine interpretation is provided to decouple the multi-molecule contributions with the assistance of genetic programming (GP) and extra-trees (ET) machine learning models, highlighting the intricate molecular features for the target outputs. The posthoc density functional theory (DFT) calculation suggests the presence of multiple hydrogen bond and anion··π surface interactions to stabilize the interfacial structures. The present ‘PEC + GP + ET + DFT’ approach is suggested to be an effective approach to design and comprehensively evaluate moleculemodified materials.