基于机器学习的钙钛矿锰氧化物材料设计

Materials Design of Perovskite Manganates Based on Machine Learning

ABO_(3)钙钛矿锰氧化物因成本低廉和稳定性好,已成为反铁磁体中最热门的存储器材料。提高ABO_(3)钙钛矿锰氧化物的奈尔温度(Néel temperature,T_(N)),使之在室温下呈现反铁磁性,具有重要的意义。利用超多面体方法对特征变量的重要性进行排序,进而结合机器学习算法来筛选特征变量,并构建了极端梯度回归(XGBoost)机器学习模型,搭建了ABO_(3)钙钛矿锰氧化物的T_(N)在线预报平台。利用高通量筛选找到了T_(N)预测值高于室温的候选材料(Sr_(0.7)Ce_(0.1)Sm_(0.2)MnO_(3),308.5 K),其TN比已知最高的样本还高6.37%。该研究方法有助于实验工作者选择最有希望的材料来做实验,可以加快新材料的研发和性能突破。

ABO_(3) perovskite manganates has become the most popular memory material in anti-ferromagnets due to its low cost and good stability.It is of great significance to improve the Néel temperature(T_(N))of ABO_(3) perovskite manganates to make it antiferromagnetic at room temperature.In this work,hyper-polyhedron method is used to rank the importance of characteristic variables,and the machine learning algorithm is integrated to screen features.The online prediction platform was built for T_(N) of ABO_(3) perovskite manganates.The XGBoost machine learning model was established to screen out the potential material(Sr_(0.7)Ce_(0.1)Sm_(0.2)MnO_(3),308.5 K)with the predicted T_(N) higher than room temperature based on high-throughput screening.The T_(N) of the potential material is 6.37% higher than the highest one known.This research method is helpful for experimental workers to select the most promising materials,which can be used to speed up the research and development of new materials with targeted performances.