Key points are not available for this paper at this time.
The process of discovering and developing new materials currently requires considerable effort, time, and expense. Machine learning (ML) algorithms can potentially provide quick and accurate methods for screening new materials. In the present work, the features of the metal organic frameworks (MOFs) as a catalyst for fixing carbon dioxide into cyclic carbonate were extracted to build a data set, which were collected from the experimental results of approximately 100 published papers. Classifiers were trained with the data set with various ML algorithms, including support vector machine (SVM), K-nearest neighbor classification (KNN), decision trees (DT), stochastic gradient descent (SGD), and neural networks (NN), to predict the catalytic performance. The ML models were trained on 80% of the data set and then tested on the remaining 20% to predict the carbon dioxide fixation ability. The trained ML model was extended to explore 1311 hypothetical MOFs, and some structures displayed a strong catalytic ability. Finally, the six best metal ions (Mn, V, Cu, Ni, Zr and Y) and four best ligands (tactmb, tdcbpp, TCPP, H3L) were determined. These six metals and four ligands could be combined into 24 MOFs, which are strongly potential catalysts for carbon dioxide fixation. Using machine learning methods can speed up the screening of materials, and this methodology is promising for application not only to MOFs as catalysts but also in many other materials science projects.
Li et al. (Tue,) studied this question.