Machine learning assisted study of the transition metal phosphides catalyst for the water electrolysis to produce hydrogen

In recent years, hydrogen production by water electrolysis has become an important strategy of energy transformation in China, where the design of efficient catalysts for hydrogen evolution reaction (HER) is a key issue. In this regard, transition metal phosphides (TMPs) are considered important non-precious metal catalysts in water electrolysis owing to their low price and high hydrogen production efficiency. However, experimental screening of highly active TMPs catalysts is time-consuming and challenging. This study provides a simple and effective method for rapidly screening highly efficient HER electrocatalysts based on machine learning and big-data analysis. Four machine learning algorithms, namely support vector regression (SVR), K-nearest neighbour (KNN), random forest regression (RF) and extreme gradient boosting (XGBoost), were developed to predict the catalytic performance of various transition metal phosphides reported in the literature in HER. After evaluating the four algorithms by RMSE and R², it was found that the RF algorithm has excellent prediction ability for overpotential, while the XGBoost algorithm predicts better for the Tafel slope. It is concluded that the contents of Ni, Co and Fe have a significant influence on the catalytic performance and highly active catalysts may be prepared by fine adjustment of their contents in the future.