Machine Learning for Accelerating Computational Study of Perovskite Alloy Materials

Researchers accelerate the computational study of perovskite alloys using machine learning

Researchers from the CEST group published a paper demonstrating that machine learning can be used to identify the best perovskite materials for solar cells. Perovskite cells are a new technology that is gaining a lot attention due to its high efficiency. They also have the potential to drastically reduce manufacturing costs compared to traditional silicon-based cells.

The rapid degradation of perovskite cells under extreme environmental conditions, like heat and humidity, has hampered their commercialization. These solar cells also contain toxic materials that are harmful to the environment. It is a continuous process to find new perovskite material that does not suffer from these issues. However, the existing experimental and computational methods are not able to test the large number of materials candidates.

Jarno Laakso, Patrick Rinke and collaborators from the University of Turku in Finland and China developed a new machine-learning-based method for predicting perovskite property quickly. This new method accelerates computations, and it can be used for studying perovskite alloys. The alloys contain many potential candidates for solar cell materials. However, conventional computational methods have made it difficult to study them.