As part of a research project at the Swiss Federal Institute of Technology in Lausanne(EPFL), a method has been developed to search large databases for potential materials for the utilisation of new solar cells. According to a press release, several promising halide perovskites were identified through the use of machine learning (ML). ML is a branch of artificial intelligence in which computers learn from data sets or databases made available to them.
Perovskites represent a promising new group of materials for future photovoltaic applications due to their simple manufacturing processes combined with low costs. It is important that the new materials have a suitable band gap so that they can utilise solar energy optimally. The band gap has the property of absorbing photons with a certain energy and then converting them into electricity.
The EPFL team led by Haiyuan Wang and Alfredo Pasquarello developed a machine learning model that was able to identify 14 completely new perovskites from 15,000 materials. These are excellent candidates for future high-efficiency solar cells. The researchers were thus able to show that the use of ML can significantly accelerate the discovery and validation of new photovoltaic materials.







