Artificial intelligence amps up perovskite development for photovoltaics

The exceptional electrical and optical properties of hybrid organic-inorganic perovskites (HOIPs) make the materials desirable for optoelectronic applications such as photovoltaics and light-emitting diodes. Since traditional methods for modeling HOIPs are time-consuming, recent studies have used machine learning to accelerate research in fabrication, characterization, device testing, and data analysis.

However, some perovskite experts may not have the requisite knowledge to perform machine learning. Hering et al. describe how machine learning can be used to accelerate the development of HOIP photovoltaics for researchers with a broad range of machine learning knowledge.

The authors discuss the prospects and challenges associated with using machine learning methods to accelerate the development of HOIP photovoltaics. They also summarize how machine learning algorithms have been successfully implemented at different stages of solar cell development, including composition screening, material fabrication, characterization, and full device testing.

“This article can complement the comprehension of scientists that have thus far focused on trial-and-error methods to fabricate and characterize perovskite solar cells,” co-author Marina Leite said, adding that experimental methods can be wearying and time-consuming due to the vast number of testable chemical compositions. “We hope to stimulate the community to embrace a new era where a ‘dataquake’ related to solar cells materials and devices is being generated and must be analyzed in a timely and informative manner.”

The team hopes their work can be used as a guideline for determining which type of machine learning should be used in future research, including on pressing questions such as determining the most stable HOIPs and their optimal applications.

Source: “Emerging opportunities for hybrid perovskite solar cells using machine learning,” by Abigail R. Hering, Mansha Dubey, and Marina S. Leite, APL Energy (2023). The article can be accessed at https://doi.org/10.1063/5.0146828 .