Australian researchers at RMIT University’s School of Science have harnessed artificial intelligence (AI) to accelerate the design and production of solar cells.
Dr Nastaran Meftahi, the lead author of the study, highlighted the global race to produce perovskite cells, emphasising its cost-effectiveness compared to traditional silicon cells and recent stability advancements, transforming the traditionally challenging and alchemical process into a rapid and reproducible method.
The team, consisting of researchers from RMIT, Monash University, and CSIRO, successfully removed human error from the equation by integrating AI into the innovation process, creating a new machine learning model capable of predicting promising chemical recipes for perovskite solar cells.
Dr. Adam Surmiak from Monash University is currently constructing a multimillion-dollar automated system for solar cell manufacturing, set to work in conjunction with the developed machine learning model, capable of predicting large volumes of chemical recipes for new perovskite solar cells.
Published in the journal Advanced Energy Materials, the combined efforts of the research team resulted in the production of reproducible perovskite solar cells with a power-conversion efficiency of 16.9 per cent, marking a significant achievement as the best-known result manufactured without human intervention.
The team’s machine learning model represents a starting point for further optimisation in terms of both power-conversion efficiency and stability, addressing the reproducibility challenge faced by human-led and other reported AI-driven perovskite cell design processes.
Dr. Meftahi outlined the potential for the machine-learning model and automated system to be applied to other types of solar cells, including those made with silicon or organic materials, expressing a keen interest in collaborating with industry partners for further testing and potential commercialisation.
For more renewable and solar news, subscribe to ecogeneration.