RMIT Researchers Help Harness AI For Solar Cell Production

What does it take to optimise solar cells? For a team of researchers led by Dr Nastaran Meftahi, harnessing the power of AI has led to the rapid development of perovskite solar cells, a groundbreaking material that has defied the efforts of researchers for decades.

The centrepiece of this research is an innovative new learning model, developed by researchers at RMIT University in conjunction with members of the Centre of Exciton Science. This machine-learning model promises to drive rapid innovation in the chemical production of perovskite solar cells.

What is Perovskite?

Perovskite is a calcium titanium oxide mineral that holds incredible promise as a material that can be used for mainstream solar power production. It was discovered in the Ural Mountains of Russia by renowned German mineralogist Gustav Rose in 1839, a celebrated member of the Rose family of chemists.

As silicon costs have risen, researchers have been looking for alternative materials that could be used as alternatives for high-efficiency solar cells. In the past decade, research has coalesced around the potential for perovskite as a replacement for silicon, which promises reduced manufacturing costs with efficiency gains higher than that of traditional silicon-based solar cells.

Reproducing the Complicated

While concentrated research has enabled researchers to increase the solar conversion rate of perovskite cells in the past decade, there have been challenges that have prevented the use of this silicon alternative in real-world situations.

Perovskite solar cells are less stable than traditional silicon cells, and as a result, researchers around the world have been investigating ways to streamline testing in a repeatable way. The solution, published in the journal Advanced Energy Materials, is rather novel.

By combining the power of machine learning with a perovskite film fabrication technique, Dr Meftahi’s team can leverage the immense power of their machine learning system to rapidly iterate, test and learn from potential perovskite solar cell candidates. Additionally, the system can be automated, allowing for the ongoing testing of numerous candidates, as well as further accelerating the pace of research.

The result is rather elegant — the team at the Centre of Excellence in Exciton Science can rapidly iterate on massive volumes of potential recipes, that are reproducible in other settings, and use those learnings to inform future model development.

Speaking about the research, the team expressed praise for the rapid development of test candidates in this setup. Meftahi noted that this model is only in its infancy, and will be further developed, highlighting that “...we’re quickly getting to the stage where we’ll be able to predict the properties of millions of cells”.

Dr Nastaran Meftahi. Image credit: ARC Centre of Excellence in Exciton Science.

The Future of Photovoltaics

Supported by the Australian Renewable Energy Agency (ARENA), the Australian Research Council (ARC) and the Australian Centre for Advanced Photovoltaics (ACAP), this crucial research in perovskites has the potential to open many doors in the photovoltaic materials testing space.

The next generations of solar cells may be developed with the help of this innovative technology, and there’s the potential for further learning and automation that can be done on other material candidates. The future of photovoltaics is bright, and as Dr Meftahi’s team looks to bring this innovative new model to the world, one must wonder what will happen next in this transforming industry. This latest research is another example of how Australian universities like RMIT and RMIT online courses continue to lead the way in learning and innovation.

Top image credit: iStock.com/audioundwerbung