Researchers at The University of Osaka have built a robotic system that watches solar cell materials form in real time. The platform uses machine learning to find what affects the quality of the final product.
The team used it to make a perovskite solar cell that reached a power conversion efficiency of 24.4%. The findings were published in the journal ACS Nano.
Why Perovskite Films Are Hard to Make
Perovskite is a material that absorbs sunlight and turns it into electricity. In solar cells, it is used as a thin layer.
That layer starts as a liquid solution. The liquid dries and crystallizes into a solid film.
Small changes during this process can cause problems. Shifts in temperature, humidity, and other conditions can affect the film and how well the solar cell works.
The researchers wanted a better way to study this step. Their new platform links film deposition with optical imaging and environmental measurements.
“We want to move beyond trial and error by measuring what happens to the film while it is actually forming,” said lead author Ryosuke Kado. He said the system lets the team connect real-time changes to film quality and find the factors that matter most.
How the Robotic System Works
The team focused on how the liquid coating turns into a crystal film. They tracked changes in the film’s color and the behavior of antisolvent droplets.
These changes happen very fast. The system can record color changes in less than a second.
It can also track the motion and impact of antisolvent droplets within a few milliseconds. This let the researchers study how droplet timing and behavior shape the film that forms next.
The platform also records environmental and processing conditions during each run. Automated tools then measure the quality of the finished film.
“The advantage of automation is that we can collect much richer information than would be practical by hand,” said senior author Akinori Saeki.
Saeki said machine learning helps the team make sense of the many variables. It can find links that are hard to see otherwise.
The researchers used these results to improve the antisolvent coating process. Their analysis found an ideal range for how the droplets should behave.
That range guided the making of a lead-based perovskite solar cell. Under the improved conditions, the device reached a power conversion efficiency of 24.4%.
The team said the result shows that data collected while a film forms can be used to improve a real solar cell. The researchers said they hope the technology will speed up photovoltaic research and lead to more practical solar cells.
