The team's robot performs a microwave conductivity measurement. (Image Credit: Akinori Saeki)
Osaka University researchers developed an AI-driven robotic measurement system that identifies optimal solar cell materials by performing photoabsorption spectroscopy, optical microscopy, and time-resolved microwave conductivity analyses. The robot then assessed 576 various thin-film semiconductor samples.
Materials other than silicon could prove more effective for solar cells. However, the materials need to be low in toxicity, extremely efficient, and made of common chemical elements. So far, there aren't many like those out there, and studies using new materials are often expensive, time-consuming, and done by hand.
"Current solar cells are made of inorganic semiconductors containing silicon and gallium, but next-generation solar cells need to reduce both cost and weight," says lead author Chisato Nishikawa. "Safety is also a concern; perovskite solar cells are efficient enough to rival silicon solar cells, but they contain toxic lead."
Illustration of the robotic measurement system. (Image Credit: JACS Au 2023, 3, 11, 3194-3203)
All samples are made from varying mixtures of tin, cesium, bismuth, and iodine. These were also treated with different organic salt additives and annealed at varying temperatures. The team used machine learning to examine the data to identify the materials' properties and automate the experiments.
"In recent years, machine learning has been very helpful in better understanding the properties of materials. These studies require vast amounts of experimental data, and combining automated experiments with machine-learning techniques is an ideal solution," says senior author Akinori Saeki.
The researchers want to improve the system's automation processing capabilities so that it's easier to identify new materials. Nishikawa says, "This method is ideal for exploring areas where there's no existing data." The robotic system has even shown some potential. With the fully automated and extremely accurate measurement process, the required work can be done in one-sixth of the normal time.
Have a story tip? Message me at: http://twitter.com/Cabe_Atwell