The approach, from the Massachusetts Institute of Technology, is expected to bypass the time-consuming steps currently needed to develop and to test new photovoltaic materials. The approach has been developed because the process of developing new solar panels is slow and out-of-step with the demands from consumers and businesses.
The new approach is centered on figuring out faster and more accurate ways for technologists to troubleshoot early-stage materials and prototype devices. The basis of this is computer modeling of the physical properties of different candidate materials. The steps with the computer model begin with making a test device; following this the current output is measured under different levels of illumination and different voltages. This enables the researchers to quantify how the performance will differ under variable and changing conditions. This approach is described as the “Bayesian inference process.”
Bayesian inference is a method of statistical inference to assess the probability for a hypothesis in real-time as more evidence or information becomes available. Bayesian inference is an important technique in statistics, and especially in mathematical statistics.
According to lead researcher Dr. Rachel Kurchin conventional methods for solar power research “”often require you to make a specialized sample, but that differs from an actual cell and may not be fully representative” of a real solar cell’s performance”, which contributes to the slow pace of most research.
The new experimental design has been used by the researchers to measure the electrical output of a sample of solar cell material. This took place under laboratory conditions of where temperature and illumination were varied. The data from the tests was used as the basis for computer modeling, designed to predict the overall performance of the material in real-world operating conditions.
The approach has been described in the journal Joule, with the peer reviewed paper headed “Rapid Photovoltaic Device Characterization through Bayesian Parameter Estimation.”
