One of the most likely materials to replace silicon in building solar cells is a material called perovskite. However, one of the biggest problems with perovskites is the tendency of the material to degrade relatively quickly. The service life of a perovskite-based solar cell has gradually improved from minutes to months, but it is much shorter than the decades that silicon solar panels can last.
An international team of researchers led by MIT scientists has developed a new approach to narrow the search for the best candidates for long-term perovskite formulations. This is a challenge due to the large number of possible combinations. The system allowed researchers to focus on a composition that in the laboratory improved existing versions of solar cells based on perovskite by a factor of 10.
The new formulation was tested in real-world conditions at the level of the total solar cell, not just in a laboratory sample. The perovskite formulation performed three times better than the latest generation formulations. Perovskites are a wide class of materials characterized by the way atoms are arranged in their layered crystalline structure. The layers within the material are described by convention as A, B and X, each consisting of several atoms or compounds.
Researchers need to research a large number of combinations to find the most likely combination that will provide longevity, efficiency, manageability and availability for the source materials. A scientist on the project says that you should consider even just three elements, the most common elements in perovskites that people insert or are not on side A of the crystalline structure of perovskite. The elements can be varied in increments of one percent in the relative composition, and the number of steps becomes absurd.
The team uses a data fusion approach. It is an iterative method that uses an automated system to guide the production and testing of various formulations. The system uses machine learning to analyze the results of these tests combined with physical modeling of first principles to guide the next round of experiments. The system repeats the process and refines the results each time.