From pv magazine Global
The number of materials with the potential for use in each of the many layers in a solar cell is enormous. And even once they have chosen one to work with, scientists need to understand its interactions with the other materials present, and the effects of changing parameters such as layer thickness, dopant concentration and a wealth of others in order to get the best out of the cells they are working on.
With so many possibilities, this can be a time-consuming process. And scientists today are increasingly able to turn to artificial intelligence to guide them in the next steps to take in practical lab work. And developing a system to do just that for solar cell design was the focus of a group of researchers at the Massachusetts Institute of Technology (MIT), who worked with experts at Google Brain to develop a system to evaluate the potential of different solar cell designs, and also predict which changes would provide improved performance characteristics. “We developed a tool that will enable others to discover more quickly other higher performance devices,” explained MIT researcher Giuseppe Romano. “Our tool can identify a unique set of material parameters that has been hidden so far because it’s very complex to run those simulations.”
The model is described in the paper ∂PV: An end-to-end differentiable solar-cell simulator, published in Computer Physics Communications. With data on a solar cell configuration, it outputs a predicted efficiency, and shows which of the input parameters affects the prediction the most. It can evaluate multiple variables for each layer including doping concentration, dielectric constant and bandgap. And its researchers point to layer thickness as a particularly important variable. “…the thickness is critical,” said MIT scientist Sean Mann. “There is a strong interplay between light propagation and the thickness of each layer and the absorption of each layer.”
MIT is making the model available as an open-source tool that can be taken up by researchers and industry. According to the institute, it is ready to be combined with other algorithms and machine learning processes to quickly assess many possible changes to device design and pull out those with the most potential.
And since it is based on open-source code, future users will be able to optimise it further for more complex cell architecture, such as tandem and multijunction devices. The current version, “can cover the majority of cells that are currently under production, and there are ways to approximate a tandem solar cell by simulating each of the individual cells,” said Romano. “But once it’s up there, the community can contribute to it, and that’s why we are really excited.”
This content is protected by copyright and may not be reused. If you want to cooperate with us and would like to reuse some of our content, please contact: firstname.lastname@example.org.