To make solar cells that can eke out every bit of energy from sunlight, researchers rely on computer modeling tools. These simulators let them assess how minor tweaks to parameters like device structure, materials used, and the thickness of different material layers can affect ultimate power output.
Several solar cell simulator packages are already freely available. But these tools remain slow, and don’t allow researchers to optimize different design parameters simultaneously. New software from a team of researchers at MIT and Google Brain could streamline solar cell improvement and discovery.
Traditional computational tools take the variables for a particular solar cell design as input, and spit out the resulting power rating as the output.
But with the new software, “we provide output but also show how efficiency would change if we change any of the input parameters,” says Giuseppe Romano, a research scientist at MIT’s Institute for Soldier Nanotechnologies. “You can change input parameters continuously and see a gradient of how output changes.”
That reduces the number of times developers have to run these time-consuming compute-heavy simulations. “You do only one simulation and automatically you have all the information you need,” he says. “That’s the beauty of this approach.”
Romano and his colleagues detailed the new software, called a differentiable solar cell simulator, in a paper published in the journal Computer Physics Communications.
Commercial solar cells have light-to-electricity efficiencies that lag behind the devices’ theoretical maximum values. Solar cell simulators let researchers understand how physical factors like material defects affect the final performance of solar cells. Simulators have already helped to improve common photovoltaic technologies such as cadmium-based thin-film cells and perovskite cells.
There are two ways the new tool should help solar cell development. The first is optimization, he says: “Say an actor in industry wants to make a high-performance solar cell but doesn’t know the effect of light-absorbing material on overall efficiency.” There’s usually an optimal thickness for this material layer to create the most charge carriers from the light it absorbs. The software would help define the optimal parameter that maximizes efficiency.
The software could similarly be used to evaluate optimal values for other variables such as the amount of doping of the material layers, the bandgap, or the dielectric constant of insulating layers.
The other way the tool helps is to reverse engineer an existing solar cell. In this scenario, researchers could measure the I-V curve—the function that gives current for each voltage—of a solar cell, and pair up these experimental measurements using the simulator. Based on the data, the software could help calculate the values of specific material parameters that are unknown.
Others might have developed similar solar cell simulators, Romano says, but “this is the first open source simulator with such nuance.” The software package is on GitHub, which should make it easy for anyone to use it and to make improvements, he says.
Researchers could couple it with their own optimization algorithms or a machine-learning system. This should speed up development of more efficient solar cells by allowing quick assessment of a wide variety of possible materials and device structures.