You can study materials of individual layers, and interfaces between the layers, but not the entire device, since that would be computationally too expensive. In fact, I do not think it makes too much sense to simulate the entire macroscopic solar cell at the atomic scale. As soon as you know microscopic, material-specific parameters of the materials and interfaces comprising the solar cell, you may consider using these parameters as inputs for macroscopic TCAD-like models of solar cells.
