Simulation of light trapping in thin-film solar cells

We study light trapping in a silicon-based thin-film solar cell setup that consists of several randomly textured layers. The focus is on amorphous and microcrystalline silicon (a-Si:H and µc-Si:H) which belong to the family of disordered semiconductors. The main characteristics of those materials is the structural disorder, which affect in an essential way the optical and electronic properties.

Figure 1. Geometrical model of the solar cell structure and composition of the different layers. Layer thicknesses are in the order of the wavelength of relevant sunlight.

Optical data of the constitutive materials have been fitted to a generalized dispersion model, which was originally intended for metals. The obtained permittivity functions for the amorphous silicon a-Si:H are plotted in Figure 2. As can be seen, this material is relatively well approximated.

Figure 2. Real and imaginary parts of the relative permittivity of amorphous silicon a-Si:H fitted to a generalized dispersion model compared to experimental data.

We choose to use Periodic Boundary Conditions (PBC) which allow to simulate artificially infinite mono-directional or bi-directional arrays while considering only one elementary pattern. To do so, cells from a periodic boundary face are matched with their neighbors on the opposite boundary of the domain. We have two possibilities to obtained such boundaries. The first one consist of symmetrizing the mesh. The main drawback of this method is the multiplication of the domain size by 4. This is not a major issue for a small model like in Figure 3, but for the full device this is not a feasible approach. The second option is to manually force the periodicity of the faces using a regular mesh. Before that, we need to modify the topography of each layer to have the same 1D border for each layer. The disadvantage here is the loss of information on the edges of the structure as can be seen in Figure 4.

Figure 3. Symmetrized model. Whitle lines indicate the limits of each submodel. Figure 4. Full model with regular periodic boundaries.