We are currently working on the attached model of a simple pn solar cell with COMSOL Multiphysics 4.3a. The model has been adapted from the old 3.5a COMSOL semiconductor diode tutorial, which is available here: http://www.uk.comsol.com/showroom/gallery/114/ Unfortunately, whenever I change something in the model, the computation does not converge. For example, if I set the generation and recombination terms to zero, the model does not converge. Another issue is that I am getting absurd values for the current density. Unfortunately COMSOL support is not helpful at all, despite the fact that we purchased a $ 8.5k license.

This model currently employs the following modules: AC/DC and Chemical Reaction. The model hasn't been coupled with the RF module yet, via the generation term.

Maybe this way of implementing a model of a solar cell is not good. Do you know the best way for implementing the drift-diffusion model for a simple pn solar cell? I was thinking about three options:

1) (Implementation via the Plasma Module (which we don't have ... :-( ));

2) Rather than using the continuity equation in the convection and diffusion nodes, where the variables are nc and pc but also the Lagrange Multipliers nc_lm and pc_lm, probably we should use a different formulation where the variables are the IMREF Efn and Efp. This was described in Malm et al., 2009, Solar Energy Materials & Solar Cells doi:10.1016/j.solmat.2008.11.058 . The authors report that . But how do we implement that on comsol multiphysics?

3) Maybe we should use the Equation Based modeling... how do we implement that on comsol?

The objective on the final modeling is to build a solar cell model of a micromorph solar cell, which we will optimize via Matlab Livelink by using the Genetic Algorithm toolbox.

Similar questions and discussions