Accurate and efficient computation of the optical properties of Nanostructures for improved Photovaoltaics
Lars Corbijn van Willenswaard
Jaap van der Vegt
Complex Photonics Systems (COPS) chair at the University of Twente
In the production of 3D nanostructures there is unavoidable fabrications disorder that cause differences between computations on idealized models and real life optical measurements. The key challenge in this project is to bring the computations closer to reality by stepping away from idealized models and including the fabrication disorder in the computation. Using data from actual nanostructures obtained by X-ray holotomography, with spatial resolution of 20nm, we aim to compute the propagation in an actual structure. Not only will this allow the comparison of computational results with real life measurements, it also gives crucial insights of what happens inside the structure. These insights could lead to improvements in the design and fabrication of the nanostructures, which can be used to trap light in photovoltaic cells for higher efficiency.
To include all the small details into a traditional finite element computation would lead to an enormous amount of elements. We therefore plan to extend our current discontinuous Galerkin discretization for Maxwell's equations to use virtual elements, which are not constraint to standard shapes like tetrahedra or hexahedra. Using the flexibility of almost arbitrary polytopes as mesh elements allows us to decouple size of the geometric details from the mesh size. This results in a large reduction in the number of elements and thus an increase the computational efficiency.