Exploring and Understanding Photovoltaic Materials Using Atomistic DFT Modelling
In the Thin Film Solar Cell group at Uppsala, we are doing research and development on photovoltaic thin film materials such as Cu(In,Ga)Se2 and Cu2ZnSn(S,Se)4 (CIGS and CZTS). Increasingly important in that work is to combine experimental studies with theory to understand the underlying physics and to help choose among materials synthesis strategies. The work is carried out within the SSF framework project “Gradient control in Thin Film Solar Cells” (Prof. Charlotte Platzer-Björkman) and is a collaboration between Uppsala University and KTH Stockholm (Prof. Clas Persson).
In this project, the crystalline structure, optical properties, phase stability, defect/cluster formations and diffusion processes will be modelled. We utilize regular LDA/GGA, the hybrid functional HSE06 and the GW approach, depending on the system size and on the specific material system in consideration. From our previous studies, we found that number of k-points in the irreducible Brillouin zone affects the detailed shape of the dielectric function significantly. This is important to point out, since most optical property calculations are performed by either HSE06 or GW, and calculation time can be increased markedly when the number of k-points is increased for both methods. We will follow existing methodologies to analyse the phase stability and defect/doping physics. The calculations of formation energies and chemical potentials will reveal fundamental crystalline and defect properties. This analysis is important also when analysing the acceptors, to avoid precipitation and defect formations between dopants. Together with analysis of disordering in CZTS, we will further develop the model of grain boundaries used in our previous studies on CIGS and will estimate their influence on solar cell parameters. Moreover, to understand the diffusion behaviour of point defects, two different models will be used. The first model will include nudged elastic band calculations for analysis of defect diffusion pathways, searching for the lowest energy and saddle configurations. The predicted configurations will be used for analysis of atomic vibrations and their contribution to the diffusivity. The second model will involve the use of ab initio molecular dynamics (AIMD) simulations for the analysis of diffusion. Here, to include the electrochemical contribution, we aim to modify the AIMD code by adding a numerical solution of Poisson's equation. Similar to other simulations, the AIMD simulations will be performed at higher temperatures to obtain better diffusion statistics for diffusivity estimation, and the results will be extrapolated to predict the diffusivity under realistic temperatures.