Controlling dopant rich layers in solar cells

Improving the efficiency of silicon (Si) solar cells relies on the understanding and optimization of individual processing steps. One such processing step, namely the diffusion of phosphorus (P), has been found to influence cell efficiencies significantly. This thesis discusses a mathematical model of P diffusion in Si based on first order approximations. Phosphorus diffusion studies have been carried out in small increments of time that facilitate the development of a diffusion model which is shown to agree reasonably well with experimental profiles over a wide range of processing conditions. It is assumed that phosphorus diffuses via vacancies in the high concentration region and via interstitials in the low concentration region. A supersaturation of interstitials due to the kick out reactions enhances dopant diffusivity in the low concentration region. However, the dissociation of phosphorus interstitial pairs into phosphorus vacancy pairs at high doping levels facilitate the suppression of excess interstitials with the help of surface injected vacancies, resulting in the lowering of the overall free energy. For the very high doping levels, the dissociation rate of phosphorus interstitials is rapid in comparison to the kick out reactions that enables the slowly injected surface vacancies to completely suppress the interstitial supersaturation, thereby restoring point defect equilibrium wherein the system attains the lowest energy configuration. The model uses the relative (to the intrinsic) concentration of point defects instead of actual point defect concentrations. This approach is particularly useful since it significantly reduces the number of fitting parameters of the model by allowing the use of an effective diffusivity in the flux equation. The effective P diffusivity is also used in the modelling of dopant precipitation kinetics, where the growth (and dissolution) of precipitates by the addition (removal) of single solute atoms is considered. Appropriate corrections are used to account for the formation of small sized clusters that are often present along with larger precipitates. The diffusion model is argued to be consistent with previous findings and is able to demonstrate very good simulation accuracies compared to all previous models. Electrically inactive P limits the low wavelength response of the solar cell. The simulation developed here, based on the finite difference method (FDM), show that high temperature diffusion processes for very short durations can achieve low levels of electrically inactive P. In addition, such processes are predicted to result in higher levels of the electrically active P thereby improving contact resistance between metal and Si.