Boron and phosphorus diffusion gettering: Efficiency, mechanisms and applicability to silicon solar cells

As it is impossible to completely prevent the presence of impurities in semiconductor device processing, various defect engineering strategies are needed. The purpose of this thesis is to increase the level of understanding of different impurity gettering techniques in silicon, especially from solar cell point-of-view. Experiments made with single crystalline silicon provide new data for both boron diffusion gettering (BDG) and phosphorus diffusion gettering (PDG). In the case of PDG, the presented steady state data at the temperature range of 650-800°C enables the determination of an accurate value for segregation coefficient of iron. For BDG, gettering efficiencies even comparable to PDG are achieved. In this case, the prevailing mechanism is found to be B-Si precipitate gettering, which requires that the boron concentration at the emitter exceeds its solubility at the annealing temperature for a sufficiently long time for nucleation and growth of B-Si precipitates. Studies made on multicrystalline silicon (mc-Si) show that both BDG and PDG lead to a significant increase in minority carrier lifetime and a substantial reduction of the red zone area. Due to the high amount of precipitated metals, the higher in-diffusion temperature of boron is found to be beneficial in mc-Si. After BDG, the average lifetime in the red zone is increased from 5 us up to 270 us. The phosphorus implantation gettering experiments reveal that the obtained gettering efficiency for implanted emitters is significantly lower in comparison to conventional PDG. However, if precipitation at the emitter is activated, considerable improvement in the gettering efficiency is achievable. This requires low gettering temperatures and/or high initial contamination level. Finally, gettering processes optimized based on the material parameter results are applied on device level in solar cell processing. With initial bulk iron concentration as high as 2 × 1014 cm-3, conversion efficiencies comparable with non-contaminated cells are obtained. The results clearly demonstrate the importance of careful process parameter design: to achieve best results, the gettering parameters used for high purity silicon should be chosen differently as for a material with high impurity content.