This thesis focuses on the effects of dopant compensation on the electrical properties of crystalline silicon relevant to the operation of solar cells. We show that the control of the net dopant density, which is essential to the fabrication of high-efficiency solar cells, is very challenging in ingots crystallized with silicon feedstock containing both boron and phosphorus such as upgraded metallurgical-grade silicon. This is because of the strong segregation of phosphorus which induces large net dopant density variations along directionally solidified silicon crystals. To overcome this issue, we propose to use gallium co-doping during crystallization, and demonstrate its potential to control the net dopant density along p-type and n-type silicon ingots grown with silicon containing boron and phosphorus. The characteristics of the resulting highly-compensated material are identified to be: a strong impact of incomplete ionization of dopants on the majority carrier density, an important reduction of the mobility compared to theoretical models and a recombination lifetime which is determined by the net dopant density and dominated after long-term illumination by the boron-oxygen recombination centre. To allow accurate modelling of upgraded-metallurgical silicon solar cells, we propose a parameterization of these fundamental properties of compensated silicon. We study the light-induced lifetime degradation in p-type and n-type Si with a wide range of dopant concentrations and compensation levels and show that the boron-oxygen defect is a grown-in complex involving substitutional boron and is rendered electrically active upon injection of carriers through a charge-driven reconfiguration of the defect. Finally, we apply gallium co-doping to the crystallization of upgraded-metallurgical silicon and demonstrate that it allows to significantly increase the tolerance to phosphorus without compromising neither the ingot yield nor the solar cells performance before light-induced degradation.
