Base doping profile control for SiGe PNP HBTs

The aim of this thesis is to investigate three aspects related to phosphorus diffusion for doping profile control in PNP SiGe HBTs: We systematically and quantitatively investigated the impact of carbon and Ge on P diffusion in strained SiGe:C up to 18% Ge and 0.32% C through experiments, which shows that the incorporation of carbon to retard P diffusion is not as effective in SiGe as it is in Si. Models were established to calculate the effective P diffusivities as a function of carbon concentration. These models can also be applied to boron, phosphorus, arsenic and antimony diffusion in Si with the presence of carbon. These results indicate that the microscopic mechanism of P diffusion in Si₀.₈₂Ge₀.₁₈ has a small but non-negligible vacancy-mediated term. An experimental study of thermal nitridation effects on phosphorus diffusion in strained Si1-xGex and strained Si1-xGex:Cy was performed. P diffusivities under thermal nitridation (vacancy injection) and the effective inert condition were compared. The result shows that thermal nitridation can retard P diffusion in SiGe with up to 18% Ge content, but the effectiveness of this retardation decreases with increasing Ge and C content. The Ge dependence can be explained by the increasing contribution from vacancy-assisted mechanism for P diffusion in strained SiGe with the increasing Ge content. P tends to segregate out of SiGe region, which happens simultaneously with diffusion. A coupled diffusion and segregation model is needed to predict the P profile evolution at thin SiGe layers. The model was re-derived theoretically, where the contributions from diffusion and segregation to dopant flux are explicitly shown. The model is generic to coupled diffusion and segregation in inhomogeneous alloys, and provides a new approach in segregation coefficient extraction. This model is especially helpful for heterostructures with lattice mismatch strains. Experiments of coupled P diffusion and segregation were performed with graded SiGe layers for Ge molar fractions up to 0.18, which are relevant to PNP SiGe HBTs. The model was shown to describe both diffusion and segregation behavior well.Applied Science, Faculty ofMaterials Engineering, Department ofGraduat