Hole mobilities in relaxed and strained undoped SiGe layers have been calculated with a one-dimensional self-consistent bipolar Monte Carlo simulation code. We have adopted a novel bandstructure model that incorporates strain effects in the alloy valence band. Both alloying and strain enhance the hole mobility compared with bulk Si and we find that alloy scattering is the dominant scattering mechanism. An alloy potential of 1.4 eV was obtained by matching our Monte Carlo data on drift mobilities to experimental Hall mobility measurements. Uncertainties in this value arise from scatter in the experimental data and a lack of detailed knowledge of the Hall factor.

Low-field hole mobilities have been calculated for heavily relaxed and strained doped SiGe alloy layers with Ge contents varying from 0 to 50%, using a novel semi-analytical bandstructure model which incorporates the effects of strain on the valence band of the alloy. We obtain poor results compared with experiment for mobilities in heavily doped Si, and attribute this to (i) a failure of the Born approximation at low carrier energies and (ii) the omission of additional effects associated with heavy doping and high carrier concentrations. For the strained doped and intrinsic alloy we observe that both the in-plane and out-of-plane hole drift mobilities increase with increasing Ge content relative to those for Si. These enhancements are due mainly to the effects of strain, and to a lesser extent due to alloying with Ge, but are offset by the presence of alloy scattering. Our results are sensitive to the details of the models used for scattering by ionized impurities; however, the large uncertainties and scatter of the experimental data preclude accurate estimates of the alloy potential. We find that an alloy potential of 2.0 eV gives an in-plane mobility consistent with experimental data for intrinsic material. Our calculations for heavily doped layers are affected by uncertainties in the value of the alloy potential and highlight a need for a better quantitative understanding of the scattering processes which are important in heavily doped alloy layers.

We present self-consistent Monte Carlo simulations of the collectors of Si and GaAs heterojunction bipolar transistors (HBTs) with wide (700 nm) collectors such as are used in power transistors. Recent experimental data on peak cut-off frequencies f T of 76 GHz for GaAs/AlGaAs HBTs have provided strong direct evidence for extended velocity overshoot in the collector under base push-out conditions and there are theoretical results using fast semi-analytic techniques confirming this explanation. Our aim is to investigate how these regions of extended velocity overshoot might form and how they would affect f T at high current densities in such structures. Differences in velocity-field characteristics and velocity overshoot between Si and GaAs collectors have large effects which are quantified by calculating f T . Comparisons are also made with data from the semi-analytic results.

A one-dimensional self-consistent bipolar Monte Carlo simulation code has been used to model carrier mobilities in strained doped SiGe and the base-collector region of Si/SiGe/Si and SiC/Si heterojunction bipolar transistors (HBTs) with wide collectors, to study the variation of the cutoff frequency f T with collector current density J C . Our results show that while the presence of strain enhances the electron mobility, the scattering from alloy disorder and from ionized impurities reduces the electron mobility so much that it is less than that of Si at the same doping level, leading to larger base transit times τ B and hence poorer f T performance for large J C for an Si/SiGe/Si HBT than for an SiC/Si HBT. At high values of J C , we demonstrate the formation of a parasitic electron barrier at the base-collector interface which causes a sharp increase in τ B and hence a dramatic reduction in f T . Based on a comparison of the height of this parasitic barrier with estimates from an analytical model, we suggest a physical mechanism for base pushout after barrier formation that differs somewhat from that given for the analytical model.