Modulation-Doped Si/SiGe and Si/SiGeC Heterostructures

The market for microelectronic components has reached a volume of more than 200 billion US$ in 2000 and is thus of utmost importance for the development of information technologies (IT) in the western societies. 95% of the microelectronic market are based on the semiconductor Si, which is the only material suited for ultra-high integration levels on computer chips. In recent years, a new market developed in the field of high-frequency communication, which was mainly driven by cellular phones and meanwhile "blue tooth" local area networks. It is rather surprising that the celebrated Si technology is not entirely capable of covering this markets. Other semiconductor materials, which are unsuited for monolithic integration with the Si-based digital circuitry, are required for a few key components in the analog front end of such applications. The aim of this project was to assess alloys and heterostructures containing the group-IV elements silicon (Si), germanium (Ge) and carbon (C) as potential materials that can boost the high frequency properties of Si without sacrificing the compatibility with large-scale integration technology. The main emphasis was put on the Si 1-y Cy /Si heterostructure on which only rudimentary knowledge existed at the beginning of the project. Just a few atomic % of C in a Si crystal can drastically modify the optical and electrical properties. However, the bonding-lengths of Si and C differ significantly, which implies that alloying even at such small C concentrations requires low-temperature epitaxy far from thermal equilibrium. In the initial states of the project we could for the first time demonstrate that it is feasible to grow Si 1-y Cy crystals with more than 2% C on Si lattice sites. Such substitutional C incorporation is the single most important prerequisite for any application. A further refinement of the growth parameters was based on layer characterization with various techniques, and finally led to Si 1-y Cy quantum wells of low defect density as demonstrated by x-ray and luminescence experiments. Further investigations were dedicated to the strain relaxation behavior of the metastable Si 1-y Cy films, which is important for device applications because of the inherent high-temperature annealing steps. We found that in contrast to some reports in the literature strain relaxation in Si 1-y Cy occurs exclusively via ß-SiC precipitation. We developed a highly sensitive experimental technique to trace ß-SiC formation by means of a characteristic infrared line in the absorption spectrum. In contrast to early expectations, the detailed assessment of the optical and electrical properties revealed that Si 1- y Cy is not well suited for high-speed field-effect transistor applications. This is due to the high local lattice distortions around C atoms and C clusters, which lead to enhanced carrier scattering and reduced mobilities. On the other hand, Si/SiGe heterostructures, which we investigated in parallel up to the device level, are very well suited for high-speed field effect transistors, but they are more complicated to grow and process than the simpler Si 1-y Cy layers would have been. Major results from the project were utilized in a subsequent industrial collaboration that was partly funded by the FFF. This will most likely lead to the use of Si 1-x-y Gex Cy layers in heterobipolar transistors, where rather small C concentrations have been shown to drastically suppress unintended dopant diffusion during process annealing steps.