The growth of high quality, thin (14-70 Å) epitaxial CoSi2 layers on Si(100) substrates have been observed. Three types of films have been grown on the Si(100) substrate; pure CoSi2(100), pure CoSi2(110), and mixed CoSi2(100)+CoSi2(110). The CoSi2(110) can exist in two 90º rotated variants on the Si(100) surface and these are observed. Methods to isolate the epitaxial orientations are described. Results from in-situ LEED and AES as well as ex-situ RBS and plan view TEM are presented.

IrSi3, ReSi2, and Rh3Si4 have been grown epitaxially on (111) and (001) Si. Transmission electron microscopy has been applied to characterize the epitaxial films. The orientation relationships between epitaxial sulicides and substrate silicon were determined from electron diffraction analysis. The average size, areal fraction of surface coverage, and fraction of epitaxy were measured. The silicide/silicon interface structures were analyzed.

The best epitaxy in the three metal/Si systems was found to be epitaxial IrSi3 in Ir/(111)Si samples annealed at 1000 °C. Epitaxial regions, as large as 40 μm in size, were observed. Three major modes of IrSi3 epitaxy on (111)Si were identified.

Chromium disilicide layers have been grown on Si(111) in a commercial molecular beam epitaxy machine. Islands are found to nucleate with more than one epitaxial orientation, and three different orientations have been identified: CrSi2(0001)//Si(111) with CrSi2 [1010]//Si[110]. CrSi2(0001)//Si(111) with CrSi2[ll20]//Si[1l0], and CrSi2(1100)//Si(111) with CrSi2[000l]//Si[112]. Very thin (2 nm) layers have been studied under a variety of growth conditions to determine factors controlling the orientation of nuclei. The degree and direction of substrate misorientation, the growth technique, and the use of a CoSi2 buffer layer are all factors affecting the morphology of CrSi2 islands.

Studies of the electrical and structural properties of TiSi2 films formed during the fabrication of shallow junctions by ion implanting As+ into p–Si through Ti films are reported here. Electrical measurements of the temperature dependent specific contact resistance and structural measurements such as transmission electron microscopy have been performed on these device structures. The specific contact resistance is seen to decrease with increasing measurement temperature. These studies have been carried out on furnace and rapid thermally annealed films.

Epitaxial regrowth of gold film on Si as a result of Au/Si eutectic reaction and epitaxial aluminum spikes forming at IC contacts during sintering have been investigated by transmission electron microscopy (TEM). For gold film on Si, three types of epitaxy were observed: (1) the crystal structures of the two lattices are exactly the same, (2) the two structures have the orientation relationship of Au(111)//Si(111) and Au[132]//Si[231] and (3) 20° misorientation from the relationship in (2). Two orientation relationships were observed in the case of Al spike in Si: (1) Al[011]//Si[123] and Al(200)//Si(11), (2) Al[001]//Si[112] and 2° misorientation between Al(200) and Si(111). Possible mechanisms are proposed to explained the observed epitaxial growth.

This paper describes preliminary device work on GexSil−x based devices and examines progress towards the realization of a silicon based heterostructure technology. Target areas for further work are identified including: An enhanced understanding of metastable-growth processes; The need for defect confinement in discommensurate epitaxy; Elimination of manufacturing barriers caused by particulate induced defects in MBE and low growth rates in low temperature CVD.

Over the last five years, studies of GexSi1−x have yielded an understanding of crystal growth mechanisms and information on electrical and optical properties such as bandstructure and heterojunction band alignment. These in turn lead to a number of preliminary device applications including optical detectors, modulation doped transistors and heterojunction bipolar transistors. In this paper, I will try to answer the question posed by the symposium organizers: In the germanium silicide system, has our understanding and experience reached the point that we have the basis for a technology, and if not, where must further work be done?

The physics governing growth and stability properties in SiGe/Si strainedlayer structures is reviewed. The role of metastability in crystal growth is outlined. Experimental data on stability limits and rates of strain relaxation are examined. We conclude that essentially all observations on relaxation of semiconductor strained-layer structures can be explained by standard models of plastic deformation adapted to the special conditions controlling dislocation dynamics in these structures.

We describe in-situ transmission electron microscope observations of the relaxation of strained layer GeSi/Si epitaxy. Dynamic observations of misfit dislocations in these structures reveal that dislocation nucleation and growth activation barriers, as well as interactions, limit the rate at which strain is relieved. The equivalence of threading and misfit dislocations in this system is demonstrated. Extension of the principles learnt from these single layer experiments to threading dislocation propagation through multilayer structures, enables us to understand the relative inefficiency of GeSi/Si strained layer superlattices in blocking threading dislocations.

Electronic properties of the n-doping--insulator--p-doping--insulator structures in ultra thin strained [001] Si - Ge superlattice have been studied theoretically. The Ge - layer is used as one of the insulating region. The Al and As atoms are treated as impurities. The superlattice ((Si)10 -(Ge)2) exhibits an indirect gap in reciprocal space and the staggered band alignment in real space. With doping, the samples show a direct band gap and staggered band alignment. The acceptor state is associated with the Al-Si bonding state, while the donor state is derived from the As s-like state. The separation of the charge carriers in the real space can be obtained.

Heteroepitaxial growth of Ge on Si(100) has been accomplished using remote plasma enhanced chemical vapor deposition at 300*#x00B0;C. Reconstructed surfaces with diffraction patterns showing non-uniform intensity variations along the lengths of the integral order streaks are observed during the first 100 Å of deposit. This observation of an atomically rough surface during the initial stages of growth is an indication of three-dimensional growth. As the epitaxial growth proceeds, the diffraction patterns become uniform with extensive streaking on both the integral and fractional order streaks. Subsequent growth, therefore, takes place in a layer-by-layer, two-dimensional mode. X-ray photoelectron spectroscopy of the early nucleation stages, less than 80 Å, show that there is uniform coverage with no evidence of island formation.

One of the approaches to the growth of GaAs on Si substrates involves the use of a heteroepitaxial buffer layer such as Ge to reduce lattice mismatch. Recently, we have had success in growing heteroepitaxial Ge films on low temperature (∼275-500°C) Si substrates by the ionized cluster beam (ICB) technique. The design of a computer-controlled phase modulated ellipsometer for rapid monitoring of the initial stages of nucleation will be described. Ellipsometric data for single crystalline, polycrystalline and amorphous morphologies are reported.

A low temperature process for cleaning Si(100) surfaces has been developed. It involves a combination of a modified hot RCA wet chemistry treatment and an in situ hydrogen treatment for the removal of oxides and carbonaceous material from the Si surface. While this treatment is successful in producing reflection high energy electron diffraction patterns which show 1/2-order reconstruction lines, subsequent Ge heteroepitaxial growth at 300°C contains a high density of microtwins. Transmission electron microscopy reveals that most of the microtwins do not propagate to the wafer surface. Furthermore, the Ge/Si interface is not abrupt, and there are regions that do not appear crystalline. This suggests that some contamination is still present on the Si(100) surface after the in situ hydrogen treatments.