The Pulsed Laser Atom Probe has been used to determine for the first time the stoichiometry of both the native oxide on Si and of the SiO2/Si interface.

(100) single crystal silicon films have been deposited onto (100) oriented Yttria-Stabilized Zirconia (YSZ) substrates by pyrolysis of SiH4 at ∼ 980°C.

The as deposited epitaxial silicon films have been characterized by Reflexion High Energy Electron Diffraction and Transmission Electron Microscopy techniques.

The as deposited silicon films have also been oxidized by oxygen transport through the substrate, resulting in a Si(100)/ amorphous SiO2/YSZ(100) structure in which the most defective part of the epitaxial silicon deposit has been eliminated. The oxidized interfaces (with SiO2 thicknesses in the 2000 Å range) have then been characterized by Transmission Electron Microscopy in order to assess the improvement in crystalline quality. Electrical measurements have also been performed on MOS-Hall bar structures.

A dual-implantation and annealing procedure has been refined which results in silicon films of single crystal perfection on sapphire substrates. This double solid phase epitaxy (DPSE) procedure consists of two self-implantation steps that serve to amorphize the defect structure of the parent silicon film with subsequent annealing steps employed to recrystallize the amorphized structure. High resolution cross-section transmission electron microscopy (HRXTEM) and ion channeling have been employed to study the microstructure of the silicon films, the sapphire substrates, and the interface. The resulting perfection of the silicon layer is shown to be a sensitive function of the first self-implant energy, dose, and silicon film thickness. Also, a correlation has been established between microtwin density and dechanneling profiles. HRXTEM of microtwins in silicon on sapphire (SOS) starting material demonstrates contrast features identified as rotational Moire' fringes between the microtwins and the matrix by means of optical diffractometry. Best results are achieved for the first implant energy of 170 keV 28Si+. Thus, the DSPE process is shown to give optimal results in defect reduction.

The capability of growing high quality strained-layer superlattices (SLS's) from lattice-mismatched semiconductors opens up new opportunities in both basic and applied material sciences. This flexibility in the choice of SLS layer materials allows the study of quantum well effects in a variety of new and interesting heterojunction systems. In addition, the large tetragonal distortions of the SLS layers allow these structures to exhibit unique features which are not found in bulk semiconductors. An exciting example is the possibility of tailoring certain SLS materials so that they exhibit small two dimensional hole effective masses and enhanced low field hole mobilities at low temperatures and low carrier concentrations. Band structure calculations indicate that the layer strains, the direct band gap magnitudes of the layers, the valence band offset, and the hole energies all can play a role in determing these small two dimensional hole effective mass values.

The low-temperature (20K) photoluminescence of InxGa1-xAs and InxGal-xAs - GaAs strained-layer superlattices (SLS's) grown by molecular beam epitaxy (MBE) is investigated. Data are presented for thick (bulk) epitaxial layers grown directly on GaAs and for relatively-thin (˜600Å) InxGa1-xAs layers under biaxial compression. Data are also presented for two series of SLS's. In the two series of SLS's, the InxGa1-xAs layer thickness (Lz) is held constant while only the GaAs layer thickness (LB) is varied. The photoluminescence (PL) spectra of the crystals are useful in analyzing the effects of biaxial strain, carrier confinement, and barrier layer thicknesses in SLS's. Results are compared with calculations based upon a modified Kronig-Penney model which incorporates the appropriate deformation potentials for SLS analysis. This type of analysis, in agreement with experimental data, suggests that the electron-to-light-hole transition can be lower in energy than the electron-to-heavy-hole transition in SLS's, depending upon layer thickness and crystal composition.

Cross-sectional TEM of GaSb/InAs superlattices grown by MBE on (100) GaAs and (100) GaSb substrates shows an unusual defect structure within the strained layers. Dislocations are present within the layers and at the interface. High-resolution TEM analysis of the structure of the InAs layers suggests that these layers grow by an island mechanism. A crystal structure different from the zinc blende, is found to be present within the GaSb layers.

This paper reviews recent work on GexSi1-x/Si(100) strained-layer epitaxy and reports new findings on GexSi1-x/Si(111) and GexSil1-x/Ge(100) growth as well as results on modulation doping. Layer synthesis and evaluation techniques are described along with tabulations of strain and critical layer thickness data. Evaluation techniques include Low Energy Electron Diffraction, Rutherford backscattering, X-ray diffraction, Raman scattering and cross-sectional Transmission Electron Microscopy. Synthesized structures range from simple heterojunctions to 100 period superlattices.

It has been demonstrated in the literature that amorphous Si (or Ge) can be transported across a metal layer and grown epitaxially on Si(Ge) single crystal substrates in the solid phase. The objective of this study is to investigate if amorphous SixGe1−x mixtures can be transported uniformly across a medium and grown epitaxially on single crystal substrates without phase separation. The samples were prepared by e-beam evaporation of thin Pd films onto Si<100> substrates, followed by co-evaporation of SixGe1−x alloyed films (0<x<1) without breaking vacuum. The samples were anneaie in vacuum at 300°C to form a Pd silicide-germanide layer at the interface, then at 500°C for transport of the alloyed layer across the Pd silicide-germanide layer and subsequent epitaxial growth on Si substrate. The samples were investigated by x-ray diffraction and by MeV ion backscattering and channeling. The results show the alloyed film transports uniformly with no phase separation detected. The channeling result shows the grown alloyed layer is epitaxial with some Pd trapped in the layer. This simple technique is potentially useful for forming lattice-matched non-alloyed ohmic contacts on III–V ternary and quaternary compounds.

High resolution electron microscopy is used to probe the atomic scale structure of interfaces and defects in the GexSi1−x/Si system. By careful quantification of lattice images, it is shown that molecular beam epitaxy may be used to grow GexSi1−x/Si (100) and (111) interfaces which are sharp on the scale of the unit cell and flat to within a few atomic planes when about 5000 Å2 of the interface are sampled. Interfacial quality is retained in single and multiple quantum well structures. Conditions for superlattice stability against misfit dislocations are discussed. It is shown that GexSi1−x/Si interfaces produced by molecular beam epitaxy at 550°C can exist in a metastable state which relaxes upon thermal annealing.

Transmission electron microscopy provides a powerful means of studying compositionally modulated materials. In such materials there is usually a local variation in electron scattering power along with a lattice dilatation wave which both accompany the local composition. The most revealing geometry for studying such materials has the lattice modulation direction lying within the plane of the thin foil. However, shear stresses accompanying the dilatation wave can be significantly relaxed by the presence of the thin foil surfaces, modifying the local atomic displacement field such that it is representative of neither the bulk, nor the free unstressed material. Two pertinent semiconductor examples which we have studied are spinodally decomposed quaternary III–V layers and strainedlayer superlattices of Si/SixGe1−x. We provide experimental evidence demonstrating relaxation in these cases and a simple elasticity model to describe it. Our data and model show a thickness dependence to relaxation and can explain previously reported ‘anomalous’ lattice parameter measurements from a strained-layer superlattice [11]. In this paper we concentrate on the effects of dilatation and relaxation on imaging and diffraction from a strained-layer superlattice.

Pseudomorphism, the epitaxial growth of metastable phases, is reviewed. Emphasis is given to recent developments in the qrowth and investigation of α-Sn films on InSb and CdTe since this is a case where there is a very large strain energy barrier to the metastable to stable phase transformation. As a result, pseudomorphic growth of α-Sn can be sustained to film thicknesses of ∼1 μm.

Progress towards stabilization of direct gap group IV semiconductors is discussed and possible applications of pseudomorphism considered.

Recent results on metastable semiconducting alloys, concerning in particular the growth of new Sn-based alloys (GaSb)1−x(Sn2)x and Gel−xSnx and the physical properties of (GaAs)1−x(Ge2)x and (GaSb)1−x(Ge2)x, are discussed. (GaSb)1−x(Sn2)x and Ge1−xSnx alloy films were grown with x-values as high as 0.20 and 0.15, respectively, well in excess of equilibrium Sn solid solubility limits (<1%) while epitaxial (GaAs)1−x(Ge2) and (GaSb)1−x(Ge2)x alloys were obtained on (100) GaAs at compositions ranging across the pseudobinary phase diagram. Low energy ion bombardment induced collisional mixing and preferential sputtering during film growth played a critical role in obtaining single phase alloys. An optimal ion energy, which depended on the ion flux and the alloy composition, was determined, allowing in most cases growth at temperatures T, sufficient for obtaining single crystal alloys on (100) GaAs and (100) Ge substrates. Decomposition of the Sn-based alloys occurred above a critical Ts- value via α-Sn-rich precipitates which were stable above the β-Sn melting point. X-ray diffraction, STEM, EXAFS, and Raman spectroscopy measurements, performed on single crystal (GaAs)1−x(Ge2)x and (GaSb)1−x(Ge2)x alloys, indicate that there is a transition in the long-range order from zincblende to diamond with increasing x while the short-range order remains perfect at all compositions, i.e. no V-V or III-Ill bonds are observed. These results are discussed in light of recent models which relate (GaAs)1−x(Ge2)x atomic structure to its band structure and optical properties.

The thermodynamics and kinetics governing the stability of compositionally modulated materials are reviewed. Typical results of interdiffusion experiments on fcc and amorphous metals are discussed to demonstrate the sensitivity and special features of the technique. Two coarsening mechanisms, and the stability of orthogonal waves in phase separating systems are analyzed.

Cantilever-beam bending and RBS channeling measurements have been used to examine implantation-induced disorder and stress buildup in In0.2Ga0.8As/GaAs SLS structures. Implantation fluences from 1011 to 1015/cm2 were used for 150 keV Si, 320 keV Kr, and 250 keV Zn in SLS and GaAs bulk materials. The critical fluence for saturation of compressive stress occurs prior to amorphous layer formation and is followed by stress relief. For all the ions the maximum ion induced stress scales with energy density into atomic processes and stress relief occurs above ∼1 × 1020: keV/cm3. Stress relief is more pronounced for the SLSs than for bulk GaAs. We suggest that stress-relief may lead to slip or other forms of inelastic material flow in SLSs, which would be undesirable for active regions in device applications. Such material flow may be avoided by limiting maximum fluences or by multiplestep implantation and annealing cycles (or hot implants) at high fluences.

We have examined the optical and transport properties of In.2Ga.8As/GaAs straled-kayer superlZotices (SLS's), which have been implanted either with 5 × 1015/cm2, 250keV Zn+ or with 5 × 1014/cm2, 70keV Be+ and annealed under an arsenic overpressure at 600 °C. For both cases, electrical activation in the implantation-doped regions equalled that of similar implants and anneals in bulk GaAs, even though the Be implant retained the SLS structure, while the Zn implant intermixed the SLS layers to produce an alloy semiconductor of the average SLS composition. Photoluminescence intensities in the annealed implanted regions were significantly reduced from that of virgin material, apparently due to residual implant damage. Diodes formed from both the Be- and the Zn-implanted SLS's produced electroluminescence intensity comparable to that of grown-junction SLS diodes in the same chemical system, despite the implantation processing and the potential for vertical lattice mismatch in the Zn-disordered SLS device. These results indicate that Zn-disordering can be as useful in strained-layer superlattices as in lattice-matched systems.

Strain measurements in AlxGa1−x As/GaAs superlattices have been carried out before and after Si ion implantation. For doses up to 5 × 1015 cm−2, no atomic intermixing of the sublayers is observed by backscattering spectrometry. However, with x-ray rocking curve measurements, significant changes in the strain profiles are detected for implantations with doses as low as 7 × 1012 cm−2. Interpretation of the rocking curves suggests that low-dose implantations release strain in the AlxGa1−x As sublayers. The strain profile recovery of the implanted samples, upon annealing at ∼ 420°C, implies that the damage caused by implantation is largely reversible.

In the context of a model system whose atoms interact via Lennard-Jones (LJ) interatomic potentials, we have studied the stability of an initially perfect strained layer superlattice interface (SLS) and the process of growth of a mismatched layer on a substrate using Monte Carlo techniques. An initially perfect SLS interface is found to be metastable up to 12% mismatch, a much higher value than is found on real SLS systems. In contrast, we find that, within the limitations of the calculations, a perfect SLS interface cannot be grown in an LJ system. The implications of these results for understanding SLS growth processes is discussed.

Misfit dislocations in epitaxial layers of Si grown by MBE at 570°C on GaP(001) substrates have been studied by TEM. It is found that layers as thick as 500 Å at least reside coherently on the substrate without misfit dislocations. In 1000 Å layers of Si the misfit strain is accommodated in part by 60-degree type dislocations with their Burgers vector inclined with respect to the interface, and by stacking faults intersecting the Si layer. The dislocations are dissociated into 30- and 90-degree Shockley partial dislocations. It is shown that in the case of a biaxial strain field, which is tensile in a (001)-plane, the 90-degree partial must be nucleated first. Only then can the 30-degree partial follow on the same glide plane. This geometrical effect explains the presence of dislocations as well as stacking faults in the Si layer.

Ion backscattering and channeling techniques, transmission and scanning electron microscopy, and secondary ion mass spectroscopy were used to investigate the reordering characteristics of implanted amorphous Si in the presence of an Al surface layer. It was found that reordering takes place at the temperature of about 400°C and is associated with an interfacial migration between Al and Si. The regrowth behavior appears to be a function of the initial annealing temperature and annealing sequence. High density of twin faults and substantial concentration of Al are observed in the regrown layers. We believe that the low temperature reordering is due to processes analogous to solid epitaxial growth with transport media.

Electrical behavior at single crystal silicide-silicon interfaces was studied. Schottky barrier heights were determined for epitaxial NiSi2 and CoSi2 layers grown under ultrahigh vacuum conditions on (111), (100) and (110) surfaces of Si. A dependence of Schottky barrier heights on interface structure was observed. These results favor intrinsic mechanisms for Schottky barrier formation. The advantages of having homogeneous metal-semiconductor interfaces for the study of Schottky barrier mechanisms are pointed out. In particular, the present epitaxial silicide-silicon interfaces represent ideal candidates for detailed theoretical investigations based on experimentally obtained atomic structures.