Surface diffusion and crystal growth processes have been studied in the Stranski-Krastanov growth system Ag/Si(111), using several UHV-SEM techniques. By depositing Ag at various rates 0.2 ≤ R ≤ 1.4 ML-min−1 through a mask of holes, surface diffusion of Ag over the intermediate layer has been observed, in competition with re-evaporation at high, and nucleation at low substrate temperatures, in the range 620 < T < 850K. The Si(111) √3Ag intermediate layer has been visualized using biassed secondary electron imaging. The surface diffusion and nucleation processes observed have been analyzed in terms of kinetic models. Comparison with experiment yields values for the adsorption, diffusion and binding (Ea′ Ed and Eb) energies of Ag on the intermediate layer. These values are approximately Ea = 2.45 ± 0.1 eV, Ed 0.35 ± 0.05 eV and Eb = 0.10 ± 0.03 eV, where the uncertainties result at least as much from lack of knowledge of pre-exponential factors in the models as in the accuracy of the experiments.

The structure and stability of thin epitaxial films have been investigated using molecular dynamics computer simulations. One issue of interest is the stability of a smooth film relative to 3-dimensional clusters. The simulation results show that the uniform film is never the lowest energy state for a system with finite misfit. However, the uniform film, produced in a layer-by-layer growth mode, can persist in a metastable state at substantial misfits, e.g. 10% at 1/2 the melting point. This is a result of the large nucleation barrier to the formation of clusters.

The second issue is the quality of the interface between the film and the substrate. At equilibrium, the critical thickness for the introduction of misfit dislocations is larger for films on the diamond cubic (100) substrate than on the (111), and differs from predictions based on continuum mechanics. We find that coherent films remain in metastable equilibrium far beyond the critical misfit calculated for full equilibrium, because a large free energy barrier inhibits the introduction of misfit dislocations.

We investigate the nucleation of epitaxial GaAs grown on vicinal Si(100) surfaces by Molecular Beam Epitaxy. We find that the vicinal surface structure is highly sensitive to the surface cleaning process, and that in-situ desorption of a volatile surface oxide at 900°C appears to result in significant step mobility, resulting in the formation of low angle surface facets. Nucleation of GaAs then occurs on these facets, raising the possibilty of influencing the heteroepitaxial nucleation and growth processes by influencing the substrate surface “template” at the surface cleaning stage.

The initial stages of GaAs growth on mirror-polished metal (W, Mo), semiconductor (Si, GaAs), and insulator (alumina) substrates have been experimentally studied in the organometallic chemical vapor deposition (OMCVD) process. Growth on the metallic substrates is often polycrystalline and nonepitaxial, while epitaxial GaAs layers with preferred orientations can be grown relatively easily on the nonmetallic substrates. The rate of nucleation of GaAs on all the substrates is found to increase with increasing arsine flow rate and with decreasing growth temperature. Our experimental results suggest that the GaAs nucleation process on all the substrates consists of an initial condensation and nucleation of gallium and subsequent formation of GaAs by arsenic supersaturation. In addition, following the nucleation of GaAs, a decrease in the substrate temperature is observed until further GaAs nucleation occurs, suggesting continued supersaturation. The nonepitaxial growth on the metallic substrates is attributed to the high rate of GaAs nucleation on these substrates.

The influences of implantation impurities, including BF2, B, F, As and P on the formation of epitaxial NiSi2 in nickel thin films on ion-implanted silicon have been investigated by transmission electron microscopy.

The presence of BF2, B, and F atoms was observed to promote the epitaxial growth of NiSi2 at low temperatures. Little or no effect on the formation of NiSi2 was found in samples implanted with As or P ions.

The results indicated that the influences of the implantation impurities are not likely to be electronic in origin. Good correlation, on the other hand, was found between the atomic size factor and resulting stress and NiSi2 epitaxy at low temperatures.

The initial stage of topotactic growth of Ni-Al spinel into Al2O3 has been examined using transmission electron microscopy. A new experimental approach to the study of solid-state reactions, which may be adapted for in-situ experiments for low-temperature systems, has been used in this study. In its present form, the technique involves heating a thin film of one oxide in the presence of a vapor of the second oxide. In the study of the growth characteristics of Ni-Al spinel phase, the orientation of the Al2O3 substrate has been found to influence greatly both the structural and morphological aspects of the spinel growth. In particular, the topotactic relationship between the spinel and the alumina are very different for (0001) and {1120} substrate orientations. The very early stage of the kinetics of the spinel growth, in which the length and the width of the spinel particles are only a few hundred angstroms, is illustrated with the results obtained from the re-heating experiments. The structure of the spinel-alumina interface has also been studied using high-resolution electron microscopy. These results are discussed in relation to the different models proposed for the spinel-alumina phase transformation.

The initial stages of interface formation are crucial for the growth of an overlayer, particularly for ultra-thin films. While growth rate, growth temperature and lattice match are important parameters in molecular beam epitaxy, growth of semiconductor heterostructures always begins on a reconstructed surface. The effect of this reconstruction has to be known before a complete understanding of the growth process is possible. Two examples will be discussed: (1) A necessary condition for perfect growth is the reordering of the substrate surface, i.e. those atoms at the interface which are originally displaced due to the reconstruction have to be brought back to bulk positions. Si(111)7×7 and Si(100)2×1 surfaces behave differently in this respect. The 2×1 orders upon deposition of Si, Ge or Sn at room-temperature whereas the 7×7 does not. This result will be correlated with the different classes of reconstruction to which these surfaces belong. The possible rela-tionship with the difference in substrate temperatures required to achieve homo-epitaxial growth on Si(100) and Si(111) surfaces will be discussed. (2) Perfect growth also relies on proper control of the growth-mode. As an example, the island-ing behavior found for deposition of Sn on Si will be discussed. While surface diffusion over macroscopic distances is observed, dramatic changes in the affinity for clustering occur depending on surface reconstruction and preparation, suggesting that clustering in this system may be controlled by appropriate choices of growth parameters.

Two distinctively different structures were found on CoSi2(111), the CoSi2-C and the CoSi2-S, which correspond to, respectively, an essentially bulk-terminated surface and one which consists of an additional Si double layer. Details of the two structures suggest new growth techniques for Si epitaxy. Single crystal Si layers with either identical orientation as the silicide or a 180°-rotated orientation can be fabricated by use of carefully designed Si templates.

The initial stage of Ge overgrowth on CaF2/Si structures was controlled by electron beam (e-beam) exposure through a room-temperature-deposited thin Ge layer on CaF2. It was found that the island growth of Ge was prevented in the e-beam exposed region and the crystalline quality and the surface flat-ness of the Ge film were much improved. From several experimental results, a growth model that e-beam dissociates the surface F atoms of CaF2 and improves the wettability between Ge and CaF2 is proposed.

A molecular dynamics simulation of the silicon adatom induced rearrangement of the silicon {100} symmetric dimer reconstructed surface has been performed. Surface diffusion is proposed to play a critical role in the reordering of this surface which leads to good epitaxy while it plays much less of a role in the reordering induced by an amorphous overlayer. These results are used to provide atomic-scale models which are consistent with high-energy ion channeling/blocking and LEED studies by Gossman and Feldman of the initial stages of silicon growth on this surface.

Computer simulations of III-V molecular beam epitaxy (MBE) show that surface reconstruction induced modulation of kinetic rates could give rise to ordering in alloys. Results are also presented for the possible influence of an external ion beam in achieving low temperature epitaxy as well as smoother growth front under usual conditions.

A new GaAs semiempirical potential energy function which utilizes both two-body and three-body contributions has been used to evaluate the surface energy and surface stress tensor for the (111) and (001)(As-terminated) surfaces as well as for ledges on the (001) surface. Both surface and ledge reconstruction patterns appear to be driven by stress tensor considerations and a long range ledge-ledge interaction arises via the stress tensor. The impact of this finding on two dimensional cluster formation and nucleation at the growing surface is very substantial.

The theory of Frank and van der Merwe (FM) in 1949 showed that a minimum energy criterion could explain the pseudomorphic growth of a deposit on a substrate of different lattice spacing and the subsequent relief of strain by misfit dislocations as the deposit thickness increases. Although the “equilibrium” theory is qualitatively correct, account must be taken of actual dislocation sources, which may be complex, and which may be more or less efficient for misfit relief than predicted by the FM model. Moreover, misfit dislocation sources may determine the morphology of the growing film, the interface topology and even the atomic structure of the deposit/substrate interface. These various roles of misfit dislocations are reviewed here with examples from work on metal/metal, semiconductor/semiconductor and metal/semiconductor systems.

A simple but general model is developed which describes the elastic relaxations in a misfitting epitaxial thin film on a rigid substrate. The film is treated as a two-dimensional, elastically anisotropic solid with an arbitrary orientation relationship to the underlying substrate. The periodic elastic displacement field, the misfit energy density, and the elastic energy density of relaxation are derived for relatively weak, periodic film-substrate interactions of a generalized form. The model is tested for a misfitting Ni-(111) monolayer in contact with a rigid Ag-(111) substrate, using computer simulations with the embedded atom method. The predicted displacement field has symmetry 3m and has a magnitude which is in close agreement with that obtained by a fully numerical solution. The dependence of the extent of relaxations on the form of the film-substrate interaction is discussed for the special case of unidirectional misfit.

We summarize high resolution transmission electron microscopy studies of interfaces with 33% misfit. We explain the existence of epitaxial interfaces for this systems by a geometrical argument similar to the 0-lattice models used to study high angle grain boundaries. Differences between systems very similar in structures are explained. We use the thus found epitaxial interfaces to build multilayered structures of the type metal/insulator/semiconductor.

Oriented deposits are commonplace in vapor deposited films. The origins of this behavior are not always clear. Various bcc metals (Fe, Nb, Mo, Ta, Cr) have been deposited onto single crystal fcc substrates (Au, Ni, MgO, Si, NaCI) which were selected with the intention of varying the character of the substrate-deposit interaction. The resulting structures have been characterized by transmission electron microscopy and grazing incidence x-ray diffraction. The observed variants of the cube-cube and Nishiyama-Wassermann orientation relationship can be understood in terms of minimization of misfit except when there is a weak interaction between substrate and deposit.

The initial stages of epitaxial silicide formation are described in terms of the nucleation properties. The reaction process is divided into the stages of interdiffusion, nucleation, and growth and coalescence. The aspects of the nucleation from the interdiffused regions are described. Results are presented for Pd and Pt deposited on clean Si surfaces. The initial thickness where silicide compound formation is observed is related to he critical cluster size of the nucleation model. The epitaxial morphology can be strongly influenced by heterogeneous nucleation involving the substrate interface, and this effect is demonstrated by reaction with metal overlayers less than the critical thickness.

In-situ observations of the growth of CoSi2 films by Co deposition in an ultrahigh vacuum transmission electron microscope are described. Results include the observation of intermediate epitaxial phases (both hexagonally distorted Co2Si and CoSi) and the observation that pinholes nucleate in CoSi2 films after growth in a metastable layer mode.

Atomically resolved and well ordered field ion images of silicon surfaces have been obtained for the first time when a carefully cleaned Si tip is annealed to 700 to 800°C for several minutes. Well ordered atomic structures develop on high index planes such as the (130), (135), (124), (234), (123), (113), (115) and (320). Some of them develop the (1×1) structure while others are reconstructed. Two structures coexist for the (130) plane. Twodimensional defects can be seen and their density is surprisingly high. We have also studied formation of thin layers of silicide on silicon surfaces.

The structure of the Au segregated Ni(110)-0.8%Au surface has been studied by scanning tunneling microscopy. The segregated Au layer forms a (7×4) structure with a c(2×4) subunit. At various coverages of Au overlayer, commensurate and incommensurate Au structures were observed. At a coverage of I monolayer, islands and facets assciated with incommensurate Au atoms were observed.