We have examined a number of MBE-grown (001) silicon thin films grown on (1012) sapphire substrates using transmission electron microscopy. We have found that for silicon films less than 0.55μm thick, microtwins are very much in evidence. For silicon films greater than 700nm thick, however, dislocations, rather than microtwins, are the predominant defect. It is our conjecture that the generation of dislocations, and the associated disappearance of microtwins in thicker MBE-grown SOS films, is analogous to the generation of misfit dislocations in silicon-germanium films grown on silicon or germanium substrates by MBE. The persistence of microtwins in CVD-grown SOS, as opposed to MBE-grown SOS, can be understood in terms of Dodson and Tsao's formulation of the kinetics of plastic flow in thin germanium-silicon films.

The modification of surface mass transport rates due to the presence of a divalent dopant in alumina has been studied. The composition of the prismatic (1010) surface of sapphire, doped with MgO by two different techniques, has been monitored with Auger electron spectroscopy. Mass transport at these doped surfaces has been measured by observing the rate of decay of a periodic surface profile. Short wavelength patterns were used so that the transport process would be dominated by solid state diffusion. The analysis assumes an isotropic surface free energy, the implications of which are discussed. It was found that during annealing in air, both surface and bulk diffusion mechanisms contribute to ionic mobility. The effect of Mg ions seems to depend on its thermodynamic activity. When the surface coverage is low, up to about 20%, the effect of the dopant is not significant. When the surface concentration of Mg increases to that corresponding to spinel (about 33%), the mass transport rate increases significantly. From these observations, it is implied that a second phase may be required to markedly enhance the surface mass transport rate of alumina in air. The results obtained in this study have been critically analyzed and are discussed in light of other data for mass transport in alumina.

In this investigation, we report on detailed microstructural characteristics of LPCVD WSi 2 films deposited on doped/undoped polysilicon and subjected to boron and phosphorus implantation and implant activation at temperatures up to 1100° C. The polysilicon films were doped with phosphorus using a POC13 source to obtain a sheet resistance of ∼ 25 Ω/ and the films were coated with about 300nm of Si rich (Si:W atomic ratio 2.6) LPCVD WSi2. The microstructures of as-deposited and high temperature annealed polycide films were studied in detail using TEM. The electrical sheet resistances were measured using four-point probes. The initial polysilicon doping level, silicide annealing conditions, subsequent implant into the polycide, and implant activation all had profound effects on the microstructural characteristics of both the silicide and polysilicon.

High spatial resolution characterization of quantum wells has been performed by monochromatic cathodoluminescence imaging and by large angle convergent beam electron diffraction. The satellite lines to sublattice reflections which have been interpreted previously in terms of kinematical diffraction theory are here explored in detail by dynamical diffraction theory.

The large angle method of convergent beam electron diffraction (LAMED) has been used to examine multi-quantum well (M1W) and single quantum well (SQW) samples of AlGaAs/GaAs and InP/InGaAs viewed approximately along the growth direction. The method combines an image of the specimen with the rocking curve, typically over about 6° for a selected reflection. LACBED patterns fran MQW samples show many orders of superlattice satellite reflections and can be used to profile compositions in both periodic and less regular structures. For SQW samples, we can measure the local quantum well thickness to near-monolayer precision.

High resolution cathodoluminescence microscopy (CL) under a magnetic field is proposed and used to study the interface structure of GaAs-AlGaAs single quantum wells. This principle called, Lorentz-force Carrier-confinement (LC), successfully improves the resolution by up to a few thousand Angstrom. Periodic light and dark stripe patterns are observed in LC-CL monochromatic images of the single quantum wells on the intentionally-tilted substrates. It is shown that the stripes represent stair-like stepped interfaces and that the observed terraces are extended and the steps are clustered. Based on the results, a new interface structure model is proposed, that takes the tilting into account.

The atomic perfection of interfaces and local impurity distribution are important parameters in the control of low dimensional semiconductor systems. Over the past few years photoluminescence excitation spectroscopy has proved to be of immense value in the detailed characterisation of such systems. This is especially the case for the GaAs/AIGaAs material technology in which tunable dye lasers can readily supply the variable wavelength excitation source needed for such work.

However this form of characterisation is much more difficult for semiconductors with smaller energy gaps e.g. the ternary InGaAs or many materials in the important antimonide based system. No readily available dye lasers exist which can be used to resonantly excite the semiconductor.

We have overcome this difficulty by using Fourier Transform Optics to source the excitation. We have successfully observed excitation spectra from InGaAs/InP and InGaAs/GaAs quantum well systems. The multiplex nature and high optical throughput of the Fourier Transform spectrometer yield high quality spectra with short measurement times.

In this paper we review the experimental technique and demonstrate its enormous potential for “low” gap semiconductors. We illustrate the technique by describing its application to various quantum well systems.

Low solute tetragonal (t) ZrO2 precipitates within a high solute cubic (c) ZrO2 matrix of a Zr-rich CaO-MgO-ZrO2 alloy were studied via high resolution electron microscopy (HREM). The precipitates underwent displacive thin foil transformations to orthorhombic (o) or monoclinic (m) symmetry during sample preparation or observation in the HREM. The orthorhombic phase grew in a facetted manner from a relatively equiaxed nucleus, consuming entire t-ZrO2 variants in multi-variant precipitates. Monoclinic laths nucleated it t/c or t/t interfaces and rapidly grew across portions of the parent phase. The crystallography of these transformations is discussed.

A review is given of current efforts to determine the structure of grain boundaries using X-ray diffraction. The general distribution of the diffracted intensity from the thin boundary region in reciprocal space is first described, and then some of the special experimental difficulties with measuring it quantitatively are discussed. These include, weak scattering and poor signal-to-noise ratio, preparation of sufficiently high quality bicrystal specimens, the presence of unwanted allowed and forbidden lattice reflections, double diffraction from the adjoining crystals, and limitations imposed by the scattering geometry. Various strategies for determining the boundary structure are then described. These include the measurement of either relative or absolute structure factors and the determination of structures based on either structure factor measurements alone or the combined use of structure factor measurements and computer modeling. The advantages of measuring absolute structure factors rather than relative structure factors when only a limited number of structure factors is measured is demonstrated. This situation may occur because of the experimental difficulties listed above. Current work, consisting of studies of absolute and/or relative structure factors of a range of [001] twist boundary structures in gold, is reviewed. Results obtained by the authors represent the first cases where good agreement has been obtained between boundary structures determined by X-ray diffraction and calculation using a physical model (i.e., the Embedded Atom Model). The large displacement model derived by Fitzsimmons and Sass for the Σ5 structure on the basis of relative structure factors is found to be incorrect. It is concluded that future progress will depend to a great extent upon the preparation of improved bicrystal specimens and the measurement of larger numbers of boundary structure factors.

The atomic structure and vibrational mean square displacements of large angle grain boundaries in Au are calculated and compared with experiment. The calculations use the Embedded Atom Method in conjunction with energy minimization and Monte Carlo computer simulation techniques. The computed structures are in good agreement with the experimental results. For tilt boundaries it is found that asymmetric boundaries can have energies comparable to symmetric boundaries. The vibrational mean square displacements of atoms at the boundary are larger than in the bulk by roughly 50%. However, significant anisotropy and site dependence is present.

Grain boundary analysis with the help of atom-probe field-ion microscopy (APFIM) needs an aimed tip-preparation technique unless the grain size of the material is very small. With the help of a dedicated FIM-tipholder for the TEM the distance grain boundary-apex is determined and reduced by a controlled backpolishing method to make the grain boundary accessible to APFIM analysis. Tilt angle and tilt axis of the grain boundary were determined analyzing the Kikuchi diagrams of the two grains. In nickel 1 at% carbon some atoms image “brighter” than others. With the help of the drop of the detector current it can be shown that these atoms are carbon. With the help of this feature the distribution of the carbon atoms in the matrix can directly be determined in the FIM. A segregation of carbon at grain boundaries was detected with the atomprobe.

Among the two basic poly-para-xylylene (PPX) crystal habits corresponding to the well known alpha and beta polymorphs, various forms of crystal/crystal associations have been observed and explained in term of three-dimensional structures. High resolution imaging and electron diffraction experiments of lamellar crystals grown from a 0.01% wt solution in 1-methylnaphthalene using the self seeding technique show strong evidence for a threedimensional “roof-like” model for the alpha/alpha twinned crystals. This model is based on the presence of tilted chains in the alpha lamellar crystals. Furthermore, observation of various forms of alpha/alpha, alpha/beta and beta/beta lamellar crystals allows a generalization of this model and a more fundamental explanation for their origin: crystal coalescence during growth in solution.

Slip patterns around <110> Knoop indentations on <001> GaAs were investigated by cross-sectioning, etching and TEM. The roles of As(g) and Ga(g) perfect and partial dislocations in different parts of the “plastic zone” were investigated. A dislocation free zone was found under indentations.

Methods suitable for measuring the extent of planar faulting in the framework structures of zeolites are considered. A simple method is outlined for simulating diffracted intensities from crystals containing coherent planar defects. The algorithm employed efficiently exploits the self-similar properties of extended faulting patterns. The simplicity of this recursion method makes it suitable for complex crystal structures such as those of zeolites, and its utility is illustrated by a study of faulting in zeolite beta. Powder diffraction pattern simulations, which are in excellent agreement with observed data, indicate a fault probability in the chiral zeolite beta stacking sequence of α ≈0.56.

A GaAs/AlxGal−xAs multiple quantum well structure was examined using low temperature (35K) cathodoluminescence in a transmission electron microscope. In this sample, stacking faults were found to occur in two different forms. The intensity of the dominant n=l(e-hh) luminescence was observed to decrease in the immediate vicinity of either form. In addition, stacking faults gave rise to several new emission features on the low energy side of the main peak. We discuss differences in cathodoluminescence spectra obtained from the two stacking fault configurations and present evidence to suggest that these features may be due to extrinsic luminescence from impurities associated with the stacking faults.

The nature and distribution of precipitates in Fe-doped InP crystals grown by metalorganic vapor phase epitaxy at 650°C are studied by transmission electron microscopy. The precipitates are spherical in shape, 4-20 nm in diameter, and uniformly distributed in the crystals. They are coherent FeP precipitates with certain orientation relationship to the InP. The precipitates are observed in crystals with Fe-doping gas flow rates of more than 20 ml/min. Their density increases from 8 × 1012 to 1 × 1014 cm-3, as the gas flow rate increases from 20-100 ml/min, while the size is almost constant. In crystals doped with less than 5 ml/min, precipitates are not observed. Crystals grown with a gas flow rate of 5 ml/min contain a Fe-concentration of about 1017 cm−3 which is the solubility limit of Fe in InP at 650°C. From these results, it is strongly suggested that when the doped Fe concentration exceeds the solubility limit, excess Fe atoms tend to condense in the matrix forming FeP precipitates.

Interfacial reactions between Ti-Ta alloy films and single crystal Si have been studied by Auger electron spectroscopy, transmission electron microscopy and X-ray diffraction. It has been found that only binary compounds, silicides of each of the metals, and not any ternary compounds, are formed. The reaction products at the early stages of the reaction are Ta5Si3, TaSi2 and the metastable (C49 structure) TiSi2. The final stage of the reaction consists of a uniform intermixing of the disilicides, TiSi2 and TaSi2. Alloying causes a shift in the temperature of silicide formation to intermediate values between those for pure Ti and Ta (500 and 700°C, respectively), and this value increases with Ta content in the alloy film.

By studying InP epitaxies it has been found that optimum growth conditions are not achieved under low growth rate regimes, which, on the contrary, yield a defect density nearly as high as under high growth rate conditions, due to the creation of a very high density of point defects from which dislocation loops originate. Such loops are generated not only at the substrate-epilayer interface but also throughout the layer and depend on the supersaturation of the feeding phases. Under conditions yielding high dislocation density both layer and hillock growth mainly occur by spiral growth.

Quantitative energy-dependent EBIC measurements have been used to calibrate the photoetching rate in HF-CrO3 aqueous solutions (DSL method: Diluted Sirtl-like etching with the use of Light) as a function of dopant concentration in LEC grown n-type GaAs samples containing growth striations. The relative DSL etching rate depends on the width of the surface depletion region associated with the semiconductor-etching solution interface, i. e. greater etch rates correspond to smaller dopant concentrations. These results are in agreement with the electrochemical model of GaAs etching in the DSL etching system.

The influence of Cu or Pd contamination on the integrity of thin SiO2 layers was studied on (100) and (111) Si substrates. Wafers were contaminated intentionally on their backsides and indiffusion of the impurities was carried out at 1200° C or 900°C by rapid thermal annealing. Both, electrical tests and cross-sectional transmission electron microscopy were applied to investigate the failure mechanisms of the oxide. Curich silicide particles as well as Pd-Si precipitates were observed at the SiO2 interface. They result in cracking and bending of the oxide film or reducing the oxide thickness. These oxide failures could be explained by a precipitation process which is associated with the emission of Si self-interstitials.