In this growth process a new strain relief mechanism operates, whereby the SiGe epitaxial layer relaxes without the generation of threading dislocations within the SiGe layer. This is achieved by depositing SiGe on an ultrathin Silicon On Insulator, SOl, substrate with a superficial silicon thickness less than the SiGe layer thickness. Initially, the thin Si layer is put under tension due to an equalization of the strain between the Si and SiGe layers. Thereafter, the strain created in the thin Si layer relaxes by plastic deformation. Since the dislocations are formed and glide in the thin Si layer, no threading dislocation is ever introduced into the upper SiGe material, which appeared dislocation free to the limit of the cross sectional Transmission Electron Microscopy (TEM) analysis. We thus have a method for producing very low dislocation, relaxed SiGe films with the additional benefit of an SO substrate. This buffer structure is significantly less than a micrometer in thickness and offers distinct advantages over the thick SiGe buffer layers presently in use.

We present an x-ray diffraction study of the variation of the tilt angle between a relaxed Si1−xGex epitaxial layer and the Si (001) substrate. Such measurements provide the basis for a new method to determine the nucleation activation energy of misfit dislocations. We show that the nucleation activation energy for 60° dislocations in the case of the multiplication mechanism observed in graded SiGe layers grown by UHV-CVD at low temperature is 4 eV.

When Ge is grown epitaxially on Si(001), the 4% mismatch between the lattice parameters of Ge and Si can produce a regular two-dimensional grid of (a/2) [1,±1,0] edge dislocations at the interface, a checkerboard with a spacing of ∼ 100 Å. We have performed classical molecular dynamical simulations of this checkerboard in large microcrystals. The results show the expected 5-fold plus 7-fold ring structure at the cores of the individual dislocations, and a new closed symmetric structure of 18 atoms at their intersections. Tetrahedral coordination is everywhere retained, with relatively small changes in the bond lengths of less than 10% and in the bond angles of less than 25%. The energetics and dislocation offset of the system are explored for the Stillinger-Weber and Tersoff potentials.

Epitaxial SiGe/Si layers are being extensively investigated for use in base regions of high-speed heterojunction bipolar-transistors (HBTs). Extended defects can be formed in SiGe/Si layers by ion-implantation. Defects, once formed in the layers, can negatively impact electrical performance and also future reliability of the HBTs. The present study investigates the interaction between selective-implant damage and strained SiGe/Si layers of sub-critical thickness. Implant-damage is observed to form dislocation-sources at the edges of implanted regions in SiGe/Si heterolayers. The dislocation sources produce glide dislocation loops. Segments of these loops glide down to SiGe/Si interfaces causing misfit dislocations to arise at interfaces in the heterolayers. Misfitdislocations are formed in directions parallel to and perpendicular to the <110> edge of the implanted region. Dislocations propagate out to a distance of ∼100-150 nm past the edge of the implant in the case of Si0.9Ge0.1/Si layers of sub-critical thickness. The origin and behavior of these defects is discussed.

The accommodation of lattice mismatch is studied in Ge0.15Si0.85 layers grown epitaxially on {111}-oriented Si substrates by chemical vapor deposition (CVD) at 1100°C. Weak beam dark field microscopy reveals a regular misfit dislocation network, which resembles the honeycomb network of edge-type dislocations anticipated by the O-lattice theory. In contrast to the latter, however, the real network exhibits extended nodes where the misfit dislocations dissociate into misfit partial dislocations. Between the partials, high resolution transmission electron microscopy (HRTEM) reveals intrinsic and extrinsic stacking faults. Owing to the presence of these stacking faults, three different atomistic structures of the GeSi/Si interface coexist and compete for the interfacial area according to their energy. The observed configuration is shown to minimize the total energy of the interface.

The present paper examines conditions which can lead to generation or blocking of structural defects as observed in the deposition of the Si-Si1−x Gex heterostructures prepared by conventional AP CVD. The lowering of the epitaxial temperature can lead to a breakdown of the epitaxial growth and to a generation of defects such as dendritically grown polycrystalline inclusions. Their generation depends on the moisture and oxygen content in the gas mixture. Fast cooling during growth of the underlying buffer layer leads to defect growth interruption and is attributed to the influence of hydrogen adsorption / desorption. A similar defect growth blocking has been found by addition of diborane to the gas mixture during CVD deposition.

A new strain relief mechanism was observed in thin films of Si0.85Ge0.15 grown by RTCVD (rapid thermal chemical vapor deposition) on patterned (001) and planar (111) Si substrates. The (001) Si substrates were lithographically patterned and anisotropically etched to produce a line pattern of V-shaped grooves running in the [110] direction where the walls of the grooves were the {111} crystal planes lying in the [110] zone. Cross-section and plan-view TEM (transmission electron microscopy) studies revealed the presence of open ended stacking fault tetrahedra in strained-layer Si0.85Ge0.15 grown both on (111) Si wafers and the {111} sidewalls of the patterned (001) Si wafers. No defects were observed in the (001) portions of the films grown on non-planar substrates.

Polycrystalline silicon-germanium alloys with compositions between 5 and 30% Ge were wet oxidized in pyrogenic steam with TCA at 700°C. TEM showed that the columnar structure and grain size, ∼500Å, were independent of composition and did not change during oxidation. Both energy dispersive x-ray spectroscopy (EDS) in a UHV STEM and RBS indicate that for low Ge containing films (≤20% Ge), Ge is rejected from the growing oxide, whereas for films with a greater Ge concentration, some Ge is incorporated in the oxide. In low Ge containing alloys, the average Ge profile was determined by RBS and the local Ge distribution was determined by EDS from the oxide interface to about 500Å deep into the alloy. A Ge rich layer was observed at the oxide/poly-SiGe interface. The Ge concentration at the grain boundaries was found to be at least three times higher than that in the grains. The mean diffusion length of Ge along the grain boundaries, measured in cross-section by EDS, was about 300Å, corresponding to an effective diffusivity of about 6×10−16 cm2/s. Analysis of RBS profiles yielded an average effective diffusivity of ∼1.3×10−14 cm2/s. Both diffusivity values suggest that Ge rejected from the oxide growth front diffuses along grain boundaries since the bulk diffusivity of Ge in Si at 700°C is on the order of 10−22 cm2s/s.

We have used high-resolution electron microscopy to examine structural changes in the amorphous - crystalline interfacial region in Ge implanted silicon due to annealing at 450°C. We observe the removal of strain in the crystalline part of the interface and a strong decrease of interfacial roughness due to annealing. In addition, we observe the nucleation of interstitial type extended defects in the crystalline region beneath the interface. These grow into {113} - stacking faults during subsequent solid-phase epitaxial regrowth. By measuring the size and density distribution of these defects we estimate the concentration profile of silicon self-interstitials in the crystalline substrate of implanted silicon.

Selective epitaxial growth has been used to produce electronically isolated devices on patterned-oxide Si substrates. The oxide / silicon interfaces in such materials are often associated with regions of poor device performance. In this study, the extended defects at the interfacial regions are examined by transmission electron microscopy, and the defects observed are correlated to electronic behavior of diodes fabricated in the selectively grown Si region. Process modifications were made to reduce the density of these defects. The nature of the successful techniques for defect reduction suggests that a large portion of the defects were due to thermal expansion mismatch, and may be avoidable.

The advantages and disadvantages of in-situ straining using both synchrotron x-ray topography and transmission electron microscopy for examining dislocation/grain boundary interactions are compared and examples given of the use of each technique. For x-ray topography, studies on ice polycrystals are discussed. Ice is well-suited for x-ray topographic studies since it has both low absorption and can be produced with a low dislocation density. Stress concentrations have been observed at grain boundaries in ice which are partially relieved by generation of 1/3<1120> dislocations. Interestingly, grain boundary generation of dislocations completely overwhelms lattice generation mechanisms. Examples of transmission electron microscope in-situ straining studies include dislocation/grain boundary interactions in L12-structured and B2-structured intermetallics. Slip transmission across grain boundaries by dislocations gliding ahead of an advancing crack is a principal feature of these studies. A significant advantage of the such studies is their inherently high resolution. However, the dislocation behavior is dominated by the inherent thinness of the specimens.

The evolution under deformation of symmetrical GBs in Si and Ge has been studied both by in-situ TEM and HREM. The influence of the strain conditions (tension or compression) on the final structure through the entrance of different types of dislocations will be discussed. The influence of the deformation temperature on the multiplicity of the structures has been observed in a particular range of misorientation angles.

Various morphologies are observed where twins meet grain boundaries in YBa2Cu3O7−δ. Twins may be “correlated” at the boundary (i.e. twin boundaries from one grain may meet a twin boundary from the other grain in a quadruple junction) and the twins may be narrowed or “constricted” at the boundary. These effects are determined by the interfacial energy. We estimate the energy of the various interfaces by determining the dislocation arrays they contain, using the constrained coincidence site lattice (CCSL) model and Bollmann's O2-lattice formalism. Our approach indicates that there are significant changes in the energy of the interfaces and is thus able to explain the variety of observed morphologies.

The macroscopic properties of many materials are controlled by the structure and chemistry at grain boundaries. A basic understanding of the structure-property relationship requires a technique which probes both composition and chemical bonding on an atomic scale. The high-resolution Z-contrast imaging technique in the scanning transmission electron microscope (STEM) forms an incoherent image in which changes in atomic structure and composition can be interpreted intuitively. This direct image allows the electron probe to be positioned over individual atomic columns for parallel detection electron energy loss spectroscopy (EELS) at a spatial resolution approaching 0.22nm. In this paper we have combined the structural information available in the Z-contrast images with the bonding information obtained from the fine structure within the EELS edges to determine the grain boundary structure in a SrTiO3 bicrystal.

The equilibrium atomic structure of the ∑=15 36.9° (210)[001] tilt boundary in rutile (TiO2) has been computed using an ionic shell model and compared to a high-resolution electron microscope image of the same boundary using image simulation. The lowest energy structure of the boundary is characterized by an in-plane translation of a/6[120] relative to the mirrorsymmetric CSL configuration in agreement with the electron microscope observations.

While crystallite-rotation experiments of the type pioneered by Gleiter and his co-workers have provided a great deal of information about the variation of grain boundary energy with crystal misorientation, certain aspects of the experiments still remain puzzling. Notably, crystal rotation rates do not follow the form predicted by consideration of the Read-Shockley equation; and Chan and Balluffi have shown that in some cases certain crystallites are able to escape from the energy cusps that trap others [5]. In this paper we re-examine the crystallite rotation mechanism and rotation rates from the perspective of the interactions of grain boundary dislocations with the free surfaces that terminate the grain boundary.

The CCSL approach for modelling dislocation configurations in high-angle grain boundaries of hexagonal materials has been re-examined. It has been shown that this model implies the existence of long range strain in both crystals and that the strain energy which would be involved makes the formation of such structures unlikely. It has been proposed that alternative methods could be developed which make no assumptions about the nature of any reference structure adopted by a boundary. If the Burgers vectors of interfacial dislocations are obtained using Pond's symmetry theory of interfacial defects and the arrangement of these defects determined using an ‘0’ lattice approach then the structure of grain boundaries may be modelled irrespective of the nature of the “reference structures”.

A mathematical model (2D) of dislocation generation at cracks on interfaces is presented, which takes into account the role of slip processes on several slip planes in the vicinity of a crack. The work investigates the effects of other incipient dislocations on the nucleation and emission of the primary dislocation that emits first and is responsible for crack-tip blunting on atomic length scales. The modeling makes use of the recently-developed Peierls-Nabarro framework for dislocation nucleation. It is found that there is a moderate increase in the critical load necessary to emit a dislocation, when incipient slip activity is allowed to occur on the prolongation of the crack plane. Furthermore, the slip at the tip, the quantity which characterizes to what extent an incipient dislocation forms before it emits, decreases when the dual slip-plane model is used. Implications for the ductile versus brittle response of Ni are discussed.

Information on the mechanisms of slip transfer across grain boundaries in an HCP α-Ti alloy has been obtained from deformation experiments performed In situ in the transmission electron microscope. Initially, lattice dislocations are accommodated within the grain boundary until a critical local dislocation density is reached. The boundary then responds by activating slip in the adjoining grain on the slip system experiencing the highest local resolved shear stress and producing the residual grain-boundary dislocation with the smallest Burgers vector. Slip on secondary slip systems may be initiated provided they reduce the magnitude of the Burgers vector of, or eliminate, the residual grainboundary dislocation. The selection rules used to predict the slip system activated by the grain boundary are the same as apply in ordered and disordered FCC materials.

We present a study of the structure and migration mechanisms of the Σ=3 {112} boundary in aluminum. Measurements of the static boundary, using both High Resolution Transmission Electron Microscopy (HRTEM) and the α-fringe technique, are consistent with the lowest energy boundary structure that is predicted by the Embedded Atom Method (EAM). HRTEM observations of the migrating boundary are also presented. The boundary is mobile and upon translation produces coherent {111} twin boundary steps. The atomistic processes that are involved in this migration are considered in terms of the required twinning shears and the results of molecular dynamics simulations using the EAM. The initial nucleation of steps in the boundary appears to be electron beam induced.