A review is presented in which existing theories of the formation of Schottky barriers are analyzed. The list includes macroscopic dielectric approaches and various microscopic quantum mechanical treatments. The central role of interface states and their different physical origins are assessed. Simple concepts, able to predict general trends in barrier heights, are examined along with detailed microscopic theories applied to individual contacts.

Over the past few years the electronic structure of transition metal-silicon and silicide-silicon interfaces (and of bulk silicide compounds) has been revealed for the first time, using surface spectroscopies (photoemission and Auger) and theoretical calculations. These investigations, which have included palladium, platinum, nickel, vanadium, chromium, molybdenum and tungsten, have elucidated the important role played by the chemical bond between the transition metal d and the Si 3p electrons. They have also shown the high chemical reactivity of the atomically clean transition metal-silicon interface, which leads to interfacial silicide formation at relatively low temperatures (i.e. markedly below those needed for bulk silicide formation on chemically cleaned silicon surfaces). As a result, silicide-like chemical bonding dominates the interface electronic structure of such contacts under a wide variety of conditions. Detailed comparisons of interface spectra (observed at low metal coverages) with those of the bulk silicide reaction products give strong evidence for additional electronic states of relatively high density (about 0.1 states per interface atom) which lie in or near the silicon band gap region. These states are believed to be true interface states associated with localized bonding configurations unique to the atoms at the interface, and they could explain the silicon-rich silicide composition of a thin (about 3–5 Å) interfacial region which has been observed by Auger composition analysis and suggested by chemical shifts in core and valence electron densities of states. Finally, metal atom diffusion into the silicon substrate has been observed in ion channeling studies and suggested from surface spectroscopy results; these impurity atoms should produce defect states localized near the interface and lying within the silicon band gap, although these electronic states have not yet been directly observed. As a whole these results present a fairly detailed picture of the electronic structure and chemistry of the silicide-silicon interface.However, correlations with the interface electrical properties are needed to ascertain which electronic features and chemical mechanisms in fact determine the Schottky barrier height of the contact.

A wide variety of the behaviors of single silicon adatoms on tungsten surfaces have been studied in the field ion microscope (FIM). Existing works are summarized, and some new results are presented. Diffusion parameters for single silicon atoms on the W{110} plane were found to be Ed=0.70±0.07eV and Do = 3.1 × 10−4 × 10±1.28 cm2 s−1, where Ed is the activation energy of diffusion and Do the diffusivity. Although chain-form atomic clusters were often formed when more than one adatom was present on a plane, the clusters were not stable during diffusion. Silicon atom transport on the W{110} plane is thus achieved dominantly by migration of single silicon atoms. The desorption field as a function of radial distance from the W{ 110} plane center was measured, and the data were related to the field distribution on the plane at a given applied voltage. The binding energy of silicon adatoms on the W{l10} plane was estimated from the desorption field. Relative pair energies for the Si-Si interaction on the W{110} plane for several bond lengths were obtained from the relative frequencies of observing these bonds in the FIM. The maximum binding occurred at a separation √2a, with the bond aligned in the [110] direction, the maximum antibinding occurred at a distance a in the [001] direction, and a weak binding occurred at 2a along the [001] direction; a was the substratelattice constant. Using the pair energy data, the relative binding energy in an adlayer for various adlayer structures was calculated. The[2√2/-√3×/√3]R35.26° superstructure, which is conventionally expressed as p(2 × 1), had the lowest energy. Our subsequent thermal equilibration experiments with deposited silicon adatoms at about 300 K confirmed the existence of such a structure. Since further heatings at about 300 K changed only the adlayer shape and composites, but did not change the structure of the superlattice, the observed superstructure was the equilibrium adlayer structure at a low degree of coverage at about 300 K. Monte Carlo computer simulations using the measured pair energies predicted the same equilibrium superstructure. In addition, many growth defects of the two-dimensional lattice, such as antiphase domains and stacking faults, were observed. The computer simulations also showed these and other defects. On the W{112} plane, no cooperative walk of silicon adatoms similar to that of metal adatoms was found. However, the staggered and straight bond configurations were observed much more often than that expected from a random distribution. When two silicon adatoms were deposited onto the same surface channel, they were mobile although the ,√3a bond separation seemed to occur more often. In addition, interactions of plane edges with silicon adatoms and some other observations are discussed.

We used UV and X-ray photoemission spectroscopies to probe the relation between the chemical and electronic structure at ultrahigh-vacuum-cleaved CdS-metal and CdSe-metal interfaces. When combined with current-voltage and capacitance-voltage studies of the same interfaces in ultrahigh vacuum, the experimental results indicate that partially dissociated cadmium cations, produced as a consequence of interfacial chemical reaction, may be the electrically active species giving rise to the observed Fermi level stabilization at these contacts. The extent of cation dissociation, a spectroscopically determined quantity, is shown to correlate inversely with the measured Schottky barrier height. An indirect and modified doping effect is suggested as one possible mechanism to explain this behavior. Features of interdiffusion, in particular regarding the interfacial distribution of dissociated cadmium, are also described.

Metal films such as gold, copper, nickel and palladium exhibit an interfacial intermixing reaction at room temperature with semiconductors with energy gaps Eg less than about 2.5 eV or dielectric constants ε larger than about 81. We have proposed a model2 of the triggering mechanism of this interfacial reaction based on the ability of a metal to screen coulombic interaction by its mobile free electrons. Such a screening may disturb the electron distribution responsible for the covalent bonding, and consequently it may make the semiconductor surface reactive towards the metal. In this respect, there must be a critical film thickness for the deposited metal film to behave as a true metal with sufficient mobile free electrons for the screening; therefore the reactivity of the semiconductor surface for the intermixing reaction must be dependent on the thickness of the deposited metal film. In favour of the screening model, we show that the reactivity of a silicon surface for intermixing or silicide formation with a gold or palladium film depends clearly on the film thickness.

The formation and structures of epitaxial CoSi2, NiSi2, Pd2 Si and PtSi films on silicon are reviewed. Polycrystalline films of reasonable epitaxial quality can be grown with sharp interfaces on Si(111) by conventional deposition and heating techniques. The interfaces on (100) substrates, however, are faceted. Perfect singlecrystal CoSi2 films can be grown on Si(111) substrates by ultrahigh vacuum techniques using conventional deposition and annealing at about 900°C and molecular beam epitaxy codeposition at about 600°C. The single-crystal films are rotated 180° with respect to the substrate. Novel defect arrays are observed at the interface. Resistivities in epitaxial films are lower than those reported in polycrystalline layers. High quality silicon films can be grown on the silicides to form semiconductor/metal/semiconductor heterostructures.

The interfaces of both epitaxial and non-epitaxial silicides and silicon were investigated by the direct lattice imaging method using cross-sectional samples. Non-epitaxial CoSi2 on silicon was observed to have a curved interface. Epitaxial CoSi2 , however, was found to be smooth within a facet. No evidence of an amorphous layer at the interface was obtained.

Epitaxial NiSi2 on Si(001) was found to be heavily faceted. The facets are on {111} and {100} planes with the former more frequently observed. The interface between Si(111) and NiSi2 is also faceted but less so than that for Si(001). The interface is very rough on a large scale. Straight boundary lines corresponding to faceted planes were observed which indicated that the interfaces on an atomic scale were quite smooth. Defect clusters and planar defects were also observed at the interfaces.

The epitaxial perfection and microstructure in thin CoSi2 (111) (and NiSi2) films on silicon has been examined using transmission electron microscopy, including high resolution cross section techniques. We find that, under ultrahigh vacuum preparation conditions only, genuine single-crystal epitaxy can occur for bothreacted and codeposited CoSi2 films. A unique interfacial defect structure is associated with this epitaxy, in which the silicide films are rotated through 180° about (111) with respect to the silicon substrates. Under less than perfectly clean conditions epitaxial perfection is never obtained for CoSi2. Moreover such perfection has never been attained under any conditions for NiSi2 on Si(111). By studying the evolution of epitaxy and defect structure in thin and thick films we have proposed a model to explain the unique quality of ultrahigh vacuum CoSi2 films. The model invokes a defect pinning mechanism to explain the dominance of the 180°-rotated epitaxy during silicide growth.

High resolution electron micrographs have been obtained from Pd2 Si-Si(111) interfaces viewed edge on and a detailed analysis of the interface structure has been carried out using computer simulations of the images. Since the stacking sequence of hexagonal Pd2Si is AαAαAα..., the question whether the interface plane of Pd2 Si is silicon rich, A, or silicon deficient, α, has been investigated. An advancing interface cannot be planar at all stages of growth; therefore the question of interface roughness along the direction of the electron beam and twinning orientations has been considered. This has been modeled by Fourier shift techniques incorporated into multislice dynamical theory and, of course, is complicated by the dynamical diffraction conditions of the structure and the contributions of high order Bragg reflections to image detail. It has been possible to access the image features present in Pd2 Si/Si micrographs and to point out the difficulties which are encountered not only in this system but in a wide variety of interface structures by high resolution electron microscopy. Perhaps the most important finding of this study is that the interface region contains a mixture of silicon-rich and silicon-deficient Pd2Si planes adjacent to the silicon side of the interface.

Radioactive 31Si (half-life, 2.62 h) was used as a marker to study Co2Si, CrSi2, TiSi2 and ZrSi2 formation. By marking the initial layer of silicide with radioactive silicon atoms and by measuring the activity profile in the silicide layer after further silicide formation, the dominant diffusing species and its mechanism of diffusion during the formation of these silicides could be determined. For Co2Si it was found that cobalt is the diffusing species, while disilicide formation was found to take place by silicon substitutional (vacancy) diffusion, with a high self-diffusion coefficient.

Transmission electron microscopy has been applied to study the formation and structure of epitaxial NiSi2 and CoSi2 thin films on silicon. Bright field and dark field imaging reveal the interface planes of faceted silicides through the strain contrast, analogous to the contrast of the precipitate-matrix interface of coherent or semicoherent precipitates. Superlattice dark field imaging depicts the distribution of twin-related and epitaxial silicides in these systems. { 111 } interfaces were found to be more prominent than {001} interfaces. Twin-related silicides were observed to cover more area on the substrate silicon than epitaxial silicides did.

In situ annealing of nickel and cobalt thin films on silicon provides a unique means of investigation of the transformation from polycrystalline to epitaxial silicides. The NiSi2 transformation was found to be very rapid at 820°C, whereas the CoSi2 transformation appeared to be very sluggish. Furnace annealing confirmed that only a small fraction of CoSi2 transforms to epitaxial CoSi2 after annealing at 850°C for 4h.

Diffraction contrast analysis has been applied to interfacial dislocations of epitaxial NiSi2/Si and CoSi2/Si systems. The dislocations were found to be of edge type with ⅙<112> and ½<110> Burgers' vectors. The average spacings are close to their respective theoretically predicted values.

Heteroepitaxial CaF2/Si and Si/CaF2/Si structures were prepared by conventional vacuum evaporation of CaF2 and silicon onto silicon substrates. The optimum conditions for obtaining good epitaxial films were investigated by changing the silicon substrate orientation, the film thickness and the substrate temperature during film deposition. From Rutherford backscattering and channelling spectroscopy it was found that CaF2 films with excellent film quality were obtained on Si(111), Si(110) and Si(100) substrates at substrate temperatures of 600– 800°C, 800°C and 500–600°C respectively. It was also found from Rutherford backscattering and channelling spectroscopy and from transmission electron microscopy that single-crystal silicon films are formed on a CaF2/Si(111) structure at a substrate temperature of 700°C. From measurements of the electrical properties of the top silicon film after the implantation of phosphorus ions at 2 ×1015 cm−2 and subsequent annealing at 750°C, an electron Hall mobility of 69cm2 V−1 s−1 was obtained.

The application of medium energy ion scattering in combination with channelling and blocking to the study of the initial stages of palladium siuicide formation is discussed. After a brief description of the experimental arrangement and method, the effects on the Rutherford backscattering spectra of depositing small quantities of palladium on clean Si(111) are reported. The uniformity and thermal stability of thin palladium silicide films grown at room temperature were measured. Finally, channelling and blocking results were used to carry out a structural analysis of thin epitaxial Pd2Si layers.

The domain structures of palladium and Pd2Si as well as their crystallographic relationship to the silicon substrates were determined on Si(111) and Si(100) samples by mapping X-ray diffraction pole figures. X-ray diffraction topography and rocking curve measurements were carried out for the silicon substrates in order to detect the presence of elastic and/or plastic deformation in the substrates caused by silicide formation. The stresses in the silicide films were determined from the bending of the silicon substrates using X-ray diffraction techniques.

The influence of electrically active n-type (75As) and p-type (11B) impurities on the solid phase epitaxial regrowth of ion-implanted amorphized Si<100> and Ge<100> has been studied for low temperature furnace annealing. Both types of impurity increase the rate of regrowth of both silicon and germanium at a concentration level of 1020 cm−3 . Above this level, 75As retards regrowth in germanium. In compensated surface layers, the regrowth rate slows down to the values observed in self-implanted or intrinsic crystals for both silicon and germanium. The results can be qualitatively explained in terms of electrically induced generation of point defects at the amorphous-crystalline interface.

High resolution ion channelling and reflection electron diffraction techniques have been used to examine details of epitaxial regrowth in Ar+-ion-implanted GaAs(100) at furnace anneal temperatures of 400°C or less. In particular, we have investigated the nature and extent of epitaxial regrowth during both isothermal and isochronal annealing for various implant energies and for implant doses above and below the amorphous threshold. Our results indicate the development of a nonplanar growth interface during annealing which may lead, ultimately, to complex near-surface crystallization processes. Consistently with our observations and recent results from other laboratories, we propose a model for the epitaxial regrowth of amorphous GaAs layers based upon non-uniform growth rates along the amorphous-crystalline interface which could arise from local stoichiometry imbalance.

High voltage electron microscopy of silicon and GaAs layers grown by liquid phase epitaxy shows that two types of growth step sources are present. These consist of single dislocations with or without a Burgers vector component parallel to the macroscopic growth direction (longitudinal or transverse step sources respectively). A simple model is used to illustrate in particular the efficient nucleation of growth steps at transverse sources. Suitably positioned dislocations create morphologically stable patterns of equidistant widely spaced surface steps with monatomic height on otherwise atomically flat surfaces. Such growth surfaces result in interfaces with minimum disorder in multilayer growth.

In the low temperature (5.5 K) photoluminescent spectra from (Al,Ga)As double heterostructures grown by liquid phase epitaxy (LPE) for laser devices, certain anomalous emission bands higher in energy than the active laye emission but lower in energy than the cladding layer emission are observed. The peak wavelengths and intensities of these anomalous emission bands vary not only from wafer to wafer but also within a given wafer indicating spatial localization of these regions. Photoluminescence measurements on cleaved edges of wafers and on taperetched wafers as well as double-crystal X-ray diffractometry measurements suggest that the anomalous emission bands originate from interfacial regions having different aluminum compositions which are formed between the active and cladding layers presumably as a result of melt carry-over during LPE growth. The integrated photoluminescent intensity of the anomalous emission bands decreases with temperature in the range 75–300 K by nearly two orders of magnitude suggesting the role of some non-radiative process. The spatially localized regions of different aluminum compositions could act as sources and/or sinks for point defects and thus could affect the reliability of devices. Further, their presence would also reflect nonuniformity in the aluminum composition in the active layer.

The solid state interfacial reaction between gold and GaAs, was found to result in small surface microstructural changes, whereas melting of the reaction product (tentatively identified as AuGa) and subsequent solidification to form the β phase resulted in aligned rectangular protrusions with sharp corners. The rectangular protrusions were found to be much finer in scale after heating in a vacuum than after heating in argon at 1 atm. High cooling rates (e.g. 40°C min−1) during solidification resulted in the formation of both rectangular protrusions and irregularly shaped protrusions with jagged boundaries and topography. The irregular protrusions were associated with a phase tentatively identified as Au2Ga. Rapid cooling at about 600°C min−1 during solidification resulted in the formation of the gold-rich terminal solid solution (α phase) which appeared as aligned rectangular protrusions with rounded corners.

The characterization of surface structural defects by low energy electron diffraction is discussed. The identification of crystal mosaic structure, random strain, antiphase domains and steps is emphasized, and it is shown that they produce characteristic effects in the angular distribution of intensity in diffracted beams. Analytical models are summarized. They are also applicable to other surfacesensitive diffraction techniques such as reflection high energy electron diffraction and grazing angle X-ray diffraction. The quantitative analysis of these several types of defects is illustrated by measurements on the surface of an epitaxially grown Ag(111) film and on a sputter-etched and annealed GaAs(110) surface.