The effects of implanting Sb into as implanted and annealed ion beam synthesised (IBS) CoSi2 layers and their subsequent behaviour during sequential isochronal annealing is investigated. For both the as implanted and annealed samples there is some Sb diffusion, towards the CoSi2 interfaces, during annealing. The extent of this diffusion is greater for the as implanted sample than for that which was annealed at 600°C/1 h + 1000°C/30 min prior to Sb implantation. During sequential annealing the Sb initially moves onto substitutional sites (below 400°C). However, once it starts to diffuse, it is found predominantly on interstitial sites. After annealing at 1000°C the crystal quality of the Sb doped layers is similar, if not superior, to that found in undoped layers, indicating that the damage introduced during the Sb implant has been entirely removed. In addition, fewer defects are observed in the silicon adjacent to the lower CoSi2/Si interface in the Sb implanted samples.

A theoretical model for the process of annealing of a layer of interstitial dislocation loops parallel and close to a planar specimen surface, such as is produced by pre-amorphization of silicon by implantation with germanium, is presented. On the assumption that the kinetics are determined by diffusion of either interstitials or vacancies and using a mean-field approximation, an equation is derived for the rate of change of each loop radius in terms of the size distribution of loops. The results of a numerical solution of this model are compared with transmission electron microscope measurements of mean loop sizes and densities in a set of specimens where the depth of loops from the surface and the annealing time were systematically varied. The results of the model are in reasonable agreement with experiment if a suitable initial loop size distribution is chosen. The model predicts the observation that removal of the loops is easier for smaller loop depths. The combined effects of loop coarsening and loss of interstitial atoms by transport to the surface are reproduced, the latter process being more important for smaller loop depths. The results imply that the initial process of nucleation of the loops has a strong influence on the subsequent behaviour during annealing. Suggestions are made as to how the loops might be avoided in device processes.

The effects of implanting dopants such as Sb and In into as implanted and annealed ion beam synthesized (IBS) CoSi2/Si layers are studied. The Sb and In ions were implanted into the centre of the IBS layers and isochronally annealed between 200 and 1000°C for 15 min. Using the results from Rutherford backscattering spectrometry (RBS) and cross sectional transmission electron microscopy (XTEM) it is found that in all cases there is some dopant segregation, towards the interfaces with annealing. The extent of the segregation is, however, determined by the redistribution kinetics of the different dopant species, being greater for Sb than In. For both Sb and In less diffusion is observed if the dopant is implanted into an annealed CoSi2 structure. Changes in the crystallinity of the CoSi2 layer can be correlated with the redistribution of the dopant, however, after annealing the crystallinity is similar to that for undoped layers, indicating that the damage introduced into the structure by dopant implantation can be entirely removed.

Buried layers of α and β iron disilicide have been fabricated by ion beam synthesis (IBS) in silicon (100) substrates. An implantation energy of 200 keV, to a dose of 4 × 1017 ions cm−2 is compared with an energy of 2 MeV, to a dose of 1 × 1018 ions cm−2. The effect of implantation energy and annealing conditions on the layer formation was studied by Rutherford backscattering spectroscopy, cross-sectional transmission electron microscopy, photoluminescence (PL) and sheet resistance measurements. A PL signal with a full width half maximum (FWHM) value of 4 meV was obtained after a 900 °C anneal for 18 h.

We have carried out photoluminescence (PL) on thin layers of silicon which have been pre-amorphised with a 400 keV germanium implant, and then subjected to rapid thermal annealing. TEM studies have established that the majority of extended defects were loops close to the original amorphous-crystalline boundary and that ∼ 90% of them are Frank type, 13a 〈111〉 dislocations. A dislocation related peak at 0.875 eV (the D2 line) occurred in most luminescence spectra. We have examined this as a function of anneal time and ambient atmosphere and we find that the D2 line strongly correlates with the Frank dislocation loops. This demonstrates that PL can be used as a diagnostic tool to detect Frank loops present in very thin silicon layers after ion implantation and annealing.

Structures of cobalt disilicide layers fabricated by ion beam synthesis on (001) silicon wafers have been studied by cross-sectional transmission electron microscopy. Implantation at 350 °C with doses of 5 and 7×1017 cm−2 of 200 keV Co+ ions was used, followed by rapid thermal annealing. For the as-implanted wafer with the lower dose, a CoSi2 layer in a parallel (A-type) epitaxial orientation was formed, and below this there were CoSi2 precipitates, some in twinned (B-type) orientations, and {113} defects. With the higher dose, polycrystalline CoSi was also present at the surface and there was substantial surface roughening. For the annealed wafers, as the annealing temperature increased from 700 to 1100 °C, the CoSi2 layer progressively increased in thickness, and the CoSi at the surface of the CoSi2 layer was eliminated. In the silicon beneath the silicide layer, the CoSi2 precipitates were greatly reduced in number and the {113} defects were eliminated.

IBS of buried α and β iron suicide layers was achieved by the implantation of 2 MeV 56Fe+ ions into (100) single crystal silicon substrates over a dose range of 3 × 1017 to 1 × 1018 cm−2followed by a high temperature anneal. No photoluminescence was observed from the as-implanted samples which contained a discontinuous layer of βFeSi2 precipitates approximately 1.5 μm below the silicon surface. Upon annealing at 700°C, a 200 nm polycrystalline βFeSi2 layer was formed which gave a PL signal centred at 1.55 μm. After a 900°C anneal, the layer transformed to αFeSix with a resistivity of approximately 280μΩcm.

Melt-spun ribbons of composition SmFe11Ti were annealed at temperatures in the range 650-825-degrees-C for 10-30 min. The maximum coercivity H(c) = 0.56 T, was obtained by annealing at 800-degrees-C for 20 min. The microstructures were determined by TEM. Lorentz electron microscopy was used to image domain walls. The specimens consist of single-domain grains of the ThMn12 structure Sm(Fe11Ti) phase, plus, in some cases, alpha-Fe. The reductions in H(c) in the under- and over-annealed specimens are attributed to irregularities in grain boundaries and growth of alpha-Fe grains, respectively.

Silicon specimens were implanted with 1 x 10(15) cm-2 of 400 keV Ge-70+ ions. Controlled etching was used to reduce the thickness of the resulting amorphous layer by various amounts. Rapid thermal annealing at 1100-degrees-C in nitrogen produced dislocation loops. TEM showed that the number of interstitial atoms per unit area in the loops decreased with increasing annealing time until the loops disappeared; this occurred more rapidly as the amorphous layer was made thinner. A model based on diffusion of self-interstitials predicts the timescale of this process to within a factor of about 2.

Transmission electron microscope studies have been made of (100) silicon wafers implanted at 500 °C with 200-keV 14N+ ions to doses of either 0.25, 0.75, or 1.4×1018 cm−2. For all of these specimens, the as-implanted wafers contained a buried amorphous layer with a damaged upper single-crystal silicon layer. For the 1.4×1018 cm−2 specimen, the amorphous layer contained bubbles. Wafers subsequently annealed at 1200 °C in order to form silicon-on-insulator structures showed the following. For the 0.25×1018 cm−2 specimen, there was a buried discontinuous polycrystalline α-Si3N4 layer, and an upper silicon layer with no observable defects. For the 0.75×1018 cm−2 specimen, there was a buried continuous polycrystalline α-Si3N4 layer containing small silicon islands, and an upper silicon layer either without defects or with microtwins adjacent to the nitride/silicon interface. For the 1.4×1018 cm−2 specimen, there was a buried multilayer structure with the middle layer consisting of substantially single-crystal α-Si3N4 free from silicon islands but containing bubbles; and an upper silicon layer with microtwins and threading dislocations. For the 0.25 and 0.75×1018 cm−2 specimens, the α-Si3N4 had often grown epitaxially in the single-crystal silicon. For the 0.75×1018 cm−2 specimen, such epitaxy had less often occurred. For the 1.4×1018 cm−2 specimen, such epitaxy was not observed. These structural results are correlated with the implantation conditions and nitrogen depth profiles obtained by secondary ion mass spectrometry. The mechanisms responsible for producing the structures are discussed.

The microstructures and domain configurations of a Pr2Fe7B3 alloy prepared in amorphous form by melt spinning and subsequently crystallized by annealing at temperatures from 606°C to 660°C were studied by transmission electron microscopy (TEM). The grain size was about 100 mm. The specimens appear to consist of Pr2Fe14B grains separated by nonmagnetic phases. The nonmagnetic phases occur in both Pr 1+eFe4B4 grains and (probably) a Pr-rich solidified eutectic, which gives a mottled contrast in TEM. Lorentz microscopy of thermally demagnetized specimens shows that the magnetic grains (at least the great majority of them) are single-domain. Small changes are seen in the magnetization configuration after applying and then removing a field of approximately 1 T, indicating magnetization reversal in some grains

Amorphous Sm(Fe11Ti) ribbons were prepared by melt spinning. After heat treatment, very fine grains (≊50 nm) of the ThMn12-structure phase were obtained with no signs of a separate intergranular phase. Each grain appears to be a single domain. The maximum room-temperature coercivity is 0.56 T, rising to 1.20 T at 11 K. A model is proposed for the coercivity which is based on an analogy with the random anisotropy model of amorphous magnetism. It gives the correct magnitude of the coercivity, and suggests it should vary as the reciprocal of the crystallite size.

Silicon specimens were implanted with 400 keV Ge+, and various thickness of the amorphised layer were removed by etching. Rapid thermal annealing at 1086-degrees-C was performed. TEM showed that dislocation loops were present. The diameters and densities were not significantly dependent on amorphous layer thickness. Coarsening of loops with increasing annealing time occurred. The number of interstitial atoms contained in the loops was constant except for an increase at the longest time (100 s). We deduce that extra interstitials are supplied from deeper in the specimen or by oxidation at the surface.

Ion beam synthesis (IBS) has been used to fabricate buried compound layers in silicon. These layers were produced by implanting combinations of O+ and N+ ions or NO+ ions at 200 keV/atom into 100 single crystal silicon maintainted at a temperature of between 510°C and 580°C. The specimens were then annealed at 1200°C for two hours.

Silicon wafers were implanted with 200 keV 14N+ ions to a dose of 0.95×1018 14N+ cm−2 at a temperature of 520 °C. These wafers were then annealed at 1405 °C for 30 min, a temperature significantly higher than that normally used (1200 °C) for annealing buried nitride, silicon‐on‐insulator (SOI) structures. Annealing leads to the formation of a well defined layer of essentially single‐crystal Si3N4 containing a few low angle grain boundaries. The Si/Si3N4 interfaces are abrupt, the lower one being almost planar while the upper one shows some irregularities due to fingerlike protrusions of Si3N4 and has an associated region of defective silicon. The silicon overlay is single crystal with no resolvable defects making the wafers good candidates for SOI substrates.

(100) silicon waferSgWere-implanted with 200keV N ions to doses of 0.95 and 1.1x1018 cm-2 at a temperature of 520°C, and then annealed at 1405°C for 30 minutes. We report here observations of the resulting microstructures. A buried ɑ-Si3N4. layer and a good quality silicon overlayer were found. Evidence of cavities in the Si3N4. layer was found in the higher dose sample. Our results suggest that the cavities are associated with the delamination which sometimes occurs during annealing.

To synthesize buried compound layers, reactive ions N+, O+ and NO+ of energy 200 keV atom−1 were implanted into (100) single crystal silicon. Implanted specimens were annealed at 1200°C for 2 h, and analysed by Rutherford backscattering and ion channelling to assess the radiation damage and impurity distributions. SIMS was used to provide further information on the impurity depth distribution and cross-sectional TEM was used to assess the quality of the silicon overlayer and the chemical composition of the buried layer. It was found that when nitrogen is implanted into a wafer which has previously received a high dose of oxygen, the nitrogen moves to the wings of the distribution to form an oxy-nitride. The presence of nitrogen during the subsequent anneal is found to lead to an improvement in the quality of the silicon overlay. A similar improvement is observed for NO+ implants into silicon where a single layer of oxy-nitride is formed. However, if a small dose of oxygen is implanted after the nitrogen, a structure containing twinned defective silicon results. From these results the mechanisms responsible for these structures can be reassessed and are discussed in terms of the solid solubility and self diffusion of the impurity within the matrix.

To synthesise buried compound layers, reactive ions C+, N− and O2 + of energy 200 keV/atom were implanted into (100) single crystal silicon at approximately 500°C. Implanted specimens were annealed at high (≥ 1200°C) temperatures. Rutherford backscattering and channelling techniques were used to study the distributions of the implanted species and annealing of the radiation damage. SIMS provided complementary information on the distribution of light elements in silicon. Cross-sectional TEM was used to identify chemical compositions of buried layers and to assess the quality of silicon overlayer after annealing. It is demonstrated that oxygen in silicon forms a well defined layer of SiO2, with a good quality single crystal silicon overlay upon annealing at 1300°C. Nitrogen in silicon can be successfully annealed at 1200°C while no annealing procedure has yet been found for carbon in silicon. Differences and similarities in both the as-implanted and high temperature annealed materials are explained by the mobilities of the impurities in bulk silicon and in the impurity rich and/or compound layers.