A. F. W. Willoughby

Diffusion of boron in germanium at 800–900°C

Diffusion of B in Ge is studied in the temperature range 800–900°C using implantation doping and B doped epitaxial Ge layers. Concentration profiles before and after furnace annealing were obtained using high resolution secondary ion mass spectroscopy (SIMS). Diffusion coefficients were calculated by fitting the annealed profiles using TSUPREM. We obtained diffusivity values which are at least two orders of magnitude lower than the lowest values previously reported in the literature. Using our values an activation energy of 4.65(±0.3)eV is calculated. Present experimental results suggest that interstitial mediated mechanism should be considered for B diffusion in Ge in accordance with recent theoretical calculations. Annealed SIMS profiles also suggest that B solid solubility in Ge is ∼2×1018cm−3 at 875°C which agrees with literature values.

Boron diffusion across silicon–silicon germanium boundaries

Most boron diffusion studies in Si–Ge have been made in regions of uniform germanium content. In this paper diffusion is observed from a boron-doped epitaxial silicon layer across surrounding Si–Ge layers. Pileup of boron in the Si–Ge layers shows that the activity coefficient for boron in Si–Ge is lower than that for pure silicon. A simple pairing model for Si–B interaction fitted the pileup quite well, with the same equilibrium constant applying to both Si0.9Ge0.1 and Si0.97Ge0.03 layers. The effect of this was simply to immobilize a significant fraction of the boron while retaining its acceptor qualities, the ratio of immobile boron to normal substitutional boron being proportional to the germanium content. Quasielectric field effects at the Si–SiGe interface have a strong effect on the results obtained.

Diffusion of ion-implanted boron in germanium

The diffusion of boron (B) in germanium (Ge) is studied. B was introduced in Ge wafers by ion implantation and concentration profiles after furnace annealing were obtained using secondary ion mass spectroscopy. The diffusion coefficient and solid solubility of B in Ge has been calculated to be 1.5(+/-0.3)x10-16 cm2/s and 5.5(+/-1.0)x1018/cm3, respectively at 850 degrees c by fitting experimentally obtained profiles. The value of diffusion coeffienc is at least two orders of magnitude lower than the minimum value reported in the literature for B diffusion in Ge. The results are significant as they question the general agreement about vacancy diffusion as the mechanism responsible for diffusion of B in Ge.

Surface charge and stress in the Si/SiO2 system

Abstract A model for the stress distribution in thermally oxidised silicon slices has been developed using beam theory and bimetallic strip theory. The stresses have been confirmed by making lattice parameter measurements on oxidised silicon using the APEX X-ray diffraction technique. The often suggested relationship between the surface charge density at the Si/SiO2 interface and the stress in the silicon surface has been investigated and shown to be inconsistent. Finally, analysis of the variation of surface charge density with oxide thickness has caused us to postulate the presence of both positively and negatively charged centres at the interface.

Self-Diffusion in gallium arsenide

The self-diffusion of arsenic in gallium arsenide has been studied over the temperature range 1000 to 1075δC using radiotracer techniques.76As was diffused into GaAs samples at known arsenic pressures in sealed capsules. After diffusion, layers were removed from the surface using anodic oxidation followed by oxide dissolution. Diffusion profiles were obtained by measuring the76As concentration in each sectioned layer by γ-radiation counting. Diffusion coefficients at PAs2 = 0.75 atm and over the temperature range 1000 to 1050δC were found to be 5.2 × 10-16cm2s-1 to 1.5 × 10-15 cm2s-1, leading to an activation energy of the order of 3.0± 0.04 eV and a pre-exponential factor of 5.5 × l0-4 ± 2.4 × 10-4 cm2s-1. Diffusion coefficients at PAs2 =3.0 atm were found to be 5.5 × 10-15 and 9.8 × 10-16 cm2 s-1 at 1050 and 1075δC, respectively. Results are discussed in terms of native point defect equilibria with the arsenic gaseous phase, and with respect to other work. It is deduced from our observed arsenic pressure dependence of the arsenic diffusivity that the most likely diffusion mechanism

Diffusion of boron in heavily doped n‐ and p‐type silicon

The diffusion of 10B in the presence of high‐concentration 11B and As doping has been studied. Dopants were introduced by ion implantation and profiles after annealing were obtained by secondary ion mass spectrometry. Diffusion coefficients were derived by comparing experimental profiles with those from a computer simulation program and results confirmed that diffusion of boron is enhanced in p+ silicon and depressed in n+ silicon. These results have been analyzed using the widely accepted vacancy model for boron diffusion and have produced values of the parameter β, which is related to the ratio of diffusivity for charged and uncharged vacancies, of 0.25 to 3.0 for the p+ and 3.0 to 7.7 for the n+ conditions. This difference cannot be ascribed to experimental error and suggests that further refinement of the vacancy model is required.

The diffusion of gold in thin silicon slices

Abstract A new theoretical analysis of the diffusion of gold in silicon is presented, in which the theory of the dissociative mechanism(1) is modified by the inclusion of vacancy generation in the bulk, the vacancies being emitted from jogs on climbing dislocations. The driving force for this climb is the depression of the vacancy concentration from its equilibrium value, by the conversion of interstitial gold atoms to the substitutional species which uses up vacancies. Equations for this model are derived and solved numerically to give concentration profiles for the case of diffusion in thin slices. It is shown that published experimental data fit the predictions of the model closely, in contrast to predictions of alternative models. A mechanism involving the reaction of di-vacancies with interstitial gold atoms to form paired substitutional gold atoms is also discussed.

Effect of fluorine implantation dose on boron thermal diffusion in silicon

This paper investigates how the thermal diffusion of boron in silicon is influenced by a high energy fluorine implant with a dose in the range 5?1014 to 2.3?1015cm-2. SIMS profiles of boron marker layers are presented for different fluorine doses and compared with fluorine profiles to establish the conditions under which thermal boron diffusion is suppressed. The SIMS profiles show significantly reduced boron thermal diffusion above a critical F+ dose of 0.9-1.4?1015cm-2. Fitting of the measured boron profiles gives suppressions of the boron thermal diffusion coefficient by factors of 1.9 and 3.7 for F+ implantation doses of 1.4?1015 and 2.3?1015cm-2 respectively. The suppression of boron thermal diffusion above the critical fluorine dose correlates with the appearance of a shallow fluorine peak on the SIMS profile in the vicinity of the boron marker layer. This shallow fluorine peak is present in samples with and without boron marker layers, and hence it is not due to a chemical interaction between the boron and the fluorine. Analysis of the SIMS profiles and cross-section TEM images suggests that it is due to the trapping of fluorine at vacancy-fluorine clusters, and that the suppression of the boron thermal diffusion is due to the effect of the clusters in suppressing the interstitial concentration in the vicinity of the boron profile.

Fracture of silicon wafers

Abstract In spite of the increasing use of silicon in applications where mechanical stresses are deliberately applied to the material, such as in transducers, and the fatal nature of cracking in silicon devices, there is very limited characterisation and understanding of the fracture behaviour of silicon wafers at room temperature. This understanding is of increasing importance with the use of larger diameter wafers in modern technology. This paper examines the fracture strength of a wide range of silicon material both as-grown and after processing. The wafers tested were from crystals grwon by float-zone and Czochralski techniques and the effects of oxidation, ion-implantation and annealing in various environments have been studied. The technique used to measure the fracture stress involved simply supporting the wafer on an aluminium ring concentric to the load axis. The load was gradually increased until the wafer fractured. This method was chosen to avoid edge effects, and has proved to have adequate reproducibility. Typical values of the fracture stress obtained by this method, for different crystals, vary between 2 and 3.5 GPa. In the first part of the study, the role of the surface on the fracture behaviour has been investigated in detail. While the surface perfection of the tensile surface has a major effect on the fracture stress (as shown in previous studies), some of the results were found to be sensitive to the compressive surface as well. In the case where the results are sensitive to the compressive surface finish the fracture stress rose from 3.7 to 8.8 GPa as the surface finish was improved while in the cases where they were not sensitive the fracture stress remained at about 3.5–4.6 GPa. Only in the float-zone material were fracture stresses approaching 8.8 GPa observed. At this level of fracture stress, the behaviour is believed to be sensitive to surface defects less than 0.01 μm in size. These results can be analyzed in terms of surface controlled defects under conditions where surface defects are dominant and bulk controlled defects where these defects are dominant. In this manner bulk effects can be isolated from surface ones. This gives the opportunity to study the effects of specific defects on the fracture stress and the results in this paper are discussed in terms of the role of surface and internal defects on the fracture stress.

The mechanisms of yield and plastic flow in HgTe and Cd x Hg1?x Te

Single crystals of HgTe and CdxHg1−x (0.18<x<0.30), oriented for single slip, have been deformed in four-point bending at strain rates ∼10−4 sec−1 and temperatures from −11 to +84° C for HgTe, and 20 to 195° C for CdxHg1−xTe. At the lowest temperatures, the stress-strain curve exhibits a sharp yield relaxation and subsequent zero work hardening regime, as commonly observed for other semiconductors. Experiments show that the yielding mechanism is that proposed by Johnston and Gilman for LiF. Possible explanations for the post-yield zero work hardening phenomenon are discussed. The influence of composition, temperature and strain rate on the stress-strain behaviour are reported. At 20° C, the upper and lower yield stresses (τuy andτ1y) increase with increasingx in qualitative agreement with our earlier hardness results. For Cd0.2Hg0.8Te,τ1y varies with temperature,T, at a strain rate of 10−4 sec−1, according toτ1y∝ exp (Q/kT) whereQ is 0.16 eV. For HgTe the comparable value is 0.11 eV. Atx=0.25 and constant temperature,τ1y depends on strain rate γ asτ1y∝ γ1/n wheren is 4. The stress level for deformation of Cd0.2Hg0.8Te at γ∼ 10−4 sec−1 and 20° C is 2–3 kg mm−2, comparable with that for InSb at 300° C or Si at 1000° C. Strain rate cycling tests on CdxHg1−xTe give values of activation volumeV* around 10b3 at 20° C, independent of plastic strain (up to 2–3%), suggesting that deformation in these alloys is controlled by the Peierls mechanism, as observed in other II–VI compounds.