We report self-diffusion studies of silicon between 855 and 1388°C in highly enriched epitaxial 28Si layers. Diffusion profiles of 30Si and 29Si are determined with high resolution secondary ion mass spectrometry (SIMS). The temperature dependence of the Si self-diffusion coefficients is accurately described with an activation enthalpy of 4.76 eV and a pre-exponential factor of 560 cm2s-1. The single activation enthalpy indicates that Si self-interstitials dominate self-diffusion over the whole temperature range investigated. Self- and interdiffusion in buried Al71GaAs/Al69GaAs/71GaAs isotope heterostructures with different Al composition is measured between 800 and 1160°C. Ga self-diffusion in AlGaAs and interdiffusion of Al and Ga at the AlGaAs/GaAs interface show that Ga diffusion decreases with increasing Al composition and that the interdiffusion coefficient depends linearly on Al concentration. Furthermore Al is found to diffuse more rapidly into GaAs than Ga diffuses in GaAs. The temperature dependence of Ga and Al diffusion in GaAs and of Ga diffusion in AlGaAs is described by a single activation enthalpy in the range of 3.6±0.1 eV, but by different pre-exponential factors. Differences found for Ga and Al diffusion in GaAs and for Ga diffusion in AlGaAs with different Al concentrations are discussed.

Self-diffusion on the As sublattice in intrinsic GaAs was investigated in a direct way by As tracer diffusion measurements using the radioisotopes 73As and 76As and in an indirect way by annealing of buried nitrogen doping layers in epitaxially grown GaAs/GaAs:N heterostructures. The latter experiments were analyzed by secondary ion mass spectroscopy and interpreted within the framework of the kick-out mechanism yielding the As diffusivities mediated by As interstitials IAs. Comparison of with tracer diffusion coefficients – including data reported in the literature–points to a substantial contribution of IAs to As diffusion in intrinsic GaAs under As-rich ambient conditions.

The advent of epitaxial-growth techniques dedicated to the growth of epitaxial layers of group-IV semiconductors and their alloys has opened up new possibilities for novel types of diffusion experiments and thus for critical tests of existing diffusion theories. It is now possible to grow test structures for diffusion consisting of well-defined narrow distributions of tracer impurities in epitaxial layers of biaxial-strained or relaxed alloys of different compositions. This enables studies of alloy effects and/or strain effects on diffusion with varying strain energy for a well-defined type of strain (tensile or compressive). A review is presented on recently published results on impurity diffusion in strained and relaxed, epitaxial SiGe alloy layers. As almost all these studies have focused on boron and antimony diffusion, these two impurities will have the leading roles in this review.

This paper presents the self-consistent tight-binding method of new generation which, unlike other tight-binding methods, allows one to calculate structural energies of multiatomic systems (molecules, clusters, defects in solids) and their spectroscopic energies in the framework of the same computational scheme and with comparable accuracy. Reliability of the method is illustrated considering defect state problems in crystalline and amorphous silicon (electron-enhanced-atomic diffusion, metastable defect creation, defects with effective-negative correlation energies, etc.) and comparing obtained results with ab initio calculations and experimental data.

The crystalline structure surrounding a single neutral vacancy in silicon is investigated through extensive first-principles total-energy calculations. The results indicate the existence of two distinct distortions of the lattice around the vacancy with essentially the same formation energies at zero pressure, but, however, with different formation volumes. The effect of hydrostatic and biaxial stresses on the relative concentration of each distortion is discussed, suggesting experimental ways to investigate the crystalline structure around the single vacancy and its role as a mediator of atomic diffusion in silicon.

Diamond films are of interest in many practical applications but the technology of producing high-quality, low-cost diamond is still lacking. To reach this goal, it is necessary to understand the mechanism underlying diamond deposition. Most reaction models advanced thus far do not consider surface diffusion, but recent theoretical results, founded on quantum-mechanical calculations and localized kinetic analysis, highlight the critical role that surface migration may play in growth of diamond films. In this paper we report a three-dimensional time-dependent Monte Carlo simulations of diamond growth which consider adsorption, desorption, lattice incorporation, and surface migration. The reaction mechanism includes seven gas-surface, four surface migration, and two surface-only reaction steps. The reaction probabilities are founded on the results of quantum-chemical and transition-state-theory calculations. The kinetic Monte Carlo simulations show that, starting with an ideal {100}-(2×1) reconstructed diamond surface, the model is able to produce a continuous film growth. The smoothness of the growing film and the developing morphology are shown to be influenced by rate parameter values and by deposition conditions such as temperature and gaseous species concentrations.

Previously reported measurements of anomalously high rates of oxygen out-diffusion in Czochralski silicon at low temperatures (T≤450°C) are confirmed. The surface concentration is shown to decrease with increasing time while the depth to which the concentration is depleted remains constant. Exposure to a hydrogen plasma under conditions known to catalyse the diffusion of isolated oxygen atoms causes an increased rate of decrease of the surface concentration without significantly affecting the depth to which the concentration is depleted. The evolution of the out-diffusion profiles cannot be explained by a catalytic mechanism operating on the isolated oxygen atoms. A slow conversion of Oi to a complex containing oxygen which can diffuse rapidly over long distances before being trapped either on the surface or in the bulk of the sample can account for both out-diffusion and simultaneous loss of [Oi] in the bulk. The conversion rate is enhanced by exposure to a hydrogen plasma, indicating that it is controlled by the diffusion rate of isolated atoms.

We present out-diffusion of Zn in Si as a new method to study properties of Si vacancies. Out-diffusion experiments were performed on homogeneously Zn-doped Si samples at 1107°C and 1154°C. The resulting concentration-depth profiles were measured by means of spreading-resistance profiling. Based on a diffusion model in which Zn migrates simultaneously via the kick-out and the dissociative mechanism all experimental profiles were modeled by computer simulations. The calculations reveal that out-diffusion of Zn from Si occurs to a considerable extent via the dissociative mechanism. Hence, vacancy properties like the equilibrium concentration and the transport capacity can be extracted from profile fittings. The results are compared with literature data deduced from in-diffusion experiments.

Interdiffusion and self-diffusion in III-V semiconductors has usually been assumed to operate through the diffusion of point defects, the concentrations of which are at thermal equilibrium values. We have studied the interdiffusion of multiple quantum well samples, grown under a range of growth conditions, each containing a thin source of vacancies. This has enabled simultaneous measurements of the interdiffusion coefficient, diffusion coefficient for vacancies and the concentration of those vacancies in a single experiment. We have shown that independent of growth conditions, within a wide window of III-V flux ratio and temperature, the diffusion at all temperatures is governed by a constant background concentration of vacancies.

The time dependence of the size distribution of arsenic precipitates during annealing for both large (~10nm) and small size (~4nm) regimes is investigated. A narrowing of the size distribution is observed in the small size regime. This improvement in size uniformity is in marked contrast to what is observed for larger precipitates, which coarsen with a widening distribution similar to that of classical Ostwald ripening. Inverse coarsening caused by an elastic interaction between small precipitates due to coherency strain is a possible mechanism for this interesting and potentially useful behavior.

The initial diffusion stage (lIDS) of Zn in InP from polymer spin-on films has been investigated. The threshold diffusion temperature of 375°C has been established. It has been found that the heavily-doped near-surface layer, that consisted of electrically neutral zinc-containing complexes, was formed at IDS, while a low part of Zn was electrically active. The Zn diffusivity during IDS has been found to be at least two orders of magnitude more than the Zn diffusivity obtained after an isothermal diffusion.

Recently, the suppression of boron diffusion due to both thermal and transient enhanced diffusion (TED) has been demonstrated through the incorporation of 0.5% substitutional carbon in the base of Si/SiGe/Si heterojunction transistor's (HBT)[1,2]. Because the devices are sensitive to diffusion on a scale less than that we can detect with SIMS, in this paper combined process and device modeling (TMA TSUPREM4 and MEDICI) are used to relate observed electrical characteristics (collector saturation currents and Early voltages) of the HBT's to boron diffusion, with a sensitivity of 20-30Å. Boron diffusivity in the SiGeC base is ~8 times slower than that of the boron diffusivity in the SiGe base without implant damage (no TED). In the case of ion implant damage in an overlying layer to cause TED the excess interstitial concentration due to ion implant damage is reduced by approximately 99% through incorporation of 0.5% substitutional carbon in the HBT SiGe bases. This demonstrates that carbon incorporation acts as an effective sink for interstitials.

An ultra-fast lateral formation of epitaxial NiSi2 line in heavily doped p+-Si diffused tub was created at room temperature ambient under the application of current densities of 106 A/cm2. A NiSi2 line of 140 μm long and I μm wide linking the cathode and anode contacts to p+-Si was formed within 1 second. Shorter lines were formed with lower current densities. The formation of shorter lines was again completed within 1 second and there was very little subsequent growth upon further stressing. The line length strongly depends upon applied current which tend to suggest that the line formation is limited by driving force rather than by kinetics. We propose that the ultra-fast silicide formation is a result of Ni interstitial diffusion in Si induced by electron-hole recombination heating and electromigration.

This work is devoted to the study of Ge diffusion in crystalline Sn1-δTe1+8 with δ=0.0065±0.0008 in temperature range T=878-973 K by electron probe microanalysis and layer by layer X-ray analysis. For the latter lattice constant dependence on composition was determined: a(Å)=a(SnTe)-(0.368±0.008)× where 0<×<0. 1. Activation energy was found to be about 1.3 eV, much less than in the case of Ge diffusion in PbTe.

Enhanced Sb diffusion in biaxially compressed Si1-xGex layers is observed. Assuming the prefactors to be stress independent the Sb diffusion coefficients in biaxially compressed Si0.9Ge0.1 and Si0.8Ge0.2 were extracted as 0.4 × 102 exp[−(3.98(eV)±0.12)/kT] and 1.3 × 102 exp[−(3.85(eV)±0.12)/kT] cm2/s, respectively. The activation volume of Sb diffusion in Si1-xGex (× ≤ 0.2) is estimated to be close to Ω, where Ω is the volume corresponding to a silicon lattice site.

The experimental and atomistic simulation methodologies by which microscopic diffusion mechanisms can be determined in solids are described. Measurement of the Haven Ratio requires evaluation of the diffusion coefficient and of the ionic conductivity for the species in pure and doped specimens and is, in practice, limited to simpler materials. Atomistic simulations using lattice statics, molecular dynamics and Monte Carlo techniques can yield very detailed information on the pathways followed by migrating ions and are being utilised more extensively for this purpose. Examples of such experimental and simulation studies are discussed.

Computational techniques are now well established tools in the study of structural properties and transport mechanisms in solid state ionic materials. This paper will highlight recent developments of such methods, with a strong emphasis on their use in the elucidation of ion diffusion mechanisms or pathways at the atomic level. In particular, contemporary work will be illustrated by accounts of calculations on oxygen ion and proton conduction in LaMO3 perovskite-type oxides, which find use in various electrochemical applications. We have used atomistic simulation, molecular dynamics (MD) and ab initio techniques to provide fresh insight as to their defect and ion diffusion properties.

The high-temperature phase of sodium ortho-phosphate, α-Na3PO4, belongs to the class of ion conducting plastic crystals, i.e., it is characterized by a dynamic rotational disorder of its poly-atomic anions and, at the same time, by a considerable translational mobility of its cations. During the past decade, the possibility, nature, and importance of a dynamic interplay between the two kinds of motion have been a subject of continued controversy. Proponents of a strong interplay coined the expression “paddle-wheel mechanism”. In our present contribution we report, for the first time, on the results of dynamic experiments probing the elementary steps of anionic and cationic motion individually. The techniques utilized in this study are coherent quasielastic neutron scattering and high-frequency conductivity spectroscopy, respectively. The data are complemented by an ab-initio molecular-dynamics simulation. Our results provide a view of the movement of anions and cations and of correlations between them. Strong dynamic coupling is detected between the octahedrally coordinated sodium ions and nearby oxygen ions. For translational sodium-ion transport, a chain mechanism appears to be operative.

Atomistic Molecular Dynamics are used to simulate diffusion of hydrocarbons inside the microporous structure of siliceous zeolite CIT-I, with chemical composition SiO2. CIT-1 is a crystalline microporous material containing channels formed by rings containing 12 and 10 Si atoms (Figure 1). The dimensions of these two channel systems are sufficient to cause substantial differences in the diffusion of para-xylene and ortho-xylene. Diffusion coefficients as a function of loading of each isomer, and activation energies have been calculated from the simulations. The effect of the isomer size in the diffusion path is also analysed.

A combination of molecular dynamics and energy barrier mapping has been used to study diffusion of xylene gas molecules in silica zeolites. Rigid ion models were created for silicalite phases and the energy barriers to xylene permeation by pore diffusion were mapped using constrained steps and energy minimization. Zeolite ZSM-5 proved to exhibit the desirable properties of a high degree of selectivity while retaining a high permeability to p-xylene. The model zeolite was then minimized with various cation dopants using a shell model to mimic ionic polarization, and changes in cell size and energy surface were examined. Long term molecular dynamics simulations showed an increased diffusion rate for the structures doped with cations of larger ionic size.