We use the Lattice Monte Carlo method to simulate the coupled motion of vacancies, interstitials and dopants during the annealing of a 50nm channel length NMOS structure. The initial defect locations are taken from Monte Carlo ion implantation simulations. The resulting defects diffuse, recombine, pair and cluster, with rates of these atomic processes calculated based on the local environment. Dopant redistribution occurs via both displacement by vacancies as well as the formation and diffusion of mobile boron and arsenic interstitials. We use these simulations to demonstrate the potential of atomistic simulations for deep submicron devices and to explore the influence of atomistic processes on macroscopic behavior.

In this paper we examine the properties of arsenic in silicon, using ab initio calculations and a statistical theory. Good agreement is found between theory and experiment for the electronic concentration as a function of temperature and total arsenic concentration. We show that for low arsenic concentrations, full activation is the equilibrium condition. In equilibrium, the neutral complex composed of a lattice vacancy surrounded by four arsenic (VAs4) is the dominant means by which high concentrations of arsenic are rendered inactive. Under constrained equilibrium conditions in which VAs4 cluster formation is prohibited, we show that VAs3Si1 cluster populations increase dramatically and can account for nearly the same degree of compensation as the VAs4 clusters. Even VAs2 clusters alone can account for substantial deactivation in the absence of VAs3 and VAs4 clusters. These smaller complexes are essential not only to the establishment of equilibrium, since SiAs4 clusters are extremely rare, but can also explain some degree of deactivation, even if the formation of VAs4 clusters are kinetically inhibited.

We have proposed a coupled and de-coupled combined method to solve partial differential equations for a transient enhanced diffusion model. In the case of a boron diffusion process, the sum of concentrations of interstitial Si and of impurity-interstitial pair, the sum of concentrations of vacancy and of impurity-vacancy pair and each chemical impurity concentration are kept constant. The charge neutrality law is also applied. This procedure has realized a robust solution system which is implemented into our in-house FEM-based 2-D process simulator, and transient enhanced diffusion simulations for a sub-quarter micron nMOSFET have been demonstrated.

2D calculations of as-implanted dopant profiles are performed using a new module of the process simulator DIOS which is based on the binary collision code Crystal-TRIM. Examples derived from MOS and trench technology are considered. Good lateral and depth resolution is achieved within an acceptable computing time by combining a trajectory splitting algorithm with a method for the lateral duplication of ion trajectories.

A computationally efficient method for ion implantation simulation is presented. The method allows two-dimensional ion implantation profiles in arbitrary shaped structures to be calculated and is valid for both amorphous and crystalline materials. It uses an extension of the one-dimensional dual Pearson approximation into the second dimension.

An alternative three-dimensional (3D) Monte Carlo (MC) dynamic simulation model for phosphorus implant into (100) single-crystal silicon has been developed which incorporates the effects of channeling and damage. This model calculates the trajectories of both implanted ions and recoiled silicons and concurrently and explicitly affects both ions and recoils due to the presence of accumulative damage. In addition, the model for room-temperature implant accounts for the self-annealing effect using our defined recombination probabilities for vacancies and interstitials saved on the unit volumes. Our model has been verified by the comparison with the previously published SIMS data over commonly used energy range between 10 and 180 keV, using our proposed empirical electronic energy loss model. The 3D formations of the amorphous region and the ultra-shallow junction around the implanted region could be predicted by using our model, TRICSI (TRansport Ions into Crystal-Silicon).

We have calculated the displacement-threshold energy Ed for point-defect production in Si and SiC using empirical potentials, tight-binding, and first-principles methods. We show that—depending on the knock-on direction—64-atom simulation cells can be sufficient to allow a nearly finite-size-effect-free calculation, thus making the use of first-principles methods possible. We use molecular dynamics (MD) techniques and propose the use of a sudden approximation which agrees reasonably well with the MD results for selected directions and which allows estimates of Ed without employing an MD simulation and the use of computationally more demanding first-principles methods. We compare our results for Ed with the available experimental values. Furthermore, we have examined the temperature dependence of Ed for C in SiC and found it to be negligible.

The annealing kinetics of implant damage in Si+ implanted Si has been investigated using in-situ and ex-situ annealing of transmission electron microscopy (TEM) samples prepared prior to annealing. The defect evolution at 800°C was studied for a Si wafer implanted with Si+ at 100keV to a dose of 2×1014 cm-2. This implant was above the sub-threshold loop formation threshold allowing one to study simultaneously the {311} defect dissolution and dislocation loop nucleation and growth. In order to study the effect on the defect evolution of using a thin sample for an in-situ annealing experiment, a pair of samples, one thick and one thinned into a TEM sample, were annealed in a furnace simultaneously. It was found that the presence of a second surface 2000Å below the implant damage did not affect the extended defect evolution. For the in-situ annealing study it was found that the {311} dissolution process and sub-threshold dislocation loop formation process was not affected by the TEM electron beam at 160kV as long as an 800°C furnace pre-anneal was done prior to in-situ annealing. The dissolution rate of the {311} defects was used to confirm the TEM holder furnace temperature. The results of both the in-situ the {311} defects is released during the 311 dissolution process and 30% comes to reside in dislocation loops. Thus, the loops appear to contain a significant fraction of the total interstitial concentration introduced by the implant.

Selective chemical etching and atomic force microscope (AFM) examination has been performed to delineate two-dimensional (2-D) dopants profiles of p/n-type well and junction areas. Selectivity strongly depended on the types of dopants and the ratio of etching solutions. Calibration showed that the carrier concentrations in both p/n-type regions could be delineated down to a level of ∼1×1017/cm3. The AFM-induced profiles were compared with the calculated data provided by the 2-D process simulators such as TRIM and SUPREM-IV.

Oxidation of thin-film Si, oxygen precipitation in bulk crystalline Si, and the formation of buried oxide layers (as in the SIMOX process for SOI) have been studied extensively for decades but the underlying atomic scale processes have remained elusive. Furthermore, common features of these phenomena have not been generally appreciated, in part because the kinetics are substantially different in the different cases. In this paper we review recent theoretical research based on atomic-scale first-principles calculations that provides a unified description of the three phenomena. In particular, we account for both the normal and enhanced mode of oxygen diffusion that leads to the formation of oxygen clusters known as thermal donors and their subsequent annealing and evolution into SiO2-like precipitates. It is proposed that a novel family of interface defects that are akin to thermal donors, composed of “frustrated” Si-O bonds, are a natural by-product of thin-film oxidation. An explicit mechanism for the emission of Si interstitials, which occurs during both oxidation and oxygen precipitation, is obtained. It is shown that emission of Si interstitials eliminates the frustrated-bond defects, thus explaining the high quality of the resulting interfaces. It is proposed that frustrated-bond defects are responsible for much of the behavior observed at the Si-SiO2 interface that cannot be accounted by dangling bonds and for the amphoteric traps that occur in high concentrations in SOI material.

A simple model of thermal dissociation and recovery of hydrogen-passivated silicon defects at the Si/SiO2 interface, such as Pb - centers, during vacuum thermal annealing has been suggested. This model considers reactions of hydrogen with defect states at the Si/SiO2 interface and diffusion of liberated atomic and molecular hydrogen in a silicon dioxide film. An excellent agreement was obtained between the experimental and numerical simulation results for oxides with different thickness (200–1024 Å), grown both (111) and (100) samples and annealed in the temperature range (480–700° C).

Nucleation of voids and vacancy-type dislocation loops in Si under vacancy supersaturation conditions has been considered. Based upon nucleation barrier calculations, it has been found that voids can be nucleated, but not dislocation loops. The homogeneous nucleation rate of voids has been calculated for different temperatures by assuming different enthalpy values of Si vacancy formation. The process of void growth due to precipitation of vacancies has been numerically simulated. Comparing results of the nucleation and the growth modeling and taking into account the competition between the two processes, the limited time available, and the crystal cooling rate after growth, it has been shown that homogeneous nucleation of voids to experimentally observed densities and void growth to observed sizes is possible if enthalpy of Si vacancy formation is within the range of 2.9 to 3.6 eV with the nucleation temperature in the range of 980–1080 °C.

The implementation and application of GaAs technology CAD for industrial research and development are described. Existing silicon-oriented software was modified extensively to make it applicable to the simulation of GaAs processes and devices. RF-oriented post-processing capabilities were implemented, and effective applications methodologies were developed. Examples of the practical application of GaAs technology CAD are presented.

We propose a new approach for modeling of impurity diffusion at semiconductor heterointerfaces. The approach is based on the notion of a common energy reference for highly localized defects. It is shown that in the kick-out process, the segregation of group II acceptors is controlled by the valence band offsets among different constituent layers of the heterostructure. Extensive numerical modeling of the diffusion provides an explanation for the experimentally observed strong segregation of Zn and Be acceptors in the lattice matched InP/InGaAs, InP/InGaAsP and GaAs/AlGaAs heterostructures.

The diffusivity of Si in GaAs shows a dependence on the cubic power of its concentration or the concentration of electrons n under both in- and outdiffusion conditions. Hence, the diffusion of Si in GaAs is consistent with the Fermi-level effect model invoking the triply-negatively-charged Ga vacancies, , as the point defect species responsible for diffusion to occur on the Ga sublattice under n-doping condition. However, the Si diffusivity values of the indiffusion cases is several orders of magnitude smaller than those of the outdiffusion cases at the same Si concentrations. This means that the two apparent Si diffusivity values under intrinsic conditions will contain also the same discrepancy, which has been previously assessed to be due to a undersaturation in indiffusion cases and a supersaturation in outdiffusion cases. In this study we have calculated the under- and supersaturation values using the known Si diffusivities. We found that the GaAs surface states play a key role in the development of the under- and supersaturations.

Heavy n-doping enhanced disordering of GaAs based III-V semiconductor superlattice or quantum well layers, as well as the diffusion of Si in GaAs have been previously explained by the Fermi-level effect model with the triply-negatively-charged group III lattice vacancies identified to be the responsible point defect species. These vacancies have a thermal equilibrium concentration proportional to the cubic power of the electron concentration n, leading to the same dependence of the layer disordering rate. In this paper, in addition, we take into account also the electric field effect produced by the material bandgap heterogeneity and/or hetero-junctions. In heavily n-doped or long time annealing cases, this effect is negligible. At low n-doping levels and for short annealing times, the layer disordering rate can be enhanced or reduced by this effect. Available experimental results of low Si-doped and very short-time annealed samples have been satisfactorily fitted using the Fermi-level effect model.

Based on the kick-out mechanism, we have modeled the time evolution of the Si self-interstitial supersaturation during gold diffusion from evaporated surface layers into silicon substrates. The model comprises a limited annihilation velocity υI of self-interstitials at the Si surface, which accounts for the gradual increase of the Au boundary concentration observed during rapid thermal annealing at higher temperatures and furnace annealing at lower temperatures. Experimental data were fitted by computer simulation of the Au diffusion process. An approximate analytical expression describing the increase of the Au boundary concentration with time is also given. We obtain υI 1.35 × 1011 exp(−3.91 eV/kBT) ms-1 for gold-covered {100}-oriented Si surfaces between 845 °C and 1119 °C.

Previous modeling of Al gettering of Au in Si indicated that, for an Al gettering layer placed on one surface of a Si wafer, Au will be gettered from both surface regions of the wafer to progressively greater depths with time. This is because, in Si, Au is a substitutional-interstitial (Aus-Aui) species with its diffusion governed by the kick-out mechanism which is mediated by Si self-inter-stitials (I). During gettering by the Al-Si liquid at one wafer surface, Aus atoms change over to Au atoms to rapidly migrate out of Si into the liquid. The changeover process consumes I. At the two wafer surface regions, the consumed I will be quickly replenished, while in the wafer interior an I undersaturation develops which hinders the Aus-Aui changeover and hence the gettering process. Experimental evidences which confirm the predictions have been obtained. Au was indiffused into a FZ Si wafer at 950°C for 16 hr. After removing the Au source and etching, samples from the in-diffused wafer were annealed at 1000°C without or with an Al layer on one surface. For samples without Al, there is no change in the net Au content, while the U-shaped indiffused profile becomes flatter. For samples with an Al layer, both wafer surface region Au concentrations were significantly decreased in 30 min while the wafer interior Au concentrations decreased only after annealing for longer times. The model predictions, the experimental results, and the implications on Si self-diffusion parameters will be discussed.

Reactive ion etching (RIE) causes substrate surface contamination, substrate damage, and induces defect reactions to produce carrier traps at depths into μm range. We have developed a model describing the indiffusion of interstitials and the subsequent reactions in the defect reaction region to predict the defect depth profiles. We formulate the reaction kinetics as a series of 1-D coupled interstitial diffusion-reaction partial differential equations (PDEs) with a moving boundary. The predicted defect depth profiles are consistent with those measured in the photoluminescence (PL) experiments. We conclude that the defect depth profiles are determined by interstitial diffusion coefficient (Di), etch rate (υE), etch time (tE), defect reaction rate (K), and background dopant and impurity concentrations ([C], [B] and [O]) in the Si substrate. The μm range defect depth profiles can be explained as: (i) fast diffuser Si self-interstitial (Sii) indiffusion to a μm depth range of (Di/K)1/2 limited by [C] and [B]; (ii) the generation of carbon and boron interstitials (Ci and Bi) through the Watkins replacement reactions, and (iii) the formation of Ci- and Bi-related defect pairs through diffusion limited pairing reactions. Ci and Bi indiffusion are limited by υE and extremely shallow (Di/υE) during a typical RIE process with υE≈1000Å/min and tE≈10min. The indiffusion of vacancies (V), contamination from the etching plasma and enhanced diffusion effects are also considered in the model.

The increase and reduction of the vacancy concentration by rapid thermal processing (RTP) was investigated by platinum diffusion. Direct experimental evidence is presented for the consumption of vacancies during oxidation and for the introduction of vacancies during processing in ammonia and nitrogen. These results confirm the indirect conclusions from dopant diffusion and from the growth and shrinkage of extended defects. In addition, it was possible to establish an upper limit for the equilibrium concentration of vacancies at 1180 °C which is lower than previously reported values.