The development of future Si device technologies will rely extensively on modeling, requiring truly predictive tools. Here we focus on the front-end processes, during which ion-implantation and annealing create 3-D impurity profiles that determine crucial electrical device parameters. The final configuration is the result of a complex interaction of dopant atoms with Si self-interstitials and vacancies, which themselves interact with each other as well as with the implantation-induced damage and interfaces. Predictive modeling requires for all these processes a solid understanding of the physical phenomena as well as accurate quantitative information. Si self-interstitials and vacancies are not observable directly in an experiment, but only via their interactions with some other physical quantity of the sample. We review our work employing dopant atoms in δ-doping superlattices (δ-DSL) that yield directly the time averaged depth profiles of Si native point defects during a particular processing sequence. This approach is uniquely suited for giving insights into the interplay of point defects in Si, providing crosschecks for atomistic calculations as well as parameters for process simulators. We describe experiments to extract interstitial and vacancy parameters and discuss the influence of intrinsic and extrinsic interstitial traps, as well as of the annealing environment, on the native point defect population. The latter allows to place certain bounds on the interstitial vacancy recombination coefficient as well as the ratio of interstitial and vacancy equilibrium concentrations.

The near-surface implantation of hyperthermal neutral atoms with (15 - 75) eV onto diamond (111) and graphite substrates is studied by molecular-dynamics (MD) using a density-functional (DF) based non-orthogonal tight-binding (TB) scheme. Depending on the initial energy and the impact point the atoms penetrate beneath the surface forming regions of local disorder and stress. The energy partition during the collision is analyzed yielding results about penetration and displacement threshold. After a final relaxation of the structures the penetration depths of the colliding particles are determined. The structural topology and the electronic properties of the induced defects and surface modifications are discussed. The penetration thresholds for noble gas atoms, hydrogen and carbon and the atomic-size dependent bulge arising after the subplantation process into a graphite substrate have been determined.

The impact of Si with incident energy Ei (0.1, 1, 5, 10, 20, and 50 eV) and angle θi (0° and 60° from the surface normal) into three model Si surfaces with varying degrees of F coverage (0 ML F, ∼ 1 ML F, and ∼2 ML F) was simulated using classical molecular dynamics (MD). From the simulation results, the probabilities for incident Si reflection and removal of surface Si and F were obtained as a function of Ei, θi, and F surface coverage. In general, these probabilities were observed to depend significantly on these parameters. This result implies that feature evolution simulations require surface reaction models with the necessary functionality in order to make quantitative predictions.

Molecular dynamics (MD) simulations were conducted of Cl+ impact (at 10, 25 and 50 eV) of an initially bare silicon surface, leading to steady state coverage of Cl in a mixed chlorosilyl layer. Our main goal in this study was to compare the MD predictions to models of ion-assisted etching involving the concept of a site balance. For the case of 50 eV Cl+ etching silicon, the coverage vs. exposure results in the simulation could be reasonably well reproduced in a site balance model, but only if the correct parameters in the model were taken from the simulation. The results of the comparison suggest that MD simulations can be helpful in the development of physically sound phenomenological models of ion-assisted etching.

Isolated Si ad-dimers nucleate predominantly on Si(100) substrate dimer rows at room temperature. Using tunneling microscopy (STM), we find that, while ad-dimers oscillate between two mutually orthogonal orientations, the geometry with the ad-dimer bond perpendicular to the substrate's dimer bonds is the more stable. This observation confirms a prediction of ab initio calculations with a gradient correction for the energies of ad-dimers in various configurations, which differs from that of the local density approximation (LDA).

Among the different effects induced by ion implantation into dielectric solids, the enhancement of the ionic mobility is the most important regarding practical applications. We simulated the Na migration in a super-ionic conductor, sodium β″-alumina, irradiated with Ar (E=200 keV) at room temperature. A theoretical model to describe the electric field originated in the bombarded region near the surface was developed. The model emphasizes the relation between the electric field and the energy deposited by the ions in the ionization process.

Due to the large concentration of electrically active defects in Hg1-xCdxTe, it is important to properly account for them when designing fabrication processes for infrared focal plane arrays (IRFPAs). Stanford University's Mercury Cadmium Telluride Process Simulator, SUMerCad, is being developed to simulate the interaction of native defects and dopants in Hg0.8Cd0.2Te in order to accurately model various processing techniques. Non-linear coupled differential equations are used to model the physics of the diffusion mechanisms, rather than extrapolating the diffusion profiles from experimental data. The fundamental constants for the point defects were obtained by exercising the simulator over the various parameter spaces and comparing the results to experimental work and best principle calculations. The methodologies used, point defect mechanisms modeled, and parameters determined are discussed.

Spectroscopic investigations of various atmospheres appearing in diamond film synthesis suggest evidence for carbon atoms, dimers or trimers. Molecular dynamics simulations with the Brenner hydrocarbon potential are being used to investigate the elementary reactions of these species on a hydrogen-terminated diamond (111) surface. In principle these types of simulations can be extended to simulations of growth morphologies, in the 1-2 monolayer regime presently.

Results from a multicomponent Monte Carlo simulation of the deposition and growth of YBa2Cu3O7 are presented and discussed. In particular, a detailed examination of the growth modes active during different morphological growth conditions is performed. At higher deposition rates, both [001] and [100] epitaxial variants (‘c’ and ‘a’ type growth, respectively) are observed to grow by modes attributed to the classic Volmer-Weber mechanism. At very low deposition rates, the film is observed to grow in a distinct, cyclic, multi-stage process. Small islands of [001] epitaxy nucleate and grow to one unit cell height followed by primarily horizontal growth or “ledge extension” until one unit cell layer has formed. This process then repeats. Simulated RHEED amplitude data from this growth process compares favorably to experimentally obtained data.

A model based on ion pairing for the simulation of donor and acceptor codiffusion in silicon is presented. The proposed model allows us to obtain a good agreement with the experimental profiles over a wide range of diffusion conditions, specially at high concentrations where the standard process simulator codes give results strongly inaccurate. Comparison of the simulations with new and literature codiffusion profiles is provided.

A point defect-based model, which has been developed to describe diffusion and electrical activation of boron in crystalline silicon during post-implantation annealing, is used to simulate the anomalous diffusion of boron implanted into silicon along the random and [100] channeling directions. The model predictions are compared to data measured by Chu et al.

In this paper is reported the development and implementation of a new local electronic stopping model for arsenic ion implantation into single-crystal silicon. Monte Carlo binary collision (MCBC) models are appropriate for studying channeling effects since it is possible to include the crystal structure in the simulators. One major inadequacy of existing MCBC codes is that the electronic stopping of implanted ions is not accurately and physically accounted for, although it is absolutely necessary for predicting the channeling tails of the profiles. In order to address this need, we have developed a new electronic stopping power model using a directionally dependent electronic density (to account for valence bonding) and an electronic stopping power based on the density functional approach. This new model has been implemented in the MCBC code, UT-MARLOWE The predictions of UT-MARLOWE with this new model are in very good agreement with experimentally-measured secondary ion mass spectroscopy (SIMS) profiles for both on-axis and off-axis arsenic implants in the energy range of 15-180 keV.

Experimental studies have shown that strains due to thermal expansion mismatch between a film and its substrate can produce very large stresses in the film that can lead to the formation of holes and hillocks. Based on a phenomenological description of the evolution of a solid surface under both capillary and stress driving forces and for surface and grain boundary self-diffusion, this article provides, for the first time, analytical and numerical solutions for surface profiles of model geometries in polycrystalline thin films. The results can explain a variety of surface morphologies commonly observed experimentally and are discussed to give some practical insights on how to control the growth of holes and hillocks in thin films.

Conformality limitations and film composition variations inside features for films deposited at low pressures are explained using examples of studies which combine transport and reaction simulations of deposition processes with experimental work. In the first example, the use of film profile information to decide between two kinetic models for the low pressure chemical vapor deposition (LPCVD) of SiO2 from tetraethoxysilane (TEOS) is described. In the second example, a simple but useful model for the plasma enhanced chemical vapor deposition (PECVD) of silicon dioxide from TEOS/oxygen mixtures (PETEOS) is discussed. In the third example, composition profiles in sputter deposited (PVD) Ti-W films are explained in terms of titanium re-emission. Combined simulation and experimental studies of film profiles and composition profiles in features is a valuable tool in efforts to arrive at useful kinetic and transport models.

In situ, ultrahigh vacuum anneals were performed to induce Cu reflow at 500°C following deposition of Cu films and a Ta barrier layer on 1 μm wide by 1 μm deep trenches. Transmission electron micrograph cross-sections show profiles which suggest that grain boundaries and surface energy anisotropy significantly affect reflow. The extent of reflow is dependent on the structure of grain boundary-surface intersections, and the surface profile consists of regions of low curvature within grains and with sharp discontinuities in curvature at grain boundaries, a structure that inhibits surface diffusion. We present results showing how the surface diffusion mediated reflow varies with grain boundary groove angle and position, and compare these results with finite-element simulations that model surface diffusion-driven reflow.

Numerical simulations are used to predict microscopic topography of deposited metal films in ionized magnetron sputter deposition experiments. The simulation takes into account ionized and neutral fluxes as deposition material sources and re–sputtering and re–deposition of the deposited material due to the ion bombardment. The shock-tracking algorithm is used to advance the moving surface boundary. Applications of this deposition technique include lining and filling of trenches, vias and dual damascene.

There is a considerable amount of evidence for the dependence of microstructure and surface morphology of films obtained by conventional thermally activated chemical vapor deposition on mass transport phenomena. This paper addresses the simulation of the influence of conveclive-diffusivc transport phenomena as well as homogeneous reactions on the deposition profile of silicon films in the case of some characteristic topological effects. For that purpose, the Si-H chemical system including both homogeneous and heterogeneous reactions is used. The evolution of the deposition profile during vapor growth is simulated for various surface defects such as a step or a trench. The influence of basic processing parameters such as temperature or flowrate is discussed.

Deposition of a thin film in the LPCVD process has been simulated by a Monte–Carlo method based on a simple model taking into account the desorption, the surface reaction, and the surface migration of the film precursor. The model has been used for the simulation of the film profile obtained in a narrow and deep trench and of the film crystallinity on a flat surface. The simulation results describe successfully those obtained by experiments under various process conditions.

Titanium (Ti) is widely used to improve the contact resistance with the underlying substrate in contact or via filling applications. As the critical dimensions of the electronic devices fall below 0.5 micron and the aspect ratio of the contact holes increases, it becomes more difficult to obtain a good coverage at the bottom of the contact. In our continuing effort to develop a better technology for the metallization of sub-micron liners [1, 2] using the sputtering process, we have constructed a feature scale model and have initiated an extensive study on the coverage of the contact liners. Our objective is to search for an alternative process technique to replace the current, widely-used collimated sputtering with the help of a realistic model that accounts for all important deposition phenomena such as the effect of the working pressure and the ion bombardment. Since the mean free path of the deposited atoms, particularly in a low pressure system, is comparable to the surface feature of the substrate, particle methods are more appropriate for the modeling of thin film formation phenomena and predictions of microstructures than the continuum approach. This is specially true in the case of sputter-deposition. In this paper, a simple two dimensional molecular dynamics model is used to predict the growth of the thin film. It allows us to study the effects of the physical parameters such as working pressure and ion energy as well as the influence of the geometric configuration, e.g., distance between the target and the substrate and the size of the contact hole. Simulations for long-throw/low-pressure sputtering and ionized magnetron sputtering are presented together with the experimental results for collimated sputtering.

A model for the texture evolution of a growing film was developed, where the tendency for the lowest possible surface energy was taken as the driving force. Two main directions were considered in the orientation distribution of the columns. An initially isotropic distribution was found to evolve to a bimodal one and finally to a distribution with a sharp peak around the orientation with the smallest surface energy. For specific initial conditions the texture turns over from one orientation to the other one. The calculated textures showed a continuous sharpening around orientations with a local or absolute minimum in the surface energy. Various calculated phenomena concerning the texture evolution could be observed on PVD deposited TiN coatings.