J. E. Rubio

Modeling arsenic deactivation through arsenic-vacancy clusters using an atomistic kinetic Monte Carlo approach

A comprehensive atomistic model for arsenic in silicon which includes charge effects and is consistent with first-principles calculations for arsenic-vacancy cluster energies has been developed. Emphasis has been put in reproducing the electrical deactivation and the annealed profiles in preamorphized silicon. The simulations performed with an atomistic kinetic Monte Carlo simulator suggest a predominant role of the mobile interstitial arsenic in deactivation experiments and provide a good understanding of the arsenic behavior in preamorphized silicon during annealing.

Physical modeling and implementation scheme of native defect diffusion and interdiffusion in SiGe heterostructures for atomistic process simulation

In order to simulate the diffusion kinetics during thermal treatments in SiGe heterostructures, a physically-based atomistic model including chemical and strain effects has been developed and implemented into a nonlattice atomistic kinetic monte carlo (KMC) framework. This model is based on the description of transport capacities of native point defects (interstitials and vacancies) with different charge states in SiGe alloys in the whole composition range. Lattice atom diffusivities have been formulated in terms of point defect transport, taking into account the different probability to move Si and Ge atoms. Strain effects have been assessed for biaxial geometries including strain-induced anisotropic diffusion, as well as charge effects due to strain-induced modifications of the electronic properties. Si-Ge interdiffusion in heterostructures has been analyzed from an atomistic perspective. A limited set of physical parameters have been defined, being consistent with previously reported ab initio calculations and experiments. The model has been implemented into a nonlattice KMC simulator and the relevant implementation details and algorithms are described. In particular, an efficient point defect mediated Si-Ge exchange algorithm for interdiffusion is reported. A representative set of simulated interdiffusion profiles are shown, exhibiting good agreement with experiments.In order to simulate the diffusion kinetics during thermal treatments in SiGe heterostructures, a physically-based atomistic model including chemical and strain effects has been developed and implemented into a nonlattice atomistic kinetic monte carlo (KMC) framework. This model is based on the description of transport capacities of native point defects (interstitials and vacancies) with different charge states in SiGe alloys in the whole composition range. Lattice atom diffusivities have been formulated in terms of point defect transport, taking into account the different probability to move Si and Ge atoms. Strain effects have been assessed for biaxial geometries including strain-induced anisotropic diffusion, as well as charge effects due to strain-induced modifications of the electronic properties. Si-Ge interdiffusion in heterostructures has been analyzed from an atomistic perspective. A limited set of physical parameters have been defined, being consistent with previously reported ab initio calculat...

Ion-beam amorphization of semiconductors: A physical model based on the amorphous pocket population

We introduce a model for damage accumulation up to amorphization, based on the ion-implant damage structures commonly known as amorphous pockets. The model is able to reproduce the silicon amorphous-crystalline transition temperature for C, Si, and Ge ion implants. Its use as an analysis tool reveals an unexpected bimodal distribution of the defect population around a characteristic size, which is larger for heavier ions. The defect population is split in both size and composition, with small, pure interstitial and vacancy clusters below the characteristic size, and amorphous pockets with a balanced mixture of interstitials and vacancies beyond that size.

Comprehensive model of damage accumulation in silicon

Ion implantation induced damage accumulation is crucial to the simulation of silicon processing. We present a physically based damage accumulation model, implemented in a nonlattice atomistic kinetic Monte Carlo simulator, that can simulate a diverse range of interesting experimental observations. The model is able to reproduce the ion-mass dependent silicon amorphous-crystalline transition temperature of a range of ions from C to Xe, the amorphous layer thickness for a range of amorphizing implants, the superlinear increase in damage accumulation with dose, and the two-layered damage distribution observed along the path of a high-energy ion. In addition, this model is able to distinguish between dynamic annealing and post-cryogenic implantation annealing, whereby dynamic annealing is more effective in removing damage than post-cryogenic implantation annealing at the same temperature.

Fermi-level effects in semiconductor processing: A modeling scheme for atomistic kinetic Monte Carlo simulators

Atomistic process simulation is expected to play an important role for the development of next generations of integrated circuits. This work describes an approach for modeling electric charge effects in a three-dimensional atomistic kinetic Monte Carlo process simulator. The proposed model has been applied to the diffusion of electrically active boron and arsenic atoms in silicon. Several key aspects of the underlying physical mechanisms are discussed: (i) the use of the local Debye length to smooth out the atomistic point-charge distribution, (ii) algorithms to correctly update the charge state in a physically accurate and computationally efficient way, and (iii) an efficient implementation of the drift of charged particles in an electric field. High-concentration effects such as band-gap narrowing and degenerate statistics are also taken into account. The efficiency, accuracy, and relevance of the model are discussed.

An improved molecular dynamics scheme for ion bombardment simulations

We have developed a method that reduces the CPU time required in molecular dynamics simulations of ion bombardment processes. This method is based on the selective integration of the particles of the system depending on their energy. Low energy particles are integrated less frequently than the high energy ones. In order to test our scheme we have carried out simulations of Ar+ bombardment of Si(100) at 300 K. Using this method gain factors up to 5.9 in computation speed have been achieved. The accuracy of the results was satisfactory in terms of total energy conservation and of the description of the individual trajectories of the atoms along the simulation.

Molecular dynamics simulation of amorphous silicon sputtering by Ar+ ions

Abstract Molecular dynamics simulations of the sputtering of amorphous silicon by 1 keV Ar + ions are reported. The amorphous sample was prepared by melting and subsequently quenching an 8000 atom (100) Si crystal. The results obtained for the sputtering yield, kinetic energy distribution, layer ejection yield and atoms per single ion ratios are compared with those obtained for silicon crystalline samples at 0 K and at 300 K.

Molecular dynamics study of the fluence dependence of Si sputtering by 1 keV Ar+ ions

The fluence dependence of the Si sputtering by 1 keV Ar+ has been studied by molecular dynamics simulations. To this purpose, previously amorphized samples with different initial argon concentration have been ion bombarded and the sputtered atoms have been analyzed. The calculated sputtering yield increases with the argon content according to the experimental results. The mechanisms involved in this sputtering enhancement are discussed.

Dose effects on amorphous silicon sputtering by argon ions: A molecular dynamics simulation

We have investigated, using molecular dynamics techniques, the sputtering yield enhancement of amorphous silicon produced by argon ion accumulation within the target. Several amorphous silicon samples, with different argon contents, were bombarded with 1 keV argon ions at normal incidence. To study the influence of the target structure, we considered samples with different argon arrangements, either uniformly distributed or within solid bubbles. We have observed that silicon sputtering yield increases linearly with dose until steady state conditions are reached. This enhancement is produced by the shallow argon atoms through the weakening of Si–Si bonds. We have also observed that argon release takes place even long after the end of the collisional phase, and it is produced by ion-induced desorption and bubble destabilization. This enhanced argon yield determines the dose where target saturation and steady state conditions are reached.

Ion implant simulations: Kinetic Monte Carlo annealing assessment of the dominant features

The atomistic physically based kinetic Monte Carlo method has been used in conjunction with the binary collision approximation (BCA) to elucidate the implant mechanisms most relevant for modeling transient-enhanced diffusion (TED). For the cases studied, we find that: (i) The spatial correlation of the interstitial, vacancy (I,V) Frenkel pairs is not critical, (ii) the interstitial supersaturation in simulations which include full I, V profiles or only the net I–V is the same, (iii) quick and noisy BCA implant I, V distributions can be directly used (or after smoothing them out) as they can still yield accurate annealing simulations, and (iv) when there is an impurity concentration comparable to the net I–V excess, the full I and V profiles have to be used in order to correctly reproduce the impurity clustering/deactivation. Finally, some practical implications for TED simulations are drawn.