I am the first of a group of seven japanese to present our Master’s Thesis on Tuesday afternoon.
It will start at 1pm sharp (GMT+9) in classroom 12-210 of Yagami Campus at Keio University, Japan.

Everyone can attend since it is open to the general public. 
However, had the presentation already being started, please refrain from entering the place.

Abstract:

Ion implantation is a technique commonly employed in the electronic industry to dope semiconductor materials. Dopant ions are accelerated and directed at high speed towards the sample’s surface. When ions penetrate in the semiconductor bulk they generate diverse forms of damage that hinder the dopant activation and enhances diffusion beyond the desired ranges. If the magnitude of this damage is increased, eventually the lattice collapses into amorphous state. This is not necessarily bad because the non-regularity of amorphous phase prevents ion channeling and, what is more important, it is easily wiped out together with the defects by the recrystallization process triggered at not-so-high temperatures. This process reconstructs the lattice leaving a high dopant activation.

However, recrystallization cannot happen in non-amorphized areas. Therefore, the layer just besides the amorphized region that contains a high concentration of defects (but not enough to get amorphized), will not be recrystallized and the damage will persist. These remaining end-of-range defects are the main cause of enhanced transient diffusion during annealing processes and current leakage in the final electronic device. The only way to control such undesirable effects is being able to predict their position inside the bulk by knowing where the amorphous/crystalline interfaces are placed after an implantation process.

We could create a database of position of amorphous/crystalline inter- faces, but to avoid the expensive requisites in time and money that experiments require, science makes use of simulators that predict the outcome of experiments by means of a model.

There have been many models proposed to calculate the span of the amorphization induced by ion implantation. Most of them work reasonably good with unreasonably bad ranges of applicability: they require a lot of fitting or tabulated parameters that have to be obtained by means of experiments, hence we can tell they are more likely designed to reproduce results rather than to predict them.

We have ideated an amorphization model that, by making use of a single fitting parameter, it is able to simulate the amorphization of a silicon bulk subsequent to a ion implantation in a great range of implant conditions. It is based in the displacement suffered by silicon atoms when the energetic ions collide with the lattice. When the average displacement is over certain threshold, the material is considered to be amorphized.

The model has been implemented with a Monte Carlo simulation in three steps. First, we obtain an atomic mixing profile from the information of vacancies generated during binary collision approximation cascades. Second, the mixing profile is convoluted with a gaussian broadening distribution in order to obtain a measure of the local displacement for each atom. At the end, it is shown the critical average displacement in the depth direction necessary to produce amorphization and hence predict the position of the amorphous/crystalline interfaces.

With our method we are able to reproduce the amorphization regions reported by a number of authors in the literature, consisting on a collection of a wide range of doses, energies and implanted species. By successfully simulating them, we demonstrated that this critical displacement is a universal value for silicon to produce amorphization.