Next: 5 Algorithmic Issues Up: An Examination Hydrogen Desorption Previous: 3.1 Theory Contents
In order to look at the problem of desorption of hydrogen from a silicon surface several things needed to be considered. First, we must consider the type of algorithm to use. Diffusion and desorption are time dependent processes with different time scales. Depending on the temperature the time for events to happen may be relatively long (ie. months) or relatively short (ie. seconds). After examining the information available on hydrogen desorption, we feel that the best methodology for attacking this project is Kinetic Monte Carlo (KMC). KMC allows us to convert from Monte Carlo time into ``real time'' for comparison with experimental data. After we decided to use KMC, the next natural step was to implement KMC using the N-fold way [1]. The N-fold way is an efficient choice since the number of types of possible moves is small. In order to implement this we must create a collection of move lists. More details on this feature of the code can be found in Section 6.2.
We also have to make a few simplifying assumptions to tackle this problem. One of the first assumptions was that the silicon surface is fixed. This is mildly unrealistic, since silicon can diffuse at the temperatures we are examining. However, the computational cost is significantly reduced if we fix the silicon surface. The silicon surface with dimer rows is modeled only in the sense that it dictates where the hydrogen atoms are allowed to move. Actual silicon atoms are not modeled. Another surface assumption made is that the B-step geometry is completely linear. In nature this is not often true, since B-steps usually are jagged. We also assume some restrictions on desorption, namely only dimer-paired atoms can desorb. Thus atoms at the base of a B-step cannot desorb since they are not paired.
Chris Siefert and Molly Moore 2002