Reliability issues including negative bias temperature instability, random telegraph noise, and time dependent break down, are traced back to some sort of tunneling mechanisms into oxide defects. This has been documented by a series of experiments, such as time dependent defect spectroscopy for instance. Also, on the theoretical side, considerable advances have been made regarding the physical description of those defects. In this context, nonradiative multiphonon processes have gained much scientific popularity during the last years, however, they rely on several parameters which depend on the detailed defect structures and are only available from atomistic simulations. For this sort of problems, density functional theory is frequently suggested as the method of choice but is eventually way to slow to mimic the processing of "amorphous'' oxides. Therefore, we have implemented an alternative method, which uses a simulated annealing approach proposed by Tu and Tersoff. One of its advantages is that it is especially suited for the production of amorphous structures. It is based on a Metropolis Monte Carlo (MMC) algorithm, in which the bond energies are considered for the acceptance of the MMC trial step. A schematic of this method is depicted in Figure 1.
Using this method, we have generated amorphous silicon dioxide structures (see Figure 2), which have been verified by pair-correlation functions and angle distributions. Furthermore, the MMC algorithm is also applied to more complex silicon dioxide interfaces, which influence the properties of the nearby oxide. Therefore, we will also investigate presence of defects, such as suboxides and protrusions along with hydrogen- and nitrogen-related defects, in order to realistically reproduce the structural properties of amorphous silicon dioxide interfaces. This will be achieved by additional potentials for these defects, whose energy contribution will be accounted for the MMC trial step. The obtained structures will be then required for further first principles investigation.