The study of degradation in Metal-Oxide-Semiconductor (MOS) transistors – specifically in the context of the negative bias temperature instability or random telegraph noise – has brought to attention the trapping of carries from the channel into defects inside the oxide. Models based on the standard Shockley-Read-Hall formulation usually fail to accurately describe the dynamics of these trapping processes, as obtained by experiments.
Electron transfer processes are of enormous importance in many fields of chemistry and, since the first half of the twentieth century, have been receiving a lot of attention. An intense study of the available literature has shed light on the physical processes involved in the tunneling of electrons to and from localized states, as is the case in MOST degradation. It shows that the thermal movement of the lattice has to be properly taken into consideration leading to a complex dependence of the capture and release rates on temperature and the electric field in the oxide. A broad spectrum of models for electron transfer is available in literature, ranging from formulations based on classical statistical mechanics (Marcus theory) to quantum theory based descriptions, based on overlaps of vibrational wave-functions (Huang-Rhys model).
The basic parameters to properly describe electron transfer are hardly accessible from experiments and have to be extracted from atomistic calculations including the electronic structure, such as Density Functional Theory (DFT) or quantum chemistry methods. State-of-the-art DFT-based electronic structure calculation methods such as SP-Korringa-Kohn-Rostoker (SP-KKR), Linearized Augmented Plain Waves (LAPW), and numerical-local orbital based DFT were evaluated in order to apply them to defect calculations.
Due to the high purity of oxides in modern semiconductor technology, the expected defects are self-defects of the materials or impurities of species that are part of the processing. As a promising defect candidate, hydrogen bridges in an amorphous silica lattice have been intensively studied using the embedded cluster method. In this approach, large (nanometer sized) atomic structures are separated into one part treated quantum-mechanically (using DFT) and one described by classical empirical potentials. The calculations were executed using the GUESS code, developed at the University College London by the group of Prof. Alex Shluger.