Copper diffusion in germanium from first principles : a first step towards high-throughput diffusivity studies in silicon and germanium
Copper diffusion in germanium from first principles : a first step towards high-throughput diffusivity studies in silicon and germaniumPromotor(en): S. Cottenier, J. Vanhellemont /15_MAT13 / Solid-state physics
Needless to say that the versatile properties of doped semiconductors lie at the heart of all advanced electronics – from your smartphone to supercomputers. Achieving ever better control over the purity, crystallinity, dopant level and dopant position is mandatory in order to produce smaller, thinner, less power-consuming and at the same time more performant devices. Whereas most of the research enabling this evolution is done experimentally, during the last years quantum simulations at the atomic scale are more and more used to complement and to a large extent even replace experiments thus leading to a considerable cost saving and reduced development time. Fig. 1 shows an example from recent work at the Center for Molecular Modeling, where density-functional theory was used to predict the most stable lattice site of 70 elements of the periodic table as dopant in germanium. In spite of 40 years of experimental work, for more than half of these elements their lattice site in germanium was not known prior to this computational study (information for the other half was used to verify that these simulations produced correct results). Fig. 1 demonstrates how quantum simulations can speed up research by allowing to select interesting dopants for further experimental study.
Fig. 1: predicted lattice site of 72 elements in Germanium (under review)
The results in Fig. 1 are about static properties: lattice positions of dopant atoms. The next even more important question is about their dynamic properties: what would, for instance, be the diffusivity of a given dopant in germanium? Diffusivity is much harder to predict from first principles. But also harder to measure experimentally. If reliable predictions can be made, this would be very useful information for the semiconductor industry.
goal You will apply a protocol for predicting diffusivity (see Acta Materialia 58 (2010) 1982 and Physical Review B 79 (2009) 054304) to the case of Cu-diffusion in germanium – this is a case that is relevant for Umicore, the industrial partner in this thesis. This protocol has been used predominantly for diffusion in metals so far, where mainly substitutional diffusion occurs. You will need to apply it for interstitial diffusion as well, which is a relevant diffusion mechanism for the very open (diamond) lattice of germanium. You will not only study the physics of Cu-diffusion, you will also use this particular case to automate the work flow and thoroughly test this automation. In this way, you will pave the way for a follow-up work, where the diffusivity of many more elements in germanium, silicon and their alloys will be studied resulting in something as Fig. 1, but now for diffusivity.