Process technology computer aided design (TCAD) is one of the essential drivers of the semiconductor industry, as it allows to predict the feasibility of fabrication of future devices without the need for expensive and time-consuming experiments. For many decades, process TCAD simulations were built on analytical models which were derived based on many years of experiments and measurements on silicon. However, as the industry moves away from silicon due to its physical limitations, and into a quasi-exploratory phase of new materials and geometries, a rethink in the way process models are devised is required. Essentially, a multi-scale approach needs to be investigated, whereby atomistic material models are used to replace the knowledge missing due to a lack of experiments. Such material models are used to provide static as well as dynamic parameters and quantities for continuum models which calculate the shape and doping profile on a more macroscopic scale. This project aims to expand on the state-of-the-art in process TCAD by introducing two additional material representations, one for topography simulations such as deposition and etching and a second one for volume problems such as implantation and diffusion.
State-of-the-art process simulators are built on a framework which defines surfaces implicitly, which makes it very difficult to store material information. In this project, the research will be focused on finding a solution which merges an implicit representation of the surface with local material information on the same grid and data structure. This will remove the need for computationally expensive and error prone conversions between implicit and explicit representations currently being applied. We will then investigate the application of this framework on physical models for highly relevant processes such as atomic layer deposition (ALD) and sputter etching, which are considerably dynamically governed by processes which have to be resolved on the atomic scale.
Device TCAD framework developers have already realized that a multi-scale approach to device simulations is essential, which is why the most advanced commercial tools provide an interface from ab-initio to compact modeling of advanced devices. This type of thinking is also essential for process TCAD, which is what we will investigate in this project, together with our industry partner SILVACO. Our research will be aimed at developing a combined molecular dynamics (MD) and level set (LS) framework to enable ab-initio predictive doping activation and motion especially following a variety of ion implantation conditions, in non-standard and not well characterized materials like for instance silicon carbide (SiC). SiC is a material which is likely to replace silicon in many high voltage and high temperature applications, due to its improved efficiency over silicon. We will then apply this multi-scale approach to investigate doping through diffusion in materials beyond silicon, such as germanium, which are likely to be used as the channel in post-silicon devices, beyond the 3nm technology node. This project will set the academic and industrial partner on the path to be at the forefront of process TCAD research since the integration of new materials in everyday devices is an inescapable reality.
