Increased CMOS logic and memory scaling has forced chip architectures into 3D design spaces such as finFET, GAA, nanosheet, etc. These applications require epitaxial structures to be grown in complicated and restrictive geometries which are composed of an increasingly different range of materials. Epi growth selectivity to conventional dielectric materials such as SiO₂ and SiN_x as well as less conventional materials like SiO_xN_1-x, Al₂O₃, and SiOCN must be maintained as usual. However changing demands on epi doping levels and thermal budgets places a large burden on conventional selectivity approaches, which most-often rely on higher HCl partial pressures or HCl etch-back steps. Cl₂ etch-back processes have also been used in limited cases as an alternative to HCl, but apart from that, new selective epi growth technology has become somewhat stagnant.

The 3D nature of advanced logic architectures means that epi must also be grown on multiple crystal surface orientations like the (100), (110), and (111). For epitaxial growth processes, which by-definition are highly sensitive to the starting surface structure, this represents unique challenges to epitaxial engineers. New approaches to achieve orientation selective growth is now also paramount. Each surface orientation can have different growth rates, etch rates, and dopant incorporation efficiencies which can result in undesirable morphology, defects, and inhomogeneous chemical compositions. Depending on the application the growth rate may need to be maximized on one surface and minimized on another, for example to facilitate “bottom-up” growth from the (100) surface. In other instances, the growth rates may need to be matched as closely as possible, for example to achieve symmetric “diamond-shaped” epi-growth on a Si fin.

These issues are at a fundamental level related to surface chemical interactions and finding new solutions will require a molecular-level understanding. In this work we present a fundamental chemical view of these challenges using both theoretical (density functional theory) calculations and in-situ surface analytical results. We analyze the reactive sites on dielectric surfaces and how various deposition precursors interact with these sites. Based on these results we will discuss alternative and/or novel routes to achieving selective epitaxial growth. We apply similar methods to analyzing the structure and reactivity of the Si (100), (110), and (111) surfaces and how deposition precursor molecules react on each surface. Our results reveal why changing Si surface orientations can result in such varied reactivity and ultimately in differing growth and etching rates. We similarly use these results to suggest methods for controlling facet/plane selective epitaxy and/or etching. Central to both dielectric and facet selectivity is precursor engineering, which entails designing new molecules to customize their reactivity (or lack of reactivity) toward specific dielectric surfaces or surface orientations.