UCLA

Semiconductor manufacturing comprises nearly 500 processing steps, where products rely on stringent design criteria to have high-performance characteristics. One of these processing steps includes the fabrication of high-$\kappa$ oxide films on the surfaces of transistors to minimize current and heat losses, and short-channel effects, which are detrimental to semiconductor longevity. These films demand thicknesses in the nanoscale that are constructed using sequential cycles of atomic layer deposition (ALD) and atomic layer etching (ALE), where precise monolayers of substrate film are deposited and exhibit self-limiting behavior. However, notable challenges in industrial practice include maintaining the accuracy of the deposition and etching processes and the uniformity of the films that are produced, identifying the operating conditions that contribute to optimal product conformation, and developing reactors that maximize the productivity of these atomic layer processes. Additionally, there is insufficient and available data for these processes in industry, which makes their characterization and optimization an obstacle for researchers. Thus, \textit{in silico} modeling has paved the way for producing data that is reflective of data observed in industrial practice. This simulated data is produced through a multiscale computational fluid dynamics framework that combined microscopic, mesoscopic, and macroscopic phases throughout various time and length scales. This work encompasses several disciplines from reaction characterization through \textit{ab initio} molecular dynamics simulations, rudimentary chemical kinetics laws, and kinetic Monte Carlo methods, reactor optimization and design through computational fluid dynamics, and feedback-based run-to-run and online process control with an application to machine learning for a plethora of atomic layer processes.