Feedback control of film composition and microstructure in high-kappa thin film growth

Fabrication of smaller feature size metal-oxide-semiconductor (MOS) devices typically requires tight control of thin film properties such as film composition and microstructure. In this dissertation, the implementation of feedback control on chemical vapor deposition (CVD) processes (especially for high-kappa materials) is studied to control variables such as impurity concentration, thickness and surface roughness.; Initially, we present a methodology for real-time control of thin film carbon content (which is considered as impurity in the film) in a plasma-enhanced chemical vapor deposition (PECVD) process using combination of on-line gas phase measurements obtained through optical emission spectroscopy (OES) and off-line (ex-situ) measurements of film composition obtained via x-ray photoelectron spectroscopy (XPS). Using this approach, a real-time control system is developed and implemented on an experimental high density PECVD system to demonstrate the effectiveness of real-time feedback control of carbon content. Experimental results of depositions and XPS analysis of deposited thin films demonstrate the advantages of operating the process under real-time feedback control in terms of robust operation and lower carbon content.; Then, motivated by recent experimental results on the growth of high-kappa dielectric thin films using PECVD, a multi-component kinetic Monte Carlo (kMC) model is developed for a conceptual deposition process which involves multiple gas phase species and is influenced by both short-range and long-range interactions. The dependence of the surface microstructure of the thin film on the deposition conditions is studied. Furthermore, kMC model-based feedback control schemes which use the substrate temperature to control the final surface roughness of the thin film are proposed.; Since kMC models are not available in closed-form and computationally expensive, we develop a systematic method for the construction of linear stochastic partial differential equation (PDE) models for feedback control of surface microstructure in thin film deposition. A linear stochastic PDE model is constructed for a generic one-dimensional thin film deposition process using surface snapshots generated by kMC simulations. An optimization-based feedback controller is designed using the constructed stochastic PDE model and applied to the kMC simulation of the deposition process to control the surface roughness.; This stochastic PDE approach is then extended to two-dimensional case and applied to multivariable control problems. A 2D linear stochastic PDE model is initially constructed which describes the spatio-temporal evolution of the film surface. A multivariable predictive control algorithm is developed which uses a finite-dimensional approximation of the stochastic PDE model to regulate the thin film thickness and surface roughness at desired levels at the end of the deposition.