Process Control for Spatial Atomic Layer Deposition

Abstract

Within the semiconductor industry, interfacial engineering at the nanoscale has been developed to manufacture transistors with feature sizes on the order of single nanometers. As an integral technique for these advances, atomic layer deposition (ALD) is a thin-film deposition technique that boasts sub-nanometer control of the thickness and chemical composition of the deposited materials. However, it is challenging to translate the vacuum chambers and long process times of ALD processes into the manufacturing lines of large-scale applications, such as displays, batteries, solar cells, and catalysts. As a variation of conventional ALD, spatial atomic layer deposition (SALD) maintains the same precise deposition control while being a faster and more scalable technique. While many SALD system designs have been reported, few have demonstrated deposition on large, complex geometries, which are required for many applications. To help advance SALD into these large-scale commercial applications, this dissertation investigates the effects of process control on SALD systems through a mechatronic experimental system and a computational model to form the elements of a digital twin.

The first focus of this thesis is the design and implementation of a novel, mechatronic SALD system to enable studies on the impact of process parameters on the deposited thin film. Sensors and actuators are used to actively maintain the gap size and parallel alignment during the deposition process through multiple-axis tilt and closed-loop feedback. Digital control of geometric process variables with active monitoring is facilitated with a custom software control package and user interface. SALD of titanium dioxide (TiO2) thin films is performed to validate self-limiting deposition with the system. A novel multi-axis printing methodology is introduced using x-y position control to define a customized motion path, which reduced variations in the film thickness from 8% to 2%.

As the second focus, a 3D computational model that incorporates laminar-flow fluid mechanics and transport of diluted species is developed to provide insight into the velocity streamlines and partial-pressure distributions within the process region of a close-proximity SALD system. The outputs of this transport model are used as the inputs to a surface reaction model that simulates the self-limiting chemical reactions. These coupled models allow for prediction of the film thickness profiles as they evolve in time, based on a relative depositor/substrate motion path. Experimental validation and model parameterization are performed using our mechatronic SALD system, which enables the direct comparison of the simulated and experimentally measured geometry of deposited TiO2 films. Characteristic features in the film geometry are identified, and the model is used to reveal their physical and chemical origins. The influence of custom motion paths on the film geometry is also experimentally and computationally investigated.

As the third focus, a reduced-order COMSOL Multiphysics® model is introduced that can predict the location of precursors in the process region. The model is used to validate the precursor location and consequential process quality during deposition. This can be of particular importance for substrate surfaces that are highly irregular or for manufacturing conditions where external factors such as temperature and ambient air speeds could change dynamically. The development of the model is discussed, and an initial experimental validation of the model is demonstrated.

Overall, this thesis focuses on understanding and improving the process control of SALD so that nanoscale interfacial engineering can be implemented for a wide range industries and applications.