Optimal design of novel precursor materials for the atomic layer deposition using computer-aided molecular design

Abstract

Nanomaterials and nanostructures with multi-functional properties found widespread applications such as electronics, optics, and coatings that can be fabricated using Atomic Layer Deposition (ALD). ALD is a vapor phase deposition technique to generate thin films of metals and metal oxides on a substrate. In this process, a precursor, which often comprises of organic functional groups that surround the depositing metal, chemisorbs on the substrate or reacts with the surface sites and with each other. Precursor chemisorption on the substrate leads ALD to be a self-limiting process. Thus, the precursor(s) should be chosen in a way to enhance deposition based on the ALD conditions. For a given application, it is practically impossible to carry out a large number of experiments using numerous precursors with varied deposition conditions to find the one that maximizes the growth rate in ALD. In addition, only existing precursors can be tested experimentally. The overall objective of this work is to develop a computational tool for the in-silico design of precursor materials using adsorbate solid solution theory (ASST). In the first part of this paper, we apply the ASST to derive properties of the functional groups present in the precursor using a new Group Contribution Method (GCM). The GCM is successfully derived and compared with the experimental data from ALD. The method shows good agreement and is useful for the design of novel materials. In the second part of this paper, using the thermodynamic properties as obtained from GCM, we develop a computer-aided molecular design (CAMD) framework for the optimal design of novel precursor materials with enhanced deposition properties for the ALD of metal oxides and metals. The metaheuristic efficient ant colony optimization algorithm (EACO) developed in-house is used for both parts. CAMD is a combinatorial optimization methodology, where molecules with optimal desired properties are generated from functional groups. The precursor selection optimization problem is posed as a mixed integer nonlinear programming problem, which is solved using EACO. The ALD growth kinetics is used as an objective of optimization and a solution is validated with thermodynamic constraints. Along with the number of groups, the temperature is also included in the optimization framework as decision variables. Forty-one novel titanium precursor molecular structures for ALD are generated with growth rates ranging from 1.23 Å/cycle to as high as 1.65 Å/cycle. Thus, these precursors have shown growth rates higher than the known titanium precursors. ALD growth rate is found to be a function of the combination of the precursor functional groups as well as temperature with a complex correlation among them.