(291c) Multiscale Simulation of Nanofeature Evolution in Atomic Layer Deposition Processes

Thin-film deposition is a key unit operation in microelectronic device, semiconductor, thin-film solar, and other energy-related manufacturing applications. These processes generally consist of a heated substrate onto which a solid, thin-film is deposited from the products of gas-phase and surface reactions. It is the interplay between gaseous-phase chemical precursor transport and the reaction processes that ultimately determine the composition, thickness, and microstructural properties of the films.

Among these thin-film processing technologies, atomic layer deposition (ALD) is particularly well-suited to modifying the geometry of nanostructured materials, such as porous membranes that serve as templates for the fabrication of nanoengineered devices. The increasing importance of ALD as a thin-film manufacturing technique and recognition that ALD does not always produce perfectly conformal films has motivated recent modeling studies. First-principles modeling has mostly been limited to the initial growth of ALD films because of the computational difficulties associated with applying large-scale molecular dynamics techniques to simulate growth of ALD films over multiple cycles. Gas-phase species transport within nanopores has been modeled by molecular dynamics and other sophisticated techniques. However, the growth and transport models have only been recently been coupled in a complete multiscale simulation [1] capable of predicting film structure over hundreds of ALD cycles.

In this paper, we describe two advances in modeling ALD processes over the cited work. First, an improved lattice Monte Carlo simulator for the atomic-scale growth is presented that reduces the anisotropy of the film characteristics, allowing for more accurate simulations of nanoscale feature evolution in 2D. Analysis of whether the ALD growth processes are stable will be presented in the context of Al₂O₃ deposition. Second, a rigorous ballistic transport model of precursor diffusion inside the high Knudsen-number nanostructures will be presented. This new transport model has the potential for much more accurate estimation of the time scales of ALD growth processes and will be shown to be crucial to accurate simulation of nanofeature evolution. Using the simulator developed, novel processing strategies aimed at maximizing precursor conversion will be developed as well as applications to energy-related nanomanufacturing processes.

[1] Adomaitis, R. A., Development of a multiscale model for an atomic layer deposition process, J. Crystal Growth, (312) 1449-1452 (2010).