Computational fluid dynamics modeling of spatial atomic layer deposition on microgroove substrates

Conformality of atomic layer deposition in microchannels: impact of process parameters on the simulated thickness profile

“…To compare the trends of the ALD thickness profile in different flow regimes in a well-defined way, we varied individual process parameters to make ALD thickness profiles 30 in free molecular flow (Kn >> 1) and transition flow (Kn 1) conditions. 25,27,28 A comparison ≈ was made with a single cycle, so that the channel filling does not influence the trends of the thickness profile. Both the scaled thickness profile and the Type 1 normalized thickness profile were used as a basis for comparison.…”

“…4,19,26,27 Diffusion-reaction models based on Fick's law of diffusion can ‡ Diffusion-reaction models relying on Fick's laws of diffusion are in the ALD literature sometimes somewhat confusingly referred to as "continuum" models; 4,31,36 in this work the term is dedicated to continuum flow conditions where the mean free path of molecules  (m) is orders of magnitude smaller than the limiting feature dimension h (m) (Knudsen number Kn = /h  10 -3 ). 26 flexibly be used in free molecular flow (Kn >> 1) as well as in transition flow (Kn  1) and even in continuum flow (Kn << 1) conditions, 25,27,28 as the effective diffusion coefficient D eff (m -2 s -1 ) can be calculated from the gas-phase diffusion coefficient D A (m -2 s -1 ) and the Knudsen diffusion coefficient D Kn (m -2 s -1 ). 17,18,20 Also Monte Carlo methods have been used for regimes other than free molecular flow, by using the mean free path  as a statistical parameter.…”

“…as (free) molecular flow, Knudsen flow, or Knudsen diffusion. 4,19,26,27 Diffusion-reaction models based on Fick's law of diffusion can flexibly be used in free molecular flow (Kn c 1) as well as in transition flow (Kn E 1) and even under continuum flow (Kn { 1) conditions, 25,27,28 as the effective diffusion coefficient D eff (m À2 s À1 ) can be calculated from the gas-phase diffusion coefficient D A (m À2 s À1 ) and the Knudsen diffusion coefficient D Kn (m À2 s À1 ). 17,18,20 Also Monte Carlo methods have been used for regimes other than free molecular flow, by using the mean free path l as a statistical parameter.…”

Can We Rationally Design and Operate Spatial Atomic Layer Deposition Systems for Steering the Growth Regime of Thin Films?

“…We considered wall thicknesses of 50–1000 μm, exhaust widths of 100–500 μm, deposition gaps of 50–500 μm, precursor flow rates of 75–300 sccm, and flow rate ratios of 1–10, for a channel width of 500 μm (Table ). These ranges were chosen to encompass the values that can be used experimentally for each parameter, ,,, with fabrication limitations guiding the choice of the minimum wall thickness (50 μm). The values considered for each parameter of interest (Table ) were combined following a full factorial design-of-experiments approach, where all the possible combinations between the different parameters generated independent simulation cases.…”

“…Some of the parameters that can be leveraged to minimize the mixing of precursors during deposition of thin films have been identified. For instance, decreasing the gap between the substrate and the deposition head and increasing the flow rate of the separation gas have been shown to prevent precursor mixing in SALD systems for porous, nonporous, − and microgroove substrates. The application of a slight vacuum at the exhausts, which makes them more efficient at purging the precursors, is also known to decrease precursor intermixing, , although some deposition head designs can operate well without vacuum at the exhaust .…”