ALD/ALE 2021 Session AF2: Precursors and Chemistry: Simulation, Modeling, and Theory of ALD

Self-terminating chemistry of atomic layer deposition (ALD) process enables one to grow pinhole free conformal thin films on high-aspect-ratio (HAR) structures. ALD has attracted ever more attention in diverse applications, such as microelectronics and nanostructured catalyst preparation.¹ Yet, it is essential to optimize ALD process parameters for the conformal deposition especially on HARs. Our previous study investigated the effect of experimental parameters on conformality using saturation profiles of archetypical trimethylaluminum-water ALD processes in lateral HARs with an aspect ratio of typically 10000:1.²

The investigation on the effect of process parameters on ALD conformality is continued by simulating ALD saturation profiles with two modeling approaches: a MATLAB simulation based on a diffusion–reaction modeling (Model A)^3,4 and a Python simulation based on a ballistic transport modelling (Model B).^5,6 These simulated saturation profiles are compared to each other in approximately the same condition. While the main features of the simulated saturation profiles are similar, differences are found in the value of 50% thickness penetration depth (PD_50%) and slope at PD_50%, as well as the shape of the tail region.

Sticking coefficient of ALD reactants describes ALD growth kinetics. A recent study by Arts et al.⁷ reported a method to back-extract the sticking coefficient value from the slope of saturation profile. By using this method, we back-extract the sticking coefficient values from the saturation profiles simulated by Models A and B and compare those values to the ones initially set for running the simulations. Interestingly, for both Models A and B, the sticking coefficient values set for the simulations differ somewhat from the ones back-extracted.

Acknowledgement

The work was supported by the Academy of Finland (ALDI consortium, decision no. 331082). Aleksi Heikkinen is acknowledged for initiating the use of Machball code (Model B).

References:

1 Cremers et al., Appl. Phys. Rev., 2019, 6, 021302.

2 Yim, Ylivaara et al., Phys. Chem. Chem. Phys., 2020, 22, 23107–23120.

3 Ylilammi, Ylivaara and Puurunen, J. Appl. Phys., 2018, 123, 205301.

4 Puurunen and coworkers, manuscript in preparation.

5 Yanguas-Gil and Elam, Theor. Chem. Acc., 2014, 133, 1465.

6 https://github.com/aldsim/machball (accessed March 4, 2021).

7 Arts et al., J. Vac. Sci. Technol. A, 2019, 37, 030908.