(371d) Multiscale Modeling of Atomic Layer Deposition Process for Optimal Operation : Integration of Kinetic Monte Carlo and Computational Fluid Dynamics

Due to the exponentially increasing data due to the continuous development of big data, deep learning, artificial intelligence, etc., there is a great demand for high integration of semiconductor memory devices. To improve integration, reduce the thickness of the thin film in the process or increase the aspect ratio of the DRAM capacitor. Accordingly, the importance of the deposition process is increasing. The chemical vapor deposition (CVD) and physical vapor deposition (PVD) methods used previously have difficulty in fine control of thin film thickness, making it difficult to produce thin films necessary for high integration. However, the atomic layer deposition (ALD) method is known to be capable of creating a thin layer through a self-limiting surface treatment method that forms a thin film layer by layer with a thickness of atomic thickness.^1-3 Consequently, ALD has shown significant potential in semiconductor industry, particularly in the field of producing 3D ultrathin films that adhere closely to surfaces.⁴ Recently, there has been increasing research at both experimental and industrial scales to optimize and commercialize ALD technology.⁵ However, these experimental approaches still suffer from limited throughput due to time and cost.⁶

Therefore, we developed a multiscale simulation model that can provide a robust guide for process operation and control. We implemented a 2-way coupled multiscale model by integrating the microscopic kinetic Monte Carlo (kMC) for reaction on the substrate surface and the macroscopic computational fluid dynamics (CFD) model for gas transport. The kMC model was implemented using Zacros, while the CFD model utilized the ANSYS Fluent program. Integration of these two scales was achieved through Python code leveraging packages provided by each respective program. Our ALD chamber model includes multiwafers to discuss issues in real industrial processes. In addition, each wafer is also divided into multiple zones and individual kMC calculations are performed in each zone. We optimized the ALD process operating conditions by applying the Si₃N₄ reaction to the model. We discussed conditions for minimizing the growth per cycle (GPC) distribution of wafers that vary by location using the rotation speed and precursor pulse time of the wafer as parameters, and also discussed operating conditions that can minimize the amount of precursor material used. Although this study was limited to the Si₃N₄ ALD process, our multiwafer ALD chamber multiscale model will be applied to a variety of thin film materials in the future to help optimize operating conditions at the actual process level.

¹ Tanner, C. M., Perng, Y. C., Frewin, C., Saddow, S. E., & Chang, J. P. (2007). Electrical performance of Al2O3 gate dielectric films deposited by atomic layer deposition on 4H-SiC. Applied Physics Letters, 91(20).

² George, S. M. (2010). Atomic layer deposition: an overview. Chemical reviews, 110(1), 111-131.

³ Shirazi, M., & Elliott, S. D. (2014). Atomistic kinetic Monte Carlo study of atomic layer deposition derived from density functional theory. Journal of computational chemistry, 35(3), 244-259.

⁴ Johnson, R. W., Hultqvist, A., & Bent, S. F. (2014). A brief review of atomic layer deposition: from fundamentals to applications. Materials today, 17(5), 236-246.

⁵ Meng, X., Byun, Y. C., Kim, H. S., Lee, J. S., Lucero, A. T., Cheng, L., & Kim, J. (2016). Atomic layer deposition of silicon nitride thin films: a review of recent progress, challenges, and outlooks. Materials, 9(12), 1007.

⁶ Elliott, S. D. (2013). ALD simulations. Atomic Layer Deposition for Semiconductors, 47-69.