Quantitative modeling of substrate velocity effects on deposition efficiency and precursor consumption in spatial ALD

Abstract

Spatial atomic layer deposition (Spatial ALD) enables highly efficient thin film deposition through increased substrate velocity. However, rapid substrate movement introduces challenges such as reduced ALD growth per cycle (GPC) and increased contamination from unwanted chemical vapor deposition (CVD), necessitating higher precursor and gas consumption. In this study, a quantitative model for analyzing CVD as a functional relationship with substrate velocity is developed, based on the kinetic theory of ALD adsorption, complemented by a coupled fluid dynamics model for precursor mass transfer and reactions. The model identifies conditions that maintain ALD performance and minimize contamination at rapid substrate velocity by enhancing gas flow rate within the micro-gap to reduce entrainment. Comprehensive analysis of coverage, CVD, utilization, and consumption revealing that reducing the micro-gap size is a more economical strategy for achieving higher film deposition rates. Furthermore, an active learning framework driven by Gaussian process regression and self-defined acquisition strategy efficiently identifies the optimal process conditions at three substrate velocities. Requiring only 115 calculations out of 77 million possible combinations, the optimization achieves a 31 % reduction in gas consumption and a 15 % reduction in precursor consumption at a substrate velocity of 0.15 m/s. Experimental validation confirms these conditions yield films with high GPC and quality comparable to thermal ALD, underscoring the model’s efficacy. This study provides valuable insights for achieving cost-effective spatial ALD at higher substrate velocities across diverse conditions.