(A) study on the ald modeling for process design of multi-component thin films

Abstract

Atomic layer deposition (ALD) enables the precise deposition of ultra-thin and highly conformal thin films with controlled composition, as a result of self-limited surface reactions. Therefore, ALD has become an essential technique for nano-scale thin films fabrication in the microelectronics industry and its applications have been extended to multi-component thin films, including high-k materials. However, the use of ALD to deposit multi-component thin films has been limited by the inability to control the initial stages of deposition, which occurs in a transient regime. Until the exposed substrate is covered by more than an atomic monolayer of the film being deposited, thickness does not have a linear relationship with the number of deposition cycles. Consequently, during the deposition of multi-component thin films, such as nano-mixed and nano-laminated films, ALD loses its most important benefit of digital thickness controllability. Further, variable growth rates complicate adjustments of the film composition. Therefore, we suggested the ALD film growth model to predict the thickness and composition of multi-component thin films. This method needed an accurate extraction of parameters through experiment. However, existing method had shown problems in accuracy and convenience. Thus, an improved method was investigated in this study. To extract precise area reduction ratios, a super-cycle concept was introduced to enhance the accuracy of extracted parameters by focusing on precursor adsorption on heterogeneous surfaces. This improved parameter extraction method was formulated and applied to $Al_2O_3-TiO_2$ binary oxide films for validation; the estimated composition was consistent with the experimental results. And then, partially saturated ALD was investigated to achieve composition adjustment in multi-component thin films. This method controlled the composition using variations in the precursor injection time rather than variation in the cycle ratios. An ad...