Role of idealized surface representations in computational AS-ALD: Assessing trimethoxypropylsilane blocking performance for Al2O3 deposition with crystalline SiO2

Area-selective atomic layer deposition (AS-ALD) using small molecule inhibitors (SMIs) holds promise for thin-film patterning but suffers from selectivity loss over multiple cycles. This study explores the mechanisms causing this loss during Al2O3 growth with SiO2 as the non-growth surface, inhibited by trimethoxypropylsilane (TMPS). We analyze SMI adsorption, maximum coverage, precursor penetration, and reactions with surface hydroxyl groups. Using a crystalline α-quartz slab to model SiO2, we compare outcomes to previous amorphous surface studies to evaluate the validity of idealized models. We find that TMPS achieves inhibitor densities similar to those in amorphous models but does not fully saturate surface hydroxyls. Precursor access through the inhibitor layer depends on molecular size, adsorption energetics, and dimerization. Larger species like TDMAA and DMAI are effectively blocked, while smaller ones such as TMA and AlCl3 reach exposed OH groups, leading to ALD reactions at these sites and selectivity loss. Although findings generally align with those from amorphous surface models, distinctions emerge: Amorphous models better reflect realistic OH densities and bonding diversity, capturing surface chemistry more accurately. Thus, while idealized (crystalline) models inform general SMI behavior, amorphous models are crucial for a more comprehensive understanding of AS-ALD selectivity and inhibitor mechanisms. Both approaches should be integrated for future studies.