Surface Reaction Kinetics for Inherent Selective Atomic Layer Deposition of Tantalum Oxide

Abstract

The chemical principal and mechanisms that enable selective atomic layer depositions are gaining rapid growing interests, which have unlocked attractive avenues for the development of novel nanostructures by depositing atoms at desired surface locations. It has found versatile applications in emerging fields beyond semiconductor industry. Yet with the continuing downscaling, it is important to expand approaches for selective ALD with atomic scale precision on nanoscale features. The self-driven selective growth is needed, modifying specific surface down to atoms, to enable unprecedented capability of tuning the structure-property relationship for emerging applications. In this talk, the inherently selective atomic layer deposition processes will be discussed. Previously, we have demonstrated facet-selective ALD processes, which based on intrinsic differences of precursors chemisorption on terraces or step edges of nanoparticles. To expand the inherently ASD, the acidity and alkaline, surface electronegativity, lattice strains could be exploited to achieve this inhibitor-free selective deposition process. Tantalum oxide was studied on a series of oxide substrates. Although the oxides have -OH groups on the surface and proposed to have similar nucleation sites, there are long nucleation delays on basic oxides. The H-transfer reaction is a key factor to influence the reaction barrier. It is hard to nucleate on basic substrates because the H-transfer reaction is blocked. Another demonstration is the redox-coupled inherently selective ALD for self-alignment of tantalum oxide on SiO2/Cu nanopatterns. By adding an in-situ ethanol reduction pulse before each traditional binary ALD cycle, and the ‘reduction-adsorption-oxidation’ ALD process increases the selectivity. Self-aligned manufacturing on nanoscale Cu/SiO2 patterns without excessive mushroom growth at the edge and undesired nucleation defects on the Cu region. The process can be reliably repeated to yield more than 5 nm-thick Ta2O5 on the SiO2 region, while no undesired deposition occurs on Cu patterns. In addition, an anisotropic growth model with the dynamical competition of expansion and dissociation of the nucleus is proposed to nucleation delay are quantitively predicted and the model provides a practical method to evaluate the selectivity of ALD theoretically. It provides a new strategy for inherently selective ALD, which will expand the selective toolbox of nanofabrication for next-generation nanoelectronic applications.