Area-Selective Deposition of Ruthenium by Area-Dependent Surface Diffusion

Abstract

Area-selective deposition (ASD) enables the growth of materials on target regions of patterned substrates for applications in fields ranging from microelectronics to catalysis. Selectivity is often achieved through surface modifications aimed at suppressing or promoting the adsorption of precursor molecules. Here, we show instead that varying the surface composition can enable ASD by affecting surface diffusion rather than adsorption. Ru deposition from (carbonyl)-(alkylcyclohexadienyl)Ru and H 2 produces smooth films on metal nitrides, and nanoparticles on SiO 2 . The latter form by surface diffusion and aggregation of Ru adspecies. Kinetic modeling shows that changing the surface termination of SiO 2 from −OH to −CH 3 , and thus its surface energy, leads to larger and fewer nanoparticles because of a 1000fold increase in surface diffusion rates. Kinetic Monte Carlo simulations show that even surface diffusion alone can enable ASD because adspecies tend to migrate from high-to low-diffusivity regions. This is corroborated by deposition experiments on threedimensional (3D) TiN−SiO 2 nanopatterns, which are consistent with Ru migrating from SiO 2 to TiN. Such insights not only have implications for the interpretation of experimental results but may also inform new ASD protocols, based on chemical vapor and atomic layer deposition, that take advantage of surface diffusion.

Inherently Area-Selective Atomic Layer Deposition of Manganese Oxide through Electronegativity-Induced Adsorption

“…It is considered that ASD is mainly originated from competition between surface chemisorption, decomposition and so on [33]. To elucidate the origin of selective growth, the density functional theory (DFT) and nudged elastic band (NEB) are adopted to calculate the decomposition energy and energy barriers of ALD precursors on Pt, Cu and SiO 2 surfaces, as shown in Figure 5a.…”

Selectivity Enhancement for Ruthenium Atomic Layer Deposition in Sub‐50 nm Nanopatterns by Diffusion and Size‐Dependent Reactivity

“…[6,24] The impact of surface diffusion and aggregation was recently also demonstrated for Ru ASD by (carbonyl)(cyclohexadienyl)Ru/H 2 CVD at a much lower deposition temperature of 250 °C, driven by high surface energy of Ru and the low surface energy of the methyl-terminated SiO 2 nongrowth surface. [12] The experimental-theoretical study indicated an average diffusion length of 140 nm for Ru adspecies on the methyl-terminated SiO 2 nongrowth substrate. Indeed, diffusionmediated Ru ASD was observed in SiO 2 -TiN line-space patterns with a critical dimension of 45 nm and SiO 2 line height of 60 nm.…”

“…[7][8][9] The roles of gas phase and surface diffusion have long been known in selective epitaxial growth by CVD. [6,7,[10][11][12][13][14][15][16][17][18][19][20] The resulting fluxes reduce deposition on the nongrowth surface and enhance the growth rate on the growth surface. As a consequence, the growth rate and selectivity depend on the pattern dimensions and geometry when the pattern dimensions are comparable or smaller than the characteristic diffusion length.…”

“…As a consequence, the growth rate and selectivity depend on the pattern dimensions and geometry when the pattern dimensions are comparable or smaller than the characteristic diffusion length. [6,12,[21][22][23] For example, diffusion processes in selective Si epitaxy by SiCl 4 /H 2 CVD at 1200 °C are characterized by diffusion lengths Area-selective deposition (ASD) is a promising bottom-up approach for fabricating nanoelectronic devices. However, a challenge is to prevent the undesired growth of nanoparticles in the nongrowth area.…”

“…Depletion of Ru nanoparticles near the interface with the growth surface and a more than two times higher growth rate for ASD compared to CVD is consistent with diffusion of Ru adspecies from the nongrowth to the growth area, where they contribute to Ru growth. [12,25] Almost no Ru deposition was present on the SiO 2 sidewalls, indicating that diffusion can be an ally in defect mitigation.…”