Modified atomic layer deposition of MoS2 thin films

Abstract

As one of the most attractive transition metal dichalcogenides (TMDs), the growth of molybdenum disulfide (MoS2) with industrial compatibility is of great importance. Atomic layer deposition (ALD) has been shown to be a promising method to achieve the growth of high-quality TMD materials. However, MoS2 films deposited by ALD often are amorphous with nonideal stoichiometry and require high-temperature post-deposition annealing. In this study, we introduce a modified ALD recipe using Mo(CO)6 and H2S, resulting in controllable linear growth behavior, a S-to-Mo ratio of 2:1, and crystalline films at a temperature as low as 190 °C. The growth mechanisms and key factors leading to this improvement are proposed and complemented by kinetics calculations. This newly developed methodology relies on aligning the process time with the reaction kinetics of carbonyl disassociation. The MoS2 films prepared herein were shown to be active hydrogen evolution reaction catalysts.

The Importance of Decarbonylation Mechanisms in the Atomic Layer Deposition of High‐Quality Ru Films by Zero‐Oxidation State Ru(DMBD)(CO)3

“…Another study by Liao and Ekerdt of both Ru 3 (CO) 12 and ruthenium bis(di-t-butylacetamidinate) dicarbonyl [Ru( t Bu-Me-amd) 2 (CO) 2 ] precursors observed that growth rate increased with temperature, without evidence of CO decomposition into other inert species such as graphitic carbon, again in line with our proposed mechanism involving simple CO desorption. [83] Further evidence for this mechanism of growth dependence on CO release is found in a recent ALD report of Mo(CO) 6 used to synthesize MoS 2 with H 2 S. [39] While that system contains a different metal center and reacts on a different growth surface, Mo(CO) 6 is a zero-oxidation state precursor containing CO ligands, similar to Ru(DMBD)(CO) 3 . Zeng et al found that at lower temperatures and purge times, the CO ligands were not completely removed and were instead incorporated into the film as impurities.…”

“…[82] Further evidence for this mechanism of growth dependence on CO release is found in a recent ALD report of Mo(CO)6 use to synthesize MoS2 with H2S. [39] While that system contains a different metal center and reacts on a different growth surface, Mo(CO)6 is a zero-oxidation state precursor containing CO ligands, similar to Ru(DMBD)(CO)3. Zeng et al found that at lower temperatures and purge times, the CO ligands were not completely removed and were instead incorporated into the film as impurities.…”

“…More generally, zero-oxidation state precursors have been reported to chemisorb to the surface through a thermally induced, surfacemediated dissociation or dissociative substitution reaction, facilitating rapid nucleation. [35][36][37][38][39] Although this pathway can reduce the self-limiting nature of their reaction with the surface and make them more challenging to apply to ALD, such precursors have been widely used in the CVD literature. [40][41][42] While there are many factors that affect nucleation and growth, the reported correlation between oxidation state and nucleation delay raises questions about the reaction mechanisms taking place.…”

“…More generally, zerooxidation state precursors have been reported to chemisorb to the surface through a thermally induced, surface-mediated dissociation (where precursor ligands remain chemisorbed to the surface) or dissociative substitution reaction (where precursor ligands may be released into the gas phase), facilitating rapid nucleation. [35][36][37][38][39] Although this pathway can reduce the self-limiting nature of their reaction with the surface and make them more challenging to apply to ALD, such precursors have been widely used in the CVD literature. [40][41][42] A range of surface chemical reactions in metal ALD and their relationships to nucleation phenomena have been discussed in greater detail elsewhere.…”

“…Although some discrepancy in GPC can be expected when using different counter-reactants (water, oxygen, hydrazine), the discrepancies in the trends of the GPC as a function of temperature and the growth of Ru islands without any counter reactant merits further investigation. Furthermore, since other studies have suggested that growth using carbonyl-based precursors (such as Mo(CO) 6 , Fe(CO) 5 , W(CO) 6 , Cr(CO) 6 , Ru 3 (CO) 12 ) is governed primarily by thermal CO dissociation (i.e., decarbonylation) at elevated temperatures, [36,39,40,52,53] one would expect that beyond the self-limiting regime (if such a regime exists), the GPC would increase sharply with increasing temperature, [36,47,48] the opposite of the reported trend in some of these works. In fact, on acidic oxides such as SiO 2 , Ru 3 (CO) 12 has been shown to decarbonylate completely at 200 °C.…”