Real-Time Observation of Atomic Layer Deposition Inhibition: Metal Oxide Growth on Self-Assembled Alkanethiols

Abstract

Through in situ quartz crystal microbalance (QCM) monitoring, we resolve the growth of a self-assembled monolayer (SAM) and subsequent metal oxide deposition with high resolution. We introduce the fitting of mass deposited during each atomic layer deposition (ALD) cycle to an analytical island-growth model that enables quantification of growth inhibition, nucleation density, and the uninhibited ALD growth rate. A long-chain alkanethiol was self-assembled as a monolayer on gold-coated quartz crystals in order to investigate its effectiveness as a barrier to ALD. Compared to solution-loading, vapor-loading is observed to produce a SAM with equal or greater inhibition ability in minutes vs days. The metal oxide growth temperature and the choice of precursor also significantly affect the nucleation density, which ranges from 0.001 to 1 sites/nm(2). Finally, we observe a minimum 100 cycle inhibition of an oxide ALD process, ZnO, under moderately optimized conditions.

Sequential Regeneration of Self‐Assembled Monolayers for Highly Selective Atomic Layer Deposition

“…Despite the successful demonstrations of AS‐ALD using SAMs, there are limitations associated with conventional methods of preparing the SAMs. First, depending on the type of SAM and the desired substrate, formation of the monolayer may require dipping the substrate into a solution containing the SAM‐forming molecules for several hours to a few days . The need for long SAM exposure in solution adds significant time to the overall AS‐ALD process.…”

“…It has been shown that the thermal stability of some well known SAMs such as thiols on metal surfaces is limited to a temperature of only ≈100 °C, which is below the growth temperature for many ALD processes. As a result, we hypothesize that after a number of ALD cycles, desorption or degradation of organic molecules from the SAM layer generates defects such as pinholes that serve as nucleation sites for ALD . Different approaches have been proposed to circumvent this problem, such as performing ALD at lower temperatures or using longer initial deposition times for the SAMs.…”

“…Since the onset of ALD that signals loss of selectivity is thought to originate from degradation or loss of the SAM from the substrate, we expect to see a significant enhancement in selectivity if the SAM can be regenerated or redeposited from the vapor without altering the properties of the deposited film on the silicon (Si) surface. Moreover, due to the different kinetics of molecular adsorption onto a surface from the vapor phase, and the fact that vapor dosing of dodecanethiol (DDT) provides an undiluted exposure of the surface to SAM molecules compared to ethanolic solution of DDT, the deposition time may be significantly reduced if carried out from the vapor phase . Vapor deposition also could provide better SAM coverage on porous or 3D structures, and could allow the SAM passivation step to be integrated with the rest of the ALD process.…”

Atomic layer deposition—Sequential self-limiting surface reactions for advanced catalyst “bottom-up” synthesis

“…Compared with Al 2 O 3 ALD, there have been far fewer studies of TiO 2 ALD on metal surfaces. In one study, Martinson et al reported "island growth" of TiO 2 ALD on Au using tetrakis(dimethylamido)titanium and water at 150 °C, where a lower growth rate was observed by in situ QCM measurements during the first 30 cycles[339]. They commented that surface oxygen contamination of the nominally bare Au might have nucleated the TiO 2 ALD.…”

Fabrication of Thin Films of α-Fe2O3 via Atomic Layer Deposition Using Iron Bisamidinate and Water under Mild Growth Conditions

“…This growth rate is more than 0.2 Å/cycle higher than that reported by Lim and co-workers 24 likely due to the lower reactor temperatures used here, which has been observed for other ALD processes. 21,27,28 Figure 2a also shows the importance of using a long purge time (30 s) in order to sufficiently clear the reactor of precursor and to avoid undesirable CVD reactions, with 15 s showing a higher average growth rate and more variations between runs. In contrast, variations in pulse time and reactor temperature have negligible effect on the growth rate with the exception of a slight profile observed for 0.1 s pulse and 130°C reactor temperature as shown in Figure 2b, c. The wall-mounted QCM used in this study was equipped with 2 ports, one near the reactor inlet and outlet to observe if any growth profile was observed during the ALD process.…”