Due to the superior thermophysical properties of silicon carbide at high temperature, silicon carbide (SiC) coatings have the potential to offer excellent resistance to both oxidation and hydriding of zircaloy-based nuclear fuel cladding used in light water reactors. Unfortunately, the current deposition technique for SiC using chemical vapor deposition (CVD) can be only achieved at relatively high substrate temperatures, which can severely degrade the performance of zircaloy cladding. As a comparison, atomic layer deposition (ALD) enables the growth of pinhole free films on large and convoluted substrates with nanometer precision that operates at low temperatures. Developing a new ALD process requires better understanding of how the film growth process takes place, which depends on the chemicals being used. The major obstacle to low-temperature ALD of SiC is to determine the right precursors. To this end, we use high-throughput ab initio calculations and state-of-the-art machine learning (ML) techniques to accelerate the precursor screening. The method involves two different types of prediction: the forward and backward predictions. The objective of the forward prediction is to train a set of machine learning models for the free energies of silicon and carbon-containing molecules from the high-throughput ab initio database. Inverting the trained forward models through Bayes’ law, we quantitatively predict the free energies of all the possible silicon and carbon-containing molecules from the PubChem compound database. By calculating the Gibbs energy of reaction using the ML predicted energy, several potential silicon and carbon precursors are predicted to be promising for ALD of SiC at low temperature. We believe the present method will be helpful to develop novel ALD precursors for other applications.

The precursor 2, 3 - dimethyl butadiene Ruthenium tri-carbonyl (Ru(DMBD)(CO)₃), is a volatile molecule with favorable properties for the deposition of both Ru and RuO₂ films via ALD.¹ An intriguing aspect of this precursor is its asymmetric molecular structure with the DMBD ligand coordinated to one side of the Ru⁴⁺ center, while the 3 carbonyl groups are bonded to the opposite side. This makes the molecule an attractive candidate to study using in situ downstream quadrupole mass spectrometry (QMS) since, the release of DMBD vs. carbonyl groups during ALD half-reactions can provide a particularly descriptive view at the atomic level of how Ru(DMBD)(CO)₃ interacts with the substrate.

In this talk, we will discuss the mechanism of ALD of Ru thin films using Ru(DMBD)(CO)₃ and H₂O. The QMS signal of the Ru(DMBD)(CO)₃ consists of three distinct species. First, the mass-to-charge ratio (m/e) of 67 corresponds to the DMBD ligand dissociating from the molecule². Second, a strong m/e = 16 is also observed as the DMBD further cracks into smaller fragments. Finally, m/e = 44 is observed and is related to CO₂ and HCOOH formation from the reaction of the tri-carbonyl groups with H₂O.³ The QMS signal of these three species will be discussed in the context of varying process parameters such as Ru(DMBD)(CO)₃ and H₂O pulse times and the temperature of the ALD process. Additionally, ex situ film characterization including atomic force microscopy, transmission electron microscopy, X-ray diffraction, X-ray photoelectron spectrum and resistivity measurements will be presented to correlate the process signatures obtained via QMS to the film type, morphology and electrical properties.

References:

1. Austin, D. Z.; Jenkins, M. A.; Allman, D.; Hose, S.; Price, D.; Dezelah, C. L.; Conley, J. F., Atomic Layer Deposition of Ruthenium and Ruthenium Oxide Using a Zero Oxidation State Precursor. Chem. Mater. 2017.

2. Chiang, C.-M.; Rowe, J.; Malic, R.; Sen, A.; Steigerwald, M.; Mills, A., A new CVD reaction for atomic layer deposition of silicon. Applied surface science 1996,107, 189-196.

3. Rosenberg, S. G.; Barclay, M.; Fairbrother, D. H., Electron beam induced reactions of adsorbed cobalt tricarbonyl nitrosyl (Co (CO) 3NO) molecules. The Journal of Physical Chemistry C 2013,117 (31), 16053-16064.

Aluminum oxide (Al₂O₃) is a suitable replacement for SiO₂ in electronic devices such as flash memory due to its wide band gap and higher dielectric constant. Atomic layer deposition (ALD) using sequential pulses of trimethylaluminum (TMA) and an oxidant is a leading method for the formation of nano-scale Al₂O₃ layers. Al₂O₃ layers grown by ALD have been demonstrated with leakage currents of less than 1 nA/cm³. The quality of the layers depends on both the deposition temperature and choice of oxidant, which is commonly water or oxygen atoms generated using a plasma.

An oxidant more reactive than water, such as H₂O₂, could produce a denser Al₂O₃ film. The growth of Al₂O₃ on hydrogen-terminated silicon using TMA and H₂O₂ was compared to that using TMA and water as a reference. Spectroscopic ellipsometry was used to determine the growth per cycle (GPC), and in situ x-ray photoelectron spectroscopy (XPS) was recorded before and after each precursor dose. The O 1s XPS peak at 531.8 eV and the Al 2p peak at 75.5 eV were used to monitor the formation of the Al-O bond and the C 1s peak at 284.5 eV was used to monitor the persistence of the Al-C bond after each half cycle. A second state in O 1s XPS at 533.3 eV assigned to the O-H bond was used to estimate the number of reactive sites present after each oxidant pulse.

The growth-per-cycle (GPC) was equal to 1.2 Å at H₂O₂ pulse times above 0.3 s, which is similar to reported ALD growth rates using water. The Al 2p XPS peak appeared after 4 ALD cycles using anhydrous H₂O₂ and after 7 cycles using water. The aluminum coverage after 10 ALD cycles was 40 % greater for anhydrous H₂O₂ compared to water, although the GPC was similar for each co-reactant. The O-H coverage doubled with each pulse of H₂O₂. The O 1s peak assigned to O-H also broadened with subsequent ALD cycles due to the presence of additional surface states. Each pulse of water increased the O-H coverage by about 1.5 times compared to the subsequent pulse. The ratio between C 1s and Si 2p peak areas showed twice as much carbon present on the surface during H₂O₂ ALD. The C 1s/Si 2p ratio increased during pulses of TMA and decreased during pulses of H₂O₂ due to surface reactions that desorb CH₄. Greater OH densities and C coverages indicate that H₂O₂ activates more surface sites for the metal precursor than water, improving Al₂O₃ film density.

Cyclic azasilanes are a promising class of reactants for atomic and molecular layer deposition due to their volatility, reactivity, and diverse set of peripheral chemical functionalities. The reactivity of this class of molecules stems from ring strain and unstable Si-N bonding. Reaction with water or OH groups results in ring opening and Si-O bond formation and is a useful chemical step for controlled film growth. This work focuses on using and understanding this class of molecules for use in atomic and molecular layer deposition ( ALD, MLD). The efficacy of the ring opening reaction and the diversity of functional groups on the cyclic azasilanes deem this precursor group highly promising for MLD. We report the growth of various MLD films based on reaction with anhydrides, metal-organics, and water. The film growth was found to be optimal near 100 °C. The two chemistries reported here are the sequential reaction of (1) a cyclic azasilane, maleic anhydride, and water and (2) a cyclic azasilane, maleic anhydride, trimethylaluminum (TMA), and water. Reaction 1 showed clear evidence of diffusion and reaction within the film as evidenced by large growth rates (~10 nm per cycle) after ~15 growth cycles, regardless of purge time. However, use of TMA in reaction 2 arrested this diffusion, presumably due to consumption of reactive groups or through densification of the film. Film properties were measured as a function of annealing temperature using ellipsometry, x-ray reflectivity, and infrared absorption measurements. These measurements indicate a densification of the films while maintaining low surface roughness. Porosity is evident after removal of organic components by a low dielectric constant for the films derived from reaction 1.These MLD chemistries offer a route to tunable hybrid thin films on high aspect ratio substrates and have potential applications in diffusion barriers, low dielectric constant layers, and passivation layers.