Atomic Layer Deposition (ALD) is a thin-film growth technique offering precise control of film thickness and the ability to coat high-aspect-ratio features such as trenches and nanopowders. Unlike other film growth techniques, ALD does not require harsh processing conditions and is not limited by line-of-sight deposition. Emerging applications for ALD materials include semiconductor devices, gas sensors, and water-diffusion barriers.
The chemistry behind ALD involves understanding how the precursors interact with surfaces to deposit the desired material. All ALD precursors need to be stable on the substrate to ensure self-limiting behavior yet reactive enough to be easily removed with the second reagent. Recent precursor development has provided many volatile organometallic compounds for most of the periodic table. As the number of precursors increases, proper precursor choice becomes crucial. This is because the film properties, growth rates, and growth temperature vary widely between the precursors. Many of the above traits can be predicted with knowledge of the precursor reaction mechanisms. This thesis aims to link surface reaction mechanisms to observed growth and nucleation trends in metal and oxide ALD systems.
The first portion of this thesis explores the mechanisms of two ALD oxide systems. First, I examine the mechanism of ALD alumina with ozone. Ozone is used as an oxidant in the semiconductor industry because the deposited Al 2O3 films possess better insulating properties and ozone is easier to purge from a vacuum system. FT-IR analysis reveals a complicated array of surface intermediates such as formate, carbonate, and methoxy groups that form during Al2O3 growth with ozone. Next, a new method to deposit thin films of Ga2O3 is introduced. Gallium oxide is a transparent conducting oxide that needs expensive solid precursors to be deposited by ALD. I show that trimethylgallium is a good high-temperature ALD precursor that deposits films of Ga2O 3 with low impurities and a good growth rate.
The second section of this thesis focuses on two metal ALD systems. One major drawback of metal ALD systems is their inability to nucleate on many oxide surfaces. This greatly limits the applications of metal ALD for interconnects and flexible electrodes. The first emphasis is an on a new palladium ALD system using palladium (II) hexafluoroacetylaceonate (Pd(hfac)2) and formalin. FT-IR studies show that the Pd(hfac)2 dissociatively adsorbs, releasing free hfacH molecules that bind to Lewis acid sites on the alumina. The observed nucleation period of Pd is linked to surface poisoning by hfacH. In a related experiment, I use trimethylaluminum exposures to remove excess hfacH from the surface. Trimethylaluminum is able to ligand exchange an easier to remove methyl group with the surface hfacH, This treatment causes palladium to nucleate much more rapidly and deposit at lower temperatures.
Finally, I examine ToRuS, a new precursor solution for Ru ALD. ToRuS, a solution of RuO4 in perfluoroethers, deposits ruthenium faster and at lower temperatures than all other Ru precursors. The mechanism for deposition and role of the perfluoroethers, however, is poorly understood. In the first study, I couple FT-IR spectroscopy with ab-initio calculations to identify the surface species formed when the perfluoroether solvent adsorbs on alumina. These surface species bind strongly to the alumina surface, creating a nonpolar, fluorinated layer. I then use these results to understand how ToRuS deposits Ru films. The fluorinated layer solvates RuO4, stabilizing it near the surface until it can be reduced by H2 gas. FT-IR and XPS analysis shows that the fluorinated layer does not leave carbonaceous impurities on the ruthenium surface or impedes metal deposition.
