UC Riverside

Platinum (Pt) has been extensively used in the fuel cells due to their excellent catalytic activity; with high work function, it is also a possible candidate for the use in memory devices as charge storage medium, and thereby has generated interest in MOS structures. The Pt cost/availability is one of the major reasons limits it’s commercialization. Scaling the film thickness or downsizing catalyst nanoparticles to single atoms could significantly increase their efficiency in energy aspect or lessen the impact of lateral coupling of charge storage layers, reduce the stress-induced leakage current in MOSFET applications.

Atomic layer deposition (ALD) is a thin film growth technique utilizing sequential self-limiting surface reactions to deposit films with atomic layer control that are usually extremely conformal to the initial substrate, and thereby is utilized in this study. However, the entire process still lacks detailed atomic-scale understanding including the formed interface and the effect of substrate used on the metal growth, this is in part because the surface science involved in the first few layer is not easy to be identified; thereby the dissertation works aimed on using surface sensitive techniques to investigate the surface reactions taking place at the initial stage of film formation.

The primary goal of the study was to examine and identify the self-limiting mechanisms that drive the surface reactions involved in ALD Pt on the Si-based and transition metal surfaces using (methylcyclopentadienyl)-trimethyl platinum (MeCpPtMe3) and O2 as the reactants. The second phase of this portion of the investigation involved establishing an optimized process window for a steady growth regime by tailoring dosing conditions. The detailed controls were based on acquisitions of chemical information per dose, which is demonstrated through in situ X-ray photoelectron spectroscopy of MeCpPtMe3 chemisorption by mapping the gradual atomic scale evolution in the surface composition, and further.

Pt thin films were grown on SiO2 films with assistance of electron impact activation and have shown an enhancement in uptake of film depositions. An atomic ratio of 6.98:1 between C:Pt and the fragmentation pattern reproduced via mass spectrometry confirm that the entire process results in loss of methyl ligands only while retaining the MeCpPt moiety intact. The depositions of MeCpPtMe3 on Si(100) with native oxide revealed a temperature window from 523 K to 573 K. The surface coverage of Pt films on 300 nm SiO2 shows a linear relationship with the exposure. Molecular oxygen atoms dosed onto the SiO2 surface after MeCpPtMe3 exposures lead to layered growth with steric hindrance during the first few cycles after which the growth per cycle shows trend becoming constant and the film thickness may increases linearly with the number of cycles.

The surface chemistry associated with the thermal ALD of platinum films on metallic nickel substrates using the identical recipe was discussed. The uptake of the MeCpPtMe3 was found to be self-limiting between 525 and 625 K, but to lead to multilayer deposition at 675 K. The calculated C:Pt ratio of the adsorbed species, suggests that all methyl moieties are lost upon activated bonding to the surface while the MeCp group remains coordinated to the adsorbed Pt atoms. Oxygen treatment of that surface leads to the complete removal of the carbon-containing species from the surface and to the formation of a thin NiO film. Further dosing with MeCpPtMe3, in the next ALD cycle, fully reduces that NiO film to metallic Ni(0) and adds more Pt to the surface. However, no Pt film buildup was seen after several ALD cycles.

The study also includes approaches of activating/passivating the starting surface by modifying the surface chemistry through liquid/gas phase pretreatment. The long nucleation delay observed from Pt/SiO2 experiments could be partially reduced by generating hydroxyl groups as new adsorption sites via piranha cleaning process. On the other hand, the selective deposition was demonstrated by applying silylation step in gas phase, and the site-blocking was shown partially effective as the XPS studies revealing the silylated samples required many more ALD cycles for nucleation and growth to achieve similar intensities compared with referenced samples.