Growth mechanism of aluminium oxide films and factors controlling it in a pulse pressure MOCVD deposition technique using different precursors in organic solvents.
Thesis DisciplineMechanical Engineering
Degree GrantorUniversity of Canterbury
Degree NameDoctor of Philosophy
The aim of this project is to investigate into the growth mechanism occurring in a pulsed pressure MOCVD deposition technique using aluminium oxide films on silicon substrates. Another objective of the study is to look into the effect of the precursorsolvent interaction on the droplet vaporisation mechanism. Commonly used aluminium oxide precursors (aluminium isopropoxide, aluminium sec butoxide, aluminium tert butoxide and aluminium acetylacetonate) and solvents (hexane and toluene) are used to deposit aluminum oxide films and their properties (film composition, surface morphology, surface roughness, growth rate and surface bonds) are studied. The deposition variables, besides the choice of precursors and solvents, includes precursor concentration, deposition temperature, presence of a shield over the substrate, and external heat to the droplets through a heat tape. The shield is used to test the effectiveness of the precursor-solvent droplet to vaporise and also ensures that aerosols or surviving droplets do not contribute to the film deposition. A model simulating the droplet behaviour and the reactor conditions, developed by Boichot et al. is used to study the aluminium oxide deposition to corroborate the experimental observations. According to the model, external heat assists in the droplet vaporisation process, though the minimum temperature at which any significant effect is observed is 220oC. At such high temperatures, it is likely that the precursors would be decomposed in the gas phase and arrive at the substrate as homogeneous particles. The enthalpy of vaporisation and the specific heat capacity of the solvents is crucial during the droplet evaporation. Solvents with a higher vapour pressure (lower enthalpy of vaporisation) release more material into the vapour phase during the flash vaporisation stage. Following the flash vaporisation, the droplets experience a process of solvent evaporation and precursor evaporation as it approaches the deposition zone. The droplets would arrive at the substrate either as aerosols or as a liquid droplet, with the latter being less likely to occur considering the high deposition temperature (>400oC). It is also possible that the decomposition of the precursors has already occurred in the gas phase, resulting in the precursors to arrive at the surface as dried up particles. A crucial pre-requisite condition is the solubility of the precursor in the solvent, either instantly or by continuous stirring. All the deposited films are found to be oxygen-rich with a considerable amount of carbon due to the absence of any oxidants, such as O2 or water vapour to react with the carbon and eliminate them. XRD for the films deposited using aluminium sec-butoxide reveals micro/nano-crystalline peaks, which coupled with the FTIR peaks suggest the presence of κ-Al2O3. The growth rate is affected by the precursor concentration, the choice of the precursor-solvent and the presence of the shield. The growth rate increases with an increase in concentration as more precursors are available for deposition. A higher growth rate is observed for the films deposited without the shield as the substrate is exposed to more precursor flux. The deposited films are smoother and uniformly distributed with nano-sized particles in the presence of the shield as the aerosols gets filtered out. Aluminium sec-butoxide is found to the best precursor among the four investigated in this study, primarily because it is a liquid at room temperature, thereby being more compatible with organic solvent and allowing higher concentration deposition. Films deposited using aluminium sec-butoxide deposition had O: Al values closest to the stoichiometric ones. Overall, the precursors can thus be ranked based on their performance (growth rate, stoichiometry, surface roughness) as sec-butoxide > tertbutoxide > acetylacetonate > isopropoxide. This ranking, however, also depends on the solvent used, with tert butoxide performing better than sec butoxide when dissolved in toluene. Hexane performs better than toluene in terms of deposition from the vapour phase, determined by the growth rate of the films under the shield. It, however, fails in chemical compatibility with several of the precursors particularly at high concentrations. The key reason for hexane showing a better performance is its high vapour pressure, which causes more material to be released into the vapour phase during the flash vaporisation stage. A condensation model has been proposed for the deposition of thin films in a pulse pressure MOCVD. It is proposed that there are 5 different mechanisms that are occurring inside the deposition chamber which affects the film morphology. Vapour phase deposition is observed under conditions of low arrival rate (low concentration or high concentration-presence of shield) resulting in a uniform film with spherical particles which is typical for aluminum oxide. For high concentrations and in the absence of the shield, the substrate experience a high localised flux of the precursors resulting in a Liedenfrost aerosol deposition mechanism forming clusters/agglomerates on the surface. The above mentioned growth mechanisms are desirable if the process is controlled leading to the formation of a solid film on the substrate. Other process that also occur inside the chamber, but are undesirable, are homogeneous particle formation and direct liquid impingement. The former occurs due to the heat from the surroundings which causes the solvent to evaporate and decomposes the precursor before it arrives at the substrate resulting in a powder formation. The direct liquid impingement is caused by a large solvent droplet arriving at the substrate, evaporating at the spot and leaving the decomposed particles in a circular pattern. The film morphology can be controlled by controlling the process parameters such as concentration, solvent choice and presence of a shield. Future work would include improving the crystallinity of the film, optimising the process to yield stoichiometric aluminium oxide film and a detailed study on the possibility of conformal coating on nanoscale patterns.