Abstract
The need to apply coatings to components with complex surface geometries, such as the blades and guide vanes in gas turbine engines, has stimulated significant interest in the development of vapor deposition techniques capable of depositing uniform coatings on such surfaces. Conventional high vacuum physical vapor deposition techniques use low deposition chamber pressures to avoid gas phase collisions, and enable vapor to reach the substrate. As a consequence, they are only able to deposit coatings on substrate surfaces that are in the line-of-sight of the vapor source. However, many of the components used in gas turbine engines contain regions that cannot be seen by such a vapor source. Recently, gas jet assisted vapor deposition techniques that utilize vapor-phase scattering collisions to transport the evaporant to a components have shown promise for enabling deposition on flux-shadowed (non-line-of-sight) regions. This dissertation has investigated the deposition of a simple nickel coating onto both single and doublet airfoil substrates that have non-line-of-sight surface regions. It has utilized multiscale Monte Carlo simulations of the coating process that are validated against experimental depositions. The study explores the viability of gas jet assisted deposition methods, and investigates the mechanisms that control atomic deposition on surfaces hidden from view of the vapor source including those found of the interior channels between closely spaced pairs of airfoils.
Coating deposition was simulated using direct simulation Monte Carlo simulations to model the vapor transport within the deposition chamber. This simulation method enabled the local flux incident on the substrate surface to be determined. It also enabled the incident angle of the vapor with the substrate to be ascertained. These two quantities were then used to simulate coating growth as a function of location on the substrate surface with a kinetic Monte Carlo simulation method. Combining the two techniques has allowed the simulation of coating thickness uniformity and microstructural development along the entirety of the substrate surface, with only the deposition configuration parameters required as input. The simulations were then used to study deposition at varying deposition chamber pressures and gas flow velocities with both stationary and rotated substrate configurations.
The incident flux over the surface of a stationary single airfoil substrate was found to depend primarily on the mean free path between vapor and carrier gas atom collisions in the vapor transporting gas jet, and the average momentum of the vapor atoms within the jet. During deposition on a single, stationary airfoil substrate, the most uniform incident flux profiles were found with a mean free path of approximately 1/10th the maximum dimension of the airfoil (3-5 mm) and low vapor atom momentum, conditions that occur at low chamber pressures and gas jet velocities. Stationary deposition resulted in substantially lower thickness coatings on non-line-of-sight regions. The coatings also had a columnar microstructure but with significant variations in columnar growth angle, thickness, and porosity around the surface of the substrate.
Rotation of the single airfoil substrates greatly improved coating thickness uniformity. During substrate rotation, the majority of flux onto a surface region was deposited while it was in the vapor source's line-of-sight. The best coating thickness uniformity was then found to occur for a process operating at a low-pressure (0.015 Pa) where gas phase scattering did not occur. However, significant variations in the coatings columnar growth angle and pore fraction were observed in the low-pressure simulated coatings, and the fraction of the evaporated atoms that condense on the substrate was low. Deposition at higher chamber pressures resulted in more uniform microstructures, and significantly improved total deposition efficiency (which was maximized at a chamber pressure near 10 Pa). However, the introduction of a inert gas jet created a non-uniform thickness profile along the convex and concave surface and resulted in the deposition of a coating whose thickness on the convex surface was 1.2 to 3 times thicker than on the concave side. An optimization method was developed to investigate non-uniform substrate rotation and source material evaporation rate variations to overcome the non-uniformity in coating thickness. Coatings simulated using an optimized rotation and evaporation rate patterns exhibited a thickness variation of 10 % or less between the convex and concave surfaces.
Deposition onto a rotated double airfoil geometry substrate introduced additional complications with deposition behavior since some regions on the interior surface were never in the line-of-sight of the vapor source. Deposition onto the outer doublet surfaces closely matched that on single airfoil substrates. However deposition to the interior surfaces relied on the scattering of vapor atoms from carrier gas streamlines that flowed through the inter-airfoil channel. Optimum deposition conditions were found where the vapor was fully depleted from the flow (and deposited onto the substrate's interior surfaces) just as the flow exited channel's rear opening. The coating thickness on the interior surfaces was found to depend on the chamber pressure and carrier gas velocity. Increasing the chamber pressure decreased the vapor's diffusion distance transverse to the carrier gas stream flow direction and allowed vapor to propagate further through the channel before deposition. At the highest chamber pressures investigated (100 Pa), most of the vapor traveled through the interior channel without depositing. The maximum deposition efficiency on the interior surfaces was found to occur at chamber pressures near 30 Pa and with high gas flow velocity. The use of higher velocity flows also improved the deposition uniformity, but had a smaller influence than variation of the chamber pressure. These observations were effected by the substrate geometry. Along the inner substrate surfaces, coatings with reasonable thickness uniformity were only possible when the mean free path was significantly smaller than the width of the channel opening. When the mean free path was of similar magnitude or greater than the channel width, deposition occurred only from line-of-sight trajectories. This prevented significant coating deposition near the inner surface midpoints.
