Process Conditions and Microstructures of Ceramic Coatings by Gas Phase Deposition Based on Plasma Spraying
- First Online:
- Cite this article as:
- Mauer, G., Hospach, A., Zotov, N. et al. J Therm Spray Tech (2013) 22: 83. doi:10.1007/s11666-012-9838-y
- 508 Views
Plasma spraying at very low pressure (50-200 Pa) is significantly different from atmospheric plasma conditions (APS). By applying powder feedstock, it is possible to fragment the particles into very small clusters or even to evaporate the material. As a consequence, the deposition mechanisms and the resulting coating microstructures could be quite different compared to conventional APS liquid splat deposition. Thin and dense ceramic coatings as well as columnar-structured strain-tolerant coatings with low thermal conductivity can be achieved offering new possibilities for application in energy systems. To exploit the potential of such a gas phase deposition from plasma spray-based processes, the deposition mechanisms and their dependency on process conditions must be better understood. Thus, plasma conditions were investigated by optical emission spectroscopy. Coating experiments were performed, partially at extreme conditions. Based on the observed microstructures, a phenomenological model is developed to identify basic growth mechanisms.
Keywordsactivation energygas phase depositionmicrostructureplasma spray-PVDstructure zone model
The very low pressure plasma spray (VLPPS) process has been developed with the aim of depositing uniform and thin coatings with large area coverage based on plasma spraying. At typical pressures of 50-200 Pa, the characteristics of the plasma jet change compared to conventional low pressure plasma spraying processes (LPPS, formerly often called vacuum plasma spraying, VPS) operating at 5-20 kPa. By VLPPS, quite thin and dense ceramic coatings can be obtained for special applications like solid oxide fuel cells (Ref 1), gas separation membranes (Ref 2), and wear protection (Ref 3).
The combination of VLPPS with enhanced electrical input power has led to the development of the plasma spray-PVD process [PS-PVD (Ref 4), initially called LPPS-TF process, TF = thin film]. At electrical currents up to 3000 A and plasma gas flow up to 200 slpm, an input power level of 180 kW could be achieved. The plasma plume expands to a length of more than 1.5 m and 200-400 mm in diameter. At appropriate parameters, it is even possible to evaporate the powder feedstock material achieving advanced microstructures and non-line of sight deposition, e.g., for thermal barrier coatings (Ref 5).
To exploit the potential of such gas-phase deposition using plasma spray-based processes, the deposition mechanisms and their dependency on process conditions must be better understood. The PS-PVD process can be pictured as occurring in three steps each with special characteristics due to the low pressure and the high power: feedstock processing in the plasma torch (1), plasma jet formation and material transport (2), deposition and coating growth (3). These sub-processes were used to structure the work described in this article.
Experiments were carried out on a Sulzer Metco Multicoat System (Sulzer Metco, Wohlen, Switzerland). It resulted from a comprehensive reconstruction of an existing conventional LPPS system. In particular, it was equipped with an additional vacuum pumping unit, a large vacuum blower to provide sufficient pumping capacity at low pressures, enlarged cooling capacity, additional power sources, a new torch transfer system and new control units. In addition to a modified single cathode O3CP gun, which was used in this work, also the F4-VB torch as well as the three-cathode TriplexPro torch can be operated.
Plasma spray parameters
Ar 35 slpm He 60 slpm
Powder feed rate
1-20 g/min (varied)
2 × 12 slpm Ar
300-1400 mm (varied)
The spectrometer used for plasma characterization was an HR2000 (Ocean Optics, Dunedin, FL, USA) scanning a wavelength range of 360-795 nm. The plasma radiation was collected through a borosilicate glass window and an achromatic lens, transferred by an optical fiber to the 25 μm entrance slit. The groove density of the grating was 600 mm−1. Wavelength calibration was carried out with a spectral Hg lamp.
Results and Discussion
Feedstock Processing in the Plasma Torch
The O3CP torch was equipped with a two-fold internal powder injection. The feedstock agglomerates are fragmented into sub-micron primary particles and heated already inside the nozzle where the plasma gas density is still high before exiting. This results in the formation of molten droplets and vapor with atomic species. The reason for the high density here is that the plasma jet is significantly supersonic. This means that the plasma gas can exit the nozzle at a pressure which is different from the chamber pressure. As the flow is faster than the pressure waves traveling in the fluid at the local speed of sound, no information on the chamber pressure is carried inside the nozzle (Ref 6).
In the experiments, the development of clusters was also observed showing neither atomic nor liquid-like behavior at deposition. Such clusters may be the product of an incomplete evaporation or nucleation and growth. Similar clusters were already produced to obtain nanostructured coatings by injecting vapor phase reactants into a plasma and rapidly quenching in a supersonic nozzle (Ref 7).
YSZ particles which were collected from the PS-PVD chamber walls showed particle sizes between 5 and 20 nm. This is significantly smaller than the YSZ feedstock primary particles which were between 70 and 130 nm. Furthermore, the XRD patterns differ significantly. They indicate monoclinic zirconia and cubic yttria for the feedstock in contrast to half tetragonal and half monoclinic YSZ for the collected particles. Thus, the collected particles can be regarded to be clusters as discussed. They play an important role in the formation of either columnar or dense coating microstructures; see below. Mixed mode deposition of clusters and liquid splats was also observed depending on the process conditions.
In this study, a helium-argon mixture was used as plasma gas. Helium has a distinctly higher ionization temperature and the associated drop of viscosity takes place at much higher temperatures compared to argon (Ref 8). This leads to an increased momentum transfer to the particles. As helium and argon are both monatomic gases, they have similar molar enthalpy characteristics.
Plasma Jet Formation and Material Transport
As the chamber pressure is significantly below the nozzle exit pressure, the jet is under-expanded. The temperature and velocity distributions in the plasma jet are at higher level and more uniform compared to conventional spray processes (Ref 6). One reason is the laminar flow of the plasma: Reynolds numbers are typically very small. Therefore, the interaction of the plasma jet with the surrounding atmosphere is weak so that it is less cooled or decelerated. The homogenous energy distribution in the plasma plume allows also coating of thin substrates which are sensitive to thermal deformation.
Based on spectroscopically measured emissions, plasma temperatures can be determined by the atomic Boltzmann plot method, which is described elsewhere (Ref 9). 32 readily identifiable Ar I atomic spectral lines were selected between 516.2285 and 763.5106 nm to draw the plots. They show linear developments and the coefficients of determination for the regression lines are always r2 > 0.88. At PS-PVD, the collision frequency is reduced and accordingly the mean free path in the plasma jet is increased. Local thermal equilibrium (LTE) may cease to be valid. Therefore, the obtained temperatures represent the excitation temperatures which are regarded to be equivalent to the electron temperatures.
Deposition and Coating Growth
Primarily, the coating microstructure is determined by the nature of the particles arriving on the substrate. The incorporation of solid particles has already been discussed above. Liquid droplets form splats piling up to build the coating. This is the general deposition mechanism for conventional plasma spraying. At PS-PVD conditions, the deposits are formed predominantly from clusters and/or vaporized atomic species. (1) Shadowing, (2) adsorption, nucleation, and growth, (3) as well as bulk recrystallization are the basic mechanisms characterizing the coating growth.
Adsorption, Nucleation, and Growth
At sufficient high temperatures, volume diffusion starts. Equiaxed grains are formed by bulk recrystallization. The coating surfaces show smooth, polyhedral structures. For high-melting coating materials, the required temperatures exceed the thermal load capacity of common substrate materials. Hence, graphite and tungsten had to be used to investigate some samples in this high temperature region.
Structure Zone Model (SZM)
For many materials, the activation energies of surface and bulk diffusion can be related to the melting point Tm. Thus, these mechanisms can be expected to dominate over different ranges of the homologous temperature of the substrate surface Ts/Tm. This is the basis of SZMs (Ref 13). The Movchan-Demchishin SZM (Ref 14) was developed based on thick coatings (0.3-2 mm) of Ti, Ni, W, Al2O3, and ZrO2 made by electron beam evaporation. Three zones were defined, each with its own structure and properties which correspond well with the characteristics of PS-PVD coatings formed by shadowing, surface, and bulk diffusion as described above. The transition temperatures for oxides were given by Ts/Tm = 0.22-0.26 between zones 1 and 2 as well as Ts/Tm = 0.45-0.5 between zones 2 and 3.
The consistence with the Thornton SZM is evident. In particular, the oblique and narrowing shape of zone T is similar. The transition line between zones 2 and 3 should be independent on the impingement rate and thus proceed vertically as bulk diffusion is not affected by reduced surface mobility. This is more obvious in the Thornton SZM.
Activation Energy for Surface Diffusion
This means that by logarithmical plotting the impingement rate versus the reciprocal temperature and calculating an exponential regression function for the data points of zone T which represent microstructures with first indications of surface diffusion, the activation energy can be determined from its slope.
Summary and Conclusion
By PS-PVD, it is possible to evaporate the powder feedstock at appropriate parameters providing advanced microstructures and non-line of sight deposition. To ensure evaporation of the feedstock, the conditions inside the nozzle in proximity to the location of injection are crucial. High-power density and accommodated feedstock characteristics are required. Besides evaporation, the formation of nano-sized clusters is observed. Having exited the nozzle, further particle heating is reduced compared to atmospheric pressure due to the low density (Nusselt numbers are smaller).
For coating growth, a SZM was developed following the work of Thornton. The analysis reveals different growth mechanisms like shadowing, surface, and bulk diffusion. Vapor deposition of compact columnar structures is only possible if the surface mobility of the adatoms is sufficient. This is obtained by high substrate temperatures and low deposition rates. The activation energy for surface diffusion was initially estimated by 141 kJ/mol for YSZ which is relatively low.