Abstract
Axial suspension plasma spraying (ASPS) is a relatively new, innovative technique with which microstructures have been produced that are similar to the ones produced by electron beam physical vapor deposition. They have a columnar structure and consist of nm- and µm-sized pores. However, so far the formation of the microstructure is not fully understood because fragmentation and vaporisation of the liquid significantly affects the deposition process. Analysis of single splats can provide important information on the phenomena controlling the coating formation process and the final coating properties. Therefore, the present study aims at providing first results of 8 wt-% yttria-stabilised zirconia single splats sprayed onto a steel substrate by use of ASPS. Scanning electron microscopy and atomic force microscopy have been used to characterise the splats with respect to appearance, shape, and size distribution.
Keywords
Introduction
Thermal barrier coatings (TBCs) are used in the hot section of gas turbines to protect the underlying components from high temperature loads [1]. They extend the lifetime of the turbine but also allow for a higher combustion temperature which enhances the engine efficiency and results in lower emissions [1]. Commonly, TBCs are produced by atmospheric plasma spray technique (APS) or by electron beam physical vapor deposition (EB-PVD). Owing to their columnar microstructure, EB-PVD coatings can accommodate thermal stresses better than APS coatings [2]. Therefore, EB-PVD coatings have longer life time but the production costs are much higher [2]. Hence, it is of interest to improve the TBC microstructure achieved by thermal spray technique to create coatings that are both strain-tolerant and have low thermal conductivity. Using suspension plasma spraying, TBC microstructures have been produced which are similar to the ones produced by EB-PVD, i.e. they have also a columnar microstructure. The injection of the suspension into the plasma plume can be either radial or axial. Axial injection shows indications that it can further enhance the performance of TBCs.
A microstructure consisting of fine pores is preferable when used as heat protective coating as it can reduce the thermal conductivity [3,4]. While in APS the minimum diameter for an injected particle is in the micrometer range, it is possible to inject nano-sized particles and achieve coatings with nm- and µm-sized pores when using a liquid feedstock [5,6]. VanEvery et al. [7] hold the smaller particle sizes and the sensitivity of the particles to the velocity of the plasma responsible for the occurrence of the columnar structure. They proposed that the small particles follow a trajectory parallel to the substrate before they impact and adhere on asperities and thereby cause the growth of multiple deposits separated by valleys [7]. Chen et al. [8] and Oberste Berghaus et al. [9] also stated that a high proportion of small particles enabled the deposition of a columnar structure during suspension plasma spraying (SPS).
Yttria-stabilised zirconia (YSZ) is commonly used as material for TBCs because of its low thermal conductivity (∼2.3 W m−1 K−1 at 1000°C) and its high coefficient of thermal expansion [1]. The YSZ powder can be dispersed in water or organic solvent (e.g. ethanol) for generating a suspension that can be used for suspension plasma spraying. Water vaporisation requires about 2.6 times more energy than ethanol, but organic solvents could form undesirable carbon particles in the coatings and are accompanied with risks during the process [10]. When the suspension is injected into the plasma plume, it breaks up and forms individual droplets, the solvent evaporates and the powder particles melt [11]. Upon impact on the substrate, the molten particles are expected to flatten out and solidify [12]. The trajectory of the YSZ droplets within the plasma jet can be divided into different paths, one path in the centre of the plasma jet and one in the periphery [13]. Droplets travelling in the periphery are probably not fully molten due to a low heat transfer between the plasma and the feedstock. On the substrate, they deposit with low impact and are recognised by their spherical shape. The splats investigated in this study originate from the droplets with trajectory in the centre of the plasma jet. They travel at high velocity, causing them to flatten on impact with the substrate. YSZ single splats sprayed by axial suspension plasma spraying (ASPS) on a steel substrate were characterised with respect to appearance, shape, and size distribution by scanning electron microscopy (SEM) and atomic force microscopy (AFM).
Methods
Process
ASPS was performed using an Axial III high power plasma torch by Northwest Mettech Corporation. A suspension consisting of 8 wt-% YSZ dispersed in ethanol with a solid loading of 25 wt-% was used. The median particle size was 500 nm giving a powder particle volume of 0.065 µm3. The suspension was sprayed onto Hastelloy X disks of 25 mm in diameter and 6 mm in thickness. The disks were placed in a horizontally rotating (300 rev min−1) fixture at a 75 mm stand-off distance, i.e. the distance between the substrate and the spray torch. The spray torch was moved vertically with a velocity of 700–1000 mm s−1 in one scan to be able to collect isolated single splats and avoid the formation of a coating.
Characterisation
The appearance of the single splats was analysed using a Leo 1550 Gemini SEM equipped with a field emission gun. Prior to investigation, the samples were gold coated. The size distribution of the YSZ splats was determined by image analysis of SEM micrographs using Axio Vision software. Cross-sections were prepared by hot mounting and polishing was performed according to standard procedure for TBCs. The height profile of the single splats was measured by AFM (Bruker Dimension ICON SPM) along two intersecting lines. The AFM data was processed by Gwyddion, a scanning probe microscopy data visualisation and analysis software.
Results
The SEM micrographs in Figures 1 and 2 show the appearance of the single splats. As can be seen, the single splats are circular or slightly elliptical and some splat surfaces have holes and/or cracks. In addition, also debris and parts of broken single splats are present.
SEM micrograph of single splats and debris produced by ASPS. (a, b) SEM micrographs of individual single splats showing cracks in the surface.
The surface area of the splats was measured manually by encircling the splats on SEM micrographs. In Figure 3, a histogram of the size distribution is given. The surface area for the 700 circular and elliptical splats varies between 0.03 and 8 µm2, with a median value of 0.65 µm2. The area distribution is positively skewed and there is an indication for a bimodal distribution with the first peak around 0.3 µm2 and the second around 1.3 µm2. The broken splats and the debris are not taken into consideration in the size distribution even though they contribute with 2/3 to the total amount of sprayed material on the substrate surface.
Histogram of the size distribution of single splats obtained by image analysis.
In addition, 100 single splats were characterised in cross-section and their diameter and height were measured. The volume could then be calculated (if assuming a circular surface area) and compared to the volume of a single particle in the suspension (0.065 µm3). Thirty splats (30%) have a volume corresponding to the size of a single particle in suspension, ±20% deviation. Figure 4(a) shows a splat with a diameter of 1.8 µm and a height of 0.25 µm. The two connected splats in Figure 4(b) have diameters of 0.36 and 0.71 µm and a height of 0.05 and 0.12 µm, respectively. Height profiles of two splats measured by AFM are given in Figure 5(a) and b. They have heights of 0.05 and 0.075 µm and diameters of 1.75 and 2 µm, respectively.
(a, b) SEM micrographs of individual single splats in cross-section. (a, b) AFM height profiles and 3D images of two different single splats.
Discussion
The investigated single splats have a circular or slightly elliptic shape and a smooth surface, even though some of the larger splats (diameter > 1 µm) have cracks or holes in the surface. In addition to the splats, there are also pieces of broken splats on the substrate surface. As deduced from their shape and appearance, the splats solidified before they broke up and pieces fell off the substrate. In Figure 1, a single splat can be seen where only a quarter of a splat is present. The other part of the splat may have fallen off the substrate due to poor adherence and crack formation at the splat surface. Chraska and King [14] discussed the reason for crack formation in plasma sprayed zirconia splats and stated that cracks occur after solidification because of temperature differences between splat and substrate. Upon cooling, the substrate will restrict the contraction of the splat. This leads to the development of stresses in the splats and when the tensile strength for YSZ is reached, a vertical crack develops in the splat. This theory holds for splats produced by APS, i.e. for larger splats. When using ASPS, cracks have only been observed in splats with a diameter larger than 1 µm, which then corresponds to the theory of Chraska and King [14]. Hence, stresses in small splats are not sufficient for cracking to occur. In another study it needs to be investigated if and how these broken splats affect the coating formation and the microstructure of the TBC.
When investigating the splats at higher magnification (Figure 2), some of the splats seem to be slightly lifted from the substrate at the outer edges. It looks like the splats have settled on the substrate with a slightly rounded base, not capable to flatten out. This impression is confirmed by the cross-section image of the single splat in Figure 4(a). As can be seen, the outer edges of the splat are not adhered to the substrate. In fact, only 1.3 µm of the 1.8 µm wide splat is attached. However, there are also splats with a base fully adhered to the substrate.
Figure 4(b) shows two splats, one very small and the other one having a diameter of 0.71 µm. Assuming a circular splat, the obtained surface area (0.4 µm2) is corresponding to the first peak in the bimodal splat area distribution in Figure 3. Joulia et al. [15] explained the bimodal size distribution of single splats produced by radial suspension plasma spraying by the occurrence of two different processes after injection of the suspension into the plasma plume: The suspension breakup into liquid drops, the solvent evaporates and, the powder particles melt and small molten droplets are formed. These small droplets correspond to the first peak of the size distribution. Some of the particles agglomerate and form a bigger solid mass when the solvent is evaporating. This solid mass can then either divide into smaller droplets, as in the first case, or they form one large molten droplet. These large droplets correspond to the second peak of the size distribution. Hence, to determine if the observed splats have been formed by one single droplet or an agglomeration of particles from the suspension, the volumes of the 100 splats investigated in cross-section have been calculated. It was found that 30% of the splats have been formed by a single molten particle from the suspension. All other splats are formed by larger droplets, i.e. agglomerates of two or more particles. The volumes of the splats analysed by AFM (Figure 5(a,b)) are calculated to be 0.12 and 0.23 µm3 and correspond to the volume of 2 and 4 agglomerated particles from the suspension, respectively.
The stand-off distance, i.e. the distance between the substrate and the spray torch, is one of the process parameters that will influence the particle impact velocity and thereby the formation of the final coating. Therefore, an upcoming article will investigate the effect of process parameters on shape, size and cracks in the single splats.
Conclusion
The SEM investigations have shown that YSZ single splats produced by ASPS have a circular or slightly elliptical shape. In larger splats (>1 µm in diameter) cracks develop because of thermal stresses. As a result, there is a large fraction of broken splats and debris on the sample surface. The size of the single splats can be related to the spraying process. The calculated volumes indicate that the small splats (∼30%) are formed by a single droplet while the large splats are formed by agglomerated suspension particles.
Footnotes
Acknowledgement
Alexei Kalaboukhov (Department of Microtechnology and Nanoscience, Chalmers University of Technology) is acknowledged for help with the AFM measurements.
Disclosure statement
No potential conflict of interest was reported by the authors.