Failure of Multilayer Suspension Plasma Sprayed Thermal Barrier Coatings in the Presence of Na2SO4 and NaCl at 900 °C
- 206 Downloads
The current investigation focuses on understanding the influence of a columnar microstructure and a sealing layer on the corrosion behavior of suspension plasma sprayed thermal barrier coatings (TBCs). Two different TBC systems were studied in this work. First is a double layer made of a composite of gadolinium zirconate + yttria stabilized zirconia (YSZ) deposited on top of YSZ. Second is a triple layer made of dense gadolinium zirconate deposited on top of gadolinium zirconate + YSZ over YSZ. Cyclic corrosion tests were conducted between 25 and 900 °C with an exposure time of 8 h at 900 °C. 75 wt.% Na2SO4 + 25 wt.% NaCl were used as the corrosive salts at a concentration of 6 mg/cm2. Scanning electron microscopy analysis of the samples’ cross sections showed that severe bond coat degradation had taken place for both the TBC systems, and the extent of bond coat degradation was relatively higher in the triple-layer system. It is believed that the sealing layer in the triple-layer system reduced the number of infiltration channels for the molten salts which resulted in overflowing of the salts to the sample edges and caused damage to develop relatively more from the edge.
Keywordscolumnar microstructure composite of gadolinium zirconate and YSZ hot corrosion suspension plasma spray
Suspension plasma spray (SPS) is a recent advancement in thermal barrier coatings deposition (Ref 1, 2). Thermal barrier coatings (TBCs), typically having a bi-layer structure with metallic layer and a ceramic top layer, are used in the hot sections of gas turbines and in diesel engines to provide components with resistance against high-temperature degradation (Ref 3-8). SPS TBCs offer a low-cost alternative to the more established, yet expensive, electron-beam physical vapor deposition (EB-PVD) TBCs. The main advantage of SPS deposition technique is its ability to provide a vertical columnar structure as in EB-PVD as well as a compact horizontal structure by changing the process parameters (Ref 9, 10). The vertical columnar structure is believed to have a superior strain tolerance during thermal cycling than by other low-cost coating deposition techniques like atmospheric plasma spraying (Ref 11).
The choice of material for the coating deposition can have a significant influence on the coating performance at high temperatures. Although yttria stabilized zirconia (YSZ), with its attractive properties, is still the industry standard for the topcoat material, the recent focus has been on other materials that can circumvent the limitations of conventional YSZ, for instance, the high-temperature phase stability of YSZ (Ref 12, 13). Among all the materials being researched for the topcoat layer in TBCs, pyrochlores of A2B2O7-type like lanthanum zirconate and gadolinium zirconate are considered to be potential candidates for the topcoat layer due to their low thermal conductivity and excellent high-temperature phase stability (Ref 14, 15). While the rare earth-based zirconates of lanthanum and gadolinium outperform YSZ when it comes to certain properties like low thermal conductivity and high-temperature phase stability (Ref 14), there is only a limited amount of data available for these materials on their performance during corrosion. Lanthanum zirconate TBCs exhibit minor damage in the presence of vanadium pentoxide while it degrades severely in the presence of sulfates of sodium and magnesium (Ref 16). Lanthanum zirconate is also known to have processing issues during which the material tends to lose its stoichiometry (Ref 17, 18). Gadolinium zirconate, deposited by SPS, on the other hand, has shown to be more susceptible to corrosion-induced damage when exposed to a salt mixture of vanadium pentoxide and sodium sulfate (Ref 19). The main degradation came from attack by the vanadium, and it was also reported in the literature that no direct chemical reaction between gadolinium zirconate and sodium sulfate was known (Ref 20). This makes gadolinium zirconate better resistant to the sulfate environments. However, when gadolinium zirconate is deposited using SPS technique, a columnar microstructure can be generated (Ref 21). The columnar microstructure, with its columnar gaps, likely has effective pathways for the molten salts, and the salts can easily reach the bond coat. The sulfates are known to degrade the bond coat material, which is typical of MCrAlY type (M is Ni and/or Co) (Ref 22, 23). Another limitation with gadolinium zirconate is its thermochemical incompatibility with the thermally grown oxide, alumina (Ref 24). This led to the development of multilayered coatings which were proven to have a better life during thermal cycling than single-layered TBCs (Ref 25-27). Gadolinium zirconate is also known to have lower fracture toughness compared to YSZ (Ref 28). This makes crack propagation much easier in gadolinium zirconate. A novel approach that was developed is to blend gadolinium zirconate and YSZ to make a composite of gadolinium zirconate and YSZ and to deposit this composite on the top of YSZ. A composite of gadolinium zirconate and YSZ has shown to have better resistance against corrosion-induced damage than pure gadolinium zirconate in the presence of V2O5 and Na2SO4 (Ref 21). It is of interest to understand how such a coating performs in the presence of a salt mixture of sodium sulfate and sodium chloride. The environment with sulfates and chlorides of sodium is common during the operation of land-based gas turbines used specially for offshore applications. As per the best of authors’ knowledge, no such work has been published before and understanding the performance of these coating systems in the presence of a salt mixture of sodium sulfate and sodium chloride is the focus of the current investigation. The term gadolinium zirconate is abbreviated as GZ hereafter.
Two coating systems, a double-layer composite of GZ + YSZ/YSZ, and a triple-layer, dense GZ (DGZ)/GZ + YSZ/YSZ are studied in this work. The expected purpose of the dense layer is to seal the columnar gaps which could have an influence on the corrosion resistance. The corrosion behavior on the TBC system as a whole is studied in this work.
Hastelloy® X in the form of disks with a diameter of 25.4 mm and thickness of 6.35 mm was used as the substrate in the present work. The substrate disks were grit blasted using alumina of 220 grit size prior to bond coat deposition to achieve a surface roughness, Ra, value of 3 μm. An MCrAlY-type bond coat, Amdry 386 with nominal composition of Ni18Co13Cr10Al0.1Y was deposited on the substrates using a M3 gun (UniqueCoat, Virginia, USA) by the high-velocity air fuel (HVAF) process.
For the top coat, three different suspensions (1 commercial and 2 experimental) manufactured by Treibacher Industrie AG (Althofen, Austria) were used in this work. The first suspension was an ethanol-based commercially available AuerCoat YSZ suspension, which had a mean particle diameter of 500 nm and a solid load content of 25 wt.%. The second suspension was an experimental type ethanol-based suspension comprising of a 50:50 wt.% mixture of GZ and YSZ. The mean particle diameter was 500 nm, and the solid load in the suspension was kept at 25 wt.%. The third is an experimental type water-based suspension comprising of GZ with a mean particle size of approximately 500 nm and the solid load content of 40 wt.%. The reason for using a 40% solid load water-based suspension was to alter the viscosity and surface tension. A higher viscosity and surface tension of the solvent lead to relatively poor atomization of the suspension droplet which eventually results in a denser coating deposition (Ref 29). The suspensions were kept on rollers overnight to obtain good dispersion of the solute in the suspension.
The bond-coated substrates were preheated prior to the topcoat deposition. Preheating of the bond-coated substrates was carried out using the Axial III Mettech gun which was operated without the suspension. The surface temperature of the bond coat was maintained at approximately 200 °C during the preheating. Preheating could help to remove the volatile dirt from the surface. Top coats were deposited using a Axial III Mettech gun (Metttech Corp, Vancouver, Canada). The spray process was stopped for about 30 min after the deposition of each layer (YSZ, GZ + YSZ, and dense GZ, respectively) to change the suspension in the feeding system for the subsequent layer. The start and stop of the spray process during a multiceramic layered TBC deposition could result in discontinuity of the TBC due to horizontal cracks at the interface of each new layer. However, in this work, before the start of each new layer deposition, the surface was preheated in order to ensure continuity of the layers and horizontal crack-free interface. Further details regarding the deposition process can be found in our previous work (Ref 21).
Corrosion tests were conducted with a salt mixture of 75 wt.% Na2SO4 + 25 wt.% NaCl at a concentration of 6 mg/cm2. The salt mixture was spread on the surface of the sample evenly without touching the edges (a clearance of 3 mm from the sample’s edge was maintained). The samples were later placed in a furnace preheated to 900 °C. The samples were held at the test temperature for 8 h after which they were removed from the furnace and allowed to cool in atmosphere (~ 80 min) till they reached the room temperature (~ 25 °C). The exposure time and cooling together constituted one corrosion cycle.
The tested samples were investigated in an x-ray diffractometer (X’pert Pro, Pan Analytical) with Cu as the source (Kα = 0.154 nm). The samples were later infiltrated with epoxy under vacuum to prevent damage to the coating during the subsequent stages of sample preparation. The samples were thereafter cut along the diameter to reveal the cross-section and prepared for cross-sectional analysis in a scanning electron microscope (SEM) according to the normal routine for the thermal barrier coating sample preparation described in (Ref 30). Energy-dispersive x-ray spectroscopy (EDS) was used to identify the elements in the coating’s cross-section.
Selected Visual Images of the Samples During Corrosion
Figure 1 shows selected visual images of the double- (GZ + YSZ/YSZ) and triple-layer (DGZ/GZ + YSZ/YSZ) samples during the different stages of corrosion. The typical salt spread region after the first and fifth cycle is marked with the black lines in the double-layer system. As shown in Fig. 1, the salt coverage is roughly circular. For up to 9 corrosion cycles, the top surface appeared to be free from damage. After the ninth cycle, the damage started to increase slowly from the edges and started to propagate toward the center. After 24 cycles, for both the double- and triple-layer samples, the visible damage was estimated at about 75% of the topcoat surface and the tests were stopped. Just with the visual inspection, there appeared to be no significant difference in the extent of damage between the double- and triple-layer samples. It has to be noted that even though there is no significant difference in the radial direction during visual inspection, there could be difference in the Z-direction (thickness) of the damaged regions. This is very difficult to observe just with visual inspection. It is emphasized here that the purpose of visual inspection, through a camera, is to illustrate the damage development from the edge to the coating center with increased exposure time. Accurate and more in-depth analysis of the coatings’ cross section was obtained from SEM analysis (discussed in the subsequent sections).
As-Received Samples Before Corrosion Tests
Double-Layer (GZ + YSZ/YSZ) System After 24 Cycles
Triple-Layer (DGZ/GZ + YSZ/YSZ) System After 24 Cycles
Interrupted Corrosion Tests-9 Cycles
Comparison Between GZ + YSZ/YSZ and DGZ/GZ + YSZ/YSZ Coating Systems
X-ray Diffraction Before and After Corrosion
After the 24 corrosion cycles in the double-layer coating, parts of the original phases (tetragonal zirconia and cubic gadolinium zirconate) were retained in the top coat. The other observed phases were oxides of Ni (NiO) and other compounds of Ni (NixFeyOz). As most of the coating was damaged, these phases come from the damaged parts of the coating. The top surface at the center of the coating (see Fig. 1) appeared to be unaffected based on the color of the coating compared to the as-sprayed samples’ top surface. At this location, the main phases of the original coating were still retained. This further proves that there is no direct reaction of the corrosive salts with both gadolinium zirconate and yttria-stabilized zirconia. Similar phases, although with different intensities, were observed with the triple-layer DGZ/GZ + YSZ/YSZ coating system. The higher intensity of the NiO signals in the triple-layer system could be because there was relatively more damage from the edge and this could have resulted in higher amounts of corrosion products at the edge compared to the double-layer system.
Hot corrosion of coatings in the presence of molten salts can be considered as a form of accelerated oxidation. Corrosion of coatings, in general, has been studied extensively and reported by other researchers (Ref 31-37). Exposure of coatings to high-temperature results in the formation of the protective TGO, α-alumina. In the case of pure oxidation, the growth of alumina scale is parabolic. The alumina scale formation is controlled by the diffusion of aluminum and oxygen along the grain boundary. During exposure to corrosive salts, the thickness of Al depletion zone was reported to follow an almost linear trend, indicating that the growth of alumina scale is different during the corrosion process (Ref 35). In the present work, the formed alumina scale is under stress due to thermal cycling of the samples (when the samples are cooled in the atmosphere to room temperature), resulting in horizontal cracks at the TGO/topcoat interface (see Fig. 4d). In addition to that, reaction with the salts could result in the alumina dissolution. Consequently, new alumina scale is formed to repair the damaged oxide scale due to both the thermal stress and dissolution. When the continuous alumina scale can no longer be formed, the salts can react directly with the bond coat and result in rapid degradation (Ref 36). Presence of chlorine, in the form of NaCl further accelerates the corrosion by forming internal voids by means of oxychlorination and chlorination/oxidation cyclic reactions (Ref 36). Salt mixtures containing NaCl can cause more severe damage compared to pure Na2SO4 (Ref 23).
In the case of the triple-layer system with a relatively dense layer on the top, the corrosion mechanism is assumed to be the same as in the case of the double-layer TBC system. It must be noted that while the top layer in the triple-layer TBC was dense, there were still few vertical cracks (columnar gaps) in the coating which could not be avoided during the deposition of the TBCs. Due to the availability of fewer infiltration channels for the molten salts in the triple-layer system, most of the salts have overflown to the edges. Due to the inert nature of gadolinium zirconate toward the corrosive salts, the salts could not be immobilized at the top surface. Thus, even with a similar mechanism for both the double- and triple-layer systems, the contribution of damage development from the edge seemed to be higher for the triple-layer system and, therefore, resulted in a relatively higher damage compared to the double layer. It has to be noted here that the large horizontal cracks in the triple-layer system may have occurred long before the tests were stopped and these delamination cracks can serve as an additional penetration channel for the molten salts.
As such, discarding the edge effect by applying some kind of protective coating to the side surfaces may result in similar behavior for the two coating systems. This is the subject of future research. Furthermore, an inert sealing layer may not be the optimum solution for reducing the corrosion damage. A reactive topcoat can restrict the salt infiltration to the upper parts of the coating. The results from the present investigation can be used as a general guideline during the design of coatings about the suitability of an inert sealing layer in the presence of molten salts.
The columnar gaps in the SPS coating served as effective pathways for the molten salt infiltration.
Due to the inert nature of the molten salts with the topcoat material, overflowing of the salts to the side surfaces occurred and resulted in damage development from the edge to the coating center.
The sealing layer in the triple-layer system proved to be ineffective due to the fact that the sealing layer reduced the number of infiltration channels for the molten salts and consequently, resulted in higher overflow of the salts and, thereby, relatively more damage.
Vinnova in Sweden is gratefully acknowledged for funding this research.
- 7.A. Thibblin, Thermal Barrier Coatings for Diesel Engines. Licentiate thesis, Kungliga Tekniska Högskolan, 2017Google Scholar
- 19.K.P. Jonnalagadda, S. Mahade, N. Curry, X.-H. Li, N. Markocsan, P. Nylén, S. Björklund, and R.L. Peng, Hot Corrosion Mechanism in Multi-layer Suspension Plasma Sprayed Gd2Zr2O7/YSZ Thermal Barrier Coatings In The Presence of V2O5 + Na2SO4, J. Therm. Spray Technol., 2017, 26(1–2), p 140-149CrossRefGoogle Scholar
- 30.K.P. Jonnalagadda, K. Yuan, X-H. Li, X. Ji, Y. Yu, and R.L. Peng, Influence of Top Coat and Bond Coat Pre-oxidation on the Corrosion Resistance of Thermal Barrier Coatings in the Presence of SO2. ASME. Turbo expo: power for land, sea and air, volume 6: ceramics, diagnostics, and instrumentation; education, manufacturing materials and metallurgy, Oslo, Norway, 2018.Google Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.