Introduction

Thermal barrier coatings (TBCs) continue to play a pivotal role in advancing energy efficiency of both propulsion and energy generating gas turbines. TBCs based on yttria-stabilized zirconia (YSZ) ceramics are now extensively applied to many critical hot section gas turbine components. Over the last two decades, this coating technology has transitioned from the role of life extension of superalloy components to prime reliant coatings for enhanced performance. Considerable progress has also been realized in understanding the performance of these coatings subjected to multitude of operational parameters. Mechanism-based models with experimentally derived insights are now enabling predictable performance of these coatings and their integration into prime reliant design.

These advancements have also pointed to limits of applicability of the YSZ system. Current YSZ-TBCs are limited to component operational temperatures of about 1200 °C. Above this temperature, two effects are observed. The stability of the beneficial t′ phase is weakened along with propensity for sintering of porous microstructure which is crucial to maintain the necessary coating compliance (Ref 1). This has resulted in a quest to identify alternative ceramics for future TBC applications.

Many considerations are given in the development of new TBC ceramics; among the primary attributes are low conductivity, thermal/phase stability, and processability (Ref 2, 3). But several additional elements have been identified including resistance to extrinsic chemical attack (e.g., calcium magnesium alumino silicate (CMAS) sand, volcanic ash, and coal ash), erosion, and foreign object damage (Ref 1). It is in this context that rare earth pyrochlore-based zirconates have emerged as promising candidate (Ref 1, 4). Among them, gadolinium zirconate (Gd2Zr2O7), hereafter referred to as GDZ, has shown to be a promising TBC material. They exhibit excellent phase stability, low thermal conductivity, high thermal expansion coefficient, high temperature sintering resistance, and ability to suppress chemical deposit induced failures (Ref 4, 5). Deficiencies include lower fracture toughness and incompatibility with the operationally evolving thermally grown aluminum oxide (TGO) layer at the bond coat-top coat interface (Ref 6). Unlike YSZ, these compounds also contain significantly higher rare-earth elements content with concomitant impact on cost. Despite these short comings, there is published evidence of their utilization in advanced aero engines fabricated via electron beam physical vapor deposition (EB-PVD) (Ref 4) and more recently in land-based power generation gas turbines.

Significant academic studies have also resulted on the zirconates processing using both EB-PVD and plasma spray processes (Ref 7-10). As is the case with YSZ, processing plays a crucial role in determining the microstructural character, properties, and performance of this new class of coatings. Thermal conductivity and compliance of zirconate TBCs, the bench mark parameters for TBC design, are also strongly dependent on the process-induced coating defects (porosity and interfaces) both in content and character. As such processing ideologies employed for YSZ may not be applicable to this new class of TBC systems. Thus, judicious control of the properties via process optimization is both a necessity and opportunity. In this paper, through systematic process map studies, the sensitivity of processing parameters on both as-sprayed and thermally aged properties of GDZ are analyzed and benchmarked with those established for YSZ system. Among the properties of interest are thermal conductivity and its stability with extended exposure, as well as mechanical compliance as quantified through recently developed measurements of non-linear elastic properties. The results of this work provide a framework for tailoring microstructure and properties for their integration within complex multi-layer TBCs.

Experimental Methods

All the YSZ and GDZ coatings were sprayed using commercially obtained powder of hollow sphere (HOSP) morphology from Saint-Gobain Ceramic Materials, Worcester, MA, USA. For YSZ coatings, powders with ensemble (10-105 µm) and fine (10-45 µm) particle size distributions were used. The GDZ coatings were sprayed using a coarse (45-105 µm) and fine (10-45 µm) powder cuts. Henceforth, these powders will be referred to as YSZ-E, YSZ-F, GDZ-C, and GDZ-F. The SEM images of the morphologies of these powders are shown in Fig. 1. The choice of these powders was based on the ability to produce coatings with significant variation in their microstructures. Figure 1 (middle) shows the particle size distribution curves of the powders obtained in the dry mode using Beckman coulter LS 13 320 (Brea, CA, USA) particle size analyzer. All the coatings were processed using an atmospheric plasma spraying (APS) Ar-H2 system F4 MB (Oerlikon Metco Inc., Westbury, NY, USA), equipped with a 8 mm nozzle.

Fig. 1
figure 1

Details on particle size distribution and micrographs of the four feedstock powders used for coating deposition, YSZ-fine, YSZ-ensemble, GDZ-coarse, and GDZ-fine. Note that the GDZ-C particles do not contain fine particles belonging to the GDZ-F powder

Full size image

The SEM micrographs of the various coatings and powders were acquired from Hitachi TM3000 (Angstrom Scientific Inc., Ramsey, NJ, USA) in backscattered electron mode. Coatings porosity calculations were performed on micrographs obtained from SEM using ImageJ software utilizing Otsu algorithm, which involves a binary conversion of each image. The microstructural features in the binary image were classified to globular pores, inter-lamellar pores, and microcracks based on their shape. The details of algorithm for such classification analysis are provided elsewhere (Ref 11). In order to obtain statistically sound porosity numbers, 10 images per coating at 500× magnification were analyzed.

Table 1 presents the processing conditions associated with the coatings designed for this study. The conditions primarily varied in terms of plasma input parameters; primary gas (Ar) and secondary gas (H2) flow rates and current. These parameters were selected based on our prior work on YSZ coatings (Ref 12-14) and resulted in significant variation in overall plasma power which introduced a substantial variance in coatings’ microstructures. The YSZ coatings, Y1, Y2, and Y3, and GDZ coatings G1, G2, and G3 were sprayed using YSZ-E and GDZ-C powder, respectively. The finer powder cut for these materials were used to produce the fourth YSZ and GDZ coatings, YF1 and GF1, respectively. For the entire deposition scheme, the powder feed rate was maintained at 30 g/min, and coatings were sprayed at a 100 mm stand-off distance.

Table 1 Details of processing parameters used for YSZ and GDZ coating deposition
Full size table

The in-flight particle diagnostics was conducted prior to the coating deposition to measure particle-state parameters, temperature (T), and velocity (V). The measurements were made using single-particle method DPV-2000 sensor (Tecnar Automation Ltd., Quebec, Canada) on 2500 particles for each processing condition. In order to achieve efficient energy transfer from plasma to injected particle, injection optimization procedure was utilized (Ref 15-17). The details of the procedure are provided elsewhere. This procedure not only results in maximum achievable particle T and V, but also ensures great process repeatability (Ref 18). The optimum carrier gas flow rates for each processing conditions are reported in Table 1.

The coatings were deposited on Al6061 aluminum substrate of three different sizes. All the specimens belonging to one coating condition were prepared during single spray session to avoid process variability associated with torch start-stop process. First, for in situ deposition stress monitoring, the coatings were deposited on an in situ beam curvature monitoring sensor (referred to as the In situ Coating Property-ICP sensor) (Ref 19, 20). Here coatings were deposited on 25.4 mm × 221 mm × 2.24 mm aluminum substrates with simultaneous monitoring of substrate curvature and temperature. Details of this procedure and interpretations of the acquired data are available in the literature. The deposition of coatings on ICP enable the in situ measurement of quenching (evolving) stresses (σev) associated with each coating formation. The coatings were deposited using a ladder-type pattern with 3 mm ladder step size covering entire substrate surface area. For each case, coatings of 400(±15) µm thickness were obtained. After deposition, the specimens were placed on ex situ coating property (ECP) sensor (Ref 21-24), and were subjected to low temperature heat cycles, room temperature to ~250 °C. This characterization technique provides two mechanical properties of coatings; elastic modulus (E) and non-linear degree (ND) (Ref 23, 24). These three properties, obtained from ICP and ECP instruments, are highly sensitive to processing parameters and, collectively, provide insights to the coating microstructure. Further details on these measurements have been provided elsewhere (Ref 21, 24).

Second set of coatings was deposited on 12.5-mm-diameter Al6061 substrates of 10 mm length. Approximately, 80% substrate length was sliced off using high speed diamond saw (Buehler Inc, Chicago, IL, USA), and the remaining substrate material was removed from coatings by dissolving it in concentrated hydrochloric acid (Fisher Scientific Inc, Pittsburg, PA, USA). Further, a thin layer of graphite (1-3 µm) was sprayed on both the free-standing coating surfaces using Aerodag G aerosol (Ascheson Colloids Company, Port Huron, MI). These graphite-coated specimens were used for direct thermal conductivity measurements using a laser flash technique (Micro Flash, Netzch Instruments Inc., Burlington, MA).

The heat treatment was performed using Thermolyne F21130 furnace (ThermoFisher Scientific, Asheville, NC, USA) under ambient conditions. For sintering studies, the free-standing coatings prepared for thermal conductivity measurements were exposed to temperatures 1050, 1100, and 1200 °C for periods between 10 and 225 h. After each incremental exposure, the graphite later was re-applied onto the specimen surfaces, and their thermal conductivity was measured at room temperature. High-temperature thermal diffusivity measurements were conducted using a multiple-furnace laser flash system (FL-5000, Anter Corporation, Pittsburgh, PA). All the high-temperature thermal diffusivity measurements were conducted in an argon atmosphere. High-temperature specific heat measurements were made on a differential scanning calorimeter (STA-449C, Netzsch Instruments Inc.). GDZ coatings with selected aging conditions, 10 and 110 h at 1200 and 1300 °C, were considered for high-temperature thermal conductivity measurements. The procedural details for this measurement are provided elsewhere (Ref 14).

X-ray diffraction spectra for coatings as well as their corresponding feedstock powders were collected with Philips diffractometer type PW 1729 (Philips Analytical Systems, Mahwah, NJ) for the 2θ range of 20-80° at the rate of 1°/minute under ambient conditions.

Compositional analysis of coatings and starting feedstock materials was conducted using inductively coupled argon-plasma atomic emission spectroscopy (ICP-AES) at Dirats Laboratories, Worcester, MA, USA. The results were provided by Saint Gobain Inc., Worcester, MA. The analysis was conducted using 10 g of coating or of their corresponding feedstock powder materials.

Third set of specimens was prepared to evaluate material’s resistance to lignite fly ash of the composition shown in Table 2. The lignite ash (in powder form) used in this study was synthesized and provided by Prof. Nitin Padture of Brown University, RI, USA. Details of the fabrication procedure and the compositional analysis of lignite ash are provided elsewhere (Ref 25). Again, acid dissolution procedure was utilized to prepare free-standing coatings. For the test, thin ash layer (~35 mg/cm2) was uniformly applied to the coatings in their free-standing form, and then was inserted into a furnace for heat treatment at 1200 °C for 24 h. The furnace was heated along with the specimens at the heating rate of 5 °C/min, and was left to cool down under natural cooling conditions after the completion of heat treatment. Post heat treatment, the specimens were mounted and polished for microstructural evaluation under SEM. Only, GDZ coatings (G1-G3 and GF1) were considered for this test, since YSZ coatings have been reported to exhibit no resistance to CMAS or lignite ash (Ref 26).

Table 2 Compositional wt.% of lignite ash (Ref 25)
Full size table

Results and Discussion

Particle Diagnostics and Process Maps

Figure 2(a) shows particles temperature distribution for the eight coating conditions, obtained from the temperature measurements of 2500 single-particle measurements by DPV sensor. The distribution provides an assessment of temperature achieved for the various sizes of particles present in the feedstock (Ref 15, 16). The average number for temperature and velocity is shown in Fig. 2(b). These two different schemes of data presentation, distribution, and averages are followed to capture particle state in detail. A comparison between YSZ and GDZ coating conditions from Fig. 2(a) reveals, as expected, that the temperature distribution for both the material shifts toward higher temperature regimes with increasing plasma power. These observations are consistent with previous work and are attributed to the higher degree of particles melting (Ref 15, 17). Differences in the distribution shape between YSZ and GDZ coating conditions can also be noted. The distribution range of YSZ conditions is significantly wider than those of GDZ conditions, indicating larger variation in melting among particles (Ref 15). These differences can be attributed to two facts:

Fig. 2
figure 2

Results from in-flight particle diagnostics of GDZ and YSZ powders performed using DPV2000 sensor. (a) Particle temperature distribution for four processing conditions used for coating depositions. The vertical dotted lines correspond to the melting point of the material. (b) First order process map developed for the two materials using average temperature and velocity of particles. The dashed arrows from left to right point to the data corresponding to coarser and finer particles at similar plasma conditions

Full size image
  1. (1)

    Somewhat lower melting temperature of GDZ powder (2570 °C) compared to YSZ (2650 °C) (Ref 27) may be helpful in the melt-ability in the plume.

  2. (2)

    Differences in particle size distributions between the two powder materials. In contrast to YSZ-E powder, the GDZ-C powder did not contain finer (10 to 40 µm) particles (Fig. 1). Finer particles will achieve temperatures more rapidly in the plume than coarser particles, and may contribute toward the broadening of the temperature distribution. Furthermore, GDZ-C powder despite having a large average size has a very narrow distribution with almost the same maximum size particles as the YSZ-E.

The temperature distribution comparison between YF1 and GF1 conditions indicated that GF1 condition not only had narrower but also occupied higher temperature region distribution. Since both the conditions were sprayed using similar particle size, the differences between the two distributions are primarily due to differences in their melting temperatures. This clearly indicates that the temperature distribution differences between Y1-3 and G1-3 conditions have a negligible contribution from the differences in their particle size ranges.

The first order process map (particle state), developed using the average particle T and V for the eight coating conditions, is shown in Fig. 2b (Ref 13). The T and V values shown in the plot are the average of the measurements on 2500 particles. For clarity, the standard deviations associated with particle state measurements are not included in figure; however, they are summarized in Table 3. The average temperature (T avg) numbers are comparable for the two powder materials for a given processing conditions, i.e., G1-3 has similar T avg as Y1-3. However, their average velocities are significantly different. YSZ coating conditions generally showed higher average velocities than GDZ coating conditions. This could be attributed to the differences in the particle sizes of the two feedstocks. The GDZ powder with larger average particle size results in increased particle dwell times in the plasma experiencing prolonged exposures. In addition, the lower melting point of GDZ material contributed further to achieve similar surface temperatures as those achieved by the YSZ particles, which had lesser dwell time. Thus, the differences observed are a manifestation of both the particle sizes as well as their melting extent.

Table 3 Average temperature and velocity of particles corresponding to different YSZ and GDZ coating processing conditions
Full size table

Compositional and Phase Analysis

Recent studies have reported a preferential loss of Gd due to vaporization from Gd2Zr2O7 during the dwell period of particles in plasma plume at high plasma enthalpy conditions (Ref 7). This compositional change in the material could result in potential phase change, altering the functionality of the material, such as CMAS resistance, thermal conductivity, etc. Therefore, in order to examine the extent of Gd loss in the coating material and any associated phase change occurred during deposition, compositional and XRD analyses were performed on the four GDZ coatings. Due to limited material availability, G1 coating was excluded from compositional analysis. The results from the ICP mass spectroscopy analyses on the coatings are provided in Table 4. Both the fine and coarse GDZ feedstock materials showed ~58 wt.%, i.e., 33 mol.%, Gd2O3 composition belonging to the pyrochlore phase composition of the GDZ materials. The corresponding coatings sprayed using these two powders showed no significant loss in Gd, and hence in Gd2O3 during processing. The Gd2O3 content in G2 (34 kW) and G3 (42 kW) coatings was lower only by 0.09 and 0.45 mol.% from GDZ-C powder. The latter suggests that higher power may be causing Gd evaporation but the effective change is still very small. Similarly, GDZ-F powder coating (GF1) showed only 0.53 mol.% loss in Gd2O3 during processing. These results, summarized in Table 4, suggest that all the GDZ coatings shared similar compositions with their feedstock materials. These results are also complemented by coatings XRD analyses (discussed later). The observed results are somewhat different than those reported by Mauer et al. (Ref 7); they have shown ~3% in Gd loss at ~47 kW of input plasma power, the highest plasma condition considered in this study. Although the plasma kilowatts used for coating deposition in this study is close to those used by Mauer et al., the use of higher enthalpy cascaded plasma torch, TriplexPro™, may have been the reason for the different conclusions between the two studies.

Table 4 Mol.% of Gd2O3 and ZrO2 of GDZ powders and selected coatings measured by ICP mass spectroscopy
Full size table

The XRD patterns obtained for the two GDZ powders and the four coatings complement the conclusion from compositional analysis (Fig. 3). Figure is indexed with fluorite (F) and pyrochlore (P) peaks for the Bragg angle range considered for measurements. There were no noticeable differences observed among all the patterns. Further, given that the mole fraction of Gd2O3 (0.5) in GDZ material belongs to the characteristic pyrochlore super-lattice structure, XRD pattern of neither powders nor coatings reflected peaks corresponding to the super-lattice pyrochlore peak P331 and P511 (Ref 27, 28). This indicates that both powders and coatings possessed only metastable fluorite (F) structures. Note that the pyrochlore lattice is highly ordered structure, and would require slow kinetics for phase formation. The HOSP-GDZ powders, which are originally plasma densified, and APS-GDZ coatings undergo additional rapid quenching step during processing. These manufacturing steps introduce high degree of disorder and localized residual stresses in the lattice structure, restricting the material to order into a super-lattice structure. The significantly large peak width in XRD patterns shown in the figure supports the above argument of higher degree of disorder structures in the materials.

Fig. 3
figure 3

X-ray diffraction records for the as-sprayed GDZ coatings as well as the feedstock material. The fluorite (F) and pyrochlore (P) peaks are indexed in the figure

Full size image

Based on the compositional analysis and XRD results, it can be concluded that the all the GDZ coatings prepared for this study did not undergo any noticeable phase and compositional change within the subjected processing parameters.

Process-Microstructure Relations

The microstructures of YSZ and GDZ coatings are shown in Fig. 4. A systematic reduction in porosity with increasing plasma power conditions can be observed for both GDZ and YSZ coatings. In general, the GDZ coatings exhibited relatively lower porosity than YSZ coatings for all four plasma conditions, except for GF1 and YF1 coatings which were sprayed using finer powder cuts resulting in similar porosity level. Despite the fact that G1-3 coatings were deposited using coarser powder cut (GDZ-C) than the Y1-Y3 coatings sprayed using ensemble powder cut which includes fines, the porosity level in the GDZ coatings was lower. It can be attributed to the lower melting temperature of GDZ material, which has shown a higher degree of melting of GDZ particles compared to YSZ, leading to denser microstructures (Fig. 2). The reduced melting of YSZ particles, on the other hand, resulted in a larger population of inter-lamellar and globular pores. The overall porosity numbers and the microcrack density as obtained from image analysis for each coating are summarized in Fig. 5.

Fig. 4
figure 4

SEM micrographs of the YSZ (Y1-3 and YF1) and GDZ (G1-3 and GF1) coatings

Full size image
Fig. 5
figure 5

Microstructural analysis of YSZ and GDZ coatings. The overall porosity and microcrack density values for the GDZ and YSZ coatings

Full size image

An interesting observation can be made regarding the microcrack density of the two coating sets. Figure 6 shows micrographs of G2 and Y2 coatings, and their binary images showing only the features classified as microcracks. The binary image was generated via fine tuning of filtering algorithm to only show the microcracks present in the acquired micrograph. The details of such algorithm are provided elsewhere (Ref 29). Both the SEM images and the binary processed images indicate the presence of significantly higher microcrack density in GDZ (G2) coatings compare to YSZ (Y2) coatings. Further quantification of microcrack density (Fig. 5) confirms the observations. These observations can primarily be associated with lower fracture toughness of the GDZ material compared to the YSZ coatings potentially attributed to YSZ’s ferroelastic switching mechanism (Ref 30). Confirmatory evidence has been obtained in a previous study comparing fracture toughness of plasma-sprayed YSZ and GDZ subjected to different processing parameters (Ref 31). The GDZ material with cubic fluorite structure lacks the benefit of this ferroelastic mechanism, therefore resulting in low fracture toughness.

Fig. 6
figure 6

The differences in microcrack densities between GDZ and YSZ coatings. SEM micrograph of Y3 and G3 coatings along with their binary images filtered for microcracks-defect only

Full size image

Thermal Conductivity

Thermal Conductivity at Ambient Temperature of Aged Coatings

Figure 7 shows the thermal conductivity numbers of the four GDZ coatings exposed to various thermal aging durations at 1200 °C. For reference and comparison, thermal conductivity of Y2 specimen is also included in figure. For each thermal exposure conditions, all the GDZ coatings show lower thermal conductivity than the reference YSZ (Y2) coating. In addition, a systematic increase in thermal conductivity of coatings was observed with increasing exposure time, which is primarily attributed to the reduction of defect population in coatings with exposure durations. These results are consistent with prior work (Ref 14), which shows not only a rapid increase in conductivity during the initial exposure (first 10 h) at 1200 °C but also a sensitivity to initial microstructure. This increase was the largest for YSZ (Y2) coating and can be attributed to the higher sintering rate of the YSZ material. The as-sprayed thermal conductivity was higher for GF1 and G3 than G1 and G2 coatings associated with initial coating densities. In fact, the two pairs individually have almost same starting thermal conductivities. With increasing aging duration, the differences in thermal conductivities between the two pairs were higher, and remained almost constant for the extended aging durations considered in this study. The aged GF1 coating exhibited largest increase in thermal conductivities among the four GDZ coatings. This coating was sprayed using fine powder and lacked microstructural defects with large opening dimensions, such as inter-lamellar and globular pores, and thus had greater propensity for sintering and concomitant increase in thermal conductivity.

Fig. 7
figure 7

Room temperature thermal conductivity data of the four GDZ and one YSZ (Y2) coatings with various exposure durations at 1200 °C

Full size image

Larson Miller Parameters

The Larson Miller parameter (LMP) incorporates the effects of aging time and temperature into a single descriptor representing an empirical fit of the exposure conditions. The parameter was first introduced by Larson and Miller in 1952 to describe the time-temperature relationship for rupture and creep stress (Ref 32). It was further extended to establish relationship between thermal conductivity and thermal exposure history of TBCs to assess their sintering rates (Ref 14, 33). The same approach is utilized in this paper. Here, the study was conducted only on the coatings (G1, G2, and G3), sprayed using GDZ-C powder. Equation 1 shows the relationship between the LMP and aging time and temperature

$$LMP = T(\ln t + C),$$
(1)

where T and t are the exposure temperature in Kelvin and time in seconds, respectively. The constant C is determined from a fit through the data, and this value was obtained to be 20 for YSZ coatings in earlier work from Tan et al. (Ref 14), and has been maintained the same for the GDZ coatings in this paper. Figure 8 shows the thermal conductivity and LMP map for the three GDZ coatings along those obtained for various YSZ coatings with varying processing conditions and microstructure (Ref 14). The data in figure correspond to several aging duration at 1200 and 1300 °C. The values associated with the YSZ band are obtained from ref (Ref 14). The figure also shows a linear fit to the data for different coatings. The slope of linear fits, summarized in Table 5, provides an assessment of coating’s sintering rate, i.e., higher the slope, faster is the sintering rate. All the GDZ coatings (G1, G2, and G3) showed substantially lower slopes than the YSZ coatings, owing to slower sintering rate of GDZ material itself. Interestingly, the differences among the three GDZ coating also appear to be significant. The sintering rate, relative to the starting microstructure of a coating, depends on the degree of reduction of microstructural defects. The coating G1 having the highest defect density among the three coatings undergoes the largest microstructural change and shows the highest slope or sintering rate. Further, the extrapolations of linear fit lines for very high ranges of LMP (>58 × 104) conditions from all three GDZ coatings seem to coincide, suggesting the equivalency of microstructural porosity achieved after an extended sintering.

Fig. 8
figure 8

LMP vs. thermal conductivity plot for G1, G2, and G3 coatings. The measurements on various YSZ coatings are shown as YSZ Band

Full size image
Table 5 Slope corresponding to the linear fit of K vs. LMP data for GDZ and YSZ coatings
Full size table

High-Temperature Thermal Conductivity

In order to examine the effect of microstructural differences on the high-temperature thermal conductivities of GDZ coatings, the measurements were performed on the three GDZ coating architecture G1, G2, and G3. Coating GF1 was not included in the study due to limited availability of instrument time at Oak Ridge National Lab. The measurements were performed only on the specimens aged for the two different durations (10, 110 h) and at the two temperatures (1200 and 1300 °C). No measurements were made on the as-sprayed coatings, as there will be microstructural changes in the coatings during measurements at high temperatures. The measured thermal conductivities of coatings are shown in Fig. 9. Several interesting observations can be noted from the figure.

Fig. 9
figure 9

High-temperature thermal conductivity measurements of GDZ (G1, G2, and G3) coatings. The measurements were performed on four sets of specimens with different thermal aging conditions, 10 and 110 h at 1200 and 1300 °C each

Full size image
  1. 1.

    For all the cases, the coating with higher porosity, G1, maintains relatively lower thermal conductivity over the entire span of temperatures compared to the denser coatings, G2 and G3, with similar aging conditions of 10 as well as 110 h at 1200 °C. These differences in the thermal conductivities appear to be not significant for G1 and G2 coatings after 1300 °C thermal aging conditions; although the G3 coating, which is the densest among the three, showed noticeable higher thermal conductivity than G1 and G2 after the equivalent thermal exposure. This indicates that the exposure to the higher temperature, i.e., 1300 °C, resulted in similar coating microstructures of G1 and G2, but not G3, coatings. On the other hand, the three coatings were able to maintain the microstructural differences at 1200 °C exposure temperature, and continued to show the trend of lower thermal conductivity corresponding to the higher porosity coatings.

  2. 2.

    In general, for each case the thermal conductivity of coatings first decreased with temperatures, and then exhibited a rise in values toward higher measurement temperatures. This behavior is consistent with what has been reported earlier by other researchers (Ref 14, 34). It was reported that the decrease in conductivity with temperature is associated with stronger phonon scattering with increasing temperature up to mid temperature range. However, with further elevation in temperature, heat conduction due to radiation dominant behavior leads to rise in thermal conductivity of the ceramics. It has been observed earlier that the transition from phonon scattering dominating to radiation dominating conditions is approximately ~700 °C range (Ref 14), which appears to be similar for GDZ coatings in this paper.

  3. 3.

    The differences between thermal conductivities of coatings aged for 10 and 110 h at 1200 °C are significantly larger than the coatings aged at 1300 °C. In fact, for the latter temperature case, there is almost no difference in thermal conductivities between the coatings aged for the two significantly different durations, especially for low-to-mid temperature measurements range (up to 700 °C). This can be explained by the larger extent of sintering with 1300 °C aging conditions.

  4. 4.

    Unlike, in the case of 1200 °C thermal exposure conditions, the thermal conductivity of all the three coatings at higher ranges of measurement temperatures (>700 °C) was lower for 110 than 10 h aging duration at 1300 °C. These differences between thermal conductivity numbers at a given measurement temperature appear to be reducing with increasing coating density; it was the highest for G1 and the lowest for G3. Although the exact mechanism associated with this observation is not clear and would require further analysis, one possible explanation can be attributed to the greater ordering degree of Gd2Zr2O7 phase with longer exposure durations. An ordered Gd2Zr2O7 phase would offer greater phonon scattering, and hence lower thermal conductivity.

Evolving Stress and Elastic Properties

Figure 10 shows the evolving stress (quenching) and non-linear elastic properties of YSZ and GDZ coatings obtained by curvature-based technique using ICP and ECP sensors. The characteristic as-recorded curves from the two measurements are shown in Fig. 10(a) and (b), representing measurements on Y2 and G2 coatings. Note that the two graphs have different abscissa, and the curves in both the graphs are shifted (vertically for (a) and horizontally for (b)) to enhance the clarity. It is evident from figure that the slope of curvature rise of ICP measurements during deposition (Fig. 10a) for G2 coating is lower than that for Y2 coating, suggesting lower net-evolving stress in the G2 coating. It is rather surprising, since, given the enhanced melting associated with GDZ material, one should expect the GDZ coating to show higher quenching stress buildup due to good splat welding than comparable YSZ coating. Note that the GDZ coatings in general exhibit denser microstructures than its YSZ counterparts (Fig. 4). The opposite trend here can be attributed to the extensive (higher) microcracking with GDZ coatings shown in Fig. 6. It has been inferred earlier that microcracking in the YSZ and GDZ coatings is majorly due to release build up stresses during deposition. The cubic phase GDZ material, unlike YSZ which possesses higher fracture toughness due to the presence of t′ phase, undergoes larger extent of microcracking compared to YSZ coatings, resulting in lower net-evolving stress in the GDZ coatings (also confirmed in Fig. 6).

Fig. 10
figure 10

Evolving coatings stress and anelastic properties of YSZ and GDZ coatings obtained from ICP and ECP measurements. (a) and (b) are characteristic curves of the two sensors showing differences in GDZ (G2) and YSZ (Y2) coatings. (c) and (d) are the computed properties of all the coatings

Full size image

The effect of microcracking is also applicable to the interpretation of non-linear curvature-temperature plots of the Y2 and G2 coatings (Fig. 10(b)). Both curves, for Y2 and G2, show a characteristic anelastic curve for plasma-sprayed crystalline ceramics (Ref 23, 24). The figure clearly indicates that the curve for GDZ coating is more non-linear than that for YSZ coating despite the higher microstructural density. Our past work has shown that the larger density of these microcracks in coatings can significantly enhance their anelastic behavior (Ref 23). Moreover, the presence of these defects also lowers the elastic modulus of the coating. These observations are quantified based on the previously established methodologies and are summarized in Fig. 10(c) for all the eight coatings. The figure points to a general trend for other YSZ and GDZ coatings—denser coatings showing higher modulus and evolving stress, although lower non-linearity. While comparison between GDZ and YSZ coatings suggest that the former generally posseses lower evolving stress, reduced modulus, and higher non-linear degree for all the spray conditions, attributed to the higher microcrack density of GDZ coatings.

Resistance to Lignite Fly Ash

The microstructural evaluation of GDZ coatings exposed to lignite fly ash at high temperature (1200 °C) for 24 h showed that the material mitigates the ash penetration into the coatings (Fig. 11). These preliminary results are consistent with those reported by many other researchers (Ref 4, 35). Figure 11 shows the microstructures of coating G2 (representing G1-3 coatings) and GF1 after lignite ash treatment. The total reaction zone depth for the coatings sprayed with the coarse GDZ-C powder, G1, G2, and G3, was approximately the same (~100 μm). However, the high density coating GF1 produced with the fine powder only showed an interaction depth of ~35 μm. This suggests that reducing porosity/crack network can complement the thermochemical interaction of Ash with GDZ to offer more effective resistance to Ash/CMAS damage in TBCs. These results are preliminary as they were only conducted at 1200 °C which is on the lower end of the reaction spectrum. Future studies will involve higher temperature exposure comparisons as well as a more detailed analysis of the microstructural attributes that govern the ash interactions.

Fig. 11
figure 11

SEM micrographs of G3 and GF1 coatings after lignite ash exposure for 24 h at 1200 °C. The approximate reaction zone depths for G3 and GF1 coatings are 100 and 30 µm

Full size image

Durability Associated Aspects of GDZ Coatings

Despite the promises offered by GDZ material such as high temperature phase stability, CMAS mitigation and low thermal conductivity, several challenges exist in terms of its insertion in a TBC system. First, the thermochemical incompatibility of the material with the alumina TGO restricts its application directly on to the bond-coated superalloy (Ref 6). In the past, a bi-layer approach has been considered by various researchers to overcome this issue, involving the deposition of a YSZ layer on bond-coat followed by GDZ top layer (Ref 36-39). Further, the differences between mechanical properties, such as fracture toughness (Ref 31), and thermal properties, such as sintering rate (this study) of the two materials, introduce significant complexities in terms of processing and development of these multi-layer YSZ-GDZ systems. For example, the differences in sintering rates of the two coating materials, as shown in this paper, can lead to strain-induced mismatch stresses at the layers interface leading toward layer spallation (during isothermal exposure such as furnace cycle testing). Additionally, the lower fracture toughness of GDZ material (Ref 31) may not only contribute toward the initiation of a premature TBC spallation crack but also significantly limit the erosion performance of the TBC. These system level challenges suggest that a complex multi-layer assembly of the coating both in terms of architecture and materials will be necessary for furnace-based testing of durability. Activities are being carried out toward such coating design and performance assessment and will be reported in the future.

Summary and Conclusions

A process-property correlation was explored on various GDZ coatings along with equivalent YSZ coatings processed over a range of plasma spray parameters. The in-flight particle diagnostics revealed a general trend that the GDZ undergoes more complete melting than YSZ possibly attributed to its lower melting point. This resulted in relatively denser microstructure of GDZ coatings compared to their YSZ counter parts for the explored parameter space. Despite the higher density, the GDZ coatings showed consistently lower thermal conductivity than YSZ coatings at both ambient and elevated temperature while also displaying parametric/microstructural sensitivities to conductivity. This trend remained consistent not only for as-sprayed GDZ coatings, but also for thermally aged coatings. The Larson Miller study confirms that the sintering rate for GDZ coatings is substantially lower than YSZ coatings.

Microstructural analysis showed the presence of higher microcrack density in GDZ coatings compared to YSZ coatings, attributed to lower intrinsic toughness of the materials, causing larger extent of microcracking in order to release quenching stresses. This explains lower stress buildup in coating during deposition as well as reduced elastic modulus and higher anelasticity in GDZ coatings compared to YSZ coatings. The phase analysis using x-ray diffraction and the compositional analysis using induction-couple-plasma mass spectroscopy revealed no significant phase and materials chemistry change for the range of plasma power used to deposit GDZ coatings. The lignite ash exposure test at 1200 °C isothermal exposure confirmed efficacy of GDZ to react with the ash and prevent their penetration into the coating. It was found that the reaction zone depth depends on the material’s density. Coating with significantly higher density (GF1) showed lower penetration depth than the other coatings (G1-G3), possibly due to interconnected porosity present in porous coatings.