APS NiCoCrAlY+HfSi Bond Coat
Figure 2 shows the SEM cross-sectional micrograph of the NiCoCrAlY+HfSi BC sprayed via the Axial III Plus (without the YSZ TBC). The maximum BC surface temperature reached during spraying was ~550 °C (monitored using an infra-red camera). The chemical composition of the BC was determined via EDX within the rectangular area (~500 µm x ~55 µm) featured over the BC and the results shown in Table 2. It is possible to observe that the chemical composition of the as-sprayed BC is found within the range of that of the MCrAlY powder feedstock (provided by the powder manufacturer). The amount oxygen in the as-sprayed BC is ~1.4 wt.%. This amount is within the range of those exhibited by NiCoCrAlY+HfSi BCs when deposited via APS at the NRC (i.e., ~0.7-1.5wt.%) and when measured by the same technique.
Table 2 Chemical composition of the MCrAlY feedstock and as-sprayed BC The BC Ra value was 13.31 ± 1.26 µm (n = 10 @λc = 2.5 mm). According to Feuerstein et al. (Ref 1), in order to provide optimum adherence for a ceramic TBC, a surface roughness (Ra) of the MCrAlY BC of at least approximately 10 µm is desirable so as to mechanically anchor the layers together. Typical Ra values of thermally sprayed MCrAlY BCs are within the ranges 7-10 µm as reported by Schmidt et al. (Ref 19) and by Eriksson et al. (Ref 20), as well as, 10-13 µm as reported by Haynes et al. (Ref 21). Consequently, the Ra value reported for the BC produced in this work is located in the upper range of the values reported in the literature and above the minimum Ra value recommended by Feuerstein et al. (Ref 1).
Schmidt et al (Ref 19), Eriksson et al. (Ref 20) and Bossmann et al. (Ref 22) demonstrated that the higher BC roughness values lead to a longer lifetime performance of YSZ APS TBCs when subjected to thermal cycling. Although the YSZ TBC cracking, failure and spallation occurs at the TGO region and a critical TGO thickness is needed for the spallation of the YSZ TBC, the critical value of TGO thickness depends on the roughness value of the BC (i.e.; the rougher the BC, the thicker the TGO critical thickness). Understanding the mechanisms of failure of TBCs is a complex issue. Besides characteristics like BC surface roughness and TGO layer thickening; factors like BC composition, BC spraying (e.g., BC as-sprayed oxidation), operating temperature, cycles, mechanical loading, stresses, phase transformations and atomic diffusion play important roles in the outcome. Nonetheless, in a “very simplistic way”, it is known that the crack path that leads to the YSZ TBC spallation originates and propagates at and/or alongside the TGO layer (which forms over the BC roughness). For this reason (for the same BC composition and deposition method) usually the rougher the surface of the BC, the longer the critical path for crack propagation to promote YSZ TBC spallation. As a consequence, typically rougher BC surfaces tend to translate into longer thermal cycle lives for APS YZ TBCs. Therefore, the high Ra value of ~13 µm reported for the BC manufactured in this work can be preliminary considered as “a desirable target”; but further thermal cycle testing is necessary to confirm this hypothesis.
By looking at Table 1, one can notice that the BC powder was sprayed at 50 g/min of powder feed rate and its DE value was 55% when sprayed at 71 kW. It is difficult to find information on the open literature about the DEs of MCrAlY BCs deposited via APS. However, based on NRC’s own experience, the combination of the powder feed rate employed (50 g/min) and resulting DE (55%) is considered to be “good” for a BC manufactured via APS. Of course, this DE value could be easily augmented by further increasing the plasma torch power (for example). On the other hand, the BC can become highly oxidized. The production of BCs via APS needs to find the “right balance” of obtaining at least “acceptable” values of powder feed rate, DE, as-sprayed BC Ra and oxidation levels.
APS YSZ TBC Architectures: Microstructure
As described in Table 1, the YSZ power feed rate and its DE value were 100 g/min and 70%, respectively. These can be considered “very high values” and show the effectiveness of a N2-based high-enthalpy APS torch to spray large amounts of YSZ and still obtaining an impressive DE level. It is important to point out that the majority of the porous APS YSZ TBC manufacturing being done today is performed by “legacy” torches, employing Ar-H2 plasmas, where the YSZ powder is typically sprayed at 60 g/min or lower and the DE values are about 40% or below.
Nonetheless, it needs to be stated that superior DE levels and powder feed rates do not necessarily yield desired results. In fact, it is the contrary; they will tend to increase even further the stresses that occur during thermal spray deposition, which can lead to unwanted segmentation cracking, inter-pass delamination and reduced bond strength. For this reason, observing the microstructures is of paramount importance.
Figure 3 shows the SEM cross-sectional micrograph of the two full architecture TBCs manufactured via the Axial III Plus for this work: (i) ~420 µm-thick YSZ TBC (Fig. 3a) and (ii) ~930 µm-thick YSZ TBC (Fig. 3b). It is important to remember that both YSZ TBCs (i) were deposited on the same ~160 µm-thick BC and (ii) sprayed using the same set of spray parameters (as per Table 1), i.e., only the YSZ thickness changes. At the YSZ DR of ~23 µm/pass, a total of 18 and 40 torch passes were employed to deposit the ~420 µm-thick and ~930 µm-thick YSZ TBCs; respectively. The maximum YSZ surface temperature (monitored using an infra-red camera) initially started to increase with each single torch pass (as expected); but stabilized at ~600 °C after about 8-9 torch passes for both YSZ top coats. Consequently, a sample temperature stabilization was reached during spraying. This sample temperature stabilization most likely influenced the mechanical properties of TBCs, such as bond strength and elastic modulus values (to be discussed in the next paragraphs).
By looking at Fig. 3, it is possible to observe that both YSZ TBCs exhibit a conventional porous microstructure. It is noticed the presence of interlamellar boundaries, globular voids and microcracks; which are the result from previously molten and semi-molten YSZ particles that flattened, overlapped and re-solidified into an assembly of lamellas during spraying. There are no major cracking segmentation, horizontal inter-layer delamination or adhesion gaps at the YSZ/BC interface. This is an important result. It shows that under controlled conditions of spraying (e.g., spray parameters, torch/substrate relative motion and maximum coating temperature during spraying), superior yield TBC manufacturing can be performed without producing unwanted defects into the TBC microstructure. The as-sprayed density and porosity values of the YSZ TBC are ~5.15 g/cm3 and ~14%, respectively (Table 3). Therefore, this APS YSZ TBC has a porosity level in the range of 10-20%, which is accepted for a porous TBC.
Table 3 As-sprayed bond strength values of TBCs manufactured in this study and complementary data As a final remark, it needs to be stressed that APS YSZ TBCs exhibiting thickness around 1000 µm (just like the one of this work) or even thicker are nothing new. They have been successfully applied in the industry, mainly for industrial gas turbines (IGTs) for many years. What this work wants to bring attention is the fact that these conventional porous TBCs can be manufactured not only at high powder feed rates, but also at high DE levels.
APS YSZ TBC Architectures: As-Sprayed Bond Strength
Bond strength testing is a very important complement feature to the microstructural characterization described in the previous session. As discussed, both YSZ top coats exhibited conventional porous microstructures and no major defects (Fig. 3). They were sprayed at 100 g/min and the DE value is 70%. These are coveted very high values. However, they may cause an over increase of residual stress levels, which could lead to low TBC adhesion values (mainly for the thicker ~930 µm-YSZ TBC) and perhaps a “weak” performance during thermal cycling.
As an example, Vaßen et al. (Ref 23) produced YSZ TBCs via APS under different YSZ powder feed rates: 4.4, 9.1, 36.4 and 54.7 g/min. All TBCs were porous (12-15%) and subjected to burner-rig testing under similar thermal gradient conditions. There was an increase of ~20% in the number of cycles to failure for the TBCs produced from 4.4 to 9.1 g/min; which was the best performing. On the other hand, from 9.1 to 54.7 g/min, there was a reduction of ~86% in the number of cycles to failure; which is considerably negative. Therefore, one may claim that elevated powder feed rates over 50 g/min are inherently negative. However, the YSZ deposited at 9.1 g/min (best performing) exhibited a DR of ~7 µm/pass; whereas that of the 54.7 g/min YSZ (worst performing) was ~31 µm/pass; i.e., 4.4X higher. Consequently, it can be hypothesized that the enhanced powder feed rate was not necessarily the main responsible for the outcome, but rather the thicker DR (i.e., increased residual stress effects). For this reason, when spraying YSZ at powder feed rates of ≥50 g/min, it is important to adjust the relative speed of the substrate surface and torch in order to avoid maximizing DR levels.
Thus, although the TBCs manufactured in this work do not exhibit unwanted defects (Fig. 3), to double-check if they were not “weakened” by residual stress effects, measuring their bond strength values is key to validate the microstructural results. The as-sprayed bond strength (ASTM C633) results and complementary data of both TBC manufactured for this study are found in Table 3. Briefly, the bond strength for the ~420 µm-thick TBC is ~13.0 MPa, whereas that of the ~930 µm-thick TBC is ~11.6 MPa. This is considered as an interesting result. Despite the fact that one YSZ TBC is twice thicker than the other, the drop of bond strength intensity for the thicker ~930 µm YSZ TBC is just 11%.
To better understand this small drop of TBC bonding, it is important to look at the data of Table 1. Initially, one needs to remember that both YSZs were sprayed over the same APS BC. As previously described, the maximum coating surface temperature during spraying grew in time with each single torch pass; however, it stabilized at ~600 °C for both YSZ top coats after approximately 8-9 torch passes. In addition, the DR values were the nearly identical (~23 µm/pass) for both YSZ TBCs. The only major difference during spraying was the additional the number of torch passes required to spray the thicker YSZ TBC; i.e., 18 (~420 µm-thick YSZ) versus 40 torch (~930 µm-thick YSZ) passes.
To complement the discussion, the snap-shots of the fracture surfaces of tested samples are shown in Fig. 4 and also help to understand this issue. First of all, it is possible to notice that for both TBCs the BC remained adhered to the substrate. In addition, all TBC samples exhibited mixed bond strength failure, i.e., a combination of adhesive (YSZ/BC interface) and cohesive (within YSZ) failures. However, for the ~420 µm-thick TBC, the failure was more adhesive; i.e., concentrated at the YSZ/BC interface. This conclusion is obtained by seeing the fracture surfaces on the substrate side (Fig 4a). The area amount of BC exposed (grey zone—YSZ/BC adhesive failure) is larger than that of the YSZ (white zone – YSZ cohesive failure). For the ~930 µm-thick TBC, the amount of white area (YSZ cohesive failure) on the substrate side (Fig. 4b) becomes more evident, but the overall mixed failure characteristic is still similar to that of Fig. 4(a). For this reason, the small variation of macrostructural features of the fracture surfaces between both TBCs also help to understand the small drop of bond strength of the thicker coating.
The elastic modulus values of both YSZ TBCs measured via instrumented indentation can also shine a light on this subject. As described in "Thermal gradient laser-rig testing" section, for each YSZ TBC, a total of 3 series zones were tested: (i) next to BC (~50 µm from BC), (ii) mid-coating thickness and (iii) next to outer surface (~50 µm from epoxy resin). A total of 25 indentations were taken for each single zone; i.e., each YSZ TBC was indented 75 times. These elastic modulus measurements can be found in Fig. 5. Essentially, the average and standard deviation values for both coatings, measured at the same equivalent zones, do not show a significant variation. The analysis of variance (ANOVA) confirmed that for each individual YSZ TBC, at the 0.01 level (i.e., 99%), the population means are not significantly different (for each indentation zone). In simple words, it is not observed a noticeable gradient or variation of elastic modulus values for both coatings and they are relatively uniform across the coating thickness, not mattering if the YSZ TBC thickness is ~420 or ~930 µm. As previously stated, the DR values were the nearly identical (~23 µm/pass) for both YSZ TBCs. In addition, the maximum coating surface temperature during deposition increased in time with each single torch pass; however, it never went above ~600 °C for both YSZ top coats during spraying after 8-9 passes. Consequently, temperature distribution of the samples plateaued and stabilized during processing. Most likely these elements contributed to the uniformity elastic modulus results shown in Fig. 5; which in turn also help to explain (i) the low difference of bond strength between both TBCs (Table 3), (ii) their same failure mode (iii) and similar fracture surfaces (Fig. 4).
It is important to stress that Shinde et al. (Ref 24) also observed a similar behavior. It was reported the progression of the elastic modulus values (in-situ beam curvature method) along the thickness for a conventional porous APS YSZ TBC. Just like what was observed in this work, the maximum surface temperature of the TBC started to increase with the number of torch passes, but it stabilized at ~550 °C after the initial 6-7 passes (of a total of 20+ passes). Within a range from approximately 100 to 700 µm, the conventional porous YSZ exhibited nearly constant elastic modulus values over all thicknesses (~50-60 GPa).
Another important point to be considered is the comparison of the bond strength values reported in this manuscript with those reported in the literature for porous and dense-vertically-cracked (DVC) APS YSZ TBCs. In simple words, it is necessary to know if the bond strength results reported in this manuscript are “good”. Table 4 displays literature values for as-sprayed porous and DVC YSZ TBCs, based on the results of Guerreiro et al. (Ref 5), Smith et al. (Ref 25), Curry et al. (Ref 26), Lee et al. (Ref 27) and Myoung et al. (Ref 28). It needs to be stressed that just like in this work, these bond strength values were also determined via the ASTM C633 method. Overall, APS YSZ TBCs are exhibiting bond strength levels ranging from ~3 to ~15 MPa within a wide variation of thicknesses (~400-2000 µm) and porosity (~3-23%) values. However, when the TBCs thicker than ~1000 µm are not considered (independently of porosity content); the bond strength values are found within the ~7-15 MPa range (Table 4). For this reason, it can be stated that the bond strength levels of the TBCs manufactured this study (i.e.; 11.6-13 MPa) are within the upper range of the values of those reported in the literature; which is a “good” result. As an industrial application curiosity, Smith et al. (Ref 25) reported that the lower bond strength acceptance limit for an industrial APS YSZ TBC exhibiting 10-15% porosity would be 7.6 MPa. Therefore, both coatings sprayed in this study would far exceed the minimum bond strength requirements for a TBC in this case.
Table 4 As-sprayed bond strength values (ASTM C633) and complementary data for different APS YSZ TBCs reported in the literature As this manuscript deals with YSZ spraying at high powder feed rates using a high-enthalpy plasma torch, it is interesting to further mention the works of Curry et al. (Ref 26), Lee et al. (Ref 27) and Myoung et al. (Ref 28); which also dealt with the same subject. Curry et al. (Ref 26) sprayed two YSZ powders at 280 g/min using the high-enthalpy 100HE APS torch (Progressive Surface, Grand Rapids, MI, USA). YSZ DE values ranging from ~40-52% were reported, and the porosity, thickness and bond strength values of these TBCs are found in Table 4. From these sets of data one can realize that YSZ powders can be sprayed at 280 g/min and respective TBCs still exhibiting porosity levels acceptable for IGTs (i.e., 20-30%) and bond strength values within the ~7-15 MPa range can be produced. Nonetheless, this elevated powder feed rate of 280 g/min most likely limited the YSZ DE levels to 52%. The YSZ TBCs of Lee et al. (Ref 27) listed in Table 4 were sprayed at 100 g/min using the legacy torch 9MB (Ar/H2 plasma) and the cascade torch TriplexPro (Ar/He plasma), both from the same company (Oerlikon Metco, Westbury, NY, USA). Unfortunately the YSZ DE values were not reported by the authors. The YSZ TBCs of Myoung et al. (Ref 28), also shown in Table 4, were also sprayed at 100 g/min using the TriplexPro (Ar/He plasma). Here, again the YSZ DE values were not reported.
APS YSZ TBC Architectures: XRD Phase Composition
Figure 6 shows the XRD diffraction patterns of the YSZ powder feedstock and both as-sprayed YSZ TBCs. To analyze the XRD spectra of YSZ, it is generally easier to split the patterns into two zones, i.e., 2θ from ~27.5 to 32.5° (Fig. 6a) and from ~72 to 76° (Fig. 6b). By looking at Fig. 6(a) one wants to know if the presence of the unwanted monoclinic (m) phase of YSZ is observed in the powder and as-sprayed TBCs. The m-YSZ phase is unwanted to due its known stress-induced martensitic transformation. Based on the YSZ phase diagram depicted by Brandon and Taylor (Ref 10), this phase transformation occurs approximately at 100 °C, i.e., very close to RT. Therefore when the TBC is heated from RT to turbine operation temperatures, this m-YSZ phase (if present) transforms into the tetragonal (t) YSZ phase, which accompanied by a volume shrinkage of 3-5%. On the contrary, when the TBC cools down to RT, the t-YSZ phase transforms again into the m-YSZ, which then is followed by a volume expansion of 3-5%. These phase transformations add an additional undesired fatigue stress mode into the TBC system, which helps to lead to early coating spallation. For example, Witz and Bossmann (Ref 29) analyzed the XRD of a delaminated YSZ TBC sample taken from an engine where accelerated TBC degradation was observed. The TBC is composed mostly of m-YSZ (>70 wt.%). For this reason, the presence of m-YSZ phase is extremely undesirable in the TBC phase composition.
The m-YSZ, if present, can be identified by its 100% and 70% highest intensity powder diffraction file (PDF) peaks (PDF #37-1484); which are found at 2θ values of ~28.2° and ~31.5° for the CuKα radiation of the XRD. By looking at the XRD pattern of Fig. 6(a), it is possible to notice a minor presence of the m-YSZ for the feedstock powder and the total absence of it for the two YSZ TBCs. The minor presence of the m-YSZ in the powder is probably related to an incomplete mixture (i.e., alloying) of ZrO2 and Y2O3 during the synthesis of the feedstock. The welcome absence of the m-YSZ phase in both as-sprayed YSZ TBCs is likely to be related to (i) the high degree of melting of the powder during spraying, which improved its alloying and (ii) the fast cooling rates of the molten YSZ splats during re-solidification upon spraying; as described by of Brandon and Taylor (Ref 10).
The other phase to be identified is the so-called non-transformable tetragonal prime (t’) phase of YSZ (PDF #48-0224). The t’-YSZ phase, although metastable, it displays a sluggish phase transformation into the regular tetragonal (t) YSZ (PDF #82-1245) and cubic (c) YSZ (PDF #30-1468) up to temperatures of approximately 1300 °C; as described by Brandon and Taylor (Ref 10). Unlike the t-YSZ phase, the t’-YSZ phase does not undergo the unwanted martensitic m-YSZ phase transformation described above. For this reason, the t’-YSZ phase is very desirable for a TBC. To properly identify the t’-phase is necessary to distinguish it from the regular tetragonal t-phase of YSZ. As both exhibit their 100% highest intensity diffraction peaks very close to each other, as per Fig. 6(a); it is necessary to look into a second zone of the XRD pattern, which is featured in Fig. 6(b). Essentially, the XRD peaks of the t’-phase and t-phase of YSZ will overlap along the pattern from 20° to 80° for the CuKα XRD radiation, except at the 2θ zone around 72-76°. The t’-YSZ phase will exhibit the diffraction peaks 004 and 220 at ~73.2° and ~74.2o (respectively); whereas the t-YSZ exhibits the diffraction peak 004/220 at ~73.7°. By looking at Fig. 6(b), it is realized that the YSZ powder has the t’-YSZ as the primary phase (with minor m-YSZ phase—Fig. 6a) and both YSZ TBCs only display the well-desired t’-YSZ phase in their compositions. More information about the importance of these YSZ phases, their formations and influence on TBC performance can be found in the work of Brandon and Taylor (Ref 10).
APS YSZ TBC Architectures: Thermal Gradients
As previously stated, TBCs reduce the temperature of the Ni-based metallic superalloy metallic components located in the hot section of gas turbine engines; which typically exhibit wall thicknesses of ~1–2 mm and melting points within the 1300-1400 °C range. According to Rolls-Royce (Ref 30), the maximum temperature that the hot gases inside combustion chamber can reach is 2000 °C (at its flame core). This gas temperature is subsequently and progressively cooled by cold air that is bled from the turbine compressor and injected along the combustor length (via dilution holes) to reach its specified turbine inlet temperature (TIT). The TIT is the temperature of the combustion gases that leave the combustion chamber and enter the turbine unit.
It needs to be stressed that a turbine engine does not always operate at its peak temperatures. For example, according to the report Commercial Aircraft Propulsion and Energy Systems Research: Reducing Global Carbon Emissions (Ref 31), aviation turbine engines operate at different power levels during each of the distinct segment of a flight. For example, the turbines of commercial jet liners at the take-off will operate at 100% of their peak power (which will last for no more than 5 min), whereas, during longest stage of the flight (i.e., cruise), they will operate at 30% of the maximum power. Regarding fighter jets, Mom and Hersback (Ref 32) exemplified that depending on the mission (e.g., from general flying to “dog-fight” mode), maximum military turbine power can be applied from few to several times during a flight mission. According to Farokhi (Ref 33), the TIT levels of modern jet engines during take-off (i.e.; max power) are reaching the range of 1500-1600 °C; whereas these TIT levels during the cruise stage are found within the 1140-1240 °C range. Regarding IGTs, Nelson and Orenstein (Ref 34) explained that they will operate at lower maximum peak temperatures than those of aviation turbines, but at much longer and continuous times (>20,000 h). As a complement note, YSZ TBCs are not supposed to temperatures higher than ~1300 °C, as discussed in "Understanding Thermal Gradients within TBCs and Metallic Components" section , even if TIT levels reach 1500-1600 °C at take-off engine power. A thin air film cooling (bled from the compressor) will flow adjacent to the TBC surface from specifically engineered cooling holes in the components, thereby providing a “cooler shielding layer” over the TBC surface (in addition to the uncoated component backside), as described in details by Rolls-Royce (Ref 30) and Farokhi (Ref 33). Finally, Witz and Bossmann (Ref 29) analyzed ex-serviced real TBC-coated components of turbine engines. Their objective of the study was to determine the real surface temperature that TBCs were subjected during engine operation (via thermal paint and XRD analysis); in order to create mappings of the TBC thermal load and a lifetime evaluation model. One of the parts analyzed was a ~20 cm x ~20 cm TBC-coated heat-tile originating from the combustion chamber. Via a thermal paint evaluation along the length of the tile, it was observed that during the life of engine operation, the TBC on this single component had experienced a variation of maximum temperature of up 150 °C across its ~20 cm extent, with the max peak located around center of the tile. Thus, based on the examples above cited, one can realize that the knowledge of the temperature drop (i.e., reduction) provided by a TBC on a coated component, over a wide range of (i) component locations, (ii) TBC thicknesses and (iii) surface temperatures, is essential for modelling, simulation and lifetime prediction. And these sets of data are scarce in the open literature.
To help addressing this need, the data profiles of the two 14% porous APS YSZ TBC samples of this study generated by the laser-rig can be found in Fig. 7 (~420 µm-thick YSZ TBC) and Fig. 8 (~930 µm-thick YSZ TBC). It includes the YSZ TBC surface temperature (T-ysz), the substrate temperature (T-sub), the laser power and the back-side air cooling for the un-coated side of the coupons. Some important characteristics are shared for both testing. After an initial laser-rig stabilization period of ~2 min, the profiles generated were essentially constant for nearly 30 min. As stated, the back cooling air jet on the uncoated side of the coupons was set at ~430 lpm for all samples. Consequently, all these TBC-coated samples are being tested under virtually identical cooling conditions. For this reason, to increase T-ysz from 1100 °C to higher values, the laser power had to be increased. On the same note, the higher the T-ysz, the higher the T-sub (as expected). It is important to remember that the T-sub was measured via a thermocouple inserted the mid-thickness of the substrate, located 1.6 mm beneath the BC/substrate interface. Therefore, the T-sub values measured in this experiment are considered to be representative of those of a back-side component (e.g., combustion chamber), if its total wall thickness was 1.6 mm.
As all ΔTs for all TBCs in the individual testing conditions depicted in Fig. 7(a)-(e) and 8(a)-(g) are consistent and steady; i.e., parallel to each other, it can be said that any significant YSZ sintering/densification events that might have occurred, were minor and did not affect the experiment in some significant way. In other words, if a significant YSZ sintering was taking place during testing, the T-ysz and T-sub profiles would become tapered towards each other. On that account, it is fair to say that the data reported in this work (Fig. 7-8) represent the ΔT temperature reduction values that occur across TBC/substrate systems for an as-sprayed 14% porous APS YSZ TBC, manufactured at two thickness levels and subjected to distinct T-ysz values but the same cooling.
As explained in the "YSZ Elastic Modulus via Instrumented Indentation Testing" section of this manuscript, the T-ysz was increased from 1100 °C in 100 °C steps (by augmenting laser power) until the temperature of the substrate (T-sub) reached around 1000 °C; which is typically considered to be the highest limit of temperature operation for static gas turbine components (e.g., combustor). Figure 9 and 10 highlight in more detail the data displayed in Fig. 7 and 8.
For example, Fig. 9 depicts the resulting T-sub values as a function of the pre-selected T-ysz levels for both TBCs. For the ~420 µm-thick YSZ TBC, when T-ysz is ~1300 °C (typically considered at the YSZ upper temperature limit), the resulting T-sub is ~875 °C; which is well below the generally accepted T-sub of 1000 °C. T-sub will only reach 1000 °C, when T-ysz approaches 1500 °C. On the other hand, for the ~930 µm-thick YSZ TBC, this 1000 °C T-sub max limit could not be reached. The reason was, the highest calibrated temperature of the laser-rig pyrometer was 1684 °C. Consequently, the laser power was elevated and stopped at 511 W, when T-ysz had reached ~1680 °C; which corresponded to a T-sub of ~880 °C (Fig. 8g).
The specific temperature drop levels (ΔTs) across the TBC/substrates thickness profile are featured in Fig. 10. Once again, when T-ysz is ~1300 °C for ~420 µm-thick YSZ TBC, the resulting temperature reduction to the substrate is ~425 °C. For ~930 µm-thick YSZ TBC, at the very same T-ysz of ~1300°C, the resulting temperature reduction to the substrate is ~600 °C. These results point up the higher thermal resistance displayed by the thicker coating, as discussed by Helminiak et al. (Ref 15). And although YSZ TBCs were not developed to work at a T-ysz of 1680 °C, for sure one cannot fail to notice the “remarkable” temperature reduction of 800 °C to the substrate yielded by the thicker TBC (Fig. 10).
Finally, based on the references discussed above, it is important to remember again that gas turbine engines operate at different temperature regimes and the thermal load over TBC-coated component will vary depending on the part location. The TBC temperature mappings, such as that proposed by Witz and Bossmann (Ref 29), are very useful to pin-point temperature profiles and hot-spots. For these reasons, a better understanding of the ΔTs that take place along the TBC/component profile is of high importance for improving the manufacturing efficiency and lifetime of TBCs. For example, the thicknesses of the TBCs can be bespoken based on the previously-mapped thermal loads across the component. Most likely this approach could be “more easily” implemented in float wall type of combustion chambers. In this configuration, the walls of a combustor are lined by a multi-assembly of individual TBC-coated heat tiles that protect the body of the combustor liner from damage due to exposure with the hot gases.
YSZ TBC Manufacturing Cost Estimation
The high powder feed rate and DE values obtained for the YSZ TBC manufacturing reported in this manuscript can be considered very promising. However, one may claim that in real production environment, the elevated plasma gas flow levels and electrical energy consumption required by a high-enthalpy torch like the Axial III Plus would overshadow its gains when compared to the performance of a legacy torch.
To address this issue, the YSZ TBC manufacturing cost estimation was based on the data gathered during this study and also on results obtained in previous works (Ref 4, 5). The estimation was based on the calculation performed to spray a 410-430 µm thick YSZ TBC on a 500 mm long cylinder component, using spray parameters developed in this research for the Axial III Plus torch, as well as, that of a legacy APS Ar/H2 torch; which was sprayed a YSZ at powder feed rate of 60 g/min, displayed a porosity level of ~13% and exhibited a DE value of 43% (Ref 5). This component part could be considered as being similar to the inner wall of an annular turbine combustion chamber. The component tangential speed, the torch transverse speed, powder feed rate, DR and DE values used to estimate the cost and time to deposit the TBC were the ones used and obtained from these studies. The cost of the consumables and energy were: 30 US$/m3 for Ar, 22 US$/m3 for N2, 38 US$/m3 for H2, 34 US$/kg for YSZ feedstock and 0.25 kWh for electricity. Other costs involved (e.g., labor) were not included in the estimation.
Briefly, by changing from an Ar/H2 legacy plasma torch to a N2-based high-enthalpy one, there is a reduction of (i) 66% in processing time, (ii) 44% in YSZ feedstock consumption, (iii) 21% in plasma gas cost, (iv) 44% in electricity cost, and (v) 60% in total manufacturing cost. As previously stated, reducing environmental footprint means reducing at the same time the amount of natural resources and energy used to produce a product. Consequently, these sets of data exemplify the possible gains of manufacturing efficiency by using N2-based high-enthalpy APS torches for YSZ TBC production.
Final Comments
Regarding manufacturing efficiency, although the results obtained in this work are promising, there are still important steps to be investigated in order to confirm the applicability of these types of high feed rate and high DE APS YSZ TBCs in industrial applications. Among these steps are (i) YSZ thermal conductivity measurements, (ii) erosion testing and (iii) TBC thermal cycling performance. Regarding thermal conductivity measurements, the as-sprayed YSZ value most likely will fall within those reported in the literature for 10-15% porous YSZ TBCs; i.e., 0.8-1.0 W/mK (Ref 1, 4, 5). Thus, no unexpected surprises should occur here. The same hypothesis can be stated for erosion testing; considering the fact that the TBC manufactured in this study exhibits a conventional porous microstructure. But for sure an investigation on the thermal cycling performance (including a comparison with a TBC benchmark) is paramount.
With respect to the thermal gradient study, it is recognized that TBCs will sinter in time in these thermal gradient conditions, as described by Zhu and Miller (Ref 35) when using a laser-rig to create thermal gradients on TBCs. Since thermal gradient tests, such as burner and laser-rig are time consuming (and thereby costly), in order to provide sufficient statistics of samples to develop TBC aging models, most likely the combination of thermal gradient testing and the Larson-Miller parameter will be necessary. Larson–Miller parameter describes the creep-like behavior of thermal conductivity increase with temperature and time for different materials. For TBC R&D, this concept has been introduced by Zhu and Miller (Ref 35) and further explored by Tan et al. (Ref 36).
Moreover, it needs to be stressed that this work is not claiming (at all) that APS YSZ TBCs can be safety operated at temperatures higher than 1300 °C for long hours in gas turbine engines. The reasons for this 1300 °C YSZ limitation are well-known, stated and discussed in "Understanding Thermal Gradients within TBCs
189 and Metallic Components" section. The objective here is to provide engineering data for a better comprehension on how hot-section components of gas turbine are protected by TBCs. Consequently, a wide range of T-ysz values were investigated, including the ones above the 1300 °C limit. It is another complement to previous study (Ref 14), where (i) the same laser-rig equipment was employed to generate thermal gradients on porous APS YSZ TBCs at a T-ysz magnitude of 1300 °C and (ii) other TBC thermal gradient results reported in the literature were summarized.
Finally, by looking at the results of the thicker YSZ TBC (Fig. 8, 9, 10), one may wonder or even speculate about a possible thermal insulation protection that a thick (e.g., ~900-1000 µm) and porous (e.g., 10-20%) multi-layer TBC assembly could offer to these metallic components. A TBC assembly, which one of the layers could even be YSZ, but with the addition to other newly-developed over-layers. New high-performance ceramics engineered to display at high temperatures a combination some key characteristics; including (i) low thermal conductivity, (ii) phase stability, (iii) CMAS-attack resistance, (iv) erosion resistance and (iv) chemical endurance against water vapor attack. If achievable, this type of multi-layer TBC assembly could potentially operate at temperatures well-over the established 1300 °C limit for YSZ and still keep the current Ni-based superalloy components at temperatures no higher than 1000 °C. And if attainable, this type of multi-layer TBC assembly and Ni-based superalloy architecture may possibly be an alternative to environmental barrier coatings (EBCs) and ceramic matrix composites (CMCs).