2020 Volume 61 Issue 8 Pages 1510-1516
Thermal barrier coatings (TBCs) are necessary to protect nickel-based alloy blades of gas turbine against oxidation and thermal fatigue in high-temperature operating conditions. Ceramic materials, which are very good natural thermal insulator, attract the most interests from engineers and scientists. However, the brittleness of ceramics is a major obstacle for utilizing them as the TBCs, which are also required very good damage tolerance against physical impacts. The cracks appear on the blade surfaces during its operation can lead to the severe failure. In this research, a composite of Y2Ti2O7 and Ni was developed as a self-crack healing material to overcome this problem. The crack-healing behavior is investigated by using Vickers indenter to create cracks on the composite surface intentionally, followed by annealing in an oxidizing environment. It is found that the main crack-healing mechanism is the filling of NiO, which was formed from the oxidation of the Ni fillers, into the cracks. Complete heal of cracks is achieved with 10 vol% Ni filler, which is confirmed by X-ray diffraction and scanning electron microscopy. Thermal conductivity and Weibull distribution for the strength of the composite were also investigated to find the appropriate volume fraction of Ni nanoparticles in this self-healing material.
Fig. 8 SEM micrograph shows a crack healed by filling and bridging.
In current aircraft gas turbine, heat-resistant alloys are widely used for turbine blades, in which nickel-based superalloy is the most popular material. This material allowed a significant increase in operating temperature which enable better performance and fuel efficiency, but the improvement has almost attained its limit due to the low metlting temperature of the alloy.1) To overcome this problem, structural ceramics with low thermal conductivity have been developed as thermal barrier coatings (TBCs) protect the underneath blades against hot corrosion and thermal fatigue in high-temperature operating conditions.2) The limit temperature of structural ceramics, ranging from 2098 to 2973 K,3) is much higher than that of super alloys. Yttria-stabilized zirconia (YSZ) has been considered a suitable material for TBC due to its low thermal conductivity.2,4) However, the intrinsic brittleness of YSZ and other structural ceramics is a problem must be solved to apply these materials. The typical fracture toughness (KIc) of ceramic is 1.5–10 MPam1/2, which is considerably lower than that of metal.5) Machining process, foreign particle impact, or thermal stress in the hot section can leave minicracks or small defects on the surface of a ceramic component, as a result considerably decreases the reliability and lifetime of that component. The defects generated by machining can be prevented by grinding and polishing, but the ones generated under operation are unavoidable. There have been many research conducted to overcome this problem, focusing on the followings: (i) improve the fracture toughness of the materials by reinforcing them with nanofiller;6) (ii) develops self-crack healing material that can repair the surface defects autonomously under operation.7–10) The second approach has attracted many interest for some advantages: it helps to save time and cost of machining process; it can improve the reliability of ceramic component; in many cases, the self-healing not only heal the crack but also enhance the strength of the ceramic.11–13)
Yttrium titanate, Y2Ti2O7 is one of important14) and the very first-studied15) members of the A2B2O7 family (in which A-site and B-site are occupied by 3+ and 4+ cations, respectively). As a member of this family, it has pyrochlore structure, which is similar to the fluorite structure.16) Under irradiation conditions, the structure of pyrochlore can be disordered to that of fluorite, inducing an enhanced radiation resistance.17) This is the reason Y2Ti2O7 was selected as one of main precipitates in the oxide-dispersed-strengthened steel used for nuclear reactor.18,19) The observation of the mesoscopic interfaces between Y2Ti2O7 crystal wafer substrate and Fe grains grown on it at 800°C could not find any significant transition layer at the interface,20) suggesting that the Y2Ti2O7 is very difficult to dissolve in and react with molten metal, in other words, this oxide has a good stability against liquid metal. Some studies have reported remarkable corrosion resistance of Y2Ti2O7 bulk21) and coat22) against molten aluminum.
In fact, Y2Ti2O7 is a promising candidate material for the outer layer of the TBC due to its low thermal conductivity.23) Furthermore, the coefficient of thermal expansion (CTE) of Y2Ti2O7 (8.36–8.39 × 10−6 K−1)24) is not much lower than that of nickel alloy (16 × 10−6 K−1),3) the major material for current gas turbine blade. This means the thermal expansion mismatch between the coating material (Y2Ti2O7) and the substrate can be diminished and hence helps to avoid failures such as vertical cracking or delamination. In addition, Y2Ti2O7 demonstrated high-temperature corrosion resistance against molten metal, not only aluminum as mentioned above but also nickel (Ni) particles.23) On the other hand, the dispersion of Ni nanoparticles in ceramic matrix, such as Al2O3, has been researched as a method to fabricate self-healing materials.25–27) Oxidation of Ni particles to form oxide filling in the cracks helps to heal the damaged surface. Therefore, Ni dispersed-Y2Ti2O7 composite (YTO-Ni), a promising candidate for TBCs due to its thermal stability and improved fracture toughness,23) can also have self-crack healing ability. However, to the best knowledge of us, there have been no research studying the healing behavior of this composite.
In this paper, the crack-healing behavior of YTO-Ni composites (with 5–20 vol% Ni) was examinated to consider its applicability for TBCs. The sintered disks of the composite was first fabricated by hot pressing method, and then were cut into specimens to test the mechanical properties and the healing ability. The crack healing behavior was investigated at 1150°C which is the typical inlet temperature of gas turbine engine blades protected with TBCs.
Figure 1 illustrates the crack-healing process in this composite. Nanosized Ni particles were dispersed homogeneously in the Y2Ti2O7 matrix (Fig. 1(a)). In order to heal the surface cracks (Fig. 1(b)), it is necessary that a positive volume variation, i.e. a volume expansion, happens upon the oxidation of the healing agent.28) The dispersed particles on the composite surface is expected to react with the oxygen in the annealing chamber (Fig. 1(b)) to form an oxide (grey color), which subsequently helps to seal the cracks on the surface (Fig. 1(c)). Here, the oxide formed from the heating process of healing particles should have volume larger than that of the original particles, to fill in the crack gap.
Schematic illustration for surface crack healing in Y2Ti2O7–Ni composite: (a) Before cracking; (b) After cracking; (c) After healing.
The following oxidation occurred in this composite during the heat-treatment:
| \begin{equation} \text{Ni} + \text{O$_{2}$} \to \text{NiO}. \end{equation} | (1) |
In this reaction, the molar ratio of Ni and NiO is 1:1. Therefore, the volume variation ΔV between the formed oxide (NiO) and the initial healing agent (Ni) can be defined as:
| \begin{equation} \Delta V = \frac{V_{ox}}{V_{ha}} = \frac{M_{ox} \times D_{ha}}{M_{ha} \times D_{ox}}. \end{equation} | (2) |
Here, Mox and Mha are the molar mass of NiO (6.67 g/cm3) and Ni (8.91 g/cm3), Dox and Dha are the density of NiO (74.7 g/mol) and Ni (58.7 g/mol), respectively. The volume variation ΔV, thus, was found as 1.7. This positive value of ΔV envisages a crack-filling behavior during annealing of this composite.
2.2 ExperimentsThe fabrication of Y2Ti2O7/Ni composite was described in a previous work.23) At first, Y2O3 (Shin-Etsu Chemical) and anatase TiO2 (Sigma Aldrich) was mixed together in stoichiometric ratio (Y2O3: TiO2 molar ratio = 1:2) and ball-milled in 99.5% ethanol media for 24 h, followed by dried in an oven at 80°C overnight and calcined in air furnace at 1200°C for 5 h to obtain Y2Ti2O7 fine powder. Then, nickel nitrate Ni(NO3)2·6H2O (98%, FUJIFILM Wako Pure Chemicals) was mixed with the synthesized Y2Ti2O7 powder and the mixture was then ball-milled, dried, grinded with a mortar and pestle before being heated in air furnace at 400°C for 2 h. During heat treatment, Ni(NO3)2·6H2O was completely decomposed to nickel oxide (NiO). The Y2Ti2O7/NiO mixture powder was then placed in an alumina boat and subjected to another heat treatment at 800°C for 2 h in a 97% argon (Ar) + 3% hydrogen (H2) environment (the flowrate of the gas mixture was 100 mL/min). The result of this treatment is a complete reduction of NiO into nickel (Ni). The Y2Ti2O7/Ni mixture was then dry ball-milled overnight to obtain very fine powder for the hot-pressing process.
Finally, the synthesized Y2Ti2O7/Ni powder mixture was hot-pressed in an Ar atmosphere at 25 MPa and 1340°C for 4 h by a sintering furnace (Fuji Dempa Kogyo, FVPHP-R-5, FRET-18) to obtain sintered body (44 mm in diameter). Both the heating and cooling rates were set at 10°C/min. The volume fraction of dispersed Ni particles in the sintered composites was 5, 10, 15, and 20 vol%, respectively. All composites have bulk density greater than or equal to 93.3%, which were confirmed by the Archimedes buoyancy method. The thermal conductivity of the composites was measured by a laser flash system (LFA 457 Microflash, Netzsch). For the measurement, samples with 10 mm in diameter and 1 mm in thickness were prepared from the sintered composites. The investigated temperatures ranged from 100 to 1000°C with 100°C intervals under Ar atmosphere at a flow rate of 120 mL/min.
To investigate the self-crack healing behaviors, pre-cracks was introduced on the mirror-polished surface of the composite by a Vickers indenter with 9.8 N and 49 N indentation loads. The dwelling time of the indenter was set at 15 s. The as-indented specimens were annealed in air using an electric furnace (FO100, Yamato Scientific) to activate the crack-healing ability of the composites. The heat treatment was conducted at 1150°C, which is considered the average temperature capability for a TBC, for 1 h. The composite’s surface and induced cracks were observed using a fluorescence microscope (BX51N, Olympus) or a field-emission scanning electron microscope (JSM-7001FA, JEOL). Energy dispersive X-ray spectroscopy (EDS) analysis was conducted to investigate the composition of phases presented on the surface by a spectrometer attaching on this scanning electron microscope (SEM). The crystalline phases on the composites’ surface, before and after heat-treatment, were identified using X-ray diffraction (XRD) analysis (RINT 2500PC, Rigaku) with Cu-Kα radiation (λ = 1.54186 Å, 40 kV, 40 mA) to investigate the oxidation.
Because nickel has a relative high thermal conductivity for a transmission metal,29) the incorporation of nickel particles into Y2Ti2O7 matrix can raise the conductivity of the composite. Thermal conductivities of the composites were measured and compared with that of 7-YSZ.4) As shown in Fig. 2, the conductivity of each composite increase with the increasing amount of Ni. In addition, the variation in conductivity according to temperature of Ni also influenced the composites. The conductivity of the metal decreases sharply in the range of 150–350°C, then increases gently in the range of 450–1050°C. When the volume fraction of Ni is small (≤10 vol%), the influence is difficult to be observed but at larger fraction of Ni (15 and 20 vol%), it become more obvious. In the range of 200–700°C (the small square in Fig. 2), the average thermal conductivities of Y2Ti2O7 and 7-YSZ are 2.39 and 2.44 W/mK, respectively. Meanwhile, the average conductivity of YTO-Ni 20 vol% composite in the same temperature range is 4.67 W/mK, nearly double that of the two ceramics. This conductivity is considered too high for a thermal barrier coating material, therefore this composite is excluded from further investigations.
Figure 3(a) shows the microstructure of the surface of YTO-Ni 15 vol% composite observed by SEM. There are only two characteristics domains which are the bright particles and the dark grains. Composition analysis conducted by EDS indicated that the bright ones (spot 1) are Ni and the dark ones (spot 2) are Y2Ti2O7, as the majority of atomic number percentage (at%) for these two spots come from Ni and (Y, Ti), respectively. The EDS spectrum for the two spots are shown in the Fig. 3(b)–(c). The count for X-rays emitted from Y-element is higher than that of Ti-element in spite of their 1:1 molar ratio (in Y2Ti2O7 molecular), because the detection limit tends to be higher for the heavier element.
SEM micrograph (a) and EDS spectra (b)–(c) for the surface of YTO-Ni 15 vol% composite. The spectrum in (b) and (c) corresponds to the analyzed spot 1 and spot 2 in (a), respectively.
The crystallinity of Y2Ti2O7 grains can be confirmed by XRD analysis. Figure 4 shows the XRD patterns of various YTO-Ni composite, before and after the healing (heat-treatment) at 1150°C for 1 h. The main phase in the fabricated composites is confirmed as Y2Ti2O7, while the intensity of Ni peaks increases with the increasing volume fraction of Ni (patterns a–c). In addition, the oxidation of Ni nanoparticles into NiO can be confirmed by XRD analysis on the surface of specimens. It is clear that after the heat treatment (pattern (d)), the Ni peaks appearing in patterns (a–c) almost disappeared. The most perceivable one is the strongest peak of Ni at 2θ = 44.51°. On the other hand, the new peaks of NiO at 2θ = 37.06°, 43.1°, 62.62°, and 79.19° can be identified in this pattern. The peak at 75.09° is overlapped with the peak at 75.25° of Y2Ti2O7 and hence make this peak look stronger than in the unhealed specimen (pattern b). These results strengthen the aforementioned theory of oxidation, which is the origin for the subsequent crack-healing behavior described in the latter part of this section.
XRD patterns for the composites, before the healing: (a) YTO-Ni 5 vol%; (b) YTO-Ni 10 vol%; (c) YTO-Ni 15 vol%, and after the healing: (d) YTO-Ni 10 vol%. The ICDD cards of Y2Ti2O7 (#42-0413), Ni (#04-0850), and NiO (#47-1049) were used to identify the crystalline phases.
Figure 5(a) shows optical image of as-sintered and annealed composites. After the heat-treatment, the surface of all annealed specimens changed from black to green, which is the characteristic color of NiO. While the surface of YTO-Ni 5 vol% was not damaged, there was a long crack at center of YTO-Ni 10 vol%. Figure 5(b) shows a micrograph of YTO-Ni 10 vol% specimen (Fig. 5(a-C)), confirming that the Vickers indentation (load = 49 N) is the origin for the crack. The self-healing effect, therefore is only evaluated for small cracks induced by 9.8 N load. On the other hand, the composite with 15 vol% Ni was broken after the heat-treatment. The excessive volume expansion of NiO, leading to a high residual thermal stress inside the matrix, is considered the main cause for the damages in these specimens.
(a) Photograph and (b) optical micrograph of specimens: (a-A) YTO-Ni 5 vol% before the heat-treatment; (a-B) YTO-Ni 5 vol%, (a-C) YTO-Ni 10 vol%, (a-D) YTO-Ni 15 vol%, after the heat-treatment; (b) specimen in (a-C) observed with optical micrograph.
Because YTO-Ni 15 vol% specimen was broken after the heat-treatment, only indentation cracks of specimens with 5 and 10 vol% Ni were observed by SEM. Figure 6 shows a Vickers indentation and the induced cracks on the surface of YTO-Ni 5 vol% specimen, before and after the healing process. Though the big cracks was not completely healed, the small cracks (marked by white arrows in Fig. 6(a)) was almost disappeared after the heat-treatment. It should be noticed that in Fig. 6(b) and Fig. 6(d), the grain boundaries of Y2Ti2O7 matrix is also visible after the treatment due to the effect of thermal etching. In the case of YTO-Ni 10 vol% composite, the crack-healing effect is more impressive. As can be seen in Fig. 7, a long crack on the left side of the Vickers indentation, was fully healed after the heat-treatment. The volume expansion can be realized when comparing the size of Ni particles in Fig. 7(c) with those of NiO in Fig. 7(d). During the heat-treatment, the healing particles were oxidized, expanded, and filled in the cracks. Because of a high volume expansion (ΔV = 1.7), the specimens are look as fully covered with a green layer of NiO (see Fig. 5), although the volume fraction of Ni was only 5–10 vol% in the original composites.
SEM micrographs of induced cracks on the surface of YTO-Ni 5 vol% composite, before (a), (c) and after (b), (d) heat-treatment. The micrographs in (c) and (d) show the cracks in (a) and (b) with higher magnification.
SEM micrographs of induced cracks on the surface of YTO-Ni 10 vol% composite, before (a), (c) and after (b), (d) heat-treatment. The micrographs in (c) and (d) show the cracks in (a) and (b) with higher magnification.
The healing process, begins at the crack tips where the gap is smallest, then develops towards the Vickers indentation. In some cases, when the crack gap is too large as shown in Fig. 8, the oxide cannot completely fill in the gap but they bridge the gap (marked by the white arrows). It is expected that longer healing time and higher healing temperature can foster the healing process and fully heal this crack. However, changing a healing parameter like healing time or temperature, etc. should be done carefully otherwise they can make the specimen broken, as in case of 15 vol% Ni particles (see Fig. 5). As mentioned above, the residual stress inside the specimen generated by the difference between the CTEs of matrix and oxide when they cooling down from annealing temperature, or the so-called thermal mismatch stress, is considered as the main cause.28) To avoid this fracture, the intrinsic strength of the composite should be greater than the total thermal stress impact on the composite, which can be calculated as follows:
| \begin{equation} \sigma_{\text{ox}} = \varphi_{\text{ox}} \times E_{\text{ox}}(1 - \nu_{\text{ox}})^{-1} \times (\alpha_{\text{ox}} - \alpha_{\text{m}}) \times \Delta T. \end{equation} | (3) |
SEM micrograph shows a crack healed by filling and bridging.
In the above equation, φox, vox, and Eox are the volume fraction, Poisson’s ratio and Young’s modulus of the oxide, αox and αm are the CTEs of the oxide and the matrix, respectively. It should be noticed that the greatest thermal stress is generated with a maximum volume fraction of the oxide (φox = φox_max). This occurs when all the healing agent particles were fully oxidized, otherwise the oxide volume fraction is smaller and hence the stress would be lessen. The maximum volume fraction φox_max of the oxide can be calculated from the volume expansion of the healing oxide ΔV as follows:
| \begin{equation} \varphi_{\text{ox_max}} = \frac{\Delta V \times \varphi_{\text{ha}}}{1 - \varphi_{\text{ha}} + \Delta V \times \varphi_{\text{ha}}}. \end{equation} | (4) |
Although there have been no reports on the value of tensile strength of Y2Ti2O7, one can deduce it from the flexural strength data. Munro30) has proved that the flexural strength/tensile strength ratio σf/σt for a ceramic is a function of Weibull modulus m of flexural strength (with m = 11 ± 4):
| \begin{equation} \frac{\sigma_{\text{f}}}{\sigma_{\text{t}}} = \left[\frac{4 (m + 1)^{2}}{m + 2}\right]^{1/m}. \end{equation} | (5) |
Our data of Y2Ti2O7 flexural strength31) was replotted as Weibull distribution in Fig. 9. It is found that the flexural strength (scale parameter) σf is 217 MPa and the Weibull modulus m, describing the variation of the strength, is approximate to 7. When substitute these values into eq. (5), the tensile strength of Y2Ti2O7 can be found as σt = 128 MPa.
Weibull distribution for the strength of Y2Ti2O7 sintered ceramic.
By substituting the elastic modulus (E = 95.8 GPa) and Poisson’s ratio (ν = 0.416) values from the literature32) to eq. (3), we calculated the maximum thermal stress generated in each composite when they cooled down from annealing temperature to room temperature. In many cases, the oxide volume fraction φox < φox_max and hence the thermal stress would be smaller than the calculated value. The calculated stress in the case of 5 vol% Ni composite is 91 MPa, much lower than the tensile strength 128 MPa of the Y2Ti2O7 matrix, therefore the specimen was not damaged. In the case of 10 vol% Ni composite, the calculated maximum stress is 176 MPa > 128 MPa, thus some severe crack can appear on the surface of the specimen as shown in Fig. 5(a-C). Heat-treatment at a higher temperature or for a longer heating time would cause this specimen broken. In the case of 15 vol% Ni composite, the thermal stress was too high (256 MPa), as double of the tensile strength and even higher than the flexural strength, thus specimen could not enduranced that stress and was broken. These results suggest that the maximum volume fraction of Ni dispersant in this composite should be 10 vol% or less.
In this research, the ceramic matrix composites of Y2Ti2O7 dispersed with Ni particles were fabricated by hot-pressing method and their self-crack healing ability was demonstrated. Surface cracks induced by a Vickers indenter can be healed after heat treatment at 1150°C in air for 1 h. The self-healing phenomenon in this composite is attributed to the volume expansion of the newly formed NiO leading to their filling in and bridging the crack’s gaps. The healing effect is more impressive when the volume fraction of Ni increased from 5 to 10 vol%. These findings are of importance for the application of Y2Ti2O7/Ni composite to the TBC of gas turbine blades. Our future research will focus on optimizing the healing parameters (time, temperature, volume fraction, etc.) to obtain more effective self-healing process. Repeatable self-healing ability, in which the self-healing behaviors can be demonstrated not only for one time but for multiple cycles of cracking-and-healing, is also planned to develop.
The authors wish to express their gratitude to Environment & Process Design Laboratory and Energy Materials Laboratory in Nagaoka Univesity of Technology for their support in composites fabrication and in thermal conductivity measurement. We also thank Mr. Takahito Eguchi and Mr. Rintaro Ninuma from National Institute of Technology, Kushiro College, for their assistance in SEM observation. This work was supported by Japan Society for Promotion of Science (JSPS) KAKENHI Grant-in-Aid for Early-Career Scientists Number 18K14086.
