Science of high performance multifunctional high temperature coatings

Diffusional mobilities in the fcc and hcp-phases in Ni-Ru have been evaluated using the Calphad method and compiled into a preliminary database. Information on Al diffusion has been added on a preliminary basis. The database can be used to calculate concentration profiles in ternary Al-Ni-Ru diffusion couples. In principle the database can be combined with other databases, e.g. the NIST-data mobility database for Ni-base alloys, to simulate diffusional reactions in commercial alloys. The database is available as database file in DICTRA format.

The t' zirconia phase is desirable in thermal barrier coatings because it combines the attributes of non-transformability, required for thermal cyclic applications, and adequate toughness. Unfortunately, the t' phase is metastable at the temperatures of interest in gas turbine operation and tends to separate into a cubic phase with low toughness, and a depleted tetragonal phase which is transformable to monoclinic upon cooling. This evolution is undesirable because it degrades the mechanical integrity of the coating. Extensive studies on zirconia partially stabilized with various tri-valent cations at a constant concentration of 7.6mol% MO1.5 (M = Sc, Yb, Y, Gd, Sm, Nd, La)revealed a well behaved trend for the high temperature stability of the metastable t' phase, which was optimal for Yb. Both smaller (Sc) as well as larger cautions (Y to La) yielded reduced stability of the t' phase relative to Yb, but for different reasons. At the composition selected, which is the standard stabilizer content for thermal barrier coatings, Sc is stable against decomposition but its tetragonal/monoclinic partitionless transformation temperature (T0[t/m]) is substantially above ambient and renders the 7.6ScO1.5 transformable upon thermal cycling. This is at variance with the claim in the literature that Sc is a better stabilizer than Y, resulting from comparing dissimilar concentrations of stabilizer. The T0[t/m] appears to be minimal for Y and increases with increasing caution size, as does the driving force for phase separation in response to shifts in the position of the tetragonal and cubic phase boundaries. Thermodynamic modelling of the ZrO2-YO1.5-GdO1.5 system has ascertained the increased in driving force with progressive substitution of Gd for Y. The combined effect rapidly decreases the stability with increasing caution size. For comparison, a 7.6%Yb doped zirconia is 4 times as durable as one doped with Y and 16 times as durable as one doped with Gd at the same concentration. Co-doping studies also revealed that combinations of Y with the smaller cautions, including Sc, are generally more durable than those with the larger ones at the same level of addition. This was again correlated to the changes in the phase equilibrium with varying caution size. In general, these results provide guidance for the design of novel coating compositions based on combinations of rare earth stabilizers.

A new measurement technique, Phase of Thermal Emission Spectroscopy, has been developed to determine the thermal properties of thermal barrier coatings. The measurement is a non-destructive infrared thermography technique in which harmonic heating is induced in the coating by laser excitation and the resulting temperature field is interrogated through its thermal emission. An original analysis of the emission signal has been developed, such that several coating quantities can be determined from the measurements. Two dimensionless thermal parameters of the film can be measured: the thermal diffusivity and the effusively contrast between the film and substrate. From these, the thermal conductivity and the volumetric heat capacity of the film can be calculated. A set of measurements is performed by recording the phase of the thermal emission as a function of laser frequency. The model is used to compare measured phase-shifts with calculated values based on a candidate set of unknown film properties. The film properties are determined by minimizing the sum of the squared errors between measured and calculated phases. The Phase of Thermal Emission Spectroscopy can be used to map characteristics of the coating, and is sensitive to interface thermal contact, structural aging, structure defects and coating thinning. This provides a means of detecting defects in a non-destructive way, and in particular decohesions near the interface, which can result in spallation. The variable thermal penetration depth of the measurement can be exploited. When contrasting maps at high and low frequencies, delaminations are seen in the low frequency map, but are absent in the high frequency map. In contrast, near surface defects are seen in both frequency maps. In this manner, the relative depth of a defect in the coating can be established.

Quantitative understanding of the mechanical properties of TBCs is essential to the elucidation of failure mechanisms and the development of performance and life prediction models. While information existed on the behaviour of TBC materials in bulk form, the property database for coatings was much more limited, especially regarding high temperature constitutive properties and fracture toughness. The problem acquired special urgency because of early evidence that most alternate materials exhibited substantially lower durability than the standard 7YSZ both against particle erosion as well as against spallation in furnace cycle tests. A new technique for high temperature testing of columnar coatings was developed within HIPERCOAT, as well as a model needed for interpreting the results. The test involves a hot impression of a coating deposited on a suitably rigid substrate, e.g. Al2O3, with a sapphire sphere, measuring the load-displacement curve from which the yield properties are subsequently calculated by deconvolution using the model. An additional advantage is that the densified zone under the hot impression can be sectioned and used to evaluate the fracture toughness of the material by indentation tests. The test has been applied to various coatings produced both in industry as well as in the academic laboratories involved in HIPERCOAT.

Rare earth zirconates are of interest in thermal barrier coatings (TBCs) because they combine low thermal conductivity with improved morphological stability relative to the standard material based on yttria stabilized zirconia (7YSZ). Based on the experimental information on the alumina-yttria-zirconia system available at the time and the similarities in phase constitution of the REO-zirconia and REO-alumina binaries, it was postulated at the start of this project that there would be a limit to the thermochemical compatibility of RE doped zirconias with the thermally grown alumina, which provides oxidation protection to the underlying super-alloy in a gas turbine component. A corollary was that the RE zirconate compositions would probably be thermochemically incompatible with alumina in all cases. Studies under this program demonstrated that interphases do form when RE zirconates are placed in contact with alumina at high temperature. This was ascertained for Gd and La, whose zirconates are of significant interest in the current TBC literature. Detailed studies on Gd zirconate deposited by EB-PVD on alumina as well as alumina-forming metallic substrates revealed that the reaction is sufficiently rapid at 1200°C (~1µm after 100h) to jeopardize the integrity of the coating. The process involves concurrent counter-diffusion of Gd and Al and yields a Gd aluminates interphase next to the alumina, overlaid by a layer of cubic zirconia whose Gd content increases with distance away from the interface. Formation of the aluminate layer is accompanied by evolution of pores at its interface with alumina, degrading adherence and compromising the integrity of the coating. It was also shown that the reaction is less severe but still significant at 1100°C (~100nm after 100h), suggesting that the maximum allowable temperature at the TGO/TBC interface should be substantially lower to avoid interphase formation over the required life of the coating system. Additional studies, still in progress, revealed that the kinetics is strongly dependent on the actual Gd content of the zirconate, which is a solid-solution pyrochlore phase. Because the thermal conductivity does not vary significantly over the same composition range, the results suggest that substoichiometric zirconate phases may be preferred for actual applications to reduce the risk of interface interactions. Compatible inter-layers are also a possible solution, as detailed in a separate result.

I. CALCULATION OF THE THERMAL CONDUCTIVITY OF COMPLEX OXIDES A method, based on a NEMD (Non Equilibrium Molecular Dynamics) technique, has been developed for calculating the thermal conductivity of complex oxides. Starting from the nature of atoms, the crystallographic structure and the inter-atomic potentials, it gives the thermal conductivity of complex oxides as a function of temperature. It is important to note that no adjustable parameter has been introduced in the inter-atomic potentials adopted (from published data). After validating the method on the yttria-zirconia system, and evaluating the possibilities and limits of this approach, the code has been run to investigate various oxide systems, some of them prospective (ternary zirconias, perovskites for example). It is used currently at ONERA within the framework of a project aiming at developing new ceramics for thermal barrier coatings. Contact: remy.mevrel@onera.fr or mathieu.fevre@onera.fr II. CALCULATION OF THE THERMAL CONDUCTIVITY OF A MULTIPHASE SOLID FROM A 2D OR 3D REPRESENTATION OF ITS STRUCTURE. A numerical method has been developed to calculate the effective thermal conductivity of heterogeneous materials directly from representations of the material. It is based on a FDM procedure applied on images (2D) or volumes (3D) representing the material. In this way, the real microstructure of the material is taken into account, thus avoiding any simplifying meshing. The effective thermal conductivity can be calculated from these representations in which each pixel (or voxel) is attributed the properties of the phase it corresponds to. A particular attention has been paid on the choice of the numerical method adopted and on the development of an original method for compacting the data so that large domains can be handled in an efficient way on common computers. For example, the 3D calculation of the thermal conductivity tensor has been performed on a 100 Mvoxels domain representing an EBPVD coating in 5 hours CPU times on a computer equipped with 2 Go RAM. It is to be noted that this code can be exploited for calculating the thermal conductivity of any heterogeneous material provided a 3D representation is available and the thermal conductivity of the individual phases are known. Contact: remy.mevrel@onera.fr or jean-marc.dorvaux@onera.fr

Activities related to property evaluation focused on measurement of thermal expansion coefficients of Ru-based aluminides (Pollock), and on martensitic transformations and mechanical properties of bond coats (Hemker). The CTE is an important parameter in determining the magnitude of the stresses in the bond coat during thermal cycling, as demonstrated by modelling of the TGO dynamics (Evans). These contribute to the evolution of morphological changes in the TGO and the associated spallation mechanisms (Evans, Fleck). Professor Pollock has an established collaboration with GE-Glocal Research that greatly benefited the research in this area. Professor Hemker has developed unique methods to probe the mechanical properties of bond coats using micro-scale specimens that allow not only mechanical testing but also in-situ observation of changes in the microstructure, with the capability of performing these tests at high temperatures. His techniques have produced substantial insight into the deformation of Pt-modified aluminide bond coats, as well as the martensitic transformations and related stresses that ensue due to thermal cycling and Al depletion by oxidation and interdiffusion. These techniques are being applied now to the Ru-modified bond coats.

The aim of this study was to understand the fine structure of the alumina-zirconia inter-face as a foundation for better insight into the TGO-TBC adherence. Extensive TEM studies have been performed at MPI-MF on YSZ films grown on sapphire at UCSB, including both the structure of the columns and associated porosity, their texture and the orientation relation-ships with the substrate. A significant finding has been that YSZ nucleates on sapphire with a strong out-of-plane epitaxy but otherwise in-plane randomness. This has provided substantial insight into the various mechanisms of evolutionary selection operating during growth, and the influence of the initial distribution of grain orientations on the column size. Concur-rent studies on the growth of YSZ on sapphire substrates containing a distribution of YSZ seeds prepared by precursor methods have shown that seed surfaces are less favourable for epitaxial growth. TEM analysis suggests that this is probably due to segregation of trace con-taminants to the surface.

A number of different coating variants and processing paths were investigated under HIPERCOAT. One processing path was conceptually similar to the established technology for Pt-modified bond coats. This consisted of depositing Ru by EB-PVD on carefully pre-pared substrates and performing a short inter-diffusion treatment (UCSB), followed by CVD aluminising (Cranfield or Howmet), microstructural analysis of pristine coating (Cranfield, Michigan), mechanical testing (JHU), and oxidation testing under isothermal (MPI-MF) and cyclic (Michigan) conditions. Control of aluminum activity during the aluminisation step of the process enabled multi- layered B2 coatings of NiAl and RuAl. An interesting design feature of this system was that processing could be adjusted to place either layer at the outer surface for TGO formation. Additionally, processing paths for hybrid Ni(Ru,Pt)Al coatings were established, wherein Ru deposition and inter-diffusion was followed by Pt electroplating in the usual fashion, prior to aluminising.

In systems comprising a two-phase NiCrAlY bond coat, the TGO develops thickness heterogeneities often referred to as pegs, which contain oxides other than alumina, such as fluorite. Delamination occurs primarily along the interface between the TGO and the bond coat. The study addresses whether the pegs adversely affect failure by acting as local stress intensification sites at the interface. The stress state in the vicinity of a peg is obtained for two types of loading: thermal expansion mismatch and TGO growth. The stress evolution with increasing number of thermal cycles is obtained and the sensitivity of these stresses to the geometry of the peg is ascertained. A fracture mechanics analysis is performed to determine the conditions under which a crack might initiate and grow. Constituent properties A suitable behaviour for the peg must be incorporated. Since its growth involves the internal formation of fluorite, it is considered to expand volumetrically. This effect is captured in the model by imposing a relatively large volumetric strain. Observations of delaminations at the TGO/Bond coat interface in NiCoCrAlY systems indicate that they form at low temperature, usually during cooling. Consequently, the cracks develop within a residual stress field induced by prior thermal cycling. Moreover, the system is essentially elastic during cracking, subject to a relatively low crack opening displacement, with no attendant plasticity. Again, by using elastic concepts of fracture, the worst-case scenario is envisaged: since the presence of plasticity local to the crack would reduce the energy release rate. Finite Element Model The peg is regarded as an isolated, semi-ellipsoidal domain consisting of TGO. The underlying super-alloy substrate is taken to be sufficiently thick (1000 times the TGO thickness) to behave as a half-space. Since the bond coat shares the same elastic properties as the substrate and the plastic zone within the bond coat does not reach the bond coat/substrate interface, the precise value of bond coat thickness is unimportant. The finite element analysis was performed using ABAQUS Standard. A cylindrical co-ordinate system is adopted for the axisymmetric model. In the first part of this study, the evolution of stress state is determined in the vicinity of the peg, absent cracking. The possibility of cracking is also addressed. The stresses are largest, and the interfacial toughness between bond coat and TGO is least at ambient temperature. Simplified finite element calculations are performed to assess the likelihood of cracking: it is envisaged that a putative penny-shaped crack is loaded by the residual stress state, and the energy release rate is calculated. The thermal loading history is taken to be spatially uniform. Initially, the TBC system is taken to be stress-free at the peak temperature (1000C). The TGO is allowed to grow at this temperature by imposing stress-free strains in accordance with a user material subroutine. As remarked above, the peg is considered to expand volumetrically by imposing a relatively large growth strain of 0.025/cycle. Due to computational limitations, a maximum of 30 thermal steps is considered. Crack formation Cracks can form within the residual stress field, and extend along the interface outside the peg as well as through the TGO internal to the peg. In order to explore the likelihood of such cracking, it is imagined that a circular penny shaped crack, radius c, develops along the adjacent interface, with its centre coincident with the axis of the peg. The energy release rate J and mode-mix have been determined. The crack is modelled by gap elements, with negligible initial opening. Since the system is elastic at ambient temperature, when the cracks form, the J-integral can be calculated by the nodal release method within the finite element code, ABAQUS. Typical results indicate that J has a characteristic variation with crack length, starting at zero, increasing to a peak, and then decreasingly rapidly with further increase in crack length, attributed to the change in the sign of the shear stress. Because of the normal compression, the crack is mode II throughout. The peak value of J increases linearly with increase in peg size, with magnitude strongly dependent on the level of friction. However, even for the largest pegs and the lowest friction, the J-values are less than the typical value of interfacial toughness, 20Jm-2. Consequently, the likelihood of cracks forming at pegs is minimal, unless the interface has been severely embrittled (for example, by the segregation of S). It is concluded that the magnitude of energy release rate is too small to form a crack unless the peg is unrealistically large (requiring a radius greater than 30 microns). Thus, pegs may serve more of a protective role in fastening the TGO layer to the underlying bond coat rather than promoting interfacial separation.

Ratcheting is a failure mechanism in thermal barrier systems motivated by a complex interaction of dynamic processes occurring in the various layers of the system. The mode of failure is typical of systems involving Pt-modified aluminide bond coats. In essence, it involves amplification of initial undulations in the bond coat as a result of stresses arising in the system, both at temperature as well as during thermal cycling, creating separations from the thermal barrier layer which eventually coalesce and cause spallation. A key feature of the mechanism is the lateral extension of the thermally grown oxide (TGO) as it thickens. This is ascribed to new aluminum oxide forming at the grain boundaries creating a growth strain that must be accommodated elastically or, at high temperature, by flow of the TGO, the bond coat, or both. Interdiffusion between the bond coat and the substrate is a major contributor to the evolution in several important ways, including swelling as a result of asymmetric fluxes, phase transformations with associated volume changes, and displacive (martensitic) transformations that change the flow characteristics of the metal. While the details of the ratcheting phenomena and associated mechanisms remain under investigation, models have been developed to explore the contributions of different parameters and exposure conditions. These models are contained in a number of publications, and some of them have been transferred to industry for further development.

This result comprises a technique for the study of the effects of composition on surface diffusion kinetics in thermal barrier oxides. The technique is an adaptation of Mullins' grain boundary grooving method to study surface diffusion in polycrystalline materials, with the novel feature that the materials are dense, thick (≥200µm) textured columnar oxide coatings produced by physical vapour deposition. The conditions are similar to those used in the deposition of thermal barrier coatings, but without rotation of the substrate to avoid segmentation of the columnar grain structure. The coatings are deposited on a substrate of similar composition to minimize thermal expansion mismatch stresses that may bias the kinetics of the process. The coatings are carefully polished parallel to the substrate plane, whereupon all boundaries are essentially normal to the surface, decreasing the uncertainty in the measurement. In principle, each boundary can be fully characterized by orientation imaging microscopy of the surrounding grains and texture analysis of the coating, enabling the correlation of orientation to the kinetics of grooving. Characterization of the grooves after heat treatment is readily accomplished by atomic force microscopy (AFM). The technique was demonstrated on the standard yttria stabilized zirconia used for thermal barrier coatings, as well as an alternate coating based on gadolinium zirconate. Tests on co-doped compositions are in progress, but preliminary results already exist for non-textured polycrystalline compacts of zirconia co-doped with both Y and Gd. The results will be published in a forthcoming publication, currently under preparation. The primary advantage of this technique is that it enables a direct and clean comparison of the effects of composition on micro-structural changes controlled by surface diffusion. This avoids some of the uncertainty of studying morphological changes in actual thermal barrier coatings, which are porous and have differences in porosity content and distribution in the as-deposited condition. Results so far reveal that the addition of Gd to YSZ slows down the surface diffusion kinetics, and that Gd zirconate has a much slower morphological evolution than the standard YSZ, which bodes well for its potential as a microstructurally stable thermal barrier material.

Gd was identified initially as one of the most promising additions, both as a co-dopant for YSZ as well as in the form of Gd2Zr2O7 as an alternate TBC material. Hence, a substantial fraction of the research effort was on modelling the ZrO2-YO3/2-GdO3/2 system. At the start of the program there was no information on the ternary, the binaries involving GdO3/2 had been experimentally studied but the reports for ZrO2-GdO3/2 were inconsistent, and the ZrO2-YO3/2 remained under some debate with only a rudimentary thermodynamic model available. All the binaries were assessed under this program and considerable experimental work was performed both on the ZrO2-GdO3/2 binary as well as the ZrO2-YO3/2-GdO3/2 ternary. The ZrO2-YO3/2 and ZrO2-GdO3/2 binaries were first assessed as part of the modelling of the respective ternary systems with AlO3/2, which is necessary to estimate the chemical compatibility between single- and co-doped zirconias and TGO. Using the complementary binary descriptions ZrO2-AlO3/2 and GdO3/2-AlO3/2 obtained under this program as well as the literature data, thermodynamic databases for the systems ZrO2-YO3/2-AlO3/2, ZrO2-GdO3/2-AlO3/2, and YO3/2-GdO3/2-AlO3/2 were derived. Phase relations in the ZrO2-GdO3/2-AlO3/2 system were first studied experimentally under this program, while the thermodynamic description of the YO3/2-GdO3/2-AlO3/2 system was obtained by extrapolation assuming that monoclinic, perovskite and garnet structures form continuous series of solid solutions. The calculations using these databases allow estimating the limit of compatibility with TGO (in terms of the dopant content in YSZ or GdSZ) at any temperature. Furthermore, it was shown that if pyrochlore phase is in contact with AlO3/2, phase with perovskite structure forms. This makes it impossible to use the pyrochlore phase as thermal barrier coating on the top of TGO, since pyrochlore and perovskite have different thermal expansion causing cracking. However, the pyrochlore phase could be used as outer layer of TBC to avoid its direct contact with alumina. Based on calculations of T0 (f/t) lines at different ratios of trivalent cations the iso-T0(f/t)-lines in the quasiternary systems were constructed, which restrict composition range where tetragonal phase can be obtained (stable or metastable). The assessed database for the ZrO2-YO3/2-GdO3/2 system has been used to calculate the liquidus surface, isothermal sections, isopleths at various Y:Gd ratios and traces of the T0(f/t) surface for the different temperatures and Y:Gd ratios. The latter bound the maximum combination of Y+Gd for which one can still produce a supersaturated tetragonal phase (t') at the temperature of interest, which is critical to the exploitation of toughening mechanisms that influence high temperature erosion. The liquidus surface is of critical relevance to the application of these materials by plasma spray. The thermodynamic database for the system ZrO2-YO3/2-GdO3/2-AlO3/2 was constructed by combining the thermodynamic descriptions of four quasiternary subsystems. The database is designed for Thermo-Calc software and can be used for various types of calculations, depending on the specific application. For example, the comparison of calculated vertical sections shows that increase of Gd+Y content from 7.6 to 15.2 mol.% decreases the stability field F+T and increases the stability fields F+AlO3/2 and F+YAG+AlO3/2. Increase of Y/(Gd+Y) ratio decrease the stability field of F+AlO3/2 and increases the stability field of F+YAG+AlO3/2. It should be mentioned that appearance of YAG phase is not desirable for thermal barrier coating, because the difference in thermal expansion between YAG and TBC results in crack formation during thermal cycling.

Using X-ray synchrotron radiation, growth strains as well as thermal mismatch strains on cooling to room temperature, have been measured in the TGO formed on FeCrAlY, NiCrAlY, NiAl single crystals of (111) and (110) orientation, and two industrial bond coats viz. General Electric's (GE) Pt-modified NiAl bond coat and Pratt and Whitney's (P & W) NiCoCrAlY bond coat. Reliability studies were done using FeCrAlY specimen. Room temperature strain obtained at the synchrotron for FeCrAlY specimen oxidized at 1000?C is in excellent agreement with the data obtained for the same specimen by conventional laboratory XRD. NiCrAlY substrates were oxidized at four different temperatures: 950, 1000, 1050, and 1100?C. Tensile growth strains develop in the oxide at all temperatures, corresponding to tensile stresses between ~100 and ~300MPa; these tensile growth strains are due almost certainly to initially formed metastable transition alumina transforming to the stable and denser A¨CAl2O3. While the strains remain persistently tensile at 950 and 1000?C, a slow creep relaxation at 1050oC and a rapid stress relaxation at 1100oC was observed. Another interesting observation in NiCrAlYs is the effect of phase transition in the NiCrAlY substrate on TGO stresses. The specimens oxidized at 1050 and 1100oC exhibited much larger compressive stresses, ~ 6 GPa, than did specimens oxidized at 950 and 1000oC, ~2.5GPa. Vacuum dilatometry shows a phase transformation in the substrate between 1000 and 1100?C with a large lattice expansion in the high temperature phase. The initial stresses, though tensile in both the cases, they show a different stress relaxation behaviour. This is clearly related to the different crystallographic texture in the TGO for the two orientations and different kinetics of E u A transformation. Further studies are being done. Prior to the in situ oxidation experiment at 1100oC, the samples had been cycled 24 times to 1150?C (heating and cooling times of ~ 2 minutes and 1 hr hold for each cycle). Growth strains are higher in P&W's NiCoCrAlY bond coat than GE's Pt-NiAl bond coat, approximately by a factor of two.

Thermal barrier systems used in gas turbines exhibit three major categories of failure: one based on oxidation and the second on impact by projectiles ingested into the gas stream and a third based on chemical attack by molten injected deposits (CMAS). Each category has been subject to a combination of experimental assessment and/or modelling. The models of oxidation-induced failure have reached a maturity that allows trends with constituent properties to be ascertained. The situation is much less mature for failure mechanisms caused by impact or CMAS. The second mode, particle impact giving rises to erosion or foreign object damage (FOD) has been extensively studied and a protocol has been developed to establish connections between material removal rates caused by particle impact and the properties of the thermal barrier material. The assessment is confined to materials deposited using electron beam physical vapour deposition (EB-PVD), which have a columnar microstructure. The output of this research is organized as follows. Basic penetration mechanics are summarized and used to establish formulae that relate the forces, stresses and penetrations to the kinetic energy of the impact. A mapping scheme is devised that provides a basis for further assessments. The results are combined in a manner that enables the derivation of scaling relations that characterize: (a) the thresholds for material removal, (b) the transitions between major mechanisms (expressed in terms of a mechanism map) and (c) some aspects of material removal at kinetic energies above the thresholds. SYNOPSIS OF MECHANISMS A combinations of large kinetic energy and high temperature, causes the material plastically deformation and density around the contact site. The deformation zones develop over a millisecond timescales, as the impacting particle decelerates. Outside the densified zone, kink bands form and extend diagonally downward, toward the interface with the thermally grown oxide (TGO). Within the bands, the columns are plastically bent, causing the boundaries of the kink band to crack, and thereby weakening the material. The similarity between impact and indentation indicates that the plasticity-based mechanisms governing material removal are not strongly affected by strain-rate. In some cases, the bands reach the interface with the thermally grown oxide (TGO). When this happens, they nucleate a delamination that extends outward from the impact site, along a trajectory within the TBC, just above the TGO. Such delaminations provide a mechanism for creating large-scale spalls, known as Foreign Object Damage (FOD). During initial impact, elastic waves are induced that interact with flaws in the columns. Within nanoseconds, bending waves propagate at the tops of the columns to accommodate the projectile as it penetrates. The localized bending causes trans-columnar cracks beneath the surface. The ensuing array, upon linkage, causes small amounts of material to be removed. Elastic waves also reflect off the bottom of the columns, becoming tensile waves that propagate back to the surface. The timescale is on the order of 60ns. These waves may also cause cracks to form and extend across the columns. This mechanism leads to small-scale loss of the outer part of the thermal barrier coating, known as Erosion. SCALING LAWS The scaling analysis provides some basic insight about the relative importance of the properties of the TBC having the greatest influence on erosion. As expected, elevating the TBC toughness has the most pervasive influence, especially through its role in elevating the cracking threshold. The corresponding role of the TBC yield strength (or hot hardness) is not transparent without guidance from models. The implication from the models is that softer materials (at high temperature) should have a substantially higher cracking threshold. This prediction has been tested by comparing the erosion trends among TBCs with different high temperature penetration resistance. Note, however, that for softer TBCs, the craters would be deeper. This would not be a problem for normal impacts, since there is no material removal below the cracking threshold. But cratering could adversely affect the material loss if a plastic ploughing mechanism were to operate when the particles arrive at high obliquity. FUTURE WORK Progress toward a mechanistic understanding at ultra-high temperatures (1000-1500C) has been limited by the absence of well-controlled experiments capable of duplicating turbine engine conditions. The challenges are the high temperatures (typically 1100C), the high impact velocities (490m/s), the size and composition of the particles (usually calcium-magnesium-alumino-silicate: CMAS). This requires a strategy, involving tests conducted within a temperature gradient, to achieve the correct surface temperatures and CMAS penetration into the ceramic to provide insight and understanding about damage mechanisms involving CMAS.

Thermal barrier coatings (TBCs) comprising bilayer or multilayer configurations have been proposed in the literature for a variety of reasons. Of particular interest here is the use of an interlayer of the standard yttria-stabilized zirconia (7YSZ) between a novel TBC material and the thermally grown oxide (TGO) that provides oxidation protection for the metallic component. Justification of this design has generally been based on the higher toughness and/or "adherence" of the 7YSZ material since spallation failures often occurs by crack propagation at or immediately above the TGO/TBC interface. The concept was extended in the HIPERCOAT program to alleviate the problem of diffusional interactions between novel TBC materials based on rare-earth zirconates and the TGO. In essence, the 7YSZ interlayer is supposed to act as a diffusion barrier, but it must also be strain tolerant so it should have a microstructure that provides high in-plane compliance. At issue is whether the segmented columnar pattern arising from EB-PVD deposition can be designed with the proper thickness to prevent inter-diffusion of the reactive oxides (Gd and Al) along its internal surfaces, and also sufficiently thin that one can take full advantage of the lower conductivity and enhanced sintering resistance of the zirconate TBC. It was demonstrated in this program that a segmented columnar YSZ layer of thickness [O] 50µm could be an effective diffusion barrier at temperatures as high as 1200°C, with no significant bulk or boundary inter-diffusion detected by analytical transmission electron microscopy. It was also found that the zirconate grows epitaxially on the 7YSZ columns with no perturbations on the microstructure other than the sharp change in composition. (A graded composition is arguably less preferable because of concerns about phase stability of Gd-rich t' compositions.) The integrity of this interface bodes well for its durability in thermal cycling.

This study was concentrated on the oxidation behaviour of NiCrAlY bond coats, of a particular interest being the formation of a mixed zone thermally grown oxide (TGO), which at the time has been reported only in single-phase â-(Ni,Pt)Al alloys. These mixed Al2O3-ZrO2 microstructures are peculiar to EB-PVD systems. Most of the study was performed on polished samples, focusing on the effects of the EB-PVD deposition environment on TGO evolution. Samples with the more typical grit-blasted surface were run in parallel to identify similarities and difference in behaviour. The research emphasized characterization by transmission electron microscopy (TEM), which provide information on a scale often absent in many of the studies reported in the literature. The implication, confirmed by comparing samples studied in this work, is that a suitable preoxidation treatment should preclude the incorporation of ZrO2 within the TGO. This study showed that the evolution of alumna scales during TBC deposition on multiphase MCrAlY alloys exhibits the same divergent behaviour between the major â and ã constituents reported in the literature for oxidation of non-coated surfaces. Absent the TBC the differences are primarily confined to the outer TGO surface layer, which is most likely the result of transient oxidation on heating. The initial oxidation morphologies are quite sensitive to pO2, especially on â that forms a much thicker outer oxide than ã in O2 at ambient pressure, but much thinner in the deposition chamber environment. The mechanistic origin appears to be the rapid growth of è-Al2O3 by outward Al diffusion at pO2 ~105 Pa, and its suppression at pO2 ~10-2 Pa, but the fundamental reasons for this behaviour are not understood. The outer oxide on ã also changes from Ni(Al,Cr)2O4 to (Al,Cr)2O3, but that does not seem to change significantly its thickness. An uderlaying columnar layer of á-Al2O3 was noted in all non-coated substrates, even after relatively short oxidation treatments. The outer oxide, however, can continue to grow concurrently with á as long as the latter is sufficiently thin to allow the requisite outward caution transport. Oxidation under a depositing TBC regenerates the divergent behaviour observed at the higher pO2, primarily by reestablishing the growth of è-Al2O3 on the â regions. This is believed to result from the dissolution of YSZ in the metastable oxide, delaying its transformation to á. When the transformation finally occurs, the dissolved ZrO2 and Y2O3 precipitate within the á-Al2O3, giving rise to a fine-grain "mixed zone" conceptually similar to those reported for Pt-modified NiAl bond coats. The overall thickness of the TGO and its constituent layers is in generally smaller under low pO2, suggesting that the growth rate of è during TBC deposition is significantly lower that that on oxidation in ambient pressure O2.

The Calphad technique has been applied to develop a thermodyanmic database for the Al-Ni-Pt and Al-Ni-Ru systems. The database can be used to calculate various types of phase diagrams, e.g. binary phase diagrams, ternary isothermal sections as well as vertical sections of interest. In addition the databases can be used to calculate the equilibrium state at different temperatures for a given alloy and to calculate driving forces for reactions. The database is available as database file in Thermo-Calc format.

Electron Beam Physical Vapour Deposited (EB-PVD) Thermal Barrier Coatings (TBCs) used in gas turbines may fail in service due to one of three mechanisms; oxidation induced failure of the bondcoat, damage induced by erosion or foreign object impact (FOD) with the columnar EB-PVD ceramic, and/or chemical attack of the ceramic part of the TBC by injested deposits (calcium magnesium alumino-silicates, CMAS). This study reports microstructural design guidelines to limit damage due to Erosion and Foreign Object Impact (FOD). EROSION Erosion is associated with small particle impact, where the near surface region - the top 20um - of individual columns is cracked due to particle impact. Damage is associated with elastic stress waves, which are generated during impact, propagate down the columns, and interact with flaws within each column of the TBC. Monte-Carlo modelling of this damage process has shown that column diameter has a major influence on material removal rates. This is associated with the impact dynamics and the likelihood and depth of cracks in individual columns. Only when a network of near surface cracks in adjacent columns is formed is material lost on subsequent impact. Thus smaller diameter columns means less material lost per impact and this has been validated by experiment, both a room temperature and elevated temperatures. The rate of small particle erosion appears to change little with temperature but does depend on the impacting particle size (specifically the ratio between the impacting particle size and the column diameter) and the impact velocity. Up to impacting particle size that is x2-x3 of the column diameter small particle erosion results. For particles greater than this one may observe ceramic compaction - densification - without material loss, under these conditions the energy density transmitted to the columns is insufficient to initiate column fracture, but sufficient to induce some local plasticity, hence densification. This mode of compaction damage is more likely to occur at increase temperatures, for small particle impaction. Glancing angle impact is also less damaging than normal impact, with material loss a function of the normal component of the impact velocity. Column inclination - the angle of growth of the EB-PVD morphology - is also a significant factor. The lowest erosion rates are associated with a 'normal' columnar microstructure. When the columnar structure is inclined, material loss due to erosion increases. This is significant for inclinations to the component surface shallower than 58deg. and most severe for inclinations below 20deg. Erosion rates increase with high temperature aging and are also dependent on ceramic composition. Small additions of ternary and quaternary additions at the 1-2 mole% level have been observed to modify the erosion performance of the TBC, with the behaviour dependent on the level and specific addition added. FORIEGN OBJECT DAMAGE (FOD) Foreign object impact produces severe damage and significant material loss down to the ceramic bond coat interface. Delamination cracks along the interface may also result giving significant ceramic material loss and therefore loss of thermal protection. The study and modelling of FOD has been a significant part of this work. Foreign object impact - damage from large particles in the gas stream - results in gross plastic damage to the columnar microstructure, densification, and the generation of shear cracks through the TBC to the ceramic/ bondcoat interface. Thus FOD develops from compaction damage when the plastic damage is excessive and shear cracks can for and propagate through the ceramic part of the TBC. Depending on test temperature, either shear cracking or column buckling may be observed. Shear cracking is observed at all test temperatures, while column buckling is only observed at elevated temperatures. Foreign Object Damage is associated with large particle (generally greater than 200um), high velocity (in excess of 100m/s) impacts. Small changes to the ceramic composition, through the addition of ternary and quaternary additions at the 1-2 mole% level do not appear to influence the TBCs damage tolerance to FOD, although this level of addition has an influence on small particle erosion.

Résultats exploitables

Partager cette page

Télécharger