Micro-scale Fracture Testing of Graded (Pt,Ni)Al Bond Coats

Abstract

PtNiAl bond coats are diffusion aluminide coatings deposited on superalloy based turbine blades for oxidation resistance and improved adhesion between the substrate and the YSZ thermal barrier coating on top. They are deposited by pack aluminisation, which makes their microstructure inherently graded and heterogeneous as well as replete with a variety of precipitates and second phase particles. The microstructure also continuously evolves during thermal cycling, because of interdiffusion with the substrate and the continuous loss of Al to the thermally grown oxide scale on top. During service, the bond coats are exposed to impact, thermal expansion mismatch, thermo-mechanical fatigue and inter-diffusion accompanied by phase transformation, which become leading causes of their failure. The bond coats being B2 crystal structures are known to be brittle at room temperature, due to which they are expected to fail during cooling, although they undergo plastic relaxation by creep above the BDTT. Little attention has been paid to the mechanical response of the bond coats, while a number of studies focus on optimizing their composition for oxidation resistance. The fracture properties of these coatings, in particular, are not very well understood due to the several different length scales of their complex microstructure playing a part. In this context, there is an interest in determination of the fracture toughness of bond coats under different loading and temperature conditions. In the present work, the fracture properties of bond coats is measured with micron-scale resolution using edge notched doubly clamped microbeam structures positioned at individual zones of the graded bond coat, subjected to bending. In order to extract the stress intensity factor for this new configuration and to determine the stress distribution and stability of this geometry under different loading conditions, extended finite element analysis (XFEM) is carried out. After establishing the microbeam geometry as a viable fracture toughness testing configuration, the contribution of different microstructural variables to toughening at room temperature is studied using SEM based in-situ testing. Since the exact composition and structure of the coating depends principally on the elements constituting the matrix-Pt, Ni and Al content, which themselves depend on the deposition parameters, we have examined in detail, coatings aluminised at different temperatures (increasing coating thickness), varying Al content in the pack mixture and starting Pt thicknesses during electro-deposition. These parameters are by no means exhaustive and there is wide scope to investigate the effect of other processing variables as well as their synergistic effects on the mechanical behavior of these coatings. Following this, the high temperature fracture behavior of the stand-alone coatings in tension is also studied to determine their brittle to ductile transition mechanism in the presence of a notch. While this covers the average behavior of the entire coating cross-section, such a study is important to establish the BDTT unambiguously since there are chances of under-estimation of these temperatures in the absence of a notch. Also free¬standing coatings without the underlying substrate offer respite from residual stresses influencing the results of such tests. The present study essentially consists of two distinct parts, one focused on the development of the testing technique to cover multiple length scales of any graded thin film or coating and the other on the determination of fracture properties of the bond coat using these methods. The thesis reads in the following way: Chapter 1 gives an introduction to the diffusion aluminised bond coats, with a focus on the failure mechanisms associated with them while underlying the need for small scale testing in these systems. The conditions driving failure in bond coats can be vast and varied and it is extremely difficult to pin-point a single important cause and also to develop predictive capabilities regarding their failure. This is described as the motivation for the present work, with an objective of finding the variation in fracture toughness values for PtNiAl bond coats of different coating thicknesses and Pt content across the temperature range spanning the BDTT of the sample. Chapter 2 describes in detail all the available literature on thermal barrier coatings in general, and diffusion aluminide bond coats in particular, while specifically highlighting its mechanical response to loads during service. The deposition parameters during pack aluminizing and the graded microstructure which develops as a consequence of the diffusion process are described. The material’s microstructure dictates its properties, but there has been limited work on the mechanical behavior of the coatings themselves due to the difficulty in preparation and testing of free-standing films of the same. Since the base matrix is that of β¬NiAl, and there has been extensive work reported on bulk NiAl in the literature, which is discussed next. This would serve as a benchmark for comparison with the properties of the bond coats themselves, which are expected to respond differently due to their continuously evolving and complex microstructure. A summary of the known mechanical properties of the coatings themselves is given next along with the failure mechanisms that have been proposed. Since the study deals with fracture properties, a short introduction of linear elastic fracture mechanics follows before elaborating on the various small scale fracture testing geometries that have been developed. There are specific differences between testing geometries, stress states as well as in the instrumentation between small scale and bulk fracture toughness tests, which are highlighted. Since these configurations are material and device specific, each group has worked out its own instrument capabilities and mechanics required to extract the mechanical properties of interest from these testing techniques. Due to these differences in addition to the differences in the size scales of the samples tested, the reported properties show a wide variation. Lack of standards add to the difficulty in interpretation of the data; moreover add to the controversy on whether a size effect exists for fracture, as it does for strength. All the non-standard small scale testing configurations require modeling and simulation to extract the desired properties from them, and the present study applies the XFEM to determine the stress distribution and calculate the stress intensity factors corresponding to the fracture loads recorded from experiments. An introduction to the XFEM method is given in the last part. Chapter 3 gives all the experimental and simulation procedures that were carried out in the present work. Since the bond coat properties need to be compared with their bulk counterparts, both the samples are characterized. The exact material compositions chosen for the study were plain NiAl, 2PtAl and 5PtAl among the pack aluminized coatings and bulk arc-melted PtNiAl samples with varying concentrations of Ni and Pt which matched the bond coat matrix compositions. The choice of the three coatings was made depending on the previously known information regarding their microstructure. The deposition conditions, temperature and times of annealing are listed, followed by a brief summary of the general characterization techniques used to study the microstructure of the bond coats before and after fracture testing. Since the micro-beams under bending were fabricated using a focused ion beam, and the micro-tensile specimen were machined by electro-discharge machining, both the micro-machining procedures are described. At such small length scales, conventional testing methods cannot be used and several modifications were incorporated to the testing geometries which are described in the next section which covers two principal fracture testing methods-microbeam bending and mini-tensile testing, along with the advantages and limitations of each. Modeling is an indispensable tool for determining stress distributions in such new geometric configurations involving material property variations, and details of the exact XFEM procedure that was implemented in ABAQUS is given in the last part of this chapter. Chapter 4 summarises the microstructure and indentation properties of the bond coat and bulk NiAl samples characterised using X-ray diffraction, electron microscopy and nanoindentation. XRD was used for phase identification, texture and determination of lattice parameters of the specimen, which confirmed β-NiAl (with no texture) as the matrix with the lattice parameter varying as a function of composition. The SEM-EPMA combination was used for probing the compositional and microstructural gradients, grain size and precipitate distribution across the coating cross-sections. The bond coat was found to have 4 distinct zones with the Ni:Al ratio gradually rising across its thickness. In addition to this, the four zones had very different grain sizes, precipitate type and distributions. Hardness and modulus values were reported from nanoindentation measurements across the coating thickness over a temperature range from 25 to 400˚C and were seen to follow the composition gradients in different ways based on the effect of the off-stoichiometric defects on these properties. The hardness was found to be a minimum for the zone with stoichiometric composition, as was the case in the bulk sample, while the modulus dropped continuously with increasing Ni content in the matrix. These are important to develop a one-to-one correlation with the fracture properties and to understand the micro-mechanisms of the same. Chapter 5 gets on with the specifics of the testing geometry. Since most of the variables of the testing technique were studied using simulation procedures, a large part of this chapter deals with the results from the modeling technique using XFEM. The XFEM is introduced in detail and its applicability in modeling of cracks and discontinuities and advantages over conventional FEM are explained. The material properties are taken from the nanoindentation data and the modeling assumes linear elastic fracture mechanics. As a validation measurement, a conventional three point beam is modeled in bending and the results compared with analytical solutions of the same. The three point beam bending geometry is also used as a benchmark to study the stability of the new geometry, now with fixed boundaries in place of a free ends. This is followed by the results from the modeling for different variables like mesh density, notch root radius, loading offsets, beam dimensions and crack length (a)/specimen width (W) ratios where both the stress distribution as well as KI are captured in 3-D for stationary cracks while crack trajectories are obtained for propagating cracks. The notch root radius is seen to not affect KI below ~300 nm and such notch radii are easily machinable in the FIB at lower currents. The crack trajectory from the experiments is seen to follow the direction of maximum tangential stress, which is also modeled very well in the XFEM. The contribution of KII to the measured stress intensity factor with increasing offsets is also calculated from the model. Stable cracking is seen for the clamped beam geometry, with KI dropping off beyond a critical a/W ratio. This was true even for a model assuming homogeneous, elastic properties with a flat R-curve under load control. This makes the clamped beam structure require higher loads for continued propagation of cracks. This critical ratio is dimension dependent, making a shorter thicker beam stable in comparison to a longer, slender one. This is unusual, especially in comparison to the three point bend geometry which shows stable cracking only in displacement control, specifically for large a/W ratios alone. Also superimposition of the load-displacement curves from simulations with those of experiments gives a good-fit. The experimental results are shown next to back¬up the claims made on geometric stability of such clamped structures. Digital Image Correlation is introduced as a means for direct measurement of crack opening displacements (COD) and fracture toughness without the aid of KI formulations. This also served as a cross¬check on the assumptions of linear elastic fracture mechanics (LEFM) made in the simulation and a good correlation is seen between the CODs measured experimentally and that obtained from the FEM analysis. Fracture toughness measurements of brittle materials with known KIC values, like fused silica glass and single crystal Si film from this proposed geometry are reported as additional validation of this geometry. Further the capabilities of in-situ testing using this geometry to measure R-curve and fatigue properties along with the initiation KIC values are shown via results from monotonic and cyclic loading under different conditions. Chapter 6 returns to address bond coat fracture at room temperature, which is the main objective of the present study. Fracture toughness is evaluated both ex-situ and in-situ, using clamped microbeam bending experiments across individual zones of the 5PtAl bond coat and for different initial Pt contents in the zone 2. KIC is seen to rise sharply with increasing Ni content of the matrix in the former case, from 5 to 15 MPam1/2 which is attributed to the change in defect chemistry with changing stoichiometry. Al rich NiAl is found to be more brittle due to vacancy hardening while Ni rich NiAl is known to increase the metallic character of the NiAl bond. Both Ni rich and Pt rich (Pt,Ni)Al give higher toughnesses among the coatings studied while the crack trajectories and toughening mechanisms distinctly depend on the precipitate morphology in individual zones. Alloying additions are seen to add to the complexity of the fracture behavior of bond coats by strengthening the matrix or by improving its ductility. Micro-kinking, grain boundary and precipitate bridging are seen in the crack wake as contributing factors to partial closure of the crack on unload. The influence of each of the microstructural variable on the fracture mode is dissected in detail before coming to an overall conclusion. The microbeams show controlled, stable cracking, which enable following of the crack trajectories across micron-length scales and make R-curve measurements possible. Both 2PtAl and 5PtAl compositions show a rising R-curve within the length scale of an individual microbeam tested. Size and geometric effects on real vs apparent R-curve behavior are discussed at the end of the chapter. Chapter 7 addresses a different area of high temperature fracture of bond coats, which becomes relevant in terms of determination of brittle to ductile transition temperature (BDTT) in notched specimen and in evaluating topography after failure across this temperature range. This set of tests is designed to measure fracture toughness and study the fracture mode along the temperature scale to exactly identify the BDTT for a given bond coat composition and strain rate, below which the coating undergoes brittle catastrophic fracture and beyond which it creeps and relaxes plastically at very low stresses. Notched free¬standing bond coat specimens are pulled in uni-axial tension to fracture and the stress at failure is used to calculate the average fracture toughness of the bond coat. The stress-strain curve shows linear elastic behavior upto the BDTT of the bond coat as expected, beyond which it becomes increasingly plastic. The KIC is seen to rise marginally upto 750˚C beyond which it showed a significant increase, from which the BDTT was calculated to be ~775˚C for notched samples. The KIC is not reported beyond the BDTT due to irrelevance of LEFM after macroscopic plasticity sets in. Fracture mode is seen to change from transgranular cleavage below the BDTT to void coalescence and ductile rupture beyond it. The experimental challenges, differences in the through thickness KIC’s obtained from tensile tests vis a vi bend tests (due to changing stress states and size scales), as well as mechanisms of ductile to brittle transition in the context of previously available literature are discussed. Chapter 8 gives the closure and important conclusions from the present work. It summarises the key results from the testing technique and highlights the proposed mechanisms which bring about a rising fracture toughness with both increasing Ni:Al ratio across the bond coat cross-section and across individual micro-beams themselves. Some new techniques and geometries which can be adopted for fracture property determination, on which work was initiated but not complete, are also proposed. The last part of the chapter deals with the future implications of the results found and some open threads and challenges on bond coat optimisiation for different properties, which are yet to be dealt with.