Coatings for High-Temperature Structural Materials: Trends and Opportunities (1996)

For coating systems in high-temperature gas-turbine applications, the primary nondestructive evaluation (NDE) issues are the ability to measure the stability and adherence of materials at elevated temperatures. Processes such as oxidation, hot corrosion, creep, and solid-state diffusion degrade performance of both coating and substrate at high temperatures with relatively high reaction rates. These processes can lead to spallation of the coating, coating/substrate interdiffusion, and crack formation in the coating or substrate. These changes can occur at the interface between coating and substrate or at other interior sites. Consequently an important criterion for selecting an NDE method is the ability to monitor the subsurface condition of coated structures. However, in view of the many goals for NDE and the wide variety of coating systems, failure modes, and component designs, no single NDE approach can meet all needs. Hence NDE methods should be selected to address specific requirements.

This appendix surveys selected NDE methods that have been, or potentially could be, applied to coated structure. Several promising, emerging methods are also discussed. All NDE methods use some external source to produce a response from the sample that can be detected without causing a permanent change in the specimen. The information extracted by the measurement is determined both by the initial source/specimen interaction and the detection method. NDE methods form families of techniques classified by either source-detection method or source/specimen interaction. Table 6-1 lists some of the primary NDE methods that have been used for coating inspection grouped primarily by detection method.

Optical methods in this discussion refer to ultraviolet, visual, and infrared spectroscopy and imaging. Optical methods measure light transmitted or reflected from a specimen as a function of wavelength or position on the sample for selected polarizations or angle of incidence. From these measurements, absorption or reflection spectra can be determined if the form of the incident light wave is known or, alternatively, if images can be formed. Since many coatings scatter light from the voids or interfaces between grains, absorption measurements made by these techniques involve contributions from intrinsic absorption and scattering.

In the ultraviolet and visible light case, the incident light does not penetrate far into the sample but is confined near the coating surface by scattering. While surface cracks and other features that have a surface expression can be detected, little information can be obtained about the coating/substrate interface. Scattering is wavelength-dependent, however, and decreases with increasing wavelength. As a result, scattering is low enough in the near-infrared that it is possible to observe reflection from the substrate and the interface between the substrate and coating and to identify regions where the coating (e.g., yttria-stabilized zirconia TBCs) has disbonded from the superalloy substrate.

Emission spectra can be obtained in the infrared region for heated samples and the optical properties of TBC materials, including the absorption coefficient, determined using Kirchoffs relations between emissivity, reflectivity, and absorptivity. Thermally stimulated emission spectra are much more difficult to obtain in the ultraviolet to visible ranges as seen from Planck's radiation law,

where Q is radiative flux per unit wave length, ¬ is the wavelength, T the temperature, and C1 and C₂ are the first and second radiation constants. For the temperature range of interest to turbine coatings applications, virtually all of the radiation is at wavelengths longer than 1 to 1.5 µm independent of the emissivity of the specimen. For yttria-zirconia and some other ceramics, the emissivity has been determined and the absorptivity derived (Thomas and Joseph, 1987; Thomas et al., 1988; Thomas, 1989). There is a relatively low intrinsic absorption in the 2- to 4-µm region, and this allows the substrate to be viewed directly if scattering is excessive. This region of transparency should be explored for other candidate TBC ceramics to assess the prospect of inspection of the interface with the substrate as well as the coating microstructure.

A second aspect of this region of high transparency is the role of radiative transport in heat transfer across a TBC. Some thermal-imaging methods described later in this appendix can determine the rate of heat transport by conduction. However, the relative importance of conductive versus radiative transport does not seem to be well understood and should be studied (see appendix B for a discussion of energy transport in these coatings). If the

gas-emission spectra follow a black-body law at the operating temperatures of existing and proposed engine temperatures, the peak emissions fall within the region of coating transparency suggesting efficient radiative transport. If the radiative load is high, then strategies to increase the thermal efficiency of the TBC must minimize both conduction and radiation.

Visual inspection is widely used for NDE of coatings. For metallic coatings, the primary goals are to determine if the coating is spalled from the substrate and if either the coating or the substrate is cracked. The major failure mechanisms of loss of oxidation protection by loss of the protective metal-oxide layer or by interdiffusion of coating and substrate can lead to the attack of the coating/substrate interface and to cracking under thermal cycling. Spallation and cracking of the coating can be identified by visual inspection in some cases. While visual inspection is not quantitative and cannot determine the useful remaining life of the coating, it can assess current damage at some level.

Enhanced visual inspection methods may improve the quantitative aspects of the method and should be applied to in-service inspection. Methods based on computer-aided image processing should be standard methods of inspection of engine components for sizing of parts and location and quantification of surface defects, such as cracks and spalls. Sizing of individual surface features may provide statistical data on qualitative changes that would aid routine inspection. The changes of surface features with service time should be monitored to document the initial condition and the rate of change with time with the goal of improving lifetime prediction. Technical aids such as coherent fiber optics in the visible and infrared might allow visual inspection of some parts to be carried out in situ.

Thermal imaging methods use a source of energy to create a temperature field in a specimen and a detection method to monitor the temperature changes. The information obtained depends on the initial source distribution, the subsequent thermal diffusion, and the detection method used. The dependence on the detection method exists because detection is based on measuring a change in some physical property of the specimen caused by temperature. Different properties may have different temporal and spatial scales as discussed below.

For optically opaque materials, amplitude-modulated (pulse or step) optical heating acts as a surface heat source equal to a flux of heat into the specimen surface. This flux induces a change in surface temperature with time based on the magnitude of the flux and the rate of thermal diffusion into the specimen.

For a layered specimen, the surface temperature initially follows the functional form of the uniform specimen but then deviates in a direction that depends on the characteristics of the second layer, the interface properties, and the depth of the second layer in the sample. The time at which the deviation occurs is the thermal transit time and can be used to determine the coating thickness if the thermal diffusivity is known, or alternatively, the thermal diffusivity if the thickness is known. For example, accuracy of measurement for the thickness of 200-µm TBC coatings is better than 5 percent.

This method can also confirm that an infrared image represents a disbonded region. The time-resolved thermal image quantitatively measures the extent of the disbond, whereas the passive infrared gives rapid views of the area. These methods may allow characterization of the coating/bondcoat/substrate interfaces as a function of time if the spatial resolution for imaging of the interface is adequate. The results suggest that infrared source and thermal detection methods can be complementary methods for monitoring the condition of ceramic TBCs on high-temperature substrates.

Other sources (i.e., non-optical) that heat internal regions of the coating or heat the substrate or bondcoat may provide the spatial resolution needed to detect substrate cracks or changes in conductivity by thermal cycling. An illustration is the localized induction heating (Lehtiniemi et al., 1991, 1993) of a plasma-spray-deposited ceramic on a high-temperature metal alloy. The inductive source heats the component resistively by currents induced in the substrate. Since cracks and other thermal features inhibit current flow and heat flow, this method can image subsurface features through the analysis of surface temperature distribution.

Microwave-source methods also have promise for TBC characterization. Many ceramic materials are relatively transparent to microwaves within the 10 GHz region. Hence microwaves can penetrate the coating with little loss and interact with specific features or constituents in the coating that may be present as contaminants that were deposited during the period of service. The resultant temperature changes may be detectable using infrared imaging methods. Thermally detected microwave spectroscopy has already been shown to have high spatial resolution (10 µm) and the ability to determine the presence of selected impurities in dielectric materials. As a speculative idea, perhaps the presence, or even the concentration, of sulfur or other fuel-related species could be determined at the coating/substrate interface by this method.

Radiometric methods are illustrative of the use of NDE methods in process control applications (Moreau et al., 1991, 1992). In plasma spraying, the heat source is ceramic particles that are accelerated toward a substrate through a hot plasma flame. The particles melt in passing through the flame. The infrared emission from the particles is monitored while in passage from flame to substrate and on the substrate itself. Using fast infrared detection methods, time-temperature profiles of individual particles on the deposited surface can be measured as they cool. The heat transfer rate depends on the thermal conductivity of the deposited layer and the characteristics of the particle, including particle size. Issues affected by the rate of cooling are lateral spread on the coating, extent of crystallization, and size of crystallites. Some changes in overall morphology can be followed. This method might be of value in providing sensor input to allow real-time control of plasma spraying so that the desired microstructure is deposited.

There are a range of thermal imaging methods that use a thermal source to induce a mechanical response by thermal expansion. Images are formed by scanning an amplitude-modulated, focused laser beam over the sample and measuring the surface displacement of the specimen either at the point of heating or at some point close by. There are several factors that affect the response, including the peak height of the surface deformation produced by the heating and the lateral spread and hence the slope of the deformation. Such methods can provide information about the spatially diverse mechanical responses of a specimen. These methods may have utility in imaging spatial changes in the stiffness of interfacial bonds.

Induced current methods use relatively low-frequency magnetic and electric fields of varied frequency as the excitation source. In the case of a magnetic source, the induced magnetic fields generated by the interaction of the source with a conducting sample are measured. This interaction takes the form of eddy currents, which are dispersive waves penetrating the specimen to a depth called the skin depth. This depth, d, can be expressed as where s is the specimen conductivity, µ the permeability, and ω is the frequency. As in the case of thermal imaging, the depth sampling provided by the skin depth, and especially its frequency dependence, allows buried regions of a specimen to be examined in a single-sided measurement. Eddy-current methods can determine the electrical properties of individual layers in multilayer conducting materials and the properties of conducting regions below a dielectric layer. They can also determine the presence of a crack in a buried conductive layer and, in some cases, the crack density. From an NDE perspective, this could be of value in assessing changes in the bondcoat or substrate with time because of crack formation or interdiffusion.

A derivative of the eddy-current method called photoinductive imaging (Moulder et al., 1992) offers higher spatial resolution (micron scale) for many materials. Eddy-current methods can also measure coating thickness with good accuracy (Smith and Stephan, 1990).

Other electromagnetic methods related to eddy currents use a variety of magnetometers and magnetometer arrays to monitor specimen condition (Goldfine et al., 1993; Wickswo et al., 1993). These advantages derive from the ability of nonsearch coil methods to operate at low frequencies to spatially pattern the fields in the specimen (Goldfine et al., 1993).

Injected current methods monitor the distribution of current injected into a material or component using sensitive magnetometer arrays. These arrays can be physical arrays, in which measurements are made in parallel, or scanned arrays. The injected currents can be of any frequency that can be detected, and hence the depth of penetration into the specimen is controlled. One goal of these methods is to monitor through-thickness changes in current distribution that result from spatial variations in conductivity that are either intrinsic (e.g., caused by deformation) or extrinsic (e.g., caused by cracks). Initial experiments suggest that these methods may have a place with further development.

Ultrasonic methods monitor the amplitude, phase, and velocity of an ultrasonic wave injected into the sample. Again, both transmission and reflection modes are available. For a layered medium, such as the coating/substrate system, in addition to a bulk compressional and two shear waves, Rayleigh waves and other interfacial waves exist. The characteristics of these surface and interface waves depend on the thickness of the coating and the stiffness of the mechanical bond between the coating and the substrate. These methods offer some promise for special applications, especially if noncontact generation and detection methods at elevated temperatures are used.

This method is related to the laser acoustic interferometric imaging discussed previously but differs in that it uses the

analysis techniques of ultrasonics. A pulsed laser source is used to produce an ultrasonic wave in the specimen, and the amplitude, phase, and arrival time of the wave are measured in a typical manner. The previous work could be viewed as the quasi-static limit of laser source ultrasonics.

X-ray methods based on radiography or tomography have clear applications to inspection of coated components. Examples include visualization of internal passages in turbine blades, the location of casting defects, and a determination of the crystallographic structure of substrates, especially directionally solidified and single-crystal materials. Synchotron-source X-ray topography has the advantage of allowing the crystal orientation and structure of an entire part to be visualized in a single measurement. While these methods require special facilities, they might be useful in process development.

Other methods are based on X-ray fluorescence and scattering (Verbinski, 1990) and are directed at the inspection of the coating. The X-ray scattering cross section is energy-dependent with elastic scattering (Compton scattering) dominating for low-atomic-number targets and medium energies and fluorescence dominating for high-atomic-number targets and low energies. In the case of fluorescence, energy-selective detection methods are used to provide some discrimination of elemental composition. Because both coating and substrate are based on high-atomic-number material, this method appears to have some promise for composition and coating density measurements.

There is a possibility that this method could be developed to include profiling of composition with depth and to monitor changes associated with exposure to the engine environment. In practice, however, the results have been of limited value because of counting-rate problems and difficulties in deconvoluting fluorescence line profiles.

Another X-ray method that has only been applied to subsurface characterization of 10-µm multilayer films uses thermal detection to monitor X-ray absorption as a function of photon energy (Masujima et al., 1987). The energy absorbed from an X-ray source of variable energy increases near the characteristic wavelength of specific elements. This increased heat flux into the sample changes the volume and surface temperature of the sample. The temperature changes can be detected by methods that monitor either the surface temperature (e.g., radiometry) or the acoustic response of the sample as discussed earlier in this appendix. Depth profiling is possible based on the arrival time for thermal diffusion from the layer to the surface or on the acoustic propagation time to the detector. This approach has not been tried for coating systems, but it has been used to characterize subsurface nickel and copper multilayers in semiconductor systems. Its potential advantages are that it is single-sided, multi-element-sensitive, preferentially responsive to high-atomic-number elements, and able to depth profile. Its disadvantages are the need for a high flux, tunable X-ray source.

Transient grating characterization methods have been recently introduced for characterization of deposited layers on surfaces. In the simplest form, they involve formation of a linear array of lines of optical or other energy on the surface of the specimen using moire or interferometric methods. Detection is also optical through either light reflection or emission methods. Because the grating is both a thermal and an acoustic source, several detection modalities are possible. In the first case, thermal properties can be determined in the plane of the coating and across the coating in a single measurement. In the second case, ultrasonic waves of known spatial and temporal structure can be launched and reflections measured at very high rates. Thin coatings and multilayered systems can be analyzed even for submicron layers, such as those present in some of the nanostructured coatings being considered for TBC applications. The array allows the spatial frequency of the measurement to be varied and the spatial scale of characteristic defects to be evaluated. These methods have advantages in being noncontact, fast, and flexible in the information presented.

An example of possible applications of transient (picoseconds to microseconds) heating in coating characterization is the use of fast radiometric heating of TBC and NiAl coatings. In both cases, the time evolution of the temperature response under fast pulse heating was a relaxation whose time dependence is

T(t) = t^-a

where t is time and a < 0.5 (the value for one-dimensional diffusion). This result suggests that the thermal diffusion in these coatings is geometrically constrained by the coating structure. It may be possible to infer structure using these methods, perhaps even in the process of layer deposition.

As discussed previously, many ceramics used for TBC applications are transparent to microwaves. Microwave reflectance measurements from the coating/substrate interface can be made to temperatures in excess of 1100°C and the reflectivity measured as a function of temperature. This raises the prospect of direct measurement of the substrate temperature through coatings. As another feature of the method, changes in reflectivity with melting and possibly with other phase changes in the material have been seen. This method is in the early phases of development but appears to have promise for application in all four areas of NDE applications.

Transient grating methods can also be produced by modulated stresses in the plane of the specimen (Lesniak and Boyce, 1993). These methods are at much slower time scales than the interferometric methods but provide information about the response in the quasi-static regime for both thermal diffusion and stress relaxation. They have proven to be of value in other materials and coating systems and should be considered for high-temperature coatings.

Goldfine, N.J., A.P. Washabaugh, S.V. Dearlove, and P.A. Guggenberg. 1993. Imposed w-k magnetometer and dielectrometer. Review of Progress in Quantitative Nondestructive Evaluation 12:1115-1122.

Lehtiniemi, R., J. Hartikainen, J. Rantala, J. Varis, and M. Luukkala. 1991. P. 441 in Review of Progress in Quantitative NDE, Vol. 11A, D.O Thompson and D.E. Chimenti, eds. New York: Plenum.

Lehtiniemi, R., J. Hartikainen, J. Rantala, J. Varis, and M. Luukkala. 1993. P. 1931 in Review of Progress in Quantitative NDE, Vol. 12B, D.O Thompson and D.E. Chimenti eds. New York: Plenum.

Lesniak, J., and B.R. Boyce. 1993. Forced diffusion thermography. Pp. 92-102 in the Proceedings of SPIE-The International Society for Optical Engineering Conference on Nondestructive Inspection of Aging Aircraft, Vol. 2001, San Diego, California. Bellingham, Washington: SPIE.

Masujima, T., H. Kawata, Y. Amemiya, N. Kamiya, T. Katsura, T. Iwamoto, H. Yoshida, H. Imai, and M. Ando. 1987. X-ray photoacoustic effect of solid materials. Chemical Letters 973-976.

Moreau, C., P. Cielo, M. Lamontagne, S. Dallaire, J.C. Krapez, and M. Vardelle. 1991. Temperature evolution of plasma sprayed niobium particles impacting on a substrate. Surface and Coatings Technology 46:173-187.

Moreau, C., M. Lamontagne, and P. Cielo. 1992. Influence of coating thickness of the cooling rates of plasma sprayed particles impinging on a substrate. Surface and Coatings Technology 53:107-114.

Moulder, J.C., E. Uzal, and J.H. Rose. 1992. Thickness and conductivity of layers from eddy current measurements. Review of Science Instruction 63:3455-3465.

Smith, K., and R. Stephan. 1990. Protective Coatings Nondestructive Evaluation. Final Report. Contract F33615-87C-5221. Wright-Patterson Air Force Base, Ohio: U.S. Air Force Wright Research and Development Center .

Thomas, M.E. 1989. A computer code for modeling optical properties of window materials. Proceedings of SPIE: Window and Dome Technologies and Materials 1112:260-267.

Thomas, M.E., and R.I. Joseph. 1987. Characterization of the complex index of refraction for sapphire, spinel, alon and yttria. Paper presented at the Meeting of the IRIS Specialty Group on Infrared Materials, June 9-10.

Thomas, M.E., R.I. Joseph, and W.J. Tropf. 1988. Infrared transmission properties of sapphire, spinel, yttria, and ALON as a function of temperature and frequency. Applied Optics 27:239-245.

Verbinski, V.V. 1990. Protective Coatings Nondestructive Evaluation. Final Report. Contract F33615-87-C-5221. Wright-Patterson Air Force Base, Ohio.: U.S. Air Force Wright Research and Development Center.

Wickswo, J., D.C. Hurley, Y.P. Ma, S. Tan, and J.P. Wikswo, Jr. 1993. P. 633 in Review of Progress in Quantitative NDE, Vol. 12A, D.O. Thompson and D.E. Chimenti, eds. New York: Plenum.