Non-destructive testing of engineering ceramics

TWI Bulletin, January 1988

Graham Edwards, BSc (Hons), DMS, MInstNDT, is a Senior Research Engineer in the NDT Research Department.

The use of ceramics is becoming more and more widespread as new and increasingly sophisticated materials become available. Traditional non-destructive testing (NDT) techniques are often unsuitable with these materials. In this comprehensive review the application of new and conventional NDT techniques to ceramic materials is examined.

There has recently been a resurgence of interest in materials science in the belief that new materials will stimulate economic growth. ^[1] In particular, ceramic materials are regarded as having an important role to play ^[2] as they exhibit highly desirable properties for use in some special environments. They tend to be chemically inert, to have high strength and rigidity and considerable resistance to the effects of high temperature. It is therefore timely to examine NDT of ceramics by examining the conventional NDT methods available and assessing the use of new testing methods.

This article concentrates on structural or engineering ceramics as distinct from electrical and traditional ones ( Fig.1). Although they represent less than 10 per cent of the ceramics market in value terms ( Table 1), ^[3] their importance is likely to grow. ^[4] These ceramics can be split into four types. Simple binary oxides that include alumina (Al ₂0 ₃), magnesia (MgO) and zirconia (ZrO ₂), the non-oxides that include silicon carbide (SiC) and silicon nitride (Si ₃N ₂), and finally, more complex mixtures that have been developed recently, of which sialon (a mixture of Si ₃N ₄ and Al ₂O ₃) is the most promising for engineering use.

Table 1. Fine ceramics market in Japan ^[3]

In the UK there have been two significant development programmes for engineering ceramics. The 'Advanced ceramics for turbines' (ACT) programme was instigated by Rolls-Royce in 1981 and is sponsored by the Department of Trade and Industry and by aerospace companies. The 'Ceramics applications for reciprocating engines' (CARE) programme was launched in January 1986 as a three year £6M programme with the aim of increasing the potential for manufacture and use of ceramic engine components. These two programmes cover the most important uses for engineering ceramics, namely turbines and reciprocating engines. Other uses include refractory linings, insulators and artificial human joints ( Table 2).

Table 2. Uses of some engineering ceramics

Quality control of ceramic components

The NDT of ceramic components is as yet poorly developed. It is therefore necessary to look at wider aspects of quality control. The quality control of manufactured goods is accomplished by measuring dimensions, properties and other characteristics and comparing these measurements with standards predetermined by quality assurance specifications and procedures ( Fig.2).

Sometimes this information can be obtained only by destroying representative samples. Where possible, there are obvious advantages in using NDT methods. For example, the integrity of every component can be inspected. On the other hand, the results of NDT cannot simulate service conditions as can mechanical tests.

The tests which form a part of quality control fall into four categories: materials, procedural, validation and standard tests.

Materials tests

As production becomes more automated, variations in the raw materials are less easily tolerated. If variations are unavoidable, then tests are needed to detect minor material changes, so that the manufacturing process can be modified instantly to maintain the quality of the final product. Minor variations in material grades can have catastrophic effects on product quality. For example, a minor change of grain size in powders of identical composition can lead to inhomogeneous densification. If no inspection is carried out until the final product is produced, then not only are expensive manufacturing processes wasted, but also the problem of finding the cause of defects becomes more difficult.

Procedural tests

Validation tests

Validation tests can involve proof tests and full scale simulation loads on the final product. Proof testing is common with ceramic components because there are no alternative test methods. The use of test loads which exceed, by a good safety margin, loads to be expected in service, is not entirely satisfactory, however. It may introduce flaws which lead to subsequent failure. Sampling has to be statistically significant and the component may be weakened during the test. It is in this area therefore that there is the most urgent need for NDT.

Standard tests

Standard tests are directed to ensure that the material, fabrication and manufacturing process have led to at least the minimum level of product quality. They are standardised and should meet national and international acceptance criteria. ^[5] Such tests are a long way from being available for ceramic components.

The aims of NDT

NDT includes all physical testing methods for detecting defects in a material or component without in any way damaging it, making it possible to inspect every component produced. Proof testing, which is more widely accepted than NDT for quality control by ceramic manufacturers, may in some cases be practicable on every component, but may induce incipient flaws that eventually cause failure in service.

The aim of NDT is to detect defects, which can be defined as flaws in the material that create a substantial risk of failure. Whether a flaw is a defect or not is determined primarily by its nature and size.

Table 3. Comparison of properties of some engineering ceramics with mild steel ^[9]

The nature of flaws in ceramics

Flaws commonly found in ceramic components include porosity, inclusions, delaminations and cracks. Porosity is inevitable, because complete densification of the ceramic during sintering is impossible. Acceptable levels of porosity can range from almost 40% in alumina refractory blocks to 0.01% in fine alumina ceramics. In fine engineering ceramics, failures can be initiated from pores no more than 10µm across.

During sintering, the loss of volatile constituents is the usual cause of porosity. Inclusions may be the result of dust from the atmosphere, or wear debris from the machinery used to crush, sieve and filter the ceramic powders.

Delaminations and cracks caused by internal stresses may arise during shaping and firing. Stresses are also set up when coating a metal with a ceramic or when welding a ceramic to a metal, because gross differences in thermal expansion occur between ceramic and metal.

Surface cracks arising during machining present a particularly difficult problem and place a severe limitation on the machining of ceramic components.

Size of flaws in ceramics

Critical flaw size is determined by material properties and the service conditions of ceramic components. An important characteristic of ceramics is their extreme brittleness. ^[6] They are therefore weak when subjected to thermal shock and suffer brittle fracture on impact. The strong bonds between atoms in ceramics make dislocation movement difficult and local high stress concentrations cannot be dissipated. In contrast, the dislocations within metals require much less energy to move so metals are ductile to some degree.

Great care is therefore needed in designing ceramic components to avoid notches and other stress concentrations, and if possible to ensure that the ceramic is under compressive loading. Small flaws cannot be tolerated in tension because fracture, once initiated, will propagate catastrophically.

To improve their toughness, studies are being carried out on transformation toughening. ^[7] There is also an interest in the incorporation of ceramics into composite materials for example by reinforcement of silicon nitride with carbon fibres. ^[8]

The identification of maximum permissible flaw size is difficult from the results of mechanical tests, because of the variability of test results. Any measurement will relate only to the particular test specimen used and test conditions. Unlike a metal, the strength of a ceramic is heavily dependent upon the size and location of the small flaws that are an inherent consequence of the material processing and preparation of the test specimens. Statistical methods are therefore necessary to establish fracture strengths under various test conditions. Strength data ( Table 3) ^[9] can be only of limited use in the design of ceramic components. This has encouraged a probabilistic approach to failure probability and assessment of critical flaw size. ^[10]

The service conditions under which engineering ceramics offer advantages tend to subject them to high temperature and high levels of abrasion. The present interest in engineering ceramics results mainly from their refractory properties ( Table 4), ability to retain mechanical strength at high temperature, and low density. For example, the inclusion of ceramic materials in reciprocating engines leads to greater fuel efficiency because of the higher operating temperatures and lower weight. ^[11]

Table 4. Engineering ceramics and melting points

Little is known about the behaviour of ceramics under various service conditions. The chemical degradation of ceramics with the 'dirty' fuels used in reciprocating engines and industrial gas turbines has to some extent handicapped their development. The 'cleaner' fuels used in aero-engines pose less of a problem.

Taking the chemical degradation of silicon nitride as an example; under normal conditions, the following reaction takes place:

Si 3N 4 + 3O 2 → 3SiO 2 + 2N 2 ↑.

The SiO 2 forms a passive coating on the Si 3N 4 inhibiting further oxidation, but if this coating is cracked or pitted, further degradation can take place.

The material properties and extreme operating conditions therefore dictate that critical flaw sizes are small. For many applications the maximum flaw dimension may be only 25µm.

Selection of NDT methods

The applicability of non-destructive tests to ceramic components can be assessed on the basis of four criteria; the nature and size of the defect, the properties of the ceramic material, the type of ceramic component and the test conditions ( Fig.3).

To define the nature of the defect is the important first step. As yet, the nature of what NDT engineers are required to find in ceramic components is poorly defined, and improved classification is necessary before NDT techniques can be developed fully.

The material properties will determine the penetration of X-rays and ultrasound and the levels of noise which may obscure defect images or signals.

Most fine engineering ceramics present few problems in this respect. Small components of complex shape are difficult to inspect with any NDT method, because the presence of geometric features may obscure defects. The problem is usually overcome by confining tests to specific critical areas of the component for which the geometric images or signals can be carefully monitored.

Finally, the conditions under which the test is carried out must be considered. A method which may be applicable in the laboratory may not fulfil its potential on the production line.

The conditions set out for selecting a new test method may include the following:

1. The test should be relevant. A test capable of detecting one type of defect may not necessarily detect another. Planar defects are a particular problem because their orientation may determine whether they are detected or not. A crack has to lie in a plane parallel to the X-ray beam, for example, to create an image on the radiograph.
2. The test should be precise. It must be able to give quantitative data on which to base simple accept/reject criteria, otherwise it is of no use in quality control.
3. The test should be reproducible. It may be the only way of communicating information about the severity of a defect and this may require comparisons with other tests.
4. The test should be accurate and reliable. If not, design margins have to become large, resulting in greater redundancy, less efficiency and greater cost.
5. The test should be practicable in the manufacturing environment, where it will be in the hands of less skilled operators.
6. The test should be cost effective.

Conventional NDT methods

Despite the small critical flaw size in some ceramic components, conventional NDT methods have a role to play in testing ceramic components with less exacting quality requirements.

Penetrant tests

Penetrant tests find widespread use in critical applications within the aerospace industry, for example on metal turbine blades. Unfortunately many ceramics are inherently porous. Defect indications may therefore lack contrast and penetrant residues may contaminate the component if post-cleaning is not thorough.

However, penetrant tests are suited for small components of complex geometry and can be automated for on-line inspection. Laser scanning techniques have been developed to examine test surfaces for fluorescent indications ^[12] and therefore dispense with the need for visual inspection.

Magnetic tests

These methods are not applicable to the current range of engineering ceramics, because the test material must be strongly ferromagnetic. There is some interest in using Barkhausen noise effects to examine substrates beneath ceramic coatings. ^[13]

Eddy current tests

Ceramic materials are in general non-conductive, although an electrically conductive sialon that can be machined by electric discharge methods has recently become available. ^[14]

Eddy current methods would be suitable for measuring the thickness of ceramic coatings on metals. Eddy current instruments can be made sensitive to the 'lift-off' effect, where a test signal is created as the coil is lifted from the surface. This effect is already used in many paint thickness gauges, where a resolution of the order of 0.001 mm is claimed.

Radiographic tests

The test procedures used on ceramics are different to those used on metals for three reasons.

1. Because they are made up of elements of low atomic number the X-ray absorption coefficient of ceramics is lower than that of metals. The image contrast improves when using low energy X-rays, typically less than 50kV.
2. The image contrast of inclusions is low, because they too are made of elements with a low atomic number.
3. The critical flaw size in ceramics may be of the order of 25µm and this is beyond the resolution capabilities of conventional radiography. Microradiography is then necessary and there are two methods:

Contact radiography

The first method uses a conventional single wall - single image technique with the object in contact with the radiographic film, but with a long film to tube focus distance. ^[15] This reduces the geometric unsharpness or penumbra to a limit, where definition of the image on the film is governed only by the energy of the X-rays and the graininess of the film. By using the radiographic films designed for lithography and spectrography, which have an ultra-fine grain size and an emulsion on one surface only, the images can be optically magnified 200 times for viewing and yet retain good image quality. However, there is some loss in image quality because of the large amounts of scattered radiation reaching the film.

Projection radiography

The second method places the object close to the X-ray tube and projects the image over a distance of typically 2m on to the radiographic film. ^[16] To reduce the geometric unsharpness to acceptable levels, the focal spot size in the X-ray tube is as small as 10-100µm in diameter. With a magnification of 40X, a lateral resolution in the image of 20µm is achievable.

The projection technique using a micro-focus X-ray tube is preferable to a contact technique using a conventional X-ray tube because there is less scattered radiation reaching the film ^[17] and this improves image quality. The method is suited for use with an image intensifier to capture a video image instead of using radiographic film. Digital image processing of the video image can improve the inherently poor defect contrast in ceramics. There is also the possibility of real-time imaging and scanning of the component.

In summary, radiography is sensitive to volumetric flaws in ceramics if the kV is low and the image can be magnified. The geometry of the component may make interpretation of the radiographic image difficult, if defect images are superimposed on images from geometric features. Unless real-time imaging is used, the method is suitable for off-line inspection only.

Ultrasonic tests

Ultrasonic tests offer the most promising methods for inspection of ceramics. Fine ceramics have a lower attenuation coefficient for ultrasound than metals. The high impedance mismatch between voids and the surrounding ceramic material ensures high reflectivity of incident ultrasound pulses, although planar defects must lie in a plane near the perpendicular to the sound beam axis if they are to be detected. To refract the sound beam into a ceramic accurately is more difficult than with metals. The high acoustic velocity in ceramics angles the refracted beam sharply away from the normal with only small changes in the angle of beam incidence.

Coupling a ceramic component to the ultrasonic probe is a problem if the ceramic is porous. Dry coupling techniques using a wheel probe have been used to inspect ceramic insulators. ^[18] However, the technique is too insensitive for engineering ceramics. Two ultrasonic test systems have recently been developed for inspecting ceramics: high frequency immersion testing and computerised ultrasonic tests.

High frequency immersion testing

An ultrasonic immersion testing system with a capability for testing ceramic turbine blades has been developed, using ultrasound at a frequency of 50MHz. Magnification is achieved by spacing the raster scans made by the probe across the test surface at only 2µm apart and then expanding the C-Scan image on a video screen. The C-Scan is a sound attenuation map of the test object created by monitoring the amplitude of a reference echo from a glass plate placed beneath the object. The C-Scan provides an image of the elastic properties of the object, which is affected by voids, cracks and other discontinuities. Problems with scanning over curved surfaces are overcome by placing the turbine blade on a computer controlled scanning table which has three axes of translational and one of rotational movement.

Computerised ultrasonic testing

A computerised ultrasonic test facility has recently been installed at Toyota Motors to inspect a ceramic pre-combustion chamber. ^[19] The system is totally automated. The components drop on to a turntable on which they rotate inside an assembly of ultrasonic probes ( Fig.4). The component and probe assembly are immersed in water to give an ultrasound coupling medium between transducers and component. As the component rotates, one set of probes moves axially down the cylindrical sides of the component and another set moves radially across the flat end surface.

The defects sought are pores on the external and internal surfaces of diameter greater than 50µm.

Five ultrasonic probes cover the critical areas. They propagate ultrasound with a frequency of 15MHz providing adequate sensitivity to detect small defects. Some are focused and others have collimated sound beams to improve spatial resolution and reduce echoes from geometric features in the component.

Control of the test sequence is by computer. The axial scans are made at 0.5mm increments. The data are taken from the A-Scan at each 1.6° angular displacement. A total of 19 350 data points is thereby collected in 8.8sec. For the radial scans, the increments are 0.25mm per revolution with a similar 1.6° angular displacement between readings of the A-Scan. This results in a further 21 725 data points acquired in 9.9sec.

The data from each of the five probes are processed and the results displayed as either a C-Scan image or a line scan. The information can be used for statistical quality control of the manufacturing process.

Summary

In general, the ultrasonic testing of ceramics is characterised by the use of test frequencies between 10 and 50MHz. The method is sensitive to most defect types, particularly planar ones, as long as the sound beam is incident along the normal to the defect plane. Material properties present few problems because of the low ultrasound attenuation coefficient. The geometry of the component can make ultrasound coupling difficult and it may not be possible to scan over complex surfaces. ^[20] Defect resolution is poor if the defect lies within a few microns of the surface. Computerised ultrasonic signal processing makes the method suitable for online inspection.

Acoustic emission tests

There is interest in the use of acoustic emission tests to evaluate the condition of plasma sprayed coatings. ^[21,22] The acoustic emissions are induced by cracking, deformation and phase changes and are readily propagated by ceramic components under proof test. Although there is considerable scope for developing the technique for ceramics, present applications have been confined to the laboratory, on simple flat specimens.

New test methods

Many new test techniques have been applied to ceramics in the laboratory: X-ray computerised tomography, acoustic microscopy, scanning laser acoustic microscopy, surface wave acoustics, thermal wave imaging, and photo-acoustic microscopy.

X-ray computerised tomography

A tomogram is constructed by rotating an object in front of an X-ray source with a highly collimated X-ray beam, and detecting the through transmitted radiation ( Fig.5). Changes in the attenuation of the beam create signals that are processed by a computer to give an image which becomes an attenuation map of the 'slice'. By moving the scanner up the axis of the object, further slices are imaged providing information in three dimensions. By using energy selective detectors that respond to low energy X-rays, high energy X-rays or charged particles, inclusions can be identified in the ceramic. ^[23]

Computerised tomography offers improved spatial resolution over conventional radiography and because of a lack of scattered radiation, achieves much higher contrast and definition in the image. Differences in object density as low as 0.025% can be detected and cracks as small as 10µm can be resolved. ^[24]

Despite these formidable levels of defect sensitivity, X-ray computerised tomography is unlikely to find widespread use in industry because of the cost which can be over £250 000. A simple instrument has been built at Surrey University for just £10 000, but it uses only one detector and a weak isotope source of radiation. Consequently, one tomogram can take 12hr to create.

Acoustic microscopy

Acoustic microscopes use ultrasound frequencies within the range 50MHz-1GHz. Acoustic microscopes operating at frequencies beyond this have been developed, but require liquid helium instead of water to couple the probe to the object ultrasonically. ^[25] The acoustic image is created by leaky Rayleigh waves which interfere with the normal reflected waves to create fringe patterns. The Rayleigh waves propagate along the test surface and therefore the microscope will detect cracks which are perpendicular to the surface ( Fig.6). At frequencies below 100MHz there is sensitivity to sub-surface flaws within a few microns of the test surface.

Scanning acoustic microscopes are available that create an acoustic image, similar to the C-Scans displayed in ultrasonic immersion tests, but at much higher magnification. ^[26]

Acoustic microscopes detect changes in the elastic properties of the test surface and these are usually more indicative of defect conditions than optical reflections.

Scanning laser acoustic microscopy

The scanning laser acoustic microscope (SLAM) uses a laser interferometer to detect ultrasound transmitted through the test sample. ^[27] The focused laser beam scans the ripple pattern produced on the test surface by continuous ultrasound with a frequency of 100-500MHz ( Fig.7). The amplitude of the surface waves is modified by the internal elastic properties of the ceramic and by voids and delaminations. Specimens have to be less than 5mm thickness and the scanning rate can be slow; for example 50min to cover a 30 x 6mm area. ^[28]

Surface acoustic wave technique

A laser beam may be used to generate ultrasound. ^[29] Laser generated ultrasound techniques have been used to inspect hot bar and billet where a non-contact method is needed. The ablation and thermal shock created by the pulsed laser set up a broad band acoustic pulse. Too much heat is generated for the technique to be non-destructive on a material at room temperature, so it has been refined to a surface acoustic wave (SAW) method. The pulsed laser beam is collimated through an axicon lens to give a 16mm diameter, 0.2mm wide annulus of heat on the specimen surface. An ultrasonic wave propagates towards the centre of the annulus, where it converges to give a higher amplitude displacement. This undulation can be detected by a laser interferometer.

Thermal wave imaging techniques

Thermal waves are found to be more sensitive to small changes in the lattice structure of materials than either electromagnetic or acoustic waves. ^[30] The technique is already used on metals and in the semiconductor industry to inspect doping levels in transistors and lattice defects in semiconductor crystals. The thermal waves have a short wavelength and are capable of resolving sub-micron size cracks. Thermal waves are difficult to image, however, as they dampen quickly, usually within one or two wavelengths.

Thermal waves can be propagated by pulsing the surface with a laser or electron beam and detecting the waves with an infrared interferometer. A thermal map of the surface is thus created. Thermal signals are complex, representing thermal wave contributions from different depths. They can be used in depth profiling by changing the laser pulse frequency and measuring the effect on the phase of the thermal signal. The method is only applicable to thin samples.

The method has been applied to the inspection of ceramic surface coatings to detect disbonds to the metal substrate. ^[31-33] Although successful on flat surfaces, it has yet to be evaluated on components of more complex geometry, where edge effects will decrease the sensitivity.

Photo-acoustic microscopy

A less expensive method of thermal wave imaging, that does not use expensive infrared cameras and interferometry, is the gas microphone cell. The mode of operation of a typical photo-acoustic imaging device is shown in Fig.8. ^[34] The sample is sealed inside the gas microphone cell and a chopped laser beam from a 4mW helium neon laser is focused to produce a 3µm spot on the sample surface, through a quartz window in the cell. The mirror deflects the laser beam in a raster scan across the sample.

The square wave light pulse from the chopped laser creates a wave-like change in the surface temperature of the sample. This creates a wave-like variation in gas pressure that is picked up by the microphone. The amplitude and phase information from the signal is digitised, processed by computer and output to the video.

The technique has been evaluated for testing sintered alpha silicon nitride turbine blades. ^[35] Flaws 75 x 30µm in dimension and 30µm below the surface were detected.

Discussion

As production methods become more rapid and ceramics enter more critical fields of application, quality control of their manufacture will become more important. As part of this quality control, NDT has a role to play. Indeed an American survey of technological development mentioned in a recent seminar on non-destructive evaluation (NDE) ^[36] suggested that of the top ten technologies identified for development, artificial intelligence was voted first, NDE second and fifth generation computing as third. Admittedly NDE extends beyond just the physical test methods. It includes an evaluation of all the parameters which affect component performance, both in the manufacturing processes and the service conditions.

The quality of engineering ceramics is generally susceptible to subtle and minute changes in chemistry and microstructure during processing and therefore NDT methods with a high level of defect sensitivity are needed. Conventional NDT methods cannot provide this level of sensitivity without considerable refinement of the techniques. This may be accomplished for example by using microfocus radiography or ultrasonic immersion tests at frequencies of 50MHz.

New test methods may provide the desired level of sensitivity but these are still under development and have yet to be proved in an industrial environment. Table 5 provides a comparison of the test methods discussed here.

Table 5 Comparison of test methods

The most versatile methods are ultrasonic. By increasing the test frequency, defects of small size can be detected, but with a loss of penetrating power of sound below the surface. SLAM overcomes this problem to some extent by using through transmitted ultrasound. Ultrasonic test methods are the most sensitive to cracks and delaminations but clusters of small defects can be difficult to resolve. The greatest drawback to ultrasonic methods, however, lies in the need to couple the transducers ultrasonically to the surface of the ceramic. This is most satisfactorily accomplished by immersing the test component in water.

Radiographic methods can only be used on bulk ceramics where there is access to both sides because the high X-ray absorption coefficient of metals would make imaging hybrid metal/ceramic components impossible. Cracks and delaminations will only be imaged on the radiograph if they lie parallel with the X-ray beam, while minimum detectable defect size will depend upon the thickness of the test component. Micro-radiography may be able to resolve spatially a dimension of 25µm, but a defect of through depth greater than this may only just attain adequate contrast for it to be visible in the radiographic image. Computerised X-ray tomography has high spatial and contrast resolution and is ideal for testing bulk ceramic components.

Thermal wave and photo-acoustic test methods show particular promise for testing ceramic coatings, but require considerable development work to engineer them for industrial use.

Acoustic emission test methods would provide a way of improving the sensitivity of current proof testing methods by providing information of sub-critical flaws.

Summary

It is the author's opinion that the refinement of current NDT practices should take precedence over the engineering of laboratory test methods, because only in extreme cases are the critical flaw sizes likely to be beyond the resolution capabilities of the improved NDT methods. In parallel with a need to develop NDT methods, there is a need for improved understanding of tolerable defect sizes for given service conditions.

Acknowledgements

This work was funded by the Research Members of The Welding Institute.

References