1. Literature Survey
The project has been started with a comprehensive literature survey on high temperature ultrasonic testing up to 1000°C to get the status of available and practically applicable methods for promoting the best high temperature ultrasonic probe development. With an up to date knowledge of the state of the art redevelopment of existing solutions can be avoided and new designs of prototype devices of ultrasonic probes (sensors) can be pursued. Evaluation of reports and literature, contact with manufacturers and users of equipment for ultrasonic testing in high temperature environment showed that at temperatures >200°C ultrasonic testing is based on electromagnetically generated ultrasound using electromagnetically excited and contactlessly working ultrasonic transducers (EMAT) [1-4] . No new paths of developing piezoelectric transducers could be found.
Concerning assembly of the components of the piezoelectric transducer (piezoelectric membrane and its protection layers, damping mass, probe's housing) three concepts to generate the acoustic coupling are in use:
Advantages and disadvantages of these concepts are as follows:
ad (a): Simple probe construction with a high surface finish requirement of the probes components to be coupled together. It is impossible in this case to use an acoustical l/2-membrane as ultrasonic probe's housing cap to suit the acoustical impedances between probe and specimen under test, since even with the aid of soft intermediate layers, e. g. gold foils, such membranes wouldn't withstand the necessary coupling pressure (Fig. 1).
Moreover even low bracings between different materials used in the ultrasonic probe can cause deformations and led to a complete failure of the probe. Theoretically air gaps with widths as low as 0.01 µm can cause almost total reflection of the ultrasonic pulse at the coupling boundary (Fig. 2), such that the ultrasonic pulse doesn't enter into the specimen under test.
ad (b): The necessary coupling pressure is low, so thin acoustical membranes can be used as probe's housing cap. Moreover in this case air gaps at boundaries are improbable to develop. However even now at the end of the project an optimal coupling fluid, concerning especially the corrosive properties of the coupling fluid, could not be made available for the temperature range under consideration.
ad (c): The optimal concept would be to presuppose that the thermal expansion coefficients of different materials could be fitted. But at this time the technology of soldering metal-ceramic joints is not yet sufficiently developed to get soldered joints stable in a large temperature range and for long periods of cyclic temperature loading.
Under aspects of technical feasibility and economy of the concept (b) has been estimated to be the most promising concept, and therefore has been followed.
2. Reference Set Up
For general purpose comparison and quantitative assessment of mechanical, thermal, and ultrasonic properties of the ultrasonic probe components in the temperature range up to 800°C, a high temperature reference set up has been designed and manufactured (Fig. 3).
In this set up the probe -except the housing-, is fixed and aligned. It consists of cup springs, damping body, guidance elements, piezo device, and terminating foils for the capacitive coupling to the electric circuitry. The gaps between piezo device, foils, and damping, are filled with glass solder. These components are located on top of a glassy ceramic as a specimen under test, which gives a back wall echo, whose amplitude and several other parameters can be traced and compared. The whole assembly is kept aligned and pressed together by means of vertically arranged threaded rods and tightly screwed up by horizontally directed mounting plates. It is designed to fit into an appropriate furnace.
Necessary adjustments, e. g. spring forces, can be determined and optimised during a heat cycle. Due to the open design, a possible breakdown of a single component can be determined, and the problem be corrected.
It allows for comparison of different damping materials, different kinds of coupling means, different kinds of means for generating coupling pressure, and relative quantitative evaluation of the sensitivity of piezo materials under stable and traceable conditions.
3. Ultrasonic Probe Components
3.1 Pressure springs
3.1.1 Cup Springs
Since for ultrasonic fluid coupling between the probe's components and the use of a l/2-membrane as an acoustical terminator a defined pressure is needed, some elements are needed which generate the necessary pressure. The Nimonic 90 cup springs which have been foreseen for this purpose in a first attempt show long-term stability only up to 600°C. Inconel and Nimonic 90 cup springs turned out to be unsuitable in the temperature range between 600 and 800°C. At 700°C Nimonic 90 cup springs showed a continuous decrease of the spring tension at a rate of 0.7 N/h. Therefore their service life will be limited to about 100 h. Service life could be increased by increasing the diameter of the cup springs. But such springs need special design and would be very expensive. So alternative devices for pressure generation have been investigated.
3.1.2 Metal Bellow
As one of the alternative solutions, the principle of temperature controlled pneumatic pressure has been selected. It can be shown theoretically, that an air column of 20mm height and 20mm diameter (Volume = 628mm³) generates a spring tension of 115 N at 800°C, which would be sufficient for ultrasonic coupling, since the couplants at that temperature have low viscosities.
The temperature controlled pneumatic pressure will be generated using vacuum-tight and temperature-stable metal-bellows from high-vacuum technology. Metal-bellows furthermore have the advantage that they allow for pre-pressurising the device at room-temperature (Fig. 4).
During the course of experiments it turned out that chemically inactive Argon had to be used as filler gas in double-walled metal bellows [11] to avoid oxidation of the bellow material by the enclosed air oxygen above 500°C. It has been calculated that 1 bar argon filled metal bellows should produce pressures high enough for ultrasonically coupling the probe's components. Adequately experiments revealed that under both conditions of short and long-term basis 1 bar argon filled metal bellows met the requirements (Fig. 5).
3.1.3 Ceramic Springs
It could be shown theoretically that ceramic springs for the intended purpose of ultrasonic coupling within the ultrasonic probe are possible to be fabricated [12] .
Investigations on ceramic springs have been started with a ceramic spring of 46 mm diameter and 23 mm height. During 257 hours 8 temperature cycles up to 800°C have been performed with a spring tension of 50 N and a spring deflection of 3.5 mm. During heating up the spring tension which had been adjusted to 50 N at room temperature decreased to 42 N. This value of the spring tension than remained constant (Fig. 6). Fig. 7 shows the stress/strain diagram of the ceramic spring before and after the first temperature cycle demonstrating the high performance and stability of the material. Further tests with four ceramic springs with sizes suitable for the ultrasonic probe finally proved that they are fit for the purpose.
3.2 Damping Bodies
In ultrasonic inspection, for good axial resolution it is desirable to use short pulses. This is achieved either by electrical means or by mechanical damping of the vibration with a damping mass on the back of the piezo-device. Since high axial resolution implies high damping and this again involves a sacrifice on sensitivity, for best ultrasonic testing results a good compromise between axial resolution and sensitivity had to be found.
In common ambient temperature applications a mixture of for instance epoxy resin with various metal powders such as aluminium, lead etc. is used for damping, but is not applicable in the temperature range of interest. Hence all sorts of different materials and arrangements had been taken under consideration and tested for their gain as damping mass. An overview of tested materials is given in table 1. It includes, to name only a few, a soft ceramic (no.1), sintered metals (2-3), carbon (4-5) and different cuts from austenitic welds. All these materials show certain aspects, which may support the idea of a good choice. But, many of them on the contrary have also exhibited properties in the experiments which could not be tolerated, and so, only a few materials have been retained for possible alternatives.
Further requirements which are to be accounted for include:
It was found that the impedance of the damping mass is allowed to have relatively low values (well below 5*10exp.6Kg/m²s) if some means of an acoustic transformation can be provided. Also, the damping properties of a material can be increased by up to 30 dB depending on the design and shape of the piece.
The test set-up for evaluation of ultrasonic properties of damping masses is given in Fig. 8 in principle. As a coupling medium in the test set-up silicon oil was used. The back wall echo from the polyimide-block is displayed on the flaw detector screen. Fig. 9 gives examples of a bad (left, austenitic weld) and a good (right, Muratherm) damping mass. It shows the screen snaps of the oscilloscope with the back wall echo at 38 % horizontal width in both cases. It should be noted that the disturbances to the left of the distinctive back wall echo in part (b) at 3.8 divisions, are due to radial vibrations of the piezo material, and are not disturbances out of the damping mass. In contrast, to the bad damping body employed, disturbing echoes either before or after the back wall echo would make the inspection difficult.
In extensive investigations many materials have proved to be of no use, such as austenitic weld, silicon carbide, passivated carbon, pulverised sintered metal, to name only a few. However, two materials are available, a ceramic with trade name Kager 9900, and a mica based material named Muratherm, which match the necessary requirements as could be seen so far. One problem of using ceramic as damping body is porosity, which may result in soaking fluid coupling into the ceramic. To prevent this, the ceramic may be coated with a thin glass film that is stable up to 800°C. CERDEC has succeeded in coating the Kager 9900 ceramic with a glass coating that matches the requirement of impermeability of the surface.
3.3 Coupling Materials
3.3.1 Glass Solder
During development of high temperature ultrasonic probes, the central problem is to solve the ultrasonic coupling between the probe's components, such as piezoceramics, damping mass, acoustic matching layers, wear plate etc.. It is essential for the efficiency of the probe. After the deciding to apply fluid coupling a suitable "fluid" for the temperature range under discussion had to be found.
Choices had to be made and balanced with respect to, state of the medium, chemical reactivity and compatibility, diffusion, evaporation, thermal expansion, mechanical behaviour under solidification etc. Low melting metals had to be excluded because short-circuits may be caused by them. Therefore low melting glasses, so-called glass solders have been chosen.
The requirements for the glass solder are:
Broad parametric studies have confirmed a glass solder named "NaPoLi", consisting of sodium oxide, phosphor oxide, and lithium oxide, to be relatively best suited for the several needs. This is mainly due to its low softening temperature (Fig. 10), and its thermal expansion coefficient which is sufficiently fitting to that of the piezoelectric transducer membrane.
In the course of continued experiments with that glass solder it revealed however that at higher temperatures the glass solder reacts chemically with the probe's components yielding FePO3. To avoid this corrosion all parts coming into contact with NaPoLi had to be protected by a coating.
New non-alkaline powdered glass solders LG-K1 and LG-K2 [13] proved to be much less chemically reactive than the glass solder NaPoLi but they were thermally stable only up to 600°C.
Variation of glass solder compositions during continuation of experiments and investigation of new mixtures showed that the glass solders LG-K15 and LG-K16 were fit-for-purpose. They were thermally stable at 800°C and revealed only low chemical reactivity against the SiO2 protection coating. Compared to NaPoLi viscosity of the new coupling medium below 600°C is higher, the temperature at which softening occurs is also higher.
Therefore NaPoLi could be used as coupling medium in the temperature range up to 500°C and LG-K16 in the temperature range between 400°C and 800°C this being solely a question of costs.
One more technical problem to be solved was to feed the glass solder during assembling the ultrasonic probe such that a layer of glass solder of uniform thickness is generated, otherwise the piezoelectric membrane might break.
It was solved by a mask, 0.2 mm thick, having an inner diameter of 10mm which was used to take up a suspension of powdered glass solder in alcohol. After evaporation of the alcohol the mask is removed leaving a smooth 0.2 mm thick and 10 mm diameter layer of glass solder enabling without any difficulty the further assembly of the ultrasonic probe.
3.3.2 Protection Coating
Since "NaPoLi" corrodes all metals except of platinum and gold (e. g. a platinum(95%)/gold(5%)-alloy is chemically resistant), all components of the ultrasonic probe which come into contact with this coupling medium must be protected by coating.
To find out experimentally a coating layer fit-for-purpose disc shaped austenitic test specimens with ø 20 mm have been used. The specimens have been typically prepared as follows:
The tests have been started by coating the test specimens by sputtering with Titanium Nitride (TiN) [14] . The coating layer thickness ranged between 1.2 and 3 mm. The specimens were then exposed to temperatures up to 800°C.
It turned out that after about 50 h at 600°C the glass solder on the open specimens had solidified, while the glass solder on the oxygen protected specimens had retained fluid even after 80 h. These results were independent of the coating layer thickness.
However, after increasing the temperature to 800°C, the glass solder solidified also on the oxygen protected specimens after 1 h. Successive experiments supported these results, i. e. with oxygen protection the glass solder appears to be stable with TiN-coating up to 600°C, above 600°C it solidifies after a short period of time.
This also holds good for Ti/TiN- coatings, and for Zirconium Nitride (ZrN) coatings.
Investigations of Titanium-Aluminium-Nitride-(TiAlN)-coatings which are said to be stable up to temperatures of 800°C, yielded no improvements using stacked specimens, since only after 40 h at 600°C the glass solder had solidified.
Further investigations with open specimens at 600°C left the glass solder fluid up to 90 h, when the experiment was terminated. It was continued then with two specimens stacked and temperature increased. The glass solder then turned out to solidify in a very short time.
Experiments with Zirconium Oxide (ZrO2) layers too failed to give satisfactory results.
After extensive enquiries into nitride and oxide thin layer properties it was concluded [14, 15, 17] that under cyclic temperature load these coatings, mainly nitride thin layers, are stable only up to 600 or 700°C. However this depends upon the coating process.
All layers investigated up to now have been deposited by physical vapor deposition (PVD-process). It has been proposed to use plasma enhanced-chemical vapor deposition (PE-CVD-process). During this process the layer is deposited on the specimen by a plasma-aided chemical reaction, which might have certain advantages compared to the PVD-process [15] .
Specimens coated by the PE-CVD-process with SiO2-layers, which are said to be stable up to 800°C, have been examined. Especially the SiOx-layer even at 800°C didn't show any alterations. But bringing NaPoLi in contact with the layers, again a chemical reaction occurred: after 40 h at 700°C the glass solder had solidified (SiNx), in an other case solid precipitations (SiOx) could be observed in the fluid coupling medium.
To overcome these difficulties a multilayer system has been tried out, e. g. an oxide-layer [16] as diffusion barrier coated by a platinum layer. However tests with specimens with SiOx - layers [15] as diffusion barriers and 0.3 µm thick Pt-layer [17] as protection coatings against the NaPoLi-couplant revealed, that the platinum coating was not thick enough and therefore porous, since it had been detached from the SiOx - layer under the effect of NaPoLi-couplant, while a 20 µm thick Pt-foil covered with NaPoLi during the same experiment didn't exhibit any changes at all.
Those specimens used for reference having had no contact with NaPoLi have been stable even at 800°C. Therefore the platinum layer which is 0.3 µm thick has been attempted to be enhanced to 20 µm galvanically.
However 60% of the specimens after sputtering revealed adhesion defects between platinum and SiO2 layer. Specimens selected by visual inspection have been temperature loaded at a rate of 100°/h. Again at 800°C two specimens revealed adhesion defects between platinum and SiO2 layer. The remaining 3 specimens have been loaded with NaPoLi couplant and again have been temperature loaded. Now again slight adhesion defects could be observed but, what is most important is that no corrosion attack by the NaPoLi couplant is observed.
The results of these experiments have been evaluated [14, 15, 17] with the aim to develop measures to improve adhesion between SiO2 and platinum layers, e. g. by introducing an adhesive agent such as a molybdenum layer. But, however, this problem which is unsolved up to now is going to be settled in the near future by BAM. In the mean time alternatively different coupling media for temperature ranges below 600°C and above that temperature will be used, see section 3.3.1.
3.4 Signal-Cable
After almost finishing the development of the ultrasonic probe, temperature resistant cables had to be found. Since the beginning of the project manufacturers appeared on the market who offer coaxial cables with one and two helixes, with outer diameters between 1 and 6 mm, and fit for temperatures up to 1200°C. These cables have been tested with the ultrasonic high temperature probes: DC resistance and capacity of the cable have been measured as a function of temperature up to 860°C since these quantities influence amplitude and form of the ultrasonic signal.
It turned out that all above cables are well suited. Cables with diameters up to 1.5 mm are preferred, because they are in addition relatively flexible. The DC resistance of a cable with 1.5 mm diameter at 800°C is more than 500 kW (Fig. 11) and has scarcely any influence either on the signal amplitude or on the form of the signal in the whole temperature range up to 800°C.
Since furthermore the cable manufacturers offer a heat resistant wall let-through being pressure resistant up to 500 bar, the ultrasonic probe design is simplified.
3.5 Ultrasonic probe's housing
Investigations on the problem of objectionable buckling of the l/2-membrane, which is closing the front of the ultrasonic probe's housing and matching the acoustic impedance, under load of coupling pressure have been performed with an Inconel 600 membrane in an austenite probe's housing. Up to 2.5-fold of the working pressure no buckling of the l/2-membrane greater than 10 micrometer could be observed, neither at room temperature nor at 800°C.
The welding of the probe's housing closing membrane is especially important, since a circular seam of small diameter (about 1 inch) is required which has to be welded vacuum-tight. Therefore welding tests with the laser-welding unit had to be performed on a large number of specimens to find out the optimal welding parameters [19] .
4. Prototype Ultrasonic Probe
The knowledge gained from investigations on the different ultrasonic probe's components has been compiled and matched to give the prototype probe shown in Fig. 12. It is designed for a circular piezo ceramic of 20 mm diameter, because this size has proved to be very effective. Smaller transducers transmit too little energy into the specimen under test at the high temperature range for which the probe is designed for.
Since the functioning of the combination of the probe's components had to be tested, the first prototype composed the following:
The experiments for probe characterisation utilised a white glassy ceramic as test specimen, that made it easier to produce a back wall echo as a test reflection indication, due to its low acoustic absorption. The probe was electrically connected to a flaw detector, and a sequence of back wall echoes were displayed. The experiments showed that at a temperature as low as 300 C the coupling began to emerge, but was subjected to changes, i. e. the height of the back wall echo of the glass ceramic changed during the increase of temperature. These changes ceased at around 470 C. At 500 C the probe showed its best sensitivity (Fig. 13), which decreased slightly up to 550 C, dropping down above this value which was 30 dB lower at 625 C. Simultaneously the pulse shape worsened severely. Frequency analysis showed a strong component at 8.2 MHz, which is probably due to the non-linear behaviour of the LiNbO3 crystal.
After cooling down it was found that the temperature sensor was displaced during the measurement, and had finally been located in the fused silica sand at the bottom of the furnace. This suggests that the measured temperatures were a bit too low, i. e. the temperature inside the probe will have been higher. This could account for the severe loss of sensitivity at 600C, where LiNbO3 single crystal should still work quite properly.
A second run with the same components yielded the similar results, but the coupling took place only at 500 C. Nevertheless, this is easy to understand as due to the use of austenitic foils a chemical reaction with the NaPoLi took place, which resulted in the formation of FePO3.
At 600 C the sensitivity reached its highest value, but dropped by 8 dB (factor 2.5) to 650 C. After a consecutive long term experiment over 18 hours at this temperature, the sensitivity dropped by another 18 dB (factor 8), which was accompanied by the same worsened pulse shape as mentioned above.
Since the prototype ultrasonic probe performed well the final ultrasonic probe shown in
Fig. 12 was designed using the following components:
The ultrasonic probe's components are acoustically coupled by high pressure (Dry coupling) [5] .
The ultrasonic probe's components are acoustically coupled using a fluid (Fluid coupling) [6,7] .
The ultrasonic probe's components are acoustically coupled by soldering (Solid coupling) [8-10] .
Fig. 1: Amplification as a function of coupling pressure
during dry coupling at constant echo height 7KB
Fig. 3: Design and parts list of the reference set up 25KB
Fig. 4: Calculation of temperature
controlled pneumatic pressure 6KB
Fig. 5: Spring tension during temperature
for a spring bellow 6KB
Fig. 6: Spring tension during temperature
for a ceramic spring 6KB
Fig. 7: Relation between spring deflection and
applied force for a ceramic spring
5KB
* application possible when acoustic impedances matchedNo.
Material Z
[10exp.6 Kg/(m² s)]
1
Kager 9900
14.67
2
Sintered metal IA
15.95
3
Sintered metal ZFK
21.09
4
Carbon power pick up perp.
6.13
5
Carbon power pick up par.
5.68
6
Aust. Weld X
42.46
7
Aust. Weld V
39.16
8
Muratherm
3.1 *
Fig. 10: Temperature dependent viscosity of glass solders 6KB
In a first series of experiments
All specimens have been provided with a SiO2 layer acting as diffusion barrier.
Fig. 11: DC resistance R vs temperature for a
new high - temperature cable Ø 1,5 mm 8KB
Fig. 12: Design and parts list of the new prototype ultrasonic probe 10KB
Fig. 2: Reflection factor at the interface between
LiNbO3 and steel upon dry coupling
(gap thickness = 0.01µm)
Fig. 8: Setup for determining the properties of
damping bodies by comparison their
screen displays on a conventional flaw detector 
Fig. 9: Left a bad example for a damping body (austenitic weld),
right a good one (Muratherm). Note that the disturbances after the
sending pulsein (b) are due to radial vibrations in the piezo material
and do notstem from the damping
Fig. 13: A-scan of a backwall echo at
500 °C from a 20 mm thick specimen
( highest sensitivity at the amplifier )
Fig. 14: Necessary increase of amplifier gain
with temperature for constant echoheight
of piezoceramic LNN 15 1100 HIP 69 #2