Bundesanstalt für Materialforschung und -prüfung

International Symposium (NDT-CE 2003)

Non-Destructive Testing in Civil Engineering 2003
Start > Contributions >Posters > Tendon Ducts: Print

Eddy Current Testing at High Temperatures for Controlling Heat Treatment Processes

Jochen Vetterlein, Foundation Institute for Materials Science, Bremen, Germany
Heinrich Klümper-Westkamp, Foundation Institute for Materials Science, Bremen, Germany
Thomas Hirsch, Foundation Institute for Materials Science, Bremen, Germany
Peter Mayr, Foundation Institute for Materials Science, Bremen, Germany

1 Introduction

Increasing quality demands have made a reliable quality control getting more and more important for manufacturing and processing industries. Especially if 100% testing is required, the advantages of the eddy current technique are used by a wide range of applications to ensure constant quality: eddy current measurements require minimal testing time, offer the possibility to measure without human intervention and are non-destructive [1].

For heat treating processes, the method is used in various applications. Aircraft industry for example uses the eddy current technique to measure the conductivity of heat treated aluminium components to check hardness and tensile strength [2]. To ensure that aluminium plates have been properly heat treated and exhibit a satisfactory degree of homogeneity, systems are developed that automatically measure conductivity on the top and the bottom of the plate as sensors scan across [3]. Multi-frequency systems make the control of ferromagnetic workpieces possible, for example to check material structure or cracks [4-6].

For all of these test procedures, the measurements are performed after the heat treatment has taken place, as there are no systems, which can be used at the heat treatment temperatures and atmospheres. Measurements during the treatment would offer much more information about the process, because not only the final state, but also all the changes during heat treatment could be monitored continuously. This additional information makes the process saver and can be used to adapt the process more easily, if changes become necessary.

To perform measurements under such difficult conditions at high temperatures the eddy current technique shows to be the most suitable NDT system because of its easy handling and the simple build-up of the sensor.

The article describes the development of an eddy current measuring system to characterise micro-structural changes during heat treatment. It can be used at temperatures up to 500°C. The construction of the sensor and the method of data analysis is illustrated. The system was used to characterise the aging process of the Al Zn Mg Cu alloy EN AW 7075. The correlation between micro-structural changes and the eddy current signal are discussed.

2 Eddy Current Measurements

2.1 How to detect the heat treatment state with eddy-current testing
Eddy currents can be used for material characterisation because the sensor signal is dependent on two material properties: the electric conductivity and the permeability. These parameters are often connected strongly to technological relevant quantities, e.g. hardness. Thus the use of eddy currents in a heat treatment process makes it possible to monitor the quantities characterising the heat treatment state continuously and in-situ.

2.2 Build up of the measuring system
The eddy current measuring system consists of three components:

  • a heat resistant eddy current sensor
  • an impedance analyser Solartron 1260A
  • a standard PC to save and to evaluate the data.

The impedance analyser supplies a sinusoidal current to the transmitting coil. At the same time the voltage of the receiving coil is measured and the complex impedance (inductance and resistance) is computed.

The sensor is a radial symmetrical two-coil-system. The receiving coil is placed between the sample and the transmitting coil where the strength of the secondary field reaches the highest values. The sensor has to be made of heat resistant materials, because it will be used inside the furnace. The winding wire of the coils is a fibreglass-insulated cooper wire. The coil former is made of machinable ceramic. The coils are encapsulated in a ceramic tube filled with ceramic sealant (Fig. 1).

Simulations were performed [7] varying the sensor's geometry and frequency to maximise its sensibility towards conductivity changes. The properties that were varied are illustrated in Fig. 2.


Fig 1:
The eddy current sensor with housing (left) and schematic drawing (right).
Fig 2: Illustration of the parameters that where varied for optimising the sensor.

The resolution of the measuring system is not only dependent on the sensor. The precision of the impedance analyser affects the resolution of the system as well. The calculations of the resolution of the sensor were made assuming an error of the absolute impedance and the phase shift of ± 0,1% and ± 0,1°, respectively. Using an analyser of a higher or lower quality class will change the absolute values of the calculated resolutions, but the general dependencies will stay the same.

Additionally, it has to be considered that the precision of an analyser is only achieved if the amplitude of the sensor signal exceeds a specific value. For example, the analyser used for the investigations demands a minimum absolute impedance of 5W .

The simulations were performed keeping the winding density of the coils constant as this parameter is fixed if a winding wire of a specific diameter is chosen. As well, the outer diameter of the sensor was not varied, because it should be chosen according to the geometry of the part to be evaluated. The results of the simulations are presented in Table 2.

variation resolution signal amplitude
decreasing the length of receiving coilincreasingdecreasing
decreasing the inner radius of transmitting or receiving coilsonly small effect increasing
increasing the length of the transmitting coil only small effectincreasing
decreasing distance between transmitting coil and receiving coilonly small effect increasing
increasing frequencyBest resolution is achieved at a specific frequency f (opt) that is proportional to the conductivity s of the workpiece.f (opt) (aluminium alloy, s "18MS/m) "5kHzincreasing
Table 2: Influence of the sensor's geometry and frequency on the resolution and signal amplitude.

The construction of the sensor is based on these dependencies. The amplitude of the sensor signal is maximised by choosing a transmitting coil of high length, a minimal inner radius of the transmitting and receiving coil and a minimal distance between transmitting and receiving coil. The length of the receiving coil is a compromise. On the one hand a minimal coil length results in the best resolution, on the other hand the coil has to be long enough so that the required signal amplitude is achieved.

As the penetration depth of the eddy currents decrease with increasing frequency this parameter is commonly used to choose the depth to be inspected. For example, the inspection of thin sheets or of coatings is usually done at high frequencies. Thus it is usually not possible to vary the frequency to optimise the sensor's resolution. The measurements presented where performed at a frequency of 10 kHz which is close to the optimum frequency of 5kHz. Simulations showed that this frequency is high enough. The finite thickness of the 10 mm samples will not interfere the measuring results.

2.3 Elimination of unwanted variables
Eddy currents are sensitive to a wide variety of parameters. This makes it necessary to eliminate the effects of unwanted influences to characterise the parameters of interest [8-13]. For material characterisation of non-ferromagnetic part as aluminium the parameter of interest is the electrical conductivity. The following methods were used to eliminate the unwanted variables:

  • The lift-off effect was eliminated by data analysis. This is possible, because a variation of the lift-off will shift the sensor signal (plotted in the impedance plane) in an other direction as a variation of the conductivity.
  • The temperature of the sensor influences the sensor signal mainly because of the ohmic resistance of the winding wire. This effect can be eliminated using a system with separate transmitting and receiving coil. Measurements should be done far below the resonance frequency.
  • The conductivity of a material is highly dependent on its temperature. Thus a change of the sample temperature will have a great effect on the sensor signal. Therefore, the sample temperature is measured simultaneously to the eddy current measurements. Assuming a constant temperature coefficient of the resitivity, the conductivities can be corrected. Another method to eliminate the effect of the sample temperature is the use of a reference sample. However, this method can only work, if the dependency between temperature and conductivity is the same for reference sample and the conductivity of the reference sample will stay constant during the measurement at heat treatment temperature.
  • The geometry of the workpiece will influence the sensor signal as well. However, the goal was to develop a sensor to determinate structural changes due to a heat treatment. This is the reason why the geometry was not varied and the shape of the samples was kept as simple as possible (see Chapter 4).

2.4 Calculation of the conductivity
In most applications the sensor signal is directly correlated to the parameters of interest. The disadvantage of this method is that every exchange of the sensor or change of frequency will demand a new correlation. Additionally, it is difficult to compare the results of two measuring systems.

To overcome these difficulties, a method was found to calculate the conductivity from the sensor signal. This makes a correlation between the conductivity and the parameters of interest independent of the frequency and the sensor.

Dodd et al. [14] found, that the normalized complex sensor impedance Z can be written asa function of the product of the sensor's frequency w and the conductivity of the workpiece s :

(1)

Assume a reference sample of known conductivity s0 will provoke a sensor impedance Z(s0,w 0) at the sensor frequency w0. A workpiece of an unknown conductivity sw will cause a sensor impedance Z(sw,w) at a sensor frequency w. According to eq. 1 the frequency w can be varied in such a way that Z(sw, w )=Z(s0,w0). In this case the unknown conductivity it given by:

(2)

However, an adjustment of the frequency to get the conductivity would be an inconvenient practice, therefore another method was used performing two measurements at the frequencies w0 and w1 provoking the impedances Z(sw,w0) and Z(sw,w1), respectively. According to eq. 1 a hypothetical measurement at the frequency w 0 of a workpiece with a conductivity

(3)

will provoke the impedance

(4)

Assuming small conductivity changes sw-sw' and s0-sw the impedance change will be proportional to the conductivity change:

(5)

Using eq. 3 and 4 you will get:

(6)

(7)

By this way the absolute value of the unknown conductivity s w can be calculated with only a single calibration measurement Z(s0,w0).

3 Age-hardening aluminium alloys

Age-hardening aluminium alloys contain at least one element that is soluble in the aluminium matrix at elevated temperature, but only to a small degree at room temperature. Heat treatment of such alloys is a three-step process:

  1. Solution heat treatment: The workpiece is annealed at a temperature high enough to dissolve the alloying elements to achieve a solid solution.
  2. Quenching: Rapidly cooling (usually to a temperature near room temperature) results in a supersaturated solution.
  3. Aging: The aging step involves the formation of finely dispersed precipitates increasing the strength of the alloy. This can be done either at room temperature (natural aging) or at elevated temperature (artificial aging).

The commercial age-hardening alloys are the Al Cu alloys (2xxx series), the Al Mg Si alloys (6xxx series) and the Al Zn Mg alloys (7xxx series). The precipitating equilibrium phases of these alloys are Al2Cu, Mg2Si and MgZn2, respectively.

Highest strengths are achieved by age-hardening alloys. This is the reason why these systems are predominantly used in light constructions, for example in the aircraft industry.

4 Experimental

The age hardening process was investigated using the Al Zn Mg Cu alloy EN AW 7075. The heat treatment state was T651 (solution heat treated, quenched, stress relieved by stretching, artificially aged). The base material was delivered as a plate with a thickness of 10 mm, which was cut to sample size (70 x 70 x 10 mm3). More details are given in [15].

The samples were solution heat treated for one hour at 456°C using an air circulation furnace and quenched in water at room temperature. Two typical aging treatments [16] were used:

  • single-step process: artificial aging at 120°C
  • two-step process: artificial aging at 100°C for four hours, followed by a heat treatment at 160°C

Artificial aging at 120°C is used to achieve maximum strength [17]. The two-step process was developed to improve the resistance to stress corrosion cracking [18]. As an additional advantage of the two-step aging, mechanical properties can be improved, if the critical quenching rate of the alloy cannot be reached, for example for workpieces of large dimensions.

In contrast to the Al Cu alloys and the Al Mg Si alloys, naturally aging is commercially not used for Al Zn Mg alloys, because no stable mechanical properties can be achieved. Strengths increase over a period of many years [17]. Nevertheless, a set of samples was aged at room temperature to get a better understanding for the metallurgical reactions.

The conductivity measurements were performed in-situ and continuously during the heat treatment using the sensor described in chapter 2. To correlate the conductivity with the heat treatment state a set of samples was made for each aging variant interrupting the heat treatment after a specific aging time by fast cooling to room temperature. The heat treatment state was then characterised by hardness testing and TEM micrographs.

5 Results

The propagation of the conductivity is presented in Fig. 3. A detailed view of the first 5 hours of aging in shown in Fig. 4. During natural aging the conductivity decreases during the hole process, while artificial aging at 120°C is associated with an increase of the conductivity.

Fig 3: Electrical conductivity of alloy EN AW 7075 during age hardening for the investigated heat treatments. Measurements were performed in-situ.

Fig 4: Electrical conductivity of alloy EN AW 7075 during the first 5 h of age hardening for the investigated heat treatments. Measurements were performed in-situ.

The conductivity of the two-step aging variant reaches a minimum after 40 min followed by a slight rise. The second aging step causes a fast increase of the conductivity.

The two-step aging variant shows a peak hardness after aging for 8 h (Fig. 5). After long aging times the TEM micrographs show the appearance of coarse particles (Fig. 7). Using single step aging, a similar hardness is reached after 24 h but there is no hardness peak. After reaching maximum hardness there is no significant change in the strength and the size of the particles (Fig. 6) differs only slightly.

Fig 5: Vicker's Hardness of alloy EN AW 7075 during age hardening for the investigated heat treatments. The measurements were performed at room temperature.

The hardness of the naturally aged samples increases in a similar way as the single-step aged samples, but at a much slower rate. A constant hardness is not reached. Even after an aging time of 1000 h the hardness is still increasing.

Fig 6: Bright field TEM micrographs of alloy EN AW 7075 after age hardening at 120°C; left: aged for 25h; right: aged for 280h.
Fig 7: Bright field TEM micrographs of alloy EN AW 7075 after two-step age hardening; left: aged for 4h at 100°C + 1h at 160°C; right: aged for 4h at 100°C + 160h at 160°C.

6 Discussion

The quenching of the aluminium alloy results in a supersaturated matrix. During aging the decomposition of the alloying elements will take place. The aging sequence is generally stated as [19-22]:

supersaturated matrix
GP Zones (coherent)
h '-phase ("MgZn2, semi coherent)
h -phase (MgZn2, noncoherent).

The variation of the electrical conductivity results from a number of contributions [23, 24] that are summarised in Table 3. Depending on witch of these processes are dominating, a rise or a fall of the conductivity is observed during aging (compare Fig. 3 and 4).

Natural Aging
At temperatures below 60°C (thus especially at room temperature) the decomposition takes place mainly by the formation of GP zones [20]. Two competing processes occur. The purification of the matrix tends to increase the conductivity. Simultaneously, there is a more significant decrease in conductivity due to the formation of GP zones. It is generally stated that for natural aging the formation of GP zones is the reason for the increase of strength. Thus the cause for the decrease of the conductivity and the increase of hardness is the same. This explains the correlation between the hardness and the conductivity and why the rate of decrease in conductivity is similar to the rate of increase in hardness (Fig. 3 and 5).

processes that tend to decrease the conductivity processes that tend to increase the conductivity
Formation of GP zones with dimensions comparable with the electron wavelength purification of the aluminium matrix
Dissolution of GP zones und conversation of GP zones in (meta)stable phases
Increase of boundary area due to coarsening of precipitates
vanishing of quenched-in vacancies
Decrease of boundary area due to growth of GP zones at a constant volume fraction of the zones
Table 3: Processes that occur during age hardening. Some of them contribute to a reduction in conductivity, others lead to a rise in conductivity.

Single-step aging at 120°C
The aging temperature of 120°C exceeds the solvus temperature of GP zones. Thus the decomposition will not start with the formation of GP zones, but with the precipitation of the h' and the h phase [21, 22]. Thus, the mechanism that causes the decrease in conductivity for the naturally aged sample will not occur for this heat treatment. Instead, GP zones that form after quenching before reaching the aging temperature as well as condensed quenched-in vacancies may promote the nucleation of the h' and h phase [21]. Simultaneously to the nucleation process a growth of the precipitations will occur. Both mechanisms will cause a purification of the aluminium matrix increasing the conductivity as it can be seen in Fig. 3.

After aging for about 10 hours the hardness keeps almost constant while the conductivity is still rising. During this period neither the processes that strengthen the alloy nor the processes that soften the material are dominating. The following strengthening processes may occur:

  • Increasing precipitations density
  • Particle growth to an average particle size reaching the ideal strengthening diameter.

Simultaneously the following mechanisms cause a decrease in strength:

  • Transformation of GP zones into h' precipitations and transformation of h' into h precipitations lowers the strength due to the increasing coherence of the particles.
  • Coarsening produces particles that exceed the ideal strengthening diameter.

As it can be seen in Table 3 the strengthening processes as well as the softening processes will tend to increase the conductivity. This is the reason why the conductivity is still raising even after the hardness has reached a constant level.

Two-step aging at 100°C and 160°C
At aging temperatures of 100°C the precipitation starts with the formation of GP I and GP II zones and ends with the appearance of GP II zones and h' particles [20]. The h ' phase will not form until at least 10 hours of aging. So during the first step of the heat treatment only GP zones are to be expected. During the first aging step there is contrary to the natural aging process only a slight decrease in conductivity during the first 40 min. Afterwards the conductivity is increasing slowly although for both aging variants GP zones are formed. The reason may be the smaller diameter of the GP zones that form at room temperature. In addition, at a temperature of 100°C GP II zones are expected whose diameter is about 80% greater than the diameter of the GP I zones. This diameter may exceed the diameter that is ideal for electron scattering. Furthermore, by a formation of larger zones the aluminium matrix will be purified to a higher degree, so that the conductivity rises.

The second aging step causes a prompt increase of conductivity. It can be assumed that the mechanisms that causes this rising are the same as for the single step temper: the purification of the matrix due to the transformation of GP zone into h' und h particles and due to the growth of the particles. The fact that these processes are controlled by diffusion explains why the rise of conductivity occurs mach faster at 160°C than during the single-step aging at 120°C.

The peak hardness is reached after 8 hours. Contrary to the singe step aging the hardness does not keep constant but decreases during further aging. Obviously the softening mechanisms are dominating. A reason for this behaviour is seen in the coarsening being the dominant mechanism after reaching peak strength. This can be observed in the TEM micrographs. After long aging times large particles can be seen, while for the single-step aging the diameter of the particles increases only slightly (Fig. 6 and 7). The resitivity rises after 5.2 h with t (-1/3) (Fig. 8). This connectivity is expected, if a coarsening process is assumed as the dominant process and if nucleation of new precipitations can be neglected [25, 26].

Fig 8: Resitivity of alloy EN AW 7075 during age hardening over t(-1/3) for the investigated aging conditions.

The fact that the conductivity rises even after peak hardness is reached confirms the assumption that the strengthening processes as well as the softening mechanisms cause an increase of the conductivity. This is the reason why the hardness peak is not connected with an extremum in the conductivity progression.

7 Conclusions

The build-up of the measuring system shows that eddy current measurements can be performed at high resolution even at high temperatures of a heat treatment process. This made it possible to record the electrical conductivity of the aluminium alloy EN AW 7075 during the aging process.

The conductivity progression can be associated with micro-structural changes of the material. During natural aging the decomposition takes place mainly by the formation of GP zones resulting in a decrease of conductivity. For temperatures above the solvus temperature of the GP zones the precipitation of h' and h particles becomes the dominating decomposition mechanism resulting in an increase of conductivity due to the purification of the aluminium matrix and the dissolution of GP zones. For long aging times at 160°C the conductivity progression reflects that the process is governed by coarsening.

The investigations on the aging of aluminium are seen as an example how the eddy current technique can be used during a heat treatment process to get continuous information about the material properties. This information can be used to make heat treatments saver and to improve their reproducibility.

8 Acknowledgments

The authors thank the Bundesministeriums für Bildung und Forschung and the Deutsche Forschungsgemeinschaft for their financial support in this project.

9 References

  1. S. Steeb: Zerstörungsfreie Werkstück- und Werkstoffprüfung, expert-Verlag 1993.
  2. H. Fischer: Keep Cool, MSR-Magazin, VTW Verlag für Technik und Wirtschaft GmbH & Co.KG, 5/2001, p. 537.
  3. C. E. Burley, J. E. Duarte: Quality Control of Heat-Treated Aluminium Plate by Electrical Conductivity Testing, Material Evaluation 42 (1984), p. 1487-1491.
  4. J. Nehring, N. Mack: Kombinierte Material-/Gefüge- und Risskontrolle - Prüfpotential kombinierter Mehrfrequenz-Wirbelstromprüfstationen im Rahmen der industriellen -on line Qualitätskontrolle", Deutsche Gesellschaft für zerstörungsfreie Prüfung e.V., DGZfP-Jahrestagung 27.-29. April 1992, Fulda, p. 128-145.
  5. B. Frühauf, G. Maier: Zerstörungsfreie Qualitätssicherung von randschichtbehandelten Bauteilen, HTM 48 (1993), p. 92-96.
  6. A. Horsch: Eddy Current Testing Often Used in Testing Hardness After Heat Treating, Proceedings of the 16th ASM Heat Treating Conference & Exposition, 19.-21. März 1996, Cincinnati, Ohio, p. 507-515.
  7. final report DFG research project KL 725/7-1: Zerstörungsfreie mikrostrukturelle Werkstoffcharakterisierung mit temperaturbeständigen Wirbelstromtastsensoren, 2002.
  8. H. Betzold: Mehrfrequenz-Wirbelstromprüfung, Optimierung der Prüfparameter und der Signal-Verknüpfungs-Algorithmen zur Erhöhung des Nachweisvermögens und zur Ertüchtigung der Interpretation der Zielgrößen, dissertation, Universität des Saarlandes 1990.
  9. M. Sellen: Neues Messverfahren für Wirbelstrom-Abstandssensoren mit weitgehender Unterdrückung des Targeteinflusses, dissertation, Universität des Saarlandes 1994.
  10. R. J. Hall: Sensor Development for Intelligent Processing of Age Hardenable Aluminium Alloys, Thesis, Naval Postgraduate School, Monterey, California 1995.
  11. Y. Wang: Trennung der Einflussgrößen von Wirbelstromsensoren durch Signalverarbeitung mithilfe von Felduntersuchung und Modellierung, Fortschritt-Berichte VDI Reihe 8 Nr. 557, Düsseldorf, VDI-Verlag 1996.
  12. F. Röper: Abstandskompensierte Wirbelstrommessung von Foliendicke, Leitfähigkeit und Temperatur, Fortschritt-Berichte VDI Reihe 8 Nr. 782, Düsseldorf, VDI-Verlag 1999.
  13. W.-J. Becker: Induktiver Wirbelstrom-Aufnehmer mit temperaturkompensiertem Spulensystem, Archiv für Elektrotechnik 73 (1990), p. 181-192.
  14. C.V. Dodd, W.E. Deeds: Analytical Solutions to Eddy-Current Probe-Coil Problems, Journal of Applied Physics 39 (1986), p. 2829-2838.
  15. J. Vetterlein, H. Klümper-Westkamp, Th. Hirsch, P. Mayr: Einsatz des Wirbelstromverfahrens zur in-situ-Kontrolle von Wärmebehandlungsvorgängen, HTM 58 (2003), p. 83-89.
  16. ASM Handbook, Vol. 4, Heat Treatment, Heat Treatment of Aluminium Alloys, 1991, p. 841-879.
  17. J. E. Hatch: Properties and Physical Metalurgy, ASM, Metals Park, Ohio, 1983, p. 184.
  18. US-patent 3,198,676: Thermal Treatment of Aluminium Base Alloy Article, 1965.
  19. P.A. Thackery: The Nature and Morphology of Precipitate in Al-Zn-Mg Alloys, Journal of Institute of Metals 68 (1968), p. 228-235.
  20. T. Ungár: The Formation of Guinier-Preston Zones in the Al-4.8 wt.% Zn-1.2 wt.% Mg Alloy Studied by X-Ray Small Angle Scattering, Zeitschrift f. Metallkunde 70 (1979), p. 739-745.
  21. A. Zahra, C.Y. Zahra, M. Laffitte, W. Lacom, H.P. Degischer: Entmischungsvorgänge in einer Al-5%Zn-1%Mg-Legierung, Zeitschrift f. Metallkunde 70 (1979), p. 172-179.
  22. J. K. Park, A. J. Ardell: Microstructures of Commercial 7075 Al Alloy in the T651 and T7 Tempers, Metallurgical Transactions 14A (1983), p. 1957-1965.
  23. M. Rosen, E. Horowitz: The Aging Process in Aluminium Alloy 2024 Studied by Means of Eddy Currents, Material Science and Engeniering 53 (1982), p. 191-198.
  24. P. L. Rossiter, P. Wells: Phil. Mag. 24 (1971), p. 425-436.
  25. P. Guyot, L. Cottignies: Precipitation Kinetics, Mechanical Strength and Electrical Conductivity of AlZnMgCu Alloys, Acta mater. 44 (1996), p. 4161-4167.
  26. R.C. Dorward: Precipitate coarsening during overaging of Al-Zn-Mg-Cu alloy, Material Science and Technology 15 (1999), p. 1133-1138.
STARTPublisher: DGfZPPrograming: NDT.net