GMR based NDT System for Defects in Magnetic Materials

Transcription

1 11th European Conference on Non-Destructive Testing (ECNDT 2014), October 6-10, 2014, Prague, Czech Republic GMR based NDT System for Defects in Magnetic Materials More Info at Open Access Database Matthias BUERKLE, Dr. Hubert GRIMM, Dr. Johannes PAUL Sensitec GmbH, Mainz, Germany Phone: , Fax: ; sensitec.com, Abstract A NDT measurement system for crack detection based on a magnetic tooth length sensor is presented. The system has compact dimensions and can also be used with complex geometries. The function is similar to a magnetic flux leakage system. Instead of measuring stray fields, the deviation of a strong applied magnetic field due to the presence of magnetic material or defects is detected. GMR multilayer sensors have been selected, because they are sensitive at high fields for measure in high magnetic field environments. Two applications are reported. We have developed a test system for magnetographic print head chips in a mass production environment. Thereby soft-magnetic pole heads are analyzed for defects caused during an electroplating process with Nickel-Iron. The second application is the detection of surface cracks in magnetic materials. The measurement results show that it is possible to detect cracks down to 10 µm deep and 81 µm wide. Because of the high sensitivity the grain size of the material is detected and might be the limitation for crack detection. Keywords: Giant magnetoresistive sensor (GMR), Magnetic flux leakage (MFL), Magnetic stray field, Crack detection, non-destructive testing (NDT) 1. Introduction Magnetic multilayers are well suited as magnetic field sensors to a wide range of magnetic applications. The antiferromagnetic coupling strength of the magnetic layers is dominated by the thickness of the nonmagnetic interface. This effect was discovered in 1988 and is called the giant magnetoresistance (GMR) [1-3]. GMR multilayer sensors are typically used at high magnetic field applications, e.g. for position measurements. Here the sensor system consists of a permanent magnet, which magnetizes e.g. toothed wheels or bars, and a GMR multilayer sensor, sensitive at fields of up to 400 ka/m. Because of the robust build-up of the sensors, consisting only of metal layers, of silicon oxide and nitride or alumina for passivation and protection, the GMR sensors can be used in harsh environments. Even high dynamic position measurements of the valve lift can be measured even in a firing engine. The use of GMR multilayers for NDT applications is a new approach. So far, GMR spin valve sensors have been used for eddy current measurement and magnetic flux leakage measurements because of their high sensitivity at low magnetic fields. 2. System Overview 2.1 Measurement System Magnetic flux leakage (MFL) systems measure the stray field which is present at an inhomogeneity near the surface of a test specimen [4]. The test specimen needs to be premagnetized by a strong applied magnetic field or with a high current flowing through it. In this work we present a measurement system which is based on a tooth length sensor system from Sensitec [5]. The system contains a permanent magnet and a GMR sensor chip which measures in a gradiometer configuration. The permanent magnet magnetizes the soft magnetic material under test (e.g. the tooth of a gear-wheel) but penetrates the GMR sensor chip, too. By means of robust GMR multilayers the GMR layers are not saturated in this field but remain sensitive. In length measurement application the GMR sensor chip measures the field 1

2 differences caused by a soft magnetic tooth bar. The field differences change when the tooth bar is moved. In a standard position measurement application, the sensor produces cosine and sine signals. Consequently the system is able to obtain position information. The GMR length sensor is designed for a tooth pitch of 634 µm (gear wheel module of m = 0.2) and is sensitive in the vertical plane of the specimen surface. The GMR sensor consists in a pattern of GMR sensing elements, each of these sensing elements have a length of only 90 µm. The sensor is placed at the bottom edge of the magnet and is exposed to a biasing field. In Figure 1 is the principle of a tooth length sensor system and its use for NDT shown. Figure 1: Left: Principle of a tooth length sensor system with permanent magnet and sensor (placed at the left side of the permanent magnet). Right: Sensor system placed above a test specimen for crack detection in NDT. As can be seen in figure 1, the GMR gradiometer is biased by the flux from the permanent magnet which enters the surface of the test specimen. At this spatial closeness to the permanent magnet the sensor must have a high working range of the magnetic field as well as a high sensitivity. These requirements can be achieved with GMR multilayer sensors. Field saturation of the GMR multilayer sensor is in the range of 500 mt. These high saturation values allow the usage of strong magnets which fully saturate the surface of the magnetic material under test. As long as there is no inhomogeneity in the surface of the specimen the signal of the sensor is constant and does not change. At any defect positions the magnetic flux lines are distorted because of the significant permeability change in this region. 2.2 Magnetography Print Heads Magnetography is a digital non-impact printing technology. The printing systems are developed and manufactured by the French company Nipson SAS. The systems are capable of high speed (up to 150 m/min) and high volume (more than 500,000 pages/month) black and white printing. An integral part of a magnetography printing system is a high-density array of microelectromagnets ( 1100 devices/cm²). Each print head chip consists of up to 616 fully functional micro-coils, magnetic poles and buried return boxes. It is fabricated on a silicon wafer by Sensitec GmbH with magnetic MEMS processes [6-7]. In a printing system are 18 chips placed as an array to get a wide printing format. Every magnetic pole generates a single dot at the printed paper. The malfunction of a single microcoil or defect in a magnetic pole results in failure of the printed image. The fully array has to be exchanged in this case, resulting in high maintenance and production downtimes of the printer. The soft-magnetic parts on the print head chips are electroplated from a Nickel Iron (NiFe) cell. The magnetic circuit consists of the soft-magnetic pole (with an area of 32 x 57 µm and about 50 µm height) above a return box and on the other side on the imaging cylinder of the hard-magnetic surface. A short electric impulse at the micro coil induces a magnetic flux 2

3 which is concentrated at the magnetic pole. The print image is transferred from the softmagnetic poles to the imaging cylinder. The imaging cylinder has a hard-magnetic surface made of CoNiP. The toner particle consists of iron oxide powder ligated in a polymer which gets adhered to a magnetized pixel at the imaging cylinder. To control each of the 616 poles, diode-chips are mounted in a flip-chip process on top of the print-head chip. An array of magnetic poles is shown in figure 2. The electroplating of the 50 µm high poles is a technical challenge because of the high aspect ratio and the ratio of the plating area to the overall wafer area is very small. In addition to reach good magnetic characteristics a constant Ni-Fe composition of 50% must be reached. During that process enclosures and imperfections inside the electroplated poles can occur. A photograph of a chip is shown in figure 3. Figure 2: Array of electroplated Cu-coils and NiFe poles. A defect of one pole corresponds to the defect of the full array. Figure 3: Print head chip with resine layer, diode chips at the left side, coils and magnetic poles in the right area. During the batch fabrication of the print head chips it was not possible to test the magnetic properties of the poles. The first time the magnetic poles on the chip can be tested is after the installation in the printer. If a pole does not work correctly the resulting dot is weak or missing completely. The measured electrical parameters of the coils and the diodes gave no indication of the pole function. To save cost it is required to test the functionality as early as possible and before the electronic packaging is done. A print head can only be delivered when every of the 616 poles are faultless. So it was necessary to develop a measurement system to test the magnetic functionality of every pole. 3. Measurement Results In this section we present the measurement results with the introduced measurement system for both magnetic print head chips and a reference test specimen with artificial cracks. 3

4 3.1 Testing of Magnetic Micro Poles for Magnetography Printers The sensor system of magnet and multilayer sensor is well capable to test arrays of magnetic circuits which are used in magnetographic printing systems as shown in the figure 5. The magnetic poles are characterized by the same principle like in position measurements: The magnetic poles are saturated by a permanent magnet, small defects in the magnetic circuits are detected by a GMR gradiometer sensor comparing each pole to his neighbors. The sensor system set-up allows very fast movements without distortion of the signals caused by induced eddy currents in the pole heads. This is a requirement to test the print head chips in a mass production environment with short testing times of about 12 seconds. Figure 5: Test setting for magnetography print head chips with GMR tooth sensor system. For a print head chip with 600 dpi resolution, 616 poles are arranged in a 7 x 88 array (480 dpi: 6 x 80 array). The sensor system moves along the 88 poles for every row. The resulting signal is a periodic function similar to the signal from a tooth bar. The signal follows not fully a sine and cosine shape because of geometric restrictions. The spacing of the tooth sensor elements is not adapted to those of the poles. In this case the GMR gradiometer sensor elements are above two pole heads with a misalignment. Because of the misalignment, each pole can be seen in the data scan. This allows the exact assignment of the location of each pole and consequently of the location of each defect. Assuming a magnetic pole head is missing the signal displays a significant maximum. Figure 6 shows a line scan for one row. Each pole is visible by one peak. A defect in one pole can be detected by the higher and deformed amplitude. By analysing the amplitudes for every pole we can create a mapping for a whole chip and make a prediction about the functionality. We have established the measurement technique for the fabrication of the print head chips in our wafer fab in Mainz. 4

5 Sensor Signal in V Figure 6: Sensor signal of a whole row with 88 pole heads from a 600 dpi chip ( , row 7). In the last third is a defect measured, see the cutout at the right side. The first and the last poles generate an explicit higher amplitude caused by the gradiometer sensor design and missing of the neighbour pole. 3.2 Reference Test Specimen Measurement Point The measurements with the tooth sensor system are undertaken on a reference test specimen from Federal Institute for Materials Research and Testing (BAM). It is made of conventional construction steel ST37 (100 mm long, 50 mm width and 10 mm height). The defects are artificial cracks made by electrical discharge machining. The depth of the cracks varies between 10 and 2240 µm and is shown in figure Width in µm Figure 7: Cross section of reference test specimen with surface crack depths [9]. The sensor has been moved over the center of the cracks with a constant velocity without position measurement as reference. The measured signals are amplified to get higher amplitudes for the analysis of the data and a correlation of the crack sizes. In figure 8 the signals from the sensor system are shown with 50 µm liftoff. All defects can be detected even the double crack in the middle of the test specimen was separated. Between the cracks the background signal is not constant but displays variations. This is not noise of the measurement system. Repeating the measurement always results in an identical background signal. Therefore it is clear the system detects material noise [9], e.g. magnetic grains. 5

6 a) Sensor Signal in V Sensor Signal in V x Measurement Point 2 x b) Measurement Point Figure 8 a): Measurement of the gradiometric field along the complete mock-up with liftoff 50 µm. b): Cutout of the four smallest cracks. It is no problem to differentiate the double crack and also the smallest crack has a clear signal shape. Figure 9 shows a comparison of the amplitudes with different liftoff heights. At smooth surfaces of the test specimen it is possible to measure in direct contact. Some amplitudes at the largest cracks are so high that the amplifier is in saturation (> 4.7 Volts). With increasing liftoff the amplitudes decreases with a 1/r²-attenuation as expected from Maxwell s equations. Defect Depth in µm Figure 9: Comparison of resulting sensor amplitudes with increasing liftoff of the sensor system. Above 4.7 Volts the amplifier is in saturation so that the signal is cut off. The amplitudes size shows the expected 1/r²-attenuation with increasing liftoff. 6

7 The signal shape measured at a crack is comparable to that of the MFL method also when the sensor is placed at the magnetization unit [10]. The signal amplitudes correlates well with crack size. With mathematical analysing methods and wavelets it is possible to extract cracks, inhomogeneity or corrosion from a signal. Ii is also possible to generate a graphical plot of a specimen and make an evaluation about the size as well as the course of cracks at the surface. Furthermore the measurement shows that the material noise is significant higher than in other presented results with MFL systems [9]. This can be explained that a magnetic field measured near the position where it enters the specimen and magnetizes it is more influenced by the grain size and grain boundaries then a stray field measured at any other point of magnetized specimen. 4. Conclusions and Outlook In this paper we introduced the use of a tooth sensor system with GMR multilayers for crack detection and failure detection in periodic structures like the pole array. It is also suitable to find imperfections in a near surface area of soft magnetic materials. The sensor system is based on a modified magnetic flux leakage method. In contrast to MFL not the stray field is detected but the distortion of a static applied field due to defects and inhomogeneities in the test specimen. The results demonstrate that defects in magnetographic print head chips in a mass production environment can be detected. Surface cracks in ST37 as small as 10 µm deep and 81 µm wide were easily detected at a test specimen with artificial cracks. It is obvious to use the system for surface scans and analyse the data with algorithms and wavelets to reconstruct defect structures. The measurement system has small dimensions so that it can be adapted to fit in shapes or holes which are difficult to access. Furthermore it is possible to adapt the GMR gradiometer sensor design to arbitrary defect structures, corrosion or inclusions. GMR sensors are able to work over a large frequency range thus we can reach high scanning rates with the system. Acknowledgement The reference test specimen was provided from Federal Institute for Materials Research and Testing (BAM) and the results can be published with the acknowledgment of Matthias Pelkner and Dr. Marc Kreutzbruck. References 1. Fert, Albert; et al.; "Giant Magnetoresistance of (001)Fe/(001)Cr Magnetic Superlattices", Physical Review Letters 61 (21): pp , November Grünberg, P.; et al.; (1989). "Enhanced magnetoresistance in layered magnetic structures with antiferromagnetic interlayer exchange". Physical Review B 39 (7): 4828, March P.A. Grünberg; 'Exchange anisotropy, interlayer exchange coupling and GMR in research and application', Sensors and Actuators A, Vol. 91, No 1, pp , June Wang, Z.D., Gu, Y.; Wang, Y.S.; A review of three magnetic NDT technologies ; Journal of Magnetism and Magnetic Materials, Vol. 324, pp , GMR Sensor-Modul zur Abtastung von Zahnstrukturen, company brochure Sensitec GmbH, März

8 6. Eltgen, J.-J. P.; Magnenet, J. G.; Magnetic printer using perpendicular recording, IEEE Trans. on Magn. 16(5), September Doms, M.; Braun, F.J.; Krämer, K.; Magnetic print head chips comprising 3D- redistribution layers, MEMS coils and magnetic circuits in a Multi-Chip-Module for serial production ; 2. GMM Workshop - Technologien und Werkstoffe der Mikrosystem- und Nanotechnik in Darmstadt; Mai visited at , 16: Pelkner, M.; Reimund V.; Kreutzbruck, M.; et.al.; 3D-GMR-Messung an Referenzbauteilen und Rekonstruktion der Rissparameter ; DGZfP-Jahrestagung, Mo.3.A.4; Pelkner, M.; Neubauer, A.; Kreutzbruck, M.; et.al.; Routes for GMR-Sensor Design in Non-Destructive Testing ; Sensors, Vol. 12, pp , September