Magneto-optic microscope for visuallydetecting subsurface defects

Zhao-Fei Zhou1,* and Yu-hua Cheng2

1Institute of Applied Laser, Sichuan University, Chengdu, Sichuan, 610065, China2School of Automotive Engineering, University of Electronic Science and Technology of China,

Chengdu, Sichuan, 610054, China

*Corresponding author: zhfzhou@scu.edu.cn

Received 11 March 2008; revised 27 May 2008; accepted 29 May 2008;posted 2 June 2008 (Doc. ID 90514); published 24 June 2008

A visual test method for detecting microdefects under fine surfaces is described. A new MO microscopethat has a laser source, a CCD camera, and an exciting coil is developed for this work. A pulse generatorsupplies an intermittent square pulse to the exciting coil, which can intensify eddy currents yet reducethe working temperature of the exciting coil and sample. The magnetic field variation produced by theimbedded defect causes a rotation of the polarization plane of the reflected beam. Therefore the reflectedbeam carries an image of the defect, which is received by a CCD camera. The optical arrangementguarantees that no light is reflected back to the laser. The system was tested with a calibrator, whichhas an artificial subsurface defect; such a test attains a visual detected image. To our knowledge this isthe first time an image of a subsurface defect has been distinctly detected with a MO sensor system.© 2008 Optical Society of America

OCIS codes: 040.0040, 110.0180.

1. Introduction

It is important to have a nondestructive and visualtest to detect subsurface defects in conjunction withthe technological development of power lasers, high-power engines, and machines working in difficult op-erating conditions. Various techniques for such testscan be used, including ultrasonic, x-ray, radioactiveray, and eddy-current methods. An ultrasonic test,for example, requires a coupling liquid betweenthe sensor and the sample surface, which makes op-eration of the test complex and slow; however, thismethod is suitable to detect larger defects in sam-ples. The operation of x-ray and radioactive raymethods is even more complicated, and they are ac-companied by safety concerns. These techniques arealso suitable to test larger defects and samples.The weakness in traditional eddy current detec-

tion is that the test result is invisible. Visual detec-

tion of microdefects under fine surfaces is of interestbut is difficult in eddy current detection. At the be-ginning of 2000, eddy current detection could only de-tect defects on a tested surface [1–4]. However, alaser profilometer can detect such defects with muchbetter resolution and easier operation [5]. Our workis to develop a visual eddy-current sensor for detect-ing defects under fine surfaces. A magneto-optic(MO) sensor is used to realize this purpose. Figure 1illustrates the effect of the MO sensor.

The incidence beam from a light source passesthrough polarizer 1 and becomes polarized. The di-rection of polarization is parallel to the plane ofthe diagram. The beam then passes through anMO crystal to which a magnetic field has been ap-plied. The polarization direction of the beam is ro-tated due to the Faraday effect. Suppose therotated angle is only 90°: then, there is no outgoingbeam passing though polarizer 2, because the polar-ization direction is perpendicular to the polarizer. Ifthe intensity of the magnetic field is weaker than theexample above, the rotation angle of the polarization

0003-6935/08/193463-04$15.00/0© 2008 Optical Society of America

1 July 2008 / Vol. 47, No. 19 / APPLIED OPTICS 3463

plane will be smaller than 90° and the outgoinglight will be brighter. Therefore the variation of mag-netic field intensity causes the intensity of the out-going beam to vary, and these variations form avisual image.

2. Design of the Magneto-Optic Experiment System

Figure 2 is a sketch map of the experiment systemthat is used to verify the operating principle of thevisualMO sensor. The light source is a semiconductorlaser that emits a red laser beam.Apolarization beamsplitter orients the polarization of the transmittedbeam to be parallel to the plane of the diagram.The beam passes through an MO crystal and reflectsfrom the sample surface. The carry table is tilted sothe reflected light does not reflect back to the laser.A 1=2λ plate rotates the polarized direction of thebeam by 90°, and then a polarization beam splitterreflects the beam and directs it to the CCD camera.A pulse generator sends a square-pulse current tothe exciting coil and induces eddy currents in thetested surface in which defects have been introduced.These defects induce a variation in the eddy currentas well as a variation in the magnetic field. The mag-netic variation depends on the local eddy current.

Finally, the CCD camera receives the reflected laserbeam that carries the image of the defect.

There are two weakness of this system: First, onlythe end part of the MO glass is close to the variatedmagnetic field, therefore the efficiency of MO trans-form is lower. Second, the hollow exciting coil (theMO crystal in the center has very low magnetic in-ductivity) cannot create a strong magnetic field. Inorder to fulfill this requirement for a MO micro-scope, the above weakness has been reformed bysucceeding research.

The interrelation of the eddy-current frequencywith penetration depth is important in decidingthe excitation frequency and power. Figure 3(a)shows the interrelation among the excitation fre-quency, sample material, and penetration depth ofthe eddy current. The longitudinal coordinate isthe relative intensity of the eddy current. Theeddy-current intensity has a skin effect. It is as-sumed that the intensity is 100% on the sample sur-face and 36.5% at a certain depth referred to as thenormalized penetration depth δ. Here,δ ¼ 1=

ffiffiffiffiffiffiffiffiffiffiπfμσ

p,

f is the excitation frequency, μ is the permeability ofthe sample, and σ is the conductivity of the sample[6]. Figure 3(b) shows the relation of the excitationfrequency and the penetration depth [6].

From Figs. 3(a) and 3(b), it can be concluded that ahigher excitation frequency is advantageous in de-tecting subsurface microdefects. However, this isonly under special conditions; in practice, there arelimitations that will be discussed below. The weak-ness of this experiment system is that it can only de-tect exposed defects on surfaces or under varnish, asfor past research [1–4]. It cannot obtain a clear imageof subsurface defects because it cannot excite a suffi-ciently strong eddy current and corresponding mag-netic field, and the optical arrangement has ratherhigh noise.

3. Magneto-Optic Microscope

Figure 4 shows a schematic diagram of the practicalMOmicroscope. In order for this microscope to detectthe subsurface defects, the intensity of the excitationpower should be strengthened considerably, thus acoil with a ferrite core and a YIG film are used.The ferrite core greatly reduces the magnetic reluc-tance of the coil. The YIG film not only has high op-tical rotating power but also can cling to the samplesurface, where the magnetic field excited by the eddycurrent is most intense. The weakness of YIG is itslow transmittance. This optical arrangement com-pletely prevents the laser beam from reflecting backto the laser source, which avoids noise being pro-duced within the laser. This is important to increasethe definition of this microscope. Finally, before theCCD camera and lens cluster, there is a polarizer,which may seem superfluous but is useful in adjust-ing the contrast of the detected image.

Another major question is what kind of excitationcurrent should be used. First, according to Figs. 3and 4, higher frequency alternating current is

Fig. 1. Operation principle of a MO sensor.

Fig. 2. Optical arrangement of the experiment system.

3464 APPLIED OPTICS / Vol. 47, No. 19 / 1 July 2008

advantageous for the subsurface detection, but forthe coil with a ferrite core, too high a frequencycauses higher impedance and higher loss, which re-duces the exciting current. Second, a square pulse ofexcitation current has a higher exciting efficiencythan does a sine wave. Another problem is thatthe exciting coil and samples are liable to overheat-ing by a powerful excitation current with higherfrequency. Therefore an intermittent square pulseis selected as the type of excitation current, whichcan reduce the working temperature while not redu-cing the exciting efficiency. Figure 5 shows thisintermittent square pulse of the exciting current.Here the pulse frequency is 2kHz; t0 is the durationof the square pulse; and t is the intermittent time,which is equal to 1:5 t0; this value is acquired fromexperiments.

4. Experiment

Samples with subsurface microdefects were madefor this work. Figure 6(a) shows a schematic dia-gram with dimensions 22mm × 10mm × 12mm,and Fig. 6(b) shows a photo of the calibrator. The di-

mensions were selected so that machining would beeasy and there would be less heat deformation. Theincisions under the top surface were made by electricspark line cutting, which pierced the whole width(10mm) of the sample.

The width of these incisions is 150–500 μm, andthey are located 15–25 μm from the top surface. Atpresent, we cannot make thinner incisions with thismethod. Figure 7 shows one of the MO-detected re-sults for an incision that was under the detected sur-face and 250 μm in width, and this sample isused as a calibrator. Figure 7(a) is the original image,and Fig. 7(b) is the result processed by adaptivefiltering [7].

Determining the resolution of the MO microscopeis problematic because, at present, a calibratorwith a subsurface “defect” a few μm in size with un-der surface machining is not available. For this rea-son, tiny grooves 15 μm× 20 μm× 1000 μm (width×depth × length) were made by photolithography onthe sample surface, and the grooves were filled withplastic, which is normally used for sample prepara-tion for a scanning electron microscope. Then, afterpolishing the surface further and after metal plating,the coating is about 2 μm in thickness. The disadvan-tage of this method is that some slight scars can beseen on the coated surface (1–2 μm concave–convex).However, this disadvantage hardly affects the detec-tion. Figure 8 shows detected images for the groove ofthe calibrator; Fig. 8(a) is the image detected by ascanning electron microscope before coating, andFig. 8(b) shows the MO detection result with the sur-face coated. The detected groove was 15 μm in width.

Fig. 3. Relation of (a) eddy current and (b) excitation frequency to penetration depth.

Fig. 4. Optical arrangement of the MO microscope. Fig. 5. Schematic diagram of the intermittent square pulse.

1 July 2008 / Vol. 47, No. 19 / APPLIED OPTICS 3465

5. Conclusions

This work is believed to be the first in which a sub-surface defect has been detected distinctly with anMOmicroscope. However, this research is not matureenough yet. For instance, how to determine the bestresolution of the MO microscope still depends on thedifficulty of calibrator fabrication; a new method ofcalibration should be found.It seems reasonable that there is the potential for

research on precision machining, because some cut-ting machining can create tears in fine machined sur-faces or enlarge crystal-boundary cracks, and thenmachining afterwards can close some cracks in thesurface. The problem is that the hidden danger offatigue failure is as before: these flaws are hard todetect. Our work provides a possibility to solve thisproblem and also has the potential for on-line detec-tion, especially for aircraft components. The minia-

turization of sensors in these components is anotherresearch task.

This project was supported by the National Natur-al Science Foundation of China.

References

1. U. Radtke, R. Zielke, and H. G. Rademacher, “Application ofmicro-optical method for real-time visualization of eddy cur-rent with high spatial resolution for nondestructive testing,”Opt. Lasers Eng. 36, 251–268 (2001).

2. P. Novotny, M. Morozov, and L. Hudec, “On the application ofMO firm in NDE,” NDT Int. 35, 227–232 (2002).

3. P. Novotny, P. Sajedl, and P. Machac, “A MO imager for NDTapplications,” NDT Int. 37, 645–649 (2005).

4. J. Pinassaud, P.-Y. Joubeet, and M. Lemistre, “Quantitativemagneto-optic imager for nondestructive evaluation,” Proc.SPIE 5768, 196–201 (2005).

5. Z. F. Zhou, T. Zhang, and W. D. Zhou, “Profilometer for mea-suring superfine surfaces,” Opt. Eng. 40, 1646–1652 (2001).

6. B. N. Jia and P. Yu, “Modeling of characteristic of magneto-optic sensor using FEM and dipole model for nondestructiveevaluation,” KSME Int. J. 16, 591–593 (2000).

7. Z. F. Zhou and T. Zhang, “Precision ranger for measuring largemechanical components,” Appl. Opt. 43, 4816–4820 (2004).

8. M. Morozov and P. Novotny, “Evaluation of eddy currentprobes based on local field excitation,” NDT Int. 35, 147–153(2002).

9. J. Lee, J. Hwang, and T. Shoji, “Modeling of characteristic ofMO sensor using FEM and dipole model for nondestructiveevaluation,” KSME Int. 16, 591–593 (2002).

Fig. 6. (a) Schematic diagram and (b) photo of the calibrator.

Fig. 7. Magneto-optical detected image and the processed result

Fig. 8. Groove image of the calibrator, which is detected by(a) scanning electron microscope (before coating) and (b) MOmicroscope (the detected surface is coated).

3466 APPLIED OPTICS / Vol. 47, No. 19 / 1 July 2008