THERMOGRAPHIC DETECTION OF CONDUCTING CONTAMINANTS IN COMPOSITE MATERIALS USING MICROWAVE EXCITATION M.W. Bowen Lockheed Aeronautical Systems Company Marietta, GA 30063 R. Osiander, J.W.M. Spicer and J.e. Murphy The Johns Hopkins University Applied Physics Laboratory Laurel, MD 20723 INTRODUCTION This paper describes microwave-source time-resolved infrared radiometry (MW -TRIR) as a method for the detection and characterization of microwave absorption by conductive fibers and other absorbing regions in dielectric materials. Due to recent technical developments in the speed, detector array size, and sensitivity of infrared focalplane arrays, time-resolved infrared radiometry has evolved into an important NDE tool which allows fast area inspection at high spatial resolution. While much prior work has focused on the detection of structural defects or disbonds in a variety of materials [1,2], the increasing importance of composite materials requires new approaches to inspection which allow characterization of local material properties. Defects in such materials may have little thermal contrast compared to the matrix material and may be invisible using conventional infrared radiometry methods. However, where the embedding material is a weak microwave absorber, localized microwave absorbing regions can be detected easily. There are three different classes of absorption processes: (1) dielectric loss (e.g. water), (2) magnetic loss, and (3) Joule heating (e.g. electromagnetic radiation interaction with conducting fibers). An example of Process (1) has been the use of time-resolved infrared radiometry (TRIR) with a microwave heating source to detect water in composite materials [3,4]. The other two processes, magnetic loss and Joule heating, have been used for microwave field imaging [5,6,7]. In a thin absorbing sample the heating pattern is related and, in some experimental conditions, is proportional to the microwave energy distribution on the surface. A thin epoxy sheet loaded with ferrite can be used as an example of a magnetic absorber and any thin metallic or carbon layer with given surface conductivity can be used to show the electric field distribution through Joule heating. When the size of the microwave absorbing region becomes smaller than the microwave wavelength (about 1" for X-Band (8-12 GHz», the heating pattern is determined by the defect geometry. In the case of a conductor, discontinuities such as an edge heat up more than the rest of the defect. As a result, small, conducting, one-dimensional structures such as metal wires or fibers are very efficient microwave absorbers. Such small absorbing structures cannot be imaged with a resolution better than about one wavelength with a microwave imaging system such as a radar range due to the Rayleigh criterium. The thermographic imaging technique allows much higher resolutions since the limiting spatial resolution of the infrared detector given by the Rayleigh criterium is about 1-10 11m. Review of Progress in Quantitative Nondestructive Evaluation. Vol. 14 Edited by D.O. Thompson and D.E. Chimenti, Plenum Press, New York, 1995 453
Such small linear conducting defects can occur in a composite fabrication environment. The quality of the composite structures depends on the purity of the different layers and is an important factor for applications in the aerospace industry. Thermographic detection with microwave excitation is a method which can detect and identify these defects. With its timeresolved aspect, TRIR [8] allows the defect depth to be determined. This can be used for evaluation of potential repairs, because defects close to the surface could be removed with minor impact on the structure. EXPERIMENTAL SETUP Figure 1 shows a diagram of the experimental setup. All the measurements described here use an HP 6890B Oscillator (5-10 GHz) to produce microwaves at a frequency of 9 GHz. This signal is amplified to a maximum power of 2.3 W by a Hughes 1277 X-band traveling wave tube amplifier and fed into a single flare horn antenna through rectangular waveguide. The antenna has a beam width of about 50 degrees and is placed about 15 cm from the sample. Both the angle of incidence and the polarization of the microwave field relative to the sample are controlled. In addition, the specimen is mounted on an X-Y-Z stage to allow accurate control of sample position. A 128x128 InSb focalplane array (Santa Barbara Focalplane) operating in the 3-5 J.!m band is used for detection of the IR radiation. The camera has a temperature resolution of about 3 mk and a frame rate as fast as 305 Hz or 3.3 ms per frame. The frame synchronization pulse of the infrared camera triggers the microwave oscillator and the sample temperature is monitored as a function of time during the microwave pulse. This allows longer observation times with low power input and hence small temperature rises, as in time resolved infrared radiometry with optical heating. Time-dependent measurements are made by recording a series of frames before, during and after the application of the microwave step heating pulse. The total measurement time is easily varied by selecting the appropriate frame rate, the number of frames to be recorded and the number of skipped frames between each recorded frame. These frames are stored in a buffer board in the microcomputer which controls the focalplane array camera and are subsequently stored on Bernouilli disks for analysis. Analysis of these buffer files is performed using LabVIEW routines on a Macintosh Quadra 840A V computer which allows a range of information to be extracted from the data. Trigger M\'I-Generator o TWT-Amp. M\'I-Horn. Sample Panel Microcomputer MW-Absorber Fig. 1............ oj '".., Experimental setup for microwave-trir measurements. 454
EXPERIMENTAL RESULTS Experiments to study the interaction of microwaves with linear conductors were performed on carbon fibers in different epoxy structures. The electromagnetic interaction depends on fiber length, thickness, and microwave polarization. The thermal response depends on the depth of the fiber in the structure and on the thermal properties of the embedding material. One advantage of infrared radiometry for detection is that a spatial resolution much better than the microwave wavelength can be obtained. Figure 2 shows an image reconstructed from a microwave hologram of a fiberglass-epoxy specimen with embedded fibers at different locations. The microwave hologram is produced from the magnitude and phase angle of the reflected microwaves when scanning an emitting and a receiving hom in a plane. The image of the reflection distribution is reconstructed digitally from this hologram. The fibers can be detected in this image and although the response depends on fiber size and orientation, these parameters cannot be imaged directly. The spatial resolution in this measurement is given by the Rayleigh criterium at the microwave wavelength. The thermal detection method, as shown in the experiments below, is sensitive to orientation, length, and its spatial resolution is limited just by the optical parameters. Dependence on Fiber D\4>th A schematic diagram showing the coordinate system used for the analytical description is shown in Fig. 3. A line source, buried at a depth z, is heated uniformly. The response measured by the IR camera is proportional to the surface temperature T(x,y,z=O). For a semiinfinite medium the solution of the thermal diffusion equation for an infinite line source with continuous heating is given by [9]: (1) Fig. 2 Image of the reflected microwave power, reconstructed from a microwave hologram of fiberglass-epoxy specimen with embedded fibers of different lengths and orientations. 455
where Ei(x) is the exponential integral and a is the thermal diffusivity. This expression can be used to fit both the time dependence and the spatial dependence of the temperature distribution to get a value for (x2+y2)/a. The thermal response of the heated fiber depends on the depth of the fiber and on the thermal properties of the embedding material. Figure 4 shows an infrared image of a carbon fiber in fiberglass-epoxy after 8 s heating. The time dependence of the temperature at the center of the fiber, taken from successive images, is shown for both fibers in Fig. 5. The solid line was calculated with Eq. 1, assuming an infinite line heating source at a given depth. Fitting Eq. 1 onto the experimental result yields a value of 0.003 cm2/s for the thermal diffusivity, knowing the depth of the fibers was 0.25 mm and 0.75 mrn. The spatial distribution of the temperature across the fibers was calculated with these values and Eq. (1). This is shown in Fig. 6 together with the experimental results for different times and the two different fiber depths. The agreement is very good. An application similar to the detection of conducting fibers in dielectric materials is shown in Fig. 7. It shows the IR images of microwave absorbing foam containing metal slivers. The slivers show up after 10 s of heating in the IR image since they are heated more than the foam. Fig. 3 Schematic diagram of coordinate system used for analytical development. Fig. 4 Infrared image of fiberglass-epoxy composite with carbon fiber contaminate after an 8 s microwave heating pulse. 456
1400 1200 () 0.75 mm IJ. 0.25 mm 1000 to) 800... ;::j... to) 0. 600 e to) E-< 400 200 Fig. 5 0 0 0.5 Squareroot (time [mk]) 1.5 2 Surface temperature as a function of square root time for a single point on fibers in 0.25 mm and 0.75 mm depth. Solid lines are calculated with Eq. 1. SZ E 1400 --d=o.25 mm, 3.9 s --d=o.75 mm, 3.9 s 1200 d=o.25 mm, 2.6 mm, 2.6 1000 800 ::J Cii 600... a. E 400 r 200 0-200 --d=o.25 mm, 1.3,. mm, 1.3 s -0.3-0.2-0.1-0 0.1 0.2 0.3 Position [em] Fig. 6 Temperature distribution across each of the fibers of Fig. 9. 457
Dependence on Fiber Orientation The microwave absorption is very sensitive to the polarization of the electric field with respect to the fiber direction. For thin fibers only the electic field component along the fiber direction (E cos e ) can induce a current in the fiber. The polarization of the microwaves can be changed by rotating the horn and measurements were made with steps of 15 deg in the polarization angle. The temperature rise at one point on the sample is plotted in Fig. 8 as a function of the angle of polarization with respect to the fiber direction and shows the expected cosine behavior. After Cut Sliver insened OK..., == l. B K Fig. 7 IR image of microwave absorbing foam after los microwave illumination with and without a metal sliver inserted. 400 350 SZ' 300 E. ::J 250 Cil... a. E 200 I-..l<: C\l 150 a.. 100 50-20 0 20 40 60 80 100 Angle of Polarization [deg] Fig. 8 Temperature at fiber center as a function of polarization angle. 458
De.pendence on Fiber Length The interaction is strongly dependent on the length of the fiber. Figure 9 shows a series of IR images for carbon fibers of different lengths in fiberglass-epoxy. In Fig. 10 the temperature distribution along each fiber is plotted for different fiber lengths. Note that for fibers longer than 12 mm the temperature shows a modal distribution along the fiber. Fig. 9 IR images for different fiber lengths. sz 1000 E- O) en a: :l 0) a. E 0) I- 500 -B--6mm --a----- 9 mm -+-12mm 18 mm -+-24mm -----tr- 30 mm o o 0.5 1.5 2 2.5 3 3.5 Position [em] Fig. 10 Temperature distribution along fiber axis for different fiber lengths. 459
CONCLUSIONS We have shown that time-resolved infrared radiometry is a very useful tool in NDE for microwave absorbing materials or microwave absorbing defects. It can be used for contaminate detection as well as for homogeneity studies. The possibility of heating buried microwave absorbers directly gives a very high contrast for such defects. Further it allows the use of deliberately implanted "defects" to be used as sensors to provide information about the material using the TRIR approach. This can be used to monitor curing of epoxy materials, deterioration of materials, or other processes such as thickness changes during coating deposition. REFERENCES 1 J.W. Maclachlan Spicer, W.D. Kerns, L.e. Aamodt and J.C. Murphy, 1. Nondestruc. Eval. 8 (2) p. 107 (1989). 2 J. W. M. Spicer, W. D. Kerns, L. e. Aamodt, J. e. Murphy, in Review of Progress in Quantitative Nondestructive Evaluation, Vol 11, pp. 433-440, edited by D. O. Thompson and D. E. Chimenti, Plenum Press, New York (1992). 3 R. Osiander, 1. W. M. Spicer, 1. e. Murphy, in Thermosense XVI, edited by J.R. Snell, Jr., SPIE 2245, pp. 111-119 (1994). 4 R Osiander, 1. W. M. Spicer, and 1. C. Murphy, to be published in Proceedings of the 8th International Topical Meeting on Photoacoustic & Photothermal Phenomena, Journal de Physique - Colloques (1994). 5 J.D. Norgard, J. Sadler, RM. Sega, E.A. Baca, and W. Prather, in Thermosense XVI, edited by J.R Snell, Jr., SPIE 2245, pp. 286-291 (1994). 6 M.R. Seiler, 1.L. Haselwood, and L.A. Stockum, in Thermosense XIV, edited by J.K. Eklund, Proc. SPIE 1682, pp. 296-307 (1992). 7 D.L. Balageas, P. Levesque, and A. Deom, in Thermosense XV, edited by J.K. Eklund, Proc. SPIE 1933, pp. 274-285 (1992). 8 L.C. Aamodt, J.W. Maclachlan Spicer and J.e. Murphy, 1. Appl. Phys. 68 (12) p. 6087 (1990). 9 H.S. Carslaw and 1.e. Jaeger, Conduction of Heat in Solids (Oxford Univ. Press, London 1959). 460