Microwaves holography revealed by photothermal films and lock-in IR thermography: Application to electromagnetic materials NDE

Transcription

1 Microwaves holography revealed by photothermal films and lock-in IR thermography: Application to electromagnetic materials NDE Daniel L. Balageas, Patrick Levesque, Mylène Nacitas, Jean-Claude Krapez, Gérard Gardette and Michel Lemistre Office National d'etudes et de Recherches Aérospatiales, Sensors, Characterization & NDE Lab., BP 72, Châtillon cedex, France ABSTRACT The amplitude and phase space distributions of electromagnetic fields (X- and Ku bands) are imaged and measured using microwaves interferometry revealed by photothermal films and lock-in infrared thermography. Such EM fields imaging is a powerful tool for NDE (non Destructive Evaluation) of dielectric and radar absorbing materials. Keywords: infrared thermography, NDE, electromagnetic windows, radar materials. PRINCIPLE OF PHOTOTHERMAL ANALYSIS OF EM FIELDS INTENSITY The absorption of an electromagnetic (EM) field E, H, by materials of complex permittivity, ε = ε' - jε'', complex permeability, µ = µ' - jµ", and real part of the conductivity, σ', produces heat with a dissipated volume power given by the well known formula: P abs = /2 [ (σ' + ωε") E 2 + ωµ" H 2 ], () where E and H are the modules of the electric and magnetic fields. The resulting heating can be detected by IR thermography (technique called EMIR, for ElectroMagnetic InfraRed). It is controlled by the materials thermal properties and by heat transfers to the surroundings. The analysis of the space distribution and time evolution of the temperature increase allows to identify the heat sources, Pabs(x,y,z), which are controlled by the strength, space distribution and orientation of the EM field, and by the EM properties of the material, which acts as an array of photothermal sensors. If these properties of the material are known, identification of these distributed heat sources can finally lead to the knowledge of the intensity distribution of the incident EM field. The identification is optimized when the photothermal transducer is a thin film. The temperature increase distribution of the thin film, which is an image of the EM field intensity, can be monitored by an infrared camera. The principle of the technique is given in Fig.. The sensitivity of the film is controlled by its EM properties. Fig. 2 presents the influence of the film square resistance on the absorption, reflection and transmission of the film. Fig. Principle of the method. Fig. 2 Electric behaviour of a thin resistive film

2 Such a technique is a close mixing of electromagnetism and thermics. The use of both thin resistive photothermal films and lock-in IR thermography allows to image, in a quantitative way, the intensity of EM fields, with a dynamic range sufficiently extended for EM purposes. The optimisation of these films and the performances of the technique: space and time modulation transfer functions and sensitivity to the electric field are given elsewhere. 2. NEED OF A PHASE MEASUREMENT The main limitation of the technique was essentially due to the fact that the photon-heat conversion only allows to measure field intensity (square of the amplitude). Such a measurement impedes to reconstruct the full field, since the phase remains unknown. It is obvious that an EMIR method allowing phase measurements would be a real progress, especially for near-field characterization of antennas. It is the reason why phase measurement was needed. Guided by optical techniques which very often use interferometric arrangements, the idea was to produce microwaves interferences for several phase values and to reveal them by photothermal films. By combining several intensity measurements it is possible to deduce both amplitude and relative phase distributions. Here the method will be very shortly presented, since we have recently presented it, with more details 2,3,4. 3. PRINCIPLE OF THE INTERFEROMETRIC METHOD The schematic configuration is given in Fig. 3 (right). We either used a long-wave (8-2 µm) mono detector camera (Agema 782), or an Amber focal plane array camera, working in the 3-5 µm range. The IR camera is monitoring a thin resistive photothermal film placed in a volume where two coherent EM fields are interfering. These fields can be generated by two antennas (one of them being the reference) or by one antenna and an EM mirror. The phase lag between the two fields is controlled by a phase lag generator in the first configuration and by the displacement of the mirror in the second one. Furthermore, the amplitude of the EM fields are modulated in view to use the lock-in detection. In the simpler configuration the photothermal film is held perpendicularly to the common axis of the two fields, at abscissa z=zo. amplitude modulated generator IR camera E antenna # modulation phase reference phase lag generator photothermal film E2 antenna #2 x y microcomputor with lock-in system z amplitude modulated generator IR camera Ei antenna modulation phase reference metallic plate photothermal film x Er y microcomputor with lock-in system z Fig. 3 Schematic configurations for microwave interferometry revealed by photothermal lock-in thermography. Left: arrangement with two antennas - Right: arrangement with a unique antenna and a metallic plate (EM mirror)

3 Let us consider the interferometric arrangement using two antennas (see Fig. 3 left). Assuming, for simplification, that the electric fields of the antennas are chosen perpendicular to the direction of propagation and that they are vertically polarized, the electric field generated, at z o, by each of the two antennas (i=,2) is: r E ( x, y) = E ( x, y).e!j" ( x,y) r. u y ; r E 2 ( x, y) = E 2 ( x, y).e j" 2 ( x,y) r. u y (2) At this point, assuming that the film is thermally thin and that the lateral thermal diffusion can be neglected, the temperature increase T(x,y) of the photothermal film (film sensitivity ξ ), is proportional to the intensity of the vectorial sum of the two incident fields: T( x, y) =! E r ( x, y) + E r 2 ( x, y) 2 =! E r ( x, y) 2 +! E r 2 ( x, y) 2 +2! E r ( x, y). E r 2 x, y ( ).cos " 2 x, y ( ( ) # " ( x, y) ) The resulting temperature increase is then a function of the individual temperature increases caused by each antenna (i=,2), which are independent of phase, T i (x,y) = ξ E r i ( x, y) 2, and of the phase lag: Δϕ = ϕ 2 (x,y) - ϕ (x,y). By combining three successive measurements: T (x,y) and T 2 (x,y) obtained respectively when antenna # and #2 are operating alone, and T(x,y) with both antennas, the phase lag ϕ can be deduced: $ T( x, y) # T!" = ar cos ( x, y) # T 2 ( x, y) ' & ) (4) (4) % 2 T ( x, y).t 2 ( x,y ) ( In order to determine the phase sign, a fourth measurement is performed, T'(x,y), with both antennas active while introducing an extra phase lag of π/2. From the new trio (T(x,y), T2(x,y) and T'(x,y)) the sine of the phase lag ϕ is deduced and it is then possible to know in which quadrant the angle ϕ is situated. Similar procedure and formulas are obtained in the second possible interferometric arrangement (one antenna and an electromagnetic mirror) 2,4. 4. EXAMPLES OF ELECTRICAL FIELDS CHARACTERIZATION To illustrate the performances of the technique, we present in this chapter some results obtained with the best set-up now operational at ONERA. The infrared imager in the present case is a focal plane array camera, an Amber 428. A lock-in system was adapted to this camera. Fig. 4 presents the intensity distribution of an ellipsoidal antenna. For all images, the scale is (-7 db, -3 db). The central zone, which is the useful part of the EM field, is between -7 db and db and appears saturated. This presentation is justified since we are looking at the method ultimate sensitivity. The comparison between the left and central images shows the strong enhancement produced by the amplitude modulation and lock-in detection. The dynamic range is much better with modulation (more than 3 db), thanks to the suppression of heat conduction effects in the film and the distribution is no more affected by the distortions caused by natural convection. Comparison between the central and right images shows the interest to use focal plane array cameras instead of more classical cameras (here an Agema 782) based on a unique detector coupled to a scanning device. Fig. 5 gives the central E-plane distribution taken from these images and compares the thermographic data to EM probe point measurements. This comparison validates the lock-in detection technique. The rest of the results presented here are obtained with the Agema camera since the use of the focal plane array in our laboratory is very recent. For this reason, we can suppose that these results will be improved when using the array camera in a near future. (3)

4 Fig. 4 Near field radiation pattern of an ellipsoidal antenna, at 2 GHz. From left to right: i) steady state measurement with the Amber camera (4 frames added); ii) lock-in detection at Hz, with the Amber camera (492 frames added); iii) idem with the Agema 782 LW (4 frames added). Relative Intensity (db) Agema 782 lock-in Hz Relative Intensity (db) Amber no modulation 2 Amber lock-in Hz 2 Amber lock-in. Hz! EM probe measurement 3 Amber lock-in Hz - -! EM probe x (cm) -3 3 x (cm) Fig. 5 E-plane distribution of electric field, from IR images of fig. 4 and from an EM probe measurement. Left: Agema and Amber cameras at Hz; right: results of Amber camera at,., and Hz For NDE purpose, using these antennas, we limit the practical volume to the central region of the generated EM field. There is roughly a 2 cm in dia. region, near the focal point, where the intensity variations are less than -3 db and where the phase is constant. This was verified by using the interferometric method previously presented. The results, obtained with Agema 782 camera, are given in Fig. 6 and 7. Fig. 6 presents both intensity (in db) and phase (in degree) in the central useful zone. The phase is found effectively constant in the -3 db zone. The interferometric measurement gives a phase noise of ± 5. Based on very recent results, not presented here, we can state that the accuracy on the phase, with the focal plane array camera, is improved by a factor of 2 to APPLICATION TO NON DESTRUCTIE TESTING It is possible to achieve NDE of EM structures by measuring directly the absorption of EM energy and its conversion into heat. The surface temperature increase, monitored by an infrared camera, allows the detection of defects (see Fig. 8 left). The interpretation of such images is not simple and their quality is not very good since the surface temperature is blurred by 3 - D heat diffusion and they depend on the

5 Fig. 6 Distribution of intensity and phase in the central zone of the E field generated by the ellipsoidal antenna used for NDE tests. Measurement by interferometric method, using Agema 782 camera, amplitude modulation at Hz and lock-in detection radial distance (cm) -5 5 relative phase ( ) Intensity (db) -45 radial distance (cm) Fig. 7 Distribution of intensity and phase in the central zone of the E field generated by the ellipsoidal antenna used for NDE tests. Measurement by interferometric method, using Agema 782 camera, amplitude modulation at Hz and lock-in detection emissivity distribution of the structure. To avoid this, it was proposed to detect defects by mapping the EM field in the near vicinity of the tested structures 6, since the EM field is distorted by the structure and locally by the possible defects. The intensity map is obtained by using a thin film positioned parallel to the structure surface. It was demonstrated that the second method was more sensitive and accurate, since the film heating is not depending on heat diffusion inside the structure. The only interaction between structure and film is due to radiation. In this case, the amplitude modulation may have a positive effect, depending on the respective nature of structure and defects (materials absorbing microwaves or not). Thus, NDE by EMIR method reduces to an EM field mapping. Its performances will directly depend on the sensitivity of the field characterization. We can suppose that defects influence both amplitude and

6 phase distribution of the EM field. Then, interferometric methods must give more information on defects than a simple intensity measurement. It appears interesting to try interferometry for NDE application and to compare it to more classical EMIR methods. Specimen microwave sensitive thermal film Specimen Microwave source Microwave source Defect Defect IR Camera IR Camera Fig. 8 Principle of NDE by EMIR methods. Left : direct method ; right : film method. Until now, this field analysis was achieved with EMIR method by just imaging the amplitude of the reflected or transmitted EM fields (see Fig. 9 left).the interferometric method and the phase analysis presented just before give new possibilities for defect detection and characterization. amplitude modulated generator photothermal film IR camera Etr modulation phase reference tested structure x Ei antenna y microcomputor with lock-in system z amplitude phase lag generator modulated generator photothermal film E2 E IR camera antenna # modulation phase reference x y z antenna #2 tested structure microcomputor with lock-in system Fig. 9 Schematic configurations for NDE by EMIR methods using a photothermal film. Left: classical EMIR arrangement with a unique antenna - Right: Interferometric EMIR arrangement, here with two antennas. 6. EXAMPLES OF NDE APPLICATIONS Two examples will be given to illustrate the advantages of the new method. They are obtained with an Agema 782 camera, with the lock-in system from Onera/Cedip 5. We use the ellipsoidal antennas whose near field was characterized by the present interferometric method (see above). The first example deals with a 5 mm-thick low absorptivity foam sample, stuck on a 5 mm-thick PC layer. The foam contains, at a depth of 25 mm, a 2 mm-dia. metallic disc (Fig. right).

7 The sample is positioned between the two antennas (Fig. 9 right). The film is placed at cm from the sample. The EM frequency is GHz. The transmission method (just the antenna #2 is active) and the interferometric one (two antennas active) are applied. Results are given in fig.. In transmission mode, the diffraction pattern due to the insert is clearly visible, but rather intricate. For a diagnostic on the shape and size of the defect, it is necessary to make theoretical calculations and compare them to the experiment. With interferometry, the thermal images show with a better accuracy the location, shape and size of the embedded defect (see intensity and phase distributions in Fig. 2) and it is possible to adjust the phase lag to enhance the contrast due to the defect. 2 cm 9 mm EPDM foam Eccosorb LS 2! r =.5 - j.69 Eccosorb LS 3! r = j 5 2 cm 7 cm 7 cm 3 mm metallic disc PC layers Fig. Samples with artificial defects. Left: radar absorbing material; right: EM transparent material 2 3 : Intensity from transmission measurement 2 : Intensity from interferometry 3 : Phase from interferometry Fig. - EMIR NDE of the low absorption sample of fig. (right). Amplitude modulation and lock-in detection at Hz with an Agema 782. Comparison between transmission and interferometric methods

8 ,5 Intensity (a.u.) Intensity (a.u.) 36 Phase ( ) 27,25,5 8 Pixels 5 5 pixels Pixels 5 5 Fig. 2 - Horizontal distributions of amplitude and phase from the images of Fig.. From left to right: i) amplitude from transmission measurement, ii) and iii) amplitude and phase by interferometric method. The second example concerns a radar absorbing material containing an artificial defect also made of a different absorber (see Fig. left). Fig. 3 presents the EMIR images obtained with the same set-up used in the first example, in transmission and interferometric modes. In spite of the intricate pattern of the field the presence of the internal non homogeneity is visible with both methods. Nevertheless, the defect presence is more clearly revealed by interferometry, especially in the phase image. The location and size of the defect is given with a better spatial resolution. 2 3 : Intensity from transmission measurement 2 : Intensity from interferometry 3 : Phase from interferometry Fig. 3 NDE of the highly absorbing sample of fig. (left). Amplitude modulation and lock-in detection at Hz with an Agema 782. Comparison between transmission and interferometric methods

9 Intensity (a.u.),8,6,4,2 x (cm),88 3,76 5,64 7,52 Phase ( ) x (cm),88 3,76 5,64 7,52 Fig. 4 NDE of the absorbing sample (fig. - left) by interferometric method. Comparison between steady state and amplitude modulated (.75 Hz) measurements. Horizontal distributions of amplitude (right) and phase (left). : steadystate; : amplitude modulation (from interferometric images of fig. 3). Fig. 4 gives amplitude and phase distributions along the horizontal central axis of the interferometric images of Fig. 3. They are compared to steady-state (no amplitude modulation of the EM field) interferometric results. The interest of amplitude modulation and lock-in detection clearly appears. This technique greatly enhances spatial resolution. This is particularly true for the phase. It also avoids distortions due to convection, which is not well visible here, since Fig. 4 presents only horizontal distributions. The interesting fact is the better signal to noise ratio in the phase image, if we consider the edge effect as a noise. In effect, the peak corresponding to the defect is near of 2 in phase and the edge effect only 4, which correspond to a ratio of 3. This pattern is similar for the intensity distribution, but weaker (the ratio is of 2). The edge effect is due to the fact that the sample is smaller than the EM field. This configuration is not favourable to interpret the images and to detect defects. A way to avoid them consists in making the difference between the fields related to the defective sample and a reference sound sample. Of course, this is not always possible. Fig. 5 presents results of such a procedure. The same sample (Fig. left) is used in conjunction with a sound sample (same shape, dimensions and material). The first row presents crude intensity and phase images obtained with the sample with defect. These images are much more noisy than those of Fig. 3. This is due to the use of a thicker film (75 µm). In effect, in modulated regime, the sensitivity of the film is inversely proportional to its thickness. Due to this poor signal to noise ratio, a classical image processing was applied, with the successive operations: - general noise reduction by morphologic low pass filter with a 7x7 kernel - low frequency and low level noise reduction by thresholding on 8/495 levels - re building of spoiled edges by 4 iterations of dilating/eroding process. The resulting images are given in the second row. The same procedure has been applied for the images resulting from the intensity and phase differences between defective and sound samples. The resulting images are given in the third row. Central horizontal and vertical distributions, taken from the processed images, are given in Fig. 6. The vertical and horizontal dimensions of the defect can be evaluated from the distributions given in this figure. The defect corresponds to a plateau (Fig. 6 b, c, d) or to a flat valley (Fig. 6 a). The horizontal and vertical defect extensions found by this means are given in Table I.

10 Fig 5 NDE of the highly absorbing sample of Fig. (left). First column : intensity maps (arbitrary linear unit), second column (phase image, arbitrary unit). First row : crude images showing the important noise due to the large thickness of the film (roughly 75 µm), second row : image processing result showing the interest of the procedure for defect detection, third row : processed difference images showing the edge effect pattern attenuation.

11 a c 256 Intensity (levels) H H x and y (cm) ! Intensity (levels) 256 H b d Relative phase (levels) 256 H H x and y (cm) ! Phase (levels) 256 H x and y (cm) x and y (cm) Fig 6 Horizontal (H) and vertical () distributions taken from images of Fig 5. a - Intensity profiles after image processing ; b - Relative phase profiles after image processing ; c - (intensity) (difference in intensity between defective and sound samples) profiles after image processing ; d - (Phase) (difference in relative phase between defective and sound samples) profiles after image processing parameters deduced sample with defect! with & without defect real defect from Fig. 5 and 6 intensity map phase map intensity map phase map horizontal extension (cm) vertical extension (cm) area (cm2) area ratio,45,585,425,52 Table I Defect extension evaluated from interferometric EMIR images and compared to the real ones. In both cases - measurement on defective samples alone and difference between defective and sound samples measurements -, location and approximate extension of the defect can be easily evaluated. The measured dimensions are always smaller than real ones. This is normal, since the film is not in the plane of the defect, but positioned at a distance of roughly cm from the defect (see Fig. 7). It is clear that the accuracy of the diagnostic is improved when the distance between film and structure surface decreases. The shape of the present defect, a square, is better correctly found on both amplitude and relative phase images with the first approach (unique measurement on defective sample). This is probably due to the lower signal to noise ratio of the difference images (not given here). On the contrary, the detection of the defect for human eyes is easier with the second approach. Thus, it can be recommended to choose the first approach for quantitative evaluation of defects and the second one for qualitative human inspection. For both approaches, the phase images lead to better evaluation of shape and dimensions of the defect than intensity images.

12 2 3 3 mm film 2 sample 3 defect Fig 7 Respective positions of film and defect inside structure (sample of Fig. left). From these two examples of NDE application of interferometry, we can state that: i) interferometry leads to better results than transmission method which only provides the field intensity, ii) phase images give information on defects with a better spatial resolution and better signal to noise ratio than amplitude images, iii) for qualitative inspection and when possible, it is worth to use differential images (difference between defective and reference sound samples). 7. CONCLUSION Since it has been demonstrated 6 that NDE by EMIR method reduces to map the EM field in the vicinity of tested structures, it was interesting to applied for NDE purposes the last developments occurred in the EM field mapping EMIR techniques, and in particular the interferometric method 2,3,4 which now permits to image both EM field intensity and phase. The new method proved to be better than classical EMIR technique. The use of phase maps instead of amplitude maps leads to a better spatial resolution and to a higher contrast for defects detection. Furthermore, preliminary tests using a focal plane array infrared camera showed a strong enhancement of the sensitivity. In the near future, this type of camera will be used for NDE purpose. A noticeable enhancement of the performances of the EMIR NDE methods can be reasonably supposed. 8. REFERENCES. D.L. Balageas, P. Levesque, M. Nacitas, J.-C. Krapez and G. Gardette, "Photothermal mapping of electromagnetic fields: state-of-the-art", Proc. IXth Int. Conf. on Photoacoustic & Photothermal Phenomena, Nanjing (China), June 996, to be published as a suppl. of Progress in Material Sc. (China). 2. M. Nacitas and P. Levesque, "Mesure d'un champ électromagnétique complexe par interférométrie microonde associée à la thermographie infrarouge", 9 e Journ. Nat. Microondes, Paris, April D.L. Balageas, P. Levesque, M. Nacitas, "Microwaves holography revealed by photo-thermal films and lock-in IR thermography", Proc. IXth Int. Conf. on Photoacoustic & Photothermal Phenomena, Nanjing (China), June 996, to be published as a suppl. of Progress in Material Sc. (China). 4. M. Nacitas, Contribution to development of non destructive testing methods for semi-transparent materials, by using microwave interferometry.combined with infrared thermography, PhD thesis, Limoges University, Nov D.L. Balageas, P.Levesque and A. Déom, "Characterization of electromagnetic fields using lock-in IR thermography", Thermosense X, SPIE vol. 933, pp , P. Levesque, A. Déom and D.L. Balageas, "NDE of absorbing materials using microwave stimulated infrared thermography", in Quantitative Infrared Thermography (QIRT'92), ed. D. Balageas, G. Busse, G.M. Carlomagno, Editions Europ. Thermique et Industrie, Paris. 92, pp , 992.