Broadband Microwave Interferometry for Nondestructive Evaluation

Transcription

1 13th International Symposium on Nondestructive Characterization of Materials (NDCM-XIII), May 2013, Le Mans, France Broadband Microwave Interferometry for Nondestructive Evaluation More Info at Open Access Database Kamel HADDADI, Tuami LASRI Institute of Electronics, Microelectronics and Nanotechnology, University Lille 1 Sciences and Technologies, Villeneuve d Ascq, France Phone: , Fax: kamel.haddadi@iemn.univ-lille1.fr Abstract A broadband microwave interferometric technique for scanning near-field microscopy applications is proposed. The method is based on the association of a vector network analyzer, a hybrid coupler, an evanescent microwave coaxial probe and a precise impedance tuner built up with a high precision programmable delay line and a variable motor-driven attenuator. Advantages such as simplicity of operation, broad frequency band capabilities and high measurement sensitivity are achieved. In particular, the measurement sensitivity, the depth and lateral resolutions of the microwave microscope are experimentally verified to evaluate the performances of the system. The usefulness and versatility of the method make it also suited for a wide range of applications. Keywords: Near-field microwave microscopy, evanescent probe, interferometry, nondestructive testing. 1. Introduction Free-space microwave sensing methods for non-destructive testing applications have been widely described in the literature and some of them are now well established [1]-[2]. In most of this kind of characterization techniques, the antenna/probe is connected to a 50Ω measurement system (i. e. vector network analyzer, reflectometer). To improve the spatial resolution of these methods, limited to a fraction of the wavelength, microwave microscopy techniques based on high impedance near-field probes have been introduced [3]-[5]. The future of near-field microwave microscopy has promising directions. In particular, micro- and nano-imaging, high speed nanodevices probing, measurement of atofarad capacitances, biological systems imaging in their saline solutions are well-known scientific challenges. Recent contributions in these domains have mainly concerned the development of probe geometries to reduce the tip stray capacitance found in many microscopes. Most of the reported microwave microscopes employ a resonator circuit to reduce the impedance mismatch and increase the measurement sensitivity. From the measurement system point of view, the use of a conventional 50 Ω instrument is generally thought as a unique solution. The major challenge of this work is to develop a new type of measurement system to meet the specific needs of nondestructive characterization tools for impedances different than the 50Ω reference impedance. The instrumentation proposed is conceived to match the impedance of the probe at any frequency [6]. In contrast with conventional resonator methods, the impedance of the probe can be adjusted precisely with a passive, low-noise and fully programmable impedance tuner. Moreover, the tip stray capacitance of the probe is also compensated by the proposed system, releasing the constraints on the design and realization of the probes. To increase the measurement sensitivity on small impedance contrasts detection, the system can operate in any measurement configuration. The principle of operation and design considerations of the proposed network analyzer, based on a broadband interferometry technique, is described in Section 2. In Section 3, the experimental set-up developed in the frequency band 1-20 GHz is presented. To evaluate the performance of the system proposed, basic experiments related to the scanning of micro-sized patterns are proposed in Section 4.

2 2. Broadband microwave interferometry technique The main measurement limitation when using evanescent near-field microwave probes is the impedance mismatch between the intrinsic impedance of the network analyzer close to 50 Ω and the impedance of near-field probes that is in the range of tens or hundreds of KΩ. The illustration is given in Fig. 1 considering the general block diagram of the vector network analyser (VNA) in a one-port measurement configuration. The technique, often called reflectometry, is based on the use of power dividers or/and directional couplers connected to the microwave source and to a device under test (DUT) with impedance Z. In Fig. 1(a), one part of the reference wave (denoted R) is collected by the power divider whereas the coupler separate the reference wave from the reflected one (denoted A). The resulting reflection coefficient measured by the VNA is expressed as Z Z = Z + Z A R Γ 0 = with Z 0 =50 Ω (1) 0 It is well-known that the measurement sensitivity is maximum for impedances around 50 Ω. But when very high impedances (or very low) are considered, most of the incident wave on the DUT is reflected back so that the VNA becomes insensitive to the variations of the reflection coefficient Γ. So, there is an urgent need to extend the VNA measurement capabilities for impedances greater than the kω. The proposed interferometric technique consists to cancel the reflected wave A at the frequency of interest. To that end, a high impedance reference device (with impedance Z REF ) is connected to the coupled port of the hybrid as illustrated in Fig. 2(b). The impedance Z REF is adjusted to cancel the total reflected wave A. Consequently, the waves reflected respectively by the device under test (with impedance Z) and the reference device have the same magnitude, but are phase-shifted by 180. The total reflected signal and the resulting reflection coefficient are theoretically zero. In fact, this technique brings the high impedance to 50 Ω. R A Z Γ (a) R A Γ REF Z Z REF (b) Figure 1. Configurations for the measurement of microwave impedances. (a) Generalized network analyzer set-up in a one-port configuration. (b) Modified network analyzer set-up based on the interferometric principle for the measurement of high impedances. Γ

3 Ideally, the cancellation technique should be done under any configuration in terms of operating frequency, impedance, and level of the magnitude of the reflection coefficient. To that end, an experimental modified VNA architecture based on the interferometric architecture is depicted in Fig. 2. PNA-X N5242A Agilent Technologies 10 MHz to 26.5 GHz 3dB, 90 hybrid coupler Anaren to 4 GHz S 21 High-resolution delay line Colby Instruments PDL 200-A DC to 18 GHz Variable attenuator Radiall R to 18 GHz Motor-driven variable attenuator ATM AF 074H to 18 GHz To high impedance Z SMA short termination Figure 2. Experimental set-up based on the proposed interferometric method for the microwave measurement of high impedances. The measurement system is a broadband network analyzer PNA-X N5242A. A coaxial 3dB, 90 hybrid coupler is connected to the port 1 of the VNA. The direct and coupled ports are connected to the high impedance device under test (with impedance Z) and to the impedance tuner respectively. The impedance tuner is built up with a high-resolution programmable delay line (Colby Instruments PDL-200A Series) connected to a motor-driven variable attenuator (ATM AF 074H-10-28). The output port of the attenuator is shortcut with a SMA short termination. A mechanically variable attenuator is inserted between the direct port of the hybrid coupler and the high impedance device under test to compensate the losses in the impedance tuner. It has to be mentioned that the directivity, source match and load match errors of the hybrid coupler are easily taken into account by the interferometric method. In other words, the measurement performance of the classical reflectometry set-up [Fig. 1(a)] depends on the electrical performance of the hybrid coupler in terms of matching properties and directivity. In the solution proposed, the directivity and mismatching effects can be easily compensated by the interferometric method, relaxing the constraints on the measurement performance of the hybrid coupler. In Section 3, a proof-of-concept near-field microwave microscopy is developed on the basis of the inteferometric principle.

4 3. Experimental near-field microwave microscope A promising application of near-field microwave microscopy methods is sensing for nondestructive testing and evaluation (NDT&E) [7]. Traditional NDT&E techniques are based on transmission line or free-space methods [8]. In these techniques, the spatial resolution achievable, set by the diffraction limit, is in the order of half the wavelength of operation. Thus, in the microwave and millimeter-wave spectrums spanning from 1 to 300 GHz, the spatial resolution is at best in the order of the millimetre. So, to bypass the diffraction limit, microscopy methods based on evanescent waves rather than propagating ones can be used [9]-[10]. This allows addressing the growing need for local characterization tools with sub-wavelength resolution. To address the urgent need of novel non-destructive techniques and dedicated instruments with high measurement performance in terms of sensitivity, accuracy and spatial resolution, a near-field microwave microscope based on microwave interferometry is proposed. A coaxial evanescent microwave probe (EMP), made in tungsten with an apex of 2 µm, is connected to the measurement port [Fig. 3(a)]. The sample to be scanned is mounted on a motor-driven xyz stage. The probe is mounted vertically over the stage. The xyz stage consists of three independent motorized linear translation sub-stages with respective travel distances of 25 cm in x/y axis and 1 cm in z axis. The minimum increment step in the three directions is 1 µm. Consequently, the sample is positioned on the chuck mounted on the stage whereas the microwave part of the microscope remains fixed. In contrast, with previous works related to the development of near-field microwave microscopes, a better stability is obtained by moving the sample under the probe tip instead of moving the probe. As the reflection coefficient S 21 is acquired from only a small part of the sample at one time, the magnitude and phase-shift images of the reflection coefficient are obtained from pixel-by-pixel scanning. (a) (b) Figure 3. Near-field microwave microscopy platform based on interferometry. (a) Photograph of the coaxial evanescent microwave probe. (b) Photograph of the microwave microscope. Concerning the software part of the platform, a National Instruments Labview interface is developed to control the position of the sample, to set the network analyzer parameters, to measure the resulting transmission coefficient S 21 and to display the results.

5 As the measurement accuracy and sensitivity are mainly governed by the intermediate frequency (IF) bandwidth of the VNA, the transmission coefficient noise floor is first determined as a function of the IF bandwidth. To that end, the source power of the VNA is set to 0 dbm and both measurement ports of the VNA are connected to match loads. The measured transmission noise floor corresponds to the average of the measured magnitude transmission on the whole frequency band 10 MHz-26.5 GHz (Fig. 4). Figure 4. Measured transmission coefficient as a function of the intermediate frequency (IF) bandwidth. From this graph, low-level noise can be reduced by decreasing the IF bandwidth with a minimum value around -105 db for an IF bandwidth of 1 Hz. In the following, the potentialities of the technique are verified through a basic experiment that consists to measure the magnitude spectra of the transmission coefficient S 21 for a wave cancellation performed at the arbitrary test frequency 2.5 GHz. In this configuration, the probe is matched in free-space. All measurements are done at room temperature around 20 C. In Fig. 5, we present the magnitude spectra of the transmission coefficient measured for different IF bandwidths. As expected, when lower IF bandwidths are considered, high quality factors are achieved with lower noise. In particular, Fig. 5(d) shows a magnitude spectrum with a zerotransmission level at -110 db. Nevertheless, there is a compromise between the measurement accuracy and the related acquisition time. Depending on the application required, the IF bandwidth, the frequency band of operation and the number of frequency points can be adjusted.

6 Transmission coefficient (db) Frequency deviation f (MHz) (a) Transmission coefficient (db) Frequency deviation f (MHz) (b) (c) (d) Figure 5. Measured magnitude spectra of the transmission coefficient S 21 at the test frequency 2.5 GHz. (a) IFBW = 100 KHz. (b) IFBW = 10 KHz. (c) IFBW = 10 Hz. (d) IFBW = 1 Hz. 4. Experimental lateral and depth resolution A major advantage of the solution proposed is the possibility to tune the resonance frequency and the quality factor at the desired distance to the material under test. In contrast with conventional resonators that exhibit a high unloaded quality factor that fails in the presence of the material under test, the interferometric process can be performed in the presence of the material to achieve a high electromagnetic coupling. In the following, the sensitivity of the transmission coefficient to the probe-to-object separation h is studied for three measurement configurations. The sample is a silicon wafer recovered by a thin gold layer. The test frequency is set to 2.5 GHz. In the first configuration, the attenuation of the impedance tuner is set to its maximum to make the transmission coefficient depending only of the probe matching properties. The second and third configurations correspond respectively to the probe matched in free-space and at 10 µm above the sample by means of the interferometric technique. The sample is moved from the distance h = 10 µm to h = 110 µm with an increment step of 10 µm. Fig. 6 presents the distance dependence of the magnitude and the phase-shift spectra of the transmission coefficient S 21 for the three cases considered.

7 (a) (b) (c) (d) (e) (f) Figure 6. Measured magnitude and phase-shift spectra of the transmission coefficient S 21 for different probe-to-sample separations h. F = 2.5 GHz. (a) and (b) Probe only. (c) and (d) Probe matched in free-space. (e) and (f) Probe matched at 10 µm above the sample. ( 10 µm, 20 µm, 30 µm, 40 µm, 50 µm, 60 µm 70 µm 80 µm 90 µm, 100 µm, 110 µm)

8 The results demonstrate that the matching properties of the probe are impacted by the probe-to-sample separation. Fig. 6(a) and Fig. 6(b) indicate that the variation of the transmission coefficient are too small to be measured with good accuracy. Indeed, as mentioned in Section 2, the VNA is not enough sensitive to measure impedance variations around the high impedance of the near-field probe. Fig. 6(c) and Fig. 6(d) show a variation of the magnitude of S 21 around 10 db. For this case, the probe that has been matched in free-space exhibits a measurement sensitivity comparable to the one offered by conventional resonant based microwave microscopes. Thus, when the distance separation decreases, the sensitivity is affected. To increase the measurement sensitivity, the transmission coefficient S 21 is tuned to a very low value when the probe is set to 10 µm above the sample [see Fig. 6(e) and Fig. 6(f)]. We note that a high microwave coupling between the sample and the probe is achieved. When the probe-to-sample separation increases, the resonance frequency is shifted and the quality factor diminishes drastically. Thus, this technique provides sensing of the local electromagnetic properties of samples with high sensitivity. In this configuration, the measured magnitude of the transmission coefficient S 21 grows with the distance separation from h = 10 µm (cancellation condition) to h = 110 µm from -74 db to -62 db at 2.5 GHz (Fig. 7). In the same manner, the phase-shift of the transmission coefficient S 21 grows from -120 to -40. In this last configuration, to benefit from the measurement sensitivity of the microscope, the stand-off distance must be kept as small as possible. Phase-shift (deg) Standoff distance (µm) (a) (b) Figure 7. Measured magnitude and phase-shift of the transmission coefficient S 21 as a function of the probe-to-sample separations h. F = 2.5 GHz. (a) Magnitude of the transmission coefficient. (b) Phase-shift of the transmission coefficient. ( Measurement nd order Polynomial Modeling). From Fig. 7, the experimental data are fitted by a 2 nd order polynomial model to derive the magnitude and phase-shift of the transmission coefficient as a function of the standoff distance h. 2 S = h h (2) arg(s 21 + db ) deg 2 == h h (3) 21 + From the relations (2) and (3), we observe a good measurement sensitivity in both magnitude and phase-shift of the transmission coefficient to the standoff distance h. When h increases, the probe-to sample electromagnetic coupling and the measurement sensitivity decrease according to the 2 nd order terms of the relations (2) and (3).

9 In the following study, we propose to show the potentialities of the system through basic experiments to yield the spatial resolution. The lateral resolution (along x-axis) is governed by the distance separation between the probe and the object under test [3]. The structure under test is a hole of 0.8 mm drilled in a copper square patch with thickness 35 µm supported by an epoxy plate (thickness 0.8 mm). In Fig. 8, we present the measured data (magnitude and phase-shift of the transmission coefficient) obtained by linearly scanning the structure under test for three distance separations. Compared to the apex size of the probe (2 µm), one is in the same order (h = 10 µm) and the two other ones are respectively h = 60 µm and h = 110 µm. The probe is preliminary matched at a standoff distance of 10 µm above the copper at a magnitude level of the transmission coefficient of db at 2.5 GHz. In the study, a linear scanning increment of 50 µm is considered to provide 32 points for a scanning distance of 1.6 mm. Transmission coefficient (db) Phase-shift (deg) Ox (µm) Ox (µm) Figure 8. One-dimensional scan of a hole of 0.8 mm drilled in a copper square patch with thickness 35 µm supported by an epoxy plate (thickness 0.8 mm) - F=2.5 GHz. ( h=10 µm, h=60 µm, h=110 µm) The measured data for the three probe-to-object separations show that the magnitude and phase-shift of the transmission coefficient allow detection and localization of the sample considered. When h is set to 60 µm and 110 µm, the magnitude and the phase-shift variations get smaller. When the probe-to-object separation increases, the collimation of the field diminishes and the one-dimensional scan gives broadened responses of the magnitude and the phase-shift. Thus, the probe-to-object separation must be kept to a relative low constant value to achieve a very good collimation of the field. In this case, the lateral resolution can be readily improved. To resume these experiments, the probe-to-object separation has to be kept in the near-zone to achieve a good depth resolution (along z-axis). The lateral resolution (along x-axis) achievable is limited by the smallest dimension of the apex size of the probe (2 µm).

10 5. Conclusion A technique for the microwave measurement of high impedances was described. The resulting instrumentation combines a vector network analyzer and a high precision interferometer. The method provides advantages such as broadband capabilities, high measurement accuracy and simplicity of operation. Based on this technique, a near-field microwave microscope for nondestructive applications has been experimentally validated. The combination of a vector network analyzer and the proposed interferometer presents a viable and promising alternative to address the needs of characterization tools for a wide range of applications. References 1. S. Kharkovsky and R. Zoughi, Microwave and millimeter wave nondestructive testing and evaluation overview and recent advances, IEEE Instrum. Meas. Mag., Vol. 10, No. 5, pp , April K. Haddadi, MM. Wang, O. Benzaim, D. Glay and T. Lasri, Contactless microwave technique based on a spread-loss model for dielectric materials characterization, IEEE Microw. Wireless Compon. Lett., Vol. 19, No. 1, pp , January S. M. Anlage, V. V. Talanov, and A. R. Schwartz, Principles of near-field microwave microscopy, Scanning Probe Microscopy: Electrical and Electromechanical Phenomena at the Nanoscale, S. Kalinin and A. Gruverman, Eds. New York: Springer Sci., pp , S. Fabiani, D. Mencarelli, A. Di Donato, T. Monti, G. Venanzoni, A Morini, T. Rozzi, and M. Farina, Broadband Scanning Microwave Microscopy investigation of graphene, Proc. IEEE MTT-S International Microw. Symp., pp. 1-4, June K. Haddadi, D. Glay, and T. Lasri, A 60 GHz scanning near-field microscope with high spatial resolution sub-surface imaging, IEEE Microw. Wireless Compon. Lett., Vol. 21, No. 11, pp , November K. Haddadi and T. Lasri, Interferometric technique for microwave measurement of high impedances, Proc. IEEE Int. Microw. Symp., pp. 1-3, Montreal, Canada, June K. Haddadi and T. Lasri, the multi-port technology for microwave sensing applications, Proc. IEEE Int. Microw. Symp., pp. 1 3, Montreal, QC, Canada, June S. Trabelsi and S. O. Nelson, Free-space measurement of dielectric properties of cereal grain and oilseed at microwave frequencies, Meas.Sci. Technol., Vol. 14, No. 5, pp , May A. Karbassi, D. Ruf, A. D. Bettermann, C. A. Paulson, D. W. Van der Weide, H. Tanbakuchi and R. Stancliff, Quantitative scanning near-field microwave microscopy for thin film dielectric constant measurement, Review of scientific instruments, Vol. 79, No. 4, pp , September M. Wang, K. Haddadi, D. Glay and T. Lasri, Compact near-field microwave microscope based on the multi-port technique, Proc. 40th European Microw. Conf., pp , Paris, France, October 2010.