Micromachined ultrasonic transducers for air-coupled

Micromachined ultrasonic transducers for air-coupled non-destructive evaluation Scan 'F. Hansen. F. Levent Degertekin. and Butrus '1'. Khuri-Yakuh Edward L. Ginzton Laboratory Stanford University Stanford. CA 9435-485 ABSTRACT Conventional methods of ultrasonic non-destructive evaluation (NDF.) use liquids to couple sound waves into the test sample. This either requires immersion of the parts to he examined or the use of complex and bulky water squirting systems that must be scanned over the structure. Air-coupled ultrasonic systems eliminate these requirements if the losses at air-solid interfaces are tolerable. Micromachined capacitive ultrasonic transducers (cmtj'l's) have been shown to have more than 1 db dynamic range when used in the bistatic transmission mode. In this paper. we present results of a pitch-catch transmission system using cmuts that achieves a 13 d13 dynamic range. Each transducer consists of 1. silicon nitride membranes of 1 l.lm diameter connected in parallel. This geometry results in transducers with a resonant frequency around 2.3 Mhz. These transducers can be used in through transmission experiments at normal incidence to the sample or to excite and detect guided waves in aluminum and composite plates. In this paper we present ultrasonic defict detection results from both through transmission and guided Lamb wave experiments in aluminum and composite plates. such as those used in aircratl. Keywords: transducer, air, NDE, Lamb waves 1. INTRODUCTION Due to the large impedance mismatch between common piezoelectric materials iid air, conventional piezoelectric transducers are not very efficient sources of ultrasound in air. Therefore many conventional ultrasonic testing systems require complicated liquid squirting systems or immersion of the sample to be examined, which may not he practical for large aerospace structures. An alternative is to place the transducer in direct contact with the samples under test. I however. defect detection or imaging can be time consuming. and fragile samples may he damaged by direct contact. In such situations, air boime ultrasound is a useful non-contact method of detecting and imaging defects in aging aircraft structures, given efficient transducers for use as transmitters and receivers. 2.1. Transducer Description 2. TRANSDUCERS Recently, several capacitive micromachined ultrasonic transducers (cmhjts) have been developed, which are capable of efficient excitation and detection of ultrasound in air4. As depicted in Fig. I. a single clement consists of a metalized.5-1 p.m thick nitride membrane suspended above a silicon substrate. Several thousand such elements are electrically connected in parallel to make the transducer, shown in Fig. 2. When the membranes are biased with a [)(' bias voltage, the transducer is capable of efficient excitation and detection of ultrasound in air. The resonant membrane structure results in a large dynamic range. in excess of 1 db. though the transducers are inherently narrow hand devices with fractional bandwidths of less than 1%. Fig. I. Schematic cross section of a single cmf Ti membrane. Fig. 2. Magiiilied view of cmi II transducer with 5 p.m membrane radii. Part of the SPIE Conference on Nondestructive Evaluation of Aging Aircraft, Airports, 31 and Aerospace Hardware Ill Newport Beach, California March 1999 SHE Vol 3586 277-786X/99/$1O.OO

2.2. System Dynamic Range Simulations Using an equivalent resonant circuit model for the impedance of the cmut, we simulate the transducer in a pitch-catch transmission configuration. Complete signal-to-noise analysis is possible using SPICE circuit simulation. This analysis accounts for non-ideal losses through capacitances, insertion losses within the transducer, and for the thermal noise of components. For the simulations, we apply a 2 V peak-to-peak tone burst through a tuning inductor to the transmitting transducer, which is under 5 V DC bias. The receiving transducer is connected through another tuning inductor to the AD6 low-noise amplifier by Analog Devices. This amplifier provides 4 db of gain with an input noise voltage of I.4 nv/'ihz. Using a generous noise bandwidth of 2 MHz around the transducer resonance, the ratio of the maximum received voltage at the amplifier input to the noise voltage represents our dynamic range. For a transmit and receive system that uses inductors to tune out the transducer reactances on send and receive, simulations predict a dynamic range of 1 9 db for the 2 MHz noise bandwidth implemented in our system. If the noise bandwidth can be further reduced to 1 MHz, the dynamic range increases to 116 db. 3.1. Transmission Through Composite Plate 3. TRANSMISSION IMAGING RESULTS Simulations of an electronic system using cmuts suggest that the dynamic range of the system is sufficient to transmit ultrasound through materials such as metal and composites in air, without the need for coupling fluids. Figure 3 shows the received tone from bistatic transmission through a four-layer carbon fiber composite plate at normal incidence. By varying the amplitude of the received signal with and without the sample present, we estimate that the composite plate and its interfaces with air attenuate the ultrasound signal by 68 db. A.7 cm air gap further attenuates the signal by 6 db at 2.3 MHz. Since the received signal without averaging has a signal-to-noise ratio of 29 db, Fig. 3 demonstrates a dynamic range of 1 3 db, 6 db less than the electronic simulation prediction. However, other sources of noise such as electromagnetic interference and supply voltage noise sources were not accounted for in the SPICE circuit simulation. Nonetheless, the system's dynamic range is more than adequate to image impact damage in a section ofthe composite sample. > E C', C ) C') ci) ci) 2 22 24 26 28 3 32 Fig. 3. Received voltage waveform through composite plate. Figure 4 shows an image produced by linearly gray-scaling the amplitude of the received signal over a section of composite. Density variations in the plate are evident by regions of light and dark. Figure 5 is produced by scanning the same section ofthe plate after it was struck in two places, which appear in the scanned image as dark spots. The damage spot on the left is not visible optically from the surface, but is clearly present in the ultrasound image. 311

6 E 5r 4-3 3 E 2-2F 7. 7.- ir- 1-1 2 3 4 5 6 7 1 2 3 4 5 6 7 o Fig. 4. Amplitude image of composite prior to Impact damage. Fig. 5. Amplitude image ol conipo ite alter impact damage int places. 3.2. Transmission Through Aluminum Plate The image in Fig. 6 shows the amplitude variations in a 3 mm thick aluminum l)late that has a.5 mm deep pattern. shown in Fig. 7, milled on the underside of the plate. The dynamic range in this transmission experiment is also 13 d13. with about 82 db of loss through the aluminum and air interfaces at 2.3 Ml lz..5 cm 3 cm cm Fig. 6. Amplitude image ofmilled pattern in aluminum plate. lig. 7. Mitled pattern on underside of plate. ((.5 mm deep. 312

4. LAMB WAVE DEFECT DETECTION The ultrasound transmission experiments of the previous section are useful for scanning small-sized samples that have both sides of the material exposed to air. Generation and detection of Lamb waves from air into plates is a more promising method for inspecting materials and structures in which only one side of the sample is accessible. Furthermore, large areas may be scanned rapidly since Lamb waves are capable of traveling relatively long distances once coupled into the solid. Any interaction of the Lamb wave with a defect can be detected by the receiving transducer, which may be near or far from the transmitter as long as it is not in the path of a specular reflection. In the experiments that follow, we present calculations and experimental results ofthe symmetric s mode propagating in an aluminum plate that is 1.2 mm thick. 4.1 Lamb Wave Propagation Calculations To excite a Lamb wave in a solid from an acoustic wave in fluid, the acoustic wave in air must strike the surface of the solid at specific incidence angles for the supported mode. The plane wave reflection coefficient calculation shown in Fig. 8 for a 1.2 mm thick aluminum plate and 2.3 MHz ultrasound suggests that three propagating modes are possible, shown by three dips in the reflection coefficient ofthe aluminum. The mode at 5.8 degrees corresponds to the lowest order symmetric so mode. I A1 Mode S A Mode -c, :, C) C3 a) ) ci) ci) O.9999S ' I I I 1 2 3 4 5 6 7 8 9 1 Angle of Incidence (degrees) Fig. 8. Reflection coefficient from 1.2 mm thick aluminum plate for range of incident angles. To estimate the conversion efficiency of acoustic power from air into the s mode, we must consider the size of the transmitting and receiving transducers. The expression for conversion efficiency r, for one-sided excitation or reception, is given by = (1 e)2 (1) where ct is the leak rate of the mode and 1 is the projected transducer length in the direction of propagation. 6 For our transducer length of 1 cm, projected length ofo.995 cm for a 5.8 angle ofincidence, and a leak rate ofo.85 Np/m for the s mode, the calculated conversion efficiency is 8.45x14. Squaring this efficiency represents conversion from air to solid, and back to air again, yielding the two-way conversion efficiency. The optimal sized transducer for efficient coupling for this attenuation constant satisfies al=l.26 and is about 12 m in length. For a more practical 1cm transducer, we expect 61.5 db 313

for the two-way coupling loss to and from the solid plate into air for this incident angle and mode. Ifwe again assume a 13 db dynamic range for our system, we expect our received signal-to-noise ratio to be 41.5 db minus any attenuation in the air gap between the solid and transducer. 4.2 Lamb wave Transmission Results The 2.3 MHz transducers are set in the configuration shown in Fig. 9, with incidence angles near 5.8 degrees to excite and detect a Lamb wave in the 1.2 mm aluminum plate. An amplified voltage signal from the receiver is shown in Fig. 1. The peak signal, which occurs about 48 ts after the transmit tone burst, represents the s Lamb wave which travels about 3 cm in the aluminum prior to detection by the receiver. Received signals at 295 ts and again at 538 ts are due to multiple reflections of the Lamb wave from the plate edges, one of which is still detected after traveling distance of more than 1 m in the plate. The group velocity calculations using these multiple echoes confirm that the Lamb wave is indeed the s mode. Transmifter [ cm Receiver I 17cm TO.5cmT 4cm air gap Lamb Wave ------------------ 1 I1 i.2 mm Aluminum Plate Fig. 9. Configuration and geometry for Lamb wave coupling into aluminum plate. > E I 2 3 4 TimeAfter Transmit Burst (1.tS) Fig. 1. Received voltage signal showing multiple reflections ofs mode Lamb wave. An expanded view of the detected Lamb wave is shown in Fig. 1 1, averaged 1 6 times to reduce the noise level. With averaging, the peak signal is 15.41 mv rms while the noise level is 147 j.tv mis, yielding a signal-to-noise ratio of 4.4 db. Therefore, we expect a factor of 4, or 12 db, decrease in signal-to-noise ratio for a signal without averaging. Accounting for 8.7 db attenuation due to the air gaps, our actual signal-to-noise ratio of 28.4 db falls about 4.4 db short of the prediction. 314

Any unaccounted signal loss due to diffraction and Lamb wave attenuation should be negligible for the Lamb wavelength and transducer spacing of 3 cm. Therefore we attribute the 4.4 db discrepancy between the predicted and actual signal to noise to imperfect angular alignment of the transducers, noise due to additional amplification, and the finite size of the transducer in two dimensions. In the future, use of phase delayed arrays of the membranes to control the desired incidence angle could improve the control over mechanical alignment. 25 2 15 1 j.. h ii II I, 3 35 4 45 5 55 6 Time After Transmit Burst (i5) 65 7 Fig. 11. Received voltage signal from air-coupled s Lamb wave, averaged 16 times. Fig. 12. Configuration for line scan of defect using air Lamb wave. 315

4.2 Defect Detection Using Lamb Waves A measured signal-to-noise ratio of 29 db is sufficient to detect the presence of a thickness change in the aluminum plate, representing a defect. Figure 12 schematically shows the configuration and the transducer with respect to a.6 mm deep rectangular 1 cm by.3 cm rectangular milled pattern on the underside of the plate. A single 5 cm line scan shown in Fig. 13 shows the drop in received signal amplitude as the Lamb wave crosses the defect. The defect effectively reduces the plate thickness and changes the propagation efficiency of the s mode. The slight increase in received signal when the defect is directly centered between the transducers is likely due to edge effects around the defect, since its size is comparable to the transducer's field of view. By taking similar line scans from a variety of angles one can use Lamb waves to locate and reconstruct the defect shape using tomography.7 1 2 3 4 5 Scan Distance (cm) Fig. 13. Received signal from s, Lamb wave line scan over defect in aluminum plate. A Defect Receiver X-YScan, 9 x 9 cm T Fig. 14. Configuration for horizontal and vertical scan of defect in Aluminum plate. Lj Transmitter Aluminum Plate 316

Figure 14 shows the configuration for a Cartesian coordinate scan, in which the transducers scan in both horizontal and ertical directions to create an image. For the best resolution and replication of defect geometry. it is advantateous to osition the transducers as close as possible to one another. As shown in the defect image of Fig. 15, even a 3 cm separation ith 1 cm wide transducers results in an image that expands the.3 cm defect to nearly 4 m in the horizontal direction. lowever. the approximate defect size and location can still he ascertained if the geometry ol the scan is known. 9 Scale (V) 8 7 6r E 3 2 1 1 2 3 4 5 6 7 8 9 cm lig IS. Scaled amplitude image of defect in aluminum plaie using air coupled s9 I anib wave. 5. CONCLUSION Electronic circuit simulation suggests that cmuts are capable of dynaniic ranges of up to 116 db with the use of filters reduce noise and tuning inductors to improve power delivery to and from the transducer. In transmission experiments wough aluminum and carbon tuber samples. cmtjts have demonstrated a dynamic range of 13 db. This dynamic range. long with cmuts ability to transmit ultrasound into air, makes them ideally suited for air-coupled NDE applications. For ne-sided inspection of aircraft materials, the transducers ma also excite and detect Lamb waves coupled from air. We emonstrate the feasibility of Lamb wave defect detection for the 5 mode in a 1.2 mm thick aluminum plate, with a received ignal-to-noise ratio of approximately 28 db. Future work includes characterization (if detects in aircraft structures for ifferent Lamb wave propagation modes in aircraft structures. 13v phase delaying arrays of membranes in the transducer, we ope to be able to electron ical Iv select the incidence angle and desired propagating mode of 1.amh waves. ACKNOWLEDGMENTS his work is supported by WPAFB and the US Office of Naval Research. We also wish to thank Igal I.adabauni tcr ibricating the transducers. 31]

REFERENCES 1 F. L. Degertekin and B. 1. Khuri-Yakub, "Hertzian contact transducers for non destructive evaluation," J. Acoust. Soc. Amer. 99, pp. 299-38, 1996. 2. M. Castaings, P. Cawley, R. Farlow, and G. Hayward, "Single sided inspection of composite material using air coupled ultrasound," I. Nondestructive Evaluation 17, pp. 37-45, 1998. 3. M. I. Hailer and B. T. Khuri-Yakub, "A surface micromachined electrostatic ultrasonic air transducer," IEEE Trans. On Ultrasonics, Ferroelectrics and Frequency Control 43, pp. 1-6, 1996. 4. D. W. Schindel and D. A. Hutchins, "Applications of micromachined capacitance transducers in air-coupled ultrasonics and nondestructive evaluation," IEEE Trans. On Ultrasonics, Ferroelectrics and Frequency Control 42, pp. 5 1-58, 1995. 5. 1. Ladabaum, X. C. Jin, H. T. Soh, A. Atalar, and B. T. Khuri-Yakub, "Surface micromachined capacitive ultrasonics transducers," IEEE Trans. On Ultrasonics, Ferroelectrics and Frequency Control 45, pp. 678-69, 1998. 6. G. Kino, Acoustic Waves: Devices, Imaging, and Analog Signal Processing, Prentice Hall, Englewood Cliffs, 1987. 7. W. Wright, D. Hutchins, D. Jansen, and D. Schindel, "Air-coupled lamb wave tomography," IEEE Trans. On Ultrasonics, Ferroelectrics and Frequency Control 44, pp.53-59, 1997. 318