AN ANALYSIS OF ELECTROMAGNETIC ACOUSTIC TRANSDUCER ARRAYS FOR NONDESTRUCTIVE EVALUATION OF THICK METAL SECTIONS AND WELDMENTS* C.M. Fortunko and R.E. Schramm Fracture and Deformation Division National Bureau of Standards Boulder, CO 80303 ABSTRACT A new type of electromagnetic-acoustic transducer (EMAT) has been developed that may be particularly suitable for use as an element of ultrasonic arrays. The new transducer can generate and receive compact ultrasonic pulses that exhibit a component of polarization parallel to the free surface. In the plane of symmetry that is normal to the free surface and bisects the EMAT (the sagittal plane), the ultrasonic signals generated by the new transducer are SH waves. In addition, the new transducer can efficiently receive ultrasonic signals from a very wide range of direction in the sagittal plane. This property is required to realize very long synthetic aperture lengths, which are needed to maximize the transverse resolution of ultrasonic inspection systems. The focusing performance of different linear synthetic array configurations using the new EMAT is compared analytically with that of a linear end-fire system using periodic-permanent-magnet (PPM) EMATs that have been used in the past in weld inspection. The advantages and inherent limitations of such systems are examined using analytical and numerical methods. Particular emphasis is placed on the analysis of arrays that can be focused obliquely with respect to the free surface and the potential usefulness of such systems in weld inspection. INTRODUCTION The use of special ultrasonic probing signals, horizontally polarized shear waves (SH waves) in nondestructive evaluation (NDE) *Contribution of the National Bureau of Standards and not subject to copyright. 283 284 C. M. FORTUNKO AND R. E. SCHRAMM has been considered in the past by several authors in the context of weld inspection. 1- 6 Historically, the interest in SH waves was generated by the fact that the directivity patterns of SH waves produced by elementary sources on the free surface of an elastic halfspace are considerably less complicated than those of vertically polarized shear waves (SV) waves and longitudinal waves (L waves).7-10 This unique property of SH waves was found to be very useful in the design of practical inspection configurations. 2 In particular, the possibility of generating SH wave signals at grazi~g angles was found to be especially attractive in the inspection of butt welded plates and shells. 3 ,6,11,12 Interest in the development of nondestructive evaluation techniques based on the use of SH waves was also generated by the realization that SH waves may be more appropriate for the inspection of certain austenitic stainless steel welds. 4 ,s In this case, it was determined that SH waves can be used to overcome certain fundamental limitations imposed by the textured, columnar grain structures of multipass stainless steel weldments. 13 However, this subject involves a discussion of elastic wave propagation in aeolotropic media and other concepts that are mainly beyond the scope of this paper. The propagation and scattering of ultrasonic waves in butt weldments modeled as isotropic and homogeneous plates can be studied in two different waves. In plates that are relatively thin compared to the ultrasonic wavelength in the bulk material, it is often advantageous to represent the ultrasonic signals in terms of propagating and evanescent plate modes. 3 ,11,12 This approach can often result in a substantial simplification of the experimental system and less complicated data processing. 6 However, the plate mode description is not at all appropriate for thick plates when compact ultrasonic pulses are used to detect and characterize surface and interior defects. 2 In such cases, the propagation and scattering of the ultrasonic signals is more adequately described using conventional diffraction and ray concepts. 14 ,lS The propagation of plane SH wave signals in plates is considerably easier to analyze than SV waves and L waves, because SH waves are completely decoupled upon reflection from signals polarized in the plane normal to the plate (the sagittal plane).16,17 As a consequence of this simplification, SH plane waves reflect from the surface of the plate without a phase shift. 16 Using superposition, it can then also be shown that the same conclusion holds for all cases involving two-dimensional diffraction of SH wave signals by surface and interior defects. However, most practical nondestructive inspection configurations involve three-dimensional diffraction. In such cases, the reflection coefficient is purely real for certain symmetrical cases of incidence in the sagittal plane. Nominally SH wave rays, that do not lie in the sagittal plane, may then generate ELECTROMAGNETIC ACOUSTIC TRANSDUCER ARRAYS FOR NDE 285 other signals by mode conversion, resulting in an increased ultrasonic background level. In the past, a number of different weld inspection schemes using SH wave probing signals have been proposed. 2-4 In the majority of cases, such schemes relied upon the periodic-permanent-magnet (PPM) electromagnetic-acoustic transducer (EMAT) configuration, shown in Fig. 1. 18 Other transducer configurations have also been proposed. 19 ,20 To overcome the inherently low conversion efficiencies, such EMATs were, in most cases, designed as linear end-fire arrays, which must be driven with sufficiently long tone bursts to attain maximum signal-to-noise performance and steer the acoustic beam in the desired direction. Obviously, the resolution along the ray connecting the center of the transducer with the point of observation (range resolution) of such transducers is severely limited by the duration of the ultrasonic signal. The resolution in the plane transverse to the ray defining the range (transverse resolution) is limited by the physical length of the array, the direction of propagation, and the ultrasonic wavelength. In addition, the performance characteristics of PPM and other EMATs of the end-fire type suffer from inadequate sidelobe suppression and inability to independently adjust important parameters of the ultrasonic beam, e.g., beam width, focal distance, beam steering angle, etc. -I 1- 1L D- / 2 3 1 Permanent Magnet Assembly M N S N S N S N S N S N S --------------~ RF Coil -t - r:::=====r:::::::J=;:==~===f-------c RF W , '{I I I I I I -: - -: -:- -: : :- -:- ~ - :I I I - , I Fig. 1. I , I I - - I I -- I - Coil I - - - - .- - - I A periodic-permanent-magnet (PPM) electromagnetic-acoustic transducer (EMAT) of horizontally polarized shear (SH) waves. C. M. FORTUNKO AND R. E. SCHRAMM 286 Sm-Co Magnet Magnet Keeper M=2 Sm-Co Magnet Elongated Spiral Coil ............... Spacer .... I t Spiral Coil ......... _ Fig. 2. Two simple permanent-magnet (PM) EMATs. Recently, a new EMAT configuration has been developed that may be particularly useful in dynamic and synthetic aperture applications. Generically, the new EMAT design is derived from the PPM EMAT shown in Fig. 1, but it uses only one (M = 1) or two (M = 2) pairs of permanent magnets. The M = 1 and M = 2 EMAT configurations are shown in Fig. 2. The M = 1 EMAT configuration is conceptually similar to a configuration described earlier in the literature. 21 However, the earlier EMATs were designed primarily to produce highly collimated ultrasonic beams propagating along the normal to the plate surface. In contrast, the M = 1 EMAT of Fig . 1 was specifically designed to produce signals with a large beam width. Like the PPM EMAT of Fig. 1, the new EMAT configurations of Fig. 2 can be used to generate and detect ultrasonic signals that exhibit a component of polarization parallel to the surface of the plate (SH waves). Because the new EMATs can be made small along the intersection of the sagittal plane and the free surface of a plate as compared to the wavelength, they can be used to generate and receive compact ultrasonic signals from a wide angular range (at least in the sagittal plane). These properties are needed to realize long synthetic aperture lengths (for high beam compression ELECTROMAGNETIC ACOUSTIC TRANSDUCER ARRAYS FOR NDE 287 ratios) that yield good range resolution performance, especially when large wavelengths must be used. The new transducers will be referred to here as permanent magnet (PM) EMATs. In this paper the subject of beam forming using the more conventional PPM EMATs is considered, first, to illustrate the considerable shortcomings of linear end-fire transducers in the context of inspection of thick metal sections and welds. Second, some general properties of synthetic aperture systems are reviewed to highlight the most desirable features of such systems in thick section inspection applications. Third, the directivity patterns in the sagittal plane of the M = 1 and M = 2, PM EMATs are discussed. Attention is then focused on the use of the simple PM EMATs in synthetic and dynamic transducer arrays with better range and transverse resolution performance than the PPM EMATs using a large number of permanent magnets. The properties of such arrays are studied using analytical and numerical methods. Finally, two possible practical implementations of arrays using PM SH wave EMATs are discussed. ULTRASONIC DIRECTIVITY PATTERNS AND RANGE RESOLUTION OF PERIODIC-PERMANENT-MAGNET EMATs The factors governing the characteristics of ultrasonic beam patterns (e.g., directivity) by periodic-permanent-magnet (PPM) EMATs are considered in detail in References 8, 9 and 18, while the principles of the transduction mechanisms are discussed in References 22, 23 and 24. The directivity pattern in the sagittal plane of the EMAT depicted in Fig. 1 can be determined approximately from a relationship derived in Reference 18: DF (¢) = {!. Si~(~Sin¢)} {Sin[~ (1 2 'ITD (-A- sin¢) i)] e-j (M'IT/4)} sin ¢ 'IT ] sin 2(~sin¢ - n) [D 2'IT (1) where the angle ¢ is defined by the ray from the center of the PPM EMAT to an observation point and the normal to the surface of the plate. As shown in Fig. 1, the PPM EMAT is composed of a periodic array of M permanent magnets with a period L = 2D and half-width W, and a spiral wire coil that is always placed b~tween the magnet assembly and the surface of the metal plate. In practice, a spacer of thickness S (not shown) is inserted betwen the magnets to reduce eddy current losses. Then, L = 2(S + D). In Eq. (1) the parameter A is the bulk ultrasonic wavelength. An examination of the structure shown in Fig. 1 reveals that the PPM EMAT is essentially composed of M/2 cells (periods) that are connected in-parallel electrically and in-series ultrasonically. As a consequence, the PPM EMAT can be effectively modeled as a linear 288 C. M. FORTUNKO AND R. E. SCHRAMM end-fire antenna. This fact is reflected in Eq. (1), which can be interpreted as a product of element ("structure") and array ("form") factors. (For clarity the element and array factors are isolated in Eq. (1) using { } braces.) However, to realize narrow beam profiles and maximum transduction efficiencies, the EMAT of Fig. 1 must be driven electrically by a tone burst of electrical current that is at least M cycles long. Then, the principal direction (direction of maximum amplitude) of the ultrasonic beam can be determined from: ~ = arc sin [~D] (2) The end-fire behavior of the PPM EMAT has been verified experimentally using the calibration configuration shown in Fig. 3. (Figure 3 shows a 152 mm radius, 203 mm wide aluminum test block of semicylindrical cross-section.) In the configuration of Fig. 3, the PPM EMAT under test (transmitter EMAT) is placed symmetrically at the intersection of the two symmetry planes of the semi-cylindrical test block. On the curved surface of t.he block, a physically small (D 2.5 mm and W = 5 mm), broadband EMAT of the type shown in Fig. 2 (M = 1) is used as a probe. (It can easily be shown that the M = 1 EMAT of Fig. 2 exhibits maximum sensitivity to SH wave signals propagating along the normal to the surface. 21 The angles e and ~, defining the coordinates of the observation point, are also shown in Fig. 3. Using the ultrasonic transducer calibration configuration of Fig. 3, the directivity patterns for a number of SH wave EMATs were evaluated. In particular, the radiation patterns of M = 16, D = 3.2 mm, and S = 0.5 mm PPM EMATs were measured in the sagittal plane at 410 and 550 kHz. The two radiation patterns are shown in Figs. 4 and 5, respectively. Also shown in Figs. 4 and 5 are the directivity patterns calculated using Eq. (1). A comparison of the experimental and theoretical results shown in Figs. 4 and 5 shows close agreement over a considerable range of ~. In particular, in Fig. 4 the maximum amplitude is measured at the grazing angle, ~ = 0°. In Fig. 5, the maximum amplitude is measured at ~ ~ 50°, also in close agreement with the simple theory. However, the appearance of two major sidelobes in Fig. 4, and the slight broadening of the principal lobes in Fig. 5 are not predicted from Eq. (1). These discrepancies are attributed to the contamination of the radiation pattern by higher-order spatial harmonic contributions and end-effects. Clearly, the sidelobes observed in Fig. 4 are not desirable in many experimental situations. However, of even greater concern is the temporal duration of the ultrasonic signals produced by long PPM EMATs. It is seen from Figs. 4 and 5 that reasonably good transverse resolution performance can be obtained by using PPM EMATs with a sufficiently large number of periods M/2. Significantly better 289 ELECTROMAGNETIC ACOUSTIC TRANSDUCER ARRAYS FOR NDE Transmitter EMAT Receiver EMAT Fig. 3. An experimental configuration used in the mapping of directivity patterns. E~~T performance levels can be obtained using linear arrays with nonuniform element spacing. 25 , 2 6 However, to obtain good transverse resolution using the end-fire construction, range resolution performance must be sacrificed. (For example, 8 rf cycles were required to produce the radiation patterns depicted in Figs. 4 and 5.) Along the principal direction of the ultrasonic beam, the range resolution ~r can be determined from Ref. 27: ~r = v 12M s (3) where v is the bulk shear wave velocity and 6f is the rms system bandwidEh (pulse-echo operation is assumed). Clearly, PPM EMATs must sacrifice good range resolution for high directivity. However, considerable improvement in range resolution can be obtained using the shorter PM EMATs in conjunction with synthetic aperture and dynamic focusing techniques, which have recently been made very practical and cost-effective by significant advances in the field of digital signal processing. 4 C. M. FORTUNKO AND R. E. SCHRAMM 290 <P (degrees) o ,.UTWun.'· 152 mm SIGNAL (mV) • Experimental - 9 =O' Fig. 4. Theoretical M=16 = 0 3 .2 mm 410 kHz Measured and theoretical SH wave directivity patterns in aluminum. PPM EMAT with (D + S) = 3.7 rom and M = 16 at 410 kHz. ENHANCEMENT OF TRANSVERSE RESOLUTION USING SYNTHETIC APERTURE FOCUSING Synthetic aperture and dynamic focusing techniques have been considered by many researchers as a means of improving the performance of medical and nondestructive evaluation systems. 28 ,29 However, in the past the use of such techniques has generally been evaluated in the context of piezoelectric-acoustic transducers, whose directivity patterns are significantly different from those of the EMATs. The purpose of this section is to review certain fundamental constraints of synthetic aperture systems that are directly influenced by the radiation patterns of the isolated transducer elements used to build up an array. Consider a linear array composed of electrically and ultrasonically isolated transducer elements, as shown in Fig. 6. In the ELECTROMAGNETIC ACOUSTIC TRANSDUCER ARRA YS FOR NDE 4> 291 (degrees) SIGNAL (mV) AkmInum, , . 152 mm • Experimental - 8 =O' Theoretical M= 16 0 3 .2 mm = 550kHz Fig. 5 . Measured and theoretical SH wave directivity patterns in aluminum. PPM EMAT with (D + S) = 3.7 rnrn and M = 16 at 550 kHz. far-field of a single transducer, the effective beam-width is determined only by the bulk wavelength A, the transducer aperture length L , and the range from the transducer to the defect R. From the construction of Fig. 6 it then follows that the rnaxigum usable length L of a synthetic aperture array is given by : s AR L s o (4 ) L A longer array does not produce further improvements in the system performance because the defect is no longer in the field of view of the a dditional transducer elements. It can then be shown that the best achievable transverse resolution 6L is given by: 27 6L A e: - R L s 0 e:L (5) C. M. FORTUNKO AND R. E. SCHRAMM 292 Scan Direction _ \4-- Fig. 6. -- L. - -'Pj Geometry used to analyze the focusing performance of synthetic-aperture, ultrasonic inspection systems. where e: = 1 for a one-way system and 1/2 for a two-way ("pulse-echo") system using the full synthetic aperture length L. Efficient synthetic aperture focusing is feasible within th~ near-field of the array, whose maximum range (R ) is approximately given by:28 L 2 0 max (Ro) max = -f- (6) The effect of synthetic aperture focusing using the time-delay focusing method is to make the synthetic array curved, with R at the radius of curvature. 29 Beam-steering is also possible byOintroducing appropriate amounts of time-delay as a function of transducer element position in the array. Both effects are illustrated in Fig. 7. The formation of focused ultrasonic beams spect to the surface is important in practical the presence of the weld reinforcement (crown) of specific defects often restricts the use of at an angle with reapplications because and the orientation symmetric arrays. ELECTROMAGNETIC ACOUSTIC TRANSDUCER ARRAYS FOR NDE 293 "----Free Surface Defect below surface Focal Point 1 Fig . 7. Geometry used to define the parameters of synthetic apertures optimized for oblique focusing. From the construction in Fig. 7, it can be shown that the effective aperture L is reduced when the focal point does not lie on the axis bisec~ing the array. As a result, the end of the near-field region of the obliquely focused array is not as far as in the normal case. The end of the near-field zone for oblique focusing [R ( ~ )] o max is approximately given by: (7) Obviously, to increase [R (~)] it is necessary to obtain as long a usable synthetic apertu~e le~£h L as possible. This is particularly true for cases of large ¢. further complication is introduced by the directivity patterns of practical transducers, which usually exhibit a limited angular beam width. Fortunately, the special characteristics of the two SH wave EMAT configurations shown in Fig. 2 can produce plane wave components at all angles ¢ in the sagittal plane and, therefore, permit the realization of longer synthetic aperture lengths than piezoelectric transducers. The directivity patterns of the simple EMATs of Fig. 2 and their effect on the ultimate resolution of practical synthetic apertures are examined in the following sections. K ULTRASONIC DIRECTIVITY PATTERNS AND TRANSVERSE RESOLUTION OF SIMPLE PM EMATs . The dir:cti~ity patterns of the two simple PM EMAT configurat10ns shown 1n F1g. 2 have been measured using the transducer calibration system illustrated in Fig. 3. The measured directivity patterns were then compared with the theory [Eq. (1)]. The measured 294 C. M. FORTUNKO AND R. E. SCHRAMM and calculated responses of the two simple EMAT configurations of Fig . 2 are shown in Figs. 8 and 9. The radiation pattern of the "half-period" (M = 1) and "full-period" (M = 2) SH wave EMATs have been measured at 520 and 880 kHz, respectively . It is seen from Figs. 8 and 9 that the correspondence between the theory and experiment is very close. Consequently, it can be assumed that Eq . (1) can be used to predict the directivity patterns of the two transducer types at other frequencies. An examination of the directivity pattern shown in Fig. 8 reveals that the "half-period" (M : 1) EMAT can be used to collect ultrasonic data from a very wide range of angles. In fact, it can be concluded that the maximum usable field of view is 180 0 when A is larger than twice the aperture length L. Experimentally, the above observation can be related directly ~o the "optical" resolution ~A of a curved array, which is given by : cP (degrees) 60 SIGNAL (mV) Aluminum, t • 1 52 mm • Experimental - 9 Fig. 8. = O· Theoretical M= 1 0 = 3 .2 mm 520 kHz Measured and theoretical SH wave directivity patterns in aluminum . at 520 kHz. "Half- period" ( M = 1), PM EMAT ~ith D = 3 . 2 mm 295 ELECTROMAGNETIC ACOUSTIC TRANSDUCER ARRAYS FOR NDE <P (degrees) 150 SIGNAL (mV) Aluminum, , • 1 52 mm • Experimental - 9 Fig. 9. M '" =o· Theoretical M=2 = 0 3.2 mm 880 kHz Measured and theoretical SH wave directivity pattern in aluminum. "Full-period" (M = 2), PM EMAT with (D + S) = 3.7 mm at 880 kHz. 2sin~/i (8) where A is the ultrasonic wavelength and ~ is the viewing angle (angular aperture) of a synthetic aperture array, as shown in Fig. 7. (The viewing angle ~ can be directly related to the numerical aperture (f/number) of a curved array.) From the discussion in the preceding section, it follows that very long synthetic aperture lengths can be obtained using the halfperiod (M = 1) EMAT of Fig. 2. This fact has an important practical consequence. Consider a defect positioned just below the free surface, as shown in Fig. 7. Assuming that the sensitivity of the system is not limited by phase errors, electronic noise, and ultrasonic and electromagnetic interferences, and that the signal amplitudes C. M. FORTUNKO AND R. E. SCHRAMM 296 are a priori compensated for diffraction and propagation losses, the effective synthetic aperture length L can be made arbitrarily large. As a consequence, the viewing angle ~scan be made to approach 90° for ¢ ~ 90°. For ¢ ~ 0°, ~ ~ 180° is theoretically permissible. The directivity pattern of the "full-period" (M = 2) EMAT of Fig. 2 is significantly different from that of the M = 1 EMAT. In particular, the amplitude vanishes for ¢ = 0° for all operating frequencies and is very small for ¢ = ±90° at 880 kHz. The former effect is a consequence of symmetry and the latter of the A/(2D + S) ratio at 880 kHz. The directivity pattern of Fig. 9 can be desirable in certain applications. In particular, it can be used effectively to reduce resonant reverberation by the ray normal to the free surfaces of a plate. However, the M = 2, PM EMAT does not permit the realization of as long synthetic aperture lengths as the M = 1 PM EMAT. Although the two PH EHATs of Fig. 2 cannot produce radiation patterns that are altogether independent of ¢ and frequency, the variations are predictable from Eq. (1). Therefore, it should be possible to adjust the gains applied to each element of the array to produce the effect of a lens with widely variable focal length R and oblique focusing angle ¢. Furthermore, the availability of expressions for the directivity patterns and diffraction losses permits the modeling of the focusing properties of arrays on the basis of Huygens principle. This can be accomplished by replacing the usual ("optical") expression for a Huygens wavelet by the appropriate directivity function. CALCULATED RESPONSES OF SYNTHETIC ARRAYS USING PM EMATs Since a synthetic array is usually focused at a particular point within an elastic solid, it must be defocused at other points. Therefore, to understand the focusing performance of a particular array, it is necessary to describe the ultrasonic fields in the neighborhood of the focal point. Here, for computational and analytical convenience it is assumed that each element of the array generates a spherical wave ~., which can be represented in the sagittal 1. plane by: DF. (¢. ,w) 1. 1. r . -+-+ e j(kor .-wt) 01. A(w) (9) 01. where k is the wave-number (2rr/A), w is the angular frequency, r . is the distance from the particular array element to the observa~ion point, and A(w) is the spectrum of the ultrasonic signal. DF.(¢.,w) is the directivity function, which may also include the effecEs 6f lift-off and changes in inductive coupling as a function of material parameters. For simplicity, Eq. (1) may be used to approximate DF (¢,w). ELECTROMAGNETIC ACOUSTIC TRANSDUCER ARRAYS FOR NDE 297 At the focal point of the array the total response is given by: ()() (10) Then. for points in the neighborhood of the focal point x f • Eq. (11) takes the form: 00 1jJ (x o ,t) (r f 1.-r.) -jwt A( w) 01 e (11) where r f . is the distance from the i'th EMAT element position to the focal point f of the array, as shown in Fig. 10. Vs is the bulk shear wave velocity. x An approximate solution to Eq. (11) can be obtained by assuming a Gaussian spectrum A(w): 1 _~ A(w) = ~w 12n e (W-W o) \-z;- 2 (12) Position Of i'th Array Element 1+1 1-1 r - r I Focal Point Observation Point Fig. 10. Geometry used for calculating responses of synthetic PM EMAT arrays in the sagittal plane. C. M. FORTUNKO AND R. E. SCHRAMM 298 The integrals in Eq. (11) can then be readily evaluated using the saddle-point method of integration. 3D The saddle-point occurs at: (13) w 0) and When all the signals are correlated at the focal point xf(t W »(l1w)2 (rf.-r .)/v o 1 1jJ (x 0 ,t) ~L i 01 S DF. (¢. ,w ) 110 2nr oi ej w 0 v s (rf·-r .) e -!:2 1 01 ~:Y (r f .-r .) 2 1 01 (14) Equation (13) can be used to study the focusing properties of synthetic arrays at maximum correlation. It is interesting to note that in the limit as I1w approaches 0, Eq. (14) is the same as for a monochromatic array. Since in many practical applications it is not possible to center the synthetic array directly above a defect, e.g., in weld inspection, it is of interest to examine the focusing properties of arrays that are focused obliquely with respect to the free surface. Oblique focusing is also required because SH waves propagating at right angles with respect to the free surface and polarized along the long defect dimensions are insensitive to planar defects that are oriented normal to the plate surfaces. In practice such defects are most likely to precipitate failure because of their orientation relative to the applied and residual stress fields. For this reason, cases of normal (on-axis) and oblique (off-axis) focusing were studied. Using Eq. (13), the focusing performance of different synthetic arrays using the PM EMATs of Fig. 2 was investigated. In particular, Fig. 11 shows contour maps of three ultrasonic field distributions (log scale) produced by a 50-mm-long array of M = 1 PM EMATs with d = 3.2 mm and w /2n = 550 kHz focused at a depth of 25 mm below the free surface? In Fig. 11, three representative cases are shown: (1) the focal point located directly below the center of the array, (2) the focal point located directly below the first element of the array, and (3) the focal point 25 mm in front of the first element of the array. The array shown in Fig. 11 has 2.5 mm elementto-element spacing and is driven by a tone burst with I1w/w = 33%. The amplitudes of the correlated signals are indicated foroeach focal point, in relation to the case of focusing directly below the center of the array. The focusing performance of the same physical 299 ELECTROMAGNETIC ACOUSTIC TRANSDUCER ARRAYS FOR NDE ~ 0.48 0.80 \.00 ~ • C> _ @] C> E ... .§ ;!: "- ... Q C> .. 2 1-0.55 MHz C· ...... . . . . . . . . . . . . . . - 50 o db • c.nl .. -1 db -2 db -3 db - 6 db -1 0 db -20 db outer , ~ ' O 3J -20 - 30 - 40 - 10 0 SURfACE POSITION E MAT lotal i on' Fig. 11. CONTOUR INTERVALS 50 In m .per IUIO 2.5 mm ,pa cing M' I .,I ~ 10 30 20 & 40 50 Imml Focusing performance of a 50-rum-long array of M = 1, PM EMATs focused at three different points 25 rum below the free surface. The array operates at w /2TI = 50 kHz and has 33 % fractional band\vidth. 0 0 .38 0 .72 0 .94 • C> -E .§ .. ... CONTOUR INTERVALS o db • clnUr -I db - 2 db - 3 db -6 db - 10 db - 20 db ' OUI" 50 mm aperlur. 2.5 mm spacing . . .. ....... ..- ~----~------~--~~~--~------~-----c.~--~ so _ 40 - 20 - 10 10 20 ~ ' o . JJ - 30 r-- • • • • • • • • • • LEMAT lOCllions Fig. 12. SURfACE POSITION 30 40 50 Imml Focusing performance of a 50-mm-long array of M = 1, PM EMATs focused at three different points 25 rum below the free surface. The array operates at w /2TI = 850 kHz and has 33% fractional bandwidth. 0 300 C. M. FORTUNKO AND R. E. SCHRAMM array, but operated at 850 kHz, is shown in Fig. 12. From the plots shown in Figs. 11 and 12 it can be verified that the transverse resolution ~ increases with frequency in accordance with Eq. (8), and decreases with ~, because the effective array length is shorter. For comparative purposes, it is useful to examine the focusing performance of an array similar to that shown in Fig. II, but with a significantly longer element-to-element spacing. Figure 13 shows a contour map of the ultrasonic field produced by a sparse (undersampled) EMAT array at w /2TI = 550 kHz when 12.5-mm element-toelement spacing is used. o A comparison of Fig. 11 with Fig. 13 shows that the contours of the main lobe, centered in the vicinity of the focal point, are essentially the same. However, in Fig. 13 the appearance of two large (grating) sidelobes can be observed. In this case, the broad spatial extent and high amplitude of the sidelobes defeats the main purpose of synthetic aperture focusing. Figure 14 shows the focusing performance of a 50-mm-long, 550 kHz array with ~w/w = 33% and 2.5-mm element-to-element spacing that uses the M = 2 ~M EMAT elements. The performance of this array should be compared directly to that shown in Fig. 11. It is seen that the performance of this array is not quite as good as that using the M = 1 PM EMATs. However, this array may be useful in practice, because the M = 2 PM EMATs produce a significantly smaller ultrasonic background level that reverberates between the free surfaces of a plate. In certain applications this may be a distinct advantage. To study the effect of array length on the focusing performance, the field distributions of a 50-mm-long and 250-mm-long M = 1 PM EMAT array , operating at 500 kHz, were compared. The results are shown in Fig. 15 for the case of focusing at a point 25 mm below the free surface and directly below the first element of the array. In Fig. IS, the amplitudes of the signals correlated at the focal point are not pre-emphasized to compensate for the effects of diffraction. It is seen that the total amplitude of the ultrasonic signal evaluated at the focal point is smaller (0.80) for the shorter array than for the longer array (1.51). However, the additional gain is not commensurate with the greatly increased array length. Also, the range and transverse resolution in the two cases are not significantly different An evaluation of the results shown in Figs. 11-15 reveals that the transverse resolution ~ agrees qualitatively with the simple theory of Eq. (8). The effect of the large angle ~ on the transverse resolution ~A is especially pronounced in Fig. 15. The quantitative differences between the simple theory of Eq. (8) and the actual transverse resolution determined directly from Figs. 11-15 (using the l/e criterion) are attributed to the effective apodization 301 ELECTROMAGNETIC ACOUSTIC TRANSDUCER ARRAYS FOR NDE ..... 0.2) -.. ~~ E ~ :z: ....... lL ~ CI CONTOUR INTERVALS ~ iJ ..- so 4M:T o M'I 50 mm ape' lure 1' 0.55 MHz ~ ' 0 . 33 - 10 12.5 mm spacing • - 30 - 20 locll i ons Fig. 13. db - unlar - I db - 2 db -) db - 6 db - 10 db - 20 db - Oull, o -10 • 10 20 so 10 30 SURFACE POSITION Imml Focusing performance of a 50-mm-long, undersampled array of M = 1, PM EMATs focused at two different points 25 mm below the free surface. The array operates at w /2n 550 kHz and has 33 % fractional band~Tidth. 0 ..... o.so 0.14 0.50 -. CONTOUR INTERVALS o db - •• nu, -I db - 2 db - ) db 50 m m Iporlun 2.5 mm spicing c· . · . · · . . .. . . . . . 1' 0.55 MHz -so - 40 [MAT lO"l i o.s Fig. 14. ~ ' O 33 -30 -20 -1 0 . .. - 6 db -10 db - 20 db - oute, 10 20 30 10 SURFACE POSITION Imml Focusing performance of a 50-mm-long array of M = 2, PM EMATs focused at three different points 25 mm below the free surface. The array operates at w /2n = 550 kHz and has 33% fractional bandwidth. 0 so C. M .FORTUNKO AND R. E. SCHRAMM 302 5\ 0 .... . . 0 0 E oS ... E oS 0 0 ...co ::: l- a.. a.. ~ Mq ..• • . I • 0. 55 MHz - --. ~ 50 mm .plrturl 2.5 mm spicing "0.55 MHz - ..... . • • • -10 10 4MAT LDClt i Dns SURFACE POSITION 250 mm .p"rture 2.5 mm spacing M' I ~ ' O . JJ - 10 Fig. 15. ... CONTOUR INTERVALS o db • cent .. -I db -2 db - J db -6 db -10 db -20 db - Duter :z: :z: ...co ::: I- 0 1.51 0.80 .. ~ ' 0 . 33 0 SURFACE POSITION (mml 10 (mml Comparison of focusing performances of two arrays, 50-mmlong and 250-mm long, of M = 1, PM EMATs focused at a point 25-mm below the free surface and directly below the first element. Both arrays operate at 500 kHz with 33% fractional bandwidth. The correlated signals are not preemphasized to compensate for diffraction effects . (taper) of the arrays by diffraction effects and the dependence of the EMAT directivity function [Eq. (1)] on the angle~. Under certain circumstances it may be possible to compensate for such effects by varying the gain of each element as a function of element position in the array. (Analytically, this can be done by evaluating Eq. (13) with the factors DF.(¢., w )/r . set to unity.) J. J. 0 oJ. For comparative purposes the focusing performance of the array shown in Fig. 11 was evaluated in the limit as ~w approaches O. The resulting contour plot is shown in Fig. 16. An examination of Fig. 16 shows that the contour patterns of Figs. 11 and 16 are considerably different. The major differences between Figs. 11 and 16 are caused by the time-limited nature of the signals used to dr'ive the array in Fig. 11. As a consequence, it can be expected that the synthetic aperture approach can offer significant improvements in practical systems. In particular, this approach promises a significant reduction of the importance of multiple echoes caused by reverberation between the free surfaces of a plate. This follows directly from the fact that the dimensions of the resolution cell, ~ r and ~A estimated from Figs. 11-15, are smaller than the thicknesses of practical plate weldments. 303 ELECTROMAGNETIC ACOUSTIC TRANSDUCER ARRAYS FOR NDE . '" 1.00 .. a _.. E ... .§ .........z: a::: 1If/&/ /" CONTOUR INTERVALS o db • Clnlor - I db -2 db -3 db - 6 db SO mm apellu," .. -50 2.5 m m .pie in g - 40 - ]0 r- • • • • • • • • • • LE MAT Lata! i on. Fig. 16. . . . . . ... .. . -20 - 10 SURFACE POSITION 10 - 10 db -20 db · oulor , 20 ]0 40 50 (mm) Focusing performance of a 50-mm-long, 550-kHz, monochroma·tic array of M = 1, PM EMATs focused at two different points 25 rom below the free surface . DISCUSSION An array can be viewed as an aperture excited in localized areas by isolated elements that, in general, exhibit nonisotropic radiation characteristics and may be driven by generators of unequal strength. Upon reception, the gain of the receiver preamplifiers may also vary, depending on the time and spatial coordinates of the elements. General methods needed to calculate the responses of practical arrays are found in the literature. 27 ,31,3 2, 33 In ultrasonic nondestructive evaluation, the ability of an array inspection system to detect and characterize a potential defect is limited principally by the resolution. High resolution can be attained either by forming an intersecting family of focused, monochromatic beams that can be scanned across the defect by applying appropriate phase shifts to individual elements, or by using timelimited ultrasonic signals that can be made to correlate at a partiular point within an elastic solid by time-delay techniques. Recent technological advances in the fields of digital processing of ultrasonic signals and electromagnetic-acoustic transducer design have made the latter approach particularly attractive . 4 For this reason the synthetic aperture alternative to improved resolution has been explored in this paper. 304 C. M. FORTUNKO AND R. E. SCHRAMM In practical applications, it is often necessary to aim a linear array at an oblique angle and, then, to scan the ultrasonic beam over a range of angles in the sagittal plane. For example, in weld inspection it is often not possible to place an array directly above the region to be inspected because of the presence of certain geometrical obstacles (e.g., the weld crown). Since the angular scanning range is restricted primarily by the element beamwidth, it is advantageous to build up the array using elements with broad angular radiation patterns. In this paper, this problem has been addressed by building up the arrays using SH wave electromagnetic-acoustic transducers with a small aperture length in relation to the ultrasonic wavelength. The use of SH waves is advantageous because the radiation patterns of elementary SH wave line-sources are omnidirectional in the sagittal plane. whereas those of shear waves polarized in the sagittal plane (SV waves) and longitudinal waves are very complicated. 7 '8 The use of SH waves has also other important practical advantages in defect detection and characterization. 3-6 Because of its inherent complexity, the subject of ultrasonic scattering from defects was omitted in this paper. Instead, it was implicitly assumed that the scatterers were ideal line or point targets. However, this idealization must not be considered a major limitation, because of the recent emphasis on long-to-medium wavelength inversion techniques in defect characterization, particularly in weld inspection.3'~'6'11 In this paper, the focusing and beam steering properties of several arrays of practical dimensions have been studied using a combined analytical-numerical approach. The performance of the arrays was simulated numerically on the basis of the empirically determined response of a new type of array element with a broad beam width. The results of the study strongly suggest that it may be feasible to fabricate practical arrays using SH wave EMATs that, although operating at relatively low frequencies (large wavelengths), would outperform in resolution and scanning range the conventional, unfocused ultrasonic inspection systems. (This implication is particularly important in weld inspection and nondestructive evaluation of thick section structures.) In practice, such arrays could either be implemented by mechanically scanning simple transmitter and receiver EMATs over the surface of the part to be evaluated, or by using physical arrays composed of many individual EMAT elements. The former approach involves essentially in-series processing of the ultrasonic data, while the latter is essentially an in-parallel signal processing approach. The main practical limitation of the in-series signal processing scheme may be caused by phase errors associated with the mechanical scanning of the single transmitter and receiver EMATs. This approach may also be limited by the total ultrasonic power that can be concentrated in the neighborhood of a defect. Although the ELECTROMAGNETIC ACOUSTIC TRANSDUCER ARRAYS FOR NDE 305 in-parallel approach can produce higher ultrasonic power densities (albeit at higher cost), it suffers from the disadvantage of rather sparse element-to-element spacing, imposed by the physical configuration of the EMATs. (This limitation is easily overcome in the inseries approach.) However, in the in-parallel approach the relative element positions are fixed. This raises the possibility of utilizing the phase information (interferometry), which can lead to even better resolution than that defined by "optical" relationships. A definitive choice between the two approaches can only be made after having taken into account the details of the expected scattering mechanisms, specimen geometry, and the cost factors related to a specific hardware implementation. ACKNOWLEDGEMENT The authors would like to acknowledge S.A. Padget who fabricated prototype SH wave EMATs and collected much of the experimental data used in this paper. This work was supported by the Center for Advanced Nondestructive Evaluation, operated by the Ames Laboratory, US DOE , for the Naval Sea Systems Command and the Defense Advanced Research Projects Agency under Contract No. W-7405-ENG-82 with Iowa State University and by the National Bureau of Standards. REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. C.M. Fortunko, C.F. Vasile and R.B. Thompson, "Electromagnetic Transducers for Weld Inspection," presented at 37th National Fall Conference of the American Society for Nondestructive Testing, Detroit, MI, October 3-6, 1977. C.M. Fortunko, "Ultrasonic Inspection of Weldments Using Frequency Scanned SH Waves," in Proc. 1979 Ultrasonic Symp., J. deKlerk and B.R. McAvoy, eds., IEEE, New York, 1979. C.M. Fortunko, R.B. King and M. Tan, "Nondestructive Evaluation of Planar Defects in Plates Using Low Frequency SH Waves," J. Appl. Phys., 53, 3450, 1982. R.K. Elsley and C.M. Fortunko, "Improvements in Flaw Detection in Austenitic Stainless Steel Weldments," Proc. 1981 Ultrasonics Symp., B.R. McAvoy, ed., IEEE, New York, 1981. D.S. Kupperman and K.J. Reimann, "Ultrasonic Wave Propagation and Anisotropy in Austenitic Stainless Steel Weld Metal, \I IEEE Trans. Sonics & Ultrasonics, 27:7, 1980. C.M. Fortunko and R.E. Schramm, "Nondestructive Evaluation of Butt Welds Using Electromagnetic-Acoustic Transducers," Weld. J., 61:39, 1982. G.F. Miller and H. Pursey, "The Field and Radiation Impedance of Mechanical Radiators on the Free Surface of a Semi-Infinite Isotropic Solid," Proc. Roy. Soc. London, Ser. A, 223:521, 1954. W.J. Pardee and R.B. Thompson, "Half-Space Radiation by EMATs," J. Nondestructive Eval., 1:157, 1980. 306 9. 10. 11. 12. l3. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. C. M. FORTUNKO AND R. E. SCHRAMM A. V. Kharitonov, "Progress and Problems in the Theory of Normal Waves in Ultrasonic Flaw Detection," Defectoskopiya, 7:59, 1979. R.B. Thompson, "The Relationship between Radiating Body Forces and Equivalent Surface Stresses: Analysis and Application of EMAT Design," J. Nondestructive Eval., 1:79, 1980. S .K. Datta, C.M. Fortunko and R.B. King, "Sizing of Surface Cracks in a Plate Using SH Waves," Proc. 1981 UltJ;asonics Symp., B.R. McAvoy, ed., IEEE, New York, 1981. N.N. Egorov and A.V. Kharitonov, "Diffraction of Shear Normal Modes by a Surface Crack in an Elastic Plate," Sov. Phys. Acousticis, 25:34. 1979. B.L. Baikie, A.R. Wagg, M.T. Whittle and D. Yapp,"Ultrasonic Inspection of Austenitic Welds," J. Br. Nucl. Energy Soc., 15: 251, 1976. L.M. Brekhovskikh, "Waves in Layered Media, Sec. Ed.," Academic Press, New York, 1980. S.K. Datta, A.H. Shah and C.M. Fortunko, "Diffraction of Medium and Long Wavelength SH Waves by Edge Cracks," J. Appl. Phys., 53:2895, 1982. B.A. Auld, "Acoustic Fields and Waves in Solids, Vol. II," WileyInterscience, New York. 1973. L. Niklas, "Group Propagation Time and Beam Displacement in Inclined Reflection: Their Effect on Practical Ultrasonic Testing of Materials" (in German). MaterialprUfung, 7:281, 1965. C.F. Vasile and R.B. Thompson, "Excitation of Horizontally Polarized Elastic Waves by Electromagnetic Transducers with Periodic Magnets," J. Appl. Phys., 50:2583, 1979. J.F. Martin and R.B. Thompson, "The Twin Magnet EMAT Configuration for Exciting Horizontally Polarized Shear Wave," Proc. 1981 Ultrasonics Symp., B.R. McAvoy, ed., IEEE, New York, 1981. R.B. Thompson, "New Configurations for the Electromagnet Generation of SH Waves in Ferromagnetic Materials," Proc. 1978 Ultrasonics Symp., J. deKlerk and B.R. McAvoy, eds., IEEE, New York, 1978. B.W. Maxfield, M. Linzer, W.B. McConnanghey and J.K. Hulbert, "Design of Permanent Magnet Electromagnetic Acoustic Wave Transducers (EMATs)," Proc. 1976 Ultrasonics Symp., J. deKlerk and B.R. McAvoy, eds., IEEE, New York, 1976. R.B. Thompson, "A Model for the Electromagnetic Generation of Rayleigh and Lamb Waves," IEEE Trans. Sonics & Ultrasonics, 20:340, 1973. R.B. Thompson, "Mechanisms of Electromagnetic Generation and Detection of Ultrasonic Lamb Waves in Iron-Nickel Alloy Polycrystals," J. Appl. Phys., 48:4942, 1977. R.B. Thompson, "A Model for the Electromagnetic Generation of Guided Waves in Ferromagnetic Metal Polycrystal," IEEE Trans. Sonics & Ultrasonics, 25:7, 1978. W.L. Weeks, "Antenna Engineering," McGraw-Hill, New York, 1968. R.H. Duhamel, "Optimum Patterns for End-fire Arrays," Proc. IRE, 41:652, 1953. ELECTROMAGNETIC ACOUSTIC TRANSDUCER ARRAYS FOR NDE 27. 28. 29. 30. 31. 32. 33. 307 A. Rihaczek, "Principles of High-Resolution Radar," McGraw-Hill, New York, 1969. E.B. Miller and F.L. Thurstone, "Linear Ultrasonic Array Design for Echosonography," J. Acoust. Soc. Am., 61:1481, 1977. G.S. Kino, D. Corl, S. Bennett and K. Peterson, "Real-time Synthetic Aperture Imaging Systems," Proc. 1980 Ultrasonics Symp., B.R. McAvoy, ed., IEEE, New York, 1980. A. Papoulis, "The Fourier Integral and Its Applications," McGrawHill, New York, 1962. B.D. Steinberg, "Principles of Aperture and Array System Design," Wiley-Interscience, New York, 1976. U. Aulenbacher and K.J. Langenberg, "Analytical Representation of Transient Ultrasonic Phased-Array Near- and Far-Fields," J. Nondestructive Eval., 1:53, 1980. H.F. Harmuth, "Acoustic Imaging with Electronic Circuits," Academic Press, New York, 1979.