Multi-Layered PZT/Polymer Composites to Rat io and...

IEEE TRANSACTIONS ON ULTRASONICS, FERROELECTRICS, AND FREQUENCY CONTROL, VOL. 49, NO. 7, JULY 2002 1005

Multi-Layered PZT/Polymer Composites to Increase Signal-to-Noise Rat io and Resolution

for Medical Ultrasound Transducers Part 11: Thick Film Technology

David M. Mills, Member, IEEE, and Stephen W. Smith, Member, IEEE

A bstract-Increasing transducer bandwidth and signal- to-noise ratio (SNR) is fundamental to improving the qual- ity of medical ultrasound images. In previous work, we have proposed the use of multi-layer 1-3 PZT/polymer composites to increase both, but have encountered significant fabrication challenges [19]. These difficulties include making the bond thickness between the layers small relative to the ultrasound wavelength and aligning the posts of the composite to increase coupling coefficient.

Thus, we have developed a multi-layer composite hybrid array that will not require post alignment. Starting from a 2-MHz, three-layer PZT-5H, thick film transducer designed for 1.5-D arrays, cuts are made only through the top layer and back-filled with epoxy, forming a composite layer on top of two ceramic layers. Finite element (PZFlex) simulations show that for a 2-MHz phased-array element with a single matching layer, the three-layer hybrid structure in- creases the pulse echo SNR by 11 dB versus a single layer PZT element and improves -6 dB pulse echo fractional bandwidth by a factor of 1.4. Composite hybrid arrays fabricated in our laboratory showed an improvement in SNR of 6 to 11 dB over a PZT control and an increase in -6 dB bandwidth by a factor of 1.1. Images from a phased-array scanner confirmed these improvements.

I. INTRODUCTION

MPROVEMENTS in SNR and the bandwidth of novel I transducer designs relative to conventional PZT transducer arrays are key areas of research for medical ultrasound imaging. Such arrays allow higher frequency imaging and tissue harmonic imaging [1]-[3] to be used for increased spatial resolution, deeper tissue penetration for frequencies currently in use, and lower excitation voltages to be used in transmit mode.

Several investigators have described the benefits of multi-layer PZT for array transducers fabricated using thick film, “green tape” technology suitable for mass pro- duction [4]-[6]. The major advantage of multi-layer PZT is improved SNR because of the better electrical match between the array element and the t,ransmitter in transmit mode and because of the increased element capacitance that drives the coaxial cable more efficiently in re-

Manuscript received May 15, 2000; accepted January 14, 2002. D. M. Mills is with the GE Global Research Center, Niskayuna,

S. W. Smith is with the Department of Biomedical Engineering, NY 12309 (e-mail: [email protected]).

Duke University, Durham, NC 27708-0281.

ceive mode. Goldberg and Smith developed a three-layer 1.5-D transducer array of 3 x 43 elements (129 total) in which they obtained 7.3 dB of SNR increase over PZT- 5H, but did not show improved bandwidth [5]. Emery and Smith fabricated a 47-element linear array with seven-layer transmit elements and single-layer receive elements. With this configuration, they obtained a 35-dB increase in SNR compared with PZT-5H in addition to significant bandwidth improvement. However, this system required new low impedance transmitters (10 0) and high impedance pre-amplifiers connected directly to the elements within the transducer assembly [7].

PZT/polymer composite transducer designs have also been investigated extensively to improve acoustic impedance matching between tissue and PZT, thus increasing transducer bandwidth [8]-[12].

Multi-layer composite transducer structures ameliorate both problems of electrical and acoustic impedance matching. Several investigators have described such stacked composite arrays [8], [13]-[16]. In the best performance, Zip- paro et al. [17] presented stacked composite plates in which coupling of ICT = 0.65 was obtained experimentally. Seyed- Bolorforosh et al. [ 181 proposed stacking piezoelectric layers with gradually decreasing acoustic impedance in the di- rection of the tissue load to obtain tunable multi-frequency arrays of greater bandwidth. However, the major problem with all of these techniques is achieving alignment of >90% of the post pitch (e.g., tolerance of 10 to 20 pm for a 1-3 composite at 2 MHz with a post pitch of 127 pm) to obtain significantly increased SNR and bandwidth relative to conventional transducer arrays [ 171.

In recent work, we presented a hybrid multi-layer transducer in which a 1-3 PZT/polymer composite top layer was manually bonded to a PZT bottom layer with a thin film epoxy bond ( 3 pm). Thus, the need for post alignment was eliminated, enabling simplified construction because the composite layer was on top of a monolithic PZT layer. As with stacked composites, the hybrid transducer yielded both improved SNR (5.2 dB) and increased -6 dB fractional bandwidth (from 23 to 30%) for the hybrid relative to a PZT-5H control [19]. (This prototype array did not include a matching layer [19] .)

In this paper, we describe analysis using finite element method (FEM) simulations and experimental results of multi-layer PZT/composite hybrid transducer arrays. To

0885-3010/$10.00 @ 2002 IEEE

1006 IEEE TRANSACTIONS ON ULTRASONICS, PERKOELEC‘l’RICS, AND FREQUENCY CONTROL, VOL. 49, NO. 7, JUIjY 2002

Elevation

Fig. 1. Three element three-layer PZT composite hybrid array dia- gram. Note azimuth arid elevation directions and the composite and matching layers. (Not to scalc.)

overcome the difficulties of thin film bonding of composite and PZT layers, this paper discusses a design in which the top layer of a thick film multi-layer transducer array is partially diced, forming a 2-2 composite layer on top of the multi-layer transduccr, as shown in Fig. 1. Our hypothesis is that by combining thc advantages of multi-laycr PZT and PZT/polymer composites, we will incrcase the SNR. and thc bandwidth relative to conventional PZT imaging arrays. The goal of thc work presented here was to de- velop a transducer material that coidd be used in existing systems as a “drop-in” replacement for current transducer materials. The paper includes design, finitc element analysis, and experimental results for a 2-MHz phased array with ”/z spacing. The “badness” criterion, which combines SNR and pulse length, previously discussed by Selfridge [20], Lockwood and Foster [al l , and Mills and Smith [I91 is used to evaluate this design. “Badness” is defined as the energy in the pulse ringdown following the center of mass of the pulse ( t o ) divided by the peak amplitude of the pulse squared.

11. METHODS

A. Modeling Methods

The details of the 2-MHz three-layer hybrid transducer array design are shown in Fig. 1. For example, LeleUatron M

3 mm (which would be typical for one row of a 4 x N 1.5-D array defined as 4 rows of N columns of transducer elements), kerf = 50 pm, and element pitch = 313 pm. We performed FEM using PZFlex (Weidlinger ASSOC., New York, NY and Los Altos, CA) to design our hybrid array, including a single matching layer compared with a corivcntional array using PZT-5H. The software packagc has been validated in previous transducer developmcnt projects [19], [22], [23]. Impedance plots and pulse echo simulations were calculated to compare the PZT control array with the hybrid array. A loaded epoxy backing, a single matching layer, and two passive (open circuit) neighboring elements were included in the model. It should be noted that we used thc same matching layer for both the

Water

Silver Foil Matching

Layer

olymer

PZT

Backing

Fig. 2. FEM mesh of five elements of the three-layer PZT composite hybrid array. The ccnter element is active.

TABLE I PZT MATERIAL PROPERTIES USED FOR MODELING

PZT-513 [24]

7820 1.37 x 10l1

2.23 x lo1’

9.23 x lo1” 2.48 x lo1”

16.1 -9.44 22.5 1310 1200 90

1.26 x 1011

8.79 x 1010

control and hybrid arrays, Le., the most suitable matching layer material that we were able to fabricate in our laboratory.

In our analysis and measurements throughout this paper, we assume the pre-amplifiers in receive mode are the dominant source of noise and that this pre-amplifier noise is independent of the transducer design. In this case, improvements in transducer signal become identical to improvements in SNR. We also assume that the cable can bc modeled as a capacitor because of the short length (cable lcngth <<”/J.

Fig. 2 shows an azimuth view ofthe finite elemcnt mesh of the transduccr design with two neighboring clements on

MILLS AND SMITH: PZT/POLYMER COMPOSITES PART 11: THICK FILM TECHNOLOGY 1007

TABLE I1 POLYMER MATERIAL PROPERTIES USED FOR MODELING

(ATTENUATION Is AT 3.5 MHz.).

Matching Composite Backing Material layer filler epoxy

P (g/cm3) 4.50 1.08 1.03 v1 ( 4 s ) 1500 2640 2680 vs ( 4 s ) 850 1150 1180 a1 (dB/mm) 5.6 3.6 3.5 as (dB/mm) 42.0 8.3 3.5 (€33)'/€0 1 1 1

TABLE I11 FEM MESH DETAILS FOR THE PZT-5H CONTROL TRANSDUCER. (THE WAVELENGTH Is BASED ON BULK ACOUSTIC VELOCITIES.)

Elements per Element wavelength at

Material size (pm) 2 MHz ~~

Water Silver foil Matching layer PZT Backing

31 x 33 x 31 13 x 33 x 31 28 x 33 x 31 61 x 33 x 31 43 x 33 x 31

25 x 23 x 25 140 x 55 x 58 27 x 23 x 24 33 x 62 x 65 31 x 41 x 43

TABLE IV FEM MESH DETAILS FOR THE HYBRID TRANSDUCER. (THE WAVELENGTH Is BASED ON BULK ACOUSTIC VELOCITIES.)

Material

Water Silver foil Matching layer Composite Polymer Composite PZT PZT Backing

Element size (pm)

25 x 26 x 31 13 x 26 x 31 18 x 26 x 31 20 x 26 x 31 55 x 26 x 31 61 x 33 x 31 33 x 26 x 31

Elements per wavelength at

2 MHz

30 x 29 x 25 140 x 69 x 58 41 x 29 x 24 66 x 50 x 42

100 x 77 x 65 37 x 77 x 65 40 x 51 x 43

TABLE V FEM ARRAY DIMENSIONS AND PROPERTIES.

PZT-5H Hybrid

Pitch (pm) 313 313 Wkerf b m ) 50 50 t (pm) 610 660 L ("1 3 3 t M L (Pm) 110 110

Fig. 3. Micrograph of the three-layer 2-MHz hybrid transducer. (Scale is 0.12 mm.)

each side. Planar symmetry was used to allow modeling of only one-half the width of the transducer in azimuth, and periodic symmetry was used to model only one period of the 2-2 composite in elevation (one-half the length of the PZT plate and the epoxy filler in elevation). Tables I and I1 contain the material properties used for this paper, Tables I11 and IV contain the mesh details. Table V con- tains the details of the array models. Ideally, 8 to 20 mesh elements per wavelength at the highest frequency of inter- est should be maintained. The time steps for the models were 11.3 nS in water and 2.8 nS elsewhere. By varying the mesh sizes and time steps in relation to the longitu- dinal velocity of sound in each material, accuracy can be maintained, if chosen properly, while improving computa- tional efficiency [22]. Pulse echo results were obtained by running a transmit problem, far-field extrapolation, and then a receive problem. These simulations were run under Windows N T on a 450-MHz Intel PentiumB I1 processor and took 1 to 3 min. Full 3-D simulations (eliminating the periodic symmetry) were also performed, but required at least 256 MB of RAM and 2 to 6 h of processor time.

B. Fabrication Methods

We fabricated transducer arrays corresponding to the design of Fig. 1, starting with a 2-NIHz three-layer PZT- 5H chip as described and used by Goldberg and Smith [5] from ceramic manufacturer "A." This chip consisted of four azimuthal rows of ceramic multi-layer material with interlaminar connections made using plated vias and edge terminations of glass frit. A single row of multi-layer material was carefully diced from the chip while preserving part of the plated vias and edge terminations needed to con- nect the layers. Then, a composite was formed by partially dicing through only the top layer, using a conservative dicing depth to avoid dicing through the first inner electrode. Then, a conductive matching layer (Ablefilm 50253, Ablestik, Rancho Domingo, CA) was placed over the top

1008 IEEE TRANSACTIONS ON ULTRASONICS, VERROELECTRICS, AND FREQUENCY CONTROL, VOL. 49, NO, 7, .JULY 2002

5.0

- 4.0 G E. (u 3.0 m S

s 2.0 .c, .- 3 = 1.0

0.0 0 2 4 6 8 I O

Frequency (MHz) (4 10 I I

-1 0 A a

-30 v

al

c 3 -50 n

-70

-90 0 2 4 6 8 10

Frequency (MHz) @)

Fig. 4. FEM complex impedance results: a) magnitude and b) phase for the PZT-5H control and hybrid array elements in water.

electrode to protect it during processing, and the composite was filled with epoxy (EPO-TEK 301-2; Epoxy Tech- nology, Inc., Billerica, MA) to obtain the PZT/composite (2-2) hybrid structure as shown in Fig. 3. The overall thickness of this structure was 660 pm with cuts approximately 200 pm deep at a pitch of 120 pm.

The chips were bonded to a loaded epoxy backing (lightly loaded with A1203 and phenolic micro-balloons) where electrical (hot) conricctions were made using a conductive epoxy (EPO-TEK) and silver wires embedded in the backing as shown in Fig. 1. The 2-MHz array was diccd into 1.5-D phased array elemcnts (Leleualron M 3 mm, kerf = 50 pni, and pitch = 313 pm). After dicing, a 12- pm silver foil was bonded to the top of the array with screen-printed conductive epoxy to form the top ground connection (not shown in Fig. 1). Elevation lenses were omitted in this study.

C. Experimental Methods

After each array was fabricated, vector impedance measurements were made with an HP 4194A impedance ana-

50 1 PZTdH Control 11

-50

0

s E -20

-60

I I

Time (pS)

(4

I * \ n l -PzT/Com~ I

0 2 4 6 8 10

Frequency (MHz) @)

Fig. 5. FEM pulse echo results: a) pulse and b) spectra for the PZT- 5H control and hybrid array elements in water.

lyher (Hewlett-Packard Co., Tokyo, Japan) connected to a single array element while the face of the array was sub- merged in water. Pitch-catch experiments to determine impulse response were performed in a water tank using a reflecting aluminum block placcd 5 cm from the transducer and an excitation pulse of 100 Vpk-plc (open circuit) [one cycle at 6 MHz from an HP-8165A signal generator (Hewlett-Packard Co., Boblingen, Germany) and a 5 0 4 EN1 amplificr (ENI, Inc., Rochester, NY)]. (As used here, pitch-catch experiments use one element for transmitting and a separate element for receiving, typically less than three elemcnts apart, while pulse echo refers to transmitting and receiving on the same element.) This pulse was identical to that of thc FEM simulations and approximated broadband cxcitation for a 2-MHz transducer with no dis- continuities. Coaxial cable (50 O), 2.2 m in length with 210 pF of capacitance per channel, was connected to each elcment. In receivc mode, the echo was recorded using a Tektronix TDS 744A (Tektronix, Inc., Wilsonville, OR) os- cilloscope with a l o x probe having 11.2 pF of capacitance and an input impedance of 10 MR.

MILLS AND SMITH: PZT/POI,YMER COMPOSITKS PART IT: THICK FILM TECIINOLOCY 1009

5.0

4.0 G Y Q) 3.0 'EI a c 2.0

v

.c1 .- CT) m = 1.0

0.0 I

30

20 n > E 10 Y

-20

0 2 4 6 8 10 -30 I Frequency (MHz) Time (pS)

-10

-30 cir Q) v m Q c p.

Y

-50

-70

-90 0 2 4 6 a 10

Frequency (MHz) @)

Q) U S U .- -

-20 a

~ i ~ . 6, ~ ~ ~ ~ ~ i ~ ~ ~ ~ ~ l colrlplcx impedance results: a) magnit,ldc arid b) phase for the PZT-5H control and hybrid array elements in water.

Fig. 7. Experimental pitch catch resiilts: a) piilsc and b) spectra for PZT-SH "yhrid array e'ernents in water.

B. Experimentu.1 Resudts 111. RESULTS

A. Modeling Results

As shown in Fig. 4, the series resonant frequency, from the PZFlcx simulated impedance plots, for both arrays, was at 2 NIHz with a magnitude of 1750 R for the PZT control element and 256 R for the hybrid clement. The resonance of the two undiced layers in the hybrid element appears near 5 MHz, the width modes appear iiear 6 MHz, and, for both array elemcnts, the third harmonic of the thickness resonance is between 7 arid 10 MHz.

Fig. 5 comparcs the simulated pulse-echo results. The PZT control element had an amplitude of 25 mV, a -6 dB bandwidth of 46%, and a normalized badness of 3.5 ( t o = 1.36 pS). The hybrid element echo amplitude was 88 rnV, with a -6 dB bandwidth of 65%, and a normalized badness of 1 ( t o = 1.35 $3). Thus, the hybrid array had an increased echo amplitude of 10.9 dB, increased bandwidth by a factor of 1.42, and decreased badness by a factor of 3.5 compared with the PZT-5H transducer.

As describcd previously, measurements in water were made to compare the hybrid and PZT-5H control arrays. The series resonant frequencies, from the experimental inipcdaiice plots in Fig. 6, for both the control arid the hybrid array elements, were 2 NIHz with a magnitude of 1980 R for the PZT control clcniciit and 251 fl for the hybrid element. The resonarice of the two iindiced layers in the hybrid clcmcnt appears near 5 NIHz, the width modes for both array clcments appear near 6 MHz, and the third harmonics of the thickness resonance are between 7 and 10 MHz. The vector impedance data arc in good agreement with the FEM predictions.

Fig. 7 shows the experiincntal pulsc-echo resiilts. The PZT control element had an amplitiide of 25 inV, a -6 dB bandwidth of 54%, and a normalized badness of 5.3 ( t o = 2.19 pS). The hybrid elcmcnt echo amplitude was 51 mV, with a -6 dB handwidth of 58%, and a normalized badness of 2.2 ( t o = 2.16 pS). Thus, the thick film processed hybrid array had an increased echo amplitude of 6.2 dB, increased bandwidth by a factor of 1.07, and dccreased badness by a factor of 2.4 compared with the PZT-5H transducer.

Because the piilsc-echo results from the thick filii1 processed array were not as good as predicted from simula-

1010 IEEE TRANSACTIONS ON ULTRASONICS, FERROELECTRICS, AND FREQUENCY CONTROL, VOL. 49! NO. 7 , JULY 2002

5.0

- 4.0 c Y

Q) 3.0 v S

E 2.0 cn m

v

4- .-

= 1.0

0.0

-1 0

-30 cn Q) v

u) m c

v -50

-70

0 2 4 6 8 10 Frequency (MHz)

(4


@>

Fig. 8. Experimental complex impedance results: a) magnitude and b) phase for the PZT-5H control and manually bonded hybrid array elements in water.

tions, we constructed another array using manual bonding techniques. A three-layer chip was fabricated by bonding together three layers of 11-MHz PZT-5H (t = 185 pm) using a low viscosity, non-conductive adhesive, Bowman FSA Adhesive #21991 (cyanoacrylate ester) (Bowman Distri- bution, Barnes Group, Inc., Cleveland, OH). As shown in Fig. 8, the series resonant frequency, from the experimental impedance plots, for the manually bonded multi-layer and the PZT control was about 2 MHz with a magnitude of 1980 R for the PZT control element and 202 s2 for the hybrid element. The single layer resonance in the hybrid element appears near 5 MHz, the width modes appear near 6 MHz, and, for both array elements the third harmonic of the thickness resonance is between 7 and 10 MHz.

Fig. 9 shows the experimental pulse-echo results for the manually bonded multi-layer versus the control. The PZT control element had an amplitude of 25 mV, a -6 dB bandwidth of 54%, and a normalized badness of 5.3 ( t o = 2.19 p S ) . The hybrid element echo amplitude was 91 mV, with a -6 dB bandwidth of 59%, and a normalized badness of 1 ( t o = 2.01 p S ) . Thus, the hybrid array had increased echo amplitude of 11.1 dB, increased bandwidth by a factor of 1.08, and decreased badness by a factor of 5 .3 compared with the PZT-5H transducer. There was good agreement

-.Jv

Time (pS) (4

n . . - PZT-5H Control

z E-10 Q) IEl 3 c, .- -

-20 a

-30 0 2 4 6 8 10

Frequency (MHz) @>

Fig. 9. Experimental pitch catch results: a) magnitude and b) phase for the PZT-5H control and manually bonded hybrid array elements in water.

between FEM predictions and experimental results in ev- ery measurement for this manually bonded array, except for the bandwidth results that are discussed subsequently.

After the water tank tests were performed, the PZT- 5H control and manually bonded hybrid arrays were connected to the Siemens SI-1200 medical ultrasound scanner (Siemens Medical Systems, Inc., Ultrasound Group, Issaquah, WA), and images were made using a CIRS (Nor- folk, VA) model 40 phantom (a = -0.5 dB/cm.MHz). From the images of the phantom [Fig. 10(a and b)], it can be seen that the hybrid array [Fig. 10(b)] has a greater depth of penetration by approximately 9 cm because of the increased SNR of 11.1 dB. (For a map of the phantom, see Fig. 11.)

I v . DISCUSSION AND CONCLUSION

This paper has shown simulated and experimental results, summarized in Tables VI and VII, of phased array transducers using PZT-5H versus two different multi-layer composite hybrids. The hybrid arrays were superior to the control array when SNR, bandwidth, and the badness pa- rameter were used as measures of performance. However,


(4 Fig. 10. Images of the CIRS model 40 phantom made with the a) PZT-5H control and b) PZT composite hybrid arrays.

0 cm

2

4

6

8

10

12

14

18

18

. J d t Y c m / ~ ~ z 15 om

Fig. 11. Map of the CIRS Model 40 phantom.

FEM simulations predicted a much greater improvement in bandwidth than what was realized experimentally.

We believe one reason for this lack of improvement is the disparity between material constants used in the FEM simulations and true material properties, including those of the ceramic, matching layer, backing, and the composite filler. Material properties were adjusted during the modeling process to improve the fit with experimental results. Future work is needed to better characterize these materials. Another reason for the lack of improvement is the effective increase in the dielectric constant (E:?) caused by

TABLE VI SUMMARY OF FEM RESULTS.

PZT-5H Hybrid

zs (a) 1750 256

-6 dB BW 46% 65% Normal badness 3.5

SNR (dB) 0 10.9

1

the edge terminations of the multi-layer, as discussed by Desilets [25]. This causes a decrease in the coupling coefficient (kA3), leading to decreased bandwidth that is not seen in the results for the control, which has no edge terminations. The effective coupling and dielectric constants for the PZT-5H control element were ICi3 = 0.75 and E&: = 910 at 5.6 MHz, as measured in isolated beams in air. The coupling coefficient for the three-layer thick film and manually bonded materials, as measured in air and calculated from the series and parallel resonant frequencies for single elements, were ki3 = 0.59 and k ig = 0.61 respectively. This matches Desilcts’ measurements from single layer edge- terminated element of IcA3 = 0.57. This decrease in coupling is caused by an increase in effective dielectric constants, which were E$$ = 1600 at 6.7 MHz for the thick film three layer and E;: = 1500 at 6.4 MHz for the manually bonded transducer. Eq. (1) shows how an increase in the effective E;: causes a decrease in ICi3 [20], [26].

(1)

Additional work is needed to improve the coupling coefficient and thus improve the bandwidth for the hybrid transducers. Desilets proposes isolation cuts to lessen the effect of the edge terminations. However, these types of cuts are not possible for a multi-layer transducer that con- tains internal electrodes. It may be possible to use isola-

1012 IEEE 'I'RANSACTIONS ON UI;L'RASONICS, FEI1IIOBLEC'I'RI~S, AND FREQUENCY CONTROL, VOL. 49, NO. 7, JULY 2002

TABLE VI1 SUMMARY OF EXPERIWIENTAL RESULTS

Hybrid Hybrid PZT-5H (thick film) (bonded)

ZS (0) 1980 251 202 SNR (dB) 0 6.2 11.1 -6 dB BW 54% 58% 59% Normal badness 5.3 2.2 1

Fig. 12. Micrograph of the three-laycr 5-MHz hybrid transducers. (Scale is 0.2 mm.)

tion cuts that are parallel to the internal electrodes or to use a low dielectric constant material to insulate the edge termination from the internal electrode.

Ringing can also be seen in the pulses from both the FEM and experimental results. Increasing the bandwidth of the transducer would tend to decrease the pulse length, but the use of at least two matching layers and better opti- mized materials would most likely be required to eliminate the ringing.

V. FUTURE WORK

We also tested the feasibility of our hybrid transducer design using higher frequency material by fabricating hybrid transducer prototypes with thick film (green tape) 5- MHz three-layer PZT chips from two manufacturers "B" and "C." These multi-layer chips were processed in a sim- ilar fashion to that described previously with the overall thickness being 300 to 400 pm and cuts approximately 80 to 100 pin deep at a pitch of 80 pm, as shown in Fig. 12. The 5-MHz arrays were diced into linear phased array elements (Lelr7,atLon M 13 mm, kerf = 0.03 mm, and pitch = 0.16 mm). However, these initial prototypes did not include matching layers.

Fig. 13 compares the 5-MHz experimental pitch-catch results for the PZT control and PZT composite hybrid arrays. The echo amplitude for the PZT array was 96 mV

100

A > 50 E v

I PZT Control 1i

-100 ' I

Time ($3)

(4 -70

h a S -80 Q) 'CI 3 CI .- - E" -90 Q

-1 00 0 2 4 6 8 10

Frequency (MHz)

tb)

Fig. 13. Experimental pitch-catch results: a) pulse and b) spectra for the PZT coiitrol and hybrid arrays in water.

with a -6 dB bandwidth of 19%. The ccho amplitude from the hybrid array was 150 mV with a -6 dB bandwidth of BO%, yielding an increase of about 4 dB in echo amplitude and increased bandwidth by a factor of 3.16. However, the predicted results from FEM were an increase of 10 dB in ccho amplitude and from about 20 to 40% in bandwidth. From the impcdance plot in air shown in Fig. 14, it can be seen that the loss tangent was higher than expected (20% for the hybrids compared with 6.6% for the control). This was a common problem for both manufacturers that caused decrease in echo amplitude. In addition, the effect of increased electrical loading (tan 6 = Q-' [27]) increased the bandwidth beyond what was expected. The cause of this increased loss tangent is unclear, but may be due to interlaminar electrodes that are too thin (1 to 2 pm in- stead of 5 pm), creating electrical resistance. It was also found that the transducer materials were not as flat as required for the composite dice and fill processing technique. Layer delamination was also discovered on several pieces upon inspection and when impedancc measurements were performed. It should be noted that the loss tangent for the 2-NlHz thick film transducers (manufacturer "A'), as measured from a single element, was 9% (Fig. 15), which was closer to that from the control at 5%, and the flatness of this material was acceptable for processing.


120 10

-1 0 100 ? [5)

C - 80 a, -30 -0 W

t -50 8 a 2

- .- 3 60 a,

3 40 c -70

0 -90 0 2 4 6 8 10

Frequency (MHz)

20

Fig. 14. Experimental impedance in air for the 5-MHz hybrid array element. Note the loss tangent of 20%.

4

h - PZTIComp. 3 Lay.

v s3 Q ‘EI

C m

s 2 .+, .-

SI 0


Fig. 15. Experimental impedance in water for the 2-MHz hybrid array element. Note the loss tangent of 9%.

In conclusion, the PZT composite hybrid structure presented in this paper shows good prospects for increasing the SNR and bandwidth for medical ultrasound transducers. However, some problems persist in the manufac- ture of hybrid transducer devices from thick film (grcen tape) materials, including reduced coupling coefficients, increased electrical loss tangent, and inadequate flatness of the multi-layer materials for dice arid fill processing.

REFERENCES

M. A. Averkiou, D. N. Roundhill, and J . E. Powers, “A new imaging technique bmed on the nonlinear properties of tis- sues,” in Proc. 1997 IEEE Ultrason. Symp., pp. 1561-1566. P. L. Mullen, “Harmonic Imaging,” Ultrasound Technology Up- date. GE Medical Systems, Milwaukee, WI, 1997. D.-L. D. Liu, P. Von Behren, and J. Kim, “Single transmit imaging,” in Proc. 1999 IEEE Ultrason. Symp., pp. 1275-1278. S. Saitoh, M. Izumi, S. Suzuki, and S. Hashimoto, “Ultrasonic probe system,” U.S. Patent 5 163 436, Nov. 1992. R. L. Goldberg and S. W. Smith, “Multilayer piezoelectric ceramics for two-dimensional array transducers,” IEEE Trans. Ul- trason., Ferroelect., Freq. Contr., vol. 41, no. 5, pp. 761-771, Sep. 1994. M. Greenstein and U. Kumar, “Multilayer piezoelectric ceramics for two-dimensional array transducers,” IEEE Trans. Ultrason., Ferroelect., Freq. Contr., vol. 43, no. 4, pp. 620-622, Jul. 1996. C. D. Emery and S. W. Smith, “Ultrasonic imaEing using a 5-

noise ratio,” IEEE Trans. Ultrason., Ferroelect., Freq. Contr., vol. 46, no. 5 , pp. 1101-1119, Sep. 1999. J . A. Hossack, “Modeling techniques for 1-3 composite transducers,” Ph.D. dissertation, Univ. Strathclyde, Glasgow, UK, 1990. T . R. Gururaja, W. A. Schulze, L. E. Cross, and R. E. Newnham, “Piezoelectric composite materials for ultrasonic transducer applications. Part 11: Evaluation of ultrasonic medical applications,” IEEE Trans. Sonics Ultrason., vol. SU-32, no. 4, pp.

[lo] T. R. Gururaja, A. Safari, R. E. Ncwnhani, and L. E. Cross, “Piezoelectric ceramic-polymer composites for transducer applications,” in Electronic Ceramic. L. M. Levinson, Ed. New York, NY: Marcel Dekker, 1987, pp. 97-128.

[I11 W. A. Smith and B. A. Auld, “Modeling 1-3 composite pieao- electrics: Thickness-mode oscillations,” IEEE Trans. liltrason., Ferroelect., Freq. Contr., vol. 38, nu. 1, pp. 40-48, Jan. 1991.

[la] H. L. W. Chan and J. Unsworth, “Simple model for piezoelectric ceramic/polymer 1-3 composites used in ultrasonic transducer applications,” IEEE Trans. Ultrason., Ferroelect., Freq. Contr., vol. 36, no. 4, pp. 434-441, Jul. 1989.

[13] D. M. Mills and S. W. Smith, “Combining multi-layers and corn- posites to increase SNR for medical ultrasound transducers,” in Proc. 1996 IEEE Ultrason. Symp., pp. 1509-1512.

[ 141 S. W. Smith, “Multi-layer composite ultrasonic transducer arrays,” U.S. Patent 5 548 564, Aug. 1996.

[15] A. Cochran, P. Reynolds, and G. Hayward, “Multilayer piezo- composite transducers for application of low frequency ultrasound,” in Proc. 1997 IEEE Ultrason. Symp., pp. 1013-1016.

[I61 T. R. Gururaja and L. A. Ladd, “Ultrasonic transductor,” U.S. Patent 5 625 149, Apr. 1997.

[17] M. Zipparo, C. G. Oakley, and M. He, “Multilayer ceramics and composites for ultrasonic imaging arrays,” in Proc. 1.999 IEEE Ultrason. Symp., pp. 947-952.

[18] M. S. Seyed-Bolorforosh, M. Greenstein, D. Harriott, and T. R. Gururaja, “Hybrid piezoelectric for ultrasonic probes,” U.S. Patent 5 638 822, Jun. 1997.

[19] D. M. Mills and S. W. Smith, “Multi-layered PZT/Polymer composites t o increase signal-to-noise ratio and resolution for medical ultrasound transducers,” IEEE Trans. Ultrason., Femoelect., Freq. Contr., vol. 46, no. 4, pp. 961-971, Jul. 1999.

[20] A. R. Selfridge, “The design and fabrication of ultrasonic transducers and transducer arrays,’’ Ph.D. dissertation, Stanford Univ., Stanford, CA, 1983.

[21] G. R. Lockwood and F. S. Foster, “Modeling and optimization of high-frequency ultrasound transducers,” IEEE Trans. Ultrason., Ferroelect., Freq. Contr., vol. 41, no. 2, pp. 225-230, Mar. 1994.

[22] N. N. Abboud, G. L. Wojcik, D. K. Vaughan, J. Mould, D. J. Powell, and L. Nikodym, “Finite element modeling for ultrasonic transducers,” in Proc. SPIE Med. Imag. 1998: Ultrason. Transducer Eng. Conf., vol. 3341, pp. 19-42.

[23] L. L. Ries and S. W. Smith, “Finite element analysis of a de- formable array transducer,” IEEE Trans. Ultrason., Ferroelect., Freq. Contr., vol. 46, no. 6, pp. 1352-1363, Nov. 1999.

[8]

[9]

499-513, ,Jd. 1985.

[24] D. J. Powell, private communication, Mar. 29, 2000. [25] C. S. Desilets, D. J . Powell, N. Abboud, and G. L. Wojcik,

“Effcct of wraparound electrodes on ultrasonic array performance,” in Proc. 1998 IEEE Ultrason. Symp., pp. 993-997.

[26] G. S. Kino, Acoustic Waves. Englewood Cliffs, NJ: Prentice- Hall, Inc., 1987.

[27] D. Berlincourt, “Piezoelectric crystals and ceramics,” in Ul- trasonic Transducer Materials. 0. E. Mattiat, Ed. New York: Plenum Press, 1971.

David M. Mills (M’02) was born in Ban- ner Elk, NC on May 5, 1972. He received the BS degree in engineering (summa cum laude) from LeTourneau University, Longview, TX, in 1994 and the Ph.D. degrec in biomedical engineering in 2000 from Duke University, Durham, NC.

While a t Duke University, he was a member of the NSF/Engineering Research Center (currently the Center for Emerging Cardiovas- cular Tedinolories) as a me-Doctoral Fellow.

- I I - I

MHz multilayer/single-laycr hybrid array for increased signal-to- His research included finite element modeling

1014 IEEE TRANSACTIONS ON ULTRASONICS, FERROELECTRICS, AND FREQUENCY CONTROL, VOL. 49, NO. 7, JULY 2002

and fabrication of multi-layer PZT polymer composites to increase SNR and bandwidth for medical ultrasound transducers.

In 2000, he joined the GE Global Research Center as a biomedical engineer to continue research and development of medical ultrasound transducers. He is a member of IEEE-UFFC.

Stephen W. Smith (M’91) was born in Cov- ington, KY on July 27, 1947. He received the BA degree in physics (summa cum laude) in 1967 from Thomas More College, Ft . Mitchell, KY; the MS degree in physics in 1969 from Iowa State University, Ames; and the Ph.D. degree in biomedical engineering in 1975 from Duke University, Durham, NC.

In 1969, he became a Commissioned Of- ficer in the U.S. Public Health Service, as- signed to the Food and Drug Administration, Center for Devices and Radiological Health;

Rockvil!e, MD, where he worked until 1990 in the study of medical imaging, particularly diagnostic ultrasound, and in the development

of performance standards for such equipment. In 1978, he became an adjunct Associate Professor of Radiology at Duke University Medi- cal Center. He is currently Professor of Biomedical Engineering and Radiology. He holds 16 patents in medical ultrasound and has au- thored 140+ publications in the field.

Dr. Smith is co-founder of Volumetrics Medical Imaging, Inc. and Memscept, Inc. He has served on the education committee of the American Institute of Ultrasound in Medicine, the executive board of the American Registry of Diagnostic Medical Sonographers, the editorial board of Ultrasonic Imaging, and the Technical Program Committee of IEEE-UFFC. He was co-recipient of the American In- stitute of Ultrasound in Medicine Matzuk Award in 1988 and 1990 and co-recipient of the IEEE-UFFC Outstanding Paper Award in 1983 and 1994.

Multi-Layered PZT/Polymer Composites to Rat io and...

Documents

Transcript of Multi-Layered PZT/Polymer Composites to Rat io and...