Вы находитесь на странице: 1из 7

ELSEVIER

Sensors and Actuators A 44 (1994) 111-117

PHYSICAL

Piezoelectric

polymer composite arrays for ultrasonic medical imaging applications


H.I. Schlaberg**, J.S. Duffyb
Univtmi~ of Bradfoni, Bradford 807 bPen.im hfefnmim Lrd, Bradford 807 IDP, UK IDQ, UK 22 February 1994

Received

28 July 1993; in revised form 11

February 1994; accepted

Abstract The predominant sensar materials used in medical acoustic imaging have traditionally been piezoelectric ceramics (PZT, LaNbO,, etc.) More recently piezoelectric polymers (PVDF) have been used in this area. Polymers present several advantages over ceramics (very flexible, easy to process, good acoustic matching to tissue, good reception constant), but also have some major drawbacks (poor emission constant, low coupling coefficient). Composite piezoelectric materials aim to incorporate the most advantageous properties of ceramics and polymers. This paper investigates the electrical and mechanical properties of two different piezoelectric polymer composite materials (Piezoflexl and Piezel) to assess their performance for medical imaging applications. Practical transducer constructions with corona-poled, PCB-backed and diced-array architectures have been manufactured with the composites, and their performance evaluated in pulse-echo mode. The results obtained, using a simple object as target, confirm the potential use of the composite materials in medical imaging applications.
Keywords Piezoelectric polymers; Ultrasonic imaging

1. Introduction

Ultrasonic imaging offers many potential advantages over other forms of medical imagining due to its nonradiative properties, low cost and simple transducer architecture. The key element in forming these images is the ultrasonic transducer, which generates the probing acoustic pulse and detects weak echoes from the objective target. These devices sustain a driving pulse of a few hundred volts, and microseconds later, detect weak echoes with a noise floor of a few microvolts. The prerequisites of an ultrasonic transducer element to be used for medical imaging are: (1) it couples energy into its measurement medium efficiently; (2) it generates 10-110 mW/cm2 of acoustic power from reasonable input voltages; (3) it provides good signal-to-noise ratio; (4) low cross-coupling between elements;
Present address: Department of Electrical of Leeds, Leeds LS2 9JT, UK. Engineering, University

(5) it generates a wide range of acoustic frequencies and pulse lengths ~2 ps in duration; (6) it is lightweight and robust. Although traditionally piezoelectric ceramics are used as array materials, they only partially meet these requirements. Ceramics have high electromechanical coupling k, (40-50%) and low losses (tan delta <3% and Q, B 50), but they have a poor acoustic match to tissue (24l-30 MRayls) and present difficulties in shaping and subdividing [l]. As an alternative to piezoelectric ceramics, the piezoelectric polymer polyvinylidene-fluoride (PVDF) has been extensively researched as an alternative sensor material for ultrasonic imaging and other acoustic sensing applications. In particular, the wide low-Q monotonic bandwidth coupled with the close acoustic impedance matching of PVDF (3 MRayls) to tissue (1.5 MRayls), compared to PZT (23 MRayls) offers potentially very high axial image resolution. Additionally the high compliance and stability of the material enables non-planar transducer architectures to be achieved. Unfortunately, whilst PVDF has a high receiving constant (PVDF g,, = - 310 X 10m3 V m/N; PZT (PC5)

0924-4247/94/$07.00 Q 1994 Elsevier SSDI 0924-4247(94)00791-F

Science S.A. All rights reserved

112

H.I. Schlaberg, IS. Duffv f Sensors and Actuators A 44 (1994) 111-117

g33=25~ 10v3 V m/N), it has a very low thicknessmode coupling coefficient (PVDFK,, = 0.12; PZT (PC5) I&= 0.70). This leads to extremely poor sensitivity and has limited to use of PVDF in transceiver applications. In an attempt to achieve an optimized ultrasonic sensor material, which maintains the inherent advantage of the wide monotonic bandwidth of PVDF, whilst offering high thickness-mode electromechanical coupling, researchers have concentrated on the development of copolymers and polymer-ceramic composites as an alternative [3-6]. This paper compares the material properties of two different polymer-ceramic piezoelectric composites and evaluates their acoustic performance for use in imaging arrays. In order to evaluate the performance of the materials in real-time application domains, several transducer array devices have been constructed and the results are reported.

2. Material properties Two forms of piezoelectric polymer composite films have been evaluated. The first, known as Piezel, comprises a lead zirconate titanate (PZT) core embedded in a matrix of the piezoelectrically active polymer polyvinylidene-fluoride (PVDF) with a volume distribution of 50-50%. The second material, Piezoflexl, also has a piezoelectrically active core of PZT but is embedded in an inactive epoxy resin matrix with a volume distribution of 5&50%. Traditional methods have been used to evaluate the mechanical, acoustic and electrical properties. Table 1 shows a comparison of such properties relative to PZT and PVDF. The measured sonic velocities in the composites (1600-2100 m/s) are between the values of the ceramic (2830 m/s) and the polymer (1400 m/s). A lower sonic velocity is preferred because it has a direct influence on the acoustic impedance of the material: Z=pc, where Z is the acoustic impedance, p the density of the material and c its sonic velocity. The ideal acoustic

impedance for a transducer material would be a value equal to that of tissue (1.5 MRayls), in order that all the acoustic energy from the transducer is transmitted into the tissue and not reflected back at the transducer-tissue interface. One of the major disadvantages of PZT is its high acoustic impedance (22 MRayls), making it necessary to add (often more than one) quarter-wavelength matching layers to reduce the effective acoustic impedance and improve the energytransmission characteristics of the transducer. The acoustic impedance of PVDF (3 MRayls) and composite materials (9-10 MRayls) compare favourably with that of tissue. The magnitude of the piezoelectric transmission coefficient d,, relates the effect of the imposed voltage along the z axis (poling direction) to the tensile strain along the x axis. The composite materials present a low d,, (Piezoflexl = 4.6 PC/N) compared to ceramics (PZT= 175 PC/N). The cross-sensitivity (d,,ld,,) of Piezel (60%), however, is larger than that of PZT (42%), while Piezoflexl has a much reduced value (18%), making it particularly suited for monolithic arrays, since the voltage applied to one element has only a reduced effect on lateral elements when they are in physical contact. The high elastic-stiffness constant of ceramics makes the manufacture of transducers difficult, especially when focusing arrangements are required and when dicing independent elements. Conversely, the high compliance of PVDF and composite films enables the construction of spherically focused radiators and the films are readily sliced to form acoustic arrays. Softer materials (low c,,) are generally more desirable, since the stiffness constant inversely affects the reception and transmission efficiency of piezoelectric materials, as can be seen from the following transmission and reception parameters [9]:
li2

Table 1 Comparison of physical properties behveen ceramic, polymer and composite materials Property Density kg/m, Sonic velocity m/s Acoustic impedance MRayls Compliance 10m9 m /N Relative petmittivity Electrical loss tan (at 1 kHz) d,, PC~N d,, PC/N g3, mV m/N PZT(PC5) 7750 2830 22 0.02 1800 0.02 410 - 175 26 PVDF 1800 1400 3 0.1 10 0.05 30 -18 340 Piezoflexl 4500 2100 10 13 32 0.08 2.5 ~ 4.6 88 Piezel 5600 1687 9.8 0.25 70 0.047 40 -24

HI. Schlaberg, J.S. Dufi

I Sensors and Actuators A 44 (1994) 111-117

113

where t is the thickness, k, is the thickness coupling coefficient, cg is the mechanical stiffness coefficient at constant charge density and & is the dielectric permittivity at constant strain. The parameters also show the importance of k,, as is intuitively obvious, whilst & has a conflicting influence. The mechanical behaviour of the composites has also been evaluated using dynamic mechanical thermal analysis (DMTA). The elastic stiffness and mechanical loss tan delta of both polymer composites have been measured and compared. Poled and unpoled samples have been examined: (i) at ambient room temperature and various mechanical frequencies ranging from 0.3 to 50 Hz, (ii) at 0.3 Hz and 50 Hz over a range of temperatures Gem 0 to 50 C. The DMTA technique assesses the structure and properties of solids and viscoelastic liquids via their dynamic moduli and damping. Changes in these parameters are studied as a function of temperature and impressed frequency. The method enables changes in internal molecular mobility, phase structure and morphology to be detected. The glass transition and secondary relaxations are also revealed. When a sinusoidal stress is applied to a perfectly elastic solid, the deformation (and hence the strain) occurs exactly in phase with the applied stress. A completely viscous material will respond with the deformation lagging 90 behind the applied stress. When a sinusoidal stress is applied to a viscoelastic material it will behave neither as a perfectly elastic nor as a perfectly viscous body and the resultant strain will lag behind the stress at some angle (6). The magnitude of the loss angle is dependent upon the amount of internal motion occurring in the same frequency range as the imposed stress. The complex dynamic modulus E* is defined as (stress E * = (strain amplitude) The complex modulus, however, does not take into account the phase and it is convenient to extract the completely elastic and completely viscous components of the deformation. The storage modulus E is defined as E, = (amplitude of stress component in phase) (strain amplitude) The loss modulus E is defined as: E= (amplitude of stress component out of phase) (strain amplitude) The storage modulus is the elastic response and corresponds to completely recoverable energy, whereas

the loss modulus is the viscous response corresponding to energy lost through internal motion. The tangent of the loss angle, tan delta, is dimensionless and is equal to the ratio of energy lost (dissipated as heat) to energy stored per cycle, where:

tan delta = (storage modulus~ =E/E


A material may be scanned over a range of temperatures and at various imposed frequencies. A loss maximum in the temperature scan is observed when the frequency of a motional process coincides with the impressed frequency. With an increase in the measurement frequency, the loss process is generally found at higher temperatures, where the molecular motion is faster. Using the described method it has been found that Piezoflexl, in its poled and unpoled states, exhibits a large tan delta loss peak (55%) at 30 C when examined at 0.3 Hz (Fig. 1). This indicates a state transition of the material occurring around the preferred application temperature, which leads to a degradation in the acoustic performance since a greater proportion of the energy is transformed into heat. The loss peak is associated with the partial relaxation of the polymer structure. This occurs near Tg (glass transition temperature) at low frequencies. At temperatures well below T, the loss is small, since nearly all the energy stored in deforming the material is quickly recovered when the stress is removed as molecular slipping and other motions are frozen in [lo]. Piezel does not show a damping peak in the temperature range considered, indicating that its T, lies beyond the range of application. When an analysis is performed over the same temperature range, but at the higher impressed mechanical frequency of 50 Hz, the same peak is found at approximately the same temperature of 30 C (Fig. 2). However, at this frequency the value of the peak is
06

(loss modulus)

04

I :

0.3

02

J
0

I 20

I 30

I 40 50

10

Temperature Fig. 1. DMTA analysis as a function of temperature at 0.3 Hz 0, Piezoflexl not poled; A, Piezoflex 1 poled; *, Piezel not poled; 0,
Piizel poled.

114

H.I. Schlabe~

IS. Duffy I Semom and Actuators A 44 (1994) 111-117

10

20

30

40

50

Temperalute Fig. 2. DMTA analysis as a function of temperature at 50 Hz 0, Piezoflex 1 poled; A, Piezoflexl not poled; * , Piezel poled; 0, Piezel not poled.
0.5 0 45

04 035 :a 03 d 0.25 1 02 015


01 0.05 0 0

10

15

20

25

30

Frequency

Fig. 3. DMTA analysis as a function of frequency at 25 C: 0, Piezoflexl; n, Piezel.

reduced to 14% from 55% at 0.3 Hz. The value of the loss in Piezel in its poled and unpoled states is also greatly reduced (from over 10% to under 4%). Analysis of the tan delta loss progression over a range of frequencies from 0.3 to 30 Hz at 25 C (Fig. 3) indicates a trend towards lower losses with increasing frequencies. Extrapolation to the operational region of 3 MHz predicts a very low loss for these frequencies. It is observed that following poling, Piezoflexl exhibits a small increase in stiffness and a large increase in brittleness. No such pronounced change was observed in Piezel before and after the poling procedure.

finements of a marketable product, these architectures show the level of success that may be obtained with the researched materials. The array architectures range from a five-element monolithic corona-poled device with a spot diameter of 5 mm to a 32-element diced array with an element width of 1 mm. S-element diced array and 16-element monolithic printed circuit board (PCB) array transducers have also been developed. The geometric limitations of the corona-poling technique restrict its practical use to the creation of relatively low numbers of elements ( -=K 20) with element size > 2 mm diameter. Corona-poled devices have been investigated by constructing several transducers (Fig. 4) using the following technique. A mask with the element arrangement and appropriate sizes (2-10 mm) is constructed from a thin sheet of aluminium. Aluminium electrodes are then vacuum evaporated onto the polymer composite material with the pattern of the mask. Each element is then poled in a field of 15 kV for a period of 30 min at a temperature of 60-80 T, after which a common electrode is evaporated onto the remaining side. Fig. 4 shows the assembly of an active transducer element array. The perspex backing provides a good matching (2=3.1 MRayls) for Piezel and with an attenuation factor of 3.2 dB/cm (at 5 MHz) provides nearly 40 dB attenuation of the signal in the backing medium, so that no echo is received from this side of the transducer. The corrosive nature of water and acoustic coupling gels necessitates the use of a protective polyethylene film or a thin Perspex sheet. The construction of diced arrays (Fig. 5) is more complex. A plate-poled sample of the composite material is bonded to the Perspex backing with a low-viscosity adhesive to create a very thin bond that will not degrade the acoustic performance of the array. At the same time, a conductive bond is created between the composite and a thin copper sheet (50 pm) that serves as the driving electrode. Conductive adhesives are usually silver-loaded epoxies, which have a high viscosity and require heating during the curing process. A second thin copper sheet is adhered to the edge of the topside to create a common electrode.
,;=?

Corona poled,,,
QlQnQntS

?T=gyJ

3. Array construction As a means of optimizing an imaging array architecture for medical use, a number of different array structures have been fabricated and their characteristic acoustic performance evaluated. While lacking the re0

i ,< I,, ,

0 0 0

,,;:;;p$:

, , , ,/ I , , 0 I
frontface

Thin

protective

Fig. 4. Corona-poled

transducer architecture.

HI.

Schlaberg J.S. Dun

I Sensors and Actuakm

A 44 (1994) Ill-117

115

Fig. 5. Diced transducer

architecture.

Etched tapper electrodes

PCBBacltng

Fig. 6. PCEt array architecture.

The bonding process is crucial to the acoustic performance of the device. Non-uniform adhesion generates differences in the response of elements and a bond that is too thick will degrade the performance of the transducer, introducing added attenuation and narrowing the bandwidth of the device [7]. Dicing of the transducer is performed to a depth that minimizes crosstalk between the elements. At a few wavelengths depth (1 mm), Rayleigh waves across the back of the cuts are suppressed by this barrier. Finally the electrical connections are made and a thin film of polyethylene is placed over the elements to protect them from corrosion and possible crosstalk resulting from infiltration of water or gel between the elements. The 16-element monolithic array using a PCB as backing (Fig. 6) was constructed using a poled sheet of Piezel [2]_ The conductive layer of evaporated aluminium was removed and this sample adhered to a PCB board where the electrode profile has been etched out of copper. The elemental profile is 2 mm X 10 mm. The magnitude of the echo obtained with this transducer was similar in amplitude of that of the diced arrays, but the simpler manufacturing process led to a noticeable reduction in the transducer ringing. However, due to the large acoustic impedance mismatch at the transducer backface, a secondary transmit pulse is generated, which interferes with echoes from the main pulse.

pulse with a peak amplitude of 120 V, a sharp falling edge of 8 ns and a recharging time of 150 ns. Fig. 7 shows that the 16-element transducer produced slightly larger and shorter acoustic signals than the 32element versions. The increase in the receiving amplitude can be attributed to the greater area of the 2 mm wide elements generating an increased force for the same unit stress. The 32-element arrays exhibited fairly uniform responses and only certain elements showed marked differences, which were attributable mainly to bonding defects. Piezel generated a shorter echo response (especially noticeable in the top traces of Fig. 7) in comparison to Piezoflexl, while the array using Piezoflexl had a greater signal amplitude but also showed increased ringing. Despite the large loss peak observed in Piezoflexl, there are only relatively small differences observed in the signal amplitudes of similar Piezoflexl and Piezel transducers (Figs. 9 and 10). This may be accounted for by the high operational frequency of the transducer, at which very reduced losses have been predicted. Fig. 9 shows the unequal responses of different elements under identical acoustic conditions. This may be attributed to manufacturing imperfections where the bonding agent is non-uniform in certain regions, giving rise to signal attenuation; alternatively, the adhesive

Fig. 7. Echo from 16.element diced Piezel array.

4. Measurement results The shape of the stimulus applied to the transducers in the following measurements was that of a negative

Fig. 8. Echo from Piezel PCJ3 array.

116

H.I. Schlaberg

LS. Dufi

1 Sensors and Actuators A 44 (1994) Ill-117

Fig. 9. Echo from 324ement

diced Piezel array.

Fig. 10. Echo from 32-element diced Piezoflexl array.

of such a test where a 12 mm thick Perspex block is positioned in front of the transducer and element number 9 is activated. A 10 mV echo is received at the emitting element, while 6 mV signals appear at the immediately adjacent elements. This large signal is, however, not due to the emission of pressure waves of these adjacent elements themselves, but results from the reflection of the signal emitted from the central element. This is supported by the observed delay in the signal atrival at elements 8 and 10 commensurate with the signal transmission path between elements 8 and 9. The effect on elements 7 and 11 is almost negligible because the beam dispersion beyond the immediately adjacent elements is very small. Linear and phased array constructions have been considered. In a phased array all the elements of the transducer are activated in sequence to generate a consistent wavefront directed at a certain angle. The time between the activation of the elements determines the angular emission of the wavefront: t=d sin(p)/v, where d is the distance between element centres, p is the emission angle and 21 is the velocity of sound in the medium [8]. To avoid the effect of grating lobes, which have a degrading influence on the acoustic results, the distance between elements should be smaller than A/2. Ihe composite material with a thickness of 200 pm generates an acoustic wave of around 3.5 MHz. This corresponds to a thickness of 500 pm in water, giving a recommended distance between element centres of 250 pm. Because of the reduced dimensions of this type of array, the linear array has been preferred for the current evaluation, leaving the phased array construction for a later stage. Several circuits have been designed to control the transducer array and transmit the results to a computer for storage and analysis for image construction, These circuits include pulser and amplifying circuits, A/D conversion and high-speed memory board and pulser control and digital I/O circuits.

5. Conclusions
Fig. 11. Crosstalk on 16-element monolithic F iezel array.

may have accumulated in a pocket causing a deformation of the polymer surface, which results in acoustic radiation oblique to the normal surface of the transducer. Results obtained from the monolithic PCB array are shown in Fig. 8. The crosstalk between elements of the monolithic transducer has been examined by activating one central element and observing the response in the elements to either side of the main emitter. The relatively low & constant of the composites indicates that only a small cross-coupling effect in neighbouring elements should be observed. Fig. 11 shows the results

The mechanical and electrical characteristics of two piezoelectric polymer composite sensor materials have been investigated. A mechanical relaxation at room temperature has been observed in the Piezoflexl material. However, in the construction of applicationoriented transducer arrays the acoustic performances of the two materials are not significantly dissimilar. The PCB-based construction of a transducer array provides a cleaner signal than the diced arrays due to the simple manufacturing process involved. However, the impedance mismatch at the air-PCB interface on the back of the transducer creates a second echo that makes this kind of array unsuitable in its present form.

&I. Schlabq, J.S. Dufi / Sensors and Actuators A 44 (1W)

III-117

117

Acknowledgements This work was supported in the form of a studentship by the Science and Engineering Research Council, UK. The authors are grateful to Morgan Matroc Ltd. for the provision of Piezoflex samples.

[7] J.D. Larson, An acoustic transducer array for medical imaging - Part 1, Hewlett-Packard J., (Oct.) (1983) 17-22. [8] J.C. Somer, Electronic sector scanning for ultrasonic diagnosis, Ultrasonics, 6 (1968) 153-159. [9] J. Callerame, R.H. Tancrell and D.T. Wilson, Transmitters and receivers for medical ultrasonics, IEEE Ultrasonics Symp. Rot., (1979) 407411. [lo] L.E. Nielsen, A4echunicnl properlies of Po@ners and Composite, Vol. 1, Marcel Dekker, New York, 1974.

References VI VI
[31 W.A. Smith and AA. Shaulov, Composite piezoelectrics: basic research to a practical device, Ferroelectics, 87 (1988) 309-320. A. Cochran, D. Reilly, G. Hayward and A. McNab, Beam forming in solids using monolithic ultrasonic arrays, IEEE Ultrasonic Symp. I+oc.., (1989) 695698. Y. Daben, Composite piezoelectric film made from PVDF polymer and PCM-PZT ferroelectric ceramics, Feroelectics, 101 (1990) 291-296. K. Ogura, K Ohya and H. Banno, Hydrostatic pressure properties of piezoelectric flexible composites, Jpn. J. Appl. Phys., 28 (1989) 60-62. H. Banno, K. Ogura, H. Sobue and K. Obya, Piezoelectric and acoustic properties of piezoelectric flexible composite, Jpn. J. AppL Phys., 26 (1987) 153-155. T.R. Gururaja, R.E. Newnham, K.A. Klicker, S.Y. Lynn, W.A. Schulze, T.R. Shrout and L.J. Bowen, Composite piezoelectric transducers, IEEE Ubasonics Symp. Boc., (1980) 576581.

Biographies Ifi& S&l&erg obtained his first degree in electronic engineering at the Escuela Universitaria de Ingenieria Tecnica, Mondragon, Spain. He received an M.Sc. in real-time electronic systems from the University of Bradford, UK, in 1989. He is currently completing a Ph.D. at the University of Bradford, UK His interests include ultrasonic composite sensors and acoustic imaging.
John Dufi graduated in electronic engineering in 1978 and received his doctorate in electronic materials science from the University of Wales in 1982. He is currently a lecturer in advanced sensor systems at the University of Bradford. His main interests are in the field of bioelectronics and in particular the application of novel sensor systems.

[41

Bl

Fl

Вам также может понравиться