IEEE TRANSACTIONS

ULTRASONICS,
ON
AND
SONICS

481

VOL. SU-32, NO. 4. JULY 1985

Piezoelectric Composite Materials for Ultrasonic

Transducer Applications. Part I: Resonant

Modes of Vibration of PZT

Rod-Polymer Composites

T. R.GURURAJA, WALTER A. SCHULZE, MEMBER,

ROBERT E. NEWNHAM,BERTRAMA.AULD,

IEEE,

LESLIE E. CROSS, FELLOW,

YUZHONG J. WANG

IEEE,

AND

Abstmcr-Theobjective
of thepresentworkwastogainadeeper
mensions of the transducer are muchsmallerthanthe
understanding of the behaviorof lead zirconate titanate (PZT) polymer acoustic wavelength [ l]-[8]. Until now there have been no
composites for applications such as ultrasonic medical diagnosis in the
studies investigating the usefulness of these composites at
megahertzfrequencyrange.Thesecompositeswereoriginally
develhigher
frequencies (1-10 MHz) for medical diagnostic and
oped for low-frequency hydrophone applications. The PZT rod-polynondestructive
testing applications. At these frequencies,
mer composites have been prepared with
five to 30 volume percent PZT
using 0.28 mm and 0.45 mm rods. In a disc of PZT rod-polymer com- the acoustic wavelength is comparable to the scale of the
posite material, there are three principal types
of resonance: the planar composite microstructure. Theacoustic impedance, bandmode, the thickness mode, and various lateral modes caused
by the regwidth, and radiation patterns of the composite transducer
ular periodicityof the PZT rod in the composite. These resonance modes
have been studied with the following techniques:
1) electrical impedance can be controlled in a manner so sophisticated that it is
measurement as a functionof frequency and 2) laser probe dilatometry impossible in single phase materials. The results of a sysof the dynamic displacementas a function of frequency and position in tematicinvestigation
of thecompositematerialsmade
the composite lattice. The observed resonance behavior is found to be frompiezoelectric lead zirconate titanate (PZT) ceramics
a result of lateral interactions in the composite through the epoxy meand piezoelectrically inactive polymer are reported here.
dium. The effect of temperature on the electromechanical properties
of electromeThisstudyfocusesontheunderstanding
of the composite has also been investigated. Implications
of these results
chanical
properties
of
the
composite
materials
in resonant
for optimizing the design of ultrasonic transducers are discussed.

configurations. The knowledge of the high-frequency dynamic behaviorof the composite was then used to evaluate
thecompositematerialsforultrasonictransducerappliI. INTRODUCTION
cations with an emphasis on medical diagnostic applicaHE DESIGN and fabrication of composite materials tions. These results are discussedin an accompanying paoptimized for a special application has been the sub- per (Part 11).
In Section I1 the requirements of a piezoelectric transject of extensive research recently. The applications range
ducer
for ultrasonic imaging applications and the
limitafrommechanicalstructures to electronic devices.In detions
of
the
existing
single
phase
transducer
materials
are
signing composite materials, primary importance is given
described.
In
Section
1
1
1
a
brief
review
of
the
earlier
works
to the proper choice of component phases and the way in
on piezoelectric ceramic-polymer composites of relevance
which they are interconnected to maximize a predefined
tothepresent
work isgiven.Advantages
of theuse of
figure of merit for the application envisaged.An important
for
ultrasonic
applications
are
pointed
out. The
composites
class of these new materials is the family of piezoelectric
criteria
used
to
select
an
appropriate
composite
structure
ceramic-polymer composite transducers. To date, most of
the work on piezoelectric ceramic-polymer composite ma- for a detailed investigation for high frequency transducer
terials has been focused on hydrophone applications in theapplications are also described. Bulk mechanical properlow-frequency(lessthan 40 KHz) range,where the di- ties of the polymer phase that strongly influence ultrasonic
properties of the composite are summarizedin Section IV.
The characterization of all the different resonance modes
ManuscriptreceivedNovember 26, 1984. This work was supported in
in the composite are discussed in Section V. Laser probe
part by North American Phillips Laboratories.
T. R. Gururaja, L. E. Cross, and R. E. Newnham are with the Materials measurements are presented in Section VI. The effect of
Research Laboratory, The Pennsylvania State University, University Park,
temperature on the resonance modesin the composite are
PA 16802, USA.
VII. Finally, the implication of the
B. A. Auld and Y.J . Wang, are with the Edward L. Ginzton Laboratory, dealt with in Section
Stanford University, Stanford, CA 94305, USA.
resultson thehigh-frequencyvibrationbehavior
of the
W. A. Schulze was with the Materials Research Laboratory, The Penncomposites
in
air
for
the
ultrasonic
imaging
applications
sylvania State University. He is now with the New York State College of
are summarized in Section VIII.
Ceramics, Alfred University, Alfred, NY 14802, USA.

482

TRANSACTIONS
IEEE

Desilets er al. [ 121 and Go11 [ 151 have further

improved
of their
the
bandwidth
and
the
sensitivity
disc
transducers
usingmultiplelayerimpedancematching
schemes. AlThe basic requirements of a piezoelectric transducer for though two layer matching schemes work very well with
ultrasonicdiagnostic imagingare an ability to generate readilyavailable materials, thetransducerfabrication is
short acoustic pulses and receive them with
high sensitiv- quite involved. As the frequency of operation is increased,
ity in the frequency range Of '-lo MHz. The transducer the thickness ofthe matching layer decreases, which makes
should also exhibit good axial and lateral resolutionto vis- the fabrication
difficult.
ualize structures with dimensions of the order of a milliThe recent development of piezoelectric polymer mameter or less L91. In terms Of the
parameters,terialssuch
aspolyvinylidenefluoride(PVF2)hasopened
these requirements can be summarized as follows: 1) The
up new possibilities for transducers operating with media
piezoelectric material should have a high electromechanof low acoustic impedance I161. PVF2 has a low acoustic
ical coupling coefficient forhighsensitivityand
2) The
impedance, reasonable piezoelectric coupling, high 933
acousticimpedance
of thetransducershouldbe
well
coefficient, and wide-bandwidth frequency response,
matched to that of the load to minimize reflection losses
which are excellent properties for acoustic imaging with a
at the interface. The acoustic impedance matching at the low-impedance load. Several intrinsic limitations of PVF2
interface determines the quality factor ( Q broadband na- transducers, as discussed in detail by Collerame et al. [ 171
ture) of the transducer.
and Hunt er al. [ 181, restrain their extensive use in ultraIt is well known that the piezoelectric ceramic materials sonic imaging applications. Major limitations are mensuch as PZTwithhighelectromechanicalcouplingsuffertioned
below. First of all, it is very difficult to produce
from a Severe disadvantage when Operated with a low- sampleswithresonancenearthetypicaloperating
fieimpedance load such as that Of the human body. The quency in medicalultrasound;although recentlysomeresearchers have beensuccessful in producing 1-2 mm thick
acoustic impedance Of PZT i' around 30 x lo6
whilethetissues have acoustic impedancenearthat
of polarized PVF2,films[191,The d33coefficient of PVF2 is
water, 1.5 kg/m2-s. ~ e n c e f o r t hall acoustic impedances only l o p C/N, and hence pVF, is a poor ultrasonic transAs mitter. In addition PVF2 has
dielectric
a
constant
of only
be
given in units Of IO6 kg/m2-s,
at KHz whichdecreases at higher frequencies. Thus,
a result of the aco.ustic impedance mismatch,thecoupling
of the acoustic energy at the transducer-load interface is the
electrical impedance of PVF2transducers is high and
Poor. The resonance behavior is thus high Q (narrow is in general more difficultto match than the piezoelectric
bandwidth). The consequence of a high Q in a transducer ceramics. The dielectric loss tangent in P V F ~is 0.25 at
a 'low Pulse-rise time and a Prolonged ring down with 2.5 MHz and acts as a major Source of absorption of eleca detrimental effect on range resolution.
tricalenergyappliedto
the transducer.Inthereceiver
In practice, two methods are used to increase the trans- mode, the dielectric loss is found to diminish the signal to
ducer bandwidth. In the first method, the resonanceof the noise ratio.
piezoelectric element is mechanically damped with a well
As discussed previously, single-phase piezoelectric mamatched backing medium [l0]. As a
a large POrtion terials(ceramic or polymer)arefarfrombeingoptimized
of the available energy is transmittedinto the backing ma- for ultrasonicimaging applications. Thus, a composite
terialresultingin
a lower sensitivity.In
the second material combining thedesirable properties of
differmethod, the frequency bandwidth Of a transducer is in- ent phases might be superior. This concept has proven to
creased
the use Of an impedancematching layer Of behighlysuccessful
in designingcompositematerials
quarter wavelength in thickness and characteristic imped- consisting of PZT and piezoelectrically inactive polymer.
ante
to the geometric
Of those Of the transThe hydrostaticpiezoelectric coefficients of thesecornNore
["I, posites were two to three orders higher than that of the
and
ducer
load [l0]. and
1121, andSouquet et al. [l3] havemade piezoelectricphaseused in the composite. Although the
Desilets er
complete theoretical and experimental analyses for obtain- design and fabrication of different types of piezoelectric
jng Optimum acoustic matching.Depending On the 'pe- compositematerialsforhydrophoneapplications
at low
cific requirementsofsensitivity,bandwidth,andimpulse
frequencies (less than 40 KHz) hasbeen
in several
papers [1]-[8], forcompletenessabriefsummary
is preresponse,the Optimum acousticimpedance of the
ing layer for a PZT ceramic disc operating into a water sented here in the next section.

equivalent load varies from four to seven. Since no readily

availablematerial has its acousticimpedancenearthis

111. COMPOSITEPIEZOELECTRICTRANSDUCERS
value, a matching layer mustbepreparedsynthetically,
which isnot always easy. However,in practice,such maPZTceramic, which is commonlyusedasapiezoelectransducer material, suffersfromseveraldisadvanterial is prepared by mixing suitablemetallicpowdersuchtric
as aluminum with some epoxy resin [lo], 1131, [ 141. The tages when used as a hydrostatic pressure sensor. The hycomposite matching layer configurations have relatively drostatic piezoelectric coefficient dh ( =d33 + 2d3,)of PZT
large attenuation coefficientandadd
to further loss of is low due to opposite signs of piezoelectric strain coeffiacoustic
cients d33and magnitudes
d 3 though
,the
,even
of both d33
energy.
11. PIEZOELECTRIC
TRANSDUCER
MATERIAL
PARAMETERS

ON SONICS AND ULTRASONICS, VOL. SU-32, NO. 4, JULY 1985

GURURAJA el ai. : PIEZOELECTRIC COMPOSITE MATERIALS-PART

483

TABLE I

DIELECTRIC
AND PIEZOELECTRICPROPERTIES

OF

PZT-POLYMER
COMPOSITES

Density
(gm/cc)

Dielectric
Constant
K

Single Phase PZT

7.6

1800

2.5

40

Coral replamine-PZT
composite
PZT-Spurns epoxy (BURPS)
PZT-silicone rubber (BURPS)

3.3

50

4.5
4.0

620
450

140
20
45

36
110
180

5040
,2200
8100

1-3

PZT rods-Spurns epoxy

PZT rods-polyurethane

1.4
1.4

54
40

56
56

27
20

1536
1100

1-3-0

PZT rods-Spurns epoxyglass spheres

PZT rods-foamed
polyurethane

1.3

78

60

41

2460

0.9

41

210

73

14600

PbTi0,-chloroprene rubber
Bi,03 modified PbTi03chloroprenerubber

40

100

35

3500

40

28

10

280

3-1

Perfodted 3-1 composite

2.6

650

30

170

5100

3-2

Perforated 3-2 composite

2.5

375

60

200

12000

Description
of Composite

3-3

0-3

and d 3 , are large. The hydrostaticvoltagecoefficient gh

(dhle&) is also small because of its high relative permittivity k . ea is the permittivity of free space.
To improve themagnitude of dh and gh, anumber of
different diphasic composites using PZT and passive polymers have been fabricated. Itwasshownthatthephase
connectivity, defined as the manner in which the individual phases areinterconnected, is the key feature in designing the composite materials. The electrical
flux pattern and mechanical stress distribution together with the
resulting physical andpiezoelectricpropertiesdepend
strongly on the phase connectivity. In a diphasic composite, there are tenpossible connectivity patterns designated
3-3. A
as 0-0, 1-0, 2-0, 3-0, 1-1,2-1, 3-1,2-2,3-2,and
1-3 connectivity pattern, for example, consists of parallel
PZT rods embeddedin a three-dimension continuouspolymer matrix [3]. In the notation used here, the piezoelectric phase appears first.
Dielectric and piezoelectric properties of different types
of PZT-polymer composite materials designed for the hydrophone applications are summarized in Table I. Hydrostatic d and g coefficients were measured by a substitution
method [7]. As listed in the table, a 3-3 composite comprised of PZT and polymer phases continuously self-connected in three dimensions to give two interlocking skeletons in intimate contact with one another [ 1, 61. A 1-3-0
compositeisa
1-3 PZTrod-polymercomposite
with a
third phase such as hollow glass spheres or pores not in
contact with each other [20]. In composite with 0-3 connectivity, piezoelectric ceramic particles (PZT, pure, and
modified lead titanate) 1-10 p in size are loaded in a polymer matrix [g]. Composites of PZT and polymer with 3-1
and 3-2 connectivity patterns have beenfabricated
by
drilling holes in sintered PZT blocks and filling the holes
with epoxy [7]. Indesigningallthesecompositemate-

g,

VmN-')

dh @CP.-')

dhgh( 1 0 - % * ~ - ' )

Reference

100

rials, the primary goal was to decouple the transverse d3,

coefficient and lower the dielectric permittivity to augment
the hydrostatic d and g coefficients. The product of dh and
g h listed in Table I was used as the figure of merit for the
hydrophone applications.
From Table I, it is evident that the hydrostatic pressure
sensitivity of some of the composites was orders of magnitude larger than the corresponding values of the piezoelectric phase used in the composites. The density of the
composite could be adjusted between the densities of the
component phases. Composites with porous polyurethane
phase were prepared to have positive buoyancy in water.
By choosingsoftpolymerssuchassiliconerubberand
polyurethane, composites were made flexible.
PZT polymer composites have several advantages over
conventional piezoceramic materials for ultrasonic applications. In composites most of the PZT (70-90 percent)
is replaced by a low-density polymer, and hence resultsin
a better acoustic impedance matching to the human body.
The composites also have a relatively low mechanical Q
(3-10) [21]. These properties of thecompositesappear
Alwell suited for thebroadbandwidthapplications.
though PZT has a high d33coefficient, the voltage coefficient g33 is low because of the high dielectric constant.
Thus, PZT is a poor receiver of ultrasound. The concept
of compositetransducerallowsdesigningacomposite
structure to enhance the g33 coefficient and improve the
sensitivity in the receiving mode. If the receiving voltage
sensitivity of the transducer is enhanced, the human body
can be interrogated at lower ultrasonic energy andthe biological effects of ultrasound, if any, can be minimized.
The following criteria were used here
to select composite materialof an appropriateconnectivity for a systematic
investigation intendedtowards
ultrasonic transducers.
Transducers for ultrasonic imaging applications are oper-

IEEE TRANSACTIONS ON SONICS AND ULTRASONICS, VOL. SU-32,N O.

484

4, JULY 1985

TABLE I1
PHYSICAL, DIELECTRIC,
A N D PIEZOELECTRIC PROPERTIES
OF PZT ROD-POLYMER
COMPOSITES
Rod
Diameter
(mm)

Volume percent
PZT
~~

5
10
0.45

10
0.28

1.78
1.26
0.90

1.41
1.64
320 2.29

0.45
0.28

0.73
0.78
0.55

3 .OS
1.62
339 2.36

0.45
0.45

20

Single Phase
PZT

Density
p(gm/cc)

Dielectric
Permittivity K

d~ Coefficient
( pUN)

20

30

Periodicity
d (mm)

400
7.6

375

19

155

I09

I94

238
3 80
94
217

183

1400

ated in a half-wavelengththicknessmodeformaximum
sensitivity. Thus, the major requirement of a composite
was to have a well defined thickness resonance with a reasonably good piezoelectric coupling coefficient and a low
Q. Pulse-echoamplitudeandbandwidthofair-backed
compositetransducersoperating with water load in the
low megahertz frequency range determined by the toneburst pulse-echo method [22] were used as additional parameters.
Amongallthedifferentcomposites,onlythose
with
PZT rods embedded in Spurrs epoxy' matrix with regular
periodicity (1-3 connectivity) appeared to be very promFig. 1. Typical 1-3 composites with PZT rods embedded in Spurrs epoxy.
isingfor the ultrasonic transducer application[21]. The
thicknesscouplingcoefficientfor
thesecomposites was
ers [3]. In brief, PZT-501A2 rods were extruded using an
determined to be about 60-70 percent. This value comorganic binder and then firedat 1285C for one half hour.
pares favorably with the thickness coupling coefficient of
Fired rods were aligned using an array
of appropriately
49 percent for the PZT-5 disc which is used
in commercial
spaced holes drilled in a pair of brass discs boltedparallel
transducers. Mechanical Q of the composites was low (3to eachother. The array of PZT rods wasthencastin
10) compared to that of PZT-5 (80). The pulse-echo amSpurrs epoxy matrix.The resultingcomposite was cut
plitude of these composite transducers was large (from 2perpendicular to the length of the rod and lapped to the
3 V to 10 V input signal) and comparable to those of comrequiredthickness.Thecomposite
wasthenelectroded
merciallyavailable transducers.The 1-3 composites had
with silver paint on the major surfaces and poled convenadditional advantages. There aremany variables including
tionally by applying an electric field of 20 kV/cm for five
PZT rod diameter andvolumepercent
of the PZT and
minutes in a heated oil bath at80C. Different composites
polymer phases, which can be vaned relatively easily to
werepreparedwithapproximatelyfive,
ten, 20, and 30
optimizetheproperties
for particular application. The
volume percent PZT using 0.45 mm rods and with ten and
continuous poling technique [23] can be used to polarize
20 volume percent PZT using 0.28 mm rods.
Table I1 gives
long PZT rods, which can then be arranged in a polymer
the periodicity (distance between the centersof neighbormatrix in such a way to obtain a desired radiation pattern.
ing PZT rods) of PZT rods in each of these composites.
Since the composite structure resembles a transducer arof
ray, the knowledge of the high-frequency dynamic behav- The composite samples were shaped as circular discs
diameter 19 mm and varying thickness from 0.5 mm to 5
ior of these composites might be helpful in resolving some
mm. A picture of typical compositesamplesisgivenin
of the problems encountered in array applications, such as
Fig.1.
minimizing the acoustic cross talk between the adjacent
Results on the average properties including density, pielements.Because of theseadvantages, 1-3 composites
ezoelectric d33 coefficient,anddielectricpermittivity
of
were chosen for an extensive investigation aimed at high
PZT rod-polymer composites have been reported earlier
frequency transducer applications.
by Klicker et al. [3]. Data for the composite samples used
A detailed procedure for fabricatingcomposites with
1-3 connectivity has been reportedby Klicker and co-work- in this work are listed in Table I1 along with the properties
of single phase PZT-5O1A for comparison.
'Trademark of Polysciences, Warrington, PA.

'Trademark of Ultrasonic Powders, Inc., South Plainfield, NJ

GURURAJA er al.: PIEZOELECTRICCOMPOSITEMATERIALS-PART

485

_--.

IV. BULKMECHANICALPROPERTIES
OF
SPURRSEPOXY
Radial Mode
in
Spurrs epoxy is the piezoelectricallyinactivephase
thecompositematerialconsideredhere.
However, its
function in thecompositetransducer
is quitecritical.
When the composite is used as an ultrasonic transmitter,
the epoxy has to effectively couple the ultrasonic energy
fromahigh-acoustic-impedance PZT to a low-acousticimpedance load. Similarly, in the receiving mode, the ul(b)
trasonic energy incident on the composite shouldbe transferred effectively to the piezoelectrically active PZT. The Fig. 2. Resonance modes in a disc-shapedpiezoelectricmaterial.(a) Radial mode. (b) Thickness mode.
importantparameters of thepolymerthatinfluencethe
performance of the composite as an ultrasonic transducer
are its acoustic impedance, longitudinal and shear wave A . Planar-ModeResonance
velocities,and attenuation coefficients and their depenThe planar mode (alsocalledradial)vibrationinthin
dence on frequency. These properties of the polymer dedisc, as illustrated in Figure 2(a), involves simultaneous
termine the interaction among the neighboring PZT rods
mechanical motion in the 1 and 2 directions driven by the
andhencethehigh-frequencydynamicbehavior
of the
(poling
direction).
electric field in the 3 direction
composite.Measurementtechniques
used to determine
Since the diameter was the largest dimension (19 mm) in
these parameters in Spurrs epoxy are described elsewhere
all the composite samples, the planar mode was thelowest
[24]. Only the results which are useful for analysis of the
frequency resonance mode.
vibration behavior of the composites are presented here.
TheIRE(nowIEEE)standardon
piezoelectricmeaThe longitudinal andtransverse
wave velocities in
surements
[26]
was
used
to
determine
the
electromechanSpurrs epoxy were determined to be 2060 m/s and 1150
ical
coefficients
and
elastic
constants.
The
most important
m/s, respectively. The acoustic impedance of the Spurrs
parameters
for
calculating
these
constants
are the series
epoxy was calculated to be 2.27 X lo6 kg/m2-s. The atand parallel resonance frequenciesf, and&, the geometric
tenuation of the transversewave in Spurrs epoxywas measured to be 6 dB/cm at 0.5 MHz. The attenuation coeffi- capacitance C,, and the impedance minimum lZ,l at rescient wasfound
to increase with thesquare
of the onance. A spectrumanalyzer(HP 3585A), which provides a circuitsimilar to the transmissionnetworkrecfrequency. The glass transition temperature Tg is another
ommended by the IRE standard, was used in measuring
characteristic parameter for polymeric material. The
glass
frequencies fm and f, corresponding to minimumand maxtransition temperature is defined as the temperaturebelow
imumimpedance. In thepresence of highmechanical
which the material behaves like a glass and above which
losses, fm andf, are quitedifferent fromfs and&. TheIRE
thematerial is rubberlike andsoft [25]. In the present
Standard recommends the following approximation to calwork Tg and its dependence on frequency was determined
culate Af = fp - fs from the measured value off, and f,
by measuring the dielectric loss tangent of Spurrs epoxy
as a function of temperature and frequency. At the glass L261
transition temperature,the loss tangentgoesthrougha
peakandcanbeeasilyidentified.ForSpurrsepoxy,
TR
was found tobe a functionof frequency varying from 70C
at 1 0 0 Hz to 121C at 1 MHz with a shift of 12-15C per
where the figure of merit M is given by
decade of frequency.
P

V. RESONANCEMODES

Resonance modes in composite materials, as in single

phase materials, are defined by the geometry and dimension of the specimen. In a circularly shaped 1-3 composite
material,theresonancemodesobservedare
theplanar
mode, the thickness mode, and a number of lateral modes
related to the regular periodicity of the PZT rods in the
composite.Characterization of thedifferentresonance
modes in 1-3 PZT rod-polymercompositewillbedescribed in thefollowingsubsections.Knowledge
of the
resonant vibration behavior of the composites in air is important and helpful in the evaluation of the composites as
anultrasonic transducer operating intowaterequivalent
load.

1
2rfJ1Co

M =

2nfm(Co + CI)lzml.

(2)

The C,, C , are the capacitance of the parallel and series

branch, and R , is the resistance of the series branch of the
equivalent circuit of the piezoelectric material near resonance. The sum (C, C , ) is the static capacitance, which
was measured at a frequency well below fundamental resonance, for instance at 1 KHz. These approximations are
valid for resonances with M > 3.
The planar coupling coefficient kp was calculated from
the expression given by [27]

l + a

IEEE TRANSACTIONS ON SONICS AND ULTRASONICS. VOL. SU-32. NO. 4. JULY 1985

486

TABLE 111
ELECTROMECHANICAL
PROPERTIES OF COMPOSITES FROM THE RADIALMODERESONANCE

Volume percent PZT

(Rod Diameter)

Thickness

5
(0.45 mm)
l0
(0.45 mm)
20
(0.45 mm)
30
(0.45 mm)
10
(0.28 mm)
20
(0.28 mm)

Thin
Thick
Thin
Thick
Thin
Thick
Thin
Thick
Thin
Thick
Thin
Thick

k,

40
17.0
21.7
36
16.6
1.1
42
1.5 to 2 . 02 3 . 0
1.2
23
21.0
1.5 to 2.3
26.5
34
1.1
20.0
3626.72.0
1.1
18.8
23.11.5
1.1
22.0
25.5
1.7

c*

N(Hzm)
(Frequency
Constants)

Longitudinal
Velocity
(mis)
1483

32

1064
1120
1011
l080
955
1043
950

36
35
32
35

1032
1097
970
1057

0.97

- 1.4

34

1561
1409
1506
l33 1
1454
1324
1458
1439
1529
1352
1474

1046

Thick = 0 . 6 mm, and thin > 3 mm.

where U is the Poissons ratio of the material. Since the

Poissons ratio of both PZT and epoxy are approximately
0.3, the Poissons ratio of the composite was assumed to
be 0.3. The value of R1 = 2.05 for U = 0.3. The f, was
approximated tofm.
The mechanical quality factor Q can be evaluated from
the relation [26]

For M > 3, (4) was simplified to

fraction of PZT. Thick composites have a higher kp comparedto thincomposites of thesame volume fraction.
Lower values of kp in thin samples indicate that the mePZT
chanicalenergy is not efficiently coupledfromthe
rods tothe polymer and from the polymer back
to the PZT
rods. Thecouplingbecomesmore
efficientforthicker
samplesindicating a greater homogeneity. A similarexplanation can be given for the difference in the frequency
constant between the thick and thin samples.
In the last column of Table 111, the longitudinal velocity
C, perpendiculartothelength
of therodislisted.
The
value of C,was calculated using the relation [27]

fs = 2.05
[(E,lp(l
27ra
The calculated values of kp and Q from (3) and ( 5 ) for
several composites are summarized in Table 111. The frequency constant Np for the planar mode (product of resonancefrequencyandthediameter)isalsolisted
in the
table. The data are divided into two groups depending on
the thickness of the sample. The first group investigated
of about 0.6
consisted of thinsampleswithathickness
mm.Thesecondgroup
was comprised of allsamples
thicker than 3 mm and was classifiedas thick samples. The
properties remained virtually constant in the two extreme
limits. The data listed in the table are an average of the
measurements of at least five samples.
As it emerges from Table 111, the figure of merit M for
the radial-mode resonance of all the composites was between one and two. As mentioned before, (3) and ( 5 ) used
for calculating kp and Q are not fully valid for values of
M < 3. Subsequently, the valuesof kpand Q listed in Table
111 are far from being exact. However, since the value of
M is approximately the same for all the samples, the uncertainty in assessing k,,, and Q is of the sameorder. Hence
the results can be used for a relative comparison of samples having different volume fraction PZT.
Table I11 indicates that the planar coupling
coefficient k,,
of the composites increases with increase in the volume

- C?)]*

where fs is the series resonance frequency of the planar

resonance. The modulus E, is the elastic modulus of the
composite perpendicularto the PZT rod axis, p is the density, and a is the radius. In ( 6 ) (E,lp)* corresponds to the
longitudinal velocity C, perpendicular to the rod axis. In
calculating C, from (6), fs was measured and D was assumed to be 0.3 as already mentioned.
An attempt was made to theoretically estimate the longitudinal velocity C, by obtaininganexpression for the
elastic modulus E, perpendicular to the rod axis as a function of volume fraction of PZT. Knowing E,. and the density p of the composite (Table 11), C, was calculated from
the expression C, = (E,/fi)2.
The effective modulusE, in the transverse direction can
[14],
becalculated by theReussconstantstressmodel
which assumesthat the two phases experience equal stress.
This assumption seems appropriate for a combination of
hard filler in a soft matrix, because the matrix should deformmorethanthefillerandthestressinsuchasolid
shouldbefairlyconstant.Theeffectivemodulus
by the
Reuss model is given by

_1 -- _U1 + -2
Et

E,

E2

(7)

GURURAJA er al. : PIEZOELECTRIC COMPOSITE MATERIALS-PART

2300-

l300

I
IO

20

30

l
40

50

Volume Percent PZT

Fig. 3. Longitudinal velocity perpendicular to the PZT rod axis as a function of volume percent PZT In composites.

487

ness decreases, approaching the Q of single-phase PZT

501A (80) [28].
The planar-mode resonance was studied as a functionof
temperature up to 100C. It was observed that the planar
couplingcoefficientremainedvirtuallyconstant
with increase in temperature. However, the amplitude of the resonancespectrumdecreasedgradually
with temperature
andwassuppressed
below 0.1 dB at 100C. Measurementscould not be made on the radial-moderesonance
above 100C. Itis interesting to note thatthis temperature
corresponds to the glass transition temperature of Spurrs
epoxy (Section IV) at the resonance frequencyf, (50 KHz).
It is conceivable that the suppression of the radial mode
above 100C is due to the very lossy nature of the Spurrs

IEEE TRANSACTIONS ON SONICS AND ULTRASONICS. VOL. SU-32, NO. 4, JULY 1985

488

TABLE IV
COMPARISON
OF IRE STANDARD METHOD
A N D VECTOR IMPEDANCE METHOD
FOR CALCULATING
ELECTROMECHANICAL COUPLING COEFFICIENTS'

Standard
Sample

Thickness
in mm

200
20 1
202
203
204
205

0.66
2.54
3.05
3.95
4.60
5.15

fm

IRE

- f,

Af= fp

-L

(KHz)

(KHz)

2.320
1.820
1.998
2.225
2.386
2.842

1.0
1.7

1.041
1.168
1.364
1.585
1.760
2.172

1.9
2.0
2.2
2.4

Vector Impedance
Method

Method

Kp (%)

24
22.6
39
22.8 0.929
37
24.7
26.41.48035
35
21.8
30.81.79230

fp -fs

kp (%)

0.350 72 12.9
48 20.3
1.184 43 22.9
25.6
27.1
1.711
27.8

39
37
37

'In a 20-percent PZT composite, with rod diameter = 0.45 mm.

thick samples while the mechanical Q showed the reverse

trend. The couplingcoefficient and the frequency constant
of thin sampleswereclosetothat
of single phase PZT
= 3700 Hz m for PZT
501A (k33 = 70 percent,and
501A) and were independentof the volume percentof PZT.
In thick thick samples both k, and NP were proportional to
the volume percent of PZT.
The observed behavior of the composites in thicknessmode resonance can be explained as follows. At the resonance frequency, PZT rods vibrate with a large amplitude. The vibrating rods act as a source of complex trans5 0 0 0 " " " " 50
" " " '
l00
I50
verse waves intheepoxypropagatingin
adirection
Temperature, ' C
perpendicular to the axis of PZT rods. These transverse
Fig. 4. Effectof temperature on the Frequency constant of radial mode res- waves generated in theepoxy interact with the array of
onance of 1-3 composites.
PZT rods. In the first approximation the interaction is assumed to involve only nearest neighbors. Effectiveness of
the interaction depends on the amplitude and phase of the
and
transverse wave relative to that of the PZT rods. For thin
Np = f p t
(10) samples resonating around 2.25 MHz, the wavelength of
where r is the thickness, andthemeaning
of theother the transverse wave is approximately 0.5 mm (taking the
terms in (8)-(IO) is given in Section V-A. The IRE Stan- transverse wave velocity to be 1150 m/s from Section IV).
dard method was used to calculate these properties of the It is also known that the Spurrs epoxy is extremely lossy
compositecorrespondingtothethickness-mode
reso- for transverse waves at this frequency. This results in only
a weak interaction between the vibration of neighboring
nance.
Theexperimentalresultsaresummarizedin
Table V. PZT rods. The relatively low Q of thin samples indicates
The data are again divided into two groups: thin samples that the rod vibrations are partly damped by the surrounding polymer.
withthickness around 0.6 mmandthicksampleswith
0.5 MHz and the
The thicksampleresonatearound
thickness above 3 mm. As it can be seenin the table, with
the exception of the five percent PZT composites, thefig- wavelength of transverse waves in the epoxy is approxiof all mately 2.3 mm. The wavelength is 1.5 to three times larger
ure of merit M for the thickness mode resonance
than the separation distance between the PZT rods in the
othersampleswasabovethree.Hence,theapproximationsrecommended by the IRE Standard to calculate k, composites of different volume percent PZT. Since the atand Q are valid. Indeed the values of the kt and Q deter- tenuation of the transverse waves in the epoxy is relatively
mined by the vector impedance method
on a selected num- low at 0.5 MHz ( - 6 dB/cm), the amplitude of the transber of samples were within two percent of the values cal- verse wave must be significant to cause a strong interaction between the rods through the epoxy medium. The
low
culated by the IRE standard method.
coupling coefficient in thick samples is probably a result
Datalisted inTable Vindicate thatforthethickness
mode resonance of thin samples (r = 0.6 mm), kt was 60- of the phase cancellation effect betweenthe vibrating rods
70 percent,and Q was three to ten.For thicksamples (due to theelectrical excitation) and the mechanical inter( t > 3 mm), k, was approximately 50 percent,and Q action between the rods through the epoxy. The low mewas 20 to 30. In general, thin composites had higher coupl-chanical loss in the epoxyis reflected in the high Q of the
ing coefficient k, and frequency constant
compared to composite.

GURURAJA er al.: PIEZOELECTRIC COMPOSITE MATERIALS-PART

489

TABLE V
THICKNESS-MODE RESONANCE
I N COMPOSITES
Longitudinal
Frequency
Velocity
Constant
Nf (Hz m)

(mis)

Volume Percent PZT

(Rod Diameter in mm)

Thickness
(mm)

Figure of
Merit M

k,
percent

5
(0.45)

0.6
4 to 6

2.4

57.1
27.5

31

1834
1327

3668
2654

10
(0.45)

0.6
3.6

4-6
4-6

68.0
40.0

7
23

1820
1492

3640
2984

20
(0.45)

0.6
3 to 5

5-7
5-7

68.9
9
1603 18
55.1

1853

3706
3203

2.2

c,

2NP

30

(0.45)

2.4

12

61.1

26

1687

3374

10
(0.28)

0.6

3-5
3.6

62.6
40.6

4
22

1956
1403

2912
2806

20

0.6
2.9 to 3.4

5-7

72.0

5-7

50.0

8
25

1874
1555

3748
3110

1.9

(0.28)

C. Lateral-ModeResonances
In additiontotheradial-andthickness-moderesonancesandtheirovertones,otherresonanceswereobserved in the frequency range 0-2 MHz. The frequencies
of these resonances were independentof sample thickness,
but were related to the lateral periodicityof the PZT rods
in the composite. Absolute value of electrical admittance
was measured on samples as a function
of frequency in the
ambience of air and waterusing a spectrum analyzer (HP
3585A). Here samples of different thickness and volume
2000

0
IO
20
30
40
50
fractionwereexaminedtoidentifythedifferentresoVolume Percent PZT
nances observed in the 0 to 2 MHz frequency range. Three
Fig. 5. Longitudinal velocity along the PZT rod axis as a function of vol- major resonances of interest are designated as fi, AI, and
ume percent PZT in composites.
A2 as categorized in Table VI.
The resonance corresponding to fi was inversely proportional to the thickness of the sample. This resonance
The longitudinal velocity Cl along the length of the rod
was heavily damped when the resonator was immersed in
canbetheoretically estimated by calculatingtheelastic
water. Such behavior suggests that resonance
at fi is the
modulus EI of the composite parallel to the PZT rods. At
longitudinal thickness vibration. The longitudinal velocity
wavelength is much
frequencieswherethetransverse
of sound, calculated usingtheformula Cl = 2tf, agrees
larger than the separation distance between the rods, the
very well withthetheoreticallyestimatedvalueforthe
Voigt constant strain model [l41 can be applied. The comcomposites of different volume fractions, as discussed in
posite modulus using the Voigt average is written as
detail in the previous section.
The resonances at&, and&? were
very similar in nature.
El = u , E ~+ u ? E ~
(11)
For a certain volume fraction, these resonances
always ocwhere El and E2 arethemoduli,and
u I and u2 a rethe curred at specified frequencies regardless of the thickness
volume fractions of the two phases (PZT and epoxy). For of the sample. As can be seen in Table VI, increase in the
PZT rods witha diameter that is small compared to length,volume fraction results in closer spacings of rods (reducA2 move to
themodulusalongthelength
is given by E = l/$: = tion of the unit cell) and the resonances and
10.5 x 10" N/m2(the Young's modulus of the rod at con- higher frequencies.
The behavior of the composites could be explained by
stant electric displacement). The longitudinal velocity Cl
simpleaveragingschemessuchastheconstantstrain
was calculated from the relation Cl = (El/p)"2.
The experimentally determined longitudinal velocity Cl, model and the constant stress model forwavelengths large
expressed as twice the frequency constant @, is given in or small compared to the lattice periodicity (Section V-A
the last column of Table V. This velocity is compared with and V-B). In the vicinity of the resonances of A, and A2,
the calculated values in Fig. 5. The excellent agreement the wavelength of the transverse shearwave is comparable
to the unit cell dimensions of the array and the analysis is
betweentheexperimentalandtheoreticalvaluesfurther
morecomplicated.
A Brillouintheory of elastic wave
supports the equal strain model for calculating the effecin a twodimensionallyperiodiclatticecortivemodulusofthecompositealongtherodaxis.propagation
40001

IEEE TRANSACTIONS ON SONICS AND ULTRASONICS, VOL. SU-32, NO. 4. JULY 1985

490

TABLE V I

R E S O N , ~ NMODES
T
IK COMPOSITE M A T E R I A L S

Resonance Frequency
.
. in
KHz
Percent
Volume

10

20

20

30

30

Rod

Thickness

Periodicity

c, = dfil

(1

= -r

v2

51
52
53
54

0.45
0.45
0.45
0.45

2.59
4.00
4.80
5 .90

604
1.58
1.76
1.76
1.76
437

450
3I2
262
222

483
449

101

I02
103

0.45
0.45
0.45

1.93
3.64
5. 15

I .27
1.27
1.27

604
410
262

804
662
644

1096
1006

2294
2984
2698

102 1
840
72 1

984
903

I
202
203
204
205

0.45
0.45
0.45
0.45
0.45

2.54
3 .OS
3.95
4.60
5.15

0.90
0.90
0.90
0.90
0.90

540
455
348
296
270

894
830
828
886
825

2742
2774
2748
2722
2780

804
747
745
792
742

I
302

0 45
0.45

2.44
5.15

0.73
0.73

569.5
276.5

1125
1188

2778
2848

82 1
867

Scanned obng rods -

(a)

(b)

Fig. 6. Unit cell geometriesandlaserscanpaths.

(a) Standing waves at the
second stopband, along thc unit cell edge. (b) Standing haves at the second
stopband.
alonr
the
unit cell diagonals.

respondingtothe PZT rod-polymercomposite was developed by Auld er ul. in 1311. Forlaterallypropagating

shear waveswith polarizationalongtherods,the
lowest
stopbandfrequencieswerecalculated.Theexistence
of
stopbands was related to Bragg-scattering from planes of
rods having various orientations in the square lattice. The
essential phenomena can be explainedby reference to Fig.
6. Suppose in Fig. 6(a) that a z-polarized, x-propagating
shear wave is incidect on the grating. At a frequency for
which the rod spacing d is one-half the shear wavelength
(X,), constructive reflection (Bragg-scattering)occurs
fromverticalplanes
of rods,andresonanceoccursbetween adjacent vertical planes. The x-propagating wave is
also scattered from planes of rods at 4.5" to the x and y
axes; butin this case, Bragg conditionsare not satisfied
A stopband exists at d =
and resonance does not occur.
h,/2 for an x-propagating wave, but there is n o resonant
scattering into y-propagating waves.A t the stopband edge,
the solution is a standing wave along x. From the square
symmetry of the lattice, a similar standing wave solution
along y also exists. This behavior corresponds to the first
stopband of the lattice.

786

691

c, = 2tfi

730

8782330
2496
25 l4
908261 8

954
853
790
769

859

When the rod spacingd is one'shear wavelength, a completely different

standing
wave pattern
results,
illusas
trated in Fig. 6(a). In this case a z-polarized, x-propagating wave is again Bragg-scattered with a phase shift
of 27r
from one vertical plane of rods to
an adjacent plane. However, it is easily verified that the wave is also Bragg-scattered at the same frequency by the 45" planes of rods, so
that resonant scattering of the x-propagating wave into a
y-propagating wave occurs.Thesey-propagating
waves
also experience resonance scattering of the same type. As
a result, the solution is a two dimensional resonant standing wave pattern as shown in Fig. 6(a). Here the rods all
vibrate in phase
and
correspond
to the
second
stopband
of
the square lattice.
For-a z-polarized z-propagating wave incident on the lattice, no Bragg scattering can occur because the structure
is uniform along z . However, a spatial modulation of the
phase fronts is produced by the difference in the properties
of the rodsandmatrix.Consequently,spaceharmonics
are generated but none are resonant. Similar conclusions
areobtained forz-propagating wavesof otherpolarizations.
In the following section, detailed laser probe measurements of the actual displacement on the composite to analyzethenatureofvibration
atthesefrequenciesarereported.
VI. LASERPROBEMEASUREMENTS
The measurements of actual displacement on the composite samples were performed using the laser heterodyne
technique [31]. The samples were fine polished (+ 1 pm)
andelectroded with highly reflecting gold electrodes for
the laserprobemeasurements.
Relativeamplitudeand
phase of the ultrasonic displacement was measured along
two scan paths: along a row of PZT rods and on the epoxy
surface in betweentwoadjacentrows
of PZT rods as

GURURAJA
M.4TERIALS-PART
er al.:
COMPOSITE
PIEZOELECTRIC

49 1

40

predictedtobezero;andthishasbeen
verified experimentally. The velocity C,, of transverse waves, as tabulated in the last two columns of Table VI is lower than the
measured velocity of 1150 m/s (Section V). Thisispossibly due to the mass loading by the PZT rods.

C . Vibration Pattern at Frequency f(

400

IO00
Frequency ( K H z )

I600

Fig. 7. Measuredelectricalinputimpedance of compositetransducer #l01

(tenpercentPZT
in Spurrsepoxy,PZT
rod diameters = 0.45 mm.
period = 1.27 mm, thickness = 1.93
mm).
Transducer
diameter
=
19.0 mm.

Fig. 10 shows plots of complex vibration pattern at the

the to
simple longitudinal
frequency fr, which corresponds
resonance
along
z. It
thickness
ofhomogeneous
the
plate
was noted earlier that no resonant space harmonic exist in
this case, so that weaker standing-wave patterns are to be
plane.
is clearly visible inthe figexpected in the

ure. The small phase variations are attributed

to the damping in the composite which was not considered in the theshown in Fig. 6. Fig. 7 giveselectricalinputimpedanceory
outlined above ,311.
data taken On One Of these
lol in
As is clearly Seen in Fig. 10, the amplitude of vibration
showing the existence Of three strongresonances.
on the epoxy at f i is about 5 to 8 dB larger than on the
show laser Scan plots Of vibration amp1itude
Figs'
PZT rods indicating an effective couplingof the ultrasonic
and phase taken at frequencies f t l ,
andfi, respec- energy from PZT to epoxy. The PZT rods vibrating due
tively, for this sample.
to electrical excitation act as the sourceof transverse waves

in the epoxy travelling in a direction perpendicular to the

axis of the PZT rods. The transverse waves produced inThe plot in Fig. 8 shows the complex vibration pattern
teract with neighboring rods to produce the resultant viatfrequency 5, correspondingtothesecondstopband
bration pattern. The vibration pattern depends very much
along the unit cell, (Fig. 6(a)), where the rods all vibrate
on the wavelengthof the transverse waves in relation to
in phase. (Vibrations at the first stopband discussed earliertheperiodicity of thelattice. As the wavelengthof the
have a 180" phase shift from one plane of the rods to the transverse waves becomes larger compared to the lattice
next and cannot be excited electricallyin a uniformly elec- periodicity,ultrasonicdisplacementonthecomposite
is
trodedsample.)Superposition
of the x and y standing expected to be more uniform.
waves illustrated shows that the z-displacement has a posFig. 10-12 compare the ultrasonic displacements on the
itive maximum at a , negative maximum at b , and zero at surface of composites 101, 102, and 103(Table VI) resoc . Because of symmetry,thesestatements
applyatall
nating in the longitudinal thickness vibration at frequenequivalent points in the lattice. In the scans of Fig. 8, the cies 620, 414, and 270 KHz, respectively. All the samples
maxima of the standing wave pattern at a and b arel80" have ten-volume-percent PZTwith a periodicityof the latout of phase, as predicted. The predicted zeros
at c in Fig. ticeequal to 1.27 mm. The only variable in the samples
8(a) and at the equivalent point in Fig. 8(b) are less sat- are the wavelengthsof transverse wave at thethickness
isfactorilyreproduced, possibly duetoimperfections
in resonance frequency, which are respectively 1.9, 2.7, and
the lattice and inaccuracy in alignment of the scan. The 4.3 mm for the above three samples (calculated using the
velocity of the transverse waves, c, = dfil is tabulated in transverse wave velocity in Spurrs epoxy of1150 m/s).
Table VI.
The difference in amplitude of vibration between the PZT
andepoxyobserved in sample 101 is greatlyreduced in
B. Vibration Pattern at Frequencyft2
sample 102. This is directly attributed to the longer waveThe plot in Fig. 9 shows the complex vibration pattern length of transverse waves in sample 102. For sample 103
at frequencyj? corresponding to the stopbandin Fig. 6(b). the transverse wavelength is more than three times the peThis is another of the higher stopbands for which the rods riodicity,and there is no noticeabledifference in amplivibrate in phase and can be excitedby uniform electrodes. tude between the PZT and epoxy. The composite vibrates
uniformly across the surface with almost equal strain on
Atf2, standing transverse waves along the unit cell diag180" thePZTandepoxy.Smallphasevariationsseenacross
onals are superposed to give maximum displacements
103. This is
out of phase at a and c , and maximum in phase displace- sampe 101 are not measurableforsample
ments at a and b. These features are clearly observable in probably because of the reduced attenuation in the epoxy
the scans of Fig. 9, except that the maximum ata is smaller atlowerfrequencies(SectionIV).Thevibrationpattern
than that predicted from the assumed standing
wave pat- observed in sample 103 is in accordance with the assumptern. This is attributedtothepresence
of higher space tion in Section V-B of the constant strain model used to
harmonicssuchthatthestanding
wave patternsare not calculate the effective modulus along the fiber axis, when
transverse wavelengths are large compared to periodicity.
simple sinusoids. It can also be seen from Fig. 6(b) that
the displacement at point d , midway between a and b , is This uniform ultrasonic displacement observed on the sur-

A . Vibration Pattern at Frequency

f,,

IEEE TRANSACTIONS ON SONICS AND ULTRASONICS, VOL.SU-32,NO.

492

4, JULY 1985

e
I
(mm)

-Position

(mm)
-60-

Position

ON THE
VII. EFFECTOF TEMPERATURE
face of the composite is a clear indicationof efficient couRESONANCE
M
ODES
pling of the mechanical energy from PZT to the epoxy.
Sincethe acousticimpedance of the epoxyis relatively
In this section the effect of temperature on the electroclose to that of the human body, an effective coupling of
acoustic energy from the composite transducer to the hu- mechanical properties of the composite is discussed. It is
clear from the preceding sections that the high-frequency
man body is ensured.

GURURAJA et a l . : PIEZOELECTRICCOMPOSITEMATERIALS-PART

493

"
'c

'Or

15

Fig. IO. Laser scan measurements of relativeamplitudeandphase

of the
acoustic vibration atfr = 620.7 KHz (sample # 101). (a) Scan along rods.
(h) Scan between rods.

.I

dynamic behavior of 1-3 composite is strongly dependent

on the mechanical properties such as the velocity and attenuation of ultrasonic waves in the Spurrs epoxy. These
properties of Spurrs epoxy are very sensitive to temperature. The Spurrs epoxy is glassy at room temperature and
goes through the glass transition temperature to a rubber
likematerial in a temperatureinterval ofonly
about
100C.Intheglassystate,
it was also shown(Section
V-A) that the compliance of the epoxy increased by approximately 65 percent in the temperature range of 0 to
100C. Thus, a study of the temperature dependence of
the resonance modes will help in understanding the elec-

tromechanical propertiesof composites with different

polymer matrices, having a wide range of elastic properties.
In the experiment, the resonance behavior of composites freely suspended in air was studied
as a function of
temperature from 0 to 150C. The thickness-mode coupling coefficient k,, mechanical Q, and the frequency constant N f of the composite were determined at 2C intervals both in heating and cooling cycles. The heating and
cooling rates were maintained at approximately 2"C/min.
Fig. 13 shows plots of absolute value of admittance as
a function of frequency from 0.1 to 1.1 MHz for sample
0.45-mm rods,
103 (ten-percent PZT compositewith

494

IEEETRANSACTIONS

ON SONICS AND ULTRASONICS, VOL. SU-32. NO. 4, JULY 1985

Fig. 12. Laser-probemeasurements of relativeamplitudeandphase of the

acoustic vibration at pi = 270 KHz of sample #103. (a) Scan on rods. (b)
Scan in between rods:

of thickness
thickness 5.15 mm) at different temperatures up to 170C. while to note that above 130"C, the amplitude
Thebehavioris typical of allthick(thickness > 3 mm) mode was increased (Fig. 13(e)-(h)). The secondary lowcomposites.Accordingtotheanalyses
given in thepre- frequencyresonancesclearlyseenat25Cdisappearat
vious section, the three major resonances observed at room
approximately100C.Thisdisappearance
is againastemperature (Fig. 13(a)) were identified as follows.
cribed to the lossy nature of the epoxy around the glass
MHz was identified as ahalf
1) Resonanceat0.262
transition temperature.
wave resonance along the thickness of the composite and
The coupling coefficient k, and Q of the thickness-mode
resonance for sample 130 are plotted as a function of temis denoted byf,.
2) Resonance at 0.644 MHz was identified as the stand- perature in Figs. 14 and 15, respectively. The value of k,
(0.48)determined atroom temperature(25C)did
not
ing wave pattern in the epoxy arising from lateral periodicity of the PZT rods. This resonance is denoted byf,,. changesubstantiallyuptoapproximately110C.
In the
3) Resonance at 1.030 MHz was identified as the third temperature range from 100 to 130"C, k, exhibited a minimum. Above 130"C, after the transverse mode had disovertone of the thickness mode.
Weak secondaryresonances
below thefrequency of appeared, k, increasedsignificantly to about 60 percent.
As shown in Fig. 15, the mechanical Q at 25C was about
thickness-mode resonance were ascribed to the overtones
20. The Q remained constant to approximately 100C and
of the planar-mode resonance of the composite disc.
decreased to a value of five at temperatures above 130C.
The effect of temperature on the above resonance modes
The observed coupling coefficient k, and Q in a thick
(Fig. 13.) is explained by the following. As the temperature
was increased from 25"C, the epoxy is expected to me- sample at room temperature were attributed to the strong
chanically soften at a faster rate than PZT. Therefore, the mechanical interaction among PZT rods through the epoxy
resonancefrequency AI, which is mainly determined by matrix. At temperatures above the glass transition temthe elastic modulus of the polymer, decreased at a faster
perature TK = llO"C), the epoxy is rubber like and soft.
rate than the resonance frequency fi, which is defined by As a result, the mechanical interaction among PZT rods
thecompositemodulus E,. As seen in Fig. 13(b)-(d),at
becomesweaker as also indicated by an increase in the
higher temperaturesf,, gradually approachesf, and finally amplitude of the resonance spectrum (Fig. 13(e)-(h)). In
merges withf, at approximately 110C.In the temperature effect, k, of the composite approach kS3 of afreely susrange 11O-13O0C, the two modes interferewith each other. pended PZT rod (70 percent) and Q decreases because of
At temperatures above 130"C, f r l passesthrough h and the damping provided by the lossy polymer. It may be reeventually disappears ata temperature around 135 "C. This called here thatfor a thin (thickness = 0.6 mm) composite
disappearance can be attributed to the fact that above the sample, k, was 60-70 percent, and Q was 3-10 (Table V).
glass transition temperature the epoxy is less viscous and These values are very closetothosemeasuredforthick
cannot support a standing wave pattern of the transverse composites above the temperature TR.It appears that the
wave (glasstransitiontemperature
of Spurrs epoxy is nature of vibration of a thick composite above TK resemaround 110C at 200 KHz, from Section IV). It is worth- bles that of a thin composite at room temperature.

CURURAJA e1 al. : PIEZOELECTRICCOMPOSITEMATERIALS-PAR7

(d)

(h)
Fig. 13. Temperature-dependenc e o f various resonances in a composite
samples (sample # 103). The x axis isthe frequency 0.1 t f l 1 . l MHz and
the y axis is the relative admit!.ance (50 dB).

495

IEEE TRANSACTIONSON

496

SONICS ANDULTRASONICS.

VOL.S U-32,

NO. 4, JULY 1985

further supports the proposed model describing the nature

of vibration for different resonance modes of 1-3 composites.
VIII.CONCLUSION
The resonance modes in circular shaped 1-3 PZT rodpolymer composites were fully characterized. In general,
the planar coupling coefficient increased with increase in
volumepercent PZT. For a givenvolumepercentPZT,
thick samples had better planar coupling coefficient than
the thin samples.
For thickness mode resonanceof thin samples, k, = 6070 percent and Q = 3-10. For thick samples, kt = 50 percent and Q = 20-30. The difference in the observed behaviorwasexplained by considering thestrength of the
interaction among the PZT rods through theepoxy. At the
thickness-moderesonance,thecoupling
of acousticenergy from PZT to epoxy was found to be most efficient,
when the transverse wavelength in epoxy at the resonance
frequency was much larger than the periodicity of the lattice.Whenthisconditionissatisfied,thecompositevibrates like a homogeneous body and it will be shown in
Part I1 of this work that the composite materials exhibit
excellentsensitivity operating into water load. From the
effect of temperature on resonancebehavior, it is clear that
theinteractionamongthePZTrodsthroughtheepoxy
matrix can either be enhanced or suppressed depending
upon the properties of the polymer phase. Compositeswith
low transverse coupling could be utilizedforlinearand
phased array systems.
In addition to these two modes
of resonances, additional
modes were observed because of the regular lateral periodicity in the structure. The resultsshowed that these resonances arose from the superposition of the standingwave
pattern of the transverse waves in the epoxy. The experimental results werein excellent agreement with the theory
of wave propagation in two dimensionally periodic solids.

IO0
200

0
0

Temperature, C

Fig. 14. Typical effect of temperature on the thicknecs coupling coefficient

of a thick composite sample (sample # 103).

o Cooling Run

H e m n q Run

ACKNOWLEDGMENT

-7

The authors are grateful to Dr. W. A. Smith and Dr. A.

Shaulov of North American Philips Laboratories and Dr.
P. A . Lewin of Drexel University (Dept. of Electrical and
Computer Engineering and Biomedical Engineering and
ScienceInstitute) for theircommentsanddiscussions,
P.
Mrs. LindaWebsterforpreparingsamples,andMr.
Moses for writing software for computer interfacing.

cc
0.60

50

l00

I50

Temperature, C

Fig. 16. Typicaleffect of temperature on thethicknesscoupling

cient of a thin composite sample ( I ) .

coeffi-

REFERENCES
In thin samples, since the coupling among the PZT rods [ I ] R . E. Newnham. D. P. Skinner,and L. E. Cross, Connectivity and

of tempiezoelectric-pyroelectric composites, Marer. Res. Bull., ~01.13,pp.

525-536,1978.

perature above TR is unlikely to affect the resonance behavior. This seems to be confirmedby the plot of k, versus I?] -, Flexiblecompositetransducers, Muter. Res. Bull., vol. 13,

pp. 599-607, 1978.

temperature for a typical thincompositesamplewith

131 K. A . Klicker, J. V. Biggers.and R . E. Newnham,Composites of
PZT and epoxy for hydrostatic transducer applications. J.Arner. Cethickness mode frequency at 2.25 MHz (Fig. 16). No sigrum. Soc., vol. 64,pp. 5-8, 1981.
nificant variation of k, was seen over the entire tempera141 K. A . Klicker. W. A . Schulze, and J. V. Biggers, Piezoelectric comture range of 0 to 150C. Thus, the temperature depenposites with 3-1 connectivity and a foamed polyurethane matrix, J.
Amer. Cerum. Soc., vol.6, pp. C208-C210, 1982.
dence of the electromechanical properties of composites

is very weak at room temperature, the increase

GURURAJA

er a ! . :

PIEZOELECTRICCOMPOSITEMATERIALS-PART

[ S ] R. E. Newnham, L. J. Bowen. K . A. Klicker. and L. E. Cross, Compositepiezoelectrictransducers,

Mater. E n g . , vol. 2, pp. 93-106,
1980.
161 K.Rittenmyer, T. R.Shrout. W. A.Schulze. and R . E. Newnham,
Piezoelectric 3-3 composites, Ferroelectrics. vol. 41. pp. 189-195,
1982.
171 A.Safari, R . E. Newnham,L. E.Cross, and W.A.Schulze,Per-

foratedPZT-polymercompositesforpiezoelectrlctransducerappli-

cations. Ferroelecfrics. vo1.41, pp. 197-205, 1982.

[8] H
. Banno,and S. Saito,Piezoelectricanddielectricproperties
of
composites of syntheticrubberandPbTiO?andPZT.
in J. Appl.
Phys. Jap., vol. 22, Suppl. 2, pp. 67-69. 1983.
191P. N. T. Wells, BiomedicalUltrasonics. London:Academic, 1977.[ 101 G . Kosoff, The effects of backings and matching on the performance
of piezoelectric ceramic transducers. IEEE Trans. Sonics Ultrason.,
vol. SU-13, pp.20-30, 1966.
[l11 J. H. Goll and B. A. Auld, Multilayer impedance-matching schemes
for broadbanding of water loaded piezoelectric transducers and high
Q electricalresonators, IEEE Trans. SonicsUlrrason., vol.SU-22,
pp. 52-53, 1975.
[l21 C. S. Desilets, J. D. Fraser, and G. S . Kino, The design of efficient
broadband piezoelectric transducers, IEEE Trans. Sonics Ultrason.,
vol. SU-25, pp. 115-125, 1978.
[l31 J. Soquet, P. Defranould, and J. Desbois, Design of low-loss widebandultrasonictransducersfor
noninvasive medical application,
IEEE Trans. Sonics Ultrason., vol. SU-26, pp. 75-81, 1979.
1141 S. Lees, and C. L. Davidson, Ultrasonic measurement of some minera1 filled plastics, IEEE Trans. Sonics U/rrasonics. vol. SU-24, pp.
222-225, 1977.
[l51 J. H. Goll, Thedesign of broadband fluid-loadedultrasonictransducers, IEEE Trans. SonicsUltrason., vol. SU-26, pp. 385-393,
1979.
[l61 G. M. Sessler, Piezoelectricity in polyvinylidene fluoride," J. Amust.
Soc. Amer., vol.70,pp. 1596-1608, 1981.
[l71 J. Callerame,R.H.Tancrell,and
W. T. Wilson,Transmitterand
receiverformedicalultrasonics,
in Proc. 1979 /E Ultrason.
Symp., 1979,pp. 407-411.
[l81 J. W. Hunt, M.Arditi,andF. S. Foster.Ultrasoundtransducersfor
pulse-echo medical imaging, IEEE Trans. Blomed. E n g . , vol. BME30, pp. 453-481, 1983.
[l91 J. C . McGrath, L . Holt, D. M. Jones, and I. M. Ward, Recent measurementsonimproved thick-film piezoelectricPVDFpolymermaterials of hydrophoneapplications,
Ferroelectrics, vol. 50, pp.
13-20, 1983.
[20] S. Y. Lynn, R. E. Newnham, K. A. Klicker, K. Rlttenmyer, A . Safari,
and W. A . Schulze, Ferroelectric composites for hydrophones, Ferroelecrrics, vol.38, pp. 955-958, 1981.
[21] T. R. Gururaja, W. A.Schulze. T. R. Shrout.A.Safari, L. Webster,
and L. E. Cross, High-frequency applications of PZT-polymer composite materials, Ferroelecfrics, vol. 39, pp. 1245-1248, 1981.
[22] K. R . Erikson, Tone-burst testing of pulse-echo transducer, / E E E
Trans. Sonics Ultrason., vol.SU-26,pp. 7-14, 1979.
[23] T. R. Gururaja, D. Christopher, R . E. Newnham, and W. A . Schulze.
Continuous poling of PZT fibers and ribbons and
its application to
new devices, Ferroelectrics, vol. 47,pp.193-200. 1983.
[24] T. R.Guraraja, W. A. Schulze.andL.E.Cross.Bulkmechanical
properties of the Spurrs epoxy and eccogel polymers, submitted to
IEEE Trans. Sonics Ultrasonics.
[25] J. D.Ferry, ViscoelasticProperties of Polymers. New York: John
Wiley. 1980.
[26] IRE standards on piezoelectric crystals: Measurements of Piezoelectricceramics, Proc. IRE, vol.49,pp. 1161-1169. 1961.
1271W.P.
Mason,Electrostrictive effect in bariumtitanateceramics,
Phys. Rev., vol. 74, pp. 1134-1147, 1948.
[28] D. A. Berlincourt, D. R. Curran. and H. Jaffe, Piezoelectric and piezomagnetic materials and their function in transducers, in Physrcal
Acoustics, vol. I , W.P.
Mason,Ed.
New York: Academic, 1964,
part A, Chap. 3.
[29]C.A.Ross,Stress
wave propagation in compositematerials, D?namic Response of Composite Materials, (Fall meeting of the Society
for Experimental Stress Analysis), pp. 1-19, 1980.
[30] R . Holland, E. P. EerNisse,Accuratemeasurement of coefficients
in a ferroelectricceramic, IEEE Trans. Sonics Ultruson., vol. SU16, pp. 173-181, 1969.
[31] B. A. Auld, H. A. Kunkel,
Y. A . Shui,and Y. Wang,Dynamicbehavior of periodic piezoelectric composites, in Proc. 1983 /E Ulfrason. Symp., 1984, pp. 554-558.

497

T. R. Gururaja was born on September 24, 1954,

in Bangalore,India.HereceivedtheB.Sc.and
M.Sc. in physics from the University of Mysore,
India. in 1974 and 1976, respectively. He received
the M.Tech. degree in materials science from the
IndianInstitute of Technology, Kanpur in 1978 and
the Ph.D. degree in solid state science from The
PennsylvaniaStateUniversity, University Park in
1984.
m:
currently
isHeResearch
Associate
with the
Materials Research Laboratory, The Pennsylvania
State University. Hisresearchinterests are in the fieldof dielectricand
piezoelectriccrystals,ceramics,andcomposites.
He is alsoworking on
ultralow permittivity materials for packaging Ga:As integrated circuits.
I

_-

>

Walter A. Schulze, Jr. (84) was born on December 8, 1943. HeobtainedtheB.S.degree

in
1965 andtheM.S.andPh.D.degrees
in solidstate science in 1968 and1973, respectively,from
The Pennsylvania State University.
He is currentlyAssociateProfessorwiththe
Ceramic Science department at Alfred University.
He worked as Senior Research Associate at The

Pennsylvania
State
University
from
1974 to 1983.
Dr. Schulze hasbeenworking
in theareas of
ferroelectric,piezoelectric,andpyroelectricmaterialsandcomposites.thick-filmmaterialsforcircuitryanddielectrics,
ceramic preparation, second harmonic generation, and electrical property
measurements.
g,

Leslie E. Cross (SM79-F84) was born on August 14, 1923, in Leeds, England. He received the
B.Sc.andPh.D.degrees
in physics fromLeeds
University in 1948 and1952,respectively.
Currently, he is Professor of ElectricalEngineeringandAssociateDirector
of the Materials
ResearchLaboratoryatThePennsylvaniaState
University. Prior to joining ThePennsylvania State
University, in 1961, he held positions with Leeds
University as Lecturer (1948-1951) and as a research associate (1954-1961).
Hisresearchinterestsincludeferroelectricmaterialsandferroic
phenomena;dielectric,piezoelectricandpyroelectriccrystals,ceramicsand
composites; electronic ceramics and their applications, electrostriction and
phase transitions.
Dr. Cross is a member of the National Academy of Engineering and a
Fellow of the American Institute of Physics and the American Ceramic Society.

Robert E. Newnham was born on March 28, 1929

in Amsterdam, NY. He received the B.S. degree
from Hartwick College, Oneonta, NY, in 1950 and
theM.S.degree from Colorado State University,
Fort Collins, in 1952. He receivedhis Ph.D. degree in physics from The Pennsylvania State University, University Park, i n 1956 and subsequently
he receivedthe Ph.D.degree in crystallography
from Cambridge University, England, in 1960.
CurrentlyDr.Newnham is Professor of solid
state science and ceramic science at The Pennsylvania State University. He is also the chairman of the solidstatescience
program. Before he joined The Pennsylvania State University, he worked
as Research Associate and Assistant Professor
i n the Massachusetts Institute of Technology, Cambridge, MA, from 1958-1966.
His research interests include crystal physics and structure-property relationships, phase transitions in ferroic materials, piezoelectric, pyroelectricandelectrostrictivephenomena,andcompositematerialsfortransducerapplications. He is author of a bookentitled StructureProperty
Relations.
Newnham
aDr.is
Fellow ofMineralogical
the
Society
of America.

498

IEEE TRANSACTIONS ON SONICS

ULTRASONICS,
AND

VOL. SU-32, NO. 4, JULY 1985