_______________________________________________3

MATERIAL AND DEVICE DESIGNING AND

FABRICATION PROCESSES

After the material design such as solid solution compositions and dopants, we need
to consider material fabrication processes. The fabrication of ferroelectric devices
generally involves two steps: preparation of the ceramic powders and sintering of
the shaped structures. Wet chemical preparation methods are utilized for producing
the ceramic powders in order to ensure reproducibility of the advanced
characteristics of the devices. Popular device designs include multilayers, bimorphs
and thin/thick films. Some necessary basic knowledge such as particle size effect
and domain contribution to ferroelectricity will also be discussed in this chapter.

3.1 MATERIAL DESIGNING

(1) Composition Selection

Let us consider an example for piezoelectric application. In general, Pb(Zr,Ti)O3

(PZT), PbTiO3 (PT), (Pb,La)(Zr,Ti)O 3 (PLZT) and PZT-based ternary ceramics are
utilized for piezoelectric applications. Their piezoelectric coefficients are
summarized in the international data book:

K. H. Hellwege et al.: Landolt - Bornstein, Group III, Vol.11, Springer-

Verlag, N.Y. (1979).

Figure 3.1 shows the composition dependence of the permittivity and the electro-
mechanical coupling factor kp for the PZT system. 1)

If we do not have this sort of comprehensive experimental data, how can we

estimate the values for the solid solutions? In general, physical properties of a solid
solution between A and B, (1 - x) A - x B, can be estimated by a phenomenological
theory.2,3) The Gibbs elastic energy of a solid solution is assumed to be a linear
combination of the Gibbs elastic energy of each component:

G1 (P,X,T) = (1/2)[(1 - x)αA + xα B] P2 + (1/4)[(1 - x)βA + xβ B] P4

+ (1/6)[(1 - x)γA + xγ B] P6
- (1/2)[(1 - x)s A + xs B] X2 - [(1 - x)QA + xQ B] P2 X.

(αA = (T - T0,A) / ε0 CA, αB = (T - T0,B) / ε0 CB ) (3.1)

57
58 Chapter 3

The solution provides reasonable first-order estimates of the Curie temperature,

spontaneous polarization and strain, permittivity, piezoelectric constant and electro-
mechanical coupling. Abe et al. reported a good example of theoretical fitting to
experimental results for the solid solution Pb(Zn1/3Nb2/3 )O3 -PbTiO3 .3) Figures 3.2-
3.5 show these fittings calculated on the basis of the data presented in Table 3.1.

Fig. 3.1 Composition dependence of the permittivity and the electromechanical

coupling factor kp in the PZT system.

Table 3.1 Coefficients for PZN (Pb(Zn1/3Nb 2/3)O3 ) and PT (PbTiO 3) used in
calculating the transition temperature and lattice parameters in (1-x)PZN-xPT.
__________________________________________________
Constants PZN PT
________________________________________________________
T 0 (oC) 130 478.8
C (105 oC) 4.7 1.5
ξ11 (107m5C-2F-1) -13.7 -29.0
ξ12 (108m5C-2F-1) -1.0 15.0
ζ111 (108m9C-4F-1) 10.3 15.6
ζ112 (108m9C-4F-1) 6.8 12.2
Q11 (10-2 m4C-2) 2.4 8.9
Q12 (10-2 m4C-2) -0.86 -2.6
Q44 (10-2 m4C-2) 1.6 6.75
ac (A) 4.058 3.957
________________________________________________________
Material and Device Designing and Fabrication Processes 59

Fig. 3.2 Phase diagram for (1 -x) PZN - x PT; (a) calculated and (b) experimental.

Fig. 3.3 Lattice parameters for xPZN-(1-x)PT; (a) calculated and (b) experimental.
60 Chapter 3

Fig. 3.4 Temperature dependence of lattice parameters for 0.91 PZN - 0.09 PT; (a)
calculated and (b) experimental.

Fig. 3.5 Temperature dependence of dielectric constant for x PZN - (1 - x) PT;

(a) calculated and (b) experimental.
Material and Device Designing and Fabrication Processes 61

(2) Dopant Effects on Ferroelectricity

Small amounts of dopants sometimes drastically change the dielectric and electro-
mechanical and electrooptic properties of ceramics. The PTC effect in
semiconductor barium titanate in Chap. 9 is the most notable example of the effect
of dopants on electrical properties.

Let us consider the crystallographic deficiencies due to impurity doping in a

perovskite crystal. Acceptor ions, such as Fe3+, introduce oxygen deficiencies (¨)
in the PZT lattice as follows:

Pb(Zry Ti1-y-xFe x)(O3-x/2¨ x/2).

This acceptor doping causes the easy reorientation of deficiency-related dipoles.

These dipoles are generated by an Fe3+ ion (effectively negative charge) and an
oxygen vacancy site (effectively positive charge). The deficiencies are produced at
a high temperature (> 1000°C) during sintering, however, the oxygen ions are still
movable even below the Curie temperature (e.g. at room temperature), because the
oxygen and the vacancy are adjacent (only 2.8 A!) and hopping easily occurs [see
Fig. 3.6(a)].

In the case of donor ions, e.g. Nb 5+ , Pb deficiency is introduced:

(Pb 1-x/2¨ x/2)(Zry Ti1-y-xNb x)O 3.

Donor doping is not very effective for generating movable dipoles, since the Pb ion
can not easily hop to an adjacent A-site vacancy due to the close oxygen
surroundings [Fig. 3.6(b)].

O Vacancy A Vacancy

easy t o mov e pr ohi bi t ed

3+
5+

(a) A ccept or Dopi ng (b) Donor Dopi ng

Fig. 3.6 Crystal deficiencies in PZT for acceptor (a) and donor (b) dopants.
62 Chapter 3

The previously mentioned crystal deficiency model helps us to understand various

ferroelectric property changes with doping. Hagimura and Uchino4) studied the
effect of impurity doping on field-induced strains using a base composition of
(Pb 0.73Ba0.27)(Zr0.75Ti0.25) O3 (soft piezoelectric). Figure 3.7(a) shows the
definitions of the maximum strain xmax, and the degree of hysteresis ∆x / xmax (%).
The degree of hysteresis is calculated from the strain deviation during the rise and
fall of the field ∆x at half of the maximum electric field (1 kV/mm) divided by the
maximum strain xmax at 1 kV/mm. Figure 3.7(b) shows the dopant effect on the
strain, in which the doping concentration is fixed at 2 at. %. It is seen that materials
incorporating donor-type ions on the B site with a large valence +4 - +6 ion (Ta 5+,
Nb 5+ , W6+ ) exhibit excellent characteristics as positioning actuators, namely,
enhancement of the strain magnitude and reduction of the hysteresis.

(a) (b)

Fig. 3.7 Dopant effect on the field-induced strain. Maximum strain and hysteresis
in Pb 0.73Ba0.27(Zr 0.75 Ti0.25)O3 -based ceramics. (a) Definition of the maximum
strain and the degree of hysteresis, and (b) dopant effect on actuator parameters.
Material and Device Designing and Fabrication Processes 63

On the contrary, the acceptor-type ions with a small valence +1 - +3 suppress the
strain and increase the hysteresis and the coercive field. Although acceptor-type
ions must be avoided in designing actuator ceramics used for positioner
applications, acceptor doping is very necessary for making "hard" piezoceramics
which are particularly suitable for ultrasonic motor applications. These dopant
effects can be attributed to the domain pinning effect.

Example Problem 3.1_________________________________________________

The coercive electric field of a ferroelectric material, which is defined as a field for
reorienting the polarization direction, is affected by dopants. Explain the dopant
effects on the "soft" and "hard" characteristics of PZT piezoelectric ceramics.

Solution

First, the "soft" and "hard" characteristics are a reflection of the coercive field EC,
in other words, the stability of the domain walls. The "hard" piezoelectric is defined
for a material with the EC > 1 kV/mm, while the "soft" one is for EC << 1 kV/mm.

Consider the transient state of a 180o domain reversal, which reveals a domain wall
front with a head-to-head polarization configuration. From Gauss's law,

div P = ρ, (ρ: charge density) (P3.1.1)

the domain wall front is very unstable in a highly insulating material, leading to
quick disappearance of this domain wall, i. e. a low coercive field. However, if the
material has movable charges, the head-to-head polarization configuration is
stabilized, leading to a high coercive field.

Next, let us consider the movable charges due to the crystallographic deficiencies.
Acceptor ions, such as Fe3+, introduce oxygen deficiencies and in the case of donor
ions, such as Nb 5+, Pb deficiency is introduced. Thus, the acceptor doping causes
domain pinning through the easy reorientation of deficiency-related dipoles, leading
to "hard" characteristics. On the contrary, the donor doping is not very effective for
domain pinning, since the Pb ion can not easily hop to an adjacent A-site vacancy
due to the close oxygen surroundings.

Moreover, it is noteworthy that lead-containing ceramics such as PZT tend to be p-

type semiconductors due to Pb evaporation during sintering. Hence, donor-type ion
doping provides rather "soft" characteristics to the piezoceramics, since donor
doping can compensate the original acceptor type deficiencies. Donor-doped
piezoceramics exhibit large piezoelectric d constants, but also a large aging effect
due to depoling.
___________________________________________________________________
64 Chapter 3

(3) High Power Characteristics

Let us consider here high power piezoelectric ceramics for ultrasonic (AC drive)
applications. In order to obtain a large output power, the ceramics are driven under
a high vibration level, namely under a relatively large AC electric field. This causes
heat generation as well as a drastic degradation in piezoelectric properties.
Therefore, the high-power device such as an ultrasonic motor requires a very "hard"
piezoelectric with a high mechanical quality factor QM , to suppress heat generation.
The QM is defined as an inverse value of the loss factor, tan δm. It is also notable
that the actual mechanical vibration amplitude at the resonance frequency is directly
proportional to this QM value.

Since the measurement technique to determine the piezoelectric characteristics at

high vibration levels was not established previously, only a few studies were made
by Uchino et al.5) Recently, Hirose et al. have succeeded in developing a method
for measuring the electromechanical properties with higher precision and stability at
high vibration levels with a constant current circuit.6)

The vibration velocity dependence of the piezoelectric constant, permittivity, elastic

compliance and electro-mechanical coupling factor was obtained for PZT based
samples.7) Figure 3.8 shows the mechanical quality factor QM and the actual
temperature rise plotted as a function of the average vibration velocity. Notice a
significant decrease in Q above a certain critical vibration level, which is the limit
of the output vibration energy. Even if the input electrical energy is increased
further, the additional energy will be converted only into heat.

QB
2000 40

QA
1 500 30

7
1000 2 20
43
TEST SAMPLE
A-type
500 10
Temperature B-type
0 Rise 0
0.01 0.02 0.05 0.1 0.2 0.5 1
Vibration Velocity v 0 (m/s)
Fig. 3.8 Vibration velocity dependence of the quality factor QM and temperature
rise for both A (resonance) and B (antiresonance) type resonances of a
longitudinally vibrating PZT ceramic transducer through d 31.
Material and Device Designing and Fabrication Processes 65

60
Undoped Pb(Zr0.54Ti0.46 )O3
0.5 wt% Nb-doped
40
0.5 wt% Fe-doped

20

0
0.01 0.05 0.1 0.5 1.0
Vibration Velocity v0 (m/s)

Fig. 3.9 Temperature rise ∆T vs. effective vibration velocity v for PZT samples.

Pb(ZrxTi1-x)O3 + 2.1 at% Fe

2000

Vibration velocity
600
v0 = 0.05 m/s

1000 v0 = 0.5 m/s 400

200

0.48 0.50 0.52 0.54 0.56 0.58

Mole Fraction of Zr (x)

Fig. 3.10 Mechanical quality factor Qm vs. mole fraction of Zr (x) at effective
vibration velocities v0 = 0.05 and 0.5 m/s for Pb(Zr xTi1-x)O 3 + 2.1 at% Fe3+
ceramics.

Figure 3.9 shows the temperature rise versus vibration velocity for undoped, Nb-
doped and Fe-doped PZT samples. The suppression of heat generation is
remarkable in the Fe-doped (acceptor-doped) ceramic.
66 Chapter 3

Figure 3.10 shows mechanical QM versus mole fraction of Zr (x) at effective

vibration velocities v0=0.05 m/s and 0.5 m/s for Pb(ZrxTi1-x)O 3 doped with 2.1
at.% of Fe3+.8) The decrease in mechanical Q with increase in vibration level is
minimum around the rhombohedral-tetragonal morphotropic phase boundary
(52/48). In other words, the worst material at a small vibration level becomes the
best at a large vibration level, and data obtained from an conventional impedance
analyzer are not relevant to high power materials.

Figure 3.11 highlights the key material-related factors affecting heat generation in
the piezoelectric material. The resistances Rd and Rm in the equivalent electrical
circuit are separately plotted as a function of vibration velocity.9) Note that Rm,
mainly related to the mechanical loss, is insensitive to the average vibration
velocity, while Rd , related to the dielectric loss, changes significantly around a
certain critical vibration velocity. Thus, the resonance loss Qm-1 at a small vibration
velocity is mainly determined by the intrinsic mechanical loss, and with increasing
vibration velocity, the intrinsic dielectric loss contribution significantly increases.
We can conclude that heat generation is caused primarily by dielectric loss (i.e., P-E
hysteresis loss). Refer to Chapter 7, Section 7.3 (2) for the equivalent circuit.

RA (directly measured)
= Rd + Rm
100

30

Rm = ω AL A/Q B

10

Rd = tan δ /ω AC d

3.0

0.03 0.1 0.3 1.0

Vibration Velocity v 0 (m/s)

Fig. 3.11 Vibration velocity dependence of the resistances Rd and Rm in the

equivalent electric circuit for a piezoelectric component.
Material and Device Designing and Fabrication Processes 67

3.2 FABRICATION PROCESSES OF CERAMICS

Ferroelectric devices are typically fabricated from polycrystalline ceramics. This
involves two steps; preparation of the ceramic powders and sintering of the shaped
structures.

(1) Preparation of Ceramic Powders

Particle shape, particle size distribution and compositional uniformity are the key
factors to be controlled in the raw powder in order to realize reproducibility of the
piezoelectric characteristics. The usual method is the oxide-mixing technique, in
which a desired chemical composition is made by firing raw oxide powders
(calcination), then crushing them into fine powders. Since the oxide-mixing
method results in difficulties in achieving microscopic compositional uniformity,
wet chemical methods (coprecipitation, alkoxide) have been employed more
recently in manufacturing ceramic devices. In this section, processes for fabricating
barium titanate (BT), lead zirconate titanate (PZT) and lead magnesium niobate
(PMN) ceramics are reviewed.10)

Solid State Reaction

Let us consider the preparation of Pb(Zr xTi1-x)O 3 powders. The raw powders
PbO, ZrO2 and TiO2 are weighed in an appropriate proportion, mixed, and calcined
at around 800 - 900°C for 1 - 2 hours. Then the sample is crushed and milled into
fine powders. The drawbacks here are that the milling process does not efficiently
give particles of size less than 1 µm, and that the contamination of the sample by
milling media is unavoidable.

BaTiO3 is prepared in principle from equi-molar quantities of raw powders of BaO

and TiO2 . In general, BaCO3 powder is recommended instead of BaO, because high
purity BaO is expensive and chemically less reactive.

A similar calcination process starting from PbO, MgO, Nb 2 O5 and TiO2 can be used
for Pb[(Mg1/3Nb 2/3) 1-xTix]O3 . However, this simple process generates a second
phase (pyrochlore) in addition to the perovskite phase. To suppress this second
phase several mol % excess PbO doped in the final sintering stage is effective.11)
Swartz et al. reported a unique method taking account of the chemical reaction
process.12) They demonstrated that the perfect perovskite phase can be obtained by
the reaction starting from columbite MgNb 2O6 and PbO:

3PbO + MgNb 2 O6 -----> 3 Pb(Mg1/3Nb 2/3)O 3

For PMN-PT, MgO, Nb 2 O5 and TiO2 are mixed and fired at 1000°C initially. Then
PbO is added to the columbite, and the sample is calcined at 800 - 900°C. Several
mol% excess MgO is particularly effective in obtaining the perfect perovskite
phase.
68 Chapter 3

Coprecipitation

Since practical piezoelectric/electrostrictive ceramics are from complex solid

solution systems, the compositional fluctuation among ceramic particles is very
important, as well as the purity. The above method involving a combination of
solid state reaction and mechanical mixing, is inevitably accompanied by this
problem. The coprecipitation method has been employed to realize enhanced
homogeneity. Adding a precipitant into the liquid solution of mixed metal salts
produces a homogeneous precipitate. Thermal dissolution makes homogeneous
powders from the precipitate.

For example, BaTiO(C2 H4 )2 .4H 2O with the Ba/Ti ratio exactly equal to 1:1, on an
atomic scale, can be precipitated from an aqueous solution of BaCl2 and TiCl4 by
dripping in oxalic acid. Thermal dissolution of this precipitate provides fine
stoichiometric BaTiO3 powders with good sintering characteristics.

For PLZT, Pb(NO3 )2 , La(NO3) 3. 6H 2 O, ZrO(NO3 )2 .2H 2O and TiO(NO3 ) 2 are

used as the starting materials.13) First, the water solutions of each nitrate are mixed
in the desired proportion, then a half volume of ethanol is added. Oxalic acid
diluted with ethanol is dripped slowly into the nitric solution, leading to the
precipitation of the PLZT oxalate. Thermal dissolution is carried out at 800°C.

In the above coprecipitation methods, thermal dissolution of the precipitate is finally

required to obtain the desired powder. However, the oxide powder can be obtained
directly through the precipitation process in special cases. A direct precipitation
method for BaTiO3 is introduced here. By dripping Ti(OR) 4 (R: propyl) into
Ba(OH) 2 water solution, fine high-purity stoichiometric BT powders with an
average particle size of 10 nm can be obtained directly.

Alkoxide Hydrolysis

When metal alkoxides M(OR)n (M: metal atom, R: alkyl) are mixed in alcohol in an
appropriate proportion, and water is added, the hydrolytic reaction produces alcohol
and the metal oxide or metal hydrate. This is sometimes called the sol-gel method.
The alkoxide method can produce very fine and very high purity powders. Since
metal alkoxides are evaporative, purification is easily accomplished through
distillation. Also high purity can be sustained during the hydrolic reaction because
no other ion doping is necessary. The mechanisms of hydrolysis and condensation
can be summarized as follows:

(a) Hydrolysis

H - O + M - OR -----> H - O - M + ROH
|
H
Material and Device Designing and Fabrication Processes 69

(b) Alkoxylation (Removal of H as an alcohol)

M - O + M - OR -----> M - O - M + ROH
|
H

(c) Oxolation (Removal of H as water)

M - O + M - OH -----> M - O - M + OH 2
|
H

In order to obtain BT powders, Ba(OC3 H 7) 2 and Ti(OC 5 H11 )4 are diluted with
isopropyl alcohol (or benzene). Choosing an appropriate hydrolysis condition (pH),
very fine stoichiometric BT powders with good crystallinity and a particle size of 10
- 100 A can be obtained (agglomerate size = 1 µm). Also the hydrolytic process
gives a specimen powder with purity more than 99.98%, leading to a remarkable
increase in the permittivity of the sintered ceramic in comparison with samples
prepared by oxide-mixing.14) Table 3.2 lists examples of ferroelectric ceramic
powders synthesized by alkoxide hydrolysis.

In the case of PZT preparation, Pb alkoxide is relatively difficult to obtain compared

to Ti and Zr alkoxides. Thus, one way is to employ a two-stage method for
convenience: (Zr,Ti)O2 is prepared by the alkoxide method, and PbO is added later
for use in the solid state reaction.15) A partial sol-gel method combined with
inexpensive ready-made nanosize powders will be a promising direction to achieve
reduction in production costs. Of course, the other way is to use Pb alkoxide;
zirconium n-butoxide Zr(O(CH 2 )3 CH3 ] 4 and titanium isopropoxide
Ti[OCH(CH 3 )2 ]4 are added into lead acetyl acetonate Pb(CH3COCHCOCH3 ) to
obtain PZT precursor.

Table 3.2 Ferroelectric powders synthesized by hydrolysis of alko-oxides.

______________________________________
Crystalline BaTiO 3
Ba(Zr,Ti)O3
(Ba,Sr)TiO 3_______
Amorphous Pb(Mg1/3Nb2/3)O3
Ba(Zn1/3Nb2/3)O3
Pb(Zr,Ti)O3
(Pb,La)(Zr,Ti)O3
______________________________________
70 Chapter 3

(2) Sintering Process

After being shaped into a desired shape, the agglomerated powder body is fired at a
high temperature (less than the melting temperature). Accelerated diffusion of the
constituent atoms on the fine particle surfaces due to the surface energy (surface
tension) promotes crystal bonding at the contact interface between the two adjacent
particles and provides sufficient mechanical strength to the ceramic without
significant distortion from the inital molded shape. This firing process is called
"sintering," which primarily eliminates pores and increases the ceramic density (see
Fig. 3.12). Notice that the physical properties of the sintered body depends not only
on the property of each fine crystalline particle, but also on the grain boundary and
the pores. An example is found in the mechanical strength: mechanical fracture in
ceramic bodies occasionally occurs at the grain boundary (intergranular type). On
the contrary, when the crystal itself has a strong cleavage character, the
polycrystalline material shows higher mechanical strength.

During sintering, the grains grow and the grain shape also changes significantly.
However, it is well recognized that the raw powder characteristics strongly affect
the manufacturing conditions and the final product characteristics. In general, the
sintering is accelerated with decreasing particle size of the raw powder (i.e., with
increased specific surface area), because the driving force of sintering is related to
the surface energy of the particles. Moreover, for fine powders, the necessary
diffusion length of the atoms for sintering becomes shorter, which accelerates pore
diffusion. This results in high density ceramics.

Diffusion

Neck

Raw powder Grain Grain boundary

Pore

Sintering

Molded Body Sintered Body

Fig. 3.12 Schematic diagram of sintering process.

Material and Device Designing and Fabrication Processes 71

(a)

(b)

Fig. 3.13 Grain growth in the PLZT ceramics 9/65/35 ceramics sintered for (a) 1
hour and (b) 16 hours.

There have been many studies on grain growth. Reference 16) is recommended for
further study. The following relationship between the grain size D and the sintering
period t is generally used:

Dβ - D0 β ∝ t . (3.2)

In the case of normal grain growth, β = 2, and for abnormal grain growth, β = 3.
Figure 3.13 shows the microphotographs of a PLZT 9/65/35 surface sintered at
1200°C for 1 and 16 hours, starting from the oxalic acid/ ethanol method.17) Figure
3.14 shows a good linear relation between the sintering period and the square of the
grain size.
72 Chapter 3

D2 - 0.601 = 0.62 t
10

0 4 8 12 16 20
Sintering time (hr)

Fig. 3.14 Grain growth in PLZT as a function of sintering time.

Doping must also be considered. The main role of dopants is to decrease the
sintering temperature, but additional effects such as suppression or enhancement of
grain growth are sometimes observed. Excess PbO or Bi2O 3 in Pb based
perovskites (PZT) inhibits grain growth. Dy doping (0.8 at.%) is very effective in
suppressing the grain size to less than 1 µm in BaTiO3 .18)

Example Problem 3.2_________________________________________________

Ions K1+, Bi3+, Zn2+ and Nb 5+ make a disordered complex perovskite crystal, the
former two occupying the A-site and the latter two occupying the B-site. Determine
the possible composition formulas.

Solution

Taking into account the charge neutrality of ABO3 = (K 1+ 1-xBi3+ x)(Zn2+1-y

Nb 5+y )O 2- 3 :
+1 .(1-x) +3. x +2 .(1-y) +5.y = +6 (P3.2.1)
Then
2x+3y=3 (0 < x < 1, 1/3 < y < 1)). (P3.2.2)

We obtain continuous solutions for x and y, which satisfy Eq. (P3.2.2). A formula
(K 3/4Bi1/4) (Zn1/6Nb 5/6)O 3 is one of the solutions.
___________________________________________________________________
Material and Device Designing and Fabrication Processes 73

(3) Single Crystal Growth

Single crystals are not commonly used in piezoelectric/electrostrictive devices. The

few popular single crystals are quartz, grown by hydrothermal synthesis, and
LiNbO3 and LiTaO3 , by the Czochralski method. Recently, Nakamura et al.
reported on a monomorph bending actuator fabricated from a thin plate of LiNbO3
crystal, in which half of the thickness is reverse polarized to function like a
bimorph.19) Although this device is fragile and the bending displacement is not
large, linearity in the displacement curve without hysteresis is attractive for some
special applications such as scanning tunneling microscopes. Single crystal growth
of PZT has been investigated intensively for some time, nevertheless no crystal size
has exceeded 1 mm3 near the morphotropic phase boundary. Recently, Pb(Zn1/3
Nb 2/3)O 3 (PZN-PT) and PMN-PT have been the focus of medical acoustic
transducer applications, because large single crystals more than 1 cm3 can be easily
grown by a simple flux method, and enormously high electromechanical coupling
factors (95 %) and piezoelectric d constants (1570 x 10-12 C/N) can be obtained
when they are poled in a special crystal direction.20,21)

3.3 DEVICE DESIGNING

We will introduce several device designs in this section, including single disks,
multilayers, composites and thin/thick films.

(1) Single Disks

Single disk devices are not practically found these days because of low efficiency in
(performance/space). However, they are still important for the laboratory
experiments.

Example Problem 3.3_________________________________________________

The apparent dielectric constant of a barium titanate based ceramic was measured to
be 500, using a 1 mm thick disk sample. However, it was found that a thin air gap
of 0.5 µm developed between the sample ceramic and the coated electrodes over
most of the electrode area on both sides, because of lack of skill in fabrication.
Estimate the actual dielectric constant.

Solution

The total capacitance of a ceramic capacitor and two air gap capacitors connected in
series was measured experimentally. When we denote the capacitor area, thickness
and the air gap by S, d and δ, respectively, the total capacitance is given by

1/C = 1/(ε0 ε S/d) + 2/(ε0 S/ δ) = (1/ε0 S)((d/ε + 2 δ) (P3.3.1)

74 Chapter 3

Since the apparent dielectric constant was calculated from

C/(ε0 S/d) = 500, (P3.3.2)

The following relation is obtained:

(1/ ε) + (2 δ/d) = 1/500. (P3.3.3)

Substituting d = 10 -3 m, δ = 0.5 x 10-6 m, we obtain the real dielectric constant of ε

= 1000.

The above mistake is occasionally found when alcohol is used to wipe the ceramic
disk after polishing, and it is not dried completely on a hot plate (above 100°C).
The reader should be careful not to make a air gap (even submicron!) during the
electroding process.
___________________________________________________________________

(2) Multilayers

To achieve a low driving voltage, miniaturization and hybridization of the devices,

ferroelectric ceramic multilayer structures have been investigated intensively for
capacitor, actuator and electrooptic applications. Key words for the future trend
will be "finer" and "hybridization." Layers thinner than 10 µm, which is currently
used in multilayer capacitors, will also be introduced in actuator devices instead of
the present 100 µm thick sheets. Non-uniform configurations or hetero-structures
of the materials, layer thickness or the electrode pattern will be adopted for practical
devices.

There are two techniques for making multilayered ceramic devices: the cut-and-
bond method and the tape-casting method. The tape-casting method has been
widely used for multilayer capacitor fabrication, and requires expensive fabrication
facilities and sophisticated techniques, but is suitable for mass-production of more
than 10 thousand pieces per month.

As shown in Fig. 3.15, a multilayer structure is composed of alternate ferroelectric

ceramic and internal electrode layers fabricated by cofiring. An adjacent pair of
electrodes composes a unit displacement element, which is connected in parallel by
the external electrode up to hundreds of layers. Figure 3.16 shows a flowchart of
the manufacturing process of the multilayer ceramic actuators. Green sheets are
prepared in two steps: slip preparation of the ceramic powder and a doctor blade
process. The slip is made by mixing the ceramic powder with solvent, deflocculant,
binder and plasticizer. The slip is cast into a film under a special straight blade, a
"doctor blade," whose distance above the carrier determines the film thickness.
After drying, the film, called a green sheet, has the elastic flexibility of synthetic
leather. The volume fraction of the ceramic in the polymer matrix at this point is
about 50%. The green sheet is then cut into an appropriate size, and internal
Material and Device Designing and Fabrication Processes 75

electrodes are printed using silver, paladium or platinum ink. Several tens to 100s
such layers are then laminated, and pressed using a hot press. After cutting into
small chips, the green bodies are sintered at around 1200°C in a furnace, taking
special care to control binder evaporation at 500°C. The sintered chips are then
polished, externally electroded, lead wires are attached, and finally the chips are
coated with a water-proof spray.

External
electrode

Polarization direction
Internal electrode

Fig. 3.15 Structure of a multilayer actuator.

Calcined powder

(Binder mixing, Vacuumization)

Slurry

(Slip casting)

Green sheet

(Punching)
(Electrode printing)
(Lamination, Press, Cutting)

Green chip

(Binder evaporation, Sintering)

(External electrode printing)

Multilayer device

Fig. 3.16 Fabrication process for a multilayer ceramic actuator.

76 Chapter 3

Example Problem 3.4_________________________________________________

For a certain applied voltage V, a multilayer structure can amplify the generative
displacement in a piezoelectric actuator with keeping its total length. Larger
displacement magnification can be expected in an electrostrictor than in a
piezoelectric. Verify this using simplest assumptions: x = d E and x = M E2 for
piezoelectric and electrostrictive cases, respectively.

Solution

Assume L, ∆L and n to be the total thickness, displacement and the number of

layers of the ceramic actuator, respectively.

(a) Piezoelectric

∆L = Lx = L d E = L d [V / (L/n)] = n d V (P3.4.1)

The generative displacement is amplified in proportion to the number of layers n.

(b) Electrostrictor

∆L = Lx = L M E2 = L M [V / (L/n)] 2 = n 2 (M/L) V2 (P3.4.2)

In this case the generative displacement is amplified in proportion to the square of

the number of layers n (more effective than in a piezoelectric!).
___________________________________________________________________

120

100

80

60

40

20

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Ve / A (volume mm3/ area mm2 )

Fig. 3.17 Temperature rise versus Ve/A (3 kV/mm, 300 Hz), where Ve is the
effective volume generating the heat and A is the surface area dissipating the heat.
Material and Device Designing and Fabrication Processes 77

Zheng et al. reported on the heat generation from multilayer piezoelectric ceramic
actuators of various sizes.22) The temperature change was monitored in actuators
driven at 3 kV/mm and 300 Hz. Figure 3.17 shows the saturated temperature
plotted as a function of Ve/A, where Ve is the effective volume (electroded portion
of the actuator) and A is the surface area. This linear relation is reasonable because
the volume Ve generates the heat and this heat is dissipated through the area A.
Thus, if you need to suppress the heat, a small Ve/A (flat and cylinder shapes are
better than cube and solid rod structures, respectively) is preferred.

(3) Bimorphs/Moonies

Unimorph and bimorph devices are defined by the number of piezoelectric ceramic
plates: only one ceramic plate is bonded onto an elastic shim, or two ceramic plates
are bonded together, respectively. The bimorphs are mainly considered here. The
bimorph causes bending deformation because two piezoelectric plates are bonded
together and each plate produces extension or contraction under an electric field.
This effect is employed for various speakers. The bimorph induces voltage for a
bending deformation, which can be used for accelerometers. Since the fabrication
process is simple (two ceramic plates are just bonded using a resin) and large
magnification of the displacement is easily obtained, this structure has been widely
used. However, the drawbacks include a low response speed (1 kHz) and low
generative force due to the bending mode. A metallic sheet (called the shim) is
occasionally sandwiched between the two piezoceramic plates to increase the
reliability, that is, the structure is maintained even if the ceramic fractures (Fig.
3.18). Also notice the tapering of the bimorph, which increases the response
frequency by maintaining the tip displacement magnitude.

Elastic shim
Piezoceramic plate
+
(a) Polarization
direction
- +

(b)

Fig. 3.18 Fundamental structure of a piezoelectric bimorph.

78 Chapter 3

There have been many reports on equations describing the tip displacement and the
resonance frequency. Summaries are provided here. Figure 3.19 illustrates two
bimorph designs without shims. Two poled piezoceramic plates with t/2 in
thickness (i.e., t is the total thickness) and L in length are bonded with their
polarization directions opposite to each other (a) or parallel to each other (b).
According to the configuration, the tip displacement δ under a voltage V is provided
as follows when one end is clamped (cantilever condition):

(a)

(b)

Fig. 3.19 Two types of piezoelectric bimorphs: (a) the anti-parallel polarization
type and (b) the parallel polarization type.

δ = (3/2) d 31 (L2/ t 2 ) V, (3.3a)

δ = 3 d 31 (L2/ t 2) V. (3.3b)

Notice that this difference comes from the electrode gap difference: t in (a) and t/2
in (b). For both cases the fundamental resonance frequency is determined by the
total thickness t as,23)

f = 0.161 (t/ L2 ) (ρ s 11E )-1/2. (3.4)

As can be anticipated, the bimorph drive is inevitably accompanied by a rotational

motion. To obtain a perfect parallel motion a special mechanism must be
employed. Figure 3.20 shows such a bimorph structure. A comp lex bimorph
proposed by Ampex has divided electrodes electrically connected oppositely at the
tip and bottom (supporting part) parts so as to compensate the canting angle at the
bottom by the opposite bend at the tip.24) The bimorph also included a sensor
function: the sensor electrode can detect the voltage generated in proportion to the
magnitude of bend.
Material and Device Designing and Fabrication Processes 79

Sensor electrode

Support Piezoactuator

Fig. 3.20 Bimorph structure for a perfectly parallel motion with a position sensing
feedback function.

Example Problem 3.5_________________________________________________

Using a PZT based ceramic with a piezoelectric constant of d31 = -300 pC/N, design
a no-shim bimorph with a total length of 30 mm (5 mm is used for cantilever
clamping) which can produce a tip displacement of 40 µm with 20 V applied.
Calculate the response speed of this bimorph. Here, the density and the elastic
compliance of the ceramic are ρ = 7.9g/cm3 and s 11E = 16x10-12 m2/N, respectively.

Solution

Considering a certain low applied voltage, type (b) in Fig. 3.14 is preferred to type
(a) in order to obtain a large displacement. Substituting L of Eq. (3.3b) with 25
mm, we get the piezoelectric plate thickness:
__________
t = L √ 3 (d 31 V/ δ)
____________________________________
= (25 x 10 m) √ (3) (300 x 10-12 C/N) (20 V) /(40 x 10 -6 m)
-3
= 530 µm. (P3.5.1)

After cutting the ceramic into plates of 265 µm in thickness, 30 mm in length and 4
- 6 mm in width, the two plates are bonded together after electroding and electrical
poling. The width of the bimorph is usually chosen as w/L < 1/5 so as not to
suppress the magnitude of bending.

The response time is estimated by the resonance period. From Eq. (3.4)

f = 0.161 (t / L2) (ρ s 11E )-1/2 ___________________________

= 0.161 [530x10 m/(25x10 m) ] / v(7.9x103 kg/m3)(16x10-12 m2/N)
-6 -3 2
= 378 Hz (P3.5.2)

Thus, roughly 2.6 msec.

___________________________________________________________________
80 Chapter 3

Example Problem 3.6_________________________________________________

When a piezoceramic plate is bonded to a metallic shim, a unimorph bending device

can be fabricated.25) The tip deflection δ of the unimorph supported in a cantilever
style is given by

δ = (d 31 E) L2 Yc tc / [Ym {t 02 - (t 0 - tm )2} +Yc {(t 0 + t c)2 - t02}] (P3.6.1)

Here E is the electric field applied to the piezoelectric ceramic, d31 , the piezoelectric
constant, L, the length of this unimorph, Yc or Ym, Young's modulus for the
ceramic or the metal, t c or t m is the thickness of each material. In addition, t 0 refers
to the distance between the strain-free neutral plane and the bonding surface, which
is represented as

t0 = [t c tm2(3 t c + 4 t m) Ym + t c4 Yc] / [ 6 t c t m(t c + t m) Ym]. (P3.6.2)

Suppose Yc = Ym, calculate the optimized condition of (t m/tc) to maximize the

deflection δ for the following conditions:

(a) for a fixed ceramic thickness t c,

(b) for a fixed total thickness t c + t m.

Solution

Setting Yc = Ym, the equations become:

δ = (d 31 E) L2 tc / [{t 0 2 - (t 0 - t m) 2} + {(t 0 + t c) 2 - t0 2}], (P3.6.3)

t0 = [t c t m2(3 t c + 4 t m) + t c4] / [ 6 t c t m(t c + t m)]. (P3.6.4)

Substituting t 0 in Eq. (3.6.3) with Eq. (P3.6.4),

δ = (d 31 E) L2 3 t m tc / (t m + t c) 3. (P3.6.5)

Then, the function f(t m) = t m tc / (t m + tc)3 must be maximized for a fixed ceramic
thickness t c (a) or for a fixed total thickness t c + t m = t tot .

(a) df(t m)/dt m = (t c - 2 t m) t c / (t m + t c )4 = 0 (P3.6.6)

Thus, the metal plate thickness should be adjusted to t m = t c / 2. t 0 = t c / 2

(b) df(t m)/dt m = (t tot - 2 t m) / t tot 3 = 0 (P3.6.7)

Both the metal and ceramic plate thickness should be adjusted to t m = tc = ttot / 2.
t0 = t tot / 3.
___________________________________________________________________
Material and Device Designing and Fabrication Processes 81

A composite actuator structure called the "moonie" has been developed to amplify
the pressure sensitivity and the small displacements induced in a piezoelectric
ceramic.26) The moonie has intermediate characteristics between the conventional
multilayer and bimorph actuators; it exhibits an order of magnitude larger
displacement (100 µm) than the multilayer, and much larger generative force (10
kgf) with quicker response (100 µsec) than the bimorph. This device consists of a
thin multilayer ceramic element and two metal plates with a narrow moon-shaped
cavity bonded together [Fig. 3.21(a)]. The moonie with a size of 5mm x 5mm x
2.5mm can generate a 20 µm displacement under 60 V, which is 8 times as large as
the generative displacement of a multilayer of the same size.27) By modifying the
end cap design (cymbal type) as shown in Fig. 3.21(b), twice the displacement can
be obtained. Also the generative displacement is rather uniform, independent of the
position from the center of the end cap.28) Another advantage of the cymbal to the
moonie is its easy fabrication process. One-step punching can make endcaps from a
metal plate.

PZT

(a) Metal Endcap (b)

Fig. 3.21 Structures of a Moonie (a) and a modified Moonie (Cymbal) (b).

(4) Flexible Composites

When needle or plate shaped piezoelectric ceramic bodies are arranged and
embedded in a polymer matrix, functional composites can be fabricated, which
provide enhanced sensitivity by keeping the actuation function. Figure 3.22(a)
shows such a 1 - 3 composite device, where PZT rods are arranged in a polymer in a
two dimensional array.

The simplest composite from a fabrication viewpoint is a 0 - 3 connectivity type,

which is made by dispersing piezoelectric ceramic powders uniformly in a polymer
matrix [Fig. 3.22(b)]. The fabrication processes are classified into a melting and a
rolling method.29) Figure 3.23 shows the flowchart for the fabrication processes.
The powders are mixed with molten polymer in the first method, while the powders
are rolled into a polymer using a hot-roller in the second method. The fabrication
processes for 1 - 3 composites are introduced in Chapter 10, Section 10.3.
82 Chapter 3

Piezoceramic fiber (Phase 1) Piezoceramic powder (Phase 1)

(a) Polymer matrix (Phase 2)

(b)
Fig. 3.22 PZT: polymer composites: (a) 1 - 3 connectivity and (b) 0 - 3
connectivity.

Melting Method Rolling Method

Melting polymer Ferroelectric powder Polymer Ferroelectric powder

(Ball milling) (Rolling)

(Film casting) (Calendering)

Composite film Composite sheet

Electroding

Poling

Piezoelectric component

Fig. 3.23 Fabrication process for PZT: polymer composites.

(5) Thin/Thick Films

Techniques for fabrication of oxide thin films are classified into physical and
chemical processes:
Material and Device Designing and Fabrication Processes 83

(a) Physical Processes

Electron beam evaporation
RF sputtering, DC sputtering
Ion beam sputtering
Ion plating
(b) Chemical Processes
Sol-gel method (dipping, spin coating etc.)
Chemical vapor deposition (CVD)
MOCVD
Liquid phase epitaxy, melting epitaxy, capillary epitaxy etc.

Sputtering has been most commonly used for ferroelectric thin films such as
LiNbO3 , PLZT,30) and PbTiO3 .31) Figure 3.24 shows the principle of a
magnetron sputtering apparatus. Heavy Ar plasma ions bombard the cathode
(target) and eject its atoms. These atoms are deposited uniformly on the substrate in
an evacuated enclosure. The sol-gel technique has also been employed for
processing PZT films.32) Applications of thin film ferroelectrics include memories,
surface acoustic wave devices, piezo sensors and micro-mechatronic devices.

Heater
Ar
Holder
Substrate
Gas Vacuum pump
O2 Plasma Magnetic
field

N S N Target
S N S

High f power supply

Fig. 3.24 Principle of a magnetron sputtering apparatus.

The thin film structure is inevitably affected by two significant parameters:

(1) Stress from the substrate -- Tensile or compressive stress is generated due to
thermal expansion mismatch between the film and the substrate, leading to
sometimes a higher coercive field for domain reorientation.
84 Chapter 3

(2) Thickness dependence of ferroelectricity -- Similar to a powder sample,

there may exist a critical film thickness below which the ferroelectricity
would disappear. No research has been made from this viewpoint.

3.4 GRAIN SIZE EFFECT ON FERROELECTRICITY

In recent years, so-called fine ceramics, ceramics which possess controlled-grain
size or very fine grains, have been manufactured. However, not many of their
properties, such as ferroelectric and piezoelectric properties, have been measured.
To understand the grain size dependence of the dielectric properties, we must
consider two size regions: The µm range in which a multiple domain state becomes
a mono-domain state, and sub-µm range in which the ferroelectricity becomes
destabilized.

Figure 3.25 shows the transverse field-induced strains of 0.8 at.% Dy doped fine
grain ceramic BaTiO 3 (grain diameter around 1.5 µm) and of the undoped coarse
grain ceramic (50 µm), as reported by Yamaji. 33) As the grains become finer,
under the same electric field, the absolute value of the strain decreases and the
hysteresis becomes smaller. This is explained by the increase in coercive field for
90o domain rotation with decreasing grain size. The grain boundaries (with many
dislocations on the grain boundary) "pin" the domain walls and do not allow them to
move easily. Also the decrease of grain size seems to make the phase transition of
the crystal much more diffuse. Figure 3.26 shows the temperature dependence of
the piezoelectric coefficient d33. Although the absolute value of d33 decreases in
the Dy-doped sample, the temperature dependence is remarkably improved for
practical applications. It should be noted that Yamaji's experiment cannot separate
the effect due to intrinsic grain size from that due to dopants.

Takasu et al. studied the effects of grain size on PLZT.34) They obtained PLZT
(9/65/35) powders by coprecipitation. Various grain sizes were prepared by hot-
pressing and by changing sintering periods. PLZT (9/65/35) shows significant
dielectric relaxation (frequency dependence of the permittivity) below the Curie
point of about 80o C, and the dielectric constant tends to be higher at lower
frequency. Figure 3.27 shows the dependence of the peak dielectric constant on
grain size. For grain size larger than 1.7 µm, the dielectric constant decreases with
decreasing grain size. Below 1.7 µm, the dielectric constant increases rapidly.
Figure 3.28 shows the dependence of the longitudinal field-induced strain on the
grain size. As the grain size becomes smaller, the maximum strain decreases
monotonically. However, when the grain size becomes less than 1.7 µm, the
hysteresis is reduced. This behavior can be explained as follows: with decreasing
grain size, (anti)ferroelectric (ferroelastic) domain walls become difficult to form in
the grain, and the domain rotation contribution to the strain becomes smaller
(multidomain - monodomain transition model). The critical size is about 1.7 µm.
Material and Device Designing and Fabrication Processes 85

However, note that the domain size is not constant, but is dependent on the grain
size, and that in general the domain size decreases with decreasing grain size.

Fig. 3.25 Field induced strain curves in Dy doped and undoped BaTiO 3 ceramics.

Fig. 3.26 Temperature dependence of the piezoelectric d33 in Dy doped and

undoped BaTiO3 ceramics.
86 Chapter 3

Fig. 3.27 Grain size dependence of the peak permittivity in PLZT 9/65/35.

Fig. 3.28 Grain size dependence of the induced strain in PLZT ceramics.
Material and Device Designing and Fabrication Processes 87

Regarding the much smaller grain/particle size range, Uchino et al. reported
previously a number of informative experiments. Figure 3.29 shows the degree of
tetragonality (c/a) change with particle size in pure BaTiO 3 at room temperature.35)
The c/a value decreases drastically below 0.2 µm and becomes 1 (i.e., cubic!) at
0.12 µm, defined as a critical particle size. Figure 3.30 shows the temperature
dependence of the c/a ratio for various particle size powders. This demonstrates the
correlation between the critical particle size and the Curie temperature, which
decreases with decreasing particle size.

Fig. 3.29 Particle size dependence of the tetragonality in BaTiO3 at room

temperature.

Fig. 3.30 Temperature dependence of the tetragonality in BaTiO3 for various

particle size samples.
88 Chapter 3

Fig. 3.31 Relationship between the critical particle size Dcrit or the critical
hydrostatic pressure and the phase transition temperature.

A similar critical particle size was also reported in (Ba,Sr)TiO3 ,36) (Ba,Pb)TiO3 37)
and for antiferroelectric PbZrO3 .36) Figure 3.31 shows the relationship between the
critical particle size Dcrit and the Curie (or Neel) temperature, TC, for these
materials. An important empirical rule was obtained:

Dcrit . (T C - Room Temp.) = Constant. (3.5)

Although there have been many reports on the critical grain/particle size, it seems to
be true that there is a kind of critical size below which the ferroelectricity disappears
(i. e. the crystal becomes cubic). No satisfactory explanation has yet been
presented. One possible explanation is based on a hydrostatic pressure model. In
general, the ferroelectric transition temperature decreases sharply (50o C /MPa) with
increasing hydrostatic pressure. Systematic data can be found in Samara's papers,
who used external hydrostatic oil pressure.38) The effective surface tension γ on a
fine particle causes a hydrostatic pressure p intrinsically:39)

p = 2γ/ R (R: particle radius) (3.6)

From the critical particle size Dcrit (= 2 Rcrit ) and the critical hydrostatic pressure
p crit , above which the cubic structure is realized at room temperature, we can
calculate the effective surface tension γ as listed in Table 3.3. The γ value is almost
constant for all perovskite ferroelectrics. The γ value, 50 times larger than that of
non-polar oxides, may be due to the additional energy from the surface charge
contribution, and from a crystallographically different phase on the surface (core-
shell model).
Material and Device Designing and Fabrication Processes 89

Table 3.3 Critical particle size, critical hydrostatic pressure and the surface tension
energy for various perovskites. (P = 2 γ / R)
________________________________________________________________
Material Curie Temp (oC) 2Rcrit (µm) Pcrit (GPa) γ (N/m)
________________________________________________________________
Ba0.9Sr0.1TiO 3 95 0.19 1.2 57
BaTiO 3 125 0.12 1.8 54
Ba0.85Pb0.15TiO 3 180 0.08 2.9 58
Ba0.5Pb0.5TiO 3 330 0.032 6.2 50
PbTiO 3 500 0.02 10 50
________________________________________________________________

3.5 FERROELECTRIC DOMAIN CONTRIBUTIONS

(1) Ferroelectric Domain Reorientation

In the phenomenological theory of the last chapter, it was assumed that the object
materials are mono-domain single crystals and the application of an electric field
does not change their state. However, in ordinary piezoelectric ceramics, these
assumptions do not strictly hold. An actual device material sometimes shows a
multiple domain structure even in a single crystal form, and a much more
complicated configuration in a polycrystalline ceramic. Figure 3.32 shows the
electric field-induced strains, one parallel and the other perpendicular to the field, in
a composition 7/62/38 of the well-known piezoelectric ceramics (Pb,La)(Zr,Ti)O3
(PLZT). In a cycle with a small maximum electric field, the field -induced strain
curve appears nearly linear, and can be called the "inverse piezoelectric effect."
However, with the increase of the maximum electric field, the hysteresis becomes
larger, and finally transforms to a symmetric butterfly shape when the electric field
exceeds a certain critical value (this is called coercive field). This is caused by the
change of polarity in ferroelectric domains under applied electric field, inducing a
different state of polarization. Strictly speaking, this PLZT includes two stages to
be taken into account:

1) domain reorientations in each grain,

2) the polycrystalline state (a complex of randomly oriented tiny crystals).

Figure 3.33 shows the domain reorientation observed in BaTiO3 single crystals
under an external field. The 90o domain walls disappear independently of each
other with increasing electric field finally to become a mono-domain state
throughout the entire crystal. However, in a polycrystalline specimen, the domain
wall motion is suppressed by the grain boundary, and a purely mono-domain state
can not be achieved.
90 Chapter 3

Fig. 3.32 Electric field-induced strain in a piezoelectric ceramic PLZT 7/62/38.

Fro m a to e, the maximum electric field increases.

Figure 3.34 shows schematically the domain reorientation in a polycrystalline body.

Suppose a sample is initially poled along the negative direction, and an external
electric field is applied in the positive direction, the crystal should shrink at first
with an increase of the field because the field is opposite in direction to the
remanent polarization. Then, the deformation reaches a minimum at a certain field
Ec (coercive field), where the polarization reversal begins in each grain. Above Ec
the crystal starts to expand until E = Emax. Near Emax, all the reversible
polarizations have been reversed, and the crystal displays "piezoelectric" behavior
once again with small hysteresis. In the process of decreasing the electric field,
there is no need for polarization reversal except in some unstable domains with
internal stresses. Until the field is reduced to zero, the strain decreases
monotonically. The final state at E = 0 is equivalent to the initial state but with the
polarization direction reversed. It can be said that the crystal is poled in the positive
direction. It is interesting to point out that the ratio σ of the transverse contraction
strain to the longitudinal expansion strain (Poisson's ratio) has similar values for the
perovskite-type piezoelectric ceramics, and is about 0.3.
Material and Device Designing and Fabrication Processes 91

(1) (2) (3)

Fig. 3.33 Domain reorientation in BaTiO 3. E direction: <---.

Strain x
4 3
1. E = 0 2. E = Ec 1
5

Field E

3. E = Emax

4. E = Ec
5. E = 0

Fig. 3.34 Schematic depiction of the strain change in a ferroelectric ceramic

associated with the polarization reorientation.
92 Chapter 3

Example Problem 3.7_________________________________________________

BaTiO3 exhibits a rhombohedral crystal symmetry at liquid nitrogen temperature

(i.e., -196o C) and the angle distorted from the cubic structure is not very large
(about 1 degree). Calculate all the possible angles between the two non-180o
domain walls.

Solution

The polarization direction of barium titanate at the low temperature is in the [111]
or its equivalent axis of a perovskite cell. Let us consider three typical directions
_ __
[111], [111] and [111] as illustrated in Fig. 3.35. The domain wall plane must be
normal to one of the following and their equivalent axes.
_
[111] - [111] = [002],
__
[111] - [111] = [022]. (P3.7.1)

Thus, the angle between two of the non-180o domain walls is calculated as follows:
_
1) (002)/(200), (022)/(022), (002)/(220)
(002). (200) = 0
= 2. 2 cosθ, θ = 90o
_
2) (022)/(220), (022)/(220)
(022). (220) = 4 or - 4
= 2√2. 2√2 cosθ, θ = 60o or 120o
_
3) (002)/(022), (002)/(022)
(002). (022) = 4 or - 4
= 2. 2√2 cosθ, θ = 45o or 135o

3 [111]

1
_
[111]

__
[111]
Fig. 3.35 Polarization directions of the rhombohedral barium titanate.
___________________________________________________________________
Material and Device Designing and Fabrication Processes 93

In the following subsections, the Uchida-Ikeda theory which treats domain

reorientation in ceramics quantitatively, and the relationship between the crystal
structure and the coercive field will be introduced.

(2) Uchida-Ikeda Model

Let us take an example, barium titanate, which has a tetragonal symmetry at room
temperature. X-ray diffraction indicates a slight elongation along the [001] direction
of a cubic perovskite cell with c/a = 1.01. Therefore, when an electric field is
applied on a monodomain single crystal along the a axis, the 90o domain
reorientation from an a to c domain is induced, subsequently leading to a strain of 1
% along the field direction. However, the situation is much more complicated in the
case of a polycrystalline sample. Uchida and Ikeda treated this problem statistically,
assuming a grain (small crystallite) being randomly oriented crystallographical-
ly.40,41)

In an unpoled homogeneous polycrystal, there must be no remnant polarization. Let

this state be the basis for zero strain. If an electric field E3 (in the 3-direction) is
applied on this sample, a polarization P3 will be induced, also the strains x1 , x2 and
x3 , where x1 = x2 = - σ x3 (σ: Poisson's ratio). Let the spontaneous polarization
and the principal strain of the individual crystallites be PS and SS, respectively. For
uniaxial crystals such as those with tetragonal and rhombohedral symmetries, the
principal strain SS is given in the direction of P3 :
SS = (c/a) - 1 (tetragonal crystal) (3.7)
SS = (3/2)(π/2 - α) = (3/2) δ (rhombohedral crystal) (3.8)

A remark should be made here on the distinction between the principal strain and
the spontaneous strain (there was some confusion in the original papers). Taking
BaTiO3 as an example, the principal strain is SS = 0.01, but the spontaneous strains
should be defined as
x3,S = (c/a0 ) - 1,
x1,S = x2, S = (a/a0) - 1 , (3.9)
where a0 is the lattice parameter of the paraelectric phase, and have the values
0.0075 and - 0.0025, respectively.

First, letting θ be the angle between the direction of the spontaneous polarization PS
of a microscopic volume ∆v in a ceramic and the direction of the electric field E3,
then the polarization P 3 is given by
____
P3 = ∫ PS cos θ dv / ∫ dv = PS cos θ, (3.10)
____
where cos θ is the average value of cos θ in all the microscopic volumes of the
ceramic. The average strain is also calculated from the orientation of the strain
ellipsoid:
94 Chapter 3

_____
x3 = S S [ ∫ cos θ dv / ∫ dv - 1/3 ] = S S (cos 2 θ - 1/3)
2 (3.11)

This model, in which the microscopic regions with spontaneous strain change only
their orientations, accompanied by no volume change, provides σ = 0.5 and
x1 = x2 = - x3 / 2 (3.12)

However, there is a serious discrepancy with experimental data.

Next, in order to find the trend for the change in induced strain with an applied
electric field, the relationship between θ and E3 has to be known. Uchida et al.
analyzed this problem by introducing a characteristic angle θ90 for non-180o
domain reorientations; in tetragonal crystals, 90o reorientation and in rhombohedral
crystals, 71o and 109o reorientations occur. But in order to simplify the explanation,
all reorientations are being represented by the former. Suppose a 90o domain
rotation occurs in a small region dv in a ceramic, and as a result, the orientation of
dv becomes θ. These authors assumed that there exists a characteristic angle θ90,
such that if θ < θ90 , a 90o rotation of the small region can occur, whereas if θ >
θ90 , no rotation will occur, and the region remains in its intial state. For a given
θ90 , which corresponds to a certain E3, by integrating Eq. (3.11) over a volume dv
for which θ < θ 90 is satisfied, the induced strains x3, x1 and x2 can be obtained as a
function of θ90. Figure 3.36 shows the relationship between θ90 and (cos2 θ - 1/3),
and Fig. 3.37(a) shows the measured values of induced strain in rhombohedral PZT
ceramics. Comparing the two θ90 and E3 figures reveals the relationship between
θ90 and E3 [Fig.3.37(b)]. It is apparent that pronounced hysteresis also appears in
the θ90 versus E3 curve.

Furthermore, by finding the polarization P3 and the field-induced strain x3 (or x1)
as a function of the electric field E3 , it is possible to estimate the volume in which a
180o reversal or a 90o rotation occurred. This is because the 180o domain reversal
does not contribute to the induced strain, only the 90o rotation does, whereas the
180o domain reversal contributes mainly to the polarization. It is shown
schematically in Fig. 3.38 that with the application of an electric field the 180o
reversal occurs rapidly whereas the 90o rotation occurs slowly.42) It is notable that
at G in the figure, there remains some polarization while the induced strain is zero,
at H the polarizations from the 180o and 90o reorientations cancel each other and
become zero, but the strain is not at its minimum. Generally in such a case, a plot
of the induced strain x3 versus polarization P3 shows large hysteresis (Fig. 3.39).43)
However, for materials whose polarization is dominated by non-180o domain
rotations, the hysteresis in the x versus P plot should hardly be observed. Such is
the case for the low temperature phase of Pb(Mg1/3Nb 2/3)O 3 which is shown in
Fig. 3.40(b).44)
Material and Device Designing and Fabrication Processes 95

_____
Fig. 3.36 Relation between θ90 and (cos2 θ - 1/3), where θ90 is a critical angle
relating to the non 180o domain reorientation and (cos2 θ - 1/3) is proportional to
the field -induced strain.

Fig. 3.37 Transverse strain x1 versus field in Pb(Zr 0.57Ti0.43)O 3 (a) and the
calculated θ90 - E3 relation (b). The measurement was done at 30o C.
96 Chapter 3

180 o Reversal

K J I H G (F,A) B C D E

90o
Reorientation

Fig. 3.38 Electric field dependence of the domain volume fraction of 180° reversal
(a) and of 90o reorientation (b). Notice the deviation of the zero fraction points, I
and G, between 180o and 90o .

Finally, the saturation values of polarization and strain of a ceramic under high
electric field are summarized:
_________________________________________________________

Tetragonal P3 --> 0.831 PS x3 --> 0.368 SS

Rhombohedral P3 --> 0.861 PS x3 --> 0.424 SS
_________________________________________________________

Fig. 3.39 P 3 - x3 relation in a tetragonal PLZT ceramic (6.25/50/50).

Material and Device Designing and Fabrication Processes 97

Fig. 3.40(a) P3 -E3, x1 - E3, (b) P3 - x1 relations in a rhombohedral crystal of

Pb(Mg1/3Nb 2/3)O3 at - 110o C.

(3) Crystal Structure and Coercive Field

In the previous subsection, a comparison was made between the tetragonal and
rhombohedral systems from the viewpoint of saturation values for ceramics under a
sufficiently large electric field. In this subsection the difference between the two
systems is illustrated with respect to the more useful quantity, the coercive field.

Table 3.4 summarizes the principal strain SS, spontaneous polarization PS, volume
percent of reoriented domains γ 90 , and coercive field Ec of PLZT ceramics which
were reported by Schmidt.43) What can be understood from this table is that in
comparison with the compositions with tetragonal symmetry, those with
rhombohedral symmetry have smaller principal strains, hence easier domain
rotation (larger γ 90 ) and smaller coercive field Ec . An expression for Ec has been
derived,43,45)

Ec = α Y S S2 γ 90 /PS (3.13)

in which Y is the Young's mudulus, and α is a factor which takes into account the
difference in domain orientation between neighboring grains, and has values of 0.1
and 0.074 for the tetragonal and rhombohedral systems, respectively.
98 Chapter 3

Table 3.4 Principal strain, spontaneous polarization, reoriented volume fraction

and coercive field in tetragonal and rhombohedral PLZT ceramics.

Crystal PLZT Principal Spontaneous Oriented Coercive Calculated

symmetry specimen strain S s polarization volume field E c Ec
(%) (µC/cm 2) fraction (%) (kV/cm) (kV/cm)

25/50/50 2.4 71 22 18 17.8

25/52/48 2.2 72 28 14.7 18.8
Tetra-
5/50/50 2.16 65 18 16.3 13
gonal
5/52/48 1.96 64.5 23 14.8 13.7
5/54/46 1.68 65 30 11.7 13

25/58/42 0.732 56.5 86.5 8.2 7

Rhombo- 25/60/40 0.74 58.5 78.5 7.6 5.4
hedral 6/65/35 0.65 45 85 5.6 5.9
6.25/60/40 0.61 49 85 5.7 4.8

Example Problem 3.8_________________________________________________

The crystal orientation dependencies of the dielectric constant ε3 and piezoelectric

constant d33 of a tetragonal PZT are schematically illustrated in Fig. 3.41. Let us
consider a uniformly oriented polycrystalline sample. Discuss the change in the ε3
and d 33 before and after poling.

[001]
PIEZOELECTRIC
CONSTANT dmax =200
PERMITTIVITY
pC/N
[001]

εmax =500
[100] [100]

(a) (b)

Fig. 3.41 Crystal orientation dependencies of the dielectric and piezoelectric

constants of a tetragonal PZT.
Material and Device Designing and Fabrication Processes 99

Solution

Before poling, because of a uniform crystalline distribution, the dielectric constant

should have an intermediate value between εmin and εmax, and the piezoelectric
constant should be zero.

Electric poling orients the polarization along the z-axis, thus, the permittivity
approaches εz, leading to a decrease in permittivity after poling. On the contrary,
the piezoelectric constant should increase monotonically with increasing poling
field, finally exhibiting a saturation of d33 above a certain poling field (close to the
coercive field).
___________________________________________________________________

CHAPTER ESSENTIALS_________________________________
1. Doping effects on ferroelectricity in PZT:
Acceptor ---> domain pinning ---> "hard" piezoelectric
Donor ---> Pb deficiency compensation ---> "soft" piezoelectric

2. Preparation of ceramic powders:

oxide-mixing technique
coprecipitation
alkoxide hydrolysis

3. Device designs:
Single disk
Multilayer
Unimorph/bimorph
Moonie/cymbal
Flexible composite
Thin/thick film

4. Comparison between multilayers and bimorphs:

1. The multilayer type does not exhibit large displacements, but has
advantages in generative force, response speed, life time and
electromechanical coupling k33 .
2. The bimorph type exhibits large displacements, but shows disadvantages in
generative force, response speed, life time and the electromechanical
coupling keff.

5. Tip displacement δ in a bimorph under a one-end clamp condition (cantilever):

δ = (3/2) d 31 (L2/ t 2 ) V or
δ = 3 d 31 (L2/ t 2) V
(according to the structure)
100 Chapter 3

Fundamental resonance frequency is determined by the total thickness t :

f = 0.16 (t/ L2 ) (ρ s 11E)-1/2 (for both structures)

6. Grain size dependence of ferroelectricity:

There seems to be a critical particle/grain size (sub micrometer), below which
the material ceases to be ferroelectric (i.e., it becomes paraelectric).

Smaller grain --> smaller polarization and strain, smaller hysteresis

--> larger mechanical strength

7. Polarization and electric field-induced strain in a polycrystalline body:

Tetragonal P3 --> 0.831 P S, x3 --> 0.368 S S
Rhombohedral P3 --> 0.861 P S, x3 --> 0.424 S S
Coercive field: Tetragonal > Rhombohedral
___________________________________________________________________

CHAPTER PROBLEMS
3.1 Using a PZT based ceramic plate with a piezoelectric constant, d31 = - 300
pC/N, size, 25 mm x 5 mm x 0.5 mm, and a phosphor bronze plate (high
Qm material!), design a unimorph with a total active length of 25 mm
which can produce the maximum tip displacement under an application of
100 V. Determine the optimized thickness of a phosphor bronze plate, then
calculate the maximum displacement. Here, the density and the elastic
compliance of the ceramic are ρ = 7.9 g/cm3 and s11E = 16 x 10-12 m2 /N,
respectively. Search for the necessary data (Young's modulus etc.) of
phosphor bronze. (Refer to Example Problem 3.6.)

3.2 Describe the difference between "silver paste" and "silver ink."

3.3 The bend curvature 1/R of a monolithic piezoelectric plate under an

electric field E is given by
to/2 to/2
1/R = [∫ Y d 31(Z)E3 (Z) Z w dZ] / [∫ Y Z2 w dZ] ,
-to/2 -to/2
where t0 and w are the thickness and the width of the plate, respectively, Y
the elastic modulus, and d 31 the piezoelectric constant. The coordinate axis
Z (plate thickness direction) has its origin at the center of the plate. When
Y and w are constant,
to/2
3
1/R = - (12/ t 0 ) [∫ d 31 (Z)E3 (Z) Z dZ] .
-to/2
If this piezoelectric plate is prepared to be a slightly n-type semiconductor,
the external voltage on the surface electrodes generates a non-uniform
Material and Device Designing and Fabrication Processes 101

distribution of the internal electric field. Like a Schottky contact, suppose

the field distribution E(Z) and the barrier thickness tb are given by

E(Z) = (q Nd / ε0 ε) (Z - t0/2 + t b ) (t 0/2 - tb < Z < t 0/2, only cathode side!)

and
tb = [(2 ε0 ε/q Nd ) (φ0 + V)] 1/2,

respectively, where q is the electron charge, Nd the donor density,

ε0 vacuum permittivity, and ε is the relative permittivity of the
monomorph. Verify that this plate bends and the curvature is given by

1/R = - β d 31 tb 2 (3 t 0 - 2 t b ) / t 03
or
1/R = - β d 31 tb 2 (2 t 0 - tb ) / t 03 ,

corresponding to the completely poled [d 31(Z) = d31 (constant)] or the

gradually poled [d 31(Z) = d31 (Z - t0/2 + tb ) / tb )] case. This bending
device is called a "monomorph." [Refer to K. Uchino et al.: Jpn. J. Appl.
Phys. 26, 1046 (1987)]

3.4 To understand the reason why the ferroelectricity will disappear with
decreasing particle size, we can consider the energy fluctuation for a nano
size ferroelectric particle as follows (Refer to Chap. Problem 1.1):
Consider a 1-dimensional finite chain of two kinds of ions +q and -q,
arranged alternately with a distance of a (see the following figure). A nano
size crystal grows gradually, starting from a single positive ion, and adding
a pair of negative or positive ions, thus keeping a crystal size 2na (n = 1, 2,
3, ... ). With increasing crystal size, the crystal Coulomb energy will be
changed as:

U 1 = (2/4πε0 ε) [- (q 2 / a)]
U 2 = (2/4πε0 ε) [- (q 2 / a) + (q 2 / 2a)]
U 3 = (2/4πε0 ε) [- (q 2 / a) + (q 2 / 2a) - (q 2 / 3a)]
...........
Un = (2/4πε0 ε) [- (q 2 / a) + (q 2 / 2a) - (q 2 / 3a) + ... ]

Calculate the minimum crystal size (2na) which maintains the crystal
energy with less than ±10 % fluctuation, even when adding or subtracting a
pair of ions (i.e., n+ 1 or n- 1). If the original crystal Coulomb energy is
not stabilized to less than this degree of fluctuation, we cannot expect the
ferroelectric phase transition to occur as a cooperative phenomenon.
102 Chapter 3

-q +q -q +q -q +q -q
(a)
-3a -2a -a 0 a +2a +3a

1 3 5 7 9 n
(b)

Madelung energy

- (2/4 πε 0)(q2/a)

1-dimensional finite chain of two kinds of ions +q and -q.

3.5 Barium titanate exhibits a tetragonal crystal symmetry at room temperature

and the distortion from the cubic structure is not very large (c/a = 1.01).
Calculate all the possible angles between the two non-180° domain walls.

3.6 In calculating Eqs. (3.10) and (3.11), the volume element dv is given by dv
= 2πr2 dr . sinθ dθ. Using this dv, calculate ∫ dv, ∫ cosθ dv and ∫ cos 2θ dv,
when the polarization is uniformly distributed with respect to θ.

REFERENCES
1) B. Jaffe, W. R. Cook and H. Jaffe: Piezoelectric Ceramics, p.142, Academic Press,
NY (1971).
2) K. Uchino and S. Nomura: Jpn. J. Appl. Phys. 18, 1493 (1979).
3) K. Abe, O. Furukawa and H. Inagawa: Ferroelectrics 87, 55 (1988).
4) A. Hagimura and K. Uchino: Ferroelectrics, 93, 373 (1989).
5) K. Uchino, H. Negishi and T. Hirose: Jpn. J. Appl. Phys., 28, Suppl. 28-2, 47
(1989).
6) S. Hirose, Y. Yamayoshi, M. Taga and H. Shimizu: Jpn. J. Appl. Phys., 30, Suppl.
30-1, 1117 (1991).
7) S. Takahashi and S. Hirose: Jpn. J. Appl. Phys., 32, Pt. 1, No.5B, 2422 (1993).
8) K. Uchino, J. Zheng, A. Joshi, Y. H. Chen, S. Yoshikawa, S. Hirose, S. Takahashi
and J. W. C. de Vries: J. Electroceramics, 2, 33 (1998).
9) S. Hirose, M. Aoyagi, Y. Tomikawa, S. Takahashi and K. Uchino: Proc.
Ultrasonics Int'l. '95, Edinburgh, p.184 (1995).
Material and Device Designing and Fabrication Processes 103

10) Kato: Fine Ceramics Technology, Vol.3 Fabrication Technology of Ceramic

Powder and Its Future, p.166, Industry Research Center, Japan (1983).
11) M. Lejeune and J. P. Boilot: Ferroelectrics 54, 191 (1984).
12) S. L. Swartz, T. R. Shrout, W. A. Schulze and L. E. Cross: J. Amer. Ceram. Soc.
67, 311 (1984).
13) Tanada, Yamamura, Shirasaki: Abstract 22nd Jpn. Ceram. Soc. Fundamental Div.
3B5, p.81 (1984).
14) Ozaki: Electronic Ceramics 13, Summer, p.26 (1982).
15) Kakegawa, Mohri, Imai, Shirasaki and Tekahashi: Abstract 21st Jpn. Ceram. Soc.
Fundamental Div. 2C6, p.100 (1983).
16) H. Abe: Recrystallization, Mater. Sci. Series 2, Kyoritsu Pub., Tokyo (1969).
17) K. Uchino and T. Takasu: Inspec. 10, 29 (1986).
18) A. Yamaji, Y. Enomoto, E. Kinoshita and T. Tanaka: Proc. 1st Mtg. Ferroelectric
Mater. & Appl. p.269, Kyoto (1977).
19) K. Nakamura, H. Ando and H. Shimizu: Jpn. J. Appl. Phys. 26, Suppl.26-2, 198
(1987).
20) J. Kuwata, K. Uchino and S. Nomura: Ferroelectrics 37, 579 (1981).
21) J. Kuwata, K. Uchino and S. Nomura: Jpn. J. Appl. Phys. 21(9), 1298 (1982).
22) J. Zheng, S. Takahashi, S. Yoshikawa, K. Uchino and J. W. C. de Vries: J. Amer.
Ceram. Soc. 79, 3193 (1996).
23) K. Nagai and T. Konno Edit.: Electromechanical Vibrators and Their Applications,
Corona Pub. (1974).
24) K. Uchino: Piezoelectric Actuators and Ultrasonic Motors, Kluwer Academic
Publishers, MA, p.241 (1997).
25) K. Abe, K. Uchino and S. Nomura: Jpn. J. Appl. Phys. 21, L408 (1982).
26) Y. Sugawara, K. Onitsuka, S. Yoshikawa, Q. C. Xu, R. E. Newnham and K.
Uchino: J. Amer. Ceram. Soc. 75, 996 (1992).
27) H. Goto, K. Imanaka and K. Uchino: Ultrasonic Techno 5,48 (1992).
28) A. Dogan: Ph. D. Thesis, Penn State University (1994).
29) Kitayama: Ceramics 14, 209 (1979).
30) M. Ishida et al.: Appl. Phys. Lett. 31, 433 (1977).
31) M. Okuyama et al.: Ferroelectrics 33, 235 (1981).
32) S. K. Dey and R. Zuleeg: Ferroelectrics 108, 37 (1990).
33) A. Yamaji, Y. Enomoto, K. Kinoshita and T. Tanaka: Proc. 1st Mtg. Ferroelectric
Mater. & Appl., Kyoto, p.269 (1977).
34) K. Uchino and T. Takasu: Inspec. 10, 29 (1986).
35) K. Uchino, E. Sadanaga and T. Hirose: J. Amer. Ceram. Soc. 72, 1555 (1989).
36) T. Yamakawa and K. Uchino: Proc. Int'l. Symp. Appl. Ferroelectrics '90, p.610
(1991).
37) K. Saegusa et al.: Amer. Ceram. Soc., 91th Ann. Mtg. (1989).
38) G. A. Samara: Ferroelectrics, 2, 277 (1971).
39) K. Uchino, E. Sadanaga, K. Oonishi and H. Yamamura: Ceramic Trans. 8, Ceramic
Dielectrics, 107 (1990).
40) N. Uchida and T. Ikeda: Jpn. J. Appl. Phys. 6, 1079 (1967).
41) N. Uchida: Rev. Elect. Commun. Lab. 16, 403 (1968).
42) N. Uchida and T. Ikeda: Jpn. J. Appl. Phys. 4, 867 (1965).
43) N. A. Schmidt: Ferroelectrics 31 , 105 (1981).
44) J. Kuwata, K. Uchino and S. Nomura: Jpn. J. Appl. Phys. 19, 2099 (1980).
45) P. Gerthsen and G. Kruger: Ferroelectrics 11, 489 (1976).
104 Chapter 3

<< Do You Know ? >>

Bimorph Family

You should distinguish the device terminology: monomorph, unimorph, bimorph

and multimorph. All are bending devices, however, the definitions are:

monomorph - single actuator ceramic plate

unimorph - single actuator plate & an elastic shim
bimorph - double actuator plates bonded together with or without
an
elastic shim
multimorph - multiple actuator plates bonded together with or without
multiple elastic shims

(a) Monomorph

(b)) Unimorph

(c) Bimorph

(d) Multimorph