Crystallization kinetics of amorphous lead zirconate titanate thin

lms in a microwave magnetic eld

Y.N. Chen, Z.J. Wang

, T. Yang, Z.D. Zhang

Shenyang National Laboratory for Materials Science, Institute of Metal Research, Chinese Academy of Sciences, 72 Wenhua Road, Shenyang 110016, China
Received 16 November 2013; received in revised form 25 February 2014; accepted 4 March 2014
Available online 25 March 2014
Abstract
We investigated the crystallization process and kinetics of lead zirconate titanate (PZT) thin lms fabricated in the magnetic eld of
microwave irradiation. The PZT thin lms were prepared by the solgel method and crystallized by microwave irradiation at 2.45 GHz.
X-ray diraction was used to identify the phases and to determine the volume fraction of the perovskite phase transformed during crys-
tallization. Transmission electron microscopy gave information on nucleation, growth and grain structure. We also discussed how the
crystallization and phase transformations correlated to the ferroelectric properties of the resultant lms. We found that an intermediate
phase formed during initial crystallization; it had a perovskite-like crystal structure, but it had a smaller lattice constant than perovskite
PZT and contained more Ti. This intermediate phase acted as a nucleation site for the perovskite PZT, which grew with a columnar grain
structure into the pyrochlore matrix throughout the lm. Using Avramis model, we found the eective activation energy for crystalli-
zation of the PZT lms by microwave irradiation to be 214 kJ mol
1
, lower than the activation energy for crystallization by conven-
tional thermal processes. These results show that microwave irradiation indeed aected the crystallization of amorphous PZT thin lms
dierently than conventional annealing.
2014 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
Keywords: PZT thin lm; Solgel; Microwave irradiation; Crystallization; Kinetics
1. Introduction
Lead zirconate titanate (Pb(Zr
x
Ti
1x
)O
3
; PZT) thin
lms have attracted interest for many applications in non-
volatile ferroelectric random-access memory [1] and micro-
electromechanical systems (MEMS) [2], such as atomic
force microscopy cantilevers [3] and MEMS shunt switches
[4]. Many techniques for depositing ferroelectric PZT thin
lms have been studied, including physical methods, such
as magnetron sputtering [5] and pulsed laser deposition
[6], as well as chemical methods, such as metalorganic
chemical vapor deposition [7] and solgel methods [8,9].
To fabricate well-crystallized PZT lms with compositions
close to the morphotropic phase boundary (Zr/Ti = 52/48),
substrate temperatures or post-deposition annealing tem-
peratures of 600750 C are often required. However,
treatment at such high temperatures causes many adverse
eects, including interdiusion between the lms and sub-
strate as well as evaporation of lead and lead oxide from
the surface, changing the stoichiometry [10,11]. Thus, fab-
ricating PZT lms at lower treatment temperatures and
times would be benecial.
The solgel method is an eective chemical technique for
fabricating PZT thin lms and has been widely studied to
achieve lower processing temperatures. This method typi-
cally involves introducing a seed layer or template layer
and modulating the Zr/Ti ratio and Pb excess in the
precursor solution. For example, by using a seed layer of
lead titanate (PbTiO
3
), Kwok and Desu decreased the
http://dx.doi.org/10.1016/j.actamat.2014.03.009
1359-6454/ 2014 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

Corresponding author.
E-mail address: wangzj@imr.ac.cn (Z.J. Wang).
www.elsevier.com/locate/actamat
Available online at www.sciencedirect.com
ScienceDirect
Acta Materialia 71 (2014) 110
crystallization temperature needed to produce perovskite
PZT (Zr/Ti = 53/47) lms to 550 C for 15 min [12]. Huang
et al. [13] reported crystallization of PZT (30/70) lms at
440 C for 100 min, attributing the lower temperature to
the formation of a Pt
x
Pb interlayer. Suzuki et al. reported
crystallization of pure perovskite PZT (53/47) lms on Pt/
Ti/SiO
2
/Si at 500 C for 2 h by using a PbTiO
3
seeding
layer with 50 mol.% excess PbO [14]. Then, by using seeded
diphasic solgel precursors, Wu et al. reported crystalliza-
tion of pure perovskite PZT (52/48) lms at 410 C for
30 h and at 550 C for 30 min [15]. More recently, by using
a low pyrolysis temperature (210 C) to circumvent the for-
mation of a pyrochlore phase, Li et al. reported crystalliza-
tion of device-quality PZT (30/70) lms at 450 C for 1 h
[16]. These reports clearly show that perovskite PZT lms
can be crystallized at low temperatures by introducing a
seed layer and modulating the Zr/Ti ratio and Pb excess
in the precursor solution; however, this process requires a
long annealing time. Another way to lower the processing
temperature is by annealing PZT lms with special meth-
ods, including laser irradiation or laser annealing [1719]
as well as microwave irradiation or microwave annealing
[2025].
Recently, microwave heating techniques have attracted
considerable attention as a new way to anneal solgel-
derived PZT lms because they oer many advantages such
as low energy cost, ne resultant microstructure, short pro-
cessing time and low processing temperature. For example,
Wang et al. crystallized perovskite PZT (52/48) lms with
good electrical properties by using multimode microwave
irradiation with a frequency of 28 GHz at 480 C for only
5 min without needing to modify the precursors [20]. In sin-
gle-mode microwave irradiation, the electric and magnetic
elds of the microwave can be spatially separated, exposing
the sample being treated to a reasonably pure electric or
magnetic eld. Bhaskar et al. reported crystallizing perov-
skite PZT lms at 450 C for 30 min in a microwave electric
eld generated by a single-mode 2.45 GHz microwave heat-
ing system with SiC rods acting as susceptors to absorb the
microwave energy and transfer the heat generated to the
lm [21,22]. By using the magnetic eld of single-mode
microwave irradiation instead of the electric eld, we crys-
tallized PZT lms similar to those of Bhaskar et al. at the
same temperature and time without needing to use SiC
rods as susceptors [2325]. In one of our previous studies,
we also found that the pyrochlore phase transforms into
the perovskite phase via an intermediate pseudo-perovskite
phase, which may be induced by microwave energy [25].
These results suggest that the nucleation and growth of
perovskite grains in PZT lms crystallized by microwave
irradiation may be dierent from those processes in PZT
lms crystallized by conventional thermal processing.
However, the mechanism by which microwave irradiation
induces amorphous PZT to transform into perovskite
PZT in thin lms is not yet fully understood. Studying
the kinetics of the formation of the perovskite phase,
including its nucleation and growth, is fundamental to
optimizing the processing conditions of crystallizing PZT,
specically reducing its processing temperature and time.
Additionally, understanding how microwave irradiation
aects the crystallization of amorphous PZT thin lms in
ways dierent from conventional thermal annealing will
improve treatments of other functional thin lms using
microwave irradiation.
In this study, we evaluated the transformation from
amorphous to perovskite in solgel-derived PZT lms
induced by microwave irradiation. Using X-ray diraction
(XRD), we determined the volume fraction of the perov-
skite phase transformed during this process. The nucle-
ation site, crystal structure and grain growth were
identied by transmission electron microscopy (TEM).
We correlated the phase information to the ferroelectric
properties of the lm, assessed the nucleation and growth
kinetics and discussed the mechanism by which the micro-
wave irradiation aected the phase transformation.
2. Experimental procedure
A B-doped Si wafer with an electrical resistivity of
47 X cm was used as the substrate. To prepare the
Pt/Ti/SiO
2
/Si stack, the Si wafer was oxidized to produce
500 nm of SiO
2
, and then Pt (200 nm)/Ti (50 nm)
stacks were sputtered. The amorphous PZT lms were
deposited on the Pt/Ti/SiO
2
/Si stacks using a solgel
method. The precursor solution was prepared from lead
acetate (Pb(CH
3
COO)
2
), zirconium-n-propoxide
(Zr(OCH
2
CH
2
CH
3
)
4
) and titanium tetraisopropoxide
(Ti((CH
3
)
2
CHO)
4
). 2-propanol [(CH
3
)
2
CHOH] was used
as the solvent. The nal concentration of the solution
was adjusted to 0.4 M with the atomic ratio of
Pb:Zr:Ti = 1.1:0.52:0.48. We coated the substrate with pre-
cursor using a spin coater, operated at 1500 rpm for 10 s
and then at 4000 rpm for 40 s. The coated lms were dried
at 120 C for 5 min and then pyrolyzed at 250 C for 5 min.
This coating, drying and pyrolizing process was repeated
once more to obtain 300 nm thick amorphous PZT lms
for microwave irradiation.
To produce the microwave irradiation, we used a single-
mode microwave (frequency = 2.45 GHz) generator with a
waveguide in TE103 mode; the wavelength (kg) of the
standing wave formed in this cavity was 12.24 cm. By mov-
ing a plunger at the end of the waveguide, the electric and
magnetic elds could be maximized at a specic position.
Each sample (10 10 mm
2
) was irradiated by placing it
at the center of the highest magnetic eld. The temperature
of the samples was measured during microwave irradiation
by using an infrared thermometer (CTLaser2 M; Optris,
Berlin, Germany). This sensor could not measure sample
temperatures below 300 C. We measured the accuracy of
the infrared pyrometer over the temperature range of
400700 C by using Zn and Al lms with known melting
point of 419.5 C and 660.4 C, respectively; all tempera-
tures reported in this paper are accurate to within 5 C.
The amorphous PZT lms coated on the substrates were
2 Y.N. Chen et al. / Acta Materialia 71 (2014) 110
irradiated in the microwave magnetic eld at 525, 550 and
575 C for 101200 s. For comparison, PZT lms were also
heated in a conventional electric furnace at 525, 550, 575
and 600 C for dierent times.
The PZT lms were analyzed by XRD (Cu K
a
radiation,
D/max-2000; Rigaku, Tokyo, Japan). The microstructures
of the lms were studied by TEM (Tecnai G
2
F20; FEI,
Eindhoven, The Netherlands) with an attached energy dis-
persive X-ray spectroscopy detector and a high-angle annu-
lar dark-eld detector. The TEM specimens were ground
and polished to a thickness of 20 lm and were then milled
with Ar ions until they were electron transparent by using
a Gatan precision ion-polishing system (PIPS 691; Gatan,
Pleasanton, USA). Pt electrodes (diameter = 0.5 mm) were
sputtered as top electrodes for electrical measurements. The
polarization vs. electric eld (PE) hysteresis loop of each
lm was measured at 1 kHz using a standard ferroelectric
testing system (TF2000E; Aixacct, Aachen, Germany).
3. Results and discussion
3.1. Crystallization of PZT lms by microwave irradiation
Fig. 1 shows the XRD patterns of PZT lms heated by
microwave irradiation at various processing temperatures
and times. All the samples are amorphous before micro-
wave irradiation. After irradiation at 525 C for 30 s, at
550 C for 10 s, or at 575 C for 1 s, a broad peak appeared
that we attribute to the (222) plane of the pyrochlore
phase. At the same irradiation conditions, a small dirac-
tion peak appeared close to the Pt (111) peak, which seems
to correspond to the (111) plane of the perovskite phase.
Increasing the irradiation time gradually increased the
intensity of this peak, allowing us to conrm that it is the
(111) perovskite peak. After irradiation at 525 C for
540 s, at 550 C for 90 s or at 575 C for 60 s, the (222)
pyrochlore peak disappeared. All the lms crystallized into
the perovskite phase and exhibited (111)-preferred orienta-
tion, properties similar to those of the lm crystallized by
conventional annealing at 600 C for 30 min, shown in
the rst pattern at the top of Fig. 1c.
For comparison, Fig. 2 shows the XRD patterns of PZT
lms heated by conventional heating. The lms heated at
525 C for 30 min and at 550 C for 15 min showed only
a broad (222) pyrochlore peak. The lm heated at
575 C for 2 min exhibited a small (111) peak for the
perovskite phase. In contrast, when using microwave irra-
diation, the small (111) perovskite peak appeared in lms
irradiated at 525 C for only 30 s, at 550 C for only 10 s
and at 575 C for only 1 s (Fig. 1ac, respectively). These
results indicate that microwave irradiation induced the for-
mation of the perovskite phase at low processing times. By
using microwave irradiation, we obtained fully crystallized
perovskite PZT lms at 525 C for 20 min and at 575 C for
only 60 s, much shorter times and lower temperatures than
those required when using conventional heating, 600 C for
30 min.
To understand the crystallites formed during initial crys-
tallization of the PZT lms heated by microwave irradia-
tion, we scanned the 2h peaks at 3642 in detail; these
results are shown in insets of Fig. 1ac. The (111)
Fig. 1. XRD patterns of PZT lms heated by microwave irradiation for
dierent times at (a) 525 C, (b) 550 C and (c) 575 C. Pv: perovskite
phase; Py: pyrochlore phase. The insets of (a), (b) and (d) show the (111)
peak of the perovskite phase shifting toward lower 2h angles as the
irradiation time increased.
Y.N. Chen et al. / Acta Materialia 71 (2014) 110 3
perovskite peaks shifted toward lower angles with irradia-
tion time, suggesting that their d-spacing increased during
crystallization. Fig. 3 also shows that the d-spacing of the
(111) planes increased with irradiation time at various tem-
peratures; note that the d-spacing of the initial crystallites
remained nearly the same across the three temperatures
shown in Fig. 3. The d-spacing of 0.2331 nm is 1% smal-
ler than the standard d-spacing of the perovskite phase
(d
111
= 0.2351 nm for PZT(52/48)). These results imply
that the initial crystallites may be an intermediate phase.
Various reports have shown that, in conventional heating,
transition intermetallic Pt
x
Pb phases form during early
crystallization of PZT thin lms deposited on Pt/Ti/SiO
2
/
Si substrates and that these intermediates lower the forma-
tion temperature of (111)-oriented perovskite [13,26,27].
Because these intermediate phases have lattice parameters
close to those of perovskite PZT, they may act as
nucleation sites (d
111
Pt
3
Ti = 0.2244 nm, d
111
Pt
x-
Ti = 0.2340 nm). To determine whether the initial crystal-
lite we observed was an intermetallic Pt
x
Pb phase, we
used TEM to study the nucleation and growth of the
perovskite phase in PZT lms heated by microwave irradi-
ation at 525 C for various times.
Fig. 4a shows a cross-sectional TEM image of the PZT
lm heated by microwave irradiation at 525 C for 30 s.
The XRD data for this lm (Fig. 1a) show that the lm
was in an early stage of crystallization. In this TEM
cross-section, the lm appears to have begun crystallizing;
we observed initial crystallites at the bottom of the PZT
lm near the PZT/Pt interface. Some of the initial crystal-
lites exhibited ferroelectric stripe-like patterns (inset of
Fig. 4a), which were also clearly present in the well-crystal-
lized perovskite PZT grains (Fig. 4c and d). This similarity
shows that the initial crystallite was ferroelectric. We also
observed ne grains in other parts of the lm, identied
as pyrochlore by using a selected-area electron diraction
pattern (inset of Fig. 4a). Fig. 4b shows a high-resolution
TEM image of the initial crystallite marked with A in
Fig. 4a. The fast Fourier transform pattern of this crystal-
lite revealed it to be a perovskite phase. The (111) plane
parallel to the PZT/Pt interface exhibited a d-spacing of
0.2330 nm, consistent with the XRD result of 0.2331 nm.
Furthermore, the d-spacing of this crystallites (110) plane
is smaller than that of the perovskite PZT. We also ana-
lyzed the lattice parameters of other initial crystallites;
these results are summarized in Table 1. The d-spacings
of these crystallites were smaller than the standard
d-spacing of perovskite PZT by 1%. From these results,
we conclude that the initial crystallites were an intermedi-
ate phase with a crystal structure similar to that of perov-
skite PZT but with smaller lattice parameters. Fig. 4c
shows a TEM cross-section of a PZT lm during interme-
diate crystallization, heated by microwave irradiation for
240 s; this image corresponds to the XRD data in
Fig. 1a. In this lm, the perovskite grains had grown to a
third of the lm thickness. Fig. 4d shows another cross-sec-
tion of a lm irradiated for 1200 s; in this lm, the colum-
nar grains, perfectly crystallized into the perovskite phase,
reached the surface of the lm.
To analyze the chemical compositions of the initial crys-
tallites, we performed semi-quantitative STEM-EDS on the
initial crystallite marked by B in Fig. 4a; these results are
shown in Fig. 5. Fig. 5a shows a high-angle annular
dark-eld image, with the initial crystallite in the center
of the image, surrounded by a box. Fig. 5bf shows the
EDS results from the initial crystallite, revealing the ele-
mental distribution of Pt, Pb, Zr, Ti and O. The Pt, Pb,
Zr and O in the initial crystallite were nearly equal to its
surroundings; in contrast, the initial crystallite contained
more Ti (Fig. 5f) than in its surroundings. Combining
our XRD and TEM results, we conclude that the initial
crystallite, an intermediate phase formed during initial
crystallization, had a perovskite structure with a smaller
lattice constant and contained more Ti than perovskite
Fig. 2. XRD patterns of PZT lms crystallized by conventional annealing
at (a) 525 C for 30 min, (b) 550 C for 15 min and (c) 575 C for 2 min.
Pv: perovskite phase; Py: pyrochlore phase.
Fig. 3. Variation of d-spacing of the (111) plane for various microwave-
irradiation times. The d-spacing deviates 1% from that of standard
perovskite value during initial crystallization. The error bars of the d-
spacing indicate the range of measured values over three measurements.
4 Y.N. Chen et al. / Acta Materialia 71 (2014) 110
PZT. We denote this intermediate phase as Pv to distin-
guish it from the perovskite PZT (Pv).
The crystallization of amorphous PZT thin lms on Pt
bottom electrodes using conventional heating has been
widely studied. Brooks et al. [28] suggested that the
Pt(111) electrode acts as a nucleation site for the perov-
skite PZT, which grows into the pyrochlore matrix. The
nucleation mechanisms for perovskite PZT during conven-
Fig. 4. TEM images of PZT lms crystallized by microwave irradiation at 525 C. (a) Cross-sectional TEM image of the lm irradiated for 30 s; (b)
HRTEM image of a crystallite from (a) at the Pt/lm interface; (c) cross-sectional TEM image of the lm irradiated for 360 s, showing that the columnar
perovskite grains grew into the pyrochlore layer; (d) cross-sectional TEM image of the lm irradiated for 1200 s, showing that the columnar perovskite
grains grew to the surface of lm.
Table 1
Lattice parameters of crystallite nuclei, calculated from XRD and TEM
data.
PDF card XRD TEM
(100) 4.0360 3.99
(110) 2.8500 2.82
(101) 2.8900 2.87
(111) 2.3510 2.3305 2.33
(201) 1.8100 1.80
Fig. 5. (a) Cross-sectional STEM image of the PZT lm; the area in the center marked with the box is a crystal nucleus. (bf) EDS elemental maps from
the nucleus shown in (a).
Y.N. Chen et al. / Acta Materialia 71 (2014) 110 5
tional heating have also been analyzed: researchers believe
that the perovskite heterogeneously nucleates on a Pt
3
Pb
intermetallic phase generated by interfacial reaction
between the lm and Pt electrode; this intermetallic is
believed to induce the phase transformation from pyroch-
lore to perovskite [13,26,27]. In contrast, in the present
study, we found that the Pv intermediate phase, formed
during initial crystallization, acted as a nucleation site for
the perovskite PZT. This intermediate phase has not been
found in crystallization of PZT lms with compositions
close to the morphotropic phase boundary [2931] or in
Ti-rich PZT lms heated by conventional thermal process-
ing [26,27,32,33]. Thus, we believe that the intermediate
phase was generated by microwave irradiation.
Several factors may have contributed to the formation
of the Pv intermediate phase. The rst mechanism is
charge transfer induced by the microwave irradiation at
the PZT/Pt interface, which consists of materials with dif-
ferent dielectric constants and conductivities. The transfor-
mation from amorphous or pyrochlore to perovskite
requires the ordering of the lead, titanium and zirconium
cations, a process likely to be sluggish. Microwave irradia-
tion may have induced interfacial space charges and
dipoles at the PZT/Pt interface, causing a new ordering
of ions and atoms on the PZT side [34]. The second mech-
anism may have to do with dierences in thermal expan-
sion. In one of our previous studies, we investigated the
heating behaviors of PZT/Pt/Ti/SiO
2
/Si, PZT/SiO
2
/Si,
Pt/Ti/SiO
2
/Si and SiO
2
/Si multilayered structures, heated
by single-mode 2.45 GHz microwave irradiation in the
magnetic eld; we found that the heating of the Pt/Ti/
SiO
2
/Si samples mainly originated from contributions of
the Pt/Ti layers and the Si substrate and that the Pt/Ti lay-
ers were more eectively heated than the Si substrate [35].
Heating during microwave irradiation can be considered
similar to ohmic heating induced by an alternating mag-
netic eld [36]; the higher electric conductivity of the Pt/
Ti layers led to a higher eddy current, explaining why the
Pt/Ti layer was more eectively heated than the Si sub-
strate. During microwave irradiation, the heat originated
from the Pt/Ti/SiO
2
/Si substrate and the heating rate was
very fast (10 C s
1
) [37]. Dierences in the thermal
expansion coecients between the PZT lm and substrate
may have led to great stress at the PZT/Pt interface, a pos-
sible driving force for ionic motion. The intermediate phase
Pv and Pt have a smaller lattice mismatch (1.7%) than do
the perovskite Pv and Pt (2.8%), which induces the forma-
tion of the intermediate phase. The third mechanism may
be related to the intermediate phase being rich in Ti: while
PbTiO
3
, PbZrO
3
and PZT have the same perovskite struc-
ture, their crystallization temperatures increase as follows:
PbTiO
3
< PbZrO
3
< PZT [38]. Furthermore, reports have
shown that Ti-rich amorphous PZT thin lms can be crys-
tallized into the perovskite phase at low temperatures
[32,33]. Although we have discussed the formation of the
intermediate phase and some possible mechanisms, further
research is needed. The presence of the Pv intermediate
phase may have changed the path of phase transitions,
reducing the activation energy of nucleating the perovskite
phase. We will discuss more on this issue in detail later in
the next section.
Based on our experimental results and analysis, we pro-
pose a tentative mechanism for the crystallization of PZT
lms during microwave irradiation; a schematic of this pro-
cess is shown in Fig. 6. First, the Pv crystallites induced by
microwave irradiation homogenously form at the bottom
of the lm near the PZT/Pt interface (Fig. 6a). Then, they
act as nuclei for the formation of the perovskite phase (Pv).
As the irradiation time increases, the columnar perovskite
grains grow vertically in vertical lamellae, and the interface
between the pyrochlore (Py) and the perovskite remains
almost at (Fig. 6b). The lamellar growth occurs because
the nuclei of the perovskite grains on the intermediate
phase (Pv) are homogenously spaced and because the
growth rates of the perovskite grains with (111)-preferred
orientation are nearly the same, which occurs because the
temperature gradient remains the same across the grains
as they linearly diverge from the electrode/substrate and
because the microwave irradiation directly aects the Py/
Pv interface. Finally, the columnar perovskite grains reach
the surface of the lm, perfectly crystallizing the lm into
the perovskite phase (Fig. 6c).
We investigated how the crystallization process inu-
enced the ferroelectric properties of the PZT lms by mea-
suring polarization vs. electric eld (PE) hysteresis loops
at 1 kHz. Fig. 7 shows the PE hysteresis loops of PZT
lms heated by microwave irradiation at 525, 550 and
575 C for various times. During initial crystallization, all
the lms exhibited quasi-linear PE and remanent polar-
Fig. 6. Schematic of the crystallization process in PZT lms induced by
microwave irradiation.
6 Y.N. Chen et al. / Acta Materialia 71 (2014) 110
izations of <1 lC cm
2
. These results occurred because,
during this crystallization step, the ferroelectric perovskite
phase was very thin and only lay at the bottom of the lm
near the PZT/Pt interface; the paraelectric pyrochlore
phase was the predominant phase, as indicated by XRD
and TEM results. As the irradiation time increased, the
hysteresis loops became tilted as the perovskite phase grew,
and the remanent polarization (P
r
) increased abruptly.
Fig. 7d shows how the remanent polarization and coercive
eld (E
c
) varied with microwave irradiation time, demon-
strating that the development of ferroelectricity in the
PZT lms correlated to the phase transition from amor-
phous to perovskite.
3.2. Activation energy of crystallization in PZT lms heated
by microwave irradiation
The crystallization kinetics in PZT lms heated by
microwave irradiation can be analyzed by following the
general approach described by the Avrami transformation
kinetics equation [39,40]:
ft 1 expkt
n
1
where f(t) is the volume fraction of the transformed perov-
skite phase, t is the irradiation time, k is a temperature-
dependent crystallization-rate constant and n is the Avrami
coecient, which is temperature independent but sensitive
to reaction conditions, including the nucleation and growth
rate.
Fig. 8a shows the relative intensity as a function of irra-
diation time, t to the Avrami model. These data were nor-
malized against the intensity of the XRD (111) peak of the
perovskite phase from lms conventionally annealed at
600 C for 30 min. Note that the transformed fraction in
the lms crystallized by microwave irradiation at 575 C
for 60 s is more than 100%; this behavior occurred because
the crystallization of the microwave-irradiated lm was
much more complete. Taking logarithms and rearranging
Eq. (1), we plotted ln{ln [1/(1 f(t))]} vs. ln t, yielding a
linear series of points with slope n and intercept ln
k; Fig. 8b shows representative ts of the data at the three
temperatures. The extracted Avrami coecient is 0.8,
which agrees well with values predicted for one-
dimensional linear growth and thickening of very large
plates [40], suggesting that the transformation from
pyrochlore to perovskite occurred via diusion-limited
Fig. 7. PE hysteresis loops of the PZT lms, heated by microwave irradiation at (a) 525 C, (b) 550 C and (c) 575 C for dierent irradiation times; (d)
remanent polarization as a function of irradiation time. The error bars denote the range of measurements after ve measurements, using dierent top
electrodes; no error bar means that the range was smaller than the height of the symbol used for the data point.
Y.N. Chen et al. / Acta Materialia 71 (2014) 110 7
growth and that the structural transformation occurred at
the phase interface. Our TEM results showed that the
columnar perovskite grains grew uniformly and vertically
from the PZT/Pt interface and that the perovskite/pyroch-
lore interface was at, like the interface of a thickening
plate (Fig. 4a, c and d). Thus, our theoretical and experi-
mental results are very consistent.
The temperature dependence of the reaction rate con-
stant k can be described by the Arrhenius equation [41]:
k Aexp
E
a
RT

2
where A is the pre-exponential factor, E
a
is the apparent
activation energy of crystallization and R is the molar
gas constant. Fig. 8c shows the plot of ln k vs. 1/T; from
the slope of this line, we obtained an activation energy of
214 kJ mol
1
for the PZT lms crystallized by microwave
irradiation. The crystallization kinetics during conven-
tional annealing of solgel-derived PZT thin lms has been
widely studied, allowing us to compare the value obtained
in the present report [4246] The present activation energy
is lower than the values obtained by Griswold et al. [43]
and Babushkin et al. [44] 275 and 310 kJ mol
1
, respec-
tively for similar solgel-derived PZT(53/47) lms crys-
tallized by conventional thermal processes. This decrease
may have been caused by the presence of the intermediate
phase, a direct eect of microwave irradiation. As discussed
before, the intermediate phase was likely induced by micro-
wave irradiation during initial crystallization. The lattice
mismatch between the intermediate phase and Pt was small
compared with that between the perovskite and Pt, reduc-
ing the activation energy of nucleation. Huang et al. re-
ported that the formation of an intermetallic Pt
x
Pb phase
during initial crystallization of a PZT thin lm deposited
on Pt/Ti/SiO
2
/Si substrate reduced the nucleation activa-
tion energy for forming the perovskite phase; they reported
an activation energy of 179 kJ mol
1
for forming the
perovskite phase in the Ti-rich PZT (30/70) lms [26].
The direct eect of microwaves on the growth of the perov-
skite grains can be considered within the framework of the
ponderomotive driving force model and may have implica-
tions for reduced energy use in solid-state processing [47].
Models of the enhanced ion transport and related eects
have suggested that these behaviors occur because of gradi-
ents in the mobility of mobile-charged species near grain
boundaries or other physical interfaces. Whittaker re-
searched diusion in microwave-heated ceramics (YbBa
2-
Cu
3
O
7d
embedded in YBa
2
Cu
3
O
7d
) and provided direct
evidence that the microwaves directly inuenced ion trans-
port during high-temperature sintering [48]. Because the
microwaves concentrated lattice defects and enhanced ion
mobility at the interface, they also increased the mass
transport. In the present study, the columnar perovskite
grains grew vertically and uniformly from the PZT/Pt
interface through the pyrochlore matrix throughout the
crystallization process. The microwaves may have directly
inuenced ion transport at the Pv/Py interface, reducing
the activation energy of growing the perovskite phase.
Fig. 9 shows a schematic of the phase transition path,
demonstrating how the Gibbs energy varies during PZT
crystallization induced by microwave irradiation and con-
Fig. 8. (a) Fraction of transformed perovskite as a function of time at
various temperatures; (b) Avrami plot of volume fraction ln{ln[1/(1 f)]}
as a function of the natural logarithm of time at various temperatures. The
error bars in (a) and (b) denote the range of the fraction of transformed
perovskite after three measurements. (c) Arrhenius plot of the reaction
constant as a function of the reciprocal temperature, used to determine the
activation energy. The error bars indicate one standard deviation of ln k
calculated by linear tting.
8 Y.N. Chen et al. / Acta Materialia 71 (2014) 110
ventional annealing. During conventional annealing, the
amorphous phase transforms to the perovskite phase
through the pyrochlore phase (the solid dark line). During
microwave irradiation, the phase transition from amor-
phous to pyrochlore still occurs, meaning the activation
energy (E
a0
) is the same as that in conventional annealing.
However, the phase transition path for pyrochlore to
perovskite diers between microwave irradiation and con-
ventional annealing. During initial crystallization, the tran-
sient phase (Pv) forms before the perovskite phase (Pv),
which then grows on Pv, reducing the Gibbs energy of
activation (E
a
) to E
a1
and E
a2
(the dashed red line). There-
fore, the emergence of Pv changes the phase transition
path, reducing the activation energy of nucleating the
perovskite phase. During grain growth of the perovskite
phase, ion transport at the Pv/Py interface is also enhanced
by the microwave irradiation. This also reduces the Gibbs
energy of activation to E
a3
(the dashed blue line). Thus,
crystallization from amorphous to perovskite in PZT lms
requires a lower activation energy when a lm is heated by
microwave irradiation.
4. Conclusions
We prepared PZT thin lms using a solgel method and
then crystallized them by using microwave irradiation at
2.45 GHz in the magnetic eld of the microwave. To study
the crystallization and kinetics of the transformation from
amorphous to perovskite PZT in the thin lms, we used
XRD to identify the phases grown and the volume fraction
of the transformed perovskite phase, and we used TEM to
investigate the grain structure, nucleation and growth. Our
results show that an intermediate phase (Pv) formed dur-
ing initial crystallization; this phase had a perovskite crys-
tal structure with a smaller lattice constant and higher Ti
content than perovskite PZT. The intermediate phase acted
as a nucleation site for the perovskite, which grew with a
columnar grain structure into the pyrochlore matrix
throughout the whole lm. We conrmed this crystalliza-
tion process by assessing how the ferroelectric properties
of the lms varied. The eective activation energy for the
formation of the perovskite phase using microwave irradi-
ation was 214 kJ mol
1
, lower than that for perovskite
crystallized by conventional thermal processes. These
results demonstrate that, in crystallization of PZT lms
heated by microwave irradiation, the microwave radiation
induced an intermediate phase, which lowered the crystal-
lization temperature and enhanced the growth rate of the
perovskite phase.
Acknowledgements
We gratefully acknowledge the partial support of this
work by the Hundred Talents Program of the Chinese
Academy of Sciences, the National Natural Science
Foundation of China (Nos. 51072202 and 51172238), the
National Basic Research Program (No. 2010CB934603)
of China and the Ministry of Science and Technology of
China.
References
[1] Scott JF. Science 2007;315:954.
[2] Smith GL, Pulskamp JS, Sanchez LM, Potrepka DM, Proie RM,
Ivanov TG, et al. J Am Ceram Soc 2012;95:1777.
[3] Bhushan B, Kwak KJ, Palacio M. J Phys: Condens Matter
2008;20:365207.
[4] Deho C, Hennings A, Kugeler C, Schneller T, Bottger U. Phys
Status Solidi A 2011;208:343.
[5] Xu Z, Chan W-H. Acta Mater 2007;55:3923.
[6] Wang ZJ, Kokawa H, Maeda R. Acta Mater 2005;53:593.
[7] Chen I-S, Roeder JF, Glassman TE, Baum TH. Chem Mater
1999;11:209.
[8] Gong W, Li J-F, Chu X, Gui Z, Li L. Acta Mater 2004;52:2787.
[9] Bastani Y, Bassiri-Gharb N. Acta Mater 2012;60:1346.
[10] Habouti S, Lahmar A, Dietze M, Solterbeck C-H, Zaporojtchenko V,
Es-Souni A. Acta Mater 2009;57:2328.
[11] Wang ZJ, Chu J, Maeda R, Kokawa H. Thin Solid Films
2002;416:66.
[12] Kwok CK, Desu SB. J Mater Res 1993;8:339.
[13] Huang Z, Zhang Q, Whatmore RW. J Mater Sci Lett 1998;17:1157.
[14] Suzuki H, Koizumi T, Kondo Y, Kaneko S. J Eur Ceram Soc
1999;19:1397.
[15] Wu A, Vilarinho PM, Reaney I, Miranda Salvado IM. Chem Mater
2003;15:1147.
[16] Li J, Kameda H, Trinh BNQ, Miyasako T, Tue PT, Tokumitsu E,
et al. Appl Phys Lett 2010;97:102905.
[17] Pan HC, Chou C-C, Tsai H-L. Appl Phys Lett 2003;83:3156.
[18] Lai SC, Lue H-T, Hsieh KY, Lung SL, Liu R, Wu TB, et al. J Appl
Phys 2004;96:2779.
[19] Muralt P, Polcawich RG, Trolier-McKinstry S. Mater Res Soc Bull
2009;34:658.
[20] Wang ZJ, Kokawa H, Takizawa H, Ichiki M, Maeda R. Appl Phys
Lett 2005;86:212903.
[21] Bhaskar A, Chang HY, Chang TH, Cheng SY. Thin Solid Films
2007;515:2891.
[22] Bhaskar A, Chang HY, Chang TH, Cheng SY. Nanotechnology
2007;18:395704.
[23] Wang ZJ, Cao Z, Otsuka Y, Yoshikawa N, Kokawa H, Taniguchi S.
Appl Phys Lett 2008;92:222905.
Fig. 9. Schematic of crystallization associated with the Gibbs energy of
activation in the PZT lm as well as the microwave-energy transfer path.
Y.N. Chen et al. / Acta Materialia 71 (2014) 110 9
[24] Wang ZJ, Otsuka Y, Cao Z, Kokawa H. Jpn J Appl Phys
2009;48(09K):A01.
[25] Wang ZJ, Chen YN, Otsuka Y, Zhu MW, Cao Z, Kokawa H. J Am
Ceram Soc 2011;94:404.
[26] Huang Z, Zhang Q, Whatmore RW. J Appl Phys 1999;85:7355.
[27] Huang Z, Zhang Q, Whatmore RW. J Appl Phys 1999;86:1662.
[28] Brooks KG, Kissurska RD, Moeckli P, Setter N. J Mater Res
1997;12:531.
[29] Kwok CK, Desu SB. Appl Phys Lett 1992;60:1430.
[30] Chen S-Y, Chen I-W. J Am Ceram Soc 1994;77:2332.
[31] Chen S-Y, Chen I-W. J Am Ceram Soc 1994;77:2337.
[32] Zhang Q, Whatmore RW, Huang Z, Vikers ME. J Phys IV France
1998;8(Pr9-7):9.
[33] Kosec M, Malic B, Mandeljc M. Mater Sci Semicond Process
2003;5:97.
[34] Von Hippel AR. Dielectrics and waves. Norwood: Artech House;
1995. p. 22834.
[35] Cao Z, Wang ZJ, Yoshikawa N, Taniguchi S. J Phys D: Appl Phys
2008;41:092003.
[36] Lou J, Hunyar C, Feher L, Link G, Thumm M, Pozzo P. Appl Phys
Lett 2004;84:5076.
[37] Chen YN, Wang ZJ. J Am Ceram Soc 2013;96:90.
[38] Wilkinson AP, Speck JS, Cheetham AK. Chem Mater 1994;6:750.
[39] Avrami MJ. Chem Phys 1940;8:212.
[40] Christian JW. The theory of transformations in metals and alloys
(part I + II). 3rd ed. Oxford: Pergamon; 2002.
[41] Vyazovkin S, Wight CA. Annu Rev Phys Chem 1997;48:125.
[42] Voigt JA, Tuitle BA, Headley TJ, Eatough MO, Lamppa DL,
Goodnow D. Mat Res Soc Symp Proc 1993;310:15.
[43] Griswold EM, Weaver L, Calder ID, Sayer M. Mater Res Soc Symp
Proc 1995;361:389.
[44] Babushkin O, Lindback T, Brooks K, Setter N. J Eur Ceram Soc
1997;17:813.
[45] Es-Souni M, Piorra A. Mater Res Bull 2001;36:2563.
[46] Lee B, Kim C, Kim SH, Shin H. Integr Ferroelectr 2004;68:247.
[47] Booske JH, Cooper RF, Freeman SA, Rybakov KI, Semenov VE.
Phys Plasmas 1998;5:1664.
[48] Whittaker AG. Chem Mater 2005;17:3426.
10 Y.N. Chen et al. / Acta Materialia 71 (2014) 110