Sensors and Actuators A 158 (2010) 106–111

Contents lists available at ScienceDirect

Sensors and Actuators A: Physical

journal homepage: www.elsevier.com/locate/sna

Current leakage and transients in ferroelectric ceramics under high

humidity conditions
Deyi Zheng a,∗ , Jonathan Swingler a , Paul Weaver b
a
School of Engineering Sciences, University of Southampton, University Road, Southampton SO17 1BJ, United Kingdom
b
National Physical Laboratory, Hampton Road, Teddington Middle TW11 0LW, United Kingdom

a r t i c l e i n f o a b s t r a c t

Article history: Lead zirconate titanate (PZT) is widely used in electromechanical devices such as ultrasonic transduc-
Received 27 July 2009 ers, sensors, ultrasonic motors, actuators and resonators. In harsh operating environments such as high
Received in revised form 28 October 2009 humidity and high temperature the electrical resistance of the ceramic can decrease resulting in leak-
Accepted 29 October 2009
age currents. These can have a significant impact on device performance, particularly for low power
Available online 26 November 2009
applications. In this paper, the increase in leakage current is investigated and characterised under high
humidity conditions. It is observed that the leakage current does not evolve smoothly over time, but is
Keywords:
characterised by current transients. Detailed measurements of these current transients are reported, and
Lead zirconate titanate
Electrical degradation
their relationship to the conduction processes in the ceramic is investigated. A mechanism of conductive
Current transients channel formation is proposed to account for the leakage current increase and current transients.
Electrical breakdown © 2009 Elsevier B.V. All rights reserved.

1. Introduction achieved by A-site vacancy substitution (e.g. La doping) which also

Lead zirconate titanate (PZT) ceramics are widely used in the
manufacture of actuators in a range of devices including security
locks, valves for fluid control, and diesel injector valves. The devel-
opment of these applications in recent years has brought with it
increased demands on performance and reliability under high elec-
trical fields and in humid environments. Recent work [1] has shown
that electromigration is the probable cause of leakage currents
under d.c. bias in humid conditions, and a model mechanism based
on the evolution of conducting pathways was proposed. Other work
[2] has shown how the conduction mechanism may be related to
the micro-structure and porosity of the ceramic.
Undoped PZT is known as a wide band gap semiconductor with
a band gap of between 2.6 eV and 3.5 eV [3]. It also exhibits p-
type conductivity due to the presence of low valence impurities
substituting for higher valence Pb ions. This causes A-site (Pb ion)
vacancies which act as electron acceptors leading to the produc-
tion of holes [4]. This means that a Schottky barrier can be formed
with the electrode material which can promote conduction in thin
films [3,5]. Tunnelling can also be a mechanism for conduction in
very thin films [5]. However, for films thicker than a few tens of
nm, conduction is determined by the bulk of the material through
Poole–Frenkel emission, where thermally excited electrons are
assisted by the electric field into the conduction band [3]. The
large piezoelectric coefficients in this class of “soft” PZT materials is
∗ Corresponding author. Tel.: +44 023 8059 5479.
E-mail addresses: d.zheng@soton.ac.uk, dz1j06@soton.ac.uk (D. Zheng).
confers high electrical resistivity [4]. Data from Ref. [4] would indi-
cate a volume resistivity of 109  cm at the temperature reported
here. For the sample size and field strength used in this study we
would expect a resistance of over 300 G which, and a leakage
current of less than 1 nA. The effects studied here are in the ␮A to
mA range, so cannot be explained by these electronic conduction
mechanisms.
There is evidence that the conduction mechanism in humid
environments is ionic in character, including the observation of
anodic breakdown features and sensitivity of the leakage current
to the relative humidity and electrode material [1]. Measurements
performed by the authors indicate that there is no significant
(in this context <250 nA is considered insignificant) leakage cur-
rent at the temperature of the measurements reported here (55 ◦ C
dry atmosphere). In fact a reduction of leakage current is usually
observed on heating to this temperature from room temperature
due to the expulsion of moisture.
Ionic silver migration from silver electrodes is commonly impli-
cated in humidity-related conduction in otherwise insulating
materials such as thick film insulators [6,7]. The proposed mech-
anism involves ionization of water at the silver cathode followed
by release of silver ions which migrate under the applied electri-
cal field. A similar mechanism has also been proposed previously
for electrical changes in PZT with Ni electrodes [1,2]. Although an
electromigration model explained the observed leakage currents,
these studies did not present micro-structural details of the pro-
posed conducting channels. The reason for this is that the amount
of conducting material required to account for the observed con-
ductivity is extremely small. The difficulty in detecting this means

0924-4247/$ – see front matter © 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.sna.2009.10.021
 D. Zheng et al. / Sensors and Actuators A 158 (2010) 106–111
Fig. 1. Experimental arrangement.
indirect methods need to be used to obtain information on the con-
duction mechanism in PZT ceramics. Significant useful information
can be obtained from the electrical characteristics of the leakage
current. Lipscomb et al. [1] analysed the general trends in the evo-
lution of the leakage current, but did not investigate the detailed
short timescale features.
This study presents detailed time-resolved measurements of the
evolution of humidity-induced leakage current in nickel electroded
PZT devices. The measurements show significant fluctuations in the
leakage current under a steady applied d.c. voltage. The detailed
characteristics of the current transients are reported, and analysed
statistically in time and frequency domains. It is shown that the cur-
rent transients behave as a switched resistance-capacitance (RC)
network, and a simple network model provides good agreement
with experimental results. The evolution of the current transients
with time and the switched RC characteristic support the proposal
of an electromigration/breakdown mechanism for the leakage cur-
rent in this type of device.
2. Experimental arrangement
Commercially available PZT samples are investigated. The sam-
ples are 2.5 cm × 0.5 cm × 0.015 cm and have nickel electrodes on
both sides of the larger surface area. The experimental arrangement
is shown in Fig. 1, which illustrates how samples are connected in
the electrical circuit.
A 240 V d.c. potential difference is supplied by the power sup-
ply. Working voltages up to a maximum of 2 kV/mm are typical
for bulk and multilayer piezoelectric actuators in applications such
as valves and piezoelectric diesel injectors [7]. This is rather lower
than the field strengths that can be sustained in thin film devices
due to the scaling up of breakdown field strength with decreasing
film thickness. A value of 250 V (1.7 kV mm−1 ) was chosen as a typi-
cal voltage for many applications. The effects of varying the applied
field on leakage current and lifetime have been reported previously
[1]. The current through the sample and the voltage across the sam-
Fig. 2. System noise evaluations with: (a) 0.5 M resistor sample; (b) 1 M resistor sample; (c) 2 M resistor sample.
107
Fig. 3. Leakage current of PZT sample under 55 ◦ C, 90 RH 240 V d.c.
ple are calculated by measuring the voltage across RV and R1 . The
resistor values are: R1 = 10 M, RV = 20 k and RI = 20 k (to a 2%
error value).
The system noise on the current measurement under ambi-
ent conditions has been evaluated by using test resistor samples
instead of PZT samples. Three different test resistor samples of
0.5 M, 1 M and 2 M were used and the current behaviour mea-
sured with the 240 V d.c. applied. These current measurements
were taken over a 10 s period with a sampling rate of 1 ms. Plots of
the probability density function of particular current values for the
three test resistor samples are given in Fig. 2.
It is found that the measurement resolution for the 0.5 M test
resistor is 4 ␮A, for the 1 M test resistor is 2 ␮A and for the 2 M
test resistor is 1 ␮A. It can be seen that, if the resistance of the
samples is from 0.5 M to 2 M, the system noise is within ±2 ␮A.
Fig. 1 also shows the PZT ceramic in the temperature/humidity
chamber. The PZT was stressed by subjecting it to a harsh envi-
ronment of 55 ◦ C, 90% relative humidity (RH) and a 240 V d.c. bias.
Voltage and current measurements were made over a 68 h period.
Measurements were taken every 1 ms for the long-term current
leakage profile investigation and every 1 ␮s for detailed data inves-
tigation. Voltage recordings (not presented here) confirmed that
there were no significant fluctuations in the applied voltage.
3. Results
A typical result of the leakage current is presented in Fig. 3,
where measurements are taken every 1 ms for 68 h.
The leakage current is observed to increase from the noise level
of 2 ␮A to approximately 700 ␮A. It can be seen in Fig. 3 that
there is an initial period of up to 3 h when the leakage current
remains at approximately the initial level. After this period the leak-
age current accelerated increases to ∼400 ␮A within the first 30 h.
Thereafter, the leakage current approaches towards the final value
of ∼700 ␮A. It can be seen that over the 68 h period the leakage
current becomes highly variable with the appearance of the cur-
rent fluctuations. These fluctuations are significantly larger than
the noise level reported in Fig. 2 and are therefore attributed to
real current transients through the sample.
108 D. Zheng et al. / Sensors and Actuators A 158 (2010) 106–111

Fig. 5. Appearance time of the current transients.

Fig. 4. Detailed features of the current transients.

3.1. Current transients

The current fluctuations on the leakage current profile in Fig. 3
have been further analysed. Two typical fluctuations are found and
shown in Fig. 4 with a “shark fin” and “square” wave profile. These
fluctuations are suggested to be the same type of transient but with
various time duration of the plateau.
The data were analysed by considering only current transients
with an amplitude higher than 30 ␮A. 4937 current transients are
obtained and the appearance times are plotted in Fig. 5.
Fig. 5 shows that the longer the PZT ceramic is stressed, the more
current transients appear, and these appear at an exponential rate.
The leakage current per current transient event is plotted in Fig. 6.
Fig. 6. Leakage current per current transient event.
3.2. Current amplitude, A, of transients
The current amplitude of the 4937 transients is investigated
and the characteristics are shown in Fig. 7. Fig. 7a is a plot of the
current amplitude of all transients in order of appearance. Fig. 7b
is a probability density function of the current amplitude. Fig. 7c
is a plot of the average current amplitude of consecutive sets of
100 data points against the number of that set. Fig. 7d is a plot
of the average current amplitude of consecutive sets of 100 data

Fig. 7. (a) Amplitude of the current transients; (b) probability distribution of the current amplitude; (c) average values of the current amplitude; (d) average values of the
current amplitude with time.
 D. Zheng et al. / Sensors and Actuators A 158 (2010) 106–111 109

Fig. 8. (a) Time intervals of the current transients; (b) probability distribution of the time intervals; (c) average values of the time intervals; (d) average values of the time
intervals with time.

points against the average value of the appearance time for that ing to the three features in the current transient as illustrated in
set. Fig. 9.
Fig. 7a shows the amplitude of all current transients with ampli-
tude higher than 30 ␮A, so the area below 30 ␮A has no data points. Step 1: switch opening
The probability distribution of the current amplitude is analysed When the switch is open, the capacitor charges until it is fully
and plotted in Fig. 7b. It can be observed that most transients are charged leading to zero i3 current. In this case, the current i through
with lower current amplitude approximately 30–100 ␮A. Fig. 7c the whole circuit is:
and d shows that the average current amplitude converges to a
V0
plateau ∼52 ␮A. This indicates that stressing the PZT samples for i = iL + i1 = iL + i2 + i3 = iL = = IL (1)
longer has no influence on the amplitude of the current transients RL
after an initial period. This initial period is ∼18 h as shown by the Step 2: switch closure
dotted vertical line in Fig. 7d.

3.3. Time interval, , of transients

The time interval  between two adjacent transients from the

4937 transients is investigated and is shown in Fig. 8. Fig. 8a is a plot
of the time interval  of all transients in order of appearance. Fig. 8b
is a probability density function of the time interval . Fig. 8c is a
plot of the average time interval  of consecutive sets of 100 data
points against the number of that set. Fig. 8d is a plot of the average
time interval  of consecutive sets of 100 data points against the
average time of appears for that set.
Fig. 8b–d shows that the time intervals  between two adjacent
current transients becomes smaller with time.

4. Discussion

Building on from previous published work and the data pre-

sented in this paper, a more comprehensive model is proposed to
describe the current leakage phenomena in PZT samples. The data
is analysed with respect to a RC network model to describe the
current behaviour.

4.1. Transient RC behaviour model

The profile of these current transients behaviour can be

described in terms of an RC network. An equivalent circuit has
been developed to model this behaviour and is shown in Fig. 9.
There are three steps to the operation of this circuit correspond- Fig. 9. Current of i1 against time.
110 D. Zheng et al. / Sensors and Actuators A 158 (2010) 106–111
Fig. 10. Schematic of breakdown process.
When the switch closes at time t0 , the voltage across the capac-
itor V1 decreases from V0 to V0 × R2 /(R1 + R2 ), so the capacitor
undergoes a discharging process, the current i3 increases from
a negative value to zero, and the current through R2 decreases
from V0 /R2 to V0 /(R1 + R2 ). The i1 current is equal to the sum of
i2 and i3 , so this increases from zero to V0 /(R1 + R2 ), and the cur-
rent through the whole circuit i logarithmically increases from IL
to [IL + V0 /(R1 + R2 )].
Step 3: switch opening
When the switch opens at t1 , the voltage across the capacitor V1
increases back from V0 × R2 /(R1 + R2 ) to V0 . The capacitor under-
goes a charging process; the current i3 decreases from positive to
zero. So the current i then exponentially decreases and converges
on IL .
The “square” wave current transient behaves in a similar fashion
to the “shark fin” current transient except that step 2 is of longer
duration. The time duration of the square switching transient is
approximately 0.5 s, and the current amplitude is around 100 ␮A.
RC network suggests that a transient can be caused by a break-
down process modelled by R2 and the switch. Additionally if step
3 does not occur in any one transient then there is a mechanism to
increase the general leakage current as in Fig. 3. With an increase in
the number of such RC network in parallel with the switch closed,
a general increase in the leakage current can be modelled.
There is a decrease in the leakage current per transient (Fig. 6)
showing that not all the transients induce the same level of this
general current increase. This indicates that the R1 and R2 values
increase over the duration of the operation.
4.2. Transient electrical breakdown model
4.2.1. Filament formation
It is proposed that metal filaments form by growing from the
cathode towards the anode after the migration of metallic ions from
the anode to the cathode during the applied field [1]. The nickel
anode reacts with water vapour and loses electrons at the anode
producing these metallic ions. These metallic ions migrate to the
cathode and build metal filaments from the cathode to the anode
as illustrated in Fig. 10.
4.2.2. Electrical breakdown
Previous work [1] identified that an ionic mechanism was
responsible for the leakage current and that the electrode material
was a major factor. However, it was not conclusive over the role of
Ni in the breakdown process – indeed, similar effect was observed,
over a longer timescale, with Au electrodes. Ceramic porosity [2],
ionic contaminants, processing conditions or the generation of ions
by electrochemical action on the ceramic all have a role to play. Sim-
ilar processes have been observed with Ag electrodes with evidence
of energetic anodic breakdown processes [8]. Current fluctuations
and transients have also been observed by the authors on systems
with Ag electrodes, but these were only investigated in detail for Ni
electrodes. Porosity is also thought to be a significant factor in the
lifetime of these devices in harsh conditions [2], and experiments
(not reported here) show that reducing porosity improves lifetime,
but that similar breakdown characteristics are still observed. In all
these cases, the observations are supportive of a similar mecha-
nism to that proposed for Ag electrodes [8]. Metallic cations are
mobilised by the solution in adsorbed water, and migrate through
the ceramic to the cathode where they are reduced to metal. These
form anodic filaments which grow back towards the anode, and
eventually result in localised breakdown events and the current
transients which are the subject of this paper. Although metallic
Ag has been directly observed in the ceramic between electrodes,
there is no direct evidence of the metallic of conductive species in
many cases. This could well be because Ag presents an extreme case
of high ionic mobility, whereas the minimum amount of material
required to produce the observed conductivity could be extremely
small, making it extremely difficult to find through material sec-
tions. In all these cases very similar phenomena were observed, so
we think this mechanism of ionic conduction and breakdown to
be applicable to a wide range of conditions, although there may
be more than one possible root cause, or combination of causes.
The current transients studied here are a highly distinctive feature
of this process. By studying their characteristics we are therefore
able to gain new information on the structure and properties of
the conductive pathways that cannot be easily obtained by other
means.
Electrical breakdown occurs on a filament when it grows to a
length s between it and the anode resulting in an electrical field
across the length s which is sufficient for the breakdown of that gap.
During breakdown the current through the filament increases from
zero as in the RC network model (this would go towards explaining
the increase step of the current transients in Fig. 4). The leakage
current through the whole PZT will increase from an initial value
to the new leakage current.
However the electrical breakdown induced current increase
will collapse and leads to the current decrease back to the orig-
inal level. The electrical breakdown can be collapsed due to the
energy concentrates in a small breakdown area causing joule heat-
ing (burn-out) [9].
5. Conclusions
Detailed measurements of current transients in a PZT piezoelec-
tric material during exposure to humidity under d.c. bias conditions
are reported for the first time. Analysis of these transients shows
that the current amplitude remains constant during exposure. The
increase in overall current is related to the increasing frequency
of the transients. The transients show a form typical of a sudden
breakdown process followed by an RC discharge. Eventually the
current path becomes open circuit although the duration of the
transient is highly variable. An RC network model is used to esti-
mate the electrical characteristics of the conducting pathways. This
is used to estimate the physical dimension of the conducting path-
way.
An electrical breakdown mechanism proposed involves the
nickel anode reacting with water vapour producing metallic ions
which migrate to the cathode. Filaments of nickel grow from these
ions back towards the anode. When a length s remains between
metal filament and the anode, breakdown occurs. Then the break-
down ceases due to the joule heating (burn-out) in the breakdown
area.
In summary, the breakdown is addressed and a detailed
mechanism is suggested in this paper. For future studies, more
 D. Zheng et al. / Sensors and Actuators A 158 (2010) 106–111
piezoelectric materials and electrode materials are needed to be
investigated to validate the proposed mechanism as a generic
description of this type of breakdown.
Acknowledgements
The authors would like to acknowledge the support of the Tech-
nology Strategy Board and the UK National Measurement Office.
We would also like to acknowledge the support of APAHOE partners
Servocell Ltd., Diameter Ltd.
References
[1] I.P. Lipscomb, P.M. Weaver, J. Swingler, J.W. McBride, The effect of relative
humidity, temperature and electrical field on leakage currents in piezo-ceramic
actuators under dc bias, Sensors and Actuators A: Physical 151 (2009) 179–
186.
[2] I.P. Lipscomb, P.M. Weaver, J. Swingler, J.W. McBride, Micro-computer tomogra-
phy – an aid in the investigation of structural changes in lead zirconate titanate
ceramics after temperature-humidity bias testing, Journal of Electroceramics
(2008), August.
[3] B. Nagaraj, S. Aggarwal, T. Song, T. Sawhney, R. Ramesh, Leakage current mech-
anisms in lead-based thin-film ferroelectric capacitors, Physical Review B, APS
59 (1999) 16022–16027.
[4] B. Jaffe, W. Cook, H. Jaffe, Piezoelectric Ceramics, Academic Press, 1971.
[5] M. Dawber, K. Rabe, J. Scott, Physics of thin-film ferroelectric oxides, Reviews of
Modern Physics 77 (2005) 1083–1130.
[6] H.C. Ling, A.M. Jackson, Correlation of silver migration with
temperature–humidity-bias (THB) failures in multilayer ceramic capacitors,
IEEE Transactions on Components, Hybrids, and Manufacturing Technology 12
(1) (1989) 130–137.
[7] R. Mock, K. Lubitz, Piezoelectric Injection Systems Piezoelectricity, Springer
Series in Materials Science, vol. 114, Springer, 2008, pp. 299–310.
111
[8] J. Thongrueng, T. Tsuchiya, K. Nagata, Lifetime and degradation mechanism of
multilayer ceramic actuator, Japanese Journal of Applied Physics Part 1 – Regular
Papers Short Notes & Review Papers 9B (1998) 5306–5310.
[9] A. Ootani, T. Hori, M. Endo, M. Furuse, I. Yamaguchi, K. Kaiho, S. Yanabu, Break-
down of YBCO thin films during current limiting and methods of improving
current limiting performance, Electrical Engineering in Japan 166 (1) (2009)
515–521.
Biographies
Mr. Deyi Zheng Received the B.Sc. degree in material sciences and technology from
the Northeastern University, China, in 2003 and the M.Eng. degree in Materials with
Business from the Queen Mary College, University of London, UK, in 2003. Since
then, he has been studying as a PhD student in the School of Engineering Sciences,
University of Southampton. Research interests include micro-mechanism and appli-
cation of function materials, electromechanical devices and micro-fluidic energy
harvesting.
Dr. Jonathan Swingler Received the B.Sc. degree in physics and chemistry from the
University of Keele, U.K., in 1990 and the Ph.D. degree in the degradation of electri-
cal contacts under low frequency fretting conditions from Loughborough University,
Loughborough, U.K., in 1994. Since then, he has been pursuing research in the area
of electrical conductance of materials at the University of Southampton, Southamp-
ton, U.K., and is currently an academic in the Electro-Mechanical Research Group. He
lectures undergraduate Electrical Systems, Electromechanical Machine, and Auto-
motive Electronics. Dr. Swingler is a Chartered Scientist, Chartered Physicist, and a
Member of the Institute of Physics.
Dr. Paul Weaver is principal research scientist with the UK’s National Physical
Laboratory’s multi-functional materials research group. Research interests include
the development of novel electromechanical devices and the application of piezo-
electrics and ferroelectrics for sensing and actuation. He is a visiting Reader at
Southampton University and holds an MA in Natural Science from Cambridge Uni-
versity, and a PhD (Aeronautics) from Southampton University. He is a chartered
engineer and member of the IET.