Invited Paper

Review of pyroelectric thermal energy harvesting and new MEMs

based resonant energy conversion techniques
Scott R. Huntera*, Nickolay V. Lavrika,b, Salwa Mostafac,
Slo Rajica and Panos G. Datskosa
a
Measurement Science and Systems Engineering Division, Oak Ridge National Laboratory,
P.O. Box 2008, Oak Ridge, Tennessee 37831-6054
b
Center for Nanophase Materials Sciences Division, Oak Ridge National Laboratory,
P.O. Box 2008, Oak Ridge, Tennessee 37831-6054
c
Dept. of Electrical Engineering and Computer Science, University of Tennessee, Knoxville,
Tennessee 37996-1600
ABSTRACT
Harvesting electrical energy from thermal energy sources using pyroelectric conversion techniques
has been under investigation for over 50 years, but it has not received the attention that thermoelectric energy
harvesting techniques have during this time period. This lack of interest stems from early studies which
found that the energy conversion efficiencies achievable using pyroelectric materials were several times less
than those potentially achievable with thermoelectrics. More recent modeling and experimental studies have
shown that pyroelectric techniques can be cost competitive with thermoelectrics and, using new temperature
cycling techniques, has the potential to be several times as efficient as thermoelectrics under comparable
operating conditions. This paper will review the recent history in this field and describe the techniques that
are being developed to increase the opportunities for pyroelectric energy harvesting.
The development of a new thermal energy harvester concept, based on temperature cycled
pyroelectric thermal-to-electrical energy conversion, are also outlined. The approach uses a resonantly
driven, pyroelectric capacitive bimorph cantilever structure that can be used to rapidly cycle the temperature
in the energy harvester. The device has been modeled using a finite element multi-physics based method,
where the effect of the structure material properties and system parameters on the frequency and magnitude of
temperature cycling, and the efficiency of energy recycling using the proposed structure, have been modeled.
Results show that thermal contact conductance and heat source temperature differences play key roles in
dominating the cantilever resonant frequency and efficiency of the energy conversion technique. This paper
outlines the modeling, fabrication and testing of cantilever and pyroelectric structures and single element
devices that demonstrate the potential of this technology for the development of high efficiency thermal-to-
electrical energy conversion devices.
Key Words: Energy harvesting, pyroelectric, bimorph cantilever, MEMS, surface micromachining
1. INTRODUCTION
1.1. Background
Presently, the U.S. uses approximately 100 Quads (∼1020Joules) of energy from all sources (primarily
petroleum, natural gas, coal and nuclear) with 40% of this energy used to generate electricity. This energy is
used to power the transportation, industrial, commercial residential sectors of the economy, and unfortunately,
over 55% of this energy is lost as low grade waste heat1. Most of this heat is generated at temperatures only a
few degrees above ambient, and the laws of thermodynamics severely limit the amount of this energy that can
be used to do useful work, such as generate electricity. On the brighter side, there are many sources of waste
heat that are generated with comparatively large temperature difference above ambient from 10s to 100s 0C
(e.g. automotive exhausts, fossil fueled power generation, petrochemical plants, metals smelting, and glass
huntersr@ornl.gov: 865-241-3995

Energy Harvesting and Storage: Materials, Devices, and Applications III,

edited by Nibir K. Dhar, Priyalal S. Wijewarnasuriya, Achyut Dutta, Proc. of SPIE Vol. 8377,
83770D · © 2012 SPIE · CCC code: 0277-786X/12/$18 · doi: 10.1117/12.920978

Proc. of SPIE Vol. 8377 83770D-1

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/21/2013 Terms of Use: http://spiedl.org/terms

 and paper manufacturing), and these temperature gradients can be used to generate commercially viable
electrical power under certain conditions. Other opportunities exist for active cooling and electrical power
generation for sensor systems on much smaller scales, such as on-chip active heat sinks, in standalone
computer systems and computer data processing centers.

Table 1. Energy Harvesting Techniques and Scavenged Power Densities2

Table 1 shows some of the more common techniques discussed in the literature to extract useful
amounts of energy from the environment, along with the energy densities that are potentially achievable with
these scavenging techniques2. The actual power available from these sources is generally quite small, except
perhaps from bright sunlight. Only the thermoelectric and pyroelectric power conversion techniques listed in
the table can be used to directly convert thermal energy to electricity, but even after several decades of intense
research, thermoelectrics are still limited to 1-5% efficiencies for realistic temperature gradients. Only the
relatively unexplored pyroelectric energy harvesting technique offers the possibility of conversion efficiencies
(and power densities) that are several time those presently achievable using thermoelectrics.
Recent modeling and experimental measurements indicate that pyroelectric thermal energy
scavenging has the potential for system level conversion efficiencies in the 10-20% range with Carnot
efficiencies approaching 50-80%. Furthermore, many pyroelectric materials are stable up to temperatures
approaching 12000C, enabling energy harvesting from high temperature sources with a much improved
thermodynamic efficiency. The primary advantage of thermoelectric
energy conversion techniques is that these devices only require a
steady state temperature gradient to operate (Figure 1a) and (a)
consequently possess no moving parts that can limit device reliability
and life. Pyroelectric converters only generate electricity when there is
a change in temperature across the device (Figure 1b), and one of the
significant challenges with the technique is to efficiently generate
rapidly time varying temperature gradients across the device from (b)
waste energy streams that possess constant or very slowly time varying
temperature gradients. Electrical current, and hence power, generation
is directly proportional to the rate of change in temperature across the
pyroelectric material, and one of the significant advantages of the
present resonantly driven pyroelectric converter is that we have
demonstrated rates of change in the temperature across a pyroelectric Figure 1. Schematics of (a) a
0 thermoelectric and (b) a pyroelectric
cantilever structure of ∼ 1000 C/sec – orders of magnitude greater than energy converter.
previously fabricated pyroelectric converters.

Proc. of SPIE Vol. 8377 83770D-2

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/21/2013 Terms of Use: http://spiedl.org/terms

 1.2 Pyroelectric Electrical Energy Generation
Pyroelectric energy generators rely on the property that the spontaneous polarization (and hence
dielectric constant) of certain materials is temperature dependent. The temperature dependent nature of the
dipole moment of pyroelectric materials has a long history where ancient treatises record that certain materials
(including tourmaline) attracted dust and straw when heated3. These materials were rediscovered in the
middle ages with the first true scientific investigations of the phenomena occurring in the 1700s. These
investigations continued through the 1800s and early 1900s,
until in 1938 a paper was published proposing the use of (a)
tourmaline crystals as current generating IR sensors in
spectroscopy applications, leading to the new field of infrared
detection which continues to this day3.
The non zero spontaneous polarization PS of
pyroelectric materials at room temperature causes the material (b)
to attract charged particles (Figure 2a). If the material is
placed between two electrodes and made the dielectric
material within a capacitor and connected to an external
circuit, then the plates of the capacitor will charge until the
(c)
surface charge on the pyroelectric material is neutralized
(Figure 2b). When the temperature across the capacitor is
raised, the dipoles within the material will begin to randomize,
reducing the polarization PS, and hence dielectric constant of
the capacitor, and a current will flow in the external circuit as (d)
long as the temperature across the capacitor is changing
(Figure 2c). A current will flow in the opposite direction once
the temperature is reduced as the dipoles in the material
realign (Figure 2d). Rapidly cycling the temperature across Figure 2. A pyroelectric material used as the
the capacitor induces an alternating current in the external dielectric in a capacitor. As the temperature of the
circuit, with the magnitude of the electrical current and energy capacitor is increased, a current flows in the
external circuit to compensate for the change in
conversion efficiency dependent on the rate of change in the bound charge at the edges of the crystal3.
temperature.
The efficiency of the thermal to electrical energy conversion process in a pyroelectric energy
harvester (and all thermal engines) is thermodynamically limited by the Carnot efficiency, ηCarnot:

ηCarnot = 1−TL /TH (1)

where TH is the temperature of the heat source and TL is the temperature of the heat sink. Consequently, a
thermal gradient of 10C at room temperatures leads a maximum Carnot efficiency of 0.33%, while a 1000C
temperature gradient gives a Carnot efficiency of 25%. Large temperature gradients not only lead to greater
power generating capacity, they also lead to higher system power conversion efficiencies. Thermal energy
conversion only becomes economically viable when source and sink temperature differences are greater than
a few 10s of degrees except under exceptional conditions where power generation is at a premium.
In real world thermal energy converters, there are additional sources of power loss that reduce the
overall system efficiency that can be achieved with the device. Pyroelectric thermal power generators
convert heat (Qin) into electrical power (Wout) with efficiency:

(2)

Proc. of SPIE Vol. 8377 83770D-3

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/21/2013 Terms of Use: http://spiedl.org/terms

 where WE is the generated electrical power, WP is the power lost in the temperature cycle, Cν is the heat
capacity of the pyroelectric device, QInt are the intrinsic heat losses in the thermal cycle and QLeak are the heat
leakages between the hot and cold sources.
Presently contemplated thermal to electrical power conversion techniques (thermoelectric and
pyroelectric) all suffer from low power conversion efficiencies, limited partly by the Carnot efficiency, but
also by the inherent limitations of the conversion technologies themselves. Pyroelectric converters remain
relatively unexplored, as early attempts in the 1960s to model and fabricate converters based on pyroelectric
operating principles gave uneconomically low conversion efficiencies (0.1-2%)4-6. The first proposed thermal
energy conversion cycle was devised by Clingman4,5, but modeled energy conversion efficiencies were well
below 1% as the energy for temperature cycling was higher than that which could be extracted from the
energy conversion cycle. A later modeling effort by van der Ziel6 in 1974 to convert solar energy to electrical
energy using a pyroelectric conversion cycle proved equally disappointing with predicted efficiencies < 0.1%.
Other contemporary modeling studies were much more encouraging however, with overall predicted power
conversion efficiencies of 10-15%7 and 20-40%8-10, with Carnot efficiencies in the range 50-80%8-12. In
contrast, thermoelectric generators have maximum Carnot efficiencies around 14-17%13-15 and overall
efficiencies around 5%15.

1.3 Ericsson Thermal Energy Conversion Cycle

The quasi-isothermal cycle shown in Figure 2 and the other early thermal conversion processes
mentioned above are very inefficient and, as a result, produce very little power. However, if the pyroelectric
capacitor is made part of a thermal engine, then much higher thermal-to-electrical conversion efficiencies and
output powers are achievable. The field was reinvigorated in the 1980’s by Olsen and his colleagues15-23 who
published a series of modeling and experimental results showing that much higher conversion efficiencies
were possible using a modified Ericsson thermal engine conversion process. The cycle Olsen developed is
shown schematically in Figure 3 and works by allowing large temperature swings across the pyroelectric
capacitor while applying alternating voltages on the capacitor electrodes.
The cycle consists of two isotherms (TL and
TH) and two constant electric fields (V1 and V2), and
TL starts at (a) in Figure 3 with the pyroelectric capacitor
at low temperature TL and the ferroelectric capacitor
charged at high voltage V2. As the temperature
increases to TH at a constant applied voltage (b), the
dielectric constant of the pyroelectric capacitor
decreases, and charge is forced to flow in the external
circuit charging the storage capacitor (Figure 3). The
TH applied voltage is then reduced to V1 at (c) and the
temperature of the pyroelectric capacitor decreased to
TL again (d), producing another, opposite sign, current
flow in the external circuit. Other thermal engine
Figure 3. The Ericsson temperature and voltage cycle
cycles include Rankin and Stirling cycles, and are
developed by Olsen15-23 to generate electrical energy in the used in steam power plants, internal combustion
pyroelectric thermal energy converter. engines and refrigerators for example.

The pyroelectric current Ip produced during the cycle shown in Figure 3 is:
dPS dT
I p = Af = Af p (3)
dt dt

Proc. of SPIE Vol. 8377 83770D-4

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/21/2013 Terms of Use: http://spiedl.org/terms

 where Af is the surface area of the pyroelectric thin film capacitor, PS (C/m2) is the pyroelectric thin film
polarization, T is the pyroelectric capacitor temperature and p is the pyroelectric coefficient. The net output
power Np from the pyroelectric capacitor is:
dT
N p = Vappl I p = Vappl pA f (4)
dt
where Vappl is the external applied voltage across the pyroelectric capacitor. The cumulative pyroelectric
conversion output work Wout from the cycle is as follows:
dT (5)
W out = ∫V appl dq = ∫N p dt = ∫ Vappl pA f
dt
dt
Equation 5 is shown schematically in Figure 3 where Wout is the integral over the area within the figure. The
greater the change in applied voltage across the pyroelectric capacitor and the wider the temperature swing,
the larger the amount of heat energy converted to useful electrical energy. Equations 3 and 5 also show that
the amount of current and electrical energy generated by this circuit is dependent on the magnitude of the
pyroelectric coefficient p, the size of the capacitor (plate area A), and very importantly, on the rate of change
in the temperature across the pyroelectric capacitor. Hence the faster the temperature can be cycled back and
forth across the device, the more efficient the energy conversion process is and the greater the amount of
electrical energy generated.
Real pyroelectric materials, however,
do not exhibit the ideal behavior shown in
Figure 3. A more typical pyroelectric
response curve is shown in Figure 4, where
the Ericsson electrical conversion cycle is
superimposed on the charge-voltage hysteresis
curve of the pyroelectric material
polyvinylidene fluoride (PVDF)19. Poling
hysteresis of the PVDF material as the
temperature is cycled across the capacitor
reduces the conversion efficiency as does the
need to operate over a limited temperature
range near, but below the material Curie
temperature - the phase transition temperature
at which the material switches from a
ferroelectric to a paraelectric state. Electrical
energy is generated when the device is
operated within the shaded area bounded by
1-2-3-4 in Figure 4 and is given by Eq. 5.
Figure 4. The Ericsson electrical conversion cycle superimposed
The first systematic attempts to
on the charge-voltage hysteresis curve for the ferroelectric
develop full scale pyroelectric energy material polyvinylidene fluoride19.
converters were by Olsen and his colleagues
in the 1980s15-23, who were able to achieve power conversion efficiencies up to 0.5%. The working principle
of their devices is shown schematically in Figure 5. Cooled fluid was pumped through a vertical stack of
pyroelectric capacitor plates and this fluid was heated at the top of the stack. The heated fluid was then
passed through the capacitor stack where it was cooled and the process repeated. Alternately heating and
cooling the plates and applying low and high voltages across the capacitor plates at appropriate points in the
conversion cycle generated an AC current in the external circuit. Figure 4 shows the input power Qin and
power generated from the capacitor system WE (blue circle) as a function of the frequency with which the hot

Proc. of SPIE Vol. 8377 83770D-5

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/21/2013 Terms of Use: http://spiedl.org/terms

 Figure 5. Operating of several previous attempts to Figure 6. The results of a recent modeling study24 to
develop efficient thermal to electric energy understand the operation and power losses from the Olsen
generators using pyroelectric materials15-23. and colleagues experiments15-23.
and cold fluid was cycled across the capacitor plates. These measurements are reprinted from a recent paper24
of a modeling study done to understand the operation and power loss processes in the Olsen work. Also
plotted in Figure 6 is the calculated power required to cycle the fluid across the capacitor plates WP (red
circle). These measurements show that as the operating frequency is increased, the output power increases
approximately linearly at first, as expected, but the power expended to cycle the fluid through the system
increases exponentially, such that at around 0.2 Hz, the power required to cycle the heat transfer fluid
approximately equals the output power from the generator, and the net power generated Wout is zero.
However, this modeling study did identify operating regimes where system efficiencies up to 3.5% were
possible with this technique24.
Table 2. Conversion efficiencies from previous pyroelectric energy harvester studies

Although a more recent attempt by Ikura and his colleagues25-29 to develop pyroelectric energy
converters by alternately shuttling hot and cold fluids across sets of pyroelectric capacitor plates achieved
much higher power conversion efficiencies (up to 5.7% system conversion efficiency with a Carnot efficiency
of 37.5%), the fabricated device suffered from the same low power generation problems25-29. The results
obtained from the Olsen and Ikura studies are summarized in Table 2 along with other recent studies by

Proc. of SPIE Vol. 8377 83770D-6

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/21/2013 Terms of Use: http://spiedl.org/terms

 Guyomar and colleagues11,12 and Pilon and his colleagues24,30-33. All these studies have similar problems and
low power conversion efficiencies due to low operating frequencies (< 1 Hz), large power requirements to
generate significant temperature cycles (Wp in Equation 2), large thermal mass capacitor systems with
relatively low breakdown strengths (i.e. low voltage differences, V2-V1) and low thermal conductivities
leading to low ΔT/Δt, and hence low ΔQ/ΔT7-12,15-38.

The MEMS based pyroelectric power generator outlined in this paper operates at much higher
frequencies (10s of Hz to 100s Hz), using thin film structures with low thermal masses and comparatively
high dielectric strengths, and high thermal conductivities (giving fast ΔT/Δt and hence large ΔQ/ΔT). The
innovative use of the heat source to power the temperature cycling through the converter using bimaterial heat
sensitive structures, and use of resonantly driven cantilever motion to rapidly move the converter through the
temperature cycle leads to high efficiency operation (i.e. WP ≈ 0 in Equation 2). Encapsulating the generator
in a partially evacuated enclosure also minimizes heat losses through gas convection and conduction
processes (i.e. QLeak ≈ 0 in Equation 2). Consequently, expected conversion efficiencies of a fully optimized
converter will be as high as 80-90% of the Carnot limit.

2. MEMS BASED PYROELECTRIC THERMAL ENERGY HARVESTER

2.1 Energy Harvester Concept
The structure of the resonating pyroelectric capacitor thermal energy converter is shown
schematically in Figure 7, with the device concept outlined in a previous paper40. Each energy converter
structure is a few hundred µm to several mm in length and width. The cantilevered pyroelectric capacitor
structure is shown in detail in Figure 8, and is fabricated with two metal films, which act as the electrodes of a
capacitor, and a pyroelectric material (e.g. copolymer of poly-vinylidene fluoride P(VDF-TrFE)) which acts
as the dielectric between the metal electrodes. Two additional small proof masses may be located at the ends
of the cantilever to increase the thermal mass of the structure and to make good thermal contacts with the hot
and cold surfaces. The cantilever can be anchored to either the hot or cold surfaces. A split anchor also
provides the capacitor electrical contacts to the external charge extraction and control circuitry (Figure 8).
These millimeter sized converters can readily be scaled up to much larger devices using 2D arrays of
individual converter elements. Arrays of up to 106 converters can be fabricated, and these arrays can
themselves be stacked to harvest electrical energy from much larger thermal waste energy sources.

Figure 7. Schematic of a pyroelectric energy harvester device, Figure 8. Schematic showing the details of the
consisting of a bimaterial cantilever structure, which alternately contacts construction of the capacitive cantilever thermal
the hot and cold surfaces, generating an electrical current in the energy converter and split anchor structures.
pyroelectric capacitor.

Proc. of SPIE Vol. 8377 83770D-7

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/21/2013 Terms of Use: http://spiedl.org/terms

 2.2 Energy Harvester Operation
The harvester operation is shown schematically in upper images in Figure 9. The cantilever structure
initially heats through the anchor, causing the cantilever to bend towards the lower cold heat sinked surface.
On contacting the cold surface, the structure rapidly loses heat, and bends back towards the upper heated
surface. On contacting the upper surface, it rapidly heats and bends away from the upper surface and again
makes contact with the lower surface and the cycle is repeated. Good thermal contact with the hot and cold
surfaces is essential to maximize the transfer of thermal energy (and thus to raise the cantilever temperature
substantially) to the pyroelectric capacitor on the cantilever.
An approximate timing diagram
for the operation of this electro-mechanical
circuit is shown in the lower images in
Figure 9. The top trace shows the
temperature in the pyroelectric capacitor as
the cantilever alternately contacts first the
lower cold surface, then the upper hot
surface. The thermal response time is
dependent on the thermal contact resistance
between the proof masses and the hot and
cold surfaces, and heat capacity of the
cantilever and capacitor structures. The
next lower trace shows the change in
pyroelectric capacitance in response to the
temperature change.
Also shown in Figure 9 is the
timing of the applied voltages across the
capacitor to implement the Ericsson
thermal energy extraction cycle. The
letters correspond to the points on the
temperature-voltage cycle diagram shown
in Figure 3. The field is switched just prior
Figure 9. Operation of the resonating cantilever structure shown in the
to the cantilever contacting the hot or cold upper images, with an approximate timing diagram (lower graphs)
surfaces. The final curve in Figure 9 showing the change in pyroelectric capacitor temperature, capacitance,
shows the expected rectified current voltage and resultant output current shown as a function of two full
extracted from the electrical circuit. temperature cycles.

3. ENERGY HARVESTER OPERATION MODELING

Extensive modeling of the mechanical and heat transfer behavior of a bimaterial cantilever placed in
proximity of a heat source was performed using finite element analysis (FEA) techniques in a multiphysics
COMSOL modeling package. The models combined solid mechanics, heat transfer physics and electrostatics
selected from the standard set of COMSOL multiphysics modules. A 2D model was used in order to maintain
a reasonably low number of degrees of freedom to be solved while defining mesh sizes sufficiently small to
accurately reflect the smallest characteristic features of our structures. Typically, the models consist of up to
1000 mesh elements and up to 10,000 degrees of freedom solved. The meshing around the contact point
between the cantilever and the heat source is particularly important and is shown in detail in Figure 10. Some
of the results from this study were published in a previous conference paper40.
The intermittent thermal contact between the heat source and the top cantilever surface was modeled
by defining the two convex surfaces as an identity pair with heat continuity boundary conditions. While this

Proc. of SPIE Vol. 8377 83770D-8

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/21/2013 Terms of Use: http://spiedl.org/terms

 represents a somewhat crude approximation of the highly complex behavior of
the actual thermal contacts between two solids surfaces, it is expected to describe
the experimental system reasonably well by empirically adjusting the thermal
conductivity of the region in vicinity of the thermal contact (Figure 10). In these
simulations, the thermal conductivity across the contact region was varied in the
range of 0.3 to 30 W/mK.
The complete cantilever structure modeled in these simulations is shown
in Figure 11. The cantilevered thermal conduction structure was a bimorph
composed of 1-10 μm silicon dioxide and 0.1-0.3 μm Al layers. The other
variables in the model included cantilever length, L = 0.5 mm to 5mm, cantilever
Figure 10. Model geometry, width from 0.5 to 2 μm, and temperature differences between the cantilever and
mesh and mesh quality in the
vicinity of the thermal
the heat source, ΔT = 10, 50 and 150 K. An optional force could also be applied
contact between the to the cantilever tip in the vertical direction, either up or down. The latter was
cantilever (bottom) and heat used to mimic the effect of electrostatic interactions that can be introduced into
source (top). our experimental system in order to control the self-oscillatory behavior.
Figures 12 and 13 show typical modeling Contact area with variable thermal
results obtained from the COMSOL simulation. Figure conductivity, k cont Heat Source
12 shows the cantilever tip displacement and the Cantilever
anchor/clamp
temperature of the cantilever at the point of contact Aluminum
with the heat source as the cantilever repeatably makes
contact with heat source. Figure 13 shows the change SiO2
in cantilever temperature at the contact point and at two
Cantilever
points along the cantilever toward the base for the same
simulation conditions. The initially large temperature
fluctuations near the cantilever tip rapidly decrease in
Figure 11. Model geometry with a color scale
magnitude away from the point of contact with the heat corresponding to thermal conductivity of the components
source. The cantilever is oscillating at around 50 Hz used in our model.
under these conditions.

Figure 12. Cantilever tip displacement (blue) and Figure 13. Temperature changes analyzed at the cantilever
temperature (red) analyzed at the cantilever tip contact point tip contact point and at points 200 µm and 500 µm from the
as a function of time for ΔT=150K, kcont = 3 W/mK and tip as a function of time for ΔT=150K, kcont = 3 W/mK and
L=1mm. L=1mm.
Figures 14 and 15 summarize the expected maximum power that can be generated with a pyroelectric
device operating under the resonant temperature cycling conditions modeled above. These results were
generated for a device with a pyroelectric material having a pyroelectric coefficient ρ = -3.5 μC/m2K (e.g.

Proc. of SPIE Vol. 8377 83770D-9

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/21/2013 Terms of Use: http://spiedl.org/terms

 P(VDF-TrFE)), and cantilever dimensions 500 μm long, 125 μm wide and 25μm thick. For a device
operating with a temperature difference of 70K (Figure 14) and a temperature cycling frequency of 80Hz
(Figure 15), then each harvester element can deliver up to 1mW with a power density of 400W/L. A
significant advantage with the present technique is that the harvester can be readily scaled to a much larger
device using microfabrication techniques. A device containing a 2D array of 100 x 100 of these individual
converters could generate up to 10 Watts under these conditions.

Figure 14. Modeled power output as a function cycled Figure 15. Modeled power output as a function of
temperature difference at a temperature cycling frequency temperature cycling frequency with a 70K temperature
of 80 Hz. difference.
These simulations demonstrated the complexity of the task required to accurately simulate and predict
the operation and efficiency of the heat transfer process. Similar calculations are required to understand
operation of the device once the pyroelectric energy conversion capacitors are incorporated in the cantilever
structures. These calculations will be performed in subsequent studies.

4. FABRICATION AND OPERATION OF RESONATING TEMPERATURE CYCLING

CANTILEVER STRUCTURES
4.1 Device Characterization
Bimorph and trimorph cantilever
structures were fabricated and tested to
demonstrate the self resonating and
temperature cycling operation of these
structures. Structures were fabricated with the
pyroelectric materials aluminum nitride and
P(VDF-TrFE), or silicon dioxide, and
incorporated Al electrodes which also acted as
one of the bimorph elements. Representative
cantilever structures are shown in Figure16 Figure 16. A 1 mm wide, 4 mm long self resonating cantilever
and were fabricated with dimensions from 0.5 structure (left) and a small array of these structures (right).
to 7 mm in length and 0.5 to 2 mm in width.
A vacuum test chamber and data collection system were assembled to temperature cycle the bimorph
test cantilever structures and to characterize cantilever thermal heat transfer structures and the pyroelectric
and electrical current generating properties of pyroelectric AlN and P(VDF-TrFE) capacitive structures. The
chamber was pumped to vacuum pressures in the range 20-50 mTorr, which is sufficient to eliminate heat loss

Proc. of SPIE Vol. 8377 83770D-10

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/21/2013 Terms of Use: http://spiedl.org/terms

 by gas conduction and convection in the vacuum chamber. The test setup is shown in Figure 17, and in
addition to the vacuum chamber, consists of a Labview controlled temperature controller and thermoelectric
(TE) cooler, a firewire video camera used to obtain video images of the moving cantilever structures
(examples are shown in Figure 19), and a precision three dimensional translation stage that can be used to
accurately position the cantilever structures between the heat source and sink. A cantilever position sensing
laser and position sensitive detector (PSD), which can be used to monitor the motion of the tip of the
cantilever structures as a function of heat source temperature40 is also shown in the left hand image of Figure
17. The right hand image of Figure17 is an enlarged view of the cantilever setup inside the chamber,
showing the bimorph cantilever mounted on the TE cooler, with the cold heat sink mounted directly above tip
of the cantilever.

Figure 17. (Left) Vacuum test chamber and optical diagnostics used to characterize the performance of the resonating
cantilever structures and (Right) a magnified image of the cantilever, TE cooler and cooled heat sink.
Figure18 shows two image frames taken by the video camera showing the cantilever oscillating with
a total tip displacement of ∼ 10μm. Finite element modeling of the cantilever motion shows a similar
cantilever tip displacement of ∼ 10μm when the oscillation has achieved steady state. These results are shown
in Figure 19, along with the modeled change in cantilever temperature as a function of time. The peak
temperature change is 15K in agreement with the estimated cantilever temperature change based on the
measured cantilever responsivity. The cantilever was oscillating at ∼ 20Hz under these conditions which
implies a rate of temperature change dT/dt ∼ 300K/sec. This is orders of magnitude greater than has been
achieved by the pyroelectric converters discussed in Section 1, and leads to the possibility of much more
efficient thermal-to-electrical energy conversion with higher power densities than previously achievable.

Figure 18. Two images taken from a short video showing the cantilever Figure 19. Modeled tip displacement and
oscillating with a tip displacement of ∼ 10μm. cantilever temperature.

Proc. of SPIE Vol. 8377 83770D-11

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/21/2013 Terms of Use: http://spiedl.org/terms

 5. DISCUSSIONS AND CONCLUSIONS
In this paper, the history and present state-of-the-art in the modeling and fabrication of pyroelectric
energy harvester devices has been summarized. This field has been relatively unexplored until recently due
primarily to the disappointing energy conversion efficiencies obtained from early modeling and experimental
studies. More recent device development studies have shown system electrical conversion efficiencies
comparable to those obtained using thermoelectric techniques, while recent modeling studies show potential
efficiencies close to the Carnot limit giving system efficiencies that are potentially several times those
obtained with competing techniques.
The development of a new innovative concept for the conversion of low temperature thermal waste
heat into electrical energy using
electromechanically resonating MEMS
based microstructures was also discussed.
This technique gives far greater rates of
temperature change across the pyroelectric
material than the previously discussed
studies were able to achieve. These
millimeter sized energy harvester structures
can be fabricated in 2D arrays and scaled
up in size allowing the devices to generate
sizable amounts of electrical power for
many applications. The layout work for a
small 2D pyroelectric self resonating
cantilever array is shown in Figure 20.
These studies indicate we can achieve our
goal of developing a compact, efficient
thermal to electrical energy pyroelectric
Figure 20. Mask and layout for a small 2D array of self resonating
harvester using electromechanical pyroelectric cantilevers.
resonating structures.
The technical challenges ahead for this technology, in being able to generate substantial amounts of
electrical energy, lie primarily in achieving good thermal conductivity across the hot and cold thermal
contacts. Large heat flow across the energy converter is essential for heat sinking and efficient power
generation applications. These are areas of active research and development by our group.

ACKNOWLEDGEMENTS
Research sponsored by the Laboratory Directed Research and Development program at the Oak
Ridge National Laboratory, managed and operated by UT-Battelle, LLC, for the U. S. Department of Energy
under Contract No. DE-AC05-00OR22725.

REFERENCES
[1] Lawrence Livermore National Laboratory, “Estimated U.S. energy use in 2009,”
https://flowcharts.llnl.gov/
[2] J. A. Paradiso and T. Starner, “Energy scavenging for mobile and wireless electronics,” Pervasive
Computing, IEEE, 4, pp. 18-27, (2005).
[3] Lang, S., “Pyroelectricity: From ancient curiosity to modern imaging tool”, Phys. Today 58, 31-36
(2005).
[4] Clingman, W. H. and R. G. Moore, "Application of ferroelectricity to energy conversion process," J.
Appl. Phys, vol. 32, pp. 675-681 (1961).

Proc. of SPIE Vol. 8377 83770D-12

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/21/2013 Terms of Use: http://spiedl.org/terms

 [5] Fattuzzo, E., H. Kiess and R. Nitshe, “Theoretical efficiency of pyroelectric power converters”, J.
Appl. Phys. 37, 510-516 (1966).
[6] van der Ziel, A., “Solar power generation with the pyroelectric effect”, J. Appl. Phys. 45, 4128
(1974).
[7] Gonzalo, A., “Ferroelectric materials as energy converters”, Ferroelectrics 11, 423-430 (1976).
[8] Drummond, A.D., “Dielectric power conversion”, Abstracts 10th Intersociety Energy Conversion
Engineering Conf., Newark, DE, 950 (1975).
[9] Drummond, A.D., “Dielectric thermal power converter”, U.S. Patent 4,220,906 (1980).
[10] Drummond, A.D., “Demonstration of a high power density electrocaloric heat engine”, Ferroelectrics
27, 213-218 (1980).
[11] Sebald, G., E. Lefeuvre and D. Guyomar, “Pyroelectric energy conversion: optimization principles”,
IEEE Trans. Ultrasonics, Ferroelectrics, & Frequency Control 55, 538-551 (2008).
[12] Guyomar, D., G. Sebald, E. Lefeuvre and A. Khodayari, “Toward energy harvesting using
pyroelectric material”, J. Intell. Mat. Sys. Struct. 20, 265-271 (2009).
[13] Bierschenk, J., “Optimized thermoelectrics for energy harvesting applications”, in Energy Harvesting
Technologies, (S. Priya and D. J. Inman, Eds.) Springer, New York, Chpt. 12, 337-351 (2009).
[14] Snyder, J.G., “Thermoelectric energy harvesting”, in Energy Harvesting Technologies, (S. Priya and
D. J. Inman, Eds.) Springer, New York, Chpt. 11, 323-336 (2009).
[15] Olsen, R.B., J. M. Briscoe, D. A. Bruno and W. F. Butler, “A pyroelectric converter which employs
regeneration”, Ferroelectrics 38, 975-978 (1981).
[16] Olsen, R.B., “Ferroelectric conversion of heat to electrical energy – a demonstration”, J. Energy 6, 91
(1982).
[17] Olsen R.B. and D. D. Brown, “High efficiency direct conversion of heat to electrical energy-related
pyroelectric measurements”, Ferroelectrics 40, 17-27 (1982).
[18] Olsen, R.B. and D. Evans, “Pyroelectric energy conversion: hysteresis loss and temperature
sensitivity of a ferroelectric material”, J. Appl. Phys. 54, 5941-5944 (1983).
[19] Olsen, R.B., J. C. Hicks, M. G. Broadhurst and G. T. Davis, “Temperature dependent ferroelectric
hysteresis study in polyvinylidene fluoride”, Appl. Phys. Lett. 43, 127-129 (1983).
[20] Olsen, R.B., D. A. Bruno, J. M. Briscoe and J. Dullea, “Cascaded pyroelectric energy converter”,
Ferroelectrics 59, 205-219 (1984).
[21] Olsen, R.B., D. A. Bruno and J. M. Briscoe, “Pyroelectric conversion cycles”, J. Appl. Phys. 58,
4709-4716 (1985).
[22] Olsen, R.B., D. A. Bruno, J. M. Briscoe and E. W. Jacobs, “Pyroelectric conversion cycle of
vinylidenefluoride-trifluoroethylene copolymer”, J. Appl. Phys. 57, 5036-5042 (1985).
[23] Olsen, R.B. “Pyroeletric energy converter element comprising vinylidene fluoride-trifluoroethylene
copolymer”, U.S. Patent 4,620,262 (1986).
[24] Vanderpool, D., J. H. Yoon and L. Pilon, “Simulations of a prototype device using pyroelectric
materials for harvesting waste heat”, Int. J. Heat Mass Transfer 51, 5052-5062 (2008).
[25] Kouchachvili, L. and M. Ikura, “Pyroelectric conversion – Effect of P(VDF-TrFE) preconditioning on
power conversion”, J. Electrostatics 65, 182-188 (2007).
[26] Kouchachvili, L. and M. Ikura, “Improving the efficiency of pyroelectric conversion”, Int. J. Energy
Res. 32, 328-335 (2008).
[27] Kouchachvili, L. and M. Ikura, “High performance P(VDF-TrFE) copolymer for pyroelectric
conversion”, U.S. Patent 7,323,506 (2008).
[28] Ikura, M., “Pyroelectric conversion system”, U.S. Patent 6,528,898 (2003).
[29] Ikura, M., “Conversion of low-grade heat to electricity using pyroelectric copolymer”, Ferroelectrics
267, 403-408 (2002).
[30] Navid, A., D. Vanderpool, A. Bah and L. Pilon, “Towards optimization of a pyroelectric energy
converter for harvesting waste heat,” Int. J. Heat Mass Trans. 53, 4060-4070 (2010).

Proc. of SPIE Vol. 8377 83770D-13

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/21/2013 Terms of Use: http://spiedl.org/terms

 [31] Nguyen, H., Navid, A. and L. Pilon, “Pyroelectric energy converter using co-polymer PVDF-TrFE
and Olsen cycle for waste heat energy harvesting,” Appl. Therm. Engin. 30, 2127-2137 (2010).
[32] Navid, A. and L. Pilon, “Pyroelectric energy harvesting using Olsen cycles in purified and porous
PVDF-TrFE thin films,” Smart Mater. Struct. 20, 025012 (2011).
[33] Lee, F. Y., S. Goljahi, I. M. McKinley, C. S. Lynch and L. Pilon, “Pyroeletric water heat energy
harvesting using relaxor foerroeletric 8/65/35 PLZT and the Olsen cycle,” Smart Mater. Struct., 21,
025021 (2012).
[34] Ujihara, M., G. P. Carman and D. G. Lee, “Thermal energy harvesting device using ferromagnetic
materials”, Appl. Phys. Lett. 91, 093508 (2007).
[35] Claude, J., Y. Lu, K. Li and Q. Wang, “Electrical storage in poly(vinlyidene fluoride) based
ferroelectric polymers”, Chem. Mater. 20, 2078-2080 (2008).
[36] Li, J., S.I. Seok, B. Chu, F. Dogan, Q. Zhang abd Q. Wang, “Nanocomposites of ferroelectric
polymers with TiO2 nanoparticles exhibiting significantly enhanced electrical energy density”, Adv.
Mat. 21, 217-221 (2009).
[37] Wie, J., P. P. Mane, C. W. Green, K. M. Mossi and K. K. Leang, “Energy harvesting by pyroelectric
effect using PZT”, Proc. ASME Conf. Smart Materials, Adaptive Structures Intelligent Systems,
Ellicott City, MY, 1-5 (2008).
[38] Chang, H. H. S. and Z. Huang, “Substantial pyroelectric effect enhancement in laminated
composites”, Appl. Phys. Lett. 92, 152903 (2008).
[40] Hunter, S. R., N. V. Lavrik, T. Bannuru, S. Mostafa, S. Rajic and P. G. Daskos, “Pyroelectric thermal
energy harvesting using MEMS based resonant structures,” Energy Harvesting and Storage:
Materials, Devices, and Applications II, Proc. SPIE, Vol. 8035, 80350V-1 (2011).
[41] Lavrik, N.V., M.J. Sepaniak, P.G. Datskos, “Cantilever transducers as a platform for chemical and
biological sensors,” Rev. Scientif. Instru. 75, 2229-2253 (2004).

Proc. of SPIE Vol. 8377 83770D-14

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 05/21/2013 Terms of Use: http://spiedl.org/terms