INTEGRATION OF DISPENSER-PRINTED ULTRALOW VOLTAGE
THERMOELECTRIC AND ENERGY STORAGE DEVICES
Zuoqian Wang1, Deepa Madan1, Alic Chen1, Rei-cheng Juang2, Michael Nill1, 3, James W Evans3,
Paul K Wright1
1
Department of Mechanical Engineering, University of California, Berkeley
2
Green Energy & Environment Research Laboratories, Industrial Technology Research Institute
3
Department of Material Science and Engineering, University of California, Berkeley
*Presenting Author: zqwang@me.berkeley.edu
Abstract: This work reports on an integrated energy harvesting prototype consisting of dispenser-printed
thermoelectric energy harvesting and electrochemical energy storage devices. The use of a commercially available
DC-to-DC converter is explored to step-up the mV output voltage of the printed thermoelectric device to several
volts for charging printable zinc-based micro-batteries. Two separate parallel-connected thermoelectric devices
were designed and fabricated, one on flexible polyimide substrates with evaporated gold contacts and another on
commercially available flexible printed circuit boards. The prototype presented in this work demonstrated the
feasibility of deploying printed energy harvesting systems in practical applications.
Keywords: integration, printed energy harvesting, energy storage, ultralow voltage and power source
INTRODUCTION
Research related to self-powered wireless sensor
systems has been an emerging area in recent years for
its wide range of potential applications [1]. Perpetual
power solutions are typically composed of three
essential components: an energy harvester, power
regulation circuitry and energy storage devices [2, 3].
A custom dispenser printer has previously been
developed for fabricating both planar thermoelectric
generators (TEG) [4-5] for thermal energy harvesting
and electrochemical micro-batteries [6-8] for energy
storage. Dispenser printing is a cost effective, additiveprocess manufacturing method to precisely deposit and
pattern layered components onto various substrates.
The printing methods are also scalable towards mass
manufacturing, including traditional screen and
flexographic printing techniques.
In previous work, a 50-couple thermoelectric
device printed on a flexible polyimide substrate was
capable of producing 10.5µW at 171.6mV for a 20K
temperature difference at matched load resistance [5].
Printed zinc-manganese dixoide microbatteries with
gel polymer electrolytes achieved an average of 1
mAh/cm2 and 1.2 mWh/cm2 for discharge rates
between C/2 – C/7 [8]. Although we have individually
demonstrated the performance of printed thermal
energy harvesting and energy storage devices, practical
applications require integrated DC-to-DC conversion.
While low currents output (µA~mA) from TEGs are
sufficient for slowly charging microbatteries, a voltage
output higher than the battery open circuit voltage is
always necessary (1.5~5V depending on the battery
types). This presents difficulties for small-scale
thermoelectric generators at low temperature
differences since device output voltages typically fall
within the mV range. Thus, we explore the use of a
commercially available voltage step-up DC-to-DC
converter (Linear Tech LTC3108) to charge a printed
microbattery using ultralow power and energy
harvested from a printed thermoelectric device. Fig. 1
and 2 show a schematic and image of the components
in an integrated circuit.
Fig. 1: A schematic of printable low voltage
thermoelectric energy harvesting and energy storage
devices integration.
THERMOELECTRIC DEVICE DESIGN
Since the DC-to-DC converter is designed as a
12Ω load resistance, the printed thermoelectric
generator was redesigned to reduce device resistance
(typically 2-5kΩ) to match the load resistance. The
reduction in device resistance was achieved by
decreasing the number of thermoelectric couples in
series and instead, placing them in parallel. However,
this results in a trade-off between voltage output and
device resistance (Fig. 3). Fig. 4 demonstrates the
optimization between voltage output, number of
couples in series and temperature difference for a
printed TEG consisting of 50 couples. To achieve a
30mV output (the minimum input required by the
converter), a 50 couple device consisting of 10 parallel
sets of 5 couples in series requires a temperature
difference of 40K. The optimized device power output
(ΔT=20K) is 22.5mW (power requirement of data
transmitting mode is typically 5/10mW).
PCB Universe, Inc.
Evaporated Polyimide Substrate
The polyimide substrates were first prepared by
shadow-mask evaporation of gold metal contacts onto
a 63.5µm thick flexible polyimide well taped on a
glass substrate. Next, the p-type and n-type elements
were dispenser printed onto the substrate to form lines
spanning across the metal contacts on both sides. The
printed devices were then cured in a vacuum oven at
250°C for six hours. Finally, the devices were
connected using silver epoxy to two copper wires to
form electrical leads. Fig. 5 shows a picture of two
printed parallel thermoelectric devices on a flexible
polyimide substrate with evaporated gold contacts. On
each device, 7 parallel sets of 5 couples in series
consisting of elements that were 5mm long was
fabricated and characterized to test performance.
Fig. 2: Proof of concept integrated devices with a LED
output.
Fig. 5: Printed two sets of thermoelectric devices in
parallel on flexible polyimide substrate, with one set
connected to copper wire.
Fig. 3: Theoretical calculations show the tradeoff
between internal resistance and open circuit voltage.
Fig. 4: An optimization of internal connection of
thermoelectric devices for highest voltage and power.
EXPERIMENTAL
Thermoelectric Devices Fabrication
Details of the thermoelectric materials and
prototype fabrication can be found in [6]. Two
different substrates were used for the printed
thermoelectric device prototypes: (1) polyimide
substrates with evaporated gold contacts and (2)
custom flexible printed circuit boards (Flex PCB) by
Flexible Printed Circuit Board
Considering the thermal inefficiencies and high
costs related to the evaporated substrate, a flexible
double layer PCB substrate was alternatively designed,
The PCB consisted of metal traces connecting the
parallel elements to the back of the substrate. The
elements in front were connected to the back through
vias in the through-thickness direction. The illustration
and a photo with both of the front and back view of the
actual device are shown in Fig. 6. On each substrate,
there are 5 parallel sets of 10 couples in series
consisting of elements that were 3 mm long. The
design was chosen to increase voltage generation and
reduce device resistance. Sets of the devices were
conveniently connected in parallel using common
copper wires through the vias as shown on Fig. 2.
Integration and Testing
Both prototypes were connected in parallel and
stacked together to reduce the overall resistance to 10
Ohms. Devices were placed on a hot plate and
connected to the voltage step-up converter. Thermal
compound was used to improve interfacial thermal
conductivity between the devices and the heat source.
A small fan provided convective flow to increase the
Fig. 6: Illustration and actual printed devices on Flex
PCB substrate.
resistances as compared to theoretical calculations.
From Fig. 7, a resistance of 8.8Ω was achieved when
10 devices were connected in parallel. As expected, a
higher open circuit voltage than the first set of devices
was achieved from this set of devices for its higher
number of couples in-series (10 couples). Temperature
differences between 30-60K provided TEG open
circuit voltages of up to 200mV, allowing for all four
preset step-up voltage outputs to be obtained.
However, as the testing temperatures increased, the
internal resistance of the device also increased. This
limited the device power and current output. Table 2
summarizes the experimental results. Fig. 8 shows the
current and circuit efficiency as a function of input
voltage for a 1M Ohm external load (similar to that of
a printed zinc microbattery). The maximum efficiency
of the circuit was 12.4% when the voltage was 36 mV.
temperature gradients. Both the input and output
voltages and currents were recorded. A 1 cm2 Zinc
based micro-battery was printed on the free space of
the conversion circuit and connected to the output
terminal [8]. When no thermal energy is available, the
microbattery discharges and provides energy for the
sensing and transmitting applications. The prototype
presented in this work uses a LED light for
demonstration purposes.
2%"
!"#$%&'()*)#$+,(&-./&
,%"
34567589:;"<=:;!>?>"
+%"
Fig. 8: Output current and efficiency vs input voltage.
@&A57?B5=C:;")5>D;C>"
$%"
/%"
1%"
!"#"+,-+./&'%-.%,"
)*"#"%-0..,"
!"#"$%&'("
)*"#"("
(%"
%"
%"
("
1"
/"
$"
+"
,"
2"
0123(4&"5&6(#&"5&7(8*,()&
."
0"
(%"
Fig. 7: Comparison of devices resistance from
theoretical calculations and experimental results.
RESULTS & DISCUSSION
Devices On Evaporated Polyimide Substrates
The first thermoelectric prototype fabricated
resulted in a higher-than-expected internal device
resistance. Each set of 5 couple in series had a device
resistance of approximately 630Ω. Thus, 56 sets of 5
couples in series were placed in parallel to reduce the
device resistance to 12Ω. Next, the device was tested
on a hot plate to achieve a 79mV open circuit voltage.
The closed circuit input voltage, however, only
reached a maximum value of 22.5mV. This limits the
maximum measurable circuit output voltage to 3.27V.
Table 1 summarizes the experimental results.
Devices On Flex PCB’s
The Flex PCB devices showed reasonable internal
Fig. 9: A printed multilayer microbatteries charging
time analysis based on an capacity of 1 mAh/cm2 for
each layer and a circuit efficiency of 12.4%.
Micro-batteries Charging Time Analysis
The 2-5V stepped-up output voltage from the
printed TEG’s is sufficient for charging the printed
electrochemical energy storage devices. Battery
charging time is a function of the current output from
the converter and the efficiency of the conversion. Fig.
9 presents a multilayer microbattery charging time
analysis based on the obtained voltage step-up circuit
efficiency and an optimized input power of 22.5mW.
As the capacity of the battery increases, the charging
time increases proportionally. For a small 1cm2
battery, the charging time will only range between a
Table 1: Measured circuit performance for thermoelectric devices on evaporated polyimide substrate, consisting of 56
parallel sets of 5 couple in series.
Open Circuit
Voltage
VOC (mV)
79
Input
Voltage
Vin (mV)
22.5
Input
Current
Iin (mA)
Circuit Output
Set Voltage
Vset (V)
Output
Voltage2
Vout (V)
Output
Current2
Iout (µA)
2.9
2.35
3.3
4.1
5
2.34
3.27
–1
–1
12.4
12.3
–1
–1
1. Input voltage was insufficient for measurable circuit output; 2. No external load applied.
Table 2: Measured circuit performance for printed thermoelectric devices on Flex PCB, consisting of 50 parallel sets of
10 couple in series.
Open Circuit
Voltage
VOC (mV)
Input
Voltage
Vin (mV)
Input
Current
Iin (mA)
Circuit Output
Set Voltage
Vset (V)
Output
Voltage1
Vout (V)
Output
Current1
Iout (µA)
110
115
115
118
120
130
150
180
200
21.5
22.2
23
23.9
25
26
27.8
31
36
2.9
3.17
3.3
3.4
3.5
4.5
4.7
5.2
5.6
2.35
2.35
2.35
2.35
2.35
3.3
3.3
4.1
5
1.175
1.224
2.05
1.43
2.35
2.78
3.3
4.1
5
1.28
1.38
2.2
1.6
2.5
3.13
3.6
4.5
5
1. Applied external load: 1 M Ohm.
few minutes. However, for large surface area batteries
and multi-layered systems, the charging time can
exceed 10 hours. Future work will be focused on
characterizing the performance of the printed battery
charged from the printed TEG described in this work.
CONCLUSIONS
An integrated dispenser-printed energy harvesting
system was investigated in this work. The system
consisted of printed thermoelectric energy generators,
a commercial voltage step-up converter and printed
electrochemical energy storage devices. The TEG
input voltages between 20-36mV were successfully
stepped up to 2.35-5V at currents between 1.28-5µA.
These values are sufficient for slow charging of
printed electrochemical batteries. Future work will
focus on improving the current output and
characterization of the charge-discharge behavior of
printed zinc micro-batteries in the integrated system.
ACKNOWLEDGEMENTS
The authors thank the California Energy
Commission for supporting this research under
contract 500-01-43. We would also like to thank
Christine Ho, Brian Mahlstedt, Michael Seidal,
Christopher Sherman, Jay Keist and Peter Minor for
their contributions.
REFERENCES
[1] James E P et al. 2004 An investigation of selfpowered systems for condition monitoring
applications Sens. and Actuators A 110 171-176
[2] Koeneman P B, Busch-Vishniac I J and Wood K
L 1997 Feasibility of micro power supplies for
MEMS J. Microelectromech. Syst. 6 355–62
[3] Harb JN, LaFollette RM, and Howell LL 2002
Microbatteries
for
self-sustained
hybrid
micropower supplies J. Power Sources 104 46–51
[4] Madan D, Chen A, Wright P K, Evans J W 2010
Dispenser printed composite thermoelectric thick
films for thermoelectric generator applications J.
Applied Physics 109 034904 (6pp).
[5] Chen A, Madan D, Wright P K, Evans J W 2011
Dispenser-printed planar thick film thermoelectric
generators J. Micromech. Microeng. 21, 389698
(8pp). In press.
[6] Ho, C.C., D.A. Steingart, J.W. Evans, and P.K.
Wright 2008 Tailoring Electrochemical Capacitor
Energy Storage Using Direct Write Dispenser
Printing ECS Transactions, Vol. 16 (1), 35-47.
[7] Ho, C.C., Murata, K., Steingart, D.A., Evans,
J.W., and Wright, P.K 2009 A super ink jet
printed zinc-silver 3D microbattery J. Micromech.
Microeng. 19, 094013 5.
[8] Ho, C.C., J.W. Evans, and P.K. Wright 2010
Direct write dispenser printing of a zinc
microbattery with an ionic liquid gel
electrolyte J. Micromech. Microeng. 20, 104009
(9pp).