ABSTRACT
SARGOLZAEIAVAL, YASAMAN. Flexible Thermoelectric Energy Harvesters Using Bulk Thermoelectric Legs and Low-Resistivity Liquid Metal Interconnects with High Thermal Conductivity Encapsulation. (Under the direction of Dr. Mehmet C. Öztürk.)
The human body continuously generates bio-heat through metabolic reactions. The
metabolic heat which is dissipated from the human skin surface, is a renewable source of
en-ergy, which can be harvested to provide sufficient power for miniaturized electronic devices.
Thermoelectric generators (TEGs) are excellent candidates for conversion of metabolic heat
to electrical power. When a temperature gradient is established across a TEG, an electrical
voltage is generated due to the Seebeck effect. Since there exists a natural temperature
difference between the human body and the ambient, TEGs can be integrated into wearable
electronics, such as fitness gadgets, as the energy source to generate the required energy. In
general, wearable TEGs can generate micro-watts to milli-watts of electrical power, which
is sufficient to continuously run wearables with low power electronics. There is a growing
interest in long-term vigilant monitoring of human health parameters as well as
environ-mental factors using wearable electronics. However, the need for frequent replacement or
recharging of the batteries is a huge challenge which makes wearable monitoring systems
less appealing to a large population of users. Recent advances in low power wearable
elec-tronics makes it feasible to realize fully self-powered systems that solely rely on harvested
energy from the human body. Therefore, incorporating flexible TEGs as the power source
into wearable electronics is a promising way to eliminate the need for batteries. With low
power sensors, electronics, and flexible TEGs, fully self-powered wearable systems can be
achievable. The focus of this thesis was to minimize the parasitic losses that occurred due
to the presence of an encapsulation layer above the EGaIn interconnects..
The main goal of this work is to improve the performance of flexible TEGs using eutectic
gallium indium (EGaIn) interconnects. Initially, the performance of flexible TEGs was
encasing layer of the EGaIn interconnects should have a thermal conductivity of about
1 W/mK (or higher) to avoid the impact of the parasitic thermal series resistance of the
encapsulation layer. Additionally, our modeling showed that incorporation of a thin metal
spreader with a thickness of less than 10µm on the cold side of the TEGs improves the
output power by enhancing heat rejection.
In order to produce flexible TEGs that can compete with commercial-off-the-shelf
(COTS) rigid TEGs, it is essential to to minimize the parasitic thermal resistances via new
materials and device design. We were able to successfully develop a high thermal
conductiv-ity (HTC) elastomer using Polydimethylsiloxane (PDMS) doped with EGaIn and graphene
nanoplatelets (GnPs). The thermal conductivity of this new elastomer was about 6X higher
than that of pure PDMS. We showed that the incorporation of the HTC elastomer as the
encasing layer of the EGaIn interconnects increased the output power of the flexible TEGs
by 1.7X compared to devices created with pure PDMS as the encapsulation layer. Finally,
we developed a method to electroplate copper on flexible TEGs that led to a further
en-hancement of 1.3X in the output power density with the presence of air flow at normal
walking speeds.
Finally, the performance of different flexible TEGs were evaluated and reported at
different ambient conditions showing best-in-class performance compared to other flexible
TEGs developed for body heat harvesting. Lastly, we identify and discuss future design
improvements and modeling pathways that can potentially impact the design, fabrication,
© Copyright 2019 by Yasaman Sargolzaeiaval
Flexible Thermoelectric Energy Harvesters Using Bulk Thermoelectric Legs and Low-Resistivity Liquid Metal Interconnects with High
Thermal Conductivity Encapsulation
by
Yasaman Sargolzaeiaval
A dissertation submitted to the Graduate Faculty of North Carolina State University
in partial fulfillment of the requirements for the Degree of
Doctor of Philosophy
Electrical Engineering
Raleigh, North Carolina
2019
APPROVED BY:
Dr. Michael Dickey Dr. Veena Misra
Dr. Alper Bozkurt Dr. Mehmet C. Öztürk
DEDICATION
I dedicate this thesis to my parents for all the support and unconditional love they
BIOGRAPHY
Yasaman Sargolzaeiaval was born in Zahedan located in southwest Iran. She completed
her Bachelors of Science in Electrical Engineering, with a focus in Electronics, at Sharif
University of Technology, Tehran, Iran. She then received her Masters of Science degree in
Electrical Engineering, majoring in nanoelectronic devices, also at Sharif University. She
started her PhD studies at NC State University in 2015 under the direction of Prof. Mehmet
C. Öztürk. In 2016, she joined the NSF funded Advanced Self-Powered Systems of Integrated
Sensors and Technologies (ASSIST) engineering research center and started working on her
PhD thesis. Her research interests include renewable energy, energy harvesting, low/
ACKNOWLEDGEMENTS
I would first like to thank my parents for the unconditional love they gave me throughout
my entire life. Without their support and guidance, I would not have been able to be the
person I am today.
I want to also express my gratitude to the students and postdocs, who have helped and
supported me through out this work including Taylor V. Neumann, Abhishek Malhotra,
Ozkan Yildiz, Farzad Mohaddes, Amin Nozariasbmarz, and Michael Lim. I would also like
to thank Francisco Suarez for initiating this work and helping me with my work during
the early days. I am forever thankful to my friend and colleague, Viswanath Padmanabhan
Ramesh, for his friendship, help, and support during these four years.
I would like to thank the ASSIST Engineering Research Center for supporting this project
and giving me the resources I needed to complete my work. I would like to thank the director
and the staff of the NCSU Nanofabrication Facility (NNF); Dr. Philip Barletta, Nicole Hedges
and Marcio Cerullo for all of their help and guidance in the cleanroom. I want to also thank
Dr. Daryoosh Vashaee for his support and help during the course of my PhD.
I would like to thank Dr. Michael Dickey for his expertise and support of my work. He
not only provided much useful feedback, but he also opened his lab to share the equipment.
I would also like to thank Dr. Veena Misra for her grand vision for the ASSIST center; and
Dr. Alper Bozkurt for his insightful comments on my work.
I would like to express my deepest thanks to my friends, Amirhassan Fallah, Elnaz
Shabani, Ghazal Kamyabjoo, Alireza Garmabi, and Mohammad Salamati who have always
been a major source of support when I needed them. My friends, who are like family to me
now, helped me get through my hardest times during these four years.
Finally, I would like to express my sincere gratitude to my advisor Prof. Mehmet C.
Öztürk for his continuous support of my work, for his patience, motivation, enthusiasm,
His technical and editorial advice was essential to the completion of this dissertation and
has taught me innumerable lessons and insights on the workings of academic research in
TABLE OF CONTENTS
LIST OF TABLES . . . viii
LIST OF FIGURES. . . ix
Chapter 1 INTRODUCTION. . . 1
1.1 History of Thermoelectric Effect and Thermoelectric Devices . . . 1
1.2 Thermoelectric Modules . . . 8
1.3 TEGs for Body Heat Harvesting . . . 11
1.4 Summary of the Thesis . . . 14
Chapter 2 Review of the Existing Methods for Fabrication of Flexible TEGs . . . . 17
2.1 Introduction . . . 17
2.2 Synthesis of Thermoelectric Materials for Flexible TEGs . . . 18
2.2.1 Physical Vapor Deposition (PVD) . . . 18
2.2.2 Electrochemical deposition . . . 20
2.2.3 Screen Printing . . . 23
2.2.4 Dispenser Printing . . . 27
2.2.5 Organic and Solution-processable Materials for Fabrication of TEGs 29 2.2.6 Fabrics . . . 32
2.3 Flexible TEGs Using Bulk Materials . . . 34
2.3.1 Flexible TEGs Using Bulk Materials and Liquid Metal Interconnects 40 2.4 Summary . . . 43
Chapter 3 High Thermal Conductivity Silicone Elastomer Doped with Graphene Nanoplatelets and Eutectic GaIn Liquid Metal Alloy (Published) . . . . 44
3.1 Abstract . . . 44
3.2 Introduction . . . 45
3.3 Experimental . . . 47
3.4 Results and Discussion . . . 49
3.5 Conclusion . . . 60
Chapter 4 Flexible Thermoelectric Generators for Heat Harvesting From the Hu-man Body - Enhanced Device PerforHu-mance Using High Thermal Con-ductivity Elastomer Encapsulation on Stretchable Liquid Metal Inter-connects . . . 62
4.1 Abstract . . . 63
4.2 Introduction . . . 63
4.3 Experimental . . . 67
4.4 Results . . . 72
4.5 Conclusion . . . 82
Chapter 5 Incorporation of Flexible Heatsinks onto TEGs . . . 86
5.2 Fabrication Process of Flexible Heatsinks . . . 86
5.3 Characterization of Flexible Heatsinks . . . 87
5.4 Incorporation of Flexible Heatsinks onto Flexible TEGs . . . 91
5.5 3D COMSOL Simulation of Flexible Heatsinks . . . 92
5.6 Conclusion . . . 94
Chapter 6 Modeling of the Human Body as the Heat source for TEGs . . . 95
6.1 introduction . . . 95
6.2 Modeling Results . . . 96
6.3 Conclusions . . . 102
Chapter 7 Conclusions and Future Work . . . 103
7.1 Conclusions . . . 103
7.2 Future Work . . . 104
BIBLIOGRAPHY . . . 106
APPENDICES . . . 123
Appendix A Steam Treatment of the Elastomer Surface for Durable Electro-plating . . . 124
LIST OF TABLES
Table 1.1 Performance of some of the best rigid TEGs for body heat harvesting. 14
Table 4.1 Properties of thermoelectric legs. . . 67 Table 4.2 Parameters used in COMSOL simulation. . . 75
Table 5.1 Total thermal resistance of the hot side of a rigid TEG with the in-corporation of different flexible heatsinks measured at different air velocities. . . 91
Table 6.1 Metabolism rate of a average-sized person at different activity levels [218]. . . 98 Table 6.2 Equations for calculating RMR from sex, age, and body mass[219].
(RMR is inO2uptake units (ml.kg´1.min´1 . . . 100 Table B.1 Properties of the conductive powders mixed with PDMS to create
LIST OF FIGURES
Figure 1.1 a) Thomas J. Seebeck. b) An illustration of the original apparatus used by Thomas Seebeck which led to the discovery of the thermo-electric effect. A and B are the junctions formed by two dissimilar conductors. The compass needle moves in response to heating one of these junctions. . . 2 Figure 1.2 a) Jean Charles Peltier. b) William Thompson (Lord Kelvin). . . 3 Figure 1.3 Dependence of the thermoelectric figure of merit (z T) of some
thermoelectric materials on the temperature[9]. a) N-type materials. b) P-type materials. . . 4 Figure 1.4 Dependence of electrical conductivity,thermal conductivity,
See-beck coefficient, and thermoelectric figure of merit, z T, on the carrier concentration for a typical thermoelectric material[9]. . . . 5 Figure 1.5 a) Maria Telkes, inventor of first TEG in 1947, and b) Abraham Ioffe
who developed the theory of thermoelectricity for semiconductors. 6 Figure 1.6 Curiosity Rover exploring Mars and relies solely on RTEGs for power. 7 Figure 1.7 Key developments in history of thermoelectrics. . . 7 Figure 1.8 Schematic of a simple thermoelectric module in a) cooling mode,
and b) power generation. . . 9 Figure 1.9 A commercial TEG made of many N-type and P-type thermoelectric
legs. a) Schematic[31]b) Real image protect[32]. . . 10 Figure 1.10 Human skin temperatures at different body locations for varying
ambient temperatures[36]. . . 12 Figure 1.11 An electrical circuit representative of a TEG on the human body . . 13
Figure 2.1 Flexible thin film TEG made with DC-sputtered thermoelectric ma-terial[95]. . . 20 Figure 2.2 Vertical flexible TEG made with planar ECD deposited
thermoelec-tric materials and vertical (through-plane) heat guides.[100]. . . 23 Figure 2.3 Vertical uni-polar flexible TEG made with planar screen-printed
thermoelectric materials. a) Schematic of the TEG. b) Fabricated TEG.[101] . . . 25 Figure 2.4 Screen-printed TEGs placed vertically on a steam pipe for waste
heat recovery.[105] . . . 25 Figure 2.5 Flexible TEG made with screen-printed thermoelectric material[106]. 27 Figure 2.6 Flexible TEG made with ink-jet printed thermoelectric material[75]. 29 Figure 2.7 A flexible and transparent TEG made with PEDOT:PSS and
ITO-PEDOT:PSS[118]. . . 31 Figure 2.8 A flexible TEG made withC60{T i S2nanosheets[120]. . . 32 Figure 2.9 Thermoelectric fiber made with thermal evaporation of nickel and
silver[94]. . . 34 Figure 2.10 a) Schematic, and b) Photograph of a woven TEG made with fibers
Figure 2.11 flexible TEG made with planar bulk thermoelectric legs and heat pipes for maintaining the temperature differential[124]. . . 36 Figure 2.12 Schematic diagram of the designed wearable TEG using bulk legs
and flexible PCB: a) TEG wears on the wrist, b) schematic view of the wearable TEG on the skin.[91]. . . 37 Figure 2.13 A flexible TEG made with bulk legs and flexible PCB[90]. . . 37 Figure 2.14 A flexible TEG made with bulk thermoelectric legs, soldered copper
strip interconnections, and PDMS encapsulation[125]. . . 38 Figure 2.15 A flexible TEG made with bulk thermoelectric legs and FPCB
elec-trodes interconnections. Bakelite holders and flexible wires were used to achieve flexibility[126]. . . 39 Figure 2.16 A bulk flexible TEG made with copper interconnects in a flexible
PDMS substrate[128]. . . 39 Figure 2.17 A bulk flexible TEG made with liquid metal (EGaIn) interconnects
[53]. . . 41 Figure 2.18 a) and b) Experimental setup for bending test. c) AC resistance of
the flexible TEG made with liquid metal interconnects as a function of number of bending cycles. “Self-healing” properties of EGaIn is highlighted in this figure.[53]. . . 42 Figure 2.19 Predicted performance of a flexible TEG made with 64
thermoelec-tric legs with an area of 4 c m2 at room temperature of 24˝C for
several potential modifications[53]. . . 43
Figure 3.1 Measured total resistance of different EGaIn/PDMS plotted against sample thicknesses for five different weight percentages of GnP (Inset:Cured GnP/EGaIn/PDMS sample.). . . 50 Figure 3.2 Thermal conductivity of samples with different GnP weight
concen-trations with and without EGaIn (50 weight%). . . 52 Figure 3.3 Thermal conductivity enhancement relative to PDMS of samples
with different GnP weight concentrations with and without EGaIn (50 weight%). . . 53 Figure 3.4 EGaIn dispersion in PDMS. . . 53 Figure 3.5 Cross-sectional SEM images of elastomers with a) GnP and b) GnP
and EGaIn inclusions. . . 55 Figure 3.6 High magnification image of an EGaIn region in a sample doped
with GnPs and EGaIn. . . 56 Figure 3.7 Young’s modulus of PDMS with EGaIn (50 %wt) and GnP inclusion. 58 Figure 3.8 Maximum strain at break of PDMS with EGaIn and GnP inclusion. . 58 Figure 3.9 Tensile strength of PDMS with EGaIn and GnP inclusion. . . 59 Figure 3.10 Viscosity measurements of PDMS with EGaIn and GnP inclusion. . 60
Figure 4.2 The process flow used in fabrication of flexible thermoelectric gener-ators. (a)The legs are placed through a template; (b) PDMS is poured between the legs, planarized and cured; (c)EGaIn interconnects are sprayed through a hard mask; (d) A thin („50µm) layer of PDMS is sprayed as temporary encapsulation; (e) The device is flipped and liquid metal interconnects are sprayed and encapsulated on the backside; (f ) Final high thermal conductivity encapsulation elas-tomer is drop-casted on both sides of the device and cured. . . 70 Figure 4.3 Thermal conductivity of the encapsulation elastomer with and
with-out EGaIn inclusion as a function of the GnP weight percentage. (Inset: A cured sample doped with 50 wt% liquid metal (EGaIn) and 2.2 wt% GnP.) . . . 71 Figure 4.4 a) A thermoelectric generator made with high thermal conductivity
encapsulation. b) cross-sectional image of the device. . . 71 Figure 4.5 a) Setup used to characterize the thermoelectric generators On the
hot plate. b) TEG worn on the wrist inside and outside of the wind tunnel. . . 73 Figure 4.6 Device performance measured for different encapsulation layers.
a) Measured open-circuit voltage versus air velocity. b) Calculated power density of TEGs with 20% fill factor. The numbers in the labels refer to the thickness of the encasing layer. The lines are just used to guide the eye. Decreasing the thickness of PDMS or increasing the thermal conductivity helps improve the power density. . . 74 Figure 4.7 a) The 3-D structure used in COMSOL simulations; b) Simulated
temperature drop across a single TE leg as a function encapsulating elastomer thickness for different elastomer thermal conductivities. 76 Figure 4.8 a) Open circuit voltage measured for thermoelectric generators
made high thermal conductivity and PDMS encapsulation with dif-ferent fill factors. b) Corresponding power density for those devices. The lines are used just to guide the eye. . . 77 Figure 4.9 Modeled∆T across a single leg in a 64 leg TEG plotted as a function
of the copper spreader thickness. . . 78 Figure 4.10 TEG with an electroplated copper spreader layer on top of HTC
elastomer. . . 79 Figure 4.11 a) Open circuit voltage measured for thermoelectric generators
made with high thermal conductivity encapsulation with and with-out a copper spreader layer. b) Open circuit voltage measured for TEGs with PDMS encapsulation with and without copper spreader. The lines are used just to guide the eye. . . 80 Figure 4.12 a) Power density for devices fabricated using high thermal
Figure 4.13 a) Image of the TEG worn on the wrist b) Open circuit voltage mea-sured on the wrist for the thermoelectric generators made with high thermal conductivity encapsulation and copper spreader with a fill factor of 20%. c) Corresponding power density for those devices. The lines are used just to guide the eye. . . 84 Figure 4.14 Setup used to measure force required to bend the devices around a
cylinder. . . 85 Figure 4.15 Force required to keep the TEGs and control samples bent around a
cylinder. . . 85
Figure 5.1 Fabrication process of flexible heatsinks. . . 87 Figure 5.2 a) Flexible heatsink and the 3D printed mold used to create it; b) A
flexible TEG with the incorporated flexible heatsink. . . 88 Figure 5.3 The rigid TEG with incorporated flexible heatsink . . . 90 Figure 5.4 The wind tunnel designed to measure interface thermal resistance
of the rigid TEG with incorporated flexible heatsink . . . 90 Figure 5.5 a) Open circuit voltage and b) power density of a TEG with 20% fill
factor with and without the flexible heatsink. . . 92 Figure 5.6 The 3D structure used to model heatsinks in COMSOL. . . 93 Figure 5.7 Dependence of the∆T across a single thermoelectric leg as a
func-tion of a) radius of the heatsink cones; b) height of the heatsink cones. . . 93
Figure 6.1 Variation of skin temperature at the wrist, forehead and fingertips with respect to environmental temperature 20˝C[44]. . . . 97
Figure 6.2 Illustration of different aspects of modeling of human body[215]. . 97 Figure 6.3 Skin thermal conductivity at an ambient temperature of 20˝C
plot-ted against different metabolic rates. . . 99 Figure 6.4 Mean skin temperature at different ambient temperatures[216]. . . 99 Figure 6.5 Skin thermal conductivity at different metabolic rates plotted against
different ambient temperatures. . . 101 Figure 6.6 Metabolic heat generation of man at the age of 25 at different walking
speeds. . . 101
Figure 7.1 Comparison between the state-of-the-art flexible TEGs and the TEGs presented in this thesis. . . 105
Figure A.1 A porous PDMS surface created by water vapor[220]. . . 125 Figure A.2 (a)The schematic and (b) the actual setup of the steaming process.
(c) An steamed sample after curing. (d) A microscope image of a steamed PDMS sample (20X magnification). . . 126
CHAPTER 1
INTRODUCTION
1.1
History of Thermoelectric Effect and Thermoelectric
De-vices
The origins of the thermoelectric effect dates back to 1821 when Thomas Johann Seebeck
(Fig. 1.1.a), a German Scientist, observed the deflection of a compass needle placed between
two dissimilar conductors. The conductors were connected together as shown in Fig. 1.1.b,
and heated by a candle at one end[1]. The deflection was explained to be due to the change
in the magnetic field produced by the current flowing through the closed electrical circuit
formed by the two electrodes[2]. Generation of an electrical voltage at the junction of two
dissimilar conductors in the presence of a temperature gradient is due to the displacement
of electrical charge carriers, which gain energy by absorbing the heat and travel to the
opposite side. The voltage generated in materials experiencing a temperature gradient is
called the “Seebeck voltage”.
(a) (b)
Figure 1.1a) Thomas J. Seebeck. b) An illustration of the original apparatus used by Thomas Seebeck which led to the discovery of the thermoelectric effect. A and B are the junctions formed by two dissimilar conductors. The compass needle moves in response to heating one of these junctions.
discovered that whenever two dissimilar conductors came into contact to form a junction
with a current passing through, heat is rejected or absorbed at the junction resulting in
local cooling or heating[3]. This mechanism is the reverse of the Seebeck effect. It was
discovered that when carriers travel across a junction of two conductors, they either need
to overcome a potential barrier or relax to a lower energy state by absorbing or releasing
heat as phonons. This mechanism is called the “Peltier effect”.
21 years later, in 1855, the connection between Peltier and Seebeck effects was
dis-covered by a British scientist, William Thompson (Fig. 1.2.b)[3]. His discovery led to the
third groundwork of the thermoelectric effect which is known as the “Thompson effect”.
Thompson discovered that when a temperature difference existed across a block of material,
the electric current absorbed the heat on the hot side and carried it to the cold side. This
The combination of Seebeck, Peltier, and Thompson effects is called the “Thermoelectric
effect" and defines the relation between the heat and the electrical energy transformation
in materials.
(a) (b)
Figure 1.2a) Jean Charles Peltier. b) William Thompson (Lord Kelvin).
The efficiency of a thermoelectric material is summarized in a parameter called
“ther-moelectric figure-of-merit: (zT)”, which was defined in 1909 by a German physicist, named
Edmund Altenkirch[2]. The figure-of-merit is expressed as, Eq. 1.1:
z T “S
2σ
κ T (1.1)
each thermoelectric material had an optimum operating temperature wherez T reached a maximum. This discovery led to extensive research to develop materials for applications
requiring different operation temperatures[4–8]. Fig. 1.3 shows the dependence ofz T on temperature for some of the popular N-type and P-type thermoelectric materials. A good
thermoelectric material must have a high Seebeck coefficient, a low thermal conductivity,
and a low electrical resistivity. Each of these three parameters are functions of carrier
con-centration and cannot be optimized independently. Fig. 1.4 shows the dependence ofσ,κ,
S, andz T on the carrier concentration for a typical thermoelectric material[9]. Shown in Fig. 1.4, as the carrier concentration increases, the Seebeck coefficient decreases while the
electrical and thermal conductivity increase. Therefore, for a maximumz T, an optimum value exists for a given carrier concentration. Early research on thermoelectric materials
was carried out by characterizing thermoelectric properties of metals and alloys which
were not suitable for the majority of applications of thermoelectric devices we have today.
(a) (b)
Figure 1.3Dependence of the thermoelectric figure of merit (z T) of some thermoelectric materi-als on the temperature[9]. a) N-type materials. b) P-type materials.
The first thermoelectric generator (TEG) was realized by a Hungarian Scientist, Maria
thermoelec-Figure 1.4Dependence of electrical conductivity,thermal conductivity, Seebeck coefficient, and thermoelectric figure of merit,z T, on the carrier concentration for a typical thermoelectric material[9].
tric effect was discovered in the early 18t hcentury, it did not have any feasible applications
until after realization of doped small band-gap semiconductors with superior
thermo-electric performances in 1950. During this period, Abraham Ioffe (Fig. 1.5.b) developed
the fundamental theory for the thermoelectric effect in semiconductors that led to
break-throughs in the development of high efficiency thermoelectric materials and modules for a
variety of applications[3].
The first application of TEGs for heat harvesting was realized in 1957 by Ken Jordan and
John Birden when they used radioisotope thermoelectric generators (RTEGs) to generate
electric power from the heat produced by the decay of radioactive materials[11]. Radioactive
materials have the capability to generate a temperature gradient as high as 600K across
a thermoelectric module[12–14]. Because of the long lifetime of such materials, they are
able to provide the required heat for many years. NASA developed several satellites and
spacecrafts that entirely relied on RTEGs as the power source. One famous example is
the “Curiosity Rover” (Fig. 1.6) that landed successfully on Mars and is currently exploring
the planet. Generally, for space exploration missions, especially at distant locations far
(a) (b)
Figure 1.5a) Maria Telkes, inventor of first TEG in 1947, and b) Abraham Ioffe who developed the theory of thermoelectricity for semiconductors.
weak sunlight rays reaching the panels; hence, RTEGs are better suited for this particular
application. Soon after realization of RTEGs, low temperature applications such as recovery
of waste heat drew attention and led to the development of new materials for efficient
performance at room temperature[15].
One of the main limitations of thermoelectric devices is their low efficiency which is
often as low as 10% of the Carnot efficiency defined as the theoretical maximum efficiency
of a heat engine working between two temperatures, which can be expresses as Eq. 1.2:
η“TH o t ´TC o l d
TH o t
(1.2)
whereTH o t andTC o l d are the temperatures on the hot and cold sides of the module
respec-tively.
This efficiency can be improved by materials optimization, which has been the main
focus of numerous studies[9, 16–20]. In 1960’s, several studies predicted the upper limit for
Figure 1.6Curiosity Rover exploring Mars and relies solely on RTEGs for power.
at the time[22]. In 1993, it was predicted that an order of magnitude improvement inz T could be possible by nanostructuring of the thermoelectric materials[23]. This work led to
an exponential increase in the amount research on the subject[24–30].
Fig. 1.7 illustrates the history of Thermoelectric effect and key devices relying on the
thermoelectric effect.
1.2
Thermoelectric Modules
Thermoelectric effect enables cooling or heating without incorporation of any moving
parts, which makes thermoelectric devices less susceptible to failure over a long period of
time. Thermoelectric devices can be used in compressorless refrigerators, which can be
very small and light weight, or as the generators of electrical energy by recovery of waste
heat.
Thermoelectric devices are typically made of N-type (where electrons are the majority
carriers) and P-type (where holes, deficiency of electrons, are the majority carriers)
semi-conductor blocks (which are referred to as thermoelectric “legs”) connected electrically
in series and thermally in parallel. When a doped semiconductor is heated on one side,
the majority carriers (electrons in N-type semiconductors and holes in P-type
semicon-ductors) absorb the thermal energy and leave the hot side resulting in accumulation of
carriers on the other side. This charge build-up generates an electric potential difference
across the semiconductor which is the basis of power generation in TEGs as described
before. The generated voltage can be increased by connecting multiple semiconductor
blocks (thermoelectric legs) in series. In the presence of the same temperature gradient,
N-type and P-type semiconductors show opposite polarities of generated Seebeck voltages.
Therefore, N-type and P-type semiconductors can be connected in series in alternating
order to achieve a net generated voltage, which is equal to sum of the voltages generated
by each individual thermoelectric leg. A simple thermoelectric couple made of a single
N-type and a single P-type semiconductor block is shown in Fig. 1.8. If the thermoelectric
couple is used as a cooler, a current (or a voltage) is applied to the module resulting in the
flow of majority careers toward one side of the semiconductor. When carriers move, they
transfer energy resulting in heat absorption at one side (cold side) and heat generation at
the other side (hot side) as shown in Fig. 1.8.a. In the power generation mode, a temperature
load as shown in Fig. 1.8.b. More N-type and P-type thermoelectric legs can be added to a
thermoelectric module to increase the output voltage.
(a) (b)
Figure 1.8Schematic of a simple thermoelectric module in a) cooling mode, and b) power gener-ation.
As shown in Fig. 1.9, rigid TEGs are made of many thermoelectric legs connected in
series and sandwiched between two thermally conductive plates (commonly referred to
as headers) made of thermally conductive ceramics such as aluminum oxide (Al2O3) or aluminum nitride (Al N) to minimize the temperature loss due to packaging. Since effective heat dissipation from the cold side is necessary to maintain a temperature differential, a
heatsink attached to the TEG is typically used, to improve heat rejection via convection.
The maximum power generated by a TEG is obtained when it is connected to a matched
load and can be calculated using Eq. 1.3:
Pm a x “
V2
O C
4RT E Ge f f
(1.3)
(a) (b)
Figure 1.9A commercial TEG made of many N-type and P-type thermoelectric legs. a) Schematic [31]b) Real image protect[32].
resistance of the TEG. Eq. 1.3 can be re-written as Eq. 1.4:
Pm a x “
z∆T2 4RT h e r m a l
(1.4)
wherez can be calculated using Eq. 1.1 and 4RT h e r m a l is the effective thermal resistance of
the TEG. Finally, the maximum conversion efficiency (ηM a x) of a TEG can be calculated
using Eq. 1.5:
ηM a x “ηC a r n o t
a
1`z Ta v g´1
a
1`z Ta v g` TC o l d
TH o t
(1.5)
wherez Ta v g is the average thermoelectric figure of merit of N-type and P-type
thermoelec-tric legs.
Even though there are many studies focused on improving the properties of
thermo-electric materials, the device design is just as important to realize a high performance TEG.
To obtain an enhanced performance, there are many challenges associated with the design
and engineering of TEGs. These challenges include electrical contact and interconnect
resistances, optimal spacing between the thermoelectric legs for a specific application (or
ther-mal parasitic resistances introduced by the package, and the therther-mal contact resistances
between the device and the heat source or the cold side and the ambient[33]. Therefore, in
order to fully benefit from high efficiency thermoelectric materials, it is essential to optimize
the design of thermoelectric devices for an application. Ideally, to maximize the output
power of a TEG, the interconnects between the thermoelectric legs should have a negligible
electrical resistance compared to the electrical resistance of the thermoelectric legs (which
minimizes the total resistance of the TEG), the top and bottom substrates should be as
thermally conductive as possible (which minimizes the parasitic thermal resistances), and
the filler material between the legs should be as thermally insulating as possible (to guide
the heat flow through the thermoelectric legs). The density of the legs is determined by the
specific application, which requires an optimization between the electrical and thermal
resistances of the module.
In wearable electronics, the main advantage of TEGs over other mature energy
har-vesting technologies such as photovoltaics is the ability of the TEGs to work continuously
during a day which would be beneficial for self-powered systems that require continuous
operation. As the technologies for internet of things (IoT) continue to flourish, there is an
inevitable need for uninterrupted communication among many electronic devices,
includ-ing sensors, storage devices, and circuits. A reliable operation of connected networks and
systems depends upon power sources, which can continuously provide the required energy
levels. Given the recent developments in low-power electronic circuits and systems, using
TEGs to harvest heat as the power source is a very promising way to realize self-powered
continuous monitoring systems.
1.3
TEGs for Body Heat Harvesting
As discussed above, successful operation of a TEG relies on establishing a temperature
much smaller temperature differentials. Given the fact that there is generally a temperature
gradient between the human body and the surrounding environment, TEGs can be utilized
to harvest the body heat and generate electric power. The first commercial device powered
by a body heat harvesting TEG was realized in 1998 by Seiko. The “Seiko Thermic” watch
solely relied on TEGs embedded in the watch as the power source[34].
Development of efficient thermoelectric materials such as Bismuth Chalcogenides
that worked at room temperature, led to the realization of TEGs as promising body heat
harvesters for powering wearable electronics. These materials possess az T value around 1 at room temperature[35]. For energy harvesting from the human body, TEGs can be
attached to various body locations to generate power. The higher the temperature at a
specific location on the body, the higher the power generated by TEGs. Fig. 1.10 shows the
temperature at different body locations for various ambient temperatures[36]. According to
Fig. 1.10, forehead, back of the neck, and chest are warmer than any other body locations.
Using TEGs to harvest body heat however, has its own challenges, since there are several
parasitic thermal resistances involved including the contact resistance between the TEG
and the skin, the contact resistance between the TEG and the ambient, and the skin thermal
resistance. Each of these resistances also vary with ambient conditions and activity levels.
To overcome these challenges, careful material and device engineering is required.
A TEG can be represented by an equivalent electrical circuit consisting of resistors and
current/voltage sources. An analogous electrical circuit of a TEG on the human body is
shown in Fig. 1.11. The temperature difference between the skin and the ambient drops
across five thermal resistors in series including the resistance of the thermoelectric legs
in parallel with the filler material, two resistors which represent the thermal resistances
of the top and the bottom headers, and the two contact resistances which represent the
contacts between the TEG and the skin, and that between the ambient and the skin. In
order to maximize the temperature differential across the legs, the parallel combination of
the legs and filler thermal resistances should be as large as possible compared to the other
thermal resistances in the circuit.
A number of studies focused on optimization of thermoelectric modules for body heat
harvesting. These studies include modeling the performance of TEGs on the human body
[33], studying the effects of the packaging on the performance[37–40], identifying the
impor-tance of thermal resisimpor-tances presented in the whole module[33, 41], maximizing the thermal
resistances of the active regions of TEGs[42], varying the geometry of the thermoelectric
legs[43], developing multistage TEGs[44], designing heatsinks for improved performance
[45, 46], identifying the optimum load resistance for maximum power generation[41], and
developing external circuitry for optimum performance[44, 47–49].
Some of the best rigid TEGs for heat harvesting from the human body are reported in
Table 1.1. In this table, reported open circuit voltages and corresponding power densities are
summarized. All the data reported in this table was obtained relying on natural convection
for heat rejection on the cold side.
Table 1.1Performance of some of the best rigid TEGs for body heat harvesting.
Year Power (µW/c m2) Ambient Temperature (˝C) Reference
2016 6.1 18.3 [38]
2014 2.2 22 [50]
2013 60 20 [51]
2012 28.5 23 [43]
2010 7.3 22 [52]
2007 20 22 [44]
1999 5.6 22 [34]
1.4
Summary of the Thesis
Due to exponential growth in the IoT industry, there is an interest in using renewable energy
an abundant source of thermal energy, flexible TEGs have been explored in depth during
the recent decade for applications in wearable electronics. The main focus of this thesis is
to improve the performance of the flexible TEGs realized in this group in 2016 by F. Suarez
[53].
In this thesis, chapter 2 will cover a comprehensive review of the existing methods for
creation of flexible TEGs including both thin film and bulk fabrication techniques. For each
method, few examples will be provided and the pros and cons of each technique will be
discussed. Finally, the methods that provide the best thermoelectric performance will be
reviewed and explained as the starting point for the direction of this thesis.
Chapter 3 will discuss the work done on development and characterization of a high
thermal conductivity (HTC) elastomer, which is compatible with the first generation of
EGaIn-based flexible TEGs. This work provides a comprehensive study of thermal as well
as mechanical properties of the developed HTC elastomer made with inclusion of EGaIn
and graphene nano-platelets (GnPs) into polydimethylsiloxane (PDMS).
Chapter 4 will cover the development of the second generation TEGs made with the
developed HTC elastomer. Results from comprehensive COMSOL modeling will be
dis-cussed to clarify the role of the HTC elastomer and a metal spreader in performance of
flexible TEGs. The characteristics and performance of flexible TEGs made with both PDMS
and HTC elastomer with different thicknesses will be reported. Additionally, TEGs with
different fill factors will be compared in terms of both voltage and power output. Finally,
the TEG that provided the best power density is characterized on the human wrist at room
temperature and the characterization results are summarized.
Chapter 5 will summarize the development and characteristics of flexible heatsinks
made with the same elastomer as the top and and bottom encapsulation for improved
performance of TEGs.
Chapter 6 will cover the efforts on the modeling of the human body as the heat source
for varying ambient conditions and activity levels. The effect of age, sex, weight, and height
of a person on the generated heat from the body was also investigated. Finally, the thermal
resistivity of the skin was calculated as a function of the ambient temperature and the
metabolic heat generated by the human body.
Finally, chapter 7 will conclude this work with a summary of the major contributions of
this thesis and directions for future studies.
Appendices A-C will summarize other studies done related to this thesis, including
steam treatment of the elastomer surface for durable electroplating, other attempts to
CHAPTER 2
REVIEW OF THE EXISTING METHODS
FOR FABRICATION OF FLEXIBLE TEGS
2.1
Introduction
Although using commercial TEGs as body heat harvesters appears as a practical solution,
they do not provide optimal thermal contact to the human body due to their rigid nature.
Consequently, the interest in flexible TEGs that can form a conformal contact to the skin has
been rapidly growing in recent years. Aesthetically pleasing flexible TEGs can be integrated
into clothing, wristbands, armbands, or other form factors, enabling harvesting from a large
area on the body. However, it is challenging to produce flexible TEGs that can compete with
the performance of their rigid counterparts due to a variety of factors including lower quality
of the new flexible thermoelectric materials, limited electrical conductivity and flexibility
of interconnect materials, parasitic thermal resistances arising from flexible packaging
materials used as headers or as the filler material between the legs.
ma-terials on flexible substrates such as fabrics[54–62], polydimethylsiloxane (PDMS)[63–69],
polyethylene terephthalate (PET)[70–73], and polyimide (PI)[74–81]. Since neither typical
fabrics nor flexible substrates could withstand elevated temperatures, low-temperature
deposition techniques, which did not require subsequent annealing at elevated
tempera-tures were employed. Different organic thermoelectric materials were also reported, which
showed promising results[82–88]. Finally, several groups took a different approach and
in-tegrated rigid thermoelectric legs in a flexible package[53, 89–91]. Flexible TEGs fabricated
using these techniques are reviewed in the following section.
2.2
Synthesis of Thermoelectric Materials for Flexible TEGs
2.2.1
Physical Vapor Deposition (PVD)
In 2005, the first flexible TEG created with PVD technique was demonstrated using thermal
evaporation of copper and nickel lines onto a flexible substrate[92]. While the generated
output power level was very low (16p W{c m2), this study provided a first example of a flexible TEG fabricated using conventional PVD techniques. Shortly afterwards, a flexible
TEG with the same architecture was demonstrated using sputteredB i2T e3that showed orders of magnitude improvement in output voltage[93]. Since then, a number of studies
focused on PVD techniques intended for flexible TEGs. These include, evaporation[78, 94],
ultra-low power sputtering[95], magnetron sputtering[79], and virtual cathode deposition
[80].
Kong et. al. used magnetron sputtering to deposit thin films ofB i2T e3andS e2T e3on polyimide substrates[79]. The entire device was then coated with the PDMS elastomer
for protection and mechanical strength, which also suppressed microcrack formation.
However, the effects of the parasitic thermal resistances introduced by the elastomer was not
µV{K and 166µV{K respectively. The electrical properties of the thermoelectric films were
tested during repeated flexing cycles with a bending radius of 7m m. These measurements revealed less than 5% variation in electrical resistance after 2000 bending cycles. The thermal
conductivity andz T values were not reported for the deposited thermoelectric films in this work. A TEG fabricated using this technique with 13 pairs of thermoelectric legs produced
an output power of„700n W at a temperature differential of∆T=24˝C.
In another study, Sevilla et. al. depositedB i2T e3andS b2T e3films by ultra-low power DC sputtering. The material was deposited on a rigid silicon substrate coated with a thermally
grownSi O2layer[95]. Photolithography and etching was used to create planar legs with very large aspect ratios (2c mˆ4µm). Subsequently, Ti/Al contacts were deposited to serve as the
electrical interconnects between the thermoelectric legs. Lithography and isotropic etching
was used to form trenches in the underlying substrate, which allowed delamination of the
18µmthick device layer. The resulting TEG is shown in Fig. 2.1. Because the deposition was
carried on a Si wafer, the technique eliminated the drawbacks of flexible substrates due to
their limited thermal stability. Reported electrical resistivity values for the depositedB i2T e3 andS e2T e3films were 638µΩ.m and 2.22µΩ.m respectively. The corresponding Seebeck coefficients were and´37µV{K and 240µV{K respectively. Thermal conductivity and
z T values for the deposited films were not reported. The open circuit voltage and the corresponding output power (per thermocouple) generated by the TEG was reported to be
approximately 40m V and 0.15µW respectively at∆T “20˝C.
Morgan et.al. used virtual cathode deposition (VCD) to depositB i2T e3/G e T e films on a flexible polyimide substrate[80]. Since the technique provided deposition rates as high as
1µm{m i n at low temperatures (below 60˝C), the authors claimed its compatibility with
roll-to-roll manufacturing. The Seebeck coefficients of N-type and P-type materials were
mechan-ical properties of the deposited films were also not reported. TEGs fabricated using this
technique exhibited an output power of 0.2n W per thermoelectric pair at a temperature differential of∆T “20˝C.
In general, the thermoelectric materials deposited using PVD techniques could not
match the properties of the best bulk materials. Furthermore, these studies suggested that
the materials were prone to crack formation since the materials are rigid and brittle in
nature.
Figure 2.1Flexible thin film TEG made with DC-sputtered thermoelectric material[95].
2.2.2
Electrochemical deposition
Electrochemical deposition (ECD), which can be used to selectively deposit thermoelectric
materials in windows defined in large-area flexible substrates has been considered by
several groups. Since ECD allows deposition of the thermoelectric legs perpendicular to the
plane of the device, stresses developed during flexing can be absorbed by the surrounding
flexible material, thus making the device less prone to crack formation.
Glatz et.al. used ECD to selectively grow copper and nickel films in windows defined in
a flexible SU-8 mold[96]. In a follow-up study, the same group used ECD to growB i2T e3, which yielded significant improved devices. The electroplating was achieved usingH T e O`
120µm were deposited. Electrical interconnections between the thermoelectric legs were created using an electroplated Au layer. The device was tested with a bending radius of 7.5
m m, no changes in electrical resistance was reported. For 99 thermocouples, the output power of„750µW/c m2was reported at∆T “51.2˝C [97].
Yousef et. al. used ion-irradiation and etching to create submicron vias on a 75µmthick
polyimide foil (Kapton HN)[81]. The ion-irradiated polyimide could be selectively etched
in a sodium hypochlorite solution. A copper layer was then evaporated in the vias to act as
the seed layer for electrochemical deposition. Antimony and nickel were then selectively
deposited in the vias to form the thermoelectric legs. Evaporated copper strips were used
as the electrical interconnections between the legs. The electrical resistance of the TEG was
reported to be 54Ωper thermocouple. The measured Seebeck coefficient of a thermocouple
was 30µV/K which is comparatively lower than those of other reported thermoelectric
materials deposited with ECD technique. The power generated by the fabricated TEG and
mechanical properties of the device were not studied.
Yamamuro et. al. used potentiostatic ECD using a nitric acid-based solution to deposit
thin films (3-4µm) ofB i2T e3andS e2T e3on a stainless steel substrate[98]. The deposited films were then transferred to a thermoplastic resin substrate using a Kapton tape.
Subse-quently, the thermoelectric legs were connected in series using Ag paste. The deposited
material suffered from small grains and cracks that developed during deposition. The
mea-sured Seebeck coefficient of N-type and P-type materials were´63µV{K and 172µV{K respectively. The electrical resistivities of the deposited materials were 79S{c mand 3.2 S{c m forB i2T e3andS e2T e3respectively. The thermal conductivity andz T values of the deposited materials were not measured in this study. An output power level of 5n W was obtained for a pair of thermoelectric legs at∆T=60˝C. The performance of this device
under mechanical stress was not studied.
Using the ECD technique, other thermoelectric materials were also explored to create
polyethylene glycol terephthalate (PET) substrate using poly[Kx(Ni-ett)]as the
thermoelec-tric material[99]. A thin layer of sputtered Pt was used as the seed layer for electroplating .
thermoelectric legs with a thickness of 3µm and an area of 1.5ˆ2.5m m2were deposited on the substrate using a patterned PDMS mask. Evaporated Au electrodes were used as
the electrical interconnects between the thermoelectric legs. This was a planar device with
both hot and cold junctions on the substrate. The material was reported to have a Seebeck
coefficient and electrical conductivity of„ ´120µV{K and 250S{c mrespectively. Ther-mal conductivity andz T of the deposited material were not reported in the study. Output power of 26n W per leg was achieved at∆T “12˝C. Performance of the device under
bending condition was not studied.
Huu et. al. used a different device architecture, which employed thermoelectric legs
parallel to the plane of the substrate, while providing off-plane heat paths on both hot
and cold sides of the device using copper heat guides[100]. The TEG was made on a
sacrificial oxidized silicon substrate using pulsed electrochemical deposition ofB i2T e3 andS e2T e3 films with a thickness of 100 µm. The deposited thermoelectric materials were then annealed at a temperature of 250˝C. Stacks of Metal layers (Ti-TiN-Au-Cu) was
deposited on the thermocouples to connect them electricaly in series. Copper heat guides
were deposited on top of the device and the topside of the device was coated with a layer of
PDMS for mechanical stability.Si O2substrate was patterned and etched on the back side of the device and heat guides were deposited on this side as well. Finally, the substrate was
etched away completely and the backside of the device was coated with another layer of
PDMS. The fabricated device is shown in Fig. 2.2. The Seebeck coefficients of N-type and
P-type materials were´150µV{K and 170µV{K respectively. The electrical resistivities
of the deposited materials were 15µΩ.m (electrical conductivity of 6.67ˆ104S{m) and 25
µΩ.m(electrical conductivity of 4ˆ104S{m) forB i
of 3µW{c m2.
The main advantage of ECD technique for fabrication of flexible TEGs is the ability of
this approach in deposition of thick films of thermoelectic materials compared to PVD
techniques.
Figure 2.2Vertical flexible TEG made with planar ECD deposited thermoelectric materials and vertical (through-plane) heat guides.[100].
2.2.3
Screen Printing
Screen printing is a low cost, simple and highly scalable technique for depositing
thermo-electric materials on flexible substrates.
measured Seebeck coefficients of the materials were´119µV{K and 134µV{K forB i2T e3 andS b2T e3respectively. The electrical resistivities of the N-type and P-type materials were 0.35Ωc mand 0.02Ωc mrespectively. The reported electrical resistivities were higher than those of the bulk materials resulting in an output power of 48n W per thermocouple at
∆T “20˝C. Thermal conductivities,z T values, and mechanical properties of the deposited
materials were not reported.
Thermoelectric legs perpendicular to the plane of the TEG were also realized using
screen printing. Suemori et. al. created a vertical TEG using screen-printed CNT-polystyrene
composite on a polyethylene naphthalate film substrate[101]. The deposited material
consisted of 35 vol.% voids, leading to a light-weight TEG. The Seebeck coefficient and
electrical conductivity of the material were 57µV{K and 2.1S{c m respectively. A uni-polar TEG (using only N-type legs) composed of 1985 individual legs was made using this
material. Au was deposited as the electrical interconnects using PVD technique. This device
is shown in Fig. 2.3. This TEG was able to generate an output power of 5.5µW{c m2at
∆T “70˝C. No changes in electrical resistance of the TEG was observed during bending
cycles with a radius of 2mm. The properties of the screen-printed thermoelectric paste was
further improved by the addition of organic materials (PEDOT:PSS)[102]as well as using
new binding materials[103].
In another attempt, several planar devices were arranged vertically between two copper
substrates[104]. The thermoelectric material was screen-printed on a paper that showed a
Seebeck coefficient of 18µV{K and an electrical conductivity of 800S{c m. The reported thermoelectric figure of merit,z T, for this material was only 0.01 which is very low com-pared to the efficient thermoelectric materials. This TEG was capable of generating an
output power 0.1µW{c m2at∆T “50˝C.
In another approach, Iezzi et. al. created flexible TEGs using screen-printing for waste
heat recovery from steam pipes. Inexpensive metallic inks of silver and nickel were deposited
Figure 2.3Vertical uni-polar flexible TEG made with planar screen-printed thermoelectric mate-rials. a) Schematic of the TEG. b) Fabricated TEG.[101]
Ag and Ni films were 4.5Ωm and 1Ωm respectively. Other thermoelectric properties of the films such as Seebeck coefficients, thermal conductivity, andz T were not reported. Several devices were placed vertically on a steam pipe as shown in Fig. 2.4. These TEGs
were capable of generating an output power of 0.73µW per thermocouple at∆T “127˝C.
In another device, thermoelectric paste made ofB i2T e3andS b2T e3was pressed into a glass fabric followed by an annealing step at 530C in nitrogen ambient to improve the
thermoelectric properties of the legs[55]. Using this technique, thermoelectric materials
infiltrated the fabric and formed vertical (through-plane) legs through the substrate. The
reported Seebeck coefficient for N-type and P-type materials were -141µV{K and 98 µV{K respectively. Electrical resistivities of theB i2T e3andS b2T e3films were 0.67ˆ105 S{m and 1.5ˆ105S{m respectively. Thermal conductivies of the screen-printed materials were 1W{m K and 1.5W{m K for N-type and P-type materials leading toz T values of 0.35 and 0.3 respectively. Electrical interconnects were then screen printed on a sacrificial
silicon wafer coated with nickel, and transferred to the legs using silver paste. Finally, the
device was coated with PDMS for mechanical integrity. This device showed orders of
mag-nitude improvement in the output power compared to the previous screen printed devices
due to a significant reduction in the total electrical resistance of the device („0.35Ω), as well as lower thermal conductivity of the thermoelectric material („1 W/mK). Worn on the wrist, the TEG showed a power density of 1µW{c m2at an ambient temperature of 15˝C. Further improvement on the same device was achieved by using ball-milled ternary
bulk thermoelectric materials, as well as replacing the silver paste with a gold paste as
the bonding material[106]. Another improvement on this device was reported later by
adopting a post-annealing step at 500˝C followed by application of an external mechanical
pressure to further densify the screen printed thermoelectric material for improved
ther-moelectric properties[107]. With these modifications as well as altering the composition
of thermoelectric materials, the group was able to reachz T values of 0.89 and 0.57 for screen-printed P-type (B i0.5S b1.5T e3) and N-type (B i2S e0.3T e2.7) materials respectively. The same group also reported an improvement on the thermoelectric properties of the screen
printed N-typeB i2S e0.3T e2.7(z T “0.9) by annealing the material in a hydrogen-rich am-bient[108]. However, the hydrogen annealing step was later found to be detrimental to
in reduced reliability and reproducibility of the material[109]. This problem was resolved
later by adding different amounts of tellurium to the thermoelectric paste and optimizing
the doping concentration. A flexible TEG made with the screen printing technique is shown
in figure Fig. 2.5.
Figure 2.5Flexible TEG made with screen-printed thermoelectric material[106].
2.2.4
Dispenser Printing
Dispenser printing is a technique that allows depositing and patterning of materials without
using a mold. This method is attractive for the fabrication of flexible TEGs due to its low
cost. Dispenser printing of thermoelectric materials was first used to printB i2T e3powder mixed with a polymer binder into the holes in a PDMS substrate[64]. The electrical contacts
were created later by using flexible PCBs attached to the top and the bottom of the TEG.
Thermoelectric properties of the deposited materials such as electrical resistivity, thermal
successful in creation of a dispenser-printed TEGs, its performance was limited due the
high electrical resistance of the device (170Ω) leading to an output power density of 0.08
µW{c m2at∆T “19˝C. Number of thermoelectric legs used in the device and their sizes
were not reported. This technique was also later explored using fabric [110]as well as
polyimide[111]substrates. The output voltage generated by these two TEGs was significant
(tens of milli-volts); however, due to very high electrical resistances, both devices suffered
from output power levels less than a nano-watt for each thermoelectric pair.
An effort to use commercially available fabrics to create flexible TEGs led to the
realiza-tion of a silk-based TEG in 2016[54]. Using dispenser printing technique, N-type (B i2T e3) and P-type (S b2T e3) thermoelectric pastes were deposited onto the holes created in a silk sheet to form vertical columns of thermoelectric materials. The Seebeck coefficients for
N-type and P-type materials were -57µV{K and 111µV{K respectively. Other thermo-electric properties were not reported for the printed materials. The thermothermo-electric legs
were connected by Ag foil and Ag paste to achieve an electrically continuous device. In this
device, the high contact resistance between the legs and silver paste (with a total device
resistance of„7 kΩ), as well as the low density of the thermoelectric materials embedded
in the silk fabric resulted in a TEG with poor performance. The TEG reported in this work
was fabricated with 11 thermocouples, and generated an output power of 15n W for a temperature differential of∆T “35˝C. The device showed less than 10% increase in the
electrical resistance during 100 bending cycles.
In another study, a mixture of thermoelectric powders (B i2S e0.3T e2.7orB i0.5S b1.5T e3) and a stabilizer solvent was deposited on a Kapton substrate using ink-jet printing (Fig. 2.6)
[75]. This method was an attractive way to use ink-jet printers for fabrication of flexible
TEGs. Using this technique, N-type and P-type thermoelectric materials showed Seebeck
coefficients of -140µV{K and 177µV{K respectively. Thermal conductivities andz T values of the thermoelectric materials were not reported. Electrical conductivities of N-type
values (compared to those of bulk materials) for both electrical conductivity and Seebeck
coefficient were attributed to the low density of the deposited films.
In conclusion, the main issue associated with dispenser printed flexible TEGs is the
high electrical resistances of the thermoelectric materials which requires further studies
for improvement.
Figure 2.6Flexible TEG made with ink-jet printed thermoelectric material[75].
2.2.5
Organic and Solution-processable Materials for Fabrication of TEGs
There is an interest in developing organic-based thermoelectric element due to their
flexi-bility, low cost, low thermal conductivity, and non-toxicity. There are many studies focused
on understanding and improving the properties of organic thermoelectric materials[84,
112–117]. Solution based processes can be used to deposit organic materials onto various
flexible substrates[117].
The first organic-based TEG was made using poly (3, 4-ethylenedioxythiophene)
polysty-rene sulfonate (PEDOT:PSS), which is a P-type material, with an electrical conductivity
of 300S{c m, a Seebeck coefficient of 780µV{K (after exposing the polymer to tetrakis
(dimethylamino) ethylene vapor), a thermal conductivity of 0.3W{m K, and az T of 0.25 at room temperature[117]. For TEG fabrication, an organic conducting salt,
the N-type material (with a Seebeck coefficient of -48µV{K and an electrical resistivity of 11
kΩ{c m). The TEG was fabricated on an SU-8 mold using Au as the electrical interconnects.
The fabricated TEG was capable of generating an output power of 0.128µW (25 nW/c m2) at∆T “10˝C. While the initial thermoelectric performance was encouraging, the oxidation
of the organic material over time caused appreciable degradation in the device properties.
Massonnet et. al. reported fabrication of a flexible TEG using organic materials to
create a flexible heat flux sensor[115]. In this study, organic thermoelectric materials were
drop-casted onto a temporary polymer substrate. After drying, the thermoelectric legs were
transferred to a flexible substrate. The initial Seebeck coefficient measured for the deposited
thermoelectric material (PEDOT:PSS layer) was 70µV{K which rapidly decreased over
time and reached a value of„40µV{K after 400 hours. These materials are currently at their infancy and much work has to be done before they can be a viable alternative to
conventional materials[94].
Another organic-based TEG was realized in 2015 by dip-coating of a polyester fabric in a
PEDOT:PSS solution[56]. A larger plain polyester substrate was used as a platform to attach
the coated polyester strips. The electrical interconnects between the strips were created
subsequently using woven silver wires. The TEG made with five thermoelectric strips (with
an area of 40m m ˆ5m m for each strip) was able to generate an output power of 12.29 n W at∆T “75.2˝C. As expected, the main limitation to the performance of the device was
the planar orientation, making it challenging for integration into wearable systems where
the temperature gradient is usually vertical to plane of the device. Other disadvantages
of the TEG include the use of single P-type legs (a unipolar TEG), poor thermoelectric
characteristics of the material (with a Seebeck coefficient of„15µV{K), and high electrical
resistance of the device („2500Ω).
Solution-processable materials can also be considered to create flexible TEGs. Examples
S{m, and N-typeW S2nanosheet with a Seebeck coefficient of -75µV{K and an electrical conductivity of 1.4ˆ103S{m) spin coated on top of an octadecyltrichlorosilane
(OTS)-treated silicon substrate, transferred to a PDMS layer, and connected electrically by Ag and
Au wires[65]; conductive PEDOT:PSS as the P-type and indium tin oxide (ITO) coated with
PEDOT:PSS as the N-type materials (with an average Seebeck coefficient of„38µV{K and an average electrical resistance of 1.7 kΩper N-P pair) (Fig. 2.7)[118]; poly(vinyl alcohol)
(PVA) gel electrolyte doped with ferric chloride and potassium ferricyanide (with an average
Seebeck Coefficient of„1.1m V{K and an average electrical conductivity of 0.008S{c m) [119]; a solution ofT i S2nanosheets exfoliated from layered polycrystalline powders, andc60 nanoparticles as a hybrid material as N-type thermoelectric legs (with a maximum Seebeck
coefficient of 115µV{K and a maximum electrical conductivity of 480 S/cm) deposited on a polyimide substrate (Fig. 2.8)[120].
Figure 2.7A flexible and transparent TEG made with PEDOT:PSS and ITO-PEDOT:PSS[118].
All of these studies demonstrated the feasibility of creation of organic-based as well
as solution-processable flexible TEGs. However, more work is needed to either lower the
Figure 2.8A flexible TEG made withC60{T i S2nanosheets[120].
2.2.6
Fabrics
For heat harvesting from the human body,TEGs with larger surface areas are desirable .
Therefore, TEGs made on a fabric platform are attractive since they can be worn comfortably
while covering a large area on the human body.
In 2002, Yamamoto et. al. realized the first fabric-based flexible TEG using Knitted metal
wires[121]. The TEG was constructed with knitted chromel and Alumel thermocouple wires
through a perforated glass-based resin. This TEG was capable of generating an output
power as high as 0.166µW (for a thermocouple pair) at∆T “26˝C.
In 2012, Hewitt et. al. created a fabric-like flexible TEG using alternating stacks of doped
multi-wall carbon nanotube (MWCNT)/polyvinylidene fluoride (PVDF) sheets[122]. The
Seebeck coefficient measured for a stack made of 72 layers of fabric was 550µV{K. This fabric-based device (with an area of„40c m2) showed an electrical resistance of 1270Ω leading to a maximum output power of 137n W at∆T “50˝C. While promising, the main
disadvantage of this TEG was the planar orientation of the thermoelectric legs with respect
to the substrate, which made it challenging to integrate the device into clothing where the
heat flow from the hot side to the cold side is perpendicular to the orientation of the legs.
In another work, Yadav et. al. created a flexible fiber coated with alternating P-type and
N-type thermoelectric segments as a single TEG[94]. In this study, Ni and Ag electrodes