Thermoelectric generators: Linking material properties and systems engineering for waste heat recovery applications

Abstract

Waste-heat recovery with thermoelectric power generators can improve energy efficiency and provide distributed electricity generation. New thermoelectric materials and material performance improvements motivate development of thermoelectric generators for numerous applications with excess exhaust and process heat. However, thermoelectric generator product development requires solving coupled challenges in materials development and systems engineering. This review discusses these challenges and indicates ways system-level performance relies on more factors than traditional thermoelectric material performance metrics alone. Relevant thermo-mechanical and chemical material properties, system components such as thermal interface materials and heat exchangers, and system form factors are examined. Manufacturing processes and total system cost components are evaluated to provide product development and commercial feasibility contexts.

Full text

Thermoelectric generators: Linking material properties and systems
engineering for waste heat recovery applications
Saniya LeBlanc ⁎
Department of Mechanical & Aerospace Engineering, The George Washington University, Washington, DC 20052, USA
abstractarticle info
Article history:
Received 24 September 2014
Received in revised form 4 November 2014
Accepted 13 November 2014
Available online 20 November 2014
Keywords:
Thermoelectric generator
Waste heat recovery
Thermoelectric system
Thermoelectric manufacturing
Thermoelectric power generation
Waste-heat recoverywith thermoelectric power generators can improve energy efficiencyand provide distribut-
ed electricity generation. New thermoelectric materials and material performance improvements motivate de-
velopment of thermoelectric generators for numerous applications with excess exhaust and process heat.
However,thermoelectric generator productdevelopment requires solving coupled challenges in materialsdevel-
opment andsystems engineering. This reviewdiscusses thesechallenges and indicates ways system-levelperfor-
mance relies on more factors than traditional thermoelectric material performance metrics alone. Relevant
thermo-mechanical and chemical material properties, system components such as thermal interface materials
and heat exchangers, and system form factorsare examined. Manufacturing processes and total system costcom-
ponents are evaluated to provide product development and commercial feasibility contexts.
© 2014 The Author. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/3.0/).
1. Introduction
Thermoelectric devices offer a unique power generation solution be-
cause they convert thermal energy into electricity without requiring
moving components. Thermoelectric generators have been proposed
for waste-heat recovery applications, and advancements in thermoelec-
tric materials development have highlighted thetechnology's energy ef-
ficiency and commercial potential. To realize this potential and improve
thermoelectric power generation feasibility, the gap between thermo-
electric materials development and generator systems engineering
must be closed. The thermoelectric generator materials characteristics
are particularly important because it is a solid-state energy conversion
device. Electron and thermal transport through multiple materials in
the device is paramount and affects overall system performance. This
review provides a systems-level perspective of thermoelectric genera-
tor development. It underscores the relationships between thermoelec-
tric materials development goals and generator system requirements.
Considerations for system components beyond the thermoelectric ma-
terials are discussed along with manufacturing and cost issues.
A thermoelectric (TE) module consists of units, or legs, of n- and p-
type semiconducting materials connected electrically in series and ther-
mally in parallel. The figure of merit ZT describes material performance.
It depends on the thermoelectric material properties Seebeck coeffi-
cient S, electrical conductivity σ, and thermal conductivity k,and
ZT =S
2
σT/kwhere Tis the temperature of the material. A TE couple
is one pair of n- and p-type legs, and a module generally has several
couples. These couples and their electrical interconnects are enclosed
by an electrical insulator, typically a ceramic. A typical off-the-shelf
module is shown in Fig. 1. As depicted in Fig. 2, a thermoelectric gener-
ator (TEG)is usually a more extensive systemthan the module. In a TEG,
the modules are connected thermally in parallel with heat exchangers
to facilitate the transfer of heat from the heat source to the module's
hot side and away from its cold side. The modules are connected to an
electrical load to close the circuit and enable electricity extraction.
Thermoelectric generators have been used to power space vehicles
for several decades [1,2], so the research and development contribu-
tions and expertise from the space industry are invaluable in the devel-
opment of terrestrial waste-heat recovery TEGs. Named radioisotope
thermoelectric generators (RTGs), the heat source in spacecraft TEGs
comes from the nucleardecay of radioactive isotopes.RTGs were select-
ed to power space vehicles since they are highly reliable, robust, and
compact. They are solid-state devices without the rotating machinery
typical of other heat engines, so RTGs do not produce noiseor vibration.
These qualities made RTGs ideal for powering autonomous space vehi-
cles with long life missions. RTGs for space power systems have unique
characteristics which differentiate this application from the waste-heat
recovery applications discussed here. The heat source temperature is
typically higher (~1000 °C) resulting in the use of thermoelectric mate-
rials suchas silicon germanium which are suitable for high temperature
power generation. The operating environments are outer space and
other planetary surfaces. Moreover, the cost–performance consider-
ations and constraints for space vehicle development are significantly
different than for waste-heat recovery applications since significant
Sustainable Materials and Technologies 1–2(2014)26–35
⁎801 22nd Street NW, Phillips Hall, Suite 739, Washington, DC 20052, USA. Tel.: +1
202 994 8436; fax: +1 202 994 0238.
E-mail address: sleblanc@gwu.edu.
http://dx.doi.org/10.1016/j.susmat.2014.11.002
2214-9937/© 2014 The Author. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/).
Contents lists available at ScienceDirect
Sustainable Materials and Technologies

value is placed on the RTG primary power generation capability and
unique suitability for the requirements of space applications.
From electronics to industrial furnaces, numerous waste-heat
sources at low- (b250 °C), mid- (~250-650 °C), and high- (N650 °C)
temperatures exist. TEGs have mostly been proposed for waste-heat re-
covery in mid- and high-temperature applications such as automotive,
engine, and industrial applications with untapped exhaust and process
heat because of the potential for appreciable power generation [3–5].
The mid- to high-temperature exhaust and process heat types of appli-
cations are the focus of the discussion here, and sample heat source
temperatures and applications are shown in Table 1. Vehicle applica-
tions include passenger vehicles and large trucks, and prototypes have
already been demonstrated [5,6]. There are a myriad of industrial pro-
cesses such as steel making and glass melting, and a comprehensive as-
sessment of industrial waste-heat opportunities in the United States
was conducted in 2008 [3]. The study notably indicated that thermo-
electric generation had not been demonstrated in U.S. industrial
applications. In Asia, particularly Japan, there have been multiple dem-
onstrations of TEG waste-heat recovery [7,8] with a recently deployed
TEG generating 250 W from a steel casting line [9].
As in space applications, key advantages of TEGs for waste-heat
recovery are their simplicity, minimal maintenance requirements, and
reliability since there is no rotatingmachinery in the system. Disadvan-
tages include low efficiencies, high costs, and systems integration
barriers. The assessment for TEG waste-heat recovery potential often
focuses on the heat source temperature where high-temperature pro-
cesses are favorable. Government-initiated studies and funding for
TEGs reflect the interest in these promising, high-temperature industri-
al process applications. Additionally, a crucial consideration for TEG
product development and commercial viability is identifying the appro-
priate fit between the productand potential markets [12]. The commer-
cial drivers for product–market fit can lead to preference for mid-
temperature applications over high-temperature applications. This
product–market fit is closely related to practical materials development
and systems engineering needs which do not necessarily correlate to
obtaining higher material ZT values.
The following discussion provides key points about the link between
TEG material properties and system performance. The target readers for
this overview are those who are outside the hermoelectrics research
and development community and/or want to understand overarching
key issues with thermoelectric materials and systems development.
Readers whoare interested in the extensive literature on thermoelectric
materials [13–16], systems design and development [11,17–20],and
specific applications [21–23] are referred to more specialized books
[16,24,25] and literature. In this review, an overview of thermoelectric
materials considerations is followed by a description of system compo-
nents and design factors and a comparison of material versus system
performance metrics. Lastly, an overview of systems manufacturing
and cost considerations is provided with particular attention paid to
materials concerns.
2. Materials overview
Thermoelectric materials are typically classified by material struc-
ture and composition. Someof the main classifications are chalcogenide,
clathrate, skutterudite, half-Heusler, silicide, and oxide. Excellent re-
views of thermoelectric materials have provided descriptions of both
the materialclassifications and therelationship between material struc-
ture and thermoelectric properties [13,14,26], so comprehensive de-
scriptions are not provided here. Chalcogenide materials have a long
history of demonstrated thermoelectric use with bismuth telluride
and lead telluride being the most prominent. Commercial, off-the-
shelf thermoelectric modules for low temperature use are primarily
made with bismuth telluride and its solid solutions with antimony or
selenium. Lead telluride has better thermoelectric properties at higher
temperatures (~500–600 °C). Materials engineering of clathrates and
skutterudites has involved introduction of void-filling or guest atoms
into a base structure. These additions can optimize electron concentra-
tion or act as phonon scattering sites. Such materials engineering to
achieve a glass-like thermal conductivity combined with good charge
carrier mobility has been termed the “phonon glass electron crystal”
approach. With one vacant sublattice in the crystal structure, the
properties of half-Heusler materials have also been improved through
void-filling as well as doping of the filled sublattices. Silicides have
generated interest due to the low cost of their abundant materials (i.e.
Fig. 1. Picturesof an off-the-shelf thermoelectric module. (a) Side view showing multiple
thermoelectric leg couples. (b) Interior of module with one substrate removed to reveal
the electrical interconnects and solder joints. The module pictured is approximately 1 in.
by 1 in. and supplied by Marlow Industries, Inc.
Fig. 2. Schematic of one thermoelectric couple in a thermoelectric generator system. A
module consists of many couples between electrical insulators. The modulesare connect-
ed to heat exchangers to interface with thermal reservoirs in the waste-heat recovery
application.
Table 1
Approximate waste-heat source temperatures are provided for sample mid- and high-
temperature TEG applications. Temperatures will vary based on TEG position in the sys-
tem, and thetemperature at the hot side of the thermoelectric will be lower than the heat
source temperature.
Application Heat source temperature Reference
Automotive exhaust 400–700 °C [10]
Diesel generator exhaust ~500 °C [11]
Primary aluminum Hall–Heroult cells 700–900 °C [3,4]
Glass melting regenerative furnace ~450 °C [3,4]
27S. LeBlanc / Sustainable Materials and Technologies 1–2 (2014) 26–35

silicon), and oxides are expected to have high temperature stability in
air. Notable advancements have been made in both the types of mate-
rials synthesized and the reportedproperties. For low temperature ther-
moelectrics, nanostructured bismuth telluride [27], polymers and
polymer–inorganic matrices [28–31], and MgAgSb-based materials
[32] have broadened the range of options. The reported material prop-
erties for high temperature thermoelectrics have demonstrated note-
worthy gains resulting in ZT values above 1. Hierarchical nano- to
meso-scale structuring [33] and new materials such as tetrahedrites
[34] have contributed to the gains.
While the materials developmentprogress is promising, the increas-
ing breadth of materials and the reports of ever-increasing ZT mask the
underlying challenges of employing the materials in devices. Material
properties are highly temperature-dependent, posing multiple chal-
lenges for application-specific materials selection. Few applications
have heat sources at one single temperature, so matchingan application
temperature with the point of peak ZT in a thermoelectric material is
unrealistic. Instead, most applications have some degree of thermal
fluctuation or cycling. If the heat source is a fluid stream, the tempera-
ture of the fluid varies along the flow direction. The temperature
decreases along the length of each thermoelectric leg, as well. Ways of
combining different materials in one device through cascading,
segmentation, and varying the material in the fluid flow direction
have been developed [35–37]. Additionally, peak ZT values for high-
temperature materials often occur at temperatures above 600 °C. How-
ever, the temperature at the hot side of the thermoelectric material it-
self is lower than the heat source temperature because there are
system component thermal resistances between the heat source and
the thermoelectric material. There are applications with heat sources
above 600 °C such as industrial process furnaces and exhaust streams,
and there are many opportunities for thermoelectric power generation
in applications with heat sources in the 250–500 °C range [12].Inboth
cases, average ZT over the application temperature range is highly rele-
vant to material selection and design. Fig. 3 presents ZT values for a high
peak ZT material as well as materials which have generated commercial
and product development interest and results [38–40].
Material stability over the full operating temperature range is rele-
vant to device engineering. Thermoelectric materials must be stable
within the filler medium. If the thermoelectric material is exposed to
air, the material must not oxidize within the operating temperature
range. Devices can be packaged in an inert gas to mediate this problem.
In low temperature devices, a solid filler is sometimes used, but solid
filler media for high temperature applications –if they currently exist
–must also be sufficiently stable and inactive with the TE material. For
example, some materials undergo sublimation within the operating
temperature range of high temperature applications. Even if the materi-
al properties are sufficient for most of the range or the nominal operat-
ing temperature, the risk associated with material sublimation during
any temperature spikes is severe device performance degradation or
even device failure [44]. Hence, reports of ever higher peak thermoelec-
tric material ZT values deceptively allude to more power generation po-
tential which may not be valid given realistic application parameters.
The translation of thermoelectric materials into devices also requires
consideration of both n- and p-type thermoelectric materials. Although
a thermoelectric unit, or couple is typically composed of n- and p-type
legs, both are not strictly necessary. For instance, if ease of synthesis,
material stability, and reliability are challenges for the n- or p-type var-
iation of a new material, faster prototype/product development, dem-
onstration, or commercialization could be achieved with one type (n
or p) coupled with a metalshunt. Using both n- and p-types of materials
significantly increases the thermopower and thus the open circuit volt-
age of the TEG since the total thermopower for a p–n leg couple, S
pn
,
equals the sum of the magnitude of each leg's thermopower. Again, ad-
vances in material ZT are thusdeceptive. For instance, an increase in the
ZT of an n-type (p-type) chalcogenide might be achieved, but a similar
increase in the properties of a corresponding p-type (n-type) chalco-
genide is not. Hence, the overall gain in ZT for the device is not propor-
tionally as high as the gain in ZT for the individual material.
Although the topic receives far less attention than thermoelectric
properties, mechanical properties of thermoelectric materials are criti-
cal for both device manufacturingand operation. Typical thermoelectric
materials are brittle and behave mechanically like ceramics. Table 2 pro-
vides a perspective of measured parameters which reflect the brittle
quality of these materials. However, it is challenging to characterize
the brittleness as measured by hardness, fracture strength, and fracture
toughness. Characterization results vary based on sample geometry and
testing technique even when the material is held constant [45,46],and
changes in composition during alloying of the same set of elements
can lead to variations in microstructure and mechanical proper-
ties [47].
The coefficient of thermal expansion (CTE) is an important parame-
ter due to the high operation temperatures and thermal cycling in po-
tential TEG applications [44,53]. Because each thermoelectric leg is
held rigidly in place by a solder joint to the electrical interconnect and
substrate, significant stresses can build up in the material, particularly
at the corners and edges. The CTE mismatch between the TE material,
interface layers, interconnects, and substrates exacerbates the problem
of stress concentration. Proposed solutions involve introduction of com-
pliant interface materials. Examples include liquid metal layers [54] and
novel structures such as carbon nanotube arrays [55].
3. System design and components
The power generation potential of a TEG depends largely on the sys-
tem design, not only the TE material. In particular, the thermal and elec-
trical impedance of the TE material relative to the impedances of the
rest of the system components influence system output and perfor-
mance [56–58]. For instance, the arrangement of the TE material within
the TE module significantly affects thermal and electrical transport in
the overall device. Although the typical unicouple formation is like the
“pi”shaped structure, alternative structures like couples which opti-
mize thermal and electrical transport have been proposed such as the
“y”shaped connectors discussed in [54]. When considering material
properties alone, the tradeoff between thermal conductivity and electri-
cal resistivity is often discussed. However, at a device-level, the conflict-
ing parameters become thermal conductance and electrical resistance
which include the physical geometry of the material in addition to the
intrinsic material properties. There is a tradeoff between the total
Fig. 3. The temperature range for typical waste-heat sources and TEG hot side operation
values is overlaid on ZT versus temperature data for a top-performing thermoelectric ma-
terial and ones which have generated commercial interest. High peak ZT values can mask
the need for large average ZT values over the application temperature range. The data
shown here for high ZT, p-type PbTe endotaxially nanostructured with SrTe (circles) are
extracted from [33] an d show a peak ZT of 2.2 at 642 °C. Skutterudite (squares, dat a
depicted for Ba
0.08
La
0.05
Yb
0.04
Co
4
Sb
12
)[41], tetrahedrite (diamonds, data depicted is for
0.5 natural mineral Cu
9.7
Zn
1.9
Fe
0.4
As
4
S
13
and 0.5 synthetic Cu
12
Sb
4
S
13
)[42], and half-
Heusler (triangles, data depicted for Nb
0.6
Ti
0.4
FeSb
0.95
Sn
0.05
)[43] materials have generat-
ed commercial interests based in part on their average ZT values.
28 S. LeBlanc / Sustainable Materials and Technologies 1–2 (2014) 26–35

thermal conductance and the electrical resistance of the TE material at
the module level, and two key features of module geometry, the fill fac-
tor and TE leg size, play a large part in this tradeoff. The fill factor is the
ratio of the amount of module surface area occupied by TE material to
the overall surface area. The total surface area is not necessarily covered
by TE material (i.e. the fill factoris not 1) due to performance and/or cost
optimization [58] and manufacturing constraints like minimum inter-
leg spacing during assembly. Likewise, the tradeoff between thermal
and electrical parameters necessitates an optimization of the TE leg
height.
The materials needs for thermoelectric devices extend beyond the
TE material alone. There are several other device components, some of
which interact directly with the TE material, affecting overall device
performance. For instance, there is a metallurgical bond between the
TE leg and the metal interconnects. There are engineering consider-
ations which influence the selection of the solder or braze material
used to form this bond. The solder/braze material must not diffuse
into the TE material; a diffusion barrier placed on the TE materialsurface
is typically required [59–62]. The electrical contact resistance between
the solder/braze material and TE material must be low. The reflow tem-
perature of the solder/braze should be higher than the device operating
temperature to prevent TE leg movement or changes in electrical resis-
tance during operation. Additionally, the thermal contact resistance be-
tween the TE leg and the metal interconnect and between the metal
interconnect and substrate will affect the temperature drop, and thus
voltage, achieved across the thermoelectric leg.
There are challenging materials requirements for the module sub-
strates and geometries. The substrates must be electrically insulating
but thermally conducting. Sincethe substrate must withstand high tem-
peratures, particularly on the hot side of the TE leg, the substrates are
typically ceramic materials, and flexible substrates for high temperature
TE devices have not been demonstrated. The TE legs are attached to the
substrate, so the substrate must have the mechanical strength to sup-
port the legs and interconnects. Module design must incorporate the
ability to shape ceramic to the desired design structure, so most
modules have a square or rectangular shape. If the application requires
a different form factor, rectangular units are often positioned to accom-
modate this form factor. There have been limited demonstrations of cy-
lindrical modules as shown in Fig. 4 [63,64].
The TE module is attached to the exterior system with thermal inter-
face materials (TIMs) to improvethe transfer of heat toand from the hot
and cold sides of the TE device. The operating temperature, particularly
on the TE system hot side, creates a critical constraint onthe TIM used as
these temperature can be above 200 °C. TIMs come in various forms in-
cluding thermal greases or pastes, gap filler pads, and phase change ma-
terials [65–67]. However, the typical maximum operating temperature
of most greases or pastes is approximately 150 °C. At hightemperatures,
greases and pastes dry out leaving air gaps which greatly increase the
thermal resistance at the interface. Gap filler pads and phase change
materials are more expensive but more reliable for hot side interfaces.
Carbon-based TIMs are able to operate at higher temperatures with typ-
ical maximum operating temperatures near 300 °C, but the thermal
conductivity is often anisotropic with the through-plane value being
lower. Novel interface materials based on carbon nanotube arrays and
composites have been proposed and experimentally demonstrated
[55,68–70], and they have the potential to provide an appropriate TIM
solution for mid- to high-temperature TEGs.
Thermoelectric systems for most applications require heat ex-
changers on both the hot and cold sides of the device. The heat ex-
changers enable sufficient heat transfer from the heat source to the TE
module hot side as well as heat rejection or cooling on the cold side.
Therefore, the effectiveness of the heat exchangers directly impacts
the temperature drop (and thus voltage) across the TE material. The
overall heat transfer capability of the heat exchanger, often denoted
by an overall heat transfer coefficient U, depends on the exchanger de-
sign and material as well as the heat exchange fluid. The value of Uis
much higher when the fluid is a liquid, but use of a liquid adds complex-
ity to the system as a closed-loop system must be implemented. In most
waste-heat recoveryapplications, the heat source is a gas streamsuch as
flue gas from a chimney or furnace, so a liquid working fluid on the hot
Table 2
Mechanical properties of selected thermoelectric materials. Measurement error and standard deviations are available in the original references.
Material type Material Young's modulus
E
(GPa)
Hardness
H
(GPa)
Fracture toughness
K
Ic
(MPa-m
1/2
)
Fracture toughness
K
c
(MPa-m
1/2
)
Fracture strength
σ
f
(MPa)
Reference
Skutterudite CeFe
3
RuSb
3
133 –1.1–2.8 –37 [45]
CoSb
3
136 –1.7 –86 [45]
PbTe-based Ag
0.43
Pb
18
Sb
1.2
Te
20
54.2 0.98 –0.34 28.1 [48]
Pb
0.95
Sn
0.05
Te–PbS 8% 54 1.28 –0.31 –[49]
Oxide Ca
3
Co
4
O
9
86 2.6 2.8 ––[50]
Ca
0.95
Sm
0.05
MnO
3
205.2 11.05 4.99 ––[51]
Antimonide Zn
4
Sb
3
75 1.65 0.68 –65 [52]
Fig. 4. Cylindrical TEG system produced by Gentherm (a) (figure reproduced from [63]) and cylindrical ingots of doped lead telluride (b) (figure reproduced from [64]).
29S. LeBlanc / Sustainable Materials and Technologies 1–2 (2014) 26–35

side is unrealistic. The heat exchangers must also be integrated into the
system with the TE module which adds system complexities such as in-
creased weight and size and additional thermal interfaces.
As indicated earlier, the optimization of TEG power output and effi-
ciency depends on TE material properties and dimensions as well as
system-level electrical and thermal resistances. Hence, the selection of
electrical and thermal contact/interface materials, substrates, and heat
exchangers directly impact TEG performance, and the intricacies of
this optimization have been explored through analytical and system
modeling. The impact of asymmetric system-level thermal resistances
on either side of the TE material has been investigated for idealized sys-
tems with constant temperature heat source, fill factor of 1, matched or
averaged n- and p-type material properties, temperature-independent
TE material performance, and negligible electrical contact and parasitic
thermal resistance approximations [57,71]. The optimum ratio of
electrical load to TE internal resistance was shown to take the form
√(1 + ZT). The optimum ratio between the hot and cold side system-
level thermal resistances took on a value of 1 for maximum power out-
put [57]. A recent study demonstrated the impact of a constant heat flux
versus constant temperature boundary condition, and it included elec-
trical contact resistances to the TE material [72]. A detailed TEG system
model indicated the optimum hot to cold side thermal resistance ratio
must be larger than 10–30 to achieve maximum TEG power output
[73]. This model was quite extensive in that it incorporated TE material
property temperature dependence, unmatched (real) n- and p-type
material properties, electrical and thermal contact resistances, parasitic
heat loss components, hot and cold side heat exchanger models, and TE
device optimization models. Thus, TEG output and performance is high-
ly dependent on the link between actual material and system-level
parameters.
4. Material vs. system metrics
Typical metrics for TEGs are ZT,efficiency, and power output. While
they are mostly reported in terms of intrinsic material properties, dis-
cussions about system-level performance point out alternative formula-
tions which account for the contribution of other device and system
components [16,58,74,75]. For instance, while ZT is often reported in
terms of TE material properties S,σ,andk, a more relevant formulation
for a module-level ZT parameter is S
pn2
T/(KR). The extrinsic parameters
Rand Kindicate electrical resistance and thermal conductance of the TE
material in a couple; these are dependent on the dimensions of the legs.
Thermoelectric efficiency is often reported on a materials basis; it is
the electrical power generated divided by the thermal energy trans-
ferred into the TE material. However, the device- or system-level effi-
ciency is actually lower since not all of the heat available to the
system is transferred to the TE material. The effectiveness of heat ex-
change between the heat source and the TE device, primarily through
the heat exchangers, determines the amount of heat transferred to the
TE material. Fig. 5 shows the difference between thermoelectric materi-
al efficiency and TEG system-level efficiency for simulated TE waste-
heat recovery systems in three combustion applications [4,76,77].The
TEG system efficiency can be considerably lower than the TE material
efficiency which is often reported withmaterial characterization results.
Only the system parameters were changed in the simulation, and the
selected thermoelectric material was held constant [76].Thedifference
between the material and system efficiencies can be reduced if the
thermoelectric material selection is optimized for the application
temperature.
Frequently, the TEG efficiency is less of a concern than the total
power output since the power generated by the TEG is the valuable
component to the end user. Sufficient understanding of the value of
the power output depends on system-level parameters. There are sys-
tem components which require power input, so the power generated
by the TE device alone is less relevant than the net power, power gener-
ated less power required for system components, of the entire system.
For instance, power is required to blow or pump a cooling fluid through
the cold side heat exchanger. Although a higher mass flow rate of work-
ing fluid in the heat exchanger results in the benefit of more effective
heat exchange, increased pumping power is required to achieve this
benefit. Additionally, power electronics may be used in a TEG system
or product. For example, these power electronics can be used for
power conversion (e.g. converting DC TEG output to AC input for an
electrical grid), and/or they can optimize the electrical load (e.g.
matching a variable circuit load resistance to the fixed power source
electrical resistance to maximize thermoelectric power generated).
These power electronics can also have a power input requirement for
operation. The net power delivered by the TEG, and thus the power gen-
erated for the end user, must account for these parasitic power require-
ments in the system. The end user may have requirements for the
minimum net power which must be delivered for the investment in
TEG power generation to be worthwhile. This is an important consider-
ation to determine the feasibility of TEG power generation in various
applications.
The net power generated must also be understood relative to system
sizing features. In most applications, there is limited space and capacity
available in which to accommodate the TEG. The total surface area and
weight of the TEG system influence the ability to integrate it into
existing systems and architectures, so the power density (power gener-
ated normalized by surface area required or available) and specific
power (power generated normalized by TEG system mass)are key met-
rics. For example, industrial furnaces have a finite surface area in which
a TEG can be incorporated. Automotive exhaust systems have finite
space to accommodate a TEG, and the additional weight of the TEG on
the vehicle entails a consequential fuel use requirement [78,79].
5. Manufacturing
The manufacturing process for thermoelectric devices varies based
on the type of thermoelectric material employed. The overview provid-
ed here applies to bulk materials and the most traditional manufactur-
ing process. It is not comprehensive for all types of TE materials and
devices but is provided as a reference to enhance understanding about
engineering challenges commonly experienced with materials. A sche-
matic of a TE device manufacturing process is shown in Fig. 6.
Process yield is a significant consideration for the manufacture of TE
devices. As opposed to materials development and characterization
where the primary yield concern corresponds to material synthesis
and the number of samples which can be characterized, there are
many more steps in the device manufacturing process. Thermoelectric
material synthesis is commonly accomplished through ball milling
powders of the constituent elements. While this can be a lengthy pro-
cess, it is fairly repeatable with high yield once a processing recipe is
Fig. 5. Thermoelectric material efficiency compared to generator system efficiency simu-
lated for three potential applications [76]. The syste m efficiency is 32%, 33%, and 59%
lower than material conversion efficiency for the water heater, automotive exhaust, and
industrial furnace applications, respectively. The variation can be reduced by selecting
the thermoelectric materials based on each application's operating temperature range.
30 S. LeBlanc / Sustainable Materials and Technologies 1–2 (2014) 26–35

established. However, establishing the process to repeatedly achieve a
specific composition can be challenging. Complete mixing is required,
and sufficient energy must be imparted to the powder particles. Ther-
moelectric materials can also be made through melting processes.
Phase diagrams [80] of TE materials demonstrate the narrow process
windows in which each phase is achieved, and the thermoelectric mate-
rial properties depend on the phase. For instance, AgSbTe
2
is in PbTe-
based materials such as TAGS and LAST thermoelectric materials in
which microstructures reportedly improve thermoelectric performance
[33,81]. The phase diagram for AgSbTe
2
(Fig. 7) demonstrates the com-
plex variations in phase, and thus microstructures, which may arise due
to process parameters [82]. Hence, process fluctuations can result in
variations of phase, microstructure formation, repeatability of the syn-
thesis process, and resulting material properties.
Once the TE material is synthesized in powder or particle form, it is
consolidated into ingots, typically through hot pressing or spark plasma
sintering. As discussed in the materials overview section, sublimation
and stability/reactivity in the process gas environment must be consid-
ered. While a given TE material might be stable at the application's oper-
ating temperature, the consolidation process (or other manufacturing
processes) may occur at a higher temperature in order to densify the
material.
The ingots are then diced to form the thermoelectric legs, and the
material brittleness is a primary concern in this process step. Any
chipping of a thermoelectric leg changes the leg surface area and influ-
ences both the leg's thermal and electrical resistance. This in turn affects
system performance, particularly when the geometry of each leg is not
the same. Chipped legs would not be used in a device, reducing
the manufacturing process yield. Moreover, dicing and subsequent
chipping can lead to crack initiation. These cracks can propagate in sub-
sequent process steps, or worse yet, during TE device operation. Even
microcracks can affect thermal and electrical transport in the TE mate-
rial. The dicing step also constrains the achievable leg dimensions
since the post-dicing leg dimensions depend on the dicing equipment
capabilities (e.g. saw blade width, dicing depth).
The advent of thin film thermoelectrics has introduced manufactur-
ing processes which are commonly associated with the semiconductor
industry. Structures such as superlattices and nanowires have been pro-
posed and are made with deposition, growth, and/or etching micro/
nanofabrication processes including molecular beam epitaxy, chemical
vapor deposition, vapor–liquid–solid growth, and electroless etching
[83]. Additionally, recent demonstrations of thermoelectric materials
Fig. 6. Schematic showing typical steps for manufacturing of thermoelectric module.
Fig. 7. Phase diagram of AgSbTe
2
, a common component in microstructured PbTe-based
thermoelectric materials. Th e complexity and uncertainties about the bounds o f the
phase diagram demonstrates how small process fluctuations can affect thermoelectric
material formation.
The figure is reproduced from [82].
31S. LeBlanc / Sustainable Materials and Technologies 1–2 (2014) 26–35

formed through solution processing techniques have been shown for
both inorganic and organic materials. Since these approaches to making
thin film thermoelectrics are quite varied, a generalizable, standard
manufacturing process flow is not yet available.
Materialcharacterization during manufacturingposes a unique chal-
lenge. Both standard and custom-built equipment exist to characterize
thermoelectric properties of TE materials [84–87]. However, these are
designed to accommodate individual samples of TE material (e.g. an
ingot or one TE leg), and specialized contacts must be made to the sam-
ple. Hence, the techniques are destructive in that the characterization
sample cannot subsequently be used for a TE device. The characteriza-
tion process is also time-consuming, so it becomes a bottleneck in the
overall manufacturing process stream. Standard, in-line characteriza-
tion techniques which could be implemented within the manufacturing
process would be valuable. They would enable characterization of the
material as it is incorporated into the device as well as the effective
properties of the material combined with its electrical and thermal con-
tacts to other device components. In-line characterization could also in-
crease process yield and reduce waste since defective devices would be
detected earlier in the manufacturing process.
6. Cost considerations
The thermoelectric materials in establisheddevices are based on bis-
muth telluride, lead telluride, and silicon germanium. The high cost of
tellurium and germanium is touted as a reason to develop thermoelec-
tric materials which do not use expensive elements [88]. The develop-
ment of polymer, silicide, oxide, and tetrahedrite TE materials is
motivated strongly by the need to lower costs and improve the com-
mercial viability of TEGs. Fig. 8 demonstrates the raw materials costs
associated with sample materials of the main types of TE material clas-
sifications. The materials cost depends heavily on material composition.
Within a given classification of material, the cost can vary by a factor of
two and sometimes even an order of magnitude, particularly when a
relatively inexpensive material is doped with an expensive element.
The doping is done to achieve better performance (e.g. higher ZT), so
it is particularly helpful to consider both cost and performance as
discussed at the end of this section.
The standard practice is to synthesize thermoelectric compounds
with high purity starting materials, and the purity of materials affects
cost with higher purityresulting inhigher costs. While the effect of ma-
terial purity on TE material properties has not been widely established
[90–93], and an up-to-date, systematic study would be useful, thepres-
ence of impurities in the TE material likely degrades the properties. The
relativedifference in material cost based on purity level is demonstrated
in Table 3 for sample materials.
While thermoelectric material costs are non-negligible, the system
component costs can overwhelm the material costs. Major contribu-
tions to system costs are due to the substrates and heat exchangers.
Substrates are typically thin ceramic pieces, often alumina-based, with
off-the-shelf costs of approximately $0.10/cm
2
. The cost of the hot and
cold side heat exchangers is particularly significant. Heat exchangers
have to be customized to the TEGin order to accommodate the form fac-
tor of the device and optimize system performance, so it is challenging
to specify cost. However, cost estimates can be determined from off-
the-shelf units. Heat exchanger cost generally scales with the overall
heat transfer coefficient [58],orU-value, as shown in Fig. 9.
Both material and device manufacturing costs are relevant for TEG
commercialization. Particularly with the advent of nanostructured ther-
moelectric materials, the cost of manufacturing techniques required to
make thermoelectric materials can vary widely. The costs vary based
on the complexity of the manufacturing process and the maturity of
each process step's technology. In particular, the costs associated with
lithographic or epitaxial deposition processes for thick/thin film mate-
rials are orders of magnitude larger than those associated with bulk ma-
terial processing [75]. While there may be a tradeoff between increased
material synthesis cost and assembly cost for thermoelectric thick/thin
film materials, such comparisons are not readily available and remain
Fig. 8. Cost of various thermoelectric materials based on the raw material costs [89] of the constituent elements.
The exact costvalues and calculations are provided in[75].
32 S. LeBlanc / Sustainable Materials and Technologies 1–2 (2014) 26–35

largely unspecified. The cost of assembly is a significant consideration.
Assembling the thermoelectric legs into the typical unicouple formation
can be performed manually or automated through pick-and-place or
shaker machines, and the cost depends on the assembly technique.
Since exact cost values for assembly steps are not publicly available,
the impact of this cost is difficult to project.
Various cost analyses for thermoelectric devices have been devel-
oped and use different approaches. An early cost calculation used fuel
cost and module construction costs for given leg thicknesses [74].The
tradeoff between cost and efficiency was investigated by first determin-
ing optimum device geometry followed by calculating material cost for
that geometry [104]. A recent analysis created a cost–performance met-
ric which coupled materials, manufacturing, and system costs with TEG
performance [58], and the cost in $/W for TEGs made with various TE
materials was calculated [75]. The analysis concluded that bulk thermo-
electric materials can achieve costs below $1/W for applications where
the thermoelectric material temperature is higher than 275 °C, and the
high costs of heat exchanger and ceramic substrate system components
present a barrier to achieving low-cost thermoelectric generators.
Particularly for high-temperature applications, it is relevant to deter-
mine whether alternative heat recovery technologies are more cost-
effective than TEGs. For instance, the higher conversion efficiencies of
organic Rankine cycle systems could outweigh the impact of their capi-
tal cost and maintenance requirements.
7. Conclusion
The abundance of waste-heat sources and increasing energy effi-
ciency goals make waste-heat recovery with thermoelectric power gen-
eration a promising technology. The realization of commercial
thermoelectric generators hinges on solving the intimately coupled
challenges with materials development and systems engineering. Mea-
suring system performance with thermoelectric material ZT alone is
insufficient for determining generator performance, and other thermo-
mechanical/chemical material properties and components strongly
impact product development. Major issues to resolve for TEG commer-
cialization are material selection based on average (not peak) ZT,mate-
rial thermal and chemical stability, engineering of interfaces and
interface materials, and optimization of hot and cold side thermal resis-
tances (e.g. heat exchangers). Moreover, the manufacturability of ther-
moelectric devices combined with the total system cost will influence
the technology's time-to-market, readiness of product supply, and cost
competitiveness. Active research and development efforts along with
the emergence of new prototypes and pilot systems indicate solutions
to the linked materials and systems challenges are well underway, and
thermoelectric generators can contribute to sustainable development.
Acknowledgments
The author gratefully acknowledges support through the NSF/DOE
Partnership on Thermoelectric Devices for Vehicle Applications (Grant
No. 1048796), Precourt Institute for Energy, Sandia National Laborato-
ries Fellowship, and Stanford DARE Fellowship. Discussions with
Dr. Matthew Scullin of Alphabet Energy were helpful in providing com-
mercialization and industry perspectives.
References
[1] R.D. Abelson, Space missions and applications, in: D.M. Rowe (Ed.), Thermoelec-
trics Handbook Macro to Nano, CRC Press, 2006, pp. 56-1–56-26.
[2] J.-P. Fleurial, T. Caillat, B.J. Nesmith, R.C. Ewell, D.F. Woerner, G.C. Carr, et al., Ther-
moelectrics: from space power systems to terrestrial waste heat recovery applica-
tions, 2nd Thermoelectrics Applications Workshop, 2011.
[3] BCS, Waste Heat Recovery: Technology and Opportunities in U.S. Industry, 2008.
[4] T. Hendricks, W.T. Choate, Engineering ScopingStudy of Thermoelectric Generator
Systems for Industrial Waste Heat Recovery, 2006.
[5] J. Fairbanks, Vehicular thermoelectrics: a new green technology, 2nd Thermoelec-
trics Applications Workshop, 2011.
[6] Gentherm, [Online]. Available: www.gentherm.com.
[7] T. Kajikawa, Thermoelectric power generation system recovering industrial
waste heat, Thermoelectrics Handbook Macro to Nano, CRC Press, 2006,
pp. 50-1–50-26.
[8] T. Kajikawa, Overview of progress in R&D for thermoelectric power generation
technologies in Japan, 3rd Thermoelectrics Applications Workshop, 2012.
[9] T. Kuroki, K. Kabeya, K. Makino, T. Kajihara, H. Kaibe, H. Hachiuma, et al., Thermo-
electric generation using waste heat in steel works, J Electron Mater 43 (6)
(Mar 2014) 2405–2410.
[10] J. LaGrandeur, D. Crane, A. Eder, Vehicle fuel economy improvement through ther-
moelectric waste heat recovery, DEER Conference, 2005.
[11] T.J. Hendricks, N.K. Karri, T.P. Hogan, C.J. Cauchy, New perspectives in thermoelec-
tric energy recovery system design optimization, J Electron Mater 42 (7)
(Feb 2013) 1725–1736.
[12] B. Alexander, A. Boscoe, M. Cabot, P. Dawsey, L.E. Barreau, R. Griffith, Alphabet En-
ergy: thermoelectrics and market entry, Calif Manage Rev 55 (1) (2012).
[13] J.R. Sootsman, D.Y.Chung, M.G. Kanatzidis, New andold concepts in thermoelectric
materials, Angew Chem Int Ed Engl 48 (46) (Jan 2009) 8616–8639.
[14] T.M. Tritt, Th ermoelectric phenomena, mate rials, and applications, Annu Rev
Mater Res 41 (1) (Aug 2011) 433–448.
[15] A. Shakouri, Recent Developments in Semiconductor Thermoelectric Physics and
Materials, Jul 2011.
[16] Thermoelectrics Handbook: Macro to Nano, CRC Press, 2006. 1014.
[17] D.T. Crane, L.E. Bell, Design to maximize performance of a thermoelectric power
generator with a dynamic thermal power source, J Energ y Resour Technol 131
(1) (Mar 2009) 012401.
[18] D.T. Crane, G.S. Jackson, Optimization of cross flow heat exchangers for thermo-
electric waste he at recovery, Energy Convers Manag 45 (9–10) (Jun 2004)
1565–1582.
[19] D.T. Crane, An introduction to system-level, steady-state and transient modeling
and optimization of high-power-density thermoelectric generator devices made
of segmented thermoelectric elements, J Electron Mater 40 (5) (Dec 20 10)
561–569.
[20] T.J. Hendricks, Thermal system interactions in optimizing advancedthermoelectric
energy recovery systems, J Energy Resour Technol 129 (3) (Sep 2007) 223.
[21] L.E. Bell, Cooling, heating, generating power, and recovering waste heat with ther-
moelectric systems, Science 321 (5895) (Sep 2008) 1457–1461.
[22] M. Zebarjadi, K. Esfarjani, M.S. Dresselhaus, Z.F. Ren, G. Chen, Perspectives on ther-
moelectrics: from fundamentals to device applications, Energy Environ Sci 5 (1)
(Jan 2012) 5147.
Table 3
Comparison of raw [89] and pure [94] material costs for select thermoelectric materials.
Representative materials from eac h main type of thermo electric material are shown.
Values are rounded to the nearest dollar. The table is adapted from [75].
Material type Material name Material composition Material cost
($/kg)
Reference Raw Pure
Chalcogenide Bi
2
Te
3
[95] 110 806
AgPb
18
SbTe
20
[96] 84 581
SiGe Si
0.8
Ge
0.2
[97] 371 7081
Clathrate Ba
8
Ga
16
Ge
28
Zn
2
[98] 615 3973
Skutterudite Yb
0.2
In
0.2
Co
4
Sb
12
[99] 24 204
Half-Heusler Ti
0.8
Hf
0.2
NiSn [100] 11 530
Silicide Mg
2
Si
0.85
Bi
0.15
[101] 7 191
Oxide (Zn
0.98
Al
0.02
)O [102] 250
Fig. 9. The cost of heat exchangers related to their overall heat transfer coefficient, or U-
value [103]. The slope is the heat exchanger cost per thermal conductance.
The figure is reproduced from [58].
33S. LeBlanc / Sustainable Materials and Technologies 1–2 (2014) 26–35

[23] J.-P. Fleurial, Thermoelectric power generation materials: technology and applica-
tion opportunities, JOM 61 (4) (Apr 2009) 79–85.
[24] D.M. Rowe (Ed.), Thermoelectrics and Its Energy Harvesting, CRC Press, 2012.
[25] S.W. Angrist, Direct Energy Conversion, Allyn and Bacon, 1982.
[26] W. Liu, X. Yan, G. Chen, Z. Ren, Recent advances in thermoelectric nanocomposites,
Nano Energy 1 (1) (Jan 2012) 42–56.
[27] B. Poudel, Q. Hao, Y. Ma, Y. Lan, A. Minnich, B. Yu, et al., High-thermoelectric per-
formance of nanostructured bismuth antimony telluride bulk alloys, Science 320
(5876) (May 2008) 634–638.
[28] Y. Sun,P. Sheng, C. Di, F. Jiao, W. Xu, D. Qiu, et al.,Organic thermoelectric materials
and devices based on p- and n-type poly(metal 1,1,2,2-ethenetetrathiolate)s, Adv
Mater 24 (7) (Feb 2012) 932–937.
[29] R. Yue, J. Xu, Poly(3,4-ethylenedioxythiophene) as promising organic thermoelec-
tric materials: a min i-review, Synth Met 1 62 (11–12) (Jul 2012) 912–917.
[30] Y. Du, S.Z. Shen, K. Cai, P.S. Casey, Research progress on polymer–inorganic
thermoelectric nanocomposite materials, Prog Polym Sci 37 (6) (Jun 2012)
820–841.
[31] O. Bubnova, X. Crispin, Towards polymer-based organic thermoelectric generators,
Energy Environ Sci 5 (11) (Oct 2012) 9345.
[32] H. Zhao, J. Sui, Z. Tang, Y. Lan, Q. Jie, D. Kraemer, et al., High thermoelectric perfor-
mance of MgAgSb-based materials, Nano Energy 7 (Jul 2014) 97–103.
[33] K. Biswas, J. He, I.D. Blum, C.-I. Wu, T.P. Hogan, D.N. S eidman, et al., Hi gh-
performance bulk thermoelectrics with all-scale hierarchical architectures, Nature
489 (7416) (Sep 2012) 414–418.
[34] X. Lu, D.T. Morelli, Y. Xia, F. Zhou, V. Ozolins, H. Chi, et al., High performance ther-
moelectricity in earth-abundant compounds based on natural mineral tetrahe-
drites, Adv Energy Mater 3 (3) (Mar 2013) 342–348.
[35] G. Snyder, T. Ursell, Thermoelectric efficiency and compatibility, Phys Rev Lett 91
(14) (Oct 2003) 148301.
[36] G.J. Snyder, Application of the compatibility factor to the design of segmented
and cascaded thermoelectric generators, Appl Phys Lett 84 (13) (Mar 2004)
2436.
[37] M.S. El-Genk, H.H. Saber, T. Caillat, Efficient segmented thermoelectric unicouples
for space power a pplications, En ergy Convers Manag 44 (11) (Jul 2003)
1755–1772.
[38] G.P. Meisner, Skutterudite thermoelectric generator for automotive waste heat re-
covery —Meisner.pdf, 3rd Thermoelectrics Applications Workshop, 2012.
[39] Alphabet Energy, [Online]. Available: www.alphabetenergy.com.
[40] GMZ Energy, [Online]. Available: www.gmzenergy.com.
[41] X. Shi, J. Yang, J.R. Salvador, M. Chi, J.Y. Cho, H. Wang, et al., Multiple-filled
skutterudites: high thermoelectric figure of merit through separately optimizing
electrical and th ermal transpo rts, J Am Chem Soc 1 33 (20) (May 2011 )
7837–7846.
[42] X. Lu, D.T. Morelli, Naturalmineral tetrahedrite as a direct source of thermoelectric
materials, Phys Chem Chem Phys 15 (16) (Apr 2013) 5762–5766.
[43] G. Joshi, R. He, M. Engber, G. Samsonidze, T. Pantha, E. Dahal, et al., NbFeSb-based
p-type half-Heuslers for power generation applications, Energy Environ Sci 7
(Oct, 2014) 4070–4076.
[44] J. Paik, E. Brandon, T. Caillat, R. Ewell, Life Testing of Yb
14
MnSb
11
for High Per for-
mance Thermoelectric Couples, Jet Propulsion Laboratory, National Aeronautics
and Space Administration, Pasadena, CA, 2011.
[45] V. Ravi,S. Firdosy, T. Caillat, B. Lerch,A. Calamino, R. Pawlik, et al., Mechanical prop-
erties of thermoelectric skutterudites, 969 (1) (2008) 656–662.
[46] J. Eilertsen, M.A. Subramanian, J.J. Kruzic, Fracture toughness of Co4Sb12 and In0.
1Co4Sb12 therm oelectric sku tterudites evaluated by three methods, J Allo ys
Compd 552 (Mar 2013) 492–498.
[47] F. Ren, E.D. Case, E.J. Timm, H.J. Schock, Hardness as a function of composition for
n-type LAST the rmoelectric ma terial, J Alloy s Compd 455 (1–2 ) (May 2008)
340–345.
[48] F. Ren, B.D. Hall, J.E. Ni, E.D. Case, J. Sootsman, M.G. Kanatzidis, et al., Mechanical
characterization of PbTe-based thermoelectric materials, MRS Proceedings, 1044,
2011 [pp. 1044–U04–04].
[49] J.E.Ni,E.D.Case,K.N.Khabir,R.C.Stewart,C.-I.Wu,T.P.Hogan,etal.,Roomtem-
perature Young's modulus, shear modulus, Poisson's ratio and hardness of
PbTe–PbS thermoelectric materials, Mater Sci Eng B 170 (1–3) (Jun 2010)
58–66.
[50] D. Kenfaui, D. Chateigner, M. Gomina, J.G. Noudem, Texture, mechanical and ther-
moelectric propertiesof Ca3Co4O9 ceramics, J Alloys Compd 490 (1–2) (Feb 2010)
472–479.
[51] J.G. Noudem, D. Kenfaui, D. Chateigner, M. Gomina, Granular and lamellar thermo-
electric oxides consolidated by spark plasma sintering, J Electron Mater 40 (5)
(Mar 2011) 1100–1106.
[52] K. Ueno, A. Yamamoto, T. Noguchi, T. Inoue, S. Sodeoka, H. Obara, Optimization of
hot-press conditions of Zn4Sb3 for high thermoelectric performance. II. Mechani-
cal properties, J Alloys Compd 388 (1) (Feb 2005) 118–121.
[53] J.E. Ni, E.D. Case, R.D. Schmidt, C.-I. Wu, T.P. Hogan, R.M. Trejo, et al., The thermal
expansion coefficient as a key design parameter for thermoele ctric materials
and its relation ship to process ing-dependent bloating, J Mater Sci 48 (18)
(May 2013) 6233–6244.
[54] D.T. Crane, L.E. Bell, Progress towards maximizing the performance of a thermo-
electric power generator, 2006 25th Int. Conf. Thermoelectr, 2006, pp. 11–16.
[55] Y. Gao, A.M. Marconnet, M.A. Panzer, S. LeBlanc, S. Dogbe, Y. Ezzahri, et al., Nano-
structured inte rfaces for therm oelectrics, J El ectron Mater 39 (9 ) (Jun 2010)
1456–1462.
[56] P.M. Mayer, R.J. Ram, Optimization of heat sink-limited thermoelectric generators,
Nanoscale Microscale Thermophys Eng 10 (2) (Jul. 2006) 143–155.
[57] K. Yazawa, A. Shako uri, Optimization of power and efficiency of thermoelectric
devices with asymmetric thermal contacts, J Appl Phys 111 (2) (Jan 2012)
024509.
[58] S.K. Yee, S. LeBlanc, K.E. Goodson, C. Dames, $ per W metrics for thermoelectric
power generation: beyond ZT, Energy Environ Sci 6 (2013) 2561–2571.
[59] Y.C. Lan, D.Z. Wang, G. Chen, Z.F. Ren, Diffusion of nickel and tin in p-type (Bi, Sb)
[sub 2]Te[sub 3] and n-type Bi[sub2](Te, Se)[sub 3] thermoelectric materials, Appl
Phys Lett 92 (10) (Mar 2008) 101910.
[60] T.Y. Lin, C.N. Liao, A.T.Wu, Evaluation of diffusion barrier between lead-free solder
systemsand thermoelectricmaterials, J Electron Mater 41 (1)(Sep. 2011) 153–158.
[61] N.-H.Bae, S. Han, K.E. Lee, B. Kim, S.-T. Kim,Diffusion at interfaces of micro thermo-
electric devices, Curr Appl Phys 11 (5) (Sep 2011) S40–S44.
[62] T. Kacsich, E. Kolawa, J.P. Fleurial, T. Caillat,M.-A. Nicolet, Films ofNi-7 at.% V, Pd, Pt
and Ta–Si–N as diffusion barriers for copper on Bi
2
Te
3
, J Phys D Appl Phys 31 (19)
(Oct 1998) 2406–2411.
[63] D. Crane, Gentherm, Thermoelectric waste heat recovery program for passenger
vehicles, U.S. Department of Energy Annual Merit Review, 2013.
[64] A. Schmitz, C. Stiewe, E. Müller, Preparation of Ring-shaped thermoelectric legs
from PbTe powders for tubular thermoelectric modules, J Electron Mater 42 (7)
(Jan 2013) 1702–1706.
[65] Laird, [Online]. Available: http://www.lairdtech.com/Products/Thermal-Manage-
ment-Solutions/Thermal-Interface-Materials/#.U867b7Ecs6U.
[66] The Bergquist Company, [Online]. Available: http://www.bergquistcompany.com/
thermal_materials/index.htm.
[67] Dow Corning, [Online]. Available: http://www.dowcorning.com/content/etronics/
etronicswet/newtim_tutorial1.asp?DCWS=Electronics&DCWSS=ThermalInterface-
WetDispensed.
[68] K.M.F. Shahil, A.A. Balandin, Graphene–multilayer graphene nanocomposites
as highly efficient thermal interface materials, Nano Lett 12 (2) (Feb 2012)
861–867.
[69] P. Kumar, S. Awasthi, Mechanical and thermal modeling of In–Cu composites for
thermal interface ma terials applicati ons, J Compos Mater 48 (11 ) (May 2013)
1391–1398.
[70] Y. Gao, A.M. Marconnet, R. Xiang, S. Maruyama, K.E. Goodson, Heat capacity, ther-
mal conductivity, and interface resistance extraction for single-walled carbon
nanotube films using frequency–domain thermoreflectance, IEEE Trans Compon
Packag Manuf Technol 3 (9) (Sep. 2013) 1524–1532.
[71] R. McCarty, Thermoelectric power generator design for maximum power: it's all
about ZT, J Electron Mater 42 (7) (Oct 2012) 1504–1508.
[72] M. Gomez, R. Reid, B. Ohara, H. Lee, Influence of electrical current variance and
thermal resistances on optimum working conditions and geometry for thermo-
electric energy harvesting, J Appl Phys 113 (17) (May 2013) 174908.
[73] T.J. Hendricks, Integrated thermoelectric–thermal system resistance optimization
to maximize power output in thermoelectric energy recovery systems, Materials
Research Society Symposium Proceedings, 2014.
[74] G. Min, D.M. Rowe, Design theory of thermoelectric modules for electrical power
generation, IEE Proc Sci Meas Technol 143 (6) (Nov 1996) 351–356.
[75] S. LeBlanc, S.K. Yee,M.L. Scullin, C. Dames, K.E. Goodson, Material andmanufactur-
ing cost considerations for thermoelectrics, Renew Sustain Energy Rev 32 (2014)
313–327.
[76] S. LeBlanc, Electrothermal Properties of Nanowire Materials for Energy Conversion
Systems, Stanford University, 2012.
[77] M.A. Karri, E.F. Thacher, B.T. Helenbrook, Exhaust energy conversion by thermo-
electric generator: Two case studies, Energy Convers Manag 52 (3) (Mar 2011)
1596–1611.
[78] D.M. Rowe, J. Smith, G. Thomas, G. Min, Weight penalty incurred in thermo-
electric recovery of automobile exhaust heat, J Electron Mater 40 (5) (Mar
2011) 784–788.
[79] C. Maranville, Overview of Ford's thermoelectric programs: waste heat recovery
and climate control, U.S. Department of EnergyThermoelectric Applications Work-
shop, 2011.
[80] A. International, Alloy Phase Diagram Database[Online]. Available: http://www1.
asminternational.org/AsmEnterprise/APD.
[81] Y. Pei, J.Lensch-Falk, E.S.Toberer, D.L. Medlin,G.J. Snyder, High thermoelectric per-
formance in PbTe due to large nanoscale Ag2Te precipitates and La doping, Adv
Funct Mater 21 (2) (Jan 2011) 241–249.
[82] J.D. Sugar, D .L. Medlin, Precipitation of A g2Te in the thermoelectric mate rial
AgSbTe2, J Alloys Compd 478 (1–2) (Jun 2009) 75–82.
[83] C.J. Vineis, A. Shakouri, A. Majumdar, M.G. Kanatzidis, Nanostructured thermoelec-
trics: big efficiency gains from smal l features, Adv Ma ter 22 (36) (Sep 20 10)
3970–3980.
[84] Netzsch, [Online]. Available: http://www.netzsch-thermal-analysis.com.
[85] ULVAC, [Online]. Available: http://www.ulvac.com.
[86] J. Martin, Apparatus for the high temperature measurement of the Seebeck coeffi-
cient in thermoelectric materials, Rev Sci Instrum 83 (6) (Jun 2012) 065101.
[87] Y.G. Yan, J. Martin, W. Wong-Ng, M. Green, X.F. Tang, A temperature dependent
screeningtool for high throughput thermoelectric characterizationof combinatori-
al films, Rev Sci Instrum 84 (11) (Nov 2013) 115110.
[88] G.G. Yadav, J.A. Susoreny, G. Zhang, H. Yang, Y. Wu, Nanostructure-based thermo-
electric conversion: an insight into the feasibility and sustainability for large-scale
deployment, Nanoscale 3 (9) (Sep 2011) 3555–3562.
[89] U.S. Geological Survey Minerals Information: Commodity Statistics and Informa-
tion[Online]. Availabl e: http://min erals.usgs.gov/minerals/pubs/commodity/
[Accessed: 19-Sep-2014].
[90] A. Miner,The industrialization of thermoelectric power generation technology, U.S.
Department of Energy Thermoelectric Applications Workshop, 2012.
34 S. LeBlanc / Sustainable Materials and Technologies 1–2 (2014) 26–35

[91] J.D. Bryan, N.P. Blake, H. Metiu, G.D. Stucky, B.B. Iversen, R.D. Poulse n, et al.,
Nonstoichiometry and chemical purity effects in thermoelectric Ba[sub 8]Ga[sub
16]Ge[sub 30] clathrate, J Appl Phys 92 (12) (Nov 2002) 7281.
[92] X.Y. Li, L.D. Chen, J.F. Fan, W.B. Zhang, T. Kawahara, T. Hirai,Thermoelectric proper-
ties of Te-doped CoSb[sub 3] by spark plasma sintering, J Appl Phys 98 (8)
(Oct 2005) 083702.
[93] R.a. Horne, Effect of oxide impurities on the thermoelectric powers and electrical
resistivities of bismuth, antimony, tellurium, and bismuth–tellurium alloys,
J Appl Phys 30 (3) (1959) 1959.
[94] Sigma-Aldrich, [Online]. Available: www.sigmaaldrich.com.
[95] N.D. Lowhorn, W. Wong-Ng, Z.-Q. Lu, J. Martin, M.L. Green, J.E. Bonevich, et al., De-
velopment of a Seebeck coefficient Standard Reference Material™,JMaterRes26
(15) (Jul 2011) 1983–1992.
[96] K.F. Hsu, S. Loo,F. Guo, W. Chen, J.S. Dyck, C. Uher, et al., Cubic AgPb(m)SbTe(2 + m):
bulk thermoelectric materials with high figure of merit, Science 303 (5659)
(Feb 2004) 818–821.
[97] M.S. El-Genk, H.H. S aber, T. Caillat, A performance comparison of SiGe and
skutterudite based segmented thermoelectric devices, AIP Conference Proceed-
ings, vol. 608, no. 1, 2002, pp. 1007–1015.
[98] D. Cederkrantz, M. Nygren, A.E.C. Palmqvist, Thermoelectric properties of partly
Sb- and Zn-substi tuted Ba[sub 8]Ga[sub 16]Ge[sub 30] clathrates, J Appl Phys
108 (11) (2010) 113711.
[99] J. Peng, J. He, P.N. Alboni, T.M. Tritt, Synthesis and thermoelectric properties of the
double-filled skutterudite Yb0.2In y Co4Sb12, J Electron Mater 38 (7) (Jan 2009)
981–984.
[100] T. Katayama, S.W. Kim, Y. Kimura, Y. Mishima, The effects of quaternary additions
on thermoelectric properties of TiNiSn-based half-Heusler alloys, J Electron Mater
32 (11) (Nov 2003) 1160–1165.
[101] S.K. Bux, M.T. Yeung, E.S. Toberer, G.J. Snyder, R.B. Kaner, J.-P. Fleurial, Mechano-
chemical synthesis and thermoelectric properties of high quality magnesium sili-
cide, J Mater Chem 21 (33) (Aug 2011) 12259.
[102] T. Tsubota, M. Ohtaki,K. Eguchi, H. Arai, Thermoelectric properties of Al-dopedZnO
as a promising oxide material for high-temperature thermoelectric conversion,
J Mater Chem 7 (1) (Jan 1997) 85–90.
[103] R.K.Shah, D.P. Sekulic, Fundamentalsof Heat Exchanger Design, John Wiley& Sons,
2003. 941.
[104] K. Yazawa, A. Shakouri, Cost–efficiency trade-off and the design of thermoelectric
power generators, Environ Sci Technol 45 (17) (Sep 2011) 7548–7553.
35S. LeBlanc / Sustainable Materials and Technologies 1–2 (2014) 26–35