Sie sind auf Seite 1von 9

Thermoelectric Energy Conversion Devices$

J Sharp, Marlow Industries, Inc., Dallas, TX, USA


r 2016 Elsevier Inc. All rights reserved.

1
Introduction
2
Fundamental Aspects and Analysis of Operation
2.1
Fundamental Aspects
2.2
Analysis
2.2.1
Generation
2.2.2
Cooling
2.2.3
Multiple stages
3
Thermoelectric Materials
3.1
Nominal Properties
3.2
Industry Materials
3.2.1
Bi2Te3 alloys
3.2.2
PbTe alloys
3.3
Other Materials
4
Device Construction and Economic Aspects
4.1
Coolers
4.2.
Generators
Acknowledgments
References

Nomenclature
S

q
T
DT
R
K
Z

Seebeck coefcient (mV K1)


thermal conductivity (W m1 K1)
electrical resistivity (mO-cm)
temperature (K, 1C)
temperature difference (K, 1C)
electrical resistance (O, mO)
thermal conductance (W K1)
gure of merit (K1)

1
1
2
2
2
3
4
5
5
5
5
5
6
6
6
7
8
8

ZT
I
V
Q
W
/, U
USD

dimensionless gure of merit (dimensionless)


electric current (A)
voltage (V)
rate of heat ow (W)
power supplied or produced (W)
coefcient of performance (none)
efciency (none)
cost (U.S. dollars)

Introduction

There are a number of ways to convert between thermal and electrical energy, and more than one of these have sometimes been
referred to as thermoelectric. Here, thermoelectric refers only to direct conversion between thermal and electrical energy by the Peltier
and Seebeck effects. An account of the physics of these related transport phenomena can be found elsewhere (Mahan, 2005). This
review discusses basic engineering and materials aspects of thermoelectric devices, without proceeding to system issues such as
heat exchange and feedback loops for precise temperature control. The discussion is limited to cooling/heating and electrical
power generation, setting aside aspects particular to thermoelectric sensors.
This article is composed of three main sections. The rst section covers fundamental aspects and analysis of device operation.
The second section discusses bulk (i.e., excluding thin lm) thermoelectric materials briey. The third section covers construction
and high-level economic aspects of bulk thermoelectric devices.

Change History: April 2015. J. Sharp increased the article from two main sections to three main sections, updated all subsections, reduced discussion of TE
materials, replaced original Figure 5, and replaced references with citations.

Reference Module in Materials Science and Materials Engineering

doi:10.1016/B978-0-12-803581-8.01093-6

Thermoelectric Energy Conversion Devices

Fundamental Aspects and Analysis of Operation

2.1

Fundamental Aspects

The fundamental aspects of a thermoelectric (TE) device can be understood by the example of idealized couples, as illustrated in
Figure 1 (Nolas et al., 2001). Two bars of semiconductor, one n-type (n) and the other p-type (p), are joined by a metal
interconnect at one end and connected in series to a voltage source or load at the other end. The two bars (a.k.a. elements,
legs, or pillars) share a common temperature at each end, and thus are thermally in parallel across a temperature difference
DT THTC.
TC, QC

TH, QH

p
+++

TH, QH

TC, QC
I

RLoad

Figure 1 Schematic illustrations of thermoelectric (TE) couples in use for cooling (left) and power generation (right). QH and QC differ by the
power delivered by the voltage source or to the load and depend on TE material properties, hot and cold temperatures, TE bar dimensions, and the
applied voltage or load.

Consider rst the cooling mode. When a current (I) is established, the metal/semiconductor junctions at one end of the couple
are reverse-biased, and the junctions at the other end are forward-biased. These are the cooling and heating junctions respectively.
At the cooling junctions, thermal energy is absorbed from the lattice to account for the mismatch in energy ux (relative to the
ground state) of the electric current in the metal compared to the semiconductors. If the two semiconductors were joined directly,
heat would be carried away from the reverse-biased junction in the extraction of electronhole pairs. At the heating junctions, the
energy mismatch is in the opposite sense, and heat is transferred from the charge carriers to the lattice.
In the power generation mode, the applied temperature gradient is opposed by a concentration gradient of the mobile carriers. This
is the Seebeck effect, and the Seebeck voltage places the hot junctions in reverse-bias and the cold junctions in forward bias. By
convention, n and p materials have negative and positive coefcients respectively, and the Seebeck coefcient of the couple is S SpSn.
The thermoelectric effect is a reversible thermodynamic process. In the absence of irreversible processes, which are ignored for
the moment, the efciency of thermoelectric energy conversion should be the Carnot limit. In a cooling device, the Peltier cooling
power is STCI. If no DT is allowed to develop, then this heat is transported without energy cost, and the coefcient of performance
is innite. With a temperature gradient, the voltage on the couple is SDT, and the ratio of heat pumped to electrical power required
is TC/DT, the Carnot limit for a refrigerator. In a power generator, the thermal voltage is SDT. If a current is owing, the heat
absorbed by the electric current at the heated junctions is STHI, and the efciency is DT/TH, again the Carnot limit. In reality,
thermoelectric devices perform much below the Carnot limits, primarily because both electrical resistance and thermal conduction
are signicant within the thermoelectric materials.

2.2

Analysis

2.2.1

Generation

Referring again to Figure 1, the series electrical and parallel thermal arrangement leads to R Rn Rp and K Kn Kp for the couple.
Relative to the ideal situation, the power delivered to the load is reduced by the internal resistance of the couple:
P IV ISDT  IR

1

The heat absorbed from the source to perform this work is increased by thermal conduction, but decreased by resistive heating:

QH SITH KDT  I2 R=2

2

Thermoelectric Energy Conversion Devices

The efciency is Z P/QH. From eqn [1], for xed S, DT, and R, the power output is maximized for I SDT/2R, which occurs
when RLoad R. The maximum power output has the value
Pmax KZDT 2 =4

3

where Z S2/RK is the thermoelectric gure of merit of the couple, with units of reciprocal temperature.
In general, Z for a couple is an extrinsic quantity because the geometry factors in R and K do not cancel. In the case that the n
and p bars share a common length-to-area ratio, Z is an intrinsic quantity that depends only on the Seebeck coefcient, electrical
resistivity (r) and thermal conductivity () of the n and p materials. Relative to one another, the length-to-area ratios of the n and p
bars can be optimized to give the best Z for the couple:
hp pi2
Zopt S2 = m rn p rp

4

The maximum efciency of a thermoelectric generator occurs when RLoad gR and is


Zmax DT g  1=TH g TC =TH ;
g

p
1 Z TC TH =2

with
5

For both power generation and cooling, Zopt determines the maximum performance that is possible for a given pair of
materials. This is consistent with the fact that Z is maximized by maximizing the thermoelectric effect while minimizing the
irreversible processes involved.
ZT varies with temperature. For power generation, DT can be so large that the couples gain performance by the use of two or
more materials in the n bar, the p bar, or both. In these segmented thermocouples, the segment materials are best chosen not just
for their individual ZT values, but also for their compatibility with regard to electrical and thermal uxes (Snyder and Ursell,
2003).
The most developed generation materials have peak ZT B1 and average ZT B0.7 for a device with a cold side near 100 1C and
DT B400 1C (which requires a system DT larger than 400 1C often signicantly larger) (Snyder, 2006). From eqn [5], the
optimum, ideal efciency for this situation is in the vicinity of 8.7%. With segmentation, if the average ZT is increased to 0.8, then
the ideal efciency for the same temperature span is B9.7%. For higher average ZT values and/or larger temperature differences,
efciencies exceeding 10% are achieved (Snyder, 2006).
The maximum power output per couple is an extrinsic quantity (eqn [3]), a system design parameter that can vary greatly. For
TE devices built into gas-to-air heat exchangers, thermal uxes can be an order of magnitude lower than for devices built into
liquid-to-liquid heat exchangers. Recent papers have addressed extrinsic (and therefore economic) aspects of TE power generation
(Yee et al., 2013; LeBlanc et al., 2014).

2.2.2

Cooling

In a real thermoelectric device, Peltier cooling is offset in part by resistive heating and thermal conduction. The net rate of
cooling is
QC STC I  I2 R=2  KDT

6

W IV IIR SDT

7

f QC =W

8

The work required to perform this cooling is

and the coefcient of performance is

Both QC and f can be maximized by varying the current, as shown in Figure 2. These current levels, which depend on material
properties and bar dimensions, are
IQ STC =R

9

If SDT=g  1R

10

and

Thermoelectric Energy Conversion Devices

Heat pumped (mW), COP (%)

70
60

Heat pumped

50
40

COP

30
20
10
0
0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4

Current (Amps)
Figure 2 Dependence of ideal heat pumping rate and ideal coefcient of performance on supplied current for a single couple with properties and
dimensions typical for temperature control of telecommunications lasers. The couple has S 420 mV K1, R 56 mO, and K103 W K1. The
length-to-area ratio for each leg (n and p) is 28 cm1. The hot side is taken as 300 K, and the DT is 40 K.

Use of eqn [9] in eqn [6] leads to




Qmax K ZTC2 =2  DT

11

Similar to power generation, the maximum heat pumping capacity of a couple is an extrinsic quantity that increases with the
area-to-length ratio of the thermoelectric bars. Power per area of thermoelectric material, Qmax/(An Ap), varies inversely with bar
length.
When DT Z(TC)2/2, no heat can be pumped. This is the largest DT that can be achieved with a single stage thermoelectric
refrigerator. For the best commercial materials operating under favorable conditions (no moisture, good heat sinking, etc.), the
maximum DT is 7075 1C for a hot-side temperature of 300 K.
For values of DT below the maximum, heat can be absorbed at the cold side, and the maximum coefcient of performance
follows from eqns [8] and [10]:
fmax gTC  TH =g 1DT

12

The curves in Figure 2 are plots of the current dependence of QC and f with TC 260 K, TH 300 K, g 1.37 and K 103 W K1.
This value of K is typical for a couple in a small cooling device. If such a device is operated near peak f, then approximately 25
couples are needed to pump one watt of thermal load.

2.2.3

Multiple stages

In cooling applications, it is common to stack multiple TE devices to achieve greater temperature differences than possible with a
single stage. Devices with six stages are able to reach temperatures as low as 170 K from 300 K, although with a quite low
coefcient of performance.
With reference to Figure 3, the efciency of a two-stage device is
F

QC

W1 W2

QC
QC
f1

C =f1
QC Q
f

1
f1

1
f1 f 1f
2

13
2

The generalization for m stages, each with the same coefcient of performance, is

F 

1
m

1
f

1

14

Thermoelectric Energy Conversion Devices

QC (cooling load)
Stage 2
QC + W2
Stage 1
QC + W2 + W1
to heat sink

Figure 3 Schematic illustration of a two-stage thermoelectric cooling device, showing the increase of total heat transfer from the cooled load to
the heat sink.

Thermoelectric Materials

3.1

Nominal Properties

Good thermoelectric materials are heavily doped semiconductors with a Seebeck coefcient in the range 150250 mV K1.
Empirically, the electrical resistivity tends to be in the range 0.53 mO cm at suitable values of the carrier concentration. Most
semiconductors do not obtain such favorable values of S and r simultaneously; high values of the power factor, S2/r, are part of
what makes good thermoelectric materials uncommon.
Even more so, it is the lattice thermal conductivity that sets thermoelectric materials apart. Good thermoelectric materials are
crystals, and yet their lattice thermal conductivity often is similar to that of amorphous silica (B1.5 W m1 K1) or, in some cases,
even as low as B0.5 W m1 K1. With the charge carriers typically contributing another 0.21.0 W m1 K1, the total thermal
conductivity is in the range 0.74 W m1 K1in most cases.
The synthesis and characterization of semiconducting compounds with high ZT values is a challenge that has attracted the
attention of researchers since the demonstration of useful capabilities during a span of several years from the late 1940s to the mid1950s (Ioffe, 1957; Goldsmid and Douglas, 1954).
In many cases, materials with good Z values are mixtures of two or more compounds that share a crystal structure. The reason
for this is the reduction of the lattice thermal conductivity by mass and elastic constant uctuations. In the discussion that follows,
such mixtures are referred to as alloys.

3.2
3.2.1

Industry Materials
Bi2Te3 alloys

Between 200 and approximately 450 K, the best n and p materials are mixtures of Bi2Te3, Sb2Te3, and Bi2Se3. These alloys support
all TE cooling/heating applications and also power generation from low-grade heat. The n material is approximately 90% Bi2Te3,
510% Bi2Se3, and 05% Sb2Te3. The p material is approximately 7580% Sb2Te3, 2025% Bi2Te3, and 05% Bi2Se3. The melting
point for the entire system of alloys is in the vicinity of 600 1C.
Directional solidication, hot pressing and plastic deformation methods mainly have been used to make the Bi2Te3 alloys,
which have a hexagonal crystal structure and anisotropic properties. Directional solidication from a melt has been the most
favored with regard to production volumes, but such materials are not satisfactory for miniature modules due to lack of
mechanical robustness. To make materials with good Z values suitable for miniature modules, hot plastic deformation methods
have become prevalent (Hayashi et al., 2007).
ZT values for industrial Bi2Te3 alloys range from 0.8 to 1.0 at room temperature. Materials tailored for power generation can
have peak ZT around 1.0 at 400 K. ZT falls above 400 K due to bipolar effects, and is less than 0.5 at 500 K. The alloys also can be
tailored for operation well below room temperature, and such variations are employed in multi-stage modules.

3.2.2

PbTe alloys

PbTe and Pb1xySnxMnyTe are industrial n and p materials respectively for operation up to about 838 K, showing a combination
of good performance, durability, and cost (LeSage et al., 1998). When protected from corrosion, the operating temperature of PbTe
alloys is limited by sublimation of Te.
PbTe and its alloys form in the cubic Halite (rock-salt) crystal structure. Since the structure and properties are isotropic, hot
pressing is a suitable manufacturing method. With melting points around 1200 K, these alloys also are compatible with melt
growth using fused silica containers, which can be a cost-effective method (Fano, 1995).
PbTe alloys can have ZT values around 1.0 at the typical hot side temperature of 400500 1C. Such alloys have ZT values in the
range of 0.20.3 at room temperature, and the ZT crossover point in comparison to industrial Bi2Te3 alloys is in the range 150
200 1C.

Thermoelectric Energy Conversion Devices

3.3

Other Materials

While Bi2Te3 and PbTe alloys have been the only industrial materials to-date, two other materials systems have been used (along
with PbTe alloys) in radioisotope thermoelectric generators for space exploration. These are Si1xGex (n-type and p-type) and
(GeTe)1y(AgSbTe2)y, a p-type material known as TAGS (Bennett, 1995).
Materials that have been investigated and shown some promise for use in thermoelectric devices largely can be grouped by their
anions: oxides, chalcogenides, pnictides, and tetrides (Zheng, 2008; Han et al., 2014). The oxides are mainly perovskites and
layered cobalt-oxides with one or two other metallic constituents occupying the spaces between layers (Walia et al., 2013; Zhan
et al., 2014). Numerous studies have been conducted on two categories of chalcogenides, those based on the familiar Bi2Te3 and
PbTe crystal structures, but with modications of the microstructure (Jariwala and Ravindra, 2014; Jood and Ohta, 2015; Eibl et al.,
2015), and those with other (and often complex) crystal structures (Wang and DiSalvo, 2001; McGuire et al., 2005; Gonalves and
Godart, 2014). The most studied pnictides are skutterudite alloys (Kleinke, 2010) and some half-Heusler compounds (Chen and
Ren, 2013). The tetrides that have been studied mainly are clathrates (Kleinke, 2010; Takabatake et al., 2014), anti-uorite
compounds (Fedorov 2009; Luo et al., 2014), Sn-based half-Heusler compounds (Chen and Ren, 2013), and some transition
metal-silicides (Fedorov, et al., 2009). In addition to these groups of semiconductors, researchers also have explored certain heavy
fermion compounds that possess moderate Seebeck coefcients despite being metals, and such materials have been imitated in
model meta-materials (Cai, 2008).
As the research aimed at higher ZT materials pushes forward, there are parallel efforts to understand the reproducibility of
reported ZT values. Many TE materials laboratories use equipment and methods that are thought to overestimate the Seebeck
coefcient by B10% (Martin et al., 2015), while there are signicant random errors in electrical resistivity and specic heat (Wang
et al., 2015). Awareness of the signicant uncertainties in reported ZT values of materials being developed can be useful when
comparing newer materials with the current industrial materials.

4
4.1

Device Construction and Economic Aspects


Coolers

Figure 4 is a side view schematic of a typical small thermoelectric cooler, with 23 couples connected in series. The thermoelectric
bars are soldered to metal connectors, copper pads that are directly bonded to ceramic plates and which contribute a small
(detrimental) fraction of the devices internal electrical resistance. The TE materials commonly occupy approximately 30% of the
module area, a ratio known as the packing fraction or ll fraction.' Though Bi2Te3 alloys have been implemented in other ways
(Stockholm, 1995), the standard module construction illustrated in Figure 4 dominates the industry in module sizes ranging from
B4 to B2500 mm2, and this has been the case for decades.

Figure 4 Side view of a commercial cooling module that contains 23 couples similar to the model couple of Figure 2. The device is
approximately 6  8  1.7 mm3. Solder has been applied to thick-lm copper on the top surface. Photograph courtesy of Marlow Industries, Inc.

The standard construction has been widely adopted for reasons that are straightforward. The ceramic plates provide mechanical
structure, low thermal resistance, and at and smooth surfaces for interfacing with heat transfer plates. The ceramic plate that is
heat sunk also accommodates the electrical leads. The copper connectors can be screen-printed (typical thickness of 50 mm) or can
be pre-fabricated pads when greater thickness is required for higher current levels. The TE materials typically can be adequately
protected from the assembly solder(s) by application of protective metal layers while the materials are in wafer form.
Some limitations of standard cooler construction have become known in the industry. Susceptibility to thermally-induced
stress can be signicant, especially at the perimeter and corners of larger devices. Two other limitations are thermal resistance of the

Thermoelectric Energy Conversion Devices

ceramic plates and electrical resistance of the connectors and interfaces; these problems become substantial for thin, high wattdensity coolers (and can be even more serious in modules based on thin-lm materials). A fourth limitation is that the assembly
solder inadvertently can wet side faces of the TE materials, leading to degradation of performance if the device is operated at a
temperature at which diffusion at the TE/solder interface is appreciable.
By use of kilogram-scale Bi2Te3 alloy ingots, alumina substrates with thick-lm copper or copper tabs, nickel diffusion barriers,
common solders, and rapid assembly methods with high yields, the TE cooling industry has achieved device cost-performance
levels that enable a wide variety of applications. Standard 40  40 mm2 modules that can manage 20 W loads while operating near
peak efciency are available for less than 10 USD (Sharp and Bierschenk, 2015).

4.2.

Generators

Figure 1 emphasizes the reversibility of thermoelectric energy conversion. In principle, the same device can be used for both
cooling and power generation. In reality, though, typical cooling devices might be used for generation only if the temperature
difference is relatively small and the heat source is at a modest temperature (o150 1C).
Certain variations in construction have been used to extend the operating range of Bi2Te3 devices for power generation. These
include use of higher temperature solder (Sharp and Bierschenk, 2015) solder-less construction (Bass, 1999) and partitioning of
ceramics for stress relief (Takaku et al., 2005). Through these changes, the operating limit can be extended to 300 1C, above which
the ZT of the Bi2Te3 alloys is quite small.
Beyond the bismuth telluride temperature range, TE power generation systems are available commercially, but devices and
materials are scarce. This situation may soon change, though, as the construction of devices for power generation with hot side
temperatures beyond 300 1C and employing non-Bi2Te3 materials is an active area of development (Aoyama et al., 2005; Guo
et al., 2012; Garca-Caadas et al., 2013; Xia et al., 2014; Salvador et al., 2014; Bartholom et al., 2014). High temperatures can
present challenges via increased oxidation/corrosion/sublimation, decreased stability of interfaces between device components,
increased stresses intrinsic to the device, and increased stresses associated with coupling to other system components. These
challenges are being addressed with development of more stable TE materials, more compatible contact metals, encapsulation
methods, sealing methods, and features to promote mechanical compliance. In addition, diffusion soldering is of interest (Yang
et al., 2013) to counter the modest availability of cost-effective brazes to support assembly and operation in the range of 400
700 1C.
Figure 5 shows a power generation device containing skutterudite alloys as the TE materials, which are joined to direct-bond
aluminum AlN ceramics using an aluminumsilicon braze. The hot-side ceramics are piecewise to minimize the devices internal
stress. A major advantage of the module is that it was made using components and methods similar to those used for making TE
cooling devices, thus representing an early attempt to make non-bismuth telluride TE generation devices cost-effectively. Concurrent with device prototyping efforts as mentioned above and exemplied in Figure 5, non-bismuth telluride modules also are
being offered for sale in small quantities (http://thermoelectric-generator.com (2015)).

Figure 5 A prototype power generation module that contains skutterudite alloys for both n and p legs. The device has a footprint of 7.4 cm2 and
can produce 56 W of power with the hot side at 500 1C and the cold side at 100 1C. The temperature limit for operation in air is B 350 1C, due
to oxidation of the p-type material. Photograph courtesy of Marlow Industries.

If the longevity demonstrated in remote power systems (LeSage et al., 1998), is maintained in versatile devices with good costperformance levels, TE power generation could become attractive in a wider range of applications. The central role of devices in the
TE cooling industry may eventually be mimicked in TE power generation, but more may be required of the modules. For example,
in the case of power generation, devices may need to maintain compliance despite being bonded to a heat exchanger on the hot

Thermoelectric Energy Conversion Devices

side, as thermal interface resistance on the hot side can signicantly degrade device output (Salvador et al., 2014). There may also
be a preference for devices that do not require sealing at the system level, due to both cost and performance concerns.
Regarding economic value, a commonly mentioned cost-performance target is 1 USD W1 at the system level for large-scale
applications, which would of course require devices that cost less than 1 USD W1. Assuming a total device cost approximately
equal to 0.5 USD cm2, as has been achieved for cooling devices, the overall cost goal leads to seeking a device power output of
approximately 1 W cm2 (at least). The required thermal ux (B10 W cm2) is consistent with reasonable/common values for TE
thermal conductivity (2.5 W m1 K1), packing fraction (25%), leg length (2.5 mm), and device DT (400 K). This estimate
suggests one path to affordable TE power generation, a path that hinges on achieving total device cost per cm2 comparable to what
is now found in the cooling industry. Other paths to affordable TE power generation, such as direct integration of the TE legs into
heat exchangers, also may have merit. Whether or not a TE generation system includes devices, TE materials are likely to be a
signicant cost component, and substantial improvements in the ratio of average ZT to material cost could enhance the overall
economic viability.

Acknowledgments
The author thanks Marlow Industries, Inc., a subsidiary of II-VI Incorporated, for support.

References
Aoyama, I., Kaibe, H., Rauscher, L., et al., 2005. Doping effects on thermoelectric properties of higher manganese silicides (HMSs, MnSi1.74) and characterization of
thermoelectric generating module using p-type (Al, Ge, and Mo)-doped HMSs and n-type Mg2Si0.4Sn0.6 legs. Japanese Journal of Applied Physics 44, 42754281.
Bartholom, K., Balke, B., Zuckermann, D., et al., 2014. Thermoelectric modules based on half-Heusler materials produced in large quantities. Journal of Electronic Materials 43
(6), 17751781.
Bass, J.C., 1999. United States Patent 5,892,656: Thermoelectric Generator. (Filed 20.05.96; issued 06.04.99).
Bennett, G., 1995. Space applications. In: Rowe, D.M. (Ed.), CRC Handbook of Thermoelectrics, rst ed. Boca Raton: CRC Press, pp. 515537.
Cai, J., 2008. The thermoelectric properties of strongly correlated systems, Ann Arbor: ProQuest.
Chen, S., Ren, Z., 2013. Recent progress of half-Heusler for moderate temperature thermoelectric applications. Materials Today 16 (10), 387395.
Eibl, O., Nielsch, K., Peranio, N., Vlklein, F. (Eds.), 2015. Thermoelectric Bi2Te3 Nanomaterials. Weinheim: Wiley-Vch.
Fano, V., 1995. Lead telluride and its alloys. In: Rowe, D.M. (Ed.), CRC Ha$ndbook of Thermoelectrics, rst ed. Boca Raton: CRC Press, pp. 257266.
Fedorov, M.I., 2009. Thermoelectric silicides: past, present, and future. Journal of Thermoelectricity 2009 (2), 5160.
Garca-Caadas, J., Powell, A.V., Kaltzoglou, A., Vaqueiro, P., Min, G., 2013. Fabrication and evaluation of a skutterudite-based thermoelectric module for high-temperature
applications. Journal of Electronic Materials 42 (7), 13691374.
Goldsmid, H.J., Douglas, R.W., 1954. The use of semiconductors in thermoelectric refrigeration. British Journal of Applied Physics 5, 386.
Gonalves, A.P., Godart, C., 2014. New promising bulk thermoelectrics: intermetallics, pnictides, and chalcogenides. European Physical Journal B 87, 42.
Guo, J.Q., Geng, H.Y., Ochi, T., et al., 2012. Development of skutterudite thermoelectric materials and modules. Journal of Electronic Materials 41 (6), 10361042.
Han, C., Li, Z., Dou, S., 2014. Recent progress in thermoelectric materials. Chinese Science Bulletin 59 (18), 20732091.
Hayashi, T., Sekine, M., Suzuki, J., Horio, Y., Takizawa, H., 2007. Thermoelectric and mechanical properties of angular extruded Bi0.4Sb1.6Te3 compounds. Materials
Transactions 48 (10), 27242728.
Ioffe, A., 1957. Semiconductor Thermoelements and Thermoelectric Cooling. London: Infosearch.
Jariwala, B., Ravindra, N.M., 2014. Process, property and performance of chalcogenide-based thermoelectric materials. Nanomaterials and Energy 3 (3), 6881.
Jood, P., Ohta, M., 2015. Hierarchical architecturing for layered thermoelectric suldes and chalcogenides. Materials 8 (3), 11241149.
Kleinke, H., 2010. New bulk materials for thermoelectric power generation: clathrates and complex antimonides. Chemistry of Materials 22 (3), 604611.
LeBlanc, S., Yee, S.K., Scullin, M., Dames, C., Goodson, K.E., 2014. Material and manufacturing cost considerations for thermoelectrics. Renewable and Sustainable Energy
Reviews 32, 313327.
LeSage, B.C., Meszaros, G., Nystrom, T., 1998. On the suitability of low cost PbTe devices for high grade waste heat power generation applications. In: Proceedings of the
XVIIth International Conference on Thermoelectrics, 1998, ICT 98. IEEE, Piscataway, pp. 426428.
Luo, X.D., Liu, H., Xu, W.L., Zhu, Y.X., 2014. Progress on Mg2Si thermoelectric materials. Advanced Materials Research 886, 7174.
Mahan, G.D., 2005. Thermoelectric effect. In: Bassani, F., Liedl, G.L., Wyder, P. (Eds.), Encyclopedia of Condensed Matter Physics, rst ed. Oxford: Elsevier, pp. 180187.
http://dx.doi.org/10.1016/B0-12-369401-9/00726-9.
Martin, J., Wong-Ng, W., Green, M.L., 2015. Seebeck coefcient metrology: Do contemporary protocols measure up? Journal of Electronic Materials 44 (6), 19982006.
doi: 10.1007/s11664-015-3640-9.
McGuire, M.A., Reynolds, T.K., DiSalvo, F.J., 2005. Exploring thallium compounds as thermoelectric materials: seventeen new thallium chalcogenides. Chemistry of Materials
17, 28752884.
Nolas, G.S., Sharp, J.W., Goldsmid, H.J., 2001. Thermoelectrics: Basic Principles and New Materials Developments. Berlin: Springer.
Salvador, J.R., Cho, J.Y., Ye, Z., et al., 2014. Conversion efciency of skutterudite-based thermoelectric modules. Physical Chemistry Chemical Physics 16 (24), 1251012520.
Snyder, G.J., 2006. Thermoelectric power generation: efciency and compatibility. In: Rowe, D.M. (Ed.), Thermoelectrics Handbook: Macro to Nano, 1st ed. Boca Raton: : CRC
Press, pp. 9-19-26.
Sharp, J., Bierschenk, J., 2015. The prevalence of standard large modules in thermoelectric applications. Journal of Electronic Materials 44 (6), 17631767.
Snyder, G.J., Ursell, T.S., 2003. Thermoelectric efciency and compatibility. Physical Review Letters 91, 148301.
Stockholm, J., 1995. Large-scale cooling: integrated thermoelectric element technology. In: Rowe, D.M. (Ed.), CRC Handbook of Thermoelectrics, rst ed. Boca Raton: CRC
Press, pp. 657665.
Takabatake, T., Suekuni, K., Nakayama, T., Kaneshita, E., 2014. Phonon-glass electron-crystal thermoelectric clathrates: experiments and theory. Reviews of Modern Physics 86,
669. (erratum Reviews of Modern Physics 86, 841).
Takaku, R., Hino, T., Shindo, T., Itoh, Y., 2005. Development of skeleton-type thermoelectric module for power generation. In: Proceedings of 23rd International Conference on
Thermoelectrics, IEEE, Piscataway.

Thermoelectric Energy Conversion Devices


Walia, S., Balendhran, S., Nili, H., et al., 2013. Transition metal oxides thermoelectric properties. Progress in Materials Science 58 (8), 14431489.
Wang, H., Bai, S., Chen, L., et al., 2015. International round-robin study on thermoelectric transport properties of n-type half-Heusler from 300 K to 773 K. Journal of
Electronic Materials. doi:10.1007/s11664-015-4006-z.
Wang, Y.C., DiSalvo, F.J., 2001. Exploring complex chalcogenides for thermoelectric applications. In: Proceedings of the Materials Research Society Symposium Proceedings,
vol. 626, MRS, , pp. Z7.6/1Z7.6/6, Pittsburgh.
Xia, H., Drymiotis, F., Chen, C.-L., Wu, A., Snyder, G.J., 2014. Bonding and interfacial reaction between Ni foil and n-type PbTe thermoelectric materials for thermoelectric
module applications. Journal of Materials Science 49 (4), 17161723.
Yang, C.L., Lai, H.J., Hwang, J.D., Chuang, T.H., 2013. Diffusion soldering of Pb-doped GeTe thermoelectric modules with Cu electrodes using a thinf-lm Sn interlayer.
Journal of Electronic Materials 42 (3), 359365.
Yee, S.K., LeBlanc, S., Goodson, K.E., Dames, C., 2013. $ per W metrics for thermoelectric power generation: beyond ZT. Energy & Environmental Science 6, 25612571.
Zhan, B., Lan, J.-L., Liu, Y.-C., et al., 2014. Research progress on oxide thermoelectric materials. Journal of Inorganic Materials 29 (3), 237244.
Zheng, J.-C., 2008. Recent advances on thermoelectric materials. Frontiers of Physics in China 3 (3), 269279.

Das könnte Ihnen auch gefallen