Renewable and Sustainable Energy Reviews 78 (2017) 1522

Contents lists available at ScienceDirect

Renewable and Sustainable Energy Reviews

journal homepage: www.elsevier.com/locate/rser

An overview of cooling of thermoelectric devices MARK

a,b, b b
Muhammad Sajid , Ibrahim Hassan , Aziz Rahman
a
School of Mechanical & Manufacturing Engineering (SMME), National University of Sciences & Technology (NUST), Islamabad, Pakistan
b
Texas A & M University at Qatar, P.O. Box 23874, Doha, Qatar

A R T I C L E I N F O A BS T RAC T

Keywords: Thermoelectric generators are an environmentally friendly source of electrical power whose applications range
Thermoelectric generator from waste heat recovery to conversion of solar energy into electricity. Low conversion eciencies however
Cooling inhibit their wide scale deployment. Research into thermoelectrics has therefore primarily focused on
Eciency improvement of material properties, leading to remarkable progress in the area which has not translated into
Thermal management
high performing thermoelectric generators. System level eciency is signicantly dependent on eective
thermal management. Heat dissipation mechanisms employed to remove waste heat from the cold side of a
thermoelectric device are reviewed in this work. The prevalent methods of cooling thermoelectric devices are
categorized and based on published experimental data, their contribution on the overall conversion eciency of
thermoelectric generators is quantied. A broad range of devices from low heat to high thermal ux have been
covered in this work and will help guide future endeavors in thermoelectric generator design and testing.

1. Introduction 2. Thermoelectric eciency

Compared with conventional electric power generators, thermo-
electric generator (TEG) oer many advantages such as reliability, no
moving parts and environmentally friendly. TEGs accomplish this by
exploiting the thermal energy of electrons (and holes) for the energy
conversion between heat and electricity. However wide scale applica-
tion of TEGs has been hindered by their low productivity, due to low
thermal-to-electrical conversion eciencies (typically < 10%) [13]
limiting their use in specialized military, medical and space elds.
Nevertheless, in applications where the thermal energy is abundant or
very low cost, as is the case with solar energy or waste heat, TEGs are
commercially viable. Two directions have been highlighted for achiev-
ing wider applications of thermoelectric devices [4], rst is to promote
the intrinsic eciencies of thermoelectric materials and the other is to
improve the way TEMs are integrated into electrical power producing
units, of which thermal management is an important part. While there
is signicant work on hot side thermal management of TEMs, focusing
on solar energy [5], automotive exhaust [6], and waste heat [7]; cold
side thermal management has received less attention. In this paper, we
review dierent methods used to dissipate heat from the cold side of
the TEM over a broad range of operating parameters and present their
performance parameters.
2.1. Theoretical eciency
The conversion eciency, TE is dependent on the temperature
dierence T and the gure of merit ZT (the eective ZT across T )
dened as ZT = (S 2T / ) and determined by the three main material
parameters: the Seebeck coecient or thermo power S (V / K ), the
electrical resistivity ( m ) and the thermal conductivity (W /(m K ))
[8]. The thermoelectric eciency is given by:
T TC 1+ZT 1
TE = H
TH 1+ZT + (TC / TH ) (1)
The term in the rst parenthesis on the right hand side is the Carnot
eciency of the thermoelectric device operating between hot and cold
temperature reservoirs of TH and TC respectively. As is evident from Eq.
(1), the conversion eciency depends on the operating temperature
and the gure of merit.
The theoretical eciency of thermoelectric power generators for
dierent ZT s are shown in Fig. 1 against the eciencies of other power
generation systems [9].
2.2. Figure of merit and eciency

During the renewed interest in thermoelectrics of the late 1900s the

Corresponding author at: School of Mechanical & Manufacturing Engineering (SMME), National University of Sciences & Technology (NUST), Islamabad, Pakistan.
E-mail addresses: m.sajid@smme.nust.edu.pk (M. Sajid), Ibrahim.hassan@qatar.tamu.edu (I. Hassan), aziz.rahman@qatar.tamu.edu (A. Rahman).

http://dx.doi.org/10.1016/j.rser.2017.04.098
Received 18 October 2015; Received in revised form 21 April 2017; Accepted 27 April 2017
1364-0321/ 2017 Elsevier Ltd. All rights reserved.
M. Sajid et al. Renewable and Sustainable Energy Reviews 78 (2017) 1522

Fig. 1. Comparison of power generation systems. From [9].

gure of merit of available materials was as low as ZT 1 [10,11]

shown in the gure in an enclosed region and these were not
competitive against conventional power generators. However because
of their noiseless operation, exibility and form factor they still found
niche applications. In search of broader applications fundamental
research in thermoelectrics has therefore focused on maximizing the
conversion eciency by developing new, high-eciency thermoelectric
materials with ZT 3 4 [12,13] which is sucient to make them
competitive against conventional power generation including geother-
mal and alkaline cells (Fig. 1). In order to accomplish this, the
challenges lie in predicting the structures of materials, their electron
and phonon band structures and transport properties, as well as in
understanding the impact of defects in the materials on transport
properties. In a recent eort, Zhang et al. [14] used complex parameter
optimization to design module structures that minimize energy losses
using Bi2Te3-based alloys and CoSb3-based skutterudites to achieve a
peak ZT 1.2 . However, their design resulted in a conversion eciency
of 12% while operating across a temperature dierence of 541 degrees Fig. 2. Ratio of thermoelectric eciency to Carnot eciency as a function of the gure of
and a cold side temperature maintained at 35 C, comparable to a merit for a range of temperature dierences.
ZT 2.0 . Similarly, Kraemer et al. [15] achieved a peak eciency of
9.6% by employing segmented thermoelectric legs, coupled with solar the eciency of Carnot cycle (a thermodynamic limit) while operating
and thermal concentration. at room temperature (30 C) and a temperature dierence of 250 C. It
Signicant boosts in the peak ZT over the benchmark value of 1 also highlights the limited performance enhancement above a tem-
have been reviewed by Z Ren [3] covering many conventional materials perature dierence of 250 C, an average increase in eciency of 0.1%
such as Bi2Te3-based alloys, PbTe, PbSe, SiGe, Mg2X (X = Si, Ge, Sn), for every 50 C temperature rise up to a maximum of 1000 C above
skutterudite, calthrate, Zintl and half-Heusler alloys. These increases room temperature. These next generation thermoelectrics (ZT 23)
have been achieved by tailoring electron transport through nanoscale have already found applications in waste heat recovery in automobiles
approach to engineering band structure and tuning phonon transport. amongst other uses [21,22]. In addition, thermoelectric are used as
Recently the ZT of Bi0.5Sb1.5Te3 (bismuth antimony telluride) was coolers to provide cold beverages, as well as in wristwatches and
increased to 1.86 0.15 at 320 K by reducing lattice thermal con- pacemakers by exploiting the very small temperature dierences within
ductivity through grain boundary and point-defect scattering targeting the body or between a body and the environment [23,24].
high-and-low frequency phonons [16]. Earlier a ZT value of 2.2 at 915
K in p-type PbTe endotaxially nanostructured with SrTe was demon- 2.3. Thermal dependence of eciency
strated by going beyond nanostructing through a panoscopic approach
to the scattering of heat-carrying phonons across integrated length While the advancement in materials development is encouraging,
scales [17]. An unprecedented ZT of 2.6 0.3 at 923 K was reported by the budding range of materials and the reports of ascending ZT s
Kanatzidis and coworkers for SnSe single crystals and attributed to its conceal the fundamental challenges of employing these materials in
ultralow lattice thermal conductivity [18]. Thus several new bulk energy conversion on an appreciable scale. Despite these promising
materials that demonstrate ZT 23 have been identied in literature results, the eciency gains at device level has yet to be demonstrated.
[13,19,20] and mechanisms for decoupling electron transport from For example the theoretical solar-to-electricity eciency of Si-Ge alloys
phonon transport in such materials through modication are under operated at a temperature of 1000 K is 12% [25] compared to
investigation for further improvement of the gure of merit. experimental ndings of eciencies <1% [26] for conversion of
To understand the contribution of ZT on the performance of the concentrated solar radiation by directly irradiated TE modules oper-
thermoelectric device, the ratio of thermoelectric eciency (Eq. (1)) ated at 900 K on the hot side. Studies reveal that with high-tempera-
and Carnot eciency are plotted against increasing ZT in Fig. 2 for ture TEMs 60% of the incident solar radiation is lost due to reradiation
temperature dierences ranging from 50 to 1000 C. The curve shows and only 20% is available for electricity conversion [27]. Another aspect
that a thermoelectric device with a ZT ~3 can provide close to 50% of is the variation in ZT with increase in temperature, which has been

16
M. Sajid et al.
Fig. 3. Temperature dependence of thermoelectric eciency with increasing hot and
cold side temperatures.
reported [28] to increase from 0.1 to 0.25 for a polycrystalline 3d-
transition metal oxide when the temperature was increased from 300 K
to 800 K.
To develop a better insight into the temperature dependence we
plot thermoelectric conversion eciency against the temperature
dierence for various ZT s (Fig. 3). For a 400% increase in hot side
temperature (from 250 C to 1000 C) there is < 5% increase in
eciency. The increase in cold side temperature has resulted in an
eciency drop by ~7 8% in comparison with Fig. 2. We isolate the
eect of increase in cold side temperature in Fig. 4. Here, the decrease
in eciency is about 3.5% for every 50 C rise in cold side temperature
with a constant hot side temperature of 1030 C and ZT ~3.
Experimental studies of the eect of incident solar radiation and cold
side temperature on eciency have found similar results [29]. As
demonstrated by He et al. [30] the electrical eciency drops from
1.625% to 1.255% when the water temperature goes from 25 C to
55 C.
These theoretical analysis shows that there is a possibility of
improving thermoelectric conversion eciency by 510% through
improved thermal management, while experimental work suggests
the prospective range to be much greater [25,26]. In another work
[21], the open circuit voltage was reported to be 15% less than
predicted values because of insucient cooling of the cold plate at
higher heat uxes.
3. Cooling of thermoelectric generators (TEG)
In its simplest form a thermoelectric power generator consists of a
Fig. 4. Eect of increasing cold side temperature on thermoelectric eciency.
17
Renewable and Sustainable Energy Reviews 78 (2017) 1522
Fig. 5. Schematic of thermoelectric generator.
receiver of thermal energy that is in thermal contact with the thermo-
electric module (TEM) and an electrically insulated cooling system as
shown in Fig. 5. Not all of the irradiated energy adsorbed by the
collector is converted to electricity and much of it ends up heating the
device and reducing the eciency, highlighting the necessity of good
thermal management of the STEG. The cooling system is necessary to
maintain the design temperature dierence across the TEM which
decreases with time due to thermal diusion.
To limit the rise in temperature of the cold side of TEM, the next
section of this paper presents the detailed review of the existing
thermoelectric cooling systems which have been employed in small to
medium scale power generation systems (Fig. 6). Apart from these
methods, heat dissipation from cold of TEM has also been attempted
using phase change materials as a heat sink [31], which have
conventionally been used to store thermal energy for the hot side of
TEMs. While the results of this work are encouraging the phenomena
requires more detailed investigation.
3.1. Air cooled
Thermoelectric power generators with low heat ux thermal
sources generally employ passive cooling and/or naturally ventilated
systems to maintain cold side temperature and have been employed in
low cost energy sources particularly waste heat and solar [3234]. A
signicant contribution in this area was by Date et al. [35] who carried
out a comparative study of active and passive air cooled TEGs with
three cooling mechanisms. They presented an analytical model to
determine the maximum theoretical heat ux capacity of thermo-
electric generator followed by a comparison of the thermal perfor-
mances of two types of commercially available TEGs with their cold
sides cooled by either bare plate, n or heat pipe heat exchanger at
wind speeds ranging from 0 m/s to 5 m/s. We summarize their ndings
Table 1.
Under natural convection cooling conditions (i.e. 0 m/s wind) the
heat pipe cooling system provided lower cold side temperature of TEG
as compared to the nned or even bare plate exchangers, however the
Fig. 6. Classication of thermoelectric cooling systems.
M. Sajid et al.
Table 1
Maximum allowable heat flux at wind speeds of 0 m/s for bare plate, 1 m/s for heat pipe and 5 m/s for nned heat exchangers (based on data from [35]).
Hot side temperature Maximum allowable heat ux (kW/m2)
Bare plate
Type A 150 C 4.0 to 16.3
Type B 250 C 8.3 to 31.1
limiting heat ux value was also lower than that achieved with the
other heat exchanger congurations. This behavior can be attributed to
the smaller n gap in heat pipe heat exchanger that restricts convection
heat transfer which is the dominant mode of heat transfer at lower
temperatures.
3.1.1. Passive cooling
The principles of dry draft cooling were employed by zdemir et al.
[32] to build and test a prototype solar TEG with an isolated wind
chimney for cooling. They presented performance data for the device
over an eight hour daytime cycle for two days in the month of august.
While the ambient temperature in the chimney varies from 30 C to
40 C the cold side temperature of TEG sees an increase from 30 C to a
maximum of 77 C before dropping later in the day. With an eective
ZT <0.2 the eciencies are quite low. In Fig. 7 Ratio of thermal
eciency to Carnot eciency for a prototype wind chimney TEG with
naturally varying & xed cold side temperature Fig. 7 we take the
experimentally measured hot side temperatures and compare the
eciency ratios obtained using observed cold side temperatures and
a xed temperature of 30 C. The maximum conversion eciency
obtained is just under 6% of Carnot eciency across the thermal
gradient while the average eciency ratio is approximately 3% across
the observed cycle. The curves indicate that eective thermal manage-
ment of cold side temperature could yield two fold increase in thermal
to electrical conversion eciency averaging just above 6% with a peak
of 12% of Carnot eciency. This analysis highlights the eect of heat
transfer irreversibility on the performance of TEG as a signicant
holdup in improvement of conversion eciency.
The eect of prevalent wind conditions on the performance of a
passively cooled solar TEG was investigated by Moraes et al. [33]. The
device features a solar evacuated tube as a heat source and a CPU
cooler as a heat sink and is exposed to 804.1 W/m2 of articial solar
radiation. Their results indicate that due to enhanced convective heat
transfer in windy conditions at the cold end of the TEM the power
production increases.
Natural convection cooling was employed by Nuwayhid et al.
[34,36] to maintain cold side temperatures on TEG used for waste
heat recovery from a woodstove. In the rst prototype the maximum
power was less than one watt per module. The commercially available
Fig. 7. Ratio of thermal eciency to Carnot eciency for a prototype wind chimney TEG
with naturally varying & xed cold side temperature (based on data from [32]).
Renewable and Sustainable Energy Reviews 78 (2017) 1522
60 mm n 80 mm n 100 mm n Heat pipe
8.1 to 21.4 9.7 to 22.1 10.6 to 23.4 32.7 to 40.3
15.3 to 48.9 18.1 to 50.3 19.5 to 52.2 59.0 to 76.7
high performance extruded aluminum heat sink used in the study gave
an average cold side temperature of 123 C for a hot side temperature
of 275 C and ambient conditions of 28 C. The indicated temperature
dierence across the TEG is 152 C while the open circuit voltage
obtained implies an actual temperature dierence of 85 C. The
considerable divergence between the two was due to the contact
resistance at the interfaces between the mounting plate and heat sink.
3.1.2. Fan cooled
The importance of managing cold side temperature in air cooled
TEG devices has been highlighted with optimization studies [3740]
with steady state and transient heat sources. Mastbergen and Willson
[41] presented a prototype TEG with a forced air-cooling for the cold
side with a 1Wfan. The net power of 4 W generated by the TE generator
was sucient to power an array of high intensity LEDs.
The used of fan cooled TEGs in automotive waste heat recovery has
been widely studied [4246]. Notably Hsu et al. [45,47] investigated a
range of heat sink congurations while employing forced air cooling of
the cold side of the device (Fig. 8). In the earlier work three
congurations comprising ten, twenty two and forty four ns in their
heat sinks were studied in order to nd the optimal conguration.
Their system generated a maximum of 44.13 W of power at which the
average temperature dierence is 88.3 K with 343.7 K at the cold side
of the TEG indicating a peak conversion eciency of 2.1%. A similar
conversion eciency was reported by Gou et al. [39] while forced air
cooling of TEG employed in industrial waste heat recovery. They were
able to restrict the peak cold side temperature increase to a 14 K
concluding that irreversibility in cold-side heat transfer is a bottle neck
in the enhancement of system performance.
Natural convection cooling systems while unable to deliver desir-
able cold side temperatures, have the advantage of no parasitic power
consumption. Whereas forced air systems employ a widely available
uid medium ensuring air cooled systems nd continued applications
in low heat ux scenarios.
3.2. Water cooled
Because of the superior heat transport properties of water, it nds
widespread application in TEGs [4852]. For instance, the thermal
eciency of a parabolic trough concentrator combined with TEMs
placed in the receiver tube with water cooling was found to be <4%
[53]. While a thermodynamic analysis of water cooled a parabolic dish
combined with a Bi-Te module predicted a solar-to-electricity eciency
of 2.81% at a temperature of 280 C [54]. The major drawback of
forced water cooling is that it parasitically draws power from the
generated amount reducing the net output of the TEG.
3.2.1. Natural convection
Due its limitations in heat transfer potential, natural convection
cooling of TEGs with water as a medium has received marginal
attention in past studies. Champier et al. [55] placed a TEM at the
bottom of an aluminum tank lled with water using it as a cold side
heat exchanger while the hot side consisted of a an aluminum ns that
extracted heat from exhaust air of a multifunctional biomass stove. The
water in the tank used for cold heat sink would be domestic hot water
in a real application. They were able to obtain up to 9.5 W from the
18
M. Sajid et al.
Fig. 8. Schematics of automotive waste heat recovery using TEG from Hsu et al. [47].
Fig. 9. Schematic of thermoelectric power generation from salinity-gradient solar pond
using a thermosyphon [57].
TEM in case of an almost matched load.
Another reasonable alternative to forced water cooling is an open
pan heat sink consisting simply of an open- topped pan full of water
[56]. While this draws on the large latent heat of vaporization of water
its downside is the risk of dry-out at higher temperatures which
requires preventative monitoring. Singh et al. [57] used natural
convection in the upper colder zone of a salinity gradient solar pond
to cool TEM which were placed in a thermosyphon (Fig. 9) placed in
between the lower hot zone and the upper cold zone of the pond. The
designed lab scale prototype was able to produce 3.2 W of power for a
temperature dierence of 27 C using sixteen thermoelectric cells.
3.2.2. Forced convection
The most widely used method of cooling TEGs is to ow cool water
over the cold side of the TEM with the only disadvantage being the
pumping cost. This has found applications in a wide range of
technologies ranging from waste heat recovery in automobiles and
industries to combined heat and power generation, some of which will
be reviewed here.
An experimental TEG unit was constructed by Niu et al. [51] that
was used to examine the inuences of the main operating conditions,
hot and cold uid inlet temperatures, ow rates and load resistance on
both power output and conversion eciency. The eect of coolant ow
rate on performance parameter was also part of the a study performed
by Chen et al. [58]. The earlier study achieved a peak conversion
eciency of 4.44% with a maximum power output of 146.5 W with cold
uid entering at 30 C and temperature dierence of 120 C. While
both researchers concluded that the performance parameters depend
more on the heating uid ow rate and on the inlet temperature of
cooling uid.
The conversion of solar energy to electrical power through TEGs
19
Renewable and Sustainable Energy Reviews 78 (2017) 1522
while employing active water cooling has been widely researched
[52,5963]. A microchannel heat sink was used by Fan et al. [63] to
cool the thermoelectric cell that produced 4.9 W at 2.9% electrical
eciency while operating across a 109 C temperature dierence.
Yazawa et al. [62] used a Fresnel lens to focus sunlight onto a
thermoelectric cell cooled by ambient water and produced 0.45 W
electrical power indicating an electrical conversion eciency of 1%
while 50% of the incident energy was captured by the water making it a
suitable cogeneration system. A peak conversion eciency of 0.78%
was reported by [61] using a parabolic dish concentrator with forced
convection cooling. An attempt to increase the solar to electric
conversion eciency was made [21] using a solar cavity receiver to
decrease re-radiation losses but only yielded a conversion eciency of
0.13% attributed to low gure of merit (0.05) of the thermoelectric cell.
A higher gure of merit (1) thermoelectric cell was used with an
evacuated tube solar collector to produce 64.8 W of power at 1.59%
electrical conversion eciency [52].
Many of these studies do not consider the parasitic losses incurred
in the pumps, piping and water-to-ambient heat exchangers. A
predictive computational model of a TEG waste heat dissipation system
was developed and tested by Aranguren et al. [48,64]. Such studies are
necessary for reducing auxiliary power consumption in water cooled
TEGs. Their study found that optimization of cooling system para-
meters (owrates, secondary heat exchanger and piping) could extend
power generation by up to 40% while operating at point of least
resistance.
3.2.3. Evaporative cooling
This methodology takes advantage of the high latent heat of
vaporization of water to provide cooling to the TEM from a thermo-
syphoning device such as a heat pipe. In the comparative study of
various cooling devices including heat pipes, cooling ns and bare
plates Date et al. [35] found that the lowest temperature at the cold end
of TEG was obtained by using a heat pipe. The maximum allowable
heat ux for these congurations has been summarized in Table 1 for
various wind conditions. The passive nature of the heat pipe design
makes it more attractive.
Another interesting attempt at employing a heat pipe on the cold
side of a TEG was made by Nuwayhid and Hamade [65] employing a
loop type conguration (Fig. 10). They were able to achieve 3 W of
power for a temperature dierence of 7080 C and a cold side
temperature exceeding 100 C leaving much to be desired from the
heat removal system.
A theoretical model was developed by Remeli et al. [66] to predict
the waste heat recovery and electrical conversion performances using
heat pipes on both hot and cold sides of a TEG. The experimental setup
(Fig. 11) included a TEG enclosed between two heat pipes to achieve
the temperature gradient for thermoelectric power generation. For a
maximum temperature dierence of 86.7 C they were able to achieve
highest conversion eciency of 2.02% with a peak 1.12 W of power
output.
A four to six times increase in power produced from TEGs was
reported by Djafar et al. [67] as a result of using heat pipes on for
M. Sajid et al.
Fig. 10. Schematic of a closed loop thermosyphonic heat pipe thermoelectric generator
[65].
Fig. 11. Outline of the heat pipe thermoelectric power generation system [66].
cooling TEMs. The use of heat pipes in conjunction with solar collectors
was investigated by He et al. [30,49]. Their study found that for water
at 45 C and solar irradiation larger than 600 W/m2 the solar heat pipe
thermoelectric system may have a thermal eciency of about 55% and
an electrical eciency above 1% compared to a 34% electrical
eciency of an organic Rankine cycle. Evaporative cooling of TEG
through heat pipes has also been employed in automobiles [44] to
replace the conventional radiator providing a combined power of 75 W
at 2.1% device eciency when employed in a 2 L engine of passenger
car. The vehicle was driving at 80 km/h during the test benetting from
enhanced convective transfer from the ns of the heat pipe and
indicated a thermal performance considerably higher than a conven-
tional radiator due to the added advantage of heat removal by
thermoelectric conversion.
4. Discussion
Essential performance characteristics from published experimental
works in TEGs are summarized in Table 1 to gain a clear perspective on
20
Renewable and Sustainable Energy Reviews 78 (2017) 1522
Fig. 12. Performance parameters (a) Eciency and (b) Power output of TEGs employ-
ing dierent heat dissipation strategies and operating between dierent temperatures.
heat dissipation abilities of the cooling methodologies detailed in
previous sections. The data listed is for conditions of peak power
output for a single TEM, eciencies have been calculated using Eq. (1).
Parameters aecting the heat exchanger like ow rates, dimensions and
n conguration have been excluded as they are not common in all
cooling devices. The cold side temperature has been chosen as the
reference in conjunction with Section 2.3, presenting analysis of
thermal dependence of conversion eciency.
An obstacle preventing the deduction of concise conclusion from
these data points is the lack of a common datum plane, i.e. each device
is functioning across varying temperature gradients with a dissimilar
cold side temperature, dierent cooling mechanism and most impor-
tantly, using diverse thermoelectric cells with variation in composition,
structure, and material properties. To simplify the comparative analy-
sis of the performance parameters of these ve heat dissipation
strategies, Fig. 12 presents the eciency and power output of studied
TEGs against a spread of cold side temperatures and thermal gradients
across the TEM.
Most of the works included in Table 2 were carried out with
commercially available TEMs which generally advertise conversion
eciencies ~4% over an average temperature gradient across the
module. However, we note two instances where a conversion eciency
> 4% was observed. In the rst case forced water cooling led to a cold
side temperature of standard atmosphere and a temperature gradient
of 120 giving an eciency of 4.44% [51]. Based on the data we estimate
a peak power output of 2.61 W/module while the parasitic losses
incurred by pumping coolants amount to 0.06 W/module resulting in
net power output of 2.55 W/module. While in the second case dry draft
cooling was used in a solar TEG to maintain the cold side of the TEM
which provided peak power of 0.83 W and an eciency of 4.59% at
M. Sajid et al. Renewable and Sustainable Energy Reviews 78 (2017) 1522

Table 2 losses of pumping and pipe friction can be cost eective. While natural
Performance characteristics of various cooling mechanisms used in TEGs from published ventilation remains advantageous because of its independence from the
works.
provision of water source despite its lower performance parameters.
Cooling Tc, Cold side T, , P, Reference
mechanism Temperature Temperature Conversion Power 5. Conclusion
(K) across TEM eciency output
(%) (W) This work summarizes the various cooling methods that have been
Passive air 349.8 110.7 4.59 0.83 [32] employed to maintain cold side temperature of a thermoelectric
cooling 298 425 2.90 0.48 [68] generator. The dependence of thermoelectric conversion eciency on
315 135 3 1.8 [69] cold side temperature was highlighted at the beginning, laying the
318 155 4 3 [70] grounds for the necessity of focusing on cooling of TEGs. The overview
396 152 2.1 4.2 [34]
of performance parameters for the various heat rejection techniques
363 150 3.2 2.4 [71]
418 135 1.2 0.57 [72] provides useful insight into prospects of integrating these systems with
Fan cooled 318 76 2.40 1.08 [73] thermoelectric generation. Evaporative heat pipe systems have the
373 30 0.34 0.52 [47] lowest cooling potential which is useful for low temperature TEGs
343.7 88.3 2.16 5.52 [45]
while forced convection of water can cater to larger thermal systems
335 17.8 2 0.85 [39]
Passive 336 27 1 3.2 [57] with higher heat uxes. The information presented can also be used for
water 363 120 Not 9.5 [55] combining the thermoelectric generators with a heat source, designing
Available the cooling system and a standalone power generation system. This
338 160 2 2.3 [74] research shows the great potential of thermoelectric generators to be
Forced 303 270 1.59 6 [52]
combined either with concentrated solar thermal systems, industrial
water 325.7 97.4 0.78 1.75 [58]
cooling 300 600 0.13 0.86 [27] heat recovery systems or any other available heat source together with
317 109 2.90 4.9 [63] passive cooling to become a feasible technology for medium to large
303 120 4.44 2.61 [51] scale power generation with the attraction of minimal moving parts.
294 63 0.80 0.34 [75]
Finally, this review highlights the potential of improving performance
291 198 2.70 3 [76]
Heat pipe 318 55 2.10 1.04 [44]
of TEGs by rening the thermal management of cold side of the
333 86.7 2.02 1.12 [66] thermoelectric cell.

13:30 [32], while the average eciency was 3.52% over the entire
duration of the device operation. This comparison indicates that forced
water cooling is more eective despite the parasitic losses in pumping
water. In this case these losses amount to a meagre 2.39% of the total
power output, whereas Aranguren et al. [48] reported a 40% increase in
power output after minimizing parasitic losses. This could be due to
dierence in number of TEMs in each system or pipe losses which are
not reported. In these cases of passive air and forced water cooling, the
TEM operates between a comparable temperature dierence of 120 K
and 110 K respectively, while there is a signicant dierence between
their cold side temperatures i.e, ~47 K. This substantiates the relation-
ship between power output and cold side thermal management.
For a broader perspective, Fig. 12a indicates overall higher
eciencies for forced water and passive air cooling, followed by fan
cooled and thermal syphoning with passive water being the least
ecient. While Fig. 12.b rearms the association of forced water
cooling with better performance in terms of power output, followed by
air cooled systems while heat pipes are trailing. The highest reported
power output of 9.5 W [55] using passive water cooling is excluded
from the discussion as the corresponding eciency was not reported by
the authors. This exclusion is justied by comparing passive water
cooling [74], active air cooling [45], and passive air cooling [34],
operating at similar eciencies of 2%, 2.16% and 2.1%, and across
temperature dierences of 160 K, 88.3 K and 152 K respectively. While
the case involving passive water cooling has the lower cold side
temperature of 338 K against 343 K for fan cooled and 396 K for
natural convection, fan based cooling gives the highest power of 5.52 W
against 4.2 W and 2.3 W for passive air and water. This can be
explained by the lower cold side temperature achieved by active air
cooling, while generally fan cooled TEG systems fail to achieve
signicant temperature dierence, 88.3 K being the highest reported.
However, the cost of operating fans has not been accounted for in these
cases which indicates that the high power output of fan based cooling is
an exception to the trend.
To summarize, heat pipes have provided the least benet in cold
side thermal management followed by passive water cooling, however
each has niche applications. Active water cooling, despite the parasitic
References
[1] Rowe DM. Thermoelectrics, an environmentally-friendly source of electrical power.
Renew Energy 1999;16:12516.
[2] LeBlanc S. Thermoelectric generators: linking material properties and systems
engineering for waste heat recovery applications. Sustain Mater Technol 2014;1
2:2635.
[3] Liu W, Jie Q, Kim HS, Ren Z. Current progress and future challenges in thermo-
electric power generation: from materials to devices. Acta Mater 2015;87:35776.
[4] Bell LE. Cooling, heating, generating power, and recovering waste heat with
thermoelectric systems. Science 2008;321:145761.
[5] Suman S, Khan MK, Pathak M. Performance enhancement of solar collectorsA
review. Renew Sustain Energy Rev 2015;49:192210.
[6] Orr B, Akbarzadeh A, Mochizuki M, Singh R. A review of car waste heat recovery
systems utilising thermoelectric generators and heat pipes. Appl Therm Eng
2016;101:4905.
[7] Gao HB, Huang GH, Li HJ, Qu ZG, Zhang YJ. Development of stove-powered
thermoelectric generators: a review. Appl Therm Eng 2016;96:297310.
[8] Tritt TM. Thermoelectric phenomena, materials, and applications. Annu Rev Mater
Res 2011;41:43348.
[9] Shakouri A, Zebarjadi M. Nanoengineered materials for thermoelectric energy
conversion. In: Thermal nanosystems and nanomaterials. ed: Springer; 2009. p.
22599.
[10] Rowe DM. Recent developments in thermoelectric materials. Appl Energy
1986;24:13962.
[11] W. C . Materials for thermoelectric energy conversion. Rep Prog Phys 1988;51:459.
[12] Zheng XF, Liu CX, Yan YY, Wang Q. A review of thermoelectrics research Recent
developments and potentials for sustainable and renewable energy applications.
Renew Sustain Energy Rev 2014;32:486503.
[13] Alam H, Ramakrishna S. A review on the enhancement of gure of merit from bulk
to nano-thermoelectric materials. Nano Energy 2013;2:190212.
[14] Zhang Q, Liao J, Tang Y, Gu M, Ming C, Qiu P, Bai S, Shi X, Uher C, Chen L.
Realizing a thermoelectric conversion eciency of 12% in bismuth telluride/
skutterudite segmented modules through full-parameter optimization and energy-
loss minimized integration. Energy Environ Sci 2017;10:95663.
[15] Kraemer D, Jie Q, McEnaney K, Cao F, Liu W, Weinstein LA, Loomis J, Ren Z,
Chen G. Concentrating solar thermoelectric generators with a peak eciency of
7.4%. Nat Energy 2016;1:16153.
[16] Kim SI, Lee KH, Mun HA, Kim HS, Hwang SW, Roh JW, Yang DJ, Shin WH, Li XS,
Lee YH, Snyder GJ, Kim SW. Dense dislocation arrays embedded in grain
boundaries for high-performance bulk thermoelectrics. Science 2015;348:10914,
[April 3, 2015].
[17] Biswas K, He J, Blum ID, Wu C-I, Hogan TP, Seidman DN, Dravid VP, Kanatzidis
MG. High-performance bulk thermoelectrics with all-scale hierarchical architec-
tures. Nature 2012;489:4148.
[18] Zhao L-D, Lo S-H, Zhang Y, Sun H, Tan G, Uher C, Wolverton C, Dravid VP,
Kanatzidis MG. Ultralow thermal conductivity and high thermoelectric gure of
merit in SnSe crystals. Nature 2014;508:3737.

21
M. Sajid et al.
[19] Chen Z-G, Han G, Yang L, Cheng L, Zou J. Nanostructured thermoelectric
materials: current research and future challenge. Progress Nat Sci: Mater Int
2012;22:53549.
[20] Shakouri A. Recent developments in semiconductor thermoelectric physics and
materials. Mater Res 2011;41:399.
[21] Weidenka A, Trottmann M, Tome P, Suter C, Steinfeld A, Veziridis A. Solar TE
converter applications. In: Koumoto K, Mori T, editors. . Thermoelectric nano-
materials, 182. Springer Berlin Heidelberg; 2013. p. 36582.
[22] Crane DT. Thermoelectric waste heat recovery program for passenger vehicles;
2013.
[23] TM T, H B, LD C. Thermoelectrics: direct solar thermal energy conversion. MRS
Bull 2008;33:366.
[24] Yang J, Caillat T. Thermoelectric materials for space and automotive power
generation. MRS Bull 2006;31:2249.
[25] Rowe DM. A high performance solar powered thermoelectric generator. Appl
Energy 1981;8:26973.
[26] Tome P, Trottmann M, Suter C, Aguirre MH, Steinfeld A, Haueter P, Weidenka
A. Thermoelectric oxide modules (TOMs) for the direct conversion of simulated
solar radiation into electrical energy. Materials 2010;3:280114.
[27] Suter C, Tome P, Weidenka A, Steinfeld A. A solar cavity-receiver packed with an
array of thermoelectric converter modules. Sol Energy 2011;85:15118.
[28] Fujita K, Mochida T, Nakamura K. High-temperature thermoelectric properties of
Na x CoO 2- single crystals. Jpn J Appl Phys 2001;40:4644.
[29] Kraemer D, Poudel B, Feng H-P, Caylor JC, Yu B, Yan X, Ma Y, Wang X, Wang D,
Muto A, McEnaney K, Chiesa M, Ren Z, Chen G. High-performance at-panel solar
thermoelectric generators with high thermal concentration. Nat Mater
2011;10:5328.
[30] He W, Su Y, Wang YQ, Riat SB, Ji J. A study on incorporation of thermoelectric
modules with evacuated-tube heat-pipe solar collectors. Renew Energy
2012;37:1429.
[31] Jaworski M, Bednarczyk M, Czachor M. Experimental investigation of thermo-
electric generator (TEG) with PCM module. Appl Therm Eng 2016;96:52733.
[32] zdemir AE, Kysal Y, zba E, Atalay T. The experimental design of solar heating
thermoelectric generator with wind cooling chimney. Energy Convers Manag
2015;98:12733.
[33] Moraes FS, Santos LC, Alencar RN, Sempels V, Sandoval V JC, Lesage FJ. Solar
thermoelectric generator performance relative to air speed. Energy Convers Manag
2015;99:32633.
[34] Nuwayhid RY, Shihadeh A, Ghaddar N. Development and testing of a domestic
woodstove thermoelectric generator with natural convection cooling. Energy
Convers Manag 2005;46:163143.
[35] Date A, Date A, Dixon C, Singh R, Akbarzadeh A. Theoretical and experimental
estimation of limiting input heat ux for thermoelectric power generators with
passive cooling. Sol Energy 2015;111:20117.
[36] Nuwayhid RY, Rowe DM, Min G. Low cost stove-top thermoelectric generator for
regions with unreliable electricity supply. Renew Energy 2003;28:20522.
[37] Nguyen NQ, Pochiraju KV. Behavior of thermoelectric generators exposed to
transient heat sources. Appl Therm Eng 2013;51:19.
[38] Wang C-C, Hung C-I, Chen W-H. Design of heat sink for improving the
performance of thermoelectric generator using two-stage optimization. Energy
2012;39:23645.
[39] Gou X, Xiao H, Yang S. Modeling, experimental study and optimization on low-
temperature waste heat thermoelectric generator system. Appl Energy
2010;87:31316.
[40] Crane DT, Jackson GS. Optimization of cross ow heat exchangers for thermo-
electric waste heat recovery. Energy Convers Manag 2004;45:156582.
[41] Mastbergen D, Willson B. Generating light from stoves using a thermoelectric
generator. In: Proceedings of the ETHOS international stove research conference;
2005.
[42] Su CQ, Xu M, Wang WS, Deng YD, Liu X, Tang ZB. Optimization of cooling unit
design for automotive exhaust-based thermoelectric generators. J Electron Mater
2015;44:187683, [06/01 2015].
[43] Ma Z, Wang X, Yang A. Inuence of temperature on characters of thermoelectric
generators based on test bed. J Nanomater 2014;2014:6.
[44] Kim S, Park S, Kim S, Rhi S-H. A thermoelectric generator using engine coolant for
light-duty internal combustion engine-powered vehicles. J Electron Mater
2011;40:8126, [05/01 2011].
[45] Hsu C-T, Yao D-J, Ye K-J, Yu B. Renewable energy of waste heat recovery system
for automobiles. J Renew Sustain Energy 2010;2:013105.
[46] Huang G-Y, Hsu C-T, Fang C-J, Yao D-J. Optimization of a waste heat recovery
system with thermoelectric generators by three-dimensional thermal resistance
analysis. Energy Convers Manag 2016;126:58194.
[47] Hsu C-T, Huang G-Y, Chu H-S, Yu B, Yao D-J. Experiments and simulations on
low-temperature waste heat harvesting system by thermoelectric power generators.
Appl Energy 2011;88:12917.
[48] Aranguren P, Astrain D, Prez MG. Computational and experimental study of a
complete heat dissipation system using water as heat carrier placed on a thermo-
electric generator. Energy 2014;74:34658.
[49] He W, Su Y, Riat SB, Hou J, Ji J. Parametrical analysis of the design and
22
Renewable and Sustainable Energy Reviews 78 (2017) 1522
performance of a solar heat pipe thermoelectric generator unit. Appl Energy
2011;88:50839.
[50] Kiemariam R, Lin C-X. Numerical simulation of integrated liquid cooling and
thermoelectric generation for self-cooling of electronic devices. Int J Therm Sci
2015;94:193203.
[51] Niu X, Yu J, Wang S. Experimental study on low-temperature waste heat
thermoelectric generator. J Power Sources 2009;188:6216.
[52] Zhang M, Miao L, Kang YP, Tanemura S, Fisher CAJ, Xu G, Li CX, Fan GZ.
Ecient, low-cost solar thermoelectric cogenerators comprising evacuated tubular
solar collectors and thermoelectric modules. Appl Energy 2013;109:519.
[53] Omer SA, Ineld DG. Design and thermal analysis of a two stage solar concentrator
for combined heat and thermoelectric power generation. Energy Convers Manag
2000;41:73756.
[54] Eswararmoorthy M, Shanmugam S. Thermodynamic analysis of solar parabolic
dish thermoelectric generator. Int J Renew Energy Technol 2010;1:34860.
[55] Champier D, Bdcarrats JP, Kousksou T, Rivaletto M, Strub F, Pignolet P. Study of
a TE (thermoelectric) generator incorporated in a multifunction wood stove. Energy
2011;36:151826.
[56] Nuwayhid RY, Moukalled F, AbuSaid R, Daaboul M. Practical design considerations
for a rural stove-top thermoelectric generator. In: Proceedings of the 19th
international conference on thermoelectrics (ICT2000), ed. Cardi, Wales: Babrow,
Wales; 2000.
[57] Singh R, Tundee S, Akbarzadeh A. Electric power generation from solar pond using
combined thermosyphon and thermoelectric modules. Sol Energy 2011;85:3718.
[58] Chen W-H, Liao C-Y, Hung C-I, Huang W-L. Experimental study on thermoelectric
modules for power generation at various operating conditions. Energy
2012;45:87481.
[59] Olsen ML, Warren EL, Parilla PA, Toberer ES, Kennedy CE, Snyder GJ, Firdosy SA,
Nesmith B, Zakutayev A, Goodrich A, Turchi CS, Netter J, Gray MH, Ndione PF,
Tirawat R, Baranowski LL, Gray A, Ginley DS. A high-temperature, high-eciency
solar thermoelectric generator prototype. Energy Procedia 2014;49:14609.
[60] Muthu G, Shanmugam S, Veerappan AR. Solar parabolic dish thermoelectric
generator with acrylic cover. Energy Procedia 2014;54:210.
[61] Eswaramoorthy M, Shanmugam S, Veerappan A. Experimental study on solar
parabolic dish thermoelectric generator. Int J Energy Eng 2013;3:62.
[62] Yazawa K, Wong VK, Shakouri A. Thermal challenges on solar concentrated
thermoelectric CHP systems. In: Proceedings of the 2012 13th IEEE intersociety
conference on thermal and thermomechanical phenomena in electronic systems
(ITherm); 2012. p. 114450.
[63] Fan H, Singh R, Akbarzadeh A. Electric power generation from thermoelectric cells
using a solar dish concentrator. J Electron Mater 2011;40:131120, [05/01 2011].
[64] Aranguren P, Astrain D, Martnez A. Study of complete thermoelectric generator
behavior including water-to-ambient heat dissipation on the cold side. J Electron
Mater 2014;43:232030.
[65] Nuwayhid RY, Hamade R. Design and testing of a locally made loop-type
thermosyphonic heat sink for stove-top thermoelectric generators. Renew Energy
2005;30:110116.
[66] Remeli MF, Tan L, Date A, Singh B, Akbarzadeh A. Simultaneous power generation
and heat recovery using a heat pipe assisted thermoelectric generator system.
Energy Convers Manag 2015;91:1109.
[67] Djafar Z, Putra N, Koestoer RA. The utilization of heat pipe on cold surface of
thermoelectric with low-temperature waste heat. Appl Mech Mater 2013:4105.
[68] Pereira A, Caro T, Lorin G, Bae T, Romanjek K, Vesin S, Kusiaku K, Duchemin
H, Salvador V, Miloud-Ali N, Aixala L, Simon J. High temperature solar thermo-
electric generator indoor characterization method and modeling. Energy
2015;84:48592.
[69] Amatya R, Ram RJ. Solar thermoelectric generator for micropower applications. J
Electron Mater 2010;39:173540, [09/01 2010].
[70] Chvez-Urbiola EA, Vorobiev YV, Bulat LP. Solar hybrid systems with thermo-
electric generators. Sol Energy 2012;86:36978.
[71] Lertsatitthanakorn C. Electrical performance analysis and economic evaluation of
combined biomass cook stove thermoelectric (BITE) generator. Bioresour Technol
2007;98:16704.
[72] Candadai AA, Kumar VP, Barshilia HC. Performance evaluation of a natural
convective-cooled concentration solar thermoelectric generator coupled with a
spectrally selective high temperature absorber coating. Sol Energy Mater Sol Cells
2016;145(Part 3):33341.
[73] Mgbemene CA, Duy J, Sun H, Onyegegbu SO. Electricity generation from a
compound parabolic concentrator coupled to a thermoelectric module. J Sol Energy
Eng 2010;132:031015, [031015].
[74] Champier D, Bedecarrats JP, Rivaletto M, Strub F. Thermoelectric power genera-
tion from biomass cook stoves. Energy 2010;35:93542.
[75] Ding LC, Akbarzadeh A, Singh B, Remeli MF. Feasibility of electrical power
generation using thermoelectric modules via solar pond heat extraction. Energy
Convers Manag 2017;135:7483.
[76] Sundarraj P, Taylor RA, Banerjee D, Maity D, Roy SS. Experimental and theoretical
analysis of a hybrid solar thermoelectric generator with forced convection cooling. J
Phys D: Appl Phys 2016;50:015501.