Solar Energy Materials & Solar Cells 157 (2016) 93–100

Contents lists available at ScienceDirect

Solar Energy Materials & Solar Cells

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

Performance evaluation of a thermally concentrated solar

thermo-electric generator without optical concentration
K.Y. Sudharshan 1, V. Praveen Kumar, Harish C. Barshilia n
Nanomaterials Research Laboratory, Surface Engineering Division, CSIR-National Aerospace Laboratories, Bangalore 560017, India

art ic l e i nf o a b s t r a c t

Article history: Recent research has shown the use of solar thermoelectric generators (STEG) to be an alternative method
Received 10 January 2016 of harnessing solar energy. They comprise of a spectrally selective absorber coupled to thermoelectric
Received in revised form materials to generate power as governed by the Seebeck effect. Since output power generated is
17 April 2016
dependent on thermal gradient in the STEG, it is important to maximize this parameter. This can be done
Accepted 14 May 2016
by increasing energy absorbed from the sun and minimizing energy lost from the system. Herein, the
former has been achieved using thermal concentration by area while the latter has been addressed by
Keywords: use of a vacuum enclosure. The effect of varying enclosure pressure on convective heat loss from the solar
Solar thermoelectric generator selective substrate (α ¼0.90, ε ¼ 0.15, area: 22.4 cm  12 cm) was studied. It was observed that at
Thermal concentration
100 mbar enclosure pressure, convective heat transfer coefficient as low as 0.6 W/m2 °C (cf., 10 W/m2 °C
Absorber coating
at atmospheric pressure) was obtained. This resulted in the rise of substrate temperature from 97 °C at
Heat transfer
ambient pressure to 142 °C at 100 mbar enclosure pressure, an observation validated by simulations.
Further study was conducted where open circuit voltage (OCV) and output power were observed and
efficiency was calculated at different enclosure pressures. In these tests the effect of using a large area
absorber was also studied. It was observed that decreasing the enclosure pressure increased the OCV,
output power and efficiency of the STEG setup. It was also noted that using a large surface area absorber
further increased the OCV and output power of the setup. During the course of the study, a maximum
OCV of 823 mV, output power of 44.2 mW, and an efficiency of 0.82% were achieved by the setup at an
enclosure pressure of 100 mbar and thermal concentration ratio of 22. The results show that the present
setup may be used as an alternative to optically concentrated STEGs which require a power consuming
solar tracking system.
& 2016 Elsevier B.V. All rights reserved.

1. Introduction Thermo-electrics generate an electrical potential gradient from

The need for researching alternative sources of energy has
arisen due to the growing demand for power and also the detri-
mental impact of the use of conventional fuels [1]. The sun being
the most widespread and abundant source of energy is therefore
the most viable option [2]. Photovoltaics and large-scale solar
thermal power systems are currently the most used means of
harnessing solar energy [3,4]. Recent research shows that spec-
trally selective absorbers coupled with thermoelectric materials
can serve as an alternative solar energy harnessing technique for
micro-power applications [5]. Focus on the latter has increased in
recent years since further growth in the field of photovoltaics is
now limited [6].
n
Corresponding author.
E-mail address: harish@nal.res.in (H.C. Barshilia).
1
Present address: Department of Mechanical Engineering, BITS Pilani,
Pilani, India.
a thermal gradient as governed by the Seebeck effect [7] expressed
as:
V OC ¼ ðSp Sn ÞðT H T C Þ ð1Þ
where Sp and Sn are Seebeck coefficient of the ‘p’ and ‘n’ junctions
and TH and TC are temperatures of the hot and cold junctions.
Thermo-electrics are mostly used as waste-heat recovery sys-
tems but can be coupled with spectrally selective substrates to
harness solar energy. These arrangements are called solar ther-
moelectric generators (STEGs). The growing interest in STEGs is
due to the fact that they require no working fluid and have no
movable parts, which greatly reduce, if not eliminate, the need for
regular maintenance [8,9]. This would be particularly beneficial for
power generation in remote locations where logistics would make
regular maintenance unviable.
Efficiency is critical for every energy system. Improving STEG
power generation efficiency can be done by increasing the thermal
gradient across the hot and cold junctions of the thermo-electric
modules, by improving the absorber (application of solar selective

http://dx.doi.org/10.1016/j.solmat.2016.05.033
0927-0248/& 2016 Elsevier B.V. All rights reserved.
94 K.Y. Sudharshan et al. / Solar Energy Materials & Solar Cells 157 (2016) 93–100
coatings) and module efficiencies (which can only be increased by
improving TEG figure of merit [5]). Maria Telkes had used optical
concentration using a Fresnel lens and achieved a system effi-
ciency of 3.36% way back in 1954 [10] and was one of the pioneers
of STEG research. Other work on STEG comes from Nattestad et al.
[11], who achieved only 2.42% system efficiency. Further work was
done by Olsen et al. [12] in 2013 where they sought to demon-
strate near 15% efficiency from their setup far outstripping the
performance of the former. Aside from using optical concentration
and vacuum enclosure, they also used a water cooling system to
cool the cold junction of the STEG module.
The main drawback in the above setups, which prevents their
widespread use, is the need for a solar tracking system. Such a
system is required due to the fact that maximum transmittance of
a Fresnel lens occurs when solar flux is incident along its principal
axis [13]. The tracking system also feeds off the already minimal
power generated by the STEGs, therefore, making the system
unviable for micro-power applications. Daniel Kraemer et al. had
come up with an alternative to the above setup and achieved
better efficiencies (4.46%) without the use of optical concentration
by using vacuum, a nano-structured absorber coating and cooling
system instead [14]. Nanostructured absorbers show higher
absorptance and lower emittance than other coatings, improving
absorber efficiency and therefore hot junction temperature. By
placing the module in vacuum, they managed to minimize heat
loss from the hot end by convection, while the cooling system
served to lower cold junction temperature. The aggregate effect
was the increase of the temperature gradient in the module and
thus higher system efficiency.
Experiments recently conducted by Candadai et al. saw them
achieve an efficiency of 1.2% at an optical concentration ratio of 65
using commercial Bi2Te3 modules by coupling to solar selective
substrates [15]. Wan et al. [16] on the other hand, coupled STEGs
with photovoltaics to get improved system efficiency of 13% in a
very innovative design. The research into TEGs has created such
excitement in the world of research with new innovative designs
such as the Flexible Thin Film TEG [17], opening new avenues for
applications. It must be noted, however, that applications are
restricted by feasibility, particularly economicity of the setup,
which is one focus of this experiment, i.e., to create a maintenance
free setup with easily available components.
In the present work we demonstrate the use of thermal con-
centration in the STEG device. The first part of the present study is
to understand the impact of enclosure pressure on the convective
heat transfer coefficient (h) in the setup. Since convective heat
transfer is the main cause of energy loss from the hot junction,
minimizing this factor would significantly improve system effi-
ciency. This is an implication of the study conducted by Hosseini
and Saidi [18], one of the few researchers to study the impact of
enclosure pressure on the convective heat transfer coefficient. This
study also serves to test the solar selective coating performance as
well as the ability of the enclosure to create and maintain vacuum.
The coefficient of convective heat transfer in this assembly is also
measured. The latter part of the study is the application of the
results achieved from the above performance tests to determine
the power and efficiency of the STEG setup in the enclosure con-
ditions. It also helps to evaluate the impact of thermal con-
centration by area.
2. Theory
2.1. Solar selectivity
The solar spectrum is comprised of three major regions – Ultra-
Violet, Visible and Infra-Red, each of which contributes to the
energy from the sun. Most of the energy is contained in the
wavelength range of 0.4–2.5 mm [19]. In order to draw maximum
energy from the sun, the absorber should ideally be almost
entirely absorptive in this range and have very low emittance. This
can be achieved by means of certain coatings atop metal sub-
strates. In the experiments that follow, an AlTiN/AlTiON/AlTiO
based absorber coating with α ¼ 0.90 and ε ¼ 0.15 on an aluminium
substrate was used [15]. The angular dependence of optical
properties of the spectrally selective absorber coating can be found
elsewhere [20,21]. The efficiency of the absorber given by Eq. (2)
[22] critically affects the efficiency of the STEG setup.
σϵT 4abs
ηabs ¼ α  ð2Þ
q
2.2. Convective heat transfer
It is the form of heat transfer in occurring in fluid medium. It is
one of the primary causes of heat loss in the STEG setup. Exposure
of hot surfaces to the atmosphere leads to heat exchange between
the surface and the medium causing a cooling. The heat exchange
by convection is expressed as follows [23]:
dT
Q_ conv ¼ hA ð3Þ
dt
The term ‘h’, is dependent on the conditions of the fluid med-
ium. It must also be noted that convective heat loss rate is also
proportional to the term dT/dt, i.e., the rate of change of surface
temperature. This term is higher for hotter surfaces and lesser for
colder ones. In other words, the hotter junction of the STEG system
would cool faster than the colder junction, which would mean that
the temperature gradient across the STEG would drop at a very
quick rate. The principle behind the present experiment is to
reduce the heat loss by convection by varying ambient conditions,
i.e., ‘h’ rather than dT/dt since the former is intrinsic to the material
and its surroundings while the latter is dependent on the tem-
perature of the system which cannot be compromised. The impact
of ambient pressure on ‘h’ has been previously studied by Hosseini
and Saidi [18] where they found that the heat loss rate decreased
with enclosure pressure.
The above phenomenon can be understood mathematically by
the means of the following expressions. Decreasing the enclosure
pressure, can be viewed as a decrease in density of enclosure gas.
Eqs. (4)–(7) show that decreasing pressure decreases the value of
‘h’, which in turn decreases the heat lost by a surface (Eq. (3)) [24].
P ¼ ρRT ð4Þ
where P is pressure of gas, ρ is its density, R is the gas constant and
T is the temperature of the gas.
Ra ¼ ρ  Gr ð5Þ
The density of the gas affects Ra which is the Rayleigh's num-
ber. Gr in the above equation is Grashof's number.
"
16 #2
1 1 f4
Nu ¼ Nu20 þ Ra6 ðPrÞ ð6Þ
300
Ra in turn affects Nu which is the Nusselt's number while f4(Pr)
is a function of the Prandtl's number.
hL
Nu ¼ ð7Þ
k
This directly impacts ‘h’ as shown in the above equation where
L is dimension factor while k is the coefficient of thermal con-
duction of the fluid. The estimation of ‘h’ is carried out by the use
of the following heat balance equation which takes into account
heat loss by convection and radiation. Conduction has been
 K.Y. Sudharshan et al. / Solar Energy Materials & Solar Cells 157 (2016) 93–100
neglected due to the minimum heat transfer involved in this
pathway.
dT s  
mC p ¼ αqAinc  ½hAðT s  T c Þ þ σεA T 4s  T 4enc  ð8Þ
dt
where m is thermal mass, Ts is temperature of the substrate and
Tenc is the temperature of the enclosure. Aside from determining
the steady state temperature of the sample, the first experiment
also seeks to prove this hypothesis and the validity in its
application.
2.3. Thermal concentration
The solar selective absorber serves as the hot end of the STEG
assembly. Assuming the STEG assembly to be similar to a Carnot
heat engine, the absorber is analogous to the heat source. To
achieve highest efficiency the maximum temperature of the
absorber must be raised. This is achieved by use of optical con-
centration where incident solar radiation is focused on a small
absorber area to obtain high temperatures. There are primarily
two drawbacks to this technique. First, an optical tracking system
is required for optimum performance, which consumes power and
second is the fact that variation in radiation intensity immediately
affects absorber temperature. Thermal concentration is a means to
overcome these drawbacks. In this technique, a large surface area
flat plate absorber is coupled to the TEG modules. The use of such
an absorber allows a higher amount of energy to be absorbed (as
absorbing area has increased).
Q in ¼ αqAin T r
where α is absorptance, q is incident heat flux, Ain is area of
incident flux and Tr is transmittance of glass. This is particularly
useful when solar intensity is low (overcast conditions) where a
small amount of energy can still be drawn despite the reduced
solar flux. Hardly any energy can be drawn in optically con-
centrated setups under similar conditions, since the energy
absorbed from incident flux to that lost by convection is insuffi-
cient to generate an adequate thermal gradient. Increasing the
absorber area presents another advantage by serving as a stable
heat source. Heat flow at the absorber-TEG interface results in a
maximum heat loss at the interface, which implies that should
incident flux be interrupted the hot end temperature quickly
reduces, causing a decrease in STEG output. The effect of these
intermittent fluctuations can be overcome by means of using the
large area absorber where heat energy from non-interface areas
compensates for local temperature variations. As a result tem-
perature measurements indicate slightly higher temperature at
interface, leading to a heat concentration-like phenomenon. This
concentration is denoted by:
Aabs
X th ¼
ASTEG
The disadvantage of the thermal concentration lies in the low
absorber temperature observed as compared to optical con-
centration by Fresnel lens making it unsuitable for large scale
power generation. However, its ability to compensate for inter-
mittent fluctuations makes it possible for application in micro
power generation.
2.4. Power output and efficiency analysis
Efficiency of the STEG system is critical in evaluating its per-
formance during operation. The STEG system comprises of a solar
selective substrate coupled with a thermo-electric generator. This
efficiency can be defined as the ratio of output power to input
power. This output power can be determined by the use of the
95
following formula where V and I are voltage and current produced
by the setup, respectively:
P out ¼ V  I ð11Þ
This value is maximum at matched load condition, i.e., when
external load matches internal resistance of the circuit [25,26]. The
input power on the other hand can be calculated as:
P in ¼ AαqX ð12Þ
where A is area of absorber, α is absorptance, q is incident solar
flux and X is the concentration ratio applied on the incident light.
For X¼ 1, the efficiency of the system is calculated by the formula:
P out
ηð%Þ ¼  100 ð13Þ
P in
Using the above equations we can derive the expression for
efficiency of the STEG system as shown in Eq. (14).
 
αqAinc  σεArad T subs 4 T surr 4 hAconv ðT subs  T surr Þ P out
η¼ ¼
αqAinc αqAinc
ð14Þ
where Ainc is the area under the incident radiation, Arad is the area
involved in re-radiation and Aconv is the area dissipating heat by
convection.
3. Experimental details
ð9Þ 3.1. Setup
The experiment was conducted in two parts with the following
objectives. The objective of the first part of the experiment was to
conduct temperature studies on solar selective substrates at var-
ious enclosure pressures as well as to test the integrity of the
enclosure. The second half of the experiment used the results from
the first to evaluate the performance of the STEG system on the
basis of open circuit voltage and output power. Due to the different
objectives, the setups for the two halves of the experiment had
certain differences. The common feature was the enclosure. The
enclosure consisted of a vacuum chamber of inner diameter 24 cm
and a height of 15.7 cm, with 4 outlets. The top of the chamber
was closed by a glass plate of thickness 1 cm which was held in
place by the vacuum in the chamber. Each of the 4 outlets of the
chamber was used for different purposes. These outlets are num-
bered as shown in Fig. 1. Outlet 1 was connected to the vacuum
pump with a butterfly valve. Outlet 2 was fitted with a vacuum
gauge. Outlet 3 was fitted with a release valve, which was used to
obtain the desired pressure in the enclosure while outlet 4 was
used to draw out the connecting wires from the PT-100 sensors
ð10Þ
Outlet 4
Outlet 3
Outlet 1
Outlet 2
Fig. 1. Experimental setup used in the present study.
96 K.Y. Sudharshan et al. / Solar Energy Materials & Solar Cells 157 (2016) 93–100
present in the enclosure to measure the substrate and enclosure
temperatures. The rest of the components vary based on the need
of the experiment.
The first part of the study involved temperature study of the
given solar selective substrate at different ambient pressures.
The solar selective substrate was an aluminium substrate with
a coating of α ¼0.90, ε ¼0.15. This substrate of dimension
22.7 cm  12 cm  0.2 cm and weight 13.015 g was placed inside
the enclosure described above. Two PT-100 sensors were placed in
the enclosure, one of which was clamped to substrate surface to
measure the substrate temperature while the second was used to
measure the enclosure temperature. The substrate itself was
placed on top of two C-shaped supports made of aluminium so
that the substrate was as close to the top of the enclosure as
possible. This helps to prevent wastage of surface area of absorber
due to shadows of the enclosure walls. The supports were covered
with teflon tape (a thermal insulator) to eliminate heat loss from
absorber by conduction.
The second part of the experiment which involves measuring
STEG efficiency with varying ambient pressure retains most of
setup including a few modifications. The solar selective substrate
was coupled with the thermo-electric modules which were
arranged in a single stack. The TEGs were coupled with the solar
selective substrate by the use of thermal grease which aside from
being a good conductor, eliminates air gaps which hamper thermal
conduction. The TEG end not coupled to the substrate was placed
on an aluminium fin which acts as a natural convective cooling
system for the STEG cool end. One of the PT-100 sensors was
removed and the wires were connected to the STEG system. The
free ends of these wires were drawn out through Outlet 4 and
connected to a multimeter to measure voltage and current.
3.2. Experiment
In the first part of the experiment, the readings of importance
are those of sample temperature and the enclosure temperature.
Steady state readings of these values were taking after placing the
setup in bright sunlight. These steady state values were then
compared with theoretical models made in COMSOL Multiphysics
to show concordance (explained in Section 3.3). Once the max-
imum temperature had been achieved, an opaque thermally
insulating object was placed in the path of the incident solar flux
to study the cooling curves of the setup. Temperature readings of
the substrate and enclosure were taken every 10 s. Slopes of the
plots at each temperature were also determined which were
eventually used to determine ‘h’. This experiment was done when
the pressure in the enclosure was maintained at 100, 200, 300,
400, 500, 700 and 913 mbar where 913 mbar was the ambient
pressure. In the second part, the critical readings were the values
of open circuit voltage, and power generated by the setup. Initially
the wires drawn from the hot and cold junctions of the STEG
system were left unconnected and the potential difference
between the two ends was measured by the use of a multimeter.
This is the open-circuit voltage (OCV) generated by the system.
Further experiments were performed by connecting a variable
resistance to the previously open ended wires drawn from the
STEG system. The resistance was set to match the internal resis-
tance of the modules, both of which were measured by the same
multimeter. Values of voltage and current were measured across
the resistor at steady state. The above data was then used to cal-
culate the efficiency of the entire STEG system by the formula Eq.
(14). The value of incident solar power was determined from the
product of area of the solar selective substrate (measured manu-
ally) and the average solar flux on the day of the experiment [27].
3.3. Simulations
The simulations were done on COMSOL Multiphysics 5.0 to
obtain a theoretical estimate of steady state temperature obtained
by the setup. A simplified model was created which consisted of
only the enclosure of specified dimensions and the substrate itself.
Heat transfer with surface to surface radiation was the physics
chosen for this test. This model presents an accurate enough pic-
ture of the setup, since care was taken to eliminate all other heat
loss pathways except natural convection and radiation which
exists only between the aforementioned components of the setup.
The pressure of the gas (material defined for enclosure) alone was
changed to match the enclosure pressures used in the experiment.
A fine mesh was created and the simulations were run.
4. Results and discussion
The sample in the focus of the study was an absorber coated
aluminium substrate of dimensions 22.7 cm  12 cm  0.4 cm with
absorptance 0.90 and emittance 0.15, which served as the hot end
of the STEG setup. This sample was prepared using the magnetron
sputtering technique and was found to be stable up to 450 oC in air
[15]. In order to determine a baseline substrate steady state tem-
perature, the sample was exposed to atmosphere and the max-
imum temperature attained by the sample was noted to be 67 oC.
This is the result of heat loss through radiation and convection
(natural and forced). In order to obtain a rough estimate of the
energy losses by these pathways the substrate was placed inside
an enclosure. In the absence of air currents (blocked by placing in
the enclosure), the maximum temperature reached by the sub-
strate increased to 97 oC, resulting from the elimination of forced
convection heat loss pathway. Vacuum enclosures have been used
by several groups [28–30] in their experiments to improve STEG
system efficiency, but they have yet to determine the actual impact
of enclosure pressure in this regard which is the focus of this
study. The enclosure pressures used are 100, 200, 300, 400, 500,
700 and 913 mbar in order to visualize a trend in the impact of this
particular physical parameter. From simulations in the COMSOL
model described in previous section, it was indicated that max-
imum sample temperature obtained in the setup does not surpass
150 oC. It was also noted that emittance which varies with sub-
strate temperature does not show significant change in the tem-
perature range of the sample [14]. As a result, in the calculation for
120
100 mbar
Sample Temperature (oC)
105 200 mbar
300 mbar
400 mbar
90 500 mbar
700 mbar
75
60
45
0 50 100 150 200 250 300
Time (s)
Fig. 2. Substrate temperature vs. time for various enclosure pressures.
 K.Y. Sudharshan et al. / Solar Energy Materials & Solar Cells 157 (2016) 93–100 97

Substrate Steady State Temperature ( C)

52
160

o
Enclosure Temperature (oC)

51 100 mbar Actual Ts

200 mbar Theoretical Ts
50 300 mbar
400 mbar 140
500 mbar
49
700 mbar

48
120

47

46 100

45
0 50 100 150 200 250 300 0 200 400 600 800 1000
Time (s) Pressure (mbar)
Fig. 3. Enclosure temperature vs. time for various enclosure pressures. Fig. 5. Steady state temperature of substrate (theoretical and actual) vs. enclosure
pressure.
Heat Transfer Coefficient(W/m2-oC)

4
55oC
70oC
3 80oC
90oC
2 ambient conditions do not impact the conditions inside the setup.
1 value of convective heat transfer coefficient decreases with a
0
100 200 300 400 500
Pressure (mbar)
Fig. 4. Heat transfer coefficient of convection vs. enclosure pressure at different
substrate temperatures.
‘h’, the value of ε was assumed as the same as that for room
temperature (ε ¼0.15).
Preliminary experiments were done to determine performance
of the absorber and the integrity of the enclosure. Fig. 2 shows the
cooling curves of the substrate at different enclosure pressures.
The plots show that rate of cooling decreases with a decrease in
enclosure pressure which is in line with observations made by
Hosseini and Saidi [18] in their study. In order to determine the
numerical value of ‘h’, enclosure temperature curves were mea-
sured with time as described in Fig. 3. This was done alongside the
cooling curve determination of substrate. Though the observations
in Fig. 3 do not have any direct impact on the rest of the experi-
ment, it brings an interesting observation to light. The tempera-
ture of the enclosure increases with a decrease in enclosure
pressure. Also the magnitude of this increase decreases with a
decrease in enclosure pressure. This implies that the energy lost by
the substrate is gained by the enclosure gas, which translates into
a rise in enclosure temperature, explained by the following
equations [31].
TE ¼ ESTEG þ EEnclosure
E ¼ mC p T ð16Þ
where TE is the total energy of the system and surrounding and E
is the energy of the individual components (STEG and the enclo-
sure). This energy is given by Eq. (16) where m is mass of the
system, Cp is the specific heat constant and T is system tempera-
ture. This observation can only be made if the substrate and the
enclosure are viewed as isolated system and surrounding, i.e.,
This proves the integrity of the enclosure. The values obtained in
the previous experiments were then used to determine the values
of ‘h’ at each enclosure pressure (Fig. 4). The plots show that the
decrease in enclosure pressure. It is noted that the lowest heat
transfer coefficient value is as low as 0.6 W/m2 oC at 100 mbar,
while the maximum value is nearly 3.2 W/m2 °C at 700 mbar. This
600 700 implies that for the same temperature gradient (substrate to
enclosure) there is 6 times more heat loss from the substrate at
700 mbar as compared to that at 100 mbar. The aforementioned
values of heat transfer coefficient are substantially lower com-
pared to those obtained by Hosseini and Saidi [18] for the corre-
sponding enclosure pressures. This lends credibility to the benefits
of vacuum enclosures in STEG setups.
Temperature studies were also conducted to determine the
steady state temperature of the substrate at different enclosure
pressures. Since it has already been proven that ‘h’ decreases with
a decrease in the enclosure pressure, it follows that the steady
state temperature is directly affected as shown in the heat balance
equation shown in Eq. (8). It is observed that the substrate steady
state temperature increases with a decrease in enclosure pressure.
In order to verify the experimental results, simulations were
modelled in COMSOL. A simplified model of the setup was made,
as described in the previous section. The C shaped supports used
in the setup were covered with Teflon tape (for thermal insula-
tion), hence thermal conduction losses through this pathway was
neglected. The ambient pressure property was varied from
100 mbar to 913 mbar to imitate experimental conditions. Fig. 5
shows the plot of steady state temperature (both experimental and
theoretical) of substrate against enclosure pressure. While there
are small variations, both plots show the same trend, viz., steady
state temperature of the substrate decreases with an increase in
enclosure pressure. The maximum steady state temperature
achieved was 142 °C attained at 100 mbar while the lowest was
ð15Þ 97 °C at ambient pressure of 913 mbar which in itself is a
98 K.Y. Sudharshan et al. / Solar Energy Materials & Solar Cells 157 (2016) 93–100

INCIDENT SOLAR RADIATION 900

X th = 1
750 X th = 22

Voc (mV)
600
2
3 450
7

300

4 150
0 200 400 600 800 1000
5 Pressure (mbar)
Fig. 7. Open circuit voltage vs. enclosure pressure for different thermal concentrations.
6
8
50 2.0
Fig. 6. Setup schematic: 1- Glass plate, 2- Solar selective substrate, 3- External Output Power X th = 1
circuit, 4- C-support, 5- Base support and fin, 6- Enclosure, 7- STEG modules, 8- Output Power X th = 22
Vacuum chamber. 40
Output Power (mW)
Efficiency X th = 1 1.5

staggering increase of nearly 50 °C. This maximum temperature
can be further increased by the use of optical concentration
techniques such as that described by Candadai et al. [15].
On a side note similar studies were conducted on two other
samples. One was black anodized aluminium (α ¼0.602, ε ¼0.81,
dimension: 22 cm  12 cm  0.6 cm) and the other was stainless
steel substrate coated with a selective coating of AlTiN (α ¼0.92,
ε ¼0.17, dimension: 10 cm  10 cm  0.6 cm). Aside from com-
parison purposes, the objective had been to determine a sample
suitable for the setup. Each sample presented its own challenges.
While economically more feasible, black anodized aluminium has
very poor α/ε ratio, which significantly lowers the absorber effi-
ciency. Thus the maximum temperature the black anodized alu-
minium reached was 82 °C at 100 mbar. While the AlTiN based
absorber coated sample on stainless steel has the better absorp-
tance amongst the samples, the high thermal mass proved detri-
mental. The consequence of using an absorber of high thermal
mass can be understood from the heat balance equation (Eq. (14)).
The high thermal mass implied that the sample heated up very
slowly. This appears as a substrate temperature (101 °C after
heating in 5 h of peak daylight) substantially lower than that of the
coated aluminium sheet used in the experiment which reached
142 °C (after 15 min heating in peak daylight). While the heating
of the substrate was slow, it was however noted that the sample
with higher thermal mass retained its heat longer than the sample
with low thermal mass. This shows that should optical con-
centration techniques be used to raise the substrate temperature,
samples with higher thermal mass might prove to be more useful
(by retaining temperature despite irregularities in solar condi-
tions), but from an efficiency perspective absorbers with lower
thermal mass would serve better. The above experiments aside
from attesting to the benefits of using vacuum and the integrity of
the enclosure also showed the advantage of using substrates of
lower thermal mass as shown above.
In the second part of the experiment the performance of the
STEG setup was evaluated and the impact of enclosure pressure
was studied. The schematic of the arrangement is shown in Fig. 6.
As shown in the figure, three TEG modules of Bi2Te3 were stacked
in series. This series stacking was chosen as the thermal gradient
Efficiency (%)
Efficiency X th = 22
30
1.0
20
0.5
10
0 0.0
0 200 400 600 800 1000
Pressure (mbar)
Fig. 8. Output power and efficiency at matched load condition vs. enclosure pressure.
across a single module is insufficient for power generation pur-
poses [15]. The packing was done by the means of thermal grease,
as was the coupling between the solar selective substrate and TEG
arrangement. In the study, two absorbers were used. Both were
made of the same material as that tested in the first part of the
experiment. The difference between the two absorbers was the
dimensions; one was 35 mm  35 mm, the same dimensions as
the TEG module while the other was 22.7 cm  12 cm. The use of a
large surface area absorber has shown to increase absorber tem-
perature and thermal efficiency of STEG system [14]. This is
referred to as the thermal concentration by area and the magni-
tude of thermal concentration is labelled Xth and is given by Eq.
(10). From this formula, we find that Xth of the 35 mm  35 mm
absorber to be 1 and 22 for the second absorber. Henceforth the
two samples will be denoted as Xth ¼1 and Xth ¼22, respectively.
The performance of the setup when each of the absorbers was
coupled with the stacked TEG was observed. The performance was
evaluated on two parameters, open circuit voltage and output
power. Open circuit voltage describes the maximum potential
difference created by the setup. Output power on the other hand
was determined by the Eq. (11), where voltage and current were
determined by the use of a multimeter. Current flows in the circuit
only in the presence of an external load. Maximum power is
generated in matched load condition. Hence the internal
 K.Y. Sudharshan et al. / Solar Energy Materials & Solar Cells 157 (2016) 93–100
resistance of the TEG system was measured. This internal resis-
tance was found to be 12.7 Ω. Manufacturer specifications present
resistance of individual modules as 4 Ω [32], which makes series
stacked resistance of the modules 12 Ω, while the additional
resistance can be attributed to internal resistance of connecting
wires and thermal grease. It was observed that the maximum OCV
generated by the setup using Xth ¼1 (called setup 1) was 272 mV at
100 mbar and generated only 150 mV at 913 mbar. This increase in
OCV is due to the decrease in enclosure pressure which decreases
system heat losses. This in turn can be intuitively related to the
heat transfer coefficient plots as well. Further experimentation at
different enclosure pressures can yield results which can be used
to determine a mathematical relation between OCV and enclosure
pressure. The maximum OCV generated by the setup using Xth ¼22
(called setup 2) was 823 mV at 100 mbar and approximately
600 mV at 913 mbar. These OCV values are nearly 4 times that of
corresponding OCVs from setup 1, the plots of which are shown in
Fig. 7. This demonstrates that the increase in absorber surface area
directly impacts STEG performance, proving the validity of thermal
concentration by area.
This is further validated by the graphs of output power against
enclosure pressure (Fig. 8). The data shows that the maximum
power generated by setup 1 was 3.78 mW at 100 mbar pressure
while it generated less than 1 mW at 913 mbar. This is sub-
stantially lower when compared to data obtained from setup 2,
where the power generated at 100 mbar was 44.2 mW while that
at 913 mbar was nearly 15 mW. This observed output surpasses
the 9.15 mW obtained by Emmanuel and Weiss [28] using their
setup which used no thermal concentration techniques. Better
outputs have been obtained by Amatya and Ram (1.8 W) [29] and
Pereira et al., (19 W) [33] in their respective arrangements. These
results were obtained due to the use of high optical concentration
ratios (4 100) and active cooling systems. The data obtained from
the present experiment also shows that the output power gener-
ated by setup 2 is  14 times that generated by setup 1.
System efficiencies were also determined for these setups.
System efficiency can be calculated in different ways. One method
can be determining absorber and TEG efficiencies independently,
whose product gives net efficiency. However, in this experiment
efficiency is calculated using Eq. (14). From the data obtained, the
efficiencies calculated are presented alongside output power
against enclosure pressure (Fig. 8). The maximum efficiency
achieved by the setup 1 is 1.62% at 100 mbar which is comparable
to that achieved by the optical concentration setup used by Can-
dadai et al. [15] at 66 suns. The experiments also show that effi-
ciency of the setup 2 is lower than setup 1, i.e., there is a decrease
of setup efficiency with increase in thermal concentration by area,
an observation that needs further experimentation before valida-
tion. This observation can be attributed to the following factors.
The calculation method used in this experiment to determine
efficiency is very simplistic. The primary issue lies in the way that
incident energy is calculated. In this experiment incident energy is
taken as product of incident irradiation flux and area of absorber,
i.e. Eq. (9). From Eqs. (9) and (10) it is clear that by using thermal
concentration by area, the increase in incident energy is propor-
tional to the area of the absorber. As a result, the increase in
magnitude of incident energy far outstrips the increase in output
power of the STEG system. Another problem with the above
approach is the fact that behavior of absorber and glass wall at
different wavelengths have not been taken into account. The
absorptance of the solar selective substrate (and emittance) varies
for each wavelength, which implies that the energy absorbed from
each wavelength differs and will not be equal to the incident
energy determined from the aforementioned formula. Also trans-
mittance of the glass plate used (taken as 90% from literature) also
varies with wavelength of incident light. This implies that the
99
incident energy calculated from the Eq. (9) is higher than the
actual incident energy. Also, the average incident solar flux taken
(from meteorological data) does not account for local variations or
variance in irradiation flux. These uncertainties together could
account for the low efficiencies obtained from the setup. Reme-
dying the aforementioned deficiencies could give a more accurate
picture of the system efficiency. Despite these uncertainties the
experiments have satisfied their objectives with conclusive proof
that enclosure pressure greatly impacts STEG efficiency while
simultaneously showing thermal concentration by area as an
alternate method of concentrating solar energy. Improving absor-
ber coatings and STEG materials will help greatly to improve the
STEG performance shown in this experiment. The ingenious setup
has also proven to be an economical method of harnessing solar
energy. With a small one-time investment, a maintenance free
setup can be created to generate power for micro-power
applications.
5. Conclusions
In this study, we have demonstrated the effects of enclosure
pressure and thermal concentration on the performance of the
Solar Thermo-Electric Generation system. The results of the study
show that in the substrate used in the setup (an absorber coating
on Al Alloy substrate: α ¼0.90, ε ¼0.15), the steady state tem-
perature attained was 97 °C in an enclosure as compared to
approximately 67 °C when placed outside. This steady state tem-
perature reaches 142 °C when the enclosure pressure is reduced to
100 mbar. COMSOL simulation results concur with the observa-
tions made experimentally. Cooling curves of the substrate at
different enclosure pressures (100, 200, 300, 400, 500, 700,
913 mbar) show that cooling rate of substrate decreases with a
decrease in enclosure pressure, proving the minimization of heat
loss by convection. The study also shows that decreasing enclosure
pressure decreases heat transfer coefficient of the enclosure gas,
which leads to the lower cooling rates observed in the substrate.
Experiments show that convective heat transfer coefficient as low
as 0.6 W/m2 °C (at 100 mbar) is obtained, proving the integrity
and effectiveness of the setup in minimizing convection losses.
A comprehensive study was also conducted to show the effect
of enclosure pressure and thermal concentration on open circuit
voltage, output power generated and efficiency of the STEG sys-
tem. The study shows that by increasing the area of the absorber
22 times, the magnitude of open circuit voltage is quadrupled,
proving the viability of using thermal concentration by area as a
method of solar energy concentration. This is further proved by
the results of the output power study which shows that output
power is multiplied nearly 14 under the conditions mentioned
above. It is also observed that decreasing enclosure pressure
shows an improvement in output power and open circuit voltage
in both the Xth ¼1 and Xth ¼22 setups. However, the magnitude of
increase in both output power and open circuit voltage is higher in
the setup involving thermal concentration. The results show that
the thermal concentration based STEGs, without tracking system,
may be used as an alternative to optically concentrated STEGs for
micro-power generation applications in remote areas.
Acknowledgements
Ms. Jyoti and Mr. Prabanjan Kulkarni are thanked for their help in
getting absorptance data and helping set up the electrical circuits,
respectively. This work is partially supported by CSIR under TAPSUN
Program (Project # NWP 0054) and BRNS (Project # U-1-125).
100 K.Y. Sudharshan et al. / Solar Energy Materials & Solar Cells 157 (2016) 93–100
References
[1] Anke Weidenkaff, Matthias Trottmann, Petr Tomes, Clemens Suter, Aldo
Steinfeld, Angelika Veziridis, Solar TE converter applications: Thermoelectric
Nanomaterials, vol. 182, Springer Series in Material Sciences, pp 365–382.
[2] B. Parida, S. Iniyan, R. Goic, A review of solar photovoltaic technologies, Renew.
Sustain. Energy Rev. 15 (2011) 1625–1636.
[3] M. Ortega, P. d elRío, E.A. Montero, Assessing the benefits and costs of
renewable electricity, Renew. Sustain. Energy Rev. 27 (2013) 294–304.
[4] J.A. Duffie, W.A. Beckman, Solar Engineering of Thermal Processes, 2nd edition,
Wiley-Interscience, New York, 1991.
[5] H.S. Lloyd, Solar thermoelectric generators, 1961, Patent No. US 2,984,696.
[6] Qing.-Jie. Zhang, Xin.-Feng. Tang, Peng.-Cheng. Zhai, Masayuki Niino,
Cyoji Endo, Recent development in nano and graded thermoelectric materials,
Mater. Sci. Forum 492–493 (2005) 135–140.
[7] P. Lee, L. Cai, P. Zhai, X. Tang, Q. Zhang, M. Niino, Design of a concentration
solar thermoelectric generator, J. Electron. Mater. 39 (2010) 1522–1530.
[8] J. Xiao, T. Yang, P. Li, P. Zhai, Q. Zhang, Thermal design and management for
performance optimization, Appl. Energy 93 (2012) 33–38.
[9] Jincan Chen, Thermodynamic analysis of a solar driven thermoelectric gen-
erator, J. Appl. Phys. 79 (1996) 2717–2721.
[10] M. Telkes, Solar thermoelectric generators, J. Appl. Phys. 265 (1954) 765–777.
[11] A. Nattestad, A.J. Mozer, M.K.R. Fischer, Y.B. Cheng, A. Mishra, P. Bäuerle, U. Bach,
Electrochemical characterization of rapid discharge sintering (RDS) NiO cathodes
for dye-sensitized solar cells of p-type, Nat. Mater. 9 (2010) 31–35.
[12] M.L. Olsen, E.L. Warren, P.A. Parilla, E.S. Toberer, C.E. Kennedy, G.J. Snyder, S.
A. Firdosy, B. Nesmith, A. Zakutayev, A. Goodrich, C.S. Turchi, J. Netter, M.
H. Gray, P.F. Ndione, R. Tirawat, L.L. Baranowski, A. Gray, D.S. Ginley, A high
temperature high efficiency solar thermoelectric generator prototype, Energy
Procedia 49 (2014) 1460–1469.
[13] Warren J. Smith, Modern Lens Design, 2nd edition, Mcgraw-Hill Professional,
New York, 2005.
[14] D. Kraemer, B. Poudel, H. Feng, J.C. Caylor, Bo Yu, X. Yan, Yi Ma, X. Wang,
D. Wang, A. Muto, K. McEnaney, M. Chiesa, Z. Ren, G. Chen, High-performance
flat-panel solar thermoelectric generators with high thermal concentration,
Nat. Mater. 10 (2011) 532–538.
[15] Aaditya A. Candadai, V. Praveen Kumar, C. Barshilia, Performance evaluation of
a natural convective cooled concentration solar thermo-electric generator
coupled with spectrally selective high temperature absorber coating, Sol.
Energy Mat. Sol. Cell 145 (2016) 333–334.
[16] Ning Wan, Li Han, Hongcai He, Nam.-Hee. Park, Kunihito Koumoto, A novel
high performance photovoltaic-thermoelectric hybrid device, Energy Environ.
Sci. 4 (2011) 3676–3679.
[17] M. Mizoshiri, M. Mikami, K. Ozaki, Flexible thin-film thermoelectric generator
inserting CR buffer layer, in: Proceedings of the Power MEMS, Atlanta, GA,
USA, 2012, pp. 2–5.
[18] Reza Hosseini, Maysam Saidi, Experimental study of air pressure effects on
natural convection from a horizontal cylinder, Heat Transf. Eng. 33 (2012)
878–884.
[19] Ralph Stair, Russell G. Johnston, Thomas C. Bagg, Spectral distribution of
energy from the sun, J. Res. Natl. Bur. Stand. 53 (2) (1954) 113–119.
[20] Viorel Badescu, J. Phys. D. : Appl. Phys. 38 (2005) 2166–2172.
[21] Viorel Badescu, Renew. Energy 30 (2005) 211.
[22] Lauryn L. Baranowski, G. Jeffrey Snyder, Eric S. Toberer, Concentrated solar
thermoelectric generator, Energy Environ. Sci. 5 (2012) 9055–9067.
[23] P.K. Nag, Engineering Thermodynamics, 5th edition, Tata Mcgraw-Hill, New
Delhi, 2013.
[24] J.P. Holman, Souvik Bhattacharya, Heat Transfer, 10th edition, Mcgraw-Hill,
India, 2010.
[25] Sarah Watzman, Design of a Solar Thermoelectric Generator (Ph.D.thesis),
Ohio State University, 2013.
[26] D.K. Paul, P. Gardner, Maximum power transfer theorem: a simplified
approach, Int. J. Electr. Eng. Educ. 35 (1998) 271–273.
[27] Synergy Environs, 〈http://www.synergyenviron.com/tools/solar_insolation.
asp〉.
[28] E. Ogbonnaya, L. Weiss, Micro Solar Thermal Power harvesting Using Ther-
moelectric Generator, in: Technical Proceedings of 38th Annual Convention on
National Society of Black Engineers, PA, 2012.
[29] R. Amatya, R.J. Ram, Solar thermoelectric generator for micro-power appli-
cations, J. Electron. Mater. 39 (2010) 1735–1740.
[30] K. McEnaney, D. Kraemer, Z. Ren, G. Chen, Modeling of concentrating solar
thermoelectric generators, J. Appl. Phys. 110 (2011) 1–6.
[31] R.E. Sonntag, Claus Borgnakke, Fundamentals of Thermodynamics, 7th edition,
Wiley Publications, Hoboken, NJ, 2008.
[32] Hi-Z Technology Inc., 〈www.hi-z.com〉.
[33] A. Pereira, T. Caroff, G. Lorin, T. Baffie, K. Romanjek, S. Vesin, K. Kusiaku,
H. Duchemin, V. Salvador, N. Miloud-Ali, L. Axila, J. Simon, High temperature
solar thermoelectric generator – Indoor characterization method and mod-
eling, Energy 84 (2015) 485–492.