Professional Documents
Culture Documents
Email: hosung.lee@wmich.edu
Abstract
This work examines the validity of formulating the effective thermoelectric material
cooler were formulated on the basis of the hot junction temperature. Then, the effective
were defined in terms of the three maximum parameters that were taken from either a
the simple standard equation with the effective material properties predicts well the
over the maximum parameters were also formulated to present the characteristics of the
2
thermoelectric coolers along with the normalized charts. The normalized charts would be
Nomenclature
T temperature (°C)
Greek symbols
Subscript
p p-type element
n n-type element
Superscript
* effective quantity
4
1. Introduction
Thermoelectric coolers have comprehensive applications [1-5] in electronic devices,
phenomena are often described by a simple standard equation, which has been widely
used in the literature [1-5], sometimes in good agreement with experiment [6,9,11]. The
simple standard equation is herein called the ideal equation, which is virtually formulated
under three assumptions that the electrical and thermal contact resistances, the Thomson
effect (temperature-dependent Seebeck coefficient), and the radiation and convection heat
transfer are negligible [11,12]. The radiation and convection heat transfer is small for the
moderate temperature differences between the hot and cold junction temperatures and the
rather slightly improves the performance [6,7]. The major errors between the
measurements and the ideal equation lie on the electrical and thermal contact resistances
[11,13].
parallel between two ceramic plates. The manufacturers usually provide the performance
curves along with the maximum parameters such as the temperature difference Tmax, the
current Imax, the cooling power Qmax, and the voltage Vmax. However, the material
properties of the modules such as the Seebeck coefficient , the electrical resistivity ,
and the thermal conductivity k are not usually provided as manufacturers’ proprietary
information. Therefore, system designers find it difficult to obtain the material properties.
5
Huang et al. [14] measured the material properties of a commercial module for the
optimum design using an evacuated and insulated test apparatus. They confirmed that the
measurements were in good agreement with the performance curves provided by the
manufacturer. Nevertheless, they were not able to fit the measured data to the ideal
equation, which was deemed mainly due to the electrical and thermal contact resistances
from the ideal equation using the definitions used by manufacturers and then expressed
the physical module properties (m, Rm, and Km) in terms of the three parameters (Tmax,
Imax, and Vmax) out of the four theoretical maximum parameters (Tmax, Imax, Qmax, and
Vmax). In this way, the module properties contain information of the number of
thermoelement couples and geometric ratio. Lineykin and Ben-Yaakov extracted the
the theoretical maximum parameters. Luo [16] used two methods to determine the
physical module properties: not only the combination (Tmax, Imax, and Vmax) used by
Lineykin and Ben-Yaakov, but also the different combination (Tmax, Imax, and Qmax) out
of the four maximum parameters. When the two methods were compared to each other
over four selected commercial modules, the physical module properties over the four
modules varied within a 5% discrepancy range. Zhang [17] obtained the physical module
properties of a commercial module using the three parameters (Tmax, Imax, and Vmax) for
Simons [21] showed a capability of the module properties (m, Rm, and Km) with a
different set of the maximum parameters (Tmax, Imax, and Qmax) to predict the
performance of an electronic module. Tan and Fok [10] evaluated the module properties
for commercial modules using the three maximum parameters (Tmax, Imax, and Vmax) and
compared the predicted results with the manufacturers’ performance curves. The
comparisons showed fair agreement and the errors increased with increasing current or
temperature difference. Recently, Ahiska and Ahiska [18] developed a new economic
based on the ideal equation by measuring the maximum parameters (Tmax, Imax, and
Most of the above mentioned works tried to extract the physical module properties
(m, Rm, and Km) from either the three parameters (Tmax, Imax, and Vmax) or the other set
of parameters (Tmax, Imax, and Qmax), which imposes the uncertainties on the cooling
power prediction. On the other hand, the present work extracts the effective material
properties (and k) from the manufacturers’ maximum parameters (Tmax, Imax, and
Qmax), which imposes the uncertainties on the voltage prediction, particularly being good
at module design for specific systems. Note that, although the differences between the
module properties and the effective material properties appear minuscule, the results and
applications are of great consequence: the module properties are constrained to have a
validity for use of the module, but the effective material properties are not, which is the
uniqueness of the present paper. The optimal design [8] using the ideal equation with the
effective material properties will now be simple and robust. The present work studies to
verify in detail the effective material properties comparing with the performance curves
7
dependence of the material properties and the subtle thermal and electrical contact
resistances can be examined with the effective material properties that are constant.
The normalized charts with the maximum parameters were presented by Buist [19]
and later Uemura [20] for the purpose of the design of thermoelectric devices. However,
theoretical formulas for the normalized charts were not found to the authors’ knowledge.
Therefore, the present work studies the normalized formulas over the maximum
parameters providing the two normalized charts, which coherently reveal the elusive
Heat Absorbed
p
n
p
n-type Semiconductor n
n
p Positive (+)
p n
p-type Semiconcuctor p
1. Ideal Equation
8
A typical thermoelectric module is shown in Figure 1. Suppose that the upper junction
1
Q c nTc I I 2 R KT (1)
2
where n is the number of thermocouples, a the Seebeck coefficient, I the current, R the
electrical resistance, K the thermal conductance, and T = (Th - Tc). From now on,
Equation (1) is called the ideal equation. The current for the optimum COP can be
T
I COP
R 1 ZT 1 (2)
T
ZT ZTh 1 (3)
2Th
2. Maximum Parameters
Let us consider a thermoelectric module shown in Figure 1 for the theoretical
maximum parameters with the ideal equation. The module consists of a number of
thermoelement couples as shown. As mentioned before, the ideal equation assumes that
there are no the electrical and thermal contact resistances, no Thomson effect, and no
radiation or convection. It is noted that the theoretical maximum parameters might differ
with the manufacturers’ maximum parameters that are usually obtained by measurement.
9
The maximum current Imax is the current that produces the maximum possible
temperature difference Tmax , which always occurs when the cooling power is at zero.
This is obtained by setting Q c = 0 in Equation (1), replacing Tc with (Th – T) and taking
derivative of T with respect to I and setting it to zero. The maximum current is finally
expressed by
1
2
1
Th Th
2
I max (4)
R Z Z
Th Tmax
I max (5)
R
The maximum temperature difference Tmax always occurs when the cooling power
(1), substituting both I and Tc by Imax and Th – Tmax, respectively, and solving for Tmax.
2
1 1 2
Tmax Th Th Th (6)
Z Z
The maximum cooling power Q c max is the maximum thermal load which occurs at
T = 0 and I = Imax. This can be obtained by substituting both I and Tc in Equation (1) by
Imax and Th (since Tc = Th ), respectively, and solving for Q c max . The maximum cooling
Q c max
n 2 Th Tmax
2 2
(7)
2R
The maximum voltage is the DC voltage which delivers the maximum possible
3. Normalized Parameters
If we divide the actual values by the maximum values, we can normalize the
1
n Th T I I 2 R KT
Q c 2
Q c max
n Th Tmax 2 R
2 2 2
(9)
which, in terms of the normalized current and normalized temperature difference, reduces
to
2
T Tmax I Tmax I T Tmax
21 1 2
Q c Tmax Th I max Th I max Tmax Th
(10)
Q c max T T T 2
1 max 1 max ZTh 1 max
Th Th
Th
where
11
2
Tmax 1 1
1 1 1 (11)
Th ZTh ZTh
T Tmax
2
T Tmax I 1 T I Tmax Th
1 1 max
Tmax Th I max 2 Th I max T
ZTh 1 max
Th
COP 2
(12)
T Tmax I T I
1 max
Tmax Th I max Th I max
V T Tmax Tmax I
1 (13)
Vmax Tmax Th Th I max
The normalized current for the optimum COP is obtained from Equation (2).
T Tmax
I COP Tmax Th
(14)
I max Tmax
1
T
1 ZT 1
h
1 T T
ZT ZTh 1 max
(15)
2 Th Tmax
Note that the above normalized values in Equations (10), (11), (12) and (13) are functions
measurements. The effective figure of merit is obtained from Equation (6), which is
2Tmax
Z (16)
Th Tmax 2
The effective Seebeck coefficient is obtained using Equations (5) and (7), which is
2Q c max
(17)
nI max Th Tmax
The effective electrical resistivity can be obtained using Equation (5), which is
Th Tmax A L
(18)
I max
13
2
k (19)
Z
The effective material properties include all the losses such as the contact resistances.
Hence, the effective figure of merit appears slightly less than the intrinsic figure of merit
as shown in Table 1. Since the material properties were obtained for a p-type and n-type
dividing by 2.
were extracted from a combination (Tmax, Imax, and Vmax) among the four manufacturer’s
maximum parameters (Tmax, Imax, Q c max , and Vmax). In the present work, the effective
material properties (, , and k*) were extracted from a different combination (Tmax,
Imax, and Q c max ) among the manufacturers’ maximum parameters. Therefore, the physical
module properties hold information of the number of thermoelement couples and the
In order to examine the status of the ideal equation with the effective material
properties, several major manufacturers were chosen as shown in Table 1. The effective
material properties were first calculated using the manufacturer’s maximum parameters
using Equations (16) - (19). The geometry of A and L were actually measured. Only one
14
set of the intrinsic material properties was provided by the manufacturer, which is shown
in Table 1.
appear very close to the intrinsic material properties. It should be noted that the
dimensionless intrinsic figure of merit of 0.803 exhibits slightly larger than the
dimensionless effective figure of merit of 0.744, which is reasonable because the contact
improvement was found even though the intrinsic material properties were used in
calculation because the contact resistances exist anyway. It is noted that the maximum
one another. The manufacturability and contact resistances may be responsible for the
inconsistency.
Table 1 Comparison of the properties and dimensions for the commercial products of
thermoelectric modules
Figure 2 depicts comparison between the calculations (solid lines) and the
in Figure 2 (a) was calculated using Equation (7) by substituting Tc by (Th –T) and using
the effective material properties. The dotted curve indicates the cooling power at the
optimal COP for which Equation (7) was used with substituting I by the current at the
optimum COP in Equation (2). It is seen in Figure 2 (a) that the calculated effective
maximum parameters (Tmax = 67°C, Imax = 3.9 A, and Q c max = 34.3 W) are in good
agreement with the manufacturer’s performance curves. On the other hand, Figure 2 (b)
decreasing the temperature difference. The errors are associated with the combination
(Tmax, Imax, and Q c max ) and partially the inherent contact resistances. The analysis
including the temperature dependence of the material properties and the thermal and
electrical contact resistances are very formidable especially for optimal system design
which requires many iterations of calculations, also not available in the literature to the
authors’ knowledge. Therefore, the errors are a nature of this work. The present work
presents a single module, but system design involves multiple modules. However, the
technique [22]. The marked data of the COP in Figure 2 (c) were not provided by the
manufacturer but generated in this work using the measured data in Figures 2 (a) and (b),
35
I = 3.9 A Prediction
30 Commercial product
3.2 A Optimal COP
Cooling Power, Qc (W)
25
2.4 A
20
1.6 A
15
10
0.8 A
5
0
0 10 20 30 40 50 60 70
10 2.4 A
8 1.6 A
6 0.8 A
0
0 10 20 30 40 50 60 70
3
Prediction
Commercial product
2.5
T = 10°C
2
COP
1.5 20 °C
1
30 °C
0.5
40 °C
50 °C
0
0 1 2 3 4
Current (A)
(c)
Figure 2. (a) Cooling power versus T, (b) Voltage versus T, as a function of current,
and (c) COP versus current as a function of T. The original performance data (triangles)
of the commercial module (Module CP10-127-05) are compared to the prediction (solid
lines). The dotted line in (a) indicates the cooling powers at the optimum COP.
Figures 3 (a) and (b) depict comparison between the calculations and the
performance data of Module RC12-4. In general, the calculations are in good agreement
with the manufacturer’s performance data. Figure 3 (c) shows only the calculations (solid
lines) wherein the COP data (dotted lines) were not able to be generated due to the lack
70
Prediction Qc = 0 W
Commercial product
60
Qc = 5 W
50
Qc = 10 W
40 Qc = 15 W
T (°C)
30 Qc = 20 W
Qc = 25 W
20
Qc = 30 W
10
Qc = 35 W
0
0 0.5 1 1.5 2 2.5 3 3.5 4
I (A)
(a)
30
Prediction, Qc = 0
Prediction, T = 0
25 Comm. product, Qc = 0
Comm. product, T = 0
20
Voltage (V)
15
Qc = 0
10
T 0
0
0 1 2 3 4 5
Current (A)
(b)
19
3
Prediction
No commercial product
2.5
T = 13.2°C
2
COP
1.5
1 26.4°C
0.5 39.6°C
52.8°C
0
0 1 2 3 4
Current (A)
Figure 3. (a) T versus current as a function of cooling powers, (b) Voltage versus
current for Qc = 0 and T = 0, respectively, and (c) COP versus current. The original
performance data (triangles and squares) in (a) and (b) of the commercial module
Figures 4 (a), (b) and (c) depict comparison between the calculations and the
35
Prediction
I = 3.6 A
30 Commercial product
Optimal COP
Cooling Power, Qc (W)
25 2.7 A
20
1.8 A
15
0.9 A
10
0
0 10 20 30 40 50 60 70
10
Voltage (V)
4
Prediction
2 Commercial product
0
0 1 2 3 4
Current (A)
(b)
21
2
Prediction
Commercial product
T = 17.25°C
1.6
1.2
COP
0.8
34.5°C
0.4
51.75°C
0
0 1 2 3 4
Current (A)
(c)
Figure 4 (a) Cooling power versus T as a function of current, (b) Voltage versus current
as a function of T, and (c) COP versus current as a function of T. The original
compared to the prediction (solid lines). The dotted line in (a) indicates the cooling
Figures 5 (a), (b) and (c) depicts comparison between the calculations and the
35
I = 3.5 A
Prediction
30 Commercial product
3A
Optimal COP
2.5 A
Cooling Power, Qc (W)
25
2A
20
1.5 A
15
1A
10
0
0 10 20 30 40 50 60 70
(a)
16
I = 3.5 A
14
3A
12
2.5 A
10
Voltage (V)
2A
8 1.5 A
6
1A
4
2 Prediction
Commercial product
0
0 10 20 30 40 50 60 70
3
T = 10°C Prediction
Commercial product
2.5
2
COP
1.5
20°C
1
30°C
0.5 40°C
50°C
0
0 1 2 3 4
Current (A)
(c)
Figure 5. (a) Cooling power versus T as a function of current, (b) voltage versus T, as
a function of current, and (c) COP versus current as a function of T. The original
compared to the prediction (solid lines). The dotted line in (a) indicates the cooling
1 1
I/Imax = 1.0 I/Imax = 1.0
0.9 0.9
0.8 0.8
0.8
0.8
0.7 0.7
0.5 0.5
Qc/Qcmax 0.4 V/Vmax
0.4
0.4 0.4
0.2
0.3 0.3
0.2
0.2 0.2
0.1 0.1
0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
T/ Tmax
Figure 6. Normalized chart I: cooling power and voltage versus T as a function of
current. The solid lines depict the data at ZTh = 0.75, while the dotted lines depict the
alternate current ratios at ZTh = 0.4. The dashed line depicts the cooling power ratios at
1 3
T/ Tmax = 0 T/ Tmax = 0
0.9
2.5
0.8
0.2 0.2
0.7
2
0.6 0.4
0.4 0.6
0.4 1
0.3
0.2 0.8
0.6 0.5
0.1
0.8
0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
I/Imax
Figure 7. Normalized chart II: cooling power and COP versus current as a function of T.
The solid lines depict the data at ZTh = 0.7, while the dotted lines depict the alternate
Figures 6 and 7 depict the normalized cooling power Q c Q c max , COP, and voltage
V Vmax , which were plotted using Equations (27), (29), and (30), respectively. The above
three dependent parameters are only functions of three independent parameters: I/Imax,
T/Tmax, and ZTh as shown in the equations. From the previous discussion, we learned
that the ideal equation with the effective material properties predicts well the real
predict well the performance with a given ZTh. The solid lines depict the predictions at
ZTh = 0.75, which is a typical dimensionless figure of merit used in the commercial
26
products. In order to see the effect of ZTh, the predictions at ZTh = 0.4 were plotted as the
dotted lines for comparison. We find from the figures that the normalized charts are not
significantly influenced by ZTh. These charts are then considered being universal to
present market. With inserting the maximum parameters provided by the manufacturers
into the charts, the reasonable cooling power, COP and voltage could be obtained as
6. Conclusions
The accuracy of the ideal equation in connection with the effective material
(which are usually obtained by the measurements), being in good agreement. Usually, the
properties and the electrical and thermal contact resistances, is very formidable. However,
when one uses the ideal equation with the effective material properties for moderate
temperature differences, the analysis becomes simple and robust and could be a platform
Normalized charts are constructed using the ideal equation and the maximum
parameters defined in this work. The normalized charts I and II represent well the
References
[1] A.F. Ioffe, Semiconductor thermoelements and thermoelectric cooling, Infoserch
[2] D.M. Rowe, CRC Handbook of Thermoelectrics, CRC Press, Boca Raton, FL, USA,
(1995).
(2010).
[4] Nolas GS, Sharp J, Goldsmid HJ, Thermoelectrics, 2001, Springer, Heidelberg,
Germany.
[5] H. Lee, Thermal design: heat sinks, thermoelectrics, heat pipes, compact heat
exchangers, and solar cells, John Wiley & Sons, Inc., Hoboken, New Jersey, USA,
(2010).
[6] C.Y. Du, C.D. Wen, Experimental investigation and numerical analysis for one-stage
thermoelectric cooler considering Thomson effect, Int. J. Heat Mass Transfer, 54: 4875-
4884, (2011).
[7] H. Leea, The Thomson effect and the ideal equation on thermoelectric coolers, Energy,
[8] H. Leeb, Optimal Design of thermoelectric devices with dimensional analysis, Appled
[10] F.L. Tan, S.C. Fok, Methodology on sizing and selecting thermoelectric cooler from
[11] S.A. Omer, D.G. Infield, Design optimization of thermoelectric devices for solar
power generation, Solar Energy Materials and Solar Cells, 53, 67-82 (1998).
[12] M.J. Huang, R.H. Yen, A.B. Wang, The influence of the Thomson effect on the
performance of a thermoelectric cooler, Int. J. Heat and Mass Transfer, 48, 413-418
(2005).
[13] G. Min, D.M. Rowe, O. Assis, S.G.K. Williams, Determining the electrical and
[14] B.J. Huang, C.J. Chin, C.L. Duang, A design method of thermoelectric cooler,
cooler from vendor datasheets, Electronics Cooling Magazine, 14, 3, 22-27 (2008).
[17] H.Y. Zhang, A general approach in evaluating and optimizing thermoelectric coolers,
[19] R.J. Buist, Universal thermoelectric design curves, 15th International Energy
[21] R. Simons, Calculation corner: Using vendor data to estimate thermoelectric module
4-6 (2010).
[22] A. Attar, H. Lee, and S. Weera, Optimal design of automotive thermoelectric air