Proceedings of HT2005 2005 ASME Summer Heat Transfer Conference July 17-22, 2005, San Francisco, California, USA

DRAFT

HT2005-72506

THREE-DIMENSIONAL ANALYSIS ON THE PERFORMANCE OF THE THERMOELECTRIC MICRO-COOLER

Kong Hoon Lee and Ook Joong Kim Thermo-uid System Department Korea Institute of Machinery and Materials Daejeon 305-343, Korea konghoon@kimm.re.kr, ojkim@kimm.re.kr

ABSTRACT Three-dimensional numerical analysis has been carried out using the FEMLAB software package to gure out the performance of the thermoelectric micro-cooler. A small-size and column-type thermoelectric cooler is considered and Bi2 Te3 and Sb2 Te3 are selected as the n- and p-type thermoelectric materials, respectively. The thickness of the thermoelectric element considered is 5 to 20 m and the thickness affects the performance of the cooler. The effect of parameters such as the temperature difference, the current, and the thickness of the thermoelectric element on the performance of the cooler has also been investigated. The coefcient of performance (COP) is the primary factor to evaluate the performance of the cooler and the COP varies with the parameters. The COP has the maximum value at a certain current and the value decreases with the temperature difference or the thickness. The predicted results also show that the performance can be improved for the thick thermoelectric element at the small temperature difference and the small current. INTRODUCTION Thermoelectric devices are used for both cooling and power generation using the Peltier and Seebeck effects, respectively [1]. The thermoelectric device contains over hundreds of n-p couples connected electrically in series, but thermally in parallel between two planar substrates. Because the Peltier and Seebeck effects are directly related, the best materials for the cooling are also optimized for power generation. Near room temperature,

the most efcient materials are heavily doped p-type and n-type (Bi,Sb)2 Te3 . Advance in the fabrication technique with microelectromechanical systems (MEMS) technology has made it possible to fabricate a lot of microdevices. Some of such devices require the precise thermal management and the compact cooling system for effective cooling within a small volume. The thermoelectric micro-cooler has become a promising candidate due to its cooling power density higher than that of the conventional bulk cooler [2,3]. Thermoelectric micro-cooler can easily be integrated and it is a suitable technique for the effective cooling of such devices as it does not have any moving parts. The thermoelectric micro-cooler requires a structure difcult to produce with conventional techniques and the fabrication with MEMS techniques has been attempted. Yao et al. [4, 5] developed the analytical model for the in-plain type thermoelectric micro-cooler and fabricated the cooler with both Si/Ge superlattice and Bi2 Te3 for sopt cooling. Snyder et al. [6] reported that they fabricated the thermoelectric microdevice using the electrochemical deposition with the photoresist mould. They used 400-m-thick oxidized silicon as a substrate and formed 20-mthick thermoelectric elements (Bi2 Te3 for p-type and Sb2 Te3 for n-type) B ttner et al. [7] fabricated the micro-cooler using their o new fabrication technique using the 4 inch wafer and reported that a net cooling of about 11 K was achieved using 800 mA with 3 p/n-junctions. da Silva and Kaviany [8] carried out the one-dimensional theoretical analysis for the column-type thermoelectric micro-cooler to be used in the cooling of the vapor sensor. They also fabricated the micro-cooler based on the anal-

Address

all correspondence to this author.

Copyright c 2005 by ASME

ysis and measured the performance of the cooler [9]. In order to gure out the performance of the thermoelectric cooler the three-dimensional numerical analysis has been carried out. The development of the model follows the one-dimensional approach of da Silva and Kaviany [8] and extends it to threedimensional situation. The FEMLAB1 is utilized in the analysis, which is a software package applicable to the multiphysics problems. The small-size and column-type thermoelectric cooler is considered. It is known that tellurium compounds currently have the highest cooling performance around room temperature. Thus, in the present study, Bi2 Te3 and Sb2 Te3 are selected as the n- and p-type thermoelectric materials, respectively. The thermoelectric element considered has the thickness of 5 to 20 m. The thickness of the element affects the thermal and electrical transport through the interface between the element and metal connector. The effect of parameters such as the temperature difference, the current, and the thickness of the thermoelectric element on the performance of the micro-cooler has also been investigated. The coefcient of performance (COP) is the primary factor to evaluate the performance of the cooler and the COP varies with the parameters. NOMENCLATURE Ate cross-sectional area of the thermoelectric element (m2 ) COP coefcient of performance Je electrical current (A) je electrical current density (A/m2 ) k thermal conductivity (W/mK) L thickness (m) P strength of electron/hole-phonon interaction Pe electrical power (W) Qc cooling rate at the cold-side substrate (W) Qh heat release rate at the hot-side substrate (W) q heat ux (W/m2 ) Re electrical resistance () Rc conduction resistance (K/W) T temperature ( C or K) V voltage (V) bulk bulk Seebeck coefcient (V/K) b boundary Seebeck coefcient (V/K) e electrical resistivity (m) Subscripts c cold, contact cold cold substrate e electron h hot, contact hot hot substrate p phonon te thermoelectric element ANALYSIS The physical domain is composed of two Silicon substrates and six thermoelectric elements sandwiched between two sub-

m 700

600 m

200 m
200 m

ate ubstr ide s old s C

ide Hot s
z y x Hot side substrate

trate subs

200 m

Cold side substrate

Fig. 1.

Schematic of the thermoelectric cooler.

strates as shown in Fig. 1. The size of each substrate is 700 m 600 m 200 m. The thermoelectric element is 20 m thick and is located between the top and bottom metal connectors of 2 m thick. The analysis is carried out with the different governing equations for the substrates and thermoelectric elements, respectively. Heat transfer in the materials which do not have the thermoelectric features is accomplished by conduction of free electrons. However, in the thermoelectric materials, the electron-phonon thermal non-equilibrium is caused near the boundary due to the phonon and electron boundary resistances at the metal and thermoelectric interface [8]. The energy conversion mechanisms such as the Joule heating and Peltier cooling/heating will also contribute to this non-equilibrium [10]. Model for the Thermoelectric Leg The model development follows the one-dimensional approach of da Silva and Kaviany [8] and extends it to threedimensional situation. The governing equations are obtained from the kinetics of electrons and phonons in an electric eld or an temperature eld which is described by the coupled Boltzmann equations for electrons and phonons [8,10,11]. The equations are generally written as (k p Tp ) = P(Te Tp ), (1)
200 m

2 (ke Te ) = e je P(Te Tp ),
1 See

200 m

2 m

Thermoelectric element

Metal connector

20 m

2 m

http://www.comsol.com

Copyright c 2005 by ASME

200 m

(2)

24 m

where je is the electrical current density. P represents the strength of the electron/hole-phonon interaction, which is obtained with the density of electrons/holes, Boltzmann constant, and electron/hole energy relaxation time [8, 10]. The electrical and thermal contact resistances at the interface between the thermoelectric element and the metal connector are considered in the model. The contact resistances cause the temperature jump through the interface and this phenomenon is treated with the interface conditions considering the resistances. The interface conditions for phonon conduction are as follows: Th Tp (Ate Rc )b,p Tp = k p z

The interface condition between the hot-side connector and the thermoelectric element is khot T z =
te L2

Th Tp (Ate Rc )b,p

+
te L2

Th Te (Ate Rc )b,e

,
te L2

(8)

and the condition in between the cold-side connector and the thermoelectric element is kcold T z =
Lte 2

Tp Tc (Ate Rc )b,p

+
Lte 2

Te Tc (Ate Rc )b,e

.
Lte 2

(9)

,
te L2

(3)

te L2

In order to solve Eq. (7), adiabatic boundary conditions are applied to the surfaces except for the top and bottom surfaces. (4) Model for the Thermal Equilibrium The bulk type thermoelectric cooler is generally studied with the thermal equilibrium conditions [12]. In such a situation, the thermal conduction within thermoelectric elements is analyzed with one equation for the equilibrium temperature. The equations for thermoelectric elements can be derived with the equilibrium temperature, T (kT = k p Tp + ke Te ), from the equations described above. Thus, the energy equations (1) and (2) are written as:
2 (kT ) = e je .

Tp Tc (Ate Rc )b,p

Lte 2

= k p

Tp z

.
Lte 2

The interface conditions for electron conduction are as follows: Th Te (Ate Rc )b,e Te z + je Te | Lte (Ate Re )b
2

te L2

= ke

te L2

2 je , 2 (5)

(10)

Te Tc (Ate Rc )b,e

Lte 2

= ke

Te z

Lte 2

+ je Te | Lte + (Ate Re )b
2

2 je

, (6)

The interface conditions for the thermoelectric elements is similarly written as follows: Th Tte (Ate Rc )b T z + je Tte | Lte (Ate Re )b
2

where = bulk b and both bulk and b are given in Table 1. The terms in the left hand side of the above equations represent the heat ow as dened by the phonon (electron) boundary resistance. The second terms in the right hand side of Eqs. (5) and (6) represent the Peltier cooling and heating, respectively. The third terms represent the Joule heating at the interfaces. Adiabatic boundary conditions are applied to all the boundary except for the interfaces indicated in Eqs. (3) through (6). Model for the Substrate and the Metal Connector Thermal energy is transferred only by conduction in the substrates and metal connectors and is described by Fouriers law.
2 (ki T ) = c je ,

= k
te L2

te L2

2 je , 2 (11)

Tte Tc (Ate Rc )b

= k
Lte 2

T z

Lte 2

+ je Tte | Lte + (Ate Re )b

2

2 je , (12) 2

and for the metal connectors, khot T z =

te L2

Th Tte (Ate Rc )b

,
te L2

(13)

(7) kcold T z Tte Tc (Ate Rc )b

where ki is kc for the metal connector, kcold for the cold-side (top) substrate and khot for the hot-side (bottom) substrate. The term related to the Joule heating of the right hand side is considered only in the metal connectors and not in the substrates since the substrates should be insulated for the electricity. The boundary conditions at the cold-side and hot-side surfaces will be the constant temperature or heat ux conditions.

=
Lte 2

.
Lte 2

(14)

SOLUTION METHOD The equations described in the previous section are solved using the FEMLAB which is the commercial software package

Copyright c 2005 by ASME

Table 1. Properties of (a) n and ptype thermoelectric elements (b) metal connectors and Silicon substrates (a) Property Unit n-type (Bi2 Te3 ) p-type (Sb2 Te3 ) P bulk b e (Ate Re )b kp (Ate Rc )b,p ke (Ate Rc )b,e W/m3 K V/K V/K m m2 W/mK K/(W/m2 ) W/mK K/(W/m2 ) Unit m W/mK 8.61 1013 228 106 187 106 1.30 105 2.6 1012 1.5 9.2 108 0.5 3.5 107 (b) Metal connector 1.7 108 400 1.761 1013 171 106 252 106 1.04 105 6.8 1012 1.5 8.0 108 0.6 9.3 107 Substrate 163

Hot side connector y x n-type TE p-type TE

7 m

1 m

Cold side connector

Fig. 2.

Schematic of the two-dimensional geometry.

Property e k

based on the nite element method and is applicable to the multiphysics problems. The thermophysical properties related to the thermoelectric elements are evaluated by referring to the values reported by da Silva and Kaviany [8]. The material of the metal connector is assumed to be copper for the sake of convenience. The thermal conductivities of Copper and Silicon are adopted from the material library of the FEMLAB software. In the calculation considering the thermal non-equilibrium near the interfaces of the thermoelectric elements, the domain for the calculation is divided into two subdomains. The rst subdomain is used for the thermoelectric elements with Eqs. (1) and (2) and the second subdomain is used for the metal connectors and substrates with Eq. (7). The rst domain has two different dependent variables such as Tp , Te and the second subdomain has only T . In the rst domain, the equations are separately applied to the n-type and p-type elements with their own properties indicated in Table 1. The temperatures for the two subdomains are connected with the interface conditions, Eqs. (3) through (6), and Eqs. (8) and (9). If the thermal equilibrium is considered, the equations described as Eqs. (10) through (12) are used for the rst subdomain and Eqs. (13) and (14) are used as the interface conditions of metal connectors.

one-dimensional analytic solutions reported in the literature [8]. The analytic solutions indicated as symbols in Fig. 3 are obtained from the one-dimensional geometry in which the thickness of each thermoelectric element is 4 m and the width is 7 m. The thickness of the metal connector is xed to 1 m and it is not important here since the temperature varies little within the connector in the y-direction. The properties of materials shown in Table 1 are used in the analysis. The predicted temperatures in the present study are nearly identical to the one-dimensional analytic solutions as indicated in Fig. 3. The result shows that the thermal non-equilibrium near the interface cause the different temperatures for the electron and the phonon. The temperature jump for electron conduction is larger than for phonon conduction due to the relatively large boundary resistance for electron conduction, (Ate Rc )b,e , indicated in Table 1. When the electric current does not ow, the temperatures for the electron and phonon are almost identical to each other. The analysis to compare the results for the electron-phonon

330 Tp 320 310 Te Tp (Analytic) Te (Analytic) Je = 30 mA Je = 15 mA

Temperature (K)

300 290 280 270 Je = 0 mA

RESULTS AND DISCUSSION Validation of the Analysis The analysis has rstly been validated with the onedimensional analytic solution reported in the literature [8] and the analysis for the validation is carried out in the twodimensional geometry as shown in Fig. 2. The dimension of the geometry is arbitrarily determined by considering that the predicted results in the present study can be compared with the

260

-3

-2

-1

4 m

Thickness (m)

Fig. 3. Comparison of the predicted temperature distributions with the one-dimensional analytic solution which is regenerated from the literature [8].

Copyright c 2005 by ASME

1 m
3

305 Non-eq. (p-type) Non-eq. (n-type) Eq. (p-type) Eq. (n-type)

by considering the thermal conduction through the interface, (Ate Rc )b = (Ate Rc )b,e (Ate Rc )b,p , (Ate Rc )b,e + (Ate Rc )b,p (15)

300

Temperature (K)

and the thermal conductivity is written for Te = Tp = T ,

295

k = ke + k p .
290

(16)

285

280

-3

-2

-1

Length, y (m)

(a)
315 310 305 Non-eq. (p-type) Non-eq. (n-type) Eq. (p-type) Eq. (n-type)

300 295 290 285 280 275

The temperatures for the n-type thermoelectric element are nearly identical for different currents but the temperatures for the p-type thermoelectric element differ from each other in Fig. 4. The average thermal conductivities are 2.0 and 2.1 W/mk, and the averaged boundary resistances are 7.29 108 and 7.37 108 m2 K/W, respectively for the n- and p-type elements. Those properties do not differ much from each other and then the temperatures for n- and p-type elements in the equilibrium condition do not also differ much from each other. However, in the non-equilibrium condition, the temperature variation near the interface for electron conduction is much higher than for phonon conduction. This is primarily due to the difference between the boundary resistances for electron and phonon conduction. For the n-type element with relatively small electron boundary resistance, the temperature obtained from the non-equilibrium analysis is similar to the temperature from the equilibrium analysis. However, the temperature differences between two analyses are large for the p-type element with relatively large electron boundary resistance. It seems to be due to the electron-phonon interaction near the interface related to the boundary resistance because the same properties excluding the averaged conductivity and resistance are used and conductivities do not differ much in two different analyses. Thus, the electron-phonon thermal nonequilibrium near the interface is considered in the analysis of the present study. Three-dimensional Analysis The three-dimensional analysis is carried out with the geometry shown in Fig. 1 but the thickness of thermoelectric elements is changed to investigate the characteristics of the proposed thermoelectric cooler with the thickness. First of all, two special boundary conditions are considered to gure out some maximum performance as a reference. One is the thermal insulation at the cold side substrate and the xed temperature at the hot-side substrate, and another is both the cold and hot sides maintained at the same temperature [1]. In the analysis, the maximum current, the maximum temperature difference, and the maximum cooling rate of the cooler are evaluated. The maximum current does not mean the maximum value of the current but the value for which the temperature difference is maximum. When the cold side is insulated and the hot-side temperature is xed to 25 C, Fig. 5 shows that the maximum current increases as the thickness of the thermoelectric element decreases and the results are summarized in Table 2. The maximum

Temperature (K)

-3

-2

-1

Length, y (m)

(b) Fig. 4. Temperature distribution for the two-dimensional geometry; (a) Je = 15 mA and (b) Je = 30 mA.

thermal equilibrium and non-equilibrium near the interface is also carried out in the two-dimensional geometry as shown in Fig. 2. Figure 4 compares the results for the thermal equilibrium and non-equilibrium analyses. For the comparison, the equilibrium temperature, T (kT = ke Te + k p Tp ), is dened for the nonequilibrium solution. The properties for the equilibrium solutions such as the boundary resistances and thermal conductivities of the thermoelectric elements are averaged with values given in Table 1 for the comparison. The boundary resistance is written

Copyright c 2005 by ASME

110

Temperature difference, Th Tc (C)

6 5

100 90

Cooling rate (W)

80 70 60 50 40 30 20 10 0 0 5 10 Lte = 20 m Lte = 15 m Lte = 10 m Lte = 5 m 15 20

Lte = 20 m Lte = 15 m Lte = 10 m Lte = 5 m

4 3 2 1 0

Current (A)

10

12

14

16

18

20

Current (A)

Fig. 5. Temperature difference between the hot and cold side for Qc = 0 and Th = 25 C.

(a)
30 25 Lte = 20 m Lte = 15 m Lte = 10 m Lte = 5 m

temperature difference at the maximum current decreases as the thickness of the element decreases and it is greater than 100 C when the thickness is greater than 15 m. For the efcient operation, it is desirable that the maximum current is small. Thus, the thicker thermoelectric element is a better choice in the view of the maximum current and temperature difference since the thinner element requests very large electric current and it is of no practical use. Figure 6 shows the cooling and heat release rate in the special condition of both the cold and hot sides maintained at the same temperature, Th = Tc = 25 C. The maximum cooling rate increases as the thickness of the thermoelectric element decreases and the current increases in the contrary of the previous case in Fig. 5. However, in spite of the large maximum cooling rate when the thickness is small, the cooler might be of no practical use since the heat release rate increases as the current increases as shown in Fig. 6(b) and the maximum cooling rate occurs at the large current. In addition, the cooling rate is larger for the thicker element at the relatively small current. Therefore, the thicker element is desirable for the efcient cooling. In order to gure out the performance of the proposed thermoelectric micro-cooler, the calculation has been carried out when the hot-side temperature is varied from 45 C to 125 C with the cold-side temperature xed to 25 C. The difference beTable 2. Maximum current, the maximum temperature difference and the maximum cooling rate. Lte (m) 10 15 20 Je,max (A) 15.0 11.2 9.0 Tc,min ( C) -67.2 -75.2 -79.6 Tmax ( C) 92.2 100.2 104.6 16.2 13.4 4.8 4.1 17.1 14.5 Je (A) Qc,max (W) Qh (W)

Heat release rate (W)

20 15 10 5 0

10

12

14

16

18

20

Current (A)

(b) Fig. 6. (a) Cooling rate at the cold side and (b) heat release rate at the hot side for Th = Tc = 25 C.

tween the hot-side and cold-side temperatures are varied from 20 C to 100 C and the predicted results are shown in Figs 7 to 10. The cooling rate at the cold-side increases with the electric current but the rate decreases after it meets its maximum value at a certain current, and the rate decreases as the hot-side temperature increases as shown in Fig. 7(a). When Lte = 20 m, the maximum cooling rate at the cold side occurs at Je = 13 A regardless of the hot-side temperature. The maximum rate decreases at the constant rate by about 0.35 W when the hot-side temperature increases by 20 C. This trend can be found for the cases with the different thickness of the thermoelectric element shown in Fig. 8. For example, when Lte = 15 m, the maximum cooling rate occurs at Je = 16 A and the rate decreases by 0.45 W when the hot-side temperature increases by 20 C as shown in Fig. 8(a). The heat release rate gradually increases with the current even after the current (Je = 13 A) at which the cooling rate

Copyright c 2005 by ASME

4 3.5

30 25 20 15 10 5 0

Cooling rate, Qc (W)

3 2.5 2 1.5 1 0.5 0 0 5 10 T = 20 C T = 40 C T = 60 C T = 80 C T = 100 C at COPmax 15 20

Heat release rate, Qh (W)

T = 20 C T = 40 C T = 60 C T = 80 C T = 100 C

10

15

20

Current (A)

Current (A)

(a)
30 25 T = 20 C T = 40 C T = 60 C T = 80 C T = 100 C 3 2.5 2

(b)
T = 20 C T = 40 C T = 60 C T = 80 C T = 100 C COPmax

Electric power (W)

20 15 10 5 0

COP
0 5 10 15 20

1.5 1 0.5 0

10

15

20

Current (A)

Current (A)

(c)

(d)

Fig. 7. Variation of (a) cooling rate at the cold side, (b) heat release rate at the hot side, (c) power, and (d) COP of the cooler (Lte = 20 m).

reaches its maximum value as shown in Fig. 7(b). It is not efcient that the cooler is used at the current greater than 13 A since the cooling rate decreases even though the current increases. The heat release rate decreases with the increase of the hot-side temperature for the current less than 7.5 A but after that the rate increases with the hot-side temperature The electric power required is evaluated by considering the thermal energy balance in the cooler as Pe = Qh Qc . It is reasonable because all the boundaries except for the hot-side and cold-side surfaces are treated as surfaces insulated to the thermal energy transfer. Since the cooling rate is fairly less than the heat release rate though the cooling rate initially increases and decreases after its maximum value with the current, the electric power similarly increases with the current as the heat release rate does as shown in Fig. 7(c). The power also increases as the hotside temperature increases.

Figure 7(d) shows the variation of the COP of the cooler with the electric current. The COP has its maximum value at the relatively small current in comparison to the case of the cooling rate. When Lte = 20 m, the maximum COP decreases from 2.88 to 0.33 its current increases from 1.4 to 6.0 A as the temperature difference increases from 20 to 100 C. The maximum COP decreases quickly as the hot-side temperature increases when the temperature is relatively low. At the current at which the maximum COP occurs, the cooling rate varies from 0.5 to 1.2 W with the hot-side temperature and it is less than the maximum cooling rate indicated in Fig. 7(a). However, in order to increase the energy efciency of the cooler, the cooler should be operated at the condition for which the maximum COP can be obtained. The cooling rates for which the maximum COPs are obtained for the different coolers are also indicated in Fig. 8. Figure 9 shows the variations of the COPs for the coolers

Copyright c 2005 by ASME

Cooling rate, Qc (W)

T = 20 C T = 40 C T = 60 C T = 80 C T = 100 C at COPmax

3 2.5 2 T = 20 C T = 40 C T = 60 C T = 80 C T = 100 C COPmax

COP
2 1 0 0 5 10 15 20

1.5 1 0.5 0

10

15

20

Current (A)

Current (A)

(a)
5 T = 20 C T = 40 C T = 60 C T = 80 C T = 100 C at COPmax 3 2.5 2

(a)
T = 20 C T = 40 C T = 60 C T = 80 C T = 100 C COPmax

Cooling rate, Qc (W)

COP
2 1 0 0 5 10 15 20

1.5 1 0.5 0

10

15

20

Current (A)

Current (A)

(b)
5 T = 20 C T = 40 C T = 60 C T = 80 C T = 100 C at COPmax 3 2.5 2

(b)
T = 20 C T = 40 C T = 60 C T = 80 C T = 100 C COPmax

Cooling rate, Qc (W)

COP
2 1 0 0 5 10 15 20

1.5 1 0.5 0

10

15

20

Current (A)

Current (A)

(c) Fig. 8. Cooling rate at the cold side for (a) Lte = 15 m, (b) Lte = 10 m, and (c) Lte = 5 m.

(c) Fig. 9. Coefcient of performance for (a) Lte = 15 m, (b) Lte = 10 m, and (c) Lte = 5 m.

with the thermoelectric elements of different thickness. When T = 20K, the maximum COPs are 2.6, 2.2, and 1.5 at the current of 1.8, 2.6, and 4.8 A for Lte = 15, 10, and 5 m, respectively, and thus the value decrease more quickly as the thickness of the

element decreases. When the thickness is small and the temperature difference is large, the current for the maximum COP largely increases and it may be of no practical use. Figure 10 shows the temperature distributions of several

Copyright c 2005 by ASME

(a)

(b)

(c)

(d)

Fig. 10. Temperature distributions of the cross-sections in (a) the yz-plane (b) the xz-plane, (c) the cold-side connector, and (d) the hot-side connector for Lte = 20 m, T = 60 C and Je = 3.8 A.

cross-section of the cooler for Lte = 20 m, T = 60 C and Je = 3.8 A for which the maximum Qc is obtained. The temperature distributions of two vertical cross-sections shown in Figs. 10(a) and (b) illustrate that the temperature variations in both the cold and hot substrates vary a little, which is coincided with the result of the two-dimensional analysis shown in Fig. 4. The temperature near the interface between the thermoelectric element and the cold-side connector is down to 20.4 C (Te = 19.2 C, T = 20.8 C) and is lower than the temperature, 25 C, in p the cold-side substrate due to the Peltier cooling. In the contrary, the temperature near the interface between the thermoelectric element and the hot-side connector is 95.5 C (Te = 97.6 C, Tp = 95.0 C) and is higher than the temperature, 85 C, in the hot-side substrate due to the Peltier heating. Figures 10(c) and (d) show the temperature distributions of the cross-sections in the cold-side and the hot-side connectors, respectively. In the cold-side connectors, the colder regions are centralized and the temperatures near the boundaries are relatively hot. The colder regions are distributed in accordance with the patterns of thermoelectric elements. The hotter regions in the hot-side connectors are also centralized and distributed in ac-

cordance with the patterns of the thermoelectric elements. However, the temperature difference between the hottest region and the coldest boundary is is about 4 C much less than the difference between the hot-side and cold-side substrate (T = 60 C).

CONCLUSION The three-dimensional numerical analysis has been carried out using the FEMLAB software package to gure out the performance of the thermoelectric micro-cooler. The small-size and column-type thermoelectric cooler is considered and Bi2 Te3 and Sb2 Te3 are selected as the n- and p-type thermoelectric materials, respectively. The thickness of the thermoelectric element considered is 5 to 20 m and the thickness affects the performance of the cooler. The two-dimensional analysis has been carried out to validate the analysis before the three-dimensional analysis and the predicted temperature is in good agreement with the onedimensional analytic solution given in the literature. The electron-phonon thermal equilibrium and non-equilibrium is considered in the two-dimensional analysis near the interface be-

Copyright c 2005 by ASME

tween the connector and the thermoelectric element. In this case, the temperatures obtained from two types of analyses shows the difference especially in the p-type element. In order to nd the maximum performance quantities such as the maximum temperature difference and cooling rate, the threedimensional analyses for two special boundary conditions is carried out. One is the thermal insulation at the cold side substrate and the xed temperature at the hot-side substrate, and another is both the cold and hot sides maintained at the same temperature. The maximum temperature difference is larger for the thicker thermoelectric element at the small value of the current (i.e., the maximum current), but the maximum cooling rate is larger for the thinner element at the large current which might be of no practical use. The effect of parameters such as the temperature difference, the current, and the thickness of the thermoelectric element on the performance of the cooler has also been investigated. The COP has the maximum value at a certain current and the value increases with the decrease of temperature difference or the increase of the thickness. The predicted results show that the performance can be improved more with the thick thermoelectric elements at the small temperature difference and the small current.

7.

8.

9.

10.

11.

12.

ACKNOWLEDGMENT This research is implemented as a part of the project Technology Development of Advanced Cooling System managed by Korea Research Council for Industrial Science & Technology. The authors gratefully appreciate the support.

MEMS-like Electrochemical Process, Nature Materials, 2(8), pp. 528531. B ttner, H., Nurnus, J., Gavrikov, A., K hner, G., J gle, M., o u a K nzel, C., Eberhard, D., Plescher, G., Schubert, A., and u Schereth, K.-H., 2004, New Thermoelectric Components Using Microsystem Technologies, Journal of Microelectromechanical Systems, 13(3), pp. 414420. da Silva, L. W., and Kaviany, M., 2004, Microthermoelectric Cooler: Interfacial Effects on Thermal and Electrical Transport, International Journal of Heat and Mass Transfer, 47(10-11), pp. 24172435. da Silva, L. W., Kaviany, M., and Asheghi, M., 2004, Measured Performance of a Micro Thermoelectric Cooler, Proceedings of ASME HT-FED 2004, Paper No. HT-FED200456412. Bartkowiak, M., and Mahan, G. D., 2001, Heat and Electricity Transport through Interfaces, Recent Trends in Thermoelectric Materials, Vol. II, Semiconductors and Semimetals, Vol. 70, Academic Press, New York, pp. 245271. Gurevich, Y. G., and Mashkevich, O. L., 1989, The Electron-Phonon Drag and Transport Phenomena in Semiconductors, Physics Reports (Review Section of Physics Letters), 181(6), pp. 327394. Yao, D.-J., Kim, C.-J., Chen, G., Fleurial, J.-P., , and Lyon, H. B., 1999, Spot Cooling by Using Thermoelectric Microcooler, Proceedings of 18st International Conference on Thermoelectrics, pp. 256259.

REFERENCES 1. Rowe, D. M., ed., 1995, Handbook of Thermoelectrics, CRC Press, Boca Ration. 2. B ttner, H., 2002, Thermoelectric Micro Devices: Current o State, Recent Developments and Future Aspects for Technological Progress and Applications, Proceedings of 21th International Conference on Thermoelectrics, pp. 511518. 3. Lim, J. R., Snyder, G. J., Huang, C.-K., Herman, J. A., Ryan, M. A., and Fleurial, J.-P., 2002, Thermoelectric Microdevice Fabrication Process and Evaluation at the Jet Propulsion Laboratory (JPL), Proceedings of 21st International Conference on Thermoelectrics, pp. 535446. 4. Yao, D.-J., Kim, C.-J., Liu, J. L., Wang, K. L., Snyder, J., Fleurial, J.-P., and Chen, G., 2001, MEMS Thermoelectric Microcooler, Proceedings of 20th International Conference on Thermoelectrics, pp. 401404. 5. Yao, D. J., Chen, G., and Kim, C.-J., 2001, Low Temperature Eutectic Bonding for In-Plane Type Micro Thermoelectric Cooler, Proceedings of International Mechanical Engineering Congress and Exposition, Paper No. IMECE2001/MEMS-23901. 6. Snyder, G. J., Lim, J. R., Huang, C.-K., and Fleurial, J.P., 2003, Thermoelectric Microdevice Fabricated by a

10

Copyright c 2005 by ASME