- Email: [email protected]
Intensive energy density thermoelectric energy conversion system by using FGM compliant pads
Pergamon PII:
www.elsevier.cotn/locate/actaastro
ActaAstronautica Vol. 51, No. 1-9. pp. 161-171, 2002 0 2002 International Astronautical Federation. Published by Elsevier Science Ltd All rights reserved. Printed in Great Britain SOO94-5765(02)00071-1 0094-5765/02 $ - see front matter
INTENSIVE ENERGY DENSITY THERMOELECTRIC ENERGY CONVERSION SYSTEM BY USING FGM COMPLIANT PADS Mitsuru KAMBE Central Research Institute of Electric Power Industry (CRIEPI) 11-1, Iwado Kita 2-chome, Komae-shi, Tokyo 201-8511, Japan E-mail:
[email protected]
Hideo SHIKATA Hitachi Powdered Metals Co., Ltd. 520, Minor&i, Matsudo-shi, Chiba, 270-2295, E-mail: [email protected]
Japan
1. INTRODUCTION
Abstract
In previous thermoelectric (TE) space power systems, such as the radioisotope thermoelectric generator (RTG) I), the TE devices were cantilevered from the heat sink (cold side). This design strategy reduced the stress in the TE device by not restricting thermal expansions, but required heat to be transferred l%om the heat source to the TE cells by radiation. As the thermal energy density available from most isotopes used in RTGs is quite low, radiation coupling of the TE cell to the heat source is an acceptable method and the RTG performance is not strongly affected by this radiation gap. However a severe mass penalty would occur if radiation coupling were to be used. Hence conductively coupling the TE module to the hot and cold heat exchangers is the most effective configuration to achieve high thermal energy density. The Conductively Coupled TE Cell with compliant pad was first developed by GE as part of the SP-100 Space Reactor Power System 2) which is designed to convert the nuclear power of a fast flux, lithium cooled space reactor into electrical power in the range of lo’s to 100’s of kW. In the SP-100 TE power conversion system 2,3,4+),TE cell uses a multicouple configuration which is conductively coupled to both the heat source and the heat sink. At operating temperatures,
In order to provide increasingly large amounts of electrical power to space and terrestrial systems with a suf6cient reliability at a reasonable cost, thermoelectric energy conversion system by using Functionally Graded Material (FGM) compliant pads has been focused. To achieve high thermal energy density in thermoelectric (TE) power conversion systems, conductively coupling the TE module to the hot and cold heat exchangers is the most effective configuration. This is accomplished by two sets of FGM compliant pads. This design strategy provides (1) a high flux, direct conduction path to heat source and heat sink, (2) the structural flexibility to protect the cell from high stress due to thermal expansion, (3) an extended durability by a simple FGM structure, and (4) manufacturing cost reduction by spark plasma sintering. High thermal energy density of more than twice as much as conventional conduction coupling TE generator is expected. TE energy conversion systems combined with FGM compliant pads
for space and terrestrial design options are proposed in this paper. o 2002 International Astronautical Federation. Published by Elsevier Science Ltd. All rights reserved.
the
161
dome
shaped
deformation
of the
TE
162
52nd IAF Congress
module, in addition to the differential thermal expansion between hot and cold exchangers, must be accommodated without inducing large stress into the sensitive cell elements. This is accomplished by two sets of the compliant pads 3,*,5).Key parameters for these compliant pads are high thermal conductivity, low stiffness in shear and compression, and the ability to limit peak stress. In the SP- 100 TE system, niobium @III) filaments planted on Nb facesheet cotiguration is adopted for the compliant pads. These compliant pads are very delicate structure because the filament diameters of the pads are roughly ten times less than the thickness of a human hair. If all the filaments in a l-inch by l-inch pad for a 100 kWe system were connected endto-end, the total filament length would span 3.5 times around the earth. This project, however, was canceled in 1993. TF system was also adopted for terrestrial applications. However, most of the systems are “conventional” conduction coupling in which no compliant pad is adopted and TE devices were placed between the hot duct (heat source> and cold duct (heat sink) with a pressure load. In this configuration, the at TE device/duct thermal resistance interface is greater, therefore poor electrical output is available. Increasing the pressure load would reduce the thermal resistance, however, it would damage fragile TE cell. Under these circumstances, the author pad design of proposed a compliant Functionally Graded Material @GM) 6. 7,8, 91 to achieve high power density TE energy conversion systems for space and terrestrial applications. FGM compliant pad consists of Cu layers for stress relaxation and an Al203 layer for electrical insulation. This design strategy provides (1) a high flux, direct conduction path to heat source and heat sink, (2) the structural flexibility to protect the celI from high stress due to thermal expansion, (3) an extended durability by a simple FGM and (4) manufacturing cost structure, reduction by powder sintering process. The most important in FGM compliant pad
design is that all the elements should be bonded together to eliminate thermal resistance at the interfaces. In this paper, design and performance of the proposed TE device with FGM compliant pads are presented. Other design options to accommodate higher operating temperature (i.e. greater thermal expansion mismatch) and severe thermal transient are also demonstrated. Proposed TE units can be applied to both space and terrestrial power systems 10). 2. TE UNIT WITH FGM COMPLIANT PADS 2.1 Design ADproach Fig. 1 illustrates multicouple SiGe TF unit with Cu/aluminaOUzOa)lCu symmetrical FGM compliant pads designed for both terrestrial and space power systems. In order to reduce edge stress of dissimilar bonded materials by differential thermal expansion, a l-inch by l-inch module, same size as that of SP- 100, has been adopted. Requirements for the compliant pad stress relaxation layer are a low stiffness in shear and compression, and a good conductor of heat. An attention was focused to select material of smaller E/ 1 where E is modulus of elasticity, and ;1thermal conductivity.
CulAlg, I cu Symmetrical FGY Graphite SiGe Insulator Graphite CulAI& I cu Symmetrical FGM
(Cold side)
Fig. 1 Concept of multicouple TE unit with Cu/AlpO&u FGM compliant pads.
52nd IAF Congress
Therefore copper (Cu) was selected. Because the melting temperature of Cu is 108X, the operating temperature of the unit is less than 800 ‘C if it is operated in the vacuum environment. For terrestrial systems, the operating temperature of less than 5OO’C (hopefully under 400°C) is recommended to avoid oxidation of Cu layer. The E/ ;1 value is independent of the porosity P of the compliant pads because their effective modulus of elasticity is roughly equal to PE, and effective thermal conductivity P ;1. Porosity of approximately 0.8 is expected in our proposed sintering process (refer to chapter 2.2). Alumina 6A.lzO3) layer has a role of an electrical insulator. Because the linear thermal expansion coefficient of Cu is twice as much as Al203, it is impossible to braze Al203 plate to Cu electrode. The essential feature of FGM is that there is no interface between Cu and A1203 layer, and cu/&03 concentration normal to the surface varies graduahy. In this structure, thermal stress due to thermal expansion mismatch of Cu and A203 can be easily accommodated 67). Graphite (Gr) layer acts as a chemical barrier between the SiGe and Cu layers. Cu/AlzOs/Cu symmetrical FGM 1i.1was designed to achieve easy assembling of the whole unit as well as to obtain structural integrity in the FGM compliant pad. In the Conventional cd~203 FGM, bonding Of an heat exchanger duct (usually made of aUSt.enitiC Steel, Inconel or Cu) and 4203 side of the FGM should be conducted carefully because the linear thermal expansion coefficient of the former is nearly twice as large as the latter. In case of symmetrical FGM, bonding of a heat exchanger duct and an outer surface of the symmetrical FGM (Cd is generally easier. Another advantage of the symmetrical FGM is that the dome shaped deformation due to the thermal expansion mismatch between Cu and Al303 layers of FGM during processing and operation could be avoided. This feature will ensure long life durability of the TE unit. though Even the thickness of the
163
symmetrical FGM is greater than that of cu/A1203 FGM, only a slight increase (l-2%) in overall thermal resistance is anticipated because most of the symmetrical FGM is composed of Cu. Accordingly the effect of the thermal resistance on the TE performance is only a few percent. 2.2 Manufacturing of FGM compliant pad Manufacturing of Cu/AhO3/Cu FGM by spark plasma sintering (SPS) has been attempted 19. SPS was recently developed to obtain a fine sintering structure at a relatively lower processing temperature by inducing plasma among material particles to The following developing issues be sir&red. should be investigated: 0 Powder sintering process should be done below the melting temperature of Cu (1083°C). l Sintering at 1250 “c under 50 MPa pressing is preferable for Al203 because higher density of A1203 is required to achieve a high flux conduction path and structural integrity. In order to meet both requirements, “solid/liquid phase” SPS has been applied. In this process, sample is gradually heated up to 1250°C while considerable attention has been made to avoid loss of liquefied Cu in the sample. In view of this, a punch pressure was held 50 MPa just prior to reach melting temperature of Cu, and after that suddenly decreased to 25 MPa. An Al203 sintering density of 80% has been achieved without any crack in the FGM layers. Sufficient electrical insulation capability in the direction normal to FGM layers has been confirmed. 2.3 Overall Assembling In order to ensure function of the TE system, all the members of TE unit should be interconnected to eliminate thermal resistance at the interfaces. Peculiar requirements for assembling TE unit are that bonding temperature should be higher than the operating temperature (~800°C) of the element to be bonded, while it should be
164
52nd IAF Congress
lower than the melting temperature of Cu (1083 “c ). Therefore bonding should be conducted in the temperature range between 850°C and 1050°C. In view of this, diffusion bonding cannot be applied. Brazing has the advantage that only lower contact loads and lower bonding temperatures are necessary compared to diffusion bonding. An active metal braze material based on the Ag/Cu/ln/Ti system was adopted. Preliminary bonding tests for ail interfaces have been made: Brazing of Gr to Cu side of FGM is a proven technique in several industrial applications, while the combination Gr/SiGe is more critical: higher difference of the thermal expansion modules, better chemical affinity between active metal braze and SiGe. Brazing of SiGe/Gr/Cu can be made using “two step brazing cycle”: first brazing of Gr to Cu at lOOO”c,including the wetting of Gr on its other side with an active metal braze, and then brazing of wetted Gr to SiGe in a second step with a lower working temperature (850°C). Activity of titanium 0% is starting at temperatures about 900-950 “c , therefore brazing of wetted Gr to SiGe can be done in a second step without damage of the primary wetting, if the brazing temperature is lower than 900°C. Brazing of Cu side of FGM to the heat exchanger duct (usually made of austenitic steel, Inconel or Cu) is also proven. Metallographic inspection revealed a sound brazing structure. In case of assembling BiTe TE unit, which is most popular for terrestrial applications, colloidal graphite (trade name HITASOL by Hitachi Powdered Metals) can be used. It is sub-micron graphite powder immersed in alcoholic solvent. HITASOL also acts as a chemical barrier between BiTe and Cu layers. Sufficient bonding strength of up to 700°C is proven. Therefore it also applied to terrestrial SiGe TE units. It features significant reduction of the thermal and
electrical resistance at the interface. The thermal resistance at a Cu/BiTe interface bonded by HITASOL is eight times less than that without HITASOL (refer to chapter 3). The de&rite advantage of HITASOL is that bonding can be done at room temperature just like glue. Sufficient bonding strength is available after one day, if dried in room temperature, or dried up in the hot air circulating furnace of 150°C for 15 minutes. HITASOL could achieve easy assembling of multicouple TE unit in which many bonding interfaces exist. Recently use of lead (Pb) braze filIer is restricted in view of the environmental protection, while nothing is as yet satisfactory as an alternative. Therefore advantage of non-toxicity is another HITASOL. Fig.2 shows unicouple BiTe TE unit with Cu/Al~0&u compliant pad assembled by HITASOL.
8mm
*
Fig.2 Sample of a unicouple TE unit with Cu/Al2O&u FGM compliant pads.
165
52nd IAF Congress
3. PERFORMANCE OF THE TE UNIT WITH FGM COMPLIANT PADS 3.1 Gap Conductance Unicouple BiTe TE units with and without FGM compliant pads were tested t.c measure temperature distribution in the unit. Fig. 3 illustrates conventional TE unit without compliant pad It consists of BiTe TE cell, Cu electrodes and a mica sheet as an electrical insulation, and assembled by a screw. l In the first test run, no interface is bonded at all to accommodate thermal expansion mismatch. l In the second test run, the hot side electrode (Cu) and BiTe are bonded by HITASOL for comparison. Usually this TE unit is sandwiched between the hot @eat source) and cold (heat sink) ducts, however, an infrared lamp is adopted instead of the hot duct in our experiment. Difference in this configuration does not affect on the measurement of gap conductance. Thermocouples of 70 ~1 m diameter are spot-welded on the TE unit members to obtain temperature proflle. Because of their very small diameter, effect of heat loss due to these thermocouples on the temperature profile is neglected. Gap conductance between the hot side electrode (Cu) and BiTe is given by the following equations.
ATi2 = <[++
;J
***(l)
qL AT23=
.*.(2>
obtains
1
6
LATiz -
-= ~TEAT
h
a-lx 0.4
4.2 AT12 -
= L
1.2
1.2 ATzs
=(3.5ATi2 /ATzs
10-3
x
-0.333)X
I
10-S
(m21QW) .-e(3)
where Temperature
difference between TCl-TC2: ATi2 ac>
Temperature
difference between TC2-TC3: ATzs (K>
Heat flux:
q(w)
Distance defined in Fig. 3: L=4.2
X
10-a (m)
6 =0.4x 10-z (m) TE cell cross section: Thermal conductivity
A= 16 x 10-G (rn2) of BiTe : ATE =1.2 (W/mIo
Gap Conductance:
h (W/m@
The first and second test runs revealed the following gap conductance between the hot side electrode (Cu) and BiTe. Therefore the thermal gap conductance at a CuBiTe interface bonded by HITASOL is eight times greater than that without HITASOL. hr=715 (W/m@: hH=5836 (W/m%):
without HITASOL with HITASOL
AilTE
Eliminating
q/A in above equations, one
It should be noted that absolute values of hx and hH are not reliable because heat loss due to radiation in radial direction is
52nd IAF Congress
166
dominating in this experimental setup (e.g. TE cell cross section of each leg is only 4 mm x 4 mm and temperature measurement is done on the TE cell surface). Therefore only a relative value of h g and hn can be discussed. Fig. 4 shows a TE unit with FGM compliant pads. All the members are bonded other by each HITASOL. Three thermocouples of 70 P m diameter are also spot-welded on the TE unit. Measured temperature is shown in the figure. Temperature profile deduced by hn described above is also plotted by dotted line in the figure. Measured and deduced temperature profiles coincide close together.
3.2 Comparison of the TE units Based on the gap conductance obtained above, performances of the TE units with and without FGM compliant pads are compared as shown in Fig. 5. l Model A is a conventional unit which consists of BiTe TE cell, Cu electrodes and mica sheets as electrical insulations, and sandwiched between the hot and cold
ducts with a pressure load. This unit has six gaps. No HITASOL layer exists. (If these gaps are filled with HITASOL, thermal stress relaxation is impossible and the TE unit would be broken.) 0 Model B is similar to Model A, however, Cu electrodes are bonded to BiTe by HITASOL. This unit has four gaps. 0 Model C is a proposed TE unit with FGM compliant pads. This unit has no gap because four interfaces are filled with HITASOL. This is a case that hot and cold duct temperature are 240 and 2o”C, respectively. Heat flux: q achievable by each model is shown in the figure. Model C can achieve heat flux of more than twice as those of model A and B. Therefore TE unit with FGM compliant pads would provide electrical power more than twice as that of conventional TE unit. Fig. 6 indicates the thermal resistance in the TE unit. Thermal resistance of gaps in model A and B dominate TE performance. Temperature gradient in each model is presented in Fig. 7.
n
Infrared Lamp
Infrared Lamp
0
...-....
ttttt
Axial Location
Experiment Calculated profile
uw cu
TCl TC2
BiTe cu Mica
TC3
I
Cooling Duct
TC4
tTested TE Unit with FGM Compliant Pads (HITASOL in all the Interfaces)
Fig.3
Experimental setup of a conventional TE unit.
Fig.4
Experimental setup of a TE unit with FGM compliant pads.
167
52nd IAF Congress
q=27.7 W/in’ q=12.9 W/in2
q=lO.5 W/id
240°C
1,Ht”“C”“tDtu”‘t
1
(0.1 mm)J
I Model A Conventional TE Unit
Cooling,Duct
I
1
Cooling Duct
1 20°C
Model C Model B Conventional TE Unit with FGM Compliant Pads (HITASOL in CuBiTe Interfaces) (HITASOL in all the Interfaces)
Fig. 5
Conduction coupling type TE unit designs.
6 Gaps
’
Fig.6
E g
MidCu
BiTe
BiTe
BiTe
Model A
ModelB
Mode/ C
Thermal resistance of the T’Eunit.
220
4 HITASOL ~ayei-s /2
*oO-
al 5
2
4 Gaps
6 Gaps
S o!
i
loo-
k?
E
X4calC”
z
BiTe Model A
Fig. 7
FGH Compliant Pads
/
2 HITASOL Lv=
‘Mica/C”
BiTe
BiTe Model B
Model %
Temperature gradient in the TE unit.
168
52nd IAF Congress
4. LIQUID
COMPLIANT
PAD
Greater the capacity of the TE energy conversion system, larger the heating and cooling ducts between that TE units are clamped. In addition the operating temperature of the TE power conversion systems for space applications is geneSally higher than those for terrestrial systems. Differential thermal expansion of the hot and cold ducts parallel to the duct surface is given by the following relation. 6==/3LAT .+0 where 6 : differential thermal expansion B : linear thermal expansion modulus of the duct material L: distance from the Axed point to the duct edge (if duct is supported at its centerline, L is half of the duct width 1 AT: difference between the hot and cold ducts temperature l
l
In case of niobium @lb) hot/cold ducts (B = 7.3 x lo-6K-1) which is operated at 906 and 5OO”c, respectively, and L=ZOO mm (duct width 400 mm), 6 will be 0.6 mm. This is a typical condition for SiGe space power system designed for SP- 100 s,4,5). In case of austenite steel (SUS316) hot/cold ducts (B = 16.7X IO-SK-i) which is operated at 240 and 2O”c, respectively, and L.=300 mm (duct width 600 mm), 6 will be 1.1 mm. This is a typical condition for BiTe TE system for terrestrial applications.
FGM compliant pad cannot accommodate such greater thermal expansion mismatch. thermal Severe transient in startup/shutdown cycle is another dif&ulty to be encountered in terrestrial TE systems. In addition vibration of the ducts normal to the duct surface would be expected if applied to the exhaust duct of the automobile. Liquid compliant pad 19) should be used together with FGM compliant pad to accommodate such severe conditions. Concept is illustrated in Fig. 8. In the cold side, FGM compliant
pads are inserted between TE cells and a cold duct as shown in the preceding chapter. In the hot side, however, we have a Cu plate and a braze filler layer as well as FGM compliant pads. In case of BiTe TE system, tin (Sn) is adopted as braze filler material. Because the melting temperature of Sn is 232°C) it is liquid during operating temperature. Due to strong affinity of Sn with Cu, the thermal gap conductance of Cu/Sn/Cu interface is enhanced, while displacement of this interface is possible because Sn is liquid. Therefore differential thermal expansion of the hot and cold ducts parallel to the duct surface can be relieved perfectly. Vibration of the ducts normal to the duct surface is also accepted. FGM compliant pad in this configuration has a role of accommodating thermal stress induced in the temperature difference of 232”Croom temperature. For space applications, CB2 braze filler (trade name of Goodfellow GmbH, 96Ag-4Ti, melting temperature 970°C) is one of the candidates under consideration. Due to the surface tension of braze filler, it could be held in the interface even in the non-gravity field. Experiment to demonstrate liquid compliant pad performance has been done. Test sample consists of two Cu cylinders of 25 mm in diameter and 30 mm thick. The temperature gradient in each sample was measured with thermocouples in 3 radial holes (1.7 mm in diameter), drilled to the center of the sample. The hot and cold side temperatures were calculated by extrapolating these data. The gap conductance is obtained as shown in Table 1. 0 For the experiment with tin (Sn> as braze filler, a ceramic felt of 0.15 mm thick (outer diameter 25 mm, inner diameter 20 mm> was clamped between the two Cu samples to keep the clearance. A Sn foil of 20 mm in diameter and 0.15 thick was also clamped between the two Cu samples, surrounded by a ceramic felt. 0 The gap conductance of CL&U interface without Sn was also measured for comparison. Because it depends on the
52nd IAF Congress
apploximately 10 times greater than that of mechanical contact. Once heated over 232”c, the gap conductance of cooled specimens (Run No.7) still remain, because brazing of Cu/Cu samples is achieved. Experiments to confirm long life durability of liquid compliant pad are now underway. Cu/Sn reaction and vaporization of the braze tXer are to be investigated
pressure load on the contact interface, a pressure load of 0.08 MPa (0.8 kg/cm2 > was kept. Results are shown in Table 1. In case Sn is inserted between the Cu specimens below 232°C (Run No.51, the gap conductance is lower than that of mechanical contact (Run No.l-4). When heated over 232°C (Run No.6 and 8), the gap conductance is enhanced to
Table 1
169
Contact resistance of the liquid compliant pad.
Thermal Flux:
295.7
296.1
297.6
299.2
197.1 298.1
198.6 298.2
261.9
269.8
279.7
290.3
1615
294.2
196.3 294.0
33.8
26.3
17.9
8.9
35.6
3.9
2.3
Cold Side Temperature: Temperature Difference TH-Tc=AT (K)
Hot Duct (~600
m
degree C)
Thermal Expansion
FGMCompliant
Pad
I:_FFq
C” plate
m
Liquid Braze i.e. Sn (Melting Point 232 degree C)
Fig.8
I
Liquid compliant pad concept.
4.2
170
S2nd IAF Congress
5. BOND-FREE
COMPLIANT
PAD
Tin (Sn) infiltrated in porous graphite (Gr) for terrestrial systems l Sn infiltrated in porous silicon carbide (Sic) for terrestrial systems l CB2 infiltrated in porous Gr for space systems The operating temperature is over 232°C for Sn, and over 970°C for CB2. l
Another measure to accommodate greater thermal expansion mismatch and severe transient is the bond-free thermal compliant pad 14 as shown in Fig. 9. In the cold side, FGM compliant pads are inserted between TE cells and cold duct as another concepts. In the hot side, however, a Cu plate and a bond-free compliant pad as well as FGM compliant pads are provided. Bond-free compliant pad is made of porous material in which braze filler material is infiltrated. If heated over the melting temperature of the braze tiller, thermal gap conductance of the interface would be enhanced due to strong a&-&y of the braze fiuer with adjacent member. Sliding of this interface is possible so long as the braze filler is liquid. The advantage of this concept is that the braze filer comes out of the pores continuously to compensate loss of braze filler mainly due to vaporization. The thermal gap conductance would be kept as much as initially obtained. Therefore the long life durability could be assured. The following combinations are now under consideration.
6. CONCLUSION Manufacturing of the FGM compliant pad and overall assembling of the TE unit have been developed. To accommodate greater thermal expansion mismatch and severe thermal transient, liquid compliant pad and bondfree compliant pad are proposed. The following results are confirmed. l The thermal gap conductance at a Cu/BiTe interface bonded by HITASOL is eight times greater than that without HITASOL. l TE unit with FGM compliant pads assembled by HITASOL would provide electrical power more than twice as that of conventional TE unit.
C/L
FGM Compliant Pad
[W]
,
Cu plate Bond-Free Compliant Pad
; Hot Duct (~600 degree C) Thermal
Fig.9
Expansion
Bond-free compliant pad concept.
I I-p
52nd IAF Congress l
l
Liquid compliant pad (Sn as braze filler> can achieve the thermal gap conductance of about 10 times greater than that of mechanical contact (pressure load 0.08 MPa). Greater displacement of the hot and cold ducts including vibration can be accepted in this concept. Bond-free compliant pad will be another measure to accommodate greater displacement of the hot and cold ducts parallel to the duct surface.
TE energy conversion units by using compliant pads will meet the increasing demands for terrestrial and space power systems.
ACKNOWLEDGMENTS This work is being performed in the joint research program of CRIEPI and Hitachi Powdered Metals Co., Ltd., while development of the bond-free compliant pad was supported by the JAERI’s Nuclear Promotion Research Program. Manufacturing of FGM samples was achieved by Prof. A. Kawasaki of Tohoku University. Liquid compliant pad test was conducted at KE Technologie GmbH.
REFERENCES 1. Angelo, J. et al, “Space Nuclear Power,” Orbit Book Company, Inc., 1985. 2. Armijo, J. et al, “SP-100, Technology Accomplishments,” Proceedings of the 6th Symposium on Space Nuclear Power Systems, Albuquerque, NM, USA January 1989. 3. Bond, J.A. et al, “Development of a High Voltage Insulator Assembly for SP- 100,” 7th Symposium on Space Nuclear Power Systems, Albuquerque, NM, USA January 1990. 4. Bond, J.A. et al, “Evolution of the SP-100 Conductively Coupled Thermoelectric Cell,” 10th Symposium on Space Nuclear Power and Propulsion, Albuquerque, NM,
171
USA January 1993. 5. Jet Propulsion Laboratory, “First SP-100 Thermoelectric Cell Produces Power,” Space Power News, JPL D-5362 Issue 4, January 1990. 6. Kambe, M. et al, “FGM Compliant Pad Design for High Efficiency Silicon Cell,” Germanium Thermoelectric Proceedings of the Third International Symposium on Structural and Functional Lausanne, Gradient Materials, Switzerland, October 1994. Density 7. Kambe, M, “High Energy Thermoelectric Energy Conversion Systems by using FGM Compliant Pads for Space and Terrestrial Applications,” Proceedings of the 48th International Astronautical Congress, Turin, Italy, October 1997. 8. Kambe, M., “Design of High Energy Density Thermoelectric Energy Conversion Unit by using FGM Compliant Pads,” Functionally Graded Materials 1998, Trans Tech Publications Ltd., (Proc. of the 5th Int. Symposium on Functionally Graded Materials), Dresden, Germany, October 1998. Pad,” 9. Kambe, M., “FGM Compliant Patent 3056047, Japan, April 14,200O. Lifetime Fast 10. Kambe, M., “Long Spectrum Reactor for Lunar Surface Power System,” Proceedings of the 10th Symposium on Space Nuclear Power and Propulsion, Albuquerque, NM, USA January 1993. FGM 11. Kambe, M., “Symmetrical Compliant Pad,” Patent pending, Japan, Heisei g-29424. 12. Otsuka, A and Kawasaki, A., “Fabrication of Cu/AI203/Cu Symmetrical Functionally Graded Material by Spark Plasma Sintering Process,” J. of the Japan Society of Powder and Powder Metallurgy, Vol.45 No.3, pp.220.224, March 1998 (in Japanese). 13. Kambe, M., “Liquid Compliant Pad,” Patent pending, Japan, 2000-384423. 14. Kambe, M. and Kawasaki. A, “Bond-Free Compliant Pad,” Patent pending, Japan, 2000-001015.
www.elsevier.cotn/locate/actaastro
ActaAstronautica Vol. 51, No. 1-9. pp. 161-171, 2002 0 2002 International Astronautical Federation. Published by Elsevier Science Ltd All rights reserved. Printed in Great Britain SOO94-5765(02)00071-1 0094-5765/02 $ - see front matter
INTENSIVE ENERGY DENSITY THERMOELECTRIC ENERGY CONVERSION SYSTEM BY USING FGM COMPLIANT PADS Mitsuru KAMBE Central Research Institute of Electric Power Industry (CRIEPI) 11-1, Iwado Kita 2-chome, Komae-shi, Tokyo 201-8511, Japan E-mail:
[email protected]
Hideo SHIKATA Hitachi Powdered Metals Co., Ltd. 520, Minor&i, Matsudo-shi, Chiba, 270-2295, E-mail: [email protected]
Japan
1. INTRODUCTION
Abstract
In previous thermoelectric (TE) space power systems, such as the radioisotope thermoelectric generator (RTG) I), the TE devices were cantilevered from the heat sink (cold side). This design strategy reduced the stress in the TE device by not restricting thermal expansions, but required heat to be transferred l%om the heat source to the TE cells by radiation. As the thermal energy density available from most isotopes used in RTGs is quite low, radiation coupling of the TE cell to the heat source is an acceptable method and the RTG performance is not strongly affected by this radiation gap. However a severe mass penalty would occur if radiation coupling were to be used. Hence conductively coupling the TE module to the hot and cold heat exchangers is the most effective configuration to achieve high thermal energy density. The Conductively Coupled TE Cell with compliant pad was first developed by GE as part of the SP-100 Space Reactor Power System 2) which is designed to convert the nuclear power of a fast flux, lithium cooled space reactor into electrical power in the range of lo’s to 100’s of kW. In the SP-100 TE power conversion system 2,3,4+),TE cell uses a multicouple configuration which is conductively coupled to both the heat source and the heat sink. At operating temperatures,
In order to provide increasingly large amounts of electrical power to space and terrestrial systems with a suf6cient reliability at a reasonable cost, thermoelectric energy conversion system by using Functionally Graded Material (FGM) compliant pads has been focused. To achieve high thermal energy density in thermoelectric (TE) power conversion systems, conductively coupling the TE module to the hot and cold heat exchangers is the most effective configuration. This is accomplished by two sets of FGM compliant pads. This design strategy provides (1) a high flux, direct conduction path to heat source and heat sink, (2) the structural flexibility to protect the cell from high stress due to thermal expansion, (3) an extended durability by a simple FGM structure, and (4) manufacturing cost reduction by spark plasma sintering. High thermal energy density of more than twice as much as conventional conduction coupling TE generator is expected. TE energy conversion systems combined with FGM compliant pads
for space and terrestrial design options are proposed in this paper. o 2002 International Astronautical Federation. Published by Elsevier Science Ltd. All rights reserved.
the
161
dome
shaped
deformation
of the
TE
162
52nd IAF Congress
module, in addition to the differential thermal expansion between hot and cold exchangers, must be accommodated without inducing large stress into the sensitive cell elements. This is accomplished by two sets of the compliant pads 3,*,5).Key parameters for these compliant pads are high thermal conductivity, low stiffness in shear and compression, and the ability to limit peak stress. In the SP- 100 TE system, niobium @III) filaments planted on Nb facesheet cotiguration is adopted for the compliant pads. These compliant pads are very delicate structure because the filament diameters of the pads are roughly ten times less than the thickness of a human hair. If all the filaments in a l-inch by l-inch pad for a 100 kWe system were connected endto-end, the total filament length would span 3.5 times around the earth. This project, however, was canceled in 1993. TF system was also adopted for terrestrial applications. However, most of the systems are “conventional” conduction coupling in which no compliant pad is adopted and TE devices were placed between the hot duct (heat source> and cold duct (heat sink) with a pressure load. In this configuration, the at TE device/duct thermal resistance interface is greater, therefore poor electrical output is available. Increasing the pressure load would reduce the thermal resistance, however, it would damage fragile TE cell. Under these circumstances, the author pad design of proposed a compliant Functionally Graded Material @GM) 6. 7,8, 91 to achieve high power density TE energy conversion systems for space and terrestrial applications. FGM compliant pad consists of Cu layers for stress relaxation and an Al203 layer for electrical insulation. This design strategy provides (1) a high flux, direct conduction path to heat source and heat sink, (2) the structural flexibility to protect the celI from high stress due to thermal expansion, (3) an extended durability by a simple FGM and (4) manufacturing cost structure, reduction by powder sintering process. The most important in FGM compliant pad
design is that all the elements should be bonded together to eliminate thermal resistance at the interfaces. In this paper, design and performance of the proposed TE device with FGM compliant pads are presented. Other design options to accommodate higher operating temperature (i.e. greater thermal expansion mismatch) and severe thermal transient are also demonstrated. Proposed TE units can be applied to both space and terrestrial power systems 10). 2. TE UNIT WITH FGM COMPLIANT PADS 2.1 Design ADproach Fig. 1 illustrates multicouple SiGe TF unit with Cu/aluminaOUzOa)lCu symmetrical FGM compliant pads designed for both terrestrial and space power systems. In order to reduce edge stress of dissimilar bonded materials by differential thermal expansion, a l-inch by l-inch module, same size as that of SP- 100, has been adopted. Requirements for the compliant pad stress relaxation layer are a low stiffness in shear and compression, and a good conductor of heat. An attention was focused to select material of smaller E/ 1 where E is modulus of elasticity, and ;1thermal conductivity.
CulAlg, I cu Symmetrical FGY Graphite SiGe Insulator Graphite CulAI& I cu Symmetrical FGM
(Cold side)
Fig. 1 Concept of multicouple TE unit with Cu/AlpO&u FGM compliant pads.
52nd IAF Congress
Therefore copper (Cu) was selected. Because the melting temperature of Cu is 108X, the operating temperature of the unit is less than 800 ‘C if it is operated in the vacuum environment. For terrestrial systems, the operating temperature of less than 5OO’C (hopefully under 400°C) is recommended to avoid oxidation of Cu layer. The E/ ;1 value is independent of the porosity P of the compliant pads because their effective modulus of elasticity is roughly equal to PE, and effective thermal conductivity P ;1. Porosity of approximately 0.8 is expected in our proposed sintering process (refer to chapter 2.2). Alumina 6A.lzO3) layer has a role of an electrical insulator. Because the linear thermal expansion coefficient of Cu is twice as much as Al203, it is impossible to braze Al203 plate to Cu electrode. The essential feature of FGM is that there is no interface between Cu and A1203 layer, and cu/&03 concentration normal to the surface varies graduahy. In this structure, thermal stress due to thermal expansion mismatch of Cu and A203 can be easily accommodated 67). Graphite (Gr) layer acts as a chemical barrier between the SiGe and Cu layers. Cu/AlzOs/Cu symmetrical FGM 1i.1was designed to achieve easy assembling of the whole unit as well as to obtain structural integrity in the FGM compliant pad. In the Conventional cd~203 FGM, bonding Of an heat exchanger duct (usually made of aUSt.enitiC Steel, Inconel or Cu) and 4203 side of the FGM should be conducted carefully because the linear thermal expansion coefficient of the former is nearly twice as large as the latter. In case of symmetrical FGM, bonding of a heat exchanger duct and an outer surface of the symmetrical FGM (Cd is generally easier. Another advantage of the symmetrical FGM is that the dome shaped deformation due to the thermal expansion mismatch between Cu and Al303 layers of FGM during processing and operation could be avoided. This feature will ensure long life durability of the TE unit. though Even the thickness of the
163
symmetrical FGM is greater than that of cu/A1203 FGM, only a slight increase (l-2%) in overall thermal resistance is anticipated because most of the symmetrical FGM is composed of Cu. Accordingly the effect of the thermal resistance on the TE performance is only a few percent. 2.2 Manufacturing of FGM compliant pad Manufacturing of Cu/AhO3/Cu FGM by spark plasma sintering (SPS) has been attempted 19. SPS was recently developed to obtain a fine sintering structure at a relatively lower processing temperature by inducing plasma among material particles to The following developing issues be sir&red. should be investigated: 0 Powder sintering process should be done below the melting temperature of Cu (1083°C). l Sintering at 1250 “c under 50 MPa pressing is preferable for Al203 because higher density of A1203 is required to achieve a high flux conduction path and structural integrity. In order to meet both requirements, “solid/liquid phase” SPS has been applied. In this process, sample is gradually heated up to 1250°C while considerable attention has been made to avoid loss of liquefied Cu in the sample. In view of this, a punch pressure was held 50 MPa just prior to reach melting temperature of Cu, and after that suddenly decreased to 25 MPa. An Al203 sintering density of 80% has been achieved without any crack in the FGM layers. Sufficient electrical insulation capability in the direction normal to FGM layers has been confirmed. 2.3 Overall Assembling In order to ensure function of the TE system, all the members of TE unit should be interconnected to eliminate thermal resistance at the interfaces. Peculiar requirements for assembling TE unit are that bonding temperature should be higher than the operating temperature (~800°C) of the element to be bonded, while it should be
164
52nd IAF Congress
lower than the melting temperature of Cu (1083 “c ). Therefore bonding should be conducted in the temperature range between 850°C and 1050°C. In view of this, diffusion bonding cannot be applied. Brazing has the advantage that only lower contact loads and lower bonding temperatures are necessary compared to diffusion bonding. An active metal braze material based on the Ag/Cu/ln/Ti system was adopted. Preliminary bonding tests for ail interfaces have been made: Brazing of Gr to Cu side of FGM is a proven technique in several industrial applications, while the combination Gr/SiGe is more critical: higher difference of the thermal expansion modules, better chemical affinity between active metal braze and SiGe. Brazing of SiGe/Gr/Cu can be made using “two step brazing cycle”: first brazing of Gr to Cu at lOOO”c,including the wetting of Gr on its other side with an active metal braze, and then brazing of wetted Gr to SiGe in a second step with a lower working temperature (850°C). Activity of titanium 0% is starting at temperatures about 900-950 “c , therefore brazing of wetted Gr to SiGe can be done in a second step without damage of the primary wetting, if the brazing temperature is lower than 900°C. Brazing of Cu side of FGM to the heat exchanger duct (usually made of austenitic steel, Inconel or Cu) is also proven. Metallographic inspection revealed a sound brazing structure. In case of assembling BiTe TE unit, which is most popular for terrestrial applications, colloidal graphite (trade name HITASOL by Hitachi Powdered Metals) can be used. It is sub-micron graphite powder immersed in alcoholic solvent. HITASOL also acts as a chemical barrier between BiTe and Cu layers. Sufficient bonding strength of up to 700°C is proven. Therefore it also applied to terrestrial SiGe TE units. It features significant reduction of the thermal and
electrical resistance at the interface. The thermal resistance at a Cu/BiTe interface bonded by HITASOL is eight times less than that without HITASOL (refer to chapter 3). The de&rite advantage of HITASOL is that bonding can be done at room temperature just like glue. Sufficient bonding strength is available after one day, if dried in room temperature, or dried up in the hot air circulating furnace of 150°C for 15 minutes. HITASOL could achieve easy assembling of multicouple TE unit in which many bonding interfaces exist. Recently use of lead (Pb) braze filIer is restricted in view of the environmental protection, while nothing is as yet satisfactory as an alternative. Therefore advantage of non-toxicity is another HITASOL. Fig.2 shows unicouple BiTe TE unit with Cu/Al~0&u compliant pad assembled by HITASOL.
8mm
*
Fig.2 Sample of a unicouple TE unit with Cu/Al2O&u FGM compliant pads.
165
52nd IAF Congress
3. PERFORMANCE OF THE TE UNIT WITH FGM COMPLIANT PADS 3.1 Gap Conductance Unicouple BiTe TE units with and without FGM compliant pads were tested t.c measure temperature distribution in the unit. Fig. 3 illustrates conventional TE unit without compliant pad It consists of BiTe TE cell, Cu electrodes and a mica sheet as an electrical insulation, and assembled by a screw. l In the first test run, no interface is bonded at all to accommodate thermal expansion mismatch. l In the second test run, the hot side electrode (Cu) and BiTe are bonded by HITASOL for comparison. Usually this TE unit is sandwiched between the hot @eat source) and cold (heat sink) ducts, however, an infrared lamp is adopted instead of the hot duct in our experiment. Difference in this configuration does not affect on the measurement of gap conductance. Thermocouples of 70 ~1 m diameter are spot-welded on the TE unit members to obtain temperature proflle. Because of their very small diameter, effect of heat loss due to these thermocouples on the temperature profile is neglected. Gap conductance between the hot side electrode (Cu) and BiTe is given by the following equations.
ATi2 = <[++
;J
***(l)
qL AT23=
.*.(2>
obtains
1
6
LATiz -
-= ~TEAT
h
a-lx 0.4
4.2 AT12 -
= L
1.2
1.2 ATzs
=(3.5ATi2 /ATzs
10-3
x
-0.333)X
I
10-S
(m21QW) .-e(3)
where Temperature
difference between TCl-TC2: ATi2 ac>
Temperature
difference between TC2-TC3: ATzs (K>
Heat flux:
q(w)
Distance defined in Fig. 3: L=4.2
X
10-a (m)
6 =0.4x 10-z (m) TE cell cross section: Thermal conductivity
A= 16 x 10-G (rn2) of BiTe : ATE =1.2 (W/mIo
Gap Conductance:
h (W/m@
The first and second test runs revealed the following gap conductance between the hot side electrode (Cu) and BiTe. Therefore the thermal gap conductance at a CuBiTe interface bonded by HITASOL is eight times greater than that without HITASOL. hr=715 (W/m@: hH=5836 (W/m%):
without HITASOL with HITASOL
AilTE
Eliminating
q/A in above equations, one
It should be noted that absolute values of hx and hH are not reliable because heat loss due to radiation in radial direction is
52nd IAF Congress
166
dominating in this experimental setup (e.g. TE cell cross section of each leg is only 4 mm x 4 mm and temperature measurement is done on the TE cell surface). Therefore only a relative value of h g and hn can be discussed. Fig. 4 shows a TE unit with FGM compliant pads. All the members are bonded other by each HITASOL. Three thermocouples of 70 P m diameter are also spot-welded on the TE unit. Measured temperature is shown in the figure. Temperature profile deduced by hn described above is also plotted by dotted line in the figure. Measured and deduced temperature profiles coincide close together.
3.2 Comparison of the TE units Based on the gap conductance obtained above, performances of the TE units with and without FGM compliant pads are compared as shown in Fig. 5. l Model A is a conventional unit which consists of BiTe TE cell, Cu electrodes and mica sheets as electrical insulations, and sandwiched between the hot and cold
ducts with a pressure load. This unit has six gaps. No HITASOL layer exists. (If these gaps are filled with HITASOL, thermal stress relaxation is impossible and the TE unit would be broken.) 0 Model B is similar to Model A, however, Cu electrodes are bonded to BiTe by HITASOL. This unit has four gaps. 0 Model C is a proposed TE unit with FGM compliant pads. This unit has no gap because four interfaces are filled with HITASOL. This is a case that hot and cold duct temperature are 240 and 2o”C, respectively. Heat flux: q achievable by each model is shown in the figure. Model C can achieve heat flux of more than twice as those of model A and B. Therefore TE unit with FGM compliant pads would provide electrical power more than twice as that of conventional TE unit. Fig. 6 indicates the thermal resistance in the TE unit. Thermal resistance of gaps in model A and B dominate TE performance. Temperature gradient in each model is presented in Fig. 7.
n
Infrared Lamp
Infrared Lamp
0
...-....
ttttt
Axial Location
Experiment Calculated profile
uw cu
TCl TC2
BiTe cu Mica
TC3
I
Cooling Duct
TC4
tTested TE Unit with FGM Compliant Pads (HITASOL in all the Interfaces)
Fig.3
Experimental setup of a conventional TE unit.
Fig.4
Experimental setup of a TE unit with FGM compliant pads.
167
52nd IAF Congress
q=27.7 W/in’ q=12.9 W/in2
q=lO.5 W/id
240°C
1,Ht”“C”“tDtu”‘t
1
(0.1 mm)J
I Model A Conventional TE Unit
Cooling,Duct
I
1
Cooling Duct
1 20°C
Model C Model B Conventional TE Unit with FGM Compliant Pads (HITASOL in CuBiTe Interfaces) (HITASOL in all the Interfaces)
Fig. 5
Conduction coupling type TE unit designs.
6 Gaps
’
Fig.6
E g
MidCu
BiTe
BiTe
BiTe
Model A
ModelB
Mode/ C
Thermal resistance of the T’Eunit.
220
4 HITASOL ~ayei-s /2
*oO-
al 5
2
4 Gaps
6 Gaps
S o!
i
loo-
k?
E
X4calC”
z
BiTe Model A
Fig. 7
FGH Compliant Pads
/
2 HITASOL Lv=
‘Mica/C”
BiTe
BiTe Model B
Model %
Temperature gradient in the TE unit.
168
52nd IAF Congress
4. LIQUID
COMPLIANT
PAD
Greater the capacity of the TE energy conversion system, larger the heating and cooling ducts between that TE units are clamped. In addition the operating temperature of the TE power conversion systems for space applications is geneSally higher than those for terrestrial systems. Differential thermal expansion of the hot and cold ducts parallel to the duct surface is given by the following relation. 6==/3LAT .+0 where 6 : differential thermal expansion B : linear thermal expansion modulus of the duct material L: distance from the Axed point to the duct edge (if duct is supported at its centerline, L is half of the duct width 1 AT: difference between the hot and cold ducts temperature l
l
In case of niobium @lb) hot/cold ducts (B = 7.3 x lo-6K-1) which is operated at 906 and 5OO”c, respectively, and L=ZOO mm (duct width 400 mm), 6 will be 0.6 mm. This is a typical condition for SiGe space power system designed for SP- 100 s,4,5). In case of austenite steel (SUS316) hot/cold ducts (B = 16.7X IO-SK-i) which is operated at 240 and 2O”c, respectively, and L.=300 mm (duct width 600 mm), 6 will be 1.1 mm. This is a typical condition for BiTe TE system for terrestrial applications.
FGM compliant pad cannot accommodate such greater thermal expansion mismatch. thermal Severe transient in startup/shutdown cycle is another dif&ulty to be encountered in terrestrial TE systems. In addition vibration of the ducts normal to the duct surface would be expected if applied to the exhaust duct of the automobile. Liquid compliant pad 19) should be used together with FGM compliant pad to accommodate such severe conditions. Concept is illustrated in Fig. 8. In the cold side, FGM compliant
pads are inserted between TE cells and a cold duct as shown in the preceding chapter. In the hot side, however, we have a Cu plate and a braze filler layer as well as FGM compliant pads. In case of BiTe TE system, tin (Sn) is adopted as braze filler material. Because the melting temperature of Sn is 232°C) it is liquid during operating temperature. Due to strong affinity of Sn with Cu, the thermal gap conductance of Cu/Sn/Cu interface is enhanced, while displacement of this interface is possible because Sn is liquid. Therefore differential thermal expansion of the hot and cold ducts parallel to the duct surface can be relieved perfectly. Vibration of the ducts normal to the duct surface is also accepted. FGM compliant pad in this configuration has a role of accommodating thermal stress induced in the temperature difference of 232”Croom temperature. For space applications, CB2 braze filler (trade name of Goodfellow GmbH, 96Ag-4Ti, melting temperature 970°C) is one of the candidates under consideration. Due to the surface tension of braze filler, it could be held in the interface even in the non-gravity field. Experiment to demonstrate liquid compliant pad performance has been done. Test sample consists of two Cu cylinders of 25 mm in diameter and 30 mm thick. The temperature gradient in each sample was measured with thermocouples in 3 radial holes (1.7 mm in diameter), drilled to the center of the sample. The hot and cold side temperatures were calculated by extrapolating these data. The gap conductance is obtained as shown in Table 1. 0 For the experiment with tin (Sn> as braze filler, a ceramic felt of 0.15 mm thick (outer diameter 25 mm, inner diameter 20 mm> was clamped between the two Cu samples to keep the clearance. A Sn foil of 20 mm in diameter and 0.15 thick was also clamped between the two Cu samples, surrounded by a ceramic felt. 0 The gap conductance of CL&U interface without Sn was also measured for comparison. Because it depends on the
52nd IAF Congress
apploximately 10 times greater than that of mechanical contact. Once heated over 232”c, the gap conductance of cooled specimens (Run No.7) still remain, because brazing of Cu/Cu samples is achieved. Experiments to confirm long life durability of liquid compliant pad are now underway. Cu/Sn reaction and vaporization of the braze tXer are to be investigated
pressure load on the contact interface, a pressure load of 0.08 MPa (0.8 kg/cm2 > was kept. Results are shown in Table 1. In case Sn is inserted between the Cu specimens below 232°C (Run No.51, the gap conductance is lower than that of mechanical contact (Run No.l-4). When heated over 232°C (Run No.6 and 8), the gap conductance is enhanced to
Table 1
169
Contact resistance of the liquid compliant pad.
Thermal Flux:
295.7
296.1
297.6
299.2
197.1 298.1
198.6 298.2
261.9
269.8
279.7
290.3
1615
294.2
196.3 294.0
33.8
26.3
17.9
8.9
35.6
3.9
2.3
Cold Side Temperature: Temperature Difference TH-Tc=AT (K)
Hot Duct (~600
m
degree C)
Thermal Expansion
FGMCompliant
Pad
I:_FFq
C” plate
m
Liquid Braze i.e. Sn (Melting Point 232 degree C)
Fig.8
I
Liquid compliant pad concept.
4.2
170
S2nd IAF Congress
5. BOND-FREE
COMPLIANT
PAD
Tin (Sn) infiltrated in porous graphite (Gr) for terrestrial systems l Sn infiltrated in porous silicon carbide (Sic) for terrestrial systems l CB2 infiltrated in porous Gr for space systems The operating temperature is over 232°C for Sn, and over 970°C for CB2. l
Another measure to accommodate greater thermal expansion mismatch and severe transient is the bond-free thermal compliant pad 14 as shown in Fig. 9. In the cold side, FGM compliant pads are inserted between TE cells and cold duct as another concepts. In the hot side, however, a Cu plate and a bond-free compliant pad as well as FGM compliant pads are provided. Bond-free compliant pad is made of porous material in which braze filler material is infiltrated. If heated over the melting temperature of the braze tiller, thermal gap conductance of the interface would be enhanced due to strong a&-&y of the braze fiuer with adjacent member. Sliding of this interface is possible so long as the braze filler is liquid. The advantage of this concept is that the braze filer comes out of the pores continuously to compensate loss of braze filler mainly due to vaporization. The thermal gap conductance would be kept as much as initially obtained. Therefore the long life durability could be assured. The following combinations are now under consideration.
6. CONCLUSION Manufacturing of the FGM compliant pad and overall assembling of the TE unit have been developed. To accommodate greater thermal expansion mismatch and severe thermal transient, liquid compliant pad and bondfree compliant pad are proposed. The following results are confirmed. l The thermal gap conductance at a Cu/BiTe interface bonded by HITASOL is eight times greater than that without HITASOL. l TE unit with FGM compliant pads assembled by HITASOL would provide electrical power more than twice as that of conventional TE unit.
C/L
FGM Compliant Pad
[W]
,
Cu plate Bond-Free Compliant Pad
; Hot Duct (~600 degree C) Thermal
Fig.9
Expansion
Bond-free compliant pad concept.
I I-p
52nd IAF Congress l
l
Liquid compliant pad (Sn as braze filler> can achieve the thermal gap conductance of about 10 times greater than that of mechanical contact (pressure load 0.08 MPa). Greater displacement of the hot and cold ducts including vibration can be accepted in this concept. Bond-free compliant pad will be another measure to accommodate greater displacement of the hot and cold ducts parallel to the duct surface.
TE energy conversion units by using compliant pads will meet the increasing demands for terrestrial and space power systems.
ACKNOWLEDGMENTS This work is being performed in the joint research program of CRIEPI and Hitachi Powdered Metals Co., Ltd., while development of the bond-free compliant pad was supported by the JAERI’s Nuclear Promotion Research Program. Manufacturing of FGM samples was achieved by Prof. A. Kawasaki of Tohoku University. Liquid compliant pad test was conducted at KE Technologie GmbH.
REFERENCES 1. Angelo, J. et al, “Space Nuclear Power,” Orbit Book Company, Inc., 1985. 2. Armijo, J. et al, “SP-100, Technology Accomplishments,” Proceedings of the 6th Symposium on Space Nuclear Power Systems, Albuquerque, NM, USA January 1989. 3. Bond, J.A. et al, “Development of a High Voltage Insulator Assembly for SP- 100,” 7th Symposium on Space Nuclear Power Systems, Albuquerque, NM, USA January 1990. 4. Bond, J.A. et al, “Evolution of the SP-100 Conductively Coupled Thermoelectric Cell,” 10th Symposium on Space Nuclear Power and Propulsion, Albuquerque, NM,
171
USA January 1993. 5. Jet Propulsion Laboratory, “First SP-100 Thermoelectric Cell Produces Power,” Space Power News, JPL D-5362 Issue 4, January 1990. 6. Kambe, M. et al, “FGM Compliant Pad Design for High Efficiency Silicon Cell,” Germanium Thermoelectric Proceedings of the Third International Symposium on Structural and Functional Lausanne, Gradient Materials, Switzerland, October 1994. Density 7. Kambe, M, “High Energy Thermoelectric Energy Conversion Systems by using FGM Compliant Pads for Space and Terrestrial Applications,” Proceedings of the 48th International Astronautical Congress, Turin, Italy, October 1997. 8. Kambe, M., “Design of High Energy Density Thermoelectric Energy Conversion Unit by using FGM Compliant Pads,” Functionally Graded Materials 1998, Trans Tech Publications Ltd., (Proc. of the 5th Int. Symposium on Functionally Graded Materials), Dresden, Germany, October 1998. Pad,” 9. Kambe, M., “FGM Compliant Patent 3056047, Japan, April 14,200O. Lifetime Fast 10. Kambe, M., “Long Spectrum Reactor for Lunar Surface Power System,” Proceedings of the 10th Symposium on Space Nuclear Power and Propulsion, Albuquerque, NM, USA January 1993. FGM 11. Kambe, M., “Symmetrical Compliant Pad,” Patent pending, Japan, Heisei g-29424. 12. Otsuka, A and Kawasaki, A., “Fabrication of Cu/AI203/Cu Symmetrical Functionally Graded Material by Spark Plasma Sintering Process,” J. of the Japan Society of Powder and Powder Metallurgy, Vol.45 No.3, pp.220.224, March 1998 (in Japanese). 13. Kambe, M., “Liquid Compliant Pad,” Patent pending, Japan, 2000-384423. 14. Kambe, M. and Kawasaki. A, “Bond-Free Compliant Pad,” Patent pending, Japan, 2000-001015.