See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/262639688

Thermal contact resistance measurements

Article  in  Journal of Enhanced Heat Transfer · January 2012

DOI: 10.1615/JEnhHeatTransf.2012006005

CITATIONS READS
6 3,808

2 authors, including:

Yun Mo
Tokyo Electron Miyagi Ltd.
6 PUBLICATIONS   33 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Yun Mo on 13 November 2017.

The user has requested enhancement of the downloaded file.

 Journal of Enhanced Heat Transfer, 19 (6): 561–569 (2012)

THERMAL CONTACT RESISTANCE MEASUREMENTS

Yun Mo∗ & Sumie Segawa

Technology Development Center, Tokyo Electron Ltd., 17 Miyukigaoka, Tsukuba-shi, Ibaraki 305-
0841, Japan

∗
Address all correspondence to Yun Mo E-mail: yun.mo@tel.com

We started with a conventional steadystate direct method to measure thermal contact resistances (TCR) of Cu/Cu,
SS304/SS304 and Al2 O3 /Al2 O3 , but later found that it is not applied to measure the TCR of thin test pieces. In the
study of thermal resistance of the interface materials with graphite sheets and gel sheets with the direct method, we found
that graphite sheet can decrease TCR at high contact pressure but not at low contact pressure while a gel sheet has a
constant thermal resistance in both vacuum and atmospheric pressure This led us to develop an indirect method which
can measure TCR without inserting temperature sensors into the test pieces. The indirect method was successfully
applied to measure the TCR of Cu/Cu. In order to completely remove the additional TCRs introduced in the indirect
method, we developed a new method by attaching additional temperature sensors directly onto the backsides of the test
pieces. The TCR of Si/Al2 O3 was measured with the new method. Both the indirect method and the new method can be
used to measure the TCR of thin test pieces.

KEY WORDS: thermal resistance, thermal contact resistance, thermal conductivity, thermal interface ma-
terials

1. INTRODUCTION come very different in those cases. Many unusual mate-

rials such as silicon, alumina, and quartz are often en-
With the increasing wafer size and decreasing design rule countered. Many surface treatments are often used such
evolutions, wafer temperature control is becoming in- as thermal spraying, electroplating, or chemical vapor de-
creasingly stringent. Because of extremely reduced cham- position to prevent contamination or corrosion. And some
ber material contamination requirements, direct wafer rare gases such as helium and argon are also fed into pro-
temperature measurements and controls are hardly car- cess chambers in very low vacuum. Thus there is no avail-
ried out in production lines to prevent metal contamina- able database or proven theories to give TCRs. Further-
tions. Susceptor temperature measurements and controls more, very light contact pressures are also far beyond the
are still mainstream. Because the high thermal contact re- application ranges of many TCR models.
sistance (TCR) between wafer and susceptor in vacuum Therefore, we have to develop an apparatus to satisfy
environments leads to high power consumption to heat those needs. We start with a conventional steadystate di-
or chill the wafer it becomes crucial to enhance the heat rect method to measure TCR of copper/copper, 304 stain-
transfer between wafer and susceptor in some way to de- less steel/304 stainless steel and alumina/alumina, and
crease the TCR Knowing the TCR between wafer and sus- thermal resistance of a graphite sheet and a gel sheet. We
ceptor is, therefore, becoming very important in thermal then develop an indirect method to measure the TCR of
design, control and simulations. copper/copper when investigating the thermal resistances
There has been much research on TCR measurement of interface materials with the direct method. Finally we
and prediction models (Ross and Stouter, 1962; Thomas invent a new method which can be used to quickly deter-
and Probert, 1972; Yovanovich and Hegazy, 1983; Tien, mine TCR and thermal resistance for thin test pieces at a
1968; Sridhar and Yovanovich, 1994). The situations very light compression load with air, nitrogen, helium and
in semiconductor manufacturing processes, however, be- argon in both vacuum and atmospheric pressure.

1065-5131/12/$35.00 ⃝
c 2012 by Begell House, Inc. 561
562 Mo & Segawa

NOMENCLATURE

a coefficient in Eq. (4), = 1.30 and 1.27 q 1 , q2 heat fluxes in the upper and the lower rod [W/m2 ]
in atmosphere and vacuum, respectively R thermal contact resistance or thermal
A area of test piece [m2 ] resistance [m2 ·K/W]
b coefficient in Eq. (4) [N], = −5199 N Ra roughness [µm]
and −39.34 N in atmosphere and RA , RB , thermal resistances [m2 ·K/W]. Also includes
vacuum, respectively RGEL , the thermal contact resistances on the
F contact load [N] RGEL1 , upper and lower surfaces for the GEL
G weight in Eq. (3) [N] RGEL2 sheets only in Eqs. (9) and (10)
k thermal conductivity [W/m·K] RA/B thermal contact resistance of A and B [m2 ·K/W]
k1 , k2 ,
thermal conductivities [W/m·K] t, tA , tB thickness [m]
kA , kB
P contact pressure [KPa] T temperature [◦ C]
q heat flux [W/m2 ] Tc1 , Tc2 interface temperature [◦ C]

2. MEASUREMENT PRINCIPLES AND due to TCR. Heat flux q through the interface is the av-
APPARATUS erage of the fluxes q1 and q2 in the upper and lower test
pieces as shown in Eq. (1) where q1 and q2 e obtained with
2.1 Direct Method the Fourier law by taking the temperature gradients from
the linear correlations of the temperatures at 10 measuring
We used a steadystate method to measure TCR as shown points as shown in Fig. 1. The interfacial surface temper-
in Fig. 1. Heat flows onedimensionally from the upper to atures of the upper and lower test pieces Tc1 and Tc2 are
the lower test piece and temperature distributes linearly obtained by extrapolating the temperatures as shown in
within each of the test pieces but jumps at the interface Fig. 1 and their difference ∆T is also shown therein The
TCR is calculated in Eq. (2).

q = (q1 + q2 )/2 (1)

R = ∆T /q (2)
Figure 2 sketches the TCR measurement apparatus #1
that we first developed in this study. The apparatus mainly
consists of heating and cooling rods, temperature sensors,
thermal insulation, heater, chiller, load handling mecha-
nisms and controllers. Five platinum resistance temper-
ature detectors (RTD) of 0.5 mm diameter are inserted
and adhered with silver paste into the holes drilled in the
heating and cooling rods in equal intervals to measure
the temperature distributions of the heating and cooling
rods. The temperature sensors were specially made and
carefully calibrated at three standard points at 0, 100 and
200◦ C with accuracy of 0.01◦ C. Five sensors in the heat-
ing and cooling rods can ensure an accurate measurement
in case one or two of them lose adhesion with the rods
FIG. 1: Schematic sketch of physical situation in vacuum There is an electric heater on the heating rod

Journal of Enhanced Heat Transfer

Thermal Contact Resistance Measurements 563

FIG. 2: Sketch of TCR measurement apparatus #1.

and a Peltier cooler under the cooling rod through heat Flower ) with an addition load cell in vacuum and atmo-
spreaders. Goldplated stainless plates as radiation shields spheric pressure, respectively. Contact pressure is the con-
surround the rods to prevent heat loss from the rod to tact load divided by the area of the cross section.
the surroundings via radiation. In order to minimize lat- The heating and cooling rods are made from copper or
eral temperature difference of the shields to the rods as 304 stainless steel, 25 mm × 25 mm in cross section and
much as possible, the radiation shields direct contact to 60 mm in length by referring to the research of Culham et
the heat spreaders. Furthermore, thermal insulation made al. (2002).
from PEEK (polyether ether ketone) covers the radiation Because of the difficulty to accurately control the con-
shields, heaters and coolers. tact load at very low contact load with Eq. (4), we devised
The heater and the heating rod with their insulations another apparatus as shown in Fig. 3, where the loads
are fixed to a frame, called a lifting arm in Fig. 2, which are manually applied by adding several weights to get the
is used to lift and press the heating rod to get a specified specified contact force. Several different weights are pre-
contact force F at the interface by controlling the com- pared in the chamber and can be caught and placed onto or
pressing force Fupper and the lifting force Flower . Fupper removed from the heating rod via a handling mechanism
and Flower are measured with an upper load cell in the outside the chamber. In order to reduce the weight of the
chamber and a lower load cell outside the chamber, which
are automatically imposed with two pneumatic cylinders.
Once Flower is manually set, Fupper is automatically cal-
culated via a microcomputer in Eq. (3) and controlled via
a controller so that a specified contact force F can be at-
tained.
F = Fupper − Flower + G (3)
where G is the weight of the lifting arm including the
heating rod, upper radiation shields and insulation. Be-
cause there are frictions between the pneumatic cylinders
and the seal parts, different in vacuum and atmospheric
pressure, not considered in Eq. (3), we used Eq. (4) to
control the contact force instead of Eq. (3)
F = a(Fupper − Flower ) + b (4)
where a and b were obtained with the leastsquare method
previously by measuring a series of F and (Fupper − FIG. 3: Sketch of TCR measurement apparatus #2.

Volume 19, Number 6, 2012

564 Mo & Segawa

heating rod we adopted an alumina heating rod, 32 mm in We found that the TCR of a gel sheet as thermal inter-
length, shorter than the cooling rod which is also made of face material (TIM) hardly varies with contact pressure
alumina. A very light ceramic plate heater of 100 W and and gas pressure when we measured the thermal resis-
2 mm in thickness was adhered on the heating rod. In- tance of the gel sheet, as will be illustrated later We then
cluding other thermal insulations, the total weight of the realized that we could obtain the TCR of the test pieces A
heating rod is 170 g, which is the least load that would be and B, often measured with the direct method as shown
applied on the interface. In this way, the vibrations of the in Fig. 4(a), indirectly by first measuring the thermal re-
contact force in controlling load can also be avoided. The sistance of a gel sheet between the heating and cooling
contact load is obtained as Eq. (5). rods as shown in Fig. 4(b) and then measuring the total
thermal resistance as shown in Fig. 4(c), provided that the
F =W +G (5) same gel sheets with the same TCR are used as that in
Fig. 4(b). The TCR with the indirect approach thus can
where G = 170 g force = 1.67 N. be obtained as the following relations:
The temperature dependence of thermal conductivity
was also considered with Eqs. (6) and (7) for copper and RA/B = Rtotal − RGEL1 − RGEL2 − RA − RB (9)
stainless steel (SS304), respectively, in calculating heat
fluxes q1 and q2 in Fig. 1 and Eq. (1) by correlating the where
data in the JSME (1986) handbook RGEL = RGEL1 = RGEL2 (10)
RA = tA /kA , RB = tB /kB (11)
kCu (T ) = −2.3571 × 10−16 T 6 − 5.7590 × 10−13 T 5
Notice that RGEL1 and RGEL2 have already included
+ 1.9695 × 10−9 T 4 − 1.6880 × 10−6 T 3
the TCRs between heating rod/gel sheet and cooling
+ 5.6712 × 10−4 T 2 − 0.12103T + 400.87 (6) rod/gel sheet as obtained in Fig. 4(b).
The merit of this indirect approach is that there is
no need to insert many temperature sensors into the test
kSS304 (T ) = −2.6400 × 10−8 T 3 + 3.8139 × 10−5 T 2
pieces A and B to get the heat flux and interface temper-
− 4.0853 × 10−4 T + 15.987 (7) atures as shown in Fig. 4(a) and measuring the TCR will
be much easier for thin test pieces to which it is diffi-
The conductivity of the alumina rods is given in Eq. (8) cult to attach temperature sensors. The thermal resistance
by fitting the experimental data provided by its maker. of a TIM between A and B can also be investigated in
this way. In addition, this indirect method can be used to
kAl2 O3 (T ) = 8.7203 × 10−10 T 4 − 1.1994 × 10−6 T 3
measure thermal conductivity of a test piece instead of A
+ 6.5912 × 10−4 T 2 − 1.9318 × 10−1 T + 36.911 (8) and B The drawback of this method is that it requires that
the thermal resistance of all gel sheets be the same, as
2.2 Indirect Method
We noted that the direct measurement of TCR as shown in
Fig. 1 is very complicated and tedious. Many holes have
to be accurately drilled and temperature sensors have to be
carefully adhered to the rods for every pair of test pieces.
It will be difficult to have every sensor really sealed in
holes because any air trapped in the holes will destroy
the contacts between the sensors and the rods in vacuum
and lead to a large measurement error although it is not
so apparent in atmospheric pressure. There is another is-
sue to measure the TCR of thin test pieces. It is very dif-
ficult, sometimes impossible, to precisely attach sensors
to a thin test piece. However, there are frequent requests
to measure the TCR of thin test pieces in many applica- FIG. 4: Direct and the indirect method to measure TCR:
tions. Therefore, we have to invent some methods to an- (a) direct method; (b) measuring the thermal resistance of
swer these requests. a gel sheet; (c) measuring the total thermal resistance.

Journal of Enhanced Heat Transfer

Thermal Contact Resistance Measurements 565

shown in Eq. (10), which may be difficult to satisfy in the measuring TCR for thin test pieces with the direct meth-
case where the surfaces of A and B facing the gels differ ods and greatly increases the efficiency of the measure-
greatly from the surfaces of the heating/cooling rods. ments. This new method of directly measuring the tem-
perature difference of the backsides of the test pieces also
2.3 A New Method removes the uncertainty errors with the indirect method
which brings additional TCRs between heating/cooling
In order to solve the issues in the indirect method, we de- rods and the test pieces with the gel sheets In addition,
veloped a new method by adhering additional two temper- this method can be used to measure thermal conductivity
ature sensors to the backsides of the test pieces as shown by attaching two temperature sensors on both sides of a
in Fig. 5. The TCR of the test pieces A and B can be ob- test piece instead of the test pieces A and B.
tained from the relations described in Fig. 5 as follows:

R = Rall − RA − RB (12) 3. EXPERIMENTAL RESULTS AND DISCUSSION

TCR and thermal resistance measurement were carried
where Rall is the thermal resistance between two temper- out in steady states, which were established after 12 h
ature sensors as follows: for the first time starting from room temperature to the
specified temperature by controlling the temperatures of
Rall = ∆T /q (13)
the heater on the heating rod and the Peltier cooler on the
RA and RB are the intrinsic thermal resistances of the test bottom of the cooling rod. About two hours are needed to
pieces, obtained with the formulas as shown in Fig. 5. ∆T reach the next steady state in the following experiments.
is the temperature difference between these two additional Contact pressure is given as follows:
temperature sensors with the black bullets as shown in
P = F/A (14)
Fig. 5 and q is the heat flux through the interface. The
heat flux q is obtained in the same way as mentioned for where A is the area of the test piece, 25 mm × 25 mm =
the direct method. 0.625×10−3 m2 in this study.
The heating and cooling rods this time just play a role
in measuring the heat flux. This method, however, largely
3.1 TCR Measurements with the Direct Method
reduces the complexity and overcomes the difficulties of
Figure 6 shows the contact pressure dependence of TCR
of copper/copper of Ra = 0.3 µm with the direct method
as shown in Figs. 1 and 2 in vacuum of 0.1 Pa and atmo-
spheric pressure, respectively. The TCR data in vacuum

FIG. 5: Principle of the new method to measure TCR:

(a) arrangement of the test pieces; (b) relations of the FIG. 6: TCR of copper/copper in vacuum and atmo-
TCR and the thermal resistances around the test pieces. spheric pressure.

Volume 19, Number 6, 2012

566 Mo & Segawa

(red circles) and atmospheric pressure (blue squares) are changes of number and size of contact spots in SS/SS con-
also fitted with exponential forms as solid red and blue tact than in Cu/Cu contact due to its larger surface hard-
lines, respectively. It can be seen that TCR in vacuum is ness. Another reason may be the macro contacts from the
largely dependent on contact pressure because the size large waviness of the stainless steel surfaces because the
and number of real contact spots largely vary with the power of the exponential −0.365 is close to −0.33 as sug-
contact pressure at low contact pressure. The TCR in at- gested by Thomas and Sayles (1975).
mospheric pressure is lower and less dependent on contact By measuring the TCRs of heating and cooling rods
pressure than that in vacuum because the heat conduc- made from copper and stainless steel as shown in Figs. 6
tion of the air confined between contact surfaces partly and 7 we found that TCRs at a very low contact pres-
decreases the TCR and does not change largely due to sure shows rather large variations. Though the hysteresis
less variation of gas volume with contact pressure. was also observed in our experiments in the loop of in-
In addition other experimental results in vacuum creasing and decreasing contact force as well as in the ex-
(shown in red triangles) with a transient experimental periments of Fenech and Rohsenow (1963), Williamson
method (Fujishiro et al., 1996) are plotted in Fig. 6 for the and Majumdar (1992), and Madhusudana and Williams
contact pressure near zero. The fitting equation in vacuum (1973), we note that the load control at very low contact
in this study can also well express the data obtained with force might cause a control accuracy problem because we
the transient method. control the contact force F by subtracting the lifting force
We found that the temperature jump at the interface Flower from the pressing force Fupper in Eq. (3) or Eq. (4)
is much larger than the temperature differences between as shown in Fig. 2, which are directly obtained from two
temperature sensors in the copper heating and cooling load cells. Any variation of Fupper may cause a large mea-
rods. This will lead to a small temperature gradient in surement error in small contact force F . We tried to solve
the rods and affect the accuracy of measuring the heat this problem by making another apparatus as shown in
flux. We modified the TCR measurement apparatus #1 in Fig. 3, in which the loads are applied by placing several
Fig. 2 by replacing the copper rods with stainless steel weights to achieve the specified value. Figure 8 shows
(SS304) rods. Figure 7 shows the TCR of the stainless the TCR of the alumina heating rod and cooling rod with
rods in vacuum with red circles and in atmospheric pres- apparatus #2 as shown in Fig. 3. The variations of TCR
sure with white squares by the direct measurement. Cor- at very low contact pressure were improved to a much
relation equations are also given in Fig. 7. The TCRs of smaller level compared to the results shown in Figs. 6 and
SS/SS are generally larger than those of Cu/Cu as shown 7.
in Fig. 6. And the contact pressure dependence of TCR of
SS/SS as shown in Fig. 7 is not as strong as Cu/Cu in vac- 3.2 Thermal Resistances of TIM
uum as shown in Fig. 6. This may result from the fewer
In many applications, it is often required to reduce the
TCR It can be achieved by inserting some TIM of high

FIG. 7: TCR of stainless steel/stainless steel in vacuum FIG. 8: TCR of alumina/alumina in vacuum and atmo-
and atmospheric pressure. spheric pressure.

Journal of Enhanced Heat Transfer

Thermal Contact Resistance Measurements 567

conductivity and high flexibility between the contact sur- TCR could not be reduced; its compliance increases as
faces. We investigated the thermal resistance of a carbon contact pressure increases and the effect of reducing TCR
sheet from Panasonic Corporation (2011) and a gel sheet appears. The graphite sheet can effectively reduce TCR in
from Taica Corporation Ltd. (2008) as TIM at very light vacuum when a high contact pressure is applied.
contact pressure below 160 KPa in vacuum and atmo- Figure 10 shows the results of inserting a gel sheet
spheric environments. Because of the high sensitivity of (product name: λ-GEL) of Taica Corporation Ltd. (2008),
the thermal resistance to the contact pressure at low con- of 0.5 mm in thickness, at the interface of copper and cop-
tact pressure a maximum contact pressure of 160 KPa was per. This gel is a kind of silicone gel mixed with ceramic
first applied in order to improve the interfacial compliance powder of high thermal conductivity. The conductivity of
before changing contact pressure. the gel sheet is 6.5 W/m·K. TCR can be greatly reduced
Figure 9 shows the results with a pyrolytic graphite due to its extreme softness and tackiness. The sheet type
sheet (PGS) from Panasonic Corporation (2011), 0.1 mm of the gel gives it a convenient usage. The TCRs of Cu/Cu
in thickness, between copper and copper rods with the di- as shown in Fig. 6 are also plotted in Fig. 10 at the same
rect method in vacuum (denoted by white diamonds) and time. We found that the TCRs of the gel sheet remain
atmospheric (denoted by asterisks) environments. Corre- constant at 2.46×10−4 m2 ·K/W no matter how small a
lation equations are also given with brown and black solid contact pressure is applied and whatever gas pressure is
lines. The TCRs of Cu/Cu in Fig. 6 are also plotted in fed after a rather large initial contact pressure of 160 KPa
Fig. 9 for the sake of comparison to the results without is applied. This is can be explained by the fact that the
the graphite sheets. gel sheet completely adheres to the surfaces and the gas
The TCRs of Cu/Cu are reduced about 25% by using trapped in the interfaces cannot be driven out even in a
the graphite sheet in atmospheric pressure. The graphite very high vacuum.
sheet, however, functions differently in vacuum, depend- The thermal resistance with the gel sheet is higher than
ing on contact pressure applied. The graphite sheet can- the TCR of Cu/Cu at atmospheric pressure. This may
not reduce the TCR at all at very low contact pressure come from the low thermal conductivity and the thickness
below 30 KPa while it greatly decreases the TCR as con- of the gel sheet itself.
tact pressure increases. The thermal resistance in vacuum Noticing the constant thermal resistance of the gel
with the graphite sheet as TIM reaches the level of that at sheet and its independence on contact pressure and gas
atmosphere when the contact pressure exceeds 160 KPa. pressure, we developed an indirect method to measure
This phenomenon may be explained by the compliance of TCR as shown in Fig. 4.
the graphite sheet at different contact pressure: its rigidity Figure 11 shows the results with the gel sheet at the in-
could not be overcome at very low contact pressure and terfaces of 304 stainless steel and 304 stainless steel, and

FIG. 9: TCR of copper/copper and the thermal resistance FIG. 10: TCR of copper/copper and the thermal resis-
with a graphite sheet (t = 0.1 mm) in vacuum and atmo- tance with a gel sheet (t = 0.5 mm) in vacuum and atmo-
spheric pressure. spheric pressure.

Volume 19, Number 6, 2012

568 Mo & Segawa

the indirect method are the same as in Fig. 10. The test
pieces A and B are copper, 3 mm in thickness and Ra =
0.3 µm. It can be seen that the TCR of copper/copper ob-
tained with the indirect method agree well with the direct
method, both in atmospheric pressure and in vacuum Us-
ing the gel sheets makes the indirect method feasible.

3.4 TCR Measurements with the New Method

Silicon and alumina are widely used materials in semi-
conductor manufacturing equipment to process silicon
wafers in annealing, chemical vapor deposition (CVD),
and etching processes. Figure 13 shows the TCRs of
Si/alumina, 3 mm in the same thickness, obtained by the
FIG. 11: TCR of stainless steel/stainless steel and the new method. The results obtained with this new method
thermal resistance with a gel sheet (t = 0.5 mm) in vac- also show the same trends of TCR variation with contact
uum and atmospheric pressure. pressure in both vacuum and atmosphere for alumina and
silicon as those with the direct and indirect methods for
Cu/Cu and SS/SS. In the range of contact pressure below
Ra = 0.3 µm. The thermal resistances of the gel sheet also
60 KPa, the TCR of alumina and Si can be fitted with the
hold constant in both vacuum and atmospheric pressure
following formulas:
but the value in vacuum (= 2.51×10−4 m2 ·K/W), close to
that within copper and copper (= 2.46×10−4 m2 ·K/W), R = 2.26 × 10−3 P −0.0705 (at 1.0 atm) (15)
is a little lower than that in atmospheric pressure
(= 3.48×10−4 m2 ·K/W). This may be due to more gas R = 1.70 × 10−2 P −0.102 (at 1.0 Pa) (16)
being contained in the interface of the stainless steel and
the gel sheet surfaces in atmospheric pressure than in vac-
uum. The gel sheets can effectively reduce TCR within 4. CONCLUSIONS
stainless steel surfaces both in vacuum and atmospheric In order to solve the unknown TCR problem in the devel-
pressure. opment of semiconductor manufacturing equipment, we
started with a conventional steadystate direct method to
3.3 TCR Measurements with the Indirect Method measure TCR of Cu/Cu, SS304/SS304 and Al2 O3 /Al2 O3 ,
Figure 12 shows the experimental result comparisons with but later found that it is not applied to measuring the TCR
the indirect and the direct method. The gel sheets used in of thin test pieces.

FIG. 12: Comparison of the TCR of copper/copper with FIG. 13: TCR of alumina/silicon in vacuum and atmo-
the direct and the indirect method. spheric pressure.

Journal of Enhanced Heat Transfer

 Thermal Contact Resistance Measurements 569

In the study of thermal resistance of interface mate- 2762, 1996.

rials with graphite sheets and gel sheets with the direct JSME, JSME Data Book: Heat Transfer (4th Ed.), in Japanese,
method, we found that graphite sheet can decrease TCR Tokyo, JSME, pp. 314–318, 1986.
at high contact pressure but not at low contact pressure Madhusudana, C. V. and Williams, A., Heat flow through metal-
while a gel sheet has a constant thermal resistance in both lic contacts-The influence of cycling the contact pressure, in
vacuum and atmospheric pressure This led us to develop Proc. of 1st Australian Conf. on Heat Mass Transfer, pp. 33–
an indirect method which can measure TCR without in- 40, Monash University, Melbourne, Australia, 1972.
serting temperature sensors into the test pieces. The in- Panasonic Corporation, PGS Graphite sheet, Retrieved May
direct method was successfully applied to measuring the 9, 2011, http://industrial.panasonic.com/www-data/pdf/
TCR of Cu/Cu. AYA0000/AYA0000CE2.pdf, 2011.
In order to completely remove the additional TCRs Ross, A. M. and Stouter, R. L., Heat transfer coefficient between
introduced in the indirect method, we developed a new UO, and Zr-2, AECL-1552, 1962.
method by attaching additional temperature sensors di- Sridhar, M. R. and Yovanovich, M. M., Review of elastic and
rectly onto the backsides of the test pieces. The TCR of plastic thermal contact conductance models: Comparison
Si/Al2 O3 was measured with the new method. Both the with experiment, J. Thermophys Heat Transfer, vol. 8, no.
indirect method and the new method can be used to mea- 4, pp. 633–640, 1994.
sure the TCR of thin test pieces. Taica Corporation Ltd., The catalogue of the GEL sheet COH
The new method we developed in this study has been series, Retrieved May 9, 2011, http://www.taica.co.jp/gel-
used to study the factors that affect TCRs and thermal re- english/products/use/index04.html, 2008.
sistances of interface material in the development of semi- Thomas, T. R. and Probert, S. D., Correlations for thermal con-
conductor manufacturing equipment. tact conductance in vacuo, ASME J. Heat Transfer, vol. 94,
pp. 276–281, 1972.
Thomas, T. R. and Sayles, R. S., Random process analysis of
REFERENCES the effect of waviness on thermal contact resistance, Prog.
Astronaut. Aeronaut., vol. 39, pp. 3–20, 1975.
Culham, J. R., Teertstra, P., Savija, I., and Yovanovich, M. M., Tien, C. L., A correlation for thermal contact conductance of
Design, assembly and commissioning of a test apparatus for nominally-flat surfaces in a vacuum, in Proc. of 7th Thermal.
characterizing thermal interface materials, in Proc. of Eighth Conductivity Conf., pp. 755–759, U.S. National Bureau of
Intersociety Conf. on Thermal and Thermomechanical Phe- Standards, Gaithersburg, MD US, 1968.
nomena in Electronic Systems, pp. 128–135, May 29-June 1, Williamson, M. and Majumdar, A., Effect of surface deforma-
2002, San Diego, CA USA, 2002. tions on contact conductance, J. Heat Transfer, vol. 114, pp.
Fenech, H. and Rohsenow, W. M., Prediction of thermal conduc- 802–810, 1992.
tance of metallic surfaces in contact, J. Heat Transfer, vol. 85, Yovanovich, M. M. and Hegazy, A. A., An accurate univer-
pp. 15–24, 1963. sal contact conductance correlation for conforming rough
Fujishiro, H., Ikebe, M., Nakasato, K., and Naito, T., Proposal of surfaces with different micro-hardness profiles, in Proc. of
three terminal method for low temperature thermal diffusivity AIAA 18th Thermophysics Conf., no. 831434, June 1–3, 1983,
measurement, Czech. J. Phys., vol. 46, suppl. S5, pp. 2761– Montreal, Canada, 1983.

Volume 19, Number 6, 2012

View publication stats