Sunteți pe pagina 1din 4

Measurement, Significance and Application of Thermal Properties of Thermal Interface Materials

Using ASTM D5470


Michael H. Bunyan and Miksa de Sorgo
Chomerics Division of Parker Hannifin Corp.
77 Dragon Court, Wobum, MA 01801
(mbunyan@parker.com, mdesorgo@parker.com)

Abstract Thermal conductivity, k, for one dimensional heat flow


through an isotropic material is defmed by Fourier’s law.
This paper describes the ASTM D5470 ( I ) test method for
measuring thermal transmission properties of interface k = ( Q / A ) x (dlAT) (2)
materials. Details of the equipment, the methodology and the Substituting (1) into (2) and rearranging yields equation
care required for accurate power and . temperature (3).
measurements are discussed. The difference between a R
= d/k (3)
material property as determined by D5470,and an application Equation 3 shows that if the thermal resistance, R, is
specific property as measured under actual in-situ conditions measured at several thickness d, the slope of the resulting R
will be discussed in detail. The paper will conclude by vs. d c w e is equal to the inverse of the thermal conductivity,
showing examples of how a thermal property determined by lk.
D8470 can be used to obtain fust order approximation of Test System Description
performance in a specific application.
The apparatus for measuring thermal resistance of
Nomenclature interface materials based on the cut bar method is shown
A area, cm2 schematically in Figure 1.
D thickness, cm Guard Heater
Q heat flow, watt Insulation
T temperature, Kelvin Heater
k thermal conductivity, Wlm-K
R material thermal resistance, K-cm2/W
Rc material contact thermal resistance, K-cm2/W Upper Calorimeter
e system thermal resistance, KiW Interface Material
Introduction
ASTM D5470 ( I ) test method has been developed and Lower Calorimeter

Y-4
promulgated over the last 18 years by some of the interface
materials suppliers for one specific reason. Both end-users
and suppliers realized that the lack of such a standard led to a Cooling Unit
variety of thepal test methods that somehow all resulted in
the same high thermal perfnrmance data for a whole range of U
thermal interface materials. The Defense General Supply Figure 1. Schematic of the thermal test fature
Center in Richmond, VA initiated the development of a
standard test method for thermal properties within the It consists of a guarded heater, upper and lower
framework of ASTM. The work was started in 1984 and the calorimeters and a thennostated cold plate. The test fixture is
test method was finally published in 1993 as D8470, placed in a press to apply a compressive load to the specimen
‘Thermal Transmission Properties of Thin, Thermally being tested.
Conductive Solid Electrical Insulation Materials”. The guarded heater consists of an aluminum guard heater.
D5470 is intended to measure these properties under separated from the main aluminum heater by a thermal
strictly controlled conditions allowing different laboratories to insulator. Both heaters are equipped with wire wound
generate equivalent thermal data. D5470 measures the cartridge heaters powered by regulated dc power supplies.
thermal resistance and the thermal conductivity of a material Each heater is equipped with a thermocouple for temperature
directly. monitoring.
Thermal resistance, R, is measured by establishing a The calorimeters are 28.7 mm diameter, 35 mm long
known, one-dimensional heat flux, Q/A, acrnss a specimen aluminum cylinders. Each is equipped with two type T
and measuring the temperature difference, AT, across the thermocouples located 25 nun apart and 5 mm from the
specimen at equilibrium. The thermal resistance of the edges. The thermocouples are inserted into 1 mm diameter
specimen is the ratio of the temperature difference to the heat wells that go to the center h e of the calorimeter and are
flux. secured in the wells with either a thermally conductive epoxy
R = AT x ( A / Q ) (1) or silicone compound. The thermocouples are used to
determine to surface temperature of the calorimeters in
contact with the test material by extrapolation as described in
ASTM D8470. The surfaces of the calorimeters are flat to
0-7803-7793-1/02/$17.00 0 2003 E E E 119 19th IEEE SEMI-THERM Symposium
about 2 micron / cm and have a surface fmish of
approximately 1.6 micron.
Thermocouples are calibrated at 0 "C in an ice bath and
near 100 "C at the boiling point of water using primary
reference standard thermometers. These thermometers are
calibrated against NIST traceable equipment and are available
from any scientific supply house.
The cooling unit is a water-cooled block of aluminum. Its
temperature is controlled by a thermostated water bath
capable of 0.1 K temperature regulation.
Test Procedure
Thermal resistance of an interface material is measured by
placing the specimen between the calorimeter surfaces,
applying the required compressive load, and applying power
to the guard and main heaters. The guard heater temperature 0 0.4 0.8 1.2
is kept within +/- 0.2 K during the test to prevent heat loss. Thickness, cm
The system is considered to have reached equilibrium when
the main heater temperature is constant over a 5-minute Figure 2. Plot of thermal resistance vs. thickness for an
period. At this point the input power and the temperature electrically insulating silicone pad.
difference across the specimen are recorded and the thermal Form-Stable Elastomers
resistance is calculated.
The thermal conductivity of an electrically insulating
Thermal conductivity is measured, by determining the silicone pad was measured using this same procedure. The
thermal resistance at different specimen thickness. Thickness
test specimen thickness was varied by plying layers of
is varied by plying up different number of layers of material
material to form stacks of 1, 2, 3, 4 and 5 layers. Thermal
to form stacks of 1,2, 3,4 and 5 layers. Sufficient pressure is resistance was measured at a 3.4 MPa compressive load. The
applied during the test to eliminate any interlayer contact thickness, d, of each compressed stack was measured after the
resistance. Unsupported material can be prepared by molding
test and was used for the calculations. The results are shown
at different thickness. Materials that are liquid at test in Table 2 and Figure 3.
temperatures will require a compression stop. The thermal
resistance of these stacks is measured and the plot of
thickness vs. thermal resistance Table 2. Thermal resistance vs. thickness of an electricallv
insulating silicone pad
Measurement Capability Layers Thickness, cm R,K-cm"V
The capability of this method can be shown by measuring 1 0.024 1.06
the thermal properties of a thermal conductivity standard. 2 0.048 1.99
Austenitic stainless steel reference material, SRM 1462,
3 0.070 2.66
obtained from NIST was machined into 2.9 cm diameter disks
of 0.25, 0.38, 0.76 and 1.02 cm thickness. Dow Coming DC 4 0.098 3.52
340 thermal grease was used to minimize contact resistance 5 0.120 4.40
between the disks and the calorimeters. The results are shown
in Table I and Figure 2. The slope of the R vs. d curve in Figure 3 is ~33.9which
corresponds to k = 2.95 W/m-K. Figure 3 also shows that at 0
Table I . Thermal resistance vs. thickness of SRMl462 thickness the thermal resistance is 0.29 K-cm21W. This
resistance is the contact resistance between the material and
Thickness, em Thermal Resistance, K-cm'/W
the calorimeters. This shows that even at compressive loads
0 0.04
as high as 3.5 MPa, there is still some contact resistance
0.25 1.87 between the silicone elastomer and the metal surfaces.
0.38 2.56 The thermal properties for the above material measured by
0.76 5.21 this method are its thermal resistance, 1.06 K-cm'm, and
1.02 6.83 thermal conductivity, 2.95 W/m-K. Thermal conductivity is
independent of fie condition of the calorimeter surfaces. The
The slope of the curve in Figure 2 is equal to the inverse thermal resistance value however is the sum of all the
of the thermal conductivity of the SRM 1462. The intercept resistance between the calorimeters, the resistance of the
is the contact resistance between the greased SRM surface and material as well as the contact resistances at the surfaces.
this set of calorimeters. Using least mean squares, the data Thus the measured thermal resistance can be apparatus
above results in 14.99 W/m-K for the thermal conductivity of dependent.
SRM 1642. The value published by NIST at 50 "C is 14.74
Wlm-K which represents a 1.7 % variance

120 19th IEEE SEMI-THERM Symposium


This material was developed for mounting discrete
S-l semiconductors to beat sinks using 2 to 4 MPa mounting
pressures. Figure 3 shows that most of the contact resistance
is eliminated above about 1.5 MPa. At low compressive loads
the contact resistance of this type of material can be much
greater than the resistance of the material itself.
Liquid I Phase Change Materials
Measuring thermal properties of liquid interface materials
requires that the sample thickness being measured be defined
by some mechanical means such as a compression stop. To
demonstrate this technique, precision gauge stainless steel
wire was used to set the gap between the calorimeters. Wire
I I I thickness of 25 to 250 microns was used. The results are
0 0.04 0.08 0.12 shown below in Table 4 and Figure 5.
Thickness, cm

Figure 3. Plot of thermal resistance vs. pressure for an


electrically insulating silicone pad.

The role of contact resistance at the calorimeter surfaces


was examined by measuring the thermal resistance of this
material at differing compressive loads. Samples were placed
between the calorimeters, a load was applied and its resistance
was measured. A new specimen was used for each load level
tested. The results are shown in Table 3 and Figure 4.

Table 3. Thermal resistance vs. compressive load


Pressure, MPa R, K-cm’N 1
0.034 5.28
0.069 4.57 0 0.01 0.02 0.03
0.103 3.66 Thickness, cm
0.172 3.04
Figure 5 . Plot of thermal resistance vs. thickness for
0.344 2.26 thermal grease.
0.690 1.68
1.379 1.30 Table 4. Thermal resistance vs. thickness of DC340
2.068 1.15 Thermal Grease
3.448 1.06
Thickness, cm R, K-cm’m
0.0024 0.47
0.0076 1.32
0.0127 2.01
0.0254 3.96

The slope of the R vs. d curve is 151.6, which corresponds


to a thermal conductivity of 0.66 Wlm-K. The intercept at
d=O is 0.11 K-cm2N and shows that there may be some
contact resistance between the grease and the calorimeter
surfaces. Alternately, non-flat calorimeter surfaces could
account for the apparent contact resistance.
Discussion
o 1 I I I I A S T M D5470 provides a direct method for measuring
O I 2 3 4
thermal conductivity and thermal resistance of interface
Pressure, MPa
materials. The method also provides an estimate of contact
resistance between a material and a metal surface as a
Figure 4. Plot of thermal resistance vs. thickness for function of pressure.
thermal grease. Parameters that are key to obtaining accurate values are
the heat flux, the temperamre difference and sample
thickness. Heat flux passing through a sample is defined as

121 19th IEEE SEMI-THERM Symposium


the electrical power applied to the cartridge heaters. Proper longer comparable in this new application. What is needed is
insulation is needed to guard against heat losses from the a Pressure vs. R curve for both products to provide additional
heater to the sample. A guard heater surrounding the heater information.
coupled with fibrous insulation along the length of the Similarly, comparison of two thermal greases of different
calorimeters are sufficient to limit heat losses to less than 2 thermal conductivity must also take into account the intended
percent at the normal test temperatures. application. The higher k grease is likely to have a higher
Accuracy of the temperature difference across a sample is viscosity then the lower k one. Depending on the
achieved by using calibrated thermocouples placed accurately compressive load, the higher viscosity grease may result in a
in the calorimeters. The calibration is performed not just on much thicker thermal joint that will negate the higher thermal
the thermocouple hut of the thermocouple / measuring conductivity.
system. Also, the dissipated power is set to provide as large a The thermal performance of phase change materials
AT as possible. (PCM) also exhibits this viscosity dependence. The objective-.
The sample thickness needs to be accurately measured in is to get the thinnest joint possible. Material properties such
order to calculate thermal conductivity. There are several as the PCM melt viscosity and filler particle size will have an
methods available to do this. The best method is to measure impact on the thermal joint thickness.
the thickness during the test when the sample is compressed Thermal data from D5470 can he used to get first order
using laser micrometers or similar methods to measure the approximation of thermal performance in specific
distance between the calorimeter surfaces. Alternately, the applications. Refer to the BN tilled silicone pad between a
compression deflection of the material can be measured on an TO220 and a heat sink. The interface material has a thermal
lnstron and the compressed thickness during the test can be resistance of 1 K-cm’N. The transistor is mounted to a heat
calculated. sink using a spring clip that applies 0.5 MPa pressure
For low modulus or liquid samples such as gap fillers, uniformly on the base. From Figure 3, the total theimal
greases or phase change materials, a compression stop of resistance including the contact resistance due to the low
some form can be used to provide the know separation pressure is 1.8 K-cm’iW. The effective heat transfer area of
between the calorimeters and thus the sample thickness. the TO220 is 1 cm2 and leads to a joint resistance of 1.8 WW.
Other variables that need to be controlled are the The experimentally measured Rc-s of this system is 2.08 WW
conditions of the calorimeter surfaces in contact with the (2) which is good agreement for such a rough approximation.
sample. They need to be maintained such that they are flat Taking into account the incomplete heat spreading within the
and have a smooth fmish. The press that provides the base of the transistor can make further refinements and closer
compressive load should apply the load evenly across the agreement with the measured results.
sample and avoid any offset loads that will cause a non- Conclusions
uniform contact resistance across a sample.
ASTM D5470 test method can be used to accurately
ASTM D5470 measures thermal resistance and thermal measure the thermal resistance and thermal conductivity of
conductivity of interface materials under well defined, interface materials. D5470 produces material properties
controlled conditions. When properly performed using obtained under controlled conditions so that they can be used
correctly operating instruments, the data generated by to compare similar materials. There are many interface
different laboratories shows good agreement, approximately 5 materials and their thermal performance in a specific
percent. D5470 data published by different suppliers on application is not limited to just their thermal properties.
similar materials can he used to compare the thermal Each application must he evaluated for the unique
properties of those materials in a specific application. requirements it places on the interface material. When all the
For example, take the case of a discrete semiconductor applicable factors are taken into account, the thermal
attached to a heat SI& using an insulating silicone pad. The performance of an interface material in a specific application
application calls out a compressive load of 2 MPa. One can be predicted.
silicone pad is Boron Nitride filed silicone rubber with a
thermal resistance of 1.0 K-cm2iW. The other is Alumina References
filled with 2.5 K-cm2iWresistance. Clearly the BN filled pad 1. D5470, “Thermal Transmission Properties of Thin,
should give the better thermal performance. This comparison Thermally Conductive Solid Electrical Insulation
is valid because both materials are similar with respect to the Materials”, American Socienty for Testing and Materials,
intended application. 2001.
Now lets take a look at two different insulating materials, 2. De Sorgo, Miksa, “Optimizing Cooling Solutions for
one the BN filled silicone pad, the other a Kapton film coated High. Power Devices”, PCIM Power Electronics
on each side with a phase change material. The thermal Proceedings, 1997.
resistance of each material is ahout 1 K-cm’iW. In the
application mentioned above, their thermal performance is
seen to be very close (2). However, if the compressive load is
reduced from 2 MPa to 0.1 MPa, the silicone pad develops a
very large contact resistance while the phase change material
has the same resistance at either compressive load. The
change in the application conditions has placed additional
requirements on the interface materials and so they are no
122 19th IEEE SEMI-THERM Symposium

S-ar putea să vă placă și