Influence of Thermal Cross Coupling at Power Modules

Influence of Thermal Cross Coupling at Power Modules
Ernö Temesi, Zsolt Gyimothy
Vincotech Kft., H-2060 Bicske, Kossuth Lajos u. 59
Abstract
The thermal impedance of the semiconductors in Power Modules is always measured for a single chip, without
the influence of other surrounding dies. This article describes the increase of the junction temperature of powersemiconductors due to the cross coupling of the Rth of components placed close to each other. The influence of
different module structures, such as baseplate-less modules, modules with baseplate, material-thicknesses and
different materials is provided.
0. Basic Module Construction and
Simulation Considerations
For the simulation the following basic module
constructions were used. The used chipset is a
150A – 158mm2 IGBT loaded with 85W and in
the case of a half-bridge configuration a 150A
FRED, loaded with 25W is used, which can be
estimated as typical for a fully loaded 150A
module in drive applications. For the different
simulations mainly the number of chips and the
position at the heatsink was alternated. In some
cases also the thickness of the substrate and the
thermal grease.
Tj
Si Chip
Sn 100um
Cu 300um
Al2O3 or AlN 640um
Cu 300um
Thermal Grease 30um
2mm
Tc
Th1
Heatsink Base 15mm
Th2
Rth to Air
Air
Figure 1: Module without baseplate
Tj
Si Chip
Sn 100um
Cu 300um
Al2O3 or AlN 380um
Cu 300um
Sn 100um
Cu 3mm
Thermal Grease
Tc
Th1
2mm
Introduction
Today power-applications are getting more and
more compact to save cost, space, and weight.
For the same reason the layout of PowerModules has to be optimized. Also the
integration factor is getting higher. In the past
single half bridges and separated rectifiers were
used, nowadays PIM´s are used. But this trend
is leading to a high concentration of thermal
losses, associated with a high influence of the
thermal cross coupling to the real thermal
impedance of the application. Since the thermal
impedance for the semiconductor of a powermodule is given only for single semiconductors
without cross coupling, it is important for the
designer of power electronic equipment to know
how much the thermal resistance will increase,
by chips and modules placed close together.
As integration and power concentration is going
forward it becomes more and more important for
the designer of frequency converters to deal also
with the influence of cross coupling. Therefore,
as long as no thermal cross coupling is provided
for Power Modules, the increase of thermal
impedance for the application has to be
estimated based on the internal distances of
semiconductors in the module.
Heatsink Base 15mm
Th2
Air
Rth to Air
Figure 2: Module with baseplate
1. Datasheet Values of the Thermal
Impedance Zth
The thermal impedance is highly influenced by
the used measurement method and used
interface materials for the characterization.
It can be measured as Tj-Tc, Tj-Th with
temperature measurement at module-back-side
or heat-sink-surface or Tj-Th with temperature
measurement inside the heat sink in a depth of
1-2mm under the module. Each method has his
pros and cons. Tj-Tc describes the module most
exact, but is highly influenced by the method of
measurement of the module backside
temperature. Tj-Th characterizes in addition the
thermal resistance from the module to the
heatsink, which it is closer to the real application,
but it is also influenced by the thickness and
material, which is used for the thermal
interconnection of the module to the heat sink.
ƒ Measurements and simulations for datasheets
were done under the following conditions:Fluid
cooled copper heat sink
ƒ Measurement of Tc at bottom surface of
module in the chip center
ƒ Measurement of Th1 at surface of heatsink,
center of the chip
ƒ Measurement of Th2 2mm deep in the heat
sink.
Type and thickness of the ceramic material
mainly influences the internal Rth of a module.
The internal resistance of a module with AL203
0,64mm without baseplate and 0,38mm AL2O3
with baseplate is nearly the same. For all
modules without baseplate the thermal
resistance of the thermal interface material is the
same. Here the module with base plate behaves
much better due to the thermal spreading of the
base plate. Also the internal Rth of the heatsink
is lower for modules with baseplate due to the
spreading of the base plate.
With a simulation in cross section the thermal
spreading of a base-plate can be clearly seen.
Rth
Tj-Tc
TjTh1
TjTh2
Al2O3 0,64mm without baseplate
grease (d=30 µm, λ=1 W/mK) best case
0,167
0,310
0,336
grease (d=50 µm, λ=0,6 W/mK) - typical
0,156
0,479
0,501
foil (d=70 µm, λ=1,2 W/mK) - typical
0,160
0,407
0,430
0,112
0,255
0,281
Al2O3 0,38mm wo baseplate,
grease (d=30 µm, λ=1 W/mK) best case
AlN without baseplate
grease (d=30 µm, λ=1 W/mK) best case
0,049
0,185
0,211
grease (d=50 µm, λ=0,6 W/mK) - typical
0,047
0,342
0,362
foil (d=70 µm, λ=1,2 W/mK) - typical
0,048
0,276
0,298
Figure 4: Al2O3 0,64mm
Al2O3 with baseplate
grease (d=30 µm, λ=1 W/mK) best case
0,159
0,220
0,230
grease (d=50 µm, λ=0,6 W/mK) - typical
0,154
0,267
0,273
foil (d=70 µm, λ=1,2 W/mK) - typical
0,156
0,248
0,256
Table 1: comparison of construction,
interface materials, and location of
temperature measurement
different module constructions
0,35
Rth inside Heatsink th1-th2
Figure 5 AlN 0,64mm
Rth thermal grease tc-th1
thermal resistance (K/W)
0,3
Rth module Tj-Tc
0,25
0,2
0,15
0,1
0,05
0
AL2O3 0,64
wo Basepl
AL2O3 0,38
wo Basepl
ALN 0,64 wo
Basepl
AL2O3 0,38
with Basepl.
Figure 3: comparison of construction and
interface materials
Figure 6 Al2O3 0,38mm, with baseplate 3mm
2. Thermal System Resistance of
Distributed and Close Chips
Six IGBT’s were mounted on Daces with and
without base plate. The arrangement was altered
from well distributed to compact packed. The
module was placed on an aluminum heat sink
200*300mm with a base thickness of 15mm and
a thermal resistance of 0,078K/W to air, which
would lead to a calculated heat sink temperature
of 80°C at 40°C ambient.
Figure 10: Al2O3 with base plate, distance
50mm
Between the chips the distance was reduced
from 100mm, to 50mm, 25mm, 10mm, 7,5mm,
5mm, and 2mm. The size of the base plate was
reduced in the same way for the modules with
base plate and DBC size respectively.
Figure 11: Al2O3 with base plate distance
25mm
At a distance of 25mm there is a overlapping of
the spreading area with equal temperature of the
baseplate between the chips.
Figure 7: Al2O3 wo. baseplate, distance
50mm
Figure 12: Al2O3 with base plate, distance
10mm
Figure 8: Al2O3 wo. baseplate, distance
25mm
At a distance of 10mm the baseplate
temperature got homogenous. A further
reduction of the distance would reduce the
spreading area around the chips and drastically
increase the thermal resistance between base
plate and heatsink.
Figure 15 will confirm the estimation. Between
100mm and 50mm distance the slope is the
same for all graphs. For less then 50mm the
slope off the chips on base plate is increasing
more compared to the modules without base
plate.
Figure 9: Al2O3 wo baseplate, distance 10mm
dist.
absolute value
Rth
Rth
Rth
TjTjTj-
Rth
Tj-
percent to 100mm distance
Rth
Rth
Rth
Rth
TjTjTjTj-Th
Th1
Th2
Th
Tc
Th1
Th2
mm
K/W
K/W
ALN wo basepl.
100
0,05
0,19
50
0,05
0,19
25
0,05
0,19
10
0,05
0,19
7,5
0,05
0,20
5
0,05
0,20
2
0,05
0,20
Al2O3 wo basepl.
100
0,17
0,32
50
0,17
0,32
25
0,17
0,32
10
0,17
0,32
7,5
0,17
0,32
5
0,17
0,32
2
0,17
0,32
Al2O3 with basepl.
100
0,16
0,20
50
0,15
0,20
25
0,15
0,20
10
0,15
0,22
7,5
0,15
0,23
5
0,16
0,24
2
0,16
0,26
K/W
K/W
%
%
%
%
0,23
0,23
0,23
0,23
0,23
0,24
0,25
0,39
0,47
0,58
0,70
0,74
0,77
0,83
0%
0%
0%
0%
0%
0%
1%
0%
0%
0%
0%
0%
1%
2%
0%
0%
0%
0%
0%
2%
7%
0%
20%
47%
80%
88%
98%
112%
0,35
0,35
0,35
0,35
0,35
0,36
0,37
0,51
0,59
0,70
0,83
0,86
0,90
0,95
0%
0%
0%
0%
0%
0%
1%
0%
0%
0%
0%
0%
1%
1%
0%
0%
0%
0%
0%
2%
5%
0%
15%
36%
61%
67%
75%
86%
0,22
0,22
0,22
0,23
0,25
0,25
0,30
0,27
0,34
0,46
0,62
0,67
0,74
0,85
0%
0%
0%
0%
0%
1%
4%
0%
0%
0%
7%
11%
17%
30%
0%
1%
2%
6%
14%
16%
36%
0%
25%
67%
129%
147%
171%
210%
1,2
thermal resistance (K/W)
Tj-Tc
Tj-Th1
0,7
AL2O3 w o Basepl
AL2O3 w ith Basepl.
0,6
0,4
0,2
20
40
60
chip distance (mm)
80
Tj-Th2
0,6
3. Thermal Resistance of a 1/2-Bridge IGBT
Module
The half-bridge simulation was carried out under
following conditions:
2* IGBT 158mm2 (Pv=85W each)
2* FRED 53mm2 (Pv=25W each)
Tested configurations:
• Al2O3 without baseplate
• ALN without baseplate
• Al2O3 with baseplate, Al2O3 0,38mm,
Cu-base plate 3mm thickness
Aluminum heatsink 100*200mm*15mm, Rth to
ambient 0,18K/W, which leads to 80°C
calculated heatsink temperature at 40°C ambient
temperature.
0,5
0,4
0,3
0,2
0,1
0,0
0
20
40
60
80
100
distance (mm)
Figure 13: Al2O3 without baseplate
1,0
Figure 16: Half Bridge without baseplate,
simulation
thermal resistance (K/W)
0,9
Tj-Th
0,8
Tj-Tc
0,7
Tj-Th1
0,6
Tj-Th2
0,5
0,4
0,3
0,2
0,1
0,0
0
20
40
60
distance (mm)
Figure 14: Al2O3 with baseplate
80
100
Figure 15: comparison of AlN, Al2O3 with,
and without base-plate
1,0
Tj-Th
ALN w o Basepl
0,8
0
All modules showed a strong increase of Rth TjTh due to the bad spreading of the heatsink.
Modules without base-plate have no increase of
Rth down to Thj2, whereas modules with baseplate have an increase due to the reduced
possible spreading of the baseplate as expected.
0,8
1
0
Table 2: decreased distance between chips
0,9
thermal resistance (K/W)
Tc
100
Table 3: Rth of Half-Bridge
Th2-Th
Th1-Th2
0,600
Tc-Th1
Tj-Tc
Figure 17: Half bridge without baseplate,
thermal measurement
thermal resistance (K/W)
0,500
0,400
0,300
0,200
0,100
0,000
Al2O3 wo basepl
AlN wo basepl
Al2O3 with basepl
Figure 19: Rth of Half-Bridges
4. Thermal Resistance of a Six Pack
Configuration
Figure 18: Half Bridge with baseplate, thermal
measurement
The module with base-plate, is a better
spreading in the base-plate compared to the
spreading in the Al heatsink. This decreases also
the thermal resistance of the spreading inside
the heatsink. It can be seen at the low Th2-Th
value of the baseplate module compared to the
modules without baseplate. The other
resistances kept nearly the same values as in
the datasheet simulation.
Rth
Rth
Rth
Rth
Tj-Tc
Tj-Th1
Tj-Th2
Th2-Th
(K/W)
(K/W)
(K/W)
(K/W)
Al2O3 wo basepl
0,157
0,287
0,326
0,259
AlN wo basepl
Al2O3 with basepl
0,046
0,153
0,168
0,217
0,206
0,241
0,254
0,186
In the following 3 * Half-Bridges, with 2* IGBTs
(Pv 85W each) and 2*FREDs (Pv 25W each) are
tested.
Tested configurations:
• ALN without baseplate
• Al2O3 without baseplate
• Al2O3 without baseplate.
• Al heat sinks 0,06K/W,
200*300mm*15mm.
Figure 20: 3 Half-Bridges w/o baseplate,
distance 50mm, thermal measurement
Figure 21: cross cut of 3 Half-Bridges w/o
baseplate simulation
datasheet and the Half-Bridge. Only the thermal
resistance for the spreading within the heatsink
was increased depending on the distance of the
Half-Bridges.
0,60
Figure 22: 3 Half-Bridges, distance 5mm,
thermal measurement
Thermal Resistance (K/W)
AlN
Al2O3
0,50
Al2O3 w. Basepl.
0,40
0,30
0,20
0,10
0,00
0
10
20
30
40
Distance (mm)
50
60
Figure 26: spreading of heat-sink Th2-Th
Figure 23: 3 Half-Bridges dist. 1mm,
simulation, simulation
Thermal Resistance (K/W)
1,00
AlN
0,90
Al2O3
0,80
Al2O3 w. Basepl.
0,70
0,60
0,50
0,40
0,30
0,20
0,10
0,00
0
10
20
30
40
Distance (mm)
50
60
Figure 27: total Rth, Tj-Th
Figure 24: cross-cut, 3 Half-Bridges, dist.
1mm, simulation
The cross-section in figure 24 shows that the
temperature doesn’t decrease so much vertically,
but in horizontal direction it decreases rapidly.
This leads to the estimation, that increasing of
the thickness of the heat-sink base or the usage
of a better thermal conducting material could
reduce the thermal resistance.
In order to proof this, a three-phase inverter
bridge with 3mm base-plate was mounted on
heat-sink bases with different thicknesses, and
different materials.
Figure 25: Six-Pack with 3mm base-plate,
DBC distance 1mm, thermal measurement
The temperature values of this simulation show
for the Rth Tj-Tc, Rth Tj-Th1 and Rth Tj-Th2
nearly the same values as specified in the
0,50
Al
Thermal Resistance (K/W)
0,45
Cu
0,40
0,35
0,30
0,25
0,20
0,15
0,10
0,05
0,00
0
10
20
30
40
50
60
Thickness Baseplate (mm)
Figure 28: increased thickness of heat-sinkbase
Increasing the base of the heat sink up to 25mm
lead to a strong reduction of the spreading Rth
inside the heatsink. Increasing the thickness up
to 50mm still decreased the Rth but the slope
was getting less steep.
5. Conclusion
For modules in the tested power range / chip
size (1200V / 150A /158mm2) without baseplate
no cross coupling within the module itself was
found.
For modules with baseplate, the cross coupling
is influencing the Rth for chips side by side at a
distance less than 5mm. The spreading of the
heat sink base is the major limiting factor and
must be considered.
If a distributed mounting of modules is not
applicable, the thickness of the heat-sink base
should be increased to values > 20mm or copper
should be considered instead of alumina.
Modules with baseplate provide the lowest
thermal resistance, since the baseplate provides
a thermal spreading before the thermal flow has
to pass the thermal interface module – heat sink
with a relative low thermal conduction. The
spreading of the base plate also improves the
spreading of the heat sink base.
However, three half-bridges without base-plate,
mounted well distributed are able to provide the
same or better performance than a dense SixPack with baseplate.