Electro-Thermal Analysis for Automotive High Power MOSFETs

Asantha Kempitiya and Wibawa Chou

Infineon Technologies Americas Corp.
101 N. Sepulveda Blvd.,
El Segundo, CA 90245
[email protected]

(PCB) which is attached to a 4mm thick Aluminum heatsink

Abstract
with thermal interface material (TIM). The system is cooled
Power dissipated within a semiconductor package must be
by natural convection as common to most auxiliary systems in
properly extracted in order to guarantee safe and reliable
automotive applications.
operation in an automotive high power application. This work
analyzes the transient thermal characteristics of two surface The simulated transient thermal impedance profiles (Zth)
mount MOSFET packages types, TOLL and PQFN 5x6 for are shown in Fig. 3 through 5. The solid lines shown in Fig. 3
various heat sinking configurations. A novel methodology is depict the transient thermal impedance obtained for TOLL
then demonstrated for comparing the relative thermal response package with various via patterns located under the device on
of these two packages under the application of transient the PCB. Here the vias were confined to the exposed die pad
power. Measurement results are provided to validate area under the surface mount device. The summary of the via
simulation results and concluded with electro-thermal pattern designs is given below in Table 1,
simulations that predict the device’s thermal response for a Simulation Via Diameter Via
Pitch (mm)
typical automotive motor drive application. The process Case (mm) Number
highlighted here allows engineers to understand the thermal 1 0.3 1.5 30
design tradeoffs associated with such devices with various 2 0.3 1 63
thermal environments and facilitate comparison and prediction 3 0.3 0.6 168
of their thermal response in performance critical applications.
Table 1: Summary of via pattern designs used for the TOLL
Keywords - Surface mount power MOSFETs, electro-thermal package.
simulations, motor drive, automotive applications.
1. Introduction
Power dissipated within a power semiconductor package
must be extracted in order to ensure system robustness and
reliability. This is especially true for power MOSFETs used
for automotive applications such as DC-DC converters and
electric power steering (EPS) where under hood ambient
temperatures can easily exceed 75°C. In this work, we
evaluate Infineon’s IPLU300N04S4-1R1 [1] power MOSFET
and compare its transient thermal response with a PQFN type
power package with equivalent die size for various heat
sinking configurations. A novel methodology is demonstrated
that analyzes the thermal impedance in the frequency domain
to predict temperature response under the application of Figure 1: Thermal simulation of six PQFN 5x6 packages on
transient power. Electro-thermal simulations are performed to PCB mounted on Aluminum heatsink (no fins) with TIM.
evaluate PQFN 5x6 vs TO leadless (TOLL) type power
packages in a motor drive application where both
semiconductor power losses and system thermal stack-up are
considered. This knowledge allows design engineers to predict
the performance of a particular power MOSFET in any
application and guarantee the reliability of its thermal
interfaces such as silicon, bondwires and die attach under
thermal stresses ensuing from worst case operation.
2. Methodology
2.1. Thermal Simulation of TOLL vs PQFN 5x6
As shown in Fig. 1 and 2, 3D Computational Fluid
Dynamics (CFD) thermal simulations are performed using
Flotherm for six power MOSFETs that form the 3-phase Figure 2: Thermal simulation of six TOLL packages on PCB
inverter mounted on a 1.57mm thick printed circuit board mounted on Aluminum heatsink (no fins) with TIM.

978-1-5386-1531-7/17/$31.00 ©2017 IEEE 302 33rd SEMI-THERM Symposium

Figure 3: The transient thermal impedance for TOLL Figure 4: The transient thermal impedance for PQFN 5x6
package for 3 distinct via patterns located under the device on package for 3 distinct via patterns located under the device on
the PCB mounted on Aluminum heat sink without fins. the PCB mounted on Aluminum heat sink without fins.

The effect of improved heat transfer through the PCB due to For each package, it is evident that after approximately 2s-
vias (with decreasing pitch / increasing density) is seeing in 3s the heat sink geometry influences the transient thermal
the time regime of 0.2s to 20s. The dotted lines show the profile for both packages. The dotted lines of the left axis
comparison between via patterns 2 and 3 with respect to via show a staggering percent reduction in Zth of approximately
pattern 1, which indicate a percent reduction in Zth that peaks 300% for the TOLL package. This improvement arises from
at approximately 20% and 37.5% respectively. the additional heat spreading and thermal capacitance present
The solid lines in Fig. 4 show the transient thermal in the TOLL package as seeing in the time regime from
impedance obtained for the PQFN 5x6 package with various approximately 20ms-3s where heat flux is primarily contained
via patterns. Here the vias were confined to the exposed die within the package.
pad area under the surface mount device. The summary of the
via pattern designs is given below in Table 2,

Simulation Via Diameter Via

Pitch (mm)
Case (mm) Number

1 0.3 1.5 9
2 0.3 1 20
3 0.3 0.6 42
Table 2: Summary of via pattern designs used for the PQFN
5x6 package.
Similar to the TOLL package, the PQFN 5x6 package
shows a percent reduction in Zth in the time regime from 0.2s
to 20s that peaks at approximately 30% and 50%. This result
also indicates that a higher improvement in heat transfer can
be achieved for cases with via patterns 2 and 3 (with respect to
Figure 5: The transient thermal impedance for TOLL and
via pattern 1) at the cost of additional vias in comparison to
PQFN 5x6 package with via pattern 1 on Aluminum heat sink
the larger TOLL package. Even though such a large amount of
with and without fins.
vias translate to higher manufacturing costs, the process
followed here allows designers to be aware of how much more 2.2. Thermal Measurement of PQFN 3x3
power can be squeezed out of a particular surface mount In order to bench mark the simulated cases, the transient
device with a certain board configuration for performance thermal impedance was measured for a single PQFN 3x3
critical applications. device on a 1.6mm thick PCB attached to a 6mm thick
Fig. 5 illustrates the transient thermal impedance (central Aluminum heatsink with thermal interface material (TIM).
axis) of both TOLL and PQFN 5x6 packages with via pattern Fig. 6 shows the measured and simulated transient thermal
1 located under the device on the PCB. Here two simulation impedance curves obtained from this study. When the
cases are considered where the Aluminum heat sink consists cumulative magnitude and time constant of the simulation
with and without fins. As expected the former provides a results was increased by 25% and 35%, a close match was
larger area for convective heat transfer thus reducing the obtained between the simulated and measured results transient
steady state thermal impedance as shown by the far right side thermal impedance curves.
of the Zth curve.
 These R-C networks are used to generate the frequency
response curves shown in Fig. 7 where the top plot shows the
magnitude of thermal impedance over frequency while the
bottom curve depicts the percent difference between the two
cases that peaks at approximately 0.1Hz. This peak denotes
that applying power in the shape of a sinusoidal frequency of
~0.1Hz provides the largest difference in transient MOSFET
junction temperature between the two thermal networks under
consideration.
3. Electro-Thermal Simulation
Electro-thermal simulations are then performed in order to
evaluate the benefit that arises from the TOLL package and
compare their performance in a typical motor drive
application. Electro-thermal models are developed [3, 4] that
consider both semiconductor power losses and the overall
Figure 6: Measured and simulated transient thermal system thermal stack-up for this purpose. A single leg of a 3
impedance for PQFN 3x3 package on PCB with vias mounted phase motor drive inverter can be represented by a MOSFET
on Aluminum heat sink with fins. half bridge as shown in Fig. 8. The center point (phase node)
of the half bridge connects to the motor winding. During each
Since the error in repetitive measurement of steady state half cycle, the sinusoidal current that flows through these
thermal resistance of a surface mount device can be as much windings flows through the switching high (low) side
as 35% [2], this process reveals that the simulation setup and MOSFET and complementary MOSFET’s low (high) side
material properties of the thermal models used for generating body diode. When a single device is considered, the MOSFET
the results in this study are within a reasonable range of channel or body diode only conducts current during separate
accuracy. half cycles.

2.3. Frequency Domain Representation of Thermal

Impedance
While the transient thermal impedance was represented
with respect to time thus far, its behavior in the frequency
domain can also be of interest. This method of representing a
thermal system is helpful especially in most power electronics
applications where voltage and current signals are defined in
terms of their amplitude and frequency. In order to facilitate
frequency domain representation, the Foster thermal R-C
network of the complete thermal stack up which includes the
heat sink with fins is extracted for the packages of interest.

Figure 8: Single leg of a 3-phase inverter realized with

MOSFETs (left), and depicts sinusoidal currents flowing
through stator windings (right).

The top curve of Fig. 9 shows the power dissipation

estimated for a MOSFET die with an on resistance of
approximately 2.57mΩ with 80A of sinusoidal current at
0.1Hz. As discussed earlier, the specific shape of the power
pulse arises due to the half cycle conduction of an individual
FET. This particular frequency and level of power dissipation
can be seeing in a motor drive applications where the motor
begins from stall and enters low speed operation at high
torque (current) and low fundamental mechanical / electrical
rotational frequency. This operation can last several seconds
depending on the specific application requirements.
The bottom curve of Fig. 9 shows the MOSFET junction
Figure 7: The frequency response of TOLL and PQFN 5x6 temperature evolution for TOLL and PQFN 5x6 packages
packages with via pattern 1 on Aluminum heatsink with fins under this power dissipation where the temperature ripples of
(top), the percent difference in thermal impedance between approximately 10°C and 43°C can be observed for either
both packages (bottom). package respectively. As a result, the PQFN package with
equivalent die size mounted on approximately similar board impedance decreases with increasing frequency as shown in
configuration provides four times higher temperature ripple. Fig. 7 (top). The percent increase in ripple temperature from
The simulation also shows that a higher peak junction TOLL to PQFN package is calculated in Fig. 10 (bottom). It is
temperature in the excess of 180°C is seen by the PQFN evident that the relationship is similar to the trend shown in
device while the peak temperature of the TOLL device is Fig. 7 (bottom). For the two thermal networks considered, this
~150°C. Both peak and ripple temperatures can have verifies the expected thermal response over frequency, and
tremendous impact on the reliability of the device’s various also confirms that the largest difference in MOSFET junction
thermal interfaces over the lifetime of operation. Thus the ripple temperature is seen at ~0.1Hz.
thermal benefits of TOLL vs a PQFN 5x6 package is clearly
4. Conclusions
demonstrated via electro-thermal simulations.
This work analyzes the thermal characteristics of a TOLL
and a PQFN 5x6 type package with equivalent die size for
various heatsinking configurations. The transient thermal
impedance is calculated for the devices mounted on a PCB
with various via patterns, assembled on an Aluminum heat
sink with TIM. A novel methodology is then demonstrated for
comparing the relative thermal response of these two packages
under the application of transient power. Simulation results
are benchmarked by measurements performed on a single
PQFN 3x3 package with similar PCB and heatsink topology.
Finally, electro-thermal simulations are carried out to predict
the thermal response of the packages under consideration and
illustrate the benefits of the TOLL package in a typical
automotive motor drive application. The techniques shown
here allow engineers to understand the design tradeoffs
associated with thermal management of surface mount devices
for performance critical applications.
References

Figure 9: The transient power dissipated in the MOSFET [1] Infineon Technologies, “OptiMOS™-T2 Power-
during low speed motoring (top), the MOSFET junction Transistor,” IPLU300N04S4-1R1 datasheet, Rev. 1.0
transient thermal response (bottom). [Revised Nov. 2014].
[2] Infineon Technologies, App. Note 994, pp.1-6.
[3] Kempitiya, A. and Chou, W., "Electro-Thermal
Simulation for High Power IGBTs for Automotive
Applications," Therminic 2016, Workshop on Thermal
Investigations of ICs and Systems, Budapest, Hungary,
2016.
[4] Ottosson, J., Alakula M. and Hagstedt, D., "Electro-
thermal simulations of a power electronic inverter for a
hybrid car," 2011 IEEE International Electric Machines
& Drives Conference (IEMDC), Niagara Falls, ON,
2011, pp. 1619-1624

Figure 10: MOSFET junction ripple temperature over power

dissipation frequency (top) percent increase in MOSFET
junction ripple temperature (bottom).

Fig. 10 (top) shows MOSFET junction ripple temperatures

over power dissipation frequency. As expected, ripple
temperature decreases as the overall transient thermal