EVS28 International Electric Vehicle Symposium and Exhibition 1

EVS28 KINTEX, Korea, May 3-6, 2015

Analysis of Power Module Reliability exposed to Real Operation Conditions observed in Electrified Vehicles

MinKi Kim, Sang Won Yoon

Department of Automotive Engineering, Hanyang University, Seoul 133-791, Korea swyoon@hanyang.ac.kr

Abstract Power modules equipped in electrified vehicles often operate at harsh conditions. Thermal stress in power

modules is known to be one of the major factors resulting performance degradations or damages in solder

joints in power modules. In this paper, the analyses of temperature distribution and lifetime were conducted

with the consideration of real operation conditions, especially ambient temperatures and coolant temperatures.

Simulation results reveal that the effects of ambient temperature increment depends on both the state of power

semiconductors and the coolant temperature. The increased temperature inevitably increases thermal stress,

reducing the lifetime of IGBT solder joint by 7%, when ambient temperature increases 100°C, which is often

observed near vehicle engines. In addition, the IGBT solder joint exhibits shorter (~25% reduction) lifetime

compared to the diode solder joint, because of higher power loss. Therefore, power module design should

simultaneously consider heat generated from devices and effects from operating conditions.

Keywords: power modules, thermal reliability, lifetime, operation conditions

1 Introduction Recently, various electrified vehicles are emerging, including EV (Electric Vehicle), HEV (Hybrid Electric Vehicle), PHEV (Plug-in Hybrid Vehicle), and FCEV (Fuel Cell Electric Vehicle). These electrified vehicles will be more popular along with the expansion of infrastructure [1-2]. Power (inverter/converter) modules are one of the core components for such vehicles, and convert battery or fuel-cell power to drive motors. The converting power continues to increases, and thus, recent power modules are demanded to be operational at high power conditions. Such high power modules are observed to suffer severe thermal reliability problems, including performance degradation [3-5], long-term fatigue [6-8], or permanent damages in power modules [9-10]. One of the

main causes of these problems is because power semiconductors under operation begin to heat up, due to power loss, generating thermal stress. The thermal stress is more severe considering that each adjacent layers consisting a power module have large mismatches in thermal expansion [3-7]. In order to solve the thermal stress problems, many studies have been reported either to reduce operation temperatures [11] or thermal stress [9], or to analyse lifetime of solder joint [6-8]. These studies generally used thermal cycling and power cycling evaluations to predict the lifetime of power modules in accelerated manners. These approaches are generally accepted, however, it should be noted that power modules are operating at even more harsh environments. Therefore, real operation conditions of vehicle, such as under hood temperatures, are another important factor deciding the performance and lifetime of power modules. As

EVS28 International Electric Vehicle Symposium and Exhibition 2

an example, the under hood temperature of conventional vehicles have been known to reach 125-140 °C [12-13]. Table 1 summarizes the severe operation conditions [12]. Thus, the reliability analysis of power modules should consider not only heat (generated from power semiconductors) but also high temperatures (observed in real operation conditions). This paper presents an analysis co-considering

changes in thermal energy coming from semiconductors and from operation conditions. To represent real situations, FEM simulations were conducted using parameters decided using commercial power modules. The FEM simulation results reveal temperature distributions and stress in various conditions, and provide estimated creep and fatigue at lead free solder joints. Table 1: Examples of temperature extremes observed

in automotive applications [12]

Location Maximum Temperature

On engine transmission 140°C

At engine (e.g., Intake manifold) 125°C

Under hood (near engine) 120°C

Under hood (remote location) 105°C

Exterior 70°C

Passenger compartments 70-80°C

2 Module descriptions

2.1 Module components A power module analysed herein is a simplified model representing the intelligent power module (IPM) used in Prius 2010 [14]. Figure 1 shows a schematic of the analysed power module. The module is composed of a direct bonded aluminium (DBA) plate, two pairs of IGBT and diode, solder, aluminium (for cooling infrastructure), and wire-bonding ribbon. The DBA is made of a patterned aluminium plate bonded on an aluminium nitride (AlN) layer. The AlN functions as heat conductor and electrical insulator, because of its good thermal conductivity and electrical insulation. Two IGBTs and two diode shown in the figure are solder bonded to the aluminium plate. The cross-section of the solder’s edge is assumed to be

trapezoid mimicking real meniscus shapes [15-16]. The top side of the solder has a size the same with that of IGBTs and diodes. Table 2 lists sizes of each component.

Table 2: Component sizes used in simulations [14-16]

Component Component Size[mm]

Width × Length × Thickness Silicon IGBT 15 × 15 × 0.3 Silicon diode 15 × 7.5 × 0.3 Aluminium

on DBA 20 × 40 × 0.6

Aluminium nitride on

DBA 22 × 44 × 0.6

SnAgCu Solder for

device bonding

- Top side: Length and width are the same with those of IGBT and diode - Bottom side:16 × 16 × 0.1

Figure 1: Schematic of the analysed power module

2.2 Problem definition When an electrified vehicle is driving, its power module begins to lose energy and generate heat. The heated module experiences thermal stress induced by mismatches of coefficient of thermal expansion (CTE) between adjacent layers of the module. Solder joint is located between silicon power semiconductors (i.e., IGBTs or diodes) and the top metal layer of a power substrate, and thus, exhibits significant thermal stress. This effect is more significant considering that power semiconductors are the major location inducing power loss and work as a heat source. Real operating conditions of power modules are another major factor deciding thermal reliability. In HEVs, their engine are equipped at harsh environment locations with vibration and high ambient temperatures. Such high ambient temperatures not only increase the maximum temperature of power module but also continuously heat up the module regardless of its operation. For

EVS28 International Electric Vehicle Symposium and Exhibition 3

this reason, high ambient temperature in HEV will be another major factor reducing power module lifetime. In addition, the high ambient temperature, in conjunction with the heat generated from power loss, continuously affects the performance and reliability of power modules. Therefore, power module design should simultaneously consider heat generated from devices and given by operating conditions.

3 FEM Simulations

3.1 Material properties In this model, the aluminium nitride in DBA substrates and aluminium in a cooling structure were assumed to be a linear-elastic model. Whereas, aluminium in DBA substrates was assumed to be a plasticity model because the stress in aluminium easily exceeds the yield strength. Also, the aluminium was modelled as no hardening plasticity, i.e., perfect plasticity. Finally, the solder, assumed to be a viscoplasticity, was assumed to have creeps and exhibits property change by temperature change by time. To import such effects, we used the Garofalo hyperbolic sine law.

Table 3: Material Properties [18-20]

Materials Al AlN Silicon Modulus

[GPa] 70 330 30

CTE [ppm/K] 23.1 4.5 4.1

Poisson Ratio 0.35 0.23 0.28 Yield Strength

[MPa] 20 Elastic

Table 4: Solder properties [18-20]

Solder name SnAgCu Temperature 298 343 373

Modulus [GPa] 53 47 44 CTE

[ppm/K] 21.3 22.15 23

Poisson ratio 0.4 Yield strength

[MPa] Strain

(0/0.01/0.1)

30 / 45 / 180

23 / 29 / 83

18 / 20 / 38

( )

sinh( )Qcr

n RTd A B edte s

-

= (1)

The is the equivalent uniaxial creep strain, T is absolute temperature, R is the gas constant; A

(creep rate coefficient), B (1/effective stress), n (Garofalo n parameter), Q (activation energy) are input constant, respectively. In this simulation, A, B, n, Q are 44100[1/sec], 0.005[1/MPa], 4.2, 44995[J/mol], respectively. These parameters are based on data from previous efforts [21-22]. The lifetime of the solder joint was calculated by the Morrow's life prediction model. The energy based model provided creep energy dissipation density as follows.

' (2 )mc f fW W ND = (2)

The ∆ is the dissipated energy density per cycles, is fatigue energy coefficient, is fatigue cycles, and m is fatigue energy exponent [23] For reality, these model parameter is imported to COMSOL simulations using previously reported experiment results [20].

0.8789260.17 (2 )c fW N -D = (3)

3.2 Simulation process The simulation was conducted by COMSOL Multiphysics FEM simulation software. Layer thicknesses vary from 150 μm (at solder) to 2.5 mm (at aluminium in the cooling structure). As each layer has different thickness, each was meshed to have different mesh size. Especially, thin layers were extremely fine meshed, including the solder layer and the contact area between IGBTs, diodes and ribbons. Other layers were relatively coarse meshed. The number of mesh elements were 44744.

Thermal boundary condition was defined for the power dissipation model. Power applied to IGBTs and diodes was 20W (low state) and 120W (high state) per cycle. Figure 2 illustrates power cycle applied to an IGBT. Diode power cycle is 1/6 times of IGBT cycles based on reported data [17]. The applied power is inevitably lost and each power device functions as a heat source.

Using convection heat flux model, ambient temperature was applied in the range from 25°C to 125°C. The range is set considering real operation conditions in conventional vehicles shown in Table 1. The heat transfer coefficient was set to be the same with that of air, 30[W/]. The cooling boundary condition (temperature of 75 °C) was applied to the bottom of the aluminium cooling structure. The heat transfer coefficient of cooling was decided to be 800 [ W/ ] based on experiment data from conventional power modules

EVS28 International Electric Vehicle Symposium and Exhibition 4

with silicon IGBTs having the maximum junction temperature of 150°C. [3-4, 17].

Figure 2: IGBT power loss cycle

Per mechanical boundary conditions, the aluminium cooling structure was fixed in all directions while and other components can freely displace to all directions. Three simulations were conducted. First, temperature distribution in a power module was calculated by different power and ambient/coolant temperatures using a time-dependent heat transfer models. Second, using the temperature distribution, creep energy dissipation density was estimated by nonlinear time-dependent analysis. Lastly, the creep density was applied to the Morrow's fatigue model and the fatigue life was estimated.

4 Results Figures 3 and 4 compare temperature distributions when ambient temperatures are 25°C and 125°C, respectively. The module temperature ranges from 129°C to 153°C and from 138°C to 162°C. Even though the ambient temperature increases by 100°C, the maximum module temperature increased only by °C. Thus, ambient temperature slightly increases module temperature by ~6% in this simulation and also increases thermal stress. However, the increment amount is small compared to the increment in ambient temperature. This small increment is because the device already generates high junction temperature (~150°C shown in Fig. 3), which is not much lower than the maximum ambient temperature.

Figures 5 and 6 show temperature distributions when ambient temperatures are 25°C and 125°C with minimum power loss (20W) in the cycle. The module temperature ranges from 86.1°C to 92.5°C and from 94.8°C to 102°C. This temperature range

is narrower than the range in Figs. 3 and 4, because the coolant dominates in low power loss conditions.

Figure 3: Temperature distribution (operation

temperature: 25°C, power loss: 120W)

Figure 4: Temperature distribution (operation

temperature: 125°C, power loss: 120W)

Figures 7 is the z-direction stress distribution at the

side boundaries of solders under an IGBT. The z-direction stress is transverse normal stress, also known as peeling stress [24]. Peeling stress is a good indicator of failures. As expected, the increased ambient temperature also induces the maximum thermal stress increment (by 9% in this simulation) at the solder. Figure 8 demonstrates the temperature distribution when coolant temperature changes. The coolant in Fig. 8 is 95°C whereas its temperature in Figs. 3-7 is 75°C. The maximum temperature elevates to 180°C (18°C increment). The maximum module

EVS28 International Electric Vehicle Symposium and Exhibition 5

temperature changes by 0.9°C with 1°C increment in coolant temperature. This increment amount is larger than that observed in ambient temperature increment, and thus, the effects from coolant temperature are more dominant that those from ambient temperature. As a result, the maximum shear stress becomes higher by 20% at the corners (~1% stress increment by 1°C increment in coolant temperature).

Figure 5: Temperature distribution (operation

temperature: 25°C, power loss: 30W)

Figure 6: Temperature distribution (operation

temperature: 125°C, power loss: 30W)

Figure 9 is a curve showing creep energy

dissipation density with different ambient temperatures at two critical points. The points are either between the IGBT and solder layer or between the diode and solder layer. When the creep dissipation energy density increases, fatigue lifetime decreases, as expected. The creep dissipation energy density is higher at the junction

between the IGBT and solder layer than at the junction between the diode and solder. Of course, the creep density is higher with higher ambient temperatures.

Figure 7: z direction stress distribution.. Operation temperature: Left: 25°C, right: 125°C (power loss:

120W)

Figure 8: Temperature distribution (operation

temperature: 125°C, power loss: 120W, coolant temperature: 95°C)

Figure 9: Creep energy dissipation density by the

functions of time

Figures 10 and 11 demonstrate the number of cycles to failure, based on the fatigue calculation using the creep energy dissipation in Fig. 9 and

EVS28 International Electric Vehicle Symposium and Exhibition 6

Morrow's energy based model. Note that the figures were depicted in log scale. One cycle is 130 minutes. The simulated number of cycles to failure is 275.42 cycles (at 125°C) and 295.12 cycles (at 25°C) at the solder joint under the IGBT. The cycles to failure decrease by about 7% when the ambient temperature changes from 25 to 125°C. Whereas, in the same ambient temperature, the cycles to failure at the solder joint of the diode is 346.73 cycles, which is 25% longer than the lifetime of the IGBT solder layer. It is because the diode has lower power dissipation, as explained beforehand.

Figure 10: Cycles to failure (Log scale; operation temperature: 25°C)

Figure 11: Cycles to failure

(Log scale; operation temperature: 125°C)

5 Conclusion The effects of operation conditions on power

modules were analysed. The maximum module temperature was dominantly affected by coolant temperatures at low power loss states, compared to the change in ambient temperature. The increased temperature also increases thermal stress. As the ambient temperature continuously heats up a module regardless of module operations, fatigue is steadily accumulated. For these reason, the failure of the solder joint accelerates. From the simulation results, the lifetime of IGBT solder joint was 25% shorter than that of diode layer. In addition, the increased ambient temperature (by 100°C) reduces solder lifetime by about ~7%.

Thus, serious reliability problems can occur by not only the module power loss but also real operation conditions in the electrified vehicles.

Acknowledgments This research was supported by Basic Science

Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (grant number 2014R1A1A2056131). Also, This research was supported by the MSIP (Ministry of Science, ICT&Future Planing), Korea, under the CITRC (Convergence Information Technology Research Center) support program (NIPA-2014-H0401-14-101) supervised by the NIPA (National IT Industry Promotion Agency).

References [1] Yilmaz, Murat, and Philip T. Krein., Review of

battery charger topologies, charging power levels, and infrastructure for plug-in electric and hybrid vehicles, Power Electronics, IEEE Transactions on 28.5 (2013), 2151-2169.

[2] Eberle, Ulrich, Bernd Müller, and Rittmar von Helmolt., Fuel cell electric vehicles and hydrogen infrastructure: status, 2012. Energy & Environmental Science 5.10 (2012), 8780-8798.

[3] Bouarroudj, M., et al., Degradation behavior of 600V–200A IGBT modules under power cycling and high temperature environment conditions, Microelectronics Reliability 47.9 (2007), 1719-1724.

[4] Song, Yantao, and Bingsen Wang., Survey on reliability of power electronic systems, Power Electronics, IEEE Transactions on 28.1 (2013), 591-604.

[5] Zhou, Weicheng, Xueqian Zhong, and Kuang Sheng., High temperature stability and the performance degradation of SiC MOSFETs, Power Electronics, IEEE Transactions on 29.5 (2014), 2329-2337

.[6] Ji, B., et al., Onboard condition monitoring of solder fatigue in IGBT power modules, Diagnostics for Electric Machines, Power Electronics and Drives (SDEMPED), 2013 9th IEEE International Symposium on. IEEE, 2013.

[7] Ye, Hua, Minghui Lin, and Cemal Basaran., Failure modes and FEM analysis of power electronic packaging, Finite Elements in Analysis and Design 38.7 (2002), 601-612.

[8] Takahashi, Tomohiro, and Qiang Yu., Precision evaluation for thermal fatigue life of power module using coupled electrical-thermal-mechanical

EVS28 International Electric Vehicle Symposium and Exhibition 7

analysis, Electronics Packaging Technology Conference (EPTC), 2010 12th. IEEE, 2010.

[9] Ning, Puqi, Khai Ngo, and Fred Wang., Thermomechanical reliability investigation of large temperature excursions in power electronics package, Energy Conversion Congress and Exposition (ECCE), IEEE, 2011.

[10] Yoon, Sang Won, Michael D. Glover, and Koji Shiozaki., Nickel–tin transient liquid phase bonding toward high-temperature operational power electronics in electrified vehicles. Power Electronics, IEEE Transactions on 28.5 (2013): 2448-2456.

[11] Charboneau, Bryan C., et al., Double-sided liquid cooling for power semiconductor devices using embedded power packaging. Industry Applications, IEEE Transactions on 44.5 (2008), 1645-1655.

[12] Fairchild, M. Ray, et al., Emerging substrate technologies for harsh-environment automotive electronics applications, SAE Technical Paper, No. 2002-01-1052 2002.

[13] Johnson, R. Wayne, et al., The changing automotive environment: high-temperature electronics, Electronics Packaging Manufacturing, IEEE Transactions on 27.3 (2004), 164-176.

[14] Burress, Timothy A., et al., Evaluation of the 2010 Toyota Prius hybrid synergy drive system, no. ORNL/TM-2010/253, Oak Ridge National Laboratory (ORNL), Power Electronics and Electric Machinery Research Facility, 2011.

[15] Bouarroudj, M., et al. Thermo-mechanical investigations on the effects of the solder meniscus design in solder joint lifetime for power electronic devices. Thermal, Mechanical and Multi-Physics Simulation Experiments in Microelectronics and Micro-Systems, 2007. EuroSime 2007. International Conference on. IEEE, 2007.

[16] Jesudoss, Pio, et al. Effect of solder volume on joint shape with variable chip-to-board contact pad ratio. Proceedings XXXII International Microelectronics and Packaging Poland Conference. IEEE Components, Packaging and Manufacturing Technology (CPMT) Society. IEEE, 2008.

[17] Xu, Zhuxian, et al., Si IGBT phase-leg module packaging and cooling design for operation at 200C in hybrid electrical vehicle applications, Applied Power Electronics Conference and Exposition (APEC), 2012 Twenty-Seventh Annual IEEE, 2012.

[18] Lei, Thomas Guangyin, et al., Effects of large- temperature cycling range on direct bond aluminum substrate, Device and Materials

Reliability, IEEE Transactions on 9.4(2009), 563-568.

[19] Matweb, www.matweb.com, accessed on 2015-02-10.

[20] Kim, Ilho, and Soon-Bok Lee., Fatigue life evaluation of lead-free solder under thermal and mechanical loads, Electronic Components and Technology Conference, 2007, ECTC'07. Proceedings. 57th. IEEE, 2007, 95-104

[21] Lau, John, W. Danksher, and Paul Vianco., Acceleration models, constitutive equations, and reliability of lead-free solders and joints, Electronic Components and Technology Conference, 2003, Proceedings. 53rd. IEEE, 2003, 229-236

[22] Hong, Bor Zen., Thermal fatigue analysis of a CBGA package with lead-free solder fillets, Thermal and Thermomechanical Phenomena in Electronic Systems, 1998.

[23] Shi, X. Q., et al., A modified energy-based low cycle fatigue model for eutectic solder alloy, Scripta Materialia 41.3 (1999), 289-296.

[24] Wen, Yujun, and Cemal Basaran., Thermomechanical stress analysis of multi-layered electronic packaging, Journal of Electronic Packaging 125.1 (2003), 134-138.

Authors

MinKi Kim received the B.S degree in automotive engineering, Kookmin University, Seoul, in 2014. Since 2014, he has been with the department of automotive engineering in Hanyang University, Seoul, where he is currently a student of integrated master's and doctor course.

Sang Won Yoon received the B.S. degree in electrical engineering from Seoul National University, Korea, and the M.S. and Ph.D. degrees in electrical engineering from University of Michigan, USA, in 2000, 2003 and 2009, respectively. After working at Toyota Research Institute of North America from 2009 to 2013, he is now assistant professor in the department of automotive engineering in Hanyang University, Seoul, Korea. His research interests include power electronics, integrated sensor systems, and the development of novel technologies for future vehicles.