Analysis of field-stressed power inverter modules from electrified vehicles

Transcription

1 XXXX Analysis of field-stressed power inverter modules from electrified vehicles Author, co-author (Do NOT enter this information. It will be pulled from participant tab in MyTechZone) Affiliation (Do NOT enter this information. It will be pulled from participant tab in MyTechZone) Abstract This paper presents a reliability study of a directly cooled IGBT module after a test drive of 85,000 Km in a fuel cell electric vehicle, as well as of an indirectly cooled IGBT module after a test drive of 200,000km in a hybrid car on public roads. At the end of the test drive, the inverter units were disassembled and analyzed with regard to the lifetime consumption. First, electrical measurements were carried out and the results were compared with the ones obtained directly after module production (End of Line test). After that, ultrasonic microscopy was performed in order to investigate any delamination in the solder layers. As a third step, an optical inspection was performed to monitor damages in the housing, formation of cracks or degradation of wire bonds. The results show none of the depicted failure modes could be found on the tested power modules after the field test. Obviously, no significant life time consumption could be observed. In order to evaluate the robustness of the components (power modules) in the real application, a directly cooled IGBT module after a test drive of Km in a fuel cell electric vehicle and an indirectly cooled IGBT module after a test drive of 200,000km in hybrid car were analyzed in detail. Failure mechanisms of a power module The main technical requirements for power converters include, among others, a very high reliability. During its life time, a power module is exposed to severe temperature cycles of different origins. The main cause of the failures is different heating of the individual areas/layers and the different thermal expansion coefficients of the material used. The typical wear-out mechanisms are bond wire liftoff and delamination (chip to ceramic or ceramic to base-plate). Figure 1 shows the typical failure mechanisms: wire bond lift off, chip solder degradation, system solder degradation between DCB and base plate. Introduction Eco-friendly vehicle like Hybrid and electric vehicles (HEV) will play a very important role in reducing CO2 emissions and reducing fuel consumption in future transportation. The energy sources of electrified vehicles, weather it is a highpower battery or a fuel cell, deliver direct current which has to be inverted into alternating current for the electric motor. The core of the main inverter in electrified vehicles is the IGBT power module in order to increase the overall efficiency of the system. During the operational lifetime, the IGBT modules are exposed to harsh environmental conditions such as severe temperature cycles. Therefore eco-friendly vehicle technology requires power modules which are highly reliable, compact, economical and rugged enough to withstand mechanical, electrical and thermal shocks. Active temperature cycles are a result of internal heating of the dies caused by inverting the direct current while driving. Passive temperature cycles are caused by variations of the ambient temperature like summer and winter cycles or by variations of the performance of the cooling system. Moisture or mechanical stresses through vibration or shock are other factors that limit the lifetime of the IGBT module. Many papers have been written about reliability testing of power modules, life time modelling and calculation [1][2][3]. However, only little has been published from experience in the field. Page 1 of 6 Figure 1. Wear out mechanisms of power modules Fuel Cell Electric Vehicles (FCEV) test car setup The fuel cell stack, electric motor, battery and hydrogen tank are the main components of the Fuel Cell Electric Vehicle (FCEV). Electricity generated from the fuel cell is first transmitted to the inverter that converts the direct current (DC) to three phase variable voltage and variable power. This changes the speed and torque of the traction motor [4]. In order to fulfill the transient requirement of vehicle propulsion, the fuel cell stack is typically coupled with a battery through a DC/DC converter forming a hybrid power system. The power modules under analysis were used in electrical power train system as shown in Figure 2. The fuel cell system is connected to motor controller directly via the DC bus, and a bidirectional

2 DC/DC converter is inserted between the DC bus and battery pack. Two IGBT-modules are used per vehicle, one for the DC/AC main inverter (FS800R07AE3, 800A nominal current, 6 switches, directly cooled by pin-fins) and the other one for the DC/DC converter (FS400R07A1E3, 400A nominal current, 6 switches, indirectly cooled). In order to evaluate the robustness of the power modules in the real application, the power module in the DC/AC main inverter was analyzed. Figure 2. Schematic diagram of Fuel cell electric vehicle (HCEV) Hybrid Electric Vehicles (HEV) test car setup The hybrid electric vehicle (HEV) in this field test is a full parallel hybrid with two energy sources, an internal combustion engine and an electric motor/generator. This system can take advantage of the benefits provided by these power sources while compensating for each other s shortcomings. [5] In such a Parallel Hybrid system, there are two parallel paths to power the wheels of the vehicle: an engine path and an electrical path. The electrical path is shown in figure 3. The battery pack is connected to the traction motor as well as to a generator through a bidirectional DC/DC converter. The HEV under test also has a starter generator which operates at the same battery voltage as the electric traction motor and the battery, but it does not provide any tractive effort to the vehicle. Instead, the starter-generator is used only to start the engine and then to charge the hybrid battery IGBT-modules are used for the main motor (FS400R07A1E3, 400A nominal current, 6 switches, indirectly cooled) and for the startergenerator motor (FS200R07A1E3, 200A nominal current, 6 switches, indirectly cooled). In order to evaluate the robustness of the power modules in the real application, the power module in the DC/AC main inverter for the traction motor was analyzed. Lifetime Simulation (Driving cycle) A lifetime simulation was carried out after the details of the for the practical test drive were available. The FTP cycle (for Federal Test Procedure) were used in this simulation which has been created by US EPA (Environmental Protection Agency) [6] to represent a commuting cycle with a part of urban driving including frequent stops and a part of highway driving. In order to predict the lifetime of a module, it is possible to emulate with dedicated tests the effects linked to mechanical stress. Those tests, with well-defined conditions, are performed by Infineon during the qualification of the modules. The module s passive lifetime is determined by temperature cycling tests without electrical stress. They mainly aim at evaluating the lifetime of the solder joint between ceramic and baseplate, and the resistance to the sudden changes in temperature that the device may experience during storage, transportation or in use (cold start). Power cycling (PC) tests address the active lifetime (under operation) are used to determine the resistance of a semiconductor device to thermal and mechanical stress linked to the power dissipation of the internal semiconductor die. Load currents periodically applied to the module cause rapid temperature changes. Depending on the cycling frequency, the mechanical stress is predominantly put on the wire bonds or on the solder between the chip and the ceramic carrier. The calculation of the lifetime of a power module based on a given driving cycle requires subsequently the computation of the temperature profile and the computation of ΔT occurrences (Infineon uses the rain-flow algorithm). The temperature profile of the module s components based on the given driving cycle, the application specific loss model and the electrical characteristics of the module has to be calculated. For every time step, those parameters are combined with the electric parameters of the module in order to derive instantaneous power losses. The calculated losses are fed into the thermal model of the module in order to derive the instantaneous temperature of the IGBT, diode and solder. Material fatigue can be quantified as the number of temperature cycles a system can sustain. In order to determine the lifetime expectancy of the module, the Coffin-Manson model is used. Bluntly speaking, those analytical models express, for a given temperature, the number of temperature swings the system can sustain. By applying those mathematical formulas to the ΔT profile computed above, it is possible to derive how much life time is used by the module over the mission time. Obviously, the used life time should always be less than 100%, which represents the total endurance of the module. Further details on lifetime calculation are available in [3] and [7]. Figure 3. Schematic diagram of Hybrid Electric Vehicles (HEV) Page 2 of 6

3 - Duration : 1374 seconds - Average speed: 34 km/h (21.2 mph) FTP 75 Highway Driving Cycle parameters: - Total distance : 16.45km (10.26 miles) - Duration : 765 seconds - Average speed: 77.7 km/h (48.3 mph) The equivalent estimated lifetime consumption estimation can be extracted from the FTP 75 driving cycle. Active ΔT is a temperature cycle occurring when the power module is powered. The active cycles are the consequence of the variations of the electrical parameters during operation of the power module (e.g. current and DC link voltage variations). (a) FTP 75 Urban Driving Cycle Passive ΔT are defined as the difference between the maximum temperature reached during the active phase of the cycle and the ambient temperature (cold start). For the simulation, passive cycles were considered for the one year test drive period only. Table 1. Lifetime consumption estimation for the traction motor module for the FCEV Active ΔT passive ΔT Total IGBT 0.160% 2.1% 2.260% Diode % 1.9% 1.989% Solder % 2.2% 2.210% As the simulation results of the FCEV traction motor module in table 1 show, there is no significant life time consumption. Table 2. Lifetime consumption estimation for the traction motor module of the HEV Active ΔT passive ΔT Total IGBT 15% 1.6% 16.6% Diode 4.8% 0.8% 5.6% Solder 0.4% 2.2% 2.6% (b) FTP 75 Highway driving cycle Figure 4. Driving cycle in the simulation The fuel cell test car was subjected to a field test during a period of almost one year. driving 85,000km mainly on highways, only a minor part of the distance was driven in urban areas. The afore-mentioned cycles are adjusted by 10% (city) and 90% (highway) to more accurately reflect real results. Figure 4. shows the used driving cycles in simulation and the following are some characteristic parameters of the each cycle. FTP 75 Urban Driving Cycle parameters: - Distance travelled: km (7.98 miles) The hybrid test car s ratio of electric mode versus combustion mode was approximately 3:7. Approximately 60% of the total distance was driven in urban areas. As table 2 shows, just like for the FCEV, the lifetime consumption for the HEV is not significant. The active lifetime calculated here has to be multiplied roughly by 0.3 because the e-motor was only switched on 30% of the time. Quality Analysis After 85,000km of operation on Fuel Cell Electric Vehicles (FCEV) and after 200,000km on Hybrid Electric Vehicles (HEV), a quality analysis was done on the electrical drivetrain system to evaluate the degree of degradation of the different components. Both power modules were analyzed with regard to the electrical performance and the degradation of the joining techniques. Page 3 of 6

4 LEGEND Measurement +5% -5% LEGEND Measurement +5% -5% Beginning the analysis with non-destructive tests, the static electrical parameters like Vcesat and VF were measured and correlated to the initial test data (End of Line test) gathered during module fabrication. Based on the measured data, correlation plots were obtained, which revealed negligible drifts in the electrical parameters. The plots of the IGBT collector emitter saturation voltage and the diode forward voltage can be regarded as an indicator for the degradation of the chip solder layers and/or the wire bond connection on top of the chips (see Figure 5 and 6). Very small drift values of 1-2% in Vcesat and VF were observed, indicating almost no degradation after 85,000Km for FCEV and 200,000Km for HEV. 1,9 1,85 1,8 1,7 1,65 Figure 6. Correlation plots of the collector emitter saturation voltage and forward voltage of the motor drive side power module of HEV 1,75 1,6 1,55 Limits 1,5 1,45 mid 1,4 1,35 1,3 1,3 1,35 1,4 1,45 1,5 1,55 1,6 1,65 1,7 1,75 1,8 1,85 1,9 VCEsat [V] before test 1,9 1,8 (a) Motor side power module in FCEV 1,7 1,6 Limits 1,5 mid 1,4 1,3 1,3 1,35 1,4 1,45 1,5 1,55 1,6 1,65 1,7 1,75 1,8 1,85 1,9 VF [V] before test Figure 5. Correlation plots of the collector emitter saturation voltage and forward voltage of the motor drive side power module of FCEV (b) Traction motor side power module in HEV Figure 6. C-SAM images after 85,000km/200,000km representing the chip solder layer. A delamination of the chip solder layers, the chip wire bonds or the ultrasonically welded power tabs could not be observed. (a) Motor side power module in FCEV Page 4 of 6

5 (b) Traction motor side power module in HEV Figure 7. C-SAM images after 85,000km/200,000km representing the system solder layer. A delamination of the system solder layer, typically starting at the edges of the ceramics, could not be observed. Subsequently to the electrical measurements, ultrasonic images (C- SAM) were obtained from the system solder layer as well as from the chip solder layer to monitor the degradation of the joining technology in both kind of modules (see Figure 7 and 8). Delamination of solder layers is a consequent result of thermal cycles in combination with a mismatch of thermal expansion coefficients of the used materials, typically starting from the edges of the ceramics. No delamination of the observed system solder layers could be found on the analyzed devices. In addition to that, neither a degradation of the chip solder layers nor lifted chip-bond wires or ultrasonically welded power tabs (HybridPACK2) could be found on the investigated power modules. The results of the ultrasonic investigation go in line with the electrical measurements of the Vcesat and VF values as an indicator for an only minimally increased electrical resistance in the conducting path caused by e.g. solder delamination or bond wire liftoff. To conclude the module analysis, an optical inspection was performed on the power module housing and of the internal components after opening the lid. The storage of the plastic frame material at elevated temperatures can lead to a brownish color whereas the exposure to thermal cycles may result in the formation of cracks in the housing or a degradation of wire bonds. However, none of the depicted failure modes could be found on the tested power modules (see Figure 8 and 9): the color of the plastic material is unchanged compared to the initial state. Cracks in the housing material were not observed. The inspection of the internal components showed no conspicuities. The appearance of the frame wire bonds within the HybridPACK1 power module, connecting the copper on top of the ceramic with the bond balconies of the frame, exhibits no degradation. The bonding connection of the bond-feet remains unchanged. (b) Traction motor side power module in HEV Figure 8. After 85,000km/200,000km, the plastic material of the housing exhibits no change in color or cracks in the frame after lid removal. Figure 9. Optical light microscopy revealed no degradation on the frame bond wires in the traction motor side power module in HEV Special attention was also paid to the pin-fin water cooling pattern on the backside of the HybridPACK2 module. No bent pins, captured particles, abrasions or corrosion could be observed during the optical investigation (see Figure 10). (a) Motor side power module in FCEV Figure 10. An optical inspection of the motor power module pin-fin area in FCEV was carried out at the end of the driving test. No bent pin-fins, particles or abrasions could be determined. Extended Quality Analysis for HEV None of the depicted failure modes could be found on the tested power modules after the field test. Obviously, no significant life time consumption could be observed. To confirm this, the indirectly cooled IGBT module in HEV system was subjected to a complete PCsec (power cycling with turn-on time of a few seconds) test after incoming inspection until end of life to check the remaining active lifetime after the km test drive. Page 5 of 6 For this purpose, the modules were mounted on a cooler and heated up to the maximum specified IGBT junction temperature of 150degC

6 within a few seconds through a load current (turn-on time). Then the load current was turned off until the junction temperature was reduced by 100K. The turn-on time was kept constant even if the junction temperature exceeded 150degC as this reflects the real conditions in the application. Under these conditions the 95%- lifetime of a new module is cycles. The end of life failure criteria was defined as maximum change ±5% of initial values of Vcesat/Vf which were online monitored during the whole PCsec test. As a result, the full power cycle capability (qualification limit) was still reached after 200,000km of field test. The detail test results are as follows. Figure 12. Result of end of life test presented in a Weibull plot (Traction motor modules) Figure 11. C-SAM images and failure mode from the end of life test with 200,000km tested module Ultrasonic images (C-SAM) were obtained after the end of life test with 200,000km tested module (see Figure 11. at left show). No delamination of the observed system solder layers could be found. Instead, chip solder degradation observed. The analyzed device reached end of life through chip bond wire liftoff (see Figure11. at right show). The end of life cycles of the modules (for traction motor and generator) were entered into a Weibull plot. As a result the 5% Weibull criteria was passed (see Figures 12) Conclusions None of the depicted failure modes could be found on the fieldtested power modules after quality analysis. Lifetime simulation of the modules revealed no significant lifetime consumption. To confirm this, the indirectly cooled IGBT module was subjected to a standard power cycling test until end of life after the 200,000km test drive. As a result, the field stressed module showed the same active lifetime as a new module from the factory. These valuable results from the field-stressed power modules were furthermore used to verify our model for lifetime calculation. References 1. A. Christmann, M. Thoben, K. Mainka, " Reliability of Power Modules in Hybrid Vehicles, " PCIM Europe 2009, Nuremberg. 2. M. Thoben, K. Mainka, R. Bayerer, I. Graf, et al, "From vehicle drive cycle to reliability testing of Power Modules for hybrid vehicle inverter," PCIM Europe 2008, Nuremberg. 3. Indrajit Paul, Laurent Beaurenaut, Frank Sauerland, Marina Stoikova, " Application based modified reliability tests and their physical correlation with lifetime assessment models," PCIM Europe 2013, Nuremberg Cell/PIP/index.html C. Castro, L. Beaurenaut, Influence of application parameters on the life time expectations of power modules used in (H)EV main inverters, EEHE 2014 Page 6 of 6