As diverse as the requirements may be in different application areas, the primary focus in all cases is the development of reliable encapsulation technology for power modules. The most common encapsulation solutions today are soldered modules with and without a backplate, and, more recently, backplate-less modules using sintering technology. These encapsulation technologies have varying advantages and disadvantages, which is why, when designing with lifespan in mind, an evaluation of these technologies is required in relation to the requirements for hybrid and electric vehicle applications. Changes in ambient temperature, for example, in the coolant water cycle, are responsible for passive thermal cycling. Furthermore, the power loss that occurs in power semiconductors produces brief (5-20 seconds) temperature rises from T = 40 °C to 60 °C. Here, the power semiconductors are heated, for example, from the coolant temperature of 70°C to over 110°C–130°C, and then cooled back down to the coolant temperature. Due to the varying coefficients of thermal expansion of the materials used, all these temperature changes result in mechanical stress. This causes fatigue in the solder joints and connections, ultimately leading to component failure.
Avoiding Soldered Connections:
In modules without a motherboard using pressure-contact technology, several methods are employed to increase module reliability. By consistently avoiding soldered connections, solder fatigue—a fundamental mechanical failure in power modules—can be completely eliminated. Here, the soldered connections between the chips and the ceramic substrate (DBC) are replaced by a sintered layer, and the connection is made using pressure-contact technology. Eliminating the motherboard has several advantages: firstly, the thickness of the thermal paste layer between the module and the heatsink can be reduced. Thermal paste is one of the main factors contributing to the overall thermal resistance of the power module; therefore, the thinnest possible layer of thermal paste should be used. In modules with a motherboard, a thermal paste layer of 75–150 µm thickness is required to compensate for the motherboard's flexibility. In modules without the motherboard, the main problem to address is how to compensate for the surface roughness of the heatsink and the DBC surface, so a 20-30 µm thick layer of thermal paste is sufficient. Removing the motherboard eliminates one of the main causes of thermal stress (see Figure 1).
Temperature-induced stress is efficiently reduced and reliability is increased, as demonstrated in passive accelerated thermal shock tests at 40°C/125°C: for sintered modules without a baseplate, the number of potential thermal shocks increased by a factor of 15. Another advantage of eliminating solder on the interconnects and baseplate is that, in modules with a baseplate, the soldered DBC areas must be minimized to reduce material fatigue at the solder joints; here, the high thermal conductivity of the baseplate ensures the necessary heat diffusion. In contrast, in a module design without a baseplate, the DBC area can be larger.
Optimal heat distribution
This paper analyzes the placement of IGBTs and a flyback diode in a 400A, 600V, three-phase inverter module. In modules with a base plate, two 200A IGBTs and two 200A flyback diodes are used per semiconductor switch. A complete phase consists of four IGBTs and four flyback diodes. The most optimized arrangement for modules without a base plate is four 100A IGBTs and two 200A flyback diodes per switch (eight IGBTs and four flyback diodes per phase). This means that the base area of a three-phase module without a base plate will be approximately 10% larger than that of a module with a base plate (see Figure 2).
When the inverter is operating, conduction and switching losses occur, meaning the power semiconductors are a local heat source. Using 3D finite element calculations, the thermal diffusion in an inverter module and heat sink for any given operating state can be calculated. For example, when a hybrid or electric vehicle accelerates, most of the power losses occur in the IGBTs, while the flyback diodes are subjected to a smaller load (see Figure 3).
Therefore, in the thermal image, the IGBT positions appear as strong heat sources. In the case of modules with a base plate, the heat is concentrated in the center of the three-phase configuration. Due to the final position of the semiconductors and the short distance between the phases, the IGBT temperature is among the highest at this point. Although in this operating state the freewheeling diodes are only subjected to a moderate load, the IGBTs cause the temperature of the diodes in the center of the module to rise considerably. At the ends of the inverter module, the diode temperature, in contrast, is 15 °C lower. Despite the base plate, the power semiconductors in the outer regions of the inverter module are progressively cooler than those in the center. This ultimately leads to a non-homogeneous heat distribution across the three phases: the average thermal load of the IGBTs in the center phase is almost 10°C higher than the average temperature of the IGBTs in the outermost phase. The difference between the maximum and minimum IGBT temperatures exceeds 20°C. The center phase limits the usable electrical power of the entire inverter module. This has two consequences: firstly, the cooling conditions and load must be chosen to prevent excessively high temperatures in the center phase; secondly, temperature-induced failure mechanisms have a stronger effect in the center phase. This means that the power circuitry design engineer for the inverter must always consider the temperature of the center phase.
In SKiM modules without a backplate, heat distribution is much more homogeneous: here, too, the IGBT positions can be seen as the strongest heat sources. However, since thermal losses are distributed across multiple positions and the distance between the DBCs is greater, more space is available for heat dissipation. The resulting losses can be effectively dissipated, reducing mutual heating between the IGBTs and diodes. Optimal heat dissipation also ensures a homogeneous load distribution across the different phases: the temperature of the IGBTs and diodes across the three inverter phases is homogeneous, and the average temperature of the IGBTs in the three phases is almost identical. The maximum temperature difference between the IGBTs is no more than 10 °C. The load is distributed evenly, allowing for optimal use of the available cooling capacity and thus simplifying the overall system design. In addition, temperature sensors on each ceramic DBC substrate allow for separate evaluation of individual phases, providing an additional monitoring option for operating temperatures.
Temperature and Lifespan:
To analyze the actual thermal load of an operating inverter, time-dependent loads must be considered. During the actual operation of a hybrid or electric vehicle, several load states occur: during vehicle acceleration, the IGBTs are under a high load, while during deceleration, when energy recuperation takes place and the electric motor battery recharges, the flyback diodes are under the highest load. To describe the time-dependent heating of the inverter module, the behavior of the power module must also be investigated for charging cycles in the 0.1 s–30 s range. The thermal resistance of the IGBTs increases over time for both configurations according to the duration of the charging pulses (see Figure 4). Heat begins to flow, spreading from the power semiconductors toward the heat sink, causing the entire module to heat up. If the charging pulses last longer than 30 seconds, the entire module will heat up, and the thermal resistance will stop increasing.
Time-dependent thermal resistance values can now be used to calculate the thermal load acting on semiconductor switches during operation. To do this, real-world charging cycles, as would occur in a real-world application, are used to simulate typical charging states and the duration of charging pulses. Let's take the example of a cycle in hybrid vehicles. During the initial start-up phase and acceleration phases, energy is drawn from the battery and used to power the electric motor. In these acceleration phases, the power output reaches up to 60 kW. The temperature of the IGBTs rises to 95 °C linearly in accordance with the inverter output (see Figure 5). During constant-speed phases, very little power is required from the inverter, and the semiconductor temperature decreases again. During deceleration, the goal is to recover as much energy as possible to feed back into the battery. In this case, the power loss of both the IGBTs and the diodes is approximately the same, while the heat to be dissipated is at its highest level, and the IGBTs reach almost 110 °C.
The maximum temperature rise of the IGBTs is ΔT = 40 °C. In terms of module lifetime, this is equivalent to 6 million charge cycles (see Figure 6). The homogeneous temperature distribution is so crucial to the inverter's service life and impacts its design that if the temperature rise is only 10 °C higher – ΔT = 50 °C – the number of possible charge cycles is reduced by a factor of three, to only 2 million cycles. To facilitate long-life design and optimize semiconductor utilization, homogeneous loss distribution is absolutely essential.
conclusion
, sintered modules without a backplate offer a range of possibilities for improving the reliability of inverters for hybrid and electric vehicles. The disadvantages of solder connections and the varying temperature expansions on the backplate are eliminated. The optimized design ensures a largely homogeneous temperature distribution across all power semiconductors during operation. This means that, in calculations of expected service life, all three phases can be considered equally, simplifying inverter design. Inverter reliability, even under considerable active and passive temperature variations, is significantly improved. This is evidenced by the diverse applications for sintered modules without a backplate, such as in electric powertrains, automobiles and commercial vehicles, as well as in demanding applications like racing cars.
Author:
Dr. Volker Demuth, SEMIKRON Product Manager
