SiC Power Module Breakthrough Enhances EV Performance and Reliability
In a significant leap forward for electric vehicle (EV) technology, researchers from Chongqing University and China Resources Microelectronics have unveiled a novel optimization strategy for silicon carbide (SiC) power modules that dramatically improves both electrical and thermal performance—critical factors in the next generation of high-efficiency motor controllers. Published in the Journal of Power Supply, the study demonstrates how a synergistic redesign of layout, interconnects, and cooling architecture can unlock the full potential of SiC semiconductors in automotive applications.
The global push toward electrification has intensified pressure on power electronics to deliver higher efficiency, greater power density, and enhanced reliability. While SiC devices inherently offer superior characteristics—such as higher breakdown field strength, wider bandgap, and better thermal conductivity—than traditional silicon (Si) counterparts, their performance in real-world modules has often been bottlenecked by legacy packaging techniques. This mismatch has limited the practical benefits of SiC, particularly in demanding environments like EV traction inverters.
Led by Rongyao Ma and Kaifeng Tang from China Resources Microelectronics (Chongqing) Limited, in collaboration with Xiaofei Pan, Zhifeng Shao, Peng Sun, and Zheng Zeng from Chongqing University, the research team tackled this challenge head-on. Their approach was not incremental but holistic—rethinking the entire module architecture from the chip level up to the heat sink.
At the heart of their innovation lies a reimagined multi-chip layout. Traditional SiC modules often adopt a linear, “single-row” arrangement inherited from Si-based designs. However, this configuration leads to uneven current distribution and localized hot spots, especially in the central chips, which suffer from higher junction temperatures and accelerated aging. The team proposed a “W-shaped” staggered layout where six 1,200 V/600 A SiC MOSFET chips are offset by half their width—approximately 2.5 mm for a standard 5 mm die. This simple geometric shift yielded profound results: finite element simulations showed an 11°C reduction in peak junction temperature compared to the conventional layout, while parasitic inductance—a critical parameter affecting switching losses and voltage overshoot—was reduced by 0.16 nH. Crucially, this improvement came without compromising electrical performance, striking a rare balance between thermal management and high-frequency switching fidelity.
Equally transformative was the adoption of copper wire bonding via the Die Top System (DTS) technology. Conventional aluminum wire bonding, long the industry standard, struggles with SiC’s smaller active area. With less surface available for interconnects, aluminum wires face excessive current density, leading to premature failure under high-load conditions. Copper, with its higher conductivity and mechanical robustness, offers a compelling alternative—but it cannot be directly bonded to typical SiC chip metallization without risking damage from ultrasonic energy during the bonding process. The DTS solution elegantly circumvents this by first sintering a thin copper foil layer onto the chip surface using silver sintering. This intermediate layer not only protects the fragile die but also improves current spreading and reduces localized heating. The result is a more reliable, higher-current-capable interconnect that aligns with the demanding lifecycle requirements of automotive applications.
Beyond the chip and interconnect, the team turned their attention to the cooling infrastructure. Most commercial modules rely on circular pin-fin heat sinks for direct liquid cooling. Recognizing the limitations of this geometry, the researchers explored elliptical pin-fins, which offer superior surface-area-to-volume ratios and more favorable flow dynamics. Using Response Surface Methodology (RSM)—a statistical and mathematical technique for modeling and optimizing complex systems—they fine-tuned four key parameters: the vertical and horizontal spacing between fins (D1 and D2), and the short and long axes of the elliptical cross-section (a and b). The optimization sought to minimize both chip temperature rise and coolant pressure drop—a classic engineering trade-off. After extensive multi-physics simulation, the optimal configuration emerged: D1 = 2.44 mm, D2 = 2.32 mm, a = 0.78 mm, and b = 0.93 mm. This design reduced the maximum junction temperature by 10°C compared to a standard elliptical layout, with only a negligible 0.2 kPa increase in pressure drop—a remarkable achievement that enhances cooling without overburdening the vehicle’s coolant pump.
To validate their theoretical and simulation work, the team fabricated two prototype modules based on the widely used HybridPACK™ Drive package standard. Module A represented the baseline: six chips connected with aluminum wires on a ZTA (zirconia-toughened alumina) direct bonded copper (DBC) substrate, cooled by a conventional circular pin-fin heat sink. Module B embodied the full suite of innovations: the W-shaped chip layout, DTS copper wire bonding, an AMB (active metal brazed) substrate for better thermal conductivity, and the optimized elliptical pin-fin cooler.
The prototypes underwent rigorous testing on two custom-built platforms. A double-pulse test setup evaluated switching behavior and parasitic inductance under 600 V/300 A conditions. While both modules exhibited excellent performance, the subtle reduction in voltage overshoot for Module B confirmed the efficacy of the layout and interconnect improvements in managing high di/dt transients. More compelling results came from the power back-to-back test, which simulates real-world inverter operation by connecting two identical inverters in a regenerative loop. Under sustained load at a coolant temperature of 65°C, Module B consistently ran cooler. At a flow rate of 5.0 liters per minute—a realistic condition for many EVs—its peak junction temperature was 6°C lower than Module A’s. This seemingly modest difference translates into a substantial extension of operational life, as power module lifetime is exponentially related to thermal cycling amplitude.
The implications of this work extend far beyond a single module design. The researchers emphasize that their methodology is not tied to the HybridPACK form factor. The principles of staggered layout, advanced interconnects, and statistically optimized cooling are universally applicable. Automotive suppliers and power electronics designers can adapt these strategies to their own packaging platforms with minimal retooling, accelerating the adoption of high-performance SiC solutions across the industry.
From a reliability standpoint, the advancements are equally critical. It is well-documented that, when packaged using traditional Si methods, SiC modules can exhibit lifespans as short as one-third that of their silicon predecessors—despite the inherent material advantages. This paradox arises from the intense localized heating and mechanical stress in the smaller die. By simultaneously addressing thermal hotspots, current crowding, and interconnect fatigue, the Chongqing team’s approach directly confronts this reliability gap. The use of silver sintering for die attachment, AMB substrates, and copper bonding all contribute to a more robust assembly capable of withstanding the harsh thermal and vibrational environment of an EV powertrain.
For the EV market, this research arrives at a pivotal moment. Automakers are racing to extend range, reduce charging times, and lower system costs. Every watt saved in the inverter translates directly into more miles per kilowatt-hour. Higher switching frequencies enabled by low-inductance SiC modules allow for smaller, lighter passive components like inductors and capacitors, further boosting power density. And perhaps most importantly, enhanced reliability reduces warranty costs and builds consumer confidence in EV technology.
The work also underscores a broader trend in power electronics: the end of “drop-in” replacements. Simply swapping a Si IGBT for a SiC MOSFET in an old package is no longer sufficient. To truly harness wide-bandgap semiconductors, the entire module must be co-designed as an integrated electro-thermal-mechanical system. This requires deep collaboration between device physicists, packaging engineers, thermal specialists, and system integrators—a multidisciplinary effort exemplified by the Chongqing consortium.
Looking ahead, the team suggests several avenues for further refinement. While copper wire bonding represents a major step forward, emerging technologies like planar interconnects or sintered copper clips could offer even lower parasitic inductance and better current sharing. Similarly, advanced cooling techniques such as jet impingement or microchannel integration might push thermal performance even further. Nevertheless, the current work provides a practical, manufacturable, and immediately impactful blueprint for the industry.
In summary, this study by Ma, Tang, Pan, Shao, Sun, and Zeng is more than an academic exercise; it is a roadmap for the next generation of automotive power modules. By harmonizing layout, materials, and thermal design, they have demonstrated a clear path to unlocking the full promise of silicon carbide—delivering the efficiency, power density, and reliability that the electric vehicle revolution demands.
This research was conducted by Rongyao Ma (School of Microelectronics and Communication Engineering, Chongqing University; China Resources Microelectronics (Chongqing) Limited), Kaifeng Tang (China Resources Microelectronics (Chongqing) Limited), Xiaofei Pan (School of Electrical Engineering, Chongqing University; China Resources Runan Technologies (Chongqing) Co., Ltd), Zhifeng Shao (China Resources Microelectronics (Chongqing) Limited), Peng Sun (School of Electrical Engineering, Chongqing University), and Zheng Zeng (School of Electrical Engineering, Chongqing University). It was published in the Journal of Power Supply, Vol. 22, No. 3, May 2024, with DOI: 10.13234/j.issn.2095-2805.2024.3.78.