Join Us and Move the World Forward

Background: Driven by carbon peak and neutrality targets, energy saving of data centers, super-fast charging of electric vehicles, and comprehensive utilization of renewable energy such as PV will be key directions of technical development in the future. The PUE of next-generation data centers will decrease towards 1.0 or lower. The output power of smart charging facilities and PV inverters will reach the MW level, and the heat flux density of chips will reach 1000 W/cm2. Heat will be a big challenge. Advanced thermal technologies should be used to achieve low thermal resistance from dies to systems. Challenges: 1. Explore microscale design methods for high thermal conductivity films and composite material to dissipate heat from high-performance chips. 2. Develop high-performance and efficient cold plates using innovative heat transfer and flow structures, materials, and processes to improve liquid cooling performance. 3. Develop technologies for two-phase heat transfer of high limit, low temperature difference, and long distance (direct or indirect cooling), including working principles, structures, materials, and processes related to two-phase heat transfer, temperature equalization, and fluid transport, to dissipate heat from high-density chips (200–1000 W/cm2) in a small space. 4. Optimize the design of efficient cooling systems and enable intelligent control. Research the mechanisms of efficient vapor-liquid separation, flow control, and pressure control, and develop the related components.

Background: IGBT/SiC power modules are critical to power conversion in high-power systems. High integration, efficiency, density, and reliability are the development trends of power modules. The junction temperature of next-generation power modules will be higher than 200°C. Performance and reliability will be major challenges. Advanced module architectures and packing technologies should be explored to continuously improve the product competitiveness. Challenges: 1. Use new structures, materials, and technologies to halve the system thermal resistance of power modules and double the power density compared with existing products in the industry. 2. Use new technologies and materials to ensure that power modules can serve for more than 25 years with the long-term junction temperature greater than 200°C. 3. Use new intelligent integration architecture (control and drive) to implement real-time monitoring, self-warning, and self-protection of power modules, keeping the failure rate less than 10 PPM. 4. Use new interconnection design and layout to ensure that the parasitic parameter of power modules is less than 2 nH and their operating frequency is twice that of exiting products.

Background: 5G is accelerating ICT convergence, bringing digital to every person, home, and organization for a fully connected, intelligent world. In the process of building the Internet of everything (IoE), data computing, storage, and transmission rates increase by tenfold to hundredfold, but the size of equipment almost keeps unchanged. The power density, conversion efficiency, and transmission efficiency of power modules should be improved. Especially for board-level power supply to supercomputing chips such as xPUs, low-voltage power supply with over 1000 A current to chips will be a major technical challenge. Challenges: In response to the continuous increase of xPU computing performance and power supply current: 1. Develop a power supply solution that features high power density and conversion efficiency (over 1000 A) for xPUs. 2. Design an innovative control control algorithm that meets the dynamic requirements of supercomputing chips for ultrahigh current (2000 A/µs) to achieve optimal dynamic performance.

Background: As global governments set their carbon peak and neutrality policies, wind-solar electrochemical energy storage and new energy vehicles are booming. High battery safety and long service life are required in these two scenarios. As complex working conditions come with large-scale battery application and long service period, the top priority is to achieve intrinsic safety throughout the battery life cycle and improve the battery service life. Challenges: 1. High battery safety: The working conditions of batteries are complex, the consistency between different batteries varies greatly, and the service period is long. A large number of batteries are connected in series and parallel in a system. Therefore, a battery failure model covering the life cycle needs to be established, and an optimal material system and structure need to be designed to effectively detect battery failures and mitigate the effects of the failures to guarantee battery safety. 2. Long battery life: Establish a high-precision aging model for different batteries that work in various conditions and are connected in series and parallel. Optimize the material system, structure design, and control policies based on the working conditions and aging model to extend the service life of batteries to over 26 years (calendar + cycle life). Develop effective methods to rapidly evaluate and predict the service life of batteries under different working conditions.

Background: Conventional battery modeling methods do not deliver satisfactory results in actual environments. Application mechanisms can be combined with big data and technologies to explore digital battery modeling under real and complex working conditions, which will bring huge benefits. For example, higher life prediction accuracy has a great impact on the design, manufacturing, and production of batteries and will facilitate the research of new-generation batteries. Challenges: 1. Battery is a complex nonlinear system that involves a number of disciplines such as material, physics, chemistry, electricity, and heat dissipation. 2. Model precision is difficult to increase due to various aging mechanisms, differences between individual batteries, variations in combinatorial topology, and dynamic yet complex working conditions. 3. Multi-objective joint optimization is necessary in various business requirements and application scenarios. 4. It is challenging to achieve high availability using a limited amount of data and test scenarios.