PCIM 研讨会 | 口述演讲：筑牢器件底层壁垒，精进功率半导体封装与可靠性技术

"芯片性能决定功率上限，封装与可靠性决定器件寿命，散热、键合、封装材料是碳化硅产业化落地的关键瓶颈。"

主持人

温旭辉

中国科学院电工研究所，中国

主持人

Naoto 

Fujishima

富士电机，日本

主持人导读

New Pressure Technology for SiC Power Modules    

演讲简介：

CN-iCuTech’s new D2M pressure tech for SiC power modules addresses thermal challenges of high-performance SiC devices (26mm² chips hit 200A/8mΩ Rdson, cost near 7th-gen IGBTs). Traditional pressure technologies have good CTE performance but poor thermal spreading and application limits. D2M uses P2D to press discrete TO SiC devices onto organic/ceramic substrates without solder/sintering, with laser tech connecting pins to a 0.5-0.8mm copper frame and novel thermoelectric paste boosting conductivity. Its integrated copper plate enables excellent thermal spreading, a laser-welded smart board ensures homogeneous current sharing. Validated on a 1000A/1200V SiC EconoDual, it optimizes thermal resistance and cost.    

演讲人简介：

Dipl. Engineer, studied power electronic and drive technology in Germany. Member of R&D associations in Germany, member of European power electronic association and board of Directors PCIM Asia. Set up inverter company in Germany and designed high power inverter (>250kW) with high power IPM in 1990. Joined Semikron-Danfoss in 2000. More than 45 years of experience in power electronic applications. 20 years in R&D, Designed high power inverter and EV inverter. Focus on packaging technology for IGBTs and SiC devices to optimize power density and reliability. Since 20 years in Asia and supporting Chinese companies in their power electronic design. Special in renewable energy application with new topologies by using novel power module packaging. In the last few years, the main focus was e-mobility, special EV inverter design with SiC devices and optimized power module packaging. Since 1st April 2025 CTO of CN-iCuTech Semiconductor HongKong and YiTong Semiconductor China    

燒結技術於SiC 功率模組之應用

演讲简介：

由於燒結技術具有強韌的熱可靠度以及卓越的電學與熱學性能，已廣泛應用於功率模組中。雖然傳統的壓力輔助銀（Ag）燒結提供了優異特性，但銀的高昂成本仍是一個顯著缺點。本研究證實，焊錫提供的晶粒剪切強度最低；而壓力輔助燒結銅所提供的晶粒剪切強度，則與壓力輔助燒結銀相當。本研究也將針對這三種界面材料，提供並討論其它的結構特性分析。

演讲人简介：

楊韶綸先生現任職於日月光集團（ASE Group）中壢廠中央研發工程部，負責領導先進封裝技術的開發專案。憑藉深厚的機械工程背景，楊先生專精於 IC 封裝的端到端整合，業務涵蓋從新材料導入到大規模量產的完整流程。

他的技術專長聚焦於 SiP 與 Power IC 的結構開發，特別是在異質整合（Heterogeneous Integration）、3D 垂直結構架構及製程優化導入等尖端領域。目前，他致力於功率模組封裝解決方案的研發，重點在於針對次世代功率電子開發關鍵整合技術。

Power cycling and thermal shock reliability of novel copper preforms

演讲简介：

In this paper, the process development and reliability of novel copper preforms are presented, which while offering a substantially reduced manufacturing footprint, ensure enhanced reliability, high thermal conductivity, excellent thermo-mechanical fatigue resistance. Further, the preforms offer the capability to realise a single material system for top side of the die, die-attach and the substrate attach, while sintering at 250°C for 5min with 20MPa sintering pressure. Such kind of a novel approach does not exist in the industry today and the reliability of such a set-up is discussed through power-cycling and high-stress thermal shock analysis.

The SiC devices assembled through this process show no failures even after 500000 cycles through power cycling and zero delamination after 1000 thermal shock cycles at +175/-65°C with a 5min dwell time, tested in a two-chamber thermal shock chamber (air-air conditions).

演讲人简介：

Sri Krishna Bhogaraju is the CEO of CuNex and a materials scientist specializing in copper sintering for high-power electronics. He has authored multiple peer-reviewed publications and holds patents on low-temperature copper sinter paste technologies, with deep expertise in microstructure–property relationships, reliability, and power electronics packaging.

用于高可靠性PEBB模块的双对称S型通道复合散热器设计

演讲简介：

本文针对直流系统场景与数据集，提出一种基于优化传统机器学习算法的轻量化电弧故障检测方案。设本文针对300 kW电力电子积木（PEBB）提出了一种双对称S型通道复合散热器（DSCH）。该设计通过优化S型通道拓扑结构，利用二次涡流增强湍流混合，使总热阻较平行通道降低了10.7% 。在300 kW原型机上的实验结果验证了该设计具有优异的温度均匀性（MATD为4.88）和低压降（1.1136 kPa），为兆瓦级功率转换系统和AI数据中心基础设施提供了高效的热管理方案 。计了一组 8 维场景专属特征，在充分利用直流电弧故障物理特性的同时，避免了复杂的卷积运算。综合实验结果表明，所提优化机器学习方法的检测精度最高可达 0.9966， 所有优化机器学习算法的参数量均显著降低，推理时间缩短约 78%，该方法可在无神经网络处理器加速的低配置微处理器上部署。

演讲人简介：

郑家豪 (Jiahao Zheng) 为中国浙江大学电气工程学院在读学生。2026 年开始，在浙江大学电气工程学院攻读博士学位 。在此期间，他深度参与了针对 300 kW 高功率电力电子积木（PEBB）原型机的热管理系统开发与实验验证工作 。他目前的研究兴趣包括：面向电网与工业应用的高压大功率变换器拓扑与控制架构设计 、用于 AI 数据中心供电的固态变压器（SST）设计 以及兆瓦级变流器的高效散热方案 。

浙江大学、英飞凌科技、

香港微电子研究院、比亚迪

最佳论文奖 | 青年工程师奖

高校科学家奖 | 优秀墙报奖

一位长期活跃在全球电力电子行业

的PCIM研讨会主席

二十年行业洞察：透视功率半导体未来，

为何PCIM Asia是不可或缺的支点？