Thermal characterization system for transient thermal impedance measurement and power cycling of IGBT modules. Issue 12 (December 2015)

• Record Type: Journal Article
• Authors: Zheng, Hanguang; Ngo, Khai D.T.; Lu, Guo-Quan
• Is Part Of: Microelectronics and reliability. Volume 55:Issue 12 Part A(2015)
• Page Start: 2575
• Page End: 2581
• Publication Date: 2015-12
• Subjects: Thermal impedance measurement -- Power cycling -- IGBT modules; Electronic apparatus and appliances -- Reliability -- Periodicals; Miniature electronic equipment -- Periodicals; Appareils électroniques -- Fiabilité -- Périodiques; Équipement électronique miniaturisé -- Périodiques; Electronic apparatus and appliances -- Reliability; Miniature electronic equipment; Periodicals; 621.3815
• DOI: 10.1016/j.microrel.2015.08.016
• Languages: English
• ISSNs: 0026-2714

Abstract

Abstract: This paper presents a custom-designed thermal characterization system for insulated-gate bipolar transistor (IGBT) modules. The main purpose of this system is to provide experimental capability for transient thermal impedance (Zth ) measurement and power cycling reliability test on IGBT modules. LabVIEW programming was applied for circuit control and data acquisition. Gate-emitter voltage (Vge ) of IGBT was used as the temperature-sensitive parameter for junction temperature (Tj ) measurement. The linear relationship between Vge and Tj for IGBT can be measured by the system. Variation coefficient of lower than 0.5% was achieved for Zth measurement. IGBT modules with the same sample geometry but different die-attach processes were used for system verification: IGBT modules with nanosilver-sintered die-attach have on the average a 12% lower Zth than SAC305 soldered modules. The power cycling capability of the system was demonstrated by cycling a nanosilver-sintered IGBT module with Tj cycled between 45 °C and 175 °C. The cycling lifetime of the IGBT module was recorded as 48, 000 cycles with the failure criteria of 20% increase on Zth . Such lifetime is 50% longer than the predicted value from lifetime prediction model.