(Invited) Wide Band-Gap on Its Hard Way up - The Trouble Starts Just Outside the Chip

Abstract

Since the market introduction of the first commercial SiC-Schottky diode in 2001 wide band-gap (WBG) semiconductor devices made significant progress. In the meantime, SiC-MOSEFTs are commercially available from several manufacturers and also GaN-HEMTs appeared on the market. Devices from both materials are available in a growing variety of packages and the components conquer more and more applications and market shares.The electrical performance of those WBG-devices is a quantum leap compared to their silicon counter parts. The specific on-resistance is at least an order of magnitude lower than in silicon and the switching speed can be so high, that the switching process is finished within nanoseconds and the associated losses are orders of magnitude lower than in silicon. Due to the lack of electron-hole-plasma the safe operating area is extremely wide, there is no risk of filamentation and other instabilities, and diodes do not show a forward recovery over-voltage [1]. On top, the devices can be operated at much higher temperatures than silicon ones. Thus, WBG-devices seem to approach the ideal switch.However, this is only true for the semiconductor chip. Of course, there are a variety of technical challenges not to mention the significantly higher cost. SiC-MOSFETs show low channel mobility and low threshold voltage with a strange temperature behaviour pointing to interfaces far from ideal, while GaN-HEMTs show the current collapse and bulk leakage requiring significant voltage derating for stable operation.But the real trouble starts just outside the chip with the package. The packaging is a limiting factor with respect to parasitics preventing fast switching. Due to the high di/dt the smallest stray inductance has a large impact and can cause massive overvoltages as well as the high dv/dt leads to large currents even through small stray capacitances. In contrast to IGBTs, this cannot be compensated anymore by smart control circuitry because the switching event is over before the circuitry can respond [2]. And it is really difficult to measure such steep transients without introducing much additional stray inductance [3].The packaging is also a limiting factor with respect to materials preventing high temperature operation beyond 200°C. Either the packaging materials are not at all suited for higher temperatures or they cannot cope with the higher temperature-swings leading to much faster degradation of joints. Furthermore, Schottky-diodes show a rather high leakage current and might be subjected to thermal runaway at elevated temperatures [4].The package is also a limiting factor with respect to long term stability because of the extremely high fields on the semiconductor surface as well as in the packaging materials. Furthermore, the usual packages are not at all hermetic, so that moisture can intrude and degrades the blocking capability over time [5].On top, the passive components in the circuitry might not cope with the high switching speeds or the high frequencies. When going to frequencies beyond 100 kHz the losses in the inductive components become significant and in some applications even the limiting factor.After all, the WBG-chips might approach the ideal switch but the advantages cannot be fully exploited yet. Significant effort is still required to gain confidence in WBG-devices’ reliability and to make them attractive for a wider range of applications, consequently profiting from the economy of scale. Only this way, WBG-devices could step out of the high end niche and could really challenge silicon, which is still a powerful competitor. And even (theoretically) better WBG-materials like Ga2O3 are at the horizon.[1] S. Rugen et al., “Investigation of the Turn-on Behaviour of Silicon pin-Diodes and SiC-MPS-Diodes and its Impact on the Anti-parallel IGBT”, accepted for publication at EPE 2016[2] C. Bödeker et al., “Investigation of an overvoltage protection for fast switching silicon carbide transistors”, IET-PEL, volume 8, issue 12, pp. 2336–2342, 2015[3] M. Adelmund et al., “Optimisation of Shunt Resistors for Fast Transients”, accepted for publication at PCIM 2016[4] C. Bödeker et al., “Criterion for the Stability against Thermal Runaway during Blocking Operation and its Application to SiC-Diodes”, IEEE-JSTPE, already available in IEEEXplore, issue tbd., 2016[5] N. Kaminski, “Progress in Silicon Based Power Electronics and its Impact on Wide Band-gap Devices - Why use SiC and GaN if Silicon can do the job?”, JST’s International Forum on Power Electronics of Advanced Wide Bandgap Semiconductors, Kyoto, 2015