Charge Trapping and Single-Defect Extraction in Gallium-Nitride Based MIS-HEMTs

Abstract

Charge Trapping and Single-Defect Extraction in
Gallium-Nitride Based MIS-HEMTs

The trend to use electricity as a primary energy source will ever increase as major sectors like the individual and public transport are going to be more and more electrified. To meet the international goals on carbon emission, not only the production and consumption but also the conversion efficiency of electricity is going to play a crucial role. The ever-growing interest in wide-bandgap semiconductors like GaN or SiC is primarily driven by their fundamental material properties, which allow building much more efficient power conversion systems when compared to silicon technology.

GaN-based transistors, commonly referred to as HEMTs or MIS-HEMTs, are among the most attractive candidates for novel power semiconductor device concepts because of their high breakdown voltage as well as their low specific on-resistance. Therefore, this work starts with an introduction of the material system, and state-of-the-art concepts for normally-on and normally-off devices. Although first devices are already commercially available, severe reliability issues – usually related to charge trapping – are still preventing them from a more widespread acceptance.

To improve the reliability of future GaN technology, a detailed physical understanding of the origin and the effects of these defects is of utmost importance. The focus of this work lies on the characterization and modeling of the defects responsible for BTI degradation in GaN MIS-HEMTs, which usually suffer from large (math image) instabilities already at nominal operating conditions. Therefore, well-established methods for BTI characterization and reliability modeling in silicon need to be carefully reassessed for their applicability to GaN technology.

One important phenomenon which is usually not considered in BTI studies for silicon technology is the electrostatic feedback of the trapped charges. Highlighting the importance of charge feedback effects on the observed BTI degradation of GaN/AlGaN MIS-HEMTs and its impact on defect modeling is one of the main findings presented in this thesis.

Another focus is the development of more robust methods for the calculation of single-defect parameters from RTN measurements in nano-scale GaN/AlGaN fin-MIS-HEMTs. The presented methods are used to obtain the characteristic time constants from the stochastic charge capture and emission events of RTN producing defects. Repeating those extractions for different bias conditions and temperatures allows to calculate other defect parameters like the vertical defect position or the trap level by assuming the transitions being governed by the NMP theory.

The insights into the physics of charge feedback caused by BTI degradation in large-area devices and the methods provided for the extraction of single-defect parameters from GaN/AlGaN MIS-HEMTs can be seen as a first step towards the identification of the microscopic defect structures responsible for BTI degradation in GaN technology. A promising path towards this goal would be the comparison of measurements on large-area and nano-scale devices with TCAD and first-principle simulations. The results of this thesis can thus be seen as a valuable contribution to future improvements of GaN technology.

Kurzfassung

Ladungsträgereinfang und Einzeldefektextraktion in
Gallium-Nitrid Basierten MIS-HEMTs

Wesentliche Sektoren wie der Individualverkehr und der öffentliche Verkehr werden in Zukunft immer stärker elektrifiziert werden, weshalb auch der Trend Elektrizität als Primärenergiequelle zu nutzen zunehmen wird. Um die internationalen CO $_2$ -Emissionsziele zu erreichen, werden nicht nur Produktion und Verbrauch, sondern auch die Umwandlungseffizienz von Elektrizität in Zukunft eine entscheidende Rolle spielen. Das ständig wachsende Interesse an Halbleitern mit großer Bandlücke, wie zum Beispiel Gallium-Nitrid (GaN) oder Silizium-Karbid (SiC), ist hauptsächlich durch ihre fundamentalen Materialeigenschaften begründet. Sie ermöglichen im Vergleich zur Siliziumtechnologie die Herstellung von wesentlich effizienteren Energieumwandlungssystemen.

GaN-basierte Transistoren, oft als HEMTs oder MIS-HEMTs bezeichnet, gehören zu den vielversprechendsten Kandidaten für neuartige Konzepte von Leistungshalbleitern, da sie hohe Durchbruchspannungen bei gleichzeitig niedrigem Einschaltwiderstand besitzen. Deswegen beginnt diese Dissertation mit einer Einführung in das Materialsystem und einer Erklärung der modernsten Konzepte für selbstleitende und selbstsperrende Transistoren. Gravierende Zuverlässigkeitsprobleme – üblicherweise verbunden mit dem Einfang von Ladungen in Defekten – verhindern derzeit eine noch größere Marktverbreitung, obwohl einige Bauteile bereits im Handel erhältlich sind.

Ein detailliertes physikalisches Verständnis des Ursprungs und der Auswirkungen dieser Defekte ist von äußerster Wichtigkeit um die Zuverlässigkeit zukünftiger GaN-Technologien zu verbessern. Der Schwerpunkt dieser Arbeit liegt daher auf der Charakterisierung und Modellierung von Defekten, die für die Degradation der Schwellspannung in GaN verantwortlich sind. Bereits unter normalen Betriebsbedingungen leiden diese Bauelemente oft schon unter großen Instabilitäten. Daher müssen auch etablierte Methoden für die Charakterisierung und Zuverlässigkeitsmodellierung von Defekten in der Siliziumtechologie nochmals sorgfältig auf ihre Anwendbarkeit auf die GaN-Technologie überprüft werden.

Die elektrostatische Rückkopplung von eingeschlossenen Ladungen ist ein wichtiges Phänomen, das bei Studien über die Bias-Temperaturinstabilität in Siliziumprozessen in der Regel nicht berücksichtigt wird. Die Bedeutung dieses Effekts für die Degradation von GaN/AlGaN MIS-HEMTs und ihre Auswirkungen auf die Modellierung von Defekten ist daher eines der wichtigsten Ergebnisse dieser Arbeit. Ein weiterer Schwerpunkt liegt auf der Entwicklung von robusteren Methoden zur Berechnung der Parameter von Einzeldefekten in GaN/AlGaN Fin-MIS-HEMTs. Die vorgestellten Methoden werden anschließend auch verwendet um die charakteristischen Zeitkonstanten aus den stochastischen Ladungseinfang- und Emissionsereignissen zu erhalten. Die Auswertung der Messungen bei verschiedene Spannungen und Temperaturen ermöglicht es auch, andere Parameter wie die vertikalen Defektpositionen oder das Energieniveau der Defekte zu berechnen. Hierbei wird angenommen, dass die Übergänge durch die NMP-Theorie beschrieben werden.

Die Physik der Ladungsrückkopplung durch Degradation und die Extraktion von Parametern aus Einzeldefekten kann als ein erster Schritt zur Identifizierung jener mikroskopischer Defekte gesehen werden, welche für Degradation in GaN-Technologien verantwortlich sind. Ein vielversprechender Ansatz für zukünftige Studien ist der direkte Vergleich von Messungen an großflächigen und nano-Bauelementen mit TCAD und First-Principle-Simulationen. Die Ergebnisse dieser Arbeit können daher als ein wertvoller Beitrag zur zukünftigen Verbesserung der GaN-Technologie angesehen werden.

Acknowledgements

First and foremost, I want to thank Prof. Tibor Grasser for teaching me how to become a researcher and for his continuous support throughout my PhD program. You surely are the most diligent and passionate researcher I ever met. You showed me that looking at the precise details can be the truly important part of research, and your talent to ask the right questions inspired large parts of this work. As my boss, I always knew that I could count on your support, even in situations which were very delicate for both of us.

Furthermore, I want to express my gratitude to be able to work in such a great place like the Institute for Microelectronics during the last years. Working in such a professional and enjoyable environment really helped me to carry on every time I was frustrated. Special thanks go to Michael Waltl: Without your efforts, none of the experimental single-defect studies would have been possible. I also want to thank Bernhard Stampfer for his great work on band-interactions within the NMP theory and for helping me to understand RTN and how to extract data from it.

I want to thank all of my colleagues for the valuable technical discussions but also for the distraction and the free time we were able to spend together. Thank you Wolfgang Gös and Franz Schanovsky for providing the basics to make my first months at the institute considerably easier. Thank you Markus Jech, Alexander Makarov, Gerhard Rzepa, Stanislav Tyaginov, Bianka Ullmann, and Yannick Wimmer for being also very good friends.

Without the support of Infineon Villach and the studies on GaN from Clemens Ostermaier, Roberta Stradiotto, and Peter Lagger, this work would not have been possible at all. Thank you very much for sharing your data and also helping me to interpret them correctly. I also want to acknowledge the great work of Global TCAD Solutions, which provided the simulation platforms necessary for my investigations. The fast and unbureaucratic support of Christian Kernstock, Ferdinand Mitterbauer, and Zlatan Stanojević really helped to keep my research on track. Last but not least, I want to express my gratitude to Ki-Sik Im and Jung-Hee Lee from Kyungpook National University, Daegu, Korea for providing the samples for my single-defect studies in GaN.

A big thank you also goes to all of my friends for their emotional support throughout the last years. Unfortunately, there is not enough space here to mention everyone, but I shall thank every single one of you in person. The good memories we share always help me to get back on track when I struggle.

Last but not least, my deepest gratitude goes to my family for supporting me in any decision of my life and for always standing behind me, no matter what happens. I doubt that I can find the right words to express how important all of you have been in order to reach this point.

Dear Renate, Alois, Christina, Ronald, Christian, and Sarah:
Thank you for all your support!