NON-TECHNICAL ABSTRACT Silicon-based electronic devices are the main component in virtually all microelectronic applications, e.g., computers and computer chips that are everywhere in cars, appliances, etc. For high-power and high-temperature uses, however, e.g., on-engine chips, power grid controls, etc. Si-based electronics are either inefficient or not usable at all. Silicon carbide is the most promising alternative, but, despite major advances in the last decade, including breakthroughs by the present team, difficult technical problems remain to be resolved. The proposed research addresses these issues with a mix of experimental and theoretical techniques. At the heart of the difficulties is the interface between silicon carbide and its native oxide, namely silicon dioxide, and the impact of the oxidation process on the underlying crystalline material. The expected results will be relevant to the broader field of the oxidation of diverse materials. Educational outreach projects will highlight to high school teachers and students the special needs of high-power, high-temperature electronics and the impact of research advances in important applications.
Silicon carbide is a promising alternative to Si for high-temperature, high-power electronics because of its larger energy gap and heat-conduction coefficient, but also because its native oxide is silicon dioxide. The SiC/SiO2 interface, however, is more complex that the Si/SiO2 interface, which is at the heart of Si-based electronics. The main problem is that oxidation releases C atoms, some of which are stuck at the interface as defects. Despite major advances in passivating defects at the SiC/SiO2 by N and H, including breakthroughs by the present team, electron mobility remains lower than desirable for applications. Recent evidence points to subtle changes in the underlying SiC substrate. The proposed research will combine state-of-the-art microscopy, electrical measurements, and theory to elucidate the oxidation process at the atomic scale, identify the origins of undesirable effects and defects, and identify new design specifications to improve carrier mobilities. Extensive education outreach will make advances accessible to high schools and the community.
The primary objective of this NSF-GOALI project was to use a combination of experimental and theoretical methods to investigate the effect of oxidation on the structural and electronic properties of the substrate and the quality of the oxide-substrate interface. Silicon carbide (SiC) was selected as a prototype material because it is a semiconductor with a large energy gap, a feature that is needed for the fabrication of electronic devices that are suitable for applications involving high power (power grid controls, all-electric ships) or high temperatures (automobile controls). Though the large energy gap, compared with the energy gap of Si, is a big advantage for high-power, high-temperature applications, other conditions need to be met. In particular, electrons in the substrate, SiC in this case, need to have a high mobility. The mobility is generally limited by scattering from defects at the interface with the oxide or by oxidation-induced defects in the substrate itself. In earlier work, the same team of investigators developed a process involving post-oxidation annealing in NO gas, which reduces significantly the defect density at the SiC-SiO2 interface. The process was patented and licensed by Cree, Inc., for the production of commercial SiC-based power electronics. Further improvements are needed, however, for this highly important industry to make advances. Motivation for this project was provided by a finding that a possible cause of electron mobility degradation in the oxidized SiC substrate is caused by a thick (>10 nm) non-stoichiometric ‘transition layer’ between the SiC and the oxide. The main objectives of the project were as follows: (1) Pursue deeper understanding of the correlations between detailed atomic structure and electronic properties of the silicon dioxide-silicon carbide interface formed by high temperature oxidation of SiC. (2) Pursue understanding different interface chemical modification processes and their impact on the performance of SiC metal-oxide-semiconductor field-effect transistors (MOSFETs) (3) Determine the key mechanisms that limit the transport of electrons in the SiC MOSFET channel and result in less than ideal performance. The main findings were as follows: (i) Theoretical and experimental work conducted under this project has conclusively determined that the interfacial region is chemically stoichiometric, and that the effects that have been attributed to a transition layer are in fact caused by interfacial roughness (Figure 1) On the other hand, carbon atoms emitted into the substrate during oxidation results in a stable carbon di-interstitial complex that causes endemic degradation of electron mobility. (ii) Following reports of high mobility produced by incorporating sodium in SiC-SiO2 structures, we provided theoretical understanding of the phenomenon and confirmed it experimentally (Figure 2). Sodium, however, is highly mobile and devices are unstable and unsuitable for applications. (iii) A comprehensive theoretical study of the oxidation mechanisms of the different SiC surfaces and comparison with the corresponding oxidation mechanisms of different Si surfaces was carried out and the results explained and reconciled diverse experimental data (Figure 3). (iii) Following a report that using phosphosilicate glass (PSG) as the gate dielectric of MOSFETs results in higher performance, we established that, while the performance is high, the stability of the device is extremely poor as a result of polarization of the dielectric. A new approach was developed using a stacked gate oxide that consists of a thin (~100 A) PSG layer at the interface which results in better stability while maintaining the high performance (Figure 4). (iv) High channel mobility has been obtained using antimony as a surface counter-dopant (Figure 5). This process is very promising for future SiC MOSFET technology. It can also serve as an experimental platform for independently studying the effects of trap passivation and counter-doping on 4H-SiC MOSFET channel mobility. All these results provide very valuable information that benefits the processing methodology of commercial SiC power devices by industrial partner Cree Inc. as well as other private companies involved in this business. On the outreach and educational front, the project funded a Masters student (Sorrie Ceesay) at nearby historically black Fisk University. Mr. Ceesay gave an oral presentation based on his Master’s thesis at the 2012 Spring meeting of the Materials Research Society in San Francisco, CA (Figure 6). Upon graduation, he entered Dayton University in Dayton, OH, as a Ph.D. graduate student in Materials Science and worked on this project for the first year, mentored by his former Fisk advisor Prof. Weijie Lu who had moved to the Air Force Laboratory near Dayton. The project also funded an annual workshop for high-school teachers who were introduced to SiC and its potential for power and high-temperature electronics at a level suitable for teaching at high schools.
