ECSCRM: Applications Steal The Show
There was a shift in emphasis at the sixth European Conference on Silicon Carbide and Related Materials (ECSCRM). The characteristics of specific substrates and devices stopped hogging the limelight, and issues surrounding silicon carbide (SiC) in commercial systems emerged as the key talking point.
This theme was taken up by Peter Friedrichs, managing director of SiCED, the German joint venture between Siemens and Infineon. In his plenary talk he asked whether SiC was a suitable technology for investment. According to Friedrichs, the first commercially available SiC products suffered from signs of overheating, caused by circuit developers using the devices beyond their rated values.
This problem has been addressed by reducing the internal electric fields within devices, and the latest products now deliver a stability comparable with that of their commercial silicon counterparts. Friedrichs believes that this advance, allied to a doubling in the cost of silicon over the last two years driven by increased demand from the photovoltaics industry, has changed the commercial landscape for SiC. "Should we invest in SiC?" is probably not the pertinent question any more, argued Friedrichs, but rather "Is now the right time to invest?"
In a session that focused on industrial applications, Roger Bassett from energy consultant Areva outlined the benefits of using SiC devices in high-voltage functions, such as power transmission and distribution. SiC offers low-loss switching of voltages in excess of 10 kV and this provides greater flexibility in the generation and distribution of electrical power. These attributes make the existing power network a major target for SiC devices – it could generate $200 million in product sales by 2020. SiC devices can offer an additional benefit over silicon, says Bassett, because they can reduce the area of land needed for construction of large-scale power conversion facilities.
The use of SiC devices in circuit applications was discussed in other talks by Ty McNutt from Northrop Grumman and Leon Tolbert from the University of Tennessee. Both of these speakers said that the promised reduction in on-state losses in circuit applications can only be delivered by making dramatic changes to packaging technology that aids heat dissipation. The two research groups have already demonstrated inverter circuits featuring SiC switches and diodes.
At previous conferences the capabilities of SiC devices at high temperatures has been a strong theme. This year s meeting was no exception. NASA s Phil Neudeck revealed that SiC FETs can run at 500 °C for more than 2400 h, which means they can be deployed into environments where traditional semiconductor technology cannot function. These could include engine exhausts, the surface of Venus and aerospace applications such as aircraft. NASA researchers have also studied behavior of this type of device in an amplifier circuit operating at 500 °C.
For several years SiC has also been touted as a suitable technology for power devices operating in high radiation environments, due to the material s wide bandgap. At this year s ECSCRM Infineon Technologies Gerald Soelkner presented the first analysis of device failure from cosmic radiation. His work demonstrates that SiC devices can deliver greater reliability than their silicon equivalents when the maximum electric field under blocking conditions is reduced enough. When this occurs, the effects of impact ionization are not important, says Soelkner.
Another application for SiC devices is gas sensing in extreme environments. Our team at Newcastle University has shown, for the first time, that gas concentrations can be recorded at high temperatures by monitoring the leakage current through the sensor s capacitor. The research shows that sensor arrays can be built that determine the concentration of various gases in a mixture. The arrays are made by varying the capacitor s dielectric for each detector, which gives every element a unique and identifiable characteristic that can be used to trace specific gases.
Improving the oxide interface
SiC MOSFETs have the potential to deliver a low on-state resistance but progress has been limited by the quality of the SiC/SiO2 interface. Typical device mobilities are 40 cm2V–1s–1, which is significantly lower than the silicon equivalents due to the high density of near-interface traps near to the conduction band edge.
Einar Sveinbjornsson, from Chalmers University, Sweden, said that this trap density is related to the presence of sodium in the oxide, which causes many deep interface traps (typically 1 × 1013 cm–2) that can lead to a large hysteresis in capacitance-voltage characteristics. The density of near interface traps can be reduced by oxidation in alumina because the sodium that is present from the processing converts these defects into deep traps, which can then be reduced by annealing in hydrogen gas. With this approach MOSFETs can be produced with mobilities in excess of 100 cm2V–1s–1.
While applications took center stage this year, progress in the quality and size of SiC substrates still arouses intense interest. Cree, which acquired Intrinsic Semiconductor this summer, picked the event to launch its zero micropipe material. Cem Basceri, from Cree s Dulles site in Virginia that previously housed Intrinsic, says that the micropipes are linked to polytype inclusions, which propagate in the [11–20] direction. With careful control of the growth conditions the formation of these inclusions is suppressed and the micropipe density can be reduced to zero, says Basceri. The company is now transferring its technology to a 100 mm platform. Recent efforts have led to 100 mm conducting substrates with just seven micropipes and high-purity semi-insulating material of the same size with micropipe densities of 2.5 cm–2.
Substrate quality is also improving at SiCrystal, says Thomas Straubinger, the firm s leader in the research and development of crystal growth. The German outfit has observed that substrate quality is dependent on the initial seed and has improved its production process by switching to a new generation of 3 inch seed crystals. Straubinger revealed that these efforts have concentrated on reducing the substrate s basal plane dislocation density because this type of defect has the greatest impact on power devices. The basal plane dislocations are linked to stress generated during the growth process and they can be substantially reduced by minimizing temperature gradients.
II-VI, a US manufacturer of SiC substrates, also reported improvements in material quality. Andy Souzis, the technology and program manager for the company s wide bandgap group, claimed that II-VI has cut the density of all types of dislocation in its 3 inch material by two orders of magnitude over the last 12 months. Typical values for comet counts are now 31 and total dislocation densities average 7 × 103 cm–2. It is a safe bet to assume that further improvements in SiC material will be reported at the next International Conference on Silicon Carbide and Related Materials, which will be held in Nara, Japan, in October 2007. However, as production of this material matures and high-quality substrates become the norm rather than the exception, the increasing focus on applications is surely set to continue.