Let's be frank, SiC devices business is not yet the most exiting money-maker activity. It is not explicitly published neither by Cree nor Infineon, but we estimate the SiC-based power electronic devices 2006 sales should have generated something around $15 million revenues (Cree showed $17M revenues on both SiC power devices AND wide-bandgap RF devices. Infineon does not publish any detail). The only product commercially available is Schottky diode (SBD), now reaching 1200V and 20A range. This component is targeting numerous possible applications (Fig.1) but is mostly used in high-end PFC (Power Factor Corrector) systems where it brings some impressive added values like better power conversion efficiency (Cree proved +50% improvement on losses), circuit board size shrinking, RF oscillations avoidance and removal of numerous passive discretes.

Theoretically, SiC can be considered as the dream technology to replace silicon-based circuits in the PFC systems. Moreover, we are talking here about a total accessible market of more than 1.3 billion PFC units in 2006 facing a +12% CAGR (Source: Darnell Group). The TOP 3 PFC manufacturers (Delta Electronics, Astec Power / Emerson and Lite-on Technology) are generating more than $2.8B revenues. So, why that tiny $15M sales for SiC devices? First reason is certainly linked to the technology implementation: It is not a chip-to-chip replacement. PFC circuits have to be redesigned to handle and catch the added value of the SiC Schottky diodes. Secondly, from an economical point-of-view, PFC market is heavily price-pressurized and introduction of a new technology is a possible risk that some companies are hesitating to take. Finally SiC Schottky diode market price itself is a limiting factor. SiC Schottky diode (SBD) cost breakdown is significant to illustrate this last point (Fig.2). To summarise, material bill accounts for ~75% of the device cost (based on current 3" wafers production), which is fundamentally different from the silicon-based technology where material is no more than 10%. The introduction of 4" material should help to make this ratio decreasing. 4" production has already been launched at Cree and Infineon may complete the 3"to 4" transition by Q3 2007.

The current market price of SiC Schottky diode is oscillating somewhere between $0.30/Amp and $0.40/Amp depending of the voltage and the lot size. According to both SiC device makers and PFC manufacturers, it seems that a price level of $0.20/Amp would give a favourable feedback for the technology penetration. First estimations of 4"-based SiC manufacturing cost lets us think that this threshold should be beat by 2008. This goes with production yield improvement, adapted SiC wafers price level ($0.15/mm² would be highly appreciated…) and fabs capacity increase.

Another parameter could boost the market: the extension of the competition landscape and the involvement of new players. The today silicon rectifiers market is leaded by companies like Vishay, Shindengen Electric, STMicroelectronics or MicroSemi. These companies are already supplying PFC manufacturers in large volumes and are benefiting from an established commercial network towards the end users. Their strategy over SiC is the following:

• Vishay has acquired the SiC technology developed since 1993 by DaimlerChrysler AG in 2000. In November 2006, Vishay has bough International Rectifiers Power Control Systems (PCS) division in which SiC developments have been achieved.
• Shindengen Electric is active in the SiC technology and has published numerous papers on it.
• STMicroelectronics has developed SiC Schottky diodes since a long time in the Tours (France) R&D center. The Italian manufacturing plant in Catania may be used for devices production
• MicroSemi has acquired APT (Advanced Power Technology) in 2006 and is now having a production activity in Oregon (US).

Diodes and switches are living quite well together and even if hybrid approach SiC-Silicon is an alternative, full SiC electronics is highly requested for numerous applications. To highlight this, hybrid electrical vehicle (HEV) is today using silicon-based IGBT and diodes in the inverter module to power the electric motor (30-50kW). This silicon chipset has to be cooled-down by a water based system to maintain device junction temperature around 85°C. However, it has to use a dedicated water-cooling system, different from the one in charge of the fuel engine that can handle higher temperatures. One of the objectives of HEV car makers is to simplify this setup by implementing SiC-based electronics that can easily wistand 150°C and more. It will then allow using a single water-cooling system instead of 2. About 15% money saving on the power module could be so achieved. That approach only goes with a full-SiC electronics chipset and wont be realistic with an hybrid solution. In terms of requirements for this particular application, 1200V/100A SiC single chips would perfectly fit… Up to now, main R&D efforts have been oriented to release the SiC MOSFET. Lots of announcements have been made (Rohm, Cree, Acreo, Toshiba…) but as a conclusion, the oxide reliability and poor electronic mobility issues remain partially unsolved.

As showed in Table 1, even if MOSFET remains the most studied device, alternatives exist and J-FET or BJT are also under the scope of prestigious R&D groups. The common characteristic of JFET and BJT is that it does not use any oxide as MOSFET does. Recent results from SemiSouth, TranSiC or GeneSiC are very encouraging: TranSiC is now fine tuning a 1200 V 6 A BJT chip with operation temperature up to 250°C, SemiSouth is focusing on a Noff version of SiC JFET and GeneSiC is concentrating on very high power SiC devices including MOSFET and JFET. It has also to be noticed that Infineon has announced its intention to produce a SiC JFET in large volume in a very near future. Early 2006, Rohm has published results concerning a SiC MOSFET rated at 900V of polarization and a Ron of 3.1 m?/cm².

It is now clear that the relative stagnation of SiC power device market is partially due to a lake of a reliable transistor technology. PFC business is the only one driving SiC device sales and perspectives for higher market penetration is mainly linked to the device cost decreasing. Others applications are requesting a complete SiC switching cell (diode and transistor). MOSFET has been largely investigated by major SiC R&D teams, but it seems more and more certain that JFET or BJT technologies may be released first on the market place. Big names are confessing to face unsolved issues with MOSFET so maybe truth is out there, out from oxides….

Yole Développement (www.yole.fr) is a market research and strategy consulting company based in Lyon, France. Dr. Philippe ROUSSEL is headed the Compound Semiconductors division since 1998. Yole is editing numerous market reports and is currently releasing their analysis in SiC, GaN, AlN, power and RF devices or HB-LED markets.