Focus on SiC power electronics business 2020

SiC is currently implemented in several power systems and is gaining momentum and credibility.

Yole Developpement stays convinced that the most pertinent market for SiC lands in high and very high voltage (more than 1.2kV), where applications are less cost-driven and where few incumbent technologies can’t compete in performance. This transition is on its way as several device/module makers have already planned such products at short term.

Thus, even though EV/HEV skips SiC, industry could expand among other apps. Now, the only question remains: Is there enough business to make so many contenders live decently? Probably, yes, as green-techs are also expanding fast, strongly requesting SiC. But, any newcomers should carefully manage strategy and properly size capex according to the market size.

Power electronics industry outlook
Electronics systems were worth $122 billion in 2012, and will likely grow to $144 billion by 2020 at a CAGR of 1.9 percent. Power inverters will grow from $41 billion in 2012 to over $70 billion by 2020 at a CAGR of 7.2 percent. Semiconductor power devices (discretes and modules) will grow from $12.5 billion in 2012 to $21.9 billion by 2020 at a CAGR of 7.9 percent. Power wafers will grow $912 million in 2012 to $1.3 billion by 2020 at a CAGR of 5.6 percent.

Looking at the power electronics market in 2012 by application and the main expectations to 2015, computer and office will account for 25 percent, industry and energy 24 percent, consumer electronics 18 percent, automotive and transport 17 percent, telecom 7 percent and others 9 percent.

The main trends expected for 2013-2015 are:
* Significant increase of automotive sector following EV and HEV ramp-up.
* Renewable energies and smart-grid implementation will drive industry sector ramp-up.
* Steady erosion of consumer segment due to pressure on price (however, volumes (units) will keep on increase).

The 2011 power devices sales by region reveals that overall, Asia is still the landing-field for more than 65 percent of power products. Most of the integrators are located in China, Japan or Korea. Europe is very dynamic as well with top players in traction, grid, PV inverter, motor control, etc. Asia leads with 39 percent, followed by Japan with 27 percent, Europe with 21 percent and North America with 13 percent.

The 2011 revenues by company/headquarter locations reveals that the big-names of the power electronics industry are historically from Japan. Nine companies of the top-20 are Japanese. There are very few power manufacturers in Asia except in Japan. Europe and US are sharing four of the top five companies. Japan leads with 42 percent, followed by Europe and North America with 28 percent each, respectively, and Asia with 2 percent.

Looking at the TAM comparison for SiC (and GaN), very high voltage, high voltage of 2kV and medium voltage of 1.2kV appear as a more comfortable area for SiC. The apps are less cost-driven and SiC added value is obvious. Low voltage from 0-900V is providing strong competition with traditional silicon technologies, SJ MOSFET and GaN. There are cost-driven apps.



SiC industry outlook
The new entrants in materials include Epiworld (China), which is proposing epi services for 3”, 4” and 6”. They are equipped with Aixtron G4 6x6” and several pieces of inspection equipment. SiCC (China), which is proposing 2, 3 and 4” n-type and SI SiC epi-ready wafers. It was established late 2010 as a spin-off of Shandong University. A third player is TYSTC (China). Tianyu Semiconductor Technology Co. Ltd was founded in Jan 2009, and produces and markets SiC epi-wafers. It has three epi-reactors running (Aixtron).

Among devices, there is Kingway Technology Co. Ltd (China), a high-tech enterprise located in Beijing, China, founded in 2010 by a group of veterans with Ph.D. Degrees and solid industrial experiences back from the US. Kingway focuses on epitaxial growth of SiC and GaN materials, design/fabrication of high power, high voltage devices and modules. Then, there is Anvil Semiconductors Ltd, (UK), a developer of small power converters using SiC power semiconductor switches. Widetronix Inc. (US) is a manufacturer of high-voltage SiC epitaxial wafers and power conversion devices. The company is located in Ithaca, New York.

Still on devices, Ascatron (SW), is a spin-off from ACREO, and is dedicated to fabrication of SiC semiconductors, from epi to diced wafers. Next, there is IBS (FR), positioned as an SiC foundry service provider. Finally, Fraunhofer IISB has been around since 2012, IISB proposes a foundry service for SiC devices, from epi to packaged chips.

The exits in material include NeoSemiTech (Korea), which has stopped its SiC activities, focusing on GaAS and Si for solar. Another firm, Caracal (US) went bankrupt. Among devices, there is SemiSouth who closed down late 2012.

As for M&A, Crysband (Kr) has been acquired by SKC (Kr), SiCed (D) that is now 100 percent owned by Infineon, and Fairchild acquired TranSiC (April 2011) for $17 million.

Moving from Si to SiC
How does SiC move the fundamental economics of a PV inverter? When moving from Si to SiC, three parameters dramatically change the fundamental economics of a PV inverter. One, SiC devices (chips) are (and will probably remain) more expensive than Silicon ones. The Bill-of-Material of the power module core will mechanically increase when using SiC.

SiC will allow gaining +1.3 points on the European efficiency with a maximum that could reach 96.7 percent (compared to 95.4 percent with silicon). Money will be saved when selling the electricity production at local feed-in tariff. Finally, SiC will allow running the inverter at a higher switching frequency (typically 32kHz, instead of 12kHz). Such a high frequency will dramatically decrease the size and so the cost of the surrounding passive devices (inductors and capacitors).

Taking into account the uncertainty of the implementation of SiC devices in EV/HEV segment, there are two scenarios with different market dynamics. In the nominal scenario, the implementation of SiC devices will start from 2015 followed by a ramp-up of SiC adoption that will take up to 11 percent shares over Si IGBT by 2020.

In a pessimistic scenario, the SiC will take off by 2016 with only a slow ramp-up in production to reach only 2.5 percent of usage by 2020. Another “worst scenario” could have been defined where SiC is totally excluded from EV/HEV. Yole stays confident that SiC will have a role to play in EV/HEV before 2020.