Number of SiC patents don't tally with revenues
Although Japan holds 72% of patents related to silicon carbide wafer growth, the US holds a larger proportion in terms of revenue
Despite a cumulative raw wafer and epi wafer market that won’t exceed $80 million in 2012, the corpus of related patents of SiC growth comprises over 1772 patent families and more than 350 companies since 1928. Eighty-three percent of patents relate to a method while 17% of them claim an apparatus.
This is according to Yole Développement'slatest report, "SiC Patent Analysis single crystal, wafer and epiwafer manufacturing".
Since 1978, the main technique used to grow bulk single-crystals of SiC is PVT (Physical Vapour Transport ). The seeded sublimation method represents 36% of published patents. The PVT technique mostly deals with the hexagonal polytype nH-SiC (n=2,4,6).
An alternative route to growing SiC is Liquid Phase Epitaxy (LPE), with early efforts dating back to 1961. This technique enables the grow of crystals with low dislocation densities at relatively low temperatures , which is particularly attractive for cubic polytype 3C-SiC.
About 37% of patents relate to a Chemical Vapour Deposition technique (CVD) which is almost exclusively used today to manufacture SiC epiwafers. Molecular Beam Epitaxy (MBE) is only mentioned in 1% of patents. The polytype (hexagonal or cubic) is explicitly claimed in 15% of patents. Numerous strategies to reduce crystal defects (micropipes, carrots …) and make semi-insulating material are proposed in 23% and 10% of patents respectively.
Roughly 350 applicants are involved in SiC crystal/epiwafer technology. They are mainly located in Japan, which holds 72% of patents and the US, which possesses 12% of patents. The five major applicants based on their patents number are Denso, Sumitomo, Nippon Steel, Bridgestone and Toyota. They represent about 35% of studied patents.
US firm Cree occupies 6th position. This balance is totally uncorrelated from the reality of the market where 75% of the SiC wafer business is generated by US-based companies, namely Cree, II-VI or Dow Corning.
Japan is only responsible for 5% of the revenues (at least before Rohm acquired SiCrystal). Similar observations are seen in Europe and Asia (not including Japan) where the number of patents/revenues ratio is very weak at the moment.
Although Japan is currently leading the IP in SiC technology, Korea and China are catching up.
Japan has become increasingly involved in SiC technology since the 1980’s. The United States was the earliest player and is still active. In contrast, only 3 Japanese companies are commercially active in SiC material:. These are Showa Denko (epiwafer), Bridgestone (wafer) and Nippon Steel (wafer and epiwafer).
During the last five years, with the establishment of some new companies, China and Korea have emerged as new players. Chinese firms include Epiworld, TianYue, TYSTC and Tankeblue. Korean firm SKC is also another contender. However, the market share held by these companies is currently very low.
It may seem obvious that IP considerations do not create a differentiating factor for success in the SiC substrate business.
Cree is leading this industry with about half the global market share, and has an excellent reputation in terms of quality, diameter and reproducibility. However, Cree does not own the widest patent portfolio. Thus, know-how and patent numbers do not seem to correlate.
The only field where number of patents and business size appears to be more balanced is in Semi-Insulating (S.I.) SiC technology where both Cree (Vanadium-free) and II-VI (Vanadium-doped) have extensively patented their respective developments.
The barriers to the entry in the SiC substrate world are very high. Today, state-of-the-art technology deals with 6” diameter wafers with very low dislocation densities. According to Yole, only Cree seems able to offer such a product today.
But why is this the case?
Firstly, Cree has been widely funded by DoD, DoE, DARPA and Navy contracts during the last 20 years, meaning the US firm has had a large advantage over its competitors. Cree has been able to invest a lot of time and money in R&D and improve the technology for both LED and Power Electronics.
So mastering SiC growth is a combination of money and development time, that cannot be compressed. Also, cross-fertilisation between its LED and Power businesses have allowed Cree to benefit from LED mass manufacturing, which is probably less stringent than power at wafer level, to fuel the power electronics side.
Apart from receiving funding to develop the technology, the only options to enter quickly in the SiC substrate battlefield appears to be through M&A (Merger & Acquisition) of an existing activity or to buy a license and related know-how, paying royalties in return.
But who is for sale?
Virtually nobody is at this current time. Beyond the top five SiC substrate leaders, Yole doesn’t see a clear positioning of companies who may want to participate in a sale or merger of their business.
Ultimately, new developments based on LPE (Liquid Phase Epitaxy), made by Toyota, Denso or Sumitomo, as well as 3C-SiC (Cubic) which may disrupt the current PVT domination, could be the deciding factor.