| From: slacker711 | 6/30/2021 10:19:27 AM | | | | | | Compound semiconductors now national strategic materials, says GlobalWafers chair
Julian Ho, Taipei; Willis Ke, DIGITIMES
Wednesday 30 June 2021
digitimes.com
Compound semiconductors will become a key industry sector reaching a national strategic level in importance, as they will be badly needed to power many critical technological applications including low-orbit satellites, 5G communications, green energy, smart power grids, EVs and high-voltage components, according to Doris Hsu, chairperson of silicon wafer specialist GlobalWafers.
Hsu made the remarks when opening an online class on compound semiconductor materials and components, in her capacity as chief of the Taiwan Compound Semiconductor and Equipment Industry-Academia Consortium under the Photonics Industry & Technology Development Association (PIDA).
Hsu said many countries including Taiwan now enforce strict export controls on compound semiconductors, requiring special permits for outbound shipments of such strategic materials including even semi-insulated silicon carbide (SiC). She stressed that Taiwan must develop its own technology prowess in the segment.
Hsu continued the compound semiconductor industry chain involves quite complicated and difficult processes, ranging from crystal growth, components manufacturing, and measuring to rollout of finished products, as well as supply of related equipment. Accordingly, her consortium has invited specialists from diverse fields seeking to build a complete compound semiconductor ecosystem covering both vertical and horizontal segments such as design, IP, raw materials and equipment.
Taiwan compound semiconductors makers are facing keen competition from international players, given that US-based Cree and II-VI now dominate the supply of related substrates and there are more than 1,000 Chinese companies engaged in the development and production of compound semiconductors, according to Hsu. |
| | Wolf speed | Stock Discussion ForumsShare | RecommendKeepReplyMark as Last Read |
|
| From: EvanG | 7/8/2021 7:28:24 PM | | | | | | STM has been using 500um thick 200mm wafers in their pilot line as opposed to 150mm wafers which are 350um thick. Does anyone know if 150mm wafers were thicker at the start of the rampup and then got thinner?
This link has the latest STM 200mm wafer quality report. Previous report said that thickness was in order to get warp and bow within spec. This report mentions both are within the final spec but doesn't mention if they will try for thinner wafers as a result.
Thicker wafers would be more lucrative for the wafer suppliers.
ec.europa.eu |
| | Wolf speed | Stock Discussion ForumsShare | RecommendKeepReplyMark as Last ReadRead Replies (2) |
|
| To: EvanG who wrote (10381) | 7/10/2021 5:14:47 PM | | From: slacker711 | | | | Fantastic find. Quite a bit of interesting information about the state of 200mm wafers. Am I correct in assuming that these wafers are from II-VI?
Overall, it appears that they are making progress on wafer quality but definitely have work to do on epi. Yields are obviously a wild card. =
Does anyone know if 150mm wafers were thicker at the start of the rampup and then got thinner?
This link has the latest STM 200mm wafer quality report. Previous report said that thickness was in order to get warp and bow within spec. This report mentions both are within the final spec but doesn't mention if they will try for thinner wafers as a result.
I started digging around and came across this oled materials catalog from ~'13. My understanding is that SiC wafers are usually produced off-axis. I couldn't find any further information about the c vs. c1 specification, but it appears that Cree was producing 500um 150mm wafers at one point.
macrogroup.ru
This article from Dow Corning talks quite a bit about the transition to 150mm SiC wafers. Ironically, since Cree is keeping the 200mm wafers in-house, this would actually hurt them until they started selling 200mm wafers.
power-mag.com
Control of the shape and flatness
Together with the reduction of extended defects, it is important to optimize the shape of the wafers via optimized crystal growth, wafering, and polishing processes. This has been specifically challenging for the 150 mm SiC substrates as the thickness of the substrates (350 µm) has been kept identical while switching from the 100 mm wafers to the 150 mm diameter. This constraint is different from what has occurred in the Silicon technology where the wafer thickness increases together with the expansion of the diameter. Maintaining thickness while increasing diameter brings additional crystal stress control requirements for 150 mm SiC.
|
| | Wolf speed | Stock Discussion ForumsShare | RecommendKeepReplyMark as Last ReadRead Replies (1) |
|
| To: slacker711 who wrote (10382) | 7/10/2021 7:14:01 PM | | From: EvanG | | | | Am I correct in assuming that these wafers are from II-VI?
Yes. Cree would need a European division to be part of Horizon 2020 project. II-VI used their German division, even though wafers come from New Jersey.
Ironically, since Cree is keeping the 200mm wafers in-house, this would actually hurt them until they started selling 200mm wafers.
Don't understand why they talk that way. Cree is probably briefing everyone on their status and everyone knows but us. Even though Cree is not a part of project it would make sense to try to qualify wafers in parallel. Still lots of time. |
| | Wolf speed | Stock Discussion ForumsShare | RecommendKeepReplyMark as Last ReadRead Replies (2) |
|
| To: EvanG who wrote (10383) | 7/11/2021 8:12:55 PM | | From: slacker711 | | | | Don't understand why they talk that way. Cree is probably briefing everyone on their status and everyone knows but us.
Sounds like OLED and blue.
Personally though, I think it is question mark whether Cree will sell 200mm wafers until competitors are closer to production (likely two years). They need to establish themselves as a first tier automotive semi supplier and the price advantage associated with 200mm may allow them to win some design slots that start shipping in '24 and beyond.
and even if they ultimately plan on selling 200mm wafers to competitors in '23, then threatening that they won't is likely a game of leverage. Not much different than STM and ON touting their internal supply or Infineon's press releases touting GTAT. |
| | Wolf speed | Stock Discussion ForumsShare | RecommendKeepReplyMark as Last ReadRead Replies (1) |
|
| To: slacker711 who wrote (10384) | 7/11/2021 9:49:58 PM | | From: EvanG | | | | if they ultimately plan on selling 200mm wafers to competitors in '23, then threatening that they won't is likely a game of leverage. Not much different than STM and ON touting their internal supply or Infineon's press releases touting GTAT.
II-VI model is very different. They will sell at all levels of supply chain. Wafer, chip, module. Regularly state their goal is to provide competitors a low enough price that they don't even try to develop their own capabilities. If Cree isn't interested they will take it.
STM will be close to pulling even with Cree in SiC revenue this year, kind of wonder if that is causing friction.
Think Infineon is using GTAT boules because they want to use their cold split technology. Not that the cold split thing makes sense to me, but they think it is a major advantage. |
| | Wolf speed | Stock Discussion ForumsShare | RecommendKeepReplyMark as Last ReadRead Replies (1) |
|
| To: EvanG who wrote (10385) | 7/11/2021 11:15:38 PM | | From: slacker711 | | | | II-VI model is very different. They will sell at all levels of supply chain. Wafer, chip, module. Regularly state their goal is to provide competitors a low enough price that they don't even try to develop their own capabilities. If Cree isn't interested they will take it.
Every single customer of Cree wants a viable second supplier. II-VI will have a fairly easy path as a merchant supplier if they can bring 200mm wafers to market in volume with decent yields.
The question for Cree is whether they can establish themselves as a first tier component/module suppliers to the auto companies. I think that a vertically integrated supplier is more likely to be a defensible long-term business model than remaining a merchant supplier of wafers. This is obviously going to cause friction with wafer customers but they will likely be forced to deal with Cree through the next 3-5 years.
If they execute, the transition to 200mm will give Cree an enormous advantage. They should use it to increase their odds of successfully becoming a first tier component supplier. |
| | Wolf speed | Stock Discussion ForumsShare | RecommendKeepReplyMark as Last ReadRead Replies (1) |
|
| To: slacker711 who wrote (10386) | 7/12/2021 12:41:02 PM | | From: EvanG | | | | This is obviously going to cause friction with wafer customers but they will likely be forced to deal with Cree through the next 3-5 years.
In a low growth environment with few players remaining I would agree but that could take decades. Expanding players will mean lower margins and lower market share. It is about surviving to be one of the few which helps if their are partners.
If they execute, the transition to 200mm will give Cree an enormous advantage.
I highly expect that not long after 200mm goes into mass production that 300mm is demonstrated. If Cree wants an advantage they need to not be outgunned on EBITDA by everyone of their competitors and that means making money, something they have consistently been unable to do. |
| | Wolf speed | Stock Discussion ForumsShare | RecommendKeepReplyMark as Last ReadRead Replies (1) |
|
| To: EvanG who wrote (10387) | 7/12/2021 2:17:30 PM | | From: slacker711 | | | | I highly expect that not long after 200mm goes into mass production that 300mm is demonstrated. If Cree wants an advantage they need to not be outgunned on EBITDA by everyone of their competitors and that means making money, something they have consistently been unable to do.
We are in limbo with Cree right now. The valuation is ludicrous based on current results and extremely rich based on the FY'24 guidance given in Nov '19. It will probably be 24 month before we can even begin to see what the results look like with the ramp of the Mohawk fab.
In other words, it is going to be a while before we can prove or disprove your skepticism.
FWIW, I doubt that they are going to fund much of their next fab. They received $500 million in subsidies before semis were considered a national priority by every region. |
| | Wolf speed | Stock Discussion ForumsShare | RecommendKeepReplyMark as Last ReadRead Replies (1) |
|
| |
