Technical Insight
SiC: survival of the fittest?
Following the demise of SemiSouth, what does the future hold for other SiC power transistor manufacturers, asks Compound Semiconductor.
In March, it announced a $18m capacity expansion; in May, it sampled the industry's first 650V SiC JFET power transistors. But in October, the US-based developer of SiC high power transistors, SemiSouth, closed.
To say the sudden shutting of the Mississippi University spin-off was a surprise and a disappointment to many in the industry would be an understatement.
Local newspapers reported the ninety job losses and key investor Power Integrations wrote off its 2010 investment. Germany-based news website, elektroniknet.de, has aired the possibility of a buyout, but chief-technology officer, Jeffrey Casady's move to SiC semiconductor heavyweight, Cree, doesn't raise many hopes.
So what will be the impact on the SiC device industry? As Philippe Roussel, analyst at France-based Yole Développement puts it: “This is bad news for the industry. SemiSouth was promising a lot, they were a key competitor to Cree, and finally they fail. Psychologically speaking, this is not a good sign.”
Roussel suggests a key reason for SemiSouth's demise was its size. With a headcount of at least 100, manpower-wise the company was significantly larger than its counterparts; for example, United Carbide and GeneSiC, both in the US, hire no more than two dozen employees.
And while many express shock, industry re-shaping has been rife with the last two years peppered with mergers and acquisitions. For example, in 2011, US semiconductor giant Fairchild bought Sweden-based TranSiC, with Germany-based semiconductor heavyweight Infineon buying SiCED, Germany, soon afterwards.
Despite industry consolidation, myriad device makers remain, jostling for market-share in an industry that arguably has yet to truly take off. While manufacturers of PV inverters have been snapping up SiC devices at an unexpectedly high rate the electric vehicle industry has yet to do the same.
The jury is still out on whether automotive-related businesses will opt for SiC transistors or skip the technology and adopt GaN or silicon variants instead. And in the meantime, Roussel expects more industry casualties.
“I have a suspicion that other companies may close down as well. A big question mark still exists whether SiC will be implemented in the electric vehicle and hybrid electric vehicle market,” he says. “Until this happens, the market size will be limited and is there enough space here for ten to fifteen companies? For the big names such as Infineon and Cree I have no doubt they will survive, for smaller companies, I just don't know.”
.jpg)
Will the likes of Toyota and other electric automotive-related manufacturers choose SiC transistors? Device manufacturers can only wait and see. Photo credit: Mariordo (Mario Roberto Durán Ortiz)
Industry winners
One factor set to influence which industry player will survive is its flavour of transistor. To date, businesses are focusing on MOSFETs, JFETs, BJTs and IGBTs.
Without a doubt, Cree leads the MOSFET pack with other developers including Mitsubishi Electric, Fuji Electric, Rohm, Toshiba, ST Microelectronics and Shindengen, Japan. Meanwhile, Shindengen and Fairchild are also working on BJTs while United SiC, for one, forges ahead with the IGBT.
SemiSouth was focussing on SiC JFETs – is this where it went wrong? The business frequently touted easier and cheaper manufacturing as a key advantage of this transistor but on the flip-side its vertical structures required non-standard driver schemes. In contrast MOSFETs are discrete devices offering chip-to-chip replacement with an IGBT or silicon MOSFET.
“JFETs are normally-on devices, and as far as we know, this is not welcomed in the industry,” says Roussel. “SemiSouth designed a normally-off device that was going through qualification. This is a guess, but maybe the product didn't reach expectations, especially compared to the MOSFET. Its JFET was very unique.”
Indeed, Infineon, also working on JFETs, claims its horizontal architecture uses an easier drive scheme. Roussel concurs, highlighting how the company's JFET uses a Si MOSFET to make it a normally-off device.
“We now have two MOSFET providers - Cree and Rohm - delivering off-the-shelf products,” he adds. “This is the holy grail for system integrators... so maybe the industry asked the question, why should I choose your JFET?”
Earlier this month, Cree unveiled its first commercially available SiC power module targeting solar inverters as well as industrial motor drives, high power converters and uninterruptible power supplies. Life in the SiC device industry goes on.
But as Roussel concludes: “The market will live without SemiSouth, but will the industry become cautious and think, 'this company was so wonderful so why did it fail?' We now have to convince manufacturers that this SiC IC technology is robust and can deliver.”
To say the sudden shutting of the Mississippi University spin-off was a surprise and a disappointment to many in the industry would be an understatement.
Local newspapers reported the ninety job losses and key investor Power Integrations wrote off its 2010 investment. Germany-based news website, elektroniknet.de, has aired the possibility of a buyout, but chief-technology officer, Jeffrey Casady's move to SiC semiconductor heavyweight, Cree, doesn't raise many hopes.
So what will be the impact on the SiC device industry? As Philippe Roussel, analyst at France-based Yole Développement puts it: “This is bad news for the industry. SemiSouth was promising a lot, they were a key competitor to Cree, and finally they fail. Psychologically speaking, this is not a good sign.”
Roussel suggests a key reason for SemiSouth's demise was its size. With a headcount of at least 100, manpower-wise the company was significantly larger than its counterparts; for example, United Carbide and GeneSiC, both in the US, hire no more than two dozen employees.
And while many express shock, industry re-shaping has been rife with the last two years peppered with mergers and acquisitions. For example, in 2011, US semiconductor giant Fairchild bought Sweden-based TranSiC, with Germany-based semiconductor heavyweight Infineon buying SiCED, Germany, soon afterwards.
Despite industry consolidation, myriad device makers remain, jostling for market-share in an industry that arguably has yet to truly take off. While manufacturers of PV inverters have been snapping up SiC devices at an unexpectedly high rate the electric vehicle industry has yet to do the same.
The jury is still out on whether automotive-related businesses will opt for SiC transistors or skip the technology and adopt GaN or silicon variants instead. And in the meantime, Roussel expects more industry casualties.
“I have a suspicion that other companies may close down as well. A big question mark still exists whether SiC will be implemented in the electric vehicle and hybrid electric vehicle market,” he says. “Until this happens, the market size will be limited and is there enough space here for ten to fifteen companies? For the big names such as Infineon and Cree I have no doubt they will survive, for smaller companies, I just don't know.”
Will the likes of Toyota and other electric automotive-related manufacturers choose SiC transistors? Device manufacturers can only wait and see. Photo credit: Mariordo (Mario Roberto Durán Ortiz)
Industry winners
One factor set to influence which industry player will survive is its flavour of transistor. To date, businesses are focusing on MOSFETs, JFETs, BJTs and IGBTs.
Without a doubt, Cree leads the MOSFET pack with other developers including Mitsubishi Electric, Fuji Electric, Rohm, Toshiba, ST Microelectronics and Shindengen, Japan. Meanwhile, Shindengen and Fairchild are also working on BJTs while United SiC, for one, forges ahead with the IGBT.
SemiSouth was focussing on SiC JFETs – is this where it went wrong? The business frequently touted easier and cheaper manufacturing as a key advantage of this transistor but on the flip-side its vertical structures required non-standard driver schemes. In contrast MOSFETs are discrete devices offering chip-to-chip replacement with an IGBT or silicon MOSFET.
“JFETs are normally-on devices, and as far as we know, this is not welcomed in the industry,” says Roussel. “SemiSouth designed a normally-off device that was going through qualification. This is a guess, but maybe the product didn't reach expectations, especially compared to the MOSFET. Its JFET was very unique.”
Indeed, Infineon, also working on JFETs, claims its horizontal architecture uses an easier drive scheme. Roussel concurs, highlighting how the company's JFET uses a Si MOSFET to make it a normally-off device.
“We now have two MOSFET providers - Cree and Rohm - delivering off-the-shelf products,” he adds. “This is the holy grail for system integrators... so maybe the industry asked the question, why should I choose your JFET?”
Earlier this month, Cree unveiled its first commercially available SiC power module targeting solar inverters as well as industrial motor drives, high power converters and uninterruptible power supplies. Life in the SiC device industry goes on.
But as Roussel concludes: “The market will live without SemiSouth, but will the industry become cautious and think, 'this company was so wonderful so why did it fail?' We now have to convince manufacturers that this SiC IC technology is robust and can deliver.”
