Technical Insight
SiC gathers momentum
With tire manufacturer Bridgestone joining the flock of companies developing SiC devices, what technological developments can we expect to see this year? Jon Newey reports.
Over the last 12 months, something of a minor gold rush seems to have taken place in the SiC world. The number of commercially available devices has expanded from the Schottky diodes offered by Cree and Infineon to include MESFETs from Rockwell Scientific and Cree. Rohm also announced that it was entering SiC Schottky diode production.
A number of other companies, including Bandgap Technologies and SemiSouth in the US, and Bridgestone in Japan, which had previously been quietly working on SiC substrates and devices, have recently announced funding, partnerships and product developments. The whole SiC community was also spurred on by the biannual International Conference on Silicon Carbide and Related Materials (ICSCRM) in October, which neatly crystallized the community s thoughts on recent progress (Compound Semiconductor December 2003).
Much has been made of the great strides taken to eliminate micropipe defects from SiC wafers. Bulk crystal growers are now setting their sights on other types of defects and maintaining high quality at larger boule diameters. Cree offers 75 mm wafers and others will follow, including Bandgap Technologies, which is using its micropipe-blocking growth technology to produce low-defect-density material. Bandgap intends to release a 75 mm product in 2004 and has demonstrated the scalability of its growth method beyond 75 mm. "We re excited because as well as blocking the micropipes, we re getting a much better substrate overall," said Jack Bonnette, Bandgap s head of business development.
The commercial drivers to produce substrates larger than 75 mm are not pressing at the moment, although Department of Defense programs are pushing companies along this route. Volumes of electronic devices are not currently large enough to warrant investment in 4 inch fabrication lines, and LED chips are relatively small, making the economies of shifting from 50 mm substrates less obvious.
At ICSCRM Cree revealed research results on a zero voltage drift bipolar device, suggesting that it had gained the ability to control basal plane dislocations, for so long the bane of those trying to fabricate reliable SiC bipolars. Does this mean that bipolar SiC devices will appear in product catalogs in 2004? "The zero forward voltage drift is an early R&D result that we need to optimize and make a repeatable process," said Cree s John Palmour. "[While] I don t foresee us, or anybody else, bringing out a bipolar device in the next year, I do anticipate that there will be a lot of progress towards that goal."
It s a similar situation for SiC MOSFETs. These are needed to replace silicon IGBTs and MOSFETs as low-power-loss switches for high-power applications such as motor drive circuits. SiC Schottky diodes have already made their presence felt as replacements for silicon devices. Not so long ago the problems of low channel mobility and gate oxide reliability seemed very daunting, but now there seems to be light at the end of the tunnel. "I don t anticipate product releases by anyone this year, but it s not that far off," said Palmour. "There s been a lot of progress, but then we had so far to go that it was hard to see when we d get to the end goal. However, I m very encouraged by progress [demonstrated] in recent results."
While the commercialization of SiC MOSFETs awaits more progress in the lab, their close cousins - SiC JFETs - appear to be closer to a market introduction. SiC epiwafer and device developer SemiSouth of Starkville, MS, plans to make a vertical JFET power switch available in late 2004. The first devices will be rated at 600 V and 20 A, with 1500 V devices on the roadmap for 2005. The company s VJFET is targeted at applications where switching at high temperatures in harsh environments, for example those experienced by equipment in oil wells, is challenging and where oxide reliability is a big issue.
Only high quality will doThe SiC marketplace is a curious one, with Cree accounting for about 85% of the market, and a growing number of small companies scrambling to relieve the leader of some of its share. One might expect that the situation for these smaller companies in a relatively small market, albeit one destined for great things, is rather precarious. At present many of them are largely surviving on US Department of Defense money, intended to drive the technology forward while building a viable SiC industry. Some of them aim to compete directly, using alternative approaches to developing high-quality materials and devices.
Other companies are forming alliances to pool their expertise in different areas. SemiSouth s tie-up with SiC substrate supplier II-VI Inc is a good example. SemiSouth is developing epitaxial processes on II-VI s substrates, allowing feedback between the two on material quality speeding up product development. "Our alliance with II-VI gives us the ability to offer a broader range of products, and provides a good second source of material," said SemiSouth s president Jeff Casady.
Cree has the advantage of having a large internal demand for substrates from its electronics program and burgeoning LED business. "It can be difficult for anyone else to get far enough up the learning curve to compete with that sort of volume," said Palmour. "However, it s always good to have competition and at some point I ve no doubt that a real second source will emerge."
Although price is crucial, at the moment the SiC market is one driven by material quality and the device results that follow. "Whoever can get rid of the defects and still grow at larger diameters will be a winner in the market," said Bonnette. "The players with the best material will be the ones that survive."
A number of other companies, including Bandgap Technologies and SemiSouth in the US, and Bridgestone in Japan, which had previously been quietly working on SiC substrates and devices, have recently announced funding, partnerships and product developments. The whole SiC community was also spurred on by the biannual International Conference on Silicon Carbide and Related Materials (ICSCRM) in October, which neatly crystallized the community s thoughts on recent progress (Compound Semiconductor December 2003).
Much has been made of the great strides taken to eliminate micropipe defects from SiC wafers. Bulk crystal growers are now setting their sights on other types of defects and maintaining high quality at larger boule diameters. Cree offers 75 mm wafers and others will follow, including Bandgap Technologies, which is using its micropipe-blocking growth technology to produce low-defect-density material. Bandgap intends to release a 75 mm product in 2004 and has demonstrated the scalability of its growth method beyond 75 mm. "We re excited because as well as blocking the micropipes, we re getting a much better substrate overall," said Jack Bonnette, Bandgap s head of business development.
The commercial drivers to produce substrates larger than 75 mm are not pressing at the moment, although Department of Defense programs are pushing companies along this route. Volumes of electronic devices are not currently large enough to warrant investment in 4 inch fabrication lines, and LED chips are relatively small, making the economies of shifting from 50 mm substrates less obvious.
At ICSCRM Cree revealed research results on a zero voltage drift bipolar device, suggesting that it had gained the ability to control basal plane dislocations, for so long the bane of those trying to fabricate reliable SiC bipolars. Does this mean that bipolar SiC devices will appear in product catalogs in 2004? "The zero forward voltage drift is an early R&D result that we need to optimize and make a repeatable process," said Cree s John Palmour. "[While] I don t foresee us, or anybody else, bringing out a bipolar device in the next year, I do anticipate that there will be a lot of progress towards that goal."
It s a similar situation for SiC MOSFETs. These are needed to replace silicon IGBTs and MOSFETs as low-power-loss switches for high-power applications such as motor drive circuits. SiC Schottky diodes have already made their presence felt as replacements for silicon devices. Not so long ago the problems of low channel mobility and gate oxide reliability seemed very daunting, but now there seems to be light at the end of the tunnel. "I don t anticipate product releases by anyone this year, but it s not that far off," said Palmour. "There s been a lot of progress, but then we had so far to go that it was hard to see when we d get to the end goal. However, I m very encouraged by progress [demonstrated] in recent results."
While the commercialization of SiC MOSFETs awaits more progress in the lab, their close cousins - SiC JFETs - appear to be closer to a market introduction. SiC epiwafer and device developer SemiSouth of Starkville, MS, plans to make a vertical JFET power switch available in late 2004. The first devices will be rated at 600 V and 20 A, with 1500 V devices on the roadmap for 2005. The company s VJFET is targeted at applications where switching at high temperatures in harsh environments, for example those experienced by equipment in oil wells, is challenging and where oxide reliability is a big issue.
Only high quality will doThe SiC marketplace is a curious one, with Cree accounting for about 85% of the market, and a growing number of small companies scrambling to relieve the leader of some of its share. One might expect that the situation for these smaller companies in a relatively small market, albeit one destined for great things, is rather precarious. At present many of them are largely surviving on US Department of Defense money, intended to drive the technology forward while building a viable SiC industry. Some of them aim to compete directly, using alternative approaches to developing high-quality materials and devices.
Other companies are forming alliances to pool their expertise in different areas. SemiSouth s tie-up with SiC substrate supplier II-VI Inc is a good example. SemiSouth is developing epitaxial processes on II-VI s substrates, allowing feedback between the two on material quality speeding up product development. "Our alliance with II-VI gives us the ability to offer a broader range of products, and provides a good second source of material," said SemiSouth s president Jeff Casady.
Cree has the advantage of having a large internal demand for substrates from its electronics program and burgeoning LED business. "It can be difficult for anyone else to get far enough up the learning curve to compete with that sort of volume," said Palmour. "However, it s always good to have competition and at some point I ve no doubt that a real second source will emerge."
Although price is crucial, at the moment the SiC market is one driven by material quality and the device results that follow. "Whoever can get rid of the defects and still grow at larger diameters will be a winner in the market," said Bonnette. "The players with the best material will be the ones that survive."
