iDeal Semi details its SuperQ tech ...
Aug 10, 2023
“We introduced the concept of SuperQ and its been interesting to see how the market reacts to working in silicon rather than wide bandgap. We are very likely the only silicon power startup in the US,” Mark Granahan, CEO and founder tells eeNews Europe Power
“There’s been a tremendous amount of surprise. Its not hard not find the silicon is dead rhetoric in power electronics and everyone assumed that was the case,” said Ryan Manack, VP of marketing at iDeal. “But the barriers to entry for WBG are high with new concepts, how to use these devices, how to drive them, so being able to ride a weave of silicon power that’s well understood and the cost and manufacturability has been a nice message. A lot of the applications have been surprised that we can deliver this performance in silicon.”
The SuperQ technology uses a cocktail of dielectric materials, etching and atomic layer deposition to improve the performance of all kinds of power devices. It enables much lower specific on resistance (RSP) for a much lower on resistance, which in turn gives an improvement in switching performance as the die is smaller.
“We believe SuperQ can continue to provide the performance improvements similar to the move to superjunction devices,” said Granahan. “The fundamental thickness of the drift region sets the breakdown voltage and that thickness adds to the resistivity of the device. Today for superjunction technologies 50% of the conduction area is n type material which is a fundamental limit of 13-15V per micron,” said Granahan.
“Our technology gives a voltage blocking of 19 to 20V/um as we have fundamentally taken the p region and nearly eliminated it using an etch capability and through a dielectric material set that allows us to provide charge balancing in a very small area. The conduction area is greatly expanded so we have more area and so improving the RSP. This gives voltage blocking of 19 to 20V/um, a 30% improvement, so our epitaxy is much thinner. The larger conduction area and higher doping concentration deliver a very effective high voltage blocking technique.”
Using more modern chip making tools also makes the process simpler and cheaper for foundries to produce.
“Our technology has been adopted from CMOS with some of the films and materials, so we are working at the nanometre level in the world of power,” said Granahan. “Rather than epi implant with over 18 masks and long process times, or trench and refill with 14 masks, our etch and ALD deposition has less than 10 masks so our capital cost is low and the process is shorter.”
“We have done a lot of work on reliability testing to prove out the technology ae go through all the well known tests for a silicon power device and at the core of it, the ability of the film to hold the charge is very important, so we have done reliability on the film itself and then on the device that shows the breakdown voltage and leakage current are stable over time.
“Our MOSFET structure is a very simple, elegant one. The mask count is about 10 to 11 and that plays to the reliability of the device.”
“Any device that needs to conductor current and block voltage can use SuperQ so diodes and MOSFETs. As we analyse the market about 90 to 95% of the space falls below 850V so it really is a massive market,” said Manack.
The single process can be used for MOSFETs and diodes from 60V to 850V rather than having to use multiple technologies to cover a wide voltage range.
“At 60 to 200Vthere is consumer, motor drive, e-bike, and the whole server and AI space that is looking to deliver lots of power and they need 120 to 200V,” he said.
“The structure of the device is optimised for fast switching,” said Granahan. “That film likes to collect holes so in switching transition those holes are held to the surface and so the gate structure experiences both lateral and gate trench structures. It’s the best of both world s with excellent DC characteristics with excellent switching. We can easily break the 150kHz range which is the upper range of power designs.”
The company is using the process to build its own 650V to 800V MOSFETs with foundry partners.
“We are a device manufacturer so we will be selling products to customers,” said Granahan. “Polar [Semiconductor in the US] is one of our manufacturing partners and acting as a foundry for us and we have subcontracted assembly and test in Asia.”
“Ultimately we will be expanding our manufacturing base and the hope is we can support a broader and higher volume solutions and expand the technology from the current 200mm to 300 mm. We have Applied Materials as a partner and we have done a lot of work with them at 300mm for the cost advantages and equipment benefits so our plan is as we expand we expand to 300mm which will require another foundry partner.
“We have the roadmap for the next 20 years to improve the technology. The expectation is that silicon in this voltage range will be a cost performance leader. The cadence is every three to four years we will put out the next generation node. It takes 3 years to fill out the portfolio at a single node and while the design teams are working on that the technology teams are working on the next node.”
“It’s about delivering high levels of efficiency in a cost effective manner. Most of the companies we engage with are looking to engage with are looking to increase the efficiency. High frequency is certainly interesting.
The SuperQ approach can be used with GaN and SIC as well as silicon.
“Our perspective is that SiC and GaN are great power materials but they have a lot of fundamental issues in manufacturability that the industry has yet to work through,” said Granahan. “The defect levels still impairs the cost level that the market needs to expand. There are some applications such as EV drivetrains that seem to be a nice spot for them as cost is not so much of an issue.”
“GaN vertical devices for moving the grid to an electrified world is interesting and SuperQ is applicable to that structure to improve the resistivity,” he said. “That’s why we believe it is a nascent technology that can be applied to silicon and to SiC and vertical GaN devices and in BCDMOS with lateral devices. Being able to integrate power and control is a huge benefit but as you get into 120V and up the power device end up being 50% of the die. If you can shrink that by 10x you can go down from a 10mm2 die to a 5mm2 die so being able to shrink the power devices is a huge cost benefit. That’s a really interesting discussion.”
“Longer term we believe that SuperQ is very applicable to WBG so we are agnostic to the materials so once the SiC world works through the fundamental defect levels we can apply the technology there. In the meantime we think we have a couple of decades of silicon.”