//php echo do_shortcode(‘[responsivevoice_button voice=”US English Male” buttontext=”Listen to Post”]’) ?>
Is silicon gone? Not yet. While gallium nitride (GaN) and silicon carbide (SiC) power transistors are now praised as the new revolution in power management, there are still challenges to be overcome, especially the high cost and, in many cases, low reliability.
Reducing power consumption and heat from electronic devices is paramount to meeting the challenges of climate change and the current energy crisis.
We’ll continue to invest in massive data centers for cloud computing and, possibly, the metaverse, as well as in new smartphones and other small electronic devices. Both SiC and GaN can help reduce size, heat, and power consumption, but it will take time for those technologies to become the norm.
At electronica 2022, in a panel discussion moderated by EE Times’ Maurizio Di Paolo Emilio, several experts talked about the current and future challenges and opportunities of GaN and SiC power transistors, focusing on production and the rate of adoption.
During the discussion, which emphasized the advantages of the two technologies, it became clear that silicon MOSFETs are not going away anytime soon. While the cost of production of GaN transistors has reached the level of MOSFETs and is even lower, catching up with volume will take many years, even decades.
“The first revolution of power happened when bipolar transistors became MOSFETs; now, we’re going through the second revolution — whether it’s GaN or SiC, silicon is gone,” said Stephen Oliver, vice president of corporate marketing and investor relations at Navitas Semiconductor. “And if anyone wants to work on the cutting edge, the leading edge, whether it’s integration, whether it’s the highest voltage, the pure-play world, gallium nitride, silicon carbide is the way to go.”
The future for the two compounds to replace silicon MOSFETs and bipolar transistors is clear. The question is when and at what cost it is going to happen.
Efficient Power Conversion CEO Alex Lidow said, “We’ll never see the end of the MOSFET. Over 10 or 15 years, the legacy applications that are there today will stay in silicon MOSFETs. MOSFETs will probably decline in growth rate, maybe even go down in unit growth rates, but go up in price like the bipolar transistor. So that’s a long cycle aspect. Gallium nitride power transistors already cost less to produce than MOSFETs.”
By 2030, the industry expects the combination of GaN and SiC to reach the market value of MOSFETs.
“Today, 95% of the market is pure silicon; of course, SiC and GaN are going at a much faster pace,” said Gerald Deboy, senior principal at Infineon Technologies. “We as a company need to have all the technologies in order to devise the advanced differentiating designs between silicon carbide and gallium nitride. Meanwhile, silicon still fills a hole, and we expect coexistence for all the technologies for at least one decade. … If you look at the speed of silicon carbide and gallium nitride, GaN is slightly behind SiC, but GaN is going at a tremendous speed, and we see lots of opportunities.”
“Last year, the power semiconductor market for silicon MOSFET discrete modules was $28 billion,” said Guy Moxey, senior director of power products at Wolfspeed, “Silicon carbide was nearly $2 billion, and I think GaN was just hovering under $1 billion. So obviously very small. If you look at what the data reports are saying for 2030, which is not that far away — two to three design cycles — the SiC market is going to be near $20 billion, and I think the GaN market’s going to be up to $5 to $6 billion, maybe north of that.”
The panelists are confident that for some low-voltage applications, GaN’s price will match that of MOSFETs next year. “On the price point, if you look at a 65-W USB Power Delivery charger, we estimate the system price would be the same for gallium nitride and silicon in the first half of next year,” said Oliver. “At the same time, it’s 3× smaller and lighter, so it’s a great value proposition for the mobile computing industry.”
GaN technology is ideal for low-voltage applications but has some reliability issues when used in automotive applications where the two technologies combine.
“The main problem that we bump into with gallium nitride is that it’s very, very fast: To use it in an inductive situation, you need to slow it down, and you need to limit the amount of parasitic inductance that you have to deal with,” said Doug Bailey, vice president of marketing and applications engineering at Power Integrations. “Our strategy is to integrate the GaN part and also the SiC part. We have silicon carbide versions of our product for automotive systems. We integrate them into the same package with the controller.”
SiC can operate at higher voltages, but it is much harder to manufacture compared with silicon. That’s why the industry is investing billions to increase capacity to meet demand.
“We’ve already known for some years that it will take time; in 2016, our president announced the highest investment in our company’s history,” said Aly Mashaly, director of the automotive segment and power systems at ROHM Semiconductor Europe. “We also have a vertical integration system in our company; we have our own fab for the substrate of the raw material and also for the fabrication of the silicon dies and also modules.”
For the panelists, the path is clear, and the technology is sound. It will take a few years to ramp up production, get more design wins, and get to the price points that the electronics industry needs to be competitive.