+44 (0)24 7671 8970
More publications     •     Advertise with us     •     Contact us
 
Loading...
Technical Insight

Magazine Feature
This article was originally featured in the edition:
Issue 4 2023

GaN: A natural playing field for innovation

News

GaN companies are re-purposing developments made in silicon over the past 70 years, while exploring new avenues for innovation and integration.

Professor Florin Udrea, CTO of Cambridge GaN Devices, gives his personal view on the extraordinary potential of GaN.

There are many ways to define innovation but the Greeks are always a good place to start. Philosopher Socrates advised: “The secret of change is to focus all of your energy, not on fighting the old, but building on the new.” Addressing the same theme from a different angle, Albert Einstein is alleged to have said: “The definition of insanity is doing the same thing over and over again but expecting different results.”

I personally believe in creating a culture of innovation — one of the founding principles of our company. If an organisation encourages creative thinking from all its employees, new ideas will emerge, challenging the old, traditional ways. Creativity sparks further creativity and great strides can be made.

All my experience - and my instinct - tells me that innovation results from a deep knowledge base. Before Cambridge Gan Devices, I ran a team at Cambridge University for more than 30 years researching different materials and technologies. Many ideas - different configurations, contexts and geometries - that were developed for silicon are now being applied to GaN. For example, superjunction technology and membrane technology are re-emerging as possibilities for GaN.

Of course, there are brand new concepts that have been developed purely for GaN too. It is this thorough and detailed understanding and appreciation of materials and device physics that is enabling new power applications based on GaN.

With our background, GaN was a natural playing ground. It is a very interesting material, but of all the materials I have studied - silicon, SiC, diamond, GaN - by far the most difficult is GaN.

GaN is not only a wide bandgap material but its incorporation in a heterojunction is what makes it special. Indeed, the use of a GaN/AlGaN structure, enables an interface quantum layer (2DEG) with concomitantly high electron charge ( ~ 1e13cm-2 ) and high mobility ( ~1700 cm2/(Vs)).

So, on one hand, GaN companies re-purpose all the developments that have been made in silicon over the past 70 years, and on the other hand GaN offers new avenues for innovation.

Furthermore, as opposed to SiC, GaN enables integration. So, for example, intelligence, protection and sensing can be produced in GaN and delivered on the same chip as the HEMT. Of course, this is also possible in silicon, but only at low power. GaN permits integration at much higher power levels and much higher frequencies.

Will GaN replace silicon in other markets? Well, for applications where high power and high frequency are required, GaN is the best material. But digital electronics will remain based on silicon, because silicon has both n-channel and p-channel transistors and can be scaled down to nm levels.

GaN has the 2DEG structure which is fantastic for making an n-channel transistor, but today there is no reliable equivalent hole gas structure that would enable the creation of a p-channel transistor in GaN. And because recombination times are very small, bipolar devices are also not possible in GaN, Therefore, for the time being at least, GaN’s best prospect is as HEMT device only. But for this purpose, GaN is a fantastic, marvellous material. It will be used for power applications from maybe 40V to about 1.2kV and is going to dominate in high-frequency applications. GaN is also interesting for RF and there are opportunities in optical applications such as LEDs.



2x pictures of the ICeGaN H2 HEMT


A novel and innovative approach to the GaN market

Cambridge Gan Devices (CGD) has taken a completely novel – innovative – approach to the GaN market: we optimise the use of GaN integration. CGD adds intelligence, sensing and protection, making the gate extremely reliable, but at the same time keeping the simplicity of a highly efficient transistor.

This approach supports the two claims we make about our ICeGaN GaN technology: that is easy to use, because our transistors can be driven in the same way as a silicon or silicon carbide device; and that it matches or even surpasses the reliability of silicon and silicon carbide.

Let’s break that down a little.

Unlike other companies that try to integrate the actual HEMT driver with the HEMT itself in one IC, CGD integrates the Miller Clamp and HEMT and an auxiliary device to regulate the voltage from a control pin (external gate) to the actual gate of the HEMT.

The Miller Clamp regulates dv/dt which is the most critical and complex issue to handle when driving a GaN HEMT. This also allows more relaxed design rules regarding how far the driver can be placed from our ICeGaN device.

Given the above, it is perfectly possible to take a standard silicon MOSFET driver IC and use it to drive CGD’s ICeGaN HEMTs without any extra circuitry. To get the best performance from an ICeGaN HEMT however, especially at high frequency, an optimised driver will be required, but all the complex slew rate design issues have been solved.



Picture of the CGD team with CEO Giorgia Longobardi holding the trophy that was awarded to the company for “Best Demo” at the Innovation Zone of TSMC’s 2023 Europe Technology Symposium.


The reason that CGD chooses not to integrate the full driver circuit is because to optimize performance, a power electronics designer must match the driver to work with their preferred controller. Also, the driver may also be used for three phase or multiple devices.

Instead of making a very complex, and perhaps sub-par, power IC that cannot be made to function optimally because of the lack of GaN p-channel transistors (meaning, for example, that it is not easily possible to integrate a totem pole type driver) CGD has chosen to perfectly integrate the essential elements that are required to make the transistor easy to use, and very reliable. Finally, if the driver and HEMT were to be fully integrated, they would also be connected thermally, leading to performance and reliability issues.

The reliability challenge is complex. Intrinsically, GaN is at least as reliable as silicon, perhaps even more so, because the intrinsic carrier concentration of GaN is very small, therefore lower leakage currents are possible. But silicon has one huge advantage: the presence of the silicon oxide. The silicon/silicon oxide interface is close-to-perfect (described by my teacher as ‘God-given’) when used to make an insulated gate. With GaN, the interface is not so good, so a p-type magnesium dopant is used to create the the so called ‘p-GaN gate’. Unfortunately, this limits the threshold of most GaN gates to around 1.3-1.6V, unless accompanied by an undesirable and significant increase in the specific on-resistance.

Another innovation that CGD has introduced in ICeGaN roughly doubles this gate threshold voltage to 3V. An auxiliary HEMT, integrated alongside the main switch and the Miller Clamp, is connected in a pass configuration between the external gate and the internal gate. The Miller Clamp addresses the fast dv/dt challenge, while the auxiliary HEMT delivers a higher threshold voltage and extends the gate voltage both in static and dynamic conditions. This means that there is no need to use a negative rail that makers of other GaN devices recommend employing when turning the transistor off to avoid retriggering the transistor. By negating the need for negative voltage rails – which have been shown to cause degradation over time – ICeGaN devices can be driven from 0-10V, 0-15V or 0-20V as preferred by the user, in the same way as silicon and SiC devices.

Robustness
Independent research by Virginia Tech University has also demonstrated that CGD’s technology is more robust than other GaN platforms. Experimental evidence presented in a paper at APEC 2023, titled ‘A GaN HEMT with Exceptional Gate Over-Voltage Robustness’, shows that ICeGaN HEMTs exhibit an exceptionally high over-voltage margin of over 70V, which is comparable to state-of-the-art traditional silicon devices.

Accidental high drive voltage is a critical concern for the gate reliability and driver design of GaN HEMT devices. Previously, other GaN HEMTs survived to around only 25V, which can be well within gate voltage overshoots in applications such as converters, resulting in device failure. Moreover, under repetitive gate voltage spikes, a discrete GaN device may see degradation well below the 25 V, as demonstrated by Virgina Tech. Until ICeGaN, higher dynamic breakdown voltage values of 70V and more, were only possible with state-of-the-art SiC and superjunction devices. ICeGaN’s hugely elevated dynamic gate breakdown capability is enabled by the integration of protection circuitry as discussed earlier.

Concluding the discussion on reliability, CGD has addressed the issue of very low temperatures, where GaN can fail. Even if a steady voltage is maintained on the outside gate, the inner terminal of the transistor will vary with temperature. So at lower temperatures, ICeGaN’s ‘Smart Robustness ‘ integrated protection circuitry clamps the device harder, to make it more reliable. No other GaN company offers this solution.

Innovation is one of the core pillars that Cambridge GaN Devices is based upon, and it runs through every aspect of our business, not just technology. This approach, I believe, often works better in small companies that are more dynamic and do not have the burden of previous business history and expectations.

For example, we have no fear of killing a profitable superjunction business with new GaN products because we don’t have those legacy devices. We have one mission, which is to deploy GaN. To achieve this, all ideas are considered…and hence we are extremely open to innovation.



Florin Udrea, CTO of Cambridge GaN Devices



×
Search the news archive

To close this popup you can press escape or click the close icon.
Logo
×
  • 1st January 1970
  • 1st January 1970
  • 1st January 1970
  • 1st January 1970
  • 1st January 1970
  • 1st January 1970
  • 1st January 1970
  • 1st January 1970
  • View all news 22645 more articles
Logo
×
Register - Step 1

You may choose to subscribe to the Power Electronics World Magazine, the Power Electronics World Newsletter, or both. You may also request additional information if required, before submitting your application.


Please subscribe me to:

 

You chose the industry type of "Other"

Please enter the industry that you work in:
Please enter the industry that you work in: