Driven by the automotive sector, the semiconductor industry keeps thriving despite the current COVID 19-pandemic. New technologies and wide bandgap materials like SiC and GaN are currently transforming the industry.
Frank Heidemann, CEO of SET GmbH, explained in an exclusive interview with German trade magazine “Markt&Technik” what potential this offers for measurement technology. The German interview can be read here in the current issue 42 of Markt&Technik, or here.
New materials are driving the market for power semiconductor testing
“A leading position in SiC translates into a real market advantage”
Markt&Technik: Mr Heidemann, when we met at the embedded world fair 2020, we already covered the topic of HiL testing. In today’s interview, we want to focus on power semiconductor testing – i.e. your company’s second pillar. Of course, there is an overlap of the two topics, but still: Where do you currently put your development focus?
Frank Heidemann: Actually, in our company, the two topics complement each other perfectly. Throughout the years, we have gained a lot of expertise on fatigue tests and testing models for power semiconductors. And consequently, we have developed a lot of understanding for validation. So, we do not only have a single development focus, we rather apply our core competencies in various areas. In the power semiconductor segment, we work with high voltages of up to 6 500 V and high currents of up to 3 500 A. In HiL, the target is to create real-time environments and to simulate models. For the moment, it is true, however, that the market for semiconductors is rapidly growing due to silicon carbide, in short SiC, and that we are very excited about these technologies as well, for we play a substantial role in driving the current market development for dynamic testing systems.
It seems that the COVID 19-pandemic has little to no effect on the power semiconductor business. Why is that?
It is obvious, that COVID 19 has hit the semiconductor industry less hard when compared to other industries. Of course, the March lockdown also had its effects on the share notations of semiconductor businesses, but we have also had a much stronger recovery since. Right now, our industry is seeing a growth cycle, which in large parts is due to the technological change currently underway. After having talked about wide bandgap materials like SiC and GaN for twenty years, we now have actual use for these products as parts of our applications. In power electronics, we can tap new potentials. Infineon, for instance, has announced, that by the end of the year, they want to expand their SiC semiconductor portfolio by 50 percent. And other companies like STMicroelectronics are also looking to massively increase their SiC turnover during the next five years.
What does the technological change toward wide bandgap mean for the market?
For power semiconductors, SiC comes with many huge advantages when compared to silicon: It is much more efficient, with a higher temperature resistance, better and more precise switching capabilities and it can be produced on a much smaller scale. Companies with a leading position in SiC technology really have a strong advantage in the market, right now. In the automotive sector, it is becoming evident, that SiC in connection with MOSFET technology can replace the IGBT for high voltage applications with up to 1 700 V. Until recently, such a development was simply not feasible, and it bears tremendous benefits, as e-mobility strongly depends on efficiency. The batteries do not grow in sync with our need for range, and so we have to invest into vehicle efficiency. We simply cannot afford to lose energy to thermal loss instead of range or electric power. Against this backdrop, the smaller thermal dissipation of SiC semiconductors is a key factor. E-mobility is growing exponentially, and our expectation is that by 2025, 10% of all new vehicles will either strongly rely on hybrid engines, or even be fully electric vehicles – which is almost equivalent to a tenfold growth of the SiC market in this domain.
And what are the consequences for measuring technology?
In connection with SiC, measuring technology is now undergoing tremendous changes as well. The new technology produces various new outages, much more complex lifespan models and it also comes with higher requirements for measuring technology. When compared to the past, real time measuring technology is now widely beneficial in order to precisely understand, when and how a semiconductor will fail. Measurements are also conducted with a much stronger focus on detail, for instance when it comes to determining the temperature of individual semiconductors with increments of 0.5 °C or when measuring individual test pieces, instead of entire batches – while covering the milli-, nano- and picoamp range.
Where do you see the biggest challenges in connection with new materials like SiC?
Silicon is an established technology that has been refined over the course of thirty years. Now, the established processes and operations need to be adapted and optimised for SiC. We are still facing a couple of major challenges, here, and are currently trying to find production technologies that deliver SiC in qualities and quantities comparable to silicon. For now, the wafers remain rather small and there are only limited SiC quantities available on the market. The big question is, how to produce high-quality wafers from grown SiC semiconductors without producing a lot of scrap. As the grown crystals are very expensive, there is a lot of potential in terms of production costs. But SiC also has quite an influence on the designs of power semiconductors. Many companies are already introducing their third or fourth generation of SiC semiconductors and use iterations to actually tap the potential available in wide bandgap. From a technological point of view, it is a tremendous challenge to translate theory into products here.
With SiC, power semiconductors can also operate at much higher temperatures, but the surrounding components are not yet compatible. Where silicon set the limit at 175 °C, silicon carbide already enables 500 °C and more. We are encountering limits in terms of bonding, moulding and baking as well as joining and housing techniques prior to the semiconductor actually reaching its limit. Furthermore, reliability plays a key role. OEMs already have their eyes on the potential of the new technology, but this must not lead to premature outages in applications. Therefore, the challenge is to introduce an entirely new technology in large quantities and to use the efficiency, without integrating massive fault effects. And in order for the lifespan of the power semiconductor to also fulfil expectations in a serial production, we need to continue developing testing methods and standards.
Which changes will the technology change bring for the automotive industry?
OEMs see a huge opportunity to provide e-mobility with a range that is both practical and relevant for consumers. With a total range beyond 500 km and charging times of less than an hour for 80% battery capacity, nobody needs to fear being stuck in the middle of nowhere anymore – provided we have a corresponding infrastructure, of course.
Due to its recent history, the automotive industry is under a lot of scrutiny, when it comes to statements concerning distances, range and lifespan. The market is much more aware and thus ensures that nobody makes overly positive statements. And at the same time, there is a new technology that actually also has to deliver in terms range and lifespan. Proving these capabilities remains a challenge for now, as the qualification bases for SiC power semiconductors keep changing constantly. With the AQG 324, ECPE has already created excellent standards for the qualification of silicon semiconductors in the automotive sector in Europe. Right now, the body is collaborating with other global standardisation committees so that these standardised testing procedures also become available for SiC and GaN.
Why is there a need to adapt current qualification standards?
When looking at reliability, we have established testing procedures and fault models for silicon, to produce the necessary shortening times equivalent to the actual useful life in application. When looking at SiC, however, we are beginning to see that the new applications and the existing testing standards do not match. The currently applicable static reliability methods fail to discover all faults and effects for SiC. Yet these static tests remain valid, as many faults still exist. But with SiC, we need to test closer to the application, i.e. more dynamically and no longer statically only.
Why do SiC power semiconductors require different testing procedures?
Let us take a look at the drift-behaviour of SiC components, which ultimately adds to the component efficiency. In silicon, we have hardly seen drifting as an effect over the course of the lifespan, because we really were in control of the semiconductor both from a technological and a process-related point of view. When testing SiC with the same static gate stress tests, there is hardly any exponential drifting either, which should be a good thing. But when stressing the same components with dynamic activation at the gate, we suddenly see completely different results. Depending on the manufacturer, these results vary from rather small drifting tendencies to relatively high drifting tendencies – and I am talking multiples here. So, we managed to clearly demonstrate, that there are some dynamic drifting and aging effects in power semiconductors, which completely elude our currently applicable static testing procedures. Now, the challenge is to develop individual physical fault models for SiC. Our research currently focusses on the available effects for SiC in dynamic areas close to the application and together with manufacturers and OEMs, we are developing new fault models. These findings then serve as the basis for standardised testing procedures.
What are the implications of such fault images for the actual application?
When a car manufacturer promises a range of 500 km, the customer expects their car to actually cover those 500 km. But if simply using the power semiconductor deteriorates its efficiency over the course of its useful life, the customer will have to face a range of only 470 km after one year. Of course, we want to avoid this scenario, as it would entail a lot pressure directed from the market towards the manufacturers.
What are the resulting challenges for testing?
We are currently working in a broad stress field: On the one hand, we have OEMs, who need to ensure that their relatively new supply chain for electric engines focusses on the right areas, so that no supplier can wilfully or negligently introduce a fault. The moving parts inside an engine are hardly worth mentioning these days, but we now have other sources of faults – one of them being power semiconductors. Every inverter you receive from your manufacturers needs to provide the same performance. The predominant question here is to define standards in qualification and to really insist on adherence to those standards.
On the other hand, we have the semiconductor manufacturers with their premium testing equipment tuned to previous standards. They may need to apply new tests, which are simply not compatible with their equipment. Of course, they understand the need for better and different testing, but investing large sums into testing equipment would also entail higher unit costs.
All parties involved wish to act in a sensible way to prevent any poorly conceived technology from becoming part of a vehicle in serial production. But they are also trying to avert total over-testing from resulting in tremendous price increases on the market for materials and components. This requires a certain sense of proportion and intensive exchanges between all parties involved, for them to jointly reach a good solution.
What is next in this exciting market?
All the major semiconductor manufacturers are currently working on high-grade SiC products, to establish a leading position in this rapidly growing market. The turnover effects of the current design-in opportunities will be visible in around two to five years, after they have entered the serial stage. This will certainly help the market take off, but all those, who fail to successfully establish SiC, will soon encounter problems in this market environment.
This also applies to manufactures, who are unable to control the new fault modes. Going forward, they may no longer be listed as second suppliers, as the influence on the application will simply be too big. Therefore, it is one of the main objectives to eliminate these effects and to prove that they no longer exist in new generations of power semiconductors. OEMs will also have to stretch a bit, in order to minimise the risk and, from a technological point of view, to move more into the direction of the application. We keep seeing more and more designated experts with the OEMs, who are deeply involved with their semiconductors. They can discuss with the manufacturers on a level playing field, and together, they can come to reliable systems fast.
All in all, power electronics is one of the big areas of growth in the semiconductor business. Going forward, we are going to see much more efficient and smaller drivetrains. This will provide us with much higher degrees of freedom in application, because we no longer need large installation space or we could use it much more flexibly, for instance with embedded technologies – and that opens up a much broader market for application. In future, we expect power semiconductors to influence every moving component. With their application in fully electric or partly electric hybrid drivetrains, the new SiC power semiconductors will also contribute to improving the ecological balance sheet of many industries and to solving one of the central topics of the future.
In this regard, what potential do you expect for the testing and measuring technologies business? And for your own company, in particular?
The new semiconductor technologies propel our growth, because right now and from a technological standpoint, we hold a leading position with quite an advantage compared to the market when it comes to dynamic stress and fatigue testing like dynamic HTGB (high temperature gates bias) or dynamic H3TRB (high humidity, voltage and temperature reverse bias). But we also see a large growth potential for measuring technology manufacturers in the domains of high voltage and high amps, when new measuring technology will be established that is currently still underrepresented. This, for instance, applies to fast and precise high voltage SMUs or high frequency measuring technology in the domain of power semiconductors. This certainly is one area of growth that will massively participate in the ongoing technology change.