Electrifying the Automotive Future: Power Semiconductors in the EV Revolution

Are you ready to supercharge your knowledge on the electrifying transformation of the automotive semiconductor industry?  
 
Fasten your seat belts for a high-octane conversation with Fabio Marino, managing director at CREA at Advantest in Italy and a trailblazer, as we drive through the accelerating shift towards green energy.  
 
We’ll be zooming in on the rapid growth of the power semiconductor market and how it's turbocharging the EV demand. Get ready to rev up your understanding of power electronics technologies like IGBTs, insulated gate bipolar transistors, silicon carbide, and gallium nitride as we put them head-to-head, comparing their specifications and power capacities. 
 
Also, we’ll shed light on how the semiconductor industry is navigating the switch from package to bare die testing, examining how it challenges the delivery of kiloamps or kilovolts through a probe card whilst maintaining efficiency and reliability.  
 
Lastly, diving into the battery testing arena, we will make sense of the implications of long lead times for probers and handlers and touch down on the maturity of silicon carbide technology and the potential influence of automated tools and the market.  
 
So buckle up, relax, and join us as we journey through the electrifying world of automotive semiconductors. 

Thanks for tuning in to "Advantest Talks Semi"!

If you enjoyed this episode, we'd love to hear from you! Please take a moment to leave a rating on Apple Podcast. Your feedback helps us improve and reach new listeners.

Don't forget to subscribe and share with your friends. We appreciate your support!

Transcript

Keith Schaub: 0:27

Welcome to another episode of Advantest Talks Semi, the podcast where we dive into the hows and whys of technological progress and the waves they create in our world. I'm your host, Keith Schaub, and today we're hitting the road with Fabio Marino, managing director at CREA, from Advantest in Italy, to talk about the automotive semiconductor industry and how electric vehicles - EVs - have seemingly electrified the industry. Over the past 10 years EVs have transformed from a niche to a mainstream market. In 2022, EV sales were up as much as 50% over 2021 and there are now more than 26 million EVs in the market. Some analysts have estimated that by 2032, as much as two-thirds of all vehicles sold in the US will be electric, and several major car companies have announced they will stop producing gas engines by 2035. What fueled this shift, what challenges and opportunities does it pose for the semiconductor industry, and where do we go from here? Let's plug in and get charged up for an electrifying conversation. Hi, Fabio, welcome to Advantest Talks Semi.

Fabio Marino: 1:41

Hi, Keith, really a pleasure being here.

Keith Schaub: 1:44

It's interesting to see how we've had electric vehicles for over a hundred years, but only in the past 10 years, has it really become mainstream. I wanted to get your take on: what are you seeing in the industry that's sort of causing this shift to move from gas power engines over to EVs?

Fabio Marino: 2:04

Well, we all understand how important it is to protect our world, and moving to green energy is somehow a road we all need to walk. Advantest, for many years, has been promoting several initiatives also in this direction, and going to electrical vehicles is exactly the right thing to do in order to minimize pollution, air pollution, and protect our world. So that was stimulating this growth.

Keith Schaub: 2:38

Fabio, what are some of the current trends that you are seeing in the EV market?

Fabio Marino: 2:42

The main trend I see at the moment is really leading the electrical vehicle market, so all the companies are really raising to gain the market share. So, the power semiconductor market is expected to go from 150 million in 2022 to 520 million in 2030. I'm talking about the testing market, not the semiconductor market.

Fabio Marino: 3:10

So, a 5x growth, and the testing market normally is in the range between 1% and 2.5% of the market, so you can calculate how big the market is, and this is driven by electrical vehicles. So, the trend I see is, first of all, for companies to win market share in that market, and I see Chinese companies investing a lot. And companies, for example, BYD, overtook even Tesla by far, by two times already during the last couple of years.

Keith Schaub: 3:45

Thanks, Fabio. What about the technological innovations that have been occurring in the industry? What's been going on in the past - let's say seven, eight, ten years that is helping that transformation?

Fabio Marino: 3:58

So, these kinds of power semiconductors, most of them are known as discreet. They have been on the market for many, many years. They were pretty common in white goods like washing machines, for example, and refrigerators. But they were really a niche for the market because they had a very low average selling price, and components, very cost sensitive, so not really appealing in the market. In reality, those devices are now back in the game thanks to the need to go to green, for electric vehicles. And we are using exactly the same components from the past. On top of those components, which are the typical IGBT, the typical power, MOSFET, power- BJT, but on top of those components recently silicon carbide and gallium nitride came into the game. Those are two new technologies made of wide-band materials with some specific features that empower the specification of those devices. But this is the only new technology that during the last few years. For the rest, we are talking about very established technology.

Keith Schaub: 5:19

Fabio, you mentioned several technologies like IGBT's insulated gate bipolar transistors and silicon carbide, and there's also gallium nitride or GANS. Can you quickly explain and contrast these new frontier technologies for us?

Fabio Marino: 5:35

I think our colleagues, but in general people, tend to put together silicon carbide and GAN as the new frontier for power electronics. But there is a substantial difference between silicon carbide and gallium nitride. From a technological perspective, gallium nitride is much more similar to the standard silicon, so this makes this technology much more mature and easy to handle by the traditional semiconductor company. And also, so integration, for example, will be much easier on gallium nitride. The other topic is that gallium nitride is also a wide-band gap material. It is much faster, so it allows really fast operation switches. So, this makes this component very good for switching, so for DC-DC converters, for example. And even the sides are smaller because of the higher frequency. So this is perfect for small components like mobile phones, for example.

Fabio Marino: 6:44

The limitation or the other difference between the silicon carbide and gallium nitride is in the specification. So normally gallium nitride, which is faster, can go up to 650 volts. There are some experiments to go beyond 1000 volts, but still on the niche side compared to the 650 volts, while the silicon carbide can go up to kilovolts. So, even the field of application is different. So normally we tend to put together silicon carbide and gallium nitride for power electronics, but this addresses two different kinds of markets. Silicon carbide is more focused on electrical vehicles. Gallium nitride is more focused on a DC-DC converter for such power-greedy applications. Actually, the main difference is in the power that those devices can handle, both in terms of voltage and current. There are standard silicon components that can handle high voltage but low current and vice versa, high current and low voltage, but in reality, those components handle mid-low power. The power semiconductor can handle both high voltage and high current, resulting in a big amount of energy.

Keith Schaub: 8:04

I'm assuming then there are specific test challenges that go along with that because of the extreme high voltages and the high currents.

Fabio Marino: 8:15

Your assumption is correct, Keith. Actually, the big challenge is in the dynamic test. So, for these kinds of components, there are two kinds of tests: static and dynamic. The static test falls in, what we can call, low power test, because the switch, the device, is either on or off. And in that condition, the overall power is low, because it can be high voltage but low current or high current and low voltage, so in this case, the total power would be low. But during the dynamic test, which tests actually the device under its transition from on to off or from off to on, the amount of energy in the game is very high, even if for a very limited amount of time. And this is the big challenge.

Keith Schaub: 9:10

I guess we're talking about the instantaneous voltage changes and the instantaneous currents. Is that correct?

Fabio Marino: 9:17

It is correct, but both occur at the same time. So, while the current is flowing, very high current is flowing, the voltage is also going high and the combination results in high power.

Keith Schaub: 9:31

And what sort of voltage and current levels are we talking about here? I've seen numbers that are in the thousands of volts and the hundreds of amps.

Fabio Marino: 9:41

It depends on the final application. It can be for a train, for example, for electrical vehicle, or for photovoltaic cells. But actually we are talking about 10 kilovolt at 10 kiloamps, for example. So we are talking about an extreme, huge amount of current and voltage.

Keith Schaub: 10:01

How many probe tips supply the currents to the wafer probe solution?

Fabio Marino: 10:06

It sounds like a surprise that a 3-pin device, 3-pad device and a bird eye needs 3,000 needles in a probe card for a single site test. And this is because each needle can only stand a limited amount of current, normally in the range of 1 amp. But we have test conditions e. g. for short circuits, where we provide 3,000 amps, and to do that you really need 3,000 pogos on the probe card. This makes the probe card business also very interesting because those consumable parts are very expensive. The testing is pretty much robust, but for good parts, the testing time is so limited that this does not represent any problem. For failing parts anyway, those parts cannot really stand that current and immediately show problems that our probe card interface technology detects, protecting then the full equipment.

Keith Schaub: 11:13

So, that brings up testing challenges. With that high of a voltage or that high of a current, how do we actually test those moving from a niche market to more of a high volume mainstream market?

Fabio Marino: 11:28

So, first of all, the first challenge is the capability right, the tester capability. So not all the tester equipment is able to test those devices during dynamic behavior. At CREA, however, we develop proprietary technology that allows that test. We can test those dynamic specifications thanks to our low- stray inductance technology and the probe card interface technology.

Keith Schaub: 11:57

Give us the background on CREA. Where did it come from, how did it get started and what's their specialty?

Fabio Marino: 12:03

CREA started in 1992 as an engineering company developing a specific product based on customer demand. Over the years, the demand was going more from an engineering need to a production need at a very high current and very high voltage. CREA developed this technology year by year. Currently, the products we have are top-notch technologies in terms of dynamic test of power semiconductors. In order to be successful, CREA had to face difficult challenges for dynamic test because, as I said before, dynamic test implies huge energy. And with this huge energy, parasitic inductance, parasitic capacitance come into the game and can impact testing quality and can even break the test equipment. That was the first challenge CREA had to face and over time we developed the low- strain inductance technology in order to minimize these parasitic values to the point that we can test those devices at very high current and voltage. The new trend is anyway to move from package test to bare die test.

Keith Schaub: 13:19

Why is that?

Fabio Marino: 13:20

The main reason for that is package test is much simpler, but the drawback is that each package has 6 to 48 single switches. And if one of those switches is broken, then the entire package has to be trashed, actually with a big amount of money wasted. Customers are now asking for bare die test in order to screen the parts before assembly and minimize the cost impact on the package side. This is the reason for that, but unfortunately, this brings another big challenge in the testing arena, which is the bare die test, where you need also a probe card to perform the test. And if something goes wrong or if a device is broken, the big amount of energy in the tester breaks the probe card, the chuck, the die and even the tester. And that was the challenge that CREA had at the phase at the beginning.

Keith Schaub: 14:21

How do we deliver kilo amps or kilovolts through a probe card reliably and efficiently?

Fabio Marino: 14:29

Actually, the solution for us was to develop a probe card interface, which is a piece of hardware that is in the middle between the tester and the probe card. This hardware monitors the current distribution for each and every needle or for each group of needles in the probe card. By doing that, we can predict if something is going wrong. If there is an abnormal current distribution, we can immediately switch off the tester and protect the equipment at the probe.

Keith Schaub: 15:06

EVs have fewer parts than combustion engines. I've seen where EVs typically have around 13,000 parts where combustion engines have 33,000 parts. Almost three times fewer or a third as many moving parts, which also reduces maintenance costs and repair costs and helps fuel the transformation and the interest in EVs. Is that something that you see when you talk to power semiconductor manufacturers?

Fabio Marino: 15:39

This is a good point! What this reduction in the moving component is bringing to the game is that many more companies are now capable of manufacturing electric vehicles, because the electric vehicles remove the technological barrier of thermal engines and go to an electric vehicle technology which is much easier. Many more companies even, for example, companies traditionally in the mobile business, now are willing to build electric vehicles and they are moving in that direction.

Keith Schaub: 16:28

Basically, the barrier to entry for automobile manufacturing is now much lower, so we have a lot more players in the market. That's going to be pretty interesting for competition, but also good for consumers, I guess.

Fabio Marino: 16:43

Exactly. This will dramatically reduce probably the costs for electric vehicles, rather than having much less components in it.

Keith Schaub: 16:52

What's coming in the future? New needs or new capabilities that are required to address this market.

Fabio Marino: 16:59

There are a couple of things that I see. One from a technological perspective is to make the silicon carbide technology mature enough for this market because this gives big advantages in terms of component size and thermal dissipation power. Silicon carbide is a wide- band gap material, and, by definition, those materials can stand higher voltage, higher current, can be smaller, and have much better thermal dissipation powers. This is the first challenge because this technology is not mature yet. The production yield is not big. So, this is a big opportunity for testing companies because the yield is not so high. But of course, semiconductor companies need to make this technology mature to make the business sustainable for them. This is the first challenge. The second technology challenge is integration as you mentioned. So, we see for the C-I ntelligent-P ower-M odule to grow a lot integrating power and some digital control or even mixed signal acquisition, for example, in a single package or in a single integrated power module.

Keith Schaub: 18:17

You mentioned the silicon carbide maturity and it's still a maturing market. Can you talk about what are some of the things that's causing that and what are some of the things that need to happen in order for that to mature?

Fabio Marino: 18:32

I see that semiconductor companies are investing. You may read in the newspapers that several companies are investing in silicon carbide factories. So the first problem silicon carbide has is, it is a very tough material to handle, in the sense that the standard silicon wafer technology does not apply as it is for silicon carbide. You need different technologies, completely different from silicon. So even singulating the dye of a wafer can have a big impact on silicon carbide and also the wafer sides at the moment are limited to six inches so why stay limited on the standard silicon when you can have 12- inches wafers? These are all the barriers that semiconductor manufacturers are facing now and need to manage in order to make this technology mature.

Keith Schaub: 19:26

The automated tools that we have in the factory cannot handle the new silicon carbide. Factories and companies are retooling for this new material but at the same time they're getting the boost from the market which is fueling that investment.

Fabio Marino: 19:43

On the integration challenge I think we are ready.

Keith Schaub: 19:47

The main takeaway is we are not a bottleneck. Our technology is essentially in place.

Fabio Marino: 19:53

Correct. You said a very important thing, Keith. We are not a bottleneck. So, if I look at the entire test line chain where we need probe cards, we need probers, we need handlers. I see that probers are not even available in the market yet for dynamic tests, only for static. Handlers have a very long lead time. There are a few handling companies for bare die that have more than a year lead time for the equipment. And probe cards: There are very few vendors and they are very expensive. So, from the tester side, it's absolutely off the shelf available with the right price, market price, and the right specs.

Keith Schaub: 21:26

So, Fabio, I want to thank you for coming on to Advantest Talks Semi, and looking forward to the next podcast about automotive power semiconductors. Thanks, Fabio.

Fabio Marino: 21:37

Thanks, Keith. It's been a pleasure being here with you.