下一波SiC浪潮：製造、供應鏈和成本

Welcome to PowerUP, a podcast show hosted by Maurizio Di Paolo Emilio that brings life to some of the stories on power electronics technologies and products featured on PowerElectronicsNews.com and through other AspenCore Media publications. In this show, you’ll hear both engineers and executives discuss news, challenges and opportunities for power electronics in markets such as automotive, industrial and consumer. Here is your host, editor-in-chief of PowerElectronicsNews.com and EEWeb.com, Maurizio Di Paolo Emilio.

歡迎收聽由Maurizio Di Paolo Emilio主持的podcast節目Powerup。本節目為PowerElectronicsNews.com和其他AspenCore Media旗下網站的內容帶來活力。在本節目中，您將聽到工程師和高層討論在汽車、工業與消費性電子等市場的電力電子相關新聞、挑戰和商機。歡迎我們的主持人，Power Electronics News和EEWeb.com主編Maurizio Di Paolo Emilio。

MAURIZIO DI PAOLO EMILIO:    Hello, everyone, and welcome to this new episode of PowerUP. Today, we will talk about silicon carbide, the next wave of SiC manufacturing, supply chain and cost. The SiC industry is growing in many markets. The electric-vehicle market is preparing to move toward SiC inverters, as Tesla has already done. Mercedes-Benz has adopted onsemi SiC technology for traction inverters as part of a strategic collaboration. So the range of SiC devices is becoming well recognized and offers a wide-bandgap alternative to traditional IGBTs. As the industry moves from internal-combustion engines to EVs, the adoption of new solutions that can increase efficiency and offer longer range and faster charging will provide benefits across the powertrain, and device manufacturers want to ensure they have access to high-quality SiC substrates to support their customers. In addition, the technical advantages, including switching speed and cost, still remain an important point. Moreover, the importance of SiC has prompted many companies to review and invest in wafer technology to define the development plans in line with the demand. There are several approaches to improving the supply side of SiC devices. These range from scaling manufacturing with larger-diameter single-crystal wafers, improving performance, defectivity and yield. All these improvements working together can help meet the exponential growth in demand foreseen for these power devices. In this podcast with Pietro Scalia, director of automotive traction solutions at onsemi, we will discover the face of manufacturing SiC solutions, design issues and what there is behind its cost. Let’s talk with Pietro.

MAURIZIO DI PAOLO EMILIO：大家好，歡迎收聽最新一集的PowerUP。今天，我們就來聊一聊碳化矽(silicon carbide)，SiC的下一波的製造、供應鏈和成本。SiC產業在許多市場都在成長中。正如特斯拉(Tesla)已在進行的，電動車(EV)市場正準備轉向SiC逆變器。賓士(Mercedes-Benz)也與安森美半導體(onsemi)策略合作，將其SiC技術用於牽引逆變器。因此，SiC元件的範圍得到了廣泛認可，並提供作為傳統IGBT的寬能隙替代品。隨著產業從內燃機轉向電動車，採用可提高效率並提供更長續航里程和更快充電速度的新解決方案，將為整個動力系統帶來好處，元件製造商也希望確保能夠獲得高品質的SiC基板以支持其客戶。包括開關速度和成本等技術優勢，仍然是重點之一。此外，SiC的重要性促使許多公司重新審視並投資晶圓技術，以定義符合需求的開發計劃。有幾種方法可以改善SiC元件的供應面，包括使用更大直徑的單晶晶圓擴展製造規模、改善性能、缺陷率和產量。結合這些改進之處將有助於滿足預期對這些功率元件需求的指數級成長。在這集podcast中，我們邀請到onsemi汽車牽引解決方案總監Pietro Scalia，一起探索SiC解決方案的製造階段、設計問題及其成本背後的原因。讓我們和Pietro好好聊聊吧！

Hi, Pietro. Thanks a lot for coming on. How are you?

嗨，Pietro。謝謝您的參與。您好嗎？

PIETRO SCALIA:     Very well, Maurizio. Thanks for having this session.

PIETRO SCALIA：我很好，Maurizio。謝謝您邀請我參加這集節目。

MAURIZIO DI PAOLO EMILIO:    Okay. So today, we will talk about SiC, the next wave of SiC in terms of manufacturing, supply chain, cost and so on. But before that, tell our power electronics community more about you.

MAURIZIO DI PAOLO EMILIO：好的。那麼，今天我們就從製造、供應鏈、成本等方面來談談SiC、SiC的下一波。但在此之前，請跟我們的Power Electronics社群介紹更多關於您自己。

PIETRO SCALIA:     Thanks, Maurizio. So let’s say, I’m not new to power. It’s about 30 years now of journeying across different companies, and I like to say as well across different technologies. So I started, of course, with silicon, across different applications, first telecommunication and then cloud, which are very big, let’s say, megatrends. And now this new automotive electrification is a very exciting journey, and we are bringing wide bandgap finally into a different, let’s say, availability status. Very exciting to drive this extra revolution, which I believe is just started but will consolidate. So I started my journey in Siemens, then I worked for Ericsson in microelectronics and then moved from Italy to Germany and to Texas Instruments. I’ve been part of wide bandgap before with Wolfspeed, and now lately in the last two years with onsemi, which is moving the industry to the next level in this wide-bandgap area in automotive escalation.

PIETRO SCALIA：謝謝Maurizio。其實，電源領域對我來說並不陌生。在大約30年的職涯旅程中，我經歷過不同的公司，可以說也橫跨不同領域的技術。當然，我從矽晶開始接觸各種不同的應用，首先是電信，然後是雲端，這些都是非常大的領域，或者可以說是大趨勢。現在，新的汽車電氣化是一個非常激動人心的旅程，我們最終將寬能隙帶入了一個不同的可利用狀況。推動這場無比的革命非常令人興奮，雖然才剛剛開始但我相信但它將會日益增強。我從加入西門子(Siemens)開始了我的職業生涯，接著在愛立信(Ericsson)從事微電子工作，然後從意大利搬到德國，後來到了德州儀器(Texas Instruments；TI)。在加入Wolfspeed之前曾經參與寬能隙團隊，最近這兩年加入了onsemi，致力於提升汽車產業到寬能隙領域的新境界。

MAURIZIO DI PAOLO EMILIO:    So we have several challenges related to the growth and processing of SiC wafers before the manufacturing of devices. So I would like to understand more. So let’s understand which are the main steps, the main phases of SiC manufacturing, maybe highlighting some points where now there is a stronger request of innovation. So how is SiC fabrication different from that of silicon? And what are the main SiC fab models?

MAURIZIO DI PAOLO EMILIO：在製造元件之前，我們面臨著與SiC晶圓生長和製程相關的幾個挑戰。那麼，我想進一步了解，哪些是SiC製造的主要步驟、主要階段，或許也能重點介紹一些目前對創新有更高要求之處。SiC製造與矽製造有何不同？以及有哪些主要的SiC晶圓製造模式？

PIETRO SCALIA:     It will be very interesting to invite the audience to have a journey through our facilities in Hudson. We started the production of this SiC. And already watching these devices, you see the incredible difference in silicon. Typically, I think everybody is familiar with this silicon, which is very, very tall. In SiC, we talk about basically inches which takes weeks to get developed. So it’s a much more difficult process to go from the bare material, the packs that are needed for the following phase of creating the lasers. So it’s a question of temperature. So it’s a very heavy process in terms of energy consumption, because the temperature required is about 300 K. This is made in special furnaces. Onsemi manufactures its own furnaces. This is also very important. The company we acquired in 2021, GT Advanced, has a lot of IPs. They started with sapphire and then they went to SiC. Both very, let’s say, hard material. The hardness of SiC is another point, right? I mean, only diamond basically is harder than that. But of course, diamond gets consumed as well. So it’s an expensive process in all the three main steps. So the back production, as we said, and then the wafering, which is also, say, quite different than silicon. We can talk a little bit more about that, because for example, onsemi, we use the way of propagating fractures, which we create by vibration. And then the third phase is more standard. The epitaxy, it’s, let’s say, to some extent similar. But in between the wafering and the epitaxi, the peculiarity here of SiC is the defects. There are several defects that are intrinsic with the materials, and they can be classified, and there is a lot of publication in literature. But what is very important as well is that some of these defects are basically killer defects. So they must be removed from the wafers created. And this is a very dense activity of screening that needs to be done with the, say, wafer level that mean activities, and this is by optical as well. And this is intrinsic new elements, very different than silicon, that creates as well a lot of challenges for the yield of the entire process.

PIETRO SCALIA：如果有機會邀請觀眾來參觀我們在哈德遜(Hudson)的製造設施，將會更有意思。我們已經開始生產這種SiC了。如果您看過這些元件，就會知道它與矽的驚人差異。通常，我認為每個人都熟悉這種非常、非常高的矽。而在SiC，我們基本上談論的是英吋，這需要幾週的時間才能開發出來。因此，從裸材料開始就是一個更困難的過程，即創建雷射的下一階段所需要的封裝。所以這是溫度的問題。因此，以能源消耗而言，這是一個非常繁重的過程，因為所需的溫度約為300K。這是在特殊的熔爐中製造的。Onsemi自行製造熔爐，這也很重要。我們在2021年收購的公司GT Advanced，帶來了很多IP。他們從藍寶石起家，然後轉向SiC，這兩者都是非常堅硬的材料。SiC的硬度是另一個重點，對吧？我的意思是，基本上只有鑽石比它更硬。但當然，鑽石也會被消耗掉。因此，在所有三個主要步驟中都是一個昂貴的過程。因此，正如我們所說，生產以及晶圓加工也與矽完全不同。這部份可以多談一點，因為，例如，在onsemi，我們使用振動產生的傳播裂縫方式。然後第三階段則更標準。例如外延層，在某種程度上是相似的。但在晶片和外延之間，SiC的特殊之處在於缺陷。有幾種材料固有的缺陷，它們是可以分類的，這部份在文獻中有很多相關出版品。但同樣重要的是，其中一些缺陷基本上是致命缺陷。因此必須將它們從創建的晶圓中移除。 這是一項非常密集的篩選，需要在晶圓級進行，這意味著活動也是透過光學進行的。這是內在固有的新元素，與矽截然不同，對整個製程的良率也提出了很多挑戰。

MAURIZIO DI PAOLO EMILIO:    Let’s do a step back. Let’s understand what are the main attributes of SiC that have led to over 25 years of investment. Because there were a lot of investments in this technology. And during these developments, we have seen the further reduction of very specific on-state resistance, in addition to the improvements of RDS(on), also a reduction in conduction losses. Conduction losses for devices with the same chip area. How much performance can be gained by using SiC in EVs for sure, but also in energy trends?

MAURIZIO DI PAOLO EMILIO：讓我們退後一步來看。首先來了解這種引領超過25年投資的SiC，其主要屬性是什麼。畢竟在這項技術上有著很多的投資。在這些發展過程中，除了改善了RDS(on)，我們還看到非常特定的導通電阻進一步降低，具有相同晶片面積的元件傳導損耗也降低了。在EV中使用SiC可以提升多少性能是確定的，而在能源領域也有這樣的趨勢嗎？

PIETRO SCALIA:     I think these two areas you touched are right at the area where we see the most mass adoption. I think it’s very different if we talked about energy generation and storage and if we talked about automotive. Let me go with some numbers into automotive. Of course, we use SiC, starting from, let’s say, the on-board chargers, going to the DC/DC and going to the AB traction. Normally, efficiency improvement is very significant. So we talk about several units and times given up to 7%. But then at the end of the story, for traction specifically, it is about range. So if we start comparing basically solutions based on IGBP and solutions based on SiC, we are able to extend the number of km that the vehicle can do. So we do this exercise every day in Tier 1 and OEMs, even by 10%. So depending, of course, on the power of the model. But now SiC is going to supply models up to 325, 350 kW. The saving is huge. You see two motors. You can see one motor per wing. Sometimes you can see even three motors in one. And of course, all this power needs the best efficiency; otherwise, it’s a big waste of energy.

PIETRO SCALIA：我認為你提到的這兩個領域正是我們看到最大規模採用的領域。我認為，如果我們談論能源生產和儲存，這與談論汽車是非常不同的。讓我們用一些數字來看汽車。當然，我們使用SiC，例如從車載充電器(OBC)開始使用，到DC/DC，再到AB循跡控制。一般來說，效率的提升是非常顯著的。 因此，我們談論最多7%的幾個單位和時間，而到最後，特別是對於循跡控制，它跟續航里程是有關的。因此，如果我們開始比較基於IGBP的解決方案和基於SiC的解決方案，就能夠延長車輛的行駛里程。因此，我們每天都在和一線(Tier 1)供應商和原始設備製造商(OEM)中進行演練，即使是10%。當然，這取決於電源的類型。但現在，SiC將提供高達325、350kW的供電模型。在節能省電方面是巨大的。你看到它有兩個馬達，每個機翼都有一個馬達。有時您甚至可以看到三個馬達合而為一。當然，所有這些電源都需要最高效率；否則，這是能源的一種巨大浪費。

MAURIZIO DI PAOLO EMILIO:    Going forward, the goal will be on developing technologies for expanding the application of these devices. Cost reduction is a must. How can SiC wafer prices be reduced faster? And what means the next step of 200-mm substrates? So SiC devices are predominantly developed on 150-mm substrates, and there is an upgrade that should offer important advantages. So when can we see these advantages? What do you think?

MAURIZIO DI PAOLO EMILIO：展望未來，目標將是開發技術以擴大這些元件的應用。降低成本勢在必行。SiC晶圓如何更快降價？200毫米(mm)基板的下一步意味著什麼？還有，SiC元件主要在150-mm基板上開發，而且還帶來可提供重要的優勢的一次升級。那麼，什麼時候才能看到這些優勢呢？您怎麼看？

PIETRO SCALIA:     I think there are three elements we need to consider. First, let’s divide into two categories. One is more market demand, which of course generates price at the end. The second one is more technical. So let’s first go fast on the market demand. Clearly, there is discussed incapacity today. So all the players on the side, six players, we are in the market. There is not enough capacity to father-in-law the entire demand. This is changing fast. Clearly, production investments are very high. Onsemi has officially announced last year 9% investment, and now even going to 15% investment, most dedicated into SiC. And the interesting thing is that even if this number is growing, say the investors are pleased with our stocks, because clearly they believe that this investment is worth it and will generate revenue for the company and profit. So the market demand at some point will be a little bit better fulfilled, and definitely after the middle of this decade, we will see an improvement in that sense. Right now, the scarcity clearly brings prices to a higher level.

PIETRO SCALIA：我認為這需要考慮三個要素。首先，讓我們分為兩類來看，其一是更多的市場需求，這當然會在最後產生價格。其次是更具技術性。因此，讓我們首先快速了解市場需求。顯然地，今天討論的是產能不足。所以，所有的六家參與業者，都在市場上。然後，並沒有足夠的產能來滿足整個需求。這正在迅速改變中。很顯然地，生產投資非常高。Onsemi去年正式宣佈9%的投資，如今甚至要做到15%的投資，大部份都會投向SiC。有趣的是，即使這個數字在成長，投資者對我們的股票感到滿意，因為他們顯然相信這項投資是值得的，而且將會為公司帶來營收和利潤。因此，市場需求在某個時候會得到更好的滿足，在這個十年中期之後，我們肯定會看到在這方面的改善。目前，由於稀缺顯然將價格推高了。

If we go on the technical side, let’s focus on two elements. The first one is the meter, right, as you discussed. I would like to hear I like how long it took, like in silicon, to move, let’s say, from one step to another. And here again, wide bandgap, you see this acceleration extremely wide. So 200 mm is now in the works for sure, at least I can talk about onsemi. Our CEO has been announcing this in the stock analyst meetings. And we are going to basically start production in 2025. We already have material that GT Advanced had already at the time of the acquisition. Now of course, scaling this up to volume takes some time, because the challenges remain, right? We talked about defects. Defects sometimes are very dense into the edge of the wafers. So of course, you don’t want to scale up and then discover that you have a lot of defects in your yield. That is not sustainable. So the yield makes the price, right?

如果我們繼續看技術方面，讓我們關注於兩項要素。第一個是儀表，就像你所討論的。我想知道從這一步到另一步需要多長時間，就像在矽一樣。 再次強調，寬能隙，你會看到這種加速非常寬。所以，200mm現在當然已在進行中，至少我可以談onsemi的情況。我們的執行長已在股票分析師會議上宣佈了這一點。基本上，我們將在2025年開始量產。我們在收購GT Advanced時取得了其所擁有的材料。現在，當然，將這個量擴大會需要一些時間，因為挑戰仍然存在，對吧？接著來談談缺陷。缺陷有時會非常密集地進入晶圓的邊緣。所以當然，你不想擴大規模後才發現你的產量中有很多缺陷。那是不可持續的。所以，產量決定價格，對嗎？

Now I introduce the third point. I think it’s also a very key element for pricing. Today, we all try to produce at least mid-yield, north of 70% that are better numbers, might have better numbers. All depends also on the size of the die. Let’s assume that 25 mm2 today is kind of a market standard, willing to be in that area I said before. So this yield should not decrease as soon as you increase the wafer size. Actually, the eight inches should have similar yield. So this is one call to action and one challenge.

接著介紹第三點。我認為這也是定價的一個非常關鍵因素。今天，我們都試圖產生至少中等產量，超過70%產量是更好的數字，也可能會有更好的數字。這一切也取決於晶片的尺寸。讓我們假設今天的25mm²是一種市場標準，願意在我之前說過的那個區域。所以當你增加晶圓尺寸時，這個產量不應該下降。實際上，8吋晶圓的產量應該差不多。因此，這是一項行動呼籲也是一項挑戰。

I think on the other side, let’s also talk about the technology. So who started the market some years ago was basically in the past, they started planar. Onsemi also started planar. And we have developed across planar different generations. We are now at the fourth generation. planar has provided a lot of possibility of protecting the gate of the structure, because this is one of the key elements of their reliability. SiC has a very high energy density, close and high voltage density. Close the gate you need to protect. Doing this in a trench structure was very difficult. So what was the impact? The other competitors have decided to go trench soon, but they paid this in terms of active area. The active area is the area of the SiC you use for conduction. Of course, when you want to pay for 25-mm2 SiC, you would like to have almost 100%. Let’s say a realistic number is between 80% and 90%, even higher possible. And doing this with the trench in the past was almost impossible. So if you analyze some competition dies, you will find a very low active area, which today it’s possible to do something different, because in the meantime, the technology enables basically the possibility of going trench.

另一方面，讓我們也談談技術。幾年前開始做市場的基本上都是過去了，他們做的是平面的。Onsemi也開始平面化。我們已經跨越不同的平面世代發展，現在來到第四代。平面提供了許多保護結構閘極的可能性，因為這是其可靠性的關鍵要素之一。SiC具有非常高的能量密度，封閉且高電壓密度。它能封閉您需要保護的閘。在溝槽結構中做到這一點非常困難。那麼影響是什麼？其他競爭者已決定很快採用溝槽結構，但其代價是主動區域。主動區域是您用於傳導的SiC區域。當然，當你想採用25mm² SiC時，幾乎可達到100%。假設一個現實的數字在80%到90%之間，甚至可能更高。而在過去用溝槽做到這一點幾乎是不可能的。所以如果你分析一些競爭晶片，你會發現一個非常低的主動區域，而今我們以做一些不同的事了，因為與此同時，該技術基本上可以實現溝槽的可能性。

Trench, like in silicon, is definitely an obvious way to go to increase density and to of course enable a better price. So what do we see in the market? Onsemi has announced the M4 will be trench, and I think other competitors who were planner are probably doing the same. So the market is moving to trench, because today we can still have an active area in the 90% using trench. This is a big revolution. So to pick up this very long path, I think I wish to highlight that there is a component given to the market. So the market capacity availability. A second element, which is definitely the 8 inches coming into place. And the third element, which is a technology shift from planar to trench without over reducing the ratio of the active area, because we have the way now to protect the gate anyway and make the technology more robust. This is of course proprietary technology. I cannot explain too much how we do it.

如同矽一樣，溝槽絕對是增加密度的明顯方式，當然也能帶來更好的價格。那麼我們在市場上看到了什麼？Onsemi已經宣布M4將採用溝槽結構，我認為其他採用平面型結構的競爭對手可能也在做同樣的事情。所以市場正在轉向溝槽，因為今天我們採用溝槽結構仍然可以擁有90%的主動區域。這是一場大革命。因此，為了走上這條漫長的道路，我想強調市場有這樣一個組成元素。所以市場還有一定的產能可用。第二個元素絕對是8吋晶片到位。第三個要素則是從平面到溝槽的技術過渡，但不會過度降低主動區域的比例，因為我們現在有辦法保護閘極並使技術更加穩健。這當然是專有技術。我無法解釋太多我們如何實現。

MAURIZIO DI PAOLO EMILIO:    Let’s talk about another big topic: reliability. So in terms of reliability, gate oxide reliability, threshold voltage instability and others have been investigated in the last years. Maybe the gate oxide will be the most important. So tell me what you think. Because the gate oxide quality determines the lifetime, operation lifetime of the SiC devices. So what are your considerations?

MAURIZIO DI PAOLO EMILIO：讓我們談談另一個重要話題：可靠性。在可靠性方面，過去幾年來一直在研究閘氧化層的可靠性、閾值電壓不穩定性等。也許閘極氧化物將是最重要的。所以，告訴我們你的看法，因為閘氧化層品質決定了SiC元件的壽命、工作壽命。那麼，您的考慮是什麼？

PIETRO SCALIA:     Yeah, absolutely correct, Maurizio. So the gate is definitely a key part here, right? I mean, at least into the DPH means at least into the RDS off, right? And as you said at the beginning, the decrease of the specific resistance is one of the goals we had. So today, the technology is in a range, let’s say, of 5 million per square centimeter, at the high temperature of 175˚C. And we are going down and down. But again, the VTH needs to be super stable. And this is, again, part of the concept I introduced before about how to protect your gate from high voltage, how to protect for the high field, and there are techniques that enable us. Actually, onsemi is very proud to have a super stable BTH. There are some tests today done now by AQG324 even with, say, dynamic gate bias, which are stressing the modules where SiC dies are placed, which are probably put in place to respond to these characteristics. We have to say that JEDEC is doing a great job, and thanks to the contribution of course from the industry and from also the semiconductor industry. And we are moving fast. I think five years ago, we thought SiC was immature in terms of robustness. Today we think the opposite, because indeed, we have a lot of hours in the field. And honestly, the level of confidence has increased a lot. We have a very few failures, and frankly speaking, it can compare. You have more failures from the field in IGBP than I ever see in carbide today.

PIETRO SCALIA：是的，Maurizio，完全正確。所以閘極絕對是這裡的關鍵部份，對吧？我的意思是，至少進入DPH系列意味著至少進入RDS(off)，對嗎？ 正如您一開始所說，降低特定電阻是我們的目標之一。所以今天，這項技術在175˚C的高溫下可以達到每平方公分500萬個。我們正在走下坡路。但同樣地，VTH需要超級穩定。再次強調，這就是我之前介紹關於如何保護閘極免受高壓、如何保護高場的概念之一部份，而且也有了一些技術可以幫助我們。實際上，onsemi非常自豪能夠擁有一個超級穩定的BTH。如今，已經有一些測試採用AQG324進行了，例如動態閘極偏置，這些測試對放置SiC晶片的模組施加壓力，這些模組可能是為了響應這些特性而放置的。我們不得不說JEDEC做得很好，這當然要感謝來自產業和半導體產業的貢獻。我們正在快速進展中。五年前，我們認為SiC在穩健性方面還不成熟。今天我們的想法恰恰相反，因為事實上，我們在這個領域有很多時間。老實說，信心水平提高了很多。我們很少會有失敗，坦地說，它也經得起比較。你在IGBT領域的失敗比我今天在碳化物中看到的還要多。

and the approach for sure to semiconductor fabrication and packaging should play, also plays an important role in device reliability? What about supply chain?

MAURIZIO DI PAOLO EMILIO：讓我們繼續談談供應鏈。盡可能以垂直整合為目標的供應鏈非常重要。晶圓(包括晶圓和元件)的供應鏈產能當然需要增加，而且晶圓的成本也必須降下來。那麼，該技術正在尋求哪些改善之處，以及半導體製造和封裝的方法應該發揮什麼作用？還有在元件可靠性方面也起著重要作用？至於供應鏈呢？

PIETRO SCALIA:     So of course, let’s say price reduction is always a delicate topic, right? It involves procurement, and it’s difficult to put without a graphic. But let me say, if we give a specific resistance constant, right, we will see cost decreasing. There is no question about that. Of course the request comes, because as soon as your technology improves, you wish to see as well the RSP decreasing. So maybe at that point, the careful price becomes flat. So I can say that the vectors we have today, one is the move to the bigger sides. The second one is the reduced pitch. We are now in the range of a couple of micro millimeters, enables definitely the trend in the direction that the marketplace demand. Coming to the supply chain is absolutely essential to control all the steps. With the acquisition of GT Advanced, we have done this step. It was a big investment but extremely strategical, because you need to control the entire flow to control capacity. Today we work with capacity under control when customers come with a new opportunity. There is not enough capacity, right? You need definitely to put investment, and you can do only if you control the entire supply chain. The customers, OEMs, Tier 1, are very happy to know that their supply is under control, not by two or three parties, but one party, right? They sign agreements, typically strategical, long term, five, ten years, right? Because people want to secure the material to be sure that they can deliver cuts. We have, during Covid, assisted with this very crazy delivery crisis of the market. Nobody wants to see anything like that when we are replacing IC models with electrical models. So the supply chain is a very key point. And the only way to secure in a very responsible way to customers the capacity is to own the entire steps, so to own boule creation, to own wafering, to know epi, and if we talk about power models as well to own the kackend part of the production.

PIETRO SCALIA：當然，降價始終是一個微妙的話題，對吧？它涉及採購，沒有圖表也很難定價。但是，我認為，如果我們給出一個特定的電阻常數，我們就會看到成本下降。這是無庸置疑旳。當然，要求出現了，因為一旦您的技術得以改善，您也希望看到RSP下降。因此，也許到那時，審慎的價格會變得持平。所以我可以說今天所擁有的向量之一是向更大的方向移動。其次是縮小間距。我們現在處於幾微毫米的範圍內，絕對可以使趨勢朝著市場需求的方向發展。進入供應鏈對於控制所有步驟是絕對必要的。 隨著GT Advanced的收購，我們已經完成了這一步。這是一項巨大的投資，但極具策略意義，因為您需要控制整個流程以控制容量。今天，當客戶帶來新機會時，我們會控制產能。產能不夠吧？你一定要投資，而且只有當你控制了整個供應鏈，才能達到目標。 客戶、OEM、Tier 1供應商很高興知道他們的供應受到控制，並不是由兩方或三方控制，而是由一方控制，對吧？ 他們簽署協議，通常是策略性的、長期的、五年、十年的協議，對嗎？因為人們想要確保材料以確保他們能夠交付。在COVID期間，我們協助解決了這場非常瘋狂的市場交付危機。當我們用電氣模型替換IC模型時，沒有人希望看到這樣的事情。所以供應鍊是一個非常關鍵的點。以對客戶非常負責任的方式確保產能的唯一方法是擁有整個步驟，因此擁有晶圓製造、擁有晶片、了解外延，如果我們談論功率模型以及擁有後端部分 的生產。

MAURIZIO DI PAOLO EMILIO:    Before my last question, I would like to talk with you about packaging. So wide-bandgap devices, I mean, GaN and SiC for sure promise higher operating temperatures and greater efficiency. This is well known. But there are thermal management concerns that designers need to take into account when designing these devices into a system. How do you see thermal management demands with increasing power density impacting the future development of process and packaging technologies? What is your strategy about these packaging technologies?

MAURIZIO DI PAOLO EMILIO：在我最後一個問題之前，我想和你談談封裝。所以，寬能隙元件，我的意思是，GaN和SiC肯定會承諾更高的工作溫度和更高的效率。這是眾所周知的。但是，在將這些元件設計到系統時，設計人員需要考慮熱管理問題。您如何看待隨著功率密度增加而影響製程和封裝技術未來發展的熱管理需求？您對這些封裝技術的策略是什麼？

PIETRO SCALIA:     So we said before, right, to try to reduce price, increase performance, we are as well reducing the pitch of the die. The die are becoming more and more dense, means a lot of works are confined in a very limited space. So we know SiC of course can resist at higher temperature, and it’s a great thermal resistance. But we need to dissipate from a very small space. So the package needs to come in help. In that sense, there are different techniques. But let me say, today’s sintering of the die is mandatory let me say state of art, that you must have in the market. And this comes top and bottom. We have what we call STM. It’s an evaluation that is soldering top, soldering/sintering top metal and bottom metal. So we can sinter the die in both directions, and I think this is essential. This enables you as well to have an isotropic dissipation inside the module. The model needs to come as well with materials, definitely that can operate with junctions up to 200 degree, because again, SiC can extend its range easily above silicon, which is traditionally limited at 175. But the epoxy of the transfer mode needs to come along. The clips are very strategical as well to extract the heat. As I said, they can be sintered. They can the soldered. And the mechanics as well is super important to spread the heat, because the current, and typically these are applications where you have multiple dies and power. The current needs to flow in a very homogeneous way. So you need to be very sure that inside your power module you have a good current sharing. And this happens not only due to the characteristics of the die, but also due to the characteristics of the encapsulation. Something we didn’t say about the die, and this is also important to remember, the die needs to become a bit smarter, in the sense that it needs to include some features, for sure temperature monitoring, but also current monitoring. And this is where you can have inside your model more intelligence. You can monitor not just voltages, temperatures we said, which is very traditional, but also current, to try to understand if there are dangerous difference of temperature. The temperature needs to be homogeneous, and current is of course a very good representation. You can really measure power at that point if you control current and voltage. We have this in our roadmap and implementation of Bare die, and this goes in combination with the power module roadmap now.

PIETRO SCALIA：我們之前說過，為了降低價格、提高性能，我們也在縮減晶片的間距。晶片越來越密集，意味著很多功能都被限制在一個非常有限的空間中。所以我們知道，SiC當然可以抵抗更高的溫度，而且是一個很好的熱阻。 但是我們必須從一個很小的空間中散熱。所以必須藉由封裝的幫忙。從這方面來說，有幾種不同的技術。但是，我認為當今的晶片燒結是強制性的，讓我談談市場上必須擁有的先進技術。這包括頂部和底部。我們有所謂的STM。這是對於焊接頂部、焊接/燒結頂部金屬和底部金屬的評估。所以我們可以在兩個方向上燒結晶片，而且我認為這是必不可少的。這也讓您能在晶片內部實現各向同性耗散。

The model needs to come as well with materials, definitely that can operate with junctions up to 200 degree, because again, SiC can extend its range easily above silicon, which is traditionally limited at 175. But the epoxy of the transfer mode needs to come along. The clips are very strategical as well to extract the heat. As I said, they can be sintered. They can the soldered. And the mechanics as well is super important to spread the heat, because the current, and typically these are applications where you have multiple dies and power. The current needs to flow in a very homogeneous way. So you need to be very sure that inside your power module you have a good current sharing. And this happens not only due to the characteristics of the die, but also due to the characteristics of the encapsulation. Something we didn’t say about the die, and this is also important to remember, the die needs to become a bit smarter, in the sense that it needs to include some features, for sure temperature monitoring, but also current monitoring. And this is where you can have inside your model more intelligence. You can monitor not just voltages, temperatures we said, which is very traditional, but also current, to try to understand if there are dangerous difference of temperature. The temperature needs to be homogeneous, and current is of course a very good representation. You can really measure power at that point if you control current and voltage. We have this in our roadmap and implementation of Bare die, and this goes in combination with the power module roadmap now.

該模型還需要材料，絕對可以在高達200度的接面運行，因為SiC可以輕鬆地將其範圍擴展到傳統上限制在175度的矽之上。但是轉移模式的外延層需要一起來進行。這些夾子也非常具有策略意義，可以提取熱量。正如我所說的，它們可以被燒結，可以焊接。機械對於傳播熱量也非常重要，因為電流，通常這些是您擁有多個晶片和電源的應用。電流需要以非常均勻的方式流動。所以你需要非常確定電源模組內部有良好的電流共享。發生這種情況不僅是由於晶片的特性，還由於封裝的特性。關於磊晶，我們並未提及，但同樣重要的是要記住，晶片需要變得更加智慧，因為它需要包含一些功能，當然還有溫度監控和電流監控。這就是您可以在模型中擁有更多智慧之處。您不僅可以監控我們所說的非常傳統的電壓、溫度，還可以監控電流，嘗試了解是否存在危險的溫差。溫度需要均勻，電流當然是一個很好的代表。如果你控制電流和電壓，真的可以在那個點測量功率。這已在我們的藍圖中並實施了Bare die，而這與電源模組發展藍圖相結合。

MAURIZIO DI PAOLO EMILIO:    In conclusion, Pietro, what are your next projects about SiC? So SiC, but also GaN, can contribute to the creation of a next-generation smart grid to solve energy problems, in particular talking about EVs. So what is the future that is waiting for us? But in particular, in the long term, how do you think SiC-based power devices should evolve to meet the next, more stringent requirements of industry?

MAURIZIO DI PAOLO EMILIO：總而言之，Pietro，關於SiC，您的下一項專案是什麼？還有，SiC和GaN都可以為創建下一代智慧電網做出貢獻，以解決能源問題，尤其是電動車。那麼，等待著我們的未來是什麼？特別是，從長遠來看，您認為基於SiC的功率元件應該如何發展才能滿足下一個更嚴格的產業要求？

PIETRO SCALIA:     I think we have to be honest, right? This decade, it’s still the decade of IGBT and silicon for this high-power application. On the other side, SiC will start dominating already the business starting 2025. Now talking also about wide bandgap, and including GaN, clearly we see that in energy and also into the front end up to 60˚C GaN, we’ll have a very good growth, even if today is far behind the SiC. In the high voltage, so clearly an application with the battery is 800 V. SiC will be the king, and it’s now spreading even at low power, because IGBT does not really pay the big, the ratio of area of IGBT needed for the same power than SiC is even higher than three. And sometimes there is this magic number that procurement did mind the SiC is three times more expensive than IGBT. So IGBT is not paying the bill. So at the end of the story SiC is invading below 150 kW, even in the very low power, and it’s going definitely up to 350 kW that we see in different models. So we have in front of us I think a very exciting second part of the decade. Still to struggle the next couple of years with capacity. Everybody is investing massively. onsemi, again, is basically building the double capacity to Roznov in Czech Republic. We have this year expanded five times Hudson the facility for the bulk of SiC boules. And [Epi], of course, is always important to invest there, because the lead time is quite important. So we are limited in capacity for the next two or three years in the market, but I think starting 2025, SiC will be ubiquitous. And I think in front of us, we have a very nice decade of development of this exciting technology.

PIETRO SCALIA：我想這必須老實說，對嗎？在接下來的十年，對於這種高功率應用來說，仍然是IGBT和矽的十年。另一方面，從2025年開始，SiC將開始主導業務。現在還談到寬能隙，包括GaN，我們清楚地看到，在能源以及高達60˚C的GaN前端，我們將會看到非常好的成長，即使今天遠遠落後於SiC。在高電壓下，很明顯電池的應用是800V。SiC將成為王者，現在即使在低功率下，它也正在普及中，因為IGBT並沒有真正付出巨大的代價，同樣需要IGBT的面積比相同功率的SiC還要高。有時會有一個神奇的數字，採購上確實介意SiC比IGBT貴三倍。所以，IGBT並未買單。而發展到最後，SiC正在侵入150kW以下，即使是在非常低的功率下，它肯定會達到350kW，我們在不同的模型中看到了這一點。所以我們面前擺著我認為十年中非常激動人心的第二部份。未來幾年仍將與產能作鬥爭。每個人都在大量投資。同樣地，onsemi基本上在捷克共和國建設Roznov的雙重產能。今年，我們將Hudson的大部份SiC 晶錠設施擴大了五倍。當然，[Epi]在那裡投資總是很重要的，因為交貨時間非常重要。所以我們在市場上未來兩三年的產能是有限的，但是我認為從2025年開始，SiC將會無處不在。而且我認為在我們面前，在這項令人興奮的技術的發展方面經歷了非常美好的十年。

MAURIZIO DI PAOLO EMILIO:    Wonderful. Thank you so much, Pietro. Thanks a lot for your support at PowerUP. Thank you.

MAURIZIO DI PAOLO EMILIO：太棒了。非常感謝，Pietro。非常感謝您對PowerUP的支持。謝謝你。

PIETRO SCALIA:     Thank you so much, Maurizio. Thanks again to be your guest.

PIETRO SCALIA：非常感謝，Maurizio。再次感謝您的邀請。

MAURIZIO DI PAOLO EMILIO:    Thank you, Pietro. So SiC is invading below 160 kW, even in very low power, and it’s going definitely up to 250 kW that we see in different models. For sure, 200 mm is now in the works, as Pietro said, and onsemi is going to basically start production in 2025. They have developed different generations. They are at the fourth generation, with a lot of possibility of protecting the gate of structure that is one of the key elements of reliability. The level of confidence is increased a lot. As Pietro said, we have very, very few failures, and in front of us, we have a very nice decade of development of this exciting technology.

MAURIZIO DI PAOLO EMILIO：謝謝Pietro。因此，SiC正在湧入160kW以下，即使是在非常低功率，它肯定會達到我們在不同模式中看到的250kW。當然，正如Pietro所說的，200mm如今正在進行中，而基本上，onsemi將在2025年開始量產。他們已經開發了不同世代的產品，如今也來到第四代了，對於可靠性關鍵要素之一的結構閘可望帶來更大的保護能力，同時也提升了信心程度。正如Pietro所說的：「我們很少很少會失敗」，因此，展現在我們面前的，我們將看到這項令人興奮的技術成就極其美好的十年發展。

That brings us to the end of this episode. Stay tuned with more news and technical aspects about power electronics. If you are listening to this on the podcast page at EETimes.com or PowerElectronicsNews.com, links to articles on topics we have discussed are shown on this page. PowerUP is brought to you by AspenCore Media. The host is Maurizio Di Paolo Emilio, and the producer is James Ede. Thank you everyone for listening. See you next episode. Stay tuned.

MAURIZIO DI PAOLO EMILIO：今天的這一集已經來到尾聲了。敬請持續關注有關電力電子的更多新聞和技術。如果您正在eetimes.com或powerelectronicsnews.com的podcast頁面上收聽此內容，那麼，在此頁面還供了與我們討論過的主題相關文章鏈接。PowerUP由AspenCore Media提供給您。主持人是Maurizio Di Paolo Emilio。製作人是James Ead。謝謝大家的聆聽。下集見。敬請關注。

參考原文：The Next Wave of SiC: Manufacturing, Supply Chain and Cost