The Rise and Fall of Silicon Carbide
Not long ago, Musk announced that each of Tesla’s vehicles will reduce the use of silicon carbide chips by 75%. Subsequently, US stocks and A-share silicon carbide concept stocks all collapsed. Wolfspeed, the global leader in silicon carbide, fell 7% that day, and it has been cut in half so far.
It is not uncommon for Musk to bring down an industry in one sentence, but silicon carbide is an exception: this originally inconspicuous industry was precisely the one that Musk brought to life.
In 2018, Musk first announced the use of silicon carbide chips in the Tesla Model 3. Immediately afterwards, BYD, Xiaopeng, and Geely followed suit and replaced silicon carbide chips in the new trams.
Since then, all models using silicon carbide chips have been labeled as “high-end” and “luxury”, and the price has risen all the way to more than 300,000 yuan.
Corresponding to the upstream of the industrial chain, a company that used to make a kind of cheap jewelry called “Moissanite” has transformed into a sexy chip concept stock because the raw material for production is silicon carbide.
Five years later, the god-maker of the year toppled the statue himself. With the end of Tesla’s investor day, all the hustle and bustle seemed to be beaten back to its original shape in an instant.
So, from cheap jewelry to semiconductor materials, how was silicon carbide held up by Musk and fell down again?
How did silicon carbide catch fire?
In 2018, Tesla began to use silicon carbide MOSFETs based on silicon carbide materials in the main drive inverter of Model 3 to replace traditional silicon-based IGBTs.
This decision caused a violent shock at the time – in the process of electric vehicles replacing fuel vehicles, the three-electric system composed of batteries, motors, and electronic controls replaced the engine and gearbox of the previous era of fuel vehicles and became the core of a car. heart. However, since the battery is direct current, the motor needs alternating current to work, which requires an inverter in the electronic control link to convert direct current into alternating current.
Among them, silicon-based IGBT, silicon carbide MOSFET and other power modules are the core technology of an inverter.
Compared with traditional silicon-based IGBTs, silicon carbide is more resistant to high temperature and high pressure, and has higher thermal conductivity. Therefore, the volume of silicon carbide devices only needs to be 1/10 of the volume of silicon devices to achieve the same power conversion requirements.
At the same time, the battery life of a tram equipped with silicon carbide MOSFET is 5% to 10% longer than that of a tram equipped with silicon-based IGBT, the loss is reduced by 75%, and the overall system efficiency is increased by 5%.
In short, silicon carbide is small in size, superior in performance, and strong in energy saving, which alleviates the problem of battery life. It has become the “popular fried chicken” of new energy vehicles in one fell swoop, and car companies have followed suit.
In 2020, BYD launched the first high-end model “Han”, which used silicon carbide power modules for the first time in the motor controller; then, Weilai also launched the ET7 in 2021, which uses silicon carbide materials on the electric drive system. In 2022, Xiaopeng will also launch the G9 model equipped with a silicon carbide electric drive platform.
The price of these models is more than 300,000 yuan, and they have entered the high-end track in the electric car field.
The upsurge of terminals has also sparked a company called CREE on the supply side.
Before silicon carbide enters the application, it has to go through a series of production links of substrate, epitaxy, device design, device manufacturing and packaging and testing, and these links can be split to different companies.
At present, there are very few companies in the world capable of completing all manufacturing links. CREE is one of them and the first company to enter the silicon carbide industry.
In 1991, CREE manufactured the world’s first commercial silicon carbide substrate and applied it in the LED field. Compared with the popularity of personal computers, MP3 players, and later smartphones, this is really not noticeable.
But what is amazing is that, over the past two decades, CREE has achieved a near-monopoly position in this niche market.
When new energy vehicles created a large incremental market, CREE won inexplicably lying down. In 2020, among the conductive silicon carbide required for automobiles, CREE alone will account for more than 60% of the global production capacity, and Keheda will only account for 4% in China.
At the technical level, CREE also maintains a leading position all year round. In wafer manufacturing, the increase in the area of a single substrate is conducive to reducing manufacturing costs. Therefore, it is generally believed in the industry that the larger the substrate size, the more advanced the research and development of the enterprise. At present, the highest level in the industry is an 8-inch silicon carbide substrate.
CREE launched an 8-inch silicon carbide substrate in 2015, and then international leading manufacturers such as Rohm and STMicroelectronics also successfully developed it in 2021. In contrast, the production capacity of the most leading manufacturers in China, such as Tianke Heda and Tianyue Advanced, is still at the 6-inch level.
Another technical difficulty lies in the yield rate of silicon carbide: in the crystal growth process of obtaining silicon carbide substrates, it only takes 3 days for traditional silicon materials to grow into a crystal rod, while it takes 7 days for silicon carbide crystal rods. Resulting in lower silicon carbide production efficiency.
Moreover, the brittle and hard properties of silicon carbide are very easy to chip when cutting wafers and dicing. Judging from the current production process, it is not easy to produce qualified, semiconductor-grade silicon carbide wafers, and the production capacity is very scarce.
With the dual blessing of a complete industrial chain + one step ahead of technology, CREE has become a popular target in the capital market. In 2020, CREE’s market value rose by 129%, ranking second in the world among major semiconductor companies during the same period, leaving Nvidia and AMD behind.
With Tesla announcing the use of silicon carbide materials, CREE has also become more focused, successively sold off the lighting business and LED business that were once regarded as the core, and changed the name of the entire company: Wolfspeed, literally translated as “wolf speed” in Chinese.
As a result, since CREE changed its name, “Wolf Speed” has become “Dog Crawl”.
On November 12, 2021, one month after the name change, Wolfspeed’s market value reached its peak, and it is only half of what it was then.
From the peak to the bottom, Musk naturally has to take responsibility. But the question is, why did silicon carbide change from a darling to an outcast in a short period of time?
Tesla: God-maker and God-destroyer
The latest plunge in Wolfspeed and silicon carbide concept stocks was due to Musk’s announcement at Tesla’s investor day in early March that bicycles would use 75% less silicon carbide.
But the market seems to have misunderstood the point: Tesla is reducing use, not not using it.
Moreover, there are many secrets behind this: despite years of development, the long crystal growth time and hard and brittle properties of silicon carbide have not been resolved. As a result, even Wolfspeed, the industry leader, has a comprehensive yield rate of only 60%, and the yield rate of domestic substrate manufacturers is even lower than 40%.
The direct result of low yield is high cost: Taking Model 3 as an example, the total cost of 48 silicon carbide MOSFETs used in its main drive inverter is 5,000 yuan, which is 3 to 5 times that of the traditional silicon-based IGBT.
For Musk, the cost-cutting madman, this is obviously intolerable. Coincidentally, at the beginning of this year, Tesla had just engaged in a wave of crazy price cuts. Musk took advantage of the situation and set a goal: to reduce the price without reducing the volume, Tesla’s performance remains the same, but the total manufacturing cost will be reduced by another $1,000.
As the core cost source of automotive electronic control, silicon carbide is naturally on the blacklist.
Musk actually has another account in mind. According to his vision: by 2030, Tesla will reach an annual output of 20 million electric vehicles, and currently every 2 Teslas use a piece of 6-inch silicon carbide. However, it is optimistically predicted that even if the annual compound growth rate is 34%, by 2030, the total production capacity of silicon carbide will only be 10.45 million to 12.18 million pieces.
In other words, if the original usage is maintained, the Tesla family will use up all the global production of silicon carbide, which is obviously impossible.
The price is unaffordable, the output cannot keep up, and Wolfspeed, which keeps dropping the chain, can only force Tesla to find another way.
On the day Tesla announced the reduction of silicon carbide, its vice president of power system engineering pointed out that Tesla has actually developed a customized module packaging technology that can maintain the same heat dissipation effect while using less silicon carbide. .
In other words, Tesla is not no longer used, but used less. And those car companies that have not mastered the “customized module packaging technology” for the time being, can only dredge the relationship with Wolfspeed in the short term, and actively look for spare tires at the same time.
For example, BYD invested in Tianyu Semiconductor and Tianke Heda, and Xiaopeng invested in Zhanxin Electronics. In 2018, the amount of domestic silicon carbide-related investment projects signed was only 5 billion yuan, and by 2020 it had reached 46.3 billion yuan, of which more than 90% were government-participated.
However, under the background of continuous price reduction in the automobile market, the domestically-produced automobiles, which are already weak in profits, can persist in their belief in silicon carbide, which is also put on a question mark.
Silicon carbide itself is nothing new, having been a material for more than 100 years. The development of technology in the commercial field has also gone through more than 30 years.
In 1884, an employee named Edward Acheson in Edison’s laboratory wanted to make “artificial diamonds” after leaving his job to start a business. As a result, a substance whose hardness was only a little less than that of diamonds and whose fire color was even brighter than diamonds was accidentally found in the reactants of a furnace in an experiment, thus opening the door to the artificial synthesis of “silicon carbide”.
More than ten years later, another scientist named Henry Mothan discovered natural silicon carbide in meteorites, and even won the 1904 Nobel Prize in Chemistry for this. Silicon carbide has therefore been given a more famous name: moissanite.
Therefore, a cold knowledge is: In fact, many companies that have become popular under the name of “silicon carbide” have been suppliers of cheap jewelry before Tesla uses silicon carbide.
For example, in January 2022, Tianyue Advanced, the first share of silicon carbide, landed on the Science and Technology Innovation Board. Until 2020, three of Tianyue Advanced’s top five clients are jewelry companies. At present, Tianke Heda, which has the highest production of silicon carbide substrates in China, is preparing to go public in 2020. In the list of major customers announced, there are also two jewelry companies that have occupied the top 5 for a long time. One of them also became Tianke Heda in 2017. largest client.
Of course, there is a more “sexy” name for silicon carbide in the capital market-third-generation semiconductors.
The reason is that silicon carbide can be used as the wafer substrate of the chip, and has the characteristics of larger forbidden band width, larger electron drift rate, larger thermal conductivity, and larger breakdown field strength in terms of physical characteristics. That is, it can adapt to harsh conditions such as high power, high frequency, high temperature, and high voltage. Therefore, the advantages are particularly obvious in scenarios such as automotive chips and high-voltage fast charging.
Corresponding to it are first-generation semiconductors and second-generation semiconductors. Among them, the first-generation semiconductor substrates are generally silicon wafers, which can be used to produce traditional CPUs, GPUs, MCUs, etc.
The substrates of the second-generation semiconductors are generally indium phosphide and gallium arsenide, which are mainly used to produce radio frequency devices, optical modules, LEDs, lasers, detectors, sensors and other devices.
It is also a chip, and it is related to Tesla and new energy vehicles. It even attracted Zhang Rujing to end in person. The hype of silicon carbide in the capital market was once praised to the height of “domestic substitution” and “curve overtaking”.
For manufacturers, as long as the silicon rods of silicon carbide in the crystal pulling process are made large enough, they can transform from the favorite moissanite supplier in Yiwu Commodity City to a third-generation semiconductor supplier.
But it is not easy to pull the ingot large enough.
The difficulty of silicon carbide production lies in: the production process is carried out in a black box with a high temperature of 2300 degrees Celsius (silicon material only needs 1600 degrees Celsius), and as the size expands, the difficulty of its production increases geometrically: there is a slight error in the control of temperature and pressure , it may lead to errors in a series of parameters such as micropipe density, dislocation density, resistivity, warpage, and surface roughness of silicon carbide materials.
Once an error occurs, this broken and deformed silicon carbide wafer will be downgraded to Moissanite.
It is precisely because of the extremely high fragmentation rate during the leap from Moissanite drilling to the third-generation semiconductor that if you want to invest in semiconductor-grade silicon carbide, the results are often very slow.
Wolfspeed developed an 8-inch silicon carbide substrate as early as 2015, but it was not until 2022, when it iterated to the third-generation wafer, that it was able to enter mass production. However, domestic companies will only start to enter 8-inch research and development in 2020.
Although “8 inches” has been shouted in the news and news issued by various factories for many years in the financial report, the mainstream substrates in the international market are still at the level of 6 inches, and the mainstream products in China are still at the level of 4 inches. substrate.
Even Wolfspeed itself is still selling substandard silicon carbide as moissanite to jewelry companies. Earlier in 2016, CREE once planned to sell the third-generation semiconductor department that developed silicon carbide to Infineon.
As a result, in 2017, CREE saw that Wolfspeed’s business was gradually improving, and its attitude turned 180 degrees: not only did it not sell it, but also shifted the focus of the entire company to the semiconductor department, and in the following years, it successively transferred the former LED and lighting Business for sale.
In 2018, CREE officially repented and announced that it refused to sell its semiconductor business to Infineon, and even turned itself into Party A and bought Infineon’s RF business.
We have told such a story.
In 2013, in order to replace Corning’s glass screen with sapphire, with the support of Apple, an American company called GT Advanced Technologies spent a total of 900 million US dollars to build a sapphire factory in Arizona.
According to the contract, Apple will prepay 578 million U.S. dollars in advance to Jite Advanced for production expansion and related equipment investment.
It is precisely because of Apple’s support that the company was extremely advanced that year. Even with financial losses, the stock price has gone out of a beautiful upward curve.
The similar warmth started, but ushered in a different ending: just one year later, because of the high cost and fragility of sapphire, Apple returned to the embrace of Corning and grew “with tears” for a whole year, once again setting a new record since its establishment. historical stock price high.
In order to repay the debt, Ji Te Advanced, who aggressively expanded production, decided to sell the furnace to repay the debt, and finally ended with the story of layoffs and bankruptcy.
Whether it is the extremely advanced bankruptcy or the ups and downs of Wolfspeed, it essentially reflects a problem: a high-volume end product often has a huge say in the upstream supply chain, and can even determine the life and death of supply chain companies.
This is actually another reason why China spares no effort to cultivate the new energy vehicle market – a new energy vehicle company with a market value of 100 billion yuan is likely to drive dozens of supply chain companies with a market value of 10 billion yuan, and behind them may be more jobs. , profits and taxes and high-income jobs.
From this perspective, the ebb and flow of silicon carbide is just a microcosm of the endless smoke in the supply chain behind new energy vehicles. The competition and pursuit of the right to speak and added value in this market are far from over.