On November 15, Beijing Tianke Heda, invested by Huawei, announced the successful development of "8-inch silicon carbide substrate", which is known as the most technically difficult substrate product in the third-generation semiconductor industry. Previously, only Infineon, A few companies such as Wolfspeed and STMicroelectronics can master it.
This is encouraging news. After the U.S. Department of Commerce’s Bureau of Industry and Security (BIS) announced targeted updates to export controls, the emergence of domestic alternatives in the field of third-generation semiconductors will undoubtedly give China’s semiconductor industry a shot in the arm. .
If I tell you that the chips used in everything from smartphones to supercomputers belong to the category of "first-generation semiconductors", then you will most likely be surprised by the development level of the domestic semiconductor industry—"The third-generation semiconductors are so powerful What's it like?"
But if this news is shared with the international third-generation semiconductor giants, they may be confused and have no idea what third-generation semiconductors are.
Because there is no such concept internationally, the standard name of the so-called "third-generation semiconductor" is "wide bandgap semiconductor". A senior practitioner in the industry told Tiger Sniff that "third-generation semiconductor" was originally a term created in China to facilitate the government's investment leaders to remember.
"There is no iterative relationship between the third-generation semiconductor and the first and second-generation semiconductors, and its application scenarios are completely incomparable with the first-generation semiconductor." The practitioner said.
Of course, this does not mean that the third-generation semiconductor has no development prospects. In fact, the third-generation semiconductors have seen a larger market demand than before, especially after the outbreak of the new energy vehicle industry.
It’s just that under this name, many people think that it can solve the problem of being “stuck” in China’s chip industry, and it is even regarded as an opportunity for China to overtake in a corner, because the third-generation semiconductor does not need to use advanced manufacturing processes—— This can get rid of the unilateral sanctions imposed by the US government.
But can this third-generation semiconductor, which has attracted capital and local governments, really help China's chip industry achieve "curve overtaking"?
A "misunderstood" name
Before answering this question, we need to clarify how the "iteration" of semiconductors is defined in China.
First of all, a concept needs to be clarified. The so-called generations of semiconductors are distinguished by the substrate material. We come into contact with chips of various shapes in our daily life, which are packaged after wafer cutting, and wafer preparation includes two major process steps of substrate and epitaxy. It can be simply understood that the substrate is the basis for processing all chips.
Traditionally, silicon-based semiconductors are regarded as the first generation of semiconductors. At present, more than 95% of the chips used in human society are still monocrystalline silicon. Due to the low price of materials and a very mature industrial chain, the growth of silicon-based semiconductors in the short term The proportion will not change much.
The second-generation semiconductors are represented by gallium arsenide and indium phosphide materials. These compound semiconductors have high frequency and radiation resistance characteristics, so they are used in national defense, aerospace, satellite communications and other fields.
As for the third-generation semiconductor mentioned in this article, its standard name should be "wide bandgap semiconductor". This type of material is represented by silicon carbide and gallium nitride, which have a higher bandgap width and are suitable for high temperature, high pressure, and high temperature. power scene. Compared with the cheap price of monocrystalline silicon, the substrates used in third-generation semiconductors are expensive, and even account for nearly 70% of the cost.
The bandgap width is a factor that determines performance such as "high voltage resistance, on-state resistance, thermal conductivity, high temperature resistance, and radiation resistance". Before the emergence of the new energy vehicle industry, the main application scenarios of the third-generation semiconductors were concentrated in 5G base stations, high-speed rail, photovoltaic inverters and other fields.
Its greatest advantage lies in energy saving. Public information shows that the application of third-generation semiconductor devices on high-speed rail can save energy by 20% while reducing the volume of the power system; in the field of photovoltaic inverters, it can reduce photoelectric conversion losses by more than 25%; intelligent The power grid field can increase power supply efficiency by more than 40% and reduce power loss by 60%.
If you look at it this way, third-generation semiconductors don't seem to have much to do with our daily lives.
Indeed, except for new energy vehicles, in most cases, the application scenarios of third-generation semiconductors are difficult to be perceived. At present, the third-generation semiconductors do not have the ability to replace silicon-based semiconductors at all.
Take the smartphone we use every day as an example. Its core is a processor chip and a coprocessor chip, both of which are logic chips in classification. As the name suggests, they are chips responsible for calculation and logic judgment. The requirements are very high, such as the "5nm chip" and "7nm chip" that we often hear.
However, as far as the development of the third-generation semiconductor is concerned, most of them are still in the "micron-level" process technology, and its characteristics are more suitable for power devices, that is, the core of electric energy conversion and circuit control in electronic devices, such as The simplest diode.
"The third-generation semiconductors are not suitable for making logic chips at all." Liao Qibo, chairman of Green Energy Xinchuang, a silicon carbide industry company, told Tiger Sniff, "Silicon carbide and gallium nitride will not replace the silicon base, and each has its own characteristics."
Therefore, the so-called relying on third-generation semiconductors to solve the problem of "stuck necks" in advanced manufacturing processes is completely nonsense, and it is not an opportunity to overtake on a curve.
But does this mean that third-generation semiconductors are useless?
Of course not, because the third-generation semiconductor material has the characteristics of high temperature resistance and good thermal conductivity, it is regarded as the best substitute for power devices on new energy vehicles.
Tesla is the "first to eat crab" company on this application.
CITIC Securities once mentioned in a research report on the dismantling of Tesla Model 3 that Model 3 uses 48 SiC MOSFETs instead of 84 IGBTs (insulated gate bipolar transistors), which greatly reduces the size and power consumption, making it the world’s largest The car has increased range.
Model 3 is the first electric vehicle to use SiC MOSFETs in the motor controller
This milestone innovation quickly spread among new energy manufacturers and ignited China's third-generation semiconductor industry.
"Tesla has boosted the development of China's third-generation semiconductors in new energy vehicles." An investor who has been investing in semiconductors for many years at home and abroad said.
Capital pours in and blossoms everywhere
"Before 2019, there was no problem of overheating investment in the domestic third-generation semiconductor industry, and it could even be said to be very miserable." A third-generation semiconductor entrepreneur said.
In fact, China's exploration of third-generation semiconductor materials is not too late. Internationally, the earliest commercial silicon carbide devices were born in the 1990s. The aforementioned investor said that decades ago, when he was working at a leading foreign semiconductor equipment manufacturer, they were already looking at third-generation semiconductor projects. They did not invest at that time because the market was too small. This point has been confirmed in the story of Tianyue Xianjin.
At the beginning of the 21st century, Academician Jiang Minhua of the Crystal Research Institute of Shandong University established a silicon carbide research group to start tackling key problems, and successfully developed the single crystal furnace required for silicon carbide materials, and mastered the processing technology of silicon carbide, realizing the process from single crystal All localization tests of growth furnace manufacturing, single crystal growth, substrate processing and application.
In May 2011, Academician Jiang Minhua passed away. Since there was no mature market for silicon carbide semiconductors at this time, few companies were willing to undertake this project, and the subject was temporarily suspended. At this time, Zong Yanmin, who is engaged in engineering equipment, learned of this project, and decisively purchased the silicon carbide material technology of Shandong University, established Shandong Tianyue Advanced Technology Co., Ltd., and started the industrialized research and development of silicon carbide material technology.
Almost at the same time, Beijing Tyco Tianrun Semiconductor also started the development of the third-generation semiconductor, and started the construction of the production line in the following year.
The most notable feature of the domestic third-generation semiconductor industry at this time is that it is almost entirely driven by private capital, with little official involvement.
To a large extent, as mentioned above, local governments are too unfamiliar with the concepts of "silicon carbide" and "gallium nitride", which drove the entrepreneurs at that time to replace "wide bandgap semiconductor" with "third "Generation Semiconductor" is an ambiguous but easy-to-understand name.
In 2016, entrepreneurs who were weak and weak ushered in the first ray of dawn. The National New Industry Development Group of the State Council has listed the tertiary semiconductor industry as a development priority, and domestic enterprises have expanded investment in tertiary semiconductor research and development projects, and the industry has officially entered a period of rapid development. Tiger Sniff, according to public statistics, has 11 third-generation semiconductor projects launched in 2016, with an average investment of over 100 million yuan.
The release of Tesla Model 3 has also pointed out the direction for other new energy car manufacturers: the original third-generation semiconductor is used in this way. On Model3, Tesla switched all silicon-based power devices to silicon carbide devices.
"Before Tesla, no car factory thought of using silicon carbide as a substitute." Liao Qibo told Huxiu that although the cost of replacing the silicon-based power devices of the entire car with silicon carbide devices is about 300 US dollars more expensive, the use of After silicon carbide devices can simplify the water cooling system, the total cost of the vehicle will be reduced instead. It is a more cost-effective business for the car factory.
Most importantly, no car manufacturer can resist the temptation to increase the cruising range by 5%-10%.
With the official endorsement and the incremental market, the domestic third-generation semiconductor industry at this time has actually completed the transition from the "introduction period" to the "growth period". Everything is ready and only owes the injection of industrial capital.
This work was led by Huawei. In August 2019, in Shandong Tianyue Advanced’s first financing list, the name of Hubble Investment, a subsidiary of Huawei, was impressively listed. From the perspective of the time, Huawei was brewing a huge car-making plan at that time.
In order not to repeat the mistakes of HiSilicon, Huawei chose to cultivate the supply chain from the source. In the ensuing 18 months, Hubble Investment successively injected capital into seven third-generation semiconductor companies, including not only substrate companies such as Tianyue Advanced, but also Tianyu Semiconductor, which is engaged in the production of epitaxial wafers, and provides third-generation semiconductor-related equipment. Suzhou Jingtuo.
The impact that Huawei has brought to Tianyue Advanced is immediate. Before Huawei’s financing, Tianyue Advanced’s valuation was 1 billion yuan, but just one year later, the company’s valuation soared to 10.095 billion yuan.
The bigger impact is that Huawei has accelerated the investment progress of the entire industry. Statistics from CASA Research, a third-party research organization, show that in 2018, there were four investments in mainland China involving third-generation semiconductors, involving a financial value of 6 billion. By 2019, there were 14 investment events, involving an amount of 22.08 billion yuan.
It was also during this period that domestic new energy vehicle manufacturers began to follow Tesla's example and use third-generation semiconductor devices in new vehicles. In 2020, the BYD Han EV model will roll off the production line. The vehicle is equipped with a silicon carbide MOSFET module, and the acceleration performance and battery life are significantly improved; in 2021, BYD Tang EV will be equipped with a silicon carbide electronic control system; The second-generation high-efficiency electric drive platform for silicon power modules.
As a result, the third-generation semiconductor industry ushered in a real market explosion stage. According to the "White Paper on Application Development and Investment Value of Third-Generation Semiconductors Under the 2020 "New Infrastructure" Trend", the market size of my country's third-generation semiconductors in 2019 was 9.415 billion yuan, and it is expected that the market size will reach 62.342 billion yuan in 2025.
Local governments are even more aggressive in introducing third-generation semiconductor projects, which are blooming everywhere. According to the analysis of articles by Xinmou Research, a third-party analysis organization, in 2020, there will be dozens of third-generation semiconductor-related policies issued in various places in my country, covering more than ten provinces (including municipalities directly under the Central Government).
However, under a thriving scene, domestic third-generation semiconductor companies do not seem to be really accepted by the market, and the bright moments brought about by massive investment do not seem to last too long.
Chicken feathers after the madness
"Hubble Investment has taken the initiative to inspect our projects, but these projects are still in the project approval stage, and we don't know how to talk to Hubble people." An industry insider told Tiger Sniff that in 2021, Hubble Investment will have a nationwide Collecting third-generation semiconductor projects within a wide range has led many silicon-based semiconductor companies to start planning to launch silicon carbide or gallium nitride projects.
However, Huawei's enthusiasm for third-generation semiconductors came to an abrupt end in early 2022.
In February 2022, Hubble invested in Tedis Semiconductor, which is mainly engaged in the R&D and manufacturing of compound semiconductor equipment. Since then, Huawei has never been involved in the third-generation semiconductor industry.
A researcher engaged in the semiconductor industry told Tiger Sniff that the current level of development of the domestic third-generation semiconductor industry may not be able to meet Huawei's car-making needs, so Huawei has temporarily stopped investing in the industry.
Here we can use BYD data as a reference. In an exchange summary in June this year, BYD disclosed the current procurement situation of the company's silicon carbide devices: in the high-power single-tube products that are mainly used in vehicle OBC (chargers), BYD is manufactured by Taiwan Hanlei, with a self-supply rate of 65%-70%, and the rest is purchased by Shenzhen Esco and other manufacturers.
In the procurement of silicon carbide module chips, 70% come from Bosch, 20% from STMicroelectronics, and a few from Cree (Wolfspeed before the name change), and none of the domestic manufacturers were selected.
The reason is that at present, the cost of silicon carbide devices has more advantages abroad. According to BYD, leading manufacturers such as Cree can achieve a yield rate of more than 75%, and the domestic yield rate is about 55%-60%. Tiger Sniff asked industry insiders for confirmation on this figure, and the other party said that "under ideal conditions, the yield rate can reach 50%, and only a small number of domestic manufacturers' products can meet the stability requirements of the car specification level."
Zhao Zhanxiang, partner and chief technology officer of Yunxiu Capital, said that because many manufacturing, device, packaging, materials and other processes on the third-generation semiconductor have not been used on a large scale, there are still many immature areas, and the second is in specific market applications. In fact, it has not been widely used in various market segments, and it will take a lot of time to stabilize them in terms of production technology and market applications.
The third-generation semiconductor industry belongs to the technology-oriented industry. In terms of design difficulty, the technical content is not high, and the difficulty lies in the technology.
Due to the special materials of the third-generation semiconductors, taking silicon carbide as an example, its Mohs hardness (a mineral hardness standard) can reach level 9.5, second only to diamond (level 10), and the processing difficulty is more difficult than silicon-based semiconductors. many.
Liao Qibo told Huxiu that on the chip side of the silicon carbide industry, although 60% to 70% of the process flow is the same as that of silicon-based semiconductors, the processing conditions of the core process are completely different: for example, silicon-based chips can be processed at room temperature. Implantation, silicon carbide material requires high temperature in this process.
Moreover, the hardness and brittleness of the material are often positively correlated, and the extremely high hardness of the silicon carbide material also puts it at risk of fragmentation and curling during processing, so the yield rate is far inferior to that of silicon-based semiconductors.
