This article analyzes the strategic significance of third-generation semiconductors and discusses the development status of China’s technology and industrialization capabilities in related fields. It elaborates that “large size and cost reduction” is the current focus of development for silicon carbide (SiC) and gallium nitride (GaN) technologies, and explores the development models of enterprises in the third-generation semiconductor industry, along with potential problems and risks. Although China has a good foundation, there are still shortcomings. It is recommended to achieve development driven by applications under the guidance of national policies, increase continuous support for production lines, systematically enrich product forms, promote the high-quality development of the third-generation semiconductor industry, and seize new opportunities for future applications.
In the early 1980s, the third-generation semiconductors began to emerge, achieving significant breakthroughs in compound lighting, and have now formed a trillion-level market scale globally. In the past three years, the development of third-generation semiconductors has been somewhat hindered due to the COVID-19 pandemic, but the global scale continues to grow at a compound annual growth rate of about 10%. With the emergence of innovative technologies such as deep ultraviolet light-emitting diodes (LEDs), Mini-LED, and Micro-LED, third-generation semiconductors are opening up new application scenarios in optoelectronics, such as new displays, smart agriculture, and healthcare, which will further expand market scale.
In 1993, the first gallium nitride (GaN) high electron mobility transistor (HEMT) device with microwave characteristics was publicly reported, and the third-generation semiconductors quickly entered the research and application fields of microwave radio frequency, especially GaN radio frequency devices, which, with their unique characteristics of high power, high efficiency, high linearity, high operating voltage, and radiation resistance, have become ideal substitutes for silicon (Si), gallium arsenide (GaAs), and other devices, playing an important role in military equipment, aerospace, and fifth-generation mobile communication (5G) technologies, and showing broad development prospects.
In the early 21st century, the United States was the first to apply silicon carbide (SiC) to equipment, represented by S-band solid-state microwave radio frequency devices. Although it was gradually replaced by GaN, its high voltage and high frequency characteristics have gained favor in the field of power electronics. It is gradually becoming a substitute for Si power electronic devices, especially after 2002, when SiC single crystal substrate technology developed rapidly, and manufacturing costs were significantly reduced. It is expected to play a major role in new energy vehicles, high-speed rail, and smart grids, supporting a market scale of over 10 trillion yuan.
The third-generation semiconductors have experienced over 20 years of development in the fields of optoelectronics, radio frequency electronics, and power electronics, supporting a trillion-level market scale, and continuously emerging new application scenarios that stimulate new development potential.
Massive Application Scenarios in the Market
5G Shows “China Speed”, GaN RF Devices Have Broad Space
5G is currently a representative and leading network information technology, characterized by high speed, ubiquitous networks, low power consumption, low latency, and high reliability, which will achieve ubiquitous interconnection of everything and deep human-machine interaction, penetrating into various industries and fields of the economy and society, becoming a key new infrastructure supporting the digital, networked, and intelligent transformation of the economy and society.
5G base stations have higher requirements for radio frequency devices. Traditional laterally diffused metal oxide semiconductor (LDMOS) cannot adapt to the high frequencies of 5G, while GaN adapts to a frequency range extending to 40GHz or even higher, meeting the high-frequency requirements of 5G; GaN features soft compression characteristics, making it easier to pre-distort and linearize, achieving higher efficiency; GaN can achieve higher power density, reaching about 4 times that of LDMOS devices; GaN packaging size is only 1/4 to 1/7 that of LDMOS, making GaN RF devices more suitable for 5G base stations. In 2010, GaN-based high-power microwave amplifier devices were first applied to high-end base station equipment with small size and high linearity, starting to enter the mobile communication market. With the comprehensive rollout of fourth-generation mobile communication (4G) wireless network infrastructure, GaN applications have significantly increased, while the market share of Si-based LDMOS devices above 2GHz has dropped from 92% to 76%. The launch of 5G has made GaN microwave power amplifiers more accepted, and at high frequencies, only GaN-based HEMT devices can be relied upon. Currently, the microwave RF technology of GaN-based HEMT has basically achieved a significant leap compared to previous generations of semiconductors (Si-based LDMOS, GaAs/InP-based pHEMT, etc.).
As the layout and promotion of 5G construction progresses, China’s 5G frequency bands have gradually expanded from 4.9, 3.5, and 2.6GHz to 2.1GHz, 700MHz, and the latest 900MHz. The structure of base stations has returned from the original large-scale dense massive multiple-input multiple-output (MIMO) antenna arrays to traditional MIMO structures, and the number of receiving and transmitting channels in base stations has decreased from the original 64 and 32 channels to 8 and 4 channels.At the same time, new requirements have been raised for GaN RF devices, requiring the output power of devices to increase from the original 100W level to 500~700W or even higher power levels. The changes in frequency, channel count, and power require innovative research and development in materials, design, processes, and packaging to meet the new base station research and production needs.
According to a report issued by the Ministry of Industry and Information Technology of the People’s Republic of China (MIIT), as of the end of May 2023, the number of 5G base stations in China has reached 2.844 million, and the number of mobile IoT terminal users has exceeded 2.05 billion. China is the first country in the world to establish a 5G SA (standalone networking) network, maximizing the end-to-end network slicing, massive connectivity, and ultra-reliable 5G characteristics. As the first of the new infrastructure, investment from 2019 to 2022 has risen year by year to 401.6 billion yuan. By 2025, the cumulative investment in China’s 5G network construction is expected to reach 1.2 trillion yuan, driving related investments exceeding 3.5 trillion yuan. According to the development characteristics of mobile communication’s “ten-year generation”, the development of 5G progresses from shallow to deep, and 5G technology continues to evolve, with 5.5G expected to enter the commercial stage in 2024.New standards will achieve enhanced capabilities, extended boundaries, and improved efficiency through the construction of six core pillars: spectrum utilization, native artificial intelligence, uplink enhancement, focus on industries, smart management, and green low-carbon, promoting the comprehensive penetration of 5G industry applications.With the saturation of low-frequency applications and the demand for wider bandwidth and greater communication capacity, 5G communication will inevitably evolve towards the millimeter-wave frequency band, and 6G mobile communication may even raise the frequency to terahertz.The product form of traditional RF devices can no longer meet the new requirements, requiring the development of millimeter-wave monolithic integrated power amplifier chips, integrating amplification, switching, and low-noise amplification into multifunctional chips for transceivers.GaN terahertz devices have advantages in the terahertz field due to their larger effective mass of electrons, higher longitudinal phonon energy, faster interband electron scattering, larger peak-to-valley current ratio in the negative resistance region, and higher two-dimensional electron gas density.Of course, when the frequency enters the terahertz frequency band, traditional transceiver modules using thick film processes will not meet the requirements for base station miniaturization, high efficiency, and high integration. Innovative research and development is needed to utilize microelectronic processes for three-dimensional packaging technology, achieving a leap from traditional two-dimensional integration to three-dimensional integration, and developing multifunctional packaging devices that integrate functions such as antennas, transceivers, control, and analog-to-digital conversion, reducing the volume and mass of devices by more than one order of magnitude and enhancing functionality by more than one order of magnitude.
The “Dual Carbon” Strategy Accelerates New Applications of SiC Power Electronic Devices
In September 2020, China clearly proposed the goals of “carbon peak” by 2030 and “carbon neutrality” by 2060. The third-generation semiconductors are known as green semiconductors. SiC power electronic devices have high breakdown voltage, high efficiency, and high frequency characteristics, with energy-saving capabilities four times that of silicon devices, making them core devices supporting the “dual carbon” strategy.Chips made from them have broad application prospects in scenarios such as new energy vehicles, photovoltaic inverters, rail transit, and smart grids.
In the field of new energy vehicles, low-voltage direct current/direct current conversion requires Schottky barrier diode (SBD) devices with voltages of 650V and below; onboard chargers (OBC) require SiC metal-oxide-semiconductor field-effect transistors (MOSFETs) below 1700V, with on-resistance levels of 25, 40, 80, and 160mΩ; in the main drive, chips with a voltage of 1200V and an on-resistance of less than 15mΩ, with a current above 100A, are developed for power modules of 400, 600, and 800A levels to meet different range requirements.At the same time, high-voltage architecture is a necessary path to achieve high-power fast charging. With the popularization of 800V fast charging systems, new SiC power electronic device products need to be developed to meet the changing requirements of voltage, current, and on-resistance, with applications in new energy vehicles evolving towards higher voltages, larger currents, and lower on-resistance.Using SiC power devices can improve the energy utilization efficiency of batteries due to their high conversion efficiency. For example, in systems equipped with 1200V SiC MOSFETs, inverter energy consumption can be reduced by more than 60%, while reducing overall vehicle energy consumption and achieving lower battery capacity requirements.Moreover, due to their large power density and high frequency, they can reduce the size and mass of power conversion modules, and their stronger tolerance to high temperatures saves heat dissipation components, achieving vehicle lightweighting.For instance, a bidirectional OBC using all-SiC MOSFETs can achieve more than double the switching frequency compared to silicon solutions, and the number of power devices and gate drivers is reduced by over 30%, improving system lightweighting and overall operational efficiency.In summary, new energy vehicles driven by SiC power devices can significantly reduce energy losses, achieving a range increase of 5% to 10% under the same battery capacity.According to the “Implementation Plan for Carbon Peak in the Industrial Field” issued by the Ministry of Industry and Information Technology and other three ministries, by 2030, the proportion of new energy and clean energy-powered transportation tools will reach about 40%, providing a broad market for the application development of SiC power electronic devices.
In the rail transit field, both domestic and international attention has been focused on the application research of SiC devices in traction converter systems, with some institutions already commercializing products and installing them on rail trains. Mitsubishi Electric has developed a 3.3kV all-SiC power module suitable for traction systems of rail trains, which can save about 30% of power compared to existing systems when applied to the main circuit system of rail train inverters. Recently, SiC has made new progress in the rail transit field, with Chengdu Metro and Xi’an Metro successively applying SiC-based converters and permanent magnet synchronous motor traction systems. Currently, China’s urban rail transit operating lines have broken through the 10,000 km mark, and the corresponding mileage is expected to reach 15,000 km by the end of the 14th Five-Year Plan, with the national railway operating mileage reaching 165,000 km, including 50,000 km of high-speed rail. This scenario highlights the advantages of medium-voltage SiC power electronic devices, which will gradually replace Si-based power electronic devices, representing medium-term market demand.
In the green energy field, the direct current generated by photovoltaic power generation needs to be inverted into alternating current to be integrated into the grid, which requires the participation of power devices in the energy conversion process. Compared to Si-based IGBTs, SiC power modules can reduce switching losses by 85%, and using SiC power devices can directly enhance the efficiency of energy conversion. According to estimates by the International Energy Agency (IEA), by 2024, if only 2% of distributed solar photovoltaic systems deploy SiC, the additional power generation could reach up to 10GW. China is a major power consumer, with the total electricity consumption in 2022 being approximately 8.6 trillion kilowatt-hours. The demand for smart grids requires third-generation semiconductor power electronic devices at the kilovolt-ampere level, with plans for commercial use by 2035, and the market demand is broader than that of new energy vehicles. The breakdown voltage and current requirements for SiC power electronic devices in this scenario are more than ten times that of new energy vehicle devices, and there are essential differences in their research and production. Innovative research and development are likewise required in materials, design, processes, and packaging, which is a long-term market demand for SiC power electronic devices.
Military Demand Accelerates New Technology Research and Application
In the global semiconductor field, the demand for new materials, devices, and processes in military equipment is an important driving force for the development of the semiconductor field. The third-generation semiconductors are key cores supporting national defense construction and are crucial elements for mastering the initiative in future warfare.The excellent performance of third-generation semiconductors can significantly reduce the size of radars, communication equipment, and guidance heads. At the same time, they can greatly enhance combat effectiveness, playing a very important supporting role in improving the unmanned, intelligent, and information levels of equipment, and have become the focus of competition in national defense science and technology among various countries.For example, after adopting third-generation semiconductor devices, radars can achieve significant improvements in detection distance and accuracy without increasing size and weight, enabling them to detect and lock onto stealth targets; through massive combined power, they can directly burn out enemy electronic devices, achieving hard kill in electronic warfare; special combat teams can achieve secure communication under radio silence conditions.The United States has widely used third-generation semiconductors in next-generation electronic jammers, long-range identification radars, guided missiles, and all-electric ship integrated power systems.
In fact, the initial industrial foundation development plans of major countries globally have shown significant military tendencies and application demands. Currently, GaN-based HEMT microwave RF technology has basically achieved a significant leap compared to previous generations of semiconductors. Key manufacturers of GaN-based semiconductor RF devices globally include Cree (now Wolfspeed), Qorvo, MACOM, and Raytheon from the United States, as well as Infineon from Germany, GaN Systems from Canada, Mitsubishi Electric from Japan, and NXP from the Netherlands. In terms of manufacturing maturity, Raytheon and Qorvo’s GaN products have reached the highest level of manufacturing maturity assessment by their Department of Defense, and the manufacturing process of GaN RF devices has met the target requirements for optimal performance, cost, and capacity, and has the capability to support full-speed production. In 2014, Raytheon announced the deployment of advanced radar using GaN modules in the “Patriot” air defense system; in 2021, it licensed its GaN-on-Si technology to GlobalFoundries to jointly develop IC processes that can handle 5G and 6G millimeter-wave signals, raising the mass production level of GaN-based RF devices to a new level and further compressing RF costs.
With new working scenarios, the demand for new forms of products such as reconfigurable multifunctional amplifiers, integrated microwave and millimeter-wave multifunctional circuits, transceiver components, digital transceiver components, terahertz chips, three-dimensional integrated multifunctional devices, ultra-high-power devices, and heterogeneous integrated devices is continuously being proposed, promoting the exploration and innovative research and industrialization of new technologies and products in the third-generation semiconductor field. New application scenarios are continuously emerging in the third-generation semiconductor field, and new standards are being established, putting forward new demands for third-generation semiconductor devices. Traditional single-function devices will be replaced by integrated devices that are high-frequency, high-power, multifunctional, highly integrated, and miniaturized. For at least the next 10 to 15 years, they will continue to support the development of new industries such as the digital economy and the trillion-level market.
Continuous Emergence of Technical Achievements, Sustained Improvement of Industrialization Level
Through continuous key support from national policies and relevant ministries, China’s third-generation semiconductor technology has seen continuous breakthroughs during the 12th and 13th Five-Year Plans, achieving breakthrough progress in basic scientific issues and obtaining a series of core intellectual property rights, laying the foundation for the development of China’s third-generation semiconductor industry and cultivating a batch of scientific research talents and innovative teams. The “14th Five-Year” national key research and development plan (2021-2023) has been announced, continuing to include third-generation semiconductors in the key special projects of “new displays and strategic electronic materials” for key support (Table 1), continuously addressing issues of usability and sustainable innovation capabilities, and promoting sustained improvement in industrialization levels.
Table 1: “14th Five-Year” Key Special Project on “New Displays and Strategic Electronic Materials” for the Third-Generation Semiconductor Direction, 2021-2023 Consultation Guide
China’s GaN industry has formed a trend of midstream enterprises extending to upstream and downstream, achieving full coverage in the fields of RF, LED, and power electronic devices. In recent years, projects in the GaN field have emerged one after another. Particularly in the substrate field, breakthroughs have been achieved in n-type doping and compensation doping for GaN single crystal material growth, producing high conductivity and semi-insulating GaN single crystals. The thickness of 4-inch (1 inch = 2.54 cm) GaN substrate products reaches (650±50) μm, with defect density <3×106cm-2; the thickness of 4-inch iron-doped GaN substrate products can reach (420±50) μm, with defect density <5×106cm-2; at the same time, the first global breakthrough of GaN crystals with a thickness of over 1 cm has been achieved. In terms of GaN RF devices, China has formed a series of GaN microwave power devices and monolithic microwave integrated circuit (MMIC) products, with operating frequencies reaching the W-band and maximum output power exceeding 500W. The development of multifunctional integrated chips for transceivers has become a hot research area, reaching international advanced levels. In terms of GaN power electronic devices, domestic companies have launched low-voltage ultra-low on-resistance and high-voltage GaN power device products rated at 650-700V.In terms of application, the development of the domestic GaN power component market is mainly driven by consumer electronics, with key applications including fast chargers, audio, wireless charging, power supplies, and other consumer-grade product application scenarios. As the technology of GaN power electronic devices matures, they are also beginning to penetrate into new energy vehicles and industrial applications.
In the SiC field, a complete industrial chain has been formed domestically, covering from equipment, materials, devices to applications. In terms of domestically produced equipment, full process segments such as crystal growth, material processing, epitaxial growth, high-temperature injection, packaging assembly, and testing have been established, involving the research and industrialization capabilities of core equipment in each link of the chain. Currently, substrate materials have transitioned from 4-inch to 6-inch SiC substrates, and 8-inch SiC substrates have been developed, achieving small-batch supply; Shandong Tianyue Advanced Technology Co., Ltd. has made breakthroughs in the preparation of high-quality low-defect 8-inch crystals by liquid phase methods; 6-inch SiC epitaxial wafers cover materials for 600-1700V SiC power electronic devices, and 8-inch epitaxial products are gradually being produced. In terms of devices, domestic manufacturers are strengthening the technical research and development and production line investment of SiC devices to accelerate domestic substitution. In terms of SiC MOSFETs, companies such as China Electronics Technology Group Corporation (CETC), San’an Optoelectronics Co., Ltd. (San’an), and Yangzhou Yangjie Electronic Technology Co., Ltd. (Yangjie) have developed 1200V series products, showing excellent characteristics like on-resistance, with some being installed in vehicles. In terms of SiC diodes, companies such as San’an, China Resources Microelectronics Limited (CR Micro), and Wingtech Technology Co., Ltd. (Wingtech) have developed SiC diodes with excellent performance data, mainly used in ultra-high performance, low-loss, and high-efficiency power supply fields. At the terminal application level, new energy vehicles account for the largest share, with multiple brands and types of new energy vehicles continuously adopting SiC solutions, and 800V is set to become the mainstream platform for electric vehicles.
“Large Size, Low Cost” Becomes the Focus of Development, Accelerating Capacity Layout
Cost-effective products are an eternal purpose of industrial development, and expanding wafer size is one of the most effective ways. The third-generation semiconductor industry chain has a significant upstream traction effect, with substrate, epitaxy, and device manufacturing being key links that distinguish it from traditional semiconductor production, where substrate and epitaxy production account for over 60% of costs.Currently, China’s GaN RF device product research and production are mainly based on a 4-inch wafer system, while SiC power electronic device products are in the product introduction and small-batch application phase, with the mainstream domestic SiC substrate, epitaxy, and production lines being 6-inch systems.According to estimates by BAE Systems, if GaN RF devices transition from 4-inch wafer production to 6-inch wafer production, the cost per square millimeter will drop from $3 to $1.5.Moreover, mature 8-inch SiC technology will effectively solve the limited yield and excessive edge waste of 6-inch SiC wafer chips, reducing the unit chip cost by 30% to 50%.
In terms of GaN substrates, the upstream raw materials of the industry chain include GaN substrates and GaN epitaxial wafers, with relatively high raw material costs. For example, GaN-on-Si wafers can produce 80% to 90% more devices than 6-inch wafers, which directly impacts device costs. Related companies in China have developed well, such as Innoscience Technology Co., Ltd., which has become the world’s largest mass production enterprise for 8-inch Si-based GaN. GaN-on-SiC is mainly restricted by the size limitations of semi-insulating SiC wafer substrates. In the substrate field, GaN substrates still face certain technical difficulties; a 2-inch GaN wafer can sell for as high as $3000 to $5000 on the international market, and the quantity is very limited. Therefore, most manufacturers use SiC or sapphire as the substrate material for GaN. While this makes the substrate process simpler and somewhat cheaper, it is prone to defects such as cracks and warping. Currently, the quality and yield of GaN epitaxial wafers almost determine the performance and yield of device products, and the cost of epitaxial wafers also accounts for nearly 50% of the entire GaN value chain. Around 2015, many foreign companies and research institutions produced 2-inch self-supporting GaN substrates with millimeter-level thickness and achieved mass production. Reports have also emerged on 4 to 6-inch GaN substrate technology. Companies like Sumitomo Chemical have achieved the same-homogeneous epitaxial growth of GaN single crystals with low dislocation density of <105cm-2. Sciocs has successfully produced prototypes of 6-inch GaN substrates and is working on establishing mass production technology, planning to triple the sales of GaN single crystal substrates within ten years. Domestic related companies are also continuously increasing their research and production capabilities for GaN substrates, gradually achieving successful research and development of 1, 2, 4, and 6-inch GaN single crystal materials since 2021.
In terms of SiC wafer substrates, they are continuously evolving towards larger sizes. The larger the substrate size, the more chips can be manufactured from a single substrate, thus lowering the unit chip cost. In 2015, Wolfspeed showcased 8-inch SiC samples; in 2019, they completed the production of the first batch of 8-inch SiC wafers; in April 2022, they became the world’s first company to officially start using 8-inch SiC wafers, and in the first quarter of 2023, they achieved shipments. At the same time, Wolfspeed has invested billions of dollars to expand its existing factory in the United States and establish an 8-inch SiC production line in Germany. Other mainstream international manufacturers, such as Coherent, also showcased 8-inch conductive SiC substrates in 2015, and in 2019, they launched semi-insulating 8-inch SiC substrates, planning to achieve mass production by 2024 and increase production capacity, including 8-inch substrates, by 5 to 10 times by 2025. Infineon announced in September 2020 that its 8-inch SiC wafer production line has been completed and will achieve mass production by 2023, with 8-inch SiC substrate device mass production by 2025, and significantly expand its factory in Malaysia to build the world’s largest 8-inch SiC wafer factory. Several companies, including ST and ROHM, have also raised their 8-inch SiC mass production plans to 2023, successively investing heavily in new large-scale factories. Currently, Wolfspeed is the only company in the world capable of mass-producing 8-inch SiC wafers, while most other international companies have set their mass production milestones around 2023. In the next 2 to 3 years, Wolfspeed’s leading position in 8-inch SiC production capacity will continue until more companies increase their capacity.
Many domestic manufacturers are also entering the 8-inch SiC team. In 2020, Shanxi ShuoKe Crystal Co., Ltd. successfully developed 8-inch substrate wafers; in 2022, they achieved small-batch production of 8-inch N-type SiC polished wafers. In 2022, Beijing Tianke Heda Semiconductor Co., Ltd. announced plans to achieve small-scale mass production of 8-inch conductive SiC substrates in 2023. At the same time, the agreements between Tianke Heda and Tianyue Advanced and Infineon will accelerate the mass production process of domestic 8-inch substrates. In February 2023, Zhejiang JingSheng Electromechanical Co., Ltd. announced that it has completed the expansion of diameter and quality iteration from 6 inches to 8 inches, achieving the development of 8-inch polished wafers and small-batch production. In May, Hantian Tiancheng Electronic Technology (Xiamen) Co., Ltd. announced that it has successfully developed 8-inch epitaxial processes with independent intellectual property rights and has the capability for mass production. In June, San’an Optoelectronics reached a new cooperation agreement with ST to invest $3.2 billion to build a joint venture factory for 8-inch SiC epitaxy and chip foundry. In the same month, China Electronics Compound Semiconductor Co., Ltd. signed a long-term supply agreement for SiC materials, including 8-inch SiC materials, with South Korea’s Power Master. In October, the first batch of 8-inch SiC substrates from Harbin Keyou Semiconductor Industry Equipment and Technology Research Institute Co., Ltd. was announced to be offline, and other companies that have made breakthroughs in 2023 include Zhejiang Dongni Electronics Co., Ltd., Shanxi Tianceng Semiconductor Materials Co., Ltd., and Hangzhou Qianjing Semiconductor Co., Ltd.In the substrate segment, various companies have announced expansion plans to further achieve domestic substitution; in the epitaxy segment, companies such as China Electronics Technology Group Corporation (CETC) Southern Group, Guangdong Tianyu Semiconductor Co., Ltd. (Dongguan Tianyu), Hebei Puxing Electronics Technology Co., Ltd. (Puxing Electronics), and Wuhu Qidi Semiconductor Co., Ltd. (Qidi Semiconductor) are also making layouts.Based on the comprehensive situation of foreign third-generation semiconductor technology and industrial development and the current status of domestic third-generation semiconductors, by the end of the “14th Five-Year Plan”, the mainstream wafer size in the GaN RF field will be 6 inches, while the mainstream wafer size for SiC power electronics will see a coexistence of 6 to 8 inches. By improving the product wafer size, the research and production costs of GaN RF devices and SiC power electronic devices in China will be further reduced, enhancing the international competitiveness of Chinese products.Leading Enterprises Choose IDM Model, Future More Fabless + Foundry Models CoexistThe history and current status of third-generation semiconductor development at home and abroad show a development model primarily based on vertically integrated manufacturing (IDM).As is well-known, the current demand for terminal applications of SiC and GaN is very large, while the substrate wafer production capacity is relatively insufficient, and the market is mainly controlled by a few suppliers.Device manufacturers cannot rely indefinitely on external substrate suppliers; therefore, they are integrating substrates into the entire manufacturing strategy of devices and technologies.Currently, major foreign third-generation semiconductor companies such as Wolfspeed, ST, ROHM, Infineon, Qorvo, and Sumitomo are developing and growing using this model. Some companies are also continuously improving their IDM model through acquisitions, such as ST’s acquisition of Norstel, ROHM’s acquisition of SiCrystal, Infineon’s acquisition of GaN Systems, and onsemi’s acquisition of GTAT.In recent years, government support for the third-generation semiconductor field in Europe and the United States has mainly focused on IDM companies with production lines. For example, in 2020, the U.S. Department of Defense supported Qorvo in building an advanced heterogeneous integrated packaging (SHIP) RF manufacturing and prototype design center. In 2021, the U.S. Department of Defense’s Advanced Research Projects Agency supported Transphorm in researching GaN epitaxial wafers on sapphire substrates. In 2023, the U.S. Air Force supported MACOM in developing existing GaN-on-SiC technology. The EU has also invested funds to support companies like BOSCH and ST in establishing a resilient European supply chain around SiC power semiconductors.Domestic major third-generation semiconductor companies are also adopting the IDM model while continuously emerging companies are transitioning to the IDM model. For example, in 2021, SIDA Semiconductor announced a fundraising of 2 billion yuan to invest in the research and industrialization projects of high-voltage specialty process power chips and SiC chips, marking its transition from a Fabless to an IDM model.This indicates that it has begun shifting from a Fabless to an IDM model. Currently, companies with good research and industrialization capabilities include China Electronics Technology Group Corporation, San’an Integrated, Hangzhou Silan Microelectronics Co., Ltd., CR Micro, and Shenzhen Basic Semiconductor Co., Ltd. The most significant feature of these IDM companies is that they established third-generation semiconductor research lines early on and continuously invested government and self-funding to enhance their research capabilities and efficiency, as well as process line processing capabilities, which are very conducive to rapid technological advancement and cost control. Conversely, leveraging production line processing capabilities to continuously explore basic materials and design potential, achieving optimal device results through the integration of materials, design, and processes. Additionally, improvements in product reliability and yield, ensuring product differentiation, cost, and long-term supply, will facilitate the transition from the innovation chain to the value chain through the IDM model. The performance and market competitiveness of SiC and GaN devices hinge on the manufacturing segment, and new entrants in the market must master this key segment to ensure their future market position. International device manufacturers are also expanding into upstream materials to seize the initiative. The choice of the IDM model by domestic major manufacturers is correct. In recent years, with the improvement of average yield, shortened device manufacturing cycles, capacity upgrades, and support from self-research and domestic equipment, the optimization of the IDM industrial cluster ecological model has achieved a revolution in industrial chain efficiency, significantly reducing device costs.Currently, although the IDM model has the advantage of internal technological integration, which is conducive to accumulating process experience, due to the cyclicality of the semiconductor industry, IDM companies are prone to being constrained by existing fixed capacities, falling into a passive situation. Additionally, companies need to cover the entire production chain from design to manufacturing to packaging and testing, requiring strong technical support and substantial financial strength, which is a barrier that most small or startup companies cannot overcome. Simultaneously, as the terminal products and applications of chips become increasingly complex, the design difficulty of chips is rapidly increasing, and the resources and costs required for research and development are continuously rising, leading to a more refined division of labor in the industry. The Fabless model without foundries has become one of the mainstream business models for chip design companies. Such companies focus on chip research and design and sales, outsourcing wafer manufacturing, packaging, testing, and other production links to foundry enterprises focused on contract manufacturing. For example, in 2023, foundry model representatives like GlobalFoundries and DBHiTek are continuously collaborating with companies to jointly develop SiC and GaN technologies and improve production capacity. With the rapid growth of future application markets and the continuous emergence of new application markets, relying solely on IDM companies may lead to supply chain shortages, as evidenced by the significant shortage of SiC automotive-grade chips. Foundry model companies specializing in SiC chip manufacturing, packaging, and testing will become particularly important, and a combination of Foundry + Fabless models will continue to emerge. In contrast, the GaN market and application scenarios and capacity are relatively small, and for the time being, it will still primarily focus on the IDM model.Currently, domestic companies are continuously emerging in various fields, from substrate epitaxy material preparation to chip processing and module production. Fabless companies need to deeply cooperate with foundries in the foundry model to promote the healthy development of the entire industry chain collaboratively. Overall, the industry will gradually present a situation of coexistence between the IDM model and the Fabless + Foundry model, each meeting different application scenarios in the terminal market, and this is also the future development direction of the business model for third-generation semiconductor companies, which can timely expand or reduce capacity according to market fluctuations and meet regional market demands nearby.China Has a Good Foundation but Still Faces Shortcomings; Development Needs to Be Driven by ApplicationsAfter more than 20 years of development, China’s third-generation semiconductors have established a foundation for technological breakthroughs and industrial collaborative development. The industry is on the verge of explosion, with Chinese companies preparing for it, and their self-controllable capabilities are continuously strengthening, and their competitive strength is continuously improving.For instance, in the automotive industry, China has currently used the development model of new energy vehicles to narrow the gap with the United States, Europe, Japan, and other countries and regions, achieving leadership in certain fields; precision manufacturing levels and supporting capabilities related to semiconductors have rapidly improved, laying a solid foundation for the localization of related equipment, supporting China’s hundreds of billions of lighting market and the deployment of 5G base stations. In the future, SiC power electronic devices will further support the trillion-level power conversion application market.The third-generation semiconductors have always been one of the important areas supported by the government. After more than a decade of development, good research results have been achieved in basic theories and materials. From equipment supply to wafer and device manufacturing, and to system integration, the rise of domestic companies will promote the shift towards local procurement. However, there is currently no clear supply leader in the ecosystem. Research and development in related fields are at a bottleneck. In the field of traditional single-function devices, the research and technology level has reached a certain height, making it difficult to achieve significant improvements. There is a certain gap in industrialization technology capabilities compared to foreign countries, and the processing capabilities of production lines cannot fully match the research and development level of new product forms, leading to research and development products lagging behind application scenario demands.Facing the enormous application market for China’s third-generation semiconductors, products and development models should be defined based on market and application-driven approaches, rather than blindly following international giants for domestic substitution. In recent years, investment and factory construction enthusiasm have surged, with too many manufacturers entering this emerging market in a short time. Many companies are conducting related research and production of third-generation semiconductors, involving single crystal substrate materials, epitaxial materials, device design, and manufacturing processes. However, it is essential to realize that there is currently an excessive investment proportion in materials, and there is still a certain gap from forming a complete and sound industrial ecosystem. There are still issues and risks such as fragmented industrial layout, severe homogenization competition, insufficient average investment in the industry, low-level repetition, investment without technology, and weak comprehensive product competitiveness. It is necessary to consider whether overly intense market competition and excess capacity will hinder the development of domestic manufacturers in the coming years.To address existing problems, it is necessary to focus on major national needs, promote the exploration of a new type of national system, aggregate scientific and technological innovation resources, concentrate on key core technologies and major application directions of third-generation semiconductors, conduct overall layouts of the entire technology system, clarify the technical points of “mature technology,” “needs to break through,” and “cutting-edge technology,” and focus on supporting common technologies and cutting-edge technology breakthroughs related to production lines, fully leveraging China’s huge market advantages and the industrial driving role of demand on the application side, linking advantageous resources across the industry, continuously optimizing the allocation of innovation resources, adhering to the dual-wheel drive of research and application, achieving technological breakthroughs and rapid transformation of scientific and technological achievements, forming future competitive advantages, promoting high-quality development in related fields, and seizing new opportunities for third-generation semiconductor applications to provide strong support for national defense security, digital economy, and the “dual carbon” strategy.
Conclusion
As a frontier field of global semiconductor technology and industry, third-generation semiconductors have significant strategic significance for national defense security and national economic development. In the current critical window period of rapid technological iteration and continuous expansion of products, the competition for international discourse power in semiconductors among countries is becoming increasingly fierce. China has a good foundation in this field, having established a relatively complete industrial chain from substrate, epitaxy, design, manufacturing, packaging, testing, to application, with continuous development in technology and industrialization capabilities, and increasing self-controllable capabilities. Achievements have been realized in large-scale applications in important fields of the digital economy, such as mobile communications and new energy vehicles. As the world’s largest third-generation semiconductor application market, China has demonstrated a strong driving force for technology innovation driven by applications, continuously developing towards “larger size, lower cost, and higher performance.” Of course, the industrialization development of China’s third-generation semiconductors also faces many challenges, and there is still a certain gap from forming a complete and sound industrial ecosystem. After the continuous development of the 12th and 13th Five-Year Plans, the overall issue of “existence” for China’s third-generation semiconductors has been resolved. During the 14th Five-Year Plan, the focus will be on addressing the issues of “usability, ease of use,” and sustainable innovation capabilities. With more than half of the 14th Five-Year Plan completed, the need for the high-quality development of the third-generation semiconductor industry is becoming more urgent. In the future, there is still a need for strategic guidance at the national level for industrial development, leveraging the advantages of a new type of national system, adhering to the dual-wheel drive of application and research and development, addressing existing issues, and leveraging the leading role of national-level innovation platforms in industrial development and core technology breakthroughs. Through joint efforts from the government, enterprises, and universities, support the healthy development of China’s third-generation semiconductor industry and seize new opportunities for future applications.
Authors:Gao Shuang, Zheng Yuting, Zhang Zhiguo
Author Profiles: Gao Shuang, Engineer at China Electronics Technology Group Corporation’s Third-Generation Semiconductor Technology Co., Ltd., research direction in wide bandgap semiconductor industrial economics; Zheng Yuting (corresponding author), Engineer at China Electronics Technology Group Corporation’s Third-Generation Semiconductor Technology Co., Ltd., research direction in wide bandgap semiconductor fundamentals.
This article is reproduced from the “Science and Technology Guide” WeChat public account.//“Physics” 50 Years Selected Articles
What is a neutron bomb? | “Physics” 50 Years Selected Articles
A historical review of crystal defect research | “Physics” 50 Years Selected Articles
Phase transitions and critical phenomena (I) | “Physics” 50 Years Selected Articles
Phase transitions and critical phenomena (II) | “Physics” 50 Years Selected Articles
Phase transitions and critical phenomena (III) | “Physics” 50 Years Selected Articles
A review and outlook on condensed matter physics | “Physics” 50 Years Selected Articles
Acoustics and ocean development | “Physics” 50 Years Selected Articles
The role of models in the development of physics | “Physics” 50 Years Selected Articles
My memories of Wu Youxun, Ye Qisun, and Sa Bendong | “Physics” 50 Years Selected Articles
The Physics Department of National Southwestern Associated University – A unique flower in the Chinese physics community during the Anti-Japanese War (I) | “Physics” 50 Years Selected Articles
The Physics Department of National Southwestern Associated University – A unique flower in the Chinese physics community during the Anti-Japanese War (II) | “Physics” 50 Years Selected Articles
Discovery of nuclear fission: history and lessons – Commemorating the 60th anniversary of the discovery of nuclear fission | “Physics” 50 Years Selected Articles
Review and outlook – Commemorating the 100th anniversary of the birth of quantum theory | “Physics” 50 Years Selected Articles
My research career – Huang Kun | “Physics” 50 Years Selected Articles
A model of collaboration between theoretical physicists and biologists in China – A historical contribution by Tang Peisheng and Wang Zhuxi to the study of water relations in plant cells (Part I) | “Physics” 50 Years Selected Articles
A model of collaboration between theoretical physicists and biologists in China – A historical contribution by Tang Peisheng and Wang Zhuxi to the study of water relations in plant cells (Part II) | “Physics” 50 Years Selected Articles
In Memory of the Forgotten – Reminiscing about Ye Qisun in his later years | “Physics” 50 Years Selected Articles
From the perspective of molecular biology – A commemoration of the 50th anniversary of the publication of the golden helix paper | “Physics” 50 Years Selected Articles
Beauty can be expressed – Mathematical equations describing floral morphology | “Physics” 50 Years Selected Articles
Einstein: A portrait on a stamp | “Physics” 50 Years Selected Articles
Fun discussions on ball sports physics | “Physics” 50 Years Selected Articles
In an instant, ninety years | “Physics” 50 Years Selected Articles
A classic work that has cultivated generations of physicists – A review of “Theory of Lattice Dynamics” | “Physics” 50 Years Selected Articles
Landau’s centenary | “Physics” 50 Years Selected Articles
In the language of heaven, explain the principles of physics | “Physics” 50 Years Selected Articles
Soft matter physics – A new discipline in physics | “Physics” 50 Years Selected Articles
Cosmology in the last 80 years | “Physics” 50 Years Selected Articles
Entropy is not a commodity – the Myth of Entropy | “Physics” 50 Years Selected Articles
Evolution phenomena in physics | “Physics” 50 Years Selected Articles
Purdue Notes – Starting from the 2010 Nobel Prize in Chemistry | “Physics” 50 Years Selected Articles
My learning and research experiences | “Physics” 50 Years Selected Articles
Weather forecasting – From experience to physical mathematics theory and supercomputing | “Physics” 50 Years Selected Articles
Commemorating the 100th anniversary of Bohr’s “Great Trilogy” and the 100th anniversary of the establishment of the Physics Department at Peking University | “Physics” 50 Years Selected Articles
The history and status of synchrotron radiation | “Physics” 50 Years Selected Articles
The origin of the concepts of Maxwell’s equations and gauge theory | “Physics” 50 Years Selected Articles
Space science – The source of exploration and discovery | “Physics” 50 Years Selected Articles
The establishment of Maxwell’s equations and their role | “Physics” 50 Years Selected Articles
New chapters in condensed matter materials – Topological quantum states beyond the Landau paradigm | “Physics” 50 Years Selected Articles
The art of thinking in physics | “Physics” 50 Years Selected Articles
Regarding the basic principles of quantum mechanics | “Physics” 50 Years Selected Articles
Singularities in spacetime and black holes – Interpretation of the 2020 Nobel Prize in Physics | “Physics” 50 Years Selected Articles
A new chapter in condensed matter physics – Topological quantum states beyond the Landau paradigm | “Physics” 50 Years Selected Articles