Scientific Observation丨Strengthening the construction of basic semiconductor capabilities and lighting up the "lighthouse" for the independent and self-reliant development of semiconductors

This article was published in "Scientific Observation", Issue 2, 2023, of "Proceedings of the Chinese Academy of Sciences"

Institute of Semiconductors, Chinese Academy of Sciences State Key Laboratory of Semiconductor Superlattice

Under "anti-globalization", the "decoupling" of industrial chains has become more and more intense. At present, my country's basic scientific and technological capabilities cannot support the national strategy of achieving high-level scientific and technological self-reliance. To this end, the report of the 20th National Congress of the Communist Party of China proposed to strengthen the construction of basic scientific and technological capabilities. Hou Jianguo, President and Party Secretary of the Chinese Academy of Sciences, wrote in the People's Daily that the scientific and technological foundation includes not only "hard conditions" such as various scientific and technological innovation organizations, scientific research facilities and platforms, scientific data, literature and periodicals, but also scientific and technological policies, institutional regulations, and innovation culture. Wait for the "soft environment". The Chinese Academy of Sciences formulated the "Ten Basic Research Points" in 2022 to clarify the strategic positioning, key layout and development goals of the Chinese Academy of Sciences' basic research, and proposed a series of topics from the aspects of topic selection mechanism, organizational model, conditional support, talent team, evaluation system, international cooperation, etc. Targeted and operable policy measures emphasize the construction of academic style, work style and academic ecology.

The technological war between China and the United States has exposed the problem of "stuck" in my country's key core technologies. Since the U.S. sanctioned ZTE in 2018, everyone has been discussing the issue of semiconductor “stuck”. From party and state leaders to ordinary people, they all agree that semiconductor technology must be vigorously developed. In particular, General Secretary Xi Jinping pointed out at the 2020 Scientists Symposium: "The root cause of many of the 'stuck neck' technical problems faced by our country is that basic theoretical research cannot keep up, and the source and underlying things are not clear." Although basic research on semiconductors It has received great attention in the past few years, but basic semiconductor capabilities, including subject settings, collaborative innovation, infrastructure, R&D investment, evaluation mechanisms, and graduate student quotas, have not been fundamentally improved, making it difficult to support high-level self-reliance and self-reliance in semiconductor technology.

Strengthening the construction of basic semiconductor capabilities is of great strategic significance

Semiconductors are the “main battlefield” in the current technological war between China and the United States. Semiconductor products with a global annual output value of US$600 billion cover thousands of chips and nearly 100,000 discrete devices, supporting various downstream electronic products with an annual output value of several trillion US dollars, as well as a digital economy with an annual output value of tens of trillions of US dollars. According to statistics, $1 of semiconductor products drives $100 of global GDP.

The semiconductor industry chain is long and broad: the upstream includes EDA software/IP modules, semiconductor equipment and materials, the midstream is chip design, manufacturing, packaging and testing, and the downstream is various electronic products, involving a large number of materials, equipment and accessories, software and IP modules . Academician Wang Yangyuan pointed out that any material, equipment or even accessory in the upstream of the semiconductor industry chain may become a means to restrict competitors. Even the United States, the birthplace of semiconductors, cannot solve the entire semiconductor industry chain independently. To this end, the United States is eager to win over Japan, South Korea and Taiwan to form the Semiconductor Quad Alliance (Chip4) to improve the security of its semiconductor supply chain while curbing my country's development of high-end chip industry. On August 11, 2022, the United States announced an embargo on my country's EDA software for next-generation GAA transistors, intending to "lock" my country's semiconductor industry in FinFET transistor technology. Fundamental research results in global semiconductor physics and microelectronics are integrated into the Process Design Kit (PDK) of EDA tools . Various chip companies in our country share the results of global semiconductor basic research by purchasing EDA companies' PDK packages, leading our country's policymakers, government personnel, and even the industry to believe that the semiconductor industry can be developed without basic semiconductor research. Today, the United States has extinguished the "lighthouse" and we have entered the "dark forest."

Semiconductor physics is the source of all semiconductor technology. The first quantum revolution gave birth to devices such as lasers and transistors, as well as semiconductor information technology including integrated circuits, optoelectronic devices, sensors, and discrete devices. Eleven achievements in the semiconductor field have won 9 Nobel Prizes in Physics. The current transistor is close to its physical limit, and "Moore's Law" is about to expire. There is an urgent need to develop new materials, new structures, new theories, new devices and new circuits that break through the performance bottleneck of CMOS devices. We are facing many basic physical problems with "no known solutions" challenge.

In the context of the Sino-US technology war and the "decoupling" of industrial chains, even if advanced chips are designed or manufactured, it will be difficult to break into the international supply chain. Through massive investment in domestic substitution, it can only achieve internal circulation or narrow the gap with the United States, but it still cannot change the "stuck neck" dilemma of "you are in me, but I am not in you". General Secretary Xi Jinping has repeatedly pointed out the strategic policy of strengthening basic research to solve the "stuck neck" problem. Currently, most high-end chips use the same FinFET transistor manufacturing technology; among the tens of thousands of FinFET transistor patents, a considerable part of the core patents come from basic research results in semiconductor physics, and these results do not rely on the most advanced technologies such as EUV lithography machines. of semiconductor manufacturing equipment. By vigorously strengthening basic semiconductor research and laying out a large number of patents in Europe and the United States around the materials, devices, and processes of next-generation transistors, we can set up "gates" in chip manufacturing, the "throat" of the global semiconductor industry chain, and form countermeasures. , is expected to solve the "stuck neck" problem of key core semiconductor technologies.

The United States is strengthening its basic semiconductor research capabilities

Under the slogan "The future of the United States depends on semiconductors," the United States passed the $280 billion "Chip and Science Act" in 2022, of which only 39 billion was used to subsidize chip manufacturing, and the rest was mainly used for research and innovation. Including: 11 billion to establish the National Semiconductor Technology Center, and a total of 169.9 billion US dollars in new funding from research funding agencies such as the National Science Foundation (81 billion) and the Department of Energy (67.9 billion) in the next five years.

During the period of great development of semiconductors from the 1960s to the 1990s, well-known universities around the world had a large team of professors in the field of semiconductors. In the 21st century, these professors gradually retired, and newly hired professors were mainly engaged in emerging directions, and basic semiconductor research gradually declined. Related Research is transferred to semiconductor corporate research facilities. This bill will allow American universities to re-recruit a large number of professors in the semiconductor field, attract a larger number of graduate students and postdoctoral fellows to the United States to engage in basic semiconductor research, and will inject strong vitality into the source innovation of semiconductor technology.

The report shows that the $67.9 billion the U.S. Department of Energy received from the bill will be mainly used for semiconductor technology research in the "post-Moore era." As early as 2016, eight national laboratories of the U.S. Department of Energy held a seminar on "post-Moore era" semiconductor technology at Sandia National Laboratories to evaluate the ability of the national laboratory's large science facilities to support microelectronics research and proposed Disruptive innovations in the new computing paradigm of the "post-Moore era" from materials and devices to system architecture and software. Lawrence Berkeley National Laboratory was reorganized in 2018. "Beyond Moore" is one of four research directions, proposing an in-depth collaborative design innovation framework from semiconductor material physics, node physics, device physics, circuits to systems.

Dilemmas faced by my country's basic semiconductor research capacity building

The National Science Conference held in 1978 called for the march towards the modernization of science and technology. After ten years of civil strife during the "Cultural Revolution", my country's science and technology work finally ushered in the "spring of science." However, at that time, there was a generational gap in technology and equipment between my country and Western developed countries, and Chinese companies were unable to "come up with topics" for basic research; basic research could only be separated from the actual needs of domestic industrial development in the process of catching up with the world's technological frontiers. After joining the WTO, the concepts of "science without borders" and "globalization" have been deeply rooted in the hearts of the people; from the "211 Project" and "985 Project" to today's "Double First-Class" construction, we have continuously strengthened the evaluation mechanism based on papers and published reviews, neglecting Due to differences in subject directions and research fields, scientific research resources are accelerating towards emerging hot spots where it is easy to publish high-end papers. Basic research closer to industrial applications is less likely to be done.

my country's first march into semiconductors began in 1956. Huang Kun, the founder of solid-state physics and semiconductor physics in my country, proposed and organized the implementation of the "Five-University Joint Specialization in Semiconductor Physics". Peking University, Fudan University, Jilin University, Xiamen University and Nanjing University The university's senior physics students and related teachers are trained at Peking University; in two years, more than 300 first-generation semiconductor professionals in my country have been trained. However, because the Ministry of Education canceled the semiconductor physics and device major in 1997, today, 67 years later, my country's basic semiconductor research talents have withered, and there are only a handful of people engaged in semiconductor theoretical research. Without a large semiconductor physics research team, it will be difficult to achieve independent innovation at the source and underlying level of semiconductor technology. Under the blockade of the United States, the development of my country's semiconductor industry will become a castle in the air.

For a long time, the United States’ annual semiconductor R&D investment has exceeded twice the sum of other countries in the world. In 2018, the U.S. federal government invested US$6 billion in semiconductor research and development, while semiconductor companies invested as much as US$40 billion, which is close to my country's central government's total scientific and technological research and development expenditure of 373.8 billion yuan. Taking the funding from the National Science Foundation of China in 2019 as an example, the funding for semiconductor science and information devices (384 million) and optics and optoelectronics (551 million) for basic semiconductor research only accounted for 2.8% of the total funding of 33 billion yuan. ; Including the 01, 02, and 03 major projects of the Ministry of Science and Technology and key projects in the semiconductor field, China's semiconductor R&D investment has long been less than 5% of that of the United States.

In addition to having a large number of world-class universities, the United States also has a large number of national laboratories as the "ballast stone" for its basic research; in addition, the major American semiconductor giants have huge basic research departments, such as Bell Labs and IBM Labs, etc. However, the number of basic semiconductor research bases in our country is scarce. The National Key Laboratory of Semiconductor Superlattice is the only national key laboratory that focuses on basic semiconductor physics. Among the established national laboratories, there are also very many personnel engaged in basic semiconductor research. There are few; no large-scale scientific equipment has been built to serve basic semiconductor research; semiconductor companies are still in the stage of localization substitution and are unable to take into account basic research.

Since the 18th National Congress of the Communist Party of China, party and state leaders have attached great importance to basic research, and the country has issued a series of documents to strengthen basic research and break the "four principles." In the field of semiconductors, the country launched the construction of model microelectronics colleges in 2014. So far, a total of 28 colleges and universities have established microelectronics colleges; in 2020, the first-level discipline of integrated circuit science and engineering will be established. However, due to the disconnect between industry and scientific research, the inertia of paper-based research will be difficult to reverse in the short term. In the second round of "double first-class" construction list announced in 2022, more than 30 materials majors in colleges and universities across the country were selected for the "double first-class" construction, including 22 chemistry institutes, 8 physics institutes, and 1 integrated circuit science institute; at the same time, Semiconductors don’t even have a subject. Basic research on traditional semiconductors not only requires large investments, high thresholds, long cycles, but also makes it difficult to publish high-end papers. Under the current paper-based evaluation mechanism, it is difficult to be selected for various talent projects and has a low input-output ratio, making it impossible to become a key development focus of universities. This has led to various exemplary microelectronics schools to focus on “changing lanes and overtaking” research in the direction of emerging hot materials.

Japan organized integrated circuit research through the "VLSI Research Alliance" in 1976, which helped Japan surpass the United States in semiconductor market share in 1986. SEMATECH (Semiconductor Manufacturing Technology Alliance) established by the United States in 1987 helped the United States regain its leadership position in the semiconductor industry. Today, IMEC in Belgium has become a world-class semiconductor innovation institution, and together with Intel Corporation and IBM Corporation in the United States, it is known as the "3I" in the global microelectronics field. A large number of professors at American universities are undertaking basic research projects commissioned by companies such as Intel, Samsung and TSMC, including research projects on semiconductor theory. However, our country has not yet established a similar organization to organize collaborative innovation in basic semiconductor research; domestic semiconductor companies are more than two generations behind the international advanced level. They mainly optimize processes to improve yield rates based on PDK provided by others, and have no time to focus on next-generation transistors. Frontier basic research is difficult to “come up with” for national strategic science and technology forces such as universities and scientific research institutes; while researchers at universities and scientific research institutes can only understand the scientific issues of semiconductor cutting-edge technology from literature and conferences, and it is difficult to find real problems and Really solves the problem.

Suggestions on strengthening basic semiconductor research capacity building

Establish and improve cross-departmental coordination mechanisms

It is recommended that the National Integrated Circuit Leading Group be renamed the National Semiconductor Leading Group to cover basic semiconductor research. Coordinate cross-departmental scientific and technological resources such as human, financial, material, and policy resources, strengthen decision-making and overall coordination, and be responsible for formulating the national semiconductor development strategy. It has an office responsible for hiring scientists from industry and academia to serve as project managers off-duty, selecting key core technologies and leading talents, supervising and implementing key research plans, assessing key research goals, formulating support policies, and other matters. It is recommended that 10% of the semiconductor output value be used as a standard to match basic semiconductor research funding. The Chinese Academy of Sciences or the Academy of Engineering should establish a semiconductor department, and the Ministry of Industry and Information Technology, the Ministry of Science and Technology, and the National Foundation of China should set up special semiconductor departments to attract the best talents to expand basic semiconductor research. Team.

Resume Semiconductor Physics Major

Introduce strong measures to make up for historical debts in basic semiconductor research. The semiconductor physics major must be restored as soon as possible, and at the same time, we must learn the strategy for the first march into semiconductors, starting from the "Five Schools Joint Semiconductor Physics Specialization", and urgently gather half of the juniors and seniors majoring in physics from the "double first-class" universities across the country. students, focus on basic theoretical courses on semiconductors, and select a group to enter doctoral programs to continue their studies. By cultivating, introducing, and stabilizing a large number of talents who have been engaged in semiconductor physics research for a long time, we strive to open up new directions, propose new theories, develop new methods, and discover new phenomena at the source and bottom of semiconductor technology.

Building a semiconductor basic research network

Encourage research-oriented universities to establish semiconductor colleges; recommend that the National Foundation for Semiconductor Basic Research add special quotas for talent projects such as the National Science Fund for Distinguished Young Scholars and innovative research groups, and establish about 10 basic scientific research centers in semiconductor physics across the country to fund 20 innovation groups and 100 research groups use the talent team effect to drive basic research back to the semiconductor field.

Establish a regional joint innovation platform

The United States is about to establish a National Semiconductor Technology Center; South Korea will establish a National Semiconductor Research Institute; and Taiwan, China, has established a Semiconductor Research Center. Our country must strengthen the construction of the national laboratory system in the semiconductor field as soon as possible. In line with the development needs of the regional semiconductor industry, about 10 large-scale regional joint innovation platforms have been established across the country to jointly tackle common technologies. Provide information sharing and academic exchange mechanisms for universities, scientific research institutes, and industries, establish extensive cooperation alliances, and promote the integration of the innovation chain and the industrial chain and the coordinated development of the upstream and downstream of the semiconductor industry chain.

Deepen the reform of the science and technology system and make good use of the "baton"

Vigorously reverse the shortcomings of pragmatism-led scientific research, dismantle the walls under the idea of ​​"small-scale farmer economy", introduce measures to protect the "death valley" innovation link with low visibility, establish a bottom-up innovation system driven by original innovation, and improve basic research support National development and security. suggestion:

1. Use funds as a means to integrate scientific research resources such as disciplines, talents, assessments, platforms, policies, etc., and cut off the hands of power that distort demand.

2. Vigorously promote the spirit of scientists who pursue originality and resist low-level repetitive follow-up research.

3. Build a diversified basic research investment mechanism with different funding targets, and give full play to the characteristics and advantages of national strategic scientific and technological forces such as national laboratories, scientific research institutes, and research universities.

4. Establish an exit mechanism for the basic research funding system. Emerging research directions are evaluated after 10 consecutive years of funding, and funding directions that have not produced significant applications are canceled, forcing basic researchers to turn to new directions and improve original innovation capabilities.

5. Institutionally ensure that post-doctoral students are more willing to engage in post-doctoral research after graduation, strengthen their independent research and interdisciplinary capabilities, and promote post-doctoral students into the main force of basic research.

6. Use discipline assessment and talent evaluation and other means to guide research universities to strengthen disciplinary diversity. Curb the repeated establishment of research teams in the same direction, break the unfavorable situation of crowding in a few popular fields, and form an original innovation source where "thousands of ships compete and hundreds of boats compete to cross the river".

7. Improve the intellectual property protection system and stimulate corporate innovation.

Strengthening the construction of basic semiconductor capabilities, stabilizing a group of basic semiconductor research teams, carrying out theoretical innovation at the source and bottom of semiconductor technology, and setting up "gates" of patents in advance at the bottom of the chip that cannot be bypassed are the key to solving the "stuck neck" of key core technologies in semiconductors. An effective strategy for difficult problems.

Director: Yang Liuchun

Editor in charge: Zhang Fan

Assistant Editor, Proofreader: PAN

Typesetting: Jiang Miao