If silicon-based comes to an end, then the global semiconductor industry must find new materials to “continue life.” In 2009, the Semiconductor Technology Development Roadmap Committee (ITRS) included carbon-based nanomaterials as a future integrated circuit technology option that continued Moore’s law, but in the subsequent time, the research progress of carbon nanomaterials did not give the industry satisfaction Answer sheet.
Recently, the team of Peng Lianmao and Zhang Zhiyong, academician of the Chinese Academy of Sciences and professor of the Department of Electronics at Peking University, announced that they have taken a big step forward from laboratory research to industrial application of carbon-based semiconductor technology. On May 22, the team published a paper “High-density Semiconductor Carbon Nanotube Parallel Array for High-Performance Electronics” in “Science” magazine, introducing its latest development of multiple purification and dimension-constrained self-assembly methods . Because it solves the problem that has long plagued the preparation of carbon-based semiconductor materials, the results of this method are exciting for the industry, but there is still a long way to go from the laboratory to the industrialization.
A major leap in high-performance carbon nanotubes
Each technology has its life cycle. The existing silicon-based chip manufacturing technology is about to touch its limit. Carbon nanotube technology is considered to be one of the important options for post-Moore technology.
Compared with traditional silicon-based CMOS transistors, carbon tube transistors have obvious comprehensive advantages in speed and power consumption. IBM’s theoretical calculations show that if the design is completely based on the existing two-dimensional plane frame, the carbon tube technology has the advantage of 15 generations and at least 30 years or more compared with the silicon-based technology. System-level simulations at Stanford University show that carbon tube technology is also expected to develop conventional two-dimensional silicon-based chip technology into three-dimensional chip technology, which will increase the overall performance of the current chip by more than 1,000 times.
The industry has high hopes for carbon nanotubes. At the TSMC IEDM conference in 2017, TSMC CTO Sun Yuancheng reported the news about carbon nanotubes.
However, carbon tube technology “ideal is very full, reality is very skinny.” The ideal value of carbon-based in theory and simulation has allowed IBM and Intel to explore this for many years, but they have encountered bottlenecks. In 2005, Intel’s device experts published a paper and concluded that carbon nanotube devices with performance surpassing silicon-based n-type transistors could not be produced. Intel subsequently abandoned carbon-based integrated circuit technology. On the technical route, both IBM and Intel chose the traditional “doping” process to prepare carbon nanotube transistors.
Academician Peng Lianmao and Zhang Zhiyong’s team entered the field in 2001 and chose a different path from Intel and IBM to develop a complete set of “undoped manufacturing technology” for carbon nanotube CMOS integrated circuits and optoelectronic devices. In 2017, a carbon tube transistor with a gate length of 5 nanometers was produced for the first time, the world’s smallest high-performance transistor so far. The overall performance is 10 times ahead of the best silicon-based transistors at the time, and it is close to the quantum limit. The paper of this achievement was published. In the 2017 issue of Science.
In 2018, the team once again broke through the traditional theoretical limit and developed a new principle of ultra-low-power Dirac source transistors, which can meet the needs of ultra-low-power integrated circuits in the future, and is the development of ultra-low-power nanoelectronics. The foundation was laid, and the paper was published in the journal Science in 2018.
In a paper published in “Science” on May 22 this year, the team of Academician Peng Lianmao and Professor Zhang Zhiyong explained their latest development of multiple purification and dimension-restricted self-assembly methods. This method solves the problems of material purity, density and area that have long plagued the preparation of carbon-based semiconductor materials. The purity has reached about 99.99997%. The density ranges from 5 nanometers to 10 nanometers, with 100 to 200 carbon nanotubes per micron. This material basically Possibility of making large-scale integrated circuits.
After the publication of this paper, Duke University professor Aaron Franklin said that 10 years ago he helped IBM determine the purity and density of carbon nanotubes, which many people thought could not be achieved. Now Peng Lianmao and Zhang Zhiyong’s team has achieved a breakthrough, “This is indeed an amazing achievement, a major leap forward for high-performance carbon nanotube transistors.” Aaron Franklin said.
The technical staff of Beijing Carbon-based Integrated Circuit Research Institute told reporters that carbon-based technology has better performance and lower power consumption than silicon-based technology, and the comprehensive advantage of performance and power consumption is 5 to 10 times, which means that carbon-based chips The performance is more than three generations ahead of silicon-based chips at the same technology node. For example, carbon-based chips using 90-nanometer technology are expected to produce silicon-based chips with performance and integration equivalent to 28-nanometer technology nodes; carbon-based chips using 28-nanometer technology can achieve silicon-based chips equivalent to 7-nanometer technology nodes. This opens another door for the global semiconductor industry, which is already on the verge of extreme values.
Regarding the breakthrough of the academician Peng Lianmao team, Chen Datong, the managing partner of Yuanhe Puhua and the chairman of the investment committee, said: “The research of the academician Peng team on carbon-based semiconductors is absolutely world-class original technology and forward-looking. There are very big advantages and opportunities in the field of materials and chips.” In 2019, Director Ye Tianchun of the Institute of Microelectronics of the Chinese Academy of Sciences said when he visited the 4-inch carbon-based semiconductor experimental line: “The research and industrialization of carbon-based semiconductors is the first in China. An indispensable and important part of the third-generation semiconductor industry.”
Carbon-based semiconductors need to cross the “death valley”
The progress of the world requires scientists to continuously discover new material laws and new theories, but turning a window into a new path requires a huge innovation chain to work together. We have been following the silicon-based technology route. On the carbon-based route, Chinese scientists have achieved world-class breakthroughs both theoretically and experimentally. Next, how do we go from the “123” in the laboratory, through the “Death Valley” of “456”, and into the “789” of industrialization?
The reporter still remembers their anxiety when interviewing Academician Peng Lianmao and Professor Zhang Zhiyong in 2017: A disruptive technology requires engineering research from the laboratory to the industry. Only engineered and mature technologies can the industry only Dare to take over.
The Beijing Institute of Carbon-Based Integrated Circuits was formally registered and established in September 2018. Its sponsors include Peking University and the Institute of Microelectronics of the Chinese Academy of Sciences. Academician Peng Lianmao serves as the dean. People in the industry know that the famous Belgian IMEC (Intercollegiate Microelectronics Research Center) laboratory was invested by the government in the early days, and now 80% of its income comes from enterprises. This top laboratory has made a huge contribution to the development of global integrated circuits. . Many industry giants, including Intel, ARM, and TSMC, are its customers. Before the new technologies of these giants enter the mass production line, the engineering of their new technologies is handed over to IMEC to complete. The development of carbon-based integrated circuits also requires such institutions. At present, the goal of Beijing Institute of Carbon-Based Integrated Circuits is to become the “IMEC” of the carbon-based integrated circuit industry.
Peng Lianmao told the “China Electronics News” reporter: “Beijing Carbon-based Integrated Circuit Research Institute hopes to continue to advance in the direction of engineering, do things with technology maturity from 4 to 8, and finally transfer the technology to enterprises. Industrialization And commercialization must be done by the company, and the research institute is doing technology research and development.”
It should be said that the launch of the “multiple purification and dimension-restricted self-assembly method” of the Beijing Institute of Carbon-Based Integrated Circuits has taken a big step forward from the laboratory to the industrialization of carbon-based technology. So what are the challenges in the next step? Which “Death Valley” needs to be traversed?
Gu Wenjun, the chief analyst of Xinmou Research, told the reporter of “China Electronics News”: “This is very exciting, but there is a long way to go from papers to new technologies to products to commodities, and we need to further increase research and development. At present, many new materials have been developed, including gallium nitride and silicon carbide, but in the long run, silicon is still difficult to replace.”
Han Xiaomin, an analyst in the semiconductor industry, believes that carbon-based is one of the important development directions of integrated circuits, but there are not many companies willing to follow up in the current industrial ecology. It is necessary to further break through the cost constraints, and to establish mature and standardized processes in terms of equipment and devices. In terms of product direction, areas that are not closely integrated with silicon-based chips are expected to be the first breakthroughs.
From laboratory to industrialization, what are the “traps” and challenges in the “valley of death” in the middle? Origin Quantum is a startup company in the field of quantum computing in China. Zhang Hui, deputy general manager of Origin Quantum, was interviewed by a reporter from China Electronics News Said that from scientific research products to industrial products, the challenges faced include continuous guarantee of funds, change of ideas and compatibility with existing industries. “Compatibility with existing industries is very important. If the existing semiconductor industry can borrow most of the instruments, equipment, and process flow, it will greatly increase the speed of industry follow-up.” He said.
In fact, the reason why the silicon-based integrated circuit industry has today’s rich and mature ecology requires huge investment in every link. Intel’s annual R&D investment accounts for more than 20% of sales revenue, and TSMC’s R&D investment in the past five years is 344 billion yuan. It is precisely because of this that it is difficult for the industrial chain to “abandon silicon and start over”, so compatibility is essential.
So, can the related technologies and process equipment processes in the silicon-based ecological chain, such as photolithography machines, software design tools, test instruments, and production process processes, be used on carbon? The answer given by Academician Peng Lianmao is: “Use The rate can reach about 80% to 90%, but the steps of cleaning and etching the carbon tube material require special treatment, and the Model of the carbon tube device needs to be established separately.”
At present, the problem of purity and density of carbon nanomaterials has been solved. “The next step also needs to ensure the process stability and uniformity of the material, and more importantly, the good compatibility with other devices and IC preparation. This is a comprehensive matter. Modern chip preparation has thousands of steps, one of which cannot be done. Good, there is no good product. Finally, there is a problem of system optimization. Materials, devices, chip design, etc. are inseparable.” Peng Lianmao said.
Carbon nanotubes have very good application prospects in the future. “Due to the special nature of carbon-based materials, it can make the circuit as soft as band-aids. If such flexible devices are used in the medical field, they will have a more comfortable examination experience for patients; in some extreme environments with high radiation and high temperature Here, robots made of carbon-based materials can better replace humans to perform high-risk tasks; if carbon-based technology is applied to smart phones, it will have lower power consumption and will extend the standby time.” Zhang Zhiyong Introduce it to reporters.
On May 26, Beijing Carbon-based Integrated Circuit Research Institute held a result release ceremony. Relevant personnel from the industrial research institutes of several large companies such as TCL were also present. What is the focus of the industry? “The industry is still concerned about when the technology will mature enough to be used, including engineering issues such as cost and reliability.” Peng Lian Spear said that this is what companies should do.
However, it is difficult for the industry to invest in a technology without seeing a return on investment. If carbon-based technology wants to be engineered, the Beijing Institute of Carbon-based Integrated Circuits needs to become the “IMEC” in the carbon-based field. The reporter learned that if we want to continue to move forward, the Beijing Carbon-based Integrated Circuit Research Institute will need about 200 million yuan of funds per year on the scale of 200 people and an experimental platform, and it needs to ensure more than ten years of capital investment. About 2 billion yuan. But until now, no company has paid attention to the value of the institute.
Not long ago, Alibaba announced that it would invest 200 billion yuan in research and development of chips and cloud operating systems in the next three years, while Tencent Cloud announced that it would invest 500 billion yuan in research and development of new infrastructure-related technologies. With the increasingly fierce competition in cloud computing, the overall layout from cloud operating systems to chips is becoming the choice of more and more giants. At present, Alibaba already has a chip company called “Pingtou Brother”. Is it possible for Tencent to enter the chip field in the future? If possible, I hope that Chinese giants can see such information and be able to pay attention to the “carbon-based “New opportunities for integrated circuits.