Before 2016, some voices in the chip industry believed that 7nm was the physical limit of silicon chip technology; Now, the 7nm process silicon chip has come out, and a new voice has emerged: 3nm is the limit.
But how to develop chips below 3nm? At present, the major semiconductor manufacturers have no clear answer.
The development of the chip industry is facing a bottleneck. The introduction of a new architecture is a major solution, but the more “fundamental” approach may be to find new materials that can replace silicon.
In fact, silicon is not the first choice from the beginning. Before silicon, the mainstream chip material is germanium, but there are three main “hard wounds” of germanium:
First, the content is low. The content of germanium in the earth’s crust is only seven parts of one million and is extremely dispersed, so it is called “rare metal”. Without a centralized “germanium mine”, the cost of germanium mining is very high, and it is difficult to realize large-scale production.
Second, it is very difficult to extract high-purity germanium. The direct result of insufficient purity is that chip performance is difficult to improve.
Third, the stability is poor. The chips using germanium transistors can only withstand the high temperature of about 80 ° C at most, while the buyers of early chips are mainly the government and military. These customers usually require the products to withstand the high temperature of 200 ° C, and the germanium chips obviously cannot.
These “hard wounds” of germanium happen to be the strength of silicon. The content of silicon in the crust is as high as 26.3%, second only to oxygen (48.6%), and is the second highest element. In terms of stability, silicon is also stronger than germanium. In terms of purification, the current technology has been able to infinitely improve the purity to nearly 100%.
Since Xiantong Company invented the plane processing technology of silicon transistor, which made the application of silicon in chips simple and efficient, silicon has gradually become the mainstream material of chip manufacturing.
What are the problems with silicon today?
The trend of chip development can be simply summarized as: smaller size, stronger performance. To develop in this direction, it is necessary to make the number of integrated components on the unit area of the chip more. The smaller the size of components, the more components can be integrated on the chip, which also means higher manufacturing difficulty.
This is like a room getting smaller and smaller, but there are more and more things to be loaded in it. No matter what kind of “storage” method is adopted, one day, the room will be “overloaded”.
Leakage and poor heat dissipation are the problems caused by silicon chip “overload”.
In fact, these two problems did not appear suddenly when the current chip manufacturing process reached 7nm. In the development of silicon chips, such problems have also appeared many times, but major manufacturers have skillfully solved them in various ways. One way is to use new materials, such as germanium and silicon as channel materials. However, how to integrate different materials onto silicon substrate is also a challenge.
Even though the current 7nm chip does not reach the physical limit of silicon chip as predicted in the past, the physical limit of silicon is inevitable.
The next “fundamental” breakthrough of chips is to find new materials.
What types of materials are expected to become mainstream?
Recently, the “2020 Top Ten Science and Technology Trends” released by the Dharma Institute predicted that the new physical mechanism of new materials will realize new logic, storage and interconnection concepts and devices, and promote the innovation of the semiconductor industry. For example, topological insulators and two-dimensional superconducting materials can realize lossless electron transport and spin transport, and can become the basis of new high-performance logic devices and interconnection devices.
In 2016, Zhang Shousheng, a famous Chinese-American scientist, announced that a new topological insulator material had been discovered and that the material had been successfully tested. If this new topological insulator material can finally be successfully applied to the semiconductor and chip industry, it will bring huge business opportunities and will also create a new era of Silicon Valley.
Of course, the application of topological insulators in the field of chip manufacturing is still at an early stage of research and development. It is known that it can solve the heating problem of electronic chips, but there are still many areas to explore how and where this material can be used in the future. In addition, topological insulators are more expected to be applied to quantum chips, but quantum chips and classical chips are totally different fields.
Another new material, two-dimensional superconductive material, is currently a hot topic in the semiconductor industry. Two-dimensional materials include graphene, phosphorene, boronene, etc. These materials are more promising to become mainstream materials.
Among them, graphene is the most prominent one.
The reason why graphene is valued is that it has another identity – carbon nano-materials besides the “two-dimensional” attribute. As early as 2012, IEEE (Institute of Electrical and Electronic Engineers) wrote in Beyond Moore that the future semiconductor industry may enter the “carbon era” from the “silicon era”. The carbon nano-material graphene may replace the original silicon-based material in the future.
Carbon nanotubes are seamless, hollow tubes rolled from graphene sheets, with excellent conductivity and thin wall. Therefore, in theory, under the same degree of integration, carbon nanotube chips can be smaller than silicon chips; In addition, carbon nanotubes produce little heat and have good thermal conductivity, which can reduce energy consumption; In addition, considering the mining cost, carbon is widely distributed and the acquisition cost is not high.
More importantly, graphene, as the first two-dimensional material discovered by human beings, has been applied in screens, batteries and wearable devices. The research of graphene has reached a relatively mature stage. Therefore, graphene is most likely to become a new mainstream chip material and replace silicon.
At present, the traditional chips independently developed by China are still in the process of commercial mass production and transformation of 28nm to 14nm manufacturing process, and there is still a large gap with the international advanced level. But this is based on silicon as the main material of chips. If carbon nanomaterials can become the mainstream in the future, will China have the opportunity to overtake at this node?
First of all, there is a small gap between China’s research on carbon nanotube devices and the international advanced level. IBM was the first to realize the preparation of carbon nanotube devices in the world. In 2017, IBM reduced the transistor size to 40 nm by using carbon nanotubes; In the same year, Peking University developed a 120nm carbon nanotube transistor. The transconductance at 0.8 volts is the highest value among the published carbon nanotube devices.
But the development of any technology will not be smooth sailing. The application of graphene in the field of chip manufacturing faces three major problems: first, it is difficult to obtain high-purity graphene at present; Secondly, the manufacturing of graphene wafers is also very difficult. Although China has taken the lead in the large-scale preparation of graphene single wafers, under the current manufacturing process, it is still prone to wrinkles, spot defects and pollution; Finally, in order to ensure that graphene incorporated into the device can continue to maintain its excellent performance, other relevant manufacturing processes and materials also need to be iterated with it to ensure the stable operation of graphene chip.
To sum up, graphene is still a relatively new material. If graphene can truly replace silicon and become the mainstream material of chips, there are still many problems to be solved in terms of manufacturing process and technical follow-up of supporting devices.
For China, there is no doubt that there is an opportunity to overtake on the curve at the node where technology changes from old to new. But the problem that needs to be faced up to is that China is often too enthusiastic about new concepts. Graphene has great potential in many fields such as chips, but from the discovery of potential to industrialization, down-to-earth research is needed, and the public’s tolerance for the failure of new technology research and development is also needed.
The curve overtaking of Chinese chips requires not only “new” materials, but also “new” industry-university-research system and a more open and inclusive investment environment.