In the field of science and technology, many breakthroughs are related to "more" and "bigger," but in the field of integrated circuits, there is a component that is constantly developing towards "smaller" size, namely the gate size of transistors.
If the central processing unit (CPU) is the heart of a computer, then the chip is its soul, determining almost all of the computer's system performance. Transistors are the core components of a chip; their smaller gate size means that more transistors can be integrated onto the chip, resulting in a leap in performance.
1. The transistor with the smallest gate length in history
Recently, Professor Ren Tianling's team from the School of Integrated Circuits at Tsinghua University published a paper entitled "Vertical molybdenum sulfide transistor with sub-1 nanometer gate length" online in the international academic journal Nature, making significant progress in the research of small-size transistors and realizing for the first time a transistor with a sub-1 nanometer gate length and good electrical performance.
Over the past decade, Ren Tianling's team has been dedicated to the research of two-dimensional material device technology, achieving innovative breakthroughs at multiple levels, including materials, device structure, processes, and system integration.
There is a famous law in the field of integrated circuits called "Moore's Law". Gordon Moore proposed in 1965 that the number of transistors that can be placed on an integrated circuit chip will double every 18-24 months, and the performance of microprocessors will double or the price will halve.
Over the past few decades, the gate size of transistors has been continuously shrinking driven by Moore's Law. However, in recent years, as the physical size of transistors has entered the nanoscale, problems such as reduced electron mobility, increased leakage current, and increased static power consumption have emerged one after another, making the development of new structures and new materials imminent.
“Currently, the gate size of transistors in the mainstream industry is above 12 nanometers, such as in the mobile phones we commonly use. But if the key dimensions of transistors can be further miniaturized, our electronic products will be more portable and have richer functions,” said Ren Tianling.
The academic community has explored ultra-short gate length transistors; however, the current limit for international research teams is only able to achieve planar molybdenum sulfide transistors with a gate length of 1 nanometer. "Transistors are the most basic starting point for chip production in both academia and industry. If we can do this smallest basic unit well, it will undoubtedly provide better support for a series of epitaxial functions," Ren Tianling told a reporter from China Youth Daily. Further breaking through the bottleneck of gate length transistors below 1 nanometer has become a key issue that the team is determined to solve.
As early as 2018, Ren Tianling's team proposed an idea to use the edges of few-layer or single-layer graphene as transistor gates to realize a new type of transistor. In the early stages of the experiment, due to the special physical properties of graphene, the team needed to further optimize the overall device process to minimize the process steps and increase its reliability under laboratory conditions.
However, in this field where few research teams have ventured, the team encountered numerous difficulties due to the lack of prior experimental references and the anticipated unknowns. Even a seemingly simple manufacturing process required the team to refine it repeatedly over a long period of time.
"Actually, we took many detours and experienced numerous failures, big and small. Working on experiments until the early hours of the morning was commonplace for team members," Ren Tianling said.
Recalling this stumbling and groping experience, Ren Tianling repeatedly mentioned two words: "perseverance" and "teamwork".
Through the tireless efforts of the team members, they ingeniously utilized the ultrathin single-atom layer thickness and excellent conductivity of graphene film as a gate, controlling the switching of the vertical MoS2 channel through the lateral electric field of graphene, finally achieving a minimum equivalent physical gate length of 0.34 nanometers. The team successfully pushed Moore's Law further to the sub-nanometer level, while also providing a reference for the application of two-dimensional thin films in future integrated circuits.
2. Smaller transistors generally have better performance.
Over the past century, electronic technology has developed rapidly. In 1920, the best medium-wave radio contained multiple vacuum tubes, many large inductors, capacitors and resistors, tens of meters of wire as the receiving antenna, and the battery pack used for power supply took up a lot of space.
Today, radios that can pick up a dozen stations can easily fit in your pocket, and their auxiliary functions are extremely rich. But the reduction in size is not just for portability: it is a key factor in achieving the high performance we expect.
The most obvious benefit of reducing component size is the ability to perform more functions within the same volume, which is especially crucial for digital circuits: more components mean you can do more in the same amount of time. For example, a 64-bit processor can theoretically perform eight times the information processing of an 8-bit processor at the same clock frequency. To achieve this, it also requires eight times the number of electronic components: registers, accumulators, bus width, and other parts will all increase eightfold. Therefore, you need a chip that is eight times larger, or the components that make up the circuit that are eight times smaller.
The same applies to memory: smaller electronic components can store more information in the same volume. Current display pixels are made of thin-film transistors, so reducing the size of the transistors increases the display's resolution. However, an even more crucial advantage comes from smaller transistors: their performance also improves significantly with smaller size.
What will the future be like?
The frequently recurring claim in technology magazines is the failure of Moore's Law, or the constant explanations for these inaccurate predictions. However, the reality is that the three giants of the semiconductor industry—Intel, Samsung, and TSMC—are still striving to compress more devices into smaller sizes and planning for future chip process improvements. While not as dramatic as the pace of improvement 20 years ago, the reduction in transistor size continues unabated.
However, for other discrete devices, we seem to have reached their natural limits: reducing size not only fails to improve performance, but also exceeds the requirements of most applications.
It seems there is no Moore's Law for discrete components. If there were, I would prefer to see someone take on the challenge of soldering these labeled components.
