Talking about the core from the shallow to the deep

　　Samsung announced that it has adopted the 11nm  process , which has improved performance by 15% compared to the previous 14nm process and reduced power consumption per unit area by 10%. If Moore's Law continues to be followed, how much room for improvement will there be in future semiconductor technology? Let's follow the mobile phone portable editor to learn about the relevant content.

　　Ten years ago, we thought that the 65nm process was the limit, because at the 65nm node, the leakage of the silicon dioxide insulation layer was intolerable. So the industry came up with HKMG, replacing silicon dioxide with high-k dielectric, and the traditional polysilicon-silicon dioxide-monocrystalline silicon structure became a metal-highK-monocrystalline silicon structure. Five years ago, we thought that the 22nm process was the limit, because at 22nm, the channel turn-off leakage was intolerable. So the industry came up with FinFET and FD-SOI, the former replaced the planar device with a three-dimensional structure to enhance the control capability of the gate, and the latter used an oxide buried layer to reduce leakage. Now we think that the 7nm process is the limit, because at the 7nm node, even FinFET is not enough to suppress leakage while ensuring performance. So the industry replaced the monocrystalline silicon channel with indium gallium arsenide to improve device performance. When we say that the process has reached its limit, we are actually saying that it has reached the limit under the existing structure, materials and equipment. However, every time a bottleneck is encountered, the industry will introduce new materials or structures to overcome the limitations of traditional processes. Of course, the cost is staggering, with the complexity and cost of each generation of technology increasing.

Source: Source Gate: Gate Drain: Drain

　　working principle

　　Millions of transistors are integrated on a chip, and a transistor is actually a switch. Transistors can process information by affecting each other's state. The gate of the transistor controls whether the current can flow from the source to the drain. The flow of electrons through the transistor is logically "1", and the absence of electrons through the transistor is "0". "1" and "0" represent the on and off states respectively. In current chips, the element connecting the source and drain of the transistor is silicon. Silicon is called a semiconductor because it can be a conductor or an insulator. The voltage on the gate of the transistor controls whether the current can pass through the transistor.

　　Moore's Law

　　In order to keep up with the pace of Moore's Law , engineers must continue to reduce the size of transistors. However, as the size of transistors decreases, the channel between the source and the gate is also shortening. When the channel is shortened to a certain extent, the quantum tunneling effect becomes extremely easy. In other words, even if there is no voltage, the source and the drain can be considered to be interconnected, then the transistor loses its own switching function, and therefore cannot realize the logic circuit. From now on, the 10nm process is achievable, 7nm also has certain technical support, and 5nm is the physical limit of the existing semiconductor process.

　　Since its inception, silicon chip technology has been developing rapidly in accordance with Moore's Law. But Moore's Law is not a true physical law after all, but more of a speculation or explanation of the phenomenon. We cannot expect semiconductor technology to develop forever in accordance with Moore's Law. However, in order to continue Moore's Law as much as possible, researchers are also trying their best, such as seeking alternative materials to silicon, to continue to improve the integration and performance of chips. Next, let's talk about several new semiconductor material solutions that may replace silicon in the future.

　　III-V Compound Materials

　　It is possible that the traditional silicon chip process will be abandoned at the 7nm node, and a new semiconductor material will be used as a successor in the next few years. At present, it seems that this new material is likely to be a III-V compound semiconductor. This semiconductor material replaces the silicon fins on the FinFET with III-V compounds. Compared with silicon, since III-V compound semiconductors have a larger energy gap and higher electron mobility, the new material can withstand higher operating temperatures and run at higher frequencies. Intel has long tried to integrate compound semiconductors of III-V compounds (indium phosphide and indium gallium arsenide) with traditional wafers. More than a year ago, IMEC (Microelectronics Research Center, whose members include semiconductor industry giants such as Intel, IBM, TSMC, and Samsung) has announced the successful integration of indium phosphide and indium gallium arsenide on 300mm 22nm wafers to develop FinFET compound semiconductors.

III-V compounds become fins on FinFETs

　　Compared with other alternative materials, III-V compound semiconductors have no obvious physical defects and are similar to the current silicon chip process. Many existing technologies can be applied to new materials, so they are also regarded as ideal materials to continue to replace silicon after 10nm. The biggest problem that needs to be solved now is probably how to increase wafer yield and reduce process costs.

　　1. Graphene

Graphene under an electron microscope shows a hexagonal structure

　　Graphene is regarded as a dream material. It has strong conductivity, bendability and high strength. These characteristics can be applied to various fields and even have the potential to change the future world. Many people also regard it as a semiconductor material to replace silicon. However, it still needs to overcome many difficulties to truly apply it to the semiconductor field.

　　2. Silicene

　　Carbon has the same chemical properties as silicon, and in fact, silicene is extremely unstable in the air, and even in the laboratory, silicene has a short shelf life. If you want to make silicene transistors, you need to try to add protective coatings and other means to ensure that silicene does not denature before it can be used in practice. Although the application of silicene faces many difficulties, it still has the hope of surpassing its big brother graphene and becoming an ideal semiconductor material.

Silicene, which has a similar structure, may be a better solution than graphene

　　Conclusion

　　Scientific research always precedes practical application by many years, and there are already many new directions trying to make breakthroughs. For example, research on trivalent and pentavalent semiconductors, carbon nanotubes, and quantum tunneling. In fact, there is still a lot of potential to be tapped in the chip architecture itself, and computing performance is not just about the process. Perhaps only quantum computing and photonic computing are the ultimate destination.

    The above is an introduction about the portability of mobile phones - a comprehensive discussion on the core of mobile phones. If you want to know more relevant information, please pay more attention to eeworld. eeworld Electronic Engineering will provide you with more complete, detailed and updated information.