Samsung announced that it has joined the 11nm process, performance than the previous 14nm increased by 15%, the power consumption per unit area reduced by 10%. If you want to follow Moore's Law to continue, the future of semiconductor technology will have much room for improvement?

        10 years ago we felt 65nm process is the limit, because to the 65nm silicon dioxide insulation layer leakage has been intolerable. So the industry out of the HKMG, with high-k medium to replace the silicon dioxide, the traditional polysilicon - silica - monocrystalline silicon structure into a metal - highK - monocrystalline silicon structure. 5 years ago we think 22nm process is the limit, because to 22nm channel off leakage has been intolerable. So the industry out of the FinFET and FD-SOI, the former with three-dimensional structure to replace the planar device to enhance the control of the gate, which uses the oxide buried layer to reduce leakage. Now we think the 7nm process is the limit, because even to the 7nm node even FinFET is not enough to ensure the performance while suppressing leakage. So the industry with indium gallium arsenide instead of monocrystalline silicon channel to improve device performance. When we say that the process to the limit, we are actually in the existing structure, materials and equipment to the limit. However, each time a bottleneck encounters, the industry will introduce new materials or structures to overcome the limitations of traditional crafts. Of course, the cost of this is also amazing, the complexity of each generation of technology and costs are rising.

Source: Gate Gate: Drain

working principle

        A chip integrates millions of transistors, and the transistor is actually a switch, the transistor can affect the state of each other to deal with information. The gate of the transistor controls whether the current flows from the source to the drain. Electronic flows through the transistor logic is "1", does not flow through the transistor is "0", "1", "0" on behalf of the open and close the two states. In the current chip, the connection transistor source and drain are silicon elements. Silicon is called a semiconductor because it can be a conductor or an insulator. The voltage across the transistor gate controls whether the current can pass through the transistor.

Th eMoore's Law

        In order to keep up with the rhythm of Moore's Law, engineers must continue to shrink the size of the transistor. But as the size of the transistor shrinks, the channel between the source and the gate is also shortened, and when the channel is shortened to a certain extent, the quantum tunneling effect becomes extremely easy, in other words, even if no voltage , The source and drain can be considered interoperable, then the transistor will lose the role of the switch itself, and therefore can not achieve the logic circuit. From now on, the 10nm process is able to achieve, 7nm also has a certain technical support, and 5nm is the physical limit of existing semiconductor technology.

        Silicon chip technology since its inception, has been followed by the rapid development of Moore's Law. But Moore's Law, after all, is not a real law of physics, and more is a speculation or explanation of the phenomenon, we can not expect the semiconductor process can always follow Moore's Law said to develop. But in order to continue as much as possible Moore's Law, researchers are also trying to find ways, such as the replacement of silicon materials to continue to improve the chip's integration and performance. Next we talk about several future possible to replace silicon, a new semiconductor materials program.

III-V compound material

        May be at the 7nm node to abandon the traditional silicon chip technology, and in the next few years to enable the new semiconductor materials as a successor, it seems that this new material is likely to be III-V compound semiconductor. The semiconductor material is a silicon-fin that replaces the FinFET with a III-V compound. As the III-V compound semiconductor has greater energy gap and higher electron mobility than silicon, the new material can withstand higher Operating temperature and running at higher frequencies. Intel has long tried to compound III-V compounds (indium phosphide and indium gallium arsenide) with traditional wafers integrated semiconductor. More than a year ago, IMEC (Microelectronics Research Center, including Intel, IBM, TSMC, Samsung and other semiconductor industry giants) has announced the successful integration of indium phosphide and indium gallium arsenide on 300mm 22nm wafers, Compound semiconductors.

The Group III-V compound becomes fins on the FinFET

        Compared with other alternative materials, III-V compound semiconductors have no obvious physical defects, and similar to the current silicon chip process, many of the existing technology can be applied to new materials, it is also considered to continue to replace after 10nm Ideal for silicon. At present, the biggest problem to be solved is probably how to increase wafer production and reduce process costs.

1、graphene

Under the electron microscope, graphene, was hexagonal structure

        Graphene is considered a fantastic material, it has a strong conductivity, can be bent, high strength, these features can be applied in various fields, Harvatek even have the potential to change the future of the world, there are many people to It is replaced by silicon, as the future of semiconductor materials. But really apply it to the semiconductor field, but also need to overcome a lot of difficulties.

2、silene

        Carbon has the same chemical properties as silicon, and in fact, in the air, the silane has a very strong instability, even in the laboratory, the silane storage time is very short. If you want to make silicate transistors, but also need to try to add protective coating and other means to ensure that silole will not degeneration, it may be applied to the actual. Although the application of silane is faced with difficulties, but it is still hope to catch up with the older brother of graphene, become the ideal semiconductor materials.

Silylenes with similar structures may be better than graphene

        Scientific research has always been in practice for many years before, and there are many new directions in trying to break through. Such as trivalent pentavalent semiconductors, carbon nanotubes, and quantum tunneling studies. In fact, the chip itself in the architecture also has great potential to be tapped, the calculation of performance is not only dead knock process. Perhaps only quantum computing, photon computing is the ultimate destination.