What does 28nm, 14nm, 7nm, 5nm mean? A look at the history of chip manufacturing reveals the first benefit of shrinking transistors: the smaller the transistor, the faster the speed, and this \"faster\" refers to the higher performance of the transistor based integrated circuit chip. Until 2004, the clock rate of the microprocessor CPU was basically increasing exponentially, and the main reason behind it was the shrinking size of the transistor.
< br >
The second benefit is increased functionality and reduced cost. When the size is reduced, the integration (the number of transistors per unit area) increases, both to increase the functionality of the chip and, according to Moore's Law, to decrease the cost as a direct result of the increase in integration.
< br >
This is also why the semiconductor industry has been pursuing Moore's law for 50 years, because if you don't meet this standard, your product will cost more than the competitor who can meet this standard, and your family will go out of business.
< br >
The third benefit is that transistor downsizing can reduce the power consumption of individual transistors, because the rules of downsizing require that at the same time, the supply voltage of the overall chip is reduced, thus reducing power consumption.
< br >
These are the main incentives for shrinking transistors, and the industry is still exploring and developing to obtain better performance, lower cost, and better function transistors.
< br >
Here is a brief history of the development of chip manufacturing enterprises:
< br >
1) In 2001, the chip manufacturing process at that time was 130 nm, and the Pentium 3 processor we used at that time was 130 nm.
< br >
2) In 2004, it was the first year of 90 nm, that year Pentium 4 used 90 nm process, and the performance was further improved.
< br >
At that time, there were many manufacturers that could achieve the 90-nanometer manufacturing process, such as Intel, Infineon, Texas Instruments, IBM, as well as UMC and TSMC.
< br >
3) In 2012, the process technology developed to 22 nm, at this time, Intel, UMC, Mediatek, GE Chip, TSMC, Samsung, etc., there are still many manufacturers in the world can reach 22 nm semiconductor process technology.
< br >
4) 2015 became a watershed in the development of chip manufacturing, when the process entered 14 nm, UMC (Taiwan UMC) stopped there.
< br >
5) In 2017, the process entered 10 nm, Intel fell in 10 nm, once Intel chip manufacturing process alone in the world, TSMC Samsung and so on are behind the chase.
< br >
However, when the process entered 10 nm, Intel's 10 nm chip could only be used on low-end machines, and Intel's main I5 and I7 processors could not be delivered due to yield problems.
< br >
In the field of 7 nm, Intel is still unable to break through, and another chip foundry giant in the United States, \"GF\
< br >
6) In 2018, the process stepped into 7 nm
< br >
Gf announced to give up 7 nm, in the previous article \"The enemy will not be merciful\
< br >
However, because of the cost and difficulty of 7nm research and development, GF finally decided to abandon 7nm.
< br >
Thus, the US government has included \"TSMC\" in the US military partners, and is ready to sign a chip foundry partnership agreement with the US government after 2024 with TSMC.
< br >
Because of 7-nanometer technology, TSMC is regarded as \"one of our own\" by the U.S. government, and in order to supply the United States for a long time, TSMC also announced a $12 billion plan to build a factory in the United States.
< br >
The United States' own OEM leader Intel fell at 10 nm, GF fell at 7 nm, and entered the more difficult 5 nm, leaving only Samsung and TSMC.
< br >
7) The introduction of 6nm mass production will be released in 2019, and the process will enter 5nm mass production in 2020
< br >
However, Samsung 5 nano was launched at the beginning of the year, and there is still a long way to go from mass production and high yield, the premise of chip foundry, first launch, trial production, and formal mass production, these three stages are more important than one.
< br >
Samsung's yield at 14 nm is not as good as TSMC's, its efficiency at 10 nm is not as good as TSMC's, and its R&D process at 7 nm is not as good as TSMC's.
< br >
You can only talk about success when you reach formal mass production and high yield, and TSMC is the only foundry in the world with the power to produce 5 nanometers.
< br >
Looking at the development of the entire chip process, it is really a lot of blood and tears, even if powerful foreign manufacturers such as IBM, Intel, and GF are also said to fall down and give up.
< br >
This is a very difficult project, there is a high probability of failure, and success requires a real sense of killing a bloody road.
< br >
8) TSMC plans to introduce 3nm into mass production in 2022, which is absolutely unique in the world
In July 2002, Gordon Moore, the founder of Moore's Law, which had declared the death penalty a few years earlier, gave an interview to reporters. However, this time, he expressed optimism, saying, The rate of doubling the number of transistors on a chip every 18 months is currently on a downward trend, but with the development of nanotechnology, Moore's Law will continue to take effect in the future.
< br >
Moore himself, it seems, is pinning his hopes on nanotechnology. Let's take a look at how nanotechnology can make nanochips.
< br >
The 20th century can be said to be the century of semiconductors, but also can be said to be the century of microelectronics, microelectronics technology refers to the use of micron and submicron fine structure technology on the semiconductor single crystal material (currently mainly silicon single crystal) sheet, the development of thousands of transistors and electronic components composed of miniature electronic circuits (called chips), And by different functions of the chip assembly into a variety of microelectronic instruments, instruments and computers. Chips can also be seen as integrated circuit blocks.
< br >
The development process of integrated circuit blocks from small scale to large scale can be seen as a process of continuous development to miniaturization. The small-scale integrated circuit developed in the late 1950s, its integration (the number of components contained in a chip) is 10 components; In the 1960s, it was developed into a medium-scale integrated circuit with an integration of 1000 components. In the 1970s, the large-scale integrated circuit was developed, and the integration reached 100,000 components; In the late 20th century, very large scale integrated circuits were developed, with an integration of more than 1 million components. Just in 1988, the United States International Business Machine Corporation (1BM) has successfully developed a storage capacity of 64 megabytes of dynamic random access memory, integrated circuit width of only 0.35 microns.
< br >
At present, the new products developed by the laboratory are 0. 25 microns, and toward 0? 1 micron march. By 2001 it had fallen to zero. One micron, or 100 nanometers. This will be the fourth major breakthrough in the history of electronic technology. Today, the integration of chips has been further improved to 10 million components. If the technology of the chip is further climbed, and the bar width of the integrated circuit is reduced, there will be a series of physical effects, which will limit the development of microelectronics technology.
< br >
In order to break through this barrier, in order to solve this difficulty, scientists have proposed the concept of nanoelectronics. This phenomenon shows that with the improvement of integrated circuit integration, the width of the chip is getting smaller and smaller, so the quality requirements of the monocrystal silicon material for the production of integrated circuits are getting higher and higher, even a particle of dust may destroy one or even several transistors, which is why Moore himself pronounced Moore's law \"death sentence\" a few years ago.
< br >
Experts predict that in the 21st century, humans will develop a computer that combines microprocessing chips with living cells. The core components of such computers are nanochips. A chip is a key component of a computer. At the heart of the development of life science and materials science, scientists are developing biochips, including protein chips and DNA chips.