Military Industry of Great Powers: Build the Strongest Flanker

102 Institute and the Shanghai Microelectronics Research Institute, plus 37 scientific research units and 62 factories across the country, a total of more than 2,000 people participated in R\u0026D and manufacturing, plus the key workbench technology exchanged with ASML, after six years Over time, six photolithography machines were dismantled, with a total investment of billions of dollars, and finally bore fruitful results.

In November 1993, the 193nm argon fluoride laser lithography machine completed technical research. and passed national demonstration.

There are two models, codenamed GK193BJ and GK193TY.

GK193TY is a projection lithography machine. The main research and development unit is the Shanghai Microelectronics Research Institute. The key technology of the workbench is the reverse ASML PA3000 lithography machine. The light source uses the argon fluoride laser light source provided by 102, and the lens system is also 102. Designed and manufactured.

The advantages of the GK193TY projection lithography machine are high precision, accurate projection, and good controllability. However, the disadvantage is that it is not very efficient. It takes dozens of times as long as a stepper lithography machine to complete the lithography process of a wafer.

In terms of commercial use, projection lithography machines are an obsolete technology, but they still have room for use in military applications where cost is not a concern, or in high-performance large computers.

As for the GK193BJ stepper lithography machine, the main technology comes from the technology exchanged with ASML, Siemens and Philips. The main R\u0026D unit is 102 Research Institute.

Contrary to the projection lithography machine, the biggest advantage of the stepper lithography machine is that it has a very high lithography efficiency without sacrificing a certain stability and ensuring the yield rate. It is most suitable for commercialization and can significantly reduce the chip cost. the cost of……

Testing of the new lithography machine has already begun. The test process is 130 nanometers, FinFET semiconductor structure, direct lithography.

Facts have proved that the FinFET semiconductor structure has excellent physical properties and can be directly applied to 130 nanometers without any changes at all.

The speed of the ARMv6 processor manufactured using the 130nm process soared by 1.5GHZ, directly breaking the 1GHZ speed mark.

The technical research team of Institute 102 continues to explore the 90-nanometer process. The GK193BJ stepper lithography machine successfully completed the exploration of the 90-nanometer process, but the yield rate dropped directly to 32%.

Using the GK193TY projection lithography machine to complete lithography in the 90-nanometer process, the yield rate can be doubled to 67%.

Using multiple exposure technology to explore the 65nm process, the GK193BJ stepper lithography machine is already very weak, and the yield rate has collapsed again. A three-inch wafer contains hundreds of chips, and one or two of them happen to be qualified, and the yield rate is already less than 1 %!

The GK193TY projection lithography machine can still maintain a yield rate of 42%.

Even on the 45nm process, stepper lithography machines are no longer useful, but the GK193TY projection lithography machine can still achieve a yield rate of 14%.

The 45nm process is already the limit of the 193nm wavelength argon fluoride laser lithography machine, and the 32nm process encounters very serious optical interference physical phenomena.

Of course, in theory, using multiple exposure technology, it is possible to complete photolithography of the 32nm process. However, in theory, using the 32nm process to manufacture an ARMv6 processor with only 20 million transistors requires more than a hundred masks. , there are more than 300 processes.

At this time, the efficiency of etching will be seriously reduced, not to mention the more than 300 processes, which will reduce the yield rate to less than 1‰. Creating a qualified chip with a 32nm process is almost like winning a lottery, and there is no point in exploring it.

Teacher Xiao Zhao is quite satisfied with these test data. The 45nm process is the limit.

For commercial use, the 130nm process can still ensure efficiency and yield, but the commercial 90nm process still needs to be explored.

Chief Engineer Qu Qinglin still has a rather vicious vision. He said to Mr. Zhao: "Mr. Zhao, our lithography machine is powerful, but the FinFET semiconductor structure is the god. When I was at Texas Instruments, the new process required key technologies. There are many, but FinFET technology can directly transition from a one-micron process to 45 nanometers..."

Qu Qinglin stretched out his fingers and counted while saying: "One micron, 800 nanometers, 500 nanometers, 350 nanometers... 65 nanometers... 45 nanometers, good guy, the manufacturing process has been improved by ten generations, and the semiconductor structure remains unchanged. Mr. Zhao, you are like this It’s a trick that’s popular all over the world, this FinFET semiconductor structure is really amazing!”

Chief Engineer Qu is a talent who has returned from overseas. He has worked for IBM and Texas Instruments for many years. He is from Bay Island and graduated from the University of California, Berkeley. Of course, he has been reviewed many times before he can get into this place and there is no problem.

Zhao Guoqing felt guilty because of what Chief Qu said just now. To be honest, if anyone carefully studies China's development in semiconductors over the years, they can find that China's semiconductor technology has developed by leaps and bounds.

A FinFET structure, in the eyes of outsiders, may be attributed to bad luck, but in the eyes of semiconductor practitioners, it is different. Teacher Xiao Zhao is no longer a locomotive leading everyone forward, but a rocket head.

But Zhao Guoqing did not take it seriously. He said: "This 193-nanometer stepper lithography machine is a major technological breakthrough. However, the international technology is changing rapidly, and we cannot slack off. The main work of the 102nd Institute of Lithography Research Laboratory is to continue research and development. A more stable workbench system, followed by...a duplex stepper lithography machine...should be put on the agenda!"

It was the first time Qu Qinglin heard the name duplex, which literally means two workstations. He didn't know what the concept was, but from what Teacher Xiao Zhao said, it must be something extraordinary. He said doubtfully: "Director Zhao, what is a duplex station?"

Zhao Guoqing said: “When the photolithography machine is working, it takes ten minutes for calibration, and it also takes dozens of minutes for photolithography. However, if the wafer is being photolithographed on one workbench, the wafer on another workbench will not be processed. Can this time be saved by calibrating the wafer? After the wafer is photolithographed, the other wafer can be calibrated just right for photolithography..."

"..." Qu Qinglin frowned slightly. After thinking about it carefully, he understood the beauty of it. With a duplex stage, one calibration and one photolithography, the efficiency will be significantly improved. If the yield rate can be guaranteed, the most direct effect is that the output can be increased. has risen significantly, while the cost of chips can continue to decline.

Zhao Guoqing saw Qu Qinglin's expression of embarrassment and said with a smile, "What, are you scared?"

Qu Qinglin shook his head and said, "I haven't figured out the simplex platform yet! Mr. Zhao, isn't it a little early to develop a duplex platform now?"

"Indeed, the simplex platform technology mainly absorbs the technology of ASML. We only know how to do some technologies, but we don't understand why ASML designed it this way!" Zhao Guoqing paused and said, "But if we continue to study the simplex platform in depth, it will not affect the duplex platform. Research, at least do some preliminary exploratory research!"

Qu Qinglin said: "Mr. Zhao, are you so optimistic about duplex technology?"

Teacher Xiao Zhao nodded. It had been a long time since anyone had questioned his decision.