Galactic Tech Empire

Chapter 85 Chip Problems

Huang Haojie has seen the construction progress of the X-ray laser base in Luhe County.

Judging from the current situation, due to the mature technology, the project is expected to be completed in November to December this year.

Then adjust and conduct preliminary experiments for two or three months, that is to say, the iron-silver room-temperature superconductor is expected to be mass-produced on a small scale in April next year.

Once there is a room temperature superconductor, the chip of Yinhe Technology must be launched.

So this time Zhang Rujing's resignation is an opportunity for Huang Haojie.

He needs someone who knows everything about the chip industry to take the helm of Galaxy Technology's chip company.

And Zhang Rujing is very suitable. The other party is very good at building factories in the chip industry, and this is exactly what Huang Haojie needs.

In addition, the character of the other party is not bad, Huang Haojie decided to contact the other party to see if he can attract Zhang Rujing to Yinhe Technology.

However, due to the very low data integrity of superconducting quantum chips, it is only about 8% so far.

Huang Haojie has decided to produce superconducting electronic chips first.

In fact, the processes of superconducting quantum chips and electronic chips are similar.

Since the popularity of computer technology in the second half of the 20th century, human beings have entered the information age.

With the increasing integration of computer chips and the smaller and smaller components, integrated circuit technology is now approaching its limit, and Moore's Law is about to expire, and the performance of electronic chips has almost reached its limit.

And although the speed of computers is increasing day by day, there are some problems that computers cannot solve at all, such as the factorization of large numbers. In theory, as long as one number is large enough, this problem is enough for the fastest computer to work for hundreds of millions of years.

So a new computer must be on the agenda.

At present, there are three main roads on the route of future computers, namely: quantum computer, photonic computer, and biological computer.

Quantum computers are also divided into many branches, and superconducting quantum computers are currently the most promising. Others include photons, ion traps, superconducting circuits, diamond color centers and semiconductor quantum dots, which are all promising to be used as qubits. That is quantum computers.

Photonic computers also have great potential. Of course, like quantum computers, materials are a big problem, and how to design them is also very troublesome.

As for the biological computer, this thing is also called DNA computer or molecular computer. Foreigners have already developed it, and the calculation speed is also extraordinary, but how to read the data is a huge problem.

And Huang Haojie now has room temperature superconductor technology in his hands, of course he is inclined to superconducting quantum computers.

In addition, the 8% quantum chip technical information is not completely unusable.

He found that one of the technologies can be used in the chip process.

Let me just talk about the current electronic computer chip technology, the chip production process and the level of Milijia (because Dongdao and the mainland are now in a state of division, so the chip technology of Dongdao is not the same as that of Dongtang).

Silicon, this thing needs to be chlorinated and then distilled to obtain silicon with high purity. Cutting into slices is the silicon wafer we want.

The evaluation index of silicon is purity. If you think about it, if there are a lot of impurities in silicon, then electrons will not be able to run smoothly.

Solar-grade high-purity silicon requires 99.9999%. More than half of this stuff in the world is produced in Dongtang, and it has long been played as the price of cabbage.

The electronic-grade high-purity silicon used in chips requires 99.999999999% (don’t count, 11 nines, also known as 9N), almost entirely dependent on imports

I heard that Xinhua Company in Suzhou Province is doing research and development, and plans to initially achieve an annual output of 0.5 tons, while Dongtang imports 15 tons a year.

The traditional overlords of high-purity silicon are still Hans Wacker (Wacker Chemicals) and Hemlock (a joint venture between Mi and Japan), and Dongtang has a long way to go.

Next is the wafer, which needs to be rotated during silicon purification, and the finished product is cylindrical.Therefore, the sliced ​​silicon wafer is also round, so it is called "wafer".

After cutting, it is necessary to assemble thousands of circuits on the wafer. The one who does this work is called a "fab".

So with the current human technology, how can this kind of operation be done?With atomic manipulation?Maybe Huang Haojie in parallel time and space can use nano-robots to complete it, as for now, just think about it.

The process of wafer processing is a bit cumbersome.

First, a layer of photosensitive material is coated on the wafer. The material melts when exposed to light. Where does the light come from?The photolithography machine can use very precise light to carve patterns on the photosensitive material, exposing the underlying wafer.

Then, by rinsing with something like plasma, the exposed wafer will be carved with many grooves. This set of equipment is called an etching machine.

Doping phosphorus in the trench, you get a bunch of N-type semiconductors.

After the completion, clean it, recoat it with photosensitive material, use a photolithography machine to carve a pattern, use an etching machine to carve a groove, and then sprinkle boron, and there will be a P-type semiconductor.

The actual process is more complicated, and this is the general principle.It's a bit like 3D printing, where wires and other devices are put in layer by layer.

So why not make the chip bigger?Wouldn't this allow more circuits to be installed?Doesn't the performance catch up with foreign countries?

The answer is surprisingly simple: money!For a wafer with a diameter of 300mm, 16 chips can be produced by 100nm process, and 10 chips can be produced by 210nm process, so the price is half cheaper. It can hold down competitors in the market, and can make money and make With more research and development, the gap widens.

However, Dongtang military chips are basically self-sufficient, because rabbits don't care about money!Chips can be made very large.

In addition, the larger the silicon wafer, the greater the probability of encountering impurities, so the larger the chip, the lower the yield rate.In general, the cost of large chips is much higher than that of small chips, but for Juntu, this is not a problem.

After all, safety comes first, spending money is better than being choked by someone.

The chip yield rate depends on the overall level of the fab, but the processing accuracy depends entirely on the core equipment, which is the "lithography machine" mentioned above.

Lithography machine, the Netherlands-ASML company (ASML) swept the world!Sorry, the output is not high, just wait slowly!Whether it is Taijidian, Sanxin, or Intel, whoever buys ASML's lithography machine first will be the first to have the 7nm process.No way, it's just so powerful!
Nikon and Canon in Sunland also make lithography machines, but their technology is far inferior to ASML. In recent years, they have been beaten by ASML and can only grab shares in the low-end market.

ASML is the only manufacturer of high-end lithography machines, each with a price of at least 1 million U.S. dollars. In 2017, only 12 units were produced. In 2018, it is expected to produce 24 units. These have been snatched up by TSMC, Samsung and Intel. In 2019 It is predicted that there will be 40 units, one of which is for Huaxin International.

Since it's so important, can't we pay more?
First: Intel has a 15% stake in ASML, Taiji Electric has 5%, and Sanxin has 3%. Sometimes, money is not everything.Second, Dongtang established the "Wassenaar Agreement", and sensitive technologies cannot be sold. Dongtang, North Colliers, Persia, and Libya are all restricted countries.

Interestingly, Shanghai Microelectronics successfully developed a 2009nm lithography machine in 90 (the core components were imported). Only when countries allow equipment above 2010nm to be sold to Dongtang, SMIC has the opportunity to miss a high-end machine.

Of course, the reasons for this are self-evident. The reason why the restrictions were lifted was mainly to hit Dongtang enterprises, so that Dongtang enterprises could not make profits, thus falling into a vicious circle.

But we don't need to be discouraged, any real estate company can easily surpass ASML in sales.

The importance of the etching machine is second only to the lithography machine. The situation of Dongtang is much better. The 16nm etching machine has been put into mass production and operation, and the 7-10nm etching machine is also on the way. Don't etcher the blockade.

Implanting elements such as boron and phosphorus on the wafer requires the use of an "ion implanter". This year, it seems that the first domestically-made commercial machine will be used in China, and the level is unknown.

70% of the market share of ion implanters is owned by Milijia Applied Materials.

Coating photosensitive materials requires the use of "glue developing machines", and Sun Country Tokyo Electronics Co., Ltd. has taken 90% of the market share.

Even the auxiliary materials such as photoresist are almost monopolized by Sun Country Shin-Etsu and Mijia Dow.

After the chip is ready, it has to be cut from the wafer, connected to the wires, installed in the casing, and tested by the way. This is called packaging and testing.

Packaging and testing is the world of East Island. ASE ranks first in the world, followed by a bunch of powerful younger brothers: Sipin, Licheng, Nanmao, Xinbang, and KYEC.

Dongtang's three packaging and testing giants, Changdian Technology, Huatian Technology, and Tongfu Microelectronics, are all doing well. After all, they are only at the end of the chip industry and their technical content is not high.

Silicon raw materials, chip design, wafer processing, packaging and testing, and related semiconductor equipment, Dongtang is still in a state of "heavy responsibilities and a long way to go" in most fields.

How long will this state of confusion last?According to the theory of "burning money and burning time", it will be about 2030!

The "Outline for the Development of the Integrated Circuit Industry" issued by the Dongtang Cabinet clearly stated that by 2030, the main links of the integrated circuit industry chain will reach the international advanced level, a number of enterprises will enter the international first echelon, and the industry will achieve leapfrog development.

At present, the overall level of Dongtang chips is almost at the stage of zero breakthrough. Although the market share is very small, every field is following up.

A long way to go!

(End of this chapter)