In Dallas, Texas, USA, Hu Weiwu's team is preparing to make a mask, or they have learned another profound lesson.

More complex circuit graphics can be summarized by far more than one mask, which mainly depends on the crisscrossing metal circuits. Only one mask will cause the circuit short circuit.

To be honest, before this, Hu Weiwu's team did not realize this problem, otherwise they would consciously separate drawings when designing.

Now they have an extra task, which is to re-draw the complex circuit diagrams that have been made.

They used to divide them into circuit diagrams based on functional circuits, but now they must be divided into metal circuit layers and changed to two to three circuit diagrams.

Current technology only supports three-layer circuit etching. With an integration of 12 million transistors, each circuit diagram is about four million transistors and the corresponding metal circuit.

The workload is very large, and Hu Weiwu's team worked day and night to modify the circuit design drawings.

Fortunately, the people brought by Hu Weiwu are all frame designers of various functional circuit diagrams, and are very familiar with their respective contents, which can be said to be clear about them.

It seems like a very complicated job, but it can actually be copied and pasted. The workload is not large. With their efforts, they re-draw the circuit diagram in just seven days.

Next is to enter the mask making process.

The mask is made of a glass substrate as the carrier and a layer of opaque ferrochromium oxide is coated with a layer of opaque chromium oxide, referred to as a chromium layer.

Then apply a layer of developing photoresist, use direct photolithography to portray the circuit pattern onto the mask like a slide.

In this way, the photoresist area that has not been developed can be etched away with a special etching solution, then rinsed with high-purity water and dried, and then re-etched with another etching solution that can etch the chromium layer, thus obtaining the desired circuit pattern.

The circuit pattern is opaque, and under the etched chrome layer is a transparent glass substrate.

The whole process can be said to be very simple, but it requires a precise lithography machine, etching machine, and professional chemical etching liquid, which has strict requirements on basic materials, optics, physics, and machinery.

In China, even a seemingly simple glass substrate cannot be made. It is not ordinary glass. It has strict requirements on light transmittance and must have nearly 100% light transmittance. It is a piece of quartz glass with high purity.

In fact, this kind of glass is also very expensive, less than a square foot, and costs tens of thousands of yuan.

Hu Weiwu swallowed his saliva when he saw those precision equipment. How much he thought that the company would have a set of such equipment one day!

But unfortunately, China was maliciously blocked by Europe and the United States, and even if you have money, it would be up to you. It would be entirely up to you to develop it yourself. I don’t know when it will be.

With an optical mask, you can directly make the chip.

The process of making a chip is similar to the process of making an optical mask, but it is more complicated.

It takes about 300 repetitive steps such as optical development, chemical etching, ion doping, cleaning and drying.

During this process, Deyi Company's 0.35 micron process is not yet mature, causing many problems.

The most embarrassing thing is the application of copper interconnection technology. Dolphin Technology does not have chip manufacturing equipment. When Hu Weiwu was doing the experiment, he only used a silicon wafer and a single crystal copper wire to simply weld it. When he found that it was sticking, it means that the experiment was successful.

But the actual application is not that simple. The transistors on the silicon wafer are very subtle and are far smaller than single-crystal copper wires, so the single-crystal copper wire can only be melted into a target material, and then sputtered to the two ends of the transistors that are not covered by photoresist using ion sputtering technology.

In this process, it is actually not single crystal copper but high-purity copper. It is necessary to find a way to make the high-purity copper sputtered to both ends of the transistor crystallize into single crystal copper.

Single crystal copper can be obtained by instant cooling, but this degree is not easy to grasp. Professional equipment must be used, otherwise the failure rate will be too high.

This requires the transformation of the ion sputtering machine to have the function of instant cooling and must be controllable.

In order to solve this problem, Deyi Company formed a strong scientific research force to carry out technical research.

The senior management of Deyi Company attached great importance to it and gave the technical research team a deadly order. They worked overtime to experiment and finally made a combined equipment after a month.

The standard for use is to experiment with the sputtered single crystal copper target, which is completely sputtered onto the transistor, and prove that the bonding is stable.

Deyi Company is already very technical, such as lithography and etching technology, which are all world-leading. The only thing that cannot be solved is the chip heating problem, so it is an immature 0.35 micron process.

After solving the key technologies of the copper interconnection process, its technical level has been greatly improved, reaching a process process level of 0.18 micron in one fell swoop.

Using a 0.18 micron process process to make a chip with a 0.35 micron process process level is Zhang Fei's bean sprouts, which is a piece of cake.

In the later packaging process, Hu Weiwu asked to use single crystal copper wire instead of gold wire to tie the wire outside, which made the technicians of Deyi Company lit up. This is a very simple question, if he didn't say it, no one would have thought of it.

But as soon as he said it, the technicians of Deyi Company knew it was feasible.

In particular, there are many memory chips such as streamlined instruction set chips, external registers, caches, etc., and there are also many wires to be built, which can save a lot of costs.

This method of connecting external wires cannot apply for a patent, so Hu Weiwu's proposal actually saved Deyi Company a lot of money.

The leadership of Deyi Company was very happy and invited Hu Weiwu's team to have a good meal.

The entire flaking process took only two months, and the next step was the mass production process.

Since Dolphin Technology has already reduced 200,000 pieces and has successfully trial-produced it, the cost of a single chip can be calculated.

The unit price of chips is affected by two main factors: one is the number of chips and the other is the chip yield rate.

Deyi Company is not a chip foundry company, so the number of chips is not an influencing factor. For them, producing ten pieces is the same as producing one hundred thousand pieces, and they do not intend to make this money.

Of course, for Dolphin Technology, there are still some impacts, because the cost of masking is very expensive, each piece costs several million, and the cost is fixed. The more chips are produced, the cheaper the unit chip will definitely be.

The second influencing factor is chip yield, which has the greatest impact on chip cost.

At the same time and cost, will each silicon wafer be qualified or only one or two chips be qualified? The unit price difference is several hundred times.

This is a question of technological maturity. When Liu Meijuan was designing the outline, it was estimated that the 0.35 micron process process was mature.

In fact, she was taking risks. Two professional OEM companies, such as TSMC and Lianhua Electronics, only have 0.5 micron process technology.

A chip with a 0.5-micron process process will produce a chip with a 12 million transistor, and the yield rate will be so low that it will be outrageously low. A whole silicon wafer may not have a qualified chip.

However, if the same chip is produced using a 0.18 micron process, the yield rate will reach an astonishing 90% of the time, which is the so-called technological maturity.

This is also why the latest chip phones are very expensive and the old phones are very cheap.

After calculation, the unit cost of the Huaxin 401 chip is about 80 US dollars, which is very cheap for a chip with the highest integration in the world.

After Hu Weiwu colluded with the country, Wang Yong directly ordered 200,000 pieces, with a total construction cost of 16 million US dollars, and then ordered a portion of the dsp chips and analog chips for the remaining money.

In short, spend 20 million US dollars to calculate the ball.
Chapter completed!