Chapter 185: The Game of Local Tyrants
When Texas Instruments was tossing and digesting the technological achievements of Huaxin Technology, a work of Huaxin Technology made Wang Anran deeply worried.
That is the disassembly of IBM's No. 1 chip production line, which lasted nearly four months, and the completion progress of the entire disassembly work is less than 50%.
The progress of a disassembly and drawing work is so difficult, and the dismantling team is currently only disassembling the easy ones first, numbering the parts, and drawing drawings and material descriptions.
And some difficult places, I didn't even dare to move, for example, lens sets, pneumatic bearings. servo compensation system, etc.
Wang Anran has repeatedly assured that you don't need to pay for the demolition, and the progress is still slow.
The crux of the problem is that most of the functional areas are closed-loop seals for the construction of these cores, and you can't find the screw buckles even if you want to disassemble them, and the only way is to disassemble them violently.
But after violent disassembly, it is basically impossible to restore it, and no one dares to open the first knife for tens of millions of devices.
For this reason, Wang Anran had no choice but to put down what he was doing and come to the town in person.
This time it was the removal of the servo compensator, which plays a vital role in the lithography process.
The reason is obvious, the wafer needs to be processed into a chip, it needs to go through multiple photolithography, physical material precipitation, etching, cleaning and other processes, precipitation, etching, cleaning and other processes are not sensitive to alignment accuracy, but multi-pass lithography involves a very difficult technical problem.
That's alignment.
The spatial position between the mask and the wafer is within 1 micron during multiple lithography processes, and the lens of the lithography machine will not be able to complete the alignment correction if this deviation is too high.
And how long is 1 micron? If you have a quick knife, you can split a hair into 50 cents in diameter, which is about 1 in 50.
To detect this length is very simple, involving the principle of a light lever, but if you want to adjust the position, it is difficult, and the compensation system is to use a series of algorithms, experimental data, etc., to apply the reverse force to the platform once the displacement is not in place, and achieve accurate positioning.
There is no doubt that this is a technology that needs to be tackled, and the duplex servo compensation system planned by Wang Anran is a hundred times more difficult than this.
Before he learned to walk, Wang Anran naturally didn't expect to be able to run, so this first step was crucial.
Zuo Taixing, the technical director, designed a set of demolition schemes, the core of which is to ask an eighth-level file worker to file off the surface metal layer by layer with a manual file, and then open the internal results.
Wang Anran agreed to this plan.
"Mr. Wang, I really arranged for Master Qin to start!"
Wang Anran waved his hand and said, "Don't worry, I will definitely repeat it again, you don't have to pay for the file if it is broken." ”
Zuo Taixing shook his head and said with a wry smile: "Mr. Wang, even if you ask me to pay, I can't afford to pay it, this system is millions of dollars, the key is that if you want to buy it, you may not be able to buy it." ”
Wang Anran came over to cheer them up, and when he came back from this trip, Wang Anran was in a very bad mood.
In the office of the chip processing factory, Wang Anran irritably lit a cigarette, smoked half of it, and snuffed out the cigarette butt in the ashtray.
The chip production line of the 1.5 micron production process is so difficult to draw now, so the success of the imitation will have to wait until what year and month, the most important thing is that Huaxin Technology submitted to the official project proposal to lead the research and development of the lithography machine, it has been several months, and there has been no reply, which makes Wang Anran complain a little in his heart.
Without official support, Wang Anran knows that in China now, those companies are stronger in lenses, and which companies have enough technology accumulation in precision machining.
Huaxin Technology alone can fully understand the technology of the entire chip production line. Even the production line, which is now somewhat outdated, is very difficult.
In Wang Anran's view, even the most optimistic estimate can be eaten within ten years, which is already burning high incense.
Ten years later, the chip processing technology of the entire industry has developed 90 nanometers, and the lithography machine has gained a firm foothold in the deep ultraviolet region.
Wang Anran has no idea, but he is still very confident in exploring the depth of the existing technology.
The BSIM mathematical modeling led by Ji Xiaoqing, on the basis of Wang Anran's opening all the way, is like riding a rocket, and in just two or three months, there has been a major breakthrough.
According to Wang Anran's estimate, the progress at this moment has exceeded the research and development progress of the original author, Professor Hu Zhengming.
In the final analysis, BSIM is an industry-standard mathematical simulation model that analyzes the substrate resistance network, saturation current, tunneling current, and stress of a specific transistor from the actual MOSFET transistor under the FinFET architecture.
The theoretical basis for the design of FinFET field-effect transistors in practical application is how to design and optimize the length of the fin (channel), the thickness of the gate wrap, and the 3D structure.
Huaxin Technology's chip processing plant has given the material basis for the realization of BSIM mathematical modeling.
Using the vapor phase physical deposition method of cash provided by Toshiba, Huaxin Technology successfully "grows" silicon fins on wafers and designs more than 200 sets of combination solutions in various sizes.
There are more than 30 sets of masks consumed, and according to the cost price of $500,000 a set, the expenditure on masks alone has reached $15 million.
Coupled with all the expenditures of the entire production line, materials, testing instruments, etc., the total has reached an astonishing more than 80 million US dollars, equivalent to more than 600 million yuan.
This is only two months, and more than 600 million yuan was burned in two months, not including labor wages, as well as the profits due from the operation of the production line.
According to the calculation method learned from the ancients, Huaxin Technology has spent more than one billion yuan on the BISM model.
Therefore, if it weren't for the local tyrant, the chip wouldn't be able to afford to play this game at all.
However, Wang Anran feels that it is worth it, and in his opinion, it is of great significance for Huaxin Technology to release the BISM mathematical model first.
This is not only an opportunity for Chinese companies to show their faces on the international stage, but more importantly, Huaxin Technology can preemptively register a large number of patents under the FinFET architecture.
It is necessary to know the BISM mathematical model, which has been set as an international standard by the EIA Compact Model Council (CMC) Electronics Industry Association.
As long as Huaxin Technology releases BISM, the entire semiconductor industry can't avoid it, especially after 45nm, leaving the FinFET field-effect transistor, the entire semiconductor industry has to be blinded.
Just when Wang Anran was planning to create a nuclear bomb for the world's semiconductor manufacturing industry, NVIDA Nvidia from the United States quietly released the company's first graphics card after its establishment, NVIDIA TNT.
Yes, TNT, high explosives, fiery performance.
Lao Huang's first knife brightened Wang Anran's dog eyes.
()