Chapter 370 MOS
In the college entrance examination in 1960, there was no such thing as checking scores. Anyway, just finish the test and wait for the results. Admission notices will start to be issued in August.
And based on Lou Xiao'e's situation, the first batch this year will definitely not be admitted, so the time will have to be a little later. The characteristic of this batch is that "confidential majors can be admitted."
So Lou Xiaoe and Gao Zhendong are not in a hurry, they just do what they have to do. Gao Zhendong goes to work, and Lou Xiaoe takes care of the housework.
——
In the office, Gao Zhendong was thinking about Factory 1274.
Due to technological development reasons, MOS (metal-oxide-semiconductor) technology was not fully formed at this time.
The integrated circuit technology of Factory 1274 still follows the PN junction isolation bipolar semiconductor path proposed in 1958. This is also the limit of Factory 1274's own capabilities, and it has even exceeded the limit.
This left Gao Zhendong in a dilemma. Should he continue to let the 1274 factory use bipolar semiconductors, or should he simply go directly to MOS technology?
The relationship between the two is not a complete substitution. Bipolar semiconductors even have obvious advantages over MOS semiconductors in many places.
For example, the switching speed of the triode tube is faster, the frequency that can be achieved is higher, the cost in the design stage is lower, the accuracy of the internal components is higher, etc.
For Gao Zhendong, what he wants is semiconductor technology that can play a greater role and have greater potential in the direction of computer technology.
So even if integrated circuits have other routes such as ultra-small assembly and film integrated circuits at this stage, Gao Zhendong has not considered these routes. These routes have their own special uses and are not without future, but in the direction of computers They have no future in the foreseeable future.
The only question is at this stage, use bipolar? Or should we take a big step and consider MOS directly?
Bipolar type has its advantages at this stage in integrating logic gate circuits.
After much thought, Gao Zhendong decided to take a big step and take MOS.
The speed of the bipolar type is faster, but at this time, no matter how fast it is, it is not much faster. It is useless if other technologies cannot support it. For computers, the tens of MHz that early MOS can achieve The speed is more than enough.
The 80286 in the 1980s was only 20MHz. In the 1960s, the faster switching speed of the bipolar type was of little significance to Gao Zhendong's needs.
Although the accuracy of bipolar internal components is higher, as a digital circuit, as long as it can complete the expected state transition under specified conditions, high component accuracy is of little use. This is not analog. circuits, the requirements are quite different.
Besides, the high accuracy of bipolar components is only relative to MOS. The real difference is not even a fifty step laughs a hundred step.
To say that the cost is low in the integrated circuit design stage is a false proposition at this stage, and labor costs are not counted at this time.
Another problem with MOS is that once a chip is finalized, it is difficult to modify it and the cost of modification is very high.
But for what Gao Zhendong wants to do, this is not a problem at all.
As chips used on a large scale, whether they are logic gate integrated circuits, CPUs, or DRAMs, they are finalized and will not be modified casually, but will be mass-produced, so this is not a problem at all in this regard. question.
The most important point is that Gao Zhendong clearly knows that from the beginning of CPU and semiconductor memory, there has been no such thing as bipolar type. Bipolar type is good for integrated logic gate circuits, but it is used for CPU and semiconductor memory are not used at all.
In other words, in this regard, from a technical and economic perspective, people have never favored bipolar semiconductors.
Intel4004, 10μm PMOS.
8008, 10μm PMOS.
The first 4Kbit DRAM, 8μm NMOS.
The first 16Kbit DRAM, 5μm NMOS.
The famous 8086/8088, 3μm NMOS.
80286, 1.5μm CMOS, which has completely consolidated Intel’s decades-long foundation.
As for why everyone unanimously chose MOS technology in this application direction, we have to talk about the advantages of MOS.
The craftsmanship of this thing is simple! Much simpler than bipolar, not a bit of a star!
Leaving aside the complex technical principles and so on, let’s briefly summarize, taking PMOS and double-diffused epitaxial bipolar type as an example, to achieve almost the same effect, the process difference between the two is very huge.
The number of PMOS epitaxy times is 1, the number of process steps is up to 45, the high temperature process is 2 steps, and the photolithography is up to 5 times.
The figures for the double diffusion epitaxial bipolar type are more than 4 times, 130 steps, 10 steps, and 8 times respectively.
Fewer procedures, simpler technology, and higher yield
For mass production, these are all money!
And for Gao Zhendong now, the fewer process steps means the higher the success rate.
The basic technologies used by the two are actually similar. The biggest difference is in the working principle of the transistor. Therefore, in terms of technical difficulty at this stage, Gao Zhendong is the porter of knowledge and the later and more advanced MOS Even lower than bipolar.
MOS technology also has a very unnatural and counter-intuitive aspect. Within the same generation, changing the design of the MOS circuit has no impact on the MOS process. Changes in the performance of the MOS circuit are achieved by changing the geometric design of the MOS field effect transistor.
In this case, the bipolar circuit needs to change the circuit performance by changing process parameters such as diffusion source, diffusion time, diffusion temperature, etc., but MOS circuits do not. The craftsmanship remains unchanged.
And this characteristic that changing the geometric design can change the performance brings another benefit of MOS integrated circuits - it is easier to implement computer-aided design and achieve semi-automatic or automated design.
In addition to the above benefits, the bipolar semiconductor itself has one biggest flaw that destined it not to go far in large-scale and ultra-large-scale integrated circuits.
——It’s not small! But MOS can!
There are many reasons for this situation.
First of all, the area of MOS tubes is inherently smaller than that of bipolar tubes.
Secondly, bipolar transistors need to isolate PN junctions or isolation wells, but MOS does not.
Third, MOS inherently provides two layers of interconnection, which makes its internal wiring more convenient.
The many advantages of MOS have also brought about some system-level derived benefits, such as higher system performance, better design predictability, better reliability and maintenance, etc.
Interestingly, the concept of MOS field effect tubes was actually proposed earlier than bipolar tubes. The reason why it became popular later is due to process constraints.
This is just like cars. The concept of electric cars is earlier than steam cars and internal combustion cars. It was in the late 19th century, but it is only in the 21st century that they are truly practical on a large scale.
But the issue of craftsmanship is not a big problem for Gao Zhendong. He can copy it!
After thinking about it, Gao Zhendong made up his mind: take a big step and start MOS!
As for the bipolar type, leave it to comrades from other factories. I will take off directly with Factory 1274!
For the future development of domestic computer technology, he is willing to take on this responsibility and risk. He wears all the clothes. When it is time to take a risk and take responsibility for the huge interests of the country and society, he still It would be pointless if you are timid.
For this time period in the early 1960s, if Gao Zhendong didn’t wear it, no one would engage in MOS. If Gao Zhendong wore it, but no one would engage in MOS, then Gao Zhendong wouldn’t wear it for nothing.
Gao Zhendong made up his mind and started to implement this matter.
He called Factory 1274 and told them that he had determined the direction of integrated circuit process research and adjustment, and then began to compile a process design guidance document for MOS technology based on the content in the book "Integrated Circuit Engineering Technology."
The design guidance document guides 1274 Factory on how to design each process of MOS technology, instead of directly writing out the process, and there is no way to write it out. This is not something Gao Zhendong is responsible for.
For Factory 1274, with this process design guidance document, they can carry out targeted and specific process design according to their own conditions, which is more operable and targeted.
It is equivalent to Gao Zhendong telling Factory 1274, what should you do in order to realize the MOS process, what are the key points of these things, what should be paid attention to, and how to calculate some important parameters in the middle.
As for every step, Factory 1274 will do it according to this guidance document.
The reason why I write this first is mainly for the sake of the comrades of the 1274 factory. The switch from bipolar transistors to MOS, although in fact the technologies used are similar, from the appearance point of view, this is a complete big change.
This kind of turn can easily cause fear for the comrades of the 1274 factory who do not understand the MOS process.
So Gao Zhendong simply came up with the design guidance document for the MOS process first. Hey, take a look. The two things have different names and different principles, but when it comes to the technology of each step used, in fact, almost.
It’s very simple, just go ahead and give it a try, you’ll be fine.
This will help the comrades of 1274 get into the state faster. After all, if you suddenly come into contact with a new technology that you know nothing about, you have to start it. Instead of directly getting all the key points of this new technology and then deepening it, For them, the sensory difference is still very big.
At 1274, after receiving a call from Gao Zhendong, Factory Director Lu and Chief Engineer Lu put aside their last worries and devoted all their energy to the production of transistors.
Since Chief Engineer Gao is in charge of adjusting the integrated circuit process, there is no need to worry here and just finish your work first.
Although China has begun to switch to DJS-60D, for 1274 Factory, the production task of transistors is still heavy.
The country is switching to 60D, but exports are still using 59. This alone is destined to make the 1274 factory full, and it can produce as much as it can.
Not to mention the domestic people’s demand for electronic products.
To put it aside, a radio requires anywhere from 4 to 9 transistors. Excluding the ones required by the DJS-59, it is simply not enough.
Anyway, everything is fine.
(End of this chapter)