PS: The first half of the book focuses on the chip manufacturing process. So those who mind can coo...... After all, it's not easy for everyone to make money! Most of the latter is the sense of accomplishment of Yu Zixian and others after the chip was manufactured. Originally, chip manufacturing didn't want to write in such detail, but as a book about the chip industry, I always feel that something is missing if I don't write it, so it shouldn't be considered water (the dog's head saves my life~.~ I won't write it like this in the future).

***

From the moment the production line is energized and put into operation, silicon begins its second journey from silicon ingots to chips.

In fact, the entire chip manufacturing process can be seen as a wonderful journey of silicon from sand to chips.

Because the raw material of the chip is silicon, which is similar to sand.

The purity of crude silicon, the raw material of semiconductors, is 98%, but the purity of silicon wafers required by chips is as high as 99.99999999%! That is, the proportion of impurities cannot exceed 1 part per billion.

For comparison, the purity of the so-called pure gold (9999 gold) is also "only" 99.99%.

Quartz sand is converted into coarse silicon, "refined" into high-purity polysilicon, and then seeded in the molten polysilicon, rotated and pulled out of a cylindrical monocrystalline silicon rod (the principle is not fundamentally different from the process of marshmallow enlargement, but the process control is very different).

The crystal embryo is then ground and polished, sliced and cut into thin wafers. The diameter of the cut wafer determines that these wafers need to be used in the corresponding fab. Generally speaking, when we talk about a few inches of wafer factory, we refer to the diameter of the silicon wafer.

The larger the wafer area, the more chips can be cut out when manufacturing chips of the same process standard, which means that the technology of this fab is agreed.

At present, wafer sizes have been developed from 2 inches (50 mm) to 3 inches (76 mm), 4 inches (101 mm, commonly known as 100 mm wafers), 6 inches (150 mm), and even 8 inches (200 mm) of technology have been achieved under laboratory conditions.

At present in 1991, 6-inch wafers (which began commercialization in 1989) have already been commercialized, while the more advanced 8-inch process is still in the laboratory, and according to historical development, it will not be put into commercial production until next year.

At present, the 4-inch wafer process of Xiangji is a process developed in the early 80s, and it is already a little backward at this time. However, within the scope of Yu Zixian's ability, this is the most advanced process that can be obtained.

Unlike "large wafers", the size of transistors and wires is reduced, that is, the line width is as small as possible, which is commonly known as "small threads". With the advancement of chip technology, the width of the electrical line is getting smaller and smaller, down to the nanometer range below the micrometer.

Both "large wafers" and "small threads" can produce more silicon dies on a single wafer, improving quality and reducing costs.

Doesn't that feel complicated? But at this point, it's not enough to engrave half of the circuit!

However, chip manufacturers like Xiangji only need to be responsible for "small threads". Because "large wafers" are generally directly monopolized by companies such as Shin-Etsu and Applied Materials of the United States.

You only need to place an order with these wafer material suppliers according to the wafer size corresponding to your production line, and they will deliver them to your door.

Once the wafer is obtained, a conductive film is coated on top of it and a layer of photosensitive agent is coated. Then, the circuit negative is placed on it, and the exposure is carried out, and the circuit part undergoes a development reaction. Subsequently, the non-reactive photosensitive layer is washed away with a chemical reagent to etch circuits into the conductive film of the wafer.

Of course, the chip circuit is complex and may contain tens of millions of transistors, even if the mask is the size of a table, and then miniaturized to a small chip, a single mask may not be able to project well at one time. Therefore, during the machining process, the circuit design is divided into multiple masks, and the above process is repeated until the etching is completed.

In this photolithography procedure, synthetic cresol resin, synthetic photosensitive (layer) agent, glue preparation, etc. are the three major components of photoresist.

After the lithography is completed, it is time for ion implantation. In the silicon wafer manufacturing process, different impurities are added at different positions, and different impurities form a field effect transistor - the smallest unit transistor of a chip according to the different concentrations/positions! Logic processor chips are composed of millions, tens of millions, or even hundreds of millions of such multi-layer transistors stacked in an organized manner like building a building.

Of course, there will be many shapes formed during the lithography process that are not actually what we need, but etched for ion implantation. Etching involves washing them off with plasma, or some structures that do not need to be carved out in the first step of photolithography.

But once it's done, it's time to clean multiple chips, not with water or anything, of course, but with plasma rinsing (bombarding the entire chip with a weaker plasma beam) – this is the plasma rinsing process.

After that, it's heat treatment. The solidification of the transistor is completed by rapid thermal annealing (that is, the whole film is instantly illuminated to more than 1200 degrees Celsius through a high-power lamp, and then slowly cooled down to make the injected ions better activated and thermally oxidized), thermal oxidation (to create silicon dioxide, that is, the gate gate of the MOSFET).

Then, the sensitive parts are coated by chemical vapor deposition (CVD) to further fine the surface of various substances, physical vapor deposition (PVD) and other processes.

Finally, after electroplating, chemical/mechanical surface treatment, bare die (or die) testing, wafer grinding and segmentation, it can be packaged at the factory.

In the entire chip manufacturing process, the lithography process is the most critical process that accounts for more than 30% of the time and cost.

In the production area of the huge clean plant of Xiangji, the personnel of each division perform their duties, staring nervously at the operation interface and monitoring interface of their equipment.

At this time, the production of the first chips is nearing completion...... Wafer testing (bare die or die) has already begun.

The semiconductor testing process of wafers belongs to the key areas of the semiconductor industry, and semiconductor testing includes CP (Circuit Probe) test, CP test, also known as wafer test, which is the first step in the post-packaging and testing of semiconductor devices, with the purpose of picking out defective chips in the wafer.

Usually, in the wafer testing step, it is necessary to electrically test the chip to ensure that the chip on the wafer is a qualified product before packaging, so the wafer test is one of the key data of chip production yield statistics.

In the wafer test area, Yu Zixian and Cao Fei, Tomoki Kashima, including Liszt and others are anxiously waiting for the data statistics of qualified chips on the first wafer.

On the testing bench, Cai Junjie personally tested the characteristics of each bare die on the wafer one by one......

Yu Zixian is a little impatient! It's been more than an hour, why hasn't a wafer been tested yet!?

The diameter of a 4-inch wafer is calculated as 100 millimeters, and the area of a chip is calculated as 100 square millimeters, so no more than 80 chips can be manufactured from a wafer. Of course, if this is calculated according to the 6502XJ91 with a 3-micron process, and if it is calculated according to the 1.2-micron process chip with a more advanced process that will be trial-produced in the next step, there are more transistors on each chip, but the area of the chip is smaller!

So, it stands to reason that this less than 80 grains won't take so long? Don't have any surprises? It can't be that there is no one that is qualified, right?

The longer Yu Zixian waits, the more bottomless his heart becomes. But at this critical moment, Yu Zixian didn't care about interrupting Cai Junjie's work rashly!

Cao Fei, who was waiting on the side, looked at Yu Zixian dangling anxiously, and also looked at Cai Junjie in the glass window!

At this time, Cao Fei has been promoted by Yu Zixian to the position of general manager of Xiangjidian. The two old-timers before are already retired.

"Boss, why don't I go in and have a look......" Cao Fei couldn't hold back his anxious waiting.

"......Huh" Yu Zixian took a deep breath, looked at the laboratory personnel who were busy, and finally sighed: "Wait, wait a quarter of an hour, if you still can't ...... I'll go in and have a look!"

Cao Fei looked at Cai Junjie inside the glass curtain wall and nodded.

In anxious waiting, fifteen minutes passed as quickly as it passed. But Cai Junjie didn't come out!

Cao Fei glanced at Yu Zixian, Yu Zixian nodded lightly, and then Cao Fei checked and sorted out the clean clothes on his body again......

Just as Cao Fei passed through the air circulation clean area and was about to step into the testing area, Yu Zixian saw Cai Junjie stop what he was doing...... With his head facing his assistant, he seemed to be saying something. Immediately after that, the ten unpackaged chips were packed in a special box and walked towards the exit.

Yu Zixian hurriedly walked to the exit.

"Director Cai, how is it?" Yu Zixian said nervously, afraid that Cai Junjie would say a number that would make him miserable - zero or 1!

"Mr. Yu, it's just amazing...... The test yield of the first wafer reached 49%!"

"50%? is so high, but how is it possible? Not the general ones are below 30%, 10% or single-digit yields are also common, and even zero is not impossible!"

"However, this 50% is too cool!" Yu Zixian raised his hand and said excitedly!

"Even so, I was very surprised and excited, and I began to wonder if I had made a mistake in the operation and made a statistical error...... So I inspected another wafer!"

"How?"

"It's just that this time it's higher, and the yield rate has reached 56%!"

This time, not only Yu Zixian was surprised, but Tomoki Kashima next to him also had a different look!

"Higher?" asked Liszt, who was present, suspiciously.

Cai Junjie saw that everyone still looked incredulous, and then said: "Later, in order to convince myself for the third time, I tested another ......"

Good guy, Cai Junjie, this guy is fighting with himself in the inside! He also said that he couldn't come out of the testing area for more than an hour, so everyone outside waited for a while!

Even if you deliver a message outside, it makes everyone nervous, and there is no movement for so long, and I think there is something wrong!

However, at this time, everyone still ignores complaining about Cai Junjie for the time being, and wants to know the grain detection yield of this third wafer.

"How much?" Yu Zixian looked at Cai Junjie and asked angrily!

"You can guess......" Cai Junjie actually became a cheap swish.

"Okay, you Cai Junjie, you actually started selling so many of us, be careful that we can't help but beat you!"

"Guess what, don't worry......"

As soon as Yu Zixian heard that Cai Junjie began to sell Guanzi, he was relieved, it seems that the yield rate is stable, at least above 49%, so he pouted and said to Cai Junjie: "Do you like to say it or not, anyway, I know it's above 49%......"

"How much, Dr. Cai, don't blame me for deducting your bonus this month, and it is estimated that the boss will not say anything disagreeable at that time!" Cao Fei on the side scolded angrily.

"The third piece...... I look at the yield, it seems to be 19%!"

"What...... Cai Junjie, is it really fake? Why is the gap so big all of a sudden?" Yu Zixian couldn't stand it at first, and asked at the head of the meeting.

"Cai Junjie, you're not mistaken, right?" Cao Fei said solemnly.

"Oh, wait, I'll take a look! I'll check again, as if I also remember that the third data is not from this......" Cai Junjie took a few pieces of paper in his hand and checked it again......

After waiting for a few seconds, he patted his head and said, "Ah! I just took the handwriting paper upside down and read it wrong, and I saw 61% as 19%!"

", Cai Junjie, your coins are gone......" Yu Zixian was furious!

"Okay, you Cai Junjie, you're really playing with us!"

……

After waiting nervously before, after Cai Junjie's trouble, everyone relaxed all of a sudden.

In addition, the yield of the data obtained greatly exceeded everyone's expectations, and the yield of the chip was about 55% for the first trial production of the three wafer boards - this is a rare victory for the people and technicians involved in the construction!

Cai Junjie took out the chip and handed it over to the laboratory personnel to be assembled, and the chip after the successful packaging will be installed on several learning machines and game consoles taken by Fu Zan, and the specific performance of the chip will be tested on the computer!

At this time, Yu Zixian was anxious, and everyone began to analyze the reason for the surprisingly high yield this time.

Just to talk about the past, in addition to the cooperation of the team and technical personnel involved in the installation and commissioning of the equipment, as well as the objective factors such as the participation of a large number of manufacturers and technical personnel, the rest is from the analysis of process technology.

Tomoki Kashima makes a compelling reason: excess process performance.

The current production line is built for a 1.2-micron process, while the 6502-XJ91 chip is a 3-micron chip, which is equivalent to a large horse-drawn trolley...... So the yield will be surprisingly high!

In the future, if the yield rate exceeds expectations for the first time in the manufacture of 1.2 micron chips, it will be a real surprise......

It's all about figuring it out, but before the results are out, who can say 1, 2, 3?

Just when they were summarizing and analyzing the yield, Fu Zan walked in with two Little Doctor learning machines and two Dabawang game consoles equipped with 6502-XJ91 chips, and the staff behind him also helped bring TVs, game controllers, keyboard lights and other auxiliary equipment.

It's time to verify the actual application of the chip!

Plug in, turn on, the game "Look at the picture and guess the idiom" in the TV screen is displayed very clearly, and there is no jamming...... Changed to a big overlord game console, and it is still smooth and comfortable.

"Oh also!"

……

"Okay!" Yu Zixian patted his thigh and shouted loudly!

The 6502-XJ91 chip that was lit up not only lit up the two self-made electronic products, but also completely ignited Yu Zixian's dream!

PS: It's 2020 soon, here, I wish all book friends in the new year: big money, small wealth, windfall, financial resources, family, love, friends, all true feelings, lovers, relatives, family, everyone is safe!

Happy New Year 2020 to everyone!

In addition, I would like to thank all the book friends who have been accompanying me and "The War of Chips" to grow up together, and QQ Reading votes and subscribes every day, thank you!

"War on Chips" and everyone, Happy New Year!

()

Set a small goal first, such as remembering in 1 second: book guest residence