Can a whole wafer be used as a CPU?

No. There are at least three factors that limit the inability to produce such a large area of CPU. For the current CPU core design, it is still necessary to consider the signal transmission delay caused by the physical distance between the transistor circuits in the chip (the same is true for GPUs, memory chips, etc.), and the size is to be expanded by thousands of times, which is currently technically impossible to solve.

The accuracy of the CPU manufacturing process has reached the nanometer level, and the slight deformation of the thermal expansion and contraction of the silicon wafer material will lead to damage, if there is such a large CPU, whether it is its own thermal expansion and contraction or the deformation caused by external force distortion and vibration, it is enough to damage its internal physical structure. The basic unit of the integrated circuit is the PN section, and each PN section will generate heat when working, if a 4 square millimeter CPU chip has a full load power of 40W, then the same integration degree of 4000 square millimeters of a single chip has exceeded the heat generated by the fierce fire stove in the back kitchen of the hotel, whether it is from the power supply or heat dissipation considerations are not feasible. <>

Silicon is used for CPUs because it has so many advantages and disadvantages that can be overcome. Although germanium also has advantages (such as turn-on voltage and carrier mobility), it has several disadvantages that are difficult to overcome.

If you ask what the CPU is made of, everyone will easily give the answer - silicon. This is true, but what about silicon from? In fact, it's the most inconspicuous sand.

It's hard to imagine that the **expensive, complex, powerful, and mysterious CPU should come from that worthless sand.

Of course, there is a complex manufacturing process that must go through in between. However, it is not just a handful of sand that can be used as a raw material, it must be carefully selected to extract the purest silicon raw materials. Imagine if you made a CPU from the cheapest and most abundant raw materials, the quality of the finished product, and you could still use a high-performance processor like today.

Wafers are made from silicon ingots that are shaped into a perfect cylinder and then cut into sheets.

Wafers are not made by the CPU, but are used to make the CPU.

Modern CPUs are made of silicon, and the purity of silicon materials is extremely high in order to make CPUs

It will be chemically purified to the point that there are almost no impurities, and it will have to be converted into silicon crystals.

The raw silicon will be melted and placed into a huge quartz furnace. At this point, a seed is placed in the furnace so that the silicon crystal grows around the seed until it becomes a near-perfect monocrystalline silicon.

The silicon ingots are fabricated and shaped into a perfect cylinder, which is then cut into sheets, called wafers. Wafers are actually used for CPU manufacturing. Generally speaking, the thinner the wafer is cut, the more finished CPU products can be made with the same amount of silicon material.

Why do you say so much? To put it simply, it is made of sand (monocrystalline silicon (sand) with a high degree of purity).

A wafer is a whole piece cut horizontally from a silicon ingot, and it has to go through many complex processes in the later stage, such as photolithography, etching, coating, processing, etching. If you don't cut, you can't process, how to make a CPU?

Yes, and someone has already done that. The commercial semiconductor industry will not do this, mainly considering the problem of yield rate, because the larger the area of a chip, the greater the possibility of defects, and for the chip, a small defect may lead to the scrapping of the entire chip.

At least in the lithography link, there is a difference. However, there is not much difference (all within tolerance). If the difference is big enough to affect the CPU system, it will be stopped for inspection early, and the materials, labor, and time of the FAB will be too expensive, and this kind of thing will never be allowed.

At the same time, other links correct the deviations caused by the lithography process. The tolerance of many lithography links is related to the error correction ability of other links. So all in all, the physique has a subtle impact, but it is not something that ordinary players need to worry about.

Differences in detailed lithography: Portrait: On big data, images near the center of the circle have better focus than the edges.

The line carved out of the round Li banquet is clear. Don't think that the images are swept out line by line, and the focus is the same. In fact, there are so many factors that humanity has done to overcome those known effects.

The edge effect is everywhere, and the center of the watermelon is sweet rather than white. For example, a high-end lithography machine in a certain country.

The silicon wafer is sucked on the silicon table, and the edge force and the center are differentThe vertical focus was on the edge, and I had to cry silently.

Horizontal: The standard deviation of the alignment of images near the center of the circle.

It's a bit better than the standard deviation near the edge. Generally, a CPU now has to go through lithography at least a dozen to dozens of times. I'm not afraid that I'm biased in this layer, because the next layer is the same, and it's still easy to use.

But I'm afraid that the two layers are relatively biased. The smaller the standard deviation, the more accurate it is. Here's an example:

The lithography machine for CPU has to be high-end, and the high-end has to be immersion. Immersion water is more likely to remain at the edges, and if it remains, it will evaporate, and if it evaporates, it will cool down, and if it cools down, it will cause the silicon wafer to deform. If it is deformed, the horizontal alignment will not be accurate.

Before this question, we have to figure out the so-called"Physique"What is the meaning. The wafers used to make chips are generally monocrystalline silicon. In the manufacturing process, the wafer will inevitably have defects.

So-called"Physique"It can be understood as the density of lattice defects per unit area. At present, the main method of producing wafers is the lifting method. The lifting method is to put the raw materials that constitute the crystal in a crucible to heat and melt, and the seed crystal is grafted on the surface of the melt to pull the melt, and under controlled conditions, the seed crystal and the melt are continuously rearranged atoms or molecules at the interface, and a single crystal is gradually solidified with cooling.

During the production process, the lifting rod is constantly rotated, and the crystals grow from the seed crystal from the inside to the outside of the middle rock. The outward alignment the higher the probability of error, the greater the probability of defects. Thus, statistically speaking, the density of lattice defects increases from the inside out.

Mapping back to the concept we began to explain, that the chip near the middle has a good constitution and the outside has a poor constitution. This also depends on the electricity used in lithography factories, such as G3258, Vietnam uses Mekong hydroelectric power to generate electricity, and the U produced is a big thunder, which cannot be on 4G. Costa Rica uses South American charcoal-fired thermal power, so it is easy to produce large eagles, and it can be up to 8g.

Germany uses the snow of the Alps to generate electricity, but because the snow power cannot effectively generate electricity, the number of G3258 produced by the German lithography factory is zero Recently, India has developed a rain CPU, using the rain of Cirapunzi as the CPU for calculation, and it is estimated that its computing power is more than 100 trillion Tianhe II, and its basic principle is: rain is a conductor, and the air between the rain is a non-conductor, so the sky that rains can be regarded as a semiconductor as a whole. In fact, making chips is actually like steaming steamed buns.

In the same step, you steam out the lump of dead noodles, why, do you know that when you steam the steamed buns, the range hood is turned on, and the quality of the steamed bread has changed. Not to mention steamed chips. As for whether the middle one is of better quality, I'm not professional, but I think there must be some differences, and this has an impact on the air pressure, temperature and other environments.

It is not an exaggeration to say that he is a Feng Shui scholar. <>