My Super Black Technology Empire

And after the chip gets bigger, the cost of a single chip will increase a lot, because in addition to the calculation and control circuit, there are communication circuits and storage circuits inside the chip. The more physical cores, the more complex these are.

After all, the wire inside the chip will become longer and the resistance will become larger when the chip is enlarged. When the voltage remains the same, the charging speed of the capacitor will slow down. If the charging is to be fast, the voltage must be increased, and the voltage increases. If the current is too large, the current will increase accordingly. If the heat generation is too large, it will be burned directly.

Today's realistic example is in front of us, AMD's thread tearer chip, which is larger than the ordinary Ryzen chip, but its price is also very high.

So this path is completely unworkable.

And Ye Fan's idea is to use carbon materials to make transistors. This is a novel genre that has been proposed by many scientists.

Why do these scientists think carbon can be used? In fact, this has a lot to do with the high-quality properties of carbon itself.

For example, the electron mobility of a transistor made of carbon nanotubes can be a thousand times that of silicon. Generally speaking, the mass base of electrons in carbon materials is better.

Another example is that the free path of electrons in carbon nanotubes is very long, that is, the activities of electrons should be more free, and it is not easy to generate heat due to friction.

Because of the advantages of these bottom layers, using carbon to make transistors, and replacing the silicon substrate layer, even without being as small as silicon transistors, can achieve the same level of performance.

For example, a study supported by the Ministry of National Defense of the United States in 2018 hopes to use 90nm carbon chips to achieve the same performance as 7nm silicon chips.

Today's quantum transistors are essentially alternative silicon transistors, except that what migrates inside them is not electrons, but quantum, but the nature of silicon itself cannot be changed.

Even if carbon is used to make chips, there are many ideas, but these ideas are still in the exploratory stage, and the closest to practicality is the carbon nanotube chip involved in this research project of Peking University. this field.

As early as 2013, Stanford University produced the world's first carbon nanotube computer, and in August 2019, the Massachusetts Institute of Technology released the world's first carbon nanotube general-purpose chip, which contains 14,000 transistors.

In the "Nature" magazine at that time, three articles were published in a row to recommend this achievement, which shows how much sensation it caused that year.

However, even this sensational research published by the Massachusetts Institute of Technology only contains 14,000 transistors, which is far from the scale of tens of billions of transistors in current mobile phone chips.

The crux of the matter lies in the four words of manufacturing process. To manufacture carbon nanotube chips with performance comparable to commercial components, an important prerequisite is to be able to manufacture high-purity, high-density, neatly arranged chips. carbon nanotube arrays.

Once the purity and density of carbon nanotubes are not high enough, or the arrangement is unsatisfactory, it will be difficult to reliably manufacture a commercial chip with hundreds of millions of transistors, because there is no guarantee that the transistor will fail.

In the study released by MIT in 2019, the purity of the carbon nanotube array used was only four nines, or 99.99%.

And people speculate that this purity is at least six nines or eight nines, so that the performance of carbon nanotube chips can match that of traditional chips.

In July, the research team led by Professor Zhang Zhiyong and Peng Lianmao from Peking University prepared a carbon nanotube array with a purity of six nines, or 99.9999%, on a 4-inch substrate through an original preparation process.

In the two important indicators of density and purity, it is 1-2 orders of magnitude higher than similar research in the past.

And based on this high-quality carbon tube array,

The researchers also mass-produced the corresponding transistors and ring oscillators to verify the mass production potential of this new process.

Through experiments, it was found that the performance of these transistors and ring oscillators surpassed the components in traditional silicon chips of the same size for the first time, proving that carbon chips may indeed be more powerful than silicon chips.

Once carbon nanotubes are applied in the industry in the future, due to their advantages in power consumption and performance, they are likely to be used in scenarios with strict energy consumption ratios such as mobile phones and 5G base stations.

If the energy consumption of the chip can continue to drop by two or two levels, it will be possible to use such very small energy sources as human body fluids for power supply, and the usage scenarios will be broader than today's consumer electronics products.

Although carbon does have many excellent properties, and some electrical properties are even better than silicon, the biggest limitation of carbon chips is actually the insulating layer in the process.

The silicon substrate as an insulating layer only needs to be oxidized to obtain silicon dioxide, but carbon materials cannot be oxidized as an insulating layer. This process is an important factor that prevents carbon from replacing silicon.

If these problems can be solved smoothly, with the strength of Datang Technology, the finished product can be mass-produced within three years, and the carbon transistors in it will be injected into the quantum transistors of today's quantum computers to realize the miniaturization of quantum computers.

After all, each of the thousand quantum computers in the basement of the Datang Science and Technology headquarters is as big as a large refrigerator, and the quantum transistors, quantum memories and other things in it are too large.

The size of a quantum transistor has reached the size of a palm. Due to the nature of its silicon, it cannot be made smaller. Therefore, there are hundreds of quantum transistors in a quantum computer, not including other parts.

This is also the main reason why quantum computers are so big. If the silicon substrate continues to be used to make chips or even transistors, then the smallest quantum transistor can only be as thin as a finger.

If the plan of carbon chips and carbon transistors can be successful, then the volume of quantum transistors can be reduced to the same size as today's mainstream electronic transistors, and hundreds of millions of small quantum transistors can even be integrated in a palm-sized chip.

In this way, the preliminary quantum chip can be completely produced, and it is possible that a quantum computer can be the size of a notebook, but its computing power is higher than that of all the computers in the world combined.

After all, today's quantum computers do not have quantum chips, but like the first computer, a large number of transistors are used to undertake data calculations.