Xiongxin No. 05, strictly speaking, this is not actually the first generation product of carbon-based chips.
Although it can now be produced in small batches and has passed a series of test experiments, this still does not change that it is only an 'experimental' product.
Yes, Xiongxin No. 05 is only one of the first batch of experimental products with relatively mature computing functions in the research and development of carbon-based chips. There is still a long way to go before commercialization.
A chip integrating 10 million carbon-based transistors per square millimeter, let alone 2025 now, even ten years ago in 2015, the number of integrated transistors could not be compared with that of silicon-based chips.
In 2015, Intel officially released its fifth-generation processor, using a 14nm process with a transistor count of up to 1.9 billion.
As for the carbon-based chip in Xu Chuan's hands, the total number of transistors has barely reached 1 billion, which is still nearly double the distance from Intel's fifth-generation products.
However, carbon-based chips and silicon-based chips are two different products in themselves and cannot be completely generalized.
A carbon-based chip constructed with one billion carbon-based transistors can match or even exceed Intel's fifth-generation processors in performance.
1 billion compared to 1.9 billion, half the number of transistors, the performance is not inferior at all, and even exceeds it in some aspects.
This is enough to demonstrate the superiority and development potential of carbon-based chips.
After all, at present, carbon-based chips are still in a state of research and development. Whether it is design software, instruction set system, chip design, etc., they are almost all derived from imitating silicon-based chips.
Although theoretically speaking, the difference between carbon-based chips and silicon-based chips is not big, and almost everything that can be used for silicon-based semiconductors can be directly applied to carbon-based semiconductors.
But after all, these two are two different element materials, how can they be 100% compatible?
Applying the design of silicon-based chips to carbon-based chips is currently just a helpless move. After all, the development of carbon-based chips is the first time in history.
In this field, they have reached the forefront of the world.
To put it simply, there are no stones in front for them to cross the river by feeling, and they need to feel their way forward bit by bit on the road behind.
If it is a real carbon-based chip, then the related circuit design, architecture, etc. need to be redesigned based on carbon-based semiconductor materials.
Now, to put it bluntly, even finished chips that can be used are only semi-finished products.
If in the future they truly rely on carbon-based semiconductor materials to develop and improve a supporting system, and achieve a true carbon-based chip...
At that time, they will have the final say in the entire industry. Silicon-based semiconductor companies such as NVIDIA, TSMC, ASML, Intel, and Qualcomm will all have to kneel down and call them daddy.
All in all, although the technology has just started, the performance of Xiongxin 05 has initially demonstrated the superiority of carbon-based chips.
Whether the thermal design power consumption is 30WTDP or the main frequency reaches 5.8Hz, it greatly demonstrates the potential of carbon-based chips, which are far beyond the reach of silicon-based chips.
.......
After briefly introducing the performance of Xiongxin No. 05, Zhao Guanggui took out another carbon-based chip from the cabinet.
Looking at the sample placed in the protective box, he said with emotion: "At present, this carbon-based chip has a certain commercial value."
"It's just that there are still many shortcomings in terms of design and preparation."
"For example, although the engraving process of carbon-based transistors is said to be a 28-nanometer process, SMIC actually uses a 65-nanometer argon fluoride lithography machine lithography machine engraving technology, which is superimposed with multiple exposure technologies.
to 28 nanometers.”
"In other words, the current upper limit of the nanometer process of our carbon-based chips is still restricted and affected by the photolithography machine to a certain extent."
Hearing this, Xu Chuan asked curiously: "But I remember that carbon-based chips seem to be able to bypass the photolithography machine? Use other engraving methods? I have read similar papers before."
After a slight pause, he continued: "And if I remember correctly, it seems that you did not use photolithography to prepare 'MOSFET metal-oxide semiconductor field effect transistor' and 'JFET junction field effect transistor' through the laboratory before.
Machine right?"
He really didn't know much about the technology related to chip preparation. After all, he was not a researcher in this field.
However, Xinghai Research Institute is studying carbon-based chips, and he has read some papers in the chip field.
For example, carbon-based chip engraving technology, circuit diagram design, etc.
Chips are known as the jewels of modern industry, and their manufacturing process involves multiple process steps.
Including oxidation lithography, ion implantation, chemical mechanical polishing, etching, deposition, metallization, cleaning, etc.
Among these process steps, photolithography technology is particularly important. It is one of the core processes of chip preparation, accounting for more than 35% of chip manufacturing costs.
Generally speaking, the main reason why chips can only be produced by photolithography machines is that photolithography technology has high resolution, high efficiency and multi-level manufacturing capabilities.
The more high-end the chip, the higher the requirements for the photolithography machine.
Currently, the only manufacturer in the world that can produce low-nano-level lithography machines is ASML in Windmill Country.
This is the overlord in the field of lithography machines. It has extreme ultraviolet (EUV) lithography machine technology and is a key equipment capable of producing the most advanced process chips. It is relied on by chip manufacturing giants such as TSMC and Samsung for the production of chips with processes of 5 nanometers and below.
.
Of course, ASML is not from a windmill country. Its lithography machine technology can be said to come from interest groups in all Western countries.
For example, the German Zeiss company provides top optical components, the Sakura country provides high-quality photoresist and single-crystal silicon wafers, the light source comes from the American Cymer company, etc.
This chapter is not finished yet, please click on the next page to continue reading the exciting content! In other words, ASML has learned the martial arts secrets from Jin Yong's novels, gathered the strengths of hundreds of schools, used them for its own use, and then used its own software to perform advanced
Integration truly achieves "one hero and three gangs". This is the cleverness of ASML.
In contrast, Nikon and Canon in Sakura Country are much more conservative, pay more attention to the Bushido spirit, and like to fight alone. This is also an important reason why Sakura Country cannot surpass ASML.
In the area of lithography machines or semiconductors, it has always been an important means for Western interest groups to suppress them.
For example, cutting-edge chips, low-nanometer lithography machines, etc. They have suffered many losses in this field.
Whether it is Huawei, Xiaomi, SMIC, BOE and other Internet/communication companies or semiconductor equipment manufacturers, they have all encountered many setbacks due to unfair treatment and malicious suppression.
If the development of carbon-based chips still cannot bypass photolithography machines, the influence, value, etc. of carbon-based chips will be limited to a certain extent.
On the opposite side, Zhao Guanggui shook his head and said: "Experimental preparation and industrial production are two completely different concepts."
"If the laboratory does not use a photolithography machine, it does not mean that the production of carbon-based chips has bypassed the photolithography machine. Chips made in a laboratory environment can be etched with circuits using instruments, and do not need to use a photolithography machine."
"But this is only a theoretical research method and cannot be mass produced."
"According to the current chip manufacturing model, all large-scale mass-produced chips are engraved with circuit patterns on silicon wafers through photolithography. The only difference is the material itself."
"The model and process are actually the same. They all require circuit etching. However, large-scale mass production of circuit etching cannot bypass the photolithography machine."
"So currently, SMIC still uses a method similar to silicon-based chips to process carbon-based chips."
After a pause, his eyes fell on the carbon-based chip held in his hand, and he continued.
"Of course, we are also organizing manpower and material resources to develop the technology that you are looking forward to, bypassing the photolithography machine and engraving carbon-based chip transistors through other methods."
"For example, arc discharge method, laser ablation method, chemical vapor deposition method and other methods are used to prepare carbon-based chips."
"But at present, these technologies are far less mature than traditional photolithography technology, and the process of preparing the chip is much larger."
"For example, we have tried to use arc discharge method and laser ablation method to prepare carbon-based chips. One of the chip processes they can achieve is at the micron level, and the other, although it has reached the nanoscale, it also exceeds 500 nanometers.
"
"It is currently almost impossible and very difficult to process and engrave carbon-based chips by bypassing the key technology of photolithography machine."
After a brief explanation, Zhao Guanggui's eyes fell on the chip in his hand.
In fact, he is not the only one who wants to bypass the photolithography machine to prepare carbon-based chips.
Not to mention others, semiconductor wafer foundries such as Warwick HiSilicon, SMIC, and even MediaTek, TSMC, Intel, etc. all want to find a way to bypass photolithography machine processing of chips.
During this time, when he was responsible for cooperating with people from Warwick HiSilicon, SMIC and other teams to produce and research carbon-based chips, he also consulted professional chip developers on this issue.
This road is not that easy to walk.
Humanity has been developing semiconductors for decades before finally settling on the path of silicon-based chips.
The reason is silicon's high cost-effectiveness, good chemical stability, excellent semiconductor properties, mature processing technology, etc.
In particular, the excellent semiconductor properties are a very critical point.
Silicon is a natural semiconductor material. It has high resistance in its pure form. After adding a small amount of impurities (doping), its conductivity can be controlled, thereby effectively switching between conductivity and insulation. This is impossible when manufacturing chips.
Required attributes.
Compared with silicon, other materials have their own shortcomings in this regard.
For example, the earliest germanium-based chips used by humans.
Germanium was the earliest material used in transistors, but due to its low content in the earth's crust, its cost was high and its stability was not as good as silicon, so it was gradually replaced by silicon.
And now they are developing carbon-based chips.
Although carbon-based materials have some advantages, such as higher operating speed and lower power consumption, their thermal conductivity is low, processing difficulty and cost are high. These problems have greatly limited the widespread use of carbon-based chips.
application.
In particular, doping circuit control and large-scale arrangement of carbon nanotubes or graphene sheets are huge problems in the production process of carbon-based chips.
In comparison, the advantages of silicon materials are much greater.
Although high-purity monocrystalline silicon, photoresist, etc. are all difficult problems, the biggest problem is the photolithography machine.
Only top-notch photolithography machines can produce lower nanometer silicon-based chips.
It is no exaggeration to say that among the chip preparation technologies currently studied by humans, silicon-based chips are the simplest.
Even the simplest one took almost the entire Western interest group decades to gradually perfect.
If you want to overtake in a curve, it is not easy to bypass the photolithography machine and directly engrave and process the chip.
It can be said that almost all the roads you can think of have been thought of and tried by researchers in the chip field.
...
Although photolithography machines are still a problem that hinders carbon-based chips, Zhao Guanggui is not too worried.
He smiled and continued: "Although the photolithography machine has a big problem, for now we don't need to worry too much."
"There are still domestic manufacturers that research lithography machines, such as Modu Microelectronics Company in Modu. They already have a mature lithography machine preparation system. They have also developed 90 nanometers, 110 nanometers, 280 nanometers and 55 nanometers.
Four major series of domestic photolithography machines.”
"And the 28-nanometer lithography machine currently being promoted is about to mature."
"For carbon-based chips, the superior physical properties of carbon-based transistors are enough to make up for our shortcomings in process technology to a certain extent."
"Theoretically, as long as we can increase the number of carbon-based transistors per square millimeter to 30 million, the performance it will explode will be comparable to the mid-to-high-end chips currently available in the world."
"This is also our next key research direction."
Xu Chuan nodded and asked, "How long will it take us to achieve this goal?"
Hearing this question, Zhao Guanggui thought carefully for a while and answered with a relatively conservative attitude: "According to the current situation, within a year at most, we should be able to produce a carbon-based chip with commercial value.
"
"Of course, the commercial value I am talking about refers to a level that is at least comparable to the mid-to-high-end silicon-based chips currently on the market."
"Um......"
After thinking for a while, Zhao Guanggui added: "If we compare it with Intel's Core series, in one year we will be able to produce at least a carbon-based chip with performance reaching the Core 10th generation level."
"Although there is still some gap between this and Core's latest 14th generation chips, with the development speed of carbon-based chips, I believe it will only take two or three years at most before we can equal or even surpass each other!"