It seems to us today that after the invention of the transistor, the emergence of the integrated circuit up to today's supermassive integrated circuits was A thing that comes naturally.
But if you go back to the early history of the semiconductor industry scene, we will find that none of the key technology breakthroughs are "inevitable".
Lithography, which is one of the key technologies that enabled the emergence of the semiconductor chip, is still the core manufacturing process of today's chips, and the lithography machine It is moreover known as the crown jewel of the semiconductor industry.
In trying to discuss how China can achieve a breakthrough in the semiconductor industry, lithography and lithography machines are always the technology we can't avoid! The hidden pains, and the technological peaks we must cross.
However, the variety of technologies involved in high-end lithography, the technical difficulty, and the complexity of the industrial chain are far beyond the imagination of the layman.
In the course of more than seventy years of the semiconductor industry, it is the continuous improvement of lithography technology that has driven the iterative upgrade of the chip structure.
At the same time, the lithography technology and the accompanying lithography machine, light source, optical components, photoresist and other materials and equipment have also formed a very high Technological barriers and an intricate industrial landscape.
Taking pictures of silicon crystals, lithography makes an appearance
There was also a huge leap from the invention of the transistor to the advent of the integrated circuit, and that was how to take a large number of electronic devices miniaturized to fit onto a small piece of circuitry.
It took another decade for the world's brightest electronic engineers to complete this leap, and that decade has become the most important in the history of electronics technology.
In the 1950s, before the advent of the chip, electronic connections were almost entirely dependent on manual work. A U.S. Navy aircraft carrier at the time had 350,000 electronic devices and required tens of millions of solder joints.
This amount of engineering made the production of electronic equipment severely inefficient, and the yield of the circuitry was entirely dependent on the operator's proficiency and Accuracy.
The electronics industry is crying out for miniaturized integrated circuits, or chips, and the process to make them is at Bell Labs Gestation.
Beginning in 1950, several Bell Labs chemists successively completed the purification of germanium and silicon crystals.
By early 1995, Henry Solo had produced silicon crystals with an impurity concentration of less than one thousandth of a percent.
At the same time, a group led by chemist Calvin Fuller developed a process for diffusing impurities in germanium crystals at high temperatures, which could be achieved by precisely Control the depth and amount of impurities entering the germanium crystal to create a PN junction.
In 1955, Fuller's group had already applied diffusion techniques to silicon crystals by injecting two types of impurities into the wafer on, forming NPN structures. Diffusion technology remains the basis for transistor fabrication to this day.
Carl Froch and Lincoln Derrick, who were also working at Bell Labs, proposed an entirely new approach to diffusion technology, namely The generation of an oxide film on the silicon wafer, on which a window pattern is etched, so that impurities can only diffuse through the window into the silicon substrate. And the areas covered with oxide film were protected.
With these basic processes in place, lithography was on the horizon.
In 1955, Jules Andrews and Walter Bond at Bell Labs began a collaboration that would be used to create photolithography for printed circuits. The technique is used for silicon wafer processing.
This is done by uniformly coating the oxide film of silica with a layer of "photoresist" (also known as photoresist), which is then applied to the wafer by means of a photoresist. The window pattern is exposed to this layer by masking, creating a precise window area.
This "window" is then chemically etched to remove any unexposed "photoresist".
Finally, the desired contaminants are diffused through these "windows" into the silicon substrate below to form the P and N types required for the semiconductor device structure to form more precise and complex semiconductor devices.
In short, the essence of photolithography is to create the electronic circuits and functional regions needed for the chip.
Photolithography machine will light source through the mask, to coated resist silicon wafers for exposure, after exposure of the photoresist changes, also "photocopying". "The graphics on top of the mask eventually led to the creation of electronic wiring diagrams on top of the wafer as well.
Purification technology, diffusion technology, oxide mask technology, and lithography, these manufacturing process technologies fill the gap from transistor discrete devices to the great divide in integrated electronics.
It wasn't long before Texas Instruments' Kilby and Fairchild Semiconductor's Noyes took these semiconductor manufacturing processes from Bell Labs' Applied to the manufacture of integrated circuits, the semiconductor industry opened the way to take off.
Lithography "Moore's Law"
Interestingly, there is another strand to the development of lithography.
At the same time that Bell Labs was making advances in semiconductor technology, two engineers working on miniaturization of solid-state circuits for the U.S. Department of Defense at the time. Jay Lathrop and James. Lyslop and James. Nall, had already begun using photoresist to make germanium transistors in 1952.
In 1957, the two advanced the lithography technique further to Bell Labs' work and created a miniaturized transistor. and ceramic hybrid circuits, and coined the term "photolithography".
In 1958, Horney of Sendong Semiconductor invented the planar process, which solved the problem of insulating and connecting transistors, while Rust and Noyce built the world's first photolithography camera for the manufacture of silicon-based crystalline transistors.
In 1959, Fairchild Semiconductor developed the world's first monostructured silicon wafer.
In 1963, the CMOS manufacturing process was developed and became the mainstream manufacturing process in today's IC industry.
In the early 1960s, the lithography technology was still very rudimentary. At that time, the mask was attached to the wafer one to one, and the wafer was only 1 inch in size. Because the principle is not complicated, just like with photography, semiconductor companies can also design their own related lithography tools and equipment, but soon specialize in it.
Because the principle is not complicated, as with photography, semiconductor companies can also design their own lithography tools and equipment, but soon specialised lithography equipment will be available. The photolithography machine appeared and immediately became one of the key devices for chip manufacturing.
It was also in 1965 that Intel founder Gordon Moore, then director of the Fairchild Semiconductor Laboratory, observed that each generation of In the early 1980s, the number of chips was almost double the capacity of the previous generation of chips, which led to Moore's Law, which was to drive the continuous upgrading of semiconductor technology.
At that time, the number of components that could be accommodated on an IC chip doubled every year on a constant price basis. 1975 In 2009, he switched to doubling it every two years.
The key to the realization of Moore's law is the lithography technology. As the size of integrated circuit components continue to shrink, and the chip integration and computing speed continue to increase, the resolution of the lithography technology The requirements are increasing.
The final realization of Moore's law is related to this optical resolution, which is determined by a Rayleigh formula.
where CD is the critical dimension for exposure, K1 is the process constant, λ is the optical wavelength, and NA is the optical numerical aperture of the projection objective.
The lower the CD value, the higher the resolution, i.e., lithography only reduces the CD by 30% to 50% every two years. Moore's law can only be fulfilled.
Therefore, there are three ways to improve the optical resolution, lowering the K1 value, improving the numerical aperture NA, and lowering the wavelength λ.
In the real technical process, the improvement of K1 and NA values is limited, while reducing the wavelength λ of the exposed light source becomes the lithography technology continuous Driving Trends.
Starting in the 1960s, semiconductor exposure light sources went through the visible, 436 nm in the 1980s, 365 nm near-ultraviolet band to the high-pressure mercury lamp light source of the 1990s, and then to the 248nm deep-ultraviolet band of the excimer KrF laser.
All the way to the 193nm ArF excimer laser in the late 1990s, which is still used today in mainstream computer chip manufacturing. DUV laser light source.
It is the 193nm wavelength that has become the watershed that determines the landscape of today's photolithography industry.
Faced with the challenge of how to break through the 193nm wavelength, both the scientific community and the lithography industry were looking for a solution to surpass it.
At that time, SVG in the U.S. and Nikon in Japan, based on the previous generation of dry lithography, opted for the seemingly more secure 157-nm F2 Laser.
The U.S. Department of Energy and Intel initiated, in conjunction with Motorola, AMD and others, the formation of EUV LLC to focus on the too-advanced 13.5nm EUV extreme ultraviolet light source.
There were also the more niche EPL, ion lithography, etc. But these attempts failed at the time.
Interestingly, Benjen Lin, an engineer from TSMC, came up with a wavelength based on 193nm in 2002.
But change the dry lithography for the immersion lithography process, that is, a thin layer of water on top of the photoresist, to the 193 nm wavelength Refraction to 134 nm, a sudden breakthrough of 157 nm barrier.
Since then, after many process improvements, the immersion lithography technology has made it to the 22nm process.
The earliest choice of immersion lithography is the "Chosen One" ASML in general.
Eventually, under the cooperation between ASML and TSMC, the 193nm immersion lithography machine was first produced, and it was also the first immersion lithography machine in the world. Three years ahead of Nikon's new products, ASML completely won the lithography market share of the vast majority.
And the collapse of Nikon was never able to come up with a better lithography machine, but only to stay in the low-end market.
After that, only ASML and the unique EUV lithography were left on the high-end lithography track. And this paragraph needs a special topic to analyze.
In the lithography technology decades of evolution process, we can actually also faintly see a lithography industry change path.
Brutal elimination game in the photolithography industry
There is no doubt that Bell Labs is the originator of semiconductor transistors and lithography.
Then, in the U.S., where the patent system is so perfect, why didn't Bell Labs and its backer, AT&T, become a major player in the semiconductor industry? leader, but rather a multitude of U.S. semiconductor companies that have risen quickly in a short period of time?
The reason this technological revolution spread quickly throughout the industry stemmed from the fact that AT&T was under antitrust pressure at the time and had to pay a The U.S. government took a stand and made semiconductor technology public.
In 1956, Bell Labs held its third sharing of semiconductor transistor technology and formally announced lithography, diffusion and oxidation layer mask technology.
This, along with the transistor manufacturing technology sold as early as 1952, directly empowered the likes of Texas Instruments, IBM, Motorola, Sony. The semiconductor technologies of companies such as Fairchild, Intel, AMD and other subsequent semiconductor giants were also created indirectly.
The announcement and proliferation of lithography technology triggered the innovation and migration of the lithography industry, which continues to this day.
In 1961, GCA Medical Technology Inc. built the first lithography machine, and in the 1970s, Kasper Instruments Inc. built the first lithography machine. In the seventies, the American Kasper Instrument Company and Perkin Elmer Company successively introduced the alignment, projection and projection lithography technology. lithography products, taking the market lead.
In 1978, GCA also introduced the Stepper, the first truly automated stepper lithography machine, with a resolution that could It reached 1 micron and gradually occupied the dominant position in the market.
Because the threshold of lithography technology was relatively low at that time, in the late 1960s, Nikon and Canon of Japan, because of the similarity of their industries, also started to set foot in the market. lithography industry.
By the 1980s, Nikon released its first commercial stepper lithograph, the NSR-1010G, with a more advanced Optical systems and self-developed lenses began to take away from GCA a series of large customers such as IBM, Intel, and AMD.
By 1984, Nikon was able to equal GCA, each with a 30% market share. Ultratech, Eaton, P&E, Canon, Hitachi and a few others were left to divide the remaining 40%.
And this was the year that ASML (Advanced Semiconductor Materials Lithography), the future dominant player in the lithography industry, was able to take a share of the market, with Philips of the Netherlands and ASML was founded with the cooperation of a small company called ASMI.
At the beginning, ASML had only 31 employees and could only work in a simple wooden room outside the Philips building. The 1980s were the "glorious years" of the Japanese semiconductor and photolithography industries, and the rise of GAC was still some time away.
With the semiconductor market slumping in 1986, GAC's new product development stalled, and it was immediately acquired, and then acquired by no one to take over. Shutter.
Ultratech stalled after a management buyout, and P&E's photolithography division was sold in 1990 to the SVG.
In the late eighties, the American photolithography trio had fallen, while Japan's Nikon and Canon had the lion's share of the market and were just getting started The ASML also only got 10% of the market share.
And in the 1990s, it was the era of the Nikon and ASML duopoly, but because of that technical route battle at the beginning of the century The position of the lithography industry hegemon is still unbreakable until today.
In general, in the sixty years of the development of lithography technology, the rapid elimination and transfer of lithography enterprises like a horse-light, in fact, behind the A very real contradiction.
As an upstream industry in chip manufacturing, the sales market for lithography was very narrow, and the sales volume was very limited. But there are dozens.
But the lithography is a highly sophisticated technology that requires huge amounts of money to keep investing in research and development and to keep updating and iterating. The smaller it is, the exponentially more difficult the technology is again.
As a result, if a firm experiences technological stagnation or breakdowns in a product, the one step ahead takes away a few semiconductor manufacturers' The vast majority of orders, and the laggards have also lost key revenue and have been unable to develop and manufacture new photolithography products, losing A chance to win the competition.
Simply put, the logic of the lithography industry is winner-takes-all, and Nikon's defeat is a lesson to be learned.
The Current Situation and Possibilities of China's Lithography Industry
Returning to the issue of China's semiconductor industry breakthrough, one of the core tasks is to realize high-end lithography, especially EUV.
However, when we understand the evolution of lithography technology and the process of migration of the lithography industry, we may be able to face the current incomparable A tough competitive landscape.
First of all, China's entry into the lithography industry is not short-lived, but our accumulation of core technologies and patents is still seriously inadequate.
The limitation of patent technology has become a huge pain that blocks the throat of China's semiconductor industry.
For a long time, Nikon, Tokyo Electron and Canon of Japan have been the major applicants of lithography patents.
After the 1990s, the number of lithography patents filed by ASML has increased dramatically, and a large number of patents have been filed in Japan as well.
In addition, ASML also has a large number of patents in Taiwan, the United States and South Korea.
In contrast, the proportion of lithography-related patent applications in China is still very low, and has not been increasing in recent years.
The monopoly of basic technology and the high threshold of technology research and development may become a major factor that makes it difficult for China's lithography industry to break through.
Secondly, just when we realize that we need to promote the autonomy of the semiconductor industry, Moore's Law of chip manufacturing is already approaching its limits. One of the major limiting factors is that the lithography process is approaching its theoretical limits.
When the chip process process evolves to less than 5nm, how to break the physical and material limits becomes a challenge for the lithography and semiconductor industry. The real-life challenges facing manufacturing companies.
In addition, in order to cope with the increasingly high cost of chip manufacturing, the chip industry is taking the way of mergers and acquisitions between enterprises.
To the most advanced chip production line belongs to only a few chip manufacturing giants such as Intel, TSMC, Samsung and Grofounders. They constitute a monopoly pattern of "you have me, I have you" with raw materials and equipment manufacturers like ASML.
For our domestic lithography industry, we are not only faced with the strict technical patent blockade, but also directly encounter close to the limits of technological evolution stage of the industry, but also to face the overwhelming advantage of ASML in a completely monopolistic position, the technical challenges we launched at this time. It really is destined to be an incredibly tough extreme challenge.
For the general public who are concerned about the breakthrough of the semiconductor industry, I am afraid that they can't be in a hurry and expect China's lithography technology in just a few years. c
We should calmly recognize the reality that lithography, as the most precise, most complex, most difficult and most expensive in chip manufacturing, is the most expensive. The development of lithography equipment has long since ceased to be a project that can be accomplished by a single country or a handful of companies.
To develop the most advanced lithography equipment, it is necessary to work with the world's top light source, optics, materials and key components manufacturers.
Even in the difficult environment where the U.S. is trying to shut down the development of China's semiconductor industry, we cannot give up our cooperation with these foreign advanced technology companies. Opportunities for exchange and cooperation.