From：
The origin of photolithography for chip manufacturing is with the United States, but why does the United States no longer produce photolithography now? How did the U.S. lithography machine, which once dominated the global market and held a technological advantage, fail in just ten years? To answer these questions, we have to mention the strong rise of Japan’s semiconductor in the 80s, just experienced nearly 40 years of economic takeoff, the capital’s housing prices straight to London New York, GDP ranked second in the world, while the government vigorously promote industrial upgrading, focus on the development of semiconductor electronics industry, to the United States chip caused a strong impact, and triggered the latter’s trade sanctions, it is the 80s Japan, in the last century This U.S.-Japan semiconductor showdown, photolithography equipment is an important battlefield for both sides fierce competition, and ultimately, the United States lost the monopoly of the global photolithography industry, but Japan is not the winner of the last laugh.
Rise of PE：
The last video we reviewed the 60s and 70s, the earliest contact progressive lithography in the ancient days of semiconductors, and the mask lithography they represent (Shadow Printing ), this primitive lithography machine, collectively known as Mask Aligner, that is, mask aligner, essentially a large ultraviolet lamp, lithography to align the mask plate and stick to the silicon wafer. Not only is it easy to leave defects on the photoresist layer of silicon wafer, resulting in low lithography yield, the production of a batch of chips, often only 10% pass rate, while photoresist and floating dust particles, but also easy to cause damage to the expensive photomask, every dozen lithography, you have to replace the new mask plate, the production costs are huge.
In order to solve the problems of mask lithography, an optical instrument factory named Perking Elmer (PE), with the financial support of the U.S. military, began to develop projection lithography technology and successfully launched the groundbreaking Micralign100 in 1974. This lithography machine, using two coaxial spherical mirrors, projected the image on the photomask onto the silicon wafer after three reflections.
Perkin Elmer’s technical innovation, so that the price of the chip from luxury to cabbage, so that the red and white machines and personal computers, the opportunity to fly into ordinary people’s homes, but also brought great commercial benefits to themselves, previously did not do photolithography, in the halfway PE, to the second year of orders will be in short supply, within 5 years, the market share from 0 to the industry’s first. The story of Micralign’s rise to prominence is the content of the first lithography war in the last episode, so I won’t repeat it here.
American Civil War：
Before the PE cross-border lithography, ruling this market is Geophysical Corporation of America, referred to as GCA, from the name you know, it originally did the map, but GCA as early as 1959, acquired a small company, David Mann, access to precision measurement technology, combined with the optical lens provided by Dr. Lund, it launched the first 971 lithography machine in 1961 and was a great success. In 1961, it launched its first 971 lithography machine and was a great success, after 10 years, Suss, K&S and other manufacturers, also launched their own lithography machine, but GCA’s market share, has been maintained at more than 60%, until the 70s, halfway to kill a Micralign. like Texas Instruments, Intel, such a large factory, and quickly and the new love of PE hit the hot, old love As one of the originators of photolithography, GCA was so upset about this gap that they wanted to develop a more advanced photolithography machine than projection aligner to regain the lost market and customers, which is the Stepper using doubled shrinking photomask.
In the early 70s, GCA’s lens supplier, has changed from Dr. Len to Nikon, a manufacturer from Japan good at doing telescope and camera lens, however, in order to get the telecentric lens needed for stepper lithography, GCA and abandoned Nikon, because they dislike the Japanese lens dish to S, so turned to Zeiss, Germany, they dislike the poor precision of the Japanese lens, lithography phase difference by The focal length is too big, and it is this abandoned supplier, a few years later will become their own gravedigger. But GCA can’t think that far ahead, at the moment, in order to beat Micralign, in addition to the lens, their stepper lithography, but also need a key thing, automated silicon wafer workpiece table.
Aligner vs Stepper：
Here is a brief explanation of the difference between aligner and stepper: aligner uses a 1:1 photomask, regardless of whether it is contact-progressive or Micralign, the circuit diagram on the mask is as big as it is on the silicon wafer. The biggest advantage of this is the speed of lithography, as the mask size is essentially the same as the silicon wafer, and a single lithography can expose a whole wafer, but the disadvantage is that the image contrast, numerical aperture and accuracy are limited.
In contrast, stepper lithography uses a scaling projection method, using an optical lens to reduce the image on the mask plate to 1/5 or 1/4 of its original size, and then project it onto the silicon wafer surface. However, the disadvantage of this is also obvious: the exposure area for lithography is now no longer the whole silicon wafer, but only a small area on the wafer, so lithography has to be reduced to zero, and for each exposure, the platform carrying the wafer must automatically pan to the next area, advancing step by step until the whole wafer is exposed, hence the name, step lithography.
Compared to large-area exposure aligners, stepper lithography sacrifices speed for accuracy, but this movable silicon workpiece table is not simple, especially when the chip process is smaller, the requirements for accuracy measurement and control, acceleration and deceleration become more stringent. Early workpiece tables were usually servo-motor driven two-dimensional platforms, but later they were an air or magnetic levitation moving system with a laser interferometer or scale measurement and control system, forming a complex precision component with optical, electronic and mechanical integration.
In 1978, GCA initially solved the technical problems of the silicon platform and launched the DSW4800, the world’s first automated stepper lithography machine, which used G-line light and a 10:1 doubled shrinking photomask. In 1981, sales soared to $110 million, which put pressure on PE, but that year they also launched the new Micralign 500, the first lithography machine with a capacity of over 100 wafers per hour. This was the first lithography machine to exceed 100 wafers per hour, maximising the speed of the aligner.
The size of chips, however, is shrinking at a rapid pace according to the script of Moore’s Law, and the accuracy of lithography is becoming more and more important, just a few years after PE came out on top, still reveling in its victory and not realising that the era of the aligner is coming to an end, and that GCA will soon return as king, step by step, with its stepper lithography.
Since then, the global lithography market has been dominated by American manufacturers, and whoever wins the American Civil War is the world’s number one. However, just as the two American manufacturers were engaged in a final showdown, in 1982, another lithography machine from overseas appeared in the factories of IBM and Texas Instruments, both in terms of optical system and silicon platform, which looked just like GCA’s stepper machine, and it was called NSR-1010g from Nikon in Japan – a former supplier of GCA.
VLSI Japan：
How did this Japanese manufacturer of lens accessories manage to break the monopoly of GCA technology and develop its own stepper lithography machine in just a few years? The story begins in 1976, when Japan’s Ministry of Communications and Industry launched the VLSI project, a four-year plan to upgrade the nation’s electronics industry, with an annual investment of 18 billion yen. These competitors are being asked to put aside their differences and work together to focus on the big things.
Regarding the VLSI project and the US-Japan semiconductor battle in the 1980s, which is a much grander topic, today we focus on one of the partial photolithography battlefields. In the four-year plan, Japan chose several technological routes to focus on breaking through, and photolithography technology and equipment was one of them, so Nikon and Canon, as old optical manufacturers, although they did not explicitly join the project, also started their own photolithography imitation tasks organized by the Ministry of International Trade and Industry in the Canon had previously focused on camera lenses and lacked in the precision measurement part, so it copied the lower threshold Micralign aligner; while Nikon was able to make both high-resolution camera lenses and precision measurement and control technology accumulated from astronomical telescopes, both of which are pre-requisite skills for unlocking step-by-step, so Nikon was more expected Nikon began to copy GCA’s lithography.
From cottage to beyond：
Although its own technology accumulation is not bad, but Nikon to achieve from scratch is not easy, good, there is the full support of friendly forces, Japan Electric bought the GCA lithography, to Nikon secretly dismantled research, how is this thing known? At that time, GCA’s production capacity was extremely limited, and the stepper lithography was given priority to US-based customers, so this imported lithography was not easily bought, and NEC could not afford to give it to them for nothing.
With reference samples, Nikon’s development process went smoothly, and in 1980 it launched its own stepper lithography machine. The first prototype had many problems, of course, but NEC and Toshiba supported domestic production and were quick to buy it and give feedback on their experience. The Americans were surprised to find that Nikon’s “copycat” machine performed better, particularly in terms of lens stability and automation, and was capable of running independently for hours, whereas GCA’s “original” lithography machine required operator intervention at all times. What’s worse, the Japanese service attitude was far superior to that of the Americans.
There was a joke in the industry at the time that if you bought a GCA lithograph, you would get a maintenance manual with the words “our best lithograph”, while if you bought a Nikon lithograph, you would get a machine care package of five engineers. The reason for this was that the single lens supplier, Zeiss, was at a low ebb at the moment, with frequent lens quality problems and frequent delays in delivery, and was unable to meet GCA’s need to expand its production capacity. I-line lithography with upgraded light sources.
In the same year another Japanese manufacturer, Canon, also launched its first stepper lithography machine, the FPA-1500, and in 1985, Nikon officially overtook GCA as the number one supplier of lithography machines, while at the same time a small company with less than 100 employees in a small Dutch town had just got its new name: ASML Asmac, and a new era began, poorly. The following year, the GCA throne had not yet been warmed up, and this year it made a huge loss of US$145 million, abandoning its low-end models and breaking its arms to survive, betting its entire fortune on high-end machines, but still unable to save itself from a broken capital chain and the defeat of Zeiss withdrawing from the partnership, and by the end of the 1980s, the American lithography industry, which had once monopolised the world, was already in a state of flux and dying; while Nikon Canon of Japan, however, had emerged strongly, firmly occupying 70% of the market.
Using history as a lesson：
Why did the Japanese lithography machines rise quickly and completely defeat the American monopoly back then?
Firstly, their technological accumulation in optical equipment and precision machinery, which was the prerequisite for Japan’s ability to catch up with the United States quickly.
Second is the government’s concentration of power to do great things, the VLSI project to pay for, shake people, set up joint laboratories, to achieve a dream linkage between upstream and downstream manufacturers and even competitors, Toshiba and NEC were willing to pay for imperfect lithography prototypes to underwrite, but also happy to provide technical support to Nikon Canon; while the U.S. chip makers were reluctant to share production details with GCA, for fear of leaking to competitors.
Third, the Japanese manufacturers are more vertically integrated, Nikon and Canon use their own stuff no matter from the lens to the platform, solving the demand from the source, communicating more quickly in R&D, more accurate product iterations and cheaper production costs, while GCA is completely dependent on Zeiss lenses, which will backfire on itself once the other side has problems with quality control or communication, a shortcoming that also appeared in the early cooperation between Asmac and Zeiss later on.
Fourthly, the American manufacturers were old-fashioned and arrogant, PE was stuck in their ways, holding on to the Micralign series all the way to the end, waiting until they lost the market before rushing to transition; GCA was so blind that when customers gave feedback that Nikon’s machines performed better, the management just dumped the pot on the sales team, not believing that they would be technically behind others.
Fifthly, the Japanese manufacturers had a better business philosophy and service support. GCA’s service team in Asia, all of whom were expatriate Americans, only started to employ local engineers in the later years, while Japan, in contrast, took the needs of its customers very seriously and set up a branch in Silicon Valley to establish its own local support network, in 1982, when Nikon sold its first stepper to the United States.
Of course, in addition to the differences between the US and Japanese manufacturers, it is also important to consider the course of history: the late 1970s were still in the pre-development stage of the semiconductor industry, various lithography technologies grew wildly, professional barriers were not deep, technological explosions were frequent, the market for lithography machines turned around, and Japan’s quick counter-attack was difficult to be replicated when put into the present day, when commercial lithography is basically mature and cutting-edge R&D has come to a standstill.
There was also the economic background at the time, the United States was still in recession caused by the oil crisis, GDP fell, unemployment rose, the Federal Reserve had to implement monetary tightening policy, while the Japanese economy was in a thriving upswing, especially the semiconductor industry, the 80s Japanese manufacturers, in the memory chip market to attack, beat the United States chip retreat.
1980 Texas Instruments revenue shrunk 14%, AMD net profit fell 60%, Intel was forced to lay off 2,000 employees, after which it was even withdrawn from the Dram storage business, but Japan is indeed a bit floating, the United States is not going to see itself being overtaken in all aspects, in addition, under the hands of a large number of little brothers, but also must not let one person eat meat, the eyes of the crowd, now technology and production are rolled up but the Japanese, that is on policy.
Conclusion：
Just in Nikon photolithography defeated GCA in 1985, the Plaza Accord was signed in New York, and in September of the following year, the US-Japan Semiconductor Agreement was introduced, imposing a mandatory 100% punitive tariff on Japanese chips, since then Japan’s chip industry began to flourish, they won the battle of photolithography, but are losing the semiconductor war, Japan spit out the ecological bit, allowing Samsung’s memory chips and TSMC’s process manufacturing However, the technological advantage in photolithography, Japan has maintained for nearly 20 years, until Europe’s Asmac feathered its nest, launched a succession of twin-scan (dual platform), Immersion (water immersion) and EUV (light source) three major battles, and finally cut the Japanese duo to the ground.
