Is the entire U.S. chip manufacturing industry being held back by just one thin film?

First, let me show my friends a picture...

On the stage, there were the US President Biden, as well as the CEOs of Nvidia, AMD, and Apple, among other tech industry bigwigs.

However, judging from their positions, it was clear that the main focus of the event was not on them, but on the three central figures from TSMC: founder Zhang Zhongmou, CEO Wei Zhejia, and chairman Liu Deyin.

Creating such a rare and significant scene was the arrival of the first machine at TSMC's US wafer fabrication plant at the end of 2022.

And it's this machine that made Biden excitedly say, "The chip manufacturing industry is coming back."

Yes, you read that right, the chip manufacturing industry in the United States was actually lost...

All along, when people talked about the strength of American chips, there was actually a qualifier: chip design. In addition to design, there are also processes such as wafer fabrication, circuit etching, testing, and packaging, all of which fall under the category of chip manufacturing.

To put it this way, since the last century, their semiconductor manufacturing share has been sliding down like a slide.

Apart from Intel, which is still persisting in making its own chips, other companies like Nvidia and AMD are almost begging Taiwan Semiconductor Manufacturing Company (TSMC) and Samsung for help.

The once semiconductor cradle has been losing its manufacturing business one after another. The reasons behind this have been thoroughly analyzed by everyone in recent years.

The mainstream viewpoint is basically that during the Cold War between the United States and the Soviet Union, the United States supported the chip technology of several East Asian countries and regions, which resulted in Japan overtaking the United States through technology theft.

The Americans found that they couldn't compete and were in the midst of deindustrialization. They placed more emphasis on the top of the value chain, chip design, and then took advantage of the industrial transfer, thinking that they could become the leader through other means and control the chip dominance.

But, does the United States really not care about manufacturing chips well? Recently, Bloomberg has revealed a new detail, perhaps providing an answer. It is said that the United States has been trying to revive its own chip manufacturing industry.

And in the extreme ultraviolet (EUV) lithography machine, the United States almost turned the tables... but because of a film, the United States missed a great opportunity.

The lithography machine should be familiar to everyone, it is a machine that etches circuits on wafers. EUV is considered the most advanced technology now. Before that, the dominant technology has always been deep ultraviolet (DUV) lithography machines.

The "strongest" Silicon Valley lithography group in the United States has been unable to surpass Japan's Nikon and Canon, and the Netherlands' ASML in making DUV lithography machines. Therefore, they are preparing to take a different path and develop EUV technology early on.

Back in 1997, the U.S. Department of Energy, in collaboration with Intel, formed the EUV LLC alliance, which not only brought together three major national laboratories, but also attracted scientists from companies such as Motorola and AMD.

With everything in place, the U.S. did not anticipate that after a comprehensive evaluation, it would be discovered that EUV lithography machines were much more difficult to manufacture than DUV lithography machines. Furthermore, their own Silicon Valley lithography group's technology was simply not up to par and could not make it work.

However, since the alliance had put in so much effort, they had to find a factory capable of producing lithography machines. The only companies with this capability were Nikon in Japan and ASML in the Netherlands.

Before the United States had fully emerged from the shadow of Japan's theft of technology, ASML naturally seized the opportunity.

However, since ASML was not a native son, before reaching a cooperation agreement, it signed a series of "selling oneself" agreements, such as having to purchase more than half of the core components of the lithography machine from the United States, and so on.

After establishing such a partnership, Intel subsequently invested $4.1 billion in ASML, becoming its largest shareholder. It can be said that the great revival of the American semiconductor industry hinges on this battle!

Of course, the fact proves that the efforts the United States has put into ASML have not been in vain. In 2018, the EUV lithography machine went into mass production.

It's time to enjoy the fruits of labor... As long as Intel says a word, I'm sure ASML will supply, but the United States has dropped the ball again. Intel just can't seem to open its mouth and make the purchase.

When ASML first started mass-producing EUV lithography machines, there was a critical technology, EUV thin film, that had not been perfected. With Intel's technology at the time, there was no way to bypass this layer of film, and they did not have the confidence to bet that this technology would be conquered in a short period of time.

Don't underestimate this technology just because of its name; it is a killer feature when it comes to the yield of chip production in lithography machines.

In fact, the operation of a lithography machine is to shrink the circuits on the mask onto the wafer by several times.

But there is a problem in this "copying" process. There may be dust particles and other dirty things in the environment. If they fall on the mask, the lithography machine will print them on the wafer as is.

You should know that the circuits on the chip are extremely tiny. If there is even a small defect, the printed chip can basically be thrown into the trash.

Moreover, according to the report from the foreign media Anandtech, because Intel only had one customer at the time, they generally used single-chip masks. In other words, only one chip circuit was etched on one mask. If there was any dust, the entire wafer would be wasted.

At this time, the role of the EUV film becomes apparent. It is installed a few millimeters above the mask, acting as a "bodyguard" to prevent dust and other things from approaching the "focus" of the mask. If it's not on the focus, it naturally won't be etched onto the chip.

But to produce this thin film is not just talk.

First, the extreme ultraviolet (EUV) film has stringent requirements for withstanding extreme ultraviolet light, which eliminates most materials. Even if the requirements are met, the film must be made extremely thin to ensure sufficient transparency.

After thinning, the film must also withstand the extremely high energy of the extreme ultraviolet light, at least within the temperature range of 600 to 1000 degrees Celsius, without any deformation.

In short, producing this thin film is extremely challenging. This is also why it took three years after the mass production of EUV lithography machines for the EUV film to barely be able to be mass-produced for commercial use.

Since they couldn't wait for EUV film, why did TSMC have the audacity to bring home the EUV lithography machine as soon as it went into mass production?

The reason is quite simple: they dared to take the risk, and they had the capital to do so. When TSMC first purchased the EUV lithography machine, there were rumors that they had developed a laminar air knife technology that was said to be as effective as the EUV film.

Moreover, TSMC has orders from various manufacturers and for various types of chips. According to Anandtech, they mostly use multi-chip masks. For example, a mask with 25 chip circuits, even without using EUV film, can still maintain a yield rate of 96% even if it encounters dust.

In comparison, at that time, Intel didn't even have a good hand in the chip-making game, so they really didn't dare to gamble.

They couldn't even place a bet, so how could Intel join the table?

According to Bloomberg, it was precisely because Intel didn't use EUV lithography machines in time that the United States missed the opportunity to return to the peak of the chip manufacturing industry.

However, from the perspective of the world, it's not that Intel's failure to purchase EUV lithography machines has caused a decline in the US chip manufacturing industry, but rather that the industry was already in decline, and the EUV lithography machines simply exposed this fact.

Of course, the US is well aware that its chip manufacturing capabilities are lacking. In recent years, they have been making both overt and covert efforts to strengthen their own chip manufacturing industry.

As for Intel, in 2021, they made a big move by announcing the IDM 2.0 plan. In summary, they are also aiming to emulate TSMC, expanding their production capacity in preparation for contract manufacturing. They have even established a dedicated contract manufacturing service division for this purpose.

Even now, with the emergence of more advanced lithography machines, Intel is not hesitating. For example, earlier this year, Intel wasted no time in placing an order for ASML's new High-NA EUV lithography machine, which has already been announced as assembled a few days ago.

On the other hand, TSMC seems to be taking a wait-and-see approach this time...

The United States has also become clever, planning to set up more baskets to hold eggs. In recent years, they have been finding ways to lure TSMC to build factories in the United States. In addition to the chip subsidy bill, the government has also been pressuring TSMC. Now, TSMC has promised to build three factories in the United States.

However, it's easy to lose the chip manufacturing industry, but it's not so simple to regain it. In 2021, when TSMC started building the factory, they encountered difficulties. The United States can't even recruit the basic talent needed for building the factory, and the production time has been pushed back from 2024 to 2025.

From the mistakes made by the Americans in chip manufacturing, it seems that there are not many opportunities for overtaking in the chip industry, but rather more of a linear advancement with the cooperation of the entire industry chain.

As for when the United States will be able to revive its own chip manufacturing industry, we don't know. At least in the short term, it seems hopeless.

Image and information sources:

Reddit, Internet

Bloomberg, The US Had a Chance to Lead in Chipmaking Tech, and Missed It

Semiengineering, EUV Pellicles Finally Ready

Anandtech, EUV Pellicles Ready For Fabs, Expected to Boost Chip Yields and Sizes

AI Chip World, In-depth | When EUV Pellicles Become the Key to "High-Precision" Chip Yield

Why did Intel's process technology, which had been leading, suddenly get stuck at 14nm for so long?