Researchers at Japan’s largest particle physics laboratory propose replacing ASML’s ultraviolet light source with a linear accelerator
If we stick to cutting-edge semiconductor manufacturing, ASML has no competition. This Dutch company is currently the only manufacturer of photolithography equipment that has in its portfolio the extreme ultraviolet (UVE) machines that are necessary to manufacture 7nm or smaller integrated circuits In an efficient way. Canon and Nikon, its main competitors, abandoned the race for the development of UVE equipment due to the enormous financial investment they had to make in the research area.
Curiously, Canon is the only company that could compete with ASML in the medium term, but it still has to prove everything. Your best asset, in any case, is your nanoimprint lithography (NIL) equipment. According to Canon, this machine allows the manufacturing of chips up to 2 nm and is ten times cheaper than a state-of-the-art EUV machine. The price difference is even greater if we compare its cost with that of a UVE and high aperture lithography equipment from ASML: 14 million euros compared to no less than 350 million euros.
Japan is flirting with particle accelerators to make chips
Canon’s NIL lithography equipment has yet to convince large semiconductor manufacturers, such as TSMC, Intel or Samsung, that they represent a real alternative to ASML’s EUV machines. However, this is not Japan’s only asset to regain its former leadership in the photolithography equipment industry. And a group of scientists from the Tsukuba High Energy Physics Research Organization with Accelerators, which is the most important particle physics laboratory in Japan, is working on a revolutionary idea.
The ultraviolet light source is responsible for transporting the geometric pattern described by the mask so that it can be transferred with great precision to the surface of the silicon wafer.
One of the most important components of UVE lithography equipment is the ultraviolet light source. It is manufactured by the American company Cymer, which has belonged to ASML since 2013, and is responsible for transporting the geometric pattern described by the mask so that it can be transferred with great precision to the surface of the silicon wafer. This light belongs to the most energetic portion of the ultraviolet region of the electromagnetic spectrum. In fact, its wavelength extends in the range from 10 to 100 nanometers (nm).
The approach of the Japanese researchers that I mentioned a few lines above is very different. And what they propose is to replace the ultraviolet light source used by ASML with a free electron laser or FEL for its name in English (Free-Electron Laser) like those used in particle accelerators. In fact, in their tests they are using a FEL laser generated by an energy recovery linear accelerator. In theory, the radiation generated by a FEL laser makes it possible to manufacture integrated circuits with a resolution comparable to that of an ultraviolet light source.
A priori it sounds good, but it seems reasonable to accept that an FEL laser linked to a particle accelerator is not exactly cheap. The reason why Japanese scientists consider their solution preferable to the one used by ASML is that a single energy recovery linear accelerator is capable of simultaneously feeding various lithography machines. This simply means that these physicists are convinced that their strategy allows them to obtain the radiation they need to manufacture integrated circuits in a much more efficient and profitable way from a strictly economic point of view.
These Japanese physicists claim that their solution is cheaper than ASML’s
There is no doubt that this proposal is very interesting from a technological perspective, but it raises very serious doubts. In 2021, these same physicists estimated that an energy recovery accelerator cost around $260 million at that time, while EUV lithography equipment was close to $200 million. Also, the accelerator is just the light source.
To its cost it is necessary to add that of the rest of the components that are necessary to manufacture the semiconductors, such as, for example, optical elements or robotic devices. Even so, these Japanese physicists assure that your solution is cheaper than ASML’s. Be that as it may, there is another parameter that is worth not overlooking: particle accelerators are enormous. And they are unlikely to be able to make them small enough to house them in a conventional chip manufacturing plant. We will see how this technology evolves, but there is no doubt that it is worth keeping track of.
Image | KEK
More information | IEEE Spectrum
In Xataka | MADMAX is at CERN and has a very ambitious purpose: to find dark matter
