ASML and its partner Carl Zeiss expect that a successor to the High NA EUV machine could be ready in about ten years. Engineers from both companies describe this in a peer-reviewed paper. The so-called Hyper NA machine is designed to print smaller structures and reuses much of the existing technology.
The paper will appear in the October issue of the Journal of Micro/Nanopatterning, Materials and Metrology. According to the authors, Hyper NA can print features as small as 5 nanometers. That is more than a third smaller than what ASML’s current High NA EUV machine can handle.
No production decision has been made yet. ASML has begun development but has not committed to actually building the machine. The Veldhoven-based company did not respond immediately. The engineers also note that challenges remain.
ASML previously indicated it would look beyond EUV. At the same time, earlier reports raised the question of whether such a Hyper NA step would ever become economically viable, now that the chip industry is approaching the limits of scaling.
Mirrors are already available
The hardware appears to be less of an obstacle than expected. According to the paper, Zeiss can already manufacture mirrors sufficient for Hyper NA. And the machine will be barely larger than the High NA systems, which are already the size of a double-decker bus.
The light source also doesn’t need to be overhauled. According to the authors, ASML’s current source can be “reused unchanged” for Hyper NA. In a footnote, they note that an alternative technique using free-electron lasers would not change the exposure system’s design.
This is relevant for American startups. xLight is developing free-electron-laser-based light sources, which Elon Musk may prefer for his Terafab chip factory in Texas. Substrate, in turn, is working on X-ray lithography.
Why are these smaller features so important? Chipmakers are increasingly improving their chips with 3D designs, but smaller structures remain critical for speed and power consumption. Lithography is a limiting factor.
After Hyper NA, things will get more challenging. According to the authors, future machines will likely need to use shorter-wavelength light. For now, this remains an academic research topic.