The semiconductor industry is relentless in its pursuit of scaling. While High-NA EUV (Numerical Aperture of 0.55) is just beginning to make its mark, ASML and the industry are already looking toward the next massive leap: Hyper-NA EUV. Targeted for deployment around 2030, this technology is designed to be the enabler for the true Angstrom era (sub-1nm processes).

What is Hyper-NA EUV?

At its core, Hyper-NA EUV represents a system with a numerical aperture (NA) of 0.75 or higher. In optical lithography, a larger NA allows the system to capture more light and print smaller, sharper features. For the sub-1nm logic nodes, Hyper-NA aims to simplify the manufacturing process by avoiding the extreme complexities and costs of excessive multiple-patterning with High-NA tools.

The Engineering Challenges

Moving from 0.55 NA to 0.75 NA is not just a linear scaling problem; it introduces severe physical and economic hurdles.

1. Imaging Contrast and Polarization: As the angles of light hitting the wafer increase significantly, polarization effects become pronounced. Light waves with certain polarizations cancel each other out, severely degrading image contrast.
2. The Need for Polarizers: To mitigate contrast loss, the system must filter out the problematic light using polarizers. However, this blocks a substantial amount of the light source, meaning the system will suffer from lower power efficiency and slower throughput unless the light source itself is dramatically improved.
3. Razor-Thin Depth of Focus (DOF): A higher NA inherently results in a much shallower depth of focus. Maintaining sharp focus across the wafer will require atomic-level surface flatness and incredibly precise mechanical control.
4. Resist and Defect Challenges: Sub-1nm processes demand much thinner photoresists, which makes etching selectivity difficult. Furthermore, secondary-electron scattering (stochastic effects) leads to a higher rate of random defects—a phenomenon sometimes referred to as the “yield cliff”.
5. Mask Reflections: Traditional mask absorbers lose reflectivity at these high incident angles, necessitating the development of entirely new mask technologies.

Economic Hurdles and Prospects

The engineering feats required for Hyper-NA come with a staggering price tag. Current estimates suggest that a single Hyper-NA EUV machine could cost over $700 million (roughly double the price of a High-NA machine and four times a standard EUV tool).

Despite the jaw-dropping cost and immense power consumption, the technology is largely deemed indispensable. The alternative—using High-NA with extensive multi-patterning—could eventually become even more expensive and prone to yield loss.

Key adopters like TSMC, Samsung, and Intel are currently conducting feasibility studies with ASML and research hubs like imec. While the foundries may try to maximize the lifespan of High-NA to delay the massive capital expenditure, the roadmap to sub-1nm logic will ultimately rely on Hyper-NA breaking through these physical barriers by the end of the decade.

Intelligence is just guided flow of electrons ⚡️. But guiding those electrons at the Angstrom scale is proving to be the ultimate engineering test.