The machine that makes every advanced chip on the planet depends on a laser most people have never heard of, and engineers are only beginning to understand whether that laser can be made dramatically more efficient. A multi-institution Japanese team has just published the most systematic study yet of what a better light source for extreme-ultraviolet (EUV) lithography could look like.
The work, posted to the arXiv preprint server on 4 June 2026 and accepted to Applied Physics Letters, comes from researchers at the University of Osaka, the National Institute for Fusion Science, the National Institutes for Quantum Science and Technology, and Osaka Metropolitan University. It is led by Professor Shinsuke Fujioka, a plasma physicist with a long record of work on laser-produced tin plasmas, the hot, dense state of matter that EUV light sources rely on to emit at the right wavelength.
The problem the team is trying to solve is a manufacturing one. Every EUV scanner in a high-volume chip fab today drives its plasma with a carbon-dioxide (CO2) laser operating at 10.6 micrometers, a wavelength that was inherited from early design choices and has stuck because it works. Those drive lasers are large, power-hungry, and account for a meaningful share of the tool's overall cost, footprint, and electricity bill. As EUV moves deeper into high-volume manufacturing, those costs scale with it. Solid-state mid-infrared lasers, similar in concept to the fiber lasers used for metal cutting, have been a research-stage alternative for years, and the open question is whether they can be made efficient enough to take over.
The Osaka group's answer is a map. Using a one-dimensional radiation-hydrodynamics code called STAR-1D, the team ran a grid search of more than 140,000 combinations of drive-laser wavelength, pulse width, and target size, then checked the predictions against existing tin-plasma experiments. The result is a publicly validated picture of where in laser-parameter space the next efficiency gains are most likely to come from, which broadens the design choices available to source engineers rather than picking a single winning architecture.
Two numbers stand out. Across the full sweep, the global conversion-efficiency maximum is 5.63%, hit at a drive wavelength of 5.5 micrometers, a figure the team says is in good agreement with recent experiments. For the practically important case of a 2-micrometer solid-state driver, the same sweep predicts a maximum of 4.64%. That is the number that matters most to chipmakers: it is the highest published, experimentally validated estimate for the architecture that real commercial tools would actually use, and it sets a target for solid-state laser development to chase.
The mechanism behind those numbers is technical but tractable. The team identifies a three-way trade-off: the drive laser has to heat a tin droplet to the electron temperature and density that produces the strongest EUV emission, while still being absorbed efficiently by the plasma, and while limiting how much of the EUV light the surrounding plasma re-absorbs before it can escape. The optimum pulse width and droplet size fall out of that balance, and they shift in predictable ways as the drive wavelength changes. Knowing the shape of that trade-off gives source engineers a way to choose laser parameters by design rather than by trial.
What the paper does not do is declare a winner. It does not show a working solid-state EUV source in a fab, and it does not claim that 4.64% conversion efficiency is a record. The numbers are model predictions, validated against experiments at a handful of operating points, and the authors are explicit that the validated regime is what supports confidence in the broader map. The work is a contribution to the design space, not a replacement for the CO2 laser.
That is still a meaningful step. The CO2 drive laser inside an EUV scanner is one of the under-discussed reasons advanced chips are expensive to make, and the search for a smaller, more efficient solid-state replacement has been gated, in part, by the lack of a shared, validated picture of where the efficiency frontier actually sits. A peer-reviewed map of that frontier, with named trade-offs and a public code path, expands the options available to the next round of EUV-source engineering. The Osaka team has not ended the search for a better EUV light source. It has narrowed it.