A Japanese team reported a 40% jump in conversion efficiency for 2 micron driven extreme ultraviolet (EUV) sources, by dividing a single laser pulse into two synchronized beams over a tin target with passive optics.
The headline number is real: a Japanese research team has shown that splitting a single 2-micron laser pulse into two synchronized beams over a tin target lifts extreme-ultraviolet (EUV) conversion efficiency from 2.6% to 3.6% in one setup, a 40% jump achieved at half the per-pulse energy. The trick is architectural rather than exotic, and it points to one of the most stubborn problems in advanced chipmaking: how to scale the EUV light source past today's power ceiling.
Conversion efficiency (CE) is the share of laser energy that ends up as usable EUV photons at 13.5 nanometers, the wavelength the most advanced chipmaking tools use to print the smallest features on silicon. Today's production EUV sources fire a high-power carbon-dioxide laser at tin droplets to make a hot plasma that radiates at that wavelength. The industry has spent years pushing CE upward because the alternative, simply building a bigger laser, runs into cost, wall-plug efficiency, and heat-handling walls. Higher CE per joule means more usable light without scaling the driver.
The new work, posted to arXiv on June 6 by Nagahama and colleagues at Utsunomiya University, RIKEN, the University of Tokyo, and Tohoku University, takes a different route. The driver is a 2-micron Ho:YAG laser firing 20-nanosecond pulses at a planar tin target, not the production 10.6-micron CO2-on-droplets setup. In a single-beam configuration, 40 millijoules of pulse energy delivered 2.6% CE. Split the same total energy into two 20-millijoule beams at matched peak intensity, and CE climbed to 3.6%, which the authors describe as the highest reported value for 2-micron-driven laser-produced plasma (LPP) EUV sources.
The mechanism is passive beam splitting: the same pulse, sent through standard optics, lands on the target in two footprints instead of one. The source stayed roughly the same size, 60 to 70 micrometers across, and the energetic-ion spectra from the plasma looked nearly identical between configurations, evidence that the dual-beam geometry was not simply remaking the plasma at higher energy density. Lower per-pulse energy matters on its own: production EUV sources want to avoid pushing single-pulse energy too high, because debris and ion damage scale with it.
The paper frames the result as a path toward multi-kilowatt-class EUV sources for high-NA and hyper-NA lithography, the next two generations of projection optics (numerical aperture is the angle at which light is collected to print finer features). That outlook belongs to the authors, not to the production line. The result is a benchtop demonstration on a planar target with a 2-micron laser, not a drop-in replacement for the CO2-and-tin-droplet architecture inside a working scanner.
Several caveats apply, and the source packet is candid about them. The 40% boost is a single-group result, with no independent replication cited. "Nearly identical" plasma conditions are the authors' own characterization, and the field is exploring parallel routes, including pre-pulse shaping of the tin target, that could compete with or complement the dual-beam approach. The 3.6% figure is a record within the 2-micron LPP category, not a record for all EUV source classes; mature CO2-driven production systems operate in a different regime. The paper is a four-page preprint with five figures, not a peer-reviewed publication.
The architectural frame is still worth holding onto. The standard way to chase higher EUV output is to scale the laser: more joules per pulse, more pulses per second, more wall-plug power. This work suggests a cheaper lever sits in the beam geometry, specifically, distributing the same total energy across multiple lower-energy beams at the target without changing the driver hardware. The authors note that the scheme extends readily to three or more beams, which would let a fixed-energy laser yield even more EUV light per shot if the scaling holds.
What to watch next is whether other groups can reproduce the 2.6%-to-3.6% lift on similar equipment, whether the dual-beam scheme still helps when the planar tin target is replaced with the tin droplets a production source would actually fire at, and how the approach stacks up against pre-pulse techniques already in the literature. None of those questions is answered here, and the source does not claim otherwise. What it does claim, on the evidence of a single arXiv preprint, is a clean constructive result: in this lab, in this configuration, splitting the pulse worked.