In an interview of Jerome Paye, CEO of TAU Systems, with TimesTech, Jerome outlines how compact particle accelerators and X-ray free-electron lasers could overcome the physical and economic limits of EUV lithography. He explains how shorter wavelengths, higher photon efficiency, and compact system design may reduce multi-patterning, improve throughput, lower wafer costs, and enable scalable manufacturing for AI-driven, next-generation semiconductor applications worldwide.
Read the full interview here:
TimesTech: EUV lithography is nearing its physical and economic limits. From a semiconductor manufacturing standpoint, what is the most critical bottleneck TAU Systems is addressing today?
Jerome: TAU Systems is developing the next generation of light sources for semiconductor manufacturing through compact particle accelerators and X-ray free-electron lasers. Our laser wakefield acceleration technology creates electron beams with energies equivalent to conventional accelerators spanning hundreds of meters, but we achieve this in just centimeters. We then send these high-energy electrons through magnetic undulators to produce tuneable X-ray lasers with wavelengths significantly shorter than current EUV systems.
Current EUV lithography machines cost around $400 million each, weigh over 300,000 pounds, and are about as evolved as they can get with current tech. Only a few percent of the light reaches the wafer, dramatically limiting throughput. At the 13.5-nanometer EUV wavelength, chipmakers must use multi-patterning to create smaller features, which adds time, decreases throughput, and increases costs. ASML’s High-NA approach of increasing numerical aperture is reaching fundamental physical and economical limits.
We’re taking the alternative path: reducing the wavelength itself. Our X-ray lasers operate at tuneable wavelengths which will be optimized for maximum transmission. Combined with wavelength-matched reflective optics offering higher reflectivity than current EUV mirrors, our technology delivers hundreds of watts of X-ray emission per compact machine. Matching or exceeding ASML’s power but at shorter wavelengths. The result is faster production, reduced multi-patterning, and dramatically improved energy efficiency.
TimesTech: Current EUV scanners are massive, power-hungry, and extremely capital-intensive. How does TAU’s compact free-electron laser fundamentally change the cost, energy, and footprint equation for fabs?
Jerome: Our goal is to replace the EUV light source for modern lithography machines with highly efficient X-ray lasers. Our light sources will have a high wall-plug efficiency from electrical to optical energy. Unused energy will be recaptured to further improve the efficiency and our X-ray laser, combined with the higher reflectivity of wavelength-matched reflective optics, will reduce operating costs by increasing production speed.Using a bright source of X-ray light will speed up production for each lithography step and reduce the need for multi-patterning. This will reduce the associated costs per wafer.
TimesTech: You’ve highlighted brighter light output and shorter wavelengths. How does this translate into real manufacturing benefits such as finer feature resolution, higher wafer yield, or faster throughput?
Jerome: The physics is straightforward: shorter wavelengths enable direct patterning of smaller features without the multi-patterning steps that define current EUV limitations. At ASML’s 13.5-nanometer wavelength, creating features significantly smaller requires printing patterns multiple times, each pass adding cycle time, reducing throughput, and introducing alignment errors that impact yield. Our tuneable X-ray wavelengths eliminate this constraint, enabling single-exposure patterning where competitors require multiple passes.
Throughput improvements derive from superior photon efficiency. Current EUV systems lose the majority of generated light through inefficient tin-droplet plasma sources and mirrors, only a few percent reaches the wafer. Our X-ray lasers combined with wavelength-matched reflective optics generate hundreds of watts per compact machine, matching or exceeding ASML’s output whilst operating at shorter wavelengths with higher-reflectivity optics. The result: faster exposure times per lithography step, eliminated multi-patterning overhead, reduced defect opportunities, and lower cost per wafer, whilst enabling the atomic-level feature control that next-generation AI and quantum computing chips demand.
TimesTech: TAU has moved laser wakefield acceleration from national labs into a shipping-container-sized commercial system. What were the key engineering challenges in making this technology fab-ready?
Jerome: The fundamental physics has been demonstrated for years, our academic colleagues delivered acceleration gradients 2,000 times stronger than conventional systems, achieving in centimeters what conventional RF technology requires hundreds of meters and vast infrastructure to accomplish. The challenge wasn’t proving the concept; it was engineering for industrial reliability. Laboratory demonstrations prioritize peak performance over consistent operation across millions of cycles. Semiconductor lithography demands extraordinary stability, electron beam energy, timing precision, and spatial characteristics must remain within tight tolerances shot after shot. We’ve invested heavily in laser stability, plasma generation control, and beam diagnostics to achieve this consistency.
System integration and footprint reduction posed equally significant challenges. National laboratory facilities occupy building-scale infrastructure; we’ve translated this to shipping-container-sized installations deployable in existing fab spaces. Our TAU Labs, California facility demonstrates this transition, validating technology whilst generating operational data. Our partnership approach with The University of Texas at Austin, Lawrence Berkeley National Laboratory, and the Extreme Light Infrastructure Nuclear Physics facility combines world-leading research expertise with commercial engineering discipline. The three-part commercialization strategy, radiation testing today, radiotherapy for manufacturing scale-up, and sustained lithography R&D investment reflects the patience required. Each application refines our core technology whilst generating revenue and operational experience for semiconductor market entry.
TimesTech: With AI and advanced computing driving demand for smaller and more complex chips, how scalable is TAU’s laser-driven accelerator technology for high-volume semiconductor manufacturing?
Jerome: Our technology’s scalability derives from several architectural advantages. Each compact accelerator unit drives a single scanner, matching or exceeding ASML’s power output at shorter wavelengths. Fab capacity scales linearly, additional scanners require additional accelerator units, but the fundamental technology remains unchanged. Unlike conventional synchrotron facilities where single large installations must serve multiple beamlines, our compact systems enable distributed deployment matching fab layout requirements. Manufacturing scalability benefits directly from our radiotherapy development pathway, which shares fundamental technology with our lithography platform but addresses a larger near-term market requiring volume production. This establishes manufacturing processes, supply chains, and quality systems at commercial scale before lithography deployment.
The economic model supports high-volume deployment. Current EUV machines cost approximately $400 million each, weigh over 300,000 pounds, and require extraordinary facilities infrastructure whilst consuming massive power relative to usable photon output. Our compact systems housed in existing fab spaces with dramatically improved efficiency offer compelling total cost of ownership advantages. Shipping-container-sized units deploy within existing fab footprints, reducing capital requirements and construction timelines. This flexibility allows fabs to add capacity incrementally rather than committing to massive infrastructure investments years in advance. Unlike ASML’s High-NA approach which faces fundamental physical limits from mirror manufacturing precision, our wavelength reduction pathway offers continuing headroom for performance enhancement as node requirements evolve.
TimesTech: Looking ahead, how do you see compact particle accelerators reshaping semiconductor manufacturing economics and fab design over the next decade?
Jerome: The current manufacturing process demands extreme capital and complexity into individual $400 million tools requiring building-scale support infrastructure. Compact accelerator technology fundamentally changes this equation through distributed deployment of shipping-container-sized systems within existing fab footprints. Fabs gain flexibility to add capacity incrementally matching production requirements rather than making massive upfront commitments, reducing financial risk whilst enabling faster market response. High wall-plug efficiency combined with energy recapture substantially reduces power consumption per wafer. Higher photon efficiency increases throughput whilst eliminating multi-patterning requirements, improving wafer economics whilst enabling more aggressive node transitions than current EUV cost structures support. Perhaps most significantly, compact accelerators provide a viable pathway beyond physical limits constraining current EUV approaches. ASML’s High-NA systems push mirror manufacturing precision near fundamental boundaries; wavelength reduction through X-ray lasers represents the physics-defined solution to atomic-level manufacturing control. Our near-term commercialization through radiation testing and radiotherapy demonstrates technology maturity whilst generating operational experience and manufacturing scale for semiconductor deployment. The industry recognizes atomic-level control ultimately requires X-rays. Compact particle accelerators won’t simply improve semiconductor manufacturing economics, they’ll enable capabilities current technology cannot deliver at any cost.
