Chinese Lab Reports 3.42% Efficiency for EUV Light Source

Researchers at the Shanghai Institute of Optics and Fine Mechanics, part of the Chinese Academy of Sciences, published a March 2025 paper reporting a laser-produced plasma (LPP) extreme ultraviolet light source that reached 3.42% conversion efficiency. The project was led by Lin Nan, an engineer who previously worked at ASML in the Netherlands.

The team drove the LPP source with a solid-state 1-micron laser and reported a peak conversion efficiency of 3.42%. That result exceeds several earlier academic benchmarks, including a 1.8% result from 2021 and a 3.2% result reported in 2019. Commercial EUV light sources typically register conversion efficiencies near 5.5%.

Power output from the laboratory setup remains below commercial levels. Technical assessments place the lab output at roughly 100 to 150 watts of EUV light, compared with about 600 watts for current commercial systems. Higher continuous optical power enables lithography tools to expose more wafers per hour, so manufacturers combine conversion efficiency and raw power to meet production throughput requirements.

Conversion efficiency measures the fraction of input laser energy converted into usable EUV light. LPP systems generate EUV by focusing a high-power laser on a target to create hot plasma that emits photons at EUV wavelengths.

Since 2019, export restrictions have limited access to the most advanced commercial EUV systems. EUV lithography is required for chip nodes below 7 nanometers, and advanced lithography equipment is concentrated among a small number of suppliers.

Separate reports indicate researchers in Shenzhen assembled an operational prototype EUV lithography machine in early 2025 using reverse-engineered elements and parts salvaged from older tools and controlled components. That prototype was not producing functional chips at the time. Researchers described the aim as: “The aim is for China to eventually be able to make advanced chips on machines that are entirely China-made.”

Chinese labs and companies have reported several technical milestones on light sources and other components. The March 2025 paper is one laboratory result among a series of efforts to develop domestic lithography tools.

On the software and compute side, an August 2026 company announcement described Qwen3.8-Max-Preview as a multimodal model with 2.4 trillion total parameters using a sparse Mixture-of-Experts design; the company has not released benchmark results, license terms, or how many parameters are active during inference. The firm’s chip unit announced plans to open-source SAIL, the software stack for its Zhenwu chips, and introduced the Zhenwu M890 on May 20, 2026, reporting 0.6 PFLOPs of FP16 performance, 144 GB of HBM3 memory, and shipments of 560,000 units to more than 400 corporate customers by April 2026.

Transitioning a laboratory light source into a production-grade lithography machine requires higher continuous EUV power, long-term component reliability, and supporting supply chains for optics, stages and control systems. Some reports set 2028 as a target for fielding domestic EUV capabilities that could approach production needs, while noting ongoing technical and sourcing hurdles.