Researchers from National Yang Ming Chiao Tung University (NYCU) and Taiwan Semiconductor Manufacturing Company (TSMC) have unveiled a breakthrough in Extreme Ultraviolet (EUV) lithography that could redefine the scaling limits of current-generation hardware. Published in the August 2026 issue of Optics Express, the study titled "High contrast EUV imaging enabled by topological quasi phase-only masks" introduces a novel mask architecture that moves away from traditional tantalum-based absorbers in favor of molybdenum (Mo)-based quasi phase-only masks (quasi-POMs). This development addresses one of the most persistent bottlenecks in semiconductor manufacturing: the degradation of image contrast and telecentricity as process nodes shrink toward the sub-10nm regime.
The Technical Challenge: The Limits of Tantalum
For years, the semiconductor industry has relied on tantalum (Ta)-based absorber materials to define circuit patterns on EUV masks. As the industry pushed toward 7nm, 5nm, and eventually 3nm nodes, the physical limitations of these absorbers became increasingly apparent. Tantalum-based masks function by absorbing EUV light to create the necessary contrast for the photoresist; however, they are inherently "lossy." They absorb a significant portion of the incident energy, which creates heat, introduces parasitic effects, and limits the resolution potential of 0.33 Numerical Aperture (NA) EUV systems.
As researchers attempted to scale patterns to 12.5nm half-pitch and below, they encountered the "shadowing effect." Because EUV light hits the mask at an angle in current reflective systems, the thickness of the tantalum absorber creates a parallax error—or telecentricity error—which causes the printed features to shift on the wafer. This misalignment reduces the common process window, making mass production increasingly difficult and yield-sensitive.
Innovation Through Topological Engineering
The research team at NYCU and TSMC proposed a paradigm shift by moving toward "quasi phase-only masks" (quasi-POMs). Unlike traditional masks that rely on thick, light-blocking tantalum, these new masks utilize molybdenum-based structures designed to manipulate the phase of the reflected EUV light rather than just its intensity.
By leveraging the optical properties of topological structures, the researchers engineered a mask that reflects light more efficiently while maintaining the phase shift required for high-contrast interference patterns. The result is a mask that behaves more like a phase-shifting lens than a simple stencil. The simulations conducted by the team demonstrated that these molybdenum quasi-POMs deliver a 35% improvement in image contrast compared to state-of-the-art tantalum absorbers.
Furthermore, the design significantly mitigates the shadowing effect. The researchers reported a 92% reduction in peak telecentricity error, which effectively stabilizes the image placement accuracy. This improvement provides a fivefold increase in the common focus window, a critical metric for chipmakers who must ensure that patterns remain crisp even when the wafer surface topography varies slightly during exposure.
Chronology of EUV Lithography Evolution
To understand the significance of this development, one must look at the timeline of EUV evolution. The transition from Deep Ultraviolet (DUV) to EUV lithography was arguably the most complex transition in the history of semiconductor manufacturing, spanning over two decades of R&D.
- 2010–2015: The foundational era of EUV, where the industry struggled with source power and pellicle durability. Tantalum-based absorbers became the standard, as they were robust and chemically stable.
- 2016–2019: The introduction of High-Volume Manufacturing (HVM) for 7nm and 5nm nodes using 0.33 NA EUV scanners. During this time, the limitations of Ta-based absorbers regarding shadowing effects were documented, leading to the development of thinner absorbers.
- 2020–2023: The rise of High-NA EUV (0.55 NA) development. Industry focus shifted toward new materials, including metal oxides and alternative absorbers to support the increasing resolution demands.
- 2024–2026: The current era, characterized by the exploration of "beyond-absorber" masks. The NYCU/TSMC paper represents the culmination of this trend, moving from physical blocking to phase manipulation.
Comparative Data Analysis
The metrics provided by the research team offer a compelling case for the adoption of Mo-based quasi-POMs. In semiconductor manufacturing, a 35% increase in image contrast is not merely an incremental gain; it is a transformative leap that allows engineers to utilize lower-dose exposures without sacrificing pattern fidelity.
The fivefold increase in the common focus window is particularly relevant for the economic viability of next-generation nodes. In high-volume manufacturing, the process window—the range of focus and dose settings that yield a defect-free pattern—is the primary driver of throughput and cost. By widening this window, TSMC could potentially reduce the number of re-work cycles required in the fab, directly impacting the cost-per-transistor.
The 92% reduction in telecentricity error is perhaps the most impressive technical achievement. Telecentricity error is a primary source of "overlay error," which is the misalignment between different layers of a chip. As nodes shrink, the tolerance for overlay error decreases proportionally. By solving this at the mask level, the researchers have effectively bought the industry more "headroom" to continue scaling without requiring more frequent or more expensive overlay correction steps.
Industry Implications and Strategic Outlook
While the study currently exists in the realm of simulation, the partnership between NYCU and TSMC suggests a clear path toward implementation. TSMC, as the world’s largest contract chip manufacturer, has a vested interest in extending the life and capability of its existing 0.33 NA EUV scanner fleet. While the industry is transitioning to 0.55 NA High-NA EUV systems for leading-edge logic, 0.33 NA scanners remain the workhorse of the industry.
By retrofitting current scanners with high-performance masks, manufacturers can extend the utility of these machines to 12.5nm half-pitch and potentially beyond. This strategy provides an economic alternative to the immediate and total transition to High-NA tools, which carry significantly higher capital expenditures.
Analysts note that the move toward phase-only masks aligns with broader trends in computational lithography. As physical optics hit their diffraction limits, the industry is increasingly relying on software and mask-side manipulation to "trick" the light into producing finer features. The integration of topological design into mask manufacturing could be the next frontier in material science for optics.
Challenges to Adoption
Despite the impressive results, the transition to molybdenum-based quasi-POMs is not without challenges. Molybdenum is highly sensitive to oxidation and environmental conditions. Transitioning from the chemically inert tantalum to a new material stack will require updates to mask cleaning processes and long-term storage protocols.
Furthermore, the manufacturing of these topological masks involves complex deposition techniques. Ensuring that these masks can be produced with defect densities low enough for commercial use—specifically the "zero-defect" standard required for EUV masks—will be the primary hurdle for the industry in the coming years.
Conclusion
The research published by Fang et al. represents a sophisticated intersection of material science and optical physics. By successfully demonstrating that quasi-phase-only masks can outperform traditional absorbers in key performance metrics, the collaboration between NYCU and TSMC has provided a clear roadmap for extending the capabilities of existing lithographic infrastructure.
As the semiconductor industry continues to push toward the angstrom era, the importance of innovations at the mask level cannot be overstated. If these molybdenum-based masks can be successfully transitioned from simulation to HVM, they will likely serve as a foundational technology for the next generation of sub-10nm chip production, providing the contrast and process latitude necessary to maintain the pace of Moore’s Law. The industry will now be watching closely to see how quickly these findings can be integrated into the pilot lines of leading foundries.
