EUV phase error has become a key mask control variable in high-NA patterning

Overview

The report was presented by Patrick Naulleau of EUV Tech Inc. during the FCMN 2026 conference, focusing on deterministic phase control, stochastic phase perturbations, and their measurement methods in EUV patterning. The report argues that EUV light sources have finite spectral and angular bandwidth, and that multilayers, absorbers, and material dispersion jointly cause rapid phase variation; after entering high-NA and smaller-node regimes, angstrom-level material effects and roughness can also translate into lithographically meaningful defects, making actinic metrology at the operating wavelength necessary.

Core views

Deterministic phase can be optimized through mask structure and material design, and phase-shift masks have been shown to improve NILS and imaging contrast; however, actual EUV phase is also affected by wavelength, incidence angle, multilayer reflection, absorber thin-film terms, and three-dimensional pattern edge effects, so a single thin-film phase cannot fully represent pattern phase. Stochastic material fluctuations further create phase roughness, which is amplified through speckle and resist MEEF into line-width roughness, statistical defects, and hotspots. Based on this, the report emphasizes that mask R&D, process control, and inspection all require actinic scattering and microscopy metrology capable of directly measuring the real EUV response.

Analysis framework

The report first uses imaging contrast and NILS cases to illustrate the value of phase control, then decomposes EUV reflection response into multilayer and absorber contributions and examines wavelength and angular dependence; it then extracts phase through broadband spectral amplitude response regression and compares thin-film phase with pattern phase. Finally, it measures phase roughness using actinic scattering, combines microscopic speckle observations with a quasi-rigorous speckle impact model, and maps mask roughness to line-width roughness and the number of statistical defects per mask.

Methodology notes

• Optical MetrologyActinic Metrology

  Directly measuring mask response at the actual EUV operating wavelength and relevant angular range.

  This method preserves multilayer, absorber, material dispersion, and three-dimensional pattern effects, and is regarded by the report as the core means needed to identify lithography-relevant mask effects.

• Phase RetrievalBroadband Spectral Amplitude Regression

  Separately measuring the responses of multilayer regions and absorber stack regions as a function of wavelength and angle, and determining phase through regression.

  This method is used to address the issue that EUV is not strictly monochromatic and that phase changes rapidly with spectrum and angle.

• Pattern MetrologyScatterometry

  Extracting pattern phase from a volumetric data space composed of angle, diffraction angle, and wavelength.

  Scatterometry can observe multilayer and absorber three-dimensional edge effects that thin-film models cannot include, and quantify the deviation between pattern phase and thin-film phase.

• Defect ModelingQuasi-Rigorous Speckle Impact Model

  Propagating phase roughness to image-plane speckle, line-width roughness, and statistical defect probability.

  The model combines roughness, resist MEEF, and imaging conditions to estimate the number of defects or hotspots per mask, revealing that risk is highly nonlinearly sensitive to roughness and MEEF.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• EUV Tech Inc.

  The company to which the report author belongs and also a direct participant in EUV actinic metrology tools and the EUV mask field.

  Strengths

  The company states that it originated from Lawrence Berkeley National Laboratory, focuses on operating-wavelength EUV metrology, and says its tools are used in the recording process for all EUV mask blanks.

  Weaknesses

  The report does not disclose financial data, customer concentration, competitive landscape, capacity, orders, or commercialization growth rates, so valuation judgment cannot be completed on this basis.

  Comparison

  Compared with non-actinic methods or methods relying only on thin-film parameters, actinic metrology can directly capture spectral, angular, three-dimensional edge, and phase roughness effects under real EUV conditions.

  Risks

  The relevant advantages mainly come from the company's own statements in its technical demonstration; equipment demand still depends on the pace of high-NA EUV mass production, customer validation, inspection standards, and the development of alternative metrology solutions.

• EUV photomask metrology and inspection industry chain

  High NA and smaller-node regimes raise the technical threshold for phase control, stochastic defect identification, and mask inspection.

  Strengths

  Both phase-contrast optimization and stochastic defect control require higher-sensitivity dedicated metrology, giving potential demand an advanced-process-driven attribute.

  Weaknesses

  Stronger technical demand does not necessarily translate synchronously into equipment purchases, and the report lacks data on market capacity and commercial penetration rates.

  Comparison

  Traditional average-parameter or non-operating-wavelength measurements have difficulty fully reflecting mask imaging performance under actual scanner conditions.

  Risks

  Delayed high-NA adoption, fluctuations in customer capital expenditure, improvements in mask processes, or maturation of alternative inspection technologies may all affect realization of tool demand.

Key data

• Optical contrast case94% versus 67%The corresponding NILS values in the case are 2.95 and 2.10.
• NILS improvement from phase-shift maskapproximately 40%Applicable to the 20 nm half-pitch contact-hole case.
• High-NA resist example9 nm half-pitch, 0.5 NAUsed to illustrate the high requirements of EUV resists for contrast and NILS.
• Phase variation within the source passbandup to 30 degreesReflects the sensitivity of EUV phase to wavelength variation.
• Phase variation within the illumination angle rangeup to 15 degreesIndicates that phase measurement and mask optimization must cover the scanner angular bandwidth.
• Line-width roughness corresponding to 75 pm roughness0.87 nmTable value; the report title summarizes it as approximately 1 nm line-width roughness.
• Statistical defect count at 75 pm roughness7.6×10^5 to 8.7×10^8Model results increase significantly as resist MEEF rises from 1.0 to 1.5, showing strong nonlinearity in defect risk.

Impact & implications

As EUV evolves toward high NA, smaller half-pitches, and tighter process windows, mask phase control will shift from managing average thin-film parameters to joint management of spectral, angular, three-dimensional pattern effects, and stochastic roughness. This is expected to increase the strategic importance of actinic mask metrology, scatterometry, and high-sensitivity inspection tools, but the report provides no market size, orders, revenue, or earnings forecasts, so the investment implications can only be judged qualitatively.

Risks

• Multilayer and absorber responses change rapidly with wavelength and angle, and simplified models may misjudge the true pattern phase.
• There is deviation between thin-film phase and pattern phase, especially at small pitches where three-dimensional edge effects are more pronounced.
• Angstrom-level material thickness variation and roughness may translate into significant phase perturbations.
• Image-plane speckle generated by phase roughness may increase line-width roughness, intrinsic defects, and stochastic hotspots.
• Statistical defect counts are highly sensitive to resist MEEF, and small parameter deterioration may cause defect counts to rise by orders of magnitude.
• The report is technical demonstration material and lacks independent validation, financial metrics, and quantification of commercial demand, so it cannot directly support securities investment conclusions.

What to watch

• The adoption progress of high-NA EUV mass-production nodes and patterning near 10 nm half-pitch.
• The scope of adoption of actinic metrology in EUV mask blanks, finished masks, and process control.
• Whether industry control thresholds and inspection standards for mask phase roughness tighten.
• Whether pattern phase measurement can stably cover the full scanner spectrum and illumination angle bandwidth.
• The NILS improvement, defect rate, and process-window performance of phase-shift EUV masks in actual mass production.
• Measured validation results for the relationship among roughness, line-width roughness, resist MEEF, and the number of hotspots per mask.