Reflective multilayer photomask structures and methods for 11.3 nm extreme ultraviolet lithography
Reflective photomasks at 11.3 nm with multilayer stacks and absorber layers address the limitations of 13.5 nm EUV lithography, enhancing imaging performance and reducing variability for improved semiconductor manufacturing.
Patent Information
- Application Number
- PCT/US2026/015798
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2025-05-09
- Filing Date
- 2026-02-19
- Publication Date
- 2026-11-12
AI Technical Summary
Current 13.5 nm EUV lithography systems face challenges in scaling to higher numerical apertures due to reflectance variation, polarization-dependent imaging asymmetries, reduced depth of focus, and increased stochastic variability, which limit further device scaling and manufacturing efficiency.
Employ reflective photomasks operating at 11.3 nm with multilayer stacks of materials like ruthenium and beryllium, incorporating absorber layers and edge barriers, and utilizing photoresist materials to mitigate these challenges, enabling high-resolution pattern formation with controlled polarization and improved depth of focus.
The solution enhances imaging performance and reduces variability, allowing for reliable high-resolution pattern formation at higher numerical apertures, improving lithographic resolution and semiconductor device progression.
Smart Images

Figure US2026015798_12112026_PF_FP_ABST
Abstract
Description
REFLECTIVE MULTILAYER PHOTOMASK STRUCTURES AND METHODS FOR 11 ,3 NM EXTREME ULTRAVIOLET LITHOGRAPHYCROSS-REFERENCE TO RELATED APPLICATIONS
[0001] This application claims the benefit of U.S. Provisional Patent Application No.63 / 802,837, filed May 9, 2025, the entire disclosure of which is hereby incorporated by reference.FIELD OF THE INVENTION
[0002] The present technology relates to extreme ultraviolet (EUV) lithography, including reflective photomasks, multilayer structures, lithographic systems, photoresist materials, and methods for high-numerical-aperture operation at wavelengths near 11.3 nm.BACKGROUND OF THE INVENTION
[0003] Semiconductor integrated circuits are fabricated in successive technology generations characterized by continued reductions in feature dimensions and increased device density. Advanced logic manufacturing includes generations commonly referred to as the 7 nm, 5 nm, and 3 nm nodes, with critical dimensions for gates, interconnects, and contacts reaching tens and single nanometers. Memory technologies such as DRAM and non-volatile memory such as three-dimensional NAND flash continue to scale through reduced dimensions, increased pattern density, and vertical integration. Semiconductor device patterning at pitch values below about 16-18 nm represents a significant inflection point in semiconductor scaling, as it corresponds to a regime in which continued dimensional reduction directly impacts achievable device density, wiring efficiency, and overall system performance. As feature sizes shrink and pattern density increases, semiconductor manufacturing faces growing challenges related to lithographic resolution, depth of focus, line-edge roughness, stochastic variability, overlay accuracy, and pattern fidelity. Continued scaling of both logic and memory devices depends on advances in lithographic exposure wavelength, optical design, and mask architecture to address the limitations encountered with existing technologies.
[0004] Extreme ultraviolet lithography (EUVL) at 13.5 nm is a primary patterning technology used to fabricate advanced semiconductor devices with dimensions below 20 nm.Current 13.5 nm EUVL systems operate in the spectral region where optical materials strongly absorb radiation and refract weakly. Because of this high absorption, transmissive optics cannot be1 50130535.1used to form images. Consequently, EUVL systems are designed to be fully reflective, relying on reflective multilayer mirror optics and reflective photomasks to direct and project the patterned light onto photoresist-coated wafers. Such reflective optical systems are known as catoptric designs.
[0005] In the early development of EUVL, the choice of multilayer mirror coatings was critical for achieving the necessary reflectance at EUV wavelengths. Constituent materials possess unique, wavelength-dependent optical properties in the EUV and soft-Xray region. The 13.5 nm wavelength was chosen because it lies beyond the L-edge absorption band of silicon (Si) at -12.5 nm (specifically the LI,3 edge) and is in a region where there is a sufficient refractive index contrast through its layered combination with molybdenum (Mo), represented as Mo / Si, with relatively low absorption. In the 1990s, alternative material combinations for other EUV wavelengths were explored. Among these, multilayer mirror coatings comprised of beryllium (Be) with various transition metals were examined for use near 11.3 nm, just beyond the Be K-edge absorption band of -11.2 nm. Multilayer mirrors were investigated for maximum reflectance at normal incidence and spectral response, which are the principal design attributes for coatings used for EUV optical components. While both Si-based systems at 13.5 nm and Be-based systems at 11.3 nm showed potential, concerns about the potential toxicity effects of beryllium compounds led to the abandonment of Be multilayer coatings and the associated 11.3 nm wavelength for widespread use in EUV lithography optics. As a result, 13.5 nm using periodic Mo / Si multilayer coatings have become the standard reflector strategy for EUVL optics. For longer-term scaling beyond the practical limits of 13.5 nm EUV, short er- wavelength soft X-ray lithography in the approximate 6-7 nm range (sometimes referred to as beyond-EUV or BEUV) has been proposed, whereas EUV lithography near 11.3 nm has received limited attention over the past several decades, in part because the incremental resolution gain relative to 13.5 nm has been perceived as modest compared to alternative next-generation approaches.
[0006] Current production 13.5 nm EUVL systems operate with numerical apertures (NA) on the order of 0.33 (or 0.33NA) and a chief-ray angle at the object (CRAO) around 6°, while higher numerical aperture systems, such as those having numerical apertures of approximately 0.55, have been developed to further improve resolution, albeit with reduced image plane focal depth. At 13.5 nm, EUVL at these NA values has reduced reliance on the multiple-patterning techniques used in deep ultraviolet (DUV) lithography by enabling smaller feature sizes to be printed in a single2 50130535.1exposure at a given technology node. In DUV processes, multiple-patterning approaches such as double, triple, or quadruple patterning are often required to achieve comparable resolution, increasing process complexity, cost, and variability. EUVL lithography simplifies process flows by decreasing the number of lithography, etch, and alignment steps required to define critical features. This reduction in process steps can improve overlay accuracy, enhance critical dimension uniformity, and lower defect risk. As a result, 13.5 nm EUV lithography has so far provided a scalable path for continued device scaling in advanced semiconductor manufacturing.
[0007] Recent generation 13.5 nm EUVL systems operating at 0.55NAhas led to increases in the maximum angles directed toward the photomask (aka the mask), collected by the projection optics, and projected into the photoresist layer. As such, these systems use anamorphic projection optics in which the mask pattern is reduced by different magnification factors in orthogonal directions. Specifically, these systems employ a 0.25X magnification (aka 4X reduction) in the one (x) direction and a 0.125X magnification (aka 8X reduction) in the other (y) direction, rather than the uniform 4X reduction in 0.33 NA EUV systems. Leading-edge device lithography using such system approaches employs mask illumination angles reaching the limits of respective optical designs, leading to large angle consequences including reflectance variation, phase effects, and polarization dependence not experienced with previous device generations.
[0008] Looking forward, the industry has begun evaluating even higher numerical apertures, such as 0.75 and above, to extend 13.5 nm EUV resolution further. These higher NA concepts aim to support continued device scaling while avoiding changes to the 13.5 nm EUV wavelength and the associated source, mask, and materials infrastructure. Such higher NAs introduce considerable challenges to depth of focus, which scales with the square of numerical aperture, reducing usable image-plane depth significantly compared to previous NA systems.
[0009] Extending 13.5 nm EUV lithography to numerical apertures above current values also introduces significant challenges associated with the reflective multilayer mask behavior at large angles. As NA increases, the angular spread of rays incident on the mask increases significantly, pushing portions of the illumination outside the angular bandwidth over which sufficient multilayer reflectivity and phase uniformity can be designed for. This leads to larger reflectance and phase variation through angle which can differ widely in x and y directions. Large incidence angles also introduce pronounced polarization effects, as s-polarized and p- polarized EUV light exhibit3 50130535.1different reflectance and phase responses in multilayer stacks. This can lead to polarizationdependent imaging asymmetries and contrast loss which can longer be ignored. In addition, as depth of focus decreases rapidly with increasing NA, substantially thinner photoresist films are needed to remain within focus tolerance, which in turn impacts etch resistance and pattern fidelity. These effects collectively impose considerable mask-related challenges to further NA scaling of 13.5 nm EUV lithography.
[0010] The limitations with scaling 13.5nm EUVL to higher NAs can be illustrated by considering the plot depicted in FIG. 1. This figure is a reflectance plot for a 13.5-nm Mo / Si mask multilayer designed for use in a reflective projection lithography system operating at a numerical aperture of 0.75, adapted from Bottiglieri et al., “Advancing semiconductor patterning with EUV hyper NA: opportunities and challenges,” in Optical and EUV Nanolithography XXXVIII, Proc. SPIE, vol. 13424, Apr. 2025, with labels and annotations added for clarity. The plot shows the unpolarized reflectance of a 40 layer-pair periodic Mo / Si multilayer comprised bilayers of 3.77 nm of Si and 3.32 nm of Mo with a 3.50 nm capping layer of Ru for mask for angles up to 20°. Mask polarization effects are not addressed by Bottiglieri et al. and are not provided as part of the design space. This Mo / Si mask multilayer is designed to be utilized at angles from 0° to 13.4° to cover the angular range illuminating the mask. Specifically, angles range from +10.8° to -10.8° in the 4X x-direction and from +1.4 to +12.2° in the 8X y-direction, rising to +13.4° at field edges. The unpolarized reflectance variance over the full angular range can be calculated from the Bottiglieri reference and is approximately 8.7%, which is dominated by the variation from 0° to about 12°. Such variation is large and the lack of a substantially flat reflectance response across the angular range can lead to degradation in image fidelity, image contrast, and x / y image symmetry, all of which are critical for semiconductor IC lithography. Reflectance variation below about 6% over the full photomask angular range is desirable. Furthermore, applying thin-film modeling using the parameters disclosed by Bottiglieri et al., the resulting difference between s-polarized and p-polarized reflectance (RS-RP) over the full mask angular range can be determined and is approximately 6.5%, which can lead to image degradation. Values below about 5% are desirable, and targets below about 4% may be preferred to maintain image performance at the highest resolution.50130535.1
[0011] A series of simulations using the design parameters described in Bottiglieri et al. further demonstrate the challenges encountered as further NA scaling is pursued. The conditions selected for modelling comprised a wavelength of 13.5nm, a numerical aperture of 0.75, telecentric dipole illumination, and both vertical (x) and horizontal (y) lines / spaces (L / S) of pitch values from 9 to 15 nm. Design guidelines from the Bottiglieri reference were followed for determining the mask absorber material and thickness. Image contrast plots resulting from the Bottiglieri conditions are shown in FIG. 2. Image contrast values for equal line / space patterns ranging from 9 nm pitch to 15 nm pitch in x and y directions are plotted for s- and p-polarization, along with the average of the two polarization states. A contrast value of 0.60 is considered the minimum value necessary to resolve images into a photoresist layer. Using such a metric, 14 nm pitch features (7 nm lines and spaces, L / S) may be resolvable with unpolarized illumination, where vertical (x) features achieve a contrast of 0.60 and horizontal (y) features achieve 0.59. For pitch values smaller than 14 nm, contrast falls below 0.60 and contrast separation between x and y features widen. Dedicated s-polarized illumination would be necessary to resolve features below 14 nm.
[0012] Image intensity variation also plays a significant role in photomask patterning.Because patterns are distributed across a mask with both x and y orientation, it is desirable that they exhibit common average intensity values in order to be recorded into a photoresist with common dimensional sizing. With a tilted reflective EUV lithography system, reflected and diffracted energy propagation will vary with feature orientation if the mask angular reflectance isn’t sufficiently flat. This is demonstrated in the image intensity plots of the configuration shown in FIG. 3. The plot shows the average image intensity for equal line / space patterns ranging from 9 nm pitch to 15 nm for s- and p-polarization in x and y directions, which experience the corresponding angular ranges cited earlier. Range-to-mean values (i.e. [max-min] / average) are calculated for each pitch case to quantify the relative spread of intensity values for both orientations with s- and with p- polarization. This metric gives an indication of the dimensional variation that could be expected in resist recorded images if no mask biasing or optical proximity correction (OPC) were used. For 14 nm pitch, the range-to-mean is large at 25%, which increases to as large as 35% with smaller pitch values.Applying mask biasing to compensate for the polarization impact on image performance can be difficult as contributions are inherently non-linear.5 50130535.1
[0013] Thus, attaining robust lithography patterning of 14 nm pitch features presents challenges as image contrast is near the minimum usable threshold, process windows become narrow, orientation-dependent effects become pronounced, photoresist exposure dose requirements rise steeply, and mask and polarization effects become limiting. As devices continue to scale, imaging stochastic effects have also become a primary limitation on lithographic performance. At these dimensions, variability arising from photon absorption statistics and energy deposition within the photoresist increasingly impacts line-edge roughness and pattern fidelity, independent of optical resolution. Accordingly, continued scaling requires strategies that address not only imaging resolution but also the fundamental stochastic behavior associated with EUV exposure.
[0014] Unless EUV lithography can reliably address these issues, continued scaling of logic and memory devices will become increasingly constrained by lithographic capability rather than by device design or integration capability. In such a scenario, further advances on the semiconductor roadmap would require additional patterning complexity or alternative patterning approaches, with corresponding impacts on cost, yield, and manufacturing scalability.
[0015] Accordingly, there is a need for improved approaches that mitigate certain difficulties associated with scaling 13.5 nm EUV lithography to larger numerical apertures, including those associated with wavelength and photomask-related challenges. Such approaches should help reduce diffraction-related limitations, increase usable depth of focus, lessen polarization-induced imaging variability, and enable more favorable performance. Meeting these objectives would enable continued advances in lithographic resolution and semiconductor device progression, while mitigating limitations associated with further scaling at 13.5 nm.SUMMARY OF THE INVENTION
[0016] The present technology relates to systems and methods for extreme ultraviolet lithography at a wavelength of approximately 11.3 nm. Lithography at or near 11.3 nm operates in a different physical and materials-interaction regime than conventional 13.5 nm EUV lithography, enabling design considerations that are not available at longer wavelengths. Disclosed embodiments include a number of related aspects directed to photomasks, photomask layers, photomask architectures, optical components, projection lithography methods, related photoresists, materials,6 50130535.1and system-level configurations that enable improved imaging performance and reduced imaging variability relative to longer-wavelength EUV systems.
[0017] For some embodiments, a photomask architecture is disclosed which mitigates mask-related limitations encountered as projection EUV lithography is scaled to larger numerical apertures. According to some embodiments, reflective photomasks operate at a wavelength of approximately 11.3 nm and numerical apertures above about 0.50, above about 0.6, or above about 0.7, while maintaining high reflectance performance over the expanded range of incidence angles present at the mask. As used herein, “approximately 11.3 nm” and “near 11.3 nm” refer to wavelength operation above the beryllium absorption edge of approximately 11.2 nm and accounting for spectral bandwidth requirements of about 2%, or 11.3 nm ± 0.1 nm. Also, as used herein, a high numerical aperture at 11.3 nm refers to numerical apertures at or above approximately 0.50. In general as used herein, “approximately” means + / -2%.
[0018] For some embodiments, reflective photomasks are disclosed which can accommodate increased angular and polarization-dependent effects at the mask when compared to current approaches. Such reflective photomasks support reliable high-resolution pattern formation at wavelengths near 11.3 nm by enabling an interference and index-contrast regime arising from operation at near 11.3 nm that differs from longer wavelength mask multilayer approaches. For such embodiments, compared to current approaches, reduced reflectance variation and reduced polarization-dependent divergence are achieved over the illumination angles incident on the photomask. Disclosed embodiments may provide flexibility in multilayer design through the use of periodic, aperiodic, or hybrid thickness profiles. Multilayer reflective stacks may be formed from material systems suitable for operation near 11.3 nm, including combinations of beryllium with transition metals or other elements, as well as compounds or alloys thereof, to permit enhanced optical performance and manufacturability.
[0019] According to some implementations, the reflective photomask includes multilayer reflective stacks comprising alternating layers of materials such as ruthenium and beryllium, rhodium and beryllium, molybdenum and beryllium, or other combinations suitable for reflection at approximately 11.3 nm. The multilayer reflective stack may include tens to more than sixty layer pairs and may employ periodic, aperiodic, or hybrid thickness profiles. Some disclosed photomask structures are configured to maintain high reflectance over the angular ranges encountered in high7 50130535.1numerical aperture 11.3 nm projection lithography while meeting requirements for reflectance uniformity and polarization control.
[0020] For some embodiments, mask absorber layers are disposed over the multilayer reflective stack and patterned to define lithographic features at a wavelength of approximately 11.3 nm. The absorber material, thickness, and complex refractive index are selected such that absorber-covered regions exhibit reduced reflectance relative to adjacent non-absorber regions, thereby enabling effective pattern formation in projection 11.3 nm lithography. Absorber materials may include one or more elements selected from nickel, silicon, tantalum, tin, tungsten, cobalt, platinum, molybdenum, zinc, rhodium, chromium, aluminum, titanium, or combinations thereof, including nitrides, borides, oxides, carbides, silicides, and chalcogenides.
[0021] According to some embodiments, a peripheral edge barrier is formed on the photomask to encapsulate exposed edge regions of a beryllium-containing multilayer reflective stack. For some embodiments, the barrier is positioned along the mask perimeter, including the beveled edge where the multilayer terminates, and is configured to isolate exposed beryllium-containing layers from environmental interaction. In certain embodiments, the barrier comprises chemically inert, low-permeability materials selected to provide stability under EUV irradiation and hydrogen-rich operating conditions. The barrier may be formed as a conformal encapsulation layer deposited selectively at the mask perimeter thereby sealing the multilayer edge without affecting the active imaging region.
[0022] Some embodiments include a pellicle structure with a membrane formed from carbon-based or silicon-based materials, and in some implementations incorporating beryllium-containing components. The pellicle may have a composite membrane structure including a beryllium-containing layer encapsulated between one or more protective capping layers. The pellicle assembly may further include a mounting frame incorporating pressure-equalization features configured to inhibit the transport of beryllium-containing species during operation. The frame may include non-line-of-sight vent structures, diffusion-limiting passages, or barrier-coated flow paths that permit pressure equilibration while restricting particulate migration, thereby providing both optical functionality and environmental containment. When used in combination with the peripheral edge barrier, the pellicle and edge barrier together provide a multi-layer containment architecture for 11.3 nm photomasks.8 50130535.1
[0023] According to some embodiments, methods are disclosed for fabricating reflective photomasks for operation at approximately 11.3 nm, such methods including forming a multilayer reflective stack on a substrate and depositing and patterning one or more absorber layers using processes compatible with the multilayer structure.
[0024] In some embodiments, methods are provided for performing reflective projection lithography using EUV radiation at a wavelength of approximately 11.3 nm in combination with a reflective photomask and a projection optical system having a numerical aperture of approximately 0.5 or greater, thereby enabling high-resolution pattern transfer while accommodating the expanded angular ranges associated with such systems.
[0025] According to some embodiments, illumination and optical configurations are disclosed that provide controlled polarization of EUV radiation at approximately 11.3 nm. Such configurations may include reflective elements exhibiting polarization-dependent reflectivity, enabling preferential selection of polarization states beneficial for high-resolution imaging. These embodiments may be implemented within the illumination or projection optics to enhance image contrast, reduce polarization-dependent variability, and support high-numerical-aperture operation at approximately 11.3 nm.
[0026] According to some embodiments, photoresist materials and lithographic methods are disclosed which exploit wavelength-dependent energy deposition behavior at approximately 11.3 nm. At this wavelength, photon-matter interactions occur in a regime in which material composition can influence not only average absorbed energy but also the statistical distribution of deposited energy at nanometer length scales, enabling material-level control of energy deposition behavior. Photoresists for 11.3 nm incorporating silicon-containing or aluminum-containing components, as described herein, may exploit these interactions to reduce stochastic variability associated with corresponding 110 eV photon absorption and electron generation that are not available at 13.5 nm. This stochastic enhancement may be achieved without relying solely on increased exposure dose or reduced resist thickness, through selection of resist composition, bonding environment, and material architecture as disclosed, allowing improved pattern fidelity and scalability at advanced feature sizes to be designed for, thereby providing a distinct advantage over photoresist systems configured for longer EUV wavelengths.9 50130535.1BRIEF DESCRIPTION OF DRAWINGS
[0027] FIG. 1 is a reflectance plot for a 13.5nm / 0.75NAmask.
[0028] FIG. 2 are image contrast plots from image simulations with a 13.5 nm projection lithography system at 0.75NA.
[0029] FIG. 3 are image intensity plots from image simulations with a 13.5 nm projection lithography system at 0.75NA.
[0030] FIG. 4 is an image intensity plot for 5 nm resolution at 13.5nm / 0.75NA conditions.
[0031] FIG. 5 is an image intensity plot for 5 nm resolution using 11.3nm / 0.63NA lithography conditions.
[0032] FIG. 6 is an illustration of a 11.3 nm reflective multilayer mask embodiment of the technology.
[0033] FIG. 7 is an illustration of a periodic 11.3 nm mask embodiment of the technology.
[0034] FIG. 8 is a reflectance plot of 5.70nm periodic multilayer embodiments employing various materials.
[0035] FIG. 9 is a reflectance plot of 5.80nm periodic multilayer embodiments employing various materials.
[0036] FIG. 10 is a reflectance plot of 5.80nm and 5.85nm periodic Ru / Be multilayer embodiments.
[0037] FIG. 11 shows design space reflectance case results for d and T combinations for Ru / Be multilayer embodiments.
[0038] FIG. 12 are reflectance plots for several period cases with E=0.40 within the Ru / Be design space.
[0039] FIG. 13 are reflectance plots for several T cases with d=5.80 nm within the Ru / Be design space.
[0040] FIG. 14 are polarized reflectance plots for several period cases with E=0.40 within the Ru / Be design space.
[0041] FIG. 15 are polarized reflectance plots for several T cases with d=5.80 nm within the Ru / Be design space.10 50130535.1
[0042] FIG. 16 shows design space reflectance case results for d and T combinations for Mo / Be multilayer embodiments.
[0043] FIG. 17 is an illustration of an aperiodic 11.3 nm mask embodiment of the technology.
[0044] FIG. 18 is a thickness layer plot of an aperiodic Ru / Be multilayer embodiment.
[0045] FIG. 19 is a reflectance plot of an aperiodic Ru / Be mask embodiment.
[0046] FIG. 20 is a thickness layer plot of an aperiodic Mo / Be multilayer embodiment.
[0047] FIG. 21 is a reflectance plot of an aperiodic Mo / Be mask embodiment.
[0048] FIG. 22 is a thickness layer plot of an aperiodic Ru / Be multilayer embodiment with localized thickness-perturbation.
[0049] FIG. 23 is a reflectance plot of an aperiodic Ru / Be mask embodiment with localized thickness-perturbation.
[0050] FIG. 24 is a thickness layer plot of a block-segmented Ru / Be multilayer embodiment.
[0051] FIG. 25 is a reflectance plot of a block-segmented Ru / Be mask embodiment.
[0052] FIG. 26 is a thickness layer plot of a mixed pair Mo / Be - Ru / Be multilayer embodiment.
[0053] FIG. 27 is a reflectance plot of a mixed pair Mo / Be - Ru / Be mask embodiment.
[0054] FIG. 28 is a thickness layer plot a 30-pair aperiodic Ru / Be multilayer embodiment.
[0055] FIG. 29 is a reflectance plot a 30-pair aperiodic Ru / Be mask embodiment.
[0056] FIG. 30 is a reflectance plot of a periodic Ru / Be mask with B4C barrier layers.
[0057] FIG. 31 is a plot of EFI response through depth for an aperiodic Ru / Be mask embodiment.
[0058] FIG. 32 is a plot reflected phase through angle for an aperiodic Ru / Be mask embodiment.
[0059] FIG. 33 is a diagram of a deposition system for formation of reflective multilayers.
[0060] FIG. 34 is a plot of lithographic simulations of image contrast vs. absorber layer thickness for various 11 ,3nm mask absorbers.
[0061] FIG. 35 is a plot of lithographic simulations of image contrast vs. absorber layer thickness for other various 11.3nm mask absorbers.
[0062] FIG. 36 is a reflectance plot of a WCo 11.3nm mask absorber.11 50130535.1
[0063] FIG. 37 is a reflectance plot of a Sn4Nis 11 3nm mask absorber.
[0064] FIG. 38 is a reflectance plot of a Ta2Si 11.3nm mask absorber.
[0065] FIG. 39 is a diagram of a photomask architecture with a peripheral edge barrier, a pellicle, and various elements.
[0066] FIG. 40 is a patterned photomask structure of the technology.
[0067] FIG. 41 is a diagram of an 11.3nm reflective projection lithography system of the technology.
[0068] FIG. 42 is a diagram of a polarization components for the beam path of an EUV illumination system.
[0069] FIG. 43 is a plot of extinction coefficients of EUV photoresist elements.
[0070] FIG. 44 are representative radial energy deposition profiles at 110 eV.
[0071] FIG. 45 are representative radial energy deposition comparative profiles for 92 eV and 110 eV.
[0072] FIG. 46 is a representative plot of effective blur parameter (o) vs. silicon content for 11.3 nm photoresist materials.
[0073] FIG. 47 is a schematic representing silicon-containing photoresist architecture embodiments.
[0074] FIG. 48 are image contrast plots from image simulations with a 11.3 nm projection lithography system at 0.63NA.
[0075] FIG. 49 are image intensity plots from image simulations with a 11.3 nm projection lithography system at 0.63NA.
[0076] FIG. 50 are image contrast plots from image simulations of periodic multilayers with a 11.3 nm projection lithography system at 0.63NA.
[0077] FIG. 51 are image intensity plots from image simulations of periodic multilayers with a 11.3 nm projection lithography system at 0.63NA.
[0078] FIG. 52 is a configuration for an EUV illumination structure with targeted polarization control.
[0079] FIG. 53 is a representative line-edge roughness (LER) vs. blur parameter plot for 11.3 nm resist exposure.12 50130535.1DETAILED DESCRIPTION OF THE INVENTION
[0080] The following Detailed Description describes representative embodiments of a reflective projection lithography platform for operation at wavelengths near 11.3 nm at high numerical apertures, including reflective multilayer photomasks and associated lithographic systems, methods, and materials. The described embodiments illustrate various aspects of the technology and are not intended to be limiting.
[0081] In projection lithography, the minimum resolution is commonly described by the Rayleigh relationship, in which the smallest resolvable feature scales with the ratio of the exposure wavelength to the NA. For a fixed resolution target, a reduction in the exposure wavelength allows a corresponding reduction in the NA required. By way of example, an imaging condition achieved using EUV radiation at a wavelength of 13.5 nm with an NA of 0.75 corresponds to the same resolution potential when operated at 11.3 nm with an NA of 0.63. Based on the Rayleigh depth-of-focus (DOF) relationship, where DOF ~ + / -0.5X / NA2, the focal depth into a photoresist layer is about 24nm for the 13.5 nm / 0.75NA case, increasing to 28.5nm for the 11.3 nm / 0.63NA case (a 19% improvement). Similarly, an imaging condition achieved using a wavelength of 13.5 nm with a NA of 0.55 (the current state of high-NAEUV lithography) corresponds to the same resolution when operated at 11.3 nm with an NA of 0.46. The corresponding DOF value for the 13.5 nm / 0.55NA case is about 44.6 nm compared to 53.4 nm for the 11.3 nm / 0.46NAcase (a 20% improvement). NA values greater than the current 0.55NAat 13.5 nm, when scaled to an 11.3 nm wavelength, correspond to values greater than approximately 0.46NA, such are examples targeted by the present technology as numerical apertures at or above 0.50NA.
[0082] The influence of polarization on image formation also becomes increasingly significant as NAs are scaled beyond current values. This is illustrated by the comparison between FIG. 4 and FIG. 5, which present aerial-image intensity plots generated under selected imaging conditions. In these particular examples, the three-dimensional effects of mask stacks are not considered as so-called thin-mask modelling is employed. FIG. 4 illustrates an EUV lithography configuration operating at a wavelength of 13.5 nm and an NA of 0.75, used here to simulate intensity images of 5 nm wide lines at a 10 nm pitch under s-polarized illumination, p-polarized illumination, and their composite average (AVG). In this case, the weak-contrast contribution of the p-polarized component at the incidence angles associated with the 13.5 nm / 0.75 NA condition13 50130535.1produces an approximately 46 percent contrast loss of the composite image relative to the s-polarized image. FIG. 5 presents a corresponding simulation for the same 5 nm line-space configuration evaluated at a wavelength of 11.3 nm and an NA of 0.63, which provides a comparable X / NA condition. Under these conditions, the contrast loss via the p-polarized component to the composite image is lower at approximately 32 percent.
[0083] Operation at the shorter 11.3 nm wavelength and lower NA values also relaxes several constraints that become increasingly concerning at high numerical apertures. With the increase in DOF from smaller NA values, there are relaxed focus tolerance and resist thickness constraints. The shorter wavelength / lower NA combinations at 11.3 nm also reduce the range of illumination angles incident on and diffracted by the reflective photomask, easing requirements on multilayer reflectivity bandwidth, phase uniformity, and polarization response across the mask. Collectively, these effects reduce mask-related challenges associated with NA scaling and provide greater flexibility in multilayer and absorber design for high-resolution EUV lithography.
[0084] EUV multilayer reflectivity is driven by the optical behavior of materials at soft-Xray wavelengths. This is due to wavelength coupling in Bragg reflectors and the layer materials' wavelength-dependent optical properties. When designing the 11.3 nm mask multilayers of the present technology, the following aspects are among those that have been considered:Phase delay, interference, and cavity effects which are controlled through layer refractive index and thickness to produce desired resonance, light coupling efficiency, and reflection.Wavelength dependency and bandwidth which are designed for high reflectance at the target wavelength while minimizing off-wavelength spectral broadening.Absorption losses which impact penetration depth and the interference efficiency through the multilayer stack. Materials with low extinction coefficient at the design wavelength are chosen to reduce absorption losses.Phase-shift upon reflection which impacts the interference of reflected light atdifferent angles, contributing to optical wavefront effects (i.e. aberrations) for coated elements and photomasks. These effects increase exponentially with increasing NA.Polarization effects and differences in reflection and reflected phase between s- and p- polarization states, which become larger as incidence angle increases.Interfacial mixing which occurs with atomic diffusion across interfaces, reducingphase contrast and efficiency. Material selection, barrier (or interface) layers, andselect deposition methods are used to enhance performance.14 50130535.1
[0085] The materials that are suitable for use in 11.3 nm mask multilayers have been chosen based on these design aspects. Chief among material properties are the optical indices, specifically refractive index (n) and the extinction coefficient (k). In the EUV and soft-Xray regime, the optical response of a material is described by the complex refractive index conventionally written as n - ik = 1-5+ ip, where 6 represents the dispersive contribution that reduces the real refractive index below unity and P = k (the extinction coefficient), corresponding to the imaginary component of the refractive index. Additional criteria affecting material choice include interface quality, interfacial mixing, thermal stability, and mechanical stability of the elements, compounds, alloys, and compositions employed.
[0086] In the EUV and soft-Xray spectral regions, the optical behavior of materials is highly wavelength-dependent. In this spectral regime, both n and k vary strongly with wavelength due to the proximity of inner-shell absorption edges. As the wavelength is shifted, for example from 13.5 nm toward the shorter wavelength of 11.3 nm, many materials transition closer to or below their relevant absorption edges, which can lead to non-linear changes in absorption (dk / dX) and dispersion (dn / dX). These changes are not monotonic across materials and the optical behavior of a given material at 11.3 nm can differ substantially from its behavior at longer wavelengths such as 13.5 nm. Accordingly, the selection of multilayer constituents, absorber materials, and barrier or interlayer compositions for 11.3 nm operation should account for wavelength-specific optical behavior and cannot rely on optical characteristics established at other wavelengths.
[0087] The optical performance of 11.3 nm multilayers is influenced by the intrinsic properties of the constituent materials and the structural configuration of the multilayer stack. In particular, the refractive index contrast (An) between alternating layers impacts the strength of constructive interference and reflectivity, while the refractive index dispersion (dn / dX) affects the spectral bandwidth by altering the wavelength dependence of reflectance. The absorption index 0 (or optical extinction coefficient k) further influences the efficiency of the multilayer by limiting the penetration depth of radiation. Beyond these material parameters, the layer-thickness distribution plays a key role. The Bragg period or “period” (d) of a multilayer is the combined thickness of one complete bilayer of alternating materials, which is a fixed value for a purely periodic structure. A thickness ratio parameter gamma (T), defined as the ratio of the low-index layer thickness to the15 50130535.1period (tj / d ), modulates the optical path within each period and influences both reflectivity and spectral response.
[0088] In reflective multilayers, high reflectance is achieved through constructive interference of partial reflections at interfaces between layers of differing refractive index and extinction coefficient. The interference condition depends on the optical constants of the materials, the individual layer thicknesses and period, and the angle of incidence. Because radiation can be decomposed into s- and p-polarized components, each polarization state also experiences different reflection coefficients and phase shifts at multilayer interfaces.
[0089] In a reflective multilayer, the reflectance is influenced by the polarization state of the incident electromagnetic field as the boundary conditions governing the electric and magnetic field components differ for s-polarized and p-polarized light at each material interface. For s-polarization, the electric field is oriented parallel to the interface, while for p-polarization the electric field includes a component normal to the interface, resulting in different Fresnel reflection coefficients and phase shifts upon reflection and transmission. As light propagates through the multilayer stack, these polarization-dependent amplitude and phase differences accumulate through multiple reflections. Interference between partial waves reflected at successive interfaces lead to distinct reflectance behavior for polarization states. In the EUV and soft-Xray regime, where the real parts of the refractive indices are close to unity and extinction coefficients are small but nonzero, the shallow index contrast and absorption make the interference conditions highly sensitive to incidence angle and polarization. Consequently, the angles and bandwidths over which constructive interference occurs differ for polarized light, producing a polarization-dependent reflectance response that becomes more pronounced at larger incidence angles.
[0090] At low incidence angles these polarization differences are small, but as the incidence angle increases, the reflectance responses of s- and p-polarized states diverge. As a result, the degree of polarization of reflected radiation increases with angle. Consequently, multilayers intended to operate over large angular ranges require selection of material n and k with tailored layer thickness distributions to maintain constructive interference and acceptable reflectance of both polarization states across the target angular range.
[0091] In the 11.3 nm photomask multilayer combinations disclosed herein, including periodic, aperiodic, and hybrid multilayer stacks, the s- and p-polarized reflectance responses evolve16 50130535.1differently through the stack as a function of angle. These differences manifest as polarizationdependent shifts in reflectance maxima, variations in angular bandwidth, and changes in the relative contribution of different regions of the multilayer stack to the overall reflectance. Some embodiments of the present technology recognize this behavior and incorporates polarization response as a design consideration in the multilayer configuration, such that the s- and p- polarized reflectance lead to acceptable performance criterion at larger incidence angles. This includes target values throughout the angular range necessary for and imposed by the operational NA value.
[0092] Accordingly, specifying performance metrics in terms of reflectance derived from the s- and p-polarized components provides physically meaningful and practically relevant metrics for mask multilayer designs intended for use at 11.3 nm over a range of non-normal incidence angles. Establishing minimum averaged reflectance thresholds over an angular range reflects a design constraint that inherently accounts for polarization effects, material optical properties, and multilayer interference behavior, and serves as an appropriate performance criterion for multilayer reflective coatings configured for extended angular operation.
[0093] For example, an averaged reflectance (i.e., the average of the s-polarized and p-polarized reflectance components) greater than about 60% is a desirable target for an 11.3-nm reflective multilayer mask operating at numerical apertures of 0.50 and higher, over incidence angles up to approximately ±10° (i.e., -10° to +10° relative to the mask surface normal). In addition, a target for reflectance variation is desirably below about 6% and the polarization divergence over the same angular range is desirably below about 5%, and may be below about 4% for some highest resolution instances. Such reflectance performance criteria provides sufficient optical throughput, image contrast, and x / y pattern fidelity for both polarization states.
[0094] Furthermore, unlike multilayer coatings developed to primarily engineer reflectance at specific angles or over a narrow angular range, large-angle 11.3nm mask multilayers preferably satisfy requirements of averaged reflectance, angular flatness, and polarization balance over a wide angular distribution determined by system numerical aperture, reduction ratio, and illumination angles. These constraints arise from the interaction between the multilayer stack and the patterned absorber layer, and impose design conditions that are not addressed by multilayer designs configured for other attributes.17 50130535.1
[0095] As described earlier, the design of 11.3 nm reflective mask multilayers comprises the use of materials that provide desirable optical performance. Optical properties of various materials at 11.3 nm are shown on Table 1. Unless otherwise noted, optical constants, extinction coefficients, and absorption coefficients reported herein for candidate multilayer, absorber, and other materials at approximately 11.3 nm are based on composition-derived values obtained from the Center for X-Ray Optics (CXRO) tabulated EUV optical-constant database and associated modeling utilities.Skin5effdepthAn w / Be w / BeMaterial n k w / Be (nm) (nm)Be 1.0089 0.0012B 0.9786 0.0023 0.0295 95 545Si 1.0058 0.0144 0.0023Y 0.9835 0.0020 0.0246 115 600Nb 0.9633 0.0040 0.0448 63 360Mo 0.9527 0.0046 0.0562 51 309Ru 0.9324 0.0067 0.0765 37 231Rh 0.9236 0.0089 0.0845 33 181Pd 0.9198 0.0135 0.0883 32 124Zr 0.9733 0.0029 0.0348 81 461SiO20.9865 0.0123 0.0216Table 1. Optical and multilayer properties of various materials at 11.3 nm wavelength.
[0096] These materials represent light elements, refractory metals, and noble metals that may be employed as individual layers, in compounds or alloys, or in combination. When deposited as nanometer-scale films, these materials exhibit distinct microstructural, mechanical, and chemical characteristics that influence multilayer performance and stability. Beryllium (Be) films typically grow with fine-grained hexagonal close-packed structure at low substrate temperatures. Boron (B) is commonly deposited as amorphous or nanocrystalline films with high hardness and low diffusivity. Yttrium (Y) and zirconium (Zr) are reactive metals that tend to form hexagonal close-packed (hep) nanocrystalline films. Molybdenum (Mo) films typically grow in a body-centered cubic structure with grain structure depending on sputter conditions. The more noble metals ruthenium (Ru),18 50130535.1rhodium (Rh), and palladium (Pd) generally form dense, nanocrystalline films (Ru hep; Rh and Pd face-centered cubic).
[0097] When layered together, these materials can form a variety of intermetallic compounds and solid solutions, depending on composition and thermal conditions, and may be considered in the choice and design of reflective multilayer coatings. Relevant compound classes include beryllides formed between Be and transition metals (e.g., Be-Mo, Be-Ru, Be-Rh systems). Such reactions may be exploited intentionally to tailor interface properties or suppressed through process control, barrier layers, or compositional grading.
[0098] Barrier layers may be used in conjunction with beryllium-containing 11.3 nm multilayers to reduce atomic interdiffusion while preserving reflectance performance. Effective barrier materials generally exhibit low mutual diffusivity with beryllium, limited chemical reactivity at deposition and operating temperatures, and stability under prolonged 11.3 nm irradiation.Materials such as boron, carbon, and silicon form dense amorphous or nanocrystalline fdms that can suppress atomic transport across interfaces while introducing minimal optical degradation when applied at sub-nanometer to nanometer level thicknesses. Compound barrier layers, including nitrides, carbides, borides, and silicides, can provide enhanced chemical bonding strength and reduced diffusion pathways, further limiting beryllium migration into adjacent reflective layers. Refractory metals and their compounds may also function as diffusion barriers when deposited as ultra-thin interlayers. The thickness of a diffusion barrier layer is kept small to balance diffusion suppression against optical impact.
[0099] Regarding the optical behavior of mask multilayer reflective coatings, the concept of a virtual reflection plane can be used to optically describe the region of an effective location in a multilayer stack if it were replaced by a single surface. This position can serve as a depth location for designing and evaluating optical effects. For 11.3 nm photomask applications, such a plane can estimate the location of the object plane.
[0100] In a reflective mask multilayer, a forward propagating field decays exponentially through a multilayer stack, with decreasing amplitude governed by refractive index contrast and material absorption. For non-absorbing materials with An small relative to n, an indication of the depth over which most interference occurs can be represented by the penetration depth, 3eff, which can be estimated as a first-order metric as:19 50130535.1
[0101] The concept of skin depth provides a first-order metric to assess loss from absorption alone as light propagates through an optical material. This depth is defined as the distance over which an initial intensity decreases by absorption to approximately 1 / e (about 37%) of its original value, which is inversely related to a material's extinction coefficient (k). This does not account for index-driven interference within the material, as occurs in a multilayer stack. For a multilayer reflective mask coating with an average extinction coefficient kavg(i.e. the average of k values for materials involved), the skin depth is given by:&Beyond the skin depth of a mask multilayer, propagation is attenuated where further increases in layer count may yield diminishing returns in reflectivity enhancement.
[0102] To account for the complex refractive index of a multilayer coating, the penetration depth can be determined by modelling the electric field intensity (EFl) through a particular multilayer stack. Propagation is determined based on the phase interference between layers, the superposition of reflected fields, and the exponential loss from absorption. More comprehensive than first-order metrics of penetration depth and skin depth which are based on material properties alone, EFI mapping can provide for the determination of more meaningful propagation distances. As the EFI decays progressively through a mask multilayer stack, the EFI 1 / e depth represents the point at which the field intensity drops to approximately 37% of its initial value. This is the depth at which layers have effectively resulted in most of the multilayer’s reflective interference, effectively representing a virtual reflecting plane.MATERIAL PROPERTY DIFFERENCES AT 13.5NM AND 11.3NM
[0103] Although multilayer optics for EUV lithography have been extensively studied at 13.5 nm, most notably using Mo / Si material systems, the optical behavior of candidate multilayer20 50130535.1materials at 11.3 nm differs in ways that fundamentally alter multilayer design considerations. As a result, design rules, material preferences, and performance tradeoffs established for 13.5 nm cannot be extrapolated or scaled for use with 11.3 nm reflective mask multilayers.
[0104] At 13.5 nm, silicon exhibits relatively low absorption and favorable refractive index contrast when paired with molybdenum, leading to the widespread adoption of Mo / Si multilayers as the standard for 13.5nm EUV mirrors and photomasks. In this regime, Mo serves as a high-index, moderately absorbing layer, while Si provides low absorption and good optical contrast. As a consequence, Mo / Si multilayers have been configured primarily for peak reflectance and spectral bandwidth near normal incidence, and performance trends have been utilized to extend reflectance behavior over a moderate range of angles.
[0105] At 11.3 nm, however, the material landscape changes substantially. Silicon exhibits significantly higher absorption, and its optical constants differ markedly from those at 13.5 nm. Beryllium, rather than silicon, becomes the low absorbing, high-index material of interest, while candidates of relatively low-index materials such as molybdenum, ruthenium, rhodium, palladium, and niobium exhibit substantially different extinction coefficients and dispersion behavior than they do at longer wavelengths. Comparative at-wavelength material data changes significantly between 13.5 nm and 11.3 nm to place these systems in a fundamentally different interference regime.
[0106] As a result, material choices used at 13.5 nm behave very differently at 11.3 nm. For example, although molybdenum is the dominant low-index material at 13.5 nm, it is not uniquely advantageous at 11.3 nm. At the shorter wavelength, alternative materials such as ruthenium, rhodium, palladium, and niobium exhibit optical constants that can produce favorable combinations of angular response, absorption depth, and polarization behavior when paired with beryllium. For example, the behavior of Ru with Be at 11.3 nm cannot be inferred from the behavior of Mo with Si at 13.5 nm. Differences in absorption coefficient, dispersion, and penetration depth lead to distinct interference buildup, altered effective optical thickness, and different sensitivity to such things as layer count, thickness ratio, and periodicity.
[0107] Table 2 shows examples of EUV optical properties and metrics for Mo and Ru combined with Si and Be at 13.5nm and at 11.3nm, comparing wavelength-specific index contrast, dispersion, absorption, and skin depth between the combinations.21 50130535.1Table 2. Index contrast, dispersion, absorption and skin depth for Mo / Si, Ru / Si, Mo / Be, and Ru / Be at 13.5nm and 11.3nm.
[0108] The optical behavior of multilayer systems formed from Be, Ru, and Mo at an exposure wavelength of 11.3 nm differs fundamentally from that of conventional Si-based multilayers operating at 13.5 nm, owing to substantial changes in both refractive index contrast and absorption characteristics across this wavelength shift. As summarized in Table 2, the transition from 13.5 nm to 11.3 nm alters not only the values of the real and imaginary components of the refractive index, but also their relative balance and dispersion behavior. For example, while Mo / Si and Ru / Si multilayers at 13.5 nm exhibit relatively high index contrast (An = 0.077 and 0.115, respectively), they are also characterized by comparatively large absorption coefficients, particularly for Ru, resulting in reduced optical skin depths, on the order of 95 nm for Ru / Si. In contrast, at 11.3 nm, Be-based multilayers such as Mo / Be and Ru / Be exhibit lower index contrast values, yet benefit from substantially reduced absorption in the low-Z beryllium layers, leading to markedly different skin depths and field penetration behavior. Notably, Ru / Be at 11.3 nm combines a moderate index contrast (An ~ 0.0765) with a skin depth over twice that of Ru / Si at 13.5 nm, fundamentally changing the number of bilayers that contribute coherently to reflection and the way interference builds through the stack.
[0109] These material-dependent differences are further compounded by changes in dispersion (dn / dX) and absorption (k) near the relevant absorption edges, which alter phase accumulation, polarization response, and angular reflectance behavior in ways that are not scalable between wavelengths. As a result, multilayer designs configured for 13.5 nm, whether periodic, depth-graded, or otherwise modified, do not provide reliable insight into the angular, polarization, or22 50130535.1interference behavior of analogous mask material systems at 11.3 nm. In particular, the interplay between reduced absorption in Be-based systems and the modified refractive index contrast at 11.3 nm places these multilayers into different interference regimes than their 13.5 nm counterparts, affecting angular flatness, polarization balance, and reflectance roll-off at larger incidence angles. Accordingly, the optical response of Be-based mask multilayers at 11.3 nm cannot be inferred from previous EUV multilayers or design practices developed for Si-based multilayers at 13.5 nm and instead requires an independent analysis specific to the shorter wavelength regime.MASK MULTILAYER INTEGRATION
[0110] Achieving the necessary precision in multilayer coatings requires angstrom-level control over layer thickness and interface quality, demanding deposition techniques and metrology systems capable of maintaining such tolerances. Mask multilayer coatings may be deposited using highly controlled physical vapor deposition (PVD) techniques to produce alternating nanoscale layers on substrates of synthetic quartz glass, fused silica, or low thermal expansion material (LTEM) glasses such as SiCL-TiCL based compositions. Among the deposition methods employed may be ion beam deposition (IBD) and magnetron sputtering. Ion beam deposition utilizes a directed ion beam to sputter atoms from elemental targets. Magnetron sputtering, including bias-assisted variants, is a method that deposits constituent layers by bombarding targets with plasma ions in a vacuum chamber. Adjusting process variables such as argon gas pressure, ion acceleration voltage, and substrate bias allows for the control of interface quality, and stress.
[0111] Multilayer stacks may be covered with a thin capping layer to protect the underlying layers from oxidation, contamination, and processing damage, with capping layer materials including silicon, ruthenium, boron carbide, carbon, or combinations thereof.
[0112] An 11.3 nm mask absorber layer is deposited after completion of the reflective multilayer stack and capping layer (if employed), using a separate physical vapor deposition process for absorber materials and the thickness desired. Absorber films may be deposited by magnetron sputtering, including DC or RF sputtering, from targets containing constituent materials employed in the layer.
[0113] One embodiment of a 11.3 nm reflective multilayer mask according to the present technology is depicted in Figure 6. Mask 100 comprises a substrate 101 a multilayer stack 102 and23 50130535.1an absorber layer 103. The multilayer stack 102 comprises a plurality of layer pairs 113, with each layer pair having alternating layers 105, 106. The alternating layers 105, 106 repeat with the same or different layer thicknesses ti and t2, t3 and U, and so forth wherein total number layer pairs 113 that make up the stack 102 may be between 20 and 70. The mask 100 may include a capping layer 104. When used in a 11.3 nm projection lithography system, radiation of a wavelength 107 illuminates the mask 100 at angles 108 and 110 associated with the numerical aperture, mask reduction ratio, and illumination angles of the projection system. The multilayer stack 102 with optional capping layer 104 reflect an amount of radiation of 11.3 nm wavelength 107 at angles 109 and 112. The absorber 103 reflects an amount of radiation of 11.3 nm wavelength 107 at angles 111. In some embodiments, the multilayer stack may comprise a plurality of layer multiplets. For example, the multilayer stack may comprise a plurality of layer triplets having a third material (not shown) or a plurality of layer quadruplets having a third material and a fourth material (not shown), or a plurality of layer multiplets comprising more than four layers.MULTILAYER STACK EXAMPLES FOR 11.3 NM
[0114] Some embodiments of the present technology are reflective masks for an EUV wavelength near 11.3 nm using multilayer combinations comprising beryllium (Be) layers.Furthermore, mask multilayer structures according to some embodiments are employed to achieve high reflectance through the angular range necessary for 11.3 nm lithography at numerical apertures at and above about 0.50. Additionally, many mask embodiments generally comprise mask blank substrates of synthetic quartz glass, fused silica, or low thermal expansion material (LTEM) glasses and the like upon which multilayer structures are coated. Furthermore, some mask embodiments comprise absorber layers over the mask multilayers which, when exposed and etched, provide the mask pattern for photoresist exposure.
[0115] As an example, an 11.3 nm EUV projection lithography system operating at a numerical aperture of 0.50 illuminates a reflective mask at angles that range between -7.2° and +7.2° in the 4X x-direction and from -2.4° to +9.6° in the 8X y-direction of anamorphic projection optics, assuming a 6° chief ray angle (CRAO) of the tilted reflective optical train. A multilayer reflective mask stack is therefore targeted such that an averaged reflectance (i.e. the average of s- polarized and p- polarized reflectance) remains greater than about 60% for 11.3nm radiation at incident at24 50130535.1angles up to at least ±9.6° relative to a surface normal of the photomask, which covers the angular ranges in both the x- and y-directions. As another example, an 11.3 nm EUV projection lithography system operating at a numerical aperture of 0.63 illuminates a reflective mask at angles that range between -9.1° and +9.1° in the 4X x-direction and from +1.5° to +10.5° in the 8X y-direction, assuming a 6° CRAO of the tilted reflective optical train. A multilayer reflective mask stack is therefore targeted such that an averaged polarized reflectance remains greater than about 60% at incident angles up to at least ±10.5° relative to a surface normal of the photomask, which covers the angular ranges in both the x- and y-directions. Accommodations for other system design configurations would lead to adjustments to these angular bandwidth ranges.PERIODIC MULTILAYER PHOTOMASK EMBODIMENTS
[0116] An 11.3 nm periodic multilayer mask embodiment is depicted in FIG. 7. In this embodiment, the mask 200 comprises periodic layer pairs 215 comprised of a first layer 205 and a second layer 206, which are layered with a fixed period defined as the sum of thicknesses of layers 205 and 206 and coated to form multilayer stack 202 on substrate 201. The total number of layer pairs 215 that mask up the stack 202 may be between 20 and 70, or 25 and 60, or 30 and 50, or 30 and 55, though numbers outside of this range still fall within the scope of the embodiment. For some embodiments, the number of layer pairs is 40. The mask 200 also includes an absorber layer 203 and may include a capping layer 204. When used in a projection lithography system, radiation of a 11.3 nm wavelength 207 illuminates at angles 208 and 210 for example between -7.2° and +9.6° for a numerical aperture of 0.50 and a CRAO of 6°, or between -9.1° and +10.5° for a numerical aperture of 0.63 and a CRAO of 6°, or other angles according to operational numerical apertures at or above 0.50NAand corresponding CRAO or other design values. The period of the periodic multilayers 215 is chosen based on thicknesses ti and t2 of layers 205 and 206, which have refractive index values m and n2 and extinction coefficient values ki and k2. According to some embodiments, the periodic multilayers 215 may additionally include a third material to form periodic layer triplets (not shown) and, optionally, a fourth material, to form periodic layer quadruplets (not shown) incorporated within the multilayer stack 202. The third, fourth, and additional multiplet materials may be provided as additional layers or interlayers disposed between the first layer 205 and the second layer 206, as thin interfacial or diffusion barrier layers between selected layer pairs, as layers inserted periodically or25 50130535.1non-periodically within the multilayer stack 202, or as modifications to one or both of the first and second layers. The first layer 205, the second layer 206, and any third or fourth materials may independently comprise one or more materials selected from beryllium (Be), boron (B), carbon (C), yttrium (Y), molybdenum (Mo), ruthenium (Ru), rhodium (Rh), palladium (Pd), silicon (Si), and zirconium (Zr) in elemental form or compounds thereof.
[0117] According to some embodiments, the periodic multilayer mask of periodic layerpairs, layer-triplets, layer-quadruplets, or layer-multiplets may have between 10 and 70, between 20 and 60, between 30 and 50, or between 35 and 45 layer-pairs, layer-triplets, layer quadruplets, or layer multiplets. For some such embodiments, the number of layer-pairs, layer-triplets, layer quadruplets, or layer multiplets may be 40.
[0118] The first layer material of a layer-pair may be beryllium (Be) and the second layer may be ruthenium (Ru), designated as Ru / Be. At this wavelength, the skin depth of Ru / Be is 23 Inm and the penetration depth is 37nm. Alternatively, the first layer may be ruthenium and the second layer may be beryllium. Alternatively, one layer material may be beryllium (Be) and the other layer may be molybdenum (Mo), designated as Mo / Be. At this wavelength, the skin depth of Mo / Be is 309 nm and the penetration depth is 5 Inm. Alternatively, one layer material may be beryllium (Be) and the other layer may be selected from rhodium (designated as Rh / Be), palladium (designated Pd / Be) or niobium (designated as Nb / Be).
[0119] FIG. 8 shows the averaged reflectance (the average of s-polarized and p-polarized reflectance) resulting from 11.3 nm illumination of examples of combinations of Mo / Be, Rh / Be, Ru / Be, Pd / Be, and Nb / Be for 40-pair periodic multilayers with periods (d) of 5.70 nm and T values of 0.41. The mask multilayer structures were designed using transfer-matrix-method (TMM) thin-film analysis software (TFCalc by Software Spectra, Inc) to compute angle-dependent reflectance characteristics. Reflectance values are plotted from 0° (i.e. normal incidence) up to 20°. For planar layers of mask multilayers, the reflectance versus incidence angle is symmetric about normal incidence as interference conditions depend on the absolute angle from the surface normal. Thus, angles plotted represent a distribution from -20° to +20° about the normal to the surface of the photomask. At normal incidence, the averaged reflectance for Mo / Be, Rh / Be, Ru / Be, Pd / Be, and Nb / Be are high at 73.1%, 76.3%, 77.5%, 71.0%, and 66.0%, respectively. Nb exhibits lower reflectance than others at normal incidence but higher reflectance at 10°. Ru exhibits the highest26 50130535.1reflectance at normal incidence with moderate reflectance at 10°. However, for an angular range of + / -1O0, the averaged reflectance values fall to 56.6%, 34.5%, 43.2%, 29.5%, and 51.9% respectively, which do not meet a through-angle 60% target criteria. It is desirable to extend angular performance to meet targeted reflectance criteria through configurations of the multilayer structure and materials.
[0120] When the multilayer period design is changed by small amounts, interference is observed to shift the reflectance maxima based on the particular properties of the materials involved. For example, as 40 layer-pair periodic multilayers of Mo / Be, Rh / Be, Ru / Be, Pd / Be, and Nb / Be are implemented with a period shift from 5.70 nm to 5.80 nm (at a T value of 0.41), as shown in FIG. 9, reflectance shifts toward angles with peak reflectance of each occurring between approximately 11° and 12°. This behavior occurs together with a reduction in normal-incidence reflectance for the evaluated material combinations, with the effect being especially pronounced for Nb / Be and Mo / Be multilayers. The distinctive through-angle reflectance behavior exhibited by each material combination is unexpected and would not be predicted from the reflectance characteristics observed at lower period values, which show substantially less variation among the material systems. Nor would such angular behavior have been obvious from the 11.3 nm optical properties of the constituent materials alone. Small changes in the design period were found to place multilayer structures in different n / k-driven interference regimes, which result in unexpected angular and polarization-dependent reflectance characteristics.
[0121] In particular, as seen in FIG. 9, as the period is increased to 5.80 nm for Nb / Be and Mo / Be multilayers, the reflection condition is observed to shift away from normal incidence and reflectance at small angles decreases rapidly, leading to a pronounced drop near normal incidence and a stronger dependence on angle. The averaged angular reflectance variation (i.e. the variation in the average of the s-polarized and p-polarized reflectance over an angular range, AR) from 0° to 10° for Nb / Be is large at about 28% and for Mo / Be it is about 13% (both above the 6% target criteria). Ru / Be multilayers exhibit flat reflectance response from normal incidence to approximately 12° as the reflectance maximum shifts to higher angles. The AR from 0° to 10° is low at about 4% and remains about 4% to 12°. In Rh / Be multilayers, the angular response also remains comparatively flat over the same angular range. The AR from 0° to 10° is about 3.1% and about 4.6% from 0° to 12°. In Pd / Be multilayers, a broadened but more strongly damped and overall reduced angular response results, producing less flatness than Ru / Be or Rh / Be despite the larger index contrast of Pd / Be at27 50130535.111.3 nm. The AR from 0° to 10° is low at about 4.2%. At incidence angles slightly beyond approximately 12°, the Mo / Be multilayer exhibits higher reflectance than the Ru / Be, Rh / Be, and Pd / Be combinations at a 5.80 nm period. In contrast, multilayers incorporating Ru, Rh, or Pd show increased attenuation at higher angles, which reduces reflectance once the angular response moves beyond the principal reflectance plateau.
[0122] As incidence angle is increased further, reflectance behavior varies among candidates, showing lower-absorbing Mo / Be and Nb / Be material combinations to exhibit higher reflectance than others above about 14° (albeit below 60%). Beyond approximately 14°, all material combinations show a decrease in reflectance as the interference conditions are apparently no longer satisfied and absorption, reduced interface coupling, and polarization effects appear to collectively suppress reflection. At these higher angles, these periodic multilayers indicate that they no longer sustain sufficient phase-matched interference regardless of material choice, resulting in a universal reduction in reflectance.
[0123] Table 3 shows the average of s-polarized and p-polarized reflectance (Ravg) for the five example multilayers of FIG. 9 at 0°, 10°, 11°, and 12° together with angular reflectance variation (AR) from 0-10° and 0-11° and maximum difference between s-polarized reflectance and p-polarized reflectance (Rs-RP) from 0-10° and 0-11°. The Rh / Be, Ru / Be, and Pd / Be examples maintain RaVgabove 60% for angles up to 10° with AR below 6% and Rs-Rpless than 5%, indeed below 4%. All three also maintain reflectance values above 60% up to 12°. Mo / Be and Nb / Be periodic multilayers at 5.80nm do not meet these performance targets.Table 3. The percent-averaged reflectance, angular variation, and polarized reflectance differences of 5.80nm periodic multilayer examples.28 50130535.1
[0124] As the incidence angle increases beyond a few degrees, polarization-dependent differences accumulate across many bilayers, producing an observable divergence in reflectance magnitude between the two polarization states. A further example of this behavior is illustrated in FIG. 10, which plots polarized reflectance versus incidence angle for 40 layer-pair Ru / Be multilayers designed with periods of 5.80 nm and 5.85 nm. For the 5.80nm-period stack case, the normal-incidence reflectance is approximately 72.5%, whereas for the 5.85 nm period stack it is about 65.7%, indicating that the interference peaks have shifted toward larger angles. The 5.80 nm period case reflectance remains flatter than the 5.85 nm case, where both designs exhibit an angular region near approximately 10.8° where the s- and p-polarized reflectance values are closely matched, with s-polarized reflectance near 77% and p-polarized reflectance near 74%. For the 5.85nm case, between 0° and 10°, the maximum difference between s- and p-polarized reflectance (Rs-Rp) is low at 2.8% but the variation in averaged reflectance (AR) is above a 6% target criteria at 9.6%, compared to 4.10% for the 5.80nm case. Beyond about 12.5°, the 5.80 nm-period case shows a reduction in both s- and p-polarized reflectance, consistent with the multilayer moving outside the angular range over which strong reflection is sustained. In contrast, the 5.85 nm-period design continues to build constructive interference to larger angles, with p-polarized reflectance increasing to approximately 74.7% near 11.7° and s-polarized reflectance reaching approximately 77.6% near 12.4°. At still larger angles, all polarization reflectance values decrease, with reflectance falling sharply beyond roughly 14.5°.
[0125] To determine the structural configurations of periodic multilayers meeting photomask performance targets, a systematic design-space study is presented. The materials employed in 11.3 nm photomasks designed to operate at large incidence angles lead to periodic multilayer structures with periods and thickness ratios that fall within a design space with a relatively narrow range.
[0126] Structural configurations of some multilayer embodiments that are disclosed herein were analyzed to determine the specific operational range of period (d) and thickness ratio (T) values to meet specific performance criteria. A design space for photomask multilayer stacks was evaluated through a systematic series of simulations across a two-dimensional parameter space to identify functional regimes that achieve target angular reflectance performance. Specifically, d and T values are varied across ranges from 5.70 nm to 5.92 nm and from 0.24 to 0.70 respectively. The resulting polarized angular reflectance performance was evaluated against a set of photomask29 50130535.1threshold criteria including an average reflectance of at least 60% over a + / - 10° angular range with an averaged reflectance variation (AR) less than 6% and a polarization-dependent reflectance difference (RS-RP) less than 5%, less than 4%, and less than 3% over the angular range. Such design criteria were chosen to achieve the desired image performance over the angular ranges associated with 11.3nm lithography at NA values of 0.50 and grater.
[0127] Notably, the design-space exploration disclosed was not directed toward maximizing peak reflectance, shaping a target angular reflectance profile, preferentially enhancing or suppressing of a particular polarization state, or meeting other material performance criteria. Instead, candidate multilayer configurations were evaluated against the set of simultaneous threshold conditions targeting high averaged reflectance, low angular variation, and limited polarization divergence across the full angular range.
[0128] FIG. 11 shows the operational design space reflectance results for d and T combinations in the specified ranges for Ru / Be multilayer configurations having 40 layer-pairs. For each combination, polarized reflectance plots were evaluated through an angular range exceeding the targeted range. Multilayers that did not result in a greater than 60% reflectance over an angular range from -10° to +10° were eliminated. Those that remained were evaluated for angular reflectance variation (AR) and the difference between s-polarized and p-polarized reflectance (Rs-RP) across the -10° to +10° range. Corresponding values are tabulated in FIG. 11. Table blocks entries for Ru / Be multilayers that correspond to the combined conditions of AR less than 6% and Rs-RPless than 3% are coded with a black background and white numbering. Table blocks for multilayers that correspond to the conditions of AR less than 6% and RS-RPmore than 3% but less than 4% are coded with black numbering and solid black outlining. Table blocks for multilayers that correspond to meeting the conditions of AR less than 6% and RS-RPless than 5% but more than 4% are coded with black numbering and dashed black outlining. In addition to the values and ranges for period and T described below, each block shown in FIG. 11 indicates the period and T of an embodiment of the present technology.
[0129] For example, in some embodiments, the multilayer stack has a period of 5.76 nm, and T in the range from 0.30 to 0.34. In some embodiments, the multilayer stack has a period of 5.78 nm, and T in the range from 0.34 to 0.42. In some embodiments, the multilayer stack has a period of 5.80 nm, and T in the range from 0.38 to 0.46. In some embodiments, the multilayer stack has a30 50130535.1period of 5.82 nm, and T in the range from 0.42 to 0.50. In some embodiments, the multilayer stack has a period of 5.84 nm, and T in the range from 0.46 to 0.52. In some embodiments, the multilayer stack has a period of 5.86 nm, and T in the range from 0.50 to 0.56. In some embodiments, the multilayer stack has a period of 5.88 nm, and T in the range from 0.54 to 0.62. In some embodiments, the multilayer stack has a period of 5.90 nm, and T in the range from 0.60 to 0.64. In some embodiments, the multilayer stack has a period of 5.92 nm, and T in the range from 0.64 to 0.68. Based on Fig. 11, some embodiments have periods ranging from 5.76 to 5.92, and T values ranging from 0.30 to 0.68.
[0130] In some embodiments, the multilayer stack has a period of 5.72 and T in the range from 0.24 to 0.26. In some embodiments, the multilayer stack has a period of 5.72 and T in the range from 0.24 to 0.28. In some embodiments, the multilayer stack has a period of 5.74 and T in the range from 0.26 to 0.34. In some embodiments, the multilayer stack has a period of 5.74 and T in the range from 0.26 to 0.36. In some embodiments, the multilayer stack has a period of 5.76 and r in the range from 0.30 to 0.40. In some embodiments, the multilayer stack has a period of 5.76 and T in the range from 0.30 to 0.42. In some embodiments, the multilayer stack has a period of 5.78 and T in the range from 0.34 to 0.46. In some embodiments, the multilayer stack has a period of 5.78 and T in the range from 0.34 to 0.48. In some embodiments, the multilayer stack has a period of 5.80 and T in the range from 0.38 to 0.50. In some embodiments, the multilayer stack has a period of 5.80 and T in the range from 0.38 to 0.52. In some embodiments, the multilayer stack has a period of 5.82 and T in the range from 0.42 to 0.54. In some embodiments, the multilayer stack has a period of 5.84 and T in the range from 0.46 to 0.58. In some embodiments, the multilayer stack has a period of 5.84 and T in the range from 0.46 to 0.60. In some embodiments, the multilayer stack has a period of 5.86 and T in the range from 0.50 to 0.58. In some embodiments, the multilayer stack has a period of 5.86 and T in the range from 0.50 to 0.62. In some embodiments, the multilayer stack has a period of 5.88 and T in the range from 0.54 to 0.64. In some embodiments, the multilayer stack has a period of 5.90 and T in the range from 0.60 to 0.66. In some embodiments, the multilayer stack has a period of 5.92 and T in the range from 0.64 to 0.68.
[0131] In some embodiments, the multilayer stack has a period of 5.70 and T in the range from 0.24 to 0.34. In some embodiments, the multilayer stack has a period of 5.72 and T in the 31 50130535.1range from 0.30 to 0.40. In some embodiments, the multilayer stack has a period of 5.74 and T of 0.24. In some embodiments, the multilayer stack has a period of 5.74 and T in the range from 0.38 to 0.46. In some embodiments, the multilayer stack has a period of 5.76 and T in the range from 0.30 to 0.40. In some embodiments, the multilayer stack has a period of 5.76 and T in the range from 0.26 to 0.28. In some embodiments, the multilayer stack has a period of 5.76 and T in the range from 0.44 to 0.50. In some embodiments, the multilayer stack has a period of 5.78 and T in the range from 0.30 to 0.32. In some embodiments, the multilayer stack has a period of 5.78 and T in the range from 0.50 to 0.52. In some embodiments, the multilayer stack has a period of 5.80 and T in the range from 0.30 to 0.36. In some embodiments, the multilayer stack has a period of 5.80 and T in the range from 0.54 to 0.56. In some embodiments, the multilayer stack has a period of 5.82 and T in the range from 0.32 to 0.40. In some embodiments, the multilayer stack has a period of 5.82 and T in the range from 0.56 to 0.58. In some embodiments, the multilayer stack has a period of 5.84 and T in the range from 0.36 to 0.44. In some embodiments, the multilayer stack has a period of 5.86 and T in the range from 0.40 to 0.48. In some embodiments, the multilayer stack has a period of 5.86 and T of 0.64. In some embodiments, the multilayer stack has a period of 5.88 and T in the range from 0.44 to 0.52. In some embodiments, the multilayer stack has a period of 5.90 and T in the range from 0.48 to 0.58. In some embodiments, the multilayer stack has a period of 5.92 and T in the range from 0.54 to 0.62.
[0132] To attain reflectance performance over a range including the target reflectance performance, for periods of 5.70 nm to 5.92 nm, T values may fall between 0.24 and 0.68. To attain reflectance performance at or near the target reflectance performance, for periods of 5.74 nm to 5.84 nm, T values may fall between 0.26 and 0.60. To attain reflectance performance at or near the target reflectance performance, for periods of 5.74 nm to 5.84 nm, T values may fall between 0.26 and 0.58. To attain reflectance performance at or near the target reflectance performance, for periods of 5.74 nm to 5.84 nm, T values may fall between 0.30 and 0.52. To attain reflectance performance at or near the target reflectance performance, for periods of 5.76 nm to 5.90 nm, T values may fall between 0.30 and 0.66. To attain reflectance performance at or near the target reflectance performance, for periods of 5.76 nm to 5.90 nm, T values may fall between 0.30 and 0.64. To attain32 50130535.1reflectance performance at or near the target reflectance performance, for periods of 5.80 nm to 5.92 nm, r values may fall between 0.38 and 0.68.
[0133] For a 3% polarization limit, there are several parameter combinations within a relatively thin asymmetric window encompassing select periods between 5.76 nm and 5.90 nm combined with select T values above 0.30. Increasing these limits to 4% and to 5% widens the parameter combination candidates. Some local trends can be observed from the data. There is an asymmetric region toward a center of the parameter space where combinations meet the aggregated specifications with wider parameter ranges than in other areas. Within central regions of the parameter space, configurations having relatively higher thickness ratios tend to exhibit reduced angular reflectance variation over the angular range of interest, while configurations having relatively larger bilayer periods tend to exhibit reduced differences between s-polarized and p-polarized reflectance.
[0134] At the same time, the mapping further shows that these tendencies are not uniform across the design space and do not evolve in a monotonic or systematic manner. Small changes in either bilayer period or thickness ratio alter the relative phase and absorption balance in ways that impact angular flatness and polarization behavior. As a result, while localized trends can be identified within portions of the parameter space, the performance of a given multilayer configuration with respect to the combined criteria of averaged reflectance, angular variation, and polarization balance are evaluated rather than inferred from nearby parameter values.
[0135] This sampling of the disclosed Ru and Be parameter space at 11.3nm reveals that functional combinations of d and T values satisfying all three criteria (angular range, reflectance variance, and polarized reflectance difference) are localized within the specified combinations, whereas combinations outside these ranges fail to meet the aggregate performance thresholds. Thus, the ranges of d and T define an operational window targeted for some periodic Ru / Be photomask multilayer embodiments for 11.3 nm projection lithography for use at numerical aperture of 0.50 or greater.
[0136] FIG. 12 and 13 are averaged reflectance plots for a subset of the Ru / Be design space, specifically for multilayers with a fixed T value of 0.4 with periods ranging from 5.70 nm to 5.86 nm (shown in FIG. 12) and a fixed 5.80 nm period with T values ranging from 0.42 to 0.52 (shown in FIG. 13). These two plots provide additional detail to the corresponding portions of the combined33 50130535.1parameter cases shown in FIG. 11. As seen in the plots, varying either of the parameters will alter the positions of maximum reflectance, reflectance drop-off, and the through-angle variation of reflectance. FIG. 14 and 15 are plots of the corresponding differences between s-polarized reflectance and p-polarized reflectance (RS-RP), which for these particular cases is larger with a 5.74nm period and a T of 0.4 as well as with a T of 0.54 and a 5.80 nm period than with others. These examples further demonstrate how small changes in period and thickness ratio can shift angular reflectance in a manner that is not proportional to the period or ratio changes. It also demonstrates the need to include polarization influences into design considerations for multilayers for use at angles beyond a few degrees.
[0137] FIG. 16 shows the design space results of Mo / Be multilayer configurations with d and T combinations in the ranges from 5.70 nm to 5.92 nm and 0.24 to 0.68 respectively. The number of layer-pairs for each case is 40. For each d and T combination, polarized reflectance plots were evaluated over a range of angles exceeding design targets. Mo / Be multilayers that did not result in a greater than 60% reflectance over an angular range from -10° to +10° were eliminated. Those that remained were evaluated for angular reflectance variation and the difference between s-polarized and p-polarized reflectance across the -10° to +10° range. Table blocks entries for Mo / Be combinations are coded in the same manner as used for the Ru / Be cases. There are several combinations that meet the aggregate performance targets for each polarization metric. The numbers of Mo / Be combinations that meet these criteria are lower than for Ru / Be, which is a result of the wavelength-specific absorption and index contrast properties of the material combinations. For a 3% polarization limit, there are several parameter combinations in an asymmetric region with periods between 5.76 nm and 5.86 nm and particular T values between 0.36 and 0.62. Combinations that meet the 4% and 5% polarization limitations increase the numbers of suitable combinations, generally toward lower period values. In addition to the values and ranges for period andT described below, each block shown in FIG. 16 indicates the period and T of an embodiment of the present technology.
[0138] For example, in some embodiments, the multilayer stack has a period of 5.76 nm, and T in the range from 0.36 to 0.40. In some embodiments, the multilayer stack has a period of 5.78 nm, T in the range from 0. 40 to 0.46. In some embodiments, the multilayer stack has a period of 5.80 nm, and T in the range from 0.46 to 0.54. In some embodiments, the multilayer stack has a34 50130535.1period of 5.82 nm, and T in the range from 0.52 to 0.56. In some embodiments, the multilayer stack has a period of 5.84 nm, and T in the range from 0.56 to 0.60 meet the target reflectance performance. In some embodiments, the multilayer stack has a period of 5.86 nm, and T of 0.62. Based on Fig. 16, some embodiments have periods ranging from 5.76 to 5.86, and T values ranging from 0.36 to 0.62.
[0139] In some embodiments, the multilayer stack has a period of 5.72 and T in the range from 0.26 to 0.34. In some embodiments, the multilayer stack has a period of 5.72 and T of 0.32. In some embodiments, the multilayer stack has a period of 5.74 and T in the range from 0.32 to 0.38. In some embodiments, the multilayer stack has a period of 5.74 and T in the range from 0.32 to 0.40. In some embodiments, the multilayer stack has a period of 5.76 and T in the range from 0.36 to 0.46. In some embodiments, the multilayer stack has a period of 5.76 and T in the range from 0.36 to 0.48. In some embodiments, the multilayer stack has a period of 5.78 and T in the range from 0.40 to 0.50. In some embodiments, the multilayer stack has a period of 5.78 and T in the range from 0.40 to 0.52. In some embodiments, the multilayer stack has a period of 5.80 and T in the range from 0.46 to 0.56. In some embodiments, the multilayer stack has a period of 5.82 and T in the range from 0.52 to 0.58. In some embodiments, the multilayer stack has a period of 5.82 and T in the range from 0.52 to 0.60. In some embodiments, the multilayer stack has a period of 5.84 and T in the range from 0.56 to 0.64. In some embodiments, the multilayer stack has a period of 5.86 and T in the range from 0.62 to 0.66.
[0140] In some embodiments, the multilayer stack has a period of 5.70 and T in the range from 0.26 to 0.32. In some embodiments, the multilayer stack has a period of 5.72 and T in the range from 0.36 to 0.42. In some embodiments, the multilayer stack has a period of 5.74 and T in the range from 0.28 to 0.30. In some embodiments, the multilayer stack has a period of 5.74 and T in the range from 0.42 to 0.48. In some embodiments, the multilayer stack has a period of 5.76 and T in the range from 0.30 to 0.34. In some embodiments, the multilayer stack has a period of 5.76 and r in the range from 0.50 to 0.52. In some embodiments, the multilayer stack has a period of 5.78 and T in the range from 0.40 to 0.50. In some embodiments, the multilayer stack has a period of 5.78 and T in the range from 0.34 to 0.38. In some embodiments, the multilayer stack has a period of 5.78 and T in the range from 0.54 to 0.56. In some embodiments, the multilayer stack has a period of 5.80 and T in the range from 0.38 to 0.44. In some embodiments, the multilayer stack 35 50130535.1has a period of 5.80 and T in the range from 0.58 to 0.60. In some embodiments, the multilayer stack has a period of 5.82 and T in the range from 0.42 to 0.50. In some embodiments, the multilayer stack has a period of 5.82 and T of 0.62. In some embodiments, the multilayer stack has a period of 5.84 and T in the range from 0.46 to 0.54. In some embodiments, the multilayer stack has a period of 5.86 and T in the range from 0.56 to 0.60.
[0141] In some embodiments of the multilayer stack, the period ranges from 5.72 nm to 5.86 nm, and T values range from 0.26 to 0.66. In some embodiments of the multilayer stack, the period ranges from 5.72 nm to 5.86 nm, and T values range from 0.32 to 0.66. In some embodiments of the multilayer stack, the period ranges from 5.72 nm to 5.80 nm, and T values are from 0.26 to 0.56. In some embodiments of the multilayer stack, the period ranges from 5.72 nm to 5.80 nm, and T values are from 0.32 to 0.56. In some embodiments of the multilayer stack, the period ranges from 5.72 nm to 5.80 nm, and T values are from 0.36 to 0.54. In some embodiments of the multilayer stack, the period ranges from 5.74 nm to 5.84 nm, and T values are from 0.32 to 0.64. In some embodiments of the multilayer stack, the period ranges from 5.74 nm to 5.84 nm, and T values are from 0.36 to 0.60. In some embodiments of the multilayer stack, the period ranges from 5.78 nm to 5.86 nm, and T values are from 0.40 to 0.62. In some embodiments of the multilayer stack, the period ranges from 5.78 nm to 5.86 nm, and T values are from 0.40 to 0.66. In some embodiments of the multilayer stack, the period ranges from 5.80 nm to 5.86 nm, and T values are from 0.46 to 0.62. In some embodiments of the multilayer stack, the period ranges from 5.80 nm to 5.86 nm, and T values are from 0.46 to 0.66.
[0142] This sampling of the disclosed Mo and Be parameter space at 11.3 nm reveals that functional combinations of d and T values satisfying all three criteria (angular range, reflectance variance, and polarized reflectance difference) are localized within the specified combinations, whereas combinations outside these ranges fail to meet the aggregate performance thresholds. Thus, the ranges of d and T define an operational window targeted for some periodic Mo / Be photomask multilayer embodiments for 11.3 nm projection lithography for use at numerical aperture of 0.50 or greater.
[0143] In addition to identifying specific periodic multilayer configurations that satisfy the aggregate performance criteria, the design-space mapping further reveals that certain regions within36 50130535.1the disclosed operational windows exhibit greater tolerance to small variations in bilayer period and thickness ratio than others. In these regions, modest deviations in one or both parameters do not result in loss of compliance with the angular reflectance, angular variation, and polarization-balance targets. As a result, selection of a particular multilayer configuration may be guided not only by nominal performance metrics, but also by the relative breadth of the surrounding parameter space over which acceptable performance is maintained. Such considerations further provide a basis for identifying periodic multilayer designs with increased parameter latitude. The existence and location of these higher-tolerance regions are not apparent from first-order estimates, material optical constants alone, or from prior multilayer designs used for other targeted performances.APERIODIC MULTILAYER STACK EMBODIMENTS
[0144] Depth-graded (or aperiodic) multilayers can provide additional degrees of freedom for further tailoring angular reflectance beyond what is achievable with strictly periodic stacks. As used herein, aperiodic multilayers refer to a multilayer structure in which the thicknesses and / or material sequence of successive layers are varied (or depth-graded) such that the coating stack does not exhibit a single, repeating period throughout its full thickness. By depth-grading the multilayer structure through a multilayer stack, additional manipulation of the angular response is possible. Design flexibility is introduced by either individually or simultaneously altering the period and the ratio of the constituent materials which can vary either monotonically or non-monotonically. In the example material systems, where the periodic designs show distinct differences in angular flatness and normal-incidence behavior among material combinations, aperiodic designs can offer a route to further redistribute reflectance across angle without fundamentally changing the material set.
[0145] For Be-based multilayers, the response to depth-grading depends on the balance between refractive index contrast and absorption in the layers involved. Material combinations like Ru / Be, Rh / Be, and Pd / Be tend to support broader angular response, while absorption limits the number of bilayers that contribute coherently. Depth grading is used herein to adjust how different portions of the multilayer contribute to reflectance at different angles, leading to an extension to angular features that arise from a single fixed period.
[0146] In some Ru / Be multilayers embodiments, which exhibit relatively flat reflectance via periodic structuring from normal incidence to approximately 12°, depth grading may further extend37 50130535.1this behavior by broadening the angular range over which constructive interference occurs. Because Ru / Be combines relatively high index contrast with moderate absorption at 11.3 nm, graded structures may preserve angular flatness while mitigating sensitivity to exact Bragg matching at normal incidence.
[0147] Some Rh / Be multilayer embodiments exhibit similar but more absorption-limited behavior. Depth-graded designs may allow the high index contrast of Rh / Be at 11.3 nm to be used more effectively by reducing the influence of the most strongly absorbing regions, extending flat angular response while accepting some reduction in peak reflectance.
[0148] Some Mo / Be multilayer embodiments, while less flat in strictly periodic designs, may benefit from depth grading in a different way. Their lower absorption at 11.3 nm is observed to allow deeper penetration of the optical field, enabling a larger fraction of the multilayer to contribute at higher angles. As a result, graded Mo / Be structures may enhance reflectance at angles beyond the primary flat-response region, consistent with the observed dominance of Mo / Be reflectance at larger angles in periodic stacks.
[0149] Overall, depth-graded aperiodic multilayers represent an approach for further improving through-angle performance in Be-based reflective multilayers. The specific benefits and limitations of such designs are material -dep endent, with Ru / Be offering potential for extended flat angular response, Rh / Be providing high-contrast opportunities at 11.3 nm, and Mo / Be enabling enhanced reflectance at larger grazing angles.
[0150] The illustrated embodiments are representative of a class of aperiodic multilayer thickness profiles that satisfy the stated performance criteria at 11.3 nm. As used herein, a “thickness profile” refers to the sequence and relative thicknesses of individual layers within the multilayer reflective stack. The technology is not limited to the specific numerical profiles shown.
[0151] Areflective 11.3 nm mask embodiment is shown in FIG. 17 comprised of depth-graded aperiodic multilayers. In the depicted embodiment, the mask 300 includes aperiodic multilayers 302 comprised of a first layer 305 with a thickness ti and a second layer 306 with a thickness t2 having a period the sum of thicknesses of the two layers. These layers are deposited over a pair of first- and second-layer materials 313 and 314 with respective thicknesses t? and t4, which are not necessarily equal to ti and t2. These layers are deposited over other pairs of first- and second-layer materials (not shown) with respective thicknesses, which are not necessarily equal to t3 and t438 50130535.1or other pairs in the stack. The multilayer stack is comprised of an aperiodic series of first- and second- material pairs coated onto substrate 301, totaling between 20 and 70, or between 25 and 60, or between 30 and 50, or between 35 and 45, though numbers outside of this range still fall within the scope of the embodiment. The aperiodic multilayer may include regions having different local periods, non-uniform or gradually varying layer thicknesses, or intentionally irregular sequences of layers. The mask 300 also includes an absorber layer 303 and may include a capping layer 304. When used in a projection lithography system, radiation of a 11.3 nm wavelength 307 illuminates at angles 308 and 310 associated with the projection system numerical aperture and mask reduction ratio. For example, this may be between about -7.2° and +9.60for a numerical aperture of 0.50NA or about -9.1° and +10.5° for a numerical aperture of 0.63NA, with other possibilities corresponding to other numerical apertures, reduction ratios, and CRAO values.
[0152] An example aperiodic mask multilayer embodiment is shown in FIG. 18, comprised of a first layer (layer number 1) which is beryllium and a second layer which is ruthenium. This aperiodic mask multilayer embodiment is herein for convenience referred to as APML1. The top layer of the stack is ruthenium (layer number 80). Multilayer tuning was performed using transfer matrix method (TMM) thin-film modeling combined with gradient descent methods to iteratively enhance reflectivity for both s- and p- polarization states at large angles, up to about 14°. Design targets were chosen to meet previously defined performance criteria of reflectance through angle range, reflectance variance, and polarization divergence. By grading the layers to have a higher fraction of high-index material near the surface, the reduced absorption allows the propagating field to penetrate deeper into the multilayer stack. The gradual variation from T values that favor the less absorbing Be over Ru at the top surface of the multilayer support higher transmission through the coating, leading to a greater influence of the Be / Ru index contrast to lead improve angular bandwidth compared to periodic designs.
[0153] In some aperiodic mask multilayer embodiments, the thickness profile includes a gradient where in some portions the period changes monotonically (by increasing or decreasing) as a function of bilayer index for at least a subset of the stack. This can be seen in FIG. 18, where viewed from the top of the stack (i.e. layer number 80) the ratio of Be to Ru starts off large decreases toward the middle of the stack, where the ratio decreased toward equal Ru and Be thicknesses. In other examples, the Be to Ru ratio decreases in parts of the stack.39 50130535.1
[0154] FIG. 19 shows the resulting s-polarized, p-polarized, and averaged reflectance at 11.3 nm from the layer structure of FIG. 18. Reflectance at 0° is 68.7% and remains above 69% at 10° for all polarization conditions. Angular reflectance remains relatively flat, where at 12°, s-polarization reflectance is 73.3% and p-polarization reflectance is 68.9%, while at 13° they are 73.1% and 67.2% respectively and at 14° they are 71.6% and 63.3% respectively. In the range from 0° to 10°, the variation in the averaged reflectance (AR) is 2.16% and the maximum difference between s-polarized reflectance and p-polarized reflectance (RS-RP) is 2.79%. In the range from 0° to 12°, AR is 2.53% and Rs-RPis 4.41%. Such performance would be able to support the mask angles associated with numerical apertures of 0.50 and higher. For example, an anamorphic 4X / 8X 11.3 nm / 0.70NA projection system with a 6.8° CRAO resulting in mask angles up to 11.8° and a 11.3 nm / 0.75NA system with the same CRAO resulting in mask angles up to about 12.2° could be supported by such a mask layer structure, with sufficient angular latitude to support system design variations that might lead to greater angular demands.
[0155] FIG. 20 shows a mask multilayer embodiment comprised of a first layer which is beryllium and a second layer which is molybdenum, referred to herein as APML2. The top layer of the stack is molybdenum. Multilayer tuning was carried out to maximize s- and p- polarized reflectance at angles up to 12°. The average multilayer period throughout the stack is 5.73nm. The resulting polarized reflectance performance is shown in FIG 21, where s-polarized reflectance is 73.3% and p-polarized reflectance is 68.9% at 12°. In the range from 0° to 10°, AR remains flat at 2.98% and Rs-RPis 2.2. In the range from 0° to 12°, AR is 6.92% and Rs-Rpis 5.68%. In Mo / Be, the intrinsic index contrast is lower than in Ru- or Rh-based stacks, so changes in layer thickness distribution and local period modulation demonstrate a weaker optical leverage on the phase and amplitude balance that governs constructive interference at larger angles. However, Mo / Be remains advantageous in circumstances where higher peak reflectance and deeper field penetration is enabled by lower Mo absorption.LOCALIZED DEVIATIONS IN MULTILAYER STRUCTURE
[0156] In the periodic, depth-graded, or hybrid multilayer reflective stacks, some embodiments also include a localized deviation, or a localized thickness-perturbation, from an otherwise smooth progression of layer thicknesses. These localized thickness-perturbation regions40 50130535.1may take the form of a temporary decrease in period (or a “notch” in the thickness profile), a temporary increase (or “bump” in the thickness profile), the omission of one or more layers, or the introduction of one or more layers having substantially increased thickness, wherein the thicknesses of one or more constituent layers depart from an otherwise constant, monotonic, or smoothly varying thickness sequence. The magnitude and shape of the deviation may vary, including abrupt or gradual transitions, provided that the resulting thickness profile exhibits a localized departure from periodic or smoothly graded behavior. Such features may extend across one or several adjacent layer pairs and may occur within an intermediate portion of the multilayer stack, including but not limited to regions closer to the substrate than to the surface. In certain embodiments, the multilayer reflective stack further includes one or more deliberately formed localized thickness-perturbation regions in which the thickness of at least one of the one or more layers deviates by at least 10% or at least 25% from an otherwise periodic, depth-graded, or hybrid thickness profile. Such localized thicknessperturbation regions may comprise a group of adjacent layers or a single layer having a thickness that differs by at least 10% in some embodiments, and by at least 25% in other embodiments, from neighboring layers to modify the reflected phase or polarization response as a function of incidence angle.
[0157] For some embodiments, the multilayer reflective stack includes a localized thicknessperturbation region comprising two or more adjacent layers for which the average thickness of the two or more adjacent layers differs by at least 10% from an average thickness of a neighboring two or more adjacent layers, wherein the two or more adjacent layers and the neighboring two or more adjacent layers have an identical number and composition of layers. In some embodiments, the average thickness of the two or more adjacent layers differs by at least 12.5% from an average thickness of a neighboring two or more adjacent layers. In some embodiments, the average thickness of the two or more adjacent layers differs by at least 15% from an average thickness of a neighboring two or more adjacent layers. In some embodiments, the average thickness of the two or more adjacent layers differs by at least 17.5% from an average thickness of a neighboring two or more adjacent layers. In some embodiments, the average thickness of the two or more adjacent layers differs by at least 20% from an average thickness of a neighboring two or more adjacent layers. In some embodiments, the average thickness of the two or more adjacent layers differs by at least 22.5% from an average thickness of a neighboring two or more adjacent layers. In some41 50130535.1embodiments, the average thickness of the two or more adjacent layers differs by at least 25% from an average thickness of a neighboring two or more adjacent layers.
[0158] In some embodiments, the difference in average thickness is that the two or more adjacent layers of the localized thickness-perturbation region have a greater thickness than average thickness of a neighboring two or more adjacent layers. In some embodiments, the two or more adjacent layers of the localized thickness-perturbation region have a lesser thickness than average thickness of a neighboring two or more adjacent layers.
[0159] For some embodiments, the multilayer reflective stack includes a localized thicknessperturbation region consisting of a single layer having a thickness that differs by at least 25% from the thicknesses of two or more adjacent layers on each side of the single layer.
[0160] Without wishing to be bound by theory, this localized deviation represents an area of phase matching and field redistribution within the multilayer stack. As radiation propagates into the structure, different depth regions contribute preferentially to reflectance over different portions of the angular response. In a multilayer design configured for extended angular performance, it can be advantageous to locally alter the optical thickness progression to adjust the relative phase of reflected fields originating from deeper regions of the stack. The resulting localized deviation effectively rebalances constructive interference across the desired angular range without requiring a global modification of the entire multilayer period profile.
[0161] The prominence of such localized deviations is dependent on the unique optical properties of the constituent materials, particularly the wavelength-specific balance between refractive index contrast and absorption at 11.3 nm. Material combinations such as Ru / Be and Rh / Be, which exhibit relatively high refractive index contrast together with moderate absorption, can tend to support strong interference contributions from multiple depth regions, which may introduce localized deviations in these systems to manage competing interference conditions and to redistribute reflectance contributions across angle. In contrast, lower-contrast or lower-absorption systems, such as Mo / Be, may exhibit smoother thickness progressions, as fewer strongly competing interference pathways are present.
[0162] Reflectance profiles associated with some multilayer stack embodiments incorporating such localized deviations may exhibit a broadened response that may include two partially overlapping maxima or a flattened plateau with a subtle inflection, rather than more single42 50130535.1sharply defined peaks. This behavior can be described as a multi-lobed or composite angular reflectance response, reflecting the attribute that different regions of the multilayer stack contribute preferentially to different angular ranges. The localized deviation acts to couple these contributions, thereby extending usable reflectance performance while maintaining overall stack thickness and manufacturability within practical limits.
[0163] The mask layer structure embodiment of FIG. 18 shows an example of such a localized, non-periodic deviation in layer thickness progression, where a reduced thickness notch 80 in the sequence is identified, extending across several layers starting at about layer number 16. This gives rise to a subtle inflection in both the s- and p-polarized reflectance as seen in FIG. 19 between about 9.5° and 10°. This localized layer deviation enables the extension of flat angular reflectance response out to larger angles. FIG. 22 shows an aperiodic mask layer structure that exhibits a localized, non-periodic deviation of increased thickness progression, identified as a bump 82 starting at about layer number 16 and extending across several layers. This embodiment is referred to herein as APML3. FIG. 23 is the resulting polarized reflectance plot, where a reflectance inflection at about 9.5° and 12° also enables the extension of flat angular reflectance response out to larger angles. In the range from 0° to 10°, AR is 3.22% and Rs-Rp is 3.02%. In the range from 0° to 12°, AR is 3.22% and s-Rp is 4.77%.BLOCK- SEGMENTED PERIODIC / APERIODIC EMBODIMENTS
[0164] In the manufacturing of 11.3 nm mask multilayers, a strictly periodic stack can support repeatable deposition processes because each bilayer has the same target thickness, potentially improving process accuracy and limiting cumulative thickness-drift error. However, fully periodic stacks offer less flexibility in tailoring angle-dependent and polarization-dependent reflectance than aperiodic designs. A block-segmented approach addresses this trade-off by dividing the multilayer into functional regions, or thickness-profile blocks. For example, one or more periodic blocks may establish a stable core that provides the dominant Bragg response, while adjacent aperiodic blocks, placed for example near the top and / or bottom of the stack, introduce controlled thickness perturbations where they can have the greatest influence on angular bandwidth, phase behavior, and polarization balance, while limiting the number of layers utilizing non-periodic tuning.43 50130535.1
[0165] For example, Ru / Be multilayer stacks may be configured to extend useful reflectance performance over a broader angular range by combining periodic and aperiodic sections within a single structure. A periodic Ru / Be multilayer may provide strong angular reflectance and the introduction of limited aperiodicity within selected regions of the stack enables broadening of the angular response while retaining the manufacturability benefits of periodic deposition. One implementation maintains a block of strictly periodic Ru / Be bilayers to supply a strong baseline reflectance contribution, while incorporating one or more adjacent graded-thickness (aperiodic) blocks above, below, or between periodic blocks, depending on the desired balance between angular broadening and overall reflectance. In such arrangements, the periodic block remains the primary reflector, and the aperiodic sections modify the angular response by reducing dependence on phase matching at a single effective period.
[0166] Placing a periodic block toward the middle or lower portion of the multilayer can help preserve reflectance strength, as incident radiation couples into the periodic region after partial interaction with the aperiodic layers, which primarily act to smooth the angular response without excessive attenuation. Alternatively, a periodic top section may be used where surface reflectance stability is desired, with underlying aperiodic sections providing additional angular broadening. Additionally, blocks may be all periodic, all aperiodic, or a combination of periodic and aperiodic.
[0167] A block-segmented Ru / Be mask multilayer is shown in FIG. 24, referred to as APML4. The bottom portion of the stack comprises a 10-pair aperiodic block with an average period of 6.13nm. This is followed by 20-pair periodic block with a period of 5.80 nm and a T of 0.46. A single 5.95 nm Ru layer separates the aperiodic segment from the periodic segment and acts as a localized deviation or bump 84. The 20-pair periodic block is followed by a 10-pair aperiodic block with an average period of 5.60 nm. The average period thickness of the entire sack is 5.85 nm. The resulting reflectance plot is shown in FIG. 25, which shows flat reflectance performance greater than about 65% for s- and p-polarizations up to about 13.5°. In the range from 0° to 10°, AR 1.34% and Rs-Rp is 3.16%. In the range from 0° to 12°, AR is 1.51% and Rs-RP5.29%.MIXED-PAIR MULTILAYER STACK EMBODIMENTS
[0168] Mixed-pair multilayer stacks that incorporate both constituent pairs present an opportunity to tailor angular reflectance performance. This can be demonstrated for example through44 50130535.1mixed pairing Mo / Be and Ru / Be bilayers. Mo / Be exhibits comparatively lower absorption enabling deeper field penetration to enhance peak reflectance at larger angles. Ru / Be provides higher index contrast and stronger coupling, resulting in a broader angular response and improved angular reflectance uniformity. Other opportunities exist based on the particular properties of the layer combinations employed, such as combinations of two or more of Mo / Be, Rh / Be, Ru / Be, Pd / Be, Nb / Be, Y / Be, and Zr / Be. Combining these material pairs within a single stack allows the reflectance response to be distributed across depth, with different portions of the stack contributing preferentially over different angular ranges.
[0169] The ordering of the material pairs within the stack provides an additional degree of freedom. For example, when Ru / Be or Rh / Be bilayers are positioned closer to the incident surface, the high index contrast promotes strong reflection at low to moderate angles, while employing underlying Mo / Be or Nb / Be bilayers can continue to contribute at higher angles due to their lower absorption and reduced damping. Alternatively, placing Mo / Be or Nb / Be layers nearer the surface allows incident radiation to propagate more deeply into the stack before significant attenuation occurs, enabling Ru / Be or Rh / Be layers at greater depth to reinforce reflectance over a broader angular range. In either configuration, the mixed-pair approach offers a means to balance absorption, index contrast, and angular bandwidth, and for designs targeting enhanced reflectance uniformity from normal incidence through to larger angles.
[0170] An example embodiment is shown in FIG. 26 where 14 aperiodic layer pairs of Ru / Be constitute a lower block-segment coated on a mask substrate followed by a period block of Ru / Be comprising 13-layer pairs of period 5.80nm followed by 13 layer pairs of aperiodic Mo / Be, with an average period over the entire stack of 5.70nm. This embodiment is referred to herein as APML5. Beryllium comprises the bottom-most and top-most layers and there is a localized beryllium layer void at layer number 13 representing a ruthenium layer between adjacent layers of total thickness 5.82nm The resulting angular reflectance is shown in FIG. 27, where the mixed-pairing of Ru / Be with Mo / Be gives rise to an increase in reflectance in angles leading up to 10°. In the range from 0° to 10°, the variation in the averaged reflectance is 2.92% and the maximum difference between s-polarized reflectance and p-polarized reflectance is 2.83%. In the range from 0° to 12°, AR is 2.92% and Rs-RPis 4.95%.MULTILAYER STACK COUNT VARIATION EMBODIMENTS45 50130535.1
[0171] Multilayer reflective coatings at 11.3 nm may be implemented using on the order of several tens of layer pairs, for example around forty layer pairs, in order to achieve high reflectance with adequate angular bandwidth and polarization control behavior. Layer-pair counts are driven by the absorption effects of the constituent materials that make up the layer pairs together with the constructive interference supported by refractive-index contrast. The multilayer skin depth provides a useful first-order metric for estimating cumulative absorption losses within a multilayer stack. For Ru / Be at 11.3 nm, the skin depth is approximately 231 nm. For a bilayer period near 5.8 nm, this corresponds to roughly 40 layer pairs representing the depth at which the initial incident intensity is reduced to approximately 1 / e of its original value. For Mo / Be with the same bilayer period, the skin depth is approximately 309 nm, corresponding to about 53 layer pairs. For Rh / Be, a depth of approximately 181 nm corresponds to about 31 layer pairs. The skin depth for Nb / Be is approximately 360 nm, corresponding to about 62 layer pairs of the same period, and for Zr / Be the skin depth is approximately 461 nm, corresponding to about 80 layer pairs. Accordingly, more or fewer layer pairs may be selected to achieve desired optical responses based on index-contrast and absorption trade-offs, and the foregoing values are illustrative rather than limiting.
[0172] In this context, Mo / Be multilayers may exhibit incremental reflectance contributions as the layer count increases toward the skin depth, as the relatively lower refractive-index contrast of the Mo / Be material system allows additional bilayers to contribute constructively before absorption effects become limiting. In contrast, Ru / Be multilayers exhibit a different reflectance accumulation behavior. Due to the higher refractive-index contrast between Ru and Be, a substantial fraction of the achievable reflectance is realized within a smaller number of layer pairs. As a result, Ru / Be multilayers having approximately thirty layer pairs may exhibit reflectance and angular response characteristics that continue to evolve as the absorption skin depth is approached. This behavior reflects the stronger optical coupling per interface in the Ru / Be system, which enables efficient formation of the reflective interference condition with fewer repetitions. Rh / Be multilayers designed around even fewer pairs, for example as few as less than thirty, may be enabled because of an even larger refractive-index contrast.
[0173] From a manufacturing perspective, the ability of Ru / Be and Rh / Be multilayers to achieve strong reflective performance with a reduced number of layer pairs may provide additional design flexibility without excluding the use of higher pair counts. Fewer total layers can reduce46 50130535.1cumulative thickness errors, interface roughness propagation, and deposition time, while also relaxing constraints on long-term thickness drift during deposition processes. At the same time, multilayers may still be implemented with higher layer counts where desired, for example to tune angular response, spectral bandwidth, or robustness to process variation. These considerations do not preclude the use of higher layer counts, which may be selected in other embodiments to satisfy optical performance objectives independent of manufacturing convenience.
[0174] An embodiment of a reduced 30-layer-pair Ru / Be aperiodic multilayer stack example is shown in FIG. 28, referred to as APML6. The average period of the entire stack is 5.80 nm, and the resulting angular reflectance plot is shown in FIG. 29. In the range from 0° to 10°, AR is 2.53%, and Rs-Rp is 2.88%. In the range from 0° to 12°, AR is 4.84%, and Rs-RPis 5.94%.
[0175] A summary of reflective performance criteria for the multilayer embodiments disclosed herein is presented in Table 4.Table 4. A summary of AR and RS-RPvalues from 0-10° and from 0-12° for aperiodic multilayers APML1 through AMPL6 (in percent reflectance).Multilayer stack embodiment layer details for each of the aperiodic multilayer examples disclosed herein are shown in Tables 5 and 6. The embodiment example configurations referred to as APML1 through APML3 are shown in Table 5 and example configurations referred to as APML4 through 6 are shown in Table 6. Layer materials and corresponding layer thickness values (in nm) are listed in columns corresponding to layer numbers, starting with layer number 1 at the bottom of the stacks over the substrate. The highest numbers represent the top of each stack.47 50130535.1Layer APML1 APML2 APML3 No. Layers nm Layers nm Layers nm 1 Be 2.77 Be 2.97 Be 2.80 2 Ru 3.07 Mo 3.02 Ru 3.09 3 Be 2.84 Be 3.00 Be 2.89 4 Ru 3.04 Mo 2.97 Ru 3.06 5 Be 2.83 Be 2.94 Be 2.87 6 Ru 3.01 Mo 2.94 Ru 3.01 7 Be 2.84 Be 2.94 Be 2.89 8 Ru 2.99 Mo 2.93 Ru 3.02 9 Be 2.85 Be 2.93 Be 2.95 10 Ru 2.94 Mo 2.91 Ru 2.98 11 Be 2.80 Be 2.93 Be 2.91 12 Ru 2.80 Mo 2.90 Ru 2.90 13 Be 2.73 Be 2.94 Be 3.00 14 Ru 2.72 Mo 2.90 Ru 2.96 15 Be 2.77 Be 2.94 Be 3.05 16 Ru 2.39 Mo 2.89 Ru 2.57 17 Be 2.60 Be 2.95 Be 2.91 18 Ru 2.55 Mo 2.88 Ru 3.69 19 Be 2.61 Be 2.95 Be 5.36 20 Ru 2.63 Mo 2.86 Ru 3.46 21 Be 2.71 Be 2.96 Be 2.84 22 Ru 2.84 Mo 2.85 Ru 2.61 23 Be 2.86 Be 2.97 Be 3.10 24 Ru 2.89 Mo 2.85 Ru 2.94 25 Be 2.87 Be 2.98 Be 3.01 26 Ru 2.88 Mo 2.84 Ru 2.85 27 Be 2.90 Be 2.98 Be 3.01 28 Ru 2.91 Mo 2.83 Ru 2.92 29 Be 2.95 Be 2.99 Be 3.04 30 Ru 2.92 Mo 2.82 Ru 2.89 31 Be 2.98 Be 3.00 Be 3.02 32 Ru 2.91 Mo 2.81 Ru 2.87 33 Be 3.00 Be 3.00 Be 3.05 34 Ru 2.89 Mo 2.80 Ru 2.87 35 Be 3.01 Be 3.01 Be 3.06 36 Ru 2.88 Mo 2.80 Ru 2.84 37 Be 3.03 Be 3.02 Be 3.0850130535.138 Ru 2.86 Mo 2.79 Ru 2.83 39 Be 3.05 Be 3.02 Be 3.09 40 Ru 2.85 Mo 2.79 Ru 2.81 41 Be 3.06 Be 3.03 Be 3.10 42 Ru 2.82 Mo 2.78 Ru 2.79 43 Be 3.06 Be 3.03 Be 3.12 44 Ru 2.81 Mo 2.78 Ru 2.78 45 Be 3.08 Be 3.04 Be 3.13 46 Ru 2.80 Mo 2.77 Ru 2.76 47 Be 3.10 Be 3.04 Be 3.15 48 Ru 2.79 Mo 2.76 Ru 2.74 49 Be 3.11 Be 3.04 Be 3.16 50 Ru 2.77 Mo 2.76 Ru 2.73 51 Be 3.12 Be 3.05 Be 3.18 52 Ru 2.76 Mo 2.75 Ru 2.71 53 Be 3.13 Be 3.05 Be 3.19 54 Ru 2.74 Mo 2.75 Ru 2.68 55 Be 3.13 Be 3.05 Be 3.20 56 Ru 2.73 Mo 2.74 Ru 2.66 57 Be 3.14 Be 3.05 Be 3.22 58 Ru 2.71 Mo 2.74 Ru 2.64 59 Be 3.15 Be 3.06 Be 3.24 60 Ru 2.71 Mo 2.73 Ru 2.61 61 Be 3.17 Be 3.07 Be 3.26 62 Ru 2.70 Mo 2.73 Ru 2.58 63 Be 3.18 Be 3.07 Be 3.29 64 Ru 2.69 Mo 2.72 Ru 2.55 65 Be 3.18 Be 3.08 Be 3.32 66 Ru 2.66 Mo 2.71 Ru 2.51 67 Be 3.17 Be 3.09 Be 3.34 68 Ru 2.65 Mo 2.70 Ru 2.46 69 Be 3.20 Be 3.12 Be 3.38 70 Ru 2.65 Mo 2.67 Ru 2.41 71 Be 3.21 Be 3.11 Be 3.41 72 Ru 2.62 Mo 2.59 Ru 2.36 73 Be 3.19 Be 3.17 Be 3.47 74 Ru 2.60 Mo 2.62 Ru 2.30 75 Be 3.21 Be 3.22 Be 3.50 76 Ru 2.59 Mo 2.27 Ru 2.22 77 Be 3.20 Be 3.02 Be 3.5650130535.178 Ru 2.50 Mo 2.10 Ru 2.1479 Be 3.21 Be 2.51 Be 3.6280 Ru 2.12 Mo 0.95 Ru 1.77Table 5. Layers and corresponding thickness values for aperiodic multilayer embodiment examples A PML 1 through A PML3.Layer APML4 APML5 AMPL6No. Layers nm Layers nm Layers nm1 Be 3.04 Be 2.78 Be 2.792 Ru 4.18 Ru 3.08 Ru 3.053 Be 3.82 Be 2.84 Be 2.864 Ru 3.33 Ru 3.08 Ru 3.025 Be 2.86 Be 2.86 Be 2.886 Ru 2.87 Ru 3.09 Ru 3.007 Be 2.84 Be 2.86 Be 2.908 Ru 2.92 Ru 3.08 Ru 2.979 Be 2.92 Be 2.86 Be 2.9210 Ru 2.93 Ru 3.17 Ru 2.9511 Be 2.94 Be 2.86 Be 2.9412 Ru 2.91 Ru 3.02 Ru 2.9213 Be 2.97 Be 0.00 Be 2.9614 Ru 2.91 Ru 2.80 Ru 2.9015 Be 3.01 Be 2.86 Be 2.9816 Ru 2.91 Ru 2.99 Ru 2.8717 Be 3.02 Be 2.91 Be 3.0018 Ru 2.90 Ru 3.01 Ru 2.8419 Be 3.04 Be 2.93 Be 3.0220 Ru 5.95 Ru 2.98 Ru 2.8221 Be 3.00 Be 2.90 Be 3.0522 Ru 2.80 Ru 2.99 Ru 2.7923 Be 3.00 Be 2.95 Be 3.0724 Ru 2.80 Ru 2.96 Ru 2.7625 Be 3.00 Be 2.94 Be 3.0926 Ru 2.80 Ru 2.93 Ru 2.7327 Be 3.00 Be 2.97 Be 3.1228 Ru 2.80 Ru 3.44 Ru 2.7029 Be 3.00 Be 3.00 Be 3.1430 Ru 2.80 Ru 2.80 Ru 2.6750130535.131 Be 3.00 Be 3.00 Be 3.17 32 Ru 2.80 Ru 2.80 Ru 2.64 33 Be 3.00 Be 3.00 Be 3.20 34 Ru 2.80 Ru 2.80 Ru 2.61 35 Be 3.00 Be 3.00 Be 3.23 36 Ru 2.80 Ru 2.80 Ru 2.58 37 Be 3.00 Be 3.00 Be 3.25 38 Ru 2.80 Ru 2.80 Ru 2.55 39 Be 3.00 Be 3.00 Be 3.28 40 Ru 2.80 Ru 2.80 Ru 2.51 41 Be 3.00 Be 3.00 Be 3.30 42 Ru 2.80 Ru 2.80 Ru 2.47 43 Be 3.00 Be 3.00 Be 3.33 44 Ru 2.80 Ru 2.80 Ru 2.44 45 Be 3.00 Be 3.00 Be 3.36 46 Ru 2.80 Ru 2.80 Ru 2.41 47 Be 3.00 Be 3.00 Be 3.39 48 Ru 2.80 Ru 2.80 Ru 2.37 49 Be 3.00 Be 3.00 Be 3.42 50 Ru 2.80 Ru 2.80 Ru 2.33 51 Be 3.00 Be 3.00 Be 3.45 52 Ru 2.80 Ru 2.80 Ru 2.30 53 Be 3.00 Be 3.00 Be 3.48 54 Ru 2.80 Ru 2.80 Ru 2.26 55 Be 3.00 Be 3.44 Be 3.51 56 Ru 2.80 Mo 2.63 Ru 2.21 57 Be 3.00 Be 3.19 Be 3.55 58 Ru 2.80 Mo 2.61 Ru 2.16 59 Be 3.00 Be 3.20 Be 3.59 60 Ru 2.80 Mo 2.59 Ru 1.85 61 Be 3.60 Be 3.2262 Ru 2.63 Mo 2.5763 Be 3.18 Be 3.2464 Ru 2.67 Mo 2.5465 Be 3.20 Be 3.2666 Ru 2.64 Mo 2.5167 Be 3.21 Be 3.2968 Ru 2.60 Mo 2.4869 Be 3.23 Be 3.3170 Ru 2.56 Mo 2.4550130535.171 Be 3.25 Be 3.3472 Ru 2.50 Mo 2.4173 Be 3.28 Be 3.3774 Ru 2.41 Mo 2.3675 Be 3.33 Be 3.4176 Ru 2.29 Mo 2.3177 Be 3.46 Be 3.4678 Ru 2.13 Mo 2.2579 Be 3.64 Be 3.2080 Ru 1.17 Mo 0.00Table 6. Layers and corresponding thickness values for aperiodic multilayer embodiment examples APML4 through APML6.APERIODIC MULTILAYER DESIGN METHODOLOGY
[0176] Some aperiodic multilayer embodiments disclosed herein were obtained through numerical multilayer modeling in which target optical performance criteria were defined in advance and used to guide adjustment of the multilayer thickness profile. In particular, target values were established for s- and p-polarized reflectance, angular reflectance variation, and polarizationdependent reflectance differences over an angular range representative of reflective photomask operation at 11.3 nm for numerical apertures of 0.50 and greater.
[0177] Starting from one or more initial multilayer configurations, individual layer thicknesses, thickness ratios, or selected groups of layers within the stack were varied and evaluated using transfer-matrix calculations at multiple incidence angles. Calculated reflectance responses were compared against the defined performance targets including (1) maintaining an average of s-and p-polarized reflectance above a threshold value, (2) limiting reflectance variation across the angular range, and (3) limiting divergence between s- and p-polarized reflectance, and the multilayer structure was adjusted iteratively to improve agreement with those targets.
[0178] In practice, this evaluation and adjustment process may be carried out using numerical techniques that allow simultaneous consideration of multiple performance objectives. Such techniques may include broader searches of the design space, for example using global or heuristic approaches, followed by refinement of selected solutions using incremental or gradient-52 50130535.1based adjustments. Performance targets may be expressed through one or more merit functions that quantify agreement with the desired reflectance, angular flatness, and polarization criteria.
[0179] The resulting multilayer configurations were further evaluated in combination with representative absorber layers to confirm that the multilayer performance remained compatible with absorber reflectance suppression requirements relevant to blank (i.e. un-patterned) and patterned photomasks.
[0180] The disclosed multilayer examples therefore represent solutions identified through objective-driven numerical evaluation and demonstrate that multilayer stacks satisfying the claimed performance criteria can be realized using established multilayer modeling and enhancement techniques.DESIGN GUIDANCE DERIVED FROM MULTILAYER EMBODIMENTS
[0181] From the foregoing embodiments, a set of practical design relationships has been identified that may assist in selecting and tuning multilayer architectures for large-angle reflective photomasks operating near 11.3 nm. These relationships arise from observed interactions at 11.3 nm and at high-NA angular regimes among refractive-index contrast, absorption, period selection, angular tuning behavior, and polarization sensitivity across periodic, aperiodic, block-segmented, and mixed-pair multilayer implementations. These interactions have not been previously recognized or exploited, including at wavelengths of approximately 11.3 nm or longer.
[0182] In periodic and aperiodic Be-based stacks, lower-absorption material combinations such as Mo / Be and Nb / Be tend to produce narrower-angle reflectance behavior, with small increases in period (i.e. on the order of O.Olnm) shifting the Bragg response toward larger angles at the expense of reduced reflectance near normal incidence. In contrast, combinations such as Ru / Be and Rh / Be, which provide higher index contrast with only moderate absorption, allow a larger effective depth of coherent contribution and can support comparatively flatter averaged reflectance across the primary angular range of interest, even when the reflectance maximum is shifted to higher angles. Higher-contrast but more strongly absorbing systems such as Pd / Be exhibit broadened but more strongly damped angular response, illustrating a design trade-off between angular bandwidth and cumulative attenuation. With increases beyond a certain point, some embodiments indicate that nearnormal reflectance may further decrease and through-angle variation may increase, with polarization53 50130535.1characteristics changing in a non-uniform manner even as peak reflectance shifts to still larger angles.
[0183] With periodic multilayers, systematic evaluation across period and thickness-ratio parameter space shows that localized combinations simultaneously satisfy targets for averaged reflectance, angular reflectance variation, and polarization balance, and that these functional regions do not change uniformly with varying period or thickness ratio. Some embodiments indicate that photomask-compatible periodic designs can be selected from a constrained operational window defined by aggregate angular and polarization performance criteria.
[0184] Some aperiodic or depth-graded multilayer embodiments demonstrate that introducing controlled thickness variations through the stack provides additional freedom to redistribute reflectance across angles beyond what is achievable with a single fixed period. By adjusting how different portions of the multilayer contribute at different incidence angles, aperiodic designs enable extension of angular bandwidth, mitigation of angular roll-off, and redistribution of polarized reflectance toward higher angles in a manner not attainable through periodic structuring alone.
[0185] Some aperiodic embodiments further indicate that controlled variation of period or layer thickness ratio within selected regions of the stack may be used to redistribute the angular response without fully sacrificing reflectance strength contributed by more periodic portions of the structure. Localized deviations or thickness perturbations inserted within limited portions of the stack can modify phase accumulation and polarization behavior in a targeted manner, while limiting sensitivity to manufacturing variation in the remainder of the multilayer. These observations provide guidance that selective aperiodicity can be effective when applied in regions where angular smoothing or polarization balancing is desired.
[0186] In block-segmented architectures, some embodiments demonstrate that maintaining a core region of substantially periodic layers may be advantageous for preserving strong baseline reflectance, while adjacent aperiodic or graded-thickness blocks can be used to broaden angular response or mitigate polarization divergence. In addition to optical benefits, the use of a predominantly periodic core may also support manufacturability by limiting the number of layers utilizing aperiodic tuning, thereby reducing cumulative thickness-drift sensitivity and simplifying process control during deposition. The relative placement of these blocks (e.g., near the top, center,54 50130535.1or lower portion of the stack) influences how the incident field couples into the structure and which layers dominate at different angles, indicating that segmentation offers a tunable mechanism for balancing reflectance uniformity and high-angle performance.
[0187] Some mixed-pair multilayer embodiments further show that combining two or more material systems within a single stack enables functional distribution of optical response with depth, where higher-contrast pairs may preferentially support low-to-moderate angles and lower-loss or lower-absorption pairs may reinforce reflectance at larger angles. The ordering and relative thickness contribution of these regions provide an additional design degree of freedom that was not evident from periodic single-pair structures alone.
[0188] For some of these embodiments, the results collectively indicate that multilayer performance near 11.3 nm is governed by coupled material-property and interference effects that do not scale linearly with small changes in period, composition, or depth distribution. The design relationships described above therefore reflect engineering insights derived from the disclosed structures and may be employed in related implementations to prioritize angular flatness, polarization balance, or high-angle reinforcement, depending on lithographic system requirements. BARRIER LAYERS
[0189] In certain embodiments, one or more ultrathin barrier layers are incorporated at selected interfaces within a multilayer reflective stack configuration. These barrier layers are inserted between adjacent constituent materials of the multilayer stack for the purpose of stabilizing the interface structure, limiting interdiffusion during or after deposition, and maintaining the optical contrast that supports constructive multilayer reflection. Barrier layers may be applied throughout the full multilayer stack or only within selected depth regions, for example near the top or bottom portion of the multilayer where thermal or radiation exposure may be greater.
[0190] In such embodiments, the barrier material is selected to exhibit chemical compatibility with the adjacent multilayer constituents, structural stability under 11.3 nm exposure conditions, and optical properties that do not substantially degrade the target reflectance response. Suitable barrier materials include, for example, boron carbide (B4C), refractory or other low-absorbing metals, carbon-rich carbides, or other chemically stable, low-absorption compounds that form continuous ultrathin interfacial layers. The barrier-layer thickness may be maintained within a55 50130535.1sub-nanometer range, such as approximately 2A to 10A, so that interface stabilization is achieved while limiting additional absorption and preserving the designed optical phase relationship within the multilayer period.
[0191] FIG. 30 illustrates the reflectance response one such embodiment in which a 5.80 nm-period Ru / Be multilayer reflective stack includes 5 A of B4C disposed at each Ru / Be interface. In this configuration, the interlayers act as localized interface-control layers that help maintain the intended bilayer structure through the full stack thickness while allowing the multilayer to retain the reflectance, angular response, and polarization performance associated with the selected Ru / Be material system at 11.3 nm. The barrier layers may be applied at every interface or at only a subset of interfaces, depending on the desired balance between interface stabilization, absorption management, and manufacturing considerations. In the range from 0° to 10°, the variation in the averaged reflectance is 3.44% and the maximum difference between s-polarized reflectance and p-polarized reflectance is 2.39%.ELECTRIC FIELD INTENSITY AND PHASE EFFECTS
[0192] FIG. 31 shows a plot of the electric field intensity (EFI) through the aperiodic Ru / Be multilayer embodiment stack of FIGS 18 and 19. at 11.3 nm. EFI data is presented for 0°, 6°, and 12°. Trend lines are fitted for 0° and 12°, indicating EFI decay through the stack when illuminated at normal incidence and at large angles. When the field intensity decreases to 1 / e of its initial value (known as the 1 / e depth), it reaches 36.8% of the original intensity, indicating significant decay. This identifies the locations of effective focal planes within the mask multilayer stack. The normal incidence (0°) 1 / e depth is 45nm, and shallower into the film stack than is commonly achieved at 13.5 nm using Mo / Si mask multilayers. The 1 / e depth for 12°, it is 51nm. The 6nm difference is low and can lead to smaller mask three-dimensional effects.
[0193] FIG. 32 shows the s- and p- polarized reflected phase (4>) for the aperiodic depth-graded Ru / Be multilayer embodiments of FIGS. 18 and 19. For a reflective photomask, the absolute phase at normal incidence is of less consequence; instead, the varying phase shift with angle is most important. The oblique angles of mask illumination together with pattern diffraction determine the angular distribution into the multilayer. For each polarization state, unique polynomial equations incorporating linear, quadratic, and fourth-order terms have been fitted, capturing the phase behavior56 50130535.1through the entire angular range (i.e., up to about 13°). When a multilayer film is coated on the planar surface of a photomask, an amount of spherical aberration inherently results, indicated by fourth-order term of the fitted polynomial. The second-order (quadratic) term represents the change in the optical path length as the angle of incidence increases, leading to an effective shift in the focus. The magnitude of these even terms indicates the symmetrical wavefront aberration of a particular design, becoming more significant at larger angles. Fit lines of the corresponding polynomials are depicted for both polarization states with low second-order and fourth-order fit terms. The sign of the fourth-order term is notably negative, suggesting that phase accumulation through the depth of the stack leads to a counteracting spherical aberration.PHOTOMASK MULTILAYER STACK DEPOSITION METHODS
[0194] The multilayer coating embodiments described herein may be deposited onto a mask substrate blank using physical vapor deposition (PVD) techniques, and in particular magnetron sputtering methods. In a magnetron sputtering system, shown in FIG. 33, the mask substrates 51a, 51b, typically a low-thermal-expansion glass or glass-ceramic material, are mounted on a rotating or translating substrate holder 52 within a high-vacuum deposition chamber 53 having a base pressure typically below 107Torr achieved via a vacuum pump 54. Deposition is performed by sequentially sputtering from multiple solid targets 55a, 55b corresponding to the constituent multilayer materials using DC and / or RF power supplies, with target power levels typically in the range of 50-500 W per target, depending on material and desired deposition rate. An inert sputtering gas 56, such as argon, is introduced at controlled flow rates through mass flow controllers 57 to maintain operating pressures generally in the 1-10 mTorr range, and one or more reactive gases 58, such as nitrogen, methane, or oxygen, may be introduced when depositing compound layers or diffusion barrier layers. Individual multilayer periods are formed by alternating exposure of the substrate to different sputtering targets 55a, 55b, with layer thickness determined by one or more of deposition time, target power, gas pressure, and substrate motion. For aperiodic or depth-graded multilayer structures, the deposition sequence is intentionally varied through the multilayer stack by adjusting one or more of layer thickness, material ratio, or target exposure order as a function of stack depth. Optional diffusion barrier or interlayers may be deposited by brief exposure to additional targets between primary layers.57 50130535.1
[0195] In certain implementations, deposition conditions are selected with particular consideration to the material behavior of beryllium in the multilayer structures. Beryllium exhibits relatively high atomic mobility and sensitivity to energetic deposition conditions, such that interface quality may be affected by excessive thermal or kinetic energy during fdm growth. Accordingly, deposition of beryllium-containing multilayers may be performed under controlled substrate temperature conditions to limit diffusion and preserve well-defined interfaces. Deposition is preferably carried out at or near ambient temperature, or under controlled cooling, to reduce thermally driven intermixing while maintaining uniform film growth.
[0196] Deposition parameters may further be selected to limit energetic particle bombardment of the growing film. In sputter-based processes, this may include adjustment of working gas pressure, target power, and substrate bias to reduce the kinetic energy of arriving species and suppress sub-surface implantation. In some embodiments, elevated working pressures are used to promote thermalization of sputtered atoms prior to arrival at the substrate, thereby favoring surface-limited growth and reducing interface mixing. These controls are particularly relevant for beryllium-containing layers, which are more susceptible to interfacial modification under energetic deposition conditions.
[0197] Deposition sequences may also be configured to promote layer-by-layer growth and minimize formation of intermixed or amorphous interface regions. Growth rates may be adjusted, and transitions between materials may be moderated through brief pauses, parameter ramping, or formation of thin transition regions. Such approaches allow stabilization of each layer prior to deposition of the next and reduce the likelihood of chemical interaction at the interface. These techniques are especially beneficial for short-period multilayers used at wavelengths near 11.3 nm, where interface sharpness strongly influences reflectance.
[0198] In multilayer systems incorporating ruthenium or molybdenum in combination with beryllium, additional attention may be given to interface control. Ruthenium layers may exhibit roughness or stress evolution if deposited under highly energetic conditions, while molybdenum can form interfacial reaction layers with beryllium if intermixing is not controlled. Accordingly, the same deposition controls — limiting energetic bombardment, moderating growth rates, and managing interface formation — are applied to maintain optical contrast and structural stability in Ru / Be and Mo / Be multilayers.58 50130535.1
[0199] In-situ metrology 59, including quartz crystal microbalances, optical monitoring, or plasma emission monitoring, may be employed to control layer thickness and uniformity across the mask surface. Deposition is typically carried out using cluster-type sputtering tools or inline mask blank coaters equipped with load-lock chambers, multiple sputtering stations, substrate motion control, and automated process sequencing, consistent with equipment and methods currently used for EUV multilayer mask blank fabrication.
[0200] Through application of these deposition approaches, beryllium-containing multilayer stacks may be formed with well-defined interfaces and stable optical properties suitable for high-reflectance operation at 11.3 nm, while remaining compatible with established thin-film deposition techniques.REFLECTANCE CHARACTERIZATION OF 11.3 NM PHOTOMASKS
[0201] The reflective properties of photomask embodiments disclosed herein may be characterized using extreme ultraviolet (EUV) reflectometry performed at a wavelength near 11.3 nm. Such measurements may be carried out using synchrotron-based EUV reflectometers or laboratory EUV reflectometry tools capable of providing narrowband radiation in the vicinity of 11.3 nm and accommodating angular and polarization-resolved measurements.
[0202] In a representative measurement configuration, a photomask sample is mounted in a vacuum or reduced-pressure chamber and illuminated with EUV radiation incident on the multilayer reflective stack at a controlled angle relative to the surface normal. The reflected EUV radiation is detected in a specular reflection geometry while the incidence angle is varied over a selected angular range, for example from approximately -20° to +20° relative to the surface normal. The reflected intensity is normalized to the incident intensity to obtain absolute reflectance as a function of angle.
[0203] To assess polarization-dependent reflectance behavior, measurements are performed with linearly polarized EUV radiation having a defined polarization state relative to the plane of incidence. In particular, reflectance data are acquired for both s-polarized radiation, in which the electric field is oriented perpendicular to the plane of incidence, and p-polarized radiation, in which the electric field lies within the plane of incidence. Polarization control may be achieved using beamline polarization optics, selectable polarization modes of the EUV source, or equivalent polarization-resolving measurement techniques.59 50130535.1
[0204] Reflectance data obtained as a function of incidence angle and polarization state provide polarization-resolved reflectance curves from which relevant performance metrics may be calculated. These metrics include the average of s-polarized and p-polarized reflectance, angular reflectance variation across the specified angular range, and the difference between s-polarized and p-polarized reflectance. The calculated values may be compared to specified thresholds to determine whether a given photomask structure satisfies defined performance criteria at approximately 11.3 nm.
[0205] Reflectometric measurements may be performed at one or more locations of the photomask to confirm uniformity of the multilayer reflective stack. Measured reflectance behavior may further be corroborated by optical modeling based on the known multilayer stack geometry, layer thicknesses, and complex refractive indices of constituent materials at the measurement wavelength, providing additional validation of the measured performance.PHOTOMASK ABSORBER LAYER EMBODIMENTS
[0206] A mask absorber is formed on the reflective multilayer mask as a patterned material layer that locally attenuates or phase-shifts incident EUV radiation to define the image contrast transferred to the wafer during lithographic exposure. Design targets for some 11.3 nm mask absorber implementations include the selection of materials with optical properties unique to the 11.3 nm wavelength so that when coated to a particular thickness, surface reflection is adequately reduced across the angles at which the mask is illuminated. Furthermore, the 11.3 nm radiation that travels through the absorber layer should interact with reflected light from the underlying layers such that it interferes to contribute to reduction effects throughout the angular range.
[0207] As described earlier, the optical properties of materials in the extreme ultraviolet and soft-Xray regions are strongly wavelength dependent as several elemental absorption edges lie within this range, such that a shift from 13.5 nm to 11.3 nm results in materially different absorption behavior and complex refractive index values that must be accounted for in optical design. Choice of the absorber layer or layers involves consideration of the necessary material refractive indexes and extinction coefficients at 11.3 nm, together with the necessary thicknesses to accomplish this. For normally incident light, the contribution from interference can be determined based on interference and absorption conditions. To take into account the angular ranges at the mask, the vacuum / 60 50130535.1absorber / mask stack can be treated with the full Fresnel / transfer-matrix formalism for complex refractive indices, then varying n, k, and thickness to enhance the calculated reflected intensity of the layer over the desired wavelength and angle range.
[0208] Because the underlying layers constitute a multilayer reflector rather than a bulk substrate, the effective reflection coefficients are a complex and angle-dependent quantity, where both the magnitude and phase of the reflectance change with angle and differ for s- and p-polarization, as demonstrated in FIG. 32. The absorber thickness and complex index therefore influence not only the surface reflection but also the coupling into and out of the multilayer cavity, so acceptable designs should be evaluated in terms of overall reflectance and phase over an angular band. Traditional reflectance formulas do not apply directly. The absorber layer should be evaluated with respect to the full stack response, including the multilayer’s angle- and polarization-dependent phase shift, rather than just based on its nominal n and k.
[0209] Furthermore, when the absorber is patterned rather than a continuous film, the optical response can no longer be treated as a simple layered stack because the 11.3 nm wavelength and illumination angles are comparable to the feature dimensions on the mask. In this regime, the mask behaves as a periodic or quasi-periodic optical structure, and diffraction, near-field coupling, comer rounding, line-edge geometry, and duty-cycle effects all contribute to the effective reflectance and phase of the reflected field. The local multilayer / absorber interaction depends not only on material n and k, but also on feature width, pitch, sidewall angle, absorber height, and aerial-image orientation and the response differs for s- and p-polarization as incidence angle increases.
[0210] Accordingly, absorber design should account for both material-stack effects and pattern-level electromagnetic behavior, typically using vector-based aerial image simulation to evaluate how the chosen absorber n, k, and thickness interact with the patterned geometry across the relevant angular and polarization distribution. In practice, candidate absorber properties can be selected to give acceptable continuous-film behavior over angle and refined in combination with realistic feature geometries to ensure that the patterned mask maintains the desired reflectance, phase, and image-transfer characteristics across the illumination range.
[0211] In addition to satisfying optical performance targets for refractive index, extinction coefficient, and thickness, suitable 11.3 nm absorber materials should also be manufacturable. This includes the ability to deposit the films uniformly over large-area substrates using processes such as61 50130535.1sputtering, while maintaining acceptable stress, adhesion, density, and surface roughness. The material must further be pattemable with anisotropic plasma etch chemistries (or through lift-off techniques) that provide adequate selectivity to the underlying multilayer stack and resist and must support critical-dimension control across fine mask features. Long-term stability under 11.3 nm exposure, thermal cycling, cleaning chemistries, and environmental handling are also important to ensure that the absorber does not degrade, oxidize, or change optical properties during mask use. Materials that meet both the optical design criteria and these practical integration constraints are therefore preferred candidates for 11.3 nm EUV mask absorbers.
[0212] As an example, with a 11.3 nm multilayer mask stack comprised of ruthenium and beryllium, an initial design point for an absorber may be with a refractive index in the approximate range of 0.90-1.0 and an extinction coefficient in the approximate range of 0.02-0.08, followed by thickness adjustment using a multilayer-aware optical model (e g., using transfer-matrix methods, TMM) that simultaneously evaluates s- and p-polarization over the target angular span. From that starting point, n, k, and thickness can be selected to trade angular uniformity, peak suppression, and manufacturability (e.g. physically realistic and attainable values). Furthermore, selection can be carried out through image evaluations via modelling and simulation of the conditions that the photomask is designed for (such as illumination wavelength, NA, CRAO, illumination conditions, and pattern sizes).
[0213] In accordance with some of the embodiments disclosed herein, the absorber layer may be formed from one or more materials selected based on their complex refractive index values at a wavelength of approximately 11.3 nm, as summarized for materials in Table 7. Depending on the selected material or combination of materials, the absorber may exhibit relatively high refractive index (n) and high extinction coefficient (k), high n and low k, low n and high k, or low n and low k characteristics. Materials or combinations may be used to achieve a desired level of attenuation and phase distribution of EUV radiation. In some embodiments, absorber properties are tailored through the use of composite or multilayer absorber stacks formed from two or more materials to achieve target n-k combinations not available from a single material alone.Element n kNi 0.955 0.050Sn 0.941 0.03550130535.1Zn 0.980 0.043Au 0.931 0.0233Ag 0.916 0.0179Al 0.997 0.027Ta 0.973 0.031Co 0.945 0.047B 0.979 0.0023Pt 0.927 0.0236Ti 0.967 0.0236Rh 0.924 0.0089Cr 0.948 0.026Mo 0.953 0.0046W 0.959 0.288Si 1.0058 0.0144Te 0.947 0.0390Table 7. Elements and complex refractive index values (n, k) at 11.3 nm
[0214] As shown in Table 7, a range of candidate constituents exhibit differing complex refractive index values, enabling selection of absorber properties across a broad n-k design space. For example, nickel (Ni) exhibits refractive index and extinction coefficient values of approximately (n, k) = (0.955, 0.050) and represents one of the more strongly absorbing materials in the table, followed by cobalt (Co) at approximately (0.945, 0.047), zinc (Zn) at approximately (0.980, 0.043), tin (Sn) at approximately (0.941, 0.035), and tantalum (Ta) at approximately (0.973, 0.031). These materials, as well as alloys, compounds, and combinations thereof, are representative of absorber compositions that provide relatively high refractive index and high extinction coefficient characteristics at 11.3 nm. In contrast, materials such as aluminum (Al) at approximately (0.997, 0.027), titanium (Ti) at approximately (0.967, 0.0141), and boron (B) at approximately (0.979, 0.00231) illustrate absorber options having high refractive index and comparatively lower extinction coefficient, which may likewise be used alone or in combination with one or more other materials, or from other materials not expressly listed in Table 7.
[0215] For optical constants of compound materials, alloys, or mixtures, the complex refractive index values also depend on the mass density of the film, as absorption and dispersion at a given wavelength scale with the number of atoms per unit volume. For higher-density films, the atomic concentration is greater, which increases both the real and imaginary contributions to the63 50130535.1refractive index, whereas a lower-density or porous film yields proportionally weaker interaction with the propagating field. Optical constants for compound films are computed by combining the elemental scattering factors according to the film’s stoichiometry and physical density, where a density representative of the thin-film microstructure produced by the deposition process is considered as part of the determination.
[0216] Several example embodiments of absorber layers for use with the present technology are described below.SILICON-CONTAINING ABSORBER LAYER EMBODIMENTS
[0217] At wavelengths below the silicon L-absorption edge of about 12.5nm, the optical response of silicon changes markedly. At 13.5 nm, elemental silicon behaves as a weak absorber (n=0.999, k=0.0018), but at 11.3 nm the real part of the refractive index rises above unity (n=1.006) and the extinction coefficient increases by almost an order of magnitude (k=0.014). This sharp increase in k means that silicon and Si-containing compounds can provide a substantially higher optical density at 11.3 nm than for longer EUV wavelengths for a given physical thickness.
[0218] In some embodiments, the absorber layer may comprise silicon-containing metal alloys or compounds formulated to take advantage of the increased absorption of silicon at 11.3 nm. In such combinations, silicon contributes additional absorption while the other components can provide high baseline optical properties and further structural stability and robustness. By adjusting the relative silicon ratio, the optical constants (n and k) of the absorber layer may be tuned such that k is increased, thereby supporting high EUV attenuation, allowing for thinner absorber fdms.
[0219] Metal silicides that form between silicon and the metallic elements discussed herein, including Ni, Co, Pt, Rh, Ag, Sn, Ta, Ti, Mo, and W, constitute a class of absorber embodiments for 11.3 nm EUV photomasks which behave differently than at other wavelengths. These silicides may be utilized to exploit their advantageous optical and mechanical properties, and to improve etch behavior relative to the pure metals. Representative silicide phases include, but are not limited to, NiSi, Ni2Si, NiSi2, CoSi, CoSi2, PtSi, PfeSi, RhSi, Rh2Si, MoSi2, Mo5Si3, Ti Si2, TisSi3, TaSi2, WSi2, and WSi, as well as mixed or metastable phases such as Si intermetallics and Si -rich silicide compounds with tailored composition, stoichiometry, and fdm morphology. Their microstructure64 50130535.1may be engineered to be nanocrystalline or amorphous depending on deposition method and thermal treatment.
[0220] Similar design considerations apply to metal-Si-nitride systems, such as TiSiN-, TaSiN-, WSiN-, CoSiN, NiSiN-, PtSiN, RhSiN-, RuSiN-, MoSiN- type films, which combine a high-Z metal with silicon and nitrogen to achieve a high extinction coefficient, a tailored refractive index, and favorable thin-film and plasma-etch characteristics suitable for 11.3 nm mask fabrication.
[0221] Additional opportunities for implementation arise in silicon-rich ceramic and hybrid systems. Absorber candidates include SiiN4, SiC, and SiON-type films driven toward Si-rich compositions, as well as metal-silicon-nitride or metal-silicon-oxynitride systems such as Ta-Si-N, Ti-Si-N, W-Si-N and Ru-Si-N an oxynitrides as well. Si N4 and SiC, when combined with high-Z metals in a single absorber layer, can provide both high attenuation and favorable mechanical and patterning properties. For example, a Ta-Si-N film with increased Si fraction may reach a target reflectance and phase behavior at reduced thickness compared to a purely metal-based nitride.
[0222] Silicon-containing absorbers may also be implemented in multi-element alloys, where silicon is introduced alongside a metal pair to tune optical constants and film stress. For example, the addition of silicon to alloy systems (for example, Ta-W-Si, Ru-Ta-Si, or Pt- W-Si-N) provides another degree of freedom as silicon increases absorption while simultaneously influencing density, intrinsic stress, and etch behavior.
[0223] Aluminum-containing materials also offer potential design space examples when combined with silicon, as aluminum can provide complementary benefits including reduced film stress, improved microstructural stability, and favorable density-to-absorption balance for thin absorber layers. Aluminum-silicon materials such as AlSixalloys, aluminum-silicon nitrides (AlSiN), and aluminum-rich silicides or mixed metal-aluminum-silicon compounds can enable tailoring of the effective optical constants to achieve strong 11.3 nm attenuation while maintaining controllable etch behavior and compatibility with chlorine- or bromine-based plasma patterning chemistries.
[0224] Silicides and silicon-containing materials may be used as single-layer absorbers or integrated into multi-component, graded, or laminated absorber stacks, where their optical and etch characteristics are combined with those of other metals or alloys to meet performance, stability, and fabrication requirements for 11.3 nm mask applications. Silicides and other silicon-containing65 50130535.1absorber approaches also provide significant advantages during plasma patterning. Tn halogen-based etch chemistries, the silicon component forms volatile species such as SiCh, enabling a chemically assisted etch mechanism that complements ion-driven removal of the metallic component. As a result, silicon-containing films typically etch more uniformly and at higher rates than refractory metals under comparable plasma conditions, while requiring lower ion energies. This reduces redeposition and the risk of damage to the underlying multilayer reflector.
[0225] Table 8 contains details of several silicides that may be used in mask absorber layers. The table lists stoichiometric formulas of thin films of such silicides, indicating densities and corresponding refractive index and extinction coefficient values at the 11.3 nm wavelength.Formula Density n kg / cm3CoSi25.30 0.990 0.0304Mg2Si 1.94 0.993 0.0171MoSi26.31 0.987 0.0163NbSi25.37 0.991 0.0141Ni2Si 7.20 0.974 0.0423PtSi 12.40 0.967 0.0217RhSi 9.60 0.959 0.0182Ru2Sh 6.96 0.978 0.0153TiSi24.02 0.992 0.0177Ta2Si 14.00 0.982 0.0301WSi29.30 0.990 0.0241ZrSi24.88 0.993 0.0129Table 8. Density, refractive index, and extinction coefficient values for various silicides at 11.3 nm.
[0226] Evaluation of the reflective mask structures for 11.3 nm lithography has been carried out for several mask absorber examples, coated over Ru / Be multilayers on a fused silica (i.e. amorphous SiCE) substrate. Lithographic imaging performance of the reflective 11.3 nm mask structures was evaluated using transfer-matrix-method multilayer modeling and vector aerial image simulation with electromagnetic mask modeling. The mask is represented as a multilayer reflective stack with patterned absorber features. Image simulations incorporate three-dimensional mask and66 50130535.1multilayer effects, capturing the interaction of incident radiation with absorber and multilayer structures across the relevant angles and polarization states.
[0227] The modeling approach determines the near-field electric-field and magnetic-field distributions at and within the mask by decomposing the patterned absorber and multilayer stack into harmonic field components and solving the coupled electromagnetic modes throughout the structure. The resulting reflected and diffracted fields are then propagated through an optical system model to obtain the aerial image intensity and phase distribution at the wafer plane. This rigorous treatment captures vector-field effects, polarization dependence, absorber thickness and multilayer phase response. This simulation methodology allows for the evaluation of absorber material choices, optical constants (n and k), thickness selections, multilayer interactions, and pattern geometries under realistic illumination conditions, enabling predictive assessment of mask performance and image-transfer characteristics for 11.3 nm lithography.
[0228] FIG. 34 and FIG. 35 are plots of lithographic simulation results for several example absorber layers comprised of various material combinations and stoichiometries. Particular imaging conditions for these scenarios included a wavelength of 11.3 nm, a numerical aperture of 0.63, off-axis dipole illumination with a center sigma of 0.60 and a radius sigma of 0.05, an anamorphic optical reduction of 4X in the x-direction and 8X in the y-direction, a chief ray angle at the object (CRAO) of 6°, and using the Ru / Be multilayer stack of FIGS. 18 and 19. The mask geometry design consists of 7.5 nm width lines and a 15 nm pitch. Imaging was carried out using s- and p- polarized illumination for absorber thicknesses ranging from 0 to 60 nm and image intensity values were recorded. Image contrast (C) was calculated as the difference between maximum intensity and minimum intensity values divided by their sum (i.e. [Imax-Imin] / [Imax+Imin]). Image contrast is plotted for each absorber case as a function of absorber layer thickness.
[0229] Among the materials plotted in FIG. 34 and FIG. 35 are the silicides Ni2Si (0.974, 0.0423), Ta2Si (0.982, 0.0301), WSi (0.990, 0.0241), PtSi (0.967, 0.0217), and CoSi2. These materials are chosen by way of example and not limitation of silicide mask absorber layers for 11.3 nm. These silicide films exhibit layer thicknesses that meet a contrast-desirability criteria of 0.60 in thickness ranges of about 20nm to 60nm.REFRACTORY-METALS AND COMPOUNDS FOR ABSORBER LAYER EMBODIMENTS67 50130535.1
[0230] In some embodiments, absorber materials include refractory metals such as tantalum (Ta), titanium (Ti), molybdenum (Mo), and tungsten (W). These materials possess comparatively moderate to large refractive index values at a 11.3 nm wavelength, in particular Ta (n=0.973), Ti (n= 0.967), Mo (n= 0.9527), and W (n=0.9593), combined with extinction coefficients that range from relatively low (Mo, Ti) to moderate (Ta) to relatively high (W). Refractory metals exhibit excellent mechanical stability and are compatible with high-temperature processing, making them suitable for multilayer stacks and for absorber layers that may undergo thermal excursions during fabrication or exposure. Refractory metals may be alloyed with more chemically active elements to improve etchability, microstructure, and absorption. For example, Sn-containing refractory-metal alloys (such as Mo-Sn, W-Sn, Ti-Sn, or Ta-Sn) benefit from the formation of volatile SnCU during plasma exposure, enabling hybrid chemical-and-sputtering etch behavior and uniform anisotropic profiles. Similarly, incorporation of Ni or Co into a Ta or W matrix can stabilize grain structure or increase mechanical hardness while modifying etch response, enabling tailored absorber behavior.
[0231] Oxide derivatives of refractory metals, including Ta2Os, TiO2, MoOx, and WOX, may be used as absorber materials or as sublayers within composite absorber stacks for 11.3 nm masks. These oxides exhibit modified refractive index and extinction coefficient values relative to their metallic counterparts and provide enhanced chemical and thermal stability under exposure. Since metal oxides may exhibit reduced etchability in halogen-based plasmas, they may be incorporated in combination with more etchable metallic layers.
[0232] Metal nitrides, particularly TaN, TiN, WN, and MoN, represent another class of absorber examples, offering high density, tunable optical constants at 11.3nm, and favorable mechanical properties. Patterning of nitride layers typically proceeds via halogen-based plasma etching, where chloride formation combined with ion-assisted desorption enables material removal.
[0233] Boride compounds, such as TiB2, M0B2, and WBX, provide extremely high density, thermal stability, and distinct optical properties at 11.3nm. Due to strong chemical bonding and the formation of low-volatility boron halides, borides may be implemented in graded compositions or alloyed systems.
[0234] Metal carbides and carbon-containing compounds represent a further class of absorber examples for 11.3nm photomasks. Carbides of refractory metals, including tantalum carbide (TaC), titanium carbide (TiC), molybdenum carbide (MoC or M02C), and tungsten carbide68 50130535.1(WC or W2C), combine high density, strong covalent bonding, and good thermal and structural stability. In addition, carbon may be incorporated into absorber layers comprising nickel, cobalt, platinum, rhodium, silver, tin, zinc, chromium, or aluminum to form metal-carbon compounds or carbon-containing alloys. These materials exhibit refractive index and extinction coefficient values at 11.3nm that differ from their corresponding pure metals and nitrides, enabling additional flexibility in absorber thickness and phase tuning.
[0235] Among the materials plotted in FIG. 34 and FIG. 35 are TaN (0.966, 0.02709), PtiMo (0.9235, 0.0217), and PtW (0.9406, 0.0277), by way of example and not limitation of refractory metal compound mask absorber layers for 11.3 nm. Tantalum nitride, platinum-molybdenum, and platinum-tungsten films exhibit layer thicknesses that meet a contrast-desirability criteria of 0.60 in thickness ranges of about 20nm to 60nm.NICKEL-CONTAINING ABSORBER LAYER EMBODIMENTS
[0236] In some embodiments, mask absorbers comprise nickel (Ni), which exhibits strong absorption with extinction-coefficient values at 11.3nm on the order of 0.05. Halogen plasma etching of Ni produces halides such as NiCE, NiBr2, and NiF2, which exhibit low volatility that generally involves ion-assisted sputtering. To retain the favorable absorption properties of Ni while enabling pattern transfer, Ni may be alloyed with elements having appreciable extinction coefficients at 11.3nm and that benefit etch behavior. Incorporation of Sn, Co, Zn, or Ta into Ni can etch volatility through formation of more volatile or mixed-halide species. By using Ni as the principal absorbing species and incorporating additional elements that modify film structure, chemical reactivity, and etch behavior, Ni-containing absorber materials may be implemented to satisfy 11.3nm absorber requirements.
[0237] Tin forms intermetallic phases with Ni, including NiiSn^ NisS , and NiiSn, which can produce uniform nanocrystalline films with stable grain size. Cobalt (Co) forms continuous solid solutions with Ni, producing fine-grained films with improved ductility. Zinc (Zn) may promote nanocrystalline or metastable phases and enhance etch responsiveness through formation of volatile zinc halides. Tantalum (Ta) contributes amorphous-forming capability and stability. Etching of Ni-Sn, Ni-Ta, Ni-Co, and Ni-Zn absorber layers may be performed using halogen-based plasma69 50130535.1processes, in which volatile metal-halide species are chemically formed and removed. Suitable chemistries include Ch, HC1, BCE, HBr, or Br2.
[0238] Additional alloying elements may be incorporated intoNi-based fdms to adjust absorber performance at 11.3nm. Germanium (Ge) may increase amorphous-phase stability; boron (B) may promote fine-grained or amorphous structures; aluminum (Al) may form ordered intermetallics that stabilize microstructure; hafnium (Hf) and zirconium (Zr) may provide high-temperature stability; molybdenum (Mo) and tungsten (W) may increase density and modify optical constants; and indium (In) and tellurium (Te) may increase EUV absorption while influencing etchproduct volatility. These elements may be used individually or in combination with Ni to produce highly absorbing films with controlled microstructure and improved patterning characteristics.
[0239] Specific multicomponent alloy systems may also be implemented. Ni-Sn-Co alloys may combine intermetallic phase stability; Ni-Sn-Zn alloys may exploit zinc halide volatility to enhance etch behavior of Ni-Sn phases; Ni-Ta-B alloys may stabilize amorphous structures while maintaining 11.3nm EUV absorption; Ni-Co-W alloys may adjust 11.3nm optical constants while retaining mechanical stability; and Ni-Sn-In, Ni-Sn-Te, Ni-Zr-Al, or Ni-Hf-B alloys may provide increased absorption. These alloys allow tailoring of absorption, density, etch response, and stability while maintaining Ni as the primary absorption contributor.
[0240] Among the materials plotted in FIG. 35 is Ni3Sn4 (0.936, 0.0446), by way of example and not limitation of nickel- and tin-containing mask absorber layers for 11.3 nm. Nickel-tin intermetallic compound films exhibit layer thicknesses that meet a contrast-desirability criteria of 0.60 in thickness ranges of about 20nm to 60nm.TIN-CONTAINING COMPOUNDS AND ALLOY ABSORBER LAYER EMBODIMENTS
[0241] In some embodiments, tin is included in 11.3nm absorber materials, as it exhibits moderate refractive index (n=0.941) and relatively strong absorption at 11.3 nm. Sn also exhibits exceptionally favorable halogen chemistry. Upon exposure to Cl radicals, Sn readily forms SnCL and related species that are volatile relatively low temperatures. This volatility enables a plasma etch mechanism dominated by chemical removal. In alloy systems such as Sn-Ni, Sn-Co, Sn-Pt, or Sn-Rh, the presence of Sn enables chemical renewal of the etch surface through formation and desorption of SnCk As the chemically active Sn is removed, the underlying alloy becomes70 50130535.1accessible to ion-enhanced removal of the inert component, enabling for anisotropic material removal.COBALT-CONTAINING COMPOUND AND ALLOY ABSORBER LAYER EMBODIMENTS
[0242] In some embodiments, cobalt is included with other materials as absorber layers as it exhibits moderate refractive index (n=0.945) and relatively strong absorption at 11.3 nm. Cobalt may be combined with other transition metals to form binary or ternary alloys such as Co-Ni, Co-Fe, Co-Cr, Co-W, Co-Mo, Co-Ta, Co-Ti, and Co-Ru, which provide increased absorption relative to elemental cobalt while enabling adjustment of stress, grain structure, and etch behavior through compositional control. For example, Co-Ni and Co-Fe alloys can yield dense, fine-grained films with higher extinction coefficient and improved mechanical stability; Co-W and Co-Mo alloys can increase mass absorption and hardness while suppressing crystallization; and Co-Cr alloys may provide oxidation resistance and favorable plasma-etch selectivity. In other embodiments, cobalt may be incorporated into metal-nitride or metal-oxynitride systems, including CoN, Co-Ti-N, Co-W-N, and Co-Ta-N, where the inclusion of nitrogen increases amorphous character and index-tuning capability while maintaining high 11.3 nm attenuation. Cobalt may further be deployed in multilayer or composite absorber stacks, for example Co-based alloys paired with higher-k metals or high-density refractory constituents, to achieve target n-k combinations and absorber thicknesses not attainable with a single-phase material.
[0243] Among the materials plotted in FIG. 34 is WCo (0.9403, 0.0466), by way of example and not limitation of cobalt-containing mask absorber layers for 11.3 nm. Tungsten-cobalt alloy fdms exhibit layer thicknesses that meet a contrast desirability criteria of 0.60 in thickness ranges of about 20nm to 60nm.ZINC-CONTAINING ABSORBER LAYER EMBODIMENTS
[0244] In some embodiments, the absorber layer includes zinc-containing materials. Zinc exhibits relatively high 11.3 nm absorption in this spectral region, with optical constants of approximately n=0.98 k=0.043. The incorporation of zinc into alloys, compounds, or multicomponent thin fdms allows the extinction coefficient to be tailored while maintaining manufacturability, film stability, and compatibility with plasma etch processes used in mask fabrication.71 50130535.1
[0245] Representative zinc-containing absorber materials include zinc-transition-metal alloys such as Zn-Ni, Zn-Co, Zn-Mo, Zn-W, and Zn-Ta, in which zinc is combined with higher-Z metal’s absorption. In some implementations, the absorber layer comprises binary or ternary zinc compounds, including zinc oxides, nitrides, and oxynitrides. In other embodiments, the absorber layer comprises zinc chalcogenide or mixed-anion compounds, including zinc sulfide (ZnS), zinc selenide (ZnSe), zinc telluride (ZnTe), and related zinc chalcogenide variants including Zn-rich chalcogenides and mixed Zn-metal chalcogenides. These compounds incorporate zinc as the primary absorbing species, while the chalcogen component enables thin-film stability, tunable stoichiometry, and compatibility with sputter deposition processes.
[0246] Among the materials plotted in FIG. 34 is ZnTe (0.960, 0.0370), by way of example and not limitation of zinc-containing mask absorber layers for 11.3 nm. Zinc telluride films exhibit layer thicknesses that meet a contrast-desirability criteria of 0.60 in thickness ranges of about 20nm to 60nm.LOW-INDEX METALLIC ABSORBER LAYER EMBODIMENTS
[0247] In some embodiments, certain metallic elements exhibiting comparatively low real refractive index values at 11.3 nm and may be used as absorber materials or absorber constituents. Representative examples include platinum (Pt), rhodium (Rh), and silver (Ag), which exhibit refractive index values in the approximate range n=0.91-0.94, lower than many other metallic absorbers at this wavelength. Their extinction coefficients span a wide range, from relatively low values (e.g., Rh) to comparatively high values (e.g., Ag), enabling absorber designs that balance attenuation efficiency, physical thickness, and phase behavior. These materials may be used individually or as alloying components, and they also influence mechanical properties, microstructure, and etch behavior relevant to EUV mask manufacturability.
[0248] Pt films form dense, nanocrystalline or amorphous structures when deposited at low substrate temperatures, minimizing surface roughness and scatter. In halogen-based plasmas, Pt forms largely nonvolatile halides, resulting in etch behavior dominated by ion-assisted sputtering. To mitigate sputter-dominated etching, Pt may be incorporated into alloys with halogen-reactive elements such as Sn, which form volatile reaction products that chemically refresh the etch front and reduce ion-energy requirements.72 50130535.1
[0249] Rh films deposited by sputtering are smooth, oxidation-resistant, and structurally stable under 11.3 nm exposure. Like Pt, Rh forms nonvolatile halides in halogen plasmas, requiring ion-enhanced removal in halogen etch chemistries. Alloying Rh with more reactive elements such as Sn or Zn enables hybrid chemical-physical etch mechanisms, improving anisotropy and patterning efficiency.
[0250] Silver-based absorbers provide the lowest refractive index of the group (n=0.916) and a comparatively high extinction coefficient, allowing efficient attenuation with reduced thickness. In halogen plasmas, Ag forms moderately stable halides removable under ion bombardment. Ag may be alloyed with other metals to improve etch behavior, microstructural stability, and stress control while preserving favorable optical properties.
[0251] Among the materials plotted in FIG. 34 and FIG. 35 are PbMo (0.9235, 0.0217), and PtW (0.9406, 0.0277), by way of example and not limitation of low-index mask absorber layers for 11.3 nm. Platinum-molybdenum and platinum-tungsten films exhibit layer thicknesses that meet a contrast-desirability criteria of 0.60 in thickness ranges of about 20nm to 60nm.ABSORBER THICKNESS AND REFLECTANCE PERFORMANCE
[0252] In the image contrast vs. layer thickness plots of FIG. 34 and FIG. 35, an image contrast criterion is set at a value of 0.60, where values above this amount are considered desirable. The responses show that image contrast values increase with increasing thickness, which is a function of attenuation by the absorber layers. Additionally, oscillations in the image contrast response occur due to interference effects between the reflections above and below the absorber layers. Contrast maxima nodes correspond to destructive interference conditions within the absorber layer. There are approximately ten such nodes within the thickness range of 0 to 60 nm. These nodes represent preferred coating thickness values for particular absorber layers, where maximum local image contrast occurs. For 11.3-nm mask absorbers, a range of approximately 20-60 nm is desirable. In this range, a balance of attenuation, etch robustness, and imaging performance can be achieved. Films thinner than roughly 20 nm may not provide sufficient attenuation and can suffer from poor etch selectivity and process stability. Conversely, absorbers thicker than about 60 nm can increase mask tel ecentri city errors and image x / y asymmetry arising from oblique illumination, which degrade aerial-image quality.73 50130535.1
[0253] Thickness values in the range of 20 to 60 nm have been tabulated for each example absorber layer corresponding to maximum contrast nodes number four through ten, as shown in Table 9. Each of the examples provides multiple selection choices to achieve image contrast values of 0.60 and above.Absorber Thickness (nm) for Contrast Node No.4 5 6 7 8 9 10Table 9. Absorber thickness values for several example absorber layers at image contrast nodes.
[0254] In reflective 11.3 nm photomasks, the absorber layer is configured to reduce the reflected intensity from absorber covered regions relative to adjacent open areas of the underlying multilayer reflector, thereby establishing the reflectance contrast needed to define the patterned image. As described earlier, the absorber is preferably thin but sufficiently thick to be compatible with pattern transfer and plasma-etch processing. For 11.3nm EUV reflective masks, absorber designs may be targeted to produce a reflectance of less than about 10% across an angular range, providing a sufficient reduction relative to the open multilayer reflectance. In some cases, absorber designs may be targeted to produce reflectance below about 6% and in some cases below about 2.5%. In the embodiments disclosed herein, the thickness of the absorber layer is selected to achieve comparable or improved reductions in reflected intensity, taking into account both the intrinsic absorption of the absorber material (driven primarily by extinction coefficient) and interference effects arising from reflections at interfaces above and below the absorber layer (driven primarily by refractive index).74 50130535.1
[0255] Absorber layer reflectance targets can also be determined based on requirements for image pattern transfer and fidelity. This can be carried out using the concept of a Contrast Transfer Function (CTF), defined as:r'-ImageCTF = - - '-Maskwhere Cimage and CMask are the contrast values of the aerial image and the mask respectively. A CTF describes how effectively the contrast from the mask is transferred to the aerial image formed by the projection system and reflects how optical effects such as diffraction, polarization dependence, numerical aperture, illumination conditions, and mask pattern structure can reduce the contrast that reaches the image plane. A CTF value closer to 1 indicates that the imaging system preserves most of the mask contrast, while lower values indicate greater loss of contrast through the lithographic imaging process. Since the CTF metric evaluates aerial image fidelity, resist effects are not involved. In practice, CTF values may fall between 0.7 and 0.9, with lower values for patterning near the Rayleigh resolution limit and for non-periodic structures. By assuming a CTF of 0.8 and using an image contrast (Cimage) of 0.60, a mask contrast (CMask) of 0.75 would suffice. If a 70% reflective multilayer is considered, an absorber reflectance less than or equal to 10% would be needed (or 14% of the reflective multilayer), and other multilayer reflectance value assumptions would alter the calculation. If a CTF of 0.70 is utilized for an image contrast of 0.60 or more, an absorber reflectance less than or equal to about 5.8% would be needed (or about 8% of the reflective multilayer reflectance). For a CTF of 0.70 and a minimum image contrast of 0.65, an absorber reflectance less than or equal to about 2.5% would be necessary (or about 3.6% of the reflective multilayer reflectance). Table 10 shows the necessary minimum thickness values of the example absorber layers to achieve reflectance values of no more than 10%.Thickness (nm)for R< 10%50130535.1Table 10. Minimum thickness requirements of absorber layers to achieve reflectance values of or below 10%.
[0256] Unless otherwise indicated, the reflectance thresholds in Table 10 refer to averaged-polarization reflectance of absorber-covered regions evaluated at approximately 11.3 nm over incidence angles from -10° to +10° relative to the surface normal, using the Ru / Be multilayer stack described herein.
[0257] Representative example plots of reflectance responses for three selected absorber layers are shown in FIGS. 36 through 38. In FIG. 36, s-polarized, p-polarized, and averaged polarized reflectance is plotted vs. incident angle for a 33 nm WCo absorber layer coated over the aperiodic Ru / Be multilayer of FIGS. 18 and 19. As seen, the reflectance at normal incidence is low at about 1.1% and rises for both polarization states to about 1.6% at 10°. FIG. 37 shows the averaged reflectance for Sn-tNis absorber layers coated to thickness values between 27nm and 51nm. In all cases, reflectance remains below about 5.8% for angles up to 10°, with thickness of 39 nm, 45 nm, and 51 nm remaining below about 2%. FIG 38 shows the averaged reflectance for Ta2Si absorber layers coated to thickness values between 37 nm and 55 nm. In all cases, reflectance remains below about 5% for angles up to 10°, with thicknesses of 49 nm and 55 nm remaining below about 2%. These representative, non-limiting examples illustrate reflectance responses for selected absorber materials.DESIGN GUIDANCE DERIVED FROM ABSORBER EMBODIMENTS
[0258] Some absorber embodiments indicate that absorber performance at approximately 11.3 nm is governed by material-dependent absorption and optical-phase behavior that do not scale predictably from absorber systems developed for longer wavelengths. In particular, the disclosed results show that the relative balance between absorption, residual reflectance, and interference-76 50130535.1related phase response within the absorber layer changes at 11.3 nm, such that absorber compositions that are conventionally configured for 13.5 nm do not exhibit comparable contrast or optical behavior at the shorter wavelength.
[0259] In silicon-containing absorbers, some embodiments demonstrate that operation near the silicon absorption edge places the absorber in a substantially different absorption and energycoupling regime than at longer wavelengths. At 11.3 nm, changes in the real and imaginary components of the refractive index of silicon-bearing materials increase field attenuation within the absorber while also modifying the phase relationship between the reflected and transmitted fields. As a result, silicon-containing absorbers that function primarily as moderate-absorption amplitude masks at 13.5 nm may, at 11.3 nm, operate in a regime in which absorption and phase-shift effects act together to further suppress residual reflectance and support improved aerial-image contrast. These embodiments therefore indicate that silicon-containing alloys, silicides, and composite absorbers at 11.3 nm constitute distinct absorber-design opportunities rather than straightforward extensions of known 13.5 nm absorber systems.
[0260] Across the evaluated absorber materials, some embodiments further show that absorber thickness interacts with unique complex refractive index values at 11.3 nm in a manner that can produce composition-dependent interference conditions resulting in reflected intensity minima. In such cases, selecting absorber thicknesses near these interference-assisted minima can reduce residual reflectance without requiring excessive absorber thickness. The location and strength of these minima vary with absorber composition due to differences in angular dispersion and extinction behavior at 11.3 nm, and are not directly inferred from thickness-performance relationships established at longer wavelengths. This behavior provides practical design guidance for identifying absorber thickness regimes at 11.3 nm in which absorption and interference effects act cooperatively, rather than relying solely on absorption-dominated thickness scaling.
[0261] Taken together, these observations indicate that absorber evaluation for 11.3 nm involves coupled absorption-edge behavior, phase-shift response, and thickness-dependent interference effects that depend on absorber composition and cannot be extrapolated from established EUV absorber materials at longer wavelengths. The absorber design relationships identified in some disclosed embodiments therefore provide guidance for selecting absorber compositions and thickness regimes according to whether enhanced absorption, reduced residual77 50130535.1reflectance, controlled phase behavior, or interference-assisted contrast improvement is prioritized for a given 11.3nm photomask implementation.ABSORBER LAYER DEPOSITION METHODS
[0262] Absorber layers may be deposited using sputtering targets selected according to the desired absorber composition and optical properties at approximately 11.3 nm. Absorbers may be formed from elemental targets, pre-alloyed targets, or compound targets, depending on whether the absorber is intended to be a pure material, an alloy, or a compound such as a nitride, oxide, or silicide. In some embodiments, co-sputtering from two or more elemental targets is employed to produce alloyed absorbers with controlled composition, enabling tuning of refractive index, extinction coefficient, stress, and etch behavior. Compound absorbers may alternatively be deposited from stoichiometric targets or formed in situ by reactive sputtering using appropriate process gases. For example, nitrogen-containing absorbers may be formed using Ni / Ar mixtures, oxygencontaining absorbers using CL / Ar mixtures, and silicon-containing absorbers by co-sputtering metal and silicon targets. Target selection may also account for target erosion behavior, compositional uniformity, and compatibility with mask-scale uniformity requirements.PHOTOMASK PERIPHERAL EDGE BARRIERS
[0263] In certain embodiments, the photomask comprises a peripheral edge barrier configured to encapsulate exposed edge regions of a multilayer reflective stack. In masks employing beryllium-containing multilayers, the termination of the multilayer at the mask edge can result in exposed beryllium-containing layers that are not present in longer wavelength EUV Mo / Si masks. The peripheral edge barrier is positioned at the physical perimeter of the mask, including the beveled or chamfered edge region where the multilayer stack terminates and where the cross-sections of the multilayer layers are exposed. The barrier is provided to isolate these exposed regions and to reduce degradation or environmental interaction of beryllium-containing material during mask handling, storage, and exposure tool use.
[0264] An 11.3 nm photomask peripheral edge barrier embodiment is depicted in Figure 39. In this embodiment, the mask substrate 70 is coated with a beryllium-containing multilayer stack 71 and any other additional layers, along with an absorber layer 72, which may or may not be patterned. A barrier layer 73 is provided at the edge of the photomask to cover the beveled or chamfered mask78 50130535.1edges and the locations of termination of the multilayer stack 71 and extending along the surface of the photomask to remain outside of an active area defined as the patterned mask region or regions.
[0265] The barrier material 73 may be selected from chemically inert, low-permeability materials including, for example, silicon nitride (SisN^, silicon carbide (SiC), diamond-like carbon (DLC), aluminum oxide (AI2O3), or transition-metal nitrides. Such materials provide resistance to chemical interaction with hydrogen-containing environments, inhibit atomic diffusion from underlying layers, and improve mechanical robustness of the photomask perimeter during handling and transport.
[0266] In certain embodiments, the peripheral edge barrier 73 is further configured to remain chemically and mechanically stable under exposure to high-energy extreme ultraviolet radiation at approximately 11.3 nm and under the hydrogen-rich environments commonly present within EUV projection lithography systems. During operation, EUV exposure tools employ hydrogen or hydrogen-containing purge gases to mitigate contamination and to assist in in-situ cleaning of optical surfaces. At photon energies corresponding to 11.3 nm, interaction of EUV radiation with exposed materials can generate energetic secondary electrons and reactive species that may promote surface modification, chemical reduction, or erosion of unprotected materials at the mask periphery. The barrier layer 73 is therefore selected and deposited to provide resistance to hydrogen-induced chemical interaction and to suppress degradation or erosion under prolonged EUV exposure. In particular, oxide- and nitride-based barrier materials deposited by atomic layer deposition (ALD) or directional deposition provide stable, low-permeability interfaces that inhibit diffusion and chemical modification of underlying beiyllium-containing layers, thereby maintaining edge integrity and reducing the risk of material release or performance drift over extended tool operation.
[0267] In some embodiments, the peripheral barrier 73 comprises a conformal encapsulation layer deposited selectively along the perimeter of the mask so as to cover the exposed multilayer cross-section while remaining outside of the active patterned region. The barrier 73 is configured to seal the multilayer termination at the edge of the substrate while leaving the optical performance of the central region unaffected. The barrier 73 may extend over the bevel and partially onto the front and / or back surfaces of the mask to form a continuous sealing region around the multilayer edge, with the overlap distance selected to remain outside a defined mask active area.79 50130535.1
[0268] In some embodiments, the peripheral barrier 73 is formed using ALD, employed to deposit a conformal, pinhole-free coating over the exposed multilayer edge with controlled thickness and uniformity. The ALD process is applied in a perimeter-selective manner such that deposition occurs at the edge region while the active mask area is shielded or otherwise excluded from coating. This selectivity may be achieved through mechanical masking, fixturing that limits precursor access to the perimeter region, or by exploiting differences in surface accessibility between the edge and the central field.
[0269] To implement perimeter-selective ALD to a photomask substrate format, the ALD equipment and handling are preferably configured for non-wafer substrates and mask-style geometry rather than conventional semiconductor wafers. In some embodiments, the mask is processed in an ALD reactor configured for large or planar substrates and equipped with a fixture that defines a perimeter exposure region. The fixture may shield the central patterned field while exposing the bevel and sidewall to precursor flow, thereby enabling selective encapsulation of the multilayer termination. In other embodiments, precursor flow and conductance are controlled such that the edge region experiences substantially greater exposure than the central field. In still other embodiments, a spatial ALD configuration is employed in which precursor zones are spatially separated and the mask is translated relative to the precursor zones, with a defined aperture limiting deposition to the perimeter band.
[0270] The ALD barrier layer may be deposited after formation of the multilayer reflective stack either prior to or after the deposition of the absorber layer. In some embodiments, the barrier is deposited prior to absorber deposition so that the multilayer edge is sealed early in the fabrication sequence. In still other embodiments, a thin ALD sealing layer is deposited at an early stage, followed by a later reinforcement coating after subsequent processing.
[0271] The barrier material is selected to provide chemical stability and low permeability while remaining compatible with the multilayer materials and downstream processing. Suitable materials include aluminum oxide, silicon oxide, silicon nitride, silicon oxynitride, aluminum nitride, or multilayer combinations thereof. Aluminum oxide deposited by ALD is a candidate material due to its conformality and barrier properties. The barrier thickness may be selected in a range of approximately 2 nm to 50 nm, depending on the desired balance between encapsulation effectiveness, mechanical robustness, and stress control.80 50130535.1
[0272] In certain embodiments, the peripheral barrier 73 is implemented as a multilayer or nanolaminate structure comprising alternating oxide and nitride layers to enhance diffusion resistance and reduce defect propagation. In other embodiments, the barrier includes a thin ALD-deposited sealing layer combined with a thicker reinforcement layer deposited by a directional technique.
[0273] In other embodiments, the barrier layer or a reinforcement layer is formed using sputtering, ion-beam deposition, or evaporation in combination with a physical shadow mask or edge-shielding fixture. The shadow mask is configured to block deposition over the active patterned region while allowing deposition onto the perimeter, bevel, and sidewall regions of the mask. In some embodiments, the photomask is oriented at an oblique angle relative to the sputter or evaporation target such that the corresponding flux impinges primarily on the edge and bevel region of the mask. In some implementations, the reinforcement layer is deposited over a previously formed ALD sealing layer, such that the ALD layer provides chemical isolation of the multilayer termination while the reinforcement layer provides abrasion resistance and mechanical durability. The shadow-masked deposition may be performed as a dedicated perimeter process step and integrated into existing mask processing flows without modification to the active field.
[0274] The peripheral barrier 73 is thereby configured to suppress release of beryllium-containing material from the multilayer edge, to reduce degradation of the multilayer stack at exposed boundaries, and to improve mask robustness during handling, cleaning, and exposure. The barrier enables practical implementation of beryllium-containing multilayer systems in photomasks for 11.3 nm lithography by addressing edge-specific reliability and containment challenges without impacting optical performance.MASK BLANK PATTERNING METHODS
[0275] Once the multilayer reflective stack, capping layer, and absorber layer (along with any other layers) have been deposited onto the mask substrate, the resulting structure constitutes what is commonly referred to as a mask blank. In certain embodiments of the present technology, the absorber layer of the mask blank is patterned using electron-beam mask writing and subsequent pattern transfer, as depicted in Figure 40. As depicted, an exemplary mask blank 60 comprises a multilayer stack 62, an optional capping layer (not shown), and an absorber layer 63. A photoresist81 50130535.165, typically a chemically amplified e-beam resist, is spin-coated to a thickness between about 100 and 300nm on the absorber 63 (optionally over an underlayer and / or hardmask layer 64), soft-baked, and then exposed in an electron-beam mask writer to define the desired pattern 66. Hardmask layers 64 such as SiCE, SiNx, TiN, or multilayer hard-mask stacks may be employed to a thickness of about 20 to 60nm to maintain pattern fidelity during pattern transfer.
[0276] Following exposure, the resist is post-exposure baked (as applicable) and developed using a developer chemistry compatible with the resist system, commonly aqueous TMAH, to form a patterned resist image. The pattern 66 is then transferred into the absorber 62 (and any intermediate layers, if present) using anisotropic plasma etching, such as inductively coupled plasma (ICP) reactive ion etch (RIE), with etch chemistries selected based on the absorber composition and configured to maintain selectivity to the capping layer and underlying multilayer reflector.
[0277] A representative halogen-based plasma process suitable for etching mask absorber layers may be carried out in an ICP reactor configured for low-pressure anisotropic etching. Inside the reactor, the mask blank is mounted on an electrostatic chuck with backside cooling, and the chuck temperature is controlled anywhere in the range of 10-50 °C, typically near ambient, to ensure stable desorption of volatile species. The chamber pressure is maintained within 2-15 mTorr, as suitable for balancing ion directionality and chemical etch components. A feed gas mixture including chlorine and / or other halogen species and an inert dilution gas such as argon is introduced through mass-flow controllers, with acceptable flow ranges, depending on desired etch rate, sidewall passivation, and microloading control. The ICP source power may be operated in a power range to establish an adequate radical density, while the RF chuck bias may be set with a 13.56 MHz or an equivalent frequency, producing ion energies typically sufficient to activate the absorber film surface and promote ion-assisted formation and removal of volatile species. Selectivity to hard-mask materials may range from 2: 1 up to 8: 1 or more, depending on the relative bias, gas ratio, and hardmask thickness. Etch profiles with sidewall angles above about 80° are achievable by operating at sufficiently low pressure to maintain ion directionality and by tuning gas species to regulate passivation. Endpoint control may be achieved by time-based processing calibrated to the absorber thickness, by optical emission monitoring of chlorine and metal-halide emission lines, or by interferometric endpoint systems integrated into the reactor.82 50130535.1
[0278] After etch, the remaining resist 65 and any hardmask layers 64 are removed by plasma stripping and / or wet stripping, followed by post-etch cleaning steps to remove residues while preserving multilayer reflectivity.
[0279] Following patterning, the mask undergoes metrology and inspection to verify pattern fidelity, critical dimension (CD) control, overlay accuracy, and defectivity. Metrology may include CD-SEM, AFM, and scatterometry to measure linewidths, edge roughness, and absorber profiles, as well as film thickness and uniformity measurements for remaining layers. Defect inspection is performed using optical and electron-beam inspection tools, and in some instances, actinic inspection at EUV wavelengths is used to identify phase, amplitude, or multilayer defects that are not detectable with non-actinic methods. Based on inspection results, localized mask repair steps may be performed using focused ion beam or electron-beam-based techniques, followed by reinspection to confirm compliance with lithographic performance requirements. A pellicle may be subsequently mounted to the patterned mask to protect the mask surface from contamination during handling and exposure.11 3NM EUV PHOTOMASK PELLICLES
[0280] In certain embodiments, the 11.3 nm photomask assembly may further comprise an EUV pellicle positioned at a defined standoff distance from the patterned absorber surface to protect the mask surface from particulate contamination and chemical degradation. An 11.3 nm photomask pellicle is depicted in Figure 39. In this embodiment, the pellicle membrane 76 is mounted to a mounting frame assembly 74 and secured to the photomask surface over the substrate 70 and subsequent layers 71, 72, and (optionally) 73. The membrane 76 comprises a freestanding thin film that may be selected from a group consisting of beryllium, carbon, and silicon, including their various allotropes and compounds. The membrane 76 may utilize carbon-based architectures such as carbon nanotubes (CNTs), graphene, or amorphous diamond-like carbon (DLC). CNT-based membranes provide a porous yet mechanically superior framework capable of withstanding high temperatures in vacuum. Alternatively, silicon-based embodiments may utilize polyciystalline silicon or silicon-rich compounds like silicon carbide (SiC) and silicon nitride Sia
[0281] In some embodiments, in order to enhance the structural and thermal characteristics of the assembly, the membrane 76 may incorporate beryllium-rich compounds such as beryllium83 50130535.1carbide BeiC, beryllium borides Be?B or BeBi, and beryllium nitride BesNi to leverage the high transmission properties beryllium at the characteristic K-absorption edge.
[0282] In certain embodiments, the pellicle membrane 76 comprises a composite structure including a beryllium-containing layer encapsulated between one or more protective capping layers. The beryllium layer serves as the primary EUV-transmitting and load-bearing membrane and is selected for its low absorption and high stiffness at a wavelength of approximately 11.3 nm.However, beryllium is chemically reactive and susceptible to oxidation, hydrogen interaction, and surface modification under EUV irradiation. Accordingly, the beryllium layer may be encapsulated by thin capping layers that isolate the beryllium from the surrounding environment while preserving high optical transmission.
[0283] The pellicle encapsulating layers may comprise carbon, silicon, silicon carbide, silicon nitride, silicon oxynitride, boron carbide, alumina, or combinations thereof, and may be deposited with controlled thickness to balance mechanical reinforcement, chemical protection, and EUV transmission. In some embodiments, at least one encapsulating layer comprises a beryllium-containing compound layer, for example a beryllium nitride, beryllium oxide, beryllium fluoride, or mixed beryllium fluoride-nitride / oxide layer, optionally used as an interfacial conversion layer in combination with one or more non-beryllium diffusion-barrier layers. In certain embodiments, the encapsulation functions as a low-permeability diffusion barrier that suppresses oxidation of beryllium-containing layers, reduces interaction with hydrogen-containing environments that may be used during EUV exposure, and limits surface roughening, erosion, or material loss during prolonged irradiation. The encapsulation may be implemented as a single layer or as a multilayer or graded encapsulation stack (e.g., an adhesion layer in combination with a barrier layer), and may have a total thickness selected, for example, from sub-nanometer thicknesses to tens of nanometers, depending on the desired balance of protection, stress, and optical performance.
[0284] In certain embodiments, the capping layers are further selected to manage mechanical stress within the pellicle structure. By appropriate choice of material and thickness, the encapsulation can compensate intrinsic stress in the beryllium fdm and reduce membrane bowing or fracture during mounting, thermal cycling, or pressure differentials. The encapsulated structure also improves compatibility with pellicle frames and mounting adhesives by providing a mechanically stable and chemically inert interface.84 50130535.1
[0285] In further embodiments, the encapsulation is intentionally asymmetric across the membrane thickness such that a first capping stack on a tool-facing side of the beryllium layer is configured primarily for chemical and irradiation durability in a hydrogen-containing EUV environment, while a second, different capping stack on a reticle-facing side is configured primarily for particle suppression, handling robustness, and compatibility with cleaning and mounting processes. The two capping stacks may differ in material composition, thickness, density, or stress state, such that the composite membrane simultaneously achieves (i) reduced beryllium reactivity and permeation, (ii) controlled membrane stress and bow, and (iii) reduced particulate generation and / or transfer in the event of damage, while maintaining a target EUV transmission at approximately 11.3 nm.
[0286] In some implementations, the composite pellicle membrane is configured such that the encapsulating layers also act as diffusion barriers, inhibiting migration of beryllium and suppressing contamination of surrounding components. The resulting composite pellicle exhibits improved durability, reduced degradation under EUV and hydrogen exposure, and enhanced operational lifetime compared to an unprotected beryllium membrane. This architecture enables practical use of beryllium -based pellicles for 11.3 nm lithography while maintaining mechanical stability, environmental resistance, and high optical performance.
[0287] The mechanical integration of the pellicle is achieved via a mounting frame assembly 74, typically composed of a low-expansion material such as invar or a specific aluminum-silicon alloy, which is bonded to the non-pattemed border of the mask substrate. This mounting frame 74 maintains a standoff distance to ensure that any contaminant particles residing on the pellicle membrane 76 are maintained significantly out of the focal plane of the lithographic tool. For high-throughput operation, the mounting frame 74 may further include filtered venting ports 75 to allow for rapid pressure equilibration during vacuum cycling without allowing particulate ingress into the protected volume.
[0288] In addition to its optical and particulate-protection functions, in certain embodiments, the pellicle may further function as a secondary barrier for materials present within the multilayer reflective stack. Suitable membrane materials may include carbon-based films, graphene-containing structures, silicon-based membranes, or composite materials incorporating diffusion-limiting constituents.85 50130535.1
[0289] In certain implementations, the pellicle mounting frame 74 may include one or more pressure-equalization features configured to accommodate vacuum cycling during lithographic tool operation. In certain embodiments, the pellicle mounting frame 74 is specifically configured to prevent the transport or release of beryllium or beryllium-containing species from the pellicle assembly during operation of an EUV lithography system. Unlike conventional pellicle frames, which primarily provide mechanical support and pressure equalization, the frame described herein incorporates structural and material features that function as a containment barrier for beryllium while still permitting pressure equalization and thermal stability of the pellicle assembly.
[0290] In these embodiments, the pellicle mounting frame 74 includes one or more vent or pressure-equalization structures 75 that are engineered not merely to allow gas flow, but to substantially inhibit transport of beryllium in atomic, molecular, or particulate form. The vent structures 75 are configured with non-line-of-sight flow paths, such as tortuous channels, stepped apertures, or labyrinthine passages, such that any species attempting to migrate from the pellicle cavity must traverse multiple directional changes before exiting the frame.
[0291] In certain implementations, the vent pathways are further lined, coated, or filled with materials selected to suppress diffusion or adsorption / desorption of beryllium. Suitable materials include silicon nitride, silicon carbide, aluminum oxide, aluminum nitride, or other dense ceramic or ceramic-like coatings that are chemically stable under EUV exposure and hydrogen-rich environments. These materials act as diffusion barriers and capture surfaces, inhibiting migration of beryllium atoms or compounds that may be generated at the pellicle surface during prolonged EUV exposure.
[0292] When used in combination with the peripheral edge barrier structure 73, the pellicle and the edge barrier together provide a multilayer containment architecture that limits exposure of materials of the multilayer stack within the photomask assembly while maintaining compatibility with reflective projection lithography operation.11.3 NM PROJECTION LITHOGRAPHY USING PHOTOMASK STRUCTURES
[0293] In certain embodiments, reflective projection lithography is performed using extreme ultraviolet radiation having a wavelength centered at approximately 11.3 nm. EUV radiation at this wavelength may be generated using a suitable EUV source and conditioned by an illumination86 50130535.1system to provide controlled angular and polarization characteristics appropriate for high-numerical-aperture imaging. The conditioned EUV radiation is directed toward a reflective photomask comprising a substrate, a multilayer reflective stack disposed on the substrate, and a patterned absorber layer disposed over the multilayer reflective stack. The multilayer reflective stack includes alternating layers comprising beryllium and one or more materials selected from ruthenium, rhodium, molybdenum, palladium, niobium, or combinations thereof, and is configured to provide constructive interference for EUV radiation at approximately 11.3 nm. In representative implementations, the multilayer reflective stack is designed to exhibit high reflectance for both s-polarized and p-polarized radiation and reduced polarization-dependent reflectance variation over an incidence-angle range extending from approximately -10° to +10° relative to a surface normal of the photomask. EUV radiation reflected from the photomask within this angular range is collected and projected by a projection optical system configured for operation at approximately 11.3 nm and having a numerical aperture of about 0.50 or greater. The projected radiation forms an aerial image on a resist-coated substrate positioned at the image plane. Exposure of the photoresist layer to the projected EUV radiation induces spatially selective chemical or structural changes within the photoresist, and subsequent post-exposure processing and development steps are performed to form a patterned resist image corresponding to the pattern defined by the absorber layer on the photomask.
[0294] In some embodiments of the present technology, a projection lithography system operating at a wavelength of approximately 11.3 nm may be configured to image photomask patterns as shown in FIG. 41. The reflective lithography system comprises a condenser subsystem 41 and a projection lens subsystem 42 configured to support numerical apertures at and above 0.50, including but not limited to 0.50, 0.55, 0.60, 0.65, 0.70, 0.75, or higher, including intermediate values. The condenser subsystem may employ a sequence of reflective optical elements arranged to collect radiation from a 11.3 nm radiation source 45 and shape the angular and spatial distribution of that radiation prior to delivery to the reflective mask 43. These condenser elements may include collector mirrors, pupil-shaping surfaces, field-conditioning reflectors, and intermediate relay mirrors, each employing multilayer reflective coatings configured for 11.3 nm operation.
[0295] The condenser subsystem 41 may further incorporate field stop apertures, slitforming elements, or arc-shaping components configured to generate an elongated or arcuate87 50130535.1illumination region suitable for scanning exposure. The arrangement and number of mirrors used in the condenser are not limited to specific counts; examples may include implementations with three, four, five, or six mirrors, though systems employing fewer or greater numbers may also be used. The positions, shapes, and reflective geometries of such elements may be selected to define the appropriate source pupil fill, partial coherence level, and off-axis angle of incidence required for high-performance imaging at the mask 43.
[0296] The mask 43 may be fabricated on substrates having lateral dimensions of approximately 6 inches by 6 inches, which are compatible with existing mask-handling, inspection, and exposure tool infrastructures. However, the concepts disclosed herein are not limited to this format, and masks may alternatively be implemented using substrates of different sizes or aspect ratios to accommodate alternative tool architectures, exposure schemes, or manufacturing requirements.
[0297] The projection optical subsystem 42 may include a plurality of multilayer-coated mirrors configured to image the patterned reticle field onto a photoresist-coated substrate 44 with high fidelity at numerical apertures at or above 0.50. The projection optical subsystem 42 may be implemented with a variety of mirror counts, which are non-limiting and may include, for example, two-mirror, four-mirror, six-mirror, eight-mirror, or greater designs. The mirrors may be aspheric or freeform surfaces configured to control aberrations across an off-axis field and to maintain acceptable image quality across the entire arcuate or line-shaped exposure region. The specific number of mirrors used in the projection optical system 42 may be selected based on desired field size, NA, working distance, manufacturability, allowable obscuration, and thermal-mechanical stability considerations. Additional mirrors may be incorporated to further correct high-order aberrations or extend usable field area, while fewer mirrors may be employed in compact or simplified embodiments.
[0298] The projection mirrors may be fabricated from low-thermal-expansion substrates coated with multilayer stacks configured for 11.3 nm reflection over the angular range corresponding to NA values of 0.50 and above. The optical configuration may be telecentric or non-telecentric, and may be symmetric or anamorphic. In some anamorphic embodiments, the magnification in the scan direction may differ from the magnification in the non-scan direction, enabling larger effective NA while maintaining manageable mask feature sizes. Magnification ratios88 50130535.1may include 4 / 8X, 3 / 6X, or other suitable values that permit high-NA imaging while preserving lithographic performance and manufacturability.
[0299] The projection optical system 42 may include pupil-shaping or pupil-filtering features configured to control the angular spectrum of light delivered to the substrate. These may be implemented through selection of mirror shapes, mirror placements, or masks located at intermediate pupil planes. In various embodiments, the projection optics may maintain strict aberration balancing across the scan field to ensure uniform imaging contrast, phase stability, and linewidth control at sub -wavelength scales. The optical train may also incorporate structures or materials that mitigate thermal drift, mechanical deformation, or EUV-induced contamination, thereby maintaining focus and overlay stability throughout operation.
[0300] In certain implementations, it may be desirable for the condenser subsystem 41 to deliver radiation to the reflective mask 43 with a controlled or preferential polarization state.Polarization-selective illumination may be employed to enhance imaging performance for smallpitch features at a wavelength of approximately 11.3 nm, particularly in reflective systems where multilayer mirrors and mask diffraction can exhibit polarization-dependent behavior. Accordingly, one or more reflective elements within the condenser may be configured not only to shape angular and spatial distribution of illumination, but also to condition the polarization state of the delivered radiation.
[0301] One example for polarization conditioning in an EUV illumination path is to employ a multilayer reflective stack operating at or near a Brewster-angle condition for the p-polarized (TM) state. At incidence angles near the Brewster angle, the reflectance of the p-polarization component can be strongly suppressed relative to the s-polarized (TE) component. For 11.3 nm operation, a beryllium-containing multilayer stack, such as Ru / Be or Mo / Be, may be designed so that this polarization-selective condition occurs near an incidence angle suitable for a turning reflection, including near 45° with s-polarized reflection between about 65% and 77% with p-polarized reflectance less than about 2%. In such an implementation, the mirror acts as a polarization-selective turning mirror within the condenser beam path, producing a reflected beam that is preferentially s-polarized while attenuating the p-polarized component. The resulting degree of polarization and throughput may be selected by multilayer design (e.g., number of layer pairs, thickness ratio, periodicity, and interface engineering) together with the chosen operating angle and illumination89 50130535.1angular spread. An implementation of a polarizing component for use in the beam path of the illumination system is shown in FIG. 42. When unpolarized beams or circularly polarized beams 80 of EUV radiation impinges on polarizer 81, the resulting reflected beams 83 are s-polarized as the p-polarized is absorbed. For circular polarization, the phase relationship that created circularity is broken. A second turning mirror 82 is used to direct the beam so to be co-axial to the initial unpolarized beam direction. The second turning mirror 82 can also be a selective Brewster angle polarizing reflector. To change the state of polarization reaching the photomask, the assembly comprising the polarizer and the second turning mirror is rotated 90° so that the s-polarized beam is oriented to the plane of the photomask to provide illumination which is p-polarized relative to the orientation prior to rotation.
[0302] Because polarization selectivity is obtained through reflective multilayer behavior rather than transmissive polarizers, this approach is compatible with EUV wavelengths. The angular bandwidth of the polarization discrimination may be tailored through multilayer adjustments to balance polarization contrast against usable field angles in the condenser.
[0303] Together, the condenser subsystem 41 and projection optical subsystems 42 described herein provide a flexible and broadly applicable optical platform for lithographic imaging at 11.3 nm wavelength with numerical apertures at or above 0.50. The non-limiting examples of mirror counts enable a wide range of design embodiments, from compact multi-mirror projection systems to extended high-correction architectures incorporating a larger number of optical surfaces. Through appropriate selection of multilayer coatings, mirror geometries, and optical configurations, the disclosed systems are capable of delivering high-resolution, high-throughput imaging suitable for semiconductor manufacturing.11.3 NM RADIATION SOURCE EXAMPLES
[0304] In some embodiments of the technology, radiation is generated by an illumination source configured to emit radiation centered at approximately 11.3 nm with spectral bandwidth, angular distribution, and output power suitable for high-throughput projection lithography.Representative source architectures include laser-produced plasma (LPP) sources and free-electron laser (FEL) sources, either of which may be integrated with the lithographic system. These source types are capable of providing stable EUV emission in the 11.3 nm band with power levels90 50130535.1compatible with reflective multilayer optics and projection imaging at high numerical aperture. Other EUV source technologies may also be employed, provided that they produce radiation in the vicinity of 11.3 nm with characteristics sufficient to support lithographic exposure.
[0305] In an embodiment employing a laser-produced plasma source, radiation near 11.3 nm is produced by irradiating a target medium with high-intensity laser pulses to create a hot, dense plasma containing multiply ionized species. When xenon or other suitable high-Z target materials are used, successive ionization stages populate partially filled inner-shell electronic states that give rise to strong emission in the vicinity of 11.3 nm through dense groups of overlapping transitions, sometimes referred to as unresolved transition arrays. The resulting output exhibits a broad but centered emission band and is typically unpolarized due to the isotropic nature of spontaneous plasma emission. LPP sources may be implemented using gas-jet, droplet, or other target-delivery architectures, and may incorporate collector optics, debris-mitigation systems, and beamconditioning elements to shape and stabilize emission for lithographic use.
[0306] In an embodiment employing a free-electron laser source, radiation near 11.3 nm is generated by passing a relativistic electron beam through an undulator structure that imposes a periodic transverse motion on the electrons, resulting in the emission and coherent amplification of radiation at a wavelength determined by the undulator parameters and beam energy. FEL architectures suitable for 11.3 nm lithography may utilize superconducting linacs, energy-recovery linacs, or other accelerator configurations to produce high-brightness beams with low emittance and high average current, enabling narrow-band, high-power operation in the 11.3 nm region. The polarization state of FEL output is defined by the undulator magnetic configuration; for example, planar undulators may generate linearly polarized radiation, while helical or phase-controlled undulators may produce circular or elliptical polarization. FEL-based sources may be operated to supply kilowatt-class average optical power, continuous or pulsed exposure modes, and, in some implementations, distribution of radiation to multiple lithography tools.
[0307] Compact FEL sources, including systems based on energy-recovery linacs, advanced accelerator gradients, or short-period undulators, may be configured to generate coherent radiation at wavelengths near 11.3 nm with reduced facility footprint compared to conventional FEL installations. Such compact FEL architectures enable integration into semiconductor manufacturing91 50130535.1environments while retaining the high brightness, narrow bandwidth, and polarization control advantages characteristic of FEL-based sources.PHOTORESIST STRATEGIES FOR 11.3 NM EUV LITHOGRAPHY
[0308] In some embodiments, the photoresist used to create patterns when exposed using 11.3 nm reflective projection lithography and the photomasks according to some embodiments of the technology include organic-polymeric resists, for example chemically amplified resists (CARs) incorporating acid-generating groups and polymer backbones based on acrylates, methacrylates, cyclic olefins, or aromatic structures. Other suitable photoresists may include inorganic or hybrid metal-oxide photoresists (MORs), such as tin-oxide, hafnium-oxide, zirconium-oxide, aluminumoxide, or mixed-metal oxide systems formulated as solution-processable precursor films that convert to a metal-oxide network upon 11.3 nm EUV exposure. Additional photoresist chemistries may include molecular resists, hybrid organic-inorganic cluster resists, and negative-tone or positive-tone nanoparticle-based resists, provided that the composition exhibits adequate absorption, pattern transfer performance, and process compatibility for lithographic imaging at approximately 11.3 nm. TAILORING 11.3 NM PHOTORESISTS FOR STOCHASTICAL BEHAVIOR
[0309] It is realized that as semiconductor device dimensions scale toward the nanometer regime, stochastic variability becomes a fundamental limitation to pattern fidelity in EUV lithography. At such dimensions, lithographic performance is increasingly governed not only by classical optical resolution limits but also by the statistical nature of photon absorption, electron generation, and energy transport within the photoresist.
[0310] With 13.5nm EUV lithography, the corresponding 92 eV photon absorption by photoresist materials is dominated by valence-band and continuum electronic interactions. Although these interactions enable efficient photoacid generation, bond scission, or other photochemical change, they also produce stochastic energy deposition at nanometer length scales, which becomes increasingly problematic as critical dimensions shrink.
[0311] In contrast, lithography performed at a wavelength of approximately 11.3 nm occupies a distinct spectral regime. At this wavelength, and the corresponding photon energy of 110 eV, the incident radiation lies above the L2,3 absorption edge of silicon. This enables electronic92 50130535.1transitions in silicon that are not accessible at longer wavelengths and can alter the fundamental physics governing energy deposition within appropriately designed photoresist materials.
[0312] Several embodiments comprising photoresist compositions are described herein that exploit this wavelength-dependent behavior, with particular emphasis on silicon-containing materials and contrast these to organic materials and to materials with higher atomic number species traditionally employed in EUV lithography.
[0313] Energy deposition in EUV lithography arises from discrete photon absorption events followed by the generation and transport of photoelectrons and secondary electrons. The local deposited energy at position x within a photoresist volume may be expressed as:where / Vi s the number of absorbed photons, E,is the energy deposited per absorption event, and Kt(x) de scribes the spatial distribution of deposited energy resulting from the i-th event.
[0314] The variance of the deposited energy is given by:Var(E) = (N)Var(EK) + Var(V)<E / <)2
[0315] The second term represents photon shot noise, which scales inversely with exposure dose. The first term represents the intrinsic variance associated with the physical mechanisms of energy deposition following photon absorption. As feature dimensions decrease, the first term becomes increasingly dominant. In this regime, stochastic defects such as line-edge roughness, linewidth variation, and feature failure arise primarily from fluctuations in the spatial and energetic distribution of deposited energy rather than from insufficient average dose. Accordingly, materials and processes that influence the variance of energy deposition per absorbed photon play a central role in determining lithographic performance at advanced technology nodes.
[0316] The plot of Figure 43 shows the wavelength-dependent extinction coefficients of representative elements relevant to 11.3 nm EUV photoresist design, including silicon, aluminum, and carbon, as well as higher-atomic-number materials such as hafnium, zirconium, zinc, and93 50130535.1titanium. As seen, the absorption behavior of these materials differs markedly at 11.3 nm compared to 13.5 nm, with silicon exhibiting a pronounced increase in absorption below its L-edge, a characteristic that enables distinct material-level design strategies when considering energy deposition and stochastic effects in advanced EUV lithography.
[0317] The silicon I >,3 absorption edge corresponds to photon energies of 99-100 eV. At a wavelength of 11.3 nm, incident 110 eV photons possess sufficient energy to induce photoionization of the silicon 2p shell. This interaction is not available at 13.5 nm, where the 92 eV photon energy lies below the silicon L-edge. Following photoionization, the resulting core hole relaxes predominantly through non-radiative processes, including Auger decay. Because both the initial excitation and subsequent relaxation processes originate from discrete atomic energy levels, the resulting photoelectrons and secondary electrons exhibit energy distributions governed by atomic structure rather than by a broad continuum of electronic states.
[0318] As a consequence, energy deposition in silicon-containing materials at 11.3 nm occurs through a constrained set of excitation and relaxation pathways. Although secondary electron generation occurs, the statistical distribution of deposited energy is influenced by the limited number of accessible electronic transitions rather than by highly delocalized electronic processes. This behavior contrasts with that observed at 13.5 nm, where silicon does not undergo core-level excitation and energy deposition proceeds through valence-band and continuum interactions with broader energy distributions.
[0319] The spatial distribution of deposited energy in a photoresist is governed by the transport of photoelectrons and secondary electrons generated during absorption. The characteristic transport length depends on electron energy, material composition, and local density. At 11.3 nm, the constrained energy spectrum associated with silicon core-level excitation leads to electron transport behavior that is intermediate between highly localized deposition and highly delocalized cascades. The resulting energy deposition kernels are narrower and more reproducible than those produced by higher-energy or continuum-driven excitation. As feature dimensions approach the electron transport length scale, even modest variations in energy deposition can produce significant variations in chemical response. Reducing the variance of energy deposition therefore can directly improve pattern fidelity without requiring increases in exposure dose or change in resist thickness.94 50130535.1
[0320] In contrast to the behavior of Si at 11 ,3nm, high-Z materials such as hafnium, zirconium, and tin exhibit broadband continuum absorption without a corresponding edge feature in this energy range. At photon energies of 92 eV and 110 eV, these materials possess multiple accessible subshells and high densities of electronic states. Photon absorption in such materials may therefore proceed through a wide range of electronic transitions, producing photoelectrons and secondary electrons with broad energy distributions and extended transport lengths. While this increases absorption, it also increases the variance of energy deposition on a per-photon basis. As a result, although high-Z materials can reduce required exposure dose, they can increase stochastic variability by broadening the distribution of energy deposition events, particularly in small features where only a limited number of absorption events occur.
[0321] Aluminum occupies an intermediate position between silicon and higher-Z materials. Aluminum exhibits electronic structure features that differ from both silicon and heavier metals and can participate in EUV absorption through distinct mechanisms. At 11.3 nm, aluminum does not exhibit the same edge-proximate behavior as silicon but can nonetheless modify energy deposition characteristics relative to higher-Z materials. Aluminum-containing photoresists can therefore be used to tune absorption efficiency, electron generation, and spatial energy deposition.
[0322] The spatial distribution of deposited energy following absorption is therefore governed by the transport of photoelectrons and secondary electrons generated during these processes. The characteristic radial extent of this energy deposition, often referred to as the electron blur radius, determines the effective spatial resolution and stochastic behavior of the photoresist.
[0323] For silicon-containing materials at approximately 11.3 nm, core-level excitation and subsequent relaxation processes (including Auger decay) contribute to the secondary-electron population. This can lead to energy deposition behavior exhibiting a more constrained set of energetically allowed processes compared with purely continuum-dominated absorption. This behavior contrasts with higher atomic number materials which exhibit multiple accessible subshells and high Auger yields, producing broader electron cascades and energy deposition.
[0324] Figure 44 is a plot showing secondary-electron energy deposition radius (or secondary-electron blur radius) at 110 eV, representing Si, Al, and high-Z elements (for example Sn and Hf) incorporated as photoresist absorption constituents, together with carbon for comparison as a fully organic based photoresist material (e.g. a CAR). As secondary-electron cascades are95 50130535.1stochastic and material -dependent, normalized radial energy-deposition profdes provide for comparative insight. The plots represent the radial extent over which energy is deposited after photoionization, Auger decay, multiple inelastic scattering events, and low-energy secondary electron generation.
[0325] TPP-2M and related IMFP approaches provide representative electron transport length scales, which can be used to construct comparative deposition kernels. The plotted profiles are illustrative, normalized distributions intended to compare relative localization behavior across material classes, rather than to specify a unique measured transport function for a particular formulation. This is primarily determined by the inelastic mean free path (IMFP), estimated applying well-established semi -empirical approaches which are used in electron spectroscopy (known as Tanuma-Powell-Penn, or TPP, and version 2M in particular) to calculate IMFPs for electrons in solids. This accounts for material density, atomic composition, bandgap, plasmon energy, and electron kinetic energy to provide representative energy deposition data.
[0326] Analysis of representative radial energy deposition profdes is shown in Table 11 for various total energy amounts deposited within a given radius. For comparative characterization, cumulative-energy radii (e.g., rso, rso, rgo) and an energy-weighted rms radius (rnns) may be computed from the normalized deposition kernel to provide relative measures of localization and tail behavior. The rmsmetric represents the energy weighted average distance from the absorption center, giving greater weight to energy deposited farther away. The rrms metric is indicative of the long tail of secondary-electron transport, which can be correlated to line-edge roughness and feature failure.50130535.1Table 11. Radial spread metrics for energy deposited in several resist representations comprising Si, Al, C, and high-Z elements.
[0327] Analysis of representative radial energy deposition distributions shows that for silicon-based resists at 11.3 nm, approximately 50% of the deposited energy can be confined within about 2 nm of the absorption site while approximately 90% of the energy remains less than about 5 nm. High-Z materials such as tin or hafnium, while highly absorbing, exhibit substantially broader secondary-electron cascades due to their dense electronic structure and multiple available decay channels. Aluminum-containing resists exhibit narrower behavior because of their electronic structure, while still remaining more localized than high-Z systems. Also included in Table 11 are radial energy deposition metrics for Si in a resist formulation exposed to 92 eV photons. Radial spread metrics are significantly larger than for 110 eV. Notably, silicon exhibits low absorption at this energy, limiting the practical relevance to lithographic performance at 13.5nm.
[0328] As photon absorption at 13.5 nm occurs primarily through valence-band and continuum transitions, photoresist compositions generally exhibit broader initial electron energy distributions, longer inelastic mean free paths, increased secondary-electron scattering, and larger spatial energy deposition kernels. This is demonstrated in Figure 45, which shows plots of radial energy deposition profiles for Si containing resist at 110 eV compared carbon-based and high-Z based resists at 92 eV. The distribution of energies for the carbon and high-Z resists are significantly broader at 92 eV than the profile for Si resists at 110 eV, indicating the differences in the mechanisms involved and the likely impact to variance of deposited energy and subsequent stochastic variability. This is from absorption at 13.5 nm by these materials that proceeds through continuum electronic excitations, populating electron distributions to deposit energy over relatively larger distances, leading to broader radial profiles. At 11.3 nm, differences in photon energy and electron-matter interaction conditions can favor more localized energy dissipation near the absorption site for otherwise similar materials, resulting in comparatively narrower deposition profiles.
[0329] To provide a quantitative framework for comparing stochastic energy-deposition behavior across silicon-containing photoresist materials at 11.3 nm, a characteristic electrontransport length scale may be represented by an effective blur parameter G. G is derived from a modeled radial energy-deposition distribution constructed using photoelectron energy spectra97 50130535.1associated with silicon L2,3-shell ionization at photon energies above approximately 100 eV, together with material-dependent inelastic mean free path values and secondary-electron scattering behavior obtained from established electron-transport models commonly used in EUV spectroscopy and electron-energy-loss analysis.
[0330] As used herein, the term electron blur radius refers to a physical measure of the spatial extent over which energy is deposited following a single photon-absorption event, as determined by the transport of photoelectrons and secondary electrons within a material. Such measures may be expressed, for example, in terms of cumulative-energy radii (r50, r80, r90) or an energy -weighted root-mean- square radius derived from an electron transport model.
[0331] By contrast, the parameter G (sigma), as used in the context of stochastic lithographic performance, refers to an effective blur parameter representing the characteristic width of an energydeposition kernel used to describe the spatial variance of deposited energy relevant to pattern formation.
[0332] The silicon atomic fractions considered in this analysis correspond to ranges achievable in silicon-containing photoresist compositions suitable for lithographic imaging at 11.3 nm, including a variety of organic, inorganic, and hybrid material systems described in further detail below. Accordingly, the o values illustrated herein represent physically attainable regimes for photoresist materials at 11.3 nm rather than abstract or hypothetical compositions, and provide a basis for comparing relative localization of energy deposition prior to discussion of specific resist architectures.
[0333] For purposes of comparison, the silicon atomic fraction referenced herein is defined relative to a photoresist matrix otherwise suitable for EUV lithographic imaging, with silicon content introduced as a controlled compositional variable. A silicon atomic fraction of 0% corresponds to a baseline photoresist composition that is substantially free of intentionally added silicon and is representative of a conventional organic EUV photoresist matrix, such as a carbon-, oxygen-, and hydrogen-based polymer or molecular resist.
[0334] Increasing silicon atomic fraction corresponds to incorporation of silicon-containing species into the resist matrix through substitution, copolymerization, clustering, network formation, or infiltration, while maintaining film-forming behavior and lithographic process compatibility. Accordingly, the silicon atomic fractions represent physically realizable photoresist compositions98 50130535.1spanning from silicon-free organic resists (0 at% Si) to silicon-containing organic, hybrid, and inorganic-dominant resist materials having progressively higher silicon content, as described in subsequent embodiments. This framework enables comparative evaluation of energy-deposition localization as a function of silicon incorporation, independent of a specific resist architecture.
[0335] FIG. 46 is a plot of the effective blur parameter G as a function of silicon atomic fraction in representative photoresist materials for exposure at a wavelength of approximately 11.3 nm. As described earlier, the parameter G represents an effective one-sigma radial width of an energy-deposition kernel describing the spatial distribution of deposited energy arising from photoelectron and secondary-electron transport following absorption of 11.3 nm EUV radiation.
[0336] For purposes of comparison, c is determined from a normalized radial energydeposition distribution K r) such that:where r is the radial distance from a photon absorption site. The energy-deposition distribution K(r) is constructed using photoelectron energy spectra associated with silicon L2,3-shell ionization at photon energies above 100 eV, together with inelastic mean free path values and secondary-electron scattering behavior obtained from electron-transport models.
[0337] The silicon atomic fractions plotted correspond to compositions attainable in silicon-containing photoresist materials suitable for use at 11.3 nm, including a range of material architectures described in subsequent embodiments. The plotted trend shows a systematic decrease in c with increasing silicon at%, reflecting increased localization of deposited energy associated with silicon core-level excitation and constrained electronic relaxation pathways at 11.3 nm. For lithographic features having critical dimensions approaching G, reductions in G correspond to reduced spatial variance of deposited energy and reduced stochastic pattern variation, including lineedge roughness and feature failure probability.
[0338] At an exposure wavelength of approximately 13.5 nm, silicon-containing photoresist materials do not provide the same degree of localization of energy deposition observed at 11.3 nm. As a result, increases in silicon content at 13.5 nm primarily affect absorption magnitude rather than99 50130535.1the spatial statistics of energy deposition, and do not produce the same systematic reduction in o illustrated in FIG. 46.
[0339] The radial energy-deposition behavior described above is expected to manifest in improvements in resist imaging performance, particularly as critical dimensions approach nanometer length scales where stochastic effects dominate. More localized energy deposition following photon absorption can reduce the variance of deposited energy within the resist volume corresponding to lithographic features, thereby reducing fluctuations in the chemical response that drive line-edge roughness, linewidth variation, and pattern failure probability.LINE-EDGE ROUGHNESS AND BLUR
[0340] As feature dimensions approach the 20 nm pitch regime and below, line-edge roughness (LER) becomes one of the dominant contributors to edge-placement error and ultimately to device yield. At these pitches, even sub-nanometer edge fluctuations represent a significant fraction of the critical dimension. Current roadmap targets for leading-edge logic identify gate-related LER values on the order of roughly 2-3 nm and trending below 2 nm over successive generations. These values reflect how allowable stochastic variability decreases as pitch shrinks, independent of classical optical resolution limits.
[0341] For purposes of this disclosure, LER is characterized in practice using a three-sigma metric, LEFUG, obtained from post-development edge measurements. In a representative measurement procedure, a critical-dimension scanning electron microscope (CD-SEM) is used to acquire edge contours over a statistically meaningful evaluation length (for example over 1 micrometer or more) for a population of nominally identical lines (for example 30 or more). The local edge position is extracted along the line, deterministic components (such as systematic tilt or curvature) are removed using established fdtering conventions, and the standard deviation of the residual edge-position distribution is calculated. The reported LERsa value corresponds to three times this standard deviation. This metric directly reflects the probability of rare edge excursions that contribute to bridging, breaks, or missing features when feature dimensions approach the stochastic length scale.
[0342] The secondary-electron blur parameters described above provide a direct physical connection to this measurable LER. In stochastic imaging models, the spatial distribution of100 50130535.1deposited energy following photon absorption establishes an effective blur length scale, which may be represented by o derived from a normalized radial energy-deposition kernel. When feature dimensions become comparable to this length scale, LER scales with both (i) the magnitude of o and (ii) the local photon-count statistics within the feature volume. A reduction in o, particularly in the long-tail behavior characterized by rgo or rrms, reduces the variance of deposited energy within a feature and correspondingly reduces both continuous roughness (LERac) and rare stochastic failure probability.
[0343] In certain embodiments, the relationship between a characteristic blur parameter G and the resulting three-sigma line-edge roughness LERaa for a dense line-and-space pattern may be modelled using threshold-crossing stochastic imaging calculations.
[0344] To relate the characteristic lateral blur parameter o to measurable line-edge roughness, it is useful to adopt an analytical description that captures the dominant physics without depending on any particular resist chemistry or stochastic simulation framework. In the context of EUV lithography as described herein, o represents the effective lateral length scale associated specifically with redistribution of deposited energy following photon absorption, including photoelectron and secondary-electron transport, approximated by a Gaussian kernel having a standard deviation o in the plane of the resist. This G therefore corresponds to the secondary-electron energy-deposition blur kernel and not to other stochastic or process-related roughness sources.
[0345] In the frequency domain, such a blur kernel may reduce image modulation according to a transfer function of the form:G( ) = exp[— (2TT<7 )2],
[0346] where is spatial frequency. For dense line-space patterns, the fundamental spatial frequency is f = 1 / p, where p is the printed pitch. The presence of this secondary-electron blur therefore reduces the modulation at the fundamental frequency, and correspondingly reduces the image log slope (ILS, which is the slope of the log of the aerial image) at the resist threshold, by a factor101 50130535.1
[0347] The edge-placement variability scales inversely with the image log slope. At the same time, the stochastic fluctuation in the latent image attributable to energy-transport statistics scales with the square root of the characteristic secondary-electron blur length. Combining these two effects leads to a closed-form expression for line-edge roughness expressed at the three-sigma level
[0348] where o is the secondary-electron energy-deposition blur parameter in nanometers and K is a proportionality constant that captures resist sensitivity, exposure conditions, and the unblurred image log slope.
[0349] Values for the constant V may be determined from experimental observation or attainable results. For a given pitch, if a resist exhibits a measured three-sigma line-edge roughness LERjc* at a known secondary-electron blur value o’ , then K is obtained by rearrangement as
[0350] Once calibrated, this relationship provides a practical means for translating changes in the secondary -electron blur kernel into expected changes in line-edge roughness under otherwise comparable imaging conditions.
[0351] Measured line-edge roughness in practical lithography systems includes additional contributions beyond secondary-electron energy deposition. These may include photon shot noise, stochastic acid generation statistics, resist molecular inhomogeneity, development variability, mask edge roughness, projection optics contributions, and metrology noise. Collectively, these effects may be represented as a baseline stochastic roughness contribution that is independent of the secondaryelectron transport blur.
[0352] These contributions to total LER may be combined with the blur-dependent term:102 50130535.1
[0353] Here, LERja.baseiine represents the residual stochastic roughness arising from nonelectron-transport and LERSG.SEB represents the roughness arising from secondary electron blur.
[0354] The foregoing expression represents one suitable analytical model for relating the secondary-electron blur parameter o to line-edge roughness for dense periodic features. The specific functional dependence arises from a Gaussian representation of the secondary-electron energydeposition kernel and evaluation of the modulation transfer at the fundamental spatial frequency corresponding to the pattern pitch. Other mathematically equivalent or numerically calibrated formulations may be employed to relate a characteristic lateral transport length to edge variability. Regardless of the specific analytical form selected, the operative relationship remains that a reduction in the secondary-electron blur parameter G reduces the transport-driven component of lineedge roughness for a given pitch and imaging condition. The model therefore provides a framework for connecting secondary-electron blur to stochastic lithographic metrics.
[0355] In some embodiments, for a representative dense pattern having a pitch of approximately 20 nm, and using a realistic carbon-based (e.g. CAR, non-CAR, or molecular glass) type resist data point representing typical resist patterning performance in which a blur parameter of approximately 3.0 nm may correspond to an LERBC, of approximately 4.6 nm, the constant T may be determined and used to evaluate LERsc for other o values under comparable processing conditions. For these example values, K is determined as 1.0076 nm1'2, with dimensional units consistent with the relationship to LER dimensional units. Solutions of the expression enable for the estimation of the expected dependence of stochastic line-edge roughness on the secondary-electron blur parameter for a given pitch, thereby providing a quantitative bridge between material-dependent energydeposition behavior and measurable lithographic LER metrics. This example provides insights into the behavior that would be expected from other observed data, which can be determined using the method described.
[0356] FIG. 53 is a LERsa vs blur o plot for 20 nm pitch dense lines determined using the representative observed data points, of LERjn = 4.6 nm, o* = 3.0 nm, and a corresponding K value of 1.0076 nm1 2. The plot shows the reduction in LERBC with decreasing o resulting from increased Si incorporation, which includes other-sources LERsa, baseline contributions of 1.5 nm (which may represent current attainable levels) and 1.0 nm (which may be achievable with improvements in103 50130535.1these other sources). As shown, the achievable LER is reduced with decreasing blur, indicating the impact of Si content in photoresist formulations exposed at a 11.3 nm wavelength.
[0357] The comparative impact of Si incorporation into photoresist for 13.5 nm exposure can be assessed based on the radial spread metrics depicted in Table 11 for 92 eV exposure of resist comprising carbon (C -based) and silicon (Si-based) formulations. At 92 eV, the radial energydeposition metrics for C- and Si-containing resist representations are similar in magnitude. In particular, Table 11 shows rgo values of 9.4 nm for both cases, and only modest differences in rjo, r<?o, and rrms(6.4 nm for C-based and 5.6 nm for Si-based at 92 eV). These values indicate that, at 13.5 nm, the lateral extent of the secondary-electron cascade is not substantially altered by the presence of silicon.
[0358] These radial spread metrics can be related to an effective secondary-electron blur parameter G by mapping a kernel representation to the normalized radial energy -deposition profile. One practical approach is to take G proportional to the energy -weighted rms radius r™.s, with the proportionality determined from a known reference condition. For dense 20 nm pitch line / space patterning at 13.5 nm, a representative carbon-based chemically amplified resist may be characterized for example by using Gc.92 = 3.0 nm and a measured LERic, = 4.5 nm under defined process conditions. From Table 11, rrmsc,92is 6.4 nm, giving a proportionality factor of approximately 0.47. Applying this same mapping to the Si-based 92 eV kernel yields a slightly lower Gc,92 = 2.6 nm.
[0359] In practice, however, silicon does not lie above its L2 absorption threshold at 92 eV, and therefore does not introduce a dominant core-level absorption pathway. The absorbed energy at 13.5 nm remains governed primarily by continuum electronic interactions of the overall resist matrix. For realistic silicon loadings in hybrid or silicon-containing CAR formulations, the resulting secondary-electron transport characteristics remain close to those of the carbon-dominated baseline. Accordingly, under otherwise identical imaging, dose, and process conditions, it is appropriate to treat both a conventional C-based resist and the same resist with silicon incorporation as having approximately G ~ 3.0 nm at 13.5 nm. Using the calibrated LER model for 20 nm pitch, this corresponds to identical LERic values no better than 4.5 nm under otherwise identical conditions. Thus, for otherwise equivalent imaging and processing conditions, silicon incorporation would not lead to a meaningful reduction in calculated LERaa at 13.5 nm.104 50130535.1
[0360] In such representative stochastic imaging calculations incorporating the materialdependent secondary-electron blur parameter o derived from the normalized radial energydeposition kernel, silicon-containing photoresist compositions exposed at approximately 11.3 nm exhibit a reduced effective blur length scale relative to the same compositions exposed at approximately 13.5 nm, where silicon core-level excitation is not accessible. When modeled for dense line-and-space patterns having pitches at or below approximately 20 nm, and evaluated under otherwise comparable optical conditions, normalized image dose, and post-exposure processing parameters, this reduction in G corresponds to a lower calculated LERsa. Accordingly, for a given silicon-containing formulation, exposure at approximately 11.3 nm can produce a measurable reduction in LERsc relative to exposure at approximately 13.5 nm, reflecting the wavelengthdependent modification of per-photon energy-deposition statistics described herein.
[0361] While the foregoing carbon-based resist examples incorporating silicon exemplify the relationship between Si content and image roughness, the analysis captures the dominant physics without depending on any particular resist chemistry. Categories of resist materials in which silicon can be incorporated into include for example chemically amplified resists, backbone scissioning polymers, and metal-oxide, -oxo cage, si-oxo and si-metal-oxo frameworks, vapor-deposited, or hybrid resists.
[0362] It should be noted that the foregoing descriptions of photon-matter interaction, electron generation, and energy-deposition behavior are provided to aid in understanding representative embodiments and are not intended to limit the technology to any particular theoretical model or mechanism of operation.STOCHASTICS REDUCTION STRATEGIES AT 11.3 NM
[0363] For lithography at nanometer length scale, where stochastic effects arise from energy deposition fluctuations comparable to the characteristic electron transport length, imaging variability is governed not by average exposure conditions but by the statistical distribution of deposited energy. Stochastic failures arise when local energy deposition falls below or exceeds thresholds required to drive chemical reactions within the resist. The probability of such failures depends on the tails of the energy distribution rather than on its mean value. The primary variance term Var(£ / <) therefore plays a central role in determining stochastic performance. Materials that produce broader105 50130535.1distributions of energy deposition exhibit larger tails in this distribution and consequently higher likelihood of line-edge roughness, missing features, or pattern collapse.
[0364] At 11.3 nm, silicon-containing materials can allow for the modification of this behavior by constraining the excitation and relaxation pathways available during photon absorption. Because energy deposition arises from a limited set of atomic-scale processes, the resulting distribution of deposited energy is narrower and more reproducible than in materials dominated by continuum absorption. Importantly, this reduction in variance does not require increased dose or thicker resist films. Instead, it arises from modification of the fundamental physics of photon-matter interaction. As feature dimensions continue to decrease, the relative importance of per-photon energy variance increases, making this effect increasingly significant at advanced technology nodes.
[0365] At 13.5 nm, mitigation of stochastic variability is commonly pursued through a combination of process optimization and material selection, including the use of higher-Z elements to increase absorption efficiency and reduce photon shot noise. While such approaches can improve dose efficiency, the underlying energy-deposition process at 13.5 nm remains dominated by continuum electronic excitations that produce broad secondary-electron cascades. As a result, material selection at 13.5 nm primarily influences the quantity of absorbed energy rather than the statistical distribution of that energy on nanometer length scales.
[0366] Consequently, improvements in stochastic behavior at 13.5 nm are typically achieved through tradeoffs such as increased exposure dose, reduced resist thickness, or constrained processing windows, rather than through intrinsic modification of the per-photon energy-deposition mechanism. In contrast, at 11.3 nm, the interaction of EUV radiation with certain materials, particularly silicon-containing compositions, occurs in a regime where core-level electronic transitions and constrained relaxation pathways play a more significant role. This enables materiallevel control over not only absorption magnitude but also the spatial statistics of energy deposition, providing an additional degree of freedom for stochastic behavior that is not available at 13.5 nm.
[0367] Lithography at approximately 11.3 nm therefore permits a different class of enhancement strategies based on material selection and electronic structure. As a result of atomic-scale absorption processes, material composition can influence not only the average absorbed energy but also the statistical distribution of deposited energy at nanometer length scales. This enables strategies that are generally unavailable at longer wavelengths. Accordingly, enhancement at 11.3106 50130535.1nm may be performed through deliberate selection of silicon-containing materials, control of silicon bonding environments, and adjustment of inorganic network composition or density to influence the variance of energy deposition per absorbed photon. In some cases, other elements exhibiting low to moderate stochastic behavior, such as aluminum and titanium, may be combined with silicon-containing materials to balance absorption strength and stochastic performance, enabling further material -level control of resist behavior.
[0368] In certain embodiments, photoresist compositions are selected such that the variance of deposited energy per absorbed photon, Var(eK), at an exposure wavelength of 11.3 nm is reduced relative to that obtained at 13.5 nm under otherwise comparable exposure and processing conditions. This reduction may be realized by incorporating silicon-containing species at concentrations and bonding environments sufficient to enable silicon core-level excitation at photon energies above the silicon L2,3 absorption edge. In practice, candidate resist formulations may be evaluated by exposing test patterns at approximately 11.3 nm and assessing stochastic lithographic metrics, including lineedge roughness, linewidth variation, and pattern failure probability, as functions of silicon content, resist thickness, and exposure dose. A reduction in these metrics without a corresponding increase in dose or loss of image contrast indicates a reduction in per-photon energy -deposit! on variance enabled by the wavelength-dependent interaction described herein.
[0369] According to some embodiments, the resist thickness is selected relative to the characteristic electron transport length associated with silicon core-level excitation at photon energies above the silicon L2,s absorption edge, such that a substantial fraction of secondary-electron energy deposition occurs within the resist volume corresponding to a single lithographic feature. In practice, this may be achieved by selecting resist film thicknesses on the order of the estimated secondary-electron transport length for the resist composition at approximately 11.3 nm and evaluating lithographic performance as a function of thickness. Thicknesses that result in reduced line-edge roughness and decreased stochastic failure probability, while maintaining acceptable image contrast and pattern transfer performance, indicate effective confinement of secondaryelectron energy deposition within the resist. In this regime, resist thickness functions as a controllable design parameter for controlling stochastic behavior rather than solely for dose efficiency or pattern collapse margin.107 50130535.1
[0370] In certain embodiments, photoresist compositions are designed and evaluated using a modeling-guided development process in which stochastic lithographic performance is evaluated through simulation prior to experimental formulation. In such embodiments, material-dependent parameters including photon absorption probability, secondary-electron transport behavior, and effective energy -deposition blur metrics are modeled as a function of resist composition, including silicon atomic fraction and bonding environment.
[0371] In such modeling-guided embodiments, stochastic lithographic performance may be evaluated using lithography simulation frameworks that couple optical imaging models with material -dep endent exposure and stochastic modules. Optical imaging may be modeled using rigorous or semi-rigorous projection lithography approaches that compute aerial image intensity, phase, and polarization effects at the wafer plane for 11.3 nm exposure. These optical models may be combined with resist exposure models that describe photon absorption statistics, photoelectron generation, and subsequent secondary-electron transport within the resist volume.
[0372] Material-dependent energy-deposition behavior may be represented using electron transport models that account for photoelectron energy spectra associated with absorption at approximately 110 eV, inelastic mean free paths, secondary-electron generation, and scattering processes within the resist matrix. Such models may employ established techniques used in electron spectroscopy and EUV resist modeling, including Monte Carlo transport approaches, continuous slowing-down approximations, or kernel-based representations derived from inelastic mean free path calculations. The resulting spatial energy-deposition distributions may be reduced to one or more effective parameters, such as a characteristic blur parameter c or cumulative-energy radii, for use in stochastic analysis.
[0373] Stochastic lithography modules may then be applied to compute predicted lithographic variability metrics, including line-edge roughness, linewidth roughness, critical-dimension variation, or pattern failure probability, based on the statistical distribution of deposited energy and the resist chemical response. In these simulations, silicon content and bonding environment are treated as input variables that influence absorption probability, electron transport length scales, and the variance of energy deposition per absorbed photon. By sweeping silicon atomic fraction and resist thickness within the simulation framework, candidate design spaces may108 50130535.1be identified that minimize stochastic variability while satisfying constraints on image contrast, sensitivity, and process window.
[0374] Using such simulations, one or more stochastic performance metrics, such as effective blur parameter G, predicted line-edge roughness, linewidth variation, or pattern failure probability, may be determined for candidate resist compositions under representative exposure conditions. The silicon content, resist thickness, and material architecture may then be selected to achieve a target balance of stochastic performance, image contrast, sensitivity, and pattern transfer capability.
[0375] According to some embodiments, the simulations are used to guide the formulation of candidate photoresist materials selected from one or more of the silicon-containing resist architectures described herein. The candidate resists are then formulated, processed, and exposed using 11.3 nm EUV radiation, and the resulting lithographic performance is experimentally characterized. Measured stochastic metrics may be compared with modeled predictions to validate the underlying energy-deposition assumptions and to further refine material selection. This iterative modeling-formulation-validation process enables systematic reduction of stochastic variability at 11.3 nm without reliance on empirical trial-and-error alone.
[0376] These approaches allow stochastic performance to be addressed at the materials-design level, rather than solely through exposure or process conditions. In this regime, material composition becomes an active design variable for stochastic control rather than a passive contributor to absorption.11 3NM PHOTORESIST EMBODIMENT EXAMPLES
[0377] Photoresist compositions are provided comprising a radiation-sensitive resist matrix and at least one dopant element selected to modify interaction with extreme ultraviolet radiation. The photoresist composition is configured such that, when exposed to extreme ultraviolet radiation having a wavelength centered at approximately 11.3 nm, the interaction between the radiation and the resist material produces a different statistical energy-deposition behavior than when the same composition is exposed to extreme ultraviolet radiation having a wavelength centered at approximately 13.5 nm under otherwise comparable exposure and processing conditions. In some embodiments, the stochastic pattern variation is characterized by one or more measurable109 50130535.1lithographic metrics including, but not limited to, line-edge roughness, line-width roughness, pattern failure probability, critical dimension variability, or combinations thereof.
[0378] In particular, the dopant element and resist composition are selected such that, at approximately 11.3 nm, atomic-scale absorption processes influence not only the average absorbed energy but also the statistical distribution of deposited energy at nanometer length scales. This wavelength-dependent interaction enables material-level control over stochastic behavior that is not available, or is substantially less effective, at longer extreme ultraviolet wavelengths, including approximately 13.5 nm.
[0379] In some embodiments, the photoresist composition comprises silicon-containing materials selected to produce a wavelength-dependent stochastic response at approximately 11.3 nm. The silicon-containing materials may be configured such that silicon bonding environment, inorganic network structure, or local atomic coordination influence the variance of energy deposition per absorbed photon, thereby reducing stochastic pattern variation at approximately 11.3 nm relative to exposure at approximately 13.5 nm.
[0380] According to some embodiments, the photoresist composition comprises one or more additional elements selected to balance absorption efficiency and stochastic behavior at approximately 11.3 nm. Such elements may include aluminum, titanium, or combinations thereof, and may be incorporated together with silicon-containing species to influence both photon absorption and the statistical distribution of secondary energy deposition within the resist.
[0381] When exposed to extreme ultraviolet radiation centered at approximately 11.3 nm, the photoresist composition exhibits a reduced spatial extent of secondary-electron-mediated energy deposition relative to exposure at approximately 13.5 nm. This reduced spatial extent contributes to improved uniformity of chemical activation within the photoresist and reduced stochastic variability in patterned features. The reduction in stochastic pattern variation observed at an exposure wavelength of approximately 11.3 nm, relative to exposure at approximately 13.5 nm under otherwise comparable conditions, arises from differences in the statistical distribution of deposited energy within the photoresist rather than from changes in exposure dose alone.
[0382] In certain embodiments, a photoresist formulation is engineered by selecting a silicon or other dopant incorporation level as a control variable to jointly reduce stochastic-driven variability and catastrophic stochastic failures. In these embodiments, the dopant content is tuned so110 50130535.1that the resist exhibits lower edge-placement variability (including reduced line-edge roughness) while also reducing the probability of rare but yield-limiting stochastic failure events (e.g., bridging, breaks, or missing features), recognizing that such failures arise from the same underlying counting statistics and per-photon energy-deposition variability that drive roughness. The dopant loading is further selected subject to manufacturable resist-performance constraints, including maintaining sufficient image contrast and resolution capability (i.e., adequate process window) and providing pattern-transfer robustness (etch resistance) appropriate for downstream processing.
[0383] In some embodiments, the photoresist composition and process conditions are selected in combination to tune both absorption-driven photon statistics and per-photon energydeposition variance at approximately 11.3 nm. For example, resist thickness, post-apply bake and post-exposure bake conditions, development process, and (when applicable) photoacid generator and quencher loading and diffusion length may be selected together with silicon incorporation level and bonding environment to provide a target balance between stochastic-driven variability (including line-edge roughness and pattern failure probability) and conventional resist performance metrics, including contrast, resolution capability, and pattern-transfer robustness.
[0384] It should be understood that silicon-containing photoresists have previously been explored in connection with extreme ultraviolet lithography at a wavelength of approximately 13.5 nm, primarily for purposes such as improving etch resistance or increased absorption to reduce dose requirements (and shot noise). At that wavelength, however, the absorption in silicon proceeds predominantly through valence-band and continuum electronic interactions. As a result, such approaches do not access silicon core-level excitation and do not provide material-level control over the statistical distribution of energy deposited per absorbed photon. In contrast, the photoresist architectures described herein are configured for exposure at approximately 11.3 nm enable corelevel photoionization followed by constrained relaxation pathways. This wavelength-dependent interaction underlies the distinct stochastic behavior exploited by the illustrated embodiments.
[0385] FIG. 47 is a schematic illustrating representative silicon-containing photoresist architecture embodiments for 11.3 nm lithography, including polymer-bound silicon, silicon-oxo clusters, hybrid organic-inorganic networks, and vapor-deposited silicon-containing films.Categories of materials in which silicon or silicon with other dopants such as aluminum can be incorporated into include chemically amplified resists, backbone scissioning polymers, and metal-111 50130535.1oxide, -oxo cage, si-oxo and si-metal-oxo frameworks, vapor-deposited, or hybrid resists. When a 110 eVEUV photon is absorbed by a photoresist matrix, it ejects a primary photoelectron and leaves behind an ionized atomic site (A+*), which in various materials relaxes predominantly through Auger decay. The emitted Auger electron and subsequent low-energy secondary electrons form a localized electron cascade that deposits energy within a nanometer-scale volume, driving the chemical reactions responsible for image formation. Subsequent reactions lead to solubility differentials between exposed and unexposed regions of the photoresist matrix, which can be based on photoacid catalysis, chain scissioning, crosslinking, or other associated mechanisms.
[0386] In certain embodiments, the photoresist composition comprises silicon-containing chemical structures selected to enhance lithographic performance at an exposure wavelength of approximately 11.3 nm. The silicon-containing species may be molecular, polymeric, coordinationbased, inorganic-organic hybrid, cluster-based, or particulate in nature, and may be present in the backbone, side-group, cage or cluster framework, ligand environment, or as a dispersed phase.
[0387] In various implementations, the silicon-containing component may include, foe example, silicon-substituted polymers, polysiloxane-containing polymers, silicon-oxo or siliconalkoxide frameworks, silicon-containing metal-oxo clusters, silicon-functionalized carboxylate complexes, silicon-containing cage or ladder structures, or silicon-containing nanoparticles or nanoclusters dispersed within an organic or hybrid resist matrix. The silicon content may be adjusted over a range sufficient to increase 11.3 nm absorption while maintaining film-forming properties, pattern transfer performance, etch resistance, and mechanical stability under lithographic processing conditions.
[0388] In some embodiments, the total silicon content of the resist material may range from approximately 1 atomic percent to 40 atomic percent silicon, for example from 5-30 atomic percent silicon in certain formulations, and from 2-20 atomic percent silicon in particular implementations. Expressed on a mass basis, the resist may contain approximately 2 weight percent to 50 weight percent silicon, for example 5-35 weight percent silicon in certain embodiments, depending on molecular architecture, ligand loading, and desired etch resistance or absorption performance.
[0389] In certain embodiments, the silicon-containing species is selected such that the resulting resist fdm exhibits a measurable increase in 11.3 nm absorption relative to a corresponding formulation without silicon. In some embodiments, the photoresist matrix comprises a copolymer112 50130535.1derived from an acrylic polymer such as methyl methacrylate and a silicon-functional methacrylate monomer, such as trimethyl silyl methacrylate (TMS-MA). The copolymer includes repeating units of methyl methacrylate and trimethylsilyl methacrylate, such that a fraction of the methacrylate side groups incorporate a -Si(CHs)3 substituent to enhance absorption and energy-coupling efficiency at a wavelength of approximately 11.3 nm. The relative fraction of silicon-functional units may be adjusted to tune the overall silicon content of the resist, for example to provide an average silicon content between about 3 atomic percent and about 30 atomic percent, depending on formulation and co-monomer selection.
[0390] In one example, the copolymer comprises repeating units of methyl methacrylate (MAA) having the repeat-unit composition CTHsO and trimethylsilyl methacrylate (TMSMA) having the repeat-unit composition CyHuChSi and 4.17 at% Si. Representative (n, k) values at 11.3nm for MMA based on material-composition modeling and are (0.9828,0.00325) with a density of 1.19g / cm3leading to an absorption of 0.0036 nm'1. The representative (n, k) values for TMSMA are (0.9858,0.00374) with a density of 1.10g / cm3leading to an absorption of 0.0042 nm'1(i.e. 17% greater absorption than MMA).
[0391] By adjusting the mole fraction of trimethyl silyl methacrylate units in the copolymer, and by optionally incorporating more silicon-rich methacrylate comonomers, the overall silicon content of the resist matrix can be tuned over a wide range. In representative examples, copolymers containing approximately 30-60 mol% silicon-functional units may exhibit an average silicon content on the order of 2-4 at%, while copolymers containing more silicon-functional units may exhibit higher levels. For example, 60 mol% TMSMA to 40 mol% MMA copolymer leads to an elemental weighting of CaiHssOioSis with 3 at% Si leading to 11.3nm complex index based on material-composition modelling of (0.9844,0.00359) and an absorption of 0.0036 nm'1.
[0392] Formulations that make use of more silicon-dense side groups or additional silicon-containing comonomers can extend the average silicon content, providing a design space for tailoring 11.3nm absorption and energy-coupling efficiency.
[0393] According to some embodiments, higher silicon loading within an acrylate- or methacrylate-based resist binder may be obtained by employing disiloxane-substituted methacrylate monomers that introduce two silicon atoms per repeat unit. One illustrative example is 3-(disiloxanyl)propyl methacrylate which may be approximated by the empirical formula CuFF^Si?113 50130535.1with 4.9 at% Si. The representative complex index at 11 ,3nm is (0.9863,0.00399) for leading to an absorption of 0.0044 nm'1(22% greater than MMA).
[0394] Other copolymers formed from disiloxane-substituted methacrylate together with non-silicon methacrylates or acrylates can allow the overall silicon content to be tuned by adjusting the mole fraction of the Si-containing comonomer. For each chosen co-monomer ratio, an effective empirical composition can be obtained as the weighted combination of the disiloxane-substituted repeat unit and the selected non-silicon comonomer and that empirical composition can be used to calculate corresponding 11.3nm optical constants and absorption. In other embodiments, still higher silicon content can be achieved using methacrylate monomers bearing a tri s(trimethyl silyljsilyl substituent (TTM-MA) with 10 at% Si or analogous groups, in which the methacrylate side chain includes a central silicon atom bonded to three trimethylsilyl groups. Copolymers incorporating TTM-MA and a non-silicon co-monomer can therefore span a yet wider range of silicon atomic fractions.
[0395] By varying the mole fraction silicon-rich comonomers, a ladder of candidate Si-containing acrylate or methacrylate binder compositions may be defined for 11.3-nm absorption performance while retaining flexibility to incorporate additional functional comonomers to control dissolution behavior, chemical amplification, or other lithographic properties.
[0396] In other embodiments, the 11.3nm photoresist comprises aluminum-containing inorganic or hybrid materials in which aluminum oxide or aluminum oxo-complexes act as one of the primary absorbing species at an exposure wavelength of approximately 11.3 nm. The resist may include aluminum-oxo clusters or networks capped with organic ligands, such as phenolate, alkoxide, or other non-methacrylate ligands, that provide solution processability and film-forming behavior while preserving a relatively high density of aluminum and oxygen within the resist matrix.
[0397] In one example, the resist material comprises aluminum oxo-phenolate clusters that may be represented by an approximate composition of C60H50O12AI4. Such a composition corresponds to an aluminum content on the order of about 3%. The increased concentration of aluminum and oxygen relative to organic-polymeric materials within this cluster network can provide enhanced 11.3-nm absorption and energy-coupling behavior compared to conventional purely organic resists. Representative (n, k) values at 11.3 nm for this composition are (0.9853,114 50130535.10.00376), corresponding to an absorption of approximately 0.0042 nm1(about 17% greater than that of the MMA reference material).
[0398] According to some embodiments, the resist may include silicon in combination with other absorbing elements (for example, tin, hafnium, titanium, or aluminum) to tailor both absorption and chemical response. In one example, the resist may comprise a mixed aluminumsilicon hybrid material having a repeat-unit composition approximated by C4HioO?AlSi, corresponding to one aluminum atom and one silicon atom per unit and an (Al+Si) content of roughly 9 at%. In such aluminosilicate-type hybrids, silicon contributes enhanced absorption and energy-coupling efficiency in the vicinity of the silicon absorption edge, while aluminum and oxygen provide additional 11.3nm absorption, network stability, and etch resistance. Representative (n, k) values at 11.3 nm are (0.9872, 0.00581), giving an absorption of approximately 0.0065 nm1(approximately 81% greater than the MMA reference). By adjusting the Al: Si ratio and organic ligand fraction, the overall composition can be tuned to balance 11.3-nm absorption, mechanical properties, and pattern-transfer performance.
[0399] Another silicon-resist embodiment includes molecular silicon-oxo cluster resists, in which discrete silicon-oxo cores are functionalized with well-defined organic ligands to enable film formation. Suitable ligands include short alkyl chains, phenyl groups, or small carboxylates, for example carboxylate-functionalized silsesquioxane clusters. Upon exposure to 11.3 nm radiation, energy absorbed in the silicon-oxo framework can induce cleavage or modification of the organic ligands and promote cluster-cluster condensation or network formation, yielding a denser, less soluble silicon-rich material in exposed regions.
[0400] In some embodiments, the silicon-containing photoresist comprises silsesquioxane-based materials, including polysilsesquioxanes, polyhedral oligomeric silsesquioxanes (POSS), polysilane-derived networks, or related silicon-rich molecular or polymeric frameworks, provided that the composition enables the wavelength-dependent energy deposition behavior described herein.
[0401] In such embodiments, silicon-oxo cage or network structures provide a silicon-rich framework that, at approximately 11.3 nm, supports the core-level excitation and constrained relaxation pathways described herein.
[0402] Silicon-centered coordination networks represent another material embodiment. In such systems, silicon atoms serve as coordination nodes bound to multidentate ligands such as115 50130535.1carboxylates, alkoxides, or small phenolates. Representative structural examples may include repeating units of the form [Si(C>2CCH3)4]n, [Si(OCH3)4]n, or [Si(O-C6Hs)4]n, or related siliconoxygen coordination frameworks. Upon exposure to 11.3-nm radiation, absorption within the silicon-containing network can lead to modification or partial cleavage of coordinated ligands and to increased crosslinking or network densification within the silicon-oxo framework, leading to a reduction in solubility in the exposed regions and enabling negative-tone resist behavior.
[0403] In certain embodiments, the photoresist may be formed in whole or in part by a vapor-phase process and comprises silicon-containing material as the primary absorbing and structure-forming component. Vapor-phase formation may be advantageous for lithography at a wavelength of approximately 11.3 nm because it enables precise control of silicon content, film density, and thickness, while reducing compositional inhomogeneity and nanoscale phase separation that can contribute to stochastic variability. Such vapor-phase processes may include chemical vapor deposition (CVD), atomic layer deposition (ALD), molecular layer deposition (MLD), or vaporphase infiltration (VPI), alone or in combination.
[0404] In some silicon-containing embodiments, the resist may be formed from volatile silicon precursors including, but not limited to, silanes, silazanes, alkoxysilanes, or cyclic siloxanes. Representative precursors include tetramethylsilane, hexamethyldisilazane (HMDS), bis-tert-butylaminosilane (BTBAS), tetraethyl orthosilicate (TEOS), or related compounds. Reaction with oxidants, moisture, plasma species, or organic co-reactants may form silicon-oxygen or siliconoxynitrogen networks containing residual organic ligands. The resulting films may comprise siliconoxo, silicon-oxynitride, or hybrid silicon-organic structures that remain processable as thin resist layers.
[0405] In some embodiments, the vapor-deposited silicon-containing film functions directly as the imaging layer. Exposure at approximately 11.3 nm induces electronic excitation within the silicon-containing network, leading to ligand cleavage, bond rearrangement, densification, or crosslinking of the silicon-oxygen framework. These exposure-induced changes produce a difference in solubility or etch resistance between exposed and unexposed regions, enabling pattern formation without the use of conventional photoacid diffusion mechanisms.
[0406] In other embodiments, silicon-containing vapor-phase species are infiltrated into a pre-formed organic resist or polymer matrix using vapor-phase infiltration or sequential infiltration116 50130535.1synthesis. In such cases, silicon-containing precursors diffuse into the polymer and react with functional groups such as hydroxyl, carbonyl, or ether moieties, forming an inorganic-rich network within the polymer volume. The degree of infdtration, silicon loading, and depth profde may be controlled through precursor exposure conditions and cycle count. Upon exposure to 11.3 nm radiation, the infdtrated silicon-containing regions undergo localized chemical modification or densification, enabling improved pattern contrast and reduced stochastic variability.
[0407] According to some embodiments, the silicon-containing vapor-phase film is deposited at a thickness less than about 50 nm, less than about 30 nm, or less than about 20 nm, such that the film thickness is comparable to the electron transport length scale. In this regime, control of silicon concentration and bonding environment enables tuning of energy deposition behavior and mitigation of stochastic effects associated with photon absorption and secondary electron generation. By adjusting precursor chemistry, deposition conditions, and post-deposition treatment, the balance between absorption, mechanical stability, and pattern transfer performance may be controlled for lithography at 11.3 nm.
[0408] In certain embodiments, the photoresist compositions described herein are formulated as liquid solutions or dispersions suitable for conventional spin-coating processes. The photoresist may be dispensed onto a substrate and distributed by spinning at a selected rotational speed to form a uniform thin film having a thickness appropriate for 11.3 nm lithographic exposure, for example in the range of approximately 10 nm to 50 nm. Following coating, the resist film may be subjected to a soft bake or post-apply bake at an elevated temperature to remove residual solvent and to stabilize the film prior to exposure. After EUV exposure, a post-exposure bake may be performed to promote the intended chemical or structural transformation within the resist, followed by development using an appropriate developer solution to form a patterned resist image. The coating, baking, and development steps may be carried out using conventional lithographic processing equipment and conditions, and may be adjusted as needed to accommodate the specific resist chemistry and film thickness employed.
[0409] In certain embodiments, a photoresist film is disposed over a silicon-containing underlayer or hardmask. Silicon-containing underlayers suitable for use in such embodiments may include, by way of example and not limitation, silicon oxide, silicon nitride, silicon oxynitride, silicon-rich dielectric films, or combinations thereof, as are commonly employed in lithographic117 50130535.1pattern transfer. In some implementations, the underlayer is substantially stoichiometric, while in other implementations the underlayer comprises a silicon-rich or sub-stoichiometric composition. At photon energies above the silicon I >,3 absorption threshold, absorption within the underlayer can generate photoelectrons and Auger electrons that contribute to the secondary-electron population within the overlying photoresist. Variations in silicon atomic density and bonding environment within the underlayer can influence the efficiency and spectral distribution of such secondary electrons. When the photoresist thickness is on the order of the electron transport length, for example between about 10 nm and about 30 nm, this interfacial energy coupling can influence the spatial and statistical distribution of deposited energy within the resist volume. In certain implementations, the underlayer composition and thickness are selected such that this contribution reduces stochastic pattern variation, including line-edge roughness and pattern failure probability, relative to otherwise similar resist stacks employing underlayers that do not provide comparable core-level excitation pathways at the exposure wavelength, while preserving resist contrast and pattern transfer performance.
[0410] The following examples illustrate certain representative photoresist architectures that may be configured to exploit the wavelength-dependent energy-deposition behavior described above at approximately 11.3 nm. These examples span organic, hybrid, and inorganic resist classes and demonstrate how silicon incorporation level, bonding environment, and resist architecture may be selected to reduce stochastic variability while maintaining lithographic performance. The examples are provided for purposes of illustration and are not intended to limit the scope of the technology. Example 1 : Hybrid Chemically Amplified Photoresist
[0411] This embodiment illustrates a silicon-containing chemically amplified photoresist suitable in which silicon incorporation is used to reduce stochastic variability while maintaining sensitivity and process compatibility. The photoresist includes a polymeric binder comprising copolymerized aromatic and silicon-functional monomers, such as hydroxy styrene units combined with disiloxane-substituted methacrylate units. The silicon-containing side groups increase absorption and modify energy-coupling behavior at 11.3 nm. A photoacid generator and quencher system may be included to enable chemically amplified solubility switching. In representative formulations, the silicon content may be selected in a range corresponding to 2-10 at% silicon, or greater when additional silicon incorporation sites are leveraged. When exposed at approximately118 50130535.111.3 nm, absorption events occurring at silicon sites contribute to a more localized secondaryelectron distribution than that observed in comparable formulations exposed at approximately 13.5 nm. As a result, the effective spatial extent of energy deposition within the resist is reduced, contributing to improved stochastic performance while retaining exposure latitude appropriate for manufacturing use.Example 2: Silicon-Rich Molecular or Cluster-Based Photoresist
[0412] This embodiment illustrates a photoresist based on discrete silicon-containing molecular or cluster structures, such as silsesquioxane or polyhedral oligomeric silsesquioxane (POSS) materials with upwards of 15-30 at% Si. In such embodiments, silicon-oxo cage structures may be functionalized with acid-labile or otherwise radiation-sensitive organic groups to enable a solubility change upon EUV exposure. When exposed at wavelengths of approximately 11.3nm, these materials preferentially undergo core-level excitation and constrained relaxation processes at silicon sites, resulting in a narrower distribution of secondary-electron transport compared with continuum-dominated absorption. This behavior can reduce stochastic edge placement variability and feature-to-feature variation relative to analogous materials exposed at longer EUV wavelengths.Example 3: Inorganic or Silicon-Dominated Network Photoresist
[0413] This embodiment illustrates a photoresist comprising a predominantly inorganic silicon-containing network, such as hydrogen silsesquioxane or related silicon-oxo frameworks, which provide high silicon atomic density and intrinsic etch resistance. In such materials, silicon atoms form the primary structural and absorbing component of the resist. At exposure wavelengths of approximately 11.3 nm, absorption within the silicon network can proceed through core-level excitation mechanisms that are inefficient or unavailable at 13.5 nm. As a result, energy deposition within the resist is dominated by localized excitation and relaxation processes associated with silicon sites. In representative implementations, high silicon content and constrained electronic relaxation pathways can produce a relatively narrow spatial distribution of deposited energy, reducing stochastic variability while simultaneously providing high pattern transfer capability. These properties make such materials suitable for applications where both stochastic control and etch resistance are critical.119 50130535.1
[0414] Collectively, these approaches demonstrate that silicon and silicon in combination with other absorbing elements can be leveraged as primary absorbers at 11.3 nm, enabling inorganic network resists, molecular cluster resists, coordination networks, and hybrid materials. Each architecture provides distinct pathways for controlling solubility contrast, network densification, and electron cascade behavior, offering multiple viable strategies for 11.3 nm photoresist materials.
[0415] In summary, as semiconductor patterning is extended toward features of smaller dimension, stochastic variability increasingly limits lithographic performance independently of optical resolution considerations. Such variability may be described in terms of the variance of deposited energy, Var(E), which includes a contribution associated with photon-count statistics and a contribution associated with the variance of energy deposited per absorbed photon. As feature dimensions shrink, the number of absorbed photons per feature volume decreases, increasing the relative contribution of the photon-count term, while simultaneously increasing the variance associated with the spatial and energetic distribution of deposited energy per absorption event becomes increasingly significant.
[0416] At a wavelength of approximately 13.5 nm, mitigation of stochastic effects is commonly achieved by increasing absorption, for example through incorporation of higher-atomic-number constituents, thereby increasing the number of absorbed photons and reducing the photoncount contribution to Var(E). However, the per-photon energy-deposition term remains governed primarily by continuum electronic interactions that produce relatively broad secondary-electron distributions. In contrast, lithography performed at approximately 11.3 nm can provide an additional degree of freedom for control of stochastic behavior, in that material selection and composition can influence not only the average absorbed energy but also the variance associated with energy deposition per absorbed photon. In certain material systems, exposure at approximately 11.3 nm can result in a reduced contribution of the per-photon variance term to Var(E), leading to improved stochastic behavior as feature dimensions below about 20 nm pitch are pursued. As scaling progresses into regimes where rare local fluctuations dominate pattern failure probability, this combined reduction of both photon-count and per-photon variance contributions can provide a more favorable path for continued device scaling than approaches relying solely on absorption enhancement at longer extreme ultraviolet wavelengths.120 50130535.1
[0417] It should be noted that these embodiments are provided for purposes of illustration and are not intended to limit the scope of the technology.RESIST SECONDARY ELECTRON BLUR AND LER METROLOGY
[0418] In certain embodiments, secondary-electron blur (SEB) for a photoresist formulation is determined using EUV interference lithography (EUV-IL) exposures performed at approximately 11.3 nm, in which the resist is exposed to a two-beam (or multi-beam) interference field that produces a well-defined periodic intensity distribution at the image plane. EUV-IL has been demonstrated as a practical resist-evaluation approach for EUVL, including for high-resolution line / space exposures, and the same physical implementation is directly extendable to 11.3 nm using a coherent beamline and an appropriate interferometric arrangement. In this approach, the interference exposure of printed periodic patterns is not limited by optical imaging aberrations, flare, pupil apodization, pupil obscuration, or mask 3D effects from a projection system. As a result, once the interference contrast is established and controlled, the dominant contributors to stochastic roughness for periodic patterns become (i) photon counting statistics (shot noise) and (ii) the resist’s lateral ener...
Claims
What is claimed is:CLAIMS1. A photomask for 11.3 nm reflective projection lithography, comprising:a substrate;a multilayer reflective stack disposed over the substrate, the multilayer reflective stack comprising a plurality of alternating layers of at least two different materials having different complex refractive indices at approximately 11.3 nm, the multilayer reflective stack comprising a thickness profile selected from a periodic thickness profile, an aperiodic thickness profile, or a combination thereof; wherein the multilayer reflective stack has a structure configured to provide constructive interference for EUV radiation at approximately 11.3 nm and to exhibit:(i) an average of s-polarized reflectance and p-polarized reflectance greater than 60% over an angular range from -10° to +10° relative to a surface normal of the photomask;(ii) an averaged angular reflectance variation AR of less than 6% across said angular range; and(iii) a difference between s-polarized reflectance and p-polarized reflectance of less than 5% over said angular range; andan absorber layer patterned over the multilayer reflective stack;wherein the absorber layer has a thickness and optical constants selected such that a reflectance of absorber-covered regions at approximately 11.3 nm is no greater than 20% of a reflectance of adjacent regions not covered by the absorber; andwherein the photomask is configured for use in a 11.3 nm projection lithography system at said wavelength operating at a numerical aperture of 0.50 or greater.
2. The photomask of claim 1, wherein the optical constants include refractive indices and extinction coefficients.
3. The photomask of claim 1, wherein the reflectance of absorber-covered regions at approximately 11.3 nm is no greater than 14% of a reflectance of adjacent regions not covered by the absorber.
4. The photomask of claim 1, wherein the reflectance of absorber-covered regions at approximately 11.3 nm is no greater than 8% of a reflectance of adjacent regions not covered by the absorber.
5. The photomask of claim 1, wherein the reflectance of absorber-covered regions at approximately 11.3 nm is no greater than 3.6% of a reflectance of adjacent regions not covered by the absorber.
6. The photomask of claim 1, wherein the plurality of alternating layers comprises materials, compounds, elements, or alloys selected from ruthenium, molybdenum, rhodium, palladium, niobium, zirconium, boron, silicon, and beryllium.
7. The photomask of claim 6, wherein at least one of the plurality of alternating layers comprises beryllium.
8. The photomask of claim 7, wherein the plurality of alternating layers comprises a repeating sequence of ruthenium layers and beryllium layers.
9. The photomask of claim 7, wherein the plurality of alternating layers comprises a repeating sequence of molybdenum layers and beryllium layers.
10. The photomask of claim 1, wherein the plurality of alternating layers comprises a substantially periodic thickness profde, wherein a period (d) of successive alternating layers falls within a range of 5.7 nm to 5.92 nm and a thickness ratio (T) falls within a range of 0.24 to 0.68.
11. The photomask of claim 1, wherein the multilayer reflective stack includes a localized thicknessperturbation region comprising two or more adjacent layers for which the average thickness of the two or more adjacent layers differs by at least 10% from an average thickness of a neighboring two or more adjacent layers, wherein the two or more adjacent layers and the neighboring two or more adjacent layers have an identical number and composition of layers.
12. The photomask of claim 1, wherein the multilayer reflective stack includes a localized thicknessperturbation region consisting of a single layer having a thickness that differs by at least 25% from the thicknesses of two or more adjacent layers on each side of the single layer.
13. The photomask of claim 11, wherein the localized thickness-perturbation region is disposed adjacent to the substrate and extends across fewer than ten neighboring layer pairs.
14. The photomask of claim 11, wherein the two or more adjacent layers of the localized thicknessperturbation region have at least 10% greater thickness than average thickness of a neighboring two or more adjacent layers.
15. The photomask of claim 11, wherein the two or more adjacent layers of the localized thicknessperturbation region have at least 10% lesser thickness than average thickness of a neighboring two or more adjacent layers.
16. The photomask of claim 1, wherein the multilayer reflective stack includes one or more barrier layers disposed between adjacent layers of the stack.
17. The photomask of claim 1, wherein the multilayer reflective stack comprises a plurality of thickness-profile blocks, each block having a thickness profile that is periodic or aperiodic and is different than the other blocks in the plurality.
18. The photomask of claim 17, wherein the plurality of thickness-profile blocks comprises a first block formed from a first material pair and a second block formed from a second, different material pair.
19. The photomask of claim 18, wherein the first material pair comprises ruthenium and beryllium and the second material pair comprises molybdenum and beryllium.
20. The photomask of claim 1, wherein the multilayer reflective stack comprises between 20 and 70 layer-pairs.
21. The photomask of claim 1, wherein the multilayer reflective stack is configured to exhibit an average of s-polarized reflectance and p-polarized reflectance greater than 60% over an angular range from -11° to +11° relative to a surface normal of the photomask.
22. The photomask of claim 1, wherein the difference between s-polarized reflectance and p-polarized reflectance is less than 4% over said angular range.
23. The photomask of claim 1, wherein the difference between s-polarized reflectance and p-polarized reflectance is less than 3% over said angular range.
24. The photomask of claim 1, wherein the averaged angular reflectance variation is less than 5% across said angular range.
25. The photomask of claim 1, wherein the photomask is configured for use in a projection lithography system having a numerical aperture of 0.60 or greater.
26. The photomask of claim 1, wherein the absorber layer comprises a material selected from the group consisting of nickel, silicon, tantalum, tin, tungsten, cobalt, platinum, molybdenum, zinc, rhodium, chromium, aluminum, and titanium.
27. The photomask of claim 26, wherein the absorber layer comprises a metal alloy or metal-based compound including a material selected from the group consisting of nickel, silicon, tantalum, tin, tungsten, cobalt, platinum, molybdenum, zinc, rhodium, chromium, aluminum, and titanium, wherein the metal-based compound is a nitride, boride, oxide, carbide, silicide, or a chalcogenide.
28. The photomask of claim 1, wherein the absorber layer has a thickness between 20 nm and 60 nm.
29. The photomask of claim 28, wherein the thickness of the absorber layer is selected to coincide with a contrast-maximum node associated with a destructive-interference condition within the absorber layer at a wavelength of approximately 11.3 nm.
30. A method of manufacturing a photomask for 11.3 nm reflective projection lithography, the method comprising:depositing, on a mask substrate, a multilayer reflective stack comprising a plurality of alternating layers of at least two different materials having different complex refractive indices at approximately 11.3 nm, the multilayer reflective stack being configured to exhibit, over an incidence-angle range from -10° to +10° relative to a surface normal of the photomask:(i) an average of s-polarized and p-polarized reflectance greater than 60 percent; and (ii) a reflectance variation of less than 6 percent;depositing an absorber layer over the multilayer reflective stack such that a reflectance of absorber-covered regions at approximately 11.3 nm is no greater than 14 percent of a reflectance of adjacent non-absorber regions; andpatterning the absorber layer to define lithographic features of the photomask.
31. The method of claim 30, wherein the reflectance of absorber-covered regions at approximately 11.3 nm is no greater than 8% of a reflectance of adjacent regions not covered by the absorber.
32. The method of claim 30, wherein the plurality of alternating layers comprises materials, compounds, elements, or alloys selected from ruthenium, molybdenum, rhodium, palladium, niobium, zirconium, boron, silicon, and beryllium.
33. The method of claim 30, wherein the plurality of alternating layers comprises a repeating sequence of ruthenium layers and beryllium layers.
34. The method of claim 30, wherein the plurality of alternating layers comprises a repeating sequence of molybdenum layers and beryllium layers.
35. The method of claim 30, wherein the plurality of alternating layers comprises a substantially periodic thickness profile, wherein a period (d) of successive alternating layers falls within a range of 5.7 nm to 5.92 nm and a thickness ratio (T) falls within a range of 0.24 to 0.68.
36. The method of claim 30, wherein the multilayer reflective stack includes a localized thicknessperturbation region comprising two or more adjacent layers for which the average thickness of the two or more adjacent layers differs by at least 10% from an average thickness of a neighboring two or more adjacent layers, wherein the two or more adjacent layers and the neighboring two or more adjacent layers have an identical number and composition of layers.
37. A method of performing reflective projection lithography, comprising:generating EUV radiation having a wavelength of approximately 11.3 nm;illuminating a reflective photomask with the EUV radiation, the photomask comprising a substrate, a multilayer reflective stack disposed on the substrate, and a patterned absorber layer disposed over the multilayer reflective stack, wherein the multilayer reflective stack comprises alternating layersincluding beryllium and at least one material selected from ruthenium, rhodium, molybdenum, palladium, and niobium, and is configured to exhibit, at angles within an incidence-angle range from -10° to +10° relative to a surface normal of the photomask:(i) an average of s-polarized and p-polarized reflectance greater than 60 percent; and (ii) a polarization-dependent reflectance difference of less than 6 percent;reflecting the EUV radiation at angles within said incidence-angle range at the photomask; and projecting the reflected EUV radiation onto a photoresist-coated substrate.
38. The method of claim 37 wherein the step of projecting comprises projecting the reflected EUV radiation through an 11.3 nm projection optical system having a numerical aperture of at least 0.50.
39. The method of claim 37, wherein the photoresist comprises a silicon-containing material selected such that the photoresist exhibits greater absorption at approximately 11.3 nm than a corresponding formulation lacking silicon.
40. The method of claim 39, wherein the silicon-containing photoresist is selected such that, for a dense line-and-space pattern having a pitch of 20 nm or less, a three-sigma line-edge roughness (LERss) is lower when exposed at a wavelength of approximately 11.3 nm than when the same photoresist composition is exposed at approximately 13.5 nm under otherwise comparable optical, dose-normalized, and processing conditions.
41. The method of claim 39, wherein the silicon-containing photoresist is selected such that a three-sigma line-edge roughness (LERio) is lower when exposed at approximately 11.3 nm than when the same photoresist composition is exposed at approximately 13.5 nm under otherwise comparable optical, dose-normalized, and processing conditions.
42. The method of claim 37, wherein the incidence-angle range includes angles up to at least ±11 degrees relative to a surface normal of the photomask.
