betaFeedback

WorksPhysics and Astronomy

A Two-Mirror Faceted Projection System for EUV Lithography

Abstract

We propose an all-reflective two-mirror projection system for extreme ultraviolet (EUV) lithography operating at exposure wavelengths of 13.513.5~nm (Mo/Si) and 11.211.2~nm (Ru/Be), delivering a fourfold (4×4\times) demagnification of the periodic mask pattern at a numerical aperture approaching unity (NAmax⁡≈0.993\mathrm{NA}_{\max} \approx 0.993). In contrast to conventional EUV projection objectives that incorporate 6–10 aspheric mirrors with an overall optical throughput of less than 15%15\%, the proposed design redirects each accepted discrete spatial diffraction order scattered by the mask onto the wafer via a dedicated pair of planar mirror facets. The number of reflections is strictly fixed at two for all accepted orders, retaining 5050–60%60\% of the power leaving the mask in each accepted order. We derive a spatial geometry providing rigorous optical path length equalization across all diffraction orders, thereby removing order-dependent propagation phase shifts. Individually optimized 30-bilayer Bragg multilayer coatings are designed for each facet using the transfer matrix method combined with global evolutionary optimization algorithms. The architecture is generalized to a three-dimensional vector formulation with a two-dimensionally periodic mask. Utilizing inverse lithography technology, Fourier parameterization, and a differentiable electromagnetic modal waveguide solver, we solve the synthesis problem for binary absorber masks (La absorber on a Ru/Be/Sr multilayer mirror). We demonstrate simulated aerial images of sub-10-nm features on the wafer (isolated peaks with a full width at half maximum (FWHM) of approximately 5.45.4~nm and line pairs with a critical dimension of 66~nm) and find that the two peaks remain resolved for the tested wafer defocus values from 00 to 55~nm along the zz-axis.

Happiness for everybody, free, and no one will go away unsatisfied!

Arkady and Boris Strugatsky, Roadside Picnic

Introduction

The modern technological world critically depends on semiconductor microchips, which lie at the heart of smartphones and personal computers, control automobiles and industrial robotics, route global internet traffic, and serve as the physical foundation for the ongoing wave of artificial intelligence systems. Global demand for computation, telecommunications, and machine perception is expanding at an explosive pace, and microelectronics has met this challenge for over half a century by doubling the number of active components on a chip approximately every two years in accordance with Moore’s Law [1]. At the foundation of this exponential progress lies the continuous downscaling of physical device dimensions, directly tied to the miniaturization of the individual transistor: smaller transistors yield higher integration density, superior operational speed, and enhanced energy efficiency. Optical photolithography serves as the primary engine of this miniaturization. It transfers the topology of the integrated circuit layout from a photomask onto a silicon wafer, thereby dictating the minimum printable feature size. The critical role of this tool in the economics of semiconductor manufacturing is reflected in its capital cost: while a 193-nm immersion lithography scanner costs on the order of 50 million euros, a state-of-the-art EUV scanner exceeds 300 million euros [2].

The minimum feature size printable via photolithography is fundamentally constrained by physical laws and obeys Rayleigh’s resolution equation, CD=k1λNACD = \frac{k_1\lambda}{NA}, where CD is the critical dimension, k1k_1 is a process-dependent factor (with a theoretical physical limit of 0.25), λ\lambda is the exposure wavelength, and NA is the numerical aperture of the projection optics [1]. As is evident from this fundamental relationship, shrinking feature dimensions requires either reducing the exposure wavelength or increasing the numerical aperture. Over six decades, the exposure wavelength transitioned from 436 nm (mercury g-line) through 365, 248, and 193 nm down to 13.5 nm, while the numerical aperture in liquid immersion systems surpassed unity to reach 1.35 [1, 3]. Along this developmental trajectory, lithographers adopted advanced techniques, such as off-axis illumination (OAI) and multiple patterning. In 2018, EUV lithography became commercially viable for high-volume manufacturing following the introduction of reliable 13.5-nm plasma sources delivering output powers exceeding 125 W [1].

The operational wavelength of 13.5 nm was selected because periodic Mo/Si Bragg multilayer coatings composed of 40–50 bilayers exhibit record near-normal reflectivities of ∼70%\sim70\% per mirror surface [4], and commercial scanner sources now deliver powers exceeding 400 W [5]. Standard NXE scanners operating at NA=0.33NA = 0.33 print a minimum half-pitch of 13 nm, serving as the core workhorses of high-volume semiconductor manufacturing. The subsequent generation of High-NA EXE:5000 systems, with NA=0.55NA = 0.55, pushes the resolution down to 8 nm. However, this is achieved via an anamorphic projection architecture (4×/8×4 \times/ 8 \times) in which the depth of focus (DoF) is reduced by a factor of 2.94, the exposure field size is halved, and mirror angles of incidence are constrained by the narrow angular reflectance bandwidth of Mo/Si coatings (±11∘\pm11^\circ) [1, 6, 7, 8]. Further increasing the numerical aperture to NA=0.75NA = 0.75 and beyond, as required for printing critical dimensions below 6 nm, will necessitate an even greater number of mirrors within the all-reflective projection channel [7, 9].

Each reflection in the illumination and projection optics absorbs a substantial fraction of the incident EUV power [2]. This motivates alternative wavelengths, including 11.2 nm with Ru/Be coatings and Xe-discharge sources [2, 10], short-wavelength multilayers in the 9–12 nm band [4, 11, 12], and architectures with fewer reflections. Shintake [9] compares a conventional system with four illumination and six projection reflections with a proposed system having two illumination and two projection reflections. At an assumed reflectance of 0.65 per surface, the corresponding transmissions are 0.6510≈1.3%0.65^{10} \approx1.3\% and 0.654≈18%0.65^4 \approx18\%. The associated estimate of a reduction from 1 MW to approximately 80 kW concerns electricity used for EUV generation, not the total scanner consumption. A later four-mirror in-line design [13] used repeated encounters with its physical mirrors, giving eight reflections for NA=0.5NA = 0.5 and twelve for NA=0.7NA = 0.7. Mirror count must therefore be distinguished from reflection count.

The design of the reflective optical systems described above relies, in most cases, on geometric ray-tracing methods, which represent crude approximations of electromagnetic wave propagation that neglect wave phenomena such as interference and diffraction. Incorporating diffractive elements directly into the projection optics has been explored previously. For instance, in a three-mirror all-reflective scheme operating at 13 nm, aspheric mirrors were replaced by reflection diffraction gratings on spherical substrates [14], and later this line of research was extended using reflective Fresnel zone plates [15, 16]. However, in those systems the diffractive structure merely reproduces the phase profile of an aspheric mirror, the total number of reflections within the channel is not reduced, and stringent monochromaticity requirements are imposed on the source. Separately, maskless EUV lithography concepts based on arrays of micro-electromechanical mirrors have been proposed as alternatives to physical masks [17], addressing mask cost, defect, and pellicle challenges, though requiring their own specialized projection architectures.

In this work, we develop an alternative concept that pushes the reduction of reflective surfaces to its logical limit. We propose a two-mirror all-reflective projection system in which each spatial diffraction order scattered by the mask is redirected onto the wafer individually by a dedicated pair of planar mirror facets. The number of reflections per accepted channel between the mask and wafer is fixed at two, independent of the numerical aperture. Thus, values of NA ∼1\sim1 are fundamentally attainable alongside a fourfold (4×4\times) demagnification of the mask pattern. We consider both two-dimensional and three-dimensional problem formulations. In the 3D case, the mask represents a two-dimensionally periodic structure (a square grating), and each propagating diffraction order is mapped to its own corresponding facet pair across both mirrors. The analysis is performed in an approximation close to geometrical optics (GO), where propagating spatial diffraction orders are described as collimated paraxial beams with uniform cross-sectional amplitude.

The paper is structured as follows. In Section 2, the electromagnetic diffraction problem for an EUV mask is formulated and the basic relations governing the two-mirror projection system are derived. Section 3 presents the two-dimensional two-mirror projection system, including the iterative design of the mirror geometry, the optimization of individually tuned Bragg multilayer coatings for each facet, and the synthesis of binary absorber masks for a target aerial image using inverse lithography technology with Fourier-parameterized projection. Section 4 extends the formulation to three dimensions, generalizing the design to a two-dimensionally periodic mask with a square lattice and demonstrating 3D mask synthesis. Section 5 discusses the advantages and disadvantages of the proposed projection system. Finally, in Section 6 concluding remarks are given. Appendix A provides a rigorous analysis of the validity of the geometrical-optics beam approximation for centimeter-scale EUV beams.

Formulation of the Problem

We begin by examining the electromagnetic diffraction problem for an EUV mask in a two-dimensional formulation, subsequently extending the analysis to the structural design of a projection system providing a 4×4\times demagnification of the mask pattern onto the wafer.

The layered mask, consisting of JJ layers, occupies the domain [−D,0][-D,0] along the zz-axis and [−L/2,L/2][-L/2,L/2] along the xx-axis in a Cartesian coordinate system (x,y,z)(x,y,z), as illustrated in Figure 1, and is periodic along the xx-direction with spatial period LxL_x. Each layer jj is homogeneous along the zz-direction and has a complex permittivity εj=εj(x)\varepsilon_j=\varepsilon_j(x). Optical constants are obtained from tabulated reference data [18, 19]. Within the mask stack, one distinguishes patterned absorber (or phase-shifting) layers that define the composition of the scattered spectrum, and the underlying multilayer Bragg mirror substrate (see Figure 2). Here, the multilayer substrate is assumed to consist of periodic three-component Ru/Be/Sr stacks [20] or two-component Ru/Be (or Mo/Si) bilayers [4, 11, 12].

Illustration

Figure 1.

Electromagnetic beam scattering by a lithographic mask

Figure 1. Electromagnetic beam scattering by a lithographic mask (schematic in the geometrical-optics approximation).

Geometry of the problem: (a) cross-section at $y=0$ with mask spatial period $L_x$, (b) perspective view of the mask stack

Figure 2. Geometry of the problem: (a) cross-section at y=0y = 0 with mask spatial period LxL_x, (b) perspective view of the mask stack.

The primary objective is to construct an all-reflective projection system that provides a 4×4\times demagnification of the mask pattern when projected onto the wafer. Under 4×4\times reduction, the spatial period on the wafer, Lx(w)L_x^{(\mathrm{w})}, satisfies Lx(w)=Lx/4L_x^{(\mathrm{w})}=L_x/4 (i.e., Lx=4Lx(w)L_x=4L_x^{(\mathrm{w})}), where both the wafer plane and the mask plane are assumed to be parallel to the xOyxOy plane. Pattern transfer from the mask to the wafer is mediated by the electromagnetic field scattered by the mask and focused onto the wafer.

Assuming the lateral mask dimension satisfies L≫LxL\gg L_x, edge effects can be neglected, allowing the mask to be treated as an infinite periodic structure during the diffraction analysis. A TE-polarized monochromatic plane wave with angular frequency ω\omega is obliquely incident on the mask. In the geometrical-optics (GO) beam approximation, the field amplitude is uniform within the beam envelope and zero elsewhere. The incident electric field is given, with exp⁡(iωt)\exp(i\omega t) time dependence dropped, by E(i)=y0E0exp⁡[−i(m~κxx−kz(i)z)]\mathbf{E}^{(i)}=\mathbf{y}_0E_0\exp\left[-i(\tilde{m}\kappa_x x-k_z^{(i)}z)\right], where E0E_0 is the electric field amplitude, m~κx\tilde{m}\kappa_x and kz(i)k_z^{(i)} are the components of the wave vector k0\mathbf{k}_0 in free space (k0=[m~2κx2+(kz(i))2]1/2k_0=\left[\tilde{m}^{2}\kappa_x^{2}+(k_z^{(i)})^2\right]^{1/2}, κx=2π/Lx\kappa_x=2\pi/L_x, k0=ω/ck_0=\omega/c, where cc is the speed of light in free space), and superscript (i)(i) denotes the incident wave. The resulting boundary-value diffraction problem is solved using modal numerical methods [20, 21, 22, 23, 24, 25]. The scattered electric field in the vicinity of the mask is expanded in a series:

Ey(r)=∑m=−∞∞Am(r)exp⁡(−iκxmx−ikz;mz),(1)E_y^{(r)}=\sum_{m=-\infty}^{\infty}A_m^{(r)}\exp(-i\kappa_xmx-ik_{z;m}z), \tag*{(1)}

where Am(r)A_m^{(r)} is the complex amplitude of the mm-th scattered spatial harmonic, and kz;m=(k02−κx2m2)1/2k_{z;m} = \left(k_0^2-\kappa_x^2m^2\right)^{1/2} (choosing the branch Im⁡kz;m≤0\operatorname{Im} k_{z;m} \le0).

For the projection-system design, only propagating diffraction orders (purely real kz;mk_{z;m}) are relevant. Thus, the scattered field of interest outside the near-field region of the mask is:

Ey(r)=∑m=−MMA~m(r)(x,z)exp⁡(−iκxmx−ikz;mz),(2)E_y^{(r)} = \sum_{m=-M}^{M} \widetilde{A}_m^{(r)}(x,z)\exp\left(-i\kappa_xmx-ik_{z;m}z\right), \tag*{(2)}

where A~m(r)(x,z)=Am(r)\widetilde{A}_m^{(r)}(x,z)=A_m^{(r)} inside the spatial beam corresponding to order mm (see Figure 1), and A~m(r)(x,z)=0\widetilde{A}_m^{(r)}(x,z)=0 outside it. The maximum propagating order index is M=⌊k0/κx⌋M=\lfloor k_0/\kappa_x\rfloor.

To achieve a 4×4\times demagnified aerial image on the wafer, the propagating field spectrum at the wafer boundary must take the form:

Ey(w)=∑n=−NNBnexp⁡[−iκx(w)nx−iswkz;n(w)ζw],(3)E_y^{(w)} = \sum_{n=-N}^{N} B_n\exp\left[-i\kappa_x^{(w)}nx-is_wk_{z;n}^{(w)}\zeta_w\right], \tag*{(3)}

where ζw=z−Zw\zeta_w=z-Z_w is measured from the wafer plane, sw=+1s_w=+1 in Examples 1, 3, and 4 and sw=−1s_w=-1 in Example 2, and BnB_n is the complex amplitude of the nn-th harmonic at ζw=0\zeta_w=0, κx(w)=2π/Lx(w)\kappa_x^{(w)}=2\pi/L_x^{(w)}, and kz;n(w)=[k02−(κx(w))2n2]1/2k_{z;n}^{(w)}=\left[k_0^2-\left(\kappa_x^{(w)}\right)^2n^2\right]^{1/2}. Under 4×4\times reduction, the maximum propagating order transmitted to the wafer is N=⌊M/4⌋N=\lfloor M/4\rfloor (ideally M=4NM=4N; the difference arises from the cutoff of orders n>Nn>N as evanescent waves).

Hence, the design task reduces to transforming the transverse spatial frequency of each mm-th harmonic scattered by the mask into harmonic n=−mn=-m on the wafer in the common Cartesian coordinate system:

mκx⟶−mκx(w).(4)m\kappa_x \longrightarrow-m\kappa_x^{(w)}. \tag*{(4)}

Because this spatial-frequency mapping is linear, it can be realized using a dual set of planar mirror facets. Specifically, the first mirror system redirects each mm-th beam along a direction parallel to the zz-axis (Figure 3), whereupon the second mirror system redirects it toward the wafer at an angle corresponding to kx=−mκx(w)k_x=-m\kappa_x^{(w)} (Figure 4). Figure 4 schematically illustrates the all-reflective system comprising two sets of planar mirror facets. The first set converts the divergent beams scattered by the mask into a parallel array of vertical beams. The second set redirects a subset of these parallel beams at the required convergence angles onto the wafer.

Reflection of the $m$-th diffraction beam from a planar mirror facet into a direction antiparallel to the $z$-axis

Figure 3. Reflection of the mm-th diffraction beam from a planar mirror facet into a direction antiparallel to the zz-axis.

Schematic of the all-reflective two-mirror faceted projection architecture

Figure 4. Schematic of the all-reflective two-mirror faceted projection architecture.

We note that because the illumination beam with spatial harmonic m~\widetilde{m} enters through an optical input aperture occupying the position corresponding to the −m~-\widetilde{m} reflection order (back reflection), order m=−m~m=-\widetilde{m} is omitted from the projected spectrum (Figure 4). Thus, at the first reflection, the amplitude A~−m~(r)(x,z)\widetilde{A}_{-\widetilde{m}}^{(r)}(x,z) is multiplied by zero.

Referring to Figure 3, the beam parameters and mirror facet orientation for vertical reflection satisfy:

cos⁡αm=mkxk0,am=Lsin⁡αm,γm=π4−αm2,(5)\begin{aligned} \cos\alpha_m &= \frac{m k_x}{k_0}, \\ a_m &= L \sin\alpha_m, \\ \gamma_m &= \frac{\pi}{4} - \frac{\alpha_m}{2}, \tag*{(5)} \end{aligned}

where αm\alpha_m is the angle of the mm-th beam measured from the mask plane (horizontal), ama_m is the beam cross-sectional width, and γm\gamma_m is the tilt angle of the facet (τm=γm\tau_m = \gamma_m in Figure 3). Each mm-th beam projects an illuminated footprint Lm(w)L_m^{(\mathrm{w})} on the wafer plane given by:

am=Lm(w)sin⁡αm(w),(6)a_m = L_m^{(\mathrm{w})} \sin\alpha_m^{(\mathrm{w})}, \tag*{(6)}

where sin⁡αm(w)=kz;m(w)/k0\sin\alpha_m^{(w)} = k_{z;m}^{(w)}/k_0 and αm(w)\alpha_m^{(w)} is the grazing angle at the wafer plane. Consequently:

Lm(w)=Lkz;mkz;m(w)=L(1−(mκxk0)21−16(mκxk0)2)1/2.(7)L_m^{(w)} = L\frac{k_{z;m}}{k_{z;m}^{(w)}} = L\left(\frac{1-\left(\frac{m\kappa_x}{k_0}\right)^2}{1-16\left(\frac{m\kappa_x}{k_0}\right)^2}\right)^{1/2}. \tag*{(7)}

To prevent beam crosstalk, each facet of the first mirror must intercept only its assigned diffraction order. This requires that adjacent beam envelopes do not overlap at the mirror plane. The intersection point of the boundaries of adjacent orders mm and m+1m+1 is derived from their ray equations:

z=kz;mκxm(x−L2),z=kz;m+1κx(m+1)(x+L2),(8)\begin{aligned} z &= \frac{k_{z;m}}{\kappa_x m}\left(x-\frac{L}{2}\right),\\ z &= \frac{k_{z;m+1}}{\kappa_x(m+1)}\left(x+\frac{L}{2}\right), \tag*{(8)} \end{aligned}

for beams of diffraction orders mm and m+1m+1, respectively. Equating the zz-coordinates yields the intersection coordinates:

x=L2(m+1)kz;m+mkz;m+1(m+1)kz;m−mkz;m+1,(9)x = \frac{L}{2}\frac{(m+1)k_{z;m}+mk_{z;m+1}}{(m+1)k_{z;m}-mk_{z;m+1}}, \tag*{(9)}
z=Lκxkz;mkz;m+1(m+1)kz;m−mkz;m+1.(10)z = \frac{L}{\kappa_x}\frac{k_{z;m}k_{z;m+1}}{(m+1)k_{z;m}-mk_{z;m+1}}. \tag*{(10)}

Figure 5 plots these boundary intersection coordinates calculated from Eqs. (9) and (10) for Lx=20λL_x = 20\lambda. Coordinates are normalized to LL. Adjacent beams overlap below these intersection points and fully decouple above them. Hence, the first-mirror facets must be positioned at zz-coordinates exceeding these critical thresholds.

Spatial intersection coordinates of adjacent beam boundaries

Figure 5. Spatial intersection coordinates of adjacent beam boundaries (markers) computed via Eqs. (9) and (10) for Lx=20λL_x = 20\lambda in the xOzxOz plane. Thin solid lines denote beam edges originating from the mask boundaries (∣x∣=L/2|x| = L/2, z=0z = 0), the dashed line traces the intersection loci, and labels indicate diffraction order pairs (m,m+1m, m+1). Coordinates are normalized to mask width LL.

The angular sector available for positioning the facet for order mm is bounded by rays connecting the mask center to the nearest adjacent intersection points, shown as blue boundary lines in Figure 6.

First-stage mirror system

Figure 6. First-stage mirror system. Blue lines designate angular spatial-frequency sectors allocated to individual diffraction harmonics.

2D Two-Mirror Projection System

Design of the Two-Mirror Projection System

The development of the two-mirror projection system proceeds through an iterative design progression:

→ Suggestions → Design of the mirror system → Discussion of disadvantages

This approach enables systematic iterative refinement of the two-mirror projection system.

Consider an illumination beam incident at 6∘6^\circ relative to the zz-axis (the industrial standard for EUV lithography). The mask lateral dimension is L=10L = 10 mm, the exposure wavelength is λ=11.2\lambda= 11.2 nm, and the mask pattern period is Lx=8λ/sin⁡6∘≈857L_x = 8\lambda/\sin6^\circ\approx857 nm. Under these conditions, the maximum propagating mask-order index is M=⌊k0/κx⌋=76M = \lfloor k_0/\kappa_x\rfloor= 76, and the maximum wafer-order index under 4×4\times demagnification is N=⌊M/4⌋=19N = \lfloor M/4\rfloor= 19.

In the configurations evaluated below, the first mirror is located at z=Z1=1000z = Z_1 = 1000 mm. The center of the mm-th facet on the first mirror is located at (±x1(m),Z1)(\pm x_1^{(m)}, Z_1), where x1(m)=Z1tan⁡θmx_1^{(m)} = Z_1 \tan\theta_m and tan⁡θm=κxm/kz;m\tan\theta_m = \kappa_x m/k_{z;m}.

Example 1: Mask and Wafer Located on the Same Side of the Second Mirror

In this baseline geometry, the second mirror system is positioned below the wafer plane (mask and wafer are on the same side of it). The tilt angle of the mm-th facet on the first mirror relative to the horizontal in this and the following example is α1(m)=θm/2\alpha_1^{(m)} = \theta_m/2. The center of the mm-th facet on the second mirror is positioned at (±x1(m),zf(m))(\pm x_1^{(m)}, z_f^{(m)}), where zf(m)=Zw−x1(m)/tan⁡θm(w)z_f^{(m)} = Z_w - x_1^{(m)}/\tan\theta_m^{(w)} with tan⁡θm(w)=4κxm/kz(w);z;m\tan\theta_m^{(w)} = 4\kappa x_m/k_z^{(w)};z;m. The tilt angle of the second-mirror facet relative to the horizontal is α2(m)=θm(w)/2\alpha_2^{(m)} = \theta_m^{(w)}/2.

Table 1 lists the structural parameters for the 19 positive-index transmitted orders.

mmθm\theta_m (°)θm(w)\theta_m^{(w)} (°)x1(m)x_1^{(m)} (mm)zf(m)z_f^{(m)} (mm)ama_m (mm)Lm(w)L_m^{(w)} (mm)α1(m)\alpha_1^{(m)} (°)α2(m)\alpha_2^{(m)} (°)
10.753.0013.1−549.710.0010.010.371.50
21.506.0026.1−548.710.0010.050.753.00
32.259.0239.2−547.19.9910.121.124.51
43.0012.0752.3−544.89.9910.211.506.03
53.7515.1565.5−541.89.9810.341.877.57
64.5018.2878.6−538.19.9710.502.259.14
75.2521.4691.8−533.69.9610.702.6210.73
86.0024.72105.1−528.39.9510.953.0012.36
96.7528.06118.4−522.29.9311.253.3814.03
107.5131.51131.8−515.09.9111.633.7515.75
118.2635.09145.2−506.79.9012.094.1317.55
129.0238.84158.8−497.29.8812.684.5119.42
139.7842.80172.4−486.19.8513.434.8921.40
1410.5447.03186.1−473.39.8314.425.2723.51
1511.3051.62199.9−458.39.8115.805.6525.81
1612.0756.74213.8−440.29.7817.836.0328.37
1712.8362.68227.8−417.79.7521.256.4231.34
1813.6070.18242.0−387.29.7228.666.8035.09
1914.3783.23256.3−330.49.6982.137.1941.61

Table 1. Mirror system configuration for Example 1 (λ=11.2\lambda= 11.2 nm, L=10L = 10 mm, Lx=857L_x = 857 nm, M=76M = 76, N=19N = 19, Z1=1000Z_1 = 1000 mm, Zw=−300Z_w = -300 mm).

The lateral span of the first mirror along the xx-axis is ±256.3\pm256.3 mm at fixed height z=Z1=1000z = Z_1 = 1000 mm, i.e., all facet centers of the first mirror lie in a single horizontal plane (facet tilt angles range from 0.37∘0.37^\circ to 7.19∘7.19^\circ, visually indistinguishable from a flat mirror). The second mirror spans an identical lateral range, but its facets lie along a curved profile zf(m)=Zw−x1(m)/tan⁡θm(w)z_f^{(m)} = Z_w - x_1^{(m)}/\tan\theta_m^{(w)} with heights spanning zf(1)=−549.7z_f^{(1)} = -549.7 mm (center) to zf(19)=−330.4z_f^{(19)} = -330.4 mm (edge). The maximum numerical aperture of the system reaches NAmax⁡=sin⁡θ19(w)=0.993\mathrm{NA}_{\max} = \sin\theta_{19}^{(w)} = 0.993.

Figure 7 displays the ray-trace diagram in the developed projection system. Each mm-th harmonic of the field scattered by the mask is converted by the first set of 2N2N planar facets into a beam parallel to the zz-axis, and then focused onto the wafer at angle θm(w)\theta_m^{(w)} to the normal by the second set of 2N2N facets. The beam width am≈10a_m \approx10 mm is preserved upon first reflection (i.e., redirection into the −z-z direction), but after reflection from the second (tilted) mirror it projects onto the wafer with a footprint enlarged by a factor of 1/cos⁡θm(w)1/\cos\theta_m^{(w)}. For m=19m = 19 (θ19(w)=83.2∘\theta_{19}^{(w)} = 83.2^\circ), the footprint reaches 82.1 mm. As the beam arrives at near-grazing incidence on the wafer, the perpendicular cross-section am=9.7a_m = 9.7 mm is stretched by a factor of 8.5.

Ray tracing in the all-reflective two-mirror projection system

Figure 7. Ray tracing in the all-reflective two-mirror projection system for N=19N = 19 orders (showing orders m=1,4,7,10,13,16,19m = 1, 4, 7, 10, 13, 16, 19). Color represents diffraction order: purple denotes m=1m = 1 (θm(w)=3.0∘\theta_{m}^{(\mathrm{w})} = 3.0^\circ), dark red denotes m=19m = 19 (θm(w)=83.2∘\theta_{m}^{(\mathrm{w})} = 83.2^\circ). Solid lines show marginal rays from the mask edges ±L/2\pm L/2. Dashed lines show axial rays. Rays trace from the mask (z=0z = 0) through mirror 1 (z=1000z = 1000 mm) and mirror 2 (z∈[−550,−330]z \in[-550, -330] mm) to the wafer (z=−300z = -300 mm).

Discussion. The analyzed system forms a demagnified pattern on the wafer via spatial harmonic transformation. The primary limitation of this design is that the wafer is situated within the ray convergence zone directed by the second mirror, severely restricting the physical clearance required for wafer stage manipulation and handling — a critical constraint in industrial lithography. Furthermore, the m=0m = 0 beam is omitted from the field arriving at the wafer, and high-order beams produce relatively large footprints on the wafer plane. The latter two aspects are secondary compared to the first constraint.

Suggestion 1. Position the wafer outside the space between the first and second mirrors.

Example 2: Wafer Positioned Below the Second Mirror

We now consider a configuration in which the second mirror system is positioned between the mask and the wafer, placing the wafer entirely below the second mirror (mask and wafer are on opposite sides of it). The first mirror remains identical (z=Z1=1000z = Z_1 = 1000 mm). The vertically descending beams are reflected by the second mirror toward the center of the wafer while continuing downward. The center of the mm-th facet of the second mirror is positioned at (±x1(m),zf(m))(\pm x_1^{(m)}, z_f^{(m)}), where

zf(m)=Zw+x1(m)tan⁡θm(w),(11)z_f^{(m)} = Z_w + \frac{x_1^{(m)}}{\tan\theta_m^{(\mathrm{w})}}, \tag*{(11)}

so that the second mirror system spans the domain from zf(1)=−50.3z_f^{(1)} = -50.3 mm (center) to zf(19)=−269.6z_f^{(19)} = -269.6 mm (edge). The height difference (219 mm) is identical to that in Example 1, but the second mirror system is now positioned above the wafer. The facet tilt angle relative to the horizontal is α2(m)=90∘−θm(w)/2\alpha_{2}^{(m)} = 90^\circ- \theta_{m}^{(w)}/2: the facets are nearly vertical, and beams strike them at grazing angles θm(w)/2\theta_{m}^{(w)}/2 measured from the facet plane (from 1.5∘1.5^\circ for m=1m = 1 to 41.6∘41.6^\circ for m=19m = 19), corresponding to incidence angles 90∘−θm(w)/290^\circ- \theta_{m}^{(w)}/2 from the facet normal.

Table 2 details the mirror configurations for the 19 positive-index orders. The final column ℓ2(m)\ell_{2}^{(m)} defines the minimum facet length required to intercept the full beam width ama_m at grazing incidence: ℓ2(m)=am/sin⁡(θm(w)/2)\ell_{2}^{(m)} = a_m/\sin(\theta_{m}^{(w)}/2). The ray trace is shown in Figure 8.

mmθm\theta_m (°)θm(w)\theta_m^{(w)} (°)x1(m)x_1^{(m)} (mm)z(m)z^{(m)} (mm)ama_m (mm)Lm(w)L_m^{(w)} (mm)α1(m)\alpha_1^{(m)} (°)α2(m)\alpha_2^{(m)} (°)ℓ2(m)\ell_2^{(m)} (mm)
10.753.0013.1−50.310.0010.010.3788.50382.5
21.506.0026.1−51.310.0010.050.7587.00191.0
32.259.0239.2−52.99.9910.121.1285.49127.1
43.0012.0752.3−55.29.9910.211.5083.9795.0
53.7515.1565.5−58.29.9810.341.8782.4375.7
64.5018.2878.6−61.99.9710.502.2580.8662.8
75.2521.4691.8−66.49.9610.702.6279.2753.5
86.0024.72105.1−71.79.9510.953.0077.6446.5
96.7528.06118.4−77.89.9311.253.3875.9741.0
107.5131.51131.8−85.09.9111.633.7574.2536.5
118.2635.09145.2−93.39.9012.094.1372.4532.8
129.0238.84158.8−102.89.8812.684.5170.5829.7
139.7842.80172.4−113.99.8513.434.8968.6027.0
1410.5447.03186.1−126.79.8314.425.2766.4924.6
1511.3051.62199.9−141.79.8115.805.6564.1922.5
1612.0756.74213.8−159.89.7817.836.0361.6320.6
1712.8362.68227.8−182.39.7521.256.4258.6618.7
1813.6070.18242.0−212.89.7228.666.8054.9116.9
1914.3783.23256.3−269.69.6982.137.1948.3914.6

Table 2. Mirror system parameters for the configuration with the wafer below the second mirror (λ=11.2\lambda= 11.2 nm, L=10L = 10 mm, Lx=857L_x = 857 nm, M=76M = 76, N=19N = 19, Z1=1000Z_1 = 1000 mm, Zw=−300Z_w = -300 mm).

Ray tracing in the configuration with the wafer below the second mirror

Figure 8. Ray tracing in the configuration with the wafer below the second mirror (N=19N = 19, showing orders m=1,4,7,10,13,16,19m = 1,4,7,10,13,16,19). The second mirror system is positioned between the mask (z=0z = 0) and wafer (z=−300z = -300 mm). Its facets are nearly vertical (tilt relative to the horizontal 90∘−θm(w)/290^\circ- \theta_m^{(w)}/2) and plotted with full length ℓ2(m)=am/sin⁡(θm(w)/2)\ell_2^{(m)} = a_m/\sin\left(\theta_m^{(w)}/2\right) required to intercept the entire beam (up to 382.5 mm for m=1m = 1).

Discussion. The lateral span of the mirrors along the xx-axis (±256.3\pm256.3 mm) and the maximum numerical aperture NAmax⁡=0.993\mathrm{NA}_{\max} = 0.993 remain identical to Example 1. The beam widths ama_m and footprint dimensions Lm(w)L_m^{(w)} on the wafer are unchanged. The key difference of this configuration is grazing incidence on the second mirror. At sufficiently small grazing angles, a suitable material can provide useful total-external-reflection reflectance without a Bragg stack. The critical angle and absorption depend on the material and wavelength and must be evaluated from the complex refractive index. No universal 15∘15^\circ threshold is assumed. At larger grazing angles, multilayer coatings require optimization for the actual incidence angle. The total optical path length of the central ray from the mask to the wafer increases from 2300.4 mm (m=1m = 1) to 2560.0 mm (m=19m = 19), corresponding to an optical path difference of ≈260\approx260 mm.

The minimum facet length ℓ2(m)=am/sin⁡(θm(w)/2)\ell_{2}^{(m)} = a_m/\sin(\theta_{m}^{(w)}/2) is large for low diffraction orders, reaching 382.5 mm for m=1m = 1, 191.0 mm for m=2m = 2, and 127.1 mm for m=3m = 3. This follows from the grazing-incidence footprint enlargement by 1/sin⁡(θm(w)/2)1/\sin(\theta_{m}^{(w)}/2).

Suggestion 2. To obtain more compact second-mirror facets, place the wafer in the base plane of the first mirror and the mask in the base plane of the second. Furthermore, this layout restores the m=0m = 0 beam to the field reaching the wafer.

Example 3: Wafer in the Plane of Mirror 1, Mask in the Base Plane of Mirror 2

In this example, the wafer is placed in the plane of the first mirror (z=Z1=1000z = Z_1 = 1000 mm) and the mask is located at the base of the second (z=0z = 0). The second mirror consists of a system of facets distributed in height: central facets (small mm) reside near the mask, while peripheral facets (large mm) ascend toward the wafer. The spatial distance from the wafer center to the central facets of the second mirror is on the order of the mask–wafer distance Z1=1000Z_1 = 1000 mm. Figure 9 shows the ray trace for this projection system.

Ray tracing with the wafer in the plane of the first mirror and the mask at the base of the second

Figure 9. Ray tracing with the wafer in the plane of the first mirror and the mask at the base of the second (N=19N = 19, showing orders m=1,4,7,10,13,16,19m = 1, 4, 7, 10, 13, 16, 19). The beam from the mask (z=0z = 0) propagates upward to mirror 1 (z=1000z = 1000 mm), reflects downward and outward at angle (θm(w)−θm)(\theta_m^{(w)} - \theta_m), intercepts the second mirror facet (zfz_f from 2 mm to 956 mm), which directs the beam upward to the center of the wafer (z=Z1z = Z_1) at convergence angle θm(w)\theta_m^{(w)}.

First mirror at z=Z1z = Z_1: facet mm is centered at (±x1(m),Z1)(\pm x_1^{(m)}, Z_1), x1(m)=Z1tan⁡θmx_1^{(m)} = Z_1 \tan\theta_m. Unlike Example 1, the first mirror does not collimate the beams vertically, but deflects them downward and outward:

(sin⁡θm,cos⁡θm)⟶[sin⁡(θm(w)−θm),−cos⁡(θm(w)−θm)].(12)\left(\sin\theta_m,\cos\theta_m\right) \longrightarrow\left[\sin\left(\theta_m^{(w)}-\theta_m\right),-\cos\left(\theta_m^{(w)}-\theta_m\right)\right]. \tag*{(12)}

The angle of incidence on the first mirror is i1(m)=θm(w)/2i_1^{(m)} = \theta_m^{(w)}/2 (1.5∘1.5^\circ–41.6∘41.6^\circ), and the facet tilt is α1(m)=θm(w)/2−θm\alpha_1^{(m)} = \theta_m^{(w)}/2-\theta_m (0.8∘0.8^\circ–27.2∘27.2^\circ relative to the horizontal).

Second mirror: facet mm is centered at (±xf(m),zf(m))(\pm x_f^{(m)},z_f^{(m)}), where

zf(m)=Z1−Z1tan⁡θmtan⁡θm(w)−tan⁡(θm(w)−θm),xf(m)=(Z1−zf(m))tan⁡θm(w).(13)z_f^{(m)} = Z_1-\frac{Z_1\tan\theta_m}{\tan\theta_m^{(w)}-\tan\left(\theta_m^{(w)}-\theta_m\right)}, \qquad x_f^{(m)} = \left(Z_1-z_f^{(m)}\right)\tan\theta_m^{(w)}. \tag*{(13)}

For small mm, tan⁡θm≪tan⁡θm(w)\tan\theta_m \ll\tan\theta_m^{(w)}, yielding zf(m)≈0z_f^{(m)} \approx0 — central facets lie near the mask. For m=19m = 19, zf≈956z_f \approx956 mm — the facet is near the wafer. The second mirror redirects the beam toward the center of the wafer:

[±sin⁡(θm(w)−θm),−cos⁡(θm(w)−θm)]⟶[∓sin⁡θm(w),cos⁡θm(w)].(14)\left[\pm\sin\left(\theta_m^{(w)}-\theta_m\right),-\cos\left(\theta_m^{(w)}-\theta_m\right)\right] \longrightarrow\left[\mp\sin\theta_m^{(w)},\cos\theta_m^{(w)}\right]. \tag*{(14)}

The angle of incidence on the second mirror is i2(m)=θm/2i_{2}^{(m)} = \theta_m/2 (0.4∘0.4^{\circ}–7.2∘7.2^{\circ}), and facet tilt angles are α2(m)\alpha_{2}^{(m)} (2.6∘2.6^{\circ}–76.0∘76.0^{\circ} relative to the horizontal). The facet lengths of the second mirror are compact: ℓ2(m)=am/cos⁡(θm/2)≈10\ell_{2}^{(m)} = a_m/\cos(\theta_m/2) \approx10 mm. Geometric configurations for all facets m=1m = 1–1919 are listed in Table 3.

mmθm\theta_m (°)θm(w)\theta_m^{(w)} (°)x1(m)x_1^{(m)} (mm)xf(m)x_f^{(m)} (mm)zf(m)z_f^{(m)} (mm)i1(m)i_1^{(m)} (°)i2(m)i_2^{(m)} (°)ℓ2(m)\ell_2^{(m)} (mm)PmP_m (mm)
10.753.0013.152.22.01.500.3710.02998.1
21.506.0026.1104.28.23.000.7510.02992.5
32.259.0239.2155.818.54.511.1210.02982.9
43.0012.0752.3206.733.06.031.5010.02969.5
53.7515.1565.5256.751.87.571.8710.02951.8
64.5018.2878.6305.574.99.142.2510.02929.8
75.2521.4691.8352.8102.610.732.6210.02903.1
86.0024.72105.1398.2134.912.363.0010.02871.2
96.7528.06118.4441.3172.114.033.389.92833.8
107.5131.51131.8481.6214.415.753.759.92790.0
118.2635.09145.2518.4262.217.554.139.92739.0
129.0238.84158.8550.9315.819.424.519.92679.6
139.7842.80172.4578.1375.721.404.899.92610.1
1410.5447.03186.1598.4442.623.515.279.92528.3
1511.3051.62199.9609.5517.325.815.659.92430.4
1612.0756.74213.8608.1601.228.376.039.82310.5
1712.8362.68227.8587.6696.531.346.429.82157.6
1813.6070.18242.0533.1807.835.096.809.81944.5
1914.3783.23256.3369.8956.141.617.199.81526.5

Table 3. Mirror system configuration with the wafer in the plane of the first mirror (i1i_1 and i2i_2 are angles of incidence on the first and second mirrors, respectively, PmP_m is total optical path length).

Discussion. The optical path length decreases with order index: from 2998 mm (m=1m = 1) to 1527 mm (m=19m = 19), creating an optical path difference of 1472 mm.

Suggestion 3. Position the facets of the second mirror such that the optical path lengths across all diffraction orders are identical.

Example 4: Equal Optical Path Length, NA ∼1\sim1

In Example 3, optical path lengths PmP_m vary substantially across orders (from 2998 mm for m=1m=1 to 1527 mm for m=19m=19), which introduces order-dependent propagation phase shifts even under coherent monochromatic illumination. In this example, the facet coordinates of the first mirror are preserved as in Example 3 (z1=Z1=constz_1=Z_1=\mathrm{const}, x1(m)=Z1tan⁡θmx_1^{(m)}=Z_1\tan\theta_m), while the facet heights zf(m)z_f^{(m)} of the second mirror and the first-mirror deflection angles βm\beta_m are adjusted so that the total optical path length PmP_m is identical for all diffraction orders (excluding m=0m=0). The maximum positive order index is N=19N=19, and NAmax⁡=0.993\mathrm{NA}_{\max}=0.993, as in Example 3.

The second mirror redirects each beam toward the center of the wafer (0,Z1)(0,Z_1) at angle θm(w)\theta_m^{(w)}. The lateral coordinate is xf(m)=(Z1−zf(m))tan⁡θm(w)x_f^{(m)}=(Z_1-z_f^{(m)})\tan\theta_m^{(w)}. The first mirror reflects the beam at a variable angle βm\beta_m (from 2.2∘2.2^\circ for m=1m=1 to 81.2∘81.2^\circ for m=19m=19), determined by the position of the corresponding facet of the second mirror:

tan⁡βm=xf(m)−x1(m)Z1−zf(m).(15)\tan\beta_m=\frac{x_f^{(m)}-x_1^{(m)}}{Z_1-z_f^{(m)}}. \tag*{(15)}

The total optical path length is:

Pm=Z1cos⁡θm+[(xf(m)−x1(m))2+(Z1−zf(m))2]1/2+Z1−zf(m)cos⁡θm(w).(16)P_m=\frac{Z_1}{\cos\theta_m}+\left[(x_f^{(m)}-x_1^{(m)})^2+(Z_1-z_f^{(m)})^2\right]^{1/2}+\frac{Z_1-z_f^{(m)}}{\cos\theta_m^{(w)}}. \tag*{(16)}

Here, the first term represents the path from the mask to the first mirror (fixed height Z1Z_1, but varying cos⁡θm\cos\theta_m due to differing kz;mk_{z;m}). The second term is the path from the first to the second mirror. The third term is the path from the second mirror to the wafer. The parameter um=Z1−zf(m)u_m=Z_1-z_f^{(m)} is determined via bisection for each order mm to enforce Pm≈2998.1P_m\approx2998.1 mm. As a result, ΔP=Pmax⁡−Pmin⁡=0\Delta P=P_{\max}-P_{\min}=0 (within numerical precision), removing order-dependent propagation phase shifts. Phase differences introduced by the mask and coatings remain. For a source with finite bandwidth, temporal coherence requires a separate assessment.

Table 4: Configuration of the equal-path mirror system (N=19N=19, NA=0.993\mathrm{NA}=0.993). Here β\beta is the reflection angle from mirror 1, i1i_1 and i2i_2 are angles of incidence, ℓ2\ell_2 is the facet length of mirror 2, and PmP_m is the total optical path length.

mmθm(w)\theta_m^{(w)} (°)zfz_f (mm)xfx_f (mm)β\beta (°)i1i_1 (°)i2i_2 (°)ℓ2\ell_2 (mm)PmP_m (mm)
13.002.052.22.21.500.3710.02998.1
26.005.4104.54.53.000.7510.02998.1
39.0211.0157.06.84.521.1110.02998.1
412.0719.0209.79.16.061.4810.02998.1
515.1529.4262.811.57.621.8310.02998.1
618.2842.4316.213.99.222.1710.02998.1
721.4658.3370.216.510.862.5010.02998.1
824.7277.1424.819.112.552.8010.02998.1
928.0699.4480.121.914.323.099.92998.1
1031.51125.4536.224.816.163.359.92998.1
1135.09155.7593.228.018.113.579.92998.1
1238.84191.1651.331.320.183.759.92998.1
1342.80232.5710.735.022.413.889.92998.1
1447.03281.4771.439.224.853.939.92998.1
1551.62339.9833.643.827.573.909.82998.1
1656.74411.4897.549.330.673.739.82998.1
1762.68502.4963.555.934.383.389.82998.1
1870.18628.21031.664.839.192.709.72998.1
1983.23869.11102.181.247.791.019.72998.1

Table 4. Configuration of the equal-path mirror system (N=19N=19, NA=0.993\mathrm{NA}=0.993). Here β\beta is the reflection angle from mirror 1, i1i_1 and i2i_2 are angles of incidence, ℓ2\ell_2 is the facet length of mirror 2, and PmP_m is the total optical path length.

The angles of incidence on the first mirror, i1=(θm+βm)/2i_1=(\theta_m+\beta_m)/2, vary from 1.5∘1.5^\circ to 47.8∘47.8^\circ. On the second mirror, i2i_2 varies from 0.4∘0.4^\circ to 3.9∘3.9^\circ, which is substantially smaller than i2=θm/2i_2=\theta_m/2 (0.4∘0.4^\circ–7.2∘7.2^\circ) in Example 3, since adjusting βm\beta_m reduces the incidence angle on the second mirror for high orders. Facet lengths are ℓ2≈10\ell_2\approx10 mm. Figure 10 shows the ray trace of the proposed equal-path projection system.

Ray tracing in the equal-path projection system

Figure 10. Ray tracing in the equal-path projection system (N=19N = 19, NA = 0.993, showing orders m=1,4,7,10,13,16,19m = 1,4,7,10,13,16,19). The first mirror lies in the plane z=Z1=1000z = Z_1 = 1000 mm. The deflection angle βm\beta_m varies from 2.2∘2.2^\circ to 81.2∘81.2^\circ. The second mirror is profiled such that total optical path length Pm≈2998.1P_m \approx2998.1 mm across all orders.

Bragg Coatings of Mirror Facets for Example 4

Each of the 2N=382N = 38 facets of the first and second mirrors in Example 4 operates at its own specific angle of incidence. Specifically, the angle of incidence on the first mirror i1(m)i_1^{(m)} ranges from 1.50∘1.50^\circ (m=1m = 1) to 47.79∘47.79^\circ (m=19m = 19), while i2(m)i_2^{(m)} on the second mirror ranges from 0.37∘0.37^\circ to 3.93∘3.93^\circ (see Tables 5 and 6, column θinc\theta_{\mathrm{inc}}). A standard Mo/Si Bragg coating (40–50 bilayers) provides maximum reflectance near a single fixed design angle. A single uniform coating cannot maintain high reflectance across the entire 0.4∘0.4^\circ–47.8∘47.8^\circ angular range. Therefore, individually optimized Bragg coatings maximizing reflectance at the exact operational angle of incidence must be designed for each facet.

Formulation of the Facet Bragg Optimization Problem

For each facet specified by mirror index j∈{1,2}j \in\{1,2\} and diffraction order m∈{1,…,19}m \in\{1,\ldots,19\}, a target angle of incidence θtarget=ij(m)\theta_{\mathrm{target}} = i_j^{(m)} (measured from the facet normal) is assigned. We seek to determine the layer thicknesses d1d_1 (absorbing material) and d2d_2 (spacer material) of a periodic Bragg mirror comprising Nb=30N_b = 30 bilayers (i.e., 2Nb=602N_b = 60 individual layers total) that maximize the reflectance R=∣r∣2R = \lvert r\rvert^2 at incident angle θtarget\theta_{\mathrm{target}}:

(d1∗,d2∗)=arg⁡max⁡d1,d2  R(d1,d2;θtarget).(17)(d_1^*,d_2^*) = \underset{d_1,d_2}{\arg\max}\; R(d_1,d_2;\theta_{\mathrm{target}}). \tag*{(17)}

Here, r=r(d1,d2;θ)r = r(d_1,d_2;\theta) is the complex field reflection coefficient calculated via the transfer-matrix method (TMM) for TE polarization. The multilayer structure is: vacuum / layer 1(d1)1(d_1) / layer 2(d2)2(d_2) / ⋯\cdots / layer 1(d1)1(d_1) / layer 2(d2)2(d_2) / vacuum (60 layers total, bounded by vacuum with ε=1\varepsilon= 1 on both sides). Optical constants (ε=(1−δ−iβ)2\varepsilon= (1-\delta-i\beta)^2, where δ\delta is the refractive index decrement and β\beta is the absorption index) are obtained from tabulated reference data [18, 19].

Transfer-matrix method (TMM). At incidence angle θ\theta, layer ii has longitudinal wavenumber kz(i)=k0(εi−sin⁡2θ)1/2k_z^{(i)} = k_0(\varepsilon_i-\sin^2\theta)^{1/2} and TE admittance pi=kz(i)/k0p_i = k_z^{(i)}/k_0. The Fresnel coefficients are rij=(pi−pj)/(pi+pj)r_{ij} = (p_i-p_j)/(p_i+p_j) and tij=2pi/(pi+pj)t_{ij}=2p_i/(p_i+p_j). Define the entering-interface matrix and the propagation matrix of layer ii by

Di−1,i=1ti−1,i(1ri−1,iri−1,i1),\mathbf{D}_{i-1,i}=\frac{1}{t_{i-1,i}}\begin{pmatrix}1&r_{i-1,i}\\r_{i-1,i}&1\end{pmatrix},
Pi=diag⁡(eiϕi,e−iϕi),ϕi=kz(i)di,\mathbf{P}_i=\operatorname{diag}(e^{i\phi_i},e^{-i\phi_i}),\qquad\phi_i=k_z^{(i)}d_i,
M=D0,1P1D1,2P2⋯D59,60P60D60,sub.(18)\mathbf{M}=\mathbf{D}_{0,1}\mathbf{P}_1\mathbf{D}_{1,2}\mathbf{P}_2\cdots\mathbf{D}_{59,60}\mathbf{P}_{60}\mathbf{D}_{60,\mathrm{sub}}. \tag*{(18)}

Here medium 0 and the substrate are vacuum. The matrices are multiplied in the written order, mapping substrate amplitudes to entrance amplitudes. The complex reflection coefficient is r=M21/M11r=M_{21}/M_{11} and R=∣r∣2R=\lvert r\rvert^2.

Optimization algorithm. The objective function f(d1,d2)=−R(d1,d2;θtarget)f(d_1,d_2)=-R(d_1,d_2;\theta_{\mathrm{target}}) is minimized via a two-stage procedure:

  1. Local optimization using the Nelder–Mead simplex algorithm, initialized near the first-order Bragg condition: d1Re⁡(ε1−sin⁡2θ)1/2+d2Re⁡(ε2−sin⁡2θ)1/2=λ/2d_1\operatorname{Re}(\varepsilon_1-\sin^2\theta)^{1/2}+d_2\operatorname{Re}(\varepsilon_2-\sin^2\theta)^{1/2}=\lambda/2.

  1. Global optimization using the differential evolution algorithm (differential_evolution in SciPy, population-size multiplier popsize=15 (30 candidates for two variables), up to 80 iterations) within bounds d1∈[0.3,5]d_1\in[0.3,5] nm, d2∈[0.3,9]d_2\in[0.3,9] nm, followed by quasi-Newton refinement using the L-BFGS-B method.

The parameters yielding the highest RR for the target angle are retained.

Configuration 1: λ=13.5\lambda=13.5 nm, Mo/Si

The absorbing material is Mo (δ=0.0763\delta=0.0763, β=0.0064\beta=0.0064) and the spacer is Si (δ=0.0010\delta=0.0010, β=0.0018\beta=0.0018). To retain the angles and numerical aperture of Example 4 at this wavelength, the mask period is scaled in proportion to λ\lambda, giving Lx=8λ/sin⁡6∘≈1033.2L_x=8\lambda/\sin6^\circ\approx1033.2 nm. At fixed Lx≈857L_x\approx857 nm the maximum wafer index would instead be N=15N=15. Optimization results for the 38 tabulated facets are summarized in Table 5.

Mirrormmθinc\theta_{\mathrm{inc}} (°)dMod_{\mathrm{Mo}} (nm)dSid_{\mathrm{Si}} (nm)Re⁡r\operatorname{Re}rIm⁡r\operatorname{Im}rR=∣r∣2R=\lvert r\rvert^2
111.502.99833.93480.7102120.4613360.717232
123.003.00003.94040.7097570.4623580.717530
134.523.00293.95000.7089820.4640860.718032
146.063.00713.96360.7078650.4665490.718741
157.623.01253.98160.7063790.4697710.719656
169.223.01944.00450.7044670.4738330.720792
1710.863.02784.03280.7020790.4787840.722149
1812.553.03804.06710.6991350.4847120.723736
1914.323.05034.10890.6954950.4918020.725582
11016.163.06504.15900.6910650.5001020.727673
11118.113.08264.21970.6856060.5098920.730045
11220.183.10394.29330.6789030.5213330.732697
11322.413.13004.38370.6705850.5347580.735561
11424.853.16284.49670.6601490.5505710.738925
11527.573.20564.64100.6468930.5692620.742530
11630.673.26434.83110.6298740.5913670.746457
11734.383.35325.09750.6074300.6178730.750738
11839.193.51245.51330.5768920.6501150.755454
11947.794.00656.50820.5258390.6969830.762293
210.372.99783.93310.7103530.4610160.717138
220.752.99793.93340.7103250.4610810.717157
231.112.99803.93400.7102800.4611820.717186
241.482.99833.93480.7102160.4613270.717229
251.832.99863.93570.7101380.4615030.717280
262.172.99893.93690.7100460.4617080.717340
272.502.99933.93810.7099430.4619410.717409
282.802.99973.93950.7098360.4621820.717479
Mirrormmθinc\theta_{\mathrm{inc}} (°)dMod_{\mathrm{Mo}} (nm)dSid_{\mathrm{Si}} (nm)Re⁡r\operatorname{Re} rIm⁡r\operatorname{Im} rR=∣r∣2R = \lvert r\rvert^2
293.093.00023.94090.7097200.4624400.717554
2103.353.00063.94230.7096070.4626940.717628
2113.573.00103.94360.7095040.4629240.717695
2123.753.00133.94470.7094150.4631240.717753
2133.883.00163.94550.7093480.4632730.717796
2143.933.00173.94580.7093210.4633320.717814
2153.903.00163.94560.7093370.4632970.717803
2163.733.00133.94450.7094250.4631000.717746
2173.383.00063.94250.7095940.4627240.717637
2182.702.99963.93900.7098730.4620990.717454
2191.012.99803.93380.7102940.4611500.717177

Table 5. Layer thicknesses of optimized Bragg mirrors for Example 4 (λ=13.5\lambda=13.5 nm, Mo/Si, 60 layers, TE polarization). dMod_{\mathrm{Mo}} and dSid_{\mathrm{Si}} are layer thicknesses, rr is complex field reflection coefficient, R=∣r∣2R=\lvert r\rvert^2 is power reflectance at target incidence angle θinc\theta_{\mathrm{inc}}.

For the first mirror, the Mo layer thickness varies from 3.00 nm (small angles) to 4.01 nm (m=19m = 19, θ=47.8∘\theta= 47.8^\circ), while the Si layer thickness varies from 3.93 to 6.51 nm. The bilayer period d=dMo+dSid = d_{\mathrm{Mo}} + d_{\mathrm{Si}} grows from 6.93 nm at normal incidence to 10.51 nm at θ=47.8∘\theta= 47.8^\circ, following the trend of the approximate Bragg condition 2dcos⁡θ≈λ2d \cos\theta\approx\lambda. Refraction is included through the layer-dependent expression used to initialize the optimization above. The reflectance RR rises from 0.717 (m=1m = 1) to 0.762 (m=19m = 19) — as the angle of incidence increases, the effective penetration depth into the absorbing medium decreases, enhancing RR. For the second mirror, all incidence angles remain small (0.37∘0.37^\circ–3.93∘3.93^\circ), so optimal thicknesses are virtually constant (dMo≈3.00d_{\mathrm{Mo}} \approx3.00 nm, dSi≈3.94d_{\mathrm{Si}} \approx3.94 nm), and RR varies only in the fourth decimal place (0.7171–0.7178). The small variation suggests that a common second-mirror coating can be a useful approximation. Its performance must be evaluated separately from the individually optimized designs tabulated here.

Configuration 2: λ=11.2\lambda= 11.2 nm, Ru/Be

The absorbing material is Ru (δ=0.0660\delta= 0.0660, β=0.0065\beta= 0.0065) and the spacer is Be (δ=−0.0122\delta= -0.0122, β=0.0013\beta= 0.0013). A negative decrement indicates that the real part of the refractive index of Be at this wavelength exceeds unity, which occurs near the Be KK-absorption edge at λ≈11.3\lambda\approx11.3 nm. Results are summarized in Table 6.

Mirrormmθinc\theta_{\mathrm{inc}} (°)dRud_{\mathrm{Ru}} (nm)dBed_{\mathrm{Be}} (nm)Re⁡r\operatorname{Re} rIm⁡r\operatorname{Im} rR=∣r∣2R = \lvert r\rvert^2
111.502.36803.29830.6690320.5486990.748674
123.002.36903.30300.6683030.5498630.748979
134.522.37063.31120.6670610.5518350.749491
146.062.37303.32280.6652740.5546400.750215
157.622.37613.33820.6629050.5583080.751151
169.222.38013.35770.6598690.5629270.752314
1710.862.38483.38170.6560920.5685510.753707
1812.552.39063.41100.6514590.5752750.755340
1914.322.39743.44660.6457570.5833040.757246
11016.162.40553.48930.6388550.5926850.759412
11118.112.41523.54110.6303950.6037260.761883
11220.182.42663.60390.6200540.6165990.764661
11322.412.44053.68100.6072720.6316650.767780
11424.852.45783.77730.5912680.6493640.771271
11527.572.47993.90050.5709210.6702340.775164
11630.672.50994.06290.5446310.6948710.779469
11734.382.55504.29100.5094010.7244000.784245
11839.192.63674.64710.4598100.7603440.789548
11947.792.90835.48540.3735910.8104250.796359
210.372.36773.29680.6692600.5483340.748579
220.752.36773.29710.6692140.5484070.748598
231.112.36783.29750.6691420.5485230.748628
241.482.36803.29820.6690390.5486880.748672
251.832.36813.29900.6689140.5488880.748724
262.172.36833.30000.6687670.5491230.748786
Mirrormmθinc\theta_{\mathrm{inc}} (°)dRud_{\mathrm{Ru}} (nm)dBed_{\mathrm{Be}} (nm)Re⁡r\operatorname{Re} rIm⁡r\operatorname{Im} rR=∣r∣2R=\lvert r\rvert^2
272.502.36863.30110.6686010.5493890.748855
282.802.36883.30220.6684290.5496630.748927
293.092.36913.30340.6682440.5499580.749003
2103.352.36933.30460.6680620.5502470.749079
2113.572.36953.30570.6678970.5505090.749147
2123.752.36973.30660.6677540.5507360.749206
2133.882.36993.30730.6676460.5509080.749251
2143.932.36993.30760.6676040.5509750.749268
2153.902.36993.30750.6676300.5509340.749258
2163.732.36973.30650.6677700.5507110.749199
2173.382.36933.30480.6680400.5502820.749088
2182.702.36873.30180.6684880.5495680.748902
2191.012.36783.29740.6691650.5484860.748619

Table 6. Layer thicknesses of optimized Bragg mirrors for Example 4 (λ=11.2\lambda= 11.2 nm, Ru/Be, 60 layers, TE polarization). Notation as in Table 5.

The Ru/Be material combination at λ=11.2\lambda=11.2 nm provides markedly higher reflectance than Mo/Si at λ=13.5\lambda=13.5 nm. RR on the first mirror ranges from 0.749 (m=1m=1) to 0.796 (m=19m=19), which is 3–4 percentage points higher than for Mo/Si (0.717–0.762). This is attributable to the lower absorption of Be (βBe=0.0013\beta_{\mathrm{Be}}=0.0013) relative to Si (βSi=0.0018\beta_{\mathrm{Si}}=0.0018) and the larger refractive index contrast of Ru/Be. As with Mo/Si, for the second mirror (small angles 0.37∘0.37^\circ–3.93∘3.93^\circ) optimal layer thicknesses are practically constant: dRu≈2.37d_{\mathrm{Ru}}\approx2.37 nm, dBe≈3.30d_{\mathrm{Be}}\approx3.30 nm, R≈0.749R\approx0.749.

Discussion

Maximizing reflectance over layer thicknesses at a fixed target angle does not force the maximum of the angular reflectance curve to occur at that angle. For example, for m=1m=1 on the second mirror, the target is 0.37∘0.37^\circ, whereas the local near-normal maxima of the tabulated Mo/Si and Ru/Be designs occur at approximately 1.90∘1.90^\circ and 2.00∘2.00^\circ, respectively. The tabulated reflectances are evaluated at the actual target angles. The angular-peak offset is not an optimization constraint.

Comparison with the typical reflectance R∼70%R\sim70\% of standard Mo/Si coatings (50 bilayers, λ=13.5\lambda=13.5 nm, incident angle ∼6∘\sim6^\circ) indicates that individually optimized 30-bilayer coatings achieve comparable or superior reflectance (R=71.7%R=71.7\%–76.2%76.2\%) with fewer layers in the ideal model. This comparison is not an experimental coating-performance validation, because the calculation omits interface roughness, interdiffusion, and fabrication errors. For Ru/Be at λ=11.2\lambda=11.2 nm, reflectance is even higher (74.9%–79.6%). The fraction of the power leaving the mask in an accepted order that is retained after the two modeled reflections is R1R2≈51.5%R_1R_2\approx51.5\% (Mo/Si, low angles) to 59.6% (Ru/Be, m=19m=19), exceeding the throughput of 6- and 10-mirror systems (12% and 2.8%, respectively) by up to a factor of ∼20\sim20.

Mask Optimization for a Target Aerial Image on the Wafer

This section is dedicated to mask optimization for a prescribed field profile on the wafer in the synthesized projection system. Mask optimization is conducted using approaches developed in [26].

Formulation of the Inverse Problem

In Example 4, the optical path lengths of the channels with 1≤n≤N=191\leq n\leq N=19 are equalized (Pn≈2998.1P_n\approx2998.1 mm), removing order-dependent propagation phase shifts. The relative phases at the wafer still depend on the mask amplitudes and on the complex reflection coefficients of both Bragg coatings. Let AnA_n denote the complex electric-field amplitude of mask order nn, and let r1,nr_{1,n} and r2,nr_{2,n} denote the corresponding facet reflection coefficients. With the common propagation phase omitted and amplitudes referenced to ζw=0\zeta_w=0, the wafer coefficient is

B−n=r2,nr1,nAn.(19)B_{-n}=r_{2,n}r_{1,n}A_n. \tag*{(19)}

Coefficients r1,nr_{1,n} and r2,nr_{2,n} were determined previously (Table 6) for λ=11.2\lambda=11.2 nm, Ru/Be. The field on the wafer is evaluated via (3). Let E(d)\mathbf{E}^{(d)} and I(d)=∣E(d)∣2I^{(d)}=\lvert\mathbf{E}^{(d)}\rvert^2 denote the target electric field distribution and intensity on the wafer. The objective of Inverse Lithography Technology (ILT) is to determine the spatial permittivity distribution εj(x)\varepsilon_j(x) of the mask layers such that the projected aerial intensity matches the target intensity:

∣E(w)∣2=∣E(d)∣2.(20)\left|\mathbf{E}^{(w)}\right|^2 = \left|\mathbf{E}^{(d)}\right|^2 . \tag*{(20)}

We consider a mask comprising a single absorber layer with permittivity ε(x)\varepsilon(x) positioned atop multilayer mirror layers that are homogeneous along xx.

Forward Problem

Scattering amplitudes AnA_n for a given profile ε(x)\varepsilon(x) are evaluated using the modal waveguide method (equivalent to RCWA). The mask consists of an absorber layer of thickness dabsd_{\mathrm{abs}} with periodically modulated permittivity ε(x)=∑mεmexp⁡(−iκxmx)\varepsilon(x) = \sum_m \varepsilon_m \exp(-i\kappa_{x m}x) (where εm\varepsilon_m are Fourier coefficients), situated on a Bragg mirror (Ru/Be/Sr, 30 periods) [20]. The solution proceeds as follows.

The electric field inside the absorber layer is represented as Ey(x,z)=X(x)Z(z)E_y(x,z) = X(x)Z(z). The transverse function X(x)X(x) is expanded in a Fourier series over plane waves ψm=exp⁡(−iκxmx)\psi_m = \exp(-i\kappa_{x m}x), and the longitudinal function is Z(z)=exp⁡(±ikzz)Z(z) = \exp(\pm i k_z z).

Substituting into the Helmholtz equation yields the algebraic eigenvalue problem:

D^Bp=kz;p2Bp,D^nm=k02εn−m−(κxn)2δnm,(21)\hat{D}\mathbf{B}_p = k_{z;p}^{2}\mathbf{B}_p,\qquad\hat{D}_{nm} = k_0^2\varepsilon_{n-m} - (\kappa_{x n})^2\delta_{nm}, \tag*{(21)}

where kz;p2k_{z;p}^{2} are the eigenvalues and kz;pk_{z;p} are the longitudinal wavenumbers of the modes, and Bp\mathbf{B}_p are the eigenvectors (components Bp,mB_{p,m} define the contribution of the mm-th plane wave to the pp-th mode). Matrix D^\hat{D} has dimensions (2M+1)×(2M+1)(2M+1) \times(2M+1).

The field in each layer is expressed via modal expansions:

Ey(j)(x,z)=k0∑p=−MM[Ap;1(j)eikz;p(j)z+Ap;2(j)e−ikz;p(j)z]∑m=−MMBp,m(j)ψm.(22)E_y^{(j)}(x,z) = k_0 \sum_{p=-M}^{M} \left[A_{p;1}^{(j)}e^{i k_{z;p}^{(j)}z} + A_{p;2}^{(j)}e^{-i k_{z;p}^{(j)}z}\right]\sum_{m=-M}^{M} B_{p,m}^{(j)}\psi_m. \tag*{(22)}

Coefficients Ap;1(j)A_{p;1}^{(j)}, Ap;2(j)A_{p;2}^{(j)} are determined from tangential field continuity (EyE_y and HxH_x) across layer boundaries, yielding the linear system:

M^X=R,(23)\hat{M}\mathbf{X} = \mathbf{R}, \tag*{(23)}

where X\mathbf{X} is the vector of unknown modal coefficients (reflected waves and waves transmitted into and through the mirror), and R\mathbf{R} describes the incident wave. Solving (23) yields modal coefficients that must be converted to electric-field amplitudes. In the implemented normalization, An=−k0Xn(r)A_n = -k_0 X_n^{(r)} (n=−M,…,Mn=-M,\ldots,M), where Xn(r)X_n^{(r)} is the reflected-wave block of X\mathbf{X}. The minus sign follows its polarization convention. Here and in (19), AnA_n denotes the converted electric-field amplitude, not the raw modal coefficient. The reflected field above the mask (z>0z>0) is:

Ey(r)(x,z)=∑n=−MMAnexp⁡(−iκxnx−ikz;nz).(24)E_y^{(r)}(x,z) = \sum_{n=-M}^{M} A_n \exp(-i\kappa_{x n}x - i k_{z;n}z). \tag*{(24)}

Optimization via Fourier-Parameterized Projection

To solve the inverse problem of finding ε(x)\varepsilon(x) from the target wafer aerial image, we apply gradient-based optimization with Fourier-parameterized mask projection [26].

Instead of optimizing hundreds of discrete pixels, a smooth latent function is introduced:

flatent(x)=Re⁡(∑m=−MMFme−iκxmx),(25)f_{\mathrm{latent}}(x) = \operatorname{Re}\left(\sum_{m=-M}^{M} F_m e^{-i\kappa_{x m}x}\right), \tag*{(25)}

where FmF_m are the complex Fourier coefficients to be optimized. The physical absorber density is obtained via a steepened sigmoid mapping:

ρ(x)=σ(βflatent(x))∈(0,1),(26)\rho(x) = \sigma\left(\beta f_{\mathrm{latent}}(x)\right) \in(0,1), \tag*{(26)}

where β\beta is a steepness parameter increased during optimization to enforce binarization. The permittivity is computed as:

ε(x)=εvac+ρ(x)(εabs−εvac),(27)\varepsilon(x) = \varepsilon_{\mathrm{vac}} + \rho(x)(\varepsilon_{\mathrm{abs}}-\varepsilon_{\mathrm{vac}}), \tag*{(27)}

where εvac=1\varepsilon_{\mathrm{vac}} = 1 and εabs\varepsilon_{\mathrm{abs}} is the absorber permittivity. The nonlinearity of σ\sigma allows a finite set of FmF_m to generate an infinite high-frequency spectrum for ε(x)\varepsilon(x), suppressing the Gibbs phenomenon.

The inverse problem is to minimize the loss functional measuring the discrepancy between the intensities of the reflected and target fields on the wafer:

ε(x)=arg min⁡ε L(ε),(28)\varepsilon(x) = \underset{\varepsilon}{\operatorname{arg\,min}}\ \mathcal{L}(\varepsilon), \tag*{(28)}

where the total loss is:

L=Lshape+Lbin+LHF+Lsupp.(29)\mathcal{L} = \mathcal{L}_{\mathrm{shape}} + \mathcal{L}_{\mathrm{bin}} + \mathcal{L}_{\mathrm{HF}} + \mathcal{L}_{\mathrm{supp}}. \tag*{(29)}

Here:

  • Lshape=MSE⁡(Iwnorm,Itargetnorm)\mathcal{L}_{\mathrm{shape}} = \operatorname{MSE}(I_w^{\mathrm{norm}}, I_{\mathrm{target}}^{\mathrm{norm}}) is the mean squared error between normalized wafer intensity and target profile, where Iw(x)=∣Ey(w)(x)∣2I_w(x) = \lvert E_y^{(w)}(x)\rvert^2 with B−n=r2,nr1,nAnB_{-n} = r_{2,n}r_{1,n}A_n.

  • Lbin=⟨ρ(1−ρ)⟩\mathcal{L}_{\mathrm{bin}} = \langle\rho(1-\rho)\rangle penalizes intermediate grayscale density values, promoting binary solutions (ρ→0\rho\to0 or 11).

  • LHF=wHF∑m=1NFRm∣Fm∣2\mathcal{L}_{\mathrm{HF}} = w_{\mathrm{HF}}\sum_{m=1}^{N_{\mathrm{FR}}}m\lvert F_m\rvert^2 penalizes unphysical high-frequency spatial components, discouraging fine-scale structure without imposing a minimum manufacturable feature size (where NFR=MN_{\mathrm{FR}} = M is the Fourier truncation order).

  • Lsupp=wsupp(α)∣A−m~∣2\mathcal{L}_{\mathrm{supp}} = w_{\mathrm{supp}}(\alpha)\lvert A_{-\tilde{m}}\rvert^2 suppresses scattering into the unused back-reflection order −m~-\tilde{m} that coincides with the input illumination aperture (this term is not present in [26]).

In our experiments, the target intensity I(d)I^{(d)} is binary, with unity in target regions and zero elsewhere. In the global wafer coordinates we prescribe Ey(d)(x)=I(d)(x)exp⁡(+4ik0xx)E_y^{(d)}(x) = \sqrt{I^{(d)}(x)}\exp(+4\mathrm{i}k_{0x}x), where k0x=k0sin⁡6∘=m~κ~xk_{0x} = k_0\sin6^\circ= \tilde{m}\tilde{\kappa}_x. The numerical plots use the reversed wafer coordinate xp=−xx_p=-x (and yp=−yy_p=-y in 3D), so the implemented phase is exp⁡(−4ik0xxp)\exp(-4\mathrm{i}k_{0x}x_p). This convention accounts for the inversion of the transverse wavevector by the projection geometry. Decomposing the target field into spatial harmonics and retaining the selected wafer orders yields the bandlimited target intensity I~(d)\tilde{I}^{(d)} used in (29). The imposed phase is a modeling choice that affects this filtered target.

The optimization procedure runs as follows: for the first 40% of epochs, only the image shape is optimized (Lbin=0\mathcal{L}_{\mathrm{bin}} = 0, β=1\beta= 1). Subsequently, β\beta is annealed up to ∼15\sim15 while the weight of Lbin\mathcal{L}_{\mathrm{bin}} increases to 5, driving the mask toward a strict binary profile without violating physical admissibility. Optimization starts from an unconstrained latent field and converges to a physical density ρ(x)\rho(x) that, after hard thresholding at 0.5, defines the synthesized mask layout.

We use the Adam optimizer with a learning rate of 5×10−35 \times10^{-3} for 500 epochs. Gradients are computed via automatic differentiation through the full physical solver: ∂L/∂Θ=(∂L/∂E(r))(∂E(r)/∂ε)(∂ε/∂Θ)\partial\mathcal{L}/\partial\Theta= (\partial\mathcal{L}/\partial E^{(r)})(\partial E^{(r)}/\partial\varepsilon)(\partial\varepsilon/\partial\Theta), where Θ={Fm}\Theta= \{F_m\}.

Optimization was conducted for λ=11.2\lambda= 11.2 nm. Mask and projection parameters:

  • Wavelength λ=11.2\lambda= 11.2 nm, incidence angle 6∘6^\circ, TE polarization.

  • Mask period Lx≈857L_x \approx857 nm, wafer period Lx(w)=Lx/4≈214L_x^{(w)} = L_x/4 \approx214 nm.

  • Maximum harmonic index on the mask M=25M = 25, on the wafer N=19N = 19.

  • Mask absorber: La (δ=−0.0440\delta= -0.0440, β=0.0159\beta= 0.0159) with thickness dabs=60d_{\mathrm{abs}} = 60 nm.

  • Bragg mirror beneath the absorber: 30 periods of Ru/Be/Sr (dRu=1.7d_{\mathrm{Ru}} = 1.7 nm, dBe=2.7d_{\mathrm{Be}} = 2.7 nm, dSr=1.34d_{\mathrm{Sr}} = 1.34 nm).

  • Mirror reflection coefficients r1,n,r2,nr_{1,n}, r_{2,n} taken from Table 6 (Ru/Be, 30 bilayers).

  • Target profile on the wafer: intensity maximum at x=0x = 0 within the cell [−107,107][-107,107] nm.

  • Fourier expansion parameter Mopt=M=25M_{\mathrm{opt}} = M = 25 (the truncation order of the latent-function expansion, NFRN_{\mathrm{FR}} in (29)), 500 epochs.

Results of Optimizations

The unfiltered target intensity is non-zero over x∈[−2.3,2.3]x \in[-2.3, 2.3] nm, corresponding to a rectangular width of 4.6 nm. The full width at half maximum (FWHM) estimated from the plotted bandlimited target is approximately 5.1 nm, whereas that of the synthesized central peak is approximately 5.4 nm. These widths differ from the unfiltered target width and are not fixed a priori to λ/2\lambda/2.

fig:11a illustrates the optimized continuous absorber density ρ(x)\rho(x). In fig:11b, the binarized mask layout (ρ(x)∈{0,1}\rho(x) \in\{0,1\}) is shown. The mask period Lx≈857L_x \approx857 nm contains several absorber features synthesized to shape the required diffraction spectrum, whose fabrication by electron-beam lithography would require separate verification of minimum widths, gaps, and fabrication tolerances.

Optimized continuous absorber density and binarized mask profile

Figure 11. (a) Absorber density distribution ρ(x)\rho(x) in the optimized mask (λ=11.2\lambda= 11.2 nm, La absorber, 500 epochs). (b) Binarized mask profile (ρ∈{0,1}\rho\in\{0,1\} at threshold ρ>0.5\rho> 0.5).

Figure 12 compares the target aerial image with the optimized result across [−Lx/2,Lx/2]=[−428.6,428.6][-L_x/2, L_x/2] = [-428.6, 428.6] nm. The field repeats with period Lx(w)=214L_x^{(w)} = 214 nm. The peak profile is reproduced with low error (Lshape=1.6×10−5\mathcal{L}_{\mathrm{shape}} = 1.6 \times10^{-5}). The peak intensity ratio reaches max⁡I/Iinc=0.947\max I/I_{\mathrm{inc}} = 0.947 (94.7% of incident intensity), which exceeds the baseline two-reflection reflectance R2≈(0.749)2≈56%R^2 \approx(0.749)^2 \approx56\% due to constructive interference among the diffracted orders.

Aerial intensity on the wafer and zoomed central profile

Figure 12. Aerial intensity on the wafer across [−Lx/2,Lx/2][-L_x/2,L_x/2]: target profile (dashed line), bandlimited target with propagating orders (dash-dotted line), and optimized result (solid line). The peak intensity ratio is max⁡I/Iinc=0.947\max I/I_{\mathrm{inc}}=0.947. Periodicity Lx(w)=214L_x^{(\mathrm{w})}=214 nm.

Figure 13 depicts the harmonic amplitude spectra ∣Am∣|A_m| (scattered by the mask) and ∣B−m∣=∣r2,mr1,mAm∣|B_{-m}| = |r_{2,m}r_{1,m}A_m| (incident on the wafer). The specular reflection order m=8m = 8 (coinciding with the incident-beam order m~=8\widetilde{m} = 8) dominates, while higher orders decay with ∣m∣|m|. The coefficients B−mB_{-m} are additionally modulated by r1,mr_{1,m} and r2,mr_{2,m} and largely follow the behavior of ∣Am∣|A_m|. The plotted wafer spectrum is indexed by its originating mask order mm, so its coefficient is B−mB_{-m} in the global-coordinate convention.

Field amplitude spectra

Figure 13. Field amplitude spectra: (a) is ∣Am∣|A_m| scattered by the mask, (b) is the wafer spectrum indexed by the originating mask order, corresponding to ∣B−m∣|B_{-m}| in global coordinates.

We next evaluated mask synthesis for an aerial image with two closely spaced intensity peaks in Example 4, targeting two 6-nm-wide lines separated by a 6-nm space within the periodic cell. fig:14a shows the simulated image and the synthesized binary mask. These are aerial-image calculations, not a resist-printing simulation.

To evaluate sensitivity to axial wafer displacement, the field was computed for wafer displacements Δz∈[0,5]\Delta z \in[0,5] nm. As shown in fig:14c, the two peaks remain resolved for the tested defocus values. The calculation does not establish a complete focus–dose process window.

3D Two-Mirror Projection System

Problem Formulation and Basic Equations

The preceding analysis addressed a 2D formulation (TE polarization, E∥y\mathbf{E}\parallel\mathbf{y}, mask invariant along yy). In three dimensions, the mask represents a 2D periodic structure with spatial periods LxL_x along xx and LyL_y along yy (assuming Lx=LyL_x=L_y, a square lattice). The mask occupies the domain [−L/2,L/2]×[−L/2,L/2]×[−D,0][-L/2,L/2]\times[-L/2,L/2]\times[-D,0] in Cartesian coordinates (x,y,z)(x,y,z). A TE-polarized beam is incident at 6∘6^\circ to the zz-axis in the xzxz-plane: k0=(k0sin⁡(π/30),0,−k0cos⁡(π/30))\mathbf{k}_0=(k_0\sin(\pi/30),0,-k_0\cos(\pi/30)) (the beam propagates toward the mask, in the −z-z direction).

Normalized aerial intensity, synthesized binary mask profile, and wafer aerial intensity under defocus

Figure 14. (a) Normalized aerial intensity on the wafer across [−Lx/8,Lx/8][-L_x/8,L_x/8]: target profile (dashed), bandlimited target (dash-dotted), and optimized result (solid). The peak intensity ratio is max⁡I/Iinc=0.818\max I/I_{\mathrm{inc}} = 0.818. Periodicity Lx(w)=214L_x^{(w)} = 214 nm. (b) Synthesized binary mask profile (ρ∈{0,1}\rho\in\{0,1\}). (c) Normalized wafer aerial intensity under wafer defocus along the zz-axis for Δz=0,…,5\Delta z = 0,\ldots,5 nm.

The transverse component of the scattered electric field outside the mask is expanded in a double Fourier series:

\begin{bmatrix} E_x^{(r)} \\ E_y^{(r)} \end{bmatrix} = \sum_{m=-\infty}^{\infty} \sum_{n=-\infty}^{\infty} \begin{bmatrix} A_{x;m,n}^{(r)} \\ A_{y;m,n}^{(r)} \end{bmatrix} \exp\left(-\mathrm{i}\kappa_x m x-\mathrm{i}\kappa_y n y-\mathrm{i}k_{z;m,n}z\right) \tag*{(30)}\end{bmatrix}

where Ax;m,n(r)A_{x;m,n}^{(r)} and Ay;m,n(r)A_{y;m,n}^{(r)} are the amplitudes of the (m,n)(m,n)-th harmonic, the superscript (r)(r) denotes the reflected (mask-side) field, κx=2π/Lx\kappa_x = 2\pi/L_x, κy=2π/Ly\kappa_y = 2\pi/L_y, and kz;m,n=(k02−κx2m2−κy2n2)1/2k_{z;m,n} = \left(k_0^2-\kappa_x^2m^2-\kappa_y^2n^2\right)^{1/2} (Im⁡kz;m,n≤0\operatorname{Im} k_{z;m,n} \le0). Propagating orders satisfy κx2m2+κy2n2≤k02\kappa_x^2m^2+\kappa_y^2n^2 \le k_0^2 (M=⌊k0/κx⌋M = \lfloor k_0/\kappa_x \rfloor, the maximum order along each axis). The propagating field in the GO beam approximation is:

\begin{bmatrix} E_x^{(r)} \\ E_y^{(r)} \end{bmatrix} = \sum_{\kappa_x^2m^2+\kappa_y^2n^2\le k_0^2} \begin{bmatrix} \widetilde{A}_{x;m,n}^{(r)}(x,y,z) \\ \widetilde{A}_{y;m,n}^{(r)}(x,y,z) \end{bmatrix} \exp\left(-\mathrm{i}\kappa_x m x-\mathrm{i}\kappa_y n y-\mathrm{i}k_{z;m,n}z\right) \tag*{(31)}\end{bmatrix}

where A~x;m,n(r)\widetilde{A}_{x;m,n}^{(r)} and A~y;m,n(r)\widetilde{A}_{y;m,n}^{(r)} are the amplitudes within the beam of order (m,n)(m,n), and zero elsewhere.

The field on the wafer under 4× demagnification (Lx(w)=Lx/4,Ly(w)=Ly/4L_x^{(w)} = L_x/4, L_y^{(w)} = L_y/4) is:

\begin{bmatrix} E_x^{(w)} \\ E_y^{(w)} \end{bmatrix} = \sum_{(m,n)\in D_N} \begin{bmatrix} B_{x;m,n}^{(w)} \\ B_{y;m,n}^{(w)} \end{bmatrix} \exp\left(-i\kappa_x^{(w)}mx-i\kappa_y^{(w)}ny-ik_{z;m,n}^{(w)}\zeta_w\right), \tag*{(32)}\end{bmatrix}

where kz;m,n(w)=[k02−(κx(w)m)2−(κy(w)n)2]1/2k_{z;m,n}^{(w)} = [k_0^2-(\kappa_x^{(w)}m)^2-(\kappa_y^{(w)}n)^2]^{1/2}, κx(w)=4κx\kappa_x^{(w)}=4\kappa_x, κy(w)=4κy\kappa_y^{(w)}=4\kappa_y, and (w) denotes the wafer-side field. Here ζw=z−Z1\zeta_w=z-Z_1, and the coefficients are referenced to the wafer plane. For the square lattice, harmonics propagating toward the wafer satisfy m2+n2≤[k0/(4κx)]2m^2+n^2 \le[k_0/(4\kappa_x)]^2. We select the smaller circular aperture m2+n2≤N2m^2+n^2 \le N^2, where N=⌊k0/(4κx)⌋=19N=\lfloor k_0/(4\kappa_x)\rfloor=19. This aperture is a design choice, not the exact propagation boundary. At the stated parameters, 24 additional propagating orders lie outside it. The selected domain is

DN={(m,n)∈Z2:m2+n2≤N2, (m,n)≠(0,0)}.(33)D_N=\{(m,n)\in\mathbb{Z}^2:m^2+n^2\le N^2,\ (m,n)\ne(0,0)\}. \tag*{(33)}

Each (m,n)(m,n)-th beam is characterized by polar angle θm,n\theta_{m,n} (from the zz-axis) and azimuthal angle φm,n\varphi_{m,n} (in the xyxy-plane):

sin⁡θm,n=[(κxm)2+(κyn)2]1/2k0,φm,n=Arg⁡(m+in).(34)\sin\theta_{m,n}=\frac{\left[(\kappa_xm)^2+(\kappa_yn)^2\right]^{1/2}}{k_0},\qquad \varphi_{m,n}=\operatorname{Arg}(m+in). \tag*{(34)}

On the wafer, the transverse wavevector, and hence sin⁡θ\sin\theta, is scaled by a factor of 4: sin⁡θm,n(w)=4sin⁡θm,n\sin\theta_{m,n}^{(w)}=4\sin\theta_{m,n}, with NAmax⁡=sin⁡θN,0(w)=0.993\mathrm{NA}_{\max}=\sin\theta_{N,0}^{(w)}=0.993.

The two-mirror system generalizes naturally to 3D. The first mirror is positioned in the plane z=Z1=1000z=Z_1=1000 mm, with the center of facet (m,n)(m,n) located at:

(x1(m,n),y1(m,n),Z1)=(Z1tan⁡θm,ncos⁡φm,n,Z1tan⁡θm,nsin⁡φm,n,Z1).(35)\left(x_1^{(m,n)},y_1^{(m,n)},Z_1\right)=\left(Z_1\tan\theta_{m,n}\cos\varphi_{m,n},Z_1\tan\theta_{m,n}\sin\varphi_{m,n},Z_1\right). \tag*{(35)}

The second mirror redirects the beam toward the wafer center (0,0,Z1)(0,0,Z_1) at angle θm,n(w)\theta_{m,n}^{(w)}. The facet center on the second mirror is positioned at:

(xf(m,n),yf(m,n),zf(m,n))=(um,ntan⁡θm,n(w)cos⁡φm,n,um,ntan⁡θm,n(w)sin⁡φm,n,Z1−um,n),(36)\left(x_f^{(m,n)},y_f^{(m,n)},z_f^{(m,n)}\right)=\left(u_{m,n}\tan\theta_{m,n}^{(w)}\cos\varphi_{m,n},u_{m,n}\tan\theta_{m,n}^{(w)}\sin\varphi_{m,n},Z_1-u_{m,n}\right), \tag*{(36)}

where um,n=Z1−zf(m,n)u_{m,n}=Z_1-z_f^{(m,n)} enforces equalized path lengths. The deflection angle βm,n\beta_{m,n} from the vertical satisfies:

tan⁡βm,n=[(xf−x1)2+(yf−y1)2]1/2um,n.(37)\tan\beta_{m,n}=\frac{\left[(x_f-x_1)^2+(y_f-y_1)^2\right]^{1/2}}{u_{m,n}}. \tag*{(37)}

The total optical path length is:

Pm,n=Z1cos⁡θm,n+[(xf−x1)2+(yf−y1)2+um,n2]1/2+um,ncos⁡θm,n(w).(38)P_{m,n}=\frac{Z_1}{\cos\theta_{m,n}}+\left[(x_f-x_1)^2+(y_f-y_1)^2+u_{m,n}^2\right]^{1/2}+\frac{u_{m,n}}{\cos\theta_{m,n}^{(w)}}. \tag*{(38)}

Parameter um,nu_{m,n} is determined by bisection to enforce Pm,n≈2998.1P_{m,n}\approx2998.1 mm. Incidence angles on the facets are i1=(θm,n+βm,n)/2i_1=(\theta_{m,n}+\beta_{m,n})/2, while i2i_2 is computed via scalar products.

For a square grating (Lx=LyL_x=L_y), all radial parameters (θ,θ(w),zf,ρf,β,i1,i2,P\theta,\theta^{(w)},z_f,\rho_f,\beta,i_1,i_2,P), where ρf=(xf2+yf2)1/2\rho_f=(x_f^2+y_f^2)^{1/2}, depend solely on the radius r=(m2+n2)1/2r=(m^2+n^2)^{1/2} and are independent of φ\varphi. The facet coordinates (x,y)(x,y) are obtained via rotation by φ\varphi:

x1=Z1tan⁡θcos⁡φ,y1=Z1tan⁡θsin⁡φ,xf=ρf(r)cos⁡φ,yf=ρf(r)sin⁡φ.(39)x_1=Z_1\tan\theta\cos\varphi,\qquad y_1=Z_1\tan\theta\sin\varphi,\qquad x_f=\rho_f(r)\cos\varphi,\qquad y_f=\rho_f(r)\sin\varphi. \tag*{(39)}

Thus, the radial design reduces to the 2D problem (Example 4) by substituting m→rm\to r. For non-integer rr (most orders with n≠0n\ne0), the radial solution is obtained numerically by bisection, and the reflectances R1,R2R_1,R_2 are interpolated linearly over rr from Table 6.

Design of the 3D Two-Mirror Projection System

Computational Parameters

Parameters match Example 4: λ=11.2\lambda=11.2 nm, L=10L=10 mm, Lx=Ly≈857L_x=L_y\approx857 nm, Z1=1000Z_1=1000 mm, 4× demagnification, M=76M=76, N=19N=19, NAmax⁡=0.993\mathrm{NA}_{\max}=0.993, and total optical path length 2998.1 mm. Harmonic (m,n)=(−8,0)(m,n)=(-8,0) (the back-reflection) does not contribute to the aerial image because it coincides with the illumination input aperture.

The number of orders in the selected wafer aperture is 1128 (all integer pairs (m,n)(m,n) within m2+n2≤192m^2+n^2 \le19^2, excluding the origin). Each order maps to one facet on mirror 1 and one on mirror 2, yielding 2256 facets total. The number of unique rr values is 132. Due to rotational symmetry, each rr corresponds to 4 to 24 individual orders that share identical Bragg multilayer parameters.

Results for Facet Layouts

Table 7 lists facet centers, incidence angles, and interpolated reflectances for 21 representative orders (m,n)(m,n). The total optical path is Pm,n=2998.1P_{m,n}=2998.1 mm across all orders. Facet tilt α1\alpha_1 depends only on rr and ranges from 0.75∘0.75^\circ (r=1r=1) to 33.4∘33.4^\circ (r=19r=19). For the second mirror, α2\alpha_2 ranges from 2.6∘2.6^\circ to 82.2∘82.2^\circ.

mmnnrrφ\varphi (°)x1x_1 (mm)y1y_1 (mm)xfx_f (mm)yfy_f (mm)zfz_f (mm)i1i_1 (°)i2i_2 (°)∣R1R2∣\lvert R_1R_2\rvert
101.0000.013.10.052.20.02.01.500.370.560
111.41445.013.113.152.252.23.22.120.530.561
212.23626.626.113.1104.652.36.53.360.830.561
222.82845.026.126.1104.6104.69.94.261.050.561
303.0000.039.20.0157.00.011.04.521.110.561
323.60633.739.226.2157.2104.815.65.451.330.561
435.00036.952.439.3210.2157.729.47.621.830.562
505.0000.065.50.0262.80.029.47.621.830.562
707.0000.091.80.0370.20.058.310.862.500.564
557.07145.065.665.6264.5264.559.510.982.520.565
859.43432.0105.365.8427.7267.3110.215.113.210.568
779.89945.092.292.2375.1375.1122.615.973.320.569
10010.0000.0131.80.0536.20.0125.416.163.350.569
10712.20735.0132.492.6543.5380.5199.120.633.780.573
13013.0000.0172.40.0710.70.0232.522.413.880.575
101014.14245.0133.0133.0551.6551.6289.125.223.930.578
16016.0000.0213.80.0897.50.0411.430.673.730.584
15817.00028.1201.0107.2850.1453.4502.434.383.380.588
18118.0283.2242.013.41031.957.3632.539.352.670.591
131318.38545.0175.0175.0748.4748.4694.941.642.270.593
19019.0000.0256.30.01102.10.0869.147.791.010.596

Table 7. Facet configurations of the first and second mirrors for the 3D case (λ=11.2\lambda=11.2 nm, Lx=Ly=857L_x=L_y=857 nm, N=19N=19, NA=0.993\mathrm{NA}=0.993, P=2998.1P=2998.1 mm). r=(m2+n2)1/2r=(m^2+n^2)^{1/2}, φ\varphi is azimuthal angle, (x1,y1)(x_1,y_1) is facet center on mirror 1 at z=Z1=1000z=Z_1=1000 mm, (xf,yf,zf)(x_f,y_f,z_f) is facet center on mirror 2, i1i_1 and i2i_2 are incidence angles, ∣R1R2∣\lvert R_1R_2\rvert is product of reflectances (interpolated from Table 6).

The centers of the 1128 facets of the first mirror lie in a common plane at z=Z1=1000z=Z_1=1000 mm within a circle of radius 256.3 mm. The 1128 facets of the second mirror lie on a surface of revolution spanning zf(r)=2.0z_f(r)=2.0 mm (r=1r=1) to 869.1 mm (r=19r=19). Figure 15 shows a 3D rendering of the projection architecture. For clarity, the silver-colored backing of the second mirror is shown in the figure only to highlight the facets against its background. In the actual system, no such underlying surface exists.

3D CAD visualization of the all-reflective two-mirror system

Figure 15. 3D CAD visualization of the all-reflective two-mirror system (viewed at 60∘60^\circ to the zz-axis). Gray planar elements: 1128 facets of mirror 1 (in the plane z=Z1=1000z = Z_1 = 1000 mm, disk radius 256.3 mm) and 1128 facets of mirror 2 (on the surface of revolution from zf=2z_f = 2 mm to 869 mm). Red 1×11 \times1 cm squares denote the mask (z=0z = 0) and wafer (z=Z1z = Z_1).

Angles of incidence on mirror 1 range from i1=1.50∘i_1=1.50^\circ (r=1r=1) to 47.79∘47.79^\circ (r=19r=19), while on mirror 2 they remain between i2=0.37∘i_2=0.37^\circ and 3.93∘3.93^\circ. The two-reflector product ∣R1R2∣\lvert R_1R_2\rvert varies narrowly between 0.560 and 0.596 (mean 0.576), retaining ~58% of the power leaving the mask in each selected order after the two reflections (4–20 times more than in 6–10-mirror schemes).

Practical limitations of the 3D case. The total of 2256 facets (1128 per mirror) makes direct fabrication and alignment technically challenging. However, due to rotational symmetry, 132 radial coating pairs are required, one per radial value rr, corresponding to up to 264 distinct multilayer structures for the two mirrors rather than one design per facet. Replacing the second-mirror coatings by a common near-normal-incidence coating is a separate approximation.

3D Mask Synthesis

Formulation of the 3D Inverse Problem

In 3D, all 1128 diffraction orders have equal optical path lengths: Pm,n≈2998.1P_{m,n} \approx2998.1 mm. The diffracted field contains two independent orthogonal transverse components, ExE_x and EyE_y, corresponding to scattering amplitudes Ax;m,n(r)A_{x;m,n}^{(r)} and Ay;m,n(r)A_{y;m,n}^{(r)} computed via the 3D waveguide method.

The harmonic amplitudes on the wafer are:

\begin{bmatrix} B_{x;-m,-n}^{(w)} \\ B_{y;-m,-n}^{(w)} \end{bmatrix} = \hat{r}_{2,m,n}\hat{r}_{1,m,n} \begin{bmatrix} A_{x;m,n}^{(r)} \\ A_{y;m,n}^{(r)} \end{bmatrix}, \qquad(m,n) \in D_N, \tag*{(40)}\end{bmatrix}

where r^1,m,n\hat{r}_{1,m,n} and r^2,m,n\hat{r}_{2,m,n} denote the complex matrix field reflection coefficients of the corresponding facets of the first and second mirrors for order (m,n)(m,n), DN={(m,n)∈Z2:m2+n2≤N2, (m,n)≠(0,0)}D_N = \{(m,n) \in\mathbb{Z}^2 : m^2 + n^2 \leq N^2,\ (m,n) \ne(0,0)\}.

For the general vector transfer, each matrix acts on global transverse electric-field components and includes the local polarization bases,

r^j=Qjoutdiag⁡(rs,j,rp,j)(Qjin)−1.\hat{r}_j = Q_j^{\mathrm{out}} \operatorname{diag}(r_{s,j},r_{p,j}) \left(Q_j^{\mathrm{in}}\right)^{-1}.

The columns of QjinQ_j^{\mathrm{in}} and QjoutQ_j^{\mathrm{out}} are the xyxy projections of the incoming and outgoing local TE- and TM-polarization unit vectors, with the reflection coefficients defined in these same bases. The first reflection therefore acts before the second. The numerical demonstration retains the scalar-channel approximation with equal TE and TM coefficients and evaluates only transverse field components, as a proof-of-concept simplification.

The field components above the wafer are:

\begin{bmatrix} E_x^{(w)}(x,y) \\ E_y^{(w)}(x,y) \end{bmatrix} = \sum_{(m,n)\in D_N} \begin{bmatrix} B_{x;m,n}^{(w)} \\ B_{y;m,n}^{(w)} \end{bmatrix} \exp\left(-i\kappa_x^{(w)}mx-i\kappa_y^{(w)}ny\right). \tag*{(41)}\end{bmatrix}

and the total aerial intensity is:

Iw(x,y)=∣Ex(w)(x,y)∣2+∣Ey(w)(x,y)∣2.(42)I_{\mathrm{w}}(x,y) = \left|E_x^{(\mathrm{w})}(x,y)\right|^2 + \left|E_y^{(\mathrm{w})}(x,y)\right|^2. \tag*{(42)}

Although the incident wave is purely TE-polarized (E∥y\mathbf{E} \parallel\mathbf{y}), the 3D mask (with a two-dimensionally periodic structure ε(x,y)\varepsilon(x,y)) generates cross-polarized scattering, so that the scattering amplitude Ax;m,n(i)A_{x;m,n}^{(i)} is non-zero (the incident field contains no cross-polarized component, Ax;m,n(i)=0A_{x;m,n}^{(i)} = 0).

The 3D latent function is:

flatent(x,y)=Re⁡(∑m=−MM∑n=−MMFm,ne−iκxmx−iκyny),(43)f_{\mathrm{latent}}(x,y) = \operatorname{Re}\left(\sum_{m=-M}^{M}\sum_{n=-M}^{M} F_{m,n} \mathrm{e}^{-\mathrm{i}\kappa_x m x-\mathrm{i}\kappa_y n y}\right), \tag*{(43)}

where Fm,nF_{m,n} are the complex Fourier coefficients to be optimized (the two-dimensional generalization of FmF_m in (25)). The physical absorber density is obtained via the sigmoid mapping ρ(x,y)=σ(βflatent)∈(0,1)\rho(x,y) = \sigma(\beta f_{\mathrm{latent}}) \in(0,1), where β\beta is a steepness parameter increased during optimization to enforce binarization, and σ\sigma is the sigmoid function defined in the 2D formulation. The permittivity is computed as ε(x,y)=εvac+ρ(x,y)(εabs−εvac)\varepsilon(x,y) = \varepsilon_{\mathrm{vac}} + \rho(x,y)(\varepsilon_{\mathrm{abs}} - \varepsilon_{\mathrm{vac}}).

Parameters and Results of Optimization

Parameters: λ=11.2\lambda= 11.2 nm, incidence angle 6∘6^\circ (TE), Lx=Ly≈857L_x = L_y \approx857 nm, Lx(w)=Ly(w)≈214L_x^{(\mathrm{w})} = L_y^{(\mathrm{w})} \approx214 nm, La absorber (dabs=60d_{\mathrm{abs}} = 60 nm) on Ru/Be/Sr Bragg mirror, Fourier expansion parameter Mopt=M=16M_{\mathrm{opt}} = M = 16 (instead of the maximum propagating order N=19N = 19), 500 epochs.

Figure 16 presents the optimization results: target emblem (University of Nizhny Novgorod logo), bandlimited target projection, continuous optimized absorber density ρ(x,y)\rho(x,y), synthesized aerial intensity, binarized mask, and aerial image from the binarized mask.

Optimization results for a 3D mask in the projection system

Figure 16. Optimization results for a 3D mask in the projection system (λ=11.2\lambda= 11.2 nm, La absorber, Mopt=16M_{\mathrm{opt}} = 16, NA ≈0.84\approx0.84, 500 epochs). Top row: target image (UNN logo), target projection onto propagating harmonics, optimized absorber density ρ(x,y)\rho(x,y). Bottom row: synthesized aerial intensity on the wafer, binarized mask (ρ>⟨ρ⟩\rho> \langle\rho\rangle), aerial intensity from the binarized mask. The mask period is Lx=Ly≈857L_x = L_y \approx857 nm. The wafer period is Lx(w)≈214L_x^{(\mathrm{w})} \approx214 nm.

The binarization penalty Lbin\mathcal{L}_{\mathrm{bin}} drops from 0.25 to 0.002, and Lshape\mathcal{L}_{\mathrm{shape}} decreases from 6.5×10−86.5 \times10^{-8} to 2.3×10−82.3 \times10^{-8}. Truncating to M=16M = 16 corresponds to NA =0.84= 0.84.

Discussion: Advantages and Disadvantages of the Proposed Projection System

Advantages

  1. The array of 2N=382N = 38 planar facets independently maps each mask diffraction harmonic to the corresponding wafer harmonic up to NAmax⁡=0.993\mathrm{NA}_{\max} = 0.993, a value unattainable in conventional multi-mirror EUV projection objectives with a limited number of reflections.

  1. For the periodic structures considered here, the common illuminated wafer region can have an area comparable to the illuminated mask area, apart from edge effects. The unit-cell period is reduced by four along each transverse direction, so more repeated cells fit within the illuminated area. This does not imply demagnification of an arbitrary finite mask pattern while preserving its image area.

  1. The finite facet apertures and angular selectivity of the Bragg coatings can act as spatial-frequency filters for individual channels. Suppression of unwanted angular components can reduce flare, but its effect on useful signal and aerial-image contrast requires a quantitative calculation with finite source bandwidth and angular extent, mask scattering, and facet acceptance.

  1. The two reflections retain approximately 50–60% of the power leaving the mask in an accepted diffraction order. This channel transmission is not the fraction of the total mask-reflected power or source power delivered to a useful image region. Comparison values of about 12% for six reflections and 2.8% for ten reflections refer only to products of mirror reflectances.

  1. Each accepted harmonic undergoes the spatial-frequency transformation mκx→−4mκxm\kappa_x \to-4m\kappa_x, corresponding to fourfold reduction and inversion of the periodic coordinate. Order-dependent complex reflection coefficients and aperture truncation modify the pattern, so exact reproduction of its shape is not implied by the linear frequency map alone.

  1. In Examples 1 and 2, all NN beams propagate strictly parallel to the zz-axis after the first reflection, simplifying mutual alignment. In Example 3, beams propagate at deflection angles (θm(w)−θm)(\theta_m^{(\mathrm{w})} - \theta_m) ranging from ∼2∘\sim2^\circ to ∼69∘\sim69^\circ, and in Example 4 at tailored angles βm\beta_m.

  1. In Examples 1 and 2, tilt angles of the first mirror do not exceed 7.2∘7.2^\circ, rendering it nearly planar and amenable to precision machining with sub-micrometer accuracy. In Example 3, tilt angles reach 27.2∘27.2^\circ, and in Example 4, 33.4∘33.4^\circ.

  1. All facets are planar, so Bragg coatings can be laterally homogeneous across each facet, eliminating the complex graded-layer deposition required on aspheric surfaces in conventional systems [7].

  1. Separate facet pairs permit channel-specific coating optimization and can allow modular changes to the accepted numerical aperture and local replacement of degraded elements. Their practical benefit depends on alignment, phase control, and thermal and mechanical stability. Replacement without optical readjustment has not been demonstrated.

Disadvantages

  1. Changing LxL_x or LyL_y changes the diffraction angles and can require new facet positions, tilts, coatings, and channel counts. Feasibility depends on the available mechanical clearance, optical acceptance, phase tolerances, and source bandwidth. It is not established for arbitrary periods.

  1. Projections of the beams on the wafer grow as Lm(w)=am/cos⁡θm(w)L_m^{(\mathrm{w})} = a_m / \cos\theta_m^{(\mathrm{w})}: values remain below 18 mm for m≤16m \le16, reach 28.7 mm for m=18m = 18, and 82.1 mm for m=19m = 19 (at θ19(w)=83.2∘\theta_{19}^{(\mathrm{w})} = 83.2^\circ, the projection is stretched by a factor of 8.5). This is not related to diffractive beam spreading but is a geometric consequence of the grazing projection of a finite-width beam (am≈10a_m \approx10 mm) onto the wafer plane. The effect can be mitigated by apodization or by sizing the apertures of the high-order facets.

  1. In Example 2, the second-mirror facets operate at grazing angles from 1.5∘1.5^{\circ} to 41.6∘41.6^{\circ}. The facet length required for full beam interception reaches 382.5 mm for m=1m = 1, whereas the corresponding lengths in Example 4 are approximately 10 mm.

  1. In Example 4, the first-mirror facet centers span ±256.3\pm256.3 mm, while the second-mirror centers reach ±1102.1\pm1102.1 mm. The complete transverse span is therefore approximately 2204.2 mm using the rounded tabulated coordinates, before mechanical supports and facet extents are included. The total optical path length is approximately 3 m.

  1. Preserving the energy balance across diffraction harmonics requires individual optimization of Bragg bilayer thicknesses for each first-mirror facet, covering incidence angles from 1.5∘1.5^{\circ} to 47.8∘47.8^{\circ}.

  1. The optical path differences in Examples 1–3 introduce deterministic relative propagation phases, rather than destroying coherence under strictly monochromatic illumination. Example 4 removes these path-dependent phase differences. Position tolerances must be set by the allowable wavefront error, while a finite source bandwidth additionally requires temporal-coherence analysis.

  1. A normal surface displacement δh\delta h produces an optical-path error of approximately 2δhcos⁡i2\delta h \cos i. For independent errors on two nearly normally reflecting facets, σP≈22σh\sigma_P \approx2\sqrt{2}\sigma_h. As an illustrative wavefront criterion, S≈exp⁡[−(2πσP/λ)2]≥0.8S \approx\exp[-(2\pi\sigma_P/\lambda)^2] \ge0.8 at λ=11.2\lambda= 11.2 nm gives σh≲0.30\sigma_h \lesssim0.30 nm per facet. The actual surface and alignment tolerances must be derived from the aerial-image requirements and error correlations, especially across 2256 facets.

  1. The validity of treating diffraction orders as bounded collimated beams is analyzed in Appendix A, confirming high precision for centimeter-scale EUV beams.

Conclusion

In this work, an all-reflective EUV projection concept has been proposed and investigated that uses two reflections per accepted channel between the mask and wafer, with each mirror composed of individual planar facets corresponding to each spatial diffraction order.

The main conclusions are:

  1. A two-mirror projection concept providing 4×4\times reduction of the periodic unit cell at NAmax⁡≈0.993\mathrm{NA}_{\max} \approx0.993 is analyzed at 13.5 nm and 11.2 nm, with Lx/λL_x/\lambda kept fixed.

  1. A spatial geometry providing rigorous equalization of optical path lengths across all diffraction harmonics is established, removing order-dependent propagation phase shifts.

  1. Multi-parameter optimization of 30-bilayer Bragg mirrors (Mo/Si at 13.5 nm and Ru/Be at 11.2 nm) demonstrates channel throughput of ∼51\sim51–60%60\%. Facets of the second mirror operate at quasi-normal incidence (≤3.93∘\le3.93^{\circ}) suggesting the possibility of a common coating as a separate approximation.

  1. The concept is generalized to 3D periodic masks. It has been shown that for the considered mask, the 2256-facet system with 1128 selected orders requires optimizing 132 radial coating pairs, or up to 264 distinct multilayer structures, due to rotational symmetry.

  1. Inverse lithography produces simulated aerial images with a central peak of approximately 5.4 nm FWHM from a 4.6-nm-wide target and with two 6-nm-wide target lines separated by 6 nm. The unused back-reflection order is suppressed in the 2D optimization, and the two peaks remain resolved for the tested defocus values up to 5 nm. Resist printing and a full process window have not been evaluated.

  1. The validity of the geometrical-optics beam approximation for centimeter-scale EUV beams over ∼1\sim1 m paths is proven (collimation length exceeds hundreds of meters, and edge blur is under a millimeter).

The modular faceted approach offers a route to reducing the number of reflections in periodic-pattern projection while using planar reflecting surfaces. Its applicability to high-resolution lithography requires further assessment of source coherence and bandwidth, polarization, fabrication and alignment tolerances, and resist response. This faceted approach can removes the fundamental technological constraints of multi-mirror EUV optics associated with the number of reflections and complex aspherization, opening a promising route toward projection lithography systems of ultra-high resolution (NA→1\mathrm{NA} \to1).

Diffractive Propagation of a Hard-Apertured Beam

Problem Formulation

As in the main text, the time dependence exp⁡(iωt)\exp(i\omega t) is assumed and omitted. Let the initial field at z=0z=0 be non-zero only over an aperture of width WW along xx:

E(x,0)=E0Π(xW)e−ikxx,Π(u)={1,∣u∣≤1/2,0,∣u∣>1/2,(44)E(x,0)=E_0\Pi\left(\frac{x}{W}\right)e^{-ik_xx},\qquad\Pi(u)=\begin{cases}1,&\lvert u\rvert\le1/2,\\0,&\lvert u\rvert>1/2,\end{cases} \tag*{(44)}

where kxk_x is the transverse wavevector component:

k=(kx,0,kz),kx2+kz2=k02,k0=2πλ.(45)\mathbf{k}=(k_x,0,k_z),\qquad k_x^2+k_z^2=k_0^2,\qquad k_0=\frac{2\pi}{\lambda}. \tag*{(45)}

The propagation angle relative to the zz-axis is α=arcsin⁡(kx/k0)\alpha=\arcsin(k_x/k_0) (cos⁡α=kz/k0\cos\alpha=k_z/k_0).

Angular Spectrum

Expanding (44) into plane waves:

E(x,z)=∫−∞∞A(q)e−iqx−iγ(q)zdq2π,γ(q)=(k02−q2)1/2,Im⁡γ(q)≤0,(46)E(x,z)=\int_{-\infty}^{\infty}A(q)e^{-iqx-i\gamma(q)z}\frac{\mathrm{d}q}{2\pi},\qquad\gamma(q)=\left(k_0^2-q^2\right)^{1/2},\qquad\operatorname{Im}\gamma(q)\le0, \tag*{(46)}

with spectral amplitude:

A(q)=E0Wsinc⁡[(q−kx)W2],sinc⁡u≡sin⁡uu.(47)A(q)=E_0W\operatorname{sinc}\left[\frac{(q-k_x)W}{2}\right],\qquad\operatorname{sinc}u\equiv\frac{\sin u}{u}. \tag*{(47)}

The spectral half-width to first zeros is Δq=2π/W\Delta q=2\pi/W. The divergence half-angle is:

θ0=Δqk0cos⁡α=λWcos⁡α=λk0Wkz=λW⊥,(48)\theta_0=\frac{\Delta q}{k_0\cos\alpha}=\frac{\lambda}{W\cos\alpha}=\frac{\lambda k_0}{Wk_z}=\frac{\lambda}{W_\perp}, \tag*{(48)}

governed by the perpendicular beam width:

W⊥=Wcos⁡α=Wkzk0.(49)W_\perp=W\cos\alpha=\frac{Wk_z}{k_0}. \tag*{(49)}

Paraxial Approximation and Beam Coordinate System

Using the beam coordinates (ξ,ζ)(\xi,\zeta) defined below, write E(x,z)=exp⁡(−ik0ζ)U(ξ,ζ)E(x,z)=\exp(-ik_0\zeta)U(\xi,\zeta). Neglecting ∂2U/∂ζ2\partial^2U/\partial\zeta^2 relative to 2ik0∂U/∂ζ2ik_0\partial U/\partial\zeta gives the paraxial envelope equation

2ik0∂U∂ζ=∂2U∂ξ2.(50)2ik_0\frac{\partial U}{\partial\zeta}=\frac{\partial^2U}{\partial\xi^2}. \tag*{(50)}

Equation (50) describes diffraction along the beam axis. In laboratory coordinates, an envelope extracted using exp⁡[−i(kxx+kzz)]\exp[-i(k_xx+k_zz)] also has a transverse transport term. The spectral curvature is γ′′(kx)=−k02/kz3\gamma^{\prime\prime}(k_x)=-k_0^2/k_z^3.

In the beam coordinate frame (ξ,ζ)(\xi,\zeta):

ζ=xsin⁡α+zcos⁡α(along the beam axis),ξ=xcos⁡α−zsin⁡α(transverse to the beam).(51)\zeta=x\sin\alpha+z\cos\alpha\quad\text{(along the beam axis)},\qquad\xi=x\cos\alpha-z\sin\alpha\quad\text{(transverse to the beam)}. \tag*{(51)}

Along the central ray, a propagation distance zz in the laboratory frame corresponds to ζ=z/cos⁡α=zk0/kz\zeta=z/\cos\alpha=zk_0/k_z along the beam axis.

To leading paraxial order, the oblique aperture is represented on the transverse beam plane by a top-hat envelope of width W⊥W_\perp. Its propagation is described by the Fresnel integral

U(ξ,ζ)=k0−2πiζ∫−W⊥/2W⊥/2U(ξ′,0)exp⁡[−ik0(ξ−ξ′)22ζ]dξ′.(52)U(\xi,\zeta)=\sqrt{\frac{k_0}{-2\pi i\zeta}}\int_{-W_\perp/2}^{W_\perp/2}U(\xi',0)\exp\left[-\frac{ik_0(\xi-\xi')^2}{2\zeta}\right]\mathrm{d}\xi'. \tag*{(52)}

Key Diffraction Scales

From (52), three characteristic scales emerge (where L≡ζL \equiv\zeta is propagation distance):

  1. Fresnel number (NFN_{\mathrm{F}}) and collimation length (LFL_{\mathrm{F}}):

NF(L)=(W⊥/2)2λL=LFL,LF=W⊥24λ=W2kz24k02λ.(53)N_{\mathrm{F}}(L) = \frac{(W_{\perp}/2)^2}{\lambda L} = \frac{L_{\mathrm{F}}}{L}, \qquad L_{\mathrm{F}} = \frac{W_{\perp}^2}{4\lambda} = \frac{W^2 k_z^2}{4 k_0^2 \lambda}. \tag*{(53)}

Near-field geometrical propagation holds for NF≫1N_{\mathrm{F}} \gg1, while Fraunhofer diffraction occurs for NF≪1N_{\mathrm{F}} \ll1.

  1. Near-Field Boundary Blur: The edge transition scale is:

Δ∼λL=2πLk0.(54)\Delta\sim\sqrt{\lambda L} = \sqrt{\frac{2\pi L}{k_0}}. \tag*{(54)}

with the 10%–90% intensity transition width given by:

Δ10–90%≈0.829λL.(55)\Delta_{10\text{--}90\%} \approx0.829\sqrt{\lambda L}. \tag*{(55)}
  1. Far-Field Divergence: The angular intensity profile is:

I(θ)=I0sinc⁡2(πW⊥θλ),(56)I(\theta) = I_0 \operatorname{sinc}^2\left(\frac{\pi W_{\perp}\theta}{\lambda}\right), \tag*{(56)}

with a main-lobe angular width between the first zeros of 2θ0=2λ/W⊥2\theta_0 = 2\lambda/W_{\perp}, a full width at half maximum of FWHM⁡≈0.886λ/W⊥\operatorname{FWHM} \approx0.886\lambda/W_{\perp}, and a first sidelobe of ≈4.7%\approx4.7\% of the maximum. The beam size expands as D(L)≈2θ0LD(L) \approx2\theta_0 L.

Near-Field Boundary Structure

When λL≪W⊥\sqrt{\lambda L} \ll W_{\perp} (NF≫1N_{\mathrm{F}} \gg1), opposite aperture edges diffract independently, matching semi-infinite knife-edge diffraction:

I(v)=I02{[C(v)+12]2+[S(v)+12]2},v=(ξedge−ξ)2λL.(57)I(v) = \frac{I_0}{2}\left\{\left[C(v) + \frac{1}{2}\right]^2 + \left[S(v) + \frac{1}{2}\right]^2\right\}, \qquad v = (\xi_{\mathrm{edge}} - \xi)\sqrt{\frac{2}{\lambda L}}. \tag*{(57)}

where C(v)C(v) and S(v)S(v) are Fresnel integrals and ξedge\xi_{\mathrm{edge}} is the right aperture edge, so v>0v > 0 points into the illuminated region. Characteristics include:

  • At the shadow boundary, I(0)=I0/4I(0) = I_0/4.

  • The first diffraction peak reaches I≈1.37I0I \approx1.37 I_0 at v≈1.22v \approx1.22.

  • In the illuminated region, decaying Fresnel fringes occur with period ∼λL/∣ξ−ξedge∣\sim\lambda L/\lvert\xi- \xi_{\mathrm{edge}}\rvert.

  • Field decays monotonically into the shadow region.

The number of Fresnel oscillations across the beam width is of the order of NFN_{\mathrm{F}}. As NFN_{\mathrm{F}} decreases to ∼10\sim10, the oscillations from the opposite edges begin to interfere with each other, distorting the central part of the plateau. At NF∼1N_{\mathrm{F}} \sim1 the profile smooths out and transitions into the far-field distribution (56).

Numerical Example: λ=11.2\lambda= 11.2 nm, W=1W = 1 cm, kx=kzk_x = k_z

For kx=kz=k0/2k_x = k_z = k_0/\sqrt{2} (α=45∘\alpha= 45^\circ):

ParameterValue
k0=2π/λk_0 = 2\pi/\lambda5.61×108 m−15.61 \times10^8\ \mathrm{m}^{-1}
kx=kzk_x = k_z3.97×108 m−13.97 \times10^8\ \mathrm{m}^{-1}
W⊥=W/2W_{\perp} = W/\sqrt{2}7.07 mm7.07\ \mathrm{mm}
θ0=λ/W⊥\theta_0 = \lambda/W_{\perp}1.58×10−6 rad≈0.33′′1.58 \times10^{-6}\ \mathrm{rad} \approx0.33^{\prime\prime}
LF=W⊥2/(4λ)L_{\mathrm{F}} = W_{\perp}^2/(4\lambda)1.1×103 m1.1 \times10^3\ \mathrm{m}

Table 8.

Table 8 and Figs. 17–19 illustrate beam evolution across distances.

LL, mNFN_FλL\sqrt{\lambda L} (mm)Δ10−90%\Delta_{10-90\%} (mm)Propagation Regime
111160.110.09Geometrical shadow
101120.330.28Near field
100111.060.88Edge-ripple overlap
10001.13.352.30Intermediate zone
100000.1110.6—Fraunhofer far field (FWHM = 14 mm)

Table 8. Diffractive beam evolution with propagation distance (λ\lambda = 11.2 nm, W⊥W_\perp = 7.07 mm). The 10%–90% widths at L≤100L \le100 m use the independent-edge approximation. The LL = 1000 m value is a numerical finite-slit estimate, where the two edge fields overlap and Eq. (55) is not applicable.

Transverse intensity profile evolution versus distance

Figure 17. Transverse intensity profile evolution versus distance (angular spectrum calculation). Dashed lines: geometrical shadow boundaries.

Conclusions of the Diffraction Analysis

  1. A centimeter-aperture EUV beam (W⊥≈7W_\perp\approx7 mm) at λ\lambda = 11.2 nm exhibits high collimation stability: LF≈1.1L_F \approx1.1 km and divergence θ0≈1.6\theta_0 \approx1.6 μ\mum. Over typical optical paths (L∼1L \sim1–3 m), the beam profile is practically indistinguishable from the geometrical projection of the mask.

  1. Edge blur obeys Δ∼λL\Delta\sim\sqrt{\lambda L}, amounting to only 0.09 mm at 1 m and 0.28 mm at 10 m. Near the boundaries, Fresnel fringes form with a local peak reaching 1.37 I0I_0, decaying monotonically into the shadow.

  1. At distances L≳LFL \gtrsim L_F, sharp edges blur and the beam transitions into a sinc2^2 distribution with sidelobe levels of ≈4.7%\approx4.7\%. The beam diameter grows linearly (D≈2θ0LD \approx2\theta_0L).

  1. Hard aperture truncation generates edge ripples. If suppression is required, apodizing elements with smoothed edge transmission/reflection can be incorporated.

Right-edge boundary diffraction profile

Figure 18. Right-edge boundary diffraction profile. The solid line shows the slit calculation. The dashed line shows the semi-infinite knife-edge analytical model, (57). Small oscillations at L=100L = 100 m arise from interference with the wave diffracted by the opposite edge.

Far-field intensity distribution

Figure 19. Far-field intensity distribution (L=10L = 10 km, NF=0.11N_{\mathrm{F}} = 0.11): the solid line shows the numerical result in normalized angular coordinate θ/θ0\theta/\theta_0. The dashed line shows the theoretical sinc⁡2(πW⊥θ/λ)\operatorname{sinc}^{2}(\pi W_{\perp}\theta/\lambda).

Nomenclature

This table summarizes the main abbreviations used in the paper.

NotationDescription
CADComputer-aided design
CDCritical dimension
DoFDepth of focus
EUVExtreme ultraviolet
FWHMFull width at half maximum
GOGeometrical optics
ILTInverse lithography technology
L-BFGS-BBound-constrained limited-memory BFGS method
Mo/SiMolybdenum/silicon multilayer Bragg coating
MSEMean squared error
NANumerical aperture
OAIOff-axis illumination
RCWARigorous coupled-wave analysis
RMSRoot mean square
Ru/BeRuthenium/beryllium multilayer Bragg coating
Ru/Be/SrRuthenium/beryllium/strontium multilayer Bragg coating
TETransverse electric polarization
TMTransverse magnetic polarization
TMMTransfer-matrix method
HFHigh-frequency penalty
UNNUniversity of Nizhny Novgorod

Table 9. Nomenclature

References

  1. [1]J. van Schoot, R. van Ballegooij, H. Butler, E. van Setten, G. Schiffelers, W. Bouman, S. van Gorp, K. Umstadter, J. Zimmermann, D. Golde, J. T. Neumann, and P. Graeupner, “Next step in Moore’s law: high NA EUV system overview and first imaging and overlay performance,” J. Micro/Nanopattern. Mater. Metrol., vol. 24, no. 1, p. 011009, 2025, doi: 10.1117/1.JMM.24.1.011009.DOI
  2. [2]N. I. Chkhalo, “New Concept for the Development of High-Performance X-ray Lithography,” Russian Microelectronics, vol. 53, no. 5, pp. 397–407, oct 2024. [Online]. Available: https://link.springer.com/10.1134/S1063739724600511
  3. [3]IEEE IRDS, “International roadmap for devices and systems (IRDS): 2024 update, More Moore,” IEEE, Tech. Rep., 2024. [Online]. Available: https://irds.ieee.org/
  4. [4]N. I. Chkhalo, S. A. Gusev, A. N. Nechay, D. E. Pariev, V. N. Polkovnikov, N. N. Salashchenko, F. Schäfers, M. G. Sertsov, A. Sokolov, M. V. Svechnikov, and D. A. Tatarsky, “High-reflection mo/be/si multilayers for euv lithography,” Opt. Lett., vol. 42, no. 24, pp. 5070–5073, Dec 2017. [Online]. Available: https://opg.optica.org/ol/abstract.cfm?URI=ol-42-24-5070
  5. [5]M. van der Kerkhof, A. Klein, P. Vermeulen, T. van der Woord, I. Donmez, G. Salmaso, and R. Maas, “High-transmission EUV pellicles supporting >400W source power,” in Proc. SPIE, vol. 12051, 2022, p. 120510B, optical and EUV Nanolithography XXXV; doi: 10.1117/12.2614262.DOI
  6. [6]J. van Schoot, E. van Setten, G. Rispens, K. Z. Troost, B. Kneer, S. Migura, J. T. Neumann, and W. Kaiser, “High-numerical aperture extreme ultraviolet scanner for 8-nm lithography and beyond,” J. Micro/Nanolith. MEMS MOEMS, vol. 16, no. 4, p. 041010, 2017, doi: 10.1117/1.JMM.16.4.041010.DOI
  7. [7]J. Kalden, J. T. Neumann, D. Jürgens, W. Graupner, W. Seitz, and P. Kürz, “EUV optics at ZEISS: status, outlook, and future,” in Proc. SPIE, vol. 12953, 2024, p. 129530Q, optical and EUV Nanolithography XXXVII.DOI
  8. [8]E. van Setten, G. Bottiglieri, L. de Winter, J. McNamara, P. Rusu, J. Lubkoll, G. Rispens, and J. van Schoot, “Edge placement error control and Mask3D effects in High-NA anarmophic EUV lithography,” in Proc. SPIE, vol. 10450, 2017, p. 104500W, doi: 10.1117/12.2280624.DOI
  9. [9]T. Shintake, “Can we improve the energy efficiency of EUV lithography?” arXiv preprint arXiv:2405.11717, 2024, presented at the Photomask Japan Symposium, Yokohama, Japan.arxiv.org/abs/2405.11717
  10. [10]I. Abramov, S. Golubev, E. Gospodchikov, A. Shalashov, A. Perekalov, A. Nechay, and N. Chkhalo, “Laser discharge in a high-pressure jet of heavy noble gas: Expansion of emitting volume promises an efficient source of euv light for lithography,” Phys. Rev. Appl., vol. 23, p. 024004, Feb 2025. [Online]. Available: https://link.aps.org/doi/10.1103/PhysRevApplied.23.024004DOI
  11. [11]R. A. Shaposhnikov, V. N. Polkovnikov, N. N. Salashchenko, N. I. Chkhalo, and S. Y. Zuev, “Highly reflective ru/sr multilayer mirrors for wavelengths 9–12 nm,” Opt. Lett., vol. 47, no. 17, pp. 4351–4354, Sep 2022. [Online]. Available: https://opg.optica.org/ol/abstract.cfm?URI=ol-47-17-4351DOI
  12. [12]V. N. Polkovnikov, N. N. Salashchenko, M. V. Svechnikov, and N. I. Chkhalo, “Beryllium-based multilayer x-ray optics,” Physics-Uspekhi, vol. 63, no. 1, p. 83, jan 2020. [Online]. Available: https://doi.org/10.3367/UFNe.2019.05.038623
  13. [13]T. Shintake, “High-NA in-line projector for EUV lithography,” arXiv preprint arXiv:2508.00433, 2025, oIST; JPN Patent Application 2025-119190.arxiv.org/abs/2508.00433
  14. [14]H. Fukuda and T. Terasawa, “Design and analysis of diffraction mirror optics for EUV projection lithography,” Microelectronic Engineering, vol. 27, pp. 239–242, 1995.DOI
  15. [15]Z. Zheng, X. Sun, P. Gu, and X. Liu, “Design of objective lens with reflective spherical Fresnel zone plate,” Frontiers of Optoelectronics in China, vol. 1, no. 1–2, pp. 178–182, 2008.DOI
  16. [16]Z. Zheng, “Design of off-axis reflective projection lens using spherical Fresnel surface,” Optik, vol. 122, pp. 145–149, 2011.DOI
  17. [17]N. Choksi, D. S. Pickard, M. McCord, R. F. W. Pease, Y. Shroff, Y. Chen, W. Oldham, and D. Markle, “Maskless extreme ultraviolet lithography,” Journal of Vacuum Science & Technology B, vol. 17, no. 6, pp. 3047–3051, 1999.DOI
  18. [18]B. Henke, E. Gullikson, and J. Davis, “X-ray interactions: Photoabsorption, scattering, transmission, and reflection at e = 50-30,000 ev, z = 1-92,” Atomic Data and Nuclear Data Tables, vol. 54, no. 2, pp. 181–342, 1993. [Online]. Available: https://www.sciencedirect.com/science/article/pii/S0092640X83710132
  19. [19]Center for X-Ray Optics. Lawrence Berkeley National Laboratory. (1993–2025) Index of refraction. [Online]. Available: https://henke.lbl.gov/optical_constants/getdb2.html
  20. [20]V. A. Es’kin and E. V. Ivanov, “Physics-informed neural systems for the simulation of electromagnetic wave diffraction from a lithography mask,” 2026. [Online]. Available: https://arxiv.org/abs/2603.15584
  21. [21]H. Tanabe, A. Jingui, and A. Takahashi, “Accelerating extreme ultraviolet lithography simulation with weakly guiding approximation and source position dependent transmission cross coefficient formula,” Journal of Micro/Nanopatterning, Materials, and Metrology, vol. 23, no. 1, pp. 014 201–014 201, 2024.
  22. [22]C.-M. Yuan and A. J. Strojwas, “Modeling optical microscope images of integrated-circuit structures,” Journal of the Optical Society of America A, vol. 8, no. 5, pp. 778–790, 1991.DOI
  23. [23]K. D. Lucas, H. Tanabe, and A. J. Strojwas, “Efficient and rigorous three-dimensional model for optical lithography simulation,” Journal of the Optical Society of America A, vol. 13, no. 11, pp. 2187–2199, 1996.DOI
  24. [24]V. Medvedev, A. Erdmann, and A. Rokopf, “3d mask simulation and lithographic imaging using physics-informed neural networks,” 2023. [Online]. Available: https://publica.fraunhofer.de/handle/publica/450903DOI
  25. [25]V. A. Es’kin and E. V. Ivanov, “Physics-informed neural networks and neural operators for a study of euv electromagnetic wave diffraction from a lithography mask,” in 2025 Days on Diffraction (DD), 2025, pp. 48–53.
  26. [26]V. A. Es’kin and E. V. Ivanov, “Gradient-based inverse lithography for euv masks via the waveguide method and a physics-informed neural operator,” 2026. [Online]. Available: https://arxiv.org/abs/2606.25753

Paper details

vlnc:W_kjkmjyx6
Version
v1 · latest
Published
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents