High Numerical Aperture Metalens for the Deep Ultraviolet

ABSTRACT

Deep‐ultraviolet (DUV) radiation is essential for advanced nanofabrication, biomedical imaging, and emerging photonic technologies, yet the absence of compact, high‐performance optical components in this spectral range has impeded progress. We demonstrate an aluminum nitride (AlN) metalens with record‐high numerical aperture (HNA) from 0.9 to 0.7, operating at 266 nm. Optical measurements confirm diffraction‐limited focusing performance. Importantly, we show a direct laser writing microscopy (DLWM) that lithographic direct writing serves microscopy purposes for the HNA metalens: besides enabling nanofabrication with 100 nm level critical dimension, it provides a universal characterization method for HNA DUV metalenses, addressing the challenges of constructing broadband HNA optical characterization systems in this wavelength range. Furthermore, the HNA AlN metalens for the first time integrates into a confocal photoluminescence system, enabling compact sub‐micrometer spectroscopic imaging with DUV excitation and a confocal photoluminescence imaging system using a metalens (NA = 0.7), resolving isolated features down to 300 nm. This work establishes AlN metalenses as a transformative and practical platform for high‐resolution DUV applications.

A record‐high‐NA AlN metalens operating at 266 nm is demonstrated for high‐resolution deep‐ultraviolet photonics. Diffraction‐limited focusing, 100 nm‐level direct laser writing, lithography‐based metalens characterization, and compact sub‐micrometer confocal photoluminescence imaging are achieved, establishing AlN metalenses as a transformative and practical platform for high‐resolution DUV applications.

1. Introduction

Deep‐ultraviolet (DUV) radiation spanning below 300 nm (<4.1 eV) forms the foundation of modern nanotechnology. Short DUV wavelengths enable semiconductor lithography, biomedical imaging, molecular spectroscopy, and food safety applications [1, 2, 3, 4, 5]. High photon energy and strong matter interactions make DUV indispensable for resolving nanoscale features and probing electronic transitions in materials. However, compact, high‐resolution DUV optical systems remain hindered by low‐loss material scarcity and conventional refractive optics limitations [6]. Most glasses and polymers exhibit strong DUV absorption, while specialized materials like magnesium fluoride (MgF₂) or calcium fluoride (CaF₂) are fragile, expensive, and challenging to nanostructure [7]. Refractive optics, though widely used, introduce bulkiness, fine surface polishing requirements, and alignment complexities that impede miniaturized system integration [8]. By leveraging subwavelength nanostructures to locally control light's phase, amplitude, and polarization, metasurfaces as transformative solutions enable ultrathin, lightweight optical elements with unprecedented design flexibility [9, 10, 11]. However, extending metasurface technology to the DUV regime demands materials combining high transparency, large refractive index (n > 2), and robust thermal and mechanical stability. That combination is rarely found in conventional dielectrics. A recent Diamond‐based metalens enables solar blind UV imaging [12]. Despite these well‐documented material advantages, the realization of DUV flat optics has historically been limited to low‐NA devices [13].

Recent advances in DUV metasurfaces have explored hafnium dioxide (HfO₂), zirconium dioxide (ZrO₂), and diamond [14, 15, 16]. While these materials offer moderate DUV transparency, practical adoption remains hampered by complex fabrication processes, limited scalability, or insufficient thermal resilience. Diamond metasurfaces require sophisticated plasma etching techniques, while polymer‐based designs degrade under high laser fluence [17]. Aluminum nitride (AlN), with its wide bandgap (∼6.2 eV), high thermal conductivity (∼320 W/mK), and semiconductor manufacturing compatibility, presents a compelling alternative [18, 19, 20, 21]. AlN's low extinction coefficient (k ∼ 0) at DUV wavelengths and exceptional thermal stability make it ideal for integrated photonics [22, 23], flat optics devices [24], and high‐power applications, such as ultrafast laser processing, where traditional materials fail [25]. Despite these advantages, experimental AlN metasurface demonstrations remain limited, particularly in achieving high numerical aperture (HNA) metalenses essential for diffraction‐limited focusing and unlocking the intrinsic resolution potential of this wavelength range [26]. Two challenges persist. First, DUV HNA metalens nanofabrication requires sub‐100 nm, sometimes sub‐50 nm feature sizes with sub‐200 nm pitch, together with high‐quality AlN films ensuring reliable etching performance. Second, characterization poses a fundamental bottleneck since higher NA optical systems, such as NA∼0.9, are impractical for most DUV wavelengths due to conventional DUV optics limitations, which further limit the optimization of nanofabrication if the device cannot be measured accurately.

Currently, strategies for achieving wavefront shaping in the DUV regime are categorized into two distinct approaches. The first, explored herein, involves the direct fabrication of linear light‐field modulation elements for fundamental DUV wavelengths. Conversely, recent theoretical and experimental advances have demonstrated that nonlinear geometric phase and diffraction can be utilized to generate and shape light in a single step [27, 28, 29]. These approaches effectively combine nonlinear wavelength conversion with simultaneous light‐field manipulation, allowing the conversion of visible or near‐infrared light into shorter wavelengths while imparting hyperbolic phase profiles. Additionally, non‐local bound states in the continuum (BICs) metasurfaces have shown high‐Q factors for DUV generation and focusing [10, 30]. However, while nonlinear and nonlocal metalenses offer significant advantages in spectral isolation and background noise suppression, their inherent energy conversion efficiency remains substantially lower than direct linear modulation, and they require ultra‐high‐peak‐power pulsed excitation. The direct linear modulation method presented in this work offers vastly superior absolute transmission and focusing efficiencies, lower fabrication complexity, and compatibility with continuous‐wave (CW) and low‐power DUV sources.

We address these challenges by demonstrating HNA AlN metalenses operating at 266 nm DUV wavelength, achieving NA up to 0.9, the highest reported for any DUV optical element with a focal spot full‐width‐at‐half‐maximum (FWHM) down to 150 nm, approaching the theoretical FWHM of 147.8 nm. This performance surpasses prior DUV metalenses, including recent AlN designs with NA = 0.2 and FWHM = 750 nm [13], and rivals state‐of‐the‐art EUV zone plates NA while offering far greater compactness and design freedom [31]. Our metalens leverages AlN's high refractive index (n = 2.2) and low loss, combined with optimized Mie‐resonant nanopillars, to enable precise phase modulation across the metalens. More importantly, besides optical characterization, we establish a direct laser writing microscopy (DLWM) as a universal and practical approach to evaluate DUV HNA metalenses. Using this lithographic method, we directly measure the metalens’ depth of focus and resolution, which agree well with optical measurements. It is worth noting that it is a broadband characterization due to the intrinsic feature of optical lithography, and the resolution is mostly limited by the properties of the photoresist used. For applications, besides the achievement of 100 nm level critical dimension in nanofabrication, AlN metalens (NA = 0.7) integration into confocal photoluminescence (PL) setups for the first time demonstrates utility for compact nanoscale spectroscopic imaging, enabling high‐resolution sub‐800 nm PL mapping using DUV excitation with visible emission.

2. Results and Discussion

2.1. HNA DUV Metalens Design and Fabrication

The schematic illustration of the HNA AlN metalens operating at a wavelength of 266 nm is presented in Figure 1a. The device is composed of periodic unit cells, hereafter referred to as meta‐atoms, each consisting of a single rectangular nanostructure fabricated from epitaxial AlN and positioned on a sapphire substrate (Figure 1b). Owing to their single‐crystalline nature, these AlN films support strong light–matter interactions for incident circularly polarized light and enable precise manipulation of the transmitted polarization state.

FIGURE 1.

Deep ultraviolet AlN metalens design and fabrication. (a) 3D schematic of AlN metalens operating at wavelength 266 nm with incident right‐circular polarized light from the backside and focused transmitted left circular polarized light. (b) 3D schematic of AlN meta‐atom with square lattice of (P_x = P_y = 200 nm), length (L = 180 nm), width (W = 40 nm), and height (H = 350 nm). (c) Simulated transmitted light and accumulated phase of light versus rotation of AlN meta‐atom around the vertical axis. (d) Simulated polarization conversion efficiency for cross‐polarized light with over 80% efficiency at a wavelength of 266 nm. (e) Simulated phase profile of HNA AlN metalens with NA = 0.9 and a diameter of 100 µm. (f) Theoretical Pancharatnam–Berry (PB) phase and accurate reconstruction of fabricated metalens via meta‐atom rotation. (g) Measured XRD of AlN epitaxial film of thickness 350 nm on sapphire substrate. (inset) Measured AFM of AlN epitaxial film of thickness 350 nm on sapphire substrate has RMS = 0.185 nm. (h) Measured transmission of AlN thin film and AlN metalens in the DUV range. (i) Scanning electron microscope (SEM) image of fabricated metalens with tilted inset images near the center and edge.

To design the meta‐atoms, 3D electromagnetic simulations were performed using finite‐difference time‐domain (FDTD) methods. A full 2π phase modulation of transmitted light was achieved by rotating the meta‐atom about its central vertical axis. Uniform and high simulated transmission efficiency was obtained for all rotation angles (Figure 1c). This high efficiency was attained through a parametric optimization of the meta‐atom's length, width, and height (Figure S2). Furthermore, cross‐polarized transmission characteristics are presented in (Figure 1d), where enhanced metalens performance is linked to increased cross‐polarization conversion efficiency (PCE). To maximize PCE, both geometric parameters and material configurations were systematically optimized, yielding efficiencies exceeding 80% at 266 nm. The high PCE correlates with the strong confinement of both electric and magnetic fields within the AlN meta‐atom at the target wavelength. Field distributions for various AlN heights are plotted in Figure S3).

Another critical challenge in HNA metaoptics is managing the steep spatial phase gradient at the periphery of the lens. According to the Nyquist–Shannon sampling theorem, avoiding phase aliasing dictates that the structural periodicity (P) must be smaller than 𝛌/2NA. For an NA = 0.9 lens operating at 266 nm, this strict criterion requires p < 147.8 nm. However, enforcing this periodicity severely restricts the physical dimensions of the rectangular meta‐atoms, precluding the high‐aspect‐ratio geometries necessary to support the strong electromagnetic mode confinement required for DUV transmission [13, 32]. Consequently, a lattice period of 200 nm was deliberately selected. While this subwavelength period marginally exceeds the strict Nyquist limit, resulting in minor phase aliasing and the redirection of a small percentage of energy into higher‐order parasitic diffraction lobes (Figure S5), it preserves the macroscopic hyperbolic phase profile required for diffraction‐limited focusing. Crucially, this geometric compromise allows for highly robust nanofabrication (Figures S1 and S6) and ensures an exceptionally near‐vanishing co‐polarization and high cross‐polarization conversion efficiency (Figure S4) that would be impossible to achieve under strictly unaliased spatial constraints.

The design and numerical modelling of DUV metalens began with the precise spatial arrangement of AlN meta‐atoms to tailor the far‐field scattering profile, ensuring constructive interference at the focal point positioned at a distance equal to the lens focal length. The required spatial phase distribution followed a near‐ideal hyperbolic phase profile, expressed as: [33]

$$\phi \left(x,y\right)=\frac{2\pi }{\lambda }\left(f-\sqrt{f^2+x^2+y^2}\right)$$ (1)

where f represents the focal length, 𝛌 = 266 nm is the operating wavelength, and the metalens diameter is fixed to 100 µm. The ideal phase map corresponding to this profile is presented in Figure 1e. The comparison between the ideal line phase profile along the radial axis of the lens and the experimentally derived phase profile is illustrated in Figure 1f. The latter was obtained by mapping the simulated phase response to specific meta‐atom orientations. A fixed meta‐atom geometry was employed across the metalens to facilitate a robust nanofabrication process and achieve high fidelity in reconstructing the designed phase distribution. The crystallinity of AlN nanopillars, confirmed by X‐ray diffraction (XRD) measurements, is shown in Figure 1g. In addition, epitaxial AlN films exhibit exceptionally low surface roughness (<0.2 nm root‐mean‐square), which minimizes scattering losses and facilitates high optical transmission. The surface morphology of AlN films, characterized by atomic force microscopy (AFM), is provided in (Figure 1g inset). The transmission spectra of both the AlN thin film and the fabricated metalens are shown in (Figure 1h), demonstrating an optical transmission exceeding 90%, which supports strong focusing efficiency at the design wavelength. A scanning electron microscopy (SEM) image of the fabricated HNA AlN metalens with NA = 0.9 is shown in Figure 1i, with vertical sidewalls of AlN nanopillars. Additional nanofabrication details are provided in the Methods section.

2.2. Optical Characterization of the HNA Metalens

The DUV focusing performance of the HNA AlN metalens was experimentally characterized using a custom‐built imaging setup. A nanosecond pulsed laser (repetition rate: 100 kHz) operating at 266 nm served as the illumination source. The beam polarization was converted to circular polarization prior to backside illumination of the metalens, which was mounted on a computer‐controlled three‐axis motorized stage. The far‐field DUV images were captured using a high‐sensitivity DUV camera, as schematically illustrated in Figure 2a. Additional details of the experimental setup are provided in the Methods section.

FIGURE 2.

Optical characterization of fabricated HNA AlN metalens. (a) 3D schematic of the optical setup used. Measured light intensity along propagation z‐axis (x‐z plane) for HNA AlN metalenses (LP) is linear polarizer, (QWP) is DUV quarter wave plate, (DUV Obj) is HNA commercial DUV optical objective, (b) NA = 0.7, (c) NA = 0.8, and (d) NA = 0.9, respectively. Experimental light intensity profile at focal plane (x‐y plane) for HNA AlN metalenses (e) NA = 0.7, (f) NA = 0.8, and (g) NA = 0.9, respectively. Measured light intensity versus the x‐axis and theoretical calculations of (h) NA = 0.7, (i) NA = 0.8, and (j) NA = 0.9, respectively. Measured modulation transfer function (MTF) versus theoretical calculations of diffraction‐limited focusing of (k) NA = 0.7, (l) NA = 0.8, and (m) NA = 0.9, respectively.

The measured focused DUV light propagation along the z‐axis of NA = 0.7 (Figure 2b), where the x–z plane is measured at each z‐value of step 200 nm. A well‐defined focal spot is observed at a working distance of 49 µm. For NA = 0.8 in (Figure 2c) the focal length is 37.2 µm. The corresponding x–z light propagation measurement for NA = 0.9 is shown in (Figure 2d), which also exhibits clear axial focusing at a distance of 25 µm. The measured focal spot intensity distributions along the x–y plane for NA = 0.7 (Figure 2e), NA = 0.8 (Figure 2f), and NA = 0.9 (Figure 2g) show highly symmetric profiles, indicating excellent nanofabrication fidelity and minimal optical loss. Quantitative analysis of the focal spots is performed to compare the measured and simulated line profiles along the x‐direction for NA = 0.7, where the full width at half maximum (FWHM) was 204 nm experimentally and 190 nm in simulation (Figure 2h). For NA = 0.8, the measured FWHM was 174 nm experimentally and 168 nm by simulation (Figure 2i). For NA = 0.9, the measured linewidth yields FWHM values of 154 nm (experiment) and 148 nm (simulation) (Figure 2j). These results demonstrate close agreement between experiment and simulation across different numerical apertures. To further evaluate the optical quality of the fabricated metalens, the modulation transfer function (MTF) was retrieved from the line spread function of the measured focal spot and compared to theoretical predictions. The MTF characterizes the contrast reproduction capability of the imaging system at different spatial frequencies [34]. MTF of NA = 0.7 (Figure 2k), NA = 0.8 (Figure 2l), and NA = 0.9 (Figure 2m) have experimental MTF cutoff frequency that closely approaches the theoretical value. The experimental MTF closely matches the theoretical prediction at low spatial frequencies. However, a slight degradation is observed at mid‐to‐high spatial frequencies. This deviation is attributed to unavoidable physical fabrication imperfections, including AlN nanopillar sidewall angles that deviate slightly from true vertical (<90°), and nanoscale sidewall roughness induced during the reactive ion etching process. These geometric perturbations introduce local phase errors that scatter light away from the primary geometric focus, reducing high‐frequency spatial contrast. Experimental focusing efficiency and encircled energy ratios are critical metrics for evaluating the practical utility of meta‐optical devices. Section S7 and Table S1 summarize these experimentally derived performance metrics. Notably, our metalenses achieved experimental focusing efficiencies up to 31.2% and encircled energy ratios up to 78.1% (for NA = 0.7). These results confirm that despite the strict fabrication and sampling constraints of the DUV regime, the devices successfully concentrate substantial energy into the focal spot, compromising loss channels of fabrication and design constraints [35]. Overall, the fabricated DUV metalenses achieved focal lengths within +5% of their theoretical design values and exhibited strong agreement with simulations in both focal spot size and MTF performance. Furthermore, SEM of fabricated metalenses with NA = 0.7 and NA = 0.8 were provided in Figures S10 and S11. The close agreement between experimental measurements and simulations underscores the potential of the proposed HNA AlN metalens platform for high‐performance optical systems operating in the DUV spectral range [36, 37, 38, 39].

2.3. DUV DLWM Demonstration

A major challenge in evaluating HNA optics at DUV wavelengths is the absence of broadband HNA refractive objectives, which makes conventional optical characterization difficult or impractical. To address this limitation, we establish a DUV DLWM for characterizing HNA metalenses. In this approach, the focusing performance of the metalens is translated directly into lithographic outcomes such as resolution and depth of focus. To ensure the highest accuracy and comparability, we use grating patterns, the gold‐standard test structure in lithography, to enable precise and quantitative assessment of the metalens performance. The metalens (NA = 0.7) was deliberately selected for the direct laser writing demonstration because its longer working distance (∼49 µm) provides crucial mechanical clearance to prevent collisions with the photoresist‐coated sample during dynamic scanning. Furthermore, its higher absolute focusing efficiency ensures optimal optical throughput for resist exposure. As shown in (Figure 3a), a nanosecond pulsed laser at 266 nm was tightly focused by an AlN metalens (NA = 0.7) onto a photoresist‐coated silicon substrate, producing well‐defined grating patterns with periods from 1 µm down to 270 nm and linewidths approaching ∼100 nm, as shown in Figure 3b. The measured values are in good agreement with the Rayleigh criterion of 244 nm derived from the measured 204 nm focal spot FWHM. Minor deviations arise from realistic lithographic effects, particularly resist behavior and thickness, which may reflect practical operating conditions rather than ideal optical limits. Furthermore, in Figure 3c, direct lithographic measurements of the depth of focus ∼0.8 µm matched perfectly with the measurements from optical setups (Figure 2b), confirming the reliability of this method. This lithography‐based DLWM thus enables not only a nanofabrication technique employing the HNA metalens but also provides a robust and broadly applicable pathway for assessing the optical properties of HNA metalenses without requiring specialized HNA refractive objectives [40, 41, 42].

FIGURE 3.

DUV direct laser writing microscopy (DLWM) demonstration. (a) Optical image of the DLWM system for characterizing HNA AlN metalens. (Inset) Zoom in to flat optics HNA AlN metalens integration and schematic of direct laser writing of sub 200 nm features on ma‐N e‐beam resist. (b) Measured the linewidth of the grating structure after resist development with different pitches between lines. (Inset) SEM images of different grating periods after development. (c) Measured linewidth of grating structure versus z‐level scan around focal length. (Inset) SEM images of the grating with a period of 1000 nm after development.

2.4. High‐Resolution DUV Spectroscopy and Imaging

High‐resolution DUV spectroscopy and imaging are particularly important for applications in semiconductor inspection [43], biomedical diagnostics [44], food safety [45], and environmental monitoring [46]. To demonstrate the nanoscale spectroscopic capability, we implemented the first scanning PL imaging experiment based on an HNA (NA = 0.7) DUV metalens. This NA offers a longer working distance and larger depth of focus for practical alignment, and a higher NA would not improve the PL imaging resolution, which is mainly limited by the longer PL emission wavelength. The metalens is integrated in a commercial confocal PL scanning system, and the supporting optical setup enables a circularly polarized, collimated DUV beam incident from the backside of the metalens, which focuses the light onto a photoluminescent sample. The emitted visible light was collected by an HNA visible objective and subsequently relayed to a spectrometer, as schematically illustrated in Figure 4a. Additional details of the imaging setup are provided in the Methods section. To evaluate the imaging performance, we fabricated test structures by patterning grating arrays into a 50 nm‐thick aluminum (Al) film deposited on a UV‐grade silica substrate using focused ion beam (FIB) milling. The gratings featured varying periods for resolution assessment. For PL imaging, the patterned metal gratings were spin‐coated with cadmium selenide (CdSe) quantum dots (QDs) to produce visible emission under DUV excitation (Figure 4b). All PL measurements were performed in transmission mode. Measured PL of thin film QDs shows peak emission at 625 nm (Figure 4c). During operation, the patterned Al‐QDs sample was illuminated from the front side with a 266 nm laser and positioned precisely at the focal plane of the metalens to ensure optimal image formation. More details in the Methods section. SEM of fabricated Al grating without QDs with pitches 1000 nm and 800 nm are plotted in Figure 4d,e. The HNA AlN metalens successfully resolved PL emission features from gratings with pitches of 1000 nm (Figure 4f) and 800 nm (Figure 4 g), using 0.5 and 0.9 NA visible objectives, respectively. The theoretical Rayleigh resolution of the 266 nm excitation, based on an experimentally extracted effective NA_eff of 0.667, is ∼244 nm. However, PL imaging relies entirely on collected emission. Because CdSe QDs emit at a wavelength of 625 nm and the signal passes through the backside substrate (introducing optical aberrations) [47], the absolute diffraction limit captured by an NA = 0.9 objective is fundamentally bounded to ∼423 nm. Despite these collection‐side constraints, the tightly confined DUV excitation from the HNA metalens successfully resolves isolated features down to 300 nm with an 800 nm full‐pitch resolution (Figure 4g). Control experiments utilizing a 532 nm laser confirmed identical resolution boundaries, proving limits are emission‐governed (Figure S14). These results highlight the superior application performance of the HNA AlN metalens for high‐resolution PL imaging under DUV excitation and pave the way for compact, high‐performance DUV spectroscopy and imaging systems enabled by flat‐optics technology [48].

FIGURE 4.

Sub one‐micrometer photoluminescence (PL) imaging system using HNA AlN metalens. (a) Schematic of PL imaging setup with CdSe QDs covering an aluminum grating structure of different periods. (b) 3D schematic of an aluminum grating covered with QDs. (c) Measured PL spectrum of QDs using 532 nm laser excitation. SEM images of Al grating before QDs coating for gratings of pitches (d) 800 nm and (e) 1000 nm, respectively. PL mapping images for HNA AlN metalens for gratings of pitches (f) 800 nm, and (g) 1000 nm using 266 nm laser excitation.

3. Conclusion

We report the first experimental realization of an AlN metalens with a record‐high numerical aperture up to 0.9 at 266 nm, delivering diffraction‐limited focusing with a ∼150 nm spot size. This achievement vastly eclipses previous DUV metalens benchmarks [12, 13, 32], quantifying a major leap in flat‐optic capabilities. Importantly, we develop a DUV DLWM platform as a quantitative characterization method that resolves a longstanding challenge in measuring DUV HNA metalenses. This approach leverages the intrinsic broadband nature of lithography and eliminates reliance on conventional HNA refractive optical systems. For applications, the DUV metalens (NA = 0.7) is integrated into a confocal PL system, achieving 300 nm isolated feature resolution and sub‐800 nm pitch resolution in spectroscopic imaging using DUV excitation. This breakthrough establishes a new performance benchmark for DUV flat optics, pushing the practical spectroscopic imaging resolution of DUV metalenses to the sub‐micrometer regime. Together, these advances bridge the gap between high‐performance DUV optics and high‐end DUV applications, positioning AlN metalenses as a transformative platform not only for compact nanofabrication, high‐resolution spectroscopy, and next‐generation DUV photonic systems [48], but also for emerging technologies such as high‐density optical data storage and quantum photonics. Owing to their ultracompact form factor, HNA, and precise phase modulation capability, HNA DUV metalenses and even arrays can tightly concentrate high‐energy photons into nanoscale volumes, which is highly attractive for next‐generation optical data storage. In addition, their integration with nonlinear crystals could enable more efficient generation and collection of entangled photon pairs through spontaneous parametric down‐conversion, providing a scalable route toward ultra‐compact quantum light sources.

4. Methods

4.1. Transmission and Phase Simulations

Optical simulations were conducted using a 3D finite‐difference time‐domain solver (Lumerical FDTD Solutions). Material refractive indices were imported from literature data. A linearly polarized plane wave, incident from the substrate side, illuminated the meta‐atom. Periodic boundary conditions are applied in the x‐y plane and a perfectly matched layer (PML) along the z‐axis. A mesh size of <3 nm ensured spatial resolution. A 2D field monitor positioned above the structure recorded transmitted fields. The transmitted phase was determined by analyzing the relative phase difference between orthogonal electric field components for each rotated angle. Simulations employed an auto‐shutoff threshold of 10⁻⁵ to guarantee numerical accuracy.

4.2. Polarization Conversion Efficiency Calculations

The polarization conversion efficiency (PCE) was computed using Lumerical FDTD Solutions by analyzing scattering parameters (S‐parameters) in two orthogonal linear polarization configurations. Two independent simulations were performed: one with a linearly x‐polarized plane wave source and another with a y‐polarized source, spanning the ultraviolet wavelength range of 220–340 nm. For each simulation, the complex transmission coefficients (S _21,x and S _21,y ) were extracted using an S‐parameter analysis box positioned in the transmission region. These components were then transformed into left‐ and right‐circularly polarized (LCP, RCP) transmission coefficients using standard Jones matrix formulations representing the Pancharatnam–Berry phase rotation:

$$S_{21,RR}=\frac{1}{2}\left(S_{21,xx}+S_{21,yy}+i{S}_{21,xy}-i{S}_{21,yx}\right)$$ (2)

$$S_{21,RL}=\frac{1}{2}\left(S_{21,xx}-S_{21,yy}-i{S}_{21,xy}-i{S}_{21,yx}\right)$$ (3)

The PCE for conversion from LCP to RCP (representing the cross‐polarized, phase‐modulated component) was rigorously calculated as:

(4)

PMLs were placed in the vertical direction, and a periodic boundary condition was applied at horizontal boundaries. The mesh size was refined near the structure to ensure simulation accuracy. Material dispersion was modeled using experimentally measured refractive indices. The optical setup used to measure focusing and encircled efficiency is plotted in Figure S7.

4.3. Nanofabrication of AlN metalens

Commercially available AlN on a sapphire substrate was used for the fabrication. The thickness of AlN is ∼350 nm (Figure S13), and the sapphire substrate thickness is 500 µm. Hydrogen silsesquioxane (HSQ) (DisChem, 12 wt% in Methyl isobutyl ketone (MIBK)) was spin‐coated onto the AlN/sapphire substrate, followed by post‐heating at 180°C for 2 min to reach a thickness of ∼150 nm to 180 nm as shown in Figure S1. Subsequently, a water‐soluble conductive polymer ESPACER TM300Z layer (Resonac) was spin‐coated onto the substrate for surface charging dissipation. The e‐beam lithography (EBL) system (ELIONIX, ELS‐7800) was used to draw an optimized metalens design at 100 kV and 500 pA, followed by developing HSQ in an aqueous mixture of NaOH and NaCl salt. A chlorine‐based recipe using an Inductive Coupled Plasma Reactive Ion Etching (ICP‐RIE) system (Oxford Instruments, OIPT Plasmalab System 100) were optimized for the AlN nanorod etching with chamber pressure of 2 Pa. After the pattern transfer process, HSQ mask was removed by buffered HF solvent.

4.4. Metalens Fabrication Characterization

Scanning electron microscopy (SEM) (Hitachi SU8220) characterization was performed under high vacuum using an acceleration voltage of 1–5 kV and a working distance of 5–10 mm. A thin chromium (Cr) film (∼15 nm) was deposited onto the sample surface using an e‐beam evaporator (Denton Explorer) to mitigate charging effects during SEM imaging. A surface profiler (KLA Tencor P16) was used to determine the thickness/height of the film/features at different fabrication stages. Optical transmission was performed using a broadband UV–vis spectrometer (Shimadzu UV3600). The DUV focusing performance and efficiency of the AlN metalenses were characterized using a home‐built 266 nm optical setup. The incident DUV beam was expanded and converted to circular polarization using a linear polarizer and a zero‐order quarter‐wave plate before illuminating the metalens. The focused beam was collected by an HNA DUV‐compatible objective and relayed to a UV‐sensitive camera for focal‐spot imaging, while a beam splitter directed part of the signal to a calibrated power meter for efficiency measurements. The focusing efficiency was evaluated using a reference‐calibrated power measurement method with aperture correction based on the measured incident beam profile, and the encircled efficiency was extracted from the focal‐plane intensity distribution by integrating the optical power within a defined region around the focal spot. Detailed optical setup, calibration procedure, aperture correction method, and efficiency definitions are provided in Section S7.

4.5. Metalens Characterization Optical Setup

The optical measurements were performed on a modified surface probe microscopy system (PSTM, WITec alpha300S) adapted for deep‐ultraviolet operation. To accurately map the point spread function (PSF), MTF, and measure the focusing dynamics of HNA AlN metalenses, the collection optics must possess an equal or greater numerical aperture. Consequently, the optical setup was rigorously calibrated utilizing a commercial DUV‐grade refractive objective with an NA of 0.9 (Nikon, DUV 266/660 100×/0.90). The utilization of this high‐end refractive objective ensures that the extreme angles of the highly convergent beam are fully captured without artificial truncation. Furthermore, a major novelty of our optimized AlN design is the structural confinement that maximizes PCE. This exceptionally high conversion efficiency acts as an intrinsic optical filter, fundamentally suppressing the transmission of unconverted co‐polarized light below 1% (Figure 1d). Optically, any residual unconverted co‐polarized light propagates as a zero‐order, uniform background. When collected by the HNA objective, this co‐polarized background is focused at the back focal plane at a very short focal length. Over the subsequent tube length prior to the camera, this beam strongly diverges, effectively preventing it from being collected as background noise at the imaging plane. This dual suppression mechanism eliminates the noise that typically plagues conventional metasurfaces, thereby enabling high‐fidelity focusing without the need for external polarization analyzers in the optical train. The original detection path, consisting of a pinhole and sensor, was substituted by a tube lens combined with a CCD camera (Allied Vision Alvium 1800 U‐812m‐UV) to enable direct image acquisition. Illumination was provided by a 266 nm pulsed laser source. A linear polarizer and a quarter‐wave plate (Thorlabs) were sequentially inserted in the incident beam path for polarization preparation. The fabricated metalens was mounted on a UV‐quartz substrate with its patterned surface facing upward. During measurement, the 266 nm beam was incident from the substrate side, and the focal spot transmitted through the metalens was collected by the objective, converted into collimated light, and relayed via the tube lens to the CMOS camera for image recording. The reported focusing efficiency is defined as the optical power enclosed within a diameter of 3× FWHM at the focal plane, divided by the total incident power on the metalens aperture. This is calculated as the product of the reference‐calibrated power‐meter efficiency and the encircled energy ratio within 3× FWHM extracted from the focal‐plane CCD image (Section S7 and Equations S3–S5). Importantly, because the reference measurement utilizes an AR‐coated commercial UV lens while the metalens is uncoated, all inherent Fresnel reflection losses at the metalens interfaces are rigorously accounted for in the final efficiency values. The power‐meter measurement and encircled‐energy extraction are described separately in Section S7.

4.6. DLWM Characterization

Silicon substrates were cleaned sequentially in acetone, isopropanol, and deionized water using an ultrasonic cleaner, dried with nitrogen, and subjected to dehydration baking at 200°C for 10 min. A photoresist layer (ma‐N 2400 series) was spin‐coated (CEE Model 200X, 4000 rpm, 60 s) and soft‐baked at 100°C for 1 min to form a uniform negative resist film suitable for laser exposure. A commercial direct laser writer platform (MetaOptics DLW system, Singapore) was employed for patterning. The fabricated AlN metalens was mounted onto the 2D lens stage with the structured surface facing downward. A 266 nm laser beam was incident from the top and focused through the metalens onto the prepared substrate, enabling maskless fabrication of nanoscale features. Following development in the ma‐D 525 developer, the samples were coated with a thin Au film using a sputter coater (JEOL JFC‐1200, 40 s deposition) to reduce charging effects. Structural characterization was performed with a dual‐beam focused ion beam system (FEI Helios Nanolab 600) to resolve the fabricated nanostructures with high‐resolution imaging.

4.7. Photoluminescence Imaging Measurement

Commercial glass substrates were coated with a 50 nm Al film using an E‐beam evaporation system (Denton Explorer). Grating structures with different pitches (1000 nm, 800 nm, and 600 nm) were fabricated on the Al surface by a dual‐beam focused ion beam (FIB) system (FEI Helios Nanolab 600). Subsequently, commercially available water‐soluble CdSe/ZnS quantum dots were spin‐coated onto the grating surface using a spin coater (CEE Model 200X, 800 rpm, 60 s) and dried on a hot plate at 60°C for 1 min. For optical measurements, the prepared samples were mounted on a hollow UV‐grade quartz substrate with the quantum dot side facing downward on the wafer stage of a PSTM system (WITec alpha300S). A 266 nm laser beam was incident from the bottom and focused by the metalens onto the grating slits to excite the red emission from quantum dots. The photoluminescence signal was collected by a 100x visible objective lens (NA = 0.9), and after alignment of the PSTM system, a selected grating region was scanned to obtain spatially resolved PL images. The collected signal was coupled into a spectrometer through the PSTM detection path, and the data were processed into PL images, which constituted the photoluminescence imaging results.

Author Contributions

Omar A. M. Abdelraouf designed HNA AlN metalenses and performed all optical simulations. Chee Leng Lay fabricated metalenses and did an SEM measurement. Haoyu Ge characterized the AlN metalenses and performed photoluminescence imaging, and Haoyu Ge, Ziyu Zhou, Youpeng Zhang, and Yingjie Liu conducted the direct laser writing measurements. Yi Fan Chen and Ai Jia Sim conducted XRD and AFM measurements. Omar A. M. Abdelraouf wrote the manuscript with the contributions from Chee Leng Lay and Haoyu Ge. Zeng Wang conceived the idea and supervised the project. All authors reviewed and approved the final version of the manuscript.

Conflicts of Interest

The authors declare no conflict of interest.

Supporting information

Supporting File: adma74047‐sup‐0001‐SuppMat.docx.

Abdelraouf O. A. M., Lay C. L., Ge H., et al. “High Numerical Aperture Metalens for the Deep Ultraviolet.” Advanced Materials 38, no. 46 (2026): e74047. 10.1002/adma.74047

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.