betaFeedback

Reading v1View latest version

WorksPhysics and Astronomy

Deep-UV All-dielectric qBIC Metasurface for Refractive Index Sensing

Abstract

All-dielectric metasurfaces supporting quasi-bound states in the continuum (qBICs) have emerged as powerful platforms for biosensing, yet their operation remains largely concentrated in the visible and near-infrared regimes. Extending these resonances into the deep-ultraviolet (DUV) to probe the intrinsic electronic absorption of biomolecules presents substantial material and nanofabrication challenges. Here, we experimentally demonstrate a hafnium dioxide (HfO2_{2}) metasurface supporting DUV qBICs in the wavelength range from 225 nm to 340 nm with quality factors reaching 500. We further demonstrate its refractive index sensing capabilities through integration with a microfluidic cell and optical characterization in reflection mode. We show that the introduction of an aqueous buffer leads to a further quality factor enhancement. Achieving a sensitivity of up to approximately 60 nm/RIU and an operational figure of merit of 40 RIU−1\mathrm{RIU}^{-1}, this low-loss HfO2_{2} metasurface provides a robust foundation for label-free DUV biosensing.

Metasurfaces, periodic arrays of subwavelength building blocks, provide versatile control over light at the nanoscale and offer compact alternatives to bulky conventional optical components [1, 2]. By tailoring the geometrical parameters of the metasurface building blocks, their arrangement, and material composition, these platforms can be engineered to alter the properties of light at the nanoscale [3, 4, 5]. Recently, low-loss all-dielectric platforms have enabled nanophotonic applications across ultraviolet (UV) [6], visible (VIS) [7] and infrared (IR) [8] spectral ranges. Metasurfaces supporting quasi-bound states in the continuum (qBICs) have attracted considerable attention owing to their high quality (QQ) factor resonances and strong field enhancement [9, 10, 11, 12, 13, 14]. These states originate from symmetry-protected bound states in the continuum (BICs) [15], which are inherently non-radiative. Perturbation of the system — such as geometric asymmetry, oblique incidence, or finite-size effects — opens a radiation channel, giving rise to qBICs [9, 10].

Despite successful experimental demonstrations in the VIS and IR wavelength ranges, extending these high-QQ resonances into the UV regime remains a formidable challenge due to material constraints and fabrication challenges. Recent theoretical studies have explored diamond-based UV metasurface [16] and dielectric nanopillars or freestanding membranes to mitigate substrate-induced losses [17]. Beyond driving nonlinear nanophotonics [18] and enhancing single-photon emission for UV quantum optics [19], a primary motivation for extending qBICs into the deep-UV (DUV, 200 nm – 290 nm) and near-UV (NUV, 290 nm – 400 nm) lies in the advanced label-free detection of biomolecules. While IR qBIC platforms excel at probing vibrational modes of chemical bonds of various analytes [20, 21], they remain insensitive to molecule-specific electronic absorption modes. Probing these intrinsic modes requires access to the DUV, since many biologically relevant molecules, including nucleic acids and aromatic amino acids, exhibit characteristic electronic absorption in this spectral region [22, 23]. To get there, one must overcome a material bottleneck, as standard high-index dielectrics, such as silicon (Si) or titanium dioxide (TiO2_2), are too absorptive in the UV. So far, UV nanophotonics has relied on alternative wide-bandgap materials, including hafnium dioxide (HfO2_2), niobium pentoxide (Nb2_2O5_5), and aluminium nitride (AlN) [6]. However, achieving high-QQ resonances with large near-field enhancement while maintaining DUV transparency and fabrication feasibility remains a challenge. Here we demonstrate that HfO2_2, with its wide bandgap exceeding 5 eV and sufficiently high refractive index (n>2n > 2), offers a promising material platform for supporting robust qBICs down to the DUV. We designed, fabricated, and experimentally investigated an all-dielectric HfO2_2 metasurface supporting resonances that span from 225 nm to 340 nm, with the QQ-factor reaching 500. Finally, we integrated the metasurface into a microfluidic cell to assess its performance in an aqueous environment. We demonstrated that introducing an aqueous buffer, which leads to a reduced refractive index contrast, results in an enhanced QQ-factor. These results establish our metasurface as a robust high-QQ platform for future label-free DUV biosensing applications.

To realize the high-QQ resonances in the UV, we first have to establish the metasurface geometry and material platform based on its fabrication feasibility, available refractive index, and absorption edge wavelength. Figure 1a illustrates the proposed metasurface unit cell, which contains two nanorods with the same width w=40w = 40 nm and height h=70h = 70 nm. The length of the first rod is fixed at L1=144L_1 = 144 nm, while the length of the second rod L2L_2 can be scaled down using an asymmetry factor AA following L2=AL1L_2 = A L_1. This asymmetry factor enables the excitation of qBIC originating from a symmetry-protected BIC based on antiparallel dipole resonances [10]. The periodicity along the xx-axis and yy-axis is px=164p_x = 164 nm and py=172p_y = 172 nm, respectively. The experimental realization of the corresponding metasurface is shown in the scanning electron microscope (SEM) image in the inset of Figure 1a.

To further investigate the physical origin of these high-QQ modes, we analyzed the electric fields within the unit cell of a metasurface with nominal lateral dimensions (S=1S = 1) and A=0.86A = 0.86 at a wavelength of λres=264\lambda_{\mathrm{res}} = 264 nm. Figure 1b shows a maximum enhancement of the electric field intensity Imax⁡I_{\max} that exceeds 700 in the x−yx-y plane positioned 35 nm above the substrate, cutting through the middle of the nanorods. More importantly, the high-intensity near-field hotspots extend outside the nanorod volume, making them easily accessible for the target bioanalytes [26]. However, realizing such strongly confined resonances in the DUV critically depends on the optical properties of the constituent material. To minimize the losses in the DUV, we have to fabricate the metasurface out of a wide-bandgap (>5> 5 eV) material. While conventional high-index dielectrics, like Si and TiO2_2, are standard materials for VIS and IR nanophotonics, their significant extinction coefficients in the UV (see Figure 1c) do not allow access to the qBICs in the desired wavelength range. We verified this by full-wave electromagnetic simulations of metasurfaces based on the double-rod unit cell made of these materials (see Methods). The simulated transmittance spectra shown in Figure 1d confirm that the strong UV absorption of Si and TiO2_2 eliminates the resonance,

Unit cell design and material platform selection for DUV metasurface supporting quasi-bound states in the continuum (qBICs)

Figure 1. Unit cell design and material platform selection for DUV metasurface supporting quasi-bound states in the continuum (qBICs). (a) Schematics of the unit cell consisting of two nanorods on a fused silica substrate. The unit cell is defined by the periodicities (px=164p_x = 164 nm, py=172p_y = 172 nm), nanorod length (L1=144L_1 = 144 nm), width (w=40w = 40 nm) and height (h=70h = 70 nm). To transform a BIC into a qBIC, the asymmetry factor A=L2/L1A = L_2/L_1 is introduced for the length of one of the nanorods. Inset shows a tilted view scanning electron microscope (SEM) image of the fabricated metasurface. The scale bar is 200 nm. (b) Simulated field enhancement profiles in the x−yx-y and x−zx-z planes of the unit cell for S=1S = 1 and A=0.86A = 0.86. The x−yx-y monitor was positioned 35 nm above the substrate through the center of the nanorods. The x−zx-z monitor was placed in the center of the shorter L2L_2 nanorod. The white arrows represent electric field vectors in an antiparallel arrangement. (c) Real (nn, solid lines) and imaginary (kk, dashed lines) parts of the complex refractive index of two conventional high-index dielectrics (Si, TiO2_2, adapted from [24] and [25], respectively) and HfO2_2 (our data, see Methods) in the UV spectral range. (d) Simulated transmittance spectra of the metasurface presented in (a) utilizing HfO2_2, TiO2_2 and Si. The strong UV absorption of Si and TiO2_2 eliminates the resonance, whereas HfO2_2 supports a sharp high-QQ resonance at the wavelength of 264 nm. (e) Table summarizing the absorption edge (AE) wavelengths (λAE\lambda_{\mathrm{AE}}) and refractive indices (nn) of UV-compatible materials. (f) Simulated transmittance spectra map as a function of the nanorod refractive index nn (k=0k = 0). The map shows the emergence of the qBIC and a secondary magnetic resonance (MR) above the diffraction limit (dashed line) as nn increases. The horizontal position of the material markers indicates the corresponding λAE\lambda_{\mathrm{AE}}.

whereas HfO2\mathrm{HfO_2} supports a sharp high-QQ qBIC at the DUV wavelength of 264 nm.

Apart from HfO2\mathrm{HfO_2} [27], we also evaluated other wide-bandgap materials that were previously used in DUV nanophotonics, including aluminium nitride (AlN\mathrm{AlN}) [28], silicon nitride (Si3N4\mathrm{Si_3N_4}) [29], tantalum pentoxide (Ta2O5\mathrm{Ta_2O_5}) [30], niobium pentoxide (Nb2O5\mathrm{Nb_2O_5}) [31] and diamond [32, 33]. The table in Figure 1e shows their absorption edges (λAE\lambda_{\mathrm{AE}}) together with refractive indices (nn) above the AE. To evaluate the functionality of these materials for qBICs in DUV, we calculated transmittance spectra of a series of hypothetical lossless double-rod metasurfaces that differed only by the nanorods’ refractive index. The resulting transmittance map in Figure 1f shows that qBIC, for the considered geometry, appears only when the refractive index is higher than 2. As it further increases, the qBIC redshifts from the diffraction region, defined as λdiff=nsubstp\lambda_{\mathrm{diff}} = n_{\mathrm{subst}}p, where nsubstn_{\mathrm{subst}} is the refractive index of the substrate [34]. This redshift is highly advantageous, as operating in the wavelength region below λdiff\lambda_{\mathrm{diff}} would cause the resonance to leak into diffraction channels propagating into the substrate, severely degrading the resonance QQ-factor. For n>2.4n > 2.4, a secondary magnetic resonance (MR) becomes prominent and also redshifts with increasing nn. Because the linewidth of this mode is broader compared to the qBIC, it falls outside the primary focus of this work and is not investigated further. The transmittance map also includes markers identifying the selected materials based on their λAE\lambda_{\mathrm{AE}} (horizontal position) and nn (vertical position). A shorter λAE\lambda_{\mathrm{AE}} favours DUV transparency, whereas a higher nn increases the spectral separation between the qBIC and the diffraction threshold. However, the redshift associated with the higher nn inadvertently pushes the resonances out of the DUV towards the NUV range. Among the presented materials, HfO2\mathrm{HfO_2} offers an advantageous combination of a relatively high refractive index and low absorption in DUV. It can thus sustain qBICs above the diffraction threshold in the presence of the glass substrate, while for lower refractive index materials the suspension in air is required [34]. At the same time, HfO2\mathrm{HfO_2} is compatible with an established damascene fabrication process [27]. This suitability is confirmed by our experimental realization of the corresponding metasurface shown in Figure 1a.

A fundamental requirement for a versatile UV label-free sensing platform is the ability to align the spectral position of the qBIC with the intrinsic absorption bands of target analytes. To achieve this, we leveraged a scaling factor SS, which scales all lateral dimensions within the unit cell while keeping the asymmetry factor AA constant. Figure 2a presents the simulated evolution of the qBIC as a function of the scaling factor SS. By systematically varying SS from 0.80 to 1.30 in our simulations, we show that our qBICs can be spectrally shifted from 225 nm to 330 nm. Additionally, the simulated resonances generally become narrower with increasing SS. Since the nanorod height remains constant, the resulting change in the aspect ratio, together with the wavelength-dependent refractive indices of the materials, contribute to this behaviour. Overall, this broad spectral tunability enables the qBIC to be matched to the intrinsic UV absorption features of various biomolecules. For example, the four fundamental DNA bases that exhibit strong absorption in the DUV spectral range, as shown in Figure 2b [35].

Spectral tailoring of the qBIC across the DUV and NUV spectral ranges

Figure 2. Spectral tailoring of the qBIC across the DUV and NUV spectral ranges. (a) Simulated transmittance spectra demonstrating broad spectral range of the qBIC by varying the unit cell scaling factor SS from 0.80 to 1.30 for a constant asymmetry factor A=0.86A = 0.86. (b) Normalized absorbance spectra of the four fundamental DNA bases (adenine, guanine, cytosine and thymine) obtained from [35], highlighting their strong intrinsic absorbance in the DUV and NUV wavelength range. Spectra are vertically offset for clarity. (c) Measured transmittance spectra illustrating broad spectral range achieved by varying the scaling factor SS for a fixed asymmetry factor A=0.86A = 0.86. The resonance shifts across from DUV to NUV spectral ranges (from ≈225\approx225 nm to 340 nm). Insets show top-view SEM images of the fabricated nanostructures. (d) Extracted total (QtotQ_{\mathrm{tot}}, pink) together with radiative (QradQ_{\mathrm{rad}}, filled gray circles) and intrinsic (QintQ_{\mathrm{int}}, open gray circles) QQ-factors as a function of the resonant wavelength.

To experimentally validate the spectral tunability predicted by our simulations, we fabricated a HfO2\mathrm{HfO_2} metasurface with systematically varying scaling factors SS. The metasurface was fabricated on a fused silica substrate following a bottom-up damascene fabrication process commonly used for the fabrication of high-aspect ratio nanostructures [27, 31, 36]. It employs electron-beam lithography followed by atomic layer deposition, with subsequent ion-beam etching and final resist removal. The optical characterization was performed using a custom-built transmittance setup consisting of a polarized xenon light source collimated onto the sample, followed by an objective collecting the light into the spectrometer (see Methods for more details on fabrication and characterization). Figure 2c shows the measured transmittance spectra as a function of the scaling factor SS. Mirroring the simulation trends, systematically scaling the metasurface unit cell dimensions shifts the qBIC across the DUV and NUV wavelength range, specifically from 225 nm to 340 nm. The inset scanning electron microscope (SEM) images show the top-view of the fabricated HfO2\mathrm{HfO_2} nanostructures, confirming the precise geometrical control across the different scaling factors. We quantified the

resonances using the temporal coupled-mode theory [37] (see Methods): the total QQ-factor (QtotQ_{\mathrm{tot}}) can be decomposed into the radiative contribution (QradQ_{\mathrm{rad}}) and non-radiative contribution (QintQ_{\mathrm{int}}) by the expression Qtot−1=Qrad−1+Qint−1Q_{\mathrm{tot}}^{-1} = Q_{\mathrm{rad}}^{-1} + Q_{\mathrm{int}}^{-1}. In this formula, QradQ_{\mathrm{rad}} represents the leakage of the qBIC into free space, while QintQ_{\mathrm{int}} embodies material absorption and other losses associated with scattering and fabrication imperfections. Figure 2d shows the values of QtotQ_{\mathrm{tot}}, QradQ_{\mathrm{rad}} and QintQ_{\mathrm{int}} as a function of the resonant wavelength extracted from Figure 2c. As the scaling factor increases, we observe an upward trend in QtotQ_{\mathrm{tot}} and QradQ_{\mathrm{rad}}, consistent with the resonance narrowing already observed in the simulations. The variations in the QQ-factors can be partially attributed to fabrication imperfections and associated parasitic losses. Based on the SEM images and the characteristics of our nanofabrication process, we expect an approximately constant absolute magnitude of structural deviations, which has a significantly larger influence on the smallest nanostructures. Consequently, structures of smaller scaling factors can suffer from higher parasitic scattering, which degrades their QQ-factors. Nevertheless, our platform still achieves total QQ-factors on the order of hundreds across the DUV and NUV range, with the highest one reaching 500. This confirms that high-QQ resonances can be successfully sustained even at wavelengths below 350 nm, where material absorption and fabrication imperfections can no longer be neglected.

In addition to spectral tailoring through the scaling factor SS, the asymmetry factor AA governs the radiative coupling of the qBIC and consequently affects its linewidth. Figure 3a presents simulated transmittance spectra for varying asymmetry factors AA at a constant scaling factor S=1.25S = 1.25. Increasing AA from 0.80 to 0.96 progressively narrows the qBIC linewidth. This is a direct consequence of the symmetry-protected nature of the mode: as the asymmetry factor approaches unity, the resonance becomes a symmetry-protected BIC and completely decouples from the radiation continuum. To experimentally evaluate the control over the resonance linewidth, Figure 3b presents measured transmittance spectra for varying asymmetry factors AA, ranging from 0.70 to 1.00, while keeping the scaling factor S=1.25S = 1.25 constant. As AA approaches unity, the qBIC indeed systematically narrows and redshifts, in agreement with our simulations. The inset SEM images show the fabricated HfO2_2 nanostructures for three different asymmetry factors, illustrating the gradual elongation of the shorter nanorod. Figure 3c quantifies the narrowing of the resonances through the extracted QtotQ_{\mathrm{tot}}, which increases substantially towards higher AA values. The corresponding QradQ_{\mathrm{rad}} increases even more pronouncedly, confirming the suppression of radiative losses as the system approaches the symmetry-protected BIC limit. In contrast, QintQ_{\mathrm{int}} exhibits a flat dependence on the asymmetry factor AA, especially when approaching the BIC limit. Together, these trends confirm that our HfO2_2 metasurface retains precise control over the resonance linewidth and robust performance under realistic fabrication constraints.

Linewidth tailoring of qBICs in HfO$_2$ metasurfaces through asymmetry factor

Figure 3. Linewidth tailoring of qBICs in HfO2_2 metasurfaces through asymmetry factor. (a) Simulated transmittance spectra demonstrating control over the resonance linewidth via altering the asymmetry factor AA with a constant scaling factor S=1.25S = 1.25. As AA approaches unity, the radiative losses are suppressed, leading to narrowing of the resonance as it nears the symmetry-protected BIC limit. (b) Measured transmittance spectra illustrating how the resonance linewidth can be tailored by varying the asymmetry factor AA from 0.70 to 1.00. Insets show top-view SEM images of the gradual elongation of the nanorod. (c) Extracted total (QtotQ_{\mathrm{tot}}, pink), radiative (QradQ_{\mathrm{rad}}, filled gray circles), and intrinsic (QintQ_{\mathrm{int}}, open gray circles) QQ-factors as a function of the asymmetry factor AA.

Having demonstrated control over both the spectral position and linewidth of the qBIC, we next investigated the performance of our HfO2_2 metasurface in aqueous environments, an essential step towards its application in UV biosensing. For this purpose, we fabricated a new HfO2_2 metasurface chip optimized for integration within a polydimethylsiloxane (PDMS) microfluidic cell with inlet/outlet tubing and a microfluidic channel (see Figure 4a and Methods). Because the PDMS exhibits strong absorption across the entire UV range, we measured in reflection mode, referencing the spectra to the bare fused silica substrate. Figure 4b and Figure 4c present the measured relative reflectance spectra as a function of the scaling factor SS when the channel is exposed to air (nair≈1n_{\mathrm{air}} \approx1) and to an aqueous buffer (naq(λ)n_{\mathrm{aq}}(\lambda)), respectively. Although the resonances remain sharp and well-defined upon the introduction of the buffer, they are noticeably shifted towards longer wavelengths. To further quantify these effects, the resonance shift with respect to that in air (Δλ\Delta\lambda) was extracted as a function of the scaling factor SS and is shown in Figure 4d. The wavelength shift Δλ\Delta\lambda generally increases from 9 nm to 20 nm as the qBIC shifts towards NUV wavelengths. Figure 4e compares QtotQ_{\mathrm{tot}}, QradQ_{\mathrm{rad}}, and QintQ_{\mathrm{int}} in air and in buffer. Because this metasurface was produced in a different fabrication run, its absolute QQ-factors cannot be directly compared to the values observed in Figure 2d. The relevant metric is the relative performance change between

Microfluidic integration and bulk refractive index sensing performance of the HfO2 metasurface across the DUV and NUV spectral range

Figure 4. Microfluidic integration and bulk refractive index sensing performance of the HfO2_2 metasurface across the DUV and NUV spectral range. (a) Schematic illustration and a photograph of the metasurface chip integrated with a polydimethylsiloxane (PDMS) microfluidic cell, enabling controlled fluid delivery through inlet/outlet tubing. (b,c) Measured relative reflectance spectra as a function of the scaling factor SS when the microfluidic channel is filled with (b) air (nair≈1n_{\mathrm{air}} \approx1) and (c) an aqueous buffer evaluated using its wavelength-dependent refractive index naq(λ)n_{\mathrm{aq}}(\lambda) [38]. Replacing air with a higher-index aqueous buffer induces a resonance wavelength shift (Δλ\Delta\lambda) towards longer wavelengths. (d) Extracted resonance wavelength shift Δλ\Delta\lambda as a function of the scaling factor SS. (e) Extracted total QQ-factor (QtotQ_{\mathrm{tot}}) and its radiative (QradQ_{\mathrm{rad}}, filled circles) and intrinsic (QintQ_{\mathrm{int}}, open circles) contributions, as a function of the scaling factor SS. Pink and black symbols represent measurements in buffer and air, respectively. (f) Bulk refractive index sensitivity (σ\sigma, pink) and the corresponding figure of merit (FOM, black) across the NUV and DUV spectral ranges.

the air and buffer environments. Interestingly, the QtotQ_{\mathrm{tot}} is consistently enhanced in the aqueous buffer. The extracted QintQ_{\mathrm{int}} generally increases in the buffer, whereas QradQ_{\mathrm{rad}} exhibits a less systematic dependence on the surrounding medium. This suggests a reduction in intrinsic losses upon introducing the buffer, potentially associated with the lower refractive index contrast (nHfO2/naqn_{\mathrm{HfO_2}}/n_{\mathrm{aq}}) and the resulting suppression of parasitic scattering [39]. Quantifying the individual contributions of the different loss mechanisms, however, would require further investigation beyond the scope of this work. Ultimately, these results highlight the mechanical and optical robustness of our HfO2_2 metasurface when operating in aqueous environments.

Finally, to evaluate the refractive index sensing performance of our integrated chip, Figure 4f shows the bulk refractive index sensitivity (σ\sigma) and the corresponding figure of merit (FOM) across the operating wavelength range. The sensitivity was estimated from the resonance wavelength shift between air and aqueous buffer as σ=Δλ/Δnaq\sigma= \Delta\lambda/\Delta n_{\mathrm{aq}}, using the wavelength-dependent refractive index of water [38] to approximate that of the buffer. While σ\sigma quantifies the spectral shift [40] per refractive index unit, the FOM =σ/FWHM= \sigma/\mathrm{FWHM} is the critical metric, as it normalizes the shift against the resonance full width at half maximum (FWHM) in the buffer [41]. Our metasurface achieves a sensitivity of up to 60 nm/RIU. While absolute sensitivity naturally scales down at shorter operating wavelengths, the narrow linewidths of the qBICs yield a FOM exceeding 40 RIU−1^{-1}. For comparison, a recently investigated UV plasmonic platform based on rhodium nanocavities on aluminum mirrors exhibited a calculated FOM below 20 [42]. Experimental refractive index sensing in the UV has also been demonstrated using all-dielectric Si3_3N4_4 metasurfaces with a reported FOM of 55 RIU−1^{-1} [43]. However, the reported resonance wavelength shift appears inconsistent with the presented experimental data, which suggest a recalculated FOM of approximately 30 RIU−1^{-1}. This comparison indicates that our experimentally realized HfO2_2 metasurface, achieving a FOM exceeding 40 RIU−1^{-1}, offers competitive sensing performance. Although high-QQ all-dielectric sensing platforms have been investigated in the VIS and IR regimes [44, 45, 46], their experimental realization in the DUV remains comparatively limited. The broad spectral tunability of our metasurface, together with its demonstrated refractive index sensing performance in an aqueous microfluidic environment, highlights its potential for future label-free UV biosensing.

In summary, we have experimentally demonstrated an all-dielectric HfO2_2 metasurface supporting high-QQ qBICs across the DUV and NUV spectral ranges. By leveraging a symmetry-broken double-nanorod unit cell, we achieved broad spectral tailoring from 225 nm to 340 nm with experimental QQ-factors reaching up to 500. Furthermore, we integrated the metasurface with a PDMS microfluidic cell and demonstrated its refractive index sensing capabilities in aqueous environments in reflection mode. Introducing an aqueous buffer led to an enhanced QQ-factor, potentially through the reduction of parasitic scattering losses associated with a lower refractive index contrast. The integrated metasurface chip exhibited a bulk refractive index sensitivity of up to 60 nm/RIU with a FOM exceeding 40 RIU−1^{-1}. These results establish HfO2_2 metasurfaces supporting qBICs as a promising platform, paving the way towards label-free DUV sensing and the direct optical probing of UV electronic transitions of biomolecules in their native spectral range.

Methods

Simulations

The transmittance spectra and electric field enhancements were obtained using finite-difference time-domain calculations in ANSYS Lumerical software. The transmittance was calculated with periodic boundary conditions along the yy-axis, anti-symmetric conditions along the xx-axis, and perfectly matched layers (PMLs) along the zz-axis. The unit cell was illuminated with a linearly polarized plane wave propagating along the zz-axis, with the electric field polarized along the xx-axis. The electric field enhancement profiles were recorded using 2D frequency-domain field monitors. The first monitor was positioned 35 nm above the substrate to record the xx-yy cross-section through the center of the nanorods. The second was placed at the center of the shorter L2L_2 nanorod to monitor its xx-zz cross-section. The dielectric function of HfO2_2 was measured and fitted using an ellipsometer (J. A. Woollam V-VASE). For the substrate, we used SiO2_2 data from the standard Lumerical database (Palik’s handbook [47]).

Fabrication

Fused silica wafers of 300 μ\mum thickness were used as substrates. The metasurfaces were prepared for electron beam lithography (EBL) by applying hexamethyldisilazane (HMDS) as a gaseous adhesion promoter, followed by spin-coating a positive tone resist (AR-P 6200.04) at 3200 rpm with an acceleration of 2000 rpm/s for 60 s, yielding a resist layer of approximately 130 nm (as measured by a reflectometer). To prevent charging effects, a conductive layer (AR-PC 5090.02) was spin-coated on top. EBL was performed using a RAITH 150 TWO system with an acceleration voltage of 30 kV and 15 μ\mum aperture. The prepared mask was subsequently filled with HfO2_2 by atomic layer deposition (Beneq), using tetrakis-dimethylamino hafnium (TDMAH) as a precursor. The top layer of HfO2_2 was sputtered off by argon-ion beam etching (SCIA). Finally, the remaining resist was removed in the microwave oxygen plasma cleaner (Diener). The fabricated samples were imaged using a scanning electron microscope (Thermo Fisher Verios). The height was estimated to be 82 nm using AFM (Bruker Dimension Icon).

Microfluidics

The polydimethylsiloxane (PDMS) microfluidic cell was fabricated by replica molding from a previously prepared silicon master mold. The central microfluidic reservoir had lateral dimensions of 7.5 ×\times 7.5 mm2^2 and a depth of 170 μ\mum. The fluid distribution was provided by tubing with a diameter of 1.5 mm, and the aqueous buffer was delivered using a syringe. To prevent any leaks, the interfaces between the metasurface chip, the PDMS cell, and the tubing were all sealed with a two-component silicone adhesive.

Optical characterization

Optical characterization was performed using a custom-built UV microscopy setup. The samples were illuminated by a polarized xenon light source (Hamamatsu L2479). The metasurface sample was systematically aligned such that this dominant polarization component was oriented along the xx-axis of the nanorods, optimizing the coupling conditions for the qBIC mode. The beam was collimated using a 300 mm lens (Thorlabs LA4855-UV), and after transmission through the metasurface it was collected by a 40×\times UV objective with a numerical aperture NA = 0.49 (Thorlabs LMU-40X-UVB), followed by a 200 mm tube lens (Thorlabs LA4102-UV). For the spectral analysis, we used a spectrometer (Andor Shamrock SR303i) equipped with a CCD camera (Andor iDus). In addition to the primary spectroscopic path, an imaging path was created for precise sample navigation, utilizing a flip-mounted mirror (Thorlabs PF20-03-G01) and a CMOS camera (Thorlabs CS165MU). For reflection-mode measurements, the beam path was modified by integrating a UV fused silica beam splitter (Thorlabs BSW20). To account for the low signal intensities characteristic of the UV regime, both the transmittance and reflectance spectra were normalized to the bare fused silica substrate, yielding the relative transmittance and reflectance values presented in our analysis, respectively.

QQ-factor evaluation

The QQ-factors were extracted by fitting the measured spectra using temporal coupled-mode theory (TCMT) [37]. The transmittance spectra of the qBIC were fitted with the following expression:

T=∣eiΦt0+γradγrad+γint+i(ω−ωres)∣2T=\left|e^{i\Phi}t_{0}+\frac{\gamma_{\mathrm{rad}}}{\gamma_{\mathrm{rad}}+\gamma_{\mathrm{int}}+i(\omega-\omega_{\mathrm{res}})}\right|^{2}

Here, eiΦt0e^{i\Phi}t_{0} is the background transmission, γ=γrad+γint\gamma=\gamma_{\mathrm{rad}}+\gamma_{\mathrm{int}} is the total loss rate, where γrad\gamma_{\mathrm{rad}} and γint\gamma_{\mathrm{int}} correspond to the radiative and intrinsic loss rates, respectively. Finally, ωres\omega_{\mathrm{res}} is the resonant frequency. The total QQ-factor was then calculated as Qtot=ωres2γQ_{\mathrm{tot}}=\frac{\omega_{\mathrm{res}}}{2\gamma}. Accordingly, the radiative and intrinsic QQ-factors were extracted as Qrad=ωres2γradQ_{\mathrm{rad}}=\frac{\omega_{\mathrm{res}}}{2\gamma_{\mathrm{rad}}} and Qint=ωres2γintQ_{\mathrm{int}}=\frac{\omega_{\mathrm{res}}}{2\gamma_{\mathrm{int}}}. For the reflectance spectra, an identical model was applied, replacing the background transmission term (eiΦt0e^{i\Phi}t_{0}) with the background reflection term (eiΦr0e^{i\Phi}r_{0}).

Refractive index sensitivity calculation

For the estimation of bulk refractive index sensitivity σ=Δλ/Δnaq\sigma=\Delta\lambda/\Delta n_{\mathrm{aq}}, the wavelength-dependent refractive index of water naq(λ)n_{\mathrm{aq}}(\lambda) was obtained from the literature [38] to approximate that of the buffer.

Acknowledgements

This work was supported by the Joint Czech-Bavarian Research Projects 2024–2026 program via project BTHA-JC-2024-49 (Bavarian-Czech Academic Agency - BTHA), project LUABA24069 (Ministry of Education, Youth and Sports of the Czech Republic). The CzechNanoLab Research Infrastructure (ID 90251), funded by MEYS CR, is gratefully acknowledged for the financial support of the measurements/sample fabrication. This project was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany’s Excellence Strategy – EXC 2089/2 – 390776260 (e-conversion 2.0) and EXC 3120/1 – 533771286 (BlueMat), and through the Emmy Noether Programme (TI 1063/1). It was also funded by the European Union (ERC, METANEXT, 101078018 and EIC, OMICSENS, 101129734). Views and opinions expressed are, however, those of the authors only and do not necessarily reflect those of the European Union, the European Research Council Executive Agency, or the European Innovation Council and SMEs Executive Agency (EISMEA). Neither the European Union nor the granting authority can be held responsible for them. The authors acknowledge support from the Center for NanoScience (CeNS) at LMU. L. de S. M. thanks for the support from the Center for Nanoscience (CeNS), Faculty of Physics, Ludwig-Maximilians-University Munich (LMU). B. I. thanks for the support from the Bavarian-Czech Academic Agency (BTHA) through mobility grant BTHA-MOB-2026-4 and from the Brno University of Technology through a specific research project CEITEC VUT-J-24-8581.

Graphical schematic showing patterned arrays, an aqueous buffer, and spectral responses

Figure 1 (PDF p. 14).

References

  1. [1]Kuznetsov, A. I. et al. ACS Photonics 2024, 11, 816–865.
  2. [2]Zeng, Y.; Zhong, H.; Long, Z.; Cao, H.; Jin, X. npj Nanophotonics 2025, 2, 39.
  3. [3]Balthasar Mueller, J. P.; Rubin, N. A.; Devlin, R. C.; Groëver, B.; Capasso, F. Physical Review Letters 2017, 118, 113901.
  4. [4]Chen, S.; Li, Z.; Zhang, Y.; Cheng, H.; Tian, J. Advanced Optical Materials 2018, 6, 1800104.
  5. [5]Chen, S.; Li, Z.; Liu, W.; Cheng, H.; Tian, J. Advanced Materials 2019, 31, 1802458.
  6. [6]Zhao, D.; Lin, Z.; Zhu, W.; Lezec, H. J.; Xu, T.; Agrawal, A.; Zhang, C.; Huang, K. Nanophotonics 2021, 10, 2283–2308.
  7. [7]Khorasaninejad, M.; Chen, W. T.; Devlin, R. C.; Oh, J.; Zhu, A. Y.; Capasso, F. Science 2016, 352, 1190–1194.
  8. [8]Arbabi, A.; Horie, Y.; Bagheri, M.; Faraon, A. Nature Nanotechnology 2015, 10, 937–943.
  9. [9]Koshelev, K.; Lepeshov, S.; Liu, M.; Bogdanov, A.; Kivshar, Y. Physical Review Letters 2018, 121, 193903.
  10. [10]Gorkunov, M. V.; Antonov, A. A.; Kivshar, Y. S. Physical Review Letters 2020, 125, 093903.
  11. [11]Shi, T.; Deng, Z.-L.; Geng, G.; Zeng, X.; Zeng, Y.; Hu, G.; Overvig, A.; Li, J.; Qiu, C.-W.; Alù, A.; Kivshar, Y. S.; Li, X. Nature Communications 2022, 13, 4111.
  12. [12]Kühner, L.; Sortino, L.; Tilmann, B.; Weber, T.; Watanabe, K.; Taniguchi, T.; Maier, S. A.; Tittl, A. Advanced Materials 2023, 35, DOI: 10.1002/adma.202209688.DOI
  13. [13]Kang, M.; Liu, T.; Chan, C. T.; Xiao, M. Nature Reviews Physics 2023, 5, 659–678.
  14. [14]Aigner, A.; Ligmajer, F.; Rovenská, K.; Holobrádek, J.; Idesová, B.; Maier, S. A.; Tittl, A.; De S. Menezes, L. Nano Letters 2024, 24, 10742–10749.
  15. [15]Hsu, C. W.; Zhen, B.; Stone, A. D.; Joannopoulos, J. D.; Soljačić, M. Nature Reviews Materials 2016, 1, 16048.
  16. [16]Zhang, C.; Nie, X.; Liu, Z.; Chen, Z.; Ma, C.; Yue, F.; Wang, L.; He, Z. Optics Communications 2023, 530, 129173.
  17. [17]Shen, S. J.; Lee, B.-R.; Peng, Y. C.; Wang, Y. J.; Huang, Y.-W.; Kivshar, Y.; Tseng, M. L. ACS Photonics 2025, 12, 2955–2964.
  18. [18]Koshelev, K.; Kruk, S.; Melik-Gaykazyan, E.; Choi, J.-H.; Bogdanov, A.; Park, H.-G.; Kivshar, Y. Science 2020, 367, 288–292.
  19. [19]Bourrellier, R.; Meuret, S.; Tararan, A.; Stéphan, O.; Kociak, M.; Tizei, L. H. G.; Zobelli, A. Nano Letters 2016, 16, 4317–4321.
  20. [20]Khan, S. A.; Khan, N. Z.; Xie, Y.; Abbas, M. T.; Rauf, M.; Mehmood, I.; Runowski, M.; Agathopoulos, S.; Zhu, J. Advanced Optical Materials 2022, 10, 2200500.
  21. [21]Lv, J.; Yao, Y.; Guo, Z.; Chen, M.; Gu, Q.; Fang, Y.; Zhang, Y.; Zhao, Y.; Ou, Q. Advanced Materials 2026, 38, e73730.
  22. [22]Jamme, F.; Kascakova, S.; Villette, S.; Allouche, F.; Pallu, S.; Rouam, V.; Réfrég iers, M. Biology of the Cell 2013, 105, 277–288.
  23. [23]Kumamoto, Y.; Taguchi, A.; Kawata, S. Advanced Optical Materials 2019, 7, 1801099.
  24. [24]Franta, D.; Dubroka, A.; Wang, C.; Giglia, A.; Vohánka, J.; Franta, P.; Ohlídal, I. Applied Surface Science 2017, 421, 405–419.
  25. [25]Franta, D.; Nečas, D.; Ohlídal, I.; Giglia, A. In SPIE Optical Systems Design, ed. by Duparré, A.; Geyl, R., Jena, Germany, 2015, 96281U.
  26. [26]Baker, R. M.; Permyakov, D.; Yu, Y. F.; Markovich, D.; Paniagua-Domínguez, R.; Gonzaga, L.; Samusev, A.; Kivshar, Y.; Luk'yanchuk, B.; Kuznetsov, A. I. Nano Letters 2015, 15, 2137–2142.
  27. [27]Zhang, C.; Divitt, S.; Fan, Q.; Zhu, W.; Agrawal, A.; Lu, Y.; Xu, T.; Lezec, H. J. Light: Science & Applications 2020, 9, 55.
  28. [28]Peng, Y. C.; Wang, Y. J.; Chen, K.-H.; Lin, Y. H.; Sakurai, H.; Chang, H.-C.; Chiang, C.-C.; Duh, R.-T.; Lee, B.-R.; Huang, C.-Y.; Shih, M.-H.; Horng, R.-H.; Konishi, K.; Tseng, M. L. Nano Letters 2025, 25, 3141–3149.
  29. [29]Kim, J.; Jeon, D.; Seong, J.; Badloe, T.; Jeon, N.; Kim, G.; Kim, J.; Baek, S.; Lee, J.-L.; Rho, J. ACS Nano 2022, 16, 3546–3553.
  30. [30]Lee, E.; Kang, H.; Yun, H.; Rho, J. Microsystems & Nanoengineering 2026, 12, 202.
  31. [31]Huang, K.; Deng, J.; Leong, H. S.; Yap, S. L. K.; Yang, R. B.; Teng, J.; Liu, H. Laser & Photonics Reviews 2019, 13, 1800289.
  32. [32]Hu, J.; Lawrence, M.; Dionne, J. A. ACS Photonics 2020, 7, 36–42.
  33. [33]Yang, X.; Wu, B.; Chen, P.-P.; Xie, Y.-Y.; Lin, C.-N.; Zhang, Z.-F.; Dou, W.-J.; Ni, P.-N.; Shan, C.-X. Laser & Photonics Reviews 2024, 18, 2400007.
  34. [34]Hirler, M.; Antonov, A. A.; Baù, E.; Aigner, A.; Heimig, C.; Hu, H.; Tittl, A. ACS Nano 2026, 20, 6722–6731.
  35. [35]Taniguchi, M.; Lindsey, J. S. Photochemistry and Photobiology 2018, 94, 290–327.
  36. [36]Devlin, R. C.; Khorasaninejad, M.; Chen, W. T.; Oh, J.; Capasso, F. Proceedings of the National Academy of Sciences 2016, 113, 10473–10478.
  37. [37]Fan, S.; Suh, W.; Joannopoulos, J. D. Journal of the Optical Society of America A 2003, 20, 569.
  38. [38]Hale, G. M.; Querry, M. R. Applied Optics 1973, 12, 555.
  39. [39]Kravets, V. G.; Kabashin, A. V.; Barnes, W. L.; Grigorenko, A. N. Chemical Reviews 2018, 118, 5912–5951.
  40. [40]White, I. M.; Fan, X. Optics Express 2008, 16, 1020.
  41. [41]Mayer, K. M.; Hafner, J. H. Chemical Reviews 2011, 111, 3828–3857.
  42. [42]Xie, Y.; Wu, Q.; Zhou, C.; Zhong, Y.; Fei, S.; Chen, J.; Guo, T.; Du, W.; Yan, Z. Optics Express 2025, 33, 30648.
  43. [43]Zheng, J.; Zhang, C.; Li, H.; Liu, X.; Huang, Y.; Zhu, J.; Yang, Z.; Li, L. Physical Chemistry Chemical Physics 2023, 25, 20026–20031.
  44. [44]Yang, Y.; Kravchenko, I. I.; Briggs, D. P.; Valentine, J. Nature Communications 2014, 5, 5753.
  45. [45]Krasnok, A.; Caldarola, M.; Bonod, N.; Alù, A. Advanced Optical Materials 2018, 6, 1701094.
  46. [46]Yang, X.; Antonov, A.; Hu, H.; Tittl, A. Nanophotonics 2025, 14, 5311–5321.
  47. [47]Palik, E. D.; Ghosh, G., Handbook of optical constants of solids; Academic Press: San Diego, 1998.

Paper details

vlnc:W_zye9cz5t
Published
Fresh — added to the Hub within the last 30 days
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