All-dielectric metasurface for high-performance structural color

The study addresses the challenge that existing structural color technologies—plasmonic and all-dielectric metasurfaces—rarely achieve, simultaneously, high brightness, narrow spectral linewidth, large color gamut, and diffraction-limited resolution required for practical applications (e.g., displays, security, storage). Plasmonic colors offer subwavelength pixels but suffer from low reflectance and small gamut; dielectric metasurfaces (e.g., TiO2) are brighter but have lower spatial resolution and limited experimental gamut relative to Rec.2020. The authors propose silicon (Si) metasurfaces augmented with a refractive index matching layer to suppress background reflection and narrow spectral bandwidth, aiming to realize an all-in-one high-performance structural color platform.

Prior work demonstrates structural colors via plasmonic nanostructures (gratings, gaps, nanoparticles) achieving vivid colors and subwavelength resolution but with limited brightness and gamut (~45% sRGB). All-dielectric metasurfaces support electric and magnetic Mie resonances for selective reflection/transmission with higher vibrancy and larger gamut; TiO2-based systems enabled full-color printing but with lower spatial resolution (~10^4 dpi) and experimental gamut only 68% of Rec.2020. Hybrid designs could simulate large gamut (~171% sRGB) but experimentally underperform (~124% sRGB) due to fabrication incompatibilities among materials. Silicon metasurfaces are CMOS-compatible, stable, and high-index, enabling compact resonators, but intrinsic Si absorption in visible leads to broadband backgrounds and limited gamut without further engineering. The paper builds on Kerker-type interference between electric and magnetic dipoles to suppress unwanted scattering and improve color quality.

Design and operating principle: Periodic arrays of Si nanodisks on sapphire (Si-on-sapphire) with fixed thickness h = 100 nm, disk radius R, and lattice period l. Colors arise from electric dipole (ED) and magnetic dipole (MD) Mie resonances; adding a refractive index matching layer reduces substrate reflection, brings ED closer to MD, and enables destructive interference outside the main peak to suppress background and narrow FWHM (Kerker effect).

Numerical simulations: Performed with Lumerical FDTD Solutions and COMSOL Multiphysics. Periodic boundary conditions in-plane; perfectly matched layers along propagation. Material indices: single-crystalline Si from software library; sapphire n = 1.76; matching layer modeled with n = 1.48. Multipolar decomposition analyzes ED/MD contributions. Color coordinates computed with custom MATLAB code and plotted on CIE 1931 chromaticity under black-body illumination.

Fabrication: Nanostructures patterned on commercial silicon-on-sapphire wafers via electron-beam lithography (PMMA A2 resist), lift-off to form Cr hard mask, followed by reactive ion etching (Oxford Plasma System 80) to transfer patterns into Si; Cr removed with chromium etchant. Structures integrated into a microfluidic channel for liquid infiltration.

Refractive index matching layer: Dimethyl sulfoxide (DMSO, n = 1.48) infiltrated into the microfluidic channel to serve as index-matching medium. Solid encapsulation demonstrated with PMMA packaging as an alternative.

Optical characterization: Bright-field optical microscopy (ZEISS Axio Scope A1), polarization and incidence controlled via home-built setup. Reflectance spectra measured with a spectrometer through a 50× objective (NA 0.55). Imaging conducted under white-light illumination; angle dependence studied (Supplementary). SEM used for morphological verification; sidewall roughness <10 nm and near-vertical (~90°) sidewalls.

Resolution tests and image printing: Pixels composed of 3×3 or 2×2 nanodisk arrays with center-to-center spacings matched to resonance wavelengths for yellow, green, blue, and purple (320, 250, 200, 190 nm). Test patterns include “Phoenix,” “Rainbow,” and a full-color “Peacock,” with pixel sizes ranging from 9×9 down to 2×2 disks.

• Simulation: Adding an index-matching layer (n = 1.48) reduces white background and narrows FWHM via ED–MD interference (Kerker-like condition). Simulated gamut reaches ~186% sRGB, 138.7% Adobe RGB, and ~99.5% Rec.2020; colors are largely angle-insensitive.
• Experiment (air): Si metasurfaces exhibit reflection peaks at 406, 433, 467–472, 526–529, 576, and 600 nm, producing multiple hues but with paler appearance and broader peaks; gamut ~78% sRGB.
• Experiment (with DMSO): Peaks shift and colors change to more vivid tones (Fandango, electric violet, dark violet, blue, cyan, bright green, yellow, red). Main-peak FWHM ~34–40 nm, about 40% narrower than in air. Background reflection suppressed to ~0. Reflectance at MD resonance maintained across visible (~74% down to ~21% depending on wavelength); specific high reflectance noted at 600 nm (~76%).
• Gamut (experiment): Expanded to ~181.8% sRGB, 135.6% Adobe RGB, and 97.2% Rec.2020, significantly surpassing prior experimental records for all-dielectric/hybrid systems.
• Fabrication quality: Single-crystalline Si process yields smooth surfaces (<10 nm roughness) and vertical sidewalls, ensuring measured reflectance/gamut closely match design; outperforms poly-Si and a-Si due to lower visible absorption (gamut 110% and 238% larger, respectively).
• Resolution: Distinct colors preserved with pixels as small as 2×2 nanodisks; center-to-center spacings of 320, 250, 200, and 190 nm for yellow, green, blue, and purple approach the diffraction limit (100×, NA 0.9). Complex images (“Phoenix,” “Rainbow,” “Peacock”) retain uniformity and color distinction down to diffraction-limited pixelation.
• Dynamic and packaging: Color can be dynamically switched by exchanging ambient (air vs DMSO). Solid-state encapsulation with PMMA reproduces the enhanced color performance, indicating applicability beyond liquids.

The index-matching layer reduces the refractive index contrast at the substrate interface, suppressing substrate reflections that otherwise introduce a white background. Simultaneously, the ED mode, more sensitive to the surrounding index than the MD mode, shifts toward the MD resonance, enabling destructive ED–MD interference outside the main band (generalized Kerker effect). This narrows the spectral FWHM and enhances color purity and brightness without significantly increasing absorption or angular sensitivity. Experimentally, the approach achieves near-Rec.2020 coverage while maintaining high reflectance and diffraction-limited pixelation—addressing the long-standing trade-offs among brightness, gamut, and resolution in structural color. Compared with previous hybrid metasurfaces whose experimental performance lagged simulations due to fabrication incompatibilities, the single-material, CMOS-compatible Si platform yields high-fidelity nanostructures and reproducible optical performance. The colors can be dynamically tuned via environmental refractive index (e.g., liquids) and preserved with solid encapsulation (PMMA), suggesting practical pathways for devices such as displays and security features.

This work introduces an all-dielectric Si metasurface with an index-matching layer that simultaneously delivers high reflectance (~76% at 600 nm), narrow spectral linewidths (~34–40 nm), negligible background reflection, record-large experimental gamut (~181.8% sRGB, 135.6% Adobe RGB, 97.2% Rec.2020), and diffraction-limited spatial resolution (down to 2×2 nanodisk pixels). The CMOS-compatible, single-crystalline Si process ensures manufacturability and durability. The concept is versatile, supporting dynamic coloration via liquids and robust performance with solid encapsulation (PMMA). Future directions include integrating electrically or thermally tunable solid-state index layers (e.g., liquid crystals, phase-change materials), scaling to wafer-level fabrication for large-area displays, optimizing pixel architectures for higher frame-rate dynamic operation, and exploring colorimetric sensing functionalities leveraging the refractive-index sensitivity of the ED mode.