Jiahao Yan

Metasurfaces enable diverse applications by controlling light's amplitude, phase, and polarization. Although deep learning‐based inverse design has revolutionized metasurface design, current models are limited by fixed operating conditions and lack universality, often requiring retraining for new wavelengths, polarizations, or application scenarios. To address this, we introduce MetasurfaceViT (Metasurface Vision Transformer), a generic AI model for inverse design. Our solution leverages a large dataset of Jones matrices, significantly expanded via physics‐informed data augmentation. By pretraining through masking wavelengths and polarization channels, MetasurfaceViT can reconstruct full‐wavelength Jones matrices, which are then used by a fine‐tuning model for inverse design. This versatility allows one‐shot structure design for arbitrary wavelength, polarization, and application requirements. We demonstrate MetasurfaceViT's capabilities in designing multiplexed printings and holograms and broadband achromatic metalenses. Prediction accuracy exceeds 99% for physically realistic designs, showcasing a significant step toward a universal optical inverse design paradigm.

Metasurfaces, sub-wavelength artificial structures, can control light's amplitude, phase, and polar ization, enabling applications in efficient imaging, holograms, and sensing. Recent years, AI has witnessed remarkable progress and spurred scientific discovery. In metasurface design, optical inverse design has recently emerged as a revolutionary approach. It uses deep learning to create a nonlinear mapping between optical structures and functions, bypassing time-consuming traditional design and attaining higher accuracy. Yet, current deep-learning models for optical design face limitations. They often work only for fixed wavelengths and polarizations, and lack universality as input-output vector size changes may require retraining. There's also a lack of compatibility across different application scenarios. This paper introduces MetasurfaceViT, a revolutionary generic AI model. It leverages a large amount of data using Jones matrices and physics-informed data augmentation. By pre-training through masking wavelengths and polarization channels, it can reconstruct full-wavelength Jones matrices, which will be utilized by fine-tuning model to enable inverse design. Finally, a tandem workflow appended by a forward prediction network is introduced to evaluate performance. The versatility of MetasurfaceViT with high prediction accuracy will open a new paradigm for optical inverse design.

To achieve optoelectronic devices with high resolution and efficiency, there is a pressing need for optical structural units that possess an ultrasmall footprint yet exhibit strong controllability in both the frequency and spatial domains. For dielectric nanoparticles, the overlap of electric and magnetic dipole moments can scatter light completely forward or backward, which is called Kerker theory. This effect can expand to any multipoles and any directions, re‐named as generalized Kerker effect, and realize controllable light manipulation at full space and full spectrum using well‐designed dielectric structures. However, the complex situations of multipole couplings make it difficult to achieve structural design. Here, generative artificial intelligence (AI) is utilized to facilitate multi‐objective‐oriented structural design, wherein the study leverages the concept of “combined spectra” that consider both spectra and direction ratios as labels. The proposed generative adversarial network (GAN) is named as DDGAN (double‐discriminator GAN) that discriminates both images and spectral labels. Using trained networks, the simultaneous design for scattering color and directivities, RGB color routers, as well as narrowband light routers is achieved. Notably, all generated structures possess a footprint <600 × 600 nm indicating their potential applications in optoelectronic devices with ultrahigh resolution.

Optical biosensors based on micro/nanofibers are highly valuable for probing and monitoring liquid environments and bioactivity. Most current optical biosensors, however, are still based on glass, semiconductors, or metallic materials, which might not be fully suitable for biologically relevant environments. Here, we introduce biocompatible and flexible microfibers from lotus silk as microenvironmental monitors that exhibit waveguiding of intrinsic fluorescence as well as of coupled light. These features make single-filament monitors excellent building blocks for a variety of sensing functions, including pH probing and detection of bacterial activity. These results pave the way for the development of new and entirely eco-friendly, potentially multiplexed biosensing platforms.

To achieve optoelectronic devices with high resolution and efficiency, there is a pressing need for optical structural units that possess an ultrasmall footprint yet exhibit strong controllability in both the frequency and spatial domains. For dielectric nanoparticles, the overlap of electric and magnetic dipole moments can scatter light completely forward or backward, which is called Kerker theory. This effect can expand to any multipoles and any directions, re-named as generalized Kerker effect, and realize controllable light manipulation at full space and full spectrum using well-designed dielectric structures. However, the complex situations of multipole couplings make it difficult to achieve structural design. Here, generative artificial intelligence (AI) is utilized to facilitate multi-objective-oriented structural design, wherein we leverage the concept of "combined spectra" that consider both spectra and direction ratios as labels. The proposed generative adversarial network (GAN) is named as DDGAN (double-discriminator GAN) which discriminates both images and spectral labels. Using trained networks, we achieve the simultaneous design for scattering color and directivities, RGB color routers, as well as narrowband light routers. Notably, all generated structures possess a footprint less than 600x600 nm indicating their potential applications in optoelectronic devices with ultrahigh resolution.

Silicon nanoparticles (Si NPs) supporting Mie resonances exhibit vivid structural colors on the subwavelength scale. For future wearable devices, next generation Si-based optical units need to be dynamic and stretchable for display, sensing, or signal processing required by human-computer interaction. Here, by utilizing the distance-sensitive electromagnetic coupling of Mie resonances, we maximize the active tuning effect of Si NP-based structures including dimers, oligomers, and NPs on WS₂, which we called Si nanopixels. Through the optical tweezers-assisted printing of Si nanopixels, patterns can be formed on arbitrary flexible substrates. The strain-sensitive tuning of scattering spectra indicates their promising application on strain sensing of various stretchable substrates via a simple "spray and test" process. In the case of Si nanopixels on polydimethylsiloxane (PDMS), local strains around 1% can be detected by a scattering measurement. Moreover, we demonstrate that the scattering intensity variation of Si nanopixels printed on wrinkled tungsten disulfide (WS₂) is pixel-dependent and wavelength-dependent. This property facilitates the application of information encryption, and we demonstrate that three barcodes can be independently encoded into the R, G, and B scattering channels through ternary logic represented by the strain-tuning effects of scattering.

Few‐layer transition metal dichalcogenides (TMDs) and their combination as van der Waals heterostructures provide a promising platform for high‐performance optoelectronic devices. However, the ultrathin thickness of TMD flakes limits efficient light trapping and absorption, which triggers the hybrid construction with optical resonant cavities for enhanced light absorption. The optical structure enriched photodetectors can also be wavelength‐ and polarization‐sensitive but require complicated fabrication. Herein, a new‐type TMD‐based photodetector embedded with nanoslits is proposed to enhance light trapping. Taking ReS₂ as an example, strong anisotropic Mie‐type optical responses arising from the intrinsic in‐plane anisotropy and nanoslit‐enhanced anisotropy are discovered. Owing to the nanoslit‐enhanced optical resonances and band engineering, excellent photodetection performances are demonstrated with high responsivity of 27 A W⁻¹ and short rise/decay times of 3.7/3.7 ms. More importantly, through controlling the angle between the nanoslit orientation and the polarization direction to excite different resonant modes, polarization‐sensitive photodetectors with anisotropy ratios from 5.9 to 12.6 can be achieved, representing one of the most polarization‐sensitive TMD‐based photodetectors. The depth and orientation of nanoslits are demonstrated crucial for optimizing the anisotropy ratio. The findings bring an effective scheme to construct high‐performance and polarization‐sensitive photodetectors.

Controlling excitonic energy transfer in 2D van der Waals (vdW) heterostructures is crucial for photonic and optoelectronic applications. Recent studies suggest that the interlayer energy transfer in vdW heterostructures is strongly correlated with the vertical interlayer spacing. However, the interlayer coupling with large separations (>20 nm) when the radiative energy transfer is dominant has not been studied yet. In this case, excitons as radiative dipole sources are able to control the light field. Here, the thickness dependency of radiative energy transfer in vertical vdW heterostructures of WS₂ (tungsten disulfide)/hBN (hexagonal boron nitride)/WS₂ is studied. The excitonic emission of WS₂/hBN and WS₂/hBN/WS₂ heterostructures is engineered with the intensity ratios of heterostructures to monolayers ranging from 5% to 250%. More importantly, by changing the stacking order to control whether forward or backward emission is collected, a controllable directivity of the excitonic emission from 0.6 to 6.0 is achieved. In theory, the tunability to high‐index/mid‐index interferences and dipole–dipole far‐field coupling is attributed. The outcomes of the study on the radiative energy transfer of vdW heterostructures containing 1Ls (monolayers) with excitonic effects and MLs (multilayers) with high refractive indices will pave the way toward the realization of all vdW nanophotonics.

Van der Waals (vdW) heterostructures based on vertical-stacking transition metal dichalcogenides (TMDCs) with tunable excitonic energies and spin-valley properties show intriguing optical and optoelectronic applications. Additionally, vdW heterostructures with high refractive indices, exciton-induced Lorentzian dispersion, and controllable structures are ideal building blocks as optical resonators for subwavelength light confinement and effective light–matter interaction, which have not been studied. Herein, we build vdW hetero-nanoslits based on tungsten disulfide (WS₂) and hexagonal boron nitride (hBN) multilayers. The multipole optical modes arise from the evolution of electromagnetic near-field distributions through engineering of refractive index and corresponding optical path differences (OPDs). More importantly, the coupling between electromagnetic multipoles with spectral and spatial overlap facilitates the directional scattering with an engineered forward-to-backward (F/B) ratio from 0.1 to 100.0 owing to generalized Kerker effects. Through further combination of WS₂ monolayers and WS₂/hBN hetero-nanoslits, the photoluminescence (PL) modulation in the range of 50% to 800% is achieved. The enhancement factor and modulation range are comparable to the best performances of single-element plasmonic or dielectric nanostructures. This work provides a different insight into designing nanophotonic devices in the visible range by solely relying on vdW heterostructures.

The delivery of optical signals from an external light source to a nanoscale waveguide is highly important for the development of nanophotonic circuits. However, the efficient coupling of external light energy into nanophotonic components is difficult and still remains a challenge. Herein, we use an external silica nanofiber to light up an organic–inorganic hybrid nanowaveguide, namely, a system composed of a polymer filament doped with MoS₂ quantum dots. Nanofiber-excited nanowaveguides in a crossed geometry are found to asymmetrically couple excitation signals along two opposite directions, with different energy dissipation resulting in different colors of the light emitted by MoS₂ quantum dots and collected from the waveguide terminals. Interestingly, rainbow-like light in the hybrid waveguide is achieved by three-in-one mixing of red, green, and blue components. This heterodimensional system of dots in waveguide represents a significant advance toward all-optical routing and full-color display in integrated nanophotonic devices.