AbstractDeep learning has become the dominant approach in artificial intelligence to solve complex data-driven problems. Originally applied almost exclusively in computer-science areas such as image analysis and nature language processing, deep learning has rapidly entered a wide variety of scientific fields including physics, chemistry and material science. Very recently, deep neural networks have been introduced in the field of nanophotonics as a powerful way of obtaining the nonlinear mapping between the topology and composition of arbitrary nanophotonic structures and their associated functional properties. In this paper, we have discussed the recent progress in the application of deep learning to the inverse design of nanophotonic devices, mainly focusing on the three existing learning paradigms of supervised-, unsupervised-, and reinforcement learning. Deep learning forward modelling i.e. how artificial intelligence learns how to solve Maxwell’s equations, is also discussed, along with an outlook of this rapidly evolving research area.

Recent introduction of data-driven approaches based on deep-learning technology has revolutionized the field of nanophotonics by allowing efficient inverse design methods. In this paper, simultaneous inverse design of materials and structure parameters of core-shell nanoparticle is achieved for the first time using deep-learning of a neural network. A neural network to learn correlation between extinction spectra of electric and magnetic dipoles and core-shell nanoparticle designs, which include material information and shell thicknesses, is developed and trained. We demonstrate deep-learning-assisted inverse design of core-shell nanoparticle for 1) spectral tuning electric dipole resonances, 2) finding spectrally isolated pure magnetic dipole resonances, and 3) finding spectrally overlapped electric dipole and magnetic dipole resonances. Our finding paves the way of the rapid development of nanophotonics by allowing a practical utilization of a deep-learning technology for nanophotonic inverse design.

Data-driven design approaches based on deep-learning have been introduced in nanophotonics to reduce time-consuming iterative simulations which have been a major challenge. Here, we report the first use of conditional deep convolutional generative adversarial networks to design nanophotonic antennae that are not constrained to a predefined shape. For given input reflection spectra, the network generates desirable designs in the form of images; this form allows suggestions of new structures that cannot be represented by structural parameters. Simulation results obtained from the generated designs agreed well with the input reflection spectrum. This method opens new avenues towards the development of nanophotonics by providing a fast and convenient approach to design complex nanophotonic structures that have desired optical properties.Keywords Nanophotonics · Inverse design · Conditional deep convolutional generative adversarial network · Deep learning Recently, data-driven design approaches have been proposed to overcome this problem. These approaches use artificial neural networks to design nanophotonic structures [10,11,12,13]. Previous studies first set the shape, such as multilayers [10] or H-antenna [11] of structures to be predicted, then trained NNs to provide output structural parameters that achieve desired optical properties. Once the NNs are trained, they provide the corresponding design parameters without additional iterative simulations. Such attempts have greatly reduced the effort and computational costs of designing nanophotonic structures. So far, these approaches have only been applied to conditions in which

Optical metamaterials have presented an innovative method of manipulating light. Hyperbolic metamaterials have an extremely high anisotropy with a hyperbolic dispersion relation. They are able to support high-k modes and exhibit a high density of states which produce distinctive properties that have been exploited in various applications, such as super-resolution imaging, negative refraction, and enhanced emission control. Here, state-of-the-art hyperbolic metamaterials are reviewed, starting from the fundamental principles to applications of artificially structured hyperbolic media to suggest ways to fuse natural two-dimensional hyperbolic materials. The review concludes by indicating the current challenges and our vision for future applications of hyperbolic metamaterials.

Passive daytime radiative cooling, which is a process that removes excess heat to cold space as an infinite heat sink, is an emerging technology for applications that require thermal control. Among the different structures of radiative coolers, multilayerand photonic-structured radiative coolers that are composed of inorganic layers still need to be simple to fabricate. Herein, we describe the fabrication of a nanoparticle-mixture-based radiative cooler that exhibits highly selective infrared emission and low solar absorption. Al 2 O 3 , SiO 2 , and Si 3 N 4 nanoparticles exhibit intrinsic absorption in parts of the atmospheric transparency window; facile one-step spin coating of a mixture of these nanoparticles generates a surface with selective infrared emission, which can provide a more powerful cooling effect compared to broadband emitters. The nanoparticle-based radiative cooler exhibits an extremely low solar absorption of 4% and a highly selective emissivity of 88.7% within the atmospheric transparency window owing to the synergy of the optical properties of the material. The nanoparticle mixture radiative cooler produces subambient cooling of 2.8 °C for surface cooling and 1.0 °C for space cooling, whereas the Ag film exhibits an above-ambient cooling of 1.1 °C for surface cooling and 3.4 °C for space cooling under direct sunlight.

In nanophotonics, multipole approach has become an indispensable theoretical framework for analyzing subwavelength meta-atoms and their radiation properties. Thus far, induced multipole moments have frequently used to illustrate the radiating properties of the meta-atoms, but they are excited at a specific illumination and do not fully represent anisotropic meta-atoms. On the other hand, dynamic polarizability (α) tensors contain complete scattering information of the metaatoms, but have not often been considered due to complicated retrieval procedures. In this study, we suggest that exact higher-order α-tensor can be efficiently obtained from T-matrix using simple basis transformation. These higher-order α-tensors are necessary to describe recently reported coupled plasmonic and high-refractive-index particles, which we demonstrate from their retrieved α-tensors. Finally, we show that description of meta-atoms using α-tensors incorporated with multiple-scattering theory vastly extends the applicability of the multipole approach in nanophotonics, allowing accurate and efficient depiction of complicated, random, multi-scale systems. Usage: Preprint.

A hyperlens is a super-resolution optical imaging device based on unique hyperbolic dispersions making the subdiffraction-limited information on objects propagate to the far-field. Here, we propose a new device consisting of a 4-inch waferscale spherical hyperlens array that allows high-throughput and easy-to-handle real-time biomolecular imaging. With this proposed device, we report the first experimental demonstration of real-time sub-diffraction-limited biomolecular imaging using a hyperlens. Hippocampal neuron cells are imaged using a hyperlens at a resolution down to 151 nm, much smaller than the diffraction limit of conventional imaging systems in the visible wavelength. These wafer-scale hyperlens devices have great potential for simple, compact, and low-cost integration with conventional optics and therefore a large variety of imaging applications in biology, pathology, medical science and general nanoscience.

Nanofabrication techniques are essential for exploring nanoscience and many closely related research fields such as materials, electronics, optics and photonics. Recently, three-dimensional (3D) nanofabrication techniques have been actively investigated through many different ways, however, it is still challenging to make elaborate and complex 3D nanostructures that many researchers want to realize for further interesting physics studies and device applications. Electron beam lithography, one of the two-dimensional (2D) nanofabrication techniques, is also feasible to realize elaborate 3D nanostructures by stacking each 2D nanostructures. However, alignment errors among the individual 2D nanostructures have been difficult to control due to some practical issues. In this work, we introduce a straightforward approach to drastically increase the overlay accuracy of sub-20 nm based on carefully designed alignmarks and calibrators. Three different types of 3D nanostructures whose designs are motivated from metamaterials and plasmonic structures have been demonstrated to verify the feasibility of the method, and the desired result has been achieved. We believe our work can provide a useful approach for building more advanced and complex 3D nanostructures.