Design Of Photonic Structures Research Articles

Exploring high-performance photonic crystal slow light waveguides through deep reinforcement learning

Supervised deep learning is not really suitable for automatic inverse design of photonic crystal waveguides. There are three reasons for this:(1) Pre-training of the network requires a large number of datasets, which leads to a lot of time spent in the physical simulation process. (2) The one-to-many mapping problem often arises during the reverse design process. (3) It is difficult to realize the inverse design of photonic structures beyond the performance of the dataset. These problems can be well addressed by deep reinforcement learning (DRL). Here, deep reinforcement learning is proposed for the first time for the design of complex slow-light photonic crystal waveguides. The design using DRL took less than 72 h to obtain a photonic crystal waveguide with a normalized delay bandwidth product of 0.449 and group index of 48.72. By increasing the scaling factor of bandwidth in the reward function to 0.25, the bandwidth value of the optimized slow-light waveguide increases to 50.01 nm with an NDBP value of 0.39. The slow-light performance of both structures is verified by FDTD. In addition, the NDBP value is further optimized to 0.46 if the size of the structural parameters is not constrained. Compared with deep learning, DRL can improve the sampling efficiency by more than an order of magnitude and is not plagued by the three problems mentioned above. This work confirms the effectiveness of DRL in the inverse design of energy band engineering. DRL can be widely used in the optimization of energy band related performance.

Temperature dependence analysis of intersubband absorption in asymmetric ZnO/MgZnO quantum wells: Potential for terahertz emission applications

In this paper, in ZnO/MgxZn1−xO quantum well structures, with 20% of Mg, we have calculated the electronic states of confined levels and the intersubband absorption coefficient with or without the effect of non-parabolicity. We use the finite difference approach to do our computations inside the envelope function formalism approximation. The results show that the wavelength range covered by the E12 transition (9–16 μm) is mid-infrared to far-infrared. According to our findings, ZnO/MgZnO quantum well structures, which have extremely similar lattice characteristics and the benefit of an unconstrained structure, have E23 transition energies that provide terahertz frequencies (1.1—20THz) that are utilized in airport security and explosives detection. This promotes intersubband photonic structure design and epitaxial development. Furthermore, we have examined the impact of temperature on the Fermi level and the load carrier population inside each level for these remarkable structures. In addition, each transition's oscillation force and intersubband absorption are considered.

Multitask Learning Deep Neural Networks Enable Embedded Design of Active Metamaterials.

In this study, we propose and implement a deep neural network framework based on multitask learning aimed at simplifying the forward modeling and inverse design process of photonic devices integrating active metasurfaces. We demonstrate and validate our approach by constructing a continuously tunable bandpass filter that is effective in the midwave infrared region. The key to this filter is the combination of a metasurface and Fabry-Perot (F-P) cavity structure of the tunable phase-change material Ge2Sb2Se4Te (GSST) and the precise control of the crystallinity of the GSST by a silicon-based heater. With the help of a deep learning framework, we are able to independently model the crystallinity and geometric parameters of the filter to maximize the use of GSST tuning for bandpass filtering. Our model discusses the self-attention mechanism and the effect of noise and compares several existing popular algorithms, and the results show that a multitask deep learning strategy can better assist the on-demand reverse design of photonic structures with phase change materials. This opens up new possibilities for personalization and functional extension of optical devices.

Adjoint method in machine learning: A pathway to efficient inverse design of photonic devices

Innovative machine learning techniques have facilitated the inverse design of photonic structures for numerous practical applications. Nevertheless, the quantity of data and the initial data distribution are paramount for the discovery of highly efficient photonic devices. These devices often require simulated data ranging from thousands to several hundred thousand data points. This issue has consistently posed a major hurdle in machine learning-based photonic design problems. Therefore, we propose a new data augmentation algorithm grounded in the adjoint method, capable of generating more than 300 times the amount of original data while enhancing device efficiency. The adjoint method forecasts changes in the figure of merit (FoM) resulting from structural perturbations, requiring only two full-wave Maxwell simulations for this prediction. By leveraging the adjoint gradient values, we can augment and label several thousand new data points without any additional computations. Furthermore, the augmented data generated by the proposed algorithm displays significantly improved FoMs. We apply this algorithm to a multi-layered metalens design problem and demonstrate that it consequently exhibits a 343-fold increase in data generation efficiency. After incorporating the proposed algorithm into a generative adversarial network, the optimized metalens exhibits a maximum focusing efficiency of 92.93%, comparable to the theoretical upper bound.

Radiative cooling textiles using industry-standard particle-free nonporous micro-structured fibers

Abstract Thermal radiation is a major dissipative pathway for heat generated by the human body and offers a significant thermoregulation mechanism over a wide range of conditions. We could use this in garment design to enhance personal cooling, which can improve the wearing comfort of garments or even result in energy savings in buildings. At present, however, radiative cooling has received insufficient attention in commercial design and production of textiles for wearable garments. Textiles that efficiently transmit the radiative heat were recently demonstrated, but either do not utilize standard weaving and knitting processes for wearable garments or require substantial process modifications. Here, we demonstrate the design and implementation of large-scale radiative cooling textiles for localized cooling management and enhanced thermal comfort using industry-standard particle-free nonporous micro-structured fibers that are fully compatible with standard textile materials and production methods. The micro-structured fibers, yarns and fabrics are part of a hierarchical photonic structure design that renders the textiles highly infrared transparent (up to > 0.8) while assuring visual opacity (up to 0.99). We design radiative cooling textiles with first-principles electromagnetic methods and fabricate them using commercial textile materials and formation facilities. Our “fabless” approach is confirmed by very good quantitative agreement between design and measurements. The resulting fabrics exhibit wearability properties expected of wearable textiles, and lower skin temperature by ≥ 3 °C compared to conventional textiles, which offers the potential for > 30 % energy savings in buildings and increases wearing comfort by significantly reducing the reliance on latent heat dissipation for thermoregulation.

Flexible Al2O3/polymethyl methacrylate composite nanofibers for high-performance sun shading and radiative cooling

Radiative cooling utilizes high solar reflectivity and high infrared emissivity to achieve temperature cooling down, which has attracted extensive research interest as a zero-energy consumption manner. However, the design of photonic structures and metamaterials for radiative cooling do not exploit commercialization prospects due to high production costs and structural complexity. Here we propose an economical and scalable electrospinning method for fabricating high-performance sun-shading and radiative cooling composite flexible film. The composite film is a non-woven fabric of nanofibers with polymethyl methacrylate (PMMA) and aluminum oxide (Al2O3) nanoparticles. The diameter of the nanofibers distributes in the range of 300–1200 nm and Al2O3 nanoparticles are wrapped in the nanofibers. The nanofibers strongly scatter sunlight to enable high solar reflectivity and back-reflection layer is no more necessary. Molecular vibrations of PMMA and phonon polarization resonance of Al2O3 nanoparticles synergistically result in high infrared emissivity. At a thickness of 0.30 mm and with an Al2O3 content of 8%, this material exhibits exceptional solar reflectance levels of approximately 95% between wavelengths of 1.1–2.5 µm. Additionally, it demonstrates an extremely low rate of solar absorption of approximately 5.5% between wavelengths of 0.3–1.1 µm, distinguishing itself as a highly effective solar selective surface even without the inclusion of a back-reflective layer. Furthermore, the material exhibits ultrahigh infrared emissivity levels of approximately 0.98 within atmospheric windows ranging from 8 to 13 µm. These remarkable characteristics position this material as a highly promising candidate for use in solar thermal energy conversion systems. As a result, the device shows a cooling performance that reaches a temperature decrease of up to 4.3 °C under 1000 W/m2 peak solar intensity and a decrease of 5.9 °C during nighttime. The composite film has enormous potential for use in vehicle shading and personal thermal management. Moreover, its mechanical properties allow for the flexibility needed to meet the demands of everyday use. Our work provides a straightforward and scalable method to manufacture products with exceptional sun shading and radiative cooling capabilities.

Climate-dependent optimization of radiative cooling structures for year-round cold energy harvesting

The potential of radiative cooling (RC) technology to passively and sustainably harvest cold energy has received a lot of interest. Since RC surface is installed outdoors for the long term and operates under time-varying environmental conditions, photonic structures constructing the RC surface need to be carefully designed based on proper theoretical modeling. However, a theoretical model for accumulated cooling performance that reflects environmental factors was scarcely developed for photonic structure design. This paper presents a more practical approach for estimating RC performance as an accumulated cold energy production (ACEP). ACEP stands for the time integration of cooling power, which is modeled to comprehensively account for a wide range of time-dependent environmental factors, including mostly neglected cloud coverage as well. The realistic performance of an RC surface at any location on Earth can be estimated without conducting any experiments. Utilizing the benefits of ACEP, photonic structures are separately optimized for 15 different locations with different climates. As a result, the universal optimum for all climates achieves 2 to 4 percent greater ACEP than the conventionally driven optimum. Based on the optimized results, we have developed a strategy for selecting radiative cooling structures based on regional climate conditions utilizing ACEP as a guiding metric. Furthermore, the impact of each environmental factor on RC performance is investigated by analyzing the ACEP components.

Integrated passive cooling fabrics with bioinspired perspiration-wicking for outdoor personal thermal management

Incorporating passive cooling technologies into personal thermal management is an effective strategy to protect the human body from the exacerbations of global warming. However, for the increasingly harsh outdoor environment, a single gentle heat dissipation route can no longer meet the cooling needs of the human body. Here, we report a passive cooling fabric integrating radiation and sweat evaporation, which consists of a polyvinyl alcohol/tetraethyl silicate (PVA/TEOS) fibrous membrane and a thin thermoplastic polyurethane (TPU) fibrous membrane. The integrated passive cooling fabric (IPCF) exhibits intense solar radiation reflectivity (98.67%) and provides sufficient atmospheric window (AW) emissivity (98.63%) due to the precise design of photonic structures and the inherent optical properties of the matrix. Benefiting from the biological structures of hierarchical porosity and surface energy gradient, the directional perspiration-wicking of IPCF can promote rapid evaporative cooling and keep the skin dry. Outdoor experiments reveal that the IPCF realizes subambient cooling of ∼8.8 °C during midday and human skin cooling of ∼14.7 °C. In the case of simulating perspiration, rapid wick-evaporation further contributes to a cooling of ∼5.7 °C. The high efficiency and energy savings of IPCF provide tremendous advantages for outdoor personal thermal management applications.

Region-specified inverse design of absorption and scattering in nanoparticles by using machine learning

Machine learning provides a promising platform for both forward modeling and the inverse design of photonic structures. Relying on a data-driven approach, machine learning is especially appealing for situations when it is not feasible to derive an analytical solution for a complex problem. There has been a great amount of recent interest in constructing machine learning models suitable for different electromagnetic problems. In this work, we adapt a region-specified design approach for the inverse design of multilayered nanoparticles. Given the high computational cost of dataset generation for electromagnetic problems, we specifically investigate the case of a small training dataset, enhanced via random region specification in an inverse convolutional neural network. The trained model is used to design nanoparticles with high absorption levels and different ratios of absorption over scattering. The central design wavelength is shifted across 350–700 nm without re-training. We discuss the implications of wavelength, particle size, and the training dataset size on the performance of the model. Our approach may find interesting applications in the design of multilayer nanoparticles for biological, chemical, and optical applications as well as the design of low-scattering absorbers and antennas.

DeepAdjoint: An All-in-One Photonic Inverse Design Framework Integrating Data-Driven Machine Learning with Optimization Algorithms

In recent years, hybrid design strategies combining machine learning (ML) with electromagnetic optimization algorithms have emerged as a new paradigm for the inverse design of photonic structures and devices. While a trained, data-driven neural network can rapidly identify solutions near the global optimum with a given data set’s design space, an iterative optimization algorithm can further refine the solution and overcome data set limitations. Furthermore, such hybrid ML-optimization methodologies can reduce computational costs and expedite the discovery of novel electromagnetic components. However, existing hybrid ML-optimization methods have yet to optimize across both materials and geometries in a single integrated and user-friendly environment. In addition, due to the challenge of acquiring large data sets for ML, as well as the exponential growth of isolated models being trained for photonics design, there is a need to standardize the ML-optimization workflow while making the pretrained models easily accessible. Motivated by these challenges, here we introduce DeepAdjoint, a general-purpose, open-source, and multiobjective “all-in-one” global photonics inverse design application framework that integrates pretrained deep generative networks with state-of-the-art electromagnetic optimization algorithms such as the adjoint variables method. DeepAdjoint allows a designer to specify an arbitrary optical design target, then obtain a photonic structure that is robust to fabrication tolerances and possesses the desired optical properties, all within a single user-guided application interface. We demonstrate DeepAdjoint for the design of infrared-controlled metasurfaces and show that a wide range of structures and absorption spectra can be achieved and optimized, including single- and multiresonance behavior through single- and supercell-class structures, respectively. Our framework, thus, paves a path toward the systematic unification of ML and optimization algorithms for photonic inverse design.

Hybrid inverse design of photonic structures by combining optimization methods with neural networks

Over the past decades, classical optimization methods, including gradient-based topology optimization and the evolutionary algorithm, have been widely employed for the inverse design of various photonic structures and devices, while very recently neural networks have emerged as one powerful tool for the same purpose. Although these techniques have demonstrated their superiority to some extent compared to the conventional numerical simulations, each of them still has its own imitations. To fully exploit the potential of intelligent optical design, researchers have proposed to integrate optimization methods with neural networks, so that they can work coordinately to further boost the efficiency, accuracy and capability for more complicated design tasks. In this mini-review, we will highlight some representative examples of the hybrid models to show their working principles and unique proprieties.

Experimental demonstration of inverse-designed silicon integrated photonic power splitters

Abstract The on-chip optical power splitter is a common and important device in photonic integrated circuits (PICs). To achieve a low insertion loss and high uniformity while splitting the guided light, multi-mode interferometer-based structures utilizing a self-imaging principle are widely used mainly in the form of a 1 × 2 configuration. Recently, an inverse design method for nanophotonic devices has emerged to overcome the limited capability of the conventional design methods and make it possible to explore the vast number of design parameters. Because of the non-intuitive shape of inverse-designed structures, they allow us to discover interesting and complex optical responses which are almost impossible to find with conventional design methods. Here, we report two kinds of inverse-designed 1 × 4 optical power splitters composed of silicon bars of different lengths, which are fabricated with a standard CMOS-compatible process. The particle swarm optimization method was used to minimize the insertion loss and divide the power evenly into each output port with finite-difference time-domain method simulation. The first optical power splitter has a compact size of 8.14 × 12 μm2 and the second optical power splitter has an even more compact size of 6.0 × 7.2 μm2. With the inverse designed structures, we fabricated the chip with a CMOS-compatible fabrication process. Experimental verification of the structures is provided and good agreement with the numerical results is obtained. The first 1 × 4 optical power splitter has a low insertion loss of less than 0.76 dB and uniformity of less than 0.84 dB, and the second more compact optical power splitter has a low insertion loss of less than 1.08 dB and uniformity of less than 0.81 dB. As the complexity of on-chip photonic systems has steadily increased, the inverse design of photonic structures holds great potential to be an essential part of advanced design tools.

Spectra dependent photonic structure design for energy harvesting by indoor photovoltaic devices

In this work, we report spectra-dependent energy harvesting by optimizing the photon management of an indoor photovoltaic device while taking into consideration the degradation of electrical transport characteristics caused by the nano-photonic structures. For the test case of a CdTe-based photovoltaic device, it has been shown that although the incorporation of dielectric-filled nanoholes in the absorber layer can enhance light absorption by about 40%, the optical-to-electrical conversion efficiency of the device is significantly diminished because of the degradation of the electrical transport characteristics. Instead, the best performance metrics are obtained when the nanostructures are incorporated in the window layer of the device alone. A finite difference time domain based numerical analysis, coupled with Poisson’s equation and continuity equation, shows that by controlling the areal density of the optimized structure in direct correlation with spectral characteristics of the illuminating light source, it is possible to maximize the overall power conversion efficiency of the indoor photovoltaic device. In the case of commercial white light-emitting diodes (LEDs), large arealdensities of holes are found to be more conducive for harvesting energy from cool-white LEDs, whereas smaller areal densities of holes provide better performances for illumination under warm-glow white LEDs.

Deep Learning for Photonic Design and Analysis: Principles and Applications

Innovative techniques play important roles in photonic structure design and complex optical data analysis. As a branch of machine learning, deep learning can automatically reveal the inherent connections behind the data by using hierarchically structured layers, which has found broad applications in photonics. In this paper, we review the recent advances of deep learning for the photonic structure design and optical data analysis, which is based on the two major learning paradigms of supervised learning and unsupervised learning. In addition, the optical neural networks with high parallelism and low energy consuming are also highlighted as novel computing architectures. The challenges and perspectives of this flourishing research field are discussed.

Photonics Empowered Passive Radiative Cooling

Passive radiative cooling has recently received renewed interest because of its unprecedented capabilities in cooling terrestrial objects below ambient air temperature without external energy consumption and greenhouse gas emission. This technology has been demonstrated as promising as replacements/complements of conventional compressed air‐based active cooling systems, which can significantly impact the global energy landscape by providing a green and efficient cooling way. The key to this success is judiciously designed photonic micro/nanostructures, which simultaneously reflect solar irradiation and emit thermal infrared emission across the atmospheric transparency window 8–13 μm. Herein, an introduction of the fundamental principles of passive radiative cooling is given, discussing the critical factors associated with the net cooling power of radiative cooling. Following this, the recently emerged photonic materials and structures (e.g., multilayer thin films, micro/nanoparticles, photonic crystals, metamaterials, metasurfaces, etc.) that facilitate radiative cooling are reviewed and fruitfully analyzed and discussed. Some possible scale‐up manufacturing ways toward the practical deployment of this energy‐efficient technology in real‐world applications are then discussed. The potential applications are also summarized and envisioned. Finally, perspectives on the future development in conjunction with artificial intelligent design of photonic structures and materials are presented and discussed.

Mechanical stress in InP and GaAs ridges formed by reactive ion etching

The mechanical deformation induced by reactive ion etching (RIE) of rectangular ridge waveguides in GaAs and InP has been investigated by photoluminescence and cathodoluminescence techniques. Several trends were identified and are discussed. First, it is concluded that the RIE process itself is the source of the mechanical deformation. A compressive volume change occurs mainly within the ridge (with a maximum close to the vertical etched sidewalls), extending outside the ridges, up to several micrometers below the bottom etched surface. An anisotropic deformation also appears, again mainly close to the etched sidewalls and below the bottom etched surface. A narrow area under tensile stress was also identified, localized outside the ridges and in a shallow region below the bottom etched surface. Cumulative, overlapping effects are seen inside the ridges where the compressive stress fields originating at the vertical etched sidewalls contribute to an overall compression inside the ridge which increases as the ridge width decreases. In addition, a tensile stress is also observed outside the ridge, strongly enhanced by the presence of neighboring ridges. These conclusions are significant for the design of photonic structures. Because of the photoelastic effect, which is important in GaAs and InP, the properties of devices such as waveguides might be affected by the mechanical stress described herein.

Inverse design of graphene-assisted metallodielectric grating and its applications in the perfect absorber and plasmonic third harmonic generation.

In this article, we propose a graphene metamaterial coupled with metallodielectric grating (GMCMG) structures to achieve plasmon induced reflection effects in the reflection spectrums. In order to enhance the light-matter interaction in the graphene, the micro-genetic algorithm is applied in the performance optimization for the GMCMG. Due to the absorption enhancement of graphene and the inverse design of photonic structures, a perfect absorber and an efficient third harmonic generator are obtained by employing optimized GMCMG structures. Compared with previous works, our design scheme provides a simple and efficient method for the optimization of photonic devices and has significant applications in optical modulators, absorbers and sensors.

Deep learning for the design of photonic structures

Innovative approaches and tools play an important role in shaping design, characterization and optimization for the field of photonics. As a subset of machine learning that learns multilevel abstraction of data using hierarchically structured layers, deep learning offers an efficient means to design photonic structures, spawning data-driven approaches complementary to conventional physics- and rule-based methods. Here, we review recent progress in deep-learning-based photonic design by providing the historical background, algorithm fundamentals and key applications, with the emphasis on various model architectures for specific photonic tasks. We also comment on the challenges and perspectives of this emerging research direction. The application of deep learning to the design of photonic structures and devices is reviewed, including algorithm fundamentals.

Deep Convolutional Mixture Density Network for Inverse Design of Layered Photonic Structures.

Machine learning (ML) techniques, such as neural networks, have emerged as powerful tools for the inverse design of nanophotonic structures. However, this innovative approach suffers some limitations. A primary one is the nonuniqueness problem, which can prevent ML algorithms from properly converging because vastly different designs produce nearly identical spectra. Here, we introduce a mixture density network (MDN) approach, which models the design parameters as multimodal probability distributions instead of discrete values, allowing the algorithms to converge in cases of nonuniqueness without sacrificing degenerate solutions. We apply our MDN technique to inversely design two types of multilayer photonic structures consisting of thin films of oxides, which present a significant challenge for conventional ML algorithms due to a high degree of nonuniqueness in their optical properties. In the 10-layer case, the MDN can handle transmission spectra with high complexity and under varying illumination conditions. The 4-layer case tends to show a stronger multimodal character, with secondary modes indicating alternative solutions for a target spectrum. The shape of the distributions gives valuable information for postprocessing and about the uncertainty in the predictions, which is not available with deterministic networks. Our approach provides an effective solution to the inverse design of photonic structures and yields more optimal searches for the structures with high degeneracy and spectral complexity.

Inverse design of photonic structures using an artificial bee colony algorithm.

We adapted the standard artificial bee colony algorithm for a binary search space in order to optimize photonic structures. The artificial bee colony algorithm in conjunction with the finite element method is applied for maximizing photonic bandgaps of photonic crystals. The proposed approach is assessed by two case studies assuming 2D photonic crystals comprised of silicon and air in a triangular lattice. The crystals were optimized to present large photonic bandgaps. The artificial bee colony algorithm presented superior performance when compared with that of a standard genetic algorithm, demonstrating an interesting approach to be used on the inverse design or optimization of photonic structures.