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.

INVESTIGATION OF THERMAL RADIATION ANGULAR AND POLARIZATION CHARACTERISTICS OF ONE-DIMENSIONAL PHOTONIC STRUCTURES ON A FINITE SUBSTRATE

One-dimensional photonic structures (PS) are the layered metamaterials, which optical properties are due to interference effects within the structure. Due to the interaction of electromagnetic waves with the periodic structure, the thermal radiation (TR) of such PS has the features of coherent: it has a selective spectrum and a petal pattern. The relevance of TR PS research is due to the search for new materials to create sources of infrared (IR) radiation in the middle and far IR range (with a wavelength > 5 μm). The second important area of application of the radiating properties of the PS is the creation of non-radiating coatings to remove unwanted TR from the heated surfaces of the IR devices. Despite the fact that the irradiative properties of PS have been studied quite well, in the literature insufficient attention is paid to the influence of the finite substrate on the parameters of TR structures. The influence of a finite incoherent substrate on the angular and polarization characteristics of the thermal radiation of the system (photonic structure)/substrate has been investigated. Systems consisting of plane-parallel Ge or Si plates as PS and incoherent BaF2, BaF2/Al and Al substrates were used in experimental studies. It has been established, the contrast and amplitude of the directional diagram petals of the TR system depend on the ratio of the optical characteristics of the photonic structure and the substrate. The presence of a strongly reflective substrate eliminates the difference between the angular dependences of the s- and p-polarized components of the TR. In the case when the PS is placed on a transparent incoherent substrate with a lower refractive index, there is an increase in the p-polarized component of the TR and the suppression of the s-polarized. The thermal imaging picture of the TR angular distribution of the Si/Al system has been recorded and it was experimentally proved that the TR of photonic systems exhibits circular patterns. The results of the work can be used in the development of narrowband emitters in the middle and far infrared range, coatings to increase radiation energy exchange and coatings with minimal emissivity that are invisible in the thermal range.

Features of thermal radiation of one-dimensional photonic structures on an absorbing substrate

The dependences of the thermal radiation (TR) lines contrast and amplitude of the systems (photonic structure)/substrate on the optical characteristics of both the photonic structure and the substrate have been investigated theoretically and experimentally. As it has been ascertained, these dependences demonstrate non-monotonic behavior, and the characteristics of the system TR can both increase and decrease depending on the ratio of individual optical parameters of the system components. The results of theoretical research were confirmed by experimental studies of TR systems Ge/GaAs, Ge/GaAs/Al, GaAs/Au and GaAs/(opaque substrate). The results of the research can be used in the development of narrow-band emitters for the middle and far IR range, which can be applied in the optical infrared devices for a wide range of applications.

Bound states in the continuum in periodic structures with structural disorder

We study the effect of structural disorder on the transition from the bound states in the continuum (BICs) to quasi-BICs by the example of the periodic photonic structure composed of two layers of parallel dielectric rods. We uncover the specificity in the robustness of the symmetry-protected and accidental BICs against various types of structural disorder. We analyze how the spatial mode localization induced by the structural disorder results in an effective reduction of the system length and limits the Q factor of quasi-BICs. Our results are essential for the practical implementation of BICs especially in natural and self-assembled photonic structures, where the structural disorder plays a crucial role.

Charge Tunable GaAs Quantum Dots in a Photonic n-i-p Diode

In this submission, we discuss the growth of charge-controllable GaAs quantum dots embedded in an n-i-p diode structure, from the perspective of a molecular beam epitaxy grower. The QDs show no blinking and narrow linewidths. We show that the parameters used led to a bimodal growth mode of QDs resulting from low arsenic surface coverage. We identify one of the modes as that showing good properties found in previous work. As the morphology of the fabricated QDs does not hint at outstanding properties, we attribute the good performance of this sample to the low impurity levels in the matrix material and the ability of n- and p-doped contact regions to stabilize the charge state. We present the challenges met in characterizing the sample with ensemble photoluminescence spectroscopy caused by the photonic structure used. We show two straightforward methods to overcome this hurdle and gain insight into QD emission properties.

Reprogrammable plasmonic topological insulators with ultrafast control

Topological photonics has revolutionized our understanding of light propagation, providing a robust way to manipulate light. So far, most of studies in this field are focused on designing a static photonic structure. Developing a dynamic photonic topological platform to switch multiple topological functionalities at ultrafast speed is still a great challenge. Here we theoretically propose and experimentally demonstrate a reprogrammable plasmonic topological insulator, where the topological propagation route can be dynamically changed at nanosecond-level switching time, leading to an experimental demonstration of ultrafast multi-channel optical analog-digital converter. Due to the innovative use of electric switches to implement the programmability of plasmonic topological insulator, each unit cell can be encoded by dynamically controlling its digital plasmonic states while keeping its geometry and material parameters unchanged. Our reprogrammable topological plasmonic platform is fabricated by the printed circuit board technology, making it much more compatible with integrated photoelectric systems. Furthermore, due to its flexible programmability, many photonic topological functionalities can be integrated into this versatile topological platform.