ABSTRACT
A long-held tenet in computer science asserts that the training of deep learning is analogous to an alchemical furnace, and its "black box" signature brings forth inexplicability. For electromagnetic metasurfaces, the related intelligent applications also get stuck into such a dilemma. Although the past 5 years have witnessed a proliferation of deep learning-based works across complex photonic scenarios, they neglect the already existing but untapped physical laws. Here, the intrinsic correlation between the real and imaginary parts of the spectra are revealed using Kramers-Kronig relations, which is then mimicked by bidirectional information flow in neural network space. Such consideration harnesses the missing spectral connection to extract crucial features effectively. The bidirectional recurrent neural network is benchmarked in metasurface inverse design and compare it with a fully-connected neural network, unidirectional recurrent neural network, and attention-based transformer. Beyond the improved accuracy, the study examines the intermediate information products and physically explains why different network structures yield different performances. The work offers explicable perspectives to utilize physical information in the deep learning field and facilitates many data-intensive research endeavors.
ABSTRACT
In this paper, a D-shaped optical fiber plasmonic sensor using planar and grating structures of silver and gold metals is simulated using the finite element method under the wave optics module of COMSOL Multiphysics. Performance defining parameters are based on (i) the transmittance curve, viz., resonance wavelength (λ r), shift in resonance wavelength (Δ λ r), minimum transmittance (T m i n ), and bandwidth (BW), and (ii) on electric field distribution of a surface plasmon wave, viz., penetration depth (PD) and propagation length (PL) obtained for the considered sensor structures. It is found that gold gives wider BW than silver (e.g., at 1.39 refractive index of the sample: 480% for the planar case and 241% for the grating case), which deteriorates sensor performance by degrading detection accuracy. However, gold gives higher Δ λ r than silver (at 1.40-1.39=0.01 change in refractive index of the sample: 18.33% for the planar case and 16.39% for the grating case), which improves sensor performance and enhances sensitivity. A grating slightly increases the BW and Δ λ r for both gold and silver. Further, with respect to silver, the sensor that contains gold demonstrates higher PD (e.g., 22.32% at 1.39 refractive index of the sample for the planar case) and lower PL (e.g., 22.74% at 1.39 refractive index of sample for the planar case). A grating increases the PD (e.g., 10% for silver at 1.39 refractive index of the sample), whereas it decreases the PL (e.g., 8.73% for silver at 1.39 refractive index of the sample). Lower PL signifies the localization of the field, whereas higher PD enables the sensor to detect larger molecules. Therefore, the sensor with grating metals provides better sensitivity with reduced detection accuracy for the detection of comparatively larger molecules.
ABSTRACT
In this research, an arc I-shaped graphene sensing structure with multi-resonance characteristics is proposed for the simultaneous detection of vibrational fingerprints with spectral separation in the terahertz range. The resonant frequencies of the sensor can be dynamically tuned by changing the gate voltage applied to the graphene arrays. The two vibrational fingerprints of lactose molecules (0.53 THz and 1.37 THz) in the transmission spectrum can be enhanced simultaneously by strictly optimizing the geometrical parameters of the sensor. More importantly, these two resonant frequencies can be tuned precisely to coincide with the two standard resonances of the lactose molecule. The physical mechanism of the sensor is revealed by inspection of the electric field intensity distribution, and the advantage of the sensor, which is its ability to operate at a wide range of incident angles, has been demonstrated. The sensing performance of the structure as a refractive index sensor has also been studied. Finally, a double arc I-shaped graphene sensor is further designed to overcome the polarization sensitivity, which demonstrates excellent molecular detection performance under different polarization conditions. This study may serve as a reference for designing graphene biosensors for molecular detection.
ABSTRACT
We theoretically investigate the magnetic-field-induced terahertz absorption enhancement of a graphene-based one-dimensional photonic crystal using the 4×4 transfer matrix method for the circular polarization of light. The results show that the magnetically tunable absorption of the structure depends on the circular polarization state, magnetic circular dichroism, and, interestingly, absorption behaviors of right-handed and left-handed circularly polarized light interchange by changing the direction of the magnetic field. These properties can be used to design the circular-polarization-based sensors.
ABSTRACT
The dependence of the optical responses of microcavity-type one-dimensional (1D) photonic crystals and 1D magnetophotonic crystals (1D-MPCs) on the arrangement order ratio (AOR), which is the ratio of the first and second layers' refractive indexes of Bragg reflectors, is studied. It is demonstrated that the optical properties of two microcavity structures resulting from exchanging the order of Bragg reflector layers so that they have the same optical contrast ratios and different AORs would be different. This difference will be more noticeable in the case of microcavity type 1D-MPCs. Regarding the arrangement-dependent magneto-optical properties of the structures, we can propose the efficient Bragg reflectors with high transmittance enhanced Faraday rotation.
