Inverse design of plasmonic metasurfaces by convolutional neural network

Lin, Ronghui; Zhai, Yanfen; Xiong, Chenxin; Li, Xiaohang

doi:10.1364/ol.387404

Cited by 73 publications

(46 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Typically, a neural network can be referred to as a forward model where the inputs are physical or geometrical parameters and outputs are optical parameters, or as an inverse model which predicts the output geometrical or physical parameters based on the input optical parameters. Some recent works have considered inverse-design-based approaches for metasurface optics [47][48][49][50][51][52][53][54][55][56][57][58], many of which use neural networks (NNs) to accelerate optimization by reducing the required number of numerical simulations [47][48][52][53][54][55][56][57][58][59][60].…”

Section: A Design Methodsmentioning

confidence: 99%

Inverse Design of On-Chip Thermally Tunable Varifocal Metalens Based on Silicon Metalines

Zarei

Khavasi

2021

IEEE Access

View full text Add to dashboard Cite

High-contrast transmitarray (HCTA) metasurfaces, named as metalines in this article, a category of lithographically defined on-chip diffractive elements with minimized scattering loss, enables parallel on-chip optical signal processing. While high-performance devices with different functionalities are reported, most of them are static, and their optical functions are fixed after fabrication. However, dynamically tunable metasurfaces are of essential need in many modern optical systems. Here, we employ an inverse design method, combining FDTD parameter sweep with Fourier-optics simulations, to design thermally-controlled varifocal metalenses. Thermal tunability is accomplished through the thermo-optical effect of silicon. Maximum focal length tunability of 12% and 24% are demonstrated for 400µm and 200µm aperture varifocal metalens doublets at the wavelength of 1.55µm, respectively. The focusing efficiency is more than 74% for the whole temperature range. The 400µm and 200µm aperture metalenses are able to concentrate the input beam to spot sizes less than one wavelength and two wavelengths, respectively. This paves the way for dynamically controlled silicon photonics computational chips.

show abstract

Section: A Design Methodsmentioning

confidence: 99%

Inverse Design of On-Chip Thermally Tunable Varifocal Metalens Based on Silicon Metalines

Zarei

Khavasi

2021

IEEE Access

View full text Add to dashboard Cite

show abstract

“…We want to note that CNNs, relying on their capability of processing large dimensional data, have become an indispensable architecture that deal with photonic devices with the high DOF. [48,49]…”

Section: Design Of High Dof Photonic Devicementioning

confidence: 99%

“…When leveraging a discriminative model for the design and optimization of photonic devices, training a surrogate model that approximates the physical responses of the devices is always necessary. For example, deep learning techniques have been leveraged for the modeling of photonic crystals, [49,50] metasurfaces, [51] and plasmonics. [52] The accuracy of the surrogate model determines the fidelity of the design and thus the additional efforts of post-processing the design.…”

Section: Efficient Modeling Of Photonic Systemmentioning

confidence: 99%

Tackling Photonic Inverse Design with Machine Learning

et al. 2021

View full text Add to dashboard Cite

Machine learning, as a study of algorithms that automate prediction and decision‐making based on complex data, has become one of the most effective tools in the study of artificial intelligence. In recent years, scientific communities have been gradually merging data‐driven approaches with research, enabling dramatic progress in revealing underlying mechanisms, predicting essential properties, and discovering unconventional phenomena. It is becoming an indispensable tool in the fields of, for instance, quantum physics, organic chemistry, and medical imaging. Very recently, machine learning has been adopted in the research of photonics and optics as an alternative approach to address the inverse design problem. In this report, the fast advances of machine‐learning‐enabled photonic design strategies in the past few years are summarized. In particular, deep learning methods, a subset of machine learning algorithms, dealing with intractable high degrees‐of‐freedom structure design are focused upon.

show abstract

“…In recent years, with the burgeoning field of metasurfaces, deep learning has emerged as a powerful tool for realising efficient inverse design of different types of plasmonic metasurfaces for different applications including spectral control, near-field design [9][10][11]. In 2018, Malkiel et al introduced a novel bidirectional DNN model which can realise both the design and characterisation of plasmonic metasurfaces [12] .…”

Section: Design Of Plasmonic Metasurfaces By Deep Learningmentioning

confidence: 99%

Deep Learning Enabled Nanophotonics

Huang¹,

Xu²,

Miroshnichenko³

2020

Advances and Applications in Deep Learning

View full text Add to dashboard Cite

Deep learning has become a vital approach to solving a big-data-driven problem. It has found tremendous applications in computer vision and natural language processing. More recently, deep learning has been widely used in optimising the performance of nanophotonic devices, where the conventional computational approach may require much computation time and significant computation source. In this chapter, we briefly review the recent progress of deep learning in nanophotonics. We overview the applications of the deep learning approach to optimising the various nanophotonic devices. It includes multilayer structures, plasmonic/ dielectric metasurfaces and plasmonic chiral metamaterials. Also, nanophotonic can directly serve as an ideal platform to mimic optical neural networks based on nonlinear optical media, which in turn help to achieve high-performance photonic chips that may not be realised based on conventional design method.

show abstract

Inverse design of plasmonic metasurfaces by convolutional neural network

Cited by 73 publications

References 33 publications

Inverse Design of On-Chip Thermally Tunable Varifocal Metalens Based on Silicon Metalines

Inverse Design of On-Chip Thermally Tunable Varifocal Metalens Based on Silicon Metalines

Tackling Photonic Inverse Design with Machine Learning

Deep Learning Enabled Nanophotonics

Contact Info

Product

Resources

About