ANN and DL based design of Nanophotonic Structures

The nanophotonic structures have potential applications in optical information processing, photonic integrated chips in high-speed data centers, and biosensing devices. We aim to simulate and design structures in TM mode whilst most of the literature present focuses on TE mode. Datasets will be gene

Project Title

ANN and DL based design of Nanophotonic Structures

Project Area of Specialization

Artificial Intelligence

Project Summary

The nanophotonic structures have potential applications in optical information processing, photonic integrated chips in high-speed data centers, and biosensing devices. We aim to simulate and design structures in TM mode whilst most of the literature present focuses on TE mode. Datasets will be generated to train neural network-based frameworks for the inverse design of targeted devices. This will reduce the computational complexity of the simulation process and will significantly reduce the time required for the optimal design of nanophotonic sructures.

Project Objectives

In this work, we employ the concepts of artificial neural networks and deep learning for nanophotonic design and Optimization. This helps in the extraction of detailed information regarding the role of nanophotonic design parameters in determining the output response of an Integrated Optical grating. Moreover, the work also aims to provide an intuitive understanding of the physics involved in the light-matter interaction in the nanophotonic structures without imposing any severe computation complexity.

Project Implementation Method

We have simulated and designed 2-D silicon integrated optical grating in COMSOL Multiphysics software/ Optics Module based on  Finite  Element  Method  (FEM)  analysis, in Transverse Magnetic(TM) mode whilst most of the literature present on nanostructure simulation focuses on Transverse Electric (TE) mode. A dataset of 90,000 samples has been generated via a MATLAB script which has successfully automated the dataset generation process, to train a neural network-based framework for inverse design of optical grating. For data samples, the grating periodicity is varied from 400 to 950 nm and the thickness of the silicon slab is varied from 200 nm to 260 nm considering the industrial lithographic limits. These data samples were used to train a 3 neural networks based framework to predict optimal values of design parameters for required transmittance. For evaluation of framework performance, predicted parameters were used for spectrum simulation, and the Mean Squared Error of the simulated spectrum and input spectrum was calculated. 
Simulation and data generation of the 3D bowtie antenna and plasmonic grating is our next step, and the data of 2-D bowtie antenna and plasmonic grating will be used to train models for inverse design of the mentioned nanostructures to obtain optimal nanophotonic structures in reduced time and lesser computational complexity.

Benefits of the Project

•    Societal and Economical Considerations:

We have seen in recent years that the world has become a global village. Due to the inter-connectivity of the world and the large transfer of data we’ve witnessed a great amount of increment in the global traffic of data as well the recent development of technology has led to an increase in the needs of a high rate of data transfer for better services as well as new technological advancements. All these data-intensive applications such as video streaming, cloud computing, and social networking require high bandwidth communication. This increase in demand of data traffic is impossible to be handled with the existing conventional ”metallic interconnects” with limited resistance and capacitance as they will reach their limit very soon with an exponential increase in data hence it is integral to shift towards ”Optical interconnects” for enabling high speed, efficient data transfer. Therefore, it is important to have a solution that can help in acquiring optimized design of nanophotonic structures integral for the development of photonic ICs to be used as optical interconnects.

•    Environmental Considerations:

The increase in demand of high data transfer rate and exponential growth in data transfer requires devices that will be capable of achieving high energy efficiency, are low in cost, and offer high bandwidth capacity. Therefore, integrated photonics has a crucial role to play as they have emerged as a promising candidate due to their specific properties for providing high bandwidth capacity along with being power efficient at a very low cost relative to their electrical counter-parts as they transfer data at the speed of light, which is the maximum speed known to mankind.

Technical Details of Final Deliverable

Design and Simulation of Nanophotonic Structures:

We have to simulate and design nanostructures using a multiphysics software - COMSOL. The structures will be simulated using its wave optics section in the optics module. An FEM analysis will be performed in the EM frequency domain for the simulation. The structures to be simulated are:
1.    Integrated Optical Grating.
2.    Plasmonic Grating.
3.    Bowtie Antenna
•    Data Generation:

For the generation of around 90,000 samples of transmittance and reflectance values of the nanophotonic structures at different values of varying parameters, we have to integrate COMSOL with MATLAB for the automatic generation of data files for acquiring quick results.

•    Implementation of Deep Learning Techniques:

A model of the neural network is to be developed for training on the acquired data. After the completion of the training of the network, it will be tested on the testing data set. A 95 - 98% accuracy is the target to be achieved.

•    Development of User Interface:

The final product will consist of an easy to use system which will allow the users to get quick results with minimum computation for varying parameters of a nanophotonic structure.

Final Deliverable of the Project

Software System

Core Industry

Telecommunication

Other Industries

Manufacturing

Core Technology

Artificial Intelligence(AI)

Other Technologies

Others

Sustainable Development Goals

Industry, Innovation and Infrastructure, Responsible Consumption and Production

Required Resources

If you need this project, please contact me on contact@adikhanofficial.com