Inverse optimization design of terahertz topological waveguides towards on-chip communication

Area of Science:

• Photonics and optical engineering
• Condensed matter physics
• Computational electromagnetics

Background:

• Topological waveguides offer robust low-loss signal transmission.
• Valley photonic crystals provide unique topological properties for waveguide design.
• Terahertz (THz) on-chip communication demands high bandwidth and low loss.

Purpose of the Study:

• To develop an inverse design framework for terahertz topological waveguides.
• To optimize waveguide structures for enhanced on-chip communication performance.
• To establish a direct link between waveguide structure and communication metrics like bit error rate (BER) and bandwidth.

Main Methods:

• Utilized valley photonic crystals for topological waveguide construction.
• Trained a deep neural network (DNN) for high-precision prediction of waveguide performance from structural parameters.
• Integrated DNN with a terahertz on-chip communication link to establish a forward design flow.
• Applied particle swarm optimization (PSO) for inverse optimization of waveguide structures.

Main Results:

• Achieved a DNN prediction error as low as 10^-6.5 after 30,000 epochs.
• Established a forward design flow mapping waveguide structure to communication performance (BER, bandwidth).
• Identified two terahertz communication windows (around 0.28 THz and 0.31 THz) supporting low-BER 3.33-Gbps communication with bandwidths of 14.81 GHz and 13.67 GHz, respectively.
• Demonstrated optimization meeting BER thresholds below 10^-3.

Conclusions:

• The proposed framework enables 'on-demand' inverse design of terahertz topological waveguides.
• Provides an efficient and convenient approach for intelligent optimization of THz on-chip communication waveguides.
• Facilitates the realization of high-performance terahertz integrated circuits for advanced communication applications.