Connect with us

With increasing upper frequency limits and heightened requirements for compact design and energy efficiency, researchers are constantly seeking new ways to improve the developments of optical antennas. These components are of the utmost importance in a multitude of practical applications, including data transmission, photonic sensing and nanoscale measurements.

Modern microchip technology offers good opportunities to reduce dimensions of signal processing circuitry. At the same time, better fiber-to-chip coupling methods are needed to maintain adequate levels of energy efficiency and directionality of the emitted signal.

In a research paper recently published on arXiv.org, the team of scientists presented an improved method for design optical phased arrays for high-density fiber-to-chip coupling applications. The suggested approach combines adjoint optimization and machine learning-based dimensionality reduction to perform multi-objective optimization with aim to discover high-performance antenna designs. Authors present a design example which illustrates how effective this methodology is when analyzing a large number of different performance-related parameters and mapping the optical array design space to a practically viable physical model of grating-based optical phased-array antenna.

In this paper we have exploited a methodology based on adjoint optimization and machine learning dimensionality reduction for the multi-objective design optimization of a grating-based micro-antenna in a 300-nm SOI platform. The compact antenna is only 3.6 mm long, has a perfectly vertical diffraction efficiency of almost 92%, and directionality of 98%. When coupled with an optical fiber with mode field diameter of 3.2 mm vertically placed on top of the antenna, a coupling efficiency of more than 81% is achieved with a wide 1-dB bandwidth of almost 158 nm. Reflection is smaller than -20 dB over the entire 1450 nm – 1650 nm wavelength range. These good performances make the antenna ideal for applications requiring dense arrays of both fiber and free-space coupling interfaces.

Link to the research article: https://arxiv.org/abs/2008.02552