21 October 2019 in Nature Communications
MDMohsen Kamandar Dezfouli
+2
Nanophotonics finds ever broadening applications requiring complex components with many parameters to be simultaneously designed. Recent methodologies employing optimization algorithms commonly focus on a single performance objective, provide isolated designs, and do not describe how the design parameters influence the device behaviour. Here we propose and demonstrate a machine-learning-based approach to map and characterize the multi-parameter design space of nanophotonic components. Pattern recognition is used to reveal the relationship between an initial sparse set of optimized designs through a significant reduction in the number of characterizing parameters. This defines a design sub-space of lower dimensionality that can be mapped faster by orders of magnitude than the original design space. The behavior for multiple performance criteria is visualized, revealing the interplay of the design parameters, highlighting performance and structural limitations, and inspiring new design ideas. This global perspective on high-dimensional design problems represents a major shift in modern nanophotonic design and provides a powerful tool to explore complexity in next-generation devices.
1 October 2025 in APL Photonics
The use of inverse design techniques is radically changing the way we conceive and design photonic integrated devices. Geometries with unprecedented complexity can take full advantage of modern fabrication technologies to demonstrate high performance and ultra-compact devices. This tutorial aims at exploring the fundamental concepts behind the use of both iterative optimization and machine learning approaches in the context of photonic design. Throughout the paper, we present the design of an integrated mode exchanger, chosen mainly as a pedagogical example, to highlight the specificity of the field, the advantages and disadvantages of the different design techniques, and to discuss the most important considerations that need to be made in choosing the right design tool.
25 March 2025 in Optics Letters
Photonic integration in a silicon nitride (SiN) platform enables low propagation losses and a wide transparency range, making it ideal for quantum applications. However, polarization control in SiN is challenging because of its low birefringence. This work demonstrates polarization beam splitters using asymmetric directional couplers in SiN, designed for 950 nm and 1300 nm wavelengths. These devices exhibit low insertion losses (<1 dB) and high polarization extinction ratios (20 dB–30 dB).
12 August 2025 in Optics Express
Incorporating the thermal dependency of silicon permittivity into topological inverse design methodologies opens the design space for generating novel compact optical devices with temperature actuation for dense photonic integrated circuitry. In this work, a thermo-optic-enabled topological optimization methodology for integrated photonic device design is proposed and experimentally validated through an ultra-compact optical switch. With a footprint of 10 µm by 6~µm, a tenth of the size of conventionally designed MZI-based switches, the device operates in the C-band region with comparable insertion loss (<2 dB) and ON/OFF extinction ratio (<14 dB) performance. At intermediate temperatures, the device operates as a tunable power splitter with an insertion loss below 2 dB over the wavelength range from 1532 to 1555 nm. The tunable splitter achieves various splitting ratios from 0:100 to 100:0 at 1543 nm with a 50:50 splitting ratio that can be maintained over a wavelength range of 23 nm with temperature changes.
3 November 2026 in IEEE Photonics Technology Letters
MMManuel A. Méndez-Rosales