A PhD dissertation submitted by the student Haider Sami Jassim Al-Dahwi, entitled:
“Design of Optical Polarization Converters and Graphene-Based Switch using Deep Learning for QKD Protocols”
was discussed at the Institute of Laser for Postgraduate Studies – University of Baghdad, on 2/9/2026, under the supervision of Assistant Professor Dr. Shilan Khosrow Tawfiq.
The examination committee consisted of Prof. Dr. Abdul Hadi Mutasher as Chairman, and members Prof. Dr. Suad Salman Ahmed, Asst. Prof. Dr. Mohammed Nazim Abbas, Asst. Prof. Dr. Tahreer Safaa Mansour, and Asst. Prof. Dr. Fatima Hamed Rajab.
The study aims to design, build, and analyze three optical devices based on Metasurfaces, using the Finite Element Method and the simulation software (COMSOL Multiphysics) and Deep Learning (DL), to be used within a polarization modulator for Quantum Key Distribution (QKD) systems, namely: a metasurface grating polarizer, an L-shaped hybrid polarization converter, and a tunable optical switch based on graphene.
By employing deep neural networks and the Surrogate Model in COMSOL, it was possible to improve the extinction ratio of the metasurface grating polarizer from 1800 to 60000 in the wavelength range of 632 nm.
As for the L-shaped polarization converter, studies were conducted to improve the structure’s ability to rotate polarization, and to make it operate within the target wavelength of 890 nm, which is a wavelength practically used in quantum channels via satellites. In addition, three new structures were designed with the aim of increasing the rotation angle of light polarization to ensure the possibility of obtaining the required polarization states in QKD protocols. With the help of deep learning, the geometric dimensions of each structure were optimized and the six required polarization states were obtained, namely: 0° and 90° and 45° and -45° and left circular polarization (LCP) and right circular polarization (RCP), with achieving high performance characteristics for the outgoing light through the metrics: polarization angle, degree of polarization, degree of linear polarization and degree of circular polarization.
Several structures corresponding to each required polarization state can be assembled on one platform to form a polarization modulator specific to QKD systems.
To operate the modulator, it requires an optical controller capable of passing light in a specific structure and blocking it in other structures. For this purpose, graphene was chosen, which is a smart material whose properties can be tuned electrically by applying a bias voltage that changes its surface conductivity, to build a tunable optical absorber that works as an optical switch, where a new metasurface structure was proposed that includes a metallic ring for the purpose of concentrating the resonant electric field within the near-infrared range, and relies on the transmission mode which is a prerequisite in the proposed system. The optimization process was carried out using deep learning on the proposed structure, achieving a high total absorptance and reflectance in the blocking state and high transmittance in the passing state.
By integrating the tunable optical absorber structure with the polarization converter structures dedicated to different polarization states, it became possible to control electrically and build an integrated system that works as a polarization modulator for quantum key distribution technologies.
The dissertation was accepted with a Very Good grade, as it fulfills the requirements for obtaining a PhD in Philosophy in Laser Applications.
