Programme

P.57 -- Poster, VCTP-49

Date: Friday, 2 August 2024

Time: 08:30 - 10:00

Machine Learning-Accelerated Design of Titanium Nitride Nanoring Metamaterial Absorbers for Broadband Solar Energy Harvesting

Nhat-Le Bui Dang1, Ngo Thi Que2, Anh D. Phan1,2, and Vu D. Lam3

1Faculty of Materials Science and Engineering, Phenikaa University, Hanoi 12116, Vietnam 2Phenikaa Institute for Advanced Study, Phenikaa University, Hanoi 12116, Vietnam 3Graduate University of Science and Technology, Vietnam Academy of Science and Technology, 18 Hoang Quoc Viet, Hanoi, Vietnam.

Metamaterial absorbers offer immense potential for applications in solar energy harvesting, radar systems, and emerging technologies. This study presents a novel two-layer metamaterial absorber design consisting of periodic titanium nitride (TiN) nanoring arrays on a TiN substrate, achieving an average absorption of 93% across the 200-3000 nm wavelength range, with a peak absorption of 99% and a total solar absorption efficiency of 95.5% under AM1.5 illumination. Recognizing the computational challenges in optimizing such nanostructured designs, we introduce a machine learning (ML)-driven approach to accelerate the design process. Three Python-based regression models (K-Nearest Neighbors, Random Forest, and Decision Tree) are trained on simulated absorption spectra and their corresponding structural parameters. Among these, the Decision Tree model demonstrates the highest accuracy with an R2 score of 0.99, enabling rapid prediction of absorption spectra from geometric parameters. This ML-accelerated approach holds promise for significantly streamlining the design and optimization of metamaterial absorbers for a wide range of applications.

Presenter: Đặng Bùi Nhật Lê