Evaluation and Design of Photothermal Conversion Performance for Complex Morphology Nanofluids Via Bidirectional Deep Neural Network

Abstract

The optical absorption and scattering of plasmonic nanoparticles are crucial for optimizing photothermal conversion efficiency, which holds great potential in applications such as plasmonic photothermal therapy, desalination, and hot-electron chemistry. However, the optimization design of photothermal conversion performance of nanofluids is a challenging task that involves multiple complex physical processes, especially for complex morphology nanofluids. To evaluate and design nanofluid photothermal conversion performance, we developed a novel method by combing electromagnetic scattering calculation, radiative transfer simulation, and machine learning. This method systematically investigates the whole process from the geometric parameters of individual nanoparticles to the radiative properties of nanofluids. We predict the absorption spectra of nanofluids and design geometric parameters to achieve desired solar absorption efficiency modeling deep neural networks bidirectionally. The results demonstrate that by adjusting the shape, material composition, and geometric characteristics of nanoparticles, it is possible to modulate resonance absorption peaks and spectral absorption. Moreover, the bidirectional neural network model achieves spectral prediction with 99% accuracy and geometric parameter design with 93% accuracy. This work provides a widely applicable and computationally efficient method that may promote the evaluation and design of nanofluids in photothermal conversion.

Keywords: nanofluids, Plasmonic nanoparticles, Radiative properties, Deep neural network, Solar photothermal conversion.