AI-Assisted Plasmonic Coupling Analysis of Spherical Gold Nanoparticles on Substrate

A method of electrostatic deposition of CTAB-stabilized gold nanoparticles on a modified APTES and PSS surface was considered. Positively charged gold nanoparticles with a spherical shape were synthesized using a one-step synthesis method with a CTAB surfactant and deposited on a negatively charged modified glass substrate surface with an APTES/PSS layer. Depending on the concentration of the gold nanoparticles, the deposition time, and the modification of the substrate, both isolated nanoparticles with a narrow plasmon peak close to the maximum position in solution, and interacting nanoparticles with varying degrees of plasmonic coupling, were obtained. We also present a deep learning approach for rapid, non-contact estimation of relative plasmon coupling (PC) in gold nanoparticles deposited on substrates using simple camera images. To obtain the training dataset, gold nanoparticles were characterized by the intensity of peaks corresponding to plasmonic coupling in the long-wavelength region of the spectrum. A fully connected neural network was trained to regress PC values from color features, minimizing the mean-squared error. The best model, retrained on the full training set, achieved R2 = 0.83, RMSE = 0.007, MSE = 0.086, and MAE = 0.050 on the test dataset.