Measurement of rod-shaped particle size using depolarized dynamic light scattering with random vector functional link...

This study presents an advanced algorithm combining random vector functional link neural networks (RVFLNN) and genetic algorithms (GA) for measuring the size of rod-shaped nanoparticles using depolarized dynamic light scattering (DDLS). RVFLNNs are trained to fit the autocorrelation functions of vertically and horizontally polarized light fields, extracting attenuation linewidths. The GA then solves the nonlinear equations to determine the particles' length and diameter. Experiments show that for aspect ratios ≤15, the inversion error is within 3%, outperforming Tikhonov's 8%. The algorithm also accurately measures particles with aspect ratios >16, with errors within 11%.