Determination of Nano- and Micro-droplet Parameters in Levitating Suspension Micro-droplets by Convolutional Neural Network-based Speckle Image Analysis
Description
The optical response of a suspension microdroplet is governed not only by the properties of the dispersed phase, but also by the finite size and optical structure of the droplet itself. As a result, the interpretation of scattered-light patterns from such systems constitutes a non-trivial inverse problem. We examined whether laser speckle images recorded from single levitating microdroplets of suspension can be used for data-driven recognition of selected droplet and suspension parameters. Experiments were performed on slowly evaporating microdroplets of monodisperse TiO2 nanoparticle (NP) suspensions in diethylene glycol confined in a linear electrodynamic quadrupole trap. Speckle images were analyzed with a convolutional neural network (CNN) trained to classify droplet diameter, nanoparticle concentration, and nanoparticle diameter, first in separate tasks and then in combined two-parameter and three-parameter classifications. The results suggest that CNN-based analysis of speckle images may provide a viable route toward multi-parameter optical diagnostics of free suspension microdroplets and, potentially, more complex aerosol-like systems. Since the training set size is over 290 GB and the independent data set is over 230 GB, here we provide only sample images (5 of each 5000-image class) used for training and, in a separate folder, from independent data set (5 of 4x1000 - one out of 4 specific microdroplets radii was selected). Each (Matlab) .mat file contains a single frame: a 640x480 array of doubles obtained from the corresponding original image by subtracting the background (a frame with no droplet), Gaussian filtering of width 2 and normalizing from min to max. Names of folders reflect the nanoparticle radius in nm, nanoparticles suspension mass concentration in mg/mL and microdroplet diameter in micrometers. All images were acquired with 20 ms exposure time. We verified that such exposure time is short enough by analysing the autocorrelation function (ACF) obtained from the dynamic light scattering (DLS) experiments performed on the studied microdroplets of suspension. The CNN code is provided in the CNN_4_Speckle_from_droplets_w_tweaks.m Matlab file. The sample ACFs from dynamic light scattering on microdroplets of 6 different diameters, with 20 mg/mL concentration of 100-nm diameter TiO2 NPs are provided in ACFs.csv file. The first line contains the column names: time, 70,60.1,81.7,107.1,131,117, where the numbers give the diameter of the microdroplets. The second line contains the units designation.
Files
Institutions
- Cardinal Stefan Wyszynski University in WarsawMazovia, Warsaw
- Institute of PhysicsMazovia, Warsaw