Size, material type, and concentration estimation for micro-particles in liquid samples

The on-site examination and characterization of microparticles are becoming crucial due to the significant rise in plastic pollution in natural resources. Hence, identifying the specific microplastic composition and quantity would enable the implementation of preventive measures. This paper presents a cost-effective setup that utilizes the Random Forest algorithm to detect the size and refractive index of micro particles, hence facilitating the identification of the material type. The system utilizes the scattering patterns of laser light from the dispersion of microparticles, namely within the concentration range of 0.05 fM to 3.00 fM. The refractive indices and particle sizes of melamine (Me8) spheres with a size of 8 μm, as well as polystyrene (PS8) spheres with a size of 8 μm and (PS10) 10 μm, were estimated using the Random Forest algorithm and recorded scattering patterns. The proposed method may deliver findings with an average deviation of 0.23 μm for particle size and 0.015 for particle refractive index. The statistical analysis indicated that there was no notable disparity between the experimental findings and the predictions derived from the machine learning system. The existing configuration can be readily converted into a point-of-use system that can be employed on-site for the purpose of monitoring and identifying microplastic contamination. [Display omitted] •In addition to size and concentration, material types of the micropollutants are crucial.•Optical scattering patterns provide critical information about material types of microspheres.•Particle size and refractive index information can be obtained by cost-effective measurement systems.•Integration of Random Forest algorithm boosts the process of Mie scattering calculations.•A portable, fast-response, and cost-effective system can be developed for detecting microplastics.