Inversion algorithm for estimating particle size

Inversion algorithm for estimating particle size information from scattering
properties.
Keywords: Particle size distribution, scattering, Vis/NIR spectroscopy, inversion algorithm, fitting
Major/minor: Biosysteemtechniek – Agrovoeding/Nanotechnologie; Statistiek
Abstract
Microstructure properties are very important quality parameters for materials in industry, medicine and
agriculture. The forward relation between the physical microstructure of materials and their scattering
properties can be well described with for instance Mie scattering theory. Nowadays, it is possible to measure
the scattering properties of various materials with accurate, non-destructive and fast techniques. Therefore,
the inversion of Mie theory would clear the way for the in-line estimation of physical microstructure
information from measured scattering properties. As scattering measurements are typically performed over a
range of wavelengths (e.g. Vis/NIR), and the effect of wavelength on the scattering properties is inversely
proportional to the effect of particle size, the scattering spectra can be consulted to estimate the particle size
distribution (PSD) of the scattering particles.
In this thesis, an analytical approximation of Mie theory will be consulted to relate the size of a spherical
scattering particle with its scattering properties. An inversion algorithm will be developed based on the
promising Philips-Twomey regularization in combination with a non-negative least squares fitting approach to
estimate the PSD from measured scattering spectra. Initially, this approach will be followed to estimate
predefined PSD’s (Weibull, lognormal, normal, …). Later on, it will be extended to also include a shapeindependent model based on spline functions. This approach will be validated on an available set of simulated
and measured scattering spectra for fat-in-water emulsions (milk, intralipid, …) and nanoparticle suspensions.
Promotor:
Prof. Wouter Saeys
BIOSYST - MeBioS (Mechatronics, Biostatistics and Sensors), KU Leuven
Kasteelpark Arenberg 30 Box 2456, B-3001 Leuven, Belgium
Tel: +32 16 328527
e-mail: [email protected]
website: http://www.biw.kuleuven.be/biosyst/mebios/biophotonics
Supervisor:
Dr. Ben Aernouts
BIOSYST - MeBioS (Mechatronics, Biostatistics and Sensors), KU Leuven
Kasteelpark Arenberg 30 Box 2456, B-3001 Leuven, Belgium
Tel: +32 16 321470
e-mail: [email protected]
Contact: [email protected]