Light extinction spectroscopy applied to polystyrene colloids: Sensitivity to complex refractive index uncertainties and to noise

•Particle size distribution and concentration of PS-DI colloids are measured via LES.•Sensitivity of LES to refractive index uncertainties & system noise is investigated.•Experimentally characterized noise profile is embedded into numerical simulations.•Wavelength spectrum limited to high SNR region improves the LES results.•In-situ calibration of complex refractive index enhances the quality of LES results.•Both mono and bi-modal size distributions are experimentally retrieved. The measurement of particle size distribution (PSD) in colloids and nanofluids presents many challenges, especially when it requires to be conducted in-situ and real-time. Our work aims to assess the capabilities of Light Extinction Spectroscopy (LES) technique to determine concentration and volumetric PSD (vPSD) of colloids and nanofluids. Numerical simulations are performed to verify robustness of LES data inversion algorithm and to identify the most relevant uncertainty sources. Experiments are carried out on well-dispersed water based colloids containing Polystyrene particles with median diameters at 120 nm and 300 nm. LES results are compared with supplier’s data, 3D-DLS measurements, and SEM images. Via sensitivity analyses, LES sensitivity to particle complex refractive index spectrum and to system noise level are examined. To improve accuracy, we limit LES data inversion at shorter wavelengths where high SNR is acquired. The noise is experimentally characterized and embedded into numerical simulations to better mimic the real conditions. We show that while noise level affects mostly vPSDs’ peak height, small inaccuracies in complex refractive index spectrum deteriorate the stability of the inversion, especially for smaller vPSDs. To overcome the latter, we carry out a preliminary in-situ calibration on particle complex refractive index spectrum. Accordingly, results are substantially improved with maximal discrepancies smaller than 3% for median and volume mean diameters, and lower than 13% for number concentrations. Besides, the retrieval of bi-modal distribution is promising. In conclusion, our methodology can be considered as a guideline to evaluate the applicability and accuracy of the LES technique in colloids.