Characterization of Bimodal Particle Size Distributions (PSD) by Scanning Electron Microscopy (SEM) and Laser Difraction (LD)

A. GALLARDO; F. CABRERA; M.A. VILLAR; V. BUCALÁ; M.J. YÁÑEZ

Praga

Congreso; 18° International Microscopy Congress; 2014

Czechoslovac Microscopy Society

Among the more popular techniques for PSDs measurement, the SEM images analysis (IA) and laser diffraction (LD) are used alone or combined. In this work, 1 g of lactose (mean dnv= 26.5 μm, particle density = 1.36 g/cm3) and 3 g polyvinylchloride (PVC) resin (mean dnv = 142.6 μm , particle density = 1.51 g/cm3) particles were mixed to obtain a bimodal PSD. Figure 1 shows a SEM micrograph of the studied particulate system. Based on the materials properties,a relationship between the number of lactose (NL) and PVC (NPVC) particles of 46:1 is expected. The NL:NPVC ratio obtained from the LD number distributions was 43:1. However, the number passing cumulative distributions obtained from IA revealed that NL:NPVC varied from 4 to 33, being this relationship strongly influenced by the SEM sample preparation. The basic calculation and the information provided by LD (which handles thousand of particles) allowedimproving the method of mounting particles on the SEM stub. Figure 2 shows the number passing cumulative functions obtained by LD and IA (using the best SEM sample preparation method). LD estimates a higher NL than that calculated from IA, probably because LD can process much more particles and it is less sensitive to segregation problems than IA.Taking into account the mass ratio of lactose and PVC and their densities, the theoretical cumulative volume of lactose (VL) and PVC (VPVC) should be around 23 % and 77 %, respectively. LD provides directly the volume cumulative distribution of the powder that is being analyzed, instead for IA, this function has to be calculated from the number cumulative function assuming a given 3D particles shape. Figure 3 shows the volume cumulative passing distributions for LD and IA, for the last method spherical particles are assumed. These resultsindicate that VL for IA and LD was 28 and 18 %, respectively; values that are between the theoretical value (23 %). Eventhough IA estimates less NL, higher VL is found for this technique.For the IA data, the calculation of the volume of each class of particles assuming spherica shape requires to elevate the diameters to a power of 3. Eventhough lower NL is predicted by IA, the diameter measurement errors or the deviation of the particles shape from a sphere may lead to the observed higher VL. The simulated bimodal PSD and its evaluation by IA and LD allows stablishing the prediction errors and modifying sample preparation procedures to optimize the PSD chacterization.