Dimensional differences between commercial asbestos and cleavage fragments of amphiboles: classification approaches...

Asbestiform fibers and cleavage fragments of the same mineral have different origins, though distinguishing between samples of fibers and fragments is often complicated. This study aims to demonstrate an efficient method for distinguishing between samples of two commercial amphibole asbestos types (crocidolite and amosite) and their non-asbestiform varieties. For the study, 15 dimensional datasets were used. For classification, two metrics were used: the fraction of elongate mineral particles with 2.99log₁₀(length)-5.82log₁₀(width)-3.80 ≥ 0, and the Pearson Index, which is the linear correlation coefficient between log₁₀(width) and log₁₀(length) of elongate mineral particles in the set. The decision boundary of CriteriaParticlesFraction ≥ 0.34PearsonIndex + 0.35 was found to predict the correct habit for amphibole datasets with an average error rate of 0.4% (Cohen's Kappa statistics 0.992). Additionally, several quantitative characteristics were used that have been demonstrated to be predictive of mesothelioma potency factors of elongate particles. These include dimensional coefficient of carcinogenicity (DCC), found as 1 - exp(-A x Surface Area^K/(B x width^T +C)), EMPA (fraction of particles longer than 5 µm with diameter not higher than 0.15 µm), EMPB (fraction of particles longer than 5 µm with diameter not higher than 0.25 µm), and aerodynamic diameter of the particles. It was demonstrated that asbestiform and non-asbestiform datasets have significantly different dimensional parameters that can be related to dissimilar toxicological effects.