Classification and quantification of minor iron-sulfide concentrations in concrete aggregate using automated mineralogy

Abstract

Iron sulfide concentrations and mineral associations triggering the internal deterioration of concrete structures are still enigmatic. Incidences of internal sulfate attacks induced by iron sulfide-containing concrete aggregates appear worldwide. Severe cases are reported from Canada, the United States of America, and Ireland. Moreover, conservative limits for the total sulfur content of aggregates increased the need to dispose of otherwise high-quality resources for concrete production. The maximum threshold values for total elemental sulfur in the European standard EN-12620 for concrete aggregates are \(\le\) 1 wt.\(\%\), and as little as \(\le\)0.1 wt.\(\%\) if the non-stoichiometric iron-sulfide pyrrhotite (Fe_(1-x)S) is present in the rock. This study investigates the potential of scanning electron microscopy-based automated mineralogy for mineral classification and the quantitative quality assessment for concrete aggregate material. Identifying the stoichiometrically closely related disulfide pyrite and monosulfide pyrrhotite is emphasized. The iron/sulfur ratio and greyscale variations in the electron backscatter images between pyrite and pyrrhotite were tested as additional differentiation criteria when acquiring mineral mapping and point-of-interest analysis. The added greyscale criterion yielded a better distinction between the two chemically similar phases. A good correlation was achieved when comparing results from energy-dispersive X-ray spectroscopy in automated mineralogy with wavelength-dispersive spectroscopy point analyses on the electron microprobe. Semi-quantification of the chemical data from automated mineralogy was computed for the total sulfur content in the petrographic samples. The total sulfur content of bulk samples, investigated by high-temperature combustion and inductively coupled plasma atomic emission spectroscopy, was consistent with the semi-quantitative results of automated mineralogy.