Microstructures Causing Structural Instability: Applying Electron Backscatter Diffraction (EBSD) to Samples of Pyrrhotite Oxidation-Induced Concrete Degradation

Electron Backscatter Diffraction (EBSD) allows for the small-scale determination of a material's crystal structure.This complements the determination of the elemental composition of a phase by Energy Dispersive Spectrometry (EDS).Further, mapping the orientation of the crystal at each pixel may highlight local differences even where the chemistry and structure remain the same.Here, we evaluate the effectiveness and applicability of EBSD in determining variations in the crystal structure of samples of the mineral pyrrhotite (Fe 1-x S where x is between 0 and 0.125) within and among grains before and after undergoing oxidation.Some residential foundations in Connecticut and Massachusetts cracked and crumbled due to the presence of the mineral pyrrhotite in the crushed stone aggregate used in their concrete costing homeowners $150,000 to $250,000 to replace (Fig 1A) [1].Pyrrhotite oxidizes, producing iron oxyhydroxide phases (e.g., geothite, ferrihydrite) as well as sulfate and hydrogen ions (Fig 1B).These products react with the cement paste, producing volumetrically larger phases and reducing the foundation's structural stability [2].To better understand the pyrrhotite reaction processes involved and to advance measurement techniques for these materials, NIST took multiple concrete core samples from a degraded foundation for structural testing, chemical composition, and microstructural characterization.Pyrrhotite bearing aggregate from this foundation, from the same Connecticut quarry as that used in the house, and from a separate quarry in Quebec, Canada related to pyrrhotite oxidation problems, were mounted and prepared for EBSD analysis.Pyrrhotite occurs as four common non-stoichiometric compositions based on replacing iron atoms with vacancies: Fe 7 S 8 , Fe 9 S 10 , Fe 10 S 11 , Fe 11 S 12 .Structurally, this is accommodated by elongation in the direction of the C axis of the base unit cell by some multiple: e.g., the C-axis is four times longer (4C) for Fe 7 S 8 , five times for 5C, etc.Each of these polytypes may be monoclinic, hexagonal, or orthorhombic due to small lattice distortions.The most iron-poor composition, 4C is generally monoclinic and is the ferrimagnetic member of the group.More iron-rich compositions are commonly hexagonal or orthorhombic and are antiferromagnetic.Rezvani et al. [3,4] applied EBSD to pyrrhotite analysis in ore samples, indicating that pyrrhotite showed fine variations on the scale of a few microns.Limiting patterns to samples identified as natural minerals, pyrrhotite structures 4C, 5C, and 6C were identified in the same grain.Rler et al. demonstrated that EBSD is applicable to cement clinker and ordinary Portland cement [5] and to the investigation of the degradation of concrete by the alkali-silica reaction (ASR) [6].Uwanyuze et al. looked at the preparation of samples of concrete degradation by pyrrhotite oxidation and the determination of major phases present [7].Samples were epoxy mounted using Epo-Tek 301 1 and polished using silicon-carbide papers to 1200 grit and polycrystalline diamond pastes to 0.25 m grit followed by one hour in a vibratory polisher using 0.04 m grit colloidal silica suspension.A thin coat of carbon (60 seconds versus our typical 120 seconds, to help achieve clear Kikuchi patterns) was applied, and conductive silver paint was used to provide an electrical ground path and secure the sample to the mount.Samples were analyzed on a JEOL 6700F SEM mounted with Bruker e -FlashHD EBSD using an accelerating voltage of 20 kV and beam current of 3 nA.As the different structures of pyrrhotite are very similar, a high-definition mode with the phosphor screen imaged at 1600 x 1200 pixels was used.Due to file size limitations at this resolution, only thin transects and small areas were collected.Kikuchi patterns were obtained from pyrrhotite in each sample, indicating the successful removal of polishing-induced strains (Fig. 2).Patterns were generally not obtainable from the iron-oxyhydroxide phase filling fractures (Fig. 3A).This may be due to the presence of non-crystalline phases or to very microcrystalline phases and/or multiple phases finely mixed.Reduced current and aperture to reduce spot size were tried but did not improve results.However, some weak patterns were evident in places, and this warrants further investigation.Pyrrhotite from the Quebec sample overall matched with monoclinic 4C (Fe 7 S 8 ).The fit of monoclinic 4C (ICSD-ID: 151765 or 151766) was only slightly different than 5C (ICSD-ID: 190012) in general, for example 0.92 vs. 0.98 misfit.Auto-matching used by the mapping function initially suggested significant variation in the grains but careful investigation of several points and manually ensuring all 12 lines were in place reduced the phases present down to 4C.Maps of crystal orientation at each pixel resulted in uniform directions indicating a continuous crystal.Pyrrhotite from the Connecticut quarry aggregate was also identified as monoclinic 4C with similar difficulty in discerning patterns.A hexagonal pattern for pyrrhotite 4C initially also matched frequently but, after optimization, was shown to have a higher misfit.Maps of crystal orientation were similar in consistency to the Quebec sample.Pyrrhotite from the Connecticut foundation core generally had weaker patterns.This sample was in a separate epoxy mount from the other two and this may be due to slight surface oxidation from sample handling or a different carbon coat thickness.Monoclinic pyrrhotite 4C was the primary polytype identified.In places where 5C and 6C were identified, it was more challenging to evaluate the polytype present.When unconstrained pattern searches on pixels were conducted, patterns associated with troilite (FeS) commonly matched more lines.Examining patterns across grains showed consistent directions (Fig. 3B) within the body of

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.