Longevity of granular iron in groundwater treatment processes: corrosion product development

Permeable reactive barriers employing iron as a reactive surface have received extensive attention. A remaining issue, however, relates to their longevity. As an integral part of a long-term column study conducted to examine the influence of inorganic cosolutes on iron reactivity toward chlorinated solvents and nitroaromatic compounds, Master Builder iron grains were characterized via scanning and transmission electron microscopy, electron energy loss spectroscopy (EELS), micro-Raman spectroscopy, and X-ray diffraction. Prior to exposure to carbonate solutions, the iron grains were covered by a surface scale that consisted of fayalite (Fe2SiO4), wüstite (FeO), magnetite (Fe3O4), maghemite (-Fe2O3), and graphite. After 1100 days of exposure to solutions containing carbonate, other inorganic solutes, and organic contaminants, the wüstite, fayalite, and graphite of the original scale partially dissolved, and magnetite and iron carbonate hydroxide (Fe3(OH)2.2CO3) precipitated on top of the scale. Raman results indicate the presence of green rust (e.g., [Fe42+Fe23+(OH)12][CO3·2H2O]) toward the column outlet after 308 days of operation, although this mineral phase disappears at longer operation times. Grains extracted from a column exposed to a high concentration (20 mM) of sodium bicarbonate were more extensively weathered than those from columns exposed to 2 mM sodium bicarbonate. An iron carbonate hydroxide layer up to 100 m thick was observed. Even though EELS analysis of iron carbonate hydroxide indicates that this is a redox-active phase, the thickness of this layer is presumed responsible for the previously observed decline in the reactivity of this column relative to low-bicarbonate columns. A silica-containing feed resulted in reduced reactivity toward TCE. Grains from this column had a strong enrichment of silicon in the precipitates, although no distinct silica-containing mineral phases were identified. The substitution of 2 mM calcium carbonate for 2 mM sodium bicarbonate in the feed did not produce a measurable reactivity loss, as the discrete calcium carbonate precipitates that formed in this system did not severely restrict access to the reactive surface.