Most explosives that occur as groundwater pollutants at Department of Defense (DoD) sites are nitro aromatic compounds (TNT, trinitrobenzene, and various di- and mono-nitrotoluenes) or nitramines (RDX, HMX, and tetryl). Under favorable conditions, most of these compounds react rapidly with zero-valent iron [Fe(0)], which suggests that permeable reactive barriers containing zero-valent iron might be useful for remediation of groundwater contaminated with explosives. However, all of the early work on explosives reduction by Fe(0) was done as batch experiments with nitroaromatic compounds, and this system produces solution-phase aromatic amines as the major products. Since these products are still substances of regulatory concern, full-scale implementation of Fe permeable reactive barriers (FePRBs) to treat explosives contaminated groundwater was delayed while complementary methods for treatment of the amines were investigated.

This work built upon research initiated under SERDP SEED project ER-1176. The overall goal of this project was to determine how operating conditions influence the distribution of products from the reduction of TNT, with the ultimate goal of defining optimal operating conditions for reaching remediation goals. Specifically there were two objectives: (1) to obtain a detailed characterization of contaminant removal using bench-scale reactors; and (2) to evaluate options for design and implementation of pilot-scale treatment zones of Fe(0).

In the first part of the project, a comprehensive mass balance of the products of TNT and RDX reaction with Fe(0) was completed. The kinetics of the reaction from adsorption of TNT and RDX were resolved, using steady-state and pulse elution experiments with small Fe(0) columns. Columns were loaded with TNT and the operating parameters were systematically varied to analyze potential desorption of TNT or products. Under the second part of the project, pilot-scale tests were performed with above-ground columns of granular Fe(0) using the pump and treat system at the Umatilla Weapons Depot. In addition, a reactive transport model was developed for use in designing full-scale, in situ FePRBs for explosives.

The disappearance of both TNT and RDX was shown to be rapid in the laboratory and ex situ field columns. Batch experiments performed with ¹⁴C-labelled TNT showed that the products of reaction with iron metal were partly sequestered on the metal (oxide) particle surfaces. Most of the bound residue could not be solubilized using a range of extraction procedures. However, batch studies over a range of experimental conditions showed considerable variability in the degradation products, while the primary TNT degradation product in column studies was TAT.

Ex situ column experiments at the Umatilla Chemical Weapons Depot were used to examine the viability of the Fe(0) under field conditions. This site was challenging because the dissolved oxygen concentration was high. A series of column experiments was conducted at the site: including columns with 10%, 20%, 30%, and 100% iron by volume in silica sand. In all of these cases, oxygen was rapidly scrubbed from the water by the iron. For the low iron cases, oxygen and RDX breakthrough occurred after several thousand pore volumes. The columns with higher iron content (30%, 100%) plugged after time. In at least some cases, this was due to particulates coming from the column. 

If the products of TNT reduction were retained on the Fe(0) surface, significant loss of reactivity (reaction-driven passivation) would be expected. However, it appears that under typical field conditions essentially all of the influent TNT will elute from the Fe(0)-bearing zone as TAT.

TAT appears to be fairly stable under conditions typical of an Fe(0)-bearing zone (i.e., no oxygen, high pH, and low natural organic matter). However, TAT is very labile to a number of transformations (hydrolysis, autoxidation, coupling) under typical aquifer conditions down-gradient from the Fe(0)-bearing zone (i.e., moderate pH, dissolved oxygen, and natural organic matter). Therefore, it is expected that an in situ Fe(0) permeable reactive barrier will fully reduce all aromatic nitro compounds, that most of these products will pass into the downgradient aquifer, and that these products will be readily removed by natural attenuation in the downgradient aquifer.