Modeling the reduction kinetics of nitroaromatic compounds and munition constituents in environmental media

Abstract

Nitroaromatic compounds (NACs) and munition constituents (MCs) are common contaminants associated with weapons testing, training and manufacturing processes and are responsible for costly remediation processes at military sites across the country. Natural attenuation as a remediation strategy provides a more cost effective treatment strategy than the excavation and pump-and-treat methods currently employed, by utilizing the natural electron content of the subsurface for contaminant degradation. The key for proper design and implementation of natural attenuation in these systems is the prediction of the timescales over which these compounds are chemically reduced by different sources of electrons in the environment. The goal of this work is to break down the complex environmental media that comprise the subsurface into its most redox active component parts and build individual models capable of predicting NAC/MC reduction kinetics for those systems. ☐ For the NAC/MC reductions, linear free energy relationships (LFERs) were utilized to relate the second order rate constants of NAC/MC reduction to an energy descriptor for a particular reaction. The energy descriptors were calculated quantum chemically with density functional theory (DFT) methods, which allowed for both the inclusion of new, emerging contaminants, as well as the investigation of multiple potential reaction pathways, namely electron affinity (EA - the transfer of the first electron), and hydrogen atom transfer (HAT - the transfer of a hydrogen atom). The results showed that calculated aqueous phase HAT energies produced LFERs that were comparable to those produced with traditional measured one-electron reduction potentials (E1H) across multiple environmentally relevant reducing systems. ☐ Hydroquinones were selected as suitable reductants for the organic fraction of the subsurface due to their simple structures and their similar redox active moieties to natural organic matter (NOM). Energy descriptors for EA and HAT oxidation reactions were calculated with DFT methods in the same manner as the NAC/MC reduction half reactions, such that the two half reactions could be combined into a unified energy LFER spanning both the oxidants and reductants. The combined full reaction LFER with aqueous HAT energies successfully fit the reduction of seven NACs with six protonated states of hydroquinones, producing a single linear relationship with a root mean squared error (RMSE) of 0.56. ☐ A major component of the organic fraction of the redox active subsurface investigated was humic acid (HA). Since HA’s complex and uncharacterized nature made direct DFT energy calculations impossible, the HA system was modeled as a collection of quinone-like functional groups, whose HA redox characteristics were fit through global optimizations of experimental HA reduction profiles (e- titrations) and NAC/MC reduction curves ([NAC/MC] vs. time). A calibration dataset consisting of three HAs and two NAC/MCs was used to fit the relative quinone-like functional groups’ distribution, as well as the slopes and intercepts to the formal potential (E0’H) to HAT oxidation energy relationship that was obtained from the hydroquinone LFER. The resulting model successfully fit both the HA reduction profiles and the NAC/MC reduction curves, and was corroborated using a validation dataset consisting of an additional HA and NAC/MC. ☐ For the inorganic fraction of redox active environmental media, iron oxides were selected as the bulk contributor of labile electrons. Ferrous iron (Fe(II)) adsorbed to the surface of an oxide was chosen as the active reductant in the system based on several previous studies showing strong correlations between the concentration of sorbed Fe(II) and observed rate constants for NAC/MC reduction. A generalized two-layer cation surface sorption model was modified for predictions of Fe(II) adsorption as a function of pH, [Fe(II)initial] and oxide loading. A dataset consisting of four NAC/MCs and five iron oxides was used to fit oxide specific sorption characteristics from measured sorbed Fe(II) and surface area normalized rate constants for NAC/MC reduction. The resulting model produced predicted surface area normalized rate constants with better fits to measured values (RMSE = 0.412) than what is currently published in the literature.