This project generated data and developed a model for the kinetics of adsorption and desorption of 2,4-dinitrotoluene and nitroglycerin to and from the nitrocellulose matrix itself. Additionally, a model was developed for the partitioning of RDX, HMX, TNT, nitroglycerin, 2,4-dinitrotoluene, and nitroguanidine and mixtures of these munitions constituents to soils of varying physical and chemical characteristics. The team developed and used a chemical probe to determine the magnitude of clay mineral binding sites and ascertained the soil composition responsible for irreversible binding. The results were modeled using polyparameter partitioning models and models for irreversible bonding using soils spanning a variety of properties including soils typical of those found at operational ranges. Initial validation of the models developed in this project was provided by comparing model results to those determined in soil column studies.

Technical Approach

Partitioning of a mixture of munitions constituents to soil was studied by analyzing the effect of kinetics and reversible/resistant behavior of the munitions constituents on the adsorption-desorption to soils of varying physical and chemical characteristics. The data was collected from batch experiments conducted near 1:1 (weight/volume) soil to solution ratios, reflecting field conditions better than the dilute soil suspensions used in most studies. Adsorption was followed by multiple desorptions simulating rainy events to quantify the resistance of munitions constituents to desorption.

Models were built by using the measured dissolved and particulate concentrations during the adsorption-desorption of munitions constituents, and the total organic carbon and other sorption phases present in the soils. Key soil properties were selected in order to use the minimal number of input parameters providing reasonable accuracy of predictions, but reflecting a wide range of soil characteristics to provide better application of the model in the field. Twenty-five soils from different places in America and Europe were used to isolate the effects of independent physical and chemical characteristics that affect sorption.

One of the models analyzed in this research was the reversible/resistant model. The innovation was to apply it to the partitioning of mixtures of munitions constituents in different soil types taking into account the effect of kinetics and the electrolyte matrix in the adsorption and desorption steps. Results indicate that the model is sufficiently simple and flexible that it can be used in the adsorption/desorption of the mixture of munitions constituents studied, because the fitting of the data was excellent even given the variation of time of equilibration and desorption and of the soil matrices.

In addition to the reversible/resistant model, a multilinear sorption model was developed to predict partitioning of munitions constituents to soils by incorporating different sorption sites in addition to organic matter to improve the predictions. Clay minerals sites, cation exchange capacity, and oxalate extractable iron were included in the partitioning model. The clay sites were used in the multilinear model in alternative forms: the particle size fraction, the cation exchange capacity, and charge sites content. To determine the charge sites, a method based on cesium sorption was refined and applied. This is the first time that this probe has been used for a wide range of soils with various characteristics to develop a model incorporating specific sorption sites for the sorption of mixtures of munitions constituents.

To provide initial validation of the results obtained in the batch equilibration studies, a flow-through column study of TNT and RDX was conducted. The major objective was to determine if the resistant binding of TNT, but not of RDX, that was found in batch studies was also found in a flow-through column. This is the first application of the reversible-resistant theory to flow-through columns and by extension to percolation of compounds though soil in the field.

The presence of the nitrocellulose matrix complicates modeling of munitions constituents release from propellant and into soil solution. In addition to the dissolution of the soluble components, the role of adsorption to nitrocellulose, and desorption from nitrocellulose, need to be taken into account. No adsorption/desorption kinetic model of munitions constituents with the nitrocellulose matrix currently exists. In this study, the kinetics of sorption and desorption of 2,4-dinitrotoluene and nitroglycerin with nitrocellulose was measured. These data can then be compared to previous studies with assembled propellants to develop more appropriate models for sorption and desorption of 2,4-dinitrotoluene and nitroglycerin with nitrocellulose.


Partitioning of munitions constituents to a number of soils indicated that organic matter in the soil was the dominant site responsible for their partitioning. Partitioning to clay was modeled as a second sorption site, which could be modeled equally well by the clay size fraction of the soil or the cation exchange capacity. For soils containing little organic matter, inclusion of clay provided significant improvement except for nitroglycerin. The cesium exchange method gave better fitting of the multilinear model, but the improvement was not enough to recommend its use due to its cumbersome procedure. Further improvement was achieved by a trilinear model that incorporated partitioning to organic matter, clay, and oxalate extractable iron. Modeling of partitioning reversibility showed that the reversibility of partitioning varied considerably among the munition constituents. The irreversible partitioning was related to binding to organic matter in the soil. The flow-through column study indicated that a portion of the RDX remained on the soil following lengthy passage of electrolyte not containing RDX. Although the batch equilibration tests indicated complete reversibility, a small non-zero resistant portion would not have been distinguished in the test.


The study of partitioning to nitrocellulose provides new understanding of the mechanism that leads to prolonged leaching of munitions constituents from propellant particles. These results will permit new modeling of constituent release using a mechanistic rather than an observational approach.

The study of partitioning of munitions constituents to soil has provided partition coefficients based on 1, 2, and 3 site models based on results for a large number of soils. The models that include partitioning to organic matter, clay, and iron oxide provide improved estimates of partitioning applicable to soils that vary widely in their properties. Consideration of both resistant partitioning as well as reversible partitioning provides constants that can be used to provide more accurate prediction of aqueous concentrations of these chemicals when coupled with hydrological software. Results of a column flow experiment show that a small irreversibly bound fraction of chemical, not observable in batch testing, may sequester a significant amount of material.