## Doctoral Dissertations

Winter 1989

Dissertation

Chemistry

#### Degree Name

Doctor of Philosophy

C L Grant

#### Abstract

An analytical method was developed to determine the concentrations of HMX, RDX, TNB, DNB, tetryl, TNT and 2,4-DNT in soil. The method relies on solvent extraction with analysis by reversed-phase liquid chromatography.

The extraction step was studied in terms of process kinetics and recovery. Two solvents (acetonitrile and methanol) and four extraction techniques (Soxhlet, ultrasonic bath, mechanical shaker and homogenizer-sonicator) were compared. Ultrasonic bath extraction with acetonitrile was selected based on extraction kinetics, overall analyte recovery, sample throughput, and instability of analytes at elevated temperature. The rate of extraction of analytes from field-contaminated soil was shown to be much slower than from spiked soils, indicating it is unwise to develop extraction procedures based solely on spiked materials.

A number of possible separations were examined. Adequate separation of the seven analytes was achieved on an LC-18 column eluted with 1:1 methanol/water with a run time under 15 minutes. Confirmation of analyte identities was recommended on LC-CN, also eluted with 1:1 methanol/water. Elution orders on the two columns were quite different due to different mechanisms of separation.

Additional tests were conducted to assess various sample processing alternatives. Removal of particulates from soil extracts was achieved by dilution of extracts 1:1 with aqueous CaCl$\sb2$. This resulted in flocculation of suspended particles, which were then easy to remove by settling and filtration. Stock standards were stable for at least a year, working standards at least 28 days, and soil extracts at least two months. Care needs to be taken to ensure that air drying of soil, prior to extraction, is not conducted in direct sunlight; otherwise losses of TNT and an increase in photochemical transformation products will result.

The overall method provides linear calibration curves over a wide range of analyte concentrations. Detection limits ranged from 0.03 to 1.27 $\mu$g/g with no extract preconcentration. Recovery of spiked analyte was better than 80% for all analytes tested. The method was successfully tested in two collaborating laboratories.

COinS