Gas chromatographic determination of N-nitrosodialkanolamines in herbicide Di- or trialkanolamine formulations

A modified method is presented to determine trace quantities of N-nitrosodiethanolamine (NDElA) and N-nitrosodiisopropanolamine (NDiPlA) in the triisopropanolamine (TiPlA) formulation of a mixture of picloram and 2,4-D. Aqueous sample is extracted with dichloromethane to remove organic interferences, and then the aqueous layer is passed sequentially through chloride anion exchange column, hydrogen cation exchange column, and Clin-Elut extraction tube. The final eluate, 10% acetone in ethyl acetate, is concentrated. The isolated nitrosamines are converted to the corresponding trimethylsilyl (TMS) derivatives and determined by gas chromatography (GC) on a DB1 column coupled with a thermal energy analyzer (GC-TEA). Eight samples of commercial TiPlA formulations are analyzed. Maximum detected levels of NDElA and NDiPlA were 0.6 and 0.9 ppm, respectively, expressed relative to total weight of active ingredients. Analysis of 13 samples of herbicide DElA formulation using a previously established method and a DB225 column gave NDElA results of 0.7-6.0 ppm. NDiPlA was not detected in those samples. Results are confirmed by GC-mass spectrometry (GC/MS) with oxygen negative chemical ionization (ONCI) detection. Detection limits for both nitrosamines are 0.05 or 0.07 ng (0.1 or 0.17 ppm) for GC-TEA detection, depending on the analytical columns used, and 20 pg (0.04 ppm) for GC/MS detection. Recoveries of NDElA are 87-109% for DElA formulation spiked at 2.6 and 3.9 ppm and 90-115% for TiPlA formulation spiked at 0.2-0.3 ppm. Similarly, recoveries of NDiPlA are 95.7-100% for the DElA formulation spiked at 0.24 and 0.48 ppm, and 82-118% for the TiPlA formulation spiked at 0.2-0.3 ppm.