Interlaboratory evaluation of an off-line supercritical fluid extraction/infrared spectrometric method for determination of petroleum hydrocarbons in solid matrixes.

A collaborative study was conducted, with 14 laboratories participating, to determine the method accuracy and precision of the proposed U.S. Environmental Protection Agency Methods 3560 and 8440. These methods involve the extraction of petroleum hydrocarbons from solid matrixes with supercritical carbon dioxide at 340 atm and 80 degrees C for 30 min (dynamic), collection of the extracted materials in tetrachloroethene (Method 3560), and analysis of the extracts by infrared (IR) spectrometry (Method 8440). The study design was based on the AOAC blind replicate design with balanced replicates. The study samples consisted of 4 solid matrixes that had petroleum hydrocarbon contents ranging from 614 to 32,600 mg/kg. Each of the 4 matrixes was extracted in triplicate, and the extracts were analyzed with 2 different IR spectrometers. In addition, each of the participating laboratories extracted a sample of unspiked clay soil, the same clay soil spiked with corn oil and reference oil at 1000 mg/kg each, and the same clay soil wetted to 30% water content and spiked with motor oil at 10,000 mg/kg (the latter 3 samples were extracted only once). Results indicated that the overall method accuracy for concentrations ranging from 614 to 32,600 mg/kg was 82.9%; the mean recoveries of petroleum hydrocarbons for each of the 4 solid matrixes ranged from 77.9 to 107% for analyses performed with the Perkin-Elmer Fourier transform IR spectrometer and from 75.9 to 101% for analyses performed with the Buck-Scientific IR spectrometer; the differences between the 2 instruments on a sample-by-sample basis were less than 17% for the total petroleum hydrocarbon determinations. The interlaboratory method precisions (RSDR) appeared to be matrix-dependent and ranged from 17.3 to 45.4% for analyses performed with the Perkin-Elmer Fourier transform IR spectrometer and from 16.7 to 47.9% for the Buck-Scientific IR spectrometer. The intralaboratory method precisions (RSDr) appeared to be less matrix-dependent and ranged from 11.5 to 17.0% for analyses performed with the Perkin-Elmer FTIR spectrometer and from 11.1 to 18.2% for the Buck-Scientific IR spectrometer. Method accuracy and precision data are also presented for the 5 laboratories that used Isco supercritical fluid extraction systems and for the 7 laboratories that used vessels with volumes of 3.5 mL or less with different supercritical fluid extraction systems.