An experimental design approach to optimise the determination of polycyclic aromatic hydrocarbons from rainfall water using stir bar sorptive extraction and high performance liquid chromatography-fluorescence detection.

Stir bar sorptive extraction (SBSE) followed by HPLC-fluorescence detection (FLD) was optimised for analysing 15 polycyclic aromatic hydrocarbons (PAHs) from water samples, especially rainfall water with low PAH content. The literature data described widely different experimental conditions for the extraction of PAHs by SBSE. A chemometric approach was therefore used to evaluate the statistically influential and/or interacting factors, among those described in the literature, and to find the best extraction and desorption conditions. Among six factors studied in a 2(6-2) fractional factorial design, only sample volume, extraction time and the interaction between both of them had significant effects on the PAH extraction recoveries. Optimal sample volume of 10 mL and extraction time of 140 min were obtained with a response surface design. For the desorption conditions, a Box-Behnken design showed that desorption time, temperature and PAH concentrations had significant effects. The best conditions were two successive desorptions with 100 microL of acetonitrile for 25 min at 50 degrees C. The optimised method was repeatable (RSD< or =5.3% for 50 ng L(-1) spiked water and < or =12.8% for 5 ng L(-1) spiked water), linear (R(2)> or =0.9956), with quantitative absolute recoveries (> or =87.8% for 50 ng L(-1) spiked water), and with the LOD between 0.2 and 1.5 ng L(-1). The optimised method was successfully applied to six-rainfall water samples collected in a suburban area. The total PAHs concentrations studied ranged from 31 to 105.1 ng L(-1). Seasonal variation was observed and on average three PAHs were at the highest concentrations (phenanthrene, fluoranthene and pyrene).