Much of the water supply industry has implemented online monitoring and warning systems for chemical or biological contaminations. When a contamination is detected, further investigations are necessary to confirm or to cancel the initial alarm. The proficiency of a non-target screening approach with solid phase extraction and GCMS in full-scan mode was examined. A selection of pesticides and industry chemicals was used for proficiency testing in extracts of raw and drinking water from Lake Constance. All total ion chromatograms (TIC) of extracted water samples showed a significant chemical background that have an adverse effect on compound identification. The TIC was evaluated with a two-dimensional search algorithm considering mass spectra similarity and retention index for identification, which was used an additional identification criteria to increase the confidence in identification. At a spiking level of 0.50 μg/l, up to 70% of the pollutants were unambiguously assignable. The sample pre-treatment was kept as simple as possible to reduce analysing time. A solid-phase extraction with extraction disks at flow rates up to 100 ml/min without any precedent filtration step reduces the sample pre-treatment time for a 1 l-sample below one hour. The recovery rates for most of the examined pollutants were above 60%.
Keywords: contamination warning system, gas chromatography mass spectrometry, non-target screening, qualitative analysis, retention index, solid-phase extraction