GAS-CHROMATOGRAPHY COUPLED WITH ATOMIC-ABSORPTION SPECTROMETRY - A SENSITIVE INSTRUMENTATION FOR MERCURY SPECIATION

New instrumentation for the speciation of mercury is described, and is applied to the analysis of natural water samples. The separation of m ercury species is effected using gas chromatography of derivatized mer cury species on a widebore capillary column. The solvent is vented usi ng a bypass valve and the separated mercury species are pyrolysed on-l ine at 800 degrees C for production of mercury atoms. These are then d etected by atomic absorption spectrometry (AAS) at the 253.7 and 184.9 nm lines simultaneously in a quartz cuvette. The use of the 184.9 nm line provides a more than five-fold increase in sensitivity compared w ith the conventional 253.7 nm line and an absolute detection limit of 0.5 pg of mercury. The dynamic range of the combined analytical lines provides a linear response over more than three orders of magnitude. A number of organic compounds not containing mercury are also detected following pyrolysis, especially at the 184.9 nm line. These background species must not co-elute at the retention times for methyl- and inor ganic mercury, as otherwise a positive interference would result. By m aximizing the chromatographic resolution and minimizing the band broad ening in the cuvette by use of a make-up gas, the retention times of i nterest are freed from co-eluting background peaks. The instrumentatio n has been applied to the determination of ng 1(-1) concentrations of methyl- and inorganic mercury in Lake Constance, Germany and within th e Lake Constance drinking water supply organization, Bodenseewasserver sorgung (BWV). The accuracy for the sum of methyl- and inorganic mercu ry has been assessed by comparison with an independent method for tota l mercury based on AAS detection implemented at BWV. Relative detectio n limits using 1 litre water samples and 15 mi injections of the final hexane extract were 0.03 ng 1(-1) for methylmercury and 0.4 ng 1(-1) for inorganic mercury based on the 3j criterion.