The objective of the work carried out was to recommend protocols for the collection, filtration (0.45 microm) and preservation of surface water samples for the subsequent determination of total 'dissolved' Hg. Cold vapour (CV) ICP-MS was employed to determine Hg; samples were acidified to a strength of 4 mol l(-1) HCl and 1% NaBH4 was used as the reducing agent in-line. Four types of 125 ml bottles were studied (Teflon, fluorinated ethene propene copolymer, FEP; high density polyethylene, HDPE; polyethylene terephthalate copolyester, PET; polypropylene, PP), together with three cleaning methods (EPA Methods 1631, 1638 and a rinse with reverse osmosis deionised water, 'MilliQ'). The transmission properties of the four materials were also studied to evaluate the potential for contamination from atmospheric Hg0. Results of this bottle study (n = 195), all below the detection limit of 0.5 ng l(-1), indicate that the bottles of choice, from an economic and time-saving perspective, are HDPE and PP, the latter being preferable if the sample is to be stored in a contaminated atmosphere. The bottles would be used on a once-only basis, negating the need for labourious and costly cleaning on repeat use. A simple rinse with MilliQ water would suffice prior to use. Twelve 0.45 microm filter systems (mostly Millipore and Gelman) were studied for (a) their potential Hg contamination properties and (b) their retention of Hg, possibly in colloidal form, during filtration. Ottawa River water, spiked at 50 ng l(-1) Hg, was used as a control sample. Again blank values were all negative, indicating contamination was not a concern but different recoveries of Hg were obtained across the different systems. The optimum systems to use, in that they provided maximum recovery (ca 80%) of Hg, are the Millipore Sterivex capsule and the Millipore Millex disc, both based on the hydrophilic Durapore membrane. The lowest recoveries (23-36%) were found with the Gelman AquaPrep systems and the Millipore HN nylon filter. Four control samples, Ottawa, Rideau and Gatineau Rivers, and a MilliQ blank, were used to study three preservation approaches, in the media: 0.5% BrCl, 2% HCl and 0.04% K2Cr2O7 in 1% HNO3. Mercury was stable for 28 days in a medium of 0.5% BrCl in all four samples but the other two media showed a loss of up to ca 20% Hg over this time frame, the loss being sample and time dependent. This may be a species transformation which is not detected by CV-ICP-MS from an HCl medium but this requires further investigation. Mercury was stable over the 28 days in the spiked blank MilliQ sample for all three preservation media.
