Investigation of Automated Sampling Techniques to Measure Total

Investigation of Automated Sampling Techniques to Measure Total Mercury in Stream-Water During Storm-Events Ami L. Riscassi*, Todd M. Scanlon Department of Environmental Sciences, University of Virginia Methods Introduction High-flow events (storms, snowmelt) are a dominant transport mechanism for total mercury (Hg. T) from the terrestrial to the aqueous environment. High-gradient headwater catchments are a primary source of downstream contamination because they store large pools of Hg in soils and sediments. Consistent, high-frequency event-sampling of headwater streams is rare, however, because of the unpredictability of high flows, remoteness of sites, and the difficulties associated with the ultra-clean sampling procedures (Shanley et al. , 2008). The use of automated sampling techniques with an ISCO® sampler has been demonstrated in several studies for trace metals and two studies for Hg (Nelson et al. , 2007; Lawson and Mason, 2001). The errors associated with using automated techniques for Hg. T sampling, however, has not been systematically evaluated in the literature. This research focuses on quantifying the relative errors associated with using an automated sampling system for streamwater Hg. T analysis and establish quality assurance guidelines for future sampling campaigns. Results All experiments were conducted with water from a local stream (Staunton River) during base flow conditions. Hg. T concentration of streamwater used in the experiments was 0. 29 ppt. Analysis of major cations/anions has not yet been completed, but mean concentrations calculated from 10 yrs of weekly baseflow samples indicate basic parameters are as follows: 6. 7 p. H, 18 μS conductivity, 86 μeq/L alkalinity, 44 μeq/L SO 4, 23 μeq/L Cl, and DOC 0. 7 mg/L. Figure 4. Staunton River, Va. Equipment test (water 5° C) : • ISCO programmed to rinse line 3 X before sample collection • ISCO samples streamwater (white vat) • Manual sample poured from vat directly to bottle (‘grab’ sample) • ISCO samples Hg-spiked streamwater (red vat) • Manual sample poured from vat directly to bottle (‘grab’ sample) • Rinse off outside of strainer with 50 ml streamwater (black vat) • ISCO samples streamwater (white vat). . . • Repeat for [Hg] spike: 0. 5, 5. 0, 25. 0, & 50. 0 ppt Equipment test Hg. T concentrations of streamwater sampled with an ISCO were elevated compared to ‘grab’ samples. Hg. T contamination ranged from 0. 00 to 0. 84 ppt. The relative percent difference (RPD) between grab and ISCO samples ranged from 0. 2 to 94%. The largest RPD’s are associated with baseflow samples immediately following a highly concentrated sample (25 or 50 ppt), a situation which is atypical in the natural environment. Empty bottle test Mean Hg. T contamination of uncapped empty bottles ranged from 0. 07 to 0. 25 ppt for the 3 and 7 day periods tested. Greater contamination was observed in open bottles with other samples (0. 29 vs 0. 10 ppt). Figure 5. Set-up of ISCO line test. Empty bottle test: Figure 2. ISCO automated sampler (right) at Paine River in Shenandoah National Park, Virginia. Figure 1. Grab sampling streamwater for Hg analysis (USGS photo taken from Shanley et al. , 2002). • 2 sets of uncapped empty bottles in ISCO without samples • 2 sets of uncapped empty bottles in ISCO with samples (0. 5 - 50. 0 ppt) • First bottle sets collected after 3 days and second sets after 7 -days, and all bottles filled with streamwater of known [Hg. T] Figure 9. Hg. T concentrations of streamwater poured into a unopened bottle, and poured into empty bottles that had been left open in an ISCO. Absolute difference (ppt) is labeled above the respective data. Uncapped sample test: Objectives 1) Quantify the cumulative errors associated with automated sampling for Hg. T analysis. Potential errors are determined for a) sequential sampling with the same strainer, sample & pump tubing b) empty sample bottles left open c) samples left open in an ISCO 2) Determine when it is appropriate to use automated sampling techniques • 2 sets of ISCO bottles filled with streamwater, and streamwater spiked with 0. 5, 5. 0, 25. 0, and 50. 0 ppt Hg • First bottle set collected after 3 -days and second at 7 -days. Figure 6. Set-up of ISCO bottle test for contamination/evasion. All samples were preserved within one hour of collection with 100% Br. Cl solution (5 ml/L). Samples were analyzed within a week of preservation for unfiltered Hg. T on a Tekran 2600 Cold Vapor Atomic Fluorescence Spectrophotometer (CVAFS) at the University of Virginia Lab. The relative percent difference (RPD) for field duplicates was <15%. Figure 7. Tekran 2600 and Dr. Babi. Materials All materials used in automated sampling are fluoropolymer (FP) (except pump tubing) and are cleaned by soaking in a heated (75 °C) 4 N HCl acid vat for 12 hrs. , rinsed 3 X with DI, dried under a clean bench, and double bagged until use. Sample bottles are cleaned according to EPA protocol. Figure 3. Teflon sample line and strainer. • • FP/Teflon® lined caps FP/Teflon® tubing (30 ft. ) FP /Teflon® strainer (made from tubing) FP /Kynar® connectors/fittings C-flex pump tubing for peristaltic pump ISCO® 2900 automated sampler glass bottles for ISCO samples FP/Teflon® bottles for grab samples Figure 8. Hg. T concentration of ‘grab’ and ISCO samples. The X-axis indicates the spiked Hg. T concentration in ppt. Absolute difference and (RPD) are labeled above the respective data. Literature Cited Lawson, N. M. , and R. P. Mason (2001), Concentration of mercury, methylmercury, cadmium, lead, arsenic, and selenium in the rain and stream water of two contrasting watersheds in Western Maryland, Water Research, 35(17), 40394052. Nelson, S. J. et al. (2007), Mass balances of mercury and nitrogen in burned and unburned forested watersheds at Acadia National Park, Maine, USA, Environmental Monitoring and Assessment, 126, 69 -80. Shanley, J. B. , et al. (2008), Comparison of total mercury and methylmercury cycling at five sites using the small catchment approach, Environmental Pollution, 154, 143 -154. Acknowledgements Amber ‘Dirty Hands’ Converse (UVA master’s student) provided analytical assistance and Kelly Hokanson (UVA undergraduate) provided field assistance. Susie Maben and Frank Deviney, of the Shenandoah Watershed Study (SWAS), provided historical streamwater chemistry data. Funding was provided through the NSF Career Grant, the DOE Graduate Assistantship in Areas of National Need (GAANN) fellowship, and the EPA Science to Achieve Results (STAR) fellowship. * Corresponding author: Ami Riscassi (alr 8 m@virginia. edu) Uncapped sample test All uncapped samples lost Hg. T ranging from 0. 1 to 4. 8 ppt. The RPD of [Hg. T] between ‘grab’ and ISCO samples ranged from 10 to 30%. Air temperature range: (25 -58°F) Figure 10. Hg. T concentrations of samples left uncapped for 3 and 7 -days in an ISCO. The X-axis indicates spiked [Hg. T]in ppt. Maximum absolute difference and RPD are labeled above the respective data. Conclusions The cumulative errors of using an ISCO automated streamwater sampler resulted in a RPD in Hg. T concentration ranging from: 8 -10% for 50 ppt, 11 -15% for 25 ppt, 2 -20% for 5 ppt, 10 -71% for 0. 8 ppt, and 5 -120% for 0. 3 ppt. ISCO automatic samplers can be used to measure streamwater for Hg. T when concentrations are above 5 ppt (220%). The EPA Method 1631 specifies that the RPD between field duplicates should be less than 20%. What’s next? (1) Repeat the bottle experiments in a heated environment (‘summertime’). (2) Concurrent manual stream sampling (middle of water column) and ISCO automated stream sampling during storm event.