United States Environmental Protection Agency Environmental Monitoring Systems Laboratory Las Vegas, NV 89193-3478 Research and Development EPA/600/SR-94/119 September 1994 EPA Project Summary Evaluation of Sampling and Field-Filtration Methods for the Analysis of Trace Metals in Ground Water Karl F. Pohlmann, Gary A. Icopini, Richard D. McArthur, and Charlita G. Rosal Selected ground water sampling and field-filtration methods were evaluated to determine their effects on field pa- rameters and trace metal concentra- tions in samples collected under sev- eral types of field conditions. The study focused on conditions where traditional approaches may produce turbid samples, which often leads to filtration of suspended particles from the sample before laboratory chemical analysis. However, filtration may also remove colloidal particles that may be impor- tant to the transport of hydrophobic organic contaminants and trace met- als. The specific sampling and filtra- tion variables investigated in this study were (1) filtration with 0.45-^m or 5.0- |am pore size filters versus no filtra- tion; (2) sampling device, specifically, bladder pump, submersible-centrifugal pump, and bailer; and (3) sampling pump discharge rate during purging and sample collection using a "low" rate of 300 mL/min and a "moderate" rate of 1000 mL/min. Three field sites were visited: an active municipal solid waste landfill in Wisconsin, a closed solid waste landfill in Washington, and a site contaminated by industrial waste in Nevada. The evaluation included three wells each at the Wisconsin and Washington sites and two wells at the Nevada site. Filtration with 5.0-jjm fil- ters was conducted only at one well at each site. The effects of field filtration were most evident for the bailer, which often produced trace metal concentrations in unfiltered samples that were orders-of- magnitude higher than in 0.45-jjm-fil- tered samples. The largest differences occurred at the most turbid wells and in samples containing the highest par- ticle concentrations. Similar effects were observed in some samples col- lected by pumps from the most turbid wells, particularly the low yield well. For most pump sampling, however, dif- ferences in concentrations between 0.45-jjm-filtered and unfiltered samples were not significant and particle con- centrations were significantly lower than those produced by the bailer. Bail- ers caused more disturbance of the sampling zone than the three pumping methods as evidenced by measure- ments of field parameters and concen- trations of particles, major ions, and trace metals. Little variation was ob- served in the analytical determinations between the pumped samples but some variation existed in the field indicator parameters—primarily, temperature, dissolved oxygen, and turbidity. Trace metal concentrations in 0.45-|am-filtered samples were generally independent of sampling method, suggesting that these constituents were present as dissolved species and not associated with par- ticles, or associated with particles smaller than 0.4 |am. At wells where 5.0-jjm filtration was conducted, physi- cal and hydrochemical conditions re- sulted in minimal differences between trace metal concentrations in the 5.0- (jm-filtered, 0.45-|jin-filtered, and unfil- tered samples. This Project Summary was developed by EPA's Environmental Monitoring Systems Laboratory, Las Vegas, NV, to announce key findings of the research ------- project that is fully documented in a separate report of the same title (see Project Report ordering information at back). Introduction Historically, ground water contaminants were considered to be partitioned between two phases, a mobile phase composed of dissolved (aqueous) solutes in water trans- ported by natural ground water flow and a normally immobile solid phase composed of the matrix materials of the water-bear- ing zone. The action of purging and sam- pling a monitoring well installed in uncon- solidated materials may provide sufficient energy to suspend matrix materials that have accumulated in the sampling zone and well bore and incorporate them in ground-water samples. Inclusion of met- als associated with these normally immo- bile matrix particles may bias analytical determinations, leading to elevated and improbable concentrations of mobile con- taminants if suspended particle concen- trations are very high. As a result, ground water samples are commonly filtered in the field to remove these suspended par- ticles. Filtration has been considered par- ticularly necessary under turbid conditions where high particle (sediment) loadings might lead to significant analytical bias through inclusion of large quantities of matrix metals in the analysis. Alternatively, the presence of particles in samples might also bias analytical determinations through removal of metal ions from solution during shipment and storage as a result of inter- actions with particle surfaces. Unfortunately, indiscriminant use of field filtration ignores the presence of particles, known as colloids, in ground water that may exist between the extremes of sol- utes and sediments. Potential association of metals with colloids has important im- plications for the practice of field filtration because the boundary between the par- ticulate and dissolved has been opera- tionally defined at 0.45|om This boundary presumes that the component retained on a 0.45 |im filter represents suspended sol- ids, while the component that passed through the filter represents dissolved met- als. Collection of ground water samples for analysis of metals concentrations is re- quired under several U.S. environmental regulations, including CERCLA (Super- fund), RCRA Subtitle C (Hazardous Waste), and RCRA Subtitle D (Solid Waste). As a result, the debate regarding ground water metals samples impacts a wide range of sampling programs and a large number of sites, suggesting the need for further research. This study was un- dertaken to investigate how concentrations of trace metals were affected by selected methods of sample collection and field- filtration. The objectives of the study were to provide a survey of the impacts of the following aspects of ground-water sam- pling: 1) Impacts of sample collection method on determinations of field parameters. 2) Impacts of filtration with 0.45-|am or 5.0 |im pore size filters versus no filtration on trace metal concentrations. 3) Impacts of sampling device—specifi- cally, bailer, bladder pump, submers- ible-centrifugal pump (at a "low" dis- charge rate of 300 mL/min), and sub- mersible-centrifugal pump (at a "mod- erate" discharge rate of 1000 mL/min), on trace metal concentrations. 4) Impacts of sampling device on par- ticle size distribution and total con- centration. The study focused on sampling in con- ventional standpipe monitoring wells un- der conditions where traditional ap- proaches to sampling may produce turbid samples. Procedure The monitoring wells sampled were con- structed of polyvinyl chloride, and were 5.08 cm in diameter, with the exception of one 10.2 cm diameter well. The top of the well screens ranged from 2 to 19 m below ground surface, with well screen lengths of 0.6 to 6.0 m. The static water level ranged from 1 to 14 m below ground sur- face. Volumes of water within the well screens ranged from 1.2 to 50 L. Although certainly not representative of geologic and hydrogeochemical conditions at all solid waste landfills and hazardous waste sites, these sites provided typical field condi- tions where traditional approaches to ground-water sampling produce turbid samples. Four methods of collecting samples from conventional standpipe monitoring wells were evaluated using three types of sam- pling devices and pump discharge rates. These methods were utilized at eight of the nine wells. The first method used a dual-check valve bailer with a volume of approximately 0.4 L. Samples were trans- ferred from the bailer directly to the sample bottles for unfiltered samples or to a filtra- tion vessel for filtered samples. Com- pressed nitrogen gas was used to drive the samples through either membrane fil- ters or disposable cartridge filters. The second sampling method was a sub- mersible-centrifugal pump (CP1) operated at a flow rate of approximately 300 ml_/ min. Filtration was conducted in-line with disposable cartridge filters. The third method was a bladder pump (BP) oper- ated at a flow rate of approximately 500 mL/min at the Wisconsin site or 1000 mL/ min at the other sites. The fourth method was a submersible-centrifugal pump (CP2) operated at a flow rate of approximately 1000 mL/min. Discharge rates were mea- sured at ground surface and were con- trolled by the pump speed rather than by flow restrictors or valves. These discharge rates were used for both purging and sam- pling. Filtration for methods three and four was conducted in the same manner as for method two. The pumps and bailer were positioned to collect samples from about 0.6 m below the top of the well screen. Measurements of turbidity, dissolved oxygen (DO), temperature, electrical con- ductivity (EC), and pH of the pump dis- charge were made in-line, while measure- ments of these parameters for the bailer discharge were made off-line. Stabiliza- tion of these parameters provided an indi- cation of equilibrium between incoming ground water, the action of the sampler, and stagnant water in the well; thereby suggesting that purging was complete. The relative values of these parameters also provided a means for comparing the sam- pling methods with respect to their ability to minimize disturbance in the sampling zone. Estimates of particle size distribu- tion were determined gravimetrically by serial ultrafiltration using microfilters of 5.0 |j,m, 0.4 |j,m, 0.1 |j,m, and 0.03 |j,m pore size. Results and Discussion The results of the study demonstrate three important factors that influence the accuracy of field parameters measure- ments during sampling from conventional standpipe monitoring wells: measurement techniques, sampling method, and hydrau- lics of the well. Impacts related to mea- surement techniques were considered mi- nor because a single individual conducted all the field measurements and all proce- dures followed established protocol. In con- trast, sampling method and well hydrau- lics had impacts on values of field param- eters in some of the sampling events that masked all other factors. The relative dis- turbance in the sampling zone caused by a sampling method was most evident in the field measurements of turbidity and DO, particularly under low-yield conditions. When the discharge rate exceeded the well yield, the increasing hydraulic gradi- ent between the formation and the well mobilized large quantities of particles, ------- thereby elevating turbidity values. Contin- ued removal of water from the well dewa- tered the filter pack, leading to gravity drainage of pore water and sediments and continually increasing turbidity values. Bailer turbidity values were further elevated by the surging action of the bailer. El- evated DO values of the bailer and BP at 1 L/min in low-yield wells reflect the for- mation of a large air-water interface which increased the potential for oxygenation of incoming ground water as the filter pack was dewatered. The bailer caused addi- tional aeration of the samples as a result of the increased exposure to the atmo- sphere during sample collection and trans- fer. The lower discharge rate of 0.3 L/ min, which was generally closer to the well yield, resulted in less variability and more representative values of turbidity and DO, as well as lower purged volumes. Somewhat less variable results were observed between sampling methods in wells where purging and sampling rate did not exceed the well yield. Under these conditions, hydraulic gradients into the well were minimal, the filter pack was not de- watered, and turbidity was generally lower. The two pumping methods produced simi- lar values of most field measurements, while the surging action of the bailer pro- duced turbidity values that were approxi- mately two orders-of-magnitude higher than those produced by the pumps. Like- wise, DO values in bailed samples were elevated with respect to the pump values, an artifact of the bailing process. The pumps produced equilibrium DO and tur- bidity conditions with relatively low purged volumes, while the bailer produced high values of these parameters and did not reach equilibrium after greater purged vol- umes. Results of the study indicate that DO is sensitive to the purging process and further suggest that DO may be an important indicator of the volume required to remove stagnant water from the sam- pling system. As with DO, turbidity exhibited a strong dependence on sampling method. The highest turbidity values were obtained with the bailers, while the lowest turbidities were obtained with the pumps. Equilibration of turbidity, like DO and oxidation-reduction conditions (Eh), is often related to sample collection method. Values of pH showed little variation be- tween pump methods with most values falling within the range of +0.2 pH units for a given well. Bailed pH values were also within this range but were usually higher than the pumped values, possibly reflecting degassing of CO2 from the samples during collection and pH mea- surement. In addition, pH reached equilib- rium at lower purged volumes than all the other parameters, independent of sam- pling method. Although pH is an important indicator of the speciation of trace metals in ground water, the relatively uniform val- ues across devices at individual wells do not alone suggest that similar metals spe- cies might be present. In almost every case, samples collected by bailer contained higher particle con- centrations than those collected by the pumps, with the greatest differences oc- curring at the most turbid wells. Further- more, the size distribution of particles in most bailed samples was highly skewed toward larger particles, with over 96% larger than 0.45 |im, and generally over 93% larger than 5.0 urn. The quantities and sizes of these particles suggest that they were not mobile in ground water un- der natural flow conditions but were pri- marily the artifacts of well construction, development, and purging and were mo- bilized by agitation in the sampling zone caused by bailing. The particle size distri- bution in samples pumped from the most turbid (low-yield) wells were also skewed toward larger particles, but total particle concentrations were much lower than in the bailed samples. In the less turbid (high- yield) wells, total particle concentrations in pumped samples were orders-of-magni- tude lower than in bailed samples, reflect- ing the lower degree of agitation caused by the pumping methods. Also, particle sizes in the pumped samples were gener- ally more uniformly distributed; approxi- mately 50% of the particles were larger than 0.45 |om Differences in metal concentrations be- tween filtered and unfiltered samples were most evident in low-yield and highly-turbid wells, particularly when the samples were collected by bailer. In fact, several metals present in unfiltered bailed samples were below detection levels in the correspond- ing filtered samples. The large differences in concentration between filtered and un- filtered bailed samples reflect the associa- tion of metals with the high concentrations of artifactual particles entrained during bail- ing. For example, iron in the sampling zone likely existed as iron hydroxide par- ticles, particles containing elemental iron, and ferrous iron sorbed to particle sur- faces. Removal of the majority of particles during filtration therefore greatly reduced iron concentrations in the filtered samples. Other metals likely existed as aqueous species sorbed to particle surfaces, or as elemental components of particles origi- nating as aquifer solids, and their concen- trations were similarly reduced by filtra- tion. Additionally, ferrous iron may have oxidized and precipitated during bailing, transfer, and filtering of the samples, and then removed during filtration. Finally, the formation of a thick filter cake during filtra- tion of bailed samples likely reduced the effective pore size of the filter membrane, thereby blocking passage of some par- ticles smaller than 0.45 urn; this would further reduce the concentrations of asso- ciated metals in the sample. Trace metal concentrations in unfiltered samples pumped from low-yield and highly turbid wells were generally lower than in unfiltered samples bailed from the same wells. This reflects the lower degree of agitation associated with pumping and, as a result, the lower artifactual particle con- centrations. Removal of the larger par- ticles in the pumped samples did, how- ever, cause filtered samples to contain lower metal concentrations than unfiltered samples, though the differences in con- centration were much lower than in bailed samples. Unfiltered metal concentrations in samples pumped at 1 L/min were often slightly higher than in samples pumped at 0.3 L/min, but the concentrations in the filtered samples from both pumps were essentially the same. Furthermore, metals concentrations in filtered pumped samples did not differ significantly from those in filtered bailed samples. In less turbid and high-yield wells, unfil- tered bailed samples usually contained the highest metal concentrations of all samples, but the differences between these concentrations and concentrations in filtered samples were much smaller than for low-yield and turbid wells. Several met- als showed only slight differences between filtered and unfiltered results in bailed samples. These results reflect the lower proportion of artifactual particles removed during filtration as compared to the low- yield and turbid wells, but also are related to the metal speciation at each well. Dif- ferences between filtered and unfiltered pumped samples were minimal, and the concentrations were essentially the same as those in the filtered bailed samples, despite the variability in proportion of par- ticles smaller than 0.45 |jm This suggests that many metals existed primarily as dis- solved species and/or were associated with particles smaller than 0.45 \im in the less turbid and high-yield wells included in this study. Conclusions and Recommendations Field determinations of unstable param- eters DO and turbidity were the most sensitive to disturbance of the sampling zone, with values produced by bailing of- ten orders-of-magnitude higher than those produced by the pumps. Variations in in- ------- dicator parameters EC and pH were insig- nificant between the four sampling meth- ods, suggesting they were less related to disturbance of the sampling zone than DO and turbidity. Temperature showed little variation between the bladder pump and bailer but was highly elevated by the operation of the submersible centrifugal pump at low discharge rates. The relationship of turbidity to particle concentration and its sensitivity to the purg- ing process, relative to other indicator pa- rameters, suggests that turbidity may be a useful indicator of relative particle con- centrations between wells and of stabili- zation of particle concentrations during monitoring well purging. If mobile particles are thought to be important to transport of contaminants in ground water, use of field parameters such as pH, temperature, or EC as criteria for determining adequate sampling conditions may result in underpurging. The effects of field filtration on trace metal concentrations were most evident when a bailer was used to sample low- yield and/or turbid wells. Concentrations in unfiltered bailed samples were up to several orders-of-magnitude higher than in filtered bailed, filtered pumped, and un- filtered pumped samples. Elevated metal concentrations in unfiltered bailed samples reflected the entrainment of large quanti- ties of normally immobile artifactual par- ticles and their associated matrix metals, and unknown quantities of contaminant metals. Pumping at low to moderate rates in low-yield and/or turbid wells resulted in less agitation in the sampling zone, lower particle concentrations, and reduced ef- fects of field filtration on metal concentra- tions. The effects of field filtration were the least evident in high-yield wells and/or low- turbidity wells. Samples bailed from these wells exhibited much smaller differences between unfiltered and 0.45-|im-filtered samples. However, bailing clearly mobi- lized artifactual particles that caused el- evated metal concentrations in most un- filtered bailed samples. Samples collected by the bailer and immediately filtered ex- hibited trace metal concentrations that were roughly equivalent to those produced by the pumps and in-line filtration. Samples pumped from these wells exhib- ited virtually no differences between unfil- tered and filtered samples, reflecting the minimal entrainment of artifactual par- ticles larger than 0.45 |im during sampling at low to moderate pumping rates. Con- centrations in filtered samples bailed from high-yield wells and/or low-turbidity wells were generally equivalent to concentra- tions in pumped samples. This reflects the removal of larger, normally immobile arti- factual particles and associated metals from the bailed samples. Although the three sample-collection methods generally produced similar re- sults when samples from less turbid wells were filtered, the pumping methods pro- duced the most consistent overall results. Most metals showed little variation be- tween filtered and unfiltered pumped samples, reflecting the minimal agitation in the sampling zone and sample during purging and sample collection. Use of sub- mersible pumps at low speeds may re- duce the uncertainty in results when col- lecting samples of inorganic ground-water constituents that have the potential to as- sociate with particles in ground water. Since this study included only a limited number of wells at three sites, it does not represent the wide variety of geologic and hydrogeochemical conditions likely to be present at all solid waste or hazardous waste landfills. As a result, more informa- tion is required from a variety of sites regarding the presence of colloidal par- ticles and the importance of these par- ticles in the transport of trace metals and other contaminants in ground water. A better understanding of colloidal transport processes in ground-water environments could be gained from research focused on describing hydrogeochemical conditions and colloid size distribution, composition, movement, and association with trace metals at a variety of solid waste and hazardous waste sites. The information in this document has been funded wholly or in part by the United States Environmental Protection Agency under Cooperative Agreement Number CR815774 to the Water Resources Cen- ter of the Desert Research Institute. It has been subjected to the Agency's peer and administrative review, and it has been ap- proved for publication as an EPA docu- ment. Mention of trade names or commer- cial products does not constitute endorse- ment or recommendation for use. ------- Charlita G. Rosal (also the EPA Project Officer, see below) is with the Environmen- tal Monitoring Systems Laboratory, Las Vegas, NV 89193-3478, K. F. Pohlman, G.A. Icopini, and R.D. McArthurare with Desert Research Institute, University of Nevada at Las Vegas, Las Vegas, NV 89119. The complete report, entitled "Evaluation of Sampling and Field-Filtration Methods for the Analysis of Trace Metals in Ground Water," (Order No. PB94-201993/AS; Cost: $19.50, subject to change) will be available only from: National Technical Information Service 5285 Port Royal Road Springfield, VA 22161 Telephone: 703-487-4650 The EPA Project Officer can be contacted at: Environmental Monitoring Assessment Laboratory U. S. Environmental Protection Agency Las Vegas, NV 89193-3478 United States Environmental Protection Agency Center for Environmental Research Information Cincinnati, OH 45268 Official Business Penalty for Private Use $300 BULK RATE POSTAGE & FEES PAID EPA PERMIT NO. G-35 EPA/600/SR-94/119 ------- |