The Essential Handbook of Ground Water Sampling - Chapter 5 pptx

Filtration involves passing a raw or bulk ground-water sample directly through a filter medium of a prescribed filter pore size either under... What Filter Pore Size to Use The most comm

Ground-Water Sample Pretreatment: Filtration and Preservation

Gillian L Nielsen and David M Nielsen

CONTENTS

Sample Pretreatment Options 131

Sample Filtration 131

What Fiter Pore Size to Use 132

Functions of Filtration 134

Which Parameters to Filter 134

Sources of Error and Bias in Filtration 136

Filtration Methods and Equipment 137

Filter Preconditioning 139

Sample Preservation 141

Objectives of Sample Preservation 141

Physical Preservation Methods 142

Chemical Preservation Methods 145

References 149

Sample Pretreatment Options

Another group of parameter-specific field protocols that must be evaluated and included

in the SAP are methods for sample pretreatment, including sample filtration and physical and chemical preservation Sample pretreatment must be performed at the wellhead at the time of sample collection to ensure that physical and chemical changes do not occur in the samples during the time that the sample is collected and after the sample container has been filled and capped ASTM International has published Standard Guides that address both types of sample pretreatment ASTM Standard D 6564 (ASTM, 2006a) provides a detailed guide for field filtration of ground-water samples, and ASTM Standard D 6517 (ASTM, 2006b) discusses physical and chemical preservation methods for ground-water samples Each type of sample pretreatment is discussed subsequently

Sample Filtration

Ground-water sample filtration is a sample pretreatment process implemented in the field for some constituents, when it is necessary to determine whether a constituent is truly ‘‘dissolved’’ in ground water Filtration involves passing a raw or bulk ground-water sample directly through a filter medium of a prescribed filter pore size either under

negative pressure (vacuum) or under positive pressure Particulates finer than the filterpore size pass through the filter along with the water to form the filtrate, which issubmitted to the laboratory for analysis Particulate matter larger than the filter pore size

is retained by the filter medium In the case of most ground-water monitoring programs,this material is rarely analyzed, although it may be possible to analyze the retainedfraction for trace metals or for some strongly hydrophobic analytes such as PCBs orPAHs Figure 5.1 illustrates a common vacuum filtration setup, andFigure 5.2illustratesone form of positive pressure filtration

What Filter Pore Size to Use

The most common method for distinguishing between the dissolved and particulatefractions of a sample has historically been filtration with a 0.45 mm filter (see, e.g., U.S.EPA, 1991) The water that passes through a filter of this pore size has, by default, becomethe operational definition of the dissolved fraction, even though this pore size does notaccurately separate dissolved from colloidal matter (Kennedy et al., 1974; Wagemann andBrunskill, 1975; Gibb et al., 1981; Laxen and Chandler, 1982) Some colloidal matter issmall enough to pass through this pore size, but this matter cannot be considereddissolved For this reason, Puls and Barcelona (1989) reported that the use of a 0.45 mmfilter was not useful, appropriate, or reproducible in providing information on metalssolubility in ground-water systems and that this filter size was not appropriate fordetermining truly dissolved constituents in ground water

The boundary between the dissolved phase and the colloidal state is transitional There

is no expressed lower bound for particulate matter and no clear cutoff point to allowselection of the optimum filter pore size to meet the objective of excluding colloidal

FIGURE 5.1

A vacuum filtration system used for ground-water samples This practice is not encouraged.

particles from the sample The best available evidence indicates that the dissolved phaseincludes matter that is less than 0.01 mm in diameter (Smith and Hem, 1972; Hem, 1985),suggesting that a filter pore size of 0.01 mm is most appropriate However, filters withsuch small pore sizes are subject to rapid plugging, especially if used in highly turbidwater, and are not practical to use in the field Kennedy et al (1974) and Puls et al (1991)provide a strong case for the use of a filter pore size of 0.1 mm for field filtration toallow better estimates of dissolved metal concentrations in samples Puls et al (1992) andPuls and Barcelona (1996) also recommend the use of 0.1 mm (or smaller) filters fordetermination of dissolved inorganic constituents in ground water Such filters areconsiderably more effective than filters with larger pore sizes (e.g., 0.45 or 1.0 mm) interms of removing fine particulate matter These filters are widely available and practicalfor use in the field for most situations, although in some highly turbid water, filterplugging may make the filtration process difficult and protracted All factors considered,0.1 mm field filtration, although it is a compromise, appears to offer the best opportunityfor collecting samples that best represent the dissolved fraction.

Yao et al (1971) indicate that colloids larger than several microns in diameter areprobably not mobile in aquifers under natural ground-water flow conditions due togravitational settling Puls et al (1991) also suggest that colloidal materials up to 2 mm aremobile in ground water systems With the upper bound for colloidal matter described bymany investigators as being between 1.0 and 10 mm, it seems reasonable to suggest that

a filter pore size of 10 mm would include all potentially mobile colloidal material andexclude the larger, clearly nonmobile artifactual fraction However, it should be noted thatusing this filter pore size, artifactual colloidal material that is finer than 10 mm in diameterwill be included in the sample Although this filter pore size is a compromise, it will lead

to conservative estimates of total mobile contaminant load while excluding at least aportion of the particulate matter that is artifactual in nature The collection and analysis ofboth filtered and unfiltered samples is sometimes suggested as a means of discriminatingbetween natural and artifactual colloidal material or between dissolved and colloidalcontaminant concentrations

FIGURE 5.2

A positive-pressure filtration system is a better option to use for ground-water samples Note the removal of sediment achieved by the cartridge filter.

Functions of Filtration

Historically, filtration of ground-water samples has served several important functions inground-water sampling programs Filtration helps minimize the problem of data bounce,which commonly results from variable levels of suspended particulate matter in samplesbetween sampling events and individual samples, making trend analysis and statisticalevaluation of data more reliable In addition, by reducing suspended particle levels,filtration makes it easier for laboratories to accurately quantify metals concentrations insamples Perhaps most importantly, filtration of samples makes it possible to determineactual concentrations of dissolved metals in ground water that have not been artificiallyelevated as a result of sample preservation (acidification), which can leach metals fromthe surfaces of artifactual or colloidal particles (Nielsen, 1996) The assumption that theseparation of suspended particulates from water samples to be analyzed eliminates onlymatrix-associated (artifactual) constituents may often be incorrect (EPRI, 1985a; Feld et al.,1987), as at least some potentially mobile natural colloidal material will be retained onmost commonly used filter pore sizes

Filtration is often performed as a post-sampling ‘‘fix’’ to exclude from samples anyparticulate matter that may be an artifact of poor well design or construction,inappropriate sampling methods (use of bailers, inertial-lift pumps, or high-speed,high-flow-rate pumps), or poor sampling techniques (agitating the water column in thewell) Filtration may be considered particularly important where turbid conditionscaused by high particulate loading might lead to significant positive bias throughinclusion of large quantities of matrix metals in the samples (Pohlmann et al., 1994).Alternatively, as discussed earlier, the presence of artifactual particulate matter insamples may also negatively bias analytical results through removal of metal ions fromsolution during sample shipment and storage as a result of interactions with particlesurfaces However, filtration is not always a valid means of alleviating problemsassociated with artifactual turbidity, as it often cannot be accomplished without affectingthe integrity of the sample in one way or another

Which Parameters to Filter

During the planning phase of a ground-water sampling program, each parameter to beanalyzed in ground-water samples should be evaluated to determine its suitability forfield filtration and the most suitable filtration medium As a general rule of thumb,parameters that are sensitive to the following effects of filtration should not be filtered inthe field:

. Pressure changes that would result in degassing or loss of volatile constituents

. Temperature changes

. Aeration and agitation that may occur during filtration processes

Table 5.1 presents a summary of parameters for which filtration may be used and ofparameters for which filtration should not be used in the field

Samples to be analyzed for alkalinity must be field filtered if significant particulatecalcium carbonate is suspected in samples, as this material is likely to impact alkalinitytitration results (Puls and Barcelona, 1996) Care should be taken in this instance,however, as filtration may alter the CO2content of the sample and, therefore, affect theresults

Filtration is not always appropriate for ground-water sampling programs If theintent of filtration is to determine truly dissolved constituent concentrations (e.g., for

geochemical modeling purposes), the inclusion of colloidal matter less than 0.45 mm inthe filtrate will result in overestimated values (Wagemann and Brunskill, 1975; Bergseth,1983; Kim et al., 1984) This result is often obtained with Fe and Al, where ‘‘dissolved’’values are obtained which are thermodynamically impossible at the sample pH (Puls

et al., 1991) Conversely, if the purpose of sampling is to estimate total mobilecontaminant load, including both dissolved and naturally occurring colloid-associatedconstituents, significant underestimates may result from filtered samples, due to theremoval of colloidal matter that is larger than 0.45 mm (Puls et al., 1991) A number ofresearchers have demonstrated that some metal analytes are associated with colloids thatare greater than 0.45 mm in size (Gschwend and Reynolds, 1987; Enfield and Bengtsson,1988; Ryan and Gschwend, 1990) and that these constituents would be removed by0.45 mm filtration Kim et al (1984) found the majority of the concentrations of rare earthelements to be associated with colloidal species that passed through a 0.45 mm filter.Wagemann and Brunskill (1975) found more than twofold differences in total Fe and Alvalues between 0.05 and 0.45 mm filters of the same type Some Al compounds, observed

by Hem and Roberson (1967) to pass through a 0.45 mm filter, were retained on a 0.10 mmfilter Kennedy et al (1974) found errors of an order of magnitude or more in thedetermination of dissolved concentrations of Al, Fe, Mn, and Ti using 0.45 mm filtration as

an operational definition for ‘‘dissolved.’’ Sources of error were attributed to passage offine-grained clay particles through the filter

Evidence from several field studies (Puls et al., 1992; Puls and Powell, 1992; Backhus

et al., 1993; McCarthy and Shevenell, 1998) indicates that field filtration does not effectivelyremedy the problems associated with artifactual turbidity in samples These and otherstudies indicate that filtration may cause concentrations of some analytes to decreasesignificantly, due to removal of colloidal particles that may be mobile under natural flowconditions Puls and Powell (1992) noted that 0.45 mm filtered samples collected with abailer had consistently lower As concentrations than samples obtained using low-flow-ratepumping They suggested that the difference may have been due to filter clogging fromexcessive fines reducing the effective pore size of the filters or adsorption onto freshlyexposed surfaces of materials brought into suspension by bailing Puls et al (1992) foundthat high-flow-rate pumping resulted in large differences in metals concentrationsbetween filtered and unfiltered samples, with neither value being representative of valuesobtained using low-flow-rate sampling Ambiguous sampling results found by McCarthyand Shevenell (1998) were attributed to analytical values for metals obtained using low-flow sampling that fell between filtered and unfiltered values from samples collected using

TABLE 5.1Analytical Parameter Filtration Recommendations

Examples of parameters that may be field filtered Alkalinity

Trace metals Major cations and anions Examples of parameters that should not be filtered VOCs

TOC TOX Dissolved gases (e.g., DO and CO2)

‘‘Total’’ analyses (e.g., total arsenic) Low molecular weight, highly soluble, and nonreactive constituents Parameters for which ‘‘bulk matrix’’ determinations are required Source: U.S EPA, 1991.

bailing or high-flow-rate pumping Discrepancies in analytical values for some metals(Al and Fe) exceeded an order of magnitude in this study They determined that filtration

of turbid samples may have occluded pores in filters, leading to removal of colloidalparticles that may be representative of the load of mobile contaminants in ground water.Puls and Barcelona (1989) also point to the removal of potentially mobile species as aneffect of filtration, indicating that filtration of ground-water samples for metals analysiswill not provide accurate information concerning the mobility of metal contaminants

If the objective of a ground-water sampling program is to determine the exposure risk

of individuals who consume ground-water from private water supply wells, filtration ofthose samples would not produce meaningful results To make this type of exposure riskdetermination, it is important to submit samples for analysis that are representative ofwater as it is consumed, and, because most people do not have 0.45 mm filters at theirtaps, unfiltered samples should be collected In addition, it is important to remember thatMCL and MCLG values set for drinking-water standards are based on unfiltered samples

Sources of Error and Bias in Filtration

The very act of filtration can introduce significant sources of error and bias into the resultsobtained from analysis of sample filtrate (Braids et al., 1987) Some of these changes insample chemistry result from pressure changes in the sample, as well as sample contactwith filtration equipment and filter media It is critical to evaluate the suitability offiltration on a parameter-specific basis and to carefully select filtration methods,equipment, and filtration media when developing site-specific filtration protocols tominimize sample bias caused by filtration The following sources of negative and positivesample bias need to be considered:

. Potential for negative bias to occur due to adsorption of constituents from thesample (U.S EPA, 1991; Horowitz et al., 1996) For example, Puls and Powell(1992) found that in-line polycarbonate filters adsorbed Cr onto the surface ofthe filter medium, resulting in an underestimation of Cr concentrations in theground-water samples being filtered

. Potential for positive bias to occur due to desorption or leaching of constituentsinto the sample (Jay, 1985; Puls and Barcelona, 1989; Puls and Powell, 1992;Horowitz et al., 1996) In the Puls and Powell (1992) study, K was observed toleach from nylon filters that were not adequately preconditioned prior to use

. Removal of particulates smaller than the original filter pore size due to filterloading or clogging as filtered particles accumulate on the filter surface(Danielsson, 1982; Laxen and Chandler, 1982) or variable particle size retentioncharacteristics (Sheldon, 1965; Sheldon and Sutcliffe, 1969)

. Removal of particulate matter with freshly exposed reactive surfaces, throughparticle detachment or disaggregation, that may have sorbed hydrophobic,weakly soluble, or strongly reactive contaminants from the dissolved phase (Pulsand Powell, 1992) This material itself may have been immobile prior to initiation

of sampling and mobilized by inappropriate sampling procedures

. Removal of solids (metal oxides and hydroxides) that may have precipitatedduring sample collection (particularly where purging or sampling methods thatmay have agitated or aerated the water column are used) and any adsorbedspecies that may associate with the precipitates Such precipitation reactions canoccur within seconds or minutes (Reynolds, 1985; Grundl and Delwiche, 1992;Puls et al., 1992), and the resultant solid phase possesses extremely high reactivity

(high capacity and rapid kinetics) for many metal species (Puls and Powell, 1992).Most metal adsorption rates are extremely rapid (Sawhney, 1966; Posselt et al.,1968; Ferguson and Anderson, 1973; Anderson et al., 1975; Forbes et al., 1976;Sparks et al., 1980; Benjamin and Leckie, 1981; Puls, 1986; Barrow et al., 1989).Additionally, increased reaction rates are generally observed with increasedsample agitation.

. Exposure of anoxic or suboxic ground water (in which elevated levels of Fe2
aretypically present) to atmospheric conditions during filtration can also lead tooxidation of samples, resulting in formation of colloidal precipitates and causingremoval of previously dissolved species (NCASI, 1982; EPRI, 1987; Puls andEychaner, 1990; Puls and Powell, 1992; Puls and Barcelona, 1996) The precipita-tion of ferric hydroxide can result in the loss of dissolved metals due to rapidadsorption or co-precipitation potentially affecting As, Cd, Cu, Pb, Ni, and Zn(Kinniburgh et al., 1976; Gillham et al., 1983; Stoltzenburg and Nichols, 1985; Kentand Payne, 1988)

. During sample filtration, care should be taken to minimize sample handling to theextent possible to minimize the potential for aeration If sample transfer vesselsare used, they should be filled slowly and filtration should be done carefully tominimize sample turbulence and agitation Stoltzenburg and Nichols (1986)demonstrated that the use of sample transfer vessels during filtration impartedsignificant positive bias for DO and significant negative bias for dissolved metalconcentrations For this reason, the use of transfer vessels is discouraged In-linefiltration is preferred because of the very low potential it poses for samplechemical alteration

Filtration Methods and Equipment

After a decision is made to field filter ground-water samples to meet DQOs for aninvestigation, decisions must be made regarding selection of the most appropriate fieldfiltration method The ground-water sample filtration process consists of several phases:(1) selection of a filtration method; (2) selection of filter media (materials of construction,surface area, and pore size); (3) filter preconditioning; and (4) implementation of fieldfiltration procedures Information on each part of the process must be presented in detail

in the SAP to provide step-by-step guidance for sampling teams to implement in the field

A wide variety of methods are available for field filtration of ground-water samples Ingeneral, filtration equipment can be divided into positive-pressure filtration and vacuum(negative pressure) filtration methods, each with several different filtration mediumconfigurations As discussed previously, ground-water samples undergo pressure changes

as they are brought from the saturated zone (where ground water is under pressure greaterthan atmospheric pressure) to the surface (where it is under atmospheric pressure),potentially resulting in changes in sample chemistry The pressure change that occurswhen the sample is brought to the surface may cause changes in sample chemistry, whichinclude loss of dissolved gases and precipitation of dissolved constituents such as metals.When handling samples during filtration operations, additional turbulence and mixing ofthe sample with atmospheric air can cause aeration and oxidation of Fe2
to Fe3
Fe3
rapidly precipitates as amorphous iron hydroxide and can adsorb other dissolved tracemetals (Stolzenburg and Nichols, 1986) Vacuum filtration methods further exacerbatepressure changes and changes due to sample oxidation For this reason, positive-pressurefiltration methods are preferred (Puls and Barcelona, 1989, 1996; U.S EPA, 1991)

Table 5.2 presents equipment options available for positive pressure and vacuumfiltration of ground-water samples.

When selecting a filtration method, the following criteria should be evaluated on a by-site basis:

site-. Possible effect on sample integrity, considering the potential for the following tooccur:

a Sample aeration, which may result in sample chemical alteration

b Sample agitation, which may result in sample chemical alteration

c Change in partial pressure of sample constituents resulting from application ofnegative pressure to the sample during filtration

d Sorptive losses of components from the sample onto the filter medium orcomponents of the filtration equipment (e.g., flasks, filter holders, etc.)

e Leaching of components from the filter medium or components of the filtrationequipment into the sample

. Volume of sample to be filtered

. Chemical compatibility of the filter medium with ground-water sample chemistry

. Anticipated amount of suspended solids and the attendant effects of particulateloading (reduction in effective filter pore size)

. Time required to filter samples Short filtration times are recommended tominimize the time available for chemical changes to occur in the sample

. Ease of use

. Availability of an appropriate medium in the desired filter pore size

. Filter surface area

. Use of disposable versus nondisposable equipment

. Ease of cleaning equipment if not disposable

. Potential for sample bias associated with ambient air contact during samplefiltration

. Cost, evaluating the costs associated with equipment purchase price, expendablesupplies and their disposal, time required for filtration, time required fordecontamination of nondisposable equipment, and QC measures

TABLE 5.2Examples of Equipment Options for Positive-Pressureand Vacuum Field Filtration of Ground-Water Samples

Positive-pressure filtration equipment In-line capsules

 / Attached directly to a pumping device discharge hose

 / Attached to a pressurized transfer vessel

 / Attached to a pressurized bailer Free-standing disk filter holders Syringe filters

Zero headspace extraction vessels Vacuum filtration equipment Glass funnel support assembly

The filtration method used for any given sampling program should be documented inthe site-specific SAP and should be consistent throughout the life of the samplingprogram to permit comparison of data generated If an improved method of filtration isdetermined to be appropriate for a sampling program, the SAP should be revised in lieu

of continuing use of the existing filtration method In this event, the effect oncomparability of data needs to be examined and quantified to allow proper data analysisand interpretation Statistical methods may need to be used to determine the significance

of any changes in data resulting from a change in filtration method

Filtration equipment and filter media are available in a wide variety of materials ofconstruction Materials of construction should be evaluated in conjunction withparameters of interest being filtered with particular regard to minimizing sources ofsample bias, such as adsorption of metals from samples (negative bias) or desorption orleaching of constituents into samples (positive bias) Materials of construction of both thefilter holder or support and the filter medium itself need to be carefully selected based oncompatibility with the analytes of interest (Puls and Barcelona, 1989) Filter holders thatare made of steel are subject to corrosion and may introduce artifactual metals intosamples Glass surfaces may adsorb metals from samples

Table 5.3presents a summary of the most commonly used filtration media available forfield filtration of water samples The potential for sample bias for these filter mediamaterials is variable, therefore, filter manufacturers should be consulted to determinerecommended applications for specific filtration media and for guidelines on the mosteffective preconditioning procedures

Large-diameter filter media (
47 mm) are recommended for ground-water samplefiltration (Puls and Barcelona, 1989) Because of the larger surface area of the filter,problems of filter clogging and filter pore size reduction are minimized High-capacity in-line filters have relatively large filter media surface areas, which may exceed 750 cm2.This can improve the efficiency of field sample filtration

Filter Preconditioning

Filter media require proper preconditioning prior to sample filtration (Jay, 1985; U.S EPA,1995; Puls and Barcelona, 1996; ASTM, 2006a) The purposes of filter preconditioning are:(1) to minimize positive sample bias associated with residues that may exist on the filtersurface or constituents that may leach from the filter, and (2) to create a uniform wettingfront across the entire surface of the filter to prevent channel flow through the filter andincrease the efficiency of the filter surface area Preconditioning the filter medium maynot completely prevent sorptive losses from the sample as it passes through the filtermedium

In most cases, filter preconditioning should be done at the wellhead immediately prior

to use (Puls and Barcelona, 1989) In some cases, filter preconditioning must be done in alaboratory prior to use (e.g., GFuF filters must be baked prior to use) Some manufacturers

‘‘preclean’’ filters prior to sale These filters are typically marked ‘‘precleaned’’ on filterpackaging and provide directions for any additional field preconditioning required prior

to filter use

The procedure used to precondition the filter medium is determined by the following:(1) the design of the filter (i.e., filter capsules or disks); (2) the material of construction ofthe filter medium; (3) the configuration of the filtration equipment; and (4) the parameters

of concern for sample analysis Filtration medium manufacturers’ instructions should befollowed prior to implementing any filter preconditioning protocols in the field to ensurethat proper methods are employed and to minimize potential bias of filtered samples

These instructions will specify filter-specific volumes of water or medium-specificaqueous solutions to be used for optimum filter preconditioning.

The volume of water used in filter preconditioning is dependent on the surface area ofthe filter and the medium’s ability to absorb liquid Many filter media become fragilewhen saturated and are highly subject to damage during handling Therefore, saturatedfilter media should be handled carefully and are best preconditioned immediately prior

to use in the field

Disk filters (also known as plate filters) should be preconditioned as follows: (1) holdthe edge of the filter with filter forceps constructed of materials that are appropriate forthe analytes of interest; (2) saturate the entire filter disk with manufacturer-recom-mended, medium-specific water (e.g., distilled water, deionized water, or sample water)while holding the filter over a containment vessel (not the sample bottle or filter holder)

to catch all run-off; (3) then place the saturated filter on the appropriate filter stand orholder in preparation for sample filtration; (4) complete assembly of the filtrationapparatus; (5) pass the recommended volume of water through the filter to completepreconditioning; (6) discard preconditioning water; and (7) begin sample filtration using aclean filtration containment vessel or flask When preconditioning disk filters, care should

be taken not to perforate the filter The filter medium should not be handled withanything other than filter forceps Otherwise, there may be a reduction in the porosity andpermeability of the filter medium In addition, care should be taken to avoid exposure ofthe filter medium to airborne particulates to minimize introduction of contaminants ontothe filter surface

Preconditioning of capsule filters requires that liquid be passed through the filter prior

to sample filtration and collection A volume of manufacturer-recommended, specific water (e.g., distilled water, deionized water, or sample water) should be passedthrough the filter while holding the capsule upright, prior to sample collection In general,large-capacity capsule filters require that 1000 ml of water be passed through the filterprior to sample collection, while small-capacity filters require approximately 500 ml ofwater to be passed through the filter

medium-Sample Preservation

The second form of pretreatment of ground-water samples is physical and chemicalpreservation As described in ASTM Standard D 6517 (ASTM, 2006b), ground-watersamples are subject to unavoidable chemical, physical, and biological changes relative to

in situ conditions when samples are brought to ground surface during sample collection.These changes result from exposure to ambient conditions, such as pressure, temperature,ultraviolet radiation, atmospheric oxygen, and atmospheric contaminants, in addition toany changes that may be imparted by the sampling device as discussed earlier in thischapter

Objectives of Sample Preservation

The fundamental objective of physical and chemical preservation of samples is tominimize further changes in sample chemistry associated with sample collection andhandling from the moment the sample is placed in the sample container to the time it isremoved from the container for extraction or analysis in the laboratory Samplepreservation methods are determined on a parameter-specific basis and must be specified

in the SAP prior to sample collection Requirements for sample container type, storageand shipping temperature, and chemical preservatives are specified in the analyticalmethod used for each individual parameter selected for analysis Sampling team