Environmental Forensics: Principles and Applications - Chapter 3 docx

Sampling procedures susceptible to chemical bias especially volatile or-ganics include the following: • Improper selection of sampling equipment relative to the analyses to be performed

of the factual information Environmental data relied upon to form an opinion should

be of a sufficient known quality to withstand the scientific and legal challengesrelative to the purpose of the data collection

In most instances, only a small percentage (about 10 to 15%) of the data in anenvironmental investigation contains elements susceptible to bias These elementsare usually associated with the geologic investigation and sample collection, analyti-cal testing, and interpretation of the horizontal and vertical extent of soil andgroundwater contamination

An important task in the forensic review of environmental data is the tion of whether a pattern of bias (systematic error) exists This bias can be due tofactually incorrect information, errors, or intentional manipulation Figure 3.1 illus-trates bias and data variability (random error) based on a sample population whosetrue concentration is about 20 parts per million (ppm) As depicted in Figure 3.1, datacan be biased negatively or positively Three specific types of biases and/or errors aredefined as follows:

determina-1 Positive bias: In a data sufficiency context, a positive bias arises when a test

incorrectly indicates contamination or an increase in contamination when there is none.

2 Negative bias: In a data sufficiency context, a negative bias occurs when

monitor-ing fails to detect contamination or an increase in the concentration of a hazardous material.

3 Erratic data: Erratic data are anomalous values which make it statistically

impos-sible to develop meaningful trends and/or correlations.

These biases result from investigative, sampling, analytical, and statistical errors.Ultimately, expert witness opinions based on incorrect information can result

3.2 GEOLOGIC CHARACTERIZATION

The geologic characterization component of a site investigation provides insightregarding contaminant distribution and transport Components of a geologic inves-tigation usually include:

• Drilling and logging of the boreholes and/or trenches

• Soil retrieval for textural classification and/or physical testing

• Soil sampling for chemical analysis

The first step is to acquire and review the original field borings and/or trench logs.Compare the information on the field logs and final logs in the report for consistency

If geologic cross-sections or fence diagrams are included in the report, examine themfor consistency with the field log/trench descriptions

When reviewing boring logs, examine their placement relative to historicalinformation, especially areas of known or suspected contamination This reviewoften provides insight as to whether additional borings and/or sampling are neces-sary Given that site access agreements among multiple parties are usually required

in order to perform additional sampling, the sooner the data sufficiency of the

FIGURE 3.1 Graphical representation of sample bias and variability (From Mishalanie, E.,

in Proc of the National Environmental Forensic Conference: Chlorinated Solvents and Petroleum Hydrocarbons, August 27–28, College of Engineering and Engineering Profes-

sional Development, University of Wisconsin, Madison, 1998, p 27 With permission.)

geologic information is identified, the sooner the site access agreements and pling can proceed The sufficiency of existing geologic information can be deter-mined via the following steps:

sam-• Ascertain whether the drilling method employed allows an accurate description of the subsurface.

• Determine whether the number of soil borings are sufficient to characterize the geologic environment relative to litigation allegations.

• Decide whether the borings are sufficiently deep to characterize the geology of interest.

• Decide whether the borings are spatially located so as not to preclude developing useful information for geologic characterization.

The drilling technology impacts the geologist’s ability to describe the soil and/orgeologic setting Reliance solely on mixed drill cuttings from air or mud rotarydrilling, for example, precludes the ability to provide detailed descriptions of strati-graphic changes Continuous hollow stem augering and/or most push technologies

that retrieve an in situ soil sample provide this level of detail.

3.2.1 B ORING L OG T ERMINOLOGY

Soil descriptions on a boring log are based on visual observations of drill cuttings orfrom physical testing (e.g., sieve analysis and hydrometer tests) The review of soiltextural descriptions requires that a uniform soil classification scheme be used or thatdifferent classifications are standardized (ASTM, 1993) The use of multiple soilclassification schemes is not uncommon where numerous environmental consultantshave performed geologic investigations at the site

An illustration of the importance of a common classification scheme is a soildescribed as a silt based on the results of a grain size analysis The particle size ofthe soil lies between 0.1 and 0.02 mm Based on these grain size results, multipleparticle size classifications are possible, as shown in Table 3.1 (Gee et al., 1986;Hillel, 1982; Wilson et al., 1998) According to the International Soil Science Societyclassification, the soil is a fine sand Other schemes classify the soil as ranging from

TABLE 3.1

Soil Classification Schemes

Particle Size Classification Classification Scheme Used

Very fine sand to silt U.S Department of Agriculture

Very fine sand to coarse silt Canada Soil Survey Committee

Fine sand International Soil Science Society

Fine sand to fines (silt and clay) American Society of Testing Materials

Fine sand to fines (silt and clay) German Standards

Fine sand to silt British Standards Institute

a silt to a fine sand All are correct for their respective classification schemes.Without adjusting these interpretations to a common standard, however, subsequentgeologic interpretations and associated diagrams can perpetuate this nonstandardizedbias In the United States, the Unified Soil Classification System developed by theCorps of Engineers (U.S Corps of Engineers, 1960) is the most commonly usedsystem (Figure 3.2).

FIGURE 3.2 Unified Soil Classification System, grain size chart, and well construction

symbols used on boring logs.

Soil color is usually recorded on a boring log If a standardized color scheme isnot used, the correlation value of this information should be considered qualitative.The most common color standard is the Munsell Soil Color Chart, which contains

196 different standard color chips (Kollmorgen Corp., 1975) The Munsell system isarranged by the following characteristics:

• Hue is the color of the soil relative to red, yellow, green, blue, and purple.

• Value indicates the lightness of the soil (0 for black and 10 for absolute white).

• Chroma is the strength of the color (0, for neutral grays, to 20) For absolute achromatic colors (pure grays, white, and black) with zero chroma and no hue, the letter N (neutral) is used in place of the hue designation.

A notation such as 5YR 5/6 on a boring log indicates use of the Munsell system In

“5YR”, 5 is the middle of the color value between yellow and red color hue (YR).The notation “5/6” is the chroma value between 5 and 6

Boring logs often contain soil terminology used by non-geologists (drillers orsoil scientists) A driller’s log may qualitatively describe a soil as light or heavy Asandy soil that is loose and well aerated is called light, while a clayey soil that tends

to absorb and retain fluid when wet is termed heavy If there is doubt about themeaning of such terms, ask the author of the boring log

When soil samples are collected at random depth intervals, ascertain whetherthere is an attempt to avoid collecting soil samples with a higher or lower probability

of detecting contamination An example is the consistent sampling of coarse sandsand the avoidance of sample collection in finer grained materials (silty or clayeysoils) through which a contaminant with a high sorption capacity has infiltrated.Conversely, soil samples collected at the interface of coarse, overlying, fine-grainedsediments (i.e., a sand overlying a clay) can result in an overestimation of theconcentration, volume, and (by extension) remediation costs for the contaminatedsoil An extension of this manipulation is the use of small sample volumes forchemical analysis which biases the chemical results due to the soil particle size notbeing representative This technique assumes that the association of a chemical in thesoil is uniquely associated with a particular particle size

A proposed approach to quantify this potential particle size bias is to examine thesample size required for analysis relative to the particle size distribution of the soilsample This method identifies the potential bias due to the grain size distributionbetween soil samples collected from similar soil textures Ramsey (1996) defines thispotential bias due to the particle size representativeness as:

S = (22.5d3/ms)1/2 (Eq 3.1)where

S = sample mass.

22.5 = sampling constant, which is an approximation and is applicable to many, but not all, hazardous waste materials.

d 3 = maximum particle diameter.

ms = sample mass in grams.

In most cases, the larger the sample volume, the smaller the potential particle sizebias If a soil sample is homogenized or sieved by the laboratory and a particularparticle size fraction is selected for extraction, a similar chemical bias can beintroduced Examination of laboratory documentation will provide this information.

If a contaminant is migrating through the soil via unsaturated flow, it canpreferentially circumvent a coarse-grained layer As a result, systematic sampling ofthese coarse-grained sediments underestimates the extent of contamination Con-versely, sampling consistently at the interface of a coarse and fine-grained sedimentthrough which a contaminant has traveled in an unsaturated state provides thegreatest opportunity of detecting a contaminant Plate 3.1* illustrates a field experi-ment where dye moving through a medium-grained glacial sand in an unsaturatedstate preferentially migrates around the coarse-grained sediment

* Plate 3.1 appears behind page 242.

FIGURE 3.3 Boring log with organic vapor analysis measurements.

Soil lithology descriptions, field measurements, and sampling locations corded on a boring log can provide insight regarding the intentional manipulation

re-of sampling locations for the purpose re-of biasing the chemical results Figure 3.3 is

a portion of a boring log containing field organic vapor analysis (OVA) and HNu™measurements The presence of a distinct layer of contamination between 45 and 50

ft is suggested by the HNu™ readings of 200 ppm; if samples were not collected forchemical testing between this interval, this could suggest intentional biasing Thistype of analysis is also useful for targeting subsequent evidentiary sampling Figure3.4 is a field boring log example that illustrates the presence of a volatile compound

at about 5 ft (OVA = 1000 ppm) that was not sampled Samples in Figure 3.4 withnon-detect and near OVA and HNu™ detection levels at 10 and 20 feet, however,were sampled In this instance, the decision not to sample at 5 feet precludes theconfirmation of a potential surface release indicated by the OVA reading of 1000ppm

FIGURE 3.4 Boring log with field measurements (OVA and HNu ).

Field measurements used to screen soil sampling locations are qualitative andsensitive to the compound detected and instrument calibration, but do not rely onfield measurements beyond this qualitative, field-screening purpose Figure 3.5 is afield log in which photoionization (PID), flame ionization (FID), and infrared (IR)detectors were used Values for the three instruments for the same soil ranged from

to create these diagrams These interpretations are important when low permeabilityhorizons, relative to vapor or liquid contaminant transport, are incorporated in thegeologic cross-sections or fence diagrams.

Geologic diagrams are created via manual interpretation (Figure 3.6) or lation by computer software Areas of inquiry (A through D) on Figure 3.6 are framedand labeled If the purpose of the cross-section is to represent the presence of acontinuous layer of clayey soils that retards contaminant transport, potential areas fordiffering interpretation are possible as described in the following text

interpo-A A contact between artificial fill (speckled fill) and a silt and clay (white space) is present midpoint between Boring 1 and MW-1 This interpretation extends the silt/ clay layer into an area where no data are available but which may be a logical assumption.

B The contact between the silty and clayey sand (dotted fill) and the silts and clays (white space) is interpreted to occur at a point that is not midpoint between MW-

1 and MW-B1 This interpretation is inconsistent with the midpoint methodology used in A.

C The extent of the silty and clayey sand (dotted fill) is interpreted to extend halfway between Boring 2 and VE-2 in one direction but only a short distance in the opposite direction between Boring 2 and VE-5 This interpretation creates a signifi- cant horizon of predominately silt and clays between C and D Another interpre- tation is to create a contact between framed areas C and D This alternative interpretation creates a thin layer of silt and clays that are less of an impediment to the vertical transport of contaminants This interpretation is also inconsistent with examples A and B, where the soil contact between two wells is interpreted as the midway point Furthermore, there is no boring located between Boring 2 and VE-4

to indicate the presence of a clay layer.

D The contact between the gravels at the bottom of VE-5 is extended toward Boring

2, where it is not encountered This is inconsistent with the method used in A and B.

E The geologic interpretation between Boring 2 and VE-4 deviates from the pattern observed in frames A to C in that the contact between the silty and clayey sands

in Boring 2 and gravels in VE-4 is not interpreted as occurring midpoint The silty and clayey soils in VE-4 are portrayed as extending just short of Boring 2, although there are no intervening data to confirm this interpretation.

Examine the horizontal and vertical scales used in cross-sections In Figure 3.6, thevertical scale is 0.4¥ of the horizontal scale If the scale is not 1:1, the viewer’s

perception may be significantly skewed The preparation of an alternate geologiccross-section that is scaled and presented as a rebuttal exhibit may be appropriate

A variation to the Figure 3.6 manual interpretation of the geologic data is assignment

of numerical values that represent different soil properties Computer software thenspatially extrapolates between these values Computer interpretations and their por-trayal in cross-sections, isopach maps, or fence diagrams can produce highly errone-ous interpretations When reviewing a computer-generated geologic diagram em-ploying this technique, you will need to:

FIGURE 3.6 Example of manually created geologic cross-section.

©2000 CRC Press LLC

• Obtain a copy of the tables and/or spreadsheets used to assign numerical values to different soil and geologic materials.

• Evaluate whether a consistent numerical value is used for identical soil and/or geologic descriptions.

• Determine how the computer software deals with and assigns geologic descriptions

to two numbers (e.g., rounded up or down).

• Identify whether multiple measurements of a geologic or soil property are cally manipulated to skew the interpolation and resulting graphical portrayal of this property; for example, combining measurements over some vertical distance and taking the average or arithmetric log of the data can mask the presence of a geologic

statisti-or soil property of interest.

3.4 SOIL COLLECTION FOR

CHEMICAL ANALYSES

Significant opportunity exists for introducing bias during the collection of soilsamples Sampling procedures susceptible to chemical bias (especially volatile or-ganics) include the following:

• Improper selection of sampling equipment relative to the analyses to be performed

• Subsampling and sample transfer

• Sample compositing

• Extended holding times

3.4.1 S OIL S AMPLING E QUIPMENT

A variety of soil sampling equipment is available with different levels of potentialchemical bias (ASTM, 1997a,b) Split-spoon sampling is probably the most commonlyused method Split-spoon barrel samplers are not recommended if there is poor samplerecovery (e.g., the metal or brass rings are not completely filled with soil) due to thepotential loss of compounds by volatilization into the headspace of the partially filledbrass tubes Confirmation that the brass tubes are decontaminated prior to use isrequired Pre-cleaned rings or tubes can be purchased with decontamination certifi-cation Recycled tubes can be cleaned at a laboratory with the requisite number of rinsatesamples and testing In the field, the sampling barrel is attached to the drive rod of thedrill rig and is driven into the soil with soil filling the sampling barrel The barrel isthen retrieved at the surface and broken open, and the soil in the brass or steel rings

is sealed or transferred into another container The exposed end of the soil in eachring should be quickly covered and sealed in the field using an inert film, such as TEF-fluorocarbon sheets that are then covered with plastic or threaded metal caps The use

of electrical tape for sealing the plastic end caps on the brass rings is discouraged (seeFigure 3.7) Permeation of toluene from the adhesives in these tapes through the plasticend caps can occur, resulting in a false bias In order to confirm the origin of thetoluene from the adhesive, it is the author’s experience that subsequent samples fromthe same location are required but without the tape used in the initial sampling

FIGURE 3.7 Improper use of electrical tape to seal brass tubes containing soil samples.

3.4.2 S UBSAMPLING AND S AMPLE T RANSFER

Subsampling is the process of “repackaging” a sample into another container tile organic compound losses occur primarily during the soil transfer from a split-spoon sampler into a 40-mL glass vial, an 8-oz glass jar, or plastic bag, throughvolatilization The amount lost is dependent on the vapor pressure of the compound,the amount of headspace in the sample container, the ambient temperature, and theamount of sample disturbance (Hewitt et al., 1992) Figure 3.8 illustrates soilsubsampling from one partially filled brass ring into a second ring so that noheadspace is present If the subsampled soil in the second brass ring is tested forvolatile organic compounds, losses can be as much as 100% of the actual value(Siegrist, 1993)

Vola-Figure 3.9 summarizes data from soil samples contaminated with TCE collectedwith different subsampling procedures and/or containers (Siegrist and Jenssen, 1990).The primary mechanism for loss was due to volatilization during collection, samplestorage, and handling The initial concentration of the spiked soil sample was 4.7 ppm.The soil sampling location coupled with an understanding regarding volatilelosses that occur during soil subsampling can be used to manipulate test results.Consider excavated soil from an underground gasoline tank removal that is stock-piled for several days The ultimate disposal decision for this soil by the regulatoryauthority is predicated on the sample results from the stockpile detecting compoundsbelow a specific action level (Plate 3.2*) BTEX results from a soil sample collectedfrom the crust of the pile has a higher probability of a lower concentration than asample collected from within the interior of the soil pile A decision based on theformer location can result in the non-detect BTEX results The stockpiled soil is thenapproved for placement into the excavation At some future time (e.g., a Phase II

FIGURE 3.8 Example of subsampling of soils in the field.

* Plate 3.2 appears behind page 242.

property transfer), soil in the former tank excavation is sampled, the samples areabove a regulatory action limit, and remediation is required This is a common

scenario, especially with aboveground ex situ remediation, that can be avoided by the

selection of appropriate sampling locations and an adequate number of confirmationsamples

When designing a soil sampling program for volatile organic compound sis, minimize the number (if any) of subsampling and/or sample transfer steps.Current soil sampling procedures specify that samples analyzed for volatile organiccompounds should be shipped to the laboratory in containers filled to capacity (i.e.,

analy-no headspace) and stored at 4∞C for no more than 14 days An option for improving

sample integrity is to use glass jars containing methanol Given that volatile pounds are more soluble in methanol than water, the longer contact time betweenthe methanol and soil results in excellent extraction efficiency In addition, extrac-tion of the volatile compounds from the soil is performed with a larger subsamplethan used in some methods (e.g., 100 g vs 5 g, or, if placed directly into the purgevessel, 1 g for gas chromatography/mass spectrometry [GC/MS] analysis) Thus, amore representative determination of the volatile compounds present results Themethanol container is usually a wide-mouth, 8-oz jar with TFE-fluorocarbon-linedlids Analytical-grade methanol (100 mL) is added to the jar, into which the soil isplaced to a predetermined level followed by immediate sealing of the jar Michigan,New Mexico, Massachusetts, and Wisconsin currently require this procedure forsoils analyzed for volatile organic compounds Considerations in using this tech-nique include investigating whether sample shipment by a commercial carrier isrestricted Coordination with the laboratory is also required if the laboratory pre-pares the methanol-filled containers prior to sampling

com-FIGURE 3.9 Loss of TCE from different sampling containers (Adapted from Siegrist, S and

P Jenssen, Environmental Science and Technology, 24(9), 1387–1392, 1990.)

When reviewing test results obtained from soil samples using this technique, beaware that methanol has a high affinity for many organic compounds Once amethanol bottle used to prepare the 8-oz jars is opened, organic compounds can berapidly adsorbed into the methanol, thereby resulting in cross-contamination Labo-ratories also purchase methanol with contaminant levels exceeding method detectionlimits (Hartman, 1998a) Testing the methanol using the same method selected forthe soil samples prior to filling the jars allows quantification of this potential bias Apotential reduction in analytical sensitivity may also occur if a gas chromatograph/Hall detector is used.

3.4.3 S OIL C OMPOSITING

Composite samples consist of multiple samples collected at various sampling tions and/or points in time The constituent information for the individual samples islost, although it may be indirectly observed through the composite measurement.Composite sampling reduces concentration variability, thereby narrowing the confi-dence interval of the population as contrasted with grab samples that maximizeconcentration variability

loca-When analyzed, composite samples produce a global average value Compositingcan result in loss of information via sample dilution, especially at near-detectionlevels, as well as the possibility of adverse physical, chemical, and biologicalinteractions resulting from the mixing process (Lancaster et al., 1988) Samplesanalyzed for volatile organic compounds should not be composited (ASTM, 1997d).Compositing can mask information otherwise useful for dating a contaminant re-lease Figure 3.10 illustrates this concept; discrete samples collected from boreholeSB2 allow the correction of historical water levels to a known release date to confirmthe release of diesel into the groundwater after 1982 Composite sampling would notprovide the depth of discrete information necessary for this interpretation Compositing,however, has value as a field screening technique for rapidly identifying whethercontamination is detectable or providing a precise estimate of the mean concentration

of a waste analyte in soil or groundwater

Composite sampling is routinely encountered in soil confirmation sampling Thisapplication has merit due to the expected contiguous and non-randomness of thecontamination, along with the assumed quantity of non-detects associated with thetesting In general, individual samples selected for compositing should be of a similarmass, although proportional sampling may be appropriate An example of propor-tional sampling is the collection of soil cores from contaminated soil overlying animpermeable zone (ASTM, 1997c) Soil cores of different length can provide anaveraged contaminant concentration value of the overlying soil, regardless of corelength

The collection time for a single composite sample should not exceed 24 hr Iflonger sampling periods are necessary, the collection of a series of compositesamples is recommended (ASTM, 1996) If composite samples are collected withoutthe opportunity to resample, a novel application using the inverse theory technique

of linear regularization may provide concentration estimates at the individual samplelevel (Lancaster and McNulty, 1998).

The use of field screening technologies rather than composite sampling canprovide a cost-efficient option to compositing X-ray fluorescence, for example, canquickly screen a soil sample for a particular element or target compound, therebyproviding the basis to identify a discrete number of samples for testing (see Table3.10) This approach is conducive to evidentiary sampling, because a large number

of samples can be tested in the field in a short period of time

3.5 GROUNDWATER CHARACTERIZATION

The hydraulic properties of an aquifer are commonly estimated or measured as part

of a groundwater characterization investigation The determination of how theseproperties are measured and their reliability is one factor in evaluating contaminanttransport and risk assessment models Hydraulic properties and their definitionsinclude:

• Hydraulic conductivity: The rate of flow of water in gallons per day through a

cross-section of 1 ft 2 under a unit hydraulic gradient at a prevailing temperature.

FIGURE 3.10 Use of discrete soil sample results and historical groundwater level data to

confirm the release of diesel into the groundwater after 1992.

• Hydraulic gradient: The rate of change in total head per unit of distance of flow

in a given direction.

• Permeability: The property or capacity of a porous rock, sediment, or soil to

transmit a fluid.

• Porosity: The percentage of the bulk volume of a rock or soil occupied by

inter-stices, whether isolated or connected.

• Transmissivity: The rate at which liquid is transmitted through a unit width of an

aquifer under a unit hydraulic gradient.

The accuracy and representativeness of these values are in part dependent on whetherthey are measured in the field or laboratory Three methods used to measure thesaturated hydraulic conductivity of an aquifer (listed from least to most representa-tive method) are laboratory permeater tests, slug tests (field), and pump tests (field).Groundwater velocity is a key input parameter used for advective and forcontaminant transport models Sources of error in acquiring this information includethe following:

1 Installing a well (to measure groundwater levels) near activities that disrupt the aquifer, such as municipal or irrigation wells that are periodically pumped and/or spreading basins used for groundwater recharge

2 Surface water bodies (e.g., streams, lakes, or reservoirs) with highly fluctuating flows located near a monitoring well

3 Tidal cycles that affect groundwater levels

4 Leaking sewers, water mains, and/or ornamental irrigation that affect localized hydraulic gradients

5 Inaccurate surveying of monitoring wells

In order to determine groundwater velocity, monitoring wells are surveyed on thehorizontal and vertical axis Wells should be surveyed to a vertical accuracy of 0.01

ft The water level depth in each well is adjusted to provide a standardized referencepoint (usually mean sea level, MSL) which is used to create a groundwater contourmap If the original vertical survey for a well is incorrect, subsequent measurementscan perpetuate this error, resulting in incorrect interpretations regarding groundwaterdirection and velocity

Incorrect water level measurements can occur as a function of the measuringpoint or the equipment The measuring point refers to the location at the groundsurface or well casing from which the depth is measured; the surface measuring pointmust be consistent In some cases, a well casing can settle over time, resulting in abiased measurement; if this is suspected, re-survey the well

Differences in water level measurements can also occur as a function of theequipment (e.g., steel tape vs a pressure transducer) (Rosenberry, 1990) There issufficient error in the two techniques to produce misleading information about thedirection of groundwater flow, especially for small groundwater gradients (@ 0.001)

Knowing what equipment was historically used for measuring the depth to water in a monitoring well and the consistency of the measurement technique mayexplain apparent anomalies in groundwater flow patterns

ground-Another potential source of error in identifying the direction and velocity ofgroundwater is combining groundwater level measurements from wells screened indiscrete aquifers, especially for unconfined and confined aquifers Using regionalmaps (e.g., 1 in = 24,000 ft) to determine groundwater flow and direction rather thaninstalling on-site wells can also result in erroneous determinations of groundwaterdirection and velocity.

Do not accept reported groundwater direction or velocity a priori without

re-viewing the actual measurements This requires plotting groundwater level ments and creating a groundwater contour map The groundwater direction obtainedfrom the graphing should be compared to the general direction of reported ground-water flow in the environmental report

measure-3.5.1 M ONITORING W ELL L OCATION

Given a sufficient understanding of the hydrogeological environment and the originand physiochemical characteristics of the chemicals released into the subsurface,monitoring wells can be designed to collect samples for the following purposes:

• Provide representative (i.e., the degree to which sample data are characteristic of

a population, variations at a point, or an environmental condition) samples.

• Avoid detecting contaminants.

• Underestimate contamination.

• Overestimate contamination.

• Generate anomalous data.

The location of a monitoring well and well screen interval has a profound impact onthe chemistry of groundwater samples collected from the well Monitoring welldesign features indicative of manipulation are summarized in Table 3.2 Othersources of potential bias include well construction materials, grouting materials,design of security covers, and drilling methods (Powell, 1997) If monitoring wellsare constructed from different materials, this might indicate an attempt to manipulatethe chemical data through well construction materials It can also indicate severalgenerations of consultants working at the site with different preferences for wellconstruction material

Elevating sample pH through improper grouting procedures or the use of ing materials containing potential contaminants can impact sample chemistry Ce-ment grout (CaCO3) can raise the pH of the surrounding soil several pH units.Elevated pH values in the vicinity of the well screen may cause precipitation ofotherwise soluble metals as they enter a halo of higher pH groundwater If thisphenomenon is suspected: (1) examine whether the pH values are high relative toother wells in the area, and (2) excessively purge the well prior to sampling (e.g., 10

grout-to 20 casing volumes) and measure the pH grout-to observe if it suddenly drops several pHunits If pH values decrease abruptly, this may suggest that the grout material hasimpacted groundwater pH in the immediate vicinity of the well

Well construction can also impact sample chemistry Poorly constructed securitycovers or valve boxes that allow contaminated surface seepage into the well canproduce anomalous chemical results (see Plate 3.3*) It is the author’s experience thatstreet runoff containing soluble lead and high total petroleum hydrocarbon concen-trations draining into a well via a cracked security cover can result in detection ofthese contaminants in groundwater samples above regulatory action limits When-ever possible, arrange a site visit to identify the existence of these types of biasesprior to examining the groundwater chemical results.

Another area of inquiry for wells drilled through multiple aquifers is whether theconsultant drilled through a confining layer, thereby introducing contamination into

a deeper, previously uncontaminated aquifer This situation is often identified on aboring log that indicates that the borehole was over-drilled with the lower portion ofthe hole backfilled with grout to seal off the penetration of the confining layer Avariation to this scenario is if the well screen intersects multiple aquifers in whichonly the upper horizon is contaminated Contaminants from the upper, contaminatedzone enter the lower, previously uncontaminated zone A reverse situation is shown

in Figure 3.11, where contamination from a lower aquifer is allowed to mix withinthe well casing during pumping Contaminants then flow into the shallower zoneswhen the pump is not operating The result is the contamination of a previouslyuncontaminated shallow water-bearing zone

TABLE 3.2

Effect of Well Location on Sample Chemistry

Impact on Sample Chemistry Design/Location Characteristics

Contaminants non-detected Wells installed cross-gradient and/or upgradient of source areas;

wells with long screens (>20 ft) and contaminants present at low concentrations

Contaminants underestimated Long screens (>20 ft); wells screened across geologic units with a

low probability of transporting contaminants Contaminants overestimated Short screens located to intersect zones with a high probability of

contamination (i.e., LNAPLs at the water table); wells screened across high- and low-contaminated zones that result in an averaged concentration that is higher than the actual concentration in the groundwater

Anomalous data Well located next to a surface water body whose chemistry is

dissimilar to the groundwater chemistry and whose presence pacts the sample chemistry in a transient and unpredictable manner; well located in an area of changing groundwater direction that results in varied chemistry depending on the direction of groundwa- ter flow.

im-* Plate 3.3 appears behind page 242.

3.5.2 I NSTALLATION OF G ROUNDWATER

M ONITORING W ELLS

Contaminant distribution in groundwater is defined by the horizontal placement ofthe monitoring well as well as the screen length and interval While federal and stateguidelines exist, the consultant or driller usually designs the well As a result,multiple well construction designs may be present which impact the chemical inter-pretation of samples collected from the network

If well construction is suspected of biasing sample chemistry in some manner, thefirst step is to review the placement of the well screen The location of the well screendetermines the vertical horizon from which a sample is collected, unless discretevertical depth sampling is performed Questions to be answered during this analysisinclude:

FIGURE 3.11 Cross-contamination of a shallow aquifer by a multiple-screened pumping

well.

FIGURE 3.12 Hydrograph indicating the presence of two aquifers.

• Are the wells screened in similar water-bearing zones?

• Does the well screen length bias sample chemistry?

• Is the well screen providing a pathway for contaminant transport from areas of high

to low contamination or from contaminated to non-contaminated zones?

The first step in answering these questions is to construct a hydrograph Ahydrograph plots time on the horizontal axis and the water level measurement foreach well on the vertical axis If sufficient information is available, the water level

in each well for a point in time is plotted and the data connected If the wells intersectthe same water-bearing zone (i.e., if they are hydraulically connected), the lines forthe various wells will follow a similar pattern over time If the wells do not followthe same general pattern, this may be evidence of multiple saturated zones that aremonitored by the well network Separate water level contour maps should be createdfor each aquifer indicated by the hydrograph Figure 3.12 is an example of a hydro-graph showing two distinct aquifers

Another technique is to sketch all of the monitoring wells on a single sheet ofpaper, adjusting for the surface elevation for each well, with the vertical axisrepresenting well depth Mark the screen interval on each well along with the waterlevel This sketch provides insight regarding the consistency of the well screensbetween wells and patterns between well screen length and water levels

Another area of inquiry is whether the screen interval biases sample chemistry

If the predominant contaminant of interest is a light non-aqueous phase liquid(LNAPL) and the well screen does not intersect the water table, the LNAPL will not

be detected Other examples of high contaminant concentrations near the water tablethat decrease sharply with depth include toluene, xylene, xylidine, dissolved oxygen,and manganese concentration profiles (Kaplan et al., 1991) and BTEX concentra-tions (Gibs et al., 1993; Martin-Hayden et al., 1991) An example of the latter is thesampling for BTEX components at the water table with a short screen (<5 to 10 ft)well vs a longer screened (>20 ft) well If this inquiry reveals potential patterns

between compound concentration depth profiles and screen length, examine thescreen length as a function of proximity to the contaminant source (if known) Ifmonitoring wells closest to the source area have long screens (possible sampledilution) and those farthest from the source are short screened, a pattern of samplechemistry manipulation and contaminant plume geometry via well screen length andproximity to the source may become apparent (Martin-Hayden and Robbins, 1997).The interpretation of the vertical distribution of a contaminant is especiallysensitive to well screen length, interval, and placement (Robbins and Martin-Hayden,1991) Detailed studies of vertical spatial and temporal gradients indicate that trans-port of contaminants is often limited; therefore, concentration profiles can be highlyvariable with depth (Barcelona et al., 1989; Garabedian et al 1987; Gibs et al., 1993;Sudicky et al., 1983) The ability of the monitoring well to provide this resolution is,then, spatially dependent, primarily on the monitoring well screen design.

An example of the impact of well screen length on source identification is shown

in Figure 3.13 The plan view in the upper panel of Figure 3.13 plots TCE trations from groundwater samples During the first round of sampling, all of thewells, except the demonstration well, were sampled (the demonstration well wasinstalled after the first round of sampling) The interpretation of the data indicated asource in the vicinity of the 1650 ppb of TCE given the upgradient well value of 120ppb TCE concentrations relied upon for this conclusion were the 120 ppb from theupgradient well, the 1650 ppb from the well near the source, and the downgradientwell with 1500 ppb The source and downgradient wells were short screened (10 ft)and completed in a silty sand The upgradient well (120 ppb) was completed with a20-foot screen that intersected the silty sand and a highly permeable sand and gravellayer Concern about the longer screened, upgradient well diluting the TCE concen-tration via the uncontaminated sand and gravel layer resulted in installation of ademonstration well screened in a manner identical to the source and downgradientwells upgradient of these three wells Sampling of the demonstration well shown inthe cross-section on the lower panel of Figure 3.13 resulted in a TCE value of 1600ppb This information is consistent with a revised interpretation that TCE migratedonto the property from an upgradient source

concen-Ideally, monitoring wells are all short screened (approximately 5 to 10 ft) andlocated at appropriate depths relative to the goal of the monitoring program In mostcases, however, monitoring wells are constructed with different screen lengths andmay not be ideally located While the potential bias associated with different moni-toring well construction can be identified, quantification of this bias cannot beassessed The qualitative impact of different well construction designs in addition to

an evaluation of groundwater purging and sampling techniques should be examined

in total and a judgment made concerning the reliability of the chemical data

3.5.3 S AMPLING P LAN

Environmental reports often include soil and groundwater sampling plans Reviewthe sampling plan and compare it with the field notes describing the actual field

practice This review can identify whether significant deviations from the samplingplan occurred A recommended step in performing this review is to compare thefollowing components of the sampling plan with what transpired during its imple-mentation:

• Could purging, sampling, and handling procedures for compounds susceptible to volatilization, precipitation, or other forms of known chemical transformation result in chemical transformations?

• Was sample filtration and/or preservation properly implemented?

• Were sufficient field blanks, travel blanks, duplicate samples, and/or equipment rinsate blanks incorporated into the sampling protocol to allow for the discrimina- tion of potential sampling introduced bias?

• Were expedited storage and transportation of samples to the laboratory mented as specified in the sampling plan?

imple-FIGURE 3.13 Impact of well screen length on source identification.

±10% in purge water pumped over at least two successive well volumes.

If a monitoring well slowly recharges, sufficient time should elapse so that atleast 95% of the purged water comes from the aquifer; wells should also be sampledwithin 6 hr of purging (Csuros, 1994) At no times should a well be purged to dryness

if the recharge rate causes the formation water to cascade down the interior of thewell screen For a well whose screen intersects a low-permeability formation,desaturating the well during purging can also result in weighting the average con-taminant concentrations near the bottom of the well and therefore underestimatingthe concentration and size of the contaminant plume (Martin-Hayden et al., 1991).Micropurging or low-flow sampling is generally considered to be between about

100 to 200 mL per minute (Puls et al., 1990) Higher rates are acceptable (i.e., 1 Lper minute) for more transmissive formations (Powell 1993) Research indicatesthat a representative groundwater sample for volatile organic sampling is achievablevia micropurging without pumping 2 to 5 well casing volumes (Kearl et al., 1994;Powell et al., 1997; Puls et al., 1995) The U.S Environmental Protection Agencyadvocates micropurging coupled with turbidity, pH, redox potential, and dissolvedoxygen measurements using downhole meters or flow-through cells until stability

is achieved These measurements are considered stabilized when the values arewithin approximately 10% over at least two measurement events Stabilization canalso occur prior to the removal of one well casing volume or may require excessivepurging (Barcelona et al., 1994; Robin et al., 1987) Research indicates that for fullyhydrogeologically characterized sites with long-term monitoring data, micropurgingand the measurement of water quality parameters such as pH, turbidity, and totalconductance may not be necessary Groundwater sampling for uranium at theFernald facility in southwestern Ohio, for example, indicated that representativesamples were collected with micropurging sampling without water quality param-eter measurements (Shanklin et al., 1995)

In a California study involving trace metal contamination, low-flow purging andsampling in addition to rigorous adherence to sampling and handling procedureswere examined When low-flow and trace metal clean techniques were employed,resultant trace element concentrations were notably lower than for values obtainedwith conventional methods The trace element concentrations from groundwaterwells were 2 to 1000 times lower than those previously reported by consultants usingconventional sampling techniques at the same wells While the consultant reportedthat cadmium and chromium concentrations exceeded the California maximumcontaminant levels (MCLs), these levels appeared to be artifacts of inappropriate,albeit standard, sampling techniques (Creasey, 1996).

The use of micropurging can bias sample chemistry due to the small capture zoneassociated with low purging rates Consider a monitoring well with phase-separateTCE in the groundwater that is 25 feet down-gradient from the well Historical TCEconcentrations in groundwater from this well are consistently in the parts-per-millionrange Purging rates of 5 gallons per minute and greater were used Subsequentsample results collected via micropurging (100 mL/min) are non-detect because thepurging and sampling capture zone does not extend to the vicinity of the phase-separate and dissolved TCE

3.5.5 G ROUNDWATER S AMPLING

When reviewing groundwater sampling procedures, identify the type of samplingequipment (e.g., bailer, submersible pump, peristaltic pump) and sampling proce-dures used Groundwater samplers vary in their impact on sample chemistry Thegreatest opportunity for negative bias occurs when sampling groundwater to betesting for volatile organic compounds Because the true value of a volatile organiccompound is unknown, the bias introduced by equipment selection and operation isrelative Sampling equipment and procedures that result in higher levels of volatilecompounds are therefore considered most accurate (i.e., unless evidence of a falsepositive bias is identified)

Numerous studies have evaluated the attributes of groundwater sampling ment and procedures on sample quality (Stolzenberg et al., 1986) Many of theseinvestigations examined the precision (i.e., a measure of the reproducibility of a set

equip-of replicate results among themselves or the agreement among repeated observationsmade under the same conditions) and recovery of volatile organic compound deter-minations on replicate samples collected in the field or laboratory Volatile com-pounds are particularly sensitive to losses by degassing at reduced pressures fromsuction devices such as peristaltic or centrifugal pumps or turbulence created bymechanical sampling devices such as gas lift samplers and bailers Volatile organiccompound losses from improper sampling equipment selection and use are cumula-tive in nature Sampling procedures that represent potential sources of volatileorganic compound loss include:

• Sampling with a bailer and aerating the sample during transfer from the bailer into the sample container

• Over-pressurizing (see Figure 3.14 ) or depressurization (Barcelona, 1990)

• Insufficient decontamination between sampling or lack of equipment blank samples between sampling, if non-dedicated sampling equipment is used

• Not placing the cap on the sampling container or immediately chilling the sampleThis level of information is usually not included in the environmental report butmay be available in field notes acquired via deposition testimony of the sampler Ingeneral, positive displacement pumps (e.g., bladder pumps) provide accurate, repro-ducible sampling performance over a range of lifts, hydraulic heads, and depths.Dedicated sampling systems suitable for both purging and sampling are also pre-ferred based on sample integrity as well as convenience and cost

3.5.6 S AMPLING E QUIPMENT AND S EQUENCE

In general, the compatibility of the sampler material(s) with the analysis to beperformed is obvious (e.g., iron bailer when testing for trace metals) Subtle chemicalbiases can be introduced, however, from a polyethylene bailer manufactured fromplastic regrinds or plastics that contain additives vs a sampler manufactured fromvirgin material A bailer manufactured from solid rod or block stock may similarlycontain potential cross-contaminants from the machining process, as opposed to abailer that is injection molded without release agents such as waxes or petroleum-based lubricants Depending on the level of scrutiny required and the potential biasrelative to the allegations in the case, this level of inquiry may or may not benecessary

FIGURE 3.14 State-of-the-art sampling equipment used to strip volatile organic compounds

from a groundwater sample by over-pressurizing the pump (Courtesy of QED; Ann Arbor, MI.)

Compounds susceptible to chemical changes due to redox shifts from the pling equipment and field procedures are most vulnerable to bias, as are compoundswith low detection limits With volatile organic compounds, losses are more likely

sam-to increase due sam-to operasam-tor procedures (Imbrigiotta et al., 1986) Table 3.3 rizes the impacts of sampling equipment (and procedures) on benzene and chloroben-zene concentrations (Blegen et al., 1988) If the opportunity exists to re-sample thewell from which suspect data were collected, the well can be re-sampled withidentical equipment but with different procedures or different equipment so that thepotential source of the bias can be identified

summa-If a monitoring well is no longer available or if the monitoring well construction

is in question, groundwater sampling using direct push technology (e.g., Hydropunch,Geoprobe, Strataprobe, etc.) can be used at the same location Standardized samplingguides are available for using direct push technologies for collecting groundwatersamples (ASTM, 1997c) Comparisons between groundwater sample chemistry ob-tained from direct push samples and standard monitoring wells are reported to bestatistically similar (Bergren et al., 1990; Church et al., 1996; Kaback et al., 1990).Sampling recommendations that minimize sampling bias, especially redox-sen-sitive and volatile compounds, include the following (Puls et al., 1989a,b; 1992):

• Isolate the sampling zone with packers to minimize the amount of purge water.

• Use low-flow pumping to minimize sample aeration and turbidity.

• Monitor water quality parameters while purging to establish baseline or state conditions to initiate sampling.

steady-• Perform filtration to estimate the total dissolved species present and collect tered samples for estimations of contaminant mobility.

unfil-In addition to potential chemical biases introduced via the selection of samplingequipment and materials of construction, the sampling sequence can also impactsample chemistry Figure 3.15 depicts a site with monitoring wells located on and offsite with the property boundary depicted by a dotted line Groundwater flows to thenorth PCE concentrations in groundwater for the previous quarter and the samplingsequence were obtained from the chain of custody (see Table 3.4) In addition, thesampling equipment and procedures were described in the work plan For on- and

TABLE 3.3 Effect of Sampling Equipment on Benzene and Chlorobenzene Concentrations

Sampling Benzene Chlorobenzene

Dedicated bladder pump 279 1463

TABLE 3.4

Summary of Sampling Results, Sampling Sequence,

and Sampling Equipment for Site Shown in Figure 3.15

Sampling Sampling Sequence Equipment Well PCE ( mmmmmg/L) On-Site Off-Site Purging Sampling

a Disposal Teflon ® bailer.

b Submersible pump operated at maximum pumping rate for purging and sampling.

FIGURE 3.15 Map illustrating the impact of sampling equipment and sequence on

ground-water chemistry results.

off-site sampling, groundwater sampling proceeded from wells with no detectablePCE to wells with higher PCE concentrations, an observation which may indicate anintent to minimize the potential for cross-contamination via the sampling sequence.The purging and sampling equipment used also suggests a knowledge of the impact

of sampling equipment on sample chemistry Equipment that introduces the leastpotential of sample chemistry bias (e.g., peristaltic pumps and Teflon® bailers) wasselected for on-site wells, while off-site groundwater samples were collected withequipment that introduced a significant potential for sample loss via volatilization(submersible pumps) A conscious attempt by the consultant to minimize the off-sitedetection of PCE may therefore be suspected

The sampling sequence may also be designed to maximize particular temporalimpacts on sample chemistry Figure 3.16 illustrates the impacts of some of thesefactors on groundwater chemistry If such relationships exist, identify whether thesampling schedule results in a systematic impact on sample chemistry, and, if so,incorporate this information into your analysis

3.5.7 E QUIPMENT D ECONTAMINATION

The chemical results from equipment decontamination samples allows an assessment

of the presence and nature of cross-contamination originating from the samplingequipment Federal, state, and American Society for Testing Materials (ASTM)standards are available which describe decontamination procedures for contact andnon-contact equipment used for soil and groundwater sampling (ASTM, 1990) Ingeneral, sampling equipment is washed with a detergent solution followed by a series

of water, desorbing agents, and deionized water rinses A decontamination procedurefor sample contact equipment includes the following tasks (Wilson 1998):

1 Wash with a detergent solution (Alconox ® or Liquinox ® or similar nonphosphate/ ammonia detergent) with a brush made or an inert material.

2 Rinse with water of a known chemical composition.

3 Rinse with an inorganic desorbing agent (10% nitric or hydrochloric acid made from reagent-grade nitric or hydrochloric acid and deionized water); this step may

be deleted if the samples will not be tested for organics.

4 Rinse with deionized water.

5 Air dry prior to next use (ascertain whether fugitive or vapors can contaminate the equipment; if this is a possibility, then air dry in an environment where airborne contamination is not a consideration).

6 Wrap the equipment for transport with an inert material such as aluminum foil.The U.S Environmental Protection Agency recommends a similar procedure forequipment used to collect samples tested for organic and inorganic constituentsaccording to the sequence in Table 3.5 (U.S EPA, 1991) Non-sample contactequipment can be rinsed with a portable power washer or steam cleaner In addition,handwashing with a brush and detergent solution may be required, followed byrinsing with water of a known chemical composition (ASTM, 1990) Examples of

FIGURE 3.16 Examples of temporal variations on groundwater chemistry and LNAPL

thickness.

noncontact equipment include drilling augers and cone penetrometer rods (see Figure3.17).

For rigorous quality assurance and quality control situations, the rinse water usedfor equipment decontamination is sampled and tested for the same compounds to beanalyzed as the sample obtained with the decontaminated sampling equipment.Referred to as an equipment rinsate blank, it is a sample of the last decontaminatedwater poured over the equipment Equipment rinsate blanks are collected from non-dedicated equipment such as pumps used for sampling, interface probes, mixingbowls, sampling scoops, split-spoon samplers, Hydropunches, bailers, and conepenetrometer testing tips Equipment rinsate blanks should be collected at a rate of

TABLE 3.5 Decontamination Procedures for Sampling Equipment Used To Collect Samples for Organic and Inorganic Analysis

Organic Compounds Inorganic Compounds

Tap water Dilute (0.1-N) hydrochloric

Organic-free reagent water or nitric acid Reagent grade acetone Tap water Pesticide-quality hexane, Reagent-grade water methyl alcohol, or

isopropanol alcohol, depending on the analysis.

FIGURE 3.17 Decontamination of groundwater sampling equipment and rinsate troughs for

cone penetrometer rods.

one blank per every 10 samples or one per day (U.S EPA, 1995) Care must be taken

to label the rinsate sample and the corresponding sample collected with this ment correctly, so that any cross-contamination from the decontamination procedurecan be identified and quantified

equip-3.5.8 S AMPLE C ONTAINERS

Recognized protocols are available that define the proper size and appropriatecontainer material for soil and water samples for a given analysis When reviewingthe chain of custody, identify whether proper containers were used and, if not,whether this represents a potential false or negative bias Generally, these types ofissues are only significant when trace concentrations are of interest Table 3.6summarizes containers relative to the analyses to be performed (U.S EPA, 1995).For liquid samples, the appropriate container for volatile compounds is a 40-mL vialsealed with a Teflon®-lined cap The vial is filled with the meniscus above the top of thevial so that when the sealed vial is inverted, there are no air bubbles Most laboratorieswill note on the sample receipt checklist log if vials are received with bubbles andmay not analyze the sample without client authorization Many laboratories will notaccept samples for analysis if the bottles were uncertified clean by a supplier orcleaned by another laboratory Table 3.7 summarizes recommended cleaning proce-dures for reused sampling containers (Wilson, 1998) If a concern exists about thechemical integrity of reused sample containers, fill several containers with distilledwater and analyze them for the compounds for which the samples will be tested

TABLE 3.6

Appropriate Sample Containers and Analysis

Container Description Analysis

80-oz amber glass bottle with Teflon ® -lined black phenolic cap Extractable organics 40-mL glass vial with Teflon ® -backed silicon septum cap Volatile organics

1-L high-density polyethylene bottle with polyethylene-lined, Metals, cyanide, and sulfide white polyethylene cap

120-ml glass vial with Teflon ® -lined, white polyethylene cap Volatile organic (soil) 16-oz wide-mouthed glass jar with Teflon ® -lined, Extractable organics/metals polylyethylene cap (water analysis)

8-oz wide-mouthed glass jar with Teflon ® -lined, black Extractable organics and polyethylene cap (water) metals in soil

4-oz wide-mouthed glass jar with Teflon ® -lined, black Extractable organics and polyethylene cap (water) metals in soil

1-L amber glass bottle with Teflon ® -lined, black polyethylene cap Extractable organics 4-L amber glass bottle with Teflon ® -lined, black phenolic cap Extractable organics 500-mL high-density polyethylene bottle with polyethylene- Metals, cyanide, and sulfide lined, baked-polyethylene cap

Sample cross-contamination can result if samples with significant differences inconcentration are shipped together For example, samples of free-phase gasoline andgroundwater samples to be analyzed for BTEX compounds should be shippedseparately The shipping container can also cause cross-contamination Cross-con-tamination originating from an ice chest can be identified via the travel blanks results.When designing a sampling plan, buy new ice chests rather than using ones provided

by the laboratory (which may have compounds sorbed into their plastic) Considersegregating highly contaminated samples (e.g., oily soil, samples with high photoion-ization detector readings) from samples without gross indications of contamination

in separate shipping containers

3.5.9 S AMPLE F ILTRATION , P RESERVATION ,

AND H OLDING T IMES

Water samples analyzed for metals may require filtration in the field Generally,samples analyzed for trace metals, inorganic anions, and cations are filtered, whilewater to be tested for total organic carbon and volatile organic compounds isunfiltered (Barcelona and Morrison, 1988) The U.S Environmental ProtectionAgency recommends collecting one unfiltered sample for total metals and onefiltered sample for dissolved metals (U.S EPA, 1986) If the contaminant concentra-tion in an unfiltered sample is higher than for a filtered sample, a portion of thecontaminant may be sorbed onto the solid particulate matter in the water Colloids

as large as 0.45 to 3 mm can be mobile and capable of transporting contaminants large

3 hr Remove the vials from the oven and allow to cool for at least 20 min Cap each vial with

a heated septum and store within a sealable (Zip-Lok ® ) bag.

distances If a chemical is transported via colloidal transport, field filtering canremove this colloid and the associated contaminant If the purpose of the sampling

is to estimate the extent of metal contamination, substantial underestimation ofcontaminant mobility can result due to metal/colloidal associations with filteredsamples At a Superfund site in California, groundwater samples tested for lead werefiltered with a 0.45-mm filter; the lead concentrations in the unfiltered samples were

20 to 600 times greater than the filtered samples For chromium, concentrations were

6 to 24 times greater in the unfiltered samples than in the filtered samples (Puls andBarcelona, 1989)

The standard 0.45- mm filter is the standard opening size, although this is an

artificial convention A 0.45-mm filter used to determine the extent of metal

contami-nation in a dissolved state can overestimate the actual concentration due to theassociation of metals with colloidal material less than 0.45 mm If the accuracy of the

dissolved metal concentrations is of concern, samples can be field filtered through a1- mm pore size filter using an in-line filter, and acidified immediately to <2 pH with

concentrated HNO3 (Puls et al., 1992) Another option is to filter the sample withmultiple filter sizes (the opening of a filter can also clog during filtering, resulting in

a diminished filter pore size) and analyze the filtrate and dissolved component.Figure 3.18 illustrates the impact of filtration with three pore sizes on elements from

a groundwater sample

Another recommendation is no filtration or a 4-mm filter for the determination

of mobile metals and in-line filtration with a large non-metallic (e.g., 142-mm),polycarbonate type, 0.1-mm pore size filter, for geochemical speciation modeling(i.e., the dissolved fraction) (Puls and Barcelona, 1989) Metal analysis is especiallysensitive to aeration introduced by filtering A small volume of oxygen introduced

in a reduced groundwater sample can result in decreases of up to 100% of lead,cadmium, zinc, arsenic, vanadium, and phosphate The amount of adsorption of atrace metal onto ferric hydroxide also depends on the extent of iron oxidation that can

be introduced via filtering It is not unusual to observe differences greater than 10%between filtered and unfiltered samples for many elements

Field filtration devices include in-line filters, positive pressure filtration, andvacuum filtration (Figure 3.19) If sample containers contain a preservative such as

FIGURE 3.18 Impact of 0.10-, 0.40-, and 10.0-mm filters on elemental analysis of a

ground-water sample (Adapted from Puls, R and Barcelona, M., Hazardous Waste and Hazardous Materials, 6(4), 385–393, 1989.)