United States
Environmental Protection
Agency
Robert S. Kerr Environmental
Research Laboratory
Ada, OK 74820
&EPA
Research and Development EPA/600/M-90/Q23 December 1990
ENVIRONMENTAL
RESEARCH BRIEF
Colloidal-Facilitated Transport of Inorganic Contaminants in
Ground Water: Part I. Sampling Considerations
Robert W. Pulsa, James H. Eychanei* and Robert M. Powell0
•Abstract
Investigations at Final Creek, Arizona, evaluated routine sampling
procedures for determination of aqueous inorganicgeochemistry
and assessment of contaminant transport by colloidal mobility.
Sampling variables included pump type and flow rate, collection
under air or nitrogen, and filter pore diameter. During well purging
and sample collection, suspended particle size and number as
well as .dissolved oxygen, temperature, specific conductance,
pH, and redox potential were monitored. Laboratory analyses of
both unfiltered samples and the filtrates were performed by
inductively coupled argon plasma, atomicabsorption with graphite
furnace, and ion chromatography. Scanning electron microscopy
with Energy Dispersive X-ray was also used for analysis of the
filtered particulates.
Suspended particle counts consistently required approximately
twice as long as the other field-monitored indicators to stabilize.
High-flow-rate pumps entrained normally nonmobile particles.
Differences in elemental concentrations using different filter-
pore sizes were generally not large with only two wells having
differences greater than 10 percent in most elemental
concentrations, although trends showed increasing
concentrations with increasing filter pore sizes in most wells.
Similar differences (> 10%) were observed for some wells when
samples were collected under nitrogen rather than in air. Fe2V
Fe^ratiosfor air-collected samples were smallerthanforsamples
"U.S. EPA, Robert S. Kerr Environmental Research Laboratory, Ada,
OK , bU.S. Geological Survey, Tucson, A2, and CNSI Technology
Services Corporation, Ada, OK.
collected under a nitrogen atmosphere, reflecting sampling-
induced oxidation.
Introduction
Ground-water samples that are representative of actual ground-
water quality are, at best, difficult to obtain (Claassen, 1982).
Disturbance of the subsurface environment ,is unavoidable
during well construction activities. Additional disturbance during
sample collection may drastically alter ground-water chemistry
due to oxidation, sorption, mixing, and turbulent flow resulting in
inaccurate estimations of contaminant loading and transport
predictions. A common study objective is to determine what
constituents are mobile in an aquifer. Many ground-water
samples are filtered to exclude particles dislodged from the local
well environment, because those particles are not mobile at
ordinary ground-water velocities. Because geochemical models
are based on the thermodynamics of dissolved constituents,
small pore-diameter filters have been preferred as the best way
to separate dissolved from particulate constituents.
In practice, 0.45-u.m filters are commonly used to balance
between the objectives of isolating dissolved constituents and
permitting reasonable use in the field. Unfortunately, particle
sizes do not have an express lower bound so that the right filter
can perfectly separate particles from solutes. Particles with
diameters from 0.003 to 10 u.m, referred to as colloids, may form
in certain environments and be mobile atground-watervelocities.
Use of 0.45-u.m filtration may exclude an important component
of the contaminant load at some waste sites, particularly where
highly toxic metals are involved (Puls and Barcelona, 1989).
^g£> Printed on Recycled Paper
-------
Many studies have demonstrated contaminant transport by
colloidal mobility (Gschwend and Reynolds, 1987; Eichholz et al.
1982; Enfield and Bengtsson, 1988; Robertson, 1984). There is
increasing concern that current methods of ground- water sample
collectfon may exclude this component of the contaminant loading
in a given system. If the purpose of sampling is to estimate
contaminant transport, substantial underestimations of mobility
may result, because of colloidal associations. Numerous studies
attesttothe strong sorptive capabilities of secondary clay minerals;
hydrous Fe, Al, and Mn oxides; and humic material of colloidal
dimensions. Takayanagi and Wong (1984) determined that
more than 70 percent of thetotal Se in riverwaters adjacent to the
Chesapeake Bay was associated with organic and inorganic
colloidal particles. Buddemeier and Rego (1986) found that
virtually all the activity of Mn, Co, Sb, Cs, Ce, and Eu was
associated with colloidal particles in ground-water samples from
underground nuclear-test cavities at the Nevada Test Site.
Colloidal particles generated in batch experiments,by Sheppard
et al. (1979) and Puls et al. (1989) were shown to retain substantial
proportions of radionuclides. Further work by Sheppard et al.
(1980) concluded that the transport of radionuclides by colloidal
clay particles should be considered in contaminant-transport
models.
Filtration is part of this concern; but otherfactors, such as sample
exposure to atmospheres different from aquifer environments
and pump-induced disturbance of the sampling zone, are also
important. Oxidation-induced precipitation and sorption
processes, many of which are kinetically rapid (seconds to
minutes), may cause previously dissolved species to be removed
during filtration, resulting in lower metal concentrations than are
actually present in the aquifer. Filter loading and clogging with
fine particles may also occur, reducing the nominal pore size of
the filter and introducing errors due to changing effective pore
size (Danielsson, 1982).
Background
A workshop was convened at the Robert S. Kerr Environmental
Research Laboratory (RSKERL) of the U.S. Environmental
Protection Agency in 1988 to examine these issues and provide
technical guidance based on currently available scientific
Information. A Superfund Ground Water Issue Paper resulting
from the workshop emphasized the importance of well construction
and-sampling methodology in obtaining representative water
chemistry data (Puls and Barcelona, 1989). Workshop
recommendations in the area of ground-water sampling are
briefly summarized below:
Purging
Water that remains in the well casing between sampling
periods is unrepresentativeof water in the formation opposite
the screened interval. It must be removed by purging or
isolated from the collected sample by a packer arrangement
prior to the collection of representative water samples. It is
important to purge the stagnant water at flow rates below
those used in development to avoid further development,
well damage orthe disturbance of accumulated corrosion or
reaction products in the well.
Isolation of Sampling Zone
Isolation of the sampling zone is necessary to minimize the
purge volume as well as to minimize air contact. This is
especially important since Eh/pH conditions of the formation
waters are often sensitive to dissolved-gas content. Inflatable
packers can be used to achieve isolation of the sampling
zone.
Pumping for Sample Collection
It k\ recommended that a positive displacement pump be
used. Other types of sample collection (e.g. bailing) may
cause displacement of non-mobile particles or substantially
alter ground-water chemistry leading to colloid formation
(e.g. vacuum pumps). Surging must be avoided, and a flow
rate close to the actual ground-water flow rate should be
employed. While an initial approximation, flow rates around
100 to 500 ml/min have been used to successfully sample
ground waters in a quiescent mode.
Assessment of Water Constituents During
Purging and Sampling
Monitoring for dissolved oxygen, temperature, specific
conductance, pH and turbidity during purging and sampling
is recommendedtodeterminebaselineground-waterquality
conditions prior to sampling.
Filtration
For estimates of contaminant mobility, filtration with coarse
filters (> 2 u.m) using the same procedures as above or
collection of unfiltered samples is recommended. Filtration
foraccurate estimations of geochemistry should be performed
in the field with in-line pressure filtration using a large (e.g.
142 mm) polycarbonate-type (thin with sharp pore-size
cutoff) 0.1 u.m filter. Air contact should be minimized and
entirely excluded for some samples. Acidification of samples
to <: pH 2 should be performed immediately. The filter holder
should be non-metallic. Holders made of steel are subject
to corrosion and may introduce non-formation metals into
samples. Prewashing of filters should be routinely performed.
In ah effort to test the efficacy of these recommendations, a joint
study by the U.S. Environmental Protection Agency and the U.S.
Geological Survey was begun in the spring of 1988. Collection
of representative unfiltered samples is quite challenging in many
systems because of the difficulty of excluding nonsuspended or
artifact particulates. Because no sampling technique is totally
passive, all contaminant-mobility estimates based on unfiltered
samples are biased toward overestimation. An attempt was
made to minimize this bias by carefully following the workshop
recommendations.
Purpose and Objectives
The specific objectives of the study were to evaluate perturbations
to the ground-water geochemistry during sample collection and,
in particular, to identify those factors that caused significant
differences in elemental concentrations or concentrations and
size distributions of suspended particles in samples collected for
analysis. Samples for both dissolved and suspended
contaminants were collected. Filters smaller than 0.45 p.m were
used to sample for dissolved constituents and for comparison
with the unfiltered or coarsely filtered samples. This document
summarizes the results of the study and addresses the efficacy
of the 1988 RSKERL filtration workshop recommendations on
ground-water sampling for metals analyses.
-------
Study Site
The study site is located at Final Creek, near Globe, Arizona,
about 130 km east of Phoenix and about 170 km north of Tucson.
Copper has been mined since 1903 from granite porphyry
adjacent to an aquifer at the site. A band of unconsolidated
alluvium 300 to 800 m wide, as much as 50 m thick, and about
20 km tong forms the upper, central part of the aquifer in a valley
along Miami Wash and Final Creek (Figures 1 and 2). Most of
the sediment in the alluvium ranges in size from fine sand to
coarse gravel, but clay and boulder lenses also are present.
Alluvial basin fill more than 100 m thick forms the remainder of
the aquifer beneath and adjacent to the unconsolidated alluvium.
Peterson (1962) described the geology of the area.
During 1940-86, acidic mining waste solutions were discarded in
an unlined lake formed behind waste and tailings piles. In 1986,
pH at the lake surface was about 2.7 and the lake volume was
about 5.5x10» m3. By May 1988, virtually all the lake water had
been spread on inactive tailings piles to evaporate. Contamination
of ground and surface waters in the area has been described by
Eychaner (1989). The distribution of pH in the aquifer was used
as a guide in selecting wells to sample for this study (Figure 1).
Water levels and chemical quality have been monitored since
1984 in several groups of observation wells (Figure 1). Each
group consists of separate wells individually completed with 10-
cm-diameter polyvinylchloride casing and a single well screen.
Most of the well screens are 0.9 m long; the longest screen in a
well sampled forthis study is 6.1 m. Most of the wells were drilled
by the hydraulic rotary method using bentonite-based drilling
mud; five wells were drilled by the hollow-stem auger method.
The annulus in the screened interval was packed with washed
pea gravel from a nearby uncontaminated area. The gravel pack
was capped with a 1 -m layer of bentonite pellets. Each well was
developed by jetting high-pressure air through the screen to
dislodge and remove fine-grained material. Comprehensive
data from the study area are available (Eychaner et al, 1989).
In the alluvium, hydraulic conductivity is on the order of 200 m/d
on the basis of cross-sectional area, hydraulic gradient, and
measured outflow (C.C. Neaville, hydrologist, U.S. Geological
Survey, written commun., 1990). For thick sections of basin fill,
hydraulic conductivity was estimated from aquifer tests of two
wells to range from 0.1 to 0.2 m/d (Neaville, written commun.,
1990).
Near the sampled wells, hydraulic conductivity was estimated on
the basis of measured water-level declines and pumping rates
during sampling periods using the solution of the unsteady
ground-water flow equation (Lohman, 1979, eq. 44). The
estimates are within an order of magnitude at best, but are useful
for comparisons among the wells because of the similarities in
construction. The estimates range from 10 to 150 m/d for wells
in the alluvium or uppermost basin fill. Estimated hydraulic
conductivity for well 105, deeper in the basin fill, was 0.5 m/d. On
the basis of hydraulic gradients that range from 0.005 to 0.008
and assumed porosity of 0.2 or 0.3, average ground-water flow
velocities near the wells range from 0.02 to 3 m/d.
1150-
Webster
Guich
Miami Wash
Final Creek
X Colloid concentration
(mg/L)
— Line of equal pH
4,5,6, and 7
0 SCALE 5 km
I i ii i I
Vertical Exaggeration 50x
800
Figure 1. Hydrogeologlc Section of the Aquifer
-------
Perennial
Streamflow
33°32'
110°47'
L
Explanation
400 Well Site and
Number
Streamflow - Sample
09498400 Site and Number
Generalized Direction
of Flow
2 Miles
3 Kilometers
33°26' -
Former
Acidic Lake
Figure 2. Plan-View of Study Site
-------
Instrumentation and Methods
Ground water was collected during two field seasons from
twelve wells selected to represent the range of pH, solute
concentration, and hydraulic conductivity abng Final Creek
(Figures 1 and 2). Three different pumps were used (Table 1).
At the lowest discharge, velocities induced at the borehole face
were estimated to range from 1 to 5 times the average ground-
water velocity close to each well in the alluvium. In the basin fill
underlying the alluvium, even the lowest discharge resulted in
velocities more than 400 times that of the ground water.
Waterthat remained in the well between sampling sessions was
purged, as it was judged to be unrepresentative of formation
water. An inflatable packer was used with the bladder and low-
rate submersible pumps to reduce necessary purge volumes.
During purging, a Hydrolab Surveyor II1 with a flow-through cell
was used to monitor temperature, specific conductance, pH,
dissolved oxygen, and oxidation-reduction potential (Ft
electrode). Samples were collected only after each indicator
reached an acceptably stable value, generally a value that
changed by less than its measurement uncertainty during one
purge volume. From 3 to 24 volumes were purged before
sampling, and the high flow rate submersible pump generally
purged the larger volumes.
During the second field season, a Malvern Autosizer lie was
used to measure suspended particles in the diameter range from
0.003 to 3 urn. The instrument determines the size distribution
of suspended particles in this size range using laser light
scattering techniques together with photon correlation
spectroscopy. Particle-concentration estimates were based on
calibration curves constructed using linear correlation (r2=0.999)
between photon counts by the instrument and known
concentrations of kaolinite, a secondary clay mineral. The
kaolinite used was a reference standard obtained from the Clay
Minerals Repository at the University of Missouri. Kaolinite was
identified by Scanning Electron Microscopy with Energy
Dispersive X-Ray (SEM-EDX) on many of the filters from the
sampled wells. Other particles captured on filters and identified
by SEM-EDX included iron oxides, smectite, jarosite, silica, and
gypsum. Although the assumption that minerals in the reference
standard adequately represent the sum total of all the colloids in
the aquifer is not entirely true, photon counts provide at least a
relative measure of suspended particle concentrations.
Colloid concentrations took longer to stabilize than other field
indicators, about 50 percent longer than dissolved oxygen or
redox potential, and about twice as long as specif ic conductance,
pH, ortemperature. Well 107 was representative of the variation
of the indicators during purging at most of the wells (Figure 3).
Stable values of the indicators at selected wells are listed in
Table 2.
Samples were collected both in air and under nitrogen using a
field glove box. Unfiltered and filtered samples were collected,
the latter using 142-mm-diameter Millipore and Nucleopore
membrane filters ranging in pore size from 0.03 to 10.0 urn.
Samples were acidified in the field immediately after filtering with
double distilled concentrated nitric acid to pH < 2. Working in the
glove box was difficult, and handling thin membrane filters with
latex gloves was particularly cumbersome.
Elemental analyses were performed with inductively coupled
plasma (ICP) for most elements; atomic absorption with graphite
furnace (AAGF)for Cd, Pb, and As; and ion chromatography (IC)
for chloride and sulfate. Analytical precision on the ICP and
AAGF were <±10 percent, and on the IC< ±5 percent. Scanning
Electron Microscopy with Energy Dispersive X-Ray (SEM-EDX)
was used to identify colloidal material captured on the membrane
filters.
Effects of Sampling Variables
Pumping Rates
Differences in pumping rates were expected to cause differences
in the concentrations and size distributions of colloidal particles
in suspensions and differences in elemental concentrations after
filtration. Ten wells were purged and sampled with as many as
three different types of pumps in June 1988 and March 1989.
Pumping rates ranged from 0.6 to 92 L/min, corresponding to
velocities of 25 to 3900 m/d at the well screens. Samples were
filtered in air through 0.4-u.m filters, filtrates were analyzed for
Table 1. Pumps used In ground water sampling.
Brand'
GeoTech
Keck
Grundfos
Type
bladder
submersible
submersible
Power
Supply
compressed
air
12Vdc
240 V ac
Diameter
(mm)
44
44
95
Discharge
(L/min)
0.6-1.1
2.8-3.8
12-92
1 Use of brand names is for identification purposes only and does not imply endorsement by any agency of the
United States Government.
-------
8
7-
6-
5-
4-
3-
2-
1-
Photon Counts (10E4/S)
Redox Potential (v)
Specific Conductance
(10E3u.S/cm)
pH (units)
Dissolved Oxygen (mg/L)
0.0
1.5 3.0
Purge Volumes
4.5
Figure 3. Changes in Water Quality Indicators During Purging of Well 107 (Keck Pump, 3/89)
Table 2. Ground-water quality indicators for selected wells.
Well:
pH (units)
Sp.Cond.(nS/cm)
Temp.(°C)
Oxygen(mg/L)
Redox Pot.(v)
Colloids (mg/L)
104
3.92
3020
18.0
0.39
0.44
—
105
6.08
4300
19.0
—
0.28
0.4
107
3.48
7070
18.4
0.14
0.44
0.3
303
4.27
3210
19.0
0.01
0.37
—
403
5.05
3200
18.8
6.07
0.38
—
451
4.73
4060
18.9
0.24
0.25
20
503
5.74
3620
18.9
6.22
0.32
0.1
-------
cations using ICP, and the filters were examined using SEM-
EDX. Particle concentrations and size distributions were
monitored in 1989 for five wells on unfiltered samples.
Cation concentrations differed by less than 10 percent between
pumping rates for seven of the ten wells. These seven wells
generally had low particle counts, and low filter loading was
observed using SEM-EDX. Well 503, in the alluvium, was
representative of the seven wells. Figure 4 illustrates changes in
water-quality indicators in well 503, where the bladder pump was
used to purge and sample, followed by use of the low-rate and
high-rate submersible pumps. The well therefore was purged
with the bladder pump prtorto placement of the lattertwo pumps.
Colloid concentration stabilized at 0.1 mg/L during pumping at
1.1 L/min and increased to 0.7 mg/L when'discharge increased
to 3.8 L/min before stabilizing again at 0.1 mg/L. When discharge
increased to 30 L/min, however, colloid concentration initially
increased to 4.4 mg/L before finally stabilizing at 0.2 mg/L.
Particle-size distributions for the final sample with each pump are
also shown in Figure 4. The low-discharge pumps produced
monomodal distributions of the same size particles. The highest
discharge produced larger and slightly more particles in a bimodal
distribution because of increased turbulence. The predominant
mineral identified on the filters from well 503 was gypsum, which
was accompanied by some iron oxide, kaolinite, and other
particles that contained Fe+AI+S. Analytical concentrations of
metals did not differ significantly but did reflect the observed
mineralogy.
For samples from the three wells where observed cation
differences exceeded 10 percent, measured particle counts and
filter loading were also significantly higher than for the other
seven wells. Particle counts differed by factors of 5 to 130
between pumping rates. Cation concentrations differed by as
much as 50 percent for both major and trace elements. Cation
concentrations were generally highest in samples with the lowest
counts (leastturbid), but some anomalous behavior was observed
for some elements (Table 3). Pump-induced entrapment of
colloidal particles could decrease dissolved cation concentrations
by sorption on freshly exposed surfaces of particles which had
been retained on filters.
Differences in cation concentrations were especially noticeable
at well 105. In March 1989, pumping at 2.8 L/min mobilized 13
times more particles and decreased Ca, Mg, Mn, and Sr
concentrations by 10 to 25 percent, compared to pumping at 0.9
L/min (Table 3). For equal volumes of filtrate, SEM photographs
showed that the proportion of the area of a 0.1 u.m filter covered
with particles was about 1 percentforthe lower pumping rate and
about 30 percent for the higher rate. In June 1988, pumping at
12 L/min decreased concentrations of Ca, Mg, Mn, Co, Ni, and
Sr by 20 to 50 percent compared to pumping at 1 L/min. Well 105
is screened in the basin fill, which has the lowest ground-water
flow velocity in this study. Even at the lowest pumping rate, the
velocity induced at the borehole face was more than 400 times
the normal ground-water velocity. Water pumped from well 105
was visibly murky at times.
Pumping well 451 at 0.8 L/min produced seven times more
particles than pumping at 3.4 L/min and decreased concentrations
of six cations by 10 to 50 percent. Again, the less turbid water
generally had the larger concentrations, but the higher pumping
rate unexpectedly produced the less turbid water. This well had
the highest colloid concentrations of any well (Table 4). The
Table 3. Cation and colloid concentration*, mg/L, after purging at different rates (March 1989, 0.4 yum filter, sampled In air).
Well
105
105
451
451
503
503
503
Discharge (L/min)
Colloids
Ca
Mg
K
Fe
Mn
Al
Cu
Co
Ni
Sr
Zn
0.9
0.3
579
149
40
<.4
6.6
<.4
<.4
<.4
<-4
1.6
0.7
2.8
4.0
478
117
37
<.4
5.0
<.4
<.4
<.4*
<.4
1.4
0,9
0.8
20
586
150
16
156
108
6.5
6.4
1.5
0.3
1.9
3.0
3.4
3.0
623
162
13
151
113
10.0
12.4
1.8
0.4
2.1
3.9
1.1
0.1
703
148
12
<.4
76
<4
<4
<.4
0.6
2.3
0.06
3.8
0.1
704
146
11
<.4
76
• <.4
<.4
<.4
0.6
2.3
0.04
30
0.2
704
147
11
<-4
73
<.4 •
<.4
<.4
0.6
2.3
0.04
-------
«. I-
I-
4-
10 -
I-
20 30
Tims (min)
40
— Photon Counts (10E5/8)
Redox Potential (v)
Specific Conductance
(10E3nS/crn)
pH (units)
— Dissolved Oxygen (mg/L)
—o-
20
Time (min)
1.0 -
50
8
•8
Sampled at
60 min.
0123
Particle Size (|n )
1.0 -
35 min.
0 12
Particle Size (
c. 20
1!
16
14
12
10
I
I
4
2
9
t
10 20
Time (min)
1.0 -
30' min.
0123
Particle Size (fn )
Figure 4a-c. Changes In Water Quality Indicators During Purging of Well 503: (a) Bladder Pump; (b) Low Speed Submersible
Pump; (c) High Speed Submersible Pump
-------
Table 4. Pumping rate data for selected wells and pumps.
Well
105
303
403
451
452
503
Date
Sampled
6-14-88
3-7-89
6-15-88
6-15-88
3-9-89
3-9-89
6-16-88
3-8-89
Pump
Discharge
(L/min)
1.0
12
0.9
2.8
0.7
24
0.8
27
0.8
3.4
0.8
28
1.0
45
1.1
3.8
30
Intake
Velocity
(m/d)
42
510
38
120
30
1000
34
1100
26
110
28
980
42
1900
47
160
1300
Formation
Velocity
(m/d)
0.012
1.33
2.93
0.25
0.75
1.63
Relative
Velocity2
460
5500
410
1300
2.9
, 99
1.5
51
5.3
22
1.8
61
3.4
150
3.7
13
100
Colloid
Concentration
(mg/L)
0.3
4.0
20
3.0
0.2
10
0.1
0.1
0.2
! Ratio of induced velocity at the borehole face to average ground-water velocity in the adjacent formation.
samples collected in March 1989 were noticeably turbid, even
after 2 hours of purging with the bladder pump. In fact, tower
particle counts by the slightly higherrate pump may have resulted
from the additional purge time, as the latter was inserted following
purging and sampling with the bladder pump. Two factors may
contribute to the high colloid concentrations at well 451:
• it is in relatively finegrained sediment inthe alluvium,
and
• it is in a part of the aquifer where pH is changing
rapidly and iron oxide coatings on colloidal clay are
dissolving.
For this data set, particle concentrations were not predictable
from pumping rate, purge volume, flow velocity at the screen, or
the ratio of velocity induced at the borehole face to local ground-
water flow velocity. Measured particle concentrations appear to
depend on interactions of these factors as well as geology, well
construction, and water chemistry.
Filtration Differences
Concentration differences among samples filtered through pore
sizes ranging from 0.1 to 10 u.m were generally less than 10
percent. Only wells 303 (Table 5) and well 503 had differences
of greaterthan 10 percent in most elemental concentrations. The
larger differences commonly were associated with use of the
high-rate submersible pump, and concentrations generally
increased with increasing filter-pore size.
Differences less than 10 percent generally were observed for
waters that have pH less than 4, which does not favor colloid
formation. The largest observed differences for well 403, for
Table 5. Cation concentrations, in mg/L, for well 303 using
different filters (June 1988,24 L/min, sampled in air).
Element
0.1
0.4 fun
10fun
Ca
Mg
K
Fe
Mn
Al
Co
Cu
Ni
Zn
391
91
4.73
171
37.7
6.74
0.68
15.0
0.68
2.75
424
100
5.49
87
40.8
7.61
0.75
16.7
0.75
3.27
492
20
9.76
211
45.5
9.93
0.86
19.2
0.88
4.13
example, were for Al, Cu, Fe, and Mg, but no consistent trend of
concentration with filter pore size is apparent (Table 6).
Filtration differencesof greaterthan 10percentwerealsogenerally
associated with use of the high-rate submersible pump because
of the increased entrainmentof particulates as observed above.
-------
Tabla 6. Cation concentrations, In mg/L, for well 403 using
different filters (June 1988,0.8 L/mln, sampled in air).
Table 7. Cation concentrations, in mg/L, for samples collected in
air and nitrogen under atmosphere (mg/L, 0.40-ftm filter,
< 1 L/mln).
Element
Ca
Mg
K
Fe
Mn
Al
Co
Cu
Ni
Zn
0.1 \m .
533
133
5.58
0.45
34.6
1.22
0.36
1.62
0.41
0.90
0.4 jim
533
113
5.47
6.63
34.9
1.91
0.36
2.14
0.42
0.95
10 urn
554
116
5.65
1.22
34.7
1.17
0.37
1.57
0.44
1.60
Oxidation of Samples
Oxidation of samples during sample collection, filtration, and
preservation general!/ resulted in substantial differences in most
wells between samples collected under nitrogen or in air. Work
by Holm et al. (1988) showed that diffusion of atmospheric gases
through pump tubing can introduce measurable concentrations
of oxygen into waters initially low in dissolved oxygen. This
source of possible contamination for both sets of samples was
minimized by collection of samples adjacent to the wellhead.
Samples collected in air were directly exposed to atmospheric
gases during filtration and acidification procedures. Significant
differences (>10 percent) were observed in many of the wells.
Variations in differences from well to well may have been caused
by a number of different factors including:
• slightly different exposure times to air, depending
on water-table depth and duration of filtration and
preservation,
• dissolved-oxygen level,
• redox potential (Eh), and
• dissolved iron concentration.
Large differences in concentrations were measured for well 303,
where dissolved iron concentration was greater than 200 mg/L
(Table 7). Differences similar to those for well 303 were also
observed in wells 51,104, and 403. In contrast, the differences
were small for well 503, where the dissolved iron concentration
was less than 0.1 mg/L.
Another indication of the extent of oxidative effects on sample
integrity was reflected in Eh values determined by various
methods for well 51 (Figure 1). The field-measured Eh value
using a R electrode was 0.43 V. A calculated Eh value, assuming
equilibrium between Fe3* and Fe(OH)3, was 0.57 V (Stollenwerk
and Eychaner,1989). In March 1989, Fez*and Feto(al for well 51
Element
Fe
Mn
Cd
Co
Cu
Ni
Zn
Well 303
air nitrogen
Well 503
air nitrogen
177
37.4
0.02
0.69
15.5
0.70
2.53
215
44.7
0.02
0.82
18.6
0.84
3.11
0.04
68.3
0.01
0.01
0.01
0.47
0.21
0.09
68.7
0.01
0.02
0.04
0.48
0.30
were determined within one week of sample collection; Fe3* was
computed by difference, and Eh was calculated from the ratio of
Fe3* to Fe2*. The calculated Eh was 0.51 V for the sample
collected and analyzed in a nitrogen atmosphere and 0.76 Vfor
the sample collected in air. Samples collected in the glove box
were transported in nitrogen-pressurized containers, and the
determinations were performed in laboratory glove boxes also
pressurized with nitrogen.
Several possible errors are associated with all these Eh evaluation
methods. Lindberg and Runnells (1984) showed that many field
Eh measurements may not reflect true redox conditions in
ground waters. However, in acidic waters such as these, field
measurements using R electrodes may be valid (Nordstrom et
al. 1979). Values calculated from equilibrium constants rely on
the assumption that Fe(OH)g is the predominant solubility
controlling phase. Stollenwerk and Eychaner (1989) used the
equilibrium expression:
Fe!>* + 3H2O-
Fe(OH)3 + 3H* log KT = -4.891,
although other values have been reported for this reaction.
Samples collected under nitrogen may have received some
exposure to oxygen during sample collection, processing, and
analysis. The Fe3* values for March 1989 were small differences
between two large numbers and are uncertain. Irrespective of
these and other limitations in estimating Eh, the large difference
observed between 0.76 V for the sample collected in air and the
other Eh values for well 51 demonstrates the extent of oxidation
that can occur if care is not taken to limit oxygen exposure during
sample -collection activities in suboxicand anoxic environments.
Conclusions
Research at this site indicates that monitoring of water-quality
indicators during well purging and sampling is important. In
addition to the indicators most often monitored, turbidity also
needs to be evaluated before collecting samples. In lieu of the
use of a turbidimeter, purging for twice the time required for
dissolved-oxygen equilibration may be a good rule of thumb.
10
-------
The use of a low flow rate pump can minimize entrapment of
nonmobile suspended participates, oxygenation of formation
water, and mixing of adjacent, possibly geochemically distinct,
ground waters. Collection and processing of anoxic or suboxic
ground water excluding atmospheric gases to the extent possible
is desirable for representative and accurate water-chemistry
data. The glove box used for collection of samples under
nitrogen was cumbersome and difficult to use, especially in
handling the thin membrane filters. If tubing of minimum length
and maximum thickness were used, in-line filtration would probably
mitigate the oxidation effects observed in the present study,
making the use of a field glove box and accompanying nitrogen
cylinders unnecessary. Although filtration differences generally
were not significant at this site, trends indicate that care needs to
be taken in selection of filter pore size and that samples need to
be filtered in the field. Additional research is needed at sites with
distinctly different hydrology, geology, and chemistry before final
recommendations can be made concerning filtration. In the
interim, collection of filtered and unfiltered samples forcomparison
purposes is suggested for at least a fraction of the samples
collected. Filtered samples are needed for accurate aqueous
geochemistry estimations, and unfiltered samples provide
conservative estimates of contaminant mobility.
The sampling recommendations proposed by the RSKERL1988
workshop participants were realistic and relatively easy to apply
in the present study. Additional time was required for purging and
sampling, but the additional care was warranted to obtain ground-
water chemistry data which were as representative as possible.
Disclaimer
The information in this document has been funded wholly or in
part by the United States Environmental Protection Agency. This
document has been subject to the Agency's peer and
administrative review and has been approved for publication as
an EPA document.
Acknowledgements
The authors gratefully acknowledge the support of Terry F. Rees,
U.S. Geological Survey, Denver, CO, forthe SEM-EDX analyses;
Donald Clark, Robert S. Kerr Environmental Research Laboratory,
Ada, OK, for the ICP and AAGF analyses; and Narong
Chamkasem, NSI Technology Services Corporation, Ada, OK,
for the 1C analyses.
References
Buddemeier, R. W. and J.H. Rego. 1986. Colloidal Radionuclides
in Groundwater. Annual Report. Lawrence Livermore National
Laboratory, Livermore, CA. UCAR 10062/85-1.
Claassen, H.C. 1982. Guidelines and techniques for obtaining
water samples that accurately represent the water chemistry of
an aquifer: U.S Geological Survey Open-file Report 82-1024,
49p.
Danielsson, L.G. 1982. On the Use of Filters for Distinguishing
Between Dissolved and Particulate Fractions in Natural Waters.
Water Res. 16:179.
Eichholz, G.G., B.G. Wahlig, G.F. Powell, and T.F. Craft. 1982.
Subsurface Migration of Radioactive Waste Materials by
Particulate Transport. Nuclear Technology 58:511.
Enf ield, C.G. and G. Bengtsson. 1988. Macromolecular Transport
of Hydrophobic Contaminants in Aqueous Environments.
Groundwater 26(1 ):64.
Eychaner, J.H. 1989. Movement of inorganic contaminants in
acidic water near Globe, Arizona, in Mallard, G.E., and Ragone,
S.E., eds., U.S. Geological Survey Toxic Substances Hydrology
Program—Proceedings of the technical meeting, Phoenix,
Arizona, September 26-30,1988: U.S. Geological Survey Water-
Resources Investigations Report 88-4220, p. 567-575.
Eychaner, J.H., M.R. Rehmann, and J.G. Brown. 1989. Chemical,
geologic, and hydrologic data from the study of acidic
contamination in the Miami Wash-Pinal Creek area, Arizona,
water years 1984-87: U.S. Geological Survey Open-File Report
89-410,105 p.
Gschwend, P.M. and M.D. Reynolds. 1987. Monodisperse Ferrous
Phosphate Colloids in an Anoxic Groundwater Plume. J. of
Contaminant Hydrol. 1:309.
Holm, T.R., G.K. George, and M.J. Barcelona. 1988. Oxygen
Transfer Through Flexible Tubing and its Effects on Ground
Water Sampling Results. Ground Water Monitoring Review. Vol.
8(3):83.
Lindberg, R.D., and D.D. Runnells. 1984. Ground Water Redox
Reactions: An Analysis of Equilibrium State Applied to Eh
Measurements and Geochemical Modeling. Science, 225:925-
927.
Lohman, S.W. 1979. Ground-water hydraulics: U.S. Geological
Survey Professional Paper 708, 70 p.
Nordstrom, D.K., E.A. Jenne, and J.W. Ball. 1979. Redox Equilibria
of Iron in Acid Mine Waters, In Chemical Modeling in Aqueous
Systems, American Chemical Society Symposium Series 93,
E.A. Jenne (ed.) p. 51 -79.
Peterson, N.P. 1962. Geology and ore deposits of the Globe-
Miami district, Arizona: U.S.Geological Survey Professional Paper
342,151 p.
Puls, R.W. and M.J. Barcelona. 1989. Ground Water Sampling
for Metals Analyses. EPA/540/4-89/001.
Puls, R.W., L.L. Ames and J.E. McGarrah. 1989. The Use of
Batch Tests as a Screening Tool for Radionuclide Sorption
Characterization Studies, Hanford, Washington, U.S.A. Applied
Geochemistry. 4:63-77.
Robertson, W.D. 1984. Contamination of an Unconfined Sand
Aquifer by Waste Pulp Liquor: A Case Study. Ground Water
22(2):191.
Sheppard, J.C., M.J. Campbell and J.A. Kittrick. 1979, Retention
of Neptunium, Americium and Curium by Diffusible Soil Particles.
Environ. Sci. Technol. 13(6), 680-684.
11
-------
Sheppard, J.C., M.J. Campbell, T. Cheng and J.A. K'rttrick. 1980.
Retention of Radionuclides by Mobile Humic Compounds.
Environ. Sci. Technol. 14(11). 1349-1353.
Stollenwerk, K.G. and J.H. Eychaner. 1989. Solubility of Aluminum
and Iron in Ground Water NearGlobe, Arizona. In U.S. Geological
Survey Toxic Substances Hydrology Program—Proceedings of
the Technical Meeting, Phoenix, Arizona, September 26-30,
1988.
Takayanagi, K. and G.T.F. Wong. 1984. Organic and Colloidal
Selenium in South Chesapeake Bay and Adjacent Waters.
Marine Chem. 14:141-148.
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati, OH 45268
BULK RATE
POSTAGE & FEES PAID
EPA PERMIT NO. G-35
Official Business
Penalty for Private Use $300
EPA/600/M-90/023
------- |