4>EPA
                          United States
                          Environmental Protection
                          Agency
                         Robert S. Kerr Environmental
                         Research Laboratory
                         Ada, OK 74820
Research and Development     EPA'600'M-91/040  July 1991

ENVIRONMENTAL
RESEARCH   BRIEF
        Facilitated Transport of Inorganic Contaminants in Ground Water:
                                Part II.  Colloidal Transport

                Robert W. Pulsa, Robert M. Powellb, Don A. Clark3 and Cynthia J. Paul6
ABSTRACT

This project consisted of both field and laboratory components.
Field studies evaluated routine sampling procedures for
determination of aqueous inorganicgeochemistry and assessment
of contaminant transport by colloidal mobility. Research at three
different metal-contaminated sites has shown that 0.45 u.m
filtration has not removed potentially mobile colloids, when samples
have been collected using low pumping flow rates (-0.2-0.3 LJ
mm). However,  when  pumping velocities greatly exceeded
formation ground-waterflow velocities, large differences between
filtered and unfiltered samples were observed, and neither were
representative of  values obtained with the  low flow-rate
pumped samples. There was a strong inverse correlation between
turbidity and representativeness of samples. Several different
sampling devices were evaluated in wells (PVC) ranging in depths
from 10 to 160 ft. Those devices which caused the least disturbance
(i.e., turbidity) also produced the most reproducible  samples
irrespective of filtration. The following  water quality indicators
were monitored  during well  purging:  dissolved O2,  pH,  Eh,
temperature, specific conductance, and turbidity. Sampling was
not initiated until all indicators had reached steady-state (usually
~ 2 to 3 casing volumes). In all cases turbidity was slowest to reach
steady-state  values, followed by dissolved oxygen and redox
potential. Temperature, specific conductance, and pH results
were generally insensitive to well purging variations.
• U.S. EPA, Robert S. Kerr Environmental Research Laboratory,
  Ada, OK.
* ManTech Environmental Technology, Inc., Robert S. Kerr
  Environmental Research Laboratory, Ada, OK.
                        In controlled laboratory experiments, the stability and transport of
                        radio-labeled Fe203 model colloids were studied using batch and
                        column techniques. Variables in the study included flow rate, pH,
                        ionic strength, electrolyte composition (anion/cation), colloid
                        concentration, and colloid size. Transport was highly dependent
                        upon colloidal stability. Iron  oxide colloids in the 100-900 nm
                        particle diameter range were not only mobile to a significant
                        extent, but under some hydrogeochemical conditions  were
                        transported faster than a conservative tracer, tritium. Extent of
                        colloidal breakthrough was dependent upon a complex variety of
                        parameters, however the highest statistical  correlation was
                        observed with particle size and anionic composition of the
                        supporting electrolyte. The dissolved transport of arsenate, a
                        ubiquitous priority metal contaminant at hazardous waste sites,
                        was compared with that of colloid-associated arsenate transport.
                        The rate of colloid-arsenate transport was over 21 times that of
                        the dissolved arsenate.

                        INRODUCTION

                        Understanding the transport and fate of inorganic contaminants
                        in the subsurface environment has been complicated by recent
                        field studies that show contaminant mobility to be greater than
                        had been predicted.  These predictions have been based on
                        properties such as speciation, solubility, ion exchange, and
                        sorption-desorption but have failed to account for the potential
                        interactions between inorganic contaminants and mobile colloids.
                        Colloids are generally considered to be particles with diameters
                        less than  10 micrometers (Stumm and Morgan, 1981), and can
                        include both organic and inorganic materials. In addition to
                        having a high surface area per unit mass and volume, particles
                                                                             Printed on Recycled Paper

-------
of dissolved organic carbon, clay minerals and iron oxides are
also extremely reactive sorbents for radionuclides and other
contaminants.  Drastic changes in aqueous geochemistry can
bring about supersaturated conditions in which inorganic colloidal
species are formed.  Decreases in pH or changes in redox
potential can cause the dissolution of soil or geologic matrix
cementing agents, promoting the release of particles. Decreases
in the ionic strength of the aqueous phase can enhance colloidal
stability and promote their transport. If mobile, these particles
may increase the mobility of sorbed contaminants above that
predicted by ourcurrentsimpledissolved-phasetransport models.

Several studies have demonstrated the presence of such colloidal
material in ground water, with indications that colloidal mobility
may facilitate the transport of contaminants in some systems.
These  studies  have also provided data on the  size range of
mobile colloids, with  evidence that particles with diameters
greater than 1  nm may actually move faster than the average
ground-water flow velocity in porous media due to effects such as
size exclusion  from smaller pore spaces. Other studies have
demonstrated  the strong binding and  enormous sorption
capacities of  colloidal particles for  inorganic and organic
contaminants.   The  significance of  colloidal mobility  as  a
contaminant transport mechanism ultimately  depends on the
presence of sufficient quantities of reactive particles in  ground
water. With current techniques, our ability to differentiate between
naturally suspended  particles and  those which are brought
artificially into suspension during  sample acquisition  is
questionable.

Inherent to this problem is the arbitrary designation of 0.45 urn as
an operational cutoff point for distinguishing between particulate
and dissolved species. If particles as large as 1 -2 n.m are mobile,
present in significant quantities,  and capable of transporting
contaminants  long distances, then sampling protocols  must
quantify this component of transport.  Transport models must
then incorporate this mechanism to provide better contaminant
migration predictions.

Colloidal Reactivity, Mobility and Size

Numerous studies demonstrate the strong adsorptive capabilities
of  secondary  clay  minerals, hydrous iron, aluminum  and
manganese oxides and humic material (Sheppard et al. 1979;
Takayanagi and Wong, 1984; Sandhu and Mills, 1987; Means
and Wijayaratne, 1982). In studies at underground nuclear-test
cavities at the Nevada Test Site, Buddemeier and Rego (1986)
found that virtually all of the activity of Mn, Co, Sb, Cs, Ce and Eu
was associated with colloidal particles in ground-water samples.
Sandhu and Mills (1987) found over 90% of the chromium and
arsenic present in a laboratory column study was associated with
colloidal iron and  manganese oxide.   Nelson et al. (1985)
determined that colloidal  organic carbon was a major factor
controlling the distribution of plutonium between the solid and
dissolved phases.

The mobility of these reactive  particles  has  already  been
demonstrated. Gschwend and Reynolds (1987) concluded that
submicron ferrous phosphate particles were suspended and
presumably mobile in a sand and gravel aquifer. These particles
were formed from sewage-derived phosphate that combined
with iron released from aquifersolids by reduction and dissolution
of ferric iron. Size distribution analyses indicated a large population
of 100 nm particles, and a lesser quantity in the range 600-800
nm. In complementary laboratory experiments with sand columns
and carboxylated polystyrene beads ranging in size from O.IOto
0.91 nm as model colloids, Reynolds (1985) recovered 45% of
the 0.91 |im size beads, and greater than 70% of 0.10 and 0.28
u,m size beads.  Field studies by Nightingale and Bianchi (1977)
showed that, under certain conditions, submicrometer-sized
particles within the surface weathered zone were mobilized for
some distance both vertically and laterally and affect ground-
water turbidity. In laboratory tests, Eicholz etal. (1982) found that
cationic nuclides were competitively adsorbed on suspended
clay particles capable of travelling at bulk water flow velocity in
porous mineral columns.   Particulate  matter of micrometer
dimensions was shown to be responsible for the transport of
radioactive sodium and ruthenium in sand beds by Champlin and
Eichholz  (1968).  As  much as 200  ug/L copper, lead, and
cadmium were found to be associated with colloidal material in
the size range  0.015-0.450 u.m by Tillekeratne et  al.  (1986).
Rapid transport of plutonium in core column  studies by Champ et
al. (1982) was attributed to colloidal transport, with 48% of the
plutonium associated with particles in the size range 0.003-0.050
u.m and 23% in the range 0.050-0.450 u,m.

Harvey et al. (1989) showed that,  in a shallow sand and gravel
aquifer, 1.35 u.m latex particles traveled faster than the 0.23 u.m
size particles.  This phenomenon was  due to size exclusion
effects (reduced  path  length), similar  to what Enfield and
Bengtsson  (1988) observed in laboratory columns with organic
macromolecules. Penroseetal. (1990)found detectable amounts
of plutonium and americium, 3390 m  downgradient  from a
source, to  be tightly or irreversibly associated with particles
between 0.025 and 0.45 u,m in size.  Champlin  and Eichholz
(1976) demonstrated that previously  "fixed" particles and
associated contaminants may be remobilized by changes in the
aqueous geochemistry of the system. Cerda  (1987) demonstrated
that the mobilization of kaolinite  fines in laboratory columns was
almost totally dependent upon the chem istry of the fluids present,
with maximum  mobility occurring under relatively high saline,
weakly alkaline pH conditions.   Repulsive colloidal forces
promoting stability were in evidence up  to 0.1  M NaCI (pH 9).
Malijevic et al. (1980) studied the stability and transport of
hematite spheres through packed-bed  columns of stainless-
steel beads as a function of pH and the concentration of a variety
of simple and complex electrolytes. Surface charge alterations of
the hematite and stainless-steel beads by the different electrolytes
was the dominant factor in colloidal deposition and detachment.

Recent estimates of colloidal concentrations in  ground  water
range as high as 63 mg/L (Buddemeier and Hunt,  1988), 60 mg/
L (Ryan and Gschwend, 1990), and 20 mg/L (Puls and Eychaner,
1990). Given the demonstrated  high binding capacity of many of
these particles, concentrations  of this magnitude could have a
significant impact on contaminant transport.

Dissolved vs. Particulate

Historically, 0.45 u,m pore size  filters have been used to
differentiate between dissolved and particulate phases in water
samples.   If the intent of the filtration is to  determine truly
dissolved constituent  concentrations for geochemical
modeling purposes, the inclusion of colloidal  material less
than 0.45 u.m in the filtrate will result in incorrect dissolved
values. This result is often observed with iron and aluminum,
where  "dissolved"  values  are   obtained  that  are
thermodynamically impossible at the sample pH. Conversely,
if the purpose of sampling is  to estimate "mobile" species in
solution,  including both  dissolved  and  particle-associated

-------
contaminants, significant underestimations of mobility may result,
due to removal of colloidal matter by 0.45 u.m filtration.

Kim et al. (1984), found the majority of the concentrations of rare
earth elements to  be associated with colloidal species that
passed through a 0.45 u.m filter. Wagemann and Brunskill (1975)
found more than two-fold differences in total iron and aluminum
values between 0.05 and 0.45 u.m filters of the same type. Some
aluminum compounds, observed to pass through a 0.45 urn filter,
were retained on a 0.10 u.m filter, by Hem and Roberson (1967).
Kennedy et al. (1974) found errors of an order of magnitude or
more in the determination of dissolved concentrations of aluminum,
iron, manganese and titanium using 0.45 |im filtration as an
operational definition for "dissolved."  Sources of error were
attributed to filter passage of fine-grained clay particles.

Sampling Objectives and Recommendations

A common and overriding ground-water sampling objective is
the acquisition of  representative  and accurate elemental
concentrations for the purpose of risk assessment at hazardous
waste sites.   In addition to dissolved species, this should
include contaminants sorbed to suspended (mobile) inorganic
and  organic  particles.   Disturbance  of the  subsurface
environment is inevitable in the process of installing monitoring
wells and collecting samples.  Artifacts or contamination of
samples can occur from the  following: poor well design or
construction;  inadequate or improper well development;
corrosion, degradation, or leaching of well  construction
materials; improper well purging, sampling, sample processing,
transportation and  storage.  Intuitively, it makes sense to
minimize disturbance of the sampling zone to  obtain
representative and accurate data.  Excessive turbidity is the
most common manifestation of disturbance.  Turbidity results
from stirred  up or  suspended sediment or foreign particles.
Natural turbidity may exist where conditions are favorable for
the production of stable suspensions (e.g. low ionic strength
waters,  geochemical supersaturation, high clay content),
whereas excessively rapid pumping or purging relative to local
hydrogeological conditions is  the  most common cause  of
artificial turbidity. Oxidation of anoxic or suboxic aquifer zones
may result from high pumping rates which impact much larger
segments of the aquifer than the interval of interest, causing
the precipitation of iron oxyhydroxide and/or mixing of chemically
distinct zones.

If  a  secondary objective is accurate "dissolved" elemental
concentrations, then samples should be filtered in the field with
in-line devices and using filter pore sizes < 0.1  u.m.

Sampling recommendations consistent with the above discussions
and  recommendations were summarized  in  a previous
Environmental Research Brief (Puls et al. 1990).  Briefly these
recommendations included:

  • Isolation of the sampling zone with packers to  minimize
    purge volume,
  • Low flow rate pumping to minimize aeration and turbidity,
  # Monitoring  of water quality  parameters while purging to
    establish baseline  or  steady-state  conditions to initiate
    sampling,
  • Maximize pumptubing wall thickness and minimize length to
    exclude atmospheric gases,
  • Filtration for estimate of dissolved species and the collection
    of unfiltered samples for estimates of contaminant mobility.
FIELD STUDIES

Sites

Between June 1988, and February 1991, three different field
sites were used to evaluate the above sampling techniques and
recommendations, emphasizing the impacts of different sample
collection  devices on sample turbidity and filtration effects  on
metals concentrations.  The first site is at Final Creek, near
Globe, Arizona, about 130 km east of Phoenix and 170 km north
of Tucson. This site and sampling results were discussed in
detail in Pulsetal. (1990).

The second is a Superfund site near Saco, Maine, about 30 km
south of Portland and 7 km inland from the coast. This site was
used for chromium waste disposal by a leather tannery from 1959
to 1977. Chromium wastes were dumped into 53 small unlined
pits and two larger lagoons (each about 0.25 hectares). The site
geology consists of  glacial sediments, underlain by a sloping
fractured bedrock surface. Sediment thickness ranges from 0 to
17 m. Analysis of ground-water flow is complicated by both the
fractured nature of the bedrock and an apparent ground-water
divide in the overburden. Upward gradients have been determined
at some locations associated with ground-water discharge to
surface drainages. Depth to water table ranges from 1 to 2 m
below ground surface.

The third site is near Elizabeth City, North Carolina, about 100 km
south of Norfolk, Virginia, and 60 km inland from the Outer Banks
of North Carolina. A chrome plating shop, in use for more than 30
years, has discharged acidic chromium wastes into the soils and
aquifer immediately below the shop. The site geology consists of
typical Atlantic coastal plain sediments characterized by complex
and variable  sequences  of surficial  sands,  silts and  clays.
Ground-water flow is generally to the northeast; however, in the
immediate vicinity of the plating shop, flow appears to be directly
toward the Pasquatank River about 90 m  north of the shop.
Ground-water flow is somewhat complicated due to wind tides.
Depth to ground water is about  2 m.  An estimated hydraulic
conductivity value of 15 m/d was  based on aquifer test data.

Site Sampling

The sampling set-up used at the Saco and  Elizabeth City sites is
depicted in Figure 1. This was similar to the arrangement used
in Globe, where a laser light scattering instrument was used  for
tracking particle concentrations instead of aturbidimeter; and a
bladder or submersible pump was used instead of a peristaltic
pump (Puls et al. 1990).  For comparison, a bailer was used in
addition to the peristaltic pump at Saco and at Elizabeth City. A
multiparameter instrument with flow-through cell was employed
in all cases to monitor pH, temperature, specific conductance,
redox potential, and  dissolved oxygen during both purging and
sample collection operations.  Sample collection was initiated
when all parameters, including turbidity, reached steady-state.
Figure 2, for well  MW101 at Saco, is typical of  the trends  for
parameter equilibration during purging. Specific conductance,
pH and temperature, although recommended by many sampling
guidelines, were the least sensitive parameters, attaining steady-
state values rather rapidly.    Corresponding contaminant
concentrations in addition to turbidity, redox, and dissolved
oxygen, are also plotted in Figure 3 for well 1 at Elizabeth City.
Chromate concentrations were shown to follow trends similar to
those for equilibration-sensitive parameters.

-------
Figure 1. Sampling set-up for Saco, Maine, and Elizabeth City,
        North Carolina, sites (shallow wells).
Suspended Particles and Sampling Devices

At the Globe, Arizona site, comparisons were made between
values of suspended particle concentration and particle size
distributions using the following pumps: bladder (0.6-1.1 L/min),
low speed submersible (2.8-3.8  L/min),  and high speed
submersible (12-92 L/min).   Particle size distributions  were
measured with a laser light scattering instrument using photon
correlation spectroscopy (Malvern AutoSizer IIC). Particles were
captured on filters and identified by scanning electron microscopy
with energy dispersive X-ray (SEM-EDX).
  20
  15
  10-
                                          ORP  (VxlOO)
             20
                     40
                              60
                                      80
                     Time Since Startup (Minutes)
                   10
                   0
12
0
                1.0
2.0         3.0

Casing Volume
                                                  4.0
Figure 2. Equilibration of ground-water quality parameters during
        well purging (well MW 101, Saco, Maine, peristaltic pump
        (-0.2 L/min)).
                                                             50
                                                             40
                                                             30
                                                             20
                                                             10
                                                                                                                 1.0
                                                                                                                 0.8
                                                                                                                 0.6  O
                                                                                                                 0.4
                                                                                                                 0.2
                                                               0    10   20   30    40    50    60    70    80    90   100
                                                                                      Time (min)


                                                               -»-ORP(Vx100) -•-  DO(mg/Lx10)  -•- Turbidity (ntu)  « Cr(mg/L)
Figure 3. Equilibration of most sensitive ground water quality
        parameters and chromate concentration during well
        purging (well 1, Elizabeth City, North Carolina, peristaltic
        pump (-0.2 L/min)).

In well 105, more than 13 times more particles were mobilized by
the low speed submersible pump compared to the bladder pump.
This well is screened in the dense basin fill, where hydraulic
conductivities are more than two orders of magnitude lower than
those in the upper alluvium.  Particles captured on filters were
identified as iron-coated  albite, gypsum and calcite.  This lower
region  of the aquifer is  saturated with  respect to calcite, but
unsaturated with gypsum. Gypsum particles were not present in
the bladder pump samples, but were present in the submersible
pump samples, probably due to mixing of the upper and lower
aquifer waters caused by the relatively high pump rate. Differences
in water chemistry from the bladder pump and  high-speed
submersible pump samples in June 1988, supportthis hypothesis
(Brown, 1990).  Interestingly, the calcite particles in the bladder
pump samples were uniformly spherical and approximately 1 u,m
in diameter.

In well 452, screened in the alluvium, over 20 times more particles
were mobilized by the high speed submersible pump compared
to the bladder pump.  These particles were captured on 0.1 (j.m
filters and identified with SEM-EDX as predominantly smectite
clays. Their presence was probably due to the fact that well 452
is screened in relatively finegrained sediment in the alluvium, and
it is near the leading edge of the acidic waste plume where pH is
decreasing rapidly and iron oxide coatings on colloidal clay are
dissolving.

In well 503, also in the upper alluvium, successive use of the
bladder, and the low speed and high speed submersible pumps
produced  increasingly more numerous and larger particles  in
suspension. The two low-discharge pumps produced monomodal
particle size distributions of approximately 500 nm.  The high
discharge submersible  pump  produced larger  particles in a
bimodal distribution centered around 800 and 2000 nm because
of increased turbulence.  The predominant mineral identified on
the filters from well 503 was gypsum.  The upper alluvial aquifer
is supersaturated with respect to gypsum due to the dissolution
of calcite by sulfuric acid-dominated wastes which have leached
into the subsurface for over 80 years.

-------
Even with the bladder pump, particles brought to the surface
were as  large as 10 u.m, probably too large to be naturally
suspended in situ, but there was clearly a significant difference
in particle population and size between the three different pumps.
Increasing pump rate generally  resulted in increased turbidity
with larger particles brought into suspension. Some anomalous
behavior in this regard was caused by the sequence of pump
comparison in a given well (Puls et al. 1990).

A  peristaltic pump was used at both the Saco,  Maine and
Elizabeth City, North Carolina sites because of the shallow depth
to ground water. As a result, lower pumping rates were possible
(0.2-0.3 L/min). Turbidity due to suspended colloids was measured
with a turbidimeter equipped with flow-through cell (Figure 1).
Turbidity is commonly measured in nephelometric turbidity units
(NTUs) and based on the comparison of the intensity of light
scattered by a sample with the intensity of  light scattered  by a
standard reference  suspension under the same  conditions.
Formazin polymer is used as the  reference.  Formazin has been
found to be more reproducible than clay or natural water (Standard
Methods for the Examination of Water and Wastewater. 1989).
Jackson candle units (JCUs)  were previously used with candle
turbidimeters, and kaolin was a standard  reference material.
Table 1 lists correlations between NTUs, JTUs, and counts/1 OOO/
sec, as measured by two turbidimeters and a laser light scattering
instrument respectively, using kaolinite as a common reference.
Counts/1000/sec are photon counts recorded from laser light
scattering measurements.

At Saco,  most wells equilibrated  at less than 5 NTUs; however,
two wells equilibrated at 10 and 58 NTUs. The latter was an older
well, and age or improper installation may explain the  high
turbidity value. At Elizabeth City, only 1 of 12 wells equilibrated
at  more  than 5 NTUs.  All of these wells  were installed and
developed  by R.S.  Kerr Laboratory  personnel using  best
available technology and guidance. The well which had the
highest equilibrated NTU value was well 8, the only well located
(screened) in a clayey zone  at the site (Figure 4).
               10    20     30     40     50

                    Tims (mm after pump on)
                  - MW 12 SH Sm
                                   . UW-9 FneSnl
Figure 4. Equilibration of turbidity levels during well purging for
        •everal well* at the Elizabeth City site (peristaltic pump,
            L/min).
A down-hole camera was used at Elizabeth City during purging
and sampling toevaluatethedisturbance caused by emplacement
and pumping.  Little impact was observed when the peristaltic
pump was turned on after both the pump tubing and the camera
had been left in the screened interval overnight.  Emplacement
of the camera itself created the greatest turbidity and required
overnight re-equilibration in the absence of pumping. Purging at
low flow rate produced approximately the same result, in terms
of turbidity, as did overnight equilibration. These observations
argue strongly for dedicated sampling equipment as the optimal
and perhaps most efficient manner of collecting representative
ground-water samples.
Table 1. Comparison of nephelometric turbidity units (NTUs),
        Jackson candle units (JCUs), photon counts from laser
        light sea tiering (cts/1000/sec) and kaolinite concentrations
        in water (mg/L).
         NTUs
JCUs
cts/1000/sec
mg/L
0.2
2.7
12.2
25.2
63.0
121.0
227.0
_
3.0
7.0
10.0
28.5
53.0
100.0
2.3
20.5
77.7
175.9
438.2
699.8
1412.2
0.1
1.0
5.0
10.0
25.0
50.0
100.0
Filtration and Sampling Devices

Filtration differences among the different sampling devices used
at the Arizona site were generally not significant. Greater than
10% differences were observed in some wells, particularly with
the high speed submersible pump, due to artifacts from  the
excessive turbidity created down-hole by the pump compared to
the natural hydrogeological conditions (Puls et al. 1990).  Similar
filtration studies at Saco and Elizabeth City produced much more
dramatic results.

Figure 5 shows chromium levels in samples from well 1 at
Elizabeth City where samples were collected both with a peristaltic
pump (200 ml/min) and with a bailer.  The purge time for water
quality parameter equilibration using the peristaltic was 1.3 hr, or
about two casing volumes (Figure 3).  Bailed samples were
collected after a standard three casing volumes had been bailed.
There were no significant differences in chromium concentrations
between unfiltered  and 5.0, 0.4 and 0.1 u.m pore size  filtered
samples with the peristaltic pump. The bailed samples were not
only significantly different, but were also 2 to 3 times higher than
the peristaltic values.  A similar response was observed in well 8.
In all twelve wells, there were no differences observed in metal
concentrations over the entire range from unfiltered to 0.1 um-
filtered,  when the samples were collected  with the peristaltic
using a low pumping rate (~200 ml/min) and the set-up depicted
in Figure 1.

-------
 Mfl/L
                    Filter Pore Size (\un)

Figure 5. Differences in chromium concentrations for samples
         collected with peristaltic pump and bailer (well 1, Elizabeth
         City).

Although the Saco site is a chrome tannery waste disposal site,
the inorganic contaminant of greatest interest in the ground water
is arsenic.  It is unclear at this time why the arsenic levels are
elevated, and this is under further investigation. Figure 6 for well
113A shows arsenic levels for samples collected using both a
peristaltic pump and a bailer. Results were similar to those for
chromium at Elizabeth City. Once again, there were no differences
observed in metal concentrations using the peristaltic pump and
different filter pore sizes; but large differences were observed
between filtered and unfiltered bailed samples. Also, these two
sets of values were generally different from the peristaltic pump
sampled values. The sampling set-up used with the peristaltic
pump (Figure 1) consistently produced the most reproducible
results, providing increased confidence that these samples were
more representative of natural geochemical conditions and particle
loading than those collected with the bailer.

It was interesting to note that there seems to be no significant
contribution to contaminant transport from suspended and mobile
particles greater than 0.1 (im at either the Elizabeth City or the
Saco site. However, this does not indicatethatcolloidalfacilitated
transport by smaller particles may not occur at these sites.

LABORATORY STUDIES

Particles with diameters of 0.1 to 2.0 urn may constitute the  most
mobile sizef raction in porous med ia. This is because the efficiency
of particle removal increases rapidly above 1 u,m diameter due to
sedimentation and/orinterception processes; whereasfor particles
smaller than 0.1 u.m diameter,  removal occurs primarily by
diffusion(Yaoetal.1971 and O'Melia, 1980). Although significant
contaminant transport by colloidal material in this size range was
not observed for the above three sites, repeated particle size
analyses using laser light  scattering with photon correlation
spectroscopy (PCS) for aqueous samples and scanning electron
microscopy (SEM) analysis for particles collected on filters at the
Globe site indicated a preponderance of particles in the size
range 0.5 to 2.0 u.m at  the lowest flow rates used for sample
collection.   Because of these and similar observations by
Gschwend and Reynolds (1987), laboratory experiments  using
alluvium from  the Globe site were performed to investigate
specific  aqueous  chemical  effects  on the  transport of
environmentally realistic colloids, in the size range of 0.1-0.9 u,m,
through natural porous media under controlled conditions.

Iron oxide particles were synthesized to specific size and shape
for use as the mobile colloidal  phase in laboratory column
experiments.  The aquifer materials from Globe (wells 107 and
452) were  used for the column  packing  or  immobile phase.
Arsenate was selected as a ubiquitous and hazardous inorganic
contaminant, to study its interaction with both the porous immobile
aquifer solids and  the mobile  inorganic colloids.   Batch
experiments were performed to evaluate colloid stability  and
assess the interactions between arsenate and colloidal Fe?O3, and
arsenate and  the aquifer matrix. Column experiments were
performed to determine the  extent of colloid transport and to
compare retardation of aqueous and colloid-associated arsenate.
Study variables included column flow rate, pH, ionic strength,
electrolyte composition (anion/cation), colloid concentration and
colloid size.

Characterization of Aquifer Solids and Colloidal Fe2O3

Core material from two locations at the site was used to pack 2.5-
cm diameter, adjustable, glass columns. Prior to packing, the
core sample was air-dried and sieved with the fraction between
106 and 2000 u.m used  in  the columns.  Subsamples were
analyzed by X-ray diffraction. The predominant mineral phases,
identified in order of intensity, were: quartz>albite» magnesium
orthoferrosilate > muscovite > samsonite > manganese oxide.

The pH2  , or pH at which the net surface charge of a solid equals
zero, iszan  important parameter affecting both colloidal stability
and the interaction of the colloids with immobile matrix surfaces.
Above the pH2  , minerals possess a net negative charge; while
below this pH, tne net charge is positive. Most sand and gravel-
type aquifer solids exhibit a net negative charge under most
environmentally-relevant pH conditions, due to the predominance
of silica (pH2   ~ 2) and other minerals such as layer silicates and
manganese oxides which have pH   's < 4.
 140



 120
 60
 40
  20
                      5           045

                   Filter Pore Size (
 Figure 6. Differences in arsenic concentrations for samples
         collected with peristaltic pump and bailer (well 11 3A,
         Saco, Maine).

-------
Spherical, monodisperse colloidal  Fe O_ (100-900  nm) was
prepared from solutions of FeCL and HCl using the method of
Matijevic and Scheiner (1978). "me method was modified by the
addition of a spike of 26Fe69CI3, prior to heating, to permit
detection of the colloidal hematite with liquid scintillation counting
techniques.  The labeled colloidal Fe.O3 allowed unequivocal
discrimination between injected particles and those mobilized
within the column packing material.

Colloid  concentration,  in milligrams per  liter (mg/L), was
determined gravimetrically by both filtration and  residue-on-
evaporatton techniques, coupled with solute analyses and mass-
balance calculations. SEMand PCS were used to determine the
particle  size. PCS was also used  to evaluate stability of the
diluted colloidal suspensions and particle size in both influent and
effluent  column suspensions.   The surface area of 200 nm
diameter uniformly spherical hematite particles was calculated to
be 5.72  m2/g using the equation,
                            SxlQ-1
                             pd
                                                  (1)
a
jj
8
3
i
S
1
1



9
«
Mobility (urn/sec f

8 -
6'
4
2
•2
-4
-6-
-8'
-10 •
4

• » 0.001 M NaCIO4
o 0.010 MNaCIO4
0
° » • -.
O
o
•
0
o
•
• .
\ 5 6 7 B 9 11
pH
where A is the geometric surface area (m2/g), p the density (g/
cm3), and d the diameter (cm).

ThepH   of the colloidal hematite was evaluated from acid-base
titrations using varying concentrations of NaCI or NaCIO4 as the
background (non-interacting) electrolytes. The plot in Figure 7a
illustrates a distinct crossover  in the titration  curves at
approximately pH 7.4 in NaCI. These titrations were performed
in a nitrogen-filled glove box. Micro-electrophoretic mobility (EM)
was  used to evaluate the pHlep (isoelectric point) of the particles.
This technique was performed in  NaCIO4.  When only non-
interacting electrolytes are present in  solution, pH   - pH. p.
Therefore, this mobility measurement served to  support the
titration  pHz   data (Figure 7b). The EM measurements were
made ontheiSench in the presence of atmospheric CO2. Colloid
stability  in both influent and effluent column suspensions as well
as in batch studies was also evaluated  using PCS to monitor
coagulation.  The colloidal suspensions were stable  in dilute
NaCIO4 and  NaCI over the pH ranges 2.0-6.5 and  7.6-11.0
Figure 7b. Electrophoretic mobility of Fe2O3 colloids as a function
         of pH In the presence of different concentrations of
         NaCI04.
(Figure 8) respectively, with a region of instability corresponding
to the estimated pH   .  Additional  experimental details are
provided in Puls et al.fl 990).

In this study,  significant enhancement of  colloid  stability  in
suspensions of 0.01 M NagHAsO,,, and 0.01 M NaH PO4 was
observed at pH values as low as 4 (« pristine pH) (Figure 9).
This observation corresponds with work by  Liang and Morgan
(1990) who observed that hematite particles bear an overall net
negative charge at  pH  « pHz   is|ine in the  presence  of
specifically sorbed anions  (e.g. 'phosphate  species). This
phenomenon has the effect of increasing the stability  region
where the particles are negatively charged.

Arsenate Adsorptlon/Desorption

Adsorption of arsenate to the aquifer solids was performed  to
determine its affinity for these surfaces and to compare batch and
column-derived solid-solution distribution values.  Adsorption
experiments with the colloidal hematite determined adsorption
capacity and strength of arsenate retention. Adsorption data for
arsenate on the aquifer solids weref itted to a Freundlich isotherm,
defined by the relation
                                                                                    S = KCf"
                                                  (2)
where K is the empirical distribution coefficient or solid surface
affinity term, S the steady-state concentration on the solids (mol/
kg), C, the steady-state solution concentration (mol/L), and n is
an empirical coefficient related to the monolayer capacity and
energy of adsorption. Steady-state, as used here,  represents the
pre-determined batch  equilibration  times (24  hr)  where no
continued  decrease in aqueous arsenate concentrations within
analytical  uncertainty were observed.   Values of n < 1  imply
decreasing energy of sorption with increasing surface coverage.
The calculated K and n values using the logarithmic form of the
above equation were 5.5 and 0.73, respectively.  These values
represent relatively weak interaction with the aquifer solids, with
the interaction primarily due to the presence of iron  oxide coatings
on mineral  grains  as  determined  by sequential  extraction
techniques (Tessier et at. 1979).

-------
      1000


       goo •


       800


       700


       600


       500 •


       400 -


       300 -


       200 .


       100
   A  A
                                               10    11   12
was added to replace the extracted supernatentfrom adsorption
batch reactors.  The  pH was the same pH used in the prior
adsorption experi ments (pH 7). Samples were equilibrated for 48
hrs.  Strong retention of the arsenate on the hematite was
observed,  since only about  2-6% of the adsorbed arsenate
fraction was released (Figure 11).  Percent  desorption was
directly proportional to the initial arsenateconcentration, indicating
a decline in the energy of adsorption as the surface  became
increasingly saturated with arsenate. This phenomenon could
have  important implications for pump and treat remediation of
highly contaminated sites.  The easily  desorbed arsenate will
result in high initial efficiency of dissolved arsenate  removal
which may significantly decline once concentration values are
reduced below the plateau portion of the adsorption isotherm and
desorption becomes less energetically favorable.
Figure 8. Stability of 150 nm Fe^O, colloids as a function of pH In
         0.005 M NaCIO4.
Arsenate adsorption data for the Fe2O3 colloids were fitted to a
Langmuir isotherm (Figure 10) defined by the relation
                           kbC.
                                                  (3)
where k is the Langmuir solid surface affinity coefficient, b the
adsorption  capacity, and  S and  C(  are  defined  above.  An
advantage  of the Langmuir model is the incorporation of the
capacity term.  The correlation coefficient for the  linearized
Langmuirform of the above equation was 0.97 and the adsorption
capacity was estimated to be 0.01 g arsenic/g Fe2O3. There was
very little  difference  in adsorption  extent between pH 4-7.
However, a gradual decrease was observed with increasing pH.

Desorption batch  experiments were performed  immediately
following arsenate adsorption to simulate passage of an arsenate
plume with subsequent contact by low- arsenate or arsenate-free
water. An equivalent  volume of arsenate-free  0.01 M  NaCIO4
                                                                  1.5
                                                                       adsorption capacity - b - 0.01 g/g
                                                            0.01
                                                                    0.02
                                                                             0.03

                                                                          Thousandths
                                                                          Cl (mol/L)
                                                                                     0.04
                                                                                             0.05
                                                                                                      0.06
                                             Figure 10. Langmuir isotherm data for arsenate adsorption on
                                                      Fe2O? colloids (pH 7,0.01 M NaCIO4, 4.5 mg:30ml, 24 hr
                                                      equilibration).
     4-


     2-
     -2-

     -4-

     -6-
    -10
                       •  0.010 M NaCIO4
                       O  0.010 M NaH2PO4
                       A  0.010 M Na2HAsO4
A
 O
          O
          f>
                          6         8
                             PH
                                              10
                                                        12
Figure 9. Electrophoretic mobility of Fe2O3 colloids as a function
         of pH in the presence of 0.01 M sodium electrolytes of
         different anionic composition.
Dissolved Arsenate Transport

Column studies of dissolved arsenate transport were performed
to compare distribution coefficients (Kd) with those derived from
the batch tests and for comparison to Fe2O3 colloid-facilitated
transport of the arsenate. Arsenate concentrations used in both
sets of experiments were comparable. The column Kd roughly
corresponds to the K value calculated from the batch tests and
is determined by:
                                                                                                                (4)
                                             where R, is the retardation factor or ratio, v/vc, of the velocity of the
                                             ground water to the velocity of the solute, pb is the bulk density,
                                             and n is the porosity. Tritiated water was used for estimating the
                                             average water velocity.

-------
        2.71E-06    1.33E-05   2.03E-05   278E-O5   343E-05


                       InMil Conc«n»«lont(mi)l/L)
                                                        100


                                                        90


                                                        80


                                                        70


                                                        «0


                                                        50


                                                        40


                                                        30


                                                        20


                                                        10


                                                        0


                                                        -10
Figure 1 1 . Adsorptlon-desorptlon data for arsenate on Fe,O, colloids
        (pH 7, 0.01 M NaCIO4, 4.5 mg:30ml, 24 hr equilibrations).

As the flow velocity was decreased,  the column  K, values
approached those of the 24-hr equilibrated  batch  K values,
indicating rate limited adsorption onto the aquifer solids at the
higher flow velocities (Table 2). These results demonstrate the
importance of generating comparative data and not relying solely
on batch sorption static equilibrium data, especially for specific
site assessment purposes. When ground-water flow velocities
are relatively rapid,  assumptions of local equilibrium  may be
invalid.

Following the arsenate transport experiments, the columns were
flushed with deionized water (~zero  ionic strength).   This
significantly increased  turbidity in the column effluent due to
dispersion of the colloidal aquifer fines. When the effluent was
analyzed, the recovered arsenate was determined to be almost
entirely colloid-associated and approximately equal tothe influent
arsenate concentration, demonstrating the potential importance
of this transport mechanism (Figure 12).
    1.2
    0.4
    0.2
                                             Drtorizwl Water
               0.1
                                             100
                        Relative Pore Volumes

                        Wflum
Figure 12.  Breakthrough curve for dissolved arsenate versus
        tritium, and mobilization of colloidal-associated arsenate
        by deionized water.
Table 2.  Comparison of distribution coefficients (K4, Ukg) for
         arsenate using Globe, AZ aquifer material In batch and
         column tests (pb = 2.65 g/cm3, n = 0.4).
                 Column

         3.4 m/d        1.7m/d

           1.4            3.0
    Batch

Steady-state

     5.5
 Colloidal Transport

 Column flow rates used were comparable to estimated ground-
 water velocities in the Globe alluvium (0.8-3.4 m/d). The injected
 colloidal hematite generally broke through at the same time or
 prior to tritiated water (Figure 13). In this case, the rate of colloid
 transport through the columns was over 21 times faster than the
 dissolved arsenate. A summary of column resultsforthe colloidal
 transport experiments has been compiled in Table 3. No colloid
 transport occurred when the particles were net positively-charged
 indicating  significant  electrostatic  interaction  with  the net
 negatively-charged matrix  material.  In  low  ionic strength
 suspensions of NaCI and NaCIO., breakthrough exceeded 50%
 of initial colloid concentrations. These electrolyte suspensions
 were  perhaps the most representative of natural conditions in
 uncontaminated ground water. There was substantially lower
 colloid recovery of sulfate-based suspensions, due to difficulties
 in maintaining colloid stability,  and apparent non-specific
 interactions of the sulfate with the hematite  surface.

 Maximum percent breakthrough occurred with phosphate and
 arsenate-based suspensions and appeared to be unaffected by
 the range of flow velocities or column lengths used. Likewise, no
 significantdifferences were observeddue to colloid concentration.
 The specific adsorption of the predominantly divalent phosphate
 and arsenate anions onto the hematite surface caused a charge
 reversal  on the initially net positively-charged surface.   The
 consequence is a lowering of the pH   below the pHz  value and
 an increase in net negative charge near the particle surface over
 a wider pH  range.  The increased negative charge increased
 repulsion between the mobile, negatively-charged particles, and
 the immobile, net negatively-charged, column matrix solids. As
 a result, the colloids were stable at a greater distance from the
 pore walls in the column matrix, where fluid  velocity is higher.

 Particle size had an inverse effect on percent breakthrough; that
 is, increasing colloidal transport was observed with  decreasing
 particle  size.  While the larger particles were still transported
there were significant differences between the 100 nm and 900
 nm size classes.   A  complicating factor  in resolving these
differences was the use of a different aquifer solid sample (well
452) for the larger hematite particles.  Sieve  analyses of the two
samples reflected differences in particle size (Figure 14), although
XRD analyses showed no significant differences in mineralogy.
The  high density  (5.3 g/cm3) of the hematite  may have
contributed to gravitational settling of the larger particles. Densities
of secondary clay minerals which  are more representative of
colloids in natural systems are on the order  of 2.6 g/cm3.

-------
Table 3. Column results of colloidal Fe2O3 transport through natural aquifer material
Size
(nm)
200
125
150
250
150
100
100
100
125
100
100
900
900
pH
3.9
8.9
8.1
8.1
8.9
7.6
7.6
7.6
7.6
7.6
7.6
7.0
7.0
Velocity
(m/d)
3.4
3.4
3.4
3.4
3.4
3.4
1.7
0.8
3.4
1.7
3.4
34
3.4
Part.Conc.
(mg/L)
10
10
5
10
5
5
5
5
10
5
5
50
50
Ionic
Strength
0.005
0.005
0.001
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
Anion
Cl
ci-
CI04
SO4*
S042
HAsO42
HAsO4z
HAsO/
HPO/'
HPO4*
HP04Z
HPO4Z
HPO4J
%C0
thru
0
54
57
17
14
97
96
93
99
99
99
33
30
Col. Length
(cm)
3.8
5.1
3.8
3.8
3.8
2.5
2.5
2.5
5.1
2.5
2.5
3.8
3.8
 All of the parameters (Table  3 headings) or  variables were
 explored in detail with the SAS program JMP, using %C
 breakthrough as the response variable.  Only colloid size and
 anion significantly impacted the %CQ breakthrough. Combining
 these two parameters into a two-way main effects analysis of
 variance (2-way ANOVA) model accounted for 98.4% of the
 variability in the colloid breakthrough results.  All other factors
 tested gave no significant correlation over the parameter ranges
 utilized  in this study.

 Given a ground water containing 10 mg/L suspended colloidal
 material, of surface area and reactivity comparable to these
 hematite particles,  0.1  mg/L  of arsenic could be  colloidally
 transported under some hydrogeochemical conditions. This is
 twice the maximum contaminant level (MCL) currently set for
 drinking water.  It should be noted that these Fe2O3 model
 colloids are relatively non-reactive and have low surface area
 compared to the more ubiquitous subsurface colloidal minerals
 (e.g. clays,  goethite).   In addition, the  suspended  solids
 concentrations of these  experiments were about one-half of
 those observed at the Arizona site and one-sixth the concentration
 observed at some other sites (Buddemeier and Hunt, 1988; Ryan
 andGschwend, 1990).
c   1.0 1

1
o
O   0.6 -
    02
                 1         2         3

                        Relative Pore Volumes
                 Phosphate Colloids
                 Arsenate Colloids
                                        • Tritium
Figure 13. Breakthrough curve for Fe2O, colloids suspended in
         0.01 M NaH,PO4 and in 0.01 M Na,HAsO4, pH 7.6,3.4 m/d,
         5 mg/L
                                                           10

-------
I
a
£
39
                   500   420    250   210   177

                         Sieve Size (urn)
 of chemical and physical variables, including but not limited to
 ionic strength, ionic composition, flow velocity, quantity, nature,
 and size of suspended colloids, geologic composition and
 structure, and ground-water chemistry. The most significant of
 these factors, under the conditions investigated in these column
 experiments, were ionic composition and particle size. Neglecting
 colloidal mobility in our predictive contaminant transport models
 may underestimate both the transport rate, maximum transport
 distance, and mass.   Chemical parameters affecting colloidal
 stability and transport must be included in transport modeling
 along with physical parameters (such as pore size distribution,
 colloidal density and size, and flow  velocity). Field sampling
 procedures must account for the possibility of colloidal transport
 and provide correct model input data. These concerns must be
 addressed during site characterization and assessment monitoring
 if colloid transport is deemed possible for the site.
 Figure 14. Sieve size fractionation of well 107 and well 452, Globe
         aquifer material.
 SUMMARY AND CONCLUSIONS

 Field results from three distinctly different sites indicate that the
 most representative and reproducible elemental concentrations
 are  obtained  by following the recommendations  proposed
 previously by  Puls  et al. (1990).  The selection  of sampling
 devices, purging and sampling flow rates and filtration procedures
 are particularly important. There is a strong inverse correlation
 between turbidity and representativeness of samples.   The
 greatest differences, both in terms of suspended  colloids and
 inorganic contaminant concentrations, were observed between
 the bladder pump and the high speed submersible  pump, in the
 deep wells, and between the peristaltic pump and the bailer in the
 shallow wells.

 Steady-state turbidity levels observed at the three  sites ranged
 from 1 -58 NTUs; and in the case of one site, turbidity differences
 were strongly related to clay mineral content. Screened intervals
 with higher clay and silt contents had  higher turbidity  values.
 While artifacts of well construction and sample collection cause
 ground-water turbidity, there are indications that high levels of
 turbidity may occur naturally due to geology and geochemistry.

 The down-hole camera indicated little artificial colloid generation
 or disturbance due to the low flow rate pumping action of the
 peristaltic pump. Emplacement of the camera, which is similar in
 size and shape  to  bladder pumps and  submersible  pumps,
 created the  greatest disturbance (turbidity). While there  has
 been no concrete  evidence of  significant  colloidal-facilitated
 transport of inorganic contaminants at any of the three sites, the
 down-hole camera  documented the existence  of significant
 concentrations of suspended colloidal material in the flow field at
 Elizabeth City. Additional research in this area continues.

 Laboratory experiments using natural aquifer material and realistic
 inorganic model colloids indicate that the transport of colloidal
 material through sand and gravel-type aquifers may be significant
 under certain hydrogeochemical conditions. Due to the strong
 reactivity of many inorganic colloids in natural subsurface systems,
the potential exists that this form of contaminant transport may be
 important at certain sites. Its significance depends on a number
 REFERENCES

 Brown, J.G. Chemical, Geologic, and Hydrologic Data from the
 Study of Acidic Contamination in the Miami Wash-Final Creek
 Area, Arizona, USGS Open File Report, 90-395, October, 1 990.

 Buddemeier, R.W. and J.H.  Rego, Colloidal  Radionuclides in
 Groundwater. FY85 Annual Report. Lawrence Livermore National
 Laboratory, Livermore, CA, UCAR 10062/85-1, 1986.

 Buddemeier,  R.W. and J.R. Hunt, Transport of  Colloidal
 Contaminants in Ground Water: Radionuclide Migration at the
 Nevada Test Site, Applied Geochemistry. 3:535-548,  1988.

 Cerda, C.M.  Mobilization of Kaolinite Fines in Porous Media,
 Colloids and Surfaces. 27: 21 9-241 , 1 987.

 Champ, D.R., W.F. Merritt and J.L. Young, Potential for Rapid
 Transport of Pu in Groundwater  as Demonstrated  by Core
 Column  Studies, In  Scientific Basis for  Radioactive Waste
 Management. Vol. 5, Elsevier Sci. Publ., NY, 1982.

 Champlin.J.B.F.andG.G.Eichholz, The Movement of Radioactive
 Sodium and Ruthenium through a Simulated Aquifer,  Water
 Resour. Res.. 4(1 ):1 47-1 58, 1968.

 Champlin.J.B.F. andG.G. Eichholz, Fixation and Remobilization
 of Trace Contaminants in Simulated Subsurface Aquifers. Health
       , 30:215-219,  1976.
Eichholz,  G.G.,  B.G. Wahlig, G.F.  Powell  and T.F. Craft,
Subsurface Migration of  Radioactive Waste Materials by
Paniculate Transport, Nuclear Tech.. 58: 511-519, 1982.

Enfield, C.G. and G. Bengtsson, Macromolecular Transport of
Hydrophobic Contaminants in Aqueous Environments, Ground
       26(1 ): 64-70, 1988.
Gschwend  P.M. and M.D. Reynolds, Monodisperse Ferrous
Phosphate  Colloids in an Anoxic Groundwater Plume,  J. of
Contaminant Hvdrol.. 1:309-327, 1987.

Harvey, R.W., L.H. George, R.L. Smith  and D.R. LeBlanc,
Transport of Microspheres and  Indigenous Bacteria through a
Sandy Aquifer: Results of Natural- and Forced-Gradient Tracer
Experiments, Environ. Sci. Technol.. 23(1):51-56, 1989.
                                                         11

-------
Hem, J.D. and C.E. Roberson, Form and Stability of Aluminum
Hydroxide Complexes in  Dilute Solution.  U.S.  Geol. Water
Supply Pap. 1827-A, A24,  1967.

Kennedy, V.C. and G.W. Zellweger, Filter Pore-Size Effects on
the Analysis of Al, Fe, Mn, and Ti in Water, Water Resour. Res..
10(4): 785-790,1974.

Kim, J.I., G. Buckau, F. Baumgartner, H.C. Moon and D. Lux,
Colloid Generation and the Actinide  Migration in  Gorleben
Groundwaters,  In Scientific Basis  for Nuclear Waste
Management. V.7, Gary L. McVay, Ed., Elsevier, NY pp. 31-40,
1984.

Liang,  L. and J.J.  Morgan, Chemical  Aspects of Iron Oxide
Coagulation in  WaterLaboratory Studies and  Implications for
Natural Systems, Aquatic Sciences. 52(1):32-55.1990.

Matijevic, E. and P. Scheiner, Ferric Hydrous Oxide Sols, sL
Colloid Interface Sci.. 63(3):509-524. 1978.

Matijivec, E., R.J. Kuo and H.  Kolny, Stability  and Deposition
Phenomena of Monodispersed  Hematite Sols, J. Colloid
Interface Sci.. 80(1):94-106. 1980.

Means, J.C. and R. Wijayaratne, Role of Natural Colloids in the
Transport of Hydrophobic Pollutants, Science. 215(19): 968-970,
1982.

Nelson, D.M., W.R. Penrose, J.O. Karttunen and P. Mehlhaff,
Effects of Dissolved Organic Carbon on the Adsorption Properties
of Plutonium in Natural Waters, Environ. Sci. Technol.. 19:127-
131,1985.

Nightingale,  H.I. and W.C. Bianchi, Ground Water Turbidity
Resultingfrom Artificial Recharge. Ground Water. 15(2^: 146-152,
1977.

O'Melia, C.R., Aquasols:  The Behavior of Small Particles in
AquaticSystems. Environ. Sci. Technol.. 14(9^:1052-1060.1980.

Penrose, W.R., W.L Polzer, E.H. Essington, D.M. Nelson and
K.A. Orlandini, Mobility of Plutonium and Americium through a
Shallow Aquifer in  a Semiarid  Region, Environ. Sci. Technol..
24:228-234, 1990.

Puls,  R.W. and J.H. Eychaner, Sampling Ground  Water for
Inorganics - Pumping Rate, Filtration, and Oxidation  Effects, In
Fourth National Outdoor Action Conference on Aquifer
Restoration. Ground Water Monitoring and Geophysical Methods.
National Water Well Association, Dublin, OH, 1990.

Puls, R.W., J.H. Eychaner and R.M. Powell,  Colloidal-Facilitated
Transport of Inorganic Contaminants in Ground  Water: Part I.
Sampling Considerations.  Environmental Research Brief,
USEPA, EPA/600/M-90/023, December 1990.

Reynolds, M.D., Colloids in Groundwater, Masters Thesis, Mass.
Inst. of Tech., Cambridge, MA, 1985.

Ryan, J.N. and P.M.  Gschwend, Colloid Mobilization in  Two
Atlantic Coastal Plain Aquifers: Field Studies. Water Resour. Res..
 1990.
Saltelli, A., A. Avogadro and G. Bidoglio, Americium Filtration in
GlauconiticSand Columns, NuclearTechnol.. 67,245-254,1984.

Sandhu, S.S. and G.L. Mills, Kinetics and Mechanisms of the
Release of Trace Inorganic Contaminants to Ground Water from
Coal Ash Basins on the Savannah River Plant. Savannah River
Ecology Lab. Aiken, SC, DOE/SR/15170-1,1987.

Sheppard, J.C., M.J. Campbell and J.A. Kittrick, Retention  of
Neptunium, Americium and Curium by Diffusible Soil Particles,
Environ. Sci. Technol.. 13(6), 680-684 1979.

Standard Methods forthe Examination of Water and Wastewater.
17th Edition, Clesceri, L.S., A.E. Greenberg and R.R. Trussell,
Eds., American Public Health Assoc., Washington D.C., 1989.

Stumm, W. and J.J. Morgan, Aquatic Chemistry. John Wiley and
Sons, Inc. N.Y., 1981.

Takayanagi, K. and G.T.F. Wong, Organic and Colloidal Selenium
in South Chesapeake Bay and Adjacent Waters, Marine Chem..
14:141-148,1984.

Tessier, A., P.G.C. Campbell, and M. Bisson, Sequential Extraction
Procedure for the Speciation of Paniculate Trace Metals, Anal.
£hejn., 51 (7): 844-849,1979.

Tillekeratne, S., T. Miwa and A. Mizuike, Determination of Traces
of Heavy Metals in Positively Charged Inorganic Colloids in Fresh
Water, Mikrochimica Acta. 6:289-296, 1986.

Wagemann, R. and G.J. Brunskill, The Effect of Filter Pore-Size
on Analytical Concentrations of Some Trace Elements in Filtrates
of Natural Water, Intern. J. Environ. Anal. Chem.. 4:75-84,1975.

Yao, K., M.T. Habibian and C.R. O'Melia, Water and Waste Water
Filtration:  Concepts and Applications, Environ. Sci. Technol..
5(11):1105-1112, 1971.

DISCLAIMER
The information in this document has been funded wholly or in
part by the United States Environmental Protection Agency. This
document  has been subjected to  the Agency's peer and
administrative review and has been approved for publication  as
an EPA document.

QUALITY ASSURANCE STATEMENT
All research projects making conclusions or recommendations
based on environmentally related measurements and funded by
the Environmental Protection Agency are required to participate
in the Agency  Quality Assurance Program. This  project was
conducted under an approved Quality Assurance Program Plan.
The procedures specified in this plan were used without exception.
Information on the plan and documental ion of Inequality assurance
activities and results are available from the Principal Investigator.

ACKNOWLEDGEMENTS
The authors wish to recognize and thank Dr. Terry F. Rees of
USGS, San Diego, CA, for SEM-EDX analyses; Bert Bledsoe,
RSKERL, for  field assistance and much helpful  technical
assistance and guidance;TerryConnally and Dick Willey, USEPA,
and Frank Blaha and Jim Vardy, USCG.
                                                         12
             •&U.S. GOVERNMENT PRINTING OFFICE: 1991 - 548-028/40045

-------