United States Office of Research and EPA/600/R-01/007b
Environmental Protection Development March 2001
Agency Washington DC 20460
xc/EPA Multispecies Reactive
O=»Si» Tracer Test in a Sand and
Gravel Aquifer,
Cape Cod, Massachusetts
Part 2
Transport of Chromium (VI) and
Lead-, Copper-, and Zinc-EDTA
Tracers
-------
EPA/600/R-01/007b
March 2001
Multispecies Reactive Tracer Test in a
Sand and Gravel Aquifer,
Cape Cod, Massachusetts
Part 2
Transport of Chromium(VI) and Lead-, Copper,
and Zinc-EDTA Tracers
J. A. Davis1, J. A. Coston1, D. B. Kent1, K. M. Hess2, J. L. Joye1,
P. Brienen1 and K. W. Campo2
1 U.S. Geological Survey
Menlo Park, CA 94025
2 U.S. Geological Survey
Northborough, MA 01532
Interagency Agreement DW14935626
Project Officer
Robert W. Puls
Subsurface Protection and Remediation Division
National Risk Management Research Laboratory
Ada, OK 74820
National Risk Management Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, OH 45268
-------
Notice
The U. S. Environmental Protection Agency through its Office of Research and
Development partially funded and collaborated in the research described here
under Interagency Agreement DW14935626. It has been subjected to the Agency's
peer and administrative review and has been approved for publication as an EPA
document. Mention of trade names or commercial products does not constitute
endorsement or recommendation for use.
All research projects making conclusions or recommendations based on environ-
mentally 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 Project Plan. The
procedures specified in this plan were used without exception. Information on the
plan and documentation of the quality assurance activities and results are available
from the Principal Investigator.
-------
Foreword
The U.S. Environmental Protection Agency is charged by Congress with protecting the Nation's land, air, and water
resources. Under a mandate of national environmental laws, the Agency strives to formulate and implement actions
leading to a compatible balance between human activities and the ability of natural systems to support and nurture
life. To meet this mandate, EPA's research program is providing data and technical support for solving
environmental problems today and building a science knowledge base necessary to manage our ecological
resources wisely, understand how pollutants affect our health, and prevent or reduce environmental risks in the
future.
The National Risk Management Research Laboratory (NRMRL) is the Agency's center for investigation of
technological and management approaches for preventing and reducing risks from pollution that threatens human
health and the environment. The focus of the Laboratory's research program is on methods and their cost-
effectiveness for prevention and control of pollution to air, land, water, and subsurface resources; protection of water
quality in public water systems; remediation of contaminated sites, sediments and ground water; prevention and
control of indoor air pollution; and restoration of ecosystems. NRMRL collaborates with both public and private sector
partners to foster technologies that reduce the cost of compliance and to anticipate emerging problems. NRMRL's
research provides solutions to environmental problems by: developing and promoting technologies that protect and
improve the environment; advancing scientific and engineering information to support regulatory and policy
decisions; and providing the technical support and information transfer to ensure implementation of environmental
regulations and strategies at the national, state, and community levels.
The use of multispecies reactive transport modeling in site assessments and remedial performance monitoring at
hazardous waste sites is to be encouraged. These can be valuable tools for simulating processes which govern
contaminant fate and transport in the subsurface. A competent reactive transport model of a site must be able to
simulate the processes that occur between water and the vapor and solid phases in contact with the water. The
accuracy of such simulations will be dependent on the database contained within such models as well as the quality
of the site-specific data collection efforts.
The lessons learned from the large-scale tracer tests and subsequent modeling investigations from this study have
implications that go far beyond the particular contaminant species studied here. For homogeneous and classical
heterogeneous reactions (e.g., dissolution equilibria), there are generally available equilibrium constants that may be
used with confidence in such models; however, solubilities for poorly crystalline or impure solid phases are generally
not available and will have to be determined using site-specific materials. Adsorption and other heterogeneous
surface reactions will generally require experimental determination of parameters that are site-specific. In addition,
chemical reactions can be rate-limited and controlled by physical factors such as diffusion or mixing of waters across
sediment layers. These studies were important in identifying the limitations of reactive transport modeling, their
utility, and where additional data collection and testing is warranted.
Stephen G. Schmelling, Acting Director
Subsurface Protection and Remediation Division
National Risk Management Research Laboratory
-------
IV
-------
Executive Summary
A comprehensive field investigation of multispecies reactive transport under variable chemical conditions was
conducted in a shallow, unconfined, sand and gravel aquifer, on Cape Cod, Massachusetts. Ten thousand liters of
groundwater with added tracers were injected into the aquifer, and the distributions of tracers were monitored for
more than one year as the tracers were transported over 200 meters (m) through an array of multilevel samplers
(MLS). Tracers included a nonreactive tracer (bromide ion) and the reactive solutes Cr(VI) and EDTA
(ethylenediaminetetraacetic acid) complexes of the divalent metal ions, Ni, Cu, Zn, and Pb. The EDTA concentration
in the injected water was in slight excess of the sum of the four metals, ensuring that the speciation of the injected
metals was dominated by their EDTA complexes. Transport of the tracers was quantified using spatial moments
computed from a series of comprehensive synoptic samplings of the tracer cloud and temporal moments of
concentration data collected frequently at two MLS (breakthrough curves).
The vertical interval over which the tracers were injected spanned three zones with contrasting chemical conditions.
The pristine zone, with high concentrations of dissolved oxygen (DO), pH values in the range 5.4-5.7, and low
concentrations of dissolved salts, overlaid the sewage-contaminated zone, with low concentrations of DO (< 3 u,M)
and dissolved manganese, pH values in the range 6.0-6.5, and increased concentrations of dissolved salts and
sewage-derived contaminants, such as phosphate and Zn. Between the pristine and sewage-contaminated zones
was the Zn-contaminated region, with elevated concentrations of dissolved and adsorbed Zn derived from the
sewage effluent. The Zn-contaminated region was approximately 3 m thick near the injection location but decreased
in thickness with distance downgradient. Concentrations of dissolved and adsorbed Zn decreased with distance
downgradient, until at 150 m away from the injection, no Zn contamination in the aquifer could be detected.
Chemical interactions between the reactive tracers and the aquifer sediments varied among the different solutes and
the different geochemical zones. More spreading and greater retardation of Cr(VI) and the metal-EDTA complexes
was observed in the pristine zone as a result of increased adsorption of these anionic solutes in that zone. The most
important factors favoring adsorption in the pristine zone relative to the other zones were the lower pH value and
lower concentrations of other anions that could compete for adsorption sites on the surfaces of the sediments.
Transport of the metal-EDTA complexes was affected by the extent and rates of various metal exchange reactions,
in which the metal ion of one of the injected metal-EDTA complexes was displaced by another cation, such as Fe, Al,
or Zn in the Zn-contaminated region of the aquifer. The free metal ions produced as a product of such metal
exchange reactions, such as Pb2+, were extensively adsorbed under the chemical conditions present in the aquifer.
Fe and Al was made available as a reactant for the metal exchange reactions by dissolution of poorly crystalline
oxyhydroxide phases present as coatings on the Cape Cod sediments.
The metal exchange reactions of Pb-EDTA complexes appeared to be the most thermodynamically favored in the
aquifer, although the extent of Zn-EDTA metal exchange was also significant outside of the region of Zn
contamination. Because of the strong adsorption of Pb2+ released in the metal exchange reactions of Pb-EDTA
complexes, the mass of dissolved Pb within the tracer cloud decreased rapidly and was not detectable 110 days
afterthe injection. Pb-EDTA complexes primarily exchanged EDTA with Zn and Fe(lll) in the first 13 days after the
injection, with the Fe derived from dissolution of sediment coatings and Zn derived from desorption within the Zn-
contaminated region. AI-EDTA complexes that had formed with excess EDTA shortly after the injection also
exchanged with Fe. Dissolved AI-EDTA complexes were not detectable 83 days after the injection. The dissolved
mass of Fe-EDTA complexes increased steadily throughout the tracer test; the rate of Fe dissolution appeared to be
rate-limited at the beginning of the tracer test.
The metal exchange reactions of Cu-EDTA complexes continued slowly throughout the tracer test, and approxi-
mately 30% of the injected Cu-EDTA was still present in the tracer cloud 300 days afterthe injection. The net mass
of dissolved Zn-EDTA in the tracer cloud increased during the first 175 days of the tracer test, due to desorption of
contaminant Zn from the sediments and the formation of dissolved Zn-EDTA complexes. However, outside of the
Zn-contaminated region, dissolved Zn-EDTA was lost from the tracer cloud as a result of metal exchange reactions
with Fe(lll). During the period 175 to 300 days after the injection, the mass of dissolved Zn-EDTA decreased rapidly
as the balance shifted to a net loss of Zn-EDTA in the metal exchange reactions. The loss of Zn-EDTA increased as
the tracer cloud sank below the Zn-contaminated region and was transported longitudinally beyond the region with
greatest Zn-contamination.
-------
The order of displacement of the metal ions (Pb»Zn>Cu»Ni) observed within the pristine and suboxic zones is
consistent with thermodynamic predictions that consider both the strength of the aqueous EDTA complexes and the
strength of adsorption of the free metal ions. At chemical equilibrium and in groundwater saturated with ferrihydrite,
the thermodynamic calculations show that the extent of the metal exchange reactions depends on both pH and the
concentration of the metal-EDTA complex. This couples the extent of the metal exchange reactions with dispersion
and other processes that lower the concentration of the metal-EDTA complexes in the groundwater.
Loss of dissolved Cr(VI) from the tracer cloud was observed as a result of the reduction of Cr(VI) to Cr(lll), and
irreversible sorption of Cr(lll). This occurred primarily in the sewage-contaminated zone, in agreement with the
results of small-scale tracer tests previously conducted at the site. Adsorption of the anionic Cr(VI) species in the
pristine zone resulted in significant retardation and spreading. Retardation of Cr(VI) was dependent on chemical
conditions that correlated well with the chemical gradients as a function of altitude. Quantitative description of the
loss of Cr(VI) from the tracer cloud will likely require accounting for the sinking of the tracer cloud across the oxic-
suboxic boundary that occurred during the course of the experiment.
The fate and transport of Zn in the sewage effluent were well described by a multispecies reactive transport model
that included surface complexation reactions to describe the pH dependence of Zn adsorption. Reasonable
quantitative predictions of the distribution of Zn contamination at the field site were made from simulations in which
the surface complexation model parameters were calibrated from the results of laboratory batch experiments. These
simulations, as well as other preliminary modeling investigations presented here, will serve as a starting point for
reactive transport modeling of the experimental results of the tracertest. The results of the field experiment provide
a well-characterized, chemically complex data set that can stimulate the development and testing of hydrogeochemical
transport models of flow coupled with chemical reactions.
VI
-------
Contents
Foreword iii
Executive Summary v
Tables viii
Figures ix
Abbreviations xi
SI Conversion Factors xii
Acknowledgments xiii
Introduction 1
Methods and Materials 1
Site Description 1
Zn Partitioning in the Zn-contaminated Region of the Aquifer 2
Tracer Test Experimental Design 3
Laboratory Batch Studies of Fe and Al Dissolution by EDTA 3
Chemical Analyses 3
Spatial and Temporal Moments 4
Results and Discussion 4
Spatial and Temporal Distribution of Reactive Tracers During Transport 4
Chromium Transport 5
Effect of Metal Exchange Reactions on Lead and Copper Transport 6
Zinc Transport 7
Laboratory Studies of the Dissolution of Iron and Aluminum from
Aquifer Sediments by EDTA 8
Implications for Reactive Transport Modeling 9
Modeling the Transport of Cr(VI) 10
Modeling Metal Exchange Reactions of Metal-EDTA Complexes 11
Modeling the Transport of Zn2+ 13
Concluding Remarks 15
References 16
VII
-------
Tables
Table 1. Data for the Breakthrough MLS, BT1 (37 m downgradient) 21
Table 2. Data for the Breakthrough MLS, BT2 (52 m downgradient) 22
Tables. Dataforthe Breakthrough MLS, BT1 (37 m downgradient) 23
Table 4. Dataforthe Breakthrough MLS, BT2 (52 m downgradient) 24
Table 5. Dissolution of Al, Fe, and Zn from Aquifer Composite
Sediments by Hydroxylamine-HCI and EDTA 25
Table 6. Stability Constants of Metal Exchange Reactions with Aqueous Metal Ions 25
Table 7. Stability Constants for Metal Ion Adsorption on Ferrihydrite 26
Table 8 Stability Constants of Metal Exchange Reactions with Adsorbed Metal Ions 26
Table 9. Issues Facing Estimation of Chemical Parameters in Reactive Transport Models 26
VIM
-------
Figures
Figure 1. Location of tracer-test site in western Cape Cod, Massachusetts, the general
path of tracers, the multilevel samplers (MLS) available for sampling during the
tracer test, the six injection MLS, the two breakthrough curve MLS, and the MLS
used to construct background chemistry transects and to define the
extent of zinc contamination 28
Figure 2. Vertical profiles of B (boron), Zn (zinc), pH, DO (dissolved oxygen), and
Mn (manganese) 1.7 m downgradient from the center of the injection taken just
prior to injection at MLS 2414A (shown in inset of Fig. 1 as filled triangle
closest to the array of injection MLS) 29
Figure 3. Schematic showing the relative locations of geochemical zones referred
to in this report 29
Figure 4. Longitudinal cross section showing the concentration of dissolved Zn (u,M)
and pH in the aquifer prior to the field experiment 30
Figure 5. Vertical profiles showing longitudinal and lateral heterogeneity in the
distribution of background dissolved Zn in the aquifer 30
Figure 6. Dissolved Zn as a function of pH. Curves show Zn concentrations versus
pH for three values of total Zn (adsorbed plus dissolved), computed from
the surface complexation model fitted to batch adsorption data
(Davis etal., 1998) 31
Figure 7. Longitudinal cross sections showing the normalized concentrations of the
tracers Br, Ni, Cr, Zn, Cu, and Pb and the dissolved concentrations of Fe and
Al, 13 days afterthe injection 32
Figure 8. Longitudinal cross sections showing the normalized concentrations of the
tracers Br, Ni, Cr, Zn, Cu, and Pb and the dissolved concentration of
Fe 83 days afterthe injection 33
Figure 9. Breakthrough curves for BT1, 37 m downgradient of the injection site 34
Figure 10. Breakthrough curves for BT2, 52 m downgradient of the injection site 36
Figure 11. Results of the spatial moments analysis showing the normalized mass
of Cr (relative to the injected Cr mass and the normalized Br mass) as a
function of time afterthe injection 38
Figure 12. Retardation factors for Cr calculated from the breakthrough-curve data,
plotted as a function of the altitude of the sampling port 38
Figure 13. Attenuation factors for Cr calculated from the breakthrough-curve data, plotted
as a function of the altitude of the sampling port 39
Figure 14. Results of the spatial moments analysis showing the moles of Ni, Pb, Cu,
Zn, Fe, and Al in the tracer cloud as a function of time afterthe injection 39
Figure 15. Concentrations of dissolved Al- and Fe-EDTA complexes and uncomplexed
(free) Zn as a function of the pH of groundwater samples collected 24 hours
afterthe injection was completed 40
Figure 16. Change in metal-EDTA complexes relative to the amount of Pb-EDTA present
for days 1 through 7 after injection 41
Figure 17. Results of the spatial moments analysis showing the normalized masses
of Ni, Cu and Pb (relative to their injected masses) as a function of time
afterthe injection 42
IX
-------
Figure 18. Results of the spatial moments analysis showing the normalized masses
of Ni and Zn (relative to their injected masses) as a function of time after
the injection 42
Figure 19. Longitudinal cross section through the portion of the tracer cloud
containing metal-EDTA complexes at 175 (upper panel) and 237 days
(lower panel) after injection; numbered contours show background Zn
concentrations (u,M, before injection) as was shown in Figure 4 43
Figure 20. Concentrations of dissolved Fe, Al, and EDTA as a function of time in batch
experiments with two samples of composite aquifer sediments mixed with
artificial groundwater solutions containing 100 u,M EDTA 44
Figure 21. Calculated extents of metal exchange between metal-EDTA complexes
and Fe(lll) dissolved from ferrihydrite as a function of pH 44
Figure 22. Calculated extents of Ni exchange between Ni-EDTA complexes and
Fe(lll) dissolved from ferrihydrite as a function of pH and the initial
Ni-EDTA concentration 45
Figure 23. Calculated extents of metal exchange between metal-EDTA complexes
and Fe(lll) as a function of pH and dissolved Fe(lll) concentration 45
Figure 24. Laboratory batch experimental data for Zn adsorption onto a composite
sample of aquifer sediments collected from the site and surface
complexation model fits 46
Figure 25. Calculated extents of metal exchange between Zn-EDTA complexes
and Fe(lll) dissolved from ferrihydrite as a function of pH and Zn-EDTA
concentration 46
Figure 26. Simulated distribution of dissolved Zn after 60 years 47
Figure 27. Calculated values of the distribution coefficient, Kd, for adsorption of Zn
as a function of pH and total Zn concentration per liter of water in the aquifer
using (A) the one-site model for Zn adsorption, and (B) the two-site model
(Davis etal., 1998) 47
-------
Abbreviations
d Day
DO dissolved oxygen
EDTA ethylenediaminetetraacetic acid
g Grams
ICP-AES inductively-coupled plasma atomic emission spectroscopy
kg Kilograms
km Kilometers
L Liters
m Meter
u,g Micrograms
u,M Micromoles per liter
mg Milligrams
mM Millimoles per liter
mm Millimeters
M Moles
MLS multilevel sampler
nm Nanometers
QC quality control
XI
-------
SI Conversion Factors
Multiply
Area:
Flow rate:
Length:
Mass:
Volume:
Temperature:
Concentration:
Pressure:
Heating value:
English (US)
Units by
1 ft2
1 in2
1 gal/min
1 gal/min
1 MGD
1 ft
1 in
1 Ib
1 Ib
1 ft3
1 ft3
1gal
1gal
°F-32
1 gr/ft3
1 gr/gal
1 Ib/ft3
1 Ib/in2
1 Ib/in2
Btu/lb
Btu/scf
Factor
0.0929
6.452
6.31 x10-5
0.0631
43.81
0.3048
2.54
453.59
0.45359
28.316
0.028317
3.785
0.003785
0.55556
2.2884
0.0171
16.03
0.07031
6894.8
2326
37260
Metric (SI)
to get Units
m2
cm2
m3/s
L/s
L/s
m
cm
g
kg
L
m3
L
m3
°C
g/m3
g/L
g/L
kg/cm2
Newton/m2
Joules/kg
Joules/scm
XII
-------
Acknowledgments
We are grateful to Drs. Barbara Bekins, John Friedly, and Juerg Zobrist for critical
technical reviews of the report. Numerous people contributed to the field and
laboratory work reported here. Technical and field assistance were provided by
Steve Coppola, Bill Carothers, Greg Granato, Matthias Kohler, Mary Kruger, John
Masterson, Tim McCobb, Bill Mitch, Brigid Rea, and Jennifer Savoie, sometimes
under unfavorable conditions. Charlie Ogle (Ogle Tooling) and Mike Fitzgerald
(Design Craft Woodworks) provided critical assistance in the design and construc-
tion of the apparatus used for processing "Chelex" samples. Sean Wallace
conducted laboratory experiments on chemical interactions between EDTA and
aquifer sediments described in the text. Chris Fuller and Linda Anderson provided
technical advice on experimental design, sampling, and analysis at various stages
during the project. The numerous contributions of Denis LeBlanc, USGS Cape Cod
Groundwater Research Site Coordinator, and Dr. Robert Puls, Project Officer, U.S.
Environmental Protection Agency, are gratefully acknowledged. Funding for the
project was provided by the U.S. Environmental Protection Agency, through
interagency agreement number DW14935626 and the U.S. Geological Survey
Toxic Substances Hydrology Program.
XIII
-------
XIV
-------
Introduction
Chemical speciation can have a major influence on the transport of inorganic solutes (Davis et al., 1993b; Yeh and
Tripathi, 1991; Liu and Narasimhan, 1989). As an example, free (uncomplexed) metal ions such as zinc (Zn) are
extensively retarded in groundwater (von Gunten et al., 1991; Kent et al., 2000). In contrast, the formation of weakly
adsorbing, anionic complexes greatly increases the mobility of metal ions, as in the cases of Zn (Kent et al., 1992) or
cobalt (Co, Zachara et al., 1995b) complexed with ethylenediaminetetraacetic acid (EDTA). Other examples of mobile
and weakly adsorbing organic complexes are found in groundwater studies of copper (Cu) and a variety of radionuclides
(McCarthy etal., 1998; McBride et al., 1997; Jacobs et al., 1988; von Gunten etal., 1991; Killeyetal., 1984). Changes
in pH along groundwater flow paths can also cause shifts in metal ion speciation that affect mobility.
In addition to complexation and acid-base reactions, redox (electron transfer) reactions may cause changes in speciation
that have a marked effect on transport. Aqueous chromium (VI) species (Cr, in the plus six oxidation state) are generally
soluble and relatively weakly adsorbing, hence they can be mobile in groundwater (Kent et al., 1995). Cr(lll), if
uncomplexed, should be highly retarded in groundwater because its aqueous species are extensively adsorbed or
insoluble (Kent et al., 1994; Anderson et al., 1994).
This report discusses the transport of a group of reactive tracers over the course of a large-scale, natural gradient tracer
test conducted at the USGS Cape Cod Toxic Substances Hydrology Research site, near Falmouth, Massachusetts
(Davis et al., 2000a,b). The overall objectives of the experiment were to demonstrate the importance of variable aquifer
chemistry on chemical reactions, including aqueous speciation changes, during transport of toxic elements in groundwa-
ter (chromate, and four metals complexed with the organic ligand EDTA: Pb, Cu, Zn, Ni). In addition, the field experiment
was designed to collect enough data to investigate the problem of modeling flow coupled with chemical reactions with a
view towards simplification of the geochemical reaction network.
The variable chemical conditions at the field site and the transport of conservative Br and weakly reactive Ni-EDTA
complexes are described in a companion report (Davis et al., 2000a). The main conclusions of the companion report
were that: 1) Br was transported conservatively; 2) EDTA and Ni-EDTA were reversibly adsorbed to the aquifer
sediments during transport, with only slight retardation relative to Br; 3) the masses of total dissolved EDTA and Ni were
nearly constant during the tracer test after the initial adsorptive equilibrium was reached; and 4) the injected metals,
except for Zn and Cr, were transported entirely as metal-EDTA complexes (Davis et al., 2000a,b). Observed tracer
retardation was most extensive under the low pH conditions of the pristine zone, a conclusion consistent with earlier
smaller scale tracer studies of metal-EDTA complexes (Kent et al., 1992; Kent et al., in preparation).
Methods and Materials
Afield site was chosen where the mobility of reactive tracers under variable chemical conditions could be compared, and
the fate of the tracers in response to different chemical processes could be contrasted. The hydrogeology, geochemistry
and injection design were discussed extensively in the companion report (Davis et al., 2000a). Only a brief description
of the pertinent details is included here.
Site Description
The study site is located in a shallow, unconfined sand and grave I aquifer, that is part of a large sand and gravel outwash
plain deposited during the retreat of a continental glacier about 15,000 years ago (Oldale and Barlow, 1986) on Cape
Cod, Massachusetts (see Figure 1, Davis et al., 2000a). The sediments were primarily Fe and Al oxyhydroxide coated
quartz (90-95%), with minor amounts of K-feldspar and ferromagnetic minerals. The fine fraction (<64 microns)
represented about 1% by weight of the sediments and the mass of organic carbon was <0.01%. Carbonate minerals
were not present in the sediments. The average aquifer porosity was 0.39.
-------
The study site has an array of over 800 multilevel samplers oriented along the average flow direction (Figure 1), which
will be discussed further below. The water table was 4 -7 m below land surface and sloped to the south at 1.6 m per 1000
m. Shallow groundwater at the site was contaminated by secondary sewage effluent that was discharged from 1936
through 1995 onto infiltration beds located about 240 m upgradientofthe injection wells. The contaminated water formed
a plume that extended up to 5 km south of the infiltration beds. Vertical concentration gradients in dissolved oxygen
(DO), pH, dissolved Zn, and other water quality parameters were observed near the water table; these gradients were
caused by the mixing of the sewage effluent with pristine, recharge water (Kent et al., 1994; Kent et al., submitted)
(Figure 2). Chemical conditions in the upper portion of the sewage-contaminated region were mildly reducing: Fe(ll)
concentrations were below detection, DO was extremely low, and there were significant concentrations of dissolved
manganese (Mn) and nitrate. These conditions are referred to as suboxic. Figure 2 shows typical chemical profiles in the
aquifer near the water table and at the top of the sewage-contamination plume. Longitudinal contours of the background
geochemical conditions in the aquifer, constructed from concentration data collected prior to the tracer test, were
published in Davis et al. (2000a,b).
For discussion purposes, we have defined three zones of chemical conditions in the groundwater: 1) a pristine zone of
uncontaminated, recharge water, 2) a sewage-contaminated zone that is suboxic, and 3) an approximately 2.5 m thick
Zn-contaminated region between the othertwo zones that includes the region where chemical gradients occurred in DO
and other solute concentrations (Figure 3). The pristine zone is the region near the water table where the groundwater
was close to or at saturation with DO (corresponding to an elevation of 12.5 m or greater). The sewage-contaminated
zone is defined as the region where the dissolved oxygen was approximately zero (<3 uM) and the background dissolved
Zn was below detection (<0.15 u,M), roughly corresponding to an elevation of 10 m or less. The Zn-contaminated region
incorporates the gradients in DO and dissolved salt concentrations, since it was coincident with the zone of mixing
between the sewage-contaminated and pristine groundwaters.
Zn Partitioning in the Zn-contaminated Region of the Aquifer
Atypical vertical profile had a peak in dissolved Zn concentration near the upper boundary of the sewage plume, in the
region where the pristine and sewage-contaminated waters mix (Figure 2). The size, shape, and Zn concentrations in
the Zn-contaminated region varied with distance downgradient (Kent et al., 2000). Figure 4 shows Zn concentrations
and pH values along a longitudinal cross section originating at the injection location. In addition to the longitudinal trends,
there were lateral variations in the shape of the Zn-contaminated region. These variations are illustrated in Figure 5.
From the center of the injection to about 75 m downgradient, both the thickness of the Zn-contaminated region and the
peak concentrations of dissolved Zn increased from west to east across the part of the aquifer traversed by the tracer
cloud. The locations of the upper and lower boundaries varied from west to east throughout the Zn-contaminated region.
Dissolved Zn concentrations in the groundwater were buffered by a reservoir of Zn adsorbed on the aquifer sediments.
Reaetal. (1991) used hydroxylamine-hydrochloride (HH) extractions of aquifer sediments to show the peak in extracted
Zn coincides with the location of the dissolved Zn peak in the groundwater. The amount of extracted Zn on the sediments
was approximately two orders of magnitude greater than that of dissolved Zn (per unit bulk volume of aquifer). Results
of small-scale tracer tests with free EDTA injected at various depths in the aquifer corroborate the HH extraction data
(Kent et al., 1992; Kent et al., in preparation). When free EDTA was injected into the Zn-contaminated region, the mass
ofZn relative to that of conservative tracers increased with distance downgradient (Davis et al, 1993b). No increase in Zn
mass was observed when free EDTA was injected deeper in the sewage plume, below the region with dissolved Zn (Kent
etal., 1992).
The distribution of Zn in the aquifer was controlled by pH-dependent adsorption onto aquifer sediments. Under the
chemical conditions present in the aquifer, Zn-bearing mineral phases, such as hydroxides, carbonates, and phos-
phates, were undersaturated by several orders of magnitude (Rea etal., 1996). Laboratory studies have shown that the
partitioning ofZn between the sediments and artificial groundwater is consistent with adsorption ofZn onto the Fe and Al
oxyhydroxide coatings on the sediment grains (Coston etal., 1995). Experimental data on partitioning ofZn onto aquifer
sediments were well described by a simple surface complexation model (Davis et al., 1998). Figure 6 shows dissolved
Zn concentration as a function of pH for various concentrations of total Zn, computed using the surface complexation
model. For a given value of total Zn (adsorbed plus dissolved Zn), the computed dissolved Zn decreases substantially
with increasing pH. Also shown in Figure 6 are data from two multilevel samplers less than 1 m apart, which were
sampled in 1991 and 1993 at the elevations where the highest dissolved Zn concentrations were observed (11.3 to 12.0
m, Figure 2). Dissolved Zn concentrations were higher and pH values lower in June of 1991 than in April of 1993.
Despite the differences in the observed dissolved Zn concentrations, the surface complexation model simulations
suggest that these field data were consistent with the Zn partitioning expected for a total Zn concentration of
approximately 80 u,M per liter of groundwater.
-------
Tracer Test Experimental Design
The test site had an array of over 809 multilevel samplers (MLS) oriented along the average flow direction (LeBlancetal.,
1991). Each MLS consisted of 15 sampling ports spaced between 25 and 76 cm apart, creating a vertically dense set of
sampling ports across the extant geochemical gradients. Groundwater, spiked with seven tracers (see below), was
injected simultaneously at 3 depths, 1 meter apart, into six MLS installed for the experiment (shown in the inset of
Figure 1). The injection was designed so that a portion of the tracer cloud traveled in each of the aquifer zones depicted
in Figure 3.
In April 1993, 9,884 liters of pristine groundwater were pumped into an above ground plastic-lined pool, and the following
tracers were added: bromide (Br, 3.43 mM), chromate (Cr(VI), 0.506 mM), and four metals (Pb, 0.248 mM; Cu,
0.266 mM; Ni, 0.256 mM; Zn, 0.266 mM) complexed with ethylenediaminetetraacetic acid (EDTA, 1.112 mM). A small
excess of EDTA (7%) was present in the injectate. The predominant species of the reactive tracers in the injectate water
were HCrGy, PbEDTA2-, CuEDTA2-, NiEDTA2-, and ZnEDTA2-. Adsorption of these anions is dependent on pH and the
concentration of competing anions, as observed in previous tracer tests (Kent et al., 1992; 1994; 1995; in preparation)
and in laboratory studies (Bowers and Huang, 1986; Huang etal., 1988; Borggaard, 1991; Nowackand Sigg, 1996). It
was expected that: 1) metal exchange reactions involving Fe(lll) and the metal-EDTA complexes would occur in the
aquifer, releasing uncomplexed tracer metal ions, e.g. Pb2+, to the system; and, 2) Cr(VI) would be reduced to the
immobile species Cr(lll) under suboxic conditions.
Synoptic samplings were conducted for nine months after the injection. A longitudinal transect with pH, and metal
speciation data was also collected during the synoptic sampling rounds. Temporal sampling was conducted at two
breakthrough MLS, 37 and 52 m downgradient from the injection site (Figure 1), where samples were collected
frequently and analyzed for Br, EDTA, total metals and Chelex-filtered metals (Davis et al., 2000a). These MLS are
referred to as BT1 and BT2, respectively. As shown in Figure 4, the sampling ports of these MLS spanned from the
pristine zone to the sewage-contaminated zone and across a pH gradient from 5.4 to 6.2. Both MLS sampled the Zn-
contaminated region at the middle ports and extended well above and below the region at the top and bottom ports.
Laboratory Batch Studies of Fe and Al Dissolution by EDTA
In order to determine the short-term rate of Fe and Al dissolution and Zn desorption by EDTA, a series of experiments
using aquifer sediments collected from the different aquifer zones were completed. The experiments were designed to
study reactions that may have occurred during the first 12 days of the tracer test, prior to the first synoptic sampling.
Core material was collected from the aquifer in 1994 along a transect 2 m east of the MLS array, using a wireline-piston
core barrel and plastic core liners approximately 1.5 m in length (Zapico et al., 1987). Cores were frozen immediately
after collection, and stored frozen until processed. Each core was sectioned at 10 cm intervals, dried under a laminar
flow hood at room temperature, and then sieved to remove grains >1 mm in diameter. A sample of the <1 mm material
from each core section was leached with HH at 50°C for 30 minutes (after Chao and Zhou, 1983) to determine the
amount of amorphous iron and aluminum oxyhydroxides. The amount of Zn extracted in each subsample was used to
determine which intervals represented the Zn-contaminated region in the cores. Three composite samples, represen-
tative of the pristine, Zn-contaminated, and sewage-contaminated zones, were prepared by mixing the appropriate
sieved and dried subsections. BET surface areas were measured by nitrogen gas adsorption (after Coston etal., 1995).
The dissolution experiments used two different artificial groundwater solutions: pristine artificial groundwater (PAGW,
Anderson et al., 1994) and sewage-contaminated artificial groundwater (SAGW, Coston et al., 1995). The composition
of the artificial ground waters approximated the average concentrations of the major ions, such as calcium, magnesium,
potassium, chloride and sulfate. pH was controlled by the addition of 2-N-morpholinoethane sulfonic acid (MES) buffer
to the artificial groundwater solutions. PAGW was adjusted to pH 5.4 and SAGW was adjusted to pH 6.5. Primary EDTA
(Na4EDTA) solutions were made up at least 2 days in advance and stored in the dark.
Since the sediments were dried, the experiments included a pre-equilibration procedure in the pH-adjusted artificial
groundwater (Coston, etal. 1995). All batch experiments had a solid to solution ratio of 400 grams of sediment per liter
of artificial groundwater. After pre-equilibration, EDTA was added from the primary standards, the tubes were covered
in foil and mounted on end-over-end rotators for the desired reaction time (0.5 to 336 hours). The total concentration of
EDTA added was 0.10 mM in each tube. At the end of a reaction period, the pH was measured, and the samples
centrifuged at 24,000g. Three aliquots of supernatant were collected from each tube for chemical analyses: 1) EDTA
concentration, 2) cation concentrations, and 3) metal speciation samples.
Chemical Analyses
Samples were analyzed by a variety of different methods to determine the total and speciated concentrations of the
tracers. Bromide was determined colorimetrically (by flow injection analysis) on unfiltered samples (Franson, 1985). An
aliquot of the same unfiltered sample was used to determine EDTA using the spectrophotometric method of Bhattacharyya
-------
and Kunda (1971). Metal analyses were made on filtered (0.45 micrometer) and acid preserved samples by ICP-AES
(Coston et al., 1998). Free metal ion tracer concentrations were determined in filtered water samples exchanged with
Chelex-100 resin, then acidified and measured by ICP-AES. The concentration of Cr(VI) was measured colorimetrically
in a small subset of anion samples by the addition ofdiphenylcarbazide and HCI (Franson, 1985). The method has been
used successfully in previous groundwater studies at the site (Kent et al., 1994, 1995; Anderson et al., 1994). The
concentration of Cr(VI) measured was always found to be equivalent to the total Cr concentration as measured by ICP-
AES. Additional details about the chemical analyses were included in the companion report to this publication (Davis et
al.,2000a).
Iron-containing colloidal material has been reported to exist within the sewage-contaminated zone at the site. Gschwend
and Reynolds (1987) detected ferrous phosphate particles approximately 0.1 micrometers in diameter in the anoxic zone
(containing dissolved Fe(ll), Kentetal., 1994) of the sewage plume. The tracer cloud did not enterthe anoxic zone, and,
therefore this material could not have caused an observance of dissolved Fe in the samples. Sampling and filtration
experiments showed that small concentrations of Fe (less than 0.5 u,M) derived from Fe-containing particles could be
detected in unfiltered samples at some locations in the suboxic zone (Kent, 1998). These were removed by 0.45
micrometer but not 8.0 micrometer filters. Laboratory column experiments conducted with sediments from the Cape Cod
aquifer have shown that a 1000-fold decrease in ionic strength did not significantly mobilize colloidal material. However,
an increase in pH from 4.8 to 6.3 did mobilize a small concentration of silica-rich colloids in the size range 0.4 to 0.8
micrometers (Roy and Dzombak, 1996, 1997). As was observed in sampling and filtration experiments, material in this
size range would have been filtered out priorto acidification of groundwater samples. Additionally, increases in pH of this
magnitude associated with passage of the tracer cloud were only observed within a few meters of the injection in the
pristine zone (Davis etal., 2000a). Fluctuations in pH during breakthrough of tracers in the suboxic zone and in all zones
farther downgradient of the injection were negligible (Davis et al., 2000a). Thus, it is very unlikely that mobilization of
colloids contributed significantly to the observations of dissolved Fe or Al later in the tracer test.
Spatial and Temporal Moments
The transport parameters of the solutes were quantified from both spatial and temporal data sets. Spatial-moments
estimated tracer mass, location, velocity, and other transport parameters for each synoptic sampling. Results for Br, Ni
and EDTA and details of the integration approach are presented in the companion report (Davis et al., 2000a).
The analysis included here does not correct the computed values of dissolved Zn mass for any Zn mass included from
the background concentrations of dissolved Zn2+. Unlike the othertracers, the boundaries of the spatial integration forZn
could not be defined by concentrations that fell below the detection limit. Instead, for each spatial sampling, the Zn
moments were calculated with the same boundaries used forthe calculation of Ni mass. It is possible that the Zn-EDTA
cloud was smaller than Ni-EDTA for some of the spatial samplings. Because the spatial integration was extended to the
side boundaries where Ni concentrations were zero, some Zn dissolved mass would have been included within that area
due to nonzero background Zn2+ concentrations. As the tracer cloud spread longitudinally with time, the number of MLS
required to establish the side tracer cloud boundaries increased, and thus the contribution of background Zn2+
concentrations to the calculated dissolved Zn mass probably increased.
Tracer was detected in all but the uppermost ports of each of the breakthrough MLS. At both MLS, some of the injected
tracers and Fe were detectable over a broader altitude range than Br. For example, at BT1, Cr, Cu, Fe and Zn were
detected at 13.21 m, 0.3 m above the highest elevation where Br was measured. At BT2, where sampling ports
extended deeper into the aquifer, Zn- and Fe-EDTA complexes were measurable up to 1 meter below the lowest
elevation with measurable Br. Due to the density contrast between the injectate solution and the ambient ground water,
center of mass of Br tracer sank about 1.9 m during transport (Davis et al., 2000a,b). The distribution of metals at the
breakthrough MLS was also consistent with some vertical retardation of the reactive tracers during sinking. The
normalized breakthrough curves (measured concentration/injected concentration: C/Co) were analyzed using the
trapezoidal integration method (Roberts et al., 1986; Harvey and Garabedian, 1991; Kent et al., 1994, 1995). These
results are presented in Tables 1-4.
Results and Discussion
Spatial and Temporal Distribution of Reactive Tracers During Transport
Longitudinal contours of the injected tracers at 13 and 83 days after injection (Figures 7 and 8, respectively) illustrate the
results of reactive transport during the experiment. Differences between the distribution of reactive and nonreactive
tracers among the aquifer zones were already apparent after 13 days (Figure 7). The most striking example was for the
Pb tracer. At 13 days, the Pb cloud presented the most compact outer contour of all the injected tracers. Seventy days
-------
later, only a small amount of Pb persisted in the sewage-contaminated zone (Figure 8). In contrast, both Ni and Cu had
a normalized concentration distribution similar to bromide, except that Ni and Cu transport were retarded in the pristine
zone. Normalized Cr concentrations were less than those of Br and Ni below the oxic/suboxic boundary (Figure 7 and
8), but dissolved Cr was present as far back as the injection MLS in the pristine zone (Figure 8).
The pattern for dissolved Zn deviated from the other tracers, especially within the Zn-contaminated region of the aquifer.
Above and below the region, the 0.002 concentration contour represents Zn added by the tracer test to the pristine and
sewage-contaminated zones. Zn transport was retarded in the pristine zone, with tailing similar to that observed for Ni
and Cu (Figure 7). The pre-existing Zn contamination (the extent of which is shown in gray in Figures 7 and 8) makes
closure of the outermost 0.002 contour uncertain within the Zn-contaminated region. Inside the Zn-contaminated region,
and early in the tracer test, the highest Zn concentration contour was 1.0, five times greater than the highest contour of
the other injected tracers (Figure 7) and the highest Zn concentrations were distributed along the base of the Zn-
contaminated region (Figure 8). Thus, there was a large, localized increase in dissolved Zn within the tracer cloud at the
bottom of the Zn-contaminated region, where the highest pre-injection dissolved Zn concentrations (>4 micromolar) were
measured (Figure 4).
Figures 7 and 8 also include concentration contours for dissolved Feand Al at 13 days, and Fe at 83 days. Thirteen days
after the injection, the outer contours of both the Al and Fe clouds were similar, and the highest concentrations of both
elements were found in the upper portion of the aquifer (the pristine zone). Most of the Al mass was concentrated at
altitudes in the pristine zone, whereas high concentrations of Fe (> 20 micromolar) were measured at lower altitudes in
the Zn-contaminated region and sewage-contaminated zones (Figure 7). These elements were not added as tracers to
the injected groundwater and were therefore evidence for the occurrence of metal exchange and mineral dissolution
reactions. The accumulation of aluminum in the tracer cloud appears to be the result of an intermediate reaction; it was
absent from the tracer cloud by Day 83. Iron continued to accumulate in the tracer cloud at all depths and the distribution
of Fe was nearly as extensive as Ni and Br after 83 days (Figure 8).
Examples of breakthrough curves from various altitudes at each MLS are shown in Figures 9 and 10. In agreement with
the synoptic data discussed above, Fe was measured at BT1 and BT2, while Al was only measured at BT1 (Table 1, data
not plotted). No Pb was measured at either breakthrough MLS; the data shown in Figure 8 are from MLS just west of
BT1.
As altitude decreased, the reactive tracer peaks appeared closer to, but always after, the bromide peaks. Metal-EDTA
complexes are known to adsorb on Fe and Al oxyhydroxide minerals (Bowers and Huang, 1986; Bryce et al., 1994;
Nowackand Sigg, 1996; Nowacket al., 1996) that coat the aquifer sediments (Coston etal., 1995). The peaks of all the
injected metal-EDTA complexes and chromate were retarded to the greatest degree in the pristine zone (Figures 9a,
10a). The anionic complexes were expected to adsorb more as pH decreased, consistent with the ligand-like adsorption
proposed by Benjamin and Leckie (1981). Breakthrough curves in the pristine zone had long trailing limbs, an
observation that is consistent with the long, drawn out tails of longitudinal contours in the pristine zone (Figure 8) and
significant retardation.
The shapes of breakthrough curves at altitudes below the pristine zone generally resembled Br curves from the same
depth; they were more symmetric and had narrower peak widths (Figures 9b-e and 10b-e). At the lower altitudes,
reactive tracers peaked within a day ortwo after bromide, indicating decreased retardation of the reactive tracers. In the
portion of the aquifer where the tracer test was conducted, pH and altitude were negatively correlated; the pH of
groundwater in the sewage-contaminated zone was greater than 6.0 (Figure 4). Also, the concentration of competing
anions (e.g., phosphate) increased (Davis etal., 2000a). Both the higher groundwater pH and phosphate caused weaker
adsorption of the anionic tracers. Thus, the more compact shape of the tracer cloud in the sewage-contaminated zone
reflects the influence of decreased retardation.
A complex breakthrough pattern emerged for Zn in the Zn-contaminated region (Figures 9b,c and 10b,c). Speciation
data indicated a peak in dissolved Zn2+ concentration that appeared consistently before the peak for Zn-EDTA2-. The size
of both the total Zn peak and the Zn-EDTA peaks exceeded bromide, indicating addition of dissolved Zn mass to the
tracer cloud in this region. In contrast to all the othertracers, Zn mass was present as both a cation (Zn2+) and anion (Zn-
EDTA2-). The ionic strength of the tracer cloud was greaterthan that of the ambient groundwater, and as the tracer cloud
advanced, equilibration of the porous medium with the higher ionic strength water of the cloud caused a slight decrease
in pH (Davis et al., 2000a). The decrease in pH and the higher ionic strength may each have contributed to the small
peaks in Zn2+.
Chromium Transport
Figure 11 shows dissolved Cr mass (normalized to initial Cr mass and relative to normalized Br mass) plotted for each
synoptic sampling round. Dissolved Cr decreased to about 50% of the injected Cr mass over the first 100 days after
injection, and continued to decline at a slower rate over the remainder of the test. The distributions illustrated in the
-------
longitudinal cross sections suggest that more Cr mass (i.e., C/Co contours > 0.2) was in the upper portion of the tracer
cloud, above the oxic/suboxic boundary (Figures 7 and 8).
Attenuation of Cr was evident from the reduction in area under the breakthrough curves relative to Br curves. For
example, at BT1 in the pristine zone, the ratio of Crto Brwas2.8 (Table 1), and decreased to a ratio slightly less than 1.0
at 12.1 meters, that is, where the chemical gradients in DO, pH, and competing anions began (Figure 2). Farther
downgradient at BT2, Cr persisted at the intermediate elevations of the transitional zone. Cr was nearly completely
partitioned to the sediments at lower altitudes at BT1 and BT2. Retardation of Cr(VI) in the pristine zone was similar to
that observed for the metal-EDTA anionic complexes (Tables 3 and 4) and was at or near 1.0 in the Zn-contaminated
region and sewage-contaminated zone (Figure 12).
Cr(VI) transport through the aquifer was affected by two major chemical processes: 1) reversible adsorption of Cr (VI),
and 2) reduction to Cr(lll) (Figure 13). The influence of these processes on Cr(VI) transport at this field site was
investigated in an earlier series of small-scale tracer tests at the field site. Two kilometers downgradient of the large-
scale tracer injection, Cr(VI) transport was controlled by weak, reversible adsorption in the pristine zone (Kent et al.,
1995). In a tracer test conducted in the suboxic region, Cr(VI) adsorption was very weak due to the higher pH and the
presence of competing anions, and little retardation was observed (Kentetal., 1994; Friedly etal., 1995). The adsorption
of competing anions has also been shown to influence the pH dependence of molybdate adsorption on aquifer sediments
at this site (Stollenwerk, 1995). Reduction fromCr(VI)toCr(lll) occurs under suboxic conditions, and Cr(lll) is so strongly
adsorbed that it can be considered immobile (Anderson et al., 1994).
Effect of Metal Exchange Reactions on Lead and Copper Transport
Adsorbed metal-EDTA complexes may undergo metal exchange reactions in which one metal is replaced by another in
the complex, including reactions where Fe(lll) dissolves from an Fe oxide and exchanges with a metal complexed by
EDTA (Nowack and Sigg, 1997). Column and batch experiments with pure mineral phases and natural sediments have
shown that metal-EDTA complexes can react with hydrous oxides to form dissolved Al- or Fe-EDTA complexes (Bowers
and Huang, 1987; Davis etal., 1993b; Girvin etal., 1993; Bryce etal. 1994; Brooks, etal. 1996; Nowack and Sigg, 1997).
The overall transport of the metal-EDTA complexes can be qualitatively explained by comparing the zeroth moment
(calculated from the synoptic datasets, see Davis et al., 2000a) for each metal (Figure 14). Although EDTA was not
measured in all synoptic samples, the sum of the dissolved metals, Al, Cu, Fe, Ni, Pb, and Zn, was very well correlated
(R2=0.985) with the EDTA concentrations measured in the temporal and transect samplings (Davis et al., 2000a).
Metal exchange reactions explain the appearance of dissolved Fe and Al within the EDTA tracer cloud, where the Fe and
Al were primarily derived from the dissolution of oxide coatings on the sediment grains in the aquifer, e.g., for Fe:
MeEDTA2- + Fe(OH)3s + >SOH + 2H+ = FeEDTA- + >SOMe+ + 3H2O (1)
where Me is Pb, Cu, Ni, orZn. Measurable concentrations of Al and Fe were dissolved from the sediments within the first
24 hours after the injection (Figure 15). During the first two days of the tracer test, dissolved Al concentrations were
greater than Fe, and the Al concentrations were inversely correlated with pH (Figure 15). The appearance of dissolved
Al in the EDTA tracer cloud was short-lived (Figure 14) and most of the Al mass appeared in the pristine zone and at the
top of the Zn-contaminated region (Figure 7). Although Fe3+forms a stronger complex than AI3+with EDTA, the formation
of Fe(lll)-EDTA complexes apparently was limited by reaction kinetics initially. The Fe(lll)- and AI-EDTA complexes
were less retarded during transport than the reactive tracers whether calculated by the temporal or spatial moment data
sets (Synoptic Rf of Al = Fe =1.10; for breakthrough MLS see Tables 3 and 4).
The mass of dissolved Fe in the EDTA tracer cloud increased over the entire experiment (Figure 14). At chemical
equilibrium, the degree of exchange of the four injected metals with Fe(lll) via the reaction shown in Equation 1 is
different for each metal and is dependent on pH, the concentration of the metal-EDTA complexes, and other chemical
factors. An important consideration was that, for a given set of conditions in the aquifer, the exchange reaction becomes
more favorable with a decrease in the concentration of metal-EDTA complexes. Thus, dilution of the tracer cloud by
dispersion enhances the exchange reaction with Fe(lll) to form Fe(lll)-EDTA.
Transport of Pb was attenuated sharply and dissolved Pb never reached BT1 or BT2 in detectable concentrations. The
retardation factor for Pb, estimated from the synoptic data, was 1.28. Dissolved Pb disappeared from the tracer cloud
completely within 111 days. Data collected during the first seven days of the tracer test along a longitudinal transect
show that significant Pb metal exchange occurred after injection (Figure 16a). If no metal exchange had occurred, i.e.
if Pb-EDTA2- and Cu-EDTA2- complexes behaved like Ni-EDTA2- complexes, then all the data shown in Figure 16a should
collapse to a value of 1.0 on each axis, as the Cu-EDTA2- data does. There was a correspondence between the loss of
dissolved Pb and the addition of dissolved Zn, as Zn-EDTA2-, within the Zn-contaminated region of the aquifer (Figure
16b). Pristine zone samples also had the highest concentrations of Fe- and AI-EDTA complexes during the first 7 days
after injection (Figure 16c). These data indicate the decrease in dissolved Pb was enhanced under pristine and Zn-
contaminated conditions, which is consistent with observations of dissolved Pb in the synoptic data (Figures 7 and 8).
-------
Cu-EDTA exchange for Zn2+ within the Zn-contaminated region of the sewage- contaminated zone probably occurred
during the tracer test (note the break in the 0.1 contour for Cu in Figure 8), but was not significant during the first 7 days
after injection (Figure 16a). Later in the test, the temporal moments calculated from BT1 and BT2 showed significant
attenuation occurred over the entire region (Table 1 and 2). The calculated retardation factors were highest in the
pristine zone and approached 1.0 at sampling ports within the sewage-contaminated zone (Table 3 and 4). Howeverthe
overall retardation, as calculated from the zeroth moments, was 1.24, 4% greater than calculated for Ni.
Cu2+ and Pb2+ formed by metal-exchange reactions (Equation 1) would be strongly adsorbed by the porous medium;
thus, the result of the metal exchange process was a decrease in the dissolved masses of Cu and Pb. In contrast, Ni-
EDTA complexes did not appear to undergo significant metal exchange; the main reaction that affected Ni transport was
adsorption of the Ni-EDTA complexes (Davis et al., 2000a,b). The transported mass of Ni remained essentially constant
throughout the tracer test, while the masses of Pb and Cu decreased significantly (Figures 14 and 17).
Zinc Transport
The effect of the pre-injection Zn2+ contamination in the aquifer on the transport of the metal-EDTA complexes was
significant. Dissolved Zn mass increased over the first 83 days of the experiment, was approximately constant for the
next 100 days, and then decreased after 175 days (Figure 14). At the plateau, the mass of dissolved Zn was greater than
the mass of Ni by a factor of 1.6 to 1.8. The increase in Zn mass must be due to exchange reactions between the metal-
EDTA complexes and the adsorbed Zn in the Zn-contaminated region (Figure 4). For the first 175 days (and 71 m) of
transport, the center of mass for Ni remained in the Zn-contaminated region (Davis et al., 2000a). Outside of the Zn-
contaminated region, dissolved Zn was missing where Ni and Fe were still present (Figure 8). Thus, during transport, Zn
mass was lost from the portion of the tracer cloud outside the Zn-contaminated region, but there was a concurrent
increase in Zn mass within this region.
Analysis of the Zn speciation data collected at BT1 and BT2 suggests the processes by which Zn mass increased during
transport. The injection was designed such that the tracers were injected across vertical gradients in dissolved and
adsorbed Zn. In addition, the tracer cloud was transported horizontally through and out of the Zn-contaminated region;
thus, there were also horizontal gradients in Zn2+ (Figure 4). ZnT/Br and Znfree/Br breakthrough curve area ratios exhibit
different trends in comparison to the other tracers (Tables 1 and 2), where ZnT refers to the total dissolved Zn
concentration and Znfree refers to that portion not complexed with EDTA . The calculated ratios for Zn^Br were as large
as 3 at the Zn-contaminated region altitudes and were consistent with a net increase in dissolved Zn mass along the
sampled flow paths. The largest increase in Zn concentrations occurred at the same altitudes with lower Cu/Br and Fe/
Br breakthrough curve area ratios and near the bottom of the Zn-contaminated region. This observation, along with the
earlier observations regarding Pb transport in the Zn region, suggested that Zn and Fe were competing as reactants in
metal exchange reactions with Cu- and Pb-EDTA complexes.
Reactions that could have increased the dissolved Zn mass in the Zn-contaminated region and early in the tracer test
include the following:
>SOZn++ CaEDTA2- + H+= >SOH + ZnEDTA2-+ Ca2+ (2)
>SOZn+ + PbEDTA2- = >SOPb+ + ZnEDTA2- (3)
>SOZn++ CuEDTA2- = >SOCu++ ZnEDTA2- (4)
where >SOH represents a generic sorption site. The reaction in Equation 2 represents the desorption ofZn from aquifer
sediments in the Zn-contaminated region by the slight EDTA excess in the injectate (shown here as complexed with Ca,
but some was complexed with Mg and Al). Equation 2 would have contributed to increased Zn mass only until Zn had
displaced Ca and other weakly bound cations from the small amount of excess EDTA, a process that was complete
within the first 13 days of the tracer test. Increases in dissolved Zn due to Equation 2 have been observed in small-scale
tracer tests (Kent et al., 1992). The influence of Equations 3 and 4 on dissolved Zn is apparent in the longitudinal
sections through the tracer cloud 13 days after injection, where normalized Zn concentrations greater than 1 can be seen
(Figure 7).
The effect of the background dissolved Zn2+on the zeroth moment for Zn has not been quantified. The injectate solution
initially had 266 u,M dissolved Zn as Zn-EDTA2- complexes (Davis et al., 2000a,b). Increases in dissolved Zn mass could
occur as the equilibrium between adsorbed and dissolved Zn was reestablished; the magnitude of this process would
depend upon the rate at which adsorption equilibrium was reestablished in comparison to the rate of movement of the
tracer cloud. All breakthrough curves from altitudes within the Zn-contaminated region (Figures 9b,c and 10b,c) had
measurable concentrations of Znfree during the period that the EDTA tracer cloud passed, suggesting thatZn desorption
was increasing the Zn2+ concentration back towards its equilibrium value on the time-scale of transport. However, the pH
also decreased slightly at the leading edge of the tracer cloud, and this probably caused some additional desorption of
Zn2+.
-------
Dissolved Zn mass continued to increase slowly after 48 days (Figure 18), primarily due to the reaction in Equation 4. Cu
concentrations were very small in comparison to Ni concentrations within the Zn-contaminated region at 175 days
(Figure 19). By 237 days, relative Cu concentrations (C/Co) were less than 0.1, while relative Ni concentrations greater
than 0.1 still occurred in the Zn-contaminated region. Because of the low Cu concentrations in the Zn-contaminated
region and a continual increase in the fraction of the tracer cloud that traveled outside of this region, the contributions to
dissolved Zn mass by Equation 4 were too small after 175 days to offset the loss of dissolved Zn by Zn-EDTA exchange
with Fe(lll). Therefore, the net result became negative, and dissolved Zn mass decreased substantially in subsequent
synoptic samplings (Figure 18).
Laboratory Studies of the Dissolution of Iron and Aluminum from Aquifer Sediments by
EDTA
Rates of mineral dissolution are known to depend on pH, ionic strength, ligand concentration, and the number of ligand
functional groups coordinated to the surface. Al oxide and oxyhydroxide solubilities are strongly influenced by pH;
however, in natural systems, dissolved silica, phosphate, and sulfate also affect Al solubility (Lindsay and Walthall,
1995). The aquifer sediments at the study site are coated with an assemblage of AI-, Fe-, and Si-bearing minerals
(Coston etal., 1995; Fuller etal., 1996), and the surface mineralogy may be altered in the sewage-contaminated zone by
the presence of various solutes (Mn, phosphate, sulfate) and higher pH. Because the dissolution rates of Fe and Al are
so critical to the fate of metal-EDTA complexes in the aquifer, a separate study of these rates in batch experiments was
undertaken.
The results suggest that numerous factors influenced the rates of metal exchange reactions with Fe and Al during the
field experiment. Figure 20 shows the concentrations of Fe, Al, and EDTA present in batch experiments conducted
under two sets of experimental conditions: 1) "pristine", using sediments from the pristine zone and artificial groundwater
at pH 5.4, with major ion composition similar to that of the pristine zone of the aquifer; and 2) "sewage-contaminated,"
using sediments from the sewage-contaminated zone and artificial groundwater at pH 6.5, with major ion composition
similar to that of the sewage-contaminated zone of the aquifer. Fe dissolution was slower than Al under pristine zone
conditions, but Fe dissolution continued slowly throughout the 14 day reaction period. The concentration of AI-EDTA
complexes peaked during the kinetic experiments (at 4 hours, pristine; at 48 hours, sewage-contaminated), and based
on mass balance considerations, the subsequent dissolution of Fe must have involved an AI-Fe exchange reaction with
AIEDTA- species.
Iron extracted by a 0.5 hour hydroxylamine-HCI (HH) extraction is presumed to represent an estimate of the amount of
amorphous iron oxyhydroxides in the sediment (Chao and Zhou, 1983). The amounts of Fe and Al dissolved by the HH
extraction were determined for each of the sediment composites. The extraction results were normalized by surface
area and compared with the apparent dissolution rates of Fe and Al determined from the EDTA batch experiments
(Table 5). The apparent dissolution rates were calculated based on linear regressions of portions of the curves shown
in Figure 20. In addition to the oxic and sewage-contaminated zone experiments, the results of EDTA batch experiments
with the Zn-contaminated composite sample from the Zn-contaminated region are shown in Table 5. The dissolution
"rates" shown are the net result of several different simultaneous reactions and should not be extrapolated to other
conditions, as the mass distribution between metal-EDTA complexes on the surface was not known, and the values
shown were solely based on the appearance of Fe- or AI-EDTA complexes in solution.
The pristine and sewage-contaminated composites contained similar amounts of HH-extractable Fe, but the Zn-
contaminated composite from the Zn-contaminated region had nearly twice the mass of amorphous Fe (Table 5). Thus,
the fast initial rate of Fe dissolution by EDTA from the Zn-contaminated composite may be due to a larger amount of
amorphous Fe material present. The rate of Fe dissolution from iron oxides by EDTA increases with increasing pH
(Chang and Matijevic, 1983; Nowack and Sigg, 1997), and the difference between the pristine and sewage-contaminated
composites which had similar amounts of amorphous Fe was consistent with the pH dependence of the reaction rate.
These observations suggested that exchange and dissolution reactions with the metal-EDTA complexes (i.e., Equa-
tion 1) were affected by a spatial variability in the abundance of amorphous Fe oxyhydroxide in the aquifer as well as the
pH gradient.
The rate of iron dissolution by EDTA is also known to be dependent on the crystallinity of iron oxide phases (Borggaard,
1976; Borggaard, 1991; Nowack and Sigg, 1997). For example, EDTA dissolution of pure ferrihydrite in the laboratory
is rapid (Rea et al., 1994) and much faster than observed for crystalline iron oxides. However, in natural sediments the
rate of Fe dissolution of poorly crystalline Fe phases by EDTA may be quite slow (Borggaard, 1976; 1979). The
difference between dissolution rates with pure mineral phases and natural sediments is probably due to the formation of
other competing metal-EDTA complexes, and also the presence of competing anions, such as PO4 (Borggaard, 1991) or
silicate (Zachara et al., 1995a) at the mineral surface.
While several studies have been published on the dissolution of Fe oxides by EDTA and metal-EDTA complexes, the
dissolution of Al oxides has not been well studied. Girvin et al. (1993) measured the concentration of dissolved Al in their
-------
experimental system (Co(ll,lll)-EDTA and y-AI2O3) and found that AI-EDTA complexes dominated the solution speciation
below pH 6.5 but were not the dominant adsorbing species. However, the amount of Al dissolved did not exhibit a
consistent trend with respect to either variable metal-EDTA complex concentration or y-AI2O3/water ratio, and the
concentration of dissolved Al was approximated based on a fit to experimental data rather than a particular mineral
solubility, such as gibbsite. In another study, although metal-EDTA complex dissociation was observed in the presence
of y-AI2O3 Bowers and Huang (1986, 1987) did not report whether AI-EDTA complexes were formed.
The slower rate of Al dissolution at the higher pH in the sewage-contaminated experiments (Figure 20) is consistent with
other work on EDTA dissolution of natural sediments at neutral pH (Aggett and Roberts, 1986). Al dissolution in the
sewage-contaminated zone experiments had both fast and slow processes which was not apparent in the experiments
conducted under pristine and Zn-contaminated conditions. There may have been a difference in the surface mineralogy
of the sewage-contaminated composite or possibly, as has been found for Fe, adsorbed phosphate or silicate also
interfered with the dissolution of Al. The high intercept value for Al dissolution under pristine zone conditions suggested
that the lower pH enhanced the solubility of aluminum minerals in the sediment. Nonetheless, these results were
consistent with the tracer test observations, that is the initial formation of AI-EDTA complexes was correlated with lower
pH values (Figure 15), and significant dissolved Al mass was observed only during the beginning of the tracer test
(Figure 14).
The computed intercept of Zn extracted by EDTA from the Zn-contaminated sediments was more than the Zn mass
extracted by the hydroxylamine-HCI extraction (Table 5). HH-extractable Zn may be an underestimate of the Zn mass
capable of forming EDTA complexes in the aquifer. As suggested by Equation 2, a portion of the injected free EDTA was
probably complexed with both Zn and Al in the Zn-contaminated region. Preliminary data from experiments in which
artificial groundwater containing Pb-EDTA complexes was equilibrated with Zn-contaminated aquifer sediments indi-
cated that metal exchange reactions involving Pb and Zn as described by Equation 3 can also occur.
In the batch experiments (Figure 20), the concentrations of Fe-EDTA complexes continued to increase slowly overtime,
suggesting that the process of Fe dissolution and metal exchange was rate-controlled by one or a series of reactions.
This behavior has been observed in other multicomponent systems (Yu and Klarup, 1993). Nowack and Sigg (1997)
demonstrated that rates of Fe dissolution were slower when EDTA was added as a metal complex (such as AI-, Zn-, or
Ni-EDTA), and the effect was more pronounced with poorly crystalline oxyhydroxide phases. It is also important to
consider how the formation and subsequent readsorption of AI-EDTA complexes (and in the Zn-contaminated region,
Zn-EDTA complexes) may effect changes in the observed rate of Fe dissolution. Bowers and Huang (1987) showed that
the addition of Zn- and Fe-EDTA complexes to a suspension of y-AI2O3 resulted in dissociation of the Zn-EDTA complex
and formation of Fe-EDTA complexes at pH values less than 7.0. Hence, the rates reported in Table 5 represent
maximum dissolution rates, as it is expected that the addition of EDTA as a metal-EDTA complex would cause the rates
of dissolution to decrease.
Implications for Reactive Transport Modeling
A principal objective of this tracer test was to create a data set that would stimulate the testing and development of
predictive models of reactive contaminant transport. Data sets generated from field studies are often too poorly
constrained with respect to source terms, boundary conditions, and available data to provide rigorous testing of reactive
transport modeling approaches. Most field and laboratory experimental studies of reactive solute transport have been
conducted under relatively uniform chemical conditions, that, although well-constrained, may not offer challenges to
reactive transport modeling approaches that are sufficiently rigorous. This tracer test was designed such that chemical
conditions in the aquifer varied over the spatial domain of the tracer test.
Spatial variability in groundwater chemistry was mostly provided by the vertical gradients across which the tracer cloud
was injected. Vertical gradients in pH and concentrations of anions (Kent et al., 1994) affected the adsorption, and
therefore the retardation ofCr(VI) and each of the metal-EDTA complexes. Vertical gradients in redox conditions resulted
in changes in the reduction of Cr(VI) with depth. The gradients in pH also affected the displacement of Cu, Zn, and Pb
from EDTA complexes by controlling the solubility of hydrous ferric oxides. In addition to the variations in groundwater
chemistry, the Zn-contaminated region was a distinctive feature of vertical and horizontal variability in chemical
properties of the porous medium. Both the thickness and extent of Zn-contamination varied along the path of the tracer
cloud.
A large network of chemical equations must be solved in reactive transport simulations of this tracertest because of the
number of reactants and products. Considering both the tracers and the chemical constituents of the aquifer, there are
at least 185 solution species, 66 solid phases, and 46 adsorbed species that could be considered in transport modeling.
However, it is possible to simplify the problem by eliminating those reactions and chemical species that do not have a
-------
significant effect on the modeling result. For example, by eliminating species that affected either the activity or total
dissolved concentration of a tracer by less than 1% in the batch geochemical model, HYDRAQL (Papelis et al., 1988),
the chemical problem could be simplified to one with 54 solution species, 3 solid phases, and 38 adsorbed species
(Davis et al., 1993a). Whether the 1% criterion is appropriate or, perhaps, too restrictive for transport modeling will
require further study.
The modeling investigations conducted by our group and collaborators to date have focused on 3 sets of chemical
processes that have been simulated independently from one another. The first modeling study addressed the influence
of redox reactions on Cr(VI) transport. The second study illustrates some of the principal factors that influenced the
extent of metal exchange reactions of the EDTA complexes. The third study demonstrates a viable approach for
modeling the influence of adsorption on transport under variable chemical conditions.
Modeling the Transport of Cr(VI)
Analysis of the results of small-scale tracer tests showed that the observed loss of Cr(VI) in the suboxic zone resulted
from rate-limited reduction of Cr(VI) to Cr(lll), which was immobilized on the sediments (Kent et al., 1994). Fe(ll)-bearing
minerals in the porous medium were the principal reductants (Anderson et al., 1994). The rate of Cr(VI) reduction
observed in well-stirred batch reactors, which was limited by diffusion into the sediment grains, had a time-scale of the
order of 1-2 days (Anderson et al., 1994). The rate of reduction observed during the field experiments, however, was
significantly slower, with a time-scale on the order of 10 days. Model simulations of both laboratory and field
experimental data led to the conclusion that Cr(VI) reduction during transport was limited by the rate of mass transfer to
the surfaces of the reductant mineral phases. On average, the rate of Cr(VI) reduction in the field could be simulated by
the rate of diffusion out of a nonreductive sediment layer approximately 22 cm thick to a thin layer (2.4 cm thick)
containing the reductant minerals (Friedly et al., 1995). Depositional features of thin strata with large abundances of
heavy minerals at this scale have been observed at the field site (Barber, 1990; Hess et al., 1992; LeBlancet al., 1991).
Differences between the results of the large-scale test reported herein and the previous small-scale experiments will
provide a basis for testing the extrapolation of such modeling approaches to large-scale contaminant transport
applications. One difference is the higher initial concentration of Cr(VI) used in the large-scale experiment (506 u,M) in
comparison to 100 u,M used in the small-scale experiments. In the small-scale field experiments, there was a five-fold
excess of reductive capacity in the porous medium compared with the moles of Cr(VI) injected, even without considering
transport. In the large-scale test, the moles of Cr(VI) injected were roughly equivalent to the reductive capacity of the
porous medium initially contacted by the groundwater with tracers. In the transport model (Friedly et al., 1995), a rate
expression that was first order with respect to Cr(VI) concentration was used to describe the mass-transfer rate into the
sediment layers containing the reductant. In accordance with this model, the initial rate of Cr(VI) mass transfer and
reduction should have been much faster during the large-scale test, and Cr(VI) should not have traveled much farther
downgradient in the suboxic zone than was observed in the small-scale tracer tests. Differences in scale between this
experiment and the small-scale tests also mean that contact times of aquifer sediments with the tracers were greater in
the large-scale tracer test because of the length of the tracer cloud.
The fact that the tracer cloud was injected across the pristine, Zn-contaminated region, and sewage-contaminated zones
adds additional challenges to modeling Cr(VI) reduction during transport. Whether Cr(VI) reduction could occur to a
measurable extent in the pristine zone is unknown. Small-scale tracer tests conducted in pristine groundwater at a site
approximately 2 km downgradient showed no evidence of reduction over the short (2 m) transport distance observed
(Kentet al., 1995). However, the large-scale tracer test was carried out much closer to the effluent disposal beds where
redox conditions likely fluctuated historically in response to changes in the quantity and quality of the sewage effluent.
Indeed, Cr(VI) reduction occurred in the suboxic zone even in the presence of dissolved oxygen concentrations high
enough to make Cr(VI) the thermodynamically stable species (Kent et al., 1994). Prolonged exposure to oxygen in the
vadose zone deactivates the Cr(VI)-reductive capabilities of Fe(ll)-containing minerals (White and Peterson, 1996), and
sediments in the pristine zone may have been unsaturated during periods of low recharge. Whether or not Cr(VI)
reduction occurred in the oxic zone, it is likely that the reductive capacity varied with depth through the Zn-contaminated
region (Figure 13). Moment computations for both Brand Ni show that an increasing fraction of the tracer cloud moved
from the oxic zone into the Zn-contaminated region and sewage-contaminated zone during the tracer test (Davis et al.,
2000a,b). Thus, a significant vertical flux of Cr(VI) from the oxic into the Zn-contaminated region and suboxic zones will
have to be accounted for in simulations of Cr(VI) reduction during transport.
Adsorption was also considered in modeling the fate and transport of Cr(VI) in the suboxic zone during small-scale tracer
tests (Friedly et al., 1995). In modeling retardation of Cr(VI) in the suboxic zone, where chemical conditions were
constant, Friedly et al. (1995) used a simple reaction of the form:
>S + Cr(VI) = >SCr(VI) (5)
in which >S represents an adsorption site and Cr(VI) represents the total dissolved Cr(VI) concentration. The stability
constant for this reaction was fit to the Cr(VI) retardation observed in the experiment. This simplified approach cannot
10
-------
be used in the large-scale tracer test because of the vertical gradient in groundwater chemistry in the region into which
the tracer cloud was injected. The gradient in pH and the increase in PO4 and other anion concentrations, resulted in
weaker Cr(VI) adsorption and less retardation in the suboxic zone (Figure 12). Kent et al. (1995) developed a model to
account for Cr(VI) adsorption in the pristine zone as a function of pH and sulfate concentration, however, reactions for
phosphate adsorption should also be included (Kent et al., 1994). Stollenwerk (1995) successfully described molybdate
adsorption onto Cape Cod sediments solids over a similar range of chemical conditions with variable phosphate. In that
study, the set of parameters that fit batch adsorption data also described the main features of molybdate transport in
column experiments conducted over a range of chemical conditions. Adsorption reactions for the metal-EDTA
complexes will also be required to account for differences in the retardation of metal-EDTA complexes observed at
different altitudes in the aquifer. Competitive adsorption of the metal-EDTA complexes may also have influenced Cr(VI)
adsorption.
Modeling Metal Exchange Reactions of Metal-EDTA Complexes
Dissolved Pb, Cu, and Zn were lost or gained within regions of the aquifer as the tracer cloud moved downgradient. The
observed losses in dissolved metals resulted from displacement of the metals from EDTA complexes by Fe(lll) or Al
dissolved or Zn desorbed from the aquifer sediments. In the regions of the aquifer without Zn contamination, the
observed rate of metal exchange decreased in the order Pb»Zn>Cu. There was no evidence that a significant mass of
Ni was lost from the tracer cloud (Davis et al., 2000a,b).
Understanding the overall result of the metal exchange reactions requires a consideration of complexation, dissolution,
and adsorption reactions. The reactions can be written as follows:
MeEDTA2- + Fe(OH)3s + 3H+ = FeEDTA- + Me2+ + 3H2O (6)
Me2+ + >SOsH = >SOsMe+ + H+ (7)
where Me represents Pb, Zn, Cu, or Ni, and >SOsH represents a strong-binding surface site available for adsorption of
the metal ions. The aqueous complexation and Fe(lll)-oxyhydroxide dissolution reactions are combined in Equation 6.
Using a solubility product of-38.8 for Fe(OH)3s (Morel and Hering, 1993) and the aqueous complexation constants of
Martell and Smith (1989) for metal-EDTA complexes, the apparent thermodynamic affinity for the metal exchange
Equation 6 increases in the order Zn»Pb>Ni>Cu (Table 6). In other words, if adsorption (Equation 7) of the metals is
unimportant, then the thermodynamic constants lead to the conclusion that ZnEDTA2- should be affected by the metal
exchange reaction the most, followed by PbEDTA2-, NiEDTA2-, and then CuEDTA2-. The values of stability constants for
adsorption reactions on the Cape Cod aquifer sediments, represented by Equation 7, are only known well for Zn2+at this
point in time (Davis etal., 1998). However, it is believed that the stability constants for Equation 7 also influence the order
of the metal exchange reactions observed in the field. To illustrate this point, consider the stability constants (Table 7)
for adsorption of the metal ions onto the surface of the poorly crystalline mineral, ferrihydrite, a commonly found Fe
oxyhydroxide coating on sediment surfaces, tabulated from Dzombak and Morel (1990). The constants in Table 7 can
only be applied consistently when used with the two-site diffuse double layer surface complexation model of Dzombak
and Morel (1990) and the surface density value of strong-sites that those authors chose. The constants illustrate that the
strength of the adsorption reactions increase in the order Pb»Cu»Zn>Ni. Thus, Pb2+ adsorbs at a lower pH value on
ferrihydrite than the other metal ions. This general preference among the metal ions for adsorption on oxide surfaces is
likely true for the Cape Cod sediments, although the thermodynamic preference for Al oxide coatings may be somewhat
different (Coston et al., 1995). In addition, the proton stoichiometry of the reactions and electrical double layer
corrections in the mass law may differ (Davis et al., 1998). By combining Equations 6 and 7, one obtains the following
reaction:
MeEDTA2-+ Fe(OH)3s + >SOsH + 2H+ = FeEDTA- + >SOsMe+ + 3H2O (8)
The stability constants of Equation 8 can be determined as the product of the constants forthe reactions in Equations 6
and 7. Using ferrihydrite as an analogue forthe Cape Cod sediment (Equation 8), one obtains the stability constants
shown in Table 8. With the two reactions combined, one obtains the same order in affinity of metal exchange as was
observed in the tracer test (outside of the Zn-contaminated region): Pb»Zn>Cu»Ni. Thus, it appears likely that the
adsorption reactions were critical components of the reaction network affecting the transport of Pb, Zn, Cu, and Ni during
the tracer test.
An equilibrium chemical model was developed to illustrate how these reactions may have controlled the evolution of the
loss of dissolved Pb, Zn, Cu, and Ni from the tracer cloud. The equilibrium model included both the strong-and weak-site
adsorption reactions of the four metal ions on the Cape Cod sediment, assuming that the stability constants were the
same as those in Dzombak and Morel (1990). Site density was estimated from the specific surface area of the sediment
(0.44 m2/g), an assumed value of 3.84 mmoles sites/m2 of surface area (Davis and Kent, 1990), also used by Dzombak
and Morel (1990), and the aquifer porosity (0.39, equivalent to 4.14 kgsediment/Lofwater). Adsorption constants for the
metal-EDTA complexes, including Fe(lll), were included in the model, and these constants were estimated from the
11
-------
retardation of metal-EDTA complex transport in the pristine zone of the aquifer. The model calculations were performed
assuming that the background chemistry was similar to that found in the sewage-contaminated zone outside of the Zn-
contaminated region.
Figure 21 shows the calculated equilibrium concentration of Fe(lll)-EDTA complexes that would be dissolved in the
aquifer as a function of pH after reaction with groundwater containing 250 u,M of one of the four injected metal-EDTA
complexes. Note that the model predicts that the favored metal exchange reaction with Fe(lll) is that of the Pb-EDTA
complexes, which are predicted to proceed to near complete exchange for pH values less than 6. Zn-EDTA complexes
are predicted to be the next to exchange with Fe(lll), but only slightly in the pH range 6.0-6.5. Cu-EDTA complexes begin
exchanging at higher pH values than Ni-EDTA complexes. The Cu-EDTA exchange with Fe(lll) has a different slope as
a function of pH than the other 3 metals; the reason for this is that the calculated Cu-Fe exchange occurs in the same pH
range as the Cu adsorption edge on the Cape Cod sand. Forthe others metals, the pH regions forthe adsorption edge
and exchange with Fe(lll) did not overlap significantly.
In addition to pH, the amount of Fe(lll) dissolved at equilibrium is sensitive to the concentration of metal-EDTA
complexes. For example, Figure 22 illustrates the calculated percent of EDTA present as dissolved Ni-EDTA complexes
in the aquifer as a function of pH and the total concentration of Ni-EDTA complexes added to the groundwater. As the
concentration of Ni-EDTA complexes decreases, the metal exchange reaction with Fe(lll) is more favored, and thus
significant Ni-Fe exchange could occur at pH 6 once the tracer cloud was diluted by factors greater than 1000. While it
was possible that dissolution and metal exchange reactions occurred with the Ni-EDTA complexes, we could not quantify
a loss of Ni mass in the field data set (Davis et al., 2000a,b).
The calculations above are meant to illustrate qualitatively that the observed order of metal exchange in the field
experiment may be consistent with the overall thermodynamics of the system. However, the calculations cannot be
considered quantitative because of several major uncertainties in the thermodynamic model:
1) The stability constants used forthe adsorption reactions (Equation 7) in these calculations were derived for
adsorption data with pure ferrihydrite only. Different values forthe metal adsorption reactions would apply
forthe Cape Cod sediments. An additional calculation is shown below forZn2+, for which apparent stability
constants forthe adsorption reactions on Cape Cod sediments have been determined.
2) The adsorption model used in the calculations included a diffuse layer electrical double layer model to
correct chemical equilibria for electrostatic energy contributions at the surface. To be consistent with the
constants used from Dzombak and Morel (1990), it was assumed that the electrostatic energy contribution
(determined from the surface acidity constants) for the Cape Cod sediment is the same as that of the
ferrihydrite surface. This is very unlikely, as the Cape Cod sediment surface should be negatively charged
due to coatings, whereas the pure ferrihydrite surface would be positively charged at pH 6. Even if the
chemical interactions were identical as assumed above, the metal ion adsorption on the Cape Cod sediment
should be stronger than pure ferrihydrite due to a more favorable electrostatic interaction. As an aside,
anion adsorption, such as that of the Cr(VI) tracer, should be weakerthan that observed for pure ferrihydrite.
3) The solubility product of the Fe oxyhydroxide coatings on the Cape Cod sediment may be different than that
used in the calculations.
4) The metal exchange reactions may be rate-limited in the field and may not be well described by equilibrium
calculations.
Forthe synoptic sampling at 13 days, the highest concentrations of dissolved Fe were observed in the pristine zone
(Figure 7). The highest concentration was about 50 u,M at a point where the initial tracer solution was diluted in the ratio
60/40% tracer solution to ambient groundwater. Figure 23 shows the expected percentages of Pb, Zn and Cu present
as EDTA complexes at equilibrium at pH 5.5 (near the values in the pristine zone). If one assumes equilibrium with
respect to dissolution of ferrihydrite is attained (as in Figures 21 and 22), the dissolved Fe concentration is computed to
be about 200 u,M. Under these conditions, the Pb-Fe exchange is essentially complete, but the Zn-Fe exchange only
occurs to a minor extent. Because significant concentrations of dissolved Al were observed (Figure 7), but are not
expected at chemical equilibrium, and because the dissolved Fe observed was only 50 u,M, it is likely that chemical
equilibrium with the tracer cloud was limited by the rate of Fe dissolution. If the equilibrium calculation is performed such
that the system is equilibrated with a dissolved Fe concentration of 50 u,M, rather than equilibrium with ferrihydrite, the
model predicts that Pb-Fe exchange should be about 30% complete at this point in the pristine zone (Figure 23). This is
closer to what was observed at the 13 day synoptic sampling.
Thus, the metal exchange reactions of the Pb-, Zn-, and Cu-EDTA complexes during transport were probably rate-
limited, although the overall order of metal exchange was likely determined by thermodynamic favorability (Table 8). The
persistence of Pb-EDTA complexes in the tracer cloud suggests that the Pb-Fe exchange reaction was rate-limited.
Preliminary reactive transport modeling of the results of small-scale tracer tests suggests that the Zn-Fe exchange
reaction in the aquifer is limited by the rate of Fe dissolution. The results of batch experiments with the Cape Cod
12
-------
sediment in our laboratory also were consistent with the hypothesis that the rates of the metal exchange reactions in the
aquifer were limited by the rate of Fe(lll) dissolution.
The rate-controlling step of exchange reactions of several transition metal-EDTA complexes with Fe(lll) in the presence
of goethite appears to be the breaking of Fe-O bonds at the mineral surface, which is independent of the particular metal-
EDTA complex present (Nowack and Sigg, 1997). However, for poorly crystalline ferrihydrite, the rate of the metal
exchange reaction was found to depend on the particular metal-EDTA complex present (Nowack and Sigg, 1997),
suggesting the rate of free metal ion desorption also influenced the overall rate of metal exchange. Compared to the total
concentration of EDTA injected (1.1 mM), an excess of poorly crystalline Fe (3.7 mmoles/liter of groundwater) was
present on the aquifer sediments contacted immediately after injection, as estimated from chemical extractions (Coston
etal., 1995; Fuller et al., 1996). Thus, if the conclusions of Nowack and Sigg (1997) apply in the aquifer, each individual
metal-EDTA complex could have a different rate for the general rate-controlling reaction (Equation 8) involving Fe(lll)
dissolution and metal-EDTA exchange.
An equilibrium adsorption model for Zn2+ adsorption on Cape Cod sediments (Davis et al., 1998) has been developed
from the results of laboratory batch experiments (Coston et al., 1995). Adsorption of Zn2+ was determined over the
relevant ranges of pH and Zn2+ concentrations that exist in the aquifer. Surface complexation of Zn2+ with "average"
surface sites on the Cape Cod sediments was described with the following two reactions:
>SsOH + Zn2+ = >SsOZn+ + H+ (9)
>S OH + Zn2+ = >S OZn+ + H+ (10)
w w v '
where >SsOH and >SwOH represent high affinity (strong) and low affinity (weak) adsorption sites, respectively. No
electrical double layeTcorrections were made to the mass laws written for these reactions. Best fits of both two-site
(Equations 9 andIO) and one-site (Equation 9 only) surface complexation models to the batch experimental data were
determined. Fits using the two-site model were superiorto those obtained with the one-site model, because of the non-
linear adsorption isotherms for Zn (Figure 24). (Note: For a linear isotherm, the slope of the adsorption data should be
one on the log-log plot of Figure 24). The total site density was estimated from the surface area of the sediment (as
described above), and the density of >SsOH in the two-site model was determined by best fit to the data (0.86% of the
total sites). Values of the stability constants for reactions in Equations 9 and 10 were log K = 0.85 and -2.40, respectively
(Davis et al., 1998). Note that these constants cannot be compared directly to those of Dzombak and Morel (1990)
because of the differing site densities and differing treatment of electrostatic energy terms.
The calculations presented in Figures 21-23 were made using the surface complexation constants for adsorption of
metal ions on ferrihydrite (Dzombak and Morel, 1990). Figure 25 compares equilibrium calculations made with
adsorption constants derived from the batch experiments with Cape Cod sediments with those of Dzombak and Morel
(1990). Adsorption of Zn is stronger on the Cape Cod sediment than on ferrihydrite, and the calculations for 250 u,M Zn-
EDTA show that the Zn exchange reaction with Fe is more favorable at a given pH value if the stronger adsorption
reactions were used in the modeling. The results show how the predicted extent of metal exchange with Feat equilibrium
depends on the values of the adsorption constants. Quantitative modeling will require that the adsorption constants for
the other metal ions on Cape Cod sediments be determined, since the value for ferrihydrite appears to underestimate the
strength of adsorption. Figure 25 also shows the calculated extent of Zn exchange with Fe at lower Zn-EDTA
concentrations using the adsorption constants for Zn on the Cape Cod sediments. Like the case presented for Ni-EDTA
complexes in Figure 22, the metal exchange reaction is favored at lower concentrations. Thus, even at the higher pH
values in the sewage-contaminated zone, Zn-EDTA exchange with Fe would be increasingly favorable as time passed
after injection, due to decreases in the concentrations of Zn-EDTA caused by dispersion and adsorption.
Modeling the Transport of Zn2+
Reactive transport modeling of the results of the tracer test will require an appropriate model for adsorption reactions.
Adsorption of the free metal ions displaced from metal-EDTA complexes resulted in their observed losses from the tracer
cloud. Most of the mass of contaminant Zn is adsorbed to the sediments, and therefore it will be necessary to quantify Zn
desorption in order to model transport of the metal-EDTA complexes in the Zn-contaminated region. Adsorption also
caused retardation of Cr(VI) and the metal-EDTA complexes. Differences in chemical conditions between the pristine
and sewage-contaminated zones affected the extent of adsorption, and thus, the retardation of these anionic solutes.
As shown in Figure 4, the leading edge of the Zn-contaminated region is approximately 155 m downgradient from the
injection. When compared to the leading edge of nonreactive solutes in the sewage plume, the location of the leading
edge of dissolved Zn suggests that Zn transport has been extensively retarded by adsorption onto the aquifer sediments
(Kent et al., 2000). The upper and lower boundaries were quite sharp, as shown in Figure 2. The results of simulations
conducted with a coupled flow and reactive transport model suggest that the sharpness of the upper boundary is
maintained by the small vertical component to flow caused by areal recharge to the aquifer. The sharp lower boundary
is caused by the increase in pH with depth, because the adsorption of Zn onto the sediments increases with increasing
pH (Coston et al., 1995; Kent et al., 2000).
13
-------
As a test of the applicability of the two-site surface complexation model (SCM) developed for Zn from batch studies
described above, the SCM adsorption reactions were included in a modeling exercise to describe Zn transport after
sewage effluent disposal. Simulations were conducted with HYDROGEOCHEM (Yeh and Tripathi, 1991), which allows
the coupling of flow modeling with transport and chemical reactions. By including the SCM approach to describe
adsorption, it was possible to test the influence of variable chemical conditions on the transport of Zn, which likely affects
the observed spatial distribution of Zn contamination in the aquifer. The vertical gradient in pH within the Zn-
contaminated region causes an increase in the extent to which Zn adsorbs on the Cape Cod sediments (Coston et al.,
1995). Zn2+ introduced at the disposal beds is transported in the local flow field, and the extent of adsorption and
retardation during transport is strongly influenced by the vertical gradient in pH. The model implemented coupled flow
with advection, dispersion, and equilibrium adsorption in a 2-dimensional vertical cross section constructed along a flow
line (Kent etal., 2000).
A steady-state flow model was constructed and the resulting flow field was used to set up the reactive transport
simulations. The model included horizontal flow across the upgradient boundary and uniform areal recharge across the
upper boundary and was calibrated using known head distributions, hydraulic conductivity measurements, and
groundwater flow rates. Longitudinal and vertical transverse dispersivities were taken from Garabedian et al. (1991).
The pH gradient was simulated by fixing the total dissolved CO2 concentration of inflowing groundwater at 1 mM
(millimoles per liter) and adjusting the ratio of H2CO3/HCO3- at each altitude to give the desired pH. The H2CO3/HCO3-
ratio in the recharge water was chosen to give a pH of 5.5. A constant and uniform concentration of dissolved Zn2+ (8 u,M)
was input along the upgradient boundary to simulate the inflow of sewage-contaminated groundwater.
Results of a 60-year simulation computed using the 2-site surface complexation model are shown in Figure 26. Results
of simulations conducted with the 1-site model were similar except that the leading edge of the Zn-contaminated region
was much more diffuse. The reactive transport simulations capture the major features of the distribution of Zn-
contamination in the aquifer as mapped in 1993 (Kent et al., submitted). Zn contamination migrated much farther at
lower pH values near the top of the sewage plume than at the higher pH values observed deeper in the sewage effluent
plume. In the downgradient region, the simulated Zn-contaminated region was relatively thin. The leading edge was
located about 350 m downgradient from the effluent disposal beds (Figure 26), which corresponds to 112 m
downgradient from the injection location in the field experiment presented in this report. This is somewhat less than was
observed in the field (ca. 145 m downgradient from the injection location, Figure 4). However, unreactive constituents of
the sewage plume migrated 7.9 km in the simulations (much farther than the observed 5.4 km), and thus the prediction
of Zn retardation was very close to that observed in the field. The simulated Zn-concentration gradient across the leading
edge was sharp in the two-site model, in agreement with the field data. Predicted Zn retardation was greater at the higher
pH values present deeper in the sewage plume (Leading edge about 125 m downgradient of the disposal beds). This
was in reasonable agreement with the few field observations made at these altitudes that suggest that Zn has been
transported a distance on the order of 50-100 m (Rea et al., 1996; Kent et al., submitted). No fitting of the field data for
Zn was done in these simulations; the predicted Zn distribution is a result of the two-site model derived from laboratory
experimental data.
The results show that the two-site model (with a small number of parameters fit to laboratory data) can reasonably predict
Zn transport for a range of chemical conditions. In comparison, a Kd modeling approach would require introducing a
spatially variable range of Kd values because of the effects of pH and Zn concentration on Kd. Figure 27 shows the
calculated dependency of Kd on pH and total Zn concentration for both the one-site and two-site surface complexation
models derived from the laboratory experimental data. Note that only the two-site model has a significant dependence
on the total Zn concentration.
The question of whether one-, two- or multiple-site adsorption models will be required in reactive transport modeling is
important, because multiple site adsorption models increase the number of equations that must be solved. Transition
metal cation adsorption on oxide minerals is characterized by constant pH isotherms that are nonlinear (e.g., Freundlich
isotherms). Describing adsorption in these systems requires a surface complexation model with two or more sites,
similar to that used for Zn on Cape Cod sediments (Davis et al., 1998). For example, Kohler et al. (1996) found that
uranium transport under variable chemical conditions was poorly described by a one-site surface complexation model.
However, a two-site model was able to describe and predict the key features of uranium transport over a range of
chemical conditions. In contrast, one-site models can often predict anion adsorption onto pure hydrous metal oxides for
a range of conditions (Dzombak and Morel, 1990). Even the effects of competition among adsorbing anions have been
successfully predicted with one-site adsorption models for a range of chemical conditions (Zachara et al., 1987; Davis
and Kent, 1990; Mesuere and Fish, 1992; Stollenwerk, 1995; Manning and Goldberg, 1996; All and Dzombak, 1996).
One-site models may also be adequate for describing the adsorption of metal ions displaced from EDTA complexes
(Equation 9), especially in the absence of significant metal contamination.
14
-------
Concluding Remarks
Lessons learned from the modeling investigations that occur as a result of this study will have implications for parameter
estimation that go far beyond the particular tracers studied here. Some important issues in reactive chemical parameter
estimation are summarized in Table 9, based on the work of Rubin (1983, 1990). The upper panel summarizes how
parameters may be estimated for chemical reactions for which local equilibrium can be assumed. For homogeneous
reactions, chemical parameters consist of equilibrium constants taken from the literature, which are generally available.
For classical heterogeneous reactions, i.e., dissolution equilibria, thermodynamic data for many crystalline solids are
also available in the literature. However, the solubilities of poorly crystalline solids, impure solids, or solid solutions are
generally not available and often will have to be determined experimentally with site-specific materials when relevant to
a particular modeling application. Adsorption and other heterogeneous surface reactions will generally require experi-
mental determination of parameters that are site-specific. Modeling of the Zn-contaminated region provides an example
of how this might be accomplished.
Chemical reactions for which the local equilibrium assumption is not valid pose major challenges to predictive modeling
of the fate and transport of reactive contaminants. For these systems, chemical reactions are classified into
homogeneous reactions, heterogeneous reactions for which the overall rate is controlled by one or more chemical
reaction steps, and heterogeneous reactions for which the overall rate is controlled by a physical process, such as
diffusion or mixing across sediment layers. An example of the latter type of reaction is the transport of Cr(VI) in the
suboxic zone, for which the rate of reduction was controlled by mixing of Cr(VI) into regions of the aquifer enriched in
reductant (Friedly et al., 1995). Because the overall rate of reduction is controlled by a process linked to the sediment
structure, rate parameters cannot be easily obtained from laboratory experiments since the sediment structure would be
disturbed. In such cases, the appropriate rate parameters may need to be determined in field experiments. In this
regard, it will be of interest to determine whether rate parameters derived from small-scale tracer tests can be applied in
modeling the results of the large-scale test.
Evaluation of the critical issues in predictive transport modeling for reactive contaminants in groundwater awaits the
testing of such models with well constrained experimental data. The design and execution of the reactive solute tracer
test described here were conducted with this in mind. The chemical complexity and heterogeneity encountered within
the spatial and temporal domain of the experiment were well characterized. Experimental data that describe the temporal
and spatial distribution of the dissolved tracers were collected to the extent feasible, along with supplementary data, such
as that describing metal speciation. Additional investigations will continue to determine which chemical reactions and
processes were most important, in order to further enhance the usefulness of these experimental data in testing
approaches to predictive modeling of the transport of reactive contaminants under variable chemical conditions.
15
-------
References
Aggett, J. and Roberts, L.S. (1986) Insight into the mechanism of accumulation of arsenate and phosphate in Hydro Lake
sediments by measuring the rate of dissolution with ethylenediaminetetraacetic acid (EDTA), Environ. Sci. Technol.,
20, 183-186.
AN, M.A. and Dzombak, D.A. (1996) Competitive sorption of simple organic acids and sulfate on goethite, Environ. Sci.
Technol..30(4). 1061-1071.
Anderson, L.D., Kent, D.B., and Davis, J.A. (1994) Batch experiments characterizing the reduction of Cr(VI) using
suboxic material from a mildly reducing sand and gravel aquifer, Environ. Sci. Technol., 28(1), 178-185.
Barber, L.B., II (1990) Geochemical heterogeneity in a glacial outwash aquifer: Effect of particle size and mineralogy on
sorption of nonionic organic solutes: Boulder, Colo., University of Colorado, Dept. of Geological Sciences,
unpublished Ph.D. thesis, 238 pp.
Benjamin, M.M. and Leckie, J.O. (1981) Conceptual model for metal-ligand-surface interactions during adsorption,
Environ. Sci. Technol., 15(9), 1050-1056.
Bhattacharyya, S.N. and Kunda, K.P. (1971) Spectrophotometric determination of EDTA, Talanta, 18, 446-449.
Borggaard, O.K. (1991) Effects of phosphate on iron oxide dissolution in EDTA and oxalate. Clays Clay Min., 39(3), 321-
325.
Borggaard, O.K. (1976) Selective extraction of amorphous iron oxides by EDTA from a mixture of amorphous iron oxide,
goethite and hematite, J. Soil Sci., 27, 478-486.
Borggaard, O.K. (1979) Selective extraction of amorphous iron oxides by EDTA from a Danish sandy loam, J. Soil Sci.,
30, 727-734.
Bowers, A.R. and Huang, C.P. (1986) Adsorption characteristics of metal-EDTA complexes onto hydrous oxides, J_,
Colloid Interface Sci., 110(2), 575-590.
Bowers, A.R. and Huang, C.P. (1987) Role of Fe(lll) in metal complex adsorption by hydrous solids, Water Res., 21(7),
757-764.
Brooks, S.C., Taylor, D.L., and Jardine, P.M. (1996) Reactive transport of EDTA-complexed cobalt in the presence of
ferrihydrite, Geochim. Cosmochim. Acta, 60, 1899-1908.
Bryce, A.L., Kornicker, W.A. Elzerman, A.W., Clark, S.B. (1994) Nickel adsorption to hydrous ferric oxide in the presence
of EDTA: Effects of component addition sequence, Environ. Sci. Technol., 28, 2353-2359.
Chang, H.C. and Matijevic, E. (1983) Interactions of metal hydrous oxides with chelating agents, J. Colloid Interface Sci.,
92(2), 479-488.
Chao, T. T. and Zhou, L. (1983) Extraction techniques for selective dissolution of amorphous iron oxides from soils and
sediments, Soil Sci. Soc. Amer. J., 47, 225-232.
Coston, J. A., Abrams, R. H., and Kent, D. B. (1998) Selected inorganic solutes, in Savoie, J. and LeBlanc, D. R., eds.,
Geochemical and hydrologic considerations in remediating phosphorus-contaminated ground water in a sewage
plume near Ashumet Pond, Cape Cod Massachusetts, U.S. Geol. Survey Water-Resources Investigations Report
97-4269, p. 11-15.
Coston, J.A., Fuller, C.C., and Davis, J.A. (1995) Pb2+ and Zn2+ adsorption by a natural aluminum- and iron-bearing
surface coating on an aquifer sand, Geochim. Cosmochim. Acta, 59(17), 3535-3547.
Davis, J. A., Hess, K. M., Coston, J. A., Kent, D. B., Joye, J. L., Brienen, P., Campo, K. W., (2000a) Multispecies reactive
tracer test in a sand and gravel aquifer, Cape Cod, Massachusetts, Part I: Experimental design and transport of
bromide and nickel-EDTA tracers, EPA/600/R-01/007a, 37 pp., U. S. Environmental Protection Agency.
Davis, J.A., Coston, J. A., Kent, D.B., and Fuller, C.C. (1998) Application of surface complexation concept to complex
mineral assemblages, Environ. Sci. Technol., 32, 2820-2828.
Davis, J.A., Kent, D.B., Coston, J.A., and Hess, K.M. (1993a) Multi-species reactive solute transport experiment at the
USGS research at Cape Cod, Massachusetts, 3. Minimization of the chemical reaction networks, EOS, Trans. Amer.
Geophys. Union, 74, 280.
Davis, J. A., Kent, D. B., Coston, J. A., Hess, K. M., and Joye, J. L. (2000b) Multispecies reactive tracer test in an aquifer
with spatially variable chemical conditions, Water Resour. Res., 36, 119-134.
Davis, J.A., Kent, D.B., Rea, B.A., Maest, A.S., and Garabedian, S.P. (1993b) Influence of redox environment and
aqueous speciation on metal transport in groundwater: Preliminary results of tracer injection studies, in Allen, H.E.,
Perdue, E.M, and Brown, D.S., Eds., Metals in Groundwater, Chelsea, Ml, Lewis Publishers, 223-273.
16
-------
Davis, J.A. and Kent, D.B. (1990) Surface complexation modeling in aqueous geochemistry, Reviews in Mineralogy, 23,
177-260.
Dzombak, D.A., and Morel, F.M.M. (1990) Surface complexation modeling: Hydrous ferric oxide, New York, Wiley-
Interscience, 393 pp.
Franson, M.A.H (1985) Standard methods for the examination of water and wastewater, 16th ed. American Public Health
Association, Washington, D.C., p. 201-204.
Friedly, J.C., Davis, J.A., and Kent, D.B. (1995) Modeling hexavalent chromium reduction in ground water in field-scale
transport and laboratory batch experiments, Water Resour. Res., 31(11), 2783-2794.
Fuller, C.C., Davis, J.A., Coston, J.A., and Dixon, E. (1996) Characterization of metal adsorption variability in a sand and
gravel aquifer, Cape Cod, Massachusetts, U.S.A., J. Contam. Hvdrol., 22, 165-187.
Garabedian, S. P., Leblanc, D. R., Gelhar, L. W., and Celia, M. A. (1991) Large-scale natural gradient tracer test in sand
and gravel, Cape Cod, Massachusetts, 2. Analysis of spatial moments for a nonreactive tracer, Water Resour. Res.,
27(5), 911-924.
Girvin, D.C., Gassman, P.L., and Bolton, H. (1993) Adsorption of aqueous cobalt ethylenediaminetetraacetate by A1203,
Soil Sci. Soc. Amer., 57, 47-57.
Gschwend, P.M., and M.D. Reynolds, (1987) Monodisperse ferrous phosphate colloids in an anoxic groundwater plume,
J. Contam. Hvdrol.. 1 (3), 309-327.
Harvey, R.W. and Garabedian, S.P. (1991) Use of colloid filtration theory in modeling movement of bacteria through a
contaminated sandy aquifer, Environ. Sci. Technol., 25(1), 178-185.
Hess, K. M, Davis, J. A., Coston, J. A., and Kent, D. B., (1999) Multispecies reactive transport under spatially variable
chemical conditions: dispersion of bromide and nickel tracers, in U.S. Geological Survey Toxic Substances
Hydrology Program — Proceedings of the Technical Meeting, Charleston, S. C., March 8-12, 1999, edited by D.W.
Morganwalp, and H.T. Buxton, U.S. Geological Survey Water-Resources Investigations Report 99-4018C, pp393-
404.
Hess, K.M., Wolf, S.H., and Celia, M.A. (1992) Large-scale natural gradient tracer test in sand and gravel, Cape Cod,
Massachusetts: 3. Hydraulic-conductivity variability and calculated macrodispersivities, Water Resour. Res., 28(8),
2011-2027.
Huang, C.P., Rhoads, E.A., Hao, O.J. (1988) Adsorption ofZn(ll) onto hydrous aluminosilicates in the presence of EDTA,
Water Res.. 22(8), 1001-1009.
Jacobs, L.A., von Gunten, H.R., Keil, R., and Kuslys, M. (1988) Geochemical changes along a river-groundwater
infiltration flow path: Glattfelden, Switzerland, Geochim. Cosmochim. Acta, 52, 2693-2706.
Kent, D.B., (1998) Effects of pumping rate and filtration on measured concentrations of inorganic solutes, J. Savoie, and
D.R. LeBlanc, Eds.,U.S. Geological Survey Water-Resources Investigations Report 97-4269, pp. 11-15.
Kent, D. B., Abrams, R. H., Davis, J. A., Coston, J. A., and LeBlanc, D. R., (2000) Modeling the influence of variable pH
on the transport of zinc in a contaminated aquifer using semi-empirical surface complexation models, Water Resour.
Res.. 36. 3411-3425.
Kent, D.B., Davis, J.A., Anderson, L.C.D., and Rea, B.A. (1995) Transport of chromium and selenium in a pristine sand
and gravel aquifer: Role of adsorption processes, Water Resour. Res., 31(4), 1041-1050.
Kent, D.B., Davis, J.A., Anderson, L.C.D., Rea, B.A., and Waite, T.D. (1994) Transport of chromium and selenium in the
suboxic zone of a shallow aquifer: Influence of redox and adsorption reactions, Water Resour. Res., 30(4), 1099-
1114.
Kent, D.B., Davis, J.A., Rea, B.A., and Anderson, L.C.D. (1992) Ligand-enhanced transport of strongly adsorbing metal
ions in the ground-water environment, in Kharaka, Y. and Maest, A.S., Eds., Water-Rock Interactions WRI-7,
Rotterdam, A.A. Balkema, 1, 805-808.
Kent, D.B., Davis, J.A., Anderson, L.C.D., Rea, B.A., and Coston, J.A., Effect of adsorbed metal contaminants on the
transport of Zn- and Ni-EDTA complexes in a sand and gravel aquifer, Environ. Sci. Technol., submitted.
Killey, R.W.D., McHugh, J.O., Champ, D.R., Cooper, E.L., and Young, J.L. (1984) Subsurface cobalt-60 migration from
a low-level waste disposal site, Environ. Sci. Technol., 18, 148-157.
Kohler, M., Curtis, G.P., Kent, D.B., and Davis, J.A. (1996) Experimental investigation and modeling of uranium(VI)
transport under variable chemical conditions, Water Resour. Res., 32(12), 3539-3551.
LeBlanc, D.R., Garabedian, S.P., Hess, K.M., Gelhar, L.W., Quadri, R.D., Stollenwerk, K.G., and Wood, W.W. (1991)
Large-scale natural gradient tracer test in sand and gravel, Cape Cod, Massachusetts: 1. Experimental design and
observed tracer movement, Water Resour. Res., 27(5), 895-910.
17
-------
Lindsay and Walthall (1995) The solubility of aluminium in soils, in Sposito, G., Ed., The Environmental Chemistry of
Aluminum, 2nd Edition, Boca Raton, FL, CRC Press, 333-361.
Liu, C.W. and Narasimhan, T.N. (1989) Redox-controlled multiple-species reactive chemical transport, 1, Model
development, Water Resour. Res., 25, 869-882.
Manning, B.A. and Goldberg, S. (1996) Modeling competitive adsorption of arsenate with phosphate and moybdate on
oxide minerals, Soil Sci. Soc. Amer., 60, 121-131.
Martell, A.E. and Smith, R.M. (1989) Critical Stability Constants, Vol. 6, Second Supplement, New York, Plenum, 604 pp.
McBride, M. B., Richards, B. K., Steenhuis, T., Russo, H. J., and Sauve, S. (1997) Mobility and solubility of toxic metals
and nutrients in soil fifteen years after sludge application, Soil Sci., 162, 487-500.
McCarthy, J. F., Czerwinski, K. R., Sanford, W. E., Jardine, P. M., and Marsh, J. D. (1998) Mobilization of transuranic
radionuclides from disposal trenches by natural organic matter, J. Contam. Hvdrol., 30, 49-77.
Mesuere, K. and Fish, W. (1992) Chromate and oxalate adsorption on goethite (a-FeOOH), Environ. Sci. Technol., 26,
2365-2370.
Morel, F.M.M. and Hering, J.G. (1993) Principles and applications of aquatic chemistry, New York, John Wiley, 588 pp.
Nowack, B., Lutzenkirchen, J., Behra, P., and Sigg, L. (1996) Modeling the adsorption of metal-EDTA complexes onto
oxides, Environ. Sci. Technol., 30, 2397-2405.
Nowack, B. and Sigg, L. (1996) Adsorption of EDTA and metal-EDTA complexes on goethite, J. Colloid Interface Sci.,
177,106-121.
Nowack, B. and Sigg, L. (1997) Dissolution of Fe(lll) (hydr)oxides by metal-EDTA complexes, Geochim. Cosmochim.
Acta. 61.951-963.
Oldale, R. N. and Barlow, R.A. (1986) Geologic map of Cape Cod and the Islands, Massachusetts, U.S. Geological
Survey, Miscellaneous Investigations Series, Map 1-1763, 1 plate.
Papelis, C., Hayes, K.F., and Leckie, J.O. (1988) HYDRAQL: A program for the computation of chemical equilibrium
composition of aqueous batch systems including surface-complexation modeling of ion adsorption at the oxide/
solution interface, Dept. of Civil Engineering, Stanford University, Technical Report, 306, 130 pp.
Rea, B.A., Kent, D.B., Anderson, L.C.D., Davis, J.A., and LeBlanc, D.R. (1996) The transport of inorganic contaminants
in a sewage plume in the Cape Cod aquifer, Massachusetts, in Morganwalp, D.W. and Aronson, D.A., Eds., U.S.
Geological Survey Toxic Substances Hydrology Program—Proceedings of the technical meeting, Colorado Springs,
Colorado, September 20-24,1993; U.S. Geological Survey Water-Resources Investigations Report, 94-4015, 191-
198.
Rea, B.A., Davis, J.A., and Waychunas, G.A. (1994) Studies of the reactivity of the ferrihydrite surface by iron isotopic
exchange and Mossbauer spectroscopy, Clays Clay Min.,42, 23-34.
Rea, B.A., Kent, D.B., LeBlanc, D.R., and Davis, J.A. (1991) Mobility of zinc in a sewage-contaminated aquifer, Cape
Cod, Massachusetts, in Mallard, G.E. and Aronson, D.A., Eds., U.S.Geological Survey Toxic Substances Hydrology
Program—Proceedings of the technical meeting, Monterey, California, March 11-15, 1991, U.S. Geological Survey
Water-Resources Investigations Report, 91-4034, 88-95.
Roberts, P.V., Goltz, M.N., and Mackay, D.M. (1986) A natural gradient experiment on solute transport in a sand aquifer,
3. Retardation estimates on mass balances for organic solutes, Water Resour. Res., 22(13), 2047-2058.
Roy, S.B., and D.A. Dzombak, Colloid release and transport processes in natural and model porous media, Colloids
SurL, 107, 245-262, 1996.
Roy, S.B., and D.A. Dzombak, Chemical factors influencing colloid-facilitated transport of contaminants in porous media,
Environ. Sci. Technol., 31, 656-664, 1997.
Rubin, J. (1983) Transport of reacting solutes in porous media: Relation between mathematical nature of problem
formulation and chemical nature of reactions, Water Resour. Res., 19,1231-1252.
Rubin, J. (1990) Solute transport with multisegment, equilibrium-controlled reactions: A feed forward simulation method,
Water Resour. Res.. 26(9), 2029-2055.
Smith, R.M. and Martell, A.E. (1989) Critical Stability Constants, 6, Second Supplement, New York, Plenum, 643 pp.
Stollenwerk, K.G. (1995) Modeling the effects of variable groundwater chemistry on adsorption of molybdate, Water
Resour. Res.. 31(2), 347-357.
von Gunten, H.R., Karametaxas, G., Krahenbuhl, U., Kuslys, M., Giovanoli, R., Hoehn, E., and Keil, R. (1991) Seasonal
biogeochemical cycles in riverborne groundwater, Geochim. Cosmochim. Acta, 55(12), 3597-3609.
18
-------
White, A.F. and Peterson, M.L. (1996) Reduction of aqueous transition metal species on the surfaces of Fe(ll)-containing
oxides, Geochim. Cosmochim. Acta, 60(20), 3799-3814.
Yeh, G.T. and Tripathi, V.S. (1991) A model for simulating transport of reactive multispecies components: Model
development and demonstration, Water Resour. Res., 27(12), 3075-3094.
Yu, J. and Klarup, D. (1993) Extraction kinetics of copper, zinc, iron, and manganese from contaminated sediment using
disodium ethylenediaminetetraacetate, Amer. Chem. Soc., 205th National meeting, Denver, CO, 33(1), 69-72.
Zachara, J.M., Gassman, P.L., Smith, S.C., and Taylor, D. (1995a) Oxidation and adsorption of Co(ll)-EDTA complexes
in subsurface materials with iron and manganese oxide grain coatings, Geochim. Cosmochim. Acta, 59(21), 4449-
4464.
Zachara, J.M., Smith, S.C., and Kuzel, L.S. (1995b) Adsorption and dissociation of Co-EDTA complexes in iron oxide-
containing subsurface sands, Geochim. Cosmochim. Acta, 59(23), 4825-4844.
Zachara, J.M., Girvin, D.C., Schmidt, R.L., and Resch, C.T. (1987) Chromate adsorption on amorphous iron oxyhydroxide
in the presence of major groundwater ions, Environ. Sci. Technol., 21, 589-594.
Zapico, M.M., Vales, S. and Cherry, J.A. (1987) A wireline piston core barrel for sampling cohesionless sand and gravel
below the water table, Ground Water Monit. Rev., 7(3), 74-82.
19
-------
Tables
20
-------
Table 1. Data for the Breakthrough MLS, BT1 (37 m downgradient) [Areas under the Breakthrough Curves for Br and the Ratio of Areas for
Reactive Tracers to Br Areas]
Area, days
Altitude, m
Br
Area(i)/
Area(Br)
EDTA
Cr
Cu
M
Zn-total
'free'Zn
Al
FeT
Pristine zone
13.51
13.21
12.90
12.60
n/s
n/s
0.44
1.24
n/s
0
1.01
0.79
0
[0.40]
2.84
1.13
0
0
0.02
0.56
0
0
0.75
0.66
0
0
0.26
0.49
n/s
n/s
0
n/s
0
0
0
0
0
[0.15]
1.31
0.47
Zn-contaminated
region
12.14
11.68
11.23
10.77
10.30
Sewage-
2.65
2.29
3.32
4.30
3.23
0.59
0.90
0.84
0.65
0.84
0.63
1.06
0.80
0.53
0.55
0.70
0.83
0.77
0.49
0.63
0.78
1.04
0.97
0.75
0.97
1.17
2.19
2.17
2.42
3.30
0.24
n/s
0.52
n/s
0.58
0.01
0
0
0
0
0.22
0.22
0.18
0.13
0.13
contaminated zone
9.85
9.39
8.93
8.47
8.02
7.56
3.75
5.12
1.24
0.57
2.27
0.16
1.04
0.94
0.86
0.58
0.70
1.20
0.40
0.06
0
0
0
0
0.75
0.98
0.77
0.15
0.55
0.84
1.09
1.07
0.86
0.31
0.74
0.98
2.52
0.85
0.23
0
0.25
0.15
n/s
n/s
n/s
n/s
n/s
n/s
0
0
0
0
0.01
0
0.16
0.19
0.20
0.13
0.24
0.26
n/s = samples not collected
[] indicates the area under the breakthrough curve of a constituent at an elevation where no Br was measurable.
21
-------
Table 2. Data for the Breakthrough MLS, BT2 (52 m downgradient) [Areas under the Breakthrough Curves for Br and the Ratio of Areas for
Reactive Tracers to Br Areas (Al Was Not Detected at this MLS)]
Altitude, m
Pristine zone
13.65
13.14
12.63
Zn-contaminated
region
12.13
11.62
11.11
10.60
10.10
Sewage-
contaminated zone
9.59
9.08
8.58
8.07
7.56
7.05
6.54
Area, days
Br
n/s
0.03
0.89
2.85
2.90
3.48
4.20
4.29
3.54
2.06
0.89
0.53
0.06
n/s
n/s
Areaft)/
Area(Br)
EDTA
n/s
0
0.99
0.65
0.84
0.72
0.92
0.98
0.98
0.88
1.54
0.97
1.03
0
0
Cr
0
0
0.79
0.75
0.72
0.57
0.69
0.39
0.12
0
0
0
0
0
0
Cu
0
0
0.14
0.57
0.62
0.48
0.67
0.57
0.75
0.70
1.30
0.69
0.19
0
0
M
0
0
0.46
0.72
1.12
0.87
1.12
1.17
1.06
0.88
1.59
0.93
0.32
0
0
Zn-total
0
1.68
0.64
1.25
2.02
2.32
3.10
2.90
1.91
0.11
0.13
0.11
0.32
0
0
'free'Zn
n/s
n/s
0.06
n/s
0.97
n/s
1.52
n/s
0.84
n/s
n/s
n/s
n/s
n/s
n/s
FeT
0
2.02
0.72
0.25
0.30
0.22
0.20
0.21
0.21
0.25
0.43
0.35
0.17
0
0
n/s = samples not collected
22
-------
Table 3.
Data for the Breakthrough MLS, BT1 (37 m downgradient) [Estimated Retardation Factors]
Altitude, m
Pristine zone
13.51
13.21
12.90
12.60
Zn-contaminated
region
12.14
11.68
11.23
10.77
10.30
Sewage-
contaminated
zone
9.85
9.39
8.93
8.47
8.02
7.56
Avg.
days
Br
n/s
n/s
61.2
56.7
67.4
82.1
77.7
67.9
71.0
85.5
94.2
97.7
93.1
90.0
79.1
Average
Rfi
/
EDTA
n/s
0
1.34
1.32
1.16
1.22
1.16
1.10
1.04
1.07
1.06
0.96
1.01
1.04
1.05
Cr
0
[159]
2.32
1.70
1.28
1.28
1.23
1.15
1.07
1.10
1.09
0
0
0
0
Cu
0
0
2.02
1.53
1.23
1.26
1.19
1.12
1.05
1.09
1.07
0.96
1.08
1.04
1.03
M
0
0
2.07
1.38
1.17
1.21
1.15
1.10
1.04
1.06
1.05
0.94
1.09
1.03
1.03
Zn-total
0
0
2.73
1.65
1.24
1.24
1.20
1.18
1.13
1.08
1.05
0.96
0
1.04
1.02
'free'Zn
n/s
n/s
0
n/s
1.14
n/s
1.29
n/s
1.14
n/s
n/s
n/s
n/s
n/s
n/s
Al
0
0
0
0
1.12
0
0
0
0
0
0
0
0
1.02
0
Fe
0
[86.9]
1.33
1.28
1.10
1.11
1.07
1.05
1.02
1.04
1.04
0.93
1.03
1.01
1.01
n/s = samples not collected
[ ]indicates the area under the breakthrough curve of a constituent at an elevation where no Br was
measurable.
23
-------
Table 4.
Data for the Breakthrough MLS, BT2 (52 m downgradient) [Estimated Retardation Factors (Al was not detected at this MLS)]
Avg.
days
Altitude, m
Pristine zone
13.65
13.14
12.63
Zn-contaminated
region
12.13
11.62
11.11
10.60
10.10
Sewage-
contaminated
zone
9.59
9.08
8.58
8.07
7.56
7.05
6.54
Br
n/s
82.7
78.3
96.9
111
98.2
95.5
114
125
134
126
110
99.1
n/s
n/s
Aver age Rf,
/
EDTA
n/s
0
1.14
1.30
1.14
1.14
1.07
1.08
1.10
1.07
1.13
1.07
1.04
0
0
Cr
0
0
1.87
1.36
1.18
1.17
1.11
1.14
1.21
0
0
0
0
0
0
Cu
0
0
1.81
1.34
1.16
1.14
1.10
1.10
1.12
1.08
1.16
1.06
1.05
0
0
Ni
0
0
1.67
1.27
1.13
1.11
1.07
1.07
1.09
1.06
1.13
1.03
1.04
0
0
Zn-total
0
2.15
1.63
1.31
1.15
1.21
1.18
1.09
1.06
1.09
1.17
1.17
1.55
0
0
'free'
Zn
n/s
n/s
1.08
n/s
1.19
n/s
1.35
n/s
1.03
n/s
n/s
n/s
n/s
n/s
n/s
Fe
0
1.34
1.17
1.14
1.06
1.06
1.04
1.03
1.05
1.03
1.07
1.01
1.02
0
0
n/s = samples not collected
24
-------
Table 5.
Dissolution of Al, Fe, and Zn from Aquifer Composite Sediments by Hydroxylamine-HCI and EDTA
Pristine Sewage- Zn-contaminated
contaminated
Surface Area, m2/g
Conditions
Hydroxylamine-HCI extraction,
umoles/m2
Apparent EDTA dissolution rates
Fast reaction, umoles/m2 - hour
Intercept, umoles/m2
R2
Time range, hours
Slow reaction, umoles/m - hour
Intercept, umoles/m2
R2
Time range, hours
0.36
Al
10.3
Fe
1.61
0.33
Al
6.55
Fe
1.70
0.24
Al
14.1
pH5.4 pH6.5
0.040
0.32
0.83
0.5-4
None
0.0019
0.16
0.96
0.5-
216
0.0005
0.42
0.95
216-
336
0.018
0.020
0.99
0.5-8
0.0014
0.16
0.98
8-48
0.007
0.076
0.84
0.5-48
0.0006
0.33
0.99
48-
336
0.014
0.066
0.85
0.5-24
None
Fe
3.08
pH6.5
0.017
0.062
0.96
0.5-24
0.001
0.39
0.83
24-
336
Zn
0.46
0.0003
0.008
0.94
0.5-48
None
none = metal-EDTA complex is undergoing resorption and exchange back reaction.
Table 6.
Stability Constants of Metal Exchange Reactions with Aqueous Metal Ions
Me
Cu2+
Ni2+
Pb2+
Zn2+
9.5
9.9
10.3
11.8
25
-------
Table 7.
Stability Constants for Metal Ion Adsorption on Ferrihydrite
_Me
Ni2+
Zn2+
Cu2+
Pb2+
0.4
1.0
2.9
4.7
Table 8.
Stability Constants of Metal Exchange Reactions with Adsorbed Metal Ions
Me
Ni
2+
Cu
2+
Zn
Pb
2+
2+
10.3
12.4
12.8
15.0
Table 9. Issues Facing Estimation of Chemical Parameters in Reactive Transport Models
Level
"Sufficiently Fast"
and Reversible
"Insufficiently Fast"
Class
Homogeneous
Heterogeneous
Classical
Heterogeneous
Surface
Homogeneous
Heterogeneous
Chemical-control
Heterogeneous
Physical-control
Parameter Estimation
Literature
Literature and using established laboratory experimental
procedures (poorly characterized participants)
Site-specific parameters, require laboratory experiments
Probably can be determined in laboratory experiments
Might be determined in laboratory experiments;
conceptual hurdles.
Process controlling rate linked to sediment structure. Field
experiments required.
26
-------
Figures
27
-------
Distance from injection center, meters
30
40
60
70
150
160
170
180
190
200
Injection Multilevel
Sampler
Breakthrough Curve
Multilevel Sampler
A Multilevel Sampler
Used to Define
Zinc Contamination
-BT2
f
1A Magnetic
North
General Path
• -I • I ... of the Tracers
i
i
i
i
20 10 0 -10 -20 -30 -40
Distance from injection center, meters
-50
Figure 1. Location of tracer-test site in western Cape Cod, Massachusetts, the general path of tracers, the multilevel samplers (MLS)
available for sampling during the tracer test, the six injection MLS, the two breakthrough curve MLS, and the MLS used to
construct background chemistry transects and to define the extent of zinc contamination.
28
-------
B, |iM DO, |iM
0 10 20 30 40 50 0 100 200 300
JD
E
~
03
CD
CO
CD
O
JD
03
CD
• f <
LJ
13
-^
H— »
m A
> • A ~
i m A -
* • A -
> • A -
• Zn " A -
A B
> • pH |A
;
0
+
4 6
Zn,
j-
-h
-h
•h
•h
•h
-h
-h
10 0
+ o
o
o
o
O
0
o
o
+
O
Mn
DO
10
Mn,
400
15
20
CD
1^
CD
CD
CO
c
CD
CD
5
o
CD
T3
13
11
5.0
5.5
6.0
PH
6.5
7.0
Figure 2. Vertical profiles of B (boron), Zn (zinc), pH, DO (dissolved oxygen), and Mn (manganese) 1.7 m downgradient from the center of
the injection taken just prior to injection at MLS 2414A (shown in inset of Fig. 1 as filled triangle closest to the array of injection
MLS). The shaded area shows the location of the Zn-contaminated region for this MLS.
N
s s / /
-------
BT 1 BT 2
20
40 60 80 100 120
Distance downgradient from injection center, meters
140
160
180
200
Figure 4. Longitudinal cross section showing the concentration of dissolved Zn (|iM) and pH in the aquifer prior to the field experiment. The
area shown is downgradient from the injection location. Location of the transect is shown in Figure 1. The small circles indicate the
locations where groundwater was collected. The filled small circles indicate the ports in which Br and reactive tracers were
detected in samples taken just after the injection. The vertical extent covered by the two breakthrough MLS (BT1 and BT2) is
indicated by the solid bars.
0>
-i— •
0>
E
ro
0>
C/5
c
ro
0>
E
0>
o
.a
ro
0>
-o
Figure 5. Vertical profiles showing longitudinal and lateral heterogeneity in the distribution of background dissolved Zn in the aquifer. Panels
A through F show sets of dissolved Zn profiles taken approximately 2, 26, 70, 115,135, and 150 meters downgradient, respec-
tively. Each panel shows vertical profiles taken west of the region traversed by the tracer cloud, near the center of the region
traversed by the tracer cloud, and east of the region traversed by the tracer cloud. Locations of MLS used to construct these
profiles are shown as filled triangles in Figure 1.
30
-------
12.0
§* 10.0
zL
^ 8.0
N
"8 6.0
_>
° 4.0
c/>
b 2.0
0.0
ZnTOT = 80 LiM
-••- ZnTOT = 90 LiM
•-• ZnTOT = 70 LiM
A 24-14A, 4/93
O 24-15,6/91
5.0
5.5
PH
6.0
6.5
Figure 6. Dissolved Zn as a function of pH. Curves show Zn concentrations versus pH for three values of total Zn (adsorbed plus dissolved),
computed from the surface complexation model fitted to batch adsorption data (Davis et al., 1998). Also shown are data from MLS
2415, sampled in June of 1991, and 2414A, sampled before the injection (Fig. 2). The two MLS are adjacent to one another and
the three data points come from the same altitude range (11.3 to 12.0 m to sea level). Location of MLS 2414A is shown in inset of
Figure 1 (filled triangle closest to the array of injection MLS).
31
-------
as
-------
"ffi
ro
o>
en
E
s
5
(C
T3
contamination
region
1° ' 0.001
oxic/suboxic
boundary 0.02
0.2-
15
10
10
5
15
10
10
10
10
0.005
0 10 20 30 40 50 60 70
Distance downgradient from injection center, meters
Figure 8. Longitudinal cross sections showing the normalized concentrations of the tracers Br, Ni, Cr, Zn, Cu, and Pb and the dissolved
concentration of Fe 83 days after the injection. The Zn-contaminated region and oxic/suboxic are same as in Figure 7.
33
-------
a.
b.
Pristine zone, altitude = 12.9 meters
37 m downgradient
0.03
0.02 -
o
c
CD
O
O
O
T3
CD
N
"ro
E
o
0.01 '
o
c
CD
O
O
O
T3
CD
N
"ro
E
o
0.00
30
120 150 180
Days since injection
210
Zn-contaminated region, altitude = 11.23 m
37 m downgradient
60 90 120 150 180
Days since injection
210
Figure 9. Breakthrough curves for BT1, 37 m downgradient of the injection site. Concentrations are normalized to the injectate concentra-
tions for each element, except Fe, which is normalized to the concentration of injected EDTA: a.) pristine zone; b.) upper Zn-
contaminated region; c.) lower Zn-contaminated region; d.) upper sewage-contaminated zone, below Zn contamination; e.) lower
sewage-contaminated zone.
34
-------
c.
Zn-contaminated region, altitude = 10.3 m
37 m downgradient
0
60
90 120 150
Days since injection
180 210
d.
Sewage-contaminated zone, altitude = 8.47 m
37 m downgradient
g
1
0
o
o
O
1
"TO
E
o
U.Ub
0.05
0.04
0.03
0.02
0.01
11111
-•- Br
-0- ZnT '
J\ -O- Ni
f 1 -A- Cu "
M
j \
/A)
r 3^^tx
0.06
0.05
0.04
0.03
0.02
0.01
n nn
-*- EDTA
-0- ZnT
-O Fe
-
-
/\
30 60 90 120 150 180 2'
Days since injection
e.
Sewage-contaminated zone, altitude = 8.02 m
37 m downgradient
o
03
O
C
o
O
"8
N
"CO
O
Figure 9.
Continued.
60 90 120 150 180
Days since injection
210
35
-------
a.
§
o
O
T3
CD
N
"ro
E
Pristine zone, altitude = 12.63 m
52 m downgradient
b.
Zn-contaminated region, altitude = 11.62 m
52 m downgradient
g
15
CD
O
O
O
N
"ro
90 120 150 180 210
Days since injection
240
90
120 150 180 210
Days since injection
240
Figure 10. Breakthrough curves for BT2, 52 m downgradient of the injection site, a.) pristine zone; b.) upper Zn-contaminated region; c.) lower
Zn-contaminated region; d.) upper sewage-contaminated zone, below Zn contamination; e.) lower sewage-contaminated zone.
36
-------
c.
Zn-contaminated region, altitude = 10.6 m
52 m downgradient
90 120 150 180 210
Days since injection
240
d.
Sewage-contaminated zone, altitude = 8.58 m
52 m downgradient
c
o
03
O
C
o
O
T3
0)
N
To
0.06
0.05
0.04
0.03
0.02
0.01
0.00
0.06
0.05
0.04
0.03
0.02
0.01
0.00
60 90 120 150 180 210 240
Days since injection
e.
Sewage-contaminated zone, altitude = 8.07 m
52 m downgradient
o
0)
o
o
O
T3
0)
N
Figure 10. continued.
90 120 150 180 210 240
Days since injection
37
-------
00
1.0'
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
Cr* = Cr mass / initial Cr mass
Br* = Br mass / initial Br mass
0 50 100 150 200 250
Days since injection
300
350
Figure 11. Results of the spatial moments analysis showing the normalized mass of Cr (relative to the injected Cr mass and the normalized Br
mass) as a function of time after the injection. The curve shown is a spline fitted to the data points.
CD
14
13
12
CO
CD ,,
(/) 11
c
CO
CD
E 10
CD
O
CD
3
£ 8
O
0<
L c*
-------
14
CD
+S 1-3
CD 1J
12
_CD
CO
CD ....
C/3 11
C
CO
CD
E 10
CD
O
03 9
CD
D
O
0
O
o
o
o
0.0 0.2 0.4 0.6 0.8
Attentuation factor
1.0
Figure 13. Attenuation factors for Cr calculated from the breakthrough-curve data, plotted as a function of the altitude of the sampling port.
Injected EDTA= 11 Moles
"EDTA" = Sum of 6 Metals
50
100 150 200 250
Days After Injection
300
350
Figure 14. Results of the spatial moments analysis showing the moles of Ni, Pb, Cu, Zn, Fe, and Al in the tracer cloud as a function of time
after the injection. The estimated moles of dissolved EDTA in the tracer cloud are shown from the sum of the moles of the six
metals at each synoptic sampling.
39
-------
g
1^3
2
(U
o
o
O
60.0
50.0
40.0
30.0
20.0
10.0
0.0
1 . 1 , 1 I I I , , , , 1 1 1 I
-
o <24 hours post injection -
'- o 0 0 0 /
x n F
1 0 00 X '1
o
o
I Oo
_ o
. OQ
0 o O ,_
o
-x o
- o o
0 D D° 00 0
D Q ..
: & n n c^
"" i9 ^ o n'""' nB
o ^ 5P x SI'S Q x
X n o D D^
• ,x ^ , ,a ^>^c^a^ ±ol?iite!i ft ! '
.0 5.5 6.0 6.5
PH
VI-EDTA I
:e-EDTA -
ree1 Zn _I
~_
~_
_
~
-
-
_
—
-
i i i
7.
Figure 15. Concentrations of dissolved Al- and Fe-EDTA complexes and uncomplexed (free) Zn as a function of the pH of groundwater
samples collected 24 hours after the injection was completed.
40
-------
3.5
< 3'°
Q 9 e
uj ^•••3
| Z°
Q 1.5 ;
UJ
d 1.0 ,
j
0.5
n n
. i i i i , i i i i i
E- o pristine
: x Zn contaminated
; D sewage-contaminated
1 1 1 • 1 1 1 1 1 .
\ : A :
; |
': '-
E i : :
I : : I
: , [ • x
t : x
= ,,,!,,,!,,
X '
O ; : -
, i , , , 1 , , , :
0.0 0.2 0.4 0.6
[Pb-EDTA]/[Ni-EDTA]
UJ
Q
UJ
0.8
0.0 0.2 0.4 0.6
[Pb-EDTA]/[Ni-EDTA]
0.8
1.0
3 < x x.x :
:x .. x^xx^
0; o- ;OCD-- ®^
, 1 , , , 1 I , , :
1.0
Q
UJ
^
Q
LJJ
d)
LL
-------
Ni mass
Cu mass
Pb mass
50
100 150 200 250
Days since injection
300
350
Figure 17. Results of the spatial moments analysis showing the normalized masses of Ni, Cu and Pb (relative to their injected masses) as a
function of time after the injection. The curves shown are splines fitted to the data points.
50
100
150 200
Days since injection
250
300
350
Figure 18. Results of the spatial moments analysis showing the normalized masses of Ni and Zn (relative to their injected masses) as a
function of time after the injection. The curves shown are splines fitted to the data points.
42
-------
co
03
"CD
E
CO
CD
CO
c
CO
Q)
E
Q)
O
.0
CO
Q)
•
16
14
12
10
8
watertable
175 Days
0
20
40
60
80
100
120
140
160
180
200
16
14
12
10
8
^
•E 6
237 Days
0 20 40 60 80 100 120
Distance downgradient from injection center, meters
140
160
180
200
Cu, Center of mass
O Ni, Center of mass
Figure 19. Longitudinal cross section through the portion of the tracer cloud containing metal-EDTA complexes at 175 (upper panel) and 237
days (lower panel) after injection; numbered contours show background Zn concentrations (|iM, before injection) as was shown in
Figure 4. White area shows the portion of the tracer cloud with Ni concentrations above the detection limit. Regions with vertical
lines show where normalized Ni concentrations exceeded 0.1. Black regions show where normalized Cu concentrations exceeded
0.1. Note that in both panels, regions with Ni C/C0 values > 0.1 enclose regions with Cu C/C0 values > 0.1. The circle and star
show the locations of the centers of mass of Ni and Cu, respectively, from the spatial moments analysis.
43
-------
100
Pristine, Fe
--SC, Fe
Pristine, Al
--A--SC, Al
Pristine, EDTA
x SC,EDTA
50
100
150 200
Time, hours
250
300
350
Figure 20. Concentrations of dissolved Fe, Al, and EDTA as a function of time in batch experiments with two samples of composite aquifer
sediments mixed with artificial groundwater solutions containing 100 nM EDTA. pH in pristine experiments was 5.4; pH in sewage-
contaminated (abbreviated as SC in legend) experiments was 6.5. The concentration of dissolved Al peaks at 4 hours under
pristine zone conditions and at 48 hours under sewage-contaminated zone conditions. Dissolution of Fe was still continuing after
336 hours (14 days).
PbEDTA
ZnEDTA
CuEDTA
• NiEDTA
At equilibrium
with ferrihydrite -
Figure 21. Calculated extents of metal exchange between metal-EDTA complexes and Fe(lll) dissolved from ferrihydrite as a function of pH.
The calculations assume chemical equilibrium and consider the adsorption of free metal ions (using the constants of Dzombak and
Morel, 1990) and the adsorption of metal-EDTA complexes (from fitting the retardation of these complexes in the pristine zone).
Calculations are shown for a suspension of ferrihydrite (with equal surface area per liter of water to that of Cape Cod sediments in
the aquifer) mixed with sewage-contaminated groundwater containing 250 nM of each metal-EDTA complex, which is close to the
conditions just after injection.
44
-------
100
Q
LU
c
CD
£
CD
Q_
250 LiM NiEDTA
100 LiM
80
60
CD
X
_CD
Q.
E
O
^ 40
....... 1 LiM
--- 0.1 LiM
At equilibrium
with ferrihydrite
20
PH
Figure 22. Calculated extents of Ni exchange between Ni-EDTA complexes and Fe(lll) dissolved from ferrihydrite as a function of pH and the
initial Ni-EDTA concentration. Other conditions in the calculations are the same as those in Figure 21.
100
.2 80
Q.
E
o
< 60
Q
LU
tfl
CO
+••
0)
o
40
20
Pristine Zone; pH 5.5
60% Tracer Injectate
40% Ambient Groundwater
Saturation with
Ferrihydrite
50
100 150
Dissolved Fe, LiM
200
Figure 23. Calculated extents of metal exchange between metal-EDTA complexes and Fe(lll) as a function of pH and dissolved Fe(lll)
concentration. The calculations assume chemical equilibrium (except for ferrihydrite dissolution) and consider the adsorption of
free metal ions and metal-EDTA complexes as in Figure 21. Calculations are shown for a solution containing the injectate solution
diluted 3:2 with pristine groundwater. Saturation of the solution with respect to ferrihydrite dissolution is achieved at a dissolved Fe
concentration near 200 |iM.
45
-------
CM
E
"55
_0)
o
TO
c
N
O)
o
0.5
0.0
-0.5
-1.0
-1.5
-2.0
-2.5
' -I •' ,j' 1 ' UTOt; -i^r-"''' '
,.-"* ^,,-**
--***' -'''I-''''
--'*'' o o m^--'' ---''
- --'' o -*i -- o o ,,-'
: ^ PH6.1^^- B^ ^.-- o pH4
,--'' --''' ,-''' • PH5
,''O,--*'' ,,-''' O PH6
--*' ,-*l5H 5-8 ,'''pH 5.0 + pH 7
"'•'''.-'''' .--'''
.»** , i -'',' 1,1,1,1
A
o--''\
.9-5.1 .
.7-5.9 -
.0-6.2 .
.1-7.4 _
-7.0
-4.5
-4.0
log ZiY
Figure 24. Laboratory batch experimental data for Zn adsorption onto a composite sample of aquifer sediments collected from the site and
surface complexation model fits, a) 1-site surface complexation model, b) 2-site surface complexation model. (From Davis et al.,
1998.)
N
100
80
Q.
O
O
^
Q
LU
"5
60
40
20
250 |iM ZnEDTA; Dz + M
250 |iM ZnEDTA
100 |iM ZnEDTA
10nM ZnEDTA
At equilibrium
with ferrihydrite
PH
Figure 25. Calculated extents of metal exchange between Zn-EDTA complexes and Fe(lll) dissolved from ferrihydrite as a function of pH and
Zn-EDTA concentration. The calculations assume chemical equilibrium and consider the adsorption of free Zn ions using the
constants derived from batch adsorption data with Cape Cod sediments. Adsorption of Zn-EDTA complexes was considered in the
same manner as in Figure 21. For comparison, the curve marked Dz + M shows the calculated extent of exchange at 250 |iM Zn-
EDTA as shown previously in Figure 21.
46
-------
50
100 150 200 250 300
Distance downgradient from disposal beds, meters
350
400
Figure 26. Simulated distribution of dissolved Zn after 60 years. Numbered contour lines show the pH gradient used in the simulation.
Shading shows the region where the Zn concentrations were greater than 1 |iM. The 0.2 |iM contour (analytical detection limit)
was within a few meters of that shown, however it was much more strongly affected by numerical oscillations, which are evident in
the 1 |iM contour near the leading edge. The abscissa shows the distance downgradient from the sewage effluent disposal beds.
The injection was located approximately 238 m downgradient from the disposal beds.
-0)
E
03
T3
O
1
O
-1
One-site SCM; Cape Cod aquifer
4.5
5.0
5.5
PH
4.8 x1Q-3M sites '
4.15kgsand/L
0.3 m2/g sand
6.0
6.5
x?
O
1
O
-1
B
4.5
5.0
Two-site SCM
4.1 x 10"5M strong sites
4.7x 10"3M weak sites
5.5
PH
6.0
6.5
Figure 27. Calculated values of the distribution coefficient, Kd, for adsorption of Zn as a function of pH and total Zn concentration per liter of
water in the aquifer using (A) the one-site model forZn adsorption, and (B) the two-site model (Davis et al., 1998). Solid matrix Zn
that is not available for desorption reactions is not included in the definition of total Zn for these calculations.
47
------- |