‘0. y E
:3: :II
. [<-

45 9.2m. 4M.
. 3.: . .

.1

.CnIL.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

u,

an???
I it

30.0.3953;

. Esau}!

a: l. .2 . 4 .
ﬁﬂa‘} #6,
K

L
I!

dun

. runs... a

nagmwmr Pumadﬁw

.. hilt): .IQII.
- I:

3....
A . ‘i

. 3130 ‘ .
5...?qu :w
. ‘. .
2.95.3... .u.
"N5:

 

0... .u’.\. '
. 1.; -u

THEM :~

This is to certify that the

dissertation entitled

Quantifying Redox Reactions in an Aquifer Contaminated
with Waste Fuel and Chlorinated Solvents

presented by

Jennifer Therese McGuire

has been accepted towards fulfillment
of the requirements for

Doctoral degree in Environmental Geosciences-
Environmental Toxicology

,ﬂaﬂ%ﬁ

M7aj0r pi‘ﬁér

 

 

Date 13 M/QC! G;

M5 U i: an Afﬁrmative Action/Equal Opportunity Institution 0.12771

 

UBRARY
Michigan State
University

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

I DATE DUE DATE DUE DATE DUE

 

lost 0 l 2006

LLLLL

 

“W“ i 8 2807

102609

 

 

 

 

 

 

 

 

 

 

 

 

6/01 cJCIRC/DaleOuopGS-p. I 5

 

QUANTIFYING REDOX REACTIONS IN AN
AQUIF ER CONTAMINATED WITH WASTE FUEL
AND CHLORINATED SOLVENTS

By

Jennifer Therese McGuire

A DISSERTATION

. Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Geological Sciences

2002

ABSTRACT
QUANTIFYING REDOX REACTIONS IN AN AQUIFER CONTAMINATED WTH
WASTE FUEL AND CHLORINATED SOLVENTS
By

Jennifer Therese McGuire

The evolution of redox conditions in an aquifer contaminated with waste fuels,
chlorinated solvents, and ﬁre ﬁghting foams was evaluated at the former Wurtsmith Air
Force Base in Oscoda, Michigan. Geochemical indicators of terminal electron accepting
processes (TEAPs) were evaluated over 5 years in conjunction with hydrogeologic and
microbiologic parameters to describe the spatial and temporal variability in redox
processes. Observed redox variability was linked to transport processes, particularly
recharge events, which deliver terminal electron acceptors (TEAS) to the reduced
groundwater plume. To understand how recharge affects groundwater biogeochemical
processes, multivariate statistics and graphical tools were used to analyze geochemical
data before and after a major recharge event (spring snow melt, 2000) at both small
(single well) and large (whole plume) scales. Results show that average TEA
concentrations increased within the aquifer and that changes in TEAPS were initiated not
in large redox zones but in smaller zones at the interfaces of plume boundaries.
Important plume boundaries include the interfaces between the reduced plume and 1)
surrounding groundwater, 2) the capillary fringe zone, and 3) recharge waters. Due to the
small and transient nature of these important zones, it is difﬁcult to directly measure

changes. Therefore, a series of ﬁeld experiments were designed to replicate as closely as

possible a recharge event in the aquifer. These experiments revealed that, when TEAs
are introduced simultaneously into an anaerobic aquifer where fermentative products
such as acetate and H2 gas have accumulated, TEAS could be consumed simultaneously
and at similar rates. However, consumption rates were more rapid than observed in
natural-gradient, large-scale experiments, likely due to a lack of readily available electron
donor (i.e., accumulation of fermentative products) in those systems.

Results ﬁom the aforementioned analyses and experiments led to the development
of a model for the evolution of redox zonation in contaminated aquifer systems. The
conceptual model proposes that during initial contaminant plume development, the rates
of TEAPs and therefore redox zonation are controlled by the production of readily
available electron donors (e. g., acetate, H2) and therefore by fermentation rates. As the
plume continues to evolve, TEAS are depleted sequentially based on thermodynamically
predicted free energies. Once a system has reached a state of maturity where organic
carbon inputs balance biodegradation reactions, a stable plume length is achieved. Under
these conditions, TEAS will continue to be depleted and fermentative products will
accumulate such that distinct redox zones down flow path cannot be sustained. Spatial
variability in TEAPS occurs in small interface zones near plume boundaries where TEAS
are introduced either by diffusion and dispersion from the sides of the plume or by
recharge from precipitation and upgradient ﬂow. TEAP rates are no longer limited by
fermentation rates but rather by TEA reduction rates. In addition, TEAS would not
necessarily be degraded sequentially given the excess of available donor. The conceptual
model was veriﬁed by incorporating its key elements and linkages in a 3-D ﬂow and

transport model, which successfully represented the early stages of plume evolution.

DEDICATION

I would like to dedicate this dissertation to Erik Wilson Smith without whose
continued support, love, and understanding this work could never have been possible.
You have been an unbelievable source of strength to me every day. Thank you for being
there through all the tears and all the excitement. There isn’t anyone else I would have

wanted to take this ride with. I love you.

iv

ACKNOWLEDGEMENTS

I would ﬁrst and foremost like to thank my advisor, mentor, and ﬁiend Dr. David
T. Long. Thank you for being there to guide me when needed, while giving me the
freedom to pursue my own ideas. I would also like to thank my other committee
members Dr. David W. Hyndman, Dr. Michael A. Velbel, Dr. Michael J. Klug, and Dr.
Sheridan Kidd Haack for all the advice, long chats about science and thoughtful reviews.
Dr. Mantha Phanikumar, without your help our numerical model could never have been
constructed. Thank you for sharing your engineering expertise. I have thoroughly
enjoyed working with all of you. This work also greatly beneﬁted from the lab assistance
rendered by, and countless research discussions with my lab mates, especially Joel Fett,
Sharon Simpson, Mary Sievert, Karen Wayland, Sarah Woodharns, Gary Icopini, and
Jonathan Kolak. Thanks guys.

My stay at MSU was greatly enhanced by the efforts of four lovely ladies, Loretta
Knutson, Cathy Caswell, Jackie Bennett, and Diane Baclawski. The department is truly
lucky to have you. Thank you Loretta for the care you put into juggling all the issues that
have come up over the years. Your smile and reassurance have meant a great deal to me.
Cathy, I will be forever be in your debt for all the last minute sampling trip orders I
slipped into your box. Thank you for always keeping things running as smoothly as
possible and getting me back on the road to collect more samples (hey. . .wait a minute!)

Jackie, thank you for keeping me somewhat sane and helping me ﬁnd the humor in

things. You’ve been a great sounding board and ﬁiend. Diane, you are a jewel. Thank
you for sharing your knowledge, time, and direction when I needed help.

A special thanks to my family for all their love and support especially during the
hard times. Mom and Dad, thank you for believing in me and always being there to listen
when I needed to talk things out. Grandma and Grandpa, thank you for your
unconditional love, endless toasted cheese sandwiches and support. Our visits always
provided me a refuge from stress and helped me to remember what’s important in life.
Jeanette, you are an amazing ﬁiend, sister, and scientist. Thank you for everything, I
hope we are always this close. I am looking forward to watching our science careers
grow together.

Finally I would like to thank the National Science Foundation for providing

ﬁmding for this project through grant EAR 9708487.

vi

TABLE OF CONTENTS

LIST OF TABLES .................................................................................. ix
LIST OF FIGURES .................................................................................. x
INTRODUCTION .................................................................................... 1
CHAPTER 1.
Temporal variations in parameters reﬂecting terminal-e1ectron-accepting processes in an
aquifer contaminated with waste fuels and chlorinated solvents .............................. 8
Reprint .................................................................................................. 9
CHAPTER II.
Geochemical characterization of the effects of recharge on an anaerobic contaminated
aquifer using multivariate statistics ............................................................... 24
Introduction ........................................................................................... 25
Material and Methods ............................................................................... 29
Study Site Description ..................................................................... 29
Chemical Analysis ......................................................................... 32
Statistical Analysis ......................................................................... 35
Results and Discussion .............................................................................. 37
Conclusions ........................................................................................... 56
References ............................................................................................. 59
CHAPTER III.
Evaluating the behavior of oxygen, nitrate, and sulfate during recharge and quantifying
reduction rates in a contaminated aquifer ........................................................ 63
Abstract ................................................................................................ 64
Introduction ........................................................................................... 64
Study Site Description .............................................................................. 70
Methods ................................................................................................ 72
Push-Pull Tests ............................................................................. 72
Surface Reaction Vessel Tests ............................................................ 74
Geochemical Analyses ..................................................................... 74
Rate Determination ........................................................................ 75
Results and Discussion .............................................................................. 77
Field vs. Laboratory N03 Measurements ............................................... 77
Push-Pull Tests ............................................................................. 79
Surface Reaction Vessel Experiments ................................................... 87
Acknowledgements ........................................................................ 90
Literature Cited ............................................................................. 91

vii

CHAPTER IV.
Evolution of redox zonation in a contaminated aquifer: Development of 3-D coupled

partial-equilibrium reactive ﬂow and transport model ........................................ 94
Introduction ......................................................................................... 95
Goals and Approach ............................................................................... 99
Study Site Description ............................................................................ 103
Methods ............................................................................................ 108
Flow Model ............................................................................... l 10
Reactive Transport Model .............................................................. 114
Geochemical Model ...................................................................... 1 15
Preliminary Phase I Modeling Results ......................................................... 117
Future Work ........................................................................................ 125
References .......................................................................................... 126
CONCLUSION .................................................................................... 135
APENDICES ....................................................................................... l 3 8
Appendix A: Procedure for Determination of Hydrogen Gas Concentrations Using the
RGA-3 ............................................................................. 139
Appendix B: Timeline of Project Events ........................................................ 144
Appendix C: Well Sampling History ............................................................ 146

viii

Table 2.1

Table 3.1

Table 3.2

Table 3.3

Table 3.4

Table 4.1

LIST OF TABLES
Rotated factor pattern of ﬁve extracted factors after varimax
rotation in January and June of 2000 .................................. 41
Initial Chemical Characterization of Selected Wells (mg/L) ....... 78
Summary of Rate Measurements for Sulfate Reduction ............ 84
Summary of Rate Measurements for Nitrate Reduction. . . . . . ..85

Summary of Available First-Order Field Rate Measurements
(day'l) ...................................................................... 85

Initial Concentrations Assigned in Flow Model (mg/L) ........... 113

ix

Figure 2.1

Figure 2.2
Figure 2.3

Figure 2.4

Figure 2.5A

Figure 2.5B

Figure 2.6

Figure 2.7

Figure 2.8

Figure 2.9

Figure 2.10

Figure 3.1

Figure 3.2

Figure 3.3

Figure 4.1

Figure 4.2

LIST OF FIGURES

Plume map for the F ire-Training Area 2 (PTA-2) at the former

Wurtsrrrith Air Force Base, Oscoda, Michigan ....................... 31
Two-dimensional well grid installed near ETA-02 ................... 33
Fluctuations in water table elevation (head) over time .............. 34

Percent change in average chemistry at the shallowest depth at all
wells between January and June 2000 (after recharge). . . . . . ....39

Cluster dendrograms for January 2000 showing ﬁve clusters. . ....45
Cluster dendrogram for June 2000 showing ﬁve clusters ........... 46

Average chemical descriptions and factor loadings of each cluster

in January 2000 ........................................................... 47
Average chemical descriptions and factor loadings of each cluster
in June 2000 ............................................................... 48
Three-dimensional spatial distribution of clusters based on factor
scores from January 2000 ............................................... 50
Three-dimensional spatial distribution of clusters based on factor
scores from June 2000 ................................................... 52
Vertical distribution of percent chemical change at a single well
(ML3) from January till June 2000 ..................................... 55
Plume map for the Fire-Training Area 2 (PTA-2) at the former
Wurtsmith Air Force Base, Michigan ................................. 71
Measured concentrations of added solutes versus time ............. 8O

Regression plot for rate coefﬁcient determination, obtained by
ﬁtting equation 3 to experimental data ................................ 81

Plume map for the Fire-Training Area 2 (ETA-2) at the former
Wurtsmith Air Force Base, Michigan ................................ 105

Two-dimensional well grid. The dashed arrows denote wells
through which chemical transects (T1 and T2) were
constructed ............................................................... 106

Figure 4.3

Figure4.4

Figure 4.5

Figure 4.6

Figure 4.7

Figure 4.8

Figure 4.9
Figure 4.10

Figure 4.11

Plot showing lack of redox zonation at WAF B as inferred by
dissolved hydrogen concentrations at two transects through the
plume roughly parallel to the direction of groundwater ﬂow in June
2000 ....................................................................... 107

Phase I conceptual model describing linkages of three existing
modeling codes. Functions will be performed at each cell with a

change in Corg at each point in time .................................... 109

Groundwater ﬂow model output showing location of natural
boundaries and approximate location of groundwater plume FTA-2
within this model framework .......................................... 111

Initial PHREEC input ﬁle; after the initial time step the total
organic (CHZO), sulfur (S), nitrogen (N), and iron (Fe) are
calculated by RT3D and included in the input ﬁle as such ........ 116

PHREEQC theoretical sample output assuming a constant input of
electron donors (Cm) ................................................... 118

Organic Matter (Corg) concentration (mol/L) at 10, 200, and 364

days ....................................................................... 120
Oxygen Concentration (mol/L) at 10, 200, and 364 days ......... 121
Nitrate Concentration (mol/L) at 10, 200, and 364 days .......... 123

Ammonium and Sulfate Concentrations (mol/L) at 10, 200, and
364 days .................................................................. 124

Some images in this dissertation are presented in color.

xi

INTRODUCTION

Understanding the means by which interconnected geochemical and
microbiological cycles combine with hydrogeological processes in evolving aqueous
systems is fundamental to identifying how environments change over time. In anaerobic
systems, one of the main controls on system evolution is the reduction-oxidation (redox)
potential of the system. Many elements in nature exist in more than one oxidation state.
The major elements (typically at concentrations greater than 1 mg/L) exhibiting redox
sensitivity include H, O, C, S, N, Fe, and Mn. Other minor (less abundant) elements
include U, Cr, As, Mo, V, Se, Sb, W, Cu, Au, Ag, and Hg. Although these minor
elements occur in lower concentrations, many can have serious effects on human health
and the environment. Considering conventions developed from traditional physical
chemistry, redox can be thought of as the tendency of particular chemicals to donate or
accept electrons. This tendency can be described in terms of the standard hydrogen
electrode (Stumm and Morgan, 1981). This convention serves to deﬁne a
thermodynamic scale to which free energies of dissolved species can be compared.
Based on this scale, at a given pH and temperature, redox reactions can be ranked into
redox reaction sequences based on free energies describing thermodynamic favorability.
In nature, microorganisms become involved in electron-transfer reactions to maintain life
functions, grow, and reproduce. These activities are often kinetically controlled and
therefore represent another important consideration besides thermodynamic free energy

when attempting to study the redox state of a system. In addition to considering

thermodynamic and kinetic biogeochemical considerations, systems may be limited by
the transport of chemical constituents necessary for systems to evolve. In groundwater
systems, the rate of supply of electron acceptors, growth substrates for microorganisms,
and nutrients are controlled by both the hydraulic gradient and aquifer properties such as
hydraulic conductivity, dispersivity, effective porosity, and sorption coefﬁcients.
Therefore, understanding the redox state of aqueous systems requires knowledge of the
interconnectivity of hydrogeological, microbiological, and geochemical processes
making it an ideal concept to study how these interconnected processes allow systems to
evolve.

The characterization of redox potential is of both scientiﬁc and practical interest.
Redox state controls the form, mobility, toxicity and persistence of many chemicals in the
environment. Without an awareness of how redox state evolves and the conditions under
which it may change, we cannot adequately evaluate health and safety concerns such as
chemical routes of exposure, natural attenuation capabilities, and the means by which we
can manage redox sensitive systems. Understanding redox evolution is particularly
important to the ﬁelds of bioremediation and natural attenuation. These ﬁelds of study
seek to ﬁnd efﬁcient ways to remediate contaminated systems. Much effort has been
expended to try to understand and implement these important remediation strategies,
which in many instances present a more effective and cost efﬁcient means of dealing with
hazardous contaminated environments. These strategies make use of redox state by
coupling organic contaminants, which serve as electron donors, to available terminal
electron acceptors (e.g., 02, N03, Fe“, and $04). It is through these microbially

mediated terminal electron accepting processes (TEAPs) that organic contaminants are

degraded. In contaminated systems, TEAPs have been observed to distribute themselves
into redox zones down ﬂow path based roughly on free energy such that the most reduced
conditions occur nearest the contaminant source (e. g., methanogenesis) and more
favorable TEAPs (e.g., N03 reduction) occur farther down ﬂow path. Unfortunately,
understanding the redox state of a system is complicated because redox is very difﬁcult to
assess and it is also both spatially and temporally dynamic.

Redox conditions and TEAPS have traditionally been estimated through
measurement of Eh or by redox couples which compare the relative concentrations of
reduced to oxidized species (e.g., Fe 2‘7 F e3 +) Such approaches are thought to be
unreliable because of problems such as redox disequilibrium and lack of speciﬁcity of Eh
measurements. The H2 concentration in aquatic systems has been proposed as a more
accurate indicator of in-situ TEAPs than the above approaches. Dissolved hydrogen is
known to be a ﬂeeting but important intermediate in the decomposition of organic matter.
Dissolved hydrogen is produced and consumed in anaerobic environments such that each
TEAP maintains a characteristic H2 concentration. However, other in-situ reactions such
as homoacetogenesis (production of acetate from hydrogen and carbon dioxide) and
dehalogenation reactions can affect concentrations of H2 in contaminated systems making
its use as a TEAP indicator more complicated to apply to real-world environments. In
addition to these complexities, even if a speciﬁc TEAP can be identiﬁed this state is not
static but rather highly dynamic both in Space and time.

An ideal study site to examine redox variability was found at the former
Wurtsmith Air Force Base (Fire Training Area 2 — FTA-OZ), where a shallow sandy

aquifer is contaminated with common petroleum hydrocarbons (BTEX compounds),

chlorinated solvents, surfactants, and ﬁre-ﬁghting foams. Given the complicated nature
of the contaminants, hypotheses regarding the possibility of additional reactions
interfering with redox indicating parameters (H2) could be tested. Hydrogeologically,
there is relatively little heterogeneity for a natural system, making transport arguments
more reliable. Geochemically and microbiologically, a database of relevant parameters at
three points in time (October 1995, June 1996, and October 1996) already existed. These
preliminary data suggested that the site was experiencing redox variability both spatially
and temporally.

The ﬁrst step was to document the variability of redox processes at PTA-02 and
attempt to link those changes to variations in hydrogeologic, microbiologic and
geochemical processes, and thereby identify the most important drivers of change in
redox state. Chapter 1 of this dissertation documents the spatial and temporal variability
of TEAPs as interpreted from combinations of the above mentioned redox determination
methods. Results suggest that TEAP processes are variable on relatively short time
scales (months) and that the observed variability was related at least in part to recharge
events delivering new terminal electron acceptors (TEAS) to the system. Results also
identify the importance of considering how changes in transport processes affect the
chemistry of the aquifer. The groundwater plume at F TA-02 has experienced relatively
cyclical changes in groundwater ﬂow direction of up to 10 degrees related to recharge
events. These shiﬁs can introduce new TEAS to previously reduced portions of the
aquifer and also have implications for evaluating changes at a static sampling point over

time. These ﬁndings suggest that, although a complicated relationship exists between

recharge events and the resulting biogeochemical processes operating in the aquifer,
recharge inputs of solutes are a major driver of change in TEAPS.

Examination of water table elevation proﬁles over time revealed that spring
snowmelt recharging the aquifer had a major impact on aquifer processes. To explore the
role of recharge, statistical and graphical tools were used to examine the groundwater
chemistry before and after a major snowmelt event in 2000. The results of these analyses
are presented in Chapter 2. Geochemical changes were evaluated at small (single well)
and large (plume) scales. Large-scale chemical averages indicated that the inﬂux of
recharge water introduced more energetically favorable TEAS to the plume. Based on
these inputs, one would expect to observe a shift in dominant anaerobic groundwater
microbial processes to more energetically favorable reactions and thus observe changes
in redox zonation. However, further data examination by factor analysis, cluster analysis,
and graphical techniques revealed that when viewed in greater detail, the changes in
TEAPS initiated by recharge may be preferentially occurring on a smaller scale.
Evaluation of chemical changes at a Single well provided further evidence that many
TEAP reactions initiated by recharge events occur in small interface zones and not in
large redox zones as we traditionally envision. Since the zone over which these reactions
occur is likely small it is very difﬁcult to directly measure these processes. Nonetheless,
recharge effects are likely most prevalent at plume boundaries, especially in the organic-
rich capillary fringe zone. This is important because initiating more favorable TEAPS in
zones with high contaminant concentrations may signiﬁcantly increase the rates of

natural attenuation and bioremediation.

The next step was to evaluate the importance of small-scale changes in TEAPS
initiated by recharge to the evolution and dynamics of redox conditions by quantifying
the observations described above with a numerical model. However, construction of a
meaningful model to test hypotheses regarding redox evolution was limited by our
understanding of speciﬁcally how TEAP reactions were initiated by recharge and the
rates at which these reactions would proceed. To address these issues a series of in-situ
ﬁeld experiments were conducted. The results of these experiments are presented in
Chapter 3. These experiments were speciﬁcally designed to evaluate reactions initiated
when a recharge solution containing various combinations of TEAS (02, N03, and 804)
comes in contact with a highly reduced aquifer. The experiments evaluate which
reactions will be initiated, the order in which acceptors will be consumed if introduced
simultaneously, and the rates of TEA consumption. In a tandem ﬁeld experiment, the
roles of the aqueous and solid phase portions of the aquifer and their associated microbial
communities on these reactions were evaluated. Results suggested that TEAPS initiated
during recharge events would not occur in the aqueous phase without the inﬂuence of the
sediments and their associated microbial communities. It was also observed that if
introduced simultaneously, TEAS could be consumed simultaneously and not according
to the traditional thermodynamic sequence based on free energy. These ﬁndings, in
conjunction with the relatively rapid rates obtained during the experiments, led to the
development of a conceptual model regarding the evolution of redox conditions in
contaminated aquifers.

This conceptual model suggests that initially, when a contaminant is introduced to

an aquifer system and a plume begins to develop, the rates of TEAPS and thus redox

zonation are controlled by donor availability and consequently by the rate of
fermentation. Ferrnentative bacteria convert large organic molecules into more readily
available electron donors such as acetate and dissolved hydrogen, to be coupled to the
traditional sequence of electron acceptors. AS the organic plume moves through the
system, TEAS are depleted sequentially according to their thermodynamically predicted
free energies. Once a system has reached a state of maturity such that organic carbon
inputs balance biodegradation reactions, a stable plume length is achieved. Within this
framework, TEAS are depleted and fermentative products such as acetate and H2 will
accumulate. Under these conditions, no mechanism exists to maintain stable, distinct
redox zones down ﬂow path within the plume. Rather, Spatial variability in TEAPS will
likely occur in small interface zones near plume boundaries where inputs of TEAS can
occur either by diffusion and dispersion from the Sides or recharge from precipitation and
upgradient ﬂow. At this point, the system has evolved to a state where it is no longer
donor (fermentation) limited in terms of rates but is instead acceptor limited in terms of
TEAP reaction rates. Under this model, TEAS are consumed rapidly and would have no
reason to be degraded sequentially given the excess of readily available donor, which is
exactly what we observed at PTA-02 during our ﬁeld experiments. To evaluate the
feasibility of this conceptual model and test hypotheses regarding redox evolution, a
three-dimensional partial equilibrium ﬂow and transport model was constructed.
Preliminary results from this modeling effort, as well as a description of the work to be

done in the near future using this mode,l is presented in Chapter 4.

CHAPTER I.

McGuire, J. T., Smith, E. W., Long, D. T., Hyndman, D. W., Haack, S. K., Klug, M. J ., and
Velbel, M. A., 2000. Temporal variations in parameters reﬂecting tenninal-elecn‘on-accepting
processes in an aquifer contaminated with waste ﬁre] and chlorinated solvents, in K. Konhauser
and G. Southam (eds), Geomicrobiology - Technical Session 06, 110th Geological Society of
America Annual Meeting. Chemical Geology, 169:471-485.

 

 

CHEMICAL
GEOLOGY

INC L1 '.DIVC
[SOTOPE GE OS C [ENC E

 

Chemical Geology 169 (2000) 471 —485

 

www.clsevier.com / locate / chemgeo

Temporal variations in parameters reﬂecting
tenrrinal-electron-accepting processes in an aquifer contaminated
with waste fuel and chlorinated solvents

Jennifer T. McGuire 3“ ’ , Erik W. Smith a, David T. Long a, David W. Hyndman a,
Sheridan K. Haack b, Michael J. King C, Michael A. Velbel a

a Department of Geological Science. Michigan State University. East Lansing. MI 48824. USA
° us. Geological Survey. East Lansing. MI 49811 USA
‘ Kellogg Biological Station. Michigan State University. Hickory Corners. MI 49060. USA

Received 1 July 1999: accepted 9 September 1999

 

Abstract

A fundamental issue in aquifer biogeochemistry is the means by which solute transport, geochemical processes. and
microbiological activity combine to produce Spatial and temporal variations in redox zonation. In this paper. we describe the
temporal variability of TEAP conditions in shallow groundwater contaminated with both waste fuel and chlorinated solvents.
TEAP parameters (including methane, dissolved iron. and dissolved hydrogen) were measured to characterize the
contaminant plume over a 3-year period. We observed that concentrations of TEAP parameters changed on different time
scales and appear to be related. in part. to recharge events. Changes in all TEAP parameters were observed on short time
scales (months). and over a longer 3-year period.

The results indicate that (l) interpretations of TEAP conditions in aquifers contaminated with a variety of organic
chemicals. such as those with petroleum hydrocarbons and chlorinated solvents. must consider additional hydrogen-consum-
ing reactions (e.g., dehalogenation); (2) interpretations must consider the roles of both in situ (at the sampling point)
biogeochemical and solute transport processes; and (3) determinations of microbial communities are often necessary to
conﬁrm the interpretations made from geochemical and hydrogeological measurements on these processes. © 2000 Elsevier
Science B.V. All rights reserved.

Keywords: Aquifer biogeochemisuy; Tenninal-electron—accepting process: Temporal variation: Redox zonation: Recharge

 

1. Introduction succession of alternate terminal electron acceptors to
support their growth using a variety of carbon sources
(Stumm and Morgan, 1981; Lovely. 1991; Lovely
and Goodwin, 1988: Chapelle, 1993). The succes-
sion of terminal electron accepting processes
(TEAPS) in order of decreasing redox potential and

' Corresponding author. free energy yield is generally: oxygen reduction.

The development of redox zones in natural aquatic
and groundwater systems is based on the activities of
microorganisms. which ﬁrst use 02. and then a

0009-2541/00/5 - see front matter © 2000 Elsevier Science B.V. All rights reserved.
Pll: $0009~2541(00)00223-0

472 J.T. McGuire et a1. / Chemical Geology 169 (2000) 471—485

nitrate reduction. manganese reduction, iron reduc-
tion, sulfate reduction. and methanogenesis (Table
1). In most closed groundwater systems. the se-
quence of TEAPS follows this succession (i.e., in-
creasingly reduced conditions occur further down
the ﬂow path) (Champ et al.. 1979). In groundwa-
ter systems contaminated with hydrocarbons. redox
zonation has been identiﬁed (Baedecker and Back.
1979; Barcelona et al.. 1989: Bennett et al.. 1993;
Vroblesky and Chapelle. 1994; Chapelle et al.. 1995).
In contaminant plumes. the sequence of TEAPS is
reversed (Lovely. 1991; Chapelle, 1993). That is. the
most reduced conditions (e.g., methanogenesis) oc-
cur near the contaminant source due to the greater
availability of electron donors. while less reducing
conditions (e.g., iron reduction) occur down flow
path from the contaminant source.

TEAP conditions and redox zonation affect the
chemical form. mobility. and persistence of many
groundwater contaminants (e.g., Stumm and Morgan,
1981; Christensen et al.. 1994: Davis et al.. 1994).
Natural biodegradation of many petroleum-derived
aromatic hydrocarbons occurs under most redox con-
ditions (Chapelle. 1993; Vogel, 1994). However. the
efﬁciency and rate of hydrocarbon and chlorinated
solvent degradation varies with redox environment
(Norris. 1994). Microbial degradation of petroleum

hydrocarbons is most efﬁcient under aerobic condi-
tions. but has also been demonstrated to occur under
methanogenic, sulfate-reducing. ferric-iron-reducing.
and nitrate-reducing conditions (Grbic-Galic and V0-
gel. 1987; Hutchins et al.. 1991; Beller et al.. 1992;
Edwards et al.. 1992; Baedecker et al.. 1993; Rabus
et al.. 1993; Lovely. 1994; Lovely and Lonergan.
1990; Lovely et al.. 1995). Redox zonation also
affects the nature and distribution of chlorinated aro-
matic and aliphatic compounds. Chlorinated ethenes
are common in groundwater and some degradation
occurs under a variety of ﬁeld and laboratory condi‘
tions (Norris. 1994).

Accurate modeling of biodegradation processes
and redox zonation in groundwater requires under-
standing the scales of temporal changes in TEAP
conditions (e.g., Essaid et al.. 1995; Hunter et al..
1998; Keating and Baht. 1998; Salvage and Yeh.
1998). Over long time scales (years). changes in
redox zonation in groundwater have been attributed
to successional changes in TEAP conditions (Bemer.
1981). For example, Baedecker et a1. (1993)
attributed the observed 5-year change in redox zona—
tion in oil-contaminated groundwater to a succes-
sional change in TEAP conditions from iron reduc-
tion to methanogenesis. Less is known about changes
in TEAP conditions on shorter (months to year) time

 

 

Table 1

Reactions and free energies discussed in this paper

Reaction Equation AG (kJ / mole)
Denitriﬁcation 2N0; + 5H: + 21-1’ _. N2 + 6H,o ~1120°
Nitrate reduction to ammonia NO; + 4H2 -+- ZH’ -0 NH] 31-120 ~600“
Manganese reduction MnO: + H2 - l\v‘ln(Ol-l)2 - 163’
Iron reduction 2Fe(OH)3 + H: -' ZFdOH): + 2H30 - 50’
Sulfate reduction 503+;er2 +H‘-’HS‘+4H30 —152~‘
CO: methanogenesis HCO; + 4H: + H’ -' CH4 + 31130 - 136‘I
Aceticlastic methanogenesis CH 3COOH -9 CH 4 + CO2 - 53h
Dechlorinatlon P‘CEl + H: -* 'I'CE2 + H’ + Cl— - 173C
Dechlorination TCE2 + 1-12 —¢ DCE" + H’ + Cl" -170‘
Dechlorination DCE" + l—l2 -9 VC" + H' 4- Cl‘ —140‘
Dechlon'nation VC‘ + H: -* ethylene + H’ Cl‘ — 150:

 

aLovley and Goodwin (1988).
bCRC (1993).

cDolfing and Beurskens (1995).
lTetrachloroethylene.

2 Trichloroethylene.

3Cis- 1.2-dichloroethylene.

’ Vinyl chloride.

10

J. T. McGuire et al. / Chemical Geology 169 (2000) 471—485

scales. but these are thought to be related to solute
transport including recharge events (e.g., LeBlanc et
al.. 1991; Hess et al.. 1992). Vroblesky and Chapelle
(1994) studied temporal variations in TEAPS at a
shallow petroleum hydrocarbon-contaminated aqui—
fer. They observed TEAP changes in individual wells
over periods as short as 2 months. Although these
observations show short-term changes in TEAP vari-
ables, they may not necessarily indicate short-term
changes in the TEAP conditions of the plume be-
cause they do not account for the inﬂuence of changes
in water table elevation or groundwater ﬂow direc-
tion (McGuire et al.. 1999). They did, however.
assess areal variations in TEAP conditions in the
plume. and reported changes over 15 months. They
attributed the temporal changes in TEAP conditions
to variations in recharge and lateral transport. which
inﬂuenced the availability of terminal electron accep—
tors such as S0,.

Solute transport and. therefore. the rate of supply
, of electron acceptors, growth substrates. and nutri-
ents is controlled by both the hydraulic gradient and
aquifer properties such as hydraulic conductivity.
dispersivity. effective porosity. and sorption coefﬁ—
cients. Spatial variations in hydraulic conductivity
determine the predominant zones of solute transport
since groundwater follows the path of least resis-
tance. Dispersion causes lateral and longitudinal so-
lute spreading due to small-scale variability in
groundwater velocity. which is a critical factor in
accurately describing the geometry of the redox
zones. Also, seasonal ﬂuctuations in the direction of
groundwater ﬂow may increase transverse dispersion
in the study area, and must also be accounted for
(Goode and Konikow, 1990).

The purpose of this paper is to describe the
temporalvariability of TEAP conditions in a Shallow
groundwater system contaminated with both waste
fuel and chlorinated solvents. Few Studies have ex-
amined TEAP conditions or changes in TEAP condi-
tions in such complex contaminant systems. We
examine changes in dissolved hydrogen (H3) con-
centrations and other selected TEAP variables (e.g..
dissolved methane (CH 4) and dissolved iron (II)
(Fe3+)) and non-TEAP variables (e.g., potassium
(K‘)). Redox conditions and TEAPS have tradition-
ally been estimated through measurement of Eh or
by the relative concentrations of reduced to oxidized

(

11

473

species (e.g., Fey/Fe“). Such approaches are
thought to be unreliable because of problems such as
redox disequilibrium (Keating and Bahr. 1998) and
lack of speciﬁcity of Eh measurements (Lovely and
Goodwin. 1998). The H2 concentration in aquatic
systems has been proposed as a more accurate indi-
cator of in situ (at the sampling point) TEAPS than
the above approaches (Lovely and Goodwin. 1988).
Dissolved hydrogen is known to be a ﬂeeting but
important intermediate in the decomposition of or-
ganic matter. Dissolved hydrogen is produced and
consumed in anaerobic environments such that each
TEAP maintains a characteristic H2 concentration
(Lovely and Phillips. 1987).

Interpretations of TEAPS from H2 concentrations,
as measured in marine and freshwater sediments and
in aquifers. are as follows: methanogenesis, greater
than 5 nM: SO4 reduction. 1—4 nM; Fe3+ reduction.
0.2-0.6 nM; and Mn (IV) or nitrate reduction, less
than 0.05—0.l nM (Lovely and Goodwin, 1988).
Chapelle et a1. (1996) demonstrated the use of H2
concentrations in delineating redox zonation in the
plume at the study site described in this paper. They
concluded that delineating redox zones based on H2
concentrations is more reliable when H2 concentra-
tion are interpreted in the context of other TEAP
variables such as Fe2+ and CH 4. a strategy we have
adopted here.

We suggest that the presence or absence of TEAP
variables in water samples from a well may not
indicate microbiological reactions that are occurring
in situ. Concentrations of TEAP variables may be the
result of their transport to the well from an upgradi-
ent source since CH 4 and Fe2+ are mobile in aque-
ous environments. Our samples were collected over a
3-year period and include quarterly measurements.
Thus, both short-term and long-term changes were
studied including inﬂuences of solute transport.

2. Description of study area

The study site is a Shallow sandy aquifer contami—
nated with petroleum hydrocarbons and chlorinated
solvents at the former Wurtsmith Air Force Base
(AFB). Michigan. USA (Fig. 1). Wurtsmith AFB lies
on a sandy plain of glacial lake sediment approxi-

474 ‘ J.T. McGuire et al. / Chemical Geology 169 (2000) 471-485

 

 

 

 

— coo — WATER-TABLE COMM-Show:
«lav-tion ofw-ter table. Contour
interval ”Eat. Duum is sea level

—- -- GROUND-M11512 plvme \\ m
't/ 3 5:. ,1
—. mourn-mm FLOW C %
hdmtcn dined ion ofﬂow l .
PTA-2 ) .3
A nas-rmmm AREA 2 ;_____,J
MIGUGAN

Fig. 1. Map showing the location of the study site.

mately 2 km west of the Lake Huron shoreline in
Michigan’s Lower Peninsula. Geologic units at the
Site consist of Mississippian sandstones, Siltstones
and shales, overlain by unconsolidated glacial sedi-
ments and surﬁcial deposits. Glacial deposits range
in thickness from 30.5 to 76.2 m and consist of
gravel, sand, silt, and clay. Surﬁcial deposits include
lacustrine sediments, ice-contact sediments. and allu-
vium. The principal water-bearing unit at the site is a
surﬁcial sand and gravel aquifer approximately 20 m
thick. The unconﬁned aquifer is underlain by at least
30.5 m of silty clay. The water table ranges from 3.5
to 5.0 m below land surface (184-186 In above sea
level) and ﬂuctuates 0.3 to 0.7 m annually. The
aquifer at the study site is comprised of highly
permeable alternating aeolian sands and glacial out-
wash material. Hydraulic conductivities of these de-
posits are on the order of 30 m / day. The water table
gradient ranges from 3 to 5 m/km. Assuming an

effective porosity of 30%, average groundwater ve-
locity is approximately 0.5 m/day.

The study site Fire Training Area-2 (PTA-2; Fig.
2) at Wurtsmith AFB was used for ﬁre training
exercises from 1952 to 1986. During these exercises,
waste fuel and solvents were ignited in the proximity
of a mock-aircraft located at the “pad” (Fig. 2).
Various compounds were used to extinguish the
ﬁres, including aqueous ﬁlm-forming foam. 3 multi~
purpose dry chemical, potassium bicarbonate based
soda. and Halon 1211. The majority of contaminants
occur within the capillary fringe and have caused an
extensive plume approximately 50 m wide and 400
m long (Fig. 2). The plume is complex in that it
contains both BTEX compounds (benzene, toluene,
ethylbenzene, and xylenes) and chlorinated solvents
(e.g., dichloroethane). Concentrations of BTEX com-
pounds in the plume range from 20 to 1000 ug /1
while concentrations of dichloroethane, chloroethane.
and vinyl chloride range from 2 to 1000 ug/l. Most
of the contamination is associated with aquifer solids
between 4.5 and 5.7 m below land surface. which
have an average concentration of total hydrocarbons

 

 

 

 

 

O IN FEET

EXPMNATDN

AREA OF mWTlON
ROAD
_ 606 .. WATER TABLE CCNTOUR-Sbws
elevation of water table in feet. Datum
is sea level.
0 MULTl-LEVEL WELL
A WATER LEVEL RECORDER

 

Fig. 2. Plume map for the Fire-Training Area 2 (PTA-2) at the
former Wurtsmith Air Force Base. Michigan. The concrete pad
was the site of ﬁre-training activities.

12

J. T. McGuire et al. / Chemical Geology 169 (2000) 471-485

of 13,650 mg/kg (U.S.G.S.. 1991). Because the
direction of groundwater ﬂow is inﬂuenced by
recharge, lateral shifts in the plume location at the
well array (see below) of up to 10 111 have been
observed (Smith et al., 1999).

3. Methods

Seasonal measurements of water levels and bio-
geochemical parameters are being collected at the
site. adding to a dataset started in 1995 by Chapelle
et al. (1996). To document the vertical distribution of
contaminants within the plume, researchers at the
National Center for Integrated Bioremediation Re-
search and Development (University of Michigan)
installed a series of multilevel sampling wells at this
site. The wells were constructed of 2.5-cm-inner-di-
ameter PVC casing. with 0.33-m PVC screens at the
bottom. The vertical spacing between the screens
varied from 0.5 to 2 m. The multilevel wells are
oriented in an array perpendicular to groundwater
ﬂow located approximately 100 m downgradient from
the suspected source. This array of four wells was
used to collect groundwater chemistry samples pre-
sented in this paper.

Each level in a well was purged with a peristaltic
pump at a rate of approximately 400 ml / min. A
ﬂow cell was used to continuously monitor tempera-
ture, pH, dissolved oxygen, Eh. and speciﬁc conduc-
tance until readings became stable. H2 samples were
taken using the bubble-stripping method (Chapelle
and McMahon. 1991). Water was continuously
pumped through a standard 400-ml-glass gas-sam-
pling bulb with an introduced nitrogen bubble. After
time. the H2 concentration in the water equilibrates
with its concentration in the gas bubble. A gas
sample was taken from the gas bubble in the bulb
with a gas-tight glass syringe at approximately 10-
min intervals to monitor the approach to steady state
H2 concentrations. Enough gas sample was taken
with each syringe for three replicates of 7.5 ml and
analyzed in the ﬁeld with a RGA3 reduction gas
analyzer (Trace Analytical). Concentrations of H2 in
the aqueous phase were calculated using Henry's
Law. The detection limit of this method is approxi-
mately 0.08 nM. Although stable readings of ﬂow
cell measurements were obtained within 20 to 30

13

475

min. H2 concentrations often required an additional
20 to 90 min to equilibrate.

In the ﬁeld. samples were collected for a variety
of geochemical measurements. Selected measure-
ments discussed in this paper include dissolved sul-
ﬁde by colorimetric spectroscopy using Chemetrics
Company Vacu-Vials’” . Fe“ by phenanthroline col-
orimetric analysis. major cations by ﬂame atomic
adsorption spectrometry, SO4 by capillary elec-
tr0phoresis. and CH 4 by ﬂame ionization detection
gas chromatography. When applicable, samples were
ﬁltered through 0.45 um polyethersulfone membrane
ﬁlters. Field blanks were collected for each sampling
trip. Samples for major cations were preserved with
ultra-pure “optima“m nitric acid. SO4 samples with
formaldehyde. and CH 2 samples with mercuric chlo-
ride crystals. Sulﬁde and Fe2+ were measured in the
ﬁeld immediately after sample collection. Water level
measurements were collected from three hourly
recording pressure transducers triangulated approxi-
mately 300 m apart about FT A-2 (Fig. 2). Individual
water level measurements were taken manually from
all available wells at ETA-2 (about 20 wells) during
sampling events.

Plots of the spatial distribution of solutes were
created using a krigged grid method with a linear
variogram model in Surfer (Golden Software). Surfer
plots of H2 were constructed with color divisions
indicating TEAP process. In addition to the TEAP
variables, spatial and temporal concentrations of K+
are presented to help deﬁne the loci of the plume
since one component of the contaminant plume is
potassium bearing ﬁre-ﬁghting foam. The contours
of the spatial plots of Fe“, CH4, and K“ reﬂect
concentrations and do not imply an interpretation of
process, with intervals chosen to best Show trends in
the well array. Spatial plots of other TEAP variables.
such as SO4 or sulﬁde, are not presented because
their concentrations were too low to generate inter-
pretable diagrams.

4. Results

Over the 3-year sampling period. H2 concentra-
tions were highest in October 1995 and generally
decreased through August 1998. Concentrations then
began to increase in November 1998 (Fig. 3). In

476 J.T. McGuire er al. / Chemical Geology 169 (2000) 471-485

ML4 ML3 MLS ML8

-. . ‘1 m . ‘ ”- ~ ' l ‘ 0 _ _ . .
r'r'rsrw T.-';.~~.:v“"f1‘.j‘r~.*‘ ‘ \v‘\ i
r . " ‘ \\.
\-. . i .- . ’ ' - K \' ‘
. . . . ' ‘ . ‘ ‘- ‘ .y\‘- I 1L
. . . u ,3

y E

l
f
i ‘ . -_ ' >5w.) .‘ 8 a é')§
l
l

 

.- . . _. .
5 . . ,sw ', . .— , _
-1 ' e ‘ A/ i

.. . . I” ‘.

_..~ , I

8 meters
¢~
l
s‘

”Riel

Lam . /

r.- .'-'

.' I a e
r. .- .o.'\v-v I ' '
.l (“A _ i 4

 

< >

5 3 meters

October 1995

 

 

 

l
“r" .‘T‘ ._ ' r
. - .. In I“ l
. t .
I s . ,0, ‘ I
' .\ \. ._, I ’9'" R‘;”.r ‘ ~ \‘ o L
I ~ ‘ ' ~— . ‘ .Q \-.
‘ ° l, i 2,; . , .. -.‘. I
. l r . r . ‘1’
-/ r l ‘ .. - ‘ ).- ‘
. / ’ '
.‘ l r - . .
1‘ 1* l/ 5 a} , . ' L
“" la ’ . \ _..
’ . ‘-._ﬂ :
‘ I
\ I] ‘ 1’ l
4
x a r L
r 1 TY r T r

 

 

 

 

v' _‘_‘ 7 ' ‘ [A ‘7 AL

I ' ‘ I
/ . _ . r“ , ,
. - l 1:9A_ L
» .- / :-_‘-v - I

£39“ “ \, . - -

.. ‘ , \ . 8 .
i r . \ - o i‘
_,., 5 x i

. l ‘ .
A / l \ [a
* ‘F Y T—

 

 

 

 

 

) r t \. : V ' 0 I
t ‘V _‘ I
4 . L r .. .. ‘ we ..,, y -. : ‘ .‘_'.: 197-1
‘3’. . :‘ ' “c r
' . ) ‘ \ K ‘ . .--"‘~ “\A “.53 l
a“. ./ \ _ . y A -‘ m l
'. '- | ,r‘ r l ‘ \ " .. 5' r
l , . 1 ' l x
._ A., .- ,J / -\ l
I. ' . n ,1. " . ' \ l
v \ —_//\\ . ‘ . . , ‘ j ‘/ _L .\~L\-
. Y r Y . V .

 

 

August 1998 November 1998

H2 concentration (nM)

Inferred Me hanogenesis Sulfate. Fe
TEAP Reduction Reduction

Fig. 3. Spatial distribution of H3 concentrations and inferred terminal electron-accepting process (TEAP) at the ETA-2 well array from
October 1995 to November 1998. Vertical axis is depth below water table and horizontal axis is lateral distance across the plume. Upper
surface of each image reflects water table posuion. Groundwater ﬂow is out of the page.

 

w..- O

   

l4

J.T. McGuire et al. / Chemical Geology I 69 (2000) 471 -485 477

 

ML4 ML3 ML5 ML8 ML4 ML3 MLS ML8
f’I—f\ '. 7." "mu” . 1 ~

«.‘

 

‘ .a.~ I»? _...-.-~

3 o W o . ; ' y‘ ?r C : 4.. 2. .‘.~“~-- ~N “ : ”(g g- ,' ": r} 4"
. Sm c ‘a 1". - 1A7 , v.1 ‘ . I . E
/% . \m'q l,// % . . "

53 meters

October 1995

8 meters

 

June 1996

 

 

 

mw_':f?'. “"“ —"'l “-‘T

1 ~13". .» , ___,’*V ‘ i.-. ‘L .
.:x.~-~-v-.‘. ;‘_-_‘___,_ 9-.“ 4.. -...~--~ . w . . ' . \~, _~"' ., a.“
“' - ‘ a. - ' ‘ '2 O i 7
V ""” ‘ Sq; ‘ h 'r . , ...-- *
//L \o “T“ “ .. 93° ‘ - ; ~l ”:69". ﬂ ’ _ r
Q r l 7'“ -‘~~- 5M” “~93" -
\\ l 1 “\ﬂ /\ A

 

 

 

 

October 1996 November 1997

.. . .- '-"‘-- i- g l i." .
' \F i I ' i 2 “ "‘ ~ ' ' i - 9
' I . \ ‘ “ Q.
‘_swr I ’ .‘ .\‘ ':'a /
3 Mk _ . . I . \ . . . /
. . . ‘ r \ ' ,./-——'/
L ‘~~ ‘ _ .- - i ? \u/ ' e i

Y f v V.

February 1998 May 1998

 

- -1.--.....Y -....

ﬁg

 

 

”l’
l
l
l.
I
l
l
i“
8
\
l
I
)t
1

 

November 1998
20 15 10 5 0

(August 199s
CH4 " Gig/I

Fig. 4. Spatial distribution of dissolved methane concentrations at the PTA-2 well array from October 1995 to November 1998. Vertical axis
is depth below water table and horizontal axis is relative distance laterally Upper surface of each image reﬂects water table position.
Groundwater flow is out of the page.

 

 

 

15

478

ML3

 

 

E "Sis-e; T: ‘2‘ "r“ . . l—. - arm
0 PM \ ‘1‘“. ""'l- "" . QM! . . 2 I?“ l
H ‘ ‘ -\ \ ‘ . _o -. f V' I is ’
l \ sar__,.__,....-1c.m ' L' (‘1‘ I
o l s \ ' “Mr, \ ‘--.‘.-" "l
E % K x __ 1” ..., '

i -
ee ! l
i A /\\A l

 

 

53 meters

October 1995

 

 

:5..— . -. 1 .. . -. I i _
17...-.. *‘x “(5‘72 .3! - - . ' .' 7- --\
ea \ “net .

\‘_ "_-‘_‘_ ‘Ta‘ (gun .1
f.\.\: “\ ~ 'k-..._.-._,‘ x .1 H .. ' ."- ,-
‘ "-—._ ~ ’ .‘ -_ A; .1
v ' N " ‘s..l.,/' _ -’1
i \/\,Q9J\ #_,/-'-"°1OM' l :
j", 0“ . ' "--—-‘ w

r\- A ‘1

October 1996

 

 

 

 

 

August 1998

F9”

403

J.T. McGuire et al. / Chemical Geology I69 (2000) 4 71—485

ML3 MLS

._ ‘in;-.3T~V‘JJEIT\} .- .
A " 1 .- . ‘f\ ’

'44—‘- , L"

.s—«' _ _ _‘:‘."..' ) .’]

 

   
  

. ‘7 f
I’d—uqow”.

‘ I r'.
’dr.’ -

- "F -c.m.._._,_,_.-_.-.o-'

 

 

 

 

 

 

 

 

 

 

 

 

\A ./A\‘

November 1998

  

v13

30-

25 20-15 10 5

h—u—u-lo

 

1

 

Fig. 5. Spatial distribution of dissolved iron concentrations at the PTA-2 well array from October 1995 to November 1998. Vertical axis is
depth below water table and horizontal axis is lateral distance across the plume. Upper surface of each image reﬂects water table position.

Groundwater ﬂow is out of the page.

16

J.T. McGuire et al. / Chemical Geology 169 (2000) 471—485 479

 

 

 

 

 

 

 

U)

‘-

O

H

3

oo

53 meters

l/MK-/ .-"3",I:t~ :xi‘f; 6 I] l ”/3 p‘, ~L,:_L 4— ,; I
. \ x4 ""' ~ ‘T‘eéf’m’ / i i \ "“‘°° . ' ‘
3 , ‘ . .
.'\ j I 1 \\ V r Y ‘ Y .I T |

 

 

October 1996 November 1997

«v..— _—‘-——- “Law-or-l

.. . .-
.\
‘
a .
’ l"
. h
'1 ‘ i
I‘,‘ r
a . I. _ -
.
1 P‘
.'
-1 ' >
. v
p. ""7“” T

 

 

May 1998

A L .I L L l ‘ 4
V . .

 

 

 

- A W‘-
\ -\ nﬁ ,h‘e ﬂ‘ “whgq‘mjv r}..’ > "31’
\ . ¢ ‘ ' o v
. . - - ‘ I ' . --

     

- x . _v_,- - __.',,

"W'-
1;...

.m . . L
i i\-’\ n c
L L —-— Y

7” r v v Y v 7 . f ﬁi

August 1998 November 1998

K+

Fig. 6. Spatial distribution of dissolved potassium concentrations at the PTA-2 well array from June 1996 to November 1998. Only limited
data were available for October 1995 to create a spartical plot. Vertical axis is depth below water table and horizontal axis is lateral distance
across the plume. Upper surface of each image reﬂects water table position. Groundwater ﬂow is out of the page.

 

 

. \
\ ' .
\- ._N r‘ ‘ ‘ _ ..
"x. xvi-4w"
"'*\-.- -

7.-. w-“ .- .u -.-_ .. ..- —.

 

420

 

l~

 

17

480 J.T. McGuire et al. / Chemical Geology 169 (2000) 471-485

October 1995. the highest H2 concentrations were
localized around the shallowest depths at ML3 and
MLS. These high concentrations indicate a potential
zone of methanogenesis by C02 reduction (Chapelle
et al., 1996). Hydrogen concentrations decrease away
from this zone suggesting a change in process from
methanogenesis to SO“t reduction to Fe3+ reduction
(Chapelle et al.. 1996). Throughout the well array.
the TEAP inferred by [‘12 measurements changed
from a large area of methanogenesis in October 1995
to predominantly SO4 reduction in June 1996. to
predominantly 1%” reduction in August 1998, and
then to predominantly SO4 reduction in November
1998. The overall decrease in H2 concentrations
from October 1995 until August 1998 suggests that a
long—term change in general TEAP conditions oc-
curred over the 3-year period across the transect.
TEAP conditions as deﬁned by 1-12 concentrations
also varied on short time scales as evidenced by the
change from predominantly Fe3*-reducing condi-
tions observed in August 1998 to SO4-reducing con-
ditions in November 1998.

Methane concentrations varied on short time scales
throughout the 3—year sampling period but did not
parallel the long-term pattern of H2 concentration
(Fig. 4). In October 1995, the highest concentrations
of CH 4 were found in the shallow wells at ML3
corresponding to the area in the well array of highest
[-12 concentrations. This is consistent with the inter—
pretation of methanogenesis by C02 reduction oc-
curring in this area (Chapelle et al., 1996). However.
changes in dissolved CH 4 concentrations in the well
array did not always parallel those predicted by I-I2
concentrations. Similar to H2, CH 4 concentrations
were highest in October 1995 and decreased in June
1996. After June 1996, [-12 concentrations continued
to decrease. while CH 4 concentrations did not.

High Fez‘ concentrations were present over the
3-year period. but the loci of highest concentrations
changed (Fig. 5). For example. in October 1995. the
highest concentrations of Fe2+ were found in the
easternmost portion of the well array at ML8. In
June 1996. the highest concentrations of Fe2+ were
found in the westernmost portion of the well array at
ML4. Similar to H2 and CH 4, the pattern of Fe2+
concentrations was found to be variable on monthly
timescales. In addition. the spatial pattern for Fe:‘
appears to be similar to that of CH 4.

Dissolved potassium concentrations in the con-
taminant plume were elevated throughout the 3-year
period compared to concentrations in the surround-
ing groundwater (Fig. 6). The source for K" in the
plume is suspected to be from the KHCO3 used to
extinguish the ﬁres during the fire training exercises.
Therefore. the elevated concentrations of K+ in the
groundwater of the well array indicate water that has
moved through the contaminant source near the pad.
As can be seen, the area of elevated K" concentra-
tions matches the spatial patterns of H2. CH 4, and
Fe2+ (Figs. 3—5). In addition. the changes in the
concentration levels of K‘ at the well array roughly
match the changes in H2 concentrations. Spatial
plots of K’ could not be constructed for groundwa-
ter samples taken in October 1995 because of a
limited dataset: however. K+ concentrations were
some of the highest recorded (29 mg/l) at ML3.
This corresponds to a time of high H2 concentrations
(Fig. 3). After October 1996. K+ decreased similarly
to H2 (Figs. 3 and 6). In November 1998. K concen-
trations increased, as did the H2 values.

5. Discussion

The inﬂuence of groundwater recharge on the
spatial and temporal variability of H3, CH 4 and
Fe2+ concentrations can be examined by comparing
the spatial maps (Figs. 3-5) to changes in the eleva-
tion of the water table at FTA-Z (Fig. 7). It can be
seen from Fig. 7 that groundwater samples from
October 1995 were collected after a period of a
steadily declining water table. Without recharge.
electron acceptors such as 03. Fe3 +. and SO 4 would
be depleted allowing for the development of an
extensive methanogenic zone within the plume re-
sulting in elevated H2 and CH 4 concentrations. In
contrast, samples from June 1996 were collected
shortly after a recharge event and 1-12 and CH 4
concentrations decreased. Dissolved hydrogen and
CH 4 concentrations in the well array might decrease
during recharge events as a result of either (1)
dilution by recharge water or (2) by limiting
methanogenesis via the introduction of more energet-
ically favorable electron acceptors such as SO4 to the
system. Lower concentrations of other ions (e.g..
Na’) during this time period were not observed.

18

J.T. McGuire er al. / Chemical Geology 169 (2000) 471—485

481

 

 

 

 

184.875 _._..
z a S MSU/USGS Sampling Events 1
g § 13mm» 0 USGS Sampling Events !
< l
a a
—l t!) 134 625 my” '
Ill Ill 1’
s a
g 2 mm‘ Jun96 l
E a) Nov 97 S\ m) ‘I
I! E 184. 375 - l
< m l
3 g Aug 98% E
184.250? l
7 l
184.125 : . . , "W 988:

Jan 95 Jan 96 Jan 97 Jan 98 Jan 99

Fig. 7. Graph depicting water table elevation as a function of time at FI‘A-Z. Data were taken from water level recorder 2 (solid line) and
manual measurements (dotted line). Periods when groundwater samples were collected are also shown. Circles represent samples collected
by the US. Geological Survey and S's represent samples collected in this study.

This suggests a change in TEAP conditions and not
dilution was the dominant cause for changes in H2
and CH 4 concentrations between October 1995 and
June 1996. Recharge events in June 1996, October
1996, November 1997, February 1998. and May
1998 could, in part. explain the low hydrogen con-
centrations during these time periods.

Based on these observations. we argue that under
conditions of declining water table, the tendency of
the system will be toward the development of more
reduced (e.g., methanogenic) TEAP conditions. while
recharge events will push the systems to less reduced
conditions (e.g., iron reduction). Although such
changes have been observed at other sites (e.g.,
Vroblesky and Chapelle, 1994). predicting speciﬁc
TEAP conditions (based on H2 concentrations) from
knowledge of when recharge occurred and the eleva-
tion of the water table remains a challenge. For
example, the August 1998 samples were collected
after a period of prolonged water—table decline simi-
lar to the conditions that occurred during the October
1995 sampling (Fig. 7). In addition. the water table
elevation was similar to that of October 1995. Thus.
it was anticipated that I-I2 concentrations in August
1998 would be similar to those of October 1995.
However, H2 concentrations in August 1998 were

19

much lower than in October 1995. Conversely, in
November 1998. with continuing decline in the water
table. l-I2 concentrations increased. A major differ-
ence between 1998 and 1995 periods was that the
recharge event preceding the declining water table in
1998 was signiﬁcantly larger than in 1995. This
suggests that the magnitude of recharge may play a
role in how H2 responds to recharge events. For
example August 1998 the water table elevation was
similar to those measured in 1995.

Methane concentrations did not parallel the over-
all decline in hydrogen concentrations over the 3-year
period. Methane and hydrogen patterns may not be
coupled because CI-I.1 may be produced by a mecha-
nism that does not rely on I-I2 such as aceticlastic
methanogenesis (Table 1). Another possibility for the
decoupling is differential transport of the solutes.
Methane and other solutes such as Fe“ can be
mobile in groundwater while [-12 is highly reactive;
thus, transport in groundwater is likely minimal
(Lovely and Goodwin. 1988).

We hypothesize that for the majority of the 3-year
period, the contaminant plume at the well array was
undergoing aceticlastic methanogenesis in combina-
tion with various intensities of SO4 and Fe“ reduc-
tion. Methane in this system may be produced by

482 J. T. McGuire et al. / Chemical Geology 169 (2000) 471—485

two separate mechanisms. Hydrogen-oxidizing me-
thanogens can produce methane from hydrogen and
carbon dioxide (CO2 reduction) and aceticlastic
methanogens can produce methane and carbon diox-
ide from acetate (aceticlastic methanogenesis). Only
in October 1995 and one sample in June 1996 were
H2 concentrations found to be high enough to indi-
cate the production of CH 4 by C02 reduction. Un-
like CO2 methanogenesis. which requires high con-
centrations of H: . aceticlastic methanogenesis has no
relationship to H2 concentrations. A result, methane
may be produced by aceticlastic methanogenesis.
while other microbial processes such as SO4 reduc-
tion occur. This hypothesis is supported by recent
studies. which indicate that at temperatures lower
than 20°C. CH 4 production occurs predominantly by
aceticlastic methanogenesis and not by C02 reduc-
tion (Nozhevnikovs et al.. 1994; Schultz and Conrad.
1996). Groundwater temperatures at the site appear
to have remained below 20°C throughout the 3-year
period. Furthermore. Dojka et al. (1998) identiﬁed
aceticlastic methanogens by a culture-independent
molecular phylogenetic approach (on sediments col-
lected from ML-3. see Fig. 2. in 1996). which sup-
ports the possibility of in situ CH 4 production.

Over the 3-year period. the threshold for H2
concentrations was maintained below that necessary
for CO2 methanogenesis. A possible explanation for
the low H2 concentrations is its consumption by
homoacetogenesis (the production of acetate from
hydrogen and carbon dioxide) (e.g.. Chin and Con-
rad. 1995) and / or dechlorination (Sulﬁta et al.. 1982)
processes. Evidence of microbiological communities
associated with both processes were found by Dojka
et al. (1998). Fennel] and Gossett (1998) examined
the competition for hydrogen in a dechlorinating
culture. They clearly show how low H2 levels driven
by dechlorination reactions favor aceticlastic
methanogenesis over C02 methanogenesis.

If H2 is consumed by in situ reactions such as
dechlorination, then its low concentration throughout
most of the 3-year period is not indicative of other in
situ processes such as SO4 and Fe“ reduction (Mc—
Guire et al.. 1999). In fact. Dojka et al. (1998) did
not ﬁnd microbial evidence for $04 and Fe3 ’ reduc-
ers in sediments collected from ML-3 in 1996. Iron
sulﬁde staining was observed on the lower portions
of the tubing used to collect samples from the multi-

level wells in November 1998, indicating that in situ
sulfate reduction has occurred in the past. However.
it is possible that this staining is related to processes
occurring locally at the well screen not related to
processes in the aquifer as a whole.

The inﬂuence of solute transport on concentra-
tions of TEAP parameters in the well array is sug-
gested by the relatively high concentrations of Fe“.
and the lack of microbial evidence for in situ Fe3+
reduction. The high concentrations of Fe2+ through-
out the well array indicate Fe3+ reduction has oc-
curred within the plume. Similarly, the low sulfate
concentrations detected in the plume at the well
array indicate sulfate reduction has occurred. How-
ever. we have found only traces of dissolved sulﬁde
in the plume. Except for 80... inputs from rain (less
than 3 ppm), there are few sources for $04; and.
thus, the high concentrations of Fe2+ effectively
remove any sulﬁde produced as iron sulﬁdes.

The high concentrations of Fe2+ in groundwater
at the well array can be explained by the fact that in
the anaerobic groundwater dissolved Fez‘ can be
transported along the ﬂow path. Dissolved iron (11)
would mainly be removed by oxidation to Fe3+ and
precipitation of oxyhydroxides. Such oxidation could
occur at the interface between the anaerobic plume
water and the oxic native groundwater (along the
fringes of the plume and at the water table). The
newly formed iron oxyhydroxides could then be-
come a substrate for iron reducers. Recharge events
and changes in water table elevation and groundwa-
ter ﬂow direction would drive iron cycling at the
interface. A similar argument can be made for the
fate, occurrence and transport of dissolved CH 4
throughout the well array. The similarity in the
spatial extents, but not necessarily concentration pat-
terns of Fe” and CI-I‘1 (Figs. 4 and 5, respectively)
support this argument. Thus. Fe“ and perhaps CH 4
could have been produced upgradient and trans-
ported to the well array and. therefore. their concen-
trations would not reflect in-situ processes.

6. Conclusions
The redox environment in a petroleum-derived

plume inﬂuences the efﬁciency. rate. and extent of
biodegradation (Vogel. 1994). Biodegradation of

20

J.T. McGuire et al. / Chemical Geology 169 (2000) 471 —485

BTEX compounds is most efﬁcient under aerobic
conditions, but has been demonstrated to occur under
most redox conditions (Baedecker et al.. 1993:
Lovely et al., 1995). In aerobic environments, BTEX
compounds are degraded to C02. In methanogenic
environments. the complete transformation of or-
ganic compounds to CO2 and CH4 requires the
combined activities of at least three different func-
tional groups of microorganisms: fermenters. fatty-
acid degraders. and methanogens. Thus, the effec-
tiveness of groundwater remediation strategies will
be inﬂuenced not only by our ability to accurately
delineate redox zones in a contaminated aquifer. but
also to determine the hydrogeological, microbiologi-
cal, and geochemical factors inﬂuencing temporal
variations in the zones.

In this paper. we described the spatial and tempo-
ral variations of H2. CH 4 and Fe2+ in shallow
groundwater contaminated with both waste fuel and
chlorinated solvents over a 3-year period. Results
indicate that changes in all parameters occurred on
short time scales (months). and that hydrogen con-
centration also exhibited long-term changes over a
longer 3-year period. Recharge events. as indicated
by changes in water-table elevation. appear to be
related to short-term changes in the concentrations of
H2. CH 4, and Fe“ and perhaps long-term changes
in the concentrations of H3. Solute transport (includ-
ing recharge), in part, controls the availability of
organic matter, nutrients. and electron donors and
acceptors to the bacteria that carry out many TEAP
reactions. These results strongly support the conclu-
sions of Chapelle et a1. (1996) that if H2 concentra-
tions are to be used as an indicator of TEAP pro-
cesses. they are best interpreted in combination with
other TEAP variables.

Our results also show that interpretations of TEAP
processes from solute concentrations must consider
the roles of in situ (at the sampling point) biogeo-
chemical and solute transport processes. The relative
importance of these processes appears to exhibit
temporal variability. which is the subject of our
on-going studies (McGuire et al.. 1998: Smith et al..
1998; Smith et al.. 1999). Microbial measurements
are often necessary to conﬁrm the interpretations
made from geochemical and hydrogeological mea-
surements on these processes. In aquifers contami-
nated with a variety of organic chemicals (e.g..

21

483

petroleum hydrocarbons, chlorinated solvents) H2
concentrations may not be indicative of TEAP pro-
cesses as they have been traditionally interpreted by
other workers such as Lovely and Goodwin (1988)
since these previous studies did not consider the
effect of dechlorination reactions on H2 partial pres-
sures.

Acknowledgements

We would like to thank Jonathan I. Kolak for his
assistance with project development and initial
ﬁeld-sample collection. The research is funded by a
grant from the National Science Foundation Environ-
mental Geochemistry and Biogeochemistry Program.
EAR-9708487. Funding is also provided by the
USGS Toxic Substances Hydrology Program. We
would also like to thank the National Center for
Integrated Bioremediation, Research, and Develop-
ment (NCIBRD) and EFX Systems. Incorporated for
their cooperation.

References

Baedecker. M.I.. Back. W.. 1979. Modern marine sediments as a
natural analog to the chemically stressed environment of a
landﬁll. I. Hydrol. 43. 393—414.

Baedecker. MJ.. Cozzarelli. I.M.. Eganhouse. R.P.. Siegel. D.I..
Bennett. PC. 1993. Crude oil in a shallow sand and gravel
aquifer: III. Biogeochemical reactions and mass balance mod-
cling in anoxic groundwater. Appl. Geochem. 8. 569-586.

Barcelona. M.J.. Holm. T.R.. Schock. M.R.. George. G.K.. 1989.
Spatial and temporal gradients in aquifer oxidation-reduction
conditions. Water Resour. Res. 25. 991-1003.

Beller. H.R.. Reinhard. M.. 1992. Metabolic by-products of anaer-
obic toluene degradation by sulfate~reducing enrichment cul-
tures. Appl. Environ. Microbiol. 58. 3192-3195.

Bennett. P.C.. Siegel. D.E.. Baedecker. Ml. Hult. M.F.. 1993.
Crude oil in a shallow sand and gravel aquifer: I. Hydrogeol-
ogy and inorganic geochemistry. Appl. Geochem. 8. 529-549.

Bemer. R.A.. 1981. Early Diagenesis. A Theoretical Approach.
Princeton Univ. Press. Princeton. 241 pp.

Champ. D.R.. Gulens. 1.. Jackson. R.E.. 1979. Oxidation—reduc-
tion sequences in groundwater ﬂow sysrems. Can. J. Earth Sci.
16. 12—23.

Chapelle. F.H., McMahon. P.B.. 1991. Geochemistry of dissolved
inorganic carbon in a coastal plain aquifer. J. Hydrol. 127.
85-108.

Chapelle. F.H., 1993. Groundwater Microbiology and Geochem-
istry. Wiley-Liss. New York. 424 pp.

484 J.T. McGuire er al. / Chemical Geology 169 (2000) 471-485

Chapelle. F.H.. McMahon. P.B.. Dubrovsky. N.M.. Fujii. R.F..
Oaksford. E.T.. Vroblesky. D.A.. 1995. Deducing the distribu-
tion of terminal electron accepting processes in hydrologically
diverse groundwater systems. Water Resour. Res. 31. 359-371.

Chapelle. F.H., Haack. S.K.. Adriaens. P.. Henry. M.A.. Bradley.
P.M.. 1996. Comparison of Eh and H2 measurements for
delineating redox processes in a contaminated aquifer. Envi-
ron. Sci. Technol. 30. 3565-3569.

Chin. K1. Conrad. R.. 1995. Intermediary metabolism in
methanogenic paddy soil and the inﬂuence of temperature.
FEMS Microbiol. Ecol. 18. 85-102.

Christensen. T.H.. Kjeldsen. P.. Albrechtsen. H.1., Heron. G..
Nielsen. P.H.. Bjerg. P.L.. Holm. P.E.. 1994. Attenuation of
landﬁll leachate pollutants in aquifers. Crit. Rev. Environ. Sci.
Technol. 24. 119-202. .

CRC. 1993. In: Lide. D.R. (Ed). Handbook of Chemistry and
Physics. 74th edn. CRC Press. Ann Arbor.

Davis. A.. Kempton. 1.H.. Nicholson. A.. 1994. Groundwater
transport of arsenic and chromium at a historical tannery.
Wobum. Massachusetts. USA. Appl. Geochem. 9. 569-582.

Dojka. M.A.. Hugenholtz. P.. Haack. S.K.. Pace. NR. 1998.
Microbial diversity in a hydrocarbon- and chlorinated-
solvent-contarninated aquifer undergoing intrinsic bioremedia-
tion. Appl. Environ. Microbiol. 64. 3869-3877.

Dolﬁng. 1.. Beurskens. 1.E.. 1995. The microbiologic and environ-
mental signiﬁcance of reductive dehalogenation. Adv. Micro-
biol. Ecol. 14. 143-160.

Edwards. E.A.. Wells. L.E.. Reinhard. M.. Grbic-Galic. D.. 1992.
Complete mineralization of benzene by aquifer microorgan-
isms under strictly anaeobic conditions. Appl. Environ. Micro-
biol. 58. 2663-2666.

Essaid. H.I.. Bekins. B.A.. Godsy. E.M.. Warren. E.. Baedecker.
M.1.. Cozzarelli. I.M.. 1995. Simulation of aerobic and anaer-
obic biodegradation processes at a crude oil spill site. Water
Resour. Res. 31. 3309-3327.

Fennell. D.E.. Gossett. I.M.. 1998. Modeling the production of
and competition for hydrogen in a dechlorinating culture.
Environ. Sci. Technol. 32. 2450-2460.

Goode. 0.1.. Konikow, LR. 1990. Apparent dispersion in tran-
sient groundwater ﬂow. Water Resour. Res. 26. 2339-2351.

Grbic-Galic. D.. Vogel. T.M.. 1987. Transformation of toluene
and benzene by mixed methanogenic cultures. Appl. Environ.
Microbiol. 53. 254-260.

Hess. K.M.. Wolf. S.H.. Celia. M.A.. 1992. Large-scale natural
gradient tracer test in sand and gravel. Cape Cod. Mas-
sachusetts: 3. Hydraulic conductivity variability and calculated
macrodispersivities. Water Resour. Res. 28. 2011-2027.

Hunter. K.S.. Wang. Y.. Van Cappellen. P.. 1998. Kinetic model-
ing of microbially-driven redox chemistry of subsurface envi-
ronments: coupling transport. microbial metabolism and geo-
chemistry. 1. Hydrol. 209. 53-80.

Hutchins. S.R.. Downs. W.C.. Wilson. J.T.. Smith. G.B.. Kovacs.
D.A.. Fine. D.D.. Douglas. R.H.. Hendrix. D.1.. 1991. Effect
of nitrate addition on biorestoration of fuel-contaminated
aquifer: ﬁeld demonstration. Ground Water 29. 571-580.

Keating. E.H.. Bahr. 1.M.. 1998. Reactive transport modeling of
redox geochemrstry: approaches to chemical disequilibrium

and reaction rate estimation at a site in northern Wisconsin.
Water Resour. Res. 34. 3573-3584.

LeBlanc. D.R.. Garabedian. S.P.. Hess. K.M.. Gelhar. L.W..
Quadri. R.D.. Stollenwerk. K.G.. 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. 895-910.

Lovely. DR. 1991. Dissimilatory Fe (III) and Mn (IV) reduction.
Microbiol. Rev. 55. 259-287.

Lovely. DR. 1994. Microbial reduction of iron. manganese. and
other metals. Adv. Agron. 54. 175-227.

Lovely. D.R.. Coates. 1.D.. Woodward. 1.C.. Phillips. E.J.P..
1995. Benzene oxidation coupled to sulfate reduction. Appl.
Environ. Microbiol. 61. 953-958.

Lovely. D.R.. Goodwin. S.. 1988. Hydrogen concentrations as an
indicator of the predominant terminal electron accepting pro-
cess in aquatic sediments. Geochim. Cosmochim. Acta 52.
11-18.

Lovely. D.R.. Lonergan. D.1.. 1990. Anaerobic oxidation of
toluene. phenol. and p-cnesol by the dissimilatory iron-reduc-
ing organisms. GS-IS. Appl. Environ. Microbiol. 52. 1858-
1864.

Lovely. D.R.. Phillips. E.J.P.. 1987. Competitive mechanisms for
the inhibition of sulfate reduction and methane production in
the zone of ferric iron reduction in sediments. Appl. Environ.
Microbiol. 53. 2636-2641.

McGuire. J.T.. Kolak. 1.1.. Long. D.T.. Hyndman. D.W.. Velbel.
M.A.. Klug,. M.1.. Haack. S.l(.. 1998. Biogeochernical obser-
vations of groundwater redox dynamics. Geol. Soc. Am. Annu.
Meet.. Abstracts with Programs 30. 303.

McGuire. J.T.. Long. D.T.. Haack. S.I(.. Klug. M.1.. Hyndman.
D.W.. Smith. E.W.. Velbel. M.A.. 1999. Challenges of assess-
ing temporal variations in redox zonation. Geol. Soc. Am.
Annu. Meet.. Abstracts with Programs 31. 324.

Norris. R.D.. 1994. Handbook of Bioremediation. US. EPA Robert
S. Kerr Environmental Research Laboratory. Lewis Publishers.
Ann Arbor.

Nozhevnikovs. A.N.. Kotsyurbenko. O.R.. Simankova. M.V..
1994. Acetogenesis at low temperature. In: Drake. H.L. (Ed).
Acetogenesis. Chapman and Hall. London. England.

Rabus. R.R.. Nordhaus. N.. Ludwig. W.. Widdel. F.. 1993. Com-
plete oxidation of toluene under strictly anoxic conditions by a
new sulfate-reducing bacterium. Appl. Environ. Microbiol. 59.
1444-1451.

Salvage. K.M.. Yeh. G.. 1998. Development of a numerical model
of kinetic and equilibrium microbiological and geochemical
reactions (BIOKEMOD). J. Hydrol. 209. 27-52.

Schultz. 5.. Conrad. R.. 1996. Inﬂuence of temperature on path-
ways to methane production in the permenantly cold profundal
sediment of Lake Constance. FEMS Microbiol. Ecol. 20.
1-14.

Smith. E.W.. Hyndman. D.W.. Kolak. 1.1.. Long. D.T.. Velbel.
M.A.. Klug. M.1.. Haack. SK. 1998. The inﬂuence of recharge
on ground water redox zonation. Geol. Soc. Am. Annu. Meet-
ing. Abstracts with Programs 30. 22.

Smith. E.W.. Hyndman. D.W.. Long. D.T.. 1999. Simulated plume
evolution using transient ﬂow conditions observed at Wurt-

22

J.T. McGuire er al. / Chemical Geology 169 (2000) 4 71 —485 485

smith AFB. Michigan. EOS Trans. Am. Geophys. Union
Spring Meet. Suppl. 155. — April 27 80. $137.

Stumm. W.. Morgan. 1.1.. 1981. Aquatic Chemistry. 2nd edn.
John Wiley and Sons. New York. 780 pp.

Sulﬁta. J.M.. Horowitz. A.. Shelton. D.R.. Tiedje. J.M.. 1982.
Dehalogenation: a novel pathway for the anaerobic biodegra-
dation of haloaromatic compounds. Science 218. 1115-1117.

U.S.G.S.. 1991. Installations Restoration Program: Phase Il-Con-
ﬁrmation / Quantiﬁcation Stage 1: Wurtsmith Air Force Base.
Michigan: Investigations of groundwater and soil contamina-
tion at selected sites. United States Geological Survey. Water
Resources Division.

23

Vogel. T.M.. 1994. Natural bioremediation of chlorinated sol-
vents. In: Norris. R.D. (Ed). Handbook of Bioremediation.
US. EPA Robert S. Kerr Environmental Research Laboratory.
Lewis Publishers. Ann Arbor.

Vroblesky. D.A.. Chapelle. EH. 1994. Temporal and spacial
changes of terminal electron accepting processes in a petroleum
hydrocarbon-contaminated aquifer and the signiﬁcance for
contaminant biodegrcdatron. Water Resour. Res. 30. 1561-
1570.

CHAPTER II.

Geochemical characterization of the effects of recharge on an anaerobic contaminated
aquifer using multivariate statistics

24

Introduction

Understanding conditions that can initiate changes in the redox state of anaerobic
systems is crucial to evaluating and managing natural and contaminated systems since
redox state controls the form, mobility, toxicity, and persistence of many natural and
anthropogenically introduced chemical constituents. Spatial and temporal variations in
groundwater redox chemistry are well documented (BARCELONA et al., 1989; CRILL and
MARTENS, 1987; KENNEDY etal., 1999; VROBLESKY and CHAPPELLE, 1994). It is
therefore essential to not only identify the distribution of redox processes at a given point
in time but also to understand the conditions under which redox state may vary. In
groundwater systems, recharge events are thought to be a signiﬁcant drive for initiating
changes in redox state (VROBLESKY and CHAPPELLE, 1994); however speciﬁc inﬂuences
and responses are not well understood.

The redox state of a system is in part controlled by the activity of
microorganisms, which ﬁrst consume oxygen, and then a succession of alternate terminal
electron acceptors (TEAS) to support their growth using a variety of carbon sources
(CHAPPELLE, 2001; LOVLEY, 1991; LOVLEY and GoonwrN, 1988; STUMM and MORGAN,
1981). The succession of terminal electron accepting processes (TEAPS) in order of
decreasing redox potential and free energy yield is generally: oxygen reduction, nitrate
reduction, manganese reduction, iron reduction, sulfate reduction, and methanogenesis.
In most closed groundwater systems, the sequence of TEAPS results in redox zones
following this succession (i.e., increasingly reduced conditions occur further down the
ﬂow path) (CHAMP et al., 1979). However, downgradient of contaminant plumes, the

sequence of TEAP zones is generally reversed (CHAPPELLE, 2001; LOVLEY, 1991). That

25

is, the most reduced conditions (e. g., methanogenesis) occur near the contaminant source
due to the greater availability of electron donors, while less reducing conditions (e. g.,
iron reduction) occur down ﬂow-path from the contaminant source. Since many
hydrocarbons, which serve as electron donors, concentrate in the capillary fringe area,
more reducing conditions often occur near the water-table interface, resulting in a plume
that has vertical as well as horizontal redox zonation (CHAPPELLE et al., 1996). In
reduced systems, recharge events can initiate changes in TEAPS by delivering more
favorable TEAS such as oxygen (02), nitrate (N 03) and sulfate (804) to the system
(VROBLESKY and CHAPPELLE, 1994). ‘

Many aspects of how recharge waters affect TEAPS and redox zonation in
reduced systems remain unknown. Direct examination of redox sensitive species, in an
aquifer before and after recharge, cannot distinguish the extent to which recharge initiates
TEAPS verses a simple dilution of aquifer water with recharge water since both processes
result in a decrease in TEA concentrations. It is also unclear the degree to which redox
zones are impacted during recharge, the depth to which recharge inﬂuences redox, and
how subtle differences in topography may enhance recharge effects. One goal of this
study is to identify the chemical signature of a recharge event at various scales within an
aquifer to evaluate the changes initiated by recharge and examine the spatial pattern of
impact. Our hypothesis is that changes in TEAPS, initiated by recharge, occur primarily
in small transition zones rather than producing changes in dominant redox zonation.
Therefore, changes in biogeochemical parameters, including those indicative of TEAPS,

were evaluated at a large plume scale and a smaller single-well scale.

26

Identifying the chemical changes associated with a recharge event is difﬁcult due
to the chemical complexity of contaminated systems. As shown by Suk and Lee (1999)
spatial and temporal measurements of chemical and physical properties alone usually do
not expose the underlying processes in complex groundwater systems. Traditional
geochemical tools such as Stiff and Piper diagrams classify water masses based on
patterns of major ion concentrations. These patterns give insight into the chemical
evolution a water mass based on water-rock interactions. In contaminated systems,
relying solely on the distributions of major ions to classify water masses is ineffective
because it neglects many important parameters that represent the hydrogeochemical
condition of the water. These traditional analysis methods do not consider redox
sensitive species or other indicators of biogeochemical processes that may control water-
chemistry variations in contaminated systems. As a result, these tools are inadequate to
reveal the underlying associations between chemical and physical parameters that lead us
to interpretations of processes during recharge events in contaminated systems.

Multivariate analyses that consider a more complete set of physical and chemical
parameters (including: major cations and anions, redox related species, alkalinity,
temperature, pH, and redox potential) can provide a better understanding of important
parameter associations in contaminated systems than traditional classiﬁcation methods.
Several researchers (e.g., (LEE et al., 2001), (SWANSON et al., 2001), (RITZI etal., 1993))
have used factor and/or cluster analysis to interpret groundwater chemical data. Suk and
Lee (1999) demonstrated the effectiveness of combining factor and cluster analyses to
reveal the governing processes and hydrogeochemical similarities between measured

parameters. One criticism of factor analysis techniques is that they are not directly

27

observable. This study employs the strategy of combining factor and cluster analysis
techniques with a visual representation of the actual chemical signature of the resulting
water masses. By examining the chemistry of the resulting water masses from the cluster
analysis an important link between the measured parameters and calculated factors is
established making process interpretations more meaningful.

Our approach involved ﬁrst performing factor analysis to determine chemical
associations, which allow for the evaluation of the dominant processes controlling
groundwater chemistry. Factor scores were then computed for each sample location,
which describe the importance (or loading) of each factor on that particular sample.
These factor scores were then used in the cluster analysis. Cluster analysis was
performed on the factor scores as opposed to the raw chemical data to reduce
classiﬁcation error due to data outliers or multicollinearity. This is effective because
factors are independent of one another and indicative of the underlying structure of the
original variables. These clusters were then plotted in 3-dimensions to identify the spatial
relationship between various water masses within the plume. Once the water masses
were deﬁned, they were examined not only in terms of the dominant factors controlling
the cluster but also examined the chemical composition to better deﬁne the factor
loadings.

The primary objective of this study is to evaluate change in a reduced aquifer
during a recharge event using statistical and graphical methods, and demonstrate the
effectiveness of multivariate statistical techniques for revealing process changes in
chemically-complex systems. Speciﬁcally, this paper seeks to 1) characterize a reduced

groundwater contaminant plume before and after a recharge event in terms of dominant

28

biogeochemical processes; 2) deﬁne the chemical changes associated with the recharge
event, and 3) determine the spatial distribution of process changes. These analyses
provide a detailed investigation of the role of recharge as a drive for redox state
dynamics, which is crucial to the proper understanding and management of natural and

contaminated reduced systems.

Materials and Methods

Study Site Description:

The study site is a shallow sandy aquifer contaminated with petroleum
hydrocarbons and chlorinated solvents at the former Wurtsmith Air Force Base (AFB),
Michigan, U.S.A. Wurtsmith AFB lies on a sandy plain of glacial lake sediment
approximately 2 km west of the Lake Huron shoreline in Michigan's Lower Peninsula.
The aquifer is approximately 20 meters thick and composed of highly permeable
alternating aeolian sands and glacial outwash material. Hydraulic conductivities of these
deposits are on the order of 30 m/day. The unconﬁned aquifer is underlain by at least
30.5 meters of silty clay. The water table varies from 3.5-5.0 m below land surface (184-
186 m above sea level), ﬂuctuates 0.3 to 0.7 m annually, and the water table gradient
ranges from 3 to 5 meters per kilometer. Assuming an effective porosity of thirty
percent, average groundwater velocity is approximately 0.3-0.5 m per day. The plume has
lateral direction shifts of 7 to 10 degrees because the groundwater ﬂow direction is

inﬂuenced by seasonal recharge at this site.

29

Fire training activities performed by the Air Force from 1952 to 1986 at this site
contaminated the groundwater. Various compounds were used during the training
exercises to ignite and extinguish the ﬁres, including waste fuels, chlorinated solvents,
aqueous ﬁlm-forming foams, a multipurpose dry chemical, potassium bicarbonate based
soda, and Halon 1211. An extensive plume approximately 50 m wide and 400 m long
has developed, with the majority of contaminant mass residing within the capillary fringe
(Figure 2.1). Most of the contamination is associated with aquifer solids between 4.5 and
5.7 m below land surface, which have an average total hydrocarbon concentration of
13,650 mg/kg (USGS, 1991). The plume is complex in that it contains both BTEX
compounds (benzene, toluene, ethylbenzene, and xylenes) and chlorinated solvents (e. g.,
dichloroethene). Concentrations of BTEX compounds in the plume range from 20 to
1,000 rig/liter while concentrations of dichloroethane, chloroethane, and vinyl chloride
range from 2 to 1,000 rig/liter.

In June 1998, a pilot ﬂow-through bioreactor system was installed near the
concrete pad to evaluate the effectiveness of stimulated microbiological activity to
remove BTEX and chlorinated solvent contamination from the site. As part of this
system, groundwater was extracted, augmented with substrates including nitrate (as
NaNO3) to serve as a TEA for microbial metabolism, and pumped through the bioreactor
system. This water was then returned to the test plot near the concrete pad. The
extraction well then re-captured the majority of “inﬂuenced” water and continued the
cycle. This operation ended in June 1999. A slug of bioreactor-inﬂuenced water, with

increased alkalinity, NH4+ and Na concentrations, reached our monitoring wells ML3 and

30

 

 

 
  

Conce _ x.“
\\\\

    
  
 

Multi-Level
Tra I : ' ct *9

0 50 METERS

0 100 FEET

 

 

 

Figure 2.1: Plume map for the Fire-Training Area 2 (PTA-2) at the former
Wurtsmith Air Force Base, Michigan. The concrete pad was the site of ﬁre-
training activities. Grey lines represent plume position during periods of low

water table; Red lines represent plume position during periods of high water table.

Blue lines are water table contours in feet above sea level.

31

MLS between February 1999 and May 1999. The bulk of the inﬂuenced water passed
through our well grid between October 1999 and January 2000. However, some remnant
water may still have been inﬂuencing the chemistry at the farthest downgradient wells
during January 2000.

The overall reducing conditions in the plume vary historically from F e—reducing
to methanogenic (as determined by H2 concentrations) on the time scale of months
(MCGUIRE et al., 2000). To document the vertical distribution of contaminants within the
plume, a series of multi-level sampling wells were installed at this site (Figure 2.2). The
wells were constructed of 2.5 cm inner diameter PVC casing, with 0.3 m long, 6.35 mm
slotted PVC screens. Vertical spacing between screens is approximately 1 m. This is an
ideal site to examine the effects of recharge on reduced systems since a detailed spatial
and temporal geochemical, hydrogeologic, and microbiological database for the site has
been established (DOJKA et al., 1998; HAACK and BEKINS, 2000; MCGUIRE et al., 2000;

REYNOLDS and HAACK, 1999).

Chemical Analysis:

The chemical composition of the groundwater at selected wells within the well
grid was evaluated in January and June of 2000 because a signiﬁcant recharge event
occurred between these two sampling events. Groundwater head measurements from 3
hourly-recording pressure transducers indicated that the hydraulic head during January

2000 was still in a seasonal decline due to lack of recharge (Figure 2.3). In May, the

32

Nor-thing (ft)

600 .

500 -

400 -

300 ~

200 ~

100 -

 

 

 

 

 

 

O FT-ML
' EFX-ML
. . "a, 318 A Pumping Well
0 .0 '
1.- 3.10
’ 330 01 EFX d ti u
02 pro no on we
12
0
” O
11 O 06
O 15
Q 14 O
13 o O 09
16 O 08
O 05
004 03
C)
O 07
10
100 200 300 400 500
Easting (ft)

Figure 2.2: Two—dimensional well grid installed near PTA-02.

33

Water Table Elevation (ft above sea level)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

moPm

 

 

mouro

 

 

mow.m

 

 

 

mowoo _ _ . _ _
oioimm oioiom AMBABG ._N\w._\m.~ oioioo oioioo 35.200

Ema—d Nb" Econ—538m E 2&9. BEG «rim—mo: gnaw 39. $89 Enema mam—w 3.30m 8332: gumam 303m.

34

groundwater head levels sharply increase due to snowmelt recharge. Peak head
conditions occurred in late May and groundwater samples were collected in early June.
Each depth within a well was purged with a peristaltic pump at a rate of
approximately 400 m1/min. A ﬂow cell (Yellow Springs Instruments) was used to
continuously monitor temperature, pH, dissolved oxygen, Eh, and speciﬁc conductance
until readings became stable. Dissolved H2 gas samples were collected using the bubble-
stripping method (CHAPPELLE and MCMAHON, 1991) and analyzed immediately in the
ﬁeld on a RGA3 reduction gas analyzer (Trace Analytical). Dissolved iron (Fey) and
sulﬁde (82') were measured in the ﬁeld, by phenanthroline and Chemetrics Vacu-VialsTM
colorimetric analyses respectively. Alkalinity was also measured in the ﬁeld by
potentiometric titration using Gran plots for graphical determinations (DREVER, 1997).
Samples for cations and anions were ﬁltered through 0.45 pm polyethersulfone
membrane ﬁlters and collected for analysis in the laboratory. Major cation samples were
preserved with ultra-pure “optimam” nitric acid and analyzed by ﬂame atomic absorption
spectrometry (Perkin Elmer). Samples collected for anions (Cl, Br, N03, and N02, 804)
and NH4+ analyses were preserved with formaldehyde and ﬂash frozen respectively, and

then analyzed by capillary electrophoresis (Hewlett Packard).

Statistical Analysis:

Factor analysis was performed using SYSTAT (SPSS Inc., 2000) on each

standardized dataset to simplify the complex relationships that exist among the observed

variables by revealing a reduced set of variables called factors. These factors are neither

35

directly observable nor expressed in terms of the original variables. These factors link
together seemingly unrelated variables to give additional insight into the underlying
processes operating in the system. In this study, the results of the groundwater chemical
analyses are the original variables and the dominant physical and biogeochemical
processes operating in the aquifer are the unobservable factors.

Our factor analysis technique involved four steps. The ﬁrst step was creating a
matrix of correlation coefﬁcients for all pairs of constituents. Results showed that
speciﬁc conductance was highly correlated to many other variables, and was thus
eliminated from the datasets prior to performing factor analysis. Measurements of Br,
N02, and 82' were also removed from the datasets due to their low/non-detectable
concentrations in the groundwater samples. The second step was estimating the factor
loadings. The principal components method was used as the parameter estimation
method to transform the observed variables into an orthogonal set of factors. However,
initial factor loadings commonly do not reveal the underlying structure of the observed
variables due to certain mathematical conditions, such as the variance properties of the
principal component (SUK and LEE, 1999). Therefore, the third step was linearly
transforming the initial set of loadings into a new set of loadings by factor rotation.
Although there are numerous rotation schemes, this study uses Kaiser’s scheme of
varimax orthogonal rotation, which yields the maximum variance of the squares of the
loadings (ABU-JABER et al., 1997). The ﬁnal step of the factor analysis was computing
the factor scores by a simple linear transformation on each of the factored variables. In
this study, the factor scores describe the importance of each of the resulting factors on the

chemistry of the wells.

36

Cluster analysis was then performed on the factor scores using JPM4 (SAS
Institute). Cluster analysis is an exploratory statistical tool used to divide the entire
groundwater system up into smaller groups or clusters. Theoretically, each group would
have similar relationships among chemical variables or a similar geochemical signature.
Factor scores were clustered instead of the measured variables to exclude unnecessary
and mutually dependant variables, which can obscure the underlying data structure.
Hierarchical clustering was performed using Ward’s method (WARD, 1963). These
clusters were then plotted in 3-Dimensions using SYSTAT to visualize the spatial pattern
of the clusters. To visualize the geochemical signature of the clusters, z-scores of the raw

data were computed and plotted using standard X-Y plots.

Results and Discussion

To examine the large-scale (plume) chemical changes initiated during a recharge
event, the average chemical concentration at the shallowest depth of all wells was
calculated in January and then again in June 2000 after a major recharge event.
Concentrations were averaged from the shallowest depths to avoid misinterpretations
based on examination of single-well measurements over time given the observed shifts in
groundwater ﬂow direction. During recharge events, groundwater head increases causing
the measured changes in the lateral direction of groundwater ﬂow. As a result of these
changes, it is often inappropriate to compare a well sampling point before and after
recharge since it may be impossible to distinguish the physical movement of a particular

water mass away from a sampling point and geochemical changes within a water mass

37

(MCGUIRE et al., 2000). By averaging chemical concentrations at the shallowest well
depths at all wells, the problem of laterally shifting water masses is reduced. This
analysis assumes that any changes in the average water chemistry are related to direct
recharge inputs and not related to the physical movement of water masses. The percent
change in average chemical values between January and June 2000 (before and aﬁer
recharge) is shown in Figure 2.4. The plot showed that after the recharge event,
temperature (T), Eh, dissolved oxygen (02), sulfate ($04) and nitrate (N 03) increased,
presumably due to the direct atmospheric inputs from precipitation. A decrease in pH
after the recharge event was coupled with increases in calcium (Ca), magnesium (Mg),
and alkalinity (primarily HCO3). This seems reasonable since the recharge water
typically has a much lower pH than the native groundwater. Thus, when mixed, recharge
would initiate dissolution of carbonate minerals and increase the Ca, Mg, and HC03
concentrations in solution. Observed increases in dissolved iron (F e”) and conductivity
were likely due to inputs from the solid phase. In the case of Fe2+, increases may be
caused by a rise in water table elevation allowing reduced water to come in contact with
previously oxidized sediments that contain Fe-oxyhydroxides coatings. These Fe (III)
rich minerals could then be available for reduction to F e2+coupled with the oxidation of
organic matter (TUCCILLo et al., 1999; VROBLESKY and CHAPPELLE, 1994). The
observed increase in Fe2+ would not occur in the presence of 02, which on average
increased after the recharge event. These seemingly contradictory observations
demonstrate that local chemical changes are important within the plume. The observed
increase in average conductivity values is in part due to the increase of solutes from

mineral dissolution but may also have contributions from organic contaminants desorbing

38

moo

 

 

 

Ame

 

 

goo

 

 

 

 

 

 

 

 

 

 

H2 [:1

 

-mo

THhOd agaKZLInnmmm4342
so- peom CMN a... wmmm
-Amo

 

 

Ems—d N3" 3802 035a 5 m<o8mo 03033 m” :5 mum—Esau" not? w” m: $65 6928: F35 m3
.38 ~08 A39. «normamov.

39

from the solid phase into the aqueous phase of the aquifer. Decreases in other solutes
such as sodium (Na), potassium (K), ammonium (NH4), bromide (Br), chloride (Cl), and
silica (SiOz) are likely due to dilution with recharge water, which is depleted with respect
to these constituents. The observed average decrease in dissolved hydrogen gas (H2)
concentrations may be due to both dilution and consumption. Dissolved hydrogen can
act as an electron donor coupled to the reduction of various TEAS (02, N03, Fe”, and
S04), which may have been introduced to the system during the recharge event. These
average percent change analyses give an estimation of the inputs to the aquifer during the
recharge event. However, this approach cannot be used to infer process and therefore
cannot distinguish dilution from consumption by TEAPS.

Factor analysis was used to assess changes initiated by recharge and attempt to
discern dilution versus biogeochemical reaction effects. Table 2.1 shows the rotated
factor pattern of the ﬁve extracted factors for the January and June 2000 sampling events.
Initial factor selection was based on the default eigenvalue criterion, which selects any
value greater than one. However, in both the January and June datasets, an additional
factor with an eignevalue very near one (0.980 and 0.983 respectively) was included to
help explain a greater percentage of the variability. Loadings on each of the factors were
interpreted such that any association greater than i075 was considered high loading with
respect to that factor, any associations between i0.5-0.7 5 were considered weaker but
still signiﬁcant loadings and any loading less than 10.5 was considered insigniﬁcant
(W AYLAND et al., in review). The sign designations suggest nothing about the strength of
the associations but rather give an indication of inverse relationships among variables if

loaded on the same factor.

40

Table 2.1: Rotated factor pattern of ﬁve extracted factors after varimax rotation in January and
June of 2000.

 

 

 

 

 

0.74

0.01

0.10

-0.03

January 00 Factor 1 Factor 2 Factor 3 Factor 4 Factor 5
T 0.05 -0.66 0.12 0.27 0.01
pH -0.86 0.01 -0.20 -0.02 0.01
Eh 0.52 0.17 0.08 -0.52 0.21
02 -0.23 0.21 0.12 -0.75 -0.12
Hz 0.79 -0.23 0.14 0.13 -0.22
Ca 0.86 0.38 -0.01 0.16 0.04
Mg 0.01 0.93 -0.07 -0.04 0.04
Na 0.11 -0.02 0.90 0.01 0.16
K 0.66 -0.23 0.44 -0.05 0.08
Fe” 0.80 -0.05 0.36 0.19 0.04
Alk 0.61 0.26 0.68 0.09 0.12
CI 0.04 -0.01 -0.08 0.02 -0.94
SC, -0.56 0.24 -0.43 -0.20 0.17
N03 -0.14 0.04 -0.16 .083 0.08
NH, 0.18 -0.23 0.73 -0.04 0.08
SiOz 0.25 -0.07 0.71 0.04 —0.18
June 00 Factor I Factor ll Factor III Factor IV Factor V

T -O.11 0.10 0.21 0.55 0.07
pH -0.48 -0.55 -0.20 0.15 -0.35
Eh 0.1 1 -0.79 0.26 -0.09 -0.31
02 -0.12 -0.91 -0.10 0.04 -0.10
Hz 0.11 0.18 0.14 -0.06 0.84
Ca 0.47 0.27 0.61 -0.13 0.32
Mg 0.06 -0.02 0.93 -0.07 -0.07
Na 0.89 -0.01 0.03 0.04 0.04
K 0.78 0.16 -0.04 -0.06 0.32
Fe2+ 0.80 0.22 0.13 002 0.39
Alk 0.74 0.21 0.51 0.01 0.28
CI 0.02 0.03 0.33 0.75 0.31
SO, -0.36 -0.09 -0.17 0.64 032
N03 -0.21 -0.82 -0.18 0.18 0.05
NH, 0.42 0.10 -0.14 0.06 0.65

-0.05

 

The January 2000 dataset (before recharge) was broken down into ﬁve factors (1 -
5). Factor 1 showed high positive loadings on Ca, Fe“, and H2, weaker positive loadings
on K and alkalinity, high negative loading on pH, and a weaker negative loading on 804.
These associations suggest active anaerobic microbial processes. In reduced systems,
anaerobic processes can allow intermediates such as H2 and various low molecular
weight organic acids, as well as end products such as Fe2+ to build up in the system
(CHAPPELLE, 2001). As a result of this activity, the pH of the system can be slightly
reduced resulting in increased dissolution of minerals such as carbonates and feldspars
(BENNETT et al., 1993). Factor 2 showed high positive loading on Mg and weak negative
loading on temperature. This factor suggests that Ca and Mg are not associated in the
system. The third factor had high positive loadings on Na, alkalinity, NH4, and SiO2.
These associations may suggest the chemical signature of water from bioreactor activities
performed at the site 2 years prior to the sampling events. Although the bulk of the
inﬂuenced water passed through the study area, as indicated by the return to background
Na concentrations, these associations suggest that these few down gradient wells may
still be experiencing a slight impact. The bioreactor activities stimulated microbial
activity by adding NaN03 to the aquifer system. Increased microbial activity resulted in
elevated values of NH, (biomass cycling and perhaps dissimilatory nitrate reduction to
NH,) and alkalinity (most likely due to increased production of organic acids) (MCGUIRE
et al., in prep). The fourth factor had high negative loadings on O2 and N03 and weaker
negative loadings on Eh. This signature suggests aerobic waters with lesser microbial

activity, likely due to low organic matter content (uncontaminated water). The ﬁnal

42

factor (5) had high negative loading with respect to C1. This is due to a single well with
much higher Cl concentrations than the others, since this factor drops out of the factor
analysis when this sample is removed. The elevated chloride concentration may be due
to localized dechlorination reactions. Unfortunately, since it is a single sample,
contamination cannot be ruled out.

The June 2000 sample set (after recharge) was also broken down into ﬁve factors
(I-V). F actor I showed high positive factor loadings on Na, K, and Fe“, weaker positive
factor loadings on alkalinity and SiO2 and weak negative loadings on pH. This signature
is interpreted to be highly reduced water impacted by recharge water. Prior to the
recharge event, the most reduced water had high concentrations of cations, alkalinity, and
H2. The post-recharge associations here are similar suggesting that the water experienced
highly anaerobic microbial activity but with one exception; high H2 values are no longer
associated with this group. Since high H2 concentrations are typically associated with
highly reduced waters, this suggests that the system has been “reset” microbially by the
introduction of more favorable TEAS (02, N03 and S04) during the recharge event.
Factor 11 was characterized by high negative loadings on 02, N03, Eh, and weaker
negative loadings on pH. These associations suggest aerobic water masses. It should be
noted that in January the aerobic factor was the fourth most important factor but in June
after the recharge event, the aerobic water signature was the second most important
factor. Factor HI has high positive loadings on Ca, Mg and alkalinity, which suggest the
dissolution of carbonate minerals. Interestingly, after the recharge event, Ca and Mg
have a higher degree of association than in the pre-recharge water. Factor IV has weak

positive loadings on C1 and S04, and weak negative loadings on temperature suggestive

43

of recharge water from snowmelt. Lower temperatures could have resulted as the cold
melt-water recharged and mixed with the aquifer water. Sulfate concentrations would
have increased due to atmospheric inputs of the precipitation (likely as snow). The C1 is
likely from the degradation of chlorinated solvents, which may have washed in with the
recharging melt water. The ﬁnal factor (V) showed a high positive loading on H2 and a
weaker positive loading of N11,. This factor likely indicates intense anaerobic microbial
activity. Dissolved hydrogen is an intermediate of microbially-mediated anaerobic
reactions which can build up in highly reduced systems and NH, would be present
primarily from the cycling of biomass cellular components. These ﬁndings suggest that
although many of the same processes are operating in the aquifer before and after the
recharge event, the relative importance of these processes have changed.

To explore the spatial distribution of process changes inferred by the factor
analysis, factor scores describing the relative importance of each of the above factors
were calculated at each individual well. Cluster analysis was then performed to group
wells together based on factor scores that were used to identify water masses with similar
biogeochemical signatures. Figure 2.5 (A&B) shows the resulting dendrograms for
January and June. In both datasets, the wells were broken down into 5 smaller clusters.
Since factors can be interpreted as processes, a plot identifying the dominant factor(s)
controlling each cluster was constructed for both January and June to interpret the
processes controlling water chemistry for each cluster (Figures 2.6 & 2.7). However, the
factors are limited in that they describe associations and not measured chemical

concentrations so these data were considered in conjunction with the average chemistry

January 2000

- M.1-21.5
= M.16-21

- ML11-21

= ML12-24
' M.14-24
' M.11-18
- M.15-24
- M.11-24
= M.12-21

- M.16-24
' M.11-27
' NL14-18
- M.14-21

= M_3-22.00
- EFX-310-24
. M.3-25.03

i: an '1") :
mi. in; 0.1.)

 

* .ii’ Elli-C. 'l .‘:'::» .ii 7
E F x. is l 512-: 7
" 5411-17.33 fl
1 3 :3 l s

- {tum—2;"
all- :3— 3. :2 5/
“ ML .5440" :‘f
° El ﬁnk) ll')--L£l.l
.. l? FIX. :5 1 err-7..)
. Ml- 1

° Ml a
.— ML5-22.64
F ML15-18

‘3 ML12-18

C? ML16-18

L" EFX-318~24
X NL5-25.67
x M.5-28.37
X M.5-33.78
-' 511114—2727

 

 

 

 

 

 

Figure 2.5A: Cluster dendrograms for J anuarv 2000 showing ﬁve clusters

45

June 2000

 

 

- M.1-20.00 ————~

- EFX-318-21 ———————l I F—
X M.1-24.50 »
X M.3-27.99 E»
X NL3-32.78 _, III
X M33759 ———~

>< EFX-310-30 —— l”
”3,2200 _ MW--. .1
:1 ML 55—2567 *ﬁl MM» "Fl
011,528.37 L...

r: ML14-18 r. l ~
ML1 5-18 — _. z -

i1 ML11-27
U ML5-22.64 , mm -~--~-» --——,
£012-18 4 IV ~—
:7 pr_310_24 ,. ___-._“_7 r

m ML 1 5-21 —« 2-.-...
m ML3-25.03 .
:2 MUS-24 J i
m ML16- 21 W_ "" (Wm
53 MLS~33.78 "H J

u EFX—310-27 m-.. NJ

. 1.112- £311:
‘ 3.1"... 12.7.1

I (.11.. 15122.5;

Hr aware:

11.1.1143? . II l‘ ”
' Mil 3.1.7

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

I, r 11 4 x')_ . 1-:

“ [1.1;le ~24

° 1.1.123-3‘?‘ . _
" 1.11.1213 \/
‘ l..ll.'l:T-~}_'7 . ‘

o 01114—277

° EH? I;

 

 

 

 

Figure 2.5B: Cluster dendrograrn for June 2000 showing ﬁve clusters

46

 

 

 

 

 

 

 

 

factor scores

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

z—scores

 

 

KalsOANwemmeachlsch>$oa~ws
111 II I

 

 

 

 

._. DI m: 00 In 0m 26 2m X man 2.» 0. m0; ZOwZIaQON

Ema—d Pa" ><anwmo 0:938: mama—.6308 Ba $92. 835%.. cm own: 0:58.. 5 35:5 Nooo.

47

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

A
S lilo—
mlv w o/ 2.2.. 0:
O // Isle:
W n lellQ<
0 \it... tire! ..o it. .. .olaoo .
t O l .... 1!: til, t, .1.
C ...... iﬂ.
.m -1
IN a a . 7
page“ _ mmaoq __ mmaoq ___ mmaoq 2 mean: <
b
w ..... O ..... Q Li.
S w . . .I. . . OH— 1....y’l/ll/011Fi3
e n +9: N
m n lilo? \5/ R\ \\
a +n< r l
m . . \w/K all -... .. 8V
F 0 $4» ‘5! 2 ,\ -.. 547d 'l-hﬂ, \. .
r if...“ .111. ........ . ..... ., . l... . I».-. ,.... V1,...
41. :11 A a i .84
a x/ : 2
~ K
N H

 

 

 

.253 u... m: 00 In On ,8 2m X man 2x 0. m9 20m 21¢ 90»

Ema—d v.4” ><o~mmo Oran—mom. manna—cage Ea @084 53me em one: 2:32. 5 :50 woos.

48

of each cluster. These clusters were then plotted in 3-dimensions to identify if a spatial
pattern would exist among the clustered wells and if so, determine the impact of various
factors (processes) on the resulting water masses. Figure2.8 depicts the distribution of
clusters in January 2000 prior to the recharge event. This plot shows that clusters of
wells with similar chemical associations also cluster together spatially forming distinct
water masses. Cluster 4 (C4) is primarily impacted by factor 1 which is characterized by
high H2, cations, alkalinity, and lower values of SO, and pH. This water is located in the
shallowest portion of the plume. This is the area of greatest contamination and enhanced
microbial activity. As microbial processes evolve, TEAs (02, F e“, N03, and S0,) are
depleted and dissolved hydrogen and Fe2+ concentrations build up. These processes
lower the pH of the water, allowing dissolution of carbonate and aluminosilicate minerals
to dissolve. Cluster 1 (C1) is located just below the water table interface and comprises
the largest area of the plume. Here the dominant factor is still factor 1 described above
but to a much lesser degree than in C4. In some wells within this cluster, factor 4,
characterized by higher concentrations of 02, N03, and higher Eh measurements, is also a
dominant factor. This suggests that these wells exhibit an overall less reduced redox state
than those in C4. Cluster 5 (C5) is the water mass that lies at the deepest sampling
points. This water mass’s most characteristic feature is a negative loading on factor 1.
This implies that this water is the least reduced of the three main water packets, which is
consistent with the deep, less contaminated, portion of the plume. Cluster 3 (C3) is most
inﬂuenced by Factor 3, which corresponds to remnant water from the bioreactor

bioremediation system. These wells are the only ones affected because they are the

49

0:588

 

Ema—d Na“ ub mega canvases c». 0.588 gang 0: @884 838 mdB H885 N25

50

farthest down gradient of the ﬁnal bioreactor circulation plot. Cluster 2 (C2) is not
necessarily a water mass. It is comprised of only one well point and characterized by
elevated Cl concentrations. As stated above, this observation may be
the result of sample contamination or may be the result of localized degradation of a
chlorinated solvent mass. It should be noted that factor 2 (high Mg, low T) and factor 4
(high 02, N03, Eh) are not dominant in any particular water mass and therefore have only
minor signiﬁcance at the plume scale.

The same assessment was repeated in June 2000, after the recharge event. Figure
2.9 shows that, again, cluster of wells with similar chemical associations also cluster
together spatially forming distinct water masses. However, the spatial distribution of the
resulting water masses is different. Cluster IV (CIV) lies primarily at the surface of the
plume but also has some wells located upgradient nearest the contaminant source area.
These wells are distinguished by higher values of H2 and NH,. Both of these parameters
indicate highly anaerobic microbial activity. It should be noted that the shallowest water
mass extends deeper than it did during the January sampling. This suggests that prior to
the recharge event (from January until May) the water likely became increasingly
reduced inﬂuencing a larger zone as anaerobic microorganisms operated without recharge
during the winter months. Vertical redox zonation has been observed at this site with the
most intense activity occurring near the water table interface (surface of plume)
(CHAPPELLE et al., 1996; MCGUIRE et al., 2000). However, the chemical signature of this
water mass is more complicated than the observed signature found in January. This
mixed chemical signature suggests that within this zone at certain localized points,

recharge has stimulated more energetically favorable microbial activity. However, on the

51

\

mac

 

 

 

 

\ =
N moo 2 2
<
m E . E
E m
0:588
O H
x a
Ammmmmmu E
2
<

 

ENE.» ”.9 we mvmaﬁ $3.358: ow 0:688 38a on 892. 828 mdB .38 woos

52

whole, the recharge event was not large enough to signiﬁcantly reset the system (i.e., the
bulk water mass still associates together as anaerobic, microbially active, reduced water).
The water mass directly beneath cluster IV is cluster 11 (C11). The CII signature suggests
that this portion of the plume is actually experiencing the most change as a result of the
introduction of recharge water. The most likely reason is that this portion of the plume
was not as deeply reduced so changes in chemistry may not have shifted microbial
processes as dramatically. The observed elevated C1 is likely from an inﬂux of
chlorinated solvents from the vadose zone as recharge water moved through the system.
Elevated SO, values and lower temperature are likely from the recharging melt water.
Cluster V (CV) is located at depth within the plume where lower values of Ca, Mg, and
alkalinity have been observed. This may be due to precipitation of carbonate minerals or
due to comparatively lesser mineral dissolution due to lesser microbial activity (and
therefore lower acid production). Cluster III (C111) is comprised of a small group of wells
located at various depths within the plume but never near the water table. Similarly (but
opposite) to cluster V, these wells have higher values of Ca, Mg, and alkalinity probably
due to the dissolution of carbonate minerals. Cluster 1 (CI) is most inﬂuenced by factor I
and occurs nearest the contaminant source (surface of plume and upgradient). These
samples were likely taken very near the water table where an inﬂux of contamination was
likely released to the plume during the recharge event. Therefore, this area likely
experienced the greatest increase in microbial activity resulting in lower pH values and
higher concentrations of most cations and alkalinity due to mineral dissolution. The lack
of H2 associations in this zone suggests that this may be a location where we are able to

observe the small interface where H2 is consumed coupled to the reduction of TEAS.

53

These ﬁndings suggest that although we know more energetically favorable TEAS
were introduced to the plume surface and that the relative importance of various
groundwater processes changed as a result of the recharge event, the system was not
microbially “reset” in large-scale redox zones within the plume. Rather, many of the
speciﬁc changes likely occur in more localized areas. To examine localized changes, a
single well (ML3) was examined at various depths to establish the vertical distribution of
chemical change. Figure 2.10 shows the percent chemical change with depth aﬁer the
recharge event at multi-level well 3. Since dissolved hydrogen gas is thought to be a
sensitive indicator of microbial process (LOVLEY and GOODWIN, 1988), this trend is of
great interest. At ﬁrst glace, it appears that shallowest depth (ML2-22.00) experienced
only a slight decrease in H2. However, since wells ML2-25, ML3-28, and ML3-33
experienced an overall increase in H2 during this period, it is reasonable to assume that
ML3-22 also experienced an increase and then a more signiﬁcant decrease during the
recharge event. This supports the ﬁndings of the factor analysis that the plume became
more reduced after January but prior to the recharge event. The observed H2 decrease at
ML3-22 and the suspected decrease at ML3-25 (lesser increase than the other wells) are
both likely due to a combination of dilution and consumption as an intermediate during
microbially mediated processes such as aerobic respiration and denitriﬁcation that may
have been initiated during the recharge event. The deepest monitoring point (ML3-3 8)
did not experience any change in H2 concentrations since historically, the concentration
has always been below the detection limit and the processes initiated by the recharge
event would only decrease the overall concentration. Interestingly, all depths

experienced a large percent increase in N03 but very low percent increases if at all in SO,

54

8
8

 

 

 

 

 

 

 

 

 

e o... .. 35B
gmmool ........_<_5.~m
m 88-1 +598 t.»
1% 58-1 2.58 > \m/
2578

m as IT \ / Q a
% So
a o.
D...

.80

-48o 4 u: N» on 003 I» am an 2» x mm» 2; 0. mo... 28 21s 20»

 

  

 

 

Ema..." ML? <38»— Emaccaos 03882: $888»: 085% m» m 358 <8: 9:3 #08 32.3 a: .38 ~08. .28 2552
£84 90 new: 888858 90 .83: a: 803 8:89 :8 maﬁa manna.

55

concentrations. Presumably, the chemical concentration of the recharge water
was uniform across the site and therefore must have contained dissolved SO, since the
percent change plots demonstrated that on a plume scale the SO, values increased. This
suggests that locally the SO, was consumed before it could reach the wells sampled,
possibly in a very narrow transition zone. These observations imply that although a zone
may be characterized as S0, reduction by H2 levels SO, may not be available for that

reaction to proceed.

Conclusions
Multivariate analyses were performed on groundwater chemistry samples

from a contaminated aquifer to characterize the biogeochemical changes initiated during
a recharge event. These changes were evaluated at a large (plume) scale and smaller
(single-well) scale. Large-scale chemical averages indicated that the inﬂux of recharge
water introduced more energetically favorable TEAS to the plume including 02, N03, and
S0,. The resulting rise in water table elevation also may have made available Fe (III)
oxyhydroxide coatings as a TEA. Based on the average inﬂux of TEAS, one would
expect to observe a shift in dominant anaerobic groundwater microbial processes to more
energetically favorable reactions and thus observe changes in redox zonation. However,
further data examination by means of factor analysis, cluster analysis, and graphical
techniques revealed that when viewed in greater detail, the scale at which changes are
initiated by recharge must be considered.

Factor analysis indicated that before and after the recharge event, the dominant

chemical associations were very Similar. However, the relative importance of each of

56

 

these processes was different after the recharge event. Therefore to visualize how these
changes in groundwater chemistry/process manifested themselves spatially in the aquifer,
cluster analysis was performed on the individual factor scores for all wells. The results of
the cluster analysis were then plotted in 3-D space to obtain a visual representation of
chemical change. These analyses demonstrated that wells with similar chemistry were
spatially distributed such that distinct water masses were observed. Comparison of these
resulting water masses before and after recharge indicated that the water masses were
altered as a result of recharge but not by initiating changes in dominant redox zones.
Rather, reactions seem to be initiated in localized areas, likely at the interface between
the reduced water and the capillary fringe zone. Evaluation of chemical changes at a
single well provided further evidence that many TEAP reactions initiated by recharge
events occur in small interface zones and not in large redox zones as we traditionally
envision.

These analyses suggest that recharge events can initiate changes in TEAPS in
anaerobic groundwater systems by introducing more energetically favorable TEAS to the
system. When coupled to organic electron donors, these more energetically favorable
reactions can lead to enhanced microbial degradation of organic contaminants,
particularly in the capillary ﬁinge zone where many organic contaminants concentrate.
However, since the zone over which these reactions occur is likely very small it is very
difﬁcult to directly measure these processes. Nonetheless, recharge effects in this
organic-rich zone may signiﬁcantly increase the rates of natural attenuation and

bioremediation. Therefore, to be able to evaluate and design the most effective

57

bioremediation or natural attenuation strategies, the importance of small-scale spatial

changes should be considered.

58

References

Abu-Jaber N. S., El Aloosy A. S., and Ali A. J. (1997) Determination of aquifer
susceptibility to pollution using statistical analysis. Environmental Geology 31.

Barcelona M. J ., Hohn T. R., Schock M. R., and George G. K. (1989) Spatial and
temporal gradients in aquifer oxidation-reduction conditions. Water Resources
Research 25(No. 5), 991-1003.

Bennett P. C., Siegel D. E., Baedecker M. J ., and Hult M. F. (1993) Crude oil in a
shallow sand and gravel aquifer-I. Hydrogeology and inorganic chemistry.
Applied Geochemistry 8, 529-549.

Champ D. R., Gulens J ., and Jackson R. E. (1979) Oxidation-reduction sequences in
groundwater ﬂow systems. Canadian Journal of Earth Sciences 127, 85-108.

Chappelle F. H. (2001) Ground- Water Microbiology and Geochemistry. John Wiley and
Sons, Inc.

Chappelle F. H., Haack S. K., Adriaens P., Henry M., and Bradley P. M. (1996)
Comparison of Eh and H2 Measurements for Delineating Redox Processes in a
Contaminated Aquifer. Environmental Science and Technology 30(No. 12), 3565-
3569.

Chappelle F. H. and McMahon P. B. (1991) Geochemistry of Dissolved Carbon in a
Coastal Plain Aquifer. Journal of Hydrology 127, 85-108.

Crill P. M. and Martens C. S. (1987) Biogeochemical cycling in an organic-rich coastal
marine basin. 6. Temporal ann spatial variations in sulfate reduction rates.

Geochimica et Cosmochimica Acta 51, 1175-1186.

59

Dojka M. A., Hugenholtz P., Haack S. K., and Pace N. (1998) Microbial divesity in a
hydrocarbon- and chlorinated solvent-contaminated aquifer undergoing intrinsic
bioremediation. Applied and Environmental Microbiology 64, 3869-3877.

Drever J. I. (1997) The Geochemistry of Natural Waters: Surface and Groundwater
Environments. Prentice Hall.

Haack S. K. and Bekins B. A. (2000) Microbial populations in contaminant plumes.
Hydrogeology Journal 8, 63-76.

Kennedy L. G., Everett J. W., Dewers T., Pickins W., and Edwards D. (1999)
Application of mineral iron and sulﬁde analysis to evaluate natural attenuation at
a fuel contaminated Site. Journal of Environmental Engineering 125, 47-56.

Lee J .-Y., Cheon J .-Y., Lee K.-K., Lee S.-Y., and Lee M.-H. (2001) Statistical evaluation
of geochemical parameter distribution in ground water system contaminated with
petroleum hydrocarbons. Journal of Environmental Quality 30, 1548-1563.

Lovley D. R. (1991) Dissimilatory Fe(III) and Mn (IV) reduction. Microbial Reviews 55,
259—287.

Lovley D. R. and Goodwin S. (1988) Hydrogen concentrations as an indicator of the
predominant terminal electron-accepting reactions in aquatic sediments.
Geochimica et Cosmochimica Acta 52, 2993-3003.

McGuire J. T., Madden A. S., Velbel M. A., Haack S. K., Long D. T., and Klug M. (in
prep) Episodic fossilization of microbes on an annual time scale in an
anthropogenically modiﬁed natural environment. Astrobiology.

McGuire J. T., Smith E. W., Long D. T., Hyndman D. W., Haack S. K., Klug M. J ., and

Velbel M. A. (2000) Temporal variations in parameters reﬂecting terminal-

60

electron-accepting processes in an aquifer contaminated with waste fuel and
chlorinated solvents. Chemical Geology 169, 471-485.

Reynolds L. A. and Haack S. K. (1999) Evaluation of RNA hybridization to assess
bacterial population dynamics at natural attenuation sites. Morganwalp DW,
Buxton HT (eds) US Geological Survey toxic substances hydrology program.
Proc Technical Meeting, Charleston, South Carolina, 8-12 March 1999. US
Geological Surv Water-Resour Invest Rep 99-4018C, 635-639.

Ritzi R. W., Wright S. L., Mann B., and Chen M. ( 1993) Analysis of temporal variability
in hydrogeochemical data used for multivariate analyses. Ground Water 31, 221-
229.

Stumm W. and Morgan J. J. (1981) Aquatic Chemistry. John Wiley and Sons.

Suk H. and Lee K.-K. (1999) Characterization of a Ground Water Hydrochemical System
Through Multivariate Analysis: Clustering into Ground Water Zones. Ground
Water 37(3), 358-366.

Swanson S. K., Bahr J. M., Schwar M. T., and Potter K. W. (2001) Two-way cluster
analysis of geochemical data to constrain spring source data. Chemical Geology
179, 73-91.

Tuccillo M. E., Cozzarelli l. M., and Herman J. S. (1999) Iron reduction in the sediments
of a hydrocarbon-contaminated aquifer. Applied Geochemistry 14, 655-667.

USGS. (1991) Installations Restoration Program: Phase II-Conﬁrmation/Quantiﬁcation
Stage 1: Wurtsmith Air Force Base, Michigan: Investigations of groundwater

and soil contamination at selected sites.

61

Vroblesky D. A. and Chappelle F. H. (1994) Temporal and spatial changes of terminal
electron—accepting processes in a petroleum hydrocarbon-contarninated aquifer
and the Signiﬁcance for contaminant biodegradation. Water Resources Research
30(No. 5), 1561-1570.

Ward J. H. (1963) Hierarchical grouping to optimize an onjective function. Journal of the
American Statistical Association 58(301), 236-244.

Wayland K., Long D. T., Hyndman D. W., Pijanowski B., Woodharns S. M., and Haack
S. K. (in review) Identifying relationships between baseﬂow chemistry and land

use with synoptic sampling and r-mode factor analysis. Journal of Environmental

Quality.

62

CHAPTER III.

McGuire, J .T., Long, D.T., Klug, M.J., Haack, S.K., Hyndman, D.W., accepted
March, 2002. Evaluating the behavior of oxygen, nitrate, and sulfate during

recharge and quantifying reduction rates in a contaminated aquifer. Environmental
Science and Technology.

63

Abstract

This study evaluates the biogeochemical changes that occur when recharge water comes
in contact with a reduced aquifer. It speciﬁcally addresses: 1) which reactions occur in-
Situ; 2) the order these reactions will occur if terminal electron acceptors (TEAS) are
introduced Simultaneously; 3) the rates of these reactions; and 4) the roles of the aqueous
and solid phase portions of the aquifer. Recharge events of waters containing various
combinations of 02, N03, and SO, were simulated at a Shallow sandy aquifer
contaminated with waste fuels and chlorinated solvents using modiﬁed push-pull tests to
quantify rates. In-situ rate constants for aerobic respiration (14.4 day ’1), denitriﬁcation
(5.04-7.44 day”) and sulfate reduction (4.32-6.48 day”) were estimated. Results show
that when introduced together, N03 and SO, can be consumed simultaneously at similar
rates. To distinguish the role of aqueous phase from that of the solid phase of the aquifer,
groundwater was extracted, amended with N03 and SO, and monitored over time.
Results indicate that neither N03 nor S0, was reduced during the course of the aqueous
phase study suggesting that NO3 and SO, can behave conservatively in highly reduced
water. It is clear that sediments and their associated microbial communities are important

in driving redox reactions.

Introduction

Identifying the biogeochemical processes responsible for, the development of and
transition between, redox zones and quantifying the rates of these processes is crucial to

understanding how many natural and anthropogenic groundwater constituents behave in

64

the environment. Understanding the dynamics associated with the redox state of a system
is important since redox controls the form, mobility, and persistence of many
groundwater contaminants. Quantifying these dynamics will allow for 1) improved
understanding of natural ﬂuctuations of redox processes in pristine and contaminated
aquifers; 2) the prediction of rates of natural microbial attenuation of contaminants; and
3) proper design and evaluation of monitored natural attenuation and biostimulation
remediation strategies in contaminated aquifers. Although recharge events are thought to
be a signiﬁcant drive for redox dynamics (1 ), their exact role is not well understood. Few
studies have looked at the introduction of recharge solutions to reduced systems and none
have examined the effects of introducing a multi-parameter recharge solution as one
might expect during a rainfall event. In addition, the roles of aqueous and solid phase
portions of an aquifer, and their associated microbial communities, on redox reactions
have not been evaluated. To further complicate our understanding of the controls on
redox dynamics, laboratory and ﬁeld measurements of redox reaction rates are often
signiﬁcantly different. This study seeks to quantify the redox reactions associated with
the aqueous and solid phase portions of a reduced aquifer during a multi-parameter

recharge event.

In pristine and contaminated aqueous systems, redox state is based on the activity
of microorganisms, which ﬁrst consume oxygen, and then a succession of alternate
terminal electron acceptors to support their grth using a variety of carbon sources (2-
5). The succession of terminal electron accepting processes (TEAPS) in order of
decreasing redox potential and free energy yield is generally: oxygen reduction, nitrate

reduction, manganese reduction, iron reduction, sulfate reduction, and methanogenesis.

65

In most closed, uncontaminated groundwater systems, the sequence of TEAPS follows
this succession (i.e., increasingly reduced conditions occur further down the ﬂow path)
(6). Downgradient of contaminant plumes, the sequence of TEAPS is reversed (2,3). That
is, the most reduced conditions (e.g., methanogenesis) occur near the contaminant source
due to the greater availability of electron donors, while less reducing conditions (e.g.,
iron reduction) occur down ﬂow path from the contaminant source. Since hydrocarbons
(which serve as electron donors) concentrate in the capillary ﬁinge area, more reducing
conditions often occur near the water-table interface resulting in a plume that has vertical
as well as horizontal redox zonation (7). In reduced systems, it is thought that recharge
events can initiate changes in TEAPS by delivering more favorable terminal electron
acceptors (TEAS) such as oxygen (02), nitrate (N03) and sulfate (S0,) to the system (1 ).
Thermodynamics would predict that when a recharge solution comes in contact with a
reduced aquifer TEAS would be consumed in the order of free energy yield. Although
thermodynamics (by evaluation of available free energies) can provide insight into the
potential metabolic attractiveness of a reaction, the actual rate and extent of any given
redox process is a function of other factors such as activation energy, metabolic kinetics,
and the presence of biotic and abiotic catalysts. This suggests that unless competitive
pressure exists (e.g., electron donor limitation), there is no reason to assume that TEAS
would be consumed in the order of free energy yield. In addition, it is unclear if other
biotic and abiotic redox reactions involving TEAS may occur when recharge water comes

in contact with reduced groundwater and its associated sediments.

To understand the biogeochemical processes that dominate during recharge

events, it is necessary to not only identify the redox reactions which may be occurring but

66

also to quantify these reaction rates. The determination of rates of TEA consumption is
challenging. Many studies use laboratory-based methods including microcosm
experiments (8-11), direct observations (radiotracers) (12,13), molecular methods (14),
geochemical modeling (5) and kinetic modeling (15). These studies have been performed
on sediments from a wide range of aqueous environments including aquifers, lakes, and
wetlands. However, these approaches can be limited by possible system disturbances,
contamination problems, and oﬁen unnaturally high concentrations of substrate. In-situ
methods to estimate rates are generally considered to be more representative of actual
subsurface conditions. These methods are not as widely used and include natural

gradient tracer experiments (16), ﬂowline estimations (I 7), and push-pull tests (18).

In general, laboratory rate-estimation methods commonly yield zeroth order rates
while in-situ ﬁeld rate estimation methods commonly yield ﬁrst order rates associated
with the TEA concentration. None of the previous laboratory or ﬁeld studies has
examined the localized in-situ TEA consumption rates that would result when natural
recharge water, containing multiple TEAS at relatively low levels, is introduced to a
reduced groundwater environment. One study by Cunningham at al(19), added multiple
TEAS (NO3 and S0,) to evaluate possible enhanced biodegradation over longer timer
scales (months) in a contaminant plume at Seal Beach, California. They found that
although N03 was utilized preferentially over $04, the two processes were not strictly
sequential. This study utilizes a relatively new technique, in-situ push-pull tests, to

examine in detail the localized processes initiated during recharge events.

Push-pull tests have been shown to be useful in obtaining quantitative information

on a wide range of in-situ microbial processes in aquifers (20,21) and lake sediments (22).

67

A test consists of three phases. In the ﬁrst phase, native groundwater is extracted from
the well for preliminary geochemical characterization. This information is necessary to
deduce which biogeochemical reactions may already be occurring in the aquifer and to
establish background concentrations of the chemicals of interest. Speciﬁcally, we wanted
to ensure that the groundwater where the tests were performed was reduced. Eh
measurements ranged from -150 to —200 mV but due to the lack of Speciﬁcity of Eh
measurements, we chose to rely on concentrations of dissolved hydrogen (H2) and other
selected TEAP variables such as 02, Fe“, N03, N02, and S0,. The H2 concentration in
aquatic systems has been proposed as an indicator of in-situ TEAPS (5). Dissolved
hydrogen is known to be a ﬂeeting but important intermediate in the decomposition of
organic matter. Dissolved hydrogen is produced and consumed in anaerobic
environments such that each TEAP maintains a characteristic H2 concentration (23).
Interpretations of TEAPS from H2 concentrations, as measured in marine and freshwater
sediments and in aquifers, are as follows: methanogenesis, greater than 5 nM; SO,
reduction, 1-4 nM; F e3 + reduction, 0.2-0.6 nM; and Mn (IV) or nitrate reduction, less than
0.05-0.1 nM (5). Chapelle et al. (7,24,25) demonstrated the use of H2 concentrations in
delineating redox zonation in the plume at the study Site described in this paper. They
concluded that delineating redox zones based on H2 concentrations is more reliable when
H2 concentration are interpreted in the context of other TEAP variables, a strategy we
have adopted here. After the geochemical characterization has been completed to
determine initial geochemical condition, the remaining phases of the test can begin. In
the second phase, or “push”, a well-mixed slug of water containing conservative and

reactive solutes is rapidly injected into the saturated zone of the aquifer. A conservative

68

 

solute is added as a tracer to account for advection and dispersion and the other added
solute(s) are reactive. The third phase, the “pull”, begins immediately after the injection
is completed. Water is Slowly extracted from the well and the concentrations of the

solutes are measured over time.

These tests can provide valuable insight into the redox reactions that may be
occurring in-situ, but they are limited in that they cannot distinguish the roles of the
aqueous and solid phases, and their associated microbiological population in these
reactions. The most widely accepted hypothesis for the inﬂuence of microbes on redox
processes in aquifers is that microbes residing on the solid phase portion of the aquifer
are responsible for the majority of the activity (26). This is thought to be due to the much
greater concentration of microorganisms in association with the solid phase than in the
aqueous phase of the aquifer. However, it is unclear what the role of the aqueous phase
is during redox reactions. Based on thermodynamics, abiotic geochemical reactions may
occur when oxidized chemical species come in contact with a reduced environment. To
determine the signiﬁcance of the abiotic and biotic redox reactions associated with the
aqueous phase of the aquifer, we designed surface reaction vessel (SRV) experiments to
run simultaneously with the push-pull tests. The goal of the SRV experiments was to
determine what reactions would occur in the aqueous phase and at what rates. Identical
control SRVS were amended with formaldehyde and run simultaneously to distinguish
microbially-mediated redox reactions from geochemically-driven reactions. The SRVS
were ﬁlled with water from the aquifer where the push-pull test was conducted and then

allowed to react over time to the same addition of solutes as introduced in the push-pull

69

tests. The vessels were designed to maintain aquifer conditions as closely as possible

including maintaining the temperature and anaerobic nature of the water.

This study evaluates the biogeochemical changes that occur when a recharge
solution comes in contact with reduced groundwater and sediments. It speciﬁcally
addresses: 1) which reactions occur in-situ; 2) the order in which these reactions will
occur if TEAS are introduced simultaneously; 3) the rates at which these reactions occur;
and 4) the roles of the aqueous and solid phase portions of the aquifer in these processes.
Push-pull tests and surface reaction vessel experiments containing various combinations
of solutes to simulate various types of recharge events were designed to address these

questions.
Study Site Description

The study Site is a Shallow‘ sandy aquifer contaminated with petroleum
hydrocarbons and chlorinated solvents at the former Wurtsmith Air Force Base (AFB),
Michigan, U.S.A. Wurtsmith AFB lies on a sandy plain of glacial lake sediment
approximately 2 km west of the Lake Huron Shoreline in Michigan's Lower Peninsula.
The aquifer is approximately 20 meters thick and composed of highly permeable
alternating aeolian sands and glacial outwash material. Hydraulic conductivities of these
deposits are on the order of 30 m/day. The unconﬁned aquifer is underlain by at least
30.5 meters of silty clay. The water table varies from 3.5-5.0 m below land surface (184-
186 m above sea level) and ﬂuctuates 0.3 to 0.7 m annually. The water table gradient
ranges from 3 to 5 meters per kilometer. Assuming an effective porosity of thirty

percent, average groundwater velocity is approximately 0.5 m per day. Because the

70

direction of groundwater ﬂow is inﬂuenced by recharge, lateral shifts in the plume

dire

 

 

 

 

 

0 $0 METERS

0 [00 FEET

EXP LAN ATION

rm AREA OF CONTAMINATION

ROAD

_ 606 — WATERTAllE CONTOUR-Shows
elevation of water table in feet. Datum
issea level.

0 WLTI-IEVEL WELL

A WATER LEVEL RECORDER

Figure 3.1 Plume map for the Fire-Training Area 2 (FTA-

 

 

2) at the former Wurtsmith Air Force Base, Michigan. The

concrete pad was the Site of ﬁre-training activities.

Fire training

activities performed
by the Air Force from
1952 to 1986 are the
source of the
contamination at this
site. Various
compounds were used
during the training
exercises to ignite and
extinguish the ﬁres,
including waste fuels,
chlorinated

solvents,

aqueous ﬁlm-forming

foams, a multipurpose dry chemical, potassium bicarbonate based soda, and Halon 1211.

The majority of the contamination resides within the capillary ﬁinge and an extensive

plume approximately 50 m wide and 400 m long has developed (Figure 1). The plume is

complex in that it contains both BTEX compounds (benzene, toluene, ethylbenzene, and

xylenes) and chlorinated solvents (e. g., dichloroethene).

Concentrations of BTEX

compounds in the plume range from 20 to 1,000 pg/IIICI' while concentrations of

dichloroethane, chloroethane, and vinyl chloride range from 2 to 1,000 pg/liter. Most of

71

the contamination is associated with aquifer solids between 4.5 and 5.7 m below land

surface, which have an average concentration of total hydrocarbons of 13,650 mg/kg (2 7).

The portion of the aquifer where the push-pull experiments were performed,
multi-level well 3 at the 25 ft (7.6 m) depth (ML3-25), remains in a reduced redox state
throughout all groundwater ﬂow regimes. The reducing conditions at this well vary
historically from Fe (II) reducing to methanogenic (as determined by H2 concentrations)
on the time scale of months (28). The subject well was constructed of 2.5 cm inner
diameter PVC casing, with 0.3 m long, 6.35 mm slotted PVC screens. Vertical spacing
between screens is approximately 1 m. This is an ideal site to examine the effects of
recharge on reduced systems since a detailed spatial and temporal geochemical,

hydrogeologic and microbiological framework for the Site has been established {28-31).
Methods
Push-Pull Tests:

Well selection was based on the initial complete geochemical characterization
including redox-indicating parameters. A summary of the initial water chemistry from
the selected wells is presented in Table 1. All activities were performed at a well with
highly reduced water (ML3-25) located roughly in the center of the plume (Figure 1).
Three push-pull tests Were performed at this well at approximately l-month intervals.
To simulate recharge ﬂuids for the push-pull tests, native uncontaminated groundwater
from a well just outside the plume, multi-level well 10 at the 37 ft (11.3 m) depth (MLlO-
37), was extracted and placed in a 20-liter Nalgene carboy. This ﬂuid was chosen to

represent a rain solution which had percolated through the unsaturated zone and

72

ultimately reached the aquifer. The solution was then amended approximately 10 mg/L
of the TEAS which would likely be introduced during a recharge event. Three different
types of recharge waters were prepared. The ﬁrst test solution represented the water
resulting from a rain event containing 02, N03, and S0,. The second test solution
represented water which had been depleted in 02, perhaps upgradient, but still contained
N03 and 80,. The third test solution represented recharge water depleted in 02 and N03,
but still containing S0,. To each of these solutions, 100 mg/L chloride (as NaCl) was
added as a conservative tracer to account for dilution from mixing, advection, and
dispersion. Nitrate and sulfate were added as NaNO; and Na2SO, respectively and hand
shaken to ensure complete dissolution and homogenization of the solution. Oxygen was
removed from test solutions two and three by stripping with inert N2 gas for
approximately 3.5 hours until ﬁeld test kits could no longer detect 02 (less than 0.03
ppm). To maintain anaerobic conditions, solutions were kept in an N2 atmosphere and
monitored for 02. Before injecting these solutions into the test well, the unscreened
interval was packed off to prevent the vertical movement of water up the well casing.
The packer was inserted to just above the well screen and inﬂated with inert N2 gas to
seal off the well. The packer remained in place throughout the test. Once the packer was
in place, the anaerobic test solution was “pushed” down the well at a rate of
approximately 1 L/min. This was followed by the injection of 2 liters of “chaser
solution” (deoxygenated, distilled deionized water) to ﬂush the test solution from the well
and force it ﬁrrther into the aquifer. The extraction or “pull” phase began immediately
after the injection phase was completed. Efﬂuent from the well was slowly extracted at a

rate of approximately 150 m1/min and monitored in the ﬁeld for Cl, 02, N03, N02, 82',

73

80,, Fe2+, temperature, pH, and Eh over time. Samples were also collected for anions

(C1, N03, N02, S0,) and NH,+ for analysis in the laboratory.
Surface Reaction Vessel Tests:

Water was pumped from the reduced well where the push-pull tests were
performed (ML3-25) into a 20L Nalgene carboy attached to a glove bag ﬁlled with N2 to
maintain anaerobic conditions. The carboy and attached glove bag were then placed into
a temperature-controlled environment (surface reaction vessel) and amended with various
TEAS. Control surface reaction vessel experiments were set up in the same manner but
inoculated with formaldehyde (to make a ﬁnal 0.4% solution) to halt microbial activity.
The water was then monitored at approximately 5-minute intervals for the duration of the
push-pull tests. Real-time measurements of 02, N03, N02, Sz’, S0,, and F e2+ were taken
in the ﬁeld. Samples were also collected for anions (Cl, N03, N02, 80,) and NH,+ for

later analysis in the laboratory.
Geochemical Analyses:

A ﬂow cell (Yellow Springs Instruments) was used to continuously monitor
temperature, pH, dissolved oxygen, Eh, and Speciﬁc conductance throughout the push-
pull tests. Dissolved H2 gas samples were collected using the bubble-stripping method
(17) and analyzed immediately in the ﬁeld on a RGA3 reduction gas analyzer (Trace
Analytical). In the ﬁeld, N03, N02, and 82', were measured by colorimetric spectroscopy
using Chemetrics Vacu-Vialsm, Fe2+ by phenanthroline colorimetric analysis, and S0, by
turbidimetric analysis using HACH AcuvacsTM. Samples were collected for anions (Cl,

N03, N02, 80,) and NH,+ analyses in the laboratory. These were preserved with

74

formaldehyde and a ﬂash-freezing technique respectively, and then analyzed by capillary

electrophoresis.
Rate Determination:

A ﬁrst-order rate equation was a better ﬁt to the experimental data than the zeroth
order rate equation. When the experimental data was plotted over time it formed a curve.
Plotting the natural log of the data over time yielded a straighter portion of the curve.
Therefore, rates were estimated assuming a ﬁrst-order process. We followed a method
established by Snodgrass and Kitanidis (32) for inferring in-situ reaction rates from push-
pull experiments. This analysis assumes the absence of sorption and negligible
background concentrations. In our case these are valid assumptions since N03 and S0,
are soluble and relatively non-sorbing and background concentrations were below
detection limits. This analysis also assumes the solutes were injected instantaneously as a
well-mixed slug. It is reasonable to assume that the injected ﬂuids were well mixed since
the test solution was well homogenized prior to injection. Our injection was not
instantaneous but this analysis assrnnes that the rapid injection was negligible compared
to the total reaction time. Our approach was modeled after single solute in-situ push-pull
tests performed by Haggerty et a1. (33) and Schroth et al. (20). We assumed that the

injected solutes were transformed within the aquifer according to a ﬁrst-order rate

equation ers/a: -kC, so that:

o—kt

C.(‘)=C.e (1)

where C,(t) is the reactive solute concentration at time t, C,” is the reactive solute

concentration at time r=0 and k [t"] is the rate coefﬁcient. Now assuming that changes in

75

saturated thickness due to injection and extraction pumping are negligible, the

breakthrough curve for a reactive solute C,(t *) is given by:

C (t at): w[e-kr _e-k(r..,-t‘)] (2)
' kt...-

where t* is the time elapsed Since the end of the test solution injection, C,,(t*) is the
breakthrough curve for a co-injected tracer (C1 in our case), and t,-,,,- is the duration of the
test solution injection. Equation 2 can be rewritten:

14%] = 1.1[1—‘5—J— k: * (3)

tr w!

This suggests that a plot of ln(C,/C,,) versus t* generates a straight line with a Slope of —-
k. Since the determination of k is based on the evaluation of ratios of C/Cm complete
tracer mass recovery is not necessary during tracer tests to obtain accurate estimates of k.
Similarly, a portion of the breakthrough curve may be used to estimate k. A non-linear
least squares regression was used to ﬁt equation 3 to experimental breakthrough data to
obtain ﬁrst order rate coefﬁcients for aerobic respiration, denitriﬁcation, and sulfate
reduction. For this purpose, C, in equation 3 is the relative concentration (i.e., the
measured concentration divided by the concentration injected in test solution) of reactive
solutes (02, N03, or 80,) and C, is the relative Cl concentration. Since the background
concentration of the reactive solutes was negligible but the background concentration of
the tracer (Cl) was detectable, 1.0-3.5 ppm depending on the test, the measured values of
chloride could not be used as a reliable indicator of the degree of mixing without

correction. Therefore, C,, was calculated as a mixing proportion between the injected Cl

76

concentration and the background chloride concentration (equation 4) to give a more

accurate description of the degree of mixing and other non-reactive processes.

 

_ C1,, 0*) — C1,
Ctr -[ X j (4)

Where Cl", (t*) is the measured chloride concentration at time t, Cl, is the measured

background chloride concentration, and X is the slope of the line generated from a plot of
the percent input solution (0-100%) versus concentration (background-injection values).
This line represents the mixing curve between the injected solution and the background
water with respect to chloride concentration. In our case, since the background
concentration was very low (< 3.5 ppm) compared to the input test solution concentration

(~100 ppm), the slope was near 1.

Results and Discussion
Field vs. Laboratory N03 Measurements:

A signiﬁcant discrepancy was observed between the N03 data obtained by ﬁeld
analyses and the N03 data obtained by laboratory analyses. Field data obtained ﬁom
samples collected within the plume gave much lower (often not detectable)
concentrations of N03 than the corresponding laboratory analyses, which accurately
represented the added initial concentrations. Field data obtained from samples collected
from outside the contaminant plume largely agreed with corresponding laboratory
analyses for N03. This suggests that some constituent of the contamination within the

plume interfered with the analyses. This ﬁeld observation was supported by laboratory

77

experiments, which showed that when a known amount of N03 was added to the
uncontaminated samples (collected outside the contaminant plume), the ﬁeld method
accurately represented the resulting concentration. However, when a known amount of
N03 was added to the contaminated samples, the ﬁeld method gave erroneously low and
inconsistent readings. One possible explanation for this discrepancy is that the N03 ﬁeld
method requires cadmium to reduce N03 to N02, which is then measured
colorimetrically. If the cadmium surfaces were coated with organic contaminants from
the plume, they may have been unavailable to reduce the N03 and as a result were not
able to be measured. For this reason, the N03 data presented in this paper will be from
laboratory analyses only. Based on these observations, we conclude that caution Should
be used when interpreting N03 data from Cd-reduction ﬁeld methods in contaminant

plumes.

Table 3.1 Initial Chemical Characterization of Selected Wells (mg/L)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Well ID Date H; 02 N03 N02 Fe? so, BTEXavg DCE
ML3-25 6/00 3.55 <05 <0.05 <0.05 7.12 <0.05 17205 5.6
ML10-37 6/00 <0.05 8.01 19.36 <0.05 <0.01 10.7 <09 <02
ML3-25 7/00 2.68 <05 <0.05 <0.05 13.7 <0.05 18-328 <02
ML10-37 7/00 <0.05 6.59 4.3 <0.05 <0.01 6.99 <09 <02
ML3-25 8/00 2.35 <05 0.41 <0.05 7.00 <0.05 9314 4.87
ML10-37 8/00 <0.05 6.43 7.13 <0.05 <0.01 3.57 <09 <02

 

DCE: cis 1,2 dichloroethene; reported in ug/L

BTEX: average benzene, toluene, ethylene, and xylene; reported in ug/L

* H2 concentrations are reported in nM

78

 

Push-Pull Tests:

The ﬁrst push-pull test (PPONS) evaluated oxygen, nitrate, and sulfate
consumption rates at well ML3-25 (June 2000). Phase 1, the geochemical
characterization, revealed that the aquifer water from that well was reduced (Table l).
PPONS involved the injection of a test solution containing approximately 100 mg/L Cl (as
a conservative tracer), 6 mg/L O2, 10 mg/L N03, and 10 mg/L S0,. Concentrations of
solutes were measured over time (Figure 3.2). AS described in the rate determination
section, these concentrations were then normalized and divided by the concentration of
conservative tracer (Cl) at that point in time to account for non-reactive processes such as
advection and dispersion. Rates were then derived from a plot of this ratio versus time
(Figure 3). Dissolved oxygen was consumed ﬁrst at a rate of approximately 0.60 hr'l. In
an anaerobic system with excess dissolved iron, oxygen can rapidly be consumed through
the abiotic oxidation of F e(II). After the 02 was consumed, N03 and SO, were consumed
simultaneously at rates of 0.21 hr'1 and 0.27 hr'l respectively. The observed lag time for
N03 and SO, reduction occurs because these reactions are biologically mediated.
Microbiological response to system change frequently shows a lag time for microbes to
adjust to new conditions (3). In addition, our simulated recharge water did not contain
electron donors, which may also have contributed to the observed lag phase. The
electron donor would have then been supplied through desorption from the solid phase
and mixing with the aquifer water. PPNS evaluated nitrate and sulfate consumption rates

at well ML3-25 (August 2000). Phase 1, the geochemical characterization, revealed that

79

the aquifer water was reduced (Table l). PPNS involved the injection of a test solution
containing approximately 100 mg/L Cl (as a conservative tracer), 10 mg/L N03, and 10
mg/L 80,. N03 and SO, were again consumed simultaneously at rates of 0.31 hr'1 and
0.18 hr'1 respectively (Figure 3.3). PPS evaluated the rate of sulfate reduction at well
ML3-25 (July 2000). Phase 1, the geochemical characterization, revealed that the well

was reduced (Table 1). PPS involved the injection of a test solution containing

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

PPONS Raw Data 10 x . ”20
41‘ .02 A30 8 1T 0 15
ED >KN03 eCl 5 L‘ X " ,, 10
v X 11 It!

:50 3 A. f... 5.41009
3% 40 i ° ' “ LL‘ 0 80 1°90
11 1'55 2:24 252 3'21 3:50 "—

5 M12K>€K XXI I? g
g 30 59K x a 60 g

8 XX
0" 20 “WA“ Xi 5‘ e 40 g
m T AA X 9 o
a. 10 Wt 20 0
Q 0 rm’09molooo o o 3 f ”ii—L31— 0 U
0"“ 0:00 1:12 2:24 3:36
elapsed time (hrzmin)

 

 

 

ﬁgure 2: Measured concentrations of added solutes versus tint. Oxygen (02), Nitrate (N03),
and sulfate (80,) concentrations are plotted on the leﬁ y-axis and chloride (Cl) concentrations
are plotted on the right y-axis.

approximately 100 mg/L Cl (as a conservative tracer) and 10 mg/l S0,. Sulfate was

consumed at a rate of 0.24 hr'1 (Figure 3.3).

80

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

PPONS Rate Data manual
0.5- * ~ ‘ * * -- » o, 0.60
4 N03021
0.0 1 “XXXXXXXX 7! X X’x‘x xx” “ﬁr 304 0.27
E 9
Q 05 - —— *\_——— v..—
i-
8 \
-1.0 - -————----~ 4
'5 A 9 DO
-15 l» — ———— — ———- -- -— — ——4 -_ ——v_-—. >1<N03 —
ASO4
“2.0 I I l G
0:00 1:12 2:24 3:36 4:48
elapsed time (hrzmin)
mama)
‘ so4 0.18
“we “aha“ﬁm‘ﬁ—h‘u
A 2N A
3: __. - _ __ _ __._._._ X >1< *
% x
U L m ,__* _ __ __, -
V X
.5
>KN03 -_.
A804
2:24 3:36 4:48 6:00
elapsed time (hrzmin)
PPS Rate Data
0.5 - MW
so4 0.24
0-0 NANA ‘—‘ ﬂ‘AK‘l—m
1': _0 5 N‘ . _
Q ' \
h
e 1081+ #2 ~ \ —-
.5
-1.5 ————— --—A————- —
'2.0 T I l l I
0:00 1 : 12 2:24 3:36 4:48 6:00

elapsed time (hr:min)

 

Figure 3.3: Regression plot for rate coefﬁcient determination, obtained by ﬁtting

equation 3 to experimental data. Lines designate portion of dataset used to determine

rates.

81

 

Obtaining representative rates of in-situ processes is often a limiting factor in the
development of geochemical models to evaluate processes in natural environments.
Literature values for N03 and SO, consumption have been calculated based on zero and
ﬁrst order assumptions and estimated primarily by laboratory studies but also by some
ﬁeld experiments. Selecting a representative rate for TEA consumption is difﬁcult since
rate estimates range several orders of magnitude (Tables 3.2 &3.3). Our ﬁeld-measured
ﬁrst-order rates for aerobic respiration and denitriﬁcation are similar to those estimated
by in-situ ﬁeld methods in other locations (Table 3.4). Estimated ﬁrst-order rates of
nitrate and sulfate reduction found by Chapelle (34) and Cunningham (19) were slower
than those observed in this study, perhaps due to differences in electron donor utilization.
These studies looked at longer-term (months) consumption of N03 and SO, down ﬂow
path, which may have allowed for utilization of a wider range of organic compounds,
including more complex fuel constituents, to serve as electron donors. Our study
examined the rates of TEA reduction, as they would be initiated during a discrete
recharge event that introduced low concentrations of electron acceptors. In this situation,
the system was depleted with respect to 02, N03, and SO, and had available microbially-
mediated intermediates such as H2 and low molecular weight organic acids to serve as
more attractive electron donors. The availability of these donors may have allowed for a
greater rate of acceptor consumption. This hypothesis is supported by the observation
that during the Cunningham (19) ﬂow path N03 augmentation study, the rate of N03
reduction decreased over time, presumably due to depletion of li'nore favorable electron

donors in the system. Given the nature of the zeroth order laboratory-based methods, it is

82

impossible to directly compare the rates found in laboratory studies to those found in
ﬁeld studies such as ours. However our in-Situ rate estimates for O2 and N03
consumption are similar to in-situ rates estimated found during push-pull tests in another
petroleum contaminated aquifer (20), suggesting that these rates may be applicable in
other reduced environments where microbial intermediates such as H2 are available in

excess of the amount of TEA introduced.

83

Table 3.2 Summary of Rate Measurements for Sulfate Reduction

 

Environment

Method

Rate (uM/day)

Order

Ref

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

UA: Black Creek, USA ﬂux 1.92-41 .1 E-5 zero (1 7)

UA: Florida, USA ﬂowpath 2.74 E-4 zero (35)

UA: Fox Hills, USA ﬂowpath 5.48 E-4 zero (36)

UA: Fuhrberg, Germany depth proﬁle 3.84 E-2 zero (3 7)

UA: Sturgeon Falls, CAN depth proﬁle 4.38 E-2 zero (38)

UA: Bocholt, Germany dgpth proﬁle 6.3 E-2 zero (13)

UA: Tuse Naes, Denmark radiotracer 0.41-1.0 zero (13)

UA: Romo, Denmark radiotracer 0.14-12.3 zero (13)
mixed and pure culture microcosms 3.04E+4 zero (39)
Nearshore sediments radiotracer 6.3 E+2 zero (12)

UA: Denmark depth 0.85-2.55 zero (40)

UA: Denmark direct 0.55-12.3 zero (40)

CA: Norman, OK Landﬁll microcosms 3.5 zero (IQ

Rate (umo_l/g -day)*

Wintergreen Lake, USA microcosms 5.50E-2 zero (9)
Wintergreen Lake, USA microcosms 3.60E-2 zero (9)
Ocean Seds, Germany microcosms 4.30E-6 zero (41)
Landﬁll: Norman, OK microcosms 6.70E-4 zero (10)
Landﬁll: Norman, OK microcosms 0.1-1.3 E-3 zero (10)
Landﬁll-Denmark microcosms 0.1-1.3 E-3 zero (11)
Landﬁll-Denmark microcosms 1.8 E-03 zero (11)
Wetland Soils microcosms 0.1-ll zero (8)

CA: petroleum push-pull test 5.28 E+3 zero (18)

Rate (day'l)

CA metroleum ﬂowpath 0.02-0.08 ﬁrst (7)

CA: ﬁtroleum microcosms 0.02-0.08 ﬁrst ( 7)

CA: petroleum aug. ﬂow path 0.1 ﬁrst (I 9)

CA: petroleum/Cl solvents push-pull test 6.48 ﬁrst this study
CA: petroleum/Cl solvents push-pull test 4.32 ﬁrst this study
CA: petroleum/Cl solvents push-pull test 5.76 ﬁrst this study

 

* umol/gram of dry weight sediment/day

UA: uncontaminated aquifer

CA: contaminated aquifer

84

 

Table 3.3 Summary of Rate Measurements for Nitrate Reduction

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Environment Method Rate (umng-dayy Order Ref.
CA: sewage (N 03) microcosms 9.6-22.4E-4 zero (16)
CA: sewage (N03) tracer tests 3.1-7.8E-4 zero (I 6)
Freshwater Lake Sediments microcosms 3.10E—01 zero (42)
Landﬁll: Denmark microcosms 1.26 E-3 - 2.33E-1 zero (I I)
Wetland soils: 0-2hrs microcosms 0.23 zero 18)
Wetland soils: l-lOdays microcosms 0.5-9.3 zero (8)
Wetland soils: lO-lOOdgys microcosms 0.2-4.7 zero (8)
Forest Soil microcosms 4.8 zero (43)
Rate (uM /dagy)
CA: petroleum push-pull tests 1.92E 3 zero (18)
Rate @111”)
CA: petroleum push-pull tests 5.28 ﬁrst (20)
CA: petroleum aug. ﬂow path 0.1-0.6 ﬁrst (19
CA: petroleum/Cl solvents push-pull tests 5.04 ﬁrst this study
CA: petroleum/Cl solvents push-pull tests 7.44 ﬁrst this study

 

 

 

 

 

 

* umol/gram of dry weight sediment/day

CA: contaminated aquifer

Table 3.4 Summary of Available First-Order Field Rate Measurements (day'l)

 

 

 

 

 

 

 

 

 

Aerobic Respiration Denitrification Sulfate reduction
Chapelle et al., 1996 (34) - - 0.02-0.08
Cunningham et al., 2000 (19) - 0.1-0.6 0.1
Schroth, et al., 1998 (20) 3.6-40.6 2.16-10.08 -
This study 14.4 5.04-7.44 4.32-6.48

 

85

 

 

Our results show that when introduced simultaneously, N03 and SO, are
consumed simultaneously at Similar rates. This is contrary to the conventional idea that
N03 will be consumed prior to S0, based on thermodynamic arguments. However, it is
consistent with the kinetic model of terminal electron accepting processes that suggests
that in a non-competitive situation there is no reason why processes should follow the
thermodynamic order. Simultaneous redox processes and overlapping redox zones have
been observed in pristine and contaminated aquifers systems previously (11,40). It has
been suggested that concentration levels of electron acceptor or donor, microniches of
various TEAPS, or variability of other parameters such as pH may explain the complex
patterns of redox processes observed. In addition to these possibilities, our results
suggest that SO, and N03 may be consumed by microbially-mediated reactions other
than the traditional respiratory pathways since N03 and SO, were consumed

simultaneously at roughly the same rates.

Four possible scenarios are proposed. One scenario assumes an abundance of
electron donors in a system limited by electron acceptors, N03 and 80,. This is a likely
scenario both at this site and many other contaminated systems. During these tests,
available H2 concentrations were high enough (2.35-3.55nM) to support both N03 and
S0, reduction. Under these circumstances, when N03 and SO, are added
simultaneously, they could be consumed Simultaneously by an abundance of sulfate
reducers in the system. Sulfate reducers have been shown to be ubiquitous in the
environment and capable of gaining energy from 02, Fe (III), and various nitrogen oxides

(8,44,45). Similarly, a second possibility is dissimilatory reduction of N03 by denitriﬁers

86

 

under these conditions. A third possible scenario, suggests that if the bacteria present are
nitrogen limited, assimilatory nitrate reduction may occur simultaneously with SO,
reduction. These processes would not necessarily follow the thermodynamic order of
processes since they would not be in competition with each other. The assimilatory N03
reduction would provide a nitrogen source to the organism while the respiratory SO,
reduction pathway would provide energy. A fourth possibility is the reduction of N03 by
green rust (GR). Nitrate has been shown to be stoichiometrically reduced to ammonium
while GR is oxidized to magnetite (46). The rate law for this reaction is ﬁrst order,
dependent on Fe (11) concentration, where k=0.072 hr'l. This rate is slower than the
observed rate for N03 removal at this site however it is not inconsequential. It is unclear
if green rust is present at the site Since it is difﬁcult to identify with standard
mineralogical procedures due to its high reactivity including extreme sensitivity to
oxidation by oxygen. However, reduced iron is present in the groundwater and magnetite
has been detected in the aquifer material. The possibility of the activity of GR is often
indicated in investigations of polluted soils, sediments, landﬁlls, and in remediation
techniques using F e(0) (47,48). Future work needs to be conducted to determine the
speciﬁc microbially-mediated TEA processes being performed at this site during recharge

events.
Surface Reaction Vessel Experiments:

Four variations of SRV experiments were performed. This water was then
amended with solutes, hand shaken to ensure it was well mixed, and then allowed to react
for the length of the corresponding push-pull test. Samples were collected at roughly 5-

minute intervals. SRVE 1 was performed using water from ML3-25 (June 2000).

87

 

Aquifer water was pumped into the SRV and amended with 100 mg/L Cl, 10 mg/L N03,
and 10 mg/L 80,. The solution was allowed to react for the length of push-pull test 1,
approximately four hours. No changes in N03 or SO, concentrations were observed over
this time period. SRVE 2 and SRVE 2K were performed using water from ML3-25 (July
2000). Aquifer water was pumped into two identical SRVS. Each was amended with 100
mg/L Cl, 20 mg/L N03, and 20 mg/L SO,. SRVE 2K was then amended with
formaldehyde (approximately 0.4% ﬁnal concentration) to serve as a killed control. The
solutions were hand shaken to ensure they were well mixed and then allowed to react for
the length of push-pull test 2, approximately 4 hours. No changes in N03 or SO,
concentrations were observed in either SRVE 2 or SRVE 2K over this time period.
SRVE 3 and SRVE 3K were performed using water from ML3-25 (July 2000). Aquifer
water was pumped into two identical SRVS. Each was amended with 100 mg/L Cl, 50
mg/L N03, and 50 mg/L SO,. SRVE 3K was then amended with formaldehyde
(approximately 0.4% ﬁnal concentration) to serve as a killed control. The solutions were
allowed to react for the length of push-pull test 2, approximately 4 hours. No changes in
N03 or SO, concentrations were observed in either SRVE 3 or SRVE 3K over this time
period. SRVE 4 and SRVE 4K were performed using water from ML3-22, which was
similar chemically to ML3-25 in all respects except it contained slightly higher DCE
concentrations (4.9 ug/L) (July 2000). Aquifer water was extracted into two identical
SRVS. Each was amended with 100 mg/L Cl, 50 mg/L N03, and 50 mg/L 80,. SRVE
4K was then amended with formaldehyde (approximately 0.4% ﬁnal concentration) to
serve as a killed control. The solutions were allowed to react for approximately 4 hours.

No changes in N03 or SO, concentrations were observed in either SRVE 4 or SRVE 4K

88

over this time period. Measured F e2+ concentration remained consistently high (11.3-
11.5 mg/L) throughout all SRVES and dissolved oxygen was below the detection limit of

our ﬁeld kit (<0.03 mg/l) suggesting that the solutions remained anaerobic.

These results indicate that any redox reactions observed in the push-pull tests do
not occur in the reduced water without inﬂuence of the sediments and their associated
microbial communities. This suggests that highly oxidized compounds such as N03 and
SO, may persist in reduced waters. This ﬁnding is analogous to the persistence of Cr3+ in
oxidized water (49). It has been shown that reduced Cr3+ is oxidized to Cr6+ at a much
greater rate by the presence of solid phase manganese oxides than by oxygen (50). In the
case of the persistence of N03 and SO, in reduced water, it is unclear if the absence of
the solid phase or its associated microbial population is allowing these oxidized
compounds to persist. However, these ﬁndings have implications for how geochemical
models of reduced systems are constructed and how we interpret the results from such
modeling efforts. Geochemical models are driven by free energy yield and consider only
the aqueous phase in predicting reactions, disequilibria and other constraints. Even with
thermodynamic and kinetic arguments to input into a model, the model would not predict
our observed data. Our data clearly demonstrate that in addition to a simple mixing of
aqueous solutions, knowledge of the reactions being controlled by the solid phase and
their associated microbial population is necessary to understand how N03 and SO, will

behave in the environment.

89

 

Acknowledgements

We thank Erik Smith, John Moss, Shane Snavely, Karen Wayland, and Sharon Simpson
for help with data collection and the MSU Physics Shop lead by Tom Palazolo for their
assistance in packer development. The research is funded by a grant from the National
Science Foundation Environmental Geochemistry and Biogeochemistry Program, EAR-
9708487. Funding is also by the USGS Toxic Substances Hydrology Program. We
thank the National Center for Integrated Bioremediation, Research, and Development

(NCIBRD) and EFX Systems, Inc. for their cooperation.

90

Literature Cited

(1)
(2)
(3)

(4)

(5)
(6)
(7)

(8)
(9)

(10)

(11)

(12)
(13)
(14)

(15)
(16)

(17)
(18)

Vroblesky, D. A.; Chappelle, F. H. Water Resour. Res. 1994, 30, 1561-1570.
Lovley, D. R. Microbiol. Rev. 1991, 55, 259-287.

Chappelle, F. H. Ground- Water Microbiology and Geochemistry; 2nd ed.; John
Wiley and Sons, Inc.: New York, 2001.

Stumm, W.; Morgan, J. J. Aquatic Chemistry; 2nd ed.; John Wiley and Sons: New
York, 1981.

Lovley, D. R.; Goodwin, S. Geochim. Cosmochim. Acta 1988, 52, 2993-3003.
Champ, D. R.; Gulens, J .; Jackson, R. E. Can. J. Earth Sci. 1979, 127, 85-108.

Chappelle, F. H.; Haack, S. K.; Adriaens, P.; Henry, M.; Bradley, P. M. Environ.
Sci. T echnol. 1996, 30, 3565-3569.

D'Angelo, E. M.; Reddy, K. R. Soil Biology and Biochemistry 1999, 31 , 815-830.

Lovley, D. R.; Dwyer, D. F .; Klug, M. J. Appl. Environ. Microbiol. 1982, 43, 1373-
1379.

Cozzarelli, I. M.; Suﬂita, J. M.; Ulrich, G. A.; Harris, S. H.; Schroll, M. A.;
Schlottrnann, J. L.; Christenson, S. Environ. Sci. T echnol. 2000, 34, 4025-4033.

Ludvigsen, L.; Albrechtsen, H.-J.; Heron, G.; Bjerg, P. L.; Christensen, T. H.
Journal of Contaminant Hydrology 1998, 33, 273-291.

Crill, P. M.; Martens, C. S. Geochim. Cosmochim. Acta 1987, 51, 1175-1186.
J akobsen, R.; Postrna, D. Geology 1994, 22, 1103-1106.

Bowman, J. P.; Jimenez, L.; Rosario, I.; Hazen, T. C.; Sayler, G. S. Appl. Environ.
Microbiol. 1993, 59, 2380-2387.

Hunter, K. S.; Wang, Y.; Van Cappellen, P. J. Hydrol. 1998, 209, 53-80.

Smith, R. L.; Garabedian, S. P.; Brooks, M. H. Environ. Sci. T echnol. 1996, 30,
3448-3456.

Chappelle, F. H.; McMahon, P. B. J. Hydrol. 1991, 127, 85-108.

Istok, J. D.; M.D., H.; Schroth, M. H.; Hyman, M. R.; O'Reilly, K. T. Ground
Water 1997, 35.

91

(19)

(20)

(21)
(22)
(23)

(24)

(25)

(26)

(27)

(28)

(29)

(30)

(31)
(32)
(33)

(34)

(35)

Cunningham, J. A.; Hopkins, G. D.; Lebron, C. A.; Reinhard, M. Biodegradation
2000, 11, 159-170.

Schroth, M. H.; Istok, J. D.; Connoer, G. T.; Hyman, M. R.; Haggerty, R.; O'Reilly,
K. T. Ground Water 1998, 36.

Istok, J. D.; Field, J. A.; Schroth, M. H. Ground Water 2000.
Luthy, L.; Fritz, M.; Bachofen, R. Appl. Environ. Microbial. 2000, 66, 712-717.
Lovely, D. R.; Phillips, E. J. P. Appl. Environ. Microbial. 1987, 53, 2636-2641.

Lofﬂer, F. E.; Tiedje, J. M.; Sanford, R. A. Appl. Environ. Microbial. 1999, 65,
4049-4056.

Chappelle, F. H.; McMahon, P. B.; Dubrovsky, N. M.; Fujii, R. F.; Oaksford, E. T.;
Vroblesky, D. A. Water Resour. Res. 1995, 31, 359-371.

Atlas, R. M.; Bartha, R. Microbial Ecology: Fundamental and Applications; 4th
ed.; Benjamin/Cummings Publishing Company, Inc.: Menlo Park, 1998.

USGS Installations Restoration Program: Phase II-Canﬁrmatian/Quantiﬁcatian
Stage 1: Wurtsmith Air F arce Base, Michigan: Investigations ofgroundwater and
soil contamination at selected sites: United States Geological Survey, Water
Resources Division, 1991.

McGuire, J. T.; Smith, E. W.; Long, D. T.; Hyndman, D. W.; Haack, S. K.; Klug,
M. J.; Velbel, M. A. Chem. Geol. 2000, 169, 471-485.

Reynolds, L. A., Haack, S. K. Marganwalp DW, Buxton HT (eds) US Geological
Survey toxic substances hydrology program. Proc Technical Meeting, Charleston,
South Carolina, 8-12 March 1999. US Geological Surv Water-Resaur Invest Rep
1999, 99-4018C, 635-639.

Dojka, M. A.; Hugenholtz, P.; Haack, S. K.; Pace, N. Appl. Environ. Microbial.
1998, 64, 3869-3877.

Haack, S. K.; Bekins, B. A. Hydrogeology Journal 2000, 8, 63-76.
Snodgrass, M. F.; Kitanidis, P. K. Ground Water 1998, 36.
Haggerty, R.; Schroth, M. H.; Istok, J. D. Ground Water 1998, 36.

Chappelle, F. H.; Bradley, P. M.; Lovley, D. R.; Vroblesky, D. A. Ground Water
1996, 34, 691-698.

Plummer, L. N. Water Resour. Res. 1977, 13, 801-812.

92

(36)

(37)

(38)

(39)

(40)
(41)

(42)
(43)

(44)

(45)

(46)

(47)

(48)

(49)

(50)

Thorstenson, D. 0; Fisher, D. W.; Croft, M. G. Water Resour. Res. 1979, 15, 1479-
1498.

Bottcher, J .; Strebel, 0.; Duynisveld, W. H. M. Gealagische Jahrbuch 1989, C51,
3-40.

Robertson, W. D.; Cherry, J. A.; Schiff, S. L. Water Resour. Res. 1989, 25, 1111-
1 123.

Spear, J. R.; Figueroa, L. A.; Honeyman, B. D. Appl. Environ. Microbial. 2000, 66,
371 1-3721.

Jakobsen, R.; Postrna, D. Geochim. Cosmochim. Acta 1999, 63, 137-151.

Knoblauch, C.; Jorgensen, B. B.; Harder, J. Appl. Environ. Microbial. 1999, 65,
4230-4233.

Nijburg, J. W.; Laanbroek, H. J. Microbial Ecology 1997, 34, 254-262.
Bengston, G.; Bergwall, C. Soil Biology and Biochemistry 2000, 32, 545-557.

Dannenberg, S.; Kroder, M.; Dilling, W.; Cypionka, H. Arch. Microbial. 1992, 158,
93-99.

Coleman, M. L.; Hedrick, D. 3.; Lovely, D. R.; White, D. C.; Pye, K. Nature 1993,
361, 436-438.

Hansen, H. C. B.; Guldberg, S.; Erbs, M.; Koch, C. B. Applied Clay Science 2001,
18, 81-91.

Cu, B.; Phelps, T. J .; Liang, L.; Dickey, M. J .; Roh, Y.; Kinsall, B. L.; Palumbo, A.
V.; Jacobs, G. K. Environ. Sci. T echnal. 1999, 33, 2170-2177.

Fredrickson, J. K.; Zachara, J. M.; Kennedy, D. W.; Dong, H.; Onstott, T. C.;
Hinman, N. W.; Li, S. M. Geochim. Cosmochim. Acta 1998, 62, 3229-3257.

Eary, L. E.; Rai, D. Environ. Sci. T echnal. 1987, 21, 1187-1193.

Rai, D.; Eary, L. E.; Zachara, J. M. Sci. T atal Environ. 1989, 89, 15-23.

93

CHAPTER IV.

Evolution of redox zonation in a contaminated aquifer: Development of 3-D coupled
partial-equilibrium reactive ﬂow-and-transport model.

94

Introduction

The concept that terminal electron accepting processes (TEAPS) distribute
themselves sequentially into redox zones is often used to interpret how organic
compounds will be degraded in the environment. Champ et al. (1979) identiﬁed redox
sequences in groundwater ﬂow systems associated with a reduction in dissolved organic
carbon. Since then, many workers have explored the application of using sequential
TEAPS to assist in the clean-up of contaminated Sites via bioremediation and natural
attenuation strategies (BAEDECKER et al., 1993; BJERG et al., 1995; KONHAUSER, 1994;
LEVINE et al., 1997; LYNGKILDE and CHRISTENSEN, 1992)[Vroblesky, 1994 #47. In fact,
natural attenuation is quickly becoming an acceptable treatment technology for managing
organic contaminant plumes (EPA, 1997). The success of such remediation strategies
depends upon a thorough understanding of how TEAPS develop into redox zones in
complex subsurface systems and under what conditions redox state and thus zonation
may change. This requires the ability to quantify interconnected hydrogeologic,
microbiologic, and geochemical processes which contribute to redox including the
evaluation of rate limiting steps. The development of tools that adequately quantify the
complete evolution of and changes in redox zonation is necessary for the effective design
of active remediation strategies and the assessment of the potential for natural attenuation
in contaminated systems.

Redox zonation is in part controlled by the activity of microorganisms, which ﬁrst
consume oxygen, and then a succession of alternate terminal electron acceptors (TEAS)

to support their grth using a variety of carbon sources (CHAPPELLE, 2001; LOVLEY,

95

1991; LOVLEY and GOODWIN, 1988; STUMM and MORGAN, 1981). The succession of
TEAPS in order of decreasing redox potential and ﬁ'ee energy yield is generally: oxygen
reduction, nitrate reduction, manganese reduction, iron reduction, sulfate reduction, and
methanogenesis. In most closed groundwater systems, the sequence of TEAPS results in
redox zones following this succession (i.e., increasingly reduced conditions occur further
down the ﬂow path) (CHAMP et al., 1979). However, down gradient of contaminant
plumes, the sequence of TEAP zones is generally reversed (CHAPPELLE, 2001; LOVLEY,
1991). That is, the most reduced conditions (e. g., methanogenesis) occur near the
contaminant source due to the greater availability of electron donors, while less reducing
conditions (e. g., iron reduction) occur father down ﬂow-path from the contaminant
source. Since many hydrocarbons, which serve as electron donors, concentrate in the
capillary ﬁinge area, more reducing conditions often occur near the water-table interface,
resulting in a plume that has vertical as well as horizontal redox zonation (CHAPPELLE et
al., 1996; VROBLESKY and CHAPPELLE, 1994).

In principle, as groundwater plumes deve10p and TEAPS progress, a steady-state
balance between source inputs and biotransformation rates can be achieved resulting in a
characteristic stabilization of plume length (BUSCHECK and ALCANTAR, 1995; MCNAB
and DOOHER, 1998). Precise mathematical descriptions of the necessary factors leading
to a characteristic plume length have not yet been worked out. Statistical analyses have
revealed that the steady-state plume length is related to the mean site groundwater
velocity, and the strength of the contaminant source (MCNAB, 2001). The TEAPS
operating in the aquifer, as well as dispersion, have also been shown to have signiﬁcant

impact on steady-state hydrocarbon plume length such that the introduction of more

96

 

favorable electron acceptors (e. g., 02) via groundwater reaeration can reduce plume
length by up to 87% (N EALE et al., 2002). Other workers have suggested that the nature
of the contaminant can also play a role in the resulting plume length. Statistical analysis
of mixed contaminant plumes including chlorinated hydrocarbons at several locations
revealed that, in plumes in which active dehalogenation was occurring, plume length was
signiﬁcantly Shorter (roughly by a factor of two) than plumes where reductive
dehalogenation was not occurring (MCNAB and RICE, 2001). Although, the uncertainty
of these factors in complex contaminant plumes makes predicting a steady-state plume
length difﬁcult, observational evidence clearly suggests that plumes do achieve a steady
state with respect to plume length on relatively short time scales (years). Therefore, as
groundwater hydrocarbon plumes evolve, they will reach a point where the front of the
plume can no longer advance based on a contaminant source of constant strength.

These concepts lead to the hypothesis that when a plume has stabilized such that it
has achieved its maximum spatial extent, horizontal redox zonation (down ﬂow path)
cannot be sustained due to the continual depletion of TEAS within the original redox
zones. Classic redox zonation patterns suggest that the most reduced zone will occur
nearest the contaminant source where more favorable TEAS such as oxygen (02), nitrate
(N03), iron (HI) Fe”, and SO, have been successively depleted. The boundary between
this highly reduced zone (e. g., methanogenic zone) and the adjacent zone where more
favorable TEAS are available (e.g., S0,) over time will not be sustained due to greater
ﬂuxes of electron donor moving out of more reduced zone into the adjacent less reduced
zone. Following this concept, it can be inferred that ultimately when a ﬁxed plume

length is achieved, remaining TEAS will continue to be consumed in each of the zones

97

 

 

until a single highly reduced zone exists. This state is, of course, not static but rather
dynamic at least in part due to inputs of TEAS from recharge events and lateral dispersion
within the plume (HESS et al., 1992). However, these inputs would result in a more
complicated redox pattern within the plume where more energetically favorable TEAPS
are stimulated in small transition zones near plume boundaries as opposed to a pattern of
larger sequential redox zones. This conceptual model for the evolution of redox state in
contaminant plumes has implications not only for basic understanding of redox processes
in nature, but also for the evaluation and design of bioremediation and natural attenuation
strategies.

Previous work conducted at the former Wurtsmith Air Force Base (WAF B) in
Oscoda, Michigan supports the idea that traditional redox zonation may not be sustained
in contaminant plumes that have had enough time to evolve to “mature” conditions where
TEAS have been successively depleted in each of the zones and changes in TEAPS likely
occur in the transition zones at plume boundaries. Multivariate statistical evaluation of
geochemical data from the contaminant plume at WAF B, which had 50 years to evolve,
has shown that water does not associate into distinct redox zones but rather, dominant
variability is inﬂuenced primarily by inputs from recharge events (Chapter 2). In
addition, microbiological measurements suggest that microorganisms that traditionally
perform more favorable TEAPS (aerobic respiration, denitriﬁcation, Fe—reduction, and
sulfate reduction) do not exist in detectable quantities within the center of the plume
(Haack, personal communication 2002, (WEST, 1999)). The primary active microbial
community in the aquifer is composed of methanogens (DOJKA et al., 1998), suggesting

that all of the more favorable TEAS have been depleted in the center of the plume. Yet,

98

 

 

we know that recharge introduces TEAS to the aquifer, and that they are consumed within
the plume because measured TEA concentrations in the aquifer remain near zero. From
this we can infer that these reactions are occurring in small regions near the boundaries of
the plume where it is difﬁcult to sample and not in large discrete redox zones.

Other workers have recognized the importance of the evolution of redox zones in
contaminant plumes as a control on the long-term potential for hydrocarbon transport and
degradation and consequently have evaluated changes in redox processes at various
scales in a contaminated aquifer (COZZARELLI et al., 2001). With the combined use of
small-scale data from aquifer cores and large-scale (plume) geochemical well data, they
demonstrated that redox evolution (Fe-reduction to methanogenesis) occurred on very
small spatial scales (tens of centimeters) in a contaminated aquifer in Bemidji,
Wisconsin. These ﬁndings further support our concept that understanding redox
evolution not only in large zones which may ultimately disappear, but also at smaller
interfaces such as at plume boundaries, is critical to understanding the fate of

hydrocarbons in groundwater systems.

Goals and Approach

To evaluate the feasibility of this conceptual model for the evolution of redox
zonation in a contaminated aquifer, a numerical model was constructed describing how
redox zones evolve after initial contaminant introduction, and how those zones may
continue to evolve over time. Numerical models are a common tool to test hypotheses

regarding the hydrogeochemical evolution of groundwater systems that link TEAPS and

99

hydrocarbon degradation (e. g., (ABRAMS and LOAGUE, 1998; COZZARELLI et al., 2000;
ESSAID et al., 1995; HUNTER et al., 1998; JAKOBSEN and POSTMA, 1999; KEATING and
BAHR, 1998; LENSING etal., 1994; SALVAGE and YEH, 1998; TEBES-STEVENS et al.,
1998). Most of the recent modeling of contaminant plume evolution assumes that
organic degradation is coupled to the sequential model of TEAPS in redox zones, which
according to our data, is not appropriate for describing the contaminant plume at WAF B
and may not be appropriate in other evolved systems (ABRAMS and LOAGUE, 1998;
ISLAM et al., 2001; PETERSON and SUN, 2000; SUNA et al., 2001).

A number of different simulation approaches have been developed, differing
mainly in how organic and inorganic processes have been coupled. A summary of these
models is provided in a recent review paper (BRUN and ENGESGAARD, 2002). The
various types of models comprise two major approaches to quantifying biogeochemical
transport; 1) those using a “one-step” biodegradation pathway, where the substrate and
electron acceptor have been identiﬁed; and 2) those using a “two-step” biodegradation
approach that divides the biogeochemical processes into two half-cell reactions, where
the ﬁrst reaction consumes substrate and donates electrons and the second reaction
consumes those electrons by coupling them to a TEA. According to Brun and
Engengaard (2001), the ﬁrst group of models (one-step) found support in the classical
biodegradation transport models, which consider complex bacterial grth relationships
and add geochemical reactions only as needed. This approach tracks a limited number of
geochemical species through a system and therefore cannot account for the more
complicated interrelations between biogeochemical cycles, which may Signiﬁcantly

impact the “real-world” chemical pathways. The second group (two-step) found support

100

in geochemical transport models that consider a complex suite of geochemical reactions
and subsequently add simple degradation reactions such as zero- or ﬁrst-order processes,
neglecting bacterial growth. Although these models do a better job of accounting for the
complete geochemistry of the system, they may oversimplify organic biodegradation
reactions by neglecting the growth and decay of microorganisms, which is an important
control on many microbially mediated reactions. Many of these two-step models adhere
to a partial equilibrium approach (PEA) where the second reaction (the TEAP) is allowed
to occur instantaneously assuming that the ﬁrst reaction is the rate-limiting step (MCNAB
and NARASIMHAN, 1994; POSTMA and JAKOBSEN, 1996). These models assume that the
initial fermentation step (electron production) is the rate-limiting step based on
thermodynamic arguments and the general occurrence of very low concentrations of
fermentation products. We have adopted this strategy for the development of our model.

Our 3-dimensional model considers both 1) the biodegradation of organic matter
based on the inﬂuence of transport on microbiological grth and 2) the complex suite of
biogeochemical reactions operating in the aquifer. Both are important to consider
because, due to the interconnectivity of biogeochemical cycles, the viability of a given
degradation pathway is a function of local geochemical conditions which ultimately
determine its energetic favorability. For example, subtle changes in pH initiated by one
chemical process (e. g. SO, reduction) can have profound effects on the pe of a system,
thus changing the stability of several other chemical constituents (LANGMUIR, 1997). By
considering a more complete geochemical system, which describes how one cycle may
impact another, predicted chemical changes may be very different than originally

anticipated. For example, the pH changes predicted above during sulfate reduction may

101

be buffered by the precipitation/dissolution reactions of calcite in the aquifer. In addition
to the geochemical processes operating in the aquifer, it is important to consider the
impact of microorganisms on the system. Microorganisms become involved in
thermodynamically favorable redox reactions to extract energy for metabolic processes.
As a result, they can inﬂuence the rates of biodegradation processes; however, they
cannot inﬂuence the overall energetic favorability of such reactions. Therefore,
geochemical thermodynamic considerations remain an important control on redox
dependent degradation particularly when considered in conjunction with the kinetics of
the rate-limiting step.

The model we have developed links the reactive transport codes MODFLOW
(MCDONALD and HARBAUGH, 1988) and RT3D (CLEMENT, 1997; CLEMENT and JONES,
1998) with the geochemical modeling code PHREEQC (PARKHURST and APPELO, 1999).
Thus ﬂow and transport variables such as advection, dispersion, rate limitation, and
microbial growth are considered with the thermodynamic geochemical framework of the
groundwater and contaminant plume system. Other attempts have been made to combine
ﬂow and transport codes with geochemical modeling codes, for example PHAST
(PARKHURST et al., 2001) links the transport code HST3D with PHREEQC. One beneﬁt
of our linked modeling code is that that many groundwater ﬂow and transport models
already exist in the MODF LOW and RT3D format so they could easily be coupled with
all the geochemical capabilities built into PHREEQC.

The goal of this study is to quantitatively evaluate a new conceptual model for the
evolution of redox zonation in aquifer systems over relatively long time scales (decades)

and demonstrate the necessity of examining the complete geochemical condition of the

102

system. Speciﬁcally, this paper 1) presents an alternative model for the ultimate fate of
redox zonation in contaminated aquifers; 2) proposes a new conceptual model describing
the importance of TEAS inputs from recharge and dispersion on the redox state of a
contaminant plume and; 3) introduces a new tool to quantify the hydrogeochemical
evolution of groundwater through the linking of three previously established modeling

codes.

Study Site Description
The study site is a shallow sandy aquifer contaminated with petroleum

hydrocarbons and chlorinated solvents at the former Wurtsmith Air Force Base (AFB),
Oscoda, Michigan, U.S.A. Wurtsmith AFB lies on a sandy plain of glacial lake sediment
approximately 2 km west of the Lake Huron shoreline in Michigan's Lower Peninsula.
The aquifer is approximately 20 meters thick and composed of highly permeable
alternating aeolian sands and glacial outwash material. Hydraulic conductivities of these
deposits are on the order of 30 m/day. The unconﬁned aquifer is underlain by at least
30.5 meters of silty clay. The water table varies from 3.5-5.0 m below land surface (184-
186 m above sea level), ﬂuctuates 0.3 to 0.7 m annually, and the water table gradient
ranges from 3 to 5 meters per kilometer. Assuming an effective porosity of thirty
percent, average groundwater velocity is approximately 0.3-0.5 m per day. The plume has
lateral direction shifts of 7 to 10 degrees because the groundwater ﬂow direction is
inﬂuenced by seasonal recharge at this site.

Fire training activities performed by the Air Force ﬁ'om 1952 to 1986 at

this site contaminated the groundwater. This suggests that the groundwater plume has

103

 

had nearly 50 years to evolve to its current redox state. Various compounds were used
during the training exercises to ignite and extinguish the ﬁres, including waste fuels,
chlorinated solvents, aqueous ﬁhn-forming foams, a multipurpose dry chemical,
potassium bicarbonate based soda, and Halon 1211. An extensive plume approximately
50 m wide and 400 m long has developed down gradient from the concrete pad where the
ﬁre training activities were performed (Figure 4.1). Most of the contamination is
associated with aquifer solids between 4.5 and 5.7 m below land surface, which have an
average total hydrocarbon concentration of 13,650 mg/kg (U SGS, 1991).

The plume is complex in that it contains both BTEX compounds (benzene,
toluene, ethylbenzene, and xylenes) and chlorinated solvents (e.g., dichloroethene).
Concentrations of BTEX compounds in the plume range from 20 to 1,000 rig/liter while
concentrations of dichloroethane, chloroethane, and vinyl chloride range from 2 to 1,000
pg/liter. To document the vertical distribution of contaminants within the plume, PVC
multi-level groundwater monitoring wells were installed at the site (Figure 4.2). The
reducing conditions of the plume vary historically from Fe-reducing to methanogenic (as
determined by H2 concentrations) both spatially and temporally on the time scale of
months (MCGUIRE et al., 2000). Dissolved H2 gas, an important intermediate in the
decomposition of organic matter, is thought to be an indicator of TEAPS. In anaerobic
environments, H2 is produced and consumed such that each TEAP maintains a
characteristic H2 concentration (LOVELY and PHILLIPS, 1987). At WAF B, TEAPS
indicated by hydrogen gas concentrations do not distribute themselves into sequential
zones down ﬂow path but rather, exhibit a more complicated likely related to recharge

inputs (Figure 4.3). Despite the complicated nature of the site, WAF B is an ideal location

104

 

 

    
  
 

Multi-Level
Tra - z - ct 4r

0 50 METERS

 

0 100 FEET

 

 

Figure 4.1 Plume map for the Fire-Training Area 2 (FTA-2) at the
former Wurtsmith Air Force Base, Michigan. The concrete pad was
the site of ﬁre-training activities. Grey lines represent plume position
during periods of low water table; Red lines represent plume position

during periods of high water table. Blue lines are water table contours
in feet above sea level.

105

 

 

 

 

 

 

 

O Fr—ML
o EFX-ML
500 _ . Q, 318 APumping Well
- '4; 310. '
o. 0 .'
0 ﬂ. 1 EFX production well
02 o, 12
400- 19.11 o
0%... .315 06
- 5.14 o
E O 13 ‘0. '0‘. 09
° 0 16 r. O
E 300 o. 0. 08
is ~ 05
z 0‘ 0....
04 "x
200 4 0 307 “u‘ T1
1“ T2
100 -
0 1 .
0 100 400 500

00 300
Easting (1‘!)

Figure 4.2: Two-dimensional well grid. The dashed arrows
denote wells through which chemical transects (T1 and T2) were

constructed.

106

”2 (nM)

Groundwater Flow Direction

V

Transect 1 (T1)

 

8 meters

 

 

 

 

 

 

310 11 14 16 3
53 meters
Transect 2 (T2)
318 l A 112 l 151 L l l 5
E T
8
d)
E
00
53 meters
4 1 0

N

Sulfate Reduction Fe Reuction

Inferred TEAP

Methanogenesis

Figure 4.3: Plot showing lack of redox zonation at WAF B as inferred by
dissolved hydrogen concentrations at two transects through the plume
roughly parallel to the direction of groundwater ﬂow in June 2000.

107

to examine the evolution of redox zonation Since detailed Spatial and temporal
geochemical, hydrogeologic, and microbiological database for the site has been
established (DOJKA et al., 1998; HAACK and BEKINS, 2000; MCGUIRE et al., 2000;

REYNOLDS and HAACK, 1999).

Methods

Development the T hree-Dimensianal Partial Equilibrium Numerical Model

The three-dimensional partial equilibrium numerical model couples an existing 3-
D ﬂow model (Sr/ﬂTH, 2002) created using MODF LOW with kinetically controlled
ferrnentations reactions (via RT3D) and thermodynamically controlled reactions (via
PHREEQC). A schematic of how these three codes work together is presented in Figure
4.4. MODF LOW is used to transport selected geochemical species including original
contaminant (Corg) from the source area. RT3D then takes the output from the ﬂow
model and allows the Corg in each cell to be reacted to produce fermentation products to
serve directly as electron donors (Cm) for the TEAPS. These outputs are then given to
PHREEQC, which evaluates the electron input, speciates the redox sensitive species
considering the complete geochemistry of the water, and returns the select chemical
Species to MODF LOW where transport processes are considered. This loop is repeated
for each cell with changes in Corg at every time step.

Initial application of the model (Phase I) is to test the hypothesis that under
realistic physico-chemical conditions at this site, a steady-state plume length can be

achieved and that once a stable plume length is established; redox zonation will no longer

108

 

.360

 

 

 

    

2.0300 E8209
84303020: 04 008 8
080830 38%

 

 

 

. . . . . 08.208.203.21... .
2.002.058 _ A ,, pImmmDO

 

 

8”... 89.00... 03.01.

 

 

 

 

4830.00... 8082.03. @8008 ._.m>_um
08008.03 01.00500... Aooav 00000 0: 0 .
0:0 0083 800303.00. 0:850 0000.02.03 04 “M59“

003038 000080

Ems—.0 A...” 3500 H 0000008”: 8002 momoagsm magmom 0». 980 0x895 Bomogm

00000. $502000 as: 8 003.0300 3 000: 00: 2:: m 035% E 020 m: 000: 00:: E :80.

109

be sustained. It should be noted that this simulation neglects inputs ﬁ'om recharge events,
which can directly deliver TEAS and also increase dispersion in increasing TEA inputs.
Examining the effects of recharge on mature plumes will be explored in the Phase (II) of
the modeling simulations (see Future Work sections). A description of speciﬁc model
construction inputs for each portion of the three components to the overall model is

described below.

Flow Model

A three-dimensional, numerical groundwater ﬂow model was constructed of the
surﬁcial aquifer in the area encompassing the former Wurtsmith Air Force Base (SMITH
et al., 1999), (SMITH, 2002). The computer program MODFLOW 96 was utilized to
solve the approximated partial differential ﬂow equation using a block-centered, ﬁnite-
difference approach throughout the modeled domain. A graphical user interface program
(GUI), GMS version 3.1, was used to construct the model grid, assign parameters, and
View model output ﬁles. The model was run as both steady-state and transient
simulations using a time-varying recharge input. Initial modeling simulations (Phase I)
use a steady state model (i.e., neglects differential recharge) and future modeling
simulations (Phase II) will use the transient model to evaluate recharge effects. The
boundaries of the model coincide with regional surface water features and groundwater
divides (Figure 4.5). To allow for accurate placement of surveyed monitoring locations
relative to model boundaries, the grid framework was referenced to the state plane

coordinate system. This was accomplished by using spatially referenced Digital Line

110

 

Figure 4.5: Groundwater ﬂow model output showing location of
natural boundaries and approximate location of groundwater plume
FTA-2 within this model framework.

111

Graphs (DLG) of surface water features to establish the model boundaries. A DLG was
also used to position an interior sink boundary (drain), representing Allen Lake. Land
surface topography was input to the model using an x,y,z data ﬁle created from Digital
Elevation Model (DEM) ﬁles. Each DEM covers a 7.5 minute unit corresponding to the
USGS 1:24,000 scale topographic quadrangle map series and is based on 30 by 30 meter
data spacing. Two DEM ﬁles, representing the Oscoda, MI and Foote Site Village, MI
7.5 minute topographic maps were processed using Arcinfo to merge the ﬁles, convert
coordinates from latitude/longitude to state plane coordinate system, and output an x,y,z
ﬁle compatible with GMS. Slightly exaggerated east-west linearity in the topography is
an unavoidable artifact of the optical scanning process the USGS uses to create the DEM
data sets. Additionally, merging the Oscoda and F oote DEM’s was complicated because
the west boundary of the Oscoda data set was approximately 15 feet lower than the
abutting east boundary of the Foote data set. Comparison to paper copies of the
corresponding 7.5 minute topographic maps showed the Oscoda data to be inaccurate
along a portion of the west edge of its coverage. This was corrected by mathematically
adjusting this portion of the generated x,y,z data set. The topography of the lowermost
(no-ﬂow) model boundary, which represents the contact with the underlying clay unit,
was generated by using the Surfer program to create a contour surface from elevation
data selected from local boring logs. GMS was used to vertically divide the model into
four layers by interpolation between the upper and lower model surfaces at each cell. An
overall horizontal cell size of 120 by 120 feet was employed with a reﬁnement to 20 by
20 feet in Fire Training Area 2 (F TA-2), located in the south-central portion of the model.

A reﬁnement factor of 0.2 was utilized in the transition between the large and small-sized

112

cells. The model consists of 265 rows by 255 columns by 4 layers for a total of 270,300
cells, approximately 60 percent of which are active, representing roughly ten square
miles. Sixty-nine observation wells located at FTA-2 were imported using their surveyed
coordinates and a “screened” elevation equivalent to the base of their actual screened
intervals.

Hydraulic parameters were initially assigned based on previously collected data
from aquifer tests, grain-size analysis and geologic reports; and then reﬁned during
model calibration exercises. Horizontal and vertical hydraulic conductivity was
estimated at 120 ﬁ/day (U SGS, 1990). An organic contaminant source area (Corg) was
added to layers 2 and 3 in the vicinity of the concrete pad (see Figure 4.1) at a
concentration of 0.001 (mol/L) based on measured total hydrocarbon concentrations.
Due to the nature of the contamination activities, it is highly unlikely that a single source
area exists but rather that it is likely comprised of several smaller source areas that
combine to contaminate the region directly surrounding the concrete pad. In addition to
the carbon source, average measured background concentrations (determined from wells
outside of the contaminant plume) were assigned to each cell in the model as shown in
Table 4.1. In the initial model, Fe3 was assigned as an immobile phase to represent solid
phase Fe-oxyhydroxide minerals. The ﬂow model tracked these ten components (Corg,
02, N03, N02, NH4, 304, 32', CH4, Fe“, Fe“) for the duration ofthe simulation.

Table 4.1 Initial concentrations assigned in ﬂow model mg/L)
Species 02 N03 N02 NH4 804 82 CH4 Fe2 Fe3K

 

 

Input 8.0 10 0.1 0 10 O O 0 20

 

 

 

 

 

 

 

 

 

 

 

 

*immobile phase

113

Reactive transport model

The reactive transport package RT3D was used to simulate the kinetically
controlled (rate-limiting) reactions occurring in the aquifer. Monod kinetics was used to
describe the fermentation of larger organic molecules (Cars) to smaller fermentative
products such as acetate and H2 (Cm), which can serve as electron donors for the TEAPS.
In the initial stage (Phase I) of the modeling, all of the TEAP reactions are considered
instantaneous when compared to the slow rate of fermentation and therefore were
evaluated and constrained by thermodynamic arguments in PHREEQC. The

fermentation of organic matter was represented by the equation:

d0... [6... J
+ C

donor org

_ —k onorX onor
dt d d Ks

 

 

- (1)
J

where C0,g is the total amount of organic matter in the aqueous phase (mol/L); k dam is

the maximum speciﬁc rate of fermentation ( 272 umol/mg of fermenting biomass-day);

X is biomass available for fermentation (0.003 mg); and stono, is the half-velocity

donor
coefﬁcient (14.1 umol/L). Initial estimates of constants were averaged from microcosm
studies (F ENNELL and GOSSETT, 1998). Aﬁer calculating the amount of fermentation
products (Cm) produced, RT3D calculates the total N, S, and Fe present in each cell and
gives this information to PHREEQC where these parameters are speciated according to

biogeochemical reactions that consider the total geochemistry of the system.

114

Geochemical model

A number of geochemical modeling programs exist for predicting inorganic
aqueous speciation and saturation based on chemical equilibrium principles, for example
WATEQ (TRUESDELL and JONES, 1974), GEOCHEM (Sposno, 1980), and MINTEQ
(FELMY et al., 1983). For our model simulations, we have chosen to use PHREEQC
because it can also calculate reversible reactions that include complexation,
mineral/amorphous solid, gas, solid solution, surface complexation, and ion-exchange
reactions including those that are irreversible reactions. The latter includes kinetically
controlled reactions, mixing of solutions, and temperature change. These additional
capabilities allow for a more realistic evaluation of hypotheses regarding the evolution of
redox state in contaminant plumes by considering a greater number of possible
geochemical interactions within the system. The initial PHREEQC chemical input ﬁle
for the model (Figure 4.6) consisted of average “background” solute concentrations at the
study site obtained from average concentrations just outside the contaminant plume. The
modeled solution was allowed to equilibrate with calcite since mineralogical studies
performed at the site permit this to be occurring (LONG et al., 2000), unpublished data
Madden, 2000). After the initial time step, total values of nitrogen (N), sulfur (S), and
iron (Fe) and the amount of electron donor (Cage) were assigned by RT3D. PHREEQC
then considers these new concentrations in conjunction with the rest of the water
chemistry, calculates chemical speciation and sends that output to MODFLOW where
transport processes are considered. Although Phase I simulations do not make use of all

the capabilities of PHREEQC (e. g., redox sensitive solid phase constraints) as they will

115

 

 

TITLE Jen 7.-Organic decomposition

 

SOLUTION 1
umts ppm
pH 7.22
density 1.0
temp 12.0
0(0) 8 02(3) -.75
Ca 150
Mg 10
Na 4
K 1
Fe 20
Mn 1
Si 8
a Cl 19
Alkalinity 150 as HCO3
S 27
N 10
EQUILIBRIUM_PHASES 1
Calcite 0.0
REACTION l
CHZO 1.0
XXX in 2 steps mmol
END

 

Figure 4.6: Initial PHREEC input ﬁle; aﬁer the initial time step the total
organic (CHzO), sulfur (S), nitrogen (N), and iron (Fe) are calculated by RT3D
and included in the input ﬁle as such

116

be in Phase II, establishing the linkage between PHREEQC and the commonly used ﬂow
and transport models MODFLOW and RT3D allows for much greater hypothesis testing

capabilities in future modeling efforts.

Preliminary Phase 1 Modeling Results

Before running the fully reactive 3-D PEA model, a sample l-D groundwater
simulation was run using PHREEQC to evaluate the effects of sequentially adding
electron donor (Cm) to our initial background water. Results show that as more electron
donor is added to the system, the TEAS present are consumed sequentially based on
thermodynamic arguments (Figure 4.7). Immediately as Cace is added to the system, the
most energetically favorable TEA oxygen [denoted 0(0)] is degraded. This is followed
by the degradation of N03 [denoted N(5)] and production of N2 initially (not shown) and
ultimately NH4 [denoted N(-3)]. After nitrate is degraded, iron oxides [denoted Fe(3)]
are reduced and Fe(2) is produced. This redox couple is followed by the consumption of
sulfate [denoted 8(6)] and the production of sulﬁde [denoted 8(2)]. These changes are
driven by changes in pe resulting from the input of electrons from the degradation of
organic matter. It should be noted that throughout this experiment, pH remained
relatively constant. Changes in pH were buffered by allowing the system to be in
equilibrium with calcite. These trends suggest that given appropriate organic input (from
MODFLOW and RT3D), the coupled model Should allow us to observe the 3-D

evolution of TEAPS and the resulting redox zonation in the aquifer.

117

 

m0_00n ﬁlmmmoo 0:65 <m_:0m

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

lxl zAmv lxl 2va lol ZA-wv It! 29 ll 2.3
III 0on 1?! mmANv tol $.0va I?! DI It] cm
which No
«.mtoA E lllll
U1 _ .0... it Am
U amt?“ Bun-t
m which i- ._o
(\ Pmtoh .
m 9m? - -- w
W. NmbA H Lu
W o no no mo 00 mo mo No mo
C

.230 A030.

Ema—.0 0.3 wiﬂmmOO 900332: 0030—0 0:65 8253i 0 00:08:” 52.: 0». £028: 00008 Amway.

pH and pe

118

The coupled 3-D reactive ﬂow and transport model was then run for 365 days to
determine if TEAPS would be initiated in the aquifer as a result of the contamination
input and to test the linkages between the three modeling codes. It should be noted that
in this initial model, RT3D was not calculating the total N, Fe, and S but rather
PHREEQC was reading the initial total input of N, Fe, and S (see Table 1) and speciating
according to the electrons produced during the fermentation step. The model output for
each of the chemical components was plotted in 3-D using Tecplot. Two-dimensional
slices through layer 2 were created from these 3-D images for ease of viewing the plume.

Figure 4.8 shows the evolution of organic matter (Corg) in the aquifer at 10, 200,
and 364 days. The initial condition of the aquifer is assumed to have a very low
concentration of organic matter. Inputs of contaminant (Corg) were then added to a
discrete “source” area and transported over time. As a result, the majority of the ﬁeld
shown in Figure 4.8 is blue corresponding to areas without Corg inputs and a smaller
distinct plume of increased Corg can be seen developing down gradient of the designated
source area. After 364 days, a contaminant plume roughly 150 feet wide by 1500 feet
long has developed. As the plume develops, we would expect that the Corg would be
fermented into usable electron donors (Cm) which could then be oxidized coupled to the
reduction of oxygen (the most favorable TEA). This would result in decreased oxygen
concentrations within the plume. Figure 4.9 demonstrates that over the 365-day
simulation, oxygen is depleted to concentrations nearing zero within the plume
presumably driven by the production of usable electron donors. This demonstrates that
the linkages between the three modeling codes are working such that the important

hydrogeochemical controls of each portion of the model are operating effectively. Based

119

Organic Matter (Corg) after 10 days (mo/L)

3.45504
1 38E-O4

O DOE-00

Organic Matter (Corg) after 200 days (mol/L)

3 45E<04

1 38E-04

O OOE-OO

."".H H l{\

Organic Matter (Corg) after 365 days (moI/L)

9 66E-04
7 59E-O4
5 52E-04
3 45E—O4
1.38504

0 OOE-OO

 

Figure 4.8: Organic Matter (C0,!) concentration (mol/L) at 10, 200, and 364 days

120

Oxygen after 10 days (moi/L)

IL“! 2H- 20 lb

21:” H.111

Oxygen after 200 days (moi/L7)

1 932-64
1525—04
1.105104 ,
5,905.05.

' 2.76505
0,005.00

1““: “1.11 14.8
~ It 2"!“ F11] 'l

Oxygen after 364 days (molIL) '

1.93504

llllllll

11,522.04
riots-04.
egos-05
2768-05
doors-00 I

If." .‘xll :l’llS

 

x.
t" :W‘l'l-li‘

Figure 4.9: Oxygen Concentration (mol/L) at 10, 200, and 364 days

121

 

 

u
r
--A-->-v~

on thermodynamics, after the oxygen has been consumed within the plume, N03 (the
second most favorable TEA) will be reduced. Figure 4.10 shows that N03 concentrations
decreased within the plume but not to concentrations below which N03 reduction cannot
be sustained (ISLAM et al., 2001). This suggests that at 365 days, the plume has reached a
state of N03 reduction. To evaluate the poised redox state at this point in time
ammonium (NH4) and sulfate (SO4) concentrations were examined (Figure 4.11). The
lack of NH4 production (concentrations near zero throughout the aquifer, denoted in blue)
and lack of SO4 reduction (high initial concentrations present throughout aquifer, denoted
in red) suggests that the system is not reduced enough to initiate those reactions and that
N03 reduction iS poised to produce N2 gas (DREVER, 1997).

These results suggest that the three modeling codes are communicating
appropriately and that the model is going through a sequential TEAP ladder based on
thermodynamic arguments. However, the rate at which the sequence of traditional redox
zones will be achieved is constrained by the kinetics of the fermentation of organic
matter. Allowing the evolution rate of the system to be limited by fermentation is
reasonable during the initial stages of redox zonation. However, once the system has
reached a state where fermentative products can build up (e. g. methanogenesis), it is no
longer appropriate to assume that fermentation is the rate-limiting step and when TEAS
are introduced to such a system (c. g. during recharge) their rates of reduction, although

comparatively fast, (see Chapter 3) become the rate limiting factor.

122

 

 

 

Nitrate after 10 days

1.93E-O4

1525-04
1,105-04
6.90E—05
2,705-05
0.001-:00
.. manna:

l~-~r—~J

.1 Furl-[Fl

Nitrate after 200 days (mollL)

1.93E-O4

1 325-04
1.1 01504
5.90505
2705.05
0005.00

Nitrate after 364 days (moi/L)

1.93E-O4

1 .5212-04
1 . 105.04
090505 .
2765205
0005-00

 

Figure 4.10: Nitrate Concentration (mollL) at 10, 200, and 364 days

123

Ammonium Profile after 364 days (mol/L)

Sulfate Profile after 364 days (mol/L)

Figure 4.11: Ammonium and Sulfate Concentrations (mollL) at 10, 200, and 364 days

 

o. ..
' . .
' a
‘v n. . .' .
. a,
' I ‘ - . u
I u
‘ ‘ .. ‘
.
. l
.

 

Future Work

Preliminary results demonstrate that combining ﬂow and transport codes with
geochemical modeling codes can provide a powerful tool to evaluate hypotheses
regarding the evolution of TEAPS in organic groundwater contaminant plumes. The
immediate next step is to continue to run the model for longer periods of time until it has
reached a “mature” state where plume length has stabilized. This will allow us to test the
hypothesis that under such conditions, redox zones will not be sustained but rather TEAS
will continue to be degraded sequentially until the plume has reached a state of
methanogenesis. Once this hypothesis has been tested, the next step is to increase the
geochemical complexity of the model such that we can incorporate realistic, redox
sensitive solid phases (e.g., pyrolusite (Mn-oxide), Fe-oxyhydroxides, and pyrite). After
these steady state hypotheses have been tested (Phase 1), Phase II will examine the more
realistic scenario evaluating changes due to recharge introducing TEAS to a mature
system. A system that has reached redox maturity will have a build up of fermentative
products (acetate and H2) and thus, it is no longer reasonable to assume that TEAP
reaction rates will be limited by fermentation reactions as during the initial evolution.
Therefore, the rates of TEA consumption become signiﬁcant. Rate estimates obtained
from push-pull tests (see Chapter 3) will be used to constrain these processes by

incorporating them into RT3D.

125

 

References

Abrams R. H. and League K. (1998) Development and testing of a
compartmentalized reaction network model for redox zones in contaminated
aquifers. Water Resources Research 34(6), 1531-1541.

Baedecker M. J., Cozzarelli I. M., Eganhouse R. P., Siegel D. E., and Bennett P. C.
(1993) Crude oil in a Shallow sand and gravel aquifer-III. Biogeochemical
reactions and mass balance modeling in anoxic groundwater. Applied
Geochemistry 8, 569-586.

Bjerg P. L., Rugge K., Pedersen J. K., and Christensen T. H. (1995) Distribution of
Redox-Sensitive Groundwater Quality Parameters Downgradient of a
Landfill (Grindsted, Denmark). Environmental Science and Technology
29(No. 5), 1387-1394.

Brun A. and Engesgaard P. (2002) Modelling of transport and biogeochemical
processes in pollution plumes: literature review and model development.
Journal of Hydrology 256, 211-227.

Buscheck T. E. and Alcantar C. M. (1995) Regression techniques and analytical
solution to demonstrate intrinsic bioremediation. In Intrinsic Bioremediation
(ed. R. E. Hinchee), pp. 109-116. Battelle Press.

Champ D. R., Gulens J., and Jackson R. E. (1979) Oxidation-reduction sequences in
groundwater flow systems. Canadian Journal of Earth Sciences 127, 85-108.

Chappelle F. H. (2001) Ground- Water Microbiology and Geochemistry. John Wiley

and Sons, Inc.

126

 

Chappelle F. H., Haack S. K., Adriaens P., Henry M., and Bradley P. M. (1996)
Comparison of Eh and H2 Measurements for Delineating Redox Processes in
a Contaminated Aquifer. Environmental Science and Technology 30(No. 12),
3565-3569.

Clement T. P. (1997) A modular computer model for simulating reactive multi-
species transport in three-dimensional groundwater systems, Pacific
Northwest National Laboratory.

Clement T. P. and Jones N. L. (1998) RT3D Tutorials for GMS Users, Paciﬁc
Northwest National Laboratory, PNNL-11805, Richland, Washington.

Cozzarelli I. M., Bekins B. A., Baedecker M. J ., Aiken G. M., Eganhouse R. P., and
Tuccillo M. E. (2001) Progression of natural attenuation processes at a
crude-oil Spill site: I. Geochemical evolution of the plume. Journal of
Contaminat Hydrology 53, 369-385.

Cozzarelli I. M., Suﬂita J. M., Ulrich G. A., Harris S. H., Schroll M. A., Schlottmann
J. L., and Christensen S. (2000) Geochemical and Microbiological Methods
for Evaluating Anaerobic Processes in an Aquifer Contaminated by Landfill
Leachate. Environmental Science and Technology 34(No. 18), 4025-4033.

Dojka M. A., Hugenholtz P., Haack S. K., and Pace N. (1998) Microbial diversity in
a hydrocarbon- and chlorinated solvent-contaminated aquifer undergoing
intrinsic bioremediation. Applied and Environmental Microbiology 64, 3869-
3877.

Drever J. I. (1997) The Geochemistry of Natural Waters: Surface and Groundwater

Environments. Prentice Hall.

127

EPA U. S. (1997) Use of monitored natural attenuation at superfund RCRA, and
underground storage tank Sites. US. EPA.

Essaid H. 1., Bekins B. A., Godsy E. M., Warren E., Baedecker M. J., and Cozzarelli
I. M. (1995) Simulation of aerobic and anaerobic biodegradation processes at
a crude oil spill site. Water Research 31, 3309-3327.

Felmy A. R., Girvin D. C., and Jenne E. A. (1983) MINTEQ: A computer program
for calculating aqueous geochemical equilibria. Final Report, EPA Contract
68-03-3089.

Fennel! D. E. and Gossett J. M. (1998) Modeling the production of and competition I L
for hydrogen in a dechlorinating culture.

Haack S. K. and Bekins B. A. (2000) Microbial populations in contaminant plumes.
Hydrogeology Journal 8, 63-76.

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 Resources Research 28,
201 1-2027.

Hunter K. S., Wang Y., and Van Cappellen P. (1998) Kinetic modeling of
microbially-driven redox chemistry of subsurface environments: coupling
transport, microbial metabolism and geochemistry. Journal of Hydrology
209, 53-80.

Islam J., Naresh S., and O'Sullivan. (2001) Modeling biogeochemical processes in
leachate-contaminated soils: a review. Transport in Porous Media 43, 407-

440.

128

J akobsen R. and Postma D. (1999) Redox Zoning, rates of sulfate reduction and
interactions with Fe-reduction and methanogenesis in a Shallow sandy
aquifer, Romo, Denmark. Geochimica et Cosmochimica Acta 63(No. 1), 137-
151.

Keating E. H. and Bahr J. M. (1998) Reactive transport modeling of redox
geochemistry: approaches to chemical disequilibrium and reaction rate
estimation at a site in northern Wisconsin. Water Resources Research 34(12),
3573-3584.

Konhauser K. O. (1994) Mineral Precipitation by Epilithic Biofilms in the Speed
River, Ontario, Canada. Applied and Environmental Microbiology 60, 549-
553.

Langmuir D. (1997) Aqueous and Environmental Geochemistry. Prentice-Hall, Inc.

Lensing H. J ., Vogt M., and Herrling B. (1994) Modeling of biologically mediated
redox processes in the subsurface. Journal of Hydrology 159, 125-143.

Levine A. D., Libelo E. L., Bugna G., Shelley T., Mayﬁeld H., and Stauffer T. B.
(1997) Biogeochemical assessment of natural attenuation of JP-4
contaminated ground water in the presence of ﬂuorinated surfactants.
Science of the Total Environment 208(3), 179-195.

Long D. T., Hyndman D. W., Haack S. K., Klug M. J., and Velbel M. A. (2000)
Understanding and modeling hydrogeological, microbiological, and
geochemical processes that control groundwater redox zonation-Year 2

Progress Report. National Science Foundation.

129

 

 

Lovely D. R. and Phillips E. J. P. (1987) Competitive mechanisms for the inhibition
of sulfate reduction and methane production in the zone of ferric iron
reduction in sediments. Applied and Environmental Microbiology 53, 2636-
2641.

Lovley D. R. (1991) Dissimilatory Fe(III) and Mn (IV) reduction. Microbial Reviews
55, 259-287.

Lovley D. R. and Goodwin S. (1988) Hydrogen concentrations as an indicator of the
predominant terminal electron-accepting reactions in aquatic sediments.
Geochimica et Cosmochimica Acta 52, 2993-3003.

Lyngkilde J. and Christensen T. H. (1992) Redox zones of a landfill leachate
pollution plume (V ejen, Denmark). Journal of Contaminant Hydrology 10,
273-289.

McDonald M. G. and Harbaugh A. W. (1988) A modular three-dimensional f‘mite—
difference ground-water flow model. US. Geological Survey Water
Techniques of Water-Resources Investigations Report 06-A1.

McGuire J. T., Smith E. W., Long D. T., Hyndman D. W., Haack S. K., Klug M. J .,
and Velbel M. A. (2000) Temporal variations in parameters reﬂecting
terminal-electron-accepting processes in an aquifer contaminated with waste
fuel and chlorinated solvents. Chemical Geology 169, 471-485.

McNab W. and Rice D. (2001) Ascertaining the effect of reductive dehalogenation
on chlorinated hydrocarbon plume lengths in groundwater: Analyses of

multisite data. Soil & Sediment Contamination 10(1), 1-19.

130

McNab W. W. (2001) A Monte Carlo simulation method for assessing
biotransformation effects on groundwater fuel hydrocarbon plume lengths.
Computers & Geosciences 27(1), 31-42.

McNab W. W. and Dooher B. P. (1998) A critique of a steady-state analytical
method for estimating contaminant degradation rates. Ground Water 36(6),
983-987.

McNab W. W. and N arasimhan T. N. (1994) Modeling reactive transport of organic
compounds in groundwater using a partial disequilibrium approach. Water
Resources Research 30(9), 2619-2635.

Neale C. N., Holder A. W., Ward C. H., and Hughes J. B. (2002) Groundwater
Reaeration and Hydrocarbon Plume Length: A model analysis. Journal of
Environmental Engineering 128(1), 40-50.

Parkhurst D. L. and Appelo C. A. J. (1999) User's guide to PHREEQC (Version 2)--
a computer program for speciation, batch-reaction, one-dimensional
transport, and inverse geochemical calculations: US. Geological Survey
Water-Resources Investigations Report 99—4259, pp. 312.

Parkhurst D. L., Kipp K. L., and Engesgaard P. (2001) PHAST—A program for
Simulating ground-water ﬂow and multi—component geochemical reactions.

Peterson J. N. and Sun Y. (2000) An analytical solution evaluating steady-state
plumes of sequentially reactive contaminants. Transport in Porous Media 41,

287-303.

131

 

Postma D. and J akobsen R. (1996) Redox Zonation: Equilibrium constraints on the
Fe(III)/SO4-reduction interface. Geochimica et Cosmochimica Acta 60(No.
17), 3169-3175.

Reynolds L. A. and Haack S. K. (1999) Evaluation of RNA hybridization to assess
bacterial population dynamics at natural attenuation sites. Marganwalp DW,
Buxton HT (eds) US Geological Survey toxic substances hydrology program.
Proc Technical Meeting, Charleston, South Carolina, 8-12 March 1999. US
Geological Surv Water-Resour Invest Rep 99-4018C, 635-639.

Salvage K. M. and Yeh G.-T. (1998) Development and application of a numerical
model of kinetic and equilibrium microbiological and geochemical reactions
(BIOKEMOD). Journal of Hydrology 209, 27-52.

Smith E. W. (2002) Groundwater ﬂow dynamics at Wurtsmith Air Force Base,
Oscoda, Michigan. MS, Michigan State University.

Smith E. W., Hyndman D. W., and Long D. T. (1999) Simulated plume evolution
using transient ﬂow conditions observed at Wurtsmith AFB, Michigan. EOS,
American Geophysical Union Spring Meeting Supplimental Issue-April 27 80,
$137.

Sposito G. (1980) GEOCHEM: A computer simulation for the calculation of
chemical equilibria in soil solutions and other natural water systems. Kearny
Foundation, University of California, Riverside, California.

Stumm W. and Morgan J. J. (1981) Aquatic Chemistry. John Wiley and Sons.

132

Suna Y., Peterson J. N ., and Bearc J. (2001) Successive identification of
biodegradation rates for multiple sequentially reactive contaminants in
groundwater. Journal of Contaminat Hydrology 51, 83-95.

Tebes-Stevens C., Valocchi A. J ., VanBriesen J. M., and Rittmann B. E. (1998)
Multicomponent transport with coupled geochemical and microbiological
reactions: model description and example simulations. Journal of Hydrology
209, 8-26.

Truesdell A. H. and Jones B. F. (1974) WATEQ, a computer program for
calculating chemical equilibria of natural waters. J. Res U.S. Geol. Survey 2,
233-274.

USGS. (1990) Installation Restoration Program: Phase II-
Confirmation/Quantiﬁcation Stage 1: Wurtsmith Air Force Base, Michigan:
Hydrogeology near Wurtsmith Air Force Base, MIchigan. United States
Geological Survey-Water Resources Division.

USGS. (1991) Installations Restoration Program: Phase II-
Conﬁrmation/Quantiﬁcation Stage 1: Wurtsmith Air Force Base, Michigan:
Investigations of groundwater and soil contamination at selected sites.

Vroblesky D. A. and Chappelle F. H. (1994) Temporal and spatial changes of
terminal electron-accepting processes in a petroleum hydrocarbon-
contaminated aquifer and the signiﬁcance for contaminant biodegradation.

Water Resources Research 30(No. 5), 1561-1570.

133

West T. G. (1999) Microbial ecology of a contaminated aquifer undergoing natural
attenuation at the Wurtsmith Air Force Base, Michigan. MS, Central

Michigan University.

134

CONCLUSION

A conceptual model for the evolution of redox zones in aqueous systems has been
proposed. This model makes the distinction between systems in which the rate of TEAPS
is controlled by the availability of electron donors (from fermentation) and those in which
there iS an accumulation of fermentative products and thus the rate of TEAPS is
controlled by the availability of electron acceptors. When a contaminant plume is
initially evolving, TEAPS will be limited by the concentration of readily available
electron donor from fermentative reactions. As the plume becomes more reduced
fermentative products begin to accumulate in the system, removing the donor limitation
on the TEAP rates. AS a result, reactions can proceed at faster rates as was observed
during the ﬁeld experiments. Newly introduced TEAS may also be consumed
Simultaneously under these conditions because there is no competitive exclusion between
TEAPS given low TEA input concentrations (e.g., typical concentrations present in
precipitation). Field experiments also demonstrated that the reduction of TEAS observed
in the aquifer did not occur in the reduced aquifer water without the inﬂuence of the solid
phase portion of the aquifer and its associated microbial community. It was also shown
that oxidized species such as N03 and S04, can persist even in highly reduced waters
where thermodynamics predict they would be reduced. This implies that when modeling
anaerobic systems, strictly thermodynamic geochemical arguments may not always be
appropriate. The geochemistry must be considered in conjunction with transport

processes and microbiological properties of the system.

135

These ﬁndings collectively imply that in a highly reduced system, much of
the important TEAPS will be initiated rapidly when TEAS are introduced to the
reduced aquifer. Under this scenario, the majority of the TEAP variability will
occur in small discrete interface zones near the boundaries of the plume. This
idea is supported by graphical and multivariate statistical analyses performed on
geochemical data obtained before and after a major recharge event (from Spring
snow melt). Statistical results demonstrate that although wells with similar
chemistry are spatially distributed near one another (denoting distinct water
masses in the system), the variability is not related to distinct redox zones in the
system. Examination of small-scale (Single well) changes suggests that the spatial
areas over which recharge has the greatest impact on TEAPS are small (less than 3
feet). It is likely that during recharge events most of these changes occur near the
water table-capillary ﬁinge interface where abundant donor and microorganisms
exist. Unfortunately, due to the small and transient nature of these zones, directly
sampling where the greatest redox variability iS occurring very difﬁcult. In
addition, our observations of geochemical changes over 5 years suggest that even
when sampling within a transition zone where TEAPS are occurring, ﬁnding an
appropriate indicator of that in-situ process may be complicated by redox
disequilibria and additional microbial reactions which impact TEAP indicating
variables such as H2.

These ﬁndings were utilized to develop a three-dimensional partial
equilibrium ﬂow and transport model to test hypotheses regarding redox

evolution. The conceptual model was veriﬁed by incorporating its key elements

136

and linkages in a 3-D ﬂow and transport model, which successfully represented
the early stages of plume evolution. This model will be used in the near future to
quantify biogeochemical changes due to recharge and will incorporate additional
real-world geochemical complexities by including realistic redox-sensitive solid
phases (e.g., pyrolusite (Mn-oxide), Fe-oxyhydroxides, and pyrite). The outputs
of these modeling efforts will not only improve our basic understanding of redox
evolution in natural and contaminated systems but will also aid in the design and

evaluation of bioremediation and natural attenuation strategies.

137

APPENDICES

138

APPENDIX A.

Procedure for Determination of Hydrogen Gas Concentrations Using the RGA-3

139

f

Set-Up

From a cold start-up, the RCA-3 requires at least 2 hours to stabilize prior to
sample analysis. The RGA—3 frequently takes longer than 2 hours to stabilize,
especially following transportation to the sampling site. Recommendation: set up
the RGA-3 at least 24 hrs. in advance of sampling.

1. Set analyzer on a stable surface protected from rain, direct sun, etc. Use a
power-strip with a surge suppressor to connect RGA-3 to power.

2. Remove plugs and caps; make sure to remove red cap from mercury scrubber
and plug from sample “out” port

3. Hook up connections to integrator (printer):

a) on integrator (back): red -) Ch A+ , black -) Ch A-
b) on RGA-3 (back): red -) 1 V out (off set) , black -) GND

4. Attach regulator to N2 cylinder; ﬂush regulator thoroughly with N2 and then
ﬁll 1-2 Tedlar bags (you will need these to make your dilutions)

5. Attach carrier gas lines, sample injection port

6. Adjust regulator to approximately 100 kPa (~20psi)

***Never turn on the detector without carrier ﬂow. In the absence of carrier
ﬂow, Hg evolved from the oxide bed may plate out onto the detector optics.

7. Turn on power switches to RGA-3 and integrator (on backs) and turn on the
RGA-3 lamp (switch located on front face).

8. Turn on heaters for column and detectors
Press “TEMP SEL” button on front panel to toggle between COL (column)
and DET (detector); press “TEMP ON/OFF” button to turn heaters on/off.
Make sure the “ON” light comes on when you turn on the heaters.

9. After 1.5 hours, turn on the lamp. You can remove the top cover of the RGA-
3 and observe a faint blue glow from the lamp housing to conﬁrm operation.
Wait at least two hours before proceeding to step 8.

10. Press “SET/AT” button to toggle between the set temperature and actual
temperature. Record set/at values for both column and detector in the logbook
(used to monitor integrity of thermocouples). ***Actual values should be
within 5 ‘C of set value and stable.

11. Press “TEST” repeatedly and toggle through the x1, x2, x4, x8, x16, and x32
positions. Record the values displayed on the LED during each position in the
logbook. These values indicate lamp output, baseline levels, etc. (see manual
for description)

12. Press “RANGE” to get out of test mode; continue to press “RANGE” until
you reach the x32 position (indicated by light); use this setting for analyses.

13. On printer, press “LEVEL”. Record initial value in logbook. On RGA-3,
press “ZERO”, then press “LEVEL” again (value Should drop to either 1000
or1500)

14. On printer, press “PT EVAL”. Wait ~ 1 min. for it to take reading and print
value. Record value in logbook. (Values greater than 500 indicate the
column might be contaminated and require reconditioning.)

15. Proceed with calibration.

 

 

140

Standby Mode

If running samples on consecutive days .....

1. At the end of the day, turn off the UV lamp, but leave the heaters on, and
leave carrier gas ﬂow on.

2. At the start of the next day, turn on the UV lamp and wait 20-30 minutes to
stabilize.

3. Carry out steps 10-15 in Set-Up.

Calibration/Sample Analyses

Record results from all analyses in logbook...

1. Check operation of valve by completing one cycle. Valve initially should be
in “load”

position (check light). On RGA-3, press VALVE to switch valve to the

“inject” position

(watch light, listen for sound of valve switching). Wait 1-2 secs, then

press VALVE

again to revert back to “load” position (watch light, listen for sound of

valve switching).
2. Analyze Lab Air:

a) Using 25 ml gas-tight syringe, open valve of syringe and push/pull
plunger several times to ﬂush the syringe

b) Fill syringe with lab air and close valve

c) Connect syringe to sample port of RGA-3, open valve on syringe, and
slowly inject 7.5 ml of sample into RGA-3.

****Sample loop capacity is ~ 1 ml; you want to inject ~6x
volume of sample 1 loop to properly flush it.

(1) Close valve on syringe; press VALVE on RGA-3 (to switch valve to
“inject” position); press INJ/END A on integrator to start run, press
start on timer (or mark your stopwatch).

e) (at 20 psi) Wait 10 seconds, then press VALVE on the RGA-3 (to
switch the valve to the “load” position). ***Be sure not to let valve
run longer: you may introduce some nasty organics which can poison
the mercuric oxide bed, reducing its efﬁciency.

0 Wait for H2 and CO peaks to come out. Retention times are
approximately 0.27 min and 0.45 min, respectively. Once they have
eluted, press INJ/END A on integrator to stop the run and calculate
peak areas.

g) Wait at least 30 seconds before running another replicate. ***You
should wait at least 3x the elution time (R I) of your analyte of interest
in order to suﬂiciently baclgﬂush the ﬁrst column.

h) After 30+ seconds, proceed with next replicate analysis of lab air.

3. Analyze N2 carrier gas (from Tedlar bags). This will serve as your calibration
blank.

l4l

d)
6)

Attach a 22 gauge needle (or ﬁner) to gas-tight syringe; insert needle into
septum ﬁtting of Tedlar bag, open valve on syringe and withdraw several
ml’s of N2. Close valve on syringe.

Remove syringe ﬁom Tedlar bag. Open valve and completely expel the

gas; close valve when ﬁnished (this serves to “ﬂush” the syringe).

Insert needle into septum ﬁtting of Tedlar bag and withdraw 25 m1 of N2;
close valve.

Remove needle from syringe adapter.

Repeat procedure outlined in steps 2c-2h (above) for N2.

4. Standards
You generally do not see H2 concentrations greater than 10ppm (8. 0 nM) in
nature, so this can serve as a preliminary upper bound for calibration.

3)
b)

c)
d)

Fill Tedlar bag with a certiﬁed H2 standard (generally 100ppm H2 in N2).
Prepare 25 ml of 10ppm H2. Attach needle to gas-tight syringe and
withdraw several m1 of H2 standard from Tedlar bag. Remove syringe
from Tedlar bag and expel the gas to ﬂush syringe. Withdraw another
aliquot of gas (several ml) from Tedlar bag and close valve. Hold syringe
vertically, open valve and slowly but steadily expel gas until plunger
reaches the 2.5 ml mark, then close valve. Withdraw enough N2 ﬁom the
Tedlar bag you ﬁlled with the carrier gas to ﬁll the syringe to 25 ml (this is
now your 10ppm H2 standard). Recommendation: If you prepare 25 ml
of 1 Oppm standard and inject 7.5 ml per replicate, you can run three
replicates and still have 2.5 ml left. You can then prepare a 1:10 dilution
of the remaining gas to make your 1 ppm H2 standard.

Analyze according to procedures outlined above.

In the same manner, prepare and analyze Ippm, 0.1 ppm, 2ppm, and
0.2ppm H2 standards. You can also prepare 4 and 0.4ppm standards if
desired.

5. Sample collection

Once well has been purged for at least 3 well volumes, connect pump
tubing to gas-stripping bulbs. Tip upright and ﬁll with groundwater.
Return bulb to online position and inject three syringe-fulls of N2 gas into
the bulb to create an inert headspace. Allow at least 15 minutes before
collecting ﬁrst sample. *Note that increasing the rate of groundwater ﬂow
will decrease the time necessary to reach equilibrium.

6. Running samples

Once you have successfully calibrated, you are ready to run samples.
Where possible, remove enough gas from the gas-stripping bulbs so that
you can run three replicates per sample. Continue to run samples until
values stabilize.

142

7. Checking calibration
a) The RGA-3 does have a tendency to drift, and the calibration should be

h)

checked every 3-4 hours. You should also check the calibration if: 1)
there is a large Shift in the baseline/integrator response, or 2) hydrocarbons
have been injected into the RGA-3 (e.g. because valve was left in “inject”
position for too long.

Check calibration by preparing and analyzing 10ppm and Ippm H2
standards. If results differ ﬁom previous calibration by more than 10%,
make a new suite of standards and go through a complete calibration.

Shut Down/Pack Up

1.
2.

9999

Turn off UV lamp and the heaters for both the detector and column.
Once temperatures are at ambient levels (i.e. detector temperature MUST
be less than 100°C, but preferably below 50°C), turn off carrier gas ﬂow.
Recommendation: Increase carrier gas flow rate (4 0-60 psi) to cool
detector more quickly.

Turn off power to the integrator and RGA-3.

DisConnect integrator.

Remove sample inlet port and carrier gas line from RGA-3.

Insert plugs into all open ports to keep particulates from entering the
RGA-3; replace red cap on mercury scrubber.

Pack RGA-3 and integrator into silver carry case; make sure all sides are
securely fastened.

143

APPENDIX B.

Timeline of Project Events

144

.02: 2:... .2: .32 .325
.32 .385 .e E: 2832. 0.22.285 .08. .335 5 wow: 2.. E 32.... .332. .. .5 8...... is. 2.5....

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

. éaoaéceeo..8..o .. _. .2... :2; 5.0 .2: 58 2::
3.2 .2... 32 2. 2.2.8.0.. m .2; .3222... Siegel
G22... .02 2... 4.82.2 82.2 00 0:... a 22.5.... :2; .3222... 881:3. ..<
u .- 2... N 2... N 2...: .3222... 82 N2...
.82.. .02 a... 400st 82.2 00 3.2 .5 .0023...» 2:25; .a .. a... . .2; ._....-......... 2.2 2:;
«N0: 3.3 set. o. 2; 22.2.8 22.2 £2.23. 00%?
«N3 2...... SEN 2: 22.250 as; 5882 momasms. 82 22:5
. . Nae 2...;ch
ﬁ.c=.0u-><.>: . 3.9:. 32 E9... 039:0. EoEEom +55 .2075: 0.80:0: Mao .00 vacuum
«0:02 302.3 .9“: ES 3: 3 0...... uczgmm Eco... c0083. mOmDSwE lag—gal
2:30.332... :2. 2.2.. 32 32. co F.22....22. 82 .2. _.<
2.22:... v32.....2. 8.. 22.. N... 8.. 28 2.0.2 5. 32.28.25.220 82 2.;

 

 

$-ka B .. 5:3. 35.55%

 

 
 
 

3.00052. $3.8 0203 REM N. t... E. mm Emu... someway— wOmQDw—z

lama 2:...

 

230 £2.02... 0822. 2.6 .3 a...

 

 

2.2.2. .02 a... 85. 32 .8 82.2; .2; .3222... 2.2.2..

 

3:0 was: 5.2% c at... “5.30m E00... .3083. m0 mDSmE
Ammo—:32 0.9“. 3532 E0... coo—S 5.52.0- 080: 03. .9233. 8000.022

   

 

23— N02
000— H «P..—uh

 

 

«.333» 020.50; 3 Tu: ~28 MEN: 5 Emma} . 5.2-5.

 

c are 22.8.8 as; 5832 03:50.).

 

m a; 22.8.8 E3 5832 000050. 2

2.... 22.2532
NS. 2.2:.

 

30E 0. «03 VS. 20.08.83 0.: {3.95 .633 2.0.0302...» kkm 0.2.: :55

 

 

 

 

 

 

 

 

wm-hN~-b..-c «we... .30.... 02520 no 2:...
x. .0 0.: «050...: 23:02.6 3: 3.0 2.0.3: .03 «3.0:: $.03 mutatu kkm 0E: RAPE
woAm _ .86 $0... .008... 02:55 wa 0:3.
v EC. M.;—925m E00... suaomom wOmQDmZ wa A02
«.500. «030.500.. 0. .05. 353:9. .0325:
m EC. mczgmm E005 somomux wOmQDmE was =..m<

 

@059. N: «ME .50 ESSENSE .36: .0: 5:: ..N£0.y\km .0 0:02 .30: SQQSXEN 31:3,... biﬁguomwx

 

 

N a: 22.22% 582 .823. mumasmz

@00— H «PEP.—

 

 

..aaN 8:22.. .2 2s 2... 8.8.. 8:22.. .2 2.3 so... 9.. s W..»; 8:22.. cam.

 

 

_ a... 22.9.8 see 58.2. 85st

 

Q53 .>0Z 3&0 00£ZU< Ewan. :anmom mUmEDwEM 2508; .

mam: baa—E.,

3.: .3522

ﬁnd. NEEmﬁ=>>

 

 

145

APPENDIX C.

Well Sampling History

146

Well Sam Ii H
uses 11/97 2/98 4/98 5/98 8/98 11/98 2199 5/99 10/99 woo 6/00

     

MM -1 7.9

ML] ~20 X
MLl-22.5 X
ML1-24.5 X
MLl-25.3

ML2-20.5

ML3-17.81
ML3-18.31
ML3-22.00
ML3-25.03
ML3-27.99
ML3-32.78
ML3-37.59

><><><><><><><
><><><><><><><
X><><><><><><
XXXXXX

><

MIA-17.50
MIA-17.75
ML4-21.44
MIA-24.44
ML4-27.42

XXXXX
><
XXXXX
XXXXX

><

ML5-18.42
ML5—18.84
ML5-22.64

ML5-25.67
ML5-28.37
ML5-33.78
ML5-38.48

><X><><><><><
XXXXXXX
XXXXXXX
XXXXXXX

ML6-12.30
ML6-l 6.02
ML6-l 9.04
ML6-22.00
ML6-28.74
ML6-32.l l

ML8-l 8.70
ML8-l 9. l 5
ML8-22.80
ML8-25.85
ML8-28.80
ML8-33.65
ML8-38.55

XXXXXXX
><><><><><><><
XXXXXXX

X

PZ-7

PX-8

PZ-9
PZ-l 5

XX

MUG-36.99

147

Well

 

USGS 11/97 2/98 4/98 5/98 8/98 11/98 2/99 5/99 10/99 1/00 6/00
MLll-18
MLll-Zl
ML11-24
MLll-27

><><><><
><><><><

ML12-l8
ML12-21
ML12-24
ML12-27

><><><><
><><><><

ML13-18
MLl3-21
MLl3-24
ML13-27

><><><><
><><><><

ML14-18
ML14-21
MLl4-24
MLl4-27

><><><><
><><><><

ML15-18
ML] 5-21
MUS-24
MUS-27

><><><><
XXXX

MLl6-l 8
ML16-21
ML16-24
ML16-27

><><><><
><><><><

148