£EPA
     United States
     Environmental Protection
     Agency
ENVIRONMENTAL

RESEARCH  BRIEF

Long-term Performance of
Permeable Reactive Barriers
Using Zero-valent Iron:
An Evaluation at Two Sites
Richard T. Wilkin*, Robert W. Puls*, and Guy W. Sewell*
                          Background

                          The permeable reactive barrier (PRB) technology is an in-situ approach for remediating groundwater
                          contamination that combines subsurface fluid flow management with a passive chemical treatment
                          zone. Removal of contaminants from a groundwater plume is achieved by altering chemical
                          conditions in the plume as it moves through the reactive barrier.  Because the reactive barrier
                          approach is a passive treatment, a large plume can be treated in a cost-effective manner relative to
                          traditional pump-and-treat systems. There have now been more than forty implementations of the
                          technology  in the past six years, which have proven that passive reactive barriers can be cost-
                          effective and efficient approaches to remediate a variety of compounds of environmental concern.
                          However, in all of the installations to date comparatively few data have been collected and reported
                          on the long-term performance of these in-situ systems, especially with respect to the buildup of
                          surface precipitates or biofouling (O'Hannesin and Gillham, 1998; McMahon et al., 1999; Puls etal.,
                          1999; Vogan, 1999; Phillips et al., 2000; Liang et al., 2000).

                          A detailed analysis of the rate of surface precipitate buildup in these types of passive, in-situ systems
                          is critical to  understanding how long these systems will remain effective and what methods may be
                          employed to extend their lifetime or to improve their performance. Different types of minerals and
                          surface coatings have been observed to form under different geochemical conditions that are
                          dictated by  aquifer chemistry and the composition of the permeable reaction zone (Powell et al.,
                          1995; Mackenzie et al., 1999; Liang et al., 2000). Microbiological impacts are also important to
                          understand  in order to better predict how long these systems will remain effective in the subsurface
                          (Scherer et al., 2000). The presence of a large reservoir of iron coupled with plentiful substrate
                          availability supports the metabolic activity of iron-reducing, sulfate-reducing, and/or methanogenic
                          bacteria.  This enhanced microbial activity may beneficially influence zero-valent iron reductive
                          dehalogenation reactions through favorable impacts to the iron surface or through direct microbial
                          transformations of the target compounds.  However, this enhancement may come at the expense of
                          faster corrosion leading to faster precipitate buildup and potential biofouling of the permeable
                          treatment zone.

                          This research brief presents findings overthe past fouryears at two sites where detailed investigations
                          by the U.S. Environmental Protection Agency (U.S. EPA) have focused on the long-term performance
                          of PRBs under a Tri-Agency Permeable Reactive Barrier Initiative (TRI). This initiative involves the
                          U.S. EPA, the Department of Defense, and the Department of Energy. The objectives of the TRI are
                          to leverage the technical and financial resources of the three agencies in order to examine the field
                          performance of multiple PRBs across the United States. A survey of existing PRBs indicated that
                          the two main challenges facing the technology were (1) uncertainties associated with the longevity
                          (geochemistry) of a PRB and (2) ensuring/verifying hydraulic performance. Therefore, this initiative
                          and our research have focused primarily  on these challenges.

                          * U.S. EPA, Office of Research and Development, National Risk Management Research
                           Laboratory, Subsurface Protection and Remediation Division, Ada, OK.

-------
Introduction

Research  described  in this research  brief  explores the
geochemical  and microbiological processes occurring within
zero-valent iron treatment zones in permeable reactive barriers
that may contribute to decreases in iron reactivity and decreases
in reaction zone permeability that, in turn, may eventually lead to
system plugging and failure. Using advanced surface analytical
techniques together with detailed coring  and water sampling
programs  at  two geographically,  hydrogeologically,  and
geochemically distinct iron barrier  installation  sites,  specific
objectives of this research project were to:

1)   Characterize  the  type  and  nature of surface precipitates
    forming over  time at the  upgradient aquifer/iron interface,
    within the iron zone, and at the downgradient/iron interface.

2)   Develop conceptual models that predict the type and rate of
    precipitate formation based on iron characteristics and water
    chemistry.

3)   Identify type and extent of microbiological activity upgradient,
    within and downgradient in at least one of the chosen sites
    to evaluate microbiological response or effects from emplaced
    iron into an aquifer system.

4)   Develop practical  and cost-effective protocols for long-term
    performance  assessments  at permeable reactive barrier
    installations.

In planning groundwater remediation systems using permeable
treatment walls, granulariron metal is the treatment material most
often proposed and used. The choice of zero-valent iron stems
from the reducing conditions created by reactions  between
groundwater and iron metal. The "ideal" iron wall would support
fast and complete contaminant transformation, and  have a
hydraulic  conductivity that could be maintained  over  time.
However, physical and chemical  properties of the iron treatment
zone will change with time as a result of the "aging" of the iron. The
lifetime of an iron wall will essentially be determined by how fast
and by what types of minerals  precipitate within the reaction
zone.  The corrosion of zero-valent iron in aqueous environments
has been widely studied (e.g., Reardon, 1995).  In water, zero-
valent iron is oxidized by many substances to ferrous ion, leading
to dissolution and volume loss of the metal. Under anaerobic
conditions,  reduction of water occurs:
 Fe°
2H2O = Fe2+
                              H  + 2Ohr
                                                       (1)
Under aerobic conditions, dissolved oxygen acts as the oxidant
and can lead to the production of ferrous and ferric iron:
2Fe° + O
     2H2O
                          = 2Fe2+ + 4Ohr
         4Fe2+ + O,  + 2H+ = 4Fe3+ + 2Ohr
                                                       (2)

                                                       (3)

Both aerobic and anaerobic iron corrosion reactions lead to an
increase in pH. Aerobic corrosion is a more rapid process, as
evidenced by the rapid loss  of dissolved oxygen in iron/water
systems. As the corrosion process proceeds, iron hydroxides
form, which increases the thickness of an iron oxide passivation
layer already present at the iron metal surface. Under anaerobic
conditions,  hydrogen gas that is formed as a product of iron
corrosion may also temporarily passivate the iron surface. Many
species abundant in groundwater can affect the  iron corrosion
process. For example, chloride, carbonate, and  sulfate can all
accelerate the corrosion of iron by increasing the dissolution rate
of the protective oxide layer.
                                              In groundwater, any of the common anions present may influence
                                              the effectiveness  of zero-valent  iron barriers  for contaminant
                                              remediation. Carbonate is of particular interest. At the high pH
                                              condition caused by the corrosion of iron, bicarbonate reacts with
                                              OH'and Ca2+ ions to form insoluble CaCO3 that will precipitate as
                                              calcite or the polymorph, aragonite.  In addition, the bicarbonate
                                              anion can precipitate  with ferrous ion to form ferrous carbonate
                                              (siderite):
                                                     Fe2+ + HCO,-+ OH- =  FeCO.(s) + H.O
                                                                                                           (4)
Under aerobic conditions, ferric oxides and  oxyhydroxides
precipitate. Under anaerobic conditions and at high pH, ferrous
hydroxide or green-rust minerals are expected to form.  Other
species in groundwater can also precipitate at the high pH,
reducing conditions created in  the iron-water system, such as
iron sulfides.  The formation  of these various precipitates has
been observed in numerous studies investigating the use of zero-
valent iron for remediating  contaminated groundwater (e.g.,
Powell et al., 1995; Mackenzie et al.,  1999; Phillips et al., 2000).
These mineral precipitates may limit access to the iron surface
and thereby decrease reactivity and  will decrease porosity and
therefore  impact flow through  the reactive media, resulting  in
decreased residence time  and  incomplete  treatment  of
contaminants.

Many studies have investigated the interactions of bacteria with
metallic iron under both aerobic and  anaerobic conditions.  The
oxidation or corrosion of zero-valent iron may be stimulated  or
inhibited by microorganisms (Morales et al., 1993; Hernandez et
al., 1994). However from an ecological perspective, metallic iron
in  subsurface environments represents a  significant energy
reservoir—an energy  supply that microorganisms will utilize if
possible.  Due to the limited solubility of oxygen in groundwater
and the rapid  reduction of molecular oxygen by Fe(0) and Fe(ll),
permeable  reactive barriers usually exist  as anaerobic
environments.  Under anaerobic conditions  molecular oxygen-
driven chemical corrosion rates may be reduced, but biologically
mediated  anaerobic  corrosion  may  occur at rates  exceeding
those seen under oxygenated conditions (Lee etal., 1995). In the
absence of oxygen, protons may serve as electron acceptors and
allow  for the formation of oxidized iron species such as Fe(ll).
Under static and aseptic conditions this process is usually limited
by the formation of a stable hydrogen film produced from the
reduction of protons on the metal surface.  However, processes
that consume hydrogen  gas  and  thereby interfere with the
formation of this hydrogen film,  such as mixing with water that is
undersatu rated with respectto hydrogen orthe activity of hydrogen-
consuming bacteria, will enhance anaerobic iron corrosion. Both
of these processes are likely to  be important in PRBs installed in
subsurface environments.

The enhancement of the  anaerobic  corrosion process and the
formation of dissolved Fe(l I) is not necessarily detrimental to PRB
performance.   If the target contaminant is reduced by Fe(ll) as
well as Fe(0), such as Cr(VI), a dispersion of aqueous Fe(ll) could
represent an increase in the size of the reaction/treatment zone
as compared to the surface contact area of Fe(0) alone.  However,
the utilization  of  dissolved hydrogen  may  result  in  bacterial
growth and biofilm formation.  The development of this biofilm in
a PRB may  be detrimental to performance through several
mechanisms.  A microbial biofilm containing iron-reducing and
sulfate-reducing bacteria as terminal  oxidizers  is a  complex
structure containing both organic and inorganic materials (Cord-
Ruwish, 2000).  These biofilms are typically only about 5% dry
weight,  suggesting a disproportionate effect on hydraulics per

-------
unit of bacterial growth. Changes in PRB hydraulic conductivity,
the masking of active sites,  the removal  of active  chemical
species, and the competition for reducing equivalents would all
seem to be  processes that could negatively affect PRB
performance.   Conversely some microbial processes  could
enhance PRB performance. In some instances bacteria may be
more effective at contaminant transformation or may degrade
compounds unaffected by PRBs.  It is therefore evident that a
clear understanding is needed of microbial/PRB interactions for
the design and efficient operation of PRBs and because inhibition
of biofilm formation  under the conditions  associated with
subsurface  PRBs appears to be technically and economically
impractical.

Site Descriptions

Two field sites were evaluated in the U.S. EPA portion of the TRI:
the U.S.  Coast Guard Support Center (USCG-SC)  site near
Elizabeth City, North Carolina, and the Denver Federal Center
(DFC) in Lakewood, Colorado.

U.S. Coast Guard Support Center

The USCG-SC is located about 100 km south of Norfolk, Virginia
and 60 km inland from the Outer Banks region of North Carolina.
The base is situated on the southern bank of the Pasquotank
River, about 5 km southeast of Elizabeth City, North Carolina. A
hard-chrome plating shop was in operation for more than 30 years
                                           in Hangar 79, which is only 60 m south of the river (Figure 1).
                                           Following its closure in 1984, soils beneath the shop were found
                                           to contain chromium  concentrations up  to  14,500 mg/kg.
                                           Subsequent site investigations by U.S. EPA personnel identified
                                           a chromate plume extending from beneath the shop to the river.
                                           The  plume has high (>10 mg/L)  concentrations of chromate,
                                           elevated sulfate(to 150 mg/L), and minor amounts of chlorinated
                                           solvent compounds (trichloroethylene, cis-dichloroethylene, vinyl
                                           chloride).   The plating shop soils and related groundwater
                                           contamination are referred to as  solid waste management  unit
                                           (SWMU) number 9 by the state of North Carolina and the USCG.
                                           Sampling results from a monitoring network consisting of more
                                           than 40  monitoring  wells and about 100  Hydropunch™  and
                                           Geoprobe™ monitoring points indicate that the Cr(VI) plume is
                                           about 35 m wide,  extends to  6.5 m below ground surface  and
                                           extends laterally about 60  m from the hangar to the Pasquotank
                                           River. Multilevel samplers  installed near the barrier wall location
                                           indicate that the bulk of the contamination resides from 4.5 to
                                           6.5 m below ground  surface.

                                           The site geology has been described in detail elsewhere (Puls et
                                           al., 1999), but essentially consists of typical Atlantic coastal plain
                                           sediments, characterized by complex and variable sequences of
                                           surficial sands, silts and clays. In  general the upper 2 m of the
                                           aquifer are sandy to  silty clays that pinch out towards the north,
                                           or near the Pasquotank River, where sandy fill predominates.
                                           Fine sands, with varying amounts of silt and clay, and silty-clay
                                           lenses form the rest  of the shallow aquifer.
                               ®MW46
                                                        MW35D
                                                           m
                                                                            N
                                   ML35
                                   MW47
                                              WC25
BUILDING 78
                                   ML34
                                   ML33
                                  X
                       4k  WC24X,^/IW50^MW49
                                                 ML14
                                   ML31

                    MW38
                                                                                PRB
MW Compliance well
©  (5-10ft. screen)

ML  Multilevel bundle
•   (6 in. screens)

WC Well cluster
•   (6 in. screens)

                                                        *


                                                             '
                                                                                            10m
                                                                                  Approx. Scale
                  HANGAR 79        I   Cp'atin9
                 	I   ^hpp   |
                                                                                    Groundwater
                                                                                    flow direction
Figure 1.    Plan view map showing compliance well, bundle and well cluster locations relative to granular iron barrier and Cr plume
            (June 1994 data).

-------
Groundwater flow velocity is extremely variable with depth, with
a highly conductive layer at roughly 4.5 to 6.5 m below ground
surface.   This layer coincides  with  the  highest  aqueous
concentrations of chromate. The groundwater table ranges from
1.5 to 2.0 m below ground  surface and the average horizontal
hydraulic gradient varies from 0.0011  to 0.0033. Slug tests
conducted on  monitoring wells with 1.5 m screened  intervals
between 3  and 6 m  below ground surface indicate  hydraulic
conductivity  values of between  0.3  to 8.6  m/d.  A multiple
borehole tracer test in wells screened between 3.9 to 5.9 m below
ground surface was conducted by Pulsetal. (1999).  Groundwater
velocities of 0.13 and 0.18 m/day were measured in this test.
Assuming an average hydraulic gradient of 0.0023 and porosity
of 0.38, these velocities correspond  to an  average  hydraulic
conductivity of about 26 m/day. The groundwater is contaminated
with  chromate and  volatile organic  compounds  including
trichloroethylene (TCE), cis-dichloroethylene (cis-DCE), and vinyl
chloride (VC).

In June of  1996, a 46 m long, 7.3 m  deep, and 0.6 m wide
permeable  reactive barrier (continuous wall configuration) of
zero-valent iron was  installed approximately 30 m  from the
Pasquotank River  (Figure  1;  Blowes  et al., 1999a,b).  The
reactive wall was designed  to remediate hexavalent chromium-
contaminated groundwater, in addition to treating  portions of a
larger overlapping plume of TCE.  A detailed monitoring network
of over 130 subsurface sampling points was installed in November
of 1996 to provide detailed  information on spatial and temporal
changes in  porewater geochemistry.

Denver Federal Center

The Denver Federal Center (DFC) is located about 10 km west of
downtown Denver,  Colorado. Aquifer materials at the site are
made up of alluvial sediments that overlie the Denver Formation.
The Denver Formation is Paleocene to Late Cretaceous in age
and consists of brown, yellowish-brown, gray, and  blue-gray
intercalated sandstone,  claystone,  siltstone, shale  and
conglomerate containing olive-brown andesitic sandstone beds.
It lies about 2 to 14m  below ground surface at the  DFC and can
attain a thickness of up to 260 m.  The Denver Formation has
been divided into two zones, the upper weathered zone and a
lower unweathered zone.  These two zones are lithologically
similar  but differ  in  color and  permeability, with the upper
weathered zone having greater fracture permeability. The upper
weathered zone is up  to 7 m thick and exhibits a grayish brown
color with yellowish orange staining while the lower unweathered
zone has a diagnostic blue color, commonly called "Denver Blue."

There are two separate  deposits of  alluvial  sediments in the
vicinity of the DFC. The Verdos Alluvium of Pleistocene age is a
poorly sorted, stratified gravel containing lenses of sand, silt, and
clay. In some drainages, the Denver Formation may be overlain
by the Piney Creek Alluvium, well-stratified, interbedded organic
sands, silts and clays  with interbedded gravels.

Groundwater at the site generally moves from west to east and
the average hydraulic velocity is about 30 cm per day in  the upper
alluvium and weathered Denver formation  (McMahon et al.,
1999).   It is contaminated with  volatile organic  compounds,
primarily  TCE,  cis-DCE,   trichloroethane  (TCA)  and
1,1-dichloroethene(DCE).  At the eastern boundary of the site,
maximum concentrations of TCE, cis-DCE, TCA, and DCEwere
about 700 ppb, 360 ppb, 200 ppb and 230  ppb, respectively
(FHWA, personal communication, 2002). At least one source of
these VOCs was a leaking underground storage  tank located
near Building  52 that was  used by  the  Federal  Highway
Administration (FHWA) to store  waste,  primarily TCA.
Groundwater flow from the aquifer discharges into Mclntyre
Gulch (Figure 2). Mclntyre Gulch is a deep channel that penetrates
the aquifer along the southern edge of the contaminant plume.
Downing Reservoir is too shallow to be influenced by the aquifer,
but the reservoir stage does affect the ground-water level.

In the fall of 1996, FHWA and GSA installed a permeable reactive
barrier atthe eastern edge of the DFC property along north-south
trending Kipling Street (Figure 2).  In contrast to the continuous
wall design used at Elizabeth City, the DFC PRB has a funnel-
and-gate design configuration.  The funnel  component of the
PRB employs metal sheet pile that was driven into unweathered
bedrock of the Denver Formation or into resistant,  weathered
layers of the Denver Formation. The depth of penetration of the
funnel ranged from about 7.0 to 10 m.  The PRB has 4 reactive
gates, each 12.2 m long, up to 9.5 m deep,  and from 1.8 m to
(Gate 1) to 0.6 m (Gates 3  and 4) wide.  The design thickness
varied because of anticipated differences of contaminant fluxes
to the PRB at different locations.

Table 1 provides a general comparison of the PRBs at Elizabeth
City and the Denver Federal Center.

Methods

Groundwater Sampling

Groundwaterwas sampled from monitoring wells using peristaltic
or submersible centrifugal pumps.  Atthe USCG-SC, 10-2" PVC
compliance wells have been monitored on a quarterly basis since
November 1996 and  up to 125  multi-level wells have been
sampled on an annual basis since November 1996. Atthe DFC
approximately 18 wells were sampled on an annual basis over
the course of this study (1999-2000). Pumping rates were always
between 150 and 250  mL/min  to  minimize chemical and
hydrological  disturbances in and around the  well  (Puls and
Powell, 1992). Prior to sample collection, waters were pumped
through a flow-through cell  containing calibrated electrodes for
pH, oxidation-reduction potential (ORP), specific conductance,
and dissolved oxygen. Stabilization of electrode readings was
assumed after 3 successive readings within ±0.10 units for pH,
±10 mV for ORP, ±3% for specific conductance, and ±10% for
dissolved oxygen.  After stabilization  of electrode  read-outs,
turbidity was generally less than 5 NTUs. Filtered samples for
cation analyses were collected using 0.45 urn cartridge filters
(Gelman aquaprep). Analyses were made using a Perkin Elmer
Optima 3300  DV inductively coupled  plasma spectrometer
(ICP-OES). Anion samples were unfiltered and unacidified and
analyses were carried  out using capillary electrophoresis (Waters
Quanta 4000E).

Colorimetric techniques were used in the field for Fe(l I), dissolved
oxygen, and hydrogen sulfide.  Ferrous iron and sulfide were
determined using the 1,10 phenanthroline and methylene blue
indicators, respectively, and  a HACH DR2010 spectrometer.
Dissolved oxygen was measured using CHEMets test kits that
employ the rhodazine D  (low range) and indigo carmine (high
range) colorimetric indicators. Alkalinitytitrations were conducted
in the  field by titration with standardized sulfuric acid  to the
bromcresol green-methyl red endpoint.  Dissolved H2 analyses
were performed atthe Denver Federal Center using the bubble
strip  method (Chapelle et al., 1997).

Core Collection and Analysis

To assess the extent of corrosion and mineral buildup on the iron
surfaces, 5 cm i.d. cores were collected using a Geoprobe™.
Core  barrels  were driven using a pneumatic hammer  to the
desired sampling location and continuous, upto  110cm, sections

-------
                                                        1690
0   m   75
Figure 2.    Plan view of Denver Federal Center PRB and groundwater elevations (after McMahon et al., 1999).
Table 1.     Comparison of PRBs Investigated in this Study
             Contaminants        PRB        Date        Iron       Iron    Ground   Groundwater,   Groundwater,
                             Configuration   Installed  Dimensions   Volume   water,         pH            DO
                                                                             SC                      (mg/L)
                                                                           (US/cm)
U.S. Coast
Guard Cr(VI)
Support
Center TCE, cis-DCE
Denver TCE, TCA, cis-
Federal DCE
Center




Continuous 6/96 46 m length
wall 7.3 m deep
0.6 m wide
Funnel-and- 10/96 Gate 1
gate 12.2 length
8.5 m deep
1.8 m wide
Gate 2
12.2 length
9.5 m deep
1.2 m wide
GateS
12.2 length
7.3 deep
0.6 m wide
201 m3
Gatel
187m3
Gate 2
139m3
GateS
53m3
325±126
(n=15)
Gatel
1236±65
(n=3)
Gate 2
1358±10
(n=3)
GateS
1306±10
(n=2)
5.94±0.44
(n=15)
Gatel
7.14±0.15
Gate 2
7.19±0.08
GateS
7.06±0.07
0.5±0.4
(n=11)
Gatel
0.5±0.2
Gate 2
0.2±0.1
GateS
<0.05
Notes:  Geochemical parameters from Elizabeth City are average values (± 1 s.d.) from upgradient monitoring well MW48.  All
        parameters monitored quarterly since 2/97 to 3/01. Geochemical parameters from DFC gates 1, 2, and 3 are average values
        from wells GSA21, GSA26, and GSA31. Gate  1 and 2 parameters were measured on a yearly basis from 5/99 to 7/01, and
        gate 3 from 7/00 to 7/01. Information for DFC gate 4 is not included here; only gates 1-3 were studied in this investigation.

-------
of iron, iron + soil, or iron + pea-gravel were retrieved.  Angle
cores (30° relative to vertical) and vertical cores were collected
in order to  assess the  spatial  distribution of mineral/biomass
buildup in the reactive media. Priorto pushing the core barrel, an
electrical conductivity profile was collected to verify the exact
position of the iron/aquifer interface (e.g., Beck et al., 2001). In
all cases core recovery was 60 to 85% of the expected  value.
Core  materials  from Elizabeth City and the Denver Federal
Center (1999-2000) were broadly similar in appearance. In all
cases the iron particles were jet black in color without any obvious
signs of cementation or oxidation. Iron grains from the upgradient
interface of DFC  gate 2 were  noticeably enriched in a  black-
colored, gel-like material.  Immediately after collection the cores
were frozen and shipped back to the Subsurface Protection  and
Remediation Division in Ada, Oklahoma, for sub-sampling  and
analysis. The frozen cores were partially thawed and then placed
in an anaerobic chamber with a maintained H2-N2 atmosphere.
Each core was logged and partitioned into 5 to 10 cm segments.
Each segment was homogenized by stirring in the glove box  and
then split into 4 sub-samples: (1) inorganic carbon analyses, (2)
sulfur analyses/X-ray diffraction (XRD), (3) Scanning electron
microscopy (SEM)/X-ray photoelectron  spectroscopy  (XPS)
analyses, and, (4) microbial assays (phospholipid fatty  acids,
PLFA). All sub-samples were retained in airtight vials to prevent
any air oxidation of redox-sensitive constituents.

Inorganic carbon analyses were conducted using a UIC carbon
coulometer system. Weighed samples were placed in a glass
reaction vessel and purged with  CO2- scrubbed air. The samples
were then acidified with hot, 5% perchloric acid and evolved CO2
gas was carried to the coulometer cell containing a CO2-sensitive
ethanolamine solution and quantitatively titrated.  Hot, dilute
perchloric acid dissolves carbonate precipitates such as siderite,
calcite,  aragonite, and  carbonate forms of green-rust.  The
perchloric acid extraction does not liberate carbide-carbon, which
is present at concentrations near 3 wt% in  Peerless iron.  Un-
reacted  Peerless  iron contains 15±5  ppm  of acid extractable
carbon.  Total sulfur measurements were made with a UIC sulfur
coulometer system. Iron samples were covered with V2O5  and
combusted in the presence of oxygen at 1050 °C.  Evolved gases
are passed through a column  of reduced Cu to quantitatively
convert all sulfur to SO2, which is then carried to  the coulometer
  cell where it is absorbed and coulometrically titrated. Un-reacted
  Peerless  iron contains 5±1 ppm of sulfur using this combustion
  method.  In addition, acid-volatile sulfide (AVS) and chromium-
  reducible sulfide (CRS) extractions were performed using hot,
  6 M HCI and 1 M CrCI2 in 0.5 M HCI, respectively (Zhabina and
  Volkov, 1978).  These acid extractions determine the quantities
  of metal  monosulfide precipitates  (AVS)  and iron  disulfide
  precipitates (i.e., pyrite; CRS).

  XPS measurements were made using a PE Model 5500 X-ray
  photoelectron spectrometer, operated with an Al Ka X-ray source
  at a power of 400 W(T. Sivavec, General Electric, Research and
  Development).  Atomic concentrations were obtained from peak
  areas using established sensitivity factors.  Depth profiles were
  measured by resting a 4 kV argon beam over an area of 2x2 mm.
  X-ray diffraction patterns were collected  of the fine-grained
  materials removed by sonication from the iron grains. Samples
  were mounted on  a quartz plate and diffraction patterns were
  obtained  using a Rigaku Miniflex diffractometer using Cu  Ka
  radiation (0.01 degree step, 0.5 degrees two-theta per minute).
  Scanning electron and  optical microscopy were  utilized  to
  determine the thickness of  surface precipitates and  evaluate
  physical morphology of grains  and extent of surface coverage.
  Prior to microscopic characterization, samples were set in epoxy
  resin, cured, and ground and polished using standard techniques.
  Samples were analyzed forcontent and distribution of phospholipid
  fatty acids (PLFA;  Microbial Insights, Inc.).  PLFA analyses are
  based on the extraction and separation of lipid classes, followed
  by quantitative analysis using  gas chromatography/mass
  spectrometry (GC/MS).

  Results and Discussion

  Groundwater Chemistry

  Major anion and cation compositions of groundwater collected
  upgradient from the iron treatment zones at Elizabeth City and the
  Denver Federal Center are shown in Figures. Groundwaterfrom
  Elizabeth City contains lower concentrations of total dissolved
  solids (<400 mg/L) compared to  groundwater from the DFC
  (1000-1200 mg/L). At both sites, the concentration of dissolved
  oxygen in upgradient groundwater is about 1  mg/L and pH is
  near-neutral. There are, however, spatial variations in Eh values
                           HCO
     Elizabeth City ML10
     Denver FC
  SOT*
Mg2
  IDS, mg/L
  O 900-1200
  o 600-900
  ° 300-600
  • 0-300
Figure 3.    Upgradient groundwater compositions (molar ratios) and TDS values forthe Elizabeth City and Denver Federal Center
            PRB sites.

-------
of groundwater entering the iron zones at both sites (Table 2).
Reactions between zero-valent iron and groundwater impact, to
some extent, the  concentration  of all  major solute species
(Figure 4). Long-term trends indicate consistent removal of TCE
and Cr(VI) at Elizabeth City as well as,  among other species,
sulfate, calcium, magnesium, and silica (Figure 4). Similartrends
are observed at the DFC and are consistent with the patterns
observed at other zero-valent iron PRB sites (Phillips etal., 2000;
Liang et  al., 2000; Battelle, 2000; Morrison et al.,  2001).  In
downgradient wells, an analysis of data collected over five years
at Elizabeth City shows little variation for most species; however,
there are subtle time-dependent increases in pH and decreases
in Eh (Figure 5).

A comparison of groundwater chemistry between upgradientand
downgradient wells indicates that the iron zones at Elizabeth City
and the DFC are long-term sinks for C, S,  Ca, Si, Mg,  N, and Mn.
Concentrations of dissolved iron are elevated in downgradient
compliance wells at Elizabeth City (Figure 5).  Dissolved iron is
not released from  the zero-valent iron but probably from the
downgradientaquiferdueto decreased redox potentials in regions
immediately downgradient from the iron media. In the region
immediately upgradient from the iron wall at Elizabeth City,
increases in the concentration of ferrous iron  have led to the
development of a reducing zone where a fraction of the hexavalent
chromium is removed from the  groundwater plume  (Figure 4).
Removal of chromium continues as the plume passes through
the iron media.

Reaction Path Modeling

Thermodynamic modeling can in principle be used as a predictive
tool for estimating the types and quantities of mineral precipitates
that form as groundwater reacts and approaches thermodynamic
equilibrium with Fe° (Gavaskar et al., 1998; Blowes and Mayer,
1999; Morrison et  al., 2001).  The Geochemist's Workbench
         program was used to simulate the incremental dissolution of Fe°
         into representative groundwater compositions  from  Elizabeth
         City and the DFC.  The primary thermodynamic database was
         modified to include solubility data and reaction stoichiometry for
         zero-valent  iron, iron monosulfides, and green-rusts (Table 3).
         Figure 6 shows the predicted changes in masses of minerals that
         precipitate and dissolve as the systems approach equilibrium
         with Fe°.  In general, equilibrium modeling predicts that Fe° will
         eventually react to form magnetite, mackinawite, magnesium
         silicate,  and/or magnesium hydroxide.  Carbonates (calcite,
         dolomite), pyrite,  quartz, and green-rust  are  predicted to be
         intermediate products.  Although dolomite formation is predicted
         based upon equilibrium modeling, the precipitation of dolomite is
         unlikely in a PRB.  Dolomite has a chemical formula CaMg(CO3)2
         and a crystallographic structure similarto that of calcite. It is well
         known that reactions to precipitate  or dissolve dolomite are
         extremely slow at  low temperatures (Hsu, 1967) so that dolomite
         precipitation  is an  unlikely cause  of the comparatively rapid
         accumulation of inorganic carbon and magnesium in PRBs.
         Interestingly, the  reaction path model predicts the formation of
         carbonate-green-rust prior to  magnetite,  consistent  with iron
         corrosion studies (McGill etal., 1976; Bonin etal., 2000).  Green-
         rusts are unstable, mixed valence Fe(ll)-Fe(lll)hydroxy-salts that
         are highly susceptible to oxidation in the presence of oxygen.
         They are known to be transient products of iron corrosion that
         precipitate as metallic iron oxidizes to form Fe(lll) oxyhydroxides
         (McGill et al., 1976; Schwertmann and Fechter,  1994).  The
         structure of green-rusts consists of sheets of Fe(ll)(OH)6 in which
         some of the  Fe(ll) is substituted for by Fe(lll).  This mixture of
         ferrous and ferric iron results in a net positive layer charge, which
         is balanced by interlayer incorporation of anions such as Cl",
         CO32",  and SO42".  Sulfate  and carbonate forms of green-rust
         appear to be the most important in natural systems (e.g., Genin
         etal., 2001).
Table 2.     Groundwater Chemistry at Elizabeth City (June, 2000; Median Values Transect 2) and the Denver Federal Center
            (July, 2000; Median Values)
                       Elizabeth City
                         aquifer        Fe°

         pH               5.84          9.51
         Eh (mV)          491           -347
         02(mg/l)          0.3           0.1
         H2 (nM)
         Na (mg/l)         50           29
         K                3.0           3.1
         Mg               9.1           3.2
         Ca               16.4          5.3
         Fe              <0.04         0.05
         Sulfate           49           <1.0
         Chloride          51           25
         Nitrate           0.9           <0.1
         Silica             10.9          <0.2
         DIG               15            8
         DOC             1.2           0.9
         IDS              290          143
DFC - gate  1
  aquifer         Fe°

    7.24         10.30
    233          -247
    0.92          0.1
    0.63          948
    179          184
    0.39          0.60
    18.1          3.9
    100          1.2
    0.19         <0.04
    234          <1.0
     52           59.7
    3.2          <0.10
    13.2          0.3
     86           60
    2.3           1.8
    1090          510
DFC - gate 2
  aquifer         Fe°

    7.80          9.90
    281           -35
    0.3           0.1
    2.38          304
    167          248
    0.27          0.95
    31            61
    114          4.1
   <0.04         <0.04
    286          359
    66           83
    2.9           1.1
    14.8          0.3
    95           46
    1.6           3.3
    1172          1011

-------
                                         Eh (mV, SHE)
                   sulfate (ppm)
                                                                                alkalinity (ppm)
                    0.0  1.0  2.0  3.0

                       calcium (ppm)
                                                          silica (ppm)
                                                       0.0  1.0  2.0  3.0
0.0  1.0  2.0  3.0     0.0  1.0  2.0  3.0      0.0   1.0  2.0  3.0
                      June, 2000
 magnesium (ppm)        iron (ppm)         chromium (ppm)      cis-DCE (ppb)
                    0.0  1.0  2.0  3.0
                                        0.0   1.0   2.0   3.0
                                                          0.0   1.0  2.0   3.0      0.0  1.0  2.0  3.0
                                                                                                 0.0   1.0   2.0   3.0
                                                          Distance, m
Figure 4.    Cross-sectional profiles at the Elizabeth City PRB for selected species (June, 2000; Transect 2).
                                              25
      9.0
      8.5 -
      8.0 -
      7.5 -
      7.0-
      6.5-
      6.0 -
      5.5 -
      5.0
                                              20-
 Q.
 Q.
 CO
0
                                          0.0
       Oct-95 Mar-97  Jul-98  Dec-99 Apr-01     Oct-95  Mar-97  Jul-98 Dec-99 Apr-01
                                            Oct-95 Mar-97  Jul-98  Dec-99 Apr-01
-o— MW13 ,
Upgradient wells
-•— MW48
-A-MW49
Downgradient wells
-A— MW47
Figure 5.    Time-dependent pH values and concentrations of calcium and iron in compliance wells from Elizabeth City (MW13, 30 m
             upgradient of PRB, screened interval 4.3-7.3 m;  MW48, 0.6 m upgradient, screened interval 4.3-7.3 m; MW49, 1.8 m
             downgradient, screened interval 4.3-7.3 m; MW47, 1.8 m downgradient, screened interval 4.3-7.3 m).

-------
Table 3.
Thermodynamic Constants Added to the Geochemist's Workbench Database (Bethke, 1998)
              Phase


              Fe°

              Mackinawite
              (ordered)

              Mackinawite
              (precipitated)

              Green Rust SO4

              Green Rust Cl

              Green Rust CO3
                          Gibbs free energy of formation
                                (AG°fi kJ/mol)

                                  0

                                  -88.4


                                  -83.7


                                  -3785.0

                                  -2145.0

                                  -3588.0
        Reference
Benning et al. (2000)


Benning et al. (2000)


Bourrie et al. (1999)

Bourrie et al. (1999)

Bourrie et al. (1999)
              Reactions

              FeS + H+ = Fe2+ + HS"

              Fe6(OH)12SO4 + 12H+ = 4Fe2+ + 2Fe3+ +  SO42" + 12H2O

              Fe6(OH)12CO3 + 12H+ = 4Fe2+ + 2Fe3+ + CO32" + 12H2O

              Fe4(OH)8CI + 8H+  = 3Fe2+ + Fe3+ + Cl" + 8H2O
Unfortunately, progress along the reaction path is arbitrary; the
x-axes in Figure 6 do not represent time or space. Reaction path
models may be used to qualitatively evaluate materials that might
form in a PRBfroman initial groundwater composition. However,
with respect to predicting the space- or time-dependent
accumulation of mineral mass or volume, reaction path modeling
is less useful because it does not account for reaction rates or
non-equilibrium processes (i.e.,  microbiological activity).
Numerical models that consider kinetic limitations are beginning
to provide reasonably accurate predictions of mineral accumulation
in PRBs(e.g., Mayer etal., 2001), although there is still a need to
collect field  data to verify and  refine these multicomponent
reactive transport simulations.

Mineral Precipitates

Solid-phase characterization studies implementing XRD, SEM-
EDX,  reflected-light microscopy, transmitted-light  microscopy,
XPS,  and chemical extractions  indicate the accumulation of
lepidocrocite, magnetite, carbonate minerals (aragonite, siderite,
or green-rust), and iron monosulfide as primary precipitates in the
Elizabeth City and DFC PRBs.  Similar authigenic precipitates in
other  Fe° PRBs were reported in previous field and laboratory
studies  (e.g., Phillips et al., 2000).   The  formation of these
minerals is predictable based on the results of thermodynamic
models discussed above.  However, as previously noted, these
equilibrium models are not able to successfully predict the rate of
mineral buildup or the position of precipitates within the reactive
media, both  essential to understanding  the impact of mineral
precipitation  on long-term  remedial  performance.
                                                XPS scans show that iron particle surfaces from Elizabeth City
                                                and the DFC contain C,O, Fe, Si, S, Mg, Ca, Mn.andN. TheXPS
                                                data indicate a surface layer dominated by iron oxyhydroxides,
                                                an intermediate layer of iron oxide, and finally zero-valent iron at
                                                the greatest sputtering depths.  Surface carbon is present as
                                                carbonate with some  detected hydrocarbon  (binding  energy
                                                284.6 eV). The oxidation state of sulfur is predominantly present
                                                as sulfide (-2) but with minor amounts of sulfate (+6).  Surface
                                                enrichment in the elements Ca, Mg, S, and Si are consistent with
                                                observed decreases  in groundwater  concentrations of these
                                                elements.  Chromium was sometimes detected by XPS  in iron
                                                samples  from  Elizabeth  City, consistent with the  reduction/
                                                precipitation mechanism  for chromium uptake (Cantrell et  al.,
                                                1995; Powell et al., 1995; Pratt et al., 1997).   Table 4 lists the
                                                elements that preferentially partition to the iron media at Elizabeth
                                                City and the Denver Federal Center, the form of each element as
                                                it is transported to the iron media,  and the possible solid-phase
                                                form of each element in the iron media. Research is continuing
                                                to identify the structure and mineralogy of precipitates that form
                                                near the surfaces of iron particles in the PRBs at Elizabeth City
                                                and the Denver Federal Center.

                                                Microscopic observations  indicate that mineral accumulation is
                                                mainly occurring on the surfaces of the iron particles where steep
                                                gradients in pH and redox potential promote mineral precipitation
                                                (Figure 7). After 3.5 years of mineral accumulation, a consistent
                                                coverage of surface material ranging in thickness from 10 to 50
                                                urn is observed on iron  grains collected near the upgradient
                                                interface  at Elizabeth City (horizontal penetration <8 cm).  At
                                                greater  penetration depths  (>8  cm), surface coatings are

-------
                              I   I   I  I   I   I   I   I   I   I   I  I   I   I   I   I   I   I  I

                                 .1     .2     .3     .4    .5     .6     .7    .8    .9     1
                                             Reaction progress
                              I   I   I  I   I   I   I   I   I   I  I   I   I   I   I   I  I   I   I
Figure 6.
                     .1     .2    .3    .4     .5     .6    .7     .8     .9     1

                                    Reaction progress

Reaction path model results showing mineral precipitation trends for representative groundwater compositions from
(a) Elizabeth City and (b) Denver Federal Center (gate 2).
Table 4.    Elements Removed from Groundwater by Fe° PRBs, Input Forms, and Possible Solid Phase Associations
element
c
s
Ca
Mg
Si
input form
HC03-
SO42"
Ca2+
Mg2+
H4Si04°
solid phase association
CaCO3, FeCO3, Fe6(OH)12CO3
FeS
CaCO3, sorbed (?)
incorporation into CaCO3; Mg-OH-Si
sorbed onto Fe(OH)3 (?)




ppts (?)

                                                     10

-------
Figure 7.    SEM photomicrographs of iron particles collected near the upgradient iron/aquifer interface (Elizabeth City, 2000).
                                                            11

-------
discontinuous and <5 urn thick.  Where mineral accumulations
are the greatest, distinctive particle morphologies include radiating
needles, characteristic of carbonates formed at high degrees of
supersaturation;  homogenous crusts  without euhedral
morphologies,  characteristic of iron  oxyhydroxides;  and thin
coatingsoraggregatesofveryfine-grained particles, characteristic
of iron monosulfides.  Textural forms and coating thickness of
surface precipitates  are broadly similar at the  DFC  and  are
comparable to those reported from the Y-12 Fe° barrier (Phillips
etal.,2000).

Eighteen cores were  analyzed to determine concentrations and
spatial distribution of inorganic carbon (1C) and sulfur accumulation
in the Elizabeth  City and the DFC Fe° PRBs.   At both  sites,
accumulation of 1C and sulfur is greatest nearthe upgradient soil-
Fe° interface, with concentrations decreasing with increasing
penetration into the iron (Figure 8). Anomalous buildup of mineral
precipitates and microbial biomass is readily apparent  in gate 2
of the DFC.  Concentrations of 1C in gate 2  are as high as 8000
ppm and total sulfur values approach 4500 ppm, or a factor of
about 4x the amounts observed in DFC gate 1 or the Elizabeth
City PRB.

Total sulfur measurements compare well with the results of acid-
volatile sulfide (AVS) extractions (Figure 9).  Sulfur in the reactive
barrier system is mainly present as labile,  iron sulfide phases.
Iron  sulfides could  potentially enhance  the degradation  of
halogenated aliphatic compounds in PRBs (Butler and Hayes,
2000). The high pH, low Eh geochemical signature in iron zones
favors the chemical  reduction  of sulfate to sulfide.   Rates of
chemical sulfate  reduction at  low temperatures  are extremely
slow(Trudingeretal.,  1985). It is expected, therefore, thatthe low
Eh (high hydrogen fugacity) environment supports metabolism
by sulfate-reducing bacteria in the iron (see Microbiology section
below). Bisulfide respired by these bacteria rapidly reacts at high
pH with ferrous iron  produced  via the dissolution of the  zero-
valent iron to form acid-volatile sulfides.  Less than about 5% of
the accumulated sulfur in the Fe° is extractable with hotchromous
chloride which suggests either that some transformation of the
iron monosulfides to pyrite (FeS2) has occurred orthat labile AVS
phases have transformed to more crystalline, acid-resistant iron
monosulfides (most likely greigite). The solubility of precipitated
mackinawite decreases with increasing pH and can be described
by:
               FeS + H+ = Fe2+ + HS-
(5)
withlogK  =-3.6(Benningetal.,2000). AtphMOandlog aFe2+ = -5
(typical of Elizabeth City water), the expected equilibrium activity
of bisulfide  is 10~8-1, which  is much lower than the nominal
detection limit of the methylene blue method (~10'58 mol/L) for
dissolved sulfide. Therefore, the detection of hydrogen sulfide in
groundwater samples collected from Fe° media is not necessarily
an indicator of whether iron sulfides are  precipitating  in the
reactive media. ConcentrationsofhydrogensulfideuptoO.5 mg/L
were observed in groundwater samples collected from the DFC.
These high  dissolved sulfide values suggest that particulates
were entrained during well purging even at the low flow rates used
in this study.

Porosity loss in the iron zones due to the precipitation of inorganic
carbon and  sulfur  minerals was estimated by integrating the
concentrations of 1C and S as a function of distance in the iron and
estimating the volume loss by using the molar volumes of pure
calcite (CaCO3), siderite (FeCO3), and mackinawite (FeS). The
proportion of 1C present as calcite and siderite was calculated by
evaluating the excess drawdown of dissolved inorganic carbon
                         5     10     15     20     25     30     35
                               10     15    20    25    30     35
                          Distance, cm

Figures.    Concentration distribution  of  (a)  solid phase
            inorganic  carbon,  (b) sulfur, and (c) calculated
            porosity lost after  ~4 years  in the Fe° media at
            Elizabeth City  and  the DFC.  Concentrations of C
            and  S were  obtained from cores  EC060300-4
            (Elizabeth City);  C1-2-71000 (DFC, gate 1); and,
            C2-17-71300  (DFC,  gate 2).  Porosity lost is
            calculated with [(initial porosity-final porosity)/(initial
            porosity)].
       relative to dissolved calcium concentrations between upgradient
       and downgradient monitoring wells.  This method suggests that
       inorganic carbon is evenly distributed between FeCO3and CaCO3
       at Elizabeth City; the DFC gates are slightly enriched in FeCO3
       (60%) relative to CaCO3(40%). Porosity gain due to the dissolution
       of Fe° was estimated by assuming that all Fe2+ released was
       precipitated as siderite and mackinawite.  Because the cores
       were collected  at an angle, the measured concentration profile
                                                          12

-------
    o>
    0)
    tJ
    8 cm.  In gate 1 of the
                                         DFC, the precipitation front is spread out over a greater distance
                                         in the iron  and  may be the result  of higher flow rates in gate 1
                                         (0.38 m/d) compared to Elizabeth City (0.15 m/d).  In gate 2,
                                         14.2% of the available porosity in first 2.5cm of iron was lost over
                                         the first 3.75 years of operation.
Table 5.
Mass Accumulation of Inorganic Carbon and Sulfur and Estimated Annual Volume Loss
   Elizabeth
   City
 date

 6/98
 6/98
 6/99
 6/00
 6/00
mass 1C
  (kg)

  28.4

  68.3
 136.6
  53.8
                        mass S
                          (kg)

                          11.7
                          7.6
                          13.1
                          28.9
                          29.8
                                          flux 1C      fluxS      VFeCC-3      V CaCO3       V FeS       V loss
                                         (kgm'V1)  (kgm'V1)   (103xcm3y~1)   (103xcm3y~1)   (103xcm3y~1)   (% total y'1)
0.058

0.093
0.140
0.055
0.024
0.016
0.018
0.030
0.031
17.3

27.7
41.6
16.4
21.9

35.1
52.7
20.7
3.9
2.5
2.9
4.8
5.0
0.05

0.08
0.12
0.05
DFC
gate 1
DFC
gate 2

7/00

7/00

34.2

481.9

41.2

179.4

0.204

2.158

0.246

0.803

13.3

188.0

11.3

158.7

7.4

32.0

0.03

0.43
   Note: V loss is the net annual porosity lost assuming that all precipitates accumulate evenly throughout the PRB.
                                                           13

-------
Mass balance on C and S was estimated  by evaluating solid
phase concentration data and the changes in the groundwater
concentrations  of  inorganic carbon  and  sulfate.  Mass
accumulation of 1C and S in the Fe° media based on changes in
groundwater solute concentrations was  estimated from Q-ACV
where Q is the volumetric flux of water through the iron media
(L/y) and AC. is the change in the concentration of dissolved 1C
and S (mg/L) between upgradient and downgradient sampling
points. Average groundwater flow velocities at Elizabeth City of
0.13 to 0.18 m/d were reported by Blowes et al. (1999). McMahon
et al. (1999) report flow median velocities in gate 1 and gate 2 at
the DFC of 0.38 m/d and 0.11  m/d, respectively,  based on heat
pulse flowmetermeasurements. Results forcarbon mass balance
(Figure 10) agree to within a factor of 1.5x for the Elizabeth City
PRB and gate 2 of the DFC, but with poor agreement in gate 1.
Sulfur mass balance is in reasonable agreement for gates 1 and
2 of the DFC.   Considerably more accumulation  of sulfur is
expected in the Elizabeth City  PRB than  has been observed on
the core materials.  In  gate 2, more solid phase C and S is
observed on the iron grains than is predicted from the observed
changes in groundwater chemistry.  These trends may indicate
that rates of mineral accumulation have decreased with time,  i.e.,
the estimates in Table 5 are based on the assumption of steady-
state conditions throughout the lifetime of the PRBs.  Several
factors lead to uncertainty in the mass balance calculations for
PRBs.  Estimates  of  accumulation based on  changes in
groundwater chemistry depend on chemical and hydrogeologic
measurements. Determination of dissolved constituents may be
analyzed at a high level of accuracy and precision.  However,
estimates of the mass flux of ground water moving through PRBs
are prone to large uncertainties.  Estimates of accumulation
based on characterization of core materials depend on analytical
measurements and also on estimates of emplaced iron density.
Spatial heterogeneity in groundwater flow velocity, concentration
of solutes, concentration of solid phase products, and emplaced
iron  density all factor into the uncertainty analysis of mass
balance calculations.
 o
              Cell 1, DFC
                Cell 2, DFC
E. City
              Cell 1, DFC
                Cell 2, DFC
E. City
 Figure 10.
Mass balance estimate for inorganic carbon and
sulfur in the Fe° media at Elizabeth City and the
DFC.  Solid  bar   based   on  solid-phase
characterization; open bar based on groundwater
composition and flow rate.
Several transformation pathways of iron minerals in Fe° PRBs are
important to evaluations of long-term performance by affecting
the surface  properties and reactivity of the  iron particles and
reducing  available porosity.   Oxidized, reduced, and  mixed-
valence iron phases form and undergo transformation in  Fe°
systems.  Oxidized phases (lepidocrocite,  goethite) apparently
are the result of reactions with dissolved oxygen; elevated  DO
concentrations can lead to rapid corrosion and clogging of Fe°
due to ferric oxyhydroxide precipitate at the upgradient interface
(Liang et al., 2000).   Akaganeite has been identified based on
X-ray diffraction analyses at the PRB from Y-12 plant site (Phillips
et al., 2000).  Akaganeite is rare  in nature; its formation is
restricted to Fe- and Cl-rich hydrothermal brines and as the result
of iron corrosion in Cl-rich fluids. Akaganeite also forms as an
alteration product  of green-rust compounds (Schwertmann and
Taylor, 1989) and therefore may be related to sample preservation
and the oxidation of green-rust compounds rather than incipient
precipitation of this  phase as an  iron corrosion  product in
groundwater systems.

Mixed valance state  iron minerals include magnetite and green-
rust compounds. Green-rust compounds are corrosion products
that are expected to form under more reducing conditions than do
iron oxyhydroxides.  Their precipitation is favored under alkaline
conditions and transformation of these compounds to magnetite
is expected based  on the thermodynamic calculations presented
in a preceding section. Iron carbonate and monosulfides are the
primary reduced iron forms and are likely to  persist as long as
reducing, alkaline conditions persist.  Although iron monosulfides
are unstable relative to pyrite, the kinetics of this transformation
process decrease  with increasing pH so that iron monosulfides
may be the  long-term solid phase sulfide  in Fe° PRBs. Similarly,
substantial iron carbonate dissolution is not expected so long as
alkaline conditions persist. Research at EPA is continuing on the
factors that govern  iron mineral transformations in PRBs and
impacts to system longevity and performance.

Microbiology

Seventy-one core  samples collected from the Elizabeth City and
DFC PRBs were analyzed  for content  and  distribution of
phospholipid fatty  acids (PLFA). These organic compounds  can
be used as lipid biomarkers to provide a quantitative means to
evaluate viable microbial biomass, community composition, and
nutritional status. The cores contained sediments from upgradient
and  downgradient locations adjacent to the  PRBs, iron  filings
from the PRB  matrix, or a mixture of sediment and iron matrix
when the cores were collected at sediment/PRB  interface. The
highest accumulations of microbial biomass  were found  at the
DFC in iron samples from gate 2.   Concentrations as high as
4,100 pmoles/g dry wt were measured in iron matrix samples from
approximately  17  ft  below ground surface in gate 2 which is
equivalent to 8.36  x 107 cells per gram of iron matrix. Sediment
samples from 0 to  5 cm upgradient of the gate 2 PRB had similar
biomass levels and composition. Microbial biomass correlates
well with  total sulfur suggesting that the accumulated biomass
contains  anaerobic,  sulfate-reducing bacterial consortia.  The
presence of this consortium is also supported by the analysis of
PLFA structural groups that  suggest the presence of sulfate-
reducing  bacteria (mid-chain branched saturate fatty  acids;
Dowling et  al., 1986; Parkes et al., 1992),  of Gram negative
anaerobic bacteria (terminally branched saturate fatty  acids;
Parkes etal., 1992; Guckertetal., 1985), and of anaerobic metal-
reducing strains (branched monenoics fatty acids; Parkes et al.,
1992; Edlund et al., 1985).  These three PFLA structural groups
comprise approximately 30% of the total PFLA mass in gate 2
samples.
                                                         14

-------
It is perhaps of value to compare the biomass and community
structure information in gate 2 samples versus gate 1 samples
due to differences in performance observed at these two PRB
systems at the DFC (Table 6). The overall distribution of PLFA
structural groups varies between gates 1  and 2.  The higher
biomass, % contribution of anaerobic bacteria, and bacteria to
eukaryote ratio in gate 2 are all consistent with an anaerobic PRB
environment with a higher biological energy sink, suggesting that
hydrogen levels would be lower and that less reducing power
would be available for chemical reduction (White et al., 1980).
This would seem to agree with the lower dissolved hydrogen
levels seen in gate 2 (Table 2), and with the breakthrough of
contaminants dueto less available reducing powerand/or biomass
impacting system hydraulics. The higher pH and lower Eh values
associated with gate 1 would also create an environment less
conducive to bacterial growth, although it is interesting to speculate
as to the conditions that govern microbial growth and the conditions
that are impacted by microbial growth.

The  availability of 45 samples from the Elizabeth City PRB,
collected as vertical,  upgradient  and downgradient interfacial
cores,  allows for some analysis  of the spatial distribution of
bacterial biomass and biomass composition. The highest biomass
levels were observed in  upgradient core subsections, which
contained the sediment/iron interface (Figure 11). These samples
contained from 600 to 900 pmoles  of PLFA per gram of sediment
and iron.  This is equivalent to 1 to 2 x 107 cells per gram of solid
matrix or less than half the maximum detected in DFC samples
from gate 2. While total biomass  peaked at the sediment/iron
interface, the contribution  of  anaerobic biomarkers identified
above  (TerBrSats, BrMonos, MidBrSats)  as a percent of total
biomass peaked 1 to 3 inches into the iron matrix (Figure 12). The
localization of metal and  sulfate reducing  activity near  the
upgradient  sediment/PRB  interface would  seem  to  correlate
reasonably well with the iron, sulfate, and Eh contours shown in
Figure 4, and with the sulfur concentrations and porosity changes
illustrated in Figure 8.  Evaluation of the microbial populations
through the PLFA data from downgradient PRB/sediment interface
is more difficult.  As  shown in Figure  13, no  clear pattern of
biomass distribution  or  population composition is  immediately
apparent, although there seems to be some enhancement of an
anaerobic population immediately down stream of the interface.
Microbial counts range from  1.85 x 106 to 2.51  x 103, thus
downgradient cores seem to contain a  10-fold reduction in  the
                                                range of values (at both the maximum and minimum) compared
                                                to the upgradient cores.  The lower counts associated with the
                                                mid-barrier and  downgradient  samples  suggest that the
                                                environment at these locations is  more challenging to bacterial
                                                growth and  survival.  Examining the geochemical conditions
                                                associated with these locations supports this hypothesis. Figure 4
                                                indicates a decrease in biologically available electron acceptors
                                                such as sulfate and  cis-DCE in the  mid-wall and downgradient
                                                locations. The higher pH along with precipitated iron and sulfur
                                                species would also tend to create a more severe environment for
                                                bacterial  growth.  Table 2 also suggests that  mid-barrier and
                                                downgradient locations may be less conducive for growth due to
                                                a higher pH and a lack of electron acceptors (oxygen, sulfate,
                                                nitrate), which would seem to negate the positive influence of an
                                                abundant electron donor,  dissolved  hydrogen.

                                                The use of cathodic hydrogen as an energy source for biofilm
                                                development has  been a staple of the corrosion  literature for
                                                some time (Von Wolzogen Kuehr and van der Vlugt, 1934).  The
                                                high surface areas of PRBs and the supply of electrochemically
                                                active species  such as sulfur oxyanions insure the development
                                                of dissolved hydrogen  in the PRB and the local subsurface
                                                environment. Using  the USCG-SC PRB as an example, we can
                                                conduct some theoretical exercises to evaluate the  potential
                                                impact of biomass development on PRB performance based on
                                                direct geochemical  measurements.  Assuming that the  PRB
                                                represents a 300 m2  cross section, and has a flow of 0.15 m/day,
                                                50 % porosity,  and an average dissolved hydrogen level of 500
                                                nm, the net hydrogen production would be 6.75  mmoles/day
                                                (Cord-Ruwish,  2000). This could yield 0.675 crrWday (assuming
                                                0.5 mol ATP/mol H2, 200 g Biomass/mol ATP, Biomass density
                                                1 g/cm3) and the Fe(l I) concentrations in Figure 4 yield 24.2 cm3/day
                                                (assuming 1 mol Fe(ll)/molH2). Whilethese estimates of biomass
                                                formation seem insignificant, the  localization of this activity in
                                                discrete zones and  interaction with inorganic precipitates  may
                                                magnify the effect. However, the bio-available energy and  thus
                                                potential biofilms generating capacity represented by the PRB is
                                                enormous.  A1 m2 surface area that corrodes at 0.1  mm per year
                                                would yield, using a  density of iron of 7800 kg/m3, 14  moles of
                                                H2/day or approximately 40  mmoles/day.  This  is equivalent to
                                                1.5 urn of biomass per day  over the entire surface area of the
                                                PRB.  This assumes uniform growth and no loss to predation or
                                                sloughing of biomass.  Nevertheless, these numbers indicate a
                                                significant potential for biofouling in  PRBs.
Table 6.
PLFA Results for DFC Gates 1 and 2
site
DFC gate 1
DFC gate 2
ave. biomass
325
2283
cells/g
6.5 x106
4.6 x107
bact/euk ratio
156
>600
% anaerobic
-12%
-30%
Notes:  Average biomass in pmoles PLFA/g dry wt.  Cell/g is estimated bacterial cells per gram dry weight of sample based on 10s
        pmoles PLFA per gram dry weight of cells and 2.0 x 1012 cells per gram dry weight cells. Bac/euk ratio is the average bacteria
        to eukaryote ratio, ratio ofpolyenoic PLFA to total- polyenoic PLFA. %Anaerobic is the percent of total PLFA composed of
        terminally branched fatty acids, branched monenoics fatty acids and mid-chain branched fatty acids. Values are average of
        12 gate 1 samples  or 14 gate 2 samples collected 7/2000.
                                                          15

-------
                  Sediment     Fe°
        i 1000n
                -30 -20  -10   0    5   10   15   20  25  30  35  40
           Subcore thickness relative to Fe°/Sediment interface, cm
Figure 11. Biomass concentration in Elizabeth City cores collected June, 2000 (upgradient aquifer/iron region). PLFA data collected
        from core EC060200-1 (open circles) and core EC060300-4 (open squares).
                  Sediment    Fe
           100 1
            75-
 D  SRBs
 •  Metal red.
 D  G+/AnG-
-O- Biomass
             0
-600
-500
 400
-300
-200
-100
T3
f
                                                                         CO
                                                                         JD
                                                                         O
                                                                         Q.
                                                                         V)
                                                                         CO
                                                                         CO
                                                                         g
                                                                         in
                -30 -20 -10   0   5  10   15  20  25  30  35  40
               Subcore thickness relative to Fe°/Sediment interface, cm
Figure 12.  Concentration and distribution of PLFA in Fe° from Elizabeth City core EC060300-4 (upgradient aquifer/iron region).
                                        16

-------
a;
EL
        0
                                                               Fe°    Sediment
                                                                                               100
                                                                                    -25
0
                                                                                                     f
                                                                                                      CO
                                                                                                     _(D
                                                                                                      O

                                                                                                      a.
                                                                                                      co
                                                                                                      CO
                                                                                                      CD
                                                                                                      E
                                                                                                      g
                                                                                                     CQ
              40   35   30   25  20   15   10    5     0     10   20   30
           Subcore thickness relative  to  Fe°/Sediment interface,
                                                                                                     cm

Figure 13.   Concentration and distribution of PLFA in Fe° from Elizabeth City (downgradient iron/aquifer region; core EC060300-6).
Summary:  PRB Long-term Performance

The types of mineral precipitates that form in Fe° barriers are
consistent with those predicted with geochemical models. The
principal factors that determinethe amount of mineral precipitation
and the spatial distribution of precipitates in reactive iron media
are flow rate, groundwater chemistry, and microbial community
structure.  After four years of operation, the Elizabeth City and
DFC reactive barriers have developed consistent patterns of
spatially  heterogeneous  mineral precipitation and microbial
activity.  The development of precipitation and biomass fronts
result from the abrupt  geochemical changes that  occur at
upgradient interface regions coupled with groundwater  mass
flux.  Upgradient regions at both sites investigated in this study
have witnessed the greatest accumulation of mineral mass and
biomass.   Neither of the sites of this study show complete filling
of available pore space  after four years,  suggesting that flow
characteristics  should  not be affected by the accumulation of
authigenic components.  However, even relatively thin coatings
of mineral precipitates that do not affect flow patterns may affect
the reactivity of iron particles with respect to the degradation of
chlorinated organic compounds by  diminishing electron flow
capacity.  At Elizabeth City, chromium concentrations have been
reduced to less that 0.01  mg/L and detectable concentrations of
chromium have not been observed in any of the downgradient
compliance wells since November, 1996. Similarly, concentrations
of chlorinated organic compounds (TCE, cis-DCE, VC) at Elizabeth
City are below regulatory target levels in downgradient compliance
wells. At the  DFC, treatment of chlorinated compounds with Fe°
has been  equally successful in gates 1 and 3. In gate 2, detection
of 1,1-DCE at downgradient sampling points has been linked to
impacts of the funnel-and-gate system on groundwater flow and
                                                 bypass of contaminants underneath the reactive zone or residual
                                                 contamination in downgradient sediments (McMahonetal., 1999).
                                                 Contaminant breakthrough, particularly of 1,1 -DCE is likely related,
                                                 either directly or indirectly, to anomalous buildup of authigenic
                                                 precipitates and biomass on the iron surfaces which could cause
                                                 a loss of effective reactivity, and therefore decreased efficiency
                                                 of contaminant degradation  processes in the reactive media.
                                                 Residual contamination downgradient of gate 2 is also a likely
                                                 source of contaminants (FHWA, 1999).  In addition to mineral/
                                                 biomass accumulation in gate 2, potential indicators of decreased
                                                 performance are increased  Eh values, decreased  dissolved
                                                 hydrogen values,  increases in relative specific  conductance
                                                 values and concentrations of certain solutes, such as sulfate and
                                                 magnesium.

                                                 Acknowledgments

                                                 WethankF. Beck, P. Clark, M. McNeil, C. Paul, F. Kahn, K. Jones,
                                                 and J. Cloud for invaluable field and laboratory assistance. We
                                                 also gratefully acknowledge the support provided by ManTech
                                                 Environmental Research Services Corp.  J. P. Messier (USCG)
                                                 is thanked for providing site assistance at the USCG-SC and
                                                 C. Eriksson (FHWA) and J. Jordan (FHWA) are thanked for site
                                                 assistance at the  DFC.

                                                 Notice

                                                 The U. S. Environmental Protection Agency through its Office of
                                                 Research and Development funded the research described here.
                                                 This research brief has been subjected to the Agency's peer and
                                                 administrative review  and approved for publication as an  EPA
                                                 document. Mention of trade names or commercial products does
                                                 not constitute endorsement or recommendation for use.
                                                        17

-------
References

Battelle. Recommendations for Modeling the Geochemical and
    Hydraulic Performance of a Permeable Reactive  Barrier,
    Prepared for Naval Facilities Engineering Service  Center:
    Port Hueneme, CA, 2000.

Beck, F., Clark, P., and Puls, R. Ground WaterMon. Rem. 2000,
    20, 55-59.

Benning, L. G., Wilkin, R. T., Barnes, H.  L. Chem. Geol. 2000,
    167, 25-51.

Bethke, C.  M.  The Geochemist's  Workbench, version 3.0, A
    users guide to Rxn, Act2, Tact, React, and Gtplot. University
    of Illinios, Urbana 1998, 184 pp.

Blowes, D.W., Gillham, R.W., Ptacek, C.J., Puls, R.W., Bennett,
    T.A., O'Hannesin, S.F., Bain, J.G., Hanton-Fong, C.J.,  Bain,
    J.G. EPA/600/R-99/095a 1999, 111 pp.

Blowes, D.W., Puls, R.W., Gillham,  R.W., Ptacek, C.J., Bennett,
    T.A., Bain, J.G., Hanton-Fong,  C.J., Paul, C.J. EPA/600/R-
    99/095b 1999,207pp.

Blowes, D. W., Mayer,  K. U. EPA/600/R-99/095c 1999,  38 pp.

Bonin, P. M. L., Odziemkowski, E. J., Reardon, E. J., Gillham, R.
    W. J. Sol. Chem. 2000, 29, 1061-1074.

Bourrie, G, Trolard, F., Genin, J., Jaffrezic, A., MaTtre,  V.,
    Abdelmoula,  M.   Geochim.  Cosmochim  Acta. 1999,
    63, 3417-3427.

Butler, E. C., Hayes, K. F. Environ. Sci. Technol.2000, 34, 422-429.

Cantrell, K.  J., Kaplan,  D. I., Wietsma, T. W. J. Haz. Mat. 1995,
    42, 201-212.

Chapelle, F. H., Vroblesky, D. A., Woodward, J. C., Lovely,  D. R.
    Environ. Sci. Technol. 1997, 31, 2873-2877.

Cord-Ruwish, R.   Environmental  Microbe-Metal Interactions,
    ASM Press: Washington, D.C., 2000, 159-173.

Dowling, N.J.E., Widdel, F., White, D. C. J. Gen. Microbiol. 1986,
    132, 1815-1825.

Edlund, A.,  Nicols, P. D., Roffey, R., White, D. C.  J. Lipid Res.
    1985, 26, 982-988.

FHWA. IM#1 Performance Evaluation Report. 1999.

Gavaskar, A. R., Gupta, N., Sass, B.  M., Jansoy, R. J., O'Sullivan,
    D. Permeable Barriers forGroundwaterRemediation Design,
    Construction,  and Monitoring,  Battelle Press: Columbus,
    OH, 1998.

Genin, J, Refait, P., Bourrie, G., Abdelmoula, M., and Trolard, F.
    Appl. Geochem. 2001, 16, 559-570.

Guckert, J.B., Antworth, C.P., Nicols, P.O., White, D.C. FEMS
    Microbiol. £co/. 1985, 31, 147-158.

Herbert, R.  B., Benner, S. G., Blowes, D. W. Appl. Geochem.
    2000,  15, 1331-1343.

Hernandez, G., Kucera, V., Thierry, D., Pedersen, A., Hermansson,
    M. Corrosion 1994, 50, 603-608

Hsu, K. J. Chemistry of dolomite formation.  In Carbonate rocks:
    Origin, Occurrence, and Classification (G. Chilingar, H. Bissel,
    and R. Fairbridge, eds.),  1967,Elsevier,  Amsterdam,
    pp. 169-191.
Lee, W., Andowski, Z.L., Nielsen, P.H., Hamilton, W.A. Biofouling
    1995, 8, 165-194.

Liang, L., Korte, N., Gu, B., Puls, R., Reeter, C. Adv. Environ. Res.
    2000, 4, 273-286.

Mackenzie, P. D., Horney, D. P., Sivavec, T. M. J. Haz. Mat. 1999,
    68, 1-17.

Mayer, K. U., Blowes, D. W., Frind, E. O. Water Resour. Res.
    2001, 37, 3091-3103.

McGill,  I. R., McEnaney,  B., Smith,  D.C.   Nature 1976,
    255,200-201.

McMahon, P. B., Dennehy,  K. F., Sandstrom,  M. W.   Ground
    Wafer 1999, 37, 396-404.

Morales, J., Esparza,  P., Gonzalez, S., Salvarezza, R., Arevalo,
    M.P. Corros. Sci. 1993,  34, 1531-1540.

Morrison, S. J.,  Metzler, D. R., Carpenter, C. E.  Environ. Sci.
    Technol. 2001, 35, 385-390.

O'Hannesin, S.  F.,  Gillham, R.  W.   Ground Water  1998,
    36, 164-170.

Parkes, R. J., Dowling,  D. C., White, D. C., Herbert, R. A., and
    Gibson, G. R. FEMS Microbiol. Ecol. 1992, 102, 235-250.

Phillips, D. H., Gu, B., Watson, D. B., Roh, Y., Liang, L., Lee, S.
    Y. Environ. Sci.  Technol. 2000, 34, 4169-4176.

Powell,  R. M., Puls,  R. W., Hightower,  S. K.,  Sabatini, D. A.
    Environ. Sci. Technol. 1995, 29, 1913-1922.

Pratt, A. R., Blowes, D. W., Ptacek, C. J.  Environ. Sci. Technol.
    1997, 31, 2492-2498.

Puls,  R. W., Powell,  R. M.   Ground  Water Mon. Rev. 1992,
    Summer, 130-141.

Puls, R. W., Paul, C.  J., Powell, R. M.  Appl. Geochem. 1999,
    14, 989-1000.

Reardon, E.  J. Environ. Sci.  Technol. 1995, 29, 2936-2945.

Scherer, M. M., Richter, S., Valentine, R. L., Alvarez, P. J. J. Crit.
    Rev. Environ. Sci. Technol. 2000,  30, 363-411.

Schwertmann, U., Fechter, H.  Clay Min. 1994,  29, 87-92.

Schwertmann, U., Taylor, R. M.  Minerals in Soil Environments,
    eds. J.B. Dixon and S.  B. Weed, Soil Science Society of
    America Book Series, No. 1, 1989, Ch. 8, pp. 379-438.

Trudinger, P. A., Chambers, L. A., Smith, J. W. Can. J. Earth Sci.
    1985, 22, 1910-1918.

Vogan, J.L., Focht, R.M., Clark, D. K., Graham, S. L. J. Haz. Mat.
    1999, 68, 97-108.

Von Wolzogen Kuehr, C.A.H, van  der Vlugt, I.S. Wafer 1934,
    18, 147-165.

White, D.C., Bobbie, R.J., Nikels, J. S., Fazio, S.  D., Davis, W. M.
    Botanica Mar. 1980, 23, 239-250.

Zhabina, N.N. and Volkov, I.I.  Environmental Biogeochemistry
    and Geomicrobiology, ed. W. Krumbein, Ann Arbor Science
    Publishers,  1978, pp. 735-746.
                                                         18

-------
x°/EPA
      United States
      Environmental Protection
      Agency
Please make all necessary changes on the below label,
detach or copy, and return to the address in the upper
left-hand corner.
If you do not wish to receive these reports CHECK HERED;
detach, or copy this cover, and return to the address in the
upper left-hand corner.

PRESORTED STANDARD
POSTAGE & FEES PAID
EPA
PERMIT No. G-35

      National Risk Management
       Research Laboratory
      Cincinnati, OH 45268

      Official Business
      Penalty for Private Use
      $300
      EPA/600/S-02/001
      March 2002

-------