Comment on "Potential Impacts of Leakage from Deep CO2 Geosequestration on Overlying Freshwater Aquifers"

Abstract

Carbon Capture and Storage may use deep saline aquifers for CO 2 sequestration, but small CO 2 leakage could pose a risk to overlying fresh groundwater. We performed laboratory incubations of CO 2 infiltration under oxidizing conditions for >300 days on samples from four freshwater aquifers to 1) understand how CO 2 leakage affects freshwater quality; 2) develop selection criteria for deep sequestration sites based on inorganic metal contamination caused by CO 2 leaks to shallow aquifers; and 3) identify geochemical signatures for early detection criteria. After exposure to CO 2 , water pH declines of 1-2 units were apparent in all aquifer samples. CO 2 caused concentrations of the alkali and alkaline earths and manganese, cobalt, nickel, and iron to increase by more than 2 orders of magnitude. Potentially dangerous uranium and barium increased throughout the entire experiment in some samples. Solid-phase metal mobility, carbonate buffering capacity, and redox state in the shallow overlying aquifers influence the impact of CO 2 leakage and should be considered when selecting deep geosequestration sites. Manganese, iron, calcium, and pH could be used as geochemical markers of a CO 2 leak, as their concentrations increase within 2 weeks of exposure to CO 2 .

Full text

Potential Impacts of Leakage from
Deep CO2Geosequestration on
Overlying Freshwater Aquifers
MARK G. LITTLE*
,†
AND
ROBERT B. JACKSON
†,‡
Center on Global Change, Duke University, Durham,
North Carolina 27708, United States, and Nicholas School of
the Environment and Biology Department, Duke University,
Durham, North Carolina 27708-0338, United States
Received July 20, 2010. Revised manuscript received
October 12, 2010. Accepted October 13, 2010.
Carbon Capture and Storage may use deep saline aquifers
for CO2sequestration, but small CO2leakage could pose a risk
to overlying fresh groundwater. We performed laboratory
incubations of CO2infiltration under oxidizing conditions for
>300 days on samples from four freshwater aquifers to
1) understand how CO2leakage affects freshwater quality; 2)
develop selection criteria for deep sequestration sites based on
inorganic metal contamination caused by CO2leaks to
shallow aquifers; and 3) identify geochemical signatures for
early detection criteria. After exposure to CO2, water pH declines
of 1-2 units were apparent in all aquifer samples. CO2
caused concentrations of the alkali and alkaline earths and
manganese, cobalt, nickel, and iron to increase by more than
2 orders of magnitude. Potentially dangerous uranium and
barium increased throughout the entire experiment in some
samples. Solid-phase metal mobility, carbonate buffering capacity,
and redox state in the shallow overlying aquifers influence
the impact of CO2leakage and should be considered when
selecting deep geosequestration sites. Manganese, iron, calcium,
and pH could be used as geochemical markers of a CO2
leak, as their concentrations increase within 2 weeks of exposure
to CO2.
Introduction
Carbon Capture and Storage (CCS) represents a suite of
technologies developed to separate, compress, transport, and
sequester securely underground the CO2produced from
power plants and other industrial facilities. Currently, power
plant flue gases containing CO2are released directly to the
atmosphere where they contribute to the steady rise in
atmospheric CO2and climate change (1, 2). However, a full
understanding of long-term environmental risks is needed
before large-scale CCS implementation is feasible (3-5).
The engineering requirements and potential environ-
mental impact of CCS technologies are currently being
investigated in many locations around the world, including
the Sleipner field in the North Sea (6), offshore Japan (7), and
Otway, Australia (8). In the U.S., the Department of Energy
has initiated seven regional CCS projects in collaboration
with industry and academia under the Regional Carbon
Sequestration Partnership (9). The potential risk of CO2
leakage has already contributed to local opposition to CCS
implementation (10-12). On time scales appropriate for
geosequestration (>1000 yrs), the bulk of CO2sequestered in
a properly chosen saline aquifer is unlikely to escape because
of solubility trapping (13, 14). However, given the hetero-
geneous nature of the subsurface, minor CO2leakage along
faults, old petroleum wells, or other pathways will persist
and will be thermodynamically difficult to seal completely
by precipitation of carbonate minerals (10, 11, 15-17).
Standards for “acceptable” levels of leakage could translate
into tons of CO2annually released from the deep storage
aquifer into intermediate and shallow strata (18). Because
freshwater aquifers used for drinking water, industry, and
agriculture lie directly above possible geosequestration
locations, leaks could form carbonic acid in groundwater
resources before surface leakage of CO2were detected (e.g.,
refs 10,19, and 20).
Although increases in carbonic acid may be buffered by
carbonate dissolution, a lowering of aquifer water pH may
release harmful metals, such as arsenic and uranium, into
the water (21-23). Such impacts on shallow aquifer com-
position have been investigated previously in field injection
studies (23, 24), model simulations (25, 26), and short-term
batch-reaction experiments (27). Our study focuses on the
long-term impacts of CO2leakage on four relatively shallow
aquifer systems overlying possible deep saline geoseques-
tration sites: Aquia and Virginia Beach in the Virginia and
Maryland tidewater region; Mahomet in Illinois; and Ogallala
in southern high plains of Texas. We performed laboratory
incubations under oxidizing conditions for more than 300
days to 1) understand how CO2leaks from deep geoseques-
tration may affect water quality in overlying shallow drinking-
water aquifers; 2) develop selection criteria for sequestration
sites based on inorganic metal contamination caused by CO2
leaks; and 3) identify geochemical signatures in affected
waters which could be used as early detection criteria.
Methods
Sample Selection. We identified freshwater aquifers that
overlie potential deep saline geosequestration target units
(13-15, 28). To prioritize the potential risks, samples were
collected only from those drinking-water aquifers where
natural, in situ concentrations of As, U, Ra, Cd, Cr, Cu, Pb,
Hg, or Se were greater than 10% of U.S. Environmental
Protection Agency’s primary maximum contaminant level
(MCL) for drinking water (29, 30). Noncontaminated ground-
water concentrations greater than 10% were interpreted as
a qualitative indication of the presence of these metals in
solid phase in the aquifer sediment. Approximately 1 kg of
dry aquifer sediment from 17 distinct locations within the
Aquia (MD), Virginia Beach (VA), Mahomet (IL), and Ogallala
(TX) aquifers was acquired from the Illinois State Geological
Survey, the United States Geological Survey, the Virginia
Department of Environmental Quality, and the Maryland
Geological Survey (Table 1, Figure 1).
Laboratory Incubations. All aquifer sediment samples
were oven-dried for 36 h at 110 °C in order to minimize
microbial activity over the long-duration (>300 days) experi-
ment. Dry samples were divided as follows: two ∼400 g
subsamples were set aside for the groundwater experiment;
one ∼5 g aliquot was powdered on a stainless steel mill for
mass spectrometry; and ∼1 g was epoxy mounted for
microprobe analysis, the latter performed on 6 samples.
Methods for the preparations of rock powders for whole-
* Corresponding author phone: (919)681-7180; fax: (919)660-7425;
e-mail: 6r4h@post.harvard.edu.
†
Center on Global Change.
‡
Nicholas School of the Environment and Biology Department.
Environ. Sci. Technol. 2010, 44, 9225–9232
10.1021/es102235w 2010 American Chemical Society VOL. 44, NO. 23, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 99225
Published on Web 10/26/2010

rock acid digestion and subsequent fluid analysis were
adapted from ref 31 and performed on each of the 17 samples.
The 17 pairs of 400-g samples were weighed precisely
and placed into bottles with 18.6 MΩ/cm nanopure makeup
water at a water-to-rock ratio of 3 to 1, were maintained at
approximately 20 °C, and are hereafter referred to as the
groundwater experiment. One set of 17 bottles was sealed
and placed in an opaque box to serve as the control, and the
TABLE 1.Sample Locations and Descriptions
sample name avg depth (m) well name aquifer reached sample description
AQ1 14.8 AA De 100 Aquia very fine sand
VB1 13 Old Pungo Ferry Road (62A-21) Virginia Beach medium quartz sand with trace
black grains
a
USGS# 363714076063501
VB2 11 Creeds Elementary (62B-15) Yorktown Confining Unit medium quartz sand with
chalcopyrite
a
USGS# 363812076021202
VB3 5 Oceana II (62C-31) Virginia Beach medium quartz sand with
unidentified black grains
a
USGS# 365046076041601
VB4 19.4 Bonney Bright North (63A-1) Yorktown-Eastover coarse quartz sand shells
a
USGS# 363337075595001
VB5 36.1 Beach Gardens Park (63C-30) Yorktown-Eastover coarse quartz sand shells
a
USGS# 365124075590701
MH1 84.7 PIAT 08 03A Mahomet sand, primarily Na/K feldspars,
quartz, dolomite, and Fe
oxides with Ti-oxide present,
trace As
b
MH2 78 CHAM 07 01A
MH3 92.2 CHAM 08 09A
MH4 79.1 CHAM 09 03A
OG1 8.4 MWR-B Ogallala fine sand, primarily quartz and
K/Na feldspars, Fe- and
Ti-oxides, coarse calcite
grains present
b
,
c
OG2 9.9 USGS# 335830102444201
OG3 12.3 MAPLE (MPL) fine sand, primarily quartz,
calcite, Ca/K feldspars, and
unidentified alumino-silicates
present, trace Fe/Ti oxides
b
,
c
OG4 19.4
OG5 29.1 USGS# 33495102545001
OG6 41.6
OG7 21.1 JWR fine sand, primarily quartz,
K/Na/Ca feldspars, Fe oxide,
calcite present, trace Ti
oxide
a
,
c
OG8 36.1
OG9 * 45.3 USGS# 334043102365501
OG10 48.3
a
Sample description from refs 32 and 33.
b
Qualitative microprobe data included in the sample description.
c
Additional
sample description from ref 45.
FIGURE 1. Areal map and vertical cross sections of central Illinois, the Texas panhandle, and Virginia and Maryland tidewater
region. Confining units are shown in dark gray. Sediment samples are from the Aquia (MD), Virginia Beach (Columbia/Yorktown)
(VA), Mahomet (IL), and Ogallala (TX) aquifers. Possible CO2sequestration strata indicated with asterisks (37-44).
9226 9ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 23, 2010

other set of 17 bottles was prepared for exposure to CO2,or
“+CO2”, as follows. A stream of 99.8% pure CO2was piped
at atmospheric pressure to each bottle through 17 individually
flow-regulated channels at a constant flow rate of 0.2 L/min
for 320 to 344 days. This flow rate is roughly equivalent to
0.005% of the CO2emissions at a typical 500 megawatt coal-
fired power plant (1). Each channel is fed into the bottles
through a 2-hole rubber stopper and delivers the CO2into
the bottles via a plastic bubble diffuser that is completely
inundated and ∼1 cm above the surface of the aquifer
sediment. The other hole in the stopper is connected to an
exhaust tube leading from the headspace of each incubation
bottle. Each bottle has an independent CO2delivery and exit
system and is covered in aluminum foil to minimize
photosynthesis.
At intervals of a week to a month (for a maximum of 344
days for the Mahomet and Ogallala samples, 320 days for
Virginia Beach, and 334 days for Aquia), ∼3 mL aliquots were
removed from each bottle and tested for pH within 10 s to
minimize the effect of degassing. A second aliquot was
removed and forced through a syringe with 2-µm filter paper
to produce 2 mL of filtered experimental groundwater from
each bottle. The 2-mL aliquots were diluted to 10 mL with
quartz distilled 18 MΩ/cm water for ICP-MS analysis. After
each sampling, nanopure water was added to the bottles to
maintain the 3:1 water-to-rock ratio. Sampling and evapora-
tion accounted for water losses in the range of 5 to 20% for
a 30-day sampling interval; thus the maximum effect of these
losses is an artificial element concentration increase of 20%.
Bottles were gently agitated for ∼10 s after each sampling
and at no other time during the experiment.
Inorganic elemental concentrations were determined by
inductively coupled plasma mass spectrometry (ICP-MS) on
a VG PlasmaQuad 3 at the Duke University Division of Earth
and Ocean Sciences. For the whole-rock samples, Li, V, Cr,
Co, Ni, Cu, Zn, Rb, Sr, Mo, Ba, and U were analyzed; for the
experimental groundwater, Li, B, Mg, Al, Ca, V, Cr, Mn, Fe,
Co, Ni, Cu, Zn, As, Se, Rb, Sr, Mo, Cd, Ba, and U were analyzed.
Instrument calibration standards were prepared from serial
dilutions of certified water standard NIST1643e. Samples and
standards were diluted to the same proportions with an
internal standard solution in 2% HNO3containing 10 parts
per billion (ppb) In, Tm, and Bi to monitor and correct for
instrument drift. All dilutions were carried out with solutions
FIGURE 2. pH in groundwater experiments plotted against time.
+CO2experiment data collection began on day ∼10. Control
experiment data collection began on day ∼45.
TABLE 2. Average Water pH and Alkali and Alkaline Earth Element Concentrations of the Groundwater Experiment, +CO2, and
Control Values
a
pH Li, ppb Mg, ppm Ca, ppm Rb, ppb Sr, ppb
MCL 6.5-8.5
+CO2ctrl +CO2ctrl +CO2ctrl +CO2ctrl +CO2ctrl +CO2ctrl
AQ1 4.47 5.79 6.6 2.7 3.4 0.27 6.2 0.5 10.2 3.2 29 1.6
VB1 4.22 4.24 25 21 12 14 8.2 7.4 1.3 1.2 113 116
VB2 4.26 5.27 2.5 0.12 2.5 0.08 5.9 0.4 2.0 0.40 175 5.1
VB3 3.11 3.12 72 73 14 16 26 18 11 9 54 35
VB4 5.18 6.17 35 22 17 23 103 106 3.3 3.2 450 381
VB5 5.73 6.72 79 53 19 24 517 501 8.0 8 4230 4243
MH1 5.64 7.41 13 2.2 15 3.6 276 18 2.4 1.1 630 82
MH2 5.64 7.36 15 7.2 14 3.0 288 18 3.3 1.7 911 124
MH3 5.63 7.56 13 4.5 14 2.4 302 12 3.0 1.2 642 63
MH4 5.85 6.89 16 12 16 17 239 88 2.6 2.4 596 277
OG2 5.65 7.21 94 41 16 6.5 275 20 2.2 1.2 1894 358
OG3 5.77 7.26 250 43 21 17 323 41 2.7 1.2 3971 1488
OG5 5.73 7.37 26 11 17 6.0 328 17 2.3 0.74 2856 377
OG6 5.80 6.84 45 17 16 7.8 273 16 2.4 0.73 3425 375
OG7 5.74 7.20 40 7.2 17 5.4 348 20 1.8 0.60 2064 358
OG8 5.69 7.36 38 16 16 11 279 23 1.7 0.92 2159 537
OG10 5.67 7.64 19 9.4 19 6.3 314 12 2.0 0.80 1938 306
a
Data averaged over the final five sampling dates for the +CO2experiment: AQ on days 194, 244, 272, 305, and 334; VB
on days 180, 230, 258, 291, and 320; MH and OG on days 204, 254, 282, 315, and 344. Data averaged over two sampling
dates for the control (ctrl): AQ on days 194 and 305; VB on days 180 and 291; MH and OG on days 204 and 315. Averages
do not include Mg data from AQ on day 194, VB on day 180, MH and OG on day 204. These days all correspond with
sampling date 12/16/09 when drastic deviations in various element concentrations in both the +CO2and control
experiments were observed.
VOL. 44, NO. 23, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 99227

prepared from deionized-quartz-distilled H2O and quartz-
distilled HNO3. Instrument drift was also monitored and
corrected for each element analyzed by analysis of the
calibration standards at regular intervals between analysis
of samples and standards. To allow for determination of U
and Th concentrations, NIST1643e was spiked with plasma-
grade single-element solutions of U and Th prior to serial
dilution. External precision for most elements is typically 4%
or less, based on replicate analysis of samples and standards.
The microprobe analysis employed an energy dispersive
spectrometry on a Cameca Camebax electron microprobe
for a qualitative identification of mineral grain cross sections.
The microprobe was operated at 15 kV accelerated potential
and 15 µAmp beam current on polished samples. Data were
reduced using 4Pi Revolution software.
Results
The chemical composition of our groundwater experiments
was significantly affected by the addition of CO2. All +CO2
groundwater experiments produced a pH below EPA’s
minimum MCL of 6.5 units (Table 2, Figure 2). Most alkali
and alkaline metal concentrations (Li, Mg, Ca, Rb, Sr) in the
+CO2experiments were >30% higher than in the control
(Table 2). For instance, Li shows a significant, time-dependent
increasing trend (Figure 3a). Decreases of Mg concentrations
in MH4, VB3, VB4, and VB5 and decreases of Ca in VB4 were
the only exceptions to the enhanced dissolution of the earth
metals in response to the addition of CO2.
Concentrations of some transition metals, including Mn,
Fe, Co, Ni, and Zn, were higher by more than 1000% in +CO2
experiments relative to the control treatments across all
aquifers (Table 3). In general, the high concentrations
observed in the +CO2experiments were apparent within 2
weeks of exposure and did not continue to increase over the
remainder of the experiment. However, Co in M1, M2, and
M3 did exhibit a significant, logarithmic time-dependent
increase over the entire experiment (Figure 3b). The remain-
ing transition metals did not behave the same across all
aquifers. Cd in the +CO2Aquia, Virginia Beach, and Mahomet
experiments was higher than the control treatments by as
much as 1000%. Yet, Cd in the +CO2Ogallala experiments
produced lower or roughly equivalent concentrations than
the control. Al, V, and Cr in +CO2Virginia Beach samples
were elevated relative to the control; however, Al, V, and Cr
in +CO2Aquia, Mahomet, and Ogallala experiments were
lower relative to the control. Cu was generally higher in the
Aquia and Ogallala +CO2experiments but lower in MH2,
MH3, VB1, VB3, and OG2 +CO2experiments relative to the
control treatments. +CO2Mo was lower than the control by
1 to 2 orders of magnitude in all aquifers and produced a
significant, time-dependent logarithmic decrease in Maho-
met samples (Figure 3c).
In response to exposure to CO2, B, Ba, Tl, and U values
were typically higher and As, Se, and Sb values were lower
in the Aquia, Mahomet, and Ogallala samples relative to the
control treatments (Figure 4, Table 4). Ba and U in some
Ogallala and Mahomet +CO2experiments produced a
significant, time-dependent logarithmic increase (Figure
3d,e), while As in some Ogallala +CO2experiments produced
a time-dependent decrease (Figure 4f). In the Virginia Beach
+CO2experiments, B and Ba were higher than in the control
while As was lower relative to the control. Se, Sb, Tl, and U
increased in some +CO2VB samples and decreased in others
relative to the control (Table 4).
FIGURE 3. +CO2concentrations plotted against time and fit to log curves unless indicated as linear. [A] Li: OG2 R2)0.84; OG3 R2)
0.84; OG5 R2)0.90; OG6 R2)0.72; OG7 R2)0.85; OG8 R2)0.86; OG10 R2)0.63. [B] Co: MH1 R2)0.78; MH2 R2)0.90; MH3 R2)
0.71. [C] Mo: MH1 R2)0.78; MH2 R2)0.47; MH3 R2)0.82; MH4 R2)0.73. [D] Ba: OG6 R2)0.76; OG7 R2)0.86; OG10 R2)0.80;
MH1 R2)0.61; MH3 R2)0.62. [E] U: OG2 R2)0.89; OG5 R2)0.79; OG6 R2)0.62; OG7 R2)0.53; OG8 R2)0.78. [F] As: OG2 R2)
0.72; OG5 R2)0.73; OG7 R2)0.83; OG8 R2)0.68; OG10 (linear) R2)0.75.
9228 9ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 23, 2010

Concentrations that exceeded their primary or secondary
U.S. EPA MCL health standard and concentrations that
exhibited a significant, time-dependent increase were of
concern because the ultimate long-term equilibrium con-
centration may exceed the MCL (29). Mn and Fe exceeded
the MCL in most samples, while Al exceeded its MCL only
in VB1, VB2, and VB3. Zn concentrations in VB3 reached
∼50% of the secondary MCL (2500 ppb) and Cd concentra-
tions in VB3 and AQ1 exceeded 50% of the primary MCL
(>2.5 ppb), but neither element exhibited a significant, long-
term increasing trend. Cr and Cu concentrations were ∼3
orders below MCL in all samples. Although As concentrations
in some control Ogallala experiments exceeded the 10 ppb
MCL, consistent with natural field measurements (34), the
addition of CO2caused a decrease in As concentrations
(Figure 3f). In some Ogallala and Mahomet samples, Ba
displayed a significant, time-dependent increase, reaching
25 to 50% of the MCL (Figure 3d). Se, Sb, and Tl did not show
a significant increasing trend; however, Se )22 ppb in VB3
(selenium MCL )50 ppb) and Tl>0.15 ppb in VB3, MH2,
and MH4 (thallium MCL )0.5 ppb). In both +CO2and control
VB5 experiments, U exceeded the 30 ppb MCL. Uranium in
the +CO2OG3 experiment also exceeded the MCL, and other
Ogallala samples exhibited a time-dependent increasing
trend without exceeding the MCL (Figure 3e). The continued
time-dependent increase of Ba and U over >300 days of
exposure to CO2justifies the need for even longer laboratory
experiments and monitoring.
The carbonate-poor experiment samples (AQ1, VB1, VB2,
VB3, and VB4) produced the lowest pH values and the lowest
Ca concentrations among +CO2and control experiments
(Tables 2 and 3). The low pH/low Ca composition may have
resulted from mixing of nanopure water with carbonate-
poor sediment and, in the case of the Virginia Beach samples
which contain chalcopyrite, from the formation of sulfuric
acid (32, 33). These samples, and particularly VB3, also
produced the highest concentrations of Al, Cr, Co, Ni, Zn,
Cd, As, and Se.
Discussion
Chemical detection of leakage into shallow aquifers from a
deep CO2geosequestration site will be an integral part of a
safe CCS system. CO2that infiltrates an unconfined freshwater
TABLE 3. Average Transition Metal Element Concentrations of the Groundwater Experiment, +CO2, and Control Values
a
Al, ppb V, ppb Cr, ppb Mn, ppb Fe, ppb Co, ppb Ni, ppb Cu, ppb Zn, ppb Mo, ppb Cd, ppb
MCL 50 100 50 300 1300 5000 5
+CO2ctrl +CO2ctrl +CO2ctrl +CO2ctrl +CO2ctrl +CO2ctrl +CO2ctrl +CO2ctrl +CO2ctrl +CO2ctrl +CO2ctrl
AQ1 5.6 29.6 0.13 0.44 0.13 2.8 109 3.9 9 63 0.73 0.03 4.6 0.16 1.84 0.12 25 36 0.13 0.06 3.0 0.12
VB1 106 162 2.0 1.5 6.1 4.4 354 393 1163 224 69 4.0 24 7.6 3.2 1.7 1299 118 0.15 0.13 0.57 0.37
VB2 108 1.4 0.06 0.13 0.21 0.08 553 63 3.6 3.7 2.8 0.05 0.47 0.13 736 28 0.13 0.02 0.67
VB3 5341 5887 1.6 0.11 16 18 611 799 4729 188 789 953 1862 2135 11.0 27 2432 1631 0.13 0.05 2.8 3.3
VB4 11.1 3.6 0.20 0.15 0.31 0.23 1723 1139 3598 187 11 2.2 18 4.4 0.66 0.9 163 42 0.54 1.6 0.11 0.04
VB5 3.7 1.8 1.6 0.66 2.3 1.1 431 20 4604 1055 4.1 0.9 24 10 1.9 2.4 53 22 23 94 0.70 0.13
MH1 2.1 45 0.14 0.54 0.41 1.17 616 12 458 30 6.8 0.03 25 0.42 2.6 1.8 69 111 0.62 31 0.18 0.03
MH2 1.6 7.0 0.16 0.35 0.42 0.23 495 62 471 29 36 0.15 80 0.82 2.2 3.7 123 32 1.5 16 0.57 0.01
MH3 2.1 29 0.28 4.3 0.41 2.5 503 5.0 517 13 15 0.16 27 0.9 1.05 2.5 117 72 0.9 37 0.25 0.03
MH4 0.8 2.2 0.05 0.05 0.46 0.32 511 240 384 170 7.2 0.25 27 1.8 1.20 0.51 15 43 0.22 0.07 0.13
OG2 2.6 3.7 1.9 70 1.0 1.2 552 2.6 481 29 12 0.06 171 1.4 13 15 26 7 20 66.95 0.07 0.07
OG3 21 2.3 3.9 91 1.0 0.69 36 0.2 577 64 1.4 0.06 14 0.7 1.3 0.42 12 32 89 79.95 0.10 0.08
OG5 6.8 8.8 14 33 1.8 4.0 461 0.3 558 26 1.7 0.02 35 0.33 2.2 0.44 129 34 3.2 48.10 0.03 0.04
OG6 2.8 6.2 17 31 7.2 5.0 444 0.4 468 22 0.46 0.04 11 0.29 1.5 0.38 87 102 2.1 57.68 0.02 0.06
OG7 3.5 11 7.9 78 0.46 0.55 214 3.9 652 30 1.3 0.08 7.8 0.42 1.7 0.91 15 76 4.8 16.92 0.01 0.01
OG8 1.6 5.4 0.34 36 0.83 4.7 694 10 1889 36 20 0.12 215 0.8 1.30 2.0 69 46 13 63.13 0.05 0.05
OG10 3.4 6.8 14 40 1.0 3.0 470 0.4 516 21 2.5 0.03 58 0.57 2.9 0.80 214 30 7.7 74.63 0.05 0.07
a
Data averaged over the final five sampling dates for the +CO2experiment: AQ on days 194, 244, 272, 305, and 334; VB on days 180, 230, 258, 291, and 320; MH and OG on days
204, 254, 282, 315, and 344. Data averaged over two sampling dates for the control (ctrl): AQ on days 194 and 305; VB on days 180 and 291; MH and OG on days 204 and 315.
Averages do not include Al data from AQ on day 194, VB on day 180, MH and OG on day 204.
FIGURE 4. Co vs Ni in +CO2Virginia Beach and Mahomet
experiments. VB1 R2)0.95; VB4 R2)0.98; VB5 R2)0.74; MH1
R2)0.93; MH2 R2)0.98; MH3 R2)0.61; MH4 R2)0.83.
VOL. 44, NO. 23, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 99229

aquifer under oxidizing conditions and atmospheric pressure
will have an immediate impact on water chemistry by
lowering pH and increasing the concentration of total
dissolved solids. Our results showed that increased Al, Mn,
Fe, Zn, Cd, Se, Ba, Tl, and U concentrations approached or
exceeded their MCL under such conditions. Additionally, Li,
Co, U, and Ba concentrations continued to rise after >300
days of exposure to CO2, indicating that long-term laboratory
experiments are important for understanding the risk of leaks
from CO2sequestration. Moreover, we found significant
fluctuations in Ca, Sr, Mn, and Ba over the course of 300+
days of analysis, which would not have been captured by an
observation window of less than 50 days. For example, Ca
in the Mahomet and Ogallala samples fluctuated between
200 and 400 ppm over the course of the experiment.
A 14-day batch experiment by Lu et al. (27) and a 30-day
MSU-ZERT field injection study (24) also showed 1 to 2 unit
pH declines and increases in Mg, Ca, Sr, Mn, Co, Zn, Cd, and
Ba after exposure to CO2. However, Rb, Fe, Ni, Cu, and U
increased in our experiment but decreased in the Lu et al.
study (Tables 2, and 3 and Figure 3e). This discrepancy may
be explained by a stable, but slowly declining pH trend in
our data (Figure 2) in contrast with a pH rebound that may
have caused reprecipitation of the elements in the Lu et al.
data. Aluminum and the oxyanion-forming trace metals (e.g.,
As, Se, Sb, Mo, V, and Cr) decreased in our study and in the
Lu et al. batch experiment but increased at the MSU-ZERT
site, consistent with the behavior of oxyanion-forming metals,
which are immobilized in moderately acidic, oxidizing
aqueous systems (34, 24, 27). The coupling of CO2plume
modeling with laboratory experiments under a range of redox
conditions should provide a robust tool for predicting the
areal extent and geochemical impact of leakage (20).
Based on our results, the relative severity of the impact
of leaks on overlying drinking-water aquifers should be
considered in the selection of CO2sequestration sites. In the
event of a CO2leak, Fe and Mn concentrations are likely to
increase, whereas the response of other potentially harmful
metals will be more varied. One primary selection criteria
should be metal availability. For instance, U concentrations
were higher in the Ogallala experiments relative to all aquifers
sampled (Table 4). The amount of carbonate available for
buffering of pH in shallow freshwater aquifers is another
important factor; the highest concentrations of Al, Cr, Co,
Ni, Zn, As, and Se for example were produced from our most-
carbonate limited, and lowest pH, samples in our experi-
ments. In addition to acidity, the redox state of the freshwater
aquifer is also important for predicting the behavior of some
elements such as U, which can be released under oxidized
conditions as shown here. In our carbonate-rich samples
(Mahomet and Ogallala), concentrations of As dropped after
CO2addition. However, Apps et al. showed that under
reducing conditions, solid phase As can provide a contami-
nation source for more than 100 years (25).
Given the potential impacts of CO2leaks, chemical
signatures in affected waters also provide an opportunity for
early detection of such leaks. In the presence of CO2, the
elements Mn, Fe, and Ca all increased by at least an order
of magnitude above control experiment concentrations
within 100 days. Therefore, all three elements should be
monitored, along with pH, as geochemical markers of CO2
leaks. Dissolved inorganic carbon and alkalinity are also
responsive and stable indicators of a leak (24, 26). Our results
also show that the chemical impact of a CO2leak may differ
within the same aquifer because of interaquifer mineralogical
heterogeneities. For example, wide variations in Virginia
Beach experimental pH and the behavior of Co and Ni, two
elements which can be mobilized as a result of desorption
from Fe and Mn oxyhydroxides (35, 36), suggest a hetero-
geneous lithology (Figure 2). Co/Ni values were similar among
all Mahomet groundwater experiments but varied among
the Virginia Beach samples (Figure 4). Therefore, groundwater
produced from lithologically different strata in aquifers above
geosequestration sites should be regularly monitored to
increase the probability of an advanced detection of CO2
leaks. Such changes may be detectable long before direct
changes in CO2are observed, if at all.
Acknowledgments
We thank C. W. Cook and G. Dwyer for assistance with the
laboratory incubations and analyses on the ICP-MS, A.
Boudreau for use of the microprobe, and D. Vinson, M.
Chandel, S. Osborn, A. Vengosh, B. Dressel (DOE NETL), and
anonymous reviewers for helpful feedback on this manu-
TABLE 4.Average Half-Metal and Other Measured Element Concentrations of the Groundwater Experiment, +CO2, and Control
Values
a
B, ppb As, ppb Se, ppb Ba, ppb Tl, ppb U, ppb
MCL 10 50 2000 0.5 30
+CO2ctrl +CO2ctrl +CO2ctrl +CO2ctrl +CO2ctrl +CO2ctrl
AQ1 55 64 0.04 0.04 0.49 2.2 25 1.5 0.007 0.000 0.03
VB1 361 80 0.81 0.39 1.5 1.4 20 16 0.012 0.012 0.13 0.10
VB2 27 17 0.16 0.12 0.29 562 5.1 0.037 0.002 0.04
VB3 10.1 7.0 5.4 4.7 19 23 22 21 0.156 0.116 5.0 9
VB4 49 44 0.9 0.39 0.05 0.12 11 6.5 0.006 0.001 0.55 0.45
VB5 292 189 6.5 1.5 1.1 1.0 45 20 0.006 0.011 102 33
MH1 70 102 0.09 0.35 4.5 15 469 28 0.035 0.005 2.6 0.83
MH2 84 64 0.23 0.61 0.34 1.2 241 32 0.193 0.050 3.6 1.3
MH3 73 81 0.12 0.77 2.0 3.4 286 13 0.051 0.005 2.0 1.3
MH4 85 77 0.18 0.17 1.6 2.1 120 46 0.200 0.169 11 5
OG2 85 53 0.32 18 0.28 1.7 504 84 0.011 0.002 19 0.61
OG3 44 21 4.1 2.9 4.3 5.3 86 34 0.010 0.002 40 9
OG5 34 20 1.4 12 0.19 1030 221 0.016 0.000 13 0.78
OG6 70 47 1.4 7.0 0.65 0.72 204 9.1 0.062 0.004 16 0.98
OG7 46 31 1.8 5.6 0.14 350 74 0.010 0.001 1.8 1.6
OG8 40 29 0.21 7.8 0.8 1.3 813 148 0.013 0.002 12 1.7
OG10 31 27 2.9 19 0.30 0.09 1013 201 0.016 0.001 8.3 0.54
a
Data averaged over the final five sampling dates for the +CO2experiment: AQ on days 194, 244, 272, 305, and 334; VB
on days 180, 230, 258, 291 and 320; MH and OG on days 204, 254, 282, 315, and 344. Data averaged over two sampling
dates for the control (ctrl): AQ on days 194 and 305; VB on days 180 and 291; MH and OG on days 204 and 315. Averages
do not include B data from AQ on day 194, VB on day 180, MH and OG on day 204.
9230 9ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 23, 2010

script. We also thank the USGS and the people who helped
us obtain sediment samples, including P. McMahon, J. J.
Smith, A. Stumpf, T. S. Bruce, T. Beach, G. Harlow, and D. W.
Bolton. Our research was funded by the Department of Energy
through the National Energy Technology Laboratory (DE-
FE0002197) and by Duke University’s Center on Global
Change. This report was prepared as an account of work
sponsored by an agency of the U.S. Government. Neither the
U.S. Government nor any agency thereof makes any warranty,
express or implied, or assumes any legal liability or respon-
sibility for the accuracy, completeness, or usefulness of any
information, apparatus, product, or process disclosed.
Note Added after ASAP Publication
There was an error in Figure 1 of the version of this paper
published October 26, 2010. The correct version published
October 29, 2010.
Supporting Information Available
S1. +CO2groundwater experiment data by sampling day,
S2. control groundwater experiment data by sampling day,
S3. natural groundwater composition,
S4. whole rock aquifer composition,
S5. geological setting of aquifer samples,
S6. map view of Aquia, Virginia Beach, Mahomet, and
Ogallala aquifers with sample locations. This material is
available free of charge via the Internet at http://pubs.acs.org.
Literature Cited
(1) Carbon Dioxide Capture and Storage; Metz, B., Davidson, O.,
de Coninck, H., Loos, M., Meyer, L., Eds.; Cambridge University
Press: Cambridge, 2005.
(2) Keeling, R. F.; Piper, S. C.; Bollenbacher, A. F.; Walker, J. S.
Atmospheric CO2records from sites in the SIO air sampling
network. Trends: A Compendium of Data on Global Change;
U.S. Department of Energy: Oak Ridge, TN, 2008.
(3) Newmark,R. L.; Friedmann, S. J.; Carroll, S. A. Water Challenges
for Geologic Carbon Capture and Sequestration. Environ.
Manage. 2010,45 (4), 651–661.
(4) Wilson, E. J.; Friedmann, S. J.; Pollak, M. F. Research for
Deployment: Incorporating Risk, Regulation, and Liability for
Carbon Capture and Sequestration. Environ. Sci. Technol. 2007,
41 (17), 5945–5952.
(5) Eccles, J. K.; Pratson, L.; Newell, R. G.; Jackson, R. B. Physical
and Economic Potential of Geological CO2Storage in Saline
Aquifers. Environ. Sci. Technol. 2009,43, 1962–1969.
(6) Gale,J.; Christensen, N. P.; Cutler, A.; Torp, T. A. Demonstrating
the potential for geological storage of CO2: the Sleipner and
GESTCO projects. Environ. Geosci. 2001,8(3), 160–165.
(7) Yamazaki, T.; Aso, K.; Chinju, J. Japanese potential of CO2
sequestration in coal seams. Appl. Energy 2006,83, 911–920.
(8) Etheridge, D.; Leuning, R.; de Vries, D.; Dodds, K.; Trudinger,
C.; Allison, C.; Fraser, P.; Prata, F.; Bernardo, C.; Meyer, M.;
Dunse, B.; Luhar, A. Atmospheric monitoring and verification
technologies for CO2storage at geosequestration sites in Australia;
CO2CRC, Report No. RPT05-0134; 2005.
(9) U.S. Department of Energy National Energy Technology Labo-
ratory Regional Carbon Sequestration Partnerships Website.
http://www.netl.doe.gov/technologies/carbon_seq/partnerships/
partnerships.html (accessed June 1, 2009).
(10) Pearce, J. M.; Shepard, T. J.; Rochelle, C. A.; Kemp, S. J.; Bouch,
J. E.; Wagner, D.; Nador, A.; Veto, I.; Baker, J.; Toth, G.; Lombardi,
S.; Annuziatellis, A.; Beaubien, S. E.; Ciotoli, G.; Pauwels, H.;
Czernichowski-Lauriol, I.; Gaus, I.; Le Nindre, Y. M.; Girard,
J.-P.; Petelet-Giraud, E.; Serra, H.; le Guern-Marot, C.; Schroot,
B.; Orlic, B.; Schuttenhelm, A.; Hatziyannis, G.; Spyridonos, E.;
Metaxas, A.; Gale, J.; Manancourt, A.; Brune, S.; Faber, E.;
Hagendorf, J.; Poggenburg, J.; Teschner, M.; Iliffe, J.; Kroos, B.;
Hildenbrand, S.; Alles, S.; Heggland, R. The Nascent Project Final
Report: Natural Analogues for the Geological Storage of CO2;
2005.
(11) Evans, J. P.; Heath, J.; Shipton, Z. K.; Kolesar, P. T.; Dockrill, B.;
Williams, A.; Kirchner, D.; Lachmar, T. E.; Nelson, S. T. Natural
Leaking CO2-charged Systems as Analogs for Geologic Seques-
tration Sites. Third Annual Conference on Carbon Capture and
Sequestration, Alexandria, VA, 2004.
(12) Haszeldine,R. S.; Quinn, O.; England, G.; Wilkinson, M.; Shipton,
Z. K.; Evans, J. P.; Heath, J.; Crossey, L.; Ballentine, C. J.; Graham,
C. M. Natural Geochemical Analogues for Carbon Dioxide
Storage in Deep Geological Porous Reservoirs, a United Kingdom
Perspective. Oil Gas Sci. Technol. 2005,60 (1), 33–49.
(13) Gilfillan,S. M. V.; Lollar, B. S.; Holland, G.; Blagburn, D.; Stevens,
S.; Schoell, M.; Cassidy, M.; Ding, Z.; Zhou, Z.; Lacrampe-
Couloume, G.; Ballentine, C. J. Solubility trapping in formation
water as dominant CO2sink in natural gas fields. Nature 2009,
458, 614–618.
(14) Benson, S. M.; Cole, D. R. CO2Sequestration in Deep Sedimentary
Formations. Elements 2008,4(5), 325–331.
(15) Damen, K.; Faaij, A.; Turkenburg, W. Health, Safety and
Environmental Risks of Underground CO2Storage - Overview
of Mechanisms and Current Knowledge. Clim. Change 2006,
74, 289–318.
(16) Lewicki, J. L.; Birkholzer, J.; Tsang, C.-F. Natural and industrial
analogues for leakage of CO2from storage reservoirs: identifica-
tion of features, events, and processes and lessons learned.
Environ. Geol. 2007,52 (3), 457–467.
(17) Kharaka,Y. K.; Thordsen, J. J.; Hovorka, S. D.; Nance, H. S.; Cole,
D. R.; Phelps, T. J.; Knauss, K. G. Potential environmental issues
of CO2storage in deep saline aquifers: Geochemical results from
the Frio-I Brine Pilot test, Texas, USA. Appl. Geochem. 2009,24
(6), 1106–1112.
(18) Hepple, R. P.; Benson, S. M. Geologic storage of carbon dioxide
as a climate change mitigation strategy: performance require-
ments and the implications of surface seepage. Environ. Geol.
2005,47 (4), 576–585.
(19) Keating, E. H.; Fessenden, J.; Kanjorski, N.; Koning, D. J.; Pawar,
R. The impact of CO2on shallow groundwater chemistry:
observations at a natural analog site and implications for carbon
sequestration. Environ. Earth Sci. 2010,60 (3), 521–536.
(20) Carroll, S.; Hao, Y.; Aines, R. Geochemical detection of carbon
dioxide in dilute aquifers Geochem. Trans.; 2009, 10 (4).
(21) Smyth,R. C.; Hovorka, S. D.; Lu, J.; Romanak, K. D.; Partin, J. W.;
Wong, C.; Yang, C. Assessing risk to fresh water resources from
long term CO2injection-laboratory and field studies. Energy
Procedia 2009,1(1), 1957–1964.
(22) Solomon, S. Carbon Dioxide Storage: Geological Security and
Environmental Issues Case Study on the Sleipner Gas field in
Norway; Bellona Report; 2007.
(23) Zheng, L.; Apps, J. A.; Spycher, N.; Birkholzer, J. T.; Kharaka,
Y. K.; Thordsen, J. J.; Kakouros, E.;Trautz, R. C. Transient Changes
in Shallow Groundwater Chemistry During the MSU-ZERT CO2
Injection Experiment, American Geophysical Union Fall Meeting
Abstract, 2009; H12B-5.
(24) Kharaka, Y. K.; Thordsen, J. J.; Kakouros, E.; Ambats, G.;
Herkelrather, W. N.; Beers, S. R.; Birkholzer, J. T.; Apps, J. A.;
Spycher, N. F.; Zheng, L.; Trautz, R. C.; Rauch, H. W.; Gullickson,
K. S. Changes in the chemistry of shallow groundwater related
to the 2008 injection of CO2at the ZERT field site, Bozeman,
Montana. Environ. Earth Sci. 2010,60, 273–284.
(25) Apps,J. A.; Zheng, L.; Zhang, Y.; Xu, T.; Birkholzer, J. T. Evaluation
of potential changes in groundwater quality in response to CO2
leakage from deep geologic storage. Transp. Porous Med. 2010,
82, 215–246.
(26) Wilkin, R. T.; Digiulio, D. C. Geochemical impacts to groundwater
from geologic carbon sequestration: controls on pH and
inorganic carbon concentrations from reaction path and kinetic
modeling. Environ. Sci. Technol. 2010,44, 4821–4827.
(27) Lu, J.; Partin, J. W.; Hovorka, S. D.; Wong, C. Potential risks to
freshwater resources as a result of leakage from CO2geological
storage: a batch-reaction experiment. Environ. Earth Sci. 2010,
60, 335–348.
(28) U.S. Department of Energy Carbon Sequestration Atlas of the
United States and Canada Online Website. http://www.
netl.doe.gov/technologies/carbon_seq/refshelf/atlas/index.ht-
ml (accessed June 1, 2009).
(29) U.S. Environmental Protection Agency Drinking Water Con-
taminants Website. http://www.epa.gov/safewater/contami-
nants/index.html (accessed June 1, 2009).
(30) U.S. Geological Survey. National Water-Quality Assessment
(NAWQA) Program Website. http://water.usgs.gov/nawqa/
index.html (accessed June 1, 2009).
(31) Meurer, W. P.; Willmore, C. C.; Boudreau, A. E. Metal redis-
tribution during fluid exsolution and migration in the Middle
Banded series of the Stillwater Complex, Montana. Lithos 1999,
47, 143–156.
(32) Bruce, T. S.; Beach, T. Virginia Beach aquifer water well drilling
logs and completion reports, unpublished, 2009.
VOL. 44, NO. 23, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 99231

(33) Smith, B. S.; Harlow, G. E. Conceptual Hydrogeologic Framework
of the Shallow Aquifer System at Virginia Beach, Virginia. U.S.
Geological Survey Water-Resources Investigations Report; 2001;
01-4262.
(34) Scanlon,B. R.; Nicot, J. P.; Reedy, R. C.; Kurtzman, D.; Mukherjee,
A.; Nordstrom, D. K. Elevated naturally occurring arsenic in a
semiarid oxidizing system, Southern High Plains aquifer, Texas,
USA. Appl. Geochem. 2009,24 (11), 2061–2071.
(35) Alloway, B. J. Heavy Metals in Soils, 2nd ed.; Blackie: Glasgow,
U.K., 1995.
(36) Teutsch, N.; Erel, Y.; Halicz, L.; Chadwick, O. A. The influence
of rainfall on metal concentration and behavior in the soil.
Geochim. Cosmochim. Acta 1999,63 (21), 3499–3511.
(37) Wilson, S. D.; Roadcap, G. S.; Herzog, B. L.; Larson, D. R.; Winstanley,
D. Hydrogeology and ground-water availability in southwest
McLean and Tazwell counties, Part 2: aquifer modeling and final
report. Illinois State Water Survey and Illinois State Geological
Survey Cooperative Groundwater Report; 1998; p 19.
(38) Herzog, B. L.; Larson, D. R.; Abert, C. C.; Wilson, S. D.; Roadcap,
G. S. Hydrostratigraphic Modeling of a Complex, Glacial-Drift
Aquifer System for Importation into MODFLOW. Ground Water
2005,41 (1), 57–65.
(39) Weibel, C. P.; Lasemi, Z. Illinois Geological Quadrangle Map:
IGQ Villa Grove-BG, Illinois State Geological Survey, 2001.
(40) Russell, R. R. Ground-Water Levels in Illinois through 1961. State
of Illinois State Water Survey Report of Investigation; 1963; p 45.
(41) Henry, M. E. Petroleum geology of the Palo Duro Basin and
Pedernal Uplift provinces as a basis for estimates of undiscovered
hydrocarbon resources. U.S. Geological Survey Open-File
Report; 1988; 87-450-U.
(42) Fogg, G. E.; Senger, R. K. Automatic Generation of Flow Nets
with Conventional Ground-Water Modeling Algorithms. Ground
Water 1985,23 (3), 336–344.
(43) Meng,A. A.; Harsh, J. F. Hydrogeologic framework of the Virginia
Coastal Plain. U.S. Geological Survey Professional Paper; 1988;
1404-C.
(44) Harsh, J. F.; Laczniak, R. J. Conceptualization and analysis of
ground-water flow system in the Coastal Plain of Virginia and
adjacent parts of Maryland and North Carolina . U.S. Geological
Survey Professional Paper; 1990; 1404-F.
(45) Lehman, T. Ogallala aquifer drilling notes, unpublished,
2001.
ES102235W
9232 9ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 23, 2010