Chemical Geology 217 (2005) 187 – 199

www.elsevier.com/locate/chemgeo

Modelling CO2 solubility in pure water and NaCl-type

waters from 0 to 300 8C and from 1 to 300 bar
Application to the Utsira Formation at Sleipner
Sandrine Portiera,b,T, Christopher Rochellec
a
Institut Français du Pétrole, Laboratoire de Géochimie, Rueil-Malmaison 92500, France
b
CGS, Institut de Géologie, Université Louis Pasteur, Strasbourg 67000, France
c
British Geological Survey, Hydrothermal Laboratory, Keyworth, Nottingham NG12 5GG, UK
Received 24 May 2003; accepted 10 December 2004

Abstract

The goal of this study is to better quantify the solubility of CO2 in brines. New experimental data on the solubility of CO2 in
a mixed salt solution at CO2 sequestration pressure and temperature conditions are presented. A thermodynamic model for
calculation of the phase equilibria of CO2–H2O–NaCl system is briefly described; notably, the solubility of the carbon dioxide
in the aqueous phase. This model was used to check his validity by comparing the calculation results with new experimental
measurements, available experimental observations and other models CO2 solubility results. Comparison with experimental
data indicates that the model can predict accurately CO2 solubility in pure water and in aqueous NaCl solutions of ionic
strengths up to 3 molal from 0 to 300 8C and from 1 to 300 bar. This model has been applied to CO2 storage at Sleipner, North
Sea. The salinity of the porewater within the Utsira Formation (the CO2 host formation) is approximately the same as seawater.
Predicted CO2 solubility is in good agreement with new experimental measurements.
D 2004 Published by Elsevier B.V.

Keywords: CO2 solubility; CO2 thermodynamics; Aqueous CO2 solution; CO2 storage at Sleipner

1. Introduction recently received much attention. The burning of

fossil fuels and other anthropogenic activity have
As the main contributor to the greenhouse effect, caused a dramatic increase in the concentration of
the impact of atmospheric CO2 on climate has atmospheric CO2 when compared to its value
recorded in ice cores during the last 400,000 years
(Houghton et al., 1996; IPCC, 2001). Capture of
* Corresponding author. Centre de Géochimie de la Surface,
Institut de Géologie, Université Louis Pasteur, 1 rue Blessig, 67000 CO2 from large point sources (e.g. power stations)
Strasbourg, France. and its storage in deep geologic formations has been
E-mail address: sandrine.portier@voila.fr (S. Portier). considered as a means to mitigate global warming
0009-2541/$ - see front matter D 2004 Published by Elsevier B.V.
doi:10.1016/j.chemgeo.2004.12.007
188 S. Portier, C. Rochelle / Chemical Geology 217 (2005) 187–199

(Omerod, 1994). Among proposed sequestration cant contributions to the long-term containment of
strategies, injection into saline aquifers represents CO2 (Pruess and Garcia, 2002).
one of the most promising alternatives. There are a The present study focuses on hydrodynamic
very large number of underground, brine-filled strata trapping, in particular the quantity of CO2 that
that could potentially be used to store CO2. Indeed, can dissolve in formation water. At equilibrium,
storage potential estimates for just the rocks beneath this will be governed by in situ pressure and
the North Sea, suggest that they have the capacity to temperature, and the composition of formation
store over 700 years of current European CO2 water. Within a CO2 storage system these will
emissions from thermal power generation (Holloway vary depending the exact position within the
et al., 1996). Over time, CO2 will dissolve in the aquifer. For example, at the injection point at
interstitial solution of the aquifer and in some Sleipner these conditions are 37 8C, 100 bar, and
formations it would slowly react with minerals to a porewater of approximately seawater salinity
form carbonates, which would lock up the CO2 (Gregersen et al., 1998). However, at the top of
permanently. Suitable aquifers would have also a the Utsira Formation pressures and temperatures
cap rock of low permeability to minimise CO2 will be slightly less.
leakage. Knowledge of the solubility in the CO2–H2O–
The Sleipner project is the world’s first commer- salts system enhances the accuracy of storage
cial-scale storage CO2 project. When injection began calculations. There have been many experimental
in 1996, it marked the first instance of CO2 being studies on the solubility of CO2 in pure water and in
stored in a geological formation because of climate- aqueous NaCl solutions. The solubility was inves-
change considerations (Backlid et al., 1996). Sleip- tigated quantitatively with some variable parameters,
ner is a natural gas field which has about 9% CO2 i.e., pressure, temperature and salinity. However, few
as an impurity in the gas. This is separated from the knowledge is available on the CO2 solubility in
methane at the surface, and then stored deep mixed salt solutions at CO2 sequestration temper-
underground rather than being emitted to the ature and pressure conditions. This study used CO2
atmosphere. Since 1996, nearly a million tons per solubility data from two sources; new data generated
year of CO2 has been injected into the large, deep, under Sleipner-specific conditions, and pre-existing
late Miocene Utsira saline aquifer. The sand-rich data obtained from the literature. First, this paper
Utsira Formation extends under a broad area of the describes the new CO2 solubility data generated
North Sea, and at Sleipner has a thickness of about under Sleipner-specific conditions. The new meas-
200 m and a minimum depth of about 800 m. The urements were made using a synthetic Utsira
dbubbleT of injected CO2 is being monitored by Formation water. The experimental conditions
repeated seismic surveys. Interpretation of the geo- chosen were between 18 and 80 8C and between
physical data and reservoir simulations have been 80 and 120 bar, with most data generated at 37 8C
undertaken as part of a European R&D project and 100 bar—in situ temperature and pressure at the
(SACS, and its follow-on projects SACS2 and injection point in the Utsira formation at Sleipner.
CO2STORE) which are led by the operating Due to the in situ conditions and the nature of the
company Statoil. During underground storage oper- fluid (saline) in which CO2 is likely to be injected,
ations in deep reservoirs, the CO2 can be trapped in modelling CO2 solubility requires taking into account
three main ways (with descriptors from Bachu et al., salinity effects in aqueous solutions as well as the
1994): as dfreeT CO2, most likely as a supercritical non-ideal behaviour of the CO2 gaseous phase. After a
phase (physical trapping); dissolved in formation short presentation of the thermodynamic model for the
water (hydrodynamic trapping); and precipitated in solubility of carbon dioxide, the second part of this
carbonate phases such as calcite (mineral trapping). paper is devoted to a comparison of calculation results
During the early stages of storage, physical trapping with experimental data and other models. On the basis
is likely to be the most dominant trapping mecha- of these results, we finally propose a quantification of
nism. However, over time, hydrodynamic trapping the dissolved CO2 that might be stored in the Utsira
and eventually mineral trapping will make signifi- Sand at the relevant conditions.
 S. Portier, C. Rochelle / Chemical Geology 217 (2005) 187–199 189

2. Materials and methods Oven at

fixed
temperature
2.1. Experimental
Stainless
The CO2 used in this study was high purity steel
(99.99%) liquid CO2 (Air Products, 4.5 Grade). This pressure
vessel Water Filter
liquid CO2 was obtained in a cylinder fitted with a dip assembly
tube and pressurised with 2000 psi (approximately
140 bar) of helium. However, the actual experimental Magnetic stirrer
pressures were controlled by ISCO syringe pumps,
which were run under dconstant pressureT mode. For
experiments below 31.1 8C, the pressure was suffi-
cient for the CO2 to be liquid. However, above this
temperature (i.e. for most of the experiments) the CO2
was a supercritical fluid.
CO2
At the start of the experimental programme it was
decided to make up a single, 25 l stock solution of
synthetic Utsira porewater (SUP), and that this
would be used for all the experiments. This solution Floating piston
sampler partly
was based upon the only analysis of the Utsira filled with
CO2 pump NaOH
solution

Table 1 Fig. 1. Schematic representation of equipment used for CO2

Composition of Oseberg, Brage and SUP porewaters (units are mg solubility studies.
l
1 except for TIC and TOC which are mol l
1)
Analyte Oseberg Average Brage Average synthetic porewater available at the start of the study—from
porewater analyses Utsira porewater
from Gregersen conducted starting solution
the Oseberg field some 200 km north of Sleipner
et al. (1998) at BGS (Gregersen et al., 1998). This synthetic porewater
pH 7.1 7.65 7.77
was close to seawater composition. Three analyses of
(c20 8C) the initial solution were made, and these averaged to
Li – 1.77 – give a representative composition (Table 1) of the
Na 10,392 9934 10,306 starting fluid for the experiments. Subsequently,
K 208 317 225 samples of Utsira porewater from the Brage field
Mg 630 664 633
Ca 426 412 432
(close to the Oseberg field) were obtained. In terms
Sr 10 9.65 10 of major element compositions, both porewaters are
Ba 0.5 6.17 0.31 very similar. Although this does not necessarily
Mn – b0.02 – mean that the porewater is of broadly uniform
Total Fe 2 0.28 1.21 composition throughout the Utsira Formation, this
Al – b0.1 0.35
Total P – b0.1 –
assumption is made given the very limited data
Total S – b2.5 1.05 available.
Si – 23.7 0.58 A schematic diagram of the dbatchT apparatus is
SiO2 – 50.7 1.23 shown in Fig. 1. The equipment consists of a
Cl 18,482 18,921 18,659 stainless steel pressure vessel with Viton O-ring
Br – 71.5 b2.00
NO3 – 8.4 –
pressure seals, containing a PTFE (polytetrafluoro-
SO4 ND b60 b2.00 ethylene) sample container. Two sizes of otherwise
HCO3 707 842 386 identical pressure vessels were used in this study,
TIC (CO2
3 ) – 0.003 – having volumes of either 100 ml or 150 ml. During
TOC b0.0003 assembly, a sample of synthetic Utsira porewater was
190 S. Portier, C. Rochelle / Chemical Geology 217 (2005) 187–199

placed into the PTFE container (filled to approx- be corrected for dilution due to the NaOH. The
imately 2/3 full), together with a magnetic stirrer dilution factor was calculated by assuming that all Cl

bead. The top of the pressure vessel was then was derived from the SUP. By knowing the starting
securely tightened down, and pressure tubing con- and final Cl
concentration, the dilution factor could
nected. The whole assembly was then placed on top be calculated. It was assumed that the experimental
of a magnetic stirrer inside a Gallenkamp PlusII fan- solutions behaved in a similar way to pure water, with
assisted oven that could be maintained to within depressurising and cooling resulting in volume
F0.5 8C. Although the base of the stainless steel changes of less than 1% (assuming no degassing).
pressure vessel was in the order of 1 cm thick, it still This change is considered relatively small compared
allowed for good dcouplingT between the magnetic to uncertainties introduced by the sampling or
stirrer and the stirrer bead. Consequently, the analytical processes. Consequently, compressibility
aqueous solution was well mixed and was in good and expansivity corrections have not been applied to
contact with the overlying CO2. A known pressure the results. Further details are provided in an internal
of CO2 was then admitted to the reactor using a high report of the British Geological Survey (Rochelle and
pressure ISCO 260D syringe pump. This pump was Moore, 2002).
set to dconstant pressureT mode such that it automati- The errors assumed for the data were set at F0.05
cally injected or withdrew CO2 as necessary. This mol.l
1 (2200 mg l
1 as CO2). These are much
minimised the impact of any leaks of CO2 from the larger than the analytical errors, but were set so as to
reactors or pipework. The system was left for at least account for dequipment handlingT variations during
24 h before any samples were taken. sampling. Larger variations caused by obvious
A sample of the CO2-rich aqueous phase was then sampling problems (as noted above) were apparent
removed (still at pressure) and reacted with 4 M from trends in the data, and caused rejection of
NaOH solution to preserve any dissolved carbon suspect data.
species. Analyses of the alkaline preserved samples Various degrees of dscatterT were found in the
were performed by potentiometric titration. Either data. Overly high solubility values were disregarded
0.01 mol l
1 (0.02 N) or 0.5 mol l
1 (1 N) sulphuric as these resulted from breakthrough of dfreeT CO2
acid was used as titrant against 2.0 ml or 1.0 ml of into the sampling tube (often after several samples
samples of pH less than 10 and greater than 11.5 had already been taken and the water–CO2 interface
respectively. The pH of the analytical sample was was relatively low in the vessel). In general, there
monitored as a function of volume of titrant added was also a hint of this when pressure connectors
and the titration was allowed to proceed until a pH were de-coupled after the sample was taken. How-
of less than 2 was achieved. Two quality control ever, it was also possible to get apparently reduced
standards were used. For the samples of pH greater concentrations of dissolved CO2—as a result of
than 11.5, a quality control standard containing 2 pressure reduction and degassing during sampling.
mol l
1 of hydroxyl ions and 0.5 mol l
1 of Although care was taken to minimise reductions in
carbonate ions was prepared by dissolving 40.00 g pressure, these could occur when opening valves to
of BDH AnalaRR sodium hydroxide and 26.4472 g equilibrate pressures in the sampling line, or too fast
of BDH AnalaRR sodium carbonate in 500 ml of a sampling flow rate through the in-line filter. It was
deionised water. For the samples of pH less than 10, relatively difficult to identify if significant pressure
a quality control standard containing 200 mg l
1 reductions occurred during sampling (compared to
bicarbonate was prepared by dissolving 0.2754 g of the cases where CO2 breakthrough occurred). In an
BDH AnalaRR grade sodium bicarbonate in 1000 ml attempt to combat such artefacts, several samples
of deionised water. The relevant QC standards were were taken at a particular pressure and temperature.
analysed at the start and finish of the analytical run The upper bound of the measured CO2 concentra-
and after not more than every ten samples. tions was taken as being more representative of
The drawT analytical data for the experimental actual CO2 solubility. A summary of measured CO2
solution/NaOH solution mixture reported CO32
con- solubility in synthetic Utsira porewater is given in
centrations in mg l
1. However, these values need to Table 2.
 S. Portier, C. Rochelle / Chemical Geology 217 (2005) 187–199 191

Table 2 pressure, the temperature, and the molar fraction of

Summary of CO2 solubility experiments in synthetic Utsira pore- the considered species in the vapour phase, according
water (SUP)
to the following equation:
Temperature (8C) Pressure (bar) CO2 solubility
(mol kg
1 H2O)
fiv ¼ ui xi P ð2Þ
18 100 1.364
18 100 1.259
35 100 1.020 The calculation of gas fugacity necessitates first to
37 80 1.005 evaluate the molar volume which depends on the
37 80 1.020 vapour phase composition, the pressure, and the
37 80 1.000
temperature. The gas fugacity is calculated using the
37 80 1.000
37 80 0.954 Peng and Robinson (1976, 1980) Equation Of State
37 90 1.020 (EOS). This EOS permit to calculate the molar
37 100 0.959 volume of a pure or mixed vapour phase as well as
37 100 1.072 the fugacity coefficient of each gas.
37 100 1.087
Similarly, the fugacity value in the aqueous phase
37 100 1.085
37 110 1.100 depends on the pressure, the temperature, and the
37 120 1.072 solution composition. For the aqueous phase, the
37 120 1.183 fugacity is calculated using a different method for
37 120 1.152 water and for other components. For the solvent, the
37 120 1.121
fugacity is calculated using its vapour pressure
50 80 0.903
50 80 0.936 (Dhima et al., 1998):
50 100 0.956
!
V̄ Haq2 Oh i
50 120 1.105
V
50 120 1.085 fHaq2 O ¼ aH2 O :PHsat2 O :exp sat
P
PH2 O ð3Þ
70 80 0.691 RT
70 80 0.671
70 90 0.690
70 100 0.710 The water activity, a i (dimensionless), is calculated
70 100 0.689 using the expression given by Helgeson (1969):
70 100 0.731
70 100 0.744 lnaH2 O ¼
0:03603:I:HNaCl ð4Þ
70 110 0.862
where H NaCl is the osmotic coefficient as evaluated by
70 120 0.814
70 120 0.890 Helgeson (1969) and I is the true ionic strength of the
80 100 0.744 solution, i.e. the nominal or stoichiometric ionic
80 100 0.814 strength corrected for the fact the NaCl is not
completely dissociated (Helgeson et al., 1981). The
saturated molar volume of the liquid water, V̄ Haq2O is
considered independent of pressure, and is calculated
2.2. Thermodynamics using the expression given by Saul and Wagner
(1987). The water vapour pressure, P Hsat2 O (in bar) is
The phase equilibrium is calculated using equality also determined as suggested by Saul and Wagner
rule of the fugacities: (1987). Since there are no vapour composition
measurements for the CO2–H2O–NaCl system in the
fiv ¼ fiaq ð1Þ temperature range of this study, we assume that water
vapour pressure of the mixtures is the same as pure
While for an ideal gas, the fugacity value in the water saturation pressure.
vapour phase will be equal to that of the partial For the fugacity of the dissolved CO2 in the
pressure (in bar), in reality the non-ideal behaviour aqueous phase, we use Henry’s law approach
should be corrected by a function depending on the (Prausnitz et al., 1986), with the Krichevsky and
192 S. Portier, C. Rochelle / Chemical Geology 217 (2005) 187–199

Kasarnovsky (1935) correction for high pressure based on a specific interaction theory for the liquid
conditions: phase (Pitzer, 1973) and a highly accurate equation of
aq H
  state for the vapour phase (Duan et al. 1992). This
fCO 2
¼ mCO2 cCO2 KCO 2
T ; PHsat2 O model is intended to calculate CO2 solubility in pure
!
V CO2h i
l
V̄ sat
water and aqueous NaCl solutions for temperatures
 exp P
PH2 O ð5Þ from 0 to 2608C and for pressures from 0 to 2000 bar
RT
and for ionic strength from 0 to 4.3 molal. Because of
sat
where K COH2 (T, P H 2O
) is the Henry’s constant of the the scarcity of CO2 solubility data for electrolyte
reaction CO2,aqX CO2,gas. The values of Henry’s systems other than NaCl, especially at high temper-
constant for carbon dioxide were evaluated by atures, the model proposed by Duan and Sun (2003)
Helgeson (1969) for the temperature range from 0 fitted to experimental measurements in the CO2–
8C to 300 8C at saturation pressure of water. The NaCl–H2O system. Nevertheless, with this specific
l
partial molar volume at infinite dilution V̄ CO 2
(T) is interaction approach, this model is able to predict CO2
calculated as suggested by Dhima et al. (1998), using solubility in other systems, such as CO2–H2O–CaCl2
an improvement of the Lyckman et al. (1965) and CO2–seawater, without fitting experimental data
equation. Estimating the activity of aqueous species from these systems.
requires the calculation of their activity coefficient
integrating the effect of salinity and temperature.
Neutral species may also behave in a non-ideal 3. Results and discussion
manner, exhibiting activity coefficients that deviate
from unity. The temperature and salinity dependence It is difficult to compare directly the CO2-synthetic
of activity coefficient for dissolved gases such as Utsira porewater dataset produced in this study with
CO2 is represented using the correlation developed in other data, as little previous experimental data could
Helgeson (1969): be found for the pressure, temperature and salinity of
interest. The more significant investigations of CO2
lncCO2 ¼ rCO2 ðT Þ:I ð6Þ solubility in water and in aqueous NaCl solutions are
listed in Table 3. The most extensive studies of CO2
where r CO2 (T) is a fitting parameter called the
solubility in pure water are those of Wiebe and Gaddy
salting coefficient. Helgeson (1969) has determined
(1939, 1940) below 100 8C and Takenouchi and
r CO2 (T) by using experimental data.
Kennedy (1964) above 110 8C, from 50 to 350 8C and
Finally, combining Eqs. (1), (2) and (5) leads to the
from 100 to 3500 bar. Natural porewaters within
following formulation for CO2 solubility in a saline
potential CO2 host formations are likely to be in the
solution:
range of 40–150 8C with NaCl concentrations up to
uCO2 xCO2 P and exceeding 3 molal. CO2 solubility has been tested
mCO2 ¼  lh i
H ðT ; P sat Þexp V̄V CO sat
over a wide range of NaCl concentrations from 25 8C
cCO2 KCO 2 H2 O RT
2
P
PH2 O to 250 8C. The most extensive studies are those of
Drummond (1981), Nighswander et al. (1989),
ð7Þ
Nicolaisen (1994) and Rumpf et al. (1994). However,
Although we use this simple semi-empirical model we note the general lack of experimental measure-
to describe the CO2 solubilities at conditions relevant ments of CO2 solubility in NaCl solutions at high
to the CO2 storage in the Northern Sea, several other pressures (100bPb300 bar), moderate temperatures
more sophisticated models on CO2 solubility have (40bTb120 8C) and in high ionic strength (up to 3 m)
been published. Based on two modified EOS of Peng– solutions. CO2 solubility data in aqueous solutions
Robinson for each of the phases (liquid and vapour), containing salts other than NaCl at elevated pressures
Soreide and Whitson (1992) presented a model (SW are very limited. The most significant investigations
model) to predict phase equilibrium of CO2 and water/ are listed in Table 4. The most extensive studies of
NaCl solution mixtures. More recently, Duan and Sun experimental CO2 solubility data in aqueous CaCl2
(2003) presented an empirical model (Duan model) solutions are those of Prutton and Savage (1945),
 S. Portier, C. Rochelle / Chemical Geology 217 (2005) 187–199 193

Table 3
Temperature, pressure and salinity conditions range of experimental data on the carbon dioxide solubility in pure water and in NaCl solutions
Aqueous solution Salt molality (mol kg
1 H2O) Temperature (8C) Pressure (bar) Authors
Pure water 0 12–100 30–800 Wiebe and Gaddy (1939, 1940)
Pure water 0 101–120 30–700 Prutton and Savage (1945)
Pure water 0 71 100–1000 Dhima et al., (1998)
Pure water 0 15–260 6.9–202.7 Gillepsie and Wilson (1982)
Pure water 0 110–260 100–700 Takenouchi and Kennedy (1964)
Pure water 0 250–350 200–3500 Toedheide and Franck (1963)
Pure water 0 100–200 3–80 Müller et al. (1988)
Pure water 0 200–330 98–490 Malinin (1959)
NaCl solution 0–2 172–330 16–93 Ellis and Golding (1963)
NaCl solution 0–6 50–400 30–266 Drummond (1981)
NaCl solution 0–6 40–160 1–100 Rumpf et al. (1994)
NaCl solution 0–3 0–25 1 Harned and Davis (1943)
NaCl solution 0–6 25–150 48 Malinin and Kurovskaya (1975)
NaCl solution 0–4 25–75 48 Malinin and Savalyeva (1972)
NaCl solution 0.1–4 0–40 1 Markham and Kobe (1941)
NaCl solution 0–6 40–160 2–96 Nicolaisen (1994)
NaCl solution 0–0.2 80–200 1–100 Nighswander et al. (1989)
NaCl solution 0–3 25 1 Onda et al. (1970)
NaCl solution 1–4.3 135–527 30–2800 Gehrig (1980)
NaCl solution 0–4.3 150–250 100–1400 Takenouchi and Kennedy (1965)
NaCl solution 0.4–5.1 15–35 1 Yasunishi and Yoshida (1979)

from 75 to 120 8C and from 15 to 885 bar with data being above seawater salinity (e.g. Ellis and
molality up to 3.9 m. For seawater salinity (approx- Golding, 1963; Takenouchi and Kennedy, 1965).
imately the same as the Utsira porewater), data are However, a somewhat limited indirect comparison is
available at lower temperatures more applicable to possible. Fig. 2 shows a comparison between liter-
oceanic disposal of liquid CO2 (e.g. Teng and ature data for CO2 solubility in seawater (having
Yamasaki, 1998), and at atmospheric pressure (Mur- approximately the same salinity as the synthetic Utsira
ray and Riley, 1971). Higher temperature data are porewater) and the experimental result at 18 8C in
limited to relatively simple solutions, with most of the SUP. We can see the accuracy between new exper-

Table 4
Temperature, pressure and salinity conditions range of experimental data on the carbon dioxide solubility in simple aqueous electrolyte solutions
others than aqueous NaCl solutions
Salt Salt molality (mol kg
1 H2O) Temperature (8C) Pressure (bar) Authors
Na2SO4 0.2–2.0 25–40 1 Markham and Kobe (1941)
1.0–2.0 40–160 0.1–98 Rumpf and Maurer (1993)
0.2–2.3 15–35 1 Yasunishi and Yoshida (1979)
CaCl2 0.0–7.0 100–150 48 Malinin and Kurovskaya (1975)
0.4–3.4 25–75 48 Malinin and Savalyeva (1972)
0.0–3.9 75–120 15–885 Prutton and Savage (1945)
0.2–4.6 25–35 1 Yasunishi and Yoshida (1979)
AlCl3 0.4–2.6 25 1 Yasunishi and Yoshida (1979)
BaCl2 0.1–1.5 25 1 Yasunishi and Yoshida (1979)
K2SO4 0.1–0.4 25 1 Yasunishi and Yoshida (1979)
KCl 0.1–4.0 0–40 1 Markham and Kobe (1941)
0.4–4.1 25–35 1 Yasunishi and Yoshida (1979)
MgCl2 0.1–3.9 15–35 1 Yasunishi and Yoshida (1979)
MgSO4 0.1–2.3 25 1 Yasunishi and Yoshida (1979)
0.5–2.0 0–40 1 Markham and Kobe (1941)
194 S. Portier, C. Rochelle / Chemical Geology 217 (2005) 187–199

2 1.6
T = 20°C
1.8 salinity ~ seawater 1.4

CO2 solubility (mol kg-1 H2O)

CO2 solubility (mol kg-1 H2O)

1.2
1.6
1
1.4
0.8
1.2
0.6 Pure water
Diaphore model at 50°C
1 Teng & Yamasaki (1998) 0.4 Rumpf & al. (1994) at 50°C
Diaphore model at 150°C
Experimental data at 18°C Takenouchi & Kennedy (1964) at 150°C
0.8 This model (pure water) 0.2 Wiebe & Gaddy (1939) at 50°C
Drummond (1981) at 50°C
This model (m NaCl = 0.7) Malinin & Kurovskaya (1975) at 150°C
0.6 0
50 100 150 200 250 0 50 100 150 200 250 300
Pressure (bar) Pressure (bar)

Fig. 2. Liquid CO2 solubility in seawater as a function of pressure. Fig. 4. CO2 solubility as a function of pressure in pure water. Model
Model calculation compared to experimental data for 20 8C. calculations compared to experimental data for 50 and 150 8C.

imental data at 188C and those of Teng and Yamasaki experimental data reported in Table 2, the CO2
(1998; CO2 solubility at 18 8C is slightly higher than solubilities have been expressed in terms of molality
solubility at 20 8C in a solution of the same salinity). (mol kg
1 H2O) and these are plotted in Fig. 3. As
All these data show CO2 solubility to be less than for shown in Fig. 3, CO2 solubility in the synthetic Utsira
pure water under similar conditions. porewater (SUP) decreases with increasing temper-
The composition of various co-existing CO2-rich ature. from the comparison between the experimental
gas phases and aqueous solutions, at equilibrium, was data and calculated results at 80, 100 and 120 bar, it
calculated according to the thermodynamic model can be seen that CO2 solubility in SUP shows a slight
presented above. increase with increasing pressure (at 50 8C, CO2
An initial series of calculations is intended to show solubility is ~15% lower at 80 bar than at 120 bar). It
the accuracy of this model in reproducing CO2 may be concluded that most experimental data on the
solubility in a mixed electrolyte aqueous solution solubility of CO2 in SUP can be represented by the
(SUP). To compare the modelling results with model adequately, and are within or close to exper-
imental uncertainty (about 7%).

1.4 1.2
This model at 120 bar
1.3 Experimental data at 120 bar
This model at 100 bar 1
CO2 solubility (mol kg-1 H2O)
CO2 solubility (mol kg-1 H2O)

1.2 Experimental data at 100 bar

This model at 80 bar
Experimental data at 80 bar
1.1 0.8
1
0.6
0.9

0.8 Drummond (1981) at 150°C

0.4
Takenouchi & Kennedy (1965) at 150°C
0.7 This model at 150°C (m NaCl = 1)
Malinin & Salvalyeva (1972) at 50°C
0.2 This model at 50°C (m NaCl = 1)
0.6 This model at 50°C (m NaCl = 3)
Drummond (1981) at 50°C
0.5 This model at 150°C (m NaCl = 3)
salinity~seawater 0
0.4 0 50 100 150 200 250 300
30 40 50 60 70 80 90 Pressure (bar)
Temperature (°C)
Fig. 5. CO2 solubility as a function of pressure in NaCl solutions of
Fig. 3. CO2 solubility in SUP as a function of temperature 1 and 3 m. Model calculations compared to experimental data for 50
calculated for 80, 100 and 120 bar. Dots are experimental results. and 150 8C.
 S. Portier, C. Rochelle / Chemical Geology 217 (2005) 187–199 195

0.04
T = 25°C coefficient calculation at standard conditions. Another
0.035 P = 1 bar important aspect pointed out by the results of Fig. 6 is
CO2 solubility (mol kg-1 H2O)

the variability of CO2 solubility according to the

0.03
nature of the electrolyte. The solubility of CO2 in
0.025 CaCl2 or MgCl2 solutions shows a pattern similar to
0.02
that expected in NaCl solution—experimental meas-
urements showing elevated solubility relative to NaCl
0.015
This model (NaCl solutions) solutions of equivalent ionic strength. Both of these
Exp. NaCl (Markham & Kobe, 1943)
0.01 Exp. NaCl (Harned & Davis, 1941)
divalent salts appear to show the same increase in
Exp. NaCl (Yasunishi & Yoshida, 1979) solubility relative to NaCl solutions of equivalent
0.005 Exp. CaCl2 (Yasunishi & Yoshida, 1979)
Exp. MgCl2 (Yasunishi & Yoshida, 1979) ionic strength. Maybe, because Ca2+ and Mg2+ tend to
0 form ion pairs very easily with the anions present in
0 0.5 1 1.5 2 2.5 3
Ionic strength (mol kg-1 H2O) solution, the salting-out effect on the solubility of CO2
in CaCl2 or MgCl2 solutions is lower than in NaCl
Fig. 6. CO2 solubility as a function of ionic strength (both in mol solutions of equivalent ionic strength. Thus, a
kg
1 H2O). Model calculation compared to experimental data for calculation performed with equivalent-NaCl solutions
25 8C.
of the same ionic strength, shows that this approxi-
mated method systematically underestimates the CO2
Figs. 4 and 5 show the agreement between the solubility in aqueous CaCl2 solutions (Figs. 6 and 7).
literature data and our model. In order to evaluate the The relative error induced by NaCl approximation
accuracy of the model in reproducing CO2 solubility (Fig. 6) becomes significantly high for higher ionic
in pure water and NaCl solutions at high temperature, strength (above 5% and up to 20%). On the basis of
we performed a series of calculation at 150 8C for the experimental data from Prutton and Savage
pressures ranging from 1 to 300 bar. The calculation (1945), we carried out other CO2 solubility calcu-
results presented on Figs. 4 and 5 can be favourably lations in 1.01 m CaCl2 solutions for moderate
compared to the measurements carried out by Taken- temperature (101 8C). The results presented on Fig.
ouchi and Kennedy (1964, 1965). As shown in Fig. 4, 7 globally confirm that the error induced by the
the solubility of CO2 in pure water increases with an equivalent-NaCl approximation has a relatively high
increase in pressure. It varies in a more complex way impact on the CO2 solubility value (less than 20%) for
with temperature. At lower pressures (b100 bar), CO2 the aqueous CaCl2 solution considered. Consequently,
solubility decreases with increasing temperature.
However, at higher pressures (N100 bar), the relation-
ship with increasing temperature changes over, such 1
1.01 m CaCl2
that at 250 bar CO2 solubility actually increases with 0.9
CO2 solubility (mol kg-1 H2O)

increasing temperature. For example, the range of 0.8

CO2 solubility in pure water is from approximately 0.7
0.75 mol kg
1 H2O at 50 8C and 50 bar to 0.6
approximately 1.15 mol kg
1 H2O at 150 8C and 0.5
200 bar. 0.4
The results in Fig. 5 indicate a systematic decrease
0.3
in the solubility of CO2 with increasing ionic strength.
0.2 This model at 101°C
For a 3 molal NaCl solution this equates to an
0.1 Prutton & Savage (1945) at 101°C
approximately twofold reduction relative to pure
water at 50 8C and 100 bar. Fig. 6 also shows a 0
0 50 100 150 200 250 300
satisfying agreement between calculated CO2 solubil- Pressure (bar)
ity and experimental measurements in NaCl for Fig. 7. CO2 solubility as a function of pressure in aqueous CaCl2
electrolyte concentrations as high as 3 molal at 1 bar solution of 1.01 m. Model calculations compared to experimental
and 25 8C, corroborating the validity of the activity data for 75.5 and 101 8C.
196 S. Portier, C. Rochelle / Chemical Geology 217 (2005) 187–199

these results show that the use of a simplified m NaCl, 0–300 8C, 1–300 bar). This range of
equivalent-NaCl approach may be acceptable only temperature, pressure and salinity is likely to encom-
for NaCl-type waters. pass a very large proportion of subsurface environ-
Fig. 8a and b show a good agreement between the ments that might be considered for the deep
results of the calculations performed with Duan underground storage of CO2.
model and the experimental measurements on the
whole pressure range. In contrast, an important
discrepancy between experimental and SW results 4. Geochemical implications
(20% at 150 8C and 200 bar in 1 molal NaCl
solution) is observed. The SW model overestimates Assuming non-reactive aquifer conditions, the
CO2 solubility in pure water and in NaCl solutions ultimate CO2 sequestration capacity in solution of
(by more than 20% at 50 8C in 2 molal NaCl an aquifer is defined as the total amount of CO2 that
solution) at high pressures (N100 bar). can theoretically dissolve to saturation in the
The results achieved up to now let us think that our formation water of this aquifer and be permanently
model notably improve CO2 solubility prediction—at trapped in a state of equilibrium (Bachu and Adams,
least for pure water and aqueous NaCl solutions (0–3 2003). To calculate the ultimate CO2 sequestration
capacity in solution, there is need to know the pore
1.6 volume of the aquifer, the formation water density
1.4
T = 50°C and the maximum CO2 content at saturation (func-
tions of temperature, pressure and salinity of the
CO2 solubility (mol kg-1 H2O)

1.2 formation water). The Utsira sand covers an area of

1 more than 26,000 km2 and ranges in depth from
pure water
about 550 to 1500 m (Chadwick et al., 2000). It
0.8
occupies two distinct depositional basins, which are
2 m NaCl
0.6 likely to be in poor hydraulic contact. The maximum
reservoir thickness is about 300 m and the estimated
0.4 This model
Duan model
pore volume of the reservoir (excluding stratigraph-
SW model
0.2 Drummond (1981) ically different, but possibly linked sand units) is
0
Wiebe & Gaddy (1939)
55.1010 m3. By using new data, our model, and
0 50 100 150 200 250 correlations given by Batzle and Wang (1992) to
Pressure (bar) calculate the Utsira porewater density, it is possible
1.6
T = 150°C
to estimate the ultimate CO2 sequestration capacity
1.4
pure water
in solution of the Utsira sand at relevant conditions.
CO2 solubility (mol kg-1 H2O)

1.2
For the following calculation, we assumed constant
in situ pressure and temperature conditions of 378C
1 and 100 bar—in situ temperature and pressure at the
0.8 injection point in the Utsira formation at Sleipner.
1 m NaCl Considering only the region of 26,000 km2 (Utsira
0.6
sand) where the injected CO2 would be a dense fluid
This model
0.4 Duan model (liquid or supercritical), the ultimate capacity of the
SW model
0.2 Takenouchi & Kennedy (1964) Utsira sand to sequester CO2 in solution in formation
Takenouchi & Kennedy (1965)
Drummond (1981) water is calculated to be 22 Gt CO2.
0 It is noteworthy that the Utsira Formation
0 50 100 150 200 250
Pressure (bar) (approximately 800 m deep) is at the shallower end
of the likely range of CO2 storage depths. Porewaters
Fig. 8. (a) CO2 solubility as a function of pressure calculated for 50
8C. SW and Duan models compared to this model. (b) CO2
within deeper aquifers or associated with hydro-
solubility as a function of pressure calculated for 150 8C. SW and carbon fields (Abbotts, 1991) could be significantly
Duan models compared to this model. more saline that the synthetic porewaters used in
 S. Portier, C. Rochelle / Chemical Geology 217 (2005) 187–199 197

these experiments. The effect of higher salinity I Ionic strength of the aqueous solution (mol
would be to lower the concentration of dissolved kg
1 H2O)
CO2 in solution and hence possibly reduce the
amount of CO2–rock reaction. Conversely however, Supercripts
deeper aquifers may be warmer, which would tend to v Vapour
increase the rate of mineral reactions which might aq Aqueous solution
dconsumeT CO2. The overall effect of competing l Infinite dilution state
processes such as these will be site-specific there-
fore, and this should be considered during initial
investigations of potential storage sites. This work Acknowledgements
highlights the need for long-term field and laboratory
experiments to test geochemical aspects of CO2 This work was supported by the EC funding (plus
sequestration. EC project code). The authors would like to acknowl-
edge Etienne Brosse (IFP). This paper is published
with the permission of the Director of the British
5. Conclusion Geological Survey. [EO]

This work provides measurements of the solubil-

ity of CO2 within the Utsira formation at Sleipner. It
References
allows comparison of the results with extrapolations
from previous experimental data. Finally, it validates Abbotts, I.L., 1991. United Kingdom oil and gas field, 25 years
a modelling approach for use in a limited range of commemorative volume. Geol. Soc. Lond., 14.
temperature, pressure and salinity. The model can Bachu, S., Adams, J.J., 2003. Sequestration of CO2 in geological
predict CO2 solubility in brines dominated by NaCl media in response to climate change: capacity of deep saline
for temperatures from 0 to 300 8C and for pressures aquifers to sequester CO2 in solution. Energy Convers. Manag.
44, 3151 – 3175.
from 1
300 bar, and for ionic strengths up to 3 Bachu, S., Gunter, W.D., Perkins, E.H., 1994. Aquifer disposal of
molal. This model is able to predict CO2 solubility CO2: hydrodynamic and mineral trapping. Energy Convers.
in more complex brines such as saline Utsira Manag. 35, 269 – 279.
porewater with good accuracy. These results will Backlid, A., KorbbL, R., Orren, G., 1996. Sleipner vest CO2
contribute to further our understanding of the disposal, CO2 injection into a shallow underground aquifer. Pap.
Soc. Pet. Eng., 36600.
reactive behavior of the supercritical CO2 under Batzle, M., Wang, Z., 1992. Seismic properties of pore fluids.
physical–chemical conditions relevant to geologic Geophysics 57 (11), 1396 – 1408.
storage and sequestration. Chadwick, R.A., Holloway, S., Kirby, G.A., Gregersen, U.,
Johannessen, P.N., 2000. The Utsira sand, Central North
Sea—an assessment of its potential for regional CO2 disposal.
5th International Conference on Greenhouse Gas Control
Notation Technologies, Cairns (Australia), August.
T Absolute temperature (Kelvin) Dhima, A., de Hemetime, J.C., Moracchini, G., 1998. Solubility of
sat
P Total pressure (bar); =P CO2+P H 2O
light hydrocarbons and their mixtures in pure water under high
x Mole fraction in vapour phase pressure. Fluid Phase Equilib. 145, 129 – 150.
Drummond, S.E., 1981. Boiling and mixing of hydrothermal fluids:
R Universal gas constant (=0.08314467 bar
chemical effects on mineral precipitation. Ph.D. dissertation, the
mol
1 K
1) Pennsylvannia States University.
m Molality in the aqueous phase (mol kg
1 H2O) Duan, Z., Sun, R., 2003. An improved model calculating CO2
u Fugacity coefficient in the vapour phase solubility in pure water and aqueous NaCl solutions from
c Activity coefficient in the aqueous phase 273 to 5333 K and from 0 to 2000 bar. Chem. Geol. 193,
f Fugacity (bar) 257 – 271.
Duan, Z., Moller, N., Weare, J.H., 1992. An equation of state
V̄ Molar volume (cm3 mol
1) for the CH4–CO2–H2O system: I. Pure systems for 0 to
K H
Henry’s constant of dissolved gas (bar mol
1 1000 8C and 0 to 8000 bar. Geochim. Cosmochim. Acta 56,
kg H2O) 2605 – 2617.
198 S. Portier, C. Rochelle / Chemical Geology 217 (2005) 187–199

Ellis, A.J., Golding, R.M., 1963. The solubility of carbon dioxide Murray, C.N., Riley, J.P., 1971. The solubility of gases in distilled
above 100 8C in water and in sodium chloride solutions. J. Am. water and seawater: IV. Carbon dioxide. Deep-Sea Res. 18,
Chem. Soc. 63, 449 – 454. 533 – 541.
Gehrig, M., 1980. Phasengleichgewichte und PVT-daten ternarer Nicolaisen, H., 1994. Phase equilibria in aqueous electrolyte
mischungen aus wasser, kohlendioxid und natriumchlorid solutions, Ph.D. dissertation, Technical University of Denmark
bis 3 kbar and 550 8C. PhD thesis, Univ. Karlsruhe, Lyngby.
Freiburg. Nighswander, J.A., Kalogerakis, N., Mehrotra, A.K., 1989.
Gillepsie, C., Wilson, G.M., 1982. Vapor–liquid and liquid–liquid Solubilities of carbon dioxide in water and 1 wt.% NaCl
equilibria: water–carbon dioxide. Rapport GPA, RR48, Provo, solution at pressures up to 10 MPa and temperatures from 80
UT. to 200 8C. J. Chem. Eng. Data 34, 355 – 360.
Gregersen, U., Johannessen, P.N., Mbller, J.J., Kristensen, L., Omerod, W.G., 1994. IEA Greenhouse Gas R&D Programme,
Christensen, N.P., et al., 1998. Saline Aquifer CO2 Storage Carbon Dioxide Disposal from Power Stations, IEA/GHG/
(S.A.C.S.) Phase Zero 1998. SR3.
Harned, H.S., Davis, R., 1943. The ionization constant of carbonic Onda, K., Sada, E., Kobayashi, T., Kito, S., Ito, K., 1970. Salting-out
acid in water and the solubility of carbon dioxide in water and parameters of gas solubility in aqueous salt solution. J. Chem.
aqueous salt solutions from 0 8C to 50 8C. J. Am. Chem. Soc. Eng. Jpn. 3, 18 – 24.
65, 2030 – 2037. Peng, D.Y., Robinson, D.B., 1976. A new two-constant equation of
Helgeson, H.C., 1969. Thermodynamics of hydrothermal systems state. Ind. Eng. Chem. Fundam. 15, 59 – 64.
at elevated temperatures and pressures. Am. J. Sci. 267, Peng, D.Y., Robinson, D.B., 1980. Two- and three-phase equili-
729 – 804. brium calculations for coal gasification and related processes.
Helgeson, H.C., Kirkham, D.H., Flowers, G.C., 1981. Theoretical Thermodynamics of Aqueous Systems in the Industrial Appli-
prediction of the thermodynamic behavior of aqueous electro- cations. ACS Symp. Ser. 393, 393 – 414.
lytes at high pressures and temperatures: IV. Calculation of Pitzer, K.S., 1973. Thermodynamics of electrolytes: I. Theoretical
activity coefficients, osmotic coefficients, and apparent molal basis and general equations. J. Phys. Chem. 77, 268 – 277.
and standard and relative partial molal properties to 600 8C and Prausnitz, J.M., Lichtenthaler, R.M., de Azevedo, E.G., 1986.
5 kb. Am. J. Sci. 281, 1249 – 1516. Molecular thermodynamics of fluid-phase equilibria, 2nd ed.
Holloway, S., Heederik, J.P., Van der Meer, L.G.H., Czerni- Prentice Hall, Upper Saddle River, NJ. 860 pp.
chowski-Lauriol, I., Harrison, R., Lindeberg, E., et al., 1996. Pruess, K., Garcia, J., 2002. Multiphase flow dynamics during CO2
The underground disposal of Carbon Dioxide—Summary disposal into saline aquifers. Environ. Geol. 42, 282 – 295.
Report. British Geological Survey Report for JOULE II Prutton, C.F., Savage, R.L., 1945. The solubility of carbon dioxide
project CT92-0031. in calcium chloride–water solutions at 75, 100, 120 8C and high
Houghton, J.T., Meira Filho, L.G., Callander, B.A., Harris, N., pressures. J. Am. Chem. Soc. 67, 1550 – 1554.
Kattenberg, A., Maskell, K., 1996. Climate Change. The Rochelle, C.A., Moore, Y.A., 2002. The solubility of CO2 into pure
Science of Climate Change. Cambridge University Press. water and synthetic Utsira porewater. British Geological Survey
IPCC, 2001. IPCC Third Assessment Report: Climate Change Commissioned Internal Report, CR/02/052, 28 pp.
2001. Rumpf, B., Maurer, G., 1993. An experimental and theoretical
Krichevsky, I.R., Kasarnovsky, J.S., 1935. Thermodynamic calcu- investigation on the solubility of carbon dioxide in aqueous
lations of solubilities of nitrogen and hydrogen in water at high solutions of strong electrolytes. Ber. Bunsenges. Phys. Chem. 1,
pressures. J. Am. Chem. Soc. 93, 1857 – 1862. 85 – 97.
Lyckman, E.W., Eckert, C.A., Prausnitz, J.M., 1965. Generalized Rumpf, B., Nicolaisen, C., Öcal, C., Maurer, G., 1994. Solubility of
reference fugacities for phase equilibrium thermodynamics. carbon dioxide in aqueous solutions of sodium chloride:
Chem. Eng. Sci. 20, 685 – 691. experimental results and correlation. Ber. Bunsenges. Phys.
Malinin, S.D., 1959. The system water–carbon dioxide at high Chem. 1, 431 – 447.
temperatures and pressures. Geokhimiya 3, 292 – 306. Saul, A., Wagner, W., 1987. International equations for the
Malinin, S.D., Kurovskaya, N.A., 1975. The solubility of CO2 in saturation properties of ordinary water substance. J. Phys.
chloride solutions at elevated temperatures and CO2 pressures. Chem. Ref. Data 16, 893 – 901.
Geochem. Int. 12, 199 – 201. Soreide, I., Whitson, C., 1992. Peng–Robinson predictions for
Malinin, S.D., Savalyeva, N.I., 1972. The solubility of CO2 in NaCl hydrocarbons, CO2, N2, H2S with pure water and NaCl brine.
and CaCl2 solutions at 25, 50 and 75 8C under elevated CO2 Fluid Phase Equilib. 77, 217.
pressures. Geokhimiya 6, 643 – 653. Takenouchi, S., Kennedy, G.C., 1964. The binary system H2O–
Markham, A.E., Kobe, K.A., 1941. The solubility of carbon dioxide CO2 at high temperatures and pressures. Am. J. Sci. 262,
and nitrous oxide in aqueous salt solutions. J. Am. Chem. Soc. 1055 – 1074.
63, 449 – 454. Takenouchi, S., Kennedy, G.C., 1965. The solubility of carbon
Mqller, G.E., et al., 1988. Das Dampf-Flqssigkeitsgleichgewicht des dioxide in NaCl solutions at high temperatures and pressures.
tern7ren Systems Ammoniak–Kohlendioxid–Wasser bei hohen Am. J. Sci. 263, 445 – 454.
Wassergehalten im Bereich zwischen 373 und 473 Kelvin. Ber. Teng, H., Yamasaki, A., 1998. Solubility of liquid CO2 in synthetic
Bunsenges. Phys. Chem. 92, 148 – 160. sea water at temperatures from 298 K to 293 K and pressures
 S. Portier, C. Rochelle / Chemical Geology 217 (2005) 187–199 199

from 6.44 MPa to 29.49 MPa, and densities of the correspon- Wiebe, R., Gaddy, V.L., 1940. The solubility of carbon dioxide in
dingaquous solutions. J. Chem. Eng. Data 43, 2 – 5. water at various temperatures from 12 to 408 and at pressures to
Todheide, K., Franck, E.U., 1963. Das Zweiphasengebiet und die 500 atm. J. Am. Chem. Soc. 62, 815 – 817.
kritische kurve im system kohlendioxid-wasser bis zu drucken Yasunishi, Y., Yoshida, F., 1979. Solubility of carbon dioxide in
von 3500 bar. Z. Phys. Chem. 37, 387 – 401. aqueous electrolyte solutions. J. Chem. Eng. Data 24, 11 – 14.
Wiebe, R., Gaddy, V.L., 1939. The solubility in water of carbon
dioxide at 508, 758 and 1008 at pressures to 700 atm. J. Am.
Chem. Soc. 61, 315 – 318.