You are on page 1of 11

Chemical Geology 424 (2016) 1–11

Contents lists available at ScienceDirect

Chemical Geology

journal homepage: www.elsevier.com/locate/chemgeo

Thermodynamic description of H2S–H2O–NaCl solutions at temperatures


to 573 K and pressures to 40 MPa
Nikolay N. Akinfiev a,⁎, Vladimir Majer b,c, Yuri V. Shvarov d
a
Institute of Geology of Ore Deposits, Petrography, Mineralogy and Geochemistry, Russian Academy of Sciences, Moscow 119017, Russia
b
Department of Chemistry, Technical University of Liberec, 46117 Liberec, Czech Republic,
c
Institute of Chemistry of Clermont-Ferrand, CNRS/Blaise Pascal University, 63170 Aubière, France
d
Geological Department, Lomonosov Moscow State University, Moscow 119992, Russia

a r t i c l e i n f o a b s t r a c t

Article history: Reliable experimental results were selected from the literature (using over 700 data) to develop a thermody-
Received 9 October 2015 namic model for calculating the solubility of hydrogen sulfide (H2S) in pure water and in aqueous NaCl solu-
Received in revised form 8 January 2016 tions between 283 and 573 K, 0.1–40 MPa and ms 0–6 mol·kg−1. Thermodynamic properties of the pure
Accepted 9 January 2016
components were calculated using highly accurate multiparametric equations of state for H2S (Lemmon and
Available online 14 January 2016
Span, 2006) and for H2O (Wagner and Pruss, 2002). Thermodynamic properties of H2S(aq) at infinite dilution
Keywords:
were based on the Henry's law constants generated from the SOCW model (Sedlbauer et al., 2000) and reported
Hydrogen sulfide by Majer et al. (2008). The determined activity coefficients of H2S in pure water and in NaCl solutions were treat-
Sodium chloride ed using the Pitzer interaction model. The Pitzer parameters for H2S in binary and ternary solutions were newly
Solubility determined while those for NaCl(aq) in the H2S-free system were adopted from the review of Archer (1992). The
Aqueous solution experimental solubilities selected for correlation are reproduced by the model with mean relative deviations of
Pitzer interaction model 5.2% and 6.1% for the H2S–H2O and for H2S–H2O–NaCl systems, respectively. These values are comparable to
the experimental uncertainty of the solubility data. The new model allows a thermodynamically consistent de-
scription of numerous other properties of the liquid phase in the ternary H2S–H2O–NaCl system, including the
activity coefficients of H2S and NaCl, the osmotic coefficients, the Setchenow constants, and the molar volume
and density of the bulk liquid. These properties can be calculated for any H2S and NaCl concentrations up to halite
saturation. The model is available as a computer code that is freely distributed.
© 2016 Published by Elsevier B.V.

1. Introduction publications of Akinfiev and Diamond (2010); Mao et al. (2013) and
of Springer et al. (2015) are representative examples that review the
Hydrogen sulfide is an important gas in natural geochemical sys- abundant sources of experimental data and modeling approaches for
tems, as exemplified by its wide occurrence in fluid inclusions. It is a CO2–H2O–NaCl. Less information is, however, available on dissolution
major component of sour reservoir fluids with concentrations attaining of hydrogen sulfide in water and in aqueous electrolytes (see Springer
several tens of percent at certain sites. Hydrogen sulfide is also present et al., 2015 for an overview). The number of measurements is much
with carbon dioxide in flue gas generated by power plants and by coal more limited due to difficulties operating with highly toxic H2S under
gasification processes. Efforts are being made to reduce the concentra- high pressures and at elevated temperatures. Prior to the publications
tion of these greenhouse gases in the atmosphere, and sequestration of Koschel et al. (2007, 2013), no reliable experimental data existed
in geological brine formations is a promising option. Thermodynamic for pressures above 15 MPa, as discussed in more detail below. Never-
data are therefore needed to model systems such as CO2–H2O–NaCl theless, several models allowing correlation and prediction of H2S solu-
and H2S–H2O–NaCl over a wide range of conditions, allowing PVT and bility in pure water and in NaCl solutions, particularly at moderately
phase equilibrium properties to be calculated for acidic gases dissolved elevated T–P conditions, have been published, and additional effort is
in deep saline aquifers. The experimental data characterizing dissolu- needed particularly for estimations over a wide range of pressures and
tion of carbon dioxide in aqueous solutions at geological conditions ionic strengths at temperatures above 400 K.
are numerous and different models permitting their correlation and ex- Barta and Bradley (1985) employed Pitzer's (1973) approach to pre-
trapolation beyond experimental conditions have been developed. The dict binary and ternary close-range interaction parameters to model gas
solubilities in brines. However, their model is restricted to H2S pressures
below 3 MPa, and the fitting procedure was mostly based on the exper-
⁎ Corresponding author. imental data of Drummond (1981), which are considered in the present
E-mail address: akinfiev@igem.ru (N.N. Akinfiev). study to be unreliable. Carroll and Mather (1989) presented a model

http://dx.doi.org/10.1016/j.chemgeo.2016.01.006
0009-2541/© 2016 Published by Elsevier B.V.
2 N.N. Akinfiev et al. / Chemical Geology 424 (2016) 1–11

for H2S solubility in pure water up to 513 K based on the cubic Peng– approach we use for this latter phase an asymmetric standard state con-
Robinson–Stryjek–Vera equation of state (EoS) and various types of vention, i.e. the activity scale of Raoult for the water solvent (xw → 1,
R
mixing rules. Because of the inherent limitations of cubic EoS, this γw →1) and the activity scale of Henry for H2S solute (xs → 0, γsH →1).
model underestimates H2S solubility at low pressures and overesti- Then we can write for water:
mates it at temperatures above 450 K. After having published a virial
  y Pφv
expansion-like EoS for H2S in NaCl solutions (Duan et al., 1996), a μ vw ðT; P; yw Þ ¼ μ ig 0
þ RT ln w 0 w ¼ μ •w ðT; P Þ þ RT ln xw γ Rw
w T; P
more recent model of Duan et al. (2007) for solubility of H2S in pure P
water and brines uses Pitzer's approach and presents an important ad- ¼ μ lw ðT; P; x w Þ; ð2Þ
vancement. However, for pure H2S they employed the EoS of Duan

et al. (1992), which is less accurate than the multiparametric EoS where μ ig 0
w(T, P ) and μ w(T,p) are the molar Gibbs free energies of water

of Lemmon and Span (2006), and the model treats the data only up to at temperature T in an ideal gas state at a reference pressure P 0 = 1
500 K. Dubessy et al. (2005) presented an asymmetric model for bar = 0.1 MPa and as pure liquid at system pressure P, respectively.
liquid-vapor equilibria in the H2S–H2O–NaCl system up to 543 K and Similarly, for hydrogen sulfide the following equation holds:
30 MPa. To compute the mixing parameters in the equation expressing   y Pφv
the excess Gibbs energy of the H2S–H2O binary, those authors used a μ vs ðT; P; ys Þ ¼ μ ig
s T; P
0
þ RT ln s 0 s ¼ μ ∞s ðT; pÞ þ RT ln xs γ H
s
Margules–Redlich–Kister formulation, while they took account of the P
¼ μ lg ðT; P; xs Þ; ð3Þ
salting-out effect by an “effective” Henry constant of H2S estimated by
an iterative fitting procedure using a Setchenow-like equation. The ∞
where μ ig 0
s (T, P ) has an analogous meaning as for water and μ s stands
Henry's law constant of the H2S binary was calculated from the model
for the partial Gibbs free energy of the solute in the standard state of in-
of Plyasunov et al. (2000). Finally, publications have appeared recently
finite dilution.
presenting the experimental data (Savary et al., 2012) or models
The Henry's law constant is a key parameter in calculations of solu-
(Zirrahi et al., 2012; Springer et al., 2015) on simultaneous dissolution
bility; in its rigorous definition (Majer et al., 2008) it is a function of
of CO2 and H2S in brines. The robust approach of Springer et al. (2015)
two independent variables (T and P):
uses a framework combining the Helgeson–Kirkham–Flowers (HKF)
EoS (Johnson et al., 1992) for standard state properties with the Pitzer  
fs
model for the excess Gibbs energy in an algorithm for solving phase kH ðT; P Þ ¼ lim s
; ð4Þ
xs →0 xs
and chemical equilibria in a multiphase system.
All of the above approaches have their strengths and weaknesses where fs is the fugacity of the solute in a mixture of mole fraction xs. It
where the better theoretically founded models with a limited number can be easily shown that kH is directly related to the Gibbs free energy
of adjustable parameters usually offer more versatility and reliability of hydration, corresponding to the difference between infinite dilution
in extrapolation, but often at the expense of precision in fitting the ex- and ideal gas standard states:
perimental data. In the present study we propose a semi-theoretical
   
γ–ϕ type model to describe the solubility of H2S in water and in NaCl so- kH ðT; P Þ
lutions, broadly following the concept developed in earlier publications RT ln 0
¼ μ ∞s ðT; P Þ−μ ig
s T; P
0
¼ ΔGhyd ðT; P Þ: ð5Þ
P
on CO2 solubility (Diamond and Akinfiev, 2003; Akinfiev and Diamond,
2010). First, emphasis is put on precise fitting of the experimental data The combination of Eqs. (3) and (5) leads to the expression
that we consider to be reliable. In particular, the recently published data
of Koschel et al. (2007, 2013) are included, presenting a valuable contri- ys Pφs
xs ¼ ; ð6Þ
bution in the range of high pressures. Second, the new model uses ro- kH γ H
s
bust multiparameter equations of state for pure H2S (Lemmon and
Span, 2006) and H2O (Wagner and Pruss, 2002), allowing highly accu- where xS and yS are the mole fractions of H2S in the water-rich and H2S-
rate description of their PVT properties. The new model also uses stan- rich phases at equilibrium, respectively. No Poynting pressure correc-
dard state properties of H2S(aq) determined earlier by simultaneous tion is required in this arrangement for kH because the Henry's law con-
fitting of various infinite-dilution properties (Sedlbauer et al., 2000; stants were generated from the SOCW model (Sedlbauer et al., 2000)
Majer et al., 2008). Third, the obtained activity coefficients of H2S in allowing calculation of the Gibbs free energy of hydration as a function
water and in NaCl solutions are then fitted using the approach of of both temperature and pressure (Majer et al., 2008). The parameters
Pitzer (1991). In addition, the results of this study are incorporated in of this model for H2S were determined by a simultaneous correlation
a broader model where the excess properties for the H2O–NaCl subsys- using several types of data: i) kH values along the saturation line of
tem are calculated as recommended by Archer (1992). This then per- water up to temperature of 533 K (27 data points) recommended by
mits calculation of a wider range of thermodynamic properties in the Fernandez-Prini et al. (2003) based on the phase equilibria measure-
ternary H2S–H2O–NaCl system by a computer code that is available on- ments of Selleck et al. (1952); Lee and Mather (1977); Gillespie and
line at http://www.geol.msu.ru/deps/geochems/soft/index_e.html. Wilson (1982) and Carroll and Mather (1989); ii) the temperature de-
rivative properties obtained by combination of data from calorimetric
2. Basic relations measurements and calculated properties of the ideal gas (one data
point for enthalpy of hydration at 298 K, and 0.1 MPa, Cox et al., 1989,
To express equilibria in a fluid system consisting of H2S-rich and and 9 values for heat capacity of hydration up to 623 K at 28 MPa,
H2O-rich phases, denoted v and l, respectively, the starting point is the Barbero et al., 1982; Hnědkovsky and Wood, 1997); iii) the partial
equality of chemical potentials μi corresponding to the partial Gibbs molar volumes at infinite dilution obtained from volumetric experi-
free energies of individual species, as follows: ments up to 623 K and 35 MPa (19 data points, Barbero et al., 1982;
Hnědkovsky et al., 1996). We are therefore confident that the model
μ vi ðT; P; yi Þ ¼ μ li ðT; P; xi Þ; ð1Þ is well conditioned and able to generate reliable kH values at tempera-
tures up to 623 K and pressures from the saturation line of water up
where yi and xi are mole fractions of a component i in the coexisting to 50 MPa.
phases. In the so-called activity – fugacity concept (γ-φ), the non- As mentioned above, most data sources for H2S–H2O phase equilib-
ideality of the H2S-rich phase is described by fugacity coefficient φi ria focus on composition of the dense water-rich phase while the
while that of the H2O-rich phase by activity coefficient γi. In our coexisting phase dominated usually by hydrogen sulfide is rarely
N.N. Akinfiev et al. / Chemical Geology 424 (2016) 1–11 3

analyzed. To circumvent this lack of experimental yS values we have directly exploitable for the correlation by the Pitzer interaction model:
used the same approximation as Duan et al. (2007), putting 10 of these studies report H2S solubilities in water (see Table 1), while
5 studies present experimental values of H2S solubility in aqueous
P−P w• NaCl solutions (see Table 2). It is apparent from the Tables that four
ys ¼ ; ð7Þ
P data sources present both binary and ternary data. First, all 606 experi-
• mental data points for binary and 623 for ternary systems were corre-
where the vapor pressure of water, PW , was calculated from the equa-
lated. Subsequently, however, 123 data points for binary and 375
tion of Wagner and Pruss (2002) recommended by the International
points for ternary systems were discarded from the final correlation
Association for Properties of Water and Steam (IAPWS).
procedure for the following reasons:
The fugacity coefficient of pure hydrogen sulfide, φs, was calculated
from the fundamental equation explicit in the Helmholz free energy
i) From Drummond's (1981) solubilities, all the data points were
(Lemmon and Span, 2006). Despite its relative complexity, this is the
first considered in preliminary fitting up to a temperature of
most reliable formulation available to date (accurate within 0.05%
603 K (110 for binary and 295 for the ternary system as indicated
over the P–T range of interest, including the near-critical region of
in Tables 1 and 2). Drummond used an isoplethic, quasi-isochoric
H2S). Using the above input we introduced the “ideal solubility” x∞
s
reversal method to study solubility of H2S in aqueous solutions of
ys Pφs 0, 1, 2, 4, and 6 NaCl molalities. Thus, temperature and total pres-
x∞s ¼ ; ð8Þ sure in the autoclave were measured at known total volume.
kH
Then to estimate H2S partial pressure in the vapor phase as
which is equal to xs in the limit of infinite dilution of H2S where γH
s is by
well as its molality in the liquid, a set of simplifications was
definition unity. Then it follows from Eq. (6): used. It follows that the experimental method employed cannot
be considered as direct. In fact, the Henry's constants reported
x∞s for H2S at temperatures above 473 K derived from the solubility
γH
s ¼ : ð9Þ
xs data are systematically higher than those given by Suleimenov
and Krupp (1994), which are in accordance with the tempera-
It is a usual practice in solution chemistry to express the composition ture dependence of kH used in the present study (see above). It
of the water-rich phase in terms of molality ms = xs/(xwMw), where is thus clear that the high-temperature data suffer from a strong
Mw = 18.0152 10−3 kg·mol−1 is the molar mass of water. When intro- systematic error. Drummond performed measurements during
ducing ms as the concentration variable into equilibrium Eq. (3) one ob- heating steps up to 673 K and then during subsequent cooling.
tains the equality: The heating steps involve H2S exsolution from supersaturated
  m solutions, whereas the cooling steps involve H2S dissolution
μ ∞s ðT; pÞ þ RT ln xs γH ∞;m 0
s ¼ μ s ðT; pÞ þ RT ln ms =m γ s ; ð10Þ into undersaturated solutions. The differences in H2S solubilities
between the heating and cooling modes especially at low tem-
where μ ∞,m
s (T, p) is the standard state potential of solute referenced to peratures are large, up to over 10% relative. This means that at
an ideal hypothetical solution of concentration m0 = 1 mol kg−1 and low temperatures, phase equilibrium in the system might not
γsm is the activity coefficient corresponding to the molality concentra- have been achieved. It is not obvious to what extent the low-
tion scale. In the limit of infinite dilution both γsH and γsm are unity, temperature data are affected. Moreover, Drummond noticed
therefore formation of pyrrhotite and H2, indicating corrosion of the
stainless-steel experimental autoclave. In addition, we encoun-
μ ∞;m ∞ 0
s ðT; pÞ ¼ μ s ðT; pÞ þ RT ln M w m : ð11Þ tered problems in reproducing the reported partial pressures
of H2O using the accurate model of Archer (1992). Finally, the
Then it follows from Eqs. (6) and (10): data of Drummond (1981), when screened during correlation,
showed in general poor consistency with solubilities from other
xs γ H m∞ sources at conditions where a comparison was possible. For
γm
s ¼
s
¼ s ; ð12Þ
ms Mw ms these reasons we have discarded both binary and ternary data
sets from the final correlation.
where m∞s stands for ii) The data points of Barrett et al. (1988) at temperatures above
353 K and pressure of 0.1 MPa (7 for binary and 78 for ternary
ys Pφs
m∞s ¼ ; ð13Þ
kH M w
Table 1
which corresponds to “ideal solubility” in terms of the molality concen- Review of representative data sources reporting solubility of H2S in water reaching above
tration scale. ambient conditions.
The same approach is basically valid for the ternary H2O–H2S–NaCl
Temperature Pressure Number of Reference
system, thus for the H2S activity coefficient in the aqueous phase we
range range experimental
can derive by an analogous procedure T (K) P (MPa) points1

m∞s 311–444 0.7–12 37 (1) Selleck et al. (1952)


γ m;t
s ¼ ; ð14Þ 433–603 0.8–2.12 14 (2) Kozintseva (1965)
mts 283–453 0.2–6.7 264 (0) Lee and Mather (1977)
304–603 0.7–19 110 (110) Drummond (1981)
where mts stands for H2S molality in the ternary system. 311–589 0.3–21 43 (1) Gillespie and Wilson (1982)
297–368 0.1 39 (7) Barrett et al. (1988)3
294–594 0.3–14 47 (2) Suleimenov and Krupp (1994)
3. Published experimental data
313 0.5–2.5 9 (0) Kuranov et al. (1996)
298–338 0.50–4.0 31 (0) Chapoy et al. (2005)
The most complete overview of the literature sources with phase 323–393 1.7–31 12 (0) Koschel et al. (2007)
equilibrium data for H2S–H2O and/or for H2S–H2O–NaCl systems is 1
Number of data points discarded from the final correlation is given in parenthesis.
given by Springer et al. (2015). For the T–P region of our interest, 11 2
Partial pressures of H2S.
studies were selected for consideration presenting the solubility data 3
Considered in the final correlation up to temperature of 354 K only.
4 N.N. Akinfiev et al. / Chemical Geology 424 (2016) 1–11

Table 2
Review of representative data sources reporting solubility of H2S in NaCl(aq) reaching above ambient conditions.

Temperature range Pressure range NaCl molality Number of Reference


T (K) P (MPa) me (mol kg−1) experimental points1

284–603 0.6–30 1–6 295 (295) Drummond (1981)


299–368 0.1 1–5 229 (78) Barrett et al. (1988)2
429–594 2.7–14 0.2–2.5 23 (0) Suleimenov and Krupp (1994)
313–393 0.3–10 4, 6 56 (0) Xia et al. (2000)
323–393 1.8–31 1, 3, 5 20 (2) Koschel et al. (2013)
1
Number of data points discarded from the final correlation is given in parenthesis.
2
Considered in the final correlation up to temperature of 354 K only.

system) were discarded because the H2S-rich data phase is in fact


composed of a substantial percentage of water that is close to its
normal boiling point. In this case the assumptions implicit in our
Eq. (7), which is used for approximative description of the vapor
phase, are not justified.
iii) From the very beginning we have excluded 14 points of Xia et al.
(2000) corresponding to the coexistence of liquid, vapor and
solid clathrate-hydrate, because these values according to the
author's comment are no longer the solubilities of H2S in the
aqueous solution. They are not considered in the data points
listed in Table 2.
iv) During fitting we have excluded several additional experimental
points that display relative deviations from the fit exceeding 20%.
This was the case for 6 and 2 outliers for the binary and ternary
systems, respectively, as indicated in parentheses in Tables 1
and 2.

A detailed listing of all the solubilities in our database with both in-
cluded and excluded data points and their deviations from the fit is
given as supplementary material in the electronic Annex. This also con-
tains the contents of H2S in the vapor phase estimated with Eq. (7).
Fig. 1a displays the distribution of the available experimental data
points in T–P projection for the binary H2S–H2O system, while Fig. 1b
shows the data in a three-dimensional T–P–mNaCl perspective for both
the binary and ternary system (mNaCl ≥ 0). The phase boundary of the
clathrate occurrence in the binary H2S–H2O system is visualized in an
insert of Fig. 1a. In the Figures, dots denote the data used in the final cor-
relation and crosses those discarded at various stages of the evaluation
and fitting procedure. It is apparent that the T–P–mNaCl region up to
10 MPa is reasonably covered by the data, yet even here the coverage
at low temperature and high salinity is not dense. At high pressures a
few measurements are available, the highest experimental pressures
reaching 31 MPa (Koschel et al., 2007, 2013) but only at temperatures
up to 393 K. At higher temperatures the upper pressure is 21 MPa.
Therefore we recommend limiting the use of the model in extrapolation
only up to 40 MPa.

4. Solubility fit for H2S–H2O–NaCl system using the Pitzer


interaction model

Pitzer's semitheoretical model (Pitzer, 1991) allows deviations of


electrolyte and nonelectrolyte solutes from ideal behavior (Henry's
law) to be described in terms of equations for activity coefficients and
their temperature and pressure derivatives. These equations are virial
expansions in concentration of solutes containing adjustable parame-
ters, reflecting interactions between individual species present in an
aqueous solution. When considering dissolved H2S as a nominal non-
electrolyte solute without taking into account its dissociation, its activi-
ty coefficient in a ternary system of a strong electrolyte is given by:
Fig. 1. T–P for binary (a) and T–P–mNaCl for ternary (b) distribution of experimental
data points for aqueous H2S solubility. Dots correspond to the experimental data
 2
ln γ m;t
s ¼ 2mts λss þ 3 mts τ sss þ 2me ðλsc þ λsa Þ ð15Þ used in the fitting procedure. Red crosses are the discarded data. Experimental points
of Drummond (1981) were discarded and are omitted from the figure for clarity. The
þ3m2e ðτscc þ 2τ sca þ τ saa Þ þ 3mts me ð2τ ssc þ 2τssa Þ; inset in Fig. a illustrates phase equilibria in the H2S–H2O system.
N.N. Akinfiev et al. / Chemical Geology 424 (2016) 1–11 5

where subscripts s, e, c, and a stand for H2S, electrolyte (NaCl), and its Table 3
cation and anion, respectively, and superscript t denotes ternary solu- Recommended values for solubility of H2S in water at temperatures up to 573 K and pres-
sures to 40 MPa.
tion. Symbols λ and τ denote coefficients for binary and ternary short-
range interactions between electrolyte and nonelectrolyte solutes. Let P/MPa T/K
us recall that molality mts in Eq. (15) is an independent variable and is 298.15 323.15 373.15 423.15 473.15 523.15 573.15
not necessarily related to the equilibrium solubility of H2S exclusively, −1
Solubility m/mol·kg
as was the case of the relationships in Section 2.
0.1 0.0994 0.0525
To simplify the notation and reduce the number of adjustable
0.5 0.505 0.293 0.130 0.0061
parameters the second and third virial coefficients are introduced 1 0.990 0.596 0.297 0.136
as Bse = λsc + λsa, Csee = τscc + 2τsca + τsaa and Csse = τssc + τssa to 2 1.188 0.635 0.405 0.114
obtain 5 2.026 1.651 1.271 0.956 0.300
10 2.100 2.703 2.752 2.622 2.045 0.522
 2 20 2.237 3.065 4.407 5.395 5.458 4.350
ln γ m;t
s ¼ 2mts λss þ 3 mts τ sss þ 2me Bse þ 3m2e C see þ 6mts me C sse : ð16Þ 30 2.369 3.363 5.097 6.63 7.11 6.41
40 3.641 5.601 7.30 7.95 7.43

Values in italics were obtained by extrapolation beyond the range of experimental data.
In the case of the binary system H2S–H2O, Eq. (15) simplifies to:

Pitzer et al. (1984) and recently by Akinfiev and Diamond (2010), the
ln γ m 2
s ¼ 2ms λss þ 3ms τ sss : ð17Þ
following equation proved to be useful:

Values for the H2S interaction parameters λss and τsss result from 100 T
λss ¼ aλ þ bλ þ cλ ð18Þ
the correlation of the binary solubility data with γm T−228 T−760
s calculated from
Eqs. (12) and (13), while the remaining parameters Bse , Csee , Csse can
with aλ = − 0.19515 ± 0.00534, bλ = 0.102822 ± 0.00409, and
in addition be obtained from the regression of the ternary data with
cλ = − 0.033726 ± 0.00212, while the ternary parameter was a con-
γm,t
s determined via Eqs. (13) and (14).
stant τsss = 0.004900 ± 0.000524 (2σ confidence level interval). The
During the fitting procedure all the experimental data sources listed
aλ and bλ parameters exhibit very low uncertainty margins and those
in Tables 1 and 2 were first considered and then the data exhibiting
for cλ and τsss are acceptable; this is also reflected by the value of the co-
large deviations from the fit were discarded. As explained above this
efficient of correlation R2 = 0.9366 for the whole set of binary experi-
was due to systematic errors (Drummond, 1981) or conditions where
mental points. The mean relative deviation of the experimental H2S
the above approximations were not justified (data of Barrett et al. at
solubilities in water from the model calculations mPit
T N 368 K); in addition several random outliers were also eliminated s

(see the electronic Annex for detailed information). vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi


uX n   2
Several variants of the Pitzer equations were tested considering u
u ms −mPit =mPit
the parameters as constants or as empirical functions of temperature t s s
i
i
sm ¼  100 ð19Þ
and pressure in an effort to obtain a good data fit with a minimum of n−1
well-determined parameters. No explicit pressure dependence of the
parameters was found. However, for the temperature dependence of was sm = 5.18% for n = 483. The analogous mean relative deviations for
λss, by analogy with the correlations by Rogers and Pitzer (1982) and the individual binary data subsets are given in the electronic Annex.

Fig. 2. Pressure dependence of H2S solubility (mol·kg−1) generated at several temperatures in comparison with the experimental data. Left panels correspond to the binary H2S–H2O and
the right panels to the ternary H2S–H2O–NaCl systems. Numbers in the lines of the right figures stand for NaCl(aq) molality. Plotted experimental points correspond to the temperatures
differing from those given in the figures by less than 10 K.
6 N.N. Akinfiev et al. / Chemical Geology 424 (2016) 1–11

Table 4a Table 4c
Recommended values for solubility of H2S in 1 m NaCl(aq) at temperatures up to 573 K Recommended values for solubility of H2S in 4 m NaCl(aq) at temperatures up to 573 K
and pressures to 40 MPa. and pressures to 40 MPa.

P/MPa T/K P/MPa T/K

298.15 323.15 373.15 423.15 473.15 523.15 573.15 298.15 323.15 373.15 423.15 473.15 523.15 573.15
−1 −1
Solubility m/mol·kg Solubility m/mol·kg

0.1 0.0869 0.0469 0.1 0.0583 0.0336


0.5 0.439 0.261 0.118 0.0056 0.5 0.292 0.185 0.0877 0.0041
1 0.857 0.527 0.267 0.123 1 0.564 0.368 0.197 0.091
2 1.039 0.567 0.363 0.102 2 0.710 0.409 0.263 0.072
5 1.752 1.437 1.110 0.831 0.260 5 1.168 0.983 0.763 0.561 0.171
10 1.814 2.301 2.311 2.177 1.689 0.433 10 1.207 1.505 1.475 1.345 1.010 0.251
20 1.930 2.596 3.639 4.441 4.490 3.521 20 1.280 1.677 2.198 2.551 2.501 1.873
30 2.042 2.840 4.225 5.59 6.06 5.41 30 1.351 1.818 2.523 3.25 3.53 3.04
40 3.068 4.673 6.26 6.91 6.43 40 1.950 2.783 3.72 4.20 3.81

Values in italics were obtained by extrapolation beyond the range of experimental data. Values in italics were obtained by extrapolation beyond the range of experimental data.

For the ternary system the best fit for Bse was assumed as follows: phase only. The data generated at high temperatures above the pressure
where experimental data are available are given in italics to emphasize
100 T that they should be considered with caution. The graphical presentation
Bse ¼ bB þ cB ð20Þ
T−228 T−760 of these data as a function of temperature at four isobars and as a func-
tion of pressure at four isotherms is given in Fig. 3. Considering the crit-
with bB = 0.03568 ± 0.00104, cB = −0.02354 ± 0.00136, the coeffi- ical parameters of H2S (Tc = 373.4 K and Pc = 8.96 MPa) the variation of
cients Csee set to zero and Csse = 0.002558 ± 0.000044 (2σ confidence H2S solubility is visualized once at subcritical, once at near critical and
level interval). The four parameters exhibit low uncertainty margins twice at supercritical pressure or temperature and this is also reflected
with the coefficient of correlation R2 = 0.9809. Use of these parameters by the forms of curves representing solubility. It can be noted that the
together with previously estimated binary ones (i.e. λss , τsss) provide a solubility of liquid H2S decreases with temperature at constant pressure
satisfactory H2S solubility description in the NaCl solutions over the while it exhibits a maximum for a supercritical fluid (Fig. 3, left). A dis-
whole temperature and pressure range. The mean relative deviation of tinction between the dissolution of gaseous and liquid H2S is particular-
the experimental H2S solubilities from the model calculation mt,Pit
s ly apparent for the 323 K isotherm (Fig. 3, right) with a steep increase in
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi solubility up to the vapor pressure of H2S and a moderate change at
uX n   2
u higher pressures. A completely different picture is observed for super-
u mts −mt;Pit =mt;Pit
t s s
i critical isotherms 473 and 573 K originating at the vapor pressure of
i
stm ¼  100 ð21Þ water with the solubility of H2S rapidly increasing.
n−1
Correspondingly, Fig. 4 shows the effect of NaCl concentration on
was stm = 6.12% for the whole set of included experimental points of solubility of H2S for the highest and lowest isobars and isotherms of
n = 248. Mean relative deviations for the individual ternary data sub- Fig. 3. The salting-out effect leading to a decrease in solubility is ob-
sets are listed similarly as for the binary system in the electronic served while the change with temperature and pressure remains analo-
Annex. A good reproduction of both binary and ternary experimental gous as for dissolution in pure water. It is also apparent that the drop in
data points by the fit is illustrated in Fig. 2. solubility for 2 m NaCl(aq) solution compared to water is somewhat
higher than that for 4 compared to 2 m NaCl(aq), confirming that the
5. Results and discussion salting out becomes less effective with increasing concentration of the
electrolyte.
The solubility data generated from the fit are listed in Tables 3 and Finally Fig. 5 shows three isopleths in a P–T projection for the 1 m
4a,4b,4c, and 4d at selected temperatures, pressures and concentrations concentration of dissolved H2S in pure water and at two concentrations
up to 573 K, 40 MPa and 6 mol·kg−1. The same information is also avail- of NaCl. It is apparent that at high temperatures the pressure increases
able in the electronic Annex. The tabulated values are limited at low steeply for dissolving the same amount H2S in water or NaCl(aq) solu-
temperatures (below 323 K) by existence of a clathrate and at high tem- tion. On the other hand approximately parallel character of the isopleths
peratures by the vapor pressure of water since we focus on the liquid suggests that the temperature does not affect significantly the pressure

Table 4b Table 4d
Recommended values for solubility of H2S in 2 m NaCl(aq) water at temperatures up to Recommended values for solubility of H2S in 6 m NaCl(aq) at temperatures up to 573 K
573 K and pressures to 40 MPa. and pressures to 40 MPa.

P/MPa T/K P/MPa T/K

298.15 323.15 373.15 423.15 473.15 523.15 573.15 298.15 323.15 373.15 423.15 473.15 523.15 573.15
−1 −1
Solubility m/mol·kg Solubility m/mol·kg

0.1 0.0761 0.0420 0.1 0.0447 0.0269


0.5 0.383 0.233 0.107 0.0050 0.5 0.223 0.147 0.072 0.0034
1 0.744 0.467 0.241 0.111 1 0.429 0.292 0.161 0.0742
2 0.913 0.508 0.326 0.0906 2 0.557 0.332 0.214 0.058
5 1.523 1.259 0.975 0.726 0.226 5 0.907 0.779 0.606 0.438 0.130
10 1.577 1.980 1.969 1.834 1.411 0.360 10 0.937 1.172 1.137 1.015 0.739 0.176
20 1.676 2.224 3.039 3.659 3.6774 2.843 20 0.992 1.299 1.652 1.851 1.755 1.260
30 1.771 2.425 3.522 4.67 5.10 4.50 30 1.0454 1.402 1.880 2.33 2.46 2.05
40 2.613 3.901 5.29 5.92 5.47 40 1.499 2.061 2.66 2.94 2.60

Values in italics were obtained by extrapolation beyond the range of experimental data. Values in italics were obtained by extrapolation beyond the range of experimental data.
N.N. Akinfiev et al. / Chemical Geology 424 (2016) 1–11 7

Fig. 3. H2S solubilities in water as a function of temperature up to 573 K at isobars of 5, 10, 20 and 40 MPa (left) and as a function of pressure up to 40 MPa at isotherms of 323, 373, 473 and
573 K (right) computed using the model in this study.

increase necessary for achieving the same degree of H2S dissolution with a detail comparison of our and Duan et al. (2007) data). It results
when the concentration of electrolyte is rising. from the comparison that for pure water agreement in solubilities at
For comparison we also include Fig. 6 where analogous plots to temperatures up to 483 K and over the whole pressure range is within
those in Fig. 4 are presented for carbon dioxide using the data of 8%, which is comparable to the experimental uncertainty. However, at
Mao et al. (2013). The solubility of CO2 is considerably lower due to its temperatures of 513 K in the lower half of the pressure range the values
nonpolar character. For instance, at 473 K and 20 MPa the solubility of Duan et al. are lower by 12 to 16% than those from our model, whereas
of H2S in pure water is 5.39 mol kg−1 while that for CO2 is only for higher pressures the differences are less than 10% and switch from
1.37 mol kg−1.The salting-out effect is therefore somewhat stronger, negative to positive. Solubilities of H2S in 2 and 4 m NaCl(aq) solutions
yet it seems to be weakening with increasing salt concentration more predicted from the two models differ by less than 10% at most tempera-
than for H2S. Otherwise the trends in the evolution of solubility with tures and pressures except at temperature of 483 K and pressure of
temperature and pressure are similar to that for H2S. However, CO2 is 20 MPa and at temperature of 513 K and pressures between 16 and
always supercritical at these conditions (Tc = 304 K, Pc = 7.4 MPa) 20 MPa, where Duan's values are higher by 14% and between 14% and
and this explains a different continuous character of curves in Fig. 6 at 26%, respectively. It is notable that our model is closer to the experimen-
the lower isobar and isotherm compared to H2S. tal data of Xia et al. (2000) for high concentrations of NaCl(aq). For
It is of interest to compare our recommended values with those gen- example, at 393 K, 10 MPa and me = 6.0 mol kg−1 the experimental sol-
erated from other representative models. For that reason we have calcu- ubility is 1.16 mol kg−1. Our model yields ms = 1.15 mol kg−1, while
lated the solubilities at 8 temperatures up to 513 K, at 15 pressures Duan et al.’s (2007) model value is ms = 1.26 mol kg−1.
to 20 MPa and at NaCl molalities of 0, 1, 2 and 4 mol·kg−1 at conditions In addition, we also compared the solubilities generated by our
where the model of Duan et al. (2007) should be valid and the recom- model with those from the model of Springer et al. (2015), using the
mended values are tabulated in their publication (see Electronic Annex data supplied by courtesy of OLI company at 7 temperatures up to

Fig. 4. H2S solubilities in water and NaCl(aq) (me = 0, 2, 4 mol·kg−1) as a function of temperature up to 573 K at isobars of 10 and 40 MPa (left) and as a function of pressure up to 40 MPa
at isotherms of 323 and 573 K (right).
8 N.N. Akinfiev et al. / Chemical Geology 424 (2016) 1–11

important deviations are for the salt-free system, where the solubilities
are predicted to be quite high. The NaCl content dampens this effect so
that the deviations between our and the OLI models in the extrapolation
region become smaller with increasing salt content.
The activity coefficient of H2S expresses the deviation from Henry's
law, valid exactly for infinitely dilute solution, and departure of γS
from unity is a measure of solution nonideality. This is strictly valid for
the binary system H2S–H2O, for which γs = 1 when ms = 0, as apparent
from Eq. (17). In the case of the ternary system, however, γts ≠ 1 even at
zero concentration of H2S, due to the presence of an electrolyte in the
solution (see Eq. (16)) whose effect is not taken into account in the
standard state properties of H2S solute. The values of γts below 1 reflect
generally hydrophilic behavior, which increases solubility compared to
ideal behavior, while values above 1 reflect repulsion forces due to the
salting-out effect. This is illustrated in Fig. 7, which shows the depen-
dence of the activity coefficient of H2S on its concentration in water
and in 2 and 4 m NaCl solutions at two temperatures. The end points
on the curves correspond to the activity coefficient of H2S at the concen-
tration where the solution is saturated with respect to hydrogen sulfide.
The tabulated values of activity coefficients corresponding to saturated
solution at temperatures, pressures and NaCl(aq) concentrations listed
Fig. 5. P–T projection of aqueous isoplethic curves for 1 m concentration of dissolved H2S in Tables 3 and 4a–4d can be found in the electronic Annex. It should
with numbers corresponding to the NaCl(aq) molalities. be noted, however, that the activity coefficients were determined
from Eqs. (12) and (14), for which two important approximations are
573 K, at 7 pressures to 40 MPa and at NaCl molalities of 0, 1, 2, 4 and inherent: i) estimation of the composition of the vapor phase from
6 mol·kg−1. The comparisons reach higher temperatures and pressures Eq. (7), which is a crude simplification, ii) supposition that the fugacity
than those with Duan et al. (2007) and therefore the differences are also of a mixture can be calculated as a product of the solute concentration in
more important. This is namely the case near the upper pressure limit the vapor phase and the fugacity of the pure fluid (ideal mixture of real
where both models are in the extrapolation regime. Differences above fluids), see Eq. (8). Therefore, the activity coefficient does not corre-
20% were observed in some cases at 473, 523 and 573 K, but a striking spond exactly to the binary and ternary interaction coefficients between
discrepancy was discovered for the solubility of H2S in water at these the solutes as defined by Pitzer, because it is affected by these two
three temperatures at 30 and 40 MPa, where the Springer et al. approximations.
(2015) model yields solubilities more than 50% higher than ours. The The change in solubility of a nonelectrolyte in the presence of an
differences then rapidly decrease with increasing concentration of electrolyte compared to its solubility in pure water is often expressed
salt at these conditions. The reason might be that, unlike our model by the empirical Setchenow equation
and the approach of Duan et al.'s (2007), the parameters in the OLI
model are constrained by both the water-rich and H2S-rich phases, af- ms m
fecting the curvature of the phase coexistence curve in the extrapola- ln ¼ kS me ð22Þ
mts
tion region. In the salt-free system, there is a critical point of the
mixture at high temperature and pressure, so the phase equilibrium
loop should close at some point. The Springer et al. (2015) model where the electrolyte-specific Setchenow constant, km S , which is posi-
seems to overanticipate the closure of the loop from below at the tive for most electrolytes, is a measure of the salting-out effect. By
highest pressures and at 473 K and above. This is also why the most combining Eqs. (12) and (14) one obtains by use of the Pitzer's

Fig. 6. Computed after Mao et al. (2013) CO2 solubilities in water and NaCl(aq) (me = 0, 2, 4 mol·kg−1) as a function of temperature up to 573 K at isobars of 10 and 40 MPa (left) and as a
function of pressure up to 40 MPa at isotherms of 323 and 573 K (right).
N.N. Akinfiev et al. / Chemical Geology 424 (2016) 1–11 9

Fig. 7. Activity coefficient of H2S in water and NaCl(aq) (me = 0, 2, 4 mol·kg−1) as a function of its concentration at temperature of 323 K (left) and 573 K (right) at pressure of 40 MPa,
computed using the current model.

Eqs. (16) and (17) H2S, it is possible to obtain the mean activity coefficient of NaCl, the os-
motic coefficient (simply related to water activity) and the molar
1 γt       
m
kS ¼ ln s ¼ 2 mts −ms =me λss þ 3
2
mts −ðms Þ2 =me τsss volume (or density) of the bulk liquid. These properties can be calcu-
me γs lated using Pitzer's concept at a given temperature and pressure for
þ 2Bse þ 3me C see þ 6mts C sse ; ð23Þ any H2S and NaCl concentrations up to halite saturation. The model is
available as a computer code. The main relations used in the code com-
where the limit for electrolyte concentration me = 0 is putations are given in the Appendix A.
Fig. 8 illustrates the temperature dependence of the density of terna-
m 2Bse þ 6C sse ms ry H2S–H2O–NaCl solutions for NaCl molalities of 0, 2, and 4 mol·kg−1
lim kS ¼ : ð24Þ
me →0 1 þ 2λss ms þ 6τ sss m2s computed using the current model. The concentration of H2S corre-
sponds to the saturation conditions at a given temperature and NaCl
The values of km S for H2S do not exhibit well pronounced temper- molality at constant pressure of 40 MPa. The overall trend reflects
ature and pressure dependencies and are in the range of 0.16– the change in solvent density with temperature. Additional decrease
0.22·kg·mol−1, decreasing for concentrations of NaCl(aq) above 2 m. of the solution density at high temperatures is due to the increasing par-
The results of this study have been incorporated in a computer code tial molar volume of H2S(aq) at these state parameters (Hnědkovsky
allowing thermodynamically consistent description of numerous prop- et al., 1996).
erties of the liquid phase in the ternary H2S–H2O–NaCl system. Thus, be-
side the solubility, activity coefficient and the Setchenow constant of
6. Conclusions

In this study we have selected reliable data from the literature pre-
senting the experimental solubility of H2S in H2O and in NaCl solutions.
We used these data to develop an activity — fugacity model that repro-
duces the experimental data close to their error margins. The model
is recommended for use in the T – P – me range to 573 K, 40 MPa
and 6 m NaCl(aq), but it should be applied with caution at pressures
above 20 MPa for temperatures higher than 473 K, as it is not supported
in this range by experimental data. The lack of reliable solubilities at
high state parameters is in general the major problem in constraining
any model and also explains the differences observed in different pre-
dictions at high temperature and pressures. Particularly, the composi-
tion of the vapor phase is rarely reported in experiments and we have
accordingly opted for a simple approximation to express it. Therefore,
the presented model can only be used to calculate H2S concentrations
in the dense (water-rich) phase. The strength of the model, compared
to earlier publications, is the use of the most representative equations
of state for pure H2S and for its Henry's law constant. An asset is
also an incorporation of the new model in a broader thermodynamic
scheme, which enables calculations not only of solubility but also a
number of other thermodynamic properties for the H2S–H2O–NaCl sys-
tem. These calculations are available using the developed computer
Fig. 8. Densities of H2S–H2O–NaCl solutions as a function of temperature (solid lines). H2S
concentrations correspond to the saturation solubility at given temperature and NaCl
code H2O–H2S–NaCl for Windows. This program is freeware; along
molality at constant pressure of 40 MPa. Dashed lines stand for density of corresponding with its detailed documentation it is available online at http://www.
binary H2O–NaCl system and are drawn for comparison. geol.msu.ru/deps/geochems/soft/index_e.html and can be used freely.
10 N.N. Akinfiev et al. / Chemical Geology 424 (2016) 1–11

Acknowledgments The molar volume, Vm of the H2S–H2O–NaCl system and hence den-
sity ρ of the bulk ternary solution at any concentration of NaCl and of
This research was supported by: i) the cooperation project CNRS- H2S up to saturation is calculated approximately with the help of the
RAS EDC25175 2012-2013 (N.A. and V.M.), ii) grant 14-17-00366 of Young's rule:
the Russian Science Foundation, (N.A.), iii) the National Programme
for Sustainability I LO 1201 and the OPR&DI project of the Centre 1=M w me
for Nanomaterials, Advanced Technologies and Innovation, CZ.1.05/ Vm ¼ V m;w þ t V ϕ;e ðA:3Þ
mts þ me þ 1=M w ms þ me þ 1=Mw
2.1.00/01.0005 at the Technical University of Liberec, both funded by mts
the Ministry of Education, Youth and Sports of the Czech Republic þ t V ∞;
ms þ me þ 1=M w s
(V.M.) The authors thank also Andre Anderko (OLI) for supplying the
data generated from the model of Springer et al. (2015) for comparison  
1 1 me mts
and for discussing the possible reasons for differences in the models. We ρ¼  þ t Me þ t Ms ;
Vm ms þ me þ 1=Mw ms þ me þ 1=Mw
t ms þ me þ 1=Mw
are very grateful to Larryn W. Diamond for his comments and correction
of the manuscript. We are also indebted to the constructive reviews by ðA:4Þ
the editor Carla Koretsky and three anonymous ChemGeo referees who
helped to improve the manuscript. where Mw, Me, and Ms. stand for molar masses (kg·mol−1) of H2O, NaCl,
and H2S, respectively, and the meaning of individual variables is as fol-
Appendix A lows: i) Vm,w is the molar volume of pure H2O computed using Wagner
and Pruss (2002) formulation. ii) Vϕ,e is the apparent molar volume of
Below are listed the main relations used in the code “H2O–H2S– NaCl in the H2S-free aqueous solution calculated as a combination of
NaCl” computations. For the mean activity coefficient of NaCl γe, it holds the partial molar volume at infinite dilution for NaCl(aq) estimated
from the HKF model (Johnson et al., 1992) and Pitzer's corrections
γ 3
ln γ e ¼ f þ me Bγe þ m2e C γe þ 2ms Bse þ 3ms me C see þ m2s C sse ; ðA:1Þ for the specified molality of NaCl with coefficients given by Archer
2 (1992). iii) V∞s is the partial molar volume of H2S at infinite dilution in
NaCl-free aqueous solution, as it holds Vφ,s = V∞
s because the coefficients
and the osmotic coefficient, ∅, is then calculated according to
λss τsss in Eq. (17) are pressure independent. The value of V∞
s is obtained

ð∅−1Þ  ðms þ 2me Þ ¼ 2f I þ 2m2e B∅



3 ∅ 2 3 from the equation of Akinfiev and Diamond (2003), whose parameters
e þ 4me C e þ ms λss þ 2ms τ sss
ðA:2Þ were adjusted consistently with the SOCW model (Sedlbauer et al.,
þ4ms me Bse þ 6m2s me C sse þ 6ms m2e C see ; 2000) used for calculation or the Henry's constant for H2S (Majer
et al., 2008).
where f γ and f ∅ are the extended Debye–Hückel terms
! Appendix B. Supplementary data
γ I0:5 2  
∅ A I 0:5
f ¼ −A∅ 0:5
þ ln 1 þ bI 0:5
; f ¼ − ∅ 0:5 ;
1 þ bI b 1 þ bI Supplementary data to this article can be found online at http://dx.
doi.org/10.1016/j.chemgeo.2016.01.006.
and the parameters Bγe , Ceγ and Be∅, Ce∅ are defined as follows:
  h i References
Bγe ≡ Be þ B∅ ð0Þ ð1Þ
e ; Be ¼ β e þ β e  g 1 α 1 I
0:5
; B∅ ð0Þ ð1Þ
e ¼ β e þ β e  exp −α 1 I
0:5
;
Akinfiev, N.N., Diamond, L.W., 2003. Thermodynamic description of aqueous nonelectro-
    lytes at infinite dilution over a wide range of state parameters. Geochim. Cosmochim.
C γe ≡ 3C ðe0Þ þ 4C ðe1Þ  g 2 α 2 I0:5 ; C∅
e ¼ C ðe0Þ þ C ðe1Þ  exp −α 2 I 0:5 ; Acta 67, 613–627.
Akinfiev, N.N., Diamond, L.W., 2010. Thermodynamic model of aqueous CO2–H2O–NaCl
solutions from −22 to 100 °C and from 0.1 to 1000 MPa. Fluid Phase Equilib. 295,
2 104–124.
g 1 ðxÞ ≡ ½1−ð1 þ xÞ expð−xÞ; g 2 ðxÞ Archer, D.G., 1992. Thermodynamic properties of the NaCl + H2O system. II. Thermody-
x2  
x4 1 namic properties of NaCl(aq), NaCl⋅2H2O(cr), and phase equilibria. J. Phys. Chem.
¼ 6− 6 þ 6x þ 3x2 þ x3 −  expð−xÞ  4 ; Ref. Data 21, 793–829.
2 x Barbero, J.A., McCurdy, K.G., Tremaine, P.R., 1982. Apparent molal heat capacities and
volumes of aqueous hydrogen sulfide and sodium hydrogen sulfide near 25 °C: the
where α1 = 2.0 kg·mol−1, α2 = 2.5 kg0.5·mol−0.5 and b = 1.2. Here, temperature dependence of H2S ionization. Can. J. Chem. 60, 1872–1880.
as usual, I denotes the ionic strength of the solution (in mol·kg−1) Barrett, T.J., Anderson, G.M., Lugowski, J., 1988. The solubility of hydrogen sulphide in
0–5 m NaCl solutions at 25–95 °C and one atmosphere. Geochim. Cosmochim. Acta
and A∅ is the Debye-Hückel slope for the osmotic coefficient 52, 807–811.
Barta, L., Bradley, D.J., 1985. Extension of the specific interaction model to include gas
   1:5
1 2πNA ρw 0:5 e2 solubilities in high temperature brines. Geochim. Cosmochim. Acta 49, 195–203.
A∅ ¼ Carroll, J.J., Mather, A.E., 1989. Phase equilibrium in the system water-hydrogen sulphide:
3 1000 4πε0 εkT experimental determination of the LLV locus. Can. J. Chem. Eng. 67, 468–470.
Chapoy, A., Mohammadi, A.H., Tohidi, B., Valtz, A., Richon, D., 2005. Experimental mea-
with NA Avogadro's number, ρw the density of water, e the electronic surements and phase behaviour modelling of hydrogen sulphide-water binary sys-
tem. Ind. Eng. Chem. Res. 44, 7567–7574.
charge, k Boltzmann's constant, ε0 the permittivity of vacuum and Cox, J.D., Wagman, D.D., Medvedev, V.A., 1989. CODATA Key Values for Thermodynamics.
ε the dimensionless dielectric constant (the relative permittivity) of Hemisphere Publishing Corp.
water (all units are SI). Values of ρw and ε in the code are computed Diamond, L.W., Akinfiev, N.N., 2003. Solubility of CO2 in water from −1.5 to 100 °C and
from 0.1 to 100 MPa: evaluation of literature data and thermodynamic modelling.
using recommendations of Wagner and Pruss (2002) and Fernández Fluid Phase Equilib. 208, 265–290.
et al. (1997), respectively. Drummond, S.E., 1981. Boiling and Mixing of Hydrothermal Fluids: Chemical Effects on
The first three right-hand terms in Eqs. (A.1) and (A.2) designate the Mineral Precipitation (Ph.D. Thesis) Department of Geosciences, Pennsylvania State
University (380 pp.).
mean ionic activity coefficient of NaCl and the osmotic coefficient in the
Duan, Z., Møller, N., Weare, J.H., 1992. An equation of state for the CH4–CO2–H2O system. I.
binary NaCl–H2O system, respectively. Their adjustable parameters Pure systems from 0 to 1000 °C and 0 to 8000 bars. Geochim. Cosmochim. Acta 56,
β(0) (1) (0) (1)
e , βe , Ce and Ce were adopted from the comprehensive study of 2605-2617.
Archer (1992) without any modification. The next terms then reflect Duan, Z., Møller, N., Weare, J.H., 1996. Prediction of the solubility of H2S in NaCl aqueous
solution: an equation of state approach. Chem. Geol. 130, 15–20.
the effect of H2S with the coefficients determined in this study as de- Duan, Z., Sun, R., Liu, R., Zhu, C., 2007. Accurate thermodynamic model for the calculation
scribed above. of H2S solubility in pure water and brines. Energy Fuel 21, 2056–2065.
N.N. Akinfiev et al. / Chemical Geology 424 (2016) 1–11 11

Dubessy, J., Tarantola, A., Sterpenich, J., 2005. Modelling of liquid-vapour equilibria in the Mao, S., Zhang, D., Li, Y., Liu, N., 2013. An improved model for calculating CO2 solubility in
H2O-CO2–NaCl and H2O–H2S–NaCl systems to 270 °C. Oil Gas Sci. Technol. Rev. IFP aqueous NaCl solutions and the application to CO2–H2O–NaCl fluid inclusions. Chem.
60, 339–355. Geol. 347, 43–58.
Fernández, D.P., Goodwin, A.R.H., Lemmon, E.W., Levelt Sengers, J.M.H., Williams, R.C., Pitzer, K.S., 1973. Thermodynamics of electrolytes. I. Theoretical basis and general equa-
1997. A formulation for the static permittivity of water and steam at temperatures tions. J. Phys. Chem. 77 (2), 268–277.
from 238 K to 873 K at pressures up to 1200 MPa, including derivatives and Pitzer, K.S., 1991. Ion interaction approach: theory and data correlation. In: Pitzer, K.S.
Debye–Hückel coefficients. J. Phys. Chem. Ref. Data 26 (4), 1125–1166. (Ed.), Activity Coefficients in Electrolyte Solutions. CRC Press Inc., pp. 75–153.
Fernandez-Prini, R., Alvarez, J., Harvey, A., 2003. Henry's constants and vapor–liquid dis- Pitzer, K.S., Peiper, J.C., Busey, R.H., 1984. Thermoynamic properties of aqueous sodium
tribution constants for gaseous solutes in H2O and D2O at high temperatures. chloride solutions. J. Phys. Chem. Ref. Data 13, 1–102.
J. Phys. Chem. Ref. Data 32, 903–916. Plyasunov, A.V., O'Connell, J.P., Wood, R.H., Shock, E.L., 2000. Infinite dilution partial molal
Gillespie, P.C., Wilson, G.M., 1982. Vapour-liquid and liquid–liquid equilibria: water- properties of aqueous solutions of nonelectrolytes. II. Equations for the standard
methane, water-carbon dioxide, water-hydrogen sulphide, water-pentane. Gas thermodynamic functions of hydration of volatile nonelectrolytes over wide range
Processors Association Research Report RR-48 Project 758-B-77. of conditions including subcritical temperatures. Geochim. Cosmochim. Acta 64,
Hnědkovsky, L., Wood, R.H., 1997. Apparent molar heat capacities of aqueous solutions 2779–2795.
of CH4, CO2, H2S, and NH3 from 304 K to 704 K at 28 MPa. J. Chem. Thermodyn. 29, Rogers, P.S.Z., Pitzer, K.S., 1982. Volumetric properties of aqueous sodium chloride solu-
731–747. tions. J. Phys. Chem. Ref. Data 11, 15–81.
Hnědkovsky, L., Wood, R.H., Majer, V., 1996. Volumes of aqueous solutions of CH4, H2S, Savary, V., Berger, G., Dubois, M., Lacharpagne, J.-C., Pages, A., Thibeau, S., Lescanne, M.,
and NH3 at temperatures from 298.15 K to 705 K and pressures to 35 MPa. J. Chem. 2012. The solubility of CO2 + H2S mixtures in waterand 2 M NaCl at 120 °C and pres-
Thermodyn. 28, 125–142. sures to 35 MPa. Int. J. Greenhouse Gas Control 10, 123–133.
Johnson, J.W., Oelkers, E.H., Helgeson, H.C., 1992. SUPCRT92: a software package for calcu- Sedlbauer, J., O'Connell, J.P., Wood, R.H., 2000. A new equation of state for correlation and
lating the standard molal thermodynamic properties of minerals, gases, aqueous prediction of standard molal properties of aqueous electrolytes and nonelectrolytes
species, and reactions from 1 to 5000 bars and 0° to 1000 °C. Comput. Geosci. 18, at high temperatures and pressures. Chem. Geol. 163, 43–63.
899–947. Selleck, F.T., Carmichael, L.T., Sage, B.H., 1952. Phase behavior in the hydrogen sulfide-
Koschel, D., Coxam, J.-Y., Majer, V., 2007. Enthalpy and solubility data of H2S in water at water system. Ind. Eng. Chem. 44, 2219–2226.
conditions of interest for geological sequestration. Ind. Eng. Chem. Res. 46, 1421–1430. Springer, R.D., Wang, P., Anderko, A., 2015. Modelling the properties of H2S/CO2/salt/H2O.
Koschel, D., Coxam, J.-Y., Majer, V., 2013. Enthalpy and solubility data of H2S in aqueous Soc. Pet. Eng. J. 20, 1120–1134.
salt solutions at conditions of interest for geological sequestration. Ind. Eng. Chem. Suleimenov, O.M., Krupp, R.E., 1994. Solubility of hydrogen sulphide in pure water and in
Res. 52, 14483–14491. NaCl solutions, from 20 to 320 °C and at saturation pressures. Geochim. Cosmochim.
Kozintseva, T.N., 1965. Rostvorimost serovodoroda v vode i solevykh rastvorakh pri Acta 58, 2433–2444.
povyshenykh temperaturakh. In: Khitarov, N.I. (Ed.), Geokhimicheskie issledovaniya Wagner, W., Pruss, A., 2002. The IAPWS formulation for the thermodynamic properties of
v oblasti povyshennykh davlenii i temperatur. Izdatelstvo Nauka, Moscow, ordinary water substances for general and scientific use. J. Phys. Chem. Ref. Data 31,
pp. 121–134 (in Russian). 387–535.
Kuranov, G., Rumpf, B., Smirnova, N.A., Maurer, G., 1996. Solubility of single gases carbon Xia, J., Perez-Salado Kamps, A., Rumpf, B., Maurer, G., 2000. Solubility of hydrogen sulfide
dioxide and hydrogen sulfide in aqueous solutions of N-methyldiethanolamine in the in aqueous solutions of the single salts sodium sulfate, ammonium sulfate, sodium
temperature range 313–413 K at pressures up to 5 MPa. Ind. Eng. Chem. Res. 35, chloride, and ammonium chloride at temperatures from 313 to 393 K and total pres-
1959–1966. sures up to 10 MPa. Ind. Eng. Chem. Res. 39, 1064–1073.
Lee, L.L., Mather, A.E., 1977. Solubility of hydrogen sulfide in water. Ber. Bunsenges. Phys. Zirrahi, M., Azin, R., Hassanzadeh, H., Moshfeghian, M., 2012. Mutual solubility of CH4,
Chem. 81, 1021–1023. CO2, H2S, and their mixtures in brine under subsurface disposal conditions. Fluid
Lemmon, E.W., Span, R., 2006. Short fundamental equations of state for 20 industrial Phase Equilib. 324, 80–93.
fluids. J. Chem. Eng. Data 51, 785–850.
Majer, V., Sedlbauer, J., Bergin, G., 2008. Henry's law constant and related coefficients for
aqueous hydrocarbons, CO2 and H2S over wide range of temperature and pressure.
Fluid Phase Equilib. 272, 65–74.

You might also like