Fluid Phase Equilibria 296 (2010) 219–229

Contents lists available at ScienceDirect

Fluid Phase Equilibria

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

Use of monomer fraction data in the parametrization of association theories

Georgios M. Kontogeorgis a,∗ , Ioannis Tsivintzelis a , Nicolas von Solms a , Andreas Grenner b ,
David Bøgh a , Michael Frost a , Anders Knage-Rasmussen a , Ioannis G. Economou c
a
Center for Energy Resources Engineering (CERE), Department of Chemical and Biochemical Engineering, Technical University of Denmark, Building 229, DK-2800 Lyngby, Denmark
b
Linde AG, Engineering Division, Dr.-Carl-von-Linde-Str. 6-14, 82049 Pullach, Germany
c
The Petroleum Institute, Department of Chemical Engineering, PO Box 2533, Abu Dhabi, United Arab Emirates

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

Article history: Association theories such as the CPA (cubic-plus-association), NRHB (non-random hydrogen bonding)
Received 9 March 2010 equations of state and the various variants of SAFT (statistical associating fluid theory) have been
Received in revised form 26 May 2010 extensively applied to phase equilibrium calculations. Such models can also be used for estimating
Accepted 31 May 2010
the monomer fraction of hydrogen bonding compounds and their mixtures. Monomer fraction data are
Available online 8 June 2010
obtained from spectroscopic measurements and they are available for a few compounds such as pure
water and alcohols as well as for some alcohol–alkane and similar mixtures. These data are useful for
Keywords:
an understanding of the capabilities and limitations of association models. The purpose of this work is
Monomer fraction
Association theories
two-fold: (i) to compare the performance of three models, CPA, NRHB and sPC-SAFT, in predicting the
CPA monomer fraction of water, alcohols and mixtures of alcohol-inert compounds and (ii) to investigate
PC-SAFT whether “improved” model parameters can be obtained if monomer fraction data are included in the
Water parameter estimation together with vapor pressures and liquid densities. The expression “improved”
Methanol implies parameters which can represent several pure compound properties as well as monomer fraction
NRHB data for pure compounds and mixtures. The accuracy of experimental monomer fraction data is discussed,
as well as the role of monomer fraction data in clarifying which association scheme should be used in
these equations of state. The results reveal that the investigated association models (CPA, sPC-SAFT and
NRHB) can predict, at least qualitatively correct, monomer fractions of associating compounds and mix-
tures. Only, small differences are observed between the models. In addition, it has been shown that, using
a suitable association scheme, a single set of parameters can describe satisfactorily vapor pressures, liq-
uid densities and monomer fractions of water and alcohols. The 4C scheme is the best choice for water,
while for methanol there is small difference between the 2B and 3B association schemes.
© 2010 Elsevier B.V. All rights reserved.

1. Introduction data can be used for testing association theories, as will be discussed
in Section 1.2. However, the interpretation of spectroscopic data
In this study a short review of the available experimental data into monomer fractions is not entirely straightforward. Monomer
for the fraction of unbonded molecules in associating pure fluids fraction data for a few compounds have been reported in the litera-
and mixtures and the modeling approaches using various associa- ture, e.g. for pure water, alcohols and for some mixtures of alcohols
tion theories is presented. The ability of three association theories or glycolethers with alkanes. These data are sometimes associated
in predicting the monomer fractions is discussed. Finally, a prelim- with uncertainties, both in terms of accuracy and interpretation,
inary investigation of the use of experimental data for monomer as will be seen later. Table 1 presents some references containing
fractions in the parametrization of association theories is pre- experimental monomer fraction data.
sented. As can be seen from Table 1, relatively few associating com-
pounds, mostly water and alcohols, have been investigated in terms
1.1. Experimental monomer fraction data of the monomer fraction and many of the data, especially those for
pure compounds are rather old. For example, the well-known data
Spectroscopy (FTIR, NMR) can be used for obtaining experimen- from Luck [1], widely used for water, methanol and ethanol are
tal data for the monomer fractions of associating compounds. These almost 30 years old.
Fig. 1 shows the experimental monomer fraction data for the
four alcohols for which we have data. In general the data follow the
∗ Corresponding author. Tel.: +45 45252859; fax: +45 88 22 58. expected trends with temperature and chain length; i.e. increas-
E-mail address: [email protected] (G.M. Kontogeorgis). ing monomer fraction with increasing temperature and alcohol

0378-3812/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.fluid.2010.05.028
220 G.M. Kontogeorgis et al. / Fluid Phase Equilibria 296 (2010) 219–229

Table 1
Experimental spectroscopic data for monomer fractions. Tr is the reduced temperature.

Compound/mixture Conditions Reference

◦ a
Water 0–375 C Luck [1] 0.5 < Tr < 0.95
Methanol 0–316 ◦ C
Ethanol 0–273 ◦ C
Propanol 15–55 ◦ C a
Lien [2]
Octanol 5–100 ◦ C a
Fletcher and Heller [3]
Palombo et al. [4]
Propanol/heptane 15–55 ◦ C Lien [2]
Methanol/hexane Ambient temperature Martinez [5]
Methanol/CCl4 20 ◦ C Prausnitz et al. [6]
2-methoxyethanol/hexane 35 ◦ C, 45 ◦ C Brinkley and Gupta [7]
2-butoxyethanol/hexane Intra- and inter-molecular hydrogen bonding
2-methoxyethanol/hexane 30–50 ◦ C Missopolinou et al. [8]
2-ethoxyethanol/hexane
Pentanol/hexane 25 ◦ C and 35 ◦ C Gupta and Brinkley [9]
Hexanol/hexane
◦
Hexanol/toluene, m-xylene 25–45 C Brinkley and Gupta [7]
Cyclohexanol/toluene 11, 61 ◦ C Brinkley and Gupta [7]
Various alcohol/alkanes 10, 25, 40 ◦ C Asprion et al. [10]
Various alcohol/alkanes 23.3 ◦ C von Solms et al. [11]
a
Data from Luck [1], Lien [2] and Fletcher and Heller [3] are also presented in reference [12].

chain length. One exception is ethanol, as the data of Luck indicate, calculated from the experimentally determined monomeric frac-
somewhat paradoxically, almost identical monomer fractions for tion. The transformation between site and monomer fraction in
methanol and ethanol. This is unlikely, considering that methanol Luck’s work must follow the assumptions adopted by Luck i.e. four
and ethanol have quite different properties, e.g. dielectric constants site for water and three site for the alcohols (more discussion on
equal to 32.6 and 24.3, respectively at room temperature (for com- association sites is provided in Section 1.2). Consequently, Luck’s
parison the dielectric constants of propanol and octanol are 20.2 data for site fractions, must be transformed into monomer frac-
and 10.3). tions if comparisons are to be made with association theories for
Luck [1] is moreover one of the few authors who present both the monomer fractions. Alternatively, the experimental site frac-
unbonded-site and monomer fractions data for water. From these tion data could be compared against the predictions of the models
data it can be concluded that at low to moderate temperatures for the site fractions. This point is not always properly understood
almost all water molecules are in aggregate form (no monomers are in the literature where sometimes the concepts of monomer frac-
present), while at higher temperatures some hydrogen bonds break tions and site fractions (from experiment and from models) are
and a fraction of water molecules exists in monomeric form. Luck confused, thus comparing unequal quantities see e.g. discussions
[1] mentioned that several early theories predicted at 0 ◦ C much by Gupta and co-workers [7,9], Kahl and Enders [14] and von Solms
higher fraction of free –OH groups (around 50–65%) in complete et al. [11,13].
disagreement to the spectroscopic data. Here it should be noted that More spectroscopic data for the hydrogen bonding degree of
Luck presented his data as fractions–OH groups, which according water have appeared after Luck. However, often contradictory
to von Solms et al. [13], should be interpreted as the fraction of free results are obtained, especially for high pressures and tempera-
hydrogen bonding sites. Furthermore, according to von Solms et al. tures. It is not clear at which temperature the number of hydrogen
[13], the fraction of monomeric molecules can be calculated from bonds approaches zero. Gorbaty and Kalinichev [15] report data for
the spectroscopically determined fraction of free hydrogen bond- the degree of hydrogen bonding ( parameter) that was obtained
ing sites. Similarly the fraction of hydrogen bonding sites can be in their laboratory during a period of more than two decades. These
data refer to supercritical water and they are based on IR absorption
and X-ray scattering. The authors report the experimental degree
of hydrogen bonding, which is equal to 1 for ice, in which each
water molecule is assumed to form 4 hydrogen bonds, and tends
to 0 as the density also tends to 0 (for low pressure, high tempera-
ture vapor). Consequently, this parameter represents the fraction of
associated sites (the fraction of sites that participate in hydrogen
bonds). The number of hydrogen bonds, in which each molecule
participates, is n = 4 [Gorbaty and Gupta [16]].
Gorbaty and Gupta [16] and Gorbaty and Kalinichev [15], in their
studies for the structural features and the degree of hydrogen bond-
ing of liquid and supercritical water as function of temperature,
also developed an equation for the dependency of  with tem-
perature ( = −8.68 × 10−4 T(◦ C) + 0.61). This equation is valid in the
range 20–550 ◦ C. At room temperature,  = 0.6, while at the critical
temperature  = 0.3, i.e. to say half the value compared to ambient
conditions, but not zero. This is an important conclusion, and unlike
what stated by some authors, there must be some degree of hydro-
gen bonding for water even at supercritical conditions. Moreover,
with respect to the gaseous state, Bondarenko and Gorbaty [17]
Fig. 1. Experimental monomer fraction data for alcohols (and NRHB predictions). state that hydrogen bonds appear at least up to 450–500 ◦ C at any
The data are from Luck [1], Lien [2], and Fletcher and Heller [3]. density exceeding zero.
 G.M. Kontogeorgis et al. / Fluid Phase Equilibria 296 (2010) 219–229 221

Hofman and Conradi [18] measured the degree of hydrogen

bonding using NMR spectrometry. They report data for an extended
pressure and temperature range, normalized using the degree of
hydrogen bonding for liquid water at 298 K and 1 bar. Consequently,
their normalized data are not directly comparable to the data of
Gorbaty and Kalinichev [15]. In order to make this comparison,
the authors converted the data considering that the fraction of
bonded protons is 0.781 at 298 K and 1 bar. The latter value was
obtained from Walrafen et al. [19]. In this way, a comparison of
the data sets obtained from Hofman and Conradi [18] and Gorbaty
and Kalinichev [15] can be obtained. However, the transformation
can be made using any other available experimental data. Finally,
Kalinichev and Bass [20], in order to calculate the degree of hydro-
gen bonding, performed Monte Carlo simulations for pure water.
As it was mentioned previously, Luck presented data for sat-
urated liquid water as % free –OH groups, which, from a careful
reading of the article, can be interpreted as the fraction of free pro-
ton donor sites [13] (XA ). From such data the fraction of hydrogen
bonded sites, or, in other words, the degree of hydrogen bonding, ,
Fig. 2. Degree of hydrogen bonding in liquid and supercritical water. Experimental
can be easily obtained. Fig. 2 shows all the aforementioned trans-
data (points) from Luck [1], Gorbaty and Kalinichev [15], Hofman and Conradi [18],
formed data for the degree of hydrogen bonding as well as the Walrafen et al. [19] simulation data [20] (dashed-dotted line) and NRHB predictions
non-random hydrogen bonding (NRHB) predictions for 30 MPa. It (solid line).
is worth mentioning that, for moderate pressures, the experimen-
tal data for liquid and supercritical water [15,18] as well as the
NRHB model show a sigmoidal change around the critical temper- that of the association energy being (almost) equal to the enthalpy
ature (no vapor liquid interface occur at pressures higher than the of the hydrogen bonding, while the association volume parame-
critical pressure). ter of the statistical associating fluid theory (SAFT)-type models
More experimental monomer fraction data are available for depends both on the entropy of hydrogen bonding and the co-
mixtures than for pure compounds. This may appear surprising but volume parameter. Table 2 also presents the association energy
an explanation is that mixture measurements are often reported at values from CPA, PC-SAFT and NRHB.
room or low temperatures.
In addition to monomer fraction data, spectroscopy can also pro- 1.2. Calculating monomer fraction data from association theories
vide values for the enthalpy (and entropy) of hydrogen bonding
compounds and mixtures and more such data are available in the Association models based on the Wertheim theory, e.g., per-
literature. Some characteristic values are shown in Table 2. Such turbation equations of state like CPA and PC-SAFT can be used for
data are also useful in conjunction with association theories as the estimating the monomer fraction of hydrogen bonding compounds
two parameters of the association term of these theories are related once the association scheme is known. Three of the most widely
to the hydrogen bonding enthalpy and volume. The clearest link is used association schemes are the two-, three- and four-site ones,

Table 2
Comparison of spectroscopic values for the hydrogen bonding energies of several compounds with the association energies from CPA, PC-SAFT and NRHB. All values are
expressed in K. References of spectroscopic values: water (Luck [1]; Koh et al. [21]), methanol (Nath and Bender [22]; Solomonov et al. [23]), alcohols (Pimentel and McClellan
[24]), phenol (Ksiazczak and Moorthi [25]). The 1051 K value for water is from molecular simulation.

Compound Spectroscopy CPAa PC-SAFT NRHBb

Methanol (2B) 2630 2957 2900 (Gross and Sadowski [26]) 3019
2090 (von Solms et al. [13])
1800 (this work)
2010 (this work)

Ethanol to hexanol (2B) 2526–3007 2590 (ethanol) 2811 (Grenner et al. [27]) 2887
2653 (ethanol) (Gross and Sadowski [26])
2472 (ethanol) (von Solms et al. [13])
2590 (ethanol) (this work)
2526 (butanol, pentanol) 2277 (propanol) (Gross and Sadowski [26]) 2887
2370 (propanol) (Kouskoumvekaki et al. [28])
3219 (octanol) 2755 (Gross and Sadowski [26]) 2887

Water (4C) 1051/1813 2003 1552 (this work) 1937

1695–1804 (Grenner et al. [29,30])
1718 (Kleiner et al. [31])
1801 (Kleiner et al. [31])
1334 (Aparicio-Martinez and Hall [32])
839 (von Solms et al. [13] (m = 3.5))
1506 (von Solms et al. [13] (m = 2.0))
2501 (Gross and Sadowski [26])
1801 (Kleiner et al. [31])
2123 (Aparicio-Martinez and Hall [32])
a
Parameters from reference [33].
b
Parameters from references [34,35].
222 G.M. Kontogeorgis et al. / Fluid Phase Equilibria 296 (2010) 219–229

Table 3
Bonding types in alcohols and water illustrating the basic association schemes, 2B, 3B and 4C which are widely used in perturbation theories. The 2B scheme implies one
proton donor and one proton acceptor per molecule, while the 4C scheme has two proton donors and two proton acceptors per molecule. The “asymmetric” 3B scheme has
either two proton donors and one proton acceptor or one proton donor and two proton acceptors per molecule.

Species Formula Type Site fractions

X A = X B ; X C = 2X A − 1
Alcohol 3B
X1 = X A X B X C

XA = XB
2B
X1 = X A X B

XA = XB = XC = XD
Water 4C
X1 = X A X B X C X D

X A = X B ; X C = 2X A − 1
3B
X1 = X A X B X C

X A = X B ; X C = 2X A − 1
3Ba
X1 = X A X B X C

XA = XB
2B
X1 = X A X B

a
Assuming one proton donor and two proton acceptors for water results to the same mathematical expression as if two proton donors and one proton acceptor are
assumed, however this is physically not likely.

summarized in Table 3. They will be hereafter in this work abbre- where  is the segment diameter, k the Boltzmann factor, while
viated as 2B, 3B and 4C, according to the terminology of Huang and εAi Bj and Ai Bj are the association energy and volume, respectively.
Radosz [36]. For example, alcohols are often modeled using the 2B It can be seen that XK is a function of the association strength,
scheme and water using the 4C scheme. which includes the two association parameters as well as the co-
Using SAFT-type EoS, the monomer mole fraction (fraction of volume parameter or the segment diameter.
molecules of compound 1 that are not bonded at any site), X1 , is Eq. (3) is valid for PC-SAFT and other SAFT variants. The associ-
given as: ation strength for CPA is given by the similar expression [33]:
  εAi Bj
 bi + bj
X1 = XK (1) Ai Bj = g() exp − 1 bij ˇAi Bj and bij = (4)
RT 2
K
where b is the CPA co-volume parameter, while εAi Bj and ˇAi Bj are
where XK is the fraction of molecules i not bonded at site K. It can the association energy and volume, respectively. Finally, g in Eqs.
be in general defined for a compound in a mixture as: (3) and (4) is the radial distribution function (more information is
⎡ ⎤−1 given in the original publications [26,33,36,37]).
 For example, using the 4C scheme, often used for water, with
X Ai = ⎣1 + j X Bj Ai Bj ⎦ (2) four equivalent sites (A = B = C = D), see Table 3, the monomer frac-
j Bj tion is:
4
where j is the molar density of j and Ai Bj is the association X1 = X A X B X C X D = (X A )
strength given by: 2
 εAi Bj
 =
2
(5)
 
Ai Bj = ij3 g hs Ai Bj exp −1 (3) 2 + 1 + 8 + 42 2 + 1 + 8
kT
 G.M. Kontogeorgis et al. / Fluid Phase Equilibria 296 (2010) 219–229 223

For fluids, in which the 3B scheme (XA = XB , XC = 2XA − 1) is 2. Prediction of monomer fraction from association
employed, we get for SAFT-type approaches: theories
2
X1 = X A X B X C = X A (2X A − 1) 2.1. Literature studies

2 2
() − 4 − 1 + ( + 1) ( + 1) + 4 Monomer fraction data have been already used in several studies
= 3
(6) in literature for testing chemical, perturbation and lattice theories
8()
and some investigations are summarized in Table 4.
while, for alcohols, using the often employed 2B scheme, we get for Economou and Donohue [40] presented predictions of various
SAFT-type approaches: association models against data from Monte Carlo simulations,
while Smits et al. [47] presented APACT (associated perturbed
2 2
X1 = X A X B = X A =  (7) anisotropic chain theory) predictions for monomeric water.
1 + 2 + 1 + 4 From Table 4, it can been seen that the ability of various theo-
ries in predicting the monomer fractions was investigated in several
Lattice equations of state, e.g. LFHB [38] (lattice–fluid hydrogen studies. However, some confusion exists in comparing experimen-
bonding) and NRHB [39], are based on somewhat different prin- tal and predicted data as it was discussed by von Solms et al. [11,13]
ciples than the Wertheim-based approaches, but both approaches and by Brinkley and Gupta [7,9]. It is important to compare the
yield similar expressions for the monomer fraction. Using the LFHB “same thing”, e.g. monomer fraction from theories with experimen-
or NRHB equations of state (which have identical hydrogen bond- tal monomer fraction data. Moreover, from the literature studies
ing terms) for pure water with four association sites (two proton presented in Table 4, it can be concluded that different theories have
donors and two proton acceptors), the expression for the monomer been tested (chemical, lattice and perturbation) and although a sys-
fraction is (notice the similarity with Eq. (5)): tematic comparison of all three approaches has not been conducted,
4 rH 4
the results appear to be qualitatively similar. This can be attributed
X1 = X A X B X C X D = X A = (1 − ) to the similar functional form for the monomer fraction, as dis-
2
cussed previously. Different models, such as the CPA, the PC-SAFT
2 and the NRHB equations of state can predict monomer fractions
=
2
(8)
  for alcohols with reasonably good accuracy and correct trends are
2K + 1 + 8K + 42 K 2 + 1 + 8K
observed with respect to chain length (for alcohols) and temper-
ature. In such models, spectroscopic monomer fraction can help
For fluids with three association sites (i.e. two proton donors and guide the scheme selection. The investigations so far with some
one proton acceptor or two proton acceptors and one proton donor, perturbation theories (CPA, PC-SAFT) point out that water is best
XA = XB , XC = 2XA − 1), the expression for the monomer fraction is represented as a four-site molecule (4C), methanol as a three site
(notice the similarity with Eq. (6)): (3B), while for ethanol and heavier alcohols there is small differ-
ence between two- and three-site schemes, with the former (2B)
2 rH 2 being possibly the best choice.
X1 = X A (2X A − 1) = (1 − ) (1 − rH )
2

(K)2 − 4K − 1 + (K + 1) (K + 1)2 + 4K 2.2. Comparison of three association equations of state for
= (9)
8(K) 3 monomer fraction prediction

while for pure alcohols with two association sites, the expression We now wish to compare systematically three equations of state
for the monomer fraction is (notice the similarity with Eq. (7)) [40]: which account explicitly for hydrogen bonding; the CPA, PC-SAFT
2 and NRHB models. The first two belong to the perturbation-type
2
X1 = X A X B = X A = (1 − rH )2 =  (10) theories with similar association terms, but CPA has a simpler term
1 + 2K + 1 + 4K than PC-SAFT for accounting for the physical interactions. NRHB
is the latest version of quasi-chemical (lattice) models. We will
In Eqs. (7) and (8)  is the molar density, r is the number of compare the three models against their ability to predict monomer
segments that each molecule occupies in the hypothetical lattice, fraction of pure compounds (water, alcohols) and mixtures of alco-
H is the number of hydrogen bonds per molecular segment and hols with hydrocarbons. The database presented in Table 1 will be
  used. The presentation will be divided into (i) water and methanol
∗ −
GH
K = exp (11) (ii) heavier alcohols and (iii) mixtures.
RT

where GH is the Gibbs free energy of association, which is given in 2.2.1. Monomer fraction of water and methanol
terms of the energy (E), volume (V), and entropy (S) change upon Fig. 3 shows calculations for the fraction of free sites, XA , of water
formation of a hydrogen bond by the equation: with CPA, sPC-SAFT and NRHB, as a function of temperature.
Water parameters were adopted from Grenner et al. [29,30],
GH = E H + PV H − TS H (12) Kontogeorgis et al. [33] and Tsivintzelis et al. [34], for sPC-SAFT,

Table 4
Investigation of association theories against spectroscopic data for monomer fractions.

System studied CPA SAFT variants Lattice theories Chemical theory

References
Alcohols [13] [13,21,34] [34,42,43,44,45] [12,41]
Water [13,29,30,32] [13,29,30,32,46] [44] –
Alkoxyalcohols/hexane – – [7] –
Alcohol–alkanes [11] [11,34] [9,34] [12]
224 G.M. Kontogeorgis et al. / Fluid Phase Equilibria 296 (2010) 219–229

Fig. 3. Experimental data [1] and modeling with CPA, sPC-SAFT and NRHB of the Fig. 5. Monomer fraction for ethanol with three association theories. Experimental
fraction of free sites as a function of temperature for water. The 4C scheme is used data [1] (points), NRHB, CPA and sPC-SAFT predictions (lines).
for water with all models.

CPA and NRHB, respectively. Typically, parameters obtained from differences exist which become more apparent at higher temper-
vapor pressure and liquid density data alone underestimate the atures. In the case of CPA, two association schemes are compared
experimental monomer fraction data. Water parameter sets which and results are better using the 3B scheme. For NRHB and sPC-SAFT
provide somewhat improved monomer fractions have been pro- the 2B association scheme is used. As shown in Fig. 4, the results
posed [e.g. von Solms et al. [13] for sPC-SAFT], but at a cost of with NRHB and CPA (3B) are overall rather similar and better than
unusual values (e.g. m = 3.5 for water) and poor phase equilibria the results with CPA (2B) and sPC-SAFT.
performance. On the other hand, best LLE results for water-alkanes
are obtained with sPC-SAFT using Grenner et al. [29,30] parameters.
Grenner et al. sPC-SAFT parameters [29,30] are physically more rea- 2.2.2. Monomer fraction of alcohols
sonable compared to those by von Solms et al. [13], but as shown Figs. 5 and 6 present comparison between experimental and
by Grenner et al. [29,30], they yield worse results for the monomer predicted monomer fraction data for ethanol and 1-propanol. The
fraction calculations. It appears that the error in monomer fraction sPC-SAFT parameters for the two alcohols are obtained from Gren-
calculations is linked to the parameter values and actually lower ner et al. [27], based on a generalized approach. This constitutes
m-values (for sPC-SAFT), which are otherwise more reasonable for of a generalized way for estimating parameters for alcohols (and
water and yield better LLE, seem to perform worse for monomer glycols). Grenner et al. [27] applied this generalized method to
fractions and vice versa. simplified PC-SAFT but the approach could be equally well used
Fig. 4 shows results for methanol with the three equations of with other SAFT variants as well (with different coefficients). One
state. Methanol parameters were adopted from Grenner et al. [35], unique set of associating parameters for all alcohols (and one for
Kontogeorgis et al. [33] and Tsivintzelis et al. [34], for sPC-SAFT, CPA all glycols) provide excellent pure compound properties as well as
and NRHB, respectively. The qualitative trend is similar but some mixture phase equilibria. The generalized equations for sPC-SAFT

Fig. 4. Monomer fraction for methanol with three association theories. Experimen- Fig. 6. Monomer fraction for 1-propanol with three association theories. Experi-
tal data [1] (points), NRHB, CPA and sPC-SAFT predictions (lines). mental data [1] (points), NRHB, CPA and sPC-SAFT predictions (lines).
 G.M. Kontogeorgis et al. / Fluid Phase Equilibria 296 (2010) 219–229 225

Fig. 9. Propanol monomer fraction in propanol-heptane mixture at three tempera-

Fig. 7. Propanol monomer fraction in 1-propanol-hexane mixture at approximately
tures. Experimental data [2] (points), NRHB, CPA and sPC-SAFT predictions (lines).
296 K. Experimental data [10,11] (points), NRHB, CPA and sPC-SAFT predictions
(lines).
show correct trends with concentration, temperature and alcohol
are for alcohols (M is the molecular weight in g/mol): chain length.

m = 0.0287M + 0.0749 2.2.3. Monomer fraction of alcohol–alkanes

m 3 = 1.6906M + 5.5449 At first, an investigation of the influence of association schemes
(2B, 3B) and in general of the parameter choice will be carried
mε/k = 7.3017M + 91.577 out. A typical example is shown in Fig. 7 and similar results are
εHB = 2811.02K obtained for the other systems. It can be seen that there are dif-
ferences between the models, even within the same equation of
HB = 0.0033 state/associating scheme using different sets of parameters. For
One unique set of associating parameters for all alcohols heavier example, in Fig. 7 the results with sPC-SAFT are based on the 2B
than methanol is also used in the NRHB equation of state [34,35] (i.e. scheme. Still, the performance of sPC-SAFT using parameters from
EH = 2886.7 K and SH /R = 3.308). Alcohol parameters for CPA were Grenner et al. [27] (equations in Section 2.2.2) is better than when
adopted from Kontogeorgis et al. [33]. Two sites on every alcohol’s the parameters from Gross and Sadowski [26] are used. For CPA,
molecule, one proton donor and one proton acceptor (2B), were the two site (2B) set performs better than the 3B set. The same con-
assumed for all models. clusion is obtained for other alcohols as well and it appears that
All the models predict qualitatively correct the dependency of for CPA, the two-site association scheme (2B) is the best choice for
monomer fraction against temperature, while their performance is alcohols heavier than methanol.
similar and somewhat inferior for ethanol but good for propanol. Next, we focus on the comparison between the three association
The worse performance for ethanol could be attributed to the fact equations of state (CPA, sPC-SAFT and NRHB). Some characteristic
that, as discussed in Section 1.1 (see also Fig. 1) the Luck’s monomer results are shown in Figs. 8–10. Alcohol parameters were adopted
fraction data for methanol and ethanol almost coincide. None of the from Grenner et al. [35], Kontogeorgis et al. [33] and Tsivintzelis et
equations of state can exhibit this trend. Instead all models show al. [34], for sPC-SAFT, CPA and NRHB, respectively.
that the monomer fraction increases with increasing chain length, Similar results are obtained for other systems, for which experi-
as would have been expected. In general, CPA, sPC-SAFT and NRHB mental data are available. It can be concluded that all three theories
(CPA, NRHB and sPC-SAFT) predict qualitatively correct the depen-

Fig. 8. 1-butanol monomer fraction in the 1-butanol-hexane mixture at 298 K. Fig. 10. Monomer fraction for hexane–ethanol. Experimental data [10,11] (points),
Experimental data [10] (points), and NRHB, CPA and sPC-SAFT predictions (lines). NRHB, CPA and sPC-SAFT predictions (lines).
226 G.M. Kontogeorgis et al. / Fluid Phase Equilibria 296 (2010) 219–229

Table 5
New sPC-SAFT parameters for water using vapor pressure (P), liquid density () and monomer fraction (X1 ) data in parameter estimation. For vapor pressures and liquid
densities data were used up to Tr = 0.9.

Scheme  (Å) ε (K) m εAB (K) AB

P (%)
 (%)
X1 (%)

2B 2.079 207.55 2.5398 2335.46 0.4327 1.89 2.57 148

3B 1.867 182.13 3.3731 2019.93 0.4440 2.00 3.39 145
4C 3.366 388.51 0.8148 1552.34 0.0184 1.42 4.83 12.5
  X cal −X exp 
%X = 1
n  i X expi  × 100 where X stands for P,  or X1 and n is the number of experimental data points.
i
1

Table 6
New CPA parameters for water using vapor pressure (P), liquid density () and monomer fraction (X1 ) data in parameter estimation. For vapor pressures and liquid densities
data were used up to Tr = 0.9.

Scheme b (L/mol) ˛0 [bar L2 /mol2 ] c1 εAB (K) ˇAB

P (%)
 (%)
X1 (%)

2B 0.01475 2.9813 0.00001 3439.77 0.0130 1.77 1.37 115

3B 0.01414 1.5096 1.5553 1924.58 0.2190 1.60 4.60 266
4C 0.01556 2.2519 0.6108 1700.40 0.0608 1.35 1.47 11.5

% deviations as in Table 5.

dency of monomer fraction against temperature, concentration and vapor pressure (P), liquid density () and monomer fraction data
alcohol chain length. The results shown are with zero interaction (X1 ). The objective function used is:
parameters i.e. they are pure predictions. In many cases, quantita-   Pexp − Pcalc 2   exp − calc 2

tive agreement between theory and experiment is achieved. In all F = v1 + v2
cases, there is excellent qualitative agreement with experimental Pexp exp
data and trends. Moreover, the performance of the three theories  X1,exp − X1,calc
2
against each other is similar. For all three equations of state, 2B is + v3 (13)
X1,exp
the best choice for alcohols heavier than methanol.
Equal weight has been given to all three properties (i.e. vi = 1).
3. Association parameters based on monomer fraction data Various association schemes are considered (2B, 3B, 4C). The water
– water parameters estimated with this procedure for the three association
schemes are shown in Table 5 for sPC-SAFT and Table 6 for CPA,
New parameters have been estimated in this work for CPA and together with the average absolute percentage deviations for the
sPC-SAFT for water and alcohols obtained from simultaneous fit of three properties used in the regression.

Table 7
sPC-SAFT and PC-SAFT parameters for methanol, ethanol, propanol and octanol from this work and literature. Those indicated as this work are based on vapor pressure (P),
liquid density () and monomer fraction (X1 ) data in parameter estimation. Tr -range used is 0.4–0.9 (0.5–0.9 for the Grenner et al. [27] parameters). For ethanol, propanol
and octanol the 2B scheme is used.

Scheme  (Å) ε (K) m εAB (K) AB

Methanol – this work 3.099 225.2 1.8538 2383.1 0.0402

Methanol (Gross and Sadowski [26]) 3.230 188.9 1.5255 2899.5 0.0352
Methanol (von Solms et al. [51]) 2.651 186.6 2.7921 2090.2 0.1460
Methanol-3B – this work (only P,) 2.411 163.2 3.5841 1795.8 0.1715
Methanol-3B – this work 2.805 198.8 2.4573 2009.1 0.0465
Ethanol – this work 3.214 209.2 2.3352 2593.1 0.0267
Ethanol (Gross and Sadowski [26]) 3.177 198.2 2.3827 2653.4 0.0324
Ethanol (Grenner et al. [27]) 4.106 316.9 1.2306 2811.0 0.0033
Propanol – this work 3.782 281.8 1.9722 2722.8 0.0040
Propanol (Gross and Sadowski [26]) 3.252 233.4 2.9997 2276.8 0.0153
Propanol (Grenner et al. [27]) 3.904 292.1 1.7996 2811.0 0.0033
Propanol (Kouskoumvekaki et al. [28]) 3.308 236.3 2.8148 2370.0 0.0146
Octanol – this work 3.796 268.8 4.0898 2737.1 0.0030
Octanol (Gross and Sadowski [26]) 3.714 262.7 4.3555 2754.8 0.0022
Octanol (Grenner et al. [27]) 3.887 273.9 3.8470 2811.0 0.0033

Table 8
CPA parameters for methanol, ethanol, propanol and octanol from this work and literature. Those indicated as this work are based on vapor pressure (P), liquid density ()
and monomer fraction (X1 ) data in parameter estimation as well as literature parameters. The Tr -range used is 0.4–0.9. For ethanol, propanol and octanol the 2B scheme is
used.

Scheme b (L/mol) ˛0 [bar L2 /mol2 ] c1 εAB (K) ˇAB

Methanol 2B (this work) 0.0323 4.3856 0.78647 2459.0 0.03580

Methanol 2B [52] 0.0310 4.0531 0.43102 2957.6 0.01610
Methanol 3B [33] 0.0334 4.5897 1.00676 1932.8 0.03440
Methanol 3B (this work) 0.0331 4.6567 0.75153 2283.1 0.01610
Ethanol [33] 0.0491 8.6716 0.73690 2589.7 0.00800
Ethanol (this work) 0.0469 6.7013 0.79873 2979.4 0.01270
Propanol [33] 0.0641 11.910 0.91709 2525.8 0.00810
Propanol (this work) 0.0642 12.688 0.70518 2877.2 0.00306
Octanol [33] 0.1485 41.582 1.14860 2872.8 0.00014
Octanol this work 0.1490 41.478 1.02375 3018.8 0.00065
 G.M. Kontogeorgis et al. / Fluid Phase Equilibria 296 (2010) 219–229 227

It can be concluded that it is possible, using both models and

the 4C scheme, to obtain good fit for both the “standard” properties
(vapor pressure, liquid density) and the monomer fractions. 4C per-
forms clearly best for all three properties than the other schemes,
while 2B and 3B perform similarly. Especially pronounced is the
improvement using the 4C scheme over the 2B and 3B ones for
temperatures below 500 K. Of the two models (CPA, sPC-SAFT), the
former represents overall the water data better, especially the liq-
uid densities (1.5% vs. about 5%). The superiority of the 4C scheme
for water verified via this investigation using monomer fraction
data is not surprising, although in another recent investigation
which also employs monomer fraction data (Aparicio-Martinez
and Hall [32]) the 2B scheme for water is found to be superior
for both CPA and PC-SAFT, and rather poor results are obtained Fig. 11. Methanol monomer fraction using CPA and sPC-SAFT with parameters fitted
especially for liquid density. In light of our results, the conclu- to vapor pressure, liquid density and monomer fraction data. Experimental data
sions by Aparicio-Martinez and Hall [32], which have also been from Ref. [1].
questioned previously [54], seem erroneous. The 4C scheme for
water is supported by other recent studies as well (Clark et al.
[46], Grenner et al. [29,30]), while some other studies (Dixit et
al. [48], Chatzidimitriou–Deismann [49]) suggest that a number
between three and four sites should be closer to reality. Moreover,
the monomer fraction results with 4C are as good as those by von
Solms et al. [13] but better compared to the results using the Gross
and Sadowski [26] and Grenner et al. [29,30] parameters.
It is worth investigating whether the representation of vapor
pressures and liquid densities is affected substantially by the inclu-
sion of monomer fraction data (X1 ) in the parameter estimation.
When X1 data are not used in parameter estimation, the deviations
in (P, ) reported in literature with PC-SAFT are: 0.9, 2.6 (Grenner et
al. [29,30]), 1.9, 6.8 (Gross and Sadowski [26] 2B), 1.0, 7.7 (Aparicio-
Martinez and Hall [32] 1.0, 7.7 with 4C, 1.0, 6.7 with 2B and 0.9, 6.4
with 3B) and between 0.4 and 0.7, 1.1 and 1.7 (von Solms et al.
[13] various 4C sets). For CPA-4C literature values, the deviation
is 0.8, 0.3. with 3B it is 1.5, 1.3 and 2B: 0.7, 0.4. Thus, it can be
concluded that inclusion of monomer fraction data does not affect
Fig. 12. Ethanol monomer fraction using CPA and sPC-SAFT with parameters fitted
very much the representation of the vapor pressures and liquid
to vapor pressure, liquid density and monomer fraction data [1] and parameters
densities. from the literature.
Finally, a comment on the parameter values for sPC-SAFT/4C:
the segment number value is close to unity which is reasonable,
a constraint optimization has been carried out with maximum devi-
but the dispersion energy and association energy values are rather
ations allowed from the optimization to be less than 1% for vapor
different from those suggested by molecular simulation and spec-
pressure and less than 2.5% in liquid density. Tables 7 and 8 show
troscopic values (around 74–160 K for dispersion, Errington et al.
the parameters of sPC-SAFT and CPA obtained in this work and com-
[50], and 1813 K for the association, Koh et al. [21]).
pared to literature values. Tables 9–12 show deviations in vapor
pressure, liquid density and monomer fraction using new and lit-
4. Association parameters based on monomer fraction data erature parameters for CPA and sPC-SAFT for methanol, ethanol,
– alcohols propanol and octanol. Figs. 11–15 show characteristic results for
monomer fractions with the parameters from this work estimated
A similar approach as for water has been followed for alco- based on all three properties for various alcohols as well as some
hols using the objective function of Eq. (13). In addition, we have alcohol–alkane mixtures.
investigated the effect of the inclusion of enthalpy of vaporiza- We can see that it is possible to obtain good fit for all three prop-
tion data in the parameter estimation as well as the importance erties using a single set of adjustable parameters. The results for
of having different weights for the different properties. Moreover, monomer fraction with the new parameters are substantially better

Table 9
Deviations in vapor pressure, liquid density and monomer fraction for methanol using CPA and sPC-SAFT with literature and new parameters.

Model Scheme Reference

P (%)
 (%)
X1 (%)

CPA 2B [52] 1.4 0.5 45

CPA 2B This work 1.0 0.5 15
sPC-SAFT 2B [26] 2.1 0.5 74
sPC-SAFT 2B [13] 1.2 0.6 20
sPC-SAFT 2B This work 1.0 2.5 16
CPA 3B [33] 0.8 0.8 44
CPA 3B This work 1.0 1.3 15
sPC-SAFT 3B This work (only P, ) 0.5 0.2 26
sPC-SAFT 3B This work 1.0 2.5 18

% deviations as in Table 5.
228 G.M. Kontogeorgis et al. / Fluid Phase Equilibria 296 (2010) 219–229

Table 10
Deviations in vapor pressure, liquid density and monomer fraction for ethanol using
CPA and sPC-SAFT with literature and new parameters. The 2B association scheme
is used.

Model Reference
P (%)
 (%)
X1 (%)

CPA [33] 2.3 0.5 >100

CPA This work 0.3 1.2 13
sPC-SAFT [26] 1.2 0.6 26
sPC-SAFT [27] 1.3 3.4 57
sPC-SAFT This work 0.2 0.7 13

% deviations as in Table 5.

Table 11
Deviations in vapor pressure, liquid density and monomer fraction for propanol
using CPA and sPC-SAFT with literature and new parameters. The 2B association
scheme is used.

Model Reference
P (%)
 (%)
X1 (%)
Fig. 13. Monomer fraction for methanol-hexane with CPA and sPC-SAFT using
CPA [33] 1.3 0.8 19
parameters from this work (based on vapor pressures, liquid densities and monomer
CPA This work 0.3 1.2 13
fraction data). Experimental data from references [10,11].
sPC-SAFT [26] 1.7 1.3 85
sPC-SAFT [28] 1.3 0.9 37
sPC-SAFT [27] 0.7 1.9 18
sPC-SAFT This work 1.0 1.6 3.0

% deviations as in Table 5.

Table 12
Deviations in vapor pressure, liquid density and monomer fraction for octanol using
CPA and sPC-SAFT with literature and new parameters. The 2B association scheme
is used.

Model Reference
P (%)
 (%)
X1 (%)

CPA [33] 1.5 1.0 110

CPA This work 1.0 0.9 11
sPC-SAFT [26] 1.2 0.6 33
sPC-SAFT [27] 1.8 0.2 32
sPC-SAFT This work 1.0 0.1 10

% deviations as in Table 5.

Fig. 14. Monomer fraction for ethanol–hexane with CPA and sPC-SAFT using param- The results are similar for sPC-SAFT and CPA. There are, more-
eters from this work (based on vapor pressures, liquid densities and monomer over, small differences between the 2B and 3B schemes, when
fraction data) and from literature. Experimental data from references [10,11]. parameters are fitted to all three properties, even for methanol.
Two hydrogen bonds per OH group is, moreover, what is found
compared to literature sets which are based solely on vapor pres- recently using molecular simulation studies with suitable transfer-
sures and liquid densities. When a fourth property is included in the able force fields (Ferrando et al. [53]); thus two sites for alcohols
parameter estimation (the enthalpy of vaporization), the results for and four sites for diols. The new parameters perform well also for
the monomer fraction are not as good. monomer fractions of mixtures but the difference from existing
parameters is small, considering the large scatter of experimental
data for mixtures.

5. Conclusions

Association models (CPA, sPC-SAFT and NRHB) can predict

monomer fraction of associating compounds and mixtures with
varying degrees of accuracy. The results are satisfactory in most
cases at least qualitatively correct, considering also the uncertain-
ties in the experimental data, and small differences are observed
between the models.
In addition, it has been shown for CPA and sPC-SAFT that, using a
suitable association scheme, a single set of parameters can describe
satisfactorily vapor pressures, liquid densities and monomer frac-
tions of water and alcohols. The 4C scheme is the best choice for
water, while for methanol there is small difference between the
2B and 3B association schemes. Both pure methanol and water
monomer fraction data used are the 30 years old data by Luck
and it would be useful to see in the future new monomer fraction
Fig. 15. Monomer fraction for propanol-heptane with CPA and sPC-SAFT using
parameters from this work (based on vapor pressures, liquid densities and monomer measurements for these and other hydrogen bonding compounds.
fraction data) and from literature. Experimental data from references [2,11]. Moreover, it remains to be seen whether including monomer frac-
 G.M. Kontogeorgis et al. / Fluid Phase Equilibria 296 (2010) 219–229 229

tion in the parameter estimation will render parameters which [8] D. Missopolinou, K. Ioannou, I. Prinos, C. Panayiotou, Z. Phys. Chem. 216 (2002)
result to improved phase behavior calculations. 1–14.
[9] R.B. Gupta, R.L. Brinkley, AIChE J. 44 (1998) 207–213.
[10] N. Asprion, H. Hasse, G. Maurer, Fluid Phase Equilib. 186 (2001) 1–25 (Supple-
List of symbols mentary material: http://www.itt.uni-stuttgart.de/).
[11] N. von Solms, L. Jensen, J.L. Kofod, M.L. Michelsen, G.M. Kontogeorgis, Fluid
b CPA fluid specific parameter
Phase Equilib. 261 (2007) 272–280.
c1 CPA fluid specific parameter [12] K. Moorthi, I. Nagata, Prop. Fluid Phase Equilib. 63 (1991) 183–210.
EH hydrogen bonding energy [13] N. von Solms, M.L. Michelsen, C.P. Passos, S.O. Derawi, G.M. Kontogeorgis, Ind.
GH Gibbs free energy of association Eng. Chem. Res. 45 (2006) 5368–5374.
[14] H. Kahl, S. Enders, Fluid Phase Equilib. 172 (2000) 27–42.
g() radial distribution function [15] Y.E. Gorbaty, A.G. Kalinichev, J. Phys. Chem. 99 (1995) 5336–5340.
ghs hard sphere radial distribution function [16] Y.E. Gorbaty, R.B. Gupta, Ind. Eng. Chem. Res. 37 (1998) 3026–3035.
k Boltzmann constant [17] G.V. Bondarenko, Y.E. Gorbaty, Mol. Phys. 74 (1991) 639–647;
G.V. Bondarenko, Y.E. Gorbaty, Geochim. Cosmochim. Acta 61 (1997)
K constant 1413–1420.
m number of molecular segments [18] M.M. Hoffmann, M.S. Conradi, J. Am. Chem. Soc. 119 (1997) 3811–3817.
M molecular weight [19] G.E. Walrafen, M.R. Fisher, M.S. Hokmabadi, W.-H. Yang, J. Chem. Phys. 85
(1986) 6970–6982.
P pressure [20] A.G. Kalinichev, J.D. Bass, J. Phys. Chem. A 101 (1997) 9720–9727.
r number of molecular segments [21] C.A. Koh, H. Tanaka, J.M. Walsh, K.E. Gubbins, J.A. Zollweg, Fluid Phase Equilib.
R gas constant 83 (1993) 51–58.
[22] A. Nath, E. Bender, Fluid Phase Equilib. 7 (1981) 275–287.
SH hydrogen bonding entropy [23] B.N. Solomonov, V.B. Novikov, M.A. Varfolomeev, A.E. Klimovitskii, J. Phys. Org.
T temperature Chem. 18 (2005) 1132–1137.
Tc critical temperature [24] G.C. Pimentel, A.L. McClellan, The Hydrogen Bond, W.H. Freeman and Co., San
Francisco, 1960.
Tr reduced temperature
[25] A. Kziazckzac, K. Moorthi, Fluid Phase Equilib. 23 (1985) 153–164.
X1 fraction of monomers [26] J. Gross, G. Sadowski, Ind. Eng. Chem. Res. 41 (2002) 5510–5515.
X Ai fraction of free sites of type A belonging to molecule i [27] A. Grenner, G.M. Kontogeorgis, N. von Solms, M.L. Michelsen, Fluid Phase Equi-
lib. 258 (2007) 83–94.
[28] I.A. Kouskoumvekaki, N. von Solms, M.L. Michelsen, G.M. Kontogeorgis, Fluid
Greek letters Phase Equilib. 215 (2004) 71–78.
 association strength [29] A. Grenner, J. Schmelzer, N. von Solms, G.M. Kontogeorgis, Ind. Eng. Chem. Res.
45 (2006) 8170–8179.
Ai Bj association strength between two sites A and B belonging [30] A. Grenner, G.M. Kontogeorgis, M.L. Michelsen, G.K. Folas, Mol. Phys. 105 (2007)
to molecules i and j, respectively 1797–1801.
˛o fluid specific parameter [31] M. Kleiner, Thermodynamic modeling of complex systems: polar and asso-
ˇAi Bj association volume for the hydrogen bond between sites ciating fluids and mixtures, Ph.D. Thesis, Technical University of Dortmund,
2008.
A and B belonging in molecules i and j, respectively [32] S. Aparicio-Martinez, K.R. Hall, Fluid Phase Equilib. 254 (2007) 112–125.
ε dispersion energy parameter [33] G.M. Kontogeorgis, M.L. Michelsen, G.K. Folas, S. Derawi, N. von Solms, E.H.
εAi Bj association energy for the hydrogen bond between sites Stenby, Ind. Eng. Chem. Res. 45 (2006) 4855–4868.
[34] I. Tsivintzelis, A. Grenner, I.G. Economou, G.M. Kontogeorgis, Ind. Eng. Chem.
A and B belonging to molecules i and j, respectively Res. 47 (2008) 5651–5659;
εHB association energy I. Tsivintzelis, A. Grenner, I.G. Economou, G.M. Kontogeorgis, Ind. Eng. Chem.
, AB association volume Res. 48 (2009) 7860.
[35] A. Grenner, I. Tsivintzelis, G.M. Kontogeorgis, I.G. Economou, C. Panayiotou, Ind.
H number of hydrogen bonds per molecular segment Eng. Chem. Res. 47 (2008) 5636–5650.
 density [36] H. Huang, M. Radosz, Ind. Eng. Chem. Res. 29 (1990) 2284–2294.
 segment diameter [37] J. Gross, G. Sadowski, Ind. Eng. Chem. Res. 40 (2001) 1244–1260.
[38] C. Panayiotou, I.C. Sanchez, J. Phys. Chem. 95 (1991) 10090–10097.
* characteristic segmental volume [39] C. Panayiotou, I. Tsivintzelis, I.G. Economou, Ind. Eng. Chem. Res. 46 (2007)
2628–2636.
Acknowledgments [40] I.G. Economou, M.D. Donohue, AIChE J. 37 (1991) 1875–1894.
[41] W. Peschel, H. Wenzel, Phys. Chem. Chem. Phys. 88 (1984) 807–812.
[42] I.V. Prikhodko, T.W. de Loos, A.I. Victorov, Int. J. Thermophys. 16 (1995)
The authors greatly appreciate the support and advice of Pro- 1287–1297.
fessor Michael L. Michelsen. Also, they wish to thank Statoil and [43] A. Deak, A.I. Victorov, T.W. de Loos, Fluid Phase Equilib. 107 (1995) 277–301.
[44] R.B. Gupta, C. Panayiotou, I.C. Sanchez, K.P. Johnston, AIChE J. 38 (1992)
Gassco (Norway), BP (UK, USA), TOTAL (France), DONG Energy and 1243–1253.
Mærsk Oil and Gas (Denmark) for supporting this work as part of [45] P. Muthukumaran, R.L. Brinkley, R.B. Gupta, AIChE J. 48 (2002) 386–392.
the CHIGP project (Chemicals in Gas Processing). [46] G.N.I. Clark, A.J. Haslam, A. Galindo, G. Jackson, Mol. Phys. 104 (2006)
3561–3581.
[47] P.J. Smits, I.G. Economou, C.J. Peters, J. de Swaan Arons, J. Phys. Chem. 98 (1994)
References 12080–12085.
[48] S. Dixit, J. Crain, W.C.K. Poon, J.L. Finney, A.K. Soper, Nature 416 (2002) 829–832.
[1] W.A.P. Luck, Angew. Chem. Int. Ed. 19 (1980) 28–41 (in English). [49] A.C. Chatzidimitriou-Dreismann, Nachr. Chim. 52 (2004) 773–776.
[2] T.R. Lien, A Study of the thermodynamic excess functions of alcohol solutions [50] J.R. Errington, G.C. Boulougouris, I.G. Economou, A.Z. Panagiotopoulos, D.N.
by IR spectroscopy, Applications to Chemical Solution Theory, PhD Thesis, Uni- Theodorou, J. Phys. Chem. B 102 (1998) 8865–8873.
versity of Toronto, 1972. [51] N. von Solms, M.L. Michelsen, G.M. Kontogeorgis, Ind. Eng. Chem. Res. 42 (2003)
[3] A.N. Fletcher, C.A. Heller, J. Phys. Chem. 71 (1967) 3742–3756. 1098–1105.
[4] F. Palombo, P. Sassi, M. Paolantoni, A. Morresi, R.S. Cataliotti, J. Phys. Chem. B [52] G.M. Kontogeorgis, I.V. Yakoumis, H. Meijer, E.M. Hendriks, T. Moorwood, Fluid
110 (2006) 18017–18025. Phase Equilib. 158–160 (1999) 201–209.
[5] S. Martinez, Spectrochim. Acta Part A: Mol. Spectrosc. 42 (1986) 531–536. [53] N. Ferrando, V. Lachet, J.M. Teuler, A. Boutin, J. Phys. Chem. B 113 (2009)
[6] J.M. Prausnitz, R.N. Lichtenthaler, E.G. de Azevedo, Molecular Thermodynamics 5985–5995.
of Fluid-Phase Equilibria, Third ed., Prentice Hall International, New Jersey, [54] M.C. dos Ramos, K.D. Goff, H. Zhao, C. McCabe, J. Phys. Chem. B 112 (2008)
1999. 9417–9427.
[7] R.L. Brinkley, R.B. Gupta, AIChE J. 47 (2001) 948–953.