16654

J. Phys. Chem. B 2008, 112, 1665416663

Physical and Chemical Absorptions of Carbon Dioxide in Room-Temperature Ionic Liquids

A. Yokozeki,*, Mark B. Shiflett, Christopher P. Junk, Liane M. Grieco, and Thomas Foo
DuPont Fluoroproducts Laboratory, Chestnut Run Plaza 711, Wilmington, Delaware 19880 and DuPont
Central Research and DeVelopment, Experimental Station, Wilmington, Delaware 19880
ReceiVed: June 16, 2008; ReVised Manuscript ReceiVed: October 14, 2008

Gaseous solubilities of carbon dioxide (CO2) in 18 room-temperature ionic liquids (RTILs) have been measured
at an isothermal condition (about 298 K) using a gravimetric microbalance. The observed pressure-temperaturecomposition (PTx) data have been analyzed by use of an equation-of-state (EOS) model, which has been successfully
applied for our previous works. Henrys law constants have been obtained from the observed (PTx) data directly
and/or from the EOS correlation. Ten RTILs among the present ionic liquids results in the physical absorption,
and eight RTILs show the chemical absorption. The classification of whether the absorption is the physical or
chemical type is based on the excess Gibbs and enthalpy functions as well as the magnitude of the Henrys constant.
In the chemical absorption cases, the ideal association model has been applied in order to interpret those excess
thermodynamic functions. Then, two types of the chemical associations (AB and AB2, where A is CO2 and B is
RTIL) have been observed with the heat of complex formations of about -11 (for AB) and from -27 to -37 (for
AB2) kJ mol-1, respectively.
Introduction
For the past several years, worldwide research on thermodynamic and transport properties of room-temperature ionic
liquids (RTILs) and their mixtures with various chemicals has
been conducted for possible applications of this new class of
compounds.1-3 Among others, one of the promising applications
with RTILs is a nonvolatile (so-called green) solvent for
capturing unwanted compounds such as CO2, SO2, H2S, etc. in
the exhaust gas stream of power plants. Although there are
commercial organic solvents to capture these flue gases at the
present, new RTIL solvents may provide viable and more
environmentally friendly alternatives. The capture and sequestration of carbon dioxide in the power plant are urgently needed
in order to reduce anthropogenic CO2 accumulations in the
atmosphere.4
As often stated elsewhere, effectively capturing acid or sour
gases from exhaust gases requires very strong absorption
because of the relatively small partial pressures (e.g., 5-15 %
v/v of CO2 at atmospheric pressure) of these gases in the gas
stream. By strong absorption, we mean that the gas absorption
needs to be chemical absorption (or reversible chemical
complex formation), instead of the simple physical absorption
(or no chemical reactions), which would be practical for highpressure gas absorption (e.g., CO2 partial pressure >525 kPa 5).
It is well-known that CO2 possesses relatively high solubility
in RTILs; for example, compared with the case of hydrocarbons.6 However, most reported cases for CO2 seem to be the
physical absorption.6-17 To the best of our knowledge, a very
high solubility of CO2 in RTIL with the chemical absorption is
only observed in the case of the CO2 + [bmim][Ac] system.4,18,19
In this report, we have examined CO2 solubilities in 18 RTILs
and identified seven more chemical absorption cases in addition
to the known [bmim][Ac] system.
The RTILs in this study are nine commercially available
RTILs: 1-ethyl-3-methylimidazolium trifluoroacetate [emim][T* To whom correspondence should be addressed. E-mail: akimichi.
yokozeki@usa.dupont.com.

DuPont Fluoroproducts Laboratory.

DuPont Central Research and Development.

FA], 1-ethyl-3-methylimidazolium acetate [emim][Ac], 1-butyl3-methylimidazolium trifluoroacetate [bmim][TFA], 1-butyl-3methylimidazolium acetate [bmim][Ac], 1-ethyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide [emim][Tf2N],
1-butyl-3-methylimidazolium hexafluorophosphate [bmim][PF6],
1-butyl-3-methylimidazolium tetrafluoroborate [bmim][BF4],
1-hexyl-3-methylimidazolium tris(pentafluoroethyl)trifluorophosphate [hmim][FAP], and 1-hexyl-3-methylimidazolium
bis(trifluoromethylsulfonyl)imide [hmim][Tf2N]. In addition,
nine new RTILs were synthesized to evaluate the interaction
with carbon dioxide: 1-butyl-3-methylimidazolium 1,1,2,2tetrafluoroethanesulfonate [bmim][TFES], 1-butyl-3-methylimidazolium propionate [bmim][PRO], 1-butyl-3-methylimidazolium isobutyrate [bmim][ISB], 1-butyl-3-methylimidazolium
trimethylacetate [bmim][TMA], tetrabutylphosphonium formate
[TBP][FOR], 1-butyl-3-methylimidazolium levulinate [bmim][LEV], 1-butyl-3-methylimidazolium succinamate [bmim][SUC], bis(1-butyl-3-methylimidazolium) iminodiacetate [bmim]2[IDA], and 1-butyl-3-methylimidazolium iminoacetic acid acetate
[bmim][IAAc].
Figure 1 provides the chemical name, abbreviation, and
structure of the 18 ionic liquids studied. The binary systems of
CO2 + [bmim][Ac], [bmim][PF6], [bmim][BF4], and [hmim][Tf2N] have been studied previously by us,8,10,19 but solubility
measurements have been repeated here in order to compare the
solubility with the present new RTILs at the same experimental
condition. Solubility (PTx: pressure-temperature-composition)
data have been collected at an isothermal condition of about
298 K, using a gravimetric microbalance method.10 The observed
PTx data show both physical (10 systems) and chemical (8
systems) absorption cases and have been successfully correlated
with an equation-of-state (EOS) model.10,20-22 These solubility
behaviors have been further examined using the thermodynamic
excess functions, as well as Henrys law constants. In the strong
absorption cases, the excess functions from the EOS correlation
are interpreted in terms of a chemical association model.23-28
Experimental Details and Results
Materials. Carbon dioxide (purity >99.99%, CAS No.
124-38-9) was purchased from MG Industries (Philadelphia,

10.1021/jp805784u CCC: $40.75 2008 American Chemical Society

Published on Web 12/03/2008

Absorption of Carbon Dioxide in Ionic Liquids

J. Phys. Chem. B, Vol. 112, No. 51, 2008 16655

Figure 1. Chemical structures of the present ionic liquids and their abbreviations.

PA). The [emim][Ac] (assay g95%, C8H14N2O2, CAS No.

143314-17-4, Lot and Filling code S25819 14804B38),
[bmim][Ac] (assay g95%, C10H18N2O2, CAS No. 284049-75-8,
Lot and Filling code S25803 444041302), [bmim][PF6] (assay
g96%, C8H15F6N2P, Lot and Filling code 1242554 15005226,
CAS registry No. 174501-64-5), [bmim][BF4] (assay g97%,
C8H15F4N2B, Lot and Filling code 1142017 31205158, CAS
registry No. 174501-65-6) were obtained from Fluka/Aldrich
(Buchs, Switzerland). The [emim][TFA] (assay g95%,
C8H11F3N2O2, CAS No. 174899-65-1, Lot S4934747 814),
[bmim][TFA] (assay g95%, C10H15F3N2O2, CAS No. 17489994-6, Lot EQ508558 616), [hmim][Tf2N] (assay g99%,
C12H19N3F6O4S2, CAS No. 382150-50-7, Lot EQ500831 642),
and [hmim][FAP] (assay g99%, C16H19N2F18P, CAS No.
713512-19-7, Lot S4872378-733) were obtained from EMD
Chemicals, Inc. (Gibbstown, New Jersey). The ionic liquid
[emim][Tf2N] (EMIIm, electrochemical grade, assay g99.5%,
C8H11F6N3O4S2, Lot and Catalog No. 259095 IL-201-20-E,
CAS registry No. 174899-82-2) was purchased from Covalent
Associates Inc. (Woburn, MA).

Nine RTILs were prepared according to the methods described in the Supporting Information of this article. The cation
containing precursor salts (purity >99%) were obtained from
Fluka/Aldrich and were used to synthesize the corresponding
RTILs using available anion precursors for this study. The
molecular structure was verified by nuclear magnetic resonance
(NMR). 19F NMR and 1H NMR spectra were recorded on a
Bruker model DRX-400 spectrometer at 376.8937 and 400.550
MHz, respectively. The purity of all RTILs synthesized in the
present work was estimated to be at least better than 95%.
Experimental Method and Results. Both the commercially
available and synthesized ionic liquids were dried by filling a
borosilicate glass tube with about 5-10 g of the ionic liquid and
pulling a coarse vacuum with a diaphragm pump (Pfeiffer, model
MVP055-3, Nashua, NH) followed by further evacuation using
a turbopump (Pfeiffer, model TSH-071) to a pressure of about 4
10-7 kPa while simultaneously heating and stirring the ionic
liquid at a temperature of about 348 K for 5 days. The final water
content was measured by Karl Fischer titration (Aqua-Star C3000,
solutions AquaStar Coulomat C and A) and all ionic liquids

16656 J. Phys. Chem. B, Vol. 112, No. 51, 2008

contained less than 1 10-3 mass fraction of water. This water
content in mass fraction translates the purity of all RTILs into better
than about 98 mol %, after the evacuation treatment, and any other
volatile impurities should have been removed in this treatment.
This purity level of RTILs is sufficient for the present purpose
(i.e., CO2 solubility measurements), although other measurements
such as viscosity, thermal and electric conductivity, etc., may
require much higher purities.
The liquid density for each RTIL was measured at 298.15 K,
and several samples were measured at higher temperatures (323.15,
348.15, and 373.15 K) using a pycnometer (Micromeritics AccuPyc
1330 with a 1 cm3 measuring cup). The uncertainty in the density
measurement was ( 0.001 g cm-3. The ionic liquids were loaded
into the pycnometer in a nitrogen-purged glovebox to minimize
exposure to moisture. The density data are provided in Table S1
of the Supporting Information.
Detailed descriptions of experimental equipment and procedures for the VLE are given in our previous reports;8,10,20
therefore, only the basic experimental techniques and measurement uncertainties are presented here.
The gas solubility (VLE) measurements were made using a
gravimetric microbalance10 (Hiden Isochema Ltd., IGA 003,
Warrington, United Kingdom). A molecular sieve trap was installed
to remove trace amounts of water from the CO2. Initially, about
65 mg of the RTIL was loaded into the sample container and heated
to 348.15 K under a vacuum of about 10-9 MPa for 10 h to remove
any trace amounts of water or other volatile impurities. The sample
temperature was measured with a type K thermocouple with an
accuracy of ( 0.1 K. The thermocouple was calibrated using a
standard platinum resistance thermometer (SPRT model 5699, Hart
Scientific, American Fork, UT, range 73-933 K) and readout
(Blackstack model 1560 with SPRT module 2560). The Blackstack
instrument and SPRT are a certified secondary temperature standard
with a NIST traceable accuracy to (0.005 K. A single isotherm
at about 298.15 K was measured for each RTIL. Pressures from
10-2 to 2.0 MPa were measured using a piezo-resistive strain gauge
(Druck, model PDCR4010) with an accuracy of (0.8 kPa. The
Druck pressure transducer was calibrated against a Paroscientific
Model 760-6K (Redmond, WA) pressure transducer (range
0-41.5 MPa, serial No. 62724). This instrument is also a NIST
certified secondary pressure standard with a traceable accuracy of
0.008% of full scale. The upper pressure limit of the microbalance
reactor was 2.0 MPa, and several isobars up to 2.0 MPa (0.01,
0.05, 0.1, 0.4, 0.7, 1.0, 1.3, 1.5, and 2.0 MPa) were measured in
the present study. In our previous reports,8,10 to ensure sufficient
time for VLE, each T,P condition was maintained for a minimum
of 3 h with a maximum time of 8 h. In this work, we found that
8 h was not sufficient to reach equilibrium at the present isothermal
condition (about 298 K). Therefore, a maximum of 20 h was set
for all sample measurements.
The instrumental uncertainties in T and P are within ( 0.1 K
and ( 0.8 kPa, respectively. These uncertainties do not cause any
significant errors in the gas solubility measurement. The total
uncertainties in the solubility data due to both random and
systematic errors have been estimated to be less than 0.006 mole
fraction at given T and P. Another large source of uncertainty in
the present solubility experiments is due to the buoyancy correction
in the data analysis. Analysis of the buoyancy effects requires an
accurate measurement of the liquid density and CO2 gas density,10
which was calculated from a highly accurate correlation.29 Liquid
density data of the present RTILs in Table S1 (Supporting
Information) were used to obtain the corrected solubility data. The
present experimental solubility (PTx) data are summarized in Tables
1-3. Observed isothermal PTx data at about 298.15 K will be

Yokozeki et al.
TABLE 1: Experimental Solubility (PTx) Data for CO2 and
Ionic Liquids
CO2 (1) + [emim][TFA] (2)

CO2 (1) + [bmim][TFA] (2)

T/K

P/MPa

100x1

T/K

P/MPa

100x1

298.1
298.1
298.1
298.1
298.1
298.1
298.1
298.1
298.1

0.0100
0.0498
0.0998
0.3999
0.6997
0.9999
1.2997
1.4998
1.9996

0.1
0.9
1.8
6.8
11.5
16.0
20.1
22.6
28.2

298.1
298.1
298.1
298.0
298.2
298.1
298.1
298.1
298.1

0.0105
0.0504
0.1001
0.3999
0.6992
0.9996
1.3004
1.4996
1.9996

0.5
1.3
2.2
7.5
12.3
17.2
21.3
23.9
30.1

CO2 (1) + [emim][Ac] (2)

CO2 (1) + [bmim][Ac] (2)

T/K

P/MPa

100x1

T/K

P/MPa

100x1

298.1
298.1
298.1
298.1
298.1
298.1
298.1
298.1
298.1

0.0100
0.0499
0.1000
0.3996
0.6995
0.9996
1.2998
1.4997
1.9998

18.9
24.6
26.7
31.3
34.0
36.2
38.4
39.8
42.8

298.1
298.0
298.2
298.1
298.1
298.1
298.3
298.1
298.1

0.0102
0.0503
0.1003
0.3994
0.7001
0.9996
1.3002
1.5000
1.9994

16.2
25.1
27.5
32.6
35.7
38.3
40.6
42.0
45.5

CO2 (1) + [emim][Tf2N] (2)

CO2 (1) + [hmim][Tf2N] (2)

T/K

P/MPa

100x1

T/K

P/MPa

100x1

298.2
298.1
298.0
297.9
298.2
298.2
298.1
298.1
298.1

0.0100
0.0500
0.0998
0.3998
0.6998
0.9997
1.2999
1.4995
1.9998

0.4
1.5
3.0
10.7
17.4
23.4
28.7
31.7
39.0

297.4
297.4
297.3
297.4
297.3
297.4
297.4
297.3
297.3

0.0091
0.0487
0.0983
0.3943
0.6922
0.9890
1.2848
1.4820
1.9748

0.6
2.0
3.7
12.6
20.3
27.0
32.6
36.0
43.3

analyzed in the following subsections using our equation-of-state

(EOS) model.
Data Analyses
Equation-Of-State (EOS) Model. As used in our previous
works, we have employed a generic RK (Redlich-Kwong) type
of cubic EOS, which is written in the following form:10,20-22

P)

RT
a(T)
V-b
V(V + b)

a(T) ) 0.427480

(1)

R2T2c
R(T)
Pc

(2)

RTc
Pc

(3)

b ) 0.08664

The temperature-dependent part of the a parameter in the

EOS for pure compounds is modeled by the following
empirical form:
e3

R(T) )

k(1/Tr - Tr)k, (Tr T/Tc)

k)0

(4)

Absorption of Carbon Dioxide in Ionic Liquids

J. Phys. Chem. B, Vol. 112, No. 51, 2008 16657

TABLE 2: Experimental Solubility (PTx) Data for CO2 and

Ionic Liquids

TABLE 3: Experimental Solubility (PTx) Data for CO2 and

Ionic Liquids

CO2 (1) + [bmim][PF6] (2)

CO2 (1) + [bmim][BF4] (2)

CO2 (1) + [hmim][FAP] (2)

T/K

P/MPa

100x1

T/K

P/MPa

100x1

T/K

P/MPa

100x1

T/K

P/MPa

100x1

298.0
298.1
298.2
298.0
298.0
298.1
298.1
298.1
298.1

0.0105
0.0504
0.1004
0.3998
0.7003
0.9997
1.2996
1.5003
1.9997

0.1
0.9
1.9
7.0
11.8
16.1
20.2
22.8
28.6

298.2
298.2
298.1
298.2
298.2
298.0
298.2
298.0
298.1

0.0100
0.0500
0.1001
0.3996
0.7002
0.9997
1.3002
1.5001
2.0002

<0.1
1.0
1.9
6.9
11.5
15.8
19.7
22.1
27.7

298.1
298.1
298.1
298.1
298.1
298.1
298.1
298.1
298.1

0.0101
0.0501
0.1001
0.3998
0.6996
0.9996
1.2997
1.4997
1.9998

0.8
2.5
4.6
15.0
23.6
31.3
37.6
41.3
49.3

298.0
298.1
298.1
298.2
298.2
298.2
297.9
298.1
298.0

0.0101
0.0501
0.1000
0.3998
0.6997
0.9999
1.2997
1.4995
1.9998

<0.1
0.8
1.7
6.8
11.4
15.8
19.7
22.2
28.5

CO2 (1) + [bmim][PRO] (2)

CO2 (1) + [bmim][ISB] (2)

T/K

P/MPa

100x1

T/K

P/MPa

100x1

298.1
298.1
298.2
298.2
298.1
298.1
298.2
298.2
298.2

0.0103
0.0503
0.1001
0.3994
0.6999
1.0001
1.2994
1.4997
1.9990

13.4
19.7
21.8
26.6
29.6
32.1
34.4
36.0
39.3

298.2
298.1
298.2
298.2
298.1
298.1
298.2
298.2
298.2

0.0102
0.0502
0.0998
0.4002
0.6998
1.0011
1.2996
1.4993
1.9999

12.3
19.6
21.8
26.7
29.9
32.5
34.9
36.5
40.3

CO2 (1) + [bmim][TMA] (2)

CO2 (1) + [TBP][FOR] (2)

T/K

T/K

P/MPa

298.1
298.1
298.1
298.1
298.1
298.1
298.1
298.1
298.1

0.0101
0.0497
0.0996
0.3999
0.6997
0.9995
1.2997
1.4999
2.0000

100x1
8.5
19.0
23.4
29.1
32.3
35.1
37.7
39.3
43.1

298.0
298.2
298.1
298.0
298.1
298.2
298.1
298.0
298.1

P/MPa
0.0100
0.0502
0.1004
0.4004
0.7004
0.9999
1.2997
1.4993
1.9996

aiajfij(T)(1 - kij)xixj, ai ) 0.427480

i,j)1

kij )

lijlji(xi + xj)
, where kii ) 0
ljixi + lijxj

100x1

T/K

P/MPa

100x1

298.1
298.1
298.1
298.1
298.1
298.1
298.1
298.1
298.1

0.0101
0.0501
0.1002
0.3996
0.6998
0.9997
1.2995
1.5001
1.9998

11.8
21.2
24.5
31.1
34.9
38.0
40.7
42.7
46.0

298.1
298.1
298.1
298.1
298.1
298.1
298.1
298.1
298.1

0.0100
0.0501
0.1002
0.3997
0.6998
0.9997
1.2998
1.4997
1.9997

<0.1
1.2
2.3
5.1
8.6
12.4
16.1
18.9
23.2

0.8
2.4
3.4
9.5
15.1
20.3
25.1
28.0
34.8

100x1

T/K

P/MPa

100x1

298.1
298.1
298.1
298.1
298.1
298.1
298.1
298.1
298.1

0.0101
0.0502
0.1002
0.3996
0.6998
0.9998
1.2995
1.4999
1.9997

2.1
5.0
7.7
14.0
19.8
25.6
30.5
34.4
39.5

298.1
298.1
298.1
298.1
298.1
298.1
298.1
298.1
298.1

0.0101
0.0502
0.1003
0.3997
0.7002
0.9996
1.2998
1.4998
1.9996

0.1
0.4
0.6
2.1
5.1
7.5
10.9
14.0
19.1

(6)

(7)

where mij ) mji, mii ) 0.

Tci: critical temperature of the ith species.

P/MPa

P/MPa

1
(b + bj)(1 - kij)(1 - mij)xixj, bi )
2 i,j)1 i
0.08664

T/K

T/K

b)

CO2 (1) + [bmim][SUC] (2)

CO2 (1) + [bmim]2[IDA] (2)

R2Tci2
R (T)
Pci i
(5)

fij(T) ) 1 + ij /T, where ij ) ji and ii ) 0

CO2 (1) + [bmim][LEV] (2)

100x1

The EOS constants for pure CO2 were taken from our
previous work,8 whereas Tc and Pc of RTILs were estimated
using Veteres method,30 assuming the critical compressibility
factor of 0.256. 1 of RTILs was determined through the
binary VLE data analysis with 0 ) 1 and 2 ) 3 ) 0, as
described in our previous reports.10,20-22
Then, the a and b parameters for general N-component
mixtures are modeled in terms of only binary interaction
parameters.

a)

CO2 (1) + [bmim][TFES] (2)

RTci
(8)
Pci

CO2 (1) + [bmim][IAAc] (2)

Pci: critical pressure of the ith species.

R: universal gas constant.
xi: mole fraction of the ith species.
The fugacity coefficient i of ith species for the present EOS
model, which is required for the phase equilibrium calculation,
is given in Appendix A, together with other EOS-derived
properties relevant to the present study.
In the above model, there are a maximum of four binary
interaction parameters: lij, lji, mij, and ij for each binary pair.
Optimal binary interaction parameters determined using a
nonlinear least-squares method for the present systems are
shown in Table 4, together with the EOS constants for pure
compounds. Observed isothermal Px plots show essentially three
types, which are illustrated in Figure 2 using the three typical
binary systems: CO2 + [emim][TFA] (Case A), CO2 +
[hmim][FAP] (Case B), and CO2 + [emim][Ac] (Case C) binary
systems. The solubility behavior of Case A is close to the case
of Raoults law (or positive deviations from Raoults law). Cases
B and C show large negative deviations from Raoults law, and
particularly in Case C, the negative deviation is so large that
the high pressure of CO2 gas becomes practically zero in the
CO2 mole fraction range of less than about 0.3. Highly negative
deviations from Raoults law suggest the chemical absorption
instead of ordinary physical absorption, i.e., chemical complex
formation.
These solubility behaviors are more clearly understood in
terms of the thermodynamic excess functions (excess Gibbs GE,
excess enthalpy HE, and excess entropy (times T) TSE energies);

16658 J. Phys. Chem. B, Vol. 112, No. 51, 2008

Yokozeki et al.

TABLE 4: EOS Constants for CO2 (1) + RTIL (2) Systems

in the Present Worka
molar
ionic liquid mass/g mol-1 Tc/K Pc/ kPa
[emim][Ac]
[emim][TFA]
[emim][Tf2N]
[hmim][FAP]
[bmim][TMA]
[bmim][IBS]
[bmim][PRO]
[bmim][PF6]
[TBP][FOR]
[bmim][TFA]
[bmim][TFES]
[bmim][Ac]
[hmim][Tf2N]
[bmim][BF4]
[bmim][SUC]
[bmim]2[IDA]
[bmim][IAAc]
[bmim][LEV]

170.11
224.18
391.31
612.28
240.34
226.32
212.29
284.18
303.44
252.23
320.30
198.26
447.42
226.02
255.31
409.52
271.31
254.33

871.3
973.8
808.8
847.9
1051.6
1030.8
1041.3
950.2
933.6
931.4
956.7
867.7
815.0
894.9
981.8
1008.0
931.8
892.3

3595
3535
2028
1393
2822
3033
3381
2645
1830
2839
2480
2942
1611
3019
2769
1645
2324
2479

l12b

l21b

1.05000 0.20430 0.24890

0.49389 0.00493 0.01054
0.89924 0.01936 0.01936
0.73769 0.05341 0.02648
0.99219 0.09990 0.18208
0.89936 0.09794 0.22365
0.72646 0.11467 0.31106
0.96012 0.01049 0.01049
1.11575 0.01138 0.01138
0.73908 0.01132 0.01132
0.79838 0.00129 0.11129
1.30810 0.18069 0.23591
0.89924 0.02323 0.02323
0.82130 0.01343 0.01343
0.90054 0.00504 0.00504
1.19556 0.02244 0.02244
0.51458 -0.05672 -0.05672
1.40421 0.12700 0.16278

a
Pure CO2 EOS constants; see ref 8. 0 ) 1 and 2 ) 3 ) 0 for
all RTILs. b Binary interaction parameters, where m12 ) m21 ) 12
) 21 ) 0.

Figure 2. Isothermal Px phase diagram at about T ) 298 K. Three

typical cases (Cases A, B, and C) are shown. Solid lines: calculated
from the present EOS model. Symbols: the present experimental data,
circles ) [emim][TFA], triangles ) [hmim][FAP], and squares )
[emim][Ac]. Dotted line: the Raoults law solubility.

see Appendix A for these functions from the present EOS.

Excess energies of Figure 3, panels a-c, shows the corresponding Cases A, B, and C in Figure 2, at T ) 298.15 K and P )
6.5 MPa. Sufficiently negative values in GE usually indicate
some chemical complex formations; the heat of mixing (or HE)
is more negative than TSE. A minimum in GE occurs around 50
mol % of the CO2 + [hmim][FAP] system (Figure 3b),
suggesting the 1:1 complex formation, whereas a GE minimum
around 33 mol % of CO2 in Figure 3c indicates the 1:2 (CO2:
[emim][Ac]) complex formation.
Plots of the thermodynamic excess functions calculated from
the present EOS are given in the Supporting Information for
all other present binary systems. Binary systems that belong to
Case C are with ionic liquids of [emim][Ac], [bmim][Ac],
[bmim][PRO], [bmim][IBS], [bmim][TMA], and [bmim][LEV].
The Case B is observed in CO2 + [hmim][FAP] and CO2 +
[bmim]2[IDA] systems, and the rest of the other systems belong
to Case A (physical absorption). It is important here to mention
that all cases in the present binary systems show the liquid-liquid
separation (immiscibility) at high CO2 concentrations (higher
than about 70 mol %), including even the highly soluble cases

Figure 3. Examples of thermodynamic excess functions calculated

from the present EOS model at T ) 298.15 K and P ) 6.5 MPa. (a)
CO2 + [emim][TFA], (b) CO2 + [hmim][FAP], and (c) CO2 +
[emim][Ac]. Similar plots for other binary systems are shown in the
Supporting Information of this article.

such as Case C, as demonstrated in our earlier works.8,19 The

highly unsymmetric shape of the excess functions, particularly
with sufficiently large positive values in GE at the high CO2
concentration, is indicative for such unusual phase behaviors;
see Figure 3 and Figures S1-S11 of the Supporting Information.
Infinite dilution properties such as Henrys law constant, kH,
are often useful for a quick assessment for the degree of gas
solubility in a solvent. Gaseous solubility (PTx) data provide
kH by a simple (conventional) analysis; see Appendix B.
Uncertainties attached to such an analysis are, however, often
quite large, depending upon the assumed polynomial form (first,
second, or third order, etc.) for the fugacity function and/or the

Absorption of Carbon Dioxide in Ionic Liquids

J. Phys. Chem. B, Vol. 112, No. 51, 2008 16659

TABLE 5: Equilibrium Constant (Ki), Enthalpy (Hi) of Complex Formation, and Henrys Constant (kH) for the binary CO2
system with various RTILs at 298.15 K
system with RTIL
[bmim][IAAc]
[bmim][SUC]
[bmim][TFES]
[bmim][BF4]
[bmim][PF6]
[emim][TFA]
[bmim][TFA]
[TBP][FOR]
[emim][Tf2N]
[hmim][Tf2N]
[hmim][FAP]
[bmim]2[IDA]
[emim][Ac]
[bmim][Ac]
[bmim][PRO]
[bmim][IBS]
[bmim][TMA]
[bmim][LEV]
a

K1

K2

physical
physical
physical
physical
physical
physical
physical
physical
physical
physical
0
0.827 (
220.3 (
222.0 (
195.5 (
186.7 (
185.7 (
219.1 (

physical
physical
physical
physical
physical
physical
physical
physical
physical
physical
1.007 ( 0.042
0
0
0
0
0
0
0

H1/kJ mol-1

-10.7 ( 0.75
0.014
13.4
6.1
9.8
21.9
23.3
5.6

H2/kJ mol-1

-28.14 ( 7.83
-30.81 ( 0.46
-37.49 ( 0.54
-30.01 ( 0.69
-26.93 ( 0.53
-28.42 ( 0.56
-28.45 ( 0.52

kH/MPa
15.1
7.08
6.09
5.29
5.26
5.20
4.94
3.80
3.18
2.57
2.28
2.10
5.12
2.22
2.29
1.37
2.86
9.93

10-3
10-3
10-4
10-3
10-3
10-3

Physical absorption.

inclusion of the theoretical origin as a data set. Furthermore, in

the present Case C mixtures, such a conventional method cannot
be applied, because any simple polynomials do not exist in
fitting the fugacity function properly as one might guess easily
from Figure 2 (Case C). We can only say qualitatively that kH
, 1 MPa in such a case. Another method is to use the EOS
correlation as described in Appendix B. Results of kH values
for the present systems are obtained from the EOS method and
are shown in the last column of Table 5; kH values from the
conventional method are summarized in Table S3 in the
Supporting Information.
Ideal Association Model. In the previous section we have
observed highly nonideal phase behaviors for some of the binary
systems, and those behaviors have been successfully correlated
with our EOS model. Here, in order to understand the physical
meaning of such highly nonideal phase behaviors (for the binary
systems of Cases B and C in the previous section), we interpret
them as a consequence of chemical absorptions.23-28 Let us
consider the following two types of chemical associations (or
complex formations) in a liquid solution of species A and B:

A + B ) AB, with an equilibrium constant K1

(9)

A + 2B ) AB2, with an equilibrium constant K2 (10)

Thus, in this solution there exist four species A, B, AB, and
AB2. When we assume that these species form an ideal solution,
the following thermodynamic excess functions can be derived:23,24

1 - zB
zB
GE
+ xB ln
) (1 - xB) ln
2
RT
xB
(1 - xB)(1 + K1zB + K2zB)
(11)
(1 - xB)zB(K1H1 + K2H2zB)
HE
)
RT
RT(1 + K1zB + K2zB2 )

(12)

where xB ) the stoichiometric (or feed) mole fraction of species

B, zB ) the true (or actually existing) mole fraction of B in the
solution, and H1 and H2 are the heats of complex formation

for the association reactions of eqs 9 and 10, respectively. The

term xB is related to zB by:23,24

xB )

(1 + K1)zB + K2zB2 (1 - zB)

1 + K1zB(2 - zB) + K2zB2 (3 - 2zB)

(13)

Using the present EOS, we can calculate the excess functions,

GE and HE (eq A17, or eqs A16 and A14 in Appendix A), at
given TPx. Here, we designate species A for CO2 and B for an
ionic liquid. The temperature is taken as 298.15 K, and P )
6.5 MPa, which is close to the vapor pressure of CO2 at 298.15
K. Then, GE and HE from the EOS correlation can be calculated
as a function of xA () 1 - xB), as shown in Figure 3. On the
other hand, GE and HE from the association model can be
evaluated similarly as a function of xB () 1-xA) with four
unknown parameters K1, K2, H1, and H2; see eqs 11-13.
These unknown parameters have been determined using a
nonlinear least-squares method by minimizing the differences
of both GE and HE functions between the EOS and the
association models; see Figure 4 for an example analysis. The
mole fraction range used in the analysis is 0 < xA < 0.6, since
solutions with higher CO2 mole fractions (about >0.7) become
immiscible liquids, as mentioned earlier. After some trial-anderror analyses, we have found that only one type of the complex
(AB or AB2) is dominating in the present liquid solution for
Cases B and C. This fact is consistent with the observation of
the minimum GE location of about xA ) 0.5 and xA ) 0.33,
mentioned in the previous section. Results of the present
analyses are shown in Table 5.
Discussion
RTILs that show a very strong chemical absorption with CO2
(Case C) all contain the anion type of [X-COO]: [bmim][Ac],
[emim][Ac], [bmim][PRO], [bmim][IBS], [bmim][TMA], and
[bmim][LEV]; see Figure 1. However, this trend cannot be
generalized, since the functional group X also plays a significant
role on the complex formation. The dramatic examples are cases
of X ) CH3, CH2CH3, CH(CH3)2, (CH3)3, and CH2CH2Cd
OCH3 (strong chemical absorption) versus X ) H, CF3,
CH2CH2CdONH2, and CH2NHCH2COOH (physical absorp-

16660 J. Phys. Chem. B, Vol. 112, No. 51, 2008

Yokozeki et al.

Figure 4. Analysis of the ideal association model using an example

of the CO2 + [bmim][Ac] system. Solid lines: the EOS model
calculation. Broken lines: results from the association model by the
least-squares analysis of GE and HE; see text for details.

tion). Thus, a simple and obvious rule for the chemical

absorption does not exist for the cases studied here. The present
observation will provide interesting challenges for the theoretical
modeling of these intermolecular interactions.
In this study, we have interpreted such strong absorptions
(highly nonideal phase behaviors) as chemical complex formations, using a classic association model: AB and AB2 complexes
in the binary A + B mixture.23,24 The present analysis and
interpretation seem quite reasonable. However, some comments
may be needed here. The thermodynamic excess functions (GE,
HE, and SE) may be sums of the physical and chemical
contributions. In the present analysis, we assumed that GE(chemical) . GE (physical), for example, neglecting the physical
contribution. Although this assumption may be valid in the
present case, the fact that the minimum location of GE is
somewhat different from that of HE (refer to Figure 3c) could
be the consequence of the present assumption. In this regard, it
is worth mentioning that GE derived from EOS (or solubility
data) is generally more reliable than the EOS-estimated HE.
Calorimetric measurements of HE (or the heat of mixing) would
provide more accurate values in HE and be desirable for the
association-model analysis.23,24
It is interesting to discuss an association-model analysis
different from that presented earlier. The present EOS, which
was developed with only a single temperature (about 298 K)
PTx data, may be reliable even for extrapolated temperatures.
We calculate GE curves for three temperatures of 273.15, 283.15
and 298.15 K at constant pressure of 6.5 MPa. These GE(T)
curves have been fitted to eq 11 (association model GE) in order
to obtain K1(T) and/or K2(T). Then, the heat of complex
formation (H1 and/or H2) can be derived from the following
thermodynamic relation, without use of HE in eq 12,

ln Ki ) const -

Hi
RT

(14)

Figure 5, panels a and b, shows an analysis of this method,

using the CO2 + [emim][Ac] system as an example. The
obtained value in H2 of -30.18 kJ mol-1 is consistent with
that in Table 5. Similar analyses have been made for the other
systems. Results are generally consistent with those in Table 5,
as presented in Table S2 of the Supporting Information.

Figure 5. Different method for the association model analysis, using

an example of the CO2 + [emim][Ac] system. (a) The temperaturedependent EOS GE(T) fitted to the association model GE using an
adjustable parameter K2(T). Solid lines: the EOS model calculations.
Dotted lines: association model GE fittings. (b) Analysis of K2(T)
obtained in Figure 5a to get the heat of the complex formation, H2.
Line: a straight-line fit. Symbols: K2(T) from Figure 5a.

Another question about the present chemical association is

to ask what type of chemical complexes they are (e.g., hydrogen
bonding or charge-transfer, etc.). The AB2 type of complexes
may mean that one CO2 molecule is sandwiched by two ionic
liquids, although the actual molecular configuration and the
nature of chemical bonding are uncertain. The magnitudes of
about -10 to -40 kJ mol-1 in H1 or H2 correspond to
various intermolecular interactions such as hydrogen bond,
ion-ion, ion-dipole, ion-quadrupole, charge-transfer bonding,
etc. No doubt, spectroscopic studies and/or ab initio MO
calculations will provide more insights on this matter. However,
we like to point out that simple solubility (vapor pressure)
measurements can provide rich and useful information about
the chemical absorption as demonstrated in the present work
and elsewhere.23-28
Some comments on Henrys law constant kH may be deserved
here. In most of the typical physical absorption cases such as
CO2 in [bmim][PF6], the conventional method to determine kH,
as described in Appendix B, works very well, if one chooses a
proper range of compositions (not including too large amounts
of the solute) and a proper-order polynomial in order to fit the
fugacity data. In some cases, the proper composition range is
not obvious, and one needs some trial-and-error analyses to get
reliable kH values. Example analyses of the conventional method
are shown in Table S3 of the Supporting Information, in order
to support the present discussion. In the case of small kH (here,
less than about 3 MPa at 298 K), the conventional method

Absorption of Carbon Dioxide in Ionic Liquids

J. Phys. Chem. B, Vol. 112, No. 51, 2008 16661

becomes unreliable; large uncertainties may result with the blind

use, depending strongly on the degree of the fitting polynomials,
range of compositions, and whether the origin (zero pressure
and zero composition) is included in the data set or not. The
method used in the present study (the EOS correlation method;
see Appendix B) works well for any cases and is quite reliable
using all experimental solubility data. Finally, it is useful to
say that a kH value less than about 2.3 MPa at 298 K would be
the case of chemical absorption for CO2 in ionic liquids, as can
be seen in Table 5.

(0.5A1A2A3 + A4A
3)(1 - kij)xixj

da
)
dT

with the following definition of symbols:

A1 (aiaj)-1/2
A2 aj

Conclusions
Solubilities of carbon dioxide in 18 ionic liquids have been
studied at a temperature of about 298 K and pressures less than
about 2 MPa. All solubility data have been successfully
correlated with our EOS model, which produced the thermodynamic excess (Gibbs, enthalpy, and entropy) functions and
Henrys law constants. Ten ionic liquids result in the physical
absorption, whereas eight of them show highly nonideal phase
behaviors, indicating the chemical absorption. These strong
absorptions have been analyzed with the ideal association model.
Two types of complex formation, AB and AB2, are identified:
A ) CO2 and B ) ionic liquid. The heats of complex formation
range from -11 to -37 kJ mol-1.
Appendices

(A6)

i)1 j)1

dai
daj
+ ai
dT
dT

(A7)

(A8)

A3 fij

(A9)

A4 1/A1

(A10)

A
3 ) fij
) -ij /T2

(A11)

Here, dai/dT (i ) i or j) above is given by:

3
dai
kik(Tci /T - T/Tci)k-1(-Tci /T2 - 1/Tci)
) A0
dT
k)0
(A12)

Appendix A.
Equation of State and the Derived Properties. The present
EOS is a generic RK (Redlich-Kwong) type of cubic equation
of state, as shown in eqs 1-8. The fugacity coefficient i of
the ith species for the present EOS model is given by:21

ln i ) ln

1
RT
a
a
+
+ bi

P(V - b)
V-b
RTb(V + b)
RTb
ai

ai

V
(A1)
+ 1 ln
a
b
V+b

ai
) 2

aiajfijxj
j)1

1 - kij -

b
i )

(b + b )(1 - m )x
i

ij

j)1

lijlji(lij - lji)xixj
(ljixi + lijxj)2

1 - kij -

lijlji(lij - lji)xixj
(ljixi + lijxj)2

(A2)

T da
V
a
H )
ln
+ RT(1 - Z)
b dT
b
V+b

R2T ci2
Pci

(A13)

The excess properties can be written using the residual properties

obtained above, with the dimensionless forms (R ) universal
gas constant):
N

xi

i)1

H i
(pure compound)
H
(mixture)
RT
RT
(A14)

-a
SE
)
R

xi

i)1

Si
(pure compound)
S
(mixture)
R
R
(A15)

-b

GE
HE
SE
)
RT
RT
R

(A3)

The residual properties (M Mideal gas - M) are useful for

calculations of the excess properties and the property changes.
Their explicit forms for the present EOS are:

A0 0.427480

HE
)
RT

where ai and bi are defined as:

N

where

(A4)

(A16)

or equivalently,

GE
)
RT

xi ln
i)1

i(species i in mixture)
i(pure species i)

(A17)

Appendix B.

1 da
V
RT
S )
ln
+ R ln
b dT V + b
P(V - b)

(A5)

The temperature derivative of the a parameter can be written

for the present EOS as

Henrys Law Constant. The Henrys constant, kH, for a

solute (here species 1) is defined as:

f V1 PV1 y1 ) kHx1

(x1 f 0)

(B1)

16662 J. Phys. Chem. B, Vol. 112, No. 51, 2008

Yokozeki et al.

where the superscript V indicates a vapor-phase property, and

x and y are liquid- and vapor-phase mole fractions of species 1,
respectively. The vapor-phase fugacity, f1V, of species 1 must
be equal to the liquid-phase fugacity, f1L, at the vapor-liquid
equilibrium (or solubility measurements):

f L1 PL1 x1

(B2)

Then, eqs B1 and B2 lead to:

kH ) PL1

(B3)

(x1 f 0)

The physical meaning of kH (normalized by a system unit

pressure) may be obtained by the direct relationship to the excess
chemical potential of solute at infinite dilution:23,32

1 - 01
kH(T, P) ) exp
RT

x1f0

( )

f V1 (T, P, y1)
df V1

x1
dx1

(B4)

(B5)
x1)0

where fV1 is the vapor phase fugacity of the pure gas (the present
case: CO2 with y1 ) 1) and can be calculated by a proper EOS
model29 at a given experimental (T, P). The fugacity for x1 ,
1 can be fitted to a proper-order polynomial of x1 in order to
use eq B5: for example, f1V ) a0 + a1x1 + a2x12, and then kH )
a1.
Another (or often more rigorous) method to obtain the
Henrys constant is to use an EOS correlation for the entire
experimental PTx data, using the relationship of eq B3. In the
present case, P 0 when x1 (solute CO2) f 0, since the solvent
(ionic liquid, species 2) is practically nonvolatile. When P )
0, the present EOS (eqs 1-8) provides an explicit form of the
EOS volume parameter, V0 as:

V0 )

)(

1 a2
- b2 1 2 RT

1-

4b2a2 /RT
(b2 - a2 /RT)2

(B6)

Then, eq B3 can be rewritten as:

ln(kH) ) ln PL1

(B8)

b
1 ) (b1 + b2)(1 - k12)(1 - m12) - b2

(B9)

Acknowledgment. The authors thank Mr. Brian L. Wells for

his assistance in the vapor-liquid equilibrium measurements
and DuPont Central Research and Development for supporting
the present work.
Supporting Information Available: (1) Synthesis of nine
RTILs, (2) liquid density data of the present RTILs, (3)
alternative analysis for the heat of complex formation, (4)
additional Henrys law constant analyses, and (5) additional plots
of thermodynamic excess functions. This information is available free of charge via the Internet at http://pubs.acs.org.
References and Notes

where 1 is the chemical potential of the solute at the infinitely

dilute solution state (in the present case, at the system T and P
f 0), and 10 is the chemical potential referring to the pure gas
(species 1) at the system T and at a pressure of 1 atm. Thus,
the Henrys constant is an important quantity for theoretical
works, which evaluate the intermolecular potential between a
solute molecule and a solvent molecule.
The Henrys law constant can be obtained conventionally
from the experimental solubility (PTx) data using the following
relation (eq B1):31

kH ) lim

a
1 ) 2a1a2f12(1 - k12) - a2

(B7)

where ln P1L is calculated by eq A1 with V ) V0, a ) a2, b )

b2, P ) 1 (with a system pressure unit, since P is eliminated
out in ln P1L) and:

(1) Ionic Liquids in Synthesis; Wasserscheid, P., Welton, T., Eds.;

Wiley-VCH: Weinheim, 2003.
(2) Plechkova, N. V.; Seddon, K. R. Chem. Soc. ReV. 2008, 37, 123
150.
(3) Magee, J. W.; Frenkel, M. IUPAC project (2003-020-2-100),
NIST Ionic liquids database, 2005; http://ilthermo.boulder.nist.gov/ILThermo/mainmenu.uix.
(4) Maginn, E. J. Design and Evaluation of Ionic Liquids as Novel
CO2 Absorbents, Quarterly Technical Report to DOE. December 31, 2004;
January 31, 2005; May 31, 2005; August 16, 2005; November 20, 2005;
January 12, 2006.
(5) The Capture and Sequestration of Carbon Dioxide, 2008; http://
www.esru.strath.ac.uk.
(6) Brennecke, J. F.; Anderson, J. L.; Maginn, E. J. Gas Solubilities
in Ionic Liquids, In Ionic Liquids in Synthesis; Wiley-VCH: Weinheim,
2003; p 81-92.
(7) Costa Gomes, M. F. J. Chem. Eng. Data 2007, 52 (2), 472475.
(8) Shiflett, M. B.; Yokozeki, A. J. Phys. Chem. B 2007, 111 (8), 2070
2074.
(9) Aki, S. N. V. K.; Mellein, B. R.; Saurer, E. M.; Brennecke, J. F.
J. Phys. Chem. B 2004, 108 (52), 2035520365.
(10) Shiflett, M. B.; Yokozeki, A. Ind. Eng. Chem. Res. 2005, 44 (12),
44534464.
(11) Kim, Y. S.; Choi, W. Y.; Jang, J. H.; Yoo, K.-P.; Lee, C. S. Fluid
Phase Equilib. 2005, 228-229, 439445.
.; Tuma, D.; Maurer, G.
(12) Kumean, J.; Perez-Salado Kamps, A
J. Chem. Thermodyn. 2006, 38 (11), 13961401.
.; Tuma, D.; Maurer, G.
(13) Kumean, J.; Perez-Salado Kamps, A
J. Chem. Eng. Data 2006, 51 (5), 18021807.
(14) Baltus, R. E.; Culbertson, B. H.; Dai, S.; Luo, H.; DePaoli, D. W.
J. Phys. Chem. B 2004, 108 (2), 721727.
(15) Muldoon, M. J.; Aki, S. N. V. K.; Anderson, J. L.; Dixon, J. K.;
Brennecke, J. F. J. Phys. Chem. B 2007, 111 (30), 90019009.
(16) Kim, Y. S.; Jang, J. H.; Lim, B. D.; Kang, J. W.; Lee, C. S. Fluid
Phase Equilib. 2007, 256 (1-2), 7074.
(17) Lee, B.-C.; Outcalt, S. L. J. Chem. Eng. Data 2006, 51 (3), 892
897.
(18) Chinn, D.; Vu, D. Q.; Driver, M. S.; Boudreau, L. C. CO2 removal
from gas using ionic liquid absorbents; U.S. Patents 20060251558A1 and
20050129598A1.
(19) Shiflett, M. B.; Kasprzak, D. J.; Junk, C. P.; Yokozeki, A. J. Chem.
Thermodyn 2008, 40 (1), 2531.
(20) Yokozeki, A.; Shiflett, M. B. AIChE J. 2006, 52 (11), 39523957.
(21) Yokozeki, A.; Shiflett, M. B. Appl. Energy 2007, 84, 12581273.
(22) Yokozeki, A. Int J. Thermophys. 2001, 22 (4), 10571071.
(23) Acree, W. E., Jr. Thermodynamic Properties of Nonelectrolyte
Solutions; Academic Press: New York, 1984.
(24) McGlashan, M. L.; Rastogi, R. P. Trans. Faraday Soc 1958, 54,
496501.
(25) Ohta, T.; Asano, H.; Nagata, I. Fluid Phase Equilib. 1980, 4, 105
114.
(26) Matsui, T.; Hepler, L. G.; Fenby, D. V. J. Phys. Chem. 1973, 77,
23972400.
(27) Handa, Y. P.; Fenby, D. V.; Jones, D. E. J. Chem. Thermodyn.
1975, 7, 337343.
(28) Hepler, L. G.; Fenby, D. V. J. Chem. Thermodyn. 1973, 5, 471
475.

Absorption of Carbon Dioxide in Ionic Liquids

(29) Lemmon, E. W.; McLinden, M. O.; Huber, M. L. NIST Reference
Fluid Thermodynamic and Transport Properties - REFPROP, v 7.0; U. S.
Department of Commerce, Technology Administration, National Institute
of Standards and Technology, Standard Reference Data Program: Gaithersburg, MD, 2002.
(30) Vetere, A. Chem Eng. J 1992, 49 (1), 2733.

J. Phys. Chem. B, Vol. 112, No. 51, 2008 16663

(31) Van Ness, H. C.; Abbott, M. M. Classical Thermodynamics of
Nonelectrolyte Solutions. McGraw-Hill: New York, 1982.
(32) Denbigh, K. The Principles of Chemical Equilibrium, 3rd ed.;
Cambridge University Press: London, 1971.

JP805784U