Water effect on CO2 absorption for hydroxylammonium based ionic liquids: A molecular dynamics study

Chemical Physics 400 (2012) 118–125

Contents lists available at SciVerse ScienceDirect

Chemical Physics journal homepage: www.elsevier.com/locate/chemphys

Water effect on CO2 absorption for hydroxylammonium based ionic liquids: A molecular dynamics study Santiago Aparicio a,⇑, Mert Atilhan b a b

Department of Chemistry, University of Burgos, 09001 Burgos, Spain Chemical Engineering Department, Qatar University, Doha, Qatar

a r t i c l e

i n f o

Article history: Received 13 September 2011 In final form 11 March 2012 Available online 21 March 2012 Keywords: Hydroxylammonium ionic liquids CO2 absorption Water Molecular dynamics

a b s t r a c t The effect of water content on CO2 absorption in 2-hydroxyethyl-trimethylammonium L-(+)-lactate and tris(2-hydroxyethyl)methylammonium methylsulfate ionic liquids was studied using classical molecular dynamics simulations. The analysis of structural and dynamic properties, together with the energy contributions, showed that molecular-level structuring of CO2–ionic liquids is not affected by the presence of water molecules. Ion–water interactions are developed while maintaining the previous fluids’ structuring. The predicted dynamic properties show decreasing molecular mobility, that should lead to increasing viscosity upon water addition for the studied concentration range. Nevertheless, water has a moderate effect on CO2 transport within the studied hydroxylammonium fluids. Ó 2012 Elsevier B.V. All rights reserved.

1. Introduction The capture of CO2 emissions from flue gases produced in fossil fuel power plants is a remarkable industrial problem because of the proposed effect of CO2 on global climate change [1]. The use of fossil fuels for electricity generation leads to 25% of the total CO2 emissions [2], and these energy related emissions are projected to increase at a 2.1% rate per year [3], because of the forecasted consumption of fossil fuels for electricity generation [3]. The current technologies used for CO2 capture from flue gases are amine based and related methods, in which a chemical absorption of CO2 is produced [4]. The well-known technological problems of amine absorption methods [5,6] has driven the attention, both in industry and academia, to less energy and more efficient alternatives. Ionic liquids have been proposed as possible candidates for CO2 capture [7,8], with available studies for several families of ionic liquids [9] including cations such as imidazolium [10,11], phosphonium [12,13], pyridinium [14] and guanidinium [15], paired with anions such as fluorinated ones [16], acetate [17] or even anion-functionalized ionic liquids [18]. The current status on the use of ionic liquids as CO2 absortive fluids was reviewed by our research group in a recent work, showing that further studies have to be performed to reach suitable alternatives to current processes [9].

⇑ Corresponding author. Address: Department of Chemistry, University of Burgos, Plaza Missael Bañuelos s.n., 09001 Burgos, Spain. Tel.: +34 947 258062; fax: +34 947 258831. E-mail address: [email protected] (S. Aparicio). 0301-0104/$ - see front matter Ó 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.chemphys.2012.03.008

The large number of possible anion–cation combinations leading to ionic liquids hinders experimental validation with all of them because of the costly absorption measurements. Therefore, the knowledge of molecular level factors controlling the gas absorption is a very important aspect for advancing in CO2 capturing technologies using ionic liquids, which can be obtained using computational chemistry methods. This computational approach has been used in the literature to analyze relevant aspects such as the partial molar volume of CO2 in ionic liquids [19], viscosity variation with CO2 absorption [20], solvation of CO2 in ionic liquids [21], molecular level structural aspects [22], or the development of computational tools for predicting CO2 solubility [23]. Hydroxylammonium-based ionic liquids are a suitable family of ionic liquids because of their remarkably adequate environmental and toxicological properties [24,25] together with their low productions costs [26]. The combination of these cations with anions such as lactate or methylsulfate leads to complex interactions, both between the involved ions together with CO2 absorbed molecules. Therefore, 2-hydroxyethyl-trimethylammonium L-(+)-lactate ([HE 3MA]LAC) and tris(2-hydroxyethyl)methylammonium methylsulfate ([3HEMA]MS) ionic liquids, Fig. 1, were selected and studied in depth by our research, using a combined experimental and computational approach, to analyze the properties of pure fluids and CO2 absorption [27,28]. Likewise, the water effect on CO2 absorption by ionic liquids was established by Husson et al. [29], who reported for imidazolium-based ionic liquids CO2 solubility decreasing (30%) with increasing quantity of water, although the molecular level mechanism was not fully understood. Therefore, as a continuation of our previous works [27,28], we report here a theoretical study using classical molecular dynamics

S. Aparicio, M. Atilhan / Chemical Physics 400 (2012) 118–125

119

according to the force field parameterization described by our group in a previous work [34]. Atomic charges of involved ions develop a pivotal role for molecular dynamics simulations, and thus, we should remark that charges used in this work were inferred according to the Merz–Singh–Kollman scheme [35], obtained from gas phase simulations of isolated ions. Parameterization for CO2 molecules was obtained from the literature [22] and SPC-E model was used for water along this work [36]. 3. Results and discussion

Fig. 1. Molecular structure of the studied ionic liquids (ionic pairs).

simulations on the molecular level properties and structure of water + CO2 + [HE3MA]LAC/[3HEMA]MS mixed fluids. The objective of the work is to analyze water effect on the CO2 absorption process from structural, energetic and dynamic viewpoints, which is very important considering the highly hygroscopic character of [HE3MA]LAC/[3HEMA]MS ionic liquids [27,28]. 2. Methods Classical molecular dynamics simulations were carried out using the MDynaMix v.5.0 molecular modelling package [30]. The simulations were carried out in the NPT ensemble using the Nosé–Hoover method to control the temperature and pressure of the simulation system [31]. The equations of motion were solved using the Tuckerman–Berne double time step algorithm [32], with long and short time steps of 1 and 0.1 fs, respectively. The Ewald summation method [33] was implemented for the Coulombic interactions with cut-off radius of 1.5 nm. The simulated systems consist of cubic boxes, with the compositions reported in Table 1, to which periodic boundary conditions were applied in the three directions to simulate an infinite system. Initial boxes were generated placing random ions, CO2 and water molecules in a FCC lattice at low density (0.2–0.3 g cm 3), then NPT simulations were performed at the selected pressure and temperature up to ensure equilibrium (checked through constant potential energy), after equilibration, 10 ns runs (time step 1 fs) were performed for the analysis of systems’ properties. Ionic liquids were described

Husson et al. [29] studied the water effect on the CO2 absorption in 1-ethyl-3-methylimidazolium bis(trifluoromethylsulfonyl)amide ionic liquid. Measurements showed dreceasing CO2 absorption with increasing water mole fraction together with decreasing fluid viscosity. The authors consider that the addition of water modifies the CO2 transport in the studied ionic liquids. As the same order of magnitude for the low-pressure solubility of CO2 is maintained with increasing water content, the decreasing viscosity may be considered as a favourable factor for the use of water + ionic liquids as CO2 capturing fluids. Goodrich et al. [37] led to analogous conclusions for amine-functionalized anion-tethered ionic liquids. In spite of these experimental results, the molecular level structure of water + CO2 + ionic liquid systems is not fully clarified, Available studies on the use of computational chemistry methods to study the possible effects of water on the CO2 absorption of ionic liquids are scarce. Perez-Blanco and Maginn [38] reported a computational study analysing the behaviour of CO2 and water molecules at the 1-n-butyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide interface. Their most remarkable conclusions were that the presence of water slows down the ionic and CO2 diffusion in the bulk phase but it has little effect on the interfacial transport dynamics of CO2 molecules. The hydroxylammonium-based ionic liquids studied in this work, [HE3MA]LAC/[3HEMA]MS, were analyzed at four different mole fractions for each ionic liquid (Table 1), in which the number of water molecules is kept constant and the number of CO2 molecules is changed from low to high. A consideration arises from the reaction of CO2 with water to form carbonic acid and its dissociation products, which leads to a lowering of the ionic liquid pH [39] This possibility was not studied in this work because we have used a non-reactive force field. We consider that the level of carbonic acid formation should be small, and thus it should not have a remarkable effect on the systems properties. In the available literature simulations studies this effect was also ignored [38]. 3.1. Water + CO2 + [HE3MA]LAC Structural properties may be firstly analysed using radial distribution functions, RDFs. The most relevant RDFs are reported in Fig. 2 for the systems water + CO2 + [HE3MA]LAC ionic liquid. The center-of-mass anion–cation RDFs do not change remarkably both with increasing water/CO2 mole fractions, Fig. 2a, and they are almost equal to that in pure [HE3MA]LAC [27]. Therefore, any of the possible structural changes inferred from the remaining RDFs reported in Fig. 2 should arise from a spatial rearrangement of involved molecules but without changes in anion–cation relative positions, and thus, through a competing effect of CO2 and water molecules for the available spaces. Water molecules interact preferentially with LAC anion, Fig. 2a, although interactions with cation are also remarkable as shown in Fig. 2c. The water–LAC interactions are developed through the oxygens in the anion carboxylate group, Fig. 2b [27], and are characterized in RDFs by a first intense, sharp and narrow peak followed by a second less intense but also

120

S. Aparicio, M. Atilhan / Chemical Physics 400 (2012) 118–125

Table 1 Systems and conditions used for molecular dynamics simulations. All simulations at 303 K and 2.5 MPa. Number of ion pairs, CO2/water molecules and mole fractions (x)

Name

[HE3MA]LAC pairs

CO2

x (CO2)

H2O

x (H2O)

125 125 125 125

14 83 14 83

0.10 0.39 0.08 0.35

7 7 31 31

0.05 0.03 0.18 0.13

W–CO2–A–[HE3MA] W–CO2–B–[HE3MA] W–CO2–C–[HE3MA] W–CO2–D–[HE3MA]

[3HEMA]MS pairs 125 125 125 125

14 83 14 83

0.10 0.39 0.08 0.35

7 7 31 31

0.05 0.03 0.18 0.13

W–CO2–A–[3HEMA] W–CO2–B–[3HEMA] W–CO2–C–[3HEMA] W–CO2–D–[3HEMA]

5

5

12

A-C (a)

HW-O3/O4(A) (b)

OW-H4(C) (c)

4

4

8 g(r)

g(r)

3

g(r)

3 2

2

4 1 0

1

0

5

10

15

0

0

0

5

10

r/Å

15

20

5

CD-CD (e)

3

3

g(r)

g(r)

12 g(r)

4

8

2

2

4

1

1

10 r/Å

15

15

HW-OD (f)

4

5

10

5

HW-OW (d)

0

5 r/Å

16

0

0

r/Å

0

0

5

10 r/Å

15

0

0

5

10

15

r/Å

W-CO2-A-[HE3MA] W-CO2-B-[HE3MA] W-CO2-C-[HE3MA] W-CO2-D-[HE3MA] Fig. 2. Radial distribution functions for the ternary systems [HE3MA]LAC + water + CO2, with compositions reported in Table 1, at 303 K and 2.5 MPa obtained from molecular dynamics simulations. A stands for anion and C for cation; O3 and O4 for COO oxygens in LAC anion; H4 for hydroxyl hydrogens in [HE3MA] cation; HW and OW for water hydrogens and oxygen, respectively; CD and OD for CO2 carbon and oxygens, respectively. In panel a, we report center of mass radial distribuition functions, for the remaining panels we report site-site functions.

important peak. These RDFs do not change neither in the position nor in the intensity with increasing water/CO2 mole fractions, and thus point to strong hydrogen bonding of water molecules with the carboxylate group in the LAC anion, even for low water mole fractions. The water/[HE3MA] cation interactions are more complex, as the RDFs reported in Fig. 2c show. They are characterised for low water mole fractions (sample W–CO2–A–[HE3MA]) by a first and second peaks, which by increasing CO2 concentration (W–CO2– B–[HE3MA]) are shifted toward lower distances. Analogous results are inferred for larger water mole fractions (comparison of W– CO2–C–[HE3MA] and W–CO2–D–[HE3MA]).Therefore, increasing CO2 concentration for a fixed water mole concentration tends to

favour the water–[HE3MA] hydrogen bonding. For a fixed CO2 concentration (comparison of W–CO2–A–[HE3MA] and W–CO2–C– [HE3MA]), increasing water mole fraction also leads to stronger cation–water interactions. Likewise, results reported in Fig. 2b in comparison with those in Fig. 2c show that larger water concentrations are required to develop remarkable water–[HE3MA] interactions than water–LAC ones. Moreover, water–LAC interactions are not affected by the CO2 concentration, contrary to water–[HE3MA]. Site–site RDFs for water–water interactions, Fig. 2d, show hydrogen bonding between water molecules only for certain concentrations. For low water mole fractions (W–CO2–A–[HE3MA]) water–water hydrogen bonding is absent, but surprisingly on going to larger

S. Aparicio, M. Atilhan / Chemical Physics 400 (2012) 118–125

CO2 concentrations for this fixed water content (comparison of W– CO2–A–[HE3MA] and W–CO2–B–[HE3MA]), water molecules develop hydrogen bonding as showed by the sharp peaks appearing for W–CO2–B–[HE3MA] in Fig. 2d. This effect is also obtained for larger water mole fractions (comparison of W–CO2–C–[HE3MA] and W–CO2–D–[HE3MA]). Therefore, water–water hydrogen bonding is produced by increasing CO2 concentrations. CO2–CO2 interactions, Fig. 2e, are affected by the presence of water molecules, increasing quantities of water decrease RDFs, but the peak shape is very similar to that in CO2–[HE3MA]LAC binary system [28]. Moreover, we report in Fig. 2f the RDFs for water–CO2 interaction showing a complex structuring but discarding water–CO2 interaction, the first peak appearing at around 4 Å points to the fact that both molecules share similar spatial regions. A more detailed picture of the spatial arrangement of the involved molecules may be obtained from spatial distribution functions, SDFs, which are reported around the anion and cation for W–CO2–D–[HE3MA] mixture in Fig. 3. Results reported in Fig. 3a show that water and cations share the same spatial regions around LAC anion: blue and green caps around the carboxylate group and the caps above the hydroxyl group. Therefore, this spatial arrangement allows water molecules to develop hydrogen bonding with LAC anion and [HE3MA] cation simultaneously in agreement with RDFs reported in Fig. 2b and c. The CO2 distribution around the LAC anion is very similar to the system in the absence of water [28], with only slight differences above and below the carboxylate group, the absorbed water interacts preferentially with carboxylate and hydroxyl LAC groups, and thus a certain rearrangement of the absorbed CO2 molecules is produced around the region. The spatial distribution around the [HE3MA] cation is reported in Fig. 3b, results show that water molecules occupy regions around the cation, which are shared with LAC anion. Moreover, a comparison with previous results in the absence of water [28] show that water molecules occupy spaces that were previously occupied by LAC molecules, and a certain increase of CO2 concentration around the cation is produced. The analysis of thermophysical properties predicted by simulations may also shed light into the fluids structuring. We report calculated density and self-diffusion coefficients for the water– CO2–[HE3MA]LAC ternary system in Table 2. Reported results show that increasing water mole fraction for a fixed CO2 concentration leads to an increasing density: 0.96% on going from sample W–CO2–A–[HE3MA] to W–CO2–C–[HE3MA], and 0.38% from

121

W–CO2–B–[HE3MA] to W–CO2–D–[HE3MA]. Therefore, the water effect is more remarkable for low CO2 concentrations. The dynamics of the ternary system is characterized by diffusion heterogeneity, with CO2 molecules diffusing faster than the remaining ones, and with water diffusion rates close to those of involved ions, Table 2. Likewise, the addition of water molecules leads to a decrease of diffusion rates for all the involved species. This is a highly anomalous behaviour (water would be expected to increase the rate of diffusion). Nevertheless, this effect was previously reported by Perez-Blanco and Maginn [38] for 1-n-butyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide, who explained it considering the strong interaction of water molecules with multiple anions, and to the poor solvation of involved ions by water molecules. They also reported that most of water molecules were isolated or in pairs. A comparison of self-diffusion coefficients for W–CO2–D– [HE3MA] system with previously reported simulated data for the same system in the absence of water [28], shows a decrease for LAC, [HE3MA] and CO2. The comparison of the simulated mixture self-diffusion coefficient for W–CO2–D–[HE3MA] (0.0363  10 9 m2 s 1, obtained from the values reported in Table 2 and the corresponding mole fractions in Table 1), and the value for the same system in the absence of water (0.0472  10 9 m2 s 1 [28]) shows a 23% decrease with water addition. This decrease in the average diffusion rate of the mixed fluid should lead to an increase in viscosity. We should have expected a decrease of fluids viscosity with increasing water concentration. Nevertheless Goodrich et al. [37] showed for amine functionalized ionic liquids that water has a small effect for ionic liquids with absorbed CO2 for low water contents as the studied in this work, leading to remarkable decrease only for water concentrations larger than 5 wt.% (W–CO2–D– [HE3MA] has a 2 wt.% water content). This effect decreases with decreasing water mole fractions. Mixture self-diffusion coefficient for W–CO2–B–[HE3MA] system (0.0413  10 9 m2 s 1, 0.42 water wt.%) is 12.4% lower than the value previously reported for the same system in the absence of water [28]. The probable molecular explanation may be inferred from the RDFs reported in Fig. 2 b and c that show strong ion–water interaction, together with RDFs in Fig. 2a showing that ion–ion interactions are not weakened by the presence of water molecules. Therefore, water molecules reinforce intermolecular forces in the studies systems, at the studied water concentrations, which should lead to a reduced mobility and to the reported decrease in self-diffusion coefficients. In spite

Fig. 3. Spatial distribution functions of (a) hydroxyl hydrogens in [HE3MA]+ cation (dark blue surfaces showing 10 times average bulk density), carbon atoms in CO2 (orange surfaces showing 5.5 times average bulk density) and hydrogen atoms in water (green surfaces showing 10 times average bulk density) around LAC anion; (b) hydroxyl hydrogens in lactate anion (blue surfaces showing 10 times average bulk density), carbon atoms in CO2 (orange surfaces showing 5.5 times average bulk density) and hydrogen atoms in water (green surfaces showing 10 times average bulk density) around [HE3MA]+ cation. Results obtained from molecular dynamics simulations for mixture W-CO2-D-[HE3MA] (see Table 1) at 303 K and 2.5 MPa. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

122

S. Aparicio, M. Atilhan / Chemical Physics 400 (2012) 118–125

Table 2 Density, q, and self-diffusion coefficient, D, for ionic liquid + CO2 + water systems at 303 K/2.5 MPa obtained from molecular dynamics simulations. Names and compositions from Table 1. 109 D/m2 s

System

q/g cm W–CO2–A–[HE3MA] W–CO2–B–[HE3MA] W–CO2–C–[HE3MA] W–CO2–D–[HE3MA] W–CO2–A–[3HEMA] W–CO2–B–[3HEMA] W–CO2–C–[3HEMA] W–CO2–D–[3HEMA]

1

3

1.1120 ± 0.0007 1.1125 ± 0.0008 1.1227 ± 0.0009 1.1167 ± 0.0006 1.3056 ± 0.0009 1.2815 ± 0.0009 1.3078 ± 0.0010 1.2861 ± 0.0009

Anion

Cation

CO2

H2O

0.0137 ± 0.0012 0.0159 ± 0.0014 0.0126 ± 0.0011 0.0176 ± 0.0016 0.0208 ± 0.0019 0.0226 ± 0.0020 0.0222 ± 0.0020 0.0226 ± 0.0020

0.0129 ± 0.0012 0.0146 ± 0.0013 0.0106 ± 0.0010 0.0138 ± 0.0012 0.0176 ± 0.0016 0.0220 ± 0.0020 0.0194 ± 0.0017 0.0183 ± 0.0016

0.0409 ± 0.0037 0.0816 ± 0.0073 0.0428 ± 0.0039 0.0687 ± 0.0062 0.0563 ± 0.0051 0.1150 ± 0.0104 0.0514 ± 0.0046 0.0730 ± 0.0066

0.0308 ± 0.0028 0.0217 ± 0.0020 0.0212 ± 0.0019 0.0314 ± 0.0028 0.0385 ± 0.0035 0.0284 ± 0.0026 0.0345 ± 0.0031 0.0439 ± 0.0040

of the predicted increase in viscosity, we should remark that selfdiffusion coefficients for CO2 molecules are not strongly affected by the presence of water molecules, the value for the systems W–CO2–B–[HE3MA] (0.0816  10 9 m2 s 1, 0.42 water wt.%) and W–CO2–D–[HE3MA] (0.0687  10 9 m2 s 1, 2 water wt.%) are very close to that in the same systems without water (0.0762  10 9 m2 s 1 [28]), which is important for the kinetic of the CO2 absorption process in this ionic liquid. These results may also be analyzed considering the intermolecular interaction energies reported in Table 3. Very subtle effects are inferred from Table 3 simulated data. A new remarkable water–ion interaction energy contribution appears, being twice to fivefold larger for LAC–water than for [HE3MA]–water interactions, which is in agreement with the structural results reported in Figs. 2 and 3. Upon water absorption ion–CO2 interactions are slightly weakened in comparison with previously reported simulated data for the same system in the absence of water [28]. Nevertheless, a global analysis of all the involved interactions shows interesting features. The calculated energy of all the interactions involving LAC anions (LAC–LAC + LAC–[HE3MA] + LAC–CO2 + LAC–water) for W–CO2– D–[HE3MA] ( 1809.3 kJ mol 1) is larger than the value obtained for the same system in the absence of water molecules ( 1672.2 kJ mol 1 [28]) which would justify the self-diffusion results reported in the previous paragraph. Results for energy of all the interactions involving [HE3MA] cation show 1708.3 kJ mol 1 for W–CO2–D–[HE3MA], which is very close to the value of the same system in the absence of water ( 1743.6 kJ mol 1). Therefore, the main effect is developed through the LAC anions; nevertheless, the LAC–[HE3MA] total interaction energy for W– CO2–D–[HE3MA] (Lennard-Jones + electrostatic contribution, 1708.3 kJ mol 1) is very close to the value for the same system

in the absence of water ( 1743.6 kJ mol 1), being even closer for W–CO2–B–[HE3MA] ( 1737.9 kJ mol 1). Therefore, from the viewpoint of interaction energies, water addition leads to a new remarkable LAC–water interaction term, and also to a lower [HE3MA]–water term, while maintaining the anion–cation interaction energy, which should lead to an increase in viscosity of the studied mixed fluids. 3.2. Water + CO2 + [3HEMA]MS The relevant calculated RDFs are reported in Fig. 4a to e. The behavior is very similar to that for water–CO2–[HE3MA]LAC system, Fig. 2: anion–cation structuring not affected by the presence of water together with remarkable anion–water, and in a lower extension cation–water interactions. The main differences rises for the water-water interactions, Fig. 4d, which show a lower degree of structuring in [3HEMA]MS system in comparison with [HE3MA]LAC system, and for the water–CO2 interactions, Fig. 4f, for which the well-defined and narrow peaks show a remarkable interaction between water and CO2 molecules in the system containing [3HEMA]MS. SDFs for the water–CO2–[3HEMA]MS are reported in Fig. 5, structuring around MS is characterized by water molecules occupying regions overlapping with hydroxyl groups of [3HEMA] cation (caps close to the oxygens in SOO group, Fig. 5a), and CO2 molecules distributed in a more diffuse way than in the system without water [28]. The distribution around [3HEMA] cation, Fig. 5b, is characterized by water molecules distributed in regions not overlapping with those from MS anions, and also a diffuse CO2 distribution. Densities reported in Table 3 shows an increasing density with increasing water concentration: 0.17% on going from sample W–CO2–A–

Table 3 Intermolecular energy contributions (sum of Lennard Jones and electrostatic contributions) to the liquid phases of ionic liquid + CO2 + water systems at 303 K/2.5 MPa obtained from molecular dynamics simulations. A stands for anion and C for cation. All values in kJ mol 1. Names and compositions from Table 1. System

A–A

W–CO2–A– [HE3MA] W–CO2–B– [HE3MA] W–CO2–C– [HE3MA] W–CO2–D– [HE3MA] W–CO2–A– [3HEMA] W–CO2–B– [3HEMA] W–CO2–C– [3HEMA] W–CO2–D– [3HEMA]

1215.37 ± 1.24

A–C

3039.30 ± 1.87 1204.02 ± 1.10

C–C

A–CO2 37.92 ± 1.05

C–CO2 35.11 ± 0.92

CO2–CO2 0.64 ± 0.11

A–H2O 75.76 ± 2.23

C–H2O 28.29 ± 2.04

CO2–H2O 2.44 ± 0.25

H2O–H2O 0.65 ± 0.10

1077.50 ± 1.08

2754.86 ± 2.03 1062.00 ± 0.69

18.45 ± 0.49

25.89 ± 0.43

2.85 ± 0.10

113.47 ± 1.54

19.13 ± 2.15

1.99 ± 0.44

1.99 ± 0.21

1208.59 ± 1.10

3016.04 ± 2.27 1190.86 ± 0.97

24.54 ± 1.04

26.04 ± 0.96

0.73 ± 0.11

105.73 ± 0.88

45.01 ± 0.67.

0.03 ± 0.24

0.55 ± 0.18

1065.32 ± 1.06

2708.36 ± 1.77 1041.21 ± 0.78

20.20 ± 0.40

24.80 ± 0.34

3.14 ± 0.11

116.07 ± 0.91

16.35 ± 0.87

2.47 ± 0.27

1.26 ± 0.24

1000.01 ± 1.36

2577.59 ± 2.08

977.60 ± 0.85

7.97 ± 1.21

51.19 ± 1.04

0.76 ± 0.11

47.95 ± 2.12

27.56 ± 2.16

0.76 ± 0.11

1.09 ± 0.14

911.31 ± 5.60

2397.13 ± 1.78

880.75 ± 0.82

8.55 ± 0.43

35.83 ± 0.32

2.05 ± 0.10

92.06 ± 2.29

21.95 ± 1.39

0.50 ± 0.50

3.04 ± 0.13

990.66 ± 1.27

2561.70 ± 2.12

967.86 ± 1.13

8.78 ± 1.08

39.97 ± 0.76

0.37 ± 0.09

60.82 ± 0.84

14.50 ± 0.73

0.65 ± 0.27

2.45 ± 0.17

879.93 ± 0.88

2349.97 ± 1.53

862.68 ± 0.71

5.10 ± 0.41

30.97 ± 0.33

2.02 ± 0.09

77.00 ± 1.07

28.43 ± 0.76

1.76 ± 0.30

1.50 ± 0.14

123

S. Aparicio, M. Atilhan / Chemical Physics 400 (2012) 118–125

5

5

12

A-C (a)

HW-O6(A) (b)

OW-H4(C) (c)

4

4

8 g(r)

g(r)

3

g(r)

3 2

2

4 1 0

1

0

5

10

15

0

0

0

5

10

r/Å

15

5

5

15

5

CD-CD (e)

4

10 r/Å

8

HW-OW (d)

HW-OD (f) 4

6

g(r)

g(r)

3

g(r)

3 4

2

2 2

1 0

0

r/Å

0

5

10

15

0

1

0

5

r/Å

10 r/Å

15

0

0

5

10

15

r/Å

W-CO2-A-[3HEMA] W-CO2-B-[3HEMA] W-CO2-C-[3HEMA] W-CO2-D-[3HEMA] Fig. 4. Radial distribution functions for the ternary systems [3HEMA]MS + water + CO2, with compositions reported in Table 1, at 303 K and 2.5 MPa obtained from molecular dynamics simulations. A stands for anion and C for cation; O6 for SO3 oxygens in MS anion; H4 for hydroxyl hydrogens in [3HEMA] cation; HW and OW for water hydrogens and oxygen, respectively; CD and OD for CO2 carbon and oxygens, respectively. In panel a, we report center of mass radial distribution functions, for the remaining panels we report site–site functions.

Fig. 5. Spatial distribution functions of (a) hydroxyl hydrogens in [3HEMA]+ cation (dark blue surfaces showing 10 times average bulk density), carbon atoms in CO2 (orange surfaces showing 5.5 times average bulk density) and hydrogen atoms in water (green surfaces showing 10 times average bulk density) around MS anion; (b) SO3 oxygens in MS anion (pink surfaces showing 10 times average bulk density), carbon atoms in CO2 (orange surfaces showing 5.5 times average bulk density) and hydrogen atoms in water (green surfaces showing 10 times average bulk density) around [3HEMA]+ cation. Results obtained from molecular dynamics simulations for mixture W-CO2-D-[3HEMA] (see Table 1) at 303 K and 2.5 MPa. In panel b, hydrogens bonded to carbon atoms are not plotted for the sake of visibility. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

[3HEMA] to W–CO2–C–[3HEMA], and 0.36% from W–CO2–B– [3HEMA] to W–CO2–D–[3HEMA], which are lower than the increasing density reported for [HE3MA]LAC in the previous section. Dynamics is also characterized by a decrease of molecular mobility by water absorption. On going from W–CO2–B–[3HEMA] (0.33

water wt.%) or W–CO2–D–[3HEMA] (1.44 water wt.%) to the same samples without water molecules, mixture self-diffusion coefficients decrease 49% and 48%, respectively, which are larger values than the ones obtained in the previous section for [HE3MA]LAC systems. Likewise, it should be remarked that water self-diffusion

124

S. Aparicio, M. Atilhan / Chemical Physics 400 (2012) 118–125

Table 4 Distribution of hydrogen bonds in W–CO2–D–[HE3MA] and W–CO2–D–[3HEMA] at 303 K and 2.5 MPa. The values reported show the average per frame. Type

W–CO2–D–[HE3MA]

W–CO2–D–[3HEMA]

Water–water Water–anion Water–cation

6 ± 0.3 45 ± 2.2 10 ± 0.5

7 ± 0.3 38 ± 2.0 20 ± 1.1

coefficients are larger in [3HEMA]MS systems than in [HE3MA]LAC, Table 3, which may be justified considering the weaker water–MS interactions in comparison with water–LAC, Table 3. The water effect on self-diffusion coefficients for CO2 molecules are not strongly affected by the presence of water molecules, the value for the systems W–CO2–B–[3HEMA] (0.1150  10 9 m2 s 1, 0.33 water wt.%) and W–CO2–D–[3HEMA] (0.0730  10 9 m2 s 1, 1.44 water wt.%) are very close to that in the same systems without water (0.1170  10 9 m2 s 1 [28]), which is important for the kinetic of the CO2 absorption process in this ionic liquid. From the analysis of intermolecular interaction energies, Table 3, we may conclude that upon water absorption all the involved interactions are slightly weakened, but this effect is balanced by the remarkable MS anion–water, and in a lesser extension [3HEMA]-water interactions. The calculated energy of all the interactions involving MS anions is 1552.14 kJ mol 1 for the W–CO2– D–[3HEMA] system (1.44 water wt.%), 1586.99 kJ mol 1 for the WCO2–B–[3HEMA] system (0.36 water wt.%), in comparison with 1501.13 kJ mol 1 for the same systems in the absence of water [28]. Therefore, MS-water interactions do not disrupt the [MS]– [3HEMA] interactions, and they control the main structural features of the system. It should be remarked that LAC-water interactions are almost twice-fold stronger than the MS-water ones, Table 3.

the case of 1-n-butyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide. Moreover, we have analyzed the hydrogen bonding distribution between water molecules and the remaining molecules, Table 4. The reported results show the clear water-anion preferential interaction, with remarkably lower water–water and water–cation hydrogen bonding. This effect is less remarkable for W–CO2–D–[3HEMA] because of the presence of three hydroxyl groups in the cation, Therefore, the hydrogen bonding analysis shows that water molecules are preferentially isolated, developing hydrogen bonds with ions, especially with anions, and thus, decreasing ionic mobility in comparison with system without water. We should remark that these results are purely theoretical, and thus, they should be considered as tentative before experimental confirmation [38]. 4. Conclusion Molecular dynamics simulations were done in this work to study the properties of the water + CO2 + [HE3MA]LAC or [3HEMA]MS mixed systems. The analysis of structural and dynamic properties, together with the predicted intermolecular energies, show that the addition of water to the studied CO2–ionic liquid systems does not disrupt the fluids’ structuring but leads to remarkable ion–water interactions, specially strong for the LAC and MS anions. The predicted self-diffusion coefficients decrease with increasing water weight percentage, being studied up to 2 wt.%, and thus, an increase in fluids’ viscosity may be expected. This effect is justified considering preferential water–ion interactions, which strengths ion–ion forces, and thus, decreases molecular mobility. The diffusion rates of CO2 molecules, and their distributions around ions are not changed by the presence of water molecules. Therefore, CO2 absorption should not be remarkably affected by the presence of water in these highly hygroscopic ionic liquids, for the studied concentration ranges.

3.3. Hydrogen bonding in water + CO2 + {[HE3MA]LAC or [3HEMA]MS } Acknowledgement The highly anomalous effect of water on ionic liquids properties deserves a further discussion on the hydrogen bonding for the studied mixed fluids. We report in Fig. 6 the sizes of water–water cluster for W–CO2–D–[HE3MA] and W–CO2–D–[3HEMA]. The reported results show water molecules tending to be isolated (non water–water hydrogen bonded) or simply forming hydrogen bonded pairs, the populations or larger clusters are almost null, for both ionic liquids. These results are in total agreement with those previously reported by Perez-Blanco and Maginn [38] for

average number per frame

25

W-CO2-D-[HE3MA] W-CO2-D-[3HEMA]

20

15

10

5

0 1

2

3

4

5

6

water-water cluster size Fig. 6. Calculated water–water clustering in W–CO2–D–[HE3MA] and W–CO2–D– [3HEMA] at 303 K/2.5 MPa. Lines are plotted for guiding purposes.

Authors acknowledge the financial support by Ministerio de Ciencia e Innovación, Spain, (Project CTQ2010-15871) and from NPRP Grant (No: NPRP-09-739-2-284) from the Qatar National Research Fund. References [1] S. Solomon, D. Qin, M. Manning, Z. Chen, M. Marquis, K.B. Averty, M. Tignor, H.L. Miller, in: S. Solomon, ;D.Qin, M. Manning (Eds.), Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change,Cambridge Univ. Press, Cambridge, UK, 2007. [2] K.A. Baumert, T. Herzog, J. Pershing, Navigating the Numbers, Greenhouse Gas Data and International Climate Policy. World Resources Institute, USA, 2005. [3] International Energy Outlook Report 2008. DOE/EIA-0484, Energy Information Administration, U.S. Department of Energy, Washington, 2008. http:// www.eia.doe.gov/oiaf/ieo/. [4] U. Desideri, A. Paolucci, Energy Convers. Manage. 40 (1999) 1899. [5] A.B. Rao, E.S. Rubin, Environ. Sci. Technol. 36 (2002) 4467. [6] S. Mamun, H.F. Svendsen, K.A. Hoff, O. Juliussen, Energy Convers. Manage. 48 (2007) 251. [7] J.F. Brennecke, B.E. Gurkan, J. Phys. Chem. Lett. 1 (2010) 3459. [8] M. Hasib, M. Siaj, F. Larachi, Chem. Eng. Process. 49 (2010) 313. [9] F. Karadas, M. Atilhan, S. Aparicio, Energ. Fuel 24 (2010) 5817. [10] C. Wang, H. Luo, X. Luo, H. Li, S. Dai, Green Chem. 12 (2010) 2019. [11] J. Palgunadi, J.E. Kang, M. Cheong, H. Kim, H. Lee, S. Kim, Bull. Korean Chem. Soc. 30 (2009) 1749. [12] Y. Zhang, S. Zhang, K. Dong, Y. Zhang, Y. Shen, X. Lu, Chem. Eur. J. 12 (2006) 4021. [13] Y. Zhang, S. Zhang, X. Lu, Q. Zhou, W. Fan, X. Zhang, Chem. Eur. J. 15 (2009) 3003. [14] S. Supasitmongkol, P. Styrring, Eng. Env. Sci. 3 (2010) 1961. [15] S. Aparicio, M. Atilhan, Ener. Fuel. 24 (2010) 4989. [16] X. Zhang, Z. Liu, W. Wang, AIChE J. 54 (2008) 2717. [17] M.B. Shiflett, D.W. Drew, R.A. Cantini, A. Yokozeki, Ener. Fuel 24 (2010) 5781.

S. Aparicio, M. Atilhan / Chemical Physics 400 (2012) 118–125 [18] B.E. Gurkan, J.C. de la Fuente, E.M. Mindrup, L.E. Ficke, B.F. Goodrich, E.A. Price, W.F. Schneider, J.F. Brennecke, J. Am. Chem. Soc. 132 (2010) 2116. [19] X. Huang, C.J. Margulis, B.J. Berne, J. Am. Chem. Soc. 127 (2005) 17842. [20] K.E. Gutowski, E.J. Maginn, J. Am. Chem. Soc. 130 (2008) 14690. [21] Y. Shim, H.J. Kim, J. Phys. Chem. B 114 (2010) 10160. [22] W. Shi, E.J. Maginn, J. Phys. Chem. B 112 (2008) 2045. [23] L.F. Vega, O. Vilaseca, F. Llovell, J.S. Andreu, Fluid Phase Equilibr. 294 (2010) 15. [24] P. Nockemann, B. Thijs, K. Driesen, C.R. Jansenn, K. van Hecke, L. van Meervelt, S. Kossmann, B. Kirchner, K. Binnemans, J. Phys. Chem. B 111 (2007) 5254. [25] D.J. Couling, R.J. Bernot, K.M. Docherty, J.K. Dixon, E.J. Maginn, Green. Chem. 8 (2006) 82. [26] Y. Fukaya, Y. Lizuka, K. Sekikawa, H. Ohno, Green Chem. 9 (2007) 1115. [27] S. Aparicio, M. Atilhan, M. Khraisheh, R. Alcalde, J. Phys. Chem. B 115 (2011) 12473. [28] S. Aparicio, M. Atilhan, M. Khraisheh, R. Alcalde, J. Fernández, J. Phys. Chem. B 115 (2011) 12487.

125

[29] P. Husson, L. Pison, J. Jacquemin, M.F. Costa-Gomes, Fluid Phase Equilibr. 294 (2010) 98. [30] A.P. Lyubartsev, A. Laaksonen, Comput. Phys. Commun. 128 (2000) 565. [31] W.G. Hoover, Phys. Rev. A 31 (1985) 1695. [32] M. Tuckerman, B.J. Berne, G.J. Martyna, J. Chem. Phys. 97 (1992) 1990. [33] U.L. Essmann, M.L. Perera, T. Berkowitz, H. Darden, H. Lee, L.G. Pedersen, J. Chem. Phys. 103 (1995) 8577. [34] S. Aparicio, M. Atilhan, M. Khraisheh, R. Alcalde, J. Phys. Chem B 115 (2011) 12473. [35] B.H. Besler, K.M. Merz, P.A. Kollman, J. Comput. Chem. 111 (1990) 431. [36] H.J.C. Berendsen, J.R. Grigera, T.P. Straatsma, J. Phys. Chem. 91 (1987) 6269. [37] B.F. Goodrich, J.C. de la Fuente, B.E. Gurkan, Z.K. Lopez, E.A. Price, Y. Huang, J.F. Brennecke, J. Phys. Chem. B 115 (2011) 9140. [38] M. Perez-Blanco, E.J. Maginn, J. Phys. Chem. B 115 (2011) 10488. [39] C.A. Ober, R.B. Gupta, Ind. Eng. Chem. Res. 51 (2012) 2524.