Understanding CO₂ capture kinetics and energetics by ionic liquids with molecular dynamics simulation

Abstract

Room temperature ionic liquids (ILs) are recognized to be potential media for CO₂ capture, but the interaction nature is less documented which hinders the future development of ILs with high CO₂ solvation capability. Here, through all atom molecular dynamics (MD) simulations, the solvation process of CO₂ with four representative ILs, [EMIM][BF₄], [BMIM][BF₄], [EMIM]CL and [BMIM]CL was systematically studied. Our data clearly reflect the fact that hydrophobic components from both cations and anions dominate CO₂ solvation because IL–CO₂ attraction is mainly driven by the van der Waals attractions. Consequently, cations with longer alkyl chain (for instance, [BMIM]⁺ than [EMIM]⁺) and anions with higher hydrophobicity (for instance, [BF₄]⁻ than CL⁻) effectively enhance CO₂ solvation. For all the ILs under study, addition of water into ILs abates CO₂ solvation through regulating the anion–CO₂ interactions. Moreover, it is worth mentioning that ILs with a hydrophobic anion ([BF₄]⁻ in this study) are more resistant to the existence of water to capture CO₂ than ILs with a hydrophilic anion (Cl⁻ in this study). Free energy decomposition analyses were conducted which support the above findings consistently. Generally, it is predicted that cations with long alkyl chain, anions with high hydrophobicity, and ILs with smaller surface tension are potentially effective CO₂ capturing media. Our present study helps the deep understanding of the CO₂ capturing mechanism by ILs and is expected to facilitate the future design and fabrication of a novel IL medium for gas capture and storage.

---

1. Introduction

Room temperature ionic liquids (ILs) are a new class of organic salts whose melting temperature falls below the conventional limit of 100 °C. Compared with conventional organic solvents, they possess many intriguing qualities, such as low vapor pressure, low toxicity, and high thermodynamic stability which have made these ionic compounds the solvents of choice for the so-called green chemistry.^1,2

ILs are regarded as potential candidates for CO₂ storage, the dominant reason for the global warming process.³ It is remarkable that, compared with the conventional amine scrubbing approach,⁴ the organic and non-volatilizing nature of ILs can effectively decrease the energy cost for CO₂ capture and avoid the concern of causing environmental pollution.⁵ For instance, Liu et al. have reported single-stage and multi-stage CO₂ adsorption processes by ILs which could effectively decrease the electrical and thermal energy costs by 42.8% and 66.04%, respectively.⁶ Moreover, several ILs demonstrated CO₂-selective capturing capability which can be utilized for the purpose of CO₂ separation and purification from gas mixture.⁷ Numerous research groups have systematically studied the capturing capacity of various ILs for various gases, including CO₂, CH₄, H₂ and so on, by both experimental and theoretical approaches.^1,2,8–15 Ramdin et al. experimentally studied the CO₂ solubility and diffusivity in numbers of ILs and proposed that ILs have great solubility capacity to CO₂.¹ Anthony et al. reported that 1-n-butyl-3-methlimidazolium hexafluorophosphates have strong dissolving ability for CO₂.² Bates et al. discovered that a new room temperature ionic liquid (TSIL) has a considerable capacity to capture CO₂ effectively, compared to the commercial amine sequestering reagents.¹³ Kerle et al. further reported that CO₂ solubility is dependent to the alkane chain length of ILs.¹² Shi et al. discovered that the ionic liquid ([HMIM][Tf₂N]) demonstrates strong interaction with CO₂, by conducting Monte Carlo simulations, which is six times stronger than the interaction between ILs and H₂.¹⁵

Albeit a large number of works that have been conducted experimentally, the effective CO₂ capture and storage are still challenges. Previous experiments have evaluated the effects of introduction of water to ILs for CO₂ capture. For instance, Bermejo et al. have systematically explored the effects of water in ILs on CO₂ solubility and found that only small ratio of water in ILs is favorable to CO₂ solvation. In contrast, high concentration of water reversely lowered the CO₂ solubility.¹⁶ Unfortunately, the molecular mechanism of the influence of water on the performance of ILs is still unknown. This situation becomes to be the main obstacle to the design and development of high-performance ILs for the aim of effective CO₂ capture and storage.

In this work, the solvation process of CO₂ by several ILs is studied by atomic molecular dynamics (MD) simulations. Four experimentally widely-studied ILs and the corresponding hydrated states were adopted as the representative solvents. Several key characters of ILs, including the specific types of cations and anions, the length of the alkyl chain, etc., have been systemically assessed on CO₂ solvation. We sincerely hope that our present study could help the better and deep understanding of CO₂ capturing mechanism by ILs and facilitate the future design and fabrication of novel IL medium for gas capture and storage.

2. Model setup and simulation details

All the MD simulations were performed with the GROMACS 5.1.3 package¹⁷ with GPU acceleration. Four types of representative ILs which are commonly studied in the experiments have been considered: [BMIM][BF₄], [EMIM][BF₄], [BMIM]CL and [EMIM]CL. The force field parameters developed by Chaban et al. through quantum mechanics calculations were used for ILs.¹⁸ TIP3P water model¹⁹ and the EMP2 CO₂ model²⁰ were used. During the simulation, a leap-frog algorithm was used to integrate Newton's equations of motion, and the time step was set to 2 fs. All the covalent bonds were constrained by the LINCS algorithm.²¹ The cut-off for the short-range electrostatics and Lennard–Jones interactions were both set to 1.0 nm and the particle mesh Ewald (PME) algorithm²² was used to deal with the long-range electrostatic interactions. The simulation systems were firstly optimized by an energy minimization with the steepest descent algorithm to generate an appropriate starting configuration, and then equilibrated for 500 ps at NPT ensemble to reach the reference temperature of 300 K and the reference pressure of 1 bar, respectively. Then 100 ns trajectory was generated for data collection at the NVT ensemble.

In addition to the normal MD simulations, we further studied the CO₂ solvation energetics in ILs by calculating the solvation free energy (ΔG) using the Free Energy Perturbation (FEP) method.^23,24 To study the effects of water contents in the ILs, we have considered a series of water molar concentrations in the range of 0–96.2%. The components in the simulation boxes were depicted in Fig. 1. In the FEP calculations, 21 values were used for the coupling parameter λ with a resolution of 0.05. The process was divided into two parts: the perturbation of electrostatic interaction and the perturbation of van der Waals (vdW) interaction.

Fig. 1 (a) Molecular structures of [BMIM]⁺, [EMIM]⁺ and [BF₄]⁻; (b) snapshot of the simulated box. The CO₂ molecule is placed at the box center.

3. Results and discussion

3.1 CO₂ solvation in pure water and ILs

As the main aim of the current work is to understand the CO₂ storage capability of ILs through a mechanism view of the solvation free energetics, the benchmark study of the CO₂ solvation free energy in pure water (abbreviated as ΔG_water) is desired. The current molecular models gave a theoretical prediction of ΔG_water = 1.80 kJ mol⁻¹ at 300 K. Compared to previous studies (Table 1), this agrees well with the experimental value of 1.005 kJ mol⁻¹.²⁵ To verify the validity of our present models of CO₂ and ILs, we have calculated the solvation free energy of CO₂ in pure [BMIM][BF₄] at the 323 K and obtained a value of −1.00 kJ mol⁻¹. This is well consistent with experimental value of −1.005 kJ mol⁻¹ reported by Cadena et al.²⁶

Table 1 The CO₂ solvation free energy in pure water from current study and previous literatures

Method	ΔGwater	Ref.
MD	1.80	This work
QM/MM	0.042	27
MM	9.54	27
Quasi chemical theory	0.25	28
Experiment	1.005	25

---

In the following, we explore the solvation characteristics of CO₂ in pure ILs at room temperature (300 K). Using the same setups in the above calculations, the solvation free energy of CO₂ in ILs (abbreviated as ΔG_IL) are calculated which are: −1.69 kJ mol⁻¹ in [EMIM][BF₄], −2.09 kJ mol⁻¹ in [BMIM][BF₄], −0.87 kJ mol⁻¹ in [EMIM]CL and −1.92 kJ mol⁻¹ in [BMIM]CL, respectively. Compared with the value in pure water of 1.80 kJ mol⁻¹, the negative values clearly indicate that CO₂ molecule is better dissolved in ILs which is consistent with experimental reports.

Deduced from four ΔG_IL, the trend of CO₂ dissolving capacities of the four ILs is as follows: [BMIM][BF₄] > [BMIM][CL] > [EMIM][BF₄] > [EMIM][CL]. Amongst them, the [BMIM][BF₄] has the highest dissolving capacity for CO₂. When comparing the cations, it is found that the length of the alkyl chain of cations has a positive correlation to CO₂ solubility when the anion is same: [BMIM][BF₄] > [EMIM][BF₄]. In detail, the ΔG_IL in [BMIM][BF₄] is −0.40 kJ mol⁻¹ lower than that of [EMIM][BF₄]. These results are consistent with the experimental reports by Camper²⁹ and Anthony.³⁰ For two Chloride ILs, the trend is same: [BMIM][CL] > [EMIM][CL]. The free energy difference is even larger, which is −1.05 kJ mol⁻¹. This is well consistent with previous experimental observations by Yunus³¹ and Aki.³² In addition, the specific types of anions also play a minor role. The ΔG_IL of two [EMIM]⁺ ILs differs by −0.82 kJ mol⁻¹ when a [BF₄]⁻ is replaced by Cl⁻. While for the two [BMIM]⁺ ILs, the difference is smaller which is −0.17 kJ mol⁻¹. This is mainly attributed to the effective volumes (in other words, the hydrophobicity) of the anions which will be presented below in detail. It is worth mention that Zhang et al. also have demonstrated that the increase in alkyl chain of cation and the fluoration of anions or cations are beneficial to capture CO₂.^33,34

In order to probe the dominating molecular components that play the key role in the solvation process of CO₂ in pure ILs, we analyzed the radial distribution functions (RDFs) of the imidazole ring, alkyl chain and anion around the center of mass (COM) of the CO₂ molecule. As depicted in Fig. 2a and b, the first interaction shell (beginning from 0.25–0.3 nm) around CO₂ is uniformly from the anions for both [BF₄]⁻ and Cl⁻. For comparison, the RDF peaks representing cation begin at a larger distance, ∼0.42 nm from CO₂. These results indicate the fact that anions form direction interactions with CO₂. Especially that for the [BF₄]⁻, it forms more intimate interactions, through the Fluorine atoms, with CO₂ with a closer distance of ∼0.25 nm. The compact anion shell, to certain extent, prevents cations from interacting with CO₂. These phenomena are consistent with experimental results by Anthony³⁰ and Zhang.³⁵ As monovalent anions, the [BF₄]⁻ clearly has a larger volume than Cl⁻. Thus, [BF₄]⁻ is relatively more hydrophobic than Cl⁻. Zhang et al. also reported that [HMIM][FEP] has higher CO₂ solubility than [HMIM][TF₂N] experimentally because of the longer fluoroalkyl chain of [FEP].³⁵

Fig. 2 The radial distribution functions of ILs atoms around CO₂ center of mass (COM). (a) [BMIM][BF₄] solid line; [EMIM][BF₄] dotted line; (b) [BMIM][CL] solid line; [EMIM][CL] dotted line.

Further, we qualitatively compared different parts of cations in contacting with CO₂. For brief, the imidazole ring of the cation is named as ‘head’ and the alkyl chain is named as ‘tail’ as depicted in the insert of Fig. 2a. For the four ILs, the heights of RDF peaks representing CO₂–tail interactions are always higher than that of CO₂–head interactions at the similar distance, regardless of the specific types of anions. Especially for the chloride ILs (Fig. 2b), there is almost no apparent peaks for the CO₂–head interactions. This points to the fact that the alkyl chains of cations play the more important role in CO₂ solvation and, more precisely, longer alkyl chain results in lower ΔG_IL.

In combination with above anion analyses, it can be concluded that CO₂ prefers to interact with hydrophobic components of cation and hydrophobic anions, although CO₂ carries a dipole. Thus, it is a feasible way to search the suitable ILs with long alkyl chains or fluorinate anions for CO₂ capture.

3.2 The effect of water in ILs on CO₂ solvation

Previous experiments have demonstrated that the existence of water highly regulates CO₂ capture. In detail, Bermejo et al. reported that small ratio of water in ILs is favorable to CO₂ dissolving. In contrast, high concentration of water will reversely lower the CO₂ solubility.¹⁶ Goodrich et al. also reported that the presence of water in ILs effectively reduced the CO₂ storage capacity.³⁶ Wang et al. reported that compared to pure triethylbutylammonium acetate ([N₂₂₂₄][Ac]), the introduction of H₂O in [N₂₂₂₄][Ac] reduced the CO₂ adsorption.³⁷ However, the underline physical mechanism that accounts for the lowered performance of ILs is still poorly documented. To explain these experimental observations, we have considered the solvation of CO₂ in a serious of hydrated ILs with molar ratios of water increased from 0% (pure ILs) to 96.2% (diluted ILs) and calculated the CO₂ solvation free energies, ΔG_h-IL.

From Fig. 3a, it is very interesting to find that the tendencies of change of ΔG_h-IL can be divided into two types according to anions. For Chloride ILs, the ΔG_h-IL increases more quickly with the introduction of water. At water concentration of 83.3%, they approach to the value of ΔG_water (1.80 kJ mol⁻¹). In contrast, ΔG_h-IL in [EMIM][BF₄] and [BMIM][BF₄] increases in a much lower manner and reaches to a plateau of 0.8–1.0 kJ mol⁻¹ when water content is as high as 96.2%.

Fig. 3 The change tendency of (a) total CO₂ solvation free energy, (b) electronic component and (c) van der Waals component in hydrated ILs with different water molar ratio. The dashed lines represent the corresponding values in pure water.

Then, the total ΔG_h-IL is decomposed into the electrostatic (ΔG_ele) and vdW (ΔG_vdW) contributions which is essential for a deep understanding of the physical interactions between CO₂ and the solvents. As shown in Fig. 3b, the ΔG_ele term is always negative which reveals that energy–favorable interactions such as charge–charge and dipole–dipole attractions between CO₂ and pure/hydrated ILs is the driving force for CO₂ capture. In hydrated ILs, the ΔG_ele term firstly increases (weakened electrostatic interactions) at low water contents and the profiles reach to the maximum at a water content of ∼75% for two [BMIM]⁺ ILs, and ∼87.5% for two [EMIM]⁺ ILs. After that, further increasing water content reversely decreases the ΔG_ele (re-strengthened electrostatic interactions) which favors CO₂ capture. This interesting transition is attributed to the formation of intimate interactions of CO₂ with excess water molecules (more discussions are presented below from RDF analyses). It is worth mention that, the water fractions in ILs are relatively small in the experiments (which is typically lower than 30%). Thus, the current results support the previous observations because the experimental conditions correspond to the weakened ΔG_ele range in Fig. 3b.

For the van der Waals interactions, it represents the energy consumption for introducing a cavity inside the solution thus the ΔG_vdW is positive. As shown in Fig. 3c, the ΔG_vdW is monotonically increasing with the addition of water. Considering the nature of van der Waals interactions, this reflects that the surface tension of the hydrated ILs increases because more energy is required to introduce the cavity for CO₂ molecule inside it. The ΔG_vdW increase is considerably larger than ΔG_ele decrease. Thus, the overall solubility of CO₂ in ILs generally decreases with addition of water. Moreover, as depicted in Fig. 3a, the CO₂ solubility in ILs with [BF₄]⁻ anion is less regulated by water. From the decomposition analyses, this is mainly caused by the slower increase of ΔG_vdW term in two [BF₄]⁻ ILs than that in Cl⁻ ILs. Thus it can be seen that smaller surface tension and stronger electrostatic interaction with CO₂ are two favorable factors to improve the CO₂ solubility in ILs.

To probe the role of water in preventing ILs from adsorbing CO₂, we analyzed spatial distributions of water, imidazole ring (head), alkyl chain (tail) and anions around the COM of CO₂. For cations in IL with water ratio from 0–87.5%, the first peaks representing both head–CO₂ and tail–CO₂ bindings uniformly decreased in [EMIM][CL] (Fig. 4a) and [EMIM][BF₄] (Fig. 4b). This reflects a fact that the cation–CO₂ interactions were suppressed, resulting in ΔG_ele increase during this range which adopted the maximum value at water ratio of 87.5% as shown in Fig. 3b. Then, when further increasing water ratio from 87.5% to 96.2%, an interesting reverse trend was observed: these peaks reversely increased and were almost comparable to that in pure ILs, revealing that the interactions of cation–CO₂ were almost recovered in diluted ILs. In addition to the formation of intimate binding of water to CO₂ (dashed lines in Fig. 4c and d), ΔG_ele was found to decrease at this range (Fig. 3b). For anions, it is also interesting to find that two anions demonstrated distinct different behaviors in response to water. For Cl⁻ (Fig. 4c), the first binding shell of Cl⁻–CO₂ dramatically decreases with addition of water. This is attributed to the strong hydrophilic nature of Cl⁻ which favors to bind to water instead of CO₂. For [BF₄]⁻ (Fig. 4d), the first binding shell increases, but in a slow manner, when water content increases, indicating that water always enhances the binding between CO₂ and [BF₄]⁻. This is attributed to the large volume of [BF₄]⁻ which is classified to be hydrophobic anion. Thus, the ΔG_h-IL of CO₂ in hydrated ILs with Cl⁻ anion increases more dramatically than [BF₄]⁻ as depicted in Fig. 3a.

Fig. 4 The radial distribution functions of solution components around CO₂ in pure, 87.5% and 96.2% hydrated ILs. (a) and (c): [EMIM][CL]; (b)and (d) [EMIM][BF₄].

By analyzing the RDFs, it is found that the dominate component to CO₂ solubility of hydrated ILs is from the anion because the influence of water is closely connected to the hydrophilicity of anions. The current theoretical results predict that ILs with hydrophobic anions are more robust to capture CO₂ when water exists.

3.3 Diffusions of ILs

One of the recognized features of ILs is the high viscosity that results in significantly lowered diffusion of the CO₂ adsorbates. Bermejo et al. reported that small ratio of water in ILs is favorable to CO₂ dissolving. In contrast, high concentration of water will reversely lower the CO₂ solubility.¹⁶ While the introduction of water into ILs regulates the diffusion of CO₂, less is known about the relationship between IL's viscosity and CO₂ solubility. In the following, we examined the diffusion coefficients (D) of cation and anion, respectively, with respect to water contents using the Einstein relation.³⁸

As summarized in Fig. 5, it is found that the values of D of both anion and cation in hydrated ILs quickly increase with the increase of water content. For the four types of pure ILs under study, the D is considerably low, with an order of ∼0.6 ×10⁻¹¹ m² s⁻¹. In diluted ILs, the D increases by around two orders. Moreover, comparing the four ILs, three aspects can be acquired: (1) the anions diffuse more quickly than cations, especially for Chloride ILs; (2) [EMIM]⁺ diffuses quicker than [BMIM]⁺ when anion is same; (3) Cl⁻ diffuses quicker than [BF₄]⁻ when cation is same. These phenomena are consistent to the Stokes–Einstein theory which predicts that molecules with smaller size can diffuse faster. Generally, from these data, CO₂ solvation in ILs is found to be negatively-correlated to the D of hydrated ILs. More specifically, hydrated ILs benefits CO₂ diffusion, simultaneously it reduces CO₂ storage capability.

Fig. 5 The diffusion coefficients of cations and anions in pure and hydrated ILs.

4. Conclusions

To summarize, by MD simulations, the key factors of ILs that determine CO₂ solvation were identified. For cations, the alkyl chain dominants CO₂ capture because longer alkyl chain could effectively enhance CO₂ solvation. While for anions, the hydrophobic species are proved to favor CO₂ solvation. From free energy decomposition analyses, this is attributed to the interacting character between CO₂ and the ILs which is mainly driven by the van der Waals attractions. The existence of water seriously abates CO₂ solvation energy for all the ILs which is attributed to the weakened anion–CO₂ interactions. More importantly, it is found that ILs with hydrophobic anions are more resistant to the existence of water to capture CO₂ than ILs with hydrophilic anions. Through our analyses, a conclusion was made that ILs with long alkyl chain, hydrophobic anions, smaller surface tension and stronger electrostatic interaction with CO₂ are indicative factors to improve the CO₂ solubility.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work is supported by the National Natural Science Foundation of China (Grant No. 11874238), the Basic Research Project of Natural Science Foundation of Shandong Province (Grant No. ZR2018MA034) and the Fundamental Research Funds of Shandong University.

References

1. M. Ramdin, T. W. de Loos and T. J. Vlugt, Ind. Eng. Chem. Res., 2012, 51, 8149–8177.
2. J. L. Anthony, E. J. Maginn and J. F. Brennecke, J. Phys. Chem. B, 2002, 106, 7315–7320.
3. S. Zeng, X. Zhang, L. Bai, X. Zhang, H. Wang, J. Wang, D. Bao, M. Li, X. Liu and S. Zhang, Chem. Rev., 2017, 117, 9625–9673.
4. H. Ahn, M. Luberti, Z. Liu and S. Brandani, Int. J. Greenhouse Gas Control, 2013, 16, 29–40.
5. M. E. Boot-Handford, J. C. Abanades, E. J. Anthony, M. J. Blunt, S. Brandani, N. Mac Dowell, J. R. Fernández, M.-C. Ferrari, R. Gross and J. P. Hallett, Energy Environ. Sci., 2014, 7, 130–189.
6. X. Liu, Y. Huang, Y. Zhao, R. Gani, X. Zhang and S. Zhang, Ind. Eng. Chem. Res., 2016, 55, 5931–5944.
7. X. Zhang, X. Zhang, H. Dong, Z. Zhao, S. Zhang and Y. Huang, Energy Environ. Sci., 2012, 5, 6668.
8. S. Budhathoki, J. K. Shah and E. J. Maginn, Ind. Eng. Chem. Res., 2015, 54, 8821–8828.
9. A. Yokozeki and M. B. Shiflett, Ind. Eng. Chem. Res., 2007, 46, 1605–1610.
10. A. Yokozeki and M. B. Shiflett, Appl. Energy, 2007, 84, 351–361.
11. D. Kerlé, R. Ludwig and D. Paschek, Z. Phys. Chem., 2013, 227, 167–176.
12. D. Kerlé, R. Ludwig, A. Geiger and D. Paschek, J. Phys. Chem. B, 2009, 113, 12727–12735.
13. E. D. Bates, R. D. Mayton, I. Ntai and J. H. Davis, J. Am. Chem. Soc., 2002, 124, 926–927.
14. J. L. Anderson, J. K. Dixon and J. F. Brennecke, Acc. Chem. Res., 2007, 40, 1208–1216.
15. W. Shi, D. C. Sorescu, D. R. Luebke, M. J. Keller and S. Wickramanayake, J. Phys. Chem. B, 2010, 114, 6531–6541.
16. M. D. Bermejo, M. Montero, E. Saez, L. J. Florusse, A. J. Kotlewska, M. J. Cocero, F. van Rantwijk and C. J. Peters, J. Phys. Chem. B, 2008, 112, 13532–13541.
17. M. J. Abraham, T. Murtola, R. Schulz, S. Páll, J. C. Smith, B. Hess and E. Lindahl, SoftwareX, 2015, 1, 19–25.
18. V. V. Chaban, I. V. Voroshylova and O. N. Kalugin, Phys. Chem. Chem. Phys., 2011, 13, 7910–7920.
19. M. F. Harrach and B. Drossel, J. Chem. Phys., 2014, 140, 174501.
20. H. Liu, S. Dai and D.-e. Jiang, Nanoscale, 2013, 5, 9984–9987.
21. B. Hess, H. Bekker, H. J. Berendsen and J. G. Fraaije, J. Comput. Chem., 1997, 18, 1463–1472.
22. T. Darden, L. Perera, L. Li and L. Pedersen, Structure, 1999, 7, R55–R60.
23. M. A. M. Latif, N. Micaêlo and M. B. A. Rahman, Chem. Phys. Lett., 2014, 615, 69–74.
24. K. Jumbri, N. M. Micaelo and M. B. Abdul Rahman, Mol. Simul., 2017, 43, 19–27.
25. J. Lide and H. P. R. Fredikse, CRC Handbook of Chemistry and Physics, CRC Press, 75th edn, 1995.
26. C. Cadena, J. L. Anthony, J. K. Shah, T. I. Morrow, J. F. Brennecke and E. J. Maginn, J. Am. Chem. Soc., 2004, 126, 5300–5308.
27. N. Prasetyo and T. S. Hofer, J. Chem. Theory Comput., 2018, 14, 6472–6483.
28. D. Jiao and S. B. Rempe, J. Chem. Phys., 2011, 134, 224506.
29. D. Camper, C. Becker, C. Koval and R. Noble, Ind. Eng. Chem. Res., 2005, 44, 1928–1933.
30. J. L. Anthony, J. L. Anderson, E. J. Maginn and J. F. Brennecke, J. Phys. Chem. B, 2005, 109, 6366–6374.
31. N. M. Yunus, M. A. Mutalib, Z. Man, M. A. Bustam and T. Murugesan, Chem. Eng. J., 2012, 189, 94–100.
32. S. N. Aki, B. R. Mellein, E. M. Saurer and J. F. Brennecke, J. Phys. Chem. B, 2004, 108, 20355–20365.
33. D. Almantariotis, T. Gefflaut, A. A. Padua, J.-Y. Coxam and M. Costa Gomes, J. Phys. Chem. B, 2010, 114, 3608–3617.
34. X. Zhang, F. Huo, Z. Liu, W. Wang, W. Shi and E. J. Maginn, J. Phys. Chem. B, 2009, 113, 7591–7598.
35. X. Zhang, Z. Liu and W. Wang, AIChE J., 2008, 54, 2717–2728.
36. B. F. Goodrich, J. C. de la Fuente, B. E. Gurkan, Z. K. Lopez, E. A. Price, Y. Huang and J. F. Brennecke, J. Phys. Chem. B, 2011, 115, 9140–9150.
37. G. Wang, W. Hou, F. Xiao, J. Geng, Y. Wu and Z. Zhang, J. Chem. Eng. Data, 2011, 56, 1125–1133.
38. Z. Tang, L. Lu, Z. Dai, W. Xie, L. Shi and X. Lu, Langmuir, 2017, 33, 11658–11669.

---

Footnote

† These authors contribute equally to this work.

---