Molecular mechanism of CO₂ absorption in phosphonium amino acid ionic liquid

Abstract

In this investigation, we examine the molecular mechanism of high pressure CO₂ absorption in tetra-butylphosphonium lysinate Ionic Liquid (IL) using molecular dynamics simulations. The calculations show that on CO₂ absorption, a structured bilayer of charges (in opposite phases) at the CO₂–IL interface is formed. The interface formation starts within a short span of tens of picoseconds, and attains saturation around ten nanoseconds, where CO₂ molecules remain absorbed in the bulk IL layers. The simulations predict a 0.9 molar absorption of CO₂ at T = 298 K and p_CO2 = 20 bar, with a further increase of 17% at a reduced temperature. The structural properties show that CO₂ molecules strongly and preferentially interact with the two terminal amine sites of the anions, where the presence of carboxylate group further enhances CO₂ absorption. The interaction and absorption of CO₂ molecules leads to enhanced mobility of the cations and anions of the IL. The mobility of anions is slightly higher than cations due to the preferential interaction of CO₂ molecules with anions. The molecular mechanism examined in this work can be used as a predictive tool to develop efficient amino acid functionalized ILs for CO₂ absorption.

---

1. Introduction

Carbon dioxide (CO₂) emissions from post-combustion pose strong challenges to the environment and health and this has motivated the search for materials for its efficient capture.^1,2 The current generation of materials which demonstrate high CO₂ uptake are amine based compounds.³ Unfortunately, the use of amines lead to poor recyclability (due to the large enthalpy of CO₂ absorption⁴) and high corrosion of steel⁵ (due to the active participation in oxidation). The use of Room Temperature Ionic Liquids (RTILs) as materials for CO₂ capture^6–8 have been proposed as a probable alternative, where the choice of cations and anions influence the amount of CO₂ absorption. Compared to cations, anions play a more significant role in absorption⁹ and their functionalization to enhance absorption has been attempted.^10–13 Among them, amino acid ionic liquids (AAILs) show potentially high^14,15 CO₂ uptake, due to the presence of amine groups. In one of the earliest attempts, Bates et al.¹⁶ employed FTIR spectroscopy and observed a 0.5 molar absorption of CO₂ (at room temperature) in an amine functionalized imidazolium based IL. Similar absorption capacity was reported by Zhang et al.¹⁷ on phosphonium AAILs, where the authors also observed enhanced absorption of CO₂ on addition of water in IL. Carvalho et al.¹⁸ used bubble point determination approach and found that phosphonium ILs have higher CO₂ solubility compared to imidazolium ILs. Using calorimetric and React-IR spectroscopy methods, Gurkan et al.¹⁹ observed an equimolar CO₂ absorption (at room temperature) in phosphonium based methioninate and prolinate AAILs. Wang et al.²⁰ measured enthalpy of absorption for a series of phosphonium based ILs and concluded that anions with large pK_a, show higher CO₂ absorption. Seo et al.²¹ synthesized a range of aprotic heterocyclic anion based phosphonium ILs, and using volumetric measurements reported an equimolar CO₂ absorption (at room temperature). Recently, Eisinger et al.²² reported a phase change (from solid to liquid) in a tetraethylphosphonium based IL. The authors concluded that heat evolved during the phase change leads to reduced power consumption associated with CO₂ capture in post combustion industrial process.

Computational investigations using Molecular Dynamics (MD) simulations to model interactions between supercritical CO₂ and [Bmim][PF₆] IL were employed by Huang et al.²³ The authors calculated structural properties and concluded that while CO₂ molecules can easily permeate into the IL phase, the ions of IL do not diffuse to the CO₂ phase. Subsequently, Perez-Blanco and Maginn²⁴ investigated absorption/desorption of CO₂ in [Bmim][Tf₂N] IL at various temperature and pressure conditions. The authors concluded that CO₂ molecules quickly absorb at the IL interface, and further diffuse to the bulk IL at longer timescales. Xing et al.²⁵ reported that the amine tethering on anion of an imidazolium IL enhances CO₂ absorption. The authors suggested that amine functionalization on the imidazolium cation leads to high interaction between cation and anion which results in low interaction of CO₂ with anion of IL, while amine functionalization on anion of IL weakens intra-IL interactions and facilitates accommodation of CO₂ absorption. Morganti et al.²⁶ and García et al.²⁷ compared the relative absorption of CO₂ and SO₂ in non-amine functionalized ILs and found preferential absorption of SO₂ over CO₂ in such non-amine ILs. Izgorodina et al.²⁸ showed that ILs containing large size anions (e.g. sulfonic, carboxylic and other bulky functional groups) are favorable candidates for CO₂ physisorption. Klähn and Seduraman²⁹ investigated several imidazolium-based ILs (with varying alkyl chain length of cation) and concluded that only the voids available in IL facilitate CO₂ absorption. The authors also concluded that ions slightly rearrange to accommodate CO₂ molecules in the bulk IL, and CO₂–IL interactions do not play any significant role.

Existing studies have provided very limited mechanistic details on CO₂ capture and for a limited type of ILs. The theoretical studies so far have not thoroughly examined the timescale of CO₂ absorption at IL interface and bulk. The actual process of CO₂ absorption in the bulk IL at very long timescales (∼100 ns) has never been explored. To examine these effects and propose a general mechanism of CO₂ absorption, we have chosen tetra-butylphosphonium lysinate [P₄₄₄₄][Lys] IL (see Fig. 1) which has large thermal stability, and the presence of multiple amine sites on the lysinate anion which can facilitate CO₂ capture. The computational details are presented in Section 2. Section 3 explores the mechanism associated with CO₂ absorption from MD simulations (using all atom force-fields). The properties calculated from MD simulations are: IL charge density, time dependent particle densities of CO₂ (at the interface and interaction with bulk IL), absorption isotherms at varying pressure and temperature, characterization of important molecular interactions between sites in IL and CO₂ via Radial Distribution Functions (RDFs)³⁰ and Spatial Distribution Functions (SDFs)³¹ and effect of CO₂ absorption on dynamics of cations and anions of IL. A summary of important findings concludes this paper.

Fig. 1 Structure of tetra-butylphosphonium lysinate (atom types defined for RDFs).

2. Computational details

The computational model required for MD simulations was initially created using a well equilibrated slab of [P₄₄₄₄][Lys] molecules, at T = 298 K and P_IL = 1 bar. A box of 512 ion pairs of IL was initially created followed by annealing and equilibration using the NpT ensemble. Various initial templates of CO₂ molecules were created and equilibrated using NpT ensemble to maintain the pressure and temperature conditions. The equilibrated IL slab was centered in a cuboidal box of dimensions 9.0 × 9.0 × z nm, where ‘z’ is derived as the length of the box (see Fig. S1 in the ESI†) required to contain a certain number of equilibrated CO₂ molecules (based on the required initial partial pressure and temperature). The IL slab was sandwiched between two layers using equilibrated CO₂ templates, where CO₂ molecules were initially separated by a vacuum layer of 1.0 nm from the surface of IL on each side of the box (see Fig. S1 in the ESI†). The initial vacuum was created to maintain an initial non-interacting behavior between CO₂ and IL based on earlier theoretical studies³² on other ILs. All MD simulations for CO₂ absorption were performed at T = 298 K and 278 K; p_CO2 = 1, 10 and 20 bar. However, in order to examine the mechanism of CO₂ absorption, unless explicitly mentioned, all results are focused on simulations performed at T = 298 K, and p_CO2 = 20 bar in this article. The GROMACS 4.6.7 (ref. 33) program was used for MD simulations. The CO₂ molecules were modeled using popularly used flexible TraPPE force field.³⁴ The force field parameters for [P₄₄₄₄][Lys] IL were taken from the work of Zhou et al.³⁵ The force field parameters of the neat IL were benchmarked by performing MD simulations and the calculated structural and dynamical properties like density, RDFs and conductivity were found to be in good agreement with previously reported experimental¹⁷ and simulation³⁵ studies (see Table S1 in the ESI†). All simulations for CO₂–IL systems were carried out for 100 ns with a time-step of 0.5 femtoseconds at the NVT ensemble with temperature maintained using the Nosé–Hoover chain thermostat^36,37 and a 1.4 nm cut-off for calculation of non-bonding interactions.

3. Results and discussion

CO₂ absorption at the interface and bulk

A time averaged charge density of [P₄₄₄₄][Lys] IL shows, on absorption of CO₂, a highly ordered and structured bilayer (of opposite phase) at the surface is formed (see Fig. 2a). The cations and anions of IL reorient themselves to facilitate the formation of the CO₂–IL interface. The area under the curve of positively charged outer region of the CO₂–IL interface is smaller than the negatively charged inner region. The existence of this bilayer facilitates the absorption of CO₂ molecules to the surface of IL. The charged bilayer formation at the CO₂–IL interface is a characteristic feature of CO₂ absorption and its formation is independent of thermodynamic conditions (see Fig. S2 in the ESI†). In comparison to vacuum–IL interface, the charge density is found more structured in CO₂–IL interface (see Fig. S3 in the ESI†). The charge bilayer formation (near the interface) supports the observations of Perez-Blanco et al.²⁴ and Dang et al.³⁸ on vacuum–[Bmim][NTf₂] IL interface.

Fig. 2 (a) Charge densities of IL before (t = 0) and after absorption (averaged from t = 5 ns to t = 10 ns). (b) Particle densities of CO₂ in the z direction of the box, where the average width of the IL layer is 4.8 nm. (c) Snapshot of CO₂–IL interface, surficial and bulk CO₂ absorption at t = 30 ns, color scheme is: green (CO₂), red (P⁺₄₄₄₄ cation) and blue (Lys⁻ anion).

Perez-Blanco et al.²⁴ calculated the particle density profile of CO₂ molecules on absorption in [Bmim][NTF₂] IL. The authors observed the interfacial crossing of CO₂ occurs with an exchange time (between CO₂ and IL phase) of 2.5 ps at the interface. Following a similar approach, we calculated the time averaged particle densities of CO₂ to examine its distribution during the processes of absorption. The particle densities are calculated using a 100 ps time window (for densities calculated till t = 1.5 ns) and 1000 ps time window, from the average time (for densities calculated after t = 2.0 ns) (see Fig. 2b). The results show that within the first few picoseconds; several layers of CO₂ molecules are absorbed near the surface of IL on both sides of the box. The intensity at the CO₂–IL interface increases till 100 picoseconds. A slight decrease in CO₂ particle density at the interface starts around 1.0 ns, as CO₂ molecules can enter the bulk IL region. Around 10.0 ns, the densities at the interface attain saturation. The time of saturation can be defined as the time beyond which particle densities at the interface remain almost constant. On saturation at the interface, existing CO₂ molecules present at the CO₂–IL interface either could move towards bulk CO₂ or could diffuse into the bulk IL layers and remains trapped. At this point, a dynamic movement of CO₂ is observed, however, due to favourability of the absorption, CO₂ molecules eventually diffuse towards the IL layer. The particle densities of CO₂ at the interface and in bulk IL remains almost constant between t = 30 ns and 100 ns, suggesting a total saturation of CO₂ density in all phases (see Fig. S4 in the ESI†). Similar features and trends in particle densities at other pressures and temperature are shown in Fig. S5 in the ESI.† A representative snapshot of absorption at t = 30 ns (see Fig. 2c) shows the presence of CO₂ molecules in the bulk IL layers. The quantitative measure of CO₂ uptake is seen from an absorption isotherm discussed in the next section. A video of simulation of CO₂ absorption in [P₄₄₄₄][Lys] IL at T = 298 K and p_CO2 = 20 bar is provided as a ESI Movie.†

Quantitative effects of pressure on CO₂ absorption

The time dependent molar absorption ratio of CO₂ : IL is calculated by an integration of CO₂ particle densities between two CO₂–IL interfacial boundaries. The absorption isotherm (see Fig. 3) shows a sharp blip at very short timescales (in few picoseconds) due to CO₂ absorption at the initially formed CO₂–IL interface. A slight decrease in absorption at timescale of 1 ns arise as CO₂ molecules at the interface will diffuse into the bulk IL layers, where absorption reaches an asymptotic value within a timescale of 10 ns. At T = 298 K and p_CO2 = 20 bar, a 0.9 molar absorption ratio (calculated as an average using time dependent molar absorption values from t = 10 ns to t = 30 ns) is achieved, and is ∼13 and 1.5 times higher than the absorption at p_CO2 = 1 bar and 10 bar respectively. At T = 278 K and p_CO2 = 20 bar, a maximum of 1.08 molar absorption of CO₂ per IL is achieved. The average molar absorption ratio values at other thermodynamic conditions are shown in Table S2 of ESI.† Due to the choice of a different cation, the theoretical values of molar absorption in this study are lower than the experimentally reported values of Goodrich et al.³⁹ (1.4 CO₂ : IL molar absorption) and Saravanamurugan et al.⁴⁰ (1.6 CO₂ : IL) with similar class of ILs.

Fig. 3 Molar CO₂ absorption in IL.

Structural view of interactions between CO₂ and IL

The interaction between CO₂ molecules and various interaction sites of the ILs are examined using calculation of RDFs and SDFs. The P[P₄₄₄₄], N1[Lys], N2[Lys] interaction sites on the IL and C[CO₂] sites are used for examination of structural features. The RDFs between the P site of the cation and N1, N2 sites of the anion for CO₂–IL and vacuum–IL systems show that interaction between cations and anions of IL decreases with CO₂ absorption (see Fig. S6 in the ESI†). A time dependent N1,N2–C RDF (see Fig. 4a) shows that structural features are similar at different times. However, the intensities of these RDFs increase till t = 10 ns, and then remains invariant till t = 30 ns, due to the saturation of CO₂ molecules at the CO₂–IL interface. Using the time period of t = 10 to t = 30 ns, we characterize various RDFs which correspond to possible interactions between cation/anionic sites of the IL and CO₂ (see Fig. 4b). The cation–CO₂ RDF show a sharp first peak with a small secondary peak. However, the first peak position appears at larger distances from the interaction site (0.56 nm) and with relatively low intensity. This is because CO₂ molecules have less preference to interact with the cation, due to a large positive charge on the phosphorous atom and steric hindrance from the bulky alkyl (butyl) group. In contrast, the anion–CO₂ RDFs show the following features: the N1–C RDF show three distinct solvation shells, a sharp first peak with the highest intensity, a first minimum at shortest distance of 0.48 nm, secondary and a tertiary shell which can accommodate CO₂ molecules. Approximately 11 CO₂ molecules interact with the N1 site within the cut-off of 1.1 nm. The N2–C RDF shows the existence of two solvation shells, which are relatively less distinct, and smaller compared to structural features of N1–C RDFs. Further, ∼4.5 CO₂ molecules can interact with the N2 sites within a cutoff of 1.1 nm. This suggests that CO₂ prefers to interact with the N1 site of the anion over N2. Our results support previous studies on conventional and AAILs. For e.g., CPMD calculations of Bhargava and Balasubramanian⁴¹ on [Bmim][PF₆] IL and Shi et al.⁴² on [Emim][OAc] IL also have shown that CO₂ molecules preferentially interact with anion. Gurkan et al.¹⁹ provided spectroscopic evidences of CO₂ binding with [P₆₆₆₁₄] based AAILs, a signature of COO⁻ formation (due to reactive amine–CO₂ interactions). Recently Luo et al.⁴³ observed multiple site cooperative interactions can enhance CO₂ capture in a pyridinium anion based IL.

Fig. 4 (a) Time dependent RDFs of anion–CO₂ interactions, (b) averaged RDFs of anion–CO₂ and cation–CO₂ interactions, (c) SDF of C[CO₂] around N1,C,O[Lys], (d) SDF of C[CO₂] around the N2[Lys] site.

To present a three-dimensional view of density distribution of CO₂ molecules around the amine sites of the anion, we calculated Spatial Distribution Functions (SDFs) using the TRAVIS³¹ program (visualized using VMD⁴⁴). The SDF of CO₂ molecules surrounding the N1 site (see Fig. 4c) shows three clouds of CO₂ densities, where two clouds are close to the carboxylate oxygen atoms, and the remaining cloud near the N1 site of the anion. However, only a single cloud of CO₂ is observed near the N2 site (see Fig. 4d) which confirms that the N1 site has preferential interaction with the CO₂ molecules, where the carboxylate group enhances CO₂ absorption.

Effect of CO₂ absorption on IL mobility

The influence of CO₂ absorption on dynamics of cations and anion of IL is examined by Mean Square Displacement (MSD)³⁰ (see Fig. 5). To compare the effect of CO₂ absorption, we simulated a vacuum–IL system, where the MSD of cations and anions are similar. Both cations and anions of IL move faster in CO₂–IL system compared to in vacuum–IL system. In comparison to cations, anions move faster and this is due to preferred interaction of anion with CO₂ (as seen from RDFs as well). Similar trends are seen at other partial pressures (see Fig. S7 in the ESI†). The MD simulations of Perez-Blanco and Maginn³² have shown that mobility of ions of IL remain unaffected with CO₂ absorption where they have calculated the dynamical properties up to 1 ns. Our MSDs supports that their observations are valid only at short timescales. At timescales beyond 1 ns (as seen in the present study), the results show that CO₂ absorption increases the mobility of cations and anions. This mobility arises due to physical interaction of CO₂ molecules at the interface and bulk IL layers. The trends in mobility of cations and anions of IL can be used as a benchmark to explore the dynamical properties of similar class of phosphonium based AAILs. Since a linear regime of MSD could not be found for cations and anions of IL, the diffusion coefficients could not be calculated.

Fig. 5 Mean Square Displacement of cations and anions of IL.

To check the reproducibility of CO₂ absorption events, we have performed MD simulations with different initial configurations and system sizes (see Fig. S8 in the ESI†) used for the CO₂–IL system. The simulations show structural and dynamical properties of CO₂–IL system remain unchanged by the choice of initial configuration (see Fig. S9–S11 in the ESI†).

4. Conclusions

A molecular mechanism of high pressure CO₂ absorption in [P₄₄₄₄][Lys] IL is investigated using MD simulations. The charge densities show a structured ordering with opposite charged layers near the CO₂–IL interface. The particle densities show that the interface is saturated with CO₂ molecules at around 10 ns. A maximum of 1.08 molar absorption of CO₂ can be achieved at T = 278 K and p_CO2 = 20 bar. The RDFs and SDFs in this study are found to be in agreement with previously reported works on other ILs. The interaction and absorption of CO₂ leads to an increase in the mobility of cations and anions of IL, where higher mobility of anions indicate interactions of CO₂ molecules largely with anions. This increase in mobility of ions of IL is supported with the decrease in cation–anion interactions after CO₂ absorption (as seen from RDFs). An immediate outcome of this study will be an examination of the process of CO₂ desorption and the effect of humidification on CO₂ absorption/desorption in IL. Such findings can motivate experimental investigations to screen, synthesize and characterize efficient ILs for CO₂ capture and also optimize the functionality of anions to enhance absorption.

Acknowledgements

The authors thank Dr Collins Assisi for computational resources. Prabhat Prakash thanks IISER Pune for graduate fellowship. AV thanks DST Nanomission (Grant No: SR/NM/NS-15/2011) and DST (SB/S1/PC-015/2013) for providing financial support in this work.

Notes and references

1. Inventory of U.S. Greenhouse Gas Emissions and Sinks: 1990-2013, US Environmental Protection Agency, Washington, DC, USA, 2015, http://https://www3.epa.gov/climatechange/ghgemissions/usinventoryreport/archive.html.
2. M. E. Boot-Handford, J. C. Abanades, E. J. Anthony, M. J. Blunt, S. Brandani, N. Mac Dowell, J. R. Fernandez, M.-C. Ferrari, R. Gross, J. P. Hallett, R. S. Haszeldine, P. Heptonstall, A. Lyngfelt, Z. Makuch, E. Mangano, R. T. J. Porter, M. Pourkashanian, G. T. Rochelle, N. Shah, J. G. Yao and P. S. Fennell, Energy Environ. Sci., 2014, 7, 130–189.
3. G. Puxty, R. Rowland, A. Allport, Q. Yang, M. Bown, R. Burns, M. Maeder and M. Attalla, Environ. Sci. Technol., 2009, 43, 6427–6433.
4. N. McCann, M. Maeder and M. Attalla, Ind. Eng. Chem. Res., 2008, 47, 2002–2009.
5. A. Veawab, P. Tontiwachwuthikul and A. Chakma, Ind. Eng. Chem. Res., 1999, 38, 3917–3924.
6. P. J. Dyson, G. Laurenczy, C. A. Ohlin, J. Vallance and T. Welton, Chem. Commun., 2003, 2418–2419.
7. J. L. Anthony, E. J. Maginn and J. F. Brennecke, J. Phys. Chem. B, 2002, 106, 7315–7320.
8. X. Zhang, X. Zhang, H. Dong, Z. Zhao, S. Zhang and Y. Huang, Energy Environ. Sci., 2012, 5, 6668–6681.
9. J. L. Anthony, J. L. Anderson, E. J. Maginn and J. F. Brennecke, J. Phys. Chem. B, 2005, 109, 6366–6374.
10. 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.
11. B. F. Goodrich, J. C. de la Fuente, B. E. Gurkan, D. J. Zadigian, E. A. Price, Y. Huang and J. F. Brennecke, Ind. Eng. Chem. Res., 2011, 50, 111–118.
12. B. E. Gurkan, T. R. Gohndrone, M. J. McCready and J. F. Brennecke, Phys. Chem. Chem. Phys., 2013, 15, 7796–7811.
13. Z.-Z. Yang, Y.-N. Zhao and L.-N. He, RSC Adv., 2011, 1, 545–567.
14. J. L. McDonald, R. E. Sykora, P. Hixon, A. Mirjafari and J. H. Davis, Environ. Chem. Lett., 2014, 12, 201–208.
15. Y. Zhang, S. Zhang, X. Lu, Q. Zhou, W. Fan and X. Zhang, Chem.–Eur. J., 2009, 15, 3003–3011.
16. E. D. Bates, R. D. Mayton, I. Ntai and J. H. Davis, J. Am. Chem. Soc., 2002, 124, 926–927.
17. J. Zhang, S. Zhang, K. Dong, Y. Zhang, Y. Shen and X. Lv, Chem.–Eur. J., 2006, 12, 4021–4026.
18. P. J. Carvalho, V. H. Álvarez, I. M. Marrucho, M. Aznar and J. A. P. Coutinho, J. Supercrit. Fluids, 2010, 52, 258–265.
19. B. E. Gurkan, J. C. de la Fuente, E. M. Mindrup, L. E. Ficke, B. F. Goodrich, E. A. Price, W. F. Schneider and J. F. Brennecke, J. Am. Chem. Soc., 2010, 132, 2116–2117.
20. C. Wang, X. Luo, H. Luo, D. E. Jiang, H. Li and S. Dai, Angew. Chem., Int. Ed., 2011, 50, 4918–4922.
21. S. Seo, M. Quiroz-Guzman, M. A. Desilva, T. B. Lee, Y. Huang, B. F. Goodrich, W. F. Schneider and J. F. Brennecke, J. Phys. Chem. B, 2014, 118, 5740–5751.
22. R. S. Eisinger and G. E. Keller, Energy Fuels, 2014, 28, 7070–7078.
23. X. Huang, C. J. Margulis, Y. Li and B. J. Berne, J. Am. Chem. Soc., 2005, 127, 17842–17851.
24. M. E. Perez-Blanco and E. J. Maginn, J. Phys. Chem. B, 2010, 114, 11827–11837.
25. H. Xing, Y. Yan, Q. Yang, Z. Bao, B. Su, Y. Yang and Q. Ren, J. Phys. Chem. C, 2013, 117, 16012–16021.
26. J. D. Morganti, K. Hoher, M. C. C. Ribeiro, R. A. Ando and L. J. A. Siqueira, J. Phys. Chem. C, 2014, 118, 22012–22020.
27. G. García, M. Atilhan and S. Aparicio, J. Phys. Chem. C, 2015, 119, 28405–28416.
28. E. I. Izgorodina, J. L. Hodgson, D. C. Weis, S. J. Pas and D. R. MacFarlane, J. Phys. Chem. B, 2015, 119, 11748–11759.
29. M. Klähn and A. Seduraman, J. Phys. Chem. B, 2015, 119, 10066–10078.
30. M. P. Allen and D. J. Tildesley, Computer Simulation of Liquids, Oxford University Press, Oxford, 1987.
31. M. Brehm and B. Kirchner, J. Chem. Inf. Model., 2011, 51, 2007–2023.
32. M. E. Perez-Blanco and E. J. Maginn, J. Phys. Chem. B, 2011, 115, 10488–10499.
33. B. Hess, C. Kutzner, D. van der Spoel and E. Lindahl, J. Chem. Theory Comput., 2008, 4, 435–447.
34. R. T. Cygan, V. N. Romanov and E. M. Myshakin, J. Phys. Chem. C, 2012, 116, 13079–13091.
35. G. Zhou, X. Liu, S. Zhang, G. Yu and H. He, J. Phys. Chem. B, 2007, 111, 7078–7084.
36. S. Nosé, Mol. Phys., 1984, 52, 255–268.
37. W. G. Hoover, Phys. Rev. A, 1985, 31, 1695–1697.
38. L. X. Dang and T.-M. Chang, J. Phys. Chem. Lett., 2012, 3, 175–181.
39. 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.
40. S. Saravanamurugan, A. J. Kunov-Kruse, R. Fehrmann and A. Riisager, ChemSusChem, 2014, 7, 897–902.
41. B. L. Bhargava and S. Balasubramanian, J. Phys. Chem. B, 2007, 111, 4477–4487.
42. W. Shi, C. R. Myers, D. R. Luebke, J. A. Steckel and D. C. Sorescu, J. Phys. Chem. B, 2012, 116, 283–295.
43. X. Luo, Y. Guo, F. Ding, H. Zhao, G. Cui, H. Li and C. Wang, Angew. Chem., Int. Ed., 2014, 53, 7053–7057.
44. W. Humphrey, A. Dalke and K. Schulten, J. Mol. Graphics, 1996, 14, 33–38.

---

Footnote

† Electronic supplementary information (ESI) available: Initial configuration of CO₂–IL slab, charge densities, particle densities of CO₂, MSD of ions, molar absorption ratio and effect of initial configurations on structural and dynamical properties. See DOI: 10.1039/c6ra09577a

---