High CO2 absorption capacity by chemisorption at cations and anions in choline-based ionic liquids

The effect of CO2 absorption on the aromaticity and hydrogen bonding in ionic liquids is investigated. Five different ionic liquids with choline based cations and aprotic N-heterocyclic anions were synthesized. Purity and structures of the synthesized ionic liquids were characterized by H and C NMR spectroscopy. CO2 capture performance was studied at 20 1C and 40 1C under three different pressures (1, 3, 6 bar). The IL [N1,1,6,2OH][4-Triz] showed the highest CO2 capture capacity (28.6 wt%, 1.57 mol of CO2 per mol of the IL, 6.48 mol of CO2 per kg of the ionic liquid) at 20 1C and 1 bar. The high CO2 capture capacity of the [N1,1,6,2OH][4-Triz] IL is due to the formation of carbonic acid (–OCO2H) together with carbamate by participation of the –OH group of the [N1,1,6,2OH] + cation in the CO2 capture process. The structure of the adduct formed by CO2 reaction with the IL [N1,1,6,2OH][4-Triz] was probed by using IR, C NMR and H–C HMBC NMR experiments utilizing C labeled CO2 gas. H and C PFG NMR studies were performed before and after CO2 absorption to explore the effect of cation–anion structures on the microscopic ion dynamics in ILs. The ionic mobility was significantly increased after CO2 reaction due to lowering of aromaticity in the case of ILs with aromatic N-heterocyclic anions.


Introduction
Ionic liquids (ILs) are typically low melting salts (usually o100 1C) comprising organic or inorganic cations and anions.
Unique physicochemical properties such as high ionic conductivity, low vapor pressure, non-flammability and functional designability make ILs suitable sorbents for CO 2 as compared to the conventional molecular liquids. Though ILs have a wide range of properties, however, viscosity is always a major concern when they are used as CO 2 sorbents. Due to the presence of cations and anions in the same system, ILs are usually more viscous than other conventional molecular liquids. [1][2][3] Often ILs are regarded as designer solvents due to their tailor-made physicochemical properties by the manipulation of cation and anion structures. Extensive research efforts have already been made to understand and predict the macroscopic physicochemical properties of the ILs on the basis of their ionic interactions, microscopic ion dynamics, etc. Despite the extensive research efforts, the influence of cationic and anionic structures on the physicochemical properties of the ILs still remains a challenge. Thus, investigation of the ionic mobility of ILs is a promising area of research for the development of high performance ILs for challenging applications. 4,5 Diffusivity (also known as self-diffusion) of molecules, ions and their aggregates is a property of any liquid as a result of thermal motion. 6 This property determines many of the physicochemical characteristics of a liquid such as viscosity, ionic conductivity, and adsorption, and also influences the mechanisms and rates of chemical reactions. In this regard, investigation of the diffusivity of ILs can help in gaining new insights and a better understanding of their ion dynamics, transport properties 7,8 and other technological processes. 7 Self-diffusion in ILs generally obeys the laws of diffusivity of molecules in liquids such as Gaussian statistics for mean-squared displacements and the Stokes-Einstein equation. 6 Moreover, selfassociation phenomena are quite typical for many ILs 9 that lead to a specific diffusivity of ions. [10][11][12]37 Absorption of gases is another type of association that also effects the diffusivity of ions. 13 Nuclear magnetic resonance (NMR) spectroscopy is an appropriate technique for the investigation of the diffusion of ions and sorbent additives in the ILs. 14 Because of the prevalent industrial applications of ILs, they are also regarded as promising CO 2 sorbents. However, the higher viscosity of ILs compared to the conventional amine sorbents remains a challenge, limiting their applicability as CO 2 sorbents. Recently, great attention has been paid by researchers to develop less viscous IL CO 2 sorbents as a possible replacement for the conventional amine sorbents. The most promising ILs as CO 2 sorbents are amino acid based ionic liquids with phosphonium and ammonium cations. 15 One of the possible reasons is that amino acid based ILs have low toxicity and faster reactivity towards CO 2 molecules. Recently we have reported that functionalized choline based amino acid ILs have excellent CO 2 capture capacity at room temperature and atmospheric pressure. 16 Additionally, the ILs with choline based cations are more attractive due to their low toxicity and biodegradability. [17][18][19] At the same time, ILs containing amino acid anions with simple choline cations are highly viscous due to hydrogen bonding. We have found that etherification of choline cations leads to a dramatically low viscosity. 16 Also by replacing amino acid anions with N-heterocyclic anions, the viscosity of the ILs decreased significantly with improved CO 2 capture capacity. 13 In addition, it was observed that most of the amino acid based ILs become highly viscous or solidified after CO 2 absorption. However, aprotic N-heterocyclic anion (AHA) based ILs remained liquid with much faster ionic mobility after reaction with CO 2 molecules. Detailed insights into the comparative physicochemical and CO 2 capture studies of amino acid and N-heterocyclic amine based ILs with a common cation [N 1,1,6,2OH ] + are recently reported. 13 CO 2 capture studies of some aprotic N-heterocyclic based anions with a phosphonium cation [P 6,6,6,14 39 The regeneration of ILs often needs heating (up to 80 1C) with a continuous flow of nitrogen gas into the system. Furthermore, these CO 2 absorption capacities are significantly lower as compared to that of the traditional 2-aminoethanol (MEA or monoethanolamine) that has 29.9 wt% (0.42 mol) as a neat sorbent and 13.1 wt% capacity can also achieved by using 30% aqueous 2-aminoethanol (MEA) solution. 42 There is an urgent need to replace 2-aminoethanol with environmentally benign ILs having higher CO 2 capture capacity.
Apart from AHA based ILs, various protic ILs have also been reported for the CO 2 capture process. The major advantage of protic ILs is their low viscosity compared to AHA based ILs. However, CO 2 capture capacity and the process of regeneration of ILs have still not been improved. Zhu 41 CO 2 capture by diamino based protic ILs has been reported by MacFarlane's group, although the maximum CO 2 uptake achieved was 12.6 wt% (0.47 mol CO 2 per mol of IL). 42 The CO 2 capture capacity of these ILs is not substantial (o20 wt%) compared to that of the traditional 2-aminoethanol (MEA) which also requires the energy demanding regeneration process to recycle the ILs. Therefore, there is a continuous quest for the development of new ionic liquids with high and reversible CO 2 absorption capacity with minimum energy input.
In this work, we investigate a comparison of the physiochemical properties as well as CO 2 capture capacities of different N-heterocyclic anion based ILs with functionalized choline as the cation. Five ILs having choline based cations and different aprotic N-heterocyclic anions were synthesized and characterized. The functionalized choline cation [N 1,1,6,2OH ] is further modified by etherification to study the effect of hydrogen bonding on the CO 2 absorption capacity, viscosity and ion mobility. FTIR, 1 H and 13 C NMR spectroscopic techniques were employed to characterize the structure of the adduct formed as a result of CO 2 reaction with the IL. In addition, 1 H NMR self-diffusion measurements were carried out on neat ionic liquids as well as after CO 2 absorption experiments using 1 H and 13 C pulsed field gradient (PFG) NMR techniques to understand the ionic mobility in these ILs. Density functional theory (DFT) calculations were performed to further support the experimental findings of the effect of CO 2 absorption on the ionic mobility of the ionic liquids.

Synthesis and characterization
All the reactants were purchased from Sigma Aldrich and used without further purification. Thin layer chromatographic analysis was carried out using silica gel 60 Å F-254 TLC plates with KMnO 4 charring solution.
All the ionic liquids were prepared using our previously reported procedures. 13 The excess of unreacted N-heterocycles was removed by adding excess of ACN and was stored over night at 0 1C. All the ILs were dried under vacuum at 60 1C for at least 2 days.
The 1 H and 13 C NMR characterization data for the synthesized ILs are given in the ESI. † Water content was measured using a Metrohm 917 Karl Fischer Coloumeter with a HYDRANAL reagent. It was found that all ILs contain a water content of o0.5%.

Physicochemical characterization
The thermogravimetric analysis (TGA) of all the five ILs was performed using a Perkin Elmer 8000 TGA instrument in the temperature range from 30 1C to 550 1C with a heating ramp of 10 1C min À1 under a nitrogen atmosphere. 2-3 mg of the ionic liquid sample was used for each experiment. All the TGA experiments were performed under nitrogen gas as the inert carrier gas.
A Lovis 2000 ME Automated Microviscometer (Anton-Paar falling ball type viscometer with 2.50 mm glass capillary, viscosity range 10-10 000 cP) was used for viscosity measurement. The viscosity of the ILs was measured in the temperature range from 20 1C to 80 1C with a step size of 5 1C using a sealed glass capillary sample tube.

NMR and FTIR spectroscopy
A Bruker Ascend Aeon WB 400 (Bruker BioSpin AG, Fällanden, Switzerland) NMR spectrometer was used with a working frequency of 400.22 MHz for 1 H and 100.65 MHz for 13 C. Bruker Topspin 3.5 software was used for the processing of data. All the spectra were recorded at 25 1C. Chemical shifts were expressed in parts per millions (d) downfield from DSS with the solvent resonance as the internal standard (D 2 O, d = 4.79) and were reported as s (singlet), d (doublet), t (triplet), q (quartet), br (broad) and m (multiplet). All coupling constants ( J) are reported in Hertz (Hz).
The 1 H and 13 C pulsed field gradient (PFG) NMR selfdiffusion experiments were performed on a Bruker Ascend Aeon WB 400 instrument using a Diff50 (60 A) z-gradient 5 mm diffusion probe. A stimulated echo pulse sequence was employed, 14,20 with the diffusion coefficient (D) determined from the signal decay using the Stejskal and Tanner equation 20 where A(g) and A(0) are the integral signal intensities obtained with and without gradient, respectively, g is the gyromagnetic ratio for a used nucleus; g is the gradient strength, d is the duration of the gradient pulse, t d = (D À d/3) is the diffusion time, and D is the delay between the gradient pulses. A sample for the study was placed in a standard 5 mm glass sample tube and closed with a plastic stopper to avoid contact with air. Before each measurement, the sample was equilibrated at the specified temperature for 20 minutes. The experimental details of the PFG NMR technique can be found elsewhere. 14 Typical experimental conditions for 1 H PFG NMR included signal averaging with 16 scans, 32 gradient steps, a total gradient pulse length of d = 1-2 ms, and a 5 s recycle delay. D was in the range from 30 to 200 ms. For the 13 C PFG NMR experiment the experimental conditions were: 3600 scans, 12 gradient steps, a total gradient length of d = 0.1-2 ms, and a 5 s recycle delay. D was 10 ms. Fourier transform infrared (FTIR) spectra were recorded on a Bruker IFS 80v vacuum Fourier transform infrared spectrometer equipped with a deuterated triglycine sulphate (DTGS) detector. All spectra were recorded at room temperature (B22 1C) using the double side forward-backward acquisition mode. A total number of 256 scans were co-added and signal-averaged at an optical resolution of 4 cm À1 . CO 2 absorption measurement CO 2 capture experiments were performed using B0.1 g of neat ILs in a 10 ml Buchi stainless steel reactor with a PTFE insert and a pressure gauge. The CO 2 capture capacities of the ILs were measured at different pressures (1-6 bar) of CO 2 (99.999 by weight%) at 20 1C and 40 1C by weighing the PTFE insert gravimetrically using a Mettler Toledo analytical balance (0.1 mg accuracy).

Computational methods
DFT calculations were performed using GAUSSIAN 09 21 software packages. The B3LYP/6-311+g (d,p) level of theory was used for molecular geometry optimization. The optimized molecular geometries were characterized to be a relative energy minimum to the potential energy surface by real frequency. For NICS calculations the aug-cc-pvtz basis set was used. However, all calculations are performed under gas phase conditions, in the current calculations cations and the solvent model are not included as the solvent properties are not available for the IL structure.

Results and discussion
Five ILs with four different aprotic heterocyclic anions (imidazole, triazole, pyrazole, succinimide) were synthesized by following our previously reported procedures. 13  [Imi] the activation energies for viscous flow are 59.7 kJ mol À1 K À1 and 57.5 kJ mol À1 K À1 , respectively. It is interesting to note that the activation energy for viscous flow significantly decreased to 50.6 kJ mol À1 K À1 upon ether functionalization of the cation [N 1,1,6,2OH ]. In addition, the activation energy for viscous flow increased significantly to 76.8 kJ mol À1 K À1 when an aromatic heterocyclic anion was replaced by a non-aromatic heterocyclic anion (succinimide anion).
It is observed that the IL with a non-aromatic heterocyclic anion (succinimide) was highly viscous at room temperature (c10 000 cP) compared to the other ionic liquids with aromatic anions. To further investigate the effect of hydrogen bonding on the viscosity, we measured the viscosity of the ether functionalized IL [N 1,1,6,2O4 ]  in the same temperature range. From our previously reported work, we envision that etherification of choline based cations will lead to reduced H-bonding interactions resulting in less viscous ILs. 16 (Fig. 2). A slight increase in the CO 2 capture capacity was observed with the increase in pressure to 6 bar. The CO 2 uptake was 19.8 wt% (1.34 mol CO 2 per mol of IL) at 6 bar. The IL [N 1,1,6,2O4 ][Pyrz], which is a semi-solid at room temperature, showed a lower CO 2 capture capacity (7.18 wt%, 0.48 mol of CO 2 per mol of IL) at 20 1C under 1 bar pressure compared to the aromatic N-heterocyclic anion based ILs. Even at higher pressure no such significant increase in the CO 2 (Fig. 3). The IL [N 1,1,6,2O4 ][Pyrz] showed a relatively lower CO 2 capture capacity (B0.55 mol of CO 2 per mol of IL, B8 wt% CO 2 ) at 40 1C, which is due to the less thermally stable carbamate anion formed after reaction with CO 2 molecules.
As the IL [N 1,1,6,2OH ]  showed the highest CO 2 capture capacity at 20 1C compared to the other studied ILs, we further studied the recyclability of this particular IL. The desorption experiments were performed using a continuous stirring under vacuum (10 À3 bar) at 20 1C for 3 hours. Interestingly, no significant decrease was observed in the CO 2 uptake performance even after the 4th cycle of sorption-desorption. The CO 2 capture capacity was B27 wt% after the 4th cycle. These data suggest that the IL [N 1,1,6,2OH ][4-Triz] can be regenerated in an energy efficient way rather than traditional nitrogen gas purging together with heating up to 80 1C (Fig. 4).
In order to gain deeper insights into the interactions of the CO 2 molecule with the IL, the [N 1,1,6,2OH ][4-Triz]-CO 2 complex was characterized after CO 2 reaction using NMR and IR spectroscopic techniques. In the 1 H NMR spectrum of the IL [N 1,1,6,2OH ] , an aromatic proton of the 1,2,4-triazolate anion  revealed a signal at 8.09 ppm before CO 2 absorption, which was shifted to 8.32 ppm after CO 2 absorption (see S5 in the ESI †). In the 13 C NMR spectrum, the signal for aromatic carbon at 153.54 ppm is shifted to 149.85 ppm along with a new signal of carbamate at 163.88 ppm after CO 2 absorption (see S11 in the ESI †). This major shift in the aromatic carbon of the 1,2,4-triazolate anion  is due to the lowering of aromaticity in the triazole ring after carbamate formation, which is further confirmed by DFT     (Fig. 6) it is observed that the N-4 nitrogen atom is the most negatively charged (À0.31) as compared with the other two nitrogen atoms (À0.24). Thus, carbamate formation is preferred at the N-4 position, which is further confirmed by NMR having one signal in the aromatic region. The IL [N 1,1,6 CO 2 atmosphere and allowed to react with the IL for several hours. Then, the 13 CO 2 enriched IL sample was placed co-axially in 10 mm NMR with D 2 O as an external lock for the 13 C NMR measurement. Interestingly, two new signals at 130.57 ppm and 158.67 ppm were observed in addition to the carbamate signal at 163.70 ppm. The new intense sharp signal at 130.57 ppm is due to the physically absorbed 13 CO 2 , whereas the signal at 158.67 ppm is probably due to the formation of carbonic acid (-OCO 2 H) with the -OH group of the [N 1,1,6,2OH ] cation (see S12 in the ESI †). 23 Surprisingly, when the 13 C NMR measurement of the 13 CO 2 enriched sample dissolved in D 2 O was carried out, the signals at 130.57 ppm and 158.67 ppm disappeared again. However, after a large number of acquisitions, a small signal at 161.51 ppm was observed due to the formation of carbonate with the -OH group of the [N 1,1,6,2OH ] cation, which was further confirmed by the 1 H-13 C 2D heteronuclear multiple bond correlation (HMBC) experiment and found to be less stable in the solution form. The 1 H-13 C HMBC spectra are shown in the ESI † (S8) (Scheme 1).
To further confirm these findings, FTIR measurements were carried out to characterize the IL-CO 2 adducts. In the FTIR spectrum of the IL [N 1,1,6,2OH ][4-Triz] after CO 2 reaction, three additional stretching bands at 1677 cm À1 , 1635 cm À1 and a small band 2338 cm À1 were observed. The band at 1677 cm À1 is due to the formation of carbamate after CO 2 reaction. The band around 1635 cm À1 is attributed to the formation of carbonic acid, which disappeared by keeping the sample under vacuum (see S12 in the ESI †). The small band at 2338 cm À1 signified the presence of physically absorbed CO 2 in the IL [N 1,1,6,2OH ] . 24 1 H NMR diffusometry was carried out to investigate the self-diffusion of ions before and after CO 2 absorption. 1 H NMR signals of stimulated echo for all the studied ILs before and after absorbed CO 2 were observed in the whole temperature range from 20 1C to 90 1C, except for the IL [N 1,1,6,2OH ][Succ], for which the signal of the 1 H NMR stimulated echo was observed only at temperatures equal to and higher than 40 1C. This is due to the accelerated T 2 NMR relaxation of protons at lower temperatures (o40 1C) for this sample. Diffusion coefficients of anions and cations were calculated from the diffusion decays of the corresponding lines in the 1 H NMR spectra. All decays were well fitted in the equation of type eqn (1). Temperature dependences of D for the cations and anions before and after CO 2 adsorption are shown in Fig. 6.
A typical approach to the analysis of temperature dependence of D of ionic liquids is the Arrhenius type equation, 6 which described the temperature dependence of Ds in the form where D 0 is a parameter that is independent of temperature, E D is the molar activation energy of diffusion and R is the gas constant. However, this equation does not fit experimental dependences in many cases. 25 More universal forms of temperature dependences taking into consideration proximity of the system temperature to their glass transition temperature, T 0 , is a Vogel-Fulcher-Tamman (VFT) VFT equation in the following form for diffusivity: 25 where T 0 and B are adjustable parameters, E D = BÁR. We have described D(T) in Fig. 6 by fitting D 0 , T 0 and B. B is a factor related to the activation energy and T 0 indicates a temperature at which free volume and mobility are reduced to zero. 43 The procedure of fitting was performed in two steps, as described previously. 44 In the first step, ln(D) was plotted against 1/(T À T 0 ) and T 0 was selected to have a dependence linear. Uncertainty in this step was AE10 K. In the second step, the dependence was fitted by a linear regression to obtain the fitting parameters (D 0 , B). Best results of the fitting are shown by solid (anions) and dotted (cations) lines in Fig. 5 and the corresponding fitting parameters as well as E D are presented in Table 1. The temperature dependence of diffusivity for ions of all studied ILs fitted well into the VFT model over the whole studied temperature range. From the 1 H diffusometry data, we observed some general features of self-diffusion of ions before and after CO 2 absorption in the studied ILs.
(1) There are two types of temperature dependences of diffusion coefficients for ILs before and after CO 2 absorption: firstly, the diffusion of cations and anions is equal for [N 1,1,6,2O4 ][Pyrz] (Fig. 5B) and [N 1,1,6,2OH ][Succ] (Fig. 5E). Secondly the diffusion of anions is higher than the diffusion of cations for other three ILs (Fig. 5A-D). The first type reveals weak dissociation of ions while the second type suggests higher dissociation of ions. This difference in dissociation of ions is apparently related to the form of temperature dependence of Ds. In the first type, the Arrhenius type of temperature dependence (eqn (2)) is observed while in the second case VFT type (eqn (3)) is demonstrated. From Table 1, it is seen that for the first type of temperature dependence of ILs is characteristic of low T 0 and higher E D as compared with  Generally, the physisorption and chemisorption of CO 2 molecules are expected to increase the mass of the diffusion particle and thus decrease the diffusion coefficients of ions. However, the observed increase in the diffusion coefficients of ions in this case is most probably due to the decrease in the inter-ion interactions after CO 2 absorption in these ILs. 13 26 The diffusion coefficient of the physically absorbed CO 2 gas is B10 6 higher than the chemically bound one (B10 À7 m 2 s À1 and B10 À13 m 2 s À1 , respectively at 293 K). It is known that the diffusion coefficient of CO 2 at normal pressure and temperature is B10 À5 m 2 s À1 . 27 Although the diffusion coefficient of CO 2 in the IL is less than that in the gas phase, however, the mobility of CO 2 is still very high compared to other components in the IL. The diffusivity of dissolved CO 2 in the IL is higher as compared with the CO 2 gas dissolved in water. The diffusivity of CO 2 dissolved in water is B10 À9 m 2 s À1 , 28 which is very close to the diffusion coefficient of water molecules. This is due to the strong network of hydrogen bonds formed in water, which incorporates CO 2 molecules and products of its dissociation. In contrast, the physically dissolved CO 2 does not interact that strongly with the IL. This is one of the plausible reasons that the dissolved CO 2 shows a relatively high translational mobility in the IL system.
In order to further support the experimental findings, we performed NICS (Nucleus-Independent Chemical Shifts) calculations using density functional theory (DFT) to evaluate the aromaticity behavior of the N-heterocyclic anions before and after CO 2 reaction. The NICS method is based on the negative of the magnetic shielding computed at the centre of the aromatic ring. A positive value implies antiaromaticity (paratropic ring current) and a negative value corresponds to aromaticity (diatropic ring current). [29][30][31][32][33] Conventionally, NICS values are at the geometrical center (GC) of the ring denoted as NICS(0) and 1 Å above/below the perpendicular plane of the ring denoted as NICS (1). In this case due to the presence of symmetry in N-heterocyclic anions, both the ring critical point (RCP) and geometrical center (GC) coincide. According to the literature, NICS(1) (1 Å above/below the plane of the ring) is the best measure of the aromaticity descriptor than NICS(0) due to aromatic ring current and spurious contributions of the in-plane tensor components at the geometrical center of the ring. Further, the minimum effect of the local s-bonding contributions is also observed at 1 Å above/below the molecular plane, thus, the out of plane tensor component of the NICS (1) values, NICS zz (1) shows an even better index of aromaticity. 34,35 Current NICS calculations were computed through the gauge-including atomic orbital method (GIAO) implemented in GAUSSIAN 09 21,36 program using the B3LYP dft functional. Due to basis set dependencies of NICS values the aug-cc-pvtz basis set was used for all NICS calculations. Ghost atoms (symbol ''Bq'' from Gaussian 09 input) were used as the NICS probe without using basis functions and placed 1 Å above along the line perpendicular to the molecular plane and at the center of the molecular plane.
It is observed that the NICS(0) and NICS (1) values are less influenced after carbamate formation with N-heterocyclic anions in the ILs (Table 2). Whereas NICS zz (1) values increased significantly after carbamate formation as a result of CO 2 reaction with the N-heterocyclic anions of the ILs. The DFT data suggested a decrease in aromaticity of the anions after carbamate formation, which results in less pi-pi stacking interactions between the anions. The DFT optimized (gas phase) structures of 1,2,4-triazolate and the carbamate anion are shown in Fig. 6. The reduced pi-pi stacking interactions lead to faster diffusion of the anions and thus result in low viscosity of the ionic liquids upon CO 2 absorption.

Conclusions
Ether functionalization of a choline based cation in the IL [N 1,1,6,2O4 ]  resulted in lowering of viscosity due to the elimination of hydrogen bonding as compared with a nonfunctionalized choline based IL [N 1,1,6,2OH ] . However, CO 2 capture performance as well as thermal stability was decreased significantly by ether functionalization on the cation. The IL [N 1,1,6,2O4 ][Pyrz] was found to be a semi-solid at room temperature and slow dissociation between cations and anions in the bulk system was observed from 1 H diffusion NMR data. The IL [N 1,1,6,2OH ][Imi] was less viscous and diffused faster as compared with the IL [N 1,1,6,2OH ]  due to the presence of lesser number of hydrogen bond acceptor atoms. A significant dissociation between cations and anions was observed in both the ILs. The IL [N 1,1,6,2OH ][Succ] with non-aromatic N-heterocyclic anions was highly viscous and showed a lower CO 2 capture capacity compared to the ILs with aromatic N-heterocyclic anions. In addition, a very slow ionic mobility with low dissociation between cations and anions in the bulk system was observed. The IL [N 1,1,6,2OH ]  showed the highest CO 2 capture capacity (28.6 wt%, 1.57 mol of CO 2 per mol of IL, 6.48 mol of CO 2 per kg of IL) at 20 1C and 1 bar pressure. in comparison with the previously reported N-heterocyclic (B8 wt%) ILs. The high CO 2 capture capacity of the IL [N 1,1,6,2OH ][4-Triz] is due to the contribution of the cholinebased cation during the CO 2 capture process by the formation of carbonic acid (-OCO 2 H) with the -OH group of the [N 1,1,6,2OH ] + cation. The [N 1,1,6,2OH ][4-Triz]-CO 2 adduct structure was confirmed by FTIR, 13 C NMR and 1 H-13 C HMBC NMR experiments utilizing 13 C labeled 13 CO 2 . The recyclability performance of [N 1,1,6,2OH ]  showed that the IL can be recycled under vacuum at room temperature without using any nitrogen gas and heating. It is worth noting that the ionic mobility was increased significantly after CO 2 reaction in the case of ILs with aromatic N-heterocyclic anions, which can open a new area of research for low viscosity ionic liquids with promising CO 2 capture performance.

Conflicts of interest
The authors declare no conflicts of interest.