Role of water in complexation of 1,4,7,10,13,16-hexaoxacyclooctadecane (18-crown-6) with Li+ and K+ in hydrophobic 1-ethyl-3-methylimidazolium bis(trifluoromethanesulfonyl)amide ionic liquid

Complexation characteristics of 1,4,7,10,13,16-hexaoxacyclooctadecane (18-crown-6, 18C6) with Li+ and K+ in a hydrophobic ionic liquid of 1-ethyl-3-methylimidazolium bis(trifluoromethanesulfonyl)amide under dry and humid conditions at 298.2 K were studied by 1H and 13C NMR chemical shifts. The comparison of the 1H and 13C chemical shifts of 18C6 molecule between the dry and humid IL solutions without the alkali metal ions showed that uncomplexed 18C6 molecules are solvated by water molecules in the humid ionic liquid solution. The changes in the 1H and 13C chemical shifts of 18C6 ligand molecule with the increases in the Li+ and K+ concentrations revealed that in both dry and humid ionic liquid solutions 18C6 molecule forms 1:1 complexes with Li+ and K+. The 1H NMR data of water molecules in the humid ionic liquid solutions demonstrated that water molecules interact with Li+-18C6 complexes and free Li+, but do not with K+-18C6 complexes and free K+. The mechanisms of the formation of the Li+ and K+ complexes in the humid ionic liquid solution are different from each other due to the differences in the complex-water interactions.


Introduction
Ionic liquids (ILs) have come to attract attention as alternatives to volatile organic solvents for solvent extraction because they have several excellent properties, e.g., low volatility, low flammability, and high thermal and chemical stabilities [1]. Crown ether molecules are one of extracting agents to effectively transfer alkali and alkaline-earth metal ions from aqueous phase to organic liquid phase [2]. A number of fundamental data on the complexation of crown ethers with ionic species and neutral molecules are available in the literature [3]. Also, there are many reports on the complexation between metal ions and crown ether molecules in hydrophobic ILs from an engineering standpoint, as reviewed [1]. Water is the key to govern the complexation because a trace of water in ILs drastically changes the physical and electrochemical properties of the entire system [4][5][6][7][8]. Dietz and co-workers have determined the structure of the complex of dicyclohexyl-18-crown-6 (DC18C6) with Sr 2? transferred from aqueous solution to hydrophobic 1-alkyl-3-methylimidazolium bis(trifluoromethanesulfonyl)amide solution by EXAFS technique [9]. They have elucidated that each of two water molecules exists in the perpendicular direction of the DC18C6 plane of the Sr 2? -DC18C6 complex. So, the effect of a trace of water in hydrophobic ILs cannot be disregarded in understanding the mechanism of the complexation of crown ether with metal ions. To clarify what role water molecules play in the complexation between crown ether and metal ions in hydrophobic ILs, it is essential to make clear the complexation in hydrophobic IL system with and without water. Although a molecular dynamics simulation has given the effect of water in hydrophobic ILs on the complex structure and the binding affinities of 1,4,7,10,13,16-hexaoxacyclooctadecane (18crown-6, 18C6) [10], any experimental evidence has not yet been reported. Hence, we tried to elucidate the effect of water on the complexation of 18C6 with Li ? and K ? in 1-ethyl-3-methylimidazolium bis(trifluoromethanesulfonyl)amide ([C 2 mim][TFSA]) solutions at 298.2 K under both dry and humid conditions using 1 H and 13 C NMR techniques. The structure of C 2 mim ? with the notation of the H and C atoms is illustrated in Fig. 1.

Chemicals
The IL used in this study was prepared and purified in our laboratory according to the same procedures as those in the previous report [11]. 18C6 (99 %) was purchased from Sigma-Aldrich Co. and used without further purification. Two alkali metal salts of lithium bis(trifluoromethanesulfonyl)amide (LiTFSA, 99.95 %) and potassium bis(trifluoromethanesulfonyl)amide (KTFSA, 99.8 %) were obtained from Sigma-Aldrich Co. and Kanto Chemical Co., respectively. The two alkali metal salts were dried for more than 24 h under reduced pressure at *353 K before solution preparations. Deuterium oxide (99.9 D atom%) as a lock solvent and sodium 3-(trimethylsilyl)-1-propanesulfonate (DSS, 97 %) as an external reference substance were purchased from Sigma-Aldrich Co. and used without further purification.

Sample preparation
The IL solvent under the dry condition was obtained by heating at *333 K under reduced pressure for more than 72 h. The water content determined by a Karl-Fisher titration was 384 ppm. 18C6 was diluted with the IL solvent to reach the 18C6 concentration of 0.139 mol kg -1 , corresponding to *0.200 mol dm -3 . Each of the alkali metal salts (LiTFSA and KTFSA) was also dissolved into the dry IL solvent at the metal ion concentration of 0.137 mol kg -1 , corresponding to *0.200 mol dm -3 . The dry IL solutions of the alkali metal ions were mixed with the dry 18C6-IL solution at desired molar ratios of the alkali metal ion to 18C6. The dry IL solutions including both 18C6 and the alkali metal ion were diluted with the dry IL solvent again. The final concentrations were 0.0133 mol kg -1 for 18C6, up to 0.0403 mol kg -1 for the alkali metal ions, and 0.0213 mol kg -1 for water. Water and neat IL were agitated for 24 h at room temperature. The humid IL solvent was then obtained by separating the IL phase from two phase mixture of water and IL after standing for 24 h at *333 K. The humid IL solvent had the water content of 2.09 9 10 4 ppm. The humid IL solutions at various molar rations of the alkali metal ions to 18C6 were prepared in a similar manner to the preparation of the dry IL solutions. The final concentrations of 18C6, the alkali metal ions, and water in the humid IL solutions were 0.0134 mol kg -1 , up to 0.0404, and 1.16 mol kg -1 , respectively. To avoid the change in water concentration with passage of time, immediately after the sample preparations, a small portion of each of the IL solutions was expeditiously sealed into an NMR inner tube (Shigemi, SC-002).

NMR measurement
A double tube constructed by a pair of the inner tube sealed with the IL solution and an outer tube filled with deuterium oxide solution including 1 wt% DSS was held for 30 min or more at 298.2 ± 0.1 K in a superconducting magnet. 1 H and 13 C NMR spectra were obtained with a JEOL ECA-500 spectrometer, where 1 H and 13 C resonance frequencies were 500.2 and 125.8 MHz, respectively. The digital resolutions of 1 H and 13 C NMR spectra were 0.14 and 0.48 Hz. The chemical shifts of 1 H and 13 C peaks of the IL sample solutions were determined using the peak of DSS as an external reference. The magnetic susceptibility correction for the dry IL systems has little effect on the 1 H and 13 C chemical shifts because the concentrations of 18C6, the alkali metal ions, and water are considerably lower than that of [C 2 mim][TFSA] (*2.5 mol kg -1 ). For the humid IL systems, in addition to the lower concentrations of 18C6 and the alkali metal ions, the volume susceptibility (-0.701 9 10 -6 at 298.2 K) of the [C 2 mim][TFSA]-water binary mixture, which is estimated from the reported values of molar susceptibility, density, and molar mass of pure [C 2 mim][TFSA] and water [11,12], is almost equivalent to the volume susceptibility of pure [C 2  (-0.700 9 10 -6 at 298.2 K) measured by applying the external double reference technique [13] within ±0.1 %. It can be considered that there is no effect of water dissolved into the [C 2 mim][TFSA] on the volume susceptibilities of the sample solutions. For these reasons, we did not make magnetic susceptibility correction to the observed 1 H and 13 C data in this work.
Formation constants (K f ) K f values can be obtained from the chemical shifts of 18C6 as a function of molar ratio of the alkali metal ion to 18C6 [14]. Briefly, in the case that 18C6 molecule forms only 1:1 complex with M ? ([M(18C6)] ? ), the 18C6 molecule was in equilibrium between complexed and uncomplexed forms.
When the exchange between complexed and uncomplexed 18C6 is faster than NMR timescale, the observed NMR peak (d obs ) of 18C6 can be expressed by using the chemical shifts of uncomplexed (d L ) and complexed forms (d ML ) as follows: where x L and x ML are the mole fractions of uncomplexed and complexed 18C6, respectively. The mole fractions x L and x ML are given by and Here, [ML] is the concentration of 1:1 M ? -18C6 complex. C L is the total concentration of 18C6. In 1:1 M ? -18C6 complexation reaction, the formation constant (K f ) is defined by where C M is the total concentration of alkali metal ion. If Eq. (5) is rearranged using Eqs, (2)-(4), In this study, the K f values of 1:1 M ? -18C6 complexes were determined from the change in 13 ) in the dry one beyond the digital resolutions. This is attributed to the difference in the solvation structures of the uncomplexed 18C6 molecules between the dry and humid IL solutions. We have recently reported that water molecules in the humid IL solutions construct waterrich domains by hydrogen bonding among them [11]. In other words, IL-rich domains may simultaneously form in the humid IL solutions. The uncomplexed 18C6 molecules in the humid IL systems may be accommodated in the water-rich domains. In contrast, 18C6 molecules are solvated by C 2 mim ? and/or TFSAin the dry IL solutions. On the assumption that 18C6 molecules are in an equilibrium between only the uncomplexed and the 1:1 metalcomplexed forms, we attempted to determine the formation constants (K f ) of the Li ? -and K ? -18C6 complexes in the IL solutions through Eq. (6). The 13 C nuclei within 18C6 molecule are more sensitive to the formation of the 18C6 complexes with the alkali metal ions than the 1 H nuclei because the C atoms bound to the ether oxygen atoms as the coordination sites for a metal ion. The d L , d ML , and K f values for the four IL systems determined by the leastsquare regression on the d C data of the 18C6 molecules with Eq. (6) are summarized in Table 1. To our knowledge, there are a few literature data of logK f of the Li ? -18C6 complex in IL systems: 1.8 in 1-ethyl-3-methylimidazolium chloride with 55 mol% aluminium(III) chloride [15] and 2.0 in 1-butyl-3-methylimidazolium bis(trifluoromethanesulfonyl)amide [16]. Judging from the K f values of the Li ? -18C6 complexes in the literature, the present values for the [C 2 mim][TFSA] systems are reasonable. On the other hand, the K f value of the K ? -18C6 complex in ILs is not available. In both dry and humid IL solutions, the K f values of the K ? -18C6 complex are larger compared with those for the Li ? -18C6 ones. This arises from the higher size fitness of K ? (ionic radius = 0.137-0.164 nm [12] ) than Li ? (ionic radius = 0.059-0.092 nm [12] ) to the 18C6 cavity (cavity radius = 0.130 nm [17] ). The important finding in the present results is that water molecules dissolved in the IL solutions hinder the formation of the Li ? -and K ? -18C6 complexes. The inhabitation of the alkali metal-18C6 complex formation on adding water is ascribed to the strong hydration of the uncomplexed 18C6 molecules by water molecules in the humid IL solutions.
Focusing on the d L values of the uncomplexed 18C6 molecule in Table 1, for both dry and humid IL solutions, the values of the Li ? -and K ? -18C6 systems are in agreement with each other. However, a comparison between the d L values for each metal system in the dry and humid IL solutions shows a slight difference, e.g., d L = 72.4 and 72.2 ppm for the Li ? system, respectively. This comes of the difference in the solvation structures of the uncomplexed 18C6 molecules between the dry and humid IL solutions. 18C6 molecules in the humid IL solutions are solvated by water molecules, as described above. In the dry IL solutions, they are solvated by C 2 mim ? and/or TFSA -.
On the other hand, the d ML values of the K ? -18C6 complex in the dry and humid IL solutions are almost equal to each other. This indicates that the structure of the K ? -18C6 complex is the same in both the dry and humid IL solutions. Hence, water molecules in the IL solutions may not contribute to the structure of the 1:1 K ? -18C6 complex. However, the significantly different K f values of the K ? -18C6 complex in the dry and humid IL solutions suggest that water molecules influence the mechanism of the complex formation. In the humid IL solution, the K ? -18C6 complex might move to the IL-rich domains after its complexation in the water-rich domains because of its hydrophobicity. Unlike in the case of the K ? -18C6 system, the d ML value of the Li ? -18C6 complex under the humid condition is quite distinct from that under the dry condition. The d ML value of the Li ? -18C6 complex in the humid IL solution is fairly larger compared to that in the dry IL solution. This shows the different structure of the Li ? -18C6 complex in the dry and humid IL solutions. The smaller d ML of the 13 C nuclei for the dry IL solution means the higher coordination state of 18C6 molecule. In contrast, the coordinating ability of Li ? to the 18C6 ligand molecule under the humid condition is lower than that under the dry one. Our previous findings showed that Li ? more preferentially interact with water molecules than with C 2 mim ? and TFSAin the humid [C 2 mim][TFSA] solutions [11]. Water molecules in the humid 18C6-IL solution may also interact with both Li ? isolated from the 18C6 ligand and incorporated into the ligand. The Li ? -water interactions may weaken the Li ? -18C6 interactions in the humid IL system. Figure 6 shows  Considering that water molecules interact with the Li ? -18C6 complex as well as the uncomplexed 18C6 molecules, the formation of the Li ? -18C6 complex is taken place in the water-rich domains in the humid IL solution. The Li ? -18C6 complexes may remain in the water-rich domains.

Conclusion
18C6 molecules are solvated by water molecules in the water-rich domains in the humid IL solutions. In the humid IL solutions, the 18C6 molecules form the alkali metal ions-18C6 complexes at the stoichiometric ratio of the unity for both Li ? and K ? systems. However, the formation mechanisms of the Li ? -and K ? -18C6 complexes are different from each other due to the difference in the interactions of water molecules with the complexes. The Li ? -18C6 complexes form in the water-rich domains and steady in the water-rich domains by the relatively strong interaction between water molecules and the Li ? -18C6 complex. On the other hand, the K ? -18C6 complexes formed in the water-rich domains may move into the ILrich domains owing to very weak interactions between water molecules and the complex.