Analytical and Bioanalytical Chemistry

, Volume 389, Issue 7, pp 2265–2275

PEG-linked geminal dicationic ionic liquids as selective, high-stability gas chromatographic stationary phases

Authors

  • Ke Huang
    • Department of Chemistry and BiochemistryThe University of Texas at Arlington
  • Xinxin Han
    • Department of Chemistry and BiochemistryThe University of Texas at Arlington
  • Xiaotong Zhang
    • Department of Chemistry and BiochemistryThe University of Texas at Arlington
    • Department of Chemistry and BiochemistryThe University of Texas at Arlington
Original Paper

DOI: 10.1007/s00216-007-1625-0

Cite this article as:
Huang, K., Han, X., Zhang, X. et al. Anal Bioanal Chem (2007) 389: 2265. doi:10.1007/s00216-007-1625-0

Abstract

It is known that room-temperature ionic liquids (RTILs) have wide applicability in many scientific and technological fields. In this work, a series of three new dicationic room-temperature ionic liquids functionalized with poly(ethylene glycol) (PEG) linkages were synthesized and characterized via a linear solvation model. The application of these ILs as new GC stationary phases was studied. The efficient separation of several mixtures containing compounds of different polarities and 24 components of a flavor and fragrance mixture indicated comparable or higher resolving power for the new IL stationary phases compared to the commercial polysiloxane and poly(ethylene glycol)-based stationary phases. In addition, the selectivities of the IL stationary phases could be quite unique. The separation of a homologous alkane and alcohol mixture displayed the “dual nature” of these ionic liquids as GC stationary phases. The thermal stability study showed the column robustness up to 350 °C. The high separation power, unique selectivity, high efficiency and high thermal stability of the new dicationic ionic liquids indicate that they may be applicable as a new type of robust GC stationary phase.

Keywords

PEGGeminal dicationic RTILGas chromatographyOrthogonal stationary phaseThermal stability

Introduction

Ionic liquids (ILs) are solvents in which the constituents consist entirely of ions. Room-temperature ionic liquids are ionic liquids with melting point below ambient temperature. They are known to have unique physicochemical properties. They possess low melting points, negligible vapor pressures, wide thermal liquid ranges, non-flammability, highly variable thermal stabilities, variable viscosities, tunable hydrophobicities and the broadest range of solvation interactions of any known solvents. Because of these traits, they enjoy a plethora of applications as the solvent medium for organic synthesis [19], organometallic syntheses [1012], liquid–liquid extractions [1317], electrolytes for electrochemical studies [1822], enzyme-catalyzed reactions [2326], matrix material for matrix-assisted laser desorption ionization (MALDI) [2730], chiral selectors [3133], and lubricant materials [34, 35]. Since ionic liquids can interact with molecules via a range of physicochemical processes, they can act like polar solvents for polar molecules, and like nonpolar solvents for nonpolar molecules. This “dual nature” of ionic liquids makes them desirable materials for GC stationary phases [3644]. In particular, the imidazolium-based ILs are known to possess wide liquid ranges and good GC column efficiencies.

A number of methods have been developed over the years in order to characterize the solvation behavior of liquids [45]. This includes the use of the empirical solvent polarity scales established to explain the solvation process of ionic liquid-mediated organic reactions [46]. Subsequently the solvatochromic dye approach developed by organic chemists [45] was adapted to characterize solvents and solvent mixtures used as mobile phases for liquid chromatography [47]. However these “single parameter polarity approaches” proved to be completely inadequate for ionic liquids since they indicated that all ionic liquids had similar polarities, i.e., close to propanol [48]. The “polarity” approach failed to explain the fact that ILs with apparently identical “polarity parameters” can exhibit completely different solvation behaviors. Clearly, the solvation interactions are more complicated than one average parameter could define. Henceforth, attempts were made with inverse GC measurements to define each individual type of interaction. Initially the Rohrschneider–McReynolds approach was used [36, 48, 49]. In this method, five judiciously selected probe molecules are used to characterize five different types of interaction. Analogous approaches were developed for liquid chromatography, such as Snyder’s P′ scale, Hoy and Karger’s “solubility parameter”, the MOSCED scale of Eckert, and the SPACE and UNIFAC models [50]. The Abraham solvation parameter model was first used to characterize ILs and used to categorize and quantify the variety of actions and interactions between ILs and dissolved solutes [38]. This represented a fundamental change in our understanding and characterization of ILs. This work resembles the Rohrschneider–McReynolds method in principle but has the advantage of perhaps better defined interaction types and it takes advantage of a larger pool of probe molecules [5154]. The linear solvation energy relationship can be described by Eq. (1).
$$ \log \,k\prime = c + eE + sS + aA + bB + lL $$
(1)

The higher case letters are the solute descriptors that can be obtained from the literature [54]. E is the extra molar refraction of the solute at 20 °C; S is the solute dipolarity/polarizabiltiy; A and B are the solute hydrogen bond acidity and basicity, respectively; L is the solute gas–hexadecane partition coefficient at 25 °C. The lower case letters are the interaction parameters that characterize the properties of the solvents (the ILs in this case) under study: c is the system constant; e is the ability of the solvent to interact with the solute probes through π and n electrons; s is the dipolarity/polarizability of the solvent; a and b are the solvent hydrogen bond basicity and acidity, respectively; and l is the dispersion forces that the solvent is able to provide. k′ is the retention factor of the probe molecules which can be measured chromatographically. By subjecting the logarithm of k′ and the solute descriptors to multiple linear regression, values of all the solvation parameters can be obtained and hence the solvation behavior of the solvent can be described (see Experimental).

Several ionic liquids have been evaluated using this method. Previously, we evaluated the properties of 17 imidazolium, pyridinium, and ammonium-based monocationic ionic liquids. The dipolar, hydrogen bonding and dispersion forces were found to be the dominant interaction forces with these ILs and they are largely influenced by the anion [38]. Subsequently, new ionic liquids with bulky substituents on imidazolium rings were synthesized [40]. These ionic liquids showed greater thermal stability, up to 260 °C. Most recently, 39 geminal dicationic RTILs with hydrocarbon linkages that showed even higher thermal stability were reported [41]. One pyrrolidinium-based dicationic liquid displayed thermal stability to over 400 °C [41]. Meanwhile the interaction parameters of these dicationic ILs were found to be similar to the corresponding monocationic type ILs, which explained the parallel or slightly better resolving power of these dicationic ILs in gas chromatography as compared to the corresponding monocationic ILs. In an effort to expand the polar solvation properties of the “ultrastable” dicationic ionic liquids, we have synthesized a series of poly(ethylene glycol) (PEG)-linked geminal dicationic ionic liquids (Fig. 1) and evaluated both their thermal stabilities and their molecular interaction parameters. They are then evaluated as GC stationary phases for the first time. Preliminary indications on analogous compounds indicate that this new class of ILs could have great potential as GC stationary phases [41, 55].
https://static-content.springer.com/image/art%3A10.1007%2Fs00216-007-1625-0/MediaObjects/216_2007_1625_Fig1_HTML.gif
Fig. 1

Structures of PEG-functionalized RTILs

Experimental

Materials

The reagents 1-methylimidazole, 1-benzylimidazole, 1-(2-hydroxyethyl) imidazole, phosphorus tribromide, tetraethylene glycol, pentaethylene glycol, hexaethylene glycol, silver trifluoromethanesulfonate, lithium bis(trifluoromethanesulfonyl)imide, ethyl acetate, toluene, alkane and alcohol homologs, and a flavor and fragrance mixture kit were obtained from Aldrich. All the probe compounds (Table 1) were also obtained from Aldrich. The GC capillaries (i.d. 250 μm) were purchased from Supelco. The polysiloxane column (RTX-5, 5% phenyl methyl polysiloxane) was purchased from Restek (Columbia, MD, USA). The Innowax column (PEG) was purchased from Agilent Technologies (Santa Clara, CA, USA).
Table 1

Probe molecules employed in the RTILs characterizations, and their solute descriptors

Probe molecules

R2

Π2H

α2H

β2H

log L16

Acetic acid

0.265

0.65

0.61

0.44

1.75

Acetophenone

0.818

1.01

0

0.49

4.501

Aniline

0.955

0.96

0.26

0.53

3.993

Benzaldehyde

0.82

1

0

0.39

4.008

Benzonitrile

0.742

1.11

0

0.33

4.039

1-Butanol

0.224

0.42

0.37

0.48

2.601

2-Chloroaniline

1.033

0.92

0.25

0.31

4.674

1-Chlorohexane

0.201

0.4

0

0.1

3.777

p-Cresol

0.82

0.87

0.57

0.31

4.312

Cyclohexanol

0.46

0.54

0.32

0.57

3.758

Cyclohexanone

0.403

0.86

0

0.56

3.792

1,2-Dichlorobenzene

0.872

0.78

0

0.04

4.518

N,N-Dimethylformamide

0.367

1.31

0

0.74

3.173

Naphthalene

1.34

0.92

0

0.2

5.161

Nitrobenzene

0.871

1.11

0

0.28

4.511

1-Nitropropane

0.242

0.95

0

0.31

2.894

1-Octanol

0.199

0.42

0.37

0.48

4.619

Octyl aldehyde

0.16

0.65

0

0.45

4.36

2-Pentanone

0.143

0.68

0

0.51

2.755

Phenetole

0.68

0.7

0

0.32

4.242

Phenol

0.805

0.89

0.6

0.31

3.766

Propionitrile

0.162

0.9

0.02

0.36

2.082

Pyridine

0.794

0.87

0

0.62

3.003

Pyrrole

0.613

0.73

0.41

0.29

2.865

Toluene

0.601

0.52

0

0.14

3.925

m-Xylene

0.623

0.52

0

0.16

3.839

o-Xylene

0.663

0.56

0

0.16

3.939

p-Xylene

0.613

0.52

0

0.16

3.839

Valeraldehyde

0.163

0.65

0

0.45

2.851

Methods

Around 30 probe molecules carrying specific functional groups were used in this work to meet the statistic requirement that a minimum number of four probe molecules are required to meaningfully define each parameter [50]. The probe molecules used are listed in Table 2 with their solute descriptors [54]. It is the most recent and widely accepted linear solvation energy relationship (LSER) model in use nowadays. In this inverse GC method, ILs are coated on GC capillaries and used as GC stationary phases. The probe molecules (Table 1; data acquired from Ref. [54]) are then individually injected and their retention factors are recorded. In this way, the stationary phase solvation behaviors are deconvoluted into five different types of individual interaction using mathematic regression methods.
Table 2

Regression parameter coefficients of Series 1, Series 2, and Series 3 PEG-linked RTILs (see Fig. 1 for the IL structures and abbreviations)

Temperature (°C)

c

e

s

a

b

l

n

R2

(MIM)2PEG3-2NTf2

50 (std. err.)

−2.99 (0.12)

0.09 (0.10)

1.91 (0.11)

2.18 (0.10)

0.11 (0.14)

0.56 (0.03)

29

0.99

70 (std. err.)

−3.07 (0.11)

0.12 (0.09)

1.80 (0.10)

1.94 (0.09)

0.07 (0.13)

0.50 (0.03)

29

0.98

100 (std. err.)

−3.21 (0.11)

0.12 (0.08)

1.68 (0.09)

1.65 (0.08)

0.03 (0.12)

0.44 (0.02)

29

0.99

(MIM)2PEG4-2NTf2

50 (std. err.)

−3.13 (0.11)

0.16 (0.09)

2.04 (0.10)

2.34 (0.09)

0.19 (0.13)

0.54 (0.03)

29

0.98

70 (std. err.)

−3.20 (0.10)

0.16 (0.08)

1.94 (0.09)

2.11 (0.08)

0.14 (0.12)

0.48 (0.03)

29

0.99

100 (std. err.)

−3.27 (0.09)

0.18 (0.07)

1.77 (0.08)

1.79 (0.07)

0.14 (0.10)

0.41 (0.02)

29

0.99

(MIM)2PEG5-2NTf2

50 (std. err.)

−2.99 (0.12)

0.31 (0.10)

1.80 (0.11)

2.16 (0.10)

0.34 (0.14)

0.54 (0.03)

29

0.99

70 (std. err.)

−3.02 (0.11)

0.26 (0.09)

1.71 (0.10)

1.93 (0.08)

0.24 (0.13)

0.48 (0.03)

29

0.99

100 (std. err.)

−3.10 (0.09)

0.23 (0.07)

1.58 (0.08)

1.64 (0.07)

0.19 (0.11)

0.41 (0.02)

29

0.99

(BzIM)2PEG3-2NTf2

50 (std. err.)

−3.06 (0.10)

0.06 (0.08)

1.86 (0.09)

2.01 (0.08)

0.21 (0.11)

0.60 (0.03)

29

0.99

70 (std. err.)

−3.09 (0.09)

0.06 (0.07)

1.76 (0.08)

1.80 (0.07)

0.15 (0.11)

0.54 (0.02)

29

0.99

100 (std. err.)

−3.19 (0.08)

0.07 (0.07)

1.62 (0.08)

1.52 (0.07)

0.10 (0.10)

0.47 (0.02)

29

0.99

(BzIM)2PEG4-2NTf2

50 (std. err.)

−2.98 (0.12)

0.08 (0.10)

1.79 (0.11)

2.04 (0.10)

0.21 (0.14)

0.59 (0.03)

29

0.99

70 (std. err.)

−3.02 (0.11)

0.07 (0.09)

1.69 (0.10)

1.81 (0.09)

0.13 (0.12)

0.54 (0.03)

29

0.99

100 (std. err.)

−3.08 (0.10)

0.09 (0.08)

1.55 (0.09)

1.54 (0.08)

0.03 (0.11)

0.46 (0.02)

29

0.99

(BzIM)2PEG5-2NTf2

50 (std. err.)

−3.03 (0.10)

0.14 (0.08)

1.80 (0.09)

2.05 (0.08)

0.29 (0.11)

0.58 (0.02)

29

0.99

70 (std. err.)

−3.09 (0.09)

0.13 (0.07)

1.71 (0.08)

1.82 (0.07)

0.21 (0.10)

0.53 (0.02)

29

0.99

100 (std. err.)

−3.18 (0.08)

0.14 (0.06)

1.56 (0.07)

1.54 (0.06)

0.17 (0.09)

0.46 (0.02)

29

0.99

(HieM)2PEG3-2NTf2

50 (std. err.)

−2.71 (0.20)

0.51 (0.15)

1.35 (0.19)

1.81 (0.15)

1.24 (0.22)

0.43 (0.05)

28

0.96

70 (std. err.)

−2.85 (0.18)

0.46 (0.14)

1.32 (0.16)

1.60 (0.13)

1.07 (0.20)

0.39 (0.04)

28

0.97

100 (std. err.)

−3.04 (0.15)

0.37 (0.12)

1.27 (0.14)

1.38 (0.11)

0.83 (0.17)

0.35 (0.04)

28

0.97

(HeIM)2PEG4-2NTf2

50 (std. err.)

−2.73 (0.17)

0.43 (0.13)

1.48 (0.16)

1.94 (0.13)

1.01 (0.19)

0.46 (0.04)

28

0.98

70 (std. err.)

−2.80 (0.14)

0.37 (0.11)

1.42 (0.13)

1.73 (0.10)

0.82 (0.15)

0.42 (0.03)

28

0.98

100 (std. err.)

−2.86 (0.15)

0.17 (0.11)

1.40 (0.14)

1.44 (0.11)

0.49 (0.16)

0.37 (0.04)

27

0.97

(HeIM)2PEG5-2NTf2

50 (std. err.)

−2.95 (0.11)

0.31 (0.09)

1.71 (0.11)

2.15 (0.09)

0.78 (0.13)

0.49 (0.03)

28

0.99

70 (std. err.)

−3.03 (0.10)

0.30 (0.08)

1.62 (0.09)

1.91 (0.08)

0.68 (0.11)

0.44 (0.02)

28

0.99

100 (std. err.)

−3.12 (0.09)

0.25 (0.07)

1.53 (0.09)

1.63 (0.07)

0.50 (0.10)

0.38 (0.02)

28

0.99

All the GC columns were made via the static coating method. The ionic liquids were dissolved in dichloromethane to make a coating solution of 0.25% (w/v). The capillary was filled with the coating solution with one end clamped tightly to ensure a complete seal. The column was placed under vacuum from another end while the temperature of the column was maintained at 40 °C in water bath. The vacuum was controlled so that the coating rate was maintained at 3–4 loops per hour. The column was then conditioned from 30–120 °C at 3° min−1. The column efficiency was tested at 100 °C with a carrier gas (helium) flow rate of 1 mL min−1. The efficiencies of these columns were approximately 2600 plates m−1. Each individual probe molecule was injected into the column at three different temperatures, 50 °C, 70 °C, and 100 °C.

(MIM)2PEG3-5-NTf2 and (BzIM)2PEG3-5-NTf2 were synthesized using similar methods to those reported in the literature [55]. (HeIM)2PEG3-5-NTf2 was synthesized by reacting poly(ethylene dibromide) with two moles of 1-hydroxylimidazole in toluene under reflux overnight. The product was dissolved in methanol and mixed with LiNTf2. The mixture was stirred at room temperature for 2 h. The product was extracted with ethylene chloride then washed with water and ether. The final product was dried in a vacuum oven with P2O5 before use. All ionic liquids were characterized using 1H NMR and electrospray ionization (ESI). All 1H NMR (400-MHz) spectroscopy was performed on solutions in deuterated DMSO. The purities of the final products were between 96 and 99%.

Multiple linear regression was performed using Analyse-it. The linearity (R2) for all evaluations was ≥0.96.

Equipment

The GC equipment used was an Agilent (Columbia, MD, USA) model 6892N (G 1540N) gas chromatograph equipped with a flame-ionization detector and Agilent ChemStation data-acquisition software. All analyses were performed with helium as carrier gas at a flow rate of 1 mL min−1 and a split ratio of 100/1. The injector and detector temperatures were 250 °C and 300 °C, respectively.

Results and discussion

Thermal stability test

The ionic liquids under study are poly(ethylene glycol) (PEG)-coupled imidazolium-based salts with bis[(trifluoromethyl)sulfonyl]imide (NTf2) or trifluoromethanesulfonate (triflate, TfO) anions. Three series of RTILs were synthesized and evaluated. Their structures are shown in Fig. 1. The imidazolium cations possess methyl (Series 1, compounds 1–3), benzyl (Series 2, compounds 4–6), or 2-hydroxylethyl (Series 3, compounds 7–9) groups on the 3 position of the imidazolium rings. All dicationic liquids were coated on short (5 m) fused-silica capillaries (see Experimental) and subjected to increasing temperature regions in order to evaluate their thermal stabilities. Typical results are shown in Fig. 2. The most thermally stable monocationic IL was reported to be 1-(4-methoxylphenyl)imidazolium triflate (MPMIM-TfO) with a bleeding temperature of 260 °C [40]. Generally, dicationic ILs possess higher bleeding temperatures than their monocationic analogues due to their higher charge, higher molecular weight, and greater intermolecular interactions [41]. Imidazolium-based (BzIM)2C12-2NTf2 and (MIM)2C9-2NTf2 (compounds 14 and 13, Fig. 1) also showed thermal stabilities exceeding 350 °C [41]. Almost the same thermal stability was found for the PEG-linked dicationic RTILs (Fig. 2). The column efficiency was retested at 100 °C after conditioning the column to 320 °C and it was found to be unchanged prior to and after conditioning. The bis(methylimidazolium)PEG series of ILs exhibited highest bleeding temperature of all RTILs tested and the linkage length apparently has only a minor influence on the thermal stability of the compounds. It was also observed that the triflate anion containing ILs were as stable as the NTf2 anion containing ILs. The extraordinary heat enduring ability of the PEG linked RTIL stationary phases suggests a potential of their development as high-temperature GC stationary phases.
https://static-content.springer.com/image/art%3A10.1007%2Fs00216-007-1625-0/MediaObjects/216_2007_1625_Fig2_HTML.gif
Fig. 2

Thermal stability of PEG-functionalized dicationic RTILs: A, (BzIM)2PEG4-2NTf2; B, (MIM)2PEG4-2NTf2; C, (HeIM)2PEG4-2NTf2; D, (MIM)2PEG4-2TfO; E, (MIM)2PEG3-2NTf2

Solvation parameter coefficient measurements

Cation effects

Dominant forces and influence of cation structure

Each ionic liquid was evaluated for their specific solvation interaction capabilities at three different temperatures, 50 °C, 70 °C, and 100 °C (see Experimental for details). The results are listed in Table 2. The temperature affects the magnitude of each parameter. As temperature increases, the solvation parameter coefficients of the ionic liquids decrease. This is because higher temperature allows higher energy and less orderly orientation of gas molecules hence weaker interactions between the probes and the liquid stationary phases [38, 50]. This is reflected by smaller K (partition coefficients) and k′ (retention factors) of probe molecules at elevated temperatures. On the other hand, lower temperature favors liquid–stationary phase interactions which are reflected in longer retention of the probes. Therefore the characterization of ionic liquids at lower temperatures is thought to produce more pronounced and accurate interaction parameters.

As is indicated by the data in Table 2, the dominant interactions for Series 1 and 2 dicationic ionic liquids are dipole–dipole interactions (s), hydrogen bond basicity (a), followed by dispersion forces (l). Due to the lack of hydrogen bond-donating moieties, these solvents show weak hydrogen bond acidity (b). The ππ and n–π electron interactions are also small relative to the other effects. These findings are in accord with previous studies where it was found that hydrocarbon-linked dicationic ILs, together with most of the tested monocationic ionic liquids, have strong dipole–dipole interactions and hydrogen bond basicities [38, 40, 41]. However, the Series 3 ionic liquids (which have hydroxyethyl groups attached to the imidazolium moiety) show some deviations from this trend (compounds 7–9, Table 2). While dipole–dipole interactions (s), hydrogen bond basicity (a), and dispersion forces (l) remain major solvation parameters, the significantly larger hydrogen bond acidity (b) also plays an important role in the probe molecule solvation processes. The increase of the b-term and e-term compared to the methylated and benzylated ILs implies a significant influence of cation structure on the hydrogen bond acidity and the ππ and n–π electron interactions of these ILs. Chromatographically, the hydroxyl groups of Series 3 act as powerful hydrogen bond donors that notably increase the retention of the basic probe molecules. For example, at 50 °C, the retention factor (k′) of pyridine increased from 2.45 on the (MIM)2PEG3-2NTf2 column to 22.31 on the (HeIM)2PEG3-2NTf2 column. However, the similarities in the interaction parameters for Series 1 and Series 2 ILs suggest a much less pronounced influence of the methyl versus the benzyl substitutions on the IL solvation behaviors (Table 2). Since most of the commercial GC stationary phases available have essentially negligible hydrogen bond acidities (b-term) [56, 57], the development of these new ILs, especially the Series 3 ILs, has a great potential to fill a huge gap of selectivities of available GC stationary phases.

Influence of “linker” chain length

As is shown in Table 2, varying the linkage from PEG3 to PEG5 does not significantly alter the solvation coefficients for the Series 1 and Series 2 RTILs. For the Series 3 RTILs, a small but steady increase in the s-terms and a-terms was discovered as the length of the linkage chain increases. This is probably due to the additional oxygen atoms on the linkage chain that make these ionic liquids better hydrogen bond acceptors and slightly more polar molecules. The progressive decrease in hydrogen bond acidities (b) with increasing linkage chain length may be due to the formation of internal hydrogen bonds which makes these ILs weaker hydrogen bond donors for probe molecules. Chromatographically, this is reflected by a steady decrease in the retention of basic compounds. For example, pyridine had a k′ of 22.31 on IL 7 at 50 °C. This value sequentially decreased to 16.62 and 7.79 on IL 8 and IL 9, respectively. The detailed mechanism for this hydrogen bond acidity decrease requires further investigation on the structures and interactions of these ionic liquids as coated GC stationary phases. However, these findings indicate a moderate influence of the linkage chain length on the solvation parameters of this specific type of ionic liquid.

Influence of linkage type

A comparison of the interaction parameters of PEG-linked and hydrocarbon-linked geminal dicationic ionic liquids with similar linkage lengths was undertaken in order to study the role of linkage types on the solvation properties of these ionic liquids (Table 3). The solvation parameter coefficients for (MIM)2C9-2NTf2 and (BzIM)2C12-2NTf2 RTILs were obtained from previous work [41]. As is suggested by the coefficients, the hydrocarbon linked stationary phases and the PEG linked stationary phases share similar solvation behavior patterns. The dipole–dipole (s) and hydrogen bonding interactions (a) are the dominant forces with dispersion interactions (l) being the third strongest force. However, the PEG-linked RTILs show slightly higher dipolarity/polarizability and hydrogen bond basicity terms due to the additional oxygens within the PEG linkers. They provide good dipole–dipole interaction sites and hydrogen bond-accepting sites. Again the Series 3 (2-hydroxylethyl-containing) ILs differ from the others by having significantly larger “e” and “b” terms. Indeed these terms appear to be somewhat larger for the hydrocarbon-linked dicationic ionic liquid as compared to the PEG linked one.
Table 3

Comparison of the interaction parameters for analogous RTILs linked by PEG versus hydrocarbons (see Fig. 1 for the IL structures and abbreviations)

Temperature (°C)

c

e

s

a

b

l

n

R2

(MIM)2PEG3-2NTf2

70 (std. err.)

−3.07 (0.11)

0.12 (0.09)

1.80 (0.10)

1.94 (0.09)

0.07 (0.13)

0.50 (0.03)

29

0.98

100 (std. err.)

−3.21 (0.11)

0.12 (0.08)

1.68 (0.09)

1.65 (0.08)

0.03 (0.12)

0.44 (0.02)

29

0.99

(MIM)2C9-2NTf2

70 (std. err.)

−2.95 (0.09)

0.11 (0.07)

1.76 (0.08)

1.75 (0.07)

0.20 (0.10)

0.51 (0.02)

33

0.99

100 (std. err.)

−3.06 (0.08)

0.11 (0.06)

1.64 (0.07)

1.50 (0.06)

0.15 (0.09)

0.43 (0.02)

32

0.99

(BzIM)2PEG4-2NTf2

70 (std. err.)

−3.02 (0.11)

0.07 (0.09)

1.69 (0.10)

1.81 (0.09)

0.13 (0.12)

0.54 (0.03)

29

0.99

100 (std. err.)

−3.08 (0.10)

0.09 (0.08)

1.55 (0.09)

1.54 (0.08)

0.03 (0.11)

0.46 (0.02)

29

0.99

(BzIM)2C12-2NTf2

70 (std. err.)

−3.07 (0.08)

0.07 (0.05)

1.62 (0.08)

1.75 (0.06)

0.57 (0.09)

0.56 (0.02)

30

0.99

100 (std. err.)

−3.12 (0.09)

0 (0.04)

1.47 (0.09)

1.44 (0.06)

0.52 (0.10)

0.46 (0.02)

30

0.99

(HeIM)2PEG3-2NTf2

70 (std. err.)

−2.85 (0.18)

0.46 (0.14)

1.32 (0.16)

1.60 (0.13)

1.07 (0.20)

0.39 (0.04)

28

0.97

100 (std. err.)

−3.04 (0.15)

0.37 (0.12)

1.27 (0.14)

1.38 (0.11)

0.83 (0.17)

0.35 (0.04)

28

0.97

(HeIM)2C9-2NTf2

70 (std. err.)

−3.04 (0.13)

0.35 (0.10)

1.49 (0.11)

1.58 (0.10)

1.03 (0.14)

0.47 (0.03)

29

0.98

100 (std. err.)

−3.08 (0.11)

0.29 (0.09)

1.44 (0.10)

1.34 (0.09)

0.76 (0.12)

0.40 (0.03)

29

0.99

Anion effects

The triflate salt of bis(methylimidazolium)PEG3 (compound 10) and bis(2-hydroxylethylimidazolium)PEG4 (compound 11) were synthesized in order to evaluate the influence of the anion on the solvation parameter coefficients (Table 4). By comparison of the bis(methylimidazolium)PEG3 NTf2 (compound 1) and bis(2-hydroxylethylimidazolium)PEG4 NTf2 (compound 8) stationary phases with their triflate analogs (compounds 10 and 11), a significant difference is noted with the s (dipolar) and a (hydrogen bond basicity) terms. At 50 °C the s term increased from 1.91 to 2.18 from compound 1 to compound 10, and from 1.48 to 2.11 from compound 8 to compound 11. The a term increased from 2.18 to 3.39 for compound 1 to compound 10, and from 1.94 to 3.19 for compound 8 to compound 11. However, while the e-term and the b-term were slightly increased by replacing the NTf2 anion with triflate anion for the bis(methylimidazolium)PEG3 IL, the change in the anion did not maintain the high e-term and b-term of bis(2-hydroxylethylimidazolium)PEG4 NTf2 IL. Despite the fact the bis(2-hydroxylethylimidazolium)PEG4 NTf2 possesses a high e-term and b-term, and the fact bis(methylimidazolium)PEG3 triflate possesses a high s-term and a-term, bis(2-hydroxyethylimidazolium)PEG4 triflate did not produce an ionic liquid with both high hydrogen bond acidity (b) and hydrogen bond basicity (a) as expected. Therefore, the solvation properties of a ionic liquid are not just simple additive effects of its individual components. The microscopic structure of a specific type of IL also has great influence on the IL properties. These results indicate that the anions play a major role in the solvation behaviors of these ionic liquids through their dominant control of the s and a properties. This same trend has been reported previously for other ionic liquids [38, 40].
Table 4

Comparison of the interaction parameters of PEG-linked RTILs with NTf2 versus triflate anions (see Fig. 1 for the IL structures and abbreviations)

Temperature (°C)

c

e

s

a

b

l

n

R2

(MIM)2PEG3-2NTf2

50 (std. err.)

−2.99 (0.12)

0.09 (0.10)

1.91 (0.11)

2.18 (0.10)

0.11 (0.14)

0.56 (0.03)

29

0.99

70 (std. err.)

−3.07 (0.11)

0.12 (0.09)

1.80 (0.10)

1.94 (0.09)

0.07 (0.13)

0.50 (0.03)

29

0.98

100 (std. err.)

−3.21 (0.11)

0.12 (0.08)

1.68 (0.09)

1.65 (0.08)

0.03 (0.12)

0.44 (0.02)

29

0.99

(MIM)2PEG3-2TfO

50 (std. err.)

−3.38 (0.15)

0.51 (0.12)

2.18 (0.13)

3.39 (0.12)

0.32 (0.17)

0.44 (0.04)

29

0.99

70 (std. err.)

−3.38 (0.15)

0.55 (0.12)

2.02 (0.13)

3.03 (0.12)

0.34 (0.17)

0.37 (0.04)

29

0.99

100 (std. err.)

−3.49 (0.11)

0.45 (0.09)

1.95 (0.98)

2.72 (0.09)

0.22 (0.13)

0.31 (0.03)

26

0.99

(HeIM)2PEG4-2NTf2

50 (std. err.)

−2.73 (0.17)

0.43 (0.13)

1.48 (0.16)

1.94 (0.13)

1.01 (0.19)

0.46 (0.04)

28

0.98

70 (std. err.)

−2.80 (0.14)

0.37 (0.11)

1.42 (0.13)

1.73 (0.10)

0.82 (0.15)

0.42 (0.03)

28

0.98

100 (std. err.)

−2.86 (0.15)

0.17 (0.11)

1.40 (0.14)

1.44 (0.11)

0.49 (0.16)

0.37 (0.04)

27

0.97

(HeIM)2PEG4-2TfO

50 (std. err.)

−3.43 (0.13)

0.30 (0.10)

2.04 (0.11)

3.12 (0.10)

0.27 (0.15)

0.50 (0.03)

29

0.99

70 (std. err.)

−3.25 (0.12)

0.30 (0.09)

1.89 (0.11)

2.83 (0.09)

0.18 (0.13)

0.44 (0.03)

29

0.99

100 (std. err.)

−3.28 (0.11)

0.27 (0.08)

1.76 (0.09)

2.42 (0.08)

0.14 (0.12)

0.36 (0.03)

28

0.99

GC application study

The PEG-functionalized dicationic RTIL stationary phases were tested for their separation abilities. Two standard mixtures were prepared. Each mixture was run on a (MIM)2PEG3-2NTf2-coated column, a (MIM)2PEG3-2TfO-coated column, a polar commercial Innowax column, and a nonpolar commercial Rtx-5 (Crossbond 5% diphenyl 95% dimethylpolysiloxane) column. Figure 3 shows the separation of a homologous series of alkanes and alcohols on the four columns. The highly polar Innowax column shows good retention for alcohols but poor selectivity for the alkanes. In contrast, the low-polarity Rtx-5 column tends to hold alkanes strongly whereas many alcohols were eluted quickly. The (MIM)2PEG3-2NTf2 and the (MIM)2PEG3-2TfO columns have their selectivity lying in between the two commercial columns. For example, 1-hexanol is retained for obviously longer on the Innowax column than on the Rtx-5 column, whereas on the (MIM)2PEG3-2NTf2 column it eluted with moderate retention. Also, octane eluted with a retention factor in between those of the Rtx-5 and the Innowax column. It is apparent that this new stationary phase has reasonable retention for both polar alcohols and nonpolar alkanes. This is another demonstration of the “dual nature” of ionic liquid-based GC stationary phases. That is, they can retain polar compounds as if they were polar stationary phases, and retain nonpolar compounds as if they were nonpolar stationary phases. The disadvantage of the (MIM)2PEG3-2NTf2 column is the tailing peaks for alcohols. Earlier studies suggested that the NTf2 columns suffer from worse mass transfer of hydrogen bond acidic molecules, probably because these molecules can more conveniently interact with the delocalized negative charge on the S–N–S moiety of the NTf2 ILs [48]. However, the (MIM)2PEG3-2TfO column shows comparable separation power to the NTf2 column but symmetric peaks for all sample components. It seems to be a more useful GC stationary phase for the separation of hydrogen-bond acidic compound mixtures.
https://static-content.springer.com/image/art%3A10.1007%2Fs00216-007-1625-0/MediaObjects/216_2007_1625_Fig3_HTML.gif
Fig. 3

Separation of homologous alkane and alcohol mixture. 1, pentane; 2, hexane; 3, dichloromethane; 4, methanol; 5, heptane; 6, ethanol; 7, octane; 8, 1-propanol; 9, 1-butanol; 10, nonane; 11, 1-pentanol; 12, decane; 13, 1-hexanol; 14, undecane; 15, dodecane; 16, tridecane; 17, tetradecane; 18, pentadecane; 19, hexadecane. Conditions: 30 °C for 3 min, 10° min−1 to 160 °C

Figure 4 shows the separation of a mixture of 24 flavor and fragrance compounds, which are primarily homologous esters. All 24 peaks were baseline separated on the (MIM)2PEG3-2NTf2 column while there was coelution of isopropyl tiglate and ethyl hexanoate on the Innowax column and partial separation of propyl butyrate and ethyl valerate on the Rtx-5 column. For this mixture, the selectivity of the (MIM)2PEG3-2NTf2 column and the (MIM)2PEG3-2TfO column lean more towards the polar Innowax column with only a few compound elution order switches. Their extraordinary resolving power for intermediate to high-polarity homologous molecules seems to be an advantage to be further explored.
https://static-content.springer.com/image/art%3A10.1007%2Fs00216-007-1625-0/MediaObjects/216_2007_1625_Fig4_HTML.gif
Fig. 4

Separation of flavor and fragrance mixture: 1, dichloromethane; 2, ethyl propionate; 3, methyl butyrate; 4, ethyl butyrate; 5, isopropyl butyrate; 6, propyl butyrate; 7, ethyl valerate; 8, allyl butyrate; 9, methyl tiglate; 10, isopropyl tiglate; 11, ethyl hexanoate; 12, propyl tiglate; 13, allyl tiglate; 14, ethyl heptanoate; 15, hexyl butyrate; 16, ethyl octanoate; 17, furfuryl propionate; 18, furfuryl butyrate; 19, hexyl tiglate; 20, furfuryl pentanoate; 21, furfuryl hexanoate; 22, benzyl butyrate; 23, furfuryl heptanoate; 24, furfuryl octanoate; 25, benzo-trans-2-methyl-2-butanoate. Conditions: 40 °C for 3 min, 10° min−1 to 150 °C

In order to study the relationship between the solvation parameters of the ILs and their corresponding chromatographic behaviors, a mixture of compounds from low polarity to high polarity was injected on the (MIM)2PEG3-2NTf2 column, the (BzIM)2PEG3-2NTf2 column, and the (HeIM)2PEG3-2NTf2 column (Fig. 5). As indicated by their solvation parameters, all three stationary phases produced similar retention behaviors and elution orders of the component compounds. However, hexanoic acid showed severely tailing peaks on the (MIM)2PEG3-2NTf2 column and the (BzIM)2PEG3-2NTf2 column. In contrast, the (HeIM)2PEG3-2NTf2 column showed a perfectly symmetric peak for hexanoic acid, while producing a slightly tailing peak for aniline. This distinct behavior is probably caused by the high hydrogen bond acidity (b-term) and correspondingly low hydrogen bond basicity (a-term) of the Series 3 type ILs. Thus, aside from anions, the cations can also be engineered to custom make ILs in order to change or improve peak symmetries.
https://static-content.springer.com/image/art%3A10.1007%2Fs00216-007-1625-0/MediaObjects/216_2007_1625_Fig5_HTML.gif
Fig. 5

Separation of polar and nonpolar compound mixture: 1, dichloromethane; 2, 1-butanol; 3, ethyl hexanoate; 4, octyl aldehyde; 5, cyclohexanone; 6, 1-octanol; 7, tridecane; 8, hexanoic acid; 9, naphthalene; 10, aniline; 11, pentadecane. Conditions for (MIM)2PEG3-2NTf2: Column: 70 °C for 3 min, 10° min−1 to 150 °C. Conditions for (BzIM)2PEG3-2NTf2 and (HeIM)2PEG3-2NTf2: Column: 50 °C for 3 min, 6° min−1 to 140 °C

Conclusions

The solvation parameter evaluation of PEG-functionalized dicationic RTILs shows that the dominant interactions for this type of ILs are the dipolar interactions (s), hydrogen bond basicity (a), followed by dispersion forces (l). The nature of the anion affects both the s-terms (dipolar interactions) and the a-terms (hydrogen bond basicity) of the ILs. The bis(2-hydroxylethylimidazolium)PEG stationary phases have higher hydrogen bond acidities than bis(methylimidazolium)PEG and bis(benzylimidazolium)PEG stationary phases. The cation structure affects both the b-terms and e-terms of these ILs. The linkage type shows slight influence on the s and a terms while the linkage length only has an effect on the Series 3 ILs. However the microscopic structure of individual type of ILs may cause some deviations of these solvation behavior patterns. Compared to commercial Innowax and Rtx-5 columns, the PEG functionalized dicationic RTILs possess somewhat better separation powers and distinct selectivities of complex samples. With their dual nature and high thermal stability, they show great potential for making a novel class of potent and robust GC stationary phases.

Acknowledgements

We gratefully acknowledge the support from the Robert A. Welch Foundation (Y0026) for their support of this work.

Copyright information

© Springer-Verlag 2007