Você está na página 1de 11

DOI: 10.1002/chem.

201404253 Full Paper

& Ionic Liquids

Benzene Solubility in Ionic Liquids: Working Toward an


Understanding of Liquid Clathrate Formation
Jorge F. B. Pereira,[a, b] Luis A. Flores,[a] Hui Wang,[a] and Robin D. Rogers*[a]

Abstract: The solubility of benzene in 15 imidazolium, pyrro- controlled by the strength of the cation–anion interactions,
lidinium, pyridinium, and piperidinium ionic liquids has been that is, the stronger the cation–anion interaction, the lower
determined; the resulting, benzene-saturated ionic liquid the benzene solubility. Other factors that were determined
solutions, also known as liquid clathrates, were examined to be important in the final amount of benzene in any given
with 1H and 19F nuclear magnetic resonance spectroscopy to liquid clathrate phase included attractive interactions be-
try and understand the molecular interactions that control tween the anion and benzene (when significant), and larger
liquid clathrate formation. The results suggest that benzene steric or free volume demands of the ions, both of which
interacts primarily with the cation of the ionic liquid, and lead to greater benzene solubility.
that liquid clathrate formation (and benzene solubility) is

1. Introduction control the solubility of the aromatic solvents in the ILs, after
Surette et al.[16] suggested that in some IL-aromatic systems,
Ionic liquids (ILs), usually defined as salts with melting points liquid clathrates (LC) were formed.
below 100 8C,[1] appeared in the last two decades as an alterna- Liquid clathrates were first described by Atwood et al.[17–19]
tive solvent for several applications,[2] such as synthesis,[3, 4] who demonstrated LC formation with highly reactive air-sensi-
electrochemistry,[5, 6] and liquid crystals,[7] to name a few. tive alkylaluminum salts. Holbrey et al.[20] investigated if LC for-
Though there are a number of applications that have taken ad- mation with ILs was a general characteristic when mixing aro-
vantage of IL characteristics, such as accessible low vapor pres- matic hydrocarbons (benzene, toluene, and xylenes) with
sures, high chemical, thermal, and electrochemical stabilities, common 1-alkyl-3-methylimidazolium ILs ([Cnmim]X). The re-
or unique solvation properties, the use of ILs in separation pro- sults demonstrated that when ILs and excess aromatic hydro-
cesses continues to be one of the most studied applications, carbons were mixed, LC phases formed spontaneously with an
because the IL’s physicochemical properties can be designed upper pure aromatic phase and a lower IL/aromatic phase with
by a proper choice of the cation and anion.[8] In this context, much lower viscosity than the neat ILs. This work suggested
the observation that several aromatic solvents, such as ben- that LC phases are formed when associative interactions be-
zene, are soluble but not completely miscible in ILs,[9] has led tween the aromatic molecules and the salt’s ions separate
to growing interest in these IL-aromatic systems. cation–anion packing interactions to a sufficient degree, allow-
Publications describing the solubility of aromatic hydrocar- ing a formation of localized cage structures. Crystal structures
bons in ILs have rapidly grown,[10] but these studies have usu- of IL-aromatic complexes, although not always indicative of
ally been aimed at the extraction of the aromatic from their the solution state, have indeed shown the formation of cage
mixtures with alkanes,[11–14] or directed at thermodynamic and structures between benzene and 1,3-dimethylimidazolium hex-
modeling studies of the binary and ternary phase diagrams.[15] afluorophosphate ([C1mim][PF6]), in which benzene is “sand-
Additionally, some experimental and modeling studies have wiched” between two imidazolium cations.
also been undertaken in order to explain the interactions that Although Holbrey et al. showed clear evidence that LC for-
mation in the IL-aromatic mixtures was based on cation–aro-
[a] Prof. Dr. J. F. B. Pereira, L. A. Flores, Dr. H. Wang, Prof. Dr. R. D. Rogers matic p–p interactions, some theoretical studies have demon-
Center for Green Manufacturing and Department of Chemistry
strated that the p-based interactions are not the predominant
The University of Alabama
Tuscaloosa, AL 35487 (USA) cause of the high solubility of the aromatics, rather that the IL-
E-mail: RDRogers@ua.edu cation and benzene preferential interactions are mainly ion-
[b] Prof. Dr. J. F. B. Pereira quadrupole, and that the ion-quadrupole interactions between
Current Address: the anion and benzene are more dominant than the cation–
Department of Bioprocess and Biotechnology
benzene interactions.[21–23] As reviewed by Weber et al.,[24] there
School of Pharmaceutical Sciences
UNESP - Univ Estadual Paulista, Araraquara, SP 14801 (Brazil) are very complex interactions found in IL-aromatic systems, in
Supporting information for this article is available on the WWW under which the ionic nature of the IL is the main driving force for
http://dx.doi.org/10.1002/chem.201404253. the high solubility of aromatics, as well as for LC formation.

Chem. Eur. J. 2014, 20, 15482 – 15492 15482  2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper

This observation was supported by studies that revealed the bond acidity, basicity, aromaticity of the anion or cation, and
ability of nonaromatic ILs to solubilize aromatic hydrocar- the length of the cationic alkyl side chains on LC formation.
bons.[25, 26] The ILs investigated included the imidazolium salts, 1-ethyl-3-
In this context, and taking into account that the molecular methylimidazolium acetate [C2mim][OAc], 1-ethyl-3-methylimi-
interactions that control IL-aromatic LC formation are not fully dazolium dicyanamide [C2mim][N(CN)2], 1-ethyl-3-methylimida-
understood, we examined and report here the solubility of zolium bis(trifluoromethanesulfonyl)imide [C2mim][NTf2], 1-(2-
benzene in various ILs as a function of varying cations and hydroxyethyl)-3-methylimidazolium bis(trifluoromethanesulfo-
anions, as well as 1H and 19F NMR spectroscopy studies aimed nyl)imide [OHC2mim][NTf2], 1-butyl-3-methylimidazolium ace-
at understanding the predominant benzene–IL interactions tate [C4mim][OAc], 1-butyl-3-methylimidazolium tetrafluorobo-
behind LC formation. The studies were designed to evaluate rate [C4mim][BF4], 1-butyl-3-methylimidazolium hexafluoro-
how LC formation is affected by cation–anion strength, prefer- phosphate [C4mim][PF6], 1-butyl-3-methylimidazolium dicyana-
ential interactions between benzene and the cation or anion, mide [C4mim][N(CN)2], and 1-butyl-3-methylimidazolium bis(tri-
and steric effects. As a result of our findings, we also present fluoromethanesulfonyl)imide [C4mim][NTf2]; the pyrrolidium
some recommendations to consider for future studies aimed salts, 1-butyl-1-methylpyrrolidinium dicyanamide [C4mpyrr]-
at selectivity in aromatic separations. [N(CN)2], 1-butyl-1-methylpyrrolidinium bis(trifluoromethane-
sulfonyl)imide [C4mpyrr][NTf2], and 1-butyl-1-methylpyrrolidini-
um tris(pentafluoroethyl)trifluorophosphate [C4mpyrr][FAP]; the
2. Results and Discussion pyridinium salts, 1-butyl-3-methylpyridinium tetrafluoroborate
Fifteen ILs (Table 1) were chosen for this study, in order to [C4mpy][BF4] and 1-butyl-3-methylpyridinium bis(trifluorome-
assess the influence of several parameters, such as hydrogen thanesulfonyl)imide [C4mpy][NTf2]; and the piperidinium salt

Table 1. Ionic liquids studied and relevant solubility and spectroscopic data.

Ionic liquid Cation Anion Benzene xwater[a] dHHx[b] DdHHx[c] DdHHB[d]


solubility[a] neat IL [ppm] [ppm]
[mol/mol] [ppm]

[C2mim][OAc] 0.55(1) 0.062(1) 10.77 0.11 0.06

[C2mim][N(CN)2] 1.07(9) 0.063(1) 9.11 0.32 0.20

[C4mim][OAc] 1.22(1) 0.097(4) 10.84 0.14 0.09

[C4mim][BF4] 1.43(1) 0.005(1) 8.52 0.14 0.03

[C4mpy][BF4] 2.00(2) 0.023(1) 8.63 0.18 0.00

[OHC2mim][NTf2] 1.48(2) 0.002(1) 8.42 0.53 0.03

[C4mim][PF6] 2.05(6) 0.013(1) 8.25 0.37 0.03

Chem. Eur. J. 2014, 20, 15482 – 15492 www.chemeurj.org 15483  2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper

Table 1. (Continued)
Ionic liquid Cation Anion Benzene xwater[a] dHHx[b] DdHHx[c] DdHHB[d]
solubility[a] neat IL [ppm] [ppm]
[mol/mol] [ppm]

[C4mim][N(CN)2] 2.15(9) 0.040(4) 9.14 0.44 0.18

[C4mpyrr][N(CN)2] 2.29(4) 0.022(3) 3.65 0.53 0.24

[C2mim][NTf2] 3.25(7) 0.012(1) 8.47 0.78 0.05

[C4mpyrr][NTf2] 4.07(5) 0.044(1) 3.51 0.72 0.09

[C4mim][NTf2] 4.27(4) 0.003(1) 8.53 0.70 0.04

[C4mpy][NTf2] 4.27(9) 0.005(1) 8.61 0.70 0.02

[C4mpyrr][FAP] 4.50(21) 0.006(1) 3.17 0.88 0.00

[C4mpip][NTf2] 4.51(10) 0.004(1) 3.32 0.67 0.08

[a] The number in parenthesis is the estimated standard deviation. [b] The proton (Hx) with the highest chemical shift (dH, ppm) in the cation, usually, but
not always the most acidic proton. [c] The difference between dH, ppm of Hx in benzene solution and that in the neat IL. [d] The difference between dH,
ppm of the benzene protons in IL solution and that in neat benzene.

1-butyl-1-methylpiperidinium bis(trifluoromethanesulfonyl)- Figure 1 indicates that increasing the water content of


imide [C4mpip][NTf2]. [C2mim][OAc] reduces the benzene solubility. This decrease in
Because this study included both hydrophilic ([OAc]and benzene solubility was expected, because water solvates the IL
[N(CN)2]) and hydrophobic ([BF4] , [PF6] , [NTf2] , and [FAP]) ions, preventing p–ion interactions with benzene. These results
ILs, we first investigated how water might influence benzene do, however, emphasize the importance of using dried ILs, to
solubility. This is primarily a concern for the hydrophilic ILs, avoid water masking the “real” influence of each IL ion on ben-
and we thus chose one of the most hydrophilic ILs, [C2mim]- zene solubility. We thus dried all of the ILs used in this study
[OAc], and prepared solutions of this IL with known water to xwater < 0.1 (Table 1), to reduce the influence of water on the
mole fractions (xwater) from 0.05 to 0.5. The solubility of ben- results as much as possible.
zene in these solutions was measured after the addition of 10 The benzene solubility in all 15 dried ILs was then deter-
molar equivalents of benzene to each water/IL mixture, vigo- mined utilizing the same procedures as noted above. The solu-
rous mixing, and equilibration under ambient conditions for bility and spectroscopic data obtained (Table 1) suggested dis-
12 h. Three aliquots of each lower LC phase were then taken tinct effects when specific cations and anions were combined,
for 1H NMR analysis, and the mole ratio between the integrat- and thus, in the discussion below, we focus on trends which
ed number of protons of benzene and IL[27] was used to deter- can be correlated to the anion or the cation.
mine the number of moles of benzene dissolved per mole of
IL in each LC (Figure 1).

Chem. Eur. J. 2014, 20, 15482 – 15492 www.chemeurj.org 15484  2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper

Figure 1. Solubility of benzene (mol/mol) in [C2mim][OAc] with different


mole fractions of water (xwater). The first point corresponds to the water con- Figure 2. Solubility of benzene (mol/mol) in several [C2mim] + (dark gray
tent of the dried IL measured by Karl Fisher titration. bars), [C4mim] + (black bars), and [C4mpyrr] + (gray bars) ILs.

2.1 Anion effects


ronment of the protons in the neat IL versus the LC phase
completely saturated with benzene. The evaluation of 1H NMR
The ability of an IL to act as solvent is often reflected in the chemical shifts in ILs (particularly in the imidazolium ILs) allows
IL’s hydrogen-bond basicity (b), mainly controlled by the IL some understanding of the interionic interactions, because the
anion.[28] One of the methods to estimate the polarity of ILs chemical shifts are proportional to the strength of such inter-
and, in particular, b, is the use of solvatochromic probes. Sever- actions.[34] For each set of experiments, the 1H NMR chemical
al researchers have reported polarity scales for different ILs shift deviations were defined according to Equation (1).
using different dyes and experimental approaches.[29–32 Jessop
et al.[33] compiled several solvatochromic parameters for ILs Dd ¼ dðIL þ benzeneÞdðILÞ ð1Þ
and observed that the b values increased in the following
trend: [NTf2](0.24) < [PF6](0.29) < [BF4](0.36) < The results presented in Table 1 suggest a relationship be-
[N(CN)2](0.52) < [OAc](0.99). Thus, in order to evaluate the tween the solubility of benzene and the difference in the
effect of different anions on LC formation, two series of imida- chemical shift of the proton of the cation with the largest
zolium-based ([C2mim] + and [C4mim] + ) and one pyrrolidinium- change in chemical shift (usually the most acidic hydrogen of
based ([C4mpyrr] + ) ILs were chosen with these anions and the the cation) in the neat IL versus the LC. In general, the addition
even more weakly basic [FAP]anion, constituting a large of benzene induces an upfield shift (negative DdH values), indi-
range of hydrogen-bond basicity. Only those ILs commercially cating more favorable interactions of the cations with benzene
available and liquid at room temperature were investigated. and more cation–anion separation. In previous work,[35] we
The benzene solubilities from Table 1 (mol benzene/mol IL) are (along with other groups)[36, 37] have shown that both increas-
compared in Figure 2. ing Coulombic attractions and increasing hydrogen bonding
As observed in Figure 2, the ILs with the most basic (and between the cation and anion leads to downfield shifts of the
most strongly coordinating) anion ([OAc]) exhibited the protons in imidazolium-based ILs. Notably, the exceptions were
lowest ability to solubilize benzene, whereas the ILs with the in the [OAc]-based ILs studied, which have very limited ben-
least basic (and most weakly coordinating) anions ([NTf2] and zene solubility, for which positive DdH deviations (downfield
[FAP]) exhibited the highest benzene solubility. Indeed, for all shift of C-2 protons) with benzene addition were observed.
three series of ILs studied, a similar trend was observed where To gain further understanding of the interactions controlling
the solubility of benzene decreased in the order [FAP]  LC formation, three ILs with the [C4mim] + cation were chosen
[NTf2] > [N(CN)2] > [PF6] > [BF4] > [OAc] , generally, but to evaluate how the benzene affects the chemical environment
not exclusively, correlated with the basicity of the anion. (The of all of the imidazolium protons. For this purpose, solutions of
exceptions, [N(CN)2]-based ILs, will be discussed later.) These different benzene/IL mole ratios, 0, 0.5, 1.0, 1.5, and 2.0 mol/mol
results would seem to indicate that, in general, the high solu- were prepared, as noted earlier and evaluated by 1H NMR
bility of benzene is controlled by the energy of binding or co- spectroscopy. The respective DdH values for [C4mim][NTf2],
ordination between the anions and cations. [C4mim][N(CN)2], and [C4mim][PF6] at different benzene/IL mol
In order to provide a deeper understanding of these results, ratios are depicted in Figure 3. Almost all of the protons in the
we used 1H NMR spectroscopy to evaluate the chemical envi- cation shift upfield with increasing concentrations of benzene.

Chem. Eur. J. 2014, 20, 15482 – 15492 www.chemeurj.org 15485  2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper

comparing the different anions, it is evident that the higher


the chemical shift deviation for a given amount of benzene,
the more total benzene that IL will solubilize (i.e., DdH and
benzene solubility increase in the order [C4mim][NTf2] >
[C4mim][N(CN)2] > [C4mim][PF6]). This suggests an important
role for the anion on the cation–p interaction and consequent-
ly on LC formation, in which the anions effectively control in-
teractions between the imidazolium cation and benzene.
The data in Figure 3 also provides clues to the behavior of
the [N(CN)2]-based ILs, which, as noted earlier, were the only
exceptions to the general trend correlating benzene solubility
with the b values of the anions. These ILs (b = 0.52) exhibited
higher ability to solubilize benzene than the ILs with
[PF6](0.29) and [BF4](0.36) anions. In this particular case, it
seems that the chemical structure of the anion has some addi-
tional influence on benzene solubility in these ILs. Looking
more closely at the DdH values of [C4mim][N(CN)2] and
[C4mim][PF6] (Figure 3), it is observed that, although both sys-
tems solubilize almost the same amount of benzene (  2
mol/mol), higher deviations in proton chemical shifts were ob-
served for [C4mim][N(CN)2].
To further assess the role of ion–benzene interactions in LCs,
we also evaluated the 1H NMR chemical shift deviations of the
benzene protons (HB) in solutions of [C4mim][NTf2], [C4mim]-
[N(CN)2], and [C4mim][PF6] at different mole ratios of benzene.
The data (Table 1, Figure 4) indicates that the largest deviations

Figure 4. Benzene proton chemical shift deviations (DdH) in [C4mim][NTf2],


[C4mim][N(CN)2], and [C4mim][PF6] at different benzene mole ratios.

Figure 3. Proton chemical shift deviations (DdH) of [C4mim][NTf2],


[C4mim][N(CN)2], and [C4mim][PF6] at different benzene/IL mole ratios.
were obtained with [C4mim][N(CN)2], followed by [C4mim]-
[NTf2], and almost no deviations for the [C4mim][PF6] solutions.
The DdH of HB values for [N(CN)2]-based ILs of approximately
The most acidic imidazolium ring hydrogen atoms exhibit the 0.2 ppm were at least fourfold higher than in the other ILs.
largest shifts, whereas the protons in the alkyl chain exhibited These results suggest an interaction between benzene and
much smaller deviations. [N(CN)2]anions that was not present for the other anions stud-
Previously, Lungwitz et al.[31, 32] showed that an IL’s b parame- ied, and this leads to an increase in the solubility of benzene.
ter correlates with the 1H NMR chemical shift of the most We also note that the chemical structure of [N(CN)2] has been
acidic proton of an imidazolium cation, and more recently previously identified to be a fairly cohesive medium due to its
Cludio et al.[28] demonstrated that this proton signal could be capability of interacting specifically through several modes, in-
related to the hydrogen-bond interaction energy in the equi- cluding lone electron pairs, dipolarity/polarizibility, and H-
molar cation–anion mixture. Interestingly in our data, when bonding.[38]

Chem. Eur. J. 2014, 20, 15482 – 15492 www.chemeurj.org 15486  2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper

Overall, the results above indicate an important role of the The origin of this difference is not entirely clear, however,
anion in IL–benzene interactions and, consequently, on the sol- longer alkyl chain length can result in lower overall polarity,
ubility of benzene. Primarily, the data suggests that anions ef- while also leading to larger entropic effects. Considering the
fectively control the coordination between a solute such as concept of LC formation in which the benzene molecules are
benzene and the cation; the stronger the cation–anion interac- entrapped by p–p stacking in the “cages” formed by two imi-
tion, the less benzene will be solubilized. However, Figure 2 dazolium cations,[20] an increase in the “free volume” of the IL
also indicates that increasing the alkyl chain length in an imi- due to steric effects would also favor increased aromatic solu-
dazolium cation (see [C2mim] + vs. [C4mim] + -based ILs in bility.
Figure 2) increases the solubility of benzene. Thus, we turn to Although we have only [OHC2mim][NTf2] as an example,
a discussion of the results aimed at understanding how the Figure 5 provides some information on the effect of functional-
cation influences benzene solubility. ization of the cation’s alkyl substituent, in particular the influ-
ence of a hydroxyl group. The presence of the hydroxyl group
halved the benzene solubility when compared with [C2mim]-
2.2 Cation effects [NTf2]. An increase in polarity by the addition of hydroxyl
groups was described previously by Chiappe et al.,[42] which, as
Three different [N(CN)2]-based ILs ([C2mim][N(CN)2], [C4mim]-
suggested by the authors, was caused by the significant in-
[N(CN)2], and [C4mpyrr][N(CN)2]) and six [NTf2]-based ILs
crease in the ability of the cation to act as a hydrogen-bond
([OHC2mim][NTf2], [C2mim][NTf2], [C4mim][NTf2], [C4mpyrr]-
donor. Increased polarity and hydrogen bonding between
[NTf2], [C4mpy][NTf2], and [C4mpip][NTf2]) were chosen to eval-
cation and anion would lead to stronger cation–anion interac-
uate the effect of the cationic core on benzene solubility and
tion and, as we noted earlier, the stronger the cation–anion in-
LC formation. A comparison of the benzene solubility is pre-
teraction, the less benzene will be solubilized.
sented in Figure 5 and reveals that the benzene solubility fol-
In contrast to the effects of the cation substituents, the influ-
lows the trend: [OHC2mim] + < [C2mim] + < [C4mim] + 
ence of the nature of the cationic core appears to be much
[C4mpyrr]  [C4mpy]  [C4mpip] + .
+ +
less dramatic. For example, in the data in Figure 5, it is ob-
served that changing the cationic core in the ILs with [N(CN)2]
anions from aromatic (imidazolium) to nonaromatic (pyrrolidi-
nium) makes little difference in benzene solubility where ap-
proximately 2.2 moles of benzene per mole of IL dissolves.
Similar behavior was observed for the [NTf2]-based ILs, for
which both aromatic ILs studied, [C4mim][NTf2] and [C4mpy]-
[NTf2], have solubilities around 4.3 mol/mol and the nonaro-
matic [C4mpip][NTf2] and [C4mpyrr][NTf2] ILs have solubilities of
4.5 and 4.1 mol/mol, respectively.
As described above when analyzing the effects of the IL
anions, to evaluate changes in the chemical environments with
different cations, benzene solutions of the [NTf2]-based ILs,
which exhibited the higher solubilities, were analyzed by 1H
and 19F NMR spectroscopy. Completely miscible IL–benzene
solutions of different mole ratios up to the maximum solubility
were prepared (0.5, 1.0, 1.5, 2.0, 2.5, 3.0, 3.5, and 4.0 mol/mol).
For each solution the 1H and 19F chemical shift deviations were
measured and the respective DdH and DdF were calculated
according to Equation (1) (for DdF the 19F NMR chemical
shift values were used). The results are presented in Figures 6
Figure 5. Solubility of benzene in different [N(CN)2] (gray bars) and [NTf2] and 7.
(black bars) ILs. As expected, the DdH values presented in Figure 6 were sim-
ilar in all of the [NTf2]-based ILs studied, in which all the pro-
tons shifted upfield with the addition of benzene. In all cases,
the protons at higher d (the acidic or a protons) exhibited the
The data suggests that both the nature of the cationic core highest deviations (ca. 0.5 ppm), and the protons of the
and the length of any appended alkyl chain can influence the butyl-alkyl chains exhibited the lowest deviations (less than or
benzene solubility. Increasing the length of the alkyl chain equal to 0.2 ppm). The 19F NMR data in Figure 7 confirm
leads to higher benzene solubility, and the [C4mim] + -based ILs these trends where Ddanion at different IL/benzene ratios exhibit
solubilize twice the number of moles of benzene than the ILs similar downfield shifts for all [NTf2]-based ILs studied. In all
with the [C2mim] + cation. These results are in accord with sev- cases, the data suggests that although the nature of each
eral studies reporting the solubilities of different aromatics in cation is quite different, the effect of benzene on the anion–
ILs.[39–41] cation interaction is almost the same.

Chem. Eur. J. 2014, 20, 15482 – 15492 www.chemeurj.org 15487  2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper

benzene is separating the cation


from the anion. As the anion is
separated from the cation,
cation–anion interactions are re-
duced and the polarizability of
the ions increased. Consequent-
ly, p-ion interactions with the
benzene are favored.
To understand what happened
to the protons of the benzene
when different cationic cores
were studied, all of the benzene
chemical shifts, DdHHB, were
compared at different mole
ratios of benzene/IL. The data
(see Figure 8) suggests two dif-
ferent trends. The nonaromatic-
based ILs induce two-fold higher
DdH of the HB values than the ILs
with aromatic cations, even
though the maximum benzene
solubility is essentially the same
in both types of ILs. This sug-
gests that the ILs with aromatic
cations are interacting with ben-
Figure 6. Proton chemical shift deviation (DdH) of [C4mim][NTf2] (top left), [C4mpy][NTf2] (top right), [C4mpyrr][NTf2] zene through p–p stacking, as in
(bottom left), and [C4mpip][NTf2] (bottom right) at different benzene/IL mole ratios. “free” benzene, whereas the in-

Figure 7. Fluorine chemical shift deviation of the anion (DdF) of [C4mim]- Figure 8. Benzene chemical shift deviation (DdH) in [C4mim][NTf2] (~),
[NTf2] (~), [C4mpy][NTf2] (^), [C4mpyrr][NTf2] (&), and [C4mpip][NTf2] (*) at dif- [C4mpy][NTf2] (^), [C4mpyrr][NTf2] (&), and[C4mpip][NTf2] (*) at different mole
ferent benzene/IL mole ratios. ratios of benzene/IL.

Bonhte et al.[37] suggested that chemical shift displace-


ments in imidazolium IL solutions with concentration can be teractions between nonaromatic cations and benzene are of
analyzed in terms of two distinct effects, namely, higher shifts the p-ion type.
with increased hydrogen bonding and lower shifts with in- Finally, the chemical shift trends were compared by calculat-
creased p-stacking of the imidazolium cation (or any aromatic ing the relative deviation, R(Dd)n, of the 1H chemical shift (DdH)
system). The [NTf2]anions in our study are among the least of Hx (the proton of the cation with highest chemical shift) and
basic anions and, therefore, have weak cation–anion interac- of the 19F chemical shift (DdF) of the [NTf2]anion (Fanion) at dif-
tion strength. Thus, as indicated above in the discussion of the ferent mole ratios (n) of benzene/IL. The chemical shift devia-
effect of anions, the lower d values determined for the cation’s tion at each benzene/IL mole ratio (DdHn or DdFn, n = 1, 2, 3, or
protons when benzene is added, seem to indicate that the 4) was compared to the maximum chemical shift deviations in

Chem. Eur. J. 2014, 20, 15482 – 15492 www.chemeurj.org 15488  2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper

saturated solutions with benzene (DdHmax or DdFmax) according


to Equations (2) and (3).

DdHn
RðDdÞn ¼ ð2Þ
DdH max

DdFn
RðDdÞn ¼ ð3Þ
DdF max

The values of R(Dd)n versus benzene mole ratio are present-


ed in Figure 9 and in Table S1 (in the Supporting Information).

Figure 10. Benzene solubility as a function of the maximum chemical shift


deviations in saturated solutions with benzene (DdHmax, ppm) of the proton
with the highest d value (Hx) in the cations: (^) [OHC2mim] + ; (~) [C2mim] + ;
(&) [C4mim] + ; (N) [C4mpy] + ; (*) [C4mpyrr] + ; (+) [C4mpip] + .

3. Conclusion

A correlation was observed between benzene solubility and


the NMR chemical shift deviations for the ILs studied. Figure 10
presents a comparison between the DdH of the proton of the
Figure 9. Relative deviation R(Dd)n of Hx (solid symbols) and Fanion (empty
cation with the highest chemical shift value (Hx) and the ben-
symbols) at different benzene/IL mole ratios.
zene solubility for different series of ILs grouped by anion,
[OAc] , [BF4] , [N(CN)2] , [PF6] , [NTf2] , and [FAP] . The data
confirm the close relationship between the deviation in the
The results indicate that the cation’s R(Dd)n is always higher chemical environment of the cation’s protons and benzene sol-
than that observed for the anion. It is also observed that the ubility. In general, the benzene solubility is dependent on the
highest deviation in the chemical-shift environment was ob- anion studied, that is, the anions that exhibit the strongest in-
served when the first mole of benzene was added, when at teractions with the cation (e.g., [OAc] , [BF4]) have the lowest
least 40 % of the total deviation occurs. With the addition of aromatic solubilities and the anions most weakly coordinated
a second mole of benzene, another high deviation is observed, (e.g., [NTf2] , [FAP]) solubilize aromatics the most.
roughly 30 % of the total. The first additions of benzene, there-
fore strongly affect the cation–anion interactions in which the Stronger cation–anion interactions lead to lower benzene
ions are displaced by the introduction of benzene. This “dis- solubility: Spectroscopic analysis supports the results dis-
tancing” of the cation and the anion is evident until 3 moles of cussed in Section 2.1, that the stronger the cation–anion inter-
benzene are added (in the case of [C2mim][NTf2] this is near actions, the less benzene can be solubilized. Interestingly,
the maximum solubility of 3.2 moles of benzene). Subsequent almost all ILs exhibit DdH within a small range of upfield chem-
additions of benzene result in only small deviations in the ical shifts, with the exception of the [OAc]-based ILs, in which
chemical shifts of the protons of each cation and the [NTf2] the cation–anion interionic interactions are much stronger,
fluorine atoms (R(Dd)n < 0.1 ppm). acting more as an ion pair. Previously, Fumino et al.[43] demon-
Overall, this data seems to indicate that benzene solvation strated that the interactions between ILs and different solvents
of the cations, due to p–ion interactions, occurs until the IL/ are dependent on two types of ion pairs, contact ion pairs
benzene ratio equals 3 moles (3 benzene per cation), after (CIPs) or solvent-separated ion pairs (SIPs). The authors sug-
which the remaining solubility is controlled by steric or free- gested that at infinite dilution the ions are solvated as CIPs in
volume effects, such as the longer alkyl-chain lengths of the weakly polar solvents, whereas for highly polar solvents they
cation discussed above. This would essentially entrap or clath- are solvated as SIPs.
rate the benzene in the LC phase, but would not strongly The importance of cation–anion interactions was also very
affect the cation–anion chemical environment. This effect clear- recently described for water solubility[44] and gas solubility.[45]
ly needs further study, including effective modeling of the sys- In the first paper, the authors observed similar trends in water
tems. solubility as observed here for benzene, in which water solubil-

Chem. Eur. J. 2014, 20, 15482 – 15492 www.chemeurj.org 15489  2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper

ity increased when strongly basic anions were combined with


a cation that possesses weaker cation–anion hydrogen-bond-
ing interactions.[44] The same trends were also observed for
CO2, SO2, and H2S solubility, and Damas et al.[45] have suggest-
ed a relationship between the increase of interaction of the
gas with the anion and the weakening of cation–anion interac-
tions. Such work is helping to provide a better understanding
of the specific ions which are most likely to interact with each
solute. For example, water should strongly interact with more
basic anions (e.g., [OAc]), while benzene appears to interact
more strongly with positively charged cations. Overall, the
growing body of literature suggests that more work is needed
to understand specific interactions with each ion and how
these interactions may overcome the specific interactions be-
tween the cation and anion in an IL.

Attractive interactions between the anion and benzene lead


to higher benzene solubilities: A second aspect of the work Figure 11. Solubility of benzene at different mole ratios (x) in [C2mim][OAc]x-
has shown that the anions can also have attractive interactions [NTf2]1x (Zero on the x axis corresponds to [C2mim][NTf2] and 1.0 to
[C2mim][OAc]).
with the aromatic solutes, as shown for the [N(CN)2]-based
ILs. The results indicate that ILs with [N(CN)2] anions have in-
creased aromatic solubility due to additional benzene–anion
interactions. Recently, Weber et al.[24] reviewed the intermolecu- be to use this knowledge to help understand the selectivity of
lar aromatic interactions with ILs, and suggested that the IL ILs for different types of aromatic compounds, such as those
anion could have an important role in IL–p interactions, due to used by the chemical industry in separations. We suggest the
a preferential coordination of the aromatic solvents by the use of multi-ionic systems (IL/benzene systems with more than
polar regions of the IL. We would suggest that further study one anion and cation, which we term double salt ILs or
using ILs with the [N(CN)2] anion and different aromatic com- DSILs[47]), in which ILs with lower and higher abilities to solubi-
pounds could help understand how the anion influences the lize benzene are combined to clathrate the aromatics.
“clathration” of aromatic solutes, as well as to determine what In our first study of this type, benzene solubility was evaluat-
molecular interactions induce such cage formation. We would ed in a DSIL we have previously characterized.[35] A strongly co-
also suggest that the IL research community continue to look ordinated IL, [C2mim][OAc], was combined with one of the
for more anions with similar behavior to [N(CN)2] in order to most weakly coordinated ILs, [C2mim][NTf2], in different mole
evaluate their potential use in increasing the selectivity for spe- ratios, in order to prepare a series of systems of the type
cific aromatic solutes. [C2mim][OAc]x[NTf2]1x. Benzene solubility in these systems was
evaluated following the procedures outlined above and the re-
Larger steric or free volume demands of the ions lead to sults are presented in Figure 11.
higher benzene solubilities: The results presented in Section The data in Figure 11 indicates that the benzene solubility
2.2 demonstrated that steric effects are also important in LC does not change linearly with [OAc]concentration. There is
formation. Previously, steric effects were reported by Shimo- a break point at x = 0.5 (1 mole of [OAc] per 2 moles of
mura et al.[46] where NMR data suggested that in [C12mim]- [C2mim] + ). When x < 0.5, benzene solubility decreased signifi-
[NTf2]/benzene solution, the imidazolium ring was “sand- cantly faster with increasing [OAc] than when observed for
wiched” between benzene molecules through cation–p inter- compositions with x > 0.5. The trends are in line with our previ-
actions above and below the imidazolium ring. Although, in ous studies of this DSIL[35] , which indicated preferential interac-
the system analyzed by these authors, cluster formation was tions (both Coulombic interactions and hydrogen bonding) be-
noted (as expected due to the longer alkyl chain lengths of tween the acetate anions compared to the [NTf2]anion with
the cation), it is interesting to note a breakpoint in their data the [C2mim] + cations. As suggested by the studies reported
correlating chemical-shift deviations at different mole ratios of here, such interactions would decrease the benzene solubility.
the [C12mim][NTf2]/benzene system and the respective benzene Overall, the data presented in Figure 11 suggest that one
solubility, which was probably due to the cation–anion interac- could tune aromatic solubility, allowing the enhancement of
tion. We also observed such behavior noted in Section 2.2. selectivity.
The use of the ILs to extract and separate aromatic com-
pounds can have significant industrial impact, and here we
4. Outlook
aimed to provide data that would bring us closer to under-
The study reported here has helped us to better understand standing what controls benzene solubility and LC formation,
the reasons for variable benzene solubility in different kinds of thus allowing better future design of specific IL systems for
ILs and the potential for forming LCs. The next challenge will this task. Using benzene as a model solute we highlighted the

Chem. Eur. J. 2014, 20, 15482 – 15492 www.chemeurj.org 15490  2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper

major interactions that we believe control the formation of LCs were examined in order to calculate the uncertainty in the deter-
including 1) the strength of the cation–anion interactions (the mination of the solubility.
main effect up to 2:1 benzene/IL mole ratio), 2) specific inter-
actions between the aromatics and anions (e.g., [N(CN)2]), and 5.3. Influence of water on benzene solubility
3) steric effects or free volumes of the ions. We hope that
The influence of water in ILs on the benzene solubility in the most
future molecular simulation and computational studies might hydrophilic IL studied, [C2mim][OAc], was evaluated. Five different
be brought to bear on the three concepts discussed above in aqueous solutions of this IL with water mole fractions (xwater) from
order to design both better ILs or IL systems and specific aro- 0.05 to 0.5 were prepared gravimetrically. Each solution was then
matic separation schemes. mixed with 10 molar equivalents of benzene, vigorously mixed,
and equilibrated under ambient conditions for 12 h. Three aliquots
of each lower LC phase were then taken to determine the benzene
5. Experimental Section solubility by 1H NMR as described above.

5.1. Chemicals 5.4. Spectroscopic studies of variable concentrations of ben-


The ILs used in this study were: 1-ethyl-3-methylimidazolium ace- zene in ILs
tate, [C2mim][OAc] (95 wt %); 1-ethyl-3-methylimidazolium dicyana-
Several molar ratios of IL-benzene mixtures (in the completely mis-
mide, [C2mim][N(CN)2] (98 wt %); 1-ethyl-3-methylimidazolium bis-
cible region) were prepared up to the maximum benzene solubili-
(trifluoromethanesulfonyl)imide, [C2mim][NTf2] (99 wt %); 1-(2-hy-
ty, and 1H NMR (and 19F NMR where appropriate) analysis of each
droxyethyl)-3-methylimidazolium bis(trifluoromethanesulfonyl)-
mixture was performed. The 1H NMR spectra were obtained utiliz-
imide, [OHC2mim][NTf2] (99 wt %); 1-butyl-3-methylimidazolium
ing a 500 MHz BrukerAvance NMR spectrometer (Karlsruhe, Germa-
acetate, [C4mim][OAc] (99 wt %); 1-butyl-3-methylimidazolium tet-
ny). Each sample was loaded solventless in a flame-sealed capillary,
rafluoroborate, [C4mim][BF4] (99 wt %); 1-butyl-3-methylimidazoli-
and the spectra were collected at 25 8C using CDCl3 as the external
um hexafluorophosphate, [C4mim][PF6] (99 wt %); 1-butyl-3-methyl-
lock. The 19F NMR spectra were obtained on a BrukerAvance
imidazolium dicyanamide, [C4mim][N(CN)2] (98 wt %); 1-butyl-3-
360 MHz NMR spectrometer (Karlsruhe, Germany), using trifluoro-
methylimidazolium bis(trifluoromethanesulfonyl)imide, [C4mim]-
acetic acid/CDCl3 solution as the external lock.
[NTf2] (99 wt %); 1-butyl-1-methylpyrrolidinium dicyanamide,
[C4mpyrr][N(CN)2] (98 wt %); 1-butyl-1-methylpyrrolidinium bis(tri-
fluoromethanesulfonyl)imide, [C4mpyrr][NTf2] (99 wt %); 1-butyl-1-
methylpyrrolidinium tris(pentafluoroethyl)trifluorophosphate,
Acknowledgements
[C4mpyrr][FAP] (high purity); 1-butyl-3-methylpyridinium tetrafluor-
oborate, [C4mpy][BF4] (99 wt %); 1-butyl-3-methylpyridinium bis(tri- We thank the Novartis-Massachusetts Institute of Technology
fluoromethanesulfonyl)imide, [C4mpy][NTf2] (99 wt %); and 1-butyl- (MIT) Center for Continuous Manufacturing (CCM) for financial
1-methylpiperidinium bis(trifluoromethanesulfonyl)imide, [C4mpip]- support.
[NTf2] (99 wt %). All the ILs were obtained from IoLiTec (Tuscaloosa,
AL) with the exception of [C4mpyrr][FAP], which was acquired from
Merck (Darmstadt, Germany). Before use, the ILs were dried to Keywords: aromatics · benzene · cation–anion strength · ionic
reduce the water content to a minimum by stirring under high liquids · clathrates
vacuum for 48 h. The water contents of the ILs were determined
with a Karl Fischer titrator (Mettler Toledo C20 coulometric KF) [1] R. D. Rogers, K. R. Seddon, Science 2003, 302, 792.
using AQUASTAR CombiCoulomat fritless methanol solution, sup- [2] T. Welton, Chem. Rev. 1999, 99, 2071.
plied by Merck as titrant. [3] M. J. Earle, K. R. Seddon, Pure Appl. Chem. 2000, 72, 1391.
[4] J. Dupont, R. F. de Souza, P. A. Z. Suarez, Chem. Rev. 2002, 102, 3667.
HPLC-grade benzene with a purity of  99.9 wt % was obtained [5] D. S. Silvester, Analyst 2011, 136, 4871.
from Sigma–Aldrich (St. Louis, MO) and used as received. Deuterat- [6] D. R. MacFarlane, M. J. Pringle, P. C. Howlett, M. Forsyth, Phys. Chem.
ed chloroform (CDCl3) was purchased from Cambridge Isotope Lab- Chem. Phys. 2010, 12, 1659.
oratories (Andover, MA). Trifluoroacetic acid used as internal sol- [7] K. Binnemans, Chem. Rev. 2005, 105, 4148.
vent for 19F NMR spectroscopy was acquired from Sigma–Aldrich [8] T. Welton, Chem. Rev. 1999, 99, 2071.
(St. Louis, MO) with a purity of 99 %. [9] L. A. Blanchard, J. F. Brennecke, Ind. Eng. Chem. Res. 2001, 40, 287.
[10] G. W. Meindersma, A. R. Hansmeier, A. B. de Haan, Ind. Eng. Chem. Res.
2010, 49, 7530.
5.2. Benzene solubility measurements [11] M. R. Heidari, B. Mokhatarani, N. Seghatoleslami, A. Sharifi, M. Mirzaei, J.
Chem. Thermodyn. 2012, 54, 310.
Experimental liquid–liquid equilibria (LLE) data were obtained by [12] G. W. Meindersma, A. J. G. Podt, A. B. de Haan, Fluid Process Technol.
mixing the respective ILs with 10 molar equivalents of benzene in 2005, 87, 59.
a vial and vortexing under ambient conditions. Spontaneous for- [13] A. Revelli, F. Mutelet, J. Jaubert, J. Phys. Chem. B 2010, 114, 4600.
mation of the LCs was observed by the increase in the IL phase [14] A. Marciniak, Fluid Phase Equilib. 2010, 294, 213.
volume. After mixing, each IL–benzene system was allowed to [15] E. J. Gonzlez, S. B. Bottini, S. Pereda, E. A. Macedo, Fluid Phase Equilib.
settle and equilibrated for 12 h for complete phase separation. 2014, 362, 163.
[16] J. K. D. Surette, L. Green, R. D. Singer, Chem. Commun. 1996, 2753.
Benzene solubility and any IL in the top benzene phase were de-
[17] J. L. Atwood, J. D. Atwood, Inorganic Compounds with Unusual Proper-
termined by 1H NMR analysis. The IL concentration in the organic ties, Advances in Chemistry Series no. 150, American Chemical Society,
phase was below the 1H NMR detection limit in all the systems Washington DC, 1976, p. 112.
evaluated. The benzene concentration in the lower LC phase was [18] J. L. Atwood, Recent Dev. Sep. Sci. 1977, 3, 195.
quantified by the molar ratio between the integrated number of [19] J. L. Atwood, in Inclusion Compounds, Vol. 1 (Eds.: J. L. Atwood, J. E. D.
protons of benzene and of IL. Three aliquots of each LC phase Davies, D. D. MacNicol), Academic Press, London, 1984.

Chem. Eur. J. 2014, 20, 15482 – 15492 www.chemeurj.org 15491  2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Full Paper

[20] J. D. Holbrey, W. M. Reichert, M. Nieuwenhuyzen, O. Sheppard, C. Har- [35] H. Wang, J. W. Brantley, G. Chatel, J. Shamshina, S. P. Kelley, J. F. B. Per-
dacre, R. D. Rogers, Chem. Commun. 2003, 476. eira, V. Debbeti, A. S. Myerson, R. D. Rogers, in preparation.
[21] C. G. Hanke, A. Johansson, J. B. Harper, R. M. Lynden-Bell, Chem. Phys. [36] D. A. Headley, N. M. Jackson, J. Phys. Org. Chem. 2002, 15, 52.
Lett. 2003, 374, 85. [37] P. Bonhte, A.-P. Dias, N. Papageorgiou, K. Kalyanasundaram, M. Grtzel,
[22] J. B. Harper, R. M. Lynden-Bell, Mol. Phys. 2004, 102, 85. Inorg. Chem. 1996, 35, 1168.
[23] K. Shimizu, M. F. Costa Gomes, A. A. H. Padua, L. P. N. Rebelo, J. N. Can- [38] A. Blahut, V. Dohal, J. Chem. Eng. Data 2011, 56, 4909.
ongia Lopes, J. Phys. Chem. B 2009, 113, 9894. [39] E. J. Gonzlez, P. F. Requejo, A. Domnguez, E. A. Macedo, Fluid Phase
[24] C. C. Weber, A. F. Masters, T. Maschemeyer, Green Chem. 2013, 15, 2655. Equilib. 2013, 360, 416.
[25] W. Meindersma, A. R. Hansmeier, A. B. de Haan, Ind. Eng. Chem. Res. [40] J. Lachwa, J. Szydlowski, A. Makowska, K. R. Seddon, J. M. S. S. Esperan-
2010, 49, 7530. Åa, H. J. R. Guedes, L. P. N. Rebelo, Green Chem. 2006, 8, 262.
[26] U. Domasnka, M. Krolikowska, Fluid Phase Equilib. 2011, 308, 55. [41] J. Grcia, J. S. Torrecilla, A. Fernndez, M. Oliet, F. Rodrguez, J. Chem.
[27] J. L. Atwood, S. G. Boot, P. C. Junk, M. T. May, J. Organomet. Chem. 1995, Thermodyn. 2010, 42, 144.
487, 7. [42] C. Chiappe, C. S. Pomelli, S. Rajamani, J. Phys. Chem. B 2011, 115, 9653.
[28] A. F. M. Cludio, L. Swift, J. P. Hallet, T. Welton, J. A. P. Coutinho, M. G. [43] K. Fumino, P. Stange, V. Fossog, R. Hempelmann, R. Ludwig, Angew.
Freire, Phys. Chem. Chem. Phys. 2014, 16, 6593. Chem. 2013, 125, 12667; Angew. Chem. Int. Ed. 2013, 52, 12439.
[29] M. A. Ab Rani, A. Brant, L. Crowthurst, A. Dolan, M. Lui, N. H. Hassan, [44] K. A. Kurnia, S. P. Pinho, J. A. P. Coutinho, Green Chem. 2014, 16, 3741.
J. P. Hallet, P. A. Hunt, H. Niedermeyer, J. M. Perez-ArÅandis, M. Schrems, [45] G. B. Damas, A. B. A. Dias, L. T. Costa, J. Phys. Chem. B 2014, 118, 9046.
T. Welton, R. Wilding, Phys. Chem. Chem. Phys. 2011, 13, 16831. [46] T. Shimomura, T. Takamuku, T. Yamaguchi, J. Phys. Chem. B 2011, 115,
[30] L. Crowhurst, P. R. Mawdsley, J. M. Perez-ArÅandis, P. A. Salter, T. Welton, 8518.
Phys. Chem. Chem. Phys. 2003, 5, 2790. [47] G. Chatel, J. F. B. Pereira, V. Debbeti, H. Wang, R. D. Rogers, Green Chem.
[31] R. Lungwitz, M. Friedrich, W. Linert, S. Spange, New J. Chem. 2008, 32, 2014, 16, 2051.
1493.
[32] R. Lungwitz, S. Spange, New J. Chem. 2008, 32, 392.
[33] P. G. Jessop, D. A. Jessop, D. Fu, L. Phan, Green Chem. 2012, 14, 1245.
[34] S. Chen, R. Vijayaraghavan, D. R. MacFarlane, E. I. Izgorodina, J. Phys. Received: July 4, 2014
Chem. B 2013, 117, 3186. Published online on October 8, 2014

Chem. Eur. J. 2014, 20, 15482 – 15492 www.chemeurj.org 15492  2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Você também pode gostar