Ionic fluids containing both strongly and weakly interacting ions of the same charge have unique ionic and chemical environments as a function of ion concentration

Article abstract of DOI:10.1002/cphc.201402894

Liquid multi-ion systems made by combining two or more salts can exhibit charge ordering and interactions not found in the parent salts, leading to new sets of properties. This is investigated herein by examining a liquid comprised of a single cation, 1-ethyl-3-methylimidazolium ([C₂mim]⁺), and two anions with different properties, acetate ([OAc]^-) and bis(trifluoromethylsulfonyl)imide ([NTf₂]^-). NMR and IR spectroscopy indicate that the electrostatic interactions are quite different from those in either [C₂mim][OAc] or [C₂mim][NTf₂]. This is attributed to the ability of [OAc]^- to form complexes with the [C₂mim]⁺ ions at greater than 1:1 stoichiometries by drawing [C₂mim]⁺ ions away from the less basic [NTf₂]^- ions. Solubility studies with molecular solvents (ethyl acetate, water) and pharmaceuticals (ibuprofen, diphenhydramine) show nonlinear trends as a function of ion content, which suggests that solubility can be tuned through changes in the ionic compositions.

Full text of DOI:10.1002/cphc.201402894

DOI: 10.1002/cphc.201402894
Articles
Ionic Fluids Containing Both Strongly and Weakly
Interacting Ions of the Same Charge Have Unique Ionic
and Chemical Environments as a Function of Ion
Concentration
Hui Wang,[a] Steven P. Kelley,[a] Jimmy W. Brantley III,[a, c] Gregory Chatel,[a, d] Julia Shamshina,[a]
Liquid multi-ion systems made by combining two or more
salts can exhibit charge ordering and interactions not found in
the parent salts, leading to new sets of properties. This is in-
vestigated herein by examining a liquid comprised of a single
cation, 1-ethyl-3-methylimidazolium ([C2mim]+), and two
anions with different properties, acetate ([OAc]À) and bis(tri-
fluoromethylsulfonyl)imide ([NTf2]À). NMR and IR spectroscopy
indicate that the electrostatic interactions are quite different
from those in either [C2mim][OAc] or [C2mim][NTf2]. This is at-
tributed to the ability of [OAc]À to form complexes with the
[C2mim]+ ions at greater than 1:1 stoichiometries by drawing
[C2mim]+ ions away from the less basic [NTf2]À ions. Solubility
studies with molecular solvents (ethyl acetate, water) and phar-
maceuticals (ibuprofen, diphenhydramine) show nonlinear
trends as a function of ion content, which suggests that solu-
bility can be tuned through changes in the ionic compositions.
Jorge F. B. Pereira,[a, e] Varun Debbeti,[a] Allan S. Myerson,[b] and Robin D. Rogers*[a]
1. Introduction
Over the past 20 years, ionic liquids (ILs, defined as liquids con-
sisting entirely of ions,[1] commonly with the arbitrary criterion
that they melt below 1008C) have received much attention as
interesting new solvents that can offer unique properties often
unavailable in traditional molecular solvents.[2] Depending on
the choice of ions, these may include negligible vapor pres-
sure; high chemical, thermal, and electrochemical stabilities;
and unique solvation properties. Perhaps even more interest-
ing is the fact that the physicochemical properties of these
“designer solvents”[3] can be tuned by the independent selec-
tion of the cation and anion to obtain the optimal solvent for
a specific application.[4]
One fundamental reason that the properties of ILs deviate
from those of molecular solvents is due to their nanoscale or-
dering. ILs are composed of ions, and hence, experience inter-
ionic interactions that yield long-lived associations.[5] Coulom-
bic interactions (i.e. electrostatic interactions) are considered to
be the major source of attraction in ILs, mainly depending on
the ion charges, the interionic distances, and the coordination
number,[6] but other interactions, such as van der Waals inter-
actions, dipole–dipole interactions, p–p stacking interactions,
and hydrogen bonding, may also be present.[7] In combination,
these interactions are responsible for organized ionic networks
and nanostructural segregation in IL media,[8] whereas delocali-
zation of the charges on the ions can reduce cation–anion in-
teractions and result in less charge ordering and nano-segrega-
tion.[9] Even if not all of the parameters affecting this domain
segregation are understood in detail yet,[10] the degree of
charge ordering is mainly governed by the nature of the
cation and anion, as well as the length/steric bulkiness of sub-
stituents on the ions.[11]
[a] Dr. H. Wang, S. P. Kelley, J. W. Brantley III, Dr. G. Chatel, Dr. J. Shamshina,
Dr. J. F. B. Pereira, V. Debbeti, Prof. R. D. Rogers
Center for Green Manufacturing and Department of Chemistry
The University of Alabama, Tuscaloosa, AL 35487 (USA)
E-mail: rdrogers@ua.edu
[b] Prof. A. S. Myerson
Novartis-MIT Center for Continuous Manufacturing
and Department of Chemical Engineering
Massachusetts Institute of Technology
77 Massachusetts Avenue
66-568, Cambridge, MA 02139 (USA)
[c] J. W. Brantley III
Current Address: 11 Forest Street Hazlehurst, GA 31539 (USA)
[d] Dr. G. Chatel
Current Address: Institut de Chimie des Milieux
et Matꢀriaux de Poitiers, Universitꢀ de Poitiers
86073 Poitiers Cedex 9 (France)
Perhaps because of the tunability and structural complexity
already present in binary ILs, relatively little attention has been
paid to ILs with three or more ions. Such systems can be pre-
pared easily by combining two or more ILs or by dissolving
solid salts in an IL. The combinations of ILs that have been
studied so far tend to show near ideal mixing behavior;[12] this
[e] Dr. J. F. B. Pereira
Current address: Department of Bioprocess and Biotechnology
School of Pharmaceutical Sciences
UNESP—Universidade Estadual Paulista
14801-902—Araraquara, SP (Brazil)
Supporting Information for this article is available on the WWW under
http://dx.doi.org/10.1002/cphc.201402894.
ChemPhysChem 0000, 00, 0 – 0
1
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Articles
makes it possible to predict certain properties of these sys-
tems, such as density.[13] Such simple behavior may seem sur-
prising in light of the importance of nanostructuring on the
properties of ILs, and indeed other properties, such as spec-
troscopy and solvating power, often show dramatic, nonlinear
changes as a function of composition.[14] We have advocated
considering these combinations as new compounds rather
than mixtures [specifically, molten analogues of crystalline
double salts or “double salt ionic liquids” (DSILs)] because the
new interactions that form between the ions of different ILs
upon mixing are ionic bonds; a form of chemical bonding.[15]
Regardless of issues with nomenclature, the ability to predict
certain physical properties, while dramatically changing the
chemical properties, makes these multi-ion IL systems very
useful for all kinds of chemical processes.
2. Results and Discussion
[C2mim][OAc] and [C2mim][NTf2] were purchased from Ionic
Liquids Technologies, Inc. (Iolitec Inc., Tuscaloosa, AL) and
dried under high vacuum at 608C for 48 h (to reduce the
water content to 1604.1 and 651.2 ppm, respectively), as de-
scribed
in
the
Experimental
section.
The
DSILs
[C2mim][OAc]x[NTf2](1Àx) (in which x is the [OAc]À/[C2mim]+
molar ratio) were prepared as approximately 6 g samples by
mass addition of the corresponding amount of each IL with
x=0.10, 0.20, 0.33, 0.50, 0.67, 0.80, and 0.90. After 1 h of mag-
netic stirring, the samples were dried under high vacuum at
608C for 24 h and stored under argon. All of the samples were
homogeneous with water contents below 1600 ppm. The den-
sity and viscosity of the DSILs were measured, and the results
(Figure S1 in the Supporting Information) showed that density
decreased with increasing acetate concentration and viscosity
increased with increasing [OAc]À/[C2mim]+ molar ratio. During
the preparation of this manuscript, two reports were published
by Stark et al. in which the physical properties and 1H NMR
spectra of the [C2mim][OAc]x[NTf2](1Àx) system were also de-
scribed.[13,14] The density and viscosity data shown in this study
are in good agreement with those reported by Stark et al.
We are interested in better understanding how competition
between ions of the same charge for the most favorable coun-
Figure 1. Structures of the ions [C2mim]+, acetate ([OAc]À), and bis(trifluoro-
methylsulfonyl)imide ([NTf2]À).
2.1. NMR Spectroscopy
1H NMR chemical shifts in imidazolium ILs reflect interionic in-
teractions and are proportional to the strength of such interac-
tions.[20] By comparing the magnitude and direction of chemi-
cal shifts in the NMR spectra of the DSILs to those of the
parent ILs as a function of the molar ratio of the ions (x), one
can suggest the probable nature, type, and strength of the in-
terionic interactions.[21] 1H NMR data for [C2mim][OAc]x[NTf2](1Àx)
were collected by loading neat samples in capillaries with
CDCl3 as the external lock and were compared with those of
[C2mim][OAc] and [C2mim][NTf2]. The spectra are shown in Fig-
ure S2 in the Supporting Information, and the chemical shifts
of the imidazolium ring protons (H-2, H-4, and H-5, see
Figure 1 for the numbering scheme) and the [OAc]À methyl
protons as a function of [OAc]À/[C2mim]+ molar ratio (x) are il-
lustrated in Figure 2.
The results of our 1H NMR spectroscopy measurements
show that all proton signals on the imidazolium ring (H-2, H-4,
and H-5) shift downfield with increasing [OAc]À/[C2mim]+
molar ratio, whereas the [OAc]À methyl protons shift upfield.
The largest shift is associated with the imidazolium H-2 proton
(downfield shift Dd of ca. 2 ppm), and this shift has a nonlinear
dependence on acetate concentration. There is a corresponding
nonlinear upfield Dd of the methyl protons in the [OAc]À
anions as the [OAc]À concentration increases. On the contrary,
Dd for the H-4 and H-5 proton signals is smaller and more line-
arly dependent on [OAc]À concentration with a downfield shift
of about 1 ppm. Lastly, the rate of change of the H-2 proton
chemical shift becomes smaller at a 1:1 ratio of [OAc]À/[NTf2]À,
before which H-2 shifts downfield at a rate more than three
times that of H-4 or H-5, and after which it is approximately as
sensitive to the addition of [OAc]À as the C-4 and C-5 protons.
terion interactions leads to new chemical environments that
might be used for specific separations applications. To study
this effect, we have conducted an investigation into the spec-
troscopic properties (1H NMR, 13C NMR, and FTIR spectroscopy)
and solubility of molecular solvents (ethyl acetate and water)
in
a
system composed of 1-ethyl-3-methylimidazolium
([C2mim]+) cations and both acetate ([OAc]À) and bis(trifluoro-
methylsulfonyl)imide ([NTf2]À) anions (Figure 1) in varying con-
centrations. Notably, due to the great difference in basicity of
the two anions, the two-ion “parent” ILs, [C2mim][OAc] and
[C2mim][NTf2], have dramatically different solvent properties
with Kamlet–Taft b parameters (representing hydrogen-bond
basicity) of 1.06 for [C2mim][OAc] and 0.23 for
[C2mim][NTf2].[16] The former is totally miscible with water,
poorly miscible with EtOAc,[17] and can dissolve biopolymers,
such as cellulose and chitin,[18] whereas the latter has very low
miscibility with water, is totally miscible with EtOAc, and
cannot dissolve the biopolymers,[19] and yet the two ILs are to-
tally miscible with each other. We have also investigated the
solubility of two active pharmaceutical ingredients (APIs; ibu-
profen and diphenhydramine) to understand how tuning the
hydrogen-bond basicity of the DSIL can lead to control over
solubility and the separation of industrially relevant com-
pounds, as well as the separation of [C2mim][OAc] from
[C2mim][NTf2]. These are in keeping with interest in applying
these systems to industrial processes, such as separations, in
which recycling of the IL is important.
&
ChemPhysChem 0000, 00, 0 – 0
www.chemphyschem.org
2
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Articles
A comparison of the known
crystal structures of small dialky-
limidazolium [OAc]À and [NTf2]À
salts (Figure 3) offers insight into
the spectroscopic changes ob-
served by NMR spectroscopy.
Crystal structures for 1,3-diethyli-
midazolium acetate ([C2C2im]
[OAc]),[22] 1,3-dimethylimidazoli-
um bis(trifluoromethylsulfonyl)i-
mide ([C1mim][NTf2]),[23] and
[C2mim][NTf2] have been report-
ed.[24] The [C2mim]+ cations (and
dialkylimidazolium cations in
general) are considered to have
five sites for interactions with
anions.[20,25] The crystal structures
of both [NTf2]À and [OAc]À salts
1
Figure 2. H NMR chemical shifts of A) the imidazolium ring protons ( : H-2, : H-4, : H-5) and B) [OAc]À methyl
!
&
*
^
protons ( : H-9) in [C2mim][OAc]x[NTf2](1Àx) (zero on the x axis corresponds to [C2mim][NTf2] and 1.0 to [C2mim][OAc]).
have a fairly isotropic distribution of anions around the imida-
zolium cations and interactions that are typical of dialkylimida-
zolium salts, with anions occupying Coulombic and hydrogen-
bonding sites (ring and a-hydrogen atoms). However, owing
to differences in the basicity of the ions and charge localiza-
tion, the contacts representative of hydrogen bonds in the
[NTf2]À salts are much longer than those in [C2C2im][OAc],
whereas the out-of-plane Coulombic interactions are slightly
shorter. The greater strength of the interactions in
[C2C2im][OAc] also appears to result in overall shorter cation–
anion contacts, and there are typically more [NTf2]À anions sur-
rounding each cation than [OAc]À ions, despite the fact that
the [NTf2]À ions are much larger. These structural features sug-
gest that, in addition to differences in interaction strength,
there are chemical differences in the nature of the interactions
preferred by [NTf2]À and [OAc]À ions.
In [C2mim][OAc]x[NTf2](1Àx), there will be competition be-
tween the anions for the most favorable interaction sites on
the cations. Because the [OAc]À anion is more basic and has
higher cohesive energy and stronger ion packing[26] than that
of the charge-diffuse [NTf2]À,[27] stronger hydrogen-bonding
and Coulombic interactions are expected between the
[C2mim]+ cation and [OAc]À.[28] As the [OAc]À concentration in-
creases (i.e. upon going from x=0 to 1.0), each [C2mim]+
cation can hydrogen bond to or interact with more [OAc]À
anions. In any compositional range other than pure [C2mim]
[OAc], the molar ratios of [C2mim]+ and [OAc]À are not 1:1,
and this preferential interaction of [C2mim]+ ions with [OAc]À
ions should lead to structures that cannot be formed in the
neat IL.
1
The H NMR chemical shifts observed for the H-2, H-4, and
H-5 protons (Figure 2A) are indicative of increased interactions
between [C2mim]+ and [OAc]À as the abundance of the [OAc]À
increases, in accordance with literature reports that the
1H NMR chemical shifts for the imidazolium cation generally
shift downfield with increasing cation–anion interaction
strength.[29] Nonlinearity observed (Figure 2) for the 1H NMR
spectroscopy signals of the cation H-2 proton and the [OAc]À
anion methyl protons can be attributed to the presence of
Figure 3. Ion packing environments showing contacts less than the sum of
the van der Waals radii. Dashed lines indicate short contacts to imidazolium
ring hydrogen and carbon atoms (a-hydrogen atom contacts excluded) with
donor–acceptor distances in ꢁ. From top to bottom: [C2C2im][OAc] (cation
and anion), [C1mim][NTf2], [C2mim][cis-NTf2], and [C2mim][trans-NTf2] (cations
only). Redrawn by using coordinates from Refs. [22–24].
ChemPhysChem 0000, 00, 0 – 0
www.chemphyschem.org
3
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Articles
both hydrogen-bonding and Coulombic interactions around
the acidic C-2 position. In this case, the chemical shift of the
H-2 proton is not only affected by directional hydrogen bond-
ing, but also by anion associations in either of the two most fa-
vorable Coulombic sites above and below the imidazolium
ring.[30] We believe the greater change in chemical shift for
proton H-2 at low [OAc]À concentrations (x<0.5) is because
[OAc]À ions are initially more likely to occupy sites near C-2
than the hydrogen-bonding sites at C-4 and C-5. This is in ac-
cordance with predicted and measured cation–anion pair dis-
tribution functions in liquid imidazolium [OAc]À[31] and [NTf2]À
ILs.[32] The chemical shifts of H-4 and H-5 are not as sensitive to
the Coulombic interaction and are likely to be only affected by
greater hydrogen bonding with [OAc]À. At x=0.5, which corre-
sponds to a 1:2 [OAc]À/[C2mim]+ ratio, the downfield shift in
the H-2 proton begins to change at the same rate as the H-4
and H-5 protons. At x=0.5 (1 acetate for every 2 cations),
there is one [OAc]À oxygen atom for every C2 position, so any
[OAc]À ions added above this concentration are equally likely
to hydrogen bond to any of the three ring protons.
13C NMR signals, although the authors did acknowledge that
this was the opposite of what was expected.[21] In agreement
1
with the H NMR spectroscopy trends noted above, the chemi-
cal shift of C-2 changes more dramatically (with a slope of ca.
5) up to x=0.5 (1 acetate for every 2 cations, or as noted
above, 1 oxygen atom per cation) than the modest shift ob-
served for the C-4 and C-5 carbon atoms. From x=0.5 to 1.0,
the downfield shift of the signals for all three carbon atoms is
approximately the same. Because the carbon atoms on the
[OAc]À ion are farther away from the interaction site, their 13C
chemical shifts are less sensitive to cation–anion interactions
and thus have been omitted from the discussion.
Taken together the NMR spectroscopy data, the preferential
interaction of [OAc]À around the C-2 position of the imidazoli-
um ring is suggested, with the in-plane hydrogen-bonding in-
teractions of secondary importance, yet still quite strong. This
is based on the observation of high- and low-sensitivity re-
gimes for the chemical shift of the C-2 proton as a function of
[C2mim]+/[OAc]À ratio with a transition from high to low sensi-
tivity at about x=0.5, at which point there is one [OAc]À
oxygen atom available for every cation. At x=0.5, each cation
could conceivably have one strong Coulombic interaction with
an [OAc]À anion, for example, in a sandwich structure with two
cations interacting with the oxygen atoms of a single [OAc]À
anion. Interestingly, in the crystal structures previously dis-
cussed, only one strong Coulombic interaction is observed per
cation. This is probably because the first interaction polarizes
the positive charge in one direction, which weakens any
others on the other side of the ring. The [OAc]À ion in crystal-
line [C2C2im][OAc] is only observed to interact Coulombically
with one cation because the 1:1 ratio of [OAc]À to [C2C2im]+
requires the neighboring cations to participate in Coulombic
The spectroscopic discussion by Stark et al. also noted the
nonideality of the spectra (the difference between the NMR
chemical shifts of each DSIL versus the linear weighted aver-
age).[13,14] The authors concluded that based on large devia-
tions for the C-2 hydrogen atom [C2mim]+ cations preferential-
ly interacted with [OAc]À due to hydrogen bonding. This was
also in agreement with their previous findings that weakly
basic ILs, such as [C2mim][NTf2], prevent biomass dissolution
on [C2mim][OAc] by tying up the basic anions.[33]
The 13C NMR spectra (Figure S3 in the Supporting Informa-
tion) provide additional support for the interactions discussed
above. The chemical shifts of the imidazolium ring carbon
atoms (C-2, C-4, and C-5) move downfield for all three carbon
atoms with increasing [OAc]À/[C2mim]+ molar ratios (Figure 4).
It has been shown through a combination of 13C NMR and X-
ray photoelectron spectroscopy (a more direct measurement
of electron density) that an increase in electron density on the
carbon atoms due to stronger interactions with more basic
anions is correlated with a downfield shift of imidazolium
interactions with other [OAc]À ions. In [C2mim][OAc]0.50[NTf2]0.50
,
the [OAc]À ion appears to “steal” cations from the weakly coor-
dinating [NTf2]À ion, leading to ion clusters that are probably
absent in the neat IL.
2.2. Infrared Spectroscopy
The DSILs of [C2mim][OAc]x[NTf2](1Àx) were further analyzed by
attenuated total reflectance Fourier transform infrared (ATR-
FTIR) spectroscopy to better understand the effects of compo-
sition changes on the anions. While the 1H NMR signal of
[OAc]À is only weakly affected by intermolecular contacts, the
frequencies of the symmetric and asymmetric stretching
modes of acetate ions are quite sensitive to their environment.
For metal acetate salts, the asymmetric and symmetric stretch-
ing modes of [OAc]À are reported to occur between n˜ =1500–
1550 and 1340–1360 cmÀ1 [34]
When protonated, these bands
.
split further into the C=O and CÀO stretching modes of acetic
acid at n˜ =1712 and 1280 cmÀ1, respectively.[35]
Because the negatively charged oxygen atoms of [OAc]À
accept hydrogen bonds from the [C2mim]+ cation, the
carbon–oxygen bonds become unequal compared with the
symmetric bonds of “free” [OAc]À ion, and instead approach
the inequivalent CÀOH and C=O bonds found in acetic acid.
Figure 4. 13C NMR chemical shifts of the imidazolium ring carbon atoms (
:
*
!
&
C-2, : C-4, : C-5) in [C2mim][OAc]x[NTf2](1Àx) (zero on the x axis corresponds
to [C2mim][NTf2] and 1.0 to [C2mim][OAc]).
&
ChemPhysChem 0000, 00, 0 – 0
www.chemphyschem.org
4
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Articles
Therefore, the IR bands most af-
fected by hydrogen bonding will
be the CÀO and C=O stretches
of [OAc]À.
The bond inequivalency and
associated effects are conse-
quences of the interactions be-
tween the [OAc]À oxygen and
imidazolium hydrogen atoms, in
which the negative oxygen atom
pulls the positive hydrogen
atom closer to it, causing polari-
zation of the imidazolium CÀH
bond (and increasing the intensi-
ty of CÀH stretching) and re-
spective polarization of the ace-
tate CÀO bond. The vacant s*
molecular orbital of the imidazo-
lium CÀH bond overlaps with
Figure 5. FTIR spectra of [C2mim][OAc]x[NTf2](1Àx) as a function of the [OAc]À/[C2mim]+ molar ratio: A) [OAc]À CÀO
symmetric stretch (from top to bottom: x=0.0 to 1.0) and B) S=O stretch (from top to bottom: x=1.0 to 0.0)
(zero corresponds to [C2mim][NTf2] and 1.0 to [C2mim][OAc]).
the filled molecular orbital of oxygen (the lone pair). Electron
density from the [OAc]À oxygen atom is transferred to the CÀH
antibonding orbital, which makes the CÀH bond weaker and
longer, whereas the [OAc]À CÀO bond is weakened and length-
ened as well (relative to “free” [OAc]À).
1373 cmÀ1 (x=1.0; Figure 5A) is observed for the CÀO sym-
metric stretch. Unfortunately, the asymmetric stretch of [OAc]À
(n˜ ꢀ1558 cmÀ1 in [C2mim][OAc]) overlaps with imidazolium
CH3(N) and CH2(N) stretches[36] at n˜ ꢀ1575 cmÀ1 (Figure S4b in
the Supporting Information). Because of interference from the
cation (the same interaction that causes a blueshift in the
anion band would cause a redshift in the cation band due to
increased carbene character[37]), we do not discuss this stretch
in detail.
Crystallographic evidence supports that hydrogen bonding
affects the [OAc]À CÀO bond more than the out-of-plane Cou-
lombic interactions. In the crystal structure of [C2C2im][OAc][22]
(Figure 3), the [OAc]À oxygen atoms that interact with C-2
through hydrogen bonding in-plane and through Coulombic
interactions out-of-plane are symmetry inequivalent. The hy-
drogen-bonded CÀ2···O distance of 3.025(2) ꢁ is shorter than
the out-of-plane CÀ2···O distance of 3.115(2) ꢁ, which is a possi-
ble indicator of the greater degree of covalency in the hydro-
gen-bonding interaction (C-2ÀH···O). The [OAc]À CÀO bonds
are asymmetric—the hydrogen-bonded oxygen has a longer
CÀO bond (1.258(1) ꢁ) than the other oxygen atom
(1.245(2) ꢁ)—which indicates that the hydrogen-bonded
oxygen atom has more sp3 character.[22] Changes in these
bonds are most easily detected by IR spectroscopy.
The S=O stretching mode of the [NTf2]À anion occurs in
a
part of the spectrum where the background from
[C2mim][OAc] is relatively flat and can be easily identified at
n˜ =1131 cmÀ1, which is in agreement with the published
value.[38] Because the cations should interact preferentially with
the [OAc]À anions as the [OAc]À ion concentration increases,
the [NTf2]À ions should interact less and less with the cations
as the [OAc]À concentration increases. This is reflected in the
IR spectroscopy data shown in Figure 5B for which a blueshift
is observed for the symmetric S=O vibration of [NTf2]À from
n˜ =1131 (x=0.0) to 1136 cmÀ1 (x=0.9); this indicates greater
double-bond character and less hydrogen bonding as [OAc]À
concentration increases. Because the [NTf2]À anion is charge
diffuse and interacts weakly with the cations, even in the ab-
sence of competition, the shift observed in the S=O vibration
is much smaller than the shift observed for the [OAc]À CÀO
symmetric stretch.
At low [OAc]À concentrations, the [OAc]À ion is less acetic
acid-like because there are so many competing cations that
the cation–anion interactions are essentially isotropic and do
not break the symmetry of the carboxylate. When the [OAc]À
concentration is higher, [OAc]À has a slightly greater negative
charge because it interacts with fewer cations, but it is more
dissymmetric. With fewer competing cations, the anion is able
to form more directional and more covalent interactions that
have the greatest effect on the structure of the [OAc]À ion.
This is consistent with the NMR spectroscopy data, which
show that cation–anion interactions get stronger as the [OAc]À
concentration increases, but is also consistent with IR spectros-
copy, which says that the acetic acid like character of the
[OAc]À ion increases with increasing acetate concentration
(again, as a result of the decreasing number of cations per ace-
tate anion).
Overall, the FTIR data supports the conclusions reached
from examining the NMR spectroscopy data, which is that the
[OAc]À anion outcompetes [NTf2]À for close contacts with
[C2mim]+. The IR spectroscopy data also indicates that, as the
number of and strength of interactions changes, the electronic
environment of the [OAc]À anion also changes (from less to
more acetic acid character as the [OAc]À/[C2mim]+ ratio in-
creases). This shows that each [C2mim][OAc]x[NTf2](1Àx) repre-
sents not only a unique combination of ions, but also one in
which the properties are based on the combined, though com-
petitive, Coulombic interactions and hydrogen bonding be-
tween each of the two anions with the cation.
In [C2mim][OAc]x[NTf2](1Àx) systems with an increasing [OAc]À/
[C2mim]+ molar ratio, a redshift from n˜ =1390 (x=0.1) to
ChemPhysChem 0000, 00, 0 – 0
www.chemphyschem.org
5
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Articles
2.3. Solubility of Small Molecules as a Probe for Chemical
Properties
The spectroscopic data given above confirms that hydrogen
bonding plays an important, although not necessarily predomi-
nant, role in DSILs with the composition [C2mim][OAc]x[NTf2](1Àx)
.
Hydrogen bonding should, however, have an even greater role
in the solvent properties of these systems. For example, it has
been suggested that the differences in the b parameter, which
represents hydrogen-bond basicity, are the main reason for dif-
fering solubilities in ILs, and that the anion plays the largest role
in most cases of solvation.[16] In the [C2mim][OAc]x[NTf2](1Àx) sys-
tems, the a parameter, which quantifies the hydrogen-bond-do-
nating ability, and the p* value (based upon changes in aroma-
ticity and various p interactions) will be approximately constant
due to the presence of a common cation. The b values of the
parent ILs (Table 1), however, are quite different because of the
different hydrogen-bond basicities of the anions. It should
therefore be possible to finely tune the hydrogen-bond basicity
in [C2mim][OAc]x[NTf2](1Àx) by controlling the [OAc]À/[C2mim]+
molar ratio. Furthermore, if solvation is controlled by aggregates
of competing ions rather than fully dissociated ions, the abrupt
changes in speciation implied by the spectroscopic study may
be manifested as abrupt changes in solubility.
~
&
Figure 6. Solubility of EtOAc ( : left axis) and H2O ( : right axis) in
[C2mim][OAc]x[NTf2](1Àx) (zero corresponds to [C2mim][NTf2] and 1.0 to
[C2mim][OAc]).
that of water (Figure 6, ~) increased with increasing
[OAc]À/[C2mim]+ molar ratio. The solubility of water increases
slowly up to a [OAc]À/[C2mim]+ ratio of approximately 0.33,
after which the solubility increases more quickly, and the
EtOAc solubility decreases rapidly until approximately x=0.67
after which the rate of decrease slows.
The solubility of EtOAc in
[C2mim][NTf2] appears to be con-
Table 1. Water content and solubility data for [C2mim][OAc], [C2mim][NTf2], and [C2mim][OAc]x[NTf2](1Àx)
.
trolled by hydrogen-bonding in-
teractions with the cation; EtOAc
is a better hydrogen-bond ac-
ceptor than [NTf2]À.[41] As the
abundance of the much stronger
hydrogen-bond acceptor [OAc]À
anion increases, the cation inter-
acts more strongly with this
more basic anion and the solubil-
ity of EtOAc decreases. The same
rationale can be used to under-
stand the increasing solubility of
water in [C2mim][OAc]x[NTf2](1Àx)
as the abundance of [OAc]À in-
creases. Water readily solvates
both [C2mim]+ and [OAc]À,
[C2mim][OAc]x[NTf2](1Àx)
Water content Solubility [mol solute per mol IL or DSIL]
[ppm]
EtOAc
H2O
Ibuprofen
Diphenhydramine
[C2mim][NTf2]
651.2
M[a]
0.45Æ0
0.03Æ0.01 2.07Æ0.07
(a=0.63, b=0.23, p*=1.00[16]
[C2mim][OAc]0.10[NTf2]0.90
[C2mim][OAc]0.20[NTf2]0.80
[C2mim][OAc]0.33[NTf2]0.67
[C2mim][OAc]0.50[NTf2]0.50
[C2mim][OAc]0.67[NTf2]0.33
[C2mim][OAc]0.80[NTf2]0.20
[C2mim][OAc]0.90[NTf2]0.10
[C2mim][OAc]
)
791.0
820.8
778.4
829.2
1246.3
992.1
M[a]
0.50Æ0.03 0.31Æ0.02 1.75Æ0.10
10.62Æ0.05 0.58Æ0.03 0.55Æ0.03 1.38Æ0.06
8.30Æ0.17
5.44Æ0.22
2.52Æ0.07
1.09Æ0.09
0.53Æ0.11
0.25Æ0.07
0.82Æ0.02 0.93Æ0.02 0.85Æ0.07
1.29Æ0.04 1.42Æ0.09 0.42Æ0.04
1.81Æ0.02 2.14Æ0.14 0.21Æ0.03
2.34Æ0.05 2.42Æ0.20 0.11Æ0.02
2.92Æ0.10 2.99Æ0.09 0.08Æ0.02
1584.7
1604.1
M[a]
3.86Æ0.11 0.06Æ0.02
(a=0.57, b=1.06, p*=0.97[16]
)
[a] M: Miscible in all proportions.
2.3.1. Solubility of Ethyl Acetate and Water
whereas the hydrophobic nature of the weak hydrogen-bond
acceptor [NTf2]À reduces water solubility.[42]
To test this hypothesis, the solubilities of both a polar protic
solvent (water, polarity index: 9)[39] and a polar aprotic solvent
(EtOAc, polarity index: 4.4)[39] were measured at 258C in
[C2mim][OAc]x[NTf2](1Àx) as a function of acetate concentration.
Each solute was added dropwise to 1.0 g of each sample until
a saturation point was reached (observed visually as a turbid
solution). Compositional analysis was performed by means of
1H NMR spectroscopy, with which the solute/[C2mim]+ molar
ratios were calculated through direct integration of appropri-
ate signals.[40]
Although more study is needed, the solubility data is sug-
gestive of a break in the solubility trends at about x=0.33 for
increasing water solubility (at which point there is 1 [OAc]À
anion for every 3 [C2mim]+ cations) and about x=0.67 for de-
creasing EtOAc solubility (at which point there are 2 [OAc]À
anions for every 3 [C2mim]+ cations). Preferential interactions
of the ions could affect the solubilities of water and EtOAc be-
cause water would prefer to interact through hydrogen bond-
ing to the [OAc]À anion[43] and EtOAc would prefer to interact
by accepting hydrogen-bond interactions from the [C2mim]+
cations.
Water is completely miscible with [C2mim][OAc], and EtOAc
is completely miscible with [C2mim][NTf2]. As expected, the sol-
ubility of EtOAc in [C2mim][OAc]x[NTf2](1Àx) (Figure 6, &) de-
creased with increasing [OAc]À/[C2mim]+ molar ratio, whereas
In a recent study, Garcꢂa et al. reported the liquid–liquid
extraction of toluene from heptane by using mixtures of
&
ChemPhysChem 0000, 00, 0 – 0
www.chemphyschem.org
6
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Articles
1-butyl-4-methylpyridinium
imide ([4-C1C4py][NTf2]) and 1-ethyl-3-methylimidazolium
1,1,2,2-tetrafluoroethanesulfonate ([C2mim][CH2F2CF2SO3]),
bis(trifluoromethylsulfonyl)-
as those with H2O and EtOAc, but did appear to occur near
specific molar ratios (3 [C2mim]+ cations per 2 [OAc]À anions
for ibuprofen and 2 [C2mim]+ cations per [OAc]À anion for di-
phenhydramine). The solubilities of both molecular solvents
and APIs in [C2mim][OAc]x[NTf2](1Àx) complement the spectro-
scopic data; this suggests that the solvent properties of
[C2mim][OAc]x[NTf2](1Àx) can be finely tuned by controlling the
abundance of unique interactions (i.e. Coulombic interactions
and hydrogen bonding) between the ions in this system.
namely, [C2mim]x[4-C1C4py](1Àx)[CH2F2CF2SO3]x[NTf2](1Àx) at 408C
and atmospheric pressure.[44] Although no chemical
explanation
was
provided
in
this
study,
[C2mim]0.70[4-C1C4py]0.30[CH2F2CF2SO3]0.70[NTf2]0.30 provided effi-
cient extraction of toluene from heptane with a higher sepa-
ration factor relative to those of sulfalone or [4-C1C4py][NTf2].
This example and our current study indicate the possibility
that DSILs can provide systems with unique and tunable solu-
bilities that might find use in many new separation tech-
niques.
2.4. Separation of [C2mim][OAc] and [C2mim][NTf2] from
[C2mim][OAc]x[NTf2](1Àx)
Separation of the ILs would be useful in chemical processes as
a way to recycle them. Above their saturation limits in a given
[C2mim][OAc]x[NTf2](1Àx), water and EtOAc will form separate
(lighter) phases. Given the different solubilities of [C2mim][OAc]
and [C2mim][NTf2] in these solvents, we hypothesized that
[C2mim][OAc] and [C2mim][NTf2] could be separated from
[C2mim][OAc]x[NTf2](1Àx) through solvent extraction. To test this,
equal masses of [C2mim][OAc]0.50[NTf2]0.50 and water were
mixed together, and the separation procedure outlined in the
Experimental Section was followed. The recovered ILs from the
top and bottom phases were analyzed by NMR spectroscopy
to identify the compositions.
2.3.2. Solubility of APIs
To apply the tunability of solvation in [C2mim][OAc]x[NTf2](1Àx)
towards a problem with industrial relevance, the solubilities of
two APIs, ibuprofen free acid, which is a good hydrogen-bond
donor, and diphenhydramine free base, which is a good hydro-
gen-bond acceptor, were determined. Each of the prepared
[C2mim][OAc]x[NTf2](1Àx) liquids were saturated with the desired
API at 258C through weight addition and constant stirring.
Concentrations of each API in [C2mim][OAc]x[NTf2](1Àx) (Figure 7)
1H NMR spectroscopy analysis of the bottom phase (Fig-
ure S6 in the Supporting Information) indicated that this phase
was [C2mim][NTf2], and no [OAc]À protons were detected (the
small signal at d=1.56 ppm was attributed to the presence of
water in CDCl3). After removal of water, the upper phase was
identified as [C2mim][OAc] with no detectable [C2mim][NTf2]
present, as indicated by signal integrations of the imidazolium
ring protons and [OAc]À methyl protons (Figure S7 in the Sup-
porting Information). To increase our detection limits, the con-
centration of [NTf2]À in recovered [C2mim][OAc] was deter-
mined by 19F NMR spectroscopy (Figure S8 in the Supporting
Information) by using trifluoroacetic acid (TFA) as an internal
standard. Based on signal integrations, the concentration of
[NTf2]À anion in recovered [C2mim][OAc] was 0.5 mol%.
We also tried to separate [C2mim][OAc] and [C2mim][NTf2]
from [C2mim][OAc]0.50[NTf2]0.50 by extraction with EtOAc
through a similar procedure, as described in the Experimental
Section. The results (Figure S9 in the Supporting Information)
showed that the molar ratio of [C2mim][OAc] to [C2mim][NTf2]
in the ILs recovered from the top EtOAc phase was 0.47:0.53,
which made this a rather inefficient way to separate the two-
~
*
Figure 7. Solubility of ibuprofen ( : right axis) and diphenhydramine ( : left
axis) in [C2mim][OAc]x[NTf2](1Àx) (zero corresponds to [C2mim][NTf2] and 1.0 to
[C2mim][OAc]). Linear trend lines with boundaries at 0.67 (for ibuprofen) and
0.50 (for diphenhydramine) have been added for visual emphasis.
were then evaluated by UV/Vis spectroscopy based on prede-
termined calibration curves (Figure S5 in the Supporting Infor-
mation).
ion ILs from [C2mim][OAc]x[NTf2](1Àx).
3. Conclusions
The APIs exhibited similar solubility behavior to that of the
molecular solvents (EtOAc and H2O), with the hydrogen-bond-
donating API (ibuprofen) becoming more soluble as the
[OAc]À/[C2mim]+ molar ratio increased and the hydrogen-
bond-accepting API (diphenhydramine) becoming more solu-
ble as the [OAc]À/[C2mim]+ ratio decreased. Again, the solubili-
ty changes were not linear as a function of [OAc]À concentra-
tion. Breaks in the solubility changes were not as pronounced
The physical properties of multi-ion fluids (e.g. IL mixtures,
high ionicity ILs, DSILs, etc.) have been the focus of most
recent studies of such systems, whereas their chemical proper-
ties, which are important in the applications of these systems,
have received much less attention. Herein, DSILs containing
three ions, one common [C2mim]+ cation and two anions
([OAc]À and [NTf2]À), with different properties, particularly hy-
ChemPhysChem 0000, 00, 0 – 0
www.chemphyschem.org
7
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Articles
after drying. Water content measurements were made once per
sample and were found to be in reasonable agreement with those
of [C2mim][NTf2] and [C2mim][OAc] used in analytical studies.[45]
drogen-bond basicity, were used as an example to investigate
the specific interactions between the ions and the solvent
properties accompanying these interactions. Spectroscopic
data revealed that the interionic interactions varied based on
the abundance of charge-localized [OAc]À anions and
[C2mim]+ cations. Spectroscopy also gave evidence for single
acetate ions forming strong interactions with multiple imidazo-
lium cations, a situation that arose from the presence of both
Coulombic interactions and hydrogen bonding, and which
could not be achieved in the 1:1 salt. Unique interactions be-
tween the ions led to nonlinear, tunable solubilities of ethyl
acetate, water, and two pharmaceuticals as a function of ionic
composition. Thus, the chemical properties of DSILs could be
finely tuned by changing the ion composition in the system.
This study suggested that unique and tunable solvent prop-
erty sets were available that were derived and controlled by
the properties and concentrations of the ions present. Not
only were the properties of the DSIL tunable, but evidence
was found for nanoscale structuring that would have been im-
possible in the parent ILs. Although further studies are neces-
sary to fully understand ion–ion specific interactions and favor-
able solvation behavior, this work shows that the ion interac-
tions with solutes in specific DSILs can be controlled by careful
choice of the chemical nature and abundance of each ion. The
fine- and coarse-scale tunability of the chemical properties of
multi-ion liquids thus offer opportunities for expanding the
range of IL solvents, particularly in the area of separations, and
an appreciation of the differences between DSILs and their
parent salts may yet lead to the development of DSILs with
truly divergent properties.
[C2mim][OAc]x[NTf2](1Àx) (in which x is the [OAc]À/[C2mim]+ molar
ratio) were prepared as about 6 g samples through mass addition
of the corresponding amount of each IL with x=0.10, 0.20, 0.33,
0.50, 0.67, 0.80, and 0.90. Each system was stirred thoroughly for
1 h, then dried under high vacuum at 608C for 24 h, and stored
under argon. All samples were found to be homogeneous.
Synthesis of Ibuprofen and Diphenhydramine
Sodium ibuprofen (15 mmol) was dissolved in DI water (15 mL),
then a 2m solution of HCl (7.5 mL, 15 mmol HCl) was added drop-
wise to the solution. The mixture was stirred at room temperature
for 2 h. The precipitated solid ibuprofen was isolated by filtration,
washed with DI water twice, and dried in an oven (Precision Econ-
otherm Laboratory Oven, Natick, MA) at 658C for 48 h. 1H NMR
(500 MHz, CDCl3): d=7.25 (d, 2H), 7.13 (d, 2H), 3.73 (q, 1H), 2.46
(d, 2H), 1.87 (m, 1H), 1.52 (d, 3H), 0.91 ppm (d, 6H).
Diphenhydramine hydrochloride ([DPH][Cl], 20 mmol) was dis-
solved in DI water (20 mL) and sodium hydroxide (20 mmol) was
dissolved in DI water (10 mL). The solution of NaOH was added
dropwise to the solution of diphenhydramine hydrochloride. The
mixture was stirred at room temperature for 2 h. Water in the solu-
tion was then evaporated by using a rotary evaporator, and aceto-
nitrile (20 mL) was added to precipitate NaCl and any starting ma-
terials that might have remained. Acetonitrile was then removed
by using a rotary evaporator, and the solution was washed with
more acetonitrile (20 mL), following by removal of the organic sol-
vent. The product was a light yellow liquid that was dried under
1
high vacuum at 508C for 24 h. H NMR (500 MHz, CDCl3 external):
d=7.16 (d, 4H), 6.96 (t, 4H), 6.85 (t, 2H), 5.11 (s, 1H), 3.31 (t, 2H),
2.31 (t, 2H), 1.92 ppm (s, 6H).
Experimental Section
Chemicals
Solubilities of EtOAc and H2O
The ILs [C2mim][NTf2] and [C2mim][OAc] were purchased from Ionic
Liquids Technologies Inc. (Tuscaloosa, AL). CDCl3 was purchased
from Cambridge Isotope Laboratories, Inc. (Andover, MA). Sodium
ibuprofen, diphenhydramine hydrochloride, and TFA (with a purity
of 99%) were obtained from Sigma–Aldrich (Milwaukee, WI). Con-
centrated hydrochloric acid (36–38%) and sodium hydroxide were
supplied by VWR International, LLC (San Dimas, CA). Deionized (DI)
water was obtained from a commercial deionizer (Culligan, North-
brook, IL) with a specific resistivity of 16.82 MWcm at 258C. All
other solvents and reagents, such as EtOAc and ethanol, were ob-
tained from Sigma–Aldrich (St. Louis, MO) and used as received.
The solubilities of EtOAc and H2O in [C2mim][OAc]x[NTf2](1Àx) were
determined by adding EtOAc or H2O dropwise to each sample
(1.0 g) until the solution just became turbid. These saturated solu-
tions were analyzed by 1H NMR spectroscopy, and the solute/
[C2mim]+ molar ratios were calculated through direct integration
of appropriate signals.[40] Solubilities of the solutes in each DSIL
were measured twice, and results were reported as the average
values with error bars.
Solubility of Ibuprofen
A calibration curve for ibuprofen was obtained by analyzing five
different solutions of known concentrations of ibuprofen. For that,
ibuprofen was weighed, transferred to a 25 mL volumetric flask,
dissolved in ethanol, and diluted to the correct volume with etha-
nol to obtain stock solutions. Then various dilutions were made by
the addition of fresh ethanol. Selected dilutions, with concentra-
tions from 0.1ꢃ10À3 to 3.0ꢃ10À3 molLÀ1, were scanned by UV/Vis
spectroscopy and the absorbance at l=263 nm was selected for
analysis. The data for absorbance versus ibuprofen concentration
were treated by linear least-squares regression to obtain a coeffi-
cient of correlation (R2) of 0.99918, with the resultant equation
Abs=0.00355+0.30603[C] (Figure S5, left, in the Supporting Infor-
mation).
Preparation of [C2mim][OAc]x[NTf2](1Àx)
Before mixing, the ILs were individually dried to minimize the
water content. Water was easily removed from [C2mim][NTf2] by
placing it under high vacuum for 48 h at 608C with magnetic stir-
ring. To dry [C2mim][OAc], additional measures were needed, and
this IL was dried through a series of toluene-based azeotropic dis-
tillations under high vacuum at 608C with magnetic stirring in
a modified Schlenk flask. Any trace solvent was finally removed by
placing [C2mim][OAc] under high vacuum for 48 h. The water con-
tent was measured by the Karl-Fischer method, and found to be
1604.1 ppm for [C2mim][OAc] and 651.2 ppm for [C2mim][NTf2]
&
ChemPhysChem 0000, 00, 0 – 0
www.chemphyschem.org
8
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Articles
The solubility of ibuprofen was tested by loading ibuprofen
(0.05 g) into a vial loaded with [C2mim][OAc]x[NTf2](1Àx) (0.5 g) and
stirring. If all added ibuprofen was dissolved, an additional amount
was added until no more could dissolve and the solvent was satu-
rated. The solution was then separated from the particulate matter
by using a polytetrafluoroethylene (PTFE) syringe filter with pore
size of 0.45 mm, and the concentration of ibuprofen was analyzed
by UV/Vis spectroscopy [Cary 3C UV/Vis spectrophotometer (Varian
Instruments, Palo Alto, CA) by using fused quartz cuvettes with
a 1 cm path length] based on the absorbance at l=263 nm and
the predetermined calibration curve.
19F NMR spectroscopy by using TFA as an internal standard com-
pound. A stock solution of TFA in [C2mim][OAc], with a concentra-
tion of 0.66ꢃ10À3 molmolÀ1 [C2mim][OAc], was prepared, and
equal volumes of the solution of TFA/[C2mim][OAc] and recovered
[C2mim][OAc] were mixed, and then used for 19F NMR spectroscopy
analysis to determine the [NTf2]À concentration.
A similar procedure was used to attempt the extraction of
[C2mim][NTf2] from [C2mim][OAc]0.50[NTf2]0.50 with EtOAc. Briefly,
[C2mim][OAc]0.50[NTf2]0.50 and EtOAc, with molar ratio of 1:10 (deter-
mined by the solubility data), were mixed together and stirred at
room temperature for 2 h. The resulting mixture was then allowed
to stand for 1 h to equilibrate, after which a biphasic system was
observed and the two phases were separated using a pipette. The
lower phase was further extracted with EtOAc twice more using
the same procedure. The top EtOAc layers were combined and the
solvent was evaporated using a rotary evaporator, followed by
drying under high vacuum at 608C for 8 h. The bottom layer was
also dried under high vacuum at 608C for 8 h.
To prepare samples for UV analysis, each of the saturated solutions
(0.01–0.02 g) were loaded into a 25 mL volumetric flask and diluted
with ethanol to the correct volume. To eliminate the effect of the
presence of ions on the absorbance, UV/Vis spectra of solutions of
[C2mim][OAc]x[NTf2](1Àx) in ethanol were recorded for baseline cor-
rection. An amount of [C2mim][OAc]x[NTf2](1Àx) corresponding to
the amount present in the saturated solutions of ibuprofen was
dissolved in ethanol to a total volume of 25 mL, and the baseline
absorbance at l=263 nm was subtracted from the absorbance of
the unknown.
Characterization
Density and viscosity: Density measurements of the DSILs were
taken by using an Anton-Paar DMA 500 density meter (Ashland,
Solubility of Diphenhydramine
Virginia, USA) at 308C (Æ0.18C) with
a
repeatability of
0.0002 gcmÀ3. The viscosity of the DSILs was measured by using
a Cambridge Viscosity viscometer, VISCOlab 3000 (Medford, MA).
Approximately 1 mL of sample was placed in the sample chamber.
The correct-sized piston corresponding to the expected viscosity
range was added and the measurement was taken at 408C.
A calibration curve for diphenhydramine was obtained by analyz-
ing five different solutions of known concentrations of diphenhydr-
amine. Diphenhydramine was weighed in a 25 mL volumetric flask,
dissolved in ethanol, and diluted to the correct volume with etha-
nol to obtain a stock solution. Then various dilutions were made
by the addition of fresh ethanol. Selected dilutions, with concentra-
tions from 0.1ꢃ10À3 to 2.5ꢃ10À3 molLÀ1, were scanned by UV/Vis
spectroscopy and the absorbance at l=258 nm was selected for
analysis. The data for absorbance versus diphenhydramine concen-
tration were treated by linear least-squares regression to obtain
a coefficient of correlation (R2) of 0.99981, with the resultant equa-
tion Abs=0.00151+0.43233[C] (Figure S5, right, in the Supporting
Information).
NMR spectroscopy: NMR spectra were taken by utilizing a Bruker
Avance NMR spectrometer (Karlsruhe, Germany) at 500 MHz for
1H NMR spectroscopy and 125 MHz for 13C NMR spectroscopy. Each
sample, except for ibuprofen, was loaded solventless in a flame-
sealed capillary, and the spectra were collected at 258C by using
CDCl3 as the external lock. Ibuprofen, a solid at 258C, was dissolved
in CDCl3 for analysis. The 19F NMR spectrum of recovered
[C2mim][OAc] was obtained on a Bruker Avance 360 MHz NMR
spectrometer (Karlsruhe, Germany) by using TFA as an internal
standard and CDCl3 as the external lock.
The solubility of diphenhydramine was tested by adding
neat diphenhydramine dropwise into
a
vial loaded with
[C2mim][OAc]x[NTf2](1Àx) (0.5 g) until the solution became turbid.
The concentration of diphenhydramine in the saturated solution
was analyzed by UV/Vis spectroscopy based on the absorbance at
l=258 nm and the predetermined calibration curve. The same
method as that described above for ibuprofen was used to elimi-
nate the influence of the presence of ions on the UV absorbance.
IR spectroscopy: IR spectroscopic measurements were taken on
neat samples by utilizing a Bruker Alpha ATR-FTIR spectrometer
(Billerica, MA), which allowed direct observation of the liquids.
Spectra were obtained in the range of nmax =400–4000 cmÀ1
.
UV/Vis spectroscopy: Solubilities of ibuprofen and diphenhydra-
mine were determined by dissolving certain amounts of solutions
of ibuprofen or diphenhydramine/DSIL in ethanol and analyzing
them with a Cary 3C UV/Vis spectrophotometer (Varian Instru-
ments, Palo Alto, CA).
Separation of [C2mim][OAc] and [C2mim][NTf2] from
[C2mim][OAc]x[NTf2](1Àx)
H2O (2.0 g) was added to [C2mim][OAc]0.50[NTf2]0.50 (2.0 g).
The mixture was stirred at room temperature for 2 h, and
then was allowed to stand for 1 h to equilibrate. A biphasic
system was formed, and the top and bottom layers were
separated by using a pipette. To extract all [C2mim][OAc] from
[C2mim][OAc]0.50[NTf2]0.50, the bottom phase was further washed
with water twice more, following the procedure described above.
The aqueous solutions were combined, and water was evaporated
by using a rotary evaporator followed by drying under high
vacuum at 608C for 8 h. The bottom layer was also dried under
high vacuum at 608C for 8 h. The ILs recovered from the top and
Acknowledgements
We thank the Novartis-Massachusetts Institute of Technology (MIT)
Center for Continuous Manufacturing (CCM) for financial support.
Keywords: concentration effects
spectroscopy · NMR spectroscopy · solvent effects
·
ionic liquids
·
IR
1
bottom phases were analyzed by H NMR spectroscopy. The con-
centration of [NTf2]À in recovered [C2mim][OAc] was determined by
[1] R. D. Rogers, K. R. Seddon, Science 2003, 302, 792–793.
[2] T. Welton, Chem. Rev. 1999, 99, 2071–2083.
ChemPhysChem 0000, 00, 0 – 0
www.chemphyschem.org
9
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Articles
[3] M. J. Earle, K. R. Seddon, Pure Appl. Chem. 2000, 72, 1391–1398.
[4] J. G. Huddleston, A. E. Visser, W. M. Reichert, H. D. Willauer, G. A. Broker,
R. D. Rogers, Green Chem. 2001, 3, 156–164.
[5] W. Zhao, F. Leroy, B. Heggen, S. Zahn, B. Kirchner, S. Balasubramanian, F.
Mꢄller-Plathe, J. Am. Chem. Soc. 2009, 131, 15825–15833.
[6] a) W. Freyland, Coulombic Fluids, Springer, Berlin, 2011, pp. 5–44; b) S.
Tsuzuki, H. Tokuda, K. Hayamizu, M. Watanabe, J. Phys. Chem. B 2005,
109, 16474–16481.
[24] a) Y. U. Paulechka, G. J. Kabo, A. V. Blokhin, A. S. Shaplov, E. I. Lozinskaya,
D. G. Golovanov, K. A. Lyssenko, A. A. Korlyukov, Y. S. Vygodskii, J. Phys.
Chem. B 2009, 113, 9538–9546; b) A. R. Choudhury, N. Winterton, A.
Steiner, A. I. Cooper, K. A. Johnson, CrystEngComm 2006, 8, 742–745.
[25] J. K. Shah, J. F. Brennecke, E. J. Maginn, Green Chem. 2002, 4, 112–118.
[26] A. Chandran, K. Prakash, S. Senapati, Chem. Phys. 2010, 374, 46–54.
[27] C. M. Forsyth, D. R. MacFarlane, J. J. Golding, J. Huang, J. Sun, M. For-
syth, Chem. Mater. 2002, 14, 2103–2108.
[7] a) A. M. Fernandes, M. A. A. Rocha, M. G. Freire, I. M. Marrucho, J. A. P.
[28] K. Fumino, K. Wittler, R. Ludwig, J. Phys. Chem. B. 2012, 116, 9507–9511.
[29] a) A. D. Headley, N. M. Jackson, J. Phys. Org. Chem. 2002, 15, 52–55;
b) P. Bonhꢅte, A. P. Dias, N. Papageorgiou, K. Kalyanasundaram, M. Grꢆt-
zel, Inorg. Chem. 1996, 35, 1168–1178.
[30] B. Qiao, C. Krekeler, R. Berger, L. Delle Site, C. Holm, J. Phys. Chem. B
2008, 112, 1743–1751.
Coutinho, L. M. N. B. F. Santos, J. Phys. Chem.
B 2011, 115, 4033–
4041b) C. Roth, T. Peppel, K. Fumino, M. Kçckerling, R. Ludwig, Angew.
Chem. Int. Ed. 2010, 10221–10224; Angew. Chem. 2010, 122, 10419–
10423; c) L. Crowhurst, N. L. Lancaster, J. M. P. Arlandis, T. Welton, J. Am.
Chem. Soc. 2004, 126, 11549–11555.
[8] A. Triolo, O. Russina, H. J. Bleif, E. Di Cola, J. Phys. Chem. B 2007, 111,
4641–4644.
[9] G. Gonfa, M. A. Bustam, Z. Man, M. I. Abdul Mutalib, Asian Trans. Eng.
2011, 1, 24–34.
[10] E. W. Castner, Jr., J. F. Wishart, J. Chem. Phys. 2010, 132, 120901.
[11] T. L. Greaves, C. J. Drummond, Chem. Soc. Rev. 2013, 42, 1096–1120.
[12] H. Niedermeyer, J. P. Hallett, I. J. Villar-Garcia, P. A. Hunt, T. Welton,
Chem. Soc. Rev. 2012, 41, 7780–7802.
[31] M. Deetlefs, C. Hardacre, M. Nieuwenhuyzen, A. A. H. Padua, O. Shep-
pard, A. K. Soper, J. Phys. Chem. B 2006, 110, 12055–12061.
[32] D. T. Bowron, C. D’Agostino, L. F. Gladden, C. Hardacre, J. D. Holbrey,
M. C. Lagunas, J. McGregor, M. D. Mantle, C. L. Mullan, T. G. A. Youngs, J.
Phys. Chem. B. 2010, 114, 7760–7768.
[33] A. Stark, M. Sellin, B. Ondruschka, K. Massonne, Sci. China Chem. 2012,
55, 1663–1670.
[34] R. L. Frost, J. T. Kloprogge, J. Mol. Struct. 2000, 526, 131–141.
[35] F. Quilꢇs, A. Burneau, Vib. Spectrosc. 1998, 16, 105–117.
[36] K. Noack, P. S. Schulz, N. Paape, J. Kiefer, P. Wasserscheid, A. Leipertz,
Phys. Chem. Chem. Phys. 2010, 12, 14153–14161.
[13] A. Stark, J. Mol. Liq. 2014, 192, 144–152.
[14] A. Stark, M. Brehm, M. Brꢄssel, S. B. C. Lehmann, A. S. Pensado, M.
Schçppke, B. Kirchner, Top. Curr. Chem. 2014, 351, 149–188.
[15] G. Chatel, J. F. B. Pereira, V. Debbeti, H. Wang, R. D. Rogers, Green Chem.
2014, 16, 2051–2083.
[37] O. Hollꢈczki, D. Gerhard, K. Massone, L. Szarvas, B. Nꢉmeth, T. Veszprꢉ-
mi, L. Nyulꢊszi, New J. Chem. 2010, 34, 3004–3009.
[16] M. A. Ab Rani, A. Brant, L. Crowhurst, A. Dolan, M. Lui, N. H. Hassan, J. P.
Hallett, P. A. Hunt, H. Niedermeyer, J. M. Perez-Arlandis, M. Schrems, T.
Welton, R. Wilding, Phys. Chem. Chem. Phys. 2011, 13, 16831–16840.
[17] H. Wang, G. Gurau, S. P. Kelley, A. S. Myerson, R. D. Rogers, RSC Adv.
2013, 3, 10019–10026.
[18] a) R. P. Swatloski, S. K. Spear, J. D. Holbrey, R. D. Rogers, J. Am. Chem.
Soc. 2002, 124, 4974–4975; b) Y. Qin, X. Lu, N. Sun, R. D. Rogers, Green
Chem. 2010, 12, 968–971.
[19] J. Jacquemin, P. Husson, A. A. H. Padua, V. Majer, Green Chem. 2006, 8,
172–180.
[20] S. Chen, R. Vijayaraghavan, D. R. MacFarlane, E. I. Izgorodina, J. Phys.
Chem. B. 2013, 117, 3186–3197.
[38] M. Herstedt, M. Smirnov, P. Johansson, M. Chami, J. Grondin, L. Servant,
J. C. Lassꢇgues, J. Raman Spectrosc. 2005, 36, 762–770.
[39] P. C. Sadek, The HPLC Solvent Guide, 2nd Ed., Wiley, Hoboken, 2002.
[40] M. Lin, M. Tesconi, M. Tischler, Int. J. Pharm. 2009, 369, 47–52.
[41] M. Reta, R. Cattana, J. J. Silber, J. Solution Chem. 2001, 30, 237–252.
[42] L. Cammarata, S. G. Kazarian, P. A. Salter, T. Welton, Phys. Chem. Chem.
Phys. 2001, 3, 5192–5200.
[43] M. Brehm, H. Weber, A. S. Pensado, A. Stark, B. Kirchner, Phys. Chem.
Chem. Phys. 2012, 14, 5030–5044.
[44] S. Garcꢂa, J. Garcꢂa, M. Larriba, A. Casas, F. Rodrꢂguez Fluid Phase Equilib.
2013, 337, 47–52.
[45] a) C. P. Fredlake, J. M. Crosthwaite, D. G. Hert, S. N. V. K. Aki, J. F. Bren-
necke, J. Chem. Eng. Data 2004, 49, 954–964; b) S. Fendt, S. Padmanab-
han, H. W. Blanch, J. M. Prausnitz, J. Chem. Eng. Data 2011, 56, 31–34.
[21] T. Cremer, C. Kolbeck, K. R. J. Lovelock, N. Paape, R. Wçlfel, P. S. Schulz,
P. Wasserscheid, H. Weber, J. Thar, B. Kirchner, F. Maier, H. P. Steinrꢄck,
Chem. Eur. J. 2010, 16, 9018–9033.
[22] G. Gurau, H. Rodrꢂguez, S. P. Kelley, P. Janiczek, R. S. Kalb, R. D. Rogers,
Angew. Chem. Int. Ed. 2011, 50, 12024–12026; Angew. Chem. 2011, 123,
12230–12232.
[23] W. M. Reichert, J. D. Holbrey, R. P. Swatloski, K. E. Gutowski, A. E. Visser,
M. Nieuwenhuyzen, K. R. Seddon, R. D. Rogers, Cryst. Growth Des. 2007,
7, 1106–1114.
Received: December 18, 2014
Published online on && &&, 2015
&
ChemPhysChem 0000, 00, 0 – 0
www.chemphyschem.org
10
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

ARTICLES
Testing the waters (or salts): Ionic envi-
ronments, and thus chemical properties,
of multi-ion fluids can be tuned by
changing the ion types and concentra-
tions. NMR and IR spectroscopy are
used to determine the electrostatic in-
teractions in a liquid comprised of
a single cation, 1-ethyl-3-methylimidazo-
lium ([C2mim]+), and two anions with
different properties, acetate ([OAc]À)
and bis(trifluoromethylsulfonyl)imide
([NTf2]-) (see figure).
H. Wang, S. P. Kelley, J. W. Brantley III,
G. Chatel, J. Shamshina, J. F. B. Pereira,
V. Debbeti, A. S. Myerson, R. D. Rogers*
&& – &&
Ionic Fluids Containing Both Strongly
and Weakly Interacting Ions of the
Same Charge Have Unique Ionic and
Chemical Environments as a Function
of Ion Concentration
ChemPhysChem 0000, 00, 0 – 0
www.chemphyschem.org
11
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ