Size matters! on the way to ionic liquid systems without ion pairing

Article abstract of DOI:10.1002/chem.201400168

Several, partly new, ionic liquids (ILs) containing imidazolium and ammonium cations as well as the medium-sized [NTf₂]^- (0.230 nm³; Tf=CF₃SO₃^-) and the large [Al(hfip)₄]^- (0.581 nm³; hfip=OC(H)(CF₃)₂) anions were synthesized and characterized. Their temperature-dependent viscosities and conductivities between 25 and 80°C showed typical Vogel-Fulcher-Tammann (VFT) behavior. Ion-specific self-diffusion constants were measured at room temperature by pulsed-gradient stimulated-echo (PGSTE) NMR experiments. In general, self-diffusion constants of both cations and anions in [Al(hfip) ₄]^--based ILs were higher than in [NTf₂] ^--based ILs. Ionicities were calculated from self-diffusion constants and measured bulk conductivities, and showed that [Al(hfip)₄] ^--based ILs yield higher ionicities than their [NTf₂] ^- analogues, the former of which reach values of virtually 100% in some cases.From these observations it was concluded that [Al(hfip) ₄]^--based ILs come close to systems without any interactions, and this hypothesis is underlined with a Hirshfeld analysis. Additionally, a robust, modified Marcus theory quantitatively accounted for the differences between the two anions and yielded a minimum of the activation energy for ion movement at an anion diameter of slightly greater than 1 nm, which fits almost perfectly the size of [Al(hfip)₄]^-. Shallow Coulomb potential wells are responsible for the high mobility of ILs with such anions. Looking for ideally non-associated ionic liquids (ILs)? By performing extensive NMR diffusion, viscosity, and conductivity measurements as well as applying a modified Marcus theory to a large set of ionic liquids (ILs), it was shown that the ideal noninteracting IL anion should have an optimum diameter of 1 nm and a fluorinated surface. Several of the here described [Al(hfip)₄]^- (hfip=OC(H)(CF₃)₂) ILs fulfill these criteria. Shallow Coulomb potential wells (see figure) are responsible for the high mobility of ILs with such anions.

Full text of DOI:10.1002/chem.201400168

DOI: 10.1002/chem.201400168
Full Paper
&
Ionic Liquids
Size Matters! On the Way to Ionic Liquid Systems without Ion
Pairing
Alexander Rupp,[a, b] Nataliya Roznyatovskaya,[c] Harald Scherer,[a] Witali Beichel,[a, b]
Petra Klose,[a] Carola Sturm,[a] Anke Hoffmann,[a] Jens Tꢀbke,[c] Thorsten Koslowski,[d] and
Ingo Krossing*[a, b]
Abstract: Several, partly new, ionic liquids (ILs) containing
imidazolium and ammonium cations as well as the medium-
sized [NTf2]ꢀ (0.230 nm3; Tf=CF3SO3ꢀ) and the large [Al-
(hfip)4]ꢀ (0.581 nm3; hfip=OC(H)(CF3)2) anions were synthe-
sized and characterized. Their temperature-dependent vis-
cosities and conductivities between 25 and 808C showed
typical Vogel–Fulcher–Tammann (VFT) behavior. Ion-specific
self-diffusion constants were measured at room temperature
by pulsed-gradient stimulated-echo (PGSTE) NMR experi-
ments. In general, self-diffusion constants of both cations
and anions in [Al(hfip)4]ꢀ-based ILs were higher than in
[NTf2]ꢀ-based ILs. Ionicities were calculated from self-diffu-
sion constants and measured bulk conductivities, and
showed that [Al(hfip)4]ꢀ-based ILs yield higher ionicities than
their [NTf2]ꢀ analogues, the former of which reach values of
virtually 100% in some cases.From these observations it was
concluded that [Al(hfip)4]ꢀ-based ILs come close to systems
without any interactions, and this hypothesis is underlined
with a Hirshfeld analysis. Additionally, a robust, modified
Marcus theory quantitatively accounted for the differences
between the two anions and yielded a minimum of the acti-
vation energy for ion movement at an anion diameter of
slightly greater than 1 nm, which fits almost perfectly the
size of [Al(hfip)4]ꢀ. Shallow Coulomb potential wells are re-
sponsible for the high mobility of ILs with such anions.
Introduction
ing for various applications, for example, as electrolytes, as
polar solvents and reaction media, and for fundamental re-
search.[3–17] For example, an ideal IL electrolyte for (electro-
chemical) energy-conversion and energy-storage devices
would show low viscosity, high conductivity and self-diffusion
constants, a very low melting point, and a large electrochemi-
cal stability window. ILs with both anions of this study fulfill
these requirements.
Ionic liquids (ILs) are classically defined as salts that melt
below 1008C. Room-temperature ILs (RTILs) accordingly melt
below 258C. Selected ILs may have characteristics that are not
found in this combination elsewhere, namely, relatively high
conductivities, low viscosities, negligible vapor pressures, low
flammability, high electrochemical resistance, good gas-storage
capability, and often low toxicity.[1,2] The potential of combin-
ing several of those properties in one IL makes them interest-
The quantities of these dynamic or transport properties of
ILs are interconnected by several relations. First, self-diffusion
constants D and viscosities h are related by the Stokes–Einstein
equation[18] [Eq. (1)],
[a] Dipl.-Chem. A. Rupp, Dr. H. Scherer, Dr. W. Beichel, P. Klose, C. Sturm,
Dr. A. Hoffmann, Prof. Dr. I. Krossing
Albert-Ludwigs-Universitꢀt Freiburg
kT
cphr
Institut fꢁr Anorganische und Analytische Chemie
Albertstrasse 21, 79104 Freiburg (Germany)
E-mail: ingo.krossing@ac.uni-freiburg.de
ð1Þ
D ¼
[b] Dipl.-Chem. A. Rupp, Dr. W. Beichel, Prof. Dr. I. Krossing
Freiburger Materialforschungszentrum
Stefan-Meier-Strasse 21
where k is the Boltzmann constant, T the absolute tempera-
ture, c a constant between 4 and 6, and r the radius.
It is still under debate whether the Stokes–Einstein equation
can be applied to IL systems, but due to its simplicity it is
often used to describe ILs.[19,20] Second, a relation between
conductivities s and self-diffusion constants is given by the
Nernst–Einstein equation [Eq. (2)],
79104 Freiburg (Germany)
[c] Dr. N. Roznyatovskaya, Dr. J. Tꢁbke
Fraunhofer-Institut fꢁr Chemische Technologie
Angewandte Elektrochemie
Joseph-von-Fraunhofer-Strasse 7, 76327 Pfinztal (Germany)
[d] Prof. Dr. T. Koslowski
Albert-Ludwigs-Universitꢀt Freiburg
Institut fꢁr Physikalische Chemie
Albertstrasse 23a, 79104 Freiburg (Germany)
z2e21NA
MkT
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/chem.201400168.
ð2Þ
s ¼
9794
ꢁ D
Chem. Eur. J. 2014, 20, 9794 – 9804
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
where z is the charge, e the elementary charge, 1 the density,
NA the Avogadro constant, and M the molecular mass.
weakly coordinating anion than the medium-sized [NTf2]ꢀ
ꢀ
anion (Tf=CF3SO3 ), shifts the IL systems towards an increas-
Third, viscosities h and molar conductivities l are connected
by Walden’s rule[21,22] [Eq. (3)], which results from Equations (1)
and (2) and is commonly accepted to be valid also for ILs.[5]
ingly ideal behavior that is governed by ion size (i.e., molecular
volumes) rather than by ion surface (i.e., intermolecular inter-
actions). These experimental findings were supported by
a modified Marcus-theory ansatz.
lh ¼ const:
ð3Þ
One drawback of all of these often-used models is that
some of their initial assumptions are not valid in the case of
pure ILs: Walden’s rule was developed for dilute solutions, the
Stokes–Einstein equation was derived for a single particle dif-
fusing in a homogeneous solvent, and self-diffusion constants
in the Nernst–Einstein equation are those at infinite dilution.
Admittedly, in most cases the adequate results that one ob-
tains when using these relations are a sufficiently good argu-
ment for their application.
Results and Discussion
Syntheses and characterization
A set of 24 ILs (full list in Supporting Information, abbreviations
in Appendix) was prepared or, where possible, purchased. The
syntheses were carried out by anion metathesis [Eq. (8)] ac-
cording to reported procedures.[4,28,29]
Li½WCAꢂ þ ½CatꢂX solvent ½Catꢂ½WCAꢂ þ LiX
ƒƒƒ!
For the quantitative description of temperature-dependent
viscosities of ILs, it has been shown by many groups that data
fitting with a Vogel–Fulcher–Tammann (VFT) ansatz [Eq. (4)]
yields excellent results,[23–26]
ð8Þ
½WCAꢂ ¼ ½NTf2ꢂꢀ, ½AlðhfipÞ4ꢂꢀ; X ¼ Cl, Br
All ILs were fully characterized by IR, Raman, and NMR spec-
troscopy (1H, 19F and 27Al). Purity was additionally always
checked by 7Li NMR spectroscopy, to ensure the absence of
lithium halides. In the case of [Al(hfip)4]ꢀ-based ILs, due to the
known decomposition pathway of the hfip group,[3] complete
removal of H2O in vacuum always results in formation of im-
purities in the 19F NMR spectra. We had to find a balance be-
tween low H2O contents (<200 ppm) and 19F NMR spectra
showing anion impurities, probably resulting from the forma-
tion of the [FAl(hfip)3]ꢀ anion[3] in amounts of at most 3%,
which, in our experience, does not alter the properties of [Al-
(hfip)4]ꢀ ILs by more than 2%. The [NTf2]ꢀ-based ILs were dried
at 908C/10ꢀ3 mbar until water contents of <20 ppm were ach-
ieved (Karl–Fischer titration). Full characterization details are
deposited in the Supporting Information.
ꢀ
ꢁ
B
h ¼ h0 exp
ð4Þ
T ꢀ T0
where h0 is the viscosity at infinite temperature, B a constant,
and T0 the Vogel temperature.
In the case of conductivities of ILs, also an Arrhenius ap-
proach [Eq. (5)] is applicable, yet VFT fits [Eq. (6)] exhibit higher
accuracy,
ꢀ
ꢁ
A
RT
s ¼ s0 exp ꢀ
ð5Þ
ð6Þ
ꢀ
ꢁ
B
Viscosity and conductivity measurements
s ¼ s0 exp ꢀ
T ꢀ T0
The temperature-dependent viscosities and conductivities of
all 24 ILs were determined between room temperature and
808C during heating in their liquid range, and fitting of the
data with a VFT ansatz was done subsequently. The results are
shown in Figures 1 and 2, respectively, while the fit parameters
are given in Tables 5 and 6 in the Appendix. All fits show R2
values greater than or equal to 0.999, and both viscosities and
conductivities are in accordance with expected results. A more
detailed description of these results can be found in the Ap-
pendix.
where s0 is the conductivity at infinite temperature, A the acti-
vation energy, B a constant, and T0 the Vogel temperature.
By combining self-diffusion and conductivity measurements,
one can assess the so-called ionicities [Eq. (7)],[27] which give
quantitative information about the degree of ion pairing
within the IL,
sexp
sdiff
sexpMkT
e21NAðDþ þ DꢀÞ
I ¼
¼
ð7Þ
Figures 1 and 2 show that [Al(hfip)4]ꢀ-based ILs almost
always show viscosities equal to or lower than and conductivi-
ties equal to or higher than those of [NTf2]ꢀ-based ILs, the only
where sexp is the experimentally measured conductivity and
sdiff the conductivity calculated from the measured self-diffu-
sion constants.
+
exceptions being [C2MIm]+, [AllylMIm]+, [C4MIm]+, and [N1123
]
salts. Clearly, the nature of the cations also has an influence on
the properties: exchanging imidazolium ions against ammoni-
um ions results in an approximately tenfold increase in viscosi-
ty. Additionally, the properties of [Al(hfip)4]ꢀ-based ILs depend
much less on the cation compared to their [NTf2]ꢀ analogues.
Herein, we present viscosity, conductivity, self-diffusion, and
ionicity data for 24 ILs, seven of which were synthesized for
the first time. Our results show that introduction of the large
[Al(hfip)4]ꢀ anion (hfip=OC(H)(CF3)2), which is an even more
Chem. Eur. J. 2014, 20, 9794 – 9804
www.chemeurj.org
9795
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
with longer alkyl chains (n>6),
apparently due to the increasing
importance of dispersive interac-
tions in these more polarizable
groups.
Comparison of the different
sets of ILs (Figure 1) shows the
expected tendency of ILs with
larger cations to have higher vis-
cosities due to more hindered
migration. This becomes even
clearer when one considers that
the migration of ions in ILs can
be described by a hole theory.[30]
Despite the much larger molecu-
lar
volume
of
[Al(hfip)4]ꢀ
(0.581 nm3) compared to [NTf2]ꢀ
(0.233 nm3),[31,32] both [CnMIm]-
[NTf2] and [NR4][NTf2] ILs are typ-
ically equally or even more vis-
cous than their [Al(hfip)4]ꢀ ana-
logues. The only exceptions to
this behavior are ILs with the
Figure 1. Viscosity data of all 24 investigated ILs. Viscosities were measured in the liquid range of the ILs from 25
to 808C.
very small [C2MIm]+ (V+ =
0.156 nm3[31,32]) and [N1123
]
(V+
+
=0.180 nm3[31,32]) cations.
This unexpected behavior
gave first hints that ILs with
a large and very weakly coordi-
nating anion exhibit almost ideal
and undisturbed physical prop-
erties. The volume of [Al(hfip)4]ꢀ
compared to that of [NTf2]ꢀ sug-
gested that ion pairing should at
least be strongly subdued to
allow for the high conductivities
and low viscosities of [Al(hfip)4]ꢀ
ILs. To shed further light on this
idea, a complete set of diffusion
measurements was done by
NMR methods.
Figure 2. Conductivity data of all 24 investigated ILs. Conductivities were measured in the liquid range of the ILs
from room temperature to 808C.
Pulsed-gradient stimulated-
echo (PGSTE) NMR measure-
ments of self-diffusion con-
stants
This feature can be explained by the different properties of the
two anions: [Al(hfip)4]ꢀ is a large and spherical ion with an
almost completely fluorinated and thus hardly polarizable sur-
face. Interactions with other nearby ions will therefore be
weak, nondirectional, and independent of the structure of the
other ion. Metaphorically speaking, cations are moving
through a more or less noninteracting network of anions that
governs the overall properties. [NTf2]ꢀ ions, on the other hand,
are to a certain extent able to interact through their more
Lewis-basic N and O atoms. These interactions are dependent
on the cation structure. Aggregation is typically favored in ILs
Ion-specific self-diffusion constants of all RTILs within the set
of ILs were measured with PGSTE-NMR experiments at room
temperature. Cation diffusion constants D+ were extracted
1
from H spectra; anion diffusion constants Dꢀ were extracted
1
from both 19F and H spectra, unless the anion signal could not
be separated properly from all cation signals. Both nuclei of
each IL were measured at least twice. Fitting was done with
the Stejskal–Tanner equation[33] [Eq. (9)] by using the integrals
of the NMR signals. The values given in Table 1 are arithmetic
means of these measurements. The total error including tem-
Chem. Eur. J. 2014, 20, 9794 – 9804
www.chemeurj.org
9796
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
supercooled for several measurements. Even more striking is
that the anion self-diffusion constants of [Al(hfip)4]ꢀ usually
exceed those of the corresponding [NTf2]ꢀ ILs. Typically, larger
particles move more slowly due to steric hindrance, increased
interaction with their surroundings, and their typically higher
molecular mass (cf. Mr([Al(hfip)4]ꢀ)=695.10, Mr([NTf2]ꢀ)=
280.15). Figure 3c and d reveal another important feature: In
these plots, the ratio of cation and anion self-diffusion con-
stants of [NTf2]ꢀ and [Al(hfip)4]ꢀ ILs, respectively, were plotted
versus the molecular volume of the cations, and a fit was done
by utilizing a generalized form of the Stokes–Einstein equation
[Eq. (1)]. The justification for this model is as follows: The con-
stant c in Equation (1) originally represented the conditions of
the diffusion process (sticky or slip conditions). From our meas-
urements, neither the value of c can be derived nor can it be
said whether the c values are the same for cations and anions
of one IL. To avoid this problem, we evaluated the ratio of cat-
ionic and anionic diffusion constants, yielding the first part of
Equation (10). In most ILs, the ions are far from spherical.
Therefore, the radius is not very suitable for this study and is
replaced by the third root of the molecular volume Vm. Also,
c+/ꢀ values are combined to yield a new parameter A*. This re-
sults in the right part of Equation (10).
Table 1. Ion-specific diffusion constants of the measured RTILs. Molecular
volumes were obtained through DFT calculations; the calculation proto-
col can be found elsewhere.[31] Diffusion constants D+/ꢀ were obtained
by signal integration.
Cation
Vmþ [nm3] D+ [10ꢀ11 m2 sꢀ1
[NTf2]ꢀ [Al(hfip)4]ꢀ
]
Dꢀ [10ꢀ11 m2 sꢀ1
]
[NTf2]ꢀ
[Al(hfip)4]ꢀ
[C2MIm]+
0.156
4.31
4.46
2.37
1.33
(1.94)[a]
4.77
2.64
2.90
1.70
0.97
1.23
1.43
0.792
(0.797)[a]
1.89
[AllylMIm]+ 0.168
[C4MIm]+
0.197
2.20[a]
2.78
1.30[a]
1.69
[C4MMIm]+ 0.218
[C6MIm]+
[C8MIm]+
0.244
0.288
0.331
0.180
0.322
0.604
0.268
0.484
1.38
2.80
2.01
1.60
1.80
1.46
1.29
not an RTIL
0.966
0.895
0.695
1.35
0.343
0.105
0.520
0.149
[C10MIm]+
+
[N1123
]
not an RTIL 1.11
1.21
0.633
2.10
0.725
+
+
+
+
[N1444
[N1888
[N2225
[N2666
]
]
]
]
0.153
too viscous 0.747
0.512
0.181
1.50
0.757
[a] [C2MIm][Al(hfip)4] and [C4MIm][Al(hfip)4] are not RTILs either, yet they
could be supercooled and measured at room temperature.
perature fluctuations was estimated by repeated measure-
ments and comparison to literature data[34–41] to be 5–10%.
ꢂ
ꢀ
ꢁꢃ
Iðd; D; gÞ
d
¼ exp ꢀg2d2g2D D ꢀ
ð9Þ
ꢀ
ꢁ1
Dþ cꢀrꢀ
Dꢀ cþrþ
Vmꢀ
Vmþ
=
3
Iðd; D; g ¼ 0Þ
3
*
ð10Þ
¼
¼ A
In Equation (9), I(d,D,g) is the signal intensity with applied gra-
dient of strength g, I(d,D,g=0) the signal intensity without gra-
dient g, g the gyromagnetic ratio, d the duration of the gradi-
ent pulse, and D the diffusion delay.
From Table 1 it is obvious that
Since in Figure 3c and d, the respective anions remain con-
À
Á
1=3
stant, we combine A* and Vmꢀ
to get a new parameter A.
Additionally, since we cannot be sure whether the Stokes–Ein-
self-diffusion constants decrease
with increasing cation size. How-
ever, the decrease seems to be
more pronounced for the cat-
ions than for the anions, as the
difference between D+ and Dꢀ
also decreases. For example, in
[C2MIm][NTf2] the cation is nearly
twice as fast as the anion,
whereas in [C10MIm][NTf2], the
two ions diffuse almost equally
fast. Furthermore, ammonium-
based ILs tend to diffuse more
slowly than imidazolium ILs of
comparable ion size. Figure 3
shows the relation of cation size
and ion diffusion constants for
all sets of ILs.
In contradiction to common
expectations, almost all cations
diffuse faster when they are
Figure 3. Results of the PGSTE NMR measurements. a) Comparison of cation self-diffusion constants of [NTf2]ꢀ and
[Al(hfip)4]ꢀ ILs with respect to the cation volume. b) Comparison of anion self-diffusion constants of [NTf2]ꢀ and
[Al(hfip)4]ꢀ ILs with respect to the cation volume. c) Combined fit of the ratios of the diffusion constants of all in-
vestigated [NTf2]ꢀ ILs. The fitting function was a generalized Stokes–Einstein ansatz of the form y=axꢀb without
combined with the larger anion
[Al(hfip)4]ꢀ. Although [C2MIm][Al-
(hfip)4] and [C4MIm][Al(hfip)4] are
not really RTILs, they could be considering [N1444][NTf2] and [C8MIm][NTf2]. d) Corresponding fit of all investigated [Al(hfip)4]ꢀ ILs.
Chem. Eur. J. 2014, 20, 9794 – 9804 www.chemeurj.org ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
9797

Full Paper
stein equation is valid in our case, the exponent is not necessa-
rily 1/3, and we replace it by a second parameter b. Hence, our
final equation is Equation (11).
icities of a small number of [NTf2]ꢀ ILs are known to be 0.5–
0.8.[27] This means that even in the very best [NTf2]ꢀ IL, namely,
the low-viscosity, highly conducting [C2MIm][NTf2], 20% of the
ions are not available for charge transport, which is an annoy-
ing feature especially for electrochemical applications. Before
we discuss the ionicity data, we note that the Nernst–Einstein
equation, which is used to calculate conductivities from self-
diffusion constants, may still be valid even if the Stokes-Ein-
stein equation does not apply in the examined systems. Also,
its successful application in IL research shows that it yields em-
pirically good results,[25,42,43] whereas the Stokes–Einstein rela-
tion clearly failed here.
Dþ
Dꢀ
À
Áꢀb
¼ A Vmþ
ð11Þ
If Equation (1) were valid, one would expect the ratios to
show a behavior of the form y=Axꢀ1/3, even when different
c factors for cations and anions are considered. However, nei-
ther [NTf2]ꢀ nor [Al(hfip)4]ꢀ ILs show such behavior. Whereas
their pre-exponential factors are very similar, they differ in the
exponents, which are 0.63 and 0.91, respectively (see Figure 3c
and d). Note that in the case of [NTf2]ꢀ ILs, fitting was done
without considering the values of [N1444][NTf2] and [C8MIm]-
[NTf2]. The former—like [N1888][NTf2]—was hardly measurable
due to its high viscosity and the resulting noncorrectable
phase errors. The latter has an unusually low ratio, but it is not
clear where this effect stems from, and we omitted this value.
Experimental self-diffusion constants also allow for the calcu-
lation of viscosities through the Stokes–Einstein equation. The
results of these calculations are deposited in the Supporting
Information. However, no general trend could be observed,
except for a tendency that viscosities are underestimated by
these calculations in the case of [NTf2]ꢀ ILs and rather overesti-
mated in the case of [Al(hfip)4]ꢀ ILs. In summary, our analysis
of the PGSTE measurements, in accordance with literature
data,[19] emphasizes that hydrodynamic models may only be
used with much caution, since they might not—or at least not
always—apply for ILs.
Table 2 lists ionicity data of all measured ILs. The estimated
overall error of the ionicities is about 10%. The values show
that [Al(hfip)4]ꢀ ILs not only show higher conductivities and dif-
Table 2. Ionicities of all measured RTILs.
Cation
Anion
[NTf2]ꢀ
[Al(hfip)4]ꢀ
[C2MIm]+
[AllylMIm]+
[C4MIm]+
[C4MMIm]+
[C6MIm]+
[C8MIm]+
0.67 (0.75)[a]
0.62
[2.10]
0.87
1.03
0.83
0.73
0.82
0.79
no D+, Dꢀ
0.87
0.56 (0.61)[a]
0.66
0.63 (0.57)[a]
0.48 (0.54)[a]
0.55
0.82
0.50
no Dꢀ
[C10MIm]+
+
[N1123
]
+
+
+
+
[N1444
[N1888
[N2225
[N2666
]
]
]
]
0.47
0.69
0.80
0.95
0.74
[a] Data given in parentheses for comparison stem from the work of Wa-
tanabe et al.[27]
Ionicities
By dividing the experimentally measured conductivities sexp by
the conductivities calculated from the measured self-diffusion
constants sdiff, it is possible to calculate the so-called ionici-
ties.[27,39] These can be seen as a parameter to describe the
degree of ion pairing in ILs. The basic assumptions for this
theory are that, in PGSTE-NMR experiments, the detected diffu-
sion is an average of all species containing the investigated
ions in solution. Ion pairs may exist but are normally too short-
lived to give rise to individual signals in the NMR spectrum.
Therefore, conductivities calculated from those diffusion con-
stants should be an upper limit, presuming that all ions con-
tribute to conductivity (=ideal limiting conductivity). The real
ideal conductivity can, however, not be measured, because the
measured diffusion constants are the mean of the diffusion
constants of unpaired and (slower) paired ions. Thus, sdiff is
close to but does not equal the ideal conductivity. Neverthe-
less, it is a good and easily accessible reference value and
allows for a quantitative comparison of degrees of ion pairing.
During bulk conductivity measurements, only net charged par-
ticles can contribute to conductivity. Ion pairs decrease the
number of available charge carriers and thus limit the detected
conductivity to an often much lower value. As a result, the ab-
sence of ion pairing would yield ionicities of 1.0 or 100%, and
complete ion pairing results in ionicities of 0.0 or 0%. The ion-
fusion constants, but also particularly high ionicities, typically
10–40% higher than those of [NTf2]ꢀ ILs with the same cations.
To the best of our knowledge, this is the first time that ILs with
ionicities of up to 100% have been reported. However, an IL
system completely without any interaction is not possible. In
this light, ion pairs may also exist in the best [Al(hfip)4]ꢀ ILs,
but are clearly less favored than in [NTf2]ꢀ ILs. This is in agree-
ment with dielectric spectroscopy, which indicated the absence
of ion pairing in [Al(hfip)4]ꢀ ILs down to the very short picosec-
ond range.[19] The unusually high ionicity of [C2MIm][Al(hfip)4]
can be attributed to subdiffusion in the supercooled melt and
a resulting underestimation of sdiff
.
The effect of the cation on ionicities can also be seen. Dis-
persion forces, steric effects, strength of Coulomb forces, and
interaction with anions differ from cation to cation. As a result,
the effect of the cation apparently is not negligible and can
lead to extraordinarily high ([C4MIm][Al(hfip)4]) or just moder-
ate ionicities ([C4MIm][NTf2]). Going from [C4MIm][Al(hfip)4] to
ILs with only slightly larger imidazolium cations supports this
finding with approximately 20% lower ionicities.
Chem. Eur. J. 2014, 20, 9794 – 9804
www.chemeurj.org
9798
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Discussion of the unusually low viscosities, high conductivi-
ties, high self-diffusion constants, and high ionicities of [Al-
(hfip)4]ꢀ ILs
1) The Stokes–Einstein equation qualitatively describes the re-
sults correctly with ratios of the cation and anion diffusion
constants that decrease with increasing cation size.
2) It fails in the quantitative description of the results. Trying
to fit the data to an exponent of b=1/3 yields R2 values of
<0.7. This is not surprising, as the requirements for its ap-
plication are at least not entirely fulfilled: The “solvent”, for
example, does not consist of a homogeneous medium.
3) [Al(hfip)4]ꢀ ILs show a more ideal behavior with an expo-
nent closer to unity. [NTf2]ꢀ ILs, on the other hand, have
a dependence on the cation molecular volume that is more
governed by additional surface-based, likely dispersive in-
teractions. This becomes manifest in a b value of roughly
2/3.
The viscosity data show two things: imidazolium ILs are less
viscous than ammonium ILs, and [Al(hfip)4]ꢀ ILs are equally or
less viscous than [NTf2]ꢀ ILs. Flat imidazolium ions only limit
ion movement in two dimensions and thus yield ILs of lower
viscosity than spherical ammonium ions.[44,45] The lower viscosi-
ties of [Al(hfip)4]ꢀ ILs can be explained by weaker interactions
in these ILs because of the more weakly coordinating nature
of [Al(hfip)4]ꢀ [4] Additionally, hydrogen-bond networks can lead
.
to a flattening of the energy landscape and thereby increase
ion mobility and dynamic properties.[46–48] Similar trends and
conclusions apply for conductivities. The higher conductivities
of ILs with small cations can be attributed to the higher
charge-carrier densities and, of course, to the lower viscosities,
which ease ion transport. Altogether, the conductivity results
resemble the viscosity measurements: less interaction, com-
bined with a network of hydrogen bonds if possible, improves
transport properties of ILs.
Self-diffusion constants of [Al(hfip)4]ꢀ-based ILs generally are
higher than those of corresponding [NTf2]ꢀ ILs. The [Al(hfip)4]ꢀ
anion, which is covered with an almost completely fluorinated
and thus nonpolarizable surface and has a much larger surface
area, has weaker coordinating characteristics than [NTf2]ꢀ. Con-
sequently, all mentioned dynamic properties crucially rely on
the ability of particles to change their position as easily as pos-
sible. Every joule of attractive forces, resulting for instance
from Coulomb or dispersive interactions, limits viscosities, con-
ductivities, and self-diffusion constants. Especially Coulomb in-
teractions should be limited in [Al(hfip)4]ꢀ ILs because of the
reciprocal dependence of the Coulomb potential EC on the
radius r [Eq. (12)],
The ionicity data underline the hypothesis that reduced in-
teractions are responsible for the good performance of [Al-
(hfip)4]ꢀ ILs as possible electrolytes. Due to this reduction, not
only is the movement of ions facilitated, but also the formation
of ion pairs is inhibited. Consequently, more ions are available
to take part in dynamic processes. Our results imply that sys-
tems almost without ion pairing are possible and can be
a starting point for the synthesis of better systems for applica-
tions in which ILs with as many free ions as possible are
needed, for instance, in electrolyte research.
To support this hypothesis, one can examine qualitatively
the energy barriers that lie between two hole positions of
a moving ion inside the IL. Figure 4 shows the difference be-
q1q2
4pe0r
EC ¼
ð12Þ
Figure 4. Energy profiles of the ions diffusing in an IL. The abscissa shows
the normalized ion positions in multiples of the distance r from the lowest-
energy equilibrium position, in relation to the average distance d from one
equilibrium position to an adjacent equilibrium position.
where q1 and q2 are the electric charges of two interacting par-
ticles, e0 is the vacuum permittivity and r the distance between
the particles.
The fit of D+/Dꢀ versus Vmþ (Figure 3) can give even more in-
sight into the intermolecular forces. We propose that the expo-
nents b [cf. Eq. (11)] are related to the degree of interactions
within the ILs. If no other interactions are present, only the
size of the ions determines the three-dimensional Coulomb
potential of the IL that limits mass transport. Therefore,
a volume-dependent term, that is, an exponent of unity should
appear in this representation. If surface-dependent interactions
occur, the exponent should also reduce to a surface term, or
Vꢀ2/3 if the contributions of these interactions are dominant. In
the light of these assumptions, three main conclusions can be
drawn for the applicability of the Stokes-Einstein equation for
IL systems:
tween [Al(hfip)4]ꢀ and [NTf2]ꢀ ILs. Assuming that the total inter-
action energy ET in an IL is mainly determined by the additive
terms of Coulomb (EC) and dispersion energies (ED), ET =EC +ED,
the energy barrier for diffusion of cations in [Al(hfip)4]ꢀ ILs
should roughly be only 74% of that of the cations in [NTf2]ꢀ
ILs. This simplified value results from the differences in the mo-
lecular volumes of the IL anions and the assumption that the
majority of ET is governed by EC.
Of course, dispersion still can have an effect, but a smaller
one compared to the effect of Coulomb interactions. However,
on extending the alkyl-chain lengths of the imidazolium cat-
Chem. Eur. J. 2014, 20, 9794 – 9804
www.chemeurj.org
9799
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
ions, dispersion increases, which ultimately leads to an increase
in intermolecular forces and thereby decreased ionicities. This
is in complete agreement with all experimental data collected
in Figures 1 and 2 as well as Tables 1 and 2.
A modified Marcus ansatz to account for the ion dynamics
To shed even more theoretical light on our results, we now
turn to the description of ion-hopping processes in the context
of the statistical mechanics of simple models of ILs. They cap-
ture the interplay of free volume, through the hard-sphere
model, and Coulomb effects represented by point charges un-
derlying many physical and chemical properties of ILs. They
provide insight by simple analytical solutions or approxima-
tions remarkably close to computer simulations. We note that
the subtleties associated with the chemical details can either
be dealt with by modifying the ion shape and intraionic
charge distributions of the constituents, or by atomically re-
solved models, both of which require large-scale computer
simulations.
When an anion is moving from one minimum position to an
adjacent minimum position, it must pass through a transition
state in which its energy is raised. In the case of the large [Al-
(hfip)4]ꢀ anion, repulsive Coulomb interactions between anions
during the motion are much weaker than for the smaller
[NTf2]ꢀ anion. Therefore, a larger anion, here further supported
by a hardly polarizable surface that is formed mainly by ali-
phatic CꢀF bonds, enables easier ion movement in this case.
This finding contradicts the common consideration that larger
ions always lead to less favorable dynamics and transport
properties. The limitations of this model are given by the
nature of the anions. [Al(hfip)4]ꢀ not only shows weak interac-
tions but also has a spherical shape, which isotropically distrib-
utes these interactions. In contrast, [NTf2]ꢀ has a more cylindri-
cal shape and thus shows more directional and less distributed
interactions. These influences must be considered, otherwise
comparison with ionicities of small, but spherical [PF6]ꢀ ILs[49]
or even with large, but more coordinating anions with stron-
ger directed or dispersive interactions can lead to errors.
To support this hypothesis and to get an at least semiquanti-
tative insight into the differences between [NTf2]ꢀ and [Al-
(hfip)4]ꢀ, we exemplarily conducted a Hirshfeld analysis[50–52] of
the crystal structures of [C4MIm][Al(hfip)4] and [C4MIm][NTf2].
Table 3 lists the results for contacts between H and F and be-
tween H and O, respectively.
To transfer an ion in a polarizable dielectric medium,
a charged particle must hop from an occupied donor position
to a vacant acceptor region. Whereas the net thermodynamic
driving force for this process vanishes, the sheer existence of
localized charges requires the presence of an activation barrier.
The energetics of the ion-hopping process can be decom-
posed into two contributions: First, a cavity accommodating
the ion must be created in the acceptor region; second,
moving the charge from the donor to the acceptor site polariz-
es the latter and depolarizes the former. Within the so-called
restricted primitive model of charged hard spheres of equal di-
ameter d and charge ꢃZe, the free energy of creating a cavity
is given by the hard-sphere contribution DGHSF [Eq. (13)],
ꢀ
ꢁ
6VkBT 3x1x2 3x32 3x3
DGHSF ffi DFHSF
¼
þ þ
Dx3
2 ln D ꢀ x0 ln D
x32
p
D
Table 3. Hirshfeld analysis of [C4MIm][Al(hfip)4] and [C4MIm][NTf2].
ð13Þ
[C4MIm][Al(hfip)4]
H–F [%]
H–O [%]
cation
anion
80.6
50.0
5.3
2.9
where DFHSF is the change in Helmholtz free energy of the
hard-sphere fluid, D=1ꢀx3 is the packing fraction of the
system and the density dependent coefficients xi are given by
Equation (14),
[C4MIm][NTf2]
cation
anion
H–F [%]
25.5
30.4
H–O [%]
28.1
35.1
p
xi ¼ 1si
ð14Þ
6
Obviously, in the [Al(hfip)4]ꢀ salt, contacts are mostly created
between H and F, whereas in the [NTf2]ꢀ salt, also H–O con-
tacts occur to a great extent. Keeping in mind that F atoms in
CF3 groups are hardly polarizable and carry less negative
charge than O atoms in the S+ꢀOꢀ bonds of the [NTf2]ꢀ anion,
H–F contacts probably account for less interaction energy than
H–O contacts. Therefore, [C4MIm][NTf2] includes more stronger
interactions than [C4MIm][Al(hfip)4], which is dominated by
weaker H–F interactions. So we can suggest that ions in the
[NTf2]ꢀ salt tend to have a higher binding energy, and accord-
ingly its ionicity is well below that of the [Al(hfip)4]ꢀ salt.
It is clear that for a complete quantitative analysis, one
would have to weigh the contacts according to the interatom-
ic distance. For such an analysis, a reference point is necessary,
as well as a function describing the energy profile, which lies
far beyond the scope of this work.
where si is the hard sphere diameter . Our numerical results do
not depend strongly on the use of Equation (13); for example,
applying the simulation data of creating a cavity in water leads
to a very similar physical picture, as described below. We
assume a packing fraction of 0.6 to be typical for liquid ILs.
The packing index of IL crystals is around 0.65 close to room
temperature.[53,54] However, on melting it further decreases and
reaches values of roughly 0.60. Setting the packing index to
higher values would not change the qualitative results at all,
and even would only change quantitative results in the range
of 5%. Experimentally determined packing indices are included
in the Supporting Information.
For the charge contribution, we make use of an expression
similar to that given by Marcus for the outer-sphere contribu-
tion to electron-transfer processes.[55,56] This quantity depends
Chem. Eur. J. 2014, 20, 9794 – 9804
www.chemeurj.org
9800
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
on the amount of charge localized on the donor and acceptor
(ZA and ZD), their diameter d and their separation R [Eq. (15)].
Note that, in contrast to electronic transport, the ions in ionic
transport phenomena do not change their diameter during
charge transport. Therefore, only one diameter appears in the
equations,
MALCOT and RENSEJ01) by measuring the mean distance be-
tween fluorine and aluminum or fluorine and nitrogen atoms,
respectively. Adding the van der Waals diameter of fluorine[59]
gave the diameter of the anions. The [NTf2]ꢀ anion is of non-
spherical shape and therefore has longitudinal (from F to F)
and lateral axes. The latter can be seen as the distance be-
tween two adjacent F atoms, between two adjacent O atoms,
or a mixture of both. Since we wanted to have a lower and an
upper limit for the diameter, we took the (shorter) distance be-
tween the O atoms and also added their van der Waals diame-
ter to obtain the lateral diameter. The values are listed in
Table 4. The aluminate shows a diameter of slightly more than
ꢀ
ꢁꢀ
ꢁ
e2
4pe0
1
e
ZD2 ZA2 ZAZD
d
DGpot ffi DFpot ffi Vpot ¼ ꢀ
1 ꢀ
þ
ꢀ
d R
ð15Þ
where Gpot is the Gibbs free energy, Fpot is the Helmholtz free
energy, and Vpot is the potential energy.
Table 4. Anion diameters of [NTf2]ꢀ and [Al(hfip)4]ꢀ calculated from the
crystal structures of their [C2MIm]+ salts and the van der Waals diameters.
We take a typical[57] IL dielectric constant of e=10 and a hop-
ping distance of one ionic diameter, and results are again
rather insensitive to the choice of these two parameters. In the
Marcus picture of a second-order expansion of the free energy,
we arrive at the picture of intersecting parabola, each mini-
mum of which corresponds to the localization of an ion at a fa-
vorable site (cf. Figures 4 and 5). As a reorganization free en-
thalpy l, we have the vertical transfer of an ion from one po-
tential energy to the other without rearranging the ions in the
neighborhood of the acceptor. It is given by the sum of DGHSF
and DGpot. It is straightforward to show that the activation bar-
rier of the ion-transfer process is given by l/4.
anion
longitudinal diameter [nm]
0.950
lateral diameter [nm]
0.549
[NTf2]ꢀ
[Al(hfip)4]ꢀ
1.166
1.1 nm, whereas the imide has a mean diameter of 0.722 nm.
As mentioned above, the activation barrier decreases with in-
creasing particle diameter up to approximately 1 nm. [Al-
(hfip)4]ꢀ lies exactly in this ideal region, whereas the [NTf2]ꢀ
anion appears to be smaller and is further away from the mini-
mum activation energy. Accordingly, [Al(hfip)4]ꢀ should be able
to change its position more easily and thus show lower viscosi-
ties, higher conductivities, and larger diffusion constants. Nev-
ertheless, our results also clearly show that [NTf2]ꢀ is a very
good anion for the synthesis of good ILs and are therefore in
agreement with the vast amount of studies in the last decade
concerning [NTf2]ꢀ ILs. This conclusion can also be supported
by comparing the herein-presented viscosity and conductivity
values with, for example, those of [PF6]ꢀ salts which have
a much smaller anion (Vmꢀ =0.107 nm3[60]). [C4MIm][PF6] has
a viscosity of 207 mPas at 258C[61] and conductivities of
1.465 mScmꢀ1 at 258C and 5.78 mScmꢀ1 at 558C.[62] Both
[C4MIm][NTf2] (3.16 mScmꢀ1 at 258C and 8.22 mScmꢀ1 at 558C)
and [C4MIm][Al(hfip)4] (6.51 mScmꢀ1 at 558C) have much
higher conductivities, and [C4MIm][NTf2] shows a much lower
viscosity of 52.3 mPas at 258C ([C4MIm][Al(hfip)4] is solid at
258C). For [C6MIm]+ salts, the differences are even larger:
[C6MIm][PF6] ( 0.62 mScmꢀ1 at 258C[63]) cannot compete with
Numerical results for the activation barrier as a function of
the hard-sphere diameter are shown in Figure 5. In the region
relevant to the anions discussed here, the barriers are slightly
less than 1 eV (ꢄ100 kJmolꢀ1), quite close to the B values in
the VFT fits of viscosities and conductivities (see Appendix)
multiplied by the Boltzmann constant to convert them to ener-
gies (cf. Tables 5 and 6). Up to about 1 nm, the activation barri-
er decreases with increasing ion diameter. Here, the monotoni-
cally decreasing polarization contribution overcompensates
the monotonically increasing cavity-formation term. In combi-
nation, they give rise to a very shallow curve around the diam-
eters of the ions discussed here.
The diameters of [Al(hfip)4]ꢀ and [NTf2]ꢀ were calculated
from the crystal structures of the [C2MIm]+ salts (CSD refcodes
[C6MIm][NTf2]
(2.09 mScmꢀ1)
and
[C6MIm][Al(hfip)4]
(2.36 mScmꢀ1). Of course, this model, due to its relative sim-
plicity and the diverse chemical nature of different classes of
ILs, cannot cover all subtleties. This becomes clear when look-
ing at viscosity and conductivity data of [BF4]ꢀ salts, for in-
stance. However, it gives a good explanation of general trends
and can shed more light on the behavior of many ionic liquids.
Figure 5. Free activation energies EA for equisized charged hard spheres as
a function of their diameter d. Solid line: hard-sphere contributions based
on Equation (13); circles and interpolated dotted line: cavity-formation free
energies in water.[58] The arrows indicate the lateral (a), average (b), and lon-
gitudinal (c) diameters of [NTf2]ꢀ and the diameter of [Al(hfip)4]ꢀ (d).
Conclusion
We have investigated the dynamic properties of 24 ILs consist-
ing of imidazolium and ammonium cations as well as [NTf2]ꢀ
Chem. Eur. J. 2014, 20, 9794 – 9804
www.chemeurj.org
9801
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
and [Al(hfip)4]ꢀ anions with the focus on ion-specific self-diffu-
sion constants and ionicities.
Table 5. Parameters for the viscosities of [Al(hfip)4]ꢀ and [NTf2]ꢀ ILs in the
VFT Equation [Eq. (4)], determined from the VFT fits in Figure 1. R2 values
were greater than or equal to 0.999 and are not given for each individual
entry.
Viscosities and conductivities show a common behavior,
which can be described by Arrhenius and VFT equations. Most
of the [Al(hfip)4]ꢀ ILs are less viscous and more conductive
than their [NTf2]ꢀ analogues. Additionally, they show a much
smaller dependence on the cation, which can be attributed to
their spherical shape, larger size, and less polarizable surface.
Ion-specific self-diffusion constants D of the ILs were mea-
sured and showed a similar trend to viscosities, that is, [Al-
(hfip)4]ꢀ ILs diffuse faster than [NTf2]ꢀ ILs, and imidazolium ILs
faster than comparable ammonium ILs. It was possible to de-
scribe the ratio D+/Dꢀ of all [NTf2]ꢀ and [Al(hfip)4]ꢀ ILs with
a single equation, respectively. The equations are a generalized
form of the Stokes–Einstein equation, which is not valid here
in its original form. The exponents of the equations showed
that ions in ILs based on [Al(hfip)4]ꢀ anions exhibit less inter-
molecular interactions than ions in [NTf2]ꢀ ILs.
IL
h0 [mPas]
B [K]
T0 [K]
[C2MIm][NTf2]
[AllylMIm][NTf2]
[C4MIm][NTf2]
[C4MMIm][NTf2]
[C6MIm][NTf2]
[C8MIm][NTf2]
[C10MIm][NTf2]
[C2MIm][Al(hfip)4]
[AllylMIm][Al(hfip)4]
[C4MIm][Al(hfip)4]
[C4MMIm][Al(hfip)4]
[C6MIm][Al(hfip)4]
[C8MIm][Al(hfip)4]
[C10MIm][Al(hfip)4]
[N1123][NTf2]
[N1444][NTf2]
[N1888][NTf2]
[N2225][NTf2]
[N2666][NTf2]
[N1123][Al(hfip)4]
[N1444][Al(hfip)4]
[N1888][Al(hfip)4]
[N2225][Al(hfip)4]
[N2666][Al(hfip)4]
0.2458
0.0659
0.3336
0.0559
0.0590
0.0241
0.0389
0.0040
0.1066
0.0430
0.0985
0.1743
0.0134
0.0069
0.0403
0.0061
0.0052
0.0505
0.0036
0.1219
0.0552
0.0102
0.0205
0.0432
632
1009
572
1061
1014
1324
1206
2402
905
169
138
185
158
155
138
148
38
144
132
157
170
112
104
147
157
127
169
140
164
163
125
142
163
1130
853
709
1486
1716
1151
1601
2008
1054
1756
837
From self-diffusion constants and bulk conductivities, ionici-
ties for most of the ILs could be calculated. It was shown that
[Al(hfip)4]ꢀ ILs in most cases have higher ionicities than their
[NTf2]ꢀ analogues, culminating in ionicities of virtually 100%.
Furthermore, variation of cations changes ionicities by altering
the degree of dispersion interaction and steric effects. These
influences, however, are more complex and less predictable.
Moreover, due to the interplay of cations and anions, the
cation effects differ from anion to anion. Since ionicities can be
seen as a measure for the fraction of free ions in a liquid elec-
trolyte, this result means that some [Al(hfip)4]ꢀ ILs with short
aliphatic side chains (nꢄ4) in the cations virtually do not show
ion pairing. To our knowledge, this is the first time that ILs
almost without ion pairing have been reported.
946
1582
1269
1005
unusually small T0 value of the VFT fit for [C2MIm][Al(hfip)4] possibly
results from the small number of data points and from its generally
peculiar behavior (see below). However, the fitting itself was very
stable in the case of [C2MIm][Al(hfip)4] and yielded the same pa-
rameters for a variety of different starting values.
We used a modified Marcus theory to describe our results
quantitatively. With increasing anion diameter, the activation
energy for movement of an ion to an empty site decreases up
to diameters of more than 1 nm. In this picture, [Al(hfip)4]ꢀ has
an almost ideal diameter of 1.17 nm, whereas [NTf2]ꢀ has
a mean diameter of 0.722 nm and therefore a higher activation
energy and worse transport properties than [Al(hfip)4]ꢀ. Alto-
gether, despite its simplicity this model gives a very good de-
scription and explanation of the experimental results, but must
of course be adapted in individual cases to properties such as
ion shape and polarizability. Especially for applications in elec-
trochemical devices, these results can be used to design high-
performance IL-based systems.
Conductivity measurements
Conductivities of the ILs were measured during heating in the tem-
perature range from room temperature up to 808C. Figure 2
shows the results, which were fitted with a VFT ansatz, except for
[C6MIm][Al(hfip)4] and [N1123][Al(hfip)4]. For these ILs, due to unsolv-
able mathematical instabilities and—in the case of the high-melt-
ing [N1123][Al(hfip)4]—few data points, VFT fits could not yield rea-
sonable parameters and an Arrhenius approach was chosen as
more applicable in these cases. For all other ILs, R2 values are
always significantly better for VFT fits, and thus VFT fits were used
to calculate conductivities at 258C (see Ionicities section). Table 6
lists the fit parameters for VFT and Arrhenius fits. In the case of
[C4MMIm][Al(hfip)4], the data point at 323.15 K was not included in
the fit. As with viscosities, [C2MIm][Al(hfip)4] again shows very dif-
ferent values of s0 and T0, which may again be attributed to the
limited number of data points as well as to the overall behavior of
this IL.
Appendix
Conductivities follow analogous trends to viscosities: The smaller
the cation, the more conductive the IL. [Al(hfip)4]ꢀ ILs are more
Although for most of the examined ILs, viscosity and conductivity
63–65]
data exist,[4,38,
we measured all ILs again on the same device
conductive than [NTf2]ꢀ ILs, again with the exceptions of the small
under the same conditions to exclude errors resulting from differ-
ent setups and to ensure reproducibility and comparability of the
data.
[C2MIm]+ and [N1123
]
cations.
+
Nomenclature
Viscosity measurements
Ammonium cations are abbreviated as [Nabcd]+, and 1-methyl-3-n-
alkylimidazolium cations as [CnMIm]+. The indices a, b, c, d, and n
describe the number of carbon atoms of the n-alkyl chains at-
VFT fitting parameters of all investigated ILs are listed in Table 5.
The experimental data points are included in Figures 1 and 2. The
Chem. Eur. J. 2014, 20, 9794 – 9804
www.chemeurj.org
9802
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
[8] O. Hollꢃczki, D. Gerhard, K. Massone, L. Szarvas, B. Nemeth, T. Veszpre-
mi, L. Nyulaszi, New J. Chem. 2010, 34, 3004–3009.
[9] X. Han, D. W. Armstrong, Acc. Chem. Res. 2007, 40, 1079–1086.
[10] G. B. Appetecchi, M. Montanino, M. Carewska, M. Moreno, F. Alessandri-
ni, S. Passerini, Electrochim. Acta 2011, 56, 1300–1307.
[11] C. Arbizzani, M. Biso, D. Cericola, M. Lazzari, F. Soavi, M. Mastragostino,
J. Power Sources 2008, 185, 1575–1579.
Table 6. Parameters for the conductivities of [Al(hfip)4]ꢀ and [NTf2]ꢀ ILs in
the VFT equation [Eq. (6)], determined from the VFT fits in Figure 2.
[C6MIm][Al(hfip)4] and [N1123][Al(hfip)4] could only be fitted sensibly with
an Arrhenius approach [Eq. (5)].
IL
s0 [mScmꢀ1
]
B [K]
T0 [K]
[12] A. Balducci, F. Soavi, M. Mastragostino, Appl. Phys. A: Mater. Sci. Process.
2006, 82, 627–632.
[13] M. Diaw, A. Chagnes, B. Carrꢄ, P. Willmann, D. Lemordant, J. Power Sour-
ces 2005, 146, 682–684.
[14] S. Passerini, F. Alessandrini, G. B. Appetecchi, M. Conte, ECS Trans. 2006,
1, 67–71.
[15] F. Wu, R. Chen, F. Wu, L. Li, B. Xu, S. Chen, G. Wang, J. Power Sources
2008, 184, 402–407.
[16] D. R. MacFarlane, M. Forsyth, P. C. Howlett, J. M. Pringle, J. Sun, G.
Annat, W. Neil, E. I. Izgorodina, Acc. Chem. Res. 2007, 40, 1165–1173.
[17] K. R. Seddon, Nat. Mater. 2003, 2, 363–365.
[18] W. Xu, E. I. Cooper, C. A. Angell, J. Phys. Chem. B 2003, 107, 6170–6178.
[19] M.-M. Huang, S. Bulut, I. Krossing, H. Weingꢂrtner, J. Chem. Phys. 2010,
133, 101101.
[C2MIm][NTf2]
[AllylMIm][NTf2]
[C4MIm][NTf2]
[C4MMIm][NTf2]
[C6MIm][NTf2]
[C8MIm][NTf2]
[C10MIm][NTf2]
[C2MIm][Al(hfip)4]
[AllylMIm][Al(hfip)4]
[C4MIm][Al(hfip)4]
[C4MMIm][Al(hfip)4]
[C6MIm][Al(hfip)4]
[C8MIm][Al(hfip)4]
[C10MIm][Al(hfip)4]
[N1123][NTf2]
[N1444][NTf2]
[N1888][NTf2]
[N2225][NTf2]
[N2666][NTf2]
[N1123][Al(hfip)4]
[N1444][Al(hfip)4]
[N1888][Al(hfip)4]
[N2225][Al(hfip)4]
[N2666][Al(hfip)4]
174.8
1024.6
267.6
1110.4
345.6
316.7
398.1
1000.2
93.1
37.89
558.4
31989[a]
154.2
294.2
2038
301
769
481
857
588
622
773
1258
241
113
204
144.0
190
163
183
186
173
68
220
264
144.6
425.6
23.65[a]
537
796
1120
1060
901
181
152
128
176
188
177
169
[20] A. W. Taylor, P. Licence, A. P. Abbott, PhysChemComm PhysChemChem-
Phys 2011, 13, 10147–10154.
1536
[21] P. Walden, Z. Phys. Chem. (Leipzig), 1906, 55, 207–246.
[22] P. Walden, Z. Phys. Chem. (Leipzig), 1906, 55, 246.
[23] H. Rodrꢅguez, J. F. Brennecke, J. Chem. Eng. Data 2006, 51, 2145.
[24] O. O. Okoturo, T. J. VanderNoot, J. Electroanal. Chem. 2004, 568, 167.
[25] H. Tokuda, K. Hayamizu, K. Ishii, M. A. B. H. Susan, M. J. Watanabe, J.
Phys. Chem. B 2005, 109, 6103–6110.
181.5
772.6
546.0
9994.4[a]
259.3
377.1
115.8
315.7
809
1050
20.46[a]
713
1020
430
868
167
155
199
165
[26] J. Jacquemin, P. Husson, A. A. H. Padua, V. Majer, Green Chem. 2006, 8,
172.
[27] K. Ueno, H. Tokuda, M. Watanabe, Phys. Chem. Chem. Phys. 2010, 12,
1649–1658.
[28] D. R. MacFarlane, P. Meakin, J. Sun, N. Amini, M. Forsyth, J. Phys. Chem.
B 1999, 103, 4164–4170.
[a] Since VFT fits did not converge to reasonable values, Arrhenius param-
eters s0 [in mScmꢀ1] and A [in kJmolꢀ1] are given.
[29] H. Matsumoto, T. Mazda, Y. Miyazaki, Chem. Lett. 2000, 29, 1430–1431.
[30] A. P. Abbott, ChemPhysChem 2004, 5, 1242–1246.
[31] U. Preiss, S. Bulut, I. Krossing, J. Phys. Chem. B 2010, 114, 11133–11140.
[32] U. P. R. M. Preiss, J. M. Slattery, I. Krossing, Ind. Eng. Chem. Res. 2009, 48,
2290–2296.
[33] E. O. Stejskal, J. E. Tanner, J. Chem. Phys. 1965, 42, 288–292.
[34] T. Umecky, M. Kanakubo, Y. Ikushima, Fluid Phase Equilib. 2005, 228–
229, 329–333.
tached to the N atom. Exceptions are 1-allyl-3-methylimidazolium
and 1-butyl-2,3-dimethylimidazolium, which are abbreviated as [Al-
lylMIm]+ and [C4MMIm]+, respectively. Further abbreviations are as
follows: hfip=hexafluoroisopropoxy, Tf=trifluoromethanesulfo-
nate.
[35] S. Seki, K. Hayamizu, S. Tsuzuki, K. Fujii, Y. Umebayashi, T. Mitsugi, T. Ko-
bayashi, Y. Ohno, Y. Kobayashi, Y. Mita, H. Miyashiro, S.-i. Ishiguro, Phys.
Chem. Chem. Phys. 2009, 11, 3509–3514.
Acknowledgements
[36] H. Liu, E. Maginn, A. E. Visser, N. J. Bridges, E. B. Fox , Ind. Eng. Chem.
Res. 2012, 51, 7242–7254.
[37] A. Noda, K. Hayamizu, M. Watanabe, J. Phys. Chem. B 2001, 105, 4603–
4610.
[38] B. Garcia, S. Lavallee, G. Perron, C. Michot, M. Armand, Electrochim. Acta
2004, 49, 4583–4588.
We would like to thank the Deutsche Forschungsgemeinschaft
(SPP1191 “Ionic Liquids” and “WCAcaps” project) for financial
support. A. Rupp wishes to thank the Fonds der Chemischen
Industrie for financial support.
[39] H. Tokuda, S. Tsuzuki, M. A. B. H. Susan, K. Hayamizu, M. Watanabe, J.
Phys. Chem. B 2006, 110, 19593–19600.
[40] G. Annat, D. R. MacFarlane, M. Forsyth, J. Phys. Chem. B 2007, 111,
9018–9024.
Keywords: ion pairs · ionic liquids · Marcus theory · transport
properties · weakly coordinating anions
[41] O. Borodin, W. Gorecki, G. D. Smith, M. Armand, J. Phys. Chem. B 2010,
114, 6786–6798.
[42] T.-Y. Wu, I.-W. Sun, S.-T. Gung, M.-W. Lin, B.-K. Chen, H. P. Wang, S.-G. Su,
J. Taiwan Inst. Chem. Eng. 2011, 42, 513–522.
[1] H. Weingꢂrtner, Angew. Chem. 2008, 120, 664–682; Angew. Chem. Int.
Ed. 2008, 47, 654–670.
[43] K. R. Harris, M. Kanakubo, Faraday Discuss. 2012, 154, 425–438.
[44] H. Matsumoto, H. Sakaebe, K. Tatsumi, J. Power Sources 2005, 146, 45–
50.
[45] S. Tsuzuki, ChemPhysChem 2012, 13, 1664–1670.
[46] S. Zahn, G. Bruns, J. Thar, B. Kirchner, Phys. Chem. Chem. Phys. 2008, 10,
6921–6924.
[2] R. D. Rogers, K. R. Seddon, Science 2003, 302, 792–793.
[3] I. Raabe, K. Wagner, K. Guttsche, M. Wang, M. Grꢂtzel, G. Santiso-Qui-
Çones, I. Krossing, Chem. Eur. J. 2009, 15, 1966–1976.
[4] S. Bulut, P. Klose, M.-M. Huang, H. Weingꢂrtner, P. J. Dyson, G. Laurenczy,
C. Friedrich, J. Menz, K. Kꢀmmerer, I. Krossing, Chem. Eur. J. 2010, 16,
13139–13154.
[5] P. Wasserscheid and T. Welton, Ionic Liquids in Synthesis, Wiley-VCH,
Weinheim, 2008.
[47] K. Fumino, A. Wulf, R. Ludwig, Angew. Chem. 2008, 120, 8859–8862;
Angew. Chem. Int. Ed. 2008, 47, 8731–8734.
[6] D. Wei, T. W. Ng, Electrochem. Commun. 2009, 11, 1996–1999.
[7] J. L. Anderson, J. K. Dixon, J. F. Brennecke, Acc. Chem. Res. 2007, 40,
1208–1216.
[48] P. A. Hunt, J. Phys. Chem. B 2007, 111, 4844–4853.
[49] H. Tokuda, K. Hayamizu, K. Ishii, M. A. B. H. Susan, M. Watanabe, J. Phys.
Chem. B 2004, 108, 16593–16600.
Chem. Eur. J. 2014, 20, 9794 – 9804
www.chemeurj.org
9803
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
[50] F. L. Hirshfeld, Theor. Chim. Acta 1977, 44, 129–138.
[51] M. A. Spackman, D. Jayatilaka, CrystEngComm 2009, 11, 19–32.
[52] J. J. McKinnon, M. A. Spackman, A. S. Mitchell, Acta Crystallogr. Sect. B:
Structural Sci. 2004, 60, 627–668.
[53] W. Beichel, P. Eiden, I. Krossing, ChemPhysChem 2013, 14, 3221–3226.
[54] W. Beichel in From the Crystal Structures of Organic Salts to Consistent
Ion Volumes and their Charge Distribution, Thesis, Albert-Ludwigs-Univer-
sitaet, Freiburg, 2013.
[60] C. Ye, J. n. M. Shreeve, J. Chem. Phys. A 2007, 111, 1456–1461.
[61] C. Frez, G. J. Diebold, C. D. Tran, S. Yu, J. Chem. Eng. Data 2006, 51,
1250–1255.
[62] O. Zech, A. Stoppa, R. Buchner, W. Kunz, J. Chem. Eng. Data, 2010, 55,
1774–1778.
[63] K. Nakamura, T. Shikata, ChemPhysChem 2010, 11, 285–294.
[64] S. Bulut in Novel Ionic Liquids with Weakly Coordinating Anions: Synthesis
Characterization, Principal Physical Properties and Their Predictions,
Thesis, Albert-Ludwigs-Universitꢂt, Freiburg, 2010.
[55] R. A. Marcus, J. Chem. Phys. 1956, 24, 966–978.
[56] R. A. Marcus, J. Chem. Phys. 1956, 24, 979–989.
[65] M. Tariq, P. J. Carvalho, J. A. P. Coutinho, I. M. Marrucho, J. N. C. Lopes,
L. P. N. Rebelo, Fluid Phase Equilib. 2011, 301, 22–32.
[57] H. Weingꢂrtner, Zeitschr. Physik. Chem. 2006, 220, 1395–1405.
[58] T. C. Beutler, D. R. Beguelin, W. F. van Gunsteren, J. Chem. Phys. 1995,
102, 3787–3793.
[59] M. Mantina, A. C. Chamberlin, R. Valero, C. J. Cramer, D. G. Truhlar, J.
Phys. Chem. A 2009, 113, 5806–5812.
Received: January 15, 2014
Published online on July 10, 2014
Chem. Eur. J. 2014, 20, 9794 – 9804
www.chemeurj.org
9804
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim