Electrolytic conductivity of four imidazolium-based room-temperature ionic liquids and the effect of a water impurity

Article abstract of DOI:10.1016/j.jct.2005.04.009

The electrolytic conductivities (κ) of four hydrophobic room-temperature ionic liquids (RTILs) were measured at temperatures from (288.15 to 323.15) K. The measurements were made with a commercial conductivity cell with platinum black electrodes. In order to exclude atmospheric moisture, the conductivity cell was modified so that it could be sealed during measurements. The RTILs studied were 1-ethyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide, 1-butyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide, 1-hexyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide, and 1-butyl-3-methylimidazolium hexafluorophosphate. The RTILs were dried under vacuum until the mass fraction of water (wH2O) was ≤10^-5. Coulometric Karl Fischer titration was used to determine wH2O in each RTIL before and after measurements of κ. Most measurements were made on dried RTILs, but some measurements were performed intentionally at higher wH2O in order to study the effect of a water impurity on κ. For {water + 1-butyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide}, κ was found to increase dramatically with increasing wH2O; for example, in going from wH2O=10-5to10-2, the fractional increase in κ was 0.36 {=(κ_wet - κ _dried)/κ_dried}. This work illustrates the importance of measuring the water content in RTILs both before and after measurements of κ. Published by Elsevier Ltd.

Full text of DOI:10.1016/j.jct.2005.04.009

J. Chem. Thermodynamics 37 (2005) 569–575
www.elsevier.com/locate/jct
Electrolytic conductivity of four imidazolium-based
room-temperature ionic liquids and the effect of a water impurity
a
Jason A. Widegren , Eric M. Saurer , Kenneth N. Marsh , Joseph W. Magee
b
c
a, ,1
*
a
Physical and Chemical Properties Division, National Institute of Standards and Technology, Boulder, CO 80305, USA
b
Department of Chemical and Biomolecular Engineering, University of Notre Dame, Notre Dame, IN 46556, USA
c
Department of Chemical and Process Engineering, University of Canterbury, Private Bag 4800, Christchurch, New Zealand
Received 7 February 2005; received in revised form 1 April 2005; accepted 9 April 2005
Abstract
The electrolytic conductivities (j) of four hydrophobic room-temperature ionic liquids (RTILs) were measured at temperatures
from (288.15 to 323.15) K. The measurements were made with a commercial conductivity cell with platinum black electrodes. In
order to exclude atmospheric moisture, the conductivity cell was modified so that it could be sealed during measurements. The
RTILs studied were 1-ethyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide, 1-butyl-3-methylimidazolium bis(trif-
luoromethylsulfonyl)imide, 1-hexyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide, and 1-butyl-3-methylimidazolium
hexafluorophosphate. The RTILs were dried under vacuum until the mass fraction of water ðwH OÞ was 610ꢀ5. Coulometric Karl
2
Fischer titration was used to determine wH O in each RTIL before and after measurements of j. Most measurements were made on
2
dried RTILs, but some measurements were performed intentionally at higher wH O in order to study the effect of a water impurity on
2
j. For {water + 1-butyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide}, j was found to increase dramatically with increas-
ing wH O; for example, in going from wH O ¼ 10ꢀ5 to 10ꢀ2, the fractional increase in j was 0.36 {=(jwet ꢀ jdried)/jdried}. This work
2
2
illustrates the importance of measuring the water content in RTILs both before and after measurements of j.
Published by Elsevier Ltd.
Keywords: Electrolytic conductivity; Room-temperature ionic liquid; Room-temperature molten salt; Specific conductance; Water impurity
1. Introduction
sure, and a wide liquid range [2–4]. As a result, RTILs
have attracted a great deal of attention for their poten-
tial use in electrochemical devices such as capacitors,
dye-sensitized solar cells, batteries, and fuel cells.
Room-temperature ionic liquids (RTILs) are salts
that melt at or below room temperature [1]. One inher-
ent difference between RTILs and conventional molecu-
lar liquids is that RTILs exhibit high electrolytic
conductivity, j (formerly called specific conductance
or ionic conductivity). They also have a variety of other
potentially useful properties, including excellent electro-
chemical and thermal stability, negligible vapor pres-
When selecting an RTIL for an electrochemical use, j
is of critical importance [2]. One impediment to the use
of RTILs is the shortage of reliable data for j [2,5]. For
many RTILs, little or no data have been published. An
example of such a case is 1-hexyl-3-methylimidazolium
bis(trifluoromethylsulfonyl)imide, abbreviated herein
as [C6mim][Tf2N], for which the only available data
are at (295 2) K [6]. For other RTILs, extensive, but
inconsistent, data have been published. An example of
such a case is 1-ethyl-3-methylimidazolium bis(trif-
luoromethylsulfonyl)imide, [C2mim][Tf2N], for which
*
Corresponding author. Tel.: +1 303 497 3298; fax: +1 303 497 5224.
E-mail address: magee@boulder.nist.gov (J.W. Magee).
1
Official contribution of the National Institute of Standards and
Technology; not subject to copyright in the United States.
0021-9614/$ - see front matter. Published by Elsevier Ltd.
doi:10.1016/j.jct.2005.04.009

570
J.A. Widegren et al. / J. Chem. Thermodynamics 37 (2005) 569–575
several groups have reported values of j in the temper-
ature range (263 to 401) K [7–14]. Unfortunately, the
data sets for [C2mim][Tf2N] are in poor agreement, as
illustrated by the published values of j at 293 K: 0.49
S Æ mꢀ1 (from a correlation) [10], 0.77 S Æ mꢀ1 (estimated
from a graph) [12], 0.82 S Æ mꢀ1 (estimated from a
graph) [7], 0.88 S Æ mꢀ1 [14], and 0.99 S Æ mꢀ1 [8,11].
Inconsistent sample purity is a likely reason for much
of the disagreement in the reported values of j for
RTILs [2]. Water is arguably the most important impu-
rity to consider for four reasons. First, atmospheric
moisture is ubiquitous. Second, even ‘‘hydrophobic’’
RTILs, which are not miscible with water, rapidly ab-
sorb water from the atmosphere [15–18] or from moist
surfaces. Hence, unless an RTIL is carefully dried and
handled, it will be contaminated with water. Third, some
data already exist that show that the presence of water
increases j in (RTIL + water) [6,19,20]. Fourth, litera-
ture reports regarding the viscosity (g) of (RTIL +
water) suggest that even trace levels of water could sig-
nificantly increase j by lowering g [21,22]. That work
2. Experimental
2.1. Materials
The
[C2mim][Tf2N],
[C4mim][Tf2N],
and
[C4mim][PF6] were commercially obtained. The
[C6mim][Tf2N] was synthesized by Dr. Mark Muldoon
at the University of Notre Dame using the following
procedure. First, 1-methylimidazole (mole fraction
x = 0.476) was mixed with a small excess of 1-bromo-
hexane (x = 0.524) at room temperature in a round-
bottom flask fitted with a condenser and N2 bubbler.
After (30 to 45) min the clear mixture started to become
cloudy, at which point the flask was placed in an ice bath
and left to stir overnight. The reaction mixture was then
heated at 343 K for 8 h, resulting in complete conversion
of 1-methylimidazole to 1-hexyl-3-methylimidazolium
bromide. The bromide salt was then washed with dry
ethyl acetate several times to remove the excess 1-
bromohexane. The residual ethyl acetate was removed
under reduced pressure at 323 K. Then the 1-hexyl-3-
methylimidazolium bromide (x = 0.476) was mixed with
a small excess of lithium bis(trifluoromethylsulfo-
nyl)imide (x = 0.524) by dissolving each in deionized
water and combining the solutions. This mixture was
stirred for 3 h, the product forming a separate phase be-
low the aqueous solution. To ensure the removal of the
excess lithium bis(trifluoromethylsulfonyl)imide and the
LiBr byproduct, the water-immiscible [C6mim][Tf2N]
was washed with deionized water until the water phase
no longer gave a precipitate upon addition of a silver ni-
trate solution. As a precaution, the [C6mim][Tf2N] was
then washed several more times with deionized water
to ensure that the last traces of the water-soluble salts
were removed. At this point, the [C6mim][Tf2N] was
dried under vacuum at 333 K to remove the residual
water. Analysis with a commercial bromide-selective
electrode showed that the mass fraction of bromide
was below the detection limit of the electrode
(wBr < 10ꢀ5). To remove traces of color and other impu-
rities, the [C6mim][Tf2N] was then dissolved in dichloro-
methane and stirred overnight with activated carbon at
343 K before being passed through a column of acti-
vated alumina (60% acidic, 20% basic, and 20% neutral).
The dichloromethane was then removed under reduced
pressure.
showed that a mass fraction of water ðwH OÞ of only
2
10ꢀ4 can significantly lower g in (RTIL + water) [21].
According to the Walden Rule [2,23], the molar conduc-
tivity (K) is inversely proportional to g,
K ꢁ g ¼ constant;
ð1Þ
where K = j/c, and c is the amount of substance concen-
tration. Hence, on the basis of the Walden Rule, we pre-
dict that trace amounts of water will significantly
increase j in (RTIL + water).
The first aim of this work was to measure j as a func-
tion of temperature for RTILs of known high purity,
especially with respect to water content. The second
aim of this work was assess the effect of small amounts
of water on j in (RTIL + water). A key to these exper-
iments was that coulometric Karl Fischer titration was
used to determine wH O both before and after measure-
2
ments of j [21].
The RTILs chosen for this study are based on the 1-n-
alkyl-3-methylimidazolium cation, abbreviated herein as
[Cimim]+, where i equals the number of carbons in the n-
alkyl side chain. This class of cations is widely used be-
cause it results in RTILs with relatively low viscosities
and relatively high thermal stabilities. The primary an-
ion chosen for this study is bis(trifluoromethylsulfo-
nyl)imide, [(F3CSO2)2N]ꢀ, abbreviated herein as
[Tf2N]ꢀ. This anion is widely used because it results in
RTILs with relatively low viscosities, and it has good
hydrolytic and electrochemical stability [14]. One RTIL
containing the hexafluorophosphate anion, [PF6]ꢀ, was
included in this study for the purposes of comparison.
In all, j was measured for four RTILs: [C2mim][Tf2N],
[C4mim][Tf2N], [C6mim][Tf2N], and [C4mim][PF6].
All four RTILs were dried thoroughly before being
used for conductivity measurements. The drying was
done at room temperature with stirring on a vacuum
line using a mechanical pump and a liquid nitrogen trap;
the ultimate pressure during drying was 60.1 Pa. After
drying, the RTILs were stored in a nitrogen atmosphere
in a glove box. Coulometric Karl Fischer titration indi-
cated that wH O 6 10ꢀ5 for all four dried RTILs. For the
2
Additionally,
[C4mim][Tf2N] and water.
j
was measured for mixtures of
three commercially obtained RTILs, silver nitrate tests
and chloride-selective electrode measurements were used

J.A. Widegren et al. / J. Chem. Thermodynamics 37 (2005) 569–575
571
to verify the manufacturerꢀs claim that wCl < 5 Æ 10ꢀ5
(The counter ion associated with a chloride impurity is
.
was determined to be (100.0 0.5) mꢀ1. Since the two
values of Kcell are the same within the estimated uncer-
tainties, we used the average value and assumed the lar-
simply the [Cimim]+.) After drying, the purities of the
1
RTILs were also checked by H and 19F nuclear mag-
ger uncertainty, Kcell = (99.9 0.7) mꢀ1
.
netic resonance (NMR) spectrometry. To check if con-
tamination or decomposition had occurred during
measurements of j, NMR spectra were also obtained
after measuring j. In every case, the purity of the RTILs
The conductivity cell was carefully cleaned and dried
before each RTIL sample was introduced. The cell was
cleaned by soaking and rinsing with ethanol, then with
deionized water with a resistivity of P18 MX Æ cm.
The cell was dried by flowing dry nitrogen through it
at room temperature for about 1 h, then placing it under
vacuum (uncapped) in the antechamber of the glove box
for about 1 h. The cell was then brought into the nitro-
gen atmosphere glove box and filled with a sample of
RTIL. An aliquot of the RTIL sample was then re-
moved from the cell for analysis by coulometric Karl
Fischer titration. At this point, the cell was capped, re-
moved from the glove box, and placed in the constant-
temperature bath. At each temperature, values of R
were collected from f = (0.8 to 5) kHz. Then R was plot-
ted as a function of fꢀ0.5 or fꢀ1, whichever was most lin-
1
was >99.5% by both H and 19F NMR.
Mixtures of ([C4mim][Tf2N] + water) were made by
adding deionized water (with a resistivity of P18
MX Æ cm) to the dried [C4mim][Tf2N] and stirring. These
mixtures were handled in the same manner as the dried
RTIL samples.
2.2. Apparatus and procedure
The ac impedance bridge technique [2,24,25] was used
to measure j as a function of T at about 0.1 MPa. Mea-
surements were made with a small-volume commercial
conductivity cell with a nominal cell constant, Kcell of
100 mꢀ1. The cell was made of borosilicate glass with
two platinum black electrodes. In order to exclude
atmospheric moisture during measurements, the cell
was modified so that it could be capped on both ends
with O-ring-sealed poly(tetrafluoroethylene) (PTFE)
plugs. A small void volume at both ends of the cell (con-
taining N2 at about 0.1 MPa) assured that the liquid
sample did not contact the O-rings or the PTFE plugs.
During measurements, the cell was immersed in a con-
stant temperature bath filled with mineral oil. The bath
temperature was measured with an ITS-90 calibrated
platinum resistance thermometer. The uncertainty in
the temperature is estimated to be 0.1 K. To avoid lead
resistance effects, the cell was connected via four wires to
a commercial precision LCR (inductance, capacitance,
and resistance) meter. With this meter, the standard
uncertainty in the measurement of electric resistance,
R, is estimated to be u = 5 Æ 10ꢀ4R. A drive voltage of
1.0 Vrms was used in the C–R series mode for all
measurements.
ear, and extrapolated to f [25]. The extrapolated value
1
of R at f and Kcell were used to calculate j from equa-
1
tion (2). Following data collection, the conductivity cell
was taken back into the glove box, and aliquots of the
RTIL sample were removed from the cell for Karl
Fischer and NMR analysis.
Coulometric Karl Fischer titration was carried out in
accordance with ASTM Standard Test Method E 1064 –
00, except sample sizes of about 1.5 g were used. The val-
ues of wH O given herein refer to the average of the values
2
obtained before and after measurements of j. In general,
the value of wH O before and after measurements of j
changed by less than the experimental uncertainty in
2
the Karl Fischer measurement. For example, the values
of wH O listed in Table 2 are the averages (rounded to
2
the nearest 1 Æ 10ꢀ5) of the following raw data from Karl
Fischer
wH O;after ¼ 8.8 ꢁ 10ꢀ6
titrations:
;
wH O;before ¼ 8.8 ꢁ 10ꢀ6
and
and
2
wH O;before ¼ 8.44 ꢁ 10ꢀ5
2
2
wH O;after ¼ 8.02 ꢁ 10ꢀ5
;
wH O;before ¼ 1.021 ꢁ 10ꢀ3 and
2
2
wH O;after ¼ 1.019 ꢁ 10ꢀ3
;
and wH O;before ¼ 8.804 ꢁ 10ꢀ3
2
2
and wH O;after ¼ 8.898 ꢁ 10ꢀ3. Uncertainties in wH O were
2
2
The cell was calibrated at T = 298.15 K using two
commercially available conductivity standards (aqueous
potassium chloride solutions) with j = (0.1 0.0005)
S Æ mꢀ1 and j = (1.0 0.0025) S Æ mꢀ1 at that tempera-
estimated from multiple measurements on commercially
available water standards for Karl Fischer titration. Not
surprisingly, the uncertainty in wH O changes signifi-
2
cantly as
a
function of the composition. At
ture. First, R at infinite frequency (f ) was determined
wH O ¼ 10ꢀ2, the expanded uncertainty (with a coverage
1
2
by measuring R from f (0.8 to 5) kHz, then plotting R
as a function of fꢀ0.5, and extrapolating the resulting line
to f [25]. For each conductivity standard, the extrapo-
factor of 2) is estimated to be U ¼ 0.04wH O; at
2
wH O ¼ 10ꢀ3, the expanded uncertainty is estimated to
2
be U ¼ 0.06wH O; and at wH O ¼ 10ꢀ4, the expanded
1
2
2
lated value of R at f was then used to calculate the va-
uncertainty is estimated to be U ¼ 0.2wH O
.
1
lue of Kcell from the equation
2
The uncertainty in the bath temperature results in a
relative standard uncertainty of ur 6 8 Æ 10ꢀ3 in the mea-
surement of j. The estimated uncertainty in Kcell results
in a relative standard uncertainty of ur = 7 Æ 10ꢀ3 in the
measurement of j. Based on the results of
([C4mim][Tf2N] + water), the estimated uncertainty in
j ¼ Kcell=R.
ð2Þ
Using the (0.1 0.0005) S Æ mꢀ1 standard, the value of
Kcell was determined to be (99.8 0.7) mꢀ1, and, using
the (1.0 0.0025) S Æ mꢀ1 standard, the value of Kcell

572
J.A. Widegren et al. / J. Chem. Thermodynamics 37 (2005) 569–575
TABLE 2
Electrolytic conductivities (j) of ([C4mim][Tf2N] + water) as a function
wH O for the dried RTILs results in a relative standard
2
uncertainty of ur = 8 Æ 10ꢀ4 in the measurement of j.
The estimated uncertainty in the measurement of R with
the precision LCR meter results in a relative standard
uncertainty of ur = 5 Æ 10ꢀ4 in the measurement of j.
Hence, using standard techniques for the propagation
of uncertainty, the expanded uncertainty (with a cover-
age factor of 2) in the measurement of j is U = 0.02 j.
of the mass fraction of water (wH O) and of the temperature
2
wH O
T/K
j/(S Æ mꢀ1
)
2
1 Æ 10ꢀ5
1 Æ 10ꢀ5
1 Æ 10ꢀ5
288.15
293.15
298.15
0.265
0.331
0.406
8 Æ 10ꢀ5
8 Æ 10ꢀ5
8 Æ 10ꢀ5
288.15
293.15
298.15
0.266
0.332
0.408
1.02 Æ 10ꢀ3
1.02 Æ 10ꢀ3
1.02 Æ 10ꢀ3
288.15
293.15
298.15
0.277
0.345
0.422
3. Results and discussion
3.1. Electrolytic conductivities of dried RTILs
8.85 Æ 10ꢀ3
8.85 Æ 10ꢀ3
8.85 Æ 10ꢀ3
288.15
293.15
298.15
0.365
0.449
0.542
For each of the four dried RTILs, j was determined
as a function of T at temperatures of (288.15, 293.15,
298.15, 308.15, and 323.15) K using the ac impedance
bridge technique (table 1). Special care was taken to
control the amount of water present in each RTIL dur-
ing these measurements, as described in Section 2. Fig-
ure 1 shows j as a function of T for all four RTILs,
as well as correlations of the data by quadratic equa-
tions (see figure 1 caption for the specific equations).
Figure 2 shows the relative deviations of j from the cor-
relations as a function of T. Figure 2 shows that the qua-
dratic equations do a better job of correlating the data
for the [Tf2N]-based RTILs, but even the correlation
for [C4mim][PF6] is within the expanded uncertainty of
the measurements. For the entire temperature range of
(288.15 to 323.15) K, the order of conductivities is
cation. At 293.15 K, for example, (j[C mim][Tf N]
ꢀ
ꢀ
2
2
j
j
[C mim][Tf N])/j[C mim][Tf N] = 0.57, and (j[C mim][Tf N]
2
2
2
4
2
4
[C mim][Tf N])/j[C mim][Tf N] = 0.48. These changes are
2
2
6
4
consistent with a literature report [14] in which it was
=
found that (j[C mim][Tf N] ꢀ j[C mim][Tf N])/j[C mim][Tf N]
2
2
2
2
4
2
0.56 at 293 K. We also find that [C4mim][PF6] has a con-
siderably lower conductivity than [C4mim][Tf2N]; for
example, at 293.15 K (j[C mim][Tf N] ꢀ j[C mim][PF ])/
2
4
4
6
j
[C mim][Tf N] = 0.67.
2
4
The relative conductivities of RTILs are typically
compared with the Stokes–Einstein relation,
Di ¼ kBT =6pgri;
ð3Þ
j
[C mim][Tf N] > j[C mim][Tf N] > j[C mim][Tf N] > j[C mim][PF6].
2
2
2
2
4
6
4
Figure 1 shows that j is quite sensitive to the size of the
2.0
1.6
1.2
0.8
0.4
0.0
TABLE 1
Temperature dependence of the electrolytic conductivities (j) of dried
room-temperature ionic liquids (RTILs)
RTIL
T/K
j/(S Æ mꢀ1
)
[C2mim][Tf2N]
288.15
293.15
298.15
308.15
323.15
0.644
0.773
0.912
1.229
1.789
[C4mim][Tf2N]
[C6mim][Tf2N]
[C4mim][PF6]
288.15
293.15
298.15
308.15
323.15
0.265
0.331
0.406
0.585
0.924
288.15
293.15
298.15
308.15
323.15
0.135
0.173
0.218
0.327
0.546
280
290
300
310
320
330
T / K
FIGURE 1. Plot of the electrolytic conductivity (j) as a function of
temperature for four RTILs. For these dried RTILs the mass fraction of
water is 10ꢀ5
.
}, [C2mim][Tf2N], correlated by {j/(S Æ mꢀ1)} =
288.15
293.15
298.15
308.15
323.15
0.079
0.109
0.146
0.247
0.477
(2.330 Æ 10ꢀ4)(T/K)2 ꢀ 0.1097(T/K) + 12.91; h, [C4mim][Tf2N], corre-
lated by {j/(S Æ mꢀ1)} = (1.878 Æ 10ꢀ4)(T/K)2 ꢀ (9.596 Æ 10ꢀ2)(T/K) +
12.32; s, [C6mim][Tf2N], correlated by {j/(S Æ mꢀ1)} = (1.408 Æ 10ꢀ4
)
(T/K)2 ꢀ (7.435 Æ 10ꢀ2)(T/K) + 9.867; , [C4mim][PF6], correlated by
{j/(S Æ mꢀ1)} = (1.939 Æ 10ꢀ4)(T/K)2 ꢀ 0.1072(T/K) + 14.86.
The mass fraction of water in these four dried RTILs is 10ꢀ5
.

J.A. Widegren et al. / J. Chem. Thermodynamics 37 (2005) 569–575
573
0.065
2
1
0.060
0.055
0.050
0.045
0.040
0
-1
-2
0.000
0.003
0.006
0.009
0.012
280
290
300
310
320
330
FIGURE 3. Plot of the viscosities (g) of ([C4mim][Tf2N] + water) as a
T / K
function of the mass fraction of water (wH O) at 293.15 K [21]. d,
2
FIGURE 2. Plot of the relative deviations of the experimental
electrolytic conductivities (jexp) from the electrolytic conductivities
given by the correlations (jcorr) in figure 1. }, [C2mim][Tf2N]; h,
[C4mim][Tf2N]; ꢂ, [C6mim][Tf2N]; , [C4mim][PF6].
Measured
values;
—,
correlation
with
the
equation
2
fg=ðPa ꢁ sÞg ¼ 80.83 ꢁ ðwH OÞ ꢀ 2.840 ꢁ wH O þ 0.06355.
2
2
where Di is the self-diffusivity of species i, kB is the Boltz-
mann constant, T is the temperature, g is the viscosity,
and ri is the radius of the species i [2,23]. In the absence
of ion pairing, Di is directly proportional to j [8]. From
equation (3) it is easy to rationalize the relative conduc-
tivities of [C2mim][Tf2N], [C4mim][Tf2N], and
[C6mim][Tf2N]. Since [C2mim][Tf2N] has the smallest
cation and the lowest g [14], it is expected to have the
largest Di and, therefore, the largest j of the three
[Tf2N]-based RTILs, which it does. [C6mim][Tf2N] has
the largest cation and the highest g, so it is expected to
have the lowest j of the three [Tf2N]-based RTILs,
which it does. Predicting the relative conductivity of
[C4mim][PF6] compared to the three [Tf2N]-based
RTILs is more difficult, since [C4mim][PF6] has the high-
est g of the four RTILs studied herein, but it also has the
smallest anion. Experimentally we find that the high g
has a greater effect than the small anion size because
[C4mim][PF6] has the lowest j of the four RTILs.
Based on the Walden Rule, we expect a decrease in g
to be accompanied by an inversely proportional increase
in K. Furthermore, given the magnitude of the change in
g with increasing wH O we expect that even trace levels of
2
water will significantly increase K (and, therefore, j) of
(RTIL + water). To test this prediction we measured j
as a function of wH O for ([C4mim][Tf2N] + water); mea-
surements were made at temperatures of (288.15, 293.15,
2
and 298.15) K and wH O values of 1 Æ 10ꢀ5, 8 Æ 10ꢀ5
,
2
1.02 Æ 10ꢀ3, and 8.85 Æ 10ꢀ3. As expected, j increases dra-
matically as wH O increases, table 2. Figure 4 shows how
2
50
[C4mim][Tf2N]
40
30
20
10
0
3.2. Electrolytic conductivities of
([C4mim][Tf2N] + water)
The viscosity of (RTIL + water) is known to depend
strongly on wH O [15,21,22,26,27]. Figure 3, which was
2
constructed from data from reference [21], shows g
-10
decreasing rapidly as wH O increases for ([C4mim]-
2
285
290
295
300
[Tf2N] + water) at 293.15 K. The data for figure 3 were
measured at atmospheric pressure with an Ubbelohde
capillary viscometers [21]. The capillaries were open to
the atmosphere, but atmospheric moisture was effec-
tively excluded by connecting drying tubes (filled with
anhydrous calcium sulfate) to the openings. The water
content was determined before and after each viscosity
measurement by coulometric Karl Fischer titration.
T / K
FIGURE 4. Plot of the relative deviations of the electrolytic conduc-
tivities (j) from the electrolytic conductivities given by the correlation
(jcorr) in figure 1 for dried [C4mim][Tf2N]. }, ([C4mim][Tf2N] + water)
with
wH = 8.85 Æ 10ꢀ3
;
,
([C4mim][Tf2N] + water)
with
2O
wH = 1.02 Æ 10ꢀ3; +, ([C mim][Tf N] + water) with wH = 8 Æ 10ꢀ5
;
2O
2O
4
2
s, dried [C4mim][Tf2N] (for which wH = 10ꢀ5); n, data point from
reference [14]; ---, correlation from reference [28].
2O

574
J.A. Widegren et al. / J. Chem. Thermodynamics 37 (2005) 569–575
TABLE 3
Electrolytic conductivities (j) of ([C4mim][Tf2N] + water) at 293.15 K, the relative change in j, the molar conductivity K, and the walden, product
(K Æ g), all as functions of the and mass fraction of water (wH
)
2O
wH O
j/(S Æ mꢀ1
)
100 Æ (j ꢀ jdried)/jdried
0.0
K/(S Æ m2 Æ molꢀ1 a
)
K Æ g/S Æ kg Æ m Æ sꢀ1 Æ molꢀ1b
2
1 Æ 10ꢀ5
0.331
0.332
0.345
0.449
9.62 Æ 10ꢀ5
9.67 Æ 10ꢀ5
1.00 Æ 10ꢀ4
1.31 Æ 10ꢀ4
6.11 Æ 10ꢀ6
6.12 Æ 10ꢀ6
6.09 Æ 10ꢀ6
5.85 Æ 10ꢀ6
8 Æ 10ꢀ5
0.4
4.1
1.02 Æ 10ꢀ3
8.85 Æ 10ꢀ3
35.8
The molar conductivity (K) and Walden product (K Æ g) are given as a function of wH O
.
2
a
The molar conductivity (K) was calculated using a literature density value [29].
b
Viscosities (g) were calculated from the correlation given in figure 3.
the conductivity of ([C4mim][Tf2N] + water) deviates
from the correlated conductivity data for the dried
wH O was greater for more viscous RTILs. That is, rela-
tively larger changes in g were observed for more viscous
2
[C4mim][Tf2N]. At 293.15 K, for example, (jexp
ꢀ
RTILs for a given value of wH O. Given that, and the
2
j
corr)/jcorr = 0.358 when the wH O = 8.85 Æ 10ꢀ3 (table 3).
adherence to the Walden Rule exhibited by
([C4mim][Tf2N] + water), we predict that larger relative
changes in j will be observed for more viscous RTILs
2
Put another way, increasing wH O by only 2.5 Æ 10ꢀ4 in-
2
creases j by 1%.
Figure 4 includes conductivity data from the litera-
for a given wH O and smaller relative changes in j will
2
ˆ
ture. For [C4mim][Tf2N], Bonhote et al. reported [14]
be observed for less viscous RTILs for a given wH O.
2
that j = 0.39 S Æ mꢀ1 at 293 K (with an estimated uncer-
tainty of U = 0.05 j), which is about 18% higher than
the value reported herein for dried [C4mim][Tf2N]. Com-
paring the literature value of j to the mixture data re-
An important conclusion from this work is that accu-
rate measurement of j requires careful control of the
water content of the RTIL. Because of the magnitude
of the effect of wH O on j, conductivity data are suspect
2
ported herein suggests that wH O for the literature
unless wH O is determined before and after the measure-
2
2
sample of [C4mim][Tf2N] was about 4 Æ 10ꢀ3. Figure 4
also shows a portion of the correlation of j as a function
of T for [C4mim][Tf2N] that was reported by Watanabe
and co-workers [28]. That correlation is lower than the
correlation reported herein, but it is not clear if the dif-
ference is significant because no estimate of uncertainty
was given.
ment of j. Checking wH O after measuring j is important
2
because it is the only way to verify that wH O has not
2
changed significantly during the measurement. To our
knowledge, this is the first time such a precaution has
been reported when measuring j for an RTIL. We rec-
ommend that this precaution be generally adopted.
It is of interest to know if the Walden Rule applies to
(RTIL + water). Density data are not available for
([C4mim][Tf2N] + water), but we can still estimate K
Acknowledgements
by assuming that the density is not a function of wH O
(this assumption is probably correct to within about
We thank Mark Muldoon, working in Joan Bren-
neckeꢀs laboratory at the University of Notre Dame,
for his careful synthesis of the [C6mim][Tf2N]. We also
thank Dwain Diller for reading the manuscript and for
useful discussions. Eric Saurer gratefully acknowledges
the support of a NIST Boulder Summer Undergraduate
Research Fellowship.
2
1% for these small values of wH O [15]). Hence, the val-
2
ues of K in table 3 were calculated from a literature va-
lue for the density of [C4mim][Tf2N] at 293.15 K and
0.1 MPa, 1442.5 kg Æ mꢀ3 [29]. The Walden products
(K Æ g) shown in table 3 were then calculated from the
correlated value of g from figure 3. Over the studied
range of compositions, K Æ g ranges from (6.12 Æ 10ꢀ6
to 5.85 Æ 10ꢀ6) S Æ kg Æ m Æ sꢀ1 Æ molꢀ1, a change of only
about 4% (compared to a 36% change in j). This is
excellent evidence that the change in g is indeed primar-
ily responsible for the change in j. The small change in
References
[1] T. Welton, Chem. Rev. 99 (1999) 2071–2083.
[2] P.C. Trulove, R.A. Mantz, in: P. Wasserscheid, T. Welton (Eds.),
Ionic Liquids in Synthesis, Wiley-VCH, Weinheim, 2003, pp. 103–
126.
K Æ g with increasing wH O could be due to a relatively
minor effect (such as a change in the extent of ion pair-
2
[3] J.S. Wilkes, ACS Sympos. Ser. 818 (2002) 214–229.
[4] J. DuPont, R.F. de Souza, P.A.Z. Suarez, Chem. Rev. 102 (2002)
3667–3692.
ing with increasing wH O). Given that K Æ g is fairly inde-
2
pendent of wH O, it is not surprising that the reported
2
[5] K.N. Marsh, J.A. Boxall, R. Lichtenthaler, Fluid Phase Equilib.
219 (2004) 93–98.
[30]
value
for
[C4mim][Tf2N]
at
293
K,
K Æ g = 5.952 Æ 10ꢀ6 S Æ kg Æ m Æ sꢀ1 Æ molꢀ1, is within the
range of our mixture values.
[6] B.D. Fitchett, T.N. Knepp, J.C. Conboy, J. Electrochem. Soc.
151 (2004) E219–E225.
Earlier work [21,22] reporting g as a function of wH O
showed that the relative decrease in g with increasing
´
[7] B. Garcia, S. Lavallee, G. Perron, C. Michot, M. Armand,
Electrochim. Acta 49 (2004) 4583–4588.
2

J.A. Widegren et al. / J. Chem. Thermodynamics 37 (2005) 569–575
575
[8] H.A. Every, A.G. Bishop, D.R. MacFarlane, G. Ora¨dd, M.
Forsyth, Phys. Chem. Chem. Phys. 6 (2004) 1758–1765.
[9] J.N. Barisci, G.G. Wallace, D.R. MacFarlane, R.H. Baughman,
Electrochem. Commun. 6 (2004) 22–27.
[21] J.A. Widegren, A. Laesecke, J.W. Magee, Chem. Commun.
(2005) 1610–1612.
[22] S. Pandey, K.A. Fletcher, S.N. Baker, G.A. Baker, Analyst 129
(2004) 569–573.
[10] A. Noda, K. Hayamizu, M. Watanabe, J. Phys. Chem. B 105
(2001) 4603–4610.
[23] W. Xu, E.I. Cooper, C.A. Angell, J. Phys. Chem. B 107 (2003)
6170–6178.
´
[24] E. Juhasz, K.N. Marsh, in: K.N. Marsh (Ed.), Recommended
[11] H. Every, A.G. Bishop, M. Forsyth, D.R. MacFarlane, Electro-
chim. Acta 45 (2000) 1279–1284.
Reference Materials for the Realization of Physicochemical
Properties, Blackwell Scientific Publications, Oxford, 1987, pp.
371–378.
[12] H. Matsumoto, M. Yanagida, K. Tanimoto, M. Nomura, Y.
Kitagawa, Y. Miyazaki, Chem. Lett. (2000) 922–923.
[13] A.B. McEwen, H.L. Ngo, K. LeCompte, J.L. Goldman, J.
Electrochem. Soc. 146 (1999) 1687–1695.
[25] R.A. Robinson, R.H. Stokes, Electrolyte Solutions, second
ed., Butterworths Scientific Publications, London, 1959 (Chap-
ter 5).
ˆ
[14] P. Bonhote, A.-P. Dias, N. Papageorgiou, K. Kalyanasundaram,
M. Gra¨tzel, Inorg. Chem. 35 (1996) 1168–1178.
[15] K.R. Seddon, A. Stark, M.-J. Torres, Pure Appl. Chem. 72 (2000)
2275–2287.
[26] J.G. Huddleston, A.E. Visser, W.M. Reichert, H.D. Willauer,
G.A. Broker, R.D. Rogers, Green Chem. 3 (2001) 156–164.
[27] C.F. Poole, B.R. Kersten, S.S.J. Ho, M.E. Coddens, K.G. Furton,
J. Chromatogr. 352 (1986) 407–425.
[16] C.D. Tran, S.H. De Paoli Lacerda, D. Oliveira, Appl. Spectrosc.
57 (2003) 152–157.
[28] H. Tokuda, K. Hayamizu, K. Ishii, M.A.B.H. Susan, M.
Watanabe, J. Phys. Chem. B 108 (2004) 16593–16600.
[29] M. Krummen, P. Wasserscheid, J. Gmehling, J. Chem. Eng. Data
47 (2002) 1411–1417.
[17] L. Cammarata, S.G. Kazarian, P.A. Salter, T. Welton, Phys.
Chem. Chem. Phys. 3 (2001) 5192–5200.
[18] J.L. Anthony, E.J. Maginn, J.F. Brennecke, J. Phys. Chem. B 105
(2001) 10942–10949.
[30] R.A. Mantz, P.C. Trulove, in: P. Wasserscheid, T. Welton (Eds.),
Ionic Liquids in Synthesis, Wiley-VCH, Weinheim, 2003, pp. 56–
68.
[19] H. Xu, D. Zhao, P. Xu, F. Liu, G. Gao, J. Chem. Eng. Data 50
(2005) 133–135.
[20] J. Zhang, W. Wu, T. Jiang, H. Gao, Z. Liu, J. He, B. Han, J.
Chem. Eng. Data 48 (2003) 1315–1317.
JCT 05-36