Physicochemical properties and structures of room temperature ionic liquids. 1. Variation of anionic species

Article abstract of DOI:10.1021/jp047480r

Room-temperature ionic liquids (RTILs) based on 1-butyl-3-methylimidazolium ([bmim]) with a variety of fluorinated anions were prepared, and the thermal behavior, density, viscosity, self-diffusion coefficients of the cations and anions, and ionic conductivity were measured over a wide temperature range. The temperature dependencies of the self-diffusion coefficient, viscosity, ionic conductivity, and molar conductivity have been fitted to the Vogel-Fulcher-Tamman equation, and the best-fit parameters for the self-diffusion coefficient, viscosity, ionic conductivity, and molar conductivity have been estimated, together with the linear fitting parameters for the density. The self-diffusion coefficients determined for the individual ions by pulsed-field-gradient spin-echo NMR method exhibit higher values for the cation compared with the anion over a wide temperature range, even if its radius is larger than that of the anionic radii. The summation of the cationic and anionic diffusion coefficients for the RTILs follows the order [bmim][(CF₃SO₂N] > [bmim][CF ₃CO₂] > [bmim][CF₃SO₃] > [bmim][BF₄] > [bmim][(C₂F₅SO ₂)₂N] > [bmim][PF₆] at 30?°C, and the order of the diffusion coefficients greatly contrasts to the viscosity data. The ionic association is proposed from the results of the ratios of molar conductivity obtained from impedance measurements to that calculated by the ionic diffusivity using the Nernst-Einstein equation. The ratio for the ionic liquids follows the order [bmim][PF₆] > [bmim][BF₄] > [bmim][(C₂F₅SO₂)₂N] > [bmim][(CF₃SO₂)₂N] > [bmim][CF ₃SO₃] > [bmim][CF₃CO₂] at 30?°C and provides quantitative information on the active ions contributing to ionic conduction in the diffusion components.

Full text of DOI:10.1021/jp047480r

J. Phys. Chem. B 2004, 108, 16593-16600
16593
Physicochemical Properties and Structures of Room Temperature Ionic Liquids. 1.
Variation of Anionic Species
†
‡
†
†,§
Hiroyuki Tokuda, Kikuko Hayamizu, Kunikazu Ishii, Md. Abu Bin Hasan Susan, and
Masayoshi Watanabe*,†
Department of Chemistry and Biotechnology, Yokohama National UniVersity, and CREST-JST, 79-5 Tokiwadai,
Hodogaya-ku, Yokohama 240-8501, Japan, and National Institute of AdVanced Industrial Science and
Technology, AIST Tsukuba Centre 5, Tsukuba 305-8565, Japan
ReceiVed: June 11, 2004; In Final Form: July 27, 2004
Room-temperature ionic liquids (RTILs) based on 1-butyl-3-methylimidazolium ([bmim]) with a variety of
fluorinated anions were prepared, and the thermal behavior, density, viscosity, self-diffusion coefficients of
the cations and anions, and ionic conductivity were measured over a wide temperature range. The temperature
dependencies of the self-diffusion coefficient, viscosity, ionic conductivity, and molar conductivity have been
fitted to the Vogel-Fulcher-Tamman equation, and the best-fit parameters for the self-diffusion coefficient,
viscosity, ionic conductivity, and molar conductivity have been estimated, together with the linear fitting
parameters for the density. The self-diffusion coefficients determined for the individual ions by pulsed-field-
gradient spin-echo NMR method exhibit higher values for the cation compared with the anion over a wide
temperature range, even if its radius is larger than that of the anionic radii. The summation of the cationic
and anionic diffusion coefficients for the RTILs follows the order [bmim][(CF
3
SO
2
)
2
N] > [bmim][CF
3 2
CO ]
>
[bmim][CF SO ] > [bmim][BF ] > [bmim][(C SO N] > [bmim][PF ] at 30 °C, and the order of the
3
3
4
2
F
5
2
)
2
6
diffusion coefficients greatly contrasts to the viscosity data. The ionic association is proposed from the results
of the ratios of molar conductivity obtained from impedance measurements to that calculated by the ionic
diffusivity using the Nernst-Einstein equation. The ratio for the ionic liquids follows the order [bmim][PF
[bmim][BF ] > [bmim][(C SO N] > [bmim][(CF SO N] > [bmim][CF SO ] > [bmim][CF CO ] at
0 °C and provides quantitative information on the active ions contributing to ionic conduction in the diffusion
components.
6
]
>
3
4
F
2 5
2
)
2
3
2
)
2
3
3
3
2
Introduction
dinium, ammonium, sulfonium and phosphonium derivatives,
and bulky and soft anions, such as [BF4], [PF6], [CF3SO3], and
Air- and water-stable room temperature ionic liquids (RTILs)
are comprised entirely of ions and are liquids at ambient or far
below ambient temperature. They possess unique physicochem-
ical properties, such as negligible vapor pressure, nonflamma-
bility, high ionic conductivity, and high thermal, chemical, and
electrochemical stability, which make them so distinct from
2
b,16-21
[(CF3SO2)2N].
Since numerous combinations of the
cationic and anionic structures are possible, physicochemical
properties of the ionic liquids can be easily tuned simply by
changing the structure of the component ions. Proper design of
the ionic liquids has, thus, been the most fascinating domain of
the current research. For the design, however, it is necessary to
accumulate data on the basic physical and chemical properties.
The high ionic concentration, resulting in the high Coulombic
interactions, implies that the differences in the ionic state and
ion dynamics significantly contribute to the difference in the
physicochemical properties of RTILs. However, microscopic
information about the ionic state and ion dynamics has been
insufficient. It has only been recently recognized that pulsed-
field-gradient spin-echo (PGSE) nuclear magnetic resonance
1
conventional molecular liquids. It is not surprising that ionic
liquids, while attracting significant attention of academicians
in multidisciplinary areas, have been considered as new solvents
and materials for practical applications. The low volatility,
together with the ease of handling and high ionic concentration,
have led the RTILs to be used as novel organic and inorganic
2
reaction media and separation solvents. Since the ionic liquids
are essentially ionic conductors, their utilization as novel
electrolytes for electrochemical devices, such as lithium second-
ary batteries, electric double layer capacitors, dye-sensitized
solar cells, fuel cells, and actuators, has also been the subject
(
NMR) method can be applied for the measurement of the self-
22-24
diffusion coefficients of the individual ionic species.
We
of intense study.3
-10
have successfully applied the technique for the determination
of the cationic and anionic self-diffusion coefficients of certain
RTILs of 1-ethyl-3-methylimidazolium ([emim]) and 1-butylpy-
ridinium ([bpy]) cations with [BF4] and [(CF3SO2)2N] anions.25
On the basis of the analysis of the ionic diffusivity, viscosity,
and conductivity, we were able to obtain useful information on
the diffusion mechanism and ion transport, which further
indicated the ion association/dissociation in the ionic liquids.
However, since the study involved ionic liquids only from two
different cationic and anionic species, a clear correlation between
the physicochemical properties and the structural difference
These developments also include hybrid
materials based on the RTILs with polymers and nano-
carbons.11-15
The RTILs are typically comprised of combinations of
organic cations, such as imidazolium, pyridinium, pyrroli-
*
To whom correspondence should be addressed. E-mail: mwatanab@
ynu.ac.jp. Fax: +81-45-339-3955.
†
Yokohama National University, and CREST-JST.
National Institute of Advanced Industrial Science and Technology.
‡
§
Permanent address: Department of Chemistry, University of Dhaka,
Dhaka 1000, Bangladesh.
1
0.1021/jp047480r CCC: $27.50 © 2004 American Chemical Society
Published on Web 09/29/2004

16594 J. Phys. Chem. B, Vol. 108, No. 42, 2004
Tokuda et al.
could not be made. This necessitates a comprehensive study
on the effect of the ionic structures on the physicochemical
characteristics of a wide variety of ionic liquids.
aluminum pans in the dry glovebox. The samples were heated
to 80 °C followed by cooling to -150 °C and then heated at a
-
1
cooling and heating rate of 10 °C min . The glass transition
temperature (Tg) and the melting point (Tm) were determined
from the DSC thermograms during the programmed reheating
steps. Thermogravimetric measurements were conducted on a
Seiko Instruments thermo-gravimetry/differential thermal ana-
lyzer (TG-DTA 6200) from room temperature to 550 °C at a
The present studies have been initiated to present reliable
data for the physicochemical properties of a series of ionic
liquids, which include thermal behavior, density, self-diffusion
coefficient of the cation and anion, viscosity, and ionic
conductivity over a wide temperature range. The ultimate goal
is, however, to know the relationship between the ionic
structures and the physicochemical properties, in order to
establish principles for the molecular design of the RTILs having
desirable properties. In this study, a series of RTILs with a fixed
cation, 1-butyl-3-methylimidazolium ([bmim]), with different
anions, [(C2F5SO2)2N], [(CF3SO2)2N], [CF3SO3], [PF6], [CF3-
CO2], and [BF4], were prepared. The physicochemical properties
were investigated, with focusing on the ion transport properties
depending on the change in the anionic structures, to know the
relationships between the ionic diffusivity, viscosity, and molar
conductivity.
-
1
heating rate of 10 K min , unless otherwise noted, under
nitrogen atmosphere with open alumina (Al2O3) pans, and for
some cases also with aluminum (Al) pans.
Density. The density measurement was performed using a
thermo-regulated density/specific gravity meter DA-100 (Kyoto
Electronics Manufacturing Co. Ltd.). The measurements were
conducted in the range of 15 to 40 °C.
Self-Diffusion Coefficients. The PGSE-NMR measurements
were conducted by using a JEOL JNM-AL 400 spectrometer
with a 9.4 T narrow-bore superconducting magnet equipped with
a JEOL pulse field gradient probe and a current amplifier. The
-
1
sine gradient pulse, providing gradient strength up to 12 Tm ,
was used throughout the measurements in this study. The self-
diffusion coefficients were measured using a simple Hahn spin-
echo sequence, (i.e., 90°-τ-180°-τ-Acquisition), incorporating
a gradient pulse in each τ period. The free diffusion echo signal
Experimental Section
Synthesis. The syntheses of the RTILs used in this study
were based on a metathesis reaction of 1-butyl-3-methylimida-
zolium chloride ([bmim][Cl]) and alkali metal or silver salts
27
attenuation, E, is related to the experimental parameters by
with different anions, following a slight modification of the
procedure reported by Welton and co-workers.26 The [bmim]-
2
2
2
2
ln(E) ) ln(S/Sg)0) ) -γ g Dδ (4∆ - δ)/π
(1)
[Cl] was prepared by reacting 1-methylimidazole with 1-chlo-
robutane in cyclohexane under a reflux condition, and the
product was purified by recrystallization from acetonitrile/ethyl
acetate solution to yield a white crystalline solid. For the
preparation of [bmim][(C2F5SO2)2N] and [bmim][(CF3SO2)2N],
the anion exchange reaction of [bmim][Cl] was made by adding
where S is the spin-echo signal intensity, δ is the duration of
the field gradient with magnitude g, γ is the gyromagnetic ratio,
D is the self-diffusion coefficient, and ∆ is the interval between
two gradient pulses. Unless otherwise noted, the measurements
were conducted with ∆ ) 50 ms. A recycle delay sufficient in
1
.2 equivalent amount of LiN(SO2C2F5)2 and LiN(SO2CF3)2 in
water, followed by repeatedly washing with water. For [bmim]-
CF3SO3], [bmim][PF6] and [bmim][BF4], 1.2 equivalent amount
allowing full relaxation (i.e., >5T ) was used between each
1
transition. The measurements for the cationic and anionic self-
1
[
diffusion coefficients in each RTIL were made by H (399.7
19
of sodium salts of the corresponding anions and [bmim][Cl]
were reacted in dichloromethane and filtered off, and then the
organic solutions were extracted several times with a small
amount of water to remove [bmim][Cl], as already described.
The ionic liquid, [bmim][CF3CO2], was prepared by the reaction
of equimolar amounts of [bmim][Cl] and CF3CO2Ag in water.
The precipitate, AgCl, was filtered off, and the filtrate was
concentrated and passed through a silica gel column to
completely remove AgCl. All of the ionic liquids were finally
dehydrated under high vacuum with heating over 48 h and were
stored in an argon atmosphere glovebox (VAC, [O2] < 1 ppm,
MHz) and F (376.1 MHz) nuclei, respectively. The measure-
ment was performed at a range of temperatures with gradual
cooling from +80 to -10 °C with the samples thermally
equilibrated at each temperature for 30 min prior to the
measurements. The samples were inserted into a 5 mm (o.d.)
NMR microtube (BMS-005J, Shigemi, Tokyo) in the glovebox
to a height of 5 mm. The sample height of some of the ionic
liquids, which were susceptible to convection due to their low
viscosities, was maintained at 2 mm, so that the relatively small
sample volume could minimize the deleterious effect of convec-
28
tion in the diffusion measurements. The reliability of the self-
diffusion coefficients for some RTILs was crosschecked by
replicate measurements by using a different system, a JEOL
GSH-200 spectrometer with a 4.7 T wide-bore magnet controlled
[
H2O] < 1 ppm). The structures of the ionic liquids were
1
13
identified by H and C NMR and fast atom bombardment mass
spectra (FAB-MS) using a JEOL JMS-AX500 mass spectrom-
eter. Chloride content in the hydrophilic [bmim][CF3SO3],
2
4,28
by TecMag Apollo.
[
bmim][CF3CO2], and [bmim][BF4] was maintained at least
Viscosity. The viscosity measurement was carried out with
a Toki RE80 cone-plate viscometer under nitrogen atmosphere.
The temperature was controlled in the range of +80 to -10 °C
by a Thomas TRL-108H circulation-type thermo-regulated water
bath.
-
1
below the solubility limit of AgCl in water (1.4 mgL ), as
checked by adding an AgNO3 solution. For the hydrophobic
ionic liquids, [bmim][(CF3SO2)2N], [bmim][(C2F5SO2)2N], and
[bmim][PF6], which are not miscible with water, chloride in
the aqueous phases in contact with the ionic liquids could not
be detected by using the AgNO3 solution. Water content of all
of the RTILs, determined by Karl Fischer titration, was below
Electrochemical Measurements. The bulk ionic conductivity
was measured by complex impedance measurements, using a
computer-controlled Hewlett-Packard 4192A LF impedance
analyzer over the frequency range from 5 Hz to 13 MHz. The
RTILs were filled in a conductivity cell constructed with
platinized platinum electrode cells (TOA Electronics, CG-511B)
with the cell constants of ca. 1 cm determined by a standard
KCl aqueous solution. The measurement was carried out at
controlled temperatures with cooling from +100 to -10 °C.
4
0 ppm. For none of the RTILs, any other impurities such as
alkali metal and silver derived from the starting materials or
byproducts could be detected in the FAB-MS measurements.
Thermal Analysis. Differential scanning calorimetry (DSC)
was carried out on a Seiko Instruments DSC 220C under
nitrogen atmosphere. The samples were tightly sealed in
-
1

Room Temperature Ionic Liquids
J. Phys. Chem. B, Vol. 108, No. 42, 2004 16595
TABLE 1: Molecular Weight and Thermal Properties
MW/gmol-
1
T
g
/°Ca
T
m
/°Ca
T
d
/°Cb
[
[
[
[
[
[
bmim][(C
bmim][(CF
bmim][CF SO
2
F
5
SO
2
)
2
2
N]
N]
519.4
419.4
288.2
284.2
252.2
226.0
-84
-87
402
423
409
433
176
425
3
SO
2
)
-3
17
10
3
3
]
bmim][PF
bmim][CF
bmim][BF
6
]
-77
-78
-83
3
CO
2
]
4
]
a
Onset temperatures of a heat capacity change (T
g
), and an
endothermic peak (T
Onset temperatures of mass loss (T ).
m
) determined by differential scanning calorimetry.
b
d
For the temperature range between 40 and 10 °C, the measure-
ment was conducted using a Lauda RE104 thermo-regulated
water bath, whereas the other measurements were conducted
in a Tabai Espec SU-220 temperature chamber.
Results
Figure 1. Density of [bmim] based ionic liquids as a function of
temperature.
Thermal Property and Density. Impurities in the RTILs
may significantly affect their properties, such as thermal
property, viscosity, conductivity, and diffusivity.29 However,
chloride and water content in the system were in the acceptable
limits, and no other impurities could be detected; therefore, the
physicochemical properties of the ionic liquids in this study have
no considerable influence from the impurities. The DSC results
during reheating scans are listed in Table 1. At the scanning
TABLE 2: Density Equation Parameters and Molar
Concentration at 30 °C (M30) Density Equation
F ) b - aT
-
4
2
M
30/10-3
mol cm
a/10
b/
R /
g cm K-1 g cm
10.3 1.81
9.40
-
3
-3
10
-1
-3
[bmim][(C
2
F
5
SO
2
)
2
2
N]
N]
9.99
9.99
10.0
9.96
9.98
9.98
2.91
3.42
4.49
4.80
4.81
5.30
-
1
[bmim][(CF
3
SO
SO
2
)
1.72
1.54
1.63
1.44
1.42
rate of 10 °C min , each RTIL in this study exhibits a heat
capacity change in the DSC thermogram, corresponding to a
glass transition temperature (Tg), with the exception of [bmim]-
[
[
[
[
bmim][CF
3
3
]
8.00
8.69
7.54
7.26
bmim][PF
bmim][CF
bmim][BF
6
]
3
2
CO ]
[
CF3SO3]. No endothermic peak, corresponding to a melting
point (Tm), was observed for [bmim][(C2F5SO2)2N], [bmim]-
CF3CO2], and [bmim][BF4] in the cycle of the measurements.
4
]
[
2
. The difference in the anions with the same [bmim] cation in
Although [bmim][CF3SO3] did not exhibit the Tg at the scan
rate of the measurement, the crystallization exotherm in the
cooling scan was found to be significantly lower than the Tm
the ionic liquids leads to different density, and the order of
density approximately follows the decreasing order of the
molecular weight of the anions. The calculated molar concentra-
tion of the RTILs at 30 °C (M30) based on the density value
and molecular weight is also tabulated in Table 2. The molar
concentration increases with decrease in the molecular weight
(by ca. 40 °C). These results indicate the very slow crystalliza-
tion rate and the fairly stable supercooling state for the ionic
liquids. Thermogravimetric results show that most of the ionic
liquids have excellent short-term thermal stability up to 400 °C
-1
of the anions, and the values range from 2.91 to 5.30 mol L .
-
1
at a scan rate of 10 °C min , except for [bmim][CF3CO2] at
76 °C (Table 1). However, the thermal stability shows scan
Self-Diffusion Coefficient and Viscosity. The PGSE-NMR
method for the determination of the self-diffusion coefficient
allows evaluation of the diffusivity of ions without the use of
any additional probe molecules, which might affect the diffusion
of the ions. Since the RTILs used in this study include NMR-
1
rate dependency, and a faster heating scan has been found to
give higher thermal stability. For instance, the onset temperatures
of mass loss (Td) of [bmim][PF6] and [bmim][CF3CO2] at a scan
-
1
rate of 10 °C min are higher by about 30 and 10 °C,
1
19
sensitive H and F nuclei in the cation and anion, respectively,
each of the cationic and anionic self-diffusion coefficients could
be independently determined. Figure 2 shows the temperature
dependency of self-diffusion coefficients of the cation (Dcation)
and anion (Danion) for [bmim][(CF3SO2)2N], [bmim][CF3SO3],
and [bmim][(C2F5SO2)2N] (Figure 2a) and [bmim][CF3CO2],
-
1
respectively, compared to those of 3 °C min . Similary, the
thermal stability of [bmim][BF4] and [bmim][PF6] showed
strong dependence on the type of the used sample pans by about
5
0 and 80 °C, respectively, where the Td using an alumina pan
was higher than those using an aluminum pan (data not shown).
This can be attributed to enhanced thermal decomposition
catalyzed by aluminum metal.30 The Td for the RTILs apparently
follows the order of [PF6] > [BF4] ≈ [(CF3SO2)2N] > [CF3-
SO3] > [(C2F5SO2)2N] . [CF3CO2]. The thermal stability of
the RTILs, therefore, depends on the anionic species. The
pyrolysis of the imidazolium based salts have been reported to
proceed most likely via SN2 process, and therefore, change in
the basicity and/or nucleophilicity of the anions is likely to bring
[bmim][BF4], and [bmim][PF6] (Figure 2b). The measuring time
(∆ in eq 1) dependency of the self-diffusion coefficients for
some of the ionic liquids was checked, and in the range of ∆
from 20 to 100 ms, the self-diffusion coefficients did not show
any notable dependency for [bmim][(CF3SO2)2N], [bmim][PF6],
and [bmim][BF4], which suggests that the ion diffusion is the
Fickian diffusion in the homogeneous fluid and the convection
effects can be disregarded in the temperature range studied under
the experimental conditions. The temperature dependency of
the diffusion coefficients in each case exhibits convex curved-
profiles; therefore, experimental data were fitted with the
3
1
about a change in the thermal stability of the RTILs. The
different Td for the variation in the nucelophilicity of the RTILs
in this study is, therefore, reasonable and in agreement with
literature.
3
2
Vogel-Fulcher-Tamman (VFT) equation for diffusivity
Temperature dependency of density for the RTILs, as depicted
in Figure 1, shows linear decrease with increasing temperature.
The best-fit parameters for the linear fitting are listed in Table
D ) D exp[-B/(T - T )]
(2)
0
0

16596 J. Phys. Chem. B, Vol. 108, No. 42, 2004
Tokuda et al.
Figure 3. Temperature dependence of simple summation of the cationic
and anionic self-diffusion coefficients (Dcation + Danion) for [bmim] based
ionic liquids.
TABLE 3: VFT Equation Parameters of Self-Diffusion
Coefficient Data
D ) D
0
0
exp[-B/(T - T )
/10- cm-2 s-1
4
B/10 K
2
D
0
0
T /K
[bmim][(C F
2 5
2 2
SO ) N]
cation
anion
1.7 ( 0.2
1.9 ( 0.1
3.5 ( 0.3
9.95 ( 0.43 160 ( 3
11.09 ( 0.17 154 ( 1
10.43 ( 0.27 157 ( 2
cation+anion
[bmim][(CF SO
3
2
)
2
N]
cation
anion
1.1 ( 0.1
1.3 ( 0.3
2.3 ( 0.3
8.45 ( 0.37 157 ( 3
9.33 ( 0.67 152 ( 6
8.84 ( 0.45 155 ( 4
cation+anion
[bmim][CF SO
3
3
]
cation
anion
cation+anion
[bmim][PF
cation
anion
0.80 ( 0.01
1.8 ( 0.2
2.3 ( 0.2
8.29 ( 0.43 163 ( 4
10.95 ( 0.34 147 ( 3
9.42 ( 0.23 156 ( 2
Figure 2. Temperature dependence of self-diffusion coefficients of
the cation and anion for (a) [bmim][(C SO N], [bmim][(CF SO N],
and [bmim][CF SO ] and (b) [bmim][PF ], [bmim][CF CO ], and
bmim][BF ].
F
2 5
2
)
6
2
3
)
2 2
6
]
3
3
3
2
[
4
1.5 ( 0.3
1.8 ( 0.3
3.2 ( 0.5
9.87 ( 0.54 169 ( 4
10.84 ( 0.54 165 ( 4
10.32 ( 0.44 167 ( 3
2
-1
cation+anion
where the constants D0 (cm s ), B (K), and T0 (K) are
adjustable parameters. The best-fit parameters of the ionic
diffusivity are summarized in Table 3. The dashed and solid
lines in Figure 2 are the calculated curves by using the best-fit
parameters and eq 2. It is worthwhile to describe that the D0
and the B parameters of the anion are larger than those of the
cation for all of the RTILs. The experimental ionic self-diffusion
coefficients for [bmim] and [PF6] in the [bmim][PF6] at 25 °C,
evaluated from the data using the best-fit parameters of the VFT
[bmim][CF CO ]
3
2
cation
anion
cation+anion
[bmim][BF
cation
anion
cation+anion
1.4 ( 0.1
1.9 ( 0.2
3.3 ( 0.3
9.19 ( 0.29 160 ( 2
10.28 ( 0.40 155 ( 3
9.70 ( 0.29 158 ( 2
4
]
1.4 ( 0.1
2.8 ( 0.4
4.0 ( 0.4
9.35 ( 0.28 162 ( 3
11.08 ( 0.47 153 ( 3
10.21 ( 0.31 158 ( 2
where η0 (mPas), B (K), and T0 (K) are constants. The best-fit
parameters for viscosity are tabulated in Table 4. The macro-
scopic viscosity values at 30 °C follows the order [bmim][PF6]
> [bmim][(C2F5SO2)2N] > [bmim][BF4] > [bmim][CF3SO3]
-
8
2
-1
equation (eq 2), are 6.8 and 4.0 × 10 cm s , respectively.
The simulated values for the self-diffusion coefficients at 25
33
°
C, reported by Morrow et al. by a molecular dynamic (MD)
-
8
2 -1
study, are 9.7 ( 4.1 and 8.8 ( 4.2 × 10 cm s , respectively,
>
[bmim][CF3CO2] > [bmim][(CF3SO2)2N], which well con-
which are fairly close to our experimental results.
trasts with that of the ionic self-diffusion coefficients.
The simple sum of the cationic and anionic self-diffusion
coefficients (Dcation + Danion) for these RTILs is shown in Figure
Conductivity. All of the six [bmim]-based ionic liquids have
relatively high ionic conductivities (σ), which exhibit temper-
ature dependency, as shown in Figure 5. The VFT equation for
conductivity is
3, and the best-fit parameters of the VFT equation are also listed
in Table 3. The sum of the cationic and anionic diffusion
coefficients for the ionic liquids follows the order [bmim][(CF3-
SO2)2N] > [bmim][CF3CO2] > [bmim][CF3SO3] > [bmim]-
[
BF4] > [bmim][(C2F5SO2)2N] > [bmim][PF6] at 30 °C.
σ ) σ exp[-B/(T - T )]
(4)
0
0
The temperature dependency of the viscosity (η) for the
RTILs and the profiles fitted to the VFT equation are depicted
in Figure 4. The corresponding VFT equation is
-1
where σ0 (Scm ), B (K), and T0 (K) are constants. The best-fit
parameters are shown in Table 5. Since the molar concentration
of these RTILs is strongly dependent on the anionic structure
(Table 2), the molar conductivity has also been calculated from
η ) η exp[B/(T - T )]
(3)
0
0

Room Temperature Ionic Liquids
J. Phys. Chem. B, Vol. 108, No. 42, 2004 16597
Figure 4. Viscosities of [bmim] based room-temperature ionic liquids
as a function of temperature.
Figure 6. Molar conductivity of [bmim] based ionic liquids: (a)
obtained from ionic conductivity and molar concentration; (b) calculated
from ionic self-diffusion coefficients and the Nernst-Einstein equation.
Figure 5. Temperature dependence of ionic conductivity for [bmim]
based ionic liquids.
TABLE 6: VFT Equation Parameters of Molar
Conductivity Data
TABLE 4: VFT Equation Parameters of Viscosity Data
η ) η
0
0
exp[B/(T - T )]
Λ
0
/102
1
2
2
-1
2
η
0
/10- mPas
B/10 K
T
0
/K
Scm mol
B/10 K
0
T /K
[
[
[
[
[
[
bmim][(C
bmim][(CF
bmim][CF SO
2
F
5
SO
2
)
2
2
N]
N]
1.7 ( 0.2
2.5 ( 0.2
3.7 ( 0.6
3.6 ( 0.5
1.1 ( 0.2
2.1 ( 0.1
7.63 ( 0.38
6.25 ( 0.22
5.70 ( 0.37
6.39 ( 0.25
7.88 ( 0.53
6.97 ( 0.69
180 ( 3
180 ( 2
193 ( 4
201 ( 2
177 ( 4
185 ( 6
Λ
imp ) Λ
0
exp[-B/(T - T
0
)]
3
SO
2
)
]
[bmim][(C
[bmim][(CF
[bmim][CF SO
2 5
F
2 2
SO ) N]
3.0 ( 0.2 8.43 ( 0.24 166 ( 2
1.5 ( 0.1 6.05 ( 0.13 175 ( 1
2.7 ( 0.2 8.41 ( 0.29 159 ( 3
3.8 ( 0.2 8.97 ( 0.20 172 ( 2
2.3 ( 0.1 7.61 ( 0.13 169 ( 1
3.2 ( 0.1 7.80 ( 0.12 171 ( 1
3
3
3
SO
2
2
) N]
bmim][PF
bmim][CF
bmim][BF
6
]
3
3
]
CO
]
2
]
[bmim][PF
[bmim][CF
6
]
3
3
CO
2
]
4
[
Λ
bmim][BF
NMR ) Λ
4
]
TABLE 5: VFT Equation Parameters of Ionic Conductivity
Data
0
exp[-B/(T - T
0
)]
[bmim][(C
2
F
5
SO
2 2
) N]
5.6 ( 0.5 8.69 ( 0.28 165 ( 2
3.6 ( 0.4 7.03 ( 0.33 165 ( 3
3.5 ( 0.3 7.67 ( 0.23 165 ( 2
5.2 ( 0.8 8.73 ( 0.43 174 ( 3
5.1 ( 0.4 7.94 ( 0.22 166 ( 2
6.2 ( 0.5 8.44 ( 0.25 166 ( 2
[
[
bmim][(CF
bmim][CF SO
3
SO
2
2
) N]
σ ) σ
0
exp[-B/(T - T
0
)]
3
3
]
/10-1 mPas
2
[bmim][PF6]
σ
0
B/10 K
0
T /K
[
[
bmim][CF
bmim][BF
3 2
CO ]
[
[
[
[
[
[
bmim][(C
bmim][(CF
bmim][CF SO
bmim][PF
bmim][CF
bmim][BF
2
F
5
SO
2
)
2
2
N]
N]
7.1 ( 0.5
4.3 ( 0.2
9.8 ( 0.8
14.7 ( 1.0
9.2 ( 0.4
13.8 ( 0.6
7.96 ( 0.22
5.65 ( 0.14
7.93 ( 0.26
8.55 ( 0.20
7.19 ( 0.13
7.41 ( 0.13
169 ( 2
178 ( 2
162 ( 3
174 ( 2
172 ( 1
174 ( 1
4
]
3
SO
2
)
]
3
3
their VFT fitting curves are shown in Figure 6a, with the best-
fit parameters listed in Table 6. The molar conductivity of the
RTILs can also be calculated from the self-diffusion coefficients
6
]
3
4
CO
]
2
]
(ΛNMR), determined by the PGSE-NMR measurements, using
the ionic conductivity and the molar concentration. The VFT
equation for the molar conductivity is
the Nernst-Einstein equation
2
2
ΛNMR ) N e (D
+ Danion)/kT ) F (D + Danion)/RT
cation
A
cation
Λ ) Λ exp[-B/(T - T )]
(5)
0
0
(6)
2
-1
where Λ0 (Scm mol ), B (K), and T0 (K) are constants. The
temperature dependency of the molar conductivity (Λimp) and
where NA is the Avogadro number, e is the electric charge on
each ionic carrier, k is the Boltzmann constant, F is the Faraday

16598 J. Phys. Chem. B, Vol. 108, No. 42, 2004
Tokuda et al.
constant, and R is the universal gas constant. The temperature
dependency of the molar conductivity calculated from the ionic
diffusion coefficient and eq 6 is shown in Figure 6b and the
best-fit parameters of the VFT equation are listed in Table 6.
The experimental molar conductivity value (Λimp) is lower than
that of the calculated molar conductivity (ΛNMR) in the whole
temperature range, which has been established as one of the
important phenomena associated with ionic liquids (vide infra).
Discussion
Ion Diffusion Mechanisms of [bmim]-Based RTILs. The
sum of the cationic and anionic diffusion coefficients determined
-
7
2
-1
for the RTILs ranges from 1.6 to 6.0 × 10 cm s at 30 °C
and is relatively low. The viscosity and self-diffusion coefficient
of a typical polar organic solvent, propylene carbonate (PC),
-
6
2
have been reported as 2.5 mPas at 25 °C and 5.8 × 10 cm
-1
22,34
s
at 30 °C, respectively.
The low ionic diffusivity of the
ionic liquids is therefore well justified in terms of their high
viscosity (from 40 to 182 mPas at 30 °C). The much lower ion
diffusion coefficients and higher viscosity for the ionic liquids
may be attributable to the very strong Coulombic forces among
each ionic species in the ionic liquids.
As mentioned above, the sum of cationic and anionic self-
diffusion coefficients and viscosities for the RTILs differs
depending on the anionic structures. In general, the diffusivity
is correlated to the fluidity (1/η) by the Stokes-Einstein
equation
kT
cπηrs
D )
(7)
where k is the Boltzmann constant, T is the absolute temperature,
c is a constant, and rs is the effective hydrodynamic (Stokes)
radius. Figure 7a,b illustrates the relationship between the Tη
Figure 7. Relationship between T η-1 (T, absolute temperature; and
-1
η, viscosity) and self-diffusion coefficients of (a) the cation and (b)
and the self-diffusion coefficient of the cation and anion,
respectively. All of the ionic liquids give approximately straight
anion for [bmim] based ionic liquids.
2
lines passing through the origin (regression factor: R > 0.988),
indicating that the ionic diffusivity in the ionic liquids basically
obeys eq 7. However, it should be noted that the slopes of the
straight lines are not identical, even for the identical cationic
relationships (Figure 7a). The Stokes law is based on the
assumption of a rigid solute sphere diffusing in a continuum of
solvent, and in the case of a large solute in a small solvent, the
factor, c, can attain the value of 6. However, if the ratio of the
solute size to solvent is increased, especially for highly viscous
media, the correlation breaks down, and the value of c in eq 7
3
5
is reduced to ca. 4. Thus, the factor c may help to understand
the microscopic ion dynamics in the RTILs. The reported van
der Waals radius from MM2 and ab initio molecular orbital
calculations is 0.330, 0.362, 0.326, 0.267, 0.254, and 0.227 nm
for [bmim], [(C2F5SO2)2N], [(CF3SO2)2N], [CF3SO3], [PF6], and
[
BF4], respectively.36,37 By using the experimental slopes in
Figure 7a and the van der Waals radius of [bmim], the factor c
for [bmim] based ionic liquids with [(C2F5SO2)2N], [(CF3-
SO2)2N], [CF3SO3], [PF6], [CF3CO2], and [BF4] can be obtained
as 3.3, 3.4, 3.4, 3.0, 3.8, and 3.3, respectively. Similarly, the c
values for the anionic diffusivity from Figure 7b are 4.1, 4.3,
Figure 8. Apparent cationic transference number, Dcation/(Dcation
anion), for [bmim] based ionic liquids plotted against temperature.
+
D
5
.0, 4.6, and 4.6 for [(C2F5SO2)2N], [(CF3SO2)2N], [CF3SO3],
Figure 8 depicts the apparent cationic transference number
[
PF6], and [BF4], respectively. Although the van der Waals
(Dcation/(Dcation + Danion)) for each RTIL as a function of
temperature. Interestingly, the cationic self-diffusion coefficient
is larger than the anionic diffusion coefficient at all of the
temperatures of the measurements, except for [bmim][BF4] at
80 °C. The transference number approximately follows the order
[(C2F5SO2)2N] > [(CF3SO2)2N] ≈ [CF3SO3] ≈ [PF6] > [CF3-
CO2] > [BF4]. The order is likely to be influenced by the size
radius of [CF3CO2] is not available in the literature, the factor
c can be approximated in the range from 5.4 to 6.0, by using
the radii of [PF6] and [BF4]. The results of smaller c values of
the cation than those of anions suggest that the anions in the
ionic liquids diffuse slower than the cation if they have the same
van der Waals radii.

Room Temperature Ionic Liquids
J. Phys. Chem. B, Vol. 108, No. 42, 2004 16599
1
Ionic Association and Ion Transport Behavior. The H and
the 19F NMR signals were always single lines for each assigned
nucleus without multiple signals, indicating that, even if there
are dissociated, paired, and/or aggregated ionic species, the rate
of exchange for the chemical equilibrium between the dissoci-
ated and associated ions in the ionic liquids is faster than the
time scale of NMR measurements. The data on the ionic
diffusivity and the molar conductivity makes it possible to
compare the ratio of the molar conductivity obtained from the
impedance measurement (Λimp) to that calculated from the ionic
diffusivity (ΛNMR). The application of the Nernst-Einstein
equation for the interpretation of the experimental results of
PGSE-NMR self-diffusion coefficients has been proved useful
by Aihara et al. for electrolyte solutions including lithium salts,
wherein they have shown that the Λimp/ΛNMR approaches unity
for complete dissociation of the electrolytes at the infinite
23
dilution. The Λimp/ΛNMR can provide useful information about
the ionic dissociation/association under equilibrium in the
RTILs, as it serves as an indicator of the percentages of ions
contributing to the ionic conduction within the diffusing
component.
Figure 9. Molar conductivity ratios (Λimp/ΛNMR) for [bmim] based
ionic liquids plotted against temperature.
of the anions. It should be mentioned here that, despite a large
disparity of the cationic (0.330 nm) and anionic (0.227 nm) radii,
the transference number of [bmim][BF4] is estimated to be
higher than 0.5 at room temperature. These results also indicate
that the cation can diffuse faster than the anions, even if the
cationic radius is larger than the anionic radii. The apparent
cationic transference number decreases almost linearly with
increasing temperature in all RTILs indicating the relatively
higher thermal acceleration of the anionic diffusion compared
with the [bmim] diffusion, which is supported by the larger B
values, in other words, the activation energies for the diffusivity
of the anions than those for the cations (Table 3) for the
individual ionic liquids.
The Λimp/ΛNMR is plotted against temperature in Figure 9
and is found to be relatively insensitive to the temperature. It
is clear that the ratios for all of the RTILs in this study are
lower than unity, indicating that not all of the diffusive species
in ionic liquids contribute to the ionic conduction. It indicates
that ionic association occurs in the liquids. The variation of the
anions in the [bmim]-based RTILs leads to the different Λimp/
ΛNMR values, and the ratio follows the order [PF6] > [BF4] >
[(C2F5SO2)2N] > [(CF3SO2)2N] > [CF3SO3] > [CF3CO2] at
30 °C. Since the cationic structure is fixed to [bmim], the order
of the Λimp/ΛNMR depends merely on the anionic character. In
the [PF6], [BF4], [(C2F5SO2)2N], and [(CF3SO2)2N], the highly
6
Figure 10. FAB-MS spectra of [bmim][PF ] for (a) positive FAB and (b) negative FAB.

16600 J. Phys. Chem. B, Vol. 108, No. 42, 2004
Tokuda et al.
electronegative fluorine atom and electron-withdrawing per-
fluorosulfonyl groups contribute to the distribution of the anionic
charge of phosphate, borate, and imide, respectively. In addition
to the anionic charge distribution, the effect of the surface-
covering of the anion backbone by fluorine atoms may be a
significant factor for weak interaction with the [bmim] cation.
On the other hand, the more pronounced anionic charge
localization in the anionic structures of [CF3SO3] and [CF3CO2],
compared to other anions, makes them interacting sites with
the [bmim]. Consequently, the interaction gives relatively lower
Λimp/ΛNMR values for [bmim][CF3SO3] and [bmim][CF3CO2].
The Λimp/ΛNMR results are in good agreement with the previous
report by Linert et al.,38 wherein the donor number of [BF4]
has been estimated to be lower than that of [CF3SO3] by
solvatochromic study. Thus, the electron-donor ability (or Lewis
basicity) determines the order of the ratios in the ionic liquids.
It should be mentioned here that the molecular and quasi-
molecular ion peaks could be observed by the FAB-MS
measurements without use of any matrixes. A typical example
of the FAB-MS spectra for [bmim][PF6] is shown in Figure
(7) (a) Kawano, R.; Watanabe, M. Chem. Commun. 2003, 330-331.
(
b) Kawano, R.; Matsui, H.; Matsuyama, C.; Sato, A.; Susan, M. A. B. H.;
Tanabe, N.; Watanabe, M. J. Photochem. Photobiol. A, Chem. 2004, 164,
8
7-92.
(8) Doyle, M.; Choi, S. K.; Proulx, G. J. Electrochem. Soc. 2000, 147,
4-37.
9) (a) Noda, A.; Susan, M. A. B. H.; Kudo, K.; Mitsushima, S.;
3
(
Hayamizu, K.; Watanabe, M. J. Phys. Chem. B 2003, 107, 4024-4033.
(b) Susan, M. A. B. H.; Noda, A.; Mitsushima, S.; Watanabe, M. Chem.
Commun. 2003, 938-939.
(
10) Lu, W.; Fadeev, A. G.; Qi, B.; Smela, E.; Mattes, B. R.; Ding, J.;
Spinks, G. M.; Mazurkiewicz, J.; Zhou, D.; Wallace, G. G.; MacFarlane,
D. R.; Forsyth, S. A.; Forsyth, M. Science 2002, 297, 983-987.
(11) (a) Carlin, R. T.; Fuller, J. Chem. Commun. 1997, 1345-1346. (b)
Fuller, J.; Breda, A. C.; Carlin, R. T. J. Electrochem. Soc. 1997, 144, L67-
L70.
12) (a) Noda, A.; Watanabe, M. Electrochim. Acta 2000, 45, 1265-
270. (b) Susan, M. A. B. H.; Kaneko, T.; Noda, A.; Watanabe, M.
submitted.
(
1
(
13) Yoshizawa, M.; Hirao, M.; Ito-Akita, K.; Ohno, H. J. Mater. Chem.
001, 11, 1057-1062.
14) (a) Snedden, P.; Cooper, A. I.; Scott, K.; Winterton, N. Macro-
2
(
molecules 2003, 36, 4549-4556. (b) Scott, M. P.; Rahman, M.; Brazel, C.
S. Eur. Polym. J. 2003, 39, 1947-1953.
(
15) Fukushima, T.; Kosaka, A.; Ishimura, Y.; Yamamoto, T.; Takigawa,
T.; Ishii, N.; Aida, T. Science 2003, 300, 2072-2074.
1
0. The mass-to-charge ratio, m/z ) 139, and 423 in the positive
FAB spectrum (Figure 10a) can be assigned by [bmim] and
bmim]2[PF6], respectively. Similarly, the values of m/z ) 145,
29, and 713 in the negative FAB spectrum (Figure 10b) are
(
16) Wilkes, J. S.; Zaworotko, M. J. Chem. Soc., Chem. Commun. 1992,
965-967.
(
17) MacFarlane, D. R.; Meakin, P.; Sun, J.; Amini, N.; Forsyth, M. J.
[
4
Phys. Chem. B 1999, 103, 4164-4170.
(
18) (a) Sun, J.; Forsyth, M.; MacFarlane, D. R. J. Phys. Chem. B 1998,
attributed to [PF6], [bmim][PF6]2, and [bmim]2[PF6]3, respec-
tively. The observation of such ion aggregates and/or clusters
102, 8858-8864. (b) Matsumoto, H.; Yanagida, M.; Tanimoto, K.; Nomura,
M.; Kitagawa, Y.; Miyazaki, Y. Chem. Lett. 2000, 922-923.
12b,39
(19) Matsumoto, H.; Matsuda, T.; Miyazaki, Y. Chem. Lett. 2000, 1430-
of imidazolium salts in a liquid matrix has also been reported.
1
431.
Interestingly, the highest molecular ions in the same measured
condition, corresponding to [bmim]3[CF3CO2]4, could be de-
tected for the [bmim][CF3CO2] ionic liquid, which gave the
lowest Λimp/ΛNMR values.
(
20) Holbrey, J. D.; Seddon, K. R. J. Chem. Soc., Dalton Trans. 1999,
2133-2139.
(
21) Bonh oˆ te, P.; Dias, A.-P.; Papageorgiou, N.; Kalyanasundaram, K.;
Gr a¨ tzel, M. Inorg. Chem. 1996, 35, 1168-1178.
(
22) Hayamizu, K.; Aihara, Y.; Arai, S.; Martinez, C. G. J. Phys. Chem.
B 1999, 103, 519-524.
23) Aihara, Y.; Sugimoto, K.; Price, W. S.; Hayamizu, K. J. Chem.
Phys. 2000, 113, 1981-1991.
24) Aihara, Y.; Bando, T.; Nakagawa, H.; Yoshida, H.; Hayamizu, K.;
Concluding Remarks
(
In this study, we propose that the ratio Λimp/ΛNMR is a useful
parameter to characterize various properties of the RTILs with
varying anionic species. In other words, the Λimp/ΛNMR repre-
sents “ionicity” of the ionic liquids, and the significant physical
and chemical properties of RTILs may be determined by this
parameter. Although this study is only concerned with anionic
effects on the ion dynamics and ionicity, it is equally important
to establish a correlation between the cationic structures and
the physicochemical properties of RTILs. Further studies would
be focused on the elucidation of such a correlation.
(
Akiba, E.; Price, W. S. J. Electrochem. Soc. 2004, 151, A119-A122.
(25) Noda, A.; Hayamizu, K.; Watanabe, M. J. Phys. Chem. B 2001,
1
05, 4603-4610.
26) Cammarata, L.; Kazarian, S. G.; Salter, P. A.; Welton, T. Phys.
Chem. Chem. Phys. 2001, 3, 5192-5200.
27) Stejskal, E. O.; Tanner, J. E. J. Chem. Phys. 1965, 42, 288-292.
(28) Hayamizu, K.; Price, W. S. J. Magn. Reson. 2004, 167, 328-333.
29) Seddon, K. R.; Stark, A.; Torres, M.-J. Pure Appl. Chem. 2000,
2, 2275-2287.
30) Ngo, H. L.; LeCompte, K.; Hargens, L.; McEwen, A. B. Thermo-
chim. Acta 2000, 357-358, 97-102.
31) (a) Fox, D. M.; Awad, W. H.; Gilman, J. W.; Maupin, P. H.;
(
(
(
7
(
(
Acknowledgment. This research was supported in part by
Grant-in-Aid for Scientific Research (Nos. 14350452 and
DeLong, H. C.; Trulove, P. C. Green Chem. 2003, 5, 724-727. (b) Awad,
W. H.; Gilman, J. W.; Nyden, M.; Harris, R. H.; Sutto, T. E.; Callahan, J.;
Trulove, P. C.; DeLong, H. C.; Fox, D. M. Thermochim. Acta 2004, 409,
1
6205024) from the Japan Ministry of Education, Science,
3
-11.
Sports and Culture. The authors acknowledge Mr. Takeo Kaneko
(
32) (a) Vogel, H. Phys. Z. 1921, 22, 645. (b) Fulcher, G. S. J. Am.
(
(
YNU) for the FAB-MS measurements and Dr. Seiji Tsuzuki
AIST) for discussions. H.T. and M.A.B.H.S. acknowledge
Ceram. Soc. 1923, 8, 339-355. (c) Tamman, G.; Hesse, W. Z. Anorg. Allg.
Chem. 1926, 156, 245-257.
(
33) Morrow, T. I.; Maginn, E. J. J. Phys. Chem. B 2002, 106, 12807-
JSPS, and K.H. acknowledges NEDO for financial support.
1
2813.
(
34) Organic SolVents, 4th ed.; Riddick, J. A., Bunger, W. B., Sakano,
References and Notes
T. K., Eds.; John Wiley & Sons: New York, 1986.
(
1) Ionic Liquids in Synthesis; Wasserscheid, P., Welton, T., Eds.;
(35) Diffusion: Mass Transfer in Fluid System; Cussler, E. L.; Cam-
bridge University Press: New York, 1984.
Wiley-VCH: Weinheim, Germany, 2003.
(
2) (a) Welton, T. Chem. ReV. 1999, 99, 2071-2083. (b) Holbrey, J.
(36) (a) Ue, M. J. Electrochem. Soc. 1994, 141, 3336-3342. (b) Ue,
M.; Murakami, A.; Nakamura, S. J. Electrochem. Soc. 2002, 149, A1385-
A1388.
D.; Seddon, K. R. Clean Prod. Process. 1999, 1, 223-236. (c) Wassersheid,
P.; Keim, W. Angew. Chem., Int. Ed. 2000, 39, 3772-3789.
(3) (a) Koch, V. R.; Nanjundiah, C.; Appetecchi, G. B.; Scrosati, B. J.
(37) The van der Waals radius of [bmim] was not available in the
literature. However, cationic radii of some ionic liquids based on imida-
zolium, pyridinium, pyrrolidinium, and ammonium cation have been
reported.3 Therefore, the radius of [bmim] may be approximated by
considering the radius of [emim] (0.303 nm) and that an increase in the
alkyl chain length (-CH2-) of the substituent of the tetraalkylammonium
cation brings about an increase of about 0.160 nm.
Electrochem. Soc. 1995, 142, L116-L118. (b) Sakaebe, H.; Matsumoto,
H. Electrochem. Commun. 2003, 5, 594-598.
6b
(4) (a) Koch, V. R.; Dominey, L. A.; Nanjundiah, C.; Ondrechen, J. J.
Electrochem. Soc. 1996, 143, 798-803. (b) Nanjundiah, C.; McDevitt, S.
F.; Koch, V. R. J. Electrochem. Soc. 1997, 144, 3392-3397.
(5) Ue, M.; Takeda, M.; Toriumi, A.; Kominato, A.; Hagiwara, R.;
Ito, Y. J. Electrochem. Soc. 2003, 150, A499-A502.
6) Papageorgiou, N.; Athanassov, Y.; Armand, M.; Bonh oˆ te, P.;
Pettersson, H.; Azam, A.; Gr a¨ tzel, M. J. Electrochem. Soc. 1996, 143, 3099-
108.
(38) Linert, W.; Jameson, R. F.; Taha, A. J. Chem. Soc., Dalton Trans.
(
1993, 3181-3186.
(39) Abdul-Sada, A. K.; Seddon, K. R. Eur. Mass Spectrom. 1997, 3,
245-247.
3