Solvent-solute interactions in ionic liquids

Article abstract of DOI:10.1039/b303095d

A range of ionic liquids was studied utilizing the Kamlet-Taft parameters hydrogen bond acidity (α), hydrogen bond basicity (β), and dipolarity/polarizability effects (π*). Some of the ionic liquids studied were methanol, acetonitrile, acetone, dichloromethane, toluene, hexane, and water. π* was high for all the ionic liquids investigated and varied with both anion and cation. α was generally moderate and depended chiefly on the cation. β was also moderate and depended mainly on the anion. Comparison was made with other polarity measurements in ionic liquids.

Full text of DOI:10.1039/b303095d

View Online / Journal Homepage / Table of Contents for this issue
Solvent–solute interactions in ionic liquidsy
Lorna Crowhurst, Philip R. Mawdsley, Juan M. Perez-Arlandis, Paul A. Salter and
Tom Welton*
Department of Chemistry, Imperial College London, Exhibition Road, London, UK SW7 2AY.
E-mail: t.welton@ic.ac.uk
Received 20th March 2003, Accepted 1st May 2003
First published as an Advance Article on the web 29th May 2003
A range of ionic liquids has been investigated using the Kamlet–Taft parameters a, b, and p*. It was
found that p* is high for all of the ionic liquids studied and varies with both anion and cation, a is generally
moderate and depends mainly on the cation, b is also moderate and depends mainly on the anion. Comparison
is made with other polarity measurements in ionic liquids.
Introduction
Ionic liquids are becoming increasingly popular as solvents
for a wide range of applications in synthesis and catalysis.
1
Although estimates vary, there is no doubt that the number
of combinations of anions and cations that will give rise to
potential ionic liquids is vast. This possibility for synthetic var-
iation has lead to ionic liquids being described as ‘‘Designer
2
Solvents’’. However, in order to be able to realise this possibi-
lity it is necessary to know exactly what is being designed, which
parameters are flexible and which are fixed. Further to this,
although the large number of ionic liquids potentially available
provides synthetic flexibility it also presents a problem, in that it
is not possible to make every possible combination of ions and
measure the properties of the resultant liquid. It is, therefore,
necessary to link the feature that is being designed to more fun-
damental properties that can be predicted. Finally, it is also
necessary to be able to compare ionic liquids to the molecular
solvents that they may one day replace.
Fig. 1 The dyes used: Reichardt’s dye (A), N,N-diethyl-4-nitroaniline
(
B) and 4-nitroaniline (C).
N
T
Table 1
E
and Kamlet–Taft values for a selection of ionic liquids
N
19
H
DG /
E
l
Cu/nm
T
ꢀ
1
Solvent
(Lit.)
p* (Lit.)
a (Lit.)
b (Lit.)
(Lit.)
kJ mol
The key characteristics of a liquid that is to be used as a
solvent are those that determine how it will interact with
potential solutes. For molecular solvents, this is most com-
monly recorded as the polarity of the pure liquid, as expressed
through its dielectric constant. The inability of this scale to
provide adequate correlations with many experimental data
has lead to the use of empirically derived measurements of sol-
11
a
(1.10 )
b
(0.46 )
[
EtNH
HO(CH
mim][Tf N]
3
][NO
3
]
(0.954)
(0.929)
(1.12)
[
2
2
) -
1187.8
1187.8
10
2
[CH
mim][Tf
bmim][SbF
3
O(CH
2
)
2
-
(0.722)
1
0
2
N]
[
6
]
0.673
0.670
1.039
1.047
0.639
0.627
0.146
0.376
525.5
565.0
1069.7
1204.5
8
3
4
4
[bmim][BF ]
vent properties. Kamlet and Taft have developed a system,
based on the comparison of effects on the UV–vis spectra of
sets of closely related dyes that were selected to probe particu-
lar solvent properties. These complimentary scales of hydrogen
bond acidity (a), hydrogen bond basicity (b) and dipolarity/
polarizability effects (p*) are the ones that we use here. There
are several sets of dyes that can be used to derive the Kamlet–
Taft parameters and many studies are quoted as an average of
the values obtained for several of these dye sets. However, here
we report the values obtained using a single set of dyes: Reich-
ardt’s dye, 4-nitroaniline and N, N-diethyl-4-nitroaniline,
which are shown in Fig. 1.
(
0.673)
8
[
bmim][PF
6
]
0.669
(0.667)
0.656
1.032
1.006
0.984
0.634
0.625
0.617
0.207
0.464
0.243
524.0
1158.1
1224.6
(516.5)
577.0
8
[bmim][TfO]
(
0.667)
0.644
0.642)
(601.5)
545.0
8
27
[
bmim][N(Tf)
2
]
1187.8
(
(546.0)
558.0
[
[
bm
2
im][BF
4
]
0.576
0.544
0.541
(0.552)
(1.00)
1.083
0.954
1.010
0.402
0.427
0.381
0.363
0.252
0.239
1204.5
1187.8
1187.8
bmpy][N(Tf)
2
]
545.0
8
[bm im][N(Tf) ]
2
2
547.0
(547.5)
(591)
3,6,28
Water
(1.33)
b
1.13 )
(1.12)
b
(1.16 )
(0.14)
b
(0.50 )
(
3,6,28
Methanol
(0.762)
0.460
0.350
0.309
(0.73)
0.799
0.704
0.791
(1.05)
0.350
0.202
0.042
(0.61)
0.370
(589)
(575)
(571)
497.5
(550)
28
Acetonitrile
28
Acetone
Results and discussion
0.539
Dichloro-
ꢀ0.014
N
T
The values of E , a, b, and p* are listed in Table 1. The litera-
ture values included in the table (in parentheses) are those
28
methane
Toluene
0.100
0.532
ꢀ0.213
0.077
(0.04)
3,7
Hexane
(0.009)
(ꢀ0.12)
(0.07)
y Electronic supplementary information (ESI) available: synthesis of
a
b
Recalculated, Average value from more than one dye set.
ionic liquids. See http://www.rsc.org/suppdata/cp/b3/b303095d/
2
790
Phys. Chem. Chem. Phys., 2003, 5, 2790–2794
This journal is # The Owner Societies 2003
DOI: 10.1039/b303095d

View Online
molecular solvents. It is also the scale that has been applied
3
obtained using the aforementioned dye set. All spectra were
ꢁ
8–11
recorded at 25 C and we were, therefore, restricted in our
to the greatest number of ionic liquids.
Measurement of
selection of ionic liquids to those that are liquid at room tem-
perature. The ionic liquids used in this work are shown (with
abbreviations) in Fig. 2.
the position of the absorption maximum of Reichardt’s dye
is a necessary part of the measurement of a as well as the
source of the E NT scale. Hence we discuss the E NT values of
the ionic liquids here.
Where the same ionic liquids have been used and so compar-
ison is possible, it can be seen that the values here are in good
agreement with those from the literature. As has been noted by
p* values
All of the p* values (Table 1) for the ionic liquids are
high in comparison with non-aqueous molecular solvents.
Although differences between the ionic liquids are small, both
the cation and the anion can be seen to affect the value. All of
5
–7
8
N
+
others, the E values for the [bmim] ionic liquids lie in the
range associated with polar hydrogen bond donor solvents
T
3
N
2 2
such as alcohols. The E values are higher for [HO(CH ) -
T
ꢀ
1
0
the [Tf
of values observed. The [bmim] ionic liquids have lower
2
N] based ionic liquids lie at the low end of the range
mim][Tf
and [EtNH
However, the value for [CH O(CH ) mim][Tf N] is also
2
N], which contains its own alcohol functionality
+
1
1
3
3
][NO ], which is an N–H hydrogen bond donor.
+
1
0
2
values than the [bm im] ionic liquids with a common anion
and the [bmpy][Tf N] ionic liquid has the lowest value of all.
2
3
2 2
2
higher than for [bmim][Tf N], even though is contains a basic
2
functionality (ether). While it also registers local electric field
In molecular solvents, p* reports the effects of the dipolarity
and polarizability of the solvent. However, it should be noted
that what is being measured is a property of the solute, namely
the differential stabilization of the more polar excited state
with respect to the ground state of the dye. Hence, p* is
derived from the change in the energy of the absorption max-
imum of the dye that is induced by the local electric field gen-
erated by the solvent. Therefore, it is no surprise that p* has
been greatly affected by the ion-dye interactions now possible
in the ionic liquid. It can be seen that as the charge on the
anions becomes delocalised over more atoms the value of p*
decreases due to the decrease in the strength of these Coulom-
bic interactions. In the cations, on the other hand, it appears
that the decrease in Coulombic interactions caused by deloca-
lising the charge around the imidazolium ring is more than
compensated for by the increased polarizability of the deloca-
lised system. The effect of delocalized p-electron systems can be
seen in the relatively high p* value of toluene in comparison to
hexane (Table 1).
effects, the E NT scale is particularly sensitive to the hydrogen
bond donor ability of the cation of the ionic liquid. This sug-
8
gests that the electron withdrawing effect of the oxygen atom
of [CH O(CH ) mim][Tf N] is actually increasing the overall
3
2 2
2
+
im] salts have significantly lower
acidity of the ion. The [bm
N
2
+
E
salts shows that there is also a clear effect of changing anion on
values, as does [bmpy][Tf N]. Comparison of the [bmim]
2
T
N
T
E . These details will be discussed below in terms of their
effects on the a values.
a values
The a values for the ionic liquids are listed in Table 1. The
values are largely determined by the nature of the cation,
but there is also a smaller anion effect. The values for the
+
5
[bmim] are generally moderately high (cf., t-BuOH). It has
long been known that all three of the imidazolium ring pro-
1
2,13
tons are acidic,
ing to solutes would be significant in the absence of hydrogen
bond accepting anions. The [bm
lowest a values, reflecting the loss of the proton on the 2-posi-
tion of the ring. [bmpy][Tf N] has a slightly higher a value
than [bm im][Tf N]. Although one might have expected this
and it was predicted that hydrogen bond-
It should be noted that the ionic liquids used in this study
are all composed of relatively similar cations. Poole et al. have
measured p* for a series of alkylammonium nitrates and thio-
cyanates (Table 1), which have still higher values, albeit with
a different set of dyes. This probably reflects the closeness of
approach to the charge centres of these salts that is possible
and hence greater ion–dye Coulombic interactions.
1
4
+
2
im] ionic liquids have the
1
1
2
2
2
latter relationship to be the reverse, since hydrogen bond
donation increases with the increasing s-character of the
1
5
C–H bond, hydrogen bonds to the a-carbons of alkyl chains
1
6
in alkylammonium salts are well known. The higher than
expected value for [bmpy][Tf N] in comparison to
[bm im][Tf N] probably reflects the number of protons that
The EN scale
T
a
2
The EN scale is considered to be a good general scale of the
solvating ability of a liquid and is known for a wide range of
T
2
2
can be involved in the hydrogen bonding to the probe
Fig. 2 The ionic liquids used in this study.
Phys. Chem. Chem. Phys., 2003, 5, 2790–2794
2791

View Online
noted that the calculated values increasingly over-estimate the
3 3
molecule in each case. The higher a value of [EtNH ][NO ],
1
9
which is an N–H hydrogen bond donor, shows that the oppor-
tunity arises to prepare strongly hydrogen bond donating ionic
liquids. The hydrogen bond donor abilities of ammonium
cations are known to be higher than that of their neutral
acidity (under-estimate DG ) as the acid gets stronger. It can
H
be seen that the hydrogen bond basicity of the ionic liquids is
generally inversely proportional to the acidities of the conju-
gate acid of the ionic liquid’s anions. The deviation from linear
1
7
amines, hence the appearance of a large Coulombic contri-
bution to the a value.
Focusing on the [bmim] salts, there is a clear anion effect
behaviour is probably a result of the under-estimation of DG
of the stronger acids.
The different b values obtained for the different [Tf N]
2
H
+
ꢀ
ꢀ
seen. With the exception of the [N(Tf)
2
]
ion, which appears
ionic liquids show that the cation has an influence, but with
the limited data set here no trend can be clearly discerned at
this stage.
not to fit the simple trend, as the anion becomes more basic
increasing b) the hydrogen bond donor ability of the ionic
(
liquid decreases. However, since the cation has been
unchanged, its ability to act as a hydrogen-bond donor has
been unchanged, so why is an effect seen at all?
Comparisons with other polarity studies
We propose that there is a competition between the anion
and probe dye solute for the proton. So, the a values of the
ionic liquids are controlled by the ability of the liquid to act
as a hydrogen-bond donor (cation effect) moderated by its
hydrogen-bond acceptor ability (anion effect). This may be
described in terms of two competing equilibria. The cation
can hydrogen bond to the anion (eqn. (1)):
A number of other methods have been used to investigate
solvent behaviour of ionic liquids.
2
0–22
Most of these have not
involved a sufficient overlap with the range of ionic liquids
used here to allow for comparison of the trends observed.
Gordon et al. have used the absorption maximum of (ace-
8
0
0
tylacetonato)(N,N,N ,N -tetramethylethylenediamine)copper
II)tetraphenylborate {[Cu(acac)(tmen)][BPh ],lCu} as a probe
of the basicity of the anions of ionic liquids and correlated this
(
þ
ꢀ
4
K1
½½bminꢂ ꢃ ꢃ ꢃ A ꢂ
þ
ꢀ
þ
ꢀ
(
½
bminꢂ þ A
+½bminꢂ ꢃ ꢃ ꢃ A ; K1 ¼
ð1Þ
þ
ꢀ
½
½bminꢂ ꢂ½A ꢂ
value with the rates (TOF) of nickel catalysed oligomerization
2
3
reactions. We have extended this study to include all of the
ionic liquids investigated here. It can be seen that there is excel-
lent agreement between this measure of basicity and b and it
can be concluded that the two measurements are reporting
similar behaviour.
The cation can hydrogen bond to the solute, in this case the
probe dye (eqn. (2)):
K2
þ
þ
(
½
bminꢂ þ solute +½b minꢂ ꢃ ꢃ ꢃ solute;
þ
ꢀ
½
½b minꢂ ꢃ ꢃ ꢃ solute ꢂ
A study of multiple solvation interactions, based on GC
measurements, that were used to characterise a range of ionic
K2 ¼
ð2Þ
þ
½
½b minꢂ ꢂ½soluteꢂ
2
4
liquids has been recently reported. Eight of the ionic liquids
used are from the same samples as those as used in this study.
The ionic liquids were prepared, purified and the sample split
into two parts one of which was sent for the GC measure-
ments and the other used for this Kamlet–Taft experiment,
so a direct comparison can be made. The non-specific inter-
actions are not directly comparable for the two methods, yet
in both studies the ionic liquids were consistently found to
be highly polar.
Clearly the concentration of the hydrogen bonded
+
[
bmim] . . .solute complex is dependent upon the availability
of the cation for bonding which, in turn, is dependent on the
value of K . Hence, as the anion becomes a better hydrogen
bond acceptor, K increases and the concentration of free
non-hydrogen bonded [bmim] decreases and, since for a given
1
1
+
+
2
solute in the [bmim] based ionic liquids K is fixed, the con-
+
centration of the [bmim] ꢃ ꢃ ꢃsolute complex falls. We have pre-
viously used these equilibria to explain the change in selectivity
of the Diels–Alder cycloaddition of cyclopentadiene and
The hydrogen bonding effects would be expected to give
similar results in both experiments. Indeed, the hydrogen bond
basicities of the ionic liquids follow the same trend in both stu-
dies, are controlled by the anion and are moderate in value.
Comparisons of hydrogen bond acidity are, however, not so
clear-cut. In the solvation study the ionic liquids’ hydrogen
bond donation behaviour as a whole was found to be domi-
nated by the hydrogen bond basicity of the anions, with a much
lesser contribution from the hydrogen bond acidity of the
cation. Low and even negative values were found for the
hydrogen bond acidity function. The only ionic liquids to display
significant hydrogen bond donor ability where those of the
1
8
methyl acrylate in various ionic liquids.
An alternative explanation is that the anions are interacting
directly with the dye itself. However, the positive charge of
Reichardt’s dye is distributed around the aromatic system
and there is no well-located site for interaction with the anions
that would play the part that the lone pairs of the oxygen atom
play for the cations.
b values
ꢀ
24
b is the hydrogen bond basicity of the solvent. The values for
the ionic liquids studied here are moderate and dominated by
the nature of the anion (Table 1). As the conjugate bases of
strong acids, the anions of the ionic liquids might be expected
to have low b-values in comparison to other solvents. However,
although those found in this study are not as high as for acet-
one, they are comparable to acetonitrile, which is thought of
[Tf N] ion. In contrast, this Kamlet–Taft study has strongly
2
emphasised the role of the hydrogen bond acidity of the cation.
A plot (Fig. 3) of the two functions, a from this study and b
from the solvation study, reveals a complex relationship.
It can be seen for the imide ionic liquids, that the hydrogen
+
bond acidity does indeed vary with cation with [bmim] being
the most acidic followed by [bmpy] and finally [bm im] in
+
+
2
5
as an electron pair donor solvent. This shows again that there
both studies. However, changing to more basic anions leads
to a dramatic drop in the acidity measurements in the solvation
study, whereas it has only a limited effect in the study
reported here. That is, the solvation measurement is anion
dominated, whereas the Kamlet–Taft measurement is cation
dominated.
We propose that this is not a simple matter of one of the
measurements being right while the other is wrong, but rather
more about what is actually being measured. The intimate
relationship that exists between solute and solvent is such that
the nature of the solute strongly influences which solvent
is an important Coulombic contribution to the hydrogen
bonds formed between the ionic liquid and solute species.
Table 1 also includes a measure of the gas phase basicity of
the anions, given as the calculated Gibbs free energy change
for the gas phase deprotonation of its conjugate acid,
1
9
DG
H
.
Experimental measurement of the acidity of such
strong Brønsted acids is extremely difficult, hence these calcu-
lated values are preferred and are available for all of the anions
used here. Calculated values also have the advantage that they
can be generated for hypothetical acids. However, it should be
2
792
Phys. Chem. Chem. Phys., 2003, 5, 2790–2794

View Online
Fig. 3 Comparison of the hydrogen bond donor b vs. a functions for the ionic liquids.z
properties will be expressed. In the solvation study the set of
probes has been deliberately selected to provide a broad range
Experimental
2
4
of interactions. Some of the probes can act as hydrogen bond
acceptors, others do not, some will interact strongly with the
cation, others with the anion and yet others with neither.
Whereas Reichardt’s dye, key in the measurement of a, has a
strongly hydrogen-bond-accepting phenoxide oxygen, which
is anionic and will strongly and exclusively interact with the
cation of the ionic liquid. Hence, this Kamlet–Taft study
reveals the interactions of which the ionic liquids are capable
in the presence of a strongly and specifically interacting solute
and how they are influenced by the nature of the two compet-
ing ions, whereas the solvation study reveals the value when
the nature of any particular solute has been deliberately
down-played by averaging over many solutes and how that
is influenced by the nature of the ions.
The syntheses of the ionic liquids used in this research have
been reported elsewhere,
supplementary information (ESI).y The dyes were used as
received. All spectra were recorded using a PC-controlled Per-
kin-Elmer Lambda 2 spectrophotometer, fitted with a thermo-
statted sample holder.
2
5,26
and are available as electronic
Acknowledgements
We would like to thank Kodak Ltd (JMPA) and GSK/
EPSRC (LC) for the provision of studentships. We would also
like to thank Professors M. H. Abraham, D. W. Armstrong
and I. A. Koppel for useful discussions.
References
Conclusions
1
(a) T. Welton, Chem. Rev., 1999, 99, 2071; (b) P. Wasserschied
and W. Keim, Angew. Chem., Int. Ed. Engl., 2000, 39, 3772; (c)
Ionic Liquids in Synthesis, ed. T. Welton and P. Wasserschied,
VCH-Wiley, Weinheim, 2002.
Comparison of the values with those of molecular solvents
show that care must be taken in claims of similarities
between ionic liquids and such solvents. While it can be said
that the E NT values of the [bmim] ionic liquids are similar
to those of short chain alcohols, all of the values of p*, a
and b are quite different. Therefore, it would not be sensible
to state that ionic liquids are ‘‘like’’ short chain alcohols,
without specifying the narrow respect in which this might
be considered to be true. This will be true of many possible
comparisons.
+
2
3
M. Freemantle, Chem. Eng. News, 1998, 76, 32.
(a) C. Reichardt, Solvents and Solvent Effects in Organic Chemis-
try, VCH, Weinheim, 2nd edn., 1990; (b) C. Reichardt, Chem.
Rev., 1994, 94, 2319.
4
(a) M. J. Kamlet and R. W. Taft, J. Am. Chem. Soc., 1976, 98,
3
77; (b) R. W. Taft and M. J. Kamlet, J. Am. Chem. Soc., 1976,
8, 2886; (c) T. Yokoyama, R. W. Taft and M. J. Kamlet, J.
9
Am. Chem. Soc., 1976, 98, 3233; (d ) M. J. Kamlet, J. L. Addoud
and R. W. Taft, J. Am. Chem. Soc., 1977, 99, 6027.
The results presented here suggest that ionic liquids are
indeed polar solvents that can, under appropriate circum-
stances, act as both hydrogen bond donors and hydrogen bond
acceptors. The p* values are uniformly high and characteristic
of ionic liquids and there is unlikely to be very much synthetic
flexibility in this parameter. a and b vary more widely with the
selection of the ions composing the ionic liquid and can be
deliberately manipulated. It is particularly exciting that the
relationships between DFT calculations of gas phase acidity,
5
6
7
8
9
M. J. Kamlet, J. L. Addoud, M. H. Abraham and R. W. Taft,
J. Org. Chem., 1983, 48, 2877.
R. M. C. Gon c¸ alves, A. M. N. Sim o˜ es, L. M. P. C. Albuquerque,
M. Ros e´ s, C. R a` fols and E. Bosch, J. Chem. Res. (S), 1993, 214.
P. Nicolet and C. Laurence, J. Chem. Soc. Perkin Trans. 2, 1986,
1
071.
M. J. Muldoon, C. M. Gordon and I. R. Dunkin, J. Chem. Soc.,
Perkin Trans. 2, 2001, 433.
K. Fletcher and S. Pandey, Appl. Spectrosc., 2002, 56, 266.
10 S. V. Dzyuba and R. A. Bartsch, Tetrahedron Lett., 2002, 43,
657.
+
4
b, [Cu(acac)(tmen)] basicity measurements and the rates of
1
1
(a) P. H. Shetty, P. J. Youngberg, B. R. Kersten and C. F. Poole,
J. Chromatogr., 1987, 411, 61; (b) S. K. Poole, P. H. Shetty and
C. F. Poole, Anal. Chim. Acta, 1989, 218, 241.
S. Tait and R. A. Osteryoung, Inorg. Chem., 1984, 23, 4352.
A. G. Avent, P. A. Chaloner, M. P. Day, K. R. Seddon and
T. Welton, J. Chem. Soc., Dalton Trans., 1994, 3405.
transition metal catalysed reactions opens the possibility for
designing the chemical function of the ionic liquids from
scratch for the first time.
1
1
2
3
z We now believe that the unexpectedly high basicity of the
[
bmim][SbF
6
] ionic liquid in the solvation study is the result of decom-
14 A. Elaiwi, P. B. Hitchcock, K. R. Seddon, N. Srinivasan, Y.-M.
Tan, T. Welton and J. A. Zora, J. Chem. Soc., Dalton Trans.,
1995, 3467.
position of the ionic liquid during transport and it has been removed
from this comparison.
Phys. Chem. Chem. Phys., 2003, 5, 2790–2794
2793

View Online
22 J. G. Huddleston, G. A. Broker, H. D. Willauer and R. D.
Rogers, in Ionic Liquids: Industrial Applications for Green Chemis-
try, ACS Symposium Series No.818, 2002.
23 P. Wasserschied, C. M. Gordon, C. Hilgers, M. J. Muldoon and
I. R. Dunkin, Chem. Commun., 2001, 1186.
24 J. L. Anderson, J. Ding, T. Welton and D. W. Armstrong, J. Am.
Chem. Soc., 2002, 124, 14 247.
25 L. Cammarata, S. G. Kazarian, P. Salter and T. Welton, Phys.
Chem. Chem. Phys., 2001, 3, 5192.
26 N. L. Lancaster, P. A. Salter, T. Welton and G. B. Young, J. Org.
Chem., 2002, 67, 8855.
27 I. A. Koppel, personal communication.
15
16
17
18
19
20
21
S. Scheiner, S. J. Grabowski and T. Kar, J. Phys. Chem., 2001,
05, 10 607.
C. E. Cannizzaro and K. N. Houk, J. Am. Chem. Soc., 2002, 124,
163.
T. Van Mourik and F. B. Van Duijneveldt, J. Mol. Struct.
THEOCHEM), 1995, 341, 63.
A. Aggarwal, N. L. Lancaster, A. R. Sethi and T. Welton, Green
Chem., 2002, 4, 517.
I. A. Koppel, P. Burk, I. Koppel, I. Leito, T. Sonoda and M.
Mishima, J. Am. Chem. Soc., 2000, 122, 5114.
A. J. Carmichael and K. R. Seddon, J. Phys. Org. Chem., 2000,
1
7
(
13, 591.
S. N. V. K. Aki, J. F. Brennecke and A. Samanta, Chem. Com-
mun., 2001, 413.
28 I. Persson, Pure Appl. Chem., 1986, 58, 1153.
2
794
Phys. Chem. Chem. Phys., 2003, 5, 2790–2794