Lancaster and Welton
the reactions of chloride and iodide are slowest. These
observations again suggest that the binding of the anion
to the cation affects the halide-cation interaction and
thus the halide nucleophilicity. Structural investigations
of the ionic liquids clearly show that the N,N-dialkylimi-
dazolium based ionic liquids can self-associate to a
greater or lesser extent through hydrogen bonding.1
For any given cation the degree of hydrogen bonding is
controlled by the hydrogen bond acceptor properties of
the anion.
we have used the Eyring equation, which allows direct
determination of both ∆H and ∆S . Therefore, these data
are reproduced here, along with the data for the effect of
2
temperature on the reaction in [bmim][N(Tf) ].
Originally the reactions of both bromide and chloride
were studied at varied temperatures. Iodide was not
studied because of the lower quality of the spectra
generated in these reactions. When the studies of the
q
q
6,20
4 6
reaction of both anions in [bmim][BF ] and [bmim][PF ]
revealed that the activation parameters for each anion
were approximately the same in each ionic liquid, sub-
sequent studies made use of chloride ion only as the
nucleophile, chloride being a more satisfactory nucleo-
phile to study in these experiments. The results for all
of the reactions studied are given in Table 5.
A comparison of k
2
against the Abraham parameter
“a” did not reveal a direct correlation between the basicity
of the ionic liquid and halide nucleophilicity.
This leads us to consider other possible factors that
would affect the halide nucleophilicities, such as solvent
viscosity. Yet, although [bmim][N(Tf)
2
] is less viscous
The activation free energy follows the trend of k values
2
than [bmim][BF ], the reaction of each halide is faster
in the latter ionic liquid than in the former.
4
already recorded, as expected. It is noted that in the ionic
q
liquids, the values of ∆G
84 kJ mol . This contrasts to the reaction in dichlo-
romethane, where the value of ∆G 2
2
98 K
are in the range of 82 to
-
1
In Table 4 we have also made comparison against some
molecular solvents, where halide nucleophilicity was
measured by nucleophilic substitution on alkylsulfonates.
Ionic liquids are often compared to acetonitrile or metha-
nol by virtue of them all having similar empirical solvent
properties. As is clear in the table, halide nucleophilicity
is affected by these molecular solvents. In the polar protic
solvent, chloride is less nucleophilic than bromide than
iodide. However, in the polar aprotic solvent the trend
q
98 K
is significantly
lower, whether the reaction is by free chloride ion or by
q
the ion-pair. The values of ∆G 2
98 K
are very similar to
1
those observed when we studied the effect of the cation.
q
The quantity ∆G 2
98 K
is calculated from the changes
in enthalpy and entropy of activation, and it is the values
q
q
of ∆H and ∆S that are most revealing. In most of the
q
ionic liquids, ∆H is large, with values in the range of 70
-
1
is approximately reversed. The example of (CF
3
)
2
CHOH
to 80 kJ mol . These are comparable to the value
observed for the reaction by the ion-pair in dichloro-
methane. This is perhaps to be expected because the
picture of the halide ion in the ionic liquid, while not a
true ion-pair, is of the anion surrounded by cations. When
is included as an example of a very strong hydrogen bond
donating molecular solvent where the reactions of the
halides (introduced as the tetramethylammonium salts)
are even slower than those that we observe in the ionic
liquids.
q
the ionic liquid is [bmim][BF ], the values of ∆H for the
4
In summary, as in molecular solvents, using different
ionic liquids results in different nucleophilicities of the
halides (both relative and absolute). There is strong
evidence of hydrogen bonding related effects but, perhaps
not surprisingly, this is not the only contribution to the
changes observed.
reactions of chloride and bromide are much lower, being
-
1
in the range 53-56 kJ mol . These values are more
similar to the activation enthalpy for the reaction of the
free ion in dichloromethane.
q
In considering ∆S for this reaction, we note that as it
proceeds via an S 2 mechanism, we would expect a
N
Activa tion P a r a m eter s. To gain further insight into
the processes affecting these reactions, the activation
parameters were determined. Given that we had shown
that there was a linear dependence of kobs upon initial
halide concentration at 25 °C with negligible intercept,
negative activation entropy. This was observed in all
cases in the ionic liquids, although there is considerable
variation in the value in different ionic liquids and it is
noted that the reactions in [bmim][BF ] were accompa-
4
nied by a particularly large entropy barrier.
the values of k
2
were determined using one value of
In earlier work, we have proposed a reaction mecha-
nism as shown in Scheme 2, where initially the halide is
fully coordinated by a number of ionic liquid cations.
[
halide] only at each temperature. From these data we
0
1
can evaluate the influence of activation enthalpy and
entropy upon the reactions studied.
There is an equilibrium by which one face of the halide
becomes uncoordinated, making it able to act as a
nucleophile and giving rise to the observed reaction.
During the activation process the remaining cations
bound to the halide become less associated as the
negative charge becomes distributed over many atoms.
This dissociation of the hydrogen-bonded cations from the
halide to a greater or lesser degree compensates for the
The effect of temperature on the rate of reaction in
[
bmim][BF
4
] has been published.6 At that time the
activation parameters were determined by use of the
Arrhenius equation. This gives an activation energy
q
(
similar to ∆H ), but the determination of the activation
entropy from this expression is less straightforward. Here
(
17) Liotta, C. L.; Grisdale, E. E.; Hopkins, H. P. Tetrahedron Lett.
loss of entropy arising from the S
associative process. However, the data from the reaction
in [bmim][BF ] show that this compensation is not
occurring. This can be interpreted as the halide dissolved
into the [bmim][BF ] being much less completely coor-
N
2 reaction being an
1
975, 4205.
(
(
18) Pearson, R. G.; Songstad, J . J . Org. Chem. 1967, 32, 2899.
19) Rodewald, R. F.; Mahendran, K.; Bear, J . L.; Fuchs, R. J . Am.
4
Chem. Soc. 1968, 90, 6698.
20) (a) Downard, A.; Earle, M. J .; Hardacre, C.; Mcmath, S. E. J .;
(
4
Nieuwenhuyzen, M.; Teat, S. J . Chem. Mater. 2004, 16, 43. (b) Saha,
S.; Hayashi, S.; Kobayashi, A.; Hamaguchi, H. Chem. Lett. 2003, 32,
dinated, giving a system of lower entropy than solutions
of halide in the other ionic liquids. Therefore the activa-
7
40. (c) Abdul-Sada, A. K.; Greenway, A. M.; Hitchcock, P. B.;
Mohammed T. J .; Seddon K. R.; Zora, J . A J . Chem. Soc., Chem.
Commun. 1986, 1753. (d) Fuller, J .; Carlin, R. T.; De Long, H. C.;
Haworth, D. J . Chem. Soc., Chem. Commun. 1994, 299.
tion entropy is much larger in [bmim][BF
4
] than in the
other ionic liquids. This suggestion also makes sense of
5
990 J . Org. Chem., Vol. 69, No. 18, 2004