Nucleophilicity in ionic liquids. 2.1 Cation effects on halide nucleophilicity in a series of bis(trifluoromethylsulfonyl)imide ionic liquids

Article abstract of DOI:10.1021/jo026113d

In this work, the nucleophilicities of chloride, bromide, and iodide have been determined in the ionic liquids [bmim] [N(Tf)₂], [bm₂im][N(Tf)₂], and [bmpy][N(Tf)₂] (where bmim = 1-butyl-3-methylimidazolium, bm₂im = 1-butyl-2,3-dimethylimidazolium, bmpy = 1-butyl-l-methylpyrrolidinium, and N(Tf)₂ = bis(trifluoromethylsulfonyl)imide). It was found that in the [bmim]⁺ ionic liquid, chloride was the least nucleophilic halide, but that changing the cation of the ionic liquid affected the relative nucleophilicities of the halides. The activation parameters ΔH^?, ΔS^?, and ΔG^? have been estimated for the reaction of chloride in each ionic liquid, and compared to a similar reaction in dichloromethane, where these parameters were found for reaction by both the free ion and the ion pair.

Full text of DOI:10.1021/jo026113d

Nu cleop h ilicity in Ion ic Liqu id s. 2.1 Ca tion Effects on Ha lid e
Nu cleop h ilicity in a Ser ies of Bis(tr iflu or om eth ylsu lfon yl)im id e
Ion ic Liqu id s
N. Llewellyn Lancaster, Paul A. Salter, Tom Welton,* and G. Brent Young
Department of Chemistry, Imperial College of Science, Technology and Medicine,
Exhibition Road, London SW7 2AY, U.K.
t.welton@ic.ac.uk
Received J une 28, 2002
In this work, the nucleophilicities of chloride, bromide, and iodide have been determined in the
ionic liquids [bmim][N(Tf)2], [bm2im][N(Tf)2], and [bmpy][N(Tf)2] (where bmim ) 1-butyl-3-
methylimidazolium, bm2im ) 1-butyl-2,3-dimethylimidazolium, bmpy ) 1-butyl-1-methylpyrroli-
dinium, and N(Tf)2 ) bis(trifluoromethylsulfonyl)imide). It was found that in the [bmim]+ ionic
liquid, chloride was the least nucleophilic halide, but that changing the cation of the ionic liquid
affected the relative nucleophilicities of the halides. The activation parameters ∆Hq, ∆Sq, and ∆Gq
have been estimated for the reaction of chloride in each ionic liquid, and compared to a similar
reaction in dichloromethane, where these parameters were found for reaction by both the free ion
and the ion pair.
In tr od u ction
In the search for alternative, environmentally benign
solvents, ionic liquids have become one of the key
focuses.2 They have been used as solvents for a wide
range of chemical processes, both stoichiometric and
catalytic. However, there has been little work directed
at what effect ionic liquids might have upon solute
reactivity.
F IGURE 1. Cations from which the ionic liquids were derived.
It is well-known that the microenvironment generated
by a solvent can change the outcome of a reaction, in
terms of both equilibria and rates.3 Since ionic liquids
have the potential to provide reaction media that are
quite unlike any other available at room temperature, it
is possible that they will have dramatic effects on
reactions in them. Indeed, there have been many claims
of great improvements in reaction yields and rates when
using ionic liquids.4 To understand how such effects may
arise, we have chosen to study a range of relatively
simple and well-understood reactions in a variety of ionic
liquids. The work described in this paper forms part of
this study.
nucleophilicity. Our methodology to determine the nu-
cleophilicity of halides in ionic liquids has been previously
reported for [bmim][BF4].5 A similar substrate has also
been used to determine halide nucleophilicity in a tet-
raalkylammonium tetraalkylboride ionic liquid.6 The only
other quantitative study of nucleophilic substitution
reported in an ionic liquid is that of chloride by cyanide
in [bmim][PF6],7 in which the effect of temperature on
the rate of reaction of potassium cyanide with benzyl
chloride was described.
The cations used to prepare the ionic liquids used in
this study are shown, with their adopted abbreviations,
in Figure 1.
The reaction chosen for study is that of methyl p-
nitrobenzenesulfonate, 1, with halide to give methyl
halide and the p-nitrobenzenesulfonate anion, 2 (Scheme
1). This system was chosen because both the substrate
and the anion product have convenient λmax values to
study this reaction by UV spectroscopy, at 253 and 275
nm, respectively. This substrate has been used in some
Ionic liquids are composed of anions and cations, either
of which may interact with solutes and therefore affect
the outcome of a reaction. In this work we have investi-
gated the effect of altering the cation upon halide
(1) Part 1: see ref 5.
(2) (a) Welton, T. Chem. Rev. 1999, 99, 2071-2083. (b) Wassercheid,
P.; Keim, W. Angew. Chem., Int. Ed. 2000, 39, 3772-3789. (c) Sheldon,
R. Chem. Commun. 2001, 2399-2407. (d) Gordon, C. M. Appl. Catal.,
A 2001, 222, 101-117. (e) Olivier-Bourbigou, H.; Magna, L. J . Mol.
Catal. A 2002, 182, 419-437. (f) Zhao, D.; Wu, M.; Kou, Y.; Min, E.
Catal. Today 2002, 74, 157-189.
(3) Reichardt, C. Solvents and Solvent Effects in Organic Chemistry,
2nd ed.; VCH (UK) Ltd.: Cambridge, U.K., 1998.
(4) Ionic Liquids in Synthesis; Wasserscheid, P., Welton, T., Eds.;
VCH Wiley: Weinheim, Germany, 2002; see also references therein.
(5) Lancaster, N. L.; Welton, T.; Young, G. B. J . Chem. Soc., Perkin
Trans. 2 2001, 2267-2270.
(6) Ford, W. T.; Hauri, R. J .; Smith, S. G. J . Am. Chem. Soc. 1974,
96, 4316-4318.
(7) Wheeler, C.; West, K. N.; Liotta, C. L.; Eckert, C. A. Chem.
Commun. 2001, 887-888.
10.1021/jo026113d CCC: $22.00 © 2002 American Chemical Society
Published on Web 11/12/2002
J . Org. Chem. 2002, 67, 8855-8861
8855

Lancaster et al.
SCHEME 1
studies in molecular solvents to determine and compare
the nucleophilicities of anions and neutral species within
those solvents.8-12 The use of alkyl sulfonates as probes
of nucleophilicity in molecular solvents and in ionic
liquids is well established.13
In molecular solvents, the reaction can proceed either
through the discrete anion (eq 1) or through an ion pair
(eq 2), where X is a halide and M is a cation.
F IGURE 2. UV spectra of [bmim]Cl (0.125 M) with 1 in
[bmim][N(Tf)2] at 25 °C taken at 150 s intervals.
X- + 1 f 2 + MeX
Mδ+ δ- + 1 f 2 + MeX
(1)
(2)
cally significant species are formed during this reaction.
When the halide was bromide, the wavelength of the
substrate could not be studied, though the isosbestic point
was observed. When iodide was used, it was necessary
to take values of absorbance from the shoulder of the
product peak (ca. 300 nm); the isosbestic point was not
observed.
X
In the ionic liquid, we would expect the anion to be
coordinated by cations (which is not necessarily the same
as an ion pair), and the discrete anion would not be
expected to be available to react in an ionic liquid. Hence,
only one reaction path is expected. To avoid the confusion
that might arise if mixtures of cations were present, the
halide was usually introduced as the salt of the cation
under study. However, [emim]I was used to study iodide
in [bmim][N(Tf)2] because [bmim]I was obtained only as
a yellow oil which could not be further purified.
Two sets of experiments were performed in each ionic
liquid. The first was to use a range of concentrations of
each halide at 25 °C to determine the second-order rate
constant for a given halide in a given ionic liquid at this
temperature. In the second set of reactions, the concen-
tration of chloride was kept constant while the temper-
ature was varied, allowing determination of ∆Hq and ∆Sq.
Deter m in a tion of Ha lid e Nu cleop h ilicity. The
observed rate constants, kobs, were found in the same way
as described previously,5 by a least-squares fitting pro-
cedure. In all of the reactions shown, the standard
deviation in the fit was always less than 5% of the quoted
value, and was more usually between 1% and 2% of the
quoted value. Values of k2 were determined by plotting
kobs against [halide]0 and finding the gradient. In each
case the intercept was allowed to float, but was always
close to zero (as expected). Comparison of these values
of k2 allows direct comparison of the relative nucleophi-
licities of the halides in a given ionic liquid. Note that
standard deviations in the values of k2 quoted (in
parentheses) are the standard deviations in the fit of the
absolute values of kobs against halide concentrations.
The results for the reactions of [bmim]Cl, [bmim]Br,
and [emim]I with 1 in [bmim][N(Tf)2] are shown in
Table 1.
Resu lts a n d Discu ssion
P r elim in a r y F in d in gs a n d Meth od ology. The first
thing to note is the UV cutoff point of the ionic liquids:
Both [bmim][N(Tf)2] and [bm2im][N(Tf)2] have a UV cutoff
of approximately 240 nm, and [bmpy][N(Tf)2] has a UV
cutoff of 200 nm, which is to be expected as this ionic
liquid’s cation does not have a delocalized π-system.
These values indicate that it is possible to study the
chosen reaction by UV spectroscopy in ionic liquids.
To each pure ionic liquid was added an aliquot of the
substrate, and spectra were recorded over 24 h. In no
case was reaction of the substrate with the ionic liquid
observed. This shows both the low nucleophilicity of the
[N(Tf)2]- ion in these ionic liquids and that any Cl-
contamination that may have been left after the synthe-
sis, as either unreacted [cation]Cl or dissolved LiCl, had
been successfully removed by the purification procedure.
Once the quality of the ionic liquids was established, it
was possible to study the reaction of 1 with halides.
Typical UV spectra for the reaction of [bmim]Cl with
1 in [bmim][N(Tf)2] are shown in Figure 2. The figure
shows the λmax values for 1 (at 253 nm) and for 2 (at 275
nm), as well as an isosbestic point at 263 nm. The
isosbestic point shows that this reaction is an example
of an “A to B” reaction, and that no other spectroscopi-
In each case there was a linear (first-order) relation-
ship between the initial concentration of halide and kobs
,
allowing assessment of k2 values. These data show that
iodide is the most nucleophilic of the halides in the
[bmim]+ ionic liquid, followed by bromide, with chloride
being the least nucleophilic.
The results for the reactions of [bm2im]Cl, [bm2im]Br,
and [bm2im]I with 1 in [bm2im] [N(Tf)2] are shown in
Table 2.
As in the [bmim]+ ionic liquid, a linear dependence of
kobs upon halide concentration was observed with each
of the halides, allowing the value of k2 to be determined.
However, these k2 values reveal that chloride is now the
most nucleophilic of the halides, with iodide being slightly
more nucleophilic than bromide. These results showing
a change in relative k2 values might appear surprising,
(8) Alluni, S.; Pero, A.; Reichenbach, G. J . Chem. Soc., Perkin Trans.
2 1998, 1747-1750.
(9) Alluni, S.; Pica, M.; Reichenbach, G. J . Phys. Org. Chem. 2001,
14, 265-270.
(10) Bunting, J . W.; Mason, J . M.; Heo, C. K. M. J . Chem. Soc.,
Perkin Trans. 2 1994, 2291-2300.
(11) Kurz, J . L.; Lu, J . Y.-W. J . Phys. Chem. 1983, 87, 1444-1448.
(12) Dietze, P. E.; Hariri, R.; Khattak, J . J . Org. Chem. 1989, 54,
3317-3320.
(13) Reference 3, Table 5-15, p 215 and references therein.
8856 J . Org. Chem., Vol. 67, No. 25, 2002

Nucleophilicity in Ionic Liquids
TABLE 1. Obser ved Ra tes of Rea ction of Ha lid es w ith 1
in [bm im ][N(Tf)2] a t 25 ˚C
TABLE 3. Obser ved Ra tes of Rea ction of Ha lid es w ith 1
in [bm p y][N(Tf)2] a t 25 °C
[halide]0/M
[1]0/mM
103kobsa/s-1
k2/M-1 s-1
[halide]0/M
[1]0/mM
103kobsa/s-1
k2/M-1 s-1
[bmim]Cl
[bmpy]Cl
0.0304
0.125
0.253
0.372
0.282
0.281
0.282
0.282
0.664
1.30
2.87
4.85
0.0312
0.124
0.248
0.373
0.296
0.311
0.303
0.307
0.958
4.04
8.36
0.0124(0.0014)
0.0391(0.0026)
14.43
[bmim]Br
[bmpy]Br
0.0317
0.127
0.253
0.378
0.281
0.284
0.291
0.288
0.446
1.66
3.65
7.30
0.0307
0.123
0.244
0.366
0.297
0.296
0.294
0.291
0.589
2.31
4.90
8.18
0.0195(0.0027)
0.0226(0.0013)
[emim]I
[bmpy]I
0.0122
0.0303
0.0607
0.0901
0.120
0.281
0.284
0.281
0.284
0.288
0.332
0.776
1.41
2.16
2.84
0.0120
0.0304
0.0607
0.0919
0.121
0.293
0.287
0.291
0.296
0.287
0.209
0.514
1.00
1.62
2.27
0.0232(0.0040)
0.0188(0.0008)
a
a
The standard deviations in the values of kobs were less than
The standard deviations in the values of kobs were less than
2% when the nucleophile was Cl- or Br- and less than 3% when
the nucleophile was I-.
5%.
TABLE 2. Obser ved Ra tes of Rea ction of Ha lid es w ith 1
in [bm 2im ][N(Tf)2] a t 25 °C
TABLE 4. Secon d -Or d er Ra te Con sta n ts for th e
Rea ction of Ha lid es w ith 1 in Ion ic Liqu id s a t 25 °C,
Rela tive Ra te Con sta n ts, a n d a Com p a r ison w ith Tw o
P ola r Molecu la r Solven ts
[halide]0/M
[1]0/mM
103kobsa/s-1
k2/M-1 s-1
[bm2im]Cl
0.0309
0.0306
0.123
0.243
0.365
0.296
0.296
0.303
0.313
0.309
0.277
0.271
2.38
6.15
12.04
k2/M-1 s-1
relative k2 values
0.0296(0.0012)
solvent
Cl-
Br-
I-
Cl- Br-
I-
[bmim][N(Tf)2]
0.0124 0.0195 0.0232
[bm2im][N(Tf)2] 0.0296 0.0221 0.0238
0.64
1.34
1.73
1.21
2.26
1
1
1
1
1
1
1.19
1.08
0.83
[bm2im]Br
[bmpy][N(Tf)2]
0.0391 0.0226 0.0188
CH2Cl2, ion paira 0.51
0.42
0.46
0.0298
0.122
0.245
0.360
0.301
0.298
0.301
0.299
0.422
1.97
5.12
7.66
CH2Cl2, free iona 1.04
0.0221(0.0011)
(CF3)2CHOHb
0.00011 0.00045 0.00039 0.24
8.67
a
Values obtained at 22 °C, using bis(triphenylphosphoranylide-
ne)ammonium chloride or bromide.8 Values obtained at 50 °C,
b
[bm2im]I
using tetramethylammonium halide.12
0.0119
0.0300
0.0592
0.0899
0.120
0.296
0.292
0.301
0.301
0.301
0.352
0.731
1.44
2.23
2.88
0.0238(0.0004)
The first point of note is that the difference between
the slowest reaction in an ionic liquid and the fastest is
approximately a factor of 3. However, the reactions in
the ionic liquids are greatly (orders of magnitude)
decelerated in comparison to the reaction in dichlo-
romethane, whether this is by the free ion or the ion pair,
and much faster than in the hexafluoropropan-2-ol. All
of the solvents considered here are considered polar, with
(CF3)2CHOH being regarded as highly ionizing and
hydrogen bond donating but weakly nucleophilic.12
In comparing the relative rates of reaction by chloride,
for example, in the ionic liquids, one might consider that
in these related solvents the differences are due to
viscosities. Recently the viscosities of the ionic liquids
used in this work have been determined;14 at 25 °C the
order of increasing viscosity is [bmim][N(Tf)2] < [bmpy]-
[N(Tf)2] < [bm2im][N(Tf)2]. The data clearly show that
the reaction rates are not dependent on the viscosity of
the ionic liquid alone. This observation is in agreement
with other studies that we have made in ionic liquids
whose viscosities are known.15 Additionally, it is noted
that while these reactions are slower than in the less
viscous dichloromethane, they are also much faster than
a
The standard deviations in the values of kobs were less than
2%.
given that the cations of the two ionic liquids considered
thus far are imidazolium based. Clearly the structure of
the cation has had an effect upon organic reactivity
within the ionic liquid.
The results for the reactions of [bmpy]Cl, [bmpy]Br,
and [bmpy]I with 1 in [bmpy][N(Tf)2] are shown in
Table 3.
As before, a linear dependence of kobs upon initial halide
concentration was observed. In this ionic liquid, the
chloride ion is by far the most nucleophilic of the halides,
but we now observe a further change in relative nucleo-
philicities. The iodide anion is now less nucleophilic than
the bromide anion.
Su m m a r y of Nu cleop h ilicity Da ta . A summary of
the relative nucleophilicities of the halides in the ionic
liquids studied in this work is shown in Table 4. Com-
parison of the data for reactions of the same substrate
with different sources of halide in dichloromethane and
hexafluoropropan-2-ol is also made. The conditions used
for the study of nucleophilicity in the molecular solvents
are not identical to ours, but are similar enough to allow
meaningful comparisons to be made.
(14) Okoturo, O. O.; Pell, T.; VanderNoot, T. J . Private communica-
tion.
(15) Lancaster, N. L.; Welton, T.; Young, G. B. Unpublished data.
J . Org. Chem, Vol. 67, No. 25, 2002 8857

Lancaster et al.
in less viscous hexafluoropropan-2-ol. Therefore, it would
appear that the activation parameters are shown to
influence the rate of reaction. These will be considered
later in the text.
[bmim][N(Tf)2] the value of k2 is less than half the value
in [bm2im][N(Tf)2] and less than one-third the value in
[bmpy][N(Tf)2]. Clearly there is some interaction between
the ionic liquid and the Cl- ion that is lowering its
nucleophilicity in [bm2im][N(Tf)2] and yet more so in
[bmim][N(Tf)2].
It is well-known that the rates of nucleophilic substitu-
tions are solvent dependent. This dependence has been
interpreted with reference to the nature of the charge
distributions in the ground state and the activated
complex in the rate-determining step by Hughes and
Ingold.16,17 In the studied reaction, the nucleophiles used
can be approximated as point charges which then com-
bine with the substrate 1 to give an activated complex
with a single negative charge distributed over several
atoms. Transferring the reaction from a nonpolar to a
polar solvent is expected to stabilize the ground state
with respect to the activated complex, thus slowing the
reaction. The energy differences between the ground and
excited states can also be influenced by other interac-
tions, such as H-bonding.
It then remains to discuss the differences between the
ionic liquids. In the ionic liquids studied bromide nucleo-
philicity, k2, was approximately constant, while iodide
showed a little variation and chloride showed more.
Therefore, we used the Br- nucleophilicity as our bench-
mark and compared the Cl- and I- values to that
(Table 4).
In [bmpy][N(Tf)2] chloride is the most nucleophilic
halide followed by bromide and finally iodide. This is in
very good agreement with Ford et al.,6 who used trieth-
ylhexylammonium triethylhexylboride ionic liquid and
found that the relative nucleophilicities of Cl-:Br-:I- were
1.8:1:0.83, respectively. It is also in agreement with the
known gas-phase nucleophilicities, which follow the order
chloride > bromide.18 This suggests that any interactions
between the ionic liquids and the reagents act equally
on all three halides in the two different ionic liquids.
In [bmim][N(Tf)2] chloride is the least nucleophilic of
the halides and iodide the most nucleophilic, a complete
reversal of the previously observed trend. This clearly
indicates that some influence of the ionic liquid is acting
differentially on the three halides, making the chloride
less reactive and the iodide more so. The [bm2im][N(Tf)2]
ionic liquid is in some way intermediate in character,
with the nucleophilicities changing in the order Cl- > I-
> Br-.
An alternative way to view the data is by how the
nucleophilicities of the different halides change in going
from one ionic liquid to another. As noted above, the k2
values for bromide in the three ionic liquids are similar,
though not identical. The k2 values for iodide are the
same in the two imidazolium-based ionic liquids and only
marginally lower in the [bmpy][N(Tf)2] ionic liquid.
However, the rate of reaction of the chloride shows a
great deal more variation. For the reaction of Cl- in
It is well established that a strong hydrogen bond is
formed between halide ions and [bmim]+ and to a lesser
extent [bm2im]+.19 The [bmpy]+ cation, on the other hand,
would not be expected to act as such a strong hydrogen
bond donor. However, studies have shown that strong
hydrogen bonds can be formed via the hydrogen atom
on a carbon R to a quaternary nitrogen (specifically
R3N+-C-H‚‚‚, a cation to which [bmpy]+ is analogous).20
Chloride is the best hydrogen bond acceptor of the halides
(hardest, most charge dense, and most coordinating). The
change in observed nucleophilicities can therefore be
explained by the degree of stabilization of the chloride
ion via hydrogen bonding to the cation of the ionic liquid.
The similarity of nucleophilicities in the tetraalkylam-
monium tetraalkylboride ionic liquid to that observed in
[bmpy]+ can also be understood by the argument that
these cations coordinate anions more poorly. Recently the
effect of hydrogen bonding in ionic liquids upon organic
reactions (specifically the Diels-Alder reaction) have
been speculated upon.21,22
Deter m in a tion of Activa tion P a r a m eter s. The
activation enthalpy (∆Hq) and activation entropy (∆Sq)
were both determined using the Eyring equation (eq 3).
k2h
∆S ∆H
ln
)
-
(3)
(
)
kBT
R
RT
Given that we had shown that there was a linear
dependence of kobs upon [chloride]0 at 25 °C, with negli-
gible intercept, the values of k2 were determined using
one value of [chloride]0 only at each temperature. From
these data, we can evaluate the influence of activation
enthalpy and entropy upon the reactions studied.
Although we might expect the reaction to be promoted
by a polar solvent because the transition complex will
be stabilized, the Hughes-Ingold approach to solvent
effects on the rates of nucleophilic substitutions16,17 states
that such reactions will be slightly slowed by polar
solvents because the point charge nucleophiles are sta-
bilized even more.
There are two good experimental reasons for studying
only the reaction of chloride in this work. The first is that
the quality of spectra for chloride were better than those
achieved for bromide or iodide. The second is that, in
other work, we found that the activation parameters were
(19) (a) Avent, A. G.; Chaloner, P. A.; Day, M. P.; Seddon, K. R.;
Welton, T. J . Chem. Soc., Dalton Trans. 1994, 3405-3413. (b) Abdul-
Sada, A. K.; Al-J uaid, S.; Greenway, A. M.; Hitchcock, P. B.; Howells,
M. J .; Seddon, K. R.; Welton, T. Struct. Chem. 1990, 1, 391-394. (c)
Elaiwi, A.; Hitchcock, P. B.; Seddon, K. R.; Srinivasan, N.; Tan, Y.-M.;
Welton, T.; Zora, J . A. J . Chem. Soc., Dalton Trans. 1995, 3467-3472.
(20) Cannizzaro, C. E.; Houk, K. N. J . Am. Chem. Soc. 2002, 124,
7163-7169.
(16) Ingold, C. K. Structure and Mechanism in Organic Chemistry,
2nd ed.; Bell: London, 1969.
(17) (a) Hughes, E. D.; Ingold, C. K. J . Chem. Soc. 1935, 244-255.
(b) Hughes, E. D. Trans. Faraday Soc. 1941, 37, 603-632. (c) Hughes,
E. D.; Ingold, C. K. Trans. Faraday Soc. 1941, 37, 657-686. (d) Cooper,
K. A.; Dhar, M. L.; Hughes, E. D.; Ingold, C. K.; MacNulty, B. J .; Woolf,
L. I. J . Chem. Soc. 1948, 2043-2049.
(18) (a) Brauman, J . I.; Olmstead, W. N.; Lieder, C. A. J . Am. Chem.
Soc. 1974, 96, 4030-4031. (b) Lieder, C. A.; Brauman, J . I. J . Am.
Chem. Soc. 1977, 99, 4219-4228. (c) Tanaka, K.; Mackay, G. I.;
Payzant, J . D.; Bohme, D. K. Can. J . Chem. 1976, 54, 1643-1659.
(21) (a) Sethi, A. R.; Welton, T. In Ionic Liquids: Industrial
Applications to Green Chemistry; Rogers, R. D., Seddon, K. R., Eds.;
ACS Symposium Series No. 818; American Chemical Society, Wash-
ington, DC, 2002; pp 241-246. (b) Aggarwal, A.; Lancaster, N. L.; Sethi,
A. R.; Welton, T. Green Chem. 2002, 4, 517-520.
(22) Dzyuba, S. V.; Bartsch, R. A. Tetrahedron Lett. 2002, 43, 4657-
4659.
8858 J . Org. Chem., Vol. 67, No. 25, 2002

Nucleophilicity in Ionic Liquids
TABLE 5. Activa tion En th a lp ies, En tr op ies, a n d F r ee En er gies for th e Rea ction of Ch lor id e w ith 1
[chloride]0/M
k2/M-1 s-1
T/K
∆Hq/kJ mol-1
∆Sq/J K-1 mol-1
∆Gq298 K/kJ mol-1
[bmim][N(Tf)2]
0.116
0.125
0.125
0.00381
0.0104
0.0273
288.15
198.15
308.15
71.8(1.6)
-42.2(5.4)
84.4(3.2)
[bm2im][N(Tf)2]
71.8(1.6)
0.122
0.123
0.121
0.00696
0.0194
0.0500
288.15
198.15
308.15
-37.3(5.3)
-43.9(6.0)
82.9(3.2)
81.6(3.6)
[bmpy][N(Tf)2]
68.5(1.8)
0.122
0.122
0.124
0.123
0.123
0.0117
0.0202
0.0326
0.0498
0.0799
288.15
293.15
298.15
303.15
308.15
CH2Cl2, ion-pair8
CH2Cl2, free ion8
79.5(2.9)
54.4(2.5)
7.9(10)
-58.6(8.4)
77.2(5.9)
71.9(5.0)
a
The values of kobs from which the k2 values were derived show standard deviations of 2% or less.
the same for both chloride and bromide.5 Values for iodide
have not been determined by us for any ionic liquid
system. The data are shown in Table 5. Note that the
standard deviations for the values of ∆Hq and ∆Sq (given
in parentheses) represent the quality of the fit of the data
to eq 3, and contain no allowance for the quality of k2.
The values of kobs from which k2 is derived typically had
a standard deviation of less than 2% of the value derived
from the fit of absorbance/time data to the first-order
F IGURE 3. Activated complex.
dominated by cations. In fact recent structural studies
in a related system confirm this expectation, and this is
discussed later.23 It is noteworthy that our data suggest
∆Hq to be more similar to that for the ion pair in
dichloromethane than that for the free ion in dichlo-
romethane. An ion pair is a cation and anion that are
coordinated to some extent and contained within a
solvent shell; in our system that solvent shell is ionic,
but is still more analogous to an ion pair than a free
anion.
kinetic model. Additionally, ∆Gq
values have been
298 K
determined for each system (eq 4); standard deviations
are given, and the same caveats apply as stated previ-
ously.
∆Gq ) ∆Hq - T∆Sq
(4)
That the ion-pair-like behavior should be observed is
not without precedent. Similar behavior is seen in the
influence of molecular solvation of ions in the gas phase,
where a single molecule of the “solvent” can interact with
the reagent and convert the reactivity to give a solution-
like behavior. In this system, the coordination of a solvent
molecule (the cation) is giving ion-pair-like behavior. It
might be argued that the number of additional cations
around the anion is not important, since our system that
contains several cations about each halide anion behaves
in an analogous mode to that for an ion pair in a
molecular solvent.
The ∆Sq values, however, are more similar to those for
the reaction by the free ion in dichloromethane than for
the reaction by the ion pair in dichloromethane.8 Since
the activation enthalpies are so similar to those for the
ion pair in dichloromethane, this needs to be explained.
This reaction follows an SN2 reaction pathway and will
form an activated complex as shown in Figure 3.
The activation step of an SN2 reaction is an associative
process, and one would expect it to have a negative
entropy. In dichloromethane the reaction by the ion pair
has a small positive activation entropy because, as the
Also given in the table are values for the chloride ion
and for the chloride ion pair in dichloromethane for the
reaction of the same substrate with chloride, where the
cation was tetrabutylammonium ion.8 It should be noted
that there is very little comparative data available in the
literature, as most studies have been more concerned
with determining the relative nucleophilicities of re-
agents in given solvents or else determining the sub-
stituent effects on the substrate.
Values of ∆Hq are similar to those for the ion pair in
dichloromethane. Therefore, the effect of the ionic liquid
on the activation enthalpy does not arise from general
“polarity” differences between the ionic liquids and
dichloromethane, but rather the association of the chlo-
ride with a cation or cations. This suggests that the same
process is occurring in the rate-limiting step in both
cases, i.e., as the chlorine-carbon bond is being formed
in the activated complex, a cation-chloride association
is being broken with an associated enthalpic cost. This
is opposed to the “free ion” case where the chlorine-
carbon bond is formed without the concomitant breaking
of another interaction, giving a lower activation enthalpy.
In an ionic liquid, where there is no molecular solvent
present to separate the ionic species, a free (molecular
solvent solvated) ion cannot occur. An ionic species would
always be expected to be coordinated by one or more
counterions of the ionic liquid, and we would expect that
the chloride will always have its first coordination sphere
(23) Hardacre, C.; Holbrey, J . D.; McMath, S. E. J .; Bowron, D. T.;
Sopev, A. K. In Ionic Liquids: Industrial Applications to Green
Chemistry; Rogers, R. D., Seddon, K. R., Eds.; ACS Symposium Series
No. 818; American Chemical Society: Washington, DC, 2002; pp 400-
412.
J . Org. Chem, Vol. 67, No. 25, 2002 8859

Lancaster et al.
SCHEME 2
in this work, some initial proposals can be made. Har-
dacre et al.23 have demonstrated by neutron diffraction
that a Cl- ion is surrounded by six cations within 6.5 Å
of the halide in the related [mmim]Cl ([mmim]+ ) 1,3-
dimethylimidazolium cation). For the reaction to occur,
the Cl- ion must first come into close proximity with the
substrate 1. To do this, the Cl- ion must dissociate from
at least one cation. Our data suggest that the order of
availability of chloride to react is [bmim][N(Tf)2] <
[bm2im][N(Tf)2] < [bmpy][N(Tf)2], which is in the reverse
order of the strength of the cation-chloride interaction.
activated complex is formed, the cation of the ion pair is
liberated as a free solvated cation. This process counter-
balances the loss of entropy associated with the formation
of the activated complex. When performed in dichlo-
romethane, the anionic leaving group 2 does not form an
ion pair in dichloromethane but becomes a free solvated
anion. In the ionic liquid 2 does associate with the cations
of the ionic liquid. It is proposed that the entropy gained
by liberating a cation from its association with the
chloride ion is cancelled out by the association of another
cation with 2. Hence, the activation entropy for the
reaction was what might be expected for an SN2 process,
with a value similar to that for reaction by the free ion
in dichloromethane.
Con clu sion s
The first thing that becomes obvious in this work is
that not all ionic liquids are the same. This work begins
to show that one cannot simply take a conventional
organic reaction and replace the solvent with a single
ionic liquid and then expect the result to be the same as
would be achieved in all other ionic liquids. However, by
the same token this work reveals that the claims made,24
that ionic liquids can be tailor-made for a given reaction,
are true. It is possible to imagine that ionic liquids can
be made to have the ideal combination of cation and
anion for a given reaction. Although this study only
concentrates on SN2 reactions, we can generalize to say
that, when using ionic liquids for reactions of highly
associating anions such as halides, reaction rates will
probably be greater in ionic liquids composed of the least
coordinating (poor hydrogen bond acids) cations.
What we cannot say is that there is an “ionic liquid
effect” on this reaction, rather that the reaction rates for
this reaction of halides with methyl p-nitrobenzene-
sulfonate are found to be within the range observed for
the same reaction in molecular solvents. In fact the
relative rates of reaction can be explained (and predicted)
by the classical Hughes-Ingold approach to examining
solvent effects on organic reactions.
We have shown that the nucleophilicity of the halides
is ionic liquid specific, just as nucleophilicity is solvent
specific in molecular solvents, with the ability of the
cation to act as a hydrogen bond donor being a key
feature of this dependence in this reaction. We have
published elsewhere the relative nucleophilicities of the
halides in [bmim][BF4]5 and found them to be (Cl-) 1.06:
(Br-) 1:(I-) 1.41. The trend of halide nucleophilicity in
[bmim][N(Tf)2] is not the same as in [bmim][BF4], and
shows that the nucleophilicity is determined by a com-
bination of cation and anion properties, and that the
explanation of nucleophilicities given above in this work
is only part of the answer. We will report the details of
the effect of the ionic liquid anions on these reactions at
a later date.
Finally, the ∆Gq
values have been compared, and
298 K
a slight trend emerges. It appears that the highest ∆Gq
value is observed in the [bmim]+ ionic liquid, and that it
falls a little for [bm2im]+, and still more for [bmpy]+. This
trend is to be expected, given that these values should
follow the same relationship as the k2 values. These ∆Gq
values are higher than those observed for reaction by
either the free ion or the ion pair in dichloromethane,
suggesting, as one would expect, that in fact this reaction
(with its charged reagent and its transition state with a
delocalized charge) is slightly less favored in ionic liquids
than in dichloromethane. However, the activation pa-
rameters alone are not sufficient to explain the observed
differences in the nucleophilicities of the halides in the
different ionic liquids.
These parameters reveal the size of the entropy and
enthalpy barriers on going from the ground state of the
available reactive forms of the reagents and the activated
complex. The ground state for the reaction has first to
be achieved to give such an available chloride ion before
the reaction can occur. The evidence shows that the
chloride is not always available to react with the sub-
strate when dissolved in the ionic liquids, though there
is still a linear relationship between the concentration
of chloride and kobs. It is likely that this relationship will
break at higher concentrations where the ratio of chloride
to [bmim]+ approaches unity. There was no evidence of
that in the concentration range studied. So, it seems that
there is an equilibrium (Scheme 2) that must be ac-
counted for.
The left-hand side of the equilibrium represents a fully
coordinated, “unavailable” chloride, whereas on the right-
hand side one face of the chloride ion is exposed to the
substrate following the dissociation of one [bmim]+ cation,
giving an “available” chloride. This loose association of
available chloride with the substrate represents the
ground state for the reaction in this system. It should be
noted that UV/vis spectroscopy showed that there is no
significant interaction between available chloride and the
substrate to give rise to a formal intermediate. The extent
to which the nucleophile is coordinated by the cation will
affect K, and thus kobs and k2. This is consistent with our
results.
Exp er im en ta l Section
Ma ter ia ls. All syntheses were performed under anaerobic
conditions using standard Schlenk techniques. All heterocycles
were distilled from potassium hydroxide, alkyl halides from
phosphorus pentoxide, and solvents from standard drying
agents before use. Methyl p-nitrobenzenesulfonate and lithium
Although the nature of the complexes proposed above,
and of the equilibrium constant, has not been determined
(24) See, for example: Freemantle, M. Chem. Eng. News 1998, 76
(34, Aug 24), 12.
8860 J . Org. Chem., Vol. 67, No. 25, 2002

Nucleophilicity in Ionic Liquids
bis(trifluoromethylsulfonyl)imide were purchased from com-
mercial sources and used as received.
solid (9.8 g, 35 mmol, 87%) after recrystallization from ethyl
acetate. H NMR:
1
1H NMR spectra (270 MHz) were recorded in deuterated
DMSO, and chemical shifts are expressed in parts per million.
Mass spectra were recorded using FAB ionization. UV/vis
spectra were recorded using a PC-controlled spectrophotom-
eter, fitted with a thermostatted sample holder.
δH (ppm) 7.64 (s, 1H), 7.61 (s, 1H), 4.08 (t, 2H, J ) 7.4 Hz),
3.72 (s, 3H), 2.56 (s, 3H), 1.64 (quintet, 2H, J ) 7.4 Hz), 1.23
(sextet, 2H, J ) 7.4 Hz), 0.84 (t, 3H, J ) 7.4 Hz). MS (FAB+):
m/z 433 ([(bm2im)2I]+, 10), 153 ([bm2im]+, 100). MS (FAB-):
m/z 407 ([(bm2im)I2]-, 60), 127 ([I]-, 100).
The preparations and spectral data of the compounds
[bmim]Cl, [bm2im]Cl, [bmim][N(Tf)2], and [bm2im][N(Tf)2] are
described elsewhere.25 The ionic liquids [bmim][N(Tf)2] and
[bm2im][N(Tf)2] both had UV cutoffs of ca. 240 nm. It should
be noted that the yields of crude products were near quantita-
tive for all compounds listed below, and that low reported final
yields occurred only where material was lost during recrys-
tallization.
1-Bu tyl-1-m eth ylp yr r olid in iu m Ch lor id e, [bm p y]Cl. In
a Schlenk flask, 1-chlorobutane (221 cm3, 2.05 mol) was added
slowly with cooling to 1-methylpyrrolidine (200 cm3, 1.92 mol)
in propan-2-ol (200 cm3). The mixture was then brought to
reflux for 24 h. After cooling, the solvent was decanted to leave
transparent crystals, which were recrystallized with propan-
2-ol, washed with ethyl acetate, and then dried under vacuum,
1
giving colorless crystals (314 g, 1.76 mol, 92%). H NMR: δH
Kin etic Stu d ies. The reactions were studied by adding
methyl p-nitrobenzenesulfonate in dichloromethane (4.7 × 10-7
mol in 0.1 cm3) to a solution of halide in ionic liquid (1.5 cm3)
at known time. The complete experimental method used for
the study of these reactions by UV spectroscopy and the
subsequent data analysis are described elsewhere.5 Most of
the reactions were monitored for 6 half-lives or more.
1-Bu tyl-3-m eth ylim id a zoliu m Br om id e, [bm im ]Br . A
solution of 1-methylimidazole (16 cm3, 0.20 mol) in toluene (25
cm3) was prepared and cooled in an ice/water bath. To this
was added slowly 1-bromobutane (25 cm3, 0.23 mol) with
stirring. After addition, the mixture was allowed to warm to
room temperature, then heated at reflux for 24 h, and then
crystallized. The solid was purified by recrystallization from
acetonitrile to give a colorless, crystalline solid (38 g, 0.17 mol,
(ppm) 3.49-3.42 (m, 4H), 3.42-3.34 (m, 2H), 3.02 (s, 3H), 2.06
(br. s, 4H), 1.66 (quintet, 2H, J ) 7.8 Hz), 1.29 (sextet, 2H, J
) 7.8 Hz), 0.91 (3, t, 3H, J ) 7.8 Hz). MS (FAB+): m/z 319
([(bmpy)2Cl]+, 15), 142 ([bmpy]+, 100). MS (FAB-): m/z 212
([(bmpy)Cl2]-, 100), 37 ([37Cl]-, 34), 35 ([35Cl]-, 40).
1-Bu tyl-1-m eth ylp yr r olid in iu m Br om id e, [bm p y]Br .
[bmpy]Br was prepared from 1-methylpyrrolidine (4.5 cm3, 43
mmol) and 1-bromobutane (5.5 cm3, 51 mmol) by the same
method as [bmpy]Cl and obtained as a colorless, crystalline
solid (2.5 g, 11 mmol, 27%) after recrystallization from propan-
1
2-ol and ethyl acetate. H NMR:
δH (ppm) 3.54-3.34 (m, 4H), 3.32 (t, 2H, J ) 7.8 Hz), 2.99
(s, 3H), 2.07 (br. s, 4H), 1.68 (quintet, 2H, J ) 7.8 Hz), 1.31
(sextet, 2H, J ) 7.4 Hz), 0.93 (t, 3H, J ) 7.4 Hz). MS (FAB+):
m/z 363 ([(bmpy)2Br]+, 10), 142 ([bmpy]+, 100). MS (FAB-):
m/z 302 ([(bmpy)Br2]-, 100), 81 ([81Br]-, 27), 79, ([79Br]-, 40).
1-Bu tyl-1-m eth ylpyr r olidin iu m Iodide, [bm py]I. [bmpy]I
was prepared from 1-methylpyrrolidine (3.8 cm3, 37 mmol) and
1-iodobutane (5.0 cm3, 44 mmol) by the same method as [bmpy]-
Cl and obtained as a colorless, crystalline solid (4.5 g, 17 mmol,
38%) after recrystallization from propan-2-ol and ethyl acetate.
1H NMR:
δH (ppm) 3.58-3.40 (m, 4H), 3.40-3.28 (m, 2H), 3.00 (s, 3H),
2.08 (br. s, 4H), 1.68 (quintet, 2H, J ) 7.4 Hz), 1.31 (sextet,
2H, J ) 7.4 Hz), 0.92 (t, 3H, J ) 7.4 Hz). MS (FAB+): m/z 411
([(bmpy)2I]+, 15), 142 ([bmpy]+, 100). MS (FAB-): m/z 396
([(bmpy)I2]-, 100), 127 ([I]-, 75).
1
86%). H NMR:
δH (ppm) 9.23 (s, 1H), 7.77 (d, 1H, J ) 1.2 Hz), 7.70 (d, 1H,
J ) 1.2 Hz), 4.13 (t, 2H, J ) 7.4 Hz), 3.81 (s, 3H), 1.70 (quintet,
2H, J ) 7.4 Hz), 1.19 (sextet, 2H, J ) 7.4 Hz), 0.82 (t, 3H, J
) 7.4 Hz). MS (FAB+): m/z 357 ([(bmim)2Br]+, 10), 139
([bmim]+, 100). MS (FAB-): m/z 299 ([(bmim)Br2]-, 100), 81
([81Br]-, 92), 79 ([79Br]-, 92).
1-Eth yl-3-m eth ylim id a zoliu m Iod id e, [em im ]I. [emim]I
was prepared from 1-methylimidazole (1.7 cm3, 40 mmol) and
iodoethane (5.0 cm3, 44 mmol) by the same method as [bmim]-
Br and purified by recrystallization from ethyl acetate to give
1
a colorless, crystalline solid (9.8 g, 35 mmol, 87%). H NMR:
δH (ppm) 9.08 (s, 1H), 7.73 (s, 1H), 7.64 (s, 1H), 4.13 (t, 2H,
J ) 7.4 Hz), 3.78 (s, 3H), 1.34 (t, 3H, J ) 7.4 Hz). MS (FAB+):
m/z 349 ([(emim)2I]+, 15), 111 ([emim]+, 100). MS (FAB-): m/z
840 ([emim)3I4]-, 5), 603 ([(emim)2I3]-, 25), 365 ([(emim)I2]-,
100), 127 ([I]-, 55).
1-Bu tyl-1-m eth ylp yr r olid in iu m Bis(tr iflu or om eth yl-
su lfon yl)im id e, [bm p y][N(Tf)2]. In a Schlenk flask, a solu-
tion of [bmpy]Cl (35.8 g, 0.202 mol) in dichloromethane (50
cm3) was added to lithium bis(trifluoromethylsulfonyl)imide
(57.4 g, 0.200 mol). The resulting suspension was stirred for
72 h and then filtered. The residual salt was washed with
further aliquots of dichloromethane, and the combined organic
extracts were washed with water until the aqueous phase was
halide free (by silver nitrate test), after which the solvent was
removed in vacuo. The resulting liquid was treated with
activated charcoal and filtered through a pad of acidic alumina
1-Bu tyl-2,3-d im eth ylim id a zoliu m Br om id e, [bm 2im ]Br .
A solution of 1,2-dimethylimidazole (4.1 g, 42 mmol) in toluene
(5 cm3) was prepared and cooled in ice/water. To this was
added 1-bromobutane (5.5 cm3, 51 mmol) with stirring. After
addition was complete the solution was allowed to warm to
room temperature and then heated to reflux for 24 h. After
cooling, a solid formed which was purified by recrystallization
from ethyl acetate to give a colorless, crystalline solid (3.7 g,
1
to give a colorless liquid (73 g, 0.17 mol, 86%). H NMR:
1
16 mmol, 38%). H NMR:
δH (ppm) 3.65-3.32 (m, 4H), 3.30-3.24 (m, 2H), 2.96 (s, 3H),
2.08 (br. s, 4H), 1.65 (quintet, 2H, J ) 7.4 Hz), 1.30 (sextet,
2H, J ) 7.4 Hz), 0.93 (3, t, 3H, J ) 7.4 Hz). MS (FAB+): m/z
564 ([(bmpy)2(N(Tf)2)]+, 1), 142 ([bmpy]+, 100). MS (FAB-): m/z
702 ([(bmpy)(N(Tf)2)2]-, 5), 280 ([N(Tf)2]-, 100).
δH (ppm) 7.62 (s, 1H), 7.60 (s, 1H), 4.05 (t, 2H, J ) 7.4 Hz),
3.69 (s, 3H), 2.52 (s, 3H), 1.62 (quintet, 2H, J ) 7.4 Hz), 1.20
(sextet, 2H, J ) 7.4 Hz), 0.82 (t, 3H, J ) 7.4 Hz). MS (FAB+):
m/z 385 ([(bm2im)2Br]+, 10), 153 ([bm2im]+, 100). MS (FAB-):
m/z 313 ([(bm2im)Br2]-, 90), 79 ([Br]-, 100).
1-Bu t yl-2,3-d im et h ylim id a zoliu m Iod id e, [b m 2im ]I.
[bm2im]I was prepared from 1,2-dimethylimidazole (3.8 g, 40
mmol) and 1-iodobutane (5.0 cm3, 44 mmol) by the same
method as [bm2im]Br and obtained as a colorless, crystalline
Ack n ow led gm en t. We thank the Leverhulme trust
for funding this project, Miss O. O. Okoturo, Mr. T. Pell,
and Dr. T. J . VanderNoot for the data used in ref 14,
and Dr. C. Hardacre for a preprint of ref 23.
(25) Cammarata, L.; Kazarian, S. G.; Salter, P. A.; Welton, T. Phys.
Chem. Chem. Phys. 2001, 3, 5192-5200.
J O026113D
J . Org. Chem, Vol. 67, No. 25, 2002 8861