Unexpected side reactions of imidazolium-based ionic liquids in the base-catalysed Baylis-Hillman reaction

Article abstract of DOI:10.1039/b203079a

Low yields are obtained when the Baylis-Hillman reaction is conducted in the presence of an imidazolium-based ionic liquid due to direct addition of the deprotonated imidazolium salt to the aldehyde. Ionic liquids are evidently not inert.

Full text of DOI:10.1039/b203079a

Unexpected side reactions of imidazolium-based ionic liquids in the
base-catalysed Baylis–Hillman reaction†
Varinder K. Aggarwal,* Ingo Emme and Andrea Mereu
School of Chemistry, University of Bristol, Bristol, UK BS8 1TS. E-mail: v.aggarwal@bristol.ac.uk
Received (in Cambridge, UK) 28th March 2002, Accepted 16th April 2002
First published as an Advance Article on the web 26th June 2002
Table 1 Conversion and yield of Baylis–Hillman reaction (Scheme 1)a
Low yields are obtained when the Baylis–Hillman reaction is
conducted in the presence of an imidazolium-based ionic
Entry BMIM Cl
3-HQD
Conversionb after 4 h Yieldc after 4 h
liquid due to direct addition of the deprotonated imidazo-
lium salt to the aldehyde. Ionic liquids are evidently not
inert.
1
2
3
4
—
0.1 eq.
0.1 eq.
0.2 eq.
0.5 eq.
12d
69
85
86
12d
25
36
37
0.5 eq.
0.5 eq.
0.5 eq.
Since the pioneering work of Seddon, ionic liquids have
emerged as a new class of stable, inert organic solvents with
unique properties which can provide considerable benefit in
synthesis.1 The immiscibility of ionic liquids with a number of
organic solvents and their zero vapour pressure aid their
recyclability. Their high polarity and ability to solubilise both
inorganic and organic materials can result in enhanced rates of
chemical processes and can provide higher/different selectiv-
ities compared to conventional solvents. Thus, as a result of
their ‘green’ credentials and potential to enhance rates and
selectivities, ionic liquids are finding increasing applications in
synthesis. They have been employed under acidic, basic and
neutral conditions and their stability under different conditions
has usually been assumed.
We have been interested in enhancing rates of Baylis–
Hillman reactions2 and some time ago considered the use of
ionic liquids as we believed that their high polarity would
provide a higher equilibrium concentration of the zwitterionic
intermediate 1 and thereby lead to higher rates (Scheme 1).
However, whilst a moderate increase in rate was indeed
observed, yields were lower than in the absence of the ionic
liquid and so we decided not to publish our results.3 We were
therefore surprised to read a recent report by Afonso on the
beneficial properties of employing imidazolium salts in Baylis–
Hillman reactions.4 As a result of this publication we decided to
carry out a more detailed study of our initial reactions and now
report the finding that imidazolium salts are not inert, but that
under mild basic conditions they are deprotonated to give
reactive nucleophiles, which consume the aldehyde leading to
lower yields. Indeed, the ease of deprotonation of this class of
ionic liquids and subsequent reaction with electrophiles seems
to have been overlooked by a considerable section of the
chemistry community. We not only address the problems
associated with the use of ionic liquids in the Baylis–Hillman
reaction, but also provide corrections and explanations of
Afonso’s results and other literature examples where imidazo-
lium salts have been employed under basic conditions.
a Conditions: Benzaldehyde (1.0 eq.), methyl acrylate (1.2 eq.), 3-hydrox-
yquinuclidine (3-HQD, 0.1, 0.2 and 0.5 eq.) and butylmethylimidazolium
chloride (BMIM Cl, 0.5 eq.). b Conversion [BH-Product / (BH-Product +
Benzaldehyde)] was measured by 1H NMR spectroscopy. c Yields were
calculated by 1H NMR using Baylis–Hillman product signals relative to
3-HQD (used as internal standard). d Measured after 6 h.
Baylis–Hillman adduct relative to starting materials after 4 h
(Table 1).
However, in analysing the NMR data we found that
benzaldehyde was consumed at a much faster rate than
consumption of methyl acrylate. Indeed, after 4 h using 0.5
equivalents of butylmethylimidazolium chloride,5 whilst ben-
zaldehyde had been mostly consumed, large amounts of the
methyl acrylate were unconverted (entry 3). We therefore
conducted control experiments in which various combinations
of different reactants were combined and discovered that
benzaldehyde actually reacted with the ionic liquid in the
presence of the mild base (3-HQD) to yield adduct 4 (Scheme
2).6 The reaction could be conducted using either DABCO or
3-hydroxyquinuclidine which were both equally effective. This
showed that imidazolium-based ionic liquids are not inert
solvents under mildly basic conditions.
In contrast to our studies, Afonso employed GC to integrate
the Baylis–Hillman adduct relative to benzaldehyde only to
obtain rate data and from this analysis concluded that the ionic
liquid led to faster reactions. They assumed that all of the
benzaldehyde was being converted to the Baylis–Hillman
product, but this is clearly incorrect as it ignores the very
significant side reaction involving the ionic liquid which
consumes benzaldehyde. In fact, the isolated yields of the
Baylis–Hillman adduct reported by Afonso were all rather low
which is in keeping with our own observations.
As further transformations of the modified ionic liquid 4 did
not occur, this substrate was tested as a potential promoter of the
Baylis–Hillman reaction. However, during the course of the
reaction, some of the original unsubstituted ionic liquid could be
detected. Of course this reverse process also provided additional
benzaldehyde which complicated our evaluation of the reaction
Initial test reactions between benzaldehyde and methyl
acrylate catalysed by 3-hydroxyquinuclidine (3-HQD) showed
that ionic liquids seemed to increase the reaction rate. An idea
of relative rate was determined by 1H NMR integration of the
Scheme 1 Baylis–Hillman Reaction.
† Electronic supplementary information (ESI) available: NMR data; details
of conditions employed by Afonso and integrations and calculations. See
http://www.rsc.org/suppdata/cc/b2/b203079a/
Scheme 2 Reaction of ionic liquid with benzaldehyde.
1612
CHEM. COMMUN., 2002, 1612–1613
This journal is © The Royal Society of Chemistry 2002

View Online
Finally, a Knoevenagel reaction has been conducted in
hexylmethylimidazolium hexafluorophosphate. In this case
malononitrile and benzaldehyde were employed together with
glycine as a very weak base. Only one example was provided
and it may be that other combinations of substrates lead to the
problem side reaction highlighted above.16
In conclusion, the use of butylmethylimidazolium ionic
liquids for the Baylis–Hillman reaction is not recommended due
to low yields resulting from direct addition of the deprotonated
butylmethylimidazolium to the aldehyde. This side reaction
dominates even if 50 mol% ionic liquids are employed. This
works highlights the need for caution when considering the use
of ionic liquids under basic conditions as they are very readily
deprotonated to give species capable of reacting with electro-
philes.
Scheme 3 Baylis–Hillman reaction in presence of modified ionic liquid
4.
rate. However, the reverse process has even more sinister
consequences to the recyclability of the ionic liquid. If the ionic
liquid from one run of a Baylis–Hillman reaction is then
employed again in a different Baylis–Hillman reaction a
mixture of products would be obtained. This was tested and was
indeed observed (Scheme 3). This clearly limits the reusability
of the ionic liquid.
Afonso reported that the ionic liquid could be reused 4 times
in the same Baylis–Hillman reaction and observed an increase
in yield with each recycle. This surprising observation was not
commented on but now is simple to understand. The ionic liquid
reacts with benzaldehyde but as more of it is converted with
each recycle, less benzaldehyde is consumed in this side
reaction and so more benzaldehyde is available for the normal
Baylis–Hillman reaction.
We thank the Deutscher Akademischer Austauschdienst
(DAAD), EPSRC, and ICI for support.
Notes and references
1 T. Welton, Chem. Rev., 1999, 99, 2071; P. Wasserscheid and W. Keim,
Angew. Chem., Int. Ed., 2000, 39, 3772; P. Wasserscheid and W. Keim,
Angew. Chem., 2000, 112, 3296; K. R. Seddon, Kinet. Catal., 1996, 37,
693; D. Zhao, M. Wu, Y. Kou and E. Min, Catal. Today, 2002, in
press.
Imidazolium salts have a pKa of around 24 in DMSO7 and so
it is perhaps surprising that they can react with aldehydes under
such mild basic conditions (the pKa of DABCO is 8.78). When
the conjugate base is used as a carbene ligand for ruthenium (the
Arduengo ligand) strong bases e.g. KOtBu are commonly
employed.9 Ionic liquids have been employed in palladium
catalysed Heck reactions and it has been suggested that using
imidazolium salts, the transformations could occur via Pd–
carbene complexes; the carbene being derived from deprotona-
tion of the imidazolium salt by a weak base.10 In such reactions,
only a very small amount of the ionic liquid is converted into the
carbene and so the transformation, if it occurs, usually goes
unnoticed. Imidazolium salts have been reacted with benzalde-
hyde in the presence of strong base (NaH typically) and then a
second electrophile.11
The initial addition of the carbene to the aldehyde is similar
to the initial addition of a thiazolium salt to benzaldehyde in the
benzoin reaction.12 However, evidently, imidazolium salts
cannot function in the same way as thiazolium salts as the
reaction stops at the adduct 4.13 Oxazolium salts have also been
studied as analogues of thiazolium salts and they behave in the
same way as the imidazolium salts towards base and benzal-
dehyde.14
The literature examples illustrate that in general, strong bases
have been employed to deprotonate imidazolium salts for
subsequent use in synthesis. We have found that mild bases with
pKa’s as low as 8–9 are strong enough to generate the
corresponding carbene.
Ionic liquids have been employed in the Horner–Wadsworth–
Emmons reactions (Scheme 4) and interestingly low yields were
obtained with ethylmethylimidazolium-based ionic liquids
(27–34%) but much higher yields were obtained with DBU-Et
salt.15 From our results it is highly likely that the low yields with
ethylmethylimidazolium-based ionic liquids were due to the
side reaction presented in Scheme 2.
2 V. K. Aggarwal, A. Mereu, G. J. Tarver and R. McCague, J. Org.
Chem., 1998, 63, 7183; V. K. Aggarwal and A. Mereu, Chem.
Commun., 1999, 2311; V. K. Aggarwal, A. Mereu, D. K. Dean and R.
Williams, J. Org. Chem., 2002, 67, 510.
3 A. Mereu, PhD Thesis, University of Sheffield, 2001, pp. 112–122.
4 J. N. Rosa, C. A. M. Afonso and A. G. Santos, Tetrahedron, 2001, 57,
4189.
5 Other imidazolium salts behaved similarly, such as BMIM PF6, EMIM
OTf, EMIM Cl, EMIM PF6 and EMIM BF4 (using 3-HQDor DABCO
as base). We have repeated Afonso’s work and his results are
reproducible: he obtained a 65% isolated yield and we obtained a 70%
NMR yield for the reaction between benzaldehyde, methyl acrylate,
DABCO and BMIM PF6 (see supplementary material†). The main
difference is that he did not observe the side reaction between the ionic
liquid and the aldehyde which in the above case consumed 22% of the
benzaldehyde.
6 The modified ionic liquid 4 has been characterised (see supplementary
material†) and peaks containing this species are observed in the crude
NMR of the Baylis–Hillman reaction (again see supplementary
material†). However, in the crude NMR we also observed an additional
set of minor signals, which may belong to the parent ionic liquid
indicating that further transformations are occurring.
7 R. W. Alder, P. R. Allen and S. J. Williams, Chem. Comm., 1995,
1267.
8 The pKa value DABCO was taken from J. Hine and Y.-J. Chen, J. Org.
Chem., 1987, 52, 2091.
9 W. A. Herrmann, C. Köcher, L. J. Gooßen and G. R. J. Artus, Chem.
Eur. J., 1996, 2, 1627; A. J. Arduengo, J. R. Goerlich and W. J.
Marshall, J. Am. Chem. Soc., 1995, 117, 11027; A. J. Arduengo, J. R.
Goerlich and W. J. Marshall, Liebigs Ann. Recueil, 1997, 365. For a
review on stable carbenes see: D. Bourissou, O. Guerret, F. P. Gabbaï
and G. Bertrand, Chem. Rev., 2000, 100, 39.
10 L. Xu, W. Chen and J. Xiao, Organomet., 2000, 19, 1123; A. J.
Carmichael, M. J. Earle, J. D. Holbrey, P. B. McCormac and K. R.
Seddon, Org. Lett., 1999, 1, 997; W. A. Herrmann and V. P. W. Böhm,
J. Organomet. Chem., 1999, 572, 141.
11 A. Miyashita, H. Matsuda and T. Higashino, Chem. Pharm. Bull., 1992,
40, 2627 and references cited therein.
12 R. Breslow, J. Am. Chem. Soc., 1958, 80, 3719.
13 N,N-diarylimidazolium salts can catalyse the benzoin-type reaction: J.
H. Teles, J.-P. Melder, K. Ebel, R. Schneider, E. Gehrer, W. Harder, S.
Brode, D. Enders, K. Breuer and G. Raabe, Helv. Chim. Acta., 1996, 79,
61.
14 H. Dugas, Biorganic Chemistry, 1989, 3rd edn., Springer, New York, p.
576H. W. Wanzlick, Angew. Chem., Int. Ed., 1962, 1, 75.
15 T. Kitazume and G. Tanaka, J. Fluorine Chem., 2000, 106, 211.
16 D. W. Morrison, D. C. Forbes and J. H. Davis Jr., Tetrahedron Lett.,
2001, 42, 6053. Professor Forbes has communicated with us that low
yields are observed with other combinations of reactants and it is likely
therefore that the side reaction presented in Scheme 2 is consuming
some of the aldehyde. His group will present their results independ-
ently.
Scheme 4 Horner–Wadsworth–Emmons reaction in ionic liquids published
by Kitazume.
CHEM. COMMUN., 2002, 1612–1613
1613