Development of catalysts for the Baylis-Hillman reaction: The application of tetramethylguanidine and attempts to use a supported analogue

Article abstract of DOI:10.1039/b106267k

We show here that tetramethylguanidine (TMG) is a useful catalyst for the Baylis-Hillman reaction of aldehydes with methyl acrylate, showing good activity with a range of aldehyde substrates and, unlike many other catalysts or catalyst mixtures, it can be

Full text of DOI:10.1039/b106267k

View Article Online / Journal Homepage / Table of Contents for this issue
Development of catalysts for the Baylis–Hillman reaction:
the application of tetramethylguanidine and attempts
to use a supported analogue
1
Nicholas E. Leadbeater* and Cornelia van der Pol†
Department of Chemistry, King’s College London, Strand, London, UK WC2R 2LS.
E-mail: nicholas.leadbeater@kcl.ac.uk
Received (in Cambridge, UK) 13th July 2001, Accepted 14th September 2001
First published as an Advance Article on the web 11th October 2001
We show here that tetramethylguanidine (TMG) is a useful catalyst for the Baylis–Hillman reaction of aldehydes
with methyl acrylate, showing good activity with a range of aldehyde substrates and, unlike many other catalysts or
catalyst mixtures, it can be used to good effect with simple aliphatic aldehydes. We show that the activity of the
catalyst is decreased when the reaction is run using solvents rather than solvent-free but, in the case where a solvent
is necessary, dichloromethane offers the best results. Attempts to use supported or derivatised TMG complexes as
catalysts for the reaction have been unsuccessful suggesting that the presence of an amine hydrogen is key to the
activity of TMG. Attempts to use stabilised aliphatic phosphines as catalysts for the reaction have proven partly
successful, with only modest yields of product being obtained with 2-(dicyclohexylphosphino)biphenyl and
2
-(di-tert-butylphosphino)biphenyl. The Verkade superbase, P(MeNCH CH ) N, proved inactive as a
2 2 3
catalyst, an adduct with the acrylate being the only product formed.
Introduction
The Baylis–Hillman reaction offers a convenient synthetic
1
route to highly functionalised molecules. The reaction involves
the coupling of an activated alkene with electrophiles (usually
aldehydes) using a base (usually an amine) as a catalyst
Fig. 1 The Baylis–Hillman reaction.
(
Fig. 1). The proposed mechanism for the reaction is shown in
up to 28 days to reach completion. Recently there has been a
resurgence in research activity based around the reaction and
there have been a number of reports of catalysts and conditions
Scheme 1. The densely functionalised products can be trans-
formed selectively, for example, into epoxides, triols and anti-
aldol products. Whilst it has been used for the synthesis of
natural products and other targets, it is not without its
problems. The reaction is very enigmatic and catalysts used are
often very substrate specific. Another major drawback to date
has been the poor reaction rates; the process taking anything
2
5–11
that lead to rate acceleration.
Although these can prove use-
3,4
ful in some situations, it is often the case that the catalysts used
are very substrate specific and so there is no really universal
catalyst that can be used and relied on for a range of substrates.
In addition, a number of the catalyst systems reported to date
either require a mixture of a number of components acting
together as a catalyst mixture or else are air- and moisture-
sensitive meaning that special conditions need to be used.
In our laboratories we have been looking at the development
of catalysts that can be used in the Baylis–Hillman reaction and
are readily available, have high activity and need no special
conditions for use. We have looked at a range of catalyst candi-
dates and at the conditions for optimal results. We report our
findings here and show that in the case of many substrates
1
,1,3,3-tetramethylguanidine (TMG) could be a catalyst of
choice. We also report our attempts to prepare solid-supported
analogues for use in the reaction.
Results and discussion
The base most often used as a catalyst for Baylis–Hillman
chemistry is 1,4-diazabicyclo[2.2.2]octane, DABCO. This
nucleophilic non-hindered base has been found to be reason-
ably versatile, working for a range of substrates. In addition, it
is easily removed from the product mixture at the end of the
reaction. More recently, however, Aggarwal and Mereu have
reported that 1,8-diazabicyclo[5.4.0]undec-7-ene, DBU, which
is normally considered as a hindered non-nucleophilic base, is
an excellent catalyst for Baylis–Hillman chemistry when used in
Scheme 1 Proposed mechanism for the amine catalysed Baylis–
Hillman reaction.
†
Study placement student from Hogeschool van Utrecht, The
Netherlands.
DOI: 10.1039/b106267k
J. Chem. Soc., Perkin Trans. 1, 2001, 2831–2835
This journal is © The Royal Society of Chemistry 2001
2831

View Article Online
Table 1 Screening of 1 and 2 for activity in the Baylis–Hillman
a
reaction
Catalyst
Reaction time/h
Product yield (%)
1
1
1
2
2
2
2
6
16
48
6
16
48
144
6
9
12
14
31
36
36
a
Using benzaldehyde and methyl acrylate as starting materials and 5
mol% phosphine catalyst. Reactions were run solvent-free and at room
temperature.
8
a 1 : 1 : 1 ratio with alkene and electrophile. The key to the
activity of DBU is thought to be the fact that the β-ammonium
enolate is stabilised through conjugation which increases its
equilibrium concentration and results in enhanced rates. In
addition to amines, recent reports suggest that some tertiary
phosphines are also good catalysts for the reaction. Aliphatic
phosphines have been shown to be the most active, in particular
tributylphosphine, reported by Rafel and Leahy to catalyse the
Fig. 2 Stabilisation of intermediates with substituted guanidines.
formed. Our attention focused on substituted guanidine
substrates as there is potential for resonance stabilisation of
the β-enolate formed with acrylates (Fig. 2). We are not the first
to use substituted guanidines as catalysts for the reaction.
Aggarwal and Mereu report the use of 1,3,4,6,7,8-hexahydro-1-
methyl-2H-pyrimido[1,2,a]pyrimidine, 4, but find it produces8
only modest yields of product (30%) over a period of two days.
10
reaction rapidly at room temperature and even faster at 0 ЊC.
The starting point for our work was to look at a number of
amine and phosphine complexes that could potentially act as
catalysts for the reaction. The problems with using aliphatic
phosphines are that they are very susceptible to aerial oxidation
and are also hard to extract from the product mixture. Recent
reports in the literature have shown that sterically crowded
dialkylbiphenylphosphines have the steric bulk and electronic
properties similar to trialkylphosphines but are significantly
12
more stable in the presence of air and moisture. Consequently,
we screened two such complexes for catalytic activity in the
Baylis–Hillman reaction. Both 2-(di-tert-butylphosphino)-
biphenyl, 1, and 2-(dicyclohexylphosphino)biphenyl, 2, showed
catalytic activity in the reaction of benzaldehyde with methyl
acrylate, although only producing product in significantly lower
yield than that reported using tributylphosphine (Table 1). This
being said, in our laboratory we were unable to reproduce the
product yields using tributylphosphine as a catalyst, finding
that product yields were not in excess of 45 to 50%. In an
attempt to increase yields of product using 1 and 2 as catalysts
we tried varying the catalyst concentration and reaction tem-
perature but neither had a significant effect on the yield of
product.
We have focused our attention on tetramethylguanidine,
1–4
TMG, (R = Me in Fig. 2) as a potential catalyst candidate.
We find that it shows good activity and low loadings of catalyst
can be used. Optimisation studies were performed using benz-
aldehyde and methyl acrylate as substrates (Tables 2–4). At
room temperature we find that the optimum yields of product
are obtained when the reaction is run for 16 h using a ratio of
benzaldehyde : methyl acrylate : TMG of 1 : 1 : 0.5 (Table 2).
We find that decreasing the catalyst loading from 25 mol% to
5
mol% results in only a slight reduction of product yield.
However, increasing the catalyst loading significantly results in
a decrease in yield rather than, as may be expected, an increase.
The exact reason for this is not fully understood but may
explain why in the reports using 4, where a benzaldehyde :
8
methyl acrylate : 4 ratio of 1 : 1 : 1 is used, yields are so low.
We were keen to shorten the reaction time to less than 16 h
and find that after 6 h the product yield is only 13% less
indicating that the majority of the product is formed in the first
few hours of reaction and as the reaction reaches an equi-
librium product formation slows. In order to try to speed up
the reaction we studied the effects of varying the temperature
on the product yield (Table 3). Rafel and Leahy report that,
when using DABCO as a catalyst, the yields of product in the
reaction are increased when the temperature is reduced from
room temperature to 0 ЊC. They suggest that this increase in
yield is as a result of the preferential formation and subsequent
rapid reaction of an ionically stabilised enolate. Using TMG
we find that the opposite is true; the optimal yield of product
is achieved at room temperature (25 ЊC) or slightly warmer
(30 ЊC). Further cooling or warming of the reaction mixture
results in a dramatic lowering in yield. The fact that the results
are so different to those reported for DABCO suggests that
the formation of an ionically stabilised enolate is not key to
the activity of TMG as a catalyst and certainly supports
the idea that the equilibrium concentration of the β-enolate
To continue the studies, the Verkade superbase, P(MeN-
13
CH CH ) N 3, was assessed for activity. Proazaphosphatranes
2
2 3
such as 3 have been used with much success as catalysts,
promoters and nonionic bases in a range of organic trans-
formations, these properties stemming from the partial
transannulation between the phosphorus and bridgehead nitro-
10
14–16
gen atoms on formation of a cation.
However, we find that
in the Baylis–Hillman reaction between benzaldehyde and
methyl acrylate, 3 forms an adduct with the acrylate and no
product is formed. This could well be attributed to the high
nucleophilicity and basicity of proazaphosphatranes, 3 having a
17
pK of 41.2 in acetonitrile and 26.8 in DMSO.
a
We decided to move away from using tertiary phosphines
as catalysts for the reaction and to look instead at substituted
amine substrates capable of stabilising the intermediates
2
832
J. Chem. Soc., Perkin Trans. 1, 2001, 2831–2835

View Article Online
a
Table 2 Optimisation of TMG catalyst loading and reaction time
Ratio of aldehyde : acrylate : TMG
Mol% TMG
Reaction time/h
Product yield (%)
1
1
1
1
1
1
1
1
: 1 : 0.1
: 1 : 0.1
: 1 : 0.25
: 1 : 0.25
: 1 : 0.5
: 1 : 0.5
: 1 : 1
5
5
12.5
12.5
25
25
50
50
6
16
6
16
6
16
6
16
58
64
66
67
68
73
31
50
: 1 : 1
a
Using benzaldehyde and methyl acrylate as starting materials and TMG as catalyst. Reactions were run solvent-free and at room temperature.
a
Table 3 Optimisation of reaction temperature
Table 5 aTMG catalysed Baylis–Hillman reaction with various
aldehydes
Mol% TMG
Temperature/ЊC
Product yield (%)
Aldehyde
Mol% TMG
Product yield (%)
5
0
0
8
12
58
66
68
73
62
67
8
5
5
1
5
2
5
1
5
0
Benzaldehyde
4-Chlorobenzaldehyde
Acetaldehyde
Propionaldehyde
Propionaldehyde
3-Phenylpropionaldehyde
(E)-Cinnamaldehyde
(E,E)-Hexa-2,4-dienal
12.5
12.5
12.5
5
12.5
12.5
25
67
65
61
63
69
55
50
25
25
25
25
30
30
40
2.5
0
5
2.5
b
5
0
a
a
Using benzaldehyde and methyl acrylate as starting materials and
Using aldehyde and methyl acrylate as starting materials and TMG
as catalyst. Reactions were run in dichloromethane and at room
TMG as catalyst. Reactions were run solvent-free and for 6 h.
b
temperature. Polymerisation occurs
a
Table 4 Assessment of the effect of solvent on the reaction
aldehydes were screened for activity in the reaction using methyl
acrylate as the activated alkene and TMG as catalyst. Reactions
were run in dichloromethane for 6 h using an aldehyde : acrylate
Solvent
Product yield (%)
None
Dichloromethane
Toluene
Hexane
THF
Dioxane
Acetonitrile
58
53
38
32
29
29
27
:
TMG ratio of 1 : 1 : 0.1. The results are shown in Table 5. As
can be seen from the data, good yields of product are obtained
with a range of aldehydes, the exception being hexa-2,4-dienal
where polymerisation occurs. This is not totally unexpected as
this substrate is known to readily polymerise in the presence of
base. The fact that the reaction works well with acetaldehyde
and propionaldehyde is particularly noteworthy. To date,
Baylis–Hillman reactions involving simple aliphatic aldehydes
are reported to give notoriously low yields of product or else
the aldehyde decomposes in the presence of the catalyst mixture
during the course of the reaction. Using TMG our results
suggest that it is possible to circumvent these problems and
obtain reasonable yields of the desired product, the remainder
of the reaction mixture being unreacted starting materials.
When using an amine catalyst, one of the drawbacks of the
Baylis–Hillman reaction is the work-up required at the end
of the reaction. It is necessary to remove the amine from the
reaction mixture by an aqueous work-up. Although this is not
difficult, it creates an extra step in the procedure and produces
quantities of aqueous waste. We wanted to try to avoid the
work-up step by immobilising the amine catalyst onto a
polymer support. The key advantage of attaching a catalyst
to a polymer support is ease of separation from the product
mixture at the end of a reaction; a simple filtration being
all that is required. In addition, as the catalyst is easily removed
from the reaction mixture, it can be re-used in subsequent
reactions.
a
Using benzaldehyde and methyl acrylate as starting materials, 5 mol%
TMG as catalyst and 5 ml solvent. Reactions were run at room
temperature for 6 h.
intermediate is increased by using a base that has the potential
for stabilisation through conjugation. We are not certain as
to why the reaction yield decreases so dramatically when the
temperature is increased from 30 to 40 ЊC, but analysis of the
product mixture shows some evidence of the TMG being
rapidly decomposed or forming a polymeric product with the
starting materials.
As both benzaldehyde and methyl acrylate are liquids, the
reaction can be run without the need for solvent. However, this
is not always the case as many potential substrates are solids.
In addition, if the reaction is to be modified for using either
polymer-supported reagents or catalysts a solvent would be
necessary to ensure appropriate swelling of the resin. We there-
fore decided to investigate the effect of solvent on the reaction
(
Table 4). We find that addition of solvent (5 ml) to the benz-
aldehyde and methyl acrylate substrate mixture results in a
decrease in overall yield. This decrease in yield is least in
dichloromethane and greatest in acetonitrile, with THF,
toluene, hexane and dioxane being in between these limits. To
our knowledge, the use of dichloromethane as a solvent
in Baylis–Hillman reactions has not been reported previously
and may offer a route to increased yields using catalyst and
procedures already reported in the literature as well as here
using TMG.
We prepared polymer-supported TMG, 5, by simple reaction
of Merrifield’s resin with TMG (Scheme 2). The resin is sus-
pended in dioxane and, after allowing time for swelling, an
excess of TMG is added. The mixture is then refluxed for 16 h
before filtering and washing away excess TMG leaving the
product in a pure form. Elemental analysis shows that a loading
of approximately 1 mmol TMG per g resin can be obtained
from Merrifield’s resin with a loading of approximately 4 mmol
Cl per g. We find that a similar loading of TMG can be
achieved using a lower loading of Merrifield’s resin indicating
To broaden the scope of the reaction, we studied the effects
of changing the aldehyde on the yields of reaction. A range of
J. Chem. Soc., Perkin Trans. 1, 2001, 2831–2835
2833

View Article Online
solvents. Reaction substrates and tetramethylguanidine were
purchased from Lancaster and Aldrich. Verkade superbase, 2-
dicyclohexylphosphino)biphenyl, 2-(di-tert-butylphosphino)-
(
biphenyl and tributylphosphine were acquired from Strem
Chemicals. Merrifield resin (200–400 mesh, crosslinked with
Ϫ1
2
% divinylbenzene, 4.3 mmol Cl g resin) was obtained
from Fluka. All commercially available chemicals were used
1
13
in reactions without further purification. H- and C-NMR
spectra were recorded using a Bruker 360 MHz NMR
spectrometer. Elemental analyses were run by Medac Ltd,
Brunel Science Park, UK. All spectra were run in CDCl using
3
TMS as a standard. The Baylis–Hillman products are all
known compounds. Physical and spectral data were compared
with those reported in the literature.
Typical procedure for catalyst screening for activity in the
Baylis–Hillman reaction
Scheme 2 Preparation of a polymer-supported analogue of TMG, 5,
and a homogeneous model, 6.
To a mixture of benzaldehyde (0.51 ml, 0.53 g, 5.0 mmol)
and methyl acrylate (0.45 ml, 0.43 g, 5.0 mmol) was added
the desired catalyst candidate. The reaction was stirred at the
required temperature for the allotted time. The reaction was
ended by addition of diethyl ether (20 ml). This solution was
washed with 2 M HCl (20 ml) and then twice with water (2 ×
that the limiting factor for replacement of Cl by TMG may well
be the steric crowding of the support.
Having prepared and characterised the supported TMG we
attempted to use it as a catalyst in the Baylis–Hillman reaction.
Our observations were disappointing, finding that 5 was com-
pletely inactive as a catalyst. The reaction was run for 48 h using
both dichloromethane and THF solutions of the reagents
and also using solvent-free conditions. In none of these cases
was any trace of product formed. We thought that one reason
for the loss of activity of TMG on immobilisation might be the
steric crowding in the environment of the polymer support. To
investigate this we prepared benzyl-TMG, 6, this acting as a
homogeneous comparison to 5 (Scheme 2). This was prepared
in a modification of the literature method in modest yield by
thermolysis of a dioxane solution of benzyl chloride and
1
5 ml). The organic extract was dried over magnesium sulfate,
the solvent and any unreacted methyl acrylate were removed
under vacuum and the product mixture was analysed.
Typical procedure for aldehyde screening for activity in the
Baylis–Hillman reaction
To a mixture of aldehyde (5.0 mmol) and methyl acrylate
(0.45 ml, 0.43 g, 5.0 mmol) in dichloromethane (1.5 ml) was
added TMG (0.06 ml, 57.5 mg, 0.5 mmol). The reaction was
stirred at room temperature for 6 h. The reaction was ended by
addition of diethyl ether (20 ml) and the work-up was as in the
case of benzaldehyde.
18
TMG. Screening the activity of 6 in the Baylis–Hillman
reaction showed that, like 5, it was inactive as a catalyst. This
therefore suggests that the amine hydrogen on free TMG is key
to the activity of this complex as a catalyst for the reaction.
Substitution of this renders the complex inactive under the
conditions used in our experiments.
Preparation of TMG-methylpolystyrene, 5
To Merrifield resin (1 g, 4.3 mmol Cl) was added dioxane
(
50 ml) and the mixture allowed to stir for 30 min to ensure
swelling of the resin. After this time TMG (2.72 ml, 2.47 g,
1.5 mmol) was added and the mixture heated at 70 ЊC for 16 h.
2
Conclusions
At the end of the reaction the beads were washed twice with
MeOH, DCM and hexane in this order. The white beads were
dried overnight in a vacuum desiccator. The loading of amine
on the support was determined by microanalysis and estimated
at approximately 1 mmol TMG g resin (analysis data: 72.69%
C, 7.79 N, 8.70 H, 10.82 Cl).
We have shown that tetramethylguanidine, TMG, is a useful
catalyst for the Baylis–Hillman reaction. It shows good activity
with a range of aldehyde substrates and, unlike other catalysts
or catalyst mixtures, it can be used with simple aliphatic
aldehydes. We have shown that the activity of the catalyst
is decreased when the reaction is run using solvents rather
than solvent-free but, in the case where a solvent is necessary,
dichloromethane offers the best results. To our knowledge this
is the first time dichloromethane has been used as a solvent for
the reaction and could prove useful for performing the reaction
using other catalysts. Attempts to use supported or derivatised
TMG complexes as catalysts for the reaction have been un-
successful suggesting that the presence of an amine hydrogen is
key to the activity of TMG. Attempts to use stabilised aliphatic
phosphines as catalysts for the reaction have proven partly
successful, with only modest yields of product being obtained
with 2-(dicyclohexylphosphino)biphenyl, 2-(di-tert-butylphos-
phino)biphenyl. The Verkade superbase proved inactive as a
catalyst, an adduct with the acrylate being the only product
formed.
Ϫ1
Preparation of 2-benzyl-1,1,3,3-tetramethylguanidine, 6
To a dioxane solution (5 ml) of benzyl chloride (0.92 ml, 1.00 g,
8.0 mmol) was added TMG (1 ml, 0.92 g, 8.0 mmol) and the
resultant mixture heated at 70 ЊC for 2 h. Diethyl ether (25 ml)
was added and the mixture washed with water (25 ml) and
dried over magnesium sulfate. The solvent was removed under
vacuum and the resultant oily product purified by flash chroma-
tography giving 3.6 mmol (44% yield) of 6 as a yellow oil.
1
H-NMR: δ 7.56–7.15 (m, 5H), 4.39 (s, 2H), 2.78 (s, 6H), 2.73
13
(s, 6H). C-NMR: δ 134.5, 128.0, 127.2, 125.8, 53.1, 39.7 and
38.9.
Procedure for using 5 and 6 in the Baylis–Hillman reaction
To a dichloromethane solution (5 ml) of benzaldehyde (0.51 ml,
0
.53 g, 5.0 mmol) and methyl acrylate (0.45 ml, 0.43 g,
Experimental
5.0 mmol) was added TMG-methylpolystyrene (0.5 g, 0.5 mmol
TMG). The reaction was stirred at room temperature and the
reaction monitored using TLC. The reaction was ended after
48 hours after no evidence was found for product formation.
General
Reactions were run in dried glassware and using distilled
2
834
J. Chem. Soc., Perkin Trans. 1, 2001, 2831–2835

View Article Online
The reaction was repeated in THF and solvent-free but again
with no success. Repeating the reactions using 6 in the place of
4 W. R. Rouch and B. B. Brown, J. Org. Chem., 1993, 58, 2151.
5
M. Shi, J.-K. Jiang, S.-C. Cui and Y.-S. Feng, J. Chem. Soc., Perkin
Trans. 1, 2001, 390.
5
again did not lead to product formation.
6
7
M. Shi, J.-K. Jiang, and Y.-S. Feng, Org. Lett., 2000, 2, 2397.
M. Kawamura and S. Kobayashi, Tetrahedron Lett., 1999, 40,
1
539.
Acknowledgements
8
9
V. K. Aggarwal and A. Mereu, Chem. Commun., 1999, 2311.
V. K. Aggarwal, A. Mereu, G. J. Tarver and R. McCague, J. Org.
Chem., 1998, 63, 7183.
The Royal Society is thanked for a University Research Fellow-
ship (NEL). Hoffman La Roche and King’s College London
are thanked for funding. Dr R. S. Grainger and Dr M. S. Salter
are thanked for useful discussions.
1
1
0 S. Rafel and J. W. Leahy, J. Org. Chem., 1997, 62, 1521.
1 V. K. Aggarwal, G. J. Tarver and R. McCague, Chem. Commun.,
1996, 2713.
1
2 J. P. Wolfe, H. Tomori, J. P. Sadighi, J. Yin and S. L. Buchwald,
J. Org. Chem., 2000, 65, 1158.
1
3 J.-S. Tang and J. G. Verkade, Angew. Chem., Int. Ed. Engl., 1993, 32,
References
8
96.
1
For recent reviews on the reaction, see: (a) E. Ciganek, Org. React.
N. Y.), 1997, 51, 201; (b) D. Basavaiah, P. D. Rao and R. S. Hyma,
14 X. Liu and J. G. Verkade, J. Org. Chem., 2000, 65, 4560.
15 Z. Wang, A. E. Wroblewski and J. G. Verkade, J. Org. Chem., 2000,
65, 2065.
16 Z. Yu and J. G. Verkade, J. Org. Chem., 1999, 64, 8021.
17 R. Schwesinger and J. G. Verkade, J. Org. Chem., 2000, 65,
5431.
(
Tetrahedron, 1996, 52, 8001; (c) S. E. Drewes and G. H. P. Roos,
Tetrahedron, 1988, 44, 4653.
See P. Langer, Angew. Chem., Int. Ed., 2000, 39, 3049.
O. B. Familoni, P. T. Kaye and P. J. Klaas, Chem. Commun., 1998,
2
3
2
563.
18 H. Kessler and D. Leibfritz, Tetrahedron, 1970, 26, 1805.
J. Chem. Soc., Perkin Trans. 1, 2001, 2831–2835
2835