Metal- and ligand-accelerated catalysis of the Baylis-Hillman reaction

Article abstract of DOI:10.1021/jo980421n

The Baylis-Hillman reaction, the coupling of an unsaturated carbonyl compound/nitrile with aldehydes, is a valuable reaction but is limited in its practicality by poor reaction rates. We have endeavored to accelerate the reaction using Lewis acids and fou

Full text of DOI:10.1021/jo980421n

J . Org. Chem. 1998, 63, 7183-7189
7183
Meta l- a n d Liga n d -Acceler a ted Ca ta lysis of th e Ba ylis-Hillm a n
Rea ction
Varinder K. Aggarwal,*,† Andrea Mereu,† Gary J . Tarver,† and Ray McCague‡
Department of Chemistry, University of Sheffield, Sheffield S3 7HF, U.K., and
Chiroscience Limited, Cambridge Science Park, Milton Road, Cambridge CB4 4WE, U.K.
Received March 5, 1998
The Baylis-Hillman reaction, the coupling of an unsaturated carbonyl compound/nitrile with
aldehydes, is a valuable reaction but is limited in its practicality by poor reaction rates. We have
endeavored to accelerate the reaction using Lewis acids and found that while conventional Lewis
acids gave reduced rates group III, and lanthanide triflates (5 mol %) gave increased rates. The
optimum metal salts were La(OTf)3 and Sm(OTf)3, which gave rate accelerations (krel) of
approximately 4.7 and 4.9, respectively, in reactions between tert-butyl acrylate and benzaldehyde
when using stoichiometric amounts of DABCO. At low loadings of DABCO (up to 10 mol %), no
reaction occurred due to association of DABCO with the metal. Use of additional ligands to displace
the DABCO from the metal was studied, and the rate of reaction was found to increase further in
most cases. Of the ligands tested, at 5 mol %, (+)-binol gave one of the largest rate accelerations
(3.4-fold) and was studied in more detail. It was found that reactions occurred even at low DABCO
concentration so that here the Lewis base and Lewis acid were able to promote the reaction without
interference from each other. While the (+)-binol (and other chiral ligands) failed to provide any
significant asymmetric induction, a substantial nonlinear effect was observed with binol. Thus,
use of racemic binol gave no effect on the rate. In seeking to maximize the rate attainable, more
soluble (liquid) ligands were studied. Diethyl tartrate and triethanolamine gave rate enhancements
of 5.2× and 3.5× at 50 mol %, respectively, versus 1.5× and 2.3× at 5 mol %. The best protocol
was to use 100 mol % DABCO, 50 mol % triethanolamine, and 5 mol % La(OTf)3. This gave overall
rate accelerations of between 23-fold and 40-fold depending on the acrylate and approximately
5-fold for acrylonitrile. A simple acid wash removed the reagents, leaving the product in the organic
phase. While triethanolamine accelerated the reaction without the lanthanum triflate (18-22-
fold at 80 mol %), the reaction in the presence of the metal salt was faster. The system was tested
synthetically on various substrates and found to give good rate accelerations with both activated
(benzaldehyde and p-nitrobenzaldehyde) and less activated aldehydes (anisaldehyde and cyclohex-
anecarboxaldehyde) with acrylates. The limited amount of dimerized acrylate in the latter reactions
is noteworthy and should extend the range of substrates that can be made by the Baylis-Hillman
reaction using our optimum conditions.
Sch em e 1. Rea ction of ter t-Bu tyl Acr yla te w ith
In tr od u ction
Ben zya ld eh yd e
The Baylis-Hillman reaction is an exquisite reaction:
cheap and readily available starting materials are con-
verted, using a suitable catalyst, into densely function-
alized products.1,2 However, the reaction suffers from
poor reaction rates, e.g., the reaction of tert-butyl acrylate
with benzaldehyde using 10 mol % DABCO catalyst takes
28 days to complete (Scheme 1).3
The mechanism of the Baylis-Hillman reaction is
shown in Scheme 2.4 Attempts have been made to in-
crease the rate of the Baylis-Hillman reaction (the RDS
is highlighted) through either physical or chemical
means, but there are disadvantages associated with most
of the methods.
vation volume: -70 cm3 mol-1), but this method requires
sophisticated equipment.5,6 Ultrasound7 and microwave
irradiation8 have been reported to give increases in rate
but again require somewhat specialized equipment.
Conducting reactions at 40 °C provides an increase in
rate of up to 2-fold,7 but a recent report that reactions
carried out at 0 °C gave rate increases of up to 40-fold
seemed the simplest and most effective method for
The reaction can be accelerated by high pressure (the
Baylis-Hillman reaction has a very high negative acti-
(1) Drewes, S. E.; Ross, G. H. P. Tetrahedron 1988, 44, 4653-4670.
(2) Basavaiah, D.; Rao, P. D.; Hyma, R. S. Tetrahedron 1996, 52,
8001-8062.
(5) Hill, J . S.; Isaacs, N. S. Tetrahedron. Lett. 1986, 27, 5007-5010.
(6) Hill, J . S.; Isaacs, N. S. J . Chem. Res. 1988, 330.
(7) Roos, G. H. P.; Rampersadh, P. Synth. Commun. 1993, 23, 1261-
1266.
(8) Kundu, M. K.; Mukherjee, S. B.; Balu, N.; Padmakumar, R.;
Bhat, S. V. Synlett 1994, 444.
(3) Fort, Y.; Berthe, M. C.; Caubere, P. Tetrahedron 1992, 48, 6371-
6384.
(4) Hill, J . S.; Isaacs, N. S. J . Phys. Org. Chem. 1990, 285-288.
S0022-3263(98)00421-6 CCC: $15.00 © 1998 American Chemical Society
Published on Web 09/29/1998

7184 J . Org. Chem., Vol. 63, No. 21, 1998
Aggarwal et al.
Ta ble 1. Ba ylis-Hillm a n Rea ction s Usin g 100 Mol %
Sch em e 2. Mech a n ism of th e Ba ylis-Hillm a n
Rea ction
DABCO a n d 5 Mol % Lew is Acid Ca ta lysta
b
b
Lewis acid
krel
Lewis acid
krel
1
Ti(OEt)4
Sc(OTf)3
<0.1
3.3
BF3‚OEt2
TiCl4
<0.1
<0.1
a
Reactions were conducted on 1 mmol scale using a 1:1:1 ratio
of tert-butyl acrylate/benzaldehyde/DABCO. Reactions stopped
b
after 24 h. Relative to reactions using 100 mol % DABCO as
catalyst and no Lewis acid.
Ta ble 2. Effect of La n th a n id e Ca ta lysts on th e Ra te of
th e Ba ylis-Hillm a n Rea ction a
b
metal catalyst
yield (%)
krel
none
3.8
11.6
12.3
13.2
12.1
16.3
15.9
1
Sc(OTf)3
Yb(OTf)3
Gd(OTf)3
Eu(OTf)3
Sm(OTf)3
La(OTf)3
3.3
3.6
3.9
3.5
4.9
4.7
increasing reaction rate.9 However, we have been unable
to reproduce this result.
Modest increases in rates have been observed in the
presence of hydrogen bond donors by using methanol as
solvent10 or using 3-hydroxyquinuclidine10,11 or ω-hydroxy
esters.12 The increase in rate by hydrogen bonding can
be attributed to either stabilization of the intermediate
aza enolate 1 (thereby increasing its concentration) or
by activation of the aldehyde or indeed both. Conducting
reactions in water13 or fluorinated solvents14 also results
in increased rates as a result of hydrophobic and fluoro-
phobic effects, respectively. The use of phosphines, which
are more nucleophilic catalysts than tertiary amines,
result in increased rates.9,15 The most active phosphines
are aliphatic phosphines, but they are also very suscep-
tible to air oxidation. Increasing the activation of car-
bonyl compounds on either the donor or acceptor also
results in increased rates. For example, trifluoroethyl
acrylate reacts ∼2.5× faster than ethyl acrylate,3 and
aldehydes undergo the Baylis-Hillman reaction more
readily than keto esters or ketones.
We have considered the possibility of using Lewis acids
to promote the Baylis-Hillman reaction. In the same
way that hydrogen bonding promotes the Baylis-Hill-
man reaction, Lewis acids should also lead to increased
rates by either stabilization of the intermediate aza
enolate (thereby increasing its concentration) or activa-
tion of the aldehyde or both. In this paper, we provide a
full account of our studies in this area.16
a
Reactions conducted on 1 mmol scale using a 1:1:1 ratio of
tert-butyl acrylate/benzaldehyde/DABCO with the addition of 5 mol
% metal catalyst and 100 µL of MeCN. Reactions stopped after 24
h. Relative to reactions using 100 mol % DABCO as catalyst
b
and no Lewis acid.
It was found that conventional Lewis acids (BF3‚OEt2,
TiCl4) resulted in deceleration of the reaction, whereas
the use of Sc(OTf)3 resulted in an acceleration. We
presume that the standard Lewis acids formed strong
DABCO-Lewis acid complexes17 which neutralized the
Lewis acid and in addition removed some of the DABCO
from the system so that there was less catalyst available
for reaction, resulting in reduced rates. In contrast, Sc-
(OTf)3, being a much harder Lewis acid, is likely to form
a weaker and more labile DABCO-Lewis acid complex.
This complex may still function as a Lewis acid as there
should still be free sites on the metal to allow coordina-
tion of the aldehyde and therefore accelerate the reaction.
We therefore screened a range of hard Lewis acids, the
lanthanides (Table 2). As some catalysts were not
completely homogeneous under neat reaction conditions,
all reactions were conducted with 100 µL of MeCN. This
proved to be the optimum solvent.
The catalysts are listed in order of increasing atomic
radii of the metal, and there is a general trend of increase
in reaction rate with increasing size of the metal cation.
La(OTf)3 and Sm(OTf)3 provided the greatest accelera-
tion, and reaction with La(OTf)3 was studied in greater
detail as it is more economical. Reactions were carried
out with varying amounts of DABCO (nucleophilic cata-
lyst) in the presence of 5 mol % La(OTf)3 and the rates
compared to reactions conducted in the absence of the
Lewis acid. The results are depicted graphically in
Figure 1 (and in the Supporting Information, Table 1).
It was found that rates did not increase linearly with
increasing DABCO concentration as might have been
expected (the reaction is first order with respect to
DABCO3,4) presumably because increasing additions of
DABCO resulted in effective dilution of the other re-
agents. In the presence of Lewis acid, it was found that
no reaction occurred until 10 mol % DABCO had been
Resu lts
Reactions were conducted using tert-butyl acrylate and
benzaldehyde with 1 equiv of DABCO and 5 mol %
different Lewis acids. After 24 h, the reactions were
stopped and the yields measured by gas chromatography
(see the Experimental Section for full details). The
results are shown in Table 1.
(9) Rafel, S.; Leahy, J . W. J . Org. Chem. 1997, 62, 1521-1522.
(10) Ameer, F.; Drewes, S. E.; Freese, S.; Kaye, P. T. Synth.
Commun. 1988, 18, 495-500.
(11) Drewes, S. E.; Freese, S. D.; Emslie, N. D.; Roos, G. H. P. Synth.
Commun. 1988, 18, 1565-1572.
(12) Basavaiah, D.; Sarma, P. K. S. Synth. Commun. 1990, 20,
1611-1615.
(13) Auge, J .; Lubin, N.; Lubineau, A. Tetrahedron Lett. 1994, 35,
7947-7948.
(14) Vojkovsky, T. Abstr. Pap. Am. Chem. Soc. 1996, 211, 288.
(15) Roth, F.; Gygax, P.; Frater, G. Tetrahedron Lett. 1992, 33,
1045-1048.
(17) It has recently been found that sulfides can catalyze the Baylis-
Hillman reaction in the presence of TiCl4. Presumably, this combina-
tion of reagents does not lead to the formation of strong Lewis acid-
Lewis base complexes. Kataoka, T.; Iwama, T.; Tsujiyama, S. J . Chem.
Soc., Chem. Commun. 1998, 197-198.
(16) Aggarwal, V. K.; Tarver, G. J .; McCague, R. J . Chem. Soc.,
Chem. Commun. 1996, 2713-2714.

Catalysis of the Baylis-Hillman Reaction
J . Org. Chem., Vol. 63, No. 21, 1998 7185
example, some diol ligands gave good rate accelerations
(Table 3, entries 2-4) but some did not (Table 3, entries
5-7). Salen, bisoxazoline, and amino alcohol ligands also
gave mixed results. Triethanolamine gave good rate
accelerations with La(OTf)3 (Table 3, entry 10), but more
substituted analogues gave lower rates (Table 3, entries
11 and 12). Similarly, hindered salens gave considerably
lower rates than less substituted salens (Table 3, entries
13-15). The optimum ligand/metal combination in terms
of rate acceleration was (R,R)-(+)-hydrobenzoin with
Yb(OTf)3 (Table 3, entry 6).
F igu r e 1. Effect of amount of DABCO on reaction rate.
Reactions were conducted on a 1 mmol scale using a 1:1 ratio
of tert-butyl acrylate/benzaldehyde with 0.1 mL of CH3CN; krel
) 1, [La(OTf)3] ) 0, [(+)-binol] ) 0, [DABCO] ) 100 mol %.
added, possibly because all the DABCO was associated
with the Lewis acid to give 3b. This is consistent with
Unfortunately, the highest enantiomeric excess ob-
tained using any of the chiral ligands was only 5% [(+)-
diisopropyl tartrate with Yb(OTf)3].
Although (R,R)-(+)-hydrobenzoin with Yb(OTf)3 gave
the highest rate, (+)-binol gave the highest rates with a
range of metals (the optimum being La(OTf)3, Table 3,
entry 2), and so a series of reactions were now per-
formed with this metal-ligand combination in order to
optimize the amount of DABCO and ligand. The results
are depicted in Figure 1 (and Supporting Information,
Table 1).
It was discovered that, now, in the presence of binol,
reactions occurred even at low concentrations (<10 mol
%) of DABCO, indicating that the metal was no longer
sequestering the amine. Also, as the graphs had steeper
gradients compared to the absence of (+)-binol a stronger
Lewis acid (12) must have been generated. The depar-
the formation of La(OTf)n(DABCO)2 complex which per-
forms its role as a Lewis acid and at concentrations above
10 mol % the DABCO can perform its role as a nucleo-
philic catalyst. The reaction is believed to occur via the
metal complex 4. The nonlinear increase in rate with
increasing concentration of DABCO could again be ac-
counted for by the diluting effect of additional DABCO.
It was thought that the binding of DABCO to the Lewis
acid to give 3b not only resulted in sequestration of the
nucleophilic catalyst but may also have moderated the
Lewis acidity of the metal due to the powerful nitro-
gen donor ligands (amine ligands are better donors
than ethers). We therefore sought to displace the nitro-
gen donor ligands with alternative groups by taking
advantage of the oxophilic nature of the metal and so
tested a broad range of oxygen-rich ligands with sev-
eral metals and discovered further rate enhancements
(Table 3).
ture from linearity at high [DABCO] presumably results
from a change in concentration of the different catalytic
species present (e.g., shift of the equilibrium from 12
toward 3b) or from dilution of the reaction. Thus, the
lanthanum catalyst with oxygen donor groups 12 was
superior to the lanthanum catalyst with nitrogen donors
3b.
No clear correlation between the structure of the ligand
and rate acceleration emerged from this study. For
These results show that the best accelerations were
obtained using stoichiometric amounts of DABCO and

7186 J . Org. Chem., Vol. 63, No. 21, 1998
Aggarwal et al.
Ta ble 3. Effect of Meta l a n d Liga n d Ca ta lysis on
Rela tive Rea ction Ra tea ,b
entry
ligand
Sc(OTf)3 Yb(OTf)3 Eu(OTf)3 La(OTf)3
1
2
3
4
none
(+)-binol
3.3
9.4
5.2
3.5
3.6
14.4
9.7
3.5
12.8
5.5
4.7
14.6
7.3
(+)-diethyl tartrate
(+)--iisopropyl
tartrate
9.5
4.6
8.1
5
6
(+)-TMTDAc
(R,R)-(+)-
4.1
3.5
8.0
16.2
3.6
5.8
4.0
5.3
hydrobenzoin
7
(+)-1,1,2-triphenyl- 3.2
5.2
4.5
2.2
3.8
5.9
4.7
ethanediol
8
9
(+)-TADDOL 5
2.9
3.3
4.65
3.3
3.1
5.7
7.0
2.31
3.6
ethylene glycol
10 triethanolamine
11 amino diol 6
10.8
4.4
3.6
12 amino triol 7
13 Salen ligand 8
14 Salen ligand 9
15 Salen ligand 10
16 bisoxazoline 11
6.3
5.8
5.2
3.2
4.0
4.4
17 N-methylephedrine 2.87
a
Reactions conducted on 1 mmol scale using a 1:1:1 ratio of
F igu r e 3. Effect of triethanolamine on reaction rate. Reac-
tions were conducted with a 1:1:1 ratio of tert-butyl acrylate,
benzaldehyde, and DABCO on a 1 mmol scale with 5 mol %
La(OTf)3; krel ) 1, [N(CH2CH2OH)3] ) 0.
tert-butyl acrylate/benzaldehyde/DABCO; 5 mol % metal catalyst,
b
5 mol % ligand and 100 µL of MeCN. krel relative to 100 mol %
DABCO without Lewis acid. c (+)-N,N-Tetramethyltartaric acid
diamide.
greater) of metal and ligand are formed. The positive
nonlinear effect indicates that the meso complex is more
stable and less reactive than the chiral complex.
Nonlinear effects are normally associated with varia-
tions in enantiomeric excess of product with variation in
enantiomeric excess of ligand.18-20 In this case, all
products were racemic. Nevertheless, for reactions that
show ligand-accelerated catalysis21 it is interesting to
apply this concept to relative reaction rates.
Effect of Liga n d Con cen tr a tion . Enantiomerically
pure binol is expensive, and so cheaper ligands were
sought that would give comparable levels of acceleration.
Diethyl tartrate and triethanolamine had been found to
give moderate rate accelerations (1.8× and 2.2×, respec-
tively) but less than binol (3.4×) at 5% loading. As higher
ligand loading led to increased rates, we decided to
investigate the effect of increased concentrations of
diethyl tartrate and triethanolamine. As these ligands
were both liquids at room temperature, much higher
loadings were indeed possible.22 The results are sum-
marized in Figures 3 and 4 and in the Supporting
Information, Table 3.
Increased amounts of ligands (triethanolamine and
diethyl tartrate) led to increased rates up to approxi-
mately 50 mol % loading of ligand. Above this level, the
rate either plateaued (as in the case with triethanol-
amine) or began to reduce (as in the case with diethyl
tartrate). Presumably, at high ligand concentrations the
metal is completely saturated and it is difficult for the
aldehyde to coordinate and so the Lewis acidity is
lowered. High ligand concentrations also dilute and
F igu r e 2. Effect of enantiomeric excess of binol on reaction
rate. Reactions were conducted with a 1:1:1 ratio of tert-butyl
acrylate, benzaldehyde, and DABCO on a 1 mmol scale with
5 mol % La(OTf)3 and 5 mol % binol; krel ) 1 for reaction with
racemic binol.
10 mol % (+)-binol. Increasing the amount of binol
further (25 mol %) did not result in further rate increases
due to reaction viscosity.
Dep en d en ce of Rea ction Ra te on En a n tiom er ic
Excess of Bin ol. Racemic binol gave much poorer
acceleration of the Baylis-Hillman reaction than enan-
tiomerically pure binol. To understand this effect, a
series of reactions were conducted using 5 mol % binol
of variable enantiomeric excesses in combination with 5
mol % La(OTf)3 (Figure 2, Supporting Information, Table
2). The results clearly showed a nonlinear relationship
between the relative rate of reaction with enantiomeric
excess of the ligand. This indicates that more than one
ligand is bonded to the metal or that a 2:2 complex (or
(18) Puchot, C.; Samuel, O.; Dunach, E.; Zhao, S.; Agami, C.; Kagan,
H. B. J . Am. Chem. Soc. 1986, 108, 2353-2357.
(19) Guillaneux, D.; Zhao, S. H.; Samuel, O.; Rainford, D.; Kagan,
H. B. J . Am. Chem. Soc. 1994, 116, 9430-9439.
(20) Kagan, H. B.; Girard, C.; Guillaneux, D.; Rainford, D.; Samuel,
O.; Zhang, S. Y.; Zhao, S. H. Acta Chem. Scand. 1996, 50, 345-352.
(21) Berrisford, D. J .; Bolm, C.; Sharpless, K. B. Angew. Chem., Int.
Ed. Engl. 1995, 34, 1059-1070.
(22) (+)-Binol loading was limited to 10%, as above that amount
the reaction became extremely viscous and no further acceleration was
observed.

Catalysis of the Baylis-Hillman Reaction
J . Org. Chem., Vol. 63, No. 21, 1998 7187
the azaenolate 1 or the aldehyde; either case would
increase the of rate the Baylis-Hillman reaction. In-
deed, the nucleophilic base DABCO may be too bulky to
occupy the ninth position, and so the Lewis acid and
Lewis base would be able to perform their respective roles
without interference from each other. However, in the
presence of an excess of triethanolamine, the ninth ligand
is more likely to be a hydroxyl group of the excess re-
agent. In this case, rate acceleration would result from
increased strengths of hydrogen bonds between the tri-
ethanolamine ligands bound to the metal and the aza-
enolate or between the triethanolamine ligands and the
aldehyde or indeed both (see below). If this is indeed the
origin of the rate acceleration, it is perhaps not surprising
that low enantioselectivity was observed with the chiral
ligands due to the lack of control in orientation of the
hydrogen bonds.
F igu r e 4. Effect of diethyl tartrate concentration on reaction
rate. Reactions were conducted on a 1 mmol scale using a 1:1:1
ratio of benzaldehyde/tert-butyl acrylate/DABCO with 5 mol
% La(OTf)3, krel ) 1, [(+)-diethyl tartrate] ) 0.
therefore slow the reaction. These results show that the
optimal loading of triethanolamine and diethyl tartrate
is ∼50 mol %. Diethyl tartrate gave the best overall
acceleration (5.2×); triethanolamine and binol gave
similar but slightly lower optimal rate increases (3.5×
and 3.9×, respectively). Due to the relative cost of
triethanolamine (£0.05/g) and its ease of removal, tri-
ethanolamine was selected as the ligand of choice in these
reactions.
A control experiment using 50 mol % triethanolamine
without La(OTf)3 was also conducted, and considerable
rate enhancements were observed. We therefore carried
out reactions with varying amounts of triethanolamine
to determine the optimum concentration for maximum
rate to determine if the metal catalyst was required
(Figure 5). It was found that 80 mol % triethanolamine
gave the highest rate accelerations (18-22-fold for dif-
ferent acrylates, Table 4) but that the rates were lower
than those using the combination of metal and ligand.
To determine whether the origin of the rate acceleration
was simply due to hydrogen bonding, a similar study was
conducted with methanol (Figure 6). In this case, it was
found that approximately 240 mol % gave the highest
rate accelerations25 (3-11-fold for different acrylates,
Table 4), but this time the rates were substantially lower
compared to triethanolamine itself. This indicated that
rate accelerations observed with triethanolamine were
not simply due to hydrogen bonding alone but must
involve a cooperative effect from multiple hydrogen
bonding sites and the amino group.
A series of reactions using different acrylates and
aldehydes were conducted on a synthetic scale to explore
the scope and limitations of this optimized process
(Tables 5 and 6).
High yields of products were obtained using a range
of acrylates over a short reaction time. We were even
able to extend the range of acceptors to include cyclohex-
anecarboxaldehyde26 and anisaldehyde.27,28 However, the
reaction is still limited to relatively reactive acceptors
A series of acrylates together with acrylonitrile were
tested using the new system (5 mol % La(OTf)3, 50 mol
% triethanolamine), and as this seemed to be the opti-
mum set of conditions, accurate rate data were obtained
by monitoring reactions at short time and determining
yields by GC (see the Experimental Section). Graphs
were plotted of yield against time (see the Supporting
Information), and the gradients of the straight line plots
gave the rate of reaction. The results are summarized
in Table 4. The expected increases in rate were observed
23
in the presence of La(OTf)3 with further rate increases
in the presence of triethanolamine for all substrates. The
optimized conditions provided rate increases of 23-40-
fold depending on the acrylate used but much smaller
rate increases for acrylonitrile.
X-ray crystal structures of lanthanide triflate-trietha-
nolamine complexes have been reported for presidium
and ytterbium24 and provide some insights into the ori-
gin of the rate acceleration observed with lanthanum.
The X-ray structures show a nine-coordinate metal center
bearing two tetradentate triethanolamine ligands and
one THF molecule. Additional THF molecules and tri-
flates hydrogen bond to the triethanolamine ligands
bound to the metal. Thus, it is possible that in the
absence of THF the ninth ligand on the metal center is
(25) It had previously been shown that 10 mol % MeOH results in
a 1.5-fold rate enhancement. No further tests were conducted. See ref
10.
(26) For reaction of cyclohexanecarboxaldehyde with methyl acrylate
see ref 7. No yield was given, but the half-life for the reaction using
20 mol % DABCO was 10 days.
(27) For reaction of anisaldehyde with methyl acrylate, see: Fou-
caud, A.; El Guemmout, F. Bull. Chim. Soc. Fr. 1989, 403-408. They
reported 90% yield after 20 days using 15 mol % DABCO.
(28) Methyl acrylate has a tendency to dimerize, and with less
reactive acceptors this is the major product (see refs 29 and 30). Under
our optimum conditions, the rate of reaction with the aldehyde is
enhanced relative to dimerization.
(23) The slightly higher relative rates observed with La(OTf)3
compared to earlier rate studies is because reactions are no longer
being diluted with 100 µL of MeCN. In addition, more accurate
measurements are now being made. We are currently obtaining five
different samples from a single reaction and analyzing them. This was
not practical in the initial stages when a very large number of different
reactions were conducted.
(24) Hahn, F. E.; Mohr, J . Chem. Ber. 1990, 123, 481-484. We thank
one of the reviewers for highlighting this paper to us.

7188 J . Org. Chem., Vol. 63, No. 21, 1998
Ta ble 4. Kin etic Stu d ies of Rea ction s of Acr yla tes w ith Ben za ld eh yd ea
Aggarwal et al.
methyl acrylate
ethyl acrylate
rate
tert-butyl acrylate
acrylonitrile
rate
rate
rate
catalytic system
(% per min)
krel
(% per min)
krel
(% per min)
krel
(% per min)
krel
DABCO
0.016
0.148
0.511
0.306
0.148
1
9.2
31.9
19.2
9.2
0.009
0.112
0.357
0.196
0.100
1
0.002
0.013
0.046
0.036
0.006
1
6.5
23
18
3
0.412
0.972
2.091
1.146
0.623
1
b
DABCO, La(OTf)3
12.4
39.7
22.0
11.1
2.36
5.07
2.78
1.51
DABCO, La(OTf)3,b N(CH2CH2OH)3
c
d
DABCO, N(CH2CH2OH)3
DABCO, CH3OHe
a
b
Reactions conducted on 1 mmol scale using 1:1:1 ratio of benzaldehyde/acrylate/DABCO. La(OTf)3 5 mol %. c Triethanolamine 50
mol %. Triethanolamine 80 mol %. e CH3OH 240 mol %.
d
acceleration (krel) of up to 4.9 when using 1 equiv of
DABCO. At low loadings of DABCO (up to 10 mol %),
no reaction occurred, and this is consistent with the
formation of
a metal-DABCO complex (La(OTf)n-
(DABCO)2) so that the DABCO was no longer available
as a nucleophilic catalyst. At the higher loadings of
DABCO, this complex, despite the amino ligands, per-
formed as a Lewis acid catalyst representing a rare
example of where a base and a Lewis acid can coexist
and cocatalyze a reaction.31
The second discovery was that a further increase in
rate was achieved when using catalytic amounts of
oxygen-rich donor ligands such as (+)-binol, L-diethyl
tartrate, and triethanolamine. Reactions now occurred
at low levels of DABCO (<10 mol %), so it is evident that
the lanthanide no longer associates with DABCO but
with the oxide ligand due to the metal’s oxophilicity. A
substantial nonlinear effect on rate with enantiomeric
composition of binol was observed. Thus, while (+)-binol
at 5 mol % gave a rate acceleration of 3.4-fold, racemic
binol gave no greater rate than in its absence. This is
consistent with the formation of a complex La(OTf)n-
(binol)2 in which the meso complex is more stable and
less reactive than the chiral complex. Given the presence
of two chiral ligands on the metal, it was disappointing
that no significant enantiomeric excesses were observed
with the chiral phenols/alcohols. We believe that the
intermediate azaenolate and aldehyde react on the
surface of the metal-ligand complex where acceleration
is provided by increased strengths of hydrogen bonds
between the hydroxyl groups of the ligands and the two
reacting species. The lack of control in orientation of the
hydrogen bonds is the probable explanation for the low
enantioselectivity.
F igu r e 5. Effect of triethanolamine concentration on reac-
tion rate. Reactions were conducted on a 1 mmol scale using
a 1:1:1 ratio of ethyl acrylate/benzaldehyde/DABCO; krel ) 1,
[N(CH2CH2OH)3] ) 0.
To maximize the reaction acceleration, ligands more
soluble than binol were needed. The protocol discovered
to give the best practicality was to use 100 mol %
DABCO, 50 mol % triethanolamine, and 5 mol % La-
(OTf)3. This gave an acceleration of up to 40-fold for an
acrylate and approximately 5-fold for acrylonitrile, and
a simple acid wash procedure allowed recovery of the
product from the reagents. While triethanolamine alone
(80 mol % was optimum) gave a rate acceleration of up
to 22-fold, the acceleration is clearly greater when the
lanthanum triflate is present. The preferred system was
demonstrated to be effective synthetically on a variety
of different acrylates and aldehydes both more and less
reactive than benzaldehyde, demonstrating the general-
F igu r e 6. Effect of methanol concentration on reaction rate.
Reactions conducted on a 1 mmol scale using a 1:1:1 ratio of
ethyl acrylate/benzaldehyde/DABCO; krel ) 1, [CH3OH] ) 0.
as ketones (cyclohexanone was attempted) gave mostly
dimerized acrylates.29,30
(29) Basavaiah, D.; Gowriswari, V. V. L.; Bharathi, T. K. Tetrahe-
dron Lett. 1987, 28, 4591-4592.
Con clu sion s
(30) Drewes, S. E.; Emslie, N. D.; Karodia, N. Synth. Commun. 1990,
20, 1915-1921.
In the above study, our principal discovery has been
that scandium and lanthanide triflate catalysts, at 5 mol
%, catalyzed the Baylis-Hillman reaction with a rate
(31) Shibasaki, M.; Sasai, H.; Arai, T. Angew. Chem., Int. Ed. Engl.
1997, 36, 1237-1256.

Catalysis of the Baylis-Hillman Reaction
Ta ble 5. Yield of Ba ylis-Hillm a n P r od u cts in Rea ction s of Acr yla tes w ith Ben za ld eh yd ea
methyl acrylate ethyl acrylate tert-butyl acrylate acrylonitrile
time (h) yield (%)
J . Org. Chem., Vol. 63, No. 21, 1998 7189
catalytic system
time (h)
yield (%)
time (h)
yield (%)
time (h)
yield (%)
c
DABCO, La(OTf)3,b N(CH2CH2OH)3
12
24
24
83
88
71
12
24
24
84
91
78
72
72
48
74
65
35
1.5
4
7
92
88
86
d
DABCO, N(CH2CH2OH)3
DABCO, CH3OHe
a
b
Reactions conducted on 1 mmol scale using 1:1:1 ratio of benzaldehyde: acrylate: DABCO. La(OTf)3 5 mol %. c Triethanolamine 50
mol %. Triethanolamine 80 mol %. e CH3OH 240 mol %.
d
Ta ble 6. Rea ction s of Ald eh yd es w ith Meth yl Acr yla te
in th e P r esen ce of 5 Mol % La (OTf)3 a n d 50 Mol %
Tr ieth a n ola m in ea
Rea ction Mon itor in g (Ta ble 4). To a stirred mixture of
acrylate (4.92 mmol, 1 equiv) and benzaldehyde (0.5 mL, 4.92
mmol, 1 equiv) at room temperature under nitrogen were
added DABCO (552 mg, 4.92 mmol, 1 equiv), lanthanum
triflate (144 mg, 0.25 mmol, 5 mol %), and triethanolamine
(0.33 mL, 2.46 mmol, 50 mol %). In kinetic studies, five
samples were taken (0.1 mL using a Pipetman Gilson syringe)
at fixed short periods of time and filtered through a short silica
gel column using a solution of the internal standard, ethyl
cinammate in CH2Cl2 (0.15 M, 0.5 mL, 0.075 mmol), and
flushed with further CH2Cl2 (∼1.5 mL), and the resulting
solutions were used for the GC analysis.
aldehyde
time
3 h
2 days
5 days
yieldb (%)
p-nitrobenzaldehyde
anisaldehyde
cyclohexanecarboxaldehyde
90
65
37c
a
Reactions conducted on 1 mmol scale using 1:1:1 ratio of
aldehyde/methyl acrylate/DABCO, 5 mol % La(OTf)3 and 50 mol
b
c
% triethanolamine. Isolated yield of pure product. ∼25% of
dimerized methyl acrylate detected by 1H NMR.
Ca lcu la tion :
ity of this method. Additionally, it is noteworthy that
reduced amounts of dimerized acrylates were obtained.
We anticipate that these new reaction conditions, by
giving enhanced reaction rates, could extend the range
of products that can practically be made by the Baylis-
Hillman reaction.
yield (% product) )
(area product)(int. stand. mol)(total weight)
100F
(area int. stand.)(acrylate mol)(sample weight)
where F ) GC correction factor (ratio between internal
standard and product peak area of a equimolar solution).
Typ ica l P r oced u r e of Op tim ized P r ocess (Ta bles 5 a n d
6). To a stirred mixture of acrylate (4.92 mmol, 1 equiv) and
aldehyde (4.92 mmol, 1 equiv) at room temperature under
nitrogen were added DABCO (552 mg, 4.92 mmol, 1 equiv),
lanthanum triflate (144 mg, 0.25 mmol, 5 mol %), and
triethanolamine (0.33 mL, 2.46 mmol, 50 mol %). After the
time indicated, the reaction was stopped by dilution with ether
(30 mL) and washed with HCl (2%, 20 mL) followed by water
(20 mL). After drying over MgSO4, filtration, and evaporation,
the crude mixture was purified by column chromatography,
eluting with petroleum ether/diethyl ether (2:1), to give the
adduct as an oil. The product had data identical to those in
the literature.
Exp er im en ta l Section
Amino diol 6,32 salen ligands 8-10,33 and bisoxazoline 1134
were prepared as described in the literature. Amino triol 735
was a gift from Dr. Nugent. All other reagents were com-
mercially available. The Baylis-Hillman products are all
known compounds.
Rea ction Mon itor in g (Ta bles 1-3, Su p p or tin g In for -
m a tion Ta bles 1-3). Reactions were conducted at 25 °C
under neat reaction conditions or diluted with 100 µL of MeCN.
Benzaldehyde (1 mmol, 102 µL), acrylate (1 mmol), and
DABCO (0.1 mmol, 11.2 mg) were mixed, and then the
required Lewis acid (0.05 equiv) was added. A control reaction
was also run without Lewis acid. After 24 h, ethyl cinnamate
(1 equiv, 168 µL) was added to the reaction mixture as internal
standard. The reaction was then diluted with ethyl acetate
and stopped by filtration through silica gel to remove both
DABCO and the Lewis acid catalyst, and the yield of the
product was determined by GC. From these data, the relative
rate of reaction was calculated using the formula krel ) [(1/(1
- Xcat.)) - 1/(1/(1 - Xuncat)) - 1], where Xcat. and Xuncat are the
concentrations of products in the catalyzed and uncatalyzed
reactions, respectively (see Supporting Information). This
equation is derived from the second-order rate equation and
gives a straight line when plotting 1/[acrylate] against time.
Ackn owledgm en t. We thank Chiroscience and Shef-
field University for a studentship (G.J .T.) and EPSRC
for studentship support (A.M.). V.K.A. thanks the
Nuffield Foundation for a Fellowship and Pfizer for
additional support.
Su p p or tin g In for m a tion Ava ila ble: Data for Figures
1-4 and graphs of experimental rate data for Table 4 (7 pages).
This material is contained in libraries on microfiche, im-
mediately follows this article in the microfilm version of the
journal, and can be ordered from the ACS; see any current
masthead page for ordering information.
J O980421N
(32) Trost, B. M.; Van Vranken, D. L.; Bingel, C. J . Am. Chem. Soc.
1992, 114, 9327-9343.
(33) Larrow, J . F.; J acobsen, E. N.; Gao, Y.; Hong, Y. P.; Nie, X. Y.;
Zepp, C. M. J . Org. Chem. 1994, 59, 1939-1942.
(34) Lowenthal, R. E.; Abiko, A.; Masamune, S. Tetrahedron Lett.
1990, 31, 6005-6008.
(35) Nugent, W. A.; Harlow, R. L. J . Am. Chem. Soc. 1994, 116,
6142-6148.