Not quite the last word on the Perkin reaction

Article abstract of DOI:10.1016/j.tet.2014.07.053

Microwave irradiation does not accelerate the rate of the Perkin reaction carried out under normal atmospheric pressure. Water is an essential yet catalytic reactant for the Perkin reaction to occur. Containment of the Perkin reaction in a sealed vessel improves the yield. Two pressure increases are observed during a 4 h reaction time. An induction period is seen in the Perkin reaction when sodium acetate is used as a base. A re-appraisal of the reaction mechanism is proposed on the basis of these observations. The use of PFA reaction vessels enables the Perkin reaction to occur under aqueous conditions for around 80 reactions/vessel.

Full text of DOI:10.1016/j.tet.2014.07.053

Tetrahedron 70 (2014) 7245e7252
Contents lists available at ScienceDirect
Tetrahedron
journal homepage: www.elsevier.com/locate/tet
Not quite the last word on the Perkin reaction
a
, ,
y
d
b
c
*
Mark Edwards
, Paul M. Rourk , Philip G. Riby , Andrew P. Mendham
a
School of Science, Grenville Building, University of Greenwich, Medway Campus, ME4 4TB, UK
James Parsons Building, Liverpool John Moores University, Byrom Street, Liverpool L3 3AF, UK
Faculty of Engineering and Science, Nelson Building, University of Greenwich, Medway Campus, ME4 4TB, UK
University of Greenwich, School of Chemical and Life Sciences, Wellington Street, Woolwich, London SE18 6PF, UK
b
c
d
a r t i c l e i n f o
a b s t r a c t
Article history:
Microwave irradiation does not accelerate the rate of the Perkin reaction carried out under normal at-
mospheric pressure. Water is an essential yet catalytic reactant for the Perkin reaction to occur. Con-
tainment of the Perkin reaction in a sealed vessel improves the yield. Two pressure increases are
observed during a 4 h reaction time. An induction period is seen in the Perkin reaction when sodium
acetate is used as a base. A re-appraisal of the reaction mechanism is proposed on the basis of these
Received 24 April 2014
Received in revised form 11 July 2014
Accepted 14 July 2014
Available online 22 July 2014
Ò
observations. The use of PFA reaction vessels enables the Perkin reaction to occur under aqueous
conditions for around 80 reactions/vessel.
Keywords:
Perkin reaction
Microwave
Ó 2014 Elsevier Ltd. All rights reserved.
Water
PFA vessels
Mechanism proposal
1. Introduction
heating method was being used. The status of experimental in-
formation on the Perkin reaction at the time could be summarised
6
Interest in applying microwave heating to organic reactions had
perfectly by Johnson’s words in his review published many years
earlier: ‘A number of studies have been made of factors influencing
the yields in the Perkin reaction, but it is difficult to draw any broad
generalizations’.
1,2
been awakened by Gedye and others in the early 1990s, and these
applications to synthesis followed indirectly from Kingston’s work
on using microwave heating for analytical sample preparation
alongside the simultaneous development of commercial ovens to
Clearly, this situation was unsatisfactory and it was important to
establish a stable and optimised baseline from which to measure
any possible improvements. Scheme 1 illustrates the particular
version of the Perkin reaction that we wished to study.
3
,4
produce the microwave heating effect by CEM amongst others.
Around 20 years ago, the Perkin reaction, which involves the use
of an inorganic acetate salt in acetic anhydride to convert a variety
of aromatic aldehydes into their corresponding 3-phenylprop-2-
enoic acids, seemed a suitable model reaction to study the heat-
5
ing effects of microwaves. The reaction is well-known, uses readily
available reactants and, since most synthetic methods describe up
to an 8 h synthesis,6 the reaction would benefit from the reaction
rate enhancement that is reported as a beneficial effect of micro-
,7
8
wave heating for organic synthesis.
Scheme 1. The particular version of the Perkin reaction under study in this paper.
2
. Results and discussion
Before embarking on a study of the microwave-assisted Perkin
reaction it was important to ensure that an optimised thermal
We used mass balance calculations, based upon benzaldehyde
input, in order to determine cinnamic acid product losses to the
waste-streams generated during work-up of the reactions. Modi-
fications to the reactions and their associated work-up procedures
were driven by the results obtained by these mass balance calcu-
lations and were important in the development of a consistently
repeatable reaction. TLC analyses against standards of known
*
Corresponding author. c/o Andy Mendham, Faculty of Engineering and Science,
University of Greenwich, Medway Campus, Kent ME4 4TB, UK; e-mail address: a.p.
mendham@gre.ac.uk (M. Edwards).
y
Retired.
http://dx.doi.org/10.1016/j.tet.2014.07.053
0
040-4020/Ó 2014 Elsevier Ltd. All rights reserved.

7246
M. Edwards et al. / Tetrahedron 70 (2014) 7245e7252
cinnamic acid concentration were used to estimate waste-stream
losses in the absence of a validated HPLC method. Table 1 shows
how these estimates compared with the actual quantities isolated
from the waste-streams examined.
Table 2
Yields obtained from a modified, 4 h Perkin reaction using sodium acetate as base
Run
number benzaldehyde:
sodium acetate: acid %
Molar ratios,
Isolated Total cinnamic Mean total
cinnamic acid % molar cinnamic
Standard
deviation,
s
nꢃ1
yield based on acid % molar
mass balance yield
calculations
acetic anhydride molar
yield
Table 1
Reliability check for TLC estimation of organic extract cinnamic acid content
1
1
1
35
36
37
38
1
3
4
1.0:1.0:2.3
13.7
13.6
13.6
17.8
17.6
17.6
18.0
17.7
17.9
24.5
24.4
25.0
23.6
23.8
23.5
23.9
23.6
23.5
24.0
0.53
Run number
TLC estimated cinnamic acid
content, % molar yield
Recovered cinnamic
acid, % molar yield
2
1
1
2
e6
7.32ꢂ0.53
9.8ꢂ0.83
7.14ꢂ0.56
9.62ꢂ1.04
3.87ꢂ2.13
0.59ꢂ0.14
0e12
3e15
2e25
4.32ꢂ2.55
0.75ꢂ0.17
3
4
9
0
The TLC estimates in Table 1 are consistent, and correlate with
the recovered yields of cinnamic acid. TLC, although a semi-
quantitative analytical method, has been reported to be useful in
the estimation of active ingredients in drug formulation studies.9
Kadin reported that in 15 TLC assays of Captopril in spiked pla-
cebo powders recovery was 100.6% with a standard deviation of
Vogel in the third edition of his world renowned book ’Practical
Organic Chemistry’.
At the time, the comparison of these early results clouded our
judgement as to the significance of the low yield obtained using
microwave irradiation with our method.
1.39% and a coefficient of variation of 1.38. Clearly TLC is useful, in
certain cases, as a quick, semi-quantitative analytical method pro-
vided that certain experimental conditions are adhered to during
its application. TLC plate loading is important and specific to the
TLC plate manufacture. Experience has shown that for the TLC
During the early stages of this exploratory work, our reaction
conditions needed to be defined more clearly than those originally
provided by Perkin himself and the many others who followed.
Using sodium acetate we found that the Perkin reaction mixture
ꢁ
plates used in this study, no more than 100
m
g loading of a waste-
reflux occurred around 147 C, whereas when potassium acetate
ꢁ
stream sample from a 5 L micropipette provides uniformly-sized
m
was used reflux occurred at 169 C. We found that excessive heating
TLC spots. The optical density of the developed waste-stream
spot, under UV 254 nm light, can be compared against a range of
different concentrations of reference standard TLC spots, which
have been carefully prepared from commercial cinnamic acid. A
process of trial and error quickly enables suitable concentrations of
TLC reference standard spots to be found and used for comparison
with each analytical sample in this semi-quantitative method. The
of reaction mixtures was deleterious to the quality of the product
obtained. Reaction mixtures heated in oil-baths at 180 C were
often dark-coloured and the crude products were often contami-
ꢁ
nated with tarry side-products, which were difficult to work-up
1
1
and eliminate. We consider that the conclusions of Bock et al.,
13
and Meyer and Beer, with respect to using a higher reaction
heating temperature, are neither conducive to ease of work-up nor
the production of good quality cinnamic acids.
reference standard TLC spots are applied using fresh, 5 mL micro-
pipettes, and are developed simultaneously alongside the waste-
stream sample, on the same TLC plate. TLC tank solvent/vapour
equilibration is important if consistent TLC spot R values and TLC
f
We had found (see Table 4) that reaction yields could be im-
proved if the stoichiometric ratios of reactants, used by Vogel, for
aromatic aldehyde, sodium acetate and acetic anhydride of
1.0:0.61:1.42 were amended in favour of 1.0:1.0:1.42. This change in
stoichiometry seemed reasonable in the light of the mechanistic
understanding for the Perkin Reaction, if sodium acetate was acting
as a base to generate the enolate anion of acetic anhydride. Al-
though the yield improvement was only about 15% relative, we
considered it helpful when running reactions for a limited time
period of 4 h that would be expected to reduce the overall yield
obtained from a reported 8 h reaction.
plate performance are required. Solvent vapour/liquid equilibration
ꢁ
times of around 20 min, for a 20ꢀ20 cm glass TLC tank at 24e25 C
were found satisfactory for TLC plate development, if chromatog-
raphy paper wicks are used inside the TLC tank in order to aid
solvent evaporation.
14
Once an estimate of the waste-stream TLC spot concentration
has been made, the result can be factored into the preparation of
the analytical sample and calculated back to the original volume of
the waste-stream under analysis.
Solubility of the metal salt would be expected to have an in-
fluence on the rate of reaction. Potassium acetate is much more
soluble in acetic anhydride than sodium acetate. However, with
sodium acetate the yield increase is not linearly proportional to the
amount of extra acetic anhydride added beyond a stoichiometric
ratio of 1.42 relative to the aromatic aldehyde. As can be seen
from Table 5 below, a 20% relative increase in yield was obtained
when the stoichiometric ratio of acetic anhydride was raised to 2.3
versus the molar quantity of benzaldehyde. This yield improvement
was achieved at a cost of increasing the yield standard deviation.
This variation in standard deviation might be expected from the
extra experimental variabilities arising from the need to decompose
excess acetic anhydride and then generate pH10 during an aqueous
work-up in the presence of the additional acetic acid formed.
The yields detailed in Table 5 were then compared with the use
of a stoichiometric amount of a phase transfer agent, tetrabuty-
lammonium hydrogen sulfate. It was hoped that metathetical ex-
change of the sodium ion by the tetrabutylammonium ion would
increase the solubility of the acetate counter-ion and thereby im-
prove the yield of the reaction. Table 6 below shows that this
We developed a reaction procedure that provided a consistent
and stable yield platform from which to measure any improve-
ments and from which we were able to critique the validity of some
of the points reported by Johnson in his review under the section
entitled ‘Selection of experimental conditions’.
The results obtained from our method shown in Table 2 are
10,11
consistent with those described by Kalnin and B o€ ck et al.
Disappointingly, Perkin reactions, using our updated 4 h version
of the method described by Vogel, when performed in standard
laboratory glassware in a CEM MARS300 modified oven did not
provide any evidence for rate acceleration by microwave heating
over that provided by standard thermal heating.
This observation with sodium acetate as base was later con-
firmed by Veverkova et al., in their microwave study of the Perkin
reaction.12
Interestingly, the average yield obtained from six microwave
reactions was almost identical with the yield obtained from runs
2
e6, in Table 4, based upon the original stoichiometry for the re-
action proposed by Perkin and published as a cognate method by

M. Edwards et al. / Tetrahedron 70 (2014) 7245e7252
7247
Table 3
Comparison of cinnamic acid yields obtained by atmospheric pressure microwave heating and thermal heating of a modified stoichiometry of the Perkin reaction described by
Vogel
Reaction number
Heating method
Microwave
Molar ratios, benzaldehyde:
sodium acetate:acetic
anhydride
Isolated cinnamic
acid % molar yield
Total cinnamic acid %
molar yield based on
mass balance calculations
Mean total cinnamic
acid % molar yield
Standard deviation,
s
nꢃ1
2
2
2
2
2
2
1
3
4
5
6
7
8
1.0:1.0:2.3
1.0:1.0:2.3
12.15
11.86
11.96
12.15
12.15
11.66
d
17.2
18.2
17.4
17.3
17.2
18.5
d
17.6
24.0
0.57
0.53
1, 13, 14, 35e40
Thermal
Table 4
Comparison of cinnamic acid yields using Vogel’s stoichiometry with that arising from a prior experimental revision
7,10
Reaction
number
Molar ratios, benzaldehyde:
sodium acetate:acetic
anhydride
Isolated cinnamic acid %
molar yield
Total cinnamic acid %
molar yield based on
mass balance calculations
Mean total cinnamic
acid % molar yield
Standard deviation,
s
nꢃ1
2
3
4
5
6
7
8
9
1.0:0.62:1.42
1.0:1.0:1.42
10.2
10.5
10.3
10.3
10.4
10.7
10.9
10.7
10.8
17.6
17.7
17.2
16.7
18.2
20.3
20.0
20.2
20.1
17.5
0.56
20.2
0.29
1
2
Table 5
Yield comparison with stoichiometric variation of acetic anhydride
Run number
Molar ratios, benzaldehyde:
sodium acetate:acetic
anhydride
Isolated cinnamic acid %
molar yield
Total cinnamic acid %
molar yield based on
mass balance
Mean total cinnamic
acid % molar yield
Standard deviation,
s
nꢃ1
calculations
7
1
1
1
3
3
3
3
3
4
e9, 12
1.0:1.0:1.42
1.0:1.0:2.3
d
d
20.2
24.0
0.29
0.53
1
3
4
5
6
7
8
9
0
13.7
13.6
13.6
17.8
17.6
17.6
18.0
17.7
17.9
24.5
24.4
25.0
23.6
23.8
23.5
23.9
23.6
23.5
hypothesis was confirmed by a 25% relative increase in overall
cinnamic acid % molar yield. Interestingly the yields obtained with
potassium acetate over the same time period of 4 h were signifi-
cantly better than those obtained with sodium acetate with or
without phase transfer agent. We suggest this is due to the greater
solubility of the potassium salt, when compared with the others.
The improved solubility results in a higher reaction mixture reflux
temperature, a faster reaction and a completely homogeneous re-
action mixture without any suspended salts such as sodium hy-
drogen sulfate.
However, because of the elevated reaction mixture reflux tem-
peratures achieved, the reaction products from potassium acetate
ꢁ
ꢁ
(169 C) and phase transfer agent modified (163 C) Perkin re-
actions contained significant amounts of a tarry-like material. The
Table 6
Comparison of cinnamic acid molar yields obtained using tetrabutylammonium hydrogen sulfate with sodium acetate, potassium acetate or sodium acetate alone
Run number
Molar ratios benzaldehyde:
metal salt:acetic anhydride
Metal salt
Isolated cinnamic
acid % molar yield
Total cinnamic acid,
% molar yield based
on mass balance
calculations
Average total cinnamic
acid % molar yield
Standard deviation,
s
nꢃ1
1
1
1
1
2
2
2
2
1, 13, 14, 35e40
1.0:1.0:2.3
1.0:1.0:1.0:2.3
NaOAc
NaOAcþBu
d
d
24.0
30.9
0.53
0.69
7
8
9
0
1
2
3
4
NHSO
4
25.8
25.0
30.2
68.3
68.7
68.9
68.0
30.5
30.5
31.7
68.9
69.2
69.7
68.5
1.0:1.0:2.3
KOAc
69.1
0.51

7248
M. Edwards et al. / Tetrahedron 70 (2014) 7245e7252
tarry crude products required prolonged heating with sodium
carbonate solutions in order to effect complete dissolution and
achievement of a stable pH value of 10 during work-up. On the
other hand, work-up of sodium acetate Perkin reactions were easier
and product quality obtained was initially better with sodium ac-
etate because negligible tarry-product was produced, as a result of
the controlled, lower reaction mixture reflux temperature.
become fused together. The reason for this fusion was not clear as
temperatures measured at the top of the reactor assembly had not
increased above room temperature at any point. The low temper-
atures recorded discounted any leakage of hot acetic acid or acetic
anhydride vapours past the condenser. Reflux of the reaction
mixture had been observed in the glass apparatus within the cavity
of the CEM MAS300 microwave oven at a much lower point than
where the guard tube was attached to the reactor condenser as-
sembly. We considered, as a possibility that an unknown, volatile,
microwave active intermediate had passed from the microwave
cavity through the condenser and reacted with the calcium chloride
granules. It was noted that the fused calcium chloride granules
smelt strongly of acetic acid when washed from the guard tube
following equipment disassembly. If escape of a highly volatile re-
active species was occurring, it might explain why the yields we
had obtained from atmospheric microwave Perkin reactions were
lower than expected (See Table 3).
ꢁ
Use of 300 C in order to prepare ‘freshly fused’ sodium acetate,
as recommended by Perkin, was found to be unnecessary by TG
analysis. TG analysis in Fig. 1 clearly shows that the majority of any
ꢁ
water of crystallisation or surface adsorption is lost by 158 C.
Moreover, the lower drying temperature we chose for sodium ac-
ꢁ
etate of 265 C, avoided the possible onset of decomposition, found
ꢁ
to be around 304e308 C during TG analysis (see Fig. 1), which we
had noted earlier in the study, when we had used a muffle-furnace
ꢁ
set-temperature of 300 C.
It was decided that with a sound set of revised reaction condi-
tions for thermal and microwave heating we should turn our in-
Ò
The PFA polymer-lined advanced composite vessels could be
sealed and this would prevent the escape of any volatile material.
Pressure and temperature measurements of a control vessel, which
was included in every microwave oven run, for programme control,
safety considerations and temperature feedback, could be made.
Unfortunately, the reliability of the control vessel yield measure-
ments was questionable when non-aqueous solvents were used.
This is because the pressure measurements relied on water to
transfer the pressure in the control vessel via a Teflon tube to
a transducer within the microwave oven control box. We thought
that the water in the pressure transfer line might react with volatile
or reactive intermediates in the control vessel or contaminate the
reaction within, rendering yield measurements meaningless.
Inspection of Table 7 above shows that the overall % molar yield
of cinnamic acid in the sealed vessels was more than twice that of
the atmospheric pressure reactions carried out in standard labo-
ratory glassware. The improvement in yield from sealed vessels,
along with the observation of fused calcium chloride granules in
the guard tubes and the lower than expected yields of the atmo-
spheric, microwave Perkin reactions provided strong evidence for
the presence of a key, gaseous, reactive intermediate, possibly ke-
tene, as postulated by Kinastowski, in his work on the mechanism
ꢁ
ꢃ1
Fig. 1. TGA analysis of anhydrous sodium acetate. Heating rate, 5 C min , under static
air with pierced aluminium pan. TGA shows about 1.59% w/w H
anhydrous sodium acetate.
2
O in this sample of
16
of the Perkin reaction when catalysed by tertiary amine.
Pressure values obtained from the control vessel, shown in
Fig. 2, provide additional evidence for this postulate through two,
clear, pressure increases in the control vessel during the reaction.
One pressure increase was observed at 1 h and the other at 3 h.
Repeatedly we had observed an induction period of about 60 min
when sodium acetate was used as the base in our Perkin reactions.
These observations had been made when conducting semi-
quantitative TLC analyses against a variety of known concentra-
tion standards. No induction period was observed, either when
potassium acetate was used as base, or when tetrabutylammonium
hydrogen sulfate was used in conjunction with sodium acetate.
Indeed, we considered that the latter observation with the imme-
diate formation of a very fine precipitate in the reaction mixtures
Ò
terest toward using perfluoroalkoxy, PFA , polymer liners that
were provided with the advanced composite vessels, ACV’s, of the
CEM MDS2100 microwave oven, rather than just standard labora-
tory glassware in an older, modified CEM MAS300 microwave oven.
Ò
Previously, PFA vessels had been shown to be advantageous in
analytical digestion work because of the increased pressure and
temperature available to the digestion process.
The PFA vessels that we used were translucent, unused and had
a nominal capacity around 75 mL.
We had observed in the earlier atmospheric pressure microwave
heated reactions that the calcium chloride granules present in the
guard tube attached to the top of the apparatus assembly had
15
Ò
Table 7
Comparison of atmospheric pressure and sealed microwave vessel Perkin reactions
Reaction
number
Reaction type
Molar ratios, benzaldehyde:
sodium acetate:acetic
anhydride
Isolated cinnamic
acid % molar yield
Total cinnamic acid
% molar yield based
on mass balance
calculations
Mean total %
molar yield
Standard deviation,
s
nꢃ1
2
2
3
3
3
3
3
3e28
Atmospheric
Sealed
1.0:1.0:2.3
1.0:1.0:2.3
d
d
17.6
36.3
0.57
0.29
9
0
1
2
3
4
28.2
28.8
29.0
28.8
28.7
28.9
36.5
36.1
36.7
36.3
35.9
36.4

M. Edwards et al. / Tetrahedron 70 (2014) 7245e7252
7249
programme it became evident very quickly that prior vessel usage
had a considerable effect on synthesis yields of the Perkin reaction
(
Table 9).
Aware of the possible impact on other synthetic work, as well as
chemical analyses, an alternative and quicker, vacuum de-
17
contamination method was developed by Riby et al.
Ò
The newer method, which involved heating PFA vessels to
ꢁ
1
55 C at 6 mmHg pressure, was used because of its time-saving
aspect of around 66 h over the saturated sodium carbonate solu-
tion decontamination soak.
The new decontamination method had a devastating effect
upon the yields of the Perkin reaction conducted in thermally
heated, PFA vessels.
Ò
Fig. 2. Perkin reaction pressure and temperature profiles obtained in the CEM
MDS2100 oven using PFA -lined ACV’s.
Ò
It can be seen from Table 10 that removal of volatile contami-
Ò
nants such as acid and water from the PFA vessel material led to no
observed reaction. This observation was a complete surprise and it
became important to ascertain what roles water and the vessel wall
material, PFA played. It is important to note that the decontami-
were good indications that a quick, metathetical exchange of so-
dium ions for tetrabutylammonium cations had occurred.
Ò
We do not know the reason(s) for this induction period with
sodium acetate. However, work published by Veverkova et al., on
microwave-stimulated Perkin reactions stated that no Cinnamic
acid was formed after 60 min of irradiation at 800 W, when sodium
nated vessels were immediately charged with reactants when
ꢁ
taken out of the vacuum oven, which was held at 155 C during the
15
acetate was used as the base, which supports our own experi-
mental observations.
Table 10
Effect of vacuum decontamination procedure on Perkin yields in sealed PFAÒ vessels
Ò
At the time we were uncertain if the PFA vessels had interacted
Reaction number
Isolated cinnamic
acid % molar yield
Total cinnamic acid %
molar yield from
mass balance
in some way with the microwave irradiation. It was decided
therefore to conduct a series of thermally heated reactions in the
same vessels using an oil-bath.
calculations
Both sets of reactions, detailed in Tables 7 and 8, were performed
54e56
57e59
Control vessel
for runs 54e56
Control vessel
for runs 57e59
0.0
0.0
0.0
0.0
0.0
10.7
ꢁ
ꢁ
at 147 C. However, the oil-bath was heated to 155 C and it took
Ò
ꢁ
around 18 min for the PFA vessels to reach 147 C. As a result of the
delay, the thermally heated vessel reactions listed in Table 8, had
a longer overall reaction time (þ7.5% relative). This discrepancy be-
tweenthetworeactionsetsadequatelyaccountsfortheslightlyhigher
yields observed (þ7.4% relative) from the thermally heated vessels.
0.0
11.0
Table 8
Comparison of thermal and microwave heating of Perkin reactions in PFAÒ vessels with benzaldehyde, sodium acetate and acetic anhydride molar ratios of 1.0:1.0:2.3
Run number
Heating
method
Isolated cinnamic
acid % molar yield
Total cinnamic acid %
molar yield based on
mass balance
Mean total cinnamic
acid % molar yield
Standard deviation
calculations
2
4
4
4
4
4
4
9e34
Microwave
Thermal
d
d
36.3
39.0
0.29
0.29
1
2
3
5
6
7
29.1
29.2
28.9
29.3
28.7
29.5
38.8
39.1
39.4
39.2
38.6
39.1
Up to this point a limited number of unused, assumed clean
decontamination procedure. Moreover, the sodium acetate used
was freshly dried at 265 C and similarly was taken directly from
the drying oven prior to charging the freshly decontaminated
vessels with reactants.
PFAÒ vessels had been utilised for this study. More PFAÒ vessels
were available but these had been used for acidic, analytical di-
gestion work. When these were incorporated into the synthetic
ꢁ
Table 9
Effect of prior acid digestion usage on Perkin reaction yields and the effect of a 72 h sodium carbonate solution decontamination soak on PFAÒ vessel performance
Reaction
number
Prior usage
Isolated cinnamic
acid % molar yield
Total cinnamic
acid % molar
yield based on
mass balance
calculations
Mean total cinnamic
acid % molar yield
Standard deviation,
s
nꢃ1
4
4
4
4
5
5
5
5
7a
7b
7c
7d
0
1
2
3
Acid digestion
Second Perkin
Third Perkin
Fourth Perkin
0.0
1.7
5.2
10.4
29.0
30.1
29.5
29.4
0.0
3.1
8.3
13.9
38.8
39.4
39.1
39.1
6.3
6.1
72 h Na
2
CO
3
decon.
39.1
0.25

7250
M. Edwards et al. / Tetrahedron 70 (2014) 7245e7252
It is clear from Table 10 above that the control vessels, which
decontamination soak. Whatever the true nature of the ‘ageing’
process is, it had a profound effect on the ability of PFA vessels to
tolerate large quantities of added water in the Perkin reaction
conditions that we had developed.
Ò
have a water transfer line and fibre optic temperature probe at-
tached, allowed the Perkin reaction to proceed, although in lower
yield than previously seen. It was postulated that water present in
the PFAÒ vessel walls had been removed by the vacuum de-
contamination procedure and that only the control vessels had
access to water via the pressure transfer line. We felt vindicated
that we had not used yield data from the control vessel when
monitoring all experimental runs.
3. Conclusions
We have found that the presence of water is essential for the
Perkin reaction to proceed. However, the amount of water required
for this reaction is not stoichiometrically related to the other re-
agent molar ratios of the reaction mixture: water behaves as
a catalyst. In a self-drying reaction mixture with fresh starting
materials, such as that found in the Perkin reaction, adventitious
ingress of water is the only way in which the reaction can proceed.
Furthermore, if a non-stoichiometric amount of water is sufficient
to allow the reaction to proceed to completion under our revised
conditions this suggests that water is regenerated during the re-
action mechanism in order to support its non-stoichiometric role.
These facts suggest that anhydrous acetic acid, the sole product
from the hydrolysis of acetic anhydride, plays a significant role in
the Perkin reaction.
It was then decided to repeat the reactions but to deliberately
add known amounts of water once the sealed vessels had cooled to
room temperature.
The results are shown in Table 11.
Table 11 shows that an equivalent yield of cinnamic acid to that
Ò
obtained fromvirgin PFA vessels was achieved when 70
mL of added
water (3.9 mmols), roughly 1/18th the scale of these reactions based
on the limiting reagent benzaldehyde. These results suggest that
water is needed in catalytic quantities in order to get the Perkin
Table 11
The effect of added water to sealed PFAÒ vessels, decontaminated by the vacuum
decontamination procedure, in the 4 h thermally heated Perkin reaction
This final conclusion contrasts strongly with the earlier con-
18
Reaction
number
Water added,
m
L
Isolated cinnamic
acid % molar yield
Total cinnamic acid %
molar yield based on
mass balance
clusions of Michael in which acetic acid was considered an in-
hibitor of the Perkin reaction, yet supports some of the unattributed
19
statements of Johnson, possibly due to his student Chappell, in his
comprehensive review.
calculations
5
6
6
6
6
4
4e59
0
0.0
10
0.0
9.6
0.0
11.8
We have shown a more than doubling of the Perkin reaction
product yield with a closed reaction vessel system. We believe that
the yield improvement in sealed vessels implicates the presence of
volatile species, which play important roles in the success of the
Perkin reaction. The two increases in pressure measurement ob-
served reinforce this claim and suggest that two volatile in-
termediates are involved in the overall transformation.
1
30
15.0
18.7
2
70
31.9
39.2
3
150
Virgin PFA
33.7
29.1 mean
41.5
39.0 mean
Ò
1e47
reaction to work. These facts when combined suggest that the cur-
rent mechanism of the Perkin reaction is incorrect as it does not take
into account our experimental observations. We thought it would be
interesting to see how much water could be tolerated by the Perkin
Preliminary, as yet unreported, kinetic work of ours suggests
that the Perkin reaction is second order when potassium acetate is
used as the base. In addition, over a period of about ten years fol-
lowing this study, we have been unable to provide any evidence for
the formation of Z-cinnamic acid under the carefully controlled
temperature conditions specified in this work. Such a finding
strongly suggests a cyclic intermediate, whose conformation en-
sures only E-cinnamic acid formation on ring-opening.
Ò
reaction when the reaction was conducted in PFA vessels (Fig. 3).
We were astonished to find that the Perkin reaction in thermally
Ò
heated PFA vessels could tolerate up to 30 mL of added water. The
final volume of water was restricted by the total, safe usable volume
Ò
available in the PFA vessels used.
We believe that this study supports the original proposition of
Kinastowski that ketene reacts with benzaldehyde to form an
oxetan-2-one intermediate. This study provides evidence for the
mechanism of formation of ketene by attack of the acetate ion on
the active methyl group of acetic acid, initially generated from the
adventitious hydrolysis of acetic anhydride causing the elimination
of water from anhydrous acetic acid across its carbonyl function.
We believe that the mechanism presented in Scheme 2 is a better
mechanistic representation of Perkin’s serendipitous reaction and
should, if proven, supplant all current, published named reaction
There was, however, a catch to this unusual reaction. After 83
consecutive reactions conducted with added water in a single set of
2
0
mechanisms.
All that we lack is the direct spectroscopic evidence for the
proposed, four membered -lactone.
b
The proposed intermediate, although racemic, should possess
interesting chemical shift and coupling data for comparison with
current H NMR spectrum prediction software packages because of
the adjacent chiral and pro-chiral centres as well as the strained
geometry of the oxetan-2-one ring system.
Fig. 3. A semi-logarithmic plot of the effect of added water on the yield of the Perkin
reaction heated thermally in sealed PFA vessels. Points above the 40% molar yield line
are the averages of at least three runs. 1snꢃ1 error bars are just visible.
1
Ò
PFAÒ vessels the tolerance to added water above 2.5 mL (2 M
Ò
equivalents) ceased. It was noted at this point that the PFA vessels
4. Experimental
4.1. General
had changed from being translucent to being dark brown coloured.
Ò
Clearly some ‘ageing’ process of the PFA polymer was taking place.
The dark colour suggested charred, organic material, which the
vacuum decontamination procedure did not remove. We did not
Acetic Anhydride, benzaldehyde and tetrabutylammonium hy-
reinvestigate the use of
a
sodium carbonate solution
drogen sulfate were purchased from SigmaeAldrich Company Ltd.

M. Edwards et al. / Tetrahedron 70 (2014) 7245e7252
7251
Scheme 2. Mechanism proposed for the Perkin reaction based on observations made during this study.
Anhydrous sodium acetate and anhydrous potassium acetate were
supplied by Lancaster Synthesis Ltd., UK. All reagents were used
directly without further purification. All solvents were supplied by
Fisher Scientific, Loughborough, UK and used without further pu-
rification. Reagents were weighed on a Sartorius BA310S top pan
balance. Anhydrous sodium acetate was dried in a Carbolite EML
0.30 and 90% CHCl
3
/9% CH
3
OH/1% HCOOH with an R
f
value of 0.36.
These systems were developed to provide cinnamic acid with
a variety of R values to ensure validation of product authenticity
f
and consistency of reaction profile. Developed TLC plates were
visualised using 254 nm UV light. Melting points were taken on
a Kofler hot-stage microscope melting point apparatus and are
uncorrected. IR spectra were obtained on a Perkin Elmer Paragon
1000 FTIR spectrometer using standard KBr disk techniques. NMR
spectra were obtained on a 270 MHz JEOL JNM-GX270 FTNMR us-
ꢁ
Eurotherm muffle-furnace at set-temperatures of 300 or 265 C for
at least 3 h. Microwave reactions were performed in either a mod-
ified CEM MAS300 800 W, or a CEM MDS2100, 1000 W multi-mode
microwave ovens. The MAS300 oven was modified by CEM UK, Ltd.,
with the middle of the top surface of the oven drilled with three
adjacent holes (2 side holesꢀ20 mm diameter and one central
holeꢀ40 mm diameter), which were fitted with three metal,
threaded chokes to prevent escape of microwave radiation. A
Holaday, HI-1501, 2450 MHz microwave oven radiation leak de-
tector survey meter was used to ensure microwave leakage was
6
ing deuteriated chloroform and DMSO-d as solvents with TMS as
internal standard.
Vacuum decontamination of PFAÒ vessels was carried out for
16 h under 6 mmHg pressure in a Gallenkamp vacuum oven at
ꢁ
155 C, fitted with an Edwards, E2M5 vacuum trolley pump oper-
ating under full gas ballast.
Sodium carbonate solution decontamination was carried out by
2
Ò
absent (<5 mW/cm ) from the modified oven during each run.
immersion of PFA vessels in a saturated sodium carbonate solu-
Atmospheric pressure microwave reactions were carried out in
tion for 72 h at room temperature.
standard laboratory glassware in the modified, CEM MAS300 oven.
Ò
Sealed microwave reactions were carried out in PFA liners con-
tained in Advanced Composite Vessels as supplied by CEM UK, Ltd.,
within the CEM MDS2100 oven. Thermogravimetric analyses were
performed using a DuPont 951 thermogravimetric analyser using
V4.10 DuPont 2000 controller software. Thermal reactions were
carried out using an IKAMAG magnetic, hotplate stirrer fitted with
an IKATRON temperature feedback probe and fuzzy logic controller.
Water addition was performed using variable volume Gilson pi-
pettes, Pipetman ClassicÔ models: P100, P200, P1000 and P5000.
TLC analyses were carried out using Merck 20ꢀ20 cm, 60F254 silica
gel plates. Glass TLC tanks were fitted with two pieces of Whatman
4.2. General procedure for atmospheric pressure microwave
reactions
Freshly dried anhydrous sodium acetate (5.65 g, 68.9 mmol) was
suspended in acetic anhydride (15 mL, 159 mmol) in a three necked
round bottom flask. Benzaldehyde (7 mL, 68.9 mmol) was added
directly to the mixture and placed in a CEM MAS300 oven and then
fitted with three air condensers in line (total dimension:
B24ꢀ1200 mm) and fitted with an adapter for thermometer and
guard tube filled with fresh calcium chloride granules. A 600 mL
glass beaker, which contained 450 mL of distilled water, was placed
in the microwave oven with the reaction assembly in order to ab-
sorb reflected microwave energy and thereby avoid damage to the
magnetron. The oven was programmed at 60% power. Every 20 min
or so water was added to the beaker in the oven to replace that lost
through boiling. The run was complete after 4 h of microwave
irradiation.
1
Chr Chromatography paper, 10ꢀ10 cm, to act as wicks for the
solvent systems. TLC solvent systems were developed using a vari-
ety of solvent mixtures selected from chloroform, methanol, ethyl
acetate, toluene and hexane solvents modified with up to 1%v/v
formic acid to prevent tailing of TLC spots. Solvent system com-
positions were: 49% hexane/40% acetone/1% HCOOH with an R
value of 0.39; 85% toluene/14% EtOAc/1% HCOOH with an R value of
f
f

7252
M. Edwards et al. / Tetrahedron 70 (2014) 7245e7252
ꢁ
The reaction mixture was allowed to cool to around 80 C and
condenser (B14ꢀ230 mm). Reactants were charged in the same
order and scale as in the earlier method. The reaction vessel was
immersed in a stirred, thermostatically controlled oil-bath at
the mixture was quenched with water (20 mL) prior to transfer to
a 500 mL conical flask. The transferred solution was maintained at
0 C on a hotplate while a hot, saturated solution of sodium car-
ꢁ
ꢁ
8
155 C. The reaction was heated at reflux for 4 h. After this time
bonate (50 mL) was carefully added. Further aliquots of hot, satu-
rated sodium carbonate solution were added, after the initial strong
effervescence, until effervescence ceased and a consistent reading
of pH10 was obtained with an indicator test-strip. The hydrolysis
and acid dissolution stage could be lengthy if the product was tarry/
oily in any way. A value of pH10 could be achieved quickly and then
this value would decrease over time as any tarry or oily product
gradually dissolved. Care must be taken to ensure that a consistent
pH10 value is achieved. The reaction mixture was allowed to cool to
room temperature and was extracted with diethyl ether (3ꢀ15 mL).
The extracts were combined, dried over sodium sulfate, which was
removed by filtration, and finally evaporated in vacuo for weighing,
TLC analyses and mass balance calculations. The remaining aqueous
solution was then acidified with concd HCl until carbon dioxide
evolution ceased. The white precipitate was recrystallized in situ.
The recrystallized solution was cooled finally in ice-water for 1 h
prior to isolation by filtration under vacuum. The colourless, crys-
talline cinnamic acid isolated was dried under vacuum overnight.
the reaction mixture was worked-up as described in the earlier
method.
Acknowledgements
Two of us would like to extend our thanks to D. Lofty of CEM UK
Ltd., for the modifications made to the CEM MAS300 microwave
oven and the loan of a Holaday, HI 1501 microwave oven leak de-
tection survey meter.
One of us would like to thank W.J.A. Chalkley for the digital
preparation of archived instrumental output data.
Professor Alexander McKillop was insistent that his students
could demonstrate a firm mechanistic understanding of the re-
actions that they studied or employed in their organic synthesis
research. This article is dedicated to the memory of Sandy’s out-
standing teaching contribution in this area and his belief that
answers lay in the laboratory not in the library. His influence
touched many and extended beyond the confines of academia to
underpin an industrial career where his insistence on the mech-
anistic understanding of organic reactions provided a solid foun-
dation for safe, cost-efficient, large-scale pharmaceutical
production.
ꢁ
ꢁ
21
Melting point 133e134 C (Lit. Mp, 133 C). IR, NMR and TLC were
consistent with authentic material. There was no depression of
melting point observed when the isolated cinnamic acid was mixed
with commercially available material.
4
.3. General procedure for sealed, microwave Perkin
reactions
References and notes
Four CEM advanced composite vessels were charged each with
freshly dried sodium acetate (5.65 g, 68.9 mmol), acetic anhydride
15 mL, 159 mmol) and benzaldehyde (7 mL, 68.9 mmol). The
1. Gedye, R. N.; Rank, W.; Westaway, K. C. Can. J. Chem. 1991, 69, 706.
(
2. Mingos, D. M. P.; Baghurst, D. R. J. Chem. Soc. Rev. 1991, 20, 1.
3. Kingston, H. M.; Jassie, L. B. Introduction to Microwave Sample Preparation:
Theory and Practice. ACS Professional Reference Book Series; American Chem-
ical Society, 1988.
control vessel was fitted with a temperature probe and topped-up
pressure transfer line. The vessels were placed in a CEM MDS2100
microwave oven and the unit was programmed to operate at 80%
4
. CEM Company History, http://www.cem.com/content78.html.
5. Perkin, W. H. J. Chem. Soc. 1868, 21 53, 181.
. Johnson, J. R. In Organic Reactions; Adams, R., Ed.; John Wiley: New York, NY,
942; Vol. 1, pp 218e265.
. Vogel, A. I. A Text-book of Practical Organic Chemistry, 3rd ed.; Longman: London,
ꢁ
power for 4 h at 147 C. After 4 h the reaction mixtures were
6
worked-up as described in the previous experimental method.
1
7
4
.4. General procedure for thermally heated, sealed Perkin
1956; pp p.712e713.
. Abramovitch, R. A.; Abramovitch, D. A.; Iyanar, K.; Tamaraselvy, K. Tetrahedron
Lett. 1991, 32, 5251.
. Kaldin, H. In Profiles of Drug Substances, Excipients and Related Methodology;
Florey, K., Ed.; Academic: New York, USA, 1982; Vol. 11; Chapter 3 Captopril,
p 114.
0. Kalnin, P. Helv. Chim. Acta 1928, 11, 977.
1. B o€ ck, F.; Lock, G.; Schmidt, K. Monatsh. Chem. 1934, 64, 401.
12. Veverkova, E.; Pacherova, E.; Toma, S. Chem. Pap. 1999, 53, 257.
13. Meyer, H.; Beer, R. Monatsh. Chem. 1913, 34, 649.
14. Sykes, P. A Guidebook to Mechanism in Organic Chemistry, 6th ed.; Longman:
Scientific & Technical: Harlow, Essex, UK, 1986; pp 227e228.
5. Kingston, H. M.; Jassie, L. B. Anal. Chem. 1986, 58, 2534.
16. Kinastowski, S.; Kasprzyk, H. Bull. Acad. Pol. Sci. 1978, 26, 907.
7. Mason, C. J.; Edwards, M.; Riby, P. Analyst 2000, 125, 327.
8. Michael, A. J. Prakt. Chem. 1889, 40, 64.
9. Chappell, F. L. Studies of the Perkin Reaction; Cornell University, 1933; (PhD
8
9
reactions
Four CEM advanced composite vessels were charged with re-
actants at the same scale as in the prior method. The control vessel
was monitored for pressure and temperature through the control
unit of the CEM MDS2100 oven. All four vessels were placed in
a thermostatically controlled, stirred, oil-bath heated at 155 C in
order to achieve reflux temperature (147 C) inside the fluoropol-
1
1
ꢁ
ꢁ
ꢁ
ymer vessels. After 4 h at 147 C the reactions were worked-up as
described in the previous experimental method.
1
1
1
1
4
.5. General procedure for thermally heated, atmospheric
Perkin reactions
Thesis).
2
0. Mundy, B. P.; Ellerd, M. G.; Favaloro, F. G., Jr. Name Reactions and Reagents in
Organic Synthesis, 2nd ed.; Wiley-Interscience, John Wiley & Sons: Hoboken,
New Jersey, USA, 2005; pp 492e493.
A three neck round bottom flask (150 mL) was fitted with
ꢁ
thermometer (0e250 C),
8
mm PTFE magnetic flea and
21. Bahadur Singh, N.; Kumar, P. J. Chem. Eng. Data 1986, 31, 406.