On the mechanism of a modified Perkin condensation leading to α-phenylcinnamic acid stereoisomers - Experiments and molecular modelling

Article abstract of DOI:10.1007/s007060070043

A modified Perkin condensation leading to α-phenylcinnamic acid stereoisomers affords predominantly or exclusively the (E)-isomer. Reaction duration, temperature, and polarity of the solvent affect the isomeric distribution only to a minor extent. However, geometry optimization of the stereoisomers by semiempirical quantum chemical methods revealed that their thermodynamic stabilities in the absolute minimum are the same. (E)-(Z)-isomerization under the conditions of the Perkin condensation resulted in an equilibrium mixture of nearly 1:1 composition, thus corroborating the results of the computations. Comparing the detailed potential energy maps of the isomers obtained at the semiempirical level of theory may explain the controversy. The (Z)-isomer has a structured potential energy surface with few but well-defined minima, whereas that of the (E)-isomer shows an extended flat plateau enabling this isomer to approach a minimum (which is close to the global one) right after its formation.

Full text of DOI:10.1007/s007060070043

Monatshefte fuÈr Chemie 131, 1097±1104 (2000)
On the Mechanism of a Modi®ed Perkin
Condensation Leading to a-Phenylcinnamic
Acid Stereoisomers ± Experiments and
Molecular Modelling
2
1
3
Â1;Ã
Â
Â
Â
Â
Istvan Palinko , Akos Kukovecz , Bela Torok , and Tamas Kortvelyesi
Â
Â
È È
È
1
Department of Organic Chemistry, University of Szeged, Szeged, H-6720 Hungary
2
Department of Applied and Environmental Chemistry, University of Szeged, Szeged, H-6720
Hungary
3
Department of Physical Chemistry, University of Szeged, Szeged, H-6720 Hungary
Summary. A modi®ed Perkin condensation leading to ꢀ-phenylcinnamic acid stereoisomers affords
predominantely or exclusively the (E)-isomer. Reaction duration, temperature, and polarity of the
solvent affect the isomeric distribution only to a minor extent. However, geometry optimization of
the stereoisomers by semiempirical quantum chemical methods revealed that their thermodynamic
stabilities in the absolute minimum are the same. (E)-(Z)-isomerization under the conditions of the
Perkin condensation resulted in an equilibrium mixture of nearly 1:1 composition, thus corroborating
the results of the computations. Comparing the detailed potential energy maps of the isomers
obtained at the semiempirical level of theory may explain the controversy. The (Z)-isomer has a
structured potential energy surface with few but well-de®ned minima, whereas that of the (E)-isomer
shows an extended ¯at plateau enabling this isomer to approach a minimum (which is close to the
global one) right after its formation.
Keywords. Isomeric distribution; Potential energy surfaces; Semiempirical quantum chemical
methods; Stereoselectivity.
Introduction
Cinnamic acid derivatives are important synthons in the production of lignins in
higher plants [1]. They are members of the shikimic acid metabolic pathway, and
their mechanism of formation is complex. Nevertheless, key reactions are
condensations (mostly of the Claisen type [1]), just as in the laboratory synthesis
(mostly of the Perkin type [2]). The usual Perkin reaction leads to the predominant
or exclusive formation of the (E)-isomer. A modi®ed Perkin condensation (Scheme
1) affords under certain conditions (low temperature, short reaction time) a mixture
of (E)- and (Z)-ꢀ-phenylcinnamic acids with (E)-isomer predominating [3±5]. Early
studies (see e.g. Ref. [6]) found this fact obvious, since the (E)-isomer was
Ã
Corresponding author

 Â
I. Palinko et al.
1098
Scheme 1
considered to be thermodynamically more stable. Indeed, reactions for prolonged
time periods afforded exclusively this isomer. However, when we started to model
the two isomers in order to ®nd a structural explanation for the large difference in
the acidities of the stereoisomers, AM1 [7], MNDO [8], and PM3 [9] semiempirical
quantum chemical calculations gave nearly identical heats of formation for the fully
optimized structures of the isolated (E)- and (Z)-isomers [10]. Therefore, it seemed
to be interesting to study the reaction from a mechanistic point of view, with the aim
of exploring factors in¯uencing the isomeric distribution. The thermal (under the
conditions of the modi®ed Perkin condensation) and photoinduced (E)-(Z) (and (Z)-
(E)) isomerizations of the acids were also investigated. The isomerization reactions
were explored experimentally as well as computationally.
Results and Discussion
The modi®ed Perkin reaction provided an isomeric mixture with a predominance of
the (E)-isomer. The maximum amount of the (Z)-diastereomer was 21%; it was
obtained when the Fieser mixture was kept under re¯ux (about 150ꢀC) for 35 min.
A compilation of literature results revealed that any modi®cation in the composition
of reactants, temperature, and duration of re¯ux resulted in the preferential or in
most cases exclusive formation of the (E)-isomer (Table 1).
Interestingly, full geometry optimization at the semiempirical quantum chemical
level of theory on the isolated molecules gave similar stability data for the two
isomers in the global minima. However, calculations also revealed that the dipole
moments were different (Table 2).
Table 1. Isomeric distribution of ꢀ-phenylcinnamic acids at various temperatures and reaction times
based on literature data
T/ꢀC
t/min
benza/mol
Ac2Ob/mol
phacc/mol
Et3Nd/mol
(E)/(Z)
30e
30e
90
90
0.74
0.74
0.74
0.74
0.18
0.30
0.40
0.30
0.30
0.30
0.30
2.10
0.21
0.84
0.20
0.20
0.20
0.20
0.11
0.18
0.40
0.20
0.20i
0.40
0.40i
1.50
0.15
0.29
100/0
100/0
100/0
100/0
96/4
65e
180
180
35
65e
140f
150g
150h
35
79/21
100/0
330
a
Benzaldehyde; b acetic anhydride; c phenylacetic acid; d triethylamine; e Ref. [11]; f Ref. [6];
g Ref. [4]; h Ref. [5]; added dropwise during the reaction
i

Stereochemistry of the Formation of ꢀ-Phenylcinnamic Acid
1099
Table 2. Standard heats of formation and dipole moments for the ꢀ-phenylcinnamic acid diaster-
eomers calculated by semiempirical quantum chemical methods
MNDO
AM1
PM3
1
1
1
ÁHf;298/kJ Á mol
115.1
ꢁ/D
ÁHf;298/kJ Á mol
100.5
ꢁ/D
ÁHf;298/kJ Á mol
98
ꢁ/D
(Z)
(E)
1.81
2.10
1.80
2.54
1.75
2.00
113.9
98.6
97.6
Table 3. Isomeric distribution in solvents of different dielectric constants, dipole moments, and
boiling points at various reaction times (the reactions were carried out under re¯ux)
Solvent
Fieser mixturea (b.p.: app. 150ꢀC)
t/min
(E)/mol%
(Z)/mol%
35
78.9
94.4
100
21.1
5.6
0
H2O (b.p.: 100ꢀC, "ꢁ20†: 80.1)
180
Diethyl ether (b.p.: 35ꢀC, "ꢁ20†: 4.34, ꢁꢁ20†: 1.15)
180
Dibutyl ether (b.p.: 142±143ꢀC, "ꢁ20†: 3.08, ꢁꢁ20†: 1.18)
40
Diglyme (b.p.: 162ꢀC, ꢁꢁ25†: 1.97)
180
100
0
100
0
60†
Benzene (b.p.: 80ꢀC, "ꢁ25†: 2.275, ꢁꢁ20
: 0)
180
100
0
Xylene (isomeric mixture; b.p.: 138±142ꢀC, "ꢁ20†: 2.37 ꢁꢁ20†: 0.30)
30
100
98.1
0
180
1.9
2.1
360
Decalin (isomeric mixture; b.p.: 192ꢀC, "ꢁ20†: 2.19, ꢁꢁ25†: 0)
97.9
35
360
540
94.2
87.4
90.5
5.8
12.6
9.5
a
See experimental part
In order to explore the possible effect of the different dipole moments upon
the reaction, solvents of different properties (varying dielectric constant and/or
hydrogen bonding capability) were used. However, the amount of the (Z)-isomer
did not increase. Moreover, the exclusive formation of the (E)-isomer can be
considered typical. The (Z)-isomer was formed (in small amounts even under these
circumstances) when high temperature, long reaction time, and nonpolar solvent
were used or the solvent was capable of hydrogen bonding (Table 3).
However, isomerization reactions, starting from either the (E)- or the (Z)-isomer,
resulted in a thermal equilibrium mixture containing 55 mol% (E)- and 45 mol%
(Z)-isomer under the conditions of the modi®ed Perkin reaction, and UV irradiation
produced a photo-equilibrium mixture of 61 mol% (E)- and 39 mol% (Z)-isomer.
These experiments nicely corroborated the results of the semiempirical calculations.

 Â
I. Palinko et al.
1100
Scheme 2
Scheme 3
Thus, the real reason of the predominance of the (E)-isomer in the formation
reaction is not due to different thermodynamic stabilities of the stereoisomers.
The modi®ed Perkin condensation consists of two parts: (1) nucleophilic
addition of the mixed anhydride formed from acetic anhydride and phenylacetic
acid to the carbonyl group of benzaldehyde via base catalysis (Scheme 2)

Stereochemistry of the Formation of ꢀ-Phenylcinnamic Acid
1101
Table 4. Absolute con®gurations of the diastereomers and those of the products after anti-elimination
Chiral centres (cf. Fig. 1)
Product after
elimination
a
b
1
2
3
4
(S)
(R)
(S)
(R)
(R)
(R)
(S)
(S)
(Z)
(E)
(E)
(Z)
and (2) elimination of acetic acid and hydrolysis of the product anhydride
(Scheme 3).
Zimmermann et al. have shown that the condensation reaction is elimination
controlled [6]. There is no reason to assume that preferential formation of any of
the four possible intermediates takes place. Assuming anti-elimination, the
intermediates are able to produce both stereoisomers (Table 4).
Semiempirical quantum chemical calculations on the possible intermediates
(Fig. 1) revealed that the heats of formation are very close to each other, and their
dipole moments are similar (Table 5). This means that the different distribution of
the ®nal stereoisomers of the reaction did not result from differences in the stability
or dipole moments of the intermediates. It is not necessary to assume syn-
elimination to account for any of the isomers either.
A comparison of the stability data of the intermediates and the ꢀ-phenyl-
cinnamic acid stereoisomers reveals that the elimination reaction is endothermic.
Strictly speaking, this holds only for gas-phase reactions; nevertheless, it should be
Fig. 1. Structure of intermediates (before elimination) in the modi®ed Perkin condensation (a, b:
stereocentres)
Table 5. Standard heats of formation and dipole moments for 1±4 obtained by the AM1 method
1
ÁHf;298/kJ Á mol
ꢁ/D
1
2
3
4
659.7
654.6
654.5
659.7
4.58
5.24
5.24
4.58

 Â
I. Palinko et al.
1102
Fig. 2. Potential energy map of (E)-ꢀ-phenylcinnamic acid calculated by the AM1 method
Fig. 3. Potential energy map of (Z)-ꢀ-phenylcinnamic acid calculated by the AM1 method
approximately valid in our system as well. Thus, the transition state is product-like,
and the potential energy surfaces of the products may therefore be used in
explaining stereoselectivity towards the (E)-isomer.
The potential energy maps of the two diastereomers differ considerably. They
were obtained by rotating the phenyl groups around the ꢂ-bond between the
ole®nic carbon and the phenyl carbon atoms (ꢃ1 and ꢄ1 correspond to position 3
(ꢅ-position) for the (E)- and (Z)-isomers, respectively, and ꢃ2 and ꢄ2 correspond in
a similar way to position 2 (ꢀ-position). The potential energy surface for the (E)-
isomer contains extensive low-energy regions close to the absolute minimum
throughout (Fig. 2). This table-like surface contains ridges of very small heights

Stereochemistry of the Formation of ꢀ-Phenylcinnamic Acid
1103
between the shallow valley-like minima. There are maxima when the hydrogens on
the phenyl groups approach each other.
In contrast, the potential energy surface for the (Z)-isomer is complex with
several local minima (Fig. 3). The absolute minima of both stereoisomers are very
close to each other. Large portions of the surface are in the higher energy region
(saddles, ridges, local maxima).
Consequently, the reason for the preferential formation of the (E)-isomer is of
kinetic origin. After formation, this molecule easily ®nds an arrangement close to
the absolute minimum, whereas it is more dif®cult for the (Z)-isomer to ®nd
minimum conformations.
Conclusions
Using the modi®ed Perkin condensation leading to ꢀ-phenylcinnamic acid isomers
as a model reaction it has been demonstrated that experimental and computational
methods together are able to give a move complete account of the observed
stereochemical features than any of them alone. Applying the two different
approaches in combination, the controversy between the isomer distribution in the
condensation reaction and that in the (E)-(Z)-isomerization was resolved, and an
explanation of the observed experimental facts could be given.
Experimental
Condensation reaction
First, the reaction was performed following the recipe of Fieser [4]. It involved heating a mixture of
2 cm3 benzaldehyde, 2.5 g phenylacetic acid, 2 cm3 acetic anhydride, and 2 cm3 triethylamine for
35 min. The mixture of products was precipitated with 4 cm3 of conc. HCl. The solid material was
dissolved in diethyl ether and then washed with 3Â10 cm3 of 3% NaOH solution. The two isomers
from the alkaline solution were obtained by selective precipitation. Acidifying with acetic acid to
pH ˆ 5 afforded the (E)-isomer, further lowering the pH to 1 with conc. HCl the (Z)-isomer. The
crude products were crystallized from diethyl ether (m.p.: (E): 446.5 K, (Z): 409.5 K, close to the
values published in the literature [4]).
In another set of experiments, duration and temperature of re¯ux were altered by variation of the
solvent. Usually, 100 cm3 of solvent was used to enforce the properties of the solvent on the reaction
mixture (the quantities of the other components remained unchanged). Duration of re¯ux was varied
from 35 min to 9 h in some cases. Products were isolated and separated as described above.
Thermal and photoinduced isomerization reactions
Thermal isomerization was performed in the Fieser mixture, i.e. 0.5 g of (E)-ꢀ-phenylcinnamic acid
were added to a 1:1 mixture of acetic anhydride and triethyl amine (2 cm3 each), and the system was
kept under re¯ux for 168 h. For determining the isomeric composition, samples were withdrawn for
GC-MS analysis at certain time intervals [12].
Photoinduced isomerization was performed with the help of an immersion mercury vapour lamp
(Hanau GmbH, intensity maximum at ꢆ ˆ 253:1 nm). The reaction was performed at room
temperature in a jacketed ¯ask (water was used for controlling the temperature) under a nitrogen
atmosphere on 1 g of (E)-ꢀ-phenylcinnamic acid dissolved in 500 cm3 of diethyl ether. The duration
of illumination was 30 h, and analysis was performed by means of the GC-MS method.

 Â
I. Palinko et al.: Stereochemistry of the Formation of ꢀ-Phenylcinnamic Acid
1104
Computational methods
Geometry optimizations for the individual molecules were performed by semiempirical quantum
chemical methods (MNDO, AM1, and PM3) included in the Spartan [13] or MOPAC93 [14]
package. The gradient norms were always less than 0.1, and the force matrices were found to be
positive de®nites verifying that minima were found.
Potential energy maps were computed for both isomers using the AM1 [7] routine in the Spartan
[13] package. In creating a detailed surface the two phenyl groups were rotated by 1ꢀ increments,
and potential energies were calculated for each conformer. The carboxyl group was allowed to
assume an optimal conformation before energy calculation. Thus, fairly detailed maps were
obtained. Potential energy values of 32580 points were calculated.
Acknowledgements
Financial support for this work provided by the Ministry of Education via grant FKFP 0286/2000 is
gratefully acknowledged.
References
[1] Mann J (1987) In: Secondary Metabolism, Clarendon Press, Oxford, p 173
[2] Johnson J (1942) In: Adams R, Bachmann WF, Johnson JR, Fieser LF, Snyder HR (eds) Organic
Reactions. Wiley, New York London, p 210
[3] Buckles RE (1955) J Chem Ed 27: 210
[4] Fieser LF, Fieser M (1955) In: Experiments in Organic Chemistry. Heath and Co., Boston, p 182
[5] Buckles RE, Bremer K (1967) Org Synt, vol 4. Wiley, New York, p 777
[6] Zimmermann H, Ahramjian L (1959) J Am Chem Soc 81: 2086
[7] Dewar MJS, Zoebisch EG, Healy EF, Stewart JJP (1985) J Am Chem Soc 107: 3902
[8] Dewar MJS, Thiel W (1977) J Am Chem Soc 99: 4899
[9] Stewart JP (1989) J Comput Chem 10: 209, 221
Â
[10] Tasi Gy, Palinko I, Kortvelyesi T, Nyerges L (1997) J Mol Struct (THEOCHEM) 391: 189
Â
Â
È
[11] Buckles RE, Cooper JA (1965) J Org Chem 30: 1588
[12] Hammond GS (1955) J Am Chem Soc 77: 334
Â
Â
Â
È È
[13] Torok B, Palinko I, Tasi Gy, Nyerges L, Bogar F (1994) J Chrom A 668: 353
[14] Spartan 4.1.1 (1995) Wavefunction, Inc, 18041 von Karman Ave. 370, Irvine, CA 92715, USA
[15] Mopac 93 (1993) Fujitsu, Ltd
Received May 7, 2000. Accepted (revised) June 13, 2000