Aromatics from pyrones: Esters of terephthalic acid and isophthalic acid from methyl coumalate

Article abstract of DOI:10.1039/c3ra42287a

The Diels-Alder reaction of methyl coumalate with alkenes bearing electron-withdrawing groups provides terephthalates or isophthalates in good yields, with the regioselectivity depending on the electron-withdrawing group. The reaction of methyl coumalate with the salt of acrylic acid gave only the monoester of isophthalic acid. Density functional theory (B3LYP/6-31 + G(d,p)) computations of the energies of the competing transition states of the para-selective Diels-Alder reactions are in good agreement with experiment. The surprising regioselectivity of methyl coumalate with activated alkenes is attributed to a secondary orbital interaction between the pyrone oxygen and the dienophile LUMO, which switches the regiochemistry expected from simple frontier molecular orbital theory arguments. The Royal Society of Chemistry 2013.

Full text of DOI:10.1039/c3ra42287a

RSC Advances
View Article Online
View Journal
PAPER
Aromatics from pyrones: esters of terephthalic acid and
isophthalic acid from methyl coumalate3
Cite this: DOI: 10.1039/c3ra42287a
ab
a
a
a
George A. Kraus,* Gerald R. Pollock III, Christie L. Beck, Kyle Palmer and Arthur
H. Winter
a
The Diels–Alder reaction of methyl coumalate with alkenes bearing electron-withdrawing groups provides
terephthalates or isophthalates in good yields, with the regioselectivity depending on the electron-
withdrawing group. The reaction of methyl coumalate with the salt of acrylic acid gave only the
monoester of isophthalic acid. Density functional theory (B3LYP/6-31 + G(d,p)) computations of the
energies of the competing transition states of the para-selective Diels–Alder reactions are in good
agreement with experiment. The surprising regioselectivity of methyl coumalate with activated alkenes is
attributed to a secondary orbital interaction between the pyrone oxygen and the dienophile LUMO, which
switches the regiochemistry expected from simple frontier molecular orbital theory arguments.
Received 9th May 2013,
Accepted 23rd May 2013
DOI: 10.1039/c3ra42287a
www.rsc.org/advances
step could be avoided if an acrylate or acrylate equivalent
could be employed in the cycloaddition step.
Introduction
Terephthalic acid is a commodity chemical produced from
petroleum feedstocks. The most common synthesis pathway
The reaction of activated alkenes with methyl coumalate
produces bicyclic lactones that are oxidized in situ to the
aromatic system using Pd/C as a catalyst, as shown in
Scheme 2. The use of an activated alkene such as methyl or
ethyl acrylate allowed us to conduct the reaction at a much
lower temperature than was possible with the unactivated
alkenes. Unfortunately, the ratio of 1,4- to 1,3-disubstituted
products was only about 3 : 1 in both cases. The yields of these
reactions were 25% and 38%, respectively. Previous reports
with pyrones by a number of researchers showed a mixture of
1
is the oxidation of para-xylene using a cobalt/manganese
catalyst in acetic acid. Terephthalic acid and dimethyl
terephthalate are employed in the preparation of polyethylene
terephthalate (PET), a thermoplastic polymer used in many
beverage and food containers and in fabrics. Global produc-
tion of terephthalic acid was almost fifty million tons. As part
of a collaborative effort to produce biorenewable chemicals
using enzyme catalysis followed by chemical catalysis, we
describe herein an environmentally benign synthesis of esters
of isophthalic acid and terephthalic acid and related aromatics
from methyl coumalate.
The Diels–Alder reaction of 2-pyrones with alkenes and
alkynes has strong literature precedent.
1
2
7
products when activated alkenes were employed. Raecke and
Ogata described reaction conditions to isomerize the potas-
sium salts of phthalic acid and isophthalic acid into
8,9
terephthalic acid, albeit under harsh conditions.
3
,4
We recently
It is clear from Scheme 2 that the two esters are closer in
the bicyclic intermediate leading to the meta-adduct than they
are in the intermediate toward the para-adduct. If the steric
demand of the ester group was significantly increased, there
may be an opportunity to direct the regioselectivity to favor the
para-diester. Reaction of acrylic acid with diisopropylethyla-
mine generated a salt which was reacted with methyl
coumalate at 140 uC. The major product of this reaction was
the mono ester of isophthalic acid in 45% yield. The rationale
reported the regioselective Diels–Alder reactions of coumalic
acid and methyl coumalate (1) with alpha-olefins, shown in
5
Scheme 1. The reaction involves a Diels–Alder reaction to
produce an oxabicyclo[2.2.2]octene intermediate that can be
dehydrogenated by a commercial Pd/C catalyst with loss of
carbon dioxide to form the para-substituted benzoate. This
reaction was successful with a variety of alpha-olefins.
Oxidation of the alkyl group R in Scheme 1 is known to
6
produce the mono ester of terephthalic acid. This oxidation
a
Department of Chemistry, Iowa State University, Ames, IA 50011, USA.
E-mail: gakraus@iastate.edu; Fax: +1 515-294-0105; Tel: +1 515-294-7794
b
NSF Engineering Research Center for Biorenewable Chemicals, Iowa State
University, Ames, IA, USA
3
Electronic supplementary information (ESI) available: Details and parameters
of computational study. See DOI: 10.1039/c3ra42287a
Scheme 1 Diels–Alder reactions with alkenes.
This journal is ß The Royal Society of Chemistry 2013
RSC Adv.

View Article Online
Paper
RSC Advances
Scheme 5 Diels–Alder reactions with acrylonitrile and acrolein.
substitution. The conversion of para-formyl benzoates into the
1
1
Scheme 2 Diels–Alder reaction with activated alkenes. para-selectivity was
terephthalates has been reported by Borhan.
2
.9 : 1.
Although we initially explored the influence of steric effects
on the partition between meta- and para-products, it is clear
from these experiments that electronic effects also play an
influential role. In order to better understand the present
results and perhaps later predict the regioselectivity of future
experiments, calculations using density functional theory were
performed.
Scheme 3 Diels–Alder reaction with an amine salt of acrylic acid.
Computational methods
for this intriguing result is not clear; however, it is possible
that the Diels–Alder reaction proceeds by way of the exo
transition state shown below in Scheme 3 to minimize non-
bonded interactions. Attempts to use the salt of coumalic acid
were limited by the poor solubility of this compound in
organic solvents.
All density functional theory (DFT) calculations were per-
formed using the Gaussian09 software suite employing the
B3LYP functional, which consists of the Becke hybrid
12
3
-parameter exchange functional with the correlation func-
13
tional of Lee, Yang, and Parr. The 3-21G basis set was
employed to identify the lowest energy product rotamers, and
the 6-31 + G(d,p) basis set was used to calculate the geometries
and energies of the starting materials, products, and transition
states. In all cases for starting materials and products,
optimized geometries were found to have zero imaginary
frequencies, indicating that the structures represent local
minima on the potential energy surfaces. Energy corrections
for the zero-point vibrational energy were added unscaled.
Transition states for forming these products were located
using the Quasi-Newton Synchronous Transit method (QST3).
Of note, transition state barriers for related Diels–Alder
reactions with this level of theory have been shown to give
Propiolic acid and its methyl ester were also examined. Use
of acetylenic esters had the advantage that the Pd/C catalyst
was not needed. The additional double bond in the bicyclic
intermediates shown in Scheme 4 enforces the proximity of
the two esters in the bicyclic intermediate leading to the meta-
product. Neither compound afforded any regioselectivity.
Propiolic acid gave a mixture of products in 64% yield while
methyl propiolate yielded 58% of the products. Use of the tert-
butyl ester led to a mixture of products involving loss of the
tert-butyl group. Since propiolic acid salts have been reported
1
0
to decompose under thermal conditions, this direction was
not pursued.
1
4
good agreement with experiment. Computed transition state
structures were all found to have one imaginary frequency that
connected the starting materials and the product.
The reaction of methyl coumalate with acrylonitrile
(Scheme 5, X = CN) provided better regioselectivity than the
propiolate system, giving the mixture of products in 60% yield
with a 1.7 : 1 ratio in favor of the para-substituted product.
Interestingly, the reaction with acrolein afforded a mixture of
adducts in 47% yield with a 4.3 : 1 ratio in favor of the para
The meta or para benzene products can arise from one of
two bicyclic intermediate stereoisomers. Additionally, each of
the two stereoisomers can exist in several different substituent
orientations (rotamers). For example, the Diels–Alder reaction
of methyl coumalate with methyl acrylate can lead to two
bicyclic stereoisomers, each of which has four different major
rotamers. These structures are depicted in Scheme 6. Similar
isomers and rotamers were found for the products of the
reactions of methyl coumalate with acrolein and acrylonitrile.
The lowest-energy rotamer for each isomer for the
carbomethoxy, oxo, and cyano-substituted Diels–Alder pro-
ducts was determined by optimizing each structure at the
B3LYP/3-21G level of theory. The lowest energy rotamer for
each stereoisomer computed at this level of theory was used as
the product for transition state searches. The geometries and
energies of the starting materials, product, and transition
Scheme 4 Diels–Alder reactions of propiolates.
RSC Adv.
This journal is ß The Royal Society of Chemistry 2013

View Article Online
RSC Advances
Paper
mixture is found experimentally with methyl propiolate,
although the methyl acrylate leads to a larger experimental
ratio of para over meta (3 : 1). But the transition state leading
to the para isomer is computed to be lower for the acrylonitrile
2
1
and acrolein dienophiles by 1.52 and 2.20 kcal mol
,
respectively, which would be expected to yield product ratios
of y6 : 1 and 13 : 1 of the para product over meta product.
The trend in the computed ratios are in reasonable agreement
with the experimentally observed ratios of 1.7 : 1 and 4.3 : 1
favoring the para isomer, respectively.
Scheme 6 Representation of the different stereoisomers and rotamers for the
In explaining the regioselectivity of these reactions, it
should be noted that inspection of the transition states shows
no evidence of steric effects being important in directing the
regiochemistry. Indeed, usually the regioselectivity of Diels–
Alder reactions can be predicted using frontier molecular
orbital theory by aligning the largest coefficient of the diene
methyl acrylate Diels–Alder intermediate.
states for each isomer were then computed at the B3LYP/6-31 +
G(d,p) level of theory.
1
5
HOMO with the largest coefficient of the dienophile LUMO.
In more rigorous terms, the orbital interaction energy for the
2
two transition states can be defined as 2(c c H + c c H ) /
Computational results
1 2 ij
3 4 ij
(
E
z
HOMO 2 ELUMO) where the c terms represent the 2p orbital
The transition states are depicted in Scheme 7. Note that for
all of the alkene dienophiles, the lowest energy transition state
yields the stereoisomer with the dienophile substituent
pointing towards the diene carbons and away from the lactone
coefficients at the bond-forming carbons, Hij represents the
resonance integral between these orbitals at the transition
state, and EHOMO and ELUMO represent the energy of the
HOMO and LUMO orbitals, respectively. Thus, this interaction
is strongest in transition states where there is a small
separation of the HOMO and LUMO and the coefficients on
the bonding carbons are large. Usually, only the closest
HOMO–LUMO interaction is considered (usually HOMO on
the diene and LUMO on the dienophile) while the other
interaction (HOMO on dienophile and LUMO on diene) with
the larger energy separation is ignored due to its smaller
energetic contribution.
(i.e. top structures in Scheme 6).
Computations suggest nearly degenerate transition states
leading to meta and para bicyclic adducts for the methyl
acrylate and acetylenic ester, which would be expected to lead
to nearly equal product mixtures (Table 1). This equal product
However, this simple Frontier Molecular Orbital (FMO)
approach for the reactions of methyl coumalate with acryloni-
trile, acrolein, and the methyl acrylate, predicts the meta
isomer should be favored rather than the para isomer that is
favored both computationally and by experiment. As expected
based on resonance arguments, the largest HOMO coefficient
on the diene p orbitals is on the 1 and 3 carbons, while the
z
largest pz LUMO coefficient on the dienophile is on the 6
carbon (see Scheme 8 for numbering). The simplistic picture
would suggest that the product would thus favor the bicyclic
structure leading to the meta regioisomer. Sometimes,
secondary orbital interactions are needed to account for the
1
6
Diels–Alder regioselectivity. Usually, the secondary interac-
tions considered are between the p orbitals of atoms 8 and 2
z
for formation of para products and 8 and 3 for formation of
meta products; however, adding these terms to the orbital
interaction expression above actually leads to a larger
predicted preference for the meta isomer over the para isomer.
Additionally, including the additional HOMOdienophile
–
LUMOdiene interaction terms into the above interaction energy
expression leads to a prediction that favors the meta isomer
over the para isomer.
Thus, some other electronic factor must lead to the
regiochemistry observed. We attribute the unexpected regios-
Scheme 7 Lowest-energy transition states found for formation of the bicyclic
products leading to the para-substituted benzenes (left) and meta-substituted
benzene products (right). The two newly-forming bond distances are indicated.
This journal is ß The Royal Society of Chemistry 2013
RSC Adv.

View Article Online
Paper
RSC Advances
Table 1 Computed transition state energies, differences in transition state energies between meta and para, and the computed ratio of para to meta for the reaction
of methyl coumalate with different dienophiles
{
21 a
b
{
21
{
21
Dienophile
DDH (kcal mol
)
Computed para–meta ratio
DH para (kcal mol
)
DH meta (kcal mol
)
2
0.07
0.9
28.10
29.09
25.50
29.53
28.03
30.60
27.70
29.74
1
2
0
.52
.20
.21
5.8
12.9
1.3
a
{
{
{
b
para–meta ratio = eˆ((DH{meta 2 DH para)/RT).
{
DDH = DH meta 2 DH para
.
electivity of the reactions to a different secondary orbital
interaction between the pyrone oxygen contribution to the
HOMO and the LUMO on the dienophile that favors the
transition state leading to the para product. This secondary
orbital interaction may exert a strong enough influence to
invert the natural electronic preference in these reactions to
form the meta product to favor the para product instead. This
interaction should favor the para isomer over the meta isomer
because the LUMO on the dienophile substituted with an
electron-withdrawing group has a larger coefficient on C6 than
on C7 and so this bonding interaction will be stronger in the
para transition state than in the meta transition state.
Several pieces of evidence favor the existence of this
secondary orbital interaction. First, inspections of the transi-
tion state geometries show that in all cases the bond length
between C4 and C6 for the para adducts is much shorter than
the bond length between C4 and C7 for the meta transition
state. This much shorter bond length is suggestive of a
stronger orbital interaction that could be attributable to the
stronger secondary overlap in the para transition state.
Second, viewing the Kohn–Sham HOMO orbitals of the
transition states shows overlap between the oxygen p
z
and
the dienophile carbon p orbital (see ESI ). Finally, it should be
z
3
noted that the difference in the coefficients in the LUMO on
C6 and C7 is much larger for acrolein than for acrylonitrile
and methyl acrylate, which is consistent with this dienophile
leading to the largest para selectivity.
The reaction of methyl coumalate with a variety of
dienophiles has been studied. The relative success using
acrolein suggests that electronic factors are more important
than steric factors in controlling the regioselectivity in favor of
the para-substituted product.
Experimental section
All starting materials were commercially available unless
otherwise noted. Methyl coumalate was prepared via methyla-
1
7
tion of coumalic acid. Due to the inherent toxicity of
acrolein, the reaction and workup were performed in a well-
ventilated fume hood. All products were either commercially
available or known in the literature. Product ratios were
1
determined by H NMR integration (300 MHz) of purified
mixtures of the isomers and comparison to known spectra.
Methyl coumalate Diels–Alder general procedure (alkenyl
dienophiles)
To a solution of methyl coumalate (0.078 g, 0.5 mmol) and
10% Pd/C (0.020 g, 25 wt%) dissolved in toluene in a sealable
tube was added alkenyl dienophile (1.5 mmol, 3 equiv.) at rt.
The mixture was sealed and stirred for 16 h at the temperature
described. The reaction was cooled to rt, opened, filtered
through Celite, washing with ethyl acetate, and concentrated
in vacuo to give the crude product which was purified via flash
Scheme 8 p
z
HOMO orbital coefficients (absolute values, sum of all p
z
orbital
coefficients in the basis set) for methyl coumalate and the p
z
LUMO coefficients
for electron-withdrawing group substituted dienophiles.
RSC Adv.
This journal is ß The Royal Society of Chemistry 2013

View Article Online
RSC Advances
Paper
column chromatography (hexanes : ethyl acetate) to give the
desired compounds as inseparable mixtures of meta and para
isomers.
(c) K. Tamao, J. Yoshida, M. Akita, Y. Sugihara, T. Iwahara
and M. Kumada, Bull. Chem. Soc. Jpn., 1982, 55, 255–260;
(d) K. Harano, T. Aoki, M. Eto and T. Hisano, Chem. Pharm.
1
Bull., 1990, 38, 1182–1191.
H NMR data was consistent with literature reported values
1
8
19
4 A. D. Carbaugh, W. Vosburg, T. J. Scherer, C. E. Castillo, M.
A. Christianson, J. Kostarellas, S. J. Gosai and M.
S. Leonard, ARKIVOC, 2007, 12, 43–54.
for methyl 3-cyanobenzoate, methyl 4-cyanobenzoate, ethyl
4
4
methyl isophthalate, ethyl methyl terephthalate, methyl
-formylbenzoate and methyl 4-formyl benzoate.
2
0
21
3
5
G. A. Kraus, S. Riley and T. Cordes, Green Chem., 2011, 13,
Methyl coumalate Diels–Alder general procedure (alkynyl
dienophiles)
2734–2736.
6 P. M. Delaney, D. L. Browne, H. Adams, A. Plant and J.
P. Harrity, Tetrahedron, 2008, 64, 866–873.
Y. Matsushita, K. Sakamoto, T. Murakami and T. Matsui,
Synth. Commun., 1994, 24, 3307–3313.
B. Raecke, Ger. 936,036.
Y. Ogata, M. Tsuchida and A. Muramoto, J. Am. Chem. Soc.,
To a solution of methyl coumalate (0.078 g, 0.5 mmol)
dissolved in toluene in a sealable tube was added alkynyl
dienophile (1.5 mmol, 3 equiv.) at rt. The mixture was sealed
and stirred for 16 h at the temperature described. The reaction
was cooled to rt, opened, filtered through Celite, washing with
ethyl acetate, and concentrated in vacuo to give the crude
product which was purified via flash column chromatography
7
8
9
1
957, 79, 6005–6008.
0 C. E. Stoner Jr. and T. B. Brill, Inorg. Chem., 1989, 28,
500–4506.
1
1
1
4
(hexanes:ethyl acetate) to give the desired compounds as
1 B. R. Travis, M. Sivakumar, G. O. Hollist and B. Borhan,
Org. Lett., 2003, 5, 1031–1034.
2 M. J. T. Frisch, et al. Gaussian09, version A02, Pittsburgh,
2009.
13 (a) A. D. Becke, J. Chem. Phys., 1993, 98, 5648; (b) A.
D. Becke, Phys. Rev. A: At., Mol., Opt. Phys., 1988, 38, 3098;
inseparable mixtures of meta and para isomers.
1
H NMR data was consistent with literature reported values
2
2
23
for mono-methyl isophthalate, mono-methyl terephthalate,
dimethyl isophthalate and dimethyl terephthalate.
2
4
25
Mono-methyl isophthalate
(c) C. Lee, W. Yang and R. G. Parr, Phys. Rev. B, 1988, 37,
To a solution of acrylic acid (0.206 mL, 3 mmol), distilled to
remove polymerized material, in toluene (5 mL, 0.2 M) was
added N,N-diisopropylethylamine (0.174 mL, 1 mmol) drop-
wise at rt. The mixture was stirred for 30 min after which
methyl coumalate (0.152 g, 1 mmol) was added, followed by
785.
14 (a) S. E. Wheeler, A. J. McNeil, P. Muller, T. M. Swager and
K. N. Houk, J. Am. Chem. Soc., 2010, 132, 3304–3311; (b) R.
S. Paton, J. L. Mackey, W. H. Kim, J. H. Lee, S.
J. Danishefsky and K. N. Houk, J. Am. Chem. Soc., 2010,
1
32, 9335–9340.
5 (a) K. N. Houk, J. Am. Chem. Soc., 1973, 95(12), 4092–4094;
b) K. N. Houk, J. Sims, C. R. Watts and L. J. Luskus, J. Am.
10% Pd/C (0.038 g, 25%/mass). The reaction was heated to 140
1
1
uC for 16 h, then cooled and quenched with sat. aq. NH Cl
4
(
solution (10 mL). The aqueous layer was extracted with EtOAc
and the combined organic extracts were washed with brine
and dried over MgSO . Filtration and concentration in vacuo
4
Chem. Soc., 1973, 95(22), 7301–7315.
6 (a) R. K. Dieter, W. H. Balke and J. R. Fishpaugh,
Tetrahedron, 1988, 44(7), 1915–1925; (b) P. V. Alston, R.
M. Ottenbrite and T. Cohen, J. Org. Chem., 1978, 43(10),
gave the crude product which was purified via flash column
chromatography (5 : 1–3 : 1 hexanes : EtOAc) to give mono-
1
1
864–1867; (c) Y. Apeloig and E. Matzner, J. Am. Chem. Soc.,
995, 117, 5375–5376.
1
methyl isophthalate in 45% yield as a light yellow solid. The H
NMR spectrum was identical to published reports for mono-
1
1
1
7 I. W. Ashworth, M. C. Bowden, B. Dembofsky, D. Levin,
W. Moss, E. Robinson, N. Szczur and J. Virica, Org. Process
Res. Dev., 2003, 7(1), 74–81.
8 J. R. Martinelli, D. A. Watson, D. M. M. Freckmann, T.
E. Barder and S. L. Buchwald, J. Org. Chem., 2008, 73(18),
2
2
methyl isophthalate.
Acknowledgements
7
102–7107.
9 O. Grossman and D. Gelman, Org. Lett., 2006, 8(6),
189–1191.
GK and GP thank the NSF Engineering Center for
Biorenewable Chemicals supported by NSF grant EEC-
1
0813570 for support for this research.
20 C. Lee, R. J. Perner, D. P. Larson, J. R. Koenig, A.
R. Gomtsyan, G. Z. Zheng, S. Didomenico, A. Stewart and
E. K. Bayburt, US 20020173665 A1.
2
1 J. Illesinghe, E. M. Campi, W. R. Jackson and A.
J. Robinson, Aust. J. Chem., 2004, 57(6), 531–536.
References
1
B. Lepoittevin and P. Roger, in Handbook of Engineering and
Specialty Thermoplastics, ed. S. Thomas and P. M. Visakh,
John Wiley & Sons, Inc., Hoboken, USA, 2011, vol. 3, pp.
22 A. Correa and R. Martin, J. Am. Chem. Soc., 2009, 131(44),
15974–15975.
23 J. Takaya, S. Tadami, K. Ukai and N. Iwasawa, Org. Lett.,
2008, 10(13), 2697–2700.
97–126.
2
3
G. A. Kraus and S. Riley, Synthesis, 2012, 44, 3003–3005.
(a) K. Afarinkia, V. Vinader, T. D. Nelson and G. H. Posner,
Tetrahedron, 1992, 48, 9111–9171; (b) E. J. Corey and A.
P. Kozlkowski, Tetrahedron Lett., 1975, 16(28), 2389–2392;
24 M. Tobisu, K. Yamakawa, T. Shimasaki and N. Chatani,
Chem. Commun., 2011, 47(10), 2946–2948.
25 E. Chenot, D. Bernardi, A. Comel and G. Kirsch, Synth.
Commun., 2007, 37, 483–490.
This journal is ß The Royal Society of Chemistry 2013
RSC Adv.