Diels-Alder reaction of vinylene carbonate and 2,5-dimethylfuran: Kinetic vs. thermodynamic control

Article abstract of DOI:10.1039/b9nj00536f

The Diels - Alder reaction between 2,5-dimethylfuran and vinylene carbonate was studied, both from an experimental and a theoretical point of view. The system was shown to slowly reach a thermodynamic equilibrium, characterized by the almost exclusive formation of the exo isomer. We rationalized these results by a comparison with classical systems involving maleic anhydride, and highlighted the different reactivity of vinylene carbonate as a dienophile. Finally, a preparative scale synthesis of pure exo isomer 4, a potentially useful synthon, ensued from this work. The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2010.

Full text of DOI:10.1039/b9nj00536f

View Article Online / Journal Homepage / Table of Contents for this issue
PAPER
www.rsc.org/njc | New Journal of Chemistry
Diels–Alder reaction of vinylene carbonate and 2,5-dimethylfuran:
kinetic vs. thermodynamic controlw
a
a
a
b
´
Celine Taffin, Glenda Kreutler, Damien Bourgeois,* Eric Clot and
Christian Perigaud
a
´
Received (in Montpellier, France) 6th October 2009, Accepted 13th November 2009
First published as an Advance Article on the web 19th January 2010
DOI: 10.1039/b9nj00536f
The Diels–Alder reaction between 2,5-dimethylfuran and vinylene carbonate was studied, both
from an experimental and a theoretical point of view. The system was shown to slowly reach a
thermodynamic equilibrium, characterized by the almost exclusive formation of the exo isomer.
We rationalized these results by a comparison with classical systems involving maleic anhydride,
and highlighted the different reactivity of vinylene carbonate as a dienophile. Finally,
a preparative scale synthesis of pure exo isomer 4, a potentially useful synthon, ensued from
this work.
the reaction there is still a kinetic preference for the endo
Introduction
adduct.6 Indeed, the reverse Diels–Alder reaction has a higher
rate than the forward reaction, as a consequence of the
stability of the starting materials. The aromaticity of the furan
heterocycle is lost during the cycloaddition process, rendering
the cycloaddition process less favourable compared to other
dienes such as butadiene or cyclopentadiene. The experimental
kinetic studies were enriched with computational results,
including DFT calculations, but the origin of the exo
selectivity is not fully understood.
The [4+2] cycloaddition between a 1,3-diene and an olefin,
known as the Diels–Alder reaction, is one of the most useful
reactions in organic chemistry.1 Its synthetic utility lies in the
availability of a large number of dienes and dienophiles which
can participate in the cycloaddition. Thus, many different
types of six-membered carbocyclic structures can be built up.
Up to four stereocenters can be formed in a single step, often
with high levels of regio- and stereocontrol. Understanding
and controlling the stereoselectivity in the installation of these
asymmetric centers constitutes a prominent goal in organic
chemistry.
Apart from the above case, where the low stability of the
[4+2] adduct overrules the kinetic preference for endo
stereoisomers, most of the work dedicated to the selective
preparation of exo stereoisomers focused on modifications of
the reaction partners to destabilize the endo transition state.7
Thus, judicious steric hindrance on dienes or dienophiles can
lead to high exo stereocontrol of the cycloaddition. The major
drawback lies in the numerous steps needed for the synthesis
of the suitable reaction partners.
The reactions involving furans as dienes are valuable
processes,2 yielding oxanorbornenes that are important inter-
mediates in the synthesis of numerous complex molecules.3
With this class of compounds, one of the most important
aspects of the Diels–Alder reaction is the understanding of the
factors influencing the exo/endo diastereoselectivity. Alder
noticed rapidly that the endo product is formed preferentially,4
and this remains generally observed in many cases. On the
other hand, the reaction of furan with maleic anhydride was
shown later to lead to the exclusive formation of the exo
adduct,5 contrary to Alder’s initial assumption.
During our synthetic studies towards nucleoside analogues,
we devised a route based on a Diels–Alder reaction with
vinylene carbonate as dienophile.8 Use of the latter is not very
common, although it leads to highly substituted cyclohexenes
or oxanorbornenes, which are precursors to the synthesis of
natural products and analogues, such as conduritols9 and
inositols.10 These syntheses are all based on the reaction with
furan. Perhaps further development was limited by the low
stereocontrol of the cycloaddition: the endo isomer is formed
with a modest selectivity (maximum 4.6/1). Also, separation of
the isomers is compulsory, either straight after the Diels–Alder
reaction, or later in the synthesis, and yields are moderate
(maximum 24%).
Extensive efforts have been made to understand the reasons
for this exception, and it appeared that the thermodynamic
control of the reaction rapidly takes place, favouring the
formation of the exo adduct, whereas at the first instants of
a
´
Institut des Biomolecules Max Mousseron (IBMM) UMR 5247
CNRS-UM 1-UM 2, Universite Montpellier 2, CC 1705,
´
`
Place Eugene Bataillon, 34095 Montpellier Cedex 5, France.
E-mail: damien.bourgeois@univ-montp2.fr;
Fax: +33 (0)4 67 04 20 29; Tel: +33 (0)4 67 14 38 79
b Institut Charles Gerhardt, CNRS 5253, Universite´ Montpellier 2,
`
CC 1501, Place Eugene Bataillon, 34095 Montpellier Cedex 5,
France. E-mail: eric.clot@univ-montp2.fr;
Fax: +33 (0)4 67 14 48 39; Tel: +33 (0)4 67 14 46 11
w Electronic supplementary information (ESI) available: 1H and 13C
NMR spectra for all new compounds, Cartesian coordinates of
optimised structures. See DOI: 10.1039/b9nj00536f
To reach the desired substitution pattern, we reacted
2,5-dimethylfuran 1 with vinylene carbonate 2 (Scheme 1).
As described in the literature,11 under thermal activation,
two diastereomers are produced, the endo adduct 3 and the
exo adduct 4, in a 40/60 ratio. The selectivity is moderate,
and opposite to the reaction with unsubstituted furan. To
circumvent the selectivity issue, the reaction sequence we
ꢀc
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2010
New J. Chem., 2010, 34, 517–525 | 517

View Article Online
Dihydroxylation of the olefin occurs exclusively via the exo
face of the molecule, in both cases, leading to diols 6 and 7.
Then deprotection of the carbonate moiety with potassium
carbonate in methanol leads to tetraols 8 and 9. The exo–exo
tetraol 9 presents two symmetry planes, and shows only 3
peaks in its 13C NMR spectrum, whereas the endo–exo
compound 8 shows 4 peaks.
Scheme 1 Diels–Alder reaction between 2,5-dimethylfuran (1) and
vinylene carbonate (2).
From the same crude reaction mixture between 1 and 2, we
also isolated bis-adduct 5 in 4% yield (Fig. 1). The formation
of such a compound was already mentioned in the first studies
of Diels–Alder reactions with vinylene carbonate.12 It is
isolated as a single diastereomer, but its structure was not
fully elucidated. A strong nuclear Overhauser effect (nOe)
between the protons located at the ring junctions implies that
they are all in an axial position. Thus, the bis-adduct derives
from the exo adduct 4, via an addition of furan 1 from its exo
side. This side is the least hindered on such systems (see also
the dihydroxylation above). The second Diels–Alder addition
can again come from an endo or an exo attack, referring to the
furan ring. Both are consistent with the spectroscopic data
obtained. The endo–exo isomer should, however, be preferred
over the exo–exo isomer, taking into consideration the closely
related works describing the bis-adducts of furan and dimethyl
acetylenedicarboxylate.13
developed is stereoconvergent, as both isomers are useful for
the synthesis. Thus, separation is not necessary, the yields are
much higher, and the methodology proved its efficiency in the
synthesis of new nucleoside analogues.
During our studies, we demonstrated that under certain
conditions, the reaction of 1 with 2 could also lead to the exo
adduct with very good yield and selectivity. Thus exo adduct 4
can be efficiently obtained on a preparative scale, and can now
be seen as a valuable synthon. The aim of the present article is
to describe a combined experimental and computational study
of the endo vs. exo selectivity of the reaction of 1 with 2.
Results
Experimental studies
These first results suggest a very moderate preference for
exo isomer, in full agreement with the selectivity previously
reported (60/40). But when we prolonged the reaction time to
increase conversion of starting materials, and thus yields of the
cycloadducts, we observed an increase in the exo/endo ratio.
After heating the same 2,5-dimethylfuran/vinylene carbonate
mixture for 48 h at 130 1C, this ratio turned to 3.8/1. And after
4 days, the selectivity was even higher than 8/1. This important
change enabled the easy isolation of pure exo material directly
after crystallisation from AcOEt. A 61% yield was obtained
on a multi-gram scale, thus making this route efficient for
selective preparation of 4 on a large scale.
Both compounds 1 and 2 are liquids and miscible, so no other
solvent was added, as this is usually the case for bi-molecular
Diels–Alder reactions. For economical reasons, as the
dienophile is far more expensive than the diene, we decided
to run the reaction with a 2/1 2,5-dimethylfuran/vinylene
carbonate ratio. This ratio was kept constant all through the
study. Not surprisingly, no reaction between both compounds
occurs at room temperature, and a minimum temperature of
90 1C is required. Even at this temperature, the reaction is very
slow. As the furan boils at 94 1C under atmospheric pressure,
we decided to conduct the reactions at higher temperature, in
sealed tubes, under autogene pressure.
These results prompted us to perform a detailed study of the
reaction. We clearly demonstrated that, at this temperature,
control of the reaction shifts from kinetics to thermodynamics.
Running the reaction at 130 1C, and analysing aliquots with
1H NMR at regular time intervals, enabled a quick study of
the exo/endo ratio of the isomers produced as a function of
time (Fig. 2).
In
a first step, we isolated and characterized both
diastereomers: complete data were never published. We
separated both isomers with flash chromatography on silica
gel, instead of fractional crystallization, which led in our hands
to mixtures of diastereomers. The important Rf difference
renders this operation easy (see experimental for details). After
24 h heating at 130 1C, adducts 3 and 4 are isolated in 25%
and 41% yield, respectively. Configuration of these adducts
was fully secured by the following procedure (Scheme 2).
The exo/endo ratio is close to 1 at the beginning of the
reaction, and it slowly and regularly evolves to reach a
selectivity greater that 10/1. Heating at 130 1C for 144 h gave
an exo/endo ratio of 13.2/1, corresponding to a dr of 86%.
After 216 h (9 days) the ratio stabilized slightly below 15/1,
Scheme 2 Chemical elucidation of relative configurations.
Fig. 1 Possible structures for bis-adduct 5.
ꢀc
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2010
518 | New J. Chem., 2010, 34, 517–525

View Article Online
Fig. 2 Time evolution of the exo/endo ratio of products during a
reaction at 130 1C.
there was no major evolution of the reaction medium, but at
this stage the bis-adduct 5 was present in an equivalent
amount to the endo adduct 3. Taking into account that the
former derives from exo adduct 4, the exo/endo ratio, taken as
1
(4 + 5)/3, becomes 16/1. Quantitative analysis by H NMR
was also less precise because the peaks broadened slightly. We
therefore did not carry on with heating of the reaction
medium.
We then performed a quantitative study of the proportions
of reagents and products during the reaction, at 110 1C and at
130 1C. We used 1,3,5-trimethoxybenzene (TMB) as an
internal standard, and analysed aliquots taken at regular time
intervals with 1H NMR. We were able to identify at least one
isolated signal for each product (i.e. 1, 2, 3, 4 and 5, plus TMB)
in the 1H NMR spectra for quantitative integration
(see Fig. 3). We used the signals at 5.83 ppm (1, dimethyl
furan), 7.13 ppm (2, vinylene carbonate), 4.64 ppm (3, endo
adduct), 4.51 ppm (4, exo adduct), 4.42 ppm (5, bis-adduct)
and 3.77 ppm (TMB).
During the course of the reaction, we could not see any
other methoxy signals that could have originated with side
products derived from TMB, the reference is thus stable. The
only other signals are multiplets at 0.8, 1.2 and 2.2 ppm,
corresponding to degradation products of the furan 1, which
may start to polymerise slowly. They are present only after
prolonged heating (more than 144 h), and, as stated before
during the exo/endo ratio study, analysis of the 1H NMR
spectra became more difficult because of peak broadening.
Concomitant with the formation of these impurities, the
reaction medium turned to black, and became slightly thicker.
Thus, the last point used for kinetic analysis is taken at 144 h
(6 days).
Fig. 3 1H NMR spectra of aliquots taken at different reaction time at
130 1C.
Quantitative analysis of reaction aliquots enabled plotting
of the conversion of 3 and 4 as a function of reaction time
(Fig. 4 and 5). Conversion of 3 and 4 corresponds to the
quantity of titrated products referred to the quantity of
starting dienophile 2 (the limiting reagent). Conversion of 2
is also plotted, and corresponds roughly to the sum of
appearing products (3, 4 and 5). This means that no other
by-product is formed, at least on an important scale.
Not surprisingly, the reaction performed at 110 1C gives
results corresponding to the reaction at 130 1C at shorter
Fig. 4 Conversion of exo and endo products at 130 1C.
reaction times. Initially, the reaction delivers a 1/1 mixture
of exo and endo adducts, even with a slight preference for
ꢀc
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2010
New J. Chem., 2010, 34, 517–525 | 519

View Article Online
Table 1 Conversion of products during retro Diels–Alder experi-
ments (Detect. means the product is detected, but at a too low level to
be quantified)
Reaction
T = 2 h
T = 40 h
3 - 1 + 2
SM (3)
1
2
SM (4)
1
2
4%
6%
4%
o1%
Detect.
Detect.
32%
29%
27%
o1%
7%
4 - 1 + 2
4%
reactants, transition states and products of the cyclo-
addition reactions described in the text were optimized at the
B3LYP/6-31G** level.15 This method has been previously
shown to be reliable on Diels–Alder reactions with furans.16
The computed structures are in very good agreement with
the previously reported X-ray structures17 for 3 and 4
(see Table 4). As Houk mentioned that in certain cases
systematic errors can occur,18 the energies of the various
systems were computed at the M05-2X/cc-pVTZ level.19 This
recently developed functional has been shown to yield better
energetics.20 The enthalpy and Gibbs free energy values given
in the text are the sum of the M05-2X/cc-pVTZ electronic
energy values and the B3LYP/6-31G** thermal corrections
at 400 K.
Fig. 5 Conversion of exo and endo products at 110 1C.
endo adduct (endo/exo ratio is 1.15 at the very beginning of the
reaction). This part ends after ca. 50% conversion of starting
materials. The second part of the reaction is characterised by a
regular disappearance of endo adduct 3, and continuation of
the formation of the exo adduct 4. After 6 days heating, there
is about 10% 2 remaining, and the presence of bis-adduct 5 in
equal amounts to endo adduct 3 suggests that secondary
reactions may become more important at this stage.
Possibly the system is close to an equilibrium point, as also
suggested by the slower evolution of the exo/endo ratio during
the last part of the reaction (see Fig. 2). To validate this
possibility, we showed that at this temperature, both adducts 3
and 4 can undergo retro Diels–Alder reaction.
Usually, differences in reactivity for cycloaddition reactions
are rationalized in the qualitative framework of FMO theory
by looking at the energies of the frontier orbitals, HOMO of
the diene and LUMO of the dienophile (Fig. 6). Furans are
electron rich dienes, forced into a reactive cisoid conformation,
with some degree of aromaticity conferred by the delocalisa-
tion of the oxygen atom lone pairs into the cycle, creating a 6-p
electron system. The reaction of furan (high HOMO) with
maleic anhydride (low LUMO) occurs at room temperature,
even in solution at moderate concentrations. The same
behaviour is observed with cyclopentadiene.
We encountered difficulties in quantifying precisely the retro
Diels–Alder reaction rates. As 1 and 2 are both volatile, a
precise titration is difficult, as the head space of the reaction
vessel may contain an important proportion of these reagents.
Also, at low conversion, as 3 and 4 are solid compounds, the
mixture is also solid at room temperature and representative
aliquots cannot be taken. Heating 3 or 4 neat (with 3% weight
TMB as internal standard) at 130 1C leads to the formation of
1 and 2 as reported in Table 1. The latter products are
analyzed after dilution of the whole reaction mixture into
CDCl3. In the case of 4, after 40 h, 1 and 2 are present in 7 and
4% quantity, respectively. We could not quantify the loss of 4.
In the case of 3, we evaluated the loss of 3 to 4% after 2 h,
and 32% after 40 h. These values correspond to the amount
of 1 and 2 detected in both cases. We conclude from these
results that the reverse reaction takes place during the
Diels–Alder reaction of 1 with 2, with a low rate for 4, and
a high rate for 3.
Substitution of furan by methyl groups increases the energy
of the HOMO, and 2,5-dimethylfuran 1 can be considered as a
‘good’ diene, slightly more electron rich than furan. However,
the substitution at the 2 and 5 positions has also increased the
steric bulk at the reactive sites.21 On the other hand, one can
expect a moderate reactivity of vinylene carbonate, a rather
electron rich dienophile as illustrated by the energy of the
LUMO (Fig. 6). The cornerstone of FMO theory is the
HOMO–LUMO gap between the two reacting molecules
and the values in Fig. 6 are consistent with the fact that furan
or dimethylfuran need to be heated neat with vinylene
carbonate at a minimum of 90 1C for a reaction to occur.
However, this qualitative approach cannot account for the
The gathered experimental data suggest a comparison with
the well known reaction of furan with maleic anhydride. There
is also a marked difference with the reaction between vinylene
carbonate 2 and unsubstituted furan, which delivers the endo
adduct. We decided to enrich this experimental study with
a computational study, to elucidate the similarities and
differences between these systems.
Computational studies
To shed more light on the experimental results, DFT calcula-
tions were carried out with the Gaussian03 package.14 The
Fig. 6 Energy levels of frontier orbitals of selected dienes and
dienophiles, obtained from B3LYP/6-31G** calculations.
ꢀc
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2010
520 | New J. Chem., 2010, 34, 517–525

View Article Online
Discussion
In order to model the observed kinetic profiles of the reaction,
we initially took into consideration only the formation of
monoadducts 3 and 4, and assumed that the kinetics are first
order (Scheme 3). It is not possible to solve exactly the system
of equations resulting from this model because of the presence
of terms linking 3 and 4. We determined the formation
constants assuming that, at the beginning of the reaction,
the reverse reactions do not take place ([3] and [4] are
very low), and that the system is close to the initial state
(i.e. [1] = [1](t = 0) and [2] = [2](t = 0), to remove the correlation
between formation of 3 and 4). The values for the formation
constants at 110 and 130 1C are given in Table 3. These values
reflect the observed slight preference for the endo adduct at the
beginning of the reaction (exo/endo ratio = 0.9, ie endo/exo
ratio = 1.1).
Fig. 7 Gibbs free energy and enthalpy (in parenthesis) values
(kcal molÀ1, T = 400 K) for the TS and products of the exo and
endo cycloaddtion reactions relative to 1 and 2, obtained from
M05-2X/cc-pVTZ//B3LYP/6-31G** calculations.
Using Eyring’s equation, we determined the activation
Gibbs free energy at 403
(DG# = 35.4 kcal molÀ1) and the endo adducts (DG#
35.3 kcal molÀ1). The computed values (Table 2) are in
excellent agreement both for the absolute (within 2 kcal molÀ1
K (130 1C) for the exo
=
difference in the reported selectivities (endo vs. exo) of the
reaction between furan or dimethylfuran and vinylene carbonate.
For the reaction between 2,5-dimethylfuran (1) and vinylene
carbonate (2), the reaction profile featuring enthalpy and
Gibbs free energy values for formation of the endo and exo
cycloaddition products is shown in Fig. 7.
)
and relative values.
We then took into account that both reactions are reversible
at the temperature of the study. Using the values shown in
Fig. 9 for the various rate constants, we could fit the observed
data quite well both at the early and late stages of the reaction
with data generated from the kinetic model. The values are
also in fair agreement with the retro Diels–Alder experiments
(k/k = 8, 4; k/k = 0.35, 3). In the case of 3, the retro
The transition states were computed using as a guess
geometry the results of a computational study for the reaction
between furan and maleic anhydride.16b Analytical calcula-
tions of the vibrational frequencies confirmed the transition
state nature of the located extrema. The geometry of the TS
(Fig. 8) does not present any striking difference compared to
other transition states between cyclic dienes and cyclic dienophiles.
The symmetrical nature of the system is reflected in the
transition state, where C–C bond formation is synchronous.
There is also no major difference between the endo and the exo
transition states. The C–C distances between both partners are
215.9 and 216.0 pm in the endo and exo systems, respectively.
We also performed similar calculations on the reactions
between furan (10) and vinylene carbonate, and between
dimethylfuran and maleic anhydride (11). The Gibbs free
energy values (T = 400 K) are reported in Table 2. The most
relevant parameters are the relative Gibbs free energy values
between the exo and endo products (DDG) and between the
endo and exo TSs (DDGa).
À1
À1
Diels–Alder reaction rate estimated from the sole experimental
s
À1, in the
point (see Table 1) gives k
= 2.0 Â 10À6
À1
same order of magnitude. The modeled rate constant
corresponds to an activation energy of 34.8 kcal molÀ1 at
403 K, again in excellent agreement with the computed value
(within 2 kcal molÀ1).
With these values for forward and reverse reaction rate
constants, we can estimate the Gibbs free energy for the
reaction forming the endo and exo cycloaddition products
(Scheme 4).
The exo product is more stable than the endo one, in
agreement with selective formation of the former upon
prolongated heating. The exo product is even more stable
than the starting reactants, while this is not the case for
the endo adduct. The calculated values are in qualitative
good agreement with the experimental values. Both products
are computed to be less stable than the starting reactant
but the computed DG values for the products are close to that
of the reactants (Table 2). More importantly, the relative DDG
value is computed to be in excellent agreement with the
experimental value (2.45 kcal molÀ1, exp; 1.9 kcal molÀ1
,
calc.).
Both experimental and theoretical results indicate that the
selectivity originates from a thermodynamic preference for the
exo isomer. In order to understand the origin of this stability,
we examined in detail the structural differences between the
computed geometries for 3 and 4. Moreover, as the X-ray
structures for 3 and 4 are available,17 this allows further test of
the accuracy of the computational methodology. In both
Fig. 8 Structure of exo and endo transition states.
ꢀc
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2010
New J. Chem., 2010, 34, 517–525 | 521

View Article Online
Table 2 Gibbs free energies (kcal molÀ1, T = 400 K) for various Diels–Alder reactions obtained from M05-2X/cc-pVTZ// B3LYP/6-31G**
calculations
Reaction
Exo TS
Exo adduct
Endo TS
Endo adduct
DDGa
DDGexo/endo
exo/endo
1 + 2
10 + 2
1 + 11
37.5
39.3
27.0
2.7
4.1
7.7
37.2
38.0
27.9
4.6
5.7
7.1
+0.3
+1.3
À0.9
À1.9
À1.6
+0.6
Table 4 Some structural features of adducts 3 and 4 (X-ray values
taken from ref. 17, n.d. = not determined)
Scheme 3 Simplified kinetic model for the Diels–Alder reaction.
Endo product (3)
Exo product (4)
Calc
X-ray
Calc
X-ray
Table 3 Formation rate constants of the Diels–Alder adducts
(values are given in molÀ1 L sÀ1
d1/pm
d2/pm
d3/pm
d4/pm
a
b
g
152.5
157.3
282.5
276.1
106.81
97.91
152.2
156.9
n.d.
153.1
156.7
269.7
288.3
104.31
100.01
114.21
111.81
153.0
156.1
n.d.
)
110 1C
130 1C
n.d.
n.d.
kexo
0.9 Â 10À7
3.7 Â 10À7
107.11
97.61
114.11
n.d.
103.61
99.81
n.d.
kendo
1.0 Â 10À7
4.1 Â 10À7
115.61
111.21
d
110.8
of the endo isomer (a angle 2.5–3.51 larger). Probably, the
deformation necessary in the endo case to completely remove
the steric interaction is too important, and thus the d3 distance
(282 pm between an O and a C atom) remains small, compared
to the d4 distance in the exo case (288 pm between two O
atoms). Furthermore, Van der Waals radii22 are higher for a C
atom (170 pm) than for an O atom (150 pm) rendering the
steric interaction in the endo isomer (illustrated by d3) more
important than the corresponding interaction in the exo
isomer (illustrated by d4).
The lower stability of the endo isomer, and its tendency to
undergo faster retro Diels–Alder reaction, is reflected by a
structure closer to the starting reagents. The distance d1 is
shorter by 0.6 pm, and the distance d2 is longer by 0.6 pm
compared with the exo isomer. Also, the distance between the
two carbon atoms of vinylene carbonate (d5) is shortened by
1.3 ppm in the endo isomer. These differences are very small,
but they all follow the same trend (Table 5).
Fig. 9 Correlation between experimental results (130 1C) and data
computed using the kinetic model depicted in Scheme 3.
Scheme 4 Gibbs free energy (T = 403 K) for the reactions forming 3
(endo) and 4 (exo).
We examined the potential effects of the methyl
substituents, but we did not notice any particular interaction.
However, the calculations indicate that the energy difference
between the two isomers is 0.3 kcal molÀ1 smaller in the case of
the reaction with unsubstituted furan 10 (Table 2). As stated
before, the Diels–Alder reaction between 10 and 2 delivers
preferentially the endo isomer. The calculations indicate
that the kinetic preference for the endo product is larger than
with 2,5-dimethylfuran (DDG# = +0.3 kcal molÀ1, 1 + 2;
DDG# = +1.3 kcal molÀ1, 10 + 2). The calculated value for
DDG# corresponds to a kinetic ratio between the endo and the
adducts, the cyclic carbonate part of the molecules is almost
completely flat. There are only minor differences between 3
and 4, listed in Table 4.
From these data, slight deformations of the oxanorbornene
structure to avoid steric interactions with the carbonate
substituent are evidenced. This substituent is moved away
from the bridgehead atom in the case of the exo isomer
(b angle 21 larger), and from the sp2 carbon atoms in the case
ꢀc
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2010
522 | New J. Chem., 2010, 34, 517–525

View Article Online
Table
5 Comparison between some structural parameters for
Conclusions
reagents and products of Diels–Alder reaction of 1 with 2(top) or 11
(bottom). SM stands for starting materials and n.a. stands for not
applicable
We have fully studied the Diels–Alder reaction between 2,5-
dimethylfuran 1 and vinylene carbonate 2. We demonstrated
that the reaction can be run under thermodynamic control, to
form almost exclusively the exo isomer. DFT calculations
showed a very good agreement with the experimental results
and allowed a rationalization of the reactivity observed. Both
experimental and computational approaches are consistent
with a slow equilibration at 130 1C, leading to the progressive
Product
d1/pm
d2/pm
d5/pm
disappearance of the endo isomer, less stable by 2.4 kcal molÀ1
.
X = O
Y = CQO
SM
136.2
152.5
153.1
136.2
152.8
152.9
n.a. (N)
157.3
156.7
n.a. (N)
159.3
159.2
133.1
154.3
155.6
133.5
153.0
154.1
The present system differs from the classical reaction between
furan and maleic anhydride on two points: (i) the activation
energies are higher and the reactions are slower; (ii) the
cycloaddition products are more stable relative to separated
reactants, so that the equilibrium is shifted towards the exo
product. Finally, we showed that the exo isomer can be easily
obtained on a preparative scale after crystallization. Our
intention is to pursue our efforts towards the synthesis of
new nucleoside analogues, deriving from this useful synthon.
endo (3)
exo (4)
SM
endo
exo
X = CQO
Y = O
exo products of ca. 5 at 400 K, in excellent agreement with the
experimental observations (endo/exo ca. 6). However, despite
this kinetic bias, the calculations indicate a clear thermo-
dynamic preference for the exo isomer. Therefore, we tried
to push further the Diels–Alder reaction between 10 and 2, in
order to reach thermodynamic equilibrium.
Experimental
General procedures
The results were somehow disappointing. We could
reproduce the predominant formation of the endo isomer after
heating at 130 1C for 12 h. Further heating yielded a very
complex mixture, which did not contain the exo isomer as the
major compound. We also observed that the endo isomer was
Unless noted otherwise, all starting materials and reagents
were obtained from commercial suppliers and were used
without further purification. Tetrahydrofuran was distilled
from sodium benzophenone ketyl. Dichloromethane, triethyl-
amine, acetonitrile and pyridine were freshly distilled from
calcium hydride. All solvents used for routine isolation of
products and chromatography were reagent grade. Reaction
flasks were dried at 100 1C. Air and moisture sensitive reactions
were performed under argon atmosphere. Flash column
chromatography was performed using silica gel 60 (230–400 mesh,
Merck) with the indicated solvents. Thin layer chromatography
was performed using 0.25 mm silica gel plates (Merck).
1
still present. Furthermore, H NMR analysis was hampered
by the presence of numerous by products. We assume that
furan polymerises more easily than dimethylfuran, and that
this system is not adapted to prolonged heating. Also, the
autogene pressure is higher with furan, and we are not
equipped to run reactions at a higher temperature (pressure
reached more than 20 bars at 180 1C).
Finally, we studied as a comparison the reaction between 1
a much better dienophile than 2 (Table 2).
Melting points were determined in open capillary tubes on a
Buchi-545 and are uncorrected. UV spectra were recorded on
¨
and 11,
The computed results are in full agreement with the experi-
mental data reported by Dewar on this reaction.23 The
activation energy is much lower for both exo and endo
adducts, as expected, consistent with the fact that the reaction
runs at room temperature. The retro Diels–Alder reaction is
also very fast in this case as illustrated by the larger relative
energies of the cycloaddition products with respect to
separated reactants. The computed data are consistent with
the reaction rates determined by Dewar23 and us. For the exo
adduct, with 11, the reverse reaction is about 1.5 fold
faster, whereas with vinylene carbonate we found it is
almost ten fold slower. We propose that this lower stability
is explained by the loss of conjugation occurring when
the double bond of 11 reacts. Thus, the Diels–Alder adducts
an Uvikon 931 (Kontron). Infrared spectra were recorded on a
Perkin–Elmer Paragon 1000 FT-IR spectrometer. 1H NMR
and 13C NMR spectra were recorded at 300 K on a Bruker 300
Avance and DRX 400, on solutions in the indicated solvent.
Chemical shifts are expressed in parts per million (ppm, d)
downfield from tetramethylsilane and are referenced to the
residual solvent peak (1H NMR, CHCl3 7.26, DMSO-d6 2.50)
or the deuterated solvent peaks (13C NMR, CDCl3 77.16,
DMSO-d6 39.52). 1H-NMR data were reported in the order of
chemical shift, multiplicity (s, singlet; d, doublet; m, multiplet
and/or multiple resonance), number of protons, and coupling
constant in hertz (Hz).
FAB mass spectra were recorded in the positive-ion or
negative-ion mode on a JEOL SX 102. The matrix was a
mixture (50 : 50, v/v) of glycerol and thioglycerol (G/T).
Electrospray (ES) mass spectra and high resolution mass
spectra were obtained on a Waters Q-TOF.
of
1 with 11 show a structure significantly distorted
towards the starting materials (Table 5). The d2 distance is
lengthened by 2–2.5 pm, and the d5 distance shortened by
1.2–1.5 pm in the adducts with 11 compared to 3 and 4. This
striking difference explains why we were able to obtain a
very good conversion at a high temperature, and why we
encountered no stability issue at room temperature with the
endo adduct 3.
Synthesis
Exo 1,7-dimethyl-3,5,10-trioxa-tricyclo[5.2.1.02,6]dec-8-en-4-
one (4). A mixture of 2,5-dimethylfuran 1 (9.8 g, 102 mmol)
ꢀc
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2010
New J. Chem., 2010, 34, 517–525 | 523

View Article Online
and vinylene carbonate 2 (4.4 g, 51 mmol) was heated in a
sealed vessel for 72 h at 130 1C. Remaining 1 was distilled off
the reaction, then the Diels–Alder adduct was crystallized
from AcOEt. The solid was filtered on a sintered glass funnel,
then washed with cold AcOEt. Exo 4 was obtained as a white
solid (5.65 g, 61%). Mp = 114–116 1C; IR (neat) n/cmÀ1 2997,
1813, 1368, 1327, 1209, 1162, 1134, 1082, 956; 1H NMR
(300 MHz, CDCl3) d (ppm) 6.22 (s, 2H), 4.51 (s, 2H), 1.58
(s, 6H); 13C NMR (75 MHz, CDCl3) d (ppm) 155.1, 139.4,
87.2, 80.3, 13.8; MS (ES+) (m/z) 183 [M + H+], 200
[M+NH4+], 224 [M + CH3CN + H+], 365 [2M + H+];
Anal. calcd for C9H10O4 C 59.34, H 5.53, found C 59.30,
H 5.28.
Exo 8,9-dihydroxy-1,7-dimethyl-3,5,10-trioxa-tricyclo [5.2.1.02,6]-
decan-4-one (7). The same procedure described above for 6
starting from 4 yielded 7 as a white solid with 73% yield.
Mp = 240–242 1C (dec.); IR (neat) n/cmÀ1 3445, 3367, 2999,
2950, 2913, 1792, 1774, 1376, 1169, 1065, 833; 1H NMR
(300 MHz, DMSO-d6) d (ppm) 4.97 (d, J = 4.2, 2H), 477
(s, 2H), 3.63 (d, J = 4.2, 2H), 1.22 (s, 6H); 13C NMR
(75 MHz, DMSO-d6) d (ppm) 154.4, 87.0, 81.6, 70.6, 10.7;
MS (ES+) (m/z) 217 [M + H+], 258 [M + CH3CN + H+],
280 [M + CH3CN + Na+], 433 [2M + H+]; HRMS (ES+)
calcd for C9H13O6 (MH+) 217.0712, found 217.0711.
Endo 1,4-dimethyl-7-oxa-bicyclo[2.2.1]heptane-2,3,5,6-tetraol (8).
To a stirred solution of 6 (40 mg, 0,18 mmol) in methanol
(1 mL) was added K2CO3 (51 mg, 0.37 mmol). After stirring
for 3 h, the reaction mixture was filtered on cotton and
concentrated in vacuo. Purification by flash chromatography
on silica gel (eluting with methanol–CH2Cl2 5/95) yielded title
compound 8 (25 mg, 73%) as a white solid. Mp = 169–171 1C;
IR (neat) n/cmÀ1 3465, 3351, 2994, 2976, 2941, 2905, 1445,
Endo 1,7-dimethyl-3,5,10-trioxa-tricyclo[5.2.1.02,6]dec-8-en-
4-one (3) and bis-adduct (5). A solution of 2,5-dimethylfuran
1 (1.115 g, 11.6 mmol) and vinylene carbonate 2 (500 mg,
5.8 mmol) was heated for 24 h at 130 1C. The residue was
chromatographed on silica gel with petrol ether–ethyl acetate
(70 : 30 then 60 : 40), yielding successively 270 mg (25%) 3 endo
as a white solid, 438 mg (41%) 4 exo as a white solid
(total yield: 66%) and 64 mg 5 (4%) as a pale yellow solid.
3: Mp = 109–111 1C; IR (neat) n/cmÀ1 2981, 1812, 1798,
1785, 1452, 1386, 1360, 1203, 1181, 1142, 1090, 990; 1H NMR
(300 MHz, CDCl3) d (ppm) 6.30 (s, 2H), 4.63 (s, 2H), 1.66
(s, 6H); 13C NMR (75 MHz, CDCl3) d (ppm) 154.8, 137.8,
87.1, 81.6, 17.2; MS (ES+) (m/z) 183 [M + H+], 200 [M +
NH4+], 224 [M + CH3CN + H+], 365 [2M + H+]; HRMS
(ES+) calcd for C9H11O4 (MH+) 183.0657, found 183.0657.
5: Mp = 193–195 1C; IR (neat) n/cmÀ1 3078, 2981, 2941,
1807, 1584, 1536, 1454, 1384, 1371, 1346, 1249, 1166, 1114,
1078, 951; 1H NMR (300 MHz, CDCl3) d (ppm) 6.24 (s, 2H),
4.41 (s, 2H), 2.12 (s, 2H), 1.59 (s, 6H), 1.35 (s, 6H); 13C NMR
(75 MHz, CDCl3) d (ppm) 152.2, 136.1, 84.7, 82.2, 81.3, 54.8,
17.6, 10.1; MS (ES+) (m/z) 279 [M + H+], 296 [M + NH4+],
320 [M + CH3CN + H+], 557 [2M + H+]; HRMS (ES+)
calcd for C15H19O5 (MH+) 279.1232, found 279.1232.
1
1328, 1254, 1202, 1166, 1150, 1108, 1068, 937, 845, 840; H
NMR (300 MHz, DMSO-d6) d (ppm) 4.71–4.70 (m, 2H),
4.39–4.37 (m, 2H, 3.94–3.92 (m, 2H), 3.45–3.43 (m, 2H),
1.17 (s, 6H); 13C NMR (75 MHz, DMSO-d6) d (ppm) 86.5,
72.5, 69.7, 15.3; MS (ES+) (m/z) 191 [M + H+], 208
[M + NH4+], 213 [M + Na+], 232 [M + CH3CN + H+],
381 [2M + H+], 403 [2M + Na+]; HRMS (ES+) calcd for
C8H15O5 (MH+) 191.0919, found 191.0919.
Exo 1,4-dimethyl-7-oxa-bicyclo[2.2.1]heptane-2,3,5,6-tetraol
(9). The same procedure described above for 8 starting from
7 yielded 9 as a white solid with 72% yield. Mp = 209–211 1C;
IR (neat) n/cmÀ1 3465, 3351, 2994, 2976, 2941, 2905, 1445,
1
1328, 1254, 1202, 1166, 1150, 1108, 1068, 937, 845, 840; H
NMR (300 MHz, DMSO-d6) d (ppm) 4.53 (d, J = 4.7, 2H),
3.45 (d, J = 4.7, 2H), 1.12 (s, 6H); 13C NMR (75 MHz,
DMSO-d6) d (ppm) 87.1, 72.4, 11.4; MS (ES+) (m/z) 191
[M + H+], 213 [M + Na+], 403 [2M + Na+]; HRMS (ES+)
calcd for C8H15O5 (MH+) 191.0919, found 191.0921.
Endo
8,9-dihydroxy-1,7-dimethyl-3,5,10-trioxa-tricyclo
Procedure for kinetic Diels–Alder experiments
[5.2.1.02,6]decan-4-one (6). To a cold (0 1C) stirred solution
of 3 (250 mg, 1.37 mmol) in a mixture of acetone (3 mL) and
water (0.75 mL) was added NMO (260 mL, 1.51 mmol, 60% wt
in water), then 0.75 mL of a 0.052 M osmium tetroxide
solution in tert-butanol. After stirring for 2 h, 5 mL of a
Na2S2O5 saturated aqueous solution was added. The mixture
was stirred for 30 min, then the solution was extracted with
dichloromethane (6 Â 20 mL), and the combined organic
layers were dried over Na2SO4 and concentrated in vacuo.
Purification by flash chromatography on silica gel (eluting
A mixture of 2,5-dimethylfuran 1 (1.23 g, 13 mmol), vinylene
carbonate 2 (562 mg, 6.5 mmol) and 1,3,5-trimethoxybenzene
(67 mg, 0.4 mmol) was stirred in a screw cap tube for 15 min at
25 1C. An aliquot was taken and used as reference for the
initial composition of the reaction mixture. The sealed vessel
was then placed at the desired temperature, and heated.
Aliquots were taken according to the following procedure:
the reaction vessel was cooled to room temperature, and left at
this temperature for at least 15 min. 100 mL were taken with a
syringe and analysed with 1H NMR in CDCl3. The remaining
reaction mixture was closed again and heated.
with methanol–CH2Cl2 4/96) yielded title compound
6
(201 mg, 68%) as a white solid. Mp = 206–208 1C (dec.);
IR (neat) n/cmÀ1 3462, 3340, 3006, 2972, 2932, 1785, 1364,
1159, 1080, 1016, 910; 1H NMR (300 MHz, DMSO-d6) d
(ppm) 4.98–4.95 (m, 2H), 4.73 (s, 2H), 3.74–3.71 (m, 2H), 1.35
(s, 6H); 13C NMR (75 MHz, DMSO-d6) d (ppm) 154.4, 86.3,
80.9, 69.9, 14.8; MS (ES+) (m/z) 217 [M + H+], 239
[M + Na+], 258 [M + CH3CN + H+], 433 [2M + H+],
455 [2M + Na+]; Anal. calcd for C9H12O6 C 50.00, H 5.59,
found C 49.91, H 5.76.
Notes and references
1 W. Carruthers, Cycloaddition Reactions in Organic Synthesis,
Pergamon Press, Elmsford, New York, 1990.
2 (a) C. O. Kappe, S. S. Murphree and A. Padwa, Tetrahedron, 1997,
53, 14179–14233; (b) B. H. Lipshutz, Chem. Rev., 1986, 86,
795–819; (c) F. I. Zubkov, E. V. Nikitina and A. V. Varlamov,
Russ. Chem. Rev., 2005, 74, 639–669; (d) B. A. Keay and
ꢀc
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2010
524 | New J. Chem., 2010, 34, 517–525

View Article Online
I. R. Hunt, Aspects of the Intramolecular Diels–Alder Reaction of a
Furan Diene (IMDAF) Leading to the Formation of 1,4-Epoxy-
decalin Systems in Advances in Cycloaddition, ed. M. Harmata, JAI
Press Inc, 1999.
3 P. Vogel, J. Cossy, J. Plumet and O. Arjona, Tetrahedron, 1999, 55,
13521–13642.
4 O. Diels and K. Alder, Chem. Ber., 1929, 62, 554–562.
5 R. B. Woodward and H. Baer, J. Am. Chem. Soc., 1948, 70,
1161–1166.
6 (a) M. W. Lee and W. C. Herndon, J. Org. Chem., 1978, 43, 518;
(b) L. Rulisek, P. Sebek, Z. Havlas, R. Hrabal, P. Capek and
A. Svatos, J. Org. Chem., 2005, 70, 6295–6302; (c) Y. Wooi Goh,
B. R. Pool and J. M. White, J. Org. Chem., 2008, 73, 151–156.
7 (a) Y. Lam, P. H. Cheong, J. M. Blasco Mata, S. J. Stanway,
V. Gouverneur and K. N. Houk, J. Am. Chem. Soc., 2009, 131,
1947–1957; (b) T. S. Powers, W. Jiang, J. Su, W. D. Wulff,
B. E. Waltermire and A. L. Rheingold, J. Am. Chem. Soc., 1997,
119, 6438–6439; (c) W. R. Roush and B. B. Brown, J. Org. Chem.,
1992, 57, 3380–3387; (d) J. Qi and W. R. Roush, Org. Lett., 2006,
8, 2795–2798; (e) T. Yoon, S. J. Danishefsky and D. De Gala,
Angew. Chem., Int. Ed. Engl., 1994, 33, 853–855; (f) T. A. Cernak
and J. L. Gleason, J. Org. Chem., 2008, 73, 102–110; (g) B. Boren,
J. S. Hirschi, J. H. Reibenspies, M. D. Tallant, D. A. Singleton and
G. A. Sulikowski, J. Org. Chem., 2003, 68, 8991–8995.
A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski,
P. Y. Ayala, K. Morokuma, G. A. Voth, P. Salvador,
J. J. Dannenberg, V. G. Zakrzewski, S. Dapprich, A. D. Daniels,
M. C. Strain, O. Farkas, D. K. Malick, A. D. Rabuck,
K. Raghavachari, J. B. Foresman, J. V. Ortiz, Q. Cui,
A. G. Baboul, S. Clifford, J. Cioslowski, B. B. Stefanov, G. Liu,
A. Liashenko, P. Piskorz, I. Komaromi, R. L. Martin, D. J. Fox,
T. Keith, M. A. Al-Laham, C. Y. Peng, A. Nanayakkara,
M. Challacombe, P. M. W. Gill, B. Johnson, W. Chen,
M. W. Wong, C. Gonzalez and J. A. Pople, GAUSSIAN 03
(Revision C.02), Gaussian, Inc., Wallingford CT, 2004.
15 (a) A. D. Becke, J. Chem. Phys., 1993, 98, 5648–52; (b) C. Lee,
W. Yang and R. G. Parr, Phys. Rev. B: Condens. Matter, 1988, 37,
785–89.
16 (a) G. A. Griffith, I. H. Hillier, A. C. Moralee, J. M. Percy, R. Roig
and M. A. Vincent, J. Am. Chem. Soc., 2006, 128, 13130–13141;
(b) S. Calvo-Losada and D. Suarez, J. Am. Chem. Soc., 2000, 122,
390–391; (c) B.S. Jursic, Tetrahedron Lett., 1997, 38, 1305–1308;
(d) E. Goldstein, B. Beno and K. N. Houk, J. Am. Chem. Soc.,
1996, 118, 6036–6043; (e) L. R. Domingo, M. J. Aurell, M. Arno
and J. A. Saez, J. Org. Chem., 2007, 72, 4220–4227;
(f) D. D. Claeys, K. Moonen, B. I. Roman, V. N. Nemykin,
V. V. Zhdankin, M. Waroquier, V. Van Speybroeck and
C. V. Stevens, J. Org. Chem., 2008, 73, 7921–7927;
(g) M. Dadwal, M. K. Kesharwani, V. Danayak, B. Ganguly,
S. M. Mobin, R. Muruganantham and I. N. N. Namboothiri,
Eur. J. Org. Chem., 2008, 6106–6118.
´
8 G. Sautrey, D. Bourgeois and C. Perigaud, Org. Biomol. Chem.,
2010, 8, 378–383.
9 A. Baran, C. Kazaz, H. Secen and Y. Sutbeyaz, Tetrahedron, 2003,
59, 3643–3648.
10 C. R. Kowarski and S. Sarel, J. Org. Chem., 1973, 38, 117–119.
11 H. J. Knops, K. Steinbeck, P. Babczinski, H. J. Santel and
R. R. Schmidt, German Patent Application, 3617715, 1987
(DE 3617715).
17 B. Bonnaud, A. J. Tenon, N. Ebby, C. N’Guessian, J. P. Aycard
and J. Lapasset, Acta Crystallogr., Sect. C: Cryst. Struct. Commun.,
1992, 48, 150–153. This reference gives no detail regarding the
conditions nor the diastereoselectivity of the reaction. Only the
X-ray structures are given, without any other spectroscopic data.
18 S. Pieniazek, F. R. Clemente and K. N. Houk, Angew. Chem., Int.
Ed., 2008, 47, 7746–7749.
12 M. S. Newman and R. W. Addor, J. Am. Chem. Soc., 1955, 77,
3789–3793.
13 J. D. Slee and E. LeGoff, J. Org. Chem., 1970, 35, 3897–3901.
14 M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria,
M. A. Robb, J. R. Cheeseman, J. A. Montgomery. Jr.,
T. Vreven, K. N. Kudin, J. C. Burant, J. M. Millam,
S. S. Iyengar, J. Tomasi, V. Barone, B. Mennucci, M. Cossi,
G. Scalmani, N. Rega, G. A. Petersson, H. Nakatsuji, M. Hada,
M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida,
T. Nakajima, Y. Honda, O. Kitao, H. Nakai, M. Klene, X. Li,
J. E. Knox, H. P. Hratchian, J. B. Cross, V. Bakken, C. Adamo,
J. Jaramillo, R. Gomperts, R. E. Stratmann, O. Yazyev,
19 For M05-2X, (a) Y. Zhao, N. E. Schultz and D. G. Truhlar,
J. Chem. Theory Comput., 2006, 2, 364–82for cc-pVTZ,
(b) T. H. Dunning, Jr., J. Chem. Phys., 1989, 90, 1007–23.
20 Y. Zhao and D. G. Truhlar, Acc. Chem. Res., 2008, 41, 157–167.
21 In the case of the reaction with fumaronitrile, 2,5-dimethylfuran
was found more reactive than furan itself: M. J. Cook and
S. J. Cracknell, Tetrahedron, 1994, 50, 12125–12132.
22 A. Bondi, J. Phys. Chem., 1964, 68, 441–451.
23 M. J. S. Dewar and A. B. Pierini, J. Am. Chem. Soc., 1984, 106,
203–208.
ꢀc
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2010
New J. Chem., 2010, 34, 517–525 | 525