On the mechanism of the thermal retrocycloaddition of pyrrolidinofullerenes (retro-prato reaction)

Article abstract of DOI:10.1002/chem.200800096

In contrast to N-methyl or N-unsubstituted pyrrolidinofullerenes, which efficiently undergo the retrocycloaddition reaction to quantitatively afford pristine fullerene, N-benzoyl derivatives do not give this reaction under the same experimental conditions. To unravel the mechanism of the retrocycloaddition process, trapping experiments of the in-situ thermally generated azomethine ylides, with an efficient dipolarophile were conducted. These experiments afforded the respective cycloadducts as an endolexo iso-meric mixture. Theoretical calculations carried out at the DFT level and by using the two-layered ONIOM (our own n-layered integrated molecular orbital and molecular mechanics) approach underpin the experimental findings and predict that the presence of the dienophile is not a basic requirement for the azomethine ylide to be able to leave the fullerene surface under thermal conditions. Once the 1,3-dipole is generated in the reaction medium, it is efficiently trapped by the dipolarophile (maleic anhydride or N-phenylmaleimide). However, for N-unsubstituted pyrrolidinofullerenes, the participation of the dipolarophile in assisting the 1,3-dipole to leave the fullerene surface throughout the whole reaction pathway is also a plausible mechanism that cannot be ruled out.

Full text of DOI:10.1002/chem.200800096

DOI: 10.1002/chem.200800096
On the Mechanism of the Thermal Retrocycloaddition of
Pyrrolidinofullerenes (Retro-Prato Reaction)
Salvatore Filippone,[a] Marta Izquierdo Barroso,[a] ngel Martín-Domenech,[a]
Sílvia Osuna,[b] Miquel Solà,*[b] and Nazario Martín*[a]
Abstract: In contrast to N-methyl or
N-unsubstituted pyrrolidinofullerenes,
which efficiently undergo the retrocy-
cloaddition reaction to quantitatively
afford pristine fullerene, N-benzoyl de-
rivatives do not give this reaction
under the same experimental condi-
tions. To unravel the mechanism of the
retrocycloaddition process, trapping ex-
periments of the in-situ thermally gen-
erated azomethine ylides, with an effi-
cient dipolarophile were conducted.
These experiments afforded the respec-
tive cycloadducts as an endo/exo iso-
meric mixture. Theoretical calculations
carried out at the DFT level and by
using the two-layered ONIOM (our
own n-layered integrated molecular or-
bital and molecular mechanics) ap-
proach underpin the experimental find-
ings and predict that the presence of
the dienophile is not a basic require-
ment for the azomethine ylide to be
able to leave the fullerene surface
under thermal conditions. Once the
1,3-dipole is generated in the reaction
medium, it is efficiently trapped by the
dipolarophile (maleic anhydride or N-
phenylmaleimide). However, for N-un-
substituted pyrrolidinofullerenes, the
participation of the dipolarophile in as-
sisting the 1,3-dipole to leave the fuller-
ene surface throughout the whole reac-
tion pathway is also a plausible mecha-
nism that cannot be ruled out.
Keywords: density functional calcu-
lations · fullerenes · nanostructures ·
reaction
mechanisms
·
retrocycloaddition
Introduction
dimensional all-carbon nanostructures in order to prepare
soluble and easy-to-handle materials in the search for new
properties for practical applications.[4] However, despite the
interest in new carbon allotropes, the chemistry of fullerenes
has not settled down and the synthesis of intriguing new
structures with unexpected properties is still possible on the
fullerene sphere.[5] In fact, the convex surface of fullerenes
formed by the highly reactive double bonds that protrude
from the spherical surface is a unique location for testing a
wide variety of challenging new reactions. Moreover, the re-
activity of fullerenes is an excellent benchmark for under-
standing the chemical behaviour of the above-mentioned
less-accessible and still scarcely known new nanoforms of
carbon.
During recent years, we have studied new reactions on
the fullerene sphere, by using 1,6-fullerenynes as a versatile
building block, such as the well-known [2+2+1] Pauson–
Khand reaction,[6] which despite its usefulness for the crea-
tion of carbocyclic rings has been almost neglected in mate-
rials science; the thermally induced [2+2] cyclization, which
forms unprecedented cyclobutene-fused carbocycles without
involving the presence of any catalyst[7] and the thermal in-
tramolecular ene reaction which leads to new and less-
known fullerenes decorated with the allene functionality.[8]
Since the discovery of fullerenes by Sir H. W. Kroto, R. F.
Curl, and the late R. E. Smalley in 1985,[1] and the closely
related carbon nanotubes by Iijima in 1991,[2] other new
forms of carbon allotropes, namely, endohedral and higher
fullerenes, nanohorns, nanoonions, and graphenes have been
reported.[3] Much effort is currently dedicated to under-
standing the chemical reactivity of these fascinating three-
[a] Dr. S. Filippone, M. I. Barroso, Dr. . Martín-Domenech,
Prof. Dr. N. Martín
Departamento de Química Orgµnica
Facultad de Ciencias Químicas, Universidad Complutense de Madrid
Ciudad Universitaria s/n 28040 Madrid (Spain)
Fax : (+34)913-944-103
E-mail: nazmar@quim.ucm.es
[b] S. Osuna, Prof. Dr. M. Solà
Institut de Química Computacional i Departament de Química
Universitat de Girona, Campus Montilivi
17071 Girona, Catalonia (Spain)
Fax : (+34)972-418-356
E-mail: miquel.sola@udg.edu
Supporting information for this article is available on the WWW
under http://www.chemeurj.org/ or from the author.
5198
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 5198 – 5206

FULL PAPER
Furthermore, in a preliminary communication we have re-
cently reported the highly efficient thermal retrocycloaddi-
tion reaction of the well-known pyrrolidinofullerenes (pyr-
rolidino[3,4:1,2,60]fullerenes).[9] These modified fullerenes
are among the most useful and studied fullerene derivatives
due to their easy synthetic preparation.[10] In contrast to
other fullerene derivatives, such as Diels–Alder cycload-
ducts[11] or Bingelꢁs methanofullerenes,[12] which undergo the
corresponding retrocycloaddition reactions, pyrrolidinoful-
lerenes have been considered to be very stable cycloadducts
and, therefore, this functionalization has been chosen as a
suitable procedure for the preparation of a wide variety of
fullerene derivatives of interest in biomedicine and in mate-
rials science.[13]
nism of this novel and versatile retrocycloaddition reaction,
we have carried out new reactions on pyrrolidinofullerenes,
in which the methyl (alkyl) group on the nitrogen atom has
been substituted by a hydrogen atom (free pyrrolidine) or
by a benzoyl group to determine the scope of the reaction.
To this end, we have also carried out trapping experiments
of the expected azomethine ylide, generated in situ as an in-
termediate 1,3-dipole, by means of the excess of dipolaro-
phile (N-phenylmaleimide) present in the reaction medium
(Scheme 1).
In this paper, we report on new experimental and theoret-
ical studies carried out on the retro-Prato reaction to deter-
mine a plausible mechanism by which these modified ful-
lerenes afford pristine fullerenes by thermal treatment in
the presence of a large excess of an efficient dipolarophile.
Whereas trapping experiments were carried out to confirm
the presence of the azomethine ylide as an intermediate 1,3-
dipole in the retrocyclization process, theoretical calcula-
tions at the DFT level by using the two-layered ONIOM
(ONIOM=our own n-layered integrated molecular orbital
and molecular mechanics) approach were performed to un-
derpin the experimental findings.
Scheme 1. Thermal retrocyclization of pyrrolidinofullerenes 1a–d.
Thus, new pyrrolidinofullerenes (1a–d) were prepared by
following the standard protocol from glycine and the respec-
tive reagents by following Pratoꢁs procedure (see the Exper-
imental Section in the Supporting Information). Compound
1b was prepared from 1a by treatment with benzoyl chlo-
ride, which allowed the N-benzoylation of the pyrrolidine
nitrogen atom. The thermal retrocycloaddition experiments
carried out with compounds 1a–d are shown in Table 1.
Results and Discussion
Synthesis and trapping experiments: In a preliminary com-
munication we have shown that the retrocycloaddition of N-
methylpyrrolidinofullerenes
(N-methyl
pyrroli-
dino[3,4:1,2,60]fullerenes) occurs quantitatively by simply
heating the starting materials at reflux in o-dicholoroben-
zene (o-DCB) for 24 h in the presence of an excess
(30 equiv) of a good dipolarophile, such as maleic anhydride
or N-phenylphthalimide. This is a very general procedure
with a broad scope of applicability.[9] In fact, the initial
brown solution of the N-methylpyrrolidinofullerenes turns
the magenta color typical of pristine [60]fullerene solutions
after the retrocyclization process. Furthermore, the reaction
has also been shown to be highly efficient with N-methylpyr-
rolidino[3,4:1,2,70]fullerenes and, even more important, it
allows the transformation of the undesired bis- and tris-cy-
cloadducts—formed as byproducts in the synthesis of pyrro-
lidinofullerenes—into the respective monoadducts, thus im-
proving significantly the yields of the Prato monoadducts.
In a general sense, the formation of N-methylpyrrolidino-
fullerenes and their subsequent quantitative thermal retro-
cyclization results in a new protection–deprotection protocol
that has already been successfully used in the separation of
the two constitutional isomers (Ih and D5h) of trimetallic ni-
tride endohedral metallofullerene Sc3N@C80.[9] We have also
reported the efficient removal of pyrrolidine rings from ful-
lerenes by controlled-potential electrochemical oxidation, to
efficiently afford pristine fullerene.[14] To unravel the mecha-
Table 1. Experimental conditions used in the retrocycloaddition reaction.
Entry Compound Dipolarophile
(30 equiv)
Catalyst Solvent
t
Yield
[h] [%]
1
2
1a
1a
–
–
–
o-DCB 24
o-DCB 24
90.0
95.0
maleic
anhydride
maleic
anhydride
N-phenyl
maleidimide
–
[a]
[a]
3
4
1a
1a
CuTf2
CuTf2
o-DCB 24
o-DCB 24
99.0
99.0
5
6
1b
1b
–
–
o-DCB 24
o-DCB 24
0.0
0.0
maleic
anhydride
maleic
anhydride
maleic
anhydride
maleic
anhydride
maleic
[a]
[a]
[a]
7
1b
1c
1c
1d
1d
CuTf2
–
o-DCB 24
27.3
8
toluene 18 quant.
toluene 18 quant.
toluene 18 quant.
toluene 18 quant.
9
CuTf2
–
10
11
anhydride
maleic
CuTf2
anhydride
[a] Copper(II) triflate (1 equiv).
Chem. Eur. J. 2008, 14, 5198 – 5206
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
5199

N. Martín, M. Solà et al.
A first and significant observation to unravel the mecha-
nism of the retrocyloaddition process in pyrrolidinoful-
lerenes is that compound 1a can undergo the thermal retro-
cyclization in the absence of the dipolarophile to afford C60
in 90% yield (entry 1). The presence of the dipolarophile in-
creases the efficiency of the process (95%, entry 2), and the
addition of copper triflate as a catalyst results in the forma-
tion of [60]fullerene in a quantitative yield.
Interestingly, whereas compounds 1a,c,d efficiently under-
go the retrocyclization process to quantitatively form pris-
tine [60]fullerene (entries 1–4, 8–11), N-benzoyl derivative
1b did not to any extent afford pristine C60 (entries 5,6).
This behaviour is in contrast to that previously observed for
N-methylpyrrolidinofullerenes which quantitatively lead to
[60]fullerene,[9] and confirms the importance of the nitrogen
lone pair in the retrocycloaddition process. Compound 1b
affords C60 in the presence of a copper catalyst (Entry 7), al-
though in a very poor yield; this could be accounted for by
previous decomposition of the N-benzoyl derivative. Be-
cause the benzoylation of the pyrrolidine ring is a reversible
process, we can take advantage of this discovery and use it
to selectively preserve a benzoylated pyrrolidine in the pres-
ence of other unsubstituted or N-methyl substituted pyrroli-
dine rings under the conditions used for the retrocycloaddi-
tion process.
The presence of an aryl substituent on C-5of the pyrroli-
dine ring with an ester group on C-2 results in an easier ret-
rocyclization process that now takes place in refluxing tolu-
ene over a shorter period of time (entries 8–11). These re-
sults clearly indicate that the simultaneous presence of sub-
stituents, such as ester and aryl groups, which stabilize the
in-situ-generated azomethine ylide, favor the retrocycloaddi-
tion process, which now occurs easily under the standard ex-
perimental conditions and, therefore, no additional catalyst
is needed.
To prove that the azomethine ylide is formed as an inter-
mediate in the retrocycloaddition process, we carried out
trapping experiments by using N-phenylmaleimide (NPM)—
instead of the less stable maleic anhydride—as the dipolaro-
phile, under the same experimental conditions. Thus, com-
pound 2 was refluxed in o-DCB in the presence of an excess
of NPM (10 equivalents). In addition to C60 (70%), the cy-
cloadduct (3), which resulted from 1,3-dipolar cyloaddition
of the in situ generated azomethine ylide to NPM, was ob-
by spectroscopic techniques and confirmed by an alternative
synthesis from the in-situ-generated azomethine ylide and
NPM as previously reported in the literature.[15]
The above trapping experiment clearly confirms that the
reaction mechanism of the retrocycloaddition process occurs
by thermal removal of the azomethine ylide as a 1,3-dipole,
which is generated in situ under the experimental conditions
used and the stability of which is highly dependent on the
nature of the substituents. Whether the dipolarophile
(NPM) is involved in the mechanism by assisting the 1,3-
dipole to leave in a concerted way or whether it acts in a
further step is an open question that we have addressed by
means of theoretical calculations (see below).
Theoretical calculations: As discussed above, under thermal
treatment and in the presence of an excess of a dipolaro-
phile, such as maleic anhydride or NPM, pyrroli-
dino[3,4:1,2,60]fullerenes can efficiently revert to [60]ful-
lerenes and azomethine ylides. This reaction is favored by
the presence of substituents that stabilize the leaving 1,3-
dipole. The effect of the substituent and the dipolarophile
may be thermodynamic in the sense that they help to stabi-
lize the final separated products and/or kinetic if the energy
barrier that needs to be surmounted is reduced. To analyze
whether the effects of the substituent and the dipolarophile
in the retro-Prato reaction are thermodynamic and/or kinet-
ic in nature, we have carried out a theoretical analysis of the
reaction mechanisms depicted in Schemes 3 and 4 for com-
pounds 4a (R1 =H) and 4b (R1 =COOCH3). We have locat-
ed every intermediate (int) and transition state (TS) in-
volved in the four reactions studied (optimized xyz coordi-
nates of all species are given as Supporting Information).
Retrocycloaddition without a dipolarophile: Our analysis
starts with retrocycloaddition in the absence of any dipolar-
ophile for the unsubstituted case. Theoretical studies of the
1,3-dipolar cycloadditions of different 1,3-dipoles (alkyl
azides,[16,17] ozone,[18] and 3-phenylphthalazinium-1-olate[19]
)
to the [6,5]- and [6,6]-bonds of C60 have been made by sever-
al authors. In all cases, the cycloadditions that take place on
[6,6]-bonds have been found to be clearly exothermic giving
rise to stable adducts on crossing a relatively small energy
barrier, whereas those that occur on the [6,5]-bonds are less
favorable. In agreement with these previous results, we
found that the retro-1,3-dipolar cycloaddition reaction start-
ing from the [6,6]-adduct 4a was endothermic by 53.1 kcal
molꢀ1 and had a high energy barrier of 45.4 kcalmolꢀ1. The
Gibbs free energy of the reaction and energy barrier are
somewhat smaller, 33.9 and 39.2 kcalmolꢀ1 as expected from
the fact that there is an increase in the entropy along the re-
action coordinate. The structure of the TS is given in
Figure 1.[20] The dark grey atoms in Figure 1 constitute the
small system treated at a high level of theory within the
ONIOM approach for all species analyzed in this work. As
can be seen in Figure 1, the reaction occurs through a con-
certed and almost synchronic TS with a distance of about
tained in 19% yield as
a 66:33 endo/exo mixture
(Scheme 2). The structure of compound 3 was determined
ꢀ
Scheme 2. Trapping experiment by using NPM as the dipolarophile.
2.71 Š between the two C atoms of the C C bonds that are
5200
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 5198 – 5206

Mechanism of the Retro-Prato Reaction
FULL PAPER
slightly shifting the geometry of
the transition state whilst fol-
lowing the transition vector in
the direction of the products.
Unfortunately, all attempts to
find this minimum structure
failed and the optimization pro-
cess yielded, in all cases, the
original reactant 4a. We think
that this behavior indicates the
existence of a very shallow min-
imum very close to the TS. This
is in agreement with the fact
that the region around the TS is
extremely flat. Indeed, the
imaginary frequency corre-
sponding to the transition
vector is only 33.8i cmꢀ1, which
is also in accordance with the
fact that any optimization we
have carried out by starting
from any structure close to the
final products (for instance by
Scheme 3. Schematic representation of the different stationary points found along the pathway of the retrocy-
cloaddition reaction assisted by the dipolarophile only at the last stage of the reaction.
ꢀ
starting from C C bond distan-
ces of 4 Š) reverts to the initial
reactant. The height for the ac-
tivation barrier together with
the endothermic nature of the
process prevents the retrocy-
cloaddition from occurring, as
found experimentally.[9]
The second reaction mecha-
nism that we have analyzed is
the retrocycloaddition for the
substituted species 4b (R1 =
COOCH3, Scheme 3) to exam-
ine the effect of the substituent
in the retro-Prato cycloaddition
without a dipolarophile (4b!
C60 +6b). As can be seen from
the values in Figure 3, the sub-
stituent favors the retrocycload-
dition by reducing the energy
barrier by about 10 kcalmolꢀ1
and the reaction energy by
about 15kcalmol ꢀ1. The reduc-
tion in the Gibbs free energies
when going from an unsubsti-
tuted to a substituted reactant
Scheme 4. Schematic representation of the different stationary points found along the pathway of the retrocy-
cloaddition reaction assisted by maleic anhydride.
being broken. It is worth noting that in all cases we have an-
alyzed both exo and endo approaches and we report here
only the most stable TS.
Since the energy of the products (C60 +6a+5) is higher
than that of the TS (see the whole energy profile in
Figure 2), a minimum structure should be found between
these two stationary points. To find such a minimum, a full
optimization starting from the transition state was done by
is somewhat lower (about 2 kcalmolꢀ1). The origin of the
stabilization (10–15kcalmol ꢀ1 along the reaction coordi-
nate) has to be attributed to the electron-withdrawing char-
acter from both inductive and resonance effects of the
COOCH3 substituent[21] that stabilizes the azomethine ylide
formed. This effect is especially important at the end of the
reaction, and for this reason the reaction energy is lowered
more than the energy barrier. To illustrate this fact in more
Chem. Eur. J. 2008, 14, 5198 – 5206
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
5201

N. Martín, M. Solà et al.
Figure 1. Optimized structure (ONIOM2(UB3LYP/6-31G(d):SVWN/
STO-3G)) for TS1a with the most relevant bond lengths [Š] and angles
[8]. H atoms and dark-grey atoms constitute the small system treated at a
high level in the ONIOM approach.
Figure 3. ONIOM2(UB3LYP/6-31G(d):SVWN/STO-3G) reaction energy
profile (Gibbs free energies in square brackets) for the retrocycloaddition
reaction of the substituted system (R1 =COOCH3) without a dipolaro-
phile, except for the last step of the reaction.
the retro-Prato cycloaddition by stabilizing the negative
charge on the carbon atom of the azomethine ylide.
The energy profile for this process (4b!C60 +6b) is given
in Figure 3, whereas Figure 4 depicts the molecular structure
of the TS for this reaction. The imaginary frequency of this
TS is 244.5i cmꢀ1. Not surprisingly, the asymmetry of the TS
ꢀ
increases for the substituted species, with C C bond lengths
between the bonds being broken of 2.24 and 2.72 Š. As ex-
Figure 2. ONIOM2(UB3LYP/6-31G(d):SVWN/STO-3G) reaction energy
profile (Gibbs free energies in square brackets) for the retrocycloaddition
reaction of the unsubstituted system without a dipolarophile, except for
the last step of the reaction.
detail, we have computed the energy barriers for two more
substituents: R1 =OH and R1 =CN. For the former, the re-
action energy increases by 5.1 kcalmolꢀ1 relative to the un-
substituted case (53.1 kcalmolꢀ1). This is in agreement with
OH being an electron-donating substituent.[21] On the other
hand, for R1 =CN, another electron-withdrawing substitu-
ent,[21] the reaction energy decreases by 15.3 kcalmolꢀ1, not
far from the 14.4 kcalmolꢀ1 reduction observed for R1 =
COOCH3. Thus, there is a clear electronic effect of the pyr-
rolidine substituent: electron-withdrawing substituents favor
Figure 4. Optimized structure (ONIOM2(UB3LYP/6-31G(d):SVWN/
STO-3G)) for TS1b with the most relevant bond lengths [Š] and angles
[8]. H atoms and dark-grey atoms constitute the small system treated at a
high level in the ONIOM approach.
5202
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 5198 – 5206

Mechanism of the Retro-Prato Reaction
FULL PAPER
ꢀ
pected from the steric hindrance, the largest C C bond
length occurs between the substituted C atom and the corre-
sponding C atom of C60. In this case, the intermediate
formed by the interaction between the 1,3-dipole and the
C60 cage is more stable because the dipole moment of the
substituted azomethine ylide (4.660 D) is significantly larger
than that of the unsubstituted 1,3-dipole (1.660 D). Despite
the overall stabilization of the TS and the products, the re-
action 4b!C60 +6b is still quite endothermic and does not
take place as found experimentally.[9] The energy profile
clearly indicates that the equilibrium between reactants 4b
and products C60 +6b is totally displaced to the reactants
side and possible products formed after severe heating will
revert to reactants.
formed after heating they do not revert to reactants.
Figure 6 shows that the TS structure for the endo approach
of the retrocycloaddition assisted by maleic anhydride is
Retro-Prato cycloaddition assisted by maleic anhydride:
Scheme 4 depicts the reaction mechanism studied in this sec-
tion. Figure 5contains the relative energies and Gibbs free
Figure 6. Optimized structure (ONIOM2(UB3LYP/6-31G(d):SVWN/
STO-3G)) for TS3a with the most relevant bond lengths [Š] and angles
[8]. H atoms and dark-grey atoms constitute the small system treated at
high level in the ONIOM approach.
slightly asymmetric. The totally symmetric TS that we have
found is about 0.28 kcalmolꢀ1 higher in energy than the
slightly asymmetric TS.[20b] Moreover, the endo approach be-
tween the maleic anhydride and the ylide has been found to
have a lower energy barrier, with a TS that is about
2.5kcalmol ꢀ1 more stable than the exo TS. It is interesting
to see how maleic anhydride provides assistance in this TS.
Figure 5. ONIOM2(UB3LYP/6-31G(d):SVWN/STO-3G) reaction energy
profile (Gibbs free energies in square brackets) for the retrocycloaddition
reaction assisted by maleic anhydride of the unsubstituted system.
ꢀ
The bond lengths between the C C bonds connecting C60
and the azomethine ylide in the reactants are shorter than
ꢀ
the new C C bonds that are being formed between maleic
energies of the TSs and products of the retrocycloaddition
assisted by maleic anhydride. For the unsubstituted system
(R1 =H), the energy barrier decreases by about 5kcalmol ꢀ1
in the retrocycloaddition assisted by maleic anhydride. How-
ever, since entropy decreases when going from 4a+5 to
TS3a, the Gibbs free energy barrier at 298 K increases by
about the same quantity. Therefore, the presence of maleic
anhydride does not improve the efficiency of the retrocy-
cloaddition by reducing the barrier. Rather, the effect we
find in this case is almost purely thermodynamic: the reac-
tion now becomes exothermic by about 16 kcalmolꢀ1 be-
cause of the increased stability of the final products. The
energy profile depicted in Figure 5clearly indicates that the
equilibrium between reactants and products is now dis-
placed to the products side and once the products are
anhydride and the azomethine ylide and, therefore, this TS
has reactant-like character. The imaginary frequency of this
TS is 179.2i cmꢀ1.
There is still another possible mechanism that can be op-
erative for the retrocycloaddition assisted by maleic anhy-
dride. In this case, the reaction would proceed from 4a to
C60 +6a and after that 6a would react with maleic anhydride
(5) to give the final nitrile oxide product 7a (see Figure 2).
Thus, in this case, the assistance of maleic anhydride does
not occur along the whole reaction path, but only at the end
of the reaction, just to stabilize the azomethine ylide
formed. All attempts to find the TS of the reaction 6a+5!
7a have been unfruitful. Indeed, any optimization we have
carried out by starting from any structure close to the final
ꢀ
products (for instance by starting from C C bond lengths of
Chem. Eur. J. 2008, 14, 5198 – 5206
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
5203

N. Martín, M. Solà et al.
4 Š) reverts to the final product 7a. This indicates that the
6a+5!7a conversion is an almost barrierless process, and
that, in the presence of a dipolarophile, once 6a is formed it
is immediately transformed into 7a. The whole energy barri-
er for 4a+5!C60 +7a indicates that the process is assisted
by the dipolarophile only at the end of the reaction
(Figure 2) and is somewhat higher in energy than that of the
retrocycloaddition assisted by maleic anhydride throughout
the whole reaction (Figure 5). However, the free-energy
profile for the latter is entropically unfavorable and, as a
consequence, the free-energy profile at 298 K favors the
mechanism in which the assistance of maleic anhydride does
not occur along the whole reaction path, but only at the end
of the reaction. Given the accuracy and reliability of the
method employed for the calculation, we consider that it is
not possible to rule out either of these two mechanisms for
the retrocycloaddition assisted by maleic anhydride for the
unsubstituted system (R1 =H), and that both mechanisms
can be operative under intense heating of pyrroli-
dino[3,4:1,2,60]fullerenes in the presence of maleic anhy-
dride.
shown in Figure 8. This TS together with intermediates
int3b and int4b reside on a relatively flat plateau. The pres-
ence of these intermediates, which were not found for the
We have also analyzed the retrocycloaddition assisted by
maleic anhydride for the substituted system (R1 =
COOCH3). The energy profile with the relative energies and
Gibbs free energies obtained for the process assisted by
maleic anhydride along the reaction coordinate is given in
Figure 7. The first TS has a lower energy barrier than the
Figure 8. Optimized structure (ONIOM2(UB3LYP/6-31G(d):SVWN/
STO-3G)) for TS3b with the most relevant bond lengths [Š] and angles
[8]. H atoms ant dark-grey atoms constitute the small system treated at
high level in the ONIOM approach.
unsubstituted case, is favored by the higher dipole moment
of the substituted azomethine ylide. In these intermediates,
ꢀ
the C C bonds that are being broken are not completely
ꢀ
separated nor are the new C C bonds totally formed. We
have performed a linear transit of five steps from int3b to
ꢀ
int4b by increasing/decreasing the four C C bonds that are
being broken or formed. The process has been carried out
by using different uniform steps for each of the four coordi-
nates. The coordinate value Rilinear transit at each step is given
in Equation (1) as:
ðRint4bꢀRint3b
Þ
Rilinear transit ¼ Rint3b þ i
ð1Þ
6
in which, i=1 to 5and Rint3b is the coordinate value for
int3b and Rint4b is the coordinate value for int4b. The re-
maining geometrical parameters are then reoptimized. As a
result we have obtained a linear transit with a continuous in-
crease of energy from int3b to int4b as depicted in Figure 6.
Therefore, we have not obtained an energy maximum from
which to initiate the search for the TS that must exist be-
tween intermediates int3b and int4b. However, with this
linear transit we have proven that this TS is probably quite
close both in energy and structurally to int4b. The last TS
Figure 7. ONIOM2(UB3LYP/6-31G(d):SVWN/STO-3G) reaction energy
profile (Gibbs free energies in square brackets) for the retrocycloaddition
reaction assisted by maleic anhydride of the substituted system (R1 =
COOCH3).
equivalent TS for the unassisted reaction, although the
Gibbs free energy barriers are similar. The structure of this
TS, which has an imaginary frequency of 16.8i cmꢀ1, is
5204
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 5198 – 5206

Mechanism of the Retro-Prato Reaction
FULL PAPER
corresponds to the formation and separation of the final ni-
trile oxide and pristine fullerene (TS5b). As compared to
reactants, the energy of this TS is as high as 46.9 kcalmolꢀ1
(Gibbs free energy of 51.7 kcalmolꢀ1) and, consequently,
overall the mechanism of retrocycloaddition assisted by
maleic anhydride for the substituted system along the reac-
tion coordinate cannot compete with the mechanism of ret-
rocycloaddition assisted by maleic anhydride only at the end
of the reaction, that is, the reaction proceeds from 4b to
C60 +6b and after that 6b reacts with maleic anhydride (5)
to give the final product 7b. The latter process has an over-
all energy barrier of only 32.9 kcalmolꢀ1 (the free-energy
value is 28.8 kcalmolꢀ1) as seen in Figure 3.
Finally, we have computed the thermodynamics of the ret-
roreaction for the N-benzoyl pyrrolidinofullerenes with R=
COPh, R1 =COOMe and R2 =H (Scheme 1). The energy
needed to release the 1,3-dipole without the presence of di-
polarophile in this system is 40.36 kcalmolꢀ1, that is,
1.6 kcalmolꢀ1 higher than that found for 4b. This result indi-
cates that the benzoyl substituent destabilizes the azome-
thine ylide generated, thus impeding the reaction.
Experimental Section
Computational details: Full geometry optimizations have been carried
out with the two-layered ONIOM approach[22] by using the Gaussian 03
program.[23] The DFT Xa exchange (a=2/3) functional[24a] in conjunction
with the electron gas correlation functional (in Vosko–Wilk–Nusair para-
metrization)[24b], that is, the so-called SVWN method together with the
standard STO-3G basis set[25] was used for the low level calculations and
the hybrid density functional B3LYP method[26] with the standard 6-
31G(d) basis set[27] was employed for the high level system. All systems
were treated with the spin-restricted formalism. The choice of DFT
methods was based on previous studies, which showed that DFT (and, in
particular, the relatively inexpensive B3LYP method combined with the
6-31G(d) or similar basis set) provides a reliable description of reaction
mechanisms of pericyclic reactions.[16,28] All stationary points found have
been characterized as either minima or TSs by computing the vibrational
harmonic frequencies; TSs have a single imaginary frequency with the
corresponding eigenvector (transition vector) related to the approach of
the two reaction centers, whereas minima have all real frequencies.
Gibbs-free-energy differences at 298.15K and 1 atm. ( DG298) were calcu-
lated from electronic energies (DE) according to Equations 2 and 3, as-
suming an ideal gas:[29]
DG298 ¼ DH298ꢀTDS298
with T=298.15K and
ð2Þ
Conclusion
298
trans
298
rot
DH298 ¼ DE þ DE0vib þ DE
þ D E þ DðDE0vibÞ298 þ DðpVÞ
ð3Þ
298
trans
298
rot
in which, DE , D E , and DE0vib are the differences between products
In summary, we have carried out new reactions on pyrrolidi-
nofullerines with different substitution patterns on the C-2
and C-5atoms of the pyrrolidine ring, which have a strong
impact on the stability of the resulting azomethine ylides
generated in the retrocycloaddition process. The nature of
the substituent on the pyrrolidine nitrogen atom also has a
great impact on the retrocyclization process. Thus, whereas
and reactants in translational, rotational and zero-point vibrational
energy, respectively; D(DEv0ib
)
is the change in the vibrational energy
298
difference as one goes from 0 to 298.15K. The molar work term D(pV) is
(Dn)RT (e.g., for two fragments combining to give one molecule Dn=
ꢀ1).
ꢀ
methyl (or alkyl) and N H groups undergo the retrocy-
cloaddition quantitatively, the presence of a benzoyl group
prevents or hinders this process.
Acknowledgements
Financial help has been provided by the Spanish MEC (projects nos.
CTQ2005–08797-C02–01/BQU, YCTQ2005–02609/BQU), the Consolid-
er–Ingenio(2010C-07–25200), the CAM (project no. S-0505/PPQ/0225),
and by the DURSI (project no. 2005SGR-00238). S.O. and S.F. thank the
MEC for a research grant and for a Ramón y Cajal contract, respectively.
We also acknowledge the Centre of Supercomputació of Catalunya
(CESCA) for partial funding of computer time.
In agreement with the experimental findings, theoretical
calculations carried out at the DFT level and by using the
two-layered ONIOM approach predict that the presence of
the dienophile is not required for removal of the azome-
thine ylide from the fullerene surface which, after in-situ
generation under thermal conditions, is trapped by the dipo-
larophile (maleic anhydride or N-phenylmaleimide). How-
ever, for the N-unsubstituted pyrrolidinofullerenes, it is pos-
sible that the dipolarophile can assist the 1,3-dipole to leave
and, therefore, this alternative mechanism cannot be totally
ruled out.
The results now reported on the scope and trapping ex-
periments of the retrocycloaddition process of pyrrolidino-
fullerenes have allowed us to unravel the mechanism of this
reaction, which has already been successfully used as a new
and efficient protection–deprotection protocol in fullerene
science, paves the way for further application of this reac-
tion in the construction of new and fascinating fullerene-
based carbon nanostructures.
[1] a) H. W. Kroto, J. R. Heath, S. C. OꢁBrien, R. F. Curl, R. E. Smalley,
Nature 1985, 318, 162–163; Nobel Lectures: b) R. E. Smalley,
Angew. Chem. 1997, 109, 1666–1673; Angew. Chem. Int. Ed. Engl.
1997, 36, 1594–1601; c) H. W. Kroto, Angew. Chem. 1997, 109,
1648–1664; Angew. Chem. Int. Ed. Engl. 1997, 36, 1578–1593;
d) R. F. Curl, Angew. Chem. 1997, 109, 1636–1647; Angew. Chem.
Int. Ed. Engl. 1997, 36, 1566–1576.
[2] S. Iijima, Nature 1991, 354, 56–58.
[3] For a recent review, see: J. L. Delgado, M. A. Herranz, N. Martín, J.
Mater. Chem. 2008, 18, 1417.
[4] For some recent books: a) A. Hirsch, The Chemistry of Fullerenes,
Wiley-VCH, Weinheim (Germany), 2005; b) Fullerenes: From Syn-
thesis to Optoelectronic Properties (Eds.: D. M. Guldi, N. Martín),
Kluwer Academic Publishers, Dordrecht (The Netherlands), 2002;
c) R. Taylor, Lecture Notes on Fullerene Chemistry: A Handbook
for Chemists, Imperial College Press, London, 1999; d) S. Reich, C.
Thomsen, J. Maultzsch, Carbon Nanotubes: Basic Concepts and
Physical Properties, Wiley-VCH, Weinheim (Germany), 2004;
Chem. Eur. J. 2008, 14, 5198 – 5206
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
5205

N. Martín, M. Solà et al.
e) Fullerenes, (Eds.: F. Langa, J.-F. Nierengarten), RSC Publishing,
Cambridge (UK), 2007.
displacements did not fully converge. b) The lack of symmetry in
this TS is not totally unexpected, see: S. M. Bachrach, P. B. White, J.
Mol. Struct (Theochem) 2007, 819, 72–78.
[5] a) N. Martín, Chem. Commun. 2006, 2093–2104; b) N. Martín, M.
Altable, S. Filippone, A. Martín-Domenech, Synlett 2007, 20, 3077–
3095; c) special issue on Functionalised Fullerene Materials, J. Mater.
Chem. 2002, 12, 1931–2159; d) special issue on the Supramolecular
Chemistry of Fullerenes Tetrahedron 2006, 9, 1905–2132.
[6] a) N. Martín, M. Altable, S. Filippone, A. Martín-Domenech, Chem.
Commun. 2004, 1338–1339; b) N. Martín, M. Altable, S. Filippone,
A. Martín-Domenech, A. Poater, M. Solà, Chem. Eur. J. 2005, 11,
2716–2729.
[7] N. Martín, M. Altable, S. Filippone, A. Martín-Domenech, M. Güell,
M. Solà, Angew. Chem. 2006, 118, 1467–1470; Angew. Chem. Int.
Ed. 2006, 45, 1439–1442.
[8] a) M. Altable, S. Filippone, A. Martín-Domenech, M. Güel, M. Solà,
N. Martín, Org. Lett. 2006, 8, 5959–5962; b) M. Güell, M. Altable,
S. Filippone, A. Martín-Domenech, N. Martín, M. Solà, J. Phys.
Chem. A 2007, 111, 5253–5258.
[9] N. Martín, M. Altable, S. Filippone, A. Martín-Domenech, L. Eche-
goyen, C. M. Cardona, Angew. Chem. 2006, 118, 116–120; Angew.
Chem. Int. Ed. 2006, 45, 110–114.
[10] a) M. Maggini, G. Scorrano, M. Prato, J. Am. Chem. Soc. 1993, 115,
9798–9799; b) N. Tagmatarchis, M. Prato, Synlett 2003, 768–779.
[11] N. Martín, J. L. Segura, F. Wudl in Fullerenes: From Synthesis to Op-
toelectronic Properties (Eds.: D. M. Guldi, N. Martín), Kluwer Aca-
demic Publishers Dordrecht (The Netherlands), 2002, Chapter 3,
pp. 81–120.
[21] C. Hansch, A. Leo, R. W. Taft, Chem. Rev. 1991, 91, 165–195.
[22] a) M. Svensson, S. Humbel, R. D. J. Froese, T. Matsubara, S. Sieber,
K. Morokuma, J. Phys. Chem. 1996, 100, 19357–19363; b) S. Dap-
prich, I. Komµromi, K. S. Byu, K. Morokuma, M. J. Frisch, J. Mol.
Struct. (Theochem) 1999, 461–462, 1–21.
[23] Gaussian 03, Revision C.01 ed., M. J. Frisch, G. W. Trucks, H. B.
Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, J. A. Mont-
gomery 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, A. J. Austin, R. Cammi, C. Pomelli, J. W.
Ochterski, P. Y. Ayala, K. Morokuma, G. A. Voth, P. Salvador, J. J.
Dannenberg, 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. Cio-
slowski, B. B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaro-
mi, 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, J. A. Pople, Gaussian, Inc. Pitts-
burgh, PA, 2003.
[24] a) J. C. Slater, Quantum Theory of Molecules and Solids, Vol. 4,
McGraw-Hill, New York, 1974; b) S. H. Vosko, L. Wilk, M. Nusair,
Can. J. Phys. 1980, 58, 1200–1211.
[12] a) C. Bingel, Chem. Ber. 1993, 126, 1957–1959; b) X. Camps, A.
Hirsch, J. Chem. Soc. Perkin Trans. 1 1997, 1595–1596.
[13] a) M. Prato, M. Maggini, Acc. Chem. Res. 1998, 31, 519–526; b) M.
Maggini, E. Menna, in Fullerenes: From Synthesis to Optoelectronic
Properties (Eds.: D. M. Guldi, N. Martín), Kluwer Academic Pub-
lishers, Dordrecht (The Netherlands), 2002, Chapter 1, pp. 1–50.
[14] O. Lukoyanova, C. M. Cardona, M. Altable, S. Filippone, A. Martín-
Domenech, N. Martín, L. Echegoyen, Angew. Chem. 2006, 118,
7590–7593; Angew. Chem. Int. Ed. 2006, 45, 7430–7433.
[15] E. Vedejs, J. W. Grissom J. Am. Chem. Soc. 1988, 110, 3238–3246.
[16] M. Cases, M. Duran, J. Mestres, N. Martín, M. Solà, J. Org. Chem.
2001, 66, 433–442.
[17] a) M. Cases, M. Duran, J. Mestres, N. Martín, M. Solà in Fullerenes
for the New Millennium (Eds.: P. V. Kamat, K. M. Kadish, D. M.
Guldi), Vol. 11, The Electrochemical Society, Pennington, 2001,
pp. 244–269; b) E. U. Wallenborn, R. F. Haldimann, F. G. Klarner,
F. Diederich, Chem. Eur. J. 1998, 4, 2258–2265; c) X. Zhuang, Z.
Yang, J. Zhang, W. Cao, J. Mol. Struct. (Theochem) 2006, 760, 45 –
52.
[25] W. J. Hehre, R. F. Stewart, J. A. Pople, J. Chem. Phys. 1969, 51,
2657–2664.
[26] a) A. D. Becke, J. Chem. Phys. 1993, 98, 5648–5652; b) C. Lee, W.
Yang, R. G. Parr, Phys. Rev. B 1988, 37, 785–789; c) P. J. Stephens,
F. J. Devlin, C. F. Chabalowski, M. J. Frisch, J. Phys. Chem. 1994, 98,
11623–11627.
[27] a) W. J. Hehre, R. Ditchfield, J. A. Pople, J. Chem. Phys. 1972, 56,
2257–2261; b) P. C. Hariharan, J. A. Pople, Theor. Chim. Acta 1973,
28, 213–222.
[28] a) M. Freccero, R. Gandolfi, M. Sarziamade, A. Rastelli, J. Chem.
Soc. Perkin Trans. 2 1998, 2413–2419; b) O. Wiest, K. N. Houk,
K. A. Black, B. Thomas, J. Am. Chem. Soc. 1995, 117, 8594–8599;
c) O. Wiest, K. N. Houk, Top. Curr. Chem. 1996, 183, 1–24; d) E.
Goldstein, B. Beno, K. N. Houk, J. Am. Chem. Soc. 1996, 118, 6036–
6043; e) T. C. Dinadayalane, R. Vijaya, A. Smitha, G. N. Sastry, J.
Phys. Chem. A 2002, 106, 1627–1633; f) H. Isobe, S. Yamanaka, K.
Yamaguchi, Int. J. Quantum Chem. 2003, 95, 532–545; g) C. Di Va-
lentin, M. Freccero, R. Gandolfi, A. Rastelli, J. Org. Chem. 2000, 65,
6112–6120.
[18] Z. Shang, Y. Pan, Z. Cai, X. Zhao, A. Tang, J. Phys. Chem. A 2000,
104, 1915–1919.
[19] L. Türker, J. Mol. Struct. (Theochem) 2002, 588, 165–174.
[20] a) We have been unable to reach full convergence for this transition
state because of the extreme flatness of the surface around this posi-
tion. The root-mean-square value for the forces converged and the
maximum force at this point was less than 1ꢂ10ꢀ3 a.u. However, the
[29] P. W. Atkins, Physical Chemistry, Oxford University Press, Oxford,
1982.
Received: January 17, 2008
Published online: April 25, 2008
5206
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 5198 – 5206