DOI: 10.1002/chem.

200902024

Double Group Transfer Reactions: Role of Activation Strain

and Aromaticity in Reaction Barriers

Israel Fernndez,*[a] F. Matthias Bickelhaupt,*[b] and Fernando P. Cosso*[c]

Abstract: Double group transfer
(DGT) reactions, such as the bimolecu-
lar automerization of ethane plus
ethene, are known to have high reac-
tion barriers despite the fact that their
cyclic transition states have a pro-
nounced in-plane aromatic character,
as indicated by NMR spectroscopic pa-
rameters. To arrive at a way of under-
standing this somewhat paradoxical
and incompletely understood phenom-
enon of high-energy aromatic transi-
tion states, we have explored six arche-
typal DGT reactions using density
functional theory (DFT) at the OLYP/
TZ2P level. The main trends in reactiv-
ity are rationalized using the activation
strain model of chemical reactivity. In
this model, the shape of the reaction
profile DE(z) and the height of the
Keywords: activation strain model ·
aromaticity · density functional
calculations · double group transfer
reactions · reactivity
overall reaction barrier DE ¼6 = DE-
ACHTUNGRE(z=zTS) is interpreted in terms of the
strain energy DEstrain(z) associated with
deforming the reactants along the reac-
tion coordinate z plus the interaction
energy DEint(z) between these de-
formed reactants: DE(z) = DEstrain(z) +
DEint(z). We also use an alternative
fragmentation and a valence bond
model for analyzing the character of
the transition states.

Introduction another in a concerted reaction pathway.[1] This definition

Double group transfer (DGT) reactions are a general class
of pericyclic reactions that occur through the simultaneous
migration of two atoms or groups from one compound to
[a] Dr. I. Fernndez
Departamento de Qumica Orgnica, Facultad de Qumica
Universidad Complutense, 28040 Madrid (Spain)
Fax: (+ 34) 913944310
E-mail: israel@quim.ucm.es
[b] Prof. Dr. F. M. Bickelhaupt
Department of Theoretical Chemistry
and Amsterdam Center for Multiscale Modeling
Scheikundig Laboratorium der Vrije Universiteit
De Boelelaan 1083, 1081 HV Amsterdam (The Netherlands)
Fax: (+ 31) 205987629
E-mail: fm.bickelhaupt@few.vu.nl
[c] Prof. Dr. F. P. Cosso
Departamento de Qumica Orgnica I-Kimika Organikoa I Saila
Facultad de Qumica-Kimika Fakultatea
Universidad del Pas Vasco-Euskal Herriko Unibertsitatea
and Donostia International Physics Center (DIPC)
P.K. 1072, 20080 San Sebastin-Donostia (Spain)
Fax: (+ 34) 943015270
E-mail: fp.cossio@ehu.es
Supporting information for this article, which includes Cartesian coor-
dinates and total energies of all stationary points, is available on the
WWW under http://dx.doi.org/10.1002/chem.200902024.
includes textbook reactions like the diimide reduction of
double or triple bonds,[2] the Meerwein–Ponndorf–Verley re-
duction (MPV) of carbonyl groups,[3] and some type II dyo-
tropic reactions that are characterized by the intramolecular
transfer of the two groups (generally hydrogen atoms).[4]
The archetypal DGT process is the thermally allowed con-
certed and synchronous transfer of two hydrogen atoms
from ethane to ethylene. The process is suprafacial on both
reaction sites and therefore these [s2s+s2s+p2s] transforma-
tions may be considered to be a thermally allowed pericyclic
reaction according to the Woodward–Hoffmann rules.[5]
Strikingly, DGT reactions share a common feature, namely,
they proceed in a concerted and synchronous fashion
through a transition state featuring a six-membered ring
that is highly in-plane aromatic,[6] as indicated by the high
negative nuclear independent chemical shift (NICS) values[7]
computed at the (3,+1) ring critical point of the electron
density.[8]
Despite the aromatic character of these transition states,
DGT reactions are associated with relatively high barriers.
So far, the origins of the high activation barriers typically
observed for DGT reactions in particular, and pericyclic pro-
cesses in general, have not been fully understood. The rela-
tively recent introduction of the so-called activation strain
model has allowed us to gain more insight into the physical

13022  2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 2009, 15, 13022 – 13032

factors that control how the activation barriers arise in dif-
ferent fundamental processes.[9] The activation strain model
is a fragment approach to understanding chemical reactions
in which the height of reaction barriers is described and un-
derstood in terms of the original reactants. Bickelhaupt and
co-workers have pioneered the systematic use of this meth-
odology for elementary reactions in organic chemistry and
catalysis.[9] They showed, for instance, how the activation
barrier of SN2 reactions originates from steric congestion
around the carbon atom that is nucleophilically attacked,
and how electronic factors modulate the barrier height.[9, 10]
Not much later, Houk et al.[11] were the first to successful-
ly apply the activation strain model (or distortion/interaction
model) to pericyclic reactions, in particular to [3+2] cycload-
dition reactions. The latter, curiously, are concerted and
occur through highly synchronous and in-plane aromatic
transition states, very similar to the situation for DGT reac-
tions.[12] Thus, Houk et al.[11a] found that the barrier heights
for the cycloadditions of a given 1,3-dipole with ethylene
and acetylene are nearly the same despite very different re-
action thermodynamics and fragment molecular orbital
(FMO) energy gaps. One of the key findings was that the
activation strain, that is, the energy to distort the 1,3-dipole
and dipolarophile to their corresponding transition-state ge-
ometries, is the major factor controlling the reactivity differ-
ences of 1,3-dipoles. Furthermore, interaction energies be-
tween the 1,3-dipole and the dipolarophile differentiate re-
activity for a series of substituted alkenes when the distor-
tion energies are nearly constant.
Herein, we aim for a deeper understanding of the origins
of the barrier heights of the fundamental class of DGT reac-
tions. We report the results of a density functional theory
(DFT) study of six archetypal examples of such transforma-
tions based on the activation strain model. It will appear
that, in good agreement with Houks findings for [3+2] cy-
cloadditions, the activation strain associated with the struc-
tural rearrangement of the reactants is also the controlling
factor for the high energy of the cyclic DGT reaction barri-
ers.
Theoretical Methods
Computational details: All calculations were performed with the Amster-
dam density functional (ADF) program developed by Baerends and
others.[13, 14] The molecular orbitals (MOs) were expanded in a large un-
contracted set of Slater-type orbitals (STOs) containing diffuse functions
(TZ2P). This basis is of triple-z quality and has been augmented by two
sets of polarization functions, that is, p and d functions for the hydrogen
atom and d and f functions for the other atoms. An auxiliary set of s, p,
d, f, and g STOs was used to fit the molecular density and to represent
the Coulomb and exchange potentials accurately in each SCF cycle. Rela-
tivistic effects were accounted for by using the zeroth-order regular ap-
proximation (ZORA).[15]
Equilibrium and transition-state geometries were fully optimized at the
OLYP[16] density functional, which involves Handys optimized exchange
(OPTX). All stationary points were confirmed by vibrational analysis:[17]
for equilibrium structures, all normal modes have real frequencies,
whereas transition states[18] have one normal mode with an imaginary fre-
Chem. Eur. J. 2009, 15, 13022 – 13032  2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
FULL PAPER
quency. Furthermore, transition states were verified to connect the sup-
posed reagent and product minima by carrying out intrinsic reaction co-
ordinate (IRC) calculations.[19]
Activation strain analyses of reaction profiles: Insight into how the reac-
tion barriers arise was obtained through activation strain analyses of the
various DGT processes.[9, 11] The activation strain model is a systematic
extension of the fragment approach from equilibrium structures to transi-
tion states as well as nonstationary points (e.g., points along a reaction
coordinate). It is a fragment approach to understanding chemical reac-
tions in which the height of reaction barriers is described and understood
in terms of the original reactants. Thus, the potential-energy surface
DE(z) is decomposed, along the reaction coordinate z, into the strain
DEstrain(z) associated with deforming the individual reactants plus the
actual interaction DEint(z) between the deformed reactants [Eq. (1)]:
DEðzÞ ¼ DEstrain ðzÞ þ DEint ðzÞ ð1Þ
Here, the reaction coordinate is defined as the projection of the IRC
(vide supra) on the HC distance between the hydrogen atom that is to
be transferred and the carbon atom to which it is transferred. This reac-
tion coordinate z undergoes a well-defined change in the course of the
reaction from 1 to the equilibrium HC distance in the reduced prod-
uct.
The strain, DEstrain(z), is determined by the rigidity of the reactants and
the extent to which groups must reorganize in a particular reaction mech-
anism, whereas the interaction, DEint(z), between the reactants depends
on their electronic structure and how they are mutually oriented as they
approach each other. It is the interplay between DEstrain(z) and DEint(z)
that determines if and at which point along z a barrier arises. The activa-
tion energy of a reaction DE ¼6 = DEACHTUNGRE(zTS) consists of the activation strain
DE6¼strain = DEstrainACHTUNGRE(zTS) plus the TS interaction DE6¼int = DEintACHTUNGRE(zTS) [Eq. (2);
see Figure 1]:
DE° ¼ DE° °
ð2Þ
strain þ DEint
Figure 1. Illustration of the activation strain model for a DGT reaction.
Molecular orbital theory and energy decomposition analysis: The interac-
tion DEint(z) between the strained reactants was further analyzed in the
conceptual framework provided by the Kohn–Sham molecular orbital
(KSMO) model.[20] To this end, it was further decomposed into three
physically meaningful terms [Eq. (3)]:
DEint ðzÞ ¼ DV elstat þDEPauli þDEorb ð3Þ
The term DVelstat corresponds to the classical electrostatic interaction be-
tween the unperturbed charge distributions of the deformed reactants
www.chemeurj.org 13023
 I. Fernndez, F. M. Bickelhaupt, and F. P. Cosso

and is usually attractive. The Pauli repulsion DEPauli comprises the desta- Table 1. Selected reaction parameters (E in kcal mol1, D in ) of DGT
bilizing interactions between occupied orbitals and is responsible for any reactions.[a]
steric repulsion (see ref. [20] for an exhaustive discussion). The orbital in-
Reaction DE° DEr Dshort
teraction DEorb accounts for charge transfer (interaction between occu-
pied orbitals on one moiety with unoccupied orbitals on the other, in-
cluding HOMO–LUMO interactions) and polarization (empty–occupied 1 45.4 0.0 2.732
orbital mixing on one fragment due to the presence of another frag-
ment).
Since the Kohn–Sham MO method of density functional theory (DFT) in 2 41.8 12.1 2.622
principle yields exact energies and, in practice, with the available density
functionals for exchange and correlation, is rather accurate, we have the
special situation that a seemingly one-particle model (an MO method) in
3 44.5 7.0 2.656
principle fully accounts for the bonding energy.[20] Through these energy
decomposition analyses (EDA), it has previously been possible, for exam-
ple, to compare the strength of conjugation, hyperconjugation, and aro-
maticity in p-conjugated systems[21, 22] and to analyze the nature of a 4 43.7 12.1 2.567
chemical bond in terms of electrostatic attraction versus s-, p-, and d-or-
bital (covalent) bonding.[23, 24]
Alternative analyses of transition states: We have also carried out an al- 5 49.0 19.6 2.655
ternative analysis of the transition states. In this approach, the model
system is partitioned, not into the two original reactants (e.g.,
C2H6+C2H4), but instead into the organic scaffold between which double 6 24.5 0.0 2.382
hydrogen transfer takes place plus the two hydrogen atoms (i.e., the two
fragments [C2H4–C2H4]··+ACHTUNGRE[H–H]··, each of which has a triplet-valence
electron configuration). To model the progress of the interactions along [a] Computed at the OLYP/TZ2P level. Dshort = the sum of the shortest
the IRC from TS to reactants in which the biradical character localizes distance of the breaking XH and forming HY bonds in the TS []
towards one of the two C2H4 moieties, a similar analysis has been carried (i.e., upper X-H-Y moiety in Figure 2).
out for the interaction between the organic scaffold of the ethane mole-
cule that is forming plus the same two hydrogen atoms (i.e., the two frag-
ments [C2H4]··+ACHTUNGRE[H–H]··, each of which again has a triplet-valence electron
configuration). The details of this approach are provided later on in the
discussion. The MO and EDA analyses in this alternative fragmentation
provide insight from a different perspective that is complementary to and
augments the activation strain analyses.
NICS values and valence bond analyses: The aromatic character of the
transition states has been confirmed by the computation of the NICS[7]
values computed at the (3,+1) ring critical point of the electron density[8]
(see below). These calculations have been carried out with the Gaussi-
an 03 suite of programs[25] using the gauge invariant atomic orbital
(GIAO) method,[26] at the OLYP level using the triple-zeta plus polariza-
tion basis sets (def2-TZVPP)[27] with the optimized OLYP/TZ2P geome-
tries. This scheme is denoted GIAO-OLYP/def2-TZVPP//OLYP/TZ2P.
For the computational details related to the valence bond analyses, see
the Results and Discussion section.

Figure 2. Geometries [] of the transition states of our DGT reactions,

computed at the OLYP/TZ2P level. See Table 1 for the position of heter-
Results and Discussion oatoms.

Structures and reaction profiles: In Table 1, we provide an
overview of our DGT model reactions together with the re-
action barriers that we have computed at the OLYP/TZ2P
level. Structures of the corresponding transition states (TS)
are depicted in Figure 2. We consider the parent reactions of
ethane with ethene and ethyne (reaction numbers 1 and 2),
and the corresponding processes involving heteroatoms, that
is, the reduction by ethane of imine and formaldehyde (reac-
tions 3 and 4) as well as nitrile (reactions 5). In addition, we
have examined the uncatalyzed Meerwein–Ponndorf–Verley
reduction of formaldehyde by methanol (reaction 6).[5]
There appears to be a correlation between the barrier
height and the geometries of the TS, in particular, the
breaking XH and forming HY distances. As can be seen
in Table 1, all reactions proceed through high energy barri-
ers of 40–50 kcal mol1. Note that the Meerwein–Ponndorf–
Verley reaction (reaction 6) has a much lower barrier of
24.5 kcal mol1. Whereas correlations should of course
always be treated with precaution, one can observe that the
reaction barrier DE° increases if the shortest of the two X-
H-Y contacts (i.e., the upper one in the TS structures of
Figure 2) becomes longer, as measured by the Dshort = rACHTUNGRE(X
H)+rACHTUNGRE(HY), which is also provided in Table 1. The correla-
tion that we find between Dshort and the barrier height
matches nicely with kinetic and structural experiments that
show that changes in Dshort of only 0.1–0.17  translate into
a rate spread of 104 s1.[28]
Figure 2 shows the corresponding transition states of the
above-mentioned transformations. In all cases, the transfor-
mations occur through highly symmetric six-membered

13024 www.chemeurj.org  2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 2009, 15, 13022 – 13032
Double Group Transfer Reactions
cyclic transition states in which the corresponding CC, C
N, and CO bonds present bond lengths that are intermedi-
ate between double and single bonds. The observed bond-
length equalization and planarity of the transition states
agree nicely with the highly in-plane aromatic character of
these species, which has been confirmed by the calculation
of the NICS(0), computed at the (3, + 1) ring critical point
of the electron density as defined by Bader,[8] and NICS(1)zz
values[29] (the out-of-plane component of the NICS comput-
ed 1  above the ring critical point, Table 2).
Table 2. Computed NICS values [ppm] of transition states TS1–TS6.
TS1 TS2 TS3 TS4 TS5
[a]
NICS(0) 27.04 25.76 26.38 27.16 25.41
NICS(1)[b] 12.77 12.23 10.97 11.50 12.25
NICS(1)zz[c] 35.74 34.77 31.90 31.41 34.63
[a] NICS(0) computed at GIAO-OLYP/def2-TZVPP//OLYP/TZ2P at the
(3,+1) ring critical point. [b] NICS(1) computed 1  above the (3,+1)
ring critical point. [c] NICS(1)zz denotes the zz tensor component of
NICS(1).
Activation strain analyses: Next we address the main pur-
pose of our study: to understand the factors that are respon-
sible for the high activation barriers of DGT reactions, using
the activation strain model. Figure 3 shows the plots of the
computed potential-energy surface along the IRC trajecto-
ries, projected onto the distance rHC (see the Theoretical
Methods section) for reactions 1 to 6 in Table 1, together
with the change of the energy contributors to DE(z),
namely, the strain DEstrain(z) and the instantaneous interac-
tion DEint(z) between the deformed reactants. All our model
systems present quite similar energy trends. Thus, we can
see that at the early stages of the process the reaction pro-
file DE monotonically becomes more and more destabilized
as the reactants approach each other. Interestingly, a sharp
increase of DE occurs in the proximity of the transition-
state region (i.e., at HC distances in the range from 2.0 to
1.6 ), thereby leading to the observed high reaction barri-
ers.
At long HC distances, the interaction energy between
the deformed reactants (DEint) becomes destabilizing and
causes the net energy DE to go up too, as can be seen in
Figure 3. The initial increase of DEint can be traced to steric
(Pauli) repulsion between the reactants in the early stages of
the reaction. Thus, before anything else happens, the reac-
tants approach and overlap occurs between closed shells, no-
tably between CH bonds of the hydrogen donor and the p
system of the hydrogen acceptor, which causes Pauli repul-
sion. This behavior of DEint is interesting as it differs from
reactions such as SN2 substitution and E2 elimination.[9a,h] In
the latter, there is of course also Pauli repulsion between
closed shells of the reactants as they approach each other.
But in addition, there is in these reactions, from the begin-
ning, a potent donor–acceptor interaction between the
HOMO of the nucleophile/base and the relatively low-
Chem. Eur. J. 205., 11, 13022 – 13032  205. Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
FULL PAPER
energy LUMO of the substrate. This means that for SN2 and
E2 reactions the net interaction DEint is stabilizing along the
entire reaction coordinate and the destabilization is con-
tained entirely in the strain DEstrain term.
If we now proceed further along the DGT reaction coor-
dinate, the trend in DEint inverts at a certain point, after
which this term becomes more and more stabilized as one
further approaches the transition state. Inspection of
Figure 3 shows that this stabilization in the DEint curve
occurs shortly after the onset of the strain curve. A decom-
position of DEint into the different terms of the interaction
energy furthermore shows that the dominant term causing
this inversion is the orbital interaction energy DEorb (see
TS6
Figure 4). Note that the electrostatic attraction DVelstat is cer-
24.48
tainly far from being negligible, but it is not the dominant
9.27
25.46 bonding term. The reason for the onset of sizable orbital in-
teractions at shorter HC distances is that the original CH
bonds are beginning to elongate. This causes the associated
s*CH orbitals to drop in energy and become more localized
on the transferring hydrogen atoms. As a result, the
HOMO–LUMO interactions between the p-electron system
of the hydrogen acceptor reactants and the s*CH orbitals of
the hydrogen donor become stronger and more stabilizing
(see Figure 4).
The reason that the overall energy DE still goes up until
reaching the TS is, of course, also the increase in the strain
energy that goes with the aforementioned elongation of the
CH bonds in the hydrogen-donor reactant. We also find a
contribution to the strain stemming from the conformational
change in the hydrogen donor (ethane or methanol) from
staggered to eclipsed. But the breaking of the two CH
bonds quickly turns into the dominant contribution of the
strain term as the TS is approached.
Interestingly, the uncatalyzed Meerwein–Ponndorf–Verley
reaction (reaction 6 in Table 1; Figure 3f) presents the
lowest computed energy barrier. This may be explained by
the following reasons: 1) the total bond-dissociation energy
required to liberate the two migrating hydrogen atoms,
thereby breaking the CH and OH bonds, is approximate-
ly 7 kcal mol1 lower in CH3OH than in CH3CH3 (see
Scheme 1); 2) we found a stabilizing effect when using
CH3OH as a “hydrogen donor,” which is not able to occur
using CH3CH3, in the formation of an OH···O hydrogen
bond between both reactants, which facilitates the hydrogen
migration if both reactants are in close proximity. The latter
stabilizing contribution is nicely reflected in the higher com-
puted DEint value (i.e., DEint  34 kcal mol1 for reaction 6
and DEint  22 kcal mol1 for the parent reaction 1,
Table 1) and in a not-so-high strain energy (i.e., DEstrain
 59 kcal mol1 for reaction 6, and DEstrain  65 kcal mol1 for
the parent reaction 1). Both effects contribute to the ob-
served decrease of the activation barrier in this system.
Based on the above discussion, it can be concluded that,
similarly to the pericyclic processes studied so far, the acti-
vation strain associated with the structural rearrangement of
the reactants is also the controlling factor for the high
energy of the DGT reaction barriers. This makes sense in
www.chemeurj.org 13025
 I. Fernndez, F. M. Bickelhaupt, and F. P. Cosso

Figure 3. Activation strain analysis of the DGT reactions (in kcal mol1, at the OLYP/TZ2P level) along the reaction coordinate projected onto the HC
bond length. a) to f) correspond to reactions 1 to 6, respectively (see Table 1).

view of the geometrically distorted nature of DGT transi-
tion states in which two CH bonds and one p bond must
first be partially broken before an equivalent number of
new bonds are formed instead.
Aromatic character according to NICS(0): The evolution of
the aromaticity (measured by the NICS(0) values) along the
reaction coordinate in the parent ethane plus ethene reac-
tion was also computed. As expected, the NICS values
become more and more negative (that is, the system be-
comes more and more aromatic) and reaches the maximum
value for the transition state. Note that the NICS(0) curve is
a mirror image of the total energy DE (Figure 5a), which
might be an indication of a correlation of both parameters.

13026 www.chemeurj.org  2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 2009, 15, 13022 – 13032
Double Group Transfer Reactions
FULL PAPER

Figure 4. Decomposition of the interaction energy between the reactants in model reactions 1–6 (a–f; see Table 1) along the reaction coordinate project-
ed onto the HC bond length, computed at the OLYP/TZ2P level.

In fact, when one plots DE values versus the NICS(0)
values, a nice second-order polynomial curve is obtained
(correlation coefficient of 0.998), thus confirming the close
relationship between both parameters (Figure 5b). It is
clearly shown in both parts of Figure 5 that DE continuously
increases with the concomitant gain in aromaticity. This fact
seems contradictory if we consider that a gain in aromaticity
is usually translated into a gain in stability. Therefore, this
apparent contradiction provides further support to the
above-mentioned conclusion that the reaction is controlled

Chem. Eur. J. 2009, 15, 13022 – 13032  2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemeurj.org 13027

Scheme 1. Computed (OLYP/TZ2P level) bond-dissociation energies (in
kcal mol1); (d) and (t) indicate doublet and triplet state, respectively.
Figure 5. a) Evolution of DE and NICS(0) (computed at the GIAO-
OLYP/def2-TZVPP//OLYP/TZ2P level) along the reaction coordinate.
b) Plot of the correlation between DE and NICS(0).
by the strain in spite of the gain in stability by aromatic de-
localization.
Valence bond perspective: Since the dyotropic ring current
present in aromatic structures is closely related to cyclic
electron stabilization, we decided to analyze the magnitude
of the resonance energy of the TS corresponding to reac-
tion 1 (see Table 1) to gain a better understanding of the re-
lationship between its high aromatic character and the
13028 www.chemeurj.org  2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
I. Fernndez, F. M. Bickelhaupt, and F. P. Cosso
energy barrier. Figure 6 shows the valence bond state corre-
lation diagram (VBSCD) for the degenerate double hydro-
gen transfer from eclipsed C2v-symmetric ethane to ethylene.
Figure 6. Valence-bond-state correlation diagram for reaction 1 (see
Table 1). The inset includes the B3LYP/6-31G* spin contributions pro-
jected onto the electron density of eclipsed ethane (C2v symmetry) and
ethylene (D2h symmetry), both in the triplet state and at the ground-state
geometry. Colors range from 0.002 a.u. (blue) to 0.000 a.u. (red).
From the VB standpoint,[30] the mechanism of this reaction
can be envisaged as the result of an avoided crossing be-
tween the adiabatic state lines YR !Y*P and Y*R !YP. The
magnitude of the activation energy DEa is given by Equa-
tion (4):
DEa ¼ DEc B ð4Þ
in which B is the resonance energy of the TS, and DEc is the
energy barrier corresponding to the crossing point. This
latter magnitude can be expressed as shown in Equation (5):
DEc ¼ fG ð5Þ
in which G is the promotion energy on going from YR to
YR*. For this kind of pericyclic reaction, G corresponds to
the vertical singlet–triplet gap required to change the elec-
tronic state of the reactants to adapt it to that of the prod-
ucts while keeping the singlet-state geometry unchanged.
Therefore, G is calculated as shown in Equation (6):
G ¼ EST ðC2v C2 H6 ÞþEST ðD2h C2 H4 Þ ð6Þ
The two terms of the right side of this expression corre-
spond to the vertical singlet–triplet (ST) gap of C2v-symmet-
ric ethane and D2h-symmetric ethylene, respectively. At the
B3LYP/6-31G* level, it is found that ESTACHTUNGRE(D2hC2H4) =
104.1 kcal mol1, which is close to the experimental value of
100.5 kcal mol1.[31] Similarly, at the same level we have
Chem. Eur. J. 2009, 15, 13022 – 13032
Double Group Transfer Reactions
found that ESTACHTUNGRE(C2vC2H6) = 241.7 kcal mol1. Therefore, G =
345.8 kcal mol1. This is a very high value that is caused by
the strongly distorted geometry of the eclipsed geometry of
ethane in the triplet state. Under these conditions, most of
the spin density of the Y*R state is concentrated in the mi-
grating hydrogen atoms of ethane and the carbon atoms of
ethylene (Figure 6), a spin distribution required for pairing
the electrons across the novel s bond. The large value of G
can be compared with that computed for the Diels–Alder
reaction between 1,3-butadiene and ethylene. In this case, it
is found that G = 173 kcal mol1 because of the much lower
EST value of 1,3-butadiene.[32] The parameter f reflects the
fraction of G that is responsible for the height of the cross-
ing energy DEc. If we assume a value of f  0.3 for thermally
allowed pericyclic reactions,[32] we obtain DEc  103.7 kcal
mol1 and therefore B  58.6 kcal mol1. This resonance
energy of the transition state is approximately 20 kcal mol1
higher than that calculated for the parent Diels–Alder reac-
tion. Therefore, we can conclude that the relatively high ac-
tivation barrier for this reaction is compatible with a highly
aromatic TS. This apparent paradox is caused by the huge
vertical energy gap required to adapt the electronic state of
the reactants to that of the products while keeping the geo-
metric features of the reactants in the ground electronic
state.
Aromaticity and stability of transition states: The above re-
sults show that a high energy of the DGT transition state
can go with a highly aromatic character. It is interesting in
this context to note that indeed cyclic delocalization causes
a stabilization with respect to the corresponding noncyclic
structure. This is well known, for example, for the p-electron
systems of benzene (cyclic delocalization: more stable) as
compared with 1,3,5-hexatriene (linear, no cyclic delocaliza-
tion: less stable). This is associated with the presence of a
sixth bond (or overlap or interaction matrix element) in the
circular system and the concomitant lowering of the kinetic
energy in the lowest-energy p level (cf. particle on a ring).[33]
In terms of simple VB resonance structures, one can illus-
trate this with the stabilizing resonance between the two
Kekul structures for benzene and, in the case of 1,3,5-hexa-
triene, the absence of low-energy resonance structures and
thus no stabilizing resonance.
The same situation occurs for the six s electrons involved
in the bond breaking and making in the ring structure of
DGT transition states. We illustrate that quantitatively by
computing the stabilization that occurs by combining the
two equivalent HCH2=CH2C radicals, of which the DGT
transition state of reaction 1 (Table 1) can be constructed in
two different ways as shown in Table 2: 1) circularly, thus
yielding the DGT transition state; and 2) in a linear fashion
by inflecting one of the two HCH2=CH2C fragments in the
indicated mirror line. The latter system is not a stationary
point. It serves merely to estimate the additional aromatic
stabilization that goes with circular conjugation as compared
with a system in which this circular conjugation is interrupt-
ed while as many as possible geometry parameters are kept
Chem. Eur. J. 2009, 15, 13022 – 13032  2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
FULL PAPER
equal. The corresponding interactions are also collected in
Table 3, together with an energy decomposition analysis.
Table 3. Analysis of the interaction between two equivalent HCH2 = CH2· frag-
ments in the circular TS of DGT reaction 1 and the exact same fragments com-
bined in a linear fashion.[a]
Circular (TS of DGT reaction 1) Linear system
DEint 60.4 31.1
DEPauli 167.6 103.2
DVelstat[b] 70.7 (31 %) 40.6 (30 %)
DEorb[b] 157.4 (69 %) 93.7 (70 %)
DEs[c] 155.7 (99 %) 92.9 (99 %)
DEp[c] 1.7 (1 %) 0.8 (1 %)
[a] In kcal mol1, computed at the OLYP/TZ2P level. [b] The percentage values
in parentheses give the contribution to the total attractive interactions
DEelstat+DEorb. [c] The percentage values in parentheses give the contribution to
the total orbital interactions DEorb.
Not unexpectedly, Table 3 shows that interrupting the aro-
matic (i.e., circular) conjugation of the six s electrons leads
to a significant weakening of the interaction, which becomes
half as strong as in the circular TS. Note that this indeed
comes mainly from a weakening of the orbital interactions,
which for 99 % of those interactions stem from the s-elec-
tron system in which the aromatic conjugation takes place.
Note, too, that the gain in DEorb from the linear to the cyclic
structure (i.e., 63 kcal mol1) is approximately equal to the
resonance B term in VB theory (58.6 kcal mol1, vide
supra). In an alternative, more qualitative manner, this may
again be represented by two VB resonance structures that
are in stabilizing, aromatic resonance in the circular DGT
transition state (see I, and the corresponding resonance
structures II and III), whereas such low-energy (nonionic,
closed-shell) resonance structures are not possible for a
linear arrangement (see IV, which has only resonance struc-
ture V, Scheme 2).
There is thus a substantial stabilization due to the aromat-
ic sextet of electrons in the circular DGT transition state.
Nevertheless, the DGT transition state has a strong tenden-
cy to localize two CH bonds and a CC p bond. One
reason is the aforementioned strain and repulsion associated
with pushing the reactants together (e.g., ethane and ethene
in reaction 1 in Table 1). There is a second aspect that needs
to be taken into consideration, namely, the question of
whether the aromatic electrons have an intrinsic propensity
to keep the molecule in its regular or circular, “aromatic”
geometry or not. It has been shown in several complementa-
ry studies that the six p electrons in benzene actually favor
a localized, cyclohexatriene-type structure.[22a,c–e, 34] Recently,
we have explained this behavior in terms of a quantitative
molecular orbital (MO) model.[22c,d] The reason is that three
www.chemeurj.org 13029
 I. Fernndez, F. M. Bickelhaupt, and F. P. Cosso

(i.e., [C2H4–C2H4]··+ACHTUNGRE[H–H]··, both fragments in their triplet

valence states) instead of the interaction between the two
original reactants (i.e., C2H6+C2H4). Figure 7 shows the
schematic orbital-interaction diagram for the DGT transi-
tion state that emerges from our quantitative Kohn–Sham
MO analyses at OLYP/TZ2P. If we move along the IRC to-
wards the reactants, the two C2H4 moieties in the organic
scaffold quickly separate and localize the triplet character
on the p and p* orbitals of one of the two C2H4 fragments.
We have modeled the interactions and overlaps through
analysis of the interacting [C2H4]·· and [H–H]·· fragments
that approach each other along the IRC of the overall reac-
tion system. The results are shown in Figure 8. Clearly, the
SOMO–SOMO overlap and orbital interactions favor locali-
zation of the CH bonds. One can therefore conclude that
the DGT transition state has a strong aromatic character in
the sense that it receives substantial stabilization from circu-
lar conjugation as compared with the less favorable linear
Scheme 2. Resonance and resonance structures in the transition state of
DGT reaction 1 (see Table 1) (I–III) and a linear analogue (IV, V).

localized (i.e., short C=C double bonds) yield effectively

more p overlap than six delocalized (i.e., somewhat elongat-
ed) ones.[22, 34]
A similar behavior can be observed in the present case of
the DGT transition state. Here, the question is whether the
two hydrogen atoms that are transferred have the propensi-
ty to keep the “aromatic geometry” of four delocalized C–
H–C bridges, or if localization towards two regular, short
CH bonds in ethane (and rupture from the remaining
ethene) yields better overlap and more stabilization. To this
end, we have analyzed for DGT reaction 1 (Table 1) the in-
teraction of the organic scaffold between which double hy-
drogen transfer takes place plus the two hydrogen atoms

Figure 8. a) Energy decomposition analysis and b) SOMO–SOMO bond

Figure 7. Schematic orbital interaction diagram between [C2H4–C2H4]·· overlaps for the CH interaction between [C2H4]·· and [H–H]·· fragments
and [H–H]·· in the TS of the DGT reaction 1 (see Table 1). along the IRC of DGT reaction 1 (see Table 1).

13030 www.chemeurj.org  2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 2009, 15, 13022 – 13032
Double Group Transfer Reactions
situation. But, at the same time, repulsion between the reac-
tants causes an overlap behavior in which the six aromatic s
electrons achieve best overlap and thus an even more stable
geometrical arrangement if the CH bonds are localized
under formation of ethane and ethene. The aromatic ring
currents, indicated by the NICS values (vide supra), simply
reflect the small HOMO–LUMO gap in the DGT transition
state that goes with the distortions (i.e., the occurrence of
partial instead of full s CH and p CC bonds) relative to
the stable reactants.
Conclusion
From the computational study reported in this paper the fol-
lowing conclusions can be drawn. 1) Similarly to other re-
ported pericyclic reactions, the activation strain associated
with the structural rearrangement of the reactants is the
controlling factor for the high activation barrier of double
group transfer reactions. 2) The gain in stability by reso-
nance and in-plane aromaticity of the transition structures
cannot compensate for the strong destabilizing effect of the
strain; 3) The in-plane aromatic character of the six s elec-
trons involved in the DGT bond rearrangement is confirmed
by the strong stabilization we compute for circular conjuga-
tion as compared with a fictitious linear system. 4) Never-
theless, the six in-plane aromatic s electrons have a propen-
sity to localize the C–H–C bridges towards two full CH
bonds (under formation of either reactants or products) be-
cause in this way even more favorable bond overlap is
ACHTUNGREachieved (this is reminiscent of the aromatic p electrons of
benzene, which have a propensity to localize the double
bonds but are overruled by the s-electron system).[22, 34]
5) The activation strain model is confirmed to be useful and
fully applicable for understanding the intricacies also of per-
icyclic reactions.
, M. Swart, D. Swerhone, G. te
Acknowledgements
We thank the Spanish Ministerio de Ciencia e Innovacin (MCINN;
CTQ2007-67528), Consolider-Ingenio 2010 (CSD2007-0006), and the
Netherlands Organization for Scientific Research (NWO-CW) for finan-
cial support. I.F. is a Ramn y Cajal fellow.
[1] S. Sankararaman, Pericyclic Reactions—A Textbook: Reactions, Ap-
plications and Theory, Wiley, Weinheim, 2005; pp. 326 – 329, and ref-
erences therein.
[2] a) S. H nig, H. R. M ller, W. Thier, Angew. Chem. 1965, 77, 368;
Angew. Chem. Int. Ed. Engl. 1965, 4, 271; b) M. Franck-Neumann,
C. Dietrich-Buchecker, Tetrahedron Lett. 1980, 21, 671; c) E. W. Gar-
bisch, Jr., S. M. Schildcrout, D. B. Patterson, C. M. Sprecher, J. Am.
Chem. Soc. 1965, 87, 2932.
[3] This latter process was recently proposed to evolve through a cyclic
six-membered transition state at elevated temperatures. See: L. So-
minsky, E. Rozental, H. Gottlieb, A. Gedanken, S. Hoz, J. Org.
Chem. 2004, 69, 1492.
[4] a) M. T. Reetz, Angew. Chem. 1972, 84, 163; Angew. Chem. Int. Ed.
Engl. 1972, 11, 130; b) M. T. Reetz, Tetrahedron 1973, 29, 2189;
Chem. Eur. J. 2009, 15, 13022 – 13032  2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
FULL PAPER
c) M. T. Reetz, Adv. Organomet. Chem. 1977, 16, 33; d) K. N. Houk,
Y. Li, M .A. McAllister, G. A. ODoherty, L. A. Paquette, W. Sie-
brand, Z. K. Smedarchina, J. Am. Chem. Soc. 1994, 116, 10895; e) G.
Frenking, F. P. Cosso, M. A. Sierra, I. Fernndez, Eur. J. Org. Chem.
2007, 5410.
[5] R. B. Woodward, R. Hoffmann, The Conservation of Orbital Sym-
metry, VCH, Weinheim, 1970, p. 141.
[6] I. Fernndez, M. A. Sierra, F. P. Cosso, J. Org. Chem. 2007, 72,
1488.
[7] Z. Chen, C. S. Wannere, C. Corminboeuf, R. Puchta, P. von R.
Schleyer, Chem. Rev. 2005, 105, 3842.
[8] R. F. W. Bader, Atoms in Molecules -A Quantum Theory, Clarendon
Press, Oxford, 1990.
[9] a) F. M. Bickelhaupt, J. Comput. Chem. 1999, 20, 114; b) A. Diefen-
bach, F. M. Bickelhaupt, J. Chem. Phys. 2001, 115, 4030; c) A. Die-
fenbach, F. M. Bickelhaupt, J. Phys. Chem. A 2004, 108, 8460; d) A.
Diefenbach, G. T. de Jong, F. M. Bickelhaupt, J. Chem. Theory
Comput. 2005, 1, 286; e) J. N. P. van Stralen, F. M. Bickelhaupt, Or-
ganometallics 2006, 25, 4260; f) G. T. de Jong, F. M. Bickelhaupt,
ChemPhysChem 2007, 8, 1170; g) G. T. de Jong, F. M. Bickelhaupt,
J. Chem. Theory Comput. 2007, 3, 514; h) A. P. Bento, F. M. Bickel-
haupt, J. Org. Chem. 2008, 73, 7290. An early study in which the re-
actant strain or distortion energy has been separated from the inter-
action energy for understanding SN2 reactions is: i) D. J. Mitchell,
H. B. Schlegel, S. S. Shaik, S. Wolfe, Can. J. Chem. 1985, 63, 1642.
[10] I. Fernndez, G. Frenking, E. Uggerud, Chem. Eur. J. 2009, 15, 2166.
[11] a) D. H. Ess, K. N. Houk, J. Am. Chem. Soc. 2008, 130, 10187. For a
different activation strain or distortion/interaction study by the
Houk group, see, for example: b) C. Y. Legault, Y. Garcia, C. A.
Merlic, K. N. Houk, J. Am. Chem. Soc. 2007, 129, 12664.
[12] a) I. Morao, B. Lecea, F. P. Cosso, J. Org. Chem. 1997, 62, 7033;
b) F. P. Cosso, I. Morao, H. Jiao, P. von R. Schleyer, J. Am. Chem.
Soc. 1999, 121, 6737; c) I. Fernndez, M. A. Sierra, F. P. Cosso, J.
Org. Chem. 2006, 71, 6178.
[13] a) Computer code ADF 2007.01, E. J. Baerends, J. Autschbach, A.
Brces, J. A. Berger, F. M. Bickelhaupt, C. Bo, P. L. de Boeij, P. M.
Boerrigter, L. Cavallo, D. P. Chong, L. Deng, R. M. Dickson, D. E.
Ellis, M. van Faassen, L. Fan, T. H. Fischer, C. Fonseca Guerra,
S. J. A. van Gisbergen, J. A. Groeneveld, O. V. Gritsenko, M. Gr n-
ing, F. E. Harris, P. van den Hoek, C. R. Jacob, H. Jacobsen, L.
Jensen, E. S. Kadantsev, G. van Kessel, R. Klooster, F. Kootstra, E.
van Lenthe, D. A. McCormack, A. Michalak, J. Neugebauer, V. P.
Nicu, V. P. Osinga, S. Patchkovskii, P. H. T. Philipsen, D. Post, C. C.
Pye, W. Ravenek, P. Romaniello, P. Ros, P. R. T. Schipper, G.
Schreckenbach, J. Snijders, M. Sol
Velde, P. Vernooijs, L. Versluis, L. Visscher, O. Visser, F. Wang,
T. A. Wesolowski, E. M. van Wezenbeek, G. Wiesenekker, S. K.
Wolff, T. K. Woo, A. L. Yakovlev, T. Ziegler, Scientific Computing
& Modeling N.V., Amsterdam, 2007; b) C. Fonseca Guerra, J. G.
Snijders, G. te Velde, E. J. Baerends, Theor. Chem. Acc. 1998, 99,
391.
[14] a) E. J. Baerends, D. E. Ellis, P. Ros, Chem. Phys. 1973, 2, 41; b) G.
te Velde, F. M. Bickelhaupt, E. J. Baerends, C. Fonseca Guerra,
S. J. A. van Gisbergen, J. G. Snijders, T. Ziegler, J. Comput. Chem.
2001, 22, 931.
[15] E. van Lenthe, E. J. Baerends, J. G. Snijders, J. Chem. Phys. 1994,
101, 9783.
[16] a) N. C. Handy, A. J. Cohen, Mol. Phys. 2001, 99, 403; b) C. Lee, W.
Yang, R. G. Parr, Phys. Rev. B 1988, 37, 785.
[17] L. Fan, L. Versluis, T. Ziegler, E. J. Baerends, W. Ravenek, Int. J.
Quantum Chem. Quantum Chem. Symp. 1988, S22, 173.
[18] L. Fan, T. Ziegler, J. Chem. Phys. 1990, 92, 3645.
[19] K. Fukui, Acc. Chem. Res. 1981, 14, 363.
[20] a) F. M. Bickelhaupt, E. J. Baerends in Reviews in Computational
Chemistry, Vol. 15 (Eds.: K. B. Lipkowitz, D. B. Boyd), Wiley-VCH,
Weinheim, 2000, pp. 1 – 86; b) M. Lein, G. Frenking in Theory and
Applications of Computational Chemistry: The First 40 Years (Eds.:
C. E. Dykstra, G. Frenking, K. S. Kim, G. E. Scuseria), Elsevier,
Amsterdam, 2005, p. 291.
www.chemeurj.org 13031

[21] Conjugation and Hyperconjugation: a) D. Cappel, S. T llmann, A.
Krapp, G. Frenking, Angew. Chem. 2005, 117, 3683; Angew. Chem.
Int. Ed. 2005, 44, 3617; b) I. Fernndez, G. Frenking, Chem. Eur. J.
2006, 12, 3617; c) I. Fernndez, G. Frenking, J. Org. Chem. 2006, 71,
2251; d) I. Fernndez, G. Frenking, Chem. Commun. 2006, 5030;
e) I. Fernndez, G. Frenking, J. Phys. Chem. A 2007, 111, 8028; f) I.
Fernndez, G. Frenking, J. Org. Chem. 2007, 72, 7367.
[22] Aromaticity: a) I. Fernndez, G. Frenking, Faraday Discuss. 2007,
135, 403; b) I. Fernndez, G. Frenking, Chem. Eur. J. 2007, 13, 5873;
c) S. C. A. H. Pierrefixe, F. M. Bickelhaupt, Chem. Eur. J. 2007, 13,
6321; d) S. C. A. H. Pierrefixe, F. M. Bickelhaupt, J. Phys. Chem. A
2008, 112, 12816; e) S. C. A. H. Pierrefixe, F. M. Bickelhaupt, Aust.
J. Chem. 2008, 61, 209.
[23] G. Frenking, K. Wichmann, N. Frçhlich, C. Loschen, M. Lein, J.
Frunzke, V. M. Rayn, Coord. Chem. Rev. 2003, 238–239, 55.
[24] A. Krapp, F. M. Bickelhaupt, G. Frenking, Chem. Eur. J. 2006, 12,
9196.
[25] These calculations were carried out using the Gaussian 03, Revision
E.01, M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria,
M. A. Robb, J. R. Cheeseman, J. A. MontgomACHTUNGREery, 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. Peters-
son, H. Nakatsuji, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Ha-
segawa, 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. Moro-
kuma, 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.
13032 www.chemeurj.org  2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
I. Fernndez, F. M. Bickelhaupt, and F. P. Cosso
Keith, M. A. Al-Laham, C. Y. Peng, A. Nanayakkara, M. Challa-
combe, P. M. W. Gill, B. Johnson, W. Chen, M. W. Wong, C. Gonza-
lez, and J. A. Pople, Gaussian, Inc., Wallingford CT, 2004.
[26] K. Wolinski, J. F. Hilton, P. Pulay, J. Am. Chem. Soc. 1990, 112,
8251.
[27] F. Weigend, R. Alhrichs, Phys. Chem. Chem. Phys. 2005, 7, 3297.
[28] a) K. Mackenzie, J. A. K. Howard, S. Mason, E. C. Gravett, K. B.
Astin, S. X. Liu, A. S. Batsanov, D. Vlaovic, J. P. Maher, J. Chem.
Soc. Perkin Trans. 2 1993, 1211; b) K. N. Houk, J. Y. Li, M. A.
McAllister, G. A. ODoherty, L. A. Paquette, W. Siebrand, Z. K.
Smedarchinal, J. Am. Chem. Soc. 1994, 116, 10895.
[29] H. Fallah-Bagher-Shaidaei, C. S. Wannere, C. Corminboeuf, R.
Puchta, P. von R. Schleyer, Org. Lett. 2006, 8, 863.
[30] S. S. Shaik, P. C. Hiberty, A Chemists Guide to Valence Bond
Theory; Wiley, New York, 2008, Chapter 6 and references therein.
[31] a) A. Vzquez-Mayagoitia, R. Vargas, J. A. Nichols, P. Fuentealba, J.
Garza, Chem. Phys. Lett. 2006, 419, 207; b) E. H. van Veen, Chem.
Phys. Lett. 1976, 41, 540.
[32] A. Ioffe, S. Shaik, J. Chem. Soc. Perkin Trans. 2 1992, 2101.
[33] P. W. Atkins, R. S. Friedman, Molecular Quantum Mechanics, 3rd
ed., Oxford University Press, Oxford, 1997, Chapter 3.
[34] a) P. C. Hiberty, S. S. Shaik, J. M. Lefour, G. Ohanessian, J. Org.
Chem. 1985, 50, 4657; b) K. Jug, A. M. Koster, J. Am. Chem. Soc.
1990, 112, 6772; c) A. Gobbi, Y. Yamaguchi, G. Frenking, H. F.
Schaefer III, Chem. Phys. Lett. 1995, 244, 27; d) S. Shaik, A. Shurki,
D. Danovich, P. C. Hiberty, Chem. Rev. 2001, 101, 1501; e) B. Kova-
cevic, D. Baric, Z. B. Maksic, T. M ller, ChemPhysChem 2004, 5,
1352; f) A. Rehaman, A. Datta, S. S. Mallajosyula, S. K. Pati, J.
Chem. Theory Comput. 2006, 2, 30.
Received: July 21, 2009
Published online: October 22, 2009
Chem. Eur. J. 2009, 15, 13022 – 13032