Switching the Spin State of Diphenylcarbene via Halogen Bonding

Article abstract of DOI:10.1021/jacs.5b12726

The interactions between diphenylcarbene DPC and the halogen bond donors CF₃I and CF₃Br were investigated using matrix isolation spectroscopy (IR, UV-vis, and EPR) in combination with QM and QM/MM calculations. Both halogen bond donors CF₃X form very strong complexes with the singlet state of DPC, but only weakly interact with triplet DPC. This results in a switching of the spin state of DPC, the singlet complexes becoming more stable than the triplet complexes. CF₃I forms a second complex (type II) with DPC that is thermodynamically slightly more stable. Calculations predict that in this second complex the DPC?I distance is shorter than the F₃C?I distance, whereas in the first (type I) complex the DPC?I distance is, as expected, longer. CF₃Br only forms the type I complex. Upon irradiation I or Br, respectively, are transferred to the DPC carbene center and radical pairs are formed. Finally, on annealing, the formal C-X insertion product of DPC is observed. Thus, halogen bonding is a powerful new principle to control the spin state of reactive carbenes.

Full text of DOI:10.1021/jacs.5b12726

Article
pubs.acs.org/JACS
Switching the Spin State of Diphenylcarbene via Halogen Bonding
Stefan Henkel,† Paolo Costa,† Linda Klute,† Pandian Sokkar,‡ Miguel Fernandez-Oliva,‡ Walter Thiel,‡
Elsa Sanchez-Garcia,*,‡ and Wolfram Sander*,†
†Lehrstuhl fur Organische Chemie II, Ruhr-Universitat Bochum, 44801 Bochum, Germany
̈
̈
‡Max-Planck-Institut fur Kohlenforschung, 45470 Mulheim an der Ruhr, Germany
̈
̈
S
* Supporting Information
ABSTRACT: The interactions between diphenylcarbene DPC and the halogen bond
donors CF3I and CF3Br were investigated using matrix isolation spectroscopy (IR, UV−vis,
and EPR) in combination with QM and QM/MM calculations. Both halogen bond donors
CF3X form very strong complexes with the singlet state of DPC, but only weakly interact
with triplet DPC. This results in a switching of the spin state of DPC, the singlet complexes
becoming more stable than the triplet complexes. CF3I forms a second complex (type II)
with DPC that is thermodynamically slightly more stable. Calculations predict that in this
second complex the DPC···I distance is shorter than the F3C···I distance, whereas in the
first (type I) complex the DPC···I distance is, as expected, longer. CF3Br only forms the
type I complex. Upon irradiation I or Br, respectively, are transferred to the DPC carbene
center and radical pairs are formed. Finally, on annealing, the formal C−X insertion product
of DPC is observed. Thus, halogen bonding is a powerful new principle to control the spin
state of reactive carbenes.
kinetics of chemical reactions. Huber et al. demonstrated that
INTRODUCTION
■
carbon−heteroatom bonds can be activated by multidentate
halogen bonding,9 and on this basis developed efficient
organocatalysts.10 Many solvents and reagents are halogenated,
and in these cases the reactivity and selectivity of reactions
might be influenced by halogen bonds stabilizing short-lived
reaction intermediates. Here, we describe how halogen bonds
entirely can change the properties of a reactive carbene
intermediate by switching its ground state from triplet to
singlet.
Since singlet carbenes R2C: are strong Lewis bases, it is
expected that these species are stabilized by halogen bonds.
Therefore, carbene chemistry should be strongly affected by
solvents and reagents X-R′ that are able to act as halogen bond
donors in carbene complexes R2C···X-R′. An early example for
a halogen bond between a stable N-heterocyclic carbene
(NHC) as halogen bond acceptor and pentafluoroiodobenzene
as donor was published by Arduengo et al. in 1991.11 The X-ray
structure analysis of this complex 1 reveals a nearly linear C−I−
C bond (178.9°) and a C−I bond length of 2.75 Å between the
iodine atom and the carbene center. The bonding was
described as “reverse ylide” with a positively charged carbene
center and a negatively charged iodine atom. A similar complex
2, in which an iodine atom interacts with two stable
nucleophilic carbenes, was described three years later.12
Other direct experimental evidence for halogen-bonded
carbenes has not been reported. A recent theoretical study
investigates the interactions between the nucleophilic carbene 3
Halogen bonding has long been recognized as an important
principle for stabilizing intermolecular complexes DX···A,
where DX is the halogen bond donor with X = Cl, Br, or I,
and A is a Lewis base as halogen bond acceptor bearing a lone
electron pair that interacts with the X atom of the donor.1−3
The stability of halogen bonds DX···A can be similar to that of
the much better studied hydrogen bonds DH···A, and therefore
halogen bonding can be the dominant intermolecular
interaction in weakly bound aggregates. In a computational
study comparing halogen with hydrogen bonds, Wolters and
Bickelhaupt concluded that both have large covalent
components stemming from HOMO−LUMO interactions.1
In many of the examples studied, the halogen bonds are
characterized by weaker electrostatic attractions and stronger
stabilizing HOMO−LUMO interactions than the hydrogen
bonds. In line with this is the observation that the directionality
of halogen bonds is even more pronounced than that of
hydrogen bonds, strongly preferring linear orientations of DX···
A.4 Halogen atoms increase the hydrophobicity of organic
molecules, and therefore halogen bonding has been considered
as a hydrophobic equivalent of the hydrophilic hydrogen
bonding.5 The large covalent contribution to halogen bonds
renders them more resistant to increasing solvent polarity, as
recently has been outlined by Hunter et al.6
Because of these properties, which complement the tradi-
tional hydrogen bonds, halogen bonds play important roles in
molecular aggregation, in particular in supramolecular assembly
and in crystallization processes.5−8 In contrast, much less is
known about the influence of halogen bonding on reactive
intermediates and transition states, and therefore on the
Received: December 5, 2015
© XXXX American Chemical Society
A
DOI: 10.1021/jacs.5b12726
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
radical-like C−H abstractions to form a radical pair, whereas S-
6 inserts into the O−H bond either concerted or stepwise via
ionic intermediates such as ylides or ion pairs. In solid
methanol above 70 K, carbene 6 reacts to give the O−H
insertion product 8,22 whereas at temperatures below 50 K it is
unreactive even on a time scale of days. In striking contrast,
carbene 6 rapidly reacts with CH3OH in argon matrices doped
with 1% of CH3OH at temperatures as low as 25 K.15 At
temperatures between 25 and 30 K solid argon is getting soft,
and small molecules such as CH3OH are allowed to diffuse and
undergo bimolecular reactions with 6. At temperatures below
15 K argon is very rigid, and thus diffusion effectively inhibited.
These observations demonstrate that (i) single molecules of
CH3OH are highly reactive toward 6 at 25 K, whereas bulk
methanol at the same temperature does not react, and (ii)
carbene 6 exclusively reacts via its (excited) singlet state S-6
even at cryogenic temperatures.
as a model acceptor with a variety of halogen bond donors.13
Interestingly, two types of complexes were predicted, depend-
ing on the bond dissociation energy (BDE) of the D−X bond.
If the BDE is large, as in most DX molecules containing C−X
or H−X bonds, the expected classical halogen-bonded
complexes with long distances between the carbene center
and the halogen atom are predicted. The binding energies of
these type I complexes are calculated between 1.5 (CH3−Cl as
donor) and 13 kcal/mol (NC-Br as donor). On the other hand,
if the BDE is small, as in the dihalogen or interhalogen
molecules, a different structure with much larger binding
energies (e.g., 48 kcal/mol for F−Br as donor) and short
distances between the carbene center and the halogen atom is
found. These type II complexes are described as ion-pair
complexes between the D− and X-3+ charged fragments.13 The
transfer of an X+ ion from the halogen bond donor to the
acceptor therefore resembles the proton transfer in hydrogen
bond complexes.
This reactivity pattern is not limited to carbene 6 and to
CH3OH: in 1% H2O-doped argon matrices the corresponding
metastable water complex S-6···HOH is formed,16 and
fluorenylidene 7 reacts with H2O to the analogous complex
S-7···HOH.17 The switching of the ground states of carbenes 6
and 7 from triplet to singlet results from the strong hydrogen
bonds with methanol or water that stabilize the singlet states,
whereas the triplet states form only weak van der Waals
interactions. If the difference in complexation energy between
the singlet and the triplet states of a carbene with triplet ground
state is larger than the S-T energy gap, the ground state is
expected to be switched to singlet by hydrogen bonding.
Halogen bonds should be able to stabilize the singlet state S-
6 in a similar way as hydrogen bonds, and we therefore
investigated the reaction of carbene 6 with CF3I and CF3Br as
halogen bond donors. These donors were selected since they
have been shown to form strong halogen bonds with various
Lewis bases,23−26 and in addition are small enough to diffuse in
solid argon during annealing. Thus, warming a CF3I-doped
argon matrix from 8 to 25 K results in the formation of van der
Waals dimers.27 This ability to diffuse in solid argon is
prerequisite for inducing bimolecular chemistry under the
conditions of matrix isolation.
An interesting case is the interaction of NHC 3 with ClCN,
where both the type I and the type II complex were predicted
to exist as minimum structures: the conventional type I
complex 4 with a binding energy of 7.7 kcal/mol and the
unconventional type II complex 5 with a negative binding
energy of −9.9 kcal/mol (with respect to the monomers 3 and
ClCN). Complex 5 lies in a very shallow minimum and is only
metastable.
The only experimentally characterized halogen-bonded
carbene complexes are the two NHC complexes 1 and 2.11,12
In addition, a few theoretical studies on carbenes as halogen
bond acceptors were published.13,14 Halogen-bonded com-
plexes of reactive carbenes are entirely unknown, and even
hydrogen-bonded complexes of these species were rarely
studied.
RESULTS AND DISCUSSION
■
DPC−CF3I Reaction: IR Spectra. Triplet diphenylcarbene
T-6 was generated by 530 nm photolysis of diphenyldiazo-
methane 9, matrix-isolated in argon doped with 1% of CF3I at 3
K. The carbene was identified by IR, EPR and UV−vis
spectroscopy.15,16,28 Subsequent annealing at 25 K allows CF3I
to diffuse in the matrix, and after 10 min at this temperature the
IR bands of T-6 and CF3I decrease in intensity and new signals
are formed (Figure 1b). The weakly bound dimers of CF3I
show only small shifts of the IR bands compared to the
monomer29 and are easily identified. Other newly formed bands
can be assigned to a complex between singlet carbene S-6 and
CF3I by comparison with the spectra of the methanol or the
water complex of S-6.15,16 Highly characteristic in the
complexes S-6···HOH and S-6···HOCH3 is the asymmetrical
CCC stretching vibration across the carbene center at 1328
We recently demonstrated that hydrogen bonds strongly
influence the reactivity of diphenylcarbene (DPC) 615,16 and
fluorenylidene 7.17 DPC 6 is a ground state triplet carbene with
a singlet triplet gap ΔGST of −4.75 kcal/mol in isooctane and of
−3.36 kcal/mol in acetonitrile.18 Since singlet S-6 is more polar
than triplet T-6, the singlet state is better stabilized in solvents
of higher polarity, and consequently, ΔGST is getting smaller
with increasing solvent polarity. However, even in methanol
glass at cryogenic temperatures, T-6 remains more stable than
S-6, which is easily demonstrated by EPR spectroscopy.
The reaction of carbene 6 with methanol has been widely
investigated19−21 as a model reaction to demonstrate the spin-
specifity of carbene reactions: T-6 is expected to undergo
B
DOI: 10.1021/jacs.5b12726
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 1. IR difference spectra showing the reaction of diphenylcar-
bene 6 with (a) H2O (red, see ref 16), (b) CF3I (black), and (c)
CF3Br (blue). The reactions are induced by warming argon matrices
containing 6 and H2O, CF3I, or CF3Br, respectively, from 3 to 25 K.
The region between 1000 and 1200 cm−1 containing strong CF3X
vibrations is omitted for clarity (see SI for the full spectrum).
Figure 2. IR difference spectra showing the reversible thermal
interconversion between complexes 10a and 11a. (a) Difference
spectra calculated at the B97-D3/def2-TZVP level of theory; black:
10a converted to 11a, red: 11a converted to 10a. (b) Experimental
difference spectra; black: cooling the matrix from 25 to 4 K, red:
warming the matrix from 4 to 25 K. The spectra were recorded after
multiple cycles of annealing to reduce matrix site effects. Bands of the
CF3 radical are labeled with an asterisk. The region between 1000 and
1200 cm−1 containing strong CF3X vibrations is omitted for clarity
(see SI for the full spectrum).
cm−1 (Figure 1a). In the S-6···ICF3 complex 10a this vibration
is slightly blue-shifted and split into two components at 1338
and 1344 cm−1.
The IR spectrum of the S-6···ICF3 complex 10a is in good
agreement with DFT calculations at the B97-D3/def2-TZVP
level of theory. The triplet carbene complex T-6···ICF3 is
predicted to be higher in energy by 3.7 kcal/mol, and the
spectrum calculated for the triplet complex does not match the
experimental spectrum. These results clearly indicate that the
interaction between triplet carbene T-6 and CF3I results in a
strong stabilization of S-6 via halogen bonding.
thermodynamically more stable than the ’classical’ halogen-
bonded type I complex 10a.
The IR spectra taken during the thermal cycling between 4
and 25 K show that the CF3 radical29 with IR absorptions at
1242 and 1249 cm−1 is also formed (Figure 2). These bands are
formed at temperatures below 10 K and disappear upon
annealing of the matrix at 25 K. When the matrix is kept at 4 K
inside the IR spectrometer, the signals of both complexes 10a
and 11a decrease, accompanied by an increase of the bands of
CF3. This conversion can be suppressed by usage of an
interference filter, blocking light above 2000 cm−1, which
indicates that complexes 10a and 11a are very photolabile, and
irradiation by the glow bar of the IR spectrometer results in
rapid photolysis.
The IR difference spectrum (Figure 1b) indicates the
formation of several species in addition to 10a. The most
prominent additional band is a very intense signal centered at
949 cm−1. This signal is assigned to an unconventional type II
halogen-bonded complex 11a which is formally described as a
−
halogen-bonded ion pair between CF3 and the diphenyl-
(iodomethyl)cation.
CF3 was previously characterized by matrix isolation
The IR absorptions of CF3 at 1242 and 1249 cm−1 agree well
with those obtained after UV photolysis of matrix isolated CF3I
(1247 and 1252 cm−1).29 The band at 1252 cm−1 was assigned
to the “free” CF3 radical, the band at 1247 cm−1 to the CF3···I
radical pair.29 Accordingly, the two signals observed in our
experiments are assigned to free CF3 and to the Ph2C−I···CF3
radical pair 12a. A good agreement between the experimental
and theoretical spectrum is found in both cases (see SI). On
annealing the matrix at 25 K the radical pair 12a is expected to
recombine, which explains the decrease of intensity of 12a
during warming the matrix.
−
spectroscopy.30 An IR band at 778 cm−1 was assigned to the
degenerate, E symmetrical CF and FCF stretching vibration,
and a band at 1050 cm−1 to the A1 symmetrical CF3 stretching
vibration. In complex 11a, the interaction between the two
fragments results in a lowering of the symmetry and lifting of
the degeneracy of the E vibration leading to two intense,
strongly blue-shifted vibrations at 1139 and 1149 cm−1 (see
below). The A1 symmetrical CF stretching vibration is red-
shifted to 949 cm−1. A comparison of these characteristic
−
vibrations of the CF3 fragment in 11a with that of CF3 and
CF3I reveals a much closer similarity to CF3I than to CF3 ,
indicating that the CF3 fragment in 11a does not carry a large
negative charge as in the anion.
The frequency of the E symmetrical IR vibration of CF3
strongly depends on the charge and shifts from 1662 cm−1 in
−
CF3 to 1284 in the CF3 radical to 778 cm−1 in CF3 . In CF3I
this vibration is found at 1175 cm−1 which suggests a small
negative partial charge (Figure 3). Complex 10a exhibits the
corresponding vibration at 1103 cm−1, and 11a at 1139 and
1149 cm−1, quite close to CF3I. This is in line with the
calculations and suggests that there is no strongly negative
charged CF3 fragment in these complexes.
+
−
Interestingly, the intensity ratio of the IR bands of complexes
10a and 11a depends on the temperature when cycling the
temperature of the matrix between 4 and 25 K. At 4 K the IR
bands of 11a increase in intensity while those of 10a decrease
in intensity, whereas at 25 K the signals of 10a increase and
those of 11a decrease (Figure 2). The thermal interconversion
of the two complexes indicates that the equilibrium between
these complexes is rapidly established even at cryogenic
temperatures, and that the activation barrier for this process
is essentially zero. Since the yield of type II complex 11a is
higher at lower temperatures, we conclude that 11a is
After the initial annealing of matrices containing T-6 and
CF3I, the formal insertion product of carbene 6 into the CI
bond of CF3I, iodo-2,2,2-trifluoro diphenylethane 13a, is only a
minor byproduct. The formation of the metastable species 10a,
11a, and 12a demonstrates that the matrix effectively inhibits
rearrangement to 13a, despite calculations (B97-D3/def2-
C
DOI: 10.1021/jacs.5b12726
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 3. Characteristic vibrations of the CF3 fragment. The
experimental frequencies of the degenerate, E symmetrical CF and
FCF stretching vibrations (black) and of the A1 symmetrical CF3
stretching vibration (red) strongly depend on the charge of the CF3
Figure 4. UV−vis spectra showing the reaction of T-6 with CF3I.
Dashed black line: Spectrum of diphenyldiazomethane 9 in argon
doped with 1% of CF3I at 10 K. Solid black line: Spectrum obtained
after irradiation of the matrix with 530 nm light. Solid red line:
Spectrum obtained after subsequent annealing of the matrix to 25 K
for 10 min. Dashed red line: Spectrum after irradiation of the annealed
matrix with 650 nm light. The cusp at 800 nm is due a grating change
of the UV−vis spectrometer.
30
30
−
group: cation CF3 , radical CF3,29 neutral CF3I, and anion CF3
+
(upper four traces) are compared to 12a, 10a and 11a (three lower
traces).
TZVP) predict the formation of 13a from T-6 and CF3I to be
strongly exothermic by 67 kcal/mol. However, prolonged
irradiation (450 nm) results in a decrease of the signals of
carbene 6, remaining 9, complexes 10a and 11a, and radical
pair 12a and formation of the final product 13a (Scheme 1).
The observed IR bands fit reasonably well to the calculated
spectrum of 13a.
DPC−CF3I Reaction: EPR Spectra. Photolysis of diazo
precursor 9 in 1% CF3I-doped Ar at 5 K produces the
characteristic signals of T-6 and a number of signals around
3400 G typical of radicals (Figure 5). The CF3 radical is
Scheme 1. Reaction of Diphenylcarbene 6 with ICF3
DPC−CF3I Reaction: UV−Vis Spectra. The formation of
10a from T-6 and CF3I could also be monitored by UV−vis
spectroscopy (Figure 4). Upon annealing of an argon matrix
containing both carbene T-6 and CF3I, the characteristic weak
visible absorption of T-6 at 458 nm31 decreases, and
simultaneously a UV absorption at 372 nm and a very broad
absorption in the visible region centered around 930 nm appear
(Figure 4). The singlet carbene complexes S-6···HOR (R = H,
CH3) exhibit a UV absorption at 360 nm assigned to singlet
carbene S-6,15,16 and therefore we assign the band at 372 nm to
10a.
Multistate CASSCF(12,12) calculations predict a near IR
transition for type 1 complex 10a at 944 nm and for the type II
complex 11a at 784 nm (see Supporting Information for the
computational details), both corresponding to HOMO−
LUMO type transitions. On this basis we assign the band at
930 nm to type I complex 10a. Upon photolysis of the matrix
with red light (650 nm) the bands assigned to 10a decrease in
intensity, in accordance with the observation in the IR
experiments.
Figure 5. EPR spectra showing the reaction of triplet diphenylcarbene
6 with CF3I. Red line: Spectrum of 6 in argon doped with 1% of CF3I
at 5 K. Solid black line: Spectrum of the same matrix after annealing to
25 K for 10 min. Dashed black line: Spectrum obtained after 650 nm
irradiation. Inset shows the radical region together with a simulated
spectrum of CF3 radicals (see SI for details).
identified by its anisotropic hyperfine coupling pattern that was
previously reported in literature.32 Warming of the matrix from
5 to 25 K reduces the intensity of T-6, as expected if the EPR
quiet singlet complexes 10a and 11a are formed. At the same
time, the radical signals disappear, which is attributed to
recombination as a result of the higher mobility of the radicals
at higher temperatures. Subsequent irradiation of the matrix at
5 K with 650 nm light does not affect the signals of the triplet
carbene, but results in the formation of an intense radical signal,
consistent with the conversion of 10a and 11a to radical pair
12a as observed in the IR experiments. The broad feature
underlying the CF3 signals is tentatively assigned to the Ph2CI
radical.
D
DOI: 10.1021/jacs.5b12726
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
DPC−CF3Br Reaction. In an effort to explore the influence
of the halogen atom on the halogen bond to carbene 6, we also
investigated the reaction with CF3Br. If a matrix containing T-6
and 0.5% of CF3Br is annealed at 25 K, the resulting IR
difference spectrum reveals the formation of S-6···BrCF3
complex 10b with a very similar IR spectrum as 10a (Figure
2c, Scheme 2).
Scheme 2. Reaction of Diphenylcarbene 6 with BrCF3
Figure 6. Relative energies and some geometric parameters of singlet
and triplet 6 and the corresponding halogen bonded complexes with
CF3I T-10a and S-10a in the gas phase and in an explicit argon matrix,
computed at the B97-D3/def2-TZVP and B97-D3/def2-TZVP//
CHARMM levels of theory. All energies are in kcal/mol and corrected
for ZPE, gas phase energies also for BSSE. The green and red bars
represent the range of QM/MM values within several snapshots.
the interaction energies of DPC complexes with methanol,
CF3I and CF3Br). When comparing the CH3OH complexes of
S- and T-6 with the XCF3 complexes, similar stabilization
energies for the interactions with triplet T-6 are predicted,
suggesting that halogen bond (with iodine) and hydrogen bond
to the triplet carbene show similar strengths. The situation is
different for the interaction with S-6, where the halogen bond is
about 5 kcal/mol stronger than the hydrogen bond. This can be
rationalized by comparing the molecular orbitals involved in
halogen and hydrogen bonding (Figure S11, SI). As expected,
the main interaction between S-6 and CF3I is found between
the HOMO of S-6 (the nonbonding orbital at the carbene
center) and the LUMO of CF3I (the σ* orbital). The
corresponding interaction is also dominating the hydrogen
bond in the S-6···HOCH3 complex. For the halogen bond,
back-donation from a lone pair at the iodine atom into the
empty p orbital at the carbene center results in an additional
stabilization of the halogen bond, which is lacking in the
hydrogen bond. This type of interaction is not possible in T-6,
where the p orbital at the carbene center contains an unpaired
electron. This can also be illustrated using NBO theory (see
SI).
When comparing the halogen bonding in 10a with that in
10b, the calculations predict a stronger interaction of the
carbene with CF3I than with CF3Br. This is in line with the
lower polarizability of the bromine atom compared to iodine,
which generally results in weaker halogen bond complexes with
bromine as donor than with iodine as donor. This can also be
rationalized by the two orbital interactions discussed above. For
the first type of interaction, the lower polarizability of bromine
causes a smaller positive partial charge at Br in direction of the
Br−C axis, leading to a weaker lone pair-σ* interaction. In
addition, since bromine is more electronegative than iodine, the
back-donation will also be less favorable, leading to an overall
weaker halogen bond with CF3Br.
In 10b the characteristic CCC stretching vibration of S-6 is
split into two components at 1330 and 1338 cm−1, slightly red-
shifted compared to 10a. Interestingly, with CF3Br a type II
complex 11b is not observed, as can be seen by the absence of
an intense vibration around 949 cm−1. Photolysis of 10b with
650 nm light leads to formation of CF3 radicals and the Ph2C−
Br···CF3 radical pair 12b, in analogy to the photochemistry of
10a. The radical pair 12b is prevented from recombination only
by the surrounding rigid matrix cage. Annealing of the matrix at
temperatures above 25 K results in the rapid formation of
bromo-2,2,2-trifluoro diphenylethane 13b, which was identified
by comparison of its IR spectrum with that of an authentic
matrix-isolated sample (see SI).
DFT, MCSCF and DFT-D/MM Calculations. At the B97-
D3/def2-TZVP level of theory, 6 is predicted to have a triplet
ground state lying 6.1 kcal/mol below the lowest singlet state.
The singlet carbene interacts strongly with CF3I, and the
calculated interaction energy of the resulting halogen bonded
complex S-10a is −15.8 kcal/mol. In contrast, the interaction of
the triplet carbene with CF3I is much weaker, and the
stabilization energy of T-10a is only −6.0 kcal/mol. Thus,
the stabilization of the singlet carbene by halogen bonding is
9.8 kcal/mol larger than that of the triplet carbene, and this
stabilization is larger than the singlet−triplet splitting of 6 of 6.1
kcal/mol. Consequently, the singlet state of the complex is
lying 3.7 kcal/mol below the triplet state. This spin inversion is
also found by QM/MM (B97-D3/def2-TZVP//CHARMM)
calculations in explicit argon matrices. There, the S-T gap is
inverted from 5.3 kcal/mol to −4.1 kcal/mol upon formation of
the halogen bonded complexes (Figure 6). Similar changes in
the S-T gap are also predicted by DFT calculations with other
functionals; for example, from 6.8 to 0.4 kcal/mol with M06-
2X/def2-TZVP and from 3.3 to −7.5 kcal/mol with BLYP-D3/
def2-TZVP (Table S22, SI).
With CF3Br, singlet carbene S-6 forms a halogen-bonded
complex with a stabilization energy of −10.7 kcal/mol (S-10b),
while T-6 forms a weaker complex with a stabilization energy of
only −4.1 kcal/mol (T-10b). Again, the singlet complex S-10b
is calculated to be more stable than the triplet complex T-10b,
however, by only 0.5 kcal/mol (Figure S10, SI).
The computed gas-phase spectra of the type II complex 11a
nicely match the experimental results, although small imaginary
modes indicate that the structure might not be a minimum in
the gas phase. The computed dipole moments of 9.14 D for
10a and 6.62 D for 11a show that 11a is of lower polarity than
the classical halogen bond complex, suggesting that it is
described better as a reversed halogen bond complex rather
than an ion-pair.
This behavior is analogous to the one observed for the
interaction of 6 with methanol and water (see Table S8, SI for
E
DOI: 10.1021/jacs.5b12726
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Although for the type II complex in the gas phase a minimum
could not be definitively located, the QM/MM calculations
reveal that in solid argon it is clearly a minimum. There, the
type II structure is 0.5 kcal/mol higher in energy than the type I
complex. Calculations of the reaction path for the inter-
conversion of the singlet complexes 10a and 11a (B97-D3/
def2-TZVP) reveal a transition state connecting both structures
lying only 0.6 kcal/mol above the type I complex 10a (Figure
S12, SI). However, the potential energy surface in this region is
very flat, and the calculations indicate nearly degenerate
structures, in agreement with the experimental results.
Multiconfigurational SCF (MCSCF) approaches have been
reported to produce reliable S-T energy gaps in various
systems.33−35 Therefore, we used MCSCF calculations to
compute the S-T energy gap and to investigate the intersystem
crossing in the reaction of 6 and CF3I. CASSCF(8,8) MP2
(referred to as CASMP2) single point calculations were
performed on the DFT-optimized geometries of S-10a and
T-10a. The 6-31G* basis set was assigned to the C, H and F
atoms and the LANL2DZ basis set was used for the iodine
atom. As indicated above, the active space was formed by 8
electrons and 8 orbitals (Figure S13, SI, for more active space
details see SI). Potential energy scans were performed for S-10a
and T-10a along the C−I distance at the DFT level and single-
point CASMP2 energies were calculated (Figure 7A). The S-T
energy gap obtained from the CASMP2 calculations is −0.9
kcal/mol for 10a, whereas the S-T energy gap of 6 was found to
be 9.2 kcal/mol, at the same level of theory. These results
support the inversion of the S-T energy gap obtained at the
DFT level. The singlet and triplet potential energy surfaces
were found to cross each other at the C−I distance of 3.25 Å.
Additionally, we calculated CASMP2 energies for the Franck−
Condon triplet state with the nuclear configuration of the
singlet state (FC triplet). The corresponding curve crosses the
singlet potential energy surface at 3.90 Å, indicating that the
intersystem crossing may occur when the C−I distance is
between 3.25 and 3.90 Å. Furthermore, the spin−orbit coupling
between the singlet and triplet states calculated at this crossing
point is 55.1 cm−1 (CASSCF(8,8) level of theory). This strong
spin−orbit coupling (due to the presence of a heavy atom, I)
may facilitate the intersystem crossing.
The CASMP2 calculations predict that 11a is more stable
than S-10a by 2.0 kcal/mol (Figure 7A). For a more precise
estimation of this energy difference, we performed the potential
energy scan with a finer reaction coordinate window (0.01 Å),
between 2.1 and 2.8 Å. As shown in Figure S15, the two
minima (S-10a and 11a) were found to be equally stable,
however with a barrier separation of 1 kcal/mol.
We also performed a 2-dimensional potential energy scan
(with the halogen bond distance and angle as reaction
coordinates) to explore the relative stability of S-10a and
11a. Interestingly, 11a (which is characterized by a C−I
distance of 2.3 Å) exhibits stability over a range of C−I−C
angles and S-10a (characterized by a nearly linear C−I−C
angle) is stable over a range of C−I distances (dC−I, Figure 7B).
These calculations also corroborate that 11a is slightly more
stable than S-10a, and support the presence of a small barrier
between these structures.
Additionally, we constructed a 2-dimensional potential
energy surface at the CASMP2//DFT level, as a function of
the distance dC−I and the carbene angle (C−C−C). For this
purpose, we performed series of geometry optimizations for a
range of C−C−C angles between 120°−140° (resolution of
2°) and dC−I distances between 2.5 and 4.0 Å (resolution of
0.02 Å). We found that the S-10a and T-10a potential energy
surfaces cross each other along a “seam” (where S and T are
degenerate),36,37 which has a sloped topology (Figure 8). The
reaction pathway outlined in Scheme 1 can be described as
follows: triplet carbene T-6 forms a relatively weak complex T-
10a upon encountering CF3I, which subsequently undergoes
intersystem crossing to form S-10a. The minimum energy
crossing point for this pathway is calculated to be at ∼4 kcal/
mol (Figure 8, θC−C−C = 125°, dC−I = 3.2 Å). Alternatively, T-
10a may avoid crossing and form 12a upon photoirradiation.
To explore if the type II complex 11a is thermally accessible
from S-10a, QM MD and QM/MM MD simulations were
performed in the gas phase and in argon matrix. The singlet
complexes 10a and 11a were used as starting structures, and
the simulations were performed at temperatures of 3, 25, 50,
and 75 K. In these simulations, all Ar atoms with a distance
larger than 12 Å from the carbene center were kept frozen,
effectively preventing the argon matrix to evaporate, even at
elevated temperatures. In the simulations at 3, 25, and 50 K
(both gas phase and argon matrix), the system remained in the
initial conformation during the whole length of the simulations
(10 ps), suggesting that at these temperatures thermal
interconversion does not take place on a time scale of
picoseconds. Interestingly, the simulation in the explicit argon
matrix at 75 K exhibited a fast and reversible conversion of S-
10a to 11a (Figure 9).
Figure 7. (A) Crossing point between singlet and triplet potential
energy surfaces of 10a. The potential energy surface was obtained by
performing constrained optimizations along the C−I distance. The S-
T energy gaps for 10a, and locations of 11a and 12a are also indicated
in the figure. (B) 2D-potential energy scan, using the halogen bonding
distance and angle as reaction coordinates. The potential energy
surfaces were constructed by single point CASMP2 calculations on the
DFT-optimized geometries.
F
DOI: 10.1021/jacs.5b12726
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
CONCLUSION
■
Halogen bonding not only influences the thermodynamic
stability and structure of stable molecular aggregates, but can
also drastically change the properties of reactive intermediates
such as carbenes. The results from IR, UV−vis, and EPR
experiments in combination with calculations at various levels
of theory allow us to draw a clear picture of how halogen
bonding changes the properties of diphenylcarbene 6:
(1) The closed shell singlet state of DPC (S-6) is an
excellent halogen bond acceptor and forms very strong
halogen bonds with both CF3I and CF3Br as halogen
bond donors. However, since DPC has a triplet ground
state (T-6), intersystem crossing is required concurrently
with the formation of the halogen bond to form the
singlet carbene complexes Ph2C···X-CF3 10a and 11a,
respectively.
(2) With both CF3I and CF3Br the halogen-bonded singlet
S-6 becomes the ground state, which clearly demon-
strates that with these donors the halogen bond
stabilization of S-6 compared to that of T-6 is larger
than the singlet−triplet gap of 6. This is in agreement
with DFT calculations which, as expected, predict for
CF3I a stronger halogen bond than for CF3Br. For CF3Br
the singlet and triplet complexes are predicted to be
almost degenerate.
Figure 8. Potential energy scan for 10a as a function of the carbene
angle and the C···I distance showing the intersection seam between
the singlet and triplet surface. The potential energy scan was
performed by a series of single point CASSCF(8,8)MP2 calculations
on the DFT-optimized geometries.
(3) The IR and UV−vis spectra of the complexes Ph2C···X-
CF3 10a and 11a clearly demonstrate that the singlet
state S-6 and not the triplet state T-6 is involved in
halogen bonding. A comparison with the hydrogen-
bonded complexes15,16 Ph2C···H−OH and Ph2C···H−
OCH3 suggests that, despite the close similarity of the
spectra of S-6, the interaction with X-CF3 is stronger
than that with H-OR. The stronger interaction results in
a noticeable blue-shift and splitting into two components
of the asymmetrical CCC stretching vibration involving
the carbene center. This is in line with the calculated
higher binding energies of the halogen-bonded com-
plexes compared to the hydrogen-bonded complexes.
(4) The most striking difference between CF3I and CF3Br as
halogen bond donors is that CF3I leads to two
energetically almost degenerate complexes 10a and 11a
that easily interconvert, whereas CF3Br forms only one
complex 10b with carbene S-6. We classify complexes
10a and 11a as conventional type I and nonconventional
type II complex, respectively, as suggested by Donoso-
Tauda et al. for a series of halogen-bonded NHC
complexes.13 According to these authors, type I
complexes are formed if the R-X bond dissociation
energy of the halogen bond donor is larger than 70 kcal/
mol, whereas type II complexes are more likely if the
BDE is below 60 kcal/mol. The BDE of CF3I was
determined to 52 kcal/mol,38 which is below the 60 kcal/
mol threshold. Since S-6 is a stronger base39,40 and a
stronger halogen-bond acceptor than 3, the abstraction
of a halogen atom from R-X is thermodynamically even
more favorable. Obviously, with 71 kcal/mol41 the BDE
of the C−Br bond in CF3Br is too large for the formation
of the type II complex 11b.
Figure 9. Snapshots from the QM MD and QM/MM MD simulations
started from the type I complex (S-10a) at 75 K, indicate that both
types of halogen bonded complexes are formed. (a) QM gas phase
simulation at the B97-D3/def2-TZVP level (b) QM/MM simulation
at the B97-D3/def2-TZVP//CHARMM level.
Additional QM MD simulations were performed at higher
temperatures (100, 200, and 300 K). At these temperatures the
gas phase simulations indicate a rapid interconversion between
type I and type II complexes, resulting in both structures being
populated independently of the starting geometry. Although
these temperatures are unrealistic in the matrix, QM/MM MD
simulations in argon were also performed at these temper-
atures. The interconversion between type I and type II
complexes was also found under these conditions. This
dynamic behavior is consistent with energetically nearly
degenerate structures connected by a very small activation
barrier, in agreement with the experiments.
Complexes 10a and 11a are in thermodynamic
equilibrium even at the lowest matrix temperature,
which implies an extremely low activation barrier for
the interconversion. The experiments also indicate that
G
DOI: 10.1021/jacs.5b12726
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
CHARMM level. In both cases, 10 ps MD simulations were performed
with a time step of 2 fs under NVT (canonical) conditions at
temperatures of 3, 25, 50, 75, 100, 200, and 300 K. A Nose−Hoover
́
photolysis of the matrix shifts the equilibrium toward
complex 10a. The nonclassical complex 11a is slightly
more stable according to the experiment. The DFT
calculations predict 10a to be more stable by
approximately 0.5 kcal/mol, but this small difference is
beyond the accuracy of these calculations. Furthermore,
our multireference CASSCF(8,8) MP2 calculations
correctly predict 11a to be more stable than 10a by 2
kcal/mol (Figure 7B).
chain thermostat56,57 was used together with a reversible noniterative
leapfrog-type integrator. In the setup for the MD simulations, the gas
phase optimized structures of the DPC - CF3I complexes were used as
starting geometry for the QM region, which was then placed in a pre-
equilibrated box of argon with 15 Å padding. The surrounding MM
atoms were equilibrated for 5 ns under NPT conditions in the
presence of a frozen QM region. Afterward, production QM/MM
MDs simulations were performed in which the QM region and all MM
atoms within 10 Å of the QM region were allowed to freely move. Ten
snapshots were taken from each of the QM/MM MD simulations and
reoptimized at the B97-D3/def2-TZVP//CHARMM level of theory.
All atoms within the active region were allowed to move in each
optimization step. The optimization was finished when the maximum
gradient component was below 0.00045 au
The two complexes 10a and 11a differ mainly in the
position of the iodine atom, and the rearrangement
formally corresponds to a transfer of an I+ ion, similar to
a proton transfer in a hydrogen bonded system. In
addition to ion transfer, halogen bonded systems can also
undergo electron transfer to produce radicals, which adds
another dimension of complexity to their chemistry.
(5) The thermal reaction of triplet DPC T-6 with CF3X to
give the final C-X insertion products 13 is a highly
complex reaction sequence involving three subsequent
intersystem crossing steps. Thus, the triplet carbene T-6
reacts to the singlet complex 10 (and in the case of CF3I
also to the second singlet complex 11a) that forms the
triplet radical pair 12, which finally recombines to give
the singlet insertion product 13 (Schemes 1 and 2). The
CASSCF(8,8)MP2 calculations for the reaction with
CF3I provide detailed insight into these processes, and in
particular rationalize the smooth formation of complex
10a.
(6) Interestingly, our experiments do not give any evidence
for the formation of an ylide or ylidic complex42 with
CF3X acting as Lewis base and the carbene S-6 with its
vacant p orbital acting as Lewis acid. Ylides formed
between singlet carbenes and ethers,43,44 alcohols,45
pyridine,46,47 and other reagents bearing lone pairs are
frequently observed as intermediates. Obviously, halogen
bonding is energetically favored over ylide formation in
the reactions investigated here. It remains an interesting
challenge for future research to find carbenes where ylide
formation is preferred over halogen bonding.
The GAUSSIAN09 suite of programs was used for the CASSCF and
CASMP2 calculations.58,59 The structures of S-10a and T-10a were
optimized at the B97-D3/def2-TZVP level. The potential energy scan
was performed along the carbene center-I distance (dC−I), in the range
from 2 to 6 Å with the step size of 0.05 Å. At each point, the positions
of C and I were fixed, and the rest of the structure was optimized.
These DFT optimized geometries were used for CASSCF and
CASMP2 single point calculations (see SI). For all the CAS
calculations, the 6-31G* basis set was used for C, H and F atoms
and the LANL2DZ basis set and Effective Core Potential (ECP) were
used for the I atom.60 Only ground state CAS calculations were
performed. The configurations having CI coefficients larger than 0.1
were used as reference states for Møller−Plesset perturbation theory
(MP2) level electron correlation.61,62 Spin−orbit coupling was
calculated at the intersection of the singlet and triplet potential
energy surfaces.63,64
ASSOCIATED CONTENT
* Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/jacs.5b12726.
■
S
Spectroscopic and computational data. (PDF)
AUTHOR INFORMATION
■
Corresponding Authors
*esanchez@kofo.mpg.de
*wolfram.sander@rub.de
In summary, halogen bonding to reactive carbenes is an
effective tool to control the spin state and the reactivity of
carbenes. The work presented here suggests that halogen
bonding has to be considered as an important mechanism for
the stabilization of reactive carbenes that might rival the long
discussed formation of ylidic complexes, and that can surpass
the stability of hydrogen-bonded complexes.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
This work was supported by the Cluster of Excellence RESOLV
(EXC 1069) funded by the Deutsche Forschungsgemeinschaft
(DFG). P.S. and E.S.-G thank Rachel Crespo-Otero for useful
discussions. E.S.-G acknowledges the support from the
Boehringer Ingelheim Foundation (Plus-3 Program) and a
Liebig-stipend from the Fonds der Chemischen Industrie, as
well as the support of the Collaborative Research Center
EXPERIMENTAL SECTION
■
General Procedures. Preparation of diphenyldiazomethane 9 and
matrix isolation measurements were carried out according to
procedures described in preceding work.15,16 Matrices were doped
with ICF3 (Fluorochem) and BrCF3 (Aldrich Chemical Co.) as
commercially available.
Computational Details. All gas phase geometry optimizations
and frequency calculations were performed using the B9748 functional
with D3 dispersion corrections49 as implemented in the program
Turbomole (version 6.4).50 The def2-TZVP51 basis set was used. NBO
analysis was performed using the NBO version 6 program.52
QM MD, QM/MM MD simulations and QM/MM optimizations
were performed using the program ChemShell53,54 as interface to
Turbomole for the QM region (DPC and CF3I) and CHARMM
31b155 for the MM region (argon matrix). QM MD simulations were
conducted at the B97-D3/def2-SVP level of theory while QM/MM
MD simulations were carried out at the B97-D3/def2-SVP//
̈
SFB1093, funded by the DFG. We thank Nesli Ozkan for
performing some of the CF3Br experiments and Stefan Huber
for introducing us to halogen bonded systems.
REFERENCES
■
(1) Wolters, L. P.; Bickelhaupt, F. M. ChemistryOpen 2012, 1, 96.
(2) Tawfik, M.; Donald, K. J. J. Phys. Chem. A 2014, 118, 10090.
(3) Desiraju, G. R.; Ho, P. S.; Kloo, L.; Legon, A. C.; Marquardt, R.;
Metrangolo, P.; Politzer, P.; Resnati, G.; Rissanen, K. Pure Appl. Chem.
2013, 85, 1711.
H
DOI: 10.1021/jacs.5b12726
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(4) Huber, S. M.; Scanlon, J. D.; Jimenez-Izal, E.; Ugalde, J. M.;
Infante, I. Phys. Chem. Chem. Phys. 2013, 15, 10350.
(5) Priimagi, A.; Cavallo, G.; Metrangolo, P.; Resnati, G. Acc. Chem.
Res. 2013, 46, 2686.
(6) Robertson, C. C.; Perutz, R. N.; Brammer, L.; Hunter, C. A.
Chem. Sci. 2014, 5, 4179.
(7) Meyer, F.; Dubois, P. CrystEngComm 2013, 15, 3058.
(8) Jungbauer, S. H.; Bulfield, D.; Kniep, F.; Lehmann, C. W.;
Herdtweck, E.; Huber, S. M. J. Am. Chem. Soc. 2014, 136, 16740.
(9) Walter, S. M.; Kniep, F.; Herdtweck, E.; Huber, S. M. Angew.
Chem., Int. Ed. 2011, 50, 7187.
(10) Kniep, F.; Jungbauer, S. H.; Zhang, Q.; Walter, S. M.; Schindler,
S.; Schnapperelle, I.; Herdtweck, E.; Huber, S. M. Angew. Chem., Int.
Ed. 2013, 52, 7028.
(44) Hoijemberg, P. A.; Moss, R. A.; Krogh-Jespersen, K. J. Phys.
Chem. A 2012, 116, 358.
(45) Xue, J.; Luk, H. L.; Platz, M. S. J. Am. Chem. Soc. 2011, 133,
1763.
(46) Jones, M. B.; Platz, M. S. J. Org. Chem. 1991, 56, 1694.
(47) Admasu, A.; Gudmundsdottir, A. D.; Platz, M. S. J. Phys. Chem.
A 1997, 101, 3832.
(48) Grimme, S. J. Comput. Chem. 2006, 27, 1787.
(49) Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. J. Chem. Phys.
2010, 132, 154104.
(50) TURBOMOLE V6.4 2012; TURBOMOLE GmbH, 2007; www.
turbomole.com.
(51) Weigend, F.; Haser, M.; Patzelt, H.; Ahlrichs, R. Chem. Phys.
̈
Lett. 1998, 294, 143.
(52) NBO 6.0; Glendening, E. D.; Badenhoop, J. K.; Reed, A. E.;
Carpenter, J. E.; Bohmann, J. A.; Morales, C. M.; Landis, C. R.;
Weinhold, F.; Theoretical Chemistry Institute, University of
Wisconsin: Madison, WI, 2013.
(11) Arduengo, A. J., III; Kline, M.; Calabrese, J. C.; Davidson, F. J.
Am. Chem. Soc. 1991, 113, 9704.
(12) Arduengo, A. J., III; Tamm, M.; Calabrese, J. C. J. Am. Chem.
Soc. 1994, 116, 3625.
(13) Donoso-Tauda, O.; Jaque, P.; Elguero, J.; Alkorta, I. J. Phys.
Chem. A 2014, 118, 9552.
(14) Lv, H.; Zhuo, H.-Y.; Li, Q.-Z.; Yang, X.; Li, W.-Z.; Cheng, J.-B.
Mol. Phys. 2014, 112, 3024.
(15) Costa, P.; Sander, W. Angew. Chem., Int. Ed. 2014, 53, 5122.
(16) Costa, P.; Fernandez-Oliva, M.; Sanchez-Garcia, E.; Sander, W.
J. Am. Chem. Soc. 2014, 136, 15625.
(17) Costa, P.; Trosien, I.; Fernandez-Oliva, M.; Sanchez-Garcia, E.;
Sander, W. Angew. Chem., Int. Ed. 2015, 54, 2656.
(18) Eisenthal, K. B.; Turro, N. J.; Sitzmann, E. V.; Gould, I. R.;
Hefferon, G.; Langan, J.; Cha, Y. Tetrahedron 1985, 41, 1543.
(19) Eisenthal, K. B.; Moss, R. A.; Turro, N. J. Science 1984, 225,
1439.
(53) ChemShell, a Computational Chemistry Shell; www.chemshell.org.
(54) Metz, S.; Kastner, J.; Sokol, A.; Keal, T. W.; Sherwood, P. Wiley
̈
Interdiscip. Rev. Comput. Mol. Sci. 2014, 4, 101.
(55) Brooks, B. R.; Brooks, C. L.; Mackerell, A. D.; Nilsson, L.;
Petrella, R. J.; Roux, B.; Won, Y.; Archontis, G.; Bartels, C.; Boresch,
S.; Caflisch, A.; Caves, L.; Cui, Q.; Dinner, A. R.; Feig, M.; Fischer, S.;
Gao, J.; Hodoscek, M.; Im, W.; Kuczera, K.; Lazaridis, T.; Ma, J.;
Ovchinnikov, V.; Paci, E.; Pastor, R. W.; Post, C. B.; Pu, J. Z.; Schaefer,
M.; Tidor, B.; Venable, R. M.; Woodcock, H. L.; Wu, X.; Yang, W.;
York, D. M.; Karplus, M. J. Comput. Chem. 2009, 30, 1545.
́
(56) Nose, S. J. Chem. Phys. 1984, 81, 511.
(57) Hoover, W. Phys. Rev. A: At., Mol., Opt. Phys. 1985, 31, 1695.
(58) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.;
Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci,
B.; Petersson, G. A.; Nakatsuji, H.; Caricato, M.; Li, X.; Hratchian, H.
P.; Izmaylov, A. F.; Bloino, J.; Zheng, G.; Sonnenberg, J. L.; Hada, M.;
Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima,
T.; Honda, Y.; Kitao, O.; Nakai, H.; Vreven, T.; Montgomery, J. A.;
Peralta, J. E.; Ogliaro, F.; Bearpark, M.; Heyd, J. J.; Brothers, E.; Kudin,
K. N.; Staroverov, V. N.; Kobayashi, R.; Normand, J.; Raghavachari, K.;
Rendell, A.; Burant, J. C.; Iyengar, S. S.; Tomasi, J.; Cossi, M.; Rega,
N.; Millam, J. M.; Klene, M.; Knox, J. E.; Cross, J. B.; Bakken, V.;
Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.;
Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Martin, R. L.;
Morokuma, K.; Zakrzewski, V. G.; Voth, G. A.; Salvador, P.;
(20) Griller, D.; Nazran, A. S.; Scaiano, J. C. J. Am. Chem. Soc. 1984,
106, 198.
(21) Wright, B. B.; Platz, M. S. J. Am. Chem. Soc. 1984, 106, 4175.
(22) Leyva, E.; Barcus, R. L.; Platz, M. S. J. Am. Chem. Soc. 1986, 108,
7786.
(23) Herrebout, W. Top. Curr. Chem. 2014, 358, 79.
(24) Nagels, N.; Herrebout, W. A. Spectrochim. Acta, Part A 2015,
136, 16.
(25) Hauchecorne, D.; Herrebout, W. A. J. Phys. Chem. A 2013, 117,
11548.
(26) Romaniello, P.; Lelj, F. J. Phys. Chem. A 2002, 106, 9114.
(27) Ito, F.; Hirabayashi, S. J. Chem. Phys. 2006, 124, 234509-1.
(28) Sander, W. W. J. Org. Chem. 1989, 54, 333.
(29) Clemitshaw, K. C.; Sodeau, J. R. J. Phys. Chem. 1988, 92, 5491.
(30) Forney, D.; Jacox, M. E.; Irikura, K. K. J. Chem. Phys. 1994, 101,
8290.
(31) Gibbons, W. A.; Trozzolo, A. M. J. Am. Chem. Soc. 1966, 88,
172.
(32) Edlund, O.; Lund, A.; Shiotani, M.; Sohma, J.; Thuomas, K. A.
Mol. Phys. 1976, 32, 49.
(33) Cui, G.; Fang, W. J. Chem. Phys. 2013, 138, 044315.
̈
Dannenberg, J. J.; Dapprich, S.; Daniels, A. D.; Farkas, O.;
Foresman, J. B.; Ortiz, J. V.; Cioslowski, J.; Fox, D. J. Gaussian 09,
Revision D.01; Gaussian Inc.: Wallingford, CT, 2009.
(59) Yamamoto, N.; Vreven, T.; Robb, M. A.; Frisch, M. J.; Schlegel,
H. B. Chem. Phys. Lett. 1996, 250, 373.
(60) Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 299.
(61) Head-Gordon, M.; Pople, J. A.; Frisch, M. J. Chem. Phys. Lett.
1988, 153, 503.
(62) McDouall, J. J. W.; Peasley, K.; Robb, M. A. Chem. Phys. Lett.
1988, 148, 183.
(63) Abegg, P. W. Mol. Phys. 1975, 30, 579.
(64) Marian, C. M. Wiley Interdiscip. Rev. Comput. Mol. Sci. 2012, 2,
187.
(34) Mai, S.; Marquetan, P.; Gonzal
9524.
́
ez, L. J. Phys. Chem. A 2015, 119,
(35) Uzunova, E. L. J. Phys. Chem. A 2009, 113, 11266.
(36) Paterson, M. J.; Robb, M. A.; Blancafort, L.; DeBellis, A. D. J.
Phys. Chem. A 2005, 109, 7527.
(37) Bernardi, F.; Olivucci, M.; Robb, M. A. Chem. Soc. Rev. 1996, 25,
321.
(38) Okafo, E. N.; Whittle, E. Int. J. Chem. Kinet. 1975, 7, 273.
(39) Keeffe, J. R.; O’Ferrall, R. A. M. ARKIVOC (Gainesville, FL, U.
S.) 2008, 183.
(40) Pliego, J. R.; DeAlmeida, W. B. J. Chem. Soc., Faraday Trans.
1997, 93, 1881.
(41) Ferguson, K. C.; Whittle, E. J. Chem. Soc., Faraday Trans. 1
1972, 68, 295.
(42) Moss, R. A. J. Phys. Org. Chem. 2011, 24, 866.
(43) Moss, R. A.; Wang, L.; Odorisio, C. M.; Zhang, M.; Krogh-
Jespersen, K. J. Phys. Chem. A 2010, 114, 209.
I
DOI: 10.1021/jacs.5b12726
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX