Photodissociation Dynamics of Cyclopropenylidene, c-C3H2

Article abstract of DOI:10.1002/chem.201501624

In this joint experimental and theoretical study we characterize the complete dynamical "life cycle" associated with the photoexcitation of the singlet carbene cyclopropenylidene to the lowest lying optically bright excited electronic state: from the initial creation of an excited-state wavepacket to the ultimate fragmentation of the molecule on the vibrationally hot ground electronic state. Cyclopropenylidene is prepared in this work using an improved synthetic pathway for the preparation of the precursor quadricyclane, thereby greatly simplifying the assignment of the molecular origin of the measured photofragments. The excitation process and subsequent non-adiabatic dynamics have been previously investigated employing time-resolved photoelectron spectroscopy and are now complemented with high-level ab initio trajectory simulations that elucidate the specific vibronic relaxation pathways. Lastly, the fragmentation channels accessed by the molecule following internal conversion are probed using velocity map imaging (VMI) so that the identity of the fragmentation products and their corresponding energy distributions can be definitively assigned.

Full text of DOI:10.1002/chem.201501624

DOI: 10.1002/chem.201501624
Full Paper
&
Astrochemistry
Photodissociation Dynamics of Cyclopropenylidene, c-C3H2
Michael S. Schuurman,*[a, b] Jens Giegerich,[c] Kai Pachner,[c] Daniel Lang,[d] Benjamin Kiendl,[d]
Ryan J. MacDonell,[b] Anke Krueger,*[d] and Ingo Fischer*[c]
Abstract: In this joint experimental and theoretical study we
characterize the complete dynamical “life cycle” associated
with the photoexcitation of the singlet carbene cycloprope-
nylidene to the lowest lying optically bright excited electron-
ic state: from the initial creation of an excited-state wave-
packet to the ultimate fragmentation of the molecule on the
vibrationally hot ground electronic state. Cyclopropenyli-
dene is prepared in this work using an improved synthetic
pathway for the preparation of the precursor quadricyclane,
thereby greatly simplifying the assignment of the molecular
origin of the measured photofragments. The excitation pro-
cess and subsequent non-adiabatic dynamics have been pre-
viously investigated employing time-resolved photoelectron
spectroscopy and are now complemented with high-level ab
initio trajectory simulations that elucidate the specific vi-
bronic relaxation pathways. Lastly, the fragmentation chan-
nels accessed by the molecule following internal conversion
are probed using velocity map imaging (VMI) so that the
identity of the fragmentation products and their correspond-
ing energy distributions can be definitively assigned.
Introduction
tion to c-C3H2 two more C3H2 isomers exist, propadienylidene,
H2C=C=C:, which was recently suggested to be a carrier of the
Cyclopropenylidene, c-C3H2 (compound 1 in Scheme 1) is a sin-
glet carbene that occurs in a surprisingly high abundance in
interstellar space. It was identified in dense and diffuse inter-
stellar clouds[1] by its rotational spectrum[2] and is suggested to
be a carrier of the unidentified infrared bands (UIB).[3] In addi-
diffuse interstellar bands,[4] and propargylene, HÀC=C=CÀH,
which has not yet been discovered in space due to its small
dipole moment.[5] All three isomers are also relevant for com-
bustion chemistry and have been detected in flames.[6] Propar-
gylene was recently studied by velocity map imaging[7] and by
negative-ion photoelectron spectroscopy.[8] The astrochemical
importance of c-C3H2 triggered laboratory work by microwave
spectroscopy,[9] matrix isolation IR spectroscopy[10]and photoio-
nization.[11] Indeed, synthetic chemists succeeded in synthesiz-
ing derivatives of c-C3H2 that could be isolated in the laborato-
ry, which have subsequently been identified as having poten-
tial in catalysis.[12]
˙
˙
Scheme 1. Pyrolysis of the quadricyclane 2 generates cyclopropenylidene
1 and benzene.
For the photochemical reactions that occur in diffuse inter-
stellar clouds, photodissociation in particular is expected to be
important due to the high UV radiation density. However, the
photodissociation of c-C3H2 has yet to be explored in detail.
Photoinitiated isomerization has been observed in rare gas ma-
trices,[10a] but since the matrix provides a bath to carry away
excess energy, the results have little bearing on the reactions
that would occur in interstellar space. The ground state is the
1A1, as both carbene electrons are in a s orbital, resulting in 2p
electrons and engendering an aromatic ring.[13] The bright B
1B1 state was identified by multiphoton ionization,[14] and its ul-
trafast internal conversion was followed by femtosecond (fs-)
photoelectron spectroscopy.[15] Subsequent H-atom loss was
observed by photofragment excitation spectroscopy, but nei-
ther a product-energy distribution nor mechanistic details
have been reported. We therefore initiated a study using ve-
locity map imaging (VMI).[16] In previous experiments on 1,
mostly chlorocyclopropene, c-C3H3Cl was used as the precur-
sor.[11b,c,14] A major challenge in such experiments on reactive
[a] Dr. M. S. Schuurman
National Research Council of Canada
100 Sussex Drive, Ottawa, ON, K1A 0R6 (Canada)
E-mail: Michael.Schuurman@nrc-cnrc.gc.ca
[b] Dr. M. S. Schuurman, R. J. MacDonell
Department of Chemistry
University of Ottawa
D’Iorio Hall, 10 Marie Curie, Ottawa, ON, K1N 6N5 (Canada)
[c] J. Giegerich, K. Pachner, Prof. Dr. I. Fischer
Institute of Physical and Theoretical Chemistry
University of Würzburg, Am Hubland, 97074 Würzburg (Germany)
E-mail: ingo.fischer@uni-wuerzburg.de
[d] Dr. D. Lang, B. Kiendl, Prof. Dr. A. Krueger
Institute of Organic Chemistry
Am Hubland, 97074 Würzburg (Germany)
E-mail: anke.krueger@uni-wuerzburg.de
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/chem.201501624.
Chem. Eur. J. 2015, 21, 14486 – 14495
14486
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
species is the unambiguous assignment of the photofrag-
ment’s parent. One has to exclude the precursor or side prod-
ucts as possible sources of the photofragment. One such side
product is c-C3HCl, generated from c-C3H2Cl2 by pyrolysis, the
latter being formed in the precursor synthesis. This possibility
can be investigated by changing the precursor. In fact, the
quadricyclane 2 has been reported as an alternative precursor
(Scheme 1).[10c,17] However, the reported overall yield of less
than 0.1% is prohibitive when substantial amounts have to be
synthesized for a series of gas-phase experiments. We there-
fore devised an improved route to synthesize the compound
to obtain a sufficient amount of 2, which is summarized in
Scheme 2.
spectroscopic investigation due to the availability of larger
amounts of 2.
Given the extent of spectroscopic interest in c-C3H2, it is per-
haps not surprising that this molecule has attracted the atten-
tion of theoretical chemistry as well. While ground-state equi-
librium structures, vibrational frequencies, ionization and verti-
cal excitation potentials have been computed previously,[14,19]
the excited state dynamics have not been studied in detail.
However, carbenes are known to possess low-lying, vibronically
coupled electronic states that can often pose a considerable
challenge to theory.[20] Again new strategies, in particular the
advent of on-the-fly molecular dynamics methods,[21] now
permit the rigorous treatment of the ultrafast dynamical as-
pects of photochemical processes in open-shell mole-
cules.
Here we present a detailed investigation of the
photochemistry and photodissociation dynamics of c-
C3H2 (1) by using quadricyclane 2 as the precursor.
The ultrafast non-adiabatic internal conversion pro-
cesses that follow photoexcitation are explored using
on-the-fly ab initio molecular dynamics simulations
and are compared to existing femtosecond time-re-
solved photoelectron spectroscopic data. In combina-
tion with the product-energy distribution obtained
by VMI, this approach allows a detailed insight into
the photodynamics of cyclopropenylidene. Only re-
cently we investigated the dynamics of propargylene,
HCCCH, by this combination of imaging spectroscopy
Scheme 2. Synthesis of spiro(cycloprop[2]ene-1,3’-tetracyclo[3.2.0.02,7.04,6]heptane) (2):
i) PhCO3tBu, CuBr, benzene; ii) 1. Ac2O, AcOH, 2. HClO4; iii) hn, acetone, n-hexane; iv) KOH,
MeOH; v) DCC (dicyclohexylcarbodiimide), DMSO, H3PO4; vi) [Cp2TiCH2Cl], THF; vii) CHBr3,
KOtBu, n-pentane; viii) nBu3SnH, THF; ix) KOtBu, DMSO.
and on-the-fly dynamics.[7]
Results and Discussion
Mass spectra and electronic spectroscopy
The synthesis is partially based on the work by Butler
et al.[17] and Hoffmann et al.[18] Here we synthesize quadricycla-
none 8 in gram quantities starting from norbornadiene in an
overall yield of 5% over five steps. This is due to the inefficient
initial reaction of norbornadiene 3 to the tert-butyl ether 4. As
the starting material is available in large quantities this step
was not optimized in this work. The subsequent steps in the
synthesis of 8 gave satisfactory yields. In the following forma-
tion of the methylene quadricyclane 9 the use of Tebbe’s re-
agent instead of the triphenylphosphane-based approach of
the Wittig reaction led to a simplified isolation of the highly
volatile 9. However, further improvement resulted from avoid-
ing any isolation of 9 before the subsequent reaction with the
dibromo carbene to form the dibromospirocyclopropane-
quadricyclane 10 directly in a yield of 49% for the two steps.
The monodebromination to 11 proceeded in 87%. The elimi-
nation of HBr to yield the desired hydrocarbon 2 led to 80%
isolated and pure product (further product could not be com-
pletely separated from the solvent). In total, 2 was synthesized
in 34% overall yield starting from 8. Until now, the target com-
pound 2 was only available as a crude product due to difficul-
ties in the isolation of this highly volatile hydrocarbon. The im-
proved synthesis and isolation enables now more advanced
In a first step, the precursor and the pyrolysis products were
characterized by photoionization TOF-MS. A difference mass
spectrum, obtained by subtracting the spectrum with pyrolysis
off from the one with pyrolysis on is depicted in Figure 1. It
was recorded with 118 nm radiation. Upon turning the pyroly-
Figure 1. Difference of the photoionization mass spectra with pyrolysis on
and off, recorded at an irradiation wavelength of l=118 nm. The increase of
the signals for the pyrolysis products at m/z: 38 and 78 is evident.
Chem. Eur. J. 2015, 21, 14486 – 14495
www.chemeurj.org
14487
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
sis on, the precursor signal at m/z: 116 decreases and two new
signals appear at m/z: 38 and 78, corresponding to c-C3H2 and
benzene, the expected pyrolysis products. Two issues should
be noted: First, the pyrolysis is clean and does not produce
further side products; second, at the employed pyrolysis tem-
peratures 90% of the precursor is transformed.
In a further series of experiments we tested the possible
sources of H-atoms. Figure 2 shows mass spectra recorded
Figure 3. H-atom photofragment excitation spectrum of c-C3H2. The upper
trace depicts the spectrum obtained using c-C3H3Cl as a precursor, the lower
trace shows the spectrum obtained using quadricyclane 2. The peak at
!
1
1
36900 cmÀ1 represents most likely the origin of the B B1 X A1 transition.
were previously observed in the m/z: 38 mass channel using
chlorocyclopropene as the precursor, they are assigned to vi-
!
1
1
bronic bands of the B B1 X A1 transition of cyclopropenyli-
dene. It corresponds to the excitation of an electron from the
HOMO into the empty p-orbital located at the carbene centre,
orthogonal to the C-C-C plane of the molecule. Note that an
absorption spectrum with a similar vibronic structure was ob-
served in a rare gas matrix and also assigned to c-C3H2.[10b] The
similarity of the action spectra demonstrates that c-C3H2 is pro-
duced from both precursors and is the source of H-atoms at
the respective peaks.
Figure 2. Photoionization mass spectra with pyrolysis at 121.6 nm (Lyman-
a resonance) and excitation laser at 271 nm on (upper trace) and off (lower
trace).Without excitation laser the H-atom signal is negligible.
with 121.6 nm photoionization and the pyrolysis turned on.
Note that Vacuum Ultraviolet (VUV) generation in Kr at this
wavelength is less efficient than in Xe at 118 nm. As visible in
Figure 2 the precursor peak is broadened and H2-loss either
due to photodissociation or dissociative photoionization
occurs, possibly by processes induced by the 365 nm funda-
mental radiation. While the excitation laser is turned off and
the pyrolysis and the ionization laser (121.6 nm, 12 mJ) are
both on (Figure 2, lower trace), there is only a negligible H-
atom signal visible. When the excitation laser (271 nm, 2 mJ) is
switched on (Figure 2, upper trace), a small C3H2 signal be-
comes visible due to resonant excitation and the H-atom
signal increases by more than two orders of magnitude. No H-
atom signal was visible without pyrolysis, that is, the quadricy-
clane does not yield H-atoms upon excitation. However, hot
benzene or pyrolysis products of benzene might be alternative
sources of H-atoms. Therefore we recorded mass spectra of
pure benzene heated in the pyrolysis tube to check for possi-
ble contributions, but found the H-atom signal at 271 nm to
be negligible.
Computational results
To develop a complete picture of the photodynamics, the in-
tramolecular processes following photoexcitation have to be
explored. Previous femtosecond time-resolved measurements
1
revealed a short lifetime of the initially excited B B1 state, but
provided only a qualitative picture. We therefore carried out
computations to provide insight into the short-time processes
and to support the experimental data. The results are com-
pared below to the earlier fs-photoionization study.[15]
Potential-energy surfaces
Ab initio optimized structures relevant to the excited state dy-
namics are presented in Figure 4, as is the atom labelling that
will be employed in the following discussion. The values of the
internal coordinates for all optimized geometries are given in
the Supporting Information. As evinced by the figure, the com-
puted vertical excitation energy to the bright S2 (1B1) state, cor-
responding to the second band in Figure 3, is 4.70 eV, which is
equal to the energy of the 267 nm pump pulse employed in
the fs-experiments.[15] The value is in very good agreement
with the previously computed values of 4.74[14] and 4.89 eV.[19d]
The initial gradient-directed relaxation on the S2 surface is seen
to be only moderate, with a (local) S2 minimum energy struc-
ture, 0.2 eV below the Franck–Condon (FC) geometry. Note
that experimentally the first band at 4.57 eV was assigned to
In the next step H-atom action spectra were recorded, that
is, the H-atom signal was monitored as a function of excitation
energy. In Figure 3 this action spectrum is compared to the
earlier one obtained using chlorocyclopropene as the precur-
sor.[14] Due to the small amount of sample available for this ex-
periment, only parts of the spectrum were reproduced. As
shown in Figure 3 the peaks at 36900 (271 nm/4.57 eV) and
37475 cmÀ1 (267 nm/4.64 eV) are present with both precursors,
while the third band around 37790 cmÀ1 is slightly shifted in
comparison with the previous spectrum. As the same peaks
Chem. Eur. J. 2015, 21, 14486 – 14495
www.chemeurj.org
14488
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Non-adiabatic dynamics
Following excitation to the
bright 1B1 state, cyclopropenyli-
dene rapidly decays to the
ground electronic state via coni-
cal intersection mediated dy-
namics. Figure 5 illustrates the
adiabatic state populations from
the multiple spawning simula-
tion as a function of time. As
seen from the S2 adiabatic popu-
lation (dashed lines in Figure 5),
decay to the ground state
begins almost immediately fol-
lowing excitation (see dashed
lines), likely engendered by the
seam of conical intersection be-
tween the S2 and S1 that en-
croaches on the FC region. The
resultant population on the S1 is
Figure 4. Relative energies at ab initio optimized stationary points on the low-lying electronic states of C3H2, de-
termined at the cc-pVTZ/MRCI level of theory. The structural motifs at each point refer to the number of the geo-
metries at the right, while the specific geometrical parameters for each point are given in the Supporting Informa-
tion.
the adiabatic excitation. In addition to an elongation of each
CÀC bond in the ring (1.42 to 1.45 Š for the CÀC bonds and
1.32 to 1.36 Š for C=C bond), this structure is found to be non-
planar (Cs symmetry), with the two hydrogen atoms displaced
out of the plane of the C3 ring via the symmetric H-wag coor-
dinate. The local minimum on the S1 surface (i.e. gradient di-
rected from the FC region) also has Cs symmetry via displace-
ments along the symmetric H-wag coordinate, but displays
shortened C1ÀC2 and C1ÀC3 bonds (1.35 Š) and a significantly
elongated C2ÀC3 bond at 1.56 Š.
The minimum-energy conical intersection (MECI) between
the S2 and S1 in the FC region is found to lie at 4.64 eV, which
is 0.1 eV above the local minimum on the S2 surface, and less
than 0.1 eV below the computed vertical excitation energies.
The MECI is planar, with C2v symmetry and involves elongation
along the single C1ÀC2 and C1ÀC3 bonds (1.48 Š), but with no
displacement along the FC active C2ÀC3 bond.
Figure 5. Time evolution of the adiabatic state populations determined from
the multiple spawning simulation for the S2 (dashed line), S1 (dotted line),
and S0 (full line) electronic states.
very short-lived (dotted line in Figure 5); however, this ampli-
tude is quickly dumped to the ground state via the second set
of conical intersections, as discussed below.
There exist a number of conical intersection types for curve
crossings between the S1 and S0 electronic states. Those rele-
vant to the current dynamics simulation are shown in Figure 4.
In particular, they fall into two classes: those involving bond
cleavage of a CÀC bond, leading to a linear isomer and those
arising due to linear combinations (symmetric and anti-sym-
metric) of the H-wag coordinates, resulting in a distorted ring.
Energetically, the two H-wag MECI (symmetric and anti-sym-
metric) are computed to lie very close in energy at 2.56 and
2.66 eV, respectively. These intersection motifs are bracketed in
energy by the double- and single-bond cleavage intersections
that are computed to lie at 1.45 and 3.90 eV, respectively.
Panels a–c of Figure 6 illustrate the integrated wavepacket
density as a function of a set of selected internal coordinates
(double-bond stretch, symmetric single-bond stretch and the
sum of the absolute values of the hydrogen out of plane
angles, denoted r1, (r2 +r3)/2 and jg1 j + jg2 j, respectively). Vi-
brations along the Franck–Condon active double-bond stretch
and symmetric single-bond stretch are observed to be remark-
ably coherent, undergoing roughly a dozen vibrational periods
before the preponderance of wavepacket amplitude is in the
ground state. Similarly, displacements along the hydrogen out
of plane coordinate, both symmetric and anti-symmetric,
begin immediately upon photoexcitation.
The relationship between these geometric distortions and
the resulting nonadiabatic transitions is more clearly elucidated
Chem. Eur. J. 2015, 21, 14486 – 14495
www.chemeurj.org
14489
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
which the C=C bond length is
greater than 2 Š. While the ma-
jority of the excited-state popu-
lation decays via the H-wag
MECI types (51% symmetric vs.
22% asymmetric), a significant
amount of population also
decays by ring-opening (~20%).
Furthermore, even the H-wag in-
tersection types display signifi-
cant elongation along the C=C
coordinate, with a mean dis-
tance of ~1.6 Š for these points
(again correlating well the MECI
optimizations). Thus, given the
amount of vibrational energy
that would be present following
transition to the S0, ring opening
shortly after reaching the
ground state can be presumed
to be likely.
Simulation of the ion-yield
signal
In the next step, we want to
compare the results of the dy-
namics calculations with earlier
fs-photoionization experiments
obtained by [1+3’] photoioniza-
tion using a 267 nm pump and
800 nm probe radiation.[15] The
time evolution of the ion yield
arising from direct ionization
from the probe pulse can be ap-
proximated by determining the
time dependence of the ioniza-
tion potential for the excited-
state trajectory basis functions.
Figure 6. Time evolution of the vibronic wavepacket. In panels a–c, the integrated wavepacket densities on S2 and
S1 are plotted as a function of selected internal coordinates. The S0 minimum value is shown as a solid line. In
panels d–f the geometric parameters for S2!S1 spawns (red circles) and S1!S0 spawns (blue squares) are shown.
The S0 minimum values are plotted as dashed lines. Those geometries corresponding to ring opened structures
fall within the solid rectangles.
in panels d–f (Figure 6). Here, the spawn geometries are plot-
The simulated signal from neutral state I is given by the ex-
ted on grids corresponding to the above-described coordi-
nates. The values at the S0 minimum geometry are shown as
dashed lines. For all of the coordinates plotted, the S2!S1 tran-
sitions (red circles) occur closer to the Franck–Condon region
than the S1!S0 transitions, as one may have expected. Most of
the displacements involve elongation of the symmetric C–C
stretch, with some H-wag character as well. This is entirely con-
sistent with the intersection optimizations which show the
S2!S1 MECI to involve primarily elongation of this coordinate
(and with no concomitant lengthening of the C2ÀC3 bond).
The S1!S0 spawn geometries show that as the wavepacket
evolves (briefly) on the S1 surface, the symmetric C–C stretch
and H-wag coordinates continue to increase, but so does the
double-bond stretch coordinate. In fact, one observes two
classes of S1!S0 spawn types, particularly in Figure 6d and f.
Ring-opening along the C=C bond is observed in the excited
state, as evinced by the clusters of spawn geometries for
pression:
Ntj ðtÞ
Z
ꢀ
ꢀ
X
2
PIðtÞ /
ð CiðtÞj hycðRðtÞÞꢀðkÞje À mynðRðtÞijidkÞ
ꢀ
j
ð1Þ
j
Âdð3hwprobe À EIPðRðtÞÞ
in which ycðRðtÞÞ and ynðRðtÞÞ are the electronic wavefunc-
tions of the cationic and neutral states, respectively, and ꢀðkÞ
is the scattering state of the outgoing electron with momen-
tum k. As defined above, CjiðtÞ is the complex amplitude for
trajectory j on adiabatic state I, while hwprobe and EIPðRðtÞÞ are
the energy of the ionizing photoelectron and the ionization
potential at geometry R(t), respectively. The prefactor on the
hwprobe term is employed since ionization (at low electron ki-
netic energies) proceeds by a three photon process. In what
Chem. Eur. J. 2015, 21, 14486 – 14495
www.chemeurj.org
14490
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
follows, the ionization matrix element (for which there is
a single energetically accessible cationic electronic state) is as-
sumed to be a geometry-independent multiplicative factor up
to which the total signal is determined.
cally possible photochemical reactions. Three reaction channels
are available in the investigated energy range:
2
ðIÞ c-C3H2 ! c-C3H ðX B2Þ þ H DRHð0 KÞ ¼ þ422:5 kJ molÀ1
The resulting simulated ion yield spectrum is shown in the
lower trace on Figure 7, while an experimental spectrum[15] is
2
ðIIÞ c-C3H2 ! l-C3H ðX PÞ þ H DRHð0 KÞ ¼ þ430:5 kJ molÀ1
1
ðIIIÞ c-C3H2 ! C3 ðX SgþÞ þ H2 DRHð0 KÞ ¼ þ306:5 kJ molÀ1
Two further reaction channels exist that are not energetically
accessible at 271 nm:
1
3
ðIVÞ c-C3H2 ! C2H2 ðX SgþÞ þ C ð PÞ DRHð0 KÞ ¼ þ451:5 kJ molÀ1
2
2
ðVÞ c-C3H2 ! CH ðX PÞ þ C2H ðX SþÞ DRHð0 KÞ ¼ þ651 kJ molÀ1
Channel (IV) is in addition spin-forbidden. The reaction en-
thalpies are based on computations carried out by Mebel et
al.[22] and Ochsenfeld et al.[23] The lowest-energy product chan-
nel (III) is associated with a reverse barrier of 107 kJmolÀ1. We
carried out RRKM computations showing that the reaction rate
for (III) at low excess energies is considerably smaller than for
H-atom loss, due to the high transition-state frequencies.
H-atom images were systematically investigated at 271 nm
excitation. Images recorded at other excitation wavelengths
confirm the conclusions obtained at 271 nm. The image seems
to be isotropic and shows a bright spot in the centre, indicat-
ing that H-atoms with low translational energy dominate. The
translational energy distribution P(ET) derived from this image
is displayed in Figure 8. It peaks at 0.03 eV, which corresponds
Figure 7. Comparison of the experimental time-dependent ion-yield ob-
tained in fs-experiments (upper trace) with a simulated one. The decay was
best fit using a bi-exponential giving rise to two time constants. As visible
there is good agreement between theory (g: computed ion signal;
*
c fit with t1 =45 fs and t2 =434 fs) and experiment ( : data, m/z: 38;
c: fit with t1 = <50 fs and t2 =180 fs).
given in the upper trace for comparison. The simulated spec-
trum was obtained by convolution of the signal computed
using Equation (1) with a Gaussian of full width half maximum
p
(FWHM) of 70 2/3 fs, which was consistent with the spectro-
scopic characterization of the experimental cross-correlation.
As evinced by Figure 7, the plot is best fit with a bi-exponen-
tial decay with time constants determined to be t1 =45 fs and
t2 =434 fs. The signal is due almost exclusively to ionization of
population on the S2 adiabatic state. As shown by the popula-
tion decay curves in Figure 5, this initial time constant is due
to the rapid initial decay of the wavepacket to the ground S0
via the S1 state. The long-time component is attributed to
those components of the wavepacket that are deflected by ini-
tial passage through the FC region of the conical intersection.
They are thus relatively long-lived given the existence of
a nearby local minimum on the S2 state. As visible, these time
constants, even given the somewhat qualitative treatment of
the ionization process are in good agreement with the experi-
mentally determined values, although t2 is somewhat shorter
in the experiment.
Figure 8. Experimental H-atom translational energy distribution (g) and
a fit based on a statistical description of the reaction (c, see text for de-
tails). The arrow indicates the maximum translational energy expected from
(I).
Imaging
The computations show that c-C3H2 returns quickly to the
ground-state potential energy surface with enough internal
energy to overcome the barrier to dissociation. To study the
photodissociation dynamics in the ground state, we recorded
velocity map H-atom images and extracted the H-atom transla-
tional energy release (TER) from it. Before the experimental re-
sults are presented, we want to summarize the thermochemi-
to 2.9 kJmolÀ1, and extends to roughly 0.3 eV. With the DRH of
422.5 kJmolÀ1 computed for channel (I), a maximum transla-
tional energy release (TER) of 19 kJmolÀ1 or 0.2 eV is expected.
However, as visible the distribution decays to zero at 0.3 eV
TER. There are two possible reasons: First, the pyrolysis might
produce vibrationally excited carbenes, given the typical vibra-
tional temperatures of 150–200 K.[7] Second, multiphoton pro-
Chem. Eur. J. 2015, 21, 14486 – 14495
www.chemeurj.org
14491
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
value of n=0.5 corresponds to a simple prior distribution[26]
and represents an ideal statistical reaction.
In the case of c-C3H2 the situation is more complicated. Two
products can be formed, c-C3H (channel I) and l-C3H (channel
II). Although channel I is lower in energy, the transition state is
tighter, making channel II competitive above the reaction
threshold. Since the vibrational density of states of the prod-
ucts is low at low excess energies, it is indispensable to include
rotations and use the rovibrational density of product states
1rovib in Equation (3) instead of the simple vibrational density.
Thus we computed two separate contributions for channels I
and II and convoluted them for the fit in Figure 8. Based on
RRKM calculations we assumed a branching ratio for channels
I/II of 85:15. The best fit to the translational energy distribution
given in Figure 8 is given by the dotted line and was obtained
with n=0.5, which is in good agreement with an ideal statisti-
cal process. Hot-band contributions were not included in the
fit. As visible the fit is not a smooth function, but shows step-
like features that are associated with the onset of product vi-
brational states. These steps are washed out in the experiment
most likely due to the rotational temperature that is assumed
to be around 150 K.[7,27]
Figure 9. Photofragment angular distribution for H-atoms with less than
0.2 eV ET is isotropic, indicating a statistical dissociation. The H-atoms with
a higher ET show an anisotropic distribution, indicating that they originate
from a different process (Left: a: H atoms with 0 eVꢀET ꢀ0.2 eV; c: fit.
Right: a: H atoms with 0.2 eVꢀET ꢀ0.3 eV; c: fit).
cesses might contribute. In fact, when the TER was recorded at
twice the excitation laser power, the distribution extended to
even higher energies, while the maximum of the distribution
did not change. This observation already suggests that H-
atoms at high TER originate from multiphoton processes.
Figure 9 depicts the angular distribution (PAD) P(q) of the H-
atom photofragments. The solid line represents a fit to the
data using the standard recoil anisotropy function:
As mentioned in the introduction, experiments were also
conducted using c-C3H3Cl as the precursor, which did not yield
a clean mass spectrum. However, the translational-energy dis-
tribution and the angular distribution of the H-atom photo-
fragment had a very similar appearance, but an inferior signal/
noise ratio.
1
4p
ð2Þ
PðqÞ ¼
½1 þ b  P2ðcos qފ
Discussion and Conclusions
The theoretical and experimental results described above now
provide a detailed description of the photochemistry of cyclo-
propenylidene. The ab initio dynamics simulations reveal the
P2 represents the second-order Legendre polynomial and b is
the anisotropy parameter which is optimized in the fit. On the
left-hand-side of Figure 9 results for H-atoms with a translation-
al energy between 0 and 0.2 eV are shown. From the opti-
mized value for b of almost 0, it follows that the H-atoms are
isotropically distributed in space and the dissociation is slower
than a rotational period, in agreement with a ground-state dis-
sociation. However, for H-atoms with ET between 0.2 and
0.3 eV (right-hand-side of Figure 9) an anisotropy of b=À0.71
is visible, supporting the assumption that the fastest H-atoms
originate from a different electronic state populated by a multi-
photon excitation.
1
following picture: Upon photoexcitation to the bright B1 state,
the molecule (i.e. the initially formed wavepacket assuming co-
herent fs-excitation) decays quickly to the S1 state given the
presence of seams of conical intersections in the Franck–
Condon region. While most of the initial wavepacket rapidly
reaches the S1 state via conical intersections, some compo-
nents may linger on the S2 surface following deflection to the
nearby local minimum. This accounts for both the initial rapid
decay, and the somewhat longer tail of the S2 population
curve in Figure 5. The time-dependent ion yield measured in
fs-time-resolved photoionization experiments is properly de-
scribed, confirming that this picture is essentially correct. How-
ever, the two time constants observed experimentally were
previously assigned to deactivation of S2 and S1. In light of the
computations this assignment has to be revised. Both time
constants reflect motion in the S2 state and the intermittent
population in the S1 state is too small to be detected. Popula-
tion on S1 decays very rapidly to the ground state by either
the H-wag or double bond cleavage, although even the former
is characterized by a significant lengthening of the double
bond. Deactivation is observed to be primarily through H-wag
intersections (~75%), although CÀC bond cleavage (ring open-
In previous experiments,[24] we modelled the P(ET) obtained
from a statistical dissociation by the following function[25]
:
PðET Þ ¼ C  ETn  1rovibðEexcess À ET Þ
ð3Þ
In this function C describes a normalization constant chosen to
scale the fit to the experimental data, provib is the rovibrational
density of states of the product c-C3H, Eexcess is the excess
energy above the activation barrier, ET is the energy released
as translation and n is a fitting parameter. The expression is in-
tegrated over all translational energies and mimics the behav-
iour described by the statistical adiabatic channel model.[25]
A
Chem. Eur. J. 2015, 21, 14486 – 14495
www.chemeurj.org
14492
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
tails on the steps leading from 3 to 9 can be found in the Support-
ing Information (SI).
ing) contributes a sizable minority (~20%) to the nonradiative
decay of the excited-state wavepacket.
In the deactivation process, cyclopropenylidene (1) thus re-
turns to the ground-state potential energy surface within less
than 1 ps and electronic excitation is converted to internal
energy. Thus the barrier to unimolecular dissociation to C3H+
H can be overcome. Velocity map imaging of the H-atom pho-
tofragments yields an isotropic angular distribution. For H-
atoms with a translational energy below 0.2 eV a b-parameter
of bꢁ0 is obtained, any anisotropy induced by the excitation
is lost. This indicates that dissociation takes significantly longer
than a rotational period and can be described by statistical re-
action theories. Note that in RRKM calculations a dissociation
rate of 4”107 sÀ1 was computed for channel I, very close to
the time resolution of our nanosecond laser system (Dt
ꢁ10 ns). Therefore it is not surprising that the rise of the H-
atom signal could not be time-resolved. At excess energies of
only 0.2 eV the number of possible vibrationally excited states
in the product is small and one might expect to see discrete
peaks in the TER distribution instead of the almost continuous
one shown in Figure 8. However, evaluation of the H-atom
translational-energy distribution has to take into account that
the internal energy is rapidly distributed over all degrees of
freedom. The computations also indicate that ring opening
takes place in a significant fraction of molecules. We therefore
conclude that both product channels, c-C3H+H (I) and l-C3H+
H (II) are accessible and thus vibrational states of both molecu-
lar products are populated in the reaction. Based on RRKM cal-
culations we assume a branching ratio of 85:15 in favour of c-
C3H formation. Furthermore rotational energy has to be includ-
ed into the product energy distribution, because pyrolytically
generated species often possess a rotational temperature of
around 150 K.[7,27] With these assumptions the translational
energy distribution was fitted by a function close to a prior dis-
tribution, which confirms that the reaction is statistical in
nature. Furthermore the existence of a substantial reverse bar-
rier for the reactions can be ruled out, because the energy of
such a reverse barrier is preferentially released as translation[28]
and leads to deviations from a prior distribution.
Synthesis
2-Dibromospiro(cyclopropane-1,3’-tetracy-
clo[3.2.0.02,7.04,6]heptane) (10) via 3-methylenequadricyclane
(9)
Under a nitrogen atmosphere, Tebbe’s reagent (25.2 mL, 3.59 g,
12.6 mmol) in toluene (0.5 m) was cooled in liquid nitrogen. The
solvent was removed under reduced pressure whilst the reagent
solution was allowed to reach room temperature. After removal of
toluene, the reagent was dissolved in dry THF and cooled to 08C.
Then ketone 8 (670 mg, 6.31 mmol) was added in small portions.
The reaction mixture was stirred for 1 h at 08C and for 45 min at
room temperature. After cooling with an ice bath, NaOH (0.1 m)
was added to the reaction mixture until foam formation stopped
(approx. 30 mL). The reaction mixture was extracted with n-pen-
tane (8” 50 mL). The combined organic phases were washed with
brine (2” 80 mL) and dried over Na2SO4. After removal of the sol-
vent under slightly reduced pressure (>900 mbar) the crude prod-
uct was dissolved in 15 mL n-pentane. KOtBu (3.55 g, 31.6 mmol)
was added and the resulting dispersion cooled to 08C. After drop-
wise addition of CH3Br (3.18 g, 1.10 mL, 12.6 mmol), the reaction
mixture was stirred at room temperature for 4 h and then poured
into ice water (30 mL). After extraction with n-hexane (5” 30 mL),
the combined organic phases were washed with ice water (1”
50 mL) and dried over Na2SO4. Removal of the solvent under re-
duced pressure and purification by column chromatography on
silica (n-pentane) gave dibromo compound 10 as a colourless crys-
talline solid (861 mg, 3.12 mmol, 49%, m.p. 70–718C). Rf (n-pen-
tane, TLC detection: UV/Vis or iodine): 0.80. 1H NMR (CDCl3,
400 MHz): d=2.06 (s, 2H), 1.96–1.91 (m, 2H), 1.85–1.81 (m, 2H),
1.35 ppm (t, J=4.5 Hz, 2H); 13C NMR (CDCl3, 100 MHz): d=47.0,
31.7, 31.7, 29.1, 18.2, 17.1 ppm; FTIR (ATR): n˜ =3062 (w), 1436 (w),
1263 (w), 1230 (m), 1194 (w), 1071 (w), 1027 (m), 994 (m), 960 (w),
927 (m), 900 (w), 795 (s), 763 (s), 729 (m), 687 (s), 670 (m), 657
(m) cmÀ1
.
2-Bromospiro[cyclopropane-1,3’-tetracyclo[3.2.0.02,7.04,6
heptane] (11)
]
Under a nitrogen atmosphere, 10 (401 mg, 1.45 mmol) was dis-
solved in dry THF (12 mL). After the addition of nBu3SnH (0.39 mL,
422 mg, 1.45 mmol), the reaction mixture was stirred for 23 h at
room temperature. The solvent was removed under reduced pres-
sure. The crude product was purified by column chromatography
on silica (n-pentane) to give 11 (248 mg, 1.26 mmol, 87%) as a col-
ourless liquid, which crystallizes upon cooling to À308C. Rf (n-pen-
tane, TLC detection: iodine): 0.70. 1H NMR (CDCl3, 400 MHz): d=
3.39 (dd, J=7.6, J=4.2 Hz, 1H), 1.84–1.77 (m, 2H), 1.76–1.72 (m,
2H), 1.54 (dd, J=6.5, J=7.6 Hz, 1H), 1.35–1.33 (m, 2H), 0.92–
0.89 ppm (m, 2H); 13C NMR (CDCl3, 100 MHz) d=37.2, 29.4, 23.5,
18.3, 16.9, 16.4, 16.1, 15.7, 15.7 ppm; FTIR (ATR): n˜ =3070 (w), 2991
(w), 1438 (w), 1289 (w), 1254 (w), 1233 (m), 1208 (w), 1183 (w),
1065 (w), 1032 (w), 1008 (w), 959 (w), 927 (w), 903 m, 845 (w), 787
In conclusion we have presented a comprehensive study on
the photochemistry of cyclopropenylidene, one of the most
important reactive organic molecules in astrochemistry, by
combining spectroscopy, ab initio simulation, and synthetic
chemistry. The intramolecular excited stated dynamics and the
unimolecular dissociation in the electronic ground state have
been described in detail. After UV excitation c-C3H2 rapidly de-
activates to the electronic ground state via conical intersec-
tions. There it loses a H-atom in a statistical fragmentation re-
action as shown by velocity map imaging.
(s), 758 (vs), 670 (w), 604 (s) cmÀ1
.
Experimental Section
Spiro[cyclopropane-3,3’-tetracyclo[3.2.0.02,7.04,6]heptane] (2)
Methods
Under a nitrogen atmosphere, 11 (212 mg, 1.08 mmol) was dis-
solved in dry DMSO (6 mL). After addition of freshly sublimated
KOtBu (242 mg, 2.16 mmol) in small portions, the solution was
stirred at room temperature for 2 h and then poured onto ice
Cyclopropenylidene 1 was produced from quadricyclane 2 as de-
picted in Scheme 1 by flash pyrolysis[29] in a jet of helium (2 bar).
The synthesis of 2 was carried out as described in Scheme 2. De-
Chem. Eur. J. 2015, 21, 14486 – 14495
www.chemeurj.org
14493
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
water (50 mL). After extraction with n-pentane (6” 30 mL), the
combined organic phases were washed with dist. water (2”50 mL)
and dried over Na2SO4. The solvent was removed under reduced
pressure (>150 mbar) and collected in a cooling trap whilst the or-
ganic phase was cooled to 08C. The collected solvent was submit-
ted to additional distillation to isolate further product. This afford-
ed 2 as colourless crystalline solid (101 mg, 869 mmol, 80% (54%
after direct isolation+27% from redistillation of the trapped sol-
vent). M.p. approx. 158C); Rf (n-pentane, TLC detection: iodine):
ple electronic states. The total wavefunction is given as a sum of
products of nuclear and electronic functions:
X
X
NIðtÞ
j¼1
Yðr, R, tÞ ¼
IYIðr; RÞ
CIjðtÞcIjðR; RjIðtÞ,PjIðtÞ,gIjðtÞÞ ð4Þ
in which CIj(t), RjI(t) and PIj(t) are the time-dependent complex ampli-
tudes, position and momenta, respectively, of basis function j, and
gIj(t) is the phase. While the amplitudes are propagated variational-
ly in the basis of trajectory functions, the position and momenta
are propagated using classical equations of motion.
1
0.80; H NMR (CDCl3, 400 MHz): d=7.75 (s, 2H), 1.70–1.68 (m, 4H),
0.64–0.62 ppm (m, 2H) ppm; 13C NMR (CDCl3, 100 MHz): d=121.5,
34.4, 27.2, 15.7 ppm; FTIR (ATR): 3068 (w), 2926 (w), 1617 (m), 1457
(w), 1357 (w), 1231 (m), 1141 (w), 1053 (w), 1014 (m), 990 (w), 964
In the present study, the electronic energies, gradients and deriva-
tive couplings that appear in Equation (4) were obtained from the
COLUMBUS electronic structure program.[33] Extrema on the poten-
tial-energy surfaces were located employing correlation-consistent
polarized triple zeta (cc-pVTZ) basis sets[34] at a Davidson-corrected
second-order multireference configuration interaction (SO-MRCI)
level of theory. The level of electronic structure theory employed
for the dynamical simulations employed a (3s2p1d/2s1p) atomic
natural orbital basis set,[35] and the wavefunction expanded at the
first-order multireference configuration interaction (FO-MRCI) level.
This lower level of theory well reproduced the energies of the
structures shown below, with differences generally less than 0.3 eV
(see SI). All MRCI computations employed the same active space
definition for the underlying CASSCF reference (8 electrons/6 orbi-
tals), comprised of 3p orbitals, a s/s* pair, and the non-bonding
lone pair orbital at the carbene centre, as defined at the ground-
state equilibrium geometry.
(w), 892 (w), 849 (w), 786 (s), 752 (vs), 720 (w), 665 (w) cmÀ1
.
Experimental methods
The photodissociation experiments were carried out in a VMI
setup[30] described previously.[24c] The precursor/seeding gas mix-
ture was expanded through a pulsed solenoid valve at a 10 Hz rep-
etition rate. The carbene was generated in a resistively heated
(25 W) SiC tube attached to the valve. The free jet was skimmed
and entered the ion optics consisting of three plates (repeller, ex-
tractor, and ground), each 15 mm apart. Fields of 2317 and
1016 VcmÀ1 were applied to extract and accelerate the ions onto
the dual stage microchannel plate (MCP) detector equipped with
a phosphorescent screen. In order to detect only hydrogen ions,
the voltage at the back plate of the detector was gated by a push/
pull switch with a gate width of 100 ns. The resulting ion signal
was imaged onto a progressive-scan camera with a 2/3“ CCD chip
by an achromatic lens. The H-atom images were accumulated over
10000 laser shots, symmetrized and reconstructed by the pBASEX
algorithm,[31] employing Legendre polynomials up to 2nd order.
The position and momenta sampled by the simulation to generate
the initial trajectory basis functions were determined from the
ground state vibrational distribution. The trajectory basis functions
were placed on the “brightest” adiabatic electronic state, as deter-
mined by the computed transition dipole moment matrix element.
The spawning simulation was performed with 23 initial conditions
that gave rise to 603 spawned basis functions.
The unfocused output of a frequency-doubled Nd:YAG laser
pumped dye laser (1–2 mJ) was used for excitation. A vertical po-
larization of the laser was ensured with a Fresnel double rhomb. A
[1+1’] multiphoton ionization (MPI) was used to detect the H-atom
2
photofragments via the Lyman-a2S! P transition at 121.6 nm. The
Acknowledgements
output of a second frequency-doubled dye laser was focused into
a cell filled with 90 Torr of Kr and the generated VUV light was
then focused into the ionization region by a 100 mm MgF2 lens
mounted at the exit of the cell. The excited H-atoms were ionized
by the remaining fundamental. The ionization laser, delayed by
10 ns with respect to the excitation laser, was scanned over
Æ6 cmÀ1 around the Lyman-a transition to cover the whole Dop-
pler profile of the H-atoms. The two laser systems were synchron-
ized electronically to within 1 ns and polarized perpendicularly
with respect to each other. A background image was recorded
without the excitation laser and subtracted from the two-color
image to minimize contributions from one-color processes. The
time of flight mass spectra (TOF-MS) were recorded by monitoring
the current at the back plate of the MCP ion detector. They were
partially recorded at an ionization wavelength of 118 nm (generat-
ed by frequency-tripling in Xe) for better detection efficiency.
This work was supported by the German Science Foundation,
DFG through contract FI 575/8-2 and the Graduate Research
School 1221. J.G. acknowledges a fellowship by the Fonds der
Chemischen Industrie. R.J.M. acknowledges a fellowship from
the National Science and Engineering Research Council of
Canada.
Keywords: ab initio molecular dynamics · flash pyrolysis · gas-
phase reactions
intermediate
·
non-adiabatic processes
·
reactive
[1] a) P. Thaddeus, C. A. Gottlieb, R. Mollaaghababa, J. M. Vrtilek, J. Chem.
Soc. Faraday Trans. 1993, 89, 2125–2129; b) M. B. Bell, P. A. Feldman,
H. E. Matthews, L. W. Avery, ApJ 1986, 311, L89–L92; c) I. H. Park, V.
Wakelam, E. Herbst, Astron. Astrophys. 2006, 449, 631–639; d) S. Spezza-
no, S. Brunken, P. Schilke, P. Caselli, K. M. Menten, M. C. McCarthy, L. Biz-
zochi, S. P. Trevino-Morales, Y. Aikawa, ApJL 2013, 769, L19.
[2] J. M. Vrtilek, C. A. Gottlieb, P. Thaddeus, ApJ 1987, 314, 716–725.
[3] L. S. Bernstein, D. K. Lynch, ApJ 2009, 704, 226–239.
[4] J. P. Maier, G. A. H. Walker, D. A. Bohlender, F. J. Mazzotti, R. Raghunan-
dan, J. Fulara, I. Garkusha, A. Nagy, ApJ 2011, 726, 41.
[5] R. A. Seburg, E. V. Patterson, R. J. McMahon, J. Am. Chem. Soc. 2009,
131, 9442–9455.
Theoretical methods
The non-adiabatic dynamics after laser excitation were simulated
employing the ab initio multiple spawning (AIMS) formalism.[21a,32]
This approach expands the time dependent wavefunction as
a product of electronic and nuclear components, with the latter
described by a direct product basis of frozen Gaussian functions.
The number of these functions changes in time in order to effi-
ciently describe regions of nonadiabatic coupling between multi-
Chem. Eur. J. 2015, 21, 14486 – 14495
www.chemeurj.org
14494
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
[6] C. A. Taatjes, S. J. Klippenstein, N. Hansen, J. A. Miller, T. A. Cool, J.-K.
Wang, M. E. Law, P. R. Westmoreland, Phys. Chem. Chem. Phys. 2005, 7,
806–813.
[7] J. Giegerich, I. Fischer, J. Petersen, R. Mitric, Phys. Chem. Chem. Phys.
2014, 16, 6294–6302.
ment (Eds.: W. Domcke, D. R. Yarkony, H. Kçppel), World Scientific, Sin-
gapore, 2011.
[22] A. M. Mebel, W. M. Jackson, A. H. H. Chang, S. H. Lin, J. Am. Chem. Soc.
1998, 120, 5751–5763.
[23] C. Ochsenfeld, R. I. Kaiser, Y. T. Lee, A. G. Suits, M. Head-Gordon, J. Chem.
Phys. 1997, 106, 4141–4151.
[24] a) H.-J. Deyerl, I. Fischer, P. Chen, J. Chem. Phys. 1999, 111, 3441–3448;
b) J. Giegerich, I. Fischer, Phys. Chem. Chem. Phys. 2013, 15, 13162–
13168; c) M. Steinbauer, J. Giegerich, K. H. Fischer, I. Fischer, J. Chem.
Phys. 2012, 137, 014303; d) B. Noller, I. Fischer, J. Chem. Phys. 2007, 126,
144302.
[25] M. Quack, Chem. Phys. 1980, 51, 353.
[26] A. Ben-Shaul, R. D. Levine, E. R. Bernstein, J. Chem. Phys. 1974, 61,
4937–4938.
[27] D. W. Minsek, J. A. Blush, P. Chen, J. Phys. Chem. 1992, 96, 2017–2025.
[28] D. H. Mordaunt, D. L. Osborn, D. M. Neumark, J. Chem. Phys. 1998, 108,
2448–2457.
[29] D. W. Kohn, H. Clauberg, P. Chen, Rev. Sci. Instrum. 1992, 63, 4003–
4005.
[30] A. T. J. B. Eppink, D. H. Parker, Rev. Sci. Instrum. 1997, 68, 3477.
[31] G. Garcia, L. Nahon, I. Powis, Rev. Sci. Instrum. 2004, 75, 4989–4996.
[32] a) T. J. Martinez, M. Ben-Nun, R. D. Levine, J. Phys. Chem. 1996, 100,
7884–7895; b) T. J. Martínez, M. Ben-Nun, R. D. Levine, J. Phys. Chem. A
1997, 101, 6389–6402; c) M. Ben-Nun, J. Quenneville, T. J. Martinez, J.
Phys. Chem. A 2000, 104, 5161–5175.
[8] D. L. Osborn, K. M. Vogelhuber, S. W. Wren, E. M. Miller, Y.-J. Lu, A. S.
Case, L. Sheps, R. J. McMahon, J. F. Stanton, L. B. Harding, B. Ruscic,
W. C. Lineberger, J. Am. Chem. Soc. 2014, 136, 10361–10372.
[9] P. Thaddeus, M. C. McCarthy, M. J. Travers, C. A. Gottlieb, W. Chen, Fara-
day Discuss. 1998, 109, 121–135.
[10] a) R. A. Seburg, R. J. McMahon, Angew. Chem. Int. Ed. Engl. 1995, 34,
2009–2012; Angew. Chem. 1995, 107, 2198–2201; b) R. A. Seburg, E. V.
Patterson, J. F. Stanton, R. J. McMahon, J. Am. Chem. Soc. 1997, 119,
5847–5856; c) H. P. Reisenauer, G. Maier, A. Riemann, R. W. Hoffmann,
Angew. Chem. Int. Ed. Engl. 1984, 23, 641; Angew. Chem. 1984, 96, 596.
[11] a) J. A. Blush, H. Clauberg, D. W. Kohn, D. W. Minsek, X. Zhang, P. Chen,
Acc. Chem. Res. 1992, 25, 385; b) H. Clauberg, D. W. Minsek, P. Chen, J.
Am. Chem. Soc. 1992, 114, 99–107; c) P. Hemberger, B. Noller, M. Stein-
bauer, I. Fischer, C. Alcaraz, B. K. Cunha de Miranda, G. A. Garcia, H.
Soldi-Lose, J. Phys. Chem. A 2010, 114, 11269–11276.
[12] a) V. Lavallo, Y. Canac, B. Donnadieu, W. W. Schoeller, G. Bertrand, Sci-
ence 2006, 312, 722–724; b) S. J. Connon, Angew. Chem. Int. Ed. 2014,
53, 1203–1205; Angew. Chem. 2014, 126, 1225–1227.
[13] B. Chen, A. Y. Rogachev, D. A. Hrovat, R. Hoffmann, W. T. Borden, J. Am.
Chem. Soc. 2013, 135, 13954–13964.
[14] P. Hemberger, B. Noller, M. Steinbauer, K. Fischer, I. Fischer, J. Phys.
Chem. Lett. 2010, 1, 228–231.
[15] P. Hemberger, J. Koehler, I. Fischer, G. Piani, L. Poisson, J.-M. Mestdagh,
Phys. Chem. Chem. Phys. 2012, 14, 6173–6178.
[16] M. N. R. Ashfold, N. H. Nahler, A. J. Orr-Ewing, O. P. J. Vieuxmaire, R. L.
Toomes, T. N. Kitsopoulos, I. A. Garcia, D. A. Chestakov, S. M. Wu, D. H.
Parker, Phys. Chem. Chem. Phys. 2006, 8, 26–53.
[17] D. N. Butler, I. Gupta, Can. J. Chem. 1982, 60, 415–418.
[18] R. W. Hoffmann, A. Riemann, B. Mayer, Chem. Ber. 1985, 118.
[19] a) D. J. deFrees, A. D. McLean, ApJ 1986, 308, L31–L35; b) T. J. Lee, X.
Huang, C. E. Dateo, Mol. Phys. 2009, 107, 1139–1152; c) K. C. Lau, C. Y.
Ng, Chin. J. Chem. Phys. 2006, 19, 29–38; d) M. C. van Hemert, E. F. van
Dishoeck, Chem. Phys. 2008, 343, 292–302.
[33] H. Lischka, R. Shepard, I. Shavitt, R. M. Pitzer, M. Dallos, T. Müller, P. G.
Szalay, F. B. Brown, R. Ahlrichs, H. J. Bçhm, A. Chang, D. C. Comeau, R.
Gdanitz, H. Dachsel, C. Ehrhardt, M. Ernzerhof, P. Hçchtl, S. Irle, G. Ked-
ziora, Kovar, V. Parasuk, M. J. M. Pepper, P. Scharf, H. Schiffer, M. Schin-
dler, M. Schüler, M. Seth, E. A. Stahlberg, J.-G. Zhao, S. Yabushita, Z.
Zhang, M. Barbatti, S. Matsika, M. Schuurmann, D. R. Yarkony, S. R. Bro-
zell, E. V. Beck, J.-P. Blaudeau, M. M. Ruckenbauer, B. B. Sellner, F. F. Plass-
er, J. J. Szymczak, in COLUMBUS, an ab initio electronic structure pro-
gram, release 7.0, 2012.
[34] T. H. Dunning Jr., J. Chem. Phys. 1989, 90, 1007–1023.
[35] K. Pierloot, B. Dumez, P.-O. Widmark, B. O. Roos, Theor. Chim. Acta 1995,
92, 149.
[20] J. F. Stanton, Faraday Discuss. 2011, 150, 331–343.
[21] a) M. Ben-Nun, T. J. Martinez, Adv. Chem. Phys. 2002, 121, 439–512; b) R.
Received: April 26, 2015
´
ˇ ´
´
Mitric, J. Petersen, V. Bonacic-Koutecky, in Advanced Series in Physical
Chemistry, Vol. 17 Conical Intersections - Theory, Computation and Experi-
Published online on August 19, 2015
Chem. Eur. J. 2015, 21, 14486 – 14495
www.chemeurj.org
14495
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim