Ultrafast electron diffraction of transient cyclopentadienyl radical: A dynamic pseudorotary structure

Article abstract of DOI:10.1016/S0009-2614(01)01476-2

Ultrafast electron diffraction (UED) is applied here in the study of the reaction of cylcopentadienyl cobalt dicarbonyl (CpCo(CO)₂) which proceeds to give product structures. These structures were probed by picosecond electron pulses. The major product of the fragmentation was found to be the cyclopentadienyl radical. The dynamic nature of the radical was best represented by a pentagonal molecular structure having D_5h symmetry with elevated mean amplitudes of vibration. Comparisons between theory and experiment are presented. The structure is that of the transition state between the compressed (dienylic) and the elongated (allylic) conformations but with longer bond distances, reflecting the dynamics of the pseudorotary surface.

Full text of DOI:10.1016/S0009-2614(01)01476-2

26 February 2002
Chemical Physics Letters 353 (2002) 325–334
www.elsevier.com/locate/cplett
Ultrafast electron diffraction of transient
cyclopentadienyl radical: A dynamic pseudorotary structure
1
Hyotcherl Ihee, Jonathan S. Feenstra, Jianming Cao , Ahmed H. Zewail
*
Laboratory for Molecular Sciences, Arthur Amos Noyes Laboratory of Chemical Physics, California Institute of Technology,
Pasadena, CA 91125, USA
Received 12 November 2001
Abstract
Ultrafast electron diffraction (UED) is applied here in the study of the reaction of cylcopentadienyl cobalt dicar-
bonyl ðCpCoðCOÞ Þ which proceeds to give product structures. These structures were probed by picosecond electron
2
pulses. The major product of the fragmentation was found to be the cyclopentadienyl radical. The dynamic nature of
the radical was best represented by a pentagonal molecular structure having D5h symmetry with elevated mean am-
plitudes of vibration. Comparisons between theory and experiment are presented. The structure is that of the transition
state between the compressed (dienylic) and the elongated (allylic) conformations but with longer bond distances, re-
flecting the dynamics of the pseudorotary surface. Ó 2002 Published by Elsevier Science B.V.
1. Introduction
distorted, lower energy ring assumes either a
compressed (dienylic) or an elongated (allylic)
conformation, which are nearly energetically de-
generate (within 0.05 kcal/mol) [2]. Structural de-
viation from D5h symmetry is large with some
previously equal bond lengths now being different
The cyclopentadienyl radical (Cp), a classic li-
gand, has long played an important role in differ-
ent areas of chemistry, both theoretically [1–4] and
experimentally [5–8]. Cp is an orbitally degenerate
doublet radical subject to a Jahn–Teller distortion
into a pair of C2v symmetric conformers joined by
a pseudorotary surface. Quantum chemical theory
predicts that the Jahn–Teller distortion from the
D5h structure stabilizes Cp by 3.5 kcal/mol [2]. The
ꢀ
by as much as 0.1 A. However, due to the near
energetic degeneracy of the two isomers, Cp exists
in a dynamic state of pseudorotation. These in-
teresting electronic properties as well as its tran-
sient nature have made Cp attractive to theoretical
and experimental inquiry.
Cp is also of interest synthetically where it is
present in many important organometallic cata-
lysts. One such compound is cyclopentadienyl
* Corresponding author. Fax: +1-626-792-8456.
E-mail address: zewail@caltech.edu (A.H. Zewail).
cobalt dicarbonyl ðCpCoðCOÞ Þ, which consists of
1
2
Present address: National High Magnetic Field Laboratory,
Florida State University, 1800 East Paul Dirac Drive, Talla-
hassee, FL 32310, USA.
a Cp ring and two carbonyl ligands bound to a
cobalt metal center. What is known about the
0009-2614/02/$ - see front matter Ó 2002 Published by Elsevier Science B.V.
PII: S0009-2614(01)01476-2

326
H. Ihee et al. / Chemical Physics Letters 353 (2002) 325–334
complex structure of ðCpCoðCOÞ Þ has been pre-
2
viously determined through studies including the
use of microwave spectroscopy [9], X-ray crystal
diffraction [10,11], and conventional gas phase
electron diffraction (ED) [12]. The photo-induced
elimination of carbonyls [13] has also been the
focus of some recent work on CpCoðCOÞ .
2
Cp, and many other radical species, are gener-
ally unstable and short lived. This often makes
conventional experiments aimed at determining
their molecular structure not applicable. With this
in mind, here we apply ultrafast electron diffrac-
tion (UED) which was developed in this labora-
tory [14–16] to provide the structural information
of transient intermediates [17–23]. These UED
studies have focused on the structural change in
the elimination of an iodine atom from 1,1,2,2-
tetrafluoro 1,2-diiodoethane ðC2F4I2Þ which re-
sults in the formation of the classical intermediate
and not the speculated ‘bridged’ structure [20].
Similarly, the elimination of carbonyls from
FeðCOÞ5 with products such as FeðCOÞ4; Fe
ðCOÞ ; FeðCOÞ were observed and found to be
2
dependent on the total energy – the transient
Scheme 1. Three possible fragmentation pathways of
CpCoðCOÞ2.
FeðCOÞ molecular structure and spin state were
4
determined [17,21]. UED has also been able to
observe transients in non-equilibrium thermal
states as in the case of the ring opening of 1,3-
cylcohexadiene (CHD) [20,22] and in the case of
energy redistribution in cylcoheptatriene (CHT)
[20,22]. Most recently, the mysterious channel-
three process in pyridine was studied and the
structure of the open-ring diradical product was
characterized [23].
2. Experimental methods
The second-generation UED apparatus [16] was
used for this experiment. It is composed of a fem-
tosecond laser, a picosecond electron gun, a free-jet
expansion source, and a two-dimensional single-
electron detection system. In a typical UED ex-
periment [16,20], femtosecond laser pulses are used
to pump the sample molecules, and similarly we did
for this reaction; the reported data, however, were
made with nanosecond laser excitation since the
product is a long-lived species and we can utilize
the much higher energy per pulse; Nd:YAG laser
pulses (532 nm, 30 Hz) were directed into a po-
tassium dihydrogen phosphate (KDP) crystal and
the resulting frequency-doubled pulses (266 nm, 3
mJ/pulse, 30 Hz) were then directed and focused
into the gas sample beneath the needle of the free-
jet expansion source in the scattering chamber.
Femtosecond laser pulses from a colliding-pulse
mode-locked ring dye laser were amplified in a
In this Letter, we present our UED results of the
fragmentation of gas-phase CpCoðCOÞ2, a large
and complex molecule of 15 atoms. There are three
possible reaction channels as illustrated in Scheme
1. Femtosecond and nanosecond multi-photon
excitation of CpCoðCOÞ results in a total detach-
2
ment of all ligands from the cobalt center thereby
leaving a bare Cp ring. A subsequent picosecond
pulse of electrons allows us to directly observe the
molecular structure of the transient species. We
were able to discern the product species and es-
tablish that it is Cp in a dynamic pseudorotary
state. We compare our results with theory and
consider the dynamics of the structural changes.

H. Ihee et al. / Chemical Physics Letters 353 (2002) 325–334
327
four-stage pulsed dye amplifier pumped by the
Nd:YAG laser. A part of the amplified pulses (620
nm, 0.5 mJ/pulse, 30 Hz, ꢀ300 fs pulse width) were
first frequency-doubled with a KDP crystal, and
then focused onto a back-illuminated photocath-
ode in the electron gun to generate ultra-short
electron pulses (ꢀ2 Â 104 electrons per pulse with a
temporal width of ꢀ15 ps). The ultrashort electron
pulses were accelerated to 18.8 keV (de Broglie
the singlet states of the possible carbonyl elimi-
nated products, CpCoCO and CpCo. The geom-
etry of the triplet state of CpCoCO was taken from
a DFT calculation given in the literature [13].
Closed-shell restricted wavefunctions were used
for the singlet states and unrestricted wavefunc-
tions were used for calculations involving multi-
plets.
ꢀ
wavelength is 0.088 A) and focused into the scat-
tering volume. The two-dimensional diffraction
images were recorded with and without the pump
laser by a charge-coupled device camera at the end
of a phosphor scintillator/fiber optic/image inten-
sifier chain in the detection chamber.
The electron pulse, laser pulse, and molecular
beam were arranged in a crossed-beam geometry,
and the alignment of the three beams was con-
trolled to within 10 lm. The time-delay between
the pumping laser pulses and the probing electron
pulses was set at 11 ns. The beam waist of both the
electron beam and laser beam was adjusted to be
about 300 lm, and the distance between the
crossed beams and the detector (camera length)
was measured to be 10.24 cm. Data processing was
performed with the aid of a computer interface.
4. Data analysis
Two-dimensional diffraction images acquired
with and without the excitation laser light were
radially averaged to generate one-dimensional to-
tal intensity curves, Itot, from which experimental
modified molecular scattering ðsMðsÞÞ curves were
obtained. Where s is the momentum transfer pa-
rameter given by s ¼ ð4p=kÞ sinðh=2Þ; k is the de
Broglie wavelength of the incident electrons and h
is the scattering angle.
off
tot
The diffraction data ðI Þ recorded without la-
ser light provides a reference signal since it origi-
nated from only the ground state parent molecules
on
tot
ðCpCoðCOÞ Þ. In contrast, the data ðI Þ obtained
2
with the laser on comprises the contributions from
both the remaining parent molecules and newly
formed products. Thus, the difference between
them selects the reaction change induced by the
laser pulse in what we have termed the diffraction–
difference method [17]. In the diffraction–difference
CpCoðCOÞ was purchased from Strem Chem-
2
ical (min. 95% purity) and used without further
purification. The sample was transferred into a cell
and degassed by three cycles of freeze-pump-thaw.
To provide enough molecular gas density in the
scattering volume, the sample cell, gas line, and
nozzle were heated and the nozzle temperature was
maintained at 151 °C. The gas pressure in the
scattering chamber during the experiment was
about 4 Â 10À4 Torr, and the pressure at the
scattering volume was estimated to be a few Torr.
signal, ðDItot ¼ Itoont À I Þ, the contributions from
off
tot
unreacted molecules and the background signal,
which do not change in the course of the chemical
reaction, are eliminated, thus highlighting the sig-
nal from structural change.
For data analysis, DItot is converted into the
corresponding DsMðsÞ curve following the con-
ventional definition of sMðsÞ [25]. The difference
radial distribution curve, Df ðrÞ, was then obtained
by the sine Fourier transformation of the DsMðsÞ
curve. As the difference curves contain equal but
opposite contributions from the parent and prod-
uct structures, it is possible to determine the rela-
tive parent and product fractions, whose values
reflect the degree of reaction progress [17]. Once
this value is obtained from the fitting of the dif-
ference curve, the contribution from solely the
3. Computational methods
All calculations were performed using the
B3LYP density functional theory (DFT) method
in the platform of GAUSSIAN 98 for windows [24].
The 6-311++G** basis set was used to calculate
the geometry, vibrational frequencies, and energies
of ground state CpCoðCOÞ as well as the isomers
2
of Cp. The 6-31+G* basis set was used to calculate

328
H. Ihee et al. / Chemical Physics Letters 353 (2002) 325–334
product structure can be isolated by adding the
appropriately scaled parent diffraction signal to
the difference curves. In this product-only curve
ðsMðsÞpoÞ, the parent component present in the
DsMðsÞ curve is cancelled out and only the signal
from the products remains [18,22].
CpCoðCOÞ obtained with the pump laser off. The
2
best fit was obtained by floating several parameters
of the molecular structure of the parent molecule
until reaching the minimum v2. The final structural
parameters are shown and compared with the re-
sults of our density functional calculations. In the
analysis of the previous conventional ED experi-
Before fitting, the raw experimental data ðItotÞ
was divided by the atomic scattering curve ob-
tained by simulating Xenon scattering data to
compensate for the s5 fall-off of the raw data. This
modification of the data made it easier to fit a
polynomial to the background and therefore ob-
tain a better starting place for structural fitting.
The fit was performed in a least-squares manner to
minimize the scaled differences ðv2Þ between the
theoretical sMðsÞ curve and the experimental one
while varying the structural parameters. The
structural fitting parameters were in the form of a
geometrically consistent z-matrix. This permitted
fitting of the whole molecule based on only a few
independent variables. The mean amplitudes of
vibration ðlÞ of the inter-nuclear distances were
approximated using Mastryukov’s semi-empirical
equations [26,27].
The geometric parameters we present are the
result of a partial refinement due to some simpli-
fications intended to eliminate parameter correla-
tion effects in the Cp ring of both the parent and
the products. Specifically, the C–H bond distances
were not fit and instead taken as given by theory.
Also, only one C–C bond and two C–C–C angles
in the ring were fit. This allows us to use the one
refined bond distance and the two refined angles to
derive the rest by geometrical considerations using
DFT-determined results for differences between
the bond distances and angles. The error bars
represent one standard deviation of the least-
squares fit. The R values [25] which are often used
to represent the goodness of fit for ED data were
less than 0.08 for all final refinements reported
here; R < 0:1 is considered to represent a good fit.
ment [12], it was assumed that the CpCoðCOÞ had
2
fivefold symmetry for the ring and twofold sym-
metry as a whole, but our diffraction analysis and
electronic structure calculations show that this
assumption is an over-simplification.
From our data, the Co–Cring distances take
ꢀ
values of 1.93 and 2.26 A for the shortest and
longest distances, respectively; the average value of
ꢀ
all five Co–Cring distances is 2.09 A. The DFT
ꢀ
values vary from 2.09 to 2.15 A (average value of
ꢀ
ꢀ
2.13 A) and a single value (2.12 A) was reported
for all Co–Cring bonds in the conventional ED [12]
analysis. The Co–Cring distances reported here are
more consistent with X-ray crystallographic data;
from one report [11], the Co–Cring bonds are given
ꢀ
ꢀ
as 2.01–2.23 A (average value of 2.09 A) and from
another report [10] the Co–Cring bonds vary from
ꢀ
ꢀ
2.06 to 2.11 A (average value of 2.08 A). The
Co–Ccarbonyl distance of this work is within one
standard deviation of the previous ED data and
shorter than the theoretically determined value by
ꢀ
0.067 A. The C–C bonds within the Cp ligand take
ꢀ
values from 1.41 to 1.46 A and the average value
ꢀ
(1.44 A) is consistent with the only Cring–Cring bond
ꢀ
distance supplied by conventional ED (1.45 A).
Finally, the Ccarbonyl–O bond length ð1:211 Æ
ꢀ
:015 AÞ is also consistent with the conventional
ꢀ
ED data ð1:191 Æ :004 AÞ [12] and longer by 0.064
ꢀ
A than what is predicted by theory. The main in-
ter-nuclear distances are assigned in the f ðrÞoff
curve in Fig. 1, with excellent agreement between
the experimental data and the refined theoretical
model.
Next, we consider the diffraction–difference
curves in order to establish which reaction chan-
nel(s) shown in Scheme 1 is(are) dominant. Fig. 2
shows the comparison between the data and the-
ory for these three possible products. For a start-
ing point, the structural parameters of these
different species were obtained from our own DFT
calculations and also from the literature [13]. The
5. Results and discussion
In Fig. 1, we present the modified molecular
scattering intensity, sMðsÞoff , and the radial dis-
tribution function, f ðrÞoff , for the parent molecule

H. Ihee et al. / Chemical Physics Letters 353 (2002) 325–334
329
Fig. 1. The refined structure (top) of the parent molecule CpCoðCOÞ along with the modified molecular scattering intensity (bottom)
2
and the radial distribution function (center) obtained with the pump laser off. Black lines are experimental curves and green lines are
the least-squares fits. Shown also at the top are our average refined parameters compared with the average parameters from DFT.
contribution from the parent molecule can be
cancelled out by adding the appropriate fraction of
the parent molecular intensity to these difference
curves [18,22]. The resulting product-only curves
are shown in Fig. 3. It is clear from both Figs. 2
and 3 that Cp is the main product and that the
other reaction product structures are nearly ab-
sent. A fit including all three structures was also
performed by floating the fraction of each species
while keeping their structural parameters fixed at
the values obtained from quantum chemical cal-
culations and Cp is overwhelmingly favored.
According to our DFT calculations, the uphill
energy for CpCoðCOÞ going to Cp þ Co þ 2CO is
2
119 kcal/mol above the ground state of the parent.
Our photon energy is 107 kcal/mol and, therefore,
with one photon at a wavelength of 266 nm, only
carbonyl moieties can be removed. That the Cp

330
H. Ihee et al. / Chemical Physics Letters 353 (2002) 325–334
Fig. 2. Comparison between theoretical and experimental fits of the diffraction–difference curve for three possible product channels.
Both difference sMðsÞ curves (right) and f ðrÞ curves (left) are shown. Black lines are experimental curves and green lines are the
unrefined theoretical fits. The theoretical curves were obtained from density functional theory calculations. The Cp curve was ap-
proximated by a half and half mixture of elongated and compressed Jahn–Teller isomers, and the fit is not as good as the refined one
shown in Fig. 4 (see text).
ring is the main product indicates the reaction is
dominated by two-photon processes, consistent
with the high pulse energy (3 mJ) used in these
experiments. We did not study whether the Cp ring
was generated by a simultaneous two-photon ab-
sorption or by a successive fragmentation of other
intermediates upon further absorption of another
photon.
A least-squares fit of the difference curve was
performed while varying the structural parameters
of the product and the result is shown in Fig. 4. A
closer examination of the Df ðrÞ curve reveals the
rich details of structural change due to the deple-
tion of CpCoðCOÞ and formation of Cp. The first
2
ꢀ
distinct peak centered at ꢀ2 A indicates the de-
pletion of Co–Ccarbonyl and Co–Cring bonds, and the

H. Ihee et al. / Chemical Physics Letters 353 (2002) 325–334
331
Fig. 3. Comparison between the theoretical and experimental fits of the product-only curve for three possible product channels. Both
product-only sMðsÞ curves (right) and f ðrÞ curves (left) are shown. Black lines are experimental curves and green lines are the unrefined
theoretical fits. The theoretical curves were obtained from density functional theory calculations. The Cp curve was approximated by a
half and half mixture of elongated and compressed Jahn–Teller isomers, and the fit is not as good as the refined one shown in Fig. 5 (see
text).
ꢀ
second peak centered just below 3 A indicates the
and the carbonyls have been separated from the
cobalt center.
depletion of the Ccarbonyl ꢁ ꢁ ꢁ Cring and Co ꢁ ꢁ ꢁ O dis-
ꢀ
tances. The third peak centered at ꢀ3.5 A indicates
The product-only structure was refined by a
least-squares fit of the product-only curve and the
results are shown in Fig. 5. In this curve, the
contributions from the parent molecules are com-
pletely absent. The product-only f ðrÞ curve, illus-
trated in Fig. 5, simply shows two peaks: one for
the depletion of the Cring ꢁ ꢁ ꢁ Ccarbonyl and Cring ꢁ ꢁ ꢁ O
distances, and the last shoulder indicates the de-
pletion of O ꢁ ꢁ ꢁ C and O ꢁ ꢁ ꢁ O distances. The first
peak corresponding to the Co–Ccarbonyl and
Co–Cring separations clearly shows that the Cp ring

332
H. Ihee et al. / Chemical Physics Letters 353 (2002) 325–334
Fig. 4. Refined best fits obtained for the difference curves for the indicated reaction at the top. The bottom-right panel shows the
diffraction–difference modified molecular scattering curve and the bottom-left panel shows the diffraction–difference radial distribution
curve. Black lines are experimental curves and green lines are the least-squares fits. The top panel shows the molecular structures
involved in the reaction.
ꢀ
the C–C bonds at ꢀ1.4 A and the other for the
used the regular pentagonal structure to represent
the dynamics of all conformational change. The
DFT electronic structure of D5h Cp was then used
as the starting condition for the refinement. The
best fit of our data was obtained when the C–C
ꢀ
indirect C ꢁ ꢁ ꢁ C distances at ꢀ2.3 A. The final
structures obtained from the fit of the diffraction–
difference and product-only curve were identical.
In order to examine the nature of dynamic
structures, we initially used a 50–50 mixture of the
elongated and compressed Jahn–Teller isomers to
fit the difference and product-only diffraction data.
However, our final least-squares refinement using
the v2 criterion yielded results less favorable of a
structure composed of two ‘static structures’, see
Figs. 3 and 5. Instead, the fit equilibrium bond
lengths ðreÞ strayed significantly from the theoret-
ically determined Jahn–Teller structures, indicat-
ing the participation of an array of many dynamic
structures. This stretched ring of D5h symmetry (a
ꢀ
bond of the D5h ring was re ¼ 1:46 Æ 0:003 A.
ꢀ
DFT gave re ¼ 1:42 A, which is consistent with the
results from rotationally resolved laser-induced
fluorescence spectroscopy [7]. The DFT calcula-
ꢀ
tions refer to the bond distance of 1.42 A of the
D5h transition state (see top Fig. 5; center), and we
ꢀ
attribute the observed 0.04 A extension of the
equilibrium bond length to the pseudorotary dy-
namics about the Cp ring.
2
Cp radical has a degenerate electronic ð E1Þ
ground state if it is planar, of D5h symmetry; the
degeneracy is lifted by Jahn–Teller distortion and
the bonds become unequal in length. Due to the
near energetic degeneracy of the two Jahn–Teller
isomers and the low barrier to pseudorotation on
the surface connecting them, the two structures
ꢀ
regular pentagon) and the large error bars (Æ0:2 A
for some of the C–C bonds) obtained with a re-
finement of the 50–50 mixture suggested that an-
other more satisfactory dynamic model was
needed to match the data. With this in mind, we

H. Ihee et al. / Chemical Physics Letters 353 (2002) 325–334
333
Fig. 5. Top: A representation of pseudorotation in the Cp radical. In the center is the D5h transition state 3.5 kcal/mol [2] above the
circular pseudorotary surface connecting the elongated and compressed Jahn–Teller isomers of Cp. The transition states along the
circular path are estimated to be elevated by energies ranging from zero to ꢀ2 kcal/mol [6]. The dotted circles indicate two of the many
other possible paths. Bottom: The refined best fits obtained for the product-only modified molecular scattering curve (sMðsÞpo) and
radial distribution curve ðf ðrÞ Þ of the D5h molecular structure of Cp at 2600 K.The dynamic nature of the radical was best repre-
po
sented by a stretched structure having D5h symmetry with elevated mean amplitudes of vibration.
among the many in Fig. 5 are not static. Estimates
of this pseudorotary barrier around the circular
landscape (not across the center) in Cp range be-
tween zero and ꢀ2 kcal/mol [6]. If no energy is
channeled into the translational energies of the
fragments, a maximum of ꢀ104 kcal/mol in in-
ternal energy for the products is available. If this
energy is then completely thermalized between and
among all modes of Cp, the vibrational ensemble
temperature would be ꢀ2900 K. In reality, how-
ever, a significant fraction of the available energy
would go into translational kinetic energy of the
fragments and the equilibrium temperature of Cp
would be lower than this value.
A refinement of the molecular temperature,
based on the mean-amplitudes of vibration ðlÞ of
the bonds, was also performed and our best-fit
temperature of the D5h product was 2600 Æ 150 K
(see Fig. 5). This apparent high temperature is a
consequence of the highly entropic phase space,
which results from the many vibrations involved in
the transformations of the structures shown (and

334
H. Ihee et al. / Chemical Physics Letters 353 (2002) 325–334
[7] L. Yu, D.W. Cullen, J.M. Williamson, T.A. Miller, J.
Chem. Phys. 98 (1993) 2682.
not shown) in Fig. 5; note that the arrowed circle
represents just one path in the basin of the energy
landscape and a few of the many others are rep-
resented by dotted circles. The long bonds of the
structure and the high vibrational temperature
harmonize with the dynamic nature of pseudoro-
tation in this species. An active pseudorotary state
involves many bonds changing lengths as the
molecule constantly converts between elongated
and compressed isomers. The rapid movement of
Cp through these pathways at high internal energy
is represented by a hot and enlarged D5h structure
as obtained by UED, from bond lengths and mean
amplitudes of vibration.
[8] V.A. Korolev, O.M. Nefedov, Russ. Chem. Bull. 42 (1993)
1497.
[9] M.A. Roehrig, Q.-Q. Chen, S.T. Haubrich, S.G. Kukolich,
Chem. Phys. Lett. 183 (1991) 84.
[10] M.Y. Antipin, Y.T. Struchkov, A.N. Chernega, M.F.
Meidine, J.F. Nixon, J. Organomet. Chem. 436 (1992) 79.
[11] I.W. Nowell, D.R. Russell, J. Chem. Soc., Dalton Trans.
(1972) 2393, 2396.
[12] B. Beagley, C.T. Parrott, V. Ulbrecht, G.G. Young, J. Mol.
Struct. 52 (1979) 47.
[13] P.T. Snee, C.K. Payne, K.T. Kotz, H. Yang, C.B. Harris, J.
Am. Chem. Soc. 123 (2001) 2255.
[14] J.C. Williamson, A.H. Zewail, Proc. Natl. Acad. Sci. 88
(1991) 5021.
[15] M. Dantus, S.B. Kim, J.C. Williamson, A.H. Zewail, J.
Phys. Chem. 98 (1994) 2782.
[16] J.C. Williamson, J. Cao, H. Ihee, H. Frey, A.H. Zewail,
Nature 386 (1997) 159.
Acknowledgements
[17] H. Ihee, J. Cao, A.H. Zewail, Chem. Phys. Lett. 281 (1997)
10.
We wish to thank the National Science Foun-
dation and the United States Air Force Office of
Scientific Research for support of this work. We
would also like to thank Dr. Boyd Goodson for his
helpful discussion.
[18] J. Cao, H. Ihee, A.H. Zewail, Chem. Phys. Lett. 290 (1998)
1.
[19] J. Cao, H. Ihee, A.H. Zewail, Proc. Natl. Acad. Sci. USA
96 (1999) 338.
[20] H. Ihee, V.A. Lobastov, U. Gomez, B.M. Goodson, R.
Srinivasan, C.-Y. Ruan, A.H. Zewail, Science 291 (2001)
458.
[21] H. Ihee, J. Cao, A.H. Zewail, Angew. Chem. Int. Ed. 40
(2001) 1532.
References
[22] C.-Y. Ruan, V.A. Lobastov, R. Srinivasan, B.M. Good-
son, H. Ihee, A.H. Zewail, Proc. Natl. Acad. Sci. 98 (2001)
7117.
[1] H. Wang, K. Brezinsky, J. Phys. Chem. A 102 (1998) 1530.
[2] K. Pierloot, B.J. Persson, B.O. Roos, J. Phys. Chem. 99
(1995) 3465.
[23] V.A. Lobostov, R. Srinivasan, B.M. Goodson, C. Ruan,
J.S. Feenstra, A.H. Zewail, J. Phys. Chem. A 105 (2001)
11159.
€
[3] T.-K. Ha, R. Meyer, H.H. Gunthard, Chem. Phys. Lett. 69
(1980) 510.
[4] W.T. Borden, E.R. Davidson, J. Am. Chem. Soc. 101
(1979) 3771.
[24] M.J. Frisch et al., Gaussian Inc., Pittsburgh, PA, 1998.
[25] I. Hargittai, M. Hargittai, Stereochemical Applications of
Gas-Phase Electron Diffraction, VCH, New York, 1988.
[26] V.S. Mastryukov, S.J. Cyvin, J. Mol. Struct. 29 (1975) 15.
[27] V.S. Mastryukov, E.L. Osina, Zhu. Struk. Khim. 17 (1976)
172.
[5] G.R. Liebling, H.M. McConnell, J. Chem. Phys. 42 (1965)
3931.
[6] L. Yu, S.C. Foster, J.M. Williamson, M.C. Heaven, T.A.
Miller, J. Phys. Chem. 92 (1988) 4263.