Evidence of a double surface crossing between open- and closed-shell surfaces in the photodissociation of cyclopropyl iodide

Article abstract of DOI:10.1021/jp0037504

Gas-phase photodissociations of cyclopropyl iodide were conducted at 266 and 279.7 nm, and the radical products were probed by multiphoton ionization, with imaging of the resulting ions and their corresponding electrons. Solution-phase photodissociations of cyclopropyl iodide were also conducted with TEMPO-trapping of the radical dissociation products. In both gas and solution phases, allyl radical was found to be a direct product of the cyclopropyl iodide photodissociation. CASSCF calculations indicate that the allyl radical could be formed directly from photoexcited cyclopropyl iodide by way of two surface crossings between open- and closed-shell potential energy surfaces. Each surface crossing represents a point of potential bifurcation in the reaction dynamics. Thus, cyclopropyl iodide that is excited to a ¹(n,σ*) state can remain on an open-shell surface and generate the cyclopropyl radical and an iodine atom or can cross to a closed-shell (ion-pair) surface. The cyclopropyl cation that results from the surface crossing can undergo barrierless ring opening to the allyl cation before crossing back to an open-shell surface to generate allyl radical and an iodine atom. In this manner, both cyclopropyl radical and allyl radical can be formed as direct products of cyclopropyl iodide photodissociation.

Full text of DOI:10.1021/jp0037504

J. Phys. Chem. A 2001, 105, 1693-1701
1693
Evidence of a Double Surface Crossing between Open- and Closed-Shell Surfaces in the
Photodissociation of Cyclopropyl Iodide
†
†
‡
,†
Pamela A. Arnold, Bogdan R. Cosofret, Scott M. Dylewski, Paul L. Houston,* and
Barry K. Carpenter*,†
Department of Chemistry and Chemical Biology, Cornell UniVersity, Ithaca, New York 14853-1301, and
School of Applied and Engineering Physics, Cornell UniVersity, Ithaca, New York 14853-2501
ReceiVed: October 12, 2000; In Final Form: January 8, 2001
Gas-phase photodissociations of cyclopropyl iodide were conducted at 266 and 279.7 nm, and the radical
products were probed by multiphoton ionization, with imaging of the resulting ions and their corresponding
electrons. Solution-phase photodissociations of cyclopropyl iodide were also conducted with TEMPO-trapping
of the radical dissociation products. In both gas and solution phases, allyl radical was found to be a direct
product of the cyclopropyl iodide photodissociation. CASSCF calculations indicate that the allyl radical could
be formed directly from photoexcited cyclopropyl iodide by way of two surface crossings between open- and
closed-shell potential energy surfaces. Each surface crossing represents a point of potential bifurcation in the
1
reaction dynamics. Thus, cyclopropyl iodide that is excited to a (n,σ*) state can remain on an open-shell
surface and generate the cyclopropyl radical and an iodine atom or can cross to a closed-shell (ion-pair)
surface. The cyclopropyl cation that results from the surface crossing can undergo barrierless ring opening to
the allyl cation before crossing back to an open-shell surface to generate allyl radical and an iodine atom. In
this manner, both cyclopropyl radical and allyl radical can be formed as direct products of cyclopropyl iodide
photodissociation.
1
. Introduction
tions of the heats of formation of allyl radical and cyclopropyl
radical, the enthalpy for ring opening has been determined to
Photoinduced homolytic cleavage of alkyl-halogen bonds
1
3,14
be between -22.8 and -25.6 kcal/mol.
However, compu-
tational studies consistently show the reaction enthalpy to be
is a commonly employed method for the generation of alkyl
radicals. However, in some instances, solution-phase irradiations
of alkyl halides have been found to result in the formation of
products that can only be explained by ionic rearrangements of
8,12
approximately -32 kcal/mol. Whether the error in the values
for the enthalpy of ring opening lies in the experimental
measurements, in the calculations, or in both is unclear.
1
the alkyl fragments. Kropp has explained these observations
Presumably due to the high activation energy, isomerization
to allyl radical has only been observed to occur from vibra-
tionally excited cyclopropyl radical in the gas phase.1 Despite
the highly exothermic nature of the ring opening, it has not been
reported to occur in solution.
in terms of an electron-transfer mechanism. He suggests that
homolytic dissociation of alkyl halides generates a solvent-caged
radical pair, which can, in some cases, undergo electron transfer
to generate an ion pair. Presumably there should be no gas-
phase counterpart to this reaction.
5,16
1
Recent efforts in our laboratories have focused on the
generation of cyclopropyl radical by the photodissociation of
cyclopropyl iodide. As described below, the results suggest
involvement of an ion pair in both gas and solution phases.
Cyclopropyl iodide is a particularly suitable substrate for this
investigation, as the differing reactivities of cyclopropyl radical
and cyclopropyl cation make it possible to distinguish between
ionic and radical processes.
Cyclopropyl cation, on the other hand, can undergo isomer-
ization to allyl cation with little to no energy barrier and an
enthalpy change of approximately -30 to -36 kcal/mol.1
This difference in reactivity between the cyclopropyl radical
and the cyclopropyl cation makes it possible to differentiate
between radical and ionic pathways under specific reaction
conditions.
2,17
On the basis of the experimental results and ab initio
calculations reported herein, we propose a mechanism for the
photodissociation of cyclopropyl iodide involving two intersec-
tions between open- and closed-shell potential energy surfaces
The isomerization of cyclopropyl radical to allyl radical has
2-8
been the subject of many computational efforts
but few
experimental studies. Experimentally determined activation
enthalpies for the ring opening of cyclopropyl radical range from
(PES). The first intersection provides a means by which the
-11
1
9.1 to 22 kcal/mol.9 Computationally determined values for
closely associated cyclopropyl radical and iodine atom cross
from the open-shell (radical-pair) to the closed-shell (ion-pair)
PES. The resulting cyclopropyl cation can then undergo
barrierless ring opening to generate allyl cation, still closely
associated with iodine anion. A second intersection between
this closed-shell surface and the open-shell surface for allyl
iodide dissociation allows allyl radical and an iodine atom to
be formed.
the enthalpy barrier are in close agreement with the experimental
values.8
,12
The reaction enthalpy for the cyclopropyl radical ring opening
is not well-established. On the basis of experimental determina-
*
To whom correspondence should be addressed.
Department of Chemistry and Chemical Biology.
School of Applied and Engineering Physics.
†
‡
1
0.1021/jp0037504 CCC: $20.00 © 2001 American Chemical Society
Published on Web 02/13/2001

1
694 J. Phys. Chem. A, Vol. 105, No. 10, 2001
Arnold et al.
2
2
The apparent detection of PES intersections in the photodis-
I( P1/2) and I( P3/2) fragments using the 2 + 1 resonance-
sociation of cyclopropyl iodide and the reported examples of
ionic rearrangements in alkyl halide photodissociations lead us
to believe that conical intersections may be prevalent in alkyl
halide photodissociations.
enhanced multiphoton ionization (REMPI) schemes at 304.0 nm
(5p P0,1,2, D2 r 5p P1/2) and 304.6 nm (5p P2 r 5p P3/2),
3
1
2
3
2
2
respectively. Due to the large Doppler width of the I( P1/2) and
2
I( P3/2) fragments it was necessary to scan the probe laser over
the resonance to ensure that the images detected all iodine atom
velocities with equal sensitivity.
2. Experimental Section
The dissociation laser radiation at 266 nm was produced by
doubling the 532 nm output from a Nd:YAG laser using an
2
.1. Synthesis of Cyclopropyl Iodide. A 1-L, three-necked,
round-bottomed flask was equipped with a stir bar, a reflux
condenser, and a 500-mL addition funnel. Magnesium turnings
3.83 g, 0.16 mol), a crystal of I2, and 50 mL of ether were
8
3° KDP crystal. Typical powers obtained were about 20 mJ/
pulse with a pulse duration of 8-10 ns. The tunable light needed
(
2
2
to probe the I( P1/2) and I( P3/2) fragments at 304.0 and 304.6
nm, respectively, was generated by frequency doubling the
output of an injection-seeded, Nd:YAG-pumped dye laser;
typical powers achieved were 2.0 mJ/pulse with a pulse width
of 8-10 ns. The dissociation and the probe beams were directed
into the vacuum chamber and focused into the interaction region
by 25 and 7.5 cm focal length planoconvex lenses, respectively.
The polarizations of both laser beams were perpendicular to
the plane defined by the molecular and laser beams. The delay
time between the pump and probe lasers was set to 17 ns.
added to the reaction flask, and a solution of 1-bromocyclo-
propane (12.0 mL, 0.15 mol) in 350 mL of ether was added to
the addition funnel. Approximately 50 mL of the 1-bromocy-
clopropane solution was added to the magnesium turnings, and
the mixture was refluxed with stirring until the brown color
from the iodine disappeared. The remainder of the 1-bromocy-
clopropane solution was then added dropwise, with refluxing
and vigorous stirring over a period of 4 h. Reflux was continued
for 1 h, then the reaction flask was cooled in an ice bath. To
the chilled mixture of cyclopropylmagnesium bromide was
added I2 (41 g, 0.16 mol) gradually via a solid addition funnel.
The reaction was allowed to warm to room temperature while
stirring was continued for 1.5 h.
The reaction was quenched by slow addition of 2 N HCl.
The layers were separated, and the organic layer was washed
five times with aqueous saturated sodium thiosulfate, one time
with aqueous saturated sodium bicarbonate, and one time with
aqueous saturated sodium chloride. The organic layer was then
dried over magnesium sulfate and filtered. Ether was removed
by distillation at atmospheric pressure.
The imaging technique uses an electrostatic immersion lens
2
2
which serves to extract the ionized fragments I( P1/2) and I( P3/2)
from the interaction region and to focus ions with equal velocity
2
0
vectors to the same point on the detector. The magnification
factor of this electrostatic lens was measured to be 1.17 ( 0.03
3
by dissociating O2 at 225.65 nm and detecting the O( P2)
3
3
fragment using the O(3p P2,1,0 r 2p P2) 2+1 REMPI scheme.
The ionized fragments were accelerated into a field-free flight
tube mounted along the axis of the molecular beam. The ions
were imaged using a position sensitive detector consisting of a
chevron double microchannel plate (MCP) assembly coupled
to a fast phosphor screen. The image on the screen was recorded
with a 640 × 480 pixel CCD camera. Both the MCP and the
camera were electronically gated to collect signal corresponding
Cyclopropyl iodide was isolated from the product mixture
by flash column chromatography, eluting with 100% pentane.
Fractions containing 97-99% cyclopropyl iodide, as determined
by gas chromatography, were combined and pentane was
removed by distillation at atmospheric pressure. The remaining
oil was then fractionally distilled at atmospheric pressure. The
fraction boiling at 95-100 °C contained 99% cyclopropyl iodide
2
2
only to the mass of the I( P1/2) and I( P3/2) fragments. Signal
levels were kept below 300 ions per frame to avoid saturation
of the MCP. Data were accumulated typically for 80 000 total
laser shots.
(
3.63 g, 14% yield). The product is a colorless oil.
.2. Solution-Phase Photodissociation of Cyclopropyl
Iodide. Two separate 0.0536 M solutions of cyclopropyl iodide
22.6 mg, 0.134 mmol) in benzene (2.5 mL) were irradiated
The experimental setup was slightly modified for electron
imaging. Both the electrostatic lens and the flight tube were
carefully wrapped in a grounded sheet of µ-metal to provide
magnetic shielding. The original time-of-flight (TOF) tube,
which produced a TOF distance of 62.5 cm, was replaced by a
shorter tube with a TOF distance of 41.5 cm to compensate for
the greater velocity of electrons. This modification allowed the
entire electron signal to be focused on the detector. In the
electron-imaging experiments, only one laser was used for both
dissociation and ionization, and the experiments were conducted
in two separate wavelength regions. The radiation in the 266-
nm region was obtained in the same manner as described above,
while the laser light radiation in the region between 279 and
2
(
with an Ace-Hanovia 450-W medium-pressure mercury-vapor
lamp for 50 h, at 9.7 °C, in the presence of a large excess of
2,2,6,6-tetramethyl-1-piperidinyloxy (TEMPO) radical trap. One
solution contained 2.46 M TEMPO (0.962 g, 6.15 mmol) and
the other solution contained 1.28 M TEMPO (0.500 g, 3.20
mmol). The reaction progress and the change in product ratios
over time were monitored by gas chromatography. The reaction
products were characterized by NMR and GC/MS. The isolated
TEMPO-trapped products were shown through independent
irradiation to be stable under the reaction conditions.
2
85 nm was obtained by frequency doubling the output of an
2
.3. Gas-Phase Photodissociation of Cyclopropyl Iodide.
injection-seeded Nd:YAG-pumped dye laser using a 74° KDP
crystal. Typical powers for both regions were kept at low values
between 2 and 3 mJ/pulse to avoid nonresonant interactions.
Due to the large velocities of electrons, their TOF is very
short and arrival at the detection source is almost instantaneous.
Thus, to detect electrons, it was necessary to electronically gate
the MCP and the camera for the duration of the laser pulse.
The gas-phase photodissociation of cyclopropyl iodide was
investigated using ion imaging and electron imaging techniques.
The technique of ion imaging has been described in more detail
1
8,19
elsewhere.
Alkyl iodides were introduced into the chamber
by bubbling He through the samples, which were cooled in ice
and acetone at -10 °C. The mixture was expanded at 17 psi
through a pulsed 250 µm diameter nozzle and collimated by a
5
00 µm diameter skimmer mounted 0.5-1 cm from the nozzle
Due to these modifications to the apparatus for electron
imaging, new calibrations were needed for the magnification
factor and electron time-of-flight. These calibrations were
achieved by carrying out state-selected ionizations of atomic
orifice. Further downstream, the molecular beam was crossed
at right angles by two counterpropagating laser beams, one used
to dissociate the alkyl iodide and the other to probe the resulting

Photodissociation of Cyclopropyl Iodide
J. Phys. Chem. A, Vol. 105, No. 10, 2001 1695
Figure 1. Raw iodine image from the 266-nm dissociation of
3
2
2
cyclopropyl iodide followed by 304.6-nm (5p P r 5p P3/2) ionization
of iodine. The polarization of the dissociation light was parallel to the
vertical direction of the figure.
iodine at 280.99, 280.69, 281.73, and 279.71 nm to produce
electrons with known energies. Jung et al. have published the
2
+ 1 REMPI lines of atomic iodine between 277 and 313 nm
+
21
and their branching ratios to the various available states of I .
These branching ratios and the energies of the I states published
+
Figure 2. Total relative translational kinetic energy distributions of
2
2
fragments from 266-nm photodissociations of cyclopropyl iodide,
by Wang et al. were used to calculate the energies of the
electrons from ionization of iodine. The electron time-of-flight
and magnification factor were then calibrated by adjusting their
values so that the experimentally obtained energy distributions
coincided with the known electron energies. These calibrated
values were then used in the processing of all electron images.
Calibration images were collected periodically to confirm that
the calibrations remained constant.
3
2
probing the 304.6-nm (5p P
iodine (s) and probing the 304.0-nm (5p P0,1,2, D
2
r 5p P3/2) REMPI transition of atomic
3
1
2
2
r 5p P1/2) REMPI
transition of atomic iodine (]). The maximum possible kinetic energy,
assuming dissociation to cyclopropyl radical and an iodine atom, is
denoted by the vertical lines. The solid line at 1.87 eV is the maximum
2
KE for formation of I P3/2 and the dashed line at 0.93 eV is the
2
maximum KE for formation of I P1/2. Horizontal error bars representing
the 0.09 eV uncertainty in the maximum kinetic energy are shown just
below the KE axis of the graph.
2
.4. Computational Methods. Calculations were carried out
23
using Gaussian 98 on a 500-MHz dual-processor Compaq
DS20E workstation and using GAMESS on a 466-MHz G3
Macintosh. Conical intersections were located by use of the
conservation of momentum. The total relative kinetic energy
distributions of the iodine and alkyl radical fragments formed
upon 266-nm dissociation of cyclopropyl iodide are shown in
Figure 2.
The anisotropy of the angular distribution is described by
the parameter â in I(θ) ∝ [1 + âP2(cos θ)], where θ is the
angle between the electric vector of the dissociation light and
24
25
routine incorporated in Gaussian 98. Orbitals were visualized
using MacMolPlt26 and GaussView 2.0.
3
. Results
.1. Gas-Phase Photodissociation of Cyclopropyl Iodide.
3
the recoil direction and P (cos θ) is the second Legendre
2
The radical fragments generated from a photodissociation
process scatter from the center of mass in three dimensions.
The velocities of the fragments relative to the center of mass
are proportional to the square root of their kinetic energies.
When these fragments are ionized and then accelerated to the
detector, the image that is observed is a two-dimensional
projection of the three-dimensional distribution of ions onto the
plane of the detector, such as shown in Figure 1. The intensity
at each pixel is proportional to the number of ions detected at
that position. The three-dimensional velocity distribution of the
fragments relative to the center of mass is reconstructed from
the two-dimensional image by taking the inverse Abel transform
polynomial. Fits to the data reveal that â ) 1.88 ( 0.27 for the
channel producing the ground state of the I atom and â )
1.91 ( 0.12 for the channel producing the spin-orbit excited
state of I.
3.1.2. Electron Imaging. Photodissociations of cyclopropyl
iodide and allyl iodide and ionizations of the resulting fragments
were independently carried out at both 266 and 279.7 nm. The
electron energy distributions resulting from the 266 nm dis-
sociations and ionizations are shown in Figure 3. It can be seen
that three distinct populations of electrons at 0.18 eV (A), 1.0
eV (B), and 2.4 eV (C) are generated from both the cyclopropyl
iodide and the allyl iodide experiments.
These energy distributions were obtained by dissociating and
ionizing near the center of the molecular beam profile, where
the density of molecules is at its greatest and formation and
ionization of clusters is highly probable. It is notable that when
the dissociations and ionizations were carried out at the more
dilute front edge of the molecular beam, populations A and C
dramatically decreased relative to B. Additionally, upon pho-
todissociation and ionization of methyl iodide under the same
experimental conditions and near the center of the molecular
beam, populations A and C were again observed, but B was
not. From this it was concluded that populations A and C were
from ionization of clusters, whereas B was from the ionization
of a common product formed from both the allyl iodide and
cyclopropyl iodide dissociations.
27,28
of the symmetrized raw image.
The angular, speed, and
kinetic energy distributions of the fragments (also relative to
the center of mass) are then derived from the reconstructed three-
dimensional distribution.
3
.1.1. Ion Imaging. The total kinetic energy distribution of
the radical fragments that were formed from dissociation of
cyclopropyl iodide at 266 nm was determined using ion imaging
techniques. The 266 nm dissociation of cyclopropyl iodide was
2
2
found to generate iodine radicals in both the P3/2 and P1/2 spin-
orbit states, which were ionized at 304.6 and 304.0 nm,
respectively. Figure 1 shows the raw image from the 304.6 nm
3
2
(
5p P2 r 5p P3/2) ionization of iodine. The relative kinetic
energy distributions of the iodine fragments were then deter-
mined from the observed relative velocity distributions of the
iodine ions, and the kinetic energy distributions of the corre-
sponding alkyl fragments were determined on the basis of
When dissociations of cyclopropyl iodide and allyl iodide
and ionizations of the resulting alkyl fragments were carried

1696 J. Phys. Chem. A, Vol. 105, No. 10, 2001
Arnold et al.
4
.1. Gas-Phase Ion Imaging. It was expected that the gas-
phase photodissociation of cyclopropyl iodide at 266 nm would
result in the formation of iodine atoms and vibrationally excited
cyclopropyl radicals. It should then be possible to determine
the internal energy distribution of the cyclopropyl radicals using
eq 1:
Eint ) hV - BDE - KEtot
(1)
In eq 1, Eint is the internal energy of the cyclopropyl fragments,
hV is the energy of the dissociating light, and BDE is the C-I
bond dissociation enthalpy of cyclopropyl iodide. KEtot is the
total relative kinetic energy of the cyclopropyl fragments and
the iodine atoms, determined as previously described.
4.1.1. Estimation of C-I Bond Dissociation Enthalpy.To our
knowledge, there are no direct experimental determinations of
the C-I bond dissociation enthalpy (BDE) of cyclopropyl
iodide. There is one estimate of 58.6 kcal/mol in the literature,29
but this is based on a value of ∆Hf° ) 61.3 kcal/mol for the
cyclopropyl radical, which was deduced by studying the kinetics
Figure 3. Electron energy distributions from 266 nm ionizations of
allyl iodide (]) and cyclopropyl iodide (_) photodissociation products.
Three distinct populations of electrons were observed at 0.18 eV (A),
30
of hydrogen atom abstraction from cyclopropane. More recent
3
1
1
.0 eV (B) and 2.4 eV (C).
measurements, based on the gas-phase acidity of cyclopropane
and the electron affinity of the cyclopropyl radical,32 bring the
heat of formation up to 67.7 kcal/mol. This correction would
raise the estimated BDE to 65.0 kcal/mol.
An alternative estimate of the C-I BDE can come from
assuming that the hypothetical reaction shown in eq 2 is
thermoneutral:
2
9
With that assumption, one can use the experimental BDEs of
3
3
34
isopropyl bromide, isopropyl iodide, and cyclopropyl bro-
3
5
mide to estimate the desired quantity. The value turns out to
be 64.0 kcal/mol.
As independent estimates, we have also carried out ab initio
calculations on the isodesmic reactions shown in eqs 3 and 4:
Figure 4. Ratio of products cyclopropyl 2,2,6,6-tetramethyl-piperidin-
-yl ester (A) and allyl 2,2,6,6-tetramethyl-piperidin-1-yl ester (B) as
1
a function of time and concentration of TEMPO radical trap: (2) 2.46
M TEMPO and (0) 1.28 M TEMPO.
out at 279.7 nm, coinciding electron distributions were again
detected, in this case at 0.80 eV. This result is again consistent
with the formation of a common product from both allyl iodide
and cyclopropyl iodide dissociations.
The calculations were carried out at the (U)MP2 level and
used a 3-21G* basis set on iodine and a 6-31G* basis set on
the hydrocarbon radicals. The calculated ∆H° values for eqs 3
and 4 were respectively -7.51 and -10.02 kcal/mol. These
values, in combination with the experimental C-I BDEs of
isopropyl iodide and methyl iodide, give values of respec-
tively 62.4 and 66.9 kcal/mol for the C-I BDE of cyclopropyl
iodide.
3
.2. Solution-Phase Photodissociation of Cyclopropyl
Iodide. The TEMPO-trapped products of solution-phase pho-
todissociation of cyclopropyl iodide were determined to be
cyclopropyl 2,2,6,6-tetramethyl-piperidin-1-yl ester (A) and allyl
2
,2,6,6-tetramethyl-piperidin-1-yl ester (B), as well as cyclo-
3
4
36
propyl iodide and allyl iodide. A plot of the ratio of products A
to B as a function of time and TEMPO concentration is shown
in Figure 4.
In summary, the four independent estimates of the C-I BDE
of cyclopropyl iodide give a value of 64.5 ( 2 kcal/mol.
Following the normal convention, this estimate of the BDE
refers to the formation of the fragments in their ground states.
However, in the gas-phase photodissociations we were able to
4
. Discussion
The gas and solution-phase results suggest that allyl radical
is formed as a direct product of cyclopropyl iodide photodis-
sociation. In the sections 4.1 through 4.4 we will provide a
detailed discussion of our interpretation of the results from each
experiment and thereby explain how this conclusion was
reached. Further details on a possible mechanism for the
formation of allyl radical from cyclopropyl iodide photodisso-
ciation were obtained through ab initio calculations. These will
be discussed in sections 4.5 and 4.6.
2
detect the iodine atom both in its ground P3/2 (which we will
2
call the I channel) state and in the spin-orbit excited P1/2 state
(the I* channel). It is consequently convenient to define a second
BDE for this latter event. Its value, 86.2 ( 2 kcal/mol, differs
from the ground-state value by an amount equal to the energy
-1
difference (7603 cm ) between the two iodine-atom spin-orbit
states.

Photodissociation of Cyclopropyl Iodide
.1.2. Determination of Eint: EVidence of an AlternatiVe
J. Phys. Chem. A, Vol. 105, No. 10, 2001 1697
4
Dissociation Mechanism.Using eq 1 and the value of 64.5 ( 2
kcal/mol (2.79 ( 0.09 eV) for the C-I BDE, a maximum
possible total kinetic energy is calculated to be 1.87 ( 0.09 eV
2
when dissociation to cyclopropyl radical and I P3/2 is carried
out at 266 nm. This maximum energy would be achieved in
the improbable event that Eint were equal to zero or, in other
words, if the cyclopropyl radical were generated with no
vibrational or rotational energy. For dissociation to cyclopropyl
2
radical and I P1/2, the maximum translational KE would be
0
.93 ( 0.09 eV. It is noteworthy that significant fractions of
the fragments exceed these maximum values in both the I and
I* channels, as shown in Figure 2. This result would imply that
some of the cyclopropyl radicals are generated with negative
internal energy! Thus, we are led to conclude that the radical
fragments of higher kinetic energy are generated from an
alternative dissociation process with a lower BDE. A direct
dissociation of cyclopropyl iodide into allyl radical and iodine
seems the most likely alternative. Adding the computed 1.39
eV exothermicity for ring opening of cyclopropyl to allyl
radical to the maximum translational KE values would shift
the vertical lines in Figure 2 up to 3.26 eV for the I channel
and 2.32 eV for the I* channel. The data reveal that none of
the experimental kinetic energies exceed these revised limits.
Figure 5. RRKM-determined rate constants for the gas-phase ring
opening of cyclopropyl radical plotted as a function of the internal
energy of the cyclopropyl radical (right curve). Internal energy
distribution of alkyl fragments formed from photodissociation of
cyclopropyl iodide in the I channel (left curve).
8
,12
TABLE 1: Input Parameters Used for the RRKM
Calculation of the Cyclopropyl Radical Ring Opening
a
cyclopropyl radical
relative energy
no. of oscillators
no. of internal rotors
[0]
18
0
4
.2. Detection of Allyl Radical by Electron Imaging. To
determine whether allyl radical was formed from the gas-phase
photodissociation of cyclopropyl iodide, we ionized the alkyl
fragments and imaged the resulting electrons. The results of
these electron imaging experiments, shown in Figure 3, indicate
that a common product is formed from the independent
photodissociations of cyclopropyl iodide and allyl iodide. To
test our postulate that this common product is the allyl radical,
we have calculated the approximate values for the electron
kinetic energies formed from ionization of allyl radical at 266
and 279.7 nm and compared these values to the observed
electron kinetic energies. The calculated values come from eq
-
1
frequencies (cm
)
697.0, 815.0, 840.0, 849.0, 999.0, 1130.0,
1169.0, 1187.0, 1216.0, 1229.0, 1300.0,
1
3
572.0, 1608.0, 3239.0, 3245.0, 3312.0,
324.0, 3355.0
transition state between cyclopropyl radical and allyl radical
relative energy (kcal/mol)
no. of oscillators
21.0
17
0
no. of internal rotors
-
1
frequencies (cm
)
464.0, 590.0, 659.0, 793.0, 879.0,
9
1
3
44.0, 1140.0, 1176.0, 1300.0,
408.0, 1561.0, 1614.0, 3251.0,
286.0, 3310.0, 3359.0, 3391.0
5
, in which KE is the kinetic energy of the electrons, hV is the
total energy of the photons absorbed, and IP is the vertical
ionization potential for the allyl radical.
2
moments of inertia (amu·Å )
16.19, 36.34, 46.09
20.94, 24.60, 38.19
2
moments of inertia (amu·Å )
a
Parameters were determined from CASPT2(3,3)/cc-pVTZ//CASS-
CF(3,3)/cc-pVTZ calculations.12
KE ) hV - IP
(5)
The vertical and adiabatic ionization energies of allyl radical
have experimentally been found to coincide at 8.13 eV,
it is formed still remains. The ion imaging results point toward
the existence of a mechanism by which cyclopropyl iodide
dissociates directly to allyl radical. However, the radical
fragments with lower kinetic energy (thus, higher internal
energy) may well come from the simple homolytic dissociation
of cyclopropyl iodide to form cyclopropyl radical and an iodine
atom. If such is the case, then these vibrationally excited
cyclopropyl radicals could also generate allyl radical by
subsequent ring opening.
3
7,38
indicating similar equilibrium geometries for the allyl radical
and cation. Although the allyl radicals produced in the photo-
dissociation of allyl iodide are vibrationally excited, because
of the similar geometries for the radical and cation, it is likely
that the Franck-Condon factors governing the ionization will
favor transitions to similarly vibrationally excited states of the
cation. Therefore, we presume that ionization of the allyl radical
generated in these studies would require approximately 8.13
eV. On the basis of this value, the energy of electrons from
Approximate rate constants for the ring opening of cyclo-
propyl radical as a function of the internal energy of the
1
+ 1 REMPI ionizations of allyl radical at 266 and 279.7 nm
39
cyclopropyl radical have been calculated using RRKM theory
are calculated to be approximately 1.19 and 0.73 eV, respec-
tively. The agreement between the expected values for electron
energies and the experimentally observed electron populations
at 1.0 and 0.80 eV provides support for the postulate that allyl
radical is the common product to which these electrons can be
assigned.
and are plotted in Figure 5. The structures, energies, frequencies,
and moments of inertia of the cyclopropyl radical and the
transition state to allyl radical were calculated at the CASPT2-
12
(
3,3)/cc-pVTZ//CASSCF(3,3)/cc-pVTZ level and are listed in
Table 1.
The observed internal energy distribution of the alkyl
4
.3. RRKM Calculations on the Ring Opening of Cyclo-
fragments formed in the I channel following 266 nm dissociation
of cyclopropyl iodide is also shown in Figure 5. It can be seen
that the cyclopropyl radicals with internal energy of 22 kcal/
mol can ring open on a time scale of approximately 2 ns, while
propyl Radical. While allyl radical does, indeed, appear to be
formed from the photodissociation of cyclopropyl iodide within
the 8-10 ns time scale of the experiments, the question of how

1698 J. Phys. Chem. A, Vol. 105, No. 10, 2001
Arnold et al.
TABLE 2: Calculated Electronic Transitions of Cyclopropyl
Iodide
principal
configuration
state
symmetry
vertical excitation
wavelength (nm)
oscillator
strength
38 f 39
37 f 39
36 f 39
35 f 39
34 f 39
A′′
A′
A′
A′′
A′
222.6
212.3
144.1
143.8
127.0
0.0043
0.0046
0.5665
0.0051
0.2841
Figure 6. Two mechanisms for the formation of allyl radical from
cyclopropyl iodide photodissociation: (1) direct formation from the
a
Principal configurations of each excited state are defined by
*
(n,σ ) photoexcited state of cyclopropyl iodide and (2) ring opening of
transitions from the ground state. The numbers in column 1 refer to
the orbitals depicted in Figure 7.
vibrationally excited cyclopropyl radical, denoted by asterisk, in
competition with vibrational cooling by collision partner M.
those with an internal energy of 30 kcal/mol can ring open on
a time scale of approximately 70 ps.
It can be concluded that the experimentally observed allyl
radical could be formed from the ring opening of vibrationally
excited cyclopropyl radical, and it could also be formed from
the direct photodissociation of cyclopropyl iodide in the gas
phase. These two pathways for formation of allyl radical are
depicted in Figure 6. Since the former process is estimated to
occur on the order of picoseconds to nanoseconds, while the
latter could occur on a femtosecond time scale, the availability
of femtosecond time-resolved electron imaging could provide
further insight into the origin of the allyl radical. Such
experiments have not been carried out at this time.
4
.4. Solution-Phase Photodissociation of Cyclopropyl
Iodide. The solution-phase photodissociation of cyclopropyl
iodide was expected to produce cyclopropyl radical, which
would then be trapped by TEMPO with a rate constant on the
order of 109
M
-1
s . With an enthalpic barrier of ap-
-1 40
Figure 7. Molecular orbitals referred to in column 1 of Table 2. ψ38
is the HOMO and ψ39 is the LUMO of ground-state cyclopropyl iodide,
according to the calculations.
proximately 22 kcal/mol, the thermal ring opening of cyclo-
propyl radical is unlikely, since collisional cooling (see Figure
6
) of any vibrationally hot radicals should be 4-5 orders of
magnitude faster than the ring opening. To our knowledge, ring
opening of cyclopropyl radical has not been observed in solution
phase.
state of cyclopropyl iodide is gained from the iodine image
shown in Figure 1. The near cos θ dependence of the recoil
2
velocity distribution on the electric vector of the dissociation
light is indicative of a parallel transition. Since excitation occurs
from an A′ state and the transition is parallel, the initially formed
photoexcited state must also be of A′ symmetry.
In an effort to determine the nature of the electronic excited-
state accessed by the 266 nm photolysis, we carried out a CIS
calculation on the MP2-optimized geometry of cyclopropyl
iodide, using the basis set described earlier. Five singlet
electronic excited states were calculated. The results are
summarized in Table 2. The orbitals cited in this table are
depicted in Figure 7.
However, upon solution-phase photodissociation of cyclo-
propyl iodide, TEMPO-trapped products from both allyl and
cyclopropyl radicals were found. As shown in Figure 4, the
product ratio was found to be only slightly dependent on the
concentration of TEMPO. If ring opening were occurring in
competition with TEMPO-trapping of the cyclopropyl radical,
then doubling the concentration of TEMPO should double the
ratio of TEMPO-trapped cyclopropyl radical to TEMPO-trapped
allyl radical. The observed product ratio showed a small
dependence on the concentration of TEMPO, but not this
expected first-order dependence. These results suggest that allyl
radical is formed as a direct product from dissociation of
photoexcited cyclopropyl iodide.
A slight time dependence was observed in the ratio of
products. While this phenomenon has not been investigated in
detail, it can be explained by recombination of C3H5 radicals
and iodine atoms (probably from solvent-caged pairs) to give
products that are susceptible to further photolysis. This process
would lead to increasing concentrations of allyl iodide in the
starting material and hence decreasing ratios of cyclopropyl to
allyl TEMPO adducts. The viability of this explanation was
demonstrated by the direct observation of allyl iodide in the
TEMPO-trapping experiments as well as by an independent
experiment showing that cyclopropyl iodide could be converted
to allyl iodide when photolyzed in absence of TEMPO.
The calculations suggest that the longest wavelength absorp-
tion of cyclopropyl iodide corresponds to a nominally forbidden
transition from the A′′ p-type lone pair of iodine to the σC-I*
orbital. Promotion of an electron from the A′ p-type lone pair
to σC-I* is predicted to be of somewhat higher energy. However,
given that the iodine image in Figure 1 shows that the state
accessed experimentally is of A′ symmetry, it is likely that these
two electronic states are really reversed in relative energy. This
disagreement, as well as the fact that the predicted absorption
maximum of 212.3 nm does not agree very well with the
experimental value of 261 nm (measured in pentane solution),
are discrepancies of a magnitude that is common for CIS
calculations, which provide only the most rudimentary descrip-
41
tion of electronic excited states. However, the results probably
are reliable enough to conclude that the allowed (σ,σ*) transition
of the C-I bond would not be accessible with the 266-nm light
used in the experiment.
4
.5. Electronic Excited States of Cyclopropyl Iodide.
Information about the nature of the initially formed photoexcited

Photodissociation of Cyclopropyl Iodide
J. Phys. Chem. A, Vol. 105, No. 10, 2001 1699
Figure 9. Geometries of the lowest energy points on the seams of
1
intersection between the closed-shell surface and the A′′ (n,σ*)
1
(
structure I) and between the closed-shell surface and the A′ (σ,σ*)
surface (structure II).
Figure 8. Schematic representation of a double surface-crossing
mechanism for the direct formation of allyl radical and an iodine atom
from photolysis of cyclopropyl iodide. The orange-red surface represents
a closed-shell electronic state. The yellow-green-blue surface represents
an open-shell singlet state, with the barrier to ring opening of the
cyclopropyl radical depicted on the front face of the box. The red arrow
1
shows the proposed path from the initially generated (n,σ*) excited
state to the allyl radical and iodine atom products.
4.6. Intersections of Closed-Shell and Open-Shell Potential
Energy Surfaces for Cyclopropyl Iodide. The experiments
described above provide strong indications that there must be a
direct pathway from the optically accessed electronic excited
state of cyclopropyl iodide to allyl radical and an iodine atom.
An appealing possibility would be that this pathway consists
of two surface crossings. The first would be a crossing between
1
the initially generated open-shell (n,σ*) state and a closed-
shell ion-pair state. Several types of high-level ab initio theory
have suggested that the cyclopropyl cation has no barrier to
1
2
ring opening, so the cyclopropyl cation/iodide ion pair would
be expected to transform spontaneously to an allyl cation/iodide
ion pair. A second surface crossing might then permit back
electron transfer, with consequent generation of an allyl radical
and an iodine atom. A double surface crossing of this kind would
Figure 10. Energy and principal configurations of the two states at
the first crossing (structure I in Figure 9).
permit the system to sidestep the 22 kcal/mol barrier to adiabatic
ring opening of the cyclopropyl radical.10,11,42 The hypothetical
The first intersection corresponds closely to that anticipated
in the schematic representation of Figure 8. However, the second
does not. Rather than crossing back to the original A′′ open-
shell surface, this conical intersection connects the closed-shell
surface with an A′ open-shell surface. Thus the diagram in
Figure 8 really should have one closed-shell and two open-
shell surfaces. However, we have retained the simpler repre-
sentation both for purposes of visual clarity and also because
the A′ and A′′ open-shell singlet states must approach degen-
eracy as the C-I distance approaches infinity.
We have looked for an intersection between the closed shell
state and the A′ open-shell state at the ring-closed geometry
but have found none. Similarly, no crossing between closed-
shell and A′′ open-shell singlet states could be found with a
ring-opened geometry. It is unclear whether the failure to locate
these crossings derives merely from limitations of the calcula-
tions or whether the crossings really do not exist.
mechanism is summarized schematically in Figure 8.
For this mechanism to be correct, it is a necessary condition
that there exist (at least) two crossings between closed-shell
and open-shell singlet states of the system. One crossing should
occur at a long C-I distance but with a more-or-less normal
distance of the C-C bond to be broken. The second should
occur at a similar C-I distance but with a substantially elongated
C-C bond. Two crossings that conform to these criteria have
in fact been found by carrying out state-averaged CASSCF-
(4,4) calculations, using the basis set described earlier.
The first crossing is a symmetry-allowed intersection of the
1
A′′ (n,σ*) state (corresponding to one of the two low-lying
excited states of cyclopropyl iodide) and an A′ closed-shell state.
The second corresponds to a same-symmetry conical intersec-
43
tion between the A′ closed-shell state and an open-shell singlet
A′ (σ,σ*) state. The geometries of the lowest energy structures
along these two seams of intersection are depicted in Figure 9.
The orbitals involved in the CASSCF calculation and the
principal configurations of each electronic state are summarized
in Figures 10 and 11.
The calculations place the second intersection at an energy
5.8 kcal/mol above the first. However, the level of theory
employed for the calculations is not sufficient for that to be
considered a reliable figure. The system has substantial internal
energy from the 266 nm (107.5 kcal/mol) photon that initiates

1700 J. Phys. Chem. A, Vol. 105, No. 10, 2001
Arnold et al.
experiences a symmetry-allowed intersection with the closed-
shell ion-pair state. The cyclopropyl cation thus generated
undergoes a barrierless ring opening to generate allyl cation.
In a second surface crossing, allyl radical and an iodine atom
are generated by a back electron transfer, occurring by way of
a same-symmetry conical intersection. It is likely that such
surface crossings are prevalent in alkyl halide photodissociations
but can only be observed when the products from radical and
ionic dissociation pathways differ. Continued studies will
address the generality of this phenomenon.
Acknowledgment. This work was supported in part by the
Department of Energy, Office of Basic Energy Sciences, under
grants DE-FG01-88ER13934 and DE-FG02-98ER14857, and
in part by the National Science Foundation, Division of
Chemistry, under grant CHE-9901065.
References and Notes
(
1) Kropp, P. J. Acc. Chem. Res. 1984, 17, 131.
(2) Szeimies, G.; Boche, G. Angew. Chem., Int. Ed. Engl. 1971, 10,
9
3
12.
(
3) Dewar, M. J. S.; Kirschner, S. J. Am. Chem. Soc. 1971, 93, 4290.
4) Farnell, L.; Richards, W. G. J. Chem. Soc., Chem. Commun. 1973,
(
34.
(
5) Merlet, P.; Peyerimhoff, S. D., Buenker, R. J.; Shih, S. J. Am. Chem.
Soc. 1974, 96, 959.
(
(
6) Dewar, M. J. S.; Kirschner, S. J. Am. Chem. Soc. 1974, 96, 5244.
7) Beran, S.; Zahradnik, R. Collect. Czech. Chem. Commun. 1976,
4
2
160.
(9) Greig, G.; Thynne, J. C. J. Trans. Faraday Soc. 1966, 62, 3338.
(
(
10) Greig, G.; Thynne, J. C. J. Trans. Faraday Soc. 1967, 63, 1369.
11) Kerr, J. A.; Smith, A.; Trotman-Dickenson, A. F. J. Chem. Soc. A
1
1
969, 1400.
Figure 11. Energy and principal configurations of the two states at
the second crossing (structure II in Figure 9).
(
(
12) Arnold, P. A.; Carpenter, B. K. Chem. Phys. Lett. 2000, 328, 90.
13) DeFrees, D. J.; McIver, R. T., Jr.; Warren, J. H. J. Am. Chem. Soc.
980, 102, 3334.
14) Berkowitz, J.; Ellison, G. B.; Gutman, D. J. Phys. Chem. 1994,
98, 2744.
the chemistry, and so a small rise in potential energy between
the first crossing point and the second would presumably pose
little problem for the reaction.
(
(
(
(
15) Dupuis, M.; Pacansky, J. J. Chem. Phys. 1982, 76, 2511.
16) Hudgens, J. W.; Dulcey, C. S. J. Phys. Chem. 1985, 89, 1505.
17) NIST Webbook: http://webbook.nist.gov.
Both of the crossing points represent potential sites of
bifurcation in the reaction dynamics. Some trajectories may
(18) Wilson, R. J.; Mueller, J. A.; Houston, P. L. J. Phys. Chem. A
1997, 101, 7593.
19) Chang, B.-Y.; Hoetzlein, R. C.; Mueller, J. A.; Geiser, J. D.;
Houston, P. L. ReV. Sci. Instrum. 1998, 69, 1665.
20) Eppink, A. T. J. B.; Parker, D. H. ReV. Sci. Instrum. 1997, 68,
3477.
1
choose to stay on the (n,σ*) surface at the first intersection.
(
They would then form vibrationally excited cyclopropyl radical
and an iodine atom. In solution, collisional cooling would almost
certainly be faster than ring opening, and so this branch would
lead to cyclopropyl-derived products. In the gas phase the hot
cyclopropyl radical would open on a time scale of nanoseconds
to the allyl radical.
(
(21) Jung, Y.-J.; Kim, Y. S.; Kang, W. K.; Jung, K.-H. J. Chem. Phys.
1997, 107, 7187.
(
22) Wang, D.; Li, Y.; Li, S.; Zhao, H. Chem. Phys. Lett. 1994, 222,
167.
(23) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb,
For reactions in the gas phase or in inert solvents, trajectories
choosing to stay on the closed-shell surface at the second
crossing point could only reform the C-C and C-I bonds to
return to cyclopropyl iodide. In a nucleophilic medium, inter-
ception of the cation by the solvent may be feasible. Such
M. A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A., Jr.;
Stratmann, R. E.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A.
D.; Kudin, K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone V.; Cossi,
M.; Cammi, R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.;
Ochterski, J.; Petersson, G. A.; Ayala, P. Y.; Cui Q.; Morokuma, K.; Malick,
D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.;
Ortiz J. V.; Baboul, A. G.; Stefanov, B. B.; Liu G.; Liashenko, A.; Piskorz,
P.; Komaromi, I.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-
Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Gonzalez, C.; Challacombe,
M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Andres, J. L.;
Gonzalez C.; Head-Gordon, M.; Replogle, E. S.; Pople, J. A. Gaussian 98
ReV. A.9; Gaussian, Inc., Pittsburgh, PA, 1998.
1
reactions are known in the photolysis of other alkyl halides.
5. Conclusions
Experimental and computational evidence has been found for
the existence of intersections between open-shell and closed-
shell potential energy surfaces in the photodissociation of
cyclopropyl iodide. In gas and solution phases, photoexcited
cyclopropyl iodide appears to dissociate directly to form an
iodine atom and an allyl radical. Ab initio calculations suggest
that this occurs by way of two crossings between open-shell
and closed-shell surfaces. The experiments show that the initially
accessed electronic state of cyclopropyl iodide is of A′ sym-
metry. If the CASSCF calculations are correct, this state must
(24) GAMESS (March 25, 2000 version for the Power Macintosh):
Schmidt, M. W.; Baldridge, K. K.; Boatz, J. A.; Elbert, S. T.; Gordon, M.
S.; Jensen, J. H.; Koseki, S.; Matsunaga, N.; Nguyen, K. A.; Su, S. J.;
Windus, T. L. J. Comput. Chem. 1993, 14, 1347.
(25) Ragazos, I. N.; Robb, M. A.; Bernardi, F.; Olivucci, M. Chem.
Phys. Lett. 1992, 197, 217.
(26) MacMolPlt 5.2.3: Bode, B. M.; Gordon, M. S. J. Mol. Graphics
Modell. 1998, 16, 133.
(27) Castleman, K. R. Digital Image Processing; Prentice Hall, New
York, 1996.
(28) Heck, A. J. R.; Chandler, D. W. Annu. ReV. Phys. Chem. 1995,
1
undergo internal conversion to the A′′ (n,σ*) state, which then
45, 335.

Photodissociation of Cyclopropyl Iodide
J. Phys. Chem. A, Vol. 105, No. 10, 2001 1701
(
(
(
29) Egger, K. W.; Cocks, A. T. HelV. Chim. Acta 1973, 56, 1516.
30) Ferguson, K. C.; Whittle, E. Trans. Faraday Soc. 1971, 67, 2618.
31) DePuy, C. H.; Gronert, S.; Barlow, S. E.; Bierbaum, V. M.;
from the boiling point. This approximation gives ∆Hvap ) 7.2 kcal/mol
and ∆Hf° ) 13.8 kcal/mol for cyclopropyl bromide in the gas phase.) and
cyclopropyl radical (see refs 31 and 32).
(36) 56.9 kcal/mol based on the heats of formation of methyl iodide
(Cox, J. D.; Pilcher, G. Thermochemistry of Organic and Organometallic
Compounds; Academic Press: New York, 1970) and methyl radical (Chase,
M. W., Jr. NIST-JANAF Themochemical Tables, 4th ed.; J. Phys. Chem.
Ref. Data, Monograph 9, 1998).
Damrauer, R. J. Am. Chem. Soc. 1989, 111, 1968.
32) Seburg, R. A.; Squires, R. R. Int. J. Mass Spectrom. Ion Proc. 1997,
67, 541.
33) 71.6 kcal/mol based on the heats of formation of isopropyl bromide
Davies, J.; Lacher, J. R.; Park, J. D. Trans. Faraday Soc. 1965, 61, 2413)
(
1
(
(
and isopropyl radical (Tsang, W. In Energetics of Organic Free Radicals;
Martinho Simoes, J. A., Greenberg, A., Liebman, J. F., Eds.; Blackie
Academic and Professional: London, 1996; p 22.)
(37) Minsek, D. W.; Blush, J. A.; Chen, P. J. Phys. Chem. 1992, 96,
2025.
(38) Houle, F. A.; Beauchamp, J. L. J. Am. Chem. Soc. 1978, 100, 3290.
(39) RRKMWSU by L. Zhu and W. L. Hase: Quantum Chemistry
Program Exchange (Indiana University) #644.
(40) Newcomb, M. Tetrahedron 1993, 49, 1151 and references therein.
(41) Foresman, J. P.; Head-Gordon, M.; Pople, J. A.; Frisch, M. J. J.
Phys. Chem. 1992, 96, 135.
(
34) 54.9 kcal/mol based on the heats of formation of isopropyl iodide
Furuyama, S.; Golden, D. M.; Benson, S. W. J. Chem. Thermodyn. 1969,
, 363) and isopropyl radical (see ref 33).
35) 80.7 kcal/mol based on the heats of formation cyclopropyl bromide
Holm, T. J. Chem. Soc., Perkin Trans. 2 1981, 464. This paper reports
Hf° for liquid cyclopropyl bromide. The enthalpy of vaporization is
apparently unknown, and so we have used Trouton’s rule to estimate it
(
1
(
(
∆
(42) Walsh, R. Int. J. Chem. Kinet. 1970, 2, 71.
(43) Yarkony, D. R. Acc. Chem. Res. 1998, 31, 511.