JOURNAL OF CHEMICAL PHYSICS
VOLUME 115, NUMBER 7
15 AUGUST 2001
Charge-transfer mediated photochemistry in alkene–O2 complexes
Gary DeBoer,a) Amy Preszler Prince, and Mark A. Youngb)
Department of Chemistry and the Optical Science and Technology Center, University of Iowa, Iowa City,
Iowa 52242
共Received 29 March 2001; accepted 30 May 2001兲
The photochemistry of a series of alkene–O2 complexes was studied in a supersonic expansion
using a resonance enhanced multiphoton ionization probe of the O( 3 P j ) photoproduct at 226 nm.
The relative yield of oxygen atoms from each complex was correlated to the ionization potential of
the alkene species and indicates that initial excitation of an intermolecular charge-transfer state
mediates the subsequent excited state chemistry. The behavior is similar to that observed previously
for the C6H6 –I2 system: a reverse electron-transfer step yields electronically excited O2 which
subsequently dissociates. The kinetic energy release of the O( 3 P j ) fragment was also measured
using a time-of-flight analysis and found to be small with an isotropic spatial distribution. No
evidence for photo-oxidation of the alkenes was observed in the mass spectra. A comparison is made
to the charge-transfer absorption spectra observed in cryogenic oxygen matrices of similar alkene
complexes. Ab initio models were used to identify the stable ground state geometry of the C2H4 –O2
complex and complete active space self-consistent-field calculations were performed to identify the
energy of the charge-transfer state for several alkene–O2 complexes. © 2001 American Institute of
Physics. 关DOI: 10.1063/1.1386784兴
␲ – ␲ system, isolated in a supersonic expansion.11 Excitation
of these complexes at 226 nm produced O-atom fragments;
without benzene in the expansion, similar excitation of isolated O2 yielded no detectable photodissociation, indicating
that dissociation in the complex is ⭓103 times more likely.
The O-atom photofragments exhibited a small and isotropic
kinetic energy release 共KER兲, as determined by a time-offlight analysis. We proposed that the observation of facile
oxygen bond cleavage was the result of the same reverse
electron transfer step identified in the C6H6 –I2 system, leaving the O2 on the repulsive part of an excited state potential.
However, identification of the active pathway was rendered
ambiguous by the possible influence of symmetry breaking
effects in the complexed oxygen molecule. The ultraviolet
共UV兲 Herzberg continuum of O2 overlaps the excitation
wavelength used in our studies and the continuum states
have been shown to be sensitive to the local environment.
The A ⬘ 3 ⌬ u and A 3 ⌺ ⫹
u states, while normally forbidden by
orbital selection rules, can gain intensity due to symmetry
relaxation in a weakly bound complex12,13 or in the solid
state.14 However, we have noted11 that in studies of donor
species, such as arenes, seeded in a cryogenic oxygen matrix,
the nominally forbidden O2 transitions appear to be much
weaker than the associated CT absorption.15
To provide a more definitive test of the active chromophore, and to further ascertain the generality of the CT
dynamics identified in the C6H6 –I2 system, we have completed a thorough study of a series of alkene–O2 ␲ – ␲ complexes. By changing the identity of the alkene, C2R4 共R⫽H,
CH3, or C2H3兲, we can systematically vary the ionization
potential 共IP兲 of the donor while the acceptor electron affinity 共EA兲 remains unchanged. While the weak ground state
intermolecular interactions will be similar across the series,
the energy of the CT state, E CT, can be expected to shift
according to the expression
I. INTRODUCTION
We have been interested in the identification of new reaction pathways and enhanced photochemical yields in
weakly bound clusters that manifest an intermolecular
charge-transfer 共CT兲 excited state. Such species should be
good candidates for interesting dynamical studies of cluster
chemistry due to the supramolecular nature of the CT transition, whereby all of the atoms in the complex must be considered. In previous liquid phase1– 4 and molecular beam
studies5– 8 of CT in aromatic-I2 species, a reverse electron
transfer mechanism was found to lead to rapid and enhanced
dissociation of the halogen molecular bond. We estimated an
enhancement of three orders of magnitude for the dissociation of I2 in the CT complex relative to photodissociation of
the isolated molecule.5 The dissociative quenching channel
has been suggested as a possible route for chemical energy
storage in solar energy conversion schemes.9 These species
are weakly interacting in the ground state and were termed
␲ – ␴ complexes by Mulliken10 since the donor molecular
orbital 共MO兲 is an aromatic ␲ orbital and the acceptor orbital
is the halogen ␴*.
These recent investigations have done much to elucidate
the detailed photochemical dynamics of CT excitation in the
complexes studied but it is not yet clear if the observed behavior is general for CT interactions, of which there are
many variations. We have endeavored to extend our studies
to encompass other complexes, particularly those involving
molecular oxygen as the potential electron acceptor species.
Initially, we studied the C6H6 –O2 complex, an example of a
a兲
Current address: Department of Chemistry and Physics, LeTourneau University, 2100 S. Mobberly, Longview, TX 75602.
b兲
Author to whom correspondence should be addressed. Electronic mail:
[email protected]
0021-9606/2001/115(7)/3112/9/$18.00
3112
© 2001 American Institute of Physics
Downloaded 15 Aug 2001 to 128.255.70.158. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp
Alkene–O2 complexes
J. Chem. Phys., Vol. 115, No. 7, 15 August 2001
e2
E CT⫽IP⫺EA⫺
,
R DA
共1兲
where the last term is the Coulombic attraction of the ion
pair in the CT state, separated by a distance R DA . Equation
共1兲 ignores the binding energy of the ground state complex,
which is generally negligible in comparison to the much
stronger CT state interaction. By noting the observed trends
in the photochemical behavior as the donor species is altered,
we can derive information regarding the influence of CT interactions on the initially excited electronic state.
There are other intriguing motivations for studying oxygen complexes, particularly those involving hydrocarbon donors. Hydrocarbon–O2 CT complexes formed in a zeolite
cavity and stabilized by the high electric fields present inside
the zeolite, have been implicated in the selective and highly
redshifted photo-oxidation observed in these systems.16 –20 A
mechanism has been proposed in which proton transfer in the
CT state from the hydrocarbon forms the HO2 radical, which
then reacts in a selective fashion.17 The details of the reaction
can be probed without the complications of cage effects and
secondary chemistry by studying similar complexes isolated
in a supersonic expansion. Oxidation of nonmethane hydrocarbons through a complicated photochemical pathway is a
key component of smog formation, leading to elevated ozone
levels and the production of potentially harmful airborne
particulates.21–25 CT-mediated photochemistry in hydrocarbon–O2 species could have an impact on the photooxidation cycle, particularly in polluted urban areas with excessively high concentrations of anthropogenic hydrocarbons, either through direct photo-oxidation in the complex or
through the production of reactive oxidative radicals.
A series of alkene–O2 complexes were generated in a
supersonic expansion and excited at a fixed wavelength with
a pulsed laser source. The relative amount of O( 3 P j ) photoproduct was determined using a multiphoton ionization
scheme within the same laser pulse and then correlated with
the IP of the alkene donor. The KER of the O-atom photofragment was also measured. The relative propensities for
O-atom formation suggest that initial excitation of a CT state
in the complex is responsible for the subsequent photochemistry, as anticipated by Eq. 共1兲. Our results are compared to
the CT absorption bands recorded for similar complexes in
cryogenic oxygen matrices. Ab initio calculations were used
to determined the ground state binding energy and structure
of the prototypical C6H6 –O2 complex. In addition, the results
of complete active space self-consistent-field 共CASSCF兲 calculations involving the ␲ orbitals for several alkene–O2
complexes are compared to the experimental measurements.
II. EXPERIMENT
The experimental apparatus consists of a supersonic expansion source coupled to a reflectron time-of-flight mass
spectrometer 共TOFMS兲. The instrument can be operated either in reflectron mode, for the collection of mass and wavelength resolved spectra, or in a pulsed extraction mode,
which allows the KER of various photoions to be deconvoluted from the characteristic flight-time spectra. The instrument has been described in greater detail in previous
3113
communications.5,26,27 The supersonic source used a commercial pulsed solenoid valve to expand premixed samples
of oxygen and alkene diluted in bulk He carrier gas. Gaseous
hydrocarbons were combined with O2 and He in measured
amounts in a stainless steel tank and allowed to mix for 24
hours prior to use. Liquid hydrocarbons were placed in a
temperature controlled bubbler through which a premixed
sample of O2 /He was allowed to flow. Typical operating conditions were adjusted so that the O2 and alkene concentrations were approximately 2% in He and the total stagnation
pressure was nominally 2300 Torr 共30 psig兲.
An excimer pumped dye laser was employed to generate
UV radiation at approximately 226 nm, separated from the
fundamental by a Pellin–Brocca prism assembly. The UV
radiation was used in a one-color experiment to both excite
the alkene–O2 complexes and to detect ground state O( 3 P j )
photoproducts via a 2⫹1 resonance enhanced multiphoton
ionization 共REMPI兲 process. The same laser wavelength was
also used to probe for other species in the mass spectrum by
nonresonant multiphoton ionization.
In the pulsed extraction mode, the photoions are generated under field free conditions and allowed to drift for a
precisely controlled delay time according to the KER imparted by the dissociation process. The ions are then extracted by a high voltage pulse and the resultant flight-time
distribution can be used to determine the photofragment
translational energy spectrum.28 –30 Information regarding the
spatial distribution of photofragments can be obtained by
rotating the laser polarization with respect to the collection
axis of the instrument. The flight-time spectrum, f ␣ (TOF), is
transformed to a laboratory frame velocity distribution,
I ␣ ( ␯ z ), according to
d 共 TOF兲
,
d共 ␯z兲
I ␣ 共 ␯ z 兲 ⫽ f ␣ (TOF兲
共2兲
where ␯ z is the component of fragment velocity along the
collection axis of the TOFMS and ␣ is the angle the laser
polarization makes with the instrument collection axis.
Our main interest was in quantifying the relative O⫹
signal as a function of the alkene donor species. To ensure
repeatable quantitative results in the face of daily variations
in instrumental sensitivity, an internal calibration technique
was employed. A sample of hexafluorobenzene, C6F6, was
subjected to a series of freeze–pump–thaw cycles and placed
in a Pyrex reservoir held at 0 °C. Vapor from the sample was
flowed through a controlling needle valve into the ionization
chamber of our TOFMS to a constant pressure of ⬇1.0
⫻10⫺7 Torr, as measured by an ionization gauge. The CF⫹
mass signal, produced by fragmentation and multiphoton
ionization of the parent, was used as a reference to normalize
the O⫹ signals produced from the alkene–O2 clusters. The
instrumental sensitivity was calibrated in this fashion on a
daily basis. A great deal of effort was devoted to maintaining
similar laser intensities during the calibration procedure and
the measurements of the O⫹ signal due to the multiphoton
nature of the excitation/ionization processes in both cases. To
ascertain the degree of reproducibility in the data, the O⫹
signal from the 1-butene alkene system was repeatedly moni-
Downloaded 15 Aug 2001 to 128.255.70.158. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp
3114
J. Chem. Phys., Vol. 115, No. 7, 15 August 2001
De Boer, Preszler Prince, and Young
TABLE I. The relative O⫹ signal from alkene–O2 complexes and the relative overlap of the 226 nm excitation
wavelength with CT absorptions measured in cryogenic matrices and calculated by the CASSCF method. All
data has been normalized to the value for propene.
tored over a period of several days and the error was found to
be within 6%. Similar errors were assumed for the other
alkenes studied.
The computational portion of the investigation was conducted using the GAUSSIAN 98 software package,31 running
on either a supercomputer 共SGI Origin2000兲 at the National
Computational Science Alliance or workstations 共4-CPU
IBM RS/6000兲 located in the Department of Chemistry at the
University of Iowa.
relative O⫹ signals we have determined for each alkene, normalized to the propene value, are listed in Table I.
The KER for the O( 3 P 2 ) fragment, which had the largest
population in our C6H6 –O2 experiments,11 was measured by
III. RESULTS
Table I contains a list of all of the alkene species studied
in the current investigation, along with their respective ionization potentials. For reference, the electron affinity of O2
also appears in Table I. Initially, a mass resolved spectrum of
a 2% O2 /He mixture was collected with the laser tuned to
225.59 nm, corresponding to resonant excitation of the
O( 3 P 2 ) fragment. Under our experimental conditions, no O⫹
共m/e⫽16兲 signal could be observed in the mass spectrum,
indicating that direct excitation of the O2 Herzberg continuum results in an undetectable level of photodissociation
in bare molecular oxygen. In contrast, all of the alkene/O2
expansions, with the exception of 2,3-dimethyl-2-butene,
yielded mass spectra with an easily measurable O⫹ signal of
varying intensity.
Representative mass spectra for the alkene mixtures
studied are collected in Fig. 1. The data has been normalized
by the CF⫹ signal according to the procedure outlined in the
experimental section 共vide supra兲. The baseline of the 2,3dimethyl-2-butene spectrum exhibits ringing due to a strong
signal at m/e⫽15, corresponding to the fragment, CH⫹
3 . We
have assumed that no O⫹ is produced for this alkene. The
FIG. 1. Mass spectra for the various alkene/O2/He mixtures showing the
region of the O⫹ signal with the excitation laser tuned to a 2⫹1 REMPI
transition in ground state O( 3 P 2 ). The alkenes are 共A兲 2,3-dimethyl-2butene, 共B兲 2-methyl-2-butene, 共C兲 cis-2-butene, 共D兲 trans-2-butene, 共E兲
2-methylpropene, 共F兲 1-butene, 共G兲 propene, and 共H兲 ethene. The baseline
noise in the 2,3-dimethyl-2-butene spectrum, 共A兲, is due to ringing from a
large fragment signal at m/e⫽15, CH⫹
3.
Downloaded 15 Aug 2001 to 128.255.70.158. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp
J. Chem. Phys., Vol. 115, No. 7, 15 August 2001
FIG. 2. The O( 3 P 2 ) photofragment velocity spectra obtained with the
propene/O2/He expansion. The laser is oriented parallel to the instrument
collection axis, ␣⫽0°, in spectrum 共A兲, and perpendicular, ␣⫽90°, in spectrum 共B兲. The points are the experimental data and the solid line is a fit to
the data assuming a Maxwell–Boltzmann function for the speed distribution.
tuning the dye laser to the appropriate wavelength and operating the TOFMS in a pulsed extraction mode. In all systems
where an O⫹ signal could be detected, the measured photofragment spectra were similar, exhibiting small and isotropic
velocity distributions. Representative photofragment velocity
spectra for the propene sample are shown in Fig. 2, with the
laser polarization oriented either parallel or perpendicular to
the collection axis of the instrument.
The appearance of the mass and energy resolved spectra
did not change noticeably when using expansions with richer
alkene concentrations. The dependence of the O⫹ signal on
the total stagnation pressure for a C2H4 /O2 /He mixture was
determined and the results shown in Fig. 3. The data can be
described by a power law of the form, 关 O⫹兴 ⬀ P n , where P is
the stagnation pressure and the best fit is with, n⫽2.4. In
contrast, a corresponding plot of the pressure in the source
region of the chamber measured by an ionization gauge, indicative of the total monomer flow from the pulsed valve
source, follows a power law with n⫽0.8. Finally, we carefully examined the mass spectra obtained from all of the
various alkene–O2 expansions for any evidence of photooxidation products. However, even using higher stagnation
pressures of up to 5400 Torr 共90 psig兲, no such fragment
masses could be observed, nor could signal corresponding to
the mass of the parent alkene-O⫹
2 be detected in our experiment.
The C2H4 –O2 complex was chosen to serve as the archetype for the theoretical studies of the ground state structure
Alkene–O2 complexes
3115
and bonding in alkene–O2 systems. The four different structures depicted in Fig. 4, all with C 2V symmetry, were utilized
as starting points for unrestricted second-order MøllerPlesset perturbation theory 共UMP2兲 optimizations with the
6-311⫹⫹G共2d,2p兲 basis set.32 Using stringent convergence
criteria 共the ‘‘Opt⫽VeryTight’’ option in GAUSSIAN 98兲, optimized structures were located for all four initial geometries.
However, only the geometry of Fig. 4共A兲, with the O2 bond
bisecting the ethene plane, was found to have a positive definite Hessian, indicating that the geometry was a true minimum on the ground state surface. Table II summarizes the
calculations and lists some of the structural parameters.
FIG. 4. Representations of the four structures for the C2H4 –O2 geometry
optimization calculations. Structure 共A兲, with the O2 bond bisecting the
ethene plane, was found to be the only stable geometry.
Downloaded 15 Aug 2001 to 128.255.70.158. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp
3116
J. Chem. Phys., Vol. 115, No. 7, 15 August 2001
De Boer, Preszler Prince, and Young
TABLE II. Summary of results for optimizations of the C2H4 –O2 complex
structures in Fig. 4 at the UMP2/6-311⫹⫹G共2d,2p兲 level of theory. The
change in length of the O2 bond, ⌬R OO , is relative to the monomer value,
1.231 Å, calculated at the same level of theory. The value for the donor–
acceptor distance, R DA , is measured from the midpoint of the ethene C–C
bond to the midpoint of the O–O bond.
Structure
A
B
C
D
Binding energy 共kcal/mole兲
0.92
0.73
0.72
0.54
Hessian
pos.definite imaginary imaginary imaginary
⌬R OO 共Å兲
⫹0.001
0.000
0.000
0.000
3.538
3.635
3.603
4.160
R DA 共Å兲
The change in the O2 bond length in the complex, ⌬R OO , is
negligible and there is little deviation from the value in isolated oxygen, R OO⫽1.231 Å, as calculated at the same level
of theory. The distance between the ethene donor and the
oxygen acceptor, R DA , was measured from the midpoint of
the C–C bond to the midpoint of the O–O bond.
The binding energy was calculated as the difference in
energy between the complex and the monomer units, separately optimized at the same level of theory. We did not
attempt to correct these energies for basis set superposition
effects. The results, contained in Table II, show that the bisected structure of Fig. 4共A兲 is also the most strongly bound,
0.92 kcal/mole, but, overall, the binding energies are less
than 1 kcal/mole and the alkene–O2 complexes are clearly
very weakly interacting species. For comparison, the binding
energy of the C6H6 –O2 complex is estimated to be about 1.6
kcal/mole33 and the dimer (O2) 2 is bound by approximately
0.2–0.5 kcal/mole.12,34 The bisected structure was also optimized at the UMP2 level with a correlation consistent polarized double-zeta basis set, aug-cc-pDVZ, augmented with
diffuse functions.35 The binding energy increased slightly to
1.12 kcal/mole. A single point energy calculation with the
corresponding triple-zeta basis set, aug-cc-pVTZ, yielded a
binding energy of 0.97 kcal/mole.35 Thus, it appears that the
6-311⫹⫹G共2d,2p兲 basis set does an adequate job of capturing most of the dynamic electron correlation. Interestingly,
we also tried a compact triple-zeta basis set optimized to
reproduce molecular polarizabilities from Sadlej36 and, while
the conclusions regarding the stable structure of the
C2H4 –O2 complex were confirmed, the results yielded unrealistically high binding energies. For instance, the bisected
structure was predicted to be bound by almost 3 kcal/mole.
We performed CASSCF calculations, with the 6-311
⫹⫹G共2d,2p兲 basis set, on several alkene–O2 complexes to
provide theoretical predictions of the CT state energy. The
active space was restricted to the doubly occupied alkene ␲ u
orbital and the singly occupied, doubly degenerate O2␲ g orbitals, a 关4,3兴-CASSCF calculation. Since the geometry of
the monomer species changes little upon complexation, we
generated structures for complexes other than C2H4 –O2 by
separately optimizing the alkene and placing the O2 in the
bisected geometry of Fig. 4共A兲 with R DA equal to that found
for the C2H4 –O2 complex 共R DA ⫽3.538 Å兲. The calculated
energies of the CT state for the ethene, propene, 1-butene,
and cis-2-butene complexes are contained in Table III. These
TABLE III. Results of 关4,3兴-CASSCF calculations with the 6-311
⫹⫹G共2d,2p兲 basis set for several alkene–O2 complexes with the bisected
geometry of Fig. 4共A兲.
Alkene
Ethene
Propene
1-Butene
cis-2-Butene
E CT 共eV兲
5.31
4.88
4.84
4.47
alkenes were chosen since the bisected geometry did not involve any possible complications due to steric interactions
between O2 and methyl groups on the alkene.
IV. DISCUSSION
It is apparent that the O⫹ signal observed by probing
alkene/O2 /He expansions derives from photochemistry in an
alkene-O2 complex or cluster. There is no corresponding
oxygen signal when alkene is absent from the mixture, confirming that direct excitation of the O2 Herzberg continuum
does not yield any detectable dissociation under our experimental conditions and that no oxygen homoclusters, 共O2) n ,
which can yield O( 3 P j ) product photofragments,13,37,38 are
produced. The pressure dependence of the O⫹ signal shown
in Fig. 3 indicates that the photofragment signal follows a
power law dependence due to the necessary formation of
cluster species in the expansion. The environment in the
alkene–O2 complex results in a dramatic increase in the efficiency for O( 3 P j ) photoproduct formation from excitation
at the UV wavelength used in our study. Based on an estimate of the degree of clustering in the expansion and the
detection limit of our instrument, we determine that O2 is 103
times more likely to photodissociate when complexed with
an alkene relative to the bare, isolated molecule.
The expansions used in the current investigation are relatively lean in alkene and O2 and, based on past experience,
we do not expect the sample to be highly clustered. We do
not find signals in the mass spectrum that might have parentage in larger cluster species. Expansions with higher alkene
concentrations 共4 –5 %兲 did not alter the photofragment velocity spectra or the mass spectra and the relative propensity
for O( 3 P j ) formation indicated in Table I did not change.
The alkene–O2 species are very weakly bound according to
our theoretical model and complex formation and clustering
is not expected to be facile. We note that we are only able to
form appreciable concentrations of oxygen homoclusters,
共O2) n , by using richer expansions, at least 10% O2 in He,
and generally higher stagnation pressures, between 60–100
psig, to generate comparable O⫹ signals.38 The O⫹ signal
measured here exhibits a power law dependence 关 O⫹兴
⬀ P 2.4, where P corresponds to the stagnation pressure. A
similar pressure scaling law, 关 dimer concentration兴 ⬀ P n , has
been found to apply to other weakly bound species, such as
共CO2) 2 , n⫽2.6, and (C2H4) 2 , n⫽2.4– 3.4.39,40 Accordingly,
we expect that the O( 3 P j ) products we observe originate
mainly from photoexcitation of the complex or small clusters.
Downloaded 15 Aug 2001 to 128.255.70.158. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp
Alkene–O2 complexes
J. Chem. Phys., Vol. 115, No. 7, 15 August 2001
3117
number of assumptions in such a comparison. Implicit is that
the relative concentration of alkene–O2 species in the expansion is similar for all of the different alkenes and that the
binding energies are comparable. We have also assumed that
the CT transition has the same intensity for all the alkene
complexes, both in determining the relative propensity for
O( 3 P j ) formation in our beam experiments and in calculating relative overlap with the CT band from the matrix data
and the CASSCF results. Finally, since we rely on the formation of O( 3 P j ) to generate the action spectrum of Fig. 5, it
is assumed that the efficiency for oxygen photodissociation
is similar across the alkene series. Nevertheless, these assumptions on the whole are reasonable and the comparison
between our results and the matrix experiments and calculations reinforce the assertion that the observed behavior is due
to initial excitation of an alkene–O2 CT band which mediates
the subsequent photochemistry.
The mechanism for dissociative quenching of the ion
pair in the CT state to produce neutral fragments can be
thought of in terms of the reverse electron-transfer process
identified earlier for the C6H6 –I2 system. The process can be
represented as
h␯
The relative propensity for O( 3 P j ) formation clearly
varies across the alkene series. The data of Table I is rendered as a plot in Fig. 5. The observed trend is what one
would expect based on absorption at a fixed wavelength into
a CT band that shifts according to the IP of the alkene donor,
as expressed by the relation in Eq. 共1兲. The data of Fig. 5
represent an absorption spectrum where the absorption band
is scanned rather than the excitation wavelength. A comparison can be made to the experimentally measured CT absorption spectra reported by Hashimoto and Akimoto for the
same alkenes isolated in cryogenic oxygen matrices.41 To
determine the overlap between the CT bands and our excitation wavelength of 226 nm, we assumed a Gaussian absorption profile centered at the peak of the CT band reported by
Hashimoto and Akimoto for each alkene and with a full
width at half maximum 共FWHM兲 of 1.1 eV, which represents
a mean value for the reported matrix spectra. Such a procedure avoids complications due to interfering monomer absorptions in the matrix spectra. Hashimoto and Akimoto also
reported that their experimental E CT was linearly related to
alkene IP with a slope near unity. We fit the E CT from the
matrix results with Eq. 共1兲 using updated values for the alkene IPs and O2 EA.42 These results were used to calculate
E CT for ethene and 1-butene, which were not reported in the
matrix experiments. The relative overlaps at 226 nm, normalized to the value for propene, are shown in Table I and plotted in Fig. 5. In addition, we have plotted similar CT band
overlap data using the results of the CASSCF calculations in
Table III but shifted by ⫹0.83 eV to better match the experimental data.
As can be seen in Fig. 5, the matrix data and the
CASSCF calculations reasonably reproduce the experimental
propensities we have measured for alkene–O2 complexes
isolated in a supersonic expansion. There are, of course, a
alkene–O2→ alkene⫹ –O⫺
2 →alkene–O2*
→alkene⫹O共 3 P j 兲 ⫹O共 3 P j 兲 .
共3兲
2
The valence electronic configuration for O⫺
2 is (3 ␴ g )
4
3
⫻(1 ␲ u ) (1 ␲ g ) as an electron is donated from the alkene
␲ u orbital to a partially filled ␲ g orbital of O2 in the CT
process. However, a different electron is back-transferred in
the second step leaving the O2 molecules in an electronically
excited state that subsequently dissociates. Most likely, the
excited O2 state is one of the Herzberg continuum states,
corresponding to an electronic configuration of (3 ␴ g ) 2
⫻(1 ␲ u ) 3 (1 ␲ g ) 3 , since the potential curves for these states
should have good overlap with the anionic O⫺
2 curve in the
Franck–Condon excitation region.
The photofragment spectra of Fig. 2 shows relatively
little energy released into O( 3 P 2 ) translation with the peak
of the measured distribution at zero velocity in the laboratory
frame. The shape and isotropic nature of the photofragment
distribution is very different from that expected for isolated
O2 dissociation, which would yield sharp peaks and a definite anisotropy due to the correlation between the laser polarization and the recoil velocity vector. Clearly, interactions
in the complex and the supramolecular nature of the CT transition effect the KER for the O( 3 P j ) fragments with some
fraction of the excess energy partitioned into internal modes
of the alkene. The data in Fig. 2 was fitted to a speed distribution, g( ␯ ), assuming an analytical Maxwell–Boltzmann
function for the distribution. The root-mean-square 共rms兲 velocity of the fitting function was 1160⫾40 m/s, corresponding to a translational temperature of 882 K. The velocitydependent anisotropy parameter, ␤ ( v z ), was taken to be zero
and with an apparent error of ⫾0.1. The fitted g( ␯ ) can then
be transformed to a center-of-mass 共c.m.兲 frame translational
energy distribution, P(E t,c.m.), according to
Downloaded 15 Aug 2001 to 128.255.70.158. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp
3118
J. Chem. Phys., Vol. 115, No. 7, 15 August 2001
De Boer, Preszler Prince, and Young
FIG. 6. The O( 3 P 2 兲 fragment translational energy distribution 共—兲, P(E t ),
using the calculated fits to the data in Fig. 2 for the propene–O2 complex. A
center-of-mass distribution, P(E t,c.m.), assuming that dissociation takes
place in the O2 c.m. frame 共---兲, is also shown. The reuslts for the C6H6 –O2
system, using the same assumption concerning O2 dissociation 共–-–兲, is
included for comparison. The arrow represents the excess energy available
at the probe laser wavelength used in our experiments.
P 共 E t,c.m.兲 ⫽
冉
冊
g 共 ␯ 兲 m P ⫺m O
,
␯
m Pm O
共4兲
⌿ N ⫽a⌿ 0 共 D⫺A 兲 ⫹b⌿ 1 共 D ⫹ ⫺A ⫺ 兲 ,
where the c.m. translational energy is given by
冉
冊
mP
m O␯ 2
E t,c.m.⫽
,
2
m P ⫺m O
tions despite the differences in the integral degrees of
freedom across the alkene series. It appears that the dynamics of dissociative quenching are very similar for all of the
alkene–O2 complexes, lending credence to our assumption
of equal O2 dissociation efficiencies in our analysis of the
propensity data 共vide supra兲.
Our ab initio calculations of the C2H4 –O2 complex confirm that these are very weakly interacting species, with
binding energies on the order of 1 kcal/mole. Mulliken explained such absorption features as arising from ‘‘contact’’
charge transfer, due to the momentary encounter of a donor
and acceptor in a thermal environment rather than from a
stable complex.10 A previous configuration interaction calculation of the C2H4 –O2 structure in Fig. 4共B兲 with the 6-31G*
basis determined a binding energy of 1.76 kcal/mole and
R DA ⫽3.8 Å,43 very close to our results. The internal degrees
of freedom of the alkene and oxygen were not allowed to
relax in this calculation and a frequency analysis was not
performed. Applying Eq. 共1兲 to the matrix measurements of
the CT bands by Hashimoto and Akimoto, and using the IP
and EA data in Table I, a mean value of R DA ⫽4.1 Å can be
estimated for the alkene–O2 complexes. Our calculated value
of R DA ⫽3.538 Å for the ground state C2H4 –O2 structure is
quite close to this value, especially considering that Eq. 共1兲
assumes transfer of unit electronic charge.
The theoretical description formulated by Mulliken to
describe the CT phenomena10 invoked a resonance wave
function to represent the ground state of the complex
共5兲
and m O is the mass of the oxygen atom and m P is the mass
of the parent species. Since the fragmentation event is likely
to be a three-body process, as implied by Eq. 共3兲, measurement of a single fragment speed distribution cannot uniquely
determine the c.m. translational energy. However, we can
estimate an upper limit to the c.m. translational energy by
assuming that dissociation takes place in the O2 c.m. frame
and that no energy is partitioned into translation of the alkene
fragment. The calculated P(E t,c.m.) is shown in Fig. 6 along
with the O atom fragment translational energy distribution,
P(E t ). The product asymptote of Eq. 共3兲 will leave 7.82
kcal/mole of excess energy available at the 226 nm probe
wavelength, as indicated by the arrow in Fig. 6, after correcting for the calculated binding energy of the complex. The
calculated distribution extends slightly beyond the energy
conservation limit, revealing the approximate nature of the
transformation. The mean energies of the experimental distributions are Ē t,c.m.⫽5.2 kcal/mole and Ē t ⫽2.6 kcal/mole.
Assuming that the O( 3 P j ) fine structure branching accounts
for about 0.2 kcal/mole, as in the C6H6 –O2 system,11 the
internal energy of the alkene can be estimated to be 2.4 –5.0
kcal/mole. The situation for the C6H6 –O2 complex was
found to be very similar and the slightly cooler P(E t,c.m.) for
this system, using the same assumptions about O2 dissociation, is shown in Fig. 6. Interestingly, all of the alkene complexes yielded very similar photofragment velocity distribu-
共6兲
where ⌿ 0 (D⫺A) is the configuration for the normal, weakly
bound complex 共the ‘‘no-bond’’ wave function兲 and
⌿ 1 (D ⫹ ⫺A ⫺ ) is the CT configuration 共the ‘‘dative bond’’
wave function兲. The orthogonal excited CT state is then
⌿ V ⫽a * ⌿ 1 共 D ⫹ ⫺A ⫺ 兲 ⫺b * ⌿ 0 共 D⫺A 兲 .
共7兲
For weakly bound complexes, 兩 b 兩 Ⰶ 兩 a 兩 , and little charge is
transferred from the donor to the acceptor in the ground
state. The bonding is dominated by normal electrostatic, dispersive, and repulsive interactions. The transition dipole moment for the CT transition in a weak complex can be approximated as
␮ VN ⬇a * b ␮ 11⫹aa * ␮ 01 .
共8兲
In the first term, ␮ 11 is the static dipole moment of the ion
pair in the CT configuration, which can be quite large, 15–20
D, at R DA values typical of van der Waals complexes. However, the mixing coefficient, b, is small for weak complexes
and, indeed, is nonzero only if ⌿ 0 and ⌿ 1 have the same
symmetry in the point group of the supramolecular complex.
The dynamic dipole, ␮ 01 , in the second term of Eq. 共8兲 is
proportional to the overlap of the donor MO, ␾ D , and the
acceptor MO, ␾ A :
␮ 01⬀ 具 ␾ D 兩 ␾ A 典 .
共9兲
Equation 共9兲 imposes a similar symmetry requirement that
the donor and acceptor MOs must have the same irreducible
representation in the supramolecular point group. Interestingly, only the stable bisected structure of Fig. 4共A兲 fulfills
Downloaded 15 Aug 2001 to 128.255.70.158. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp
Alkene–O2 complexes
J. Chem. Phys., Vol. 115, No. 7, 15 August 2001
FIG. 7. Picture representing the nonzero overlap between the C2H4 donor
␲ u orbital and the O2 acceptor ␲ g orbital, both of which have b 1 irreducible
representation in the C 2V symmetry of the bisected complex geometry of
Fig. 4共A兲.
these requirements; the ground state is of B 2 symmetry in the
C 2V point group as is one component of the CT state and the
ethene b 1 ␲ u -donor orbital can overlap with the oxygen b 1
␲ g -acceptor orbital. Figure 7 illustrates the situation for the
donor and acceptor MOs. The transition moment for the CT
excitation will be directed along R DA . Note that the reverse
electron-transfer process of Eq. 共3兲 to yield one of the
Herzberg continuum states would involve the O2 ␲ u orbitals,
neither of which have the proper irreducible representation,
a 1 and b 2 species, to overlap with the ethene b 1 ␲ u orbital.
However, due to very weak binding in the C2H4 –O2 complex, zero point fluctuations may lead to a relatively nonrigid
structure, reducing the symmetry and facilitating CT from
any of the structures shown in Fig. 4.
We observed no evidence for photo-oxidation products
in the mass spectra for any of the alkene–O2 complexes studied, as we noted in Sec. III. In contrast, our studies of the
C6H6 –O2 system we did show evidence for the O-atom addition product, C6H6O, under conditions that favored the formation of larger clusters of the form (C6H6兲n – O2.11 We proposed that the benzene molecules formed a solvent cage
which effectively trapped recoiling O( 3 P j ) products and promoted secondary chemical reactions. It may be that even if
larger alkene–O2 clusters are formed in the expansion, they
do not yield an effective solvent cage due to geometry considerations or a smaller collision cross-section for O atoms.
The data also suggest that no concerted photo-oxidation
takes place in these complexes. However, the nonresonant
MPI method employed for detection is not ideal and other
techniques, such as single photon vacuum ultraviolet photoionization, may provide greater insight.
V. CONCLUSIONS
The photochemistry of a series of alkene–O2 complexes
isolated in a supersonic expansion was studied by laser excitation at 226 nm. The same laser pulse also photoionized
any resultant O( 3 P j ) product fragments by a 2⫹1 REMPI
process. The normalized O⫹ signal was found to vary across
the series, suggesting that initial excitation is into a CT band
of the complex that shifts according to the IP of the alkene
donor species. Comparison of the experimentally measured
O atom yields with the overlap of the 226 nm excitation
wavelength with CT bands measured for similar alkene complexes in cryogenic oxygen matrices and from CASSCF cal-
3119
culations confirm our interpretation. The photodissociation
of O2, which is enhanced by three orders of magnitude in the
complex relative to isolated oxygen, results from a reverse
electron-transfer step in the initial CT state that yields excited O*
2 which subsequently dissociates. A similar mechanism was observed in the C6H6 –I2 complex, indicating that
the dissociative quenching mechanism may constitute fairly
general behavior for CT complexes. The O atom photofragment velocity spectra reveal that interactions in the supramolecular complex result in a relatively small KER with an
isotropic spatial distribution. The velocity spectra are similar
for all of the alkene complexes which produce measurable
O( 3 P j ) product and the dissociation dynamics appear to be
similar. Ab initio models of the C2H4 –O2 complex identify
the stable structure as one with the O2 bond bisecting the
alkene plane but with a small binding energy of 0.92 kcal/
mole. The structure has the correct symmetry to facilitate
charge transfer between the donor alkene ␲ u orbital and the
acceptor oxygen ␲ g orbital. Examination of the mass spectra
showed no evidence for photo-oxygenation products, either
through a concerted reaction or via secondary chemistry induced by cage effects in larger clusters.
ACKNOWLEDGMENTS
The authors would like to gratefully acknowledge the
donors of the Petroleum Research Fund, administered by the
ACS, for support of this work and the NSF–CRIF 共CHE9974502兲 program for support of the computational facilities
used.
1
E. Lenderink, K. Duppen, and D. A. Wiersma, Chem. Phys. Lett. 211, 503
共1993兲.
2
S. Pullen, L. A. Walker II, and R. J. Sension, J. Chem. Phys. 103, 7877
共1995兲.
3
L. A. Walker, S. Pullen, B. Donovan, and R. J. Sension, Chem. Phys. Lett.
242, 177 共1995兲.
4
E. Lenderink, K. Duppen, F. P. X. Everdij, J. Mavri, R. Torre, and D. A.
Wiersma, J. Phys. Chem. 100, 7822 共1996兲.
5
G. Deboer, J. W. Burnett, A. Fujimoto, and M. A. Young, J. Phys. Chem.
100, 14882 共1996兲.
6
P. Y. Cheng, D. Zhong, and A. H. Zewail, J. Chem. Phys. 103, 5153
共1995兲.
7
P. Y. Cheng, D. Zhong, and A. H. Zewail, Chem. Phys. Lett. 242, 369
共1995兲.
8
P. Y. Cheng, D. Zhong, and A. H. Zewail, J. Chem. Phys. 105, 6216
共1996兲.
9
S. B. Karki, J. P. Dinnocenzo, S. Farid, J. L. Goodman, I. R. Gould, and T.
A. Zona, J. Am. Chem. Soc. 119, 431 共1997兲.
10
R. S. Mulliken and W. B. Person, Molecular Complexes, 1st ed. 共WileyInterscience, NewYork, 1969兲.
11
G. DeBoer and M. A. Young, J. Chem. Phys. 106, 5468 共1997兲.
12
A. Horowitz, W. Schneider, and G. K. Moortgat, J. Phys. Chem. 93, 7859
共1989兲.
13
L. Brown and V. Vaida, J. Phys. Chem. 100, 7849 共1996兲.
14
J. Goodman and L. E. Brus, J. Chem. Phys. 67, 1482 共1977兲.
15
A. J. Rest, K. Salisbury, and J. R. Sodeau, J. Chem. Soc., Faraday Trans.
2 73, 265 共1977兲.
16
F. Blatter, F. Moreau, and H. Frei, J. Phys. Chem. 98, 13403 共1994兲.
17
F. Blatter, H. Sun, and H. Frei, Catal. Lett. 35, 1 共1995兲.
18
H. Frei, F. Blatter, and H. Sun, CHEMTECH 26, 24 共1996兲.
19
S. Vasenkov and H. Frei, J. Phys. Chem. B 101, 4539 共1997兲.
20
H. Sun, F. Blatter, and H. Frei, Catal. Lett. 44, 247 共1997兲.
21
B. J. Finlayson-Pitts and J. N. Pitts, Jr., Atmospheric Chemistry: Fundamentals and Experimental Techniques 共Wiley-Interscience, New York
1986兲.
22
J. H. Seinfeld, in Progress and Problems in Atmospheric Chemistry, edited
Downloaded 15 Aug 2001 to 128.255.70.158. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp
3120
J. Chem. Phys., Vol. 115, No. 7, 15 August 2001
by J. R. Barker 共World Scientific, Singapore, 1995兲, Vol. 3, p. 34.
R. Atkinson, S. M. Aschmann, J. Arey, and B. Shorees, J. Geophys. Res.
97, 6065 共1992兲.
24
S. N. Pandis, R. A. Harley, G. R. Cass, and J. H. Seinfeld, Atmos. Environ., Part A 26A, 2269 共1992兲.
25
F. M. Bowman, C. Pilinis, and J. H. Seinfeld, Atmos. Environ. 29, 579
共1995兲.
26
M. A. Young, J. Chem. Phys. 102, 7925 共1995兲.
27
M. A. Young, J. Phys. Chem. 98, 7790 共1994兲.
28
H. J. Hwang and M. A. El-Sayed, Chem. Phys. Lett. 170, 161 共1990兲.
29
H. J. Hwang, Ph.D., UCLA, 1991.
30
H. J. Hwang and M. A. El-Sayed, J. Chem. Phys. 94, 4877 共1991兲.
31
M. J. Frisch, H. B. Schlegel, G. E. Scuseria et al., GAUSSIAN 98, Revision
A.7, Gaussian, Inc., Pittsburgh, PA, 1998.
32
J. A. Pople, J. Chem. Phys. 72, 650 共1980兲.
33
J. R. Grover, G. Hagenow, and E. A. Walters, J. Chem. Phys. 97, 628
共1992兲.
23
De Boer, Preszler Prince, and Young
C. A. Long and G. E. Ewing, J. Chem. Phys. 58, 4824 共1973兲.
R. A. Kendall, J. T. H. Dunning, and R. J. Harrison, J. Chem. Phys. 96,
6796 共1992兲.
36
A. J. Sadlej, Collect. Czech. Chem. Commun. 53, 1995 共1988兲.
37
T. G. Slanger, Science 241, 945 共1988兲.
38
G. DeBoer, Ph.D., University of Iowa, 1998.
39
E. A. Walters, J. R. Grover, J. T. Clay, P. Cid-Aguero, and M. V. Willcox,
J. Phys. Chem. 96, 7236 共1992兲.
40
J. R. Grover, W. J. Herron, M. T. Coolbaugh, W. R. Peifer, and J. F.
Garvey, J. Phys. Chem. 95, 6473 共1991兲.
41
S. Hashimoto and H. Akimoto, J. Phys. Chem. 91, 1347 共1987兲.
42
NIST Chemistry Web Book, NIST Standard Reference Database Number
69 - February 2000 Release, http://webbook.nist.gov/chemistry/
43
B. F. Minaev, V. V. Kukueva, and H. Ågren, J. Chem. Soc., Faraday Trans.
90, 1479 共1994兲.
34
35
Downloaded 15 Aug 2001 to 128.255.70.158. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/jcpo/jcpcr.jsp