Hydride abstraction by NO+ from ethanol: effects of collision energy and ion rotational state

Article abstract of DOI:10.1063/1.1286917

The collision and rotational energy effects on reaction of NO⁺ with ethanol were studied over the energy range from near thermal to several electron volts. The reactant NO⁺ was prepared using a 1+1′ photon mass-analyzed threshold ionization (MATI) scheme. Sharp collision energy dependence was observed. This effect was attributed to a tight bottleneck for H^-1 abstraction.

Full text of DOI:10.1063/1.1286917

Hydride abstraction by NO + from ethanol: Effects of collision energy and ion rotational
state
Richard J. Green, Jun Qian, Ho-Tae Kim, and Scott L. Anderson
Citation: The Journal of Chemical Physics 113, 3002 (2000); doi: 10.1063/1.1286917
View online: http://dx.doi.org/10.1063/1.1286917
View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/113/8?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
The study of state-selected ion-molecule reactions using the vacuum ultraviolet pulsed field ionization-photoion
technique
J. Chem. Phys. 125, 132306 (2006); 10.1063/1.2207609
Vibrational mode and collision energy effects on reaction of H 2 CO + with C 2 H 2 : Charge state competition
and the role of Franck-Condon factors in endoergic charge transfer
J. Chem. Phys. 123, 204313 (2005); 10.1063/1.2128670
Proton-transport catalysis, proton abstraction, and proton exchange in HF+HOC + and H 2 O+HOC + and
analogous deuterated reactions
J. Chem. Phys. 118, 6222 (2003); 10.1063/1.1559480
The effects of vibrational mode and collision energy on the reaction of formaldehyde cation with carbonyl sulfide
J. Chem. Phys. 117, 8292 (2002); 10.1063/1.1514053
Guided-ion beam study of the O 2 + +C 2 H 2 charge-transfer and chemical reaction channels
J. Chem. Phys. 110, 4291 (1999); 10.1063/1.478312
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.143.1.222 On: Fri, 12 Dec 2014 18:54:49

JOURNAL OF CHEMICAL PHYSICS
VOLUME 113, NUMBER 8
22 AUGUST 2000
Hydride abstraction by NO¿ from ethanol: Effects of collision energy
and ion rotational state
Richard J. Green, Jun Qian, Ho-Tae Kim, and Scott L. Andersona)
Chemistry Department, University of Utah, Salt Lake City, Utah 84112
͑Received 5 April 2000; accepted 22 May 2000͒
The effects of NOϩ rotational state and collision energy on the reaction NOϩϩC2H5OH
→HNOϩC2H4OHϩ were studied in a guided-ion-beam instrument over the collision energy range
from 50 meV to 3.7 eV. Integral cross sections for the reaction are presented. NOϩ is prepared in
specific rotational levels ͑Nϩϭ0,1 and Nϩϭ10͒ by means of mass-analyzed threshold ionization.
Ab initio calculations were used to probe stationary points on the potential energy surface. The
reaction is sharply inhibited by collision energy, suggesting a bottleneck for reaction. If rotational
energy had a similar effect, ϳ50% inhibition from Nϩϭ10 excitation would be observed at low
collision energy. Instead, rotation is found to have no effect within experimental error. A precursor
complex mechanism is proposed to explain the results. © 2000 American Institute of Physics.
͓S0021-9606͑00͒00232-4͔
I. INTRODUCTION
transitions in the ionization step. Because the photoelectron
can carry away orbital angular momentum, the typical result
is a narrow distribution of ion rotational states. For example,
Fujii, Ebata, and Ito4 used laser-induced fluorescence to
study the nascent distribution of COϩ produced using ͑2ϩ1͒
REMPI via the E 1⌸ intermediate state of CO. They found
that ionization through the Niϭ2 intermediate state pro-
duced only ions in the Nϩϭ2 level but that ionization via
Niϭ4 produced ions in a mixture predominantly composed
of Nϩϭ3,5, and 7. Selection rules for MPI of diatomic mol-
ecules were derived by Xie and Zare.5 Despite the fact that
REMPI only rarely produces ions in a single rotational state,
the technique is a useful, and relatively simple method to
study the effects of the ion’s rotation on reaction.
Hydride (HϪ) abstraction by NOϩ from hydrocarbons is
a common reaction used in chemical ionization mass spec-
trometry. As a consequence, the thermal rate constants for
many NOϩϩhydrocarbon reaction have been measured.1
The rate constants are quite sensitive to hydrocarbon struc-
ture, with different isomers often showing two orders of
magnitude difference. This report focuses on the collision
and rotational energy effects on reaction of NOϩ with etha-
nol, studied over the energy range from near thermal to sev-
eral electron volts.
The effects of ion rotational state on ion–molecule reac-
tions have been studied for only a few reactions. The first
such study was reported in 1968 by Chupka, Russell, and
Refaey.2 They used single-photon vacuum ultraviolet ͑VUV͒
photoionization to prepare Hϩ2 with rotational quantum num-
ber (Jϩ) varying from 0 to 2. In the reaction: H2ϩϩH2
→Hϩ3 ϩH, they found the effect of rotation to be less than
10%. Again, using VUV photoionization, Liao, Liao, and
Ng3 studied symmetric charge transfer between H2ϩ and H2 ,
using the autoionization propensity rules to control the ion
rotational state over the range from Jϩϭ0 to 2. Within their
experimental uncertainty, no effects of rotational state were
observed. In both experiments the rotational energy and an-
gular momentum were small compared to the collision en-
ergy and collisional angular momentum. The rotational auto-
ionization method is practically limited to preparation of
Hϩ2 , where autoionization transitions are easily resolved and
intense compared to the continuum.
The clearest effects of rotational state have been ob-
served by Gerlich and co-workers, capitalizing on their ca-
pability for studying reactions at subthermal collision
energies—a regime in which rotational energy is a signifi-
cant fraction of the available energy. Gerlich and Rox6 used
REMPI to produce COϩ ͑ ϭ0, Nϩϭ0–7͒ and studied the
v
effects of both collision energy and rotational state on the
association reaction: COϩ(Nϩ)ϩ2CO→͑CO͒2ϩϩCO. The
collision energy ranged down to about 0.5 meV, whereas the
rotational energy could be varied from 0 to 10 meV. Both
rotational and collision energy were found to decrease the
association rate, and the effect was nonspecific, i.e., the de-
crease depends only on total energy. Gerlich, Jerke, and
Schweizer7 compared the effects of rotational excitation
(Nϩϭ0–5, Erotϭ0Ϫϳ120 meV͒ with vibrational excitation
at a collision energy of 10 meV, on the reaction
Hϩ( ,Nϩ)ϩH →HϩϩH. Both vibration and rotation were
Resonance-enhanced multiphoton ionization ͑REMPI͒
has been used for several rotationally state-selected ion–
molecule reaction studies. Here, a single rotational state is
used as the intermediate in a REMPI scheme, and rotational
selection in the ion relies on propensity rules for rotational
v
2
2
3
found to suppress reactivity, but rotation had roughly twice
the effect for a given amount of energy. Glenewinkel-Meyer
and Gerlich8 recently reported another study of the H2ϩϩH2
→Hϩ3 ϩH reaction, in which both REMPI and pulsed-field
ionization were used to prepare HϪ2 ͑ ϭ0–1, Nϩϭ0–4͒.
v
a͒
Electronic mail: anderson@chemistry.utah.edu
They reported that rotational energy hinders reaction, and
0021-9606/2000/113(8)/3002/9/$17.00
3002
© 2000 American Institute of Physics
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.143.1.222 On: Fri, 12 Dec 2014 18:54:49

J. Chem. Phys., Vol. 113, No. 8, 22 August 2000
Hydride abstraction by NOϩ
3003
that the effect is less than 10% for rotational excitation up to
Nϩϭ5. Using a potential surface calculated by Eaker and
Schatz9 they suggested that the hindrance was caused by one
of four possible mechanisms: increased centrifugal barrier in
the entrance channel, a rotational effect on the decay of the
collision complex back to reactants, steric hindrance, or a
transition to a nonreactive surface.
Mass-analyzed threshold ionization ͑MATI͒ is a variant
on pulsed-field ionization, developed by Zhu and Johnson10
and used by many groups for ion spectroscopy studies.11–16
MATI allows formation of mass- and state-selected ions,
with rotational selection possible in many cases. MATI spec-
tra of the NO molecule have been reported by several groups
including Mackenzie et al.,17 using the B 2⌸ intermediate
state, and Sato and Kimura18 using the A 2⌺ intermediate
state. The first use of MATI for an ion–molecule reaction
study was reported by Mackenzie et al. They studied the
reaction H2ϩ (Nϩϭ0–3)ϩH2→H3ϩϩH,11,19,20 placing an
upper limit on any rotational effect to be less than 10% at
collision energies between 10 and 250 meV. This observa-
tion is consistent with the experiments by Chupka, Russell,
and Refaey,2 and by Gerlich and co-workers.
To date, the range of systems for which ion rotational
effects have been studied is narrow, comprising only sym-
metric diatomic association and proton transfer reactions.
The NOϩϩC2H5OH system reported on here differs in some
significant ways. For symmetric systems, the degeneracy of
the MϩϩM and MϩMϩ reactant charge states guarantees
that nonadiabatic behavior will be possible as the intermo-
lecular interaction mixes the asymptotic states. The ioniza-
tion energy of NO is more than 1 eV lower than that of
ethanol, thus it is less likely that nonadiabatic effects are
important. The neutral reactant has many more degrees of
freedom, allowing reactions mediated by longer-lived colli-
sion complexes. In addition, because the neutral reactant is a
physically larger molecule, the moments of inertia in any
complexes that form are substantially larger than those in
diatomic–diatomic systems. This changes the relation be-
tween angular momentum and rotational energy as com-
plexes form and rearrange. Finally, both neutral and ion in
our system have significant dipole moments, raising the pos-
sibility of stronger anisotropic effects.
action; however, it does so by a factor of four times less than
predicted by theory.
Viggiano and co-workers have reported several studies
in their selected-ion flow drift tube ͑SIFDT͒ in which the
effects of the neutral reactant’s rotation were studied by in-
dependently varying the temperature and the average kinetic
energy ͑i.e., drift velocity͒ of the reactants.24,25 For example,
in the reaction of Oϩ with HD,25 it was found that the overall
rate constant was independent of temperature or average ki-
netic energy in the range studied, but that the branching ratio
into ODϩ and OHϩ at a given average kinetic energy varied
as a function of temperature. This effect was attributed to the
changes in HD rotational temperature. The theoretical work
of Dateo and Clary on the effects of rotation on long-range
capture26 accounted for the observations. The physical pic-
ture is that the HD has different centers of mass and polar-
izability, causing the H end to be oriented toward the Oϩ.
Rotation decreases the orientation effect. Another example
was found in a study of Krϩ(2P3/2) with HD, in which it was
found that the KrHϪ/KrDϩ branching ratio was affected by
rotation in the opposite sense from the effect of kinetic
energy.27 Viggiano and co-workers have recently reviewed
studies of a variety of reactions using this method.28
The reaction studied here is hydride abstraction by NOϩ
from ethanol:
NOϩ Nϩ͒ϩC H OH→HNOϩC H Oϩ.
͑1͒
͑
2
5
2
5
The collision energy was varied from 50 meV to 3.7 eV, and
the reactivity of two NOϩ rotational states was examined.
The first state is an unresolved mixture of Nϩϭ0 and 1, and
the second state is Nϩϭ10. The rotational and centrifugal
distortion constants of NOϩ are 1.997 cmϪ1 and 5.64
ϫ10Ϫ6 cmϪ1, respectively,29 thus the energies of the reac-
tant states are 0, 3.98, and 218.6 cmϪ1, or 0, ϳ0.5, and ϳ27
meV. At our lowest collision energy, the Nϩϭ10 rotational
energy is about 50% of the collision energy. The thermal
energy in the EtOH reactant is 76 meV, approximately
equally divided between rotation and vibration. In terms of
total energy, exciting Nϩϭ10 increases the total energy at
our lowest experimental point by 21%.
Williamson and Beauchamp30 have studied reaction ͑1͒
in an ion-cyclotron-resonance spectrometer ͑ICR͒, and
Spanel and Smith31 studied the reaction in a ͑SIFDT͒. The
results of these studies are discussed in the following.
In addition to the previous studies of ion rotational state
effects, several groups have studied the effects of the rota-
tional energy of the neutral reactant in ion–molecule reac-
tions. In general, these studies have either used para and
normal H2 as the neutral, or have varied the temperature of
the neutral reactant. In the first category, Marquette, Rebrion,
II. EXPERIMENTAL METHODOLOGY
The guided-ion-beam instrument used in these studies
has been described previously, along with calibration and
data analysis procedures.32–35 For this study, reactant NOϩ is
prepared using a 1ϩ1Ј photon MATI scheme. The first pho-
ton excites the neutral molecule to the A 2⌺ state of NO via
a particular rotational transition in the 0–0 band. The second
photon pumps the A-state molecules into an ensemble of
high-lying Rydberg states that converge upon the ion rota-
tional state of interest. The Rydberg molecules are then field
ionized to generate NOϩ in the desired rotational state. To
produce low rotational state ions, we excite the R21(0.5)
transition to the A state, and to produce high Nϩ we excite
the R21(8.5) line. Figure 1 shows MATI spectra from both
21
and Rowe studied the reaction of Nϩ with H2 concluding
that neutral rotation is important in driving this endoergic
reaction. The Gerlich group has studied a number of MϩϩH2
reactions using this approach.22 Their study of NϩϩH2 com-
pared the effects of rotation and collision energy. The results
are consistent with those of Marquette, Rebrion, and Rowe
and show that rotation is at least as effective as collision
energy in overcoming the activation barrier. In reaction of
CHϩ3 and CDϩ3 with hydrogen and helium, Gerlich and Ka-
effer have studied the effect of neutral rotation upon the ter-
nary rate coefficient.23 They found that rotation inhibits re-
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.143.1.222 On: Fri, 12 Dec 2014 18:54:49

3004
J. Chem. Phys., Vol. 113, No. 8, 22 August 2000
Green et al.
electric field prior to field ionization. Neutral NO is injected
into the source as a pulsed, seeded beam. Multiphoton exci-
tation is carried out between a pair of plates separated by
12.7 mm, with apertures to allow beam transmission, and
with a series of intermediate electrodes that allow us to cre-
ate a uniform field across the interplate gap. In the first 100
ns after the laser pulse, this region is field free, and then a
field of approximately 1.5 V/cm is applied, oriented such that
it retards the prompt ions produced in the laser pulse. As
reported by Sato and Kimura18 we find that the MATI signal
is enhanced by the presence of this separation field and the
field strength is adjusted to maximize signal. Vrakking and
Lee36 reported enhancement in the lifetime of high-lying Ry-
dberg states in the presence of an electric field, which they
attribute to l mixing induced by the field. Merkt and Zare
have published a model for such a mechanism.37
The NO neutrals, including the high-lying Rydbergs, are
not retarded, and pass into the second region of the source
͑7.6 mm long͒, where the Rydbergs are field ionized by a 20
ns, 50 V/cm pulse. The resulting MATI ions are focused by
a set of lenses into a quadrupole ion guide, where a combi-
nation of focusing properties and time-of-flight ͑TOF͒ gating
is used to velocity filter the beam and reject any prompt ions
or ion fragments that might be present. The ions are next
injected into a two-section octapole ion guide. In the first
section, the collision energy is set and the NOϩ beam is
guided through a 11.3 cm cell containing 0.1 mTorr of eth-
anol vapor ͑Quantum Chemical Company͒. To allow correc-
tion for reactions occurring outside the cell, measurements
are also made with the same ethanol flow rate dumped di-
rectly into the vacuum chamber surrounding the cell. Product
ions, along with unreacted NOϩ, are collected by the octa-
pole and passed into the second section of the ion guide. For
these experiments this section was operated at a dc potential
ϳ1.5 V below that of the first section of the guide to improve
collection of slow product ions. TOF through the guide sys-
tem is used to measure the primary ion beam velocity distri-
bution, and also to measure recoil velocity distributions for
the product ions. Finally, the ions are collected and injected
into a conventional quadrupole mass filter, and counted by a
Daly detector/multichannel scalar combination. The cross
sections are calculated from the product signal with ethanol
filling the cell minus the signal with the ethanol flow di-
verted into the chamber. This subtraction compensates for
any reactions taking place in the second octapole segment.
The chief problem with using MATI for rotational state
selection is that angular momentum conservation dictates
that to produce rotationally hot NOϩ, the NO neutral precur-
sor must also be rotationally hot. MATI also requires that the
neutrals have a well-defined and reasonably high velocity to
allow Rydberg/ion separation, dictating use of a seeded neu-
tral beam. The concomitant rotational cooling makes produc-
tion of high-Nϩ state ions quite difficult. In the experiments
here, we are able to produce substantial intensities ͑ϳ5000
ions/s͒ for the Nϩϭ0,1 beam, but only 200–300 ions/s for
Nϩϭ10. In order to eliminate any intensity-dependent sys-
tematic errors in comparing the Nϩϭ0,1 and Nϩϭ10 re-
sults, the laser was attentuated for the Nϩϭ0,1 experiments
FIG. 1. MATI spectra of NO via the A 2⌺ intermediate state. The sharp
structure is the np Rydberg series converging on the ionization threshold for
each rotational state of the ion. Ions are produced by scanning the second
laser while the first laser excites a rotational line of the 0–0 band of the
A
R
2⌺–X 2⌸ transition. In the upper spectrum the first laser excites the
21(0.5) line. In the lower, it excites R21(8.5).
intermediate states. The sharp structure is the np series of
high-lying Rydberg states converging to each ion rotational
state. In the following we report reactivity of NOϩ in the
unresolved mixture of Nϩϭ0 and 1, and in Nϩϭ10.
The Rydberg excitation process also generates ions that
are not in the desired state, and these are rejected by a weak
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.143.1.222 On: Fri, 12 Dec 2014 18:54:49

J. Chem. Phys., Vol. 113, No. 8, 22 August 2000
Hydride abstraction by NOϩ
3005
so that the ion beam intensity was equally poor for the entire
data set.
Two products are observed in the reaction, at m/q 45
and m/q 47. The former is the major C2H5Oϩ product, and,
as in the ICR experiments of Williamson and Beauchamp,30
the latter is attributed to a secondary reaction of the major
product ion: C2H5OϩϩC2H5OH→C2H7OϩϩC2H4O (⌬Hrxn
ϭϪ0.04 eV). It is not practical to eliminate the secondary
reaction, because the primary reaction generates some
C2H5Oϩ that is very slow in the laboratory frame, with high
secondary reaction probability. Owing to the very low pri-
mary ion intensity, we did not attempt to measure the pres-
sure dependence of the cross sections. Taking advantage of
the fact that the only source of m/q 47 is secondary reaction
of the primary product ion, we simply summed the m/q 47
and 45 signals to calculate the reaction cross section.
One potentially important issue is the possibility that ro-
tationally inelastic collisions might perturb the reactant rota-
tional state distributions prior to reaction. We can estimate
the importance of nonreactive collisions by observing
changes in the NOϩ beam pulse when the scattering cell is
filled with ethanol vapor. Reactive collisions lead to loss of
NOϩ intensity, but nonreactive collisions lead to NOϩ that is
shifted to lower laboratory energies relative to the sharp re-
actant beam pulse. We estimate that the total fraction of ions
having a single collision ranges from 34% at low energies to
24% at high energies. This fraction includes reactive colli-
sions and any elastic or inelastic collisions that result in sig-
nificant change in the lab velocity. Based on these fractions,
we estimate that ϳ10% of the reactive signal could result
from reactant ions that had previously undergone a nonreac-
tive collision. While 10% multiple collisions is undesirable,
it would not obscure a substantial rotational effect on the
reaction cross section. Ordinarily we operate at lower scat-
tering gas densities to avoid the issue, but the low primary
beam intensity and small reaction cross section mandate high
target densities.
FIG. 2. The velocity distribution of the reactant NOϩ (Nϩϭ0,1) MATI
beam at a nominal energy of 0.3 eV, together with a Gaussian fit.
energy for scattering cell empty and full͒. Ordinarily, we
simply average data from each cycle together, and because
the cycle-to-cycle collision energy distribution is reasonably
constant ͑20–30 meV͒, the effect on collision energy resolu-
tion is small.
For the NOϩϩEtOH reaction at low collision energies,
the cross section is such a steep function of the collision
energy that even small shifts in the collision energy distribu-
tion can lead to significant errors in comparing data at the
same nominal energy. We were, therefore, forced to abandon
averaging, and treat each acquisition cycle separately. For
each cycle we have analyzed the primary beam TOF, reject-
ing any cycles where the reactant velocity distribution was
broadened or distorted. The cross section for each cycle is
then determined from the primary and product signals from
that cycle, and plotted against the collision energy taken as
the mean of a Gaussian fit to the NOϩ velocity distribution
measured during that cycle. This procedure results in having
a well-defined collision energy associated with each cross-
section point, but with minimal signal averaging to compen-
sate for shot-to-shot intensity fluctuations. The combination
of low primary intensities and short averaging times results
in badly scattered cross-section points ͑including negative
apparent cross sections due to the subtraction procedure, de-
scribed previously͒. Conclusions regarding the collision en-
ergy and rotational state dependence of the cross sections
must, therefore, be taken from fits to the data.
Figure 2 shows a velocity profile of the reactant MATI
beam at 0.3 eV nominal collision energy and also a Gaussian
fit used to calculate the mean collision energy ( E
͗
͘
col
ϭ0.304 eV). The beam velocity distribution is determined
by a number of factors, including the potential distributions
in the ionization volume and scattering octapole, the homo-
geneity of the field ionization pulse, and the neutral precur-
sor velocity distribution. Because factors such as the laser
alignment, pulsed valve-laser timing, surface potentials, and
time/spatial profile of the field ionization pulse can vary from
experiment to experiment, the actual collision energy distri-
bution associated with a particular nominal energy varies.
The variations can be large ͑0.2 eV͒ if the laser is realigned
or some experimental setting is changed, but during the
course of one set of experiments the collision energy is gen-
erally quite stable ͑20–30 meV͒. Multiphoton ionization pro-
duces beams with both shot-to-shot and long-term intensity
fluctuations. To compensate for shot-to-shot fluctuations,
lengthy signal averaging is required, and to avoid systematic
errors from the long term drift, the averaging is done as a set
of quick cycles through a complete data set ͑primary and
product ion intensities and TOFs as a function of collision
III. RESULTS
Cross sections for hydride abstraction are shown in Fig.
3 over the collision energy range from ϳ50 to ϳ500 meV.
The inset shows the full range of collision energy studied
͑ϳ50 meV to ϳ3.7 eV͒. Data are shown for reaction of a
mixture of Nϩϭ0 and 1 ͑low Nϩ͒, and for Nϩϭ10 ͑high
Nϩ͒. At low collision energy, the cross section approaches
25 Å2, but falls off rapidly as the collision energy increases.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.143.1.222 On: Fri, 12 Dec 2014 18:54:49

3006
J. Chem. Phys., Vol. 113, No. 8, 22 August 2000
Green et al.
substantial populations of ϩϭ0–4. Even modest vibra-
v
tional enhancement would account for the higher rate con-
stant measured in the ICR.
Comparison of the data shows immediately that there is
no major effect of the rotational states studied. The scatter in
the data points is so large, however, that a small effect would
be obscured. Indeed, the confidence limits on an uncon-
strained statistical analysis ͑e.g., two dimensional
Kolmogorov–Smirnov40͒ are still too large to be helpful in
deciding if there might be a small rotational effect. To look
for possible small effects, we adopted the following proce-
dure. We start with the assumption that the cross sections for
low and high Nϩ are, in fact, identical. We then fit the com-
bined set of low and high Nϩ data points to a single power
law expression: ␴ϭK•Encol , with best fit parameters K
ϭ0.52Ϯ0.07 and nϭϪ1.3. This functional form was chosen
because it gives an acceptable fit with only two parameters.
Note that it is thermal averaging of this expression that gives
a rate constant in good agreement with the Spanel and Smith
SIFDT experiment, confirming that the cross section really
does rise sharply at very low collision energies.
We then fit the low NϪ and high Nϩ distributions inde-
pendently using the single parameter expression: ␴ϭK
•EϪco1l .3 , where the exponent ͑Ϫ1.3͒ was taken from the fit to
the combined data set. In essence, this analysis makes the
assumption that rotation might affect the magnitude of the
cross section, but not the collision energy dependence. The
values of the scale factor K are 0.50Ϯ0.14 for low Nϩ, and
0.54Ϯ0.07 for high Nϩ, where the uncertainties are 95%
confidence limits. The small difference in scale factors is
equivalent to 0.60 Å2 at 0.1 eV and is certainly not signifi-
cant. Similar analysis was done in which we ignored outly-
ing points in the fits. Because of the scatter in the data, we
cannot rule out the possibility that rotational energy may
enhance reactivity, i.e., ignoring the outliers causes the K
value for high-NϩNOϩ at low energy to be ϳ20% larger
than that for low Nϩ.
FIG. 3. A scatter plot of the cross section for the hydride abstraction reac-
tion as a function of collision energy for the rotational states studied. The
inset shows the full collision energy range studied, while the main figure
shows the low energy range. The curves are power laws fits to the data
Ϫ1.3
col
(␴ϭK•E
). See the text for details.
By Ecolϭ1.5 eV the cross section is essentially zero within
the rather poor sensitivity limits dictated by the low Nϩ
ϭ10 intensity.
The conclusion is that rotation certainly does not inhibit
reaction for the states studied, but we cannot rule out the
possibility that rotation slightly enhances reactivity. In any
case, any effect is certainly within the experimental error.
We also cannot rule out the unlikely case of a nonmonotonic
dependence on rotational state that accidentally vanishes at
the states studied. The lack of rotational dependence is in
contrast to the strong inhibitory effect of collision energy. At
our lowest collision energy ͑50 meV͒, adding 27 meV of
rotational energy (Nϩϭ10) has no detectable effect. Adding
the same amount of collision energy decreases the cross sec-
tion by nearly a factor of 2. The collision energy effect is
particularly striking in light of the fact that, on average, the
thermal energy of the ethanol reactant is 76 meV. Thus, 27
meV represents only a 21% increase in the total energy.
One issue in developing a reaction mechanism is
whether Hϩ is abstracted from methyl, methylene, and/or
hydroxyl positions on EtOH. The most energetically favor-
able C2H5Oϩ product is CH3CHOHϩ, and in this case the
reaction is exoergic by ϳ16 kcal/mol.41 Williamson and
Beauchamp30 showed that for their thermal energy collisions,
hydride abstraction occurred only from the alkyl moiety. Our
The thermal rate constant reported by Williamson and
Beauchamp30 ͑8.0ϫ10Ϫ10 cm3/molecules s) is 30% of the
capture rate constant for ion–ethanol collisions, calculated
using Troe’s statistical adiabatic channel theory.38 The result
of Spanel and Smith31 (5ϫ10Ϫ10 cm3/molecules s) corre-
sponds to 20% of the capture rate. Our cross section at the
lowest experimental energy ͑ϳ50 meV͒ is only 7% of the
capture cross section. By extrapolating the functional fit ͑see
the following͒ of the collisional energy dependence of our
cross section to thermal energies and averaging over a
Maxwell–Boltzmann distribution of collision energies, we
derive a rate constant of 4.3ϫ10Ϫ10 cm3/molecules s, in rea-
sonable agreement with the result of Spanel and Smith. The
origin of the discrepancy with the rate of Williamson and
Beauchamp is not clear, but vibrational excitation seems a
likely culprit. In our experiment and in the SIFDT experi-
ments of Spanel and Smith, the reactant ions are in their
vibrational ground state, while in the ICR experiments of
Williamson and Beauchamp, the NOϩ may have had some
vibrational excitation. Indeed, estimating the ionization
Franck–Condon factors from the photoelectron spectrum39
of NO, suggests that electron-impact ionization will produce
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.143.1.222 On: Fri, 12 Dec 2014 18:54:49

J. Chem. Phys., Vol. 113, No. 8, 22 August 2000
Hydride abstraction by NOϩ
3007
lowest energies are in the range present in ICR experiments,
thus abstraction from the alkyl moiety should dominate for
us, as well. To get a rough estimate of the relative difficulty
of abstracting HϪ from CH3 , OH, and CH2 groups, we com-
A more likely scenario involves a precursor complex. In
ion–molecule systems, there often are weakly bound reac-
tantlike structures that can support an intermediate complex.
Generally, there is no energetic barrier to formation of such
reactantlike complexes and the transition state (TSR) sepa-
rating reactants from the complex is loose—near the orbiting
limit. If a precursor complex forms, it can either dissociate
back to reactants via the loose transition state (TSR), or pass
through the rate-limiting transition state (TSHA) and on to
products.
pared the energies of single point calculations at the
42
MP2/6-31G level using GAUSSIAN 98 for CH3CHOHϩ,
*
CH3CH2Oϩ, and CH2CH2OHϩ where the structures were
frozen in the ethanol geometry. In all cases, the singlet states
are substantially lower in energy than the triplets. Among the
singlets, CH3CHOHϩ is more stable than CH2CH2OHϩ by
1.75 eV, and more stable than CH3CH2Oϩ by 4.4 eV. These
frozen geometry calculations certainly do not give the true
difference in reaction path energetics for abstraction from the
three sites, however, they do suggest that abstraction from
the methylene site probably dominates the reaction mecha-
nism, at least in the low collision energy range of greatest
interest.
*
MP2/6-31G calculations were carried out to look for
ϩ
complexes in the NO–EtOH system. Figure 4 shows the
͓
͔
six complexes found and their energies relative to reactants,
including zero point energy corrections. Both top and side
views are shown except in the case of Ex1, where the heavy
atoms are roughly coplanar. The complexes can be charac-
terized as reactantlike ͑Er1, En2, En3͒, productlike ͑Ex1͒,
and insertion complexes ͑I1, I2͒. Of these complexes, the
reactantlike are the lowest in energy, and are obvious candi-
dates to serve as precursor complexes.
IV. DISCUSSION
All three reactantlike complexes have the NO cation
complexed to the hydroxyl O atom—the most negatively
charged atom in EtOH. En1 and En2 both have the more
positive N atom in NOϩ coordinated to the O atom in EtOH,
and the NO orientation is roughly parallel to the direction of
the dipole moment in free EtOH. In this context it should be
noted that EtOH has two conformers, calculated to be nearly
At high collision energies, reaction mechanisms tend to
be direct, because binding energies are too weak compared to
the collision energy to stabilize long-lived complexes. The
small cross section observed at high energies indicates that
direct HϪ abstraction has an efficiency less than 1% at ener-
gies over 1.5 eV. The balance of the discussion will focus on
the reaction mechanism in the more interesting low-
collision-energy regime.
From a dynamical point of view, the most interesting
observation is the unusually strong collision energy depen-
dence. At low collision energies, ion–molecule reactions
with no activation barriers often occur with cross sections
that are a substantial fraction of the collision cross section.
At our lowest collision energy, the experimental reaction ef-
ficiency, defined here as the ratio of the measured cross sec-
tion to the capture cross section,38 is roughly 7%. Efficiency
at Ecolϭ0.2 eV has dropped to ϳ3.5%, and above 1 eV the
efficiency is less than 1%.
*
isoenergetic ͑⌬EϽ5 meV at both the MP2/6-31G and G3
levels of theory͒. The barrier to interconversion is 43 meV at
*
MP2/6-31G . One conformer has OH oriented coplanar with
the CC bond, as in En1, and the other has OH oriented such
that the CCOH dihedral angle is ϳ65°, roughly as in En2. In
essence, the En1 and En2 complexes have NOϩ coordinated
in geometries that maximize ion–dipole and dipole–dipole
forces for the two conformers of EtOH. In En3 the CCOH
dihedral angle is reduced to ϳ10°, and more important, NOϩ
is bound with its oxygen atom closest to the hydroxyl O
atom, resulting in somewhat lower binding energy.
The fact that these reactantlike complexes have geom-
etries that appear to maximize the ion–dipole ͑and for En1
and En2, the dipole–dipole͒ attractions, suggests that long-
range forces during approach should tend to orient the reac-
tants to maximize complex formation, with the specific com-
plex formed depending on approach geometry. Note that the
binding energies of these complexes are more than five times
the collision energy over the entire dynamically interesting
energy range (EcolϽ0.25 eV). With these properties, forma-
tion of the reactantlike complexes is expected to be quite
efficient, as required for a precursor mechanism.
Such a sharp decrease in reaction efficiency with in-
creasing energy is usually the result of a bottleneck ͑tight
transition state͒ that inhibits reaction. Reaction clearly pro-
ceeds with no activation energy, as shown by the Ecol depen-
dence of the cross section, thus the rate-limiting transition
state for hydride abstraction (TSHA) must be below the en-
ergy of the reactants. There are at least two ways in which a
tight geometric constraint can result in reaction efficiency
that decreases with increasing energy. For example, it might
be that NOϩ must approach EtOH in the correct orientation
for reaction to occur. Collisions at low energies might be
efficient because long-range forces orient the reactants as
they approach, while at high energies there is insufficient
time for the long-range torques to have a significant effect. In
the NOϩ–EtOH system, however, this ‘‘approach geom-
etry’’ scenario is unlikely. The dominant long-range force is
the ion–dipole force that tends to orient EtOH. The dipole
moment of ethanol is such that ion–dipole interaction favors
orienting the hydroxyl O atom toward the ion, whereas one
would expect that the reactive geometry must have one of
the labile methylene H atoms directed at NOϩ.
En3 has the NOϩ moiety correctly oriented for nitrogen
attack on the methylene CH bond, whereas En1 and En2
require more extensive rearrangement to allow hydride ab-
straction. Mechanistically, it is important to know whether
the more stable En1 and En2 complexes can easily convert to
En3. Rather than searching for the transition states for inter-
conversion, we did a series of calculations on structures mid-
way between En2 and En3, varying the distance and angle of
NO relative to EtOH. The lowest energy geometry was only
ϳ0.22 eV above the energy of En3, or 0.55 eV above En2,
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.143.1.222 On: Fri, 12 Dec 2014 18:54:49

3008
J. Chem. Phys., Vol. 113, No. 8, 22 August 2000
Green et al.
the reactantlike complexes are long enough to be mechanis-
tically significant. To address this issue we performed Rice–
Ramsperger–Kassel–Marcus ͑RRKM͒ calculations,43 using
*
scaled frequencies from the MP2/6-31G calculations and
assuming an orbiting transition state44–46 for TSR . At the
lowest total energies in the experiment, both En1 and En2
complexes are quite long lived ͑tens of nanoseconds͒, and
the lifetime is still hundreds of picoseconds at Ecolϭ0.25 eV.
En3, because it is less strongly bound and also less floppy, is
shorter lived, with lifetimes ranging from a few nanoseconds
at Ecolϭ0.05 eV to a few picoseconds at 0.25 eV. These
RRKM results neglect the rapid interconversion between
complexes, and the true lifetime is determined by the aggre-
gate density of states in the set of complexes. As most of the
density is found in the En1 and En2 wells, the resulting
lifetime should be at least hundreds of picoseconds over the
energy range of interest (EcolϽ0.25 eV). The RRKM calcu-
lations also neglected the rate of the HϪ abstraction reaction,
however, the small experimental cross section suggests that
this assumption is reasonable. Note that the fraction of the
aggregate complex lifetime that is spent in each potential
well depends on the ratios of the density of states in the
wells. The fraction varies somewhat with energy and angular
momentum, but is roughly 0.4, 0.6, and 0.003, for En1, En2,
and En3, respectively.
The conclusion that the precursor complex lifetime is
long is consistent with experiment. We have measured the
axial recoil velocity distributions for the product ions, and
while the small cross section and poor kinematics makes
quantitative interpretation impossible, the measurements are
consistent with forward–backward symmetric recoil
velocities.33 Such velocity distributions are a necessary con-
dition for concluding that reaction is mediated by a complex
with lifetime greater than a few picoseconds.
The above-mentioned considerations strongly suggest
that formation of a long-lived precursor ͑i.e., the set of inter-
converting reactantlike complexes͒ is facile at low collision
energies. Once the precursor has formed, the actual mecha-
nism for hydride abstraction is unclear. For reaction at low
energies, a likely mechanism is ‘‘insertion–elimination,’’
where NOϩ inserts in a methylene CH bond, followed by
HNO elimination. Indeed, the very low reactivity observed at
high collision energies, where direct reaction is expected to
dominate, tends to suggest that direct abstraction is ineffi-
cient. In an insertion–elimination mechanism, the complexes
I1 and I2 are likely candidates for insertion intermediates.
Both complexes have NO inserted into a methylene CH
bond; however, I2 is much lower in energy and can dissoci-
ate directly to HNOϩC2H5Oϩ products. I1, in contrast, is a
high-energy geometry, and also would require HON→HNO
isomerization to reach the ground state products. ͑The HON
isomer is 2.17 eV higher in energy than HNO, thus we can
exclude the HONϩC2H5OϪ product channel.͒ Given these
considerations, the rate-limiting transition state (TSHA) for
an insertion–elimination mechanism is almost certainly at
the critical configuration for insertion to form I2.
*
FIG. 4. Geometries of stable complexes optimized at the MP2/6-31G level
of theory. Reactantlike complexes ͑En1, En2, En3͒, insertion-type com-
plexes ͑I1, I2͒, and a productlike complex ͑Ex1͒ are shown. A top and side
view of each complex is shown with the exception of Ex1 in which the
heavy atoms are roughly planar. The energy of each complex is shown
relative to the energy of the separated reactants.
and its energy should be a reasonable upper limit on the
energy of the true transition state. As this energy is well
below the available energy in the complexes, it is clear that
interconversion between complexes should be facile.
The other obvious question is whether the lifetimes for
The alternative to insertion–elimination is direct abstrac-
tion. As noted, direct abstraction is quite inefficient at high
collision energies, but it is conceivable that there is a con-
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.143.1.222 On: Fri, 12 Dec 2014 18:54:49

J. Chem. Phys., Vol. 113, No. 8, 22 August 2000
Hydride abstraction by NOϩ
3009
certed abstraction pathway that might be efficient at low en-
ergies, where the collision time is long. In such a concerted
mechanism, the transition state (TSHA) would presumably be
at the barrier for attack of NOϩ on the hydrogen, and the
nascent HNO would not be chemically bonded to the
CH3CHOHϩ moiety. Products might either separate imme-
diately, or possibly pass through a productlike complex such
as Ex1. We did a number of calculations to probe the ener-
getics of NOϩ attack on the methylene hydrogen. Optimiza-
tions were started with NOϩ positioned at several distances
͑ranging down to ϳ1.7 Å͒ and orientations relative to the
methylene CH bond. These geometries all converged back to
reactantlike complexes, suggesting that the transition state
for this mechanism lies at short NO–EtOH separations ͑and
high energy͒.
We made numerous attempts to optimize transition
states for both insertion and direct abstraction. Optimizations
were attempted using the transit-guided quasi-Newton and
Berny algorithms, implemented in GAUSSIAN98. The only
transition states found were for uninteresting motions such as
torsions. The higher order saddle points found were substan-
tially above the reactant energy, and therefore poor approxi-
mations to TSHA, which must lie below the reactant energy.
We also attempted to distort second-order saddle points into
reasonable starting geometries for TS optimizations. All at-
tempts failed.
For either mechanism, TSHA must be a relatively com-
pact structure with rotational constants not radically different
from those of the precursor complexes. The real unknowns
are the vibrational tightness of TSHA and its energy with
respect to reactants. The TSHA transition state, particularly
for an insertion reaction, is likely to be substantially stiffer
than either En3 or I2. In contrast, the orbiting transition state
(TSR) is very loose, with rotations replacing the low fre-
quency vibrations in the complexes. Because we have only
estimates for the properties of TSHA, we cannot calculate
reaction efficiency using RRKM theory. Nonetheless, the
qualitative picture suggests that vibrational stiffness should
contribute to a substantial bottleneck inhibiting reaction.
It is also clear that some of the bottleneck must result
from angular momentum conservation, as follows. As the
precursor forms, the angular momentum of the collision must
be converted to tumbling rotation of the complex, and be-
cause the moment of inertia in the complex is substantially
lower than at the orbiting transition state (TSR), considerable
energy is tied up in rotation of the complex. In passage
through a compact TS, such as TSHA, this rotational energy
is not available to drive reaction, whereas in dissociation
back through TSR , much of the rotational energy is con-
verted back to translation to drive dissociation. As a conse-
quence, reaction is suppressed in high angular momentum
collisions. Because angular momentum increases with colli-
sion energy, reactivity tends to be inhibited by increasing
energy.
ture model, i.e., that all collisions with LϽLmax lead to
capture, we can also estimate the average L. At our lowest
collision energy ͑0.05 eV͒ Lmax is ϳ200ប and Lavg is ϳ140ប,
but Lavg increases quite slowly with increasing Ecol . Increas-
ing the collision energy by a factor of 6 to 0.3 eV increases
Lmax and Lavg by only a factor of 1.5, but decreases the
reaction cross section by an order of magnitude. For a colli-
sion at 0.05 eV with LϭLavg , the energy going into tum-
bling rotation of the complex is ϳ0.19 eV, compared to the
total available energy EcolϩEinternal͑EtOH͒ϩwell depth of
͓
͔
1.50 eV ͑for En2͒. Similarly, for the average collision at 0.3
eV, the rotational energy in the precursor is ϳ0.44 eV, com-
pared to Etotalϭ1.75 eV. Certainly the rotational energy in
the complex is not insignificant, but given the slow increase
in L with Ecol , it seems surprising that angular momentum
conservation could completely account for the bottleneck to
reaction. We conclude that both vibrational tightness and an-
gular momentum effects probably contribute to suppressing
reaction at higher energies.
In this regard, it is somewhat puzzling that there is no
rotational effect. There are three possible explanations. One
possibility is that angular momentum conservation is, indeed,
the major constraint on reaction, and that the rotational an-
gular momentum is simply too small to have a significant
effect. The rotational angular momentum is only 10 ប, com-
pared to the collisional Lavg of Ͼ100 ប, and because it adds
vectorially, has little effect on the total angular momentum.
A second possibility is that the precursor complex potential
might be isotropic with respect to NO rotation, such that the
nascent rotational energy does not couple in the complex and
is not available to drive decomposition. Given the fact that
both EtOH and NOϩ have substantial dipole moments, such
isotropic interaction is implausible. We tested this idea with
a series of ab initio single point calculations, in which the
NO–EtOH separation was fixed, and the NO was rotated in
three orthogonal planes. As expected, the binding is quite
anisotropic and we estimate that the barrier to NOϩ rotation
within the complex is 0.8–1.25 eV. A third possibility is to
invoke some role for dynamics in the precursor complex
mechanism discussed previously. As the precursor complex
forms, NOϩ rotation will transform into intermolecular
bending vibrations. While the additional energy in these vi-
brations will tend to shorten the complex lifetime, it is pos-
sible that bending motion is particularly effective at driving
the system through the critical configuration for HϪ abstrac-
tion ͑i.e., TSHA͒.
In summary, the most likely explanation for the sharp
collision energy dependence is a tight bottleneck for HϪ ab-
straction. The properties of the complexes in this system sug-
gest a contribution to the bottleneck both from vibrational
tightness and angular momentum conservation constraints.
Too little is known about the nature of the bottleneck to say
with much confidence, why rotation is ineffective. Further
quantum chemistry, and perhaps trajectory studies of
NOϩ–EtOH collisions could presumably resolve this
question.
We can estimate the maximum orbital angular momen-
tum in collisions that can form complexes as Lmaxϭ␮•
vrel
•bmax , where ␮ is the reduced mass,
is the relative ve-
vvel
locity of the reactants, and bmax is the maximum impact pa-
rameter leading to capture. Given the assumption of the cap-
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.143.1.222 On: Fri, 12 Dec 2014 18:54:49

3010
J. Chem. Phys., Vol. 113, No. 8, 22 August 2000
Green et al.
25 A. A. Viggiano, J. M. V. Doren, R. A. Morris, J. S. Williamson, P. L.
Mundis, and J. F. Paulson, J. Chem. Phys. 95, 8120 ͑1991͒.
26 C. E. Dateo and D. C. Clary, J. Chem. Soc., Faraday Trans. 2 85, 1685
͑1989͒.
ACKNOWLEDGMENTS
The authors would like to thank Stephen Klippenstein,
Alex Boldyrev, William Hase, and Jack Simons for helpful
suggestions. This work was supported by the National Sci-
ence Foundation under Grant No. CHE-9807625. We are
also grateful for a grant of computer time from the Utah
Center for High Performance Computing. The SGI Origin
2000 used was funded in part by the SGI Supercomputing
Visualization.
27 I. Dotan, A. A. Viggiano, and R. A. Morris, J. Chem. Phys. 96, 7445
͑1992͒.
28 A. A. Viggiano and R. A. Morris, J. Phys. Chem. 100, 19227 ͑1996͒.
29 K. P. Huber and G. Herzberg, Molecular Spectra and Molecular Struc-
ture. IV. Constants of Diatomic Molecules ͑Van Nostrand Reinhold, New
York, 1979͒.
30 A. D. Williamson and J. L. Beauchamp, J. Am. Chem. Soc. 97, 5714
͑1975͒.
31 P. Spanel and D. Smith, J. Chem. Phys. 104, 1893 ͑1996͒.
32 J. Qian, H. Fu, and S. L. Anderson, J. Phys. Chem. 101, 6504 ͑1997͒.
33 Y.-h. Chiu, H. Fu, J.-t. Huang, and S. L. Anderson, J. Chem. Phys. 102,
1199 ͑1995͒.
1 Y. Ikezoe, S. Matsuoka, M. Takebe, and A. Viggiano, Gas Phase Ion-
Molecule Reaction Rate Constants through 1986 ͑Mass Spectroscopy So-
ciety of Japan, Tokyo, 1987͒.
2 W. A. Chupka, M. E. Russell, and K. Refaey, J. Chem. Phys. 48, 1518
͑1968͒.
34 Y.-h. Chiu, H. Fu, J.-t. Huang, and S. L. Anderson, J. Chem. Phys. 102,
1188 ͑1995͒.
3 C. L. Liao, C. X. Liao, and C. Y. Ng, J. Chem. Phys. 81, 5672 ͑1984͒.
4 A. Fujii, T. Ebata, and M. Ito, Chem. Phys. Lett. 161, 93 ͑1989͒.
5 J. Xie and R. N. Zare, J. Chem. Phys. 93, 3033 ͑1990͒.
6 D. Gerlich and T. Rox, Z. Phys. D: At., Mol. Clusters 13, 259 ͑1989͒.
7 S. L. Anderson, Adv. Chem. Phys. 82, 177 ͑1992͒.
8 T. Glenewinkel-Meyer and D. Gerlich, Isr. J. Chem. ͑in press͒.
9 C. W. Eaker and G. C. Schatz, J. Phys. Chem. 89, 2612 ͑1985͒.
10 L. Zhu and P. Johnson, J. Chem. Phys. 94, 5769 ͑1991͒.
11 T. P. Softley, S. R. Mackenzie, F. Merkt, and D. Rolland, Adv. Chem.
Phys. 101, 667 ͑1997͒.
35 H.-T. Kim, R. J. Green, J. Qian, and S. L. Anderson, J. Chem. Phys. 112,
5717 ͑2000͒.
36 M. J. J. Vrakking and Y. T. Lee, J. Chem. Phys. 102, 8818 ͑1995͒.
37 F. Merkt and R. N. Zare, J. Chem. Phys. 101, 3495 ͑1994͒.
38 J. Troe, Chem. Phys. Lett. 122, 425 ͑1985͒.
39 K. Kimura, S. Katsumata, Y. Achiba, T. Yamazaki, and S. Iwata, Hand-
book of HeI Photoelectron Spectra of Fundamental Organic Molecules
͑Japan Scientific Societies Press, Tokyo, 1981͒.
40 G. Fasano and A. Franceschini, Mon. Not. R. Astron. Soc. 225, 155
͑1987͒.
12 C. Jouvet, C. Dedonder-Lardeux, S. Martrenchard-Barra, and D. Solgadi,
Chem. Phys. Lett. 198, 419 ͑1992͒.
41 S. G. Lias, J. E. Bartmess, J. F. Liebman, J. L. Holmes, and R. D. Levin,
J. Phys. Chem. Ref. Data Suppl. 17, 1 ͑1988͒.
13 K. F. Willey, C. S. Yeh, and M. A. Duncan, Chem. Phys. Lett. 211, 156
͑1993͒.
42 M. J. Frisch, G. W. Tucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J.
R. Cheeseman, V. G. Zakrzewski, J. A. Montgometry, R. E. Stratmann, J.
C. Burant, S. Dapprich, J. M. Millam, A. D. Daniels, K. N. Kudin, M. C.
Strain, O. Farkas, J. Tomasi, V. Barone, M. Cossi, R. Cammi, B. Men-
nucci, C. Pomelli, C. Adamo, S. Clifford, J. Ochterski, G. A. Peterson, P.
Y. Ayala, Q. Cui, K. Morokuma, D. K. Malick, A. D. Rabuck, K. Ragha-
vachari, J. B. Foresman, J. Cioslowski, J. V. Ortiz, B. B. Stefanov, G. Liu,
A. Liashenko, P. Piskorz, I. Komaromi, R. Gomperts, R. L. Martin, D. J.
Fox, T. Keith, M. A. Al-Laham, C. Y. Peng, A. Nanayakkara, C. Gonza-
lez, M. Challacombe, P. M. W. Gill, B. G. Johnson, W. Chen, M. W.
Wong, J. L. Andres, M. Head-Gordon, E. S. Replogle, and J. A. Pople,
GAUSSIAN 98, Gaussian, Inc., Pittsburgh, PA, 1998.
14 H. Krauss and H. J. Neusser, J. Chem. Phys. 97, 5923 ͑1992͒.
15 C. W. Hsu, K. T. Lu, M. Evans, Y. J. Chen, C. Y. Ng, and P. Heimann, J.
Chem. Phys. 105, 3950 ͑1996͒.
16 G. Lembach and B. Brutschy, Chem. Phys. Lett. 273, 421 ͑1997͒.
17 S. R. Mackenzie, E. J. Halse, E. Gordon, D. Rolland, and T. P. Softley,
Chem. Phys. 209, 127 ͑1996͒.
18 S. Sato and K. Kimura, Chem. Phys. Lett. 249, 155 ͑1996͒.
19 S. R. Mackenzie and T. P. Softley, J. Chem. Phys. 101, 10609 ͑1994͒.
20 S. R. Mackenzie, E. J. Halse, F. Merkt, and T. P. Softley, Proc. SPIE
2548, 293 ͑1995͒.
21 J. B. Marquette, C. Rebrion, and B. R. Rowe, J. Chem. Phys. 89, 2041
͑1988͒.
43 L. Zhu and W. L. Hase, Quantum Chemical Program Exchange ͑QCPE͒,
computer program 644.
22 D. Gerlich, J. Chem. Soc., Faraday Trans. 89, 2199 ͑1993͒.
23 D. Gerlich and G. Kaffer, Astrophys. J. 347, 849 ͑1989͒.
24 A. A. Viggiano, R. A. Morris, and J. F. Paulson, J. Chem. Phys. 90, 6811
͑1989͒.
44 M. T. Rodgers, K. M. Ervin, and P. B. Armentrout, J. Chem. Phys. 106,
4499 ͑1997͒.
45 M. T. Rodgers and P. B. Armentrout, J. Chem. Phys. 109, 1787 ͑1998͒.
46 E. V. Waage and B. S. Rabinovitch, Chem. Rev. 70, 377 ͑1970͒.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.143.1.222 On: Fri, 12 Dec 2014 18:54:49