Dynamics of reactions O((1)D)+C(6)H(6) and C(6)D(6)

Hui-Fen Chen, Chi-Wei Liang, Jim J. Lin, Yuan-Pern Lee, J. F. Ogilvie, Z. F. Xu, and M. C. Lin

Citation: The Journal of Chemical Physics 129, 174303 (2008); doi: 10.1063/1.2994734 View online: http://dx.doi.org/10.1063/1.2994734

View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/129/17?ver=pdfcov Published by the AIP Publishing

Articles you may be interested in

Measurement of the differential cross section of the photoinitiated reactive collision of O ( D 1 ) + D 2 using only one molecular beam: A study by three dimensional velocity mapping

J. Chem. Phys. 132, 244308 (2010); 10.1063/1.3427534

Reaction dynamics of O H + ( 3 ) + C 2 H 2 studied with crossed beams and density functional theory calculations

J. Chem. Phys. 125, 133117 (2006); 10.1063/1.2212417

Reactive quenching of OH ( A + 2 ) by D 2 studied using crossed molecular beams J. Chem. Phys. 124, 201106 (2006); 10.1063/1.2206779

Reaction dynamics of C N + O 2 N C O + O ( P 2 3 ) J. Chem. Phys. 124, 094307 (2006); 10.1063/1.2173261

Dynamics of the C ( D 1 ) + D 2 reaction: A comparison of crossed molecular-beam experiments with quasiclassical trajectory and accurate statistical calculations

J. Chem. Phys. 122, 234309 (2005); 10.1063/1.1930831

Dynamics of reactions O

_„

1

D

_…+C

₆

H

₆

and C

₆

D

₆

Hui-Fen Chen,1Chi-Wei Liang,2Jim J. Lin,3,4,a兲Yuan-Pern Lee,3,4,a兲 J. F. Ogilvie,5 Z. F. Xu,6and M. C. Lin3,6,a兲

1

Department of Chemistry, National Tsing Hua University, Hsinchu 30013, Taiwan

2

Department of Chemistry, National Taiwan University, Taipei 10617, Taiwan

3

Department of Applied Chemistry and Institute of Molecular Science, National Chiao Tung University, Hsinchu 30010, Taiwan

4

Institute of Atomic and Molecular Sciences, Academia Sinica, Taipei 10617, Taiwan

5

Escuela de Qumica, Universidad de Costa Rica, Ciudad Universitaria Rodrigo Facio, San Pedro de Montes de Oca, San Jose 2060, Costa Rica

6

Department of Chemistry, Emory University, Atlanta, Georgia 30322, USA

共Received 8 July 2008; accepted 15 September 2008; published online 4 November 2008兲 The reaction between O共1_{D兲 and C6H}

6 共or C6D6兲 was investigated with crossed-molecular-beam reactive scattering and time-resolved Fourier-transform infrared spectroscopy. From the crossed-molecular-beam experiments, four product channels were identified. The major channel is the formation of three fragments CO+ C5H5+ H; the channels for formation of C5H6+ CO and C6H5O + H from O共1D兲+C6H6 and OD+ C6D5 from O共1D兲+C6D6 are minor. The angular distributions for the formation of CO and H indicate a mechanism involving a long-lived collision complex. Rotationally resolved infrared emission spectra of CO共1ⱕ␷ⱕ6兲 and OH 共1ⱕ␷ⱕ3兲 were recorded with a step-scan Fourier-transform spectrometer. At the earliest applicable period 共0–5 ␮s兲, CO shows a rotational distribution corresponding to a temperature of ⬃1480 K for ␷

= 1 and 920–700 K for ␷= 2 – 6, indicating possible involvement of two reaction channels; the vibrational distribution of CO corresponds to a temperature of⬃5800 K. OH shows a rotational distribution corresponding to a temperature of⬃650 K for␷= 1 – 3 and a vibrational temperature of ⬃4830 K. The branching ratio of 关CO兴/关OH兴=2.1⫾0.4 for O共1_D兲+C6H

6 and关CO兴/关OD兴⬎2.9 for O共1_D兲+C6D

6 is consistent with the expectation for an abstraction reaction. The mechanism of the reaction may be understood from considering the energetics of the intermediate species and transition states calculated at the G2M共CC5兲 level of theory for the O共1_D兲+C6H

6 reaction. The experimentally observed branching ratios and deuterium isotope effect are consistent with those predicted from calculations. © 2008 American Institute of Physics.关DOI:10.1063/1.2994734兴

I. INTRODUCTION

The reactions of singlet oxygen atoms O共1

D兲 are

impor-tant in atmospheric chemistry because O共1_{D兲 is highly} reactive toward small molecules in the atmosphere. Reac-tions of O共1_{D兲 with hydrogen,}1–3

water,4–8methane,9–15 and higher saturated hydrocarbons16–23 have been extensively investigated. Previous experiments on the dynamics of formation of OH from reactions of O共1

D兲 with saturated

hy-drocarbons indicated two possible paths: insertion of O共1_D兲 and direct abstraction of H.16,24–26 Several theoretical in-vestigations16–19,27–30 on the reaction mechanisms of O共1D兲

+ CH4, C2H6, and c-C3H6supported the reported experimen-tal observations.

In contrast, reactions between O共1_{D兲 and unsaturated} hydrocarbons have been investigated to a less extent.31–35 Sato and Cvetanović31 and Kajimoto et al.33 reported the formation of both enols, via an insertion of O共1_{D兲 into a} C–H bond, and epoxides, via an addition of O共1_{D兲 into a} C = C double bond, in these reactions. Based on the observed

relative product yields, Kajimoto et al.33 concluded that the epoxide channel occurs more readily than the enol channel. Honma34 employed laser-induced fluorescence 共LIF兲 to de-termine the distributions of rotational and vibrational states of OH produced from the reaction of O共1D兲 with C2H₄under flow conditions at low pressure and reported bimodal rota-tional distributions of OH 共␷= 0 and 1兲. Gonzalez et al.35 reported bimodal rotational distributions for OH共␷= 0 and 3兲 and unimodal ones for OH共␷= 1 and 2兲 determined with LIF; they suggested that, in contrast to what was proposed by Kajimoto et al.,33the reaction evolves preferentially via in-sertion of O共1

D兲 into the C⫺H bond, yielding internally

cold OH through slow decomposition of an enol-type inter-mediate and internally excited OH by rapid elimination be-fore the relaxation of the internal energy of the intermediate. They proposed also that the production of rotationally cold but vibrationally hot OH 共␷= 3兲 occurred via an abstraction channel.

Several experimental36–40 and theoretical41,42 investiga-tions on reacinvestiga-tions of O共3

P兲 with aromatic compounds have

been reported. Sibener et al.36 investigated the reaction of O共3_P兲+C6H

6 with crossed-molecular beams and concluded that the initially formed triplet biradical C6H6O either

de-a兲_{Authors to whom correspondence should be addressed. Electronic}

ad-dresses: jimlin@gate.sinica.edu.tw, yplee@mail.nctu.edu.tw, and

chemmcl@emory.edu.

0021-9606/2008/129共17兲/174303/14/$23.00 129, 174303-1 © 2008 American Institute of Physics

composes to eliminate a hydrogen atom关reaction共1兲兴 or be-comes stabilized likely via a nonradiative transition to the ground state共S0兲 manifold of phenol 关reaction共2兲兴,

O共3_{P兲 + C6H} 6→ C6H5O + H, 共1兲 O共3 P兲 + C6H6→ C6H5OH, 共2兲 O共3P兲 + C6H₆→ C₅H₆+ CO, 共3兲 O共3_{P兲 + C6H} 6→ C6H5+ OH. 共4兲

Barry et al.38 investigated the reaction of a crossed-molecular beam of O共3

P兲+C6H6 at a collision energy of 16.5 kcal mol−1 and reported little 共0.8 kcal mol−1兲 rota-tional excitation of the OH product detected by LIF; the results indicate that the reaction might proceed directly via an O⫺H⫺C collinear transition structure. Theoretical calculations41,42indicated that reactions共1兲and共2兲are major channels, whereas reaction 共4兲 becomes important at high temperatures with an estimated branching ratio of 50% at 2000 K. Reaction 共3兲 was predicted to be a minor channel with a yield of⬍5% even under flame conditions.

There is no report on the kinetics or dynamics of the reaction O共1D兲+C6H₆. On the basis of the present under-standing of reactions of O共3P兲 with benzene and of O共3P兲

and O共1D兲 with alkanes and alkenes, the reaction of O共1D兲

with benzene is expected to occur more readily than of O共3

P兲 with benzene and to produce both singlet phenol

共C6H5OH兲 and benzene epoxide 共O⬍C6H6兲 as intermedi-ates, in which “O⬍” in the formula indicates an epoxide structure. The mechanisms established previously for the thermal and photolytic decomposition of phenol might also assist us in understanding the reaction mechanism of O共1

D兲+C6H6.43,44

In this work, we have investigated the reactions of O共1

D兲+C6H6 and C6D6through the determination of trans-lational energy distributions and the branching ratios of vari-ous channels, O共1D兲 + C6H₆→ CO + C₅H₆, 共5a兲 O共1_{D兲 + C6H} 6→ CO + C5H6ⴱ→ CO + C5H5+ H, 共5b兲 O共1 D兲 + C6H6→ H + C6H5O, 共6a兲 O共1D兲 + C6H₆→ H + C₆H₅Oⴱ→ H + C₅H₅+ CO, 共6b兲 O共1_{D兲 + C6H} 6→ OH + C6H5, 共7兲

with crossed-molecular beams and by measurements of internal-state distributions and branching ratios of CO and OH with time-resolved Fourier-transform infrared 共FTIR兲 emission.45,46 Reactions共5a兲 and共5b兲produce CO as a pri-mary product; some C5H6 are stable, listed as reaction共5a兲,

and some C5H6 might have enough internal energy 共indi-cated as C5H6ⴱ兲 to dissociate further to C5H5+ H, listed as reaction 共5b兲. Similarly, reaction 共6a兲 indicates the produc-tion of H and stable C6H5O, whereas reaction共6b兲indicates that energetic C6H5Oⴱ further decomposes to yield C5H5 + CO. The products of reactions共5b兲and共6b兲 are identical, although they are dynamically produced from two distinct reaction paths. We have also performed electronic structure calculations to predict the energetics of the reaction interme-diates and transition states on the potential-energy surfaces 共PES兲 of the O+benzene reaction and used them to predict, with statistical rate calculations, the rate coefficients and branching ratios.

II. EXPERIMENTS

A. Crossed-molecular-beam experiments

As most features of the crossed-molecular-beam appara-tus have been described previously,47 only the relevant part of the experimental setup is described here. An atomic beam of O共1D兲 was generated upon laser photolysis at 157.6 nm of

a skimmed molecular beam of O2,48

O2+ h␯共157.6 nm兲 → O共1D兲 + O共3P兲. 共8兲 The output of a F2excimer laser共Lambda Physik, LPX 210i, F2 version, 30– 50 mJ pulse−1兲 was focused with a special spherical-cylindrical MgF2lens to a spot size of 3⫻3 mm2. Under such conditions, O共3_{P兲 atoms were generated also in} approximately equal proportions. The O共1_{D兲 atomic beam} had a narrow velocity distribution 共⬍2%兲 and an angular divergence of about ⫾4° 关full width at half maximum 共FWHM兲兴. The O共1_D/3_{P兲 atomic beam has a mean speed of} 2290 m s−1_.

Even though the reactivity of an O共3_{P兲 atom toward} benzene is expected to be smaller than that of O共1D兲,49,50the contribution from O共3P兲 was carefully examined. The output

of an ArF excimer laser 共Lambda Physik, LPX 210i, 20– 25 mJ pulse−1_{兲, focused with two cylindrical} fused-silica lenses to a spot size of 4⫻4 mm2_{, photodissociated a} molecular beam of SO₂ to generate a beam of O共3P兲 with

negligible O共1D兲,

SO₂+ h␯共193 nm兲 → SO共3⌺−_{兲 + O共}3_{P兲. 共9兲} Because diverse vibration-rotational states of SO photofrag-ments are populated upon photolysis of SO2,51 the atomic beam of O共3_{P兲 had a broad velocity distribution 共⬃33%,} FWHM兲 with an angular divergence of about ⫾6° 共FWHM兲. Since the velocity distribution of the O共3_{P兲 atom was} intrin-sically broad, the SO2beam was not skimmed to enhance the intensity. The mean speed of the O共3_{P兲 atomic beam was} about 2300 m s−1_{. To achieve similar conditions for} com-parison, the molecular beam of O2was also not skimmed in some experiments. S18_O

2 共97% 18O兲, synthesized on burn-ing sulfur powder under18O2gas, was employed to produce 18_O共3_P兲.

A molecular beam of benzene was generated on expand-ing a premixed sample共2% in Ne兲 through a pulsed Even-Lavie valve52 with its head heated to 488 K to diminish formation of clusters. This valve produces a benzene pulse with width of ⬃40 ␮s at the interaction region, hence di-minishing significantly the effusive background gases from the beam source. A sharp-edged skimmer共Beam Dynamics, diameter of 2 mm兲 served to define the angular divergence of about ⫾1.8°. Perdeuterated benzene 共C6D6, isotopic purity ⬎99.95%, ACROS兲 was used in D-isotopic experiments. The mean speed of the benzene beam was 1050 m s−1with a distribution of 5%.

The two reactant beams crossed each other at 90°; the collision energy was tuned to⬃10 kcal mol−1 upon adjust-ing the velocity of the O atomic beam. Products scattered from the reaction center traveled 24 cm before being de-tected with a time-resolved quadrupole mass filter. The hous-ing of the electron-impact ionizer of the mass filter is differ-entially pumped in three sections to 10−12 _{Torr so that the} background signal from residual gases and scattered gas is diminished. Velocity distributions of the product were de-rived from the time-of-flight 共TOF兲 spectra of the nascent products, recorded with a multichannel scaler共EG&G, Turbo MCS兲. The angular distribution of products was measured on rotating the detector. A computer program employs trial dis-tributions of translational energy P共ET兲 and angular

disper-sion P共␪兲 of products in the center-of-mass 共CM兲 frame to simulate the TOF spectra in the laboratory frame using for-ward convolution.53P共ET兲 and P共␪兲 were adjusted iteratively

until a satisfactory fit to the experimental TOF spectra and angular distribution was attained. Instrumental functions used in the program were determined from calibration ex-periments, including photolysis of O₂ at 157.6 nm and O共1_{D兲+Xe quenching/elastic scattering.}

B. Time-resolved IR emission experiments

The apparatus employed to obtain step-scan time-resolved Fourier-transform spectra 共TR-FTS兲 has been de-scribed previously;54–56 only a summary is given here. A telescope mildly focused the photolysis beam from a KrF laser 共248 nm兲 to an area of ⬃6⫻22 mm2 _{at the reaction} center to yield a fluence of ⬃50 mJ cm−2_{. Filters passing} either 1700– 2800 cm−1 _{共for the detection of CO兲 or} 2840– 4000 cm−1_{共for the detection of OH兲 were employed} to minimize the number of scan steps. The transient signal from an InSb detector with a rise time of 0.7 ␮s was pream-plified with a gain factor of 105 _{V A}−1 _{共EG&G Judson,} PA9–50, 1.5 MHz bandwidth兲, followed by further amplifi-cation with a factor of 500 共bandwidth of 1 MHz兲 before being digitized with an internal data-acquisition board 共16 bits兲 at a resolution of 5 ␮s. Data were typically averaged over 60 laser pulses at each scan step; 2508 or 4881 scan steps were performed to yield an interferogram resulting in a spectrum with resolution of 1.0 or 0.3 cm−1for OH and CO detection, respectively. To improve the signal to noise ratio 共S/N兲 of the spectrum, we averaged six sets of time-resolved spectra under the same experimental conditions to yield

sat-isfactory spectra. The temporal response function of the in-strument was determined with a pulsed IR laser beam, as described previously.57

Ozone 共O3兲 and C6H6 were injected into the reaction chamber separately; to decrease the collisional quenching of CO and OH, a minimal pressure yielding acceptable signals was used: PO₃= 0.072– 0.097 Torr and PC₆H₆= 0.020 − 0.092 Torr. Flow rates were FO₃= 1.7– 2.4 SCCM and

FC₆H₆= 0.4– 2.4 SCCM; SCCM denotes cubic centimeter per minute under standard conditions 共273 K and 760 Torr兲. A large fraction 共⬃60%兲 of O3 was dissociated upon irradia-tion at 248 nm based on the reported absorpirradia-tion cross secirradia-tion of 1.5⫻10−17 _cm2_molecule−1 _{for O}

3 at 248 nm.58 The depletion of O₃ after each laser pulse was modest, as was confirmed by the negligible variation in the signal when we decreased the repetition rate of the photolysis laser from 19 to 12 Hz.

C₆H₆共Fluka, ⱖ99.5%兲 was used without purification ex-cept for degassing at 77 K. O₃ was produced from O₂ 共Scott Specialty Gases, 99.995%兲 with an ozone generator 共Polymetrics, model T-408兲, stored over silica gel at 196 K, and eluted from the trap with a small flow of He 共Scott Specialty Gases, 99.999%兲. The partial pressure of O3 was determined from the absorption of Hg emission at 254 nm in a cell with length of 7.0 cm; the cross section for absorption of O3at 254 nm was taken to be 1.15⫻10−17 cm2.59 III. COMPUTATIONAL METHODS

The potential-energy diagram for the reaction system O共1_D兲+C6H

6 is extended from those established previously on the thermal decomposition43 and the photofrag-mentation44 of C6H5OH based on energies predicted at the highest level of the modified Gaussian-2 method, G2M共CC5兲.60

In the G2M calculation, the geometries of re-action intermediates and transition states on the ground elec-tronic surface of C6H5OH were optimized with the

GAUSS-IAN 03 program61 at the B3LYP/6-311G共d,p兲 level of

theory.62,63

Calculations of rate coefficients were performed with the

VARIFLEX program64 based on the microcanonical Rice–

Ramsperger–Kassel–Marcus共RRKM兲 theory and variational transition-state theory65–70 with corrections for Eckart tunneling71 and multiwell reflection of the reaction flux.72 The energy increment was fixed at 10 cm−1 _{in all} calcula-tions of sums of states and densities of states that were per-formed using the modified Beyer–Swinehart algorithm.73 The component rates were evaluated at the E/J-resolved level and the pressure dependence was treated with calcula-tions based on a one-dimensional master equation using the Boltzmann probability of the reaction complex for the

J-distribution. A simple exponential quenching model was

employed to calculate the coefficients of collision energy transfer.74 An average step size of 120 cm−1 _{for energy} transfer per collision 具⌬E典down was employed for the He buffer gas. The Lennard-Jones 共LJ兲 parameters of buffer gases 共␧/kB= 10.2 K and ␴= 2.56 Å for He兲 and complex

共␧/kB= 450 K and ␴= 4.50 Å, the same as C6H5OH兲 were taken from the literature.75 For a barrierless association 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:

or decomposition, a fitted Morse function, V共R兲

= De兵1−exp关−␤共R−Re兲兴其2, was used in conjunction with an

anisotropic potential function to represent the minimum-energy path共MEP兲 for variational calculations of rate coef-ficients.

IV. RESULTS

A. Distribution of product translational energy in crossed beams of O_„1D_…+C6H6

Because both O共1_{D兲 and O共}3_{P兲 atoms react with C6H} 6, one important issue in this experimental investigation is to distinguish their individual contribution. As the photolyzed O2 beam yields O共1D兲 and O共3P兲 atoms in equal propor-tions, we need another O atom source which provides a dif-ferent ratio of these two atomic states in order to sort out the individual contributions. Photolysis of SO2 at 193 nm pro-duces O atoms in its 3P state; the photon energy is

insuffi-cient to produce any O共1

D兲 atom within the one-photon

limit. With the photolyzed SO2 source, we found that the reaction O共3_P兲+C6H

6 produced C6H6O 关reaction 共2兲兴 and H + C6H5O 关reaction 共1兲兴 but no detectable CO, C5H6, or C5H5 products, consistent with the previous results of crossed-molecular beam using a discharge source.36The con-tribution of the O共1

D兲+C6H6 reaction can then be obtained by subtracting the O共3_{P兲 contribution obtained with the} pho-tolyzed SO2 beam from the results obtained with the photo-lyzed O2beam. We deliberately tuned the velocities of both O atom sources to be similar and quantified their relative intensities with the 70 eV electron-impact ionizer by assum-ing that the ionization cross sections of O共1

D兲 and O共3P兲 are

about equal.

Although reactions 共1兲 and共6a兲 yield the same C6H5O + H products, we found that the reactivity of O共1

D兲 is at least

five times that of O共3_{P兲 for the formation of C6H}

5O. Thefore, the error from this subtraction process is small for re-action共6a兲. Furthermore, the C6H6O product from reaction

共2兲is unique for having a zero recoil velocity and hence, can

be easily separated from the products of the O共1_D兲+C6H 6 reaction. The CO formation channels from the O共1D兲

+ C6H6 reaction is unaffected by the O共3P兲+C6H6 reaction due to its negligible contribution. In the following, we focus our discussions on the data which have been adequately corrected to represent the products from the O共1D兲+C6H₆ reaction.

As formation of CO+ C5H6 is expected from the O共1

D兲+C6H6 reaction,42–44 we first searched for a signal of C5H6. TOF spectra of mass 66 at three representative labo-ratory angles are shown in frames共a兲⫺共c兲 of Fig.1; in these spectra three components are observed 关Fig.1共c兲兴. The two slower components 共designated as ␤ and ␥兲 arise from the naturally abundant13C-isotopic C₅H₅+. As the signals at mass 65 共C5H₅+兲 are much larger than those at mass 66, this 13

C12C4H5 +

signal has intensity comparable to that of12C5H6 + . After subtraction of the contribution from the 13C isotope, only the most rapid component 共designated as ␣兲 remains, which we adopted to be due to channel共5a兲. To ensure the validity of the subtraction for the 13C-species, we also per-formed the experiment O共1D兲+C6D₆, in which there is no such isotopic contamination. The distribution of kinetic en-ergy at mass 72共C5D₆+兲 is nearly identical to that of the rapid component at mass 66 in the experiment of O共1_D兲+C6H

6. Frames 共d兲⫺共f兲 of Fig. 1 show the signals at mass 65 observed in the experiments of O共1_D兲+C6H

6. The signal might arise from three possible sources: 共1兲 reactions 共5b兲 and 共6b兲to produce CO+ C5H5+ H, 共2兲 the daughter ion of C5H6 that was produced from reaction 共5a兲, and 共3兲 the daughter ion of C6H5O that was produced from reaction共6a兲. Figure 2 shows the Newton diagram for the observed products of reactions共5a兲,共6a兲,共5b兲, and共6b兲. The Newton circle of C6H5O is expected to be small because the H atom coproduct carries away almost 99% of the total translational energy. The component ␥ at mass 65关Fig.1共f兲兴 has almost

identical TOF and angular distributions to those for the sig-nal at mass 93 共C6H5O+兲. A comparison of their angular distributions in the laboratory frame is shown in Fig.3. The

0.0 0.5 1.0 1.5 2.0 2.5 3.0

(d)

m/z=65

30

o



0.0 0.2 0.4 0.6



(a)

m/z=66

20

o 0 1 2 3 4 5

(e)

40

o 0.0 0.2 0.4 0.6 0.8



(b)

R

e

l.

Intens

ity

40

o 0 100 200 300 400 500 0 10 20 30

( f )

60

o

neutral time of flight /



s

0 100 200 300 400 0.0 0.5 1.0 1.5 2.0 2.5











(c)

60

o

FIG. 1. Representative TOF spectra in the

crossed-beam experiment with O共1_D兲+C

6H6at collision energy

of 10 kcal mol−1_{for mass 66关panels 共a兲–共c兲兴 and mass}

65关共d兲–共f兲兴 at selected laboratory angles. The ordinate

scales vary. Three components ␣, ␤, and ␥ are

dis-cussed in text.

similarity of angular distributions and their TOF spectra at masses 65 and 93 leads us to conclude that the␥component at mass 65 represents a daughter ion of C₆H₅O.

The second possible source of mass 65, the daughter ion of C₅H₆, can contribute only slightly to the signal because the TOF spectra of those two masses are disparate. Here only a minor portion of signals at mass 65 关designated as ␣ in Figs.1共d兲–1共f兲兴 is attributed to the daughter ion of C₅H₆. The relative contribution of component␣ is obtainable from the momentum-matched CO coproduct 共see discussion below兲. After considering two of the three possible sources for the observed C5H5+TOF spectra, the most reasonable assignment for the remaining ␤ component is a product from channels

共5b兲and共6b兲.

According to quantum-chemical calculations共discussed below兲, 1,3-cyclopentadiene is the most likely isomer for the C5H6 product. The ionization energy of this species is 8.57 eV and the threshold for its dissociative ionization to form C5H5++ H is 12.62 eV,

76

indicating that ion C5H6+ is quite stable; about 93 kcal mol−1is required to break a C–H bond of C5H6

+

. If 1,3-cyclopentadiene were produced, then there

would be no difficulty in observing its parent ion with the electron-impact detector.77 The fact that the proportion of C5H6+observed was much smaller than that of C5H5+suggests that channels共5b兲and共6b兲dominate over channel共5a兲.

Because the background signal of C16O was non-negligible, it was difficult to investigate CO product directly. Instead, we replaced the 16O共1D兲 source with 18O共1D兲 and

detected directly the C18O product at mass 30. As shown in Fig.4, two components,␣

⬘

and␤

⬘

, fit the TOF spectra of the C18O product. The signal ␥ was observable only at angles near the CM angle and is attributed to a daughter ion of C6H5O produced from reaction共6a兲. The more rapid compo-nent,␣

⬘

, is momentum-matched to C5H6共the␣component兲 and the slow one, ␤

⬘

, to C5H5+ H 共the ␤ component兲. The C18O data also indicate that channels共5b兲and共6b兲are domi-nant.

Figure5 shows the primary P共ET兲 used to fit reactions

共5a兲,共5b兲,共6b兲, and共6a兲. The P共ET兲 of channels 共5a兲,共5b兲,

and 共6b兲 fit satisfactorily data from both O共1D兲+C6H6 and O共1_D兲+C6D

6 experiments; whereas for reasons of reso-lution, the P共ET兲 of channel共6a兲was determined only from

the experiment of O共1_D兲+C6D

6. For the three-fragment channels 共5b兲 and 共6b兲, the momentum exerted by the H atom is negligible because of its small mass; we can thus analyze only the momentum-matching condition for C5H5 and CO. Practically high background at m/z=1 makes detec-tion of the H product unattainable. Nevertheless, we can see from Fig.10that there is no reverse barrier for these H atom loss processes. Therefore, the kinetic energy of the H atom product is expected to be small. The momentum of the H atom product would be relatively minor in comparison with those of the C5H5and CO products. In the analysis, we only used the two masses of C5H5and CO. That is, the presented

P共ET兲 for channels共5b兲 and共6b兲 includes only the

transla-tional energies of the C5H5and CO products, with the small amount of translational energy of the H atom excluded. The

FIG. 2. Newton diagram for the crossed-beam experiment with O共1D兲

+ C6H6at collision energy of 10 kcal mol−1. Three representative Newton

circles are for C5H6product of channel共5a兲at ET,peak= 28 kcal mol−1, C5H5

product of channels共5b兲and共6b兲at ET,peak= 6 kcal mol−1, and C6H5O

prod-uct of channel共6a兲at E_T,peak= 12 kcal mol−1_.

0 10 20 30 40 50 60 70 80 90 0.0 0.2 0.4 0.6 0.8 1.0 1.2
component of C5H5 + C6H5O + Normalized Intensity

LAB angle / deg.

FIG. 3. Angular distributions of C6H5O+and the␥component of C5H5

+_关Fig.

1共f兲兴 in the laboratory frame.

0.0 0.5 1.0 1.5 R e l. Intens ity

neutral time of flight /
s

(a)

m/z=30

20

o 0.0 0.5 1.0 1.5 2.0



'



'

₄₅

o

(b)

0 100 200 300 400 500 0.0 0.5 1.0 1.5 2.0 2.5



'



'





'



'

(c)

67

o

FIG. 4. Typical TOF spectra at mass 30共C18_O+_{products兲 in the}

crossed-beam experiment of18_O共1_D兲+C

6H6at collision energy of 10 kcal mol−1.

Three components␣⬘,␤⬘, and␥are discussed in text.

P共␪兲 for these channels are either isotropic or

forward-backward symmetric, indicating a reaction mechanism asso-ciated with an long-lived complex.

It is difficult to investigate the channel for formation of OH from O共1_D兲+C6H

6 because the background signal from residual H2O gas is large; hence, we investigated this chan-nel in the reaction of 18O共1D兲+C6D6. A weak 18OD signal was detected within a limited range of laboratory angles. The TOF spectrum and P共ET兲 are shown in Fig. 6. In these

ex-periments, a slightly different collision energy was used to increase the intensity of the O共1_{D兲 beam. As P共␪兲 cannot be} accurately determined from the limited data, we assumed an isotropic P共␪兲 in the preliminary analysis.

B. Infrared emission of CO from the flow experiments To maintain a feasible condition as nearly collisionless as practicable, we decreased the partial pressure of O3 and

C6H6 while maintaining a satisfactory signal to noise ratio. This ratio for the OH bands is superior to that for CO be-cause a decreased spectral resolution was required for OH and the Einstein coefficients of OH are, in general, greater. Satisfactory spectra of CO were hence obtained on averaging six spectra that were recorded in separate experiments under similar conditions.

We assigned lines of CO based on spectral parameters reported by Ogilvie et al.78and employed values of Einstein

A coefficients of CO calculated previously.79,80The spectrum exhibits emission of CO with J

⬘

up to 30 and ␷

⬘

up to 6. Each vibration-rotational line was normalized with the rela-tive instrument response factors and divided by its respecrela-tive Einstein coefficient to yield a relative population P_␷共J

⬘

兲. Partially overlapped lines of CO, such as J

⬘

= 11, 22, 28 of

␷

⬘

= 1, J

⬘

= 7 , 12, 22, 25 of␷

⬘

= 2, J

⬘

= 3 , 8 , 14, 21 of␷

⬘

= 3, and

J

⬘

= 2 , 14, 20 of␷= 4, were deconvoluted to yield their inten-sities.

Semilogarithmic plots of P_␷共J

⬘

兲/共2J

⬘

+ 1兲 versus

J

⬘

共J

⬘

+ 1兲 for CO 共␷

⬘

= 1 – 6兲 recorded 0–5 ␮s upon photoly-sis of O3 appear in Fig.7. Fitted Boltzmann-type rotational distributions of CO, derived from the spectrum recorded in the range of 0 – 5 ␮s, yielded rotational temperatures of 1480⫾140, 920⫾100, 860⫾60, 850⫾80, 810⫾90, and 700⫾110 K for ␷

⬘

= 1 – 6, respectively; unless spec-ified otherwise, listed error limits represent one standard deviation in fitting. An average rotational energy of

Er= 1.9⫾0.3 kcal mol−1 was observed for CO 共␷= 1 – 6兲. In

our previous work on O共1_D兲+CO,79

we observed that rotational quenching of CO is non-negligible under our ex-perimental conditions 共PCO= 0.058 and PO3= 0.016 Torr兲 even at 5 ␮s; hence, we fitted the rotational temperature of CO at varied periods upon photolysis to an exponential de-cay and estimated the nascent rotational temperature to be 0 10 20 30 40 50 60 70 80

(c)

5b+6b

5b+6b

(b)

kinetic energy / kcal mol-1

(a)

5a

kinetic energy / kcal mol-1

P(E

T

)

0 30 60 90 120 150 180 CM angle / deg.

P(

)

0 10 20 30 40 50 60 70 80

6a

FIG. 5.共a兲 Kinetic-energy distribution P共ET兲 used to fit channels共5a兲and

共5b兲⫹共6b兲; 共b兲 P共ET兲 used to fit channel 共6a兲; and 共c兲 P共␪兲 used to fit

channels共5b兲and共6b兲. Boundaries of the shaded area indicate limits of the

distributions which still give acceptable fits to the experimental data. An

isotropic angular distribution was used to fit channels共5a兲and共6a兲.

0 10 20 30 40 50 60

(b)

R el .I nten si ty P(E T )

kinetic energy / kcal mol-1

0 25 50 75 100 125

(a)

neutral time of flight /
s

FIG. 6. 共a兲 TOF spectra and 共b兲 the corresponding P共ET兲 of OD product in

crossed-beam experiments with O共1D兲+C6D6 at collision energy of

12 kcal mol−1_. 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 0 200 400 600 800 1000 1200 1400 1600 0 1 2 3 0 1 2 3 v = 1 T_r= 1480 ± 140 K v = 2 T_r= 920 ± 100 K ln[P v /( 2J+1 )] v = 3 T_r= 860 ± 60 K v = 5 T_r= 810 ± 90 K J(J+1) v = 6 T_r= 700 ± 110 K v = 4 T_r= 850 ± 80 K

FIG. 7. Semilogarithmic plots of relative rotational populations of CO共␷

= 1 – 6, circle兲 upon photolysis of a flowing mixture of O3共0.097 Torr兲 and

C6H6共0.020 Torr兲 at 248 nm. Solid lines represent least-squares fits.

1890⫾120, 1180⫾110, 1090⫾90, 1040⫾70, and 1000⫾80 K for CO 共␷= 1兲 to CO 共␷= 5兲, respectively. After applying a correction factor of 1.26 for rotational quenching based on decay in the rotational temperature, we estimated a nascent rotational energy of 2.4⫾0.4 kcal mol−1 based on the observed data.

We assumed a Boltzmann distribution and associated an interpolated population with overlapped lines. The relative populations obtained on counting levels up to the observed

J_max in each vibrational level were normalized to yield a relative vibrational population 共␷= 1兲:共␷= 2兲:共␷= 3兲: 共␷= 4兲:共␷= 5兲:共␷= 6兲=37.6:29.5:15.1:8.8:5.6:3.0, corre-sponding to a vibrational temperature of 5800⫾330 K. Assuming a Boltzmann distribution, we estimated the population of ␷= 0 to be 2.2⫾0.2 times that of ␷= 1. The vibrational distribution of CO normalized for ␷= 0 − 6 is thus 共␷= 0兲:共␷= 1兲:共␷= 2兲:共␷= 3兲:共␷= 4兲:共␷= 5兲:共␷= 6兲 = 45.2: 20.7: 17.0: 8.2: 4.7: 2.9: 1.3, as shown in Table ⌱ and Fig.8共a兲. The average vibrational energy of CO thus derived is E_␷= 8.0⫾0.7 kcal mol−1_{. Vibrational quenching is} negli-gible within 5 ␮s; the correction is less than 3%.

C. Infrared emission of OH from the flow experiments Emission spectra of OH, at a resolution of 1.0 cm−1_, were recorded 0 – 5 ␮s after photolysis of O3 共0.097 Torr兲 and C6H6共0.020 Torr兲. Assignments were based on spectral parameters reported by Colin et al.81The spectrum exhibits emission from OH with values of K

⬘

up to 9 and␷

⬘

up to 3. Each vibration-rotational line in the P branch was analyzed to yield a relative population P_␷共K

⬘

兲, using Einstein coeffi-cients reported by Holtzclaw et al.82 Semilogarithmic plots of P_␷共K

⬘

兲/共2K

⬘

+ 1兲 versus K

⬘

共K

⬘

+ 1兲 for OH 共␷= 1 – 3兲 re-corded 0 – 5 ␮s after photolysis of O3 are shown in Fig. 9. There is a negligible variation in the population of OH for the two spin-orbit components. Fitted rotational distributions of Boltzmann type for the P1 and P2 branches of OH 共␷

⬘

= 1 – 3兲 yield rotational temperatures of 660⫾20, 570⫾20 and 690⫾40 K, as listed in Table I. An average rotational energy of Er= 1.5⫾0.2 kcal mol−1 for OH

共␷= 1 – 3兲 observed 0–5 ␮s after photolysis is derived. Based on the derived rotational temperatures of OH

共␷= 1 – 2兲 as a function of reaction periods, we estimated the average nascent rotational temperatures to be 680⫾10 and 610⫾10 K for OH 共␷= 1兲 and OH 共␷= 2兲, respectively. The nascent average rotational energy of OH is thus

Er= 1.6⫾0.3 kcal mol−1.

The relative vibrational population of OH was derived to be 共␷= 1兲:共␷= 2兲:共␷= 3兲=60.6:30.9:8.5, corresponding to a vibrational temperature of 4830⫾230 K. Assuming a Boltzmann distribution, we estimated the population of

␷= 0 to be 3.2⫾0.3 times that of␷= 1. The vibrational dis-tribution of OH normalized for ␷= 0 – 3 is thus 共␷= 0兲: 共␷= 1兲:共␷= 2兲:共␷= 3兲=66.1:20.6:10.5:2.9, as shown in Table⌱ and Fig.8共b兲. The average vibrational energy of OH thus derived is E_␷= 5.0⫾1.0 kcal mol−1.

D. Branching ratios and their D-isotopic effect

We searched for the HCO signal in the crossed-beam experiments but detected no signal at mass 31 共HC18O+_兲. Because the background at this mass is small and the HCO+ ion is stable, any neutral HCO product, if formed, is hence detectable as HCO+_{; we concluded that the channel yielding} HCO is negligible.

We assumed similar detection efficiencies of C5H6 and C5H5 in the electron-impact ionization/detection and deter-mined the branching ratio of channels 共5a兲/关共5b兲⫹共6b兲兴 by analyzing the TOF spectra of C5H6+and C5H5+, together with their daughter ions down to C₅+. Daughter ions smaller than C₅+ are expected to have negligible contributions as the C5-ring is quite stable. The Jacobian factor in the transfor-mation from a laboratory frame to the CM frame is included in the analysis. For O共1

D兲+C6H6 and O共1D兲+C6D6 reac-tions, the branching ratios of channels共5a兲/关共5b兲⫹共6b兲兴 were determined to be 0.12⫾0.03 and 0.13⫾0.02, respectively. A similar procedure is applicable to channel共6a兲, for which the branching ratio of channels共6a兲/关共5b兲⫹共6b兲兴 was deduced to be 0.38⫾0.06 for the O共1

D兲+C6D6reaction. The large mass 10 100 0 1 2 3 4 5 6 10 100

Relative Population /% (a) CO (b) OH

FIG. 8. Relative vibrational distributions of CO 关panel 共a兲兴 and OH

关panel 共b兲兴 upon photolysis of a flowing mixture of O3共0.097 Torr兲 and

C6H6共0.020 Torr兲 at 248 nm. The populations of␷= 0 are estimated from

Boltzmann distributions. 1 2 3 4 5 1 2 3 4 5 0 20 40 60 80 100 1 2 3 4 5 P1 (
= 1) P2 (
= 1) K(K+1) ln[ Pv /( 2K +1 )] P1 (
= 2) P2 (
= 2) P1 (
= 3) P2 (
= 3)

FIG. 9. Semilogarithmic plots of relative rotational populations of OH

共␷= 1 – 3; P1 branch, circle; P2 branch, open triangle兲 upon photolysis of a

flowing mixture of O3共0.097 Torr兲 and C6H6共0.020 Torr兲 at 248 nm. Solid

lines represent least-squares fits.

ratio of the products limits the resolution of P共ET兲 for

chan-nels 共6a兲 and 共6b兲; the error bar of this branching ratio is consequently larger.

Only a weak 18OD signal was detected from 18O共1D兲

+ C6D6 within a limited range of laboratory angles. As no OD signal was detected from the reaction with the O共3

P兲

source from SO2 photolysis, we can exclude the possibility that the 18OD signal arises from the reaction of18O共3P兲; we

conclude that the formation of18OD is minor, with a branch-ing ratio of⬍0.1.

The relative branching ratios of the channels to form CO

and OH were also determined with TR-FTS on summing the total populations of the products, including estimates of

␷= 0. Care was taken to correct the small background signal of OH measured when no C6H6was added; the signal might result from reactions of O共1D兲 with background H2O and trace hydrocarbons in the system. After correction for this interference, the ratio of 关CO兴/关OH兴 was derived to be 2.1⫾0.2.

In the O共1D兲+C6D₆experiment, IR emission of CO but not OD was detected. The signal to noise ratio of CO emission in this experiment was ⬃4.2. Although accurate

-47.0 -107.4 C5H4CO + H2 -62.7 -77.9 C6H4+ H2O -133.0 -95.1 -105.0 -60.0 C6H5O + H -63.0 -126.9 -89.8 -94.8 -73.3 -89.0 TS1 C6H5OH TS4 TS5 TS6 TS8 TS9 -65.1 -42.7 TS10 -134.4 M4 M2 M3 M1 TS11 TS12 M5 M6 M7 TS13 TS14 TS15 -65.7 -106.8 -100.2 -90.2 -101.4 -69.6 -66.7 -36.8 C5H6+CO C6H6+ O(1D) -104.5 -149.1 TS2 -36.8 TS16 M9 -64.7 -44.4 TS17 M8 TS7 -37.5 C6H5+ OH -80.4TS3 (-0.2) TS18 (-2.7) M10 (-36.6) C6H5+ OH C6H6+ O(1D) (0.0) C5H5+H+CO -43.6 TS19 -9.1 0.0 M1 (C_s) 1.218 1.470 1.532 1.350 1.456 1.341 1.497 M2 1.159 1.322 1.457 1.348 1.463 1.337 M3 (Cs) 1.238 1.465 1.436 1.465 1.468 1.370 1.403 M4 (C2V) 1.221 1.480 1.338 1.496 M5 (Cs) 1.432 1.518 1.467 1.350 1.450 M6 (Cs) 1.390 1.337 1.455 1.350 M7 1.195 1.469 1.479 1.594 1.497 1.534 1.514 1.333 M10 (Cs) 1.367 0.963 1.396 1.397 1.388 1.393 1.396 1.393 3.356 1.076 1.385 1.392 1.392 1.392 1.384 1.393 C6H5OH (Cs)

FIG. 10. Energy profile for C6H6+ O共1D兲 at the G2M/ /B3LYP/6-311G共d,p兲 level for the association/isomerization/decomposition channels and at the

CIPT2/ /CAS共8,8兲/6-311G共d,p兲 level for the H-abstraction channel 共separate line on top with ordinates shifted upward slightly and energy listed in

parentheses兲. Energies are in kcal mol−1_{. Structures of some key intermediates are also shown. Bond lengths are in Å.}

Einstein coefficients of OD are unavailable, we estimated these based on the theory that they are proportional to the cube of frequency and to the square of the matrix element83 Assuming roughly

冑

2 as the ratio of reduced masses of OD and OH, the matrix elements for⌬␷= 1 transitions of OH and OD have a ratio of 1/

冑

2, and the frequency factor ratio is approximately 共1/

冑

2兲3_{. The maximal Einstein coefficient of} OD is hence ⬃18% those of OH, which are, in turn, ⬃7% those of CO. The signal to noise ratio 共S/N兲 for OH in our experiments with O共1

D兲+C6H6was about 15; hence, we ex-pect that, under similar conditions, this ratio for OD in the reaction O共1

D兲+C6D6to be about 2.7 if the same amount of OD were produced. The fact that we observed no detectable emission of OD共S/N⬍2兲 indicates that there is a substantial deuterium isotopic effect with 关OH兴/关OD兴⬎1.4, that is, 关CO兴/关OD兴⬎2.9 in the reaction of O共1

D兲+C6D6.

E. PES for the reaction

The predicted potential-energy diagram and some impor-tant intermediate structures are presented in Fig.10; all sym-bols given in the figure are identical to those given previ-ously for the decomposition of phenol.43,44 New product channels characterized by M9, TS16, TS17, M10, and TS18 are added into this figure. The uncertainties of the calculated enthalpy of reaction are estimated to be about ⫾3 kcal mol−1_{when we compare the calculated values with} known experimental results. As the mechanism for the uni-molecular decomposition of C6H5OH, a key intermediate in the O共1_D兲+C6H

6 reaction, has been previously discussed in detail,43,44 our discussion of the present system focuses on the initial bimolecular processes and the isomerization of the excited intermediate to C6H5OH.

The geometric parameters of various new intermediates and transition states are available from the Electronic Phys-ics Auxiliary Publication Service共EPAPS兲.84The reaction of O共1_{D兲 with C6H}

6 can occur along two distinct paths—an addition to one C = C bond to give benzene oxide共M5兲 and a direct H abstraction 共via TS18兲 to give OH and C6H5. Our repeated searches for the C⫺H insertion product C6H5OH always resulted in the ring-addition intermediate M5. This addition reaction to produce M5 occurs with no barrier and

is exothermic by 106.8 kcal mol−1_{. The barrierless path} of minimum energy 共MEP兲, calculated with the B3LYP/6-311G共d,p兲 method by manually stretching the length of bond O⫺C from 1.4 Å at M5 to 4.6 Å, corresponding to a structure asymptotic to the reactants, is presented well by the Morse function, V共RO–C兲=108.8 兵1−exp关−1.949共RO–C− 1.400兲兴其2 _{kcal mol}−1_{, in which R}

O–C in unit of angstroms denotes one of the two stretching O¯C6H6 isosceles bonds. The MEP was employed to evaluate the branching ratio for production of CO relative to OH, to be discussed later.

Benzene oxide 共M5兲 can isomerize to 2,4-cyclohexadienone 共M1兲 or oxepin 共M6兲 via transition states TS11 共−65.7 kcal mol−1 _{relative to the reactants兲 or TS12} 共−100.2 kcal mol−1_{兲, respectively. The latter} ring-enlargement isomerization with a small barrier of 6.6 kcal mol−1 _{apparently occurs more readily. The former} concerted isomerization reaction via TS11 has a large barrier because it involves both C⫺O bond breaking and H migra-tion. Further isomerization from M6 to M9 is, however, more difficult than those from M1 to M2, M3, and C6H5OH because the energy of TS16 共−36.8 kcal mol−1_{兲 is much} greater than those of TS4 共−95.1 kcal mol−1_{兲, TS6} 共−89.8 kcal mol−1_{兲, and TS3 共−80.4 kcal mol}−1_{兲. The} domi-nant reaction channels are hence predicted to be the isomerization/decomposition paths through M1, as in the case of the thermal and photolytic decomposition reactions studied previously;43,44the reaction channel from M6 to M9 via TS16 is expected to be kinetically noncompetitive in the reaction O共1_D兲+C6H

6.

As discussed previously,43,44M1 can decompose to form C₅H₆+ CO via two channels: one through M2 and TS5 and the other through M3, TS13, M7, and TS14. Although the barrier for M1 to form M2 via TS4 has an energy of 5.3 kcal mol−1less than that for M1 to form M3 via TS6, the latter channel is predicted to be dominant because, in the product outlet, the energy of transition state TS5 is 10.3 kcal mol−1 _{greater than that of TS14. The transition} state TS3 for isomerization of M1 to phenol 共C6H5OH兲 has energy greater than those of TS6 and TS4 by 9.4 and 14.7 kcal mol−1, respectively. Furthermore, transition states TS2 and TS1 for the decomposition channels of C6H5OH also have energies greater than those of transition states

TABLE I. Fitted rotational temperature Trot, average rotational energy Erot, and vibrational population of CO共␷兲

and OH共␷兲 recorded 0–5 ␮s upon irradiation of a flowing mixture of O3共0.097 Torr兲 and C6H6共0.02 Torr兲 at

248 nm.

␷

CO OH

Trot共K兲 Erot共kcal兲 Population Trot共K兲 Erot共kcal兲 Population

0 共0.452兲a 共0.661兲a 1 1480⫾140 2.63 1.000共0.207兲 660⫾20b 0.98 1.000共0.206兲 2 920⫾100 1.63 0.785共0.170兲 570⫾20 0.79 0.509共0.105兲 3 860⫾60 1.43 0.402共0.082兲 690⫾40 0.88 0.138共0.029兲 4 850⫾80 1.17 0.234共0.047兲 5 810⫾90 1.10 0.160共0.029兲 6 700⫾110 0.69 0.080共0.013兲

a_{Normalized population with␷= 0 predicted from Boltzmann distribution.}

b_{Average rotational temperature for P1 and P2 branches.}

TS14 and TS5. In addition to these channels, M1 might also decompose to form C6H5O + H via a loose variational transi-tion state with the dissociatransi-tion energy of 73.0 kcal mol−1_. Similarly, C6H5OH might decompose to form C6H5O + H and, to a lesser extent, C6H5+ OH, via loose transition states. Nevertheless, to estimate the rate coefficients for production of H and OH, we computed the MEP of both M1→C6H5O + H and C₆H₅OH→C₆H₅+ OH at the B3LYP/6-311G共d,p兲 level of theory; they are described with Morse functions

V共RC–H兲=80.4兵1−exp关−4.68共RC–H− 1.35兲兴其2 _{kcal mol}−1 _and

V共RC–O兲=117.1兵1−exp关−2.413共RC–O− 1.467兲兴其2 _{kcal mol}−1_, respectively. Further decomposition of C₆H₅O to C₅H₅ and CO involves a large potential barrier of 50.9 kcal mol−1 _via a three-step isomerization, which was discussed in detail by Liu et al.85The elimination of H from C5H6involves a loose transition state that is endothermic by 83.3 kcal mol−1_{, as} shown in Fig.10. Because the fragments C5H6+ CO+ H have less energy than the reactants by 43.6 kcal mol−1_{, this} composition is expected to occur more readily than the de-composition of C6H5O.

The direct H abstraction occurs via transition state TS18 to form a hydrogen-bonded complex, O¯HC6H5 共M10兲. These structures were optimized with various methods, BH&HLYP, MP2, and CAS共8,8兲 共Ref. 86兲 using the

6-311G共d,p兲 basis set. The B3LYP method failed to locate either M10 or TS18, which is predicted to have Cssymmetry

with an imaginary wavenumber of 730i cm−1_{at BH&HLYP,} 1989i cm−1at MP2, and 4789i cm−1at CAS共8,8兲. In TS18, the breaking C⫺H bond is predicted to have lengths of 1.18, 1.23, and 1.21 Å, and the evolving O⫺H bond to have lengths of 1.22, 1.24, and 1.28 Å, respectively, at the BH&HLYP, MP2, and CAS共8,8兲 levels. In M10, the O¯HC6H₅ bond is predicted to have a length of 2.80 Å at the BH&HLYP level and 3.06 Å at the CAS共8,8兲 level. These critical geometric parameters are similar to those of the hydrogen-abstraction channel in the reaction of O共1

D兲+C2H6 reported by Sun et al.28 The energies of M10 and TS18 have been refined by the CIPT2 method87 based on the geometric parameters optimized at the CAS共8,8兲/6-311G共d,p兲 level. To evaluate the energy rela-tive to the reactants, we calculated the energy of the super-molecule 共O¯HC6H5兲 with a separation of 25 Å between the two fragments, C6H6 and O共1D兲. The results show that M10 and TS18 lie at −2.7 and −0.2 kcal mol−1_{, respectively,} relative to the reactants at the CIPT2 level. At the same level of theory, another supermolecule 共OH¯C6H5兲 with the separation of 25 Å between two radicals C6H5 and OH, ap-proximately considered to be the C₆H₅+ OH products, was calculated to lie at −36.6 kcal mol−1, 0.9 kcal mol−1 less than that calculated with the G2M method. These results were employed for calculations of the relative product yields of the reaction of O共1

D兲+C6H6. Reaction共7兲is insufficiently exothermic for secondary dissociation to occur; as predicted at the G2M level, C₆H₅→H+C₆H₄ requires an activation energy of 77 kcal mol−1.

V. DISCUSSION

A. Reaction channels and thermal rate coefficients For the channel to form CO from the thermal decompo-sition of C6H5OH, the rate coefficients determined with a flow tube88 and a shock tube89 are quantitatively accounted for with the following paths:43

C6H5OH C5H6+ CO M3 M1 M7 M2 Path A Path B 共10兲

The dynamics of formation of CO in the photofragmentation of C6H5OH at 248 or 193 nm is consistent with this mechanism.44 In the photofragmentation of C6H5OH, addi-tional product channels producing C6H5O + H and H2O 共at 193 nm only兲 were detected.

The reaction of O共1

D兲+C6H6 is expected to occur pri-marily via routes of two types:共1兲 a low-energy path via M1, followed by isomerization/decomposition to produce C5H6 + CO, C6H5O + H, and C6H5+ OH, and 共2兲 an abstraction path via M10 to produce C6H5+ OH.

O共1 D兲 + C6H6↔ M5ⴱ↔ M1ⴱ→ C6H5O + H 共11a兲 ↔C6H5OHⴱ→ C6H5+ OH 共11b兲 ↔M2ⴱ_{→ C} 5H6+ CO 共11c兲 ↔M3ⴱ_{↔ M7}ⴱ_{→ C} 5H6+ CO 共11d兲 C6H6+ O共1D兲 ↔ M10 → C6H5+ OH. 共12兲 As discussed in Sec. IV E, the major channel for production of OH is the direct abstraction关reaction共12兲兴, rather than the addition/decomposition in reaction 共11b兲. For the formation of CO via channels 共11c兲 and 共11d兲, the latter is dominant because the energy of TS14 is predicted to be less than that of TS5 by 9.7 kcal mol−1_{; at 300 K, k}

共11d兲is about five to six times k_共11c兲, whereas at 1400 K, the ratio decreases to 2.

The rate coefficients for various channels under low-pressure conditions were calculated with the RRKM theory and variational transition-state theory with the VARIFLEX code of Klippenstein et al.64The pressure dependence of the individual and total thermal rate coefficients for the forma-tion of CO, H, and OH at 300 K over diverse pressures using He as a bath gas are shown in Fig.11. The total rate coeffi-cient is independent of pressure below 1000 atm. The indi-vidual rate coefficients k共CO兲, k共H兲, and k共OH兲, in which the reaction product is shown in parentheses, remain nearly con-stant under pressures less than ⬃600 Torr; experimental conditions in this work are thus at the low-pressure limit.

Figure 12共a兲 shows the temperature dependence of the individual and total thermal rate coefficients at the low-pressure limit in the temperature range of 200–2000 K. The total rate coefficient exhibits a small positive temperature dependence and is expressed as

k共total兲 = 2.42 ⫻ 10−9_T0.1.08_{exp共− 23/T兲}

cm3molecule−1s−1. 共13兲

The total thermal rate coefficient is mainly contributed by those for formation of CO and H,

k共CO兲 = 4.37 ⫻ 10−9_T−0.087_{exp共19/T兲} cm3molecule−1s−1 共T = 200 – 500 K兲, 共14a兲 k共CO兲 = 5.67 ⫻ 10−12T0.924exp共− 504/T兲 cm3molecule−1s−1 共T = 500 – 2000 K兲, 共14b兲 k共H兲 = 3.76 ⫻ 10−8T−0.383exp共12/T兲 cm3molecule−1s−1, 共15兲

respectively. The rate coefficient for formation of k共OH兲, in-cluding both decomposition关reaction 共11b兲兴 and abstraction 关reaction共12兲兴 channels,

k共OH兲 = 5.78 ⫻ 10−6T−1.47exp共163/T兲

cm3molecule−1s−1 共16兲

is about one-tenth that of k共CO兲 and k共H兲. k共M1兲 makes a negligible contribution to the total rate coefficient under our experimental conditions. Individual branching ratios are given in Fig.12共b兲. As the temperature increases, the yield of CO decreases whereas that of H increases; the branching ratio of OH remains small even at high temperatures. B. Channels to produce CO

As multistep isomerization is required to form CO, a protracted duration and an approximately statistical distribu-tion of energy are expected, consistent with the experimental observations. CO was observed in the crossed-molecular-beam experiment to be a major product, but the expected

counterproduct C5H6 tends to decompose further to C5H5 + H 关reaction 共5b兲兴. Similarly, some C6H5O produced with internal energy exceeding the dissociation barrier共for TS 19兲 might decompose further to C5H5 and CO 关reaction 共6b兲兴. The three-fragment channel, C5H5+ CO+ H, is exothermic by 43.6 kcal mol−1_{, but there is a barrier for the secondary} pro-cess C6H5Oⴱ→C5H5+ CO. For channel共6a兲, the primary H product carries no internal energy and the primary transla-tional energy release is expected to be small because there is no reverse barrier for the formation of C6H5O + H. At this point, it is difficult to estimate the branching ratio between reactions共5b兲and共6b兲from these experiments.

In TR-FTS experiments, nascent rotational temperatures of 1890⫾120, 1180⫾110, 1090⫾90, 1040⫾70, and 1000⫾80 K for CO 共␷= 1兲 to CO 共␷= 5兲 were derived; the rotational temperature of CO 共␷= 1兲 appears to be greater than the rotational temperatures of CO共␷= 2 – 5兲. One possi-bility is that CO produced from reaction 共6b兲 is populated only at␷= 0 and 1 because of the smaller exothermicity and large barrier; the greater rotational temperature for CO pro-duced in this channel might be related to the geometry of the transition state. The torque angle of TS19 共⬃56°兲 is much larger than that of TS14共⬃13°兲; it is therefore expected that the CO produced from channel 共6b兲 would have a greater rotational excitation than that from channels 共5a兲 and 共5b兲,

0 1 2 3 4 5 -12 -11 -10 -9 -8 (a) k_{T otal} k(O H) k(CO ) k(H) k(M1) Log (k /c m 3 molecule -1 s -1 ) 1000 T-1 / K-1 0 500 1000 1500 2000 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 (b)
(com plex)
(O H)
(CO)
(H) Branching ratio T/ K

FIG. 12. 共a兲 Total and individual thermal rate coefficients for the reaction

O共1_D兲+C

6H6as a function of temperature at the low-pressure limit.共b兲

Individual branching ratio as a function of temperature at the low-pressure limit. -6 -4 -2 0 2 4 6 -12 -11 -10 -9 -8 k(OH) k(H) k(CO) k_total k(M1) Log (k /c m 3 molecule -1 s -1 ) Log (P / Torr)

FIG. 11. Total and individual thermal rate coefficients for the reaction

O共1_D兲+C

6H6as a function of pressure at 300 K.

whereas that from channels共5a兲and共5b兲would have a less rotational excitation but a greater vibrational excitation.

In the photolysis of phenol at 193 nm, the available energy of 148 kcal mol−1 _{above C}

6H5OH is near that of the reaction O共1_D兲+C6H

6, 149+ 10= 159 kcal mol−1; the kinetic energy of O共1_{D兲, produced upon photolysis of O3at} 248 nm, with respect to the CM for O共1_D兲+C6H

6 was re-ported to be 10.2 kcal mol−1_.90

Previous experiments on photolysis of phenol at 193 nm indicated a nascent rotational temperature of⬃4600 K for CO 共␷= 1 – 4兲 and an observed vibrational distribution of 共␷= 1兲:共␷= 2兲:共␷= 3兲:共␷= 4兲 = 64.3: 22.2: 9.1: 4.4 corresponding to a vibrational tempera-ture of 3350⫾20 K; an average rotational energy of 6.9⫾0.7 kcal mol1 _{and vibrational energy of} 3.8⫾0.7 kcal mol−1 _{were derived.}44

In the present work on O共1_D兲+C6H

6, the CO product shows less rotational excita-tion共2.4⫾0.4 kcal mol−1_{兲 but greater vibrational excitation} 共8.0⫾0.7 kcal mol−1_{兲. A separate phase space might be} sampled for these two photolytic and bimolecular processes, even though they might be expected to proceed across the same PES.

C. Channels to produce OH

Most OH is produced via a direct abstraction; the behav-ior from a single channel is expected. In TR-FTS experi-ments, nascent rotational temperatures of OH were deter-mined to be 680⫾10 and 610⫾10 K for OH 共␷= 1 – 2兲, respectively.

The O⫺H bond length in transition state TS18 is 1.279 Å, and the angle C⫺H⫺O is almost linear. The small torque angle in the transition state implies little rotational excitation of OH, consistent with our observation of average rotational energy of only 1.6⫾0.3 kcal mol−1_{. The O⫺H} bond length in TS18, which is much elongated relative to the

value of 0.97 Å for diatomic OH, indicates that the OH product might be highly vibrationally excited, consistent with our observation of population of OH up to␷= 3 with a vibrational energy of⬃29 kcal mol−1_{. The observed average} vibrational energy of OH, 5.0⫾1.0 kcal mol−1_{, is 10.7% of} the total available energy. This fraction is large if one con-siders the complexity of the counterproduct C6H5.

D. Branching ratios and D-isotopic effect

The microcanonical rate coefficients k and the branching ratios ␣ for the formation of various products are shown in

Fig. 13共a兲, with kH共E兲⬎kCO共E兲⬎kOH-共12兲共E兲⬎kOH-共11b兲共E兲;

the latter two correspond to the formation of OH via reac-tions共11b兲and共12兲, respectively. These rate coefficients take no account of secondary dissociation channels. Reactions

共11c兲and共11d兲hence correspond to reactions共5a兲and共5b兲, reaction 共11a兲 corresponds to reactions 共6a兲 and 共6b兲, whereas reactions共11b兲and共12兲correspond to reaction共7兲.

As the excitation energy of the O共1

D兲+C6H6 reaction increases from 5 to 40 kcal mol−1, the branching ratio␣共H兲 increases slightly from 0.54 to 0.60, ␣共OH兲 increases from 0.07 to 0.19, whereas␣共CO兲 decreases from 0.39 to 0.21, as indicated in Fig. 13共b兲. The reaction system O共1D兲+C6D6 shows a substantial deuterium kinetic isotopic effect; kCO共E兲 and kOH共E兲 increase whereas kH共E兲 becomes smaller upon deuteration of C₆H₆.

In the beam experiments, the observed branching ratio for production of OD from O共1_D兲+C6D

6 is less than 0.1, consistent with the predicted branching ratio 0.04 at

E = 10 kcal mol−1_{关Fig.13共b兲兴. Because of secondary} decom-positions, it is difficult to make a direct comparison between the calculated and experimental branching ratios for the H and CO channels.

As the energy increases from 5 to 40 kcal mol−1_,

kCO共E兲/kOH共E兲 decreases from 5.27 to 1.13 and kCO共E兲/kOD共E兲 decreases from 15.37 to 2.85; the ratios for

the reaction of O共1_D兲+C6D

6 are 2.9–2.5 times greater than those for O共1_D兲+C

6H6. At an energy of 10.2 kcal mol−1, the kinetic energy of O共1D兲 with respect to the CM upon

pho-tolysis of O₃ at 248 nm, ␣共CO兲:␣共OH兲=0.35:0.08=4.3, agrees qualitatively with the experimental value of

␣共CO兲/␣共OH兲=2.1⫾0.2 determined with TR-FTS. Some observed CO might be produced via reaction共6b兲, involving secondary dissociation of C₆H₅O but the proportion is ex-pected to be small, and likely a major part has been taken into account in performing the extrapolation of the popula-tions to CO共␷= 0兲.

For the reaction of O共1D兲+C6D₆, the ratio

␣共CO兲:␣共OD兲 is predicted to be 0.55:0.04 at an energy of 10.2 kcal mol−1; the ratio of ␣共CO兲/␣共OD兲 is ⬃2.9 times that of ␣共CO兲/␣共OH兲 in the reaction of O共1D兲

+ C6H6. Our experimental observation of an isotopic ratio

␣共OH兲/␣共OD兲⬎1.4 is consistent with this result. According to our calculations, the ratio kOH共E兲/kOD共E兲 for the hydrogen abstraction is predicted to be about 2.1 in the energy range from 5 to 40 kcal mol−1, as expected from the deuterium isotopic effect. -10.0 -9.5 -9.0 -8.5 _(a) E/ kcal mol-1 Branching ratio
Log (k /cm 3 molecule -1 s -1 (CO + C₅H₆) (H + C6H5O) (OH + C₆H₅) (OH + C6H5,) from H-abs (CO + C5D6) (D + C₆D₅O) (OD + C6D5) (OD + C6D5,) from D-abs 0 10 20 30 40 50 60 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 (CO + C5H6) (H + C6H5O) (OH + C6H5) (CO + C5D6) (D + C6D5O) (OD + C6H5) (b)

FIG. 13.共a兲 Individual product microcanonical rate coefficients of the

reac-tion O共1_D兲+C

6H6as a function of energy.共b兲 Individual product

microca-nonical branching ratios as a function of energy.