Kinetic Studies of the Reactions of IO Radicals Determined by Cavity Ring-Down Spectroscopy

Article abstract of DOI:10.1021/jp9902497

We demonstrate the application of cavity ring-down spectroscopy (CRDS) to the measurement of concentrations of IO radical and of reaction rate coefficients for the reaction systems, IO + IO and IO + NO, using the source reaction, O(³P) + CF₃I. By monitoring IO radicals, we obtain the 295 K rate coefficients, k(O + CF₃I → IO + CF₃) = (5.8 ± 1.5) × 10^-12 cm³ molec^-1 s^-1; k(IO + IO) = (1.0 ± 0.3) × 10^-10 cm³ molec^-1 s^-1, and k(IO + NO) = (1.9 ± 0.5) × 10^-11 cm³ molec^-1 s^-1 at the pressures of 1250 Pa (9.4 Torr) and 4000 Pa (30.1 Torr). For the IO A ²Π_3/2 - X ²Π_3/2 (2, 0) bandhead at 445.04 nm we have determined an absorption cross-section, σ = (7.3 ± 0.7) × 10^-17 cm². Error limits indicate the confidence of two standard deviations and propagate the uncertainty in the absorption cross-section.

Full text of DOI:10.1021/jp9902497

J. Phys. Chem. A 1999, 103, 6173-6180
6173
Kinetic Studies of the Reactions of IO Radicals Determined by Cavity Ring-Down
Spectroscopy
Dean B. Atkinson† and Jeffrey W. Hudgens*
Physical and Chemical Properties DiVision, National Institute of Standards and Technology,
Gaithersburg, Maryland 20899
Andrew J. Orr-Ewing
School of Chemistry, UniVersity of Bristol, Cantock’s Close, Bristol BS8 1TS, U.K.
ReceiVed: January 21, 1999
We demonstrate the application of cavity ring-down spectroscopy (CRDS) to the measurement of concentrations
of IO radical and of reaction rate coefficients for the reaction systems, IO + IO and IO + NO, using the
source reaction, O(3P) + CF3I. By monitoring IO radicals, we obtain the 295 K rate coefficients, k(O + CF3I
f IO + CF3) ) (5.8 ( 1.5) × 10-12 cm3 molec-1 s-1; k(IO + IO) ) (1.0 ( 0.3) × 10-10 cm3 molec-1 s-1,
and k(IO + NO) ) (1.9 ( 0.5) × 10-11 cm3 molec-1 s-1 at the pressures of 1250 Pa (9.4 Torr) and 4000 Pa
2
2
(30.1 Torr). For the IO A Π3/2 - X Π3/2 (2, 0) bandhead at 445.04 nm we have determined an absorption
cross-section, σ ) (7.3 ( 0.7) × 10-17 cm2. Error limits indicate the confidence of two standard deviations
and propagate the uncertainty in the absorption cross-section.
1. Introduction
CRDS has been used to study reactions of the phenyl,22-26
NH2,27 and propargyl radicals,28 the kinetics of the C2H5/
Sensitive, precise, and quantitative methods for studying the
reactions of IO are necessary to clarify the role of this radical
in the chemistry of earth’s troposphere and stratosphere.1
Atmospheric iodine compounds are principally biogenic in
origin, and CH3I from marine algal emissions is rapidly
photolyzed in the troposphere, liberating iodine which may react
with ozone to form IO. The subsequent reactions of IO can
influence tropospheric ozone concentrations,2 and elevated IO
concentrations are suspected to arise in the lower troposphere
over the Arctic in the springtime, coinciding with reduced ozone
episodes.3,4 Rapid convective transport of CH3I to the upper
troposphere and lower stratosphere has been postulated3 and
would result in formation of IO at altitudes where it may
participate in catalytic stratospheric ozone destruction cycles.
The possible existence of these processes is evidenced by the
recent detection of the IO radical in the marine boundary layer.5
Several studies have been performed of the kinetics of
reactions of IO using laser photolysis and detection of IO via
absorption spectroscopy6-9 or laser induced fluorescence
(LIF).10-15 Such methods have established values of rate
coefficients for the IO + IO self-reaction,7-9 and for reaction
of IO with NO,13,15,16 Br,11 BrO,7,12 ClO,14 CHCl2CF2Cl,17
HCOOH,17 and dimethyl sulfide,18 to name a few. The current
study, however, is the first use of cavity ring-down spectroscopy
(CRDS)19-21 for detection of IO and determination of its reaction
rates, and demonstrates the suitability of combining CRDS with
laser photolysis methods for accurate kinetics studies of species
such as IO. Since the introduction of CRDS detection to the
field of kinetics by M. C. Lin,22,23 CRDS detection has become
increasingly used for rate coefficient determinations. To date,
29
C2H5O2 and vinoxy/O2 systems,30 and the chlorine and
bromine adducts of propargyl halides31 and allene.32 Where
comparisons have been made with well-established rate coef-
ficients determined by methods other than CRDS, the agreement
has been excellent,29 but validation of the CRDS method for
kinetics studies has not been extensively investigated. The
current study provides a direct comparison of reaction rate
coefficients measured both for detection of IO via a sharply
structured, rotationally resolved spectrum and via a diffuse,
unstructured absorption spectrum. The results show quantitative
agreement with the previous LIF and absorption spectroscopy
studies. We concentrate on the reactions of IO + IO and IO +
NO because these have been the subject of several previous
studies, thereby allowing a careful comparison of our CRDS-
based results with other experimental methods, and because of
their importance in linking the atmospheric chemistry of iodine
with that of NOx species and in determining the partitioning of
atmospheric iodine between I and IO. The use of CRDS
detection to determine rates of reaction of IO suggests a sensitive
method for studying the reactions of the BrO radical. BrO
2
2
exhibits no A Π - X Π LIF spectrum because of rapid
predissociation of the A 2Π state, but the effectiveness of CRDS
detection of this radical has already been demonstrated.33
2. Experimental
The laser photolysis/cavity ring-down kinetic apparatus used
to measure the rates of reactions of IO has been described in
detail elsewhere28,29 and will only be summarized briefly here.
Experiments were conducted in a quartz flow tube with high
reflectivity (HR) mirrors located at either end; these mirrors
define the ring-down cavity (RDC). The HR mirrors were held
in adjustable mounts attached to the flow tube via flexible
bellows and positioned 1.1 m apart. Reaction was initiated by
* Author for correspondence. E-mail: jeffrey.hudgens@nist.gov. Fax:
301-975-3672.
† Current address: Chemistry Department, Portland State University,
Portland, OR 97207-0751.
10.1021/jp9902497 CCC: $18.00 © 1999 American Chemical Society
Published on Web 07/14/1999

6174 J. Phys. Chem. A, Vol. 103, No. 31, 1999
Atkinson et al.
laser photolysis and reaction products were probed at a variable
delay after initiation by cavity ring-down absorption detection
using a second laser pulse. The extent of absorption was
determined by monitoring the ring-down decay rate of the laser
pulse within the cavity, and typical 1/e decay times for the ring-
down were 1 µs. The variation of IO absorption with time delay
between pulses can be analyzed (assuming Beer-Lambert
absorption behavior) to give the desired reaction rates. During
these experiments the ambient temperature of the laboratory and
reactor was 295 K.
experimental conditions reduced by the random, rather than
systematic, variation of the delay times. We measured the base
system absorbance immediately before each kinetic absorbance
measurement at a given reaction time. This zeroing procedure
eliminated system drift and confined the measurement of
absorbance change to the 6-cm long photolysis zone.
The kinetic experiments produce transient absorption (which
is readily converted to radical concentration) vs time. To derive
rate coefficients from these data, the kinetic mechanism is
designed and the absorption data is computationally simulated.
This simulation is compared to the observed data and the ø2
value is computed. Using the Levenberg-Marquardt proce-
dure,34 selected rate coefficients and concentrations are varied
to minimize this value. The simulation software comprises of a
custom graphic interface that spawns satellite shells executing
the original FORTRAN code of the ACUCHEM program35 that
simulates kinetics through numerical forward integration of a
reaction mechanism. The accuracy of this program has been
verified.35
IO was produced in the RDC by 193 nm photolysis of N2O
in the presence of CF3I (>97% and used without further
purification) and argon buffer gas. Flows of these gases into
the RDC were regulated by mass flow controllers. An ArF
excimer laser was used to produce ca. maximum energy ) 100
mJ, 20 ns duration pulses of 193 nm UV at 5 Hz repetition rate
which after beam steering and shaping placed a flux of e 6 mJ
cm-2 into the photolysis zone of the reactor. The laser pulses
were shaped into 6-cm long, 1-cm high stripes by a cylindrical
lens and a beam mask and intersected the RDC perpendicular
to the axis defined by the HR mirrors. UV photolysis of N2O
generated O(1D) which underwent rapid collisional quenching
to form O(3P). Contamination of the mirrors over the duration
of the experiment was minimized by flushing argon gas through
separately pumped regions of the RDC immediately in front of
the mirrors. The flow of Ar through these regions was regulated
so that the pressure matched the total pressure in the central
region of the RDC, as has been described in detail previously.
For reactions involving NO, the NO (either neat or ca. 5% in
Ar) was passed through a copper coil immersed in a dry ice/
acetone slush bath to remove contaminant NO2 and was
introduced into the RDC via a mass flow controller.
3. Results and Analyses
The 193 nm photolysis of N2O in the mixture of N2O, CF3I,
NO, and argon results in the formation and loss of IO and NO
via the following kinetic scheme:
N2O + hν (λ ) 193 nm) f N2 + O(1D)
(1)
(2)
O(1D) + M f O(3P) + M (M ) Ar, N2O, CF3I, NO)
O(1D) + N2O f NO + NO
f N2 + O2
(3a)
(3b)
A pulsed dye laser pumped by the output of a XeCl excimer
laser produced the probe laser light at wavelengths in the range
between 445 and 470 nm. The dye laser operated on coumarin
440 dye, and the output energy was typically between 1 and 5
mJ per pulse at 5 Hz repetition rate. The probe laser was injected
into the RDC through one end mirror. For a beam waist of 0.32
mm and a mirror reflectivity of R ≈ 99.9%, the maximum cavity
flux is e5 mJ cm-2. The mirror alignment was carefully adjusted
to give an exponential decay of the light intensity within the
cavity, with maximum suppression of interference effects caused
by beating between transverse cavity modes as described in
detail elsewhere.29 Light exiting the other end mirror was imaged
through a pinhole onto a quartz diffuser and photomultiplier
tube. For kinetics experiments the laser wavelength cor-
responded either to excitation of IO on the bandhead of the
highly structured IO A-X (2, 0) band at λCRDS ) 445.04 nm
or at the bandhead of the much weaker and diffuse IO A-X
(1, 0) band at λCRDS ) 455.17 nm.
O(3P) + NO + Ar f NO2 + Ar
(4)
O(3P) + CF3I f IO + CF3
f CF2O + IF
(5a)
(5b)
(5c)
(6a)
f CF3OI
IO + IO f I2 + O2
f I +OIO
f I2O2
(6b)
(6c)
f 2I + O2
(6d)
(7)
IO + NO f I + NO2
IO + O(3P) f O2 + I
IO f products
(8)
(9)
For each laser shot, the exponential decay of the laser light
intensity within the cavity, as detected by the photomultiplier,
was digitized using a digital oscilloscope and transferred to a
personal computer via an IEEE-488 interface for analysis. Each
decay trace was fitted separately to a first-order exponential
decay, and the resultant ring-down rates were averaged for each
time delay between the two lasers. Ring-down decay rates were
converted to absorbances via standard relationships.29 The time
delay between photolysis and probe lasers was selected ran-
domly within a predetermined range by a programmed pulse
delay generator, and 10 laser shots were averaged at that delay
time to produce one data point. Typically, data points at 100
different delay settings were accumulated to define each IO
temporal decay trace, with the effects of small drifts in the
The unimolecular process in reaction 9 corresponds to wall loss
or diffusion.
Reactions 2 and 3 describe the competition between the
collisional deactivation of O(1D) and its reaction with N2O.
Reactions 4 and 7 produce NO2. We have accounted for NO2
production because its optical absorption continuum overlaps
with the IO spectrum. The reaction rate coefficients of k3 and
the termolecular rate coefficient k4 are known36 as are the
deactivation rate coefficients of O(1D) by Ar,37 NO,38 and N2O.36
The deactivation rate coefficient of O(1D) by CF3I is unknown;
however, extrapolation of the homologous series of k2(CF3Cl)
) 4.2 × 10-12 cm3 molec-1 s-1 39 and k2(CF3Br) ) 5.9 × 10-11
cm3 molec-1s-1 40 would predict that k2(CF3I) is considerably

IO Radical Reactions Determined by CRDS
J. Phys. Chem. A, Vol. 103, No. 31, 1999 6175
larger than k2(CF3Br). Thus, as in previous studies,7,8,10 the large
global deactivation rate will quench the O(1D) population to
O(3P) within nanoseconds and allow us to ignore the effects of
any residual NO from reaction 3. This conclusion was confirmed
during our modeling of the IO absorption data.
Reaction 5 has three exothermic channels. Gilles et al.41
determined that channel 5a accounts for 83% of the products.
Little consensus exists regarding the products of reaction 6. Two
studies have concluded that channel 6a constitutes less than 5%
of the total reaction.7,42 The observation of a spectrum of OIO
suggests that channel 6b may be active; however, the mechanism
for OIO production was not established.43 To fit data for the
reaction of IO + BrO, Laszlo et al.6 postulated that between 30
and 40% of the products followed channel 6c. Harwood et al.8
also concluded that channel 6c is important and that the sum of
channels 6b and 6d accounts for as much as 50% of the
reactivity.
Kinetic traces of the rise and fall of the IO signal were
obtained over intervals corresponding to the delay between
photolysis and probe lasers of up to 50 ms. Because a numerical
model of diffusion between the CRD axis and the reactor walls
predicted that wall loss (reaction 9) may begin to affect the CRD
signal for times exceeding 20 ms, we did not use data observed
beyond this time for derivation of reaction rate coefficients.
2
2
Figure 1. CRD spectra of (a) the IO A Π3/2 - X Π3/2 (1, 0) band
and (b) part of the IO A 2Π3/2 - X 2Π3/2 (2, 0) band. Combs above the
spectrum indicate the rotational line assignments. The relative vertical
axes differ between Figure 1a and 1b. See text.
Although data beyond 20 ms were not used to derive rate
coefficients, we obtained such data and examined it as part of
our larger effort to ensure that no aerosols were forming during
these experiments. Previous studies of IO radical reactions have
noted that these systems have a propensity to form aero-
sols,7,41,42,44 and Cox et al.44 have proposed that these aerosols
are composed of I4O9. The presence of aerosols will corrupt
determinations of rate coefficients by accelerating the loss of
IO radicals. During a CRDS experiment the light scattering
induced by aerosols would cause an apparent increase of
absorbance, and this signature would also be evidenced during
the kinetic modeling of absorption data. During all experiments
we observed the decay of IO absorbance from its maximum
over intervals up to 50 ms and never detected an increase in
system absorbance. The absence of an optical signature evidenc-
ing aerosols showed that aerosols, if formed at all, were
produced after the experimental measurement process had
concluded. The absence of deposits on the reactor walls is
evidence that aerosols did not form during the 200 ms interval
between each experiment. Finally, even if such aerosols were
formed, our experimental procedure assured that they were
removed before each measurement. During a recent study of
propargyl radical kinetics, a kinetic system where soot deposits
occur, we were able to experimentally measure the time required
for our apparatus to remove stable, volatile, photolytic reaction
products from the CRDS cavity.28 Armed with this knowledge,
we adjusted our reactor flow conditions and laser repetition rate
to ensure that the reactor contents were completely refreshed
between each laser pulse. In short, no physical or optical data
evidencing aerosols or particulates were ever observed under
the experimental conditions of the present study.
the photolysis and CRD laser pulses was 500 µs. The IO (1, 0)
band is weak and diffuse, showing no discernible rotational
structure, and is not observed by LIF because of rapid
2
predissociation of IO A Π3/2 V ) 1. The detailed analyses of
these and other IO CRD spectra and the information they contain
about the predissociation of the A-state are presented else-
where.45
The broad, unstructured nature of the IO (1, 0) band makes
it well suited, in principle, to CRDS detection for kinetics
measurements, whereas use of the sharp rotational structure of
the IO (2, 0) band requires greater care. To apply CRDS
detection to the accurate measurement of rate coefficients, we
must ensure that all cavity modes excited by the probe laser
decay at the same rate. If the absorption feature used to monitor
concentrations of a radical or molecule is narrower than the
laser bandwidth, the cavity modes excited by the laser and
overlapped by the absorption line will decay (ring down) faster
than those not spanned by the absorption. The resultant ring-
down decay measured for each laser pulse will thus not be a
single exponential, and ring-down rates will not be straightfor-
wardly related to the concentration of the absorber.46 The
bandhead of the IO (2, 0) band, although discrete, is sufficiently
broad to extinguish all cavity modes excited by the probe laser
equally, but isolated rotational lines have Doppler and pressure-
broadened widths comparable to the laser bandwidth. Kinetics
measurements were thus made by tuning the probe laser either
to the bandhead of the IO (2, 0) band at λCRDS ) 445.04 nm or
the IO (1, 0) band at λCRDS ) 455.17 nm. The latter measure-
ments will be less sensitive because of the inherent weakness
of the IO (1, 0) band.
3.1. IO CRD Spectra and Absorption Cross-Sections. CRD
spectra of the IO A2Π3/2 - X2Π3/2 (2, 0), (1, 0), and (0, 0)
bands were recorded with typical gas flows of 60 standard cubic
centimeters per minute (sccm) of N2O, 3 sccm of CF3I, and
1000 sccm of Ar, with the pump throttled to give a total pressure
of 4000 Pa (30.1 Torr). As an example of the sensitivity of
CRDS detection, Figure 1 shows representative spectra of the
To derive rate coefficients of the second-order reaction 6,
IO + IO, we need accurate absorption cross-sections, σλ, for
the IO (1, 0) and IO (2, 0) bands. Although two previously
reported, low resolution spectra (∆λ ≈ 0.8 nm fwhm7 and ∆λ
≈ 0.44 nm fwhm8) obtained with monochromators do not
exhibit rotational fine structure, these spectra contain the
2
2
IO (1, 0) and IO (2, 0) bands from the A Π3/2 - X Π3/2
transition. These spectra were obtained when the delay between

6176 J. Phys. Chem. A, Vol. 103, No. 31, 1999
Atkinson et al.
intensity data that enable the renormalization of the rotationally
resolved CRD spectrum (∆λ ≈ 0.0013 nm) to find σλ.
Absorption experiments measure an average absorption cross-
section, σeλff, that is weighted by a normalized instrument
function, S(τ), e.g.,
σeλff
)
-∆∆λλ/2/2I(λ + τ) Sinstrument(τ)dτ
(10)
∫
where I(λ) is the spectrum of the molecule and ∆λ is the spectral
width of nonzero values of S(τ). For a CRD experiment, SCRD(τ)
is nonzero over the width of a single cavity mode. (As
mentioned above, when more than one cavity mode is active,
the spectrum must observe equal σλs to attain good absorbance
measurements.) Under the conditions that permit good CRDS
measurements, I(λ) is constant over ∆λ and eq 10 reduces to
σλ ≡ σeλff ) I(λ)dλo, where dλo is the unit length. This
condition occurs for the CRD spectra of the IO (2, 0) and IO
(1, 0) bands.
For measurements involving a monochromator, Smono(τ) is
the slit function. The monochromators used in previous studies
could not resolve the rotational spectrum of the IO (2, 0) band.
Thus, the fine structure of I(λ) was convoluted with Smono(τ),
and each observed spectrum was simply a broadened, weighted
average of the rotational intensity envelope. Since the Smono(τ)
of each monochromator system is unique, the previously
reported absorption cross-sections of the IO (2, 0) bands7-9,44
are effective cross-sections, σeλff, that are suitable only for
concentration measurements within the specific apparatus.
Using knowledge of Smono(τ), we can convolute the I(λ)
observed by CRDS and compute the band profile observed by
the monochromator. Since the maximum of this computed
profile is proportional to σeλff, we can renormalize I(λ) to
absolute cross-section, thus, determining σλ for each rotational
line within the IO (2, 0) band, hence enabling us to determine
the concentration of IO with the CRDS spectrum. To obtain
this normalization factor, we used the calibration slit function
of the 0.25 m monochromator and σeλff ) 1.05 × 10-17 cm2
measured by Laszlo et al.7 during their study of the IO spectrum.
We also confirmed this Smono(τ) by rescanning the mono-
chromator over the 435.8 nm mercury emission line. This
Smono(τ) (0.7 nm fwhm and 1.2 nm full-width) was used in eq
Figure 2. Kinetic measurements that determine k6(IO + IO). Squares
show absorbance as a function of the time delay between the photolysis
laser pulse that initiates IO formation and the CRDS laser pulse that
probes absorbance. The reagent flows of both measurements are
identical and the pressure is 4015 Pa. The fits to the data, shown by
the solid lines, give individual determinations of k6(IO + IO): (a)
Kinetic measurement while monitoring IO concentration at the (2, 0)
bandhead; (b) Kinetic measurement while monitoring IO concentration
at the (1, 0) band.
because the rotational lines of the IO (1, 0) band are homo-
geneously broadened into a continuum and because the resolu-
tion of both experiments (∆λ ≈ 0.44 nm8 and ∆λ ) 0.01 nm,9
respectively) were of sufficient resolution to simplify eq 10 to
σλ ) I(λ)dλo. The accord between the CRDS and these
conventional absorption measurements of σI4O55 increases confi-
dence in the accuracy of our derived cross-section at the
bandhead
IO (2, 0), σ
of
IO
445
.
10 to convolute the CRD spectrum, I(λ). The renormalized
In separate experiments we used a 0.25 m monochromator
IO
spectrum found σ ) (7.3 ( 0.7) × 10-17 cm2 for the IO (2,
to measure the absorption cross-sections of NO2 at 445.04 and
445
NO2
445
455.17 nm and found σ
) 5.5 × 10-19 cm2 and σN45O52 ) 4.4
0) bandhead at 445.04 nm, where the uncertainty represents two
standard deviations. An intuitive way to look at this renormal-
ization process is to view the rotational spectrum as a set of
boxes, each of which contains either a rotational line or no
intensity at all. The ratio of σI4O45/σeλff ) 7 indicates that only
∼1/7 of all boxes contain a rotational line. The low resolution
monochromator is averaging many resolution boxes of which
6 of every 7 contains zero absorbance.
× 10-19 cm2, respectively. At both wavelengths the spectrum
of NO2 is a continuum and σλ ) σeff. In the following, these
λ
cross-sections are used to account for NO2 production in our
data analysis model. The relevant cross-sections of I2 are about
σI4245 ≈ 3 × 10-20 cm2 and σ4I255 ≈ 4 × 10-20 cm2 , which are
sufficiently small that we can ignore the contributions to total
absorbance by I2.47
The absorption cross-section of the IO (1, 0) band is
determined from the ratio of the transient absorbance maxima
of the IO (2, 0) and IO (1, 0) bands obtained from transient
absorption signals observed under the same flow reactor
conditions. Figure 2 shows an example of these transient signal
measurements. The ratio of the transient maxima at t ≈ 180 µs
is AI4O45(2,0)/AIO(1,0) ) 0.1 ( 0.01, which leads to σIO ) (7.3 (
3.2. Determination of k5(O + CF3I) and k6(IO + IO).
Determinations of the rate coefficients k5(O + CF3I) and
k6(IO + IO) were made under a variety of pressure and gas-
composition conditions (Table 1), both to check for pressure
dependence of the rate coefficients and to assess the likelihood
of aerosol formation within the reactor. Data sets were recorded
in argon buffer (with no added NO) at total pressures of 1290
Pa (9.7 Torr) and 4000 Pa (31.2 Torr). Since the reaction rate
coefficients obtained at both pressures were statistically the
same, we report their averages.
455
455
0.7) × 10-18 cm2 for the cross-section of the IO (1, 0) band at
455.17 nm. For comparison, Harwood et al.8 obtained σ4IO55 ) 8
× 10-18 cm2 and the laser absorption experiment of Stickel et
al.9 obtained σ4IO55 ) 6.2 × 10-18 cm2. The CRDS measurement
may be compared directly with these prior measurements
Figure 3 shows representative absorbance data and the fit of
these data that extracts one individual rate for k5(O + CF3I)

IO Radical Reactions Determined by CRDS
J. Phys. Chem. A, Vol. 103, No. 31, 1999 6177
TABLE 1: Summary of the Experimental Conditions and Derived Rate Coefficients Obtained at 295 K for the O + CF3I
Reaction System
partial press
CF3I, Pa
derived k5,b
derived k6,b
derived k7,b
transitiona
total press., Pa
N2O, Pa
NO, Pa
cm3 molec-1 s-1
cm3 molec-1 s-1
cm3 molec-1 s-1
IO (2, 0)
IO (2, 0)
IO (2, 0)
IO (2, 0)
IO (2, 0)
IO (2, 0)
IO (2, 0)
IO (1, 0)
IO (2, 0)
IO (2, 0)
IO (2, 0)
IO (2, 0)
IO (2, 0)
IO (2, 0)
IO (2, 0)
IO (2, 0)
IO (2, 0)
IO (2, 0)
IO (2, 0)
IO (2, 0)
IO (2, 0)
IO (2, 0)
IO (2, 0)
IO (1, 0)
IO (1, 0)
IO (1, 0)
IO (1, 0)
IO (1, 0)
IO (1, 0)
1252
1290
1302
1264
4010
4015
4015
4015
1241
1264
4010
1241
1264
4014
1241
1241
1264
4016
1241
1264
4016
1264
1264
4014
4016
4016
4019
4022
4020
34.4
34.4
34.1
34.6
83.4
83.5
83.5
83.5
34.1
34.6
83.4
34.1
34.6
83.5
34.1
34.1
34.6
83.5
34.1
34.6
83.5
34.6
34.6
83.5
83.5
83.5
83.6
83.6
83.6
4.9
4.9
4.8
(6.0 ( 0.4) × 10-12
(10.7 ( 0.8) × 10-11
(11.3 ( 0.5) × 10-11
(12.0 ( 0.7) × 10-11
(8.2 ( 0.2) × 10-11
(8.3 ( 0.2) × 10-11
(9.8 ( 0.4) × 10-11
(8.8 ( 0.8) × 10-11
(6.0 ( 0.8) × 10-12
(4.5 ( 1) × 10-12
(6.2 ( 0.4) × 10-12
(6.4 ( 0.5) × 10-12
9.8
11.7
11.8
11.9
11.9
4.8
9.8
11.7
4.8
9.8
11.8
4.8
4.8
9.8
11.8
4.8
9.8
11.8
9.8
9.8
11.8
11.8
11.8
11.8
11.8
11.8
0.13
0.14
0.16
0.20
0.20
0.24
0.27
0.27
0.27
0.31
0.33
0.34
0.39
0.41
0.47
0.08
0.16
0.24
0.31
0.39
0.47
(2.4 ( 0.1) × 10-11
(1.87 ( 0.08) × 10-11
(1.48 ( 0.06) × 10-11
(2.2 ( 0.1) × 10-11
(1.78 ( 0.08) × 10-11
(1.75 ( 0.08) × 10-11
(2.21 ( 0.1) × 10-11
(2.21 ( 0.09) × 10-11
(1.74 ( 0.08) × 10-11
(1.71 ( 0.07) × 10-11
(2.27 ( 0.09) × 10-11
(1.73 ( 0.08) × 10-11
(1.92 ( 0.07) × 10-11
(1.73 ( 0.07) × 10-11
(1.7 ( 0.2) × 10-11
(1.9 ( 0.3) × 10-11
(1.3 ( 0.2) × 10-11
(1.8 ( 0.3) × 10-11
(1.5 ( 0.2) × 10-11
(2.0 ( 0.2) × 10-11
(2.1 ( 0.2) × 10-11
a IO (1, 0) denotes measurements at λCRDS ) 455.17 nm and IO (2, 0) denotes measurements at λCRDS ) 445.04 nm. b Rate coefficient lists the
expanded uncertainty (2σ) of the precision.
Under all of the conditions employed, the model predicts that
IO formation was >95% complete within 500 µs.
In a previous study, CRDS was used to detect rotational lines
3
1
of the IF B Π (0+)-X Σ+(0+) transition with good sensitiv-
ity.45 Since the spectrum of IF overlaps the IO (2, 0) and IO (1,
0) bands and is readily observed by CRDS, the absence of any
IF signal indicates that the production of IF by reaction 5b is
insignificant.
Figure 2 shows representative absorbance data obtained while
measuring IO concentrations with the IO (2, 0) and IO (1, 0)
bands. The solid lines are the fits that obtain individual
measurements of k6(IO + IO). The fits were also computed with
a numerical model comprising of reactions 5 and 6. The average
of data (Table 1) obtained by monitoring the IO (2, 0) bandhead
yielded k6(IO + IO) ) (1.0 ( 0.3) × 10-10 cm3 molec-1 s-1
(Table 2). Averages of data measured at 1250 and 4000 Pa were
identical. When monitoring IO radicals with the (1, 0) band,
we obtained k6(IO + IO) ) (8.8 ( 1.2) × 10-11 cm3 molec-1
s-1, where this determination of k6(IO + IO) is based on our
determination of σI4O55. The stated uncertainty of each determi-
nation of k6(IO + IO) denotes two standard deviations and
contains the uncertainty in the absorption cross-section.
Figure 3. Kinetic measurement that determines k5(O + CF3I). The
squares show absorbance at 445.04 nm as a function of the time delay
between the photolysis laser pulse that initiates the IO production
process (reactions 1, 2, and 5) and the CRDS laser pulse that probes
absorbance by IO radicals. The fit to the data, shown by the solid line,
gives one determination of k5(O + CF3I). The pressure in the experiment
is 1252 Pa.
When computing fits of the kinetic data, adding reaction 8
to the model had little effect and produced essentially the same
reaction rate coefficient. The insensitivity of the model to
reaction 8 occurs because the concentrations of O(3P) and IO
are small relative to that of CF3I. We also attempted to fit the
IO radical decay data using models that explicitly accounted
for the quenching of O(1D) via reaction 2 and for NO produced
by reaction 3a. Presumably, this more extensive model would
account for any enhanced loss of IO from a cross-reaction with
NO (reaction 7), where the NO is generated by reaction 3a.
while monitoring the transient absorbance at the IO (2,0)
bandhead at λCRDS ) 445.04 nm. The average of all determina-
tions from individual experiments yielded k5(O + CF3I) ) (5.8
( 1.5) × 10-12 cm3 molec-1 s-1 (Table 2), where the uncertainty
denotes two standard deviations. The model used to fit these
data comprised reactions 5 and 6. According to the model, the
photolysis event dissociates between 0.01 and 0.05% of the
irradiated N2O to yield maximum IO concentrations ranging
between 1.0 × 1012 radicals cm-3 and 1.2 × 1013 radicals cm-3
.

6178 J. Phys. Chem. A, Vol. 103, No. 31, 1999
Atkinson et al.
TABLE 2: Summary of the Methods, Conditions, and Rate Coefficients at 295 K of this Work and Previous Studies of the
Primary Reactions in the O + CF3I Reaction System. Rate Coefficients are Listed for Temperatures of 298 K unless Otherwise
Indicateda
298 K rate
coefficient,a
techniqueb
(species
detected)
cm3 molec-1 s-1
pressure, Pa
ref
comments
k5 (O(3P) + CF3I)
(5.8 ( 1.5) × 10-12
(1.1 ( 0.3) × 10-11 c
(5.3 ( 0.08) × 10-12
(4.3 ( 0.2) × 10-12
1250, 4000
800
6700
CRDS (IO) this work T ) 295 K.
RF (O3P)
LIF (IO)
RF (O3P)
57
10
41
Temperature dependence over range 230 to 310 K and a rate expression are reported.
Branching ratio, temperature dependence over the range 213 to 364 K and a rate
expression for k5 are reported.
13300
LIF (IO)
LIF (IF)
(6.5 ( 1.5) × 10-12 c
1330
58
IF was a product of complex chemistry and not from reaction 5.
k6 (IO + IO)
(1.0 ( 0.3) × 10-10
(9.9 ( 3.0) × 10-11
(8.0 ( 1.8) × 10-11
(6.6 ( 2) × 10-11 c
(5.6 ( 1.2) × 10-11
4 × 10-10
1250, 4000
53200
CRDS (IO) this work T ) 295 K.
UVA (IO)
7980-79800 UVA (IO)
8
7
9
42
44
59
60
T ) 250 to 320 K.
T ) 295 K
T ) 300 K.
T ) 250 to 373 K.
T ) 303 K. See text.
101080
270-93100
101080
240-670
133
UVA (IO)
UVA (IO)
UVA (IO)
MS (IO)
(5.5 ( 0.8) × 10-11
(3.0 ( 1.1) × 10-11 c
MS (IO)
k7 (IO + NO)
(1.9 ( 0.5) × 10-11
1250, 4000
CRDS (IO) this work T ) 295 K.
(2.17 ( 0.22) × 10-11 5330-26700 UVA (IO)
16
Temperature dependence over the range 242 to 359 K and a rate expression are reported.
Here k6 is independent of pressure.
Temperature dependence over the range 237 to 346 K and a rate expression are reported.
(1.9 ( 0.1) × 10-11
670-13330
53-106
LIF (IO)
LIF (IO)
MS (IO)
15
13
61
(2.8 ( 0.2) × 10-11
(1.67 ( 0.16) × 10-11 270
a Uncertainties denote two standard deviations. b CRD ) cavity ring-down, RF ) resonance fluorescence, LIF ) laser induced fluorescence,
UVA ) UV-visible absorption, MS ) mass spectrometry. c Uncertainty not explained but is presumed to be two standard deviations.
According to this reaction scheme, NO production will vary as
functions of Ar, CF3I, and N2O densities within the reactor.
However, neither k5(O + CF3I) nor k6(IO + IO) exhibited
statistically meaningful dependencies as total pressure of argon
buffer changed from 1250 to 4000 Pa, as CF3I density was
varied by a factor of 3, and as N2O density was varied by a
factor of 10. Since no effect was observed in the data, we
discarded reactions 2, 3a, and 7 from the final model.
and IO (1, 0) bands, we conclude that reaction 11b played no
role under the conditions of our kinetic studies. Gilles et al.41
attempted to observe CF3O during their study without success.
Although their search was directed toward characterizing
reaction 5, this negative result suggests that reaction 11a is
comparatively slow. We conclude that CF3 radicals are unlikely
to have perturbed the temporal evolution of IO during the present
experiments.
Previously, Laszlo et al.7 noted that increased CF3I concentra-
tion also increased the propensity for aerosol formation. Aerosols
were observed to increase the apparent rate of k6(IO + IO).
During the present work k6(IO + IO) did not vary as a function
of initial CF3I concentration. This invariance is additional
evidence that aerosol formation was minimal or absent during
our kinetics determinations.
Evaluating the effects of I atoms upon determinations of k6(IO
+ IO) is complicated by our incomplete knowledge of reaction
6. By monitoring I2 absorption, Harwood et al.8 determined that
all I atoms and IO radicals eventually revert to I2. The growth
of I2 absorption followed second order kinetics with relatively
slow rate constants ranging between 3 × 10-13 cm3 molec-1
s-1 and 1.2 × 10-12 cm3 molec-1 s-1. Laszlo et al.7 observed
similar I2 signals and noted their pressure dependence. To
account for these observations, Harwood et al.8 proposed the
reaction scheme
In our model the decay of IO is presumed to occur only
through reaction 6; however, we can also consider the effects
of CF3 and of I atoms upon the determination of k6(IO + IO).
Reaction rate coefficients of CF3 with itself, with O(3P),48,49
with I atoms,50 and with CF3I 51 are known. These reactions
are slow, and significant concentrations of CF3 persist. Presum-
ably the reaction
IO + I (+M) f I2O
I + I2O f I2 + IO
(12)
(13)
The formation of I2 through reactions 12 and 13 shows a second-
order behavior because the IO production reaction
CF3 + IO f CF3O + I
∆rH2o98.15 ) -206 kJ mol-1 52-54 (11a)
O(3P) + I2 f IO + I
(14)
f CF2O + IF
∆rH2o98.15 ) -388 kJ mol-1
(11b)
creates equal concentrations of IO and I atoms.
Because the present study monitors IO radicals, reactions 12
and 13 will tend to form a null cycle with little net effect on
the absorption trace. Furthermore, during this study the impor-
tance of reactions 12 and 13 is greatly diminished, owing to
the absence of I atoms at early times. Using the absorption cross-
sections of CF3I 55 and N2O 36 at 193 nm, we estimate that the
initial molar concentration of I atoms is < 2.5% of the initial
might perturb rate coefficient determinations. Adding the known
rate coefficients to the model and refitting the data to find
k11(CF3 + IO) degrades the quality of the fits for k5(O + CF3I)
and k6(IO + IO) and increases the accumulated error. No
experimental data support the activity of reaction 11. Since we
do not observe IF absorption bands overlapping the IO (2, 0)

IO Radical Reactions Determined by CRDS
J. Phys. Chem. A, Vol. 103, No. 31, 1999 6179
via the termolecular reactions these radicals do not affect
CF3 + NO (+M) f CF3NO
I + NO (+M) f INO
(15)
(16)
determinations of k7(IO + NO). For these simulations we have
used the k5 and k6 of this work (Table 2). Reaction 7 generates
NO2 that has an increasing absorbance contribution at λCRDS
.
Consequently, the numerical model incorporates the appropriate
optical absorption coefficients and simulates the combined
absorbance by IO and NO2. The average of all individual kinetic
traces obtained while monitoring the IO (2, 0) bandhead is k7-
(IO + NO) ) (1.9 ( 0.5) × 10-11 cm3 molec-1 s-1. The average
of all kinetic traces obtained while monitoring the IO (1, 0)
band is k7(IO + NO) ) (1.8 ( 0.6) × 10-11 cm3 molec-1 s-1
.
Since the relative cross-sections of IO and NO2 differ for the
IO (2, 0) and IO (1, 0) bands, the good agreement of k7(IO +
NO) derived at each wavelength is a measure of consistency of
the overall model.
4. Discussion
Table 2 summarizes the ambient temperature (295 K) rate
coefficients determined during this study of the reaction system
O(3P) + CF3I. Table 2 also lists the results reported by previous
studies; first, by detection method for IO, and then, by
chronology. Except as noted, these ambient temperature rate
coefficients are for 298 K. Uncertainties listed with each rate
coefficient denote two standard deviations and propagate the
uncertainty in the absorption cross-sections. Previous determina-
tions of k6(IO + IO) and k7(IO + NO) have used different
sources for O(3P), different iodine atom donors, pressure ranges,
and detection techniques. As seen in Table 2, the present rate
coefficients derived from CRDS measurements are in excellent
agreement with the more current determinations and add to
consensus among these rate coefficients. The only serious
disagreement is with the rate coefficient of Inoue et al.13 for
k7(IO + NO). Cox and Coker44 indicated that their anomalously
high k6(IO + IO) (Table 2) contains contributions from
termolecular and heterogeneous processes. Our measurements
of k7(IO + NO) at 1200 and 4000 Pa total pressure show no
significant variation with pressure, thereby supporting the data
of Daykin and Wine16 over the pressure range 5330 and 26700
Pa (40 and 200 Torr) and of Turnipseed et al.15 over the pressure
range 670 and 13330 Pa (5 and 100 Torr). These accumulated
data indicate that k7(IO + NO) is independent of pressure for
pressures above 1330 Pa (10 Torr). Our rate coefficients also
show no dependence on the concentration of CF3I, which
provides additional evidence that aerosol formation in the reactor
can be discounted under our experimental conditions.
Figure 4. Kinetic measurements that determine k7(IO + NO). The
squares show absorbance as a function of the time delay between the
photolysis laser pulse that initiates IO formation and the CRDS laser
pulse that probes absorbance by the IO radical and NO2. The fits of
the data, shown by the solid lines, give individual determinations of
k7(IO + NO): (a) Kinetic measurement while monitoring IO concentra-
tion at the (2, 0) bandhead. The NO density is 6.77 × 1013 molec cm-3
;
(b) Kinetic measurement while monitoring IO concentration at the (1,
0) band. The NO density is 1.13 × 1014 molec cm-3
.
O(3P) atoms. Our source of IO radicals (reaction 5) does not
produce I atoms. Reactions 6b and 6d provide a comparatively
meager source of I atoms. Thus, the I atoms in our present study
are expected to have little effect on the observed rate coef-
ficients.
3.3. Determination of k7(IO + NO). Kinetics measurements
of reaction 7, IO + NO, were undertaken at a total pressure in
the reaction cell of near 1250 Pa (9.4 Torr) and 4000 Pa (30.1
Torr) (Table 1). At pressures of ∼1250 Pa, typical flow rates
of the constituent gases were 750 sccm Ar, 30 sccm N2O, and
3 sccm CF3I, and neat NO was added at flow rates varying
from 0.1 to 0.25 sccm. At pressures near 4000 Pa typical flow
rates were 1000 sccm Ar, 30 sccm N2O, and 3 sccm CF3I, and
a mixture of 4.80% NO in argon was added with flow rates
from 0.5 to 3.0 sccm. The system absorbance was monitored
either at the IO (2, 0) bandhead at 445.04 nm or with the IO (1,
0) band at 455.17 nm. Detection of IO at the bandhead of the
A-X (1, 0) band gave poorer quality data sets because the
absorption via this band is weaker than via the A-X (2, 0)
band. Also, the effects of NO2 absorption at longer times are
more pronounced since the ratio of the IO/NO2 absorption cross-
sections is closer to unity.
To date, the validation of CRDS for determination of rate
coefficients has been sparse and hence the current investigation
constitutes a contribution to the evidence supporting the use of
this recently developed spectroscopic technique for such studies.
Moreover, this study provides validation that CRDS can
accurately determine rate coefficients for reactions studied in
first order (k5 and k7) and in second order (k6).
Figure 4 shows the absorbance data and the fit that obtains
individual measurements of k7(IO + NO). Kinetic traces
obtained while monitoring the IO (2, 0) and IO (1, 0) bands are
shown. Each data set exhibited a linear pseudo-first-order plot
of kobs as a function of NO concentration; however, to
compensate for side reactions, we computationally fit the data
directly using a kinetic model comprising of reactions 5, 6, and
7. Since added NO rapidly depletes CF3 radicals56 and I atoms36
This study also provides practical data regarding the range
of absorption coefficients that are of use for kinetics measure-
ments with CRDS. Using the IO (2, 0) band and assuming a
minimum measurable absorbance of 10-7 cm-1, we estimate
that the detection limit for IO radicals is 1.4 × 109 molec cm-3
.
By obtaining readily available mirrors of higher reflectivity this
limit could be improved by another factor of 10. However, the
more interesting aspect is that measurements can be obtained

6180 J. Phys. Chem. A, Vol. 103, No. 31, 1999
Atkinson et al.
IO
455
under less than ideal conditions. At σ ) 7.3 × 10-18 cm2,
(19) O’Keefe, A.; Deacon, A. G. ReV. Sci. Instrumen. 1988, 59, 2544.
(20) Scherer, J. J.; Paul, J. B.; O’Keefe, A.; Saykally, R. J. Chem. ReV.
1997, 97, 25.
the IO (1, 0) absorption cross-section is an order of magnitude
smaller than the IO (2, 0) bandhead. Furthermore, measurements
using the IO (1, 0) band suffer greater interference from an
increasing population of absorbing NO2 than when using than
IO (2, 0) band. Despite these hindrances, we were able to follow
the decay of IO for about three 1/e lifetimes using the IO (1, 0)
band for peak radical densities of ∼5 × 1012 cm-3. Importantly,
agreement is found for the determinations of k6 and k7 at both
wavelengths. As expected, rate coefficients determined using
the IO (1, 0) band have larger uncertainties and any study using
such small cross-sections can examine only a restricted range
of experimental conditions. Even so, good rate coefficients can
be obtained.
(21) Wheeler, M. D.; Newman, S. M.; Orr-Ewing, A. J.; Ashfold, M.
N. R. J. Chem. Soc. Faraday Trans. 1998, 94, 337.
(22) Yu, T.; Lin, M. C. Int. J. Chem. Kinet. 1993, 25, 875.
(23) Yu, T.; Lin, M. C. J. Am. Chem. Soc. 1994, 116, 9571.
(24) Yu, T.; Lin, M. C. J. Phys. Chem. 1995, 99, 8599.
(25) Yu, T.; Lin, M. C. Int. J. Chem. Kinet. 1994, 26, 771.
(26) Yu, T.; Lin, M. C. J. Phys. Chem. 1994, 98, 9697.
(27) Diau, E. W.; Yu, T.; Wagner, M. A. G.; Lin, M. C. J. Phys. Chem.
1994, 98, 4034.
(28) Atkinson, D. B.; Hudgens, J. W J. Phys. Chem. A, 1999, 103, 4242.
(29) Atkinson, D. B.; Hudgens, J. W. J. Phys. Chem. A 1997, 101, 3901.
(30) Zhu, L.; Johnston, G. J. Phys. Chem. 1995, 99, 15114.
(31) Atkinson, D. B.; Hudgens, J. W. J. Phys. Chem. A, submitted.
(32) Atkinson, D. B.; Hudgens, J. W., in preparation.
(33) Wheeler, M. D.; Newman, S. M.; Ishiwata, T.; Kawasaki, M.; Orr-
Ewing, A. J. Chem. Phys. Lett. 1998, 285, 346.
(34) Nonlinear LeVenberg-Marquardt fitting Virtual instrument,
Labview 4.0, National Instruments Corporation, Austin, TX, 1994-1995.
(35) Braun, W. Int. J. Chem. Kinet. 1988, 20, 51.
(36) DeMore, W. B.; Sander, S. P.; Golden, D. M.; Hampson, R. F.;
Kurylo, M. J.; Howard, C. J.; Ravishankara, A. R.; Kolb, C. E.; Molina,
M. J. Chemical kinetics and photochemical data for use in stratospheric
modeling. EValuation number 12, JPL publication 97-4, 1997.
(37) Davidson, J. A.; Schiff, H. I.; Brown, T. J.; Streit, G. E.; Howard,
C. J. J. Chem. Phys. 1978, 69, 1213.
Acknowledgment. The visit of A.J.O.E. to NIST was made
possible by the award of a NATO travel grant (Ref. CRG
950105). A.J.O.E. thanks the Royal Society for the award of
the Eliz. Challenor Research Fellowship. We thank Drs. R. E.
Huie, B. Laszlo, and A. Fahr for experimental assistance in the
course of this study and Dr. Michael Kurylo for helpful
comments and suggestions during the preparation and revision
of this manuscript.
(38) Doroshenko, V. M.; Kudryavtsev, N. N.; Smetanin, V. V. High
Energy Chem. 1992, 26, 227.
(39) Takahashi, K.; Wada, R.; Matsumi, Y.; Kawasaki, M. J. Phys.
Chem. 1996, 100, 10145.
References and Notes
(40) Thompson, J. E.; Ravishankara, A. R. Int. J. Chem. Kinet. 1993,
25, 479.
(41) Gilles, M. K.; Turnipseed, A. A.; Talukdar, R. K.; Rudich, Y.;
Villalta, P. W.; Huey, L. G.; Burkholder, J. B.; Ravishankara, A. R. J. Phys.
Chem. 1996, 100, 14005.
(1) Wayne, R. P.; Poulet, G.; Biggs, P.; Burrows, J. P.; Cox, R. A.;
Crutzen, P. J.; Hayman, G. D.; Jenkin, M. E.; Le Bras, G.; Moortgat, G.
K.; Platt, U.; Schindler, R. N. Halogen oxides: radicals, sources and
reserVoirs in the laboratory and in the atmosphere, Air Pollution Research
Report 55, Directorate-General Science, Research and Development
Environment research programme, European Commission, 1995.
(2) Davis, D.; Crawford, J.; Liu, S.; McKeen, S.; Bandy, A.; Thornton,
D.; Rowland, F.; Blake, D. J. Geophys. Res. 1996, 101, 2135.
(3) Solomon, S.; Garcia, R. R.; Ravishankara, A. R. J. Geophys. Res.
1994, 99, 20491.
(4) Sturges, W. T.; Barrie, L. A. Atmos. EnViron. 1988, 22, 1179.
(5) Alicke, B.; Hebestreit, K.; Platt, U.; Carpenter, L. J.; Sturges, W.
Ann. Geophys. 1998, 16 (Suppl. II), C716.
(6) Laszlo, B.; Kurylo, M. J.; Huie, R. E. J. Geophys. Res. 1997, 102,
1523.
(7) Laszlo, B.; Kurylo, M. J.; Huie, R. E. J. Phys. Chem. 1995, 99,
11701.
(8) Harwood, M. H.; Burkholder, J. B.; Hunter, M.; Fox, R. W.;
Ravishankara, A. R. J. Phys. Chem. A 1997, 101, 853.
(9) Stickel, R. E.; Hynes, A. J.; Bradshaw, J. D.; Chameides, W. L.;
Davis, D. D. J. Phys. Chem. 1988, 92, 1862.
(42) Sander, S. P. J. Phys. Chem. 1986, 90, 2194.
(43) Himmelmann, S.; Orphal, J.; Bovensmann, H.; Richter, A.;
Ladsta¨tter-Weissenmayer, A.; Burrows, J. P. Chem. Phys. Lett. 1996, 251,
330.
(44) Cox, R. A.; Coker, G. B. J. Phys. Chem. 1983, 87, 4478.
(45) Newman, S. M.; Howie, W.; Lane, I. C.; Upson, M.; Orr-Ewing,
A. J. J. Chem. Soc., Faraday Trans. 1998, 94, 2681.
(46) Zalicki, P.; Zare, R. N. J. Chem. Phys. 1995, 102, 2708.
(47) Calvert, J. G.; Pitts, J. N. Photochemistry; Wiley: New York, 1966.
(48) Ryan, K. R.; Plumb, I. C. J. Phys. Chem. 1982, 86, 4678.
(49) Tsai, C.; Belanger, S. M.; Kim, J. T.; Lord, J. R.; McFadden, D.
L. J. Phys. Chem. 1989, 93, 1916.
(50) Skorobogatov, G. A.; Seleznev, V. G.; Slesar, O. N. Dokl. Phys.
Chem. (Engl. Transl.) 1976, 231, 1292.
(51) Andreeva, T. L.; Kuznetsova, S. V.; Maslov, A. I.; Sobel’man, I.
I.; Sorokin, V.N. High Energy Chem. 1972, 6, 368.
(52) Zachariah, M.; Westmoreland, P.; Burgess, D. R.; Tsang, W. J.
Phys. Chem. 1996, 100, 9737.
(10) Ho¨lscher, D.; Fockenberg, C.; Zellner, R. Berichte Der Bunsen-
Ges. Phys. Chem. 1998, 102, 716.
(53) Chase, M. W., Jr. J. Phys. Chem. Ref. Data 1996, 25, 1297.
(54) Chase, M. W., Jr. J. Phys. Chem. Ref. Data, Monograph 9, 1998.
(55) Fahr, A.; Nayak, A. K.; Huie, R. E. Chem. Phys. 1995, 199, 275.
(56) Ley, L.; Masanet, J.; Caralp, F.; Lesclaux, R. J. Phys. Chem. 1995,
99, 1953.
(57) Addison, M. C.; Donovan, R. J.; Garraway, J. Chem. Soc. Faraday
Discuss. 1979, 67, 286.
(11) Gilles, M. K.; Turnipseed, A. A.; Burkholder, J. B.; Ravishankara,
A. R. Chem. Phys. Lett. 1997, 272, 75.
(12) Gilles, M. K.; Turnipseed, A. A.; Burkholder, J. B.; Ravishankara,
A. R.; Solomon, S. J. Phys. Chem. A, 1997, 101, 5526.
(13) Inoue, G.; Suzuki, M.; Washida, N. J. Chem. Phys. 1983, 79, 4730.
(14) Turnipseed, A. A.; Gilles, M. K.; Burkholder, J. B.; Ravishankara,
A. R. J. Phys. Chem. 1997, 101, 5517.
(58) Watson, T. A.; Addison, M.; Wittig, C. Chem. Phys. 1983, 78, 57.
(59) Barnes, I.; Bastian, V.; Becker, K. H.; Overath, R. D. Int. J. Chem.
Kinet. 1991, 23, 579.
(60) Martin, D.; Jourdain, J. L.; Laverdet, G.; Le Bras, G. Int. J. Chem.
Kinet. 1987, 19, 503.
(15) Turnipseed, A. A.; Gilles, M. K.; Burkholder, J. B.; Ravishankara,
A. R. Chem. Phys. Lett. 1995, 242, 427.
(16) Daykin, E. P.; Wine, P. H. J. Phys. Chem. 1990, 94, 4528.
(17) Buben, S. N.; Larin, I. K.; Trofimova, E. M. Kinet. Catal. 1995,
36, 745.
(18) Daykin, E. P.; Wine, P. H. J. Geophys. Res. 1990, 95, 18547.
(61) Ray, G. W.; Watson, R. T. J. Phys. Chem. 1981, 85, 2955.