Pulsed laser photolysis kinetics study of reactions between small radical species


Material Information

Pulsed laser photolysis kinetics study of reactions between small radical species
Physical Description:
vi, 140 leaves : ill. ; 29 cm.
Nicovich, John Michael, 1950-
Publication Date:


Subjects / Keywords:
Laser photochemistry   ( lcsh )
Free radicals (Chemistry)   ( lcsh )
Chemistry thesis Ph. D
Dissertations, Academic -- Chemistry -- UF
bibliography   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 1993.
Includes bibliographical references (leaves 135-139).
Statement of Responsibility:
by John Michael Nicovich.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001890695
oclc - 29642244
notis - AJW5919
System ID:

Full Text







To begin with, I must recognize the assistance, encouragement and

patience of my research advisor, Dr. R. J. Hanrahan, and my coworkers

at the Georgia Tech Research Institute (GTRI), in particular, Drs. P. H.

Wine and A. R. Ravishankara. The research reported here was supported

financially by the National Aeronautics and Space Administration (NASA)

by subcontracts through the Jet Propulsion Laboratory (JPL). Additional

support for the O(3P) + C10 study was provided by the Chemical

Manufacturers Association (CMA). Partial financial support for the

author was received from GTRI through a tuition reimbursement program

and other funds. The efficient and friendly typing service provided by

Gail Tucker is gratefully acknowledged. Finally, I thank my family for

putting up with me for all this time.



ACKNOWLEDGMENTS .................................. ii

ABSTRACT ............................................ v


I. INTRODUCTION .............................. 1


Introduction ................................. 11
Experimental Section .......................... 15
Actinom etry ................................. 18
Kinetic Measurements ......................... 24
Results and Discussion ......................... 28

Secondary Reactions that Effect [HO2] ........ 33
Comparison with Other Studies .............. 44
Comments on the Pressure Independence
of k I ....................... ... .. .... 46

RATE COEFFICIENT ......................... 49

Introduction ................................. 49
Experimental Section .......................... 50
Results and Discussion ......................... 57

Secondary Chemistry ...................... 59
Summary of Results ...................... 68
Comparison with Previous Work ............. 71
Implications for Atmospheric Chemistry ....... 74

OF THE O(3P) + CIO REACTION ................

Introduction .........
Experimental ........
Results .............
Discussion ..........
Summary ...........


. 76
. 78
. 83

............ 113


METHODS ..........

Introduction ........
Flow Systems ......
Resonance Lamps .
Fluorescence Detection
Signal Acquisition ....
UV Photometry ......
Numerical Simulations


REFERENCES ...............




............ 118


............ 133

............ 135

. . ... 140


Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy



John Michael Nicovich

May 1993

Chairman: Robert J. Hanrahan
Major Department: Chemistry

The kinetics of the important stratospheric reaction O(3P) + HO2 -.

OH + 02 (1) has been studied at 298 K as a function of N2 diluent gas

pressure from 10 to 500 Torr and also as a function of temperature from

266 to 391 K in 80 Torr of N2. Pulsed 248.5 nm KrF laser photolysis of

Og/H202/N2 mixtures produced O(1D) and OH. The O('D) was rapidly

quenched to O(3P) by N2 and HO2 was produced by reaction of OH with

H202. O(3p) was monitored by time-resolved resonance fluorescence. The

concentration of the excess species, HO2, was calculated from experiment-

ally measured parameters. This is the first study of this reaction under

pseudo-first order conditions at pressures greater than a few Torr and the

first time the described techniques have been employed for studying

radical-radical reactions.

The temperature and pressure dependence of another important

stratospheric reaction, O(3P) + C10 -* Cl + 02 (2), has been studied over

the ranges 231 to 367 K and 25 to 500 Torr. Pulsed 351 nm XeF laser

photolysis of C12/O3/N2 mixtures produced Cl in excess over 03. After a

delay sufficient for the reaction Cl + 03 C10 + 02 to go to completion,

a small fraction of the C10 was photolyzed by a 266 nm Nd:YAG laser

pulse. The decay of O(3P) in an excess of a known concentration of C10

was followed by resonance fluorescence spectroscopy. A few measure-

ments of the rate coefficient for the reaction C10 + C10 -. products were

also performed.

The rate coefficients for both reactions (1) and (2) were found to be

independent of pressure. Our results are described by the following

Arrhenius expressions in units of 10-11 cm3 molecule"1 s-1: kl(T) = (2.91

+ 0.70)exp[(228 + 75)/T] and k2(T) = (1.55 + 0.33)exp[(263 + 60)/T].

The errors are 2a and represent precision only. The results are compared

with previous measurements and the implications of these findings to

atmospheric chemistry are discussed.



The kinetics of gas phase reactions between free radicals, species

containing one or more unpaired electrons, are of importance from both

basic and practical standpoints. Radical-radical reactions often proceed

on an attractive potential energy surface with no barrier to the formation

of a bound intermediate. These reactions therefore offer relatively simple

systems that have, over the past two decades, become amenable to

theoretical studies. An in-depth discussion of the theoretical basis for the

observed kinetics of radical-radical reactions is beyond the scope of this

work but can be found elsewhere.1 From a more applied point of view,

radical-radical reactions play critical roles in complex chemical systems

such as in combustion and in the chemistry of the Earth's atmosphere.

Experimentally, the kinetics of radical-radical reactions are difficult

to study. Radical-radical reactions are usually fast and in order to study

the kinetics it is necessary to generate and quantify two reactive

intermediates. Note that under the definition given here for free radicals

there are a few species, such as NO2, NO and 02, that do not have to be

generated in situ. It is often the case that the self-reactions of the


radicals, as well as the reactions of the radicals with reaction products,

are rapid, making it more difficult to quantify radical concentrations. The

most commonly employed techniques for directly studying the kinetics of

radical-radical reactions are variations of flow-tube experiments. In flow-

tube studies, the concentration of one species is monitored at a known

distance downstream from where it is mixed with an excess of a second

species. Knowledge of the flow velocity allows the distance to be

converted to a reaction time and variation of the distance and the

concentrations leads to the kinetic information. A major advantage of

flow-tube experiments is adaptability. Many different detectors can be

used to obtain reactant concentrations and a wide variety of radicals can

be generated, often in an electrical discharge or by thermal decomposition.

Standard discharge-flow methods are restricted to pressures in the 1 to

10 Torr range, although some recent studies have reported experiments

performed at higher pressures.23 A major disadvantage is the possibility

of undetected heterogeneous effects. A more detailed description of flow-

tube experiments can be found elsewhere.4 In part, the present studies

were undertaken to develop an alternate experimental method for the

study of radical-radical reactions. Previous results on the two reactions

studied in the present work are derived mainly from flow-tube

experiments. Our adaptation of the laser flash photolysis-resonance

fluorescence (LFP-RF) technique to the study of radical-radical reactions

has allowed these two reactions to be studied at higher pressures than

before and without the same systematic errors inherent in discharge flow

studies. A discussion of the innovations in the LFP-RF technique, as well

as comparison with previous results, will appear in subsequent chapters.

The two reactions studied in this work

O(3P) + HO OH + 02 (1)

O(3p) + CO1 Cl + 02 (2)

are both of importance in the earth's stratosphere. The major

stratospheric chemistry problem relates to ozone which, along with

molecular oxygen, forms the protective ultraviolet absorption layer under

which life evolved on Earth. Our knowledge of the natural dynamic

balance between solar-initiated photochemistry, free radical reactions, and

the background gases that are precursors of the free radicals in the

stratosphere can be found described in detail in many sources.5-10 Much

of this knowledge has been gained in the last two decades from research

spurred on by the recognition that the activities of man are altering the

"natural" composition of the atmosphere. Of particular importance is the

problem of how the introduction of additional trace species to the

stratosphere changes the balance between the formation and destruction

of ozone.

The formation of a roughly constant level of ozone in the strato-

sphere was first explained by Chapman11 via the following mechanism:

02 + hv (< 190nm) 2 0 (3)

O+O 02+ M O3+M (4)

03 + hv (< 310 nm) O + 02 (5)

O + O3 202 (6)

where M represents a third-body collider (N2 and 02). This oxygen-only

mechanism predicts concentrations of odd oxygen, 03 and 0, that are too

large when compared with measured levels. It is now known that other

major loss processes are extant in the stratosphere. These include the

catalytic cycles

X + 03 -* XO + 02

XO + O-X +02

O + 03 -* 02 net

where X = OH, H, NO, Cl or Br. It should be noted that the net result of

this cycle is the equivalent of reaction (6). At lower altitudes, where the

partitioning between "odd" oxygen species (03 and 0) is shifted toward

the molecular species, the efficient ozone destruction cycle becomes:

OH + 03 HO2 + 02 (7)

HO2 + 3 OH + 202 (8)

2 03 -* 302 net

The efficiency of each cycle toward odd oxygen destruction is

dependent upon the rate coefficient for the rate-determining step in each

cycle and the coupling with reactions that remove the catalytic radicals.

Examples of the latter include

Cl + CH4 HC1 + CH3 (9)

and OH + HO2 H20 + 02 (10)

The major hydrogen-oxygen free radicals, OH and HO2, play critical

roles in the chemistry of the stratosphere either in catalytic cycles or

through reactions that partition other species between free radicals and

chemically stable forms. The dominant source of HO. radicals throughout

the stratosphere is the reaction

H20 + O('D) 2 OH


where the excited oxygen atoms are formed by the photolysis of 03.

Reaction (1) is the rate-determining step in a catalytic cycle that leads to

03 destruction

OH + 03 -HO2 + 02 (7)

O + HO2-* OH + 02 (1)

O + O3 202 net

Figure 1 (taken from reference 12) shows the conversion rate between OH

and HO2 as a function of altitude for all major reactions thought to

interconvert these species. Note that reaction (1) is the dominant

reaction converting HO2 to OH above 40 km, i.e., in the upper


In 1974, Stolarski and Cicerone13 suggested that chlorine introduced

into the atmosphere in rocket exhaust could destroy 03 by another

catalytic cycle

Cl + 03 C10 + 02 (12)

0 + C10- Cl + 02 (2)

O + 03 -* 2 02





U) 0

+U C+J + C
0 0

V3~~0 0 mco )C

= = C4cq 7c


o o

(WN) 3 niii-l

In this cycle, reaction (2) is rate-determining. Subsequently, Molina and

Rowland14 identified chlorofluorocarbons (CFCs) as another, much larger

anthropogenic source of stratospheric Cl. In general, CFCs are defined

as hydrocarbons in which all the hydrogen atoms have been replaced by

either chlorine, fluorine or a combination of both. Since World War II,

CFCs have been produced for use in a variety of industrial applications

such as solvents and refrigerants. Relatively inert to oxidation or other

loss processes in the troposphere, the CFCs mix into the stratosphere

where their ultraviolet photolysis produces free chlorine atoms.

Approximately 80% of the present atmospheric load of chlorine is from

CFCs with the remainder coming from naturally occurring CH3C1.

Recent measurements have clearly shown that indeed the level of

stratospheric 03 has been negatively perturbed. Trend estimates in the

loss of ozone in the mid-latitudes as measured by several methods and

predicted by model calculations are shown in Figure 2 (taken from

reference 15). The measurement techniques include ground based

(Umkehr), balloon-borne (Payerne and Hohenpeissenberg), and satellite-

borne instrumentation (SAGE). As can be seen in the figure, the

measured loss of 03 is greatest in the lower stratosphere between 15 and

20 km. More dramatic losses in the ozone column have been observed for

the past several years over the continent of Antarctica in the austral

spring. Heterogeneous processes that take place on polar stratospheric

E 0 3o

m m>
C o
Sa ~

\-- (N 'S-S

I /.
"" ,' t o g o
_ (1 a CS "- _. +-

o ba

er) 4, .--
S) 00o8
o o 0

0 4i

) I E-I 8

t LO 00 -

a I 0 o C)

a 0 1.0


clouds freeing highly photoactive chlorine species (HOCI and C12) followed

by reactions of the oxides of chlorine are thought to be responsible for the

so-called Antarctic "ozone hole."

The specific roles of reactions (1) and (2) in the chemistry of the

Earth's atmosphere are discussed further in the following chapters. The

study of the room temperature rate coefficient for the 0 + HO2 reaction

is presented in Chapter II, followed by the temperature dependence study

of the same reaction in Chapter III. The study of the temperature

dependence of the 0 + C10 rate coefficient is described in Chapter IV.

Except for minor editorial changes necessary to integrate the three

separate studies into a single thesis, the material in each of these chapters

is taken, by permission, from three separately published journal

articles.16-18 A concluding chapter contains a summary of the work

presented here as well as a discussion of pertinent information that has

appeared in the literature subsequent to the publication of the three

studies. Finally, there are two appendices. The first contains additional

details of the experimental and calculational techniques and the second

appendix lists, in numerical order, the complete set of reactions.




Bimolecular reactions involving two free radicals are of great interest

because both reactants have unpaired electrons and hence could interact at

distances longer than those typical of radical-molecule encounters. Also,

because of the attractive nature of the encounter, the energy barriers for the

reaction can be nonexistent or small; i.e., the reaction can proceed on a purely

attractive potential surface. Consequently, association reaction channels can

become possible even when there exist other pathways such as simple atom

transfers. The majority of radical-radical reactions have, until now, been

studied using the low pressure discharge flow technique4 which is extremely

well suited for these types of reactions since the two free radicals can be

created in physically separate regions and then mixed to observe the reaction

of interest. However, the pressure range of discharge flow studies is presently

limited to a few Torr. (Currently, a few laboratories are developing high

pressure discharge flow apparatus.) Therefore, it is obvious that if the

association pathways in radical-radical reactions are to be observed, especially

when they are in competition with other pressure independent channels, these

reactions will have to be studied at much higher pressures.

In addition to the above mentioned rationale for studying radical-radical

reactions at high pressures, there is also a more practical reason for such

studies. Most radical-radical reactions that are important in the atmosphere

and in combustion systems take place at pressures much greater than a few

Torr. If one intends to model such complex systems, it is necessary to ensure

that the input rate constants are those applicable under the pressure conditions

actually encountered. Also, one cannot ignore the possibility of surface

enhanced reactions influencing the results of discharge flow studies even

though, in principal, experimental checks for such effects can be made. Lastly,

multiple studies using vastly different techniques are necessary to recognize

and then (hopefully) minimize systematic errors. The last point is especially

important since a great deal of reliance is placed on the absolute accuracy of

kinetic data in modeling atmospheric and combustion processes.

In the past, a few high pressure studies of radical-radical reactions have

been carried out. However, except for the cases of self-reactions, i.e., R- + R"

- products (where R- = free radical), most of these studies have not been able

to isolate the reaction of interest and therefore had to rely on modeling a

complex scheme of reactions to extract the rate coefficient of interest. To

worsen the situation, in many cases the modeling had to include reactions

which were themselves not well understood. In some of these studies the

modeling calculations showed that the experimentally measured parameters

were not even very sensitive to the value of the rate constant of interest.

Therefore, it is clear that experimental techniques which can directly measure

radical-radical reaction rate constants at high pressures are needed.

In our laboratory, we are developing a method based on laser photolysis

to selectively produce free radicals in the homogeneous gas phase in such a way

as to isolate the reaction of interest and subsequently follow the course of the

reaction using spectroscopic techniques. This method has become feasible

because of advances in laser technology as well as improvements in our

understanding of the photochemistry of precursors and kinetics of reactions

from which the radicals of interest are generated. The present paper describes

the first of these studies where the rate coefficient for the reaction of O(3P)

with HO2,

O(3p) + HO2 OH + 02 (1)

has been measured at N2 pressures ranging from 10 to 500 Torr.

Reaction (1) was chosen for the initial study for the following four

reasons: (a) there is only one set of products possible for this reaction, at least

at low pressures; (b) the reaction could be studied using only one photolysis

laser; (c) there have been two recent studies of reaction (1) at low pressures19,20

which are in agreement (see below) and thus provide a good basis for

comparison; and (d) the chemistry in the system could be kept under very good

control. Based on the above description, one might get the impression that this

technique is of limited use. However, it will, in time, become evident that with

carefully thought out systems and further improvements in our understanding

of the photochemistry of important radical precursors, this technique will prove

to be extremely powerful.

Reaction (1) has been studied at 298 K by many investigators.19-23 The

measured values of k1 range from 2.7 to 7.0 x 10-11 cm3 molecule-1 s-1. All

studies,19-23 except one,23 have been carried out using discharge flow methods

of pressures lower than 4 Torr. Some of these investigations21,22 were indirect,

i.e., kI was measured relative to other rate coefficients. The rate constants for

the reference reactions themselves have recently been revised.24 The high

pressure study23 was very indirect and involved the pulse radiolysis of a

mixture of 02, H2, and Ar followed by extensive modeling to extract k1. The

most recent low pressure studies by Keyser19 and Sridharan et al.,20 however,

were carried out such that reaction (1) was isolated and k1 measured under

pseudo-first order conditions to be (6.1 0.4) x 10-11 and (5.4 + 0.9) x 10-11

cm3 molecule-Is-1, respectively. Keyser has also measured k, as a function of

temperature and obtained a small negative activation energy for reaction (1).

In the study described here, 0(3P) and HO2 were produced by cophotolysis

of 03 and H202 in N2 at 248.5 nm using a KrF excimer laser. Reaction (1) was

studied under pseudo-first order conditions in [O(3P)] with [HO2] > [O(3P)].

[HO2] was calculated from the knowledge of various parameters. The rate

coefficient for reaction (1) was measured at seven different total pressures

between 10-500 Torr where the bulk of the gas was N2 and kI was found to be

independent of pressure with an average value of (6.2 1.1) x 10-11 cm3


Experimental Section

Reaction (1) was investigated at high pressures (10-500 Torr) where the

reaction was isolated and was studied under pseudo-first order conditions in

[O(3p)] with [HO2] in excess over [O(3P)]. The experimental approach used in

this investigation involved 248.5 nm KrF excimer laser photolysis of a mixture

of H202 and 03 in N2 (or Ar) to produce known concentrations of HO2 and

O(3P). The temporal profile of [O(3p)] was subsequently followed using time

resolved resonance fluorescence detection. The cophotolysis of H202 and 03

in excess N2 at 248.5 nm produces HO2 and O(3P) via the following reactions:

H202 --> 20H (13)

03 248.5 > O(1D) + 02(alAg) (14a)

> O(3P) + 02(X3yg-) (14b)

OH + H202 -> HO2 + H20 (15)

O('D) + N2 0(3P) + N2


[O(3p)] decays due to the following reactions:

O(3P) + HO2 -* OH + 02 (1)

O(3p) + H20 > OH + HO2 (17a)

-p H20 + 02 (17b)

O(3p) loss from detection zone due to diffusion
and reaction with background impurities (18)

The concentration of HO2 was not directly measured but was calculated

based on experimentally measured and other known parameters. The criteria

for producing a known concentration of HO2 are listed below and it is worth

pointing out that the accuracy of the measured value of kI directly depends on

how well each one of these criteria is met: (i) the concentration of the

photolytic precursor [H202] should be measured in the system; (ii) the

absorption cross section of H202 at the photolysis wavelength a(H202248.5

nm) should be known; (iii) OOH, the quantum yield for the production of OH,

the precursor of HO2, should be known at the photolysis wavelength; (iv) F, the

laser fluence should be measured; and (v) the laser fluence should be "spatially

uniform." If all these criteria are met, since the concentration of H202 was

such that the system was optically thin at 248.5 nm, the concentration of OH

formed can be calculated using the relation,

[OH]o = F x [H202] x u(H202, 248.5 nm) x (OH

The concentration of H202 was directly measured using absorption of

202.6 nm zinc lamp radiation in a 76 cm long cell which was traversed by the

reaction mixture after it exited the reactor. The absorption cross section of

H202 at 202.6 nm24 is 4.30 x 10-19 cm2. The absorption cross section of H202

at 248.5 nm24 is 8.63 x 10-20 cm2 and the quantum yield for OH production

from H202 at 248.5 nm25,26 is 2. In addition, it has been shown that OH is

produced in the electronic and vibrational ground state.27 In the presence of

excess N2, the rotational and translational temperature of OH would be relaxed

to that of the bath gas within a microsecond, a period very short compared to

the time scale of the experiments.

Spatial uniformity of laser beams is not easy to obtain. The beam has to

be reproducibly uniform for each pulse and not so merely on the average. If

there is a spatial gradient (as in the case of a Gaussian beam) or if there are

regions of high fluence (hot spots) then reaction (1) would proceed at different

rates in different parts of the beam. The O(3P) detection zone is of finite size

(~ 2 cm3) and hence the measured O(3P) decay rate will be an unknown

weighted average of the rates of reaction (1) in the various volumes in the

detection zone. Moreover, since one cannot directly measure the fluence in the

detection zone itself, if the beam were nonuniform, then the fluence measured

outside the reactor could be an erroneous representation of the fluence in the

detection zone. To overcome all of these problems we used a device known as

a segmented aperture optical integrator to make the photolysis beam spatially

uniform. The operation of this device and the extent of uniformity of the

integrated beams has been described in an earlier publication from our

laboratory.28 The spatial uniformity of the beam used in the present

investigation was better than 5% over its entire cross section. The beam was

made to essentially fill the reactor, and the laser fluence was constant over the

entire (-20 cm) length of the reactor.

The laser beam fluence was measured as the beam exited the reactor using

an EG and G photodiode based radiometer capable of measuring individual

pulses. Even though the calibration of this instrument is traceable to NBS

standards, we carried out actinometry experiments to check the calibration.

These actinometry experiments are described below.


Preliminary experiments using conventional actinometry techniques such

as the aqueous phase ferrioxalate method or gas phase ozone photolysis

followed by end product analysis showed very poor precision as well as fluence

dependence. Therefore, it was clear that such methods are ill suited for

measuring fluence of high powered, short pulse width lasers. To overcome this

problem, we employed a very simple method where we measured the time

resolved loss of 03 due to photolysis under geometrical conditions identical to

those employed in the actual kinetics experiments. The experimental set up is

shown in Figure 3. A 5 cm diameter cell equipped with 2" quartz windows on

opposite faces to transmit the photolysis beam and 1" windows on tubes

attached to the sides of the cell to transmit the analysis beam (253.7 nm) at 900

to the photolysis beam was used. A mixture of 03 in N2 was slowly flowed



Hg LAMP n, ,-

L: -lj





A schematic diagram of the apparatus used for
actinometry experiments involving ozone loss
measurements in real time. The gas mixture containing
ozone and N2 were premixed from measured flow rates
and entered at the port marked gases in. The pressure
in the system was measured at the port marked gases
out. The hatched area represents the 248.5 nm
photolysis beam of square cross section from the
segmented aperture optical integrator. The dashed line
represents the 253.7 nm Hg radiation, L is the distance
the 253.7 nm beam traverses in the path of the
photolysis beam and L' is the length over which ozone


Figure 3.

through the cell. The 2" faces were covered by a precisely cut 3 cm x 3 cm

aperture to clearly define the area of the photolysis laser beam which traversed

the cell. A beam of 253.7 nm radiation from a mercury pen ray lamp run by a

regulated dc power supply traversed the cell through the 1" windows, thereby

crossing the 3 cm wide photolysis beam. It was passed through a

monochromator and detected by a quartz enveloped 1P28 photomultiplier tube.

The output of the photomultiplier tube was monitored using a signal average

operating in the peak height analysis mode.

The transmitted 253.7 nm light intensity Io was measured with the cell

flushed with N2. Then a constant amount of ozone was added and the

transmitted intensity I' measured. The ratio I'/Io was used to calculate the

concentration of ozone flowing through the cell. The photolysis laser was

turned on and the integrated beam allowed to photolyze the ozone in the cell.

The area of cross section of the beam was adjusted to fill the 3 cm x 3 cm

aperture in front of the cell, thereby photolyzing 03 in an identical area inside

the cell. The signal average was pretriggered and the transmitted 253.7 nm

light intensity monitored. The flow rate of the OgfN2 mixture through the cell

was such that the cell was completely replenished between laser flashes

(repetition rate = 0.1 Hz). The observed temporal behavior of the transmitted

253.7 nm light intensity is shown in Figure 4. Depending on the signal level,

64 to 500 flashes were averaged. As seen in Figure 4, synchronous with the

laser pulse there is a sudden jump in the transmitted intensity to I, followed

by a slower increase to a maximum 12. Following this maximum, the intensity






I ~


0 5

msec msec sec


The temporal profile of the intensity of the 253.7 nm
beam transmitted through the cell including the photoly-
sis area. Io is the intensity with no ozone in the cell. I'
is the intensity after ozone has been added and its
concentration has equilibrated. Notice the discontinui-
ties in both Y and X axes. The arrow marks the time at
which the photolysis laser was fired. I1 and 12 are the
intensities immediately after photolysis and after all
chemistry has stopped. The decay from 12 to I', in the
time scale of 0-8 s, is due to refilling of the cell. For the
next laser pulse, the sequence starts at time 0. Ii/I'
yields the initial ozone loss and consequently the fluence
while I'/Io yields the concentration of ozone in the cell.

Figure 4.

4 8

. ----- 2

slowly decreases back to I' as the contents of the cell are swept out and filled

with a fresh mixture.

The interpretation of the above result is straightforward. The sudden

jump is due to the photolysis of 0329 to produce O(ID) + 02(alAg) and 0(3p)

+ 02(X3:). The primary 03 loss is assumed to occur with a quantum yield

of unity.30 The slower ozone loss is due to the following chemistry:

03 + hv > O(1D) + 02(alAg) (14a)
O 0(3P) + 02(X31g) (14b)

O(1D) + N2 -* O(3P) + N2 (16)

O(3P) + 03 2 (6)

02(alAg) + 03 -* 0(3P) + 202 (19)

It has been shown that 03 photolysis at 248.5 nm leads to both O(1D) +

02(alAg) and 0(3p) + 02(X31-).29 The ratio of the quantum yields of 0(3p)

and O(1D), cO(3P)/qO(1D) has been shown29 to be -0.1. Therefore, the net

loss of 03 due to the secondary chemistry relative to the primary process, i.e.,

reaction (14) should be -2.8. The value we obtained is 2.9 + 0.4 which

confirms that the system is behaving as expected.

The primary quantity of interest, of course, is I,/I'. This quantity is

directly related to the photon fluence in the monitoring zone of the cell by the


ln(I1/I') = [O311ot x U(03,253.7 nm) x L (II)

where [Og311ot is the decrease in [03] due to photolysis, U(03,253.7 nm) is the
03 absorption cross section at 253.7 nm taken30 to be 1.12 x 10-17 cm2, and L
is the width of the photolysis laser beam (3.0 cm). Measurement of [O3]lost
leads to the photon fluence F via the relationship,

F = (III)
o(03,248.5 nm) x [03]

Simultaneous with these measurements, the transmitted laser fluence F'
(corrected as described below) was monitored by the EG and G radiometer
masked by a known area (1.32 cm2). The ratio of F/F' was the quantity we
were after since using it the photon fluence in the kinetics experiments could
be directly calculated.
Four runs with varying amounts of 03 and fluence were carried out. The
average measured value of F/F' was 1.04 + 0.08 where the error is 2a. (F was
corrected upwards by 7.6% for the light reflected back into the cell by the back
window and F' corrected by 8% for the loss due to transmission through 03
and the back window.) In all kinetics experiments the fluence measured by the
radiometer was corrected to take the actinometry results as the primary
method for fluence measurements.

The elegance of this method is clearly shown by substituting relation II
into III and replacing [Og] by In(Io/I')/{a(O3,253.7 nm) x L'}, to yield,

In(I1/Io) x L'
F= (IV)
a(03,248.5 nm) x L

I,, Io, L', and L are all directly measured very accurately. (It should be
noted that the spread in measured intensity was -0.005%). The only quantities
whose accuracies need to be relied on are a(03,248.5 nm) and the quantum
yield for ozone dissociation. Both of these quantities are undoubtedly accurate
to better than 5% each; this leads to an accuracy of 11% in measuring F.
Combining this with possible errors due to slight nonuniformities in the laser
beam yields an accuracy of -12% for the photon fluence measurements.
Kinetic Measurements
A schematic diagram of the apparatus used for carrying out kinetic
measurements is shown in Figure 5. The reaction cell was 20 cm long, 3.8 cm
i.d., and made of brass. Its inside was coated with FEP teflon and over coated
with halocarbon wax. A mixture of H202, 03 and N2 was flowed through the
cell. The H202 content of this mixture was measured via its absorption of
202.6 nm Zn+ radiation from a Zn lamp in a 76 cm long absorption cell through
which the reaction mixture flowed after exiting the reactor. In a previous study
in our laboratory31, it was shown that H202 does not decompose to a
measurable extent in this reactor. 03 does not interfere with the H202
measurement since (a) its concentration is very low and (b) the absorption cross
section of 03 at 202.6 nm is quite small.30 All gas transfer lines between the
source mixtures and the reactor were made of teflon connected by a few
stainless steel cajon fittings to minimize H202 decomposition. The H202/N2



Iu I -

CC _j
p< m

C 1- Uj




)- -
0 g --
c l^
0 c

+a a
0 *>



a a)

U >

w) o



o u

a) *- a
2o <

0n 0

t) a) 0)

00 "0S

0 4|


rI4.-A w


mixture was generated by bubbling N2 through a bubbler containing H202.
Ninety percent pure H202 (from FMC corporation) was used after N2 had been
bubbled through it for 24 to 36 h to remove H20 and 02.
The 03/N2 mixture was stored in a 12 liter bulb. The concentration of 03
in this mixture was measured using 253.7 nm absorption. The 03 was prepared
by passing 99.99% pure 02 (Matheson Gas Products) through an ozonizer and
collected on silica gel at 195 K Before use ozone was liquified at 77 K and 02
pumped out. N2 (99.9995%) and Ar (99.9995%) were both purchased from
Matheson Gas Products and used as supplied. The pressure in the system was
measured as the mixture exited the reactor using a capacitance manometer.
The 03/N2 mixture and pure N2 were flowed through mass flow meters and
their flow rates controlled by needle valves. Since H202 will decompose if
passed through flow meters, a measured amount of N2 was bubbled through the
bubbler containing H202 and the mixture thus generated was added to the
03/N2 stream flowing into the reactor.
The reaction cell was equipped with four ports in its middle. Two ports
were used to transmit the output of an oxygen resonance lamp which was
filtered by a CaF2 window to remove Lyman-a radiation. The resonance
fluorescence excited by the resonance lamp beam was collected at 900 by a
MgF2 lens and focused onto the photocathode of an EMR-542G solar blind
photomultiplier tube. The output of the PM tube was amplified, discriminated
against noise, and fed into a signal average operating in the multichannel
scaling mode.
A typical kinetics experiment consisted of flowing 0.3-1 x 1013 cm-3 of 03
in 10-500 Torr of N2 through the reaction cell and the absorption cell. Jo, the
intensity of the 202.6 nm Zn+ line transmitted through the absorption cell was
measured. Next, the H202/N2 mixture was introduced into the stream of 03/N2


mixture, and, after equilibration of H202 with the walls, J, the transmitted
202.6 nm light intensity, was measured. Using the values of J and Jo, [H2021
was directly calculated; it ranged from 0.3 to 2 x 1016 cm-3. J was continuously
monitored throughout the course of the experiments. The signal average was
pretriggered 10 ms before the photolysis laser was fired to obtain the
background count rate and the delay was precisely controlled by a delay
generator. The photolysis beam, which had been made spatially uniform by the
optical integrator, entered the reactor and initiated the chemistry. [O(3P)] was
monitored as a function of time. Typically, 16 to 512 individual O(3p) temporal
profiles were averaged to obtain one curve. During the averaging period, the
fluence of each pulse of the photolysis beam was measured after the beam
exited the reactor, using the (calibrated) E G and G radiometer. The laser
fluence was then varied to produce a different concentration of HO2. Typically,
five to eight fluence values were used to create 1-10 x 1012 cm-3 of HO2. After
this series of runs were completed, the H202 flow was turned off and Jo
remeasured. It should be noted that in these series of runs, for each value of
HO2, the ratio of [HO2]o/[O3P)]o is not altered. To change this ratio, the
composition of the mixture was varied.
During the course of each kinetic run, it is necessary that the fluence (per
pulse) be constant. The KrF excimer laser used in the present experiments was
equipped with a gas processor which continuously circulates the gas mixture
through the laser cavity, cleans up the mixture, and replenishes F2 such that
the laser energy is extremely stable. The measured peak to peak fluctuations
were typically -2%, and the average of 10 to 512 pulses much more stable.
This stability made it unnecessary to include an energy monitoring device into
the circuitry triggering the signal average to discard any individual temporal

profile obtained for a given laser pulse which had not met the energy
Results and Discussion
All experiments were carried out under pseudo-first order conditions in
[O(3P)] with [HO2] in excess over [O(3P)]. As will be pointed out later, in this
particular system it is not essential to keep [HO2] much greater than [O(3p)].
If the postphotolysis chemistry is governed only by reactions (15) (18) and
reaction (1), then the temporal behavior of [O(3P)] (after all OH has been
converted to HO2) is given in Equation (V).

-- = (kj[HO2] + k'd)[O(3p)] (V)

where k'd = k17[H202] + k18. For constant concentrations of H202 and HO2,
Equation (V) can be integrated to yield

[O(3P)] = [O(3P)]oexp{(-ki[HO2] + k'd)t} (VI)

Figure 6 shows plots of ln[O(3p)] versus time for three different values of
[HO2] at fixed values of [H202] and [03]. It is clear that Equation (VI) is
obeyed. The slope of these lines yield k' = kj[HO2] + k'd. Figure 7 shows a
plot of k' versus [HO2] and the slope of this line yields k1. The value of k1
measured using N2 as the diluent gas under various experimental conditions
are listed in Table I. Each of these values were derived by measuring at least
five different k' values.





0 10
TIME, msec

Temporal profile of O(3p) following the photolysis of a
mixture of H202, 03 and 100 Torr N2 at 298 K. The
concentrations of HO2 were varied by changing the
fluence, and are shown next to each curve. For the sake
of visual clarity, the curves are displaced on the Y axis.
Different numbers of traces were averaged for each
curve. The slope of each line gives k', the measured
pseudo-first order rate constant.


Figure 6.


0 5

EH02J,10'2 cm-3

Plot of k' (the measured pseudo-first order rate
constant) versus [HO2] at 100 Torr N2 pressure and 298
K. Each [HOg] has been corrected for the occurrence of
the OH + HO2 reaction as described in the text. The
slope of the line yields kI. The intercept represents kd
= k18 + k17 [H202].

Figure 7.



*- N N N C
0 '-4 l- -4 r-4 -
S+1 +1 +1 +1 +1 +1 +1
ht 'd to (0 -4 (0 0R

G 00 a0 10 O t- Q00 N
d3 tod 4 t o o q N
0 -0 Q Q 0 0
+ +1 +1 +1 +1 +1 +1 +1
t- 1 0 = N t- t- 1-4
v o o o -4 o C- 00
00 g C^i 0O '.0

0 <<_________________________________

0 -V Go to toc o c wt 2t 0 0 w
0RC Cl* *^l L- wl cq t-o ic nCD-S tO'- c'v a-' t- oo LOm O q a

+1 +1 +1 +1 +1I +1 +1+ ++ I +1 +1 +1+1 T +1 +1 +1+1
4) 0 t I"$ O v WO cc m o '4'00 Ct-t- IV eq t- C- toto
S 00 co 00-4 -W m 00 =00to v-4COINO00 0 eq 00
9. t6 C t to C0i t Os. L6 to6t

r-4 t 10)
+1 +1 +1

to tor ko
4-I (D i

eq0000 -
eqeqcoto 5

+1 +1 +1 +1
00 0 vco
0 LO to c?

+1 +1 +1 +1I
idi idi

eso eq f
+1 +1 +1
1q 04 10
ko LO Od

+1 +1
04 00

.- 0

tPHo V. a o 0 mN m 0 t- 0 0

S-ito 0000 to U5 Go cc-toCD wIt- to goco to i CD




0 g
0 a.
0 ON


eq '

-4 o)fl ^
.g o

"S^ 3


One series of experiments was carried out with 100 Torr of Ar as the
diluent gas. In these experiments, a large fraction of the O(1D) produced by 03
photolysis reacted with H202 to produce HO2,

O(1D) + H202 OH + HO2 (20)

OH + H202 -* H20 + HO2 (15)

This HO2 production could not be helped since argon is a very poor
quencher32,33 of O(WD) and hence reaction (20) was competitive with the
quenching. The extra production of HO2 (2 HO2's for each O(1D)) accounted
for -12% of the net HO2 produced. The value of k1 obtained in 100 Torr of Ar
(after correcting for certain reactions affecting [HO2]o which will be discussed
later) is (6.27 + 0.68) x 10-11 cm3 molecule-1 s-1.
It is necessary to mention that some preliminary measurements of k1 were
carried out using 266 nm photolysis (4th harmonic Nd:Yag) in 75 Torr of N2.
These experiments yielded a value of (7.96 + 0.65) x 10-11 cm3 molecule1s-1,
(errors are 2a and do not include systematic errors) which were reported at a
recent meeting.34 Since then we have discovered that two corrections have to
be made to this number, both due to the measured fluence. First, the
actinometry experiments decrease this value of k1 by 4%. Second, we have
neglected to correct for the loss in measured fluence due to the back window
as well as the increase of the fluence in the reaction zone due to the reflection
off the back window; this further decreases the value by -16%, yielding k1 =
(6.5 + 0.6) x 10-11 cm3 molecule-ls-1 in reasonably good agreement with our
other results.


Secondary Reactions that Effect [HO21
As mentioned earlier, during the course of this work [HO2] was not
directly measured but calculated. Therefore, a great deal of attention was paid
toward recognizing, eliminating, and when unable to eliminate, accounting for
HO2 production or depletion reactions. Each of the reactions that can alter the
calculated [HO2] is considered and its possible effects on the measured value
of k1 are discussed in this section.
One of the nice features of the chemical system used in this study is that
HO2 should be regenerated after consumption in reaction (1):

O(3) + HO2 -* OH + 02 (1)

OH + H202 -* HO2 + H20 (15)

To check the effectiveness of the regeneration, a few experiments were carried
out with [HO2]o/[O(3P)] ratio as low as 3. Indeed, the O(3p) decay plots were
exponential and the value of k1 thus obtained was, within experimental error,
identical to those measured at higher values of this ratio (see Table I).
However, most of the experiments were carried out with [HO2]o/[O(3P)]o
greater than 10, with an average value of 15.
Even though at first glance, looking at the exponential decays of O(3p)
shown in Figure 6 and the linear variation of k' with [HO2]o shown in Figure
7, it seems that the system is well behaved, a detailed consideration of the
chemistry in the system shows that there are small contributions by some
secondary reactions that do effect the calculated HO2 concentration and they
cannot be ignored. The contributions of these secondary reactions were

calculated and the measured value of k1 corrected for these complications. The
corrections due to the secondary reactions are individually discussed below.
OH + HO2 reaction. During the course of the reaction, OH was generated
in reaction (1) and consumed by reaction (15). The steady state concentration
of OH in the system was typically 2% of the [O(3p)] and therefore no error was
introduced due to the occurrence of the OH + HO2 reaction,

OH + HO2 H20 + 02 (21)

However, immediately after the creation of OH by H202 photolysis the
concentration of OH free radicals was very high and the contribution from
reaction (21) cannot be ignored. The consequence of reaction (21) is that some
of the OH reacts with HO2 and not with H202, thereby leading to an [HO21o
value which is less than that calculated by neglecting reaction (21). To take
this error into account, a numerical integration of the kinetics in the post-
photolysis chemistry was carried out and the [HO2] that is actually produced
was thus calculated. The integration technique is discussed in the final section
of Appendix A. The following two reaction scheme was used in the calculation:

OH + H202 HO2 + H20 (15)

OH + HO2 -* H20 + O2 (21)

The recommended value24 of k21,

k21 = (7 + 4 Ptm) x 10-11 cm3 moleculels-1

where Patm is pressure in atmospheres, was used. The integration showed
that [HO2]o calculated assuming reaction (21) to be negligible was as much as
8% higher than that actually produced in some of our experiments. To check
for this effect, one series of experiments were carried out where -5 x 1015 cm-3
of H202 was photolyzed at various fluence values to produce different ratios of
[H202]/[OH]o, some as low as 800. Figure 8 shows a plot of measured k' versus
calculated [HO21o (open circles). It is clear that the plot deviates from a
straight line at higher [HO2] (i.e., lower [H2021/[OH]o). However, when
allowance is made for reaction (21), as estimated by the integration scheme, the
plot is linear. Each value of [HO2] (in the kinetic runs) was corrected to allow
for the occurrence of reaction (21). The corrections to [HO2] varied from 0.5%
to 8% and was on the average -3%. It should be noted that the corrections
were larger for higher [HO2] since it was at these values that the [H2021/[OH]o
ratios were lowest. The contribution due to reaction (21) could be made
negligible by making the H202 concentration very large. However, three
factors discouraged such increases. First, H202 absorbs 130.3 nm oxygen
resonance radiation and increasing [H202] beyond -2 x 1016 cm-3 resulted in
prohibitively degraded sensitivity for oxygen atom detection. Second, the loss
of O(3p) due to reaction with H202 (reaction 17) would also increase, thereby
making k'd large. Third, the fraction of O(1D) reacting with H202 relative to
that being quenched by N2 would increase. Therefore, [H202] was always held
at less than 2 x 1016 cm-3.
O(D) + H2gO reaction. 03 photolysis at 248.5 nm produces O(1D) 90%
of the time.29 In the presence of a large concentration of N2 the majority of
O(ID) is quenched to O(3p). However, a small fraction of O(1D) can react with
H202 to produce two molecules of HO2 for each O(1D) undergoing such a
reaction. The extent to which reaction (20) can contribute towards the net





[H021,1012 cm-3

Figure 8.

Plot of k' versus [HO2] at 250 Torr and 298 K. The
open circles represent [HO2] uncorrected for the
occurrence of the OH + HO2 reaction. The filled circles
were obtained by correcting the HO2 concentration as
described in the text. The slope of the line yields k1.

HO2 produced depends on the ratio [H202]/[N2] as well as the ratio

[O3]/[H202]. In all experiments where N2 was the diluent gas, except those
with 10 Torr of N2, [HO2] produced via the reaction sequence (20) followed by
reaction (15) could be made less than 0.05% of that produced via H202
photolysis. In the 10 Torr experiments, however, the [HO2] produced by O(1D)
reaction ranged from 0.6% to 2% of the total and this contribution was taken
into account. The worst situation was in the case of one series of experiments
where argon was used as the diluent gas as discussed earlier. This reaction can
be made negligible by working with high [HO2]/[O(3P)]o as was the case in
most experiments.
O(3p) + H2OQ reaction. The reaction of O(3P) with H202,

O(3p) + H202 > OH + HO2 (17a)
> H20 + 02 (17b)

can also increase the [HO2] during the course of reaction (1). However, this
reaction is very slow, k17 = -1.45 x 10-15 cm3 molecule-'1s1 31; so within the
time O(3p) decays to 10% of its initial value, the maximum amount of HO2
produced would be less than -2% in the worst case, i.e., higher [H202] (-2 x
1016 cm"3) and the lowest measured decay rates (-70 s-1). Therefore, the

contribution due to this reaction was neglected.
HO2 reactions with impurities. HO02 is a reactive free radical and hence
can react with impurities in the system thereby leading to an unrecognized
depletion of its concentration. Keeping in mind that the only diagnostic at our
disposal is the measurement of O(3p) temporal profiles, in order to check for
loss of HO2 because of reactions with impurities in the system, we varied the
[H202] while creating the same initial concentration of HO2 by correspondingly


varying the fluence. The measured O(3p) decay rates were not affected thereby
indicating the lack of impurities coming in with H202. The HO2 + O3 reaction
is too slow to deplete [HO2] and hence can be ignored. Variations in [03] had
no effect on measured O(3p) decay rates which again indicated the
unimportance of HO2 reaction with 03 as well as with any impurities coming
in with 03. To check for impurities being swept in with the diluent gas N2 the
flow rate of the gas mixture through the cell was varied, and found to have no
effect. This test, of course, does not eliminate the presence of impurities in N2
itself. We used N2 of the highest available purity and it seems unlikely to
contain any reactive impurities. In addition, the measured value of k1 was
independent of N2 pressure from 10 to 500 Torr. (Of course, it is possible that
there can be an accidental cancellation if k1 increases with pressure and
compensates for decreases in [HO2] due to reaction of impurities. This
possibility seems highly unlikely.)
HO + HO2 reaction. During the course of the kinetic measurements,

many observations indicated the importance of the HO2 + HO2 reaction. When
the O(3P) detection sensitivity was high (i.e., when all windows were clean and
the resonance lamp tuned for minimal line reversal), it was clear that O(3P)
decays were getting slightly nonexponential at longer reaction times with the
decays getting slower as reaction (1) proceeded. The observed reaction time
where this departure was evident always occurred after two 1/e times and was
independent of [HO21o. Also, these deviations were slightly more pronounced
at higher pressures of N2 than at lower pressures. Such a decrease in decay
rates of [O(3p)] can be due to either the HO2 concentration decreasing as
reaction (1) proceeds or due to generation of O(3P) via secondary reactions. As
will be discussed later, the second possibility can be ignored.

To take HO2 loss via the self-reaction into account we derive the following
expression for the 0(3P) temporal profile based on reactions (1), (17), (18), and

O(3P) + HO2 -* HO + 02 (1)

HO2 + HO2 -* H202 + 02 (22)

O(3P)--> loss due to diffusion and reaction (17).

[O(3P)] k,
In = ln(1 + 2k22[HO2]ot) + kd't (VII)
[O(3P)]o 2k22

The logarithm can be expanded into a series since 1 < 2k22[H02]ot < 1, and
the cubic and higher terms neglected to obtain:

In P- = kj[H02]ot x (1 k22[HO2]ot) + k'dt (VIII)

If [HO2] is indeed time independent and equal to [HO2]o, then a plot of
ln[O(3p)] versus time should be a straight line and its slope should equal
(kl[H021]o + kd'). In Figure 9, we have plotted In([O(3p)]/[O(3p)]o) versus time
(curve a, open circles) for one experiment at 250 Torr N2. It is seen that up to
two 1/e times the decays do look exponential as does the data shown in Figure
6. However, at longer reaction times, the curve deviates from a straight line.
When each [O(3P)]t is corrected for the depletion of [HO2] using expression













<^0 sba--2

o --3 0
0 L-J
0 00 0 0

5 10

(a) Plot of ln[O]/[O]o versus reaction time (open circles).
The scale is on the right as shown by the arrow. The
solid line is the least squares fit of this data up to two
1/e times (i.e., down to ln[O]/[O]o = -2). The slope of
this line is (424 + 20)s-1 where the error is 2a. Notice
the deviation of the points from this line at reaction
times longer than -6 ms, (b) Plot of (ln[O]/[O]o)/(1-
k22[HO2]t) versus reaction time t. This correction
accounts for the loss in HO2 due to its self-reaction
during the course of reaction (1). The scale is on the
left. The solid line is the least squares fit of the data up
to 10 ms. The slope of the line is (453 + 13) s-1 where
the error is 2a. Notice the linearity of the points. The
curve a has been shifted up by 1 ms for the sake of
visual clarity.


Figure 9.

VIII and we plot ln([O(3p)]/[O(3P)]o)/(1 k22[HO21ot) versus time, the
calculated points indeed fall on a straight line (curve b). In addition, it is seen
that the slope of line b is 6.8% larger than that of curve a calculated using
points up to two 1/e times. It is worth noting that curve b shows what the
[O(3p)] decay would be if HO2 was not continuously decreasing due to reaction
(22), but stayed constant at [HO21o. The increase in the calculated value of k'1
using the corrected decay curves over the uncorrected curve is ~7.8%, i.e.,
(k'uncorrected k'd)/(k'corrected k'd) = 1.078. This is exactly what we would get
if we used the [HO2] at the point where reaction (1) has proceeded to its 1/e
point. Since in all kinetic measurements we used the O(3p) decays up to two
1/e times in calculating k'1 values, we can simply correct the [HO21o value to
represent that at the 1/e point. The value [HO2] at the 1/e point for reaction
(1), [HO2]'o, is related to [HO21o by the simple relation

[HO2]'o = [HO]o/{1 (2k22/k1)} (IX)

Most of our kinetic data were corrected using this method rather than
correcting each decay curve point by point. One set of data was analyzed by
correcting each decay curve and the obtained value of k1 was indistinguishable
from that obtained using the above simpler method. The correction of k1 for
this HO2 loss process ranged from 5.4% at 10 Torr to 9.7% at 500 Torr. Both
the uncorrected (column 5) and corrected values (column 6) of k1 are listed in
Table I.
Loss of HO2 due to diffusion. It is possible for HO2 to diffuse from the
detection zone to the walls of the reactor and be lost during the course of the
reaction. To minimize this possibility, the photolysis beam was made to nearly
fill the reactor. [O(3P)] was monitored at the center of the beam so that HO2

loss due to diffusion would be minimal. Therefore, HO2 loss due to diffusion
had to be quite rapid to be important. It seems unlikely to be faster than that
of O(3P) which was observed to be less than -5 s-1 in this system. In addition,
if this loss was important, at low [HO2] the observed O(3P) temporal profiles
would be nonexponential, and be unobservable at high [HO2]. As discussed
earlier, slightly nonexponential decays were observed at all [HOg] and were
completely accounted for by reaction (22). Therefore, we conclude that HO2
loss due to diffusion was unimportant and [HO2] was not corrected to take this
process into consideration.
Secondary reactions involving O(3P). If O(3P) is either removed by
reactions other than (1) and (17) or created during the course of reaction (1),
our measured value of k1 would be erroneous. The reaction of O(3P) with
H202 [reaction (17)] does occur and, as pointed out earlier, merely contributes
to an increased intercept. The concentration of 03 used in these experiments
(typically < 1 x 1013 cm-3) cannot deplete O(3P) via the O(3p) + 03 202
reaction. Just after photolysis, a small fraction of O(3P) could react with OH,

O(3p) + OH O2 + H (23)

However, the amount of OH thus depleted is negligible (-0.05%) and
inconsequential since the O(3P) temporal profiles for the first 100 ps after the
laser pulse was not followed. Also, any O(3P) lost by reaction (23) will
regenerate an equivalent amount of HO2 due to subsequent reactions,

H + HO -* 2 OH (H2 + 02, O + H20)20 (24)

OH + H202 HO2 + H20



Reaction (23) cannot be an important contributor to the O(3P) loss during
reaction (1) since the steady state concentration of OH is very low (-108 109
cm-3) as discussed earlier. Also, since k23 k1,24 reaction of O(3p) with OH
would be indistinguishable from that with HO2. The only reaction that can
produce O(3P) during our experiments is that due to 02(alAg) + 03 O(3P)
+ 202. 02(alAg) is formed by ozone photolysis. The rate coefficient for this
reaction24 is too slow to be important under our experimental conditions which
employed only 1 x 1013 cm43 of ozone. Also, variations of [03] had no effect on
the measured value of kI implying the negligible importance of 02(alAg).
Possible effects of H20. Reactions of HO2 are known to be enhanced by
the presence of H20.24 In our experiments, H202 was the source of HO2 and
it is hard to remove water from the system since H202 decomposition leads to
H20 formation. Therefore, to check for possible H20 effects, two series of
experiments using 100 Torr N2 were carried out where up to 1.5 Torr of H20
was added. The measured values of k1 were, with the combined experimental
error, unaltered and did not show any trends with H20 concentration. We
could not use higher concentrations of H20 since it absorbs 130.3 nm oxygen
resonance radiation, thereby severely decreasing O(3p) detection sensitivity.
These experiments did show, however, that our measured values of k1 could not
have been affected by the presence of H20 since the maximum amount of H20
that could have been in the system was ~10% of the H202 which was usually
< 0.3 Torr.
Total error in the determination of k1. As mentioned earlier, the total
estimated error in calculating the fluence is 12%. If we add to this error,
possible errors in the knowledge of [H202], the quantum yield of OH
production, and the uncertainties in absorption cross sections of H202 at 248.5
and 202.6 nm (5% each), the total uncertainty in our calculation of [OH]


produced upon H202 photolysis would be -15%. It is important to note here
that any systematic errors in the knowledge of o(H202,248.5 nm) will be
canceled out, at least partially, since we are using a(H202,202.6 nm) and as
long as both numbers were taken from the same set of data the ratio
o(H202,248.5 nm)/U(H202,202.6 nm) will be much more accurate. It is this
ratio that is important in calculating [OH]o. However, we have conservatively
assigned 5% error to each quantity.
As we discussed earlier, not all OH is converted to HO2. The possible
reduction due to reaction (21) was 8% in the worst case. In addition, the HO2
loss due to reaction (22) varied from 5% to 9.7% with a mean value of 7.5%.
These losses have been well accounted for, as described earlier. However, to
be on a generously safe side, we assume that the errors in each correction could
be 100%. Using all these errors in an error propagation analysis yields the net
error in HO2 concentration to be 18%. This error has been added to the
precision of the measurements of kI and the total estimated errors are shown
in the last column of Table I.
Comparison with Other Studies
Table II lists all measurements of kI to date. This table is essentially a
duplication of that presented by Keyser19 except that the recent results of
Sridharan et al.20 and those of the present investigation are included. The
possible reasons for the discrepancies between k1 values measured by Keyser
and Sridharan et al. on one hand and all other previous measurements on the
other have been discussed before19,20 and need not be repeated. The agreement
between our results and that of Keyser, (6.1 0.4) x 10-11 cm3 molecule-1s"1,
is extremely good. The errors quoted above for Keyser's value do not include
systematic errors which he estimates to be + 25%. The results of Sridharan
et al.,19 (5.4 + 0.9) x 10-11 cm3 molecule-ls-1 are also in good agreement with

04 eq eq cq C4 eq C4-

404 0 0~ 0

a0. .) ... 6.
eq0 ~A A A
eq eq e

o A A A A A~

A 0 -0

.- -. .-
A A 0 .0 .0

0 0 i-
6-.e ++11+



4 eq

0) N t-0

Z0 0

0 0 tCiL
E--~q 4 oo 00 t-(


.- C4 m*




our results considering the error limits of the two studies. (It is worth pointing
out that Keyser and Sridharan et al. have not corrected their data for the
occurrence of reaction (22) since they measured [HOg] and the change in O(3P)
decay rates due to the continuously changing concentration of HO2 will be
small as was discussed for our measurements.) Based on the k1 value measured
by Keyser, Sridharan et al., and us, it appears that the value of k1 is well
established at 298 K and the average value is (5.9 + 1.0) x 10-11 cm3
molecule-Is-1, independent of pressure. The atmospheric implication of this
higher value of k1 compared to previous results have been discussed by
Keyser19 and Sridharan et al.20 and need not be repeated.
Comments on the Pressure Independence of k1
In this study we have measured k1 over a pressure range of 10 to 500 Torr
of N2, a factor 50 change. A plot of k1 as a function of pressure is shown in
Figure 10. The error bars that are shown represent precision only. The reason
for doing so is that systematic errors are essentially the same at all pressures
and therefore the relative values of k1 at various pressures are unaffected by
the systematic errors. From Figure 10, it is clear that the value of k1 is
essentially the same at all pressures within the experimental precision. The
maximum range of k1 values we observe is 5.81-6.47 x 10-11 cm3 molecule-1s-1.
The average value falls within the error bounds of each one of the measured k1
values (including that in 100 Torr of Ar). We prefer to quote a pressure
independent value of (6.2 + 1.1) x 10-11 cm3 molecule-ls-1 for k1. Therefore, it
is clear and worth reiterating that the value of k1 is independent of pressure
from 10 to 500 Torr.
The reaction of O(3P) with HO2 is too fast to be understood in terms of
a simple H atom abstraction reaction proceeding through a tight transition
state.19,35 O(3p) can attack HO2 either on the H atom end to abstract the H








0 0
=- a)

= o

1= 0 )
cd o
2 !4


S- s -einoelo uwluo'L 3


atom or on the oxygen atom end to form an H-O-O...O0 (not necessarily linear)
complex which can decompose to give OH and 0219. The latter route can
proceed on a purely attractive potential energy surface since the free radical
site of HO2 is involved. If this complex is formed, it might be possible to
stabilize it and decrease its decomposition back to O(3P) + HOg. If this
happens, then the observed value of kI should increase at high pressures. The
absence of the effect of pressure on k, up to 500 Torr seen here indicates that,
if formed, the lifetime of the complex towards decomposition into O(3P) + HO2
should be a nanosecond or less. This short lifetime is not difficult to
comprehend since the complex would be very energy rich and the energy is
distributed over relatively few degrees of freedom. (A discussion of subsequent
experiments that clarify the reaction pathway is presented in the next chapter.)




The reaction of ground state oxygen atoms with hydroperoxyl radicals

O(3P) + HO2 OH + 02

is a major odd oxygen destruction pathway in the upper stratosphere and

mesosphere. Along with the reactions

O(3P) + OH H + 02


H + 02 + M HO2 + M

H +03 OH+ 02,



reaction (1) plays a major role in controlling the partitioning among H, OH,

and HO2 radicals in the upper atmosphere. Hence, accurate kinetic data for

reaction (1) are needed in order to model upper atmospheric chemistry.

Several kinetics studies of reaction (1) are reported in the literature.16,19-
23,36 Four recent direct measurements of k1 at 298 K are in excellent

agreement,16,19,20,36 with reported rate coefficients all within the range (5.2-6.2)

x 10-11 cm3 molele-ules1. However, there has been only one investigation of

the temperature dependence of k1. Keyser19 studied reaction (1) over the

temperature range 229 372 K in a discharge flow system at 1 Torr total

pressure and observed that kl(T) increased with decreasing temperature; his

reported "negative activation energy" was ~ 0.4 kcal/mole. Although Keyser's

study of reaction (1) was a high quality experiment which appears to be free of

significant systematic errors, the uncertainty in his reported activation energy

will remain undesirably high37 until independent confirmation is reported.

Several years ago, we developed a pulsed laser photolysis technique for

studying the kinetics of radical-radical reactions at pressures up to 1 atm. We

first applied this technique to investigate reaction (1) at 298 K over the

pressure range 10 500 Torr.16 In this paper, we report the results of a pulsed

laser photolysis resonance fluorescence study of the temperature dependence

of k1. Our results, obtained using a much different experimental approach than

that employed by Keyser,19 confirm his reported temperature dependence and

also demonstrate that k1 is independent of pressure at subambient


Experimental Section

With a few modifications, the experiments were carried out in the same

manner as our previous room temperature study.16 A review of the

experimental approach, along with details pertinent to this investigation, is

given below. A schematic of the apparatus is shown in Figure 11.

All experiments were carried out under "slow flow" conditions using a

jacketed Pyrex reactor with an internal volume of 320 cm3. The cell was

maintained at a constant temperature by circulating ethylene glycol from a

thermostated bath through the outer jacket. A copper-constantan thermocouple

with a stainless steel jacket was inserted into the reaction zone through a

vacuum seal, thus allowing measurement of the gas temperature under the

precise pressure and flow conditions of the experiment.

Reactants were produced with HO2 in excess using 248.5 nm pulsed laser

photolysis of H202/O3/N2 mixtures:

H202 + hv(248.5nm) 2 OH (13)

O(D) + O2(alAg) (14a)

03 + hv(248.5nm)

O(3P) + 02(X3Z ) (14b)

OH + H202 HO2 + H20 (15)

O(1D) + N2 O(3P) + N2



w m an C

0 ;%0 0


o ~co


0 ba C)

~cf 004~W
oL bD~- 40
0 o~


N~ a)

a .) C :*- 4

0 c0
-4 0 0
cd4 0
lcd ;d 4d


0 C
$og4 o


A Lambda Physik model 200E KrF excimer laser was used as the photolysis

light source. Kinetic data was obtained by time-resolved resonance fluorescence

detection of O(3p). The laser pulsewidth was 20 ns while, under our

experimental conditions, reactions (15) and (16) proceeded at rates of (3-10) x

103 and 7 x 107s-1, respectively. O(3p) decay rates were typically in the range

30 500 s-1.

As in our previous study, the concentration of the excess reactant, HO2,

was not directly measured but was calculated based on experimentally

measured and other known parameters. To obtain [HO2] one must determine

[OH]o, the initial concentration of photolytically produced OH, and the yield of

HO2 from reaction (15) and competing side reactions. The chemistry of

conversion of OH to HO2 is discussed in detail in a later section. Since the

concentrations of H202 and 03 were such that the system was optically thin at

248.5 nm, [OH]o could be calculated from the following relationship

[OH]0 = (OH x a(H202,248.5 nm,T) x [H202] x F (X)

where (DOH is the quantum yield for OH production from 248.5 nm photolysis

of H202, a(H202,248.5 nm,T) is the absorption cross section for H202 at 248.5

nm and temperature T, and F is the laser photon fluence. The determination

of each factor in equation (I) is discussed in detail below.

4)OH- It is known25,26,38 that (OH =2.

g(H2Og,248.5 nm.T). The absorption cross section for H202 at 248.5 nm

is known to be 8.8 x 10-20 cm2 at 298 K.37 We have recently measured

temperature dependent absorption cross sections for hydrogen peroxide over

the wavelength range 193-350 nm.39 At 248.5 nm, the following expression

describes the observed temperature dependence (units are cm2 molecule-1):

a(H202,248.5 nm, T) = 1.023 x 10-19 exp{(-45 + 20)/T} (XI)

HFI2Ogl. Hydrogen peroxide can be lost in the slow flow system either by

decomposition (particularly at higher temperatures) or by condensation (at

lower temperatures). To ensure that the H202 concentration in the reactor was

known, we monitored H202 by UV photometry before the gas mixture entered

the reactor and after the gas mixture exited the reactor. The absorption cells

were 216.2 and 90.0 cm in length. The monitoring wavelengths were 228.8 nm

in the longer cell (Cd line) and 202.6 nm in the shorter cell (Zn+ line). Both

cells were kept at ambient temperature, and the absorption cross sections

needed to convert absorbance data to H202 concentration were obtained by

interpolation from current NASA recommendations:37 1.86 x 10-19 cm2 at 228.8

nm and 4.31 x 10-19 cm2 at 202.6 nm. When the reactor temperature was 350

K or below, the difference in H202 concentration measured in the two

absorption cells was never more than a few percent. At 391 K, the highest

temperature at which experiments were performed, 15-20% of the H202 was

lost upon traversal of the reactor. The H202 concentration in the reaction zone


was always taken to be the temperature-corrected average of the concentrations

measured in the two absorption cells. Under our experimental conditions,

absorption by ozone was negligible compared to absorption by H202 at 202.6

nm. At 228.8 nm, ozone made a small but significant contribution to the total

absorbance. Care was taken to ensure that [03] was constant during the I and

Io measurements required for the [H202] determination.

F. The photolysis laser beam was made spatially uniform through use of

a segmented aperture optical integrator.16,28 Virtually the whole cross-sectional

area of the cell was irradiated, and the depth of focus of the integrated beam

was such that radical concentrations were nearly uniform down the entire

length of the cell. The laser beam fluence was measured as the beam exited the

reactor by using an EG&G photodiode based radiometer capable of measuring

individual pulses. Pulse-to-pulse stability, a requirement for signal averaging

in this type of experiment, was found to be very good. Only an occasional pulse

energy deviated from the average by more than +_ 5%. In order to avoid having

to correct the measured fluence for reflection off the back window, an

antireflection coated window with > 99.5% transmission at 248.5 nm was

employed. The radiometer was calibrated by using a novel ozone actinometry

method which has been described in detail previously.16

As mentioned above, all experiments were carried out under slow flow

conditions. The linear flow velocity through the reactor was typically 12 cm s-1

and the laser repetition rate was typically 0.4 Hz. The reactor was 23 cm in

length, so the reactor volume was completely replenished with a fresh gas

mixture between laser pulses. All experiments employed nitrogen as the buffer

gas at a total pressure of 80 Torr. Data were obtained over the temperature

range 266-391 K. The temperature range was limited at the low end by the

vapor pressure of H202 and at the high end by the thermal instability of H202.

The gases used in this study had the following stated minimum purities:

N2, 99.999%; 02, 99.99%. Hydrogen peroxide was 90 wt % in water. It was

concentrated further by bubbling N2 through the sample for several days before

experiments were undertaken and continuously during the course of the

experiments. To prevent significant decomposition of H202, all components

traversed by H202 between the bubbler and the exit from the last absorption

cell were Pyrex or Teflon with the exception of a few stainless steel fittings.

The needle valve and flowmeter in the H202 line were positioned so that N2

flowed through these components before entering the bubbler. Ozone was

prepared by passing 02 through a commercial ozonator and was stored on silica

gel at 195 K. Before use it was degased at 77 K to remove 02. Dilute O3/N2

mixtures were prepared in 12 liter Pyrex bulbs for use in experiments.

A typical experiment was initiated by flowing (3-10) x 1012 03 molecules

cm-3 in 80 Torr of N2 through the reactor and absorption cells. I0 and Jo, the

intensities of 228.8- and 202.6-nm light transmitted through the absorption

cells, were measured. Next, H202 was introduced into the gas flow, the N2 flow

was reduced so the total flow rate and total pressure were the same after

addition of H202 as before addition of H202, and I and J, the reduced

intensities of 228.8- and 202.6 nm light transmitted through the reactor, were

measured. Using the measured values of I, Io, J, and Jo, the concentration of

H202 in the gas stream entering and exiting the reactor was calculated; it

ranged from (2-6) x 1015 molecules cm"3. I and J were continuously monitored

during the course of an experiment. The multichannel analyzer was

pretriggered before the photolysis laser fired to obtain the background count

rate. The concentration of O(3p) was monitored as a function of time after the

photolysis pulse. A total of 50 200 laser shots were averaged to obtain one

pseudo-first order kinetic decay. The fluence of each laser pulse was measured

using the radiometer and a calibrated aperture. The laser fluence was varied

at constant [Og] and [H202] to obtain pseudo-first order decays as a function

of [HO2]. Five to ten fluence values were employed to determine each

bimolecular rate constant, k1, from the slope of a k' versus [HO2] plot (k' the

O(3p) pseudo-first order decay rate). After the fluence variations were

completed, the H202 flow was turned off, the total pressure and total flow rate

were readjusted, and 10 and Jo were remeasured. All HO2 concentrations

employed in the k1 determinations were in the range (0.15-8.5) x 1012 molecules

cm-3. It should be noted that, in a series of runs involving variation of the laser

fluence at constant [03] and [H202], the ratio [OH]o/[O(3P)]o is not altered; to

change this ratio, the composition of the reaction mixture must be varied.

Results and Discussion

In the absence of competing side reactions which affect the HO2

concentration, the kinetic system is inherently pseudo-first order, i.e., all HO2

lost via reaction (1) is rapidly regenerated via reaction (15). However, as will


be discussed in detail below, the importance of certain side reactions is

suppressed when [HO2] > > [O(3p)]. Most of our experiments were carried out

under experimental conditions where [HO2] = 10[O(3P)]o, although this ratio

was varied as a check on the kinetic model used to extract k1. At low

temperature, where relatively low H202 concentrations had to be employed, the

HO2 to O(3P) ratio was typically somewhat lower than that employed at higher


Photolytically produced O(3p) can be lost via the following pseudo-first

order processes:

O(3P) + HO2 OH + 02 (1)

O(3P) + H202 OH + HO2 (17)

O(3P) + 03 202 (6)

O(3P) -* loss by diffusion from the detector field of

view and reaction with background impurities (18)

Under our experimental conditions, reaction (1) dominated O(3P) removal

except at very low HO2 levels. Reaction (6) was of negligible importance while

reaction (17) was minor but not negligible.31. The value of k18 was directly


measured to be -5 s-1, 1 to 2 orders of magnitude slower than kj[H02] under

most experimental conditions.

In the absence of competing side reactions which affect the concentrations

of O(3P) or HO2, removal of O(3p) should obey first order kinetics

In{[O(3P)]o/[O(sp)]} = (kl[HO2] + kd)t = k'expt (XII)


kd k17[H202] + k6[03] + k18 (XIII)

Figure 12 shows typical plots of In [O(3p)] versus time. Equation XII does

indeed appear to be obeyed. If side reactions were unimportant, the

bimolecular rate coefficient would be obtained from the slope of a k'expit versus

[OH]o plot, such as shown in Figure 13.

Secondary Chemistry

In the paper describing our 298K study of reaction (1), possible side

reactions were considered in detail.16 It was concluded that two reactions could

significantly affect the HO2 temporal profile:

OH + HO2 -* O2 + H20 (21)

HO2 + HO2 H202 + 02





time (ms)

Figure 12.

Typical 0(3P) temporal profiles observed following 248.5
nm pulsed laser photolysis of O3/H202/N2 mixtures.
Experimental conditions: T = 281 K, P = 80 Torr,
[H202] = 3.9 x 1015 molecules cm-3, [O3] = 6 x 1012
molecules cm-3, laser fluence (in units of mJ cm-2) = (a)
1.95, (b) 4.55, and (c) 6.31. Solid lines are obtained
from least squares analyses of the first two 1/e times of
O(3P) decay and give the following pseudo-first order
decay rates (k'expti): (a) 111 s-1, (b) 210 s-1, (c) 303 s-1.


Figure 13.

CH02O (1012molecules per cm3)

Typical plots of k'exptl versus [OH]o (open circles) and k'
versus [HO2]ma (closed circles). Experimental conditions: T
= 281 K, P = 80 Torr, [H202] = 3.9 x 1015 molecules cm-3,
[03] = 6 x 1012 molecules cm"3. The dashed line is obtained
from a linear least squares analysis of the k'exptl versus [OH]o
data and gives the "uncorrected" rate coefficient (5.24 + 0.38)
x 10-11 cm3 molecule"1 s"1. The solid line is obtained from a
linear least squares analysis of the k' versus [HO2]max data
and gives the "corrected" rate coefficient (6.15 + 0.54) x 10-11
cm3 molecule-1 s-1.

To quantify the roles of reactions (21) and (22) in our kinetics

experiments, a series of computer simulations were carried out where the

temporal profiles of key species were calculated under a variety of experimental

conditions by numerical integration of the appropriate rate equations. (The

final section of Appendix A contains a discussion of the integration procedure.)

For completeness, a number of reactions which were expected to play very

minor roles in the O(3P) and HO2 kinetics were included in the mechanism.

The complete set of reactions and rate coefficients used in the simulations is

given in Table III.

Reaction (21) competes with reaction (15) during the period immediately

after the laser pulse when OH is being converted to HO2. Each time reaction

(21) occurs, two HO2 radicals are lost which otherwise would have been present

to react with O(3P). Reaction (21) is most important at high laser fluence (i.e.,

high [OH]o/[H202]) and at low temperature. From the computer simulations,

a set of curves were constructed which relate [HO21]ma, the peak amount of

HO2 present after all photolytically produced OH has reacted away but before

appreciable HO2 loss via processes such as reaction (22) has occurred, to [OH]o.

Representative correction curves are plotted in Figure 14. For the 101

experiments used in the kl(T) determinations, the average value for

[HO21max/[OH]o was 0.947 while the minimum value under any set of

experimental conditions was 0.845.

As pointed out above, loss of O(3P) via reactions (1), (6), (17) and (18)

occurs on a time scale which is long compared to the time scale for HO2

Table III. Reaction Set for Computer Simulations.

Reaction Rate Coefficient(ab)

O + HO2 OH + 02 kI = 3 x 10-11 exp(200/T)

O + OH -* H + 02 k23 = 2.2 x 10-11 exp(117/T)

H + 03 OH + 02 k26 = 1.4 x 10-10 exp(-470/T)

OH + H202 HO2 + H20 k15 = 3.1 x 10-12 exp(-187/T)

O + H202 OH + HO2 k17 = 1.4 x 10-12 exp(-2000/T)

O + 03 202 k6 = 8 x 10-12 exp (-2060/T)

0 loss k18 = 5 s-1

OH + HO2 02 + H20 k21 = 1.7 x 10-11 exp(416/T) + 3 x
10-31[M] exp(500/T)

HO2 + HO2 H202 + 02 k22 = 2.3 x 10-13 exp(590/T) + 1.7 x
10-33M] exp(1000/T)

HO2 loss k27 = 5 s-1

HO2 + 03 OH + 2 02 ks = 1.4 x 10-14 exp(-580/T)

H + HO2 2 OH k24 = 6.4 x 10-11 exp(0/T)

(a) All rate coefficients are taken from reference 37 except k18 which was
measured and k27 which was set equal to k18.

(b) All rate coefficients except k18 and k27 are in units of cm3 molecule-1 s1.

0 5 10

EOH]o (l012molecules per cm3)

Figure 14.

Typical correction curves (obtained from computer
simulations) which relate [HO21ma to [OH]o. All curves
shown in the figure are from simulations with [H202] =
4 x 1015 molecules cm-3. Temperature: I, 400 K; II, 300
K; III, 260 K. [OH]o/[O(3p)]o: a, 104; b, 10; c, 4.






formation. Hence, if the concentration of HO2 were constant over the time

period of O(3p) removal, k, could be obtained from the slope of a k'exptl versus

[HO21max plot. Unfortunately, small but significant time variation of [HO2] can

result from the occurrence of reactions (17), (22), and (27). Reactions (17) and

HO2 loss by diffusion from the reaction zone and

reaction with background impurities (27)

(27) are most important at low laser fluences, i.e., low [HO2]. Reaction (17) is

strongly temperature dependent and increases in importance at high

temperature. Reaction (17) is also a more important source of HO2 when

[OH]o/[O(3P)]o is relatively low. Reaction (22) becomes an important HO2

removal mechanism at relatively high HO2 concentrations. To avoid large

corrections for HO2 loss via reaction (22), all experiments were carried out with

[HO21m.x < 9 x 1012 molecules cm-3 and all data analyses were restricted to

two i/e times of O(3P) decay; i.e., no data where [O(3p)]/[O(3p)]o < 0.13 were

used in the data analysis. Computer simulations were carried out under a

variety of experimental conditions. The first two 1/e times of the slightly

nonexponential computer generated O(3P) temporal profiles were least squares

fit to an exponential decay to obtain k'sim, the simulated decay rate. The

simulated decay rates were then compared to the "real" decay rates, k'r, to

obtain a set of correction curves, some of which are shown in Figure 15.

7 W




o o


0 O

WIS)1 / J;I



*S M



- 4

**> C?

0 >-

o 0


.- 0
2 o

0 > 0)-

44 -4

*- ..- <4 0
*M -

a) 0 "-
p.4 '

k'r kj[HO2]maz + k17[H202] + k6[O3] + kis (XIV)

The first two 1/e times of each experimental O(3P) temporal profile were least

squares fit to equation (XII) to obtain k'exptl. A corrected value of k' was then

computed from the expression

k' = k'expti(k'r/k'sim) (XV)

One source of uncertainty in the computer simulations concerns the fact that

k27 was not measured; it was estimated that k27 = k18 5 s-1. Lack of

knowledge of the exact value of k27 increases the uncertainty in the k'r/k'sim

factors at low concentrations of HO2. The low [HO2] data are relatively

unimportant in defining kj, so the uncertainty of k27 makes only a minor

contribution to the overall undertainty in k1. For the 101 experiments used in

the kl(T) determinations, the average value of k'r/k'sim was 1.06 and the

maximum value under any set of experimental conditions was 1.18.

As discussed above, k, values corrected for secondary chemistry were

obtained from the slopes of plots of k' versus [HO2] ma. Typical data are

shown in Figure 13. For all 15 rate coefficients measured, the corrected k1

values were larger than the values which would have been obtained from plots

of k'exptl versus [OH]o (see Figure 13, for example). The magnitude of the

secondary chemistry corrections was largest at the lowest temperatures. For

the 15 rate coefficients reported, the average difference between the corrected

and uncorrected bimolecular rate coefficients was 16% while the largest

difference was 26%.

Summary of Results

The experimental results are summarized in Table IV. Errors quoted for

individual "corrected" k1 determinations are 2a and refer only to the precision

of the k' versus [HO21,, data. The absolute accuracy of our k1 determinations

is limited not only by precision but also by uncertainties in measurement of the

laser photon fluences (F), the H202 concentration, the correction factor used

to obtain [HO2max from [OH]o, the correction factor used to obtain k' from

k'elpt1 (i.e., k'l/k'sim), and unidentified systematic errors. We estimate the

pertinent 2a uncertainties to be as follows: F, 10%; [H202], 5%; [HO2]m=/[OH]o

5%; k'r/k'uim, 10% at 266K and 5% at 298K and above; unidentified systematic

errors, 5%. Precision did not appear to be temperature dependent and

averaged 9%. Hence, the absolute accuracy of an individual k1 determination

is estimated to be + 23% at 266 K and + 21% at 298 K and above.

An Arrhenius plot of our results is shown in Figure 16. An unweighted

linear least squares analysis of the In k, versus 1/T data gives the following

expression (units are cm3 molecule1 s-1):

kl(T) = (2.91 + 0.70) x 10-11 exp[(228 + 75)/T]


= ----- -; =; =

- .43 -4.-4- 4.- 4. I I 4. 4. 4. 4. 4.-*-*- 4-

0 O 0 0 1- t 0 -- o to '0 '

N S O m t0 0 o m 0 CO '0 0 O O- t c- -S

oo) to t O
Lo CD 1D 1 04 1-0 1 0 1 4

00 L- 14t -4 o t
to 0 0

r4 0 0 10 C- 0
ci ci c0i ci^ c^ 0- <0 0O 0"- 00 0"" 0"" o"l '"

^. ^ Oi W Oi N O OO N W lO tO t-
m a m m m m m e s rt o O c ^ ~ o w w << -' 'r '


0 0


1 0

o 4


Q0 M
I E.
S. 00g

5 M "
S <




J9 --o-- KEYSER


4 -I I 4
2.5 3.5 4.5

1 O00/T(K)

Figure 16. Arrhenius plot for the O(3P) + HO2 reaction.

The uncertainties quoted in the above expression are 2a and represent precision


Comparison with Previous Work

Our results are compared with those reported by other investigators in

Table V. The 298 K value reported in this paper is identical with our

previously reported value16 and at the upper end of the group of recent direct

determinations16,19,20,36 which span the range (5.2-6.2) x 10-11 cm3 molecule-1

s-1. Other than our studies, the other recent direct measurements all employed

low pressure discharge flow systems, all measured not only O(3p) but also HO2

(either directly or indirectly by conversion to OH), and all carefully considered

the role of competing side reactions. Hence, there is no obvious reason to

prefer one value over another. The early work of Burrows et al.22 and Hack

et al.21 give k1(298 K) values which are considerably lower than those reported

in references 16, 19, 20, and 36. Possible reasons for the apparently erroneous

results reported in references 22 and 21 are discussed elsewhere.19 Lii et al.23

obtained a value for k1 by comparing computer simulations with HO2 and O3

concentration profiles observed following pulsed radiolysis of 02/H2/Ar

mixtures. Their reported rate coefficient agrees well with our results.

However, as pointed out by Keyser,19 the experimental data of Lii et al. are

very insensitive to the value of kl; hence, their error limits should be 500%

rather than the reported 30%.

Very little temperature dependent data are available for reaction (1).

Values for k1 of 8 x 10-11 cm3molecule-'s-1 at 1600 K40 and 6 x 10-11cm3

*gq e q N 0 o V M

0 o '- <3 C o 0
.S "^ eq__ -4.d5 0 <0
kO0 vU0
+1 + +1 + 1 + +1 +1

C o 10
C4a w

+1 +

Ci t
+1 +1

gl~~t As *--* |S--| asaa
00 AA A A A A A A
S 'A A A A A A

0 00 0 0q 0 0

,g NI *a / -l ".^ o a^ o

om aE ^^ E a
S 2

0 t--: C0 A 0 0
K T- oeqe -i eqi .^ C6
ON5 4M c 2
0 eq O
gE-4 cq CM C4 W Cl q m
Co Co0 e COC

It I

il I

'a- o '
ozo a

4 ~44
14 d

iII Is

molecule-'s-1 at 1050 K41 have been inferred from flame studies. The only

previous temperature dependence study in the atmospheric temperature regime

is that of Keyser.19 As seen from the comparisons in Table V and Figure 14,

our results are in excellent agreement with those of Keyser. Since our

experiments were very different in methodology from Keyser's and are subject

to different sources of systematic errors, the uncertainty of kl(T) for

atmospheric modeling purposes is now greatly reduced. Keyser's study

employed helium buffer gas at a pressure of 1.0 Torr while our study employed

nitrogen buffer gas at a pressure of 80 Torr. Thus, agreement between the two

studies strongly suggests that k1 is independent of pressure over the relevant

upper atmospheric temperature and pressure ranges.

Reaction (1) could proceed via a hydrogen abstraction mechanism (la) or

via formation of an energized HOOO "pseudo-intermediate" (ib):

0 + HO2 ..... HO2- OH + 02 (la)

O + O2H ..... OOH -O2 + OH (Ib)

Sridharan et al.42 recently reported an elegant experiment in which 160H and
180H products from the reaction of 180 with HO2 were monitored in a

discharge flow system. They found that only 16OH was produced, implying that

O reacts with HO2 via channel (Ib). Thermochemical estimates suggest that

HOOO is bound relative to 0 + HO2, but is ~ 12 kcal mol"1 less stable than the

OH + 02 products.43'44 Mozurkewich45 points out that since HOOO is bound

relative to 0 + HO2, we might expect to find a long-range interaction that

would produce a transition state for reaction (1) very similar to that expected

from formation of HOOO. Mozurkewich's RRKM calculations yield a rate

constant for formation of HOOO of 5.5 x 10-11cmmolecule'ls-1 independent of

temperature.45 The observed negative activation energy for reaction (1) could

probably be reproduced if a small barrier were assumed in the HOOO -* OH

+ OH2 reaction path.46

Implications for Atmospheric Chemistry

Model calculations of OH and HO2 concentration profiles in the upper

stratosphere are very-sensitive to the choice of kl(T). Kaye and Jackman,47

considering both sensitivity of their model to various parameters and

uncertainties in these parameters, have concluded that the uncertainty of k1

contributes more to the uncertainty in [OH] and [HO2] at 350N, 40 km altitude

than any parameter in their model except k21. The currently recommended

value37 for kl(T) is

kl(T) = 3.0 x 10-11exp[(200+200)/T] cm3nolecule-'s-1 (XVII)

The results reported in this paper will have little effect on the recommended

A factor and activation energy, but will substantially reduce the uncertainty in

the above expression; this will, in turn, significantly reduce the overall uncer-

tainty in model calculations of upper stratospheric OH and HO2 concentrations


and facilitate meaningful comparisons of stratospheric measurements with

photochemical models.

Jackman et al.48 have recently compared 03 concentrations at 5N and 43

km altitude measured by LIMS (limb infrared monitor of the stratosphere) with

those calculated from a photochemical model using LIMS measurements of

H20, HNO3, NO2, and temperature and SAMS (stratospheric and mesospheric

sounder) measurements of CH4 as input. The model predicted lower 03 levels

than those actually observed. A sensitivity analysis showed that the overall

uncertainty in the model calculation was a factor of 1.7, about the same

magnitude as the discrepancy between model and measurement. Of the many

parameters used in the model, the uncertainty of k1 was found to make the

sixth largest contribution to the overall uncertainty of the calculation. Our

results will therefore result in a small but significant reduction in the model





The reaction of ground state oxygen atoms O(3P) with CO1 radicals is the

rate determining step in the dominant catalytic cycle via which chlorine atoms

destroy odd oxygen in the middle stratosphere:

O + CO1 02 + Cl (2)

Cl + 03 CO1 + 02 (12)

0 + 03 -- 202 net

The primary source of stratospheric chlorine atoms is the photolysis of

anthropogenic chlorofluorocarbons.

Seven measurements of k2(298 K) are reported in the literature.49"56 There

is agreement among the five most recent studies that k2(298 K) lies in the range

3.5 4.2 x 10-11 cm3molecule-1s"1. The activation energy for reaction (2) is

known to be small,50-56 but its value is not as well defined as would be desirable

for such an important stratospheric reaction. In fact, it is not clear if k2(T)

increases or decreases with decreasing temperature. In addition to the

abovementioned studies of reaction (2) at atmospheric temperatures, one high

temperature (1250 K) shock tube measurement of kI has been reported,57 as has

one theoretical calculation of k2(T) over the temperature range 220-1000 K 58

Because predictions of chlorine catalyzed ozone loss are very sensitive to

the value of k2(T) used in model calculations, it is important that this rate

coefficient be determined with high precision at stratospheric temperatures.

Studies employing a variety of experimental techniques are desirable in order to

uncover possible systematic errors. All previous studies of reaction (2) at

ambient and subambient temperatures49-56 employed discharge flow systems

which were limited to total pressures of 10 Torr or less. It is interesting to note

that reaction (2) occurs on a potential energy surface with a minimum along the

reaction coordinate, i.e., the intermediate complex C1OO is a bound species

whose ground state correlates with O(3p) + ClO(X2j)59. Reactions which occur

on potential energy surfaces of this type often exhibit negative activation

energies and pressure dependent rates.

We have recently developed a pulsed laser photolysis method for carrying

out direct kinetics studies of radical-radical reactions at pressures up to one

atmosphere, and applied this method to study the temperature and pressure

dependence of the 0 + HO2 reaction.16,17 Using an extension of the technique

employed in the 0 + HO2 investigations, we have studied the kinetics of reaction

(2) in N2 buffer gas over the temperature and total pressure ranges 231-367 K

and 25-500 Torr. Our results, which include observation of a significant negative

activation energy, are reported in this paper.


A schematic of the apparatus appears in Figure 17. The two radical species

were created via a scheme involving two separate photolysis lasers. Under "slow

flow" conditions, a gas mixture containing Cl2 and 03 in a large excess of N2

buffer gas was first subjected to photolysis by a XeF excimer laser,

C12 + hv (351 nm) 2C1 (28)

The combination of the excimer laser fluence and [C12] was always large enough

for the condition [Cl]o > [03]0 to hold. During a predetermined delay period,

t', the reaction

Cl + 03 CO1 + 02 (12)

was allowed to go to completion. At this time the ozone in the reaction cell had

effectively been titrated by Cl atoms and the initial value of [03]O could be

related to [CIO]t'. At the end of this delay a second laser pulse, the fourth

harmonic of the fundamental wavelength from a Nd:YAG laser, photolyzed a

small fraction of the C10,

C10 + hv(266 nm) -* Cl + O


a >

o 0)

CI 4 0

ho 0

(1 0)'

ri! 0) 4 ) o~
0) Q.4 bi

0 4-2

0d C. 0t 'a 04

0 r0
> ej



The decay of oxygen atoms in the presence of excess CIO was followed by

monitoring the time dependence of fluorescence signal which was continuously

excited by a microwave discharge resonance lamp. The lamp was operated with

a low pressure of helium (<1 Torr) containing a small fraction of O2.

The ozone storage bulb contained a mixture of 1% to 2% 03 in nitrogen,

while the C12 was stored neat. These species were leaked through needle valves

into the main gas flow. Ozone in the gas flow was measured in a 35.0 cm

absorption cell that was placed within a multipass optical arrangement. Using

modified White cell optics, 30 passes of the 254 nm Hg line from a pen-ray lamp

were sent through the absorption cell for an effective path length of 10.50 m.

The chlorine was measured in a 216 cm absorption cell using a single pass of the

366 nm Hg line, also from a pen-ray lamp. Both atomic lines were isolated using

suitable band pass filters. Typically, the absorption cells were upstream from the

reaction cells, although in a few experiments the C12 and 03 were measured after

the flow exited the reaction cell. Because C12 absorbance at 254 nm was not

totally negligible [a = 1.6 x 10-21 cm2 (Ref. 60)], the reference light intensity for

the [03] determination was always measured with C12 flowing. The C12

absorption cross section at 366 nm and the 03 absorption cross section at 254

nm were taken to be 1.01 x 10-19 cm2 (Ref. 60) and 1.147 x 10-17 cm2 (Ref. 61-

63), respectively.

The Pyrex reaction cell measured 16 cm along its longer axis and had an

internal diameter of 4 cm. The two laser beams counterpropagated along the

longer axis. Around the middle of the cell were four 1.5 cm diameter side arms,


each perpendicular to the long axis of the cell and at 900 to each other. The

resonance lamp radiation entered the cell through one of the side arms and the

fluorescence signal was collected through a neighboring arm. The central

portion of the cell was surrounded by a jacket through which thermostated

liquids were flowed to control the temperature of the gas mixture inside the

reactor. The gas mixture entered the cell through several ports very near the

window at one end of the long axis and exited the cell through similar ports near

the opposite window. Because the chemistry initiated by the excimer laser beam

completely titrated one component (Og) of the gas mixture within much of the

cell volume, the cell was designed to have minimum total volume and low dead

space, i.e., gas flowed through all volume elements of the cell at approximately

equal rates. The typical linear flow rate through the cell was 14 cm s-1 and the

repetition rate of the two laser sequence was usually 0.4 Hz. Therefore, the gas

mixture within the entire volume of the reaction cell was replenished between

excimer laser pulses. The temperature of the gas mixture was measured by

replacing one of the end windows with an acrylic flange through which a

copper-constantan thermocouple could be inserted. The errors in the reported

temperatures are estimated to be no more than + 1.0 K at the extreme

temperatures and less at intermediate temperatures.

Oxygen resonance lamp radiation was focused into the reaction zone by a

2-inch focal length MgF2 lens. The reaction zone was viewed by a solar blind

photomultiplier tube through a similar lens. The volumes between the resonance

lamp and the reaction cell, and between the reaction cell and the photomultiplier

tube were purged with a mixture of 1% 02 in nitrogen. This excluded room air

and also acted as a filter of extraneous emissions from the resonance lamp. A

CaF2 window between the cell and the photomultiplier tube eliminated the

possibility of hydrogen atom detection. Fluorescence signals were accumulated

using photon counting techniques in conjunction with multichannel scaling.

Each sweep of the analyzer was triggered simultaneously with the excimer laser.

From 50 to 500 flashes were averaged to obtain sufficient signal to noise ratio

for quantitative kinetic analysis.

The total pressure in the flow system was measured with a capacitance

manometer. Due to the necessarily fast flow rate and the small (4.0 mm i.d.)

tubing connecting the various components of the flow system, there were

measurable pressure gradients between the absorption cells and the reaction cell.

Quantitative adjustments were made for these gradients in the calculation of the

Cl2 and O3 concentrations in the reaction cell under each set of conditions. The

magnitude of the adjustment was largest at the lowest pressure (15% at 25 Torr)

and negligible at the highest (1% at 200 Torr). The nitrogen buffer gas

comprised at least 94% of the mixture for all experiments and its flow rate was

monitored using a calibrated electronic mass flowmeter.

The concentration of CIO, the excess species in this technique, was derived

from the concentration of ozone as measured in situ. Therefore, it was not

necessary to have an absolute calibration of the photolysis laser fluences.

However, it was important to know that following the excimer laser pulse the

condition [Cl] > [O0] held throughout the reaction zone. Therefore, the excimer

laser fluence was measured in each experiment. As will be discussed below, it

was also important to monitor the 266 nm laser fluence. Both of these quantities

were determined using a photodiode-based calibrated radiometer. With the front

optic of the excimer laser approximately 2 m from the center of the reaction cell,

the beam was rectangular in cross section and measured 2.0 cm by 4.0 cm.

When measured through a 0.5 cm diameter aperture, the fluence peaked at the

center of the beam and dropped off by 10% per 0.5 cm distance from the center

in both the vertical and horizontal directions. The beam from the Nd:YAG laser

was aligned at the center of the volume irradiated by the excimer laser. It was

estimated to be 0.4 + 0.1 cm in diameter.

The reagent purities and sources were as follows: N2 (99.999%, Spectra

Gases, Inc.); Cl2 (99.9%, Matheson Gas Products, Inc.); 02 (99.99%, Spectra

Gases, Inc.). Ozone was prepared in a commercial ozonator using UHP oxygen.

It was stored at 195 K on silica gel and degassed at 77 K before use. The other

gases were used without further purification.


In the absence of competing reactions that either deplete or enhance the

ground state oxygen atom (O(3P)) concentration, the temporal behavior of

[O(3p)] following the 266 nm laser pulse can be described by the relationship

In {[O(3P)]t/[O(3P)]t,} = (k2[CIO] + k'd)(t t') = k'(t-t') (XVIII)


k'd = k30oC12] + k18


In Equation XIX, k30 and k18 are the rate coefficients for the following processes:

O(3P) + C12 products- (30)

O(3p) -* loss by diffusion from the viewing
zone and reaction with background
impurities (18)

A typical experimental O(3P) temporal profile is shown in Figure 18.

Unexpectedly, a significant build-up and decay of O(3P) occurred before the 266

nm laser fired; possible sources of this O(3p) and its implications for our study

of reaction (2) are discussed below. It should be noted that the vertical axis in

Figure 18 has units of concentration. To construct Figure 18, the fluorescence

signal before the 266 nm laser fired was scaled to account for the fact that the

351 nm laser photolyzed the entire field of view of the detection system, while

the 266 nm laser photolyzed only 15% of the detector viewing zone. The size of

the viewing zone was estimated by placing a series of apertures in front of the

351 nm beam and noting the variation of fluorescence signal strength with beam


Typical decays of O(3P) generated by the 266 nm laser pulse are shown in

Figure 19. At each temperature and pressure a minimum of five and an average

of eight experiments were performed at various values of [03]o. Because the

Nd:YAG laser fluence was measured in each experiment, the amount of CO1 lost

via 266 nm photolysis could be quantified. The expression


Excimer fires
li I I I




-Nd:YAG fires


TIME (ms)

Figure 18.

An experimental 0 atom temporal profile obtained
under the following conditions: [Cl2] = 1.40 x 1016
molecules cm-3; [03]o = 7.27 x 1013 molecules cm-3; [Cl]o
= 1.82 x 1014 molecules cm-3; total pressure = 25 Torr;
temperature = 298 K; multichannel scalar dwell time =
25 microseconds.





r i I r 1

1'' I "I I
0 3.0 6.0 9.0

Figure 19.

Typical 0 atom temporal profiles. These experiments
were carried out under the following conditions (all
concentrations expressed in molecules cm-3): 25 Torr
total pressure; T = 298 K; [C12] = 9.8 x 1015; [O3]o =
1.73 x 1013 (A), 5.15 x 1013 (B), 8.76 x 1013 (C); and [Cl]o
= 1.10 x 1014 (A), 1.66 x 1014 (B), 2.29 x 1014 (C).





[C10]t =[03]o [O(SP)]t, (XX)

was used to calculate the CIO concentration under the assumption that no loss

of C10 occurred during the delay between photolysis laser pulses. The ratio

[O(3p)]/[CIO] immediately following the 266 nm laser pulse was determined for

each experiment, and was typically in the range 0.01- 0.04. Hence, it was

appropriate to correct each measured decay rate for slight deviations from

pseudo-first order conditions. These corrections were derived from computer

simulations of reactions (2), (30) and (18) under the range of experimental

conditions employed. (A discussion of the computer simulations appears in

Appendix A.) Best fit values for each decay rate (k') were obtained from linear

regression analyses of the experimental data over at least two 1/e times. Each

value of k' was then increased by 2% or less using the appropriate nonpsuedo-

first order correction. Representative plots of [CIO] versus k' are shown in

Figure 20. These data were subjected to linear least squares analyses to give

values for k2. The results are presented in Table VI. Note the separate columns

for k2(uncorrected) and k2(corrected). The former represents the best fit to the

data when k' was not corrected for nonpseudo-first order behavior and [CIO] was

set equal to [O3]o less the amount photolyzed to produce [O(3P)]. The latter k2

values include the small nonpseudo-first order correction to k' and additional

corrections to [CIO] discussed below.

The absorption cross section used to calculate the amount of CIO

photolyzed by the Nd:YAG laser was estimated experimentally. The signal level









0 5 10


(101 molecules cm3)

Figure 20.

A k' versus [CIO] plot of typical data taken at 25 Torr
total pressure of N2 and at three temperatures. Note
that the 298 and 231 K data have been displaced upward
by 1000 and 2000 s-1, respectively. Solid lines are
obtained from linear least squares analyses and give the
following rate coefficients (in units of 10-11 cm3
molecule-1 s-1): 4.99 at 231 K, 4.07 at 298 K, 3.44 at 367

Table VI. Summary of k2 Determinations

Temperature Pressure k2 x 1011 k2 x 1011
(K) (Torr) (uncorrected) (ab) (corrected)(a)
(cm3molecule-ls-1) (cm3molecule-ls-1)

231 25 4.64 + 0.16 4.99 + 0.17
238 25 4.25 + 0.43 4.47 + 0.43
252 200 3.10 + 0.22 3.97 + 0.30
255 25 4.15 + 0.20 4.42 + 0.20
255 200 3.25 + 0.20 3.95 + 0.32
257 25 3.86 + 0.15 4.10 + 0.11
257 25 4.30 + 0.15 4.47 + 0.19
275 25 3.94 + 0.17 4.10 + 0.16
298 25 3.55 + 0.21 3.84 + 0.14
298 25 3.47 + 0.25 3.59 + 0.25
298 25 3.51 + 0.14 3.62 + 0.13
298 25 3.87 + 0.36 3.93 + 0.36
298 25 3.89 + 0.15 4.07 + 0.17
298 50 3.91 + 0.23 4.07 + 0.24
298 50 3.62 + 0.15 3.98 + 0.18
298 50 3.70 + 0.12 3.91 + 0.11(c)
298 50 3.53 + 0.19 3.76 + 0.14
298 50 3.76 + 0.16 3.98 + 0.14
298 100 3.73 + 0.68 3.97 + 0.67
298 200 3.39 + 0.27 3.73 + 0.24

(a) Errors are 2a.
(b) Uncorrected values have not been adjusted for non pseudo-first order
conditions and [CIO] loss by reaction (31). See text for details.
(c) Carried out with 283 nm photolysis of CO1, under "reversed" and "normal"
flow conditions.

Table VI. (Continued)

Temperature Pressure k2 x 1011 k x 1011
(K) (Torr) (uncorrected)(a~) (corrected)(a)
(cm3moleculels-1) (cm3molecule-ls-1)
298 200 3.39 + 0.28 3.71 + 0.24
298 200 3.55 + 0.12 3.76 + 0.13
298 500 3.18 + 0.16 4.03 + 0.20
338 25 3.16 + 0.60 3.26 + 0.12
359 200 2.99 + 0.14 3.09 + 0.15
360 25 2.89 + 0.13 2.96 + 0.12
367 25 3.35 + 0.13 3.44 + 0.13

(a) Errors are 2a.
(b) Uncorrected values have not been adjusted for non pseudo-first order
conditions and [CIO] loss by reaction (31). See text for details.

immediately after the Nd:YAG laser fired was directly proportional to the

concentration of oxygen atoms. If the Nd:YAG laser was not preceded by a pulse

from the excimer laser, then the photolyte was 03. If all the 03 had been

converted to CO1 via reaction with Cl atoms created in the 351 nm excimer

pulse, then the signal was due to CIO photolysis. Therefore, in back-to-back

experiments with constant [C12], [03] and 266 nm fluence, the ratio of 0(3P)

signal with and without 351 nm photolysis should be identical to the ratio of the

CIO and 03 absorption cross sections at the Nd:YAG laser wavelength (the

O(1D) product of 03 photolysis is rapidly quenched to O(3P) by N2). This

experiment resulted in a value of 0.35 for the signal ratio. If U(03,266 nm) is

taken to be 9.0 x 10-18 cm2 (Ref. 60) then a(CIO,266 nm) is approximately 3.1 x
10-18 cm2. This is somewhat lower than the low resolution cross section at this

wavelength that is reported in the literature.64 However, we carried out the

identical signal level comparison near 283 nm, a wavelength where the high

resolution cross section has been characterized.65 Using a frequency doubled,

Nd:YAG pumped tunable dye laser the rotational structure in the CO1 spectrum

was reproduced by observing the resonance fluorescence signal as the wavelength

was scanned. At 282.95 nm, the peaks of the R(19.5) and P(16.5) lines of the

A2x3/2 X2;3/2 9-0 band, we observed a factor of 1.65 + 0.20 more fluorescence

signal when the excimer laser photolyzed C12 than when the excimer laser beam

was blocked. Based on the dye laser linewidth employed and the literature

values for the ozone and CIO cross sections, we expected a ratio of 1.80 + 0.40.

This result confirms our value for u266(CIO).

As mentioned above, the delay between the two laser pulses, t', was

adjusted to be long enough for reaction (12) to go to completion. The value of

the Cl + 03 rate constant is reasonably well known" so an appropriate delay

time could be calculated. As a check the delay time was varied until a constant

signal and decay rate were observed, indicating that all the O3 had been

converted to CIO. The majority of experiments were carried out with delay

periods of either 3.4 ms or 6.4 ms.

The temporal behavior of C10 could be monitored by following both the

O(3p) signal level produced by 266 nm photolysis and the measured value of k'.

For some conditions (lower temperature, higher pressures) it was observed that

C10 was decreasing on the time scale of the delay between the two lasers. The

disappearance of C10 is probably due to self-reaction, a process that has not been

completely characterized.66 In order to make a correction for the amount of C10

lost during the delay between laser pulses and also during the 0 atom decay, a

series of experiments were carried out in which the delay time was varied at

fixed [03], [C12] and laser fluences.

For the process

C10 + C10 products (31)

the time dependence of [C10] is described by the relationship

2k31t = [ClO]t- [C10]-1



If we define k31, to be the second order rate coefficient for C10 removal under

our experimental conditions, then plotting [CIO]t-1 versus time should yield a

straight line of slope 2k31' and intercept equivalent to [CO1]o-1. The absolute

concentration of C10 needed to determine k31 was deduced in two manners. In

the first method, Equation XVII was rearranged to calculate [C10]. Here k2 was

taken from fixed delay experiments at the same experimental temperature and

pressure. k31, was determined from Equation XXI. Then the values for [C10]

in the fixed delay experiments were adjusted to account for the loss of C10

during the delay. This procedure was iterated until constant values of k2 and

k31, were obtained. The second method involved relating the signal level

immediately following the 266 nm laser pulse to the value of [C10]. The signal

calibration could be arrived at by two procedures. When enough data was

available under the same experimental conditions, a [C10] versus signal

calibration curve was constructed. Otherwise, signal versus delay time was

plotted and extrapolated to time zero and that signal level normalized to [03]o.

This normalization factor was then used to put the signal level at each delay

time on an absolute concentration scale. Because of the reciprocal relation in

Equation XXI, the value of k31, is quite sensitive to the scaling factor used in

setting the value of [C10]. Also the signal level is quite dependent on the

operating conditions of the resonance lamp. Therefore, the values of k31,

determined from the first method are more precise. However, when using k' as

a measure of [C10], we assume that 0 atoms are removed only by reaction with

C10 and C12; this assumption appears to be justified. A typical plot of [C10]"1

versus time appears in Figure 21. The end results of a number of such

experiments appear in Table VII. These results are presented only as a measure

of the phenomenological loss rate of C10 in our system, not as a definitive

measurement of k31.

Hayman et al.66 report 298 K values for k31 which are substantially faster

than the values for k31,(298 K,P) we have determined. However, our 255 K

results appear to agree reasonably well with the low temperature values of k31

of Hayman et al. According to Hayman et al.,66 reaction (31) has an important

branch to form a weakly bound dimer which can either decompose or react

rapidly with chlorine atoms. The relevant reaction scheme is given below:

C10 + C10 + M '* (C10)2 + M (31a,-31a)

CIO + CIO -* OClO + Cl (31b)

C10 + C10 -* other products (31c)

Cl + (C10)2 -* C12 + C100 (32)

C1OO + M Cl + O2 + M (33)

Cl + C1OO -* C12 + 02 (major) (34a)

Cl + C100 -i 2C10 (minor) (34b)

Cl + OCIO -. 2CIO