Effects of secondary reactions upon the determination of the biomolecular rate constant for several alkane and OH reactions

Material Information

Effects of secondary reactions upon the determination of the biomolecular rate constant for several alkane and OH reactions
Slemp, Gary Michael
Publication Date:
Physical Description:
xi, 109 leaves : illustrations ; 29 cm


Subjects / Keywords:
Free radicals (Chemistry) ( lcsh )
Chemical kinetics ( lcsh )
Chemical kinetics ( fast )
Free radicals (Chemistry) ( fast )
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )


Includes bibliographical references (leaves 108-109).
General Note:
Submitted in partial fulfillment of the requirements for the degree, Master of Science, Department of Chemistry
Department of Chemistry
Statement of Responsibility:
by Gary Michael Slemp.

Record Information

Source Institution:
|University of Colorado Denver
Holding Location:
|Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
31250632 ( OCLC )
LD1190.L46 1994m .S54 ( lcc )

Full Text
Gary Michael Slemp
B.S., Colorado State University, 1977
A thesis submitted to the
Faculty of the Graduate School of the
University of Colorado at Denver
of the requirements for the degree of
in partial fulfillment
Master of Science

This thesis for the Master of Science
degree by
Gary Michael Slemp
has been approved for the
Department of
Robert Damrauer

Slemp, Gary Michael (M.S., Chemistry)
Effects of Secondary Reactions Upon the
Determination of the Bimolecular Rate Constant
for Several Alkane and OH Reactions.
Thesis directed by Dr. Larry G. Anderson
The influence of OH + alkyl radical reactions
on the determination of the bimolecular rate
constant for the reaction of the OH radical with a
series of alkanes was studied using an Acuchem
modeling program. The modeling was for an
experimental investigation of these reactions in a
discharge flow resonance fluorescence system. An
alkyl radical is produced from the reaction of OH +
alkane. These alkyl radicals compete with the
original alkane for OH radicals. These secondary
processes are of some significance in the
measurement of the bimolecular rate constant of the
initial OH + alkane reaction. At very high OH
concentrations, the OH + alkane bimolecular rate
constant used in the modeling simulation and the

bimolecular rate constant determined by the model
differed by as much as 85%. The results demonstrate
that a significant impact upon the determined
bimolecular rate constant was observed with
increasing OH radical concentrations.
The model includes a variety of radical
reactions. The model results showed that the OH +
alkyl radical had the most significant effect on the
determined bimolecular rate constant.
Researchers acknowledge the presence of
secondary effects in these types of studies, however
most deny that these effects have any significant
impact upon the determination of the bimolecular
rate constant for the OH + alkane reaction under
their experimental conditions. In this study, a
flash photolysis experiment was modeled. The model
showed that secondary effects, if present, did not
significantly effect the determination of the
bimolecular rate constant under normal experimental
conditions. These experiments were done at lower OH
concentrations than our flow system experiments.
When the flash experiments were modeled using higher
OH concentrations, the results demonstrated that

secondary effects did have an impact on the
determined bimolecular rate constant.
Based upon the results of this research,,
we find that the secondary reaction of OH + alkyl
radical does have a significant influence upon the
determination of the bimolecular rate constant for
the OH + alkane reaction. Although this effect can
be observed at low OH concentrations, it becomes
increasingly more important at higher OH
This abstract accurately represents the content of
the candidate's thesis. I recommend its publication.

1. INTRODUCTION................................. 1
Intent of Research........................... 1
Background.................................. 2
Sources of OH Radicals in the
Troposphere.................................. 4
Sources of Hydrocarbons in the
Troposphere................................ 9
2. METHOD...................................... 10
Experimental Approach....................... 10
Laboratory Experiments...................... 15
Modeling Experiment....................... . 16
3 RESULTS..................................... 2 6
Modeling Results............................ 2 6
Reaction Mechanism Importance............... 3 8
Curvature Effects........................... 42
4. COMPARISONS WITH LITERATURE................. 47
5. CONCLUSIONS............................... 55
B. EXPERIMENTAL DESIGN.................... 61
C. KINETIC MODELING PLOTS................ 8 5
E. EXPERIMENTAL RESULTS................... 94
F. OH FORMATION / LOSS PLOTS.............. 9 8

1. Acuchem Input File........................ 17
2. Acuchem Output File......................... 20
3. Reactant Concentration Determination........ 22
4. Reaction Time Determination................. 24
5. OH Decay Plot.............................. 3 0
6. Kinetic Results for Reaction Effects:
Very High OH Concentrations: Isobutane..... 3 9
7. OH Loss with OH + Butane Reaction.......... 4 3
8. OH Loss without OH + Butane Reaction........ 44
9. Kinetics Modeling Results for Flash
Photolysis Experiment: Butane............... 48
10. Kinetics Modeling Experiment for Flash
Photolysis Experiment: Butane with
Increased OH Concentrations................. 50
11. OH Loss: OH + Butane Reaction............... 53
12. OH Loss: OH + Butyl Reaction................ 53
1. Discharge Flow Resonance System............ 65
2. Orientation of Detection System........... 7 0
3. Titration Curve............................ 7 5
4. OH Decay Plot............................... 78

5. Propane Experimental Data 07/01/91........81
6. Propane Experimental Data 06/30/91....... 83
7. Propane Experimental Data 07/01/91....... 83
8. Butane Experimental Data 08/18/91........ 84
9. Butane Experimental Data 08/30/91........ 84
1. Kinetics Results: Ethane............... 85
2. Kinetics Results: Propane................ 8 6
3. Kinetics Results: Butane................ 87
4. Kinetics Results: Isobutane............. 88
5. Kinetics Results: Neopentane............ 89
1. OH Formation: H02 + NO................... 98
2. OH Formation: 0 + H02................... 98
3. OH Formation: 0 + H202.................. 99
4. OH Formation: N03 + H02.................. 99
5. OH Formation: H '+ H02...................100
6. OH Formation: 0 + Butane.................100
7. OH Formation: N02 + H.................. 101
8. OH Loss: OH + 0......................... 101
9. OH Loss: OH + H02....................... 102
10. OH Loss: OH + Butene.................... 102
11. OH Loss: OH + Octane.................... 103

12. OH Loss: OH + OH........................ 103
13. OH Loss: OH + H202...................... 104
14. OH Loss: OH + H0N02..................... 104
15. OH Loss: OH + H2........................ 105
16. OH Loss: OH + N03....................... 105
17. OH Loss: OH + ROH....................... 106
18. OH Loss: OH + N02....................... 106
19. OH Loss: OH + HONO...................... 107
20. OH Loss: OH + NO........................ 107

1. Hubler Study............................... 8
2. Comparison of OH + Isobutane Under
Four Modeling Conditions.................. 41
3. Comparison of OH + Isobutane Under
Four Modeling Conditions
(two point curve)......................... 41
4. OH Decay Comparison....................... 45
1. Core Reactions............................ 58
2. Ethane Reactions......................... 59
3. Propane Reactions......................... 59
4. Butane Reactions.......................... 60
5. Isobutane Reactions....................... 60
6. Neopentane Reactions...................... 60
1. Comparison of Experimental Rate Constant
for OH + Propane to Literature Values.... 80
1. OH + Ethane Model Results................. 90
2. OH + Ethane Model Results
(two point curve)......................... 90
3. OH + Propane Model Results................ 9 0

4. OH + Propane Model Results
(two point curve)....................... 91
5. OH + Butane Model Results............... 91
6. OH + Butane Model Results
(two point curve)....................... 91
7. OH + Isobutane Model Results............ 92
8. OH + Isobutane Results
(two point curve)....................... 92
9. OH + Neopentane Model Results........... 92
10. OH + Neopentane Model Results
(two point curve)....................... 93
1. Experimental Results for
OH + Propane Reaction................... 94
2. Experimental Results for
OH + Butane Reaction..................... 9 6

Intent of Research
Absolute rate constants for the reaction of the
hydroxyl radical with various alkanes have been
determined experimentally by several researchers
including Frank Tully1,2,3 and Roger Atkinson.4,5,6 The
first step in this reaction involves the abstraction of
a hydrogen atom from the alkane resulting in an alkyl
radical and water. These alkyl radicals can further
compete with the original alkane to react with the OH
radicals still present in the system. When high levels
of OH radicals are present, these secondary effects may
impact the determination of the rate constant for the
initial OH + alkane reaction. The importance of
secondary kinetic reactions upon these determined rate
constants has been considered insignificant by most
In this investigation, flow tube experiments along
with complex chemical modeling were used to determine
the rate constant for the reaction of OH radicals with
various alkanes. The effect of secondary interactions
upon these determined absolute rate constants were
observed. The results in this study demonstrate that
under certain conditions, the determined rate constant

is affected significantly by these secondary
The study of reactions between OH radicals and
hydrocarbons has become increasingly significant from
the standpoint of atmospheric science and combustion
chemistry. Determination of the hydroxyl radical
reaction rate constant is highly important in assessing
the atmospheric lifetimes of many organic compounds.
In 1961, Leighton suggested that free radicals such as
OH may be a key intermediate in photochemical air
pollution.7 Greiner later determined the rate
constants for the OH radical reactions with several
alkanes using a flash photolysis-kinetic spectroscopy
technique.8 His results further supported the
importance of these radicals in air pollution processes
such as photochemical smog formation.
When organic compounds are emitted into the
atmosphere, the degradation of these compounds occur as
a result of four major reaction pathways: photolysis,

reaction with ozone, reaction with N03, and, most
importantly in the lower atmosphere, reaction with the
OH radical.9
Following Greiner's proposal10, two research
groups proposed a chain mechanism in which the hydroxyl
radical could in theory convert NO to N02.4
OH + CO -+ H + C02
H + 02 + M -+ H02 + M
H02 + NO -* OH + N02
Later studies demonstrated that this mechanism was
significant in photochemical air pollution only when CO
concentrations are sufficiently high that the rate of
the OH radical reaction with CO is comparable to its
rate with hydrocarbons.4 In polluted urban
environments, hydrocarbons were found to be involved in
a similar role to CO in the oxidation of NO to N02.4
With hydrocarbons present, chain reactions are
initiated by an OH radical attack on the parent
hydrocarbon. Organic Peroxy radicals as shown below for
CH4 are involved in the oxidation of NO to N02:

OH + CH4 - H20 + CH3
ch3 + o2 -+ ch3o2
ch3o2 + NO CH30 + NO-
ch3o + 2 -> HCHO + H02
ho2 + NO -* OH + N02
Longer hydrocarbon chains (C>4) would not follow the
reaction sequence indicated above. The products
resulting from reactions involving longer hydrocarbon
chains would become involved in other processes.11
Sources of OH Radicals in the Troposphere
The primary source of OH radicals in the unpolluted
troposphere is from the photodissociation (wavelength
<310 nm) of 03. The photolysis of 03 forms molecular
oxygen and atomic oxygen, either or both of which may be
in an excited state, depending upon the excitation
energy.11 03
03 + hv (<310nm) -* 02 + O(-'-D)

In the troposphere, the production of 0(1D) is most
significant as a source of OH free radicals through its
reaction with H20 vapor which is always present in the
0 (2D) + H20(g) -* 2OH
This very fast reaction occurs in competition with
the deactivation of the excited oxygen atom by air.4,11
0(1D) + air - 0(3P) + air
In 1 atmosphere of air at 50 % relative humidity and
298 K, approximately 10 % of the 0(1D) produced reacts
with H20 to produce OH.11
Other significant sources of OH are the reactions of
0(1D) with other hydrogen containing molecules.
O^D) + H2 -* H + OH
0(1D) + CH4 CH3 + OH
The photolysis of HONO and H202 directly produce OH in

polluted environments.
HONO + hv (<400 run) - OH + NO
H202 + hv (<360 nm) -* 2 OH
Photolysis is the primary fate of HONO.11 The
mechanisms by which HONO can be formed in the
atmosphere is not well established at this time.11
Atkinson and Lloyd12 investigated the recombination of
OH and NO as the source of HONO. The reaction rate was
determined to be 6.6 x 10-12 cm3/molec-sec. This
reaction would produce a small steady state
concentration of HONO as a result of photolysis of the
molecule. Other researchers have investigated the
following reactions as being sources of HONO
NO + N02+ H20 ** 2H0N0
2N02 + H20 * HONO + HN03
H02 + N02 -* HONO + 02
Hydrogen peroxide (H202) is typically produced in

the atmosphere through the reaction of two hydroperoxyl
free radicals.11
HC>2 + HO2 H202 + O2
Other tropospheric sources of OH are those that
are linked to the production of H atoms or H02 radicals
directly such as the photolysis of formaldehyde during
daylight hours.
HCHO + hv -* H + HCO
H + 02 -* H02
HCO + 02 - H02 + CO
This H02 formation is followed by the reaction of H02
with NO to produce OH and N02. Because H02 is closely
tied to OH through this reaction, when adequate
concentrations of NO are present, the loss of H02 is a
source of OH.
H02 + NO OH + N02
At higher altitudes in the stratosphere and mesosphere,

the photodissociation of 02 and N20 are sources of the
O(-'-D) atoms, while the photodissociation of H20 yields
OH radicals directly along with H atoms.11
In recent years, experimental and modeling
techniques have been used to estimate the concentration
of OH radicals in the troposphere.11
In 1984 Hubler et al.13, using optical absorption
techniques determined the concentration of OH radicals,
in the troposphere (Table 1).
Table 1: Hubler Study13
Environment Concentration (molecules/cm3)
Remote regions 0.984 9.84 X 105
Rural regions 0.246 2.46 X 105
Moderately polluted 1.20 9.84 X 106
Heavily polluted > 9.84 X 106
Because of its reactivity and potential for high
concentrations in polluted environments, the study of
hydroxyl radical reaction rates has become a
significant area of research in pollution chemistry.

Sources of Non-methane Hydrocarbons in the Troposphere
Although hydrocarbon emissions do occur in the
environments from natural sources such as natural gas
seepage, bacterial fermentation, and plants, the
majority of hydrocarbon emissions come from man.
Incomplete incineration, industrial solvent leakages,
unburned fuel from automobiles, incomplete combustion
of coal and wood, and petroleum processing, transfer
and use are major contributors to global hydrocarbon
emissions. Petroleum refining, and oil and gas
production account for 25% of the total non-methane
hydrocarbons released in the United States. Organic
solvents account for another 25%.11

Experimental Approach
In this study, we have observed the reaction of OH
radicals and five different alkanes described by the
R-H + OH -+ R + H20
The hydroxyl radical abstracts a hydrogen atom
from the alkane resulting in an alkyl radical and
water. At high OH concentrations with large quantities
of alkyl radicals, secondary reactions between these
two radicals will occur and have an impact upon the
determination of the OH + alkane rate constant.14 In
order to observe this effect of secondary kinetics,
modeling experiments were run using variable
concentrations of OH with five alkanes. The C2 to C5
alkanes were selected in linear and branched
configurations. A discharge flow system was used to
determine the experimental bimolecular rate constant

under pseudo-first-order conditions. The bimolecular
reaction rate is expressed as follows:
Bimolecular Reaction Rate = k[OH][alkane]
where k = rate constant for the reaction between the OH
radical and the alkane. For pseudo-first order
conditions to exist, the alkane concentration is more
than a factor of 10 higher than the OH radical
concentration allowing the alkane concentration to
remain essentially constant throughout the reaction.
The rate of loss of OH over time is equal to the rate
constant (k) times the concentration of each species:
-d[OH]/dt = k[OH][alkane]
The alkane concentration will remain essentially
constant. This value can be factored into the rate
constant k making it k1.
k' = k[alkane]

Integrating the results gives us
d[OH]/[OH] = k'dt or
ln[OH] = k't + c
which is a linear equation. Plotting the natural log
of the OH concentration versus time will allow us to
determine the pseudo-first order rate constant for the
reaction assuming a constant concentration of alkane.
The slope of the plot k' vs [alkane] is equal to the
bimolecular rate constant between OH + alkane. This
method corrects for OH wall losses in the system or
The kinetics of the reaction occurring in the flow
system was observed by measuring the change in the
concentration of the reactant in smaller concentration
with the time of reaction. In studying the kinetics of
the OH + alkane reaction, the reactant monitored is
typically the hydroxyl radical. The concentration of
the hydroxyl radical can be measured by fluorescence in
the 310 nm wavelength region. The study of OH + alkane
kinetics can be observed by a variety of techniques.

This investigation involves the use of a discharge flow
technique which will be summarized in Appendix B.
Other experimental methods include flash photolysis,
pulse radiolysis and modulation phase shift. A variety
of detection devices can be used to observe the OH
concentration. These include resonance absorption,
resonance fluorescence, electron paramagnetic
resonance, mass spectrometry, laser-induced
fluorescence and laser magnetic resonance. The
detection limits of these methods vary from 109 to 1014
molecules/cm3, with resonance fluorescence, the most
widely used method, being near the low end range and
resonance absorption on the high end.
Current data available for studying the rate
constants for OH + alkanes at room temperature are not
consistant.14 Discrepancies in the evaluation of rate
constants are often attributed to uncontrollable
effects which occur during the experiment. One
potentially significant effect is that of secondary
reactions which occur between the OH radical and the
alkyl radical product. Other factors accounting for
this variability may include systematic experimental

errors or system impurities.
In reviewing similar studies in the literature for
the determination of the OH + alkane rate constants,
Bonilla14 concluded that secondary effects of the OH +
alkyl radical were not observed due to the experimental
conditions used by other researchers. Secondary
effects were not seen because the range of the OH
concentrations used experimentally was too low.
The effects of secondary reactions have been
considered negligible in past kinetic studies of the OH
+ alkane reaction. According to Anderson and
Stephens15, the initial OH concentrations must be
somewhat high to show this secondary effect.
In order to better evaluate the impact of
secondary kinetics upon the determination of the rate
constant, a complex modeling scheme was run for five
branched and unbranched alkanes. By varying the initial
OH concentration, the significance of the secondary
effects would be seen by examining the apparent
bimolecular rate constant for the reaction of the
alkanes + OH radical. Without the influence of
secondary effects, the apparent bimolecular rate

constant should remain relatively unchanged at all OH
concentrations. Variation in the apparent bimolecular
rate constant would indicate that secondary reactions
have a significant effect. In addition, we will look
more closely at why other researchers may have not seen
the significance of these effects by attempting to
model the experiments done in their laboratories.
Laboratory Experiments
Details of the laboratory portion of this research
can be found in Appendix B. The experimental design
initiated by Juan Bonilla14, was followed in an attempt
to collect laboratory data using a discharge flow
resonance system for the same five alkanes used in the
modeling simulations of this research. A series of
system difficulties ended the research prematurely.
Particulars are given in Appendix B.

Modeling Experiment
To better understand the possible mechanisms and
rate constants which could affect the OH + alkane
reactions in the flow system and to better estimate the
initial concentrations of N02 and H atoms to be used in
the flow system experiment, a computer modeling
experiment was run for each of the hydrocarbons that
would be studied using the flow system.
Acuchem/Acuplot is a computer program for modeling
complex reactions. It was developed at the National
Bureau of Standards in Gaithersburg, Maryland, by
Walter Braun and John Herron of the Chemical Kinetics
Division and David Kahaner of the Scientific Computing
Division. The Acuchem program reads an input file
which may contain up to 80 different reactions
comprised of up to 40 different chemical species. The
selected reactions, reactant concentrations and output
time for the reactions to occur are entered into an
Acuchem input file (Figure 1). A model was constructed
which would most closely recreate the conditions that
exist in the experimental flow system. Five models

Figure 1: Acuchem Input File
1) ,N02+H=N0-H3H, 1.32E-10
2) ,0H+0H=H20-0,1.8E-12
3) ,0H+N02=K0KC2,0.83E-11
4) ,OH+NO=HONO,.34E-11
5) ,0H+0H=H2,02,C.28E-11
6) ,0H+0=H+C2,3.3 E-ll
7) ,0H+H02=H20+02,1.1S-10
8) ,OH=WALLOH,2.5
9) ,0+N02=NO-02,9.7E-12
10) ,0+H02=OH-F02,5.7E-11
11) ,0+N02=NC3,0.9E12
12) , O-t-N0=NC2,0.94E-12
13) , HO2-rNO=CH-rN02,8.3E-12
14) ,HO2+NO2=HC2NO2,0.96E-12
15) ,H02+H02=H2C2-r02,1.6E-L2
16) ,H02=KALLH02,2.5
17) , 0+H202=OH-rH02,1.7E15
18) , 0K-rH2C2=H20H02,1.7E-12
19) ,O+NO3=C2-FNC2,1.0E-ll
20) ,NO+N03=NC2-rNC2,2.7E-ll
21) ,H+02=HC2,1.79E-15
22) ,N02-FH03=N205,1.2E-12
23) ,N205=N02-rN03,3.8E-3
24 ) 0H+H02NC2=H20+DUMMY, 0.5E-11
25) , DUMMY=N02*r02 lE-i-20
26) ,HONO-rHON02=H2O-i-DUMMY2,2.71E-17
27) , DUMMY2=NC2-rN02,1E-F20
28) , N03=02-F-N0,2.7E16
. 29) N03-rN02=N0-rDUMMY3,4 OE-16
30) DUMMY3=N02-!-C2,1E-F-20
31) , N03+N03=02-FDUMMY4,2.3E-16
32) ,DUMMY4=N02-rN02,1E+20
33) ,H02N02=H02+N02,3.9E-4
34) , HO2+N02=H0N0-F02,3 OE-15
35) ,H02-r-N03=H0N02-r02,4.3E-12
36) ,H02+N03=0H+DUMMY5,4.3E-12
37) ,DUMMY5=N02+02,lE-r20
38) ,H=WALLH,2.5
39) , 0=WALL0,2.5
40) ,H+H02=H2-F02,5.6E-12
41) ,H+H02=OH+OH,7.2E-11
42) ,H+H02=H20+0,2.4E-12
43) ,0H+H2=H20-FH, 6.7E-15
44) , OH+HON02=H20-i-N03,9.91-14
45) , 0H+H0N0=H20+NC-2,4.9E-12
46) , 0H+N03=H02-rN02,2.3E-11
47) ,0H+neoC5H12=neoC5Hll-H2O,0.984E-12
48) ,0H+neoC5Hll=neoC5H110H,3E-11
49) ,O('D)-FH2O=OH-rOH,2.3E-10
50) , O ('D) -rK20=N2-rG2,4.4E-11
51) ,0('D)-rN20=N0-rN0,7.2E-ll

Figure 1: Acuchem Input File
52) , 0 (' D)+H20=0+H20,1.95E12
53) ,0('D)+N20=OtN2O,1.17E12
54) ,0('D)+He=0+He,6E-16
H, 3.77E+10;
NO,9.44 001E+12;
OH, 1.2E4-12;
H0N02,2.421E+12 ;
WALLOH, 2.273E-rll;
H02N02,4.2 65E+08;
DUT4MY2,1.625E-12 ;
neoCSHHOH, IE-30;
OUTPUT TIMES (last time followed by semi-colon)

were developed that would represent the reactions
occurring in the flow system for OH and five different
alkanes including ethane, propane, butane, isobutane,
and neopentane. The chemistry used in the model
included a core of 46 reactions containing 19 different
species that would most likely occur under the
experimental flow system conditions (Table 1, Appendix
A) as well as various reactions involving interactions
between the alkane of interest and these core
reactants. The input file is processed and through a
variety of iterations, the program solves the resulting
system of differential equations. Acuchem then
generates an output file containing species
concentration versus time for the various reactants.
The Acuplot portion of this program displays the output
file containing the modeling results (Figure 2).
The number of hydrocarbon reactions modeled by the
software are variable for each alkane depending upon
experimental rate constants available in the
literature. Appendix A also contains the reactions
used in the model for each of the five hydrocarbons
studied. Very little information was available for the

Figure 2: Acuchem Output File
0. OOOE+OO 4.5B3E+11 1.2B7E+11
3.900E-03 4. 212E+11 1.055E+11
TIME 0 H0N02
0.OOOE+OO 3.649E+10 1.854E+11
3.900E-03 3.686E+10 1.9S4E+11
0.OOOE+OO 9.114E+08 7. 041E+10
3. 900E-03 1.042E+09 7. 937E+10
0.OOOE+OO 2.474E+08 2. 422E-13
3.900E-03 2. 481E+08 2.174E-13
0.OOOE+OO 1.290E-14 2. 797E+10
3.900E-03 1.618E-14 2. 911E+10
TIME neoCSHi1 neoC5Hl
0.OOOE+OO 1.000E-30 1. 000E-30
3.900E-03 2.448E+11 3. 67BE+10
0.OOOE+OO 8. 030E+14
3.900E-03 B. 0B0E+14
1.719E+12 1. 114E+12 7. 407E + 10
1.725E+12 7.404E+11 3.&53E+11
HONO H2D2 02
1.374E+11 9.347E+10 2. 391E+10
1.558E+11 1.022E+I1 2.929E+10
N03 H02N02 WALLH02
3.291E+0B 4.348E+06 2.072E+07
3.614E+0S 5. 873E+06 3. 032E+07
6. 903E-15 6.033E-16 2. 492E-19
8.378E-15 6.0BBE-16 3. 003E-19
WALLO H2 neoC5H12
1.215E+09 9.548E+06 8. OOCE-t-13
1.575E+09 1.205E+07 7.972E+13
1 0 ( 'D) N2D N2
1.OOOE-30 1.OOOE-30 1. 000E-30
1.0O0E-30 1.000E+30 1.000E-30

neopentane modeling mechanisms (Table 6, Appendix A).
The neopentyl radical + OH rate constant was estimated
by a comparison to the ethyl radical + OH rate
constant. The slight difference was due to the
neopentyl radical being much more sterically hindered.
To accurately portray the chemistry in the flow
system at each of the eight ports when the alkane gas
is added to the system, the 46 core reactions were run
in a modeling experiment without the addition of the
alkane. This modeling experiment was run at three
different levels of OH concentration. Reaction times
for this modeling experiment were determined by the
amount of time it took for the gas to travel from the
point where N02 was added into the system to its
arrival at each of the eight alkane inlet ports.
Knowing the distances traveled and the gas velocity
through the flow system, the model could predict a
snapshot of the chemistry that would exist in the
system at each port when the alkane is added (Figure
3). The gas velocity through the system had previously
been determined experimentally using the flow system

Figure 3: Reactant Concentration Determination
Determination of Reactant Concentrations
H in Modeling Simulation

From the above modeling results, eight different
input files are created for each of the three OH
concentrations. Each of these twenty-four input files
is modeled using nine concentrations of the alkane
studied. The alkane concentrations range from 1 to 9 x
1013 molecules/cm3.
The reaction times for these experiments are
determined from the distance between each alkane inlet,
port and the point of detection. Once again these
times can be determined mathematically by the gas
velocity through the system and the distance travelled
to the point of detection (Figure 4).
In determining the pseudo-first order rate
constant from the modeling experiment, the natural log
of each OH concentration is plotted against reaction
time for the nine alkane concentrations. The slopes of
each of these nine plots correspond to the pseudo-first
order rate constant at each alkane concentration. The
plot of the pseudo-first order rate constant versus the
alkane concentration should be a linear relationship.
The bimolecular rate constant for the OH + alkane
reaction is equal to the slope of this line.

Figure 4: Reaction Time Determination
Determination cf Reaction Times
in Modeiina Simulation
Velocity through System = 2704 cm/sac
2 3 4 5 6
7 8
5.3 10.5 CM
0.CC57 sec ------
0.C073 soc ------------
0.0C93 sec ------------------
0.0112 sec ------------------------
0.0131 sec ------------------------------
0.01 SO sec ------------------------------------
0.0163 sec -------------------------------------------

In addition, modeling experiments were run to
determine the significance of various reactions upon
the determination of the bimolecular rate constant. By
eliminating specific alkyl radical reactions from the
mechanism and determining the bimolecular rate constant
for the model without these reactions, an assessment of
how each of these reactions impacts the determined
bimolecular rate constants is made.
In the literature, other researchers1'9'17 have
failed to see the importance of secondary effects upon
the determined bimolecular rate constant for OH +
alkane. A modeling simulation was set up to determine
why these secondary effects were not seen in these
previous experiments.

Modeling Results
The kinetics results from the modeling runs were
obtained after extracting OH concentration values from
the Acuplot files. Using a Quattro Pro spreadsheet,
the natural log of entered OH concentrations was
determined. The dependent and independent variables
were designated as In[OH] and time respectively.
Modeling simulations were run for each alkane at three
OH concentrations. Nine concentrations of each alkane
were examined. Data were collected for each modeling
simulation at eight reaction times which approximated
flow system conditions. A linear regression analysis
was done for the In[OH] versus reaction time for nine
alkane concentrations. The slopes of these plots were
equal to the pseudo-first order rate constants. The
apparent bimolecular rate constant for the OH + alkane
reaction was determined from the slope of a linear
regression analysis of the pseudo-first order rate

constants versus the alkane concentrations.
Rate constants for the simulation of the core
reactions and for each OH + alkane simulation were
obtained from various literature sources.14'19'20'21,22
Many of the rate constants were estimated values based
upon similar reaction mechanisms of known rate
constants (Appendix A). Estimations for rate constant
values were often determined by comparisons to other
similar mechanisms. For the OH + isobutyl reaction
rates, W. Tsang22 indicated that no experimental data
exist for these reactions. The rate constant for OH +
isobutyl - isobutene + H20 was designated as 2 x 10-11
molec/cm3-sec. This mechanism is considered a
disproportionation reaction and Tsang recommends a rate
constant one half of that for OH + n-propyl radical.
For the OH + isobutyl -* isobutyl alcohol, Tsang
assigns a rate constant for the combination process of
4 x 10-11 molec/cm3-sec. At sufficiently high
temperatures, the hot butanol adduct may decompose to
form isopropyl and hydroxymethyl radicals. This
possibility was ignored in this work. Similar
estimations were made for other rate constants used in

the model as specified in the mechanism tables in
Appendix A.
The selection of reaction mechanisms used in the
modeling simulation was based upon those reactions
which could potentially occur in the experimental flow
system. The Acuchem model is designed to handle
bimolecular reactions resulting in two products.
Several reaction mechanisms which resulted in three
products were relevant to the modeling simulation. In
order to fit these reactions into the Acuchem model,
dummy compounds were used to represent two of the
products in a three product reaction. The reaction
immediately following these reactions would show the
dummy compound forming two products in a very fast
reaction. This would allow the reaction mechanism to
stay within the format of the model. The core reaction
mechanisms can be seen in Table 1, Appendix A.
In order to evaluate the significance of secondary
effects upon the determined bimolecular rate constant
in this study, a comparison of the bimolecular rate
constant calculated from the model with the absolute
rate constant used in the modeling simulation must be

made. If no secondary effects exist, the calculated
bimolecular rate constant and the absolute bimolecular
rate constant for the OH + alkane used in the model
should be the same. It is also apparent from the
results of this study that the concentration of OH
present in the model has an important impact on the
calculated bimolecular rate constant. Without the
influence of secondary reactions, the determined rate
constants would be consistent at all OH concentrations.
The results of the modeling simulation reveal a .
difference in the calculated bimolecular rate constant
between the three levels of OH concentrations used in
this study for all cases.
In Figure 5 an OH decay plot for OH + ethane is
shown. The OH decay plot compares the OH signal for
low, high and very high OH concentrations at an ethane
concentration of 9 x 1013 molecules/cm3. The model
concentrations were normalized with respect to the
concentration at the smallest reaction time within each
set of points and the natural log of this signal was
plotted against reaction time. The non-linear effect
of OH loss at very high OH concentrations compared to

ln[OH] (Normalized)
Figure 5: OH Decay Plot
OH + Ethane
Time (sec.)
Low [OH] + High [OH] -^Very High [OH]
[Ethane] = 9E13 molec/cm3

the loss at low OH concentrations is quite apparent.
The curvature appears even more obvious at longer
reaction times since more ethyl radicals will be formed
at longer reaction times. The OH radicals will be more
likely to react with the ethyl radicals and therefore a
decrease in OH concentration will be observed. This
behavior is consistent for all alkanes studied.
Appendix C contains plots of the pseudo-first
order k versus alkane for each of the five modeling
experiments. As previously mentioned, the slopes of
these plots are equal to the determined bimolecular
rate constant for each OH + alkane reaction. Appendix D
contains two tables for each of the OH + alkane
modeling experiments studied. The first table in each
set shows the determined bimolecular rate constants at
each OH concentration based upon a linear least squares
analysis of the complete modeling data. The second
table for each set shows the bimolecular rate constants
for the same modeling experiments based upon an initial
slope analysis of the first two points. The initial
slope analysis was performed on each modeling data set
in order to give a better estimate of the determined

bimolecular rate constant. Due to the apparent
curvature of the plots in Appendix C, a linear least
squares analysis would not give an accurate
representation of the determined bimolecular rate
constant. At low alkane concentrations, the alkyl
radicals have a more significant impact on the
chemistry of OH + alkane. At higher alkane
concentrations, an abundance of alkane molecules may
mask the full effect of the alkyl radicals. This can be
seen by observing the plots in Appendix C. The most
change or steepest slope is seen at low alkane
concentrations. As the alkane concentration increases,
the slope of the plot begins show curvature. Thus, an
initial slope analysis on the first two points in the
curve would give a more accurate assessment of the
determined bimolecular rate constant when it is most
impacted by the alkyl radicals.
The tables in Appendix D show that OH
concentration has a significant impact upon the
determined bimolecular rate constant. Without the
presence of secondary effects, the bimolecular rate
constant should be the same at all OH concentrations.

One also observes a consistant increase in the
determined bimolecular rate constants in going from
modeling experiments run at low OH concentration to
modeling experiments run at very high OH
concentrations. In all cases, the determined
bimolecular rate constants become increasingly larger
with increased OH concentration. In the case of propane
where the initial slope analysis was not used (Table 3-,
Appendix D), the curvature of the line may have
impacted the calculation at high OH concentration. At
very high OH concentrations, the differences are most
significant and range from a 5.5 to 85.9% difference as
compared to the k used in the modeling experiments.
The OH + ethyl radical reaction rate constant in
the model is approximately 155 times faster than the
rate constant for OH + ethane. At higher ethane
concentrations, more ethyl radicals are produced
causing the competition between the ethyl radical and
ethane for OH radicals to become somewhat significant.
Closer examination of Figure 1, Appendix C
indicates almost no curvature in the plots of the
pseudo-first order k versus ethane concentration. The

curvature becomes slightly more pronounced with
increasing OH concentration. Compared to other alkanes
in this study, the curvature in the ethane + OH
reaction plots is minimal. This may be due to the rate
constant of OH + ethane being small relative to the
rate constant for OH + ethyl radical. Despite the lack
of curvature in these plots, the differences in the
determined bimolecular rate constants between each OH
concentration is significant.
Figure 2 of Appendix C shows the plot of the
determined pseudo-first order rate constant versus
propane concentration. Tables 3 and 4 of Appendix D
compares the rate constants determined from the
modeling experiments to the rate constant used in the
The calculated bimolecular rate constants may vary
depending upon the reaction mechanism used in the
model. Some of the reactants observed in this modeling
study had more reactions available in the literature
than others. Note that the determined bimolecular rate
constant with propane at very high OH concentration was
only slightly higher than the rate constant at propane

and high OH. The rate constant for propyl + OH is
approximately 72 times faster than the rate constant
for propane + OH. The increased number of reaction
mechanisms used in this model may have had an impact
upon the results.
The modeling simulation between butane and the
three levels of OH concentration showed a behavior that
was much different from the other modeling experiments.
As the OH concentration was increased, the determined
bimolecular rate constant increased more significantly
than was seen in other modeling experiments. Tables 5
and 6 in Appendix D shows the differences between these
calculated bimolecular rate constants and the
bimolecular rate constant used in the model. Figure 3
of Appendix C shows that a higher degree of curvature
exists in the three plots of the determined pseudo-
first order rate constants versus butane concentration
as compared to other models. Part of these differences
from other models could be explained in terms of the
source of the OH + butyl rate constant. This modeling
simulation was based upon the work of Juan Bonilla14.
The rate constant which was used in his experiments for

the butyl radical with OH is very fast compared to the
butane + OH rate constant. The other alkyl radical and
OH rate constants used in this modeling simulation were
estimated by Tsang20'21,22.
Tables 7 and 8 of Appendix D compares the rate
constants between the k in the model and the calculated
bimolecular rate constant for isobutane + OH. Some
unusual behavior is occurring in this modeling scheme..
Two of the calculated rate constants are lower than the
k used in the model. Although the very high OH rate
constants differ the least from the model rate
constant, Figure 4 of Appendix C reveals a higher
degree of curvature in the calculated pseudo-first
order rate constant versus the isobutane concentration
plots. This curvature is most likely responsible for
the unusual behavior seen in the table above. The
results determined using the initial slopes analysis
clarifies the confusion.
The neopentane model contains 2 reactions in
addition to the core 46 reactions used in the model
simulation. Although the neopentane + OH reaction rate
used in the model was taken from the literature, the

neopentyl + OH reaction rate had to be estimated.
Neopentane is a sterically hindered molecule in terms
of its reactivity. The best estimate for the neopentyl
reaction rate was based upon its similarity to ethyl +
OH model in reactivity. The modeling simulation results
for the three levels of OH are seen in Figure 5 of
Appendix C. Table 5 of Appendix D compares the
calculated bimolecular rate constant to the rate
constant used in the model.
In all cases, the determined bimolecular rate
constant for alkane and low OH did not differ
significantly from the absolute bimolecular rate
constant used in the modeling simulation. Secondary
kinetic effects are often regarded as insignificant in
other research. These results demonstrate that higher
concentrations of OH are necessary for significant
secondary reactions to be observed.
The curvature in the pseudo-first order k versus
alkane plots became more pronounced with increased
carbon number and in some cases with increased
branching in the alkane. This did not hold true for
neopentane which is similar to ethane in reactivity.

Reaction Mechanism Importance
An assumption has been made with respect to the
source of the secondary effects in these experiments.
The fast reactions of the OH + alkyl radicals are
thought to be key factors in the determination of the
bimolecular rate constant. In order to verify that the
OH + alkyl reaction, was the reaction which most
impacted the OH + alkane reaction rate, a modeling
simulation was designed using isobutane and very high
OH. The same model which was run for isobutane + very
high OH in the above simulations was run three more
times. In each of these three simulations, one or more
of the reactions involving a reactant + isobutyl was
omitted from the model. Figure 6 illustrates the
pseudo-first order k versus isobutane plots for each of
these modeling schemes. Table 2 shows the calculated
bimolecular rate constants under the three modeling
conditions. The plots in Figure 6 illustrate that
under two of the conditions, the modeling results are
nearly unchanged compared to the complete very high OH
+ isobutane modeling run. A good amount of curvature

Pseudo First Order K (1/sec)
Figure 6:
Kinetics Results for Reaction Effects
Very High OH Concentrations: Isobutane
OH + Isobutane K = 2.19E-12
VHOH Full Run
*"VHOH less ISB + ISB
--VHOH less N02/N03
"VHOH less OH+lsobutl
[Isobutane]E13 moJ/cm3

still exists in these plots. Small differences are
seen in the calculated bimolecular rate constants. In
the modeling simulation where the OH + isobutyl
reaction was omitted, a near linear relationship exists
between the pseudo-first order k and isobutane
concentration. From these plots it has been shown that
secondary effects are due to the presence of the OH +
isobutyl reaction in the modeling simulation.
It had been assumed that if secondary effects were
not present in the model, the determined bimolecular,
rate constant for OH + alkane should be very close to
the rate constant used in the model. The calculated
bimolecular rate constant is almost 20 percent smaller
than the model k value. Other processes may be
occurring which produce OH radicals at a relatively
fast rate. Without the presence of the fast OH +
isobutyl reactions in the model, more OH radicals will
be present in the model to follow other reaction

Table 2: Comparison of OH + Isobutane Under
Four Modeling Conditions.
Model k = 2.19 x 10"12 cm3/molecule-sec
Model Variation Very High OH Bimolecular Rate Constant (xlO-12) (cm3/molec-sec) % Difference from Model k
Full Model 2.07 -5.5
Less Isobutyl + Isobutyl Rxn 2.35 +7.2
Less Isobutyl + NO, N02 Reactions 2.42 +9.9
Less Isobutyl + OH Reaction 1.79 -21.8
Table 3: Comparison of OH + Isobutane Under Four
Modeling Conditions.
(based upon two point curve)
Model k = 2.19 x 10~12 cm3/molecule-sec
Model Variation Very High OH Bimolecular Rate Constant (xlO-12) (cm3/molec-sec) % Difference from Model k
Full Model 2.52 +15.1
Less Isobutyl + Isobutyl Rxn 2.87 +31.1
Less Isobutyl + NO, N02 Reactions 2.81 +28.3
Less Isobutyl + OH Reaction 1.76 -19.6

The curvature seen in the experimental modeling
plots is attributed to the effects of secondary
reactions. The apparent curvature was investigated
further to determine its cause.
The determined bimolecular rate constants in Table
3 show the results based upon a two point initial slope
analysis. The differences in the slopes are even more
pronounced in this more accurate representation of how
secondary effects impact the determination of the
bimolecular rate constant.
Curvature Effects
A modeling simulation was carried out involving OH
+ butane with and without the OH + butyl radical
reaction. The two OH loss plots are seen in Figures 7
and Figure 8. The results demonstrate that the OH loss
is greater for reactions involving high butane
concentrations when the OH + butyl radical is present
as seen by the scale change on the y-axis. In plotting
the In[OH] versus time, the slope is equal to the
pseudo-first order rate constant or the loss of OH /sec.

rr> ('i
Figure 7:
7.5 9.5
11-2 13.1 15.0 15.S
time (msec')
[Butane] fx

C 2 5 G
:yl R=

II101 ill
No G
Figure 8:

Table 4: OH Decay Comparison.
[Butane] molecules/cm3 OH Decay with OH + Butyl Rxn OH Decay Without OH + Butyl Rxn
1E+13 33.1 14.2
2E+13 64.7 28.3
3E+13 95.2 42.5
4E+13 124.3 56.8
5E+13 152.1 71.3
6E+13 178.0 85.7
7E+13 201.3 100.1
8E+13 221.8 114.3
9E+13 239.4 128.2
Table 4 shows the OH decay or loss for
the modeling simulation with and without the OH +
butyl reaction. The middle column shows the OH loss as
seen in the modeling simulations. The third column
represents what OH loss would be without the secondary
effects of the OH + butyl reaction. The higher OH loss
is thought to be a result of not only the reaction
between OH + butane but also between the highly
competitive reaction of OH + butyl radical. This loss
is almost twice what would be expected for a reaction
free from secondary effects. The determined

bimolecular rate constant for OH + butane excluding the
OH + alkyl reaction was 1.42 x 10-12 cm3/molec-sec.
This rate constant is somewhat lower than the rate
constant used in the model in Appendix A.

F. Tully1'2'3 has investigated several reaction
rates for various alkanes and the OH radical.
Experimentally, he repeatedly acknowledged that
secondary effects do occur, however, he indicated that,
their impact upon the determined bimolecular rate
constant is insignificant. To determine why Tully was
unable to observe secondary effects, a modeling
simulation was created which would closely simulate his
flash photolysis experiments. Using the modeling
reaction mechanisms for butane above, conditions were
altered to reflect the flash photolysis experiment
carried out by F. Tully and A. T. Droege1 at Sandia
National Laboratories in 1986. Reaction times were
altered to coincide with the times measured in a flash
photolysis experiment. Initial concentrations of n-
butane, 0('D), H20 and N20 were matched to those
specified in the literature. The results are displayed
in Figure 9. The determined bimolecular rate constant

Pseudo First Order K (1/sec)
Figure 9:
Kinetics Modeling Results for Flash
Photolysis Experiment: Butane
OH + Butyl K = 0.5E-9 cm3/molec-sec
OH + Butane K = 1.7E-12 cm3/molec-sec
[OH] 0.1 10 msec
[Butane] (x E14 molec/cm3)
Determined Rate Constant (K) = 1.8E-12 cm3/mo!ec-sec

of 1.8 x 10-12 cm3/molec-sec. is slightly higher than
the rate constant used in the model. This difference
can probably be explained in terms of slight secondary
effects. The relationship between the pseudo-first
order rate constant and butane concentration appears to
be linear. Modeling simulations were then run with the
concentration of OH increased by a factor of five and
then by a factor of ten. These results can be seen in
Figure 10. With increasing OH concentration, an
increased bimolecular rate constant was calculated.
Curvature in the line is only slightly more apparent.
The differences in the calculated bimolecular rate
constants is a significant indicator of secondary
effects. It is assumed that the concentration of OH
radicals used in the original experiment was too low to
see these secondary effects.
The Tully experiments begin with a larger amount
of butane. Because the flash photolysis process is
faster, more butane is present to compete with butyl.
This will also minimize the secondary effects of the OH
+ Butyl process.
In looking more closely at these plots, it becomes

eudo First Order K (1/sec)
Figure 10:
Kinetics Modeling Results for Flash
Photolysis Experiment: Butane
OH + Butyl K = 0.5E-9 cm3/moloc-coc
""Low OH Medium OH ^ High OH

apparent that the intercept on the y-axis is higher
compared to the intercept in the flow system
experiment. In the flow system experiment, the
intercept is related to the wall loss of OH radicals.
This wall loss had experimentally been determined to be
2.5/sec. and corresponds to the y-intercept of the
pseudo-first order rate constant versus alkane
concentration plot. The modeling experiments for OH +
butane showed the y-intercept to be 3.3, 10.2 and 15.4
for low, high and very high OH respectively. In the
flash photolysis modeling experiments the y-intercept
was found to be 76 and 84 for the medium and high OH.
An intercept of about 30 existed for the low OH
experiment. Experimentally, these higher intercepts
can be attributed to diffusion effects in the flash
photolysis process, however, the modeling reaction
mechanisms did not include a mechanism for diffusion
effects. These higher intercepts were present by
virtue of the experimental conditions and initial
concentrations without diffusion effects being
To better understand the differences in the

results between the flow system model used in this
experiment and F. Tully's flash photolysis results3, a
series of first order decay plots were created
depicting OH losses and OH formations for various
reaction processes. Graphs of all the OH loss and OH
formations reactions can be seen in Appendix E. The
most important OH loss reactions for both systems are
the OH + butane (Figure 11) and the OH + butyl (Figure.
12) reactions. Note that these are the only plots
where the both systems have similar scales. Other OH
reaction processes appear less significant in the flash
photolysis experiment even if they remains somewhat
important for the flow system experiment. The
significance of this lies in the fact that Tully used
lower concentrations of OH in his experiments. Figures
11 and 12 show that the reaction time is longer in the
flow system experiments. There is a significantly
higher level of OH present in the flow system model at
4 milliseconds compared to the flash photolysis
experiment. Along with this observation is a more
dramatic decrease in OH through the course of the flow
system experiment. Appendix F shows that a variety of

I(|0M| |I3i lliIf II*|
Figure 11:
OH -r Butene Reaction
- i
0 1 2 3 4 5 6/59 lb l'l 12 13 14 15 15 17 15
..i~c >
c: u c.-
i iuio-i i uv
:< ox
I \. O i;uC>ii i ^j
Figure 12

OH Eutyi Reaction

I -
n c i
0 1 2
5 6
7 S -9 10 1
12 13 1^15 15 17 15

behaviors exists in OH formations and losses between
the two experimental methods. Generally, OH is still
being formed at the end of the flash photolysis
experiment. In the flow system experiments, much higher
levels of OH exist quite early and a decrease in OH
formation is observed through the course of the
experiment. A similar observation holds true in terms
of the OH loss plots. A higher level of OH exists early
in the experiment and a dramatic decrease is observed
through the course of the experiment.

Modeling results consistently showed that
secondary effects will impact the determined
bimolecular rate constant for the reaction between the
OH radical and the five alkanes observed in this study.
The results demonstrate that the overall effect is most
pronounced at very high OH concentration levels. The
variations in the bimolecular rate constants in the
literature along with what was observed in this
research clearly indicate that an uncertainty exists in
these OH alkane rate constants. In this study we
observe that the fast OH + butyl process causes the
most significant change in k from the k used in the
model. The OH butane process in this research was the
only simulation where the rate constant for OH + butyl
(i.e. alkyl) was from a source other than the Tsang
references.20'21'22 Bonilla's14 very fast OH + butyl
rate constant reveals how large an impact secondary
processes can have on the bimolecular rate constant
under certain conditions. In each simulation an

observation can be made as to how higher OH
concentrations effect the determined rate constant.
The OH + alkyl radical reaction has been shown to
have the most dramatic effect upon the determination of
the apparent bimolecular rate constant in this modeling
simulation. Omitting this reaction from the model
produced a near linear representation of the pseudo-
first order k versus alkane concentration plot. The
removal of other reactions involving the alkyl radical
from the simulation had little or no effect upon the
determined bimolecular rate constant.
The experiments of F. Tully did not verify the
significance of secondary processes on the
determination of the bimolecular rate constant. This
was most likely due to both the experimental method
used as well as and the concentration of OH used in
these experiments. The Tully experiment was modeled
based upon the information available in the literature.
Secondary effects were not evident in our modeling
results. With the same experiment modeled at
concentrations of OH five and ten times higher,
secondary effects became more evident. It is evident

that Tully would not be able to see the impact that
secondary effects had upon his determined rate
constants at the concentrations used in his research.

Table 1: Core Reactions9'15
Reactions cm3/molec-sen
N02 + H = NO + OH 1.39E-10
OH + OH = H20 + 0 1.80E-12
OH + N02 = HONO 2 8.30E-12
OH + NO HONO 3.40E-12
OH + OH = H202 2.80E-12
OH + 0 = H + 02 3.30E-11
OH + H02 = H20 + 02 1.10E-10
OH + = WALLOH 0.25E+01
0 + N02 = NO + 02 9.70E-12
0 + H02 = OH + 02 5.90E-11
0 + N02 = N03 9.00E-12
0 + NO NO 2 9.40E-12
H02 + NO = OH + N02 8.30E-12
HO 2 + N02 = H02N02 9.60E-13
H02 + H02 = H202 + 02 1.60E-12
H02 + =s WALLH02 0.25E+01
0 + H202 OH + HO 2 1.70E-15
OH + H202 = H20 + HO 2 1.70E-12
0 + N03 = 02 + NO 2 l.OOE-11
NO + N03 = N02 + N02 2.90E-11
H + 02 = H02 1.79E-15
N02 + N03 = N205 1.20E-12
N205 + = N02 + N03 3.80E-03
OH + H02N02 = H20 + DUMMY 4.60E-12
DUMMY = NO 2 + 02 1.00E+20
HONO + H0N02 = H20 + DUMMY2 2.71E-17
DUMMY2 = N02 + N02 1.00E+20
N03 + = 02 + NO 2.70E-16
N03 + N02 = NO + DUMMY3 4.00E-16
DUMMY3 = N02 + 02 1.00E+20
N03 + N03 = 02 + DUMMY4- 2.30E-16
DUMMY4 = N02 + N02 1.00E+20
H02N02 = HO 2 + N02 3.90E-04
H02 + N02 = HONO + 02 3.00E-15
H02 + N03 = 02 + HONO 2 4.10E-12
H02 + N03 = OH + DUMMY5 4.30E-12
DUMMY5 = N02 + 02 1.00E+20
H + = WALLH 0.25E+01
0 + = WALLOH 0.25E+01
H + HO 2 = H2 + 02 5.60E-12
H + HO 2 = OH + OH 7.20E-11
H + H02 = H20 + 0 2.40E-12
OH + H2 = H20 + H 6.70E-15
OH + HONO 2 = H20 + N03 4.60E-12
OH + HONO = H20 + N02 4.90E-12
OH + N03 = H02 + N02 2.30E-11

Table 2: Ethane Reactions.
Reactions cm3/molec-sec
OH + ETHANE s ETHYL + H20 2.57E-13
OH + ETHYL = ETHYLENE + H20 4.00E-11*
OH + BUTANE = BUTYL + H20 1.70E-12
0 + ETHANE = ETHYLENE + OH 2.65E-16
ETHYL + H = ETHYLENE + H2 3.00E-12
ETHYL + H202 = ETHANE + HO 2 2.80E-15*
ETHYLENE + OH C2H3 + H20 1 4 4 E-16
Table 3: Propane Reactions.
Reactions cm3/molec-sec
H + PROPANE = PROPYL + H2 4.70E-17
0 + PROPANE = PROPYL + ' OH 2.58E-15
OH + PROPANE = PROPYL + H20 1.10E-12
H2 + PROPYL = PROPANE + H 6.31E-21
02 + PROPYL = C3H702 5.64E-12
H + PROPYL = PROPYLENE + H2 3.00E-12*
HO 2 + PROPANE = PROPYL + H202 1.29E-25*
0 + PROPYL = HCHO + C2H5 2.29E-11
0 + PROPYL = C2H5CHO + H 1.37E-10
H202 + PROPYL = HO 2 + PROPANE 6.17E-17
- Values based on estimation.

Table 4: Butane reactions.
Reactions cm3/molec-sec
OH + BUTANE = BUTYL + H2 0 1.70E-12
OH + BUTYL = ROH 5.00E-09
OH + BUTYL = BUTENE + H20 5.00E-09
OH + ROH = ALCOHYL + H20 7.00E-12
OH + OCTANE = OCTYL + H20 9.10E-12
OH ' + BUTENE = ALCOHYL 3.00E-11
0 + BUTANE = BUTYL + OH 2.20E-14
N02 + BUTYL = BUTYLNO 2.50E-11
Table 5: Isobutane Reactions.
Reactions cm3/molec-sec
OH + ISOBUTYL = ROH 4.00E-11
OH + ROH = ALCOHYL + H20 7.00E-12
OH + OCTANE = OCTYL + H20 9.10E-12
ISOBUTANE + HO 2 = ISOBUTYL + H20 2.75E-14
Table 6: Neopentane reactions
Reactions cm3/molec-sec

Comparison of Two Widely Used Techniques
Flash Photolysis versus Discharge Flow Systems
In the study of gas phase reactions for the
determination of absolute rate constants, a variety of
laboratory techniques can be seen in the literature.
The most prevalent of these techniques are discharge
flow and flash photolysis techniques. In flash
photolysis, a flash lamp emits a pulse of light which
dissociates a parent molecule resulting in reactive
species. Immediately upon the formation of the
reactive species, a time decay of the reactive species
is monitored using a technique such as fluorescence,
where the species observed is excited by a resonance
lamp. First order kinetics are followed because the
concentration of the stable reactant is in great excess
compared to the more reactive species being observed.
Flow systems such as the one used in this
experiment, consist of a tube in which two reactants

are mixed in the presence of a large amount of inert
gas. As the mixture travels down the length of the
tube at relatively fast speeds, the species are allowed
to react. By observing the decay of one of the
reactants along the length of the tube, kinetics data
are collected, and the absolute rate constant for the
reaction is determined. Fluorescence is used to monitor
the decay of OH in the flow system. Detection methods
such as optical absorption, electron spin resonance,
laser magnetic resonance, chemiluminescence or mass
spectrometry can also be used to monitor species decay
during the course of an experiment.11
Both techniques offer certain advantages.16 Flow
systems are more limited in terms of experimental
pressure ranges in order to maintain somewhat stable
flow dynamics in the tube. Even with this limitation,
flow systems have the ability to study reactions with
pressures as low as 0.5 torr. Flash photolysis
experiments can be run as low as 5 Torr.
With larger reaction cells and higher pressures,
wall losses do not have to be addressed with the flash
photolysis technique. Wall losses tend to be a serious

problem with flow systems. Coating the inner walls of
the flow tube with a halocarbon wax can minimize wall
loss problems but cannot eliminate them. Flow systems
offer more flexibility in terms of detection systems.
Flash photolysis detectors must respond rapidly on a
microsecond time scale making them both limited and
expensive. Flow system experiments are steady state
experiments where the reaction time is determined from
the point of mixing two species in relation to the
distance traveled to the detector. Detector speeds are
less important in this technique. Flow systems are
better equipped to study reaction mechanisms. Using
laser magnetic resonance or mass spectrometry in
combination with flow systems allows for the
identification of intermediate reactants and products.
The types of reactants which can be studied are
limited in flash photolysis. The reactant to be studied
must be able to be generated by flash photolysis.
Using this technique, one typically studies the
reactions between a stable molecule and a more reactive
species. Flow systems are able to study radical -
radical reactions.11,15

Discharge Flow Resonance Fluorescence Experiment
A detailed description of the experimental flow
tube used in this study (Figure 1) can be found in Juan
Bonilla's work.14 The system was designed using a 1.25
inch i.d. copper pipe and fittings soldered together in
various orientations to allow for an introduction of
gases into the system, connections for readout devices
and detection equipment as well as for easy
Several modifications of the flow system were made
through the course of this research in an effort to
attain useful data.
In the Bonilla experiment, the system was
controlled and monitored using an IBM personal
computer. The advantages to computer interfacing are
unlimited in terms of monitoring system conditions,
running repetitive experiments and acquiring data. Both
hardware and software modifications were made. A
Zenith personal computer was used in place of the IBM
computer. The modified basic program used to control
the flow system can be found on attached disc.


Physical changes in the flow system were made in an
effort to resolve problems which occurred after the
system had been relocated to a new laboratory. An
increased background signal developed early in this
research. Boxes were constructed from balsa wood to
enclose both microwave discharge cavities in an attempt
to control scattered light problems. The position of
the lenses used to optimize the focal point of light
inside the flow tube also was modified. The lenses
were set at optimal distances between the light source
and detector within a metal tube. The inner walls of
this tube were painted flat black. The tube extended
from the light source to the guartz window on the flow
system tube directly under the detector. Black felt
material was wrapped and taped around each end of the
tube containing the lenses at the point where it
connected to either the light source or to the flow
system. The seam between the photomultiplier tube and
the plate on top of the flow system was covered with
black electrical tape to block out another potential
source of scattered light. Aluminum foil was used to
shield power sources which may have been contributing

to the high background noise. Most connectors to the
power supplies and between the amplifier discriminator
and photomultiplier tube were changed and cables were
shortened if possible.
The main body of the flow system tube was coated
with halocarbon wax twice within the course of this
research. The quartz tubes in the microwave discharge
cavities were replaced.
The carrier gas used in the experiment was Grade
4.5 helium. A hydrogen/helium mixture (8.6% H2)
supplied the hydrogen to the system. In running low OH
experiments, the helium diluent gas alone was run
through the microwave discharge cavity. Enough
impurities exist in this gas to supply H atoms to the
system for specific low OH experiments.14 A 0.56% N02
mixture of nitrogen dioxide/helium supplied the N02 to
the system. A 99.5% purity hydrocarbon in lecture
bottle cylinders was used as the hydrocarbon source for
the system. Depending upon the behavior of the flow
controllers, the H2/He mixture, the N02/He mixture
and/or the hydrocarbon gas would have to be diluted
prior to setting up the experimental run. Dilutions

were made by first rinsing empty propane cylinders with
helium. The clean cylinders were then evacuated to
about 4 torr using a small vacuum pump. Dilution
factors were calculated taking atmospheric pressure
into account. The cylinders were carefully filled with
the reactant gas to the appropriate pressure. The
cylinder was closed and the appropriate amount of
helium diluent was allowed to fill the cylinder. Once
the proper amount of helium entered the system, the
flow was stopped and the cylinder was closed. Gas
regulators controlled the amount of reactant gas to
flow into the system through the flow controllers.
The source of OH in the flow tube was a result of
the reaction of hydrogen atoms with N02.
H* + N02 -+ OH + NO
The source of hydrogen atoms was the breakdown of
hydrogen molecules as they passed through the microwave
discharge cavity.
Flow controllers were calibrated using a soap
bubble method. Ten different voltages were individually

supplied to each flow controller. The rate at which a
bubble travels a known distance in a calibrated tube
was recorded for each voltage setting. A linear
regression analysis on the determined flow rate through
the flow controller versus the voltage setting was used
to determine gas concentrations in the system. A log
book was maintained in order to evaluate the
consistency of each flow controller's performance over
time. Changes in slope signified changes in the
performance of the flow controller. From this
calibration curve, the flow of the gas through the flow
controller can be related to gas concentration. The
calibrations were completed prior to running each
experiment in order to ascertain consistent
Figure 2 indicates the orientation of the
detection components with respect to each other. Point
a to point b identifies the direction of flow through
the system toward the vacuum pump. Points e and f are
the location of the Wood's horns. These horns create a
"black hole" to trap excess scattered light in the
detection area of the flow system. At point c, a six

inch copper plate was welded with a 1 cm hole in the
direct center to which a quartz window is epoxied. The
opening of this window is directly aligned to the
photomultiplier tube. The copper disc serves as a base
to the photomultiplier tube. A narrow band interference
filter is placed between the quartz window and the
photomultiplier tube which allows only light in the 280
- 325 nm wavelength region to pass through to the
detector. At point d, another quartz window is epoxied
to the opening of the 6 port fitting. This window is
aligned with the center or most intense portion of the
resonance lamp tube some 15 inches away. Point g was
the location of the most intense light within the flow
tube where the OH radicals would be excited and
Fluorescence was detected using a Thorn EMI
photomultiplier. Voltage to the detector was supplied
by a separate 15 volt power supply which was then
amplified to a low current and high voltage and
connected by a coaxial cable to the PMT housing. The
PMT output line was connected via a BNC cable to an
EMI GenCom model APED-11 amplifier discriminator used

to reduce noise and the amplify the electron bursts
into pulses. The APED-11 amplifier discriminator was
designed to operate along with charge output devices
which generate a single width-defined output pulse for
each negative input current pulse greater than a fixed
comparator threshold from the photomultiplier tube.
The amplifier discriminator is powered by a 5 volt
source and is connected to both a computer counter as
well as an independent counter for signal observations.
The signal is determined by subtracting the
background signal from the fluorescence signal. The
background signal is defined as the system signal when
the OH is not present in the system or when the N02 gas
is not turned on. The fluorescence signal is the
result of the following. As H20 passed through a second
microwave cavity, it dissociates into H and excited OH
radicals. These excited OH radicals emit light with a
strong component near 310 nm. This emitted light is
directed into the flow tube under the detector. This
light is used for fluorescence excitation of OH
radicals within the flow system. This fluorescence
signal is the result of the OH radicals in the flow

system being excited by OH radicals in the microwave
discharge cavity. The excited OH radicals in the flow
system are then detected by a photomultiplier tube
located directly over the area of fluorescence.
Background readings were taken before and after each
signal reading.
Experimental Procedure
Prior to performing an experiment, the system was
allowed to condition by flowing OH radicals through the
system for several hours. If the system had not been
running for several weeks or months, this conditioning
and equilibration of the flow system would continue for
up to 3 days before consistent signal could be
observed. In order to determine the OH concentration, a
constant concentration of H atoms was titrated into the
system with an increasing concentration of N02. During
this time, the OH signal would be observed. The
reaction rate constant between H and N02 is quite fast
at 1.3 X 10-10 cm3/molecule-sec.14

N02 + H -* OH + NO
Stoichiometrically, one OH radical is produced for
each N02 molecule added to the system. As more N02 is
added to the system, the OH concentration
proportionately increases until all the H atoms in the
system have reacted. At this point, further addition
of N02 will not produce more OH radicals and the signal
remains the same. The plot of this observation is
considered the titration curve (Figure 3). The point
at which the curvature of the plot changes slope
determines the equivalence point where the
concentration of H atoms is equal to the concentration
of N02 molecules. This concentration is also the OH
concentration in the system. A titration curve was
generated each day before and after experimental runs
to observe the consistency of the OH signal. Past
experiments determined that the flow of the H2/Helium
mixture passing through the discharge is inversely
proportional to the efficiency of H atom production.
Faster flows coincide with shorter residence times in
the discharge tube.

Figure 3:
xitraLiun Curve for deteraimiLiou uf Oil Coucentruv i'-ri

Three different concentrations of OH were observed
experimentally. Low OH concentrations varied from 1 to
3 x 10-11 molecules/cm3. High OH concentrations varied
from 1 to 3 x 10-12 molecules/cm3. Very high OH
concentrations varied from 8 to 16 x 10-12
molecules/cm3. Wall loss rates were determined from the
intercepts of the pseudo-first order rate constant
versus time plots.
The reaction rates for a particular concentration
of hydrocarbon and OH was measured by observing the
signal after the hydrocarbon was introduced through
each of the eight inlet ports which correspond to a
different reaction time. Each hydrocarbon
concentration was varied from 1 to 9 x 1013
molecules/cm3 and a pseudo-first order rate constant
was determined for nine different hydrocarbon
concentrations for each of the three OH concentration
levels. Experiments were performed at room temperature
and a flow system pressure of approximately 1 torr.
Bimolecular rate constants were determined from
the slope of the plot of the various pseudo-first order
rate constants versus hydrocarbon concentration. These

rate constants were determined at each OH
Experimental Results / Discussion
Experimental results for pseudo-first order decay
were obtained from a data output file following each
experimental run. An example of a pseudo-first order
decay plot can be seen in Figure 4. The slope of this
plot is equal to the pseudo-first order rate constant.
The determination of the apparent bimolecular rate
constant for each data set was the result of a linear
least squares analysis of each pseudo first order rate
constant versus each alkane concentration. The flow
system used in this experiment produced a minimum amount
of useable data through the duration of the
investigation. Many factors caused inconsistent results
to be obtained. One significant factor was the inability
of the flow controllers to function at their lowest flow
capacities. Attempts were made to clean the flow
controllers by flushing the flow chambers with freon and
allowing the controller to dry overnight by purging with

Figure 4: OH Decay Plot
KINETICS 12:26:19 06-30-1991
n 8,48
0 8.08
L 7,44
0 7,11 '
E 6.79
C 6.47
H 6.15
E 5,83
\ 1 1 1 1 1 1 1 1 1
- \ -
\ m

K m

ij'.. \ o 0
1 1 1 1 1 1 1 t
0,00200.80410,00610,00810,91020,01229.8.1420,01630,0133Q, 0283

minimal helium flow. Gases were diluted in an attempt to
compensate for poorly functioning flow controllers but
without success. Flow controllers were adjusted in an
attempt fine tune the settings. Some improvement
resulted but not enough to reach the low flow levels
needed. In the summer of 1991, several experimental runs
were completed using propane and butane at high OH
As previously discussed, discharge flow resonance
fluorescence experiments are affected by wall loss
activity. Experimentally, OH radical wall loss reaction
rates were determined by adding N02 directly to the flow
system through at least three of the eight fixed reactant
inlets rather than at the position designated earlier
immediately after the discharge region. This allowed the
OH radicals to be exposed to the walls of the tube for
various lengths of time. First order wall loss for OH
under these conditions were determined by measuring the
decay of OH fluorescence. The wall loss rates were found
to be about 2.5 s-1. Assumptions were made that the OH
was formed immediately upon the addition of excess N02 to
the flow tube.

The best experimental data obtained for the propane
experiment are seen graphically in Figure 5. The
experimental bimolecular rate constant for high OH +
propane was determined to be 1.08 X 10-12 cm3/molecule-
sec. Table 1 compares this determined rate constant to
literature values. The raw data can be seen in Appendix
C. The literature results differed from the results of
this research from 1.8% to 14%.
Table 1: Comparison of Experimental Rate Constant for OH
+ Propane to Literature Values.
Research Values Rate Constant (X10~12) (cm3/molecule-sec.)
This research 1.08
Droege and Tully1 1.10
Atkinson et al.9 1.22
Schiffman et al.17 1.02
N.R. Greiner10 1.26
Overend et al.18 1.22
The rate constants determined through the
modeling portion of this research resulted in an
effective bimolecular rate constant of 1.20 X 10-12
cm3/molecule-sec for the propane with high OH experiment

.>=>.ulr. i c.t Cfdor V (1/i
Figure 5: Propane Experimental Data
Uvncnmarifc! EC's o' iitq f\~7 /C i
l-ajjCi ii i iisi i Lt; i iSSuiio / / w i / w1 i
him rr 1 Hicn us-; / rrccane
1 4U-
60 to 1 CO
fFrcpsi.el (xH!5) '
l~rr.rr- U'Z'Z)

and an effective bimolecular rate constant of 1.09 X 10
12 cm3/molecule-sec. for the propane with low OH
Other OH and propane experiments in this study did
not generate useable results. Figures 6 and 7 show high
variability in the results of two other experimental
runs. Experimental data collected for the OH + butane
reaction was inconsistent and not used for comparisons in
this investigation. These results can be seen in Appendix
C. Figures 8 and 9 show the wide variability in these
experiments. No conclusive information can be offered
for the failure to obtain consistent data.
A vacuum pump failure ultimately ended further
investigations. With the resolution of this problem,
further studies should be undertaken.

,c>udci 1 s.t Ordtir k (l/seo)
Figure 6: Propane Experimental Data
1__vncnmcnT;! 1!oi iIto AS ,'t5n /Q i
: y\.n i i -w i ii_i l i uai ,vw'/ vw'S-'/ 1
nun ;-.icn om / I^iC'Og.p.w
o' (. i i!

i .*V\J

jai i
jS i50-i
6 1
i vru
vj . W J
a i
vJ *
i j

i ij 1 to
1 iu i/Ci ISO 190 210 2:0 220
[rrcpenel ix El 4)
ITirof:?; 1 l -E > 21

Figure 7: Propane Experimental Data
jnrpcnT^i pOQi ilTo OV/Pm /Q i
' i I 1 1 ' 1 1 wl I 1 *wl 1 Lw t / \J 1 ( W I
Hi or. OH ; P^ccc.r-0
i 21 CK m
c i 12-:-: a a ! j l
1 50-; B l i j
a TO ; 1 / B i 1
1 CM- 1 =:--; = n 14Gi a 1 1 i i i 1
i or- i i 1
1.2 1 : 1.4 i.o 1.5 i./ 1.S l.y 2 21 22
fFrccane'i ;'x El -Ti

isudci 1 £.t Order k (I /<
Figure 8: Butane Experimental Data
. 1vncnmcr'^i U'cc'i lit? fOC 70'
I /Lfw'CJl 11 1 H-11 I ICCUILO >w/*w/ i w -s^i Hinn i j.1-. / rajT.rr.g
G lO W
3. 17u-
h.* 1 S-fW 1
C. G 4 C.-N : w'V -
i-s 1 \y. ;
-8 < or_ 1 wW
ifl 120-
LL i i 1
.>A S
I V* J-
i CO 1:0 100 130 i ^0
i Vimr-fi i1 ;E!* 2!
Figure 9: bzX|3 r Butane Experimental Data imsntai Results 08/30/91
Run ^; r-!ich QH / Butane
4 .Sri
i l 4 ia ; > i l 1 1
l i a n i
120- i !
i 4 .-..-i: 1 1 i i j
^ 1 caj- l ** i
i GG~ 1 \ ! i s 1
10 20 £0 40 c-j tO 70 80 £0 i 00
(Times OP! 2}

Pcoudo First Order K (1/soc)
appendix c
Figure 1:
Kinetics Results for-Low, High and Very
High OH Concentrations: Ethane
OH -f Ethane H = 2.57E-13 c:u3/inoI-sec
Low OH
High OH
Very Hi OH
[t thane] (x E13 moIec/cm3)

PGOtido Pirn I Order K
Figure 2:
Kinetics Results for Low. High and Very-
High OH Concentrations: Propane
CK -r Propane K = 1.I0S-12
Low OH
High OH
Very Hi OH
1 23456789
[Pre-cans] (x E13 mo I sc/c m2)

Hnlft 1c determined
Figure 3:
Kinetics Results for Low, High and Very
Hi^h OH Concentrations: Butane
OH -j- Butane K = 1.72E-12 cmF/mol-sec
Low OH
-*-High OH
Very Hi OH

Kate k determined
Figure 4:
Kinetics Results for Lovr. High and Very
High OH Concentrations: isobutane "
'OH + IscbufaneK =
2.19E-12 m3/molec-sec
Low OH
High OH
5 Very Hi OH

Pooudo First Ordor K (l/aec)
Figure 5:
Kinetics Results for Low, High and Very
High OH Concentrations: Neopentane
Oa, -r XcopentantlC =
9.54.1,-13 c=:3/iaolec-scc