EPA-650/4-75-026
June 1975
Environmental Monitoring Series
MATHEMATICAL MODELING
OF SIMULATED
PHOTOCHEMICAL SMOG
P0
ID
-------
EPA-650/4-75-026
MATHEMATICAL MODELING
OF SIMULATED
PHOTOCHEMICAL SMOG
by
Paul A. Durbin, Thomas A. Hecht, and Gary Z. Whitten
Systems Applications, Inc.
950 Northgate Drive
San Rafael, California 94903
Contract No. 68-02-0580
ROAP No. 21AKC - 23
Program Element No. 1A1008
EPA Project Officer: Marcia C. Dodge
Chemistry and Physics Laboratory
National Environmental Research Center
Research Triangle Park, North Carolina 27711
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, D. C. 20460
June 1975
-------
EPA REVIEW NOTICE
This report has been reviewed by the National Environmental Research
Center - Research Triangle Park , Office of Research and Development,
EPA, and approved for publication. Approval does not signify thru the
contents necessarily reflect the views and policies of the Environmental
Protection Agency, nor does mention of trade names or commercial
products constitute endorsement or recommendation for use.
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environ-
mental Protection Agency, have been grouped into series. These broad
categories were established to facilitate further development and applica-
tion ol environmental technology Elimination of traditional grouping was
consciously planned to foster technology transfer and maximum interface
in related fields. ThfSr series are:
1. ENVIRONMENTAL HEALTH EFFECTS RESEARCH
2. ENVIRONMENTAL PROTECTION TECHNOLOGY
3. ECOLOGICAL RESEARCH
4. ENVIRONMENTAL MONITORING
5. SOCIOECONOMIC ENVIRONMENTAL STUDIES
6. SCIENTIFIC AND TECHNICAL ASSESSMENT REPORTS
9. MISCELLANEOUS
This report has been assigned to the ENVIRONMENTAL MONITORING
series. This series describes research conducted to develop new or
improved methods and instrumentation for the identification and quanti-
fication of environmental pollutants at the lowest conceivably significant
concentrations. It also includes studies to determine the ambient concen-
trations of pollutants in the environment and/or the variance of pollutants
as a function of time or meteorological factors.
This document is available to the public for sale through the National
Technical Information Service, Springfield, Virginia 22161.
Publication No. EPA-650/4-75-026
11
-------
n
PREFACE
As part of its program to clarify the roles of organic compounds and
oxides of nitrogen in the production of photochemical smog, the U.S. Environ-
mental Protection Agency (EPA) is supporting the study of irradiation-
induced air pollution in environmental chambers and the determination of
the rate constants and mechanisms of elementary reactions thought to be
important in smog formation. To complement this experimental effort, the
EPA is sponsoring SAI's work on the development of a chemical kinetic mech-
anism for photochemical smog formation. This mechanism, incorporating ex-
perimentally measured rate constants, is presently being compared with data
obtained from smog chamber experiments. Ultimately, the mechanism should
be capable of predicting the kinetics of the chemical transformations that
take place in photochemical smog. Our initial efforts to formulate and eval-
uate a kinetic mechanism for photochemical smog formation were summarized in
a detailed planning document (Seinfeld et al., 1973), in a 1973 final report
(Hecht et al., 1973), and in a 1974 final report (Hecht et al., 1974a).
-------
in
ABSTRACT
This report deals with the continued development and testing of a
kinetic mechanism for photochemical smog formation. An "explicit" mechanism
is developed and validated on smog chamber data taken at the University of
California at Riverside, Battelle, and the National Air Pollution Control
Association. After critical review of recent observations of reactions be-
lieved to occur in photochemical smog, a mechanism is developed for each of
the following systems: propylene-NOx, butane-NCy propylene-S02-NOx, and
toluene-NO . In addition, the report demonstrates that some chamber effects,
such as photolysis and surface reactions, potentially play a critical role in
smog chamber experiments. Finally, the report discusses the application of
kinetic simulation to a study of hydrocarbon reactivity and ozone production
in smog systems.
-------
CONTENTS
PREFACE . . . IT
ABSTRACT iji
LIST OF ILLUSTRATIONS vi
LIST OF TABLES x
I INTRODUCTION
1
A. Smog Chamber Simulation and Elementary
Reaction Kinetics
2
B. Explicit Mechanistic Approach 2
C, Hydrocarbon Reactivity .
II CHEMICAL KINETICS 12
A. Inorganic Chemistry 12
1. Heterogeneous HN02 Chemistry 12
2. Heterogeneous HNOs Formation 17
3. 03 Decay 19
B. Organic Chemistry 19
1. Propylene-OH- Reactions 19
2. Unimolecular Decomposition of AUoxyl Radicals 22
3. Alkoxyl Radical-02 Reactions 23
4. Propylene-Oa Reactions 23
5. Radical-Radical Reactions 25
6. PAN Chemistry 26
C. Photochemistry 27
1. Photolysis Rate Constants 27
2. Spectrum Decay 29
D. S02 Oxidation 31
1. Some Observations 34
2. Kinetic Mechanism for S02 Oxidation 35
E. The Toluene-NO -Air System 40
1. Toluene + 0(3P) 40
2, Toluene + OH- 40
3. The Proposed Mechanism 41
-------
Ill SMOG CHAMBER SIMULATIONS 45
A. Simulations of UCR Data 45
1. Propylene-NOx- Air System 46
2. Butane-N0x-Air System 68
B. Simulations of Battelle Data 87
1. Instrumentation 87
2. Mechanism Used 92
3. Results and Discussion 92
C. Simulations of the NAPCA Toluene-N0x Data 105
1. Mechanics Used • 105
2. Results and Discussion 105
IV HYDROCARBON REACTIVITY 114
A. Survey of Reactivity Measures 114
1. Temporal Measures H^
2. Concentration Measures H5
3. Combined Temporal and Concentration Measures 116
B. Measure Assessment 118
1. Scope and Procedure H8
2. Measure Study 121
3. The Measures Selected 127
4. Mixture Study 130
5. Derivation of Some Properties of T^ 133
C. Relation of the Above Considerations to
Ozone Production 142
1. Ozone Isopleths 142
2. Chemical Dynamics . . . 150
V CONCLUDING REMARKS 151
REFERENCES -153
-------
VI
ILLUSTRATIONS
1 Smog Profiles for Different Values of kio, the Rate Constant
for the Reaction 2H20 + NO + NO? + 2HN02 + HgO . . . . ...... 15
2 Smog Profiles for Different Initial Concentrations of HNCu .... 16
3 Smog Profiles for Different Values of kg, the Rate Constant
for the Heterogeneous Formation of HMOs, NzOs + H20 •+ 2HN03 • • • 18
4. Spectra of Sunlight and U.C. Riverside Solar Simulator ...... 28
5 Effect of a 20 nm Filter Shift on Smog Profiles ......... 32
6 Effect of a 20 nm Filter Shift on N02 Behavior for EC-60 ..... 33
7 Propylene-N0v Factorial Block .................. 47
A
8 EC-11 Simulation Results and UCR Data for N09, 0,, and NO .... 52
C. O
9 EC-11 Simulation Results and UCR Data for Propylene
and Formaldehyde ......................... 53
10 EC-11 Simulation Results and UCR Data for Acetone and PAN .... 54
11 EC-11 Simulation Results and UCR Data for Acetaldehyde ...... 55
12 EC-12 Simulation Results and UCR Data for 03, NO, and N0« . . . . 56
13 EC-12 Simulation Results and UCR Data for Propylene
and Formaldehyde ......................... 57
14 Simulation Results and UCR Data for PAN and Acetaldehyde ..... 58
15 EC-16 Simulation Results and UCR Data for 03> NO, and N02 . . . . 59
16 EC-16 Simulation Results and UCR Data for Acetone,
Acetaldehyde, and PAN ...................... 60
17 EC-16 Simulation Results and UCR Data for Propylene
and Formaldehyde ......................... 61
18 EC-18 Simulation Results and' UCR Data for Propylene
and Acetaldehyde ......................... 62
19 EC-18 Simulation Results and UCR Data for NO, N02> and PAN .... 63
-------
vii
20 EC-18 Simulation Results and UCR Data for 03 and Formaldehyde ... 64
21 EC-21 Simulation Results and UCR Data for Propylene
and Acetaldehyde 65
22 EC-21 Simulation Results and UCR Data for NO, N02>
and Formaldehyde 66
23 EC-21 Simulation Results and UCR Data for 03 and PAN 67
24 n-Butane/NOv Factorial Block 71
A
25 EC-39 Simulation Results and UCR Data for Og, NO, and N02 72
26 EC-39 Simulation Results and UCR Data for Butane 73
27 EC-39 Simulation Results and UCR Data for Acetaldehyde
and Formaldehyde 74
28 EC-39 Simulation Results and UCR Data for MEK and PAN 75
29 EC-41 Simulation REsults and UCR Data for 03> NO, and N02 76
30 EC-41 Simulation Results and UCR Data for Butane 77
31 EC-41 Simulation Results and UCR Data for MEK and PAN ....... 78
32 EC-41 Simulation Results and UCR Data for Acetaldehyde
and Formaldehyde 79
33 EC-42 Simulation Results and UCR Data for Butane, NO, and N02 . . . 80
34 EC-42 Simulation Results and UCR Data for MEK, Formaldehyde
and Acetaldehyde 81
35 EC-42 Simulation Results and UCR Data for Ozone 82
36 EC-44 Simulation Results and UCR Data for NO, N02, and
and Acetaldehyde 83
37 EC-44 Simulation Results and UCR Data for Ozone 84
38 EC-44 Simulation Results and UCR Data for Butane 85
39 EC-44 Simulation Results and UCR Data for Formaldehyde
and MEK 86
40 EC-44 Simulation Results, Using k]0 = 1-3 x 10~12ppnf3 nrin"
0.024 ppnH min"1, for NO and N02
41 EC-44 Simulation Results, Using kiQ = 1.3 x 10~12ppnf3 mirT1
and kn = 0.024 ppnH mitr1, for Ozone 89
-------
42 EC-44 Simulation Results, Using k10 = 1.3 x 10 12 ppm~3 min"1
and kn = 0.024 ppm"1 min"1, for Butane ............. 90
43 EC-44 Simulation Results, Using k]0 = 1.3 x 10"12 ppm"3 min"1
and k-|-| = 0.024 ppm-"1 min-', for MEK, Acetaldehyde,
and Formaldehyde ........................ 91
44 S-107 Simulation Results and Battelle Labs Data for
Propylene, N02, and S02 ................ ..... 95
45 S-107 Simulation Results and Battelle Labs Data for
03, NO, and SOg Aerosol ..................... 95
46 S-110 Simulation Results and Battelle Labs Data for
Propylene, N02, and S02 ..................... 97
47 S-110 Simulation Results and Battelle Labs Data for
03, NO, and S03 Aerosol ..................... 98
48 S-113 Simulation Results and Battelle Labs Data for
Propylene, N02, and S02 ..................... "
49 S-113 Simulation Results and Battelle Labs Data for
03, NO, and S03 Aerosol ..................... 10°
50 S-114 Simulation Results and Battelle Labs Data for
Propylene, 03, NO, and N02 ................... 101
51 S-115 Simulation Results and Battelle Labs Data for
Propylene, 03> NO, and N02 ................... "°2
52 EPA-258 Simulation Results and NAPCA Data for NO, N02, and 03 . . 107
53 EPA-258 Simulation Results and NAPCA Data for Toluene ...... 108
54 EPA 272 Simulation Results and NAPCA Data for NO, N02, and 03 . . 109
55 EPA-272 Simulation Results and NAPCA Data for Toluene ...... 110
56 EPA-305 Simulation Results and NAPCA Data for NO, N02> and 03 . . Ill
57 EPA-305 Simulation Results and NAPCA Data for Toluene ...... 112
58 The Use of Kinetic Simulations to Assess Reactivity ....... 120
59 Typical Smog Profile ...................... 136
60 T as a Function of Initial Hydrocarbon Concentration ...... 139
m
61 Lines of Constant 03 (in ppm) After 1 Hour of Simulation .... 143
62 Lines of Constant 03 (in ppm) After 2 Hours of Simulation .... 144
-------
63 Lines of Constant 0., (in ppm) After 5 Hours of Simulation 145
O '
64 Lines of Constant 0 (in ppm) After 8 Hours of Simulation 146
65 Lines of Constant 03 (in ppin) After 9 Hours of Simulation 147
66 Time of the N09 Peak (in Minutes) 148
-------
TABLES
1 Summary of Reactions 5
2 Summary of Rate Constants 8
3 Photolysis Constant Changes from UV Loss 30
4 Rate Constants for S02 Oxidation 36
5 UCR Propylene-N0y Experiments: Initial Concentrations
of Primary Reactants 47
6 The Propylene Oxidation Mechanism 48
7 Changes Made in Mechanism for Butane Simulations . 69
8 UCR Butane-N0x Experiments: Initial Concentrations
of Primary Reactants and Values of k, 71
9 Analytical Characteristics of Battelle Experimental
Setup \ 93
10 S02 Oxidation Mechanism 94
11 Battelle Propylene-N0x-S02 Experiments: Initial
Concentrations and Values of k, 103
12 Rates of S02 Oxidation by Various Oxidants
(from S-107 Simulation) 104
13 Toluene Oxidation Mechanism 106
14 NAPCA Toluene-N0x Experiments: Initial Concentrations .... 113
15 Definitions of Reactivity Measures 123
16 Results of the Measure Study: Reactivities Relative
to Propylene 124
17 Initial Concentrations for Experiments Listed in Tables
Tables 16 and 18 125
18 Results of the Mixture Study: Mixture Reactivities
Relative to Propylene 131
-------
I INTRODUCTION
The problem of photochemical air pollution has received considerable
attention from scientists, legislators, and the public during the past two
decades. Although the problem was originally manifested in Los Angeles, it
has become increasingly evident in other urban and even nonurban areas
and is worthy of widespread concern. As its name indicates, a fundamental
characteristic of photochemical air pollution is the role played by sunlight-
initiated chemical transformations. Primary pollutants are often hazardous
themselves; however, these secondary chemical processes greatly exacerbate
the problem. They lead to the production of phytotoxicants, lacrymators, and
carcinogens.
The present document reports on the continuation of an ongoing research
effort to isolate and model the complex chemical reactions that occur in
polluted atmospheres. This effort is aimed at producing a photochemical
kinetic mechanism for smog formation having sufficiently accurate kinetics
to provide realistic predictions of pollution production while, at the same
time, being simple enough for practical use in large computer models of urban
airsheds. Past efforts (Hecht et al., 1973; Hecht et al., 1974a) toward
these ends have resulted in the development of a "generalized" kinetic mech-
anism. The generalized mechanism has proved to have great utility, making
urban airshed modeling feasible (Reynolds et al., 1974). However, in the
present report, a digression from the generalized approach has been made. To
take full advantage of smog chamber data and kinetic studies, we employed an
"explicit" mechanism. The explicit approach and the reasons for its incor-
poration are discussed below, after a review of some background information.
A final topic, the application of the kinetic mechanism to an investigation
of hydrocarbon reactivity, is discussed at the end of this introduction.
-------
A. SMOG CHAMBER SIMULATIONS AND ELEMENTARY REACTION KINETICS
Smog chamber investigations and elementary kinetics studies provide
the basic inputs to the present kinetic mechanism. In smog chamber studies,
clean air and pollutants (usually NO, N02, and a hydrocarbon) are irradiated
in a reactor. Measurements of reactant and product concentrations as a func-
tion of time provide smog profiles, which the kinetic mechanism should repro-
duce. These profiles constitute quantitative and qualitative descriptions of
the macroscopic features of smog. Thus, the conversion of NO to N02, the oxi-
dation of hydrocarbons, and the production of oxidants, as observed in the at-
mosphere, are reproduced in a controlled laboratory environment. But smog
chamber studies cannot provide a knowledge of the microscopic features of smog
formation. Independently, kineticists study elementary reactions that could
be important in this process.
Kinetics studies are performed to elucidate the rates and mechanisms of
particular reactions that occur within the overall process of smog formation.
These studies differ from smog chamber studies in that the reactants and
other conditions are carefully chosen to isolate or emphasize the reactions
of interest. The observed variations in reactant and product concentrations
with time are used to formulate rate equations and rate constants. If com-
plicating reactions are absent or well characterized, these rate equations
and constants have universal validity. Mechanistic information conies from
the rate expression as well as from observed reaction products.
Ideally, a kinetic mechanism would be simply the assemblage of results
from kinetic studies of all reactions that occur. In reality, not all of
the reactions that could occur have been studied, and often orders of magni-
tude of uncertainty may be associated with those that have been studied. In
addition, the number of possible reactions in smog is very large. Hence, a
complete mechanism is neither practical, because it would include an enormous
number of reactions, nor feasible, because the needed kinetic information is
unavailable. The best one can do to obtain a closed system of chemical equa-
tions and rate constants is to use available kinetic information and methods
-------
for estimating other information, to draw analogies, and to make other sim-
plifying assumptions. The kinetic mechanism that results may be a fairly
accurate, albeit simplified, description of reality. The system of equations
constituting the current mechanism has largely been discussed previously
(Hecht and Seinfeld, 1972; Hecht et al., 1974b). However, several changes
and additions have been made since last year's report. These new features,
and the kinetic studies supporting them, are reviewed in Chapter II of the
present report.
In addition to the reactions characterized by universal kinetic expres-
sions, there are processes, both physical and chemical, occurring in smog
chambers that are peculiar to a given chamber. These chamber effects include
surface-catalyzed reactions, adsorption and desorption of chemicals, dilution
due to sampling, inhomogeneous concentrations, and spatial and spectral varia-
bility of light sources. Of the surface reactions, probably the most important
is the heterogeneous formation of nitrous and nitric acid. The heterogeneous
production of HNOX is investigated in Chapter II. A possible spectral
variation of the light source used in smog chamber simulations done at the
University of California at Riverside is also considered from a theoretical
standpoint. The results show the critical influence of surface and light
properties on smog profiles and point to a need for more detailed character-
izations of smog chambers.
In Chapter III, predictions of smog profiles made by numerical simulation
are compared with experimental data. To model the smog chamber results com-
pletely, we had to incorporate some of the chamber effects mentioned above.
The techniques for doing so are described in Chapter III.
B. EXPLICIT MECHANISTIC APPROACH
The most straightforward approach to formulating a kinetic mechanism
is to assemble the most important reactions that occur. The mechanism's com-
plexity is then dictated by the criterion for "important." A technique des-
cribed by Hecht et al. (1974b) even further simplifies the formulation. In
this technique, groups of reactions are "lumped" into single reactions; these
reactions are then combined to form a generalized mechanism. But the lumping
-------
approach introduces nonphysical lumped rate constants, stoichiometric co-
efficients, and chemical species. Because of this last feature, the predic-
tion of detailed product yields is not feasible, and because of the first
two, kinetic information cannot be used directly. Thus, we used the classi-
cal, explicit approach instead, to take advantage of recent kinetic studies
and the full range of product measurements available.
Formulating the explicit mechanism involves simply expanding the
lumped reactions of the generalized mechanism. Conversely, reformulation
of a generalized mechanism involves contracting the explicit mechanism.
Thus, in the present approach, uncertainties associated with the lumping
procedure have been eliminated, while a kinship with the generalized mech-
anism has been retained. After validation, the explicit mechanism can then
serve as a basis for rederiving a generalized mechanism and for checking
the accuracy of lumping techniques. We should emphasize that the explicit
approach is an interim step. We hope that it will serve to further clarify
smog kinetics and thus to lay the groundwork for the more practical general-
ized approach.
A review of the reactions and rate constants used in the mechanisms
contained herein appears in Tables 1 and 2. The lumped format (i.e., use
of the R group) was used to represent a class of reactions, which appear
explicitly in the later mechanisms (Chapter III).
C. HYDROCARBON REACTIVITY
i
Pollution control strategists must know the reactivity of various hydro-
carbons to predict the potential impact of different emissions sources. The
conceptualization of reactivity can, however, take various forms. Corres-
pondingly, there are many ways of quantifying reactivity in smog systems.
Probably the most obvious and commonplace measure of reactivity is the time
required for N0£ to reach its peak concentration, because the NOg peak is a
distinctive feature of smog profiles. Many other measures have appeared in
-------
Table 1
SUMMARY OF REACTIONS
;. Inorganic H-N-0 Compound Chemistry
A. The N02-NO-03 Cycle
1. NOe + hv + NO + 0
2. 0+02+M+03+M
3. 03 + NO + N02 + 02
B. Other NOX Chemistry
4. N02 + 0 -> NO + 02
5. N02 + 03 •+ NOs + 02
6. NOs + NO •*• 2N02
7. NOs + N02 -> N205
8. N205 -1 NOs + N°2
9. H20 + N205 surface> 2HN03
C. HNO? Chemistry
10. NO + N02 + 2H20 surface> 2HN02 + H20
11. 2HN02 surface> NO + N02 +
12. HN02 + hv •* NO + OH-
D. OH- and HO^ Reactions with NOX
13. OH- + NO -1 HN02
14. OH- + N02 -^ HN03
15. H02 + NO -> N02 + OH-
11. 03 Inorganic Chemistry
16. 03 + hv + 02 + 0(3P)
17. 03 + hv -»- 02 + 0(1D)
-------
18. 0(]D) + M + 0(3P)
19. OpD) + H20 -> 20H-
20. 03 + OH- + H02 + 02
21. 03 + H02' -> OH- + 202
22. 03 surface> products
III. Organic Oxidation Reactions
A. Butane
°2
23. C4H10 + 0 -£*- R02' + OH-
24. C4Hio + OH- — ^ n-R02 + H20
25. C4H10 + OH- —2+- s-R02 + H20
B. Propylene
26. CH + OH- •> n-RO-
27. C3H6 + OH- -*• s-RO-
°2
28. C3H6 + 03 —=*-'CH3CHO + H02 + OH- + CO
29. C3H6 + 03 -^H2CO + CH3C(0)02 + OH-
30. C3H6 + 0 —£- R02 + RC(0)02
°2
31. C3H6 + 0 —V R02 + H02 + CO
C. A1dehydes
32. RCHO + OH- -X- RC(0)02 + H20
D. Toluene
02
33. C6H5CH3 + OH- *~ C6H5CH202 + H20
34. C5H5CH3 + OH- -^ C6H4(CH3)(OH) + H02
34a. C6H5CH3 + OH- —^
E. Organic Radicals
CL
35. RO- —^ R02 + H2CO
36. RO- + 02 ^ ALD + H02
-------
IV. Other Photolysis Reactions
37. H202 + hv -v 20H-
°2
38. RCHO + hv — ^R02 + CO + H02
39. RCHO + hv + paraffin + CO
202
40. H2CO + hv — H-CO + 2H02'
41 . H2CO + hv -*- H2 + CO
V. Reactions of Organic Free Radicals with NOX
A. NO Oxidation
42. R02 + NO + RO- + N02
°?
43. RC(0)02 + NO —+ R02 + C02 + N02
B. PAN Chemistry
44. RC(0)02 + N02 -> RC(0)02N02 (PAN)
45. PAN --*-N03 + C02 + R02
VI. Radical -radical Recombination Reactions
46. H02 + H02 ^ H202 + 02
47. H02' + R02 -v R02H + 02
48. RC(0)02 + H02 ->• RC(0)02H
VII. SO? Chemistry
A. SO? Oxidation
49. N03 + S02 •> SOs + N02
50. H02 + S02 -»• SOs + OH-
51. R02 + S02 -> SOa + RO-
Op
52. RC(0)02 + S02 — ^S03 + C02 + R02
°2
53. S02 + OH- —
54. HSOs + NO -»• HS04 + N02
55. HS04 + H02 -v HoS04 + 02
56. HS04 + N02 -^ H2S04 + HN03
-------
Table 2
SUMMARY OF RATE CONSTANTS
(ppnH rnin-1 unless noted)
Rate Constant
Reaction Number
1
2 (ppm-2 min-1)
3
4
5
6
7
8 (min-1)
9
10 (ppnT3 min"3)
•11
12
13
14
15
16
17
Garvin and
Hampson (1974)
2.08 x TO'5
25.2
1.34 x 104
5.0 x ID'2
1.3 X 104
5.6 x 103
21.9
< 1.5 x 10-5C
< 10-13C
2.95 x 103
1.2 x 104
2.95 x 102
Uncertainty
Factor This Study3
Experimental
1.2 2.08 x 10'5
1.3 25.2
1.2 1.34 x 10*
1.3 5.0 x ID'2
5.0 1.3 x 104
2.5 5.6 x 103
2.0 24.0
5.0 x 10'6
1.3 x 10-11
• 2.6 x 10-1
Experimental
2.0 3.0 x 103
2.5 l.OxlO4
3.2 8.0 x 102
Experimental
Experimental
Others
2.04 x 10"5
(Wu and Nlki, 1975)
27.0
(Wu and Niki, 1975)
1.39 x 104
(Wu and Niki, 1975)
6.8 x 10-2
(Wu and Niki, 1975)
1.5 x 104
(Wu and Niki, 1975)
4.5 x 103
(Wu and Niki, 1975)
ks/kg = 4.2 x TO'3
(Benson, 1968)b
10-10 . 10-12
(Noch et al., 1974)
ho H20/k11 = 9.7 x
(Demerjian et al.,
8.9 x 103
(Cox, 1974)
•v, 1.2 x 103
(Mabey and Hendry,
10-7
1974)
1974)
-------
Rate Constant
Garvin and
Reaction Number Hampson (1974)
18 8.6 x 104d
19 5.17 x 105
20 8.71 x 106
21 2.4
22
23
24, 25 3.47 x 103
26, 27 2.14 x 104
28, 29 0.02
30, 31 5.3 x 103
32 RCHO; R=0,l
RCHO; R=2,3
33,34
35 (nrirr1)
36 ' -v 4.4 x 10-3
37-41
42, 43
44
45
46 8.5 x 103
47, 48
49
Uncertainty
Factor This Study3
1.4 8.6 x 104
1.3 5.1 x 105
2.0 8.7 X 101
2.0 2.4
Experimental
64.0
1.2 3.4 x 103
1.2 2.5 x 104
0.02
1.2 5.3 x 103
2.1 x 104
4.5 x 104
9.2 x 103
0.6 - 1.3 x 104e
0.04 - 0.2e
Experimental
103 (estimate)
3 x 102 (estimate)
3 x ID'3
2.0 6.0 x 103
3 x 103 (1/2 k44)
14.0
Others
12.6
(Johnson et al., 1970)
2.5 x 104
(Morris et al., 1971)
0.026
(Becker et al . , 1974)
2.3 x 104
(Morris and Niki, 1971;
Morris et al . , 1971)
4.6 x 104
(Morris and Niki, 1971)
9.0 x 103
(Davis, 1974)
0.8 - 12.0 x 104
(Batt et al., 1974)
^ 0.8
(Mendenhall et al., 1974)f
< 14.7
(Davis, 1974)
-------
10
Rate Constant
Garvin and Uncertainty
Reaction Number Ijampson (1974) Factor
50
51, 52
53
54
55
56
This Study6
1.3
1.5 (estimate)
9 x 102
8 x 102
(estimate)
9 x 103 (3/2
1 x 104
(estimate)
Others
1.35
(Davis, 1974)
1.35 x 103
(Davis, 1974)
a At 303°K
b Calculated using Keq ^ expf-AGjgg/RT), iG from Benson (1966}
c Surface-dependent
^ Combination of values for M = 02 and M = N£
e Depends on carbon skeleton
f Combination of BuO- and MeO- data
-------
11
the literature. A review of these reactivity indices and an evaluation of
the properties of some of them appear in Chapter IV of this report.
Ultimately, reactivity should be related to the production of harmful
components of photochemical air pollution. For this reason, we investi-
gated the relationship between reactivity measures and ozone production. In
this report, ozone yield is shown to be relatively insensitive to reactivity
for a group of olefins, though it does depend on initial NOX and hydrocarbon
concentrations.
An important aspect of photochemical smog production in the atmosphere
is its dynamic behavior. An understanding of the interaction of the time
scales for reaction, transport, and dispersion is necessary to master the
pollution problem. The dynamics of smog formation are, to a significant
extent, determined by the time required for N02 to reach its peak. Chapter IV
investigates this interrelationship.
-------
12
II CHEMICAL KINETICS
A. INORGANIC CHEMISTRY
An excellent review of inorganic (and some organic) reactions occurring
in the stratosphere has been prepared by Garvin and Hampson (1974). A
majority of the inorganic reactions thought to occur in polluted atmospheres
can also be found in that review. The rate constant rec&mmendations made by
Garvin and Hampson have generally been adhered to in the present study. The
few exceptions to this rule are cases where the results of more recent studies
have become available or where the reactions are surface dependent. The
latter set of reactions is considered below because of their possible signifi-
cance (and elusiveness) in smog chamber simulations.
1. Heterogeneous HN02 Chemistry
Of the reactions that can take place on smog chamber walls, perhaps
the most significant is the heterogeneous formation of HN02:
2H20 + NO + N02 surface> 2HN02 + HgO (10)*
The occurrence of this reaction following the introduction of NOX reactants
could produce significant amounts of HN02- OH radicals produced by HN02
photolysis play a major role early in the reaction. Thus, the occurrence
of Reaction (10) can have a profound effect on the length of the induction
period. It is important to note that Reaction (10) is a thermal reaction
and that it can occur in the approximately 45 minute dark mixing time allowed
at the beginning of Riverside experiments. During this time, Reaction (10)
will compete with Reaction (11):
* This report uses two numbering systems. Chemical reactions used in modeling
have simple Arabic numbers taken from Table 1 and previous work (Hecht et al.,
1974b). Other chemical reactions and mathematical equations are numbered
sequentially within each chapter, e.g., (II-l), (II-2).
-------
13
2HN02 surfac? H20 + NO + N02
(ID
A solution of the rate expression derived from Reactions (10) and (11) for
the concentration of HN02 at time t is
1
[HN02]t = ,
1 A + L
B tanhf-,
(II-D
[HN02]eq tanh
where
.1/1 +
"™ I. I r ttf\ i •
\[NO]Q [N02],
B = [HN02]
eq
[H20]2 ([N0]0 - [N02]Q)
2n
[NO]Q [N02]Q
1/2
^/2
(2[HO]0 [N02]Q)1/2 k1Q [H20]2
2
([NO] - [N02])2 (k1Q [H20]2)
- 1/2
= (keq [NO]
eq l"UJ0
2
eq
-------
14
Taking [N0]0 = 1.0 ppm, [N02]Q = 0.1 ppm, [H20] = 2 x 104 ppm, and
keq = 1.9 x 10"2 gves
[HN02leq = 4,4 x 10"2 ppm
In their study, Noeh et al. (1974) found kio * 10'12 ppnr3 min"1 (we have
adjusted their third-order results to fourth-order kinetics) in a quartz
cylinder and kio * 10'10 with metal present. Garvin and Hampson report
k-jg < 1Q-13 for gas phase reactions.
Taking kio = 10"13, 10~12, 10'11, and 1(T10 ppnr3 min"1 as a represen-
tation of this range gives the following respective values of T:
T = 5500, 550, 55, and 5.5 minutes. Thus, at t = 45 minutes,
[HN02l/[HN02l = 0.008, 0.08, U.67 and 1.0, respectively. Depending on
, the dark reaction will produce from 0.8 to 100 percent of the equili-
brium concentration of HN02 in 45 minutes. The rate constant listed for
"this study" in Table 2 leads to a time scale (T) of 41.9 minutes (or ap-
proximately 1 hour), and to an HN02 concentration of 79 percent of the equi
librium value in 45 minutes. Naturally, these results depend on [NO]Q,
and the dark mixing time. In interpreting the conclusions for large k-|0»
one should realize that the approximation used in Eq. (II-l) may break down
when t is on the order of T.
Figure 1 illustrates the effect of setting k10 at 10'13, 10'12, and
10"10 ppm-3 min'1 (and varying kn to keep keq constant) on the propylene-
NOX system. The shortening of the induction period is readily visible in
the N02 curves. Figure 2 shows the results of setting kio ec1ual to
10~13 p'pm"3 min'1 and introducing 0, 10, or 100 percent of the equilibrium
amount of HN02 as an initial condition. Again, HN02 shortens the induction
period. For both sets of figures, the experimental conditions were identi-
cal, except for the variable being studied.
It is conceivable that all the HN02 formed at the reactor walls does
not vaporize. Inadequate cleaning of chamber walls might then allow a
"memory effect" on subsequent experiments. Jaffee and Smith (1974) have
experimentally demonstrated the effects of inadequate cleaning, as well as
other chamber effects. The anomalous OH source observed recently in the
-------
1.00 -r-
-80 --
a.
OL
£
•M
0>
U
C
o
o
.60 --
-MO --
.20 --
0-00
0-0 50-0 100-0
• 10-12
i% ki° = lcHO
150-0 200 0 250-0
Time (minutes)
300-0 350-0 400-0
FIGURE 1. SMOG PROFILES FOR DIFFERENT VALUES OF ki
-------
1.00 -r-
.80
5: .so
o
•M
c
CD
U
C
o
.MO --
.50 -
0-00
NOg
03
N02
03
N02
03
[HN02]Q = 100% of Equilibrium
[HN02]Q = 10% of Equilibrium
[HN02]Q = 0
Time (minutes)
FIGURE 2. SMOG PROFILES FOR DIFFERENT INITIAL CONCENTRATIONS OF HN02
-------
17
Riverside chamber (Winer, 1975) could be due to HN02 retention by the cham-
ber walls (see Chapter III).
2. Heterogeneous HNCh Formation
Morris and Niki (1973) found that, for the reaction
N205 + H20 Surface> 2HN03 (9)
the upper limit of the rate, kg, is 1.5 x 10~5 ppnr1 min"^ in the gas
phase. They also showed that this reaction has both homogeneous and hetero-
geneous components. Spicer and Miller (1974) have presented evidence that
the primary mode of HN03 production in their chamber is the heterogeneous
reaction. They achieved nearly a 100 percent nitrogen balance in the gas
phase and thus concluded that HN03 vaporizes after formation.
Because the sequence of reactions
N02 + 03 -»• N03 + 02 (5)
N03 + N02 * N205 (7-8)
will result in the highest ^05 concentrations only after significant 03
has formed, Reaction (9) is important primarily after the N02 peak.
Figure 3 shows the effect of this reaction. For the two sets of profiles
in Figure 3, kg was taken to be 1.5 x 10"3 ppm~^ min"^ (Jaffee and Ford's
value, 1967) and 5 x 10~5 ppm'l min"^. The higher value results in a
markedly greater consumption of N02 and correspondingly lower production
of 03.
This theoretical evaluation of heterogeneous HNOX formation has shown
its potential significance. Obviously, the extent of these reactions de-
pends on surface type as well as on surface-to-volume ratio. The need for
experimental evaluation of kg and \C\Q in individual smog chambers is evident.
-------
1.00
.80 --
o.
0.
o
•r-
-M
to
.60 --
c .MO --
.20 --
0.00
0-0
* N02 )
X 03 >kg = 5.0 x TO'6 ppnr1 nriiH
X Propylene 1
X N02 \
^ 03 Hg = 1.5 x 10"^ ppnH tm'n'l
^ Propylene 7
50-0 100.0
150:0 200:0 Z5U-0
Time (minutes)
300-0
35U.Q MOO-Q
FIGURE 3. SMOG PROFILES FOR DIFFERENT VALUES OF kg, THE RATE CONSTANT
FOR THE HETEROGENEOUS FORMATION
OF HNOs, NgOs + H£0 + 2HN03
•00
-------
19
3. 03 Decay
The rate of 03 destruction on chamber walls [Reaction (22), Table 1]
used in the calculations in Chapter III was obtained from 03 half-lives
observed in each smog chamber. Decay rates were exponential; first-order
decay was assumed.
B. ORGANIC CHEMISTRY
The oxidation of organic molecules in polluted air has been described
previously (Demerjian et al., 1974a; Hecht et al., 1974a). In the following
sections, we discuss recent studies in this area in relation to the development
of the kinetic mechanism. Those reactions not dealt with here are described
in the above references.
1. Propylene + OH«
Early investigators of the olefin-NOx-air photochemical reactions dis-
covered that, in addition to ozone and oxygen atoms, an unknown oxidant was
participating strongly in smog formation (see Lei.ghton, 1961). Several spec-
ulations were made as to the identity of this reactant, and it now appears
(e.g., Demerjian et al., 1974a or Hecht et al., 1974a) that the OH radical
is the most likely of these possibilities. In fact, OH is probably the
most important olefin oxidant in smog, accounting for well over 50 percent
of the olefin disappearance rate (Calvert and McQuigg, 1975). Hence, the
success of a kinetic mechanism rests heavily on its representation of the
OH-olefin (or, in general, OH-organic) reaction (including the ensuing chain).
There are two likely alternatives for the initial step in the OH-propylene
reaction:
> Abstraction of an allylic hydrogen
> Addition of OH to the carbon-carbon double bond.
-------
20
Slagle et al. (1974) studied the propylene-OH reaction in their crossed
molecular beam reactor. Product analysis by photoionization mass spectrom-
etry yielded a dominant ion signal corresponding to the abstraction product.
Hence, they suggested that the propylene-OH reaction proceeds primarily by
abstraction of an allylic hydrogen. In our simulations, this reaction was
assumed to be followed immediately by the addition of QZ to tne a11^ radical
OH- + CH2=CHCH3
(H-2)
This could be followed by NO oxidation to form N02 and acrolein:
CH2=CHCH202 + NO -> CH2=CHCH20- + N02
CH2=CHCH20- + 02 -»• CH2=CHCHO +
(1 1-3)
(1 1-4)
but the large production of acrolein that would result from this reaction
is contrary to the observed products.
Another speculative reaction pathway was considered, in which 2 03
molecules add to the ally! radical to form a five membered cyclic peroxy
radical .
CH2=CHCH3
OH
NO
+ CO + 2HoCO
NO
r\
y
k
(II-5)
However, this formulation could not match the rapid reaction process that
was observed in the UCR chamber, nor could it account for the large acetalde-
hyde product yield. Unless further experimental evidence arises to indicate
-------
21
that the abstraction mechanism is important in smog, we shall assume that
the OH-propylene reaction proceeds by addition.
Morris et al. (1971) investigated the OH-propylene reaction in a flow-
discharge reactor coupled to a time-of-flight mass spectrometer. Adduct
peaks were observed. It was concluded that OH adds to propylene and that
the adduct is collisionally stabilized. In the C3D6 + OH reaction, they
observed that H was retained in the final aldehyde product, while D was lost,
Furthermore, acetaldehyde was a major product of OH + C2H4, and propional-
dehyde was observed in the reaction of OH + C3H6. These results indicate
that addition followed by hydride shift is a principal mechanism for the
elementary OH-olefin reaction:
CH3CHCH2OH + CH3CH2CH20- , (26)
CH,CH=CH9 + OH-
3 * \ OH 0-
CH3CHCH£ •»• CH3CHCH3 (27)
Reaction (26) corresponds to terminal addition, and Reaction (27) to internal
addition. Preliminary results obtained by Slagle et al. (1974) also indicate
that, to the extent that OH addition products were observed, the hydroxyl hydro-
gen is retained in the product. However, both these studies were done in the
absence of 02. For Reactions (26) and (27) to occur in air, the hydride shift
must be so fast that it precludes the addition of 02. By incorporating these
reactions in our mechanism as presented above, we have implicitly assumed
that this is true.
Although the experiments cited above provide mechanistic insights,
they do not indicate the relative importance of Reactions (26) and (27). One
would expect terminal addition to dominate because it yields the thermody-
namically favored secondary radical. The best agreement between model predic-
tions and the data was obtained with a terminal/internal ratio of 4. The
present OH-propylene mechanism predicts the formation of propionaldehyde and
minor amounts of acetone in accord with the UCR product measurements.
-------
22
2. Unimolecular Decomposition of Alkoxyl Radicals
Carbonyl product yields are determined to a significant extent by the
mode of reaction of alkoxyl radicals, short-lived intermediate products of
hydrocarbon oxidation. In addition to Reactions (26) and (27), these species
are formed from alkylperoxy radicals in reactions with NO or SOe [Reactions
(42) and (51) in Table 1]. The fates of alkoxyl radicals that have been con-
sidered are decomposition [Reaction (35)], reaction with 02 [Reaction (36)],
and reactions with NO or N02- The last two reactions were summarily investi-
gated and found to be relatively unimportant. Yields of nitrites and nitrates
were very small,* and so RO- + NOX reactions are not considered here. In -this
section and the next, the first two modes of reaction are discussed.
Recent experiments by Batt et al. (1974) have resulted in the first
absolute measurement of rate constants for alkoxyl radical unimolecular decom-
position at high pressure. By thermally decomposing alky! nitrates, monitoring
fractionation products, and using their own rate constants for RO + NO reac-
tions, they were able to determine rate constants for RO decomposition. They
did not estimate the accuracy of these rate constants. But, considering other
figures they reported, we believe that an uncertainty factor of 8 is reasonable.
Even with this uncertainty, their new values are several orders of magnitude
higher than most previous estimates (reviewed in Batt et al., 1974).
Batt et al. (1974) obtained rate constants only for the decomposition of
i-CsHBO-, s-C4HgO-, and t-C^gO-, whereas other isomers of these species and
shorter alkoxyl radicals appear in the mechanisms of Chapter III. We estimated
additional rate constants when necessary by using smog chamber data and noting
that, in Batt et al.'s results, the unimolecular decomposition rate decreases
with decreasing numbers of skeletal carbon atoms.
* Alkyl nitrates have been detected at UCR by Fourier interferometry, but
available literature (e.g., Spicer and Miller, 1974; Kopczynski et al.,
1974) leads us to believe that quantitative yields are indeed low. An
unknown oxidant peak observed during butane-NOx experiments at UCR was
attributed to butyl nitrate, but the estimated 6-hour concentration was
on the order of 1 pphm, which corresponds to only about 1 percent of the
reacted butane.
-------
23
3. Alkoxyl Radical— 0? Reactions
The individual rates of Reactions (35) and (36) are of little concern
in developing a smog mechanism. But their ratio determines the course
of the overall reaction. Unfortunately, no such ratio measurements are
available. Those used in the present formulation were determined largely
from UCR product distributions.
Available rate constants for the reaction
CH30- + 02 -> H2CO + H02- (36.1)
vary over two orders of magnitude. Garvin and Hampson (1974) have recom-
mended that k36 1 % 4.5 x 10~3 pprn-l min"1, whereas Mendenhall et al. (1974)
have estimated that k3g 1 % 0.8 ppnr1 min"1. In the latter study, k36 1
was determined from measurements of t-butylnitrate pyrolysis and a (very
uncertain) rate constant ratio for the reactions of 02 and NO with CH30-.
The authors concluded that a more direct measurement of k35 ^ was desirable.
Therefore, while Mendenhall et al.'s value serves as a guideline, considerable
freedom has been taken in the present study in adjusting rate constants for
the RO- + Q£ reactions. For smog modeling, the need for rate ratio measure-
ments is even greater.
4. Propylene + 03
As noted in Subsection 1, the importance of the 03-olefin reaction in
photochemical smog has long been recognized. Consequently, a large number
of kinetic studies have been performed, and reasonable agreement on the
rate constant for propylene + 03 has been reached (Becker et al., 1974;
Stedman et al., 1973; Garvin and Hampson, 1974). The mechanism for this reac-
tion is in a greater state of flux. However, there is much evidence in favor
of a recent mechanism postulated by O'Neal and Blumstein (0-B) (O'Neal and
Blumstein, 1973; Finlayson et al., 1974). The 0-B mechanism supplants the
Criegee mechanism. The Criegee mechanism has proved to be a satisfactory
-------
24
explanation of 03-olefin chemistry in solution (Leighton, 1961), but not
in the gas phase, where the formation of a "zwitterion" intermediate is not
as appealing.
The 0-B mechanism proceeds through the formation of an equilibrium
between a molozonide and an oxy-peroxy biradical. This biradical may
undergo many transformations, but, based on O'Neal and Blumstein's estimates,
the most likely is an a-hydrogen abstraction:
0
(I
»CHCHCH,
biradical-
OOH
OOH
0
Both Routes a and b could occur, depending on the nature of the biradical
intermediate. The a-keto hydroperoxide products of Reaction (II-6) will be
formed in an excited state. Their fractionation results in the following
overall reactions:
CH2=CHCH3 + 03-
CH3C02 + OH-
>CH3CHO 4 CO + H02 + OH-
(28)
(29)
These products are consistent with the mass spectra obtained by Becker et al.
(1974). The split between Routes (28) and (29) is determined by which hydro-
peroxide is formed in Reaction (II-6). In the present work, k29/k20 = 2 was
chosen arbitrarily to improve model predictions. Combining Reactions (28)
and (29) with this rate constant ratio gives
-------
25
CH0=CHCH- + 0-3-4-0.33 H9CO + 0.33 CHJXk + 0.67 CH-CHO + 0.67 HO.
£.63 £. J C. 3 *-
+ 0.67 CO + 1.0 OH-
This reaction can be compared with Reaction (33) in Table 8 of last year's
report:
CH2=CHCH3 + 03 4 0.75 HgCO + 0.25 CH^Og + 0.5 CH3CHO + 1.0 H02
+ 0.75 OH- + 0.75 CO
The new reaction differs from last year's in showing reduced formaldehyde,
H02, and CO yields in favor of increased OH', acetaldehyde, and CH3C(0)0;j.
Becker et al.'s (1974) yields of CO from propylene + 03 was 0.6 molecules
of CO produced per molecule of propylene consumed. Because secondary reac-
tions of aldehydes could have contributed to this yield, the stoichiometric
coefficient of 0.67 is probably an overestimate.
5. Radical-Radical Reactions
The reactions discussed so far have been hydrocarbon-organic chain ini-
tiation and propagation reactions. Chain termination occurs through radical-
radical reactions, such as NO + OH- •> HN02 [Reaction (13)], N02 + OH- + HNOs
[Reaction (14)], or H02 + H02 + H202 + 02 [Reaction (46)]. To provide ade-
quate damping, one must include similar reactions for organic radicals. This
can be done with reactions analogous to.those just cited.. In the present for-
mulation, the production of organic hydroperoxides,
R02 + H02 * ROOH + 02 , (47-48)
has been chosen. On an effective collision probability basis (likelihood
of an R02-H02 collision) these reactions were assigned a rate constant of
one-half that of the reaction H02 + HOg •»• H202 + 02 [Reaction (46)1.
Further thermal or photolytic reactions of organic hydroperoxy species have
not been included.
-------
26
6. PAN Chemistry
The formation of peroxyacetylnitrate (PAN), and its homologs, occurs by
another radical-radical reaction. Reaction (44),
0 0
RCO^ + N02 -». RC02N02 , (**)
was discussed in last year's final report (Hecht et al., 1974b). It was
speculated there that PAN might hydrolyze on the walls of the UCR chamber.
Although this undoubtedly could occur, an analogy to N20s suggests that gas
phase collisional destruction could be several orders of magnitude faster
than surface reactions under ambient conditions. Thus, we presently pro-
pose that PAN may undergo a thermal decomposition reaction, resulting in
the rupture of the peroxy and carbon-carbon bonds:
°9
(Mf) PAN -5 N03 + C02 + CH302 (+M) (45)
Based on data contained in Benson (1968) and Doma.lski (1971), this reaction
is exothermic by about 14 kcal mole"1.
The occurrence of Reaction (45) is supported by the experiments of
Schuck et al. (1972) and recent PAN decay experiments in the Riverside cham-
ber (UCR monthly report No. 4). In the former study, PAN was found to oxi-
dize NO to N02- The reaction was first order in PAN and zeroth order in NO.
The ratio of C02 produced to PAN consumed was nearly 1. The ratio of N02
formation to PAN consumption was approximately 2 in a nitrogen atmosphere,
but was much greater than 2 in an oxygen atmosphere. This is further evi-
dence for the occurrence of Reaction (45), followed by NOs + NO + 2N02
[Reaction (6)] in an N2 atmosphere, and Reaction (6) plus
CH302 + NO -»• CHaO- + N02 [Reaction (42)], CH30- + 02 •* H2CO + H0£ [Reaction
(46)], and H02 + NO -> OH- + N02 [Reaction (15)] in an 02 atmosphere.
Schuck et al.'s rate constant, k45 = 2.06 x 10'2 min"1, is 10 times that
obtained from the half-lives observed in the UCR chamber. This difference
-------
27
may be due to additional wall decomposition in Schuck et al.'s reactor.
The Riverside half-lives of 5.7 ± 0.1 hours in the light and 5.5 ± 0.4
hours in the dark provide further confirmation that PAN does not photo-
decompose at an appreciable rate (Leighton, 1961).
C. PHOTOCHEMISTRY
A distinctive feature of "Los-Angeles-type" air pollution is the role
played by sunlight in its causation. The free radicals that initiate the
process of oxidant production in Los Angeles air come from photolytic
splitting of molecular bonds. In smog chamber studies of the type considered
in Chapter III, sunlight is replaced by artificial illumination. Aside from
the overall intensity of the light source, the spectral distribution of pho-
ton flux is the major light source characteristic. The spectrum of the
Riverside solar simulator and the solar spectrum are reproduced in Figure 4.
The solar simulator consists of a light source and a light filter. The dark
solid line in Figure 4 represents the filtered spectrum.
1. Photolysis Rate Constants
Given a light source spectrum, such as that in Figure 4, rate constants
for the various photolysis reactions included in Table 1 can be computed.
For this purpose, quantum yields and absorption coefficients for the absorbing
molecules must be known. Photolysis rate constants can then be computed from
k = /I <(. a dv, (H-7)
J vyv v
and
I = 1° F , (II-8)
V V V
where
1° = photon flux provided by the light source at wavelength v,
av = absorption coefficient for the absorbing molecule,
Fv = filter factor,
4>v = quantum yield (molecules dissociated per photon absorbed),
-------
28
E
c
o
0>
in
C\J
I
c_
O
l/l
c.
o
o
CL
ST
r—
O
X
3
Sunlight
U.C. Riverside
Solar Simulator
Simulator with
20 nm Filter Shift
300
400
500
€00
nm
FIGURE 4. SPECTRA OF SUNLIGHT AND U.C. RIVERSIDE SOLAR SIMULATOR
-------
29
and where integration extends over the entire light spectrum. Rate constants
for the present investigation were computed from Eq. (II-7) using I values
V ^
obtained when UCR first installed their solar simulator in the chamber, and
values of ay and ^ extracted from Calvert and Pitts (1967) and Johnston and
Graham (1974). The values of these rate constants, normalized by k,, for
N02 -*• NO + 0 [Reaction (1)] are presented in Table 3.
Measurements of !q are made periodically at UCR, and k^p: 9/4 k]) is
obtained from light meter readings during the Battelle simulations. The
procedure used in carrying out the computations presented in Chapter III
was to multiply the measured values of Iq by the ratios in Table 3 to obtain
the needed photolysis rate constants. The shortcomings of this procedure
are critically assessed below. Obviously, periodic measurements of Iv
would facilitate a more accurate approach.
2. Spectrum Decay
For most of the photoabsorbers participating in smog reactions, mole-
cular dissociation occurs primarily as a result of the absorption of ultra-
violet light. As is to be expected, different chemical species have different
absorption spectra. For example, the absorption by N02, resulting in NO and
0, is largest and fairly uniform over the 300 to 400 nm range, whereas the
absorption by acetaldehyde, to form either Cffy and CO or CH3 and HCO, exhi-
bits a peak in the 250 to 300 nm range and very little absorption elsewhere.
Hence, the ratio of k-j to other photolysis rate constants will be very sensi-
tive to the intensity and distribution of UV light from the irradiation source.
Periodic measurements of k-| by experimentalists at UCR have shown a
consistent light deterioration. Between smog chamber experiments EC-38 and
EC-60 k-j decreased by about 40 percent (UCR monthly report No. 5). The cause
of this reduction was not investigated, but there are two obvious possibili-
ties. The first is a reduction in light source emission, perhaps resulting
from a buildup of UV-absorbing material on the inside of the light bulb
(Burton, 1975). The second is increased UV absorption by mirrors. Absorption
.by mirrors could slowly impair their reflection properties. Indeed, the
Data obtained through private communications.
-------
30
Table 3
PHOTOLYSIS CONSTANT CHANGES FROM UV LOSS
Reaction
N02 -> NO + 0
HNOg •+ OH- + NO
03 •> 02 + 0(1D)
03 •*• 02 + 0(3P)
H2CO -> H- + HCO-
HgCO -v H£ + CO
CHaCHO -> Products^
CH3CH2CHO + Products^
H202 -*• 20H-
MEK -> CHaCHO + CH3C(0)'
standard
Ratio
1.0
0.070
0.026
0.035
0.0049
0.011
0.0077
0.0085
0.0036
0.0036
5 nm Shift
1*
0.070
0.013
0.034
0.0040
0.010
0.0060
0.0065
0.0031
0.0024
10 nm Shift
1*
0.070
0.006
0.033
0.0031
0.0089
0.0045
0.0047
0.0028
0.0015
20 nm Shift
1*
0.069
0.0009
0.0031
0.0018
0.0069
0.0022
0.0022
0.0021
0.0005
Iq itself was reduced 1.9 percent by a 5 nm shift, 4.2 percent by a 10 nm
shift, and 10 percent for a 20 nm shift. Reduced ratios listed have been
renormalized to new Iq.
These products include radicals and stable species, as shown in Table 1.
Precise quantum yields are not known. We have assumed that the quantum
yield for stable products is approximately one-third of that for radicals.
-------
31
mirrors were observed to be damaged and were sent out to be recoated after
Run EC-61.) This reduced reflection and increased absorption would probably
be most pronounced in the UV region.
Although spectrum decay must ultimately be determined experimentally,
its potential effect can be assessed theoretically. For this purpose, rate
constants were recomputed using the following variation of
k = fl'et 4 d , (II-9)
J v V V V
where
I1 = F -1° , (11-10)
v v-y v
and y is the magnitude of a shift in the filter factor to lower frequencies.
Replacing Iv in Eq. (11-7) by 1^ in Eq. (11-9) reduces the UV intensity. The
spectrum change due to a 20 nm shift in the filter factor is illustrated in
Figure 4. Percentage reductions in rate constant ratios, computed for v = 5,
10, and 20 nm, appear in Table 3, along with their unshifted values. In
Figure 5, the effect of a 20 nm shift on smog profiles is illustrated. The
spectrum deterioration clearly delays the N02 peak, as observed experiment-
ally at UCR. Although Experiments EC-16 and EC-60 were nominally the same,
EC-60 took almost twice as long to reach the N0£ peak. Using only the re-
ported initial conditions and values of k-j, we could not reproduce this
delay by computer simulation (see Figure 6). However, a simulation employing
a 20 nm filter shift showed that the observed delay could easily be accounted
for by UV spectrum deterioration. We concluded that more complete spectrum
characterization in smog chambers is needed.
D. S02 OXIDATION
In the previous two years, our smog chemistry modeling efforts have
focused on systems containing olefins, paraffins, and NOX. During this
year, we also considered systems containing S02 and toluene. The next two
sections are devoted to a discussion of chemical processes related to these
two species.
-------
1-00
.80
Q.
Q.
O
4J
m
01
O
O
.60 --
.20 —
0-DO •-*«
0-0
;Propylene
Time (minutes)
y& Propylene
X N02
9K 03
X Propylene
X 03
3DQ-Q
No shift
20 nm shift
350-D
40Q-0
FIGURE 5. EFFECT OF A 20 NM FILTER SHIFT ON SMOG PROFILES
to
.f\>
-------
1.20
0.90
a.
0.
c
o
P 0.60-
OJ
o
o
o
0.30-
0,0
0.0
OBSERVED DATA
NO nm SHIFT SIMULATION
20 nm SHIFT SIMULATION
50.0
100.0
150,0 200..0
Time (minutes)
250.0
300.0
350.0
400.0
CO
CO
FIGURE- 6. EFFECT OF A 20 nm FILTER SHIFT ON NOo BEHAVIOR FOR EC-60
-------
34
1. Some Observations
Past smog chamber studies of irradiated S02-NOx-hydrocarbon and
S02-03-olefin mixtures have resulted in a variety of qualitative and quanti-
tative observations, many of which are inconclusive or even contradictory.
Of particular significance is an uncertainty about the effect of S02 on the
yields of carbonyl products (see reviews by Leighton, 1961; Wilson and Levy,
1970) and ozone (Wilson and Levy, 1970; Altshuller et al., 1968). Wilson and
Levy (1970) have shown that the overall reaction is strongly dependent on
relative humidity. Undoubtedly, chamber effects also play a role in causing
these inconsistencies. There is agreement, however, on the observation that
the addition of S0£ to the hydrocarbon-NOx-air system results in increased
aerosol production and that this aerosol formation occurs after the N02 peak.
It has also been generally observed that the addition of propylene to an irra-
diated S02-NOX mixture increases S02 consumption and aerosol production. The
aerosol is thought to be chiefly sulfuric acid and water, though it may contain
small concentrations of organic and nitrite-type material (Filby and Penzhorn,
1974; Bufalini, 1971).
An examination—which appears later—of the Battelle data (S-110 and S-115)
permits two further observations. The addition of ^ 0.5 ppm S02 to an irra-
diated N0x-olefin-air mixture does not substantially alter either the ozone
production or the propylene oxidation rate.
A free radical mechanism for S02 oxidation is in accord with these
observations. Because oxidation occurs mainly after the N02 peak and because
added hydrocarbon increases aerosol production, the radicals that oxidize S02
are probably the same as those oxidizing NO. These radicals must be either
organic or of organic origin. The low reactivity of S02 accounts for the
smallness of its effect on both hydrocarbon oxidation and ozone production.
Sulfuric acid is probably the main oxidation product of S02-
-------
35
2. Kinetic Mechanisms for SO? Oxidation
Recent investigators of atmospheric S02 chemistry (e.g., Davis et al . ,
1974; Calvert and McQuigg, 1975; Castleman et al., 1974) have suggested that
the oxidation of S02 by OH' and HO;? is the primary means of S02 removal in
the gas phase. Several recent measurements of the S02-OH- rate constant are
reviewed in Table 4. They cover a wide range, but the value of 900 ppm~l min"'
selected for the current mechanism encompasses the full range with a 50 percent
uncertainty.
As shown in Table 4, Davis (1974) measured a rate constant of
l<50 = I-3 PPm~^ min'l for
H02 + S02 + OH- + S03 . (50)
Because alkyl and acyl peroxy radicals should be more reactive than HO^, we
have incorporated Reactions (51) and (52) into the kinetic mechanism with
rate constants of 1.5 ppm~l min"^, slightly higher than
R02 + S02 -> RO- + S03 , (51)
R02 + C02 + S03 (52)
Reactions of N03 and N205 with S02 were postulated by Wilson and Levy
(1969) to explain a rapid reaction observed between N02» 03, and S02- As
shown in Table 4, this could not be explained by a ; direct reaction of S02
with 03. Thus, the following sequence was invoked:
NQ2 + 03 + N03 + 02; N03 + N02 ^ N^ , (5, 7 and 8)
N03 + S02 -> S03 + N02 , (49)
N2°5 + S02 * S03 + 2N02
-------
RATE CONSTANTS FOR S02 OXIDATION.
(In ,.„„,-1 r.in-1)
-
S02
Reaction
» OH - HSOj
+ o -iL so3
Davis (1974)
1350 i 130
(7.6±0.9)exp(-2240/RT)
Calvert
and HcQuiqg
(1975)'
Castleman
et al.
(1974)
590
Others
C60 ± 120
(COX, 1974)
3000
(Leighton, 1964)
Present
Mechanism
900
S02 + N03
N02
< 14.7
10-5
270 ± 50
(Mulcahy et al., 1967)
14
S02 «• H02 --SO'3 + OH-
1.35 ± 0.2
S02
S02
S02
+ RC(0)02 - R + CO + S03 )
+ ROJ •* RO- + S03 j
S02 + N205 - 2fi02 + S03
S02 + 03 •» $03 + 02
$03 + H20 JU K25:04
R02 + S02 » ROS03
RC+00- + S02 - S03 + RCHO
HSOj + 02 - HSOs * S03 + H02
HSOJ + NO -> HSO^ + N02
HS04 + HOJ •* H2504 + 02
HSOi * N02
H20
• H2S04 + HN03
<. 1.-2 10-8
3 x 10'7
10-4k(OH- * S02).
if HS04 is a product
% HOj + S02
< 6.2 10'8
< 1.2 ID'10
l°"4k(OH- + S02)
740
i. 10"4 min-1
(Leighton, 1961)f
1.3
1.5
•v 6.3
(Cox and Penkett, 1972)
Immediate
See text
800
9000
1 x 104
* Quantum yields for $03 production ($$03) measured in an S02 atmosphere vary from 3 x 10"4 (Cox, 1972) to 0.1 (Chung et al., 1975). Chung
f. al. (1975) explained this discrepancy by considering secondary reactions. Extrapolation of these results to atmospheric condition 1s
not possible at present because $50, varies with atmospheric pressure and type of diluent.
CO
en
-------
37
The upper limits of the N20,- + S02 reaction rate (Table 4) indicate its in-
significance. In contrast, the upper limits recorded for k»9 (Table 4)
are in severe disagreement. We can estimate k.g assuming reactions of H02
and N03 with S02 and NO are similar (all are "oxygen abstraction" reactions).
Using the ratio, k(s02 + H02VHNO + H0p)> to be approximately 10" (from
Tables 2 and 4) and k(^g + NO) - 1° PPm~ m'in~ (from Table 2) we find
3
k49 = k(S09 + H09)/k(NO + HOA) ' k(NOQ + NO) ~ 10 ppm" min"
L ^ *J 6 (11-11)
This estimate is in closest agreement with Davis1 (1974) reported value.
Reaction (49) could therefore be significant, and it has been initially
included in the mechanism with k4g = 14 ppm min"1 (Davis1 value).
Cox and Penkett (1972) found that SOp added to the olefin-03 system
was oxidized to sulfuric acid aerosol. They also found that, when the ozone-
olefin consumption ratio in the absence of S02 was greater than 1, the addi-
tion of S02 reduced this ratio. The addition of S02 also enhanced carbonyl
product yields. From these and other results, Cox and Penkett inferred that
a short-lived intermediate of the ozone-olefin reaction, which they specu-
lated might be a zwitterion, was responsible for SOp oxidation. However, in
light of the discussion of the mechanism for 63 + propylene in Subsection 4
and the data in Table 4, it seems more reasonable to account for Cox and
Penkett's observations in terms of reactions of S02 with hydroxyl, hydroper-
oxyl, peroxyalkyl, and peroxyacyl radicals rather than zwitterions.
The final modes of S02 oxidation included in Table 4 are direct oxidation
by 0 atoms and photooxidation. The former can be ruled out, because the maxi-
mum 0 concentration is only about 10~9 ppm. With S02 in ppm concentrations,
this gives rates < 10~6 min"^. Photooxidation has been widely investigated
(Cox, 1972; Sidebottom et al., 1972; Demerjian et al., 1974b; Smith and Urone,
1974); the experimental results show that it is an unimportant oxidation pro-
cess in polluted atmospheres. The inefficiency of photooxidation is due, in
part, to the forbiddenness of direct excitation of S02 from the ground state
to its reactive triplet state SOg^B-j) (Sidebottom et al., 1972). Sidebottom
et al. (1972) estimated an upper limit of 2 percent S02 consumption per hour
for gas phase photooxidation in the atmosphere. However, as they noted,
-------
38
this could be an extreme upper IJmit because the efficiency for singlet-
triplet intersystem crossing may have been overestimated, and the assump-
tion that all quenching of 3S02 by 02 was chemical rather than physical was
probably faulty. Because of the results of the investigations cited above,
S02 photooxidation does not appear in the present mechanism.
Direct reaction between 3S02 and hydrocarbons has been reported at high
reactant pressures (Badcock et al . , 1971; Sidebottom et al . , 1971; Filby and
Penzhorn, 1974). However, at ppm levels, Demerjian et al . (1974b) found
that the primary result of the interaction of 3S02 with 2-butene was the
interconversion of cis and trans isomers. Thus, a direct chemical reaction
between ^sc and olefins under atmospheric conditions can also be ruled out.
As shown in Tables 1 and 4, $03 is produced by the reaction of
RC(0)02, H02, and N03 with S02. Castleman et al . (1974) found that S03
rapidly combines with water to form ^SO/p Presumably, ^864 then nucleates
and adds more water to form aerosol droplets (H2S04-nH20) . The studies car-
ried out by Cox and Penkett (1972) and the more recent mass spectrometric
measurements made by Schulten and Schurath (1975) indicate that ^$04 is
the major sulfur-containing product of the olefin/03/S02 reactions. In
Bufalini's (1971) survey, she reported that the aerosol produced as a result
of the irradiation of 3 ppm of lower olefins in the presence of 1 ppm N02 and
0.5 S02 at 50 percent relative humidity was primarily composed of sulfuric
acid. Accordingly, no other products have been accounted for in the present
mechanism. Furthermore, we assumed that $03 is rapidly converted to
Davis et al. (1974) and Calvert and McQuigg (1975) have speculated on
the formation of compounds containing S, 0, H, and N when NOX is present.
Presumably, the formation of these nitrogen-containing species results from
radical -radical reactions with oxidation products of the OH--S02 adduct. For
example, HS04, formed as shown in Table 4, could combine with N02 to form
HS04N02- Subsequent hydrolysis at the wall would produce nitric and sulfuric
acid:
-------
39
HS°4 + N02 1U?r H2S04 + HN03
This represents one possible fate of the HS04 molecule produced in the
following series:
02
OH- + S02 -»• HSOg — ^HS05 > (50)*
HSOg + NO -> HSO^ + N02 . (54)
Other possible reactions of HS04 (and HS05) have been assembled by
Calvert and McQuigg (1975), including hydrogen abstractions from alkanes
and H02* and addition to olefinic double bonds. Davis et al . (1974) proposed
that HS06 could be produced by the addition of Q£ to HStty. It could then
oxidize two'NO molecules, reforming HS04, which would then repeat the process
This mechanism was presented to explain an ozone bulge found by Davis et al .
(1974) in the plume downwind of the Potomac Electric Power Company's power
plant at Morgantown. However, laboratory studies have shown that the addition
of S02 to the hydrocarbon-NOx system actually causes either little change
or even a reduction in 03 production (Wilson and Levy, 1970). As stated
at the outset of this discussion, we assumed in the present mechanism
formulation that S02 has only minor effects on 03 yield; so Davis et al.'s
suggestion has not been incorporated. In addition to Reaction (56), a
simple termination reaction,
HS04 -> H2S04 + 02 , (55)
* An alternative to the last step in Reaction (50), HS03 + 02 -> H02' +
seems implausible thermodynamically. AH is approximately 0 to 20 kcal
mole-1 for this step, compared with AH ^ -50 kcal mole-1 for the last step
In Reaction (50).
-------
40
has been included. In light of the current lack of experimental studies,
this seems a pragmatic and adequate closure to the S02 oxidation mechanism.
E. THE TOLUENE-NOX-AIR SYSTEM
The development of a toluene oxidation mechanism has been greatly hin-
dered by a lack of information on the pertinent reactions of toluene and by
a dearth of detailed product analyses of smog chamber experiments. For
these reasons, the mechanism proposed in Subsection 3 below is sketchy and
possibly premature. The first two following subsections briefly review the
currently available literature relating to the smog reactions of toluene.
1. Toluene + 0(3P)
Jones and Cvetanovic (1961) and Atkinson and Pitts (1975) have determined
rate constants for this reaction. According to the latter, k (toluene + 0)
^ 120 ppnf' mirr1. At a typical 0 concentration of ^10~9 ppm, this reaction
is of little importance. In the present mechanism, OH- is taken to be the
sole oxidant.
2. Toluene + OH-
Davis et al. (1975) measured a rate constant of 9.0 ± 0.6 x 103 ppm'1
min~' for the toluene-OH- reaction under a pressure of 100 torr (M = He),
where the reaction was still in the pressure fall-off region. However, extra-
polation to room pressure via a Lindemann. (1/k versus 1/P) plot indicated
that this value was very close to the high pressure limit. The pressure
effect Davis et al. observed (k at 100 torr was nearly twice its value at
3 torr) is evidence that a significant fraction of the reaction proceeds ini-
tially through the formation of a collisibnally stabilized adduct. The rate
constant measured at 3 torr (k = 5.3 ± 0.4 x 10~3 ppm"1 min'1) implies that
addition to the ring accounts for at least one-half of the toluene-OH-
reaction.
-------
41
Doyle et al. (1975) used a 6000-liter glass smog chamber to measure the
disappearance rates of several aromatic compounds relative to that of butane.
The hydrocarbons, along with ^0.27 ppm NO and ^0.04 ppm N02, were irradiated
and aromatic concentrations were measured after 1 and 2 hours. After pre-
senting evidence that OH- was the primary hydrocarbon oxidant in their system,,
Doyle et al. obtained k(toiuene + OH.) = 6-2 ± 2-3 x lo3 PPm min , though,
of course, this value depends on the value of the rate constant used for
butane + OH. Doyle's measurements of relative rates for the three isomers
of xylene indicate that the primary mechanism for aromatic hydrocarbon-OH-
reactions is electrophilic addition to the ring (see Section 3).
Although neither of these two investigations give direct evidence of the
existence of an abstraction route, the resonant stability of the phenalkyl
radical, produced by hydrogen abstraction from the methyl group, makes this
an appealing pathway. That this process does indeed occur (to some extent)
is implied by the observation of benzaldehyde and peroxybenzolnitrate (PBZN)
in the toluene-NOx system by Heuss and Glasson (1968). Following Davis et al.
(1975), we have set a limit of 50 percent for the relative frequency of
H-abstraction to OH-addition.
Another noteworthy observation was made by Altshuller et al. (1970), who
identified .formaldehyde as a product of the same (toluene-NOx) system. Simul-
taneously, they detected only trace amounts of acetaldehyde. Formaldehyde
could be formed subsequent to oxidation of the methyl group on toluene or
after displacement of the methyl group by OH.*
The investigations just cited are the basis for our current toluene oxi-
dation mechanism. Obviously, more product measurements are sorely needed.
3. The Proposed Mechanism
Reactions (33) and (34) are the initial H abstraction and OH addition
reactions:
* Another possibility, not included in the current mechanism, is the opening
of the aromatic ring with the subsequent formation of fractionation products.
As these are likely to include acetylene, testing for this possibility experv
mentally should be straightforward.
-------
42
0,
LQ) + OH-
fH2°2
(O) + H20
(33)
OH 0,
-> [0
OH
HO
(34)
In both cases, an immediate reaction with 02 has been assumed. The o-cresol
product of Reaction (34) is taken as a prototype for o, p (i.e., electro-
philic addition) cresols.*
The peroxy phenalkyl radical, produced in Reaction (33), could react
as follows, to produce benzaldehyde and PBZN.
:HO
NO + \O]
N02 + CQ)
H02 + (O]
(42, 36)
:HO
L Ic
NO,
CO] + OH--^H20 + (g) -^2_^ [0]
(32, 44)
Other radical reactions, selected in analogy to those included in Table 1
are
* The procedure of using o- as a prototype for o, p- was used throughout.
-------
43
(0} - H?CO .
(35)
NO
(43)
both of which produce a peroxy-phenyl radical. Subsequent reactions [Nos,
(42'), (42")] produce (o,p) quinones:
. (42', 42")
Because of the diradical character of quinones, they could react further, but
such speculation has not been included here.
If in Reaction (34) OH- has combined with toluene at the ring-methyl
bond, Reaction (34) would become
(Oj + OH-
0,
3H
(Q) +
(34')
-------
44
Phenol and cresols [from Reactions (34) and (34')] should be very reac-
tive. Gitchell et al. (1974) found that addition of phenol had an inhibiting
effect on propylene-NOx systems. Therefore, they suggested that the
OH- + phenol reaction proceeds by hydrogen abstraction from the phenolic
hydroxyl group. This, and their further speculation that NO and N0£ add
to the phenoxy radicals thus produced, seem reasonable. However, in an
attempt to match the predictions of the kinetic mechanism with the NAPCA
smog chamber profiles (see Chapter III), -OH hydrogen abstraction from
phenol and cresols were assumed to be followed by 02 addition as in
f3
+ OH ^>H,0 +
2
The above reactions, along with appropriate radical + H02 reactions,
constitute the present kinetic mechanism for toluene oxidation (Table 13,
Chapter III). Rate constants have been selected on the basis of reaction
analogies.
-------
45
III SMOG CHAMBER SIMULATIONS
Smog chamber studies done at the University of California, Riverside
(UCR), Battelle Memorial Institute, and the National Air Pollution Control
Administration (NAPCA) serve as the data base for validating the kinetic
mechanisms. Validation involves comparing smog profiles obtained by numeri-
cal integration of chemical rate equations with those obtained experimen-
tally. Numerical integration and the derivation of rate expressions from
stoichimetric equations and rate constants were done on a CDC-7600 digital
computer. The programs used (MODKIN and CHEMK) were developed under previous
EPA contracts by SAI and by G. Z. Whitten at Lawrence Berkeley Laboratory
specifically for this purpose (see Appendix A of Hecht et al., 1974a).
Unfortunately, rate constants and reaction mechanisms were not accurately
known for many of the elementary steps comprising the smog mechanisms. Addi-
tionally, chamber characterization was usually incomplete. Thus, an unavoid-
able aspect of validation was "tuning." Tuning involves adjusting parameters
within their range of uncertainty for the sole purpose of improving the fit
between experiments and predictions. As in the past, we attempted to minimize
the amount of tuning and to maximize the dependence on kinetic data.
Chamber characteristics that were usually available are k,, dilution
rate and heterogeneous 03 loss. The last two were incorporated as simple
first-order reactions. Rate constants for all photolysis reactions were
determined from k, through the use of Table 3. Other procedural aspects
and the simulation results appear in the ensuing subsections.
A. SIMULATIONS OF UCR DATA
The operating characteristics of the UCR chamber and peripheral equip-
ment were described in detail in last year's report (Hecht et al., 1974a).
The only addendum to be made here is to note calibration errors in 0^ and
-------
46
NOX measurements discovered during the year. The following corrections
to the data appearing in last year's report had to be made:
V "V1-2 '
NO1 = 1.2 NO ,
N02' = 1.2 N02 + 0.2 PAN
Primes represent the new values, and the absence of primes indicates old values.
1. Results for the Propylene-NO^-Air Block
A
The initial concentrations for the propylene-NOx block appear in Table 5
and Figure 7. The mechanism used to simulate these experiments is presented
in Table 6. Photolysis rate constants used in the simulations were computed
from a 10 nm filter-shifted spectrum. Chapter II presents these rate con-
stants, normalized by k]. The choice of a 10 nm shift was somewhat arbitrary,
but more complete spectrum decay characterization was not available.
Figures 8 through 23 present the results of UCR experiments and computer
simulations thereof. Throughout this chapter, experimental data are represented
by plotted symbols, and computer curves by unbroken lines. With the exception
of Run EC-12, the predictions of propylene, N02, NO, and ozone behavior are
fairly good. The exceptions are the lack of correspondence between the asymptotic
N02 levels in EC-18, and, consequently, low 03 predictions and a small delay in
the N02 peak of EC-11, which is also mirrored in the 03 profile.
The predictions of carbonyl products (except PAN) are consistently low. At
least part of this discrepancy might be attributed to spuriously high measurements
(Darnell, 1974), especially at low concentrations.
-------
47
Table 5
PROPYLENE-NOV EXPERIMENTS
A
(In ppm)
Initial Concentrations
EC
Run
11
12
16
18
21
Propylene
0.447
0.082
1.036
0.972
0.104
NO
0.115
0.106
1.12
0.106
0.558
N02
0.020
0.012
0.156
0.0142
0.066
NOX
0.135
0.118
1.27
0.148
0.624
llVSo/ MV/v
Ratio
0.15
0.10
0.12
0.12
0.11
kl
0.223
0.223
0.223
0.223
0.223
1.0
1
Q.
C
O
(0
c
| 0.5
o
o
ox
~ 0.1
If) X
x 21
-
x 12 x 11 18 x
i i i
0.1 0.5 1.0
Initial Propylene Concentration (ppm)
Note: The EC run number is given next to each point
FIGURE 7. PROPYLENE/ NOV FACTORIAL BLOCK
A
-------
48
*
Table 6
THE PROPYLENE OXIDATION MECHANISM
Rate Constant
__ _ Reaction _ (ppnH mln"1 )
N02 + hv -v NO + 0 0.223f
0 + 02 + M + 03 + M 2.08 x 10~5
03 + NO + N02 + 02 25.2
N02 + 0 -> NO + 02 1.34 x 104
N02 + 03 -»• N03 + 02 0.05
+ NO •> 2N02 1.3 x 104
+ N02 •*• N205 5.6 x 103
^- N03 + N02 24f
H20 + N205 -»• 2HN03 5 x 10'6
NO + N02 + 2H20 -> 2HN02 + H20 1.3 x 1Q-11§
2HN02 -> NO + N02 + H20 0.24
HN02 + hv ->• NO + OH- 1.5 x lO'2"1"
OH- + NO -»• HN02 3 x 103
OH- + N02 -»• HN03 1 x 104
H02 + NO -*• N02 + OH- 8 x 102
-------
49
Rate Constant
_ Reaction _ (ppm""1 min"1)
03 + hv + 02 + 0(3P) 7.3 x 10-3
03 + hv -> 02 + 0(]D) 1.3 x 10-3
0(]D) + M -> 0(3P) 8.6 x 104
0(]D) + H20 -> 20H- 5.1 x 105
03 + OH- -»• H02 + 02 87
03 + H02 •* OH- + 202 2<4
03 * wall 1
10"3t
C3H5 + OH- •*• CH3CH(0-)CH3 5.0 x 103
C3He + OH- •*• CH3CH2CH20- 2.0 x 104
03 -»• H2CO + CH3C(0)02' + OH- 0.013
C3H6 + 03 + CH3CHO + H02 + OH- 0.007
3
C3H6 + 0 -> CH3CH20- + HO? + CO 2.66 x 10
C3H6 + 0 •* CH302' + CH3C(0)02 2.66 x 103
H2CO + OH- -*• H20 + H02 + CO 2.1 x 104
CH3CHO + OH- -> H20 + CH3CO(0)02 2.1 x 104
CH3CH2CHO + OH- -> H20 + CH3CH2C(0)02 4.5 x 104
CH3CH(0-)CH3 -> CH302 + CH3CHO 8.35 x
-------
50
Rate Constant
Reaction (ppnH min~1)
CH3CH2CH20- -> CH3CH202 + H2CO 8.5 x 103t
CH3CH20- -> CH302 + H2CO 6.0 x 103f
CH3CH(0-)CH3 + 02 -> CH3C(0)CH3 + H02 0.04
CH3CH2CH20- + 02 -* CH3CH2CHO + H02 0.06
CH3CH20- + 02 -> CH3CHO + H02 0.04
CH30- + 02 -»• H2CO + H02 0.04
H202 + hv -> 20H- 1.06 x 10-3t
H2CO + hv -> H2 + CO 2.1 x 10-3t
H2CO + hv -> 2HO^ + CO 6.9 x lO'41"
CH3CHO + hv -v CH4 + CO 3 x lQ-4t
CH3CHO + hv •*• CH302 + H02 + CO 7 x 10'4t
CH3CH2CHO + hv •*• CH3CH3 + CO 3 x lQ-4t
CH3CH2CHO + hv -> CH3CH202 + H02 + CO 7 x 10"4t
CH302 + NO •*• N02 + CH30- 1 x 103
CH3CH202 + NO -»• N02 + CH3CH20- 1 x 103
CH3CH2CH202 + NO -> N02 + CH3CH2CH20- 1 x 103
CH3C(0)02 + NO + N02 + CH302 + C02 1 x 103
-------
51
Rate Constant
Reaction (ppm-1 mirr"*)
CH3CH2C(0)02 + NO + N02 + CH3CH202 + C02 1 x TO3
CH3C(0)02 + N02 -* CH3C(0)02N02 3 x 102
CH3CH2C(0)02 + N02 -»• CH3CH2C(0)02N02 3 x 102
CH3C(0)02N02 -> CH302 + N03 + C02 3 x l(T3t
CH3CH2C(0)02N02 -»• CH3CH202 + N03 + C02 3 x lO'31"
H02 + H02 * H202 +02 6 x 103
H02 + CH302 .-> CH3OOH + 02 3 x 103
H02 + CH3CH202 -> CH3CH2OOH + 02 3 x 103
H02 + CH3C(0)02 -»• CH3C(0)OOH + 02 3 x 103
H02 + CH3CH2C(0)02 -v CH3CH2C(0)OOH + 02 3 x
*
§
•
ppm~J mm
-------
X ND2
O 03
X NC
.30 T-
.24
.16
fO
OJ
o
o
o
•12 ~-
.06
0.00
100 0
SO - 0
3DB-0
350-0
Time (minutes)
MC-G-O
tn
ro
FIGURE 8. EC-11 SIMULATION RESULTS AND UCR DATA FOR N02> Og, AND NO
-------
.50 -r-
.MO
I- -30
o
a
.20 --
.10 -
o.oo
FORMALDEHYDE
PROPYLENE
0-0
SQ.O 100-0
15'J-O
cDQ.O 2CU-0
Time (minutes)
FORMALDEHYDE
• ..
PROPYLENE
300 :o 150 :o HO^O
tn
FIGURE 9. EC-11 SIMULATION RESULTS AND UCR DATA FOR PROPYLENE AND FORMALDEHYDE
-------
10 -r-
.08 --
K RCETGNE
X PRN
c
-------
.25 -p-
.20
ex.
O.
CD PCETFLCEHYCE
a
.10
.OS —
0-00
O.G
100-0
SOOiO 250.0
Time (minutes)
300 - 0
150-0
FIGURE 11. EC-11 SIMULATION RESULTS AND UCR DATA FOR ACETALDEHYDE
cr
tn
-------
.20 -T-
.16
a.
Q-
.12
o
§ .06
o
.OH
CD 03
H NO
S N02
0.00
0-0
50-0
i oo.o i50.o 200-0 ajiu.'ti 300 :o
Time (minutes)
FIGURE 12. EC-12 SIMULATION RESULTS AND UCR DATA FOR O, NO, and
MOO . U
en
-------
FORMALDEHYDE
PROPYLENE
.10 T-
.08 • —
I'.06
.04
o
o
.02
C.CO
Q.Q 5Q-0 IOG.0 15'J-O 200. Q 250 . 0 300. Q
Time (minutes)
FIGURE 13. EC-12 SIMULATION RESULTS AND UCR DATA FOR PROPYLENE AND FORMALDEHYDE
01
-------
.05
.04
.03
3 RCETRLDEHYK.
X PPN
-------
I .50 -r-
1.20 --
.90 --
to
§ GO
(J -CUI
O
.3D -- Z
X NO
S NO
O.CC
Time (minutes)
en
FIGURE 15. EC-16 SIMULATION RESULTS AND UCR DATA FOR 03, NO, AND N02
-------
o.
in
s-
CJ
c.
o
.36
.12
C.CQ
G-C
SO - 0
X RCtTONE
CD cCf.TFLCF.HYCE
X PPN
100 0
ISQ.'O "00-'0 250-0
Time (minutes)
:300.0
ACETALCEHYDE
PAN
?rJ-:-.l!
FIGURE 16. EC-16 SIMULATION RESULTS AND UCR DATA FOR ACETONE, ACETALDEHYDE, AND PAN
-------
I.SO -T-
1.20
.90
.
O» '
(J
o
o
.60
.3D --
0.00
X FORMALDEHYDE
* PROPYLENE
Time (minutes)
FIGURE 17. EC-16 SIMULATION RESULTS AND UCR DATA FOR PROPYLENE AND FORMALDEHYDE
-------
1.00 -r-
.80
-5 -60 --
to
-»->
(U
o
o
o
.40 --
.20 --
0-00
* Propylene
<2) Acetaldehyde
O.Q SQ:Q 100:0 ISQ:C 2uc:c 250:0 300-0
Time (minutes)
35U-0 HOQ-0
ro
FIGURE 18. EC-18 SIMULATION RESULTS AND UCR DATA FOR PROPYLENE AND ACETALDEHYDE
-------
.IB -r-
.12 --
,£ -09
o
•M
to
0)
o
o
.06 --
.03
0.00
X NO
X NOz
O PAN
X X « K H __%_ X K_
0-0 50-0 100.0
15U-Q 200-0 25U-Q
Time (minutes)
300-0 350-0 MOQ-0
CTI
CO
FIGURE 19. EC-18 SIMULATION RESULTS AND UCR DATA FOR NO, NOg, AND PAN
-------
.50 -r-
.40
QL
Q.
.30 --
c
-------
Propylene
Acetaldehyde
.15 -r
.12
^ .09
o
•r—
4J
10
-------
X NO
X N02
X Formaldehyde
FORMALDEHYDE
o.oo
en
FIGURE 22. EC-21 SIMULATION RESULTS AND UCR DATA FOR NO, N02, AND FORMALDEHYDE
-------
.010 -.-
.008 --
a.
Q.
IO
O>
U
.006
.004
.002
0.000
2°3
X PAN
0-0 50.0 IDO.O
aoc.'o
Time (minutes)
300 :o
40". C
FIGURE 23. EC-21 SIMULATION RESULTS AND UCR DATA FOR 03 AND PAN
-------
68
PAN predictions are sometimes high (EC-18 and EC-11) and sometimes low
(EC-16 and EC-12). Calibration of the panalyzer has been an enigma; thus,
data values may be suspect. But the chemistry of PAN is still largely unex-
plored, and this is very likely a large part of the problem.
The results for EC-12 are inexplicable. The data indicate a total
absence of an induction period, but mechanistic predictions completely
disagree with this observation. If the current mechanism's validity is
accepted, this completely disparate behavior can be resolved only by blam-
ing wall effects. The reactant concentrations in EC-12 are extremely low,
and the wall effects are hence, relatively, at their maximum. A small
source of radicals, such as photolysis of off-gassed aldehyde or HN02,
could provide the initial impetus to this system.* A recent experiment
done at UCR provides further evidence in favor of this explanation. CO
oxidation to CC^ was observed upon irradiation of a supposedly clean chamber
filled with dry synthetic air. The oxidation rate of O.I ppm per hour
with CO present at 100 ppm indicates that the OH- concentration in this
system could have reached ^10"^ ppm. Presumably, this was due to a wall
source of radicals.
2. The Butane-N0x-Air System
a. The Mechanism Used
The butane mechanism was considerably less successful than the propylene
mechanism, even though they are similar. . The modifications to the propylene
mechanisms to adopt it to butane included adding reactions pertiment to butane
and reflecting changes in k^ Photolysis rate constants used in the butane
simulations were computed from a 10 nm filter-shifted spectrum (as with propene)
The reaction changes are listed in Table 7. Most of the changes included in
Table 7 represent merely the addition of four-carbon species. Most notable is
the ratio of 2.4 to 1 for internal to terminal hydrogen abstraction by OH« from
* In these simulations, initial concentrations of 0.01 for formaldehyde and
acetaldehyde were used. The carbon balance obtained from product measure-
ments (excluding CO) at UCR is usually 60 to 80 percent. However, in EC-12,
this balance was about 110 percent—hence, the suspicion that unaccounted-
for sources (or initial concentrations) were present.
-------
69
Table 7
REACTION CHANGES HADE FOR THE BUTANE SIMULATIONS
(With the propylene chemistry eliminated and all photolysis rates scaled to k,)
Reaction Rate Constant
1 1 -1
C4H1Q + OH- —*• H20 + C4Hg02 1.0 x 10J ppm"1 min
°9 Til
C.H10 + OH- -^~H20 + CH3CH2CH(02)CH3 2.4 X ID"3 ppm"1 min
°9 11
C,H,n + 0 —^*-CH,CH,CH(0:)CH, + OH- 64 ppm"1 min"1
4 1U 3 L. t. *5
°9 All
CH3CH2CH2CHO + OH- —£- H20 + CHjCHgCHgCtOjOg 4.5 x 10* ppm min"1
0, All
CH3CH2C(0)CH3 + OH- -4- CH3C(0)OH + CH3CH202 1 x 10A ppm"1 min"1
CH3CH2CH2CH20- —*- CH3CH2CH202 + HgCO 8.5 x 103 min"1
CH3CH2CH2CH20- + 02 —- CH3CH2CH2CHO + H02 0.2 ppm"1 min"1
CH,CH9CH(0-)CH. + 0? —*- CH3CH?C(0)CH, + HOj 0.2 ppm"1 min"
Ob O t we. «3t
CH3CH2CH(0-)CH3 —»• CH3CH202 + CH-jCHO 1.3 x 104 min"1
20.
CH3CH2C(0)CH3 + hv —^ CH3CH202 + CH3C(0)02 7.8 x 10 min"
CH3CH2CH2CHO + hv —*- CH3CH2CH3 + CO 3.1 x 10"4 min"1
202 -4 -1
CH3CH2CH2CHO + hv —^ CH3CH2CH202 + H02 + CO 7.2 x 10 H min
CH3CH2CH2CH202 + NO —»- N02 + CH-jCHgCHgCHgO- 1.0 X 103 ppm"1 min"1
CH3CH2CH(0')CH3 +'NO —•- N0£ + CH3CH2CH(0-)CH3 1.0 x 103 ppm"1 min"1
v/\ *i 1 1
CH3CH2CH2C(0)02 + NO -^ N02 + CH3CH2CH202 + C02 1.0 x 103 ppm"1 min"1
CH3CH2CH2C(0)02 + N02 —•- CH3CH2CH2C(0)02N02 3.0 x 102 ppm"1 min'1
°2 3 -1
CH3CH2CH2C(0)02N02 -^ CH3CH2CH202 + N03 + C02 3.0 x 10° min
CH3CH2CH2C(0)02 + N02 —^ CH3CH2CH2C(0)02H + 02 3.0 x 103 ppm"1 min"1
-------
70
C.HIQ. This is lower than the ratio for abstraction by Cl atoms (Morrison and
Boyd, 1971), although one might expect it to be higher because OH- has less
electron affinity than Cl (Hefter et al., 1972) and should therefore be more
selective in its attack. A ratio measurement is desirable.
Another element of Table 7 worth commenting on is the reaction of OH-
with CH3CH2C(0)CH3 (methlyethylketone, or MEK) . The rate of hydrogen abstrac-
tion from MEK can be computed from Greiner's formula for paraffins (Oohnson
et al., 1970), with a modification to take into account the lowering of the
secondary C-H bond dissociation energy. Scaled to the current value of
3.7 x 103 ppnf1 min'1 for C^g + OH-, the rate constant for CH3CH2C(0)CH3
+ OH- •* abstraction products is about 2 x 103 ppm~^ min~^. However, an
analogy with propylene suggests that addition to the C-0 double bond may
be a faster reaction. Although repeated experiments were performed by UCR
(i.e., experiments with the same initial N0x/butane ratio), we chose only
one experiment for each of the different NO /butane ratios.
A
0-
I
CH3CH2C(0)CH3 -> CH3CH2C(OH)CH3 (32')
followed by
CH3CH2C(OH)CH
The net products are acetic acid and an ethyl peroxy radical. The lower
limit to k32' was taken to be 2 x 103 ppm'1 rnin'1, and 1 x 104 ppnT1 min"1
was used in the simulations.
The carbon balance (excluding CO and C02) of the butane data was usually
about 20 to 40 percent, which indicates undetected carbon-containing species.
Acetic acid may have been one of them.
b. Results
The initial concentrations and values of k] for the butane-NOx block
(Figure 24) are recorded in Table 8. Model predictions, along with UCR data,
appear in Figures 25 through 39. Generally speaking, the mechanism predicts
-------
71
an overall rate greatly in excess of that observed, the only exception being
the low butane run, EC-42. Excluding EC-42, the universal over-prediction of
1.0
0.
c
o
-------
NO
03
N32
.40
.30
o
U
<=
O
o
.20
.10 --
K
3.00
0-0
SQ.Q
Si Z
NO,
100.0
-Hi.
800-'Q .
Time (minutes)
FIGURE 25. EC-39 SIMULATION RESULTS AND UCR DATA FOR 03, NO, AND N0£
ro
-------
5-00 -T-
£3.00
o
aj -
o 2.00
o
1 -00
O.GG
* X W
* X X X
0-0 50-Q 100-0
150-Q 200-0 25U-Q
Time (minutes)
CO
FIGURE 26. EC-39 SIMULATION RESULTS AND UCR DATA FOR BUTANE
-------
X
KCfTRLCF.HYDf
.10 -T-
-OB
I -06
10
o
o
O.CQ
0-0
SU-Q
FORMALDEHYDE
ACETONE
x
1.00-Q
151). 0
COO - 0
Time (minutes)
300-0
FIGURE 27. EC-39 SIMULATION RESULTS AND UCR DATA FOR ACETALDEHYDE AND FORMALDEHYDE
-------
x
.10 T-
-08 •--
O
•r-1
«S
£ i. Of
O
O
.02 --
O.C!G J
o-o so.o IQQ:O iso-'a 200:0 250 :o .300:0 3?:o:c HCC.Q
Time (minutes)
FIGURE 28. EC-39 SIMULATION RESULTS AND UCR DATA FOR MEK AND PAN
-------
.60 -r-
.MB
A :-36
•»->
o
o
o
.24 ~-
.12
i-CD
3-t
CD 1)3
2 NG2
X NC
iSO-0 201? .'0 250 .'0
Time (minutes)
oro o
0
HOt
FIGURE 29. EC-41 SIMULATION RESULTS AND UCR DATA FOR 03, NO, AND N02
-------
S.OO -i-
4-00
S" 3-00
;2.oo
1.00 --
O.OD
so .'Q
15Q.Q ?OQ-Q 2SU-Q
Time (minutes)
433.0
FIGURE 30. EC-41 SIMULATION RESULTS AND UCR DATA FOR BUTANE
-------
.20
-IS
-------
5K FQRf1SL.DF.WDE
VD FiCcTRLCF.HYuF
.10 -r-
.08
t:-oe
(O
+»'.
C
o> •
o •
C'
o •
CJ •
.02 ~-
a.cc
FORMALDEHYDE
- ACETALDEHYDE
GO-Q
IOQ.Q
150-Q
EQQ.Q 2SU-Q
Time (minutes)
3QQ-Q
to
FIGURE 32. EC-41 SIMULATION RESULTS AND UCR DATA FOR ACETALDEHYDE AND FORMALDEHYDE
-------
* BUTHNf
X ND2
X NO
-6U -i-
.36
01
o
o
.24
.12 --
c.oo
BUTANE
NO
Q.O 50-0 100.Q 1SO-0
SQQ.Q 250.Q
Time (minutes)
3'JJ.O SDD-C
00
o
FIGURE 33. EC-42 SIMULATION RESULTS AND UCR DATA FOR BUTANE, NO, AND NO,
-------
.05 -T-
.04 •••
.03 --
o
o
.01
a.ca
o-u
X ME*
FORMALDEHYUh
MEK
ACETALDEHYUb
s ....
z
.-—-"ft"
SU-Q
100-0
150-0 COO-0
Time (minutes)
00
FIGURE 34. EC-42 SIMULATION RESULTS AND UCR DATA FOR MEK, FORMALDEHYDE, AND ACETALDEHYDE
-------
.010 T-
,008 --
£
&. .006
5
(O
o .004
o
o
03
.002 - -
CD
0.000
0.0
50-0
100-0
-f
150-0 200-0 250 .'3
Time (minutes)
3QO-0
35Q • 0
—.1 n
>. v.1 • v.r
CO
ro
FIGURE 35. EC-42 SIMULATION RESULTS AND UCR DATA FOR OZONE
-------
W NO
X N32
VD PCFTRLGEHYDf
I .GO -r-
O.GO
Q-0
:O.Q 103.0
3?:3-0 4.23-0
Time (minutes)
oo
oo
FIGURE 36. EC-44 SIMULATION RESULTS AND UCR DATA FOR NO, N02, AND ACETALDEHYDE
-------
.36 --
CJ tn
o La
.09 - -
O
250.0
300-0
Time (minutes)
00
FIGURE 37. EC-44 SIMULATION RESULTS AND UCR DATA FOR OZONE
-------
fc RUTflrtF.
s.on -i-
4.80
3.60
a
t-
S 2.MQ
o
1 .20
.00
o-o so.o IOQ.Q
200-0 <
Time (minutes)
FIGURE 38. EC-44 SIMULATION RESULTS AND UCR DATA FOR BUTANE
CD
01
-------
.16
o
•r~
•P
J_
•M
-------
87
formaldehyde shows that the chain processes whereby butane is broken down
into oxygenated products are improperly represented. Correction of this
fault must await more complete product measurements and improved kinetic
data. Because rate constants for reactions of most of the less-than-four-
carbon radicals were determined from propylene simulations, it is possible
that 04 radical reactions are inadequately represented, but this conclusion
cannot be drawn on the basis of the results alone.
Of course, the traditional scapegoats, uncertainties of surface and photo-
lysis reactions, can also be blamed for the disparity between model predictions
and experimental results. Although they probably exacerbate the problem, they
are not the sole culprits. The results of EC-44 are reproduced in Figures
40 through 43, with k-|Q and k-|] (NO + N02 + H20 -> 2HN02) reduced by a factor
of 10. The reaction is greatly delayed, and the agreement with the data
is improved. However, once the reaction begins, the net rate of NO oxidation
is obviously still too fast.
B. SIMULATIONS OF BATTELLE DATA
To clarify the interaction of S02 with hydrocarbon-NOx-air pollutants,
EPA has sponsored a series of smog chamber experiments performed at Battelle
Memorial Laboratories in Columbus, Ohio. We have received the results of five
of these experiments and report here on the kinetic simulation of those data.
1. Instrumentation
The Battelle simulations were carried out in a 17.8 cubic meter environ-
mental chamber. The surface-to-volume ratio of this chamber is 2.6 meters"1.
Ozone concentrations were measured using a chemiluminescent method; ethylene,
NO, and N02 were measured using an automated Saltzman method; S02 using a
Beckman 906 analyzer; and propylene using gas chromotography with a flame
ionization detector. $03 concentrations were inferred from aerosol size
distributions measured with a Thermo Systems electrical aerosol analyzer, by
assuming that equilibrium existed between sulfuric acid aerosol in the con-
densed and vapor phases., This method was tested by independent chemical
-------
X NO
X NOi!
1.50 -r
1.20
.90
o
£
o
o
.60 --
.30 --
Q.CO
Q-Q
5U-0
loo-o isu-'u 200:0 2so :o
Time (minutes)
300-0
FIQURE 40, EC-44 SIMULATION RESULTS, USING kln = 1.3 x 10"12 ppm"3 and k,, = 0.024 ppm"1 min"1,
FOR NO AND NO,
00
00
-------
.05 T-
O.CQ
25U-0
300-0
350-'0
40C-0
Time (minutes)
FIGURE 41. EC-44 SIMULATION RESULTS, USING k1Q = 1.3 x 10"12 ppm"3 min"1
AND k,, = 0.024 ppm"' min"1, FOR OZONE
CO
<£>
-------
6.CQ -r-
.80
BUTRNE
o. 3.60 - -
2.MD
c
o
1.20 - -
•0-00
0-0
50-Q 1.00-0
eou-o
Time (minutes)
300-Q
FIGURE 42. EC-44 SIMULATION RESULTS, USING Iqo = 1.3 x TO"12: ppnf3 min"1
AND k = 0.024 ppnf min", FOR BUTANE
«£»
O
-------
X
.IS -r
.12 --
.09 --
-M
c
OJ
u
o
o
.06 --
.03 --
o.co
0-Q
5U-0
100-0
15U:0 200:Q 25U-Q
Time (minutes)
300-0
MEK
FORMALDEHYDE
ACETALDEHYDE
350 0
'FIGURE 43. EC-44 SIMULATION RESULTS, USING Mo = 1.3 x 10'12 ppm'3 min'1 AND
. kn = 0.024 ppra-1 min-1, FOR MEK, ACETALDEHYDE, AND FORMALDEHYDE
-------
92
measurements of S03 and found to be fairly accurate. The N02 photodissocia-
tion constant, kd, was determined from photocell readings, the photocell
having been previously calibrated. The relation k] = 4/9 kd (Wu and N1ki,
1975) has been used in our model. Other instrument and chamber character-
istics are summarized by Table 9 (Miller 1975).
2. The Mechanism Used
The propylene mechanism used in the UCR simulations forms the core of
the mechanism used to simulate the Battelle data. The reactions listed in
Table 10 were added to this core. In the absence of adequate light spectrum
information, photolysis rates were computed from the unshifted ratios included
in Table 3. Additionally, a rate constant of 2 x 10'11 ppnr3 min'1 was
assigned to the heterogeneous reaction of NO, N02, and H20 (Reaction 10,
Table 1), 0.4 ppm'1 min"1 was assigned to the reverse Reaction (11), and
c -1 i \ / >
1 x 10' ppnr1 min'1 was assigned to the heterogeneous reaction of N205 and
H20 (Reaction 9). This procedure is in accord with the discussion of heter-
ogenous HNOX chemistry in Chapter II and was used solely to improve N0?
profile predictions.
A recent rate constant measurement of 9.0 x 103 ppm'1 min'1 (Cox, 1974;
Atkinson, 1975) for the reaction OH- + NO + HN02 was used in these simula-
tions (but in none of the others).*
3. Results and Discussion
The initial concentrations and value of k] for the propylene-NOx-S02
block are contained in Table 11. Two experiments contained no S02 (S-114
and S-115); three did contain S02 (S-107, S-110, and S-113). Of the three
containing S02, S-113 was performed with reduced light intensity. All
photolysis rates for model predictions were scaled down accordingly.
* These new rate constant determinations were called to our attention late
in the contract year Although their incorporation in the Battelle simulation
2p?L9?S thpSNOtS^eirTh-feCt Vhe UCR Sl™la«°ns ^s an undesiraE e
delay in the N02 peak. This result was expected because rate constant
consan
adjustments, made to give the best fit to UCR data, correspond to the lower
^SnS A* 10Tu?pnrT mirr1' The OH< + N0 Action is simply the reverb
of HNOz + hv. This suggests that a modification of heterogeneous HNOp
ijf uncertainty bounds) would compensate the delayed
-------
93
Table 9
ANALYTICAL AND CHAMBER CHARACTERISTICS OF
THE BATTELLE EXPERIMENTAL SETUP
(a) Analytical Characteristics
Analysis
°3
N02
NO
so2
C3H6
S03
Dew point
Temperature
(b)
Lag time Dark time
(sec) (sec)
< 4 < 4
•v 500 % 20
* 500 ^ 20
% 300
—
—
< 5
< 1
Chamber Characteristics
Criteria
N02 photodissociation
kd, 0.38 min
k,, 0.16 min"
Uncertainty
Factor
0.1
0.2
0.2
0.15
0.15
0.5
0.1
0.05
Uncertainty
Factor
0.15
0.15
03 half-life (may vary with conditions)
Dark, 6-8 hrs
Light, 3-4 hrs
Dilution (variable with
sampling)
0.25
0.25
0.1-0.25
S/V ratio, meters =2.6
Background conditions (slightly variable)
CH4 < 2 ppm
Nonmethane hydrocarbons < 0.2 ppm C
CO < 3 ppm
-------
Table 10
94
ADDITIONAL REACTIONS USED IN SIMULATING
THE S02-OXIDATION DATA FROM BATTELLE
Reaction
S02 H
S02 H
S02 H
S02 n
H HO; — *- so, + OH.
£ 0
f- CH0OA — »- SOo + CH-jO-
32 33
3 C- C. r\ ^ 3 £• £
OP
1- CH,CH9C(0)02 -=*- SO^ + CH.CH90; + C09
3 L. •• *3 0 ^ c c
so2 +
HS05
HS04
HS04
S00 +
CH3C(0)0'
OH- -X-
«- NO —*- NO
)3 + NO,
H02
N00
H90
HN0
H2S04
Rate Constant
(ppm min" )
1.3
1.5
1.5
1.5
1.5
14
2
9 x 10^
8 x 102
9 x 103
1 x 10
Immediate
Predictions and experimental data are illustrated in Figures 31 through
51. As shown in these figures, the inability of the mechanism to follow the
propylene data—in sharp contrast to the UCR propylene simulations—is somewhat
disconcerting. In the early stages, the data are followed quite closely, but
the mechanism shows gradually tapering decay asymptotically, whereas the measured
propylene concentration drops off rapidly to zero. The NO and S02 predictions
are fairly accurate. Since the latter is relatively unreactive, a sizeable
fraction of its disappearance is due solely to dilution. The dilution rate
was approximately 10 percent per hour. Thus, 65 percent of the S02 disappear-
ance in S-107 and S-110 and 87 percent in S-113 was due to dilution.
-------
i .GO V
Q.
Q.
C
o
id
u
C
o
o
.80
.60
.MO --
.2D --
0-00
Propylene
N02
S0
0-0
50-0
10Q-0
15U-U
200-0
25il.U
300-U
350
40G.U
Time (minutes)
FIGURE 44. S-107 SIMULATION RESULTS AND BATTELLE LABS
DATA FOR PROPYLENE, N02> AND S02
UD
cn
-------
.MO -i-
i.
D.
C
o
•f—
4J
fO
S-
Ol
o
C
o
o
.32
.16 --
.08 --
0.00
3
X NO
S03 Aerosol
5U-U
100.U
isu-i
200 o
Time (minutes)
250-0
300-U
350-D
H'JO-Q
iO
FIGURE 45. S-107 SIMULATION RESULTS AND BATTELLE LABS
DATA FOR 03, NO, AND S03 AEROSOL
-------
* Propylene
X N02
X S0
1.50 -r-
o
s_
o
c
o
.60
0.00
50-U lOO-U 15U-U 200-U 2SU-U 30Q-U 350-U HOQ-U
Time (minutes)
FIGURE 46. S-110 SIMULATION RESULTS AND BATTELLE LABS
DATA FOR PROPYLENE, N02, AND S02
vo
-------
.50 -r-
.40 -r
Q.
Q.
.30 --
O)
o
c
o
CJ
.20 --
.10 --
0.00
0-0
•o 03
X NO
S03 Aerosol
o o
o o o o o
O (DO
100-0 150-0 200-0 250-0 300-0 3CJO-U HuQ-0
Time (minutes)
FIGURE 47. S-110 SIMULATION RESULTS AND BATTELLE LABS
DATA FOR 03, NO, AND S03 AEROSOL
00
-------
1.00 -r-
-------
.32 --
.21* —
o>
o
c
o
.16 —
.08 --
0.00
0.
3
W NO
SO- Aerosol
0-0
o
5U-0 iOO-Q
ISd-U 200-U
Time (minutes)
25U-0
300-0 350-0 400-0
FIGURE 49 S -113 SIMULATION RESULTS AND BATTELLE LABS
'DATA FOR o3, NO, AND so3 AEROSOL
o
o
-------
1.00 -r-
.80
.60 --
-------
Q.
03
O
c
o
0.00
* Propylene
o o3
H NO
S NO,,
250-U
300- 0
350-U
400-Q
Time (minutes)
FIGURE 51. S-115 SIMULATION RESULTS AND BATTELLE LABS
DATA FOR PROPYLENE, 0^ NO, AND N02
(Thir, experiment w.r.
-------
103
The N02 predictions are good, but it is only fair to reiterate that
heterogeneous rate constants were "tuned" to this curve. It was impossible,
however, to fit all the data by adjusting these reactions. In S-107 and S-110
post-peak N02 predictions are high, whereas in S-113, S-114, and S-115 they
are low. 03 is always low, usually by at least a factor of 2. This could be
due simply to under-prediction of propylene oxidation (and consequently over-
prediction of NO), though such a large difference is hard to explain. S03
appears too late and in too great a quantity, being about a factor of 2
too high in both S-107 and S-110. With proper propylene oxidation and 03
production, the discrepancy in $03 yield would undoubtedly grow even worse,
but the time delay would be shortened. One possible source of error could be
an overestimate of the rate of reaction of N03 with S02. The rate constant
of 14 pprrf1 min'1 is, after all, Davis1 (1974) upper limit. Table 12 sum-
marizes the magnitude of the various modes of S02 oxidation as a function
of time. N03 does not account for more than 10 percent of the net oxidation
rate until after 3 hours. Then it accounts for one-fourth to one-third
of the total rate. Since the discrepancy between the data and predictions
appears in the form of an overshoot late in the reaction, N03 is a likely
culprit. Thus, a measurement of the N03 + 862 rate constant is needed.
Table 11
BATTELLE PROPYLENE-NO-S09 EXPERIMENTS
A t
Initial concentrations of Reactants
Run No.
S-107
S-110
S-113
S-114
S-115
NO
0.328
0.392
0.409
0.414
0.417
N02
0.113
0.099
0.099
0.095
0.108
so2
0.474
0.480
0.482
0
0
Propylene
1.03
1.10
0.96
0.95
0.97
kl
0.175
0.177
0.079
0.172
0.176
-------
104
Table }?.
RATCS OF SO? OXIDATION UY VAHK)!!', OXtDAtiTS
(From the- S-107 Simulation)
Tlmo
(min)
CO
120
180
240
300
360
Oxidant
OH-
ROj
RC(0)02
HOJ
N03
Total
OH-
R02
RC(0)02
H02
N03
Total
OH-
R02
RC(0)02
HOJ
N03
Total
OH-
R02
RC(0)02
HOJ
N03
Total
OH-
R02-
RC(0)02
H02
N03
Total
OH-
ROj,
RC(0}02
HOg
N03
Tottl
Rate
8.8 x 10-5
5.6 x 10-5
6.1 x lO-6
1.1 x 10"4
. 7.2 x 10-6
2.7 x lO'4
4.5 x ID'5
1.4 x 10'4
1.8 x 10-5
1.8 x 10'4
3.2 x 10-5
4.2 x lO"4
2.1 x 10-5
1.9 x ID'4
3.8 x 10-5
1.2 x 10'4
4.9 x ID'5
4.2 x 10'4
1.1 x ID'5
2.4 x ID"4
8.0 x 10'5
6.8 x ID'5
1.2 x ID'4
5.2 x 10-*
7.1 x ID"6
1.7 x 10"4
2.1 x ID'5
4.6 x 10-5
9.5 x ID'5
3.3 x lO"4
5.4 x 10-6
1.2 x 10'4
2.3 x ID'5
3.3 x 10'5
6.6 x ID"5
2.5 x 10-<
-------
105
Another possible source of the $03 discrepancy could be an unmeasured
buildup of products on walls. Hence, not all 863 would be contained in
gas phase aerosols. In the current mechanism, only the reaction
HSO^ + N02 ->• H2S04 + N0,3 (Reaction 56) was assumed to leave products on the
walls.
C. SIMULATIONS OF THE NAPCA TOLUENE-NOX DATA
The NAPCA data served as a base for our previous two years of mechanism
development (Hecht et al., 1974b; Hecht et al., 1973). The latter reference
describes the chamber and experimental techniques and discusses the data.
1. The Mechanism Used
The mechanism for toluene oxidation, described in Chapter II, is pre-
sented in Table 13. The inorganic reactions, not included there, are the
same as those in Table 6. The formaldehyde chemistry was also taken from
Table 6. The photolysis reactions were adjusted to kj = 0.266, and the fol-
lowing changes were made in heterogeneous HNOX chemistry:
kg = 1 x 10"3
k1Q = 3 x 10"11
k^ = 6 x 10"1
kg is now Jaffee and Ford's (1967) value, and k^g has been brought to within
a factor of 3 of Noeh et al.'s (1974) value for a metal surface.
2. Results and Discussion
The results of three simulations are displayed in Figures 52 through
57. The N02 data for EPA-272 (Figures 54 and 55) between 100 and 300 minutes
appear to be erroneous, probably as a result of instrument failure. The
Initial concentrations and values of k, are given in Table 14.
-------
106
Table 13
TOLUCNK OXIDATION ML'CHAIIISM
Rate Constant
B.yp.y.SP (PP"| min'')
CgH5CH3 + OH- -£- CgH5CH202 + H.,0 4.5.x 103
0,
CgH5CH3 + OH- —C C6H4(CH3)(OH) + H02 4.0 x 103
CgH4(CH3)(OH) + OH- _^c5H4(CH3)(0-){02) +'H20 3.0 x 10*
C6H5CH3 + OM> —^"C6H5OH + CM3°2 7'° x 10?
°2 /,
C,H,OH + OH -i- C,H,(0)0, t H,0 l.OxlO4
65 -652 .,
°2 a
gHgCHO + OH' -^ CgH5C(0}02 + H20 1.0 x 104
it
4
°2
C6H5CH20- -£+ C6H50^ + H2CO 1.5 x 10
CgHjCHgO- + 02 -. CgHjCHO + H02 0.4
CgHjCH^ + HO - CgHjCHgO- + NOg 1.0 x 103
0, ,
CgH4(CH3)(0)02 + HO -i- N02 + CgH3fCH3)(0)0 * HO^ 1.0 x 103
Me-quinone
0? 3
CgH5(0)0^ + HO — £- N02 + CgH4(0)0 + HO^ 1.0 x 10
Qulnone
C6H5°2 + K0 - N02 * C6
0,
CgH5C(0)0^ + NO — ^-N02 + CgHjO^ + C02 1.0 x 103
CgH5C{0)02 + N02 - C6H5C(0)02N02(PB2N) 3.0 x 102
°7 »
C6H5C(0)02N02 -i- CgH50^ + N03 + C02 3.0 X 10"3
HOj + C6H5C(0)Oj - CgH5C(0)02H 3.0 x 103
H02 + C6H5CH2°2 "* C6H5CH2°2H * °2 3.0 x 103
H02 + C6H5°2 " C6H5°2H + °2 3'° x 1()3
HOJ + C6H5(0)Cj -. C6H5(0)02H 3.0 x 103
H02 * WCH3W°2 - C6H4(OH3)(0)02H 3.0 x 103
-------
.50 -r
.'-10
1 -30
£
+*
t~
g 20
.12 --
C.OO
X ND
X NC?
•3 03
Time (minutes)
FIGURE 52. EPA-258 SIMULATION RESULTS AND NAPCA DATA FOR NO, N02, AND
-------
S.CG -T-
H.CO --
g, -3.CO --
ta
g 2.0C
o
o
1 -GO
O.J
0-0
150-0 200-Q
2SQ-Q
3GQ-0
Time (minutes)
FIGURE 53. EPA-258 SIMULATION RESULTS AND NAPCA DATA FOR TOLUENE
o
00
-------
• --Jv* **
.MO --
o. .3D
§ .20 - -
o
o
.Id
x
X NQ
Time (minutes)
FIGURE 54. EPA-272 SIMULATION RESULTS AND NAPCA DATA FOR NO, N02» AND 0.
-------
2.en -r-
£ ' 1.20 - -
2
•p
or
o
cr
o
o
fc nri
0-0
50-0
100-0 150-0 2DO-Q ;
Time (minutes)
3DO-0
FIGURE 55. EPA-272 SIMULATION RESULTS AND NAPCA DATA FOR TOLUENE
-------
£
-p
01
u
o
o
O.CD
X KC
X NC2
13 03
KSO -r-
1 -SO
.90
.60
.30
Time (minutes)
FIGURE 56. EPA-305 SIMULATION RESULTS AND NAPCA DATA FOR NO, N02> AND 03
-------
TOLUENf.
S.CO -r-
t.CD --
5. 3.CO --
IO
4->
C
o>
u
c
o
o
? O
C. -l_.L
1-CD - -
C -Ci
.
50-0
ICQ-Q
2DQ-Q 250-0
Time (minutes)
FIGURE 57. EPA-305 SIMULATION RESULTS AND NAPCA DATA FOR TOLUENE
ro
-------
113
Table 14
INITIAL CONCENTRATIONS AND VALUES OF kn FOR
THE NAPCA TOLUENE-NOX EXPERIMENTS
EPA Run No.
258
272
305
NO
0.33
0.30
1.36
N02
0.04
0.04
0.08
Toluene
2.88
1.10
3.14
kl
0.266
0.266
0.266
The current mechanism cannot match the observed toluene oxidation rate,
and, as a result, it under-predicts ozone yields. The hypothetical chain
processes for ring oxidation to quinones and methyl group oxidation to benzalde-
hyde and PBzM are apparently too short. Product analyses currently being
performed at UCR may help to clarify the true mechanism of toluene oxidation.
-------
114
IV HYDROCARBON REACTIVITY
A. SURVEY OF REACTIVITY MEASURES
The concept of "reactivity" is familiar to every chemist, yet a pre-
cise definition of the term can often be very difficult. While one might
intuitively feel that a given species is more reactive than another, quan-
tification of reactivity requires careful definition of (1) the physical
conditions of the system in which reactivity is being determined, (2) the
concentrations of the reactants, and (3) the time scale of the reactions.
If we restrict consideration to the photochemical reactions occurring in
smog chamber experiments, the apparent reactivity of hydrocarbons will de-
pend on the physical characteristics of the chamber system, e.g., light in-
tensity, temperature, and surface-to-volume ratio. The rate of disappearance
of the hydrocarbon (or the appearance rate of products) will also be a func-
tion of the concentrations of (1) the hydrocarbon, (2) the oxidants of the
hydrocarbon, and (3) the other species that either react with or lead to the
formation of the oxidants. And, since the chemical state of the system
changes continuously, a measure of reactivity based on instantaneous rates
of formation or disappearance of chemical species will consequently be a
function of time.
We summarize below the criteria that have been used to quantify reac-
tivity. In considering these measures, we assume that the physical state
of the system, the initial concentrations of reactants, and the time scales
are all suitably defined and controlled, so that a meaningful comparison of
the reactivity values for different hydrocarbons can be made. The measures
fall into three classifications based on the physical dependence of the index:
temporal, concentration, or combined temporal and concentration criteria.
-------
115
1. Temporal Measures
Many indices have been constructed to characterize reactivity in terms
of the temporal occurrence of chemical or associated physiological events
during the smog formation process.
The most common of these include:
> The time of the N0£ peak (Tmax) (Altshuller and Cohen, 1963).
> Tmax for oxidant or ozone (Altshuller and Bufalini, 1971).
> The time required for one-half or one-quarter of the initial
hydrocarbon to be oxidized (Altshuller and Bufalini, 1971).
> The threshold time for eye irritation (Heuss and Glasson, 1968).
Other criteria that might be considered are the threshold times for "harmful"
effects measurable in terms of biological indicators (Feldstein, 1974). How-
ever, such effects—for example, the onset of eye irritation—are difficult
to quantify.
2. Concentration Measures
The intensity of smog formation is often assessed in terms of the con-
centrations of major primary and secondary pollutants. As a result, some
investigators have chosen to base hydrocarbon reactivity measures on the maxi-
mum concentrations of products that ultimately form. Their criteria include:
> Maximum (or asymptotic) oxidant (Heuss and Glasson, 1968;
Altshuller and Bufalini, 1971; Dimitriades and Wesson, 1972),
> Maximum eye irritation*—a function of the concentrations of
lachrymators and other irritants (Heuss and Glasson, 1968;
Altshuller and Bufalini, 1971).
* Yeung and Phillips (1973) have attempted to relate reactivity to eye
Irritation through the use of a "biological effect factor":
Relative Chrtc.1 Reactivity -
-------
116
Because air pollution really consists of all the products listed above and
more, a reactivity measure based on the concentrations of smog components
might best be expressed in the form
R = E Vi
where
R = reactivity measure,
i = an index of all harmful components in the system,
c-j = the maximum concentration of the itn species,
ai = a species weighting factor relating the toxicity (or
other harmful effects) of Cj.
3. Combined Temporal and Concentration Measures
Investigators have also characterized reactivity using measures that
depend on both concentration and time. The most common of these criteria
are dosage (ppm-min), rate (ppm min-1), rate constants (ppnr1 min'1). and
percentage of hydrocarbon or NOX consumed at a fixed time (dimensionless).
a. Dosage
Dosage is defined as
A
J c. dt
AHshuller et al. (1970) and Dimitriades and Wesson (1972) have used N02,
oxidant, PAN, and formaldehyde dosages as measures of hydrocarbon reactivity,
The choice of tf, the ending period of the integration, is important. One
might wish to use the characteristic residence time of pollutants in major
-------
117
air basins, the time to the N0£ peak, or the time to the oxidant peak. The
beginning time of the integration, tg, is usually the initiation of irradia-
tion, but some later time might also be chosen. In the case of oxidant dosage,
for instance, it might be more practical to take tg as the time of the N02
peak. Consideration of maximum one-hour dosages (e.g., tf - tg = 60 minutes)
might also be appropriate in view of existing federal air quality standards.
b. Rate
The rates of the chemical transformations in polluted air change con-
tinuously, and both instantaneous and average rates have been used as mea-
sures of hydrocarbon reactivity. The instantaneous rate, (dci/dt)^^, has
been considered by Altshuller and Cohen (1963), using N02 as c-j and
tj = t(i/2 N02 max) on botn tne ascending and descending portions of the
N02 versus time profile. One could also evaluate the slopes at tj = 0 or
at the time of maximum slope. Average formation rates for N02 can be de-
fined as l/2(NO)o/t(l/2 NO conversion) or as (N02)m,v/tN(v> ma><- The former
11 Id A £
has been taken as a reactivity measure by Heuss and Glasson (1968) and
Glasson and Tuesday (1970). Other average rate measures using oxidant
rather than N02 have been reported by Heuss and Glasson (1968) and Altshuller
and Bufalini (1971).
c. Rate Constants
Hydrocarbons in the atmosphere are oxidized principally through reac-
tions with 0, OH-, and 03. Of these three species, OH- is thought to be
the most important oxidant of all classes of hydrocarbons, and the rate con-
stant for the OH-hydrocarbon reaction has been used by Niki et al. (1972) to
characterize reactivity. They found that kg^ correlated much better with
the reactivity measures of Glasson and Tuesday (1970) and Altshuller and
Cohen (1963) than did kg or kQ3» the respective rate constants for the
0 and 63 oxidation reactions of hydrocarbons.
-------
118
d. Percent Hydrocarbon Oxidation
A final measure of reactivity is the percentage of initial hydrocarbon
oxidized at a fixed time (Altshuller and Bufalini, 1971). As in the case
of dosage, one might consider a time span equivalent to the residence time
of an air mass in a polluted air basin, the time of the N02 peak, the time
of the ozone peak, or some other relevant period.
B. MEASURE ASSESSMENT
Since an evaluation of all of the measures listed above would be imprac-
tical, this study has been limited to a group representative of the simplest
and most practical of those proposed. The evaluation process is described
in Section 2; the final selections are reported in Section 3.
A primary application of reactivity measures is the prediction of the
smog formation potential of mixtures of hydrocarbons emitted as automobile
exhaust and solvent fumes. Predictive ability is essential to effective
control-strategy planning and evaluation (Dimitriades, 1973). As part of
the present study, several smog simulations using mixtures of olefins and
NOX in air were performed to determine whether mixture reactivity can be
predicted for these simple cases. The results of the mixture study appear
in Section 4. A semi-theoretical justification for some experimental ob-
servations made during this study is presented in Section 5. The section
Immediately following describes experimental methods.
1. Scope and Procedure
a. Mathematical Simulation
The data used in this study were generated by the mathematical »del
for Smog station presented in Table 8 of the Second Annual Report (Hecht
« .1., 197*). The use of nu^rical rather than physical expends i,
unusual, but it offers definite advantages. Initial concentrations can be
-------
119
specified precisely; ambiguities due to chamber effects are absent; the
concentrations of all species (including free radicals) are known at any
given time; "experiments" are quick (a few seconds of computation), easy,
and inexpensive to carry out; and instrumentation is not needed to measure
species concentrations.
The principal drawback to modeling Is Its possible inaccuracy in rep-
resenting reality. For the present purpose, which is the theoretical eval-
uation of reactivity measures, it is not mandatory that the kinetic mec an-
ism be absolutely accurate, since all comparisons of measures are made be-
tween coated results. If the mechanism were infallib e, reacts uld
be evaluated directly by simulation. As an example of d,rect s a ,
consider an industrial process that normally results ,n atmo pher
trations of 1 ppm propylene, 0.2 pom NO, and 0.02 ppm 02. Supp , d
tion, that at a slightly higher cost, about one-half th,s pro * e
oxidized to form a!dehyde (and C02). Kopczynsk, et .1 (1974) ™
and Wesson (1972) showed that aldehydes can be as reacts as e , o he
effect of this partial oxidation is not, a priori obvious e tility •
this proposed approach to hydrocarbon emissions reduct,on wo 1 d th e
be of co cern to control strategists. Figure 58 demonstra r
Simulation can be used. As this figure shows, in th,s s,m e ^
production of smog constituents (typified here by
easily and the cost-effectiveness of the proposed s rate y t hus
Unfortunately, obtaining a perfected mechanism for he t r tr,
emitted pollutants would retire an overwhelming - ^ " 'of ....
present study requires only that the mo , ^ s , / /^l to be
sure comparison. The model has been tet 6 ^^ smg chamber data
capable of providing reasonably accurate pred ctions
(see the Second Annual Report, Hecht et al., 1974D).
b. Purpose
nt of factors pertinent to reactivity determination can
The assessment of factors pe efficiency of data
help m the planning of laboratory expends, improve
-------
1.00
-------
121
collection, and provide heuristic guidelines for impact evaluation. Toward
this end, the kinetic mechanism provides a basis for conclusions, strengthens
arguments, and tests hypotheses. Although the mechanism has been validated
for a restricted number of hydrocarbons, by using it as a "laboratory" we
have obtained results that hopefully have broad significance.
The purpose of this study was not to provide absolute quantification
of hydrocarbon reactivity. Such quantification, if it is possible, must be
obtained from well-controlled laboratory experiments. Extensive tabulations
of laboratory results can be found in Heuss and Glasson (1968), Glasson and
Tuesday (1970), and an MSA Research Corporation report (1972).
c. Procedure
By inputting desired initial conditions and integrating the appropriate
kinetic equations, we obtained concentration-versus-time curves for a given
hydrocarbon-NOx-air system. It was then necessary to select the relevant
points, such as peak concentrations or times of peaks, from the computer
output to determine values for the various reactivity measures. More de-
tails on what these points were and how they were used are given in subse-
quent sections.
2. Measure Study
a. Criteria for the Evaluation of Measures
While criteria for a good measure are basically intuitive, it is worth
mentioning a few of them here. Ideally, a measure of a given hydrocarbon's
reactivity would be independent of initial reactant concentrations. Unfor-
tunately, because of the complexity of smog systems, such independence is
not realizable. However, some measures will show less variability than
others. If the inevitable variations show a consistent trend, it may be
possible to specify their functional dependence and, hence, to develop a
very useful predictive ability. Within this report, a measure that shows a
-------
122
small predictable trend is termed "self-consistent." Aside from being self-
consistent, satisfactory measures must also be consistent with each other
and not widely disparate with accepted reactivity values.
Three pragmatic requirements we have imposed are that the measure be
(1) directly related to the production of harmful smog components,
(2) clearly defined, and (3) easily and accurately measurable. Without
these properties, the applicability of the measure would be severely
limited. Other criteria will be developed as needed.
b. Normalization
The results from a study of the measures defined in Table 15 are tabu-
lated in Table 16, and the initial conditions for the experiments reported
are shown in Table 17. The entries in Table 16 were normalized by the reac-
tivity of propylene; i.e., they are in units of propylene equivalents. P.ro-
pylene simulations were carried out at initial NO, N02> and hydrocarbon con-
centrations corresponding to each row, and the times, concentrations, and
rates corresponding to each column were determined. These were then used to
normalize values obtained for other olefins. Since the rate constants are in-
versely proportional to time, for the time scales, an inverse ratio was used,
in which the relative reactivity of a given hydrocarbon (HC) is given by the
ratio Tpropylene^HC' wfiere T is t^ie appropriate time scale. For other measures,
a direct ratio was used.
c. The Elimination Process
The data in Table 16 do not provide a sufficient basis for choosing
the best measure of reactivity without additional considerations. But an
inspection of these data does permit a rapid elimination of three of the
criteria. The % HCt=ioo obviously fails the self-consistency test because
it produces a wide range of values at various initial concentrations.
While showing self-consistency, N02(max) and 03(max) are quite inconsistent
with other measures. In fact, they are so close to unity that, in light of
model inaccuracy, the differences in reactivity between the various olefins
-------
123
Table 15
DEFINITIONS OF REACTIVITY MEASURES*
Reactivity Measure
HCt=100
03 (max)
Scaling
N02 rate
conversion
Definition
conversion (°r
100 x HC(t=loO min)/HCo
Peak concentration of N02
Peak 03 concentration or asymptotic
concentration
Reciprocal of the HC concentration
required to obtain a Tj^fmax) equal
to that of 1 ppm C^\s
N02(max)/TN02(max)
Time for the N02 concentration to
reach the value [N0210 + 1/2[NO]0
Time to the N02 peak
Time when [HC] = 0.75[HC]0
* Reactivities relative to C3Hs are given by the reciprocal ratio of
time scales and direct ratio of all other measures.
-------
124
Table 16 •
RESULTS OF THE MEASURE STUDY: REACTIVITIES* RELATIVE TO PROPYLENE
Experiment*
1
2
3
4
5
6
7
8
9
10
11
12
G&TS
A&B**
ASCtt
% HC10Q
0.12
0.10
0.16
0.27
0.29
0.18
0.06
0.17
N02(max)
--
0.91
0.93
0.95
0.96
0.92
0.91
0.84
Q
3(max) Scaling
< 0.25
0.25
0.89
0.90
0.90
0.90
—
0.27
1.16
1.12
1.10
1.09
N02 rate
—
0.23
0.23
0.28
0.28
0.28
0.27
0.23
(H NO conversion)
—
0.25
0.22
0.20
0.21
0.25
0.29
0.24
0.49
N02(max)
—
0.25
0.25
0.29
0.29
0.30
0.30
0.27
1.45
3.27
1.83
3.98
0.36
0.23,0.31
\ HC
* 0.2
0.23
0.23
0.24
0.26
0.24
0.24
0.26
1.68
3.84
2.11
5.23
0.48
The reactivity measures are defined in Table 15.
** Altshuller and Bufalini (1970) values for ethylene.
t The Initial concentrations are given in Table 16.
tt Altshuller and Cohen (1953) values for ethylene.
5 Glasson and Tuesday (1971) values for ethylene.
-------
125
Table 17
INITIAL CONCENTRATIONS FOR EXPERIMENTS LISTED
IN TABLES 16 AND 18
Experiment
1
2+
3
4
5
6
7
8
9+
10+
11 +
12+
13+
14
15
16+
17
[NO]Q
0.4
0.4
0.4
0.4
0.4
0.2
0.48
0.5
0.4
0.4
0.4
0.4
0.4
0.3
0.48
0.4
0.4
[NOJ
* 0
0.1
0.1
0.1
0.1
0.1
0.05
0.02
0.1
0.1
0.1
0.1
0.1
0.1
0.075
0.02
0.1
0.1
[HC]Q
0.5
1.0
2.0
3.0
4.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
0.6
Initial HC Composition
HC5*
HC5
HC5
HC5
HC5
HC5
HC5
HC5
HC1
HC,
£.
HC3
HC4
0.2 each HC] , HC2, HC3> HC4, and HCg
0.2 each HC] , HC2, HC3, HC4> and HCg
0.25 each HC2, HC4, HCg, and HCg
0.25 each HC2, HC4> HCg, and HCg
0.15 each HC0, HC., HCC, and HCC
£. <\ 0 D
* In this study H^ s 1-butene; HC2 = cis-2-butene; HC3 = 2-me-l-butene;
HC4 s 2-me-2-butene; HCs = ethylene; HC6 = propylene.
t These initial conditions are defined as "standard." Experiments 2 and
9 through 13 provided data for the mixture study.
-------
126
are insignificant when they are based on either W^fmax) or 03(max). The
investigators who successfully used 03(max) probably defined it either as
a peak or, when no peak occurred, as the concentration of 63 at the end of
a time-limited experiment; they are usually not clear on this point. The
entries omitted from Table 16 correspond to experiments in which 03 had
reached neither a peak nor an asymptotic value by the end of 400 minutes.
In these omitted cases, the final value indicated a reactivity significantly
less than one. Thus, the definition of 03(max) varies, depending on whether
a peak is reached. The values of reactivity obtained will thus depend on the
length of the experiment; they are neither clearly defined nor consistent.
We surmise that 03(max) and N02(max) are insufficient measures.
As shown in Section 5, scaling can be demonstrated to be equivalent
to TUJ. Their theoretical equivalence is borne out by experimental evidence,
as shown in Table 16.
Relative reactivity based on the N02 rate is defined as NOaCmax) times
TJJ, (it is multiplicative because of the inverse normalization of time scales)
Since the N02(max) values are all close to unity, the difference between Tm
and the N0£ rate should be small. The added complexity in rate determina-
tion hardly seems worthwhile. Because it is a "combined spatio-temporal"
measure, the error in determination of the N02 rate is the sum of errors of
its component parts. Clearly, if one of these components is a good measure,
accuracy as well as simplicity can be gained by using it alone. A last, and
possibly undesirable, property of the N02 rate is that it is a difference
approximation to the rate of N02 formation. Therefore, it does not repre-
sent the actual rate at any point on the NOg curve; it is instead an average
value.
conversion ">s a widely used measure, usually appearing in the
guise of "NO photooxidation rate," defined as NOo/2T^ (c.f. , Glasson and
Tuesday, 1970). Since, in the present study, relative reactivities were
computed at a given NOg, this column in Table 16 could just as easily have
been labeled "R(NO photooxidation)"- Interpreted as such, \ also has the
possible shortcoming of being a difference approximation, although it is a
-------
127
very good one when the induction period is negligibly short. Ti^NO conversion*
as defined in Table 15, is not the half-time for N02 formation, rather it
falls somewhere between the half-time and peak time of N02- Its location
relative to the time of product formation is therefore ill defined. Thus,
the most objectionable quality of \ is its lack of direct correlation with
03 production or other harmful smog constituents. In the application of
reactivity criteria to pollution control, this is indeed a serious shortcoming.
TN02(max) and Tj,HC conversion' the on^ measures left, are the measures
we have selected for use. Because of their importance to the present study,
they are discussed in detail in Section 3.
The results of other investigations are also included in Table 17 for
the purpose of indicating the need to eventually combine the work contained
in this report with laboratory investigations. Meaningful comparisons of
our values with those of the other investigators cannot be made now because
experimental conditions generally differ considerably.
3. The Measures Selected
a. Practicality
The measures found to be most conducive to the quantification of smog
effects are the time of the N02 peak, TNo2(max) (hereafter denoted as Tm) ,
and the time required for one-quarter of the initial hydrocarbon to be oxi-
dized (Tjj). In the terminology introduced in Section A, these are temporal
measures. By being one dimensional, they avoid the increased measurement
error inherent in the combined concentration and temporal measures. Both
are simply and clearly defined. Although the N02 peak may not be sharply
resolved in practice, interpolation methods along with extremum theory pro-
vide for its accurate determination. A parabolic curve fit can be used for
this purpose.
-------
128
As shown by the results of Experiments 1 through 5 in Table 16, the
relative reactivity of ethylene based on Tm (hereafter abbreviated RR-j-J
varies by about 15 percent as the initial hydrocarbon concentration (HCg)
is changed by a factor of 4. There is a visible trend toward increasing
RRj with increasing HCg. Although any variability is undesirable, the fact
that it is small indicates that the RRTm, measured as a single HCQ, may be
applicable throughout a wide range of HCo's (at fixed N00 and N02Q). !t may
even be possible to capitalize on the consistency of the trend to estimate
the accuracy of a constant value in this range.
At fixed HCQ, increasing N0o/N020 by a factor of 6 caused an increase
in RRj of almost 20 percent (see Experiments 2 and 7). In Section 5, this
behavior is shown to be attributable to induction period effects. Because
of its role in determining RRjm, further study of the induction period would
be useful.
As shown by the data in Table 17, RRT% exhibits the same trend with
increasing HCg as that observed for RRTm. But variability with N0o/N020
is almost absent. An increase in this ratio by a factor of 6 caused vir-
tually no change in RRTj,. The slightly erratic behavior shown in Experi-
ment 8 is most probably caused by inaccuracy in this experiment. (Unfortu-
nately, the need to incorporate an interpolation scheme did not become ap-
parent until the late stages of this study.) \ is therefore partially in-
consistent with Tm. The apparent absence of induction period effects on
RR-|> is interesting and deserves further investigation. It is probably due
to the smaller initial slope of the hydrocarbon curve.
b. Usefulness
Aside from considerations of simplicity and consistency, a useful
measure must ultimately be related to objectionable pollution effects.
Dimitriades et al. (1970) have discussed this issue and concluded that,
while no one index is fully satisfactory, "...there is evidence that the
over-all [sic] level of activity in the photosimulated hydrocarbon/NOx
-------
129
system is reflected in the pattern of N02 formation." It is well known
that the N0£ peak is correlated with the formation of 03 and PAN, two haz-
ardous components of smog. In any "time-limited" system, such as an urban
airshed with a characteristic residence time, the amount of 03 present due
to chemical reaction is directly related to Tm. Aldehydes also contribute
to the deleterious effects of smog. Because aldehyde appearance is comple-
mentary to hydrocarbon disappearance, 1% is an indication of their importance
in a time-limited system. T^ is also useful for evaluating the magnitude
of synergistic effects in mixture reactions. Altshuller and Bufalini (1971)
define a synergistic effect as "...one in which the reactivity or the amount
of product produced by a given compound is affected by the presence of a
second." Since, in general, the oxidation of several hydrocarbons in a par-
ticular mixture will lead to the same or similar products, synergism is not
easily determined from product measures. A much simpler means, particu-
larly when using a numerical model, is to monitor the rate of hydrocarbon
disappearance. For a given hydrocarbon, the change in TV from its value
in an individual hydrocarbon-NOx reaction system to that in a multihydrocarbon-
NOX reaction system provides an indication of interactive effects.
c. Measurability
Because laboratory techniques must ultimately be used either to mea-
sure reactivity or to obtain empirical constants needed for its prediction,
a satisfactory measure must have the additional property that accurate and
reliable instrumentation be available for its determination. N0£ and hydro-
carbon concentrations are routinely monitored in smog chamber experiments
with reasonable accuracy. Hydrocarbon concentration can be measured with
as little as 1 percent error by gas chromatographic methods. In contrast,
N02 is obtained from the difference between NOX (after conversion to NO) and
NO concentrations. These are measured by chemiluminesence with about 95
percent accuracy. When N02 concentrations are low, the percentage error in
the difference of these values may be large. When [NOel is at its peak and,
consequently, [NO] is low, the error in the difference will be at a minimum.
The accuracy in [N02] at the peak should therefore be close to 95 percent.
-------
130
Of course, only the temporal location (Tm), not the peak value itself is
used. The cycling time for NO measurement is only a minute or two; so a
high density of points and, hence, sharp resolution of Tm can be obtained.
With reasonably accurate values of concentration and the time of measurement
of each point well known, an accurate determination of Tm should be possible.
4. Mixture Study
a. Mixtures Used
Reactivities were computed for five different olefin-NOx mixtures at.
the initial concentrations indicated in Table 16. Experimental and pre-
dicted results are shown in Table 18. The reactivities are all relative to
that of propylene at standard initial conditions (N0o=0.4, N020=0.1, HCo=1.0).
The olefin mixture used in Experiments 13 and 14 is composed of the five most
reactive olefins studied and is a typical "highly reactive" mixture. A
second mixture, used in Experiments 15 through 17, contains the two most
reactive and the two least reactive olefins and is characteristic of a wide
reactivity range mixture.
b. Results
The first row of entries for each experiment in Table 18 is labeled
"mixture." In Columns 3 and 5 (Tm and Tj, measured) of this row, relative
reactivities of the olefin mixtures are measured. Tm, as before, is the
time of the NOg peak. Tj. is now the time required for the total olefin con-
centration to drop by 25 percent. Below the \ entry are the time for each
component olefin to reach 75 percent of its initial concentration. As
before, all results are normalized by propylene experiments at the given
initial concentrations.
c. Predictions
There are also three prediction columns in Table 18. These are pre-
dictions of mixture reactivity computed by the "linear summation" method,
defined by:
-------
131
Table 18
RESULTS OF THE MIXTURE STUDY: MIXTURE REACTIVITIES RELATIVE TO PROPYLENE
Tm
Experiment Hydrocarbons Simulated LSM
13 Mixture 2.42 2.31
HC,*
HC2
HC3
HC4
HC6
14 Mixture 2.64
HC,
HC2
HC3
.HC4
HC6
15 Mixture 1.54
HC2
HC4
HC5
HCfi
16 Mixture 2.16 2.13
HC2
HC4.
HC5
HC6
17 Mixture 1.26 1.28
HC2
HC4
HC5
"6
LSM
Uneorrected
Simulated Prediction
2.86 2.77
2.30
3.42
2.71
4.27
1.37
2.71 -
2.15
3.39
2.53
4.44
1.21
1.94
2.33
2.83
0.44
1.20
2.58 2.56
3.58
4.31
0.47
2.58
1.98 1.54
2.52
3.15
0.44
1.20
\
Simulated
Reactivity of
Hydrocarbons
In Mixture
1.68
3.84
2.11
5.23
1.0
1.68
3.84
2.11
5.23
1.0
3.84
5.23
0.23
1.0
3.84
5.23
0.23
1.0
3.84
5.23
0.23
1.0
Prediction
Corrected for
Synerqlsm
2.81
2.76
1.70
2.73
1.83
• In this study. HC] i 1-butene; HC2 i cit-2-butene; HC3 i 2-oe-l-butene; HC4 a 2-me-2-butene; HC5 i ethylene;
HC( i propylene.
-------
132
RLS ' E C1R1
where
RLS = mixture reactivity by linear summation,
n = number of organic mixture components,
cj = initial concentration of the ith hydrocarbon,
R-j = reactivity of the ith hydrocarbon.
The linear summation technique has been discussed by Glasson and Tuesday
(1971) and Dimitriades et al. (1970).
The values of R-j, based on either Tm or T^, were obtained from the
experiments in Table 16 as follows: R-j comes from Experiment 9, R2 from
Experiment 10, R$ from Experiment 11, R4 from Experiment 12, R$ from Experi-
ment 2, HC5 is the reference olefin (propylene) with a reactivity defined
to be 1. (Note, from the definitions of HCs 1 through 6 in the footnote to
Table 17, that R-j increases with increased substitution at the double bond.
Column 4 of Table 18 gives the mixture reactivity based on Tm as com-
puted by the linear summation method. The values shown agree very well
with observations. Experiments 14 and 15 were done under nonstandard ini-
tial NO and N02 values, and, therefore, no predictions could be made.
Although the initial olefin concentration in Experiment 17 was not the
standard value of 1 ppm, prediction could be made by "scaling." This pro-
cedure consists simply of multiplying RLS by the ratio of the initial
hydrocarbon concentration to its standard value (in this case the RRLS of
Experiment 16 times 0.6). Hence, the slower rate at lower hydrocarbon con-
centrations is compensated for multiplicatively. Justification for the
application of the scaling technique to RRTm is given in Section 5. Its
application to RRj^ is not really justifiable because the approximately
exponential hydrocarbon decay rate indicates a nonlinear dependence of
on HCg. This explains the low value listed.
-------
133
d. Synergism
Columns 6 and 7 of Table 18 show both the linearly predicted reactivity
of the mixtures and the standard reactivities of their components. The lat-
ter were copied from Table 16. Comparison of the standard reactivities in
Column 7 with the observed values in Column 5 indicates the presence of
synergistic effects. It is evident that in all cases the two most reactive
olefins, HC2 and HC4, experienced a decrease in relative reactivity, whereas
the less reactive olefins experienced an increase. This behavior can be ex-
plained qualitatively in the following manner. The competition for available
oxidant between higher and lower reactivity olefins depressed the reaction
rate of the former. At the same time, the high rate of oxidant production
by HC2 and HC4 accelerated the consumption of less reactive olefins.
Linear summation was also applied to the synergistically modified values
of T^J (Tjj measured). The resulting mixture reactivities are listed in the
last column of Table 18. Although these predictions are close to observed
values, we cannot state that they are always an improvement over the unmodi-
fied predictions.
Another inconsistency between Tm and Tjj is contained in Table 18.
Whereas decreasing NOXQ from Experiment 13 to the value used in Experiment
14 increased RRr , it decreased RRT. . Many explanations of this behavior
'HI. "Z
can be offered, including the changed 03 production and the radical-
scavenging ability of N02. It is the complex interaction of all these ef-
fects that leads to the discrepant behavior.
5. Derivation of Some Properties of Tm
a. Derivation
In this section, the analytical solution for the dependency of Tm on
Initial hydrocarbon and NO concentrations is obtained through a consideration
of simplistic photochemical smog kinetics. Kinetic equations and empiricism
based on the observed shape of smog profiles are used toward this end.
-------
134
O'Brien (1974) has demonstrated that, except in the late stages of the
photolytic hydrocarbon-NOx-air reaction, the concentrations of NO, 03, and
N02 are related by the approximation
k,
[N02] *f- [N0][03] , (IV-D
where k-| and k3 are rate constants for the reactions
i
N02 + hv —!*- NO + 0 (1)
k-
NO + 03 —i- N02 + 02 • . (3)
Existence of the photostationary state, expressed by Eq. (IV-1), in smog
profiles computed using the Hecht-Seinfeld-Dodge kinetic model has been
demonstrated by Liu (1974).
In addition to Reaction (3), the conversion of NO to NOg is accomplished
through the reaction
R0£ + NO —^RO- + N02 , • (42)
where R is usually an alkyl group or hydrogen atom. Hence, in the period
before the N02 peak, the rate of NO production and consumption is governed
by Reactions (1), (3), and (42). Thus,
j - k3[NO][03] - k42[R02][NO]
Equation (IV-1) can be used to simplify Eq. (IV-2):
-------
135
This expression represents the perturbation to photos tationarity introduced
by free radical oxidation.
In a completely generalized mechanism, ROg includes all free radical
oxidation products of hydrocarbons. It is formed in the following reaction
where Ox represents a general oxidant (primarily OH- in the time prior to
the N02 peak), HC-j a hydrocarbon, and ct^ an appropriate stoichiometric co-
efficient (see Hecht and Seinfeld, 1972).
Using Reactions (IV-4), (42), and the steady-state assumption for R02
leads to the following equation:
O x]
x
*
Substituting for RO^ in Eq. (IV-3) and using Reaction (IV-4) gives
d[HC.]
The rate terms in Eq. (IV-5) can be used to relate Tm to HCg and NOg
through the use of empirical observation. Figure 59 is an illustration of
a typical, through idealized smog profile at initial NO and N02 concentra-
tions for which there is a very short induction period. It is apparent
that the curves labeled NO and hydrocarbon are approximately linear during
the early stages of the reaction. Good approximations in this linear
period are
* Liu (1974) has demonstrated that this assumption becomes valid within a
few seconds of reaction initiation. The neglect of termination reactions
1n the steady-state expression is valid early in the reaction.
-------
136
Hydrocarbon
50
100
Minutes
150
FIGURE 59. TYPICAL SMOG PROFILE
-------
137
m
and
[HC.] . [HC.] (1 - b.t) , (iv-7)
where b^ is a constant related to the slope;
The introduction of HCQ. as a multiplicative factor can be justified
by a direct integration of the hydrocarbon rate equation derived from
Reaction (IV-4):
[HC.] = [HC.] exp |"-/k0 [Ox] dtj * [HC.] /I - / kQ [Oxl dt) ; t < t,
U 1» X. -J U > X« /
(IV-8)
Comparison of Eq. (IV-8) with Eq. (IV-7) shows that bj can be related to
the average oxidant concentration,
but such an interpretation is not essential to this derivation. There is
no reason to believe that ^ will be independent of N00, N02o, or HCi
Furthermore, the evidence indicates that bl is proportional to light inten-
sity (Niki et al., 1972; Glasson and Tuesday, 1970). Other factors not
considered may also influence ty. However, the dependence on N0xn and HC0
1s assumed to be negligible.
Substituting Eqs. (iv-6) and (IV-7) in Eq. (iv-5) and rearranging
gives
-------
138
[N0]fl
Tm a b' [HC.l ; bi ~= aibi •
1 0
b. Verification and Application
At a fixed initial NO concentration, Eq. (IV-10) states that the time to
the N02 peak should be inversely proportional to the initial hydrocarbon con-
centration. In Figure 60, the observed values of Tm are plotted as a func-
tion of HC-JQ for HCi = propylene at one NOg value and HCi = ethylene at two
values of NOg. All curves can be fit by the form a/HC0. From Eq. (IV-10),
a = N00/bj. For propylene, the curve fit has been drawn in, whereas for the
two ethylene curves it has only been indicated. When the two values of a
for ethylene are divided by the corresponding values of NOg, the results are:
a/NOo = 500 and a/NOn. = 496. Therefore, b\ for ethylene is 2 x 10'3.* The
data presented in Figure 60 thus confirm Eq. (IV-10).
Equation (IV-10) can be used to predict reactivity relative to propylene
RRT - Tm prop _ bi HCJ
- "
Once RR-rm has been determined at one HCiQ, it can be predicted at other
HCig's by applying Eq. (IV-11). Recall that this is exactly the procedure
(called "scaling") used to predict the mixture reactivity in Experiment 17.
The equivalence of scaling and Tm as reactivity measures can be shown
as follows: Equation (IV-11), with Tm HC. = Tm propylene and [prop]0 = 1 ppm,
states that
K)"1 •
bi
B1 • (IV-12)
prop
* As an order of magnitude check using Eq. (IV-9) and taking Ox as OH-,
0.75 x 104 ppm-1 min-1, OH- = 1.5 x 10-7 ppm; thus
1 x 10-3 min'1 (^ is then 2).
-------
139
250 -
Experimental Conditions
0.1
N00 = 0.5, N02_ = 0.1
N00 = 0.4, N02()
Curve
Fit
HCg (ppm)
FIGURE 60. Tm AS A FUNCTION OF INITIAL HYDROCARBON CONCENTRATION
-------
140
This, by definition (Table 15), is the scaling reactivity of HCj. Because
RRTn) is normalized, HC1(J = [prop]0. Thus, from Eq. (IV-11),
The equivalence of RRTm and scaling follows directly.
The derivation of Eq. (IV-10) can be extended to multihydrocarbon
smog systems by summing over the index 1 from Eq. (IV-4) onward. Equation
(IV-10) then becomes (upon inversion)
HCb
where n is the number of hydrocarbons. Assuming the values of b] are those
obtained in individual hydrocarbon simulations, the linear summation method
results directly. In light of this required assumption, it is surprising
that synergistically modified predictions (Table 18) are not better than
unmodified values. However, the modified values were based on RRT
'%
whereas the above derivation is based on Tm and is not rigorously applicable
to Tjj.
c. The Induction Period
It is worth reiterating that Eqs. (IV-6) and (IV-7) are valid only if
induction period effects are negligible. For this condition to hold, the
initial HC/NOX and N02/N0 ratios must be relatively high (lower bounds have
not been established but 2 and 0.1, respectively, seem reasonable). In
addition, the individual values of HC0, N00, and N02g may themselves be
important.
-------
141
A striking, and possibly disconcerting, feature of Eq. (IV-10) is the
absence of explicit dependence on N02Q. Dimitriades (1972) found that, for
irradiated auto exhaust with an N02o/N00 ratio of about 0.1, the rate of NO
photooxidation was independent of N02Q as long as N02Q was above 0.03 ppm.
The major effect of low values of N02o is to cause a nonnegligible induction
period. It seems reasonable to speculate that an induction period of length
TI will simply cause a shift in the start of the "linear period" by Tj.
Tm can then be replaced by
where T° is given by Eq. (IV-10). RR-^ is defined as follows:
T1 T° + T
, s = - , }
°
mi Ira1 T 'Ij
I prop
m
where Tj/TJjj « 1 has been assumed. If hydrocarbon i is less reactive
than propylene (RRQ < 1) and has an induction period about equal to that
m
or propylene, the correction to RRjO will be positive.* Conversely, for
more reactive hydrocarbons (RR-j-jj) > 1), the correction will be negative.
The induction period, therefore, always has the effect of shifting the
relative reactivity toward unity. The increase in RRTm with decreasing
initial [N02] from a value of 0.25 in Experiment 2 (Table 17) to 0.30 in
Experiment 7 can be cited as evidence of this tendency.
Since atmospheric concentrations include the range of NOg, N02n, and
HC0 for which there is an induction period, methods for predicting TI are
needed. Apparently, an inverse dependence on N02Q is indicated.
* Under these conditions, RRT. % RRTm + (Ti/TJj.) (1 - RRTo)
m Q \ v m
-------
142
C. RELATION OF THE ABOVE CONSIDERATIONS TO OZONE PRODUCTION
The preceding discussion of the issue of relative hydrocarbon reac-
tivity has shown that indices directly related to ozone production are in-
adequate for reactivity assessment. The production of ozone entered only
indirectly through its relation to Tm. In the following discussion, this
relationship is explored in greater detail. As a prefacing note, the dis-
tinction between individual hydrocarbon reactivity (at a fixed NOX concen-
tration) and hydrocarbon-NOx system reactivity (with both hydrocarbon and NOX
being variable) should be emphasized. From a control strategist's point of
view, this distinction is between emissions composition and total emissions.
For the former, which was the topic of the previous section, net ozone pro-
duction is not a sufficient characterization, whereas for the latter, ozone
production (as shown below) is a distinct characteristic.
1. Ozone Isopleths
Isopleths relating the concentration of 03 and the time of the NO?
peak, Tm, to initial hydrocarbon and NO concentrations at various reaction
times are shown in Figures 61 through 66.* These figures are based on the
same computer output used to generate the isopleths contained in the final
report for the first contract year (Hecht et a!., 1973). The initial con-
ditions were as follows:
> [HC]0 = 75 percent n-C^io and 25 percent C3H6
> [N0}0 = as stated on each figure
[N02]Q = 0.1 [NOX]
= 0.35 mlrr1.
As might be expected from the derivation in the preceding section of this
report, the lines of constant Tm (Figure 66) are nearly straight, and their
slope increases as Tm decreases.
* The dashed portions of these figures have been obtained by extrapolation.
-------
143
2.0
i.
a.
o
o
10
4J
O
1.6
1.2
0.8
0.4 -
FIGURE 61. LINES OF CONSTANT 03 (IN PPM) AFTER 1 HOUR OF SIMULATION
-------
144
0.5 0.4
20
o.
o.
o
o
10
0.8
FIGURE 6.2. LINES OF CONSTANT 03 (IN PPM) AFTER 2 HOURS OF SIMULATION
-------
145
i.
o.
(O
4->
O
0.4 -
0 L
FIGURE 63. LINES OF CONSTANT 03 (IN PPM) AFTER 5 HOURS OF SIMULATION
-------
146
a.
a.
o
o
as
4->
o
.0
0.9
FIGURE 64. LINES OF CONSTANT 03 (IN PPM) AFTER 8 HOURS OF SIMULATION
-------
147
O
O
re
to
•P
FIGURE 65. LINES OF CONSTANT 03 (IN PPM) AFTER 9 HOURS OF SIMULATION
-------
148
o
o
•x.
to
4->
O
240
360
540
N00 (ppm)
FIGURE 66. TIME OF THE N02 PEAK (IN MINUTES)
-------
149
In Figures 61 through 65, there are two characteristic regions. A
line drawn along the ridge line of the surface represented by the set of
isopleths in each of these figures would divide these regions. In the re-
gion to the right of the ridge line, the isopleths are fairly straight, and
the super-imposition of Figure 66 would indicate that, especially at earlier
times, they are nearly parallel to the lines of constant Tm. The line
dividing the regions is also nearly a line of constant Tm. The region to
the left is characterized by almost vertical isopleths, though in Figures
64 and 65 they curve back on themselves noticeably.
The features of these figures are not at all surprising; they simply.
reflect the characteristics of 03-versus-time profiles observed in smog
chambers (see, for example, the UCR profiles contained in this report).
The empty space in the lower right-hand corner of Figures 61 through 65
(03 < 0.1 ppm) reflects the finite time that elapses before 03 begins to
build up. At Tm, 03 begins to accumulate almost linearly with time—hence,
the closely spaced isopleths that parallel lines of Tm = constant (Figure 66),
Eventually, [03] approaches an asymptotic level. Correspondingly, the
spacing of isopleths widens, and they turn to the vertical. The reactions
N02 + 03 •*• N03 + 02 , (5)
N03 + N02 * N205 , (7)
N205 + H20 -> 2HN03 , (9)
and
03 + HC •*• Products
along with photolysis and destruction on surfaces, cause 03 depletion late
in the reaction, resulting in the backward curvature of the isopleths.
-------
150
2. Chemical Dynamics
These observations indicate that the characteristics of Figures 61
through 65 are prescribed by the chemical dynamics of smog formation. In
the atmosphere, where chemistry interacts with the mechanical processes of
dispersion and transport, a consideration of dynamics is essential to con-
trol strategy planning. For example, consider the upper region in the 8-
hour isopleths (Figure 64). Figure 62 indicates that, at a fixed level of
NO, a reduction in [HC]0 would have very little effect on 03 production.
The results presented previously show that maximum 03 levels (at fixed NOX
concentrations) are also almost independent of hydrocarbon reactivity (for
a set of olefins). However, an examination of Figure 66 (and the data in
Section B above) shows that a reduction in [HC]0 (or HC reactivity) has a
significant effect on increasing Tm. Thus, a reduction in [HC]0 could slow
down 03 production, even though this decrease may have little effect on the
expected net yield. In the atmosphere, where pollutants can be rapidly dis-
persed, the predicted maximum yield, based on simulations of smog chamber
experiments, may never be realized. The peak 03 level achieved is therefore
closely related to the expected Tm.
From the preceding results, one can conclude that both HC and NO must
be taken into account when attempting to select optimum 03 abatement strate-
gies. However, because of the complex interaction of mechanical and chemi-
cal processes in the atmosphere, it is difficult to extrapolate such results
as those presented in this report directly to atmospheric emissions. To
evaluate the effect of control strategies directly, one would need to imbed
the kinetic mechanism in an airshed model that takes atmospheric conditions
into consideration. In isolation, the kinetic mechanism can only provide
"rules of thumb."
-------
151
V CONCLUDING REMARKS
In closing, some mention should be made of the implications of our
findings on air pollution modeling. First, we briefly summarize these
findings.
Kinetic mechanisms for the chemical transformations occurring in irra-
diated propylene, butane, toluene-NOx, and propylene-NOK-S02 systems were
postulated and used to simulate smog chamber data. Varying degrees of cor-
respondence between predicted and observed profiles were obtained. In general,
the propylene-NO and propylene-NQX-S02 mechanisms were the most successful.
For the most part, their predictions of propylene, 03, NO, N02, and S02 were
reasonably good. Although the accuracy was not very good, the propylene
mechanisms were still able to follow the behavior of each species. The
butane mechanism predicted too rapid NO oxidation and the toluene mechanism
predicted insufficient oxidation of toluene. Butane oxidation products con-
taining structures of two or more carbon atoms were apparently produced in
greater quantity than the mechanism indicated; the low carbon recoveries in
the UCR butane runs support this conclusion. However, more kinetic and smog
chamber data are needed before the toluene mechanism can be assessed and
revised.
We demonstrated that uncertainty in the magnitude of surface reactions
and light source spectrum decay, as well as other chamber effects, could ac-
count for a great deal of the discrepancy between data and theory, although
erroneous rate constants and reaction mechanisms contributed also. Instrument
error, a topic barely touched upon here, is another ever-present source of
ambiguity.
Unlike purely gas-phase thermal reactions, surface and photolytic reac-
tions are chamber-dependent; furthermore, for a given chamber they may vary
from experiment to experiment. Their proper treatment would require a con-
siderable and continuous effort toward chamber characterization. Thus, smog
-------
152
chamber experiments, which are meant to clarify the kinetics by eliminating
some of the complexities present in the atmosphere, have introduced their
own problems, although they are not as complex as those in the atmosphere.
We are not denying the value of smog chamber experiments; instead, we are
emphasizing the intricacy of this analysis.
Upon being confronted with the important role chamber effects play in
the laboratory, one cannot help wondering whether atmospheric counterparts
exist, and if so, how to incorporate them into an airshed model. Heterogen-
eous (pseudo-gas-phase) rate constants are functions of surface-to-volume
ratio, as well as the surface's catalytic efficiency, both of which are not
known for the urban environment. Solar irradiation depends on the state of
the upper atmosphere, as well as on meteorological conditions, and has a di-
urnal and seasonal periodic variation. Hopefully, parametric representation
of the variability of the solar spectrum will make the characterization of
the spectrum feasible. Clearly, the spectrum itself affects numerous reac-
tions, and the variation of a single rate constant (such as k]) cannot ade-
quately account for the effects of spectrum variability.
The approach to modeling heterogeneous chemistry used in the present work
was to represent local surface reactions as pseudo-gas-phase reactions. The
pattern of N02 formation was shown to reflect the value assigned to rate con-
stants for heterogeneous (HNOX) chemistry. Capitalizing on this relation-
ship, we determined heterogeneous rate constants by "tuning" to the N02
curve. When applying the mechanism to the atmosphere, one can take a simi-
lar approach. In the absence of requisite kinetic data, tuning to atmos-
pheric N02 data may be possible. Assuming gas phase kinetics are accurately
represented, this approach would provide a practical means of evaluating the
heterogeneous reactions.
The mechanism's applied utility was demonstrated, in Chapter IV, in a
study of hydrocarbon reactivity and ozone formation.* Thus, the kinetic
mechanism can be a useful tool to investigators of photochemical air pollution,
either in the explicit form or in the streamlined, generalized format.
* There the issue of chamber effects was avoided by presenting results on a
relative.basis.
-------
153
REFERENCES
Altshuller, A. P., and J. Bufalini (1971), Environ. Sci. Techno!., Vol. 5,
p. 39. ~~
Altshuller, A. P., and I. R. Cohen (1963), Int. J. Air Hater Pollut..
Vol. 7, p. 787. --
Altshuller, A. P., S. L. Kopczynski, W. A. Lonneman, T. L. Becker, and
D. L. Wilson (1968), Environ. Sci. Technol . , Vol. 2, pp. 696-698.
Altshuller, A. P., S. L. Kopczynski, W. A. Lonneman, F. D. Sutterfield, and
D. L. Wilson (1970), Environ. Sci. Techno!., Vol. 4, pp. 44-49.
Atkinson, R. (1975), unpublished.
Atkinson, R., and J. N. Pitts, Jr. (1975), J. Phys. Chem. . Vol 79
pp. 295-298. - -
Badcock, C. C., H. W. Sidebotton, J. G. Calvert, 6. W. Reinhardt, and
E. K. Darmon (1971), J. Am. Chem. Soc.. Vol. 93, pp. 3115=3121.
Batt, L., R. D. McCulloch, and R. T. Milne (1974), "Thermochemical and
Kinetic Studies of Alkyl Nitrites (RONO)," Department of Chemistry,
University of Meston Walk, Aberdeen, Scotland.
Becker, K. H., V. Schurath, and H. Seitz (1974), Int. J. Chem. Kinet.,
Vol. 6, pp. 725-739. ' -
Benson, S. W. (1968), Thermochemical Kinetics (John Wiley & Sons, New York).
Bufalini, M. (1971), Environ. Sci. Techno!., Vol. 8, pp. 685-700.
Burton, S. (1974), private communication.
Calvert, J. G., and R. D. McQuigg (1975), Int. J. Chem. Kinet.. to be
published. ~~ -
Calvert, J. G , and J N. Pitts, Jr. (1967), Photochemistry (John Wiley
& Sons, New York). ~~~ -
Castleman, A. W Jr RE. Davis, H. R. Munkelwitz, I. N. Tang, and
/,', Wood <1974)' "Abstracts of Symp. on Chem. Data for the Lower
and Upper Atmos.," Warrenton, Virginia.
'."7?'pp' 161*182* and °' W' Bottenhe1m (1975>» Int. J. Chem. Kinet..
-------
154
Cox, R. A. (1972), J. Phys. Chem.. Vol. 76, p. 814.
(1972), J. photo. Chem.. Vol. 3, pp. 291-304.
Cox, R. A., and S. A. Penkett (1972), J. Chem. Soc.. Faraday Trans. I,
Vol. 68, pp. 1736-1753.
Darnell, K. (1974), private communication.
Davis, D. D. (1974), "Absolute Rate Constants for Elementary Reactions of
Atmospheric Importance: Results from the University of Maryland's
Gas Kinetics Laboratory," Chemistry Department, University of Maryland,
College Park, Maryland.
Davis, D. D., W. Bellinger, and S. Fischer (1975), J. Phys. Chem., Vol. 79,
pp. 293-294.
Davis, D. P., G. Smith, and G. Klauber (1974), Science, Vol. 179, p. 280.
Demerjian, K. L., A. J. Kerr, and J. G. Calvert (1974a), "The Mechanism of
Photochemical Smog Formation," in Advances in Environmental Science
Technology, J. N. Pitts and R. L. Metcalf, eds., John Wiley & Sons
New York.
Demerjian, K. L., J. G. Calvert, and D. L. Thorsell (1974b), Int. J Chem
Kinet., Vol. 6, pp. 829-848.
Dimitriades, B. (1973), draft for EPA policy paper.
(1972), Environ. Sci. Techno!.. Vol. 6, p. 253.
Demitriades, B., B. H. Eccleston, and R. W. Hum (1970), J. Air Pollut
Contr. Assoc.. Vol. 20, p. 150. ~
Dimitriades, B., and T. C. Wesson (1972), J. Air Pollut. Contr. Assoc.,
Vol. 22, p. 33. "
Domalski, E. S. (1971), Environ. Sci. Techno!.. Vol. 5, pp. 443-444.
Doyle, G. T., A. C. Lloyd, K. R. Darnall, A. M. Winer, and J. N. Pitts, Jr
(1975), Environ. Sci. Techno!.. Vol. 3, pp. 237-241.
Feldstein, M. (1974), J. Air Pollut. Contr. Assoc.. Vol. 24, p. 469.
Filby, W. G., and H. D. Penzhorn (1974), "Kinetics of the Photoreaction
of S02 with Alkanes~The Diversion of Intermediate Radicals by Oxygen,"
in Abstracts of Symposium on Chemical Data for the Lower and Upper
Atmosphere," Warrenton, Virginia.
-------
155
Finlayson, B. J., J. N. Pitts, Jr., and R. Atkinson (1974), Contribution
#I6PP-UCR-74-3, from Department of Chemistry and Statewide Air Pollu-
tion Research Center, University of California, Riverside, California.
Garvin, D., and R. F. Hampson (1974), "Chemical Kinetics Data Survey VII:
Tables of Rate and Photochemical Data for Modeling of the Stratosphere,"
NBSIR 74-430, National Bureau of Standards, Washington, D. C.
Gitchell, A., R. Simonaitis, and J. Heicklen (1974), JAPCA, Vol. 24,
pp. 357-361.
Glasson, W., and C. Tuesday (1970), Environ. Sci. Techno!., Vol. 4, p. 916.
(1971), Environ. Sci. Technol., Vol. 5, p. 155.
Hecht, T. A., and J. H. Seinfeld (1972), Environ. Sci. Technol., Vol. 6,
p. 47. ~~
Hecht, T. A., P. M. Roth, and J. H. Seinfeld (1973), "Mathematical Simulation
of Atmospheric Photochemical Reactions," Report R73-28, Systems Appli-
cations, Incorporated, San Rafael, California.
Hecht, T. A., M. K. Liu, and D. C. Whitney (1974a), "Mathematical Simulation
of Smog Chamber Photochemical Experiments," Report R74-9, Systems
Applications, Incorporated, San Rafael, California.
Hecht, T. A., J. H. Seinfeld, and M. C. Dodge (1974b), Environ. Sci. Technol.,
Vol. 8, p. 327.
Hefter, J. H., T. A. Hecht, and G. S. Hammond (1972), J. Am. Chem. Soc.,
Vol. 94, pp. 2793-2797.
Heuss, J. M., and W. A. Glasson (1968), Environ. Sci. Technol., Vol. 2,
pp. 1109-1116.
Jaffee, S., and H. W. Ford (1967), J. Phys. Chem.. Vol. 71, p. 1832.
Jaffee, R. J., and F. C. Smith, Jr. (1974), Presentation at 67th Annual
Meeting of the Air Pollution Control Association, Denver, Colorado.
Johnston, H. S., and R. Graham (1974), Canad. J. Chem., Vol. 52, pp. 1415-1423.
Johnston, H. S., J. N. Pitts, Jr., J. Lewis, L. Bufonte, and T. Mottershead
(1970), "Project Clean Air," University of California Task Force
Assignments, Vol. 4.
Jones, G.R.H., and R. J. Cvetanovic (1961), Canad. J. Chem., Vol. 39,
pp. 2444-2451.
Kopczynski, S. L., A. P. Altshuller, and F. D. Sutterfield (1974), Environ
Sci. Techno!.. Vol. 8, pp. 909-918. "
-------
156
Leighton, P. A. (1961), Photochemistry of Air Pollution (Academic Press,
New York). - *
Liu, M. K. (1974), private communication.
Mabely, W. R and A G. Hendry (1974), Stanford Research Institute,
Menlo Park, California.
Mendenhall, G. D., D. W. Golden, and S. W. Benson (1974), "The Very Low-
Pressure Pyrolysis (VLPP) of n-Propyl Nitrite, tert-Butyl Nitrite
n2n^™yl ^^^"^te. Constants for Some Alkoxy Radical Reactions,"
Department of Thermochemistry and Chemical Kinetics, Stanford Research
Institute, Menlo Park, California.
Miller, D. F. (1975), private communication.
Morris, E. D. , Jr., and H. Niki (1971), J. Phys. Chem. . Vol. 75, pp. 3640-3641
_ (1973), J. Phys. Chem.. Vol. 77, pp. 1929-1932.
> J- ^' Chem. Soc..
Morrison R. T and R. N. Boyd (1971), Organic Chemistry. Second Edition
(Allyn and Bacon, Incorporated, Boston, Massachusetts).
Mulcahy, M.F.R. , JR. Stephen, and J. C. Ward (1967), J. Phys. Chem..
vol. /I, pp. 2124-2131.
Niki HE. E Daby,'and B Weinstock (1972), "Photochemical Smog and Ozone
Reactions," American Chemical Society, Washington, D. C.
Noeh, S., B. H. Nordstrom, and H. S. Johnson (1974), "Kinetic Study of the
Gas-Phase Formation of Nitrous Acid,", Department of Chemistry, University
of California, Berkeley, California.
O'Brien, R. J. (1974), Environ. Sci. Techno! .. Vol. 8, p. 579.
O'Neal, H. E., and C. Blumstein (1973), Int. J. Chem. Kinet.. Vol. 5, pp. 397-413.
Reynolds, S. D., M. K Liu, T. A. Hecht, P. M. Roth, and J. H. Seinfeld (1974),
Atmos. Environ., Vol. 8, pp. 563-596.
Schuck, E. A., E. R. Stephens, M. A. Price, and K. R. Darnall (1972), "Reaction
of Peroxyacetyl Nitrate with Nitric Oxide," Statewide Air Pollution
Research Center, University of California at Riverside.
Schulten, H. R., and U. Schurath (1975), J. Phys. Chem.. Vol. 79, pp.. 51-57.
-------
157
Seinfeld, J. H., T. A. Hecht, and P. M. Roth (1973), "Existing Needs in the
Experimental and Observational Study of Atmospheric Chemical Reactions,"
GPA Report EPA-R4-73-031, Systems Applications, Incorporated, San Rafael,
California.
Sidebottom, H. W., C. C. Badcock, 0. G. Calvert, B. R. Rabe, and E. K. Damon
(1971), J. Am. Chem. Soc.. Vol. 93, pp. 3121-3128.
Sidebottom, H. W., C. C. Badcock, G. W. Jackson, J. G. Calvert, G. W.
Reinhardt, and E. K. Damon (1972), Environ. Sci. Techno!., Vol. 6, pp. 72-79.
Slagle, I. R., J. R. Gilbert, R. E. Grahm, and D. Gutman (1974), "Direct
Identification of Reactive Channels in the Reactions of Hydroxyl Radicals
with Allene, Propylene, and 2-Butene," Department of Chemistry, Illinois
Institute of Technology, Chicago, Illinois.
Smith, J. P., and P. Urone (1974), Environ. Sci. Techno!., Vol. 8, pp. 742-746.
Spicer, C. W., and D. F. Miller (1974), "Nitrogen Balance in Smog Chamber Studies,"
presented at the 67th Annual Meeting of the Air Pollution Control
Association, Denver, Colorado.
Stedman. D. H., C. H. Wu, and H. Niki (1973), J. Phys. Chem., Vol. 77, pp. 2511-
2514.
UCR Monthly Report No. 4 (1974), "A Systematic Investigation of Varibles in
Photochemical Systems in Smog Chamber: An Experimental Basis for Modeling
Photochemical Air Pollution, " Statewide Air Pollution Research Center,
University of Calfornia, Riverside, California.
Wilson, W. E., Jr., and A. Levy (1969), "A Study of Sulfur Dioxide in Photo-
chemical Smog II," 2nd Annual Progress Report to American Petroleum
Institute, Project S-ll.
(1970), JAPCA, Vol. 20, pp. 385-390.
Winer, A. (1975), private communication.
Wu, C. H., and H. Niki (1975), Environ. Sci. Techno!.. Vol. 9, pp. 46-52
Yeung, C.K.K., and C. R. Phillips (1973), Atmos. Environ., Vol. 7, p. 551.
-------
TECHNICAL REPORT DATA
(I lease read Inunctions on the reverse before completing)
EPA-650/4-75-026
4. TITLE AND SUBTITLE
Mathematical Modeling of Simulated
Photochemical Smog
3. RECIPIENT'S ACCESSION-NO.
5. REPORT DATE
June 1975
6. PERFORMING ORGANIZATION CODE
Paul A. Durbin, Thomas A. Hecht, and Gary Z. Whitten
8. PERFORMING ORGANIZATION REPORT NO
EF75-62
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Systems Applications, Inc.
950 Northgate Drive
San Rafael, CA 94903
10. PROGRAM ELEMENT NO
1A1008
11. CONTRACT/GRANT NO.
12. SPONSORING AGENCY NAME AND ADDRESS
U. S. EPA
Office of Research And Development
National Environmental Research Center
Trianglg Park, M r. 977H
68-02-0580
13. TYPE OF REPORT AND PERIOD COVERED
Final (June'74 -June '75)
14. SPONSORING AGENCY CODE "*
15.
JOTES
16. ABSTRACT
The continued development and testing of a kinetic mechanism for photochemical
smog formation is described. Detailed mechanisms containing the individual
chemical reactions occurring in irradiated propylene, n-butlne. toluene-
Nh^h^dHpropylfn!:NY5?2 ?ystems were Postulated and used to simulate smog
chamber data A theofetieal evaluation was made of the contribution of such
chamber effects as light source spectrum decay and surface reactions ?n tho
reactivity of the chamber mixture. The applicatl of kineti? s mu at on to
reactlvi^ and ozone production in smog systems is
17.
_. T
3.
DESCRIPTORS
KEY WORDS AND DOCUMENT ANALYSIS
Photochemical Modeling
Chemical Kinetics
Atmospheric Chemistry
^DISTRIBUTION STATEMENT
Unlimited
Form 2220-1 (9-73)
b.lDENTIFIERS/OPEN ENDED TERMS
19. SECURITY CLASS (THIS Report)'
I. SECURITY CLASS I This page)
Unclassified
c. COSATI Field/Group
21. NO.OF PAGES
160
22. PRICE
------- |