DEVELOPMENT AND TESTING OF THE CBM-IV
FOR URBAN AND REGIONAL MODELING
by
M. w. Gery
G. Z. Whltten
Systems Applications, Inc.
101 Lucas Valley Road
San Rafael, California 94903
and
J. P. Klllus
433 Michigan Avenue
Berkeley, California 94707
Contract No. 68-02-4136
Project Officer
Dr. Marc1 a C. Dodge
Atmospheric Chemistry and Physics Division
Atmospheric Sciences Research Laboratory
Research Triangle Park, North Carolina 27711
ATMOSHERIC SCIENCES RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U. S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NC 27711
-------
REVISIONS TO THE CB4 MECHANISM
The following revisions should be made to the CB4 mechanism listed in
TABLE 1-3:
Reaction (46) C2O3 + NO •* NO2 + XO2 + FORM + HO2
Change the Reaction Rate Data to the following:
Pre-factor Temp. Factor Rate Constant (5) 298K
5.15 E + 04 EXP(- 180/T) 2.82 E+04
Reaction (47) C2O3 + NO2 ->• PAN
Change the Reaction Rate Data to the following:
Pre-factor Temp. Factor Rate Constant (a) 298K
3.84 E +03 EXP(+ 380/T) 1.37 E + 04
Reaction (48) PAN -*. C2O3 + NO2
Change the Reaction Rate Data to the following:
Pre-factor Temp. Factor Rate Constant (5) 298K
1.20 E+18 EXP(-13500/T) 2.54 E-02
Reaction (82)
Add the following reaction:
XO2 + H02 -> (No Products)
With the following Reaction Rate Data:
Pre-factor Temp. Factor Rate Constant (5) 298K
1.134E + 02 EXP(+ 1300/T) 8.90 E +03
-------
BiblioFile Catalog
Title: Development and Testing of the CBM-IU
-------
DEVELOPMENT AND TESTING OF THE CBM-IV
FOR URBAN AND REGIONAL MODELING
by
M. W. Gery
G. Z. WhHten
Systems Applications, Inc.
101 Lucas Valley Road
San Rafael, California 94903
and
J. P. Klllus
433 Michigan Avenue
Berkeley, California 94707
Contract No. 68-02-4136
Project Officer
Dr. Marda C. Dodge
Atmospheric Chemistry and Physics Division
Atmospheric Sciences Research Laboratory
Research Triangle Park, North Carolina 27711
ATMOSHERIC SCIENCES RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U. S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK. NC 27711
-------
DISCLAIMER
This report has been reviewed by the Atmospheric Sciences Research Labora-
tory, U. S. Environmental Protection Agency, and approved for publica-
tion. Approval does not signify that the contents necessarily reflect the
view and policies of the U. S. Environmental Protection Agency, nor does
mention of trade names or commercial products constitute endorsement or
recommendation for use.
ii
-------
ABSTRACT
A new chemical kinetics mechanism for simulating urban and
regional photochemistry based on the carbon-bond method of hydrocarbon
condensation was developed and evaluated in this project. The Carbon-
Bond Mechanism-IV (CBM-IV) is a condensed version of an expanded mech-
anism (CBM-X) that was initially developed using currently available
laboratory and smog chamber data. In addition to a general updating of
the mechanism to include the most recent kinetic, mechanistic, and photo-
lytic information, the CBM-IV comprises extensive improvements to the
chemical representations of aromatic species, biogenic hydrocarbons and
PAN. CBM-IV performance in predicting ozone, formaldehyde, and PAN
concentrations was evaluated against the results of approximately 160
experiments from four different smog chambers. Both the maximum predicted
concentrations and the time to the maximum were compared. Other para-
meters such as hydrocarbon and NOX decay rates were also addressed. The
results of these evaluations indicate substantial improvement in the
ability of the CBM-IV to simulate aromatic and isoprene systems. Formald-
ehyde predictions for the isoprene experiments were also very good. The
CBM-IV overpredicted maximum ozone concentrations by 2% and underpredicted
formaldehyde by 9% for 68 different hydrocarbon/NOx mixture experiments
from three different smog chambers.
-------
CONTENTS
List of Illustrations vi1
List of Tables xx
1 INTRODUCTION 1
Chronological Development of the CBM 1
Surrogate Hydrocarbon Approximations 5
Description of the Carbon Bond Mechanlsm(s) 7
Specific Improvements to the Mechanisms 22
2 MECHANISM DEVELOPMENT METHODOLOGY 25
Formulation of the Mechanism 25
Mechanism Evaluation and Adjustment 26
Mechanism Condensation 29
Atmospheric Applications 30
3 THE INORGANIC AND CARBONYL REACTION SET 33
Inorganic Chemistry 33
Formaldehyde Reactions 46
Acetaldehyde and PAN Reactions 49
Photolysis Reactions 70
Condensation of Inorganic and Carbonyl Reactions 95
4 REACTIVE HYDROCARBON CHEMISTRY 99
Alkyl Group (Paraffin) Chemistry 99
Development of Alkyl Chemistry for the CBM-X 116
Olefin and Ethene Chemistry .- 130
-------
Aromatic Chemistry 139
I soprene 156
5 REPRESENTATION OF SMOG CHAMBER PROCESSES
AND CHARACTERISTICS 159
Photolytlc Processes 159
Wall-Related Processes 175
6 EVALUATION AND DEMONSTRATION OF THE CBM-IV 206
Aldehydes and PAN 207
Ethene and Olefins 210
Aromat ics 212
Biogenlc Hydrocarbons 215
Mixtures 215
Summary of Simulation Results 220
7 SUMMARY AND CONCLUSIONS 395
References 400
vi
-------
FIGURES
1-1 Schematic of Carbon-Bond Mechanism development 3
1-2 Schematic description of the hierarchy of species for
the Carbon Bond Mechanism-IV 9
2-1 Schematic representation of the development, condensation,
evaluation and application of a photochemical kinetics
mechani sm 27
3-1 Simulation results and experimental data for the UNC
biacetyl-NOx experiment of 8 August 1980 54
3-2 Simulation results and experimental data for the UNC
acetaldehyde-NOx experiment of 8 August 1980 55
3-3 Simulation results and experimental data for the UNC
acetaldehyde-NOx experiment of 26 December 1977 56
3-4 Simulation results and experimental data for the UNC
acetaldehyde-NOx experiment of 19 November 1977 57
3-5 Simulation results and experimental data for the UNC
acetaldehyde-NOx experiment of 20 November 1977 59
3-6 Simulation results and experimental data for the UCR
acetaldehyde-NOx experiment EC-254 60
3-7 Simulation results and experimental data for the UCR
acetaldehyde-NOx experiment EC-253 61
3-8 Temperature dependence functions for the peroxyacetyl
radical reactions with NO and N02 63
3-9 m-Xylene concentration traces for UCR EC-345 comparing
the effects of Including radical disproportionation
products in the H02 plus R02 reactions .' 67
vii
-------
3-10 Actinic flux for 20 and 60 degree solar zenith angles
1n bin sizes of 1 and 5 nm 72
3-11 Photoactlon spectra for N02 using 1 and 5 nm bins 76
3-12 Comparison of absorption cross sections for formaldehyde
from Moortgat (top) and Bass (bottom) 86
3-13 Photoaction spectra for formaldehyde to stable products
(top) and formaldehyde to radicals (bottom) using the
10 nm bins of NASA (1855) and the 1 nm bins from Bass
(1980) 87
3-14 Photoaction spectra for acetaldehyde using 1 and 5 nm bins.... 89
3-15 Comparison of relative j-value curve shapes with respect
to solar zenith angle for five major photolytic species 90
3-16 Comparison of photoaction spectra for jNQ2 and JQ^Q at
20 and 60 degrees solar zenith angles 91
4-1 Schematic representation of the PAR reaction scheme 121
4-2 Rate constants for OH plus aromatics 143
4-3 Experimental results for JN2784B 146
4-4 Experimental resuls for JN2784R 147
5-1 Comparison of experimental data for EC-232 and EC-246 164
5-2 Data points are daily average atmospheric turbidity in
North Carolina 168
5-3 Specific photolysis rate of nitrogen dioxide and Incidental
total solar radiation and ultraviolet radiation as a
function of time for 13 October 1976 170
5-4 Data points from Zafonte 173
5-5 Data points from Stedman 174
5-6 Simulation results for UNCR 72179 181
5-7 Estimated hydroxyl radical concentrations for UCR EC-238
using two different radical estimation techniques 186
vm
-------
5-8 Estimated radical concentrations for a set of UCR EC
experiments [[[ 187
5-9 Estimated radical concentrations for two UCR ITC
experiments [[[ 188
5-10 Ozone and PAN traces for ITC-627, an acetaldehyde-only
experiment [[[ 191
5-11 Experimental data from the two-day, ITC-631 experiment
with the dark period removed and the two light periods
merged [[[ 196
5-12 Estimated radical source strengths for five UNC smog
chamber experiments using two different estimation
techni ques [[[ 198
5-13 Comparison of estimated radical Input rates and kj
for chamber characterization experiments in the UNC chamber... 201
5-14 Histogram used to estimate the relationship of radical
input rate to kj (JNQ2) ....................................... 203
6-1 Simulation results for UNC experiment OC0984B
(symbols are experimental measurements) ....................... 222
6-2 Simulation results for UNC experiment OC0984R
(symbols are experimental measurements) ....................... 223
6-3 Simulation results for UNC experiment AU0179B
(symbols are experimental measurements) ....................... 224
6-4 Simulation results for UNC experiment AU0279B
(symbols are experimental measurements) ....................... 225
6-5 Simulation results for UNC experiment AU0479B
(symbols are experimental measurements) ....................... 226
-------
6-8 Simulation results for UNC experiment JN1482R
(symbols are experimental measurements) 229
6-9 Simulation results for UNC experiment AU2482B
(symbols are experimental measurements) 230
6-10 Simulation results for UNC experiment JN1482B
(symbols are experimental measurements) 231
6-11 Simulation results for UNC experiment AU2482R
(symbols are experimental measurements) 232
6-12 Simulation results for UNC experiment AU0479R
(symbols are experimental measurements) 233
6-13 Simulation results for UNC experiment OC0584B
(symbols are experimental measurements) 234
6-14 Simulation results for UNC experiment OC0584R
(symbols are experimental measurements) 235
6-15 Simulation results for UNC experiment AU1078B
(symbols are experimental measurements) 236
6-16 Simulation results for UNC experiment AU1078R
(symbols are experimental measurements) 237
6-17 Simulation results for UNC experiment AU2378B
(symbols are experimental measurements) 238
6-18 Simulation results for UNC experiment AU2378R
(symbols are experimental measurements) 239
6-19 Simulation results for UNC experiment JL0880R
(symbols are experimental measurements) 240
6-20 Simulation results for UNC experiment JL0986B
(symbols are experimental measurements) 241
6-21 Simulation results for UNC experiment JL0986R
(symbols are experimental measurements) 242
6-22 Simulation results for UNC experiment JL1386B
(symbols are experimental measurements) ,. 243
-------
6-23 Simulation results for UNC experiment JL1386R
(symbols are experimental measurements) 244
6-24 Simulation results for UNC experiment OC1184B
(symbols are experimental measurements) 245
6-25 Simulation results for UNC experiment OC1284B
(symbols are experimental measurements) 246
6-26 Simulation results for UNC experiment AU1679R
(symbols are experimental measurements) 247
6-27 Simulation results for UNC experiment OC1278B
(symbols are experimental measurements) 248
6-28 Simulation results for UNC experiment OC2578B
(symbols are experimental measurements) 249
6-29 Simulation results for UNC experiment SE2383B, and SE2383R
(symbols are experimental measurements) 250
6-30 Simulation results for UNC experiment SE2583B, and SE2583R
(symbols are experimental measurements) 251
6-31 Simulation results for UNC experiment SE2783B, and SE2783R
(symbols are experimental measurements) 252
6-32 Simulation results for UCR experiment EC-177
(symbols are experimental measurements) 253
6-33 Simulation results for UCR experiment EC-121
(symbols are experimental measurements) 254
6-34 Simulation results for UCR experiment EC-278
(symbols are experimental measurements) 255
6-35 Simulation results for UCR experiment EC-123
(symbols are experimental measurements) 256
6-36 Simulation results for UCR experiment EC-124
(symbols are experimental measurements) 257
6-37 Simulation results for UNC experiment JN2779B .
(symbols are experimental measurements) 258
XI
-------
6-38 Simulation results for UNC experiment JL3080R
(symbols are experimental measurements) 259
6-39 Simulation results for UNC experiment JN2784B
(symbols are experimental measurements) 260
6-40 Simulation results for UNC experiment AU1579B
(symbols are experimental measurements) 261
6-41 Simulation results for UNC experiment AU0183R
(symbols are experimental measurements) 262
6-42 Simulation results for UNC experiment OC2782R
(symbols are experimental measurements) 263
6-43 Simulation results for UNC experiment AU2782B
(symbols are experimental measurements) 264
6-44 Simulation results for UCR experiment EC-266
(symbols are experimental measurements) 265
6-45 Simulation results for UCR experiment EC-271
(symbols are experimental measurements) 266
6-46 Simulation results for UCR experiment EC-327
(symbols are experimental measurements) 267
6-47 Simulation results for UCR experiment EC-340
(symbols are experimental measurements) 268
6-48 Simulation results for UCR experiments EC-264,EC-265,
EC-269, and EC-273 (symbols are experimental measurements).... 269
6-49 Simulation results for UNC experiment JN2784R
(symbols are experimental measurements) 270
6-50 Simulation results for UCR experiment EC-346
(symbols are experimental measurements) 271
6-51 Simulation results for UCR experiment EC-345
(symbols are experimental measurements) 272
6-52 Simulation results for UCR experiment EC-344
(symbols are experimental measurements) 273
xii
-------
6-53 Simulation results for UNC experiments JL3080B,
AU2782R, AU0183B, and OC2782B (symbols are
experimental measurements) 274
6-54 Simulation results from the CALL (Carter et al., 1986)
and earlier CBMX (Whltten et al., 1985) mechanisms
from UNC JL3080R 275
6-55 Simulation results for UNC experiment JL1480B
(symbols are experimental measurements) 276
6-56 Simulation results for UNC experiment JL1480R
(symbols are experimental measurements) 277
6-57 Simulation results for UNC experiment JL1680B
(symbols are experimental measurements) 278
6-58 Simulation results for UNC experiment JL1680R
(symbols are experimental measurements) 279
6-59 Simulation results for UNC experiment JL1780B
(symbols are experimental measurements) 280
6-60 Simulation results for UNC experiment JL1780R
(symbols are experimental measurements) 281
6-61 Simulation results for UNC experiment JN1780B
(symbols are experimental measurements) 282
6-62 Simulation results for UNC experiment JN1780R
(symbols are experimental measurements) 283
6-63 Simulation results for UNC experiment JN2080R
(symbols are experimental measurements) 284
6-64 Simulation results for UNC experiment JN2280B
(symbols are experimental measurements) 285
6-65 Simulation results for UNC experiment JN2280R
(symbols are experimental measurements) 286
6-66 Simulation results for UNC experiment JN2380B
(symbols are experimental measurements) 287
6-67 Simulation results for UNC experiment JL2580B
(symbols are experimental measurements) 288
xiii
-------
6-68 Simulation results for UNC experiment JL2580R
(symbols are experimental measurements) 289
6-69 Simulation results for UNC experiment JN1580B
(symbols are experimental measurements) 290
6-70 Simulation results for UNC experiment JN1580R
(symbols are experimental measurements) 291
6-71 Simulation results for UNC experiment JL1580B
(symbols are experimental measurements) 292
6-72 Simulation results for UNC experiment JL1580R
(symbols are experimental measurements) 293
6-73 Simulation results for UNC experiment JN1480B
(symbols are experimental measurements) 294
6-74 Simulation results for UNC experiment JN1480R
(symbols are experimental measurements) 295
6-75 Simulation results for UNC experiment JN2180R
(symbols are experimental measurements) 296
6-76 Simulation results for UNC experiment AU2284R
(symbols are experimental measurements) 297
6-77 Simulation results for UNC experiment AU2584B
(symbols are experimental measurements) 298
6-78 Simulation results for UNC experiment ST0184B
(symbols are experimental measurements) 299
6-79 Simulation results for UNC experiment ST0284B
(symbols are experimental measurements) 300
6-80 Simulation results for UNC experiment ST0284R
(symbols are experimental measurements) 301
6-81 Simulation results for UNC experiment ST0384R
(symbols are experimental measurements) 302
6-82 Simulation results for UNC experiment AU0484J*
(symbols are experimental measurements) 303
xiv
-------
6-83 Simulation results for UNC experiment AU0584B
(symbols are experimental measurements) 304
6-84 Simulation results for UNC experiment AU0584R
(symbols are experimental measurements) 305
6-85 Simulation results for UNC experiment AU0684B
(symbols are experimental measurements) 306
6-86 Simulation results for UNC experiment AU0684R
(symbols are experimental measurements) 307
6-87 Simulation results for UNC experiment AU0784R
(symbols are experimental measurements) 308
6-88 Simulation results for UNC experiment AU0884B
(symbols are experimental measurements) 309
6-89 Simulation results for UNC experiment AU0984R
(symbols are experimental measurements) 310
6-90 Simulation results for UNC experiment ST0884B
(symbols are experimental measurements) 311
6-91 Simulation results for UNC experiment ST0884R
(symbols are experimental measurements) 312
6-92 Simulation results for UNC experiment ST1784R
(symbols are experimental measurements) 313
6-93 Simulation results for UNC experiment ST2184B
(symbols are experimental measurements) 314
6-94 Simulation results for UNC experiment ST2184R
(symbols are experimental measurements) 315
6-95 Simulation results for UCR experiment EC-231
(symbols are experimental measurements) 316
6-96 Simulation results for UCR experiment EC-232
(symbols are experimental measurements) 317
6-97 Simulation results for UCR experiment EC-233
(symbols are experimental measurements) 318
-------
6-98 Simulation results for UCR experiment EC-237
(symbols are experimental measurements) 319
6-99 Simulation results for UCR experiment EC-238
(symbols are experimental measurements) 320
6-100 Simulation results for UCR experiment EC-241
(symbols are experimental measurements) 321
6-101 Simulation results for UCR experiment EC-242
(symbols are experimental measurements) 322
6-102 Simulation results for UCR experiment EC-243
(symbols are experimental measurements) 323
6-103 Simulation results for UCR experiment EC-245
(symbols are experimental measurements) 324
6-104 Simulation results for UCR experiment EC-246
(symbols are experimental measurements) 325
6-105 Simulation results for UCR experiment EC-247
(symbols are experimental measurements) 326
6-106 Simulation results for UCR experiment ITC-630
(symbols are experimental measurements) 327
6-107 Simulation results for UCR experiment ITC-631
(symbols are experimental measurements) 328
6-108 Simulation results for UCR experiment ITC-633
(symbols are experimental measurements) 329
6-109 Simulation results for UCR experiment ITC-635
(symbols are experimental measurements) 330
6-110 Simulation results for UCR experiment ITC-637
(symbols are experimental measurements) 331
6-111 Simulation results for UNC experiment OC0382B
(symbols are experimental measurements) 332
6-112 Simulation results for UNC experiment OC0382R
(symbols are experimental measurements)..... 333
xvi
-------
6-113 Simulation results for UNC experiment AU3181B
(symbols are experimental measurements) 334
6-114 Simulation results for UNC experiment AU3181R
(symbols are experimental measurements) 335
6-115 Simulation results for UNC experiment JL2281B
(symbols are experimental measurements) 336
6-116 Simulation results for UNC experiment JL2281R
(symbols are experimental measurements) 337
6-117 Simulation results for UNC experiment JN2581B
(symbols are experimental measurements) 338
6-118 Simulation results for UNC experiment JN2581R
(symbols are experimental measurements) 339
6-119 Simulation results for UNC experiment DE0782B
(symbols are experimental measurements) 340
6-120 Simulation results for UNC experiment DE0782R
(symbols are experimental measurements) 341
6-121 Simulation results for UNC experiment SE1682B
(symbols are experimental measurements) 342
6-122 Simulation results for UNC experiment SE1682R
(symbols are experimental measurements) 343
6-123 Simulation results for UNC experiment NV1182B
(symbols are experimental measurements) 344
6-124 Simulation results for UNC experiment NV1182R
(symbols are experimental measurements) 345
6-125 Simulation results for UNC experiment SE1882B
(symbols are experimental measurements) 346
6-126 Simulation results for UNC experiment SE1882R
(symbols are experimental measurements) 347
6-127 Simulation results for both sides of UNC experiments
SE1481, SE2081, and SE2981 (symbols are experimental
measurements) 348
xvn
-------
6-128 Simulation results for both sides of UNC experiments
SE0381 and SE1081 (symbols are experimental
measurements) 349
6-129 Simulation results for UNC experiment AU2681B
(symbols are experimental measurements) 350
6-130 Simulation results for UNC experiment AU2681R
(symbols are experimental measurements) 351
6-131 Simulation results for UNC experiment SE1984B
(symbols are experimental measurements) 352
6-132 Simulation results for UNC experiment SE1984R
(symbols are experimental measurements) 353
6-133 Simulation results for UNC experiment OC0483B
(symbols are experimental measurements) 354
6-134 Simulation results for UNC experiment OC0783B
(symbols are experimental measurements) 355
6-135 Simulation results for UNC experiment OC0783R
(symbols are experimental measurements) 356
6-136 Scatter diagrams comparing predicted maximum ozone
concentration versus measured values for aldehyde
(top) and alkene (bottom) smog chamber experiments 357
6-137 Scatter diagrams comparing predicted maximum ozone
concentration versus measured values for aromatic
(top) and blogenlc hydrocarbon (bottom) smog chamber
experiments 358
6-138 Scatter diagrams comparing predicted maximum ozone
concentration versus measured values for reactive
hydrocarbon mixture experiments (top) and all
simulated experiments (bottom) 359
6-139 Scatter diagrams comparing predicted formaldehyde
concentration versus measured values for aldehyde
(top) and alkene (bottom) smog chamber experiments 360
xvi i 1
-------
6-140 Scatter diagrams comparing predicted maximum formaldehyde
concentration versus measured values for aromatic
(top) and biogenic hydrocarbon (bottom) smog chamber
experiments
6-141 Scatter diagrams comparing predicted maximum formaldehyde
concentration versus measured values for reactive
hydrocarbon mixture experiments (top) and all simulated
experiments (bottom) 362
xix
-------
TABLES
1-1 The Carbon-Bond Mechanism series 2
1-2 The Carbon Bond Mechanism-X 10
1-3 The Carbon Bone Mechan1sm-IV 17
3-1 Absorption cross sections and quantum yields for NC^
photolysis 73
3-2 Results of jNQ2 calculations 75
3-3 Absorption cross sections and quantum yields for
formaldehyde photolysis to radical products 78
3-4 Absorption cross sections and quantum yields for
formaldehyde photolysis to stable products 79
3-5 Absorption cross sections and quantum yields for ozone
photolysis to O^D) 80
3-6 Absorption cross sections and quantum yields for
acetaldehyde photolysis to radical products 81
3-7 Absorption cross sections and quantum yields for
nitrous acid photolysis 82
3-8 Absorption cross sections and quantum yields for hydrogen
peroxide photolysis 83
3-9 Results of j-value calculations 85
4-1 Average concentration (ppb) of each hydrocarbon species
identified in ambient air by the ERT and WSU methods for
23 common analyses 100
4-2 Calculated nitrate yields compared to measured yields
for various alkanes 115
xx
-------
4-3 Explicit alkyl group chemistry 117
4-4 Carbon and bond types 122
4-5 Alkyl group chemistry of the Carbon Bond Mechan1sm-X 125
4-6 Species Involved 1n the oxidation of paraffin 126
5-1 Relative spectral distribution for the blacklights in the
SAPRC 6400-liter Indoor Teflon Chamber (ITC) 161
5-2 Summary of average relative spectral distributions used
1n modeling the SAPRC Evacuable Chamber (EC) experiments
by Carter et al. (1986) 163
5-3 Comparison of total NOX with measured MONO 1n the UCR-EC 179
5-4 Estimated radical levels using two different methodologies
for the UCR-ITC 190
5-5 Chamber background radical sources as a fraction of total
radical Inputs for the UCR-EC 193
5-6 Chamber background radical sources as a fraction of total
radical Inputs for smog chambers other than the UCR-EC 194
5-7 Radical Inputs for CO/NOX experiments in the UNC
smog chamber 200
6-1 Initial conditions for UNC formaldehyde smog chamber
simulations 363
6-2 Results of UNC formaldehyde simulations 364
6-3 Initial conditions for acetaldehyde and propionaldehyde
smog chamber simulations 365
6-4 Results of acetaldehyde and propionaldehyde simulations 366
6-5 Initial conditions for UNC ethene smog chamber
simulations 368
6-6 Results of UNC ethene simulations 369
6-7 Initial conditions for olefin smog chamber simulations 370
xxi
-------
6-8 Results of olefin simulations
6-9 Initial conditions for toluene smog chamber simulations 373
6-10 Results of toluene simulations 374
6-11 Initial conditions for xylene smog chamber simulations 376
6-12 Results of xylene simulations 377
6-13 Initial conditions for UNC isoprene and a-pinene
smog chamber simulations 379
6-14 Results of UNC isoprene and a-pinene simulations 380
6-15 Initial conditions for UNC SynUrban and SynAuto
mi xture s imu 1 at i ons 382
6-16 Results of UNC SynUrban and SynAuto mixture
simulations 383
6-17 Initial conditions for UCR-EC seven component
hydrocarbon mixture simulations 385
6-18 Results of UCR-EC seven component hydrocarbon
mixture simulations 386
6-19 Initial conditions for UCR-ITC multi-day simulations 387
6-20 Results of UCR ITC multi-day experiment simulations 388
6-21 Initial conditions for UNC hydrocarbon reactivity
experiment simulations 390
6-22 Results of UNC hydrocarbon reactivity experiment
simulations 393
xxn
-------
1 INTRODUCTION
For over a decade, the United States Environmental Protection Agency (EPA)
has supported the development and testing of photochemical kinetics mech-
anisms used 1n urban and regional air quality simulation models (AQSMs).
This report describes the development and evaluation of the most recent
version of the Carbon-Bond Mechanism (CBM), known as the CBM-IV. As a
technical basis for this condensed version, a full (expanded) version, the
CBM-X, was first developed. The CBM-X is suitable for use in less complex
and computationally demanding AQSMs or chemical kinetic models, but serves
primarily as the basis for the CBM-IV, which was developed for use in
larger models such as the Urban Airshed Model (UAM), the EPA Regional
Oxidant Model (ROM), the EPA OZIPM/EKMA model, and various Systems Appli-
cations models [i.e., Regional Transport Model (RTM)].
The remainder of this section contains an overview of the CBM. We briefly
consider the chronology of its development and discuss the fundamental
difference between the CBM approach and those of mechanisms that use other
hydrocarbon surrogate approximations. Section 2 describes the methodology
of formulation, condensation, and evaluation used 1n the CBM. Sections 3
and 4 document the chemical characteristics of the individual components
of the chemical mechanisms, Including photolysis rate calculations and
descriptions and specific examples of condensation techniques used to
formulate the CBM-IV. Section 5 discusses smog chamber characteristics
and artifacts. Section 6 summarizes model evaluation results.
CHRONOLOGICAL DEVELOPMENT OF THE CBM
Since 1976, Systems Applications has been developing the CBM for use in
urban and regional AQSMs. The series of mechanisms developed to date is
given in Table 1-1; a flow chart depicting the logical succession of
mechanisms 1s shown in Figure 1-1. The first version of the CBM was
formulated in 1977 (Whitten and Hogo, 1977) and was highly parameterized
due to the limited state of knowledge at the time. Since then, our know-
ledge of photochemical processes has expanded considerably, primarily as a
result of the efforts of the EPA and other organizations to gather experi-
mental data, particularly chemical kinetic and smog chamber data. This
Increased knowledge was first reflected in CBM-11'and CBM-III. All three
-------
TABLE 1-1. The Carbon-Bond Mechanism series.
Mechanism
Description
Number of
Reactions/Species
Reference
CBM-I Original CBM
CBM-II Update/Expansion
of CBM-I
CBM-III Update of CBM-II
with improved
aromatlcs chemistry
CBM-X Expanded version of
CBM suitable for
use in EKMA
CBM-XR CBM-X with isoprene
chemistry for
regional scale
modeling
CBM-RR Condensed version
of CBM-XR
CBM-IV Original, highly
condensed version
of CBM-RR
CBM-X Update of 1985 CBM-X
CBM-IV Latest version of
condensed CBM
35/20
65/27
75/36
146/67
170/78
113/47
70/28
164/71
81/33
Whltten and Hogo (1977)
Whitten et al. (1980a)
Whltten et al. (1980b)
Killus and Whitten (1982a)
Whitten et al. (1985a)
Whltten et al. (1985b)
Whltten et al. (1985b)
Whltten and Gery (1986)
This work
This work
-------
1
Current |
L
V
i
n
CBM-I
x
CBM-II
i
CBM-III
. CBM-XR
— J CBn X p andCBM-RR
"1 ir
CBM-IY
, Updated j Early
1 CBM-X 1 LBH-IV
j
Condensed Expanded Original ROM
lechanisms Mechanisms Mechanisms
FIGURE 1-1. Schematic of Carbon-Bond
Mechanism development.
-------
of these early CBM mechanisms were developed for use 1n sophisticated
AQSMs and are, therefore, highly condensed.
In 1985, Whltten et al. (1985a) presented an expanded version of the CBM
known as the CBM-X for use in the OZIPM-3/EKMA model. The CBM-X contained
a more complete kinetic description than was previously employed with that
version of OZIPM, but its use was feasible because the computational
demands of OZIPM were much less than those of larger AQSMs. The CBM-X
provided the basis on which previous versions had been developed; it was
validated by comparing the chemistry of species common to almost all
organic systems (e.g., formaldehyde, acetaldehyde, PAN, and CO) with the
results of numerous smog chamber experiments containing those compounds.
In addition, the first formulations of ethene (ETH), 1-olefins (OLE),
mono-alkylbenzenes (TOL), multi-alkylbenzenes (XYL), and paraffins (PAR)
were developed and tested and the complete mechanism was evaluated with
smog chamber data from the University of North Carolina (UNC) facility
(Jeffries et al., 1985a) for a series of authentic and synthetic auto-
mobile exhaust/NOx experiments; and for a series of aromatic/NOx experi-
ments in the Battelle chamber.
The CBM-X was then further expanded for regional use by the inclusion of
isoprene (ISOP) chemistry (Whltten et al.. 1985b). Following this addi-
tion, the complete mechanism was again evaluated with a set of smog cham-
ber data from the University of California at Riverside (UCR) (Carter et
al., 1985). These experiments provided a multi-day testing environment in
which nighttime reactions could be investigated for their accuracy and
importance to condensation procedures. The use of a large data set from
this second outdoor chamber also provided a clearer description of
significant smog chamber characteristics. The mechanism was then con-
densed for use in the first version of the EPA/ROM. As shown in Table
1-1, the resulting mechanism (CBM-RR) contained 113 reactions and 47
species. Although the CBM-RR was appropriate for use in the ROM, it was
still a very large and complex mechanism. As a result of changes in the
ROM mechanism Input requirements, we developed a more condensed version of
the CBM-RR nominally Identified as CBM-IV (Whitten and Gery, 1986).
Through the development process just described and shown in Figure 1-1, it
can be seen that the initial version of the CBM-IV was based on the origi-
nal CBM-X and the ensuing evaluations with smog chamber data. Throughout
this lengthy evaluation process, certain aspects of the original CBM-X
became better understood. As a result, the work reported here was under-
taken to update the technical basis of the CBM-X and to transfer this new
representation to the condensed CBM-IV. We have also evaluated CBM-IV
performance against smog chamber data and provided information and altera-
tions needed to Implement the CBM-IV in the ROM and the newly revised
OZIPM-4/EKMA.
-------
SURROGATE HYDROCARBON APPROXIMATIONS
The CBM-IV developed 1n this project uses a combination of Inorganic
chemistry and explicit and surrogate organic reaction schemes. The
Inorganic and Carbonyl Reaction Set (ICRS) explicitly treats Inorganic
chemistry and the reactions of some product species (e.g., formaldehyde,
acetaldehyde, and PAN) common to several hydrocarbon mechanisms. Reactive
hydrocarbons are usually treated within a surrogate approximation; how-
ever, some specific hydrocarbons that (1) cannot easily be included into
surrogate schemes and (2) are common constituents of polluted atmospheres
(I.e., ethene, isoprene, and to a lesser degree, toluene, xylene, and
acetaldehyde) warrant individual treatment. The determination of species
to be treated 1n either surrogate or explicit reaction schemes was per-
formed by Whitten et al., (1985a) for a large number of species reportedly
measured in the atmosphere. To test the significance of Individual hydro-
carbon species, Whitten et al. used three ranking schemes: the OH reac-
tion rate, relative concentration, and fraction of reacted hydrocarbon
after 12 hours. This work established the importance of ethene, toluene,
and the xylenes (and possibly isoprene). The reader 1s referred to that
work (Whitten et al., 1985a) for further details.
The large number of hydrocarbons Involved in tropospheric photochemistry
precludes a fully explicit chemical treatment of each individual compound,
causing problems of generalization and categorization in kinetic mechanism
development for organic species. Surrogate approaches, 1n which indi-
vidual entities are used to represent a variety of similarly reacting com-
pounds, offer sets of natural categories for hydrocarbon classification;
however, each surrogate approximation inevitably involves departures from
the true chemistry of individual compounds. If the classification error
becomes too great, then another category must be added to the mechanism,
and the simulation process becomes less compact and analytically tractable
and more expensive.
Several techniques have been devised to generalize and classify reactive
hydrocarbons and their chemistry. Molecular lumping 1s outwardly the
simplest (e.g., Atkinson et al., 1982; Carter et al., 1986; NCAR, 1987).
In this methodology, a general hydrocarbon category is developed from the
average behavior of a number of different compounds. The principal
requirement for a molecular lumping scheme is that the behavior of the
Individual compounds not depart too greatly from the average behavior of
the category. Rate constants for each generalized hydrocarbon class
-------
should, 1n theory, be some appropriate average rate of the Individual com-
pounds contained in each class.*
Noninteger stoichiometric parameters are also required to describe reac-
tion pathways for the lumped molecular species and the lumped inter-
mediates generated by the chemistry. The choice of the proper values for
the stoichiometric parameters poses an additional problem similar to the
averaging of rate constants. In practice, these mechanisms usually
specify some default value that may or may not approximate the ensemble
average. If the hydrocarbon mixture to be modeled is sufficiently dis-
similar to the mixture used to create the default values, an erroneous
prediction of chemical dynamics could result. The proper conversion of
such dissimilar mixtures into lumped molecular concentrations is often a
complicated problem that may not have a unique solution. The stoichio-
metric parameters that describe the average molecular size and product
distribution of the lumped species can also affect both reactivity and
carbon balance. The specification of these parameters to simultaneously
satisfy all of these constraints can be difficult, and some lumped
mechanisms have been shown to violate the reactive-carbon-conservation
requirement (Whitten, 1983).
Another technique that has been used to reduce the difficulty of the
generalization problem is lumped structural reactivity classification. In
this technique categorization is based on the similarity of unique struc-
tural groups common to many different molecules. The source of this
approach is the explicit decoupling of submolecular reactivity components
from the method of carbon conservation. The physical basis of such an
* The averaging of rate constants can be a difficult and potentially seri-
ous problem. The range of values over which such averaging is performed
must be minimized so that the effects of highly reactive compounds
will not be underpredicted at the beginning of simulations and the
effects of the less reactive compounds remaining at the end of the simu-
lations will not be exagerated. Most smog chamber experiments involve
initial concentrations of precursors with no additional precursors added
during the course of the experiment; the most reactive compounds
strongly affect the chemistry in the early parts of these experiments.
A generalized mechanism using average rate constants based on a wide
range of initial reactivities might test well against smog chamber data,
but still not have appropriate averages for accurate atmospheric simula-
tions. A generalized mechanism using average rate constants based on
narrow ranges of reactivities will be reliable over a wide spectrum of
conditions.
-------
empirical finding appears to be that the forces between atoms in the same
or different molecules are very "short range"; that is, they are appreci-
able only over distances of the order of 0.1 to 0.3 nm (Benson, 1976; Ben-
son and Bass, 1958). Benson and Bass have shown that it is possible to
describe many molecular properties using a hierarchical system of addi-
tivity laws in which the simplest- or "zeroth-order" law would be the law
of additivity of atom properties. In an atom-additivity scheme, one
assigns partial values for the property in question to each atom in the
molecule. The molecular property is thereby the sum of all the atom con-
tributions. Thus, in the lumped structure approach, the carbon atom
structures within the hydrocarbon molecules are the lumping category. For
example, one lumped species might be used to represent all the single-
bonded carbon atoms that occur not only in alkanes, but also in other
organics such as olefins and substituted aromatics. Another lumped
species might be used to represent pi-bonded carbon in the double-bond
groups that are found in olefins.
The lumped structure approach offers several advantages over the lumped
molecular approach. Because similar structures have similar reaction rate
constants, the averaging problem is minimized. Carbon conservation can be
expressly maintained, and carbon balance during the simulation can be
monitored. Because the number of structural carbon bonding groups is
relatively small, a large number of compounds can be uniquely assigned to
carbon-bond concentrations. Some problems will occur in attempting to
assign structures not expressly accounted for (e.g., esters, ethers, alco-
hols, amines, mercaptans, etc.) in the CBM. The principal disadvantage is
that intramolecular processes such as decomposition can be difficult to
treat. Also, because the lumped structure approach is less intuitive than
the lumped molecule approach, extensive documentation is required to
enable others to use mechanisms based on this approach. The structural
reactivity generalization method has been the basis for our mechanism
design philosophy because it eliminates many of the troublesome features
inherent in the molecular lumping method.
DESCRIPTION OF THE CARBON BOND MECHANISM(S)
CBM-IV and CBM-X were developed and evaluated on the basis of relation-
ships among hierarchical species in the overall chemical mechanism. The
core of this hierarchy includes both a set of inorganic reactions and the
reactions of carbonyl species common to most tropospheric photooxidation
systems, the ICRS. The remaining reactions are made up of various organic
reaction subsets that describe either specific hydrocarbons or condensed
forms of the chemistry of other hydrocarbons (or hydrocarbon struc-
tures). The hierarchical approach to mechanism development was first out-
lined in Whitten et al. (1979); a more recent description of mechanism
-------
development and evaluation based on the independent development and
evaluation of individual components of a chemical mechanism in a hier-
archical manner is contained in Gery (1987). The most important objective
of this approach is the minimization of compensating errors in the final,
overall chemical mechanism. We now focus on the hierarchical relationship
between general reactivity groups of the CBM as illustrated in Figure
1-2. The specific reaction mechanisms of CBM-X and CBM-IV are given in
Tables 1-2 and 1-3. The subgroups of these mechanisms that appear in the
hierarchical structrue are briefly described next.
Inorganic Chemistry
Reactions 1 through 36 describe the atmospheric chemistry of species that
contain only oxygen, nitrogen, and hydrogen atoms and one reaction involv-
ing CO.
Common Carbonyl and Methane Chemistry
Reactions 37 through 73 in the CBM-X (Table 1-2) describe the detailed
chemistry of the common carbonyl compounds formaldehyde, acetaldehyde, and
acetone. Other carbonyls such as methylglyoxal and benzaldehyde, are
considered along with their respective parent hydrocarbons. The common
carbonyl chemistry has been condensed to reactions 37 through 50 in the
CBM-IV (Table 1-3). CBM-X reaction 74 (reaction 51 in the CBM-IV)
describes the chemistry of methane. Since methane 1n the atmosphere does
not decay from an assumed constant level of 1.85 ppm, the concentration of
methane has been lumped into the rate constant. The reactions to this
point provide enough overall chemistry to generate photochemical ozone and
PAN; therefore, this set can be Independently tested against smog chamber
experiments. Section 3 describes the development of this reaction set and
describes some of these tests.
Paraffinic or Single-Bonded Carbon Chemistry
CBM-X reactions 75 through 92 are designed to describe the chemistry of a
typical atmospheric mixture of paraffins. The single-bonded constituents
of other types of molecules, such as olefins and aromatics, are also
treated as the one-carbon PAR species. This reaction scheme has been con-
densed to reactions 52 through 55 in the CBM-IV.
8
-------
I PARAFFINS J I OLEFINS | j^AROMATICS I Hydrocarbons
Carbonyla
PAN Compounds
Formaldehyde
[CO]
NO NOa NO, NjO,
MONO HNO, HO HO, H2O,
Inorganics
Loire* t Lerel
FIGURE 1-2. Schematic Description of the Hierarchy of Species for the
Carbon Bond Mechanism-IV.
-------
TABLE 1-2. The Carbon Bond MechanUM-X.
Nu*>er
1)
2)
3)
4)
5)
6)
7)
8)
9)
10)
11)
12)
1 3\
14)
15)
16)
17)
18)
19)
20)
21)
22)
23)
24)
25)
26)
27)
28)
29)
30)
3D
32)
33)
34)
35)
36)
37)
38)
39)
40)
41)
^* /
42)
43)
44)
45)
46)
47)
48)
49)
Reaction Rate Data
Reaction1 Pre-factor
03
0
0
0
03
03
03
N03
N03
N03
NO
NO
OH
OH
HONO
OH
OK
H02
H02
OH
H02
H02
OH
OH
FORM
FORM
FORM
FORM
FROX
FROX
ALD2
ALD2
ALD2
4
4
4
4
4
4
*
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
N02
0
NO
N02
N02
NO
N02
03
03
01D
010 + H20
OH
H02
N03
NO
N02
N02
N205 4 H20
N205
NO
N02 4 H20
NO
HOMO
HONO
HONO
N02
HN03
NO
N02
PNA
PNA
H02
H02 4 H20
H202
H202
CO
OH
FORM
FORM
0
N03
H02
FROX
H02
NO
0
OH
N03
AL02
-hvl->
_ __ _»y
_»___>
— — >
,
*
____.*
-hvl-'
-hv4-'
_- — f
>
>
—hvl-*
>
>
>
>
>
>
_.__.>
. ..->
-hvl-*
>
>
>
>
>
>
>
>
>
>
-hvS-*
>
>
>
-hv2->
-hv3->
>
>
r
T
»
>
>
-hv5->
NO 4 0
03
N02
NO
N03
N02
N03
0
DID
0
2.000H
H02
OH
0.89N02 4 0.890 4 0.11NO
2.00N02
NO 4 N02
N205
2.00HN03
NO 3 + N02
2. 00 NO 2
2. 00 HO NO
HONO
OH 4 NO
N02
NO 4 N02
HN03
N03
OH 4 N02
PNA
H02 4 N02
N02
H202
H202
2.000H
H02
H02
H02 4 CO
2.00H02 4 CO
CO
OH 4 H02 4 CO
totrn 4 NO? * en
FWJ ^ nu£ ^ w
FROX
H02 4 FORK
PROX
N02 4 H02 4 FACD
C203 4 OH
C203
C203 4 HM03
ME02 4 H02 4 CO
8.383
2.643
1.375
2.303
3.233
1.760
5.300
1.147
3.260
2.344
2.100
3.390
1.909
3.660
7.849
1.900
2.110
2.600
1.600
6.554
1.975
9.770
1.500
1.537
7.600
5.482
1.640
2.876
1.909
8.739
7.690
2.550
4.720
3.220
1.500
4.302
Q Wl
1.480
9.000
9.600
1.040
1.739
1.037
3.700
£404
£403
£404
E+02
E+02
E+02
E-02
£+05
£+03
E+Ol
E+Ol
E+04
E+Ol
£+02
E-06
E+16
E-05
E-ll
£402
E-01
£+03
E-05
£403
£403
E+02
E+15
£403
£+01
£-10
£-01
£403
£402
£404
£404
Em
— wl
£401
£401
£403
£404
£404
£404
Temp. Factor Rate Constant « 298 K
exp((-E/R)/T) k^ (ppm-^ln'1)
*£XP(
*£XP(-
*£XP(
*EXP{
•EXP(-
*£XP(
•£XP(-
*EXPf-
fcAr i
*EXP(
*£XP(-
*EXP(
1175/T)
1370/T)
687/T)
602/T)
2450/T)
390/T)
940/T)
580/ri
t^JV/ t g
250/T)
1230/T)
256/T)
*EXP(-10897/T)
*EXP(
*EXP(
*EXP(
•£XP(
*£XP(
*£XP(
530/T)
806/T)
713/T)
1000/T)
240/T)
749/T)
*£XP(-10121/T)
*EXP(
*EXP(
*EXP(
*EXP(-
*EXP(-
•£XP(-
•EXP(
380/T)
1150/T)
5800/T)
187/T)
1550/T)
986/T)
250/T)
see notes
4.323 E406
2.664 E+Ol
1.375 E404
2.309 E+03
2.438 E+03
4.731 E-02
5.300 E-02.
see notes
4.246 E+05
3.260
1.000 £402
2.999
3.390 E+Ol.
4.416 E+04
5.901 £-01
1.853 E+03
1.900 E-06
2.776
1.539 £-04
1.600 £-11
9.799 £403
1.975 £-01.
9.770 E403
1.500 E-05
1.682 £404
2.179 £402
1.227 £404
2.025 £+03
5.115
6.833 £403
4.144 £403
2.180 £-01
2.550 £-01.
2.520 £403
3.220 £402
1.500 £404
see notes
see notes
2.370 E402
9*nn r ni
. JUU t -U 1
1.480 £401
9.000 £401
9.600 £403
1.040 £404
6.360 £402
2.400 £404
3.700
see notes
"l
"l
kl
k39
(Continued)
10
-------
TABLE 1-2. (Continued).
Number
|
Reaction1
Reaction Rate Data
Pre-factor
Te«.p.
Factor Rate Constant • 298K
(ppnflBln~J) exp((-E/R)A) kjgg (ppjT^I
50)
51)
52)
53)
54)
55)
56)
57)
58)
59)
60)
61)
62)
63)
64)
65)
66)
67)
68)
69)
70)
71)
72)
73)
74)
75)
76)
77)
78)
79)
80)
81)
82)
83)
84)
85)
86)
87)
88)
19)
90)
91)
92)
93)
94)
96)
97)
C203
C203
ME02
ME02
MEO
MEO
MEO
MEN3
MNIT
ME02
ME02
ME02
C203
ME02
ME02
C203
C203
AONE
AN02
PAR
PAR
R02
R02
R02R
R02R
ROR
X
D
D
D
D
A02
0
0
0
0
OH
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
NO
N02
PAN
N02
MPNA
NO
NO
NO
N02
MEO
OH
OH
MNIT
MEOZ
ME02
C203
C203
H02
H02
H02
H02
AONE
OH
NO
OH
OH
OH
NO
NO
NO
NO
N02
ROR
ROR
ROR
ROR
PAR
PAR
PAR
PAR
«T
NO
RET
OLE
OLE
OLE
OLE
OLE
> W2 4 ME02
> PAN
> C203 4 N02
> MPNA
> «02 4 N02
_____> MFD * Hf\?
-----> MN IT
> FORM 4 H02 4 NO
> NEN3
> FORM 4 H02
> FORM + N02
> FORM + NO
-hvl-> MEO + NO
> 2.00MEO
> FORM + MEOH
> ME02 4 MEO
> 2.00ME02
> PROX
> MEO + OH
> pROX
> NE02 4 OH
-hvl-> ME02 4 C203
> AN02
> C203 + FORM + N02
> ME02
> R02
> R02R
> N02 4 H02 4 AL02 + X
> NTR
> N02 4 ROR
_ — > NTR
> NTR
> RET 4 H02
> KET 4 0
> ALD2 4 D 4 X
> AONE 404 2.00X
>
> R02
> A02 4 2.00X
> R02R
> C203 4 X
> N02 4 AONE 4 H02
-hvl-> C203 4 R02 4 2.00X
> 2 PAR
> ALD2
> H02 4 CO 4 R02
> R02 4 X 4 CO 4 FORM 4 OH
> MEQ2 4 ALD2 4 X
7.915
1.180
5.616
3.810
1.190
6.140
2.296
1.920
2.217
2.605
4.000
3.000
3.000
1.910
3.560
4.400
3.700
2.550
8.541
2.000
7.600
4.000
5.800
1.200
6.521
1.360
1.067
1.140
6.584
1.033
1.836
2.200
9.546
1.236
2.394
2.621
1.000
7.500
2.250
2.500
1.000
1.200
3.000
6.146
6.146
1.756
3.512
7.740
E+03
E-04
E+18
E+02
E+17
E+03
E+04
E+03
E+04
E+07
E+02
E+02
E-01
E+02
E+02
E+03
E+03
E+01
E+01
E+03
E+03
E-05
E+02
E+04
E+03
E+02
E+03
E+04
E+04
E+04
E+05
E+04
E+04
E+16
E+16
E+16
E+04
E+03
E+03
E+02
E+04
E+04
E-04
E+03
E+03
E+03
E+03
E+03
•EXP(
*EXP(
250A)
5500A)
*EXP(-14000/T)
*EXP(
821A)
*EXP(-10371/T)
•EXP(
*EXP(
*EXP{-
*EXP(
*EXP(
,
*EXP(-
*EXP(-
*EXP(-
«EXP(-
*EXP(-
*EXP(-
*EXP(-
•EXP(-
*CXP(-
*EXP(-
«£XP(
180A)
200/T)
1200A)
1300A)
1300A)
1710A)
1400A)
1400A)
8000A)
8000A)
8000A)
324A)
324A)
324A)
324A)
504A)
1.831
1.223
2.220
5.990
9.145
.123
.443
.920
.217
.645
.000
3.000
3.000
1.910
3.560
4.400
3.700
2.000
6.700
2.000
7.600
4.000
5.800
1.200
2.100
.360
.067
.140
.000
.033
.673
2.200
9.546
2.706
5.250
5.749
1.000
7.500
2.250
2.500
1.000
1.200
3.000
2.072
2.072
5.920
1.184
4.200
E+04
E+04
E-02
E+03
E+02
E+04
E+04
E+03
E+04
E+05
E+02
E+02
E-01.
E+02
E+02
E+03
E+03
E+03
E+03
E+03
E+03
E-05.
E+02
E+04
E+01
E+02
E+03
E+04
E+02
E+04
E+03
E+04
E+04
E+04
E+04
E+04
E+04
E+03
E+03
E+02
E+04
E+04
E-04.
E+03
E+03
E+02
E+03
E+04
" )
ki
A
k.
A
kl
(Continued)
11
-------
TABLE 1-2. (Continued).
NMber
Reaction1
Reaction Rate Data
Pre-factor
Te«p.
Factor Rate Constant » 298 K
(ppnT'fcln"1) exp((-E/R)/T) k^g (pi
98)
99)
100)
101)
102)
103)
104)
105)
106)
107)
108)
109)
110)
111)
112)
113)
114)
115)
116)
117)
118)
119)
120)
121)
122)
123)
124)
125)
126)
127)
128)
129)
130)
131)
132)
133)
134)
135)
136)
137)
138)
139)
140)
141)
142)
143)
144)
145)
146)
147)
03
03
03
03
N03
PH02
PN02
0
0
OH
ET02
ET02
03
03
CRIG
CRIG
CRIG
CRIG
MCRG
MCRG
MCRG
MCRG
OH
OH
OH
B02
OH
BZ02
BZ02
PH02
PHO
OH
OH
N03
CRO
CR02
CR02
T02
T02
4 OLE
4 OLE
4 OLE
4 OLE
4 OLE
4 NO
4 NO
4 ETH
4 ETH
4 ETH
4 NO
4 NO
4 ETH
4 ETH
HOTA
HOTA
HOTA
HTMA
HTMA
HTMA
HTMA
HTMA
4 NO
4 H20
4 FORM
4 ALD2
4 NO
4 H20
4 FORM
4 AL02
4 TOL
4 TOL
4 TOL
4 NO
BZA
4 BZA
4 NO
4 N02
PBZN
4 NO
4 N02
4 CRES
4 CRES
4 CRES
4 N02
4 NO
4 NO
4 NO
4 NO
T02
> ALD2
> FORM
, ALD2
, FORM
, PN02
> ON IT
> FORM
> ME02
» FORM
> FJ02
> N02
> NQ2
> HCHO
> HCHO
__,-->
, CO
4 CRIG
4 MCRG
4 HOTA
4 HTMA
4 AL02
4 H02
4 H02
4 X
4 X
4 X
4 X
4X4 2.00N02
4 CO
4 CO 4 OH
4 2.00FORM 4 H02
4 ALD2
4 CRIG
4 HOTA
4 H02
> 2.00H02
»
, NE02
, ME02
, FORM
> ME02
> N02
> FACD
> OZD
> OZD
> N02
> ACAC
> OZD
> OZD
> 802
> CRES
> T02
, N02
-hvl->
> BZ02
> N02
> PBZN
> BZ02
> N02
> NPHN
> CSO
> CR02
> CRO
, KCRE
__ „, ttny
> N02
_-_.-> unj
> NTR
> H02
4 CO 4
4 H02
4 CO 4
4 H02
4 FORM
4 H20
4 ALD2
4 H20
4 H02
OH
2.00H02
4 BZA 4 H02
4 PH02
4 N02
4 PHO
4 HN03
4 OPEN
4 ACID
4 OPEN
4 CRES
4 CO
4 H02
4 H02
4 H02
4.208
4.208
6.312
6.312
1.135
1.000
1.000
1.078
4.620
3.000
9.360
2.640
7.424
1.114
2.000
7.000
1.000
2.000
3.200
3.200
6.000
8.000
1.035
5.900
2.950
2.950
1.035
5.900
2.950
2.950
2.393
1.115
1.751
1.200
4.000
2.000
3.700
2.500
5.566
1.200
2.000
2.440
3.660
3.250
2.000
6.000
6.000
1.080
1.200
2.500
£401
£403
£404
E404
E403
£403
£403
£403
£401
£401
£401
£401
£401
£401
£401
£404
£-01
£403
£403
£404
£-01
£403
£403
£402
£403
£403
£404
E-03
£404
£403
£403
£418
£404
£404
£404
£404
£404
£404
£403
£403
£404
£403
£402
*EXP(-
*EXP(-
*EXP(-
*EXP(-
*EXP(-
*EXP(-
*£XP(
*EXP(-
*£XP(-
*EXP(
*£XP(
*EXP(
2105/T)
2105/T)
2105/T)
2105/T)
792/T)
792/T)
411/T)
2633/T)
2633/T)
322/T)
322/T)
322/T)
*EXP(-14000/T)
3.600
3.600
5.400
5.400
1.135
1.000
1.000
7.560
3.240
1.192
9.360
2.640
1.080
1.620
2.000
7.000
1.000
2.000
3.200
3.200
8.000
8.000
1.035
5.900
2.950
2.950
1.035
5.900
2.950
2.950
7.050
3.285
5.160
1.200
4.000
2.000
3.700
2.500
2.200
1.200
2.000
2.440
3.660
3.250
2.000
6.000
6.000
1.080
1.200
2.500
pm-'fcin-1)
E-03
E-03
E-03
E-03
£401
£403
E404
£402
£402
£404
£403
£403
E-03
E-03
£401
£401
E401
£401
£401
£401
£+04
E-01
£403
£403
£404
£-01
£403
£403
£402
£+03
£403
£404
E-03«k,
E404
£403
£403
£-02
£404
£404
£404
£404
£404
£404
£403
£403
E404
£403
E402
(Continued)
12
-------
TABLE 1-2. (Continued).
ftnfcer
148)
149)
150)
151)
152)
153)
154)
155)
156)
157)
158)
159)
160)
161)
162)
163)
164)
OH
OH
OH
OH
XL02
XINT
OH
NGPX
OH
OPPX
OPEN
OPEN
OPEN
OPEN
OPEN
4 XYL
4 XYL
4 XYL
4 XYL
4 NO
4 NO
4HGLY
4 NO
NGLY
4 OPEN
4 NO
OPEN
4 03
4 03
4 03
4 03
4 03
«
Reaction1
> XL02
> CRES
> T02
> XINT
> K02
> N02
> NGPX
> NQ2
-hv2-> C203
> OPPX
> N02
-hv2-> C203
> ALD2
> FORM
> MGLY
> C203
>
Reaction Rate Data .
4 PAR 4
4 H02 4
4 H02 4
4C203
4 CO 4
4 C2034
4 FORM 4
4 CO 4
4 NGPX 4
4 CO 4
4 FORM 4
H02
BZA
Pre-factor
2.453 E+03
4.906 E+03
7.358 E+03
4 PAR
2 NGLY 4 2 PAR
H02
H02
H02
H02
4 CO
4 CO
FORM 4 CO
OH
H02
4 2 H02
4 CO
.810 E+03
.200 E+04
.200 E+04
.600 E+04
.200 E+04
.640
.400 E+04
.200 £404
.040
.409 E-03
.424 E-03
.606 E-02
.738 E-02
8.030 E-03
Te»p. Factor Rate Constant 9 298K
*xp((-E/R)/T) kjgg (ppn-^iln-1)
*EXP(
*EXP{
•EXP(
*EXP(
*EXP(-
*EXP(-
*£XP(-
*EXP(-
*£XP(-
116/T)
116/T)
116/T)
116/T)
500/T)
500/T)
500/T)
500/T)
500/T)
3.620 £403
7.240 E+03
1.086 E+04
1.448 E+04
1.200 E+04
1.200 E+04
2.600 E+04
1.200 E+04
9.640 .
4.400 E+04
1.200 E+04
9.040
4.500 E-04
1.200 E-03
3.000 E-03
8.850 E-03
1.500 E-03
k38
"38
(Continued)
13
-------
TABLE 1-2 (continued).
Notes:
1. Pressure dependent values for M, 02 and CH4 are included 1n the rate
constant data (see text).
[M] = 1 x 106 ppm, (02] = 2.095 x 105 ppm, and [CH41 = 1.85 ppm.
The symbols hvl through hv5 in the reaction listing indicate
photolysis reactions with rates dependent on solar irradiation. The
basis for these rates is discussed in Section 3; the symbols
represent:
« hv2 = JHCHO (to radicals)« hv3 = JHCHO (to H
hv4 = j03 (to 02 + O^D)). hv5 = JCH3CHO-
Other photolysis reactions are ratioed to these functions and noted
in the K2gg column of the mechanism listing.
2. Chemical species in the CBM-X are:
_ Species Name _ Representation
Nitric Oxide NO
Nitrogen Dioxide N02
Nitrogen Trioxide (nitrate radical) N03
Dinitrogen Pentoxide N205
Nitrous Acid HONO
Nitric Acid HN03
Peroxynitric acid (H02N02) PNA
Oxygen Atom (singlet) 01D
Oxygen Atom (triplet) 0
Hydroxyl Radical OH
Water H20
Ozone 03
Hydroperoxy Radical H02
Hydrogen Peroxide H202
Carbon Monoxide CO
Formaldehyde (CH2=0) FORM
Hydroxymethylperoxy Radical (HOCH200*) FROX
Organic Peroxide (ROOH) PROX
Formic Acid (HCOOH) FACD
14
-------
TABLE 1-2 (continued).
Species Name Representation
High Molecular Weight Aldehydes (RCHO, R>H) ALD2
Peroxyacyl Radical (CH3C(0)00') C203
Peroxyacyl Nitrate (CH3C(0)OON02) PAN
Methylperoxy Radical (CH300') ME02
Methylperoxy Nitric Add (CH302N02) MPNA
Methoxy Radical (CH30') MEO
Methyl Nitrite (CH3ONO) MNIT
Methyl Nitrate (CH3ON02) MEN3
Methanol (CH3OH) MEOH
Acetone (CH3C(0)CH3) AONE
Acetylmethylperoxy Radical (CH3C(0)CH200*) AN02
Paraffin Carbon Bond (C-C) PAR
Primary Organic Peroxy Radical R02
Secondary Organic Peroxy Radical R02R
Organic Nitrate NTR
Secondary Organic Oxy Radical ROR
Ketone Carbonyl Group (-C(O)-) KET
Dimethyl Secondary Organic Peroxide Radical A02
Olefinic Carbon Bond (C=C) OLE
Cregiee Biradlcal (H2COO'j CRIG
Methyl Cregiee Biradical (CH3(H)C"00') MCRG
"Excited" Formic Acid HOTA
"Excited" Acetic Add HTMA
Nitrated Organic Peroxy Radical
(-CH(ON02)-CH(00*)-) PN02
C2 01 nitrate Group DNIT
Ethene (CH2=CH2) ETH
Ethanol Peroxide Radical (CH2OH-CH200') ET02
Ozonide and further products OZD
Acetic Add (CH3COOH) ACAC
Toluene (CgH5-CH3) TOL
Benzylperoxy Radical B02
Cresol and higher molecular weight Phenols CRES
Toluene-Hydroxyl Radical Adduct T02
Benzaldehyde BZA
Peroxybenzoyl Radical BZ02
Peroxybenzoyl Nitrate PBZN
Phenylperoxy Radical PH02
Phenoxy Radical PHO
Nitrophenol NPHN
Methylphenylperoxy Radical CR02
15
-------
TABLE 1-2 (concluded).
Species Name Representation
Methylphenoxy Radical CRO
Nitrocresol NCRE
High Molecular Weight Aromatic Oxidation
Ring Fragment OPEN
Aromatic Ring Fragment Acid ACID
Xylene (C6H4-(CH3)2) XYL
Methylbenzylperoxy Radical XL02
Xylene-Hydroxyl Radical Adduct XINT
Methylglyoxal (CH3C(0)C(0)H) MGLY
Peroxide Radical of MGLY (CH3C(0)C(0)00') MGPX
Peroxide Radical of OPEN OPPX
Paraffin Loss Operator (-PAR) X
Paraffin-to-Peroxy Radical Operator D
TOTAL = 71
16
-------
TABLE 1-3. The Carbon Bond Mechan1s»-IV.
Umber
1)
2)
3)
4)
5)
6)
7)
8)
9)
10)
11)
12)
13)
14)
15)
16)
17)
18)
19)
20)
21)
22)
23)
94 \
25)
26)
27)
28)
29)
30)
3D
32)
33)
34)
35)
36)
37)
38)
39)
40)
41)
42)
43)
44)
45)
46)
47)
48)
49)
50)
51)
03
0
0
0
03
03
03
N03
N03
N03
NO
NO
OH
OH
V*'
HONO
OH
OH
H02
H02
OH
H02
H02
OH
OH
FORM
FORM
FORM
ALD2
ALD2
AL02
C203
C203
C203
C203
4
4
4
4
4
4
4
4
4
4
4
4
4
.
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
N02
0
NO
N02
N02
NO
N02
03
03
DID
DID
OH
H02
N03
NO
N02
N02
N205
N205
NO
N02
NO
HONO
HONO
HONO
N02
HN03
NO
N02
PNA
PNA
H02
H02
H202
H202
CO
OH
FORM
FORM
0
N03
0
OH
N03
ALD2
NO
N02
PAN
C203
N02
OH
*
Reaction1
-hvl->
— — _>
r
.-.-_>
•»—.»
>
>
— hwl->
-h«4->
>
4 H20 >
— - — >
>
-h«l->
>
>
>
4 H20 >
>
>
4 H20 >
>
-hvl->
>
— — >
>
__ >
— >
___>
4 H20 >
-hv3->
>
_ — >
>
-hv2->
-h«3-'
— — >
__.>
>
— — >
— — >
-hv5->
->
_ — ->
.->
___-->
__ — ->
>
NO 4 0
03
N02
NO
N03
N02
N03
0
010
0
2.000H
H02
OH
0.89N02 4 0.890 4 0.11NO
2.00N02
NO 4 N02
N205
2.00HN03
N03 4 N02
2.00N02
2.00HONO
HONO
OH 4 NO
NO?
IVU£
NO 4 N02
HN03
N03
OH 4 N02
PNA
H02 4 N02
N02
H202
H202
2.000H
H02
H02
H02 4 CO
CO 4 2.00H02
CO
OH 4 N02 4 CO
HN03 4 H02 4 CO
C203 4 OH
C203
C203 4 HN03
FORM 4 X02 4 CO 4 2.00H02
N02 4 X02 4 FORM 4 H02
PAN
C203 4 N02
2.00FORM 4 2.00X02 4 2.00H02
0.79FORM 4 0.79X02 4 0.79H02 4
0.790H
X02 4 FORM 4 H02
Reaction Rate Data
Pre-factor
(ppa'^ln'1]
8.383
2.643
1.375
2.303
3.233
1.760
5.300
1.147
3.260
2.344
2.100
3.390
1.909
3.660
7.849
1.900
2.110
2.600
1.600
6.554
1.975
9.770
1.500
1.537
7.600
5.482
1.640
2.876
1.909
8.739
7.690
2.550
4.720
3.220
1.500
4.302
9.300
1.739
1.037
3.700
7.915
1.180
5.616
3.700
9.600
6.521
E+04
E+03
E+04
E+02
E+02
E+02
E-02
E+05
E+03
E+01
£401
E+04
E+01
E+02
E-06
£+16
E-05
E-ll
£402
E-01
r . f*i+
E-05
£403
£403
E+02
E+15
£403
£401
£-10
E-01
£403
£402
£404
£404
E-01
£404
£404
£403
£-04
£+18
£403
£403
£403
leap. Factor Rate Constant 9 298 K
1 exp((-E/R)/T) kjgg (ppn'^ln'1)
*£XP(
*£XP(-
•EXP(
*£XP(
*£XP(-
*EXP(
*EXP(-
*EXP(-
*EXP(
*£XP(-
*£XP(
1175/T)
1370/T)
687/T)
602/T)
2450/T)
390/T)
940/T)
580/T)
250/T)
1230/T)
256/T)
*EXP(-10897/T)
*EXP(
*EXP(
*EXP(
*EXP(
*£XP(
*£XP(
S30/T)
806/T)
713/T)
1000/T)
240/T)
749/T)
*EXP(-10121/T)
*EXP(
*£XP(
•EXP(
*EXP(-
*EXP(-
*CXP(-
*£XP(
*EXP(
•EXP(
380/T)
1150/T)
5800/T)
187/T)
1550/T)
986/T)
250/T)
250/T)
5500/T)
*£XP(-14000/T)
*EXP(-
1710/T)
see notes
4.323 £+06
2.664 E+01
1.375 E+04
2.309 E+03
2.438 E+03
4.731 E-02
5.300 E-02.
see notes
4.246 E+05
3.260
1.000 E+02
2.999
3.390 E+01.
4.416 E+04
5.901 E-01
1.853 E+03
1.900 E-06
2.776
1.539 £-04
1.600 E-ll
9.799 E+03
1.975 E-01.
977n F«/)i
. / /U LWJ
1.500 E-05
1.682 E+04
2.179 £+02
1.227 £+04
2.025 E+03
5.115
6.833 £403
4.144 E+03
2.181 E-01
2.550 E-01.
2.520 E+03
3.220 E+02
1.500 E+04
see notes
see notes
2.370 E+02
9.300 E-01
6.360 E+02
2.400 E+04
3.700
see notes
1.831 E+04
1.223 E+04
2.220 E-02
3.700 E+03
9.600 E+03
2.100 E+01
kl
i
|[.
kl
k39
(Continued)
17
-------
TABLE 1-3. (Continued)
Nuaber
52)
53)
54)
55)
56)
57)
58)
59)
60)
61)
62)
63)
64)
65)
66)
67)
68)
69)
70)
71)
72)
73)
74)
75)
76)
77)
78)
79)
80)
81)
PAR
ROR
0
OH
03
N03
0
OH
03
OH
T02
OH
CRES
CRO
OPEN
OPEN
OH
OH
0
OH
03
N03
X02
X02
X02N
4 OH
ROR
ROR
4 N02
4 OLE
4 OLE
4 OLE
4 OLE
4 ETH
4 ETH
4 ETH
4 TOL
4 NO
T02
4 CRES
4 N03
4 N02
OPEN
4 OH
4 03
4 XYL
4 MGLY
MGLY
4 ISOP
4 ISOP
4 ISOP
4 ISOP
4 NO
4 X02
4 NO
Reaction1
> 0.87X02 4 0.13X02H 4 0.11H02 4
0.11ALD2 4 0.76ROR - 0.11 PAR
> 1.10AL02 4 0.96X02 4 0.94H02 4
0.04X02N 4 0.02ROR - 2.10PAR
, H02
' 0.63ALD2 4 0.38H02 4 0.28X02 4
0.30CO 4 0.20FORM 4 0.02X02N 4
0.22PAR 4 0.200H
> FORM 4 ALD2 4 X02 4
H02 - PAR
> 0.50AID2 4 0.74FORM 4 0.33CO 4
0.44H02 4 0.22X02 4 0.100H
- PAR
> 0.91X02 4 FORM 4 AL02 4
0.09X02N 4 N02 - PAR
> FORM 4 0.70X02 4 CO 4
1.70H02 4 0.300H
> X02 4 1.56FORM 4 H02 4
0.22ALD2
> FORM 4 0.42CO 4 0.12H02
> 0.08X02 4 0.36CRES 4 0.44H02 4
0.56T02
' 0.90M02 4 0.90H02 4 0.900PEN
' CRES 4 H02
' 0.40CRO 4 0.60X02 4 0.60H02 4
0.300PEN
, CRO 4 HM03
>
-tw2-> C203 4 H02 4 CO
> X02 4 2.00CO 4 2.00H02 4
C203 4 FORM
> 0.03AL02 4 0.62C203 4 0.70FORM 4
0.03X02 4 0.69CO 4 0.060H 4
0.76H02 4 0.20MGLY
> 0.70H02 4 0.50X02 4 0.20CRES 4
0.80MGLY 4 1.10PAR 4 0.30T02
> X02 4 C203
-hv2-> C203 4 H02 4 CO
> 0.60H02 4 O.BOAL02 4 0.550LE 4
0.50X02 4 0.50CO 4 0.45ETH 4
0.90PAR
> X02 4 FORM 4 0.67H02 4
0.40MGLY 4 0.20C203 4 l.OOETH 4
0.20ALD2 4 0.13X02N
> FORM 4 0.40AL02 4 0.55ETH 4
0.20MGLY 4 0.10PAR 4 0.06CO 4
0.44H02 4 0.100H
> X02N
> N02
__-->
Reaction Rate Data
Pre-factor Temp. Factor Rate Constant * 298K
(pprn'^ln"1) exp((-E/R)A) k^ge (PP"""""^"1)
1.203 E403
6.250 E416 nXP(- 8000A)
9.545 £404
2.200 £404
1.756 £404 *EXP(- 324A)
7.740 £403 *EXP( 504 A)
2.104 E401 *EXP(- 2105A)
1.135 £401
1.540 £404 *£XP(- 792 A)
3.000 £403 *EXP( 411 A)
1.856 E401 *£XP(- 2633 A)
3.106 £403 *EXP( 322A)
1.200 £404
2.500 £402
6.100 £404
3.250 £404
2.000 £404
9.040
4.400 £404
8.030 £-02 *EXP(- SOOA)
2.453 £404 *EXP( 116A)
2.600 £404
9.640
2.700 £404
1.420 £405
1.800 £-02
4.700 £402
1.200 £404
2.550 £401 *EXP( 1300A)
1.000 £403
1.203 £403
1.371 E405
9.545 £404
2.200 £404
5.920 E403
4.200 E404
1.800 £-02
1.135 £401
1.080 £403
1.192 E404
2.700 E-03
9.150 E403
1.200 £404
2.500 £402
6.100 £404
3.250 E404
2.000 E+04
9.040 «k3g
4.400 £404
1.500 £-02
3.620 £404
2.600 E404
9.640 .kjg
2.700 £404
1.420 E405
1.800 £-02
4.700 £402
1.200 £404
2.000 £403
1.000 £403
(Continued)
18
-------
TABLE 1-3 (continued).
Notes:
1. Pressure dependent values for M, 02 and CH4 are Included 1n the rate
constant data (see text).
IM] = 1 x 106 ppm, [02] = 2.095 x 105 ppm, and
[CH4] = 1.85 ppm.
The symbols hvl through hv5 1n the reaction listing indicate
photolysis reactions with rates dependent on solar Irradiation. The
basis for these rates is discussed in Section 3; the symbols
represent:
hvl = jN02, hv2 = jHCHO (to radicals), hv3 = jHCHO (to H2 + CO),
(to 02 + O^D)), hv5 = JCH3CHO'
Other photolysis reactions are ratioed to these functions and noted
in the K298 column of the mechanism listing.
2. Chemical species 1n the CBM-IV are:
_ Species Name _ Representation
Nitric Oxide NO
Nitrogen Dioxide N02
Nitrogen Tr1ox1de (nitrate radical) N03
Dinltrogen Pentoxlde N205
Nitrous Acid HONO
Nitric Add HN03
Peroxynitrlc add (H02N02) PNA
Oxygen Atom (singlet) 01D
Oxygen Atom (triplet) 0
Hydroxyl Radical OH
Water H20
Ozone 03
Hydroperoxy Radical H02
Hydrogen Peroxide H202
Carbon Monoxide CO
Formaldehyde (CH2=0) FORM
High Molecular Weight Aldehydes (RCHO, R>H) ALD2
Peroxyacyl Radical (CH3C(0)OQ-) . C203
19
-------
TALBE 1-3 (concluded).
Species Name Representation
Peroxyacyl Nitrate (CH3C(0)OON02) PAN
Paraffin Carbon Bond (C-C) PAR
Secondary Organic Oxy Radical ROR
Oleflnic Carbon Bond (C=C) OLE
Ethene (CH2=CH2) ETH
Toluene (C6Hc-CH3) TOL
Cresol and higher molecular weight Phenols CRES
Toluene-Hydroxyl Radical Adduct T02
Methylphenoxy Radical CRO
High Molecular Weight Aromatic Oxidation
Ring Fragment OPEN
Xylene (C6H4-(CH3)2) XYL
Methylglyoxal (CH3C(0)C(0)H) MGLY
Isoprene ISOP
NO-to-N02 Operat1on X02
NO-to-N1trate Operation X02N
TOTAL = 33
20
-------
Olefin and Ethene Chemistry
Reactions 93 through 127 1n the CBM-X and reactions 56-62 1n the CBM-IV
constitute the subset of reactions specific to molecules containing double
bonds. Since such molecules can have a wide range of reactivity, the
atmospheric mix of olefins in the CBM-X is treated 1n three different
ways:
(1) Terminal olefins such as propene and 1-butene are handled in
condensed form via the species OLE. OLE is a two-carbon lumped-
structure species that represents the carbon bond found in 1-
olefins. These reactions are condensed to reactions 56 through
59 1n the CBM-IV.
(2) Ethene, which 1s the least reactive of the olefins, 1s treated
explicitly. The ethene mechanism has little significant conden-
sation, and 1s described by reactions 60 through 62 of the CBM-
IV.
(3) Highly reactive olefins such as those with internal double bonds
(e.g., c1s-2-butene) are treated by assuming that they have
already reacted in the atmosphere to form two higher aldehydes
(i.e., 2 ALD2). This condensation has been successfully
demonstrated 1n Whitten et al. (1979) and was recommended for
use 1n earlier versions of the CBM (Whitten et al., 1980a;
Whitten et al., 1980b).
Aromatic Chemistry
The chemistry of all aromatic structures 1s separated into monofunctlonal
and multifunctional rings. The aromatic structure chemistry of molecules
with only one functional group is represented by the reactions of the
toluene molecule (TOL), whereas the m-xylene (XYL) reactions are used for
multifunctional aromatic rings. Reactions 128 through 164 in the CBM-X
(63 through 74 in the CBM-IV) represent the reactions of these species and
their products which Include two highly reactive dicarbonyls, methyl-
glyoxal (MGLY) and a lumped, higher molecular weight dicarbonyl (OPEN).
Isoprene Chemistry
Isoprene (ISOP) chemical reactions are included in the CBM-IV as a surro-
gate for biogenic, terpenoid hydrocarbon species. The chemistry of
Isoprene is represented by reactions 75 through 78 in the CBM-IV.
21
-------
SPECIFIC IMPROVEMENTS TO THE MECHANISMS
The following major improvements have been implemented in the CBM:
(1) New chemical kinetics and stoichiometric information have been
included in both versions to update both inorganic and organic
components.
(2) Water is now explicitly represented in the CBM-IV mechanism (it
is no longer "hidden" from the user by embedding its concentra-
tion in the kinetic expression).
(3) The chemistry of 0(1D) is explicitly represented in the CBM-
IV. Again, the "hidden" kinetics are now visible.
(4) The species ROR is now explicitly represented in the PAR-plus-OH
reaction sequence of the CBM-IV. This alteration, along with
alterations 2 and 3, was performed to eliminate the complicated
empirical stoichiometric and kinetic representations that were
necessarily included in the previous version of the CBM-IV.
Provided that the earlier algorithms were correctly implemented
by users, no functional differences should result from these
changes.
(5) The temperature-dependence of acetylperoxy radical reactions
with NO and N02 has been improved in both mechanisms.
(6) The product distribution and kinetics of the acetylperoxy radi-
cal reactions with R02 radicals have been improved to better
simulate experimental evidence from environmental chambers.
(7) The sensitivity of X02 chemistry has been further tested, and
the reactions of that species have been updated in the CBM-IV.
(8) The chemistry of ETH was altered to account for glycolaldehyde
formation in both mechanisms.
(9) The reaction of ALD2 with H02 was removed from both mechanisms
because there is no substantial proof of its occurrence.
(10) The chemical equilibrium between H02, N02 of peroxynitric acid
(PNA) is listed in both the mechanisms. In our earlier work,
this reaction was considered optional for warm-temperature con-
ditions, which enhance the PNA decomposition reaction; however,
we feel its inclusion will prevent it from being ignored under
conditions for which it is needed.
22
-------
(11) The use of FORM as surrogate for 6LY has been eliminated 1n the
CBM-IV, making the species FORM an explicit representation of
formaldehyde and allowing the CBM-IV to be used for formal-
dehyde-specific simulations.
(12) The chemistry of aromatic species (TOL and XYL) has been signi-
ficantly altered 1n both mechanisms to account for the sensi-
tivity of ozone formation to NMHC/NOX (particularly for TOL).
(13) A new condensed isoprene (ISOP) mechanism has been developed and
evaluated for the CBM-IV.
(14) A new carbon fractionatlon scheme for a-p1nene was developed and
evaluated.
23
-------
24
-------
MECHANISM DEVELOPMENT METHODOLOGY
In this section we briefly describe our mechanism development methodology
1n order to clearly define relevant sources of uncertainty. In addition,
because the photochemical kinetics mechanisms used to simulate atmospheric
chemistry are (1) developed and evaluated using data of different types
and quality, (2) condensed on the basis of differing philosophies, and
(3) Implemented 1n different ways, a general description will provide the
reader with a better understanding of the components of mechanism develop-
ment not evident from a description of only the final mechanism.
FORMULATION OF THE MECHANISM
Photochemical kinetics mechanisms such as the CBM are formulated by review
and evaluation of fundemental kinetic data. In the development of the
CBM, we have drawn heavily on such published review articles as the work
of NASA (DeMore et al., 1985). CODATA (Baulch et al., 1984), and Atkinson
and Lloyd (1984). Such review 1s a time-consuming process and portions of
these works are somewhat dated; therefore, we have supplemented this
Information with a review of more current Information available 1n chemi-
cal journals. Since we will continue to add the results of future experi-
mental studies and mechanism development efforts to this basic data pool,
1t 1s necessary to clearly document the methodology used so that new
Information can be easily integrated. In our mechanism formulation (which
1s documented 1n the following sections), we have attempted to (1) base
all condensations or approximations on elementary reactions and experi-
mentally or thermodynamlcally determined rate constants, (2) reference all
sources of basic kinetic and mechanistic data, and (3) describe all rele-
vant decisions concerning our choices and uses of these data. In this
way, the basic data available for model development at this time have been
formally selected and are no longer considered to be adjustable para-
meters. As our discussion will illustrate, however, the assumptions used
to combine these fundamental data into a working mechanism will evolve as
additional Information becomes available.
25
-------
MECHANISM EVALUATION AND ADJUSTMENT
The evaluation of chemical mechanisms against smog chamber data using a
simple stepwlse hierarchy of simulations was first suggested by Demerjian,
Kerr. and Calvert (1974), who noted that smog chamber data were most
effectively analyzed by starting with simple systems and Increasing the
complexity of the hydrocarbon components in a stepwise fashion. Whitten
(1979) later put forth the principles of the hierarchy of species
approach, a method of model development and testing intended to Identify
sources of simulation uncertainty. Figure 1-2 shows the hierarchy of
species for the CBM-X. The basic concept of this hierarchy is that
mechanisms should be validated starting from the lowest level. In the
case of smog chamber data, this level includes chamber background and
dilution experiments. Once acceptable agreement between simulation
results and measurements is obtained, no changes in rate constants and
reaction stoichlometry can be made for the part of the mechanism that has
been tested. Simulations then increase to higher hierarchical levels.
Thus, if disagreement between simulation results and measurements occurs
at a higher hierarchical level and all lower species data have already
been simulated successfully, 1t is most likely that the new chemistry for
the higher level species is in error. This methodology minimizes the pos-
sibility of fortuitous agreement between simulations and measurements
caused by compensating errors that can occur 1n a mechanism created from
only basic kinetic data. Jeffries and Arnold (1987) point out that the
hierarchical approach narrows the macroscopic nature of data from smog
chamber experiments so that it more clearly confronts Individual portions
of the theory. The opposite method, direct creation of a mechanism to
simulate a mixture of hydrocarbons and NOX 1n the atmosphere without simu-
lating the products of the hydrocarbon mixture, 1s less certain because of
the possibility of compensating errors.
Smog chamber experiments provide one source of additional information.
While our understanding of fundamental atmospheric processes provides the
basic descriptions of chemical variability used in initial mechanism
development, evaluation of the mechanism through comparison with smog
chamber data allows refinement derived from a real, controlled test
environment. That 1s, whereas fundamental chemical data may lack informa-
tion on unstudied reactions, data obtained from smog chamber experiments
Inherently contain much of this information because these experiments are
designed to reproduce many of the chemical processes in the atmosphere.
Therefore, the iterative adjustment of mechanism assumptions through simu-
lation of smog chamber experiments combines the strengths of both types of
data so that uncertainties in each become more apparent and can be elimi-
nated. This iterative type of approach is shown in Figure 2-1, which is
an idealized representation of the mechanism development process, as the
26
-------
ro
Kinetic Date
Mechanistic Data
Theoretical Calculations
Data Evaluation
Assumptions
(Deletions)
1on\
Mechanistic Compilation
of Best Available
Information
Condensation
Process
Evaluation
Verlflca
vs. Data
on j
Released/Published
Version of
Explicit Chemistry
Condensation
Process
Condensed Version
of Current
Explicit Chemistry
f Evaluation, \
[ Verification I-
V vs: Data /
Condensed/Operational
Version of Explicit
Chemistry
Application
Validation J
Prediction \
FIGURE 2-1. Schematic representation of the development, condensation,
evaluation and application of a photochemical kinetics mechanism.
-------
first loop 1n the top left portion of the drawing. The boxes 1n the
figure represent real (though sometimes changing) data, compilations, and
mechanism development points; 1n short, they are starting, nodal, and
stopping points. Listings of data, results, and mechanisms can be
obtained at any of these points to represent the current development
status. The ovals represent processes that often require a certain level
of Ingenuity, test data, and validation. These processes are the sources
of uncertainty that must be clearly described 1n formulating a predictive
chemical kinetics mechanism from basic data, and clearly documented in
each application.
The evaluation and adjustment of current explicit mechanisms, as shown in
Figure 2-1, 1s the primary objective of projects such as the one reported
here. Since the smog chamber experiments performed at different facili-
ties (and even at the same facility) are extremely variable, this portion
of the mechanism development effort can accumulate significant uncer-
tainty. Therefore, 1t 1s essential to establish a robust data set prior
to any mechanism evaluation and adjustment. In the work reported here, we
have utilized data from smog chambers at the University of North Carolina
(UNC), the University of California at Riverside (UCR), and Battelle
Columbus Research Laboratories for a variety of single hydrocarbon species
and mixtures, and different initial conditions (especially HC/NOX ratios)
and times of year (for outdoor chambers). For the UNC dual outdoor cham-
bers data, we have also attempted to always simulate both sides of an
experiment at the same time, using identical chamber artifact and light
conditions for each side so that the objective of each specific experiment
could be studied through comparison of the results for each side.
It is also critical to estimate or define the basic characteristics and
conditions of each chamber. Although these features will be discussed in
a later section, we note here that they are still very uncertain, and
that no meaningful comparison of individual mechanism characteristics can
be conducted 1f those mechanisms were developed using different artifact
assumptions. This is because artificial Inorganic chamber chemistry may
be included as part of the mechanism's organic reactivity, or too much or
too little artificial chemistry might be assumed, resulting in an organic
chemistry section that contains inaccurate treatment of reactivity. At
the level of atmospheric modeling, where these assumptions are supposedly
eliminated from the mechanisms, two mechanisms using different artifact or
light assumptions that match chamber data equally well will give different
atmospheric predictions because one will have excluded or included more
chamber artifacts than the other.
28
-------
MECHANISM CONDENSATION
A photochemical kinetics mechanism used to predict the chemical behavior
of a complex mixture of hydrocarbons under atmospheric conditions must be
condensed enough to be Implemented in an AQSM. The process by which a
condensed mechanism 1s formulated and verified 1s a stepwlse loop similar
to the smog chamber evaluation of explicit mechanisms. We show the con-
densation process 1n Figure 2-1 as a set of parallel loops leading to a
verified, condensed mechanism. The solid lines describe the condensation
of an explicit mechanism that has first been tested with smog chamber
data, while the dashed lines Imply production of an explicit mechanism
from fundamental kinetic data, followed by evaluation of a condensed
mechanism with experimental data. In both cases, the evaluation with smog
chamber data is an essential test against real atmospheric data in order
to establish the completeness of the explicit compilation. It is our pre-
ference to proceed along the path that provides a validated explicit
mechanism (solid lines) prior to condensation since (1) compensating
errors in the mechanism will not be as easily masked by condensation prior
to evaluation with smog chamber, and (2) a complete explicit mechanism
that has been verified against all available data can be used to produce a
number of different levels of condensation for different applications.
The alternate pathway requires a new condensation for every iterative step
in the smog chamber evaluation loop, resulting 1n a more complicated task
to produce the same results.
The objective of the condensation process leads to a more clearly defined
methodology than that of the smog chamber evaluation process. This is
because the object of condensation is not to simulate smog chamber experi-
ments correctly, or to Improve the theory, but to duplicate the ability of
the explicit mechanism in a more widely applicable form. The condensed
mechanism should not simulate the smog chamber results more accurately
since 1t is based only on the explicit mechanism. Therefore, because the
object of a condensation evaluation test is to duplicate the results of a
more explicit set of reactions, 1t is not always necessary to use real
test conditions since the reference standard is not the smog chamber data,
but the results of the explicit mechanism simulation. Thus, we usually
attempt both to verify that the condensed mechanism is an accurate map of
the explicit formulation within the range of physical and chemical condi-
tions available in smog chamber data, and to test the condensation results
1n more extreme conditions than will be used in the intended applica-
tion. Therefore, it 1s often necessary to fabricate these conditions to
achieve such a comparison.
In Sections 3 and 4 of this report we present the theoretical basis for
our development of explicit mechanisms, followed by a discussion of the
development of the CBM-X and the condensation of that mechanism to the
29
-------
CBM-IV. The CBM-X 1s primarily an explicit mechanism, though 1t does
employ hydrocarbon surrogate assumptions, whereas the CBM-IV 1s highly
condensed. (The reader 1s referred to Sections 3 and 4 for discussion of
specific techniques and results.) In both mechanisms, the Inorganic por-
tion of the chemistry 1s maintained 1n the explicit form. Our primary
strategy for development of a CBM-IV mechanism is shown by the solid line
1n Figure 2-1. After evaluation of the fundemental kinetic and mechanis-
tic data, an initial formulation of the CBM-X was attempted. We then
optimized the agreement of the CBM-X with smog chamber data by Improving
our mechanistic assumptions. We proceeded to condense the CBM-X to the
CBM-IV in the stepwlse manner documented in Sections 3 and 4. This
overall process was revisited at various points throughout the project
before the final versions of these mechanisms were decided upon. In both
the development of the CBM-X and the condensation to the CBM-IV. the
hierarchy of species within the photochemical systems to be modeled
allowed compensating errors 1n these steps to be minimized.
ATMOSPHERIC APPLICATIONS
Although this report 1s devoted mainly to a discussion of the condensation
of the CBM-X Into the CBM-IV, we note one final area of uncertainty in
mechanism development and application. Even though a condensed mechanism
with "minimal" uncertainty has been developed using fundamental and smog
chamber data and a well-devised condensation scheme, it is not necessarily
valid for all Intended applications. For Instance, one might say that
even with an Ideal fit to all smog chamber tests and a very good condensa-
tion technique, the chemical mechanism has only been validated for the
chemical and physical conditions exhibited in those data. Different
ranges of concentrations and meteorological conditions are applicable in
the atmosphere; there are also other applications for photochemical
kinetics mechanisms than the prediction of ozone formation.
To address the Issue of uncertainty 1n mechanism application for specific
AQSMs and ambient conditions, we must consider a few facts. First, the
description of conditional variation 1n the chemical mechanisms is not
based solely on ozone-related smog chamber data. Although the smog cham-
ber simulations are used to test the completeness of the mechanism, the
inherent conditional variations are usually based on kinetic studies with
much wider physical ranges than the smog chamber experiments. Therefore,
though the explicit mechanism is only verified against real measurements
within the range of conditions available in a smog chamber, there is no
reason to expect a rapid deterioration of the chemical description for
conditions somewhat beyond those of a smog chamber. As noted, however, 1t
is necessary to confirm that the condensed mechanism sucessfully dupli-
cates the explicit mechanism over the entire range of application.
30
-------
Although smog chamber and ambient concentration ranges can be vastly dif-
ferent, we must rely on our development effort to make the mechanism rea-
sonably complete so that 1t can handle chemical situations that are
untested 1n smog chambers. We know the concentration characteristics of
such conditions from ambient measurements. For instance, the effects of
background species such as methane and CO, which are generally of little
importance in high concentration smog chamber experiments, must be inclu-
ded in atmospheric models. Reactions expected to be significant for con-
ditions of extremely low NOX and VOC, must also be Included to account for
such conditions in rural applications. We have attempted to include the
chemistry for these conditions 1n the CBM.
31
-------
32
-------
3 THE INORGANIC AND CARBONYL REACTION SET
This section describes the Inorganic and Carbonyl Reaction Set (ICRS).
The ICRS 1s the portion of the CBM that contains (1) the basic Inorganic
chemistry of oxygen, nitrogen, and hydrogen species plus one reaction of
CO and (2) the chemistry of simple oxygenated organic species. These
chemical reactions are common to virtually all tropospherlc chemistry and
are the foundation of the CBM. In the current reevaluatlon of the
mechanism, the basic kinetic and mechanistic information contained in the
ICRS was reviewed and updated. Our description of the ICRS is divided
Into a discussion of (1) Inorganic chemistry. (2) formaldehyde reactions,
(3) acetaldehyde and PAN reactions, (4) photolysis reactions, and (5) con-
densation of the ICRS. The carbonyl reactions of higher molecular weight
species (such as acetone and dicarbonyls) are considered 1n later discus-
sions of the hydrocarbon classes from which they are formed.
INORGANIC CHEMISTRY
The Inorganic portion of the ICRS 1s composed of the first 36 reactions
listed 1n Tables 1-2 and 1-3.* The ICRS contained 1n these versions of
the CBM Includes recently published Information; therefore, a brief dis-
cussion of the Individual chemical reactions 1s in order here. The ICRS
Includes both photolytic and thermal reactions. Calculation of photolytic
reaction rates will be discussed 1n detail later in this section. The
current discussion 1s limited to a description of photolytic mechanisms.
The reaction rates of thermal reactions and their temperature dependencies
1n the range of tropospherlc Interest are discussed first. Reaction rates
are presented for 298 K; temperature-dependent expressions are presented
1n Arrhenlus form:
k = A exp((-E/R)(l/T)).
* Reaction numbers given 1n the text refer to the CBM-IV unless they are
preceeded with an X; an X Indicates that a particular reaction 1s from
the CBM-X (Table 1-2).
33
-------
Rate constants for reactions of the type
A + B < — > [AB]* —M--> AB,
which are 1n the fall-off regime between blmolecular and termolecular
reactions, were Initially determined from the following formula (DeMore et
al., 1985):
k0(T)[M] {1
k(M,T) = ( -------------------- ) 0.6
The third order expression 1s given by
= k300(T/300)-n cmV1,
where the value 1s suitable for air as a third body. The high-pressure
rate constant 1s given by:
k.(T) - k.300(T/300)-m cmV1.
The values used 1n this discussion are taken primarily from the NASA
review (DeMore et al.t 1985) (the reader 1s referred to that document for
further description of the foregoing expressions). The curves resulting
from these expressions were then fit with the Arrhenlus formula for the
temperature range of Interest 1n the atmosphere and smog chambers (260 K
to 320 K). In addition, the CBM-IV reported here is designed for simula-
tion of photochemistry near the earth's surface. It should not be used
for high altitude (low pressure) calculations without implementation of
the appropriate pressure-dependent rate expressions.
The photolysis of N02 leads to the production of 0(3P):
N02 — hv--> NO + 0(3P). (1)
0(3P) will subsequently be referred to as 0. The rate of this photolysis
reaction is usually used to gauge the magnitude of light intensity in
experimental facilities and in the atmosphere. However, it 1s really a
measure of the actinic flux only in the region of N02 absorption;
therefore, the photolysis rates of all other species are not uniquely
determined by the rate of reaction 1. Calculation of k(l) and Us
relationship to other photolytic rates will be discussed later 1n this
section.
34
-------
0 atoms react with oxygen molecules through the three-body process:
0 + 02 + M ------ > 03 + M.
This 1s the only significant source of ozone 1n the earth's atmosphere.
We used the low pressure limit of NASA (DeMore et al.t 1985) and the high-
pressure value from CODATA (Baulch et al., 1984):
k0 = 6.0 x NT34 (T/300)'2-3 cm^olecule^s'1,
and
k. = 2.8 x NT12 cm3molecule-1s'1.
For temperature-dependent values of M and 02 (at 1 atmosphere), the reac-
tion can be simplified to
0 (+ 02 + M) ------ > 03 (+ M), (2)
resulting 1n an Arrhenlus expression of
k(2) = 8.383 x 104 exp(1175/T) mlrT1,
and
k(2)298 = 4.32 x 106 mln'1.
In the subsequent discussion and the listing given 1n Tables 1-2 and 1-3,
we have simplified our descriptions by Including the temperature-dependent
values of M and 02 1n the rate expression, noting the simplifications
given next by the use of parentheses in the equation. In a similar way,
products Included in parentheses are not tracked in the mechanism
(generally, because they are species with nearly constant concentrations,
e.g., M, 02, and C02, or because their subsequent reactions are not fur-
ther treated in the CBM, e.g., nitrates, nltrocresols, and organic acids).
The third reaction 1n the ICRS, which along with the first two forms the
basis of the photostationary state relationship, 1s
03 + NO ------ > N02 (+02). (3)
The NASA review (DeMore et al., 1985) places the rate expression at
k(3) = 2.643 x 103 exp(-1370/T)
and
k(3)298 = 26.6 ppnf Am1n~A ,
with an uncertainty of about 20 percent.
35
-------
0 atoms react with N02 1n two ways:
0 + N02 ------ > NO (+02), (4)
and
0 + N02 (+M) ------ > N03 (+M). (5)
For reaction 4, NASA (DeMore et al., 1985) recommends
k(4)298 = 1.375 x 104 ppnf imin'1,
with an uncertainty of about 10 percent. The fall-off values for reaction
5 are also from NASA (DeMore et al., 1985):
k0 = 9.0 x KT32 (T/300)-2'0 cnAnolecule'V1,
and
k = 2.2 x 10'11 cn^molecule'V1,
which lead to an effective Arrhenlus expression of
k(5) = 2.303 x 102 exp(687/T) ppnT^Irr1.
and
k(5)298 = 2.31 x 103 ppm'^ln'1.
The similar three-body reaction for NO 1s
0 + NO (4M) ------ > N02 (4fl). (6)
The NASA (DeMore et al., 1985) low- and high-pressure limit values for
reaction 6 are
k0 = 9.0 x 10'32 (T/300)-1-5 cm6molecule-2s-1,
and
k,,, = 3.0 x 10"11 cm^olecule'^'1,
which lead to an effective Arrhenius expression of
k(6) = 3.233 x 102 exp(602/T)
and
k(6)298 = 2.44 x 103 ppm-^in"1.
The reaction of 03 and N02 can be significant when most of the NO has been
converted to N02. This is an important scenario downwind of an urban area
late in the day or at night, since this reaction represents a nonphoto-
lytic source of radicals:
N02 + 03 > N03 (+02). (7)
36
-------
NASA (DeMore et al., 1985) and Atkinson and Lloyd (1984) use an Arrhenius
expression of
k(7) = 1.760 x 102 exp(-2450/T)
and
k(7)298 = 4.73 x 1(T2
assuming an uncertainty of approximately 15 percent.
Ozone photolysis occurs 1n two channels with photoactlon spectra 1n
significantly different spectral locations (see the discussion of photoly-
sis rate later 1n this section). The higher energy (short wavelength)
process results 1n the formation of 0(^0) , while the process that occurs
at longer wavelengths forms 0(3P) (referred to as 0):
03 — hv— > 0 (+ 02) (8)
03 — hv— > O^D) (+ 02) (9)
In the ICRS, the photolytlc rate of reaction 8 1s Implemented as a ratio
to k(l) (since N02 absorbs at long wavelengths), while photolytlc rate 9
must be explicitly Input with respect to zenith angle because Its short
wavelength energy source does not vary uniquely with J^Q .
The 0(*D) formed 1n reaction 9 can be converted to the lower electronic
energy state of 0 via energy transfer to oxygen and nitrogen molecules:
0(n) (+02) ------ > 0 (
and
O^D) (+ N2) ------ > 0 (+ N2).
Following the recommendations of NASA (DeMore et al., 1985), we combine
these reactions to yield
O^D) (+ M) ------ > 0 (+ M). (10)
The combined kinetic expression for reaction 10 is
k(10) = 1.147 x 1010 exp(390/T) min"1,
and 1n .
k(10)298 = 4.25 x 101U min'1,
with an uncertainty of about 20 percent.
37
-------
Atmospheric water vapor can compete with M for 0( D) through reaction 11:
0(1D) + H20 — — > 2 OH. (11)
Although an Insignificant reaction loop results from the combination of
reactions 8 through 10, reaction 11 is an important hydroxyl -radical -form-
ing process. One way of thinking of this is that the presence of ozone in
daylight can convert water vapor to hydroxyl radicals. This process is
important near the period of ozone buildup, when radical concentrations
are increasing. At that point, reaction 11 tends to provide a "bootstrap"
function, in which the formation of new ozone promotes the formation of
new hydroxyl radicals. The rate of reaction 11 is temperature-independent
and NASA (DeMore et a~l., 1985) recommends
k(ll)298 = 3.26 x 105 ppm^min-1.
This reaction rate 1s also uncertain to about 20 percent. Because (1)
reactions 10 and 11 are the only important reactions of 0( D), (2) are
strongly dependent on the 0(^0) formation rate, and (3) deplete any 0( D)
extremely rapidly, a steady-state concentration is easily established for
0(*D) under all daylight conditions. This steady state can be represented
equally as well in the ICRS if we lower k(10) and k(ll) by an identical
factor, thus minimizing some stiffness in the solution. In the ICRS,
therefore, both rates have been divided by 10 . The time constant for
decay via reactions 10 and 11 1s of the order of 10 minutes. Thus, a
significant change 1n 0(AD) in even 1 minute is still accurately simulated
to about 6 significant figures with this stiffness reducing factor of 10 .
The temperature dependence of the reactions of ozone with OH and H02,
03 -i- OH - > H02 (+ 02), (12)
and
03 + flOj --- > OH (+ 202), (13)
are taken from the NASA review (DeMore et al., 1985). For reaction 12,
k(12) = 2.344 x 103 exp(-940/T) pprn'^in'1,
and
k(12)298 = 1.00 x 102 ppm^min"1,
with an uncertainty of approximately 30 percent.
38
-------
For reaction 13,
k(13) = 2.100 x 101 exp(-580/T)
and
k(13)298 = 3.00 ppnf^min'1,
with an uncertainty of approximately 50 percent at 298 K.
The N03 radical undergoes photolysis in the visible region of the surface
solar spectrum; therefore, its j-value 1s determined through a ratio to
jNO (see the discussion of this rate at the end of this section). Two
possible channels occur:
N03 —hv—> N02 + 0,
and
N03 — hv—> NO + 02.
These have been combined into reaction 14:
N03 —hv—> 0.89 N02 + 0.89 0 + 0.11 NO (+ 0.11 02). (14)
The reaction of N03 with NO,
N03 + NO > 2N02> (15)
has been recently studied by two separate research groups and the accepted
rate constant has been increased. We have used the NASA rate (DeMore et
al., 1985), which yields
k(15) = 1.909 x 104 exp(250/T) ppm'^in"1,
and A 1 1
k(15)298 = 4.42 x 104 ppm^min-1.
In light of the values recently determined by Sander and Kircher (1986;
k(15J298 = 3.54 x 104 ppm^min'1) and Hammer et al. (1986; k(15)30Q = 4.36
x 10 ppm min ), we assume an uncertainty of 25 percent at 298 K.
The next four reactions (16 through 19) in the ICRS represent a set of NOX
reactions that often prove to be an important source of nitric acid
through the reaction of the equilibrium species dinitrogen pentoxide.
First, N02 and N03 can react in two ways:
N03 + N02 > NO + N02 (+ 02), (16)
and
N03 + N02 (+ M) > N205 (+ M). - (17)
39
-------
The rate expressions are
k(16) = 3.66 x 101 exp(-1230/T) ppn
with,
k(16)298 = 0.590 ppirtirT1 (Atkinson and Lloyd, 1984),
and ? 1 1
k(17) = 7.849 x 10^ exp(256/T) ppm'^in'1,
with,
k(17)29g = 1.853 x 103 ppnT-W-1 (DeMore et al., 1985).
Reaction 17 obviously appears to be more important. However, reaction 16
cannot be ruled out because of its somewhat unique reduction of N03 to
NO. For conditions of low NO, such as at night when 03 titrates all of
the NO to N02 and N03, reaction 16 is the only source of NO in the ICRS.
N20g chemistry consists of reactions 18 and 19:
N205 + H20 > 2 HN03, (18)
and
N205 (+ M) > N02 + N03 (+ M). (19)
Reaction 19 is the reverse of reaction 17; the combination of these reac-
tions represents the N20g equilibrium. The Ke_ value and its temperature
dependence are highly uncertain and the subject of much recent research.
The rate for reaction 19 is based on the K-- value of NASA (DeMore et al.,
1985) and the k(17) expression: ^
k(19) = k(17)/Keq,
where
k(17) = 5.346 x 10'13 exp(256/T) ciAolec-V1 ,
and
Keq = 1.52 x 10'27 exp(11153/T) cn^molec'1.
Then,
k(19) = 3.52 x 1014 exp(-10897/T) s'1,
or
= 2.11 x 1016 exp(-10897/T) miiT1,
with
k(19)298 = 2.78 min"1.
This value is more closely related to the Graham and Johnston (1978)
determination suggested by Perner et al. (1985).
40
-------
The value for k(18) is highly uncertain and the process may be heterogen-
ous. The value selected was recommended for atmospheric use by Atkinson
and Lloyd (1984):
k(18)298 = 1.9 x 10~6 ppm^min'1.
The termolecular self reaction of NO in the presence of oxygen,
NO -•- NO (+ 02) > 2 N02, (20)
1s generally insignificant in the atmosphere, but is included in the ICRS
to correctly model conditions of high NO. The rate constant is that of
Atkinson and Lloyd (1984):
k(20) = 2.60 x 10'5 exp(530/T) ppm'^in'1.
and
k(20)2g8 = 1.539 x 10~4 ppm^min"1,
with an uncertainty of about 25 percent.
The chemical cycle of nitrous acid is included in the ICRS as reactions 21
through 25:
NO + N02 + H20 > 2 MONO, (21)
OH + NO (+ M ) > MONO (+ M), (22)
HONO —hv--> OH + NO, (23)
OH + HONO > N02 (+ H20), (24)
HONO + HONO > NO + N02 (+ H20). (25)
The gas-phase equilibrium of HONO, N02, NO, and H20 is represented by
reactions 21 and 25; the values of k2j and k25 are taken from the study by
Kaiser and Wu (1977):
k(21) = 1.60 x 10'11 pprrf^in'1,
and «; 1 1
k(25) = 1.50 x 10'b ppm^min-1,
with an unknown temperature dependence. The predominant formation route
for nitrous acid is through the reaction of OH, NO, and a third body
(reaction 22). The high- and low-pressure fall-off parameters for this
reaction are from Atkinson and Lloyd (1984):
k0 = 6.7 x 10'31 (T/300)'3'3 ciAolecule'V1,
k = 3.0 x 10'11 (T/300)-1'0 cn^molecule'V1,
41
-------
which lead to an effective Arrhenius expression of
k(22) = 6.554 x 102 exp(806/T) ppm'^in'1,
and ill
k(22)29g = 9-80 x ICr ppnf Ani1n .
The photolysis rate of MONO (reaction 23) is represented as a ratio to the
j-value of N02. This reaction generally dominates as the loss mechanism
for MONO, though the reaction of OH and MONO (reaction 24) cannot be dis-
counted because of its possible role as a sink for OH. The temperature-
independent rate of that reaction (Atkinson and Lloyd, 1984) is estimated
at
k(24)298 = 9.77 x 103 ppm'^in'1,
with an uncertainty factor of 2.
Nitric acid formation and destruction through OH reactions are represented
in reactions 26 and 27:
OH + N02 (+ M) > HN03 (+ M). (26)
and
OH + HN03 (+ M) > N03 (+ H20 + M). (27)
The high- and low-pressure fall-off parameters for reaction 26 are those
of NASA (DeMore et al., 1985):
k0 = 2.6 x 10"30 (T/300)'3-2 cn^molecule'V1,
and
km = 2.4 x KT11 (T/300)'1-3 cn^molecule'V1,
which lead to an effective Arrhenius expression of
k(26) = 1.537 x 103 exp(713/T) pprn'^in'1,
and
k(26)298 = 1.68 x 104 ppm~*min~*.
For reaction 27, the pressure and temperature dependence were fit by com-
bining a low-pressure (bimolecular) limit with a Lindemann-Hinshelwood
expression for pressure dependence (DeMore et al., 1985):
42
-------
k3[M]
k(M,T) = k0 + —:
k3[M]
1 +
k.
where
k0 = 7.2 x 1CT15 exp(785/T).
k2 = 4.1 x 1CT16 exp(1440/T),
k3 = 1.9 x 1(T33 exp(725/T).
The Arrhenius expression fit to these data is
k(27) = 7.600 exp(1000/T) ppm^min'1,
with
k(27)298 * 2.18 x 102 ppm^min"1.
The bimolecular reaction of H02 and NO (reaction 28) is
H02 + NO > N02 + OH. (28)
The Arrhenius expression used is from NASA (DeMore et al., 1985):
k(28) = 5.482 x 103 exp(240/T) ppm'^in'1,
and aii
k(28)2g8 = 1.23 x HT ppnT^IrT1,
with an uncertainty at 298 K of approximately 20 percent.
Reactions 29 through 31 represent peroxynitric acid (PNA) chemistry in the
ICRS:
H02 + N02 (+ M) > PNA (+ M), (29)
PNA (+ M) > H02 + N02 (+ M), (30)
OH + PNA > N02 (+ H20 + 02). (31)
The forward rate of the PNA gas-phase equilibrium (reaction 29) is taken
from the fall-off parameters given in Atkinson and Lloyd (1984):
k0 = 2.3 x 10"31 (T/300)"4'6 oAiolecule'V1,
k, = 4.2 x 10'12 (T/300)"1"0'2 cn^molecule'V1,
43
-------
which lead to an effective Arrhenius expression of
k(29) = 1.64 x 102 exp(749/T) ppm^min'1,
and 311
k(29)2gg = 2.03 x 10 ppnf nmin .
Two alternate pathways have been suggested for these reactants:
H02 + N02 + H20 > PNA -i- H20,
and
H02 + N02 > MONO + 02.
The former pathway was studied by Sander and Peterson (1984), but it is
highly uncertain; the latter, if it occurs at all, apparently occurs at a
rate that is insignificant under all conditions of interest (Atkinson and
Lloyd, 1984). Therefore, neither reaction is included in the ICRS.
The rate for PNA decomposition (reaction 30) is based on the Keq value of
NASA (DeMore et al.t 1985) and the k2g expression:
k(30) = k(29)/Keq,
where
k(29) = 1.117 x 10~13 exp(749/T) ai^molecule'1*'1,
Keq = 2.33 x KT27 exp(10870/T) cm^olecule'1.
Then,
k(30) = 4.79 x 10iJ exp(-10121/T) s ,
or
= 2.88 x 1015 exp(-10121/T) min'1,
with
k(30)298 = 5.11 min'1.
Reaction 31, the OH-plus-PNA reaction, is generally unimportant compared
to PNA decomposition, but is included for low-temperature simulations.
The product distribution is not clear, but the one used here is often
given the highest possibility. Reaction rate information is taken from
NASA (DeMore et al., 1985), though both the rate and activation energy
values are highly uncertain:
k(31) = 1.909 x 103 exp(380/T) pprn'^in'1,
31
k(31)298 = 6'83 x 10 ppm
44
-------
Hydrogen peroxide chemistry is shown in reactions 32 through 35.
H02 + H02 > H202 (+ 02), (32)
H02 + H02 + H20 > H202 (-»- 02 + H20), (33)
H202 --hv--> 2 OH, (34)
OH + H202 > H02 (+ H20). (35)
The H02 self reaction is enhanced by the presence of water vapor; thus,
two reactions (32 and 33) are used to represent this process. ICRS rate
constants are from Atkinson and Lloyd (1984). At surface atmospheric
pressure,
k(32) = 8.739 x 101 exp(1150/T) ppm^min"1,
and
k(32)298 = 4.14 x 103 ppm^min'1.
For the reaction with water,
k(33) = 7.690 x 10'10 exp(5800/T) ppm^min'1,
and
k(33)2op = 2.18 x 10 ppm min .
The photoaction spectrum of H202 photolysis is near that of formaldehyde
photolysis (to stable products); therefore, the photolysis rate, k(34), is
determined by ratioing to jur-HO* at varyin9 zenith angles (as opposed to
The reaction of OH can be an important loss mechanism for H202 during some
daylight conditions. The reaction rate used in the ICRS is from NASA
(DeMore et al., 1985), which recommends
k(35) = 4.720 x 103 exp(-187/T) ppnT^IrT1.
and qii
k(35)298 = 2.52 x 10-3 ppm'^in'1.
The last reaction in the ICRS is the oxidation of CO by OH:
OH + CO > H02 (-1- C02). (36)
NASA (DeMore et al., 1985) recommends the formula:
k(36) = 1.5 x 10~13 (l+0.6Patm) cn^molecule'V1,
k(36) = 2.217 x 102 (l+0.6Patm) ppm'^in"1.
45
-------
At 640 m, the temperature-independent rate is approximately
k(36) = 3.22 x 102 ppm^min'1.
FORMALDEHYDE REACTIONS
Formaldehyde (CH2=0) is the simplest carbonyl species in polluted atmo-
spheres and is also a key organic oxidation product of almost all larger
organic species. Since it is capable of being both oxidized and photo-
lyzed and is therefore an important source of radicals in photochemical
systems, careful consideration must be given to the representation and
uncertainties of these chemical processes. In addition, because formal-
dehyde is considered a toxic species, we have chosen to explicitly repre-
sent its chemical reactions. As will be apparent in later discussions of
hydrocarbon representations, this choice entailed the removal of FORM as a
surrogate for some carbonyl species, in particular, glyoxal formed from
aromatics.
In the reaction with formaldehyde (designated FORM in the CBM), the
hydroxyl radical abstracts one of the hydrogens to form water, leaving the
formyl radical (0=CH):
FORM + -OH > 0=CH + H20
In the troposphere (at surface concentrations of 02) it is a good approxi-
mation to assume that the only reaction of formyl radical is:
0=CH + 02 > CO + H02*
These reactions are combined to yield
FORM + OH ( + 02) > CO + H02 (+ H20), (37)
with a rate constant of
k(37)298 = I5000 ppm min~*,
and an uncertainty of about 25 percent (DeMore et al., 1985).
Two formaldehyde photolysis reactions occur in the middle ultraviolet
region of the surface spectrum. Assuming that all product H* radicals
react exclusively with tropospheric 02, these are
46
-------
FORM —hv~> 2 H02 + CO, (38)
and
FORM —hv—> CO (-1- H2). (39)
The ratio of photolysis rates between j38 and j3g varies with the spectral
distribution of surface light (which changes with the zenith angle of the
sun). The calculation of these photolysis rates is discussed later in
this report. At present, we note that the rate of reaction 39 is always
greater than that of reaction 38 for solar irradiation. However, at low
zenith angles near solar noon, the relative importance of the radical-pro-
ducing reaction becomes greater than it is at higher zenith angles.
The reaction of 0(3P) with formaldehyde is usually unimportant, but we
include it because we already treat 0(3P) in the mechanism; thus, the
inclusion of this reaction is not a computational burden. In addition, it
is an organic source of *OH, since the 0 atom extracts a formaldehyde
hydrogen atom. Again, assuming all H* produced reacts exclusively with
02, the reaction is
FORM + 0 > OH + H02 + CO. (40)
The reaction rate used in the ICRS is that of NASA (DeMore et al., 1985):
k(40) = 4.302 x 104 exp(-1550/T) ppm^mlrT1,
and ? 1 1
k(40)298 = 2.37 x 10Z ppm'^iln'1.
The N03 radical also reacts with formaldehyde through the abstraction of a
hydrogen atom to produce nitric acid. Assuming all of the H* intermediate
reacts with atmospheric 02, the reaction is
FORM + N03 > HN03 + H02 + CO, (41)
with a rate constant used by Cantrell et al. (1985) for 298 K:
k(41)298 = 0.93 ppm'^in"1.
Although this rate is closely supported in the review by NASA [DeMore
et al. (1985) use a rate of 0.88 ppm^min'1], both groups suggest high
uncertainty limits. Those of NASA are 50 percent.
A final reaction scheme for formaldehyde that may be important under some
limited conditions is the addition (and later regeneration) of the H02'
radical. This reaction, and postulated subsequent reactions, were origi-
nally presented by Su et al. (1979a; 1979b). The overall scheme is
47
-------
HCHO + H02' < > HOOCH20'
HOOCH20' < > HOCH200'
HOCH200' + H02' > H02CH2OOH + 02
HOCH200' + NO > N02 + HOCH20'
HOCH20* (^ 02) > H02' + HCOOH
The initial equilibrium shows the presence of a transition intermediate
that appears capable of rearrangement to a more stable peroxy radical (Su
et al., 1979b) through a subsequent equilibrium. We have followed the
example of Veyret et al. (1982) and represent these two equilibria by
HCHO + H02' > HOCH200',
HOCH200' > HCHO + H02*.
where the peroxy radical results from the combined equilibria. We also
assume that the last reaction is rapid and is the only loss mechanism for
the HOCH20* radical. Hence,
HOCH200' + NO (+ 02) > N02 + H02' + HCOOH.
In our listing of the CBM-X, we use FROX to represent the peroxy radical
and PROX to represent total non-H202 peroxides. The overall scheme is
FORM + H02 > FROX, (X-42)
FROX > FORM + H02, (X-43)
FROX + H02 > PROX, (X-44)
FROX + NO > N02 + H02 + FACD. (X-45)
The rate of the initial reaction has been reported by a number of
researchers as 14.8 ppm~^min~^ (Su et al, 1979a; 1979b), 110.9 ppm~^min~^
(Veyret et al., 1982), and 162.6 pprn'^in'1 (Barnes et al., 1985). The
back reaction to the original products was determined to be 90 min"* (Su
et al, 1979a; 1979b). and 1800 min'1 (Veyret et al., 1982). These are
relatively rapid reverse rates, and appear to force the equilibria to the
original products under all conditions. This is especially true if one
notes that the reaction of H02 and NO (reaction 28) is fast, resulting in
little HOo for high-NO conditions and a relatively ineffectual reaction of
NO with FROX for low-NO conditions. We use the reaction rates of Su et
al. (1979b) for the equilibrium reactions in the CBM-X; the remaining two
rates are
48
-------
k(X-44) = 9.60 x 103 CH3C- + 'OH.
In the CBM, we assume that the peroxyacetyl radical forms rapidly (exclu
sively) under tropospherlc conditions through reaction with 02:
CH3C- + 02 ------ > CH3COO-
Combining these reactions results 1n
CH3CH + 0 (+02)--> CH3COO- + -OH. (42)
49
-------
The reaction rate used in the ICRS is derived from Atkinson and Lloyd
(1984):
k(42) = 1.739 x 104 exp(-986/T)
and
k(42)298 = 6.36 x
The reaction of the hydroxyl radical with acetaldehyde is also expected to
proceed by abstraction of the carbonyl hydrogen atom to form water.
Again, by assuming that only C203 is formed from the acetyl radical, the
CBM representation of the reaction is
CH3CH + OH ------ > CH3COO- (+ H20). (43)
The kinetic expression we have used in the past is very similar to that
recommended by Atkinson (1985). We therefore use
k(43) = 1.037 x 104 exp(250/T) pprn'^in'1,
and
k(43)298 = 2.40 x 104 ppm-1 min-1.
As we note in our discussion of the CBM-IV condensation steps, the reac-
tion with OH is a key oxidation reaction for higher molecular weight alde-
hydes, and acetaldehyde (designated ALD2) is used to describe the chemis-
try of the aldehyde functional group for higher molecular weight alde-
hydes.
Acetaldehyde reaction with N03 radical is expected to proceed in a manner
analogous to these reactions, following the scheme
0 0
II II
CH3CH + N03 ------ > CH3COO' + HN03. (44)
We derive our rate constant for this reaction on an estimate by Kerr and
Calvert (1985) of
k(44)2g8 = 3.70 ppm^min'1.
This reaction is often unimportant in the daytime, when N03 concentrations
are usually low. However, in nighttime situations of high ozone and NOX,
reaction 44 can result in the elimination of acetaldehyde and the produc-
tion of PAN.
50
-------
The photolysis of acetaldehyde occurs at wavelengths near the surface cut-
off of ultraviolet light in a region similar to ozone photolysis to 0(1D).
The three primary processes under tropospheric conditions (after reaction
of initial products with 02) are
0
II
CH3CH~hv—> CH300' + H02 + CO, (I)
—hv--> CH4 + CO, (II)
0
II
__hv--> CH3COO' + H02. (Ill)
On the basis of the quantum yields reported by Horowitz and Calvert
(1982), we assume that only photodissociation channel I is important for
conditions common in the lower troposphere. Therefore, we use
ALD2 —hv—> ME02 + H02 + CO, (X-49)
where CH300' is designated ME02. The photolysis of acetaldehyde is also
included in the CBM-IV, but as we explain later in this section, a univer-
sal peroxy operator (X02), FORM and H02 are used to represent ME02 in the
condensed mechanism.
The calculation of the acetaldehyde photolysis rate is discussed later in
this section. We note here, however, that though the hydroxyl radical
reaction rate constant is larger for acetaldehyde (and higher molecular
weight aldehydes) than for formaldehyde, the opposite is true for photo-
lysis rates. Hence, the hydroxyl radical reaction, which forms C203 pro-
ducts, is relatively more important than photolysis (not yielding C203)
for higher molecular weight aldehydes than it is for formaldehyde.
As noted, the chemistry of PAN (and higher molecular weight, PAN-like com-
pounds) must be accurately described to ensure realistic simulation of
radical dynamics when NOX and C203 or PAN are present. We know from
experience that good representations of the PAN chemistry almost always
lead to better simulation of the relationships among ozone, organics, and
NOX, both temporally and with respect to measured concentrations. There-
fore, one of the goals of this CBM evaluation was to examine and, if pos-
sible, enhance the PAN reaction set. The following discussion describes
the formulation of our new reaction scheme. As a'starting point, we
51
-------
examined the scientific literature for sources of basic kinetic informa-
tion and to determine the degree of uncertainty associated with that
information. We were also aware that the smog chamber data from both UNC
and UCR comprised key information thus far not integrated with the basic
data in any systematic way. Therefore, we also discuss an evaluation of
that smog chamber data in combination with the published information to
provide a more complete description of the PAN and C203 reaction scheme.
As noted, the decomposition of PAN,
PAN > C203 + N02, (48)
is strongly temperature-dependent. We utilize the kinetic information
reported by Atkinson and Lloyd (1984) to derive
k(48) = 5.616 x 1018 exp(-14000/T) min'1,
or
k(48)298 = 2.22 x 10~z min'1.
Another critical aspect of the association between PAN and ozone chemis-
try, however, is the strong dependence upon the ratio of the C203 reac-
tions with NO and N02:
C203 + NO > N02 + ME02, (X-50)
and
C203 + N02 > PAN. (X-51)
Examination of the UNC data set revealed three acetaldehyde-NOx experi-
ments conducted during the fall and winter of 1977 that showed good evi-
dence of the temperature-dependent relationship between competing peroxy-
acetyl reactions X-50 and X-51. Previous attempts to simulate these
experimental data (see, for example, Whitten et al., 1980b) resulted in
either greatly overpredicted ozone, overpredicted PAN, or both. Whitten
et al. (1983) attempted to improve these simulations by reducing acetal-
dehyde photolysis at low temperatures; this procedure improved the fits
for ozone, but yielded poor results for PAN.
For the experimental conditions of these experiments (low temperature and
a high ratio of acetaldehyde to NOX), PAN formation is the major radical
termination process in the photochemically reactive system. This is
because the PAN decay rate is small, while the availabilty of OH for the
key OH-plus-N02 reaction is suppressed by the presence of excess acetalde-
hyde. Thus, PAN production at low temperatures is strongly sensitive to
total radical input from any source (chamber radical production should be
less significant at low temperatures, but even where chamber sources are
significant, their existence would not alter such a sink phenomenon).
52
-------
For the three low- temperature acetaldehyde-NOx experiments in the UNC
chamber, we found that the production of ozone, and the total conversion
of NO to N02, was strongly related to the ratio of C203 reaction with NO
and N02. In addition, the ratio of C203 reaction with NO and N02 over the
temperature range of 266 to 290 K seemed to be significantly higher than
that for experiments at 300 to 315 K. A probable cause of a variation in
this ratio is a negative activation energy for the reaction of C203 plus
N02 (reaction X-51), since this reaction 1s known to be pressure-dependent
(Basco and Parmar, 1987). (In the following discussion and in Tables 1-2
and 1-3, note that a negative activation energy is listed as a positive
exponent since the exponent is the value of -E/R.) In the following
analysis, we describe the changes in the ratio of k(X-50)/k(X-51) with
temperature in terms of negative activation energy for reaction X-51.
However, for negative activation energies, the Arrhenius formulation must
break down at some point of temperature reduction because it will predict
an infinitely increasing rate constant.
The best simultaneous fits of PAN and ozone for the high-temperature (298
to 312 K) acetaldehyde-NOx and biacetyl-NOx experiments of 8 August 1980
and the two low- temperature acetaldehyde-NOx experiments (UNCB-122677 and
UNCB-111978) and were achieved using
k(X-50) = 7.915 x 103 exp(250/T) pprn'^in'1,
and /iii
k(X-50)298 = 1.83 x 10* ppnf -^lin'1,
for the reaction of C203 with NO (reaction X-50). The corresponding best-
fit Arrhenius expression for the reaction of C203 with N02 (reaction X-51)
is
k(X-51) = 1.180 x 10'4 exp(5500/T) ppm^mlir1.
and /iii
k(X-51)298 = 1.22 x 10* ppnf -hnin-1.
The results of these simulations are shown in Figures 3-1 through 3-4.
For the high-temperature experiments (Figures 3-1 and 3-2), the PAN pre-
dictions are somewhat higher than reported concentrations, but the decay
rates are correctly simulated. The measured PAN sometimes exceeds the
measured NOX (generally assumed to be N02 plus PAN) and we suggest that
the PAN calibration factors may have been high on that date. Since the
modeled N02 plus PAN for both experiments is nearly exact, we consider
this simulation to be as close as the data allow. We also note that for
the acetaldehyde-NOx experiment of 8 August 1980, .it was necessary to use
20 ppb of MONO as an initial condition to Initiate the early morning
53
-------
NO, N02 and 03
0.1
200
400
•00
•00
BIAC, FORM and PAN
too
400
•00
•00
FIGURE 3-1. Simulation results and experimental data for the UNC
b1acetyl-NOx experiment of B August 1980 (lines with many symbols are
simulation results).
-------
NO, N02 and 03
BOO
BIAC, FORM and PAN
•oo
•oo
FIGURE 3-2. Simulation results and experimental data for the UNC
acetaldehyde-NOx experiment of 8 August 1980 (lines with many symbols are
simulation results).
55
-------
OJS
0.3 -
0.25 -
0.2 -
0.16 -
O.1 -
OflS -
NO, PAN, N02 (plus PAN)
UNC8 ACOALDEHYDE
•00
O.O4S
CODS -
OZONE
CORRECTED FOR 13 SEC UNE DELAY
200
400
•00
MINUTES
FIGURE 3-3. Simulation results and experimental data for the UNC
acetaldehyde-NOx experiment of 26 December 1977 (symbols are data)
56
-------
NO, PAN, N02 (plus PAN)
eoo
OZONE
0.2
0.1* -
o.ie -
0.17 -
oie -
O.15 -
0.14 -
0.13 -
0.12 -
O.11 -
O.I -
oat -
oat -
0.07 -
oat -
O.05 -
0.04 -
0.03 -
0112 -
oa\ -
o
200
4OO
MINUTES
•00
•00
FIGURE 3-4. Simulation results and experimental data for the UNC
acetaldehyde-NOx experiment of 19 November 1977 (symbols are data)
57
-------
reactivity. This may indicate that the current estimates of in-chamber
acetaldehyde photolysis rate variation with zenith angle results in too
low a radical production rate at high zenith angles (reactivity for high-
zenith-angle fall and winter experiments is often underpredicted with the
current photolysis functions); however, we lack sufficient information
about early morning chamber characteristics to make a judgement at this
time. In any case, this does not affect the overall evaluation of simula-
tions, nor the implications for the k(X-50)/k(X-51) variation.
The low-temperature experiments (Figures 3-3 and 3-4) also fit well with
these rate expressions. Experimental and simulation results from a third
low-temperature experiment of 20 November 1977 (Figure 3-5) were somewhat
underpredicted by the mechanism (Figure 3-5a), while a better fit can be
obtained with the same k(X-51) activation energy and an Arrhenius pre-
factor that is 32 percent lower (Figure 3-5b). We take this as one indi-
cation of the uncertainty of these results. We also note that the NO
decay in all three low-temperature simulations tends to lag behind the
observed values, which might indicate an underestimation of k(X-50)/k(X-
51). Fitting the NO decay rate invariably results in overestimation of
ozone, however. Therefore, there appears to be either an inconsistency in
the simulation input or experimental data or another process at work that
we have not yet considered. [The low ozone simulations of 26 December and
20 November 1977 have been corrected for a 13-second sample line delay in
the UNC ozone monitoring system (Jeffries, 1987); the magnitude of this
correction is noted in Figure 3-5.]
There have been few previous estimates of the ratio of C203 reaction with
NO and N02. Atkinson and Lloyd (1984) based their recommendation of 1.5
on the 1979 modeling study of Carter et al., which derived the ratio from
a single smog chamber experiment (UCR EC-164). During the course of this
experiment, temperature control was apparently lost and the temperature
increased to 312 K, with a relative humidity of about 75 percent (nearly
40000 ppm of H20). Under conditions of elevated temperature and humidity,
the UCR evacuable chamber has been shown to desorb large amounts of con-
taminants (Carter et al., 1982), so any conclusions drawn from modeling
this experiment might be suspect. Moreover, the ratio of k(X-50)/k(X-51)
derived from this experiment is directly contradicted by other experiments
in the same chamber that suggest a ratio of 2.06 (see Figures 3-6 and 3-7
for EC-254 and EC-253). Chamber contamination is also important in these
experiments, especially for EC-253, since no NOX was added in this experi-
ment and thus the only source of NOX for PAN formation was chamber desorp-
tion and initial NOX bypassing the clean air system. The amount of NOX
necessary to explain the PAN formation in EC-253 is about 20 percent
greater than the radical source derived from the study of Carter et al.
(1982). Thus the "unknown chamber radical source11 may primarily be
MONO. For these UCR EC simulations we used the following chamber reac-
tions:
58
-------
NO. PAN. N02 (plus PAN)
NO. PAN. N02 (plu* PAN)
OZONE
OZONE
<£>
(a)
(b)
FIGURE 3-5. Simulation results and experimental data for the UNC acetaldehyde-NOx experiment of 20 November
1977 (symbols are data), (a). Simulation based on rates 1n Table 1-2. (b) Best fit simulation using a
k(X-51) Arrhenlus pre-factor reduced by 32 percent.
-------
EC-254
NO. N02, AND OS
XX*XV^XXXXXXXXXX X X
400
PAN
4OO
FIGURE 3-6. Simulation results and experimental data for
the UCR acetaldehyde-NOy experiment EC-254 (symbols are
data).
60
-------
0.15
EC253
OZONE
400
0.045
EC253
PAN
004 •
0 035 -
0.03 -
O.D25 -
OD2 -
O.016 -
OD1 -
0.006 -
0 H
100
200
MINUTES
300
400
FIGURE 3-7. Simulation results and experimental data for the UCR
acetaldehyde-NOx experiment EC-253 (symbols are data).
61
-------
(wall) > MONO k = 0.00039 x k(l),
(wall) > NO k = 0.00008 x k(l),
N02 > MONO k = 0.00137 x k(l).
NQ2 > NO k = 0.00040 x k(l),
NQ2 > wall k = 0.00230 x k(l),
03 > k = 0.00180.
There is an observed decrease -in the PAN production rate after 200 minutes
in both of these experiments, which suggests that both the NOX and radical
source from the chamber walls is finite and diminishes after this time.
Since the data reported in Carter et al. (1982) were all taken from
periods shorter than 200 minutes, the finite nature of the wall source
might not have been observed. This phenomenon may have important implica-
tions in chamber experiments involving longer reaction times.
The other available estimates for kX-50/kX-51 are from Cox and Roffey
(1977) and Cox et al. (1975). Although Cox and Roffey judged that their
data indicated no obvious temperature dependence, we can see from Figure
3-8 that this is due to data scatter as much as to any other factor. Two
regression lines are plotted in Figure 3-8a. The line with the smaller
slope (indicitave of an activation energy (E/R) of about -1500 K) is
associated with a very poor correlation coefficient (r = 0.21). However,
the highest temperature point (T = 328 K) is an outlier; removal of this
point gives the second line a correlation coefficient of 0.56 and indi-
cates an activation energy of -3000 K. Since the data of Cox and Roffey
were obtained by adding NO to higher concentrations of PAN, it is possible
that features of PAN decay other than NO and N02 competition for C203,
especially radical-radical reactions, resulted in a loss of oxidizing
potential (i.e., PAN decay can occur without the oxidation of NO),
especially at high temperatures. This oxidation loss would then produce
a falloff of the NO conversion rate at increasing temperatures which is
consistent with the data of Cox and Roffey.
Figure 3-8b shows the full data set available to us including our esti-
mates of k(X-50)/k(X-51) for the beginning and ending temperatures of the
UNC low-temperature acetaldehyde experiments. The lower slope line is the
least squares estimate based on the entire data set, excluding the highest
temperature point for Cox and Roffey. The slope indicates an activation
energy of -4000 K. The higher slope line uses our best estimate of the
temperature dependence of reaction X-51. Given the probability of high
temperature rolloff in the Cox and Roffey data, we plan to use this esti-
mate until more data become available.
62
-------
TEMPERATURE DEPENDENCE OF K46/K47
1.2
FROU COX AND RDFFET (1B77)
1 -
0.7 -
O.6 -
0.6-
0.4 -
0.3-
O.2 -
0.1
IDE 326K
DATAF1T
3.04 3.06 3.12 3.16 32 324 326 3.32
1000/K
(a) using the data from Cox and Roffey (1977)
3.36
1.9
1.6-
1.4 -
PEROXYACETYL REACTION
TEMPERATURE DEPENDENCE OF K46/K47
1
OB
O.6
O.4
0.2
-O.2 -
-O.4 -
-O.6-
-O.6 -
-1 -
-1.2-
—1.4 -
-1.C
3,4
10DO/K
ZJt
(b) using all available data as described in the text.
FIGURE 3-8. Temperature dependent functions for the peroxyacetyl
radical reactions with NO and NCU: (a) using the data from Cox and
Roffey (1977); (b) using all available data [DCox and Roffey (1977),
+ Cox et al. (1981), OEC-253 and EC-254, A UNC low-temperature
-------
At this point, we must also describe our evaluation of the radical-radical
reactions important in the C203 reaction scheme. In this and earlier
studies (Whitten et al., 1983; Killus and Whitten, 1984) we noted the
importance of radical-radical reactions in the complete description of PAN
decay. As shown in Figures 3-1 through 3-5, changing PAN decay rates are
frequently observed in the UNC chamber as the temperature Increases during
the day. The accurate simulation of PAN decay through the changing condi-
tions of a smog chamber experiment requires careful consideration of radi-
cal recombination rates. In the CBM-X, C203 plus radical reactions are
represented by
C203 + ME02 > ME02 + MEO, (X-65)
C203 + C203 > 2 ME02, (X-66)
C203 + H02 > PROX, (X-69)
C203 + H02 > ME02 + OH. (X-70)
The rates for the first two reactions are based on the results of Addison
et al. (1980):
k(X-65) = 4.40 x 103 ppm'^in'1,
and
k(X-66) = 3.70 x 103 ppm'^in'1,
independent of temperature.
The methylperoxy (ME02) plus H02 reactions analogous to these H02 reac-
tions with C203 are
ME02 + H02 > PROX, (X-67)
and
ME02 + H02 > MEO + OH. (X-68)
These four H02 reactions are very important in the simulation of oxidant
chemistry during the afternoon because H02 generally occurs at concentra-
tions higher than those of any individual organic radicals. For several
years, we have used rate constants for the C203 reactions based on the
results of Addison et al. (1980):
k(X-69) = 2.00 x 103
and
k(X-70) = 7.60 x 103
64
-------
We also use the rates determined by Cox and Tyndall (1980) for the ME02
reactions with H02:
k(X-67) = 2.550 x 101 exp(1300/T) ppm^min'1,
with
k(X-67)298 = 2.00 x 103 ppm^mln"1,
and
k(X-68) = 8.541 x 101 exp(1300/T) ppnf ^in'1,
with
k(X-68)298 = 6.70 x 103 ppnf ^lin'1.
We now consider the combined rates of the radical loss reactions (k(X-67)
+ K(X-68) and k(X-69) + k(X-70). These values are larger than some other
estimates (e.g., Kan et al., 1980; Niki et al., 1982) by up to a factor of
five. However, the C203 plus H02 study of Addison et al., and the ME02
plus H02 study of Cox and Tyndall were the only investigations that
directly measured the actual rates of peroxyacetyl and methylperoxy radi-
cal decay. All other studies reviewed inferred the radical-radical reac-
tion rates from observed products such as organic peroxides. Thus, those
studies could have suffered from underprediction of the reaction rate if
the observed products were not representative of the entire product yield.
Additional direct measurements have recently become available for methyl-
peroxy (ME02) and analogous ethylperoxy radical reactions with H02 (McAdam
et al., 1987 and Cattell et al., 1986). These data also support the
higher rates based on earlier radical decay studies, and indicate that the
discrepancies between the values obtained from observation of the radical
decays and observation of product (peroxide) yields could indicate the
existence of alternate product pathways for these reactions. We suggest
that the observed peroxide products could form at the lower rates
determined from observation of those products, whereas the differences
between the direct and product rates could be due to undetected dispropor-
tionation reactions:
C203 + H02 ------ > ME02 + OH (+ 02 + C02),
ME02 + H02 ------ > MEO + OH (+ 02),
ET02 + H02 ------ > ETO + OH (+ 02).
Our simulations suggest that the failure to represent such reactions in
the PAN reaction mechanism is responsible for the general failure of many
current kinetic mechanisms to adequately account for hydrocarbon decay
subsequent to the ozone peak, as NO concentrations become very low (see
for example, Leone and Seinfeld, 1984, Figure 8; or Atkinson et al.,
1982). Under such conditions, the production of OH by
65
-------
NO + H02 ------ > N02 + OH (28)
is limited, and the other major source of OH at the time 1s the reaction
03 + H02 ------ > OH (-(-2 02). (13)
The rate constant for this reaction has been recently remeasured and is
apparently more than a factor of four too low to account for the observed,
post-ozone-maximum hydrocarbon losses. Other possible sources of OH would
include ozone photolysis to 0(*D) and chamber wall emissions of HONO. In
smog chamber simulations, both of these sources are an order of magnitude
too low to be responsible for the "missing OH source" and would, moreover,
have a major effect upon the simulations prior to the ozone peak.
If we include the postulated disproportlonatlon reactions in the mechan-
ism, we obtain excellent fits to post ozone peak hydrocarbon decays for a
number of different hydrocarbon types. An example of this improvement is
shown in Figure 3-9 for a typical UCR-EC m-xylene trace. This example and
others using the CBM-IV condensed mechanism can be found throughout the
plots given in Section 6. The effect on post-ozone hydrocarbon decay is
usually most visible for aromatic hydrocarbons. However, the "missing OH
source" is not a special feature of aromatic oxidation, but 1s more
clearly shown in that system because (1) aromatics have high rates of
reaction with OH and (2) they have low rates with competing oxidizers such
as ozone (the former being necessary to distinguish OH loss from chamber
dilution and the latter being a confounding effect). The effect appears
directly related to PAN levels and decay and applies across a variety of
hydrocarbon mixes and reactivity levels.
We briefly discuss the additional ME02 and MEO reactions in the CBM-X
next. The methylperoxynitric acid (MPNA) equilibrium reactions are expec-
ted to be analogous to those of PNA (reactions 29 and 30):
ME02 + N02 (+ M) ------ > MPNA (+ M), (X-53)
and
MPNA (+ M) ------ > ME02 + N02 (+ M) . (X-54)
The rate of MPNA formation is taken from the falloff parameters given by
NASA (DeMore et al., 1985):
k0 = 1.5 x HT30 (T/300)-4-0
and
= 6.5 x lO'12 (T/300)-2'0
66
-------
O.4 -
0 J -
O.2 -
0.1 -
100 200
Vim*
*00
400
(a) without radical disproportionatlon products.
O.4 -
O J -
O.1 -
100
SOD
soo
4OO
(b) with radical disproportlnatlon products.
FIGURE 3-9. m-Xylene concentration traces for UCR EC-345 comparing the
effects of Including radical disproportionate products 1n the H02 plus
R02 reactions.
67
-------
which lead to an Arrhenius expression of
k(X-53) = 3.810 x 102 exp(821/T)
and Til
k(X-53)298 = 5.99 x 1CT ppm^min .
The rate of decomposition of MPNA is also based on the equilibrium con-
stant data of NASA (DeMore et al., 1985):
k(X-54) = k(X-53)/keq,
where
k(X-53) = 2.583 x 10'13 exp(821/T) cn^molecule'V1,
and
k = 1.30 x 10'28 exp(11192/T) cm^molecule'1,
then M
k(X-54) = 1.99 x 1015 exp(-10371/T) s"1,
or
k(X-54) = 1.19 x 1017 exp(-10371/T) min'1,
with
k(X-54)298 = 9.15 x 102 min'1.
The additional reactions of ME02 and MEO (reactions X-55 through X-64) are
generally noncompetitive channels that ultimately yield a distribution of
stable products and NO-to-NO? conversions. These reactions are briefly
described next. The reader is referred to the review of Atkinson and
Lloyd (1984) and the kinetic compilations of NASA (DeMore et al., 1985)
and CODATA (DeMore et al., 1985) for a more complete description of the
formulation and kinetics of these reactions.
The remaining reactions of ME02 (recall that the ME02 plus H02 reactions
were discussed previously) are
ME02 + NO > MEO + N02, (X-55)
ME02 + ME02 > 2 MEO, (X-63)
ME02 + ME02 > FORM + MEOH, (X-64)
where MEOH is methanol and MEO is the methoxy radical (CH3*). The kinetic
expression for reaction X-55 is from NASA (DeMore et al. 1985):
k(X-55) = 6.140 x 103 exp(180/T) pprn'^in'1,
and
k(X-55)298 = 1.12 x 104 ppm^min'1.
68
-------
The rate for the self reaction of ME02 was taken from CODATA (Baulch et
al., 1984); the ratio between the two product pathways (k(X-63) and k(X-
64)] is also taken from that review. Thus,
k(X-63) = 1.91 x 102 pprn'^in'1,
and
k(X-64) = 3.56 x 102 ppm'^in'1.
The reactions of MEO are
MEO + NO > MNIT, (X-56)
MEO + NO > FORM + H02 + NO, (X-57)
MEO + N02 > MEN3, (X-58)
MEO (+ 02) > FORM + H02. (X-59)
The initial three kinetic expressions are taken from Atkinson and Lloyd
(1984); their rate for methylnitrite (MNIT) formation is
k(X-56) = 2.296 x 104 exp(200/T) ppm'^in'1,
and
k(X-56)298 = 4.44 x 104 ppnf •hnlrT1.
The rate for the alternate reaction products (reaction X-57) is
k(X-57) = 1.92 x 103 ppm^min-1.
The reaction rate constant for reaction with N02 is
k(X-58) = 2.22 x 104 ppm'^in'1.
The recent NASA evaluation (DeMore et al., 1985) uses the temperature-
dependent kinetic expression for reaction X-59 of
k(X-59) = 2.605 x 107 exp(-1200/T) min'1 ,
and .
k(X-59)298 = 4.64 x 10D min A.
Finally, methyl nitrate and nitrite losses are given in reactions X-60
through X-62:
MEN3 + OH > FORM + N02, (X-60)
MNIT + OH > FORM + NO, (X-61)
MNIT ~hv--> MEO + NO. . (X-62)
69
-------
The reaction of methyl nitrate with the hydroxyl radical is relatively
slow for a hydroxyl radical rate constant, as are the per-carbon reaction
rate constants of higher molecular weight alkyl nitrates (Atkinson et al.,
1982). For k(X-60), we use (Atkinson and Lloyd, 1984)
k(X-60) = 4.00 x 102 ppm'^in'1.
The reaction of methyl nitrite with OH (X-61) is relatively unimportant
compared with photolysis (X-62). As recommended by Atkinson and Lloyd, we
use
k(X-61) = 3.00 x 102 ppm'^in'1,
3-°° x 10 m1n" •
and
PHOTOLYSIS REACTIONS
Calculations of specific photolysis rates (j-values) are required because
the photolysis of key photoreceptor species provides radical products that
must be accurately described to correctly simulate both smog chamber data
sets and atmospheric conditions. Photolysis of stable molecules is the
major source of new radicals in the gas phase. The accurate determination
of the photolysis rates, especially early 1n the day when the "radical
pool" is limited, is critical in achieving the proper representation of
the balance with radical termination reactions. In this subsection we
describe the data and calculation methods used to determine in-chamber
photolysis rates. In addition, we compare atmospheric photolysis rates
calculated with our data and procedures with those recently determined by
Jeffries and Sexton (1987) at UNC. The UNC rates have been recommended
for use with the CBM-IV during atmospheric simulation with OZIPM-4 (Hogo
and Gery, 1988). We felt that comparison of these recently reported
values with data and methods used at Systems Applications would be useful
in understanding the ranges of uncertainty of different calculation
methods and data.
The j-value of a specific molecular process is defined at any given time
by the integral of the triple product, Io», over the wavelength interval
of interest,
xmax
Jn - / I(x)o(x)«(x)dx .
r max
J Xmin
where I is the actinic flux, o is the molecular absorption cross section,
and * is the quantum yield for a specific photolysis process (n). From
70
-------
this simple formulation, it is apparent that uncertainties in j-value cal-
culations result from two sources: (1) the measure of "in-chamber"
actinic flux and its spectral distribution and variation with time and
conditions, and (2) the uncertainties associated with the experimental
determination of the absorption cross section and quantum yields. The
actinic flux calculations have been a source of controversy and uncer-
tainty because of the many variables that must be included in site-
specific descriptions. Although the UNC facility has the best characteri-
zation of any outdoor chamber, certain aspects have yet to be determined.
The most important of these are the changes in spectral distribution due
to attenuation and reflection by the Teflon film at different solar zenith
angles and the differences in cloud attenuation caused by different cloud
types. These remaining uncertainties are the object of experimental
studies at UNC and must currently be estimated for modeling purposes. The
values and assumptions we have used to determine the actinic flux within
the UNC chambers are described in Section 5.
The experimentally determined photolysis rate of N02 (JNQZ or kl) is o^60
used in photochemical kinetics models as the indicator of the amount of
light energy available at the given time of day being simulated. It must
be kept in mind, however, that NO^ absorbs light and photolyzes in a
specific (rather long wavelength) region compared to other key photorecep-
tors. As the zenith angle of the sun changes (and therefore, the spectral
distribution of the solar irradiation), the photolysis rates of the other
species will not directly follow the JNQ2 value. This is because the
effects of different atmospheric processes (e.g., molecular absorption and
scattering; albedo and aerosol scattering) vary with spectral distribu-
tion. Therefore, though the jN02 value can be used to track the relative
amount of energy available for photolysis throughout the day. another fac-
tor related to the changing spectral distribution must be considered so
that the absolute amount of energy available to species that absorb in
other spectral regions (such as ozone and formaldehyde) can be determined.
For this, we use the zenith angle of the solar disk as an indicator of the
spectral distribution. This results in a set of nonlinear ratios that can
be multipled by the experimental j^Q2 value for a given time of day to
determine the specific j-values at that time of day (and zenith angle).
If time-varying and wavelength-resolved actinic flux measurements become
available, it will be beneficial to directly calculate the j-values of all
important species. At present, however, these data are not available and
the ratio of the clear-sky relationships to experimental jNQ2 data is
required.
The ratios of j-values to jN02 can be determined by calculating absolute
photolysis rate constants for each photolysis process using a standard set
of actinic flux distributions for a range of zenith angles. In this case
the clear-sky, ground-level values calculated with the model of Demerjian
71
-------
i
c
0
c
I
A.
n
I
SEt-02
4E+02 -
4E+02 -
3E+02 -
Aciimc Flux for 20 and 60 Degree ZAs
(Sauro*: D«m«r|lan «t erf, 1960}
280
300
320
340
Wov»l«ngtt> (run)
360
400
420
FIGURE 3-10. Actinic flux for 20 and 60 degree solar zenith angles in bin
sizes of 1 and 5 nm. Areas under curves are equal. Source: Demerjian
et a!., 1980.
72
-------
TABLE 3-1. Absorption cross sections and quantum yields for NC^ photolysis;
N02 -«-* NO + 0
Cross Section
Wavelength (cm2/molec.)
(nm) (x l.E+20)
280
281
282
283
284
285
286
287
288
289
290
291
292
293
. 294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
5.54
5.58
5.36
5.36
6.25
6.99
7.29
7.37
7.66
7.89
8.18
9.90
9.37
9.75
9.48
9.68
9.30
12.17
11.72
12.57
11.72
12.28
13.87
15.92
15.96
16.56
15.81
16.33
16.18
18.38
17.56
18.78
19.61
20.35
19.42
Quantum
Yield
0.984
0.984
0.984
0.984
0.984
0.984
0.984
0.984
0.984
0.984
0.984
0.984
0.984
0.984
0.984
0.984
0.983
0.982
0.982
0.981
0.980
0.980
0.978
0.978
0.977
0.976
0.975
0.974
0.973
0.973
0.972
0.971
0.970
0.970
0.969
Wavelength
(nm)
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
Cross Section
(cm2/molec.)
(x l.E+20)
22.50
21.31
23.29
24.81
23.14
25.37
26.53
26.49
27.68
26.75
27.86
28.79
29.09
30.77
29.98
29.87
30.50
30.05
37.27
29.80
34.52
35.08
34.63
34.78
39.88
38.80
41.66
38.32
35.45
40.21
40.70
42.93
42.78
48.21
46.12
Quantum
Yield
0.968
0.967
0.966
0.966
0.965
0.964
0.963
0.962
0.962
0.961
0.960
0.959
0.958
0.958
0.957
0.956
0.955
0.954
0.954
0.953
0.952
0.951
0.950
0.950
0.949
0.948
0.947
0.946
0.946
0.945
0.944
0.943
0.942
0.942
0.941
(Continued)
73
-------
TABLE 3-1 (Concluded)
NO + 0
Wavelength
(nm)
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
Cross Section
(cm2/molec.) Quantum Wavelength
(x l.E+20) Yield (nm)
40.99
45.20
44.38
39.88
50.40
51.30
46.05
55.80
50.37
45.53
45.13
53.87
50.40
51.23
48.74
57.81
53.98
51.86
53.42
51.82
54.20
52.12
59.81
55.02
52.12
53.53
62.35
56.69
51.74
54.68
59.86
56.62
56.36
53.72
59.67
0.940
0.939
0.938
0.938
0.937
0.936
0.935
0.934
0.934
0.933
0.932
0.931
0.930
0.930
0.929
0.928
0.912
0.896
0.881
0.865
0.849
0.833
0.817
0.802
0.786
0.770
0.780
0.920
0.820
0.870
0.900
0.810
0.700
0.680
0.700
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
Cross Section
(cm2/molec.)
(x l.E+20)
59.41
53.20
56.02
59.78
60.23
60.01
58.29
60.49
54.54
55.42
58.89
61.45
56.65
64.06
56.03
67.59
65.25
57.10
51.04
60.67
63.17
53.90
47.28
62.61
59.00
57.73
58.78
53.65
70.04
59.41
60.41
48.47
53.12
55.17
52.79
57.72
Quantum
Yield
0.770
0.840
0.750
0.810
0.780
0.800
0.880
0.840
0.900
0.900
0.840
0.830
0.820
0.770
0.780
0.680
0.650
0.620
0.570
0.420
0.320
0.330
0.250
0.200
0.190
0.150
0.100
0.090
0.080
0.080
0.070
0.060
0.050
0.040
0.030
0.020
74
-------
TABLE 3-2. Results of JNQ2
calculations. (Albedo = 0.05)
Zenith jNQ2 (min )
Angle
(deg) SAI UNC
0 0.485 0.503
10 0.481 0.499
20 0.469 0.487
30 0.448 0.464
40 0.415 0.430
50 0.366 0.379
60 0.295 0.306
70 0.194 0.202
78 0.095 0.099
86 0.020 0.019
75
-------
Pholoaclion Specira for N02
3
«
O
«
C
O
n
"5
16.0
14.0 -
11.0 -
12.0 -
11.O -
10.0 -
9.0 -
8.0 -
7.0 -
«.0 -
8.0 -
4.0 -
3.0 -
2.0 -
1.0 -
0.0
(at Zan'th Anp<« of 20
280 300 320 340 360
Wuv.l«n9*» (nm)
380
400
420
FIGURE 3-11. Photoaction spectra for NO? using 1 nm bins (this report)
and 5 nm bins (Jeffries and Sexton, 1987).
76
-------
o
et al. (1980) were used. These data provide the number of photons cnfc
sec"1 in wavelength bins of 5 nm for ten zenith angles from 0 to 86
degrees. The calculations that produced these data assumed one specific
aerosol distribution, surface albedo (0.05), and ozone column density
(0.292 cm-atm). Because we perform our numerical integration of j-values
1n 1 nm bins, we have interpolated the larger bins of Demerjian and co-
workers to 1 nm values, verifying at all times that the sum of the energy
within five 1 nm bins is equal to the energy in the original 5 nm bin.
The spectral distributions at two zenith angles (20 and 60 degrees) for 1
nm and 5 nm bins are shown in Figure 3-10.
The photolysis rates for the process:
N02 ^ NO + 0 (1)
were calculated using the actinic fluxes of Demerjian et al. (1980) and
the absorption cross sections and quantum yields listed in Table 3-1.
These results are given in Table 3-2 and compared to values recently
determined by UNC (Jeffries and Sexton, 1987). The rates determined by
Systems Applications are between 2 and 4 percent lower than those of
UNC. In both cases, the data were taken from the NASA (DeMore et al.,
1985) review; however, the UNC calculations are based on 5 nm bin averages
of quantum yield and cross section, whereas our calculations were per-
formed at 1 nm. The resulting photoaction spectra (the product Io« versus
wavelength) are shown in Figure 3-11 for a 20° zenith angle. Our calcula-
tions using 1 nm and 5 nm bins show that the variation in methodologies
results in the minor j^ differences. Such a difference in JNQ£ is not a
significant deviation and is certainly bounded by the uncertainty of the
experimentally derived data.
The absorption cross sections and quantum yields used for formaldehyde,
ozone, acetaldehyde (ALD2), nitrous acid, and hydrogen peroxide are pre-
sented in Tables 3-3 through 3-8. In the following discussion, the pro-
cesses we refer to are
HCHO ^ 2H- + CO (38)
JHCHOS:
HCHO ^ H2 + CO (39)
hu
CH3CHO — H- + CO + CH300- (45)
77
-------
TABLE 3-3. Absorption cross sections and quantum yields for formaldehyde
photolysis to radical products:
HCHO ----*
Wavelength
(nm)
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
Cross Section
(cm2/molec.)
(x l.E+20)
2.34
1.65
0.76
0.46
3.93
3.46
2.32
0.95
2.32
2.50
1.43
1.32
0.66
5.22
4.30
3.21
1.59
1.96
3.66
1.55
0.72
1.51
0.74
4.35
4.79
4.94
3.02
1.16
2.18
2.25
Quantum
Yield
0.560
0.580
0.600
0.620
0.630
0.650
0.670
0.680
0.700
0.710
0.720
0.730
0.750
0.760
0.760
0.770
0.780
0.790
0.790
0.790
0.800
0.800
0.800
0.800
0.800
0.790
0.790
0.790
0.780
0.770
2H- + CO
Wavelength
(nm)
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
Cross Section
(cm2/molec.)
(x l.E+20)
1.03
0.81
1.49
1.55
3.99
2.88
2.79
3.59
1.65
0.73
1.71
1.32
0.43
0.60
0.75
2.19
3.44
1.75
1.01
3.03
1.96
0.79
0.32
0.15
0.17
0.02
0.17
0.32
1.93
2.15
1.07
Quantum
Yield
0.760
0.750
0.740
0.730
0.720
0.700
0.690
0.670
0.650
0.630
0.610
0.590
0.570
0.540
0.510
0.490
0.460
0.430
0.390
0.360
0.330
0.290
0.250
0.210
0.170
0.130
0.083
0.038
0.000
0.000
0.000
78
-------
TABLE 3-4. Absorption cross sections and quantum yields for formaldehyde
photolysis to stable products:
HCHO --^-* CO + H
Wavelength
(nm)
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
Cross Section Cross Section
(cm2/molec.) Quantum Wavelength (cm2/molec.)
(x l.E+20) Yield (nm) (x l.E+20)
2.34
1.65
0.76
0.46
3.93
3.46
2.32
0.95
2.32
2.50
1.43
1.32
0.66
5.22
4.30
3.21
1.59
1.96
3.66
1.55
0.72
1.51
0.74
4.35
4.79
4.94
3.02
1.16
2.18
2.25
1.03
0.81
1.49
1.55
3.99
2.88
2.79
3.59
1.65
0.73
0.440
0.420
0.400
0.380
0.370
0.350
0.330
0.320
0.300
0.290
0.280
0.270
0.250
0.240
0.240
0.230
0.220
0.210
0.210
0.210
0.200
0.200
0.200
0.200
0.200
0.210
0.210
0.210
0.220
0.230
0.240
0.250
0.260
0.270
0.280
0.300
0.310
0.330
0.350
0.370
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
1.71
1.32
0.43
0.60
0.75
2.19
3.44
1.75
1.01
3.03
1.96
0.79
0.32
0.15
0.17
0.02
0.17
0.32
1.93
2.15
1.07
0.31
0.94
1.37
0.57
0.12
0.04
0.04
0.07
0.03
0.03
0.09
0.90
1.17
0.72
0.26
0.05
0.03
0.04
0.03
Quantum
Yield
0.390
0.410
0.430
0.460
0.490
0.510
0.540
0.550
0.570
0.580
0.590
0.600
0.610
0.620
0.620
0.620
0.620
0.620
0.610
0.610
0.600
0.590
0.570
0.560
0.540
0.520
0.500
0.470
0.450
0.420
0.390
0.360
0.330
0.300
0.260
0.230
0.200
0.160
0.130
0.100
-------
TABLE 3-5. Absorption cross sections and quantum yields for ozone photolysis to
-tiiU 0(1D)
Wavelength
(nm)
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
Cross Section
(cm2/molec.)
(x l.E+20)
397.00
360.00
324.00
301.00
273.00
244.00
221.00
201.00
176.00
158.00
141.00
126.00
110.00
98.90
86.20
76.70
66.40
58.80
51.00
45.20
Quantum
Yield
0.900
0.900
0.900
0.900
0.900
0.900
0.900
0.900
0.900
0.900
0.900
0.900
0.900
0.900
0.900
0.900
0.900
0.900
0.900
0.900
Wavelength
(nm)
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
Cross Section
(cm2/molec.)
(x l.E+20)
39.20
34.40
30.30
26.30
23.50
20.20
18.00
15.60
13.60
12.30
10.30
9.27
8.00
6.92
6.29
5.22
4.78
4.04
3.72
2.91
Quantum
Yield
0.900
0.900
0.900
0.900
0.900
0.884
0.848
0.800
0.740
0.660
0.560
0.450
0.340
0.250
0.180
0.120
0.080
0.050
0.020
0.000
-------
TABLE 3-6. Absorption cross sections and quantum yields for acetaldehyde
photolysis to radical products:
Wavelength
(nm)
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
Cross Section
(cm2/molec.)
(x l.E+20)
4.50
4.54
4.58
4.62
4.66
4.70
4.74
4.78
4.82
4.86
4.90
4.82
4.74
4.66
4.58
4.50
4.46
4.42
4.38
4.34
4.30
4.12
3.94
3.76
3.58
ALD2 -te-~
Quantum
Yield
0.580
0.575
0.570
0.565
0.560
0.555
0.550
0.545
0.540
0.535
0.530
0.520
0.510
0.500
0.490
0.480
0.470
0.460
0.450
0.440
0.430
0.418
0.406
0.394
0.382
H- + CO + CH3
Wavelength
(nm)
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
*
Cross Section
(cm2/molec.)
(x l.E+20)
3.40
3.27
3.14
3.01
2.88
2.75
2.62
2.49
2.36
2.23
2.10
2.04
1.98
1.92
1.86
1.80
1.66
1.52
1.38
1.24
1.10
1.18
0.94
0.85
0.77
0.69
Quantum
Yield
0.370
0.350
0.330
0.310
0.290
0.270
0.250
0.230
0.210
0.190
0.170
0.156
0.142
0.128
0.114
0.100
0.088
0.076
0.064
0.052
0.040
0.032
0.024
0.016
0.008
0.000
81
-------
TABLE 3-7. Absorption cross sections and quantum yields for nitrous acid
photolysis:
MONO ----* OH + NO
Wavelength
(nm)
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
Cross Section Cross Section
(cm2/molec.) Quantum Wavelength (cm2/molec.)
(x l.E+20) Yield (nm) (x l.E+20)
2.09
2.00
1.90
1.81
1.71
1.62
1.54
1.46
1.39
1.31
1.23
1.17
1.11
1.05
0.99
0.93
0.89
0.84
0.80
0.75
0.71
0.67
0.64
0.61
0.57
0.54
0.51
0.49
0.47
0.44
0.42
0.40
0.38
0.36
0.34
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
0.32
0.30
0.29
0.27
0.26
0.25
0.23
0.22
0.21
0.20
0.19
0.18
0.17
0.16
0.15
0.14
0.14
0.13
0.12
0.12
0.11
0.11
0.10
0.09
0.09
0.08
0.08
0.08
0.07
0.07
0.06
0.06
0.06
0.05
0.05
0.05
Quantum
Yield
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
82
-------
TABLE 3-8. Absorption cross sections and quantum yields for hydrogen peroxide
photolyis:
Hn "^ o nu
~U0 >• e. DM
Wavelength
(nm)
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
Cross Section Cross Section
(cm2/molec.) Quantum Wavelength (cm2/molec.)
(x l.E+20) Yield (nm) (x l.E+20)
0.00
0.00
0.20
0.42
0.46
0.42
0.30
0.46
3.60
6.10
2.10
4.27
4.01
3.93
4.01
4.04
3.13
4.12
7.55
6.64
7.29
8.70
13.80
5.91
5.91
6.45
5.91
4.58
19.10
16.30
10.50
8.70
33.50
20.10
10.20
8.54
8.32
8.20
7.49
7.13
6.83
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
17.40
11.40
37.10
49.60
24.60
11.90
9.35
7.78
7.29
6.83
6.90
7.32
9.00
12.10
13.30
21.30
35.20
45.00
29.30
11.90
9.46
8.85
7.44
4.77
2.70
1.90
1.50
1.90
5.80
7.78
11.40
14.00
17.20
19.90
19.00
11.90
5.65
3.20
1.90
1.20
0.50
Quantum
Yield
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
-------
JOID:
0, -^ 0(1D) + 0, (9)
Both the absolute photolysis rates (calculated with the actinic flux of
Demerjian et al., 1980) and the zenith-angle-dependent ratios for these
j-values to jN02 are given in Table 3-9 to facilitate comparison of these
rates for different types of photochemical kinetics simulation models.
For instance, the Photochemical Kinetics Simulation System (PKSS)
developed by Jeffries (1986) allows input of absolute photolysis rates,
while the CHEMK program, which is used at Systems Applications and is the
basis of the OZIPP computer code, utilizes zenith-angle-dependent ratios
to jN02 to internally determine absolute rates. At present, this is pri-
marily a format difference. That is, even though absolute rates can be
input in PKSS, they are currently derived by the ratio-to-j^g? process
since there are no spectrally resolved experimental actinic flux measure-
ments that would allow direct calculation of the other j-values.
The formaldehyde photolysis rates and ratios in Table 3-9 are considerably
lower than the values recommended by Jeffries and Sexton (1987). Our
^HCHOr anc* JHCHOS rates are approximately 90 and 70 percent of the UNC-
determined values, and the differences are due almost completely to the
choices of quantum yield and absorption cross section data. Our quantum
yield data for both formaldehyde photolysis channels are 1 nm values that
were fit by Calvert (1980) through his data and that of Moortgat and co-
workers (1978, 1979). Ten nm averages of these data are approximately
zero to ten percent lower than the 10 nm values used by UNC. The two
major sources of absorption cross section data for formaldehyde are from
Bass (1980) and Moortgat (1986). We have communicated with both
researchers and obtained their data. As shown in Figure 3-12, the results
differ by more than thirty percent, especially at the longer wavelength
end. After considering both data sets for the UNC chamber simulations, we
found that use of the 0.05 nm resolved data of Bass to derive 1 nm bin
averages provided the in-chamber absolute rates needed to model measured
formaldehyde decay. This does not imply that the Bass data are more cor-
rect, but only that when combined with other chamber-dependent factors
these data allowed good simulation of simple formaldehyde systems. Com-
pared to our 1 nm data from Bass (1980) UNC utilized 10 nm bin values
given by NASA (DeMore et al., 1985), which were determined by averaging
the two sets of absorption cross section data. Figure 3-13 shows the
photoaction spectra for both formaldehyde photolysis channels using our
1 nm values, the UNC 10 nm bins, and a 5 nm light source (the amount of
light was equal for both photoaction spectra). The noted differences in
j-values above result almost completely from the differences in quantum
yield and absorption cross section data; the significant differences
84
-------
TABLE 3-9. Results of j-value calculations.
Absolute Photolysis Rates (j-values)
Zenith
Angle (xlOOO, min , no chamber corrections)
(deg)
0
10
20
30
40
50
60
70
78
86
Ratios to Jj
Zenith
Angle
(deg)
0
10
20
30
40
50
60
70
78
86
HCHOr
1.575
1.552
1.470
1.330
1.135
0.885
0.592
0.295
0.111
0.017
-------
£
o
I • 111 • •' I—"'"'•'"••' "I ... I,,.. I ... I .. I
270 283 250 JOC 310
2 -
1 -
rsi
I I I I I I I I I I I I I I I I I
*700 MOO WOO XOO 3100 MOO 3X>0 ' >40O 9SOO
WAVELENGTH. A
FIGURE 3-12. Comparison of absorption cross sections for
formaldehyde from Moortgat (top) and Bass (bottom).
86
-------
O£8
Pholoaciion Speclra HCHOs
-------
between the Bass (1980) and Moortgat (1986) cross section data lead to the
larger variation for JHCHOS* since the discrepancy between these data is
mainly towards the longer wavelength region.
Our determination of JQ^Q is approximately 10 percent higher than that
recommended by UNC, again resulting from the use of slightly different
data sources. Our absorption cross section data were calculated by
obtaining 1 nm averages of high-resolution data provided by Dr. Arnold
Bass of the National Bureau of Standards (1985). Quantum yield data were
obtained from Atkinson and Lloyd (1984). Our acetaldehyde j-values differ
significantly from those recommended by UNC (our values are approximately
20 percent larger). Again, this indicates the range of variability in
data sets. Our j/\|_g2r va^ues were derived from the quantum yields and
absorption cross sections reported in the latest CODATA review (Baulch et
al., 1984). UNC used values from Carter et al. (1986), who utilized the
quantum yields listed in Atkinson and Lloyd (1984) and absorption cross
sections from Calvert and Pitts (1966). Figure 3-14 shows the differences
in photoaction spectra resulting from these different values.
These j-values, namely jN02, JHCHOr- %HOs» JQ1D and JALD2* represent
important, short wavelength tropospheric photolysis reactions. Each pro-
cess responds somewhat differently to diurnal changes in solar spectra (as
represented by changes in zenith angle), depending on the proximity of
their absorption cross section to the more highly attenuated blue side of
the solar spectrum. Figure 3-15 shows the differences in j-value curve
shapes, and the need for model input of either direct j-values or zenith-
angle-dependent ratios to J^- To illustrate the resulting differences,
Figure 3-16 shows photoaction spectra, JQ^Q and JNQZ* at solar zenith
angles of 20° and 60° (solar spectra shown in Figure 3-10). While jN02
diminishes by only 37 percent, JQ^Q decreases by over 80 percent because a
large fraction of the light energy needed for 0( D) production (below 300
nm) is not avaliable at zenith angles of 60 degrees.
As noted earlier, our in-chamber calculations of j-values were performed
at a high wavelength resolution (1 nm bins) using the best absorption
cross section and quantum yield data available. However, since the
j-values previously calculated by Jeffries and Sexton (1987) are recom-
mended for use with the CBM-IV in OZIPM-4 atmospheric simulations, we have
compared the two sets of calculated j-values to show the differences in
atmospheric values that can occur as a result of the technique and data
used. This was done by eliminating the chamber-dependent factors (such as
Teflon reflection and absorption) from our in-chamber rates and comparing
the results with the rates of Jeffries and Sexton (1987). Those differ-
ences have already been noted.
-------
O
w
0.03
0.03 -
0.03 -
0.03 -
0.02 -
0.02 -
0.02 -
0.02 -
0.02 -
O.01 -
0.01 -
O.O1 -
0.01 -
O.O1 -
0.00 -
0.00 -
0.00
2»0
Phoioadion Spedra ALD2r
(crt Zcntttt An9l« of 20 ct*gr»««)
310
330
350
370
3»0
(nm)
FIGURE 3-14. Photoaction spectra for acetaldehyde using 1 nm bins (this
report) and 5 nm bins (Jeffries and Sexton, 1987).
89
-------
Relaiive j—value Curves
(tarpvt |-MiliM (ZA-0) Equal to 1.0}
2
c
O N02
FIGURE 3-15. Comparison of relative j-value curve shapes with respect
to solar zenith angle for five major photolytic species.
9P
-------
Pholoaciion Specira for 03 and N02
'i
!
1.0
o.»
o.e
0.7
0.6
0.5
O.4
0.3
0.2
0.1
0.0
(at ZA-2O and 2A-6O)
280
I
300
i
320
340
3«0
380
400
420
(twn)
FIGURE 3-16. Comparison of photoaction spectra for
20 and 60 degrees solar zenith angles.
and JQ,D at
91
-------
However, the recommended atmospheric rates and our in-chamber j-values
used for mechanism development are mutually exclusive because of the
existence of poorly defined chamber factors; thus neither data set can be
used to support the use of the other for the opposing application. For
example, though the Bass formaldehyde absorption cross section data com-
bined with our current understanding of chamber-dependent radiation fac-
tors provides good agreement with formaldehyde rates obtained 1n the smog
chamber, this fact alone is not sufficient to justify use of the Bass data
in atmospheric simulations because the chamber factors are highly uncer-
tain. It does, however, indicate that we have a good representation of
the absolute in-chamber formaldehyde photolysis rate so that we can pro-
ceed with a hierarchical mechanism development process. If current
efforts to better define chamber radiation factors are fruitful, it will
be both instructive and necessary to utilize one set of quantum yield and
cross section data for both in-chamber and atmospheric j-value calcula-
tions. Such calculations would also benefit from an expanded effort to
minimize uncertainties in the molecular and solar distribution data
used. At present, however, we see no problem in using two slightly dif-
ferent data sets for mechanism development and atmospheric simulation pro-
vided that (1) the calculated in-chamber absolute rates allow accurate
simulation of simple experimental photolytic systems, and (2) the uncer-
tainty associated with the quantum yield and cross section data is con-
sidered in atmospheric and smog chamber j-value calculations.
In addition to the photolysis processes just described, we represent four
other processes,
JHONO:
HONO JX HO- + NO (23)
03 JX 0(JP) + 02 (8)
JN03:
N03 ^ N02 + 0
(14)
—* NO + 02
by fixed ratios to JNQ2- Quantum yields and absorption cross sections for
N03 were taken from the review of Atkinson and Lloyd (1984). The NASA
review (DeMore, 1985) was the source of the ozone quantum yields and HONO
cross sections. Finally. HONO quantum yields were taken from Baulch et
al. (1984) and a high resolution ozone cross section was obtained from the
National Bureau of Standards (Bass, 1985).
92
-------
The absolute j-values and ratios to jNg2 are shown in Table 3-9.
tracks the shape of the upper curve in Figure 3-15 (JNQZ) Qul'te well, as
is apparent in Table 3-9 from the generally constant ratios to JNQZ- Tak
Z-
ing the mean of the ratios to jNQ2 for the values between 20° and 70°
zenith angles, we find
JHONO = °-197 (± °-009) x JN02»
where the associated error is two standard deviations. For both JQ3P and
i the values are
J03p = 0.053 (± 0.012) x JNQ2,
and,
JN03 = 33-9 <* 9-9> x JN02»
where jNQ3 represents the combined N03 photolysis channels, since the
ratios between them are independent of zenith angle, resulting in
N03 ^ 0.89 (N02 + 0) + 0.11 (NO + 02). (14)
The good correspondence of JHONO» ^03P» an<^ JN03 w^th ^N02 (seen ^n Table
3-9) is not surprising since these species absorb light at wavelengths
longer than most of the molecules discussed, and are, therefore, near to
the spectral region of N02 absorption. The correlation of JQ3p and
to jN02 is not as good as that of JHQNO* nowever» these species will
respond to changes in spectral distribution (zenith angle) much as N02
does.
Hydrogen peroxide photolyzes at a comparatively slow rate through the pro-
cess JH202:
H2°2 ^ 20H*
The absorption cross section is given by CODATA (Baulch et al., 1984) and
the quantum yield is assumed equal to unity over the spectral region of
interest. Unlike MONO and N03, hydrogen peroxide absorbs light at shorter
wavelengths and has a zenith angle response similar to that of the
photolysis of formaldehyde to stable products (CBM-IV reaction 39). It is
therefore possible to formulate a fixed ratio factor for determining JH202
from either jN02 or our JHCHOS- These are calculated to be
JH2£)2 = 0.255 (± 0.040) x JHCHOs,
and
JH2Q2 = 8.33 (± 4.09) x 10"4 x JNQ2.
-------
Care must be exercised in using ratios such as these because the reference
j-values (JNQ2 and JHCHOs^ may vary ^or dlf^erent applications. For
example, the ratio of 0.255 is based on JucHOs calculations used to m°del
smog chamber experiments. As noted, our J^CHOs va1ues di^er ^rom tne
results of Jeffries and Sexton (1987) by approximately 30 percent. Hence,
if their calculated formaldehyde rates are used (as in atmospheric model-
ing with OZIPM-4), the appropriate relationship is
jH202 = °-189 x JHCHOs '
For cases in which it is difficult to determine the relationship to
JHCHO§» the ratio to JNQ2 snould be used since that value is somewhat more
certain. The values given in Tables 1-2 and 1-3 show H202 photolysis
ratioed to our HCHOs rates.
Two other photolysis reactions occur in the CBM-IV:
%EN:
OPEN ----- > C203 -i- H02 + CO (69)
MGLY ----- > C203 + H02 + CO (74)
Little is known about the photolysis of these species (particularly that
of OPEN, since it is composed of highly reactive aromatic ring decomposi-
tion products). The absorption cross section of MGLY has been reported
(Carter et al., 1986), but since a wavelength-resolved quantum yield func-
tion has not yet been determined, it is difficult to apply this informa-
tion to solar spectra. Instead, these rates have been estimated through
smog simulations that will be described in a later section. For our
applictions, we have used
= 9'04 x J'
= 9-64 x J
When these rates are used with the JncHOr values from Jeffries and Sexton
(1987) in OZIPM-4, the following ratios should be applied:
8'40 x
8-96 x
Tables 1-2 and 1-3 contain rate constant ratios to our JHCHOr va1ues-
94
-------
CONDENSATION OF INORGANIC AND CARBONYL REACTIONS
In previous versions of the CBM-X, the third-body effect of molecular
nitrogen and oxygen was explicitly represented in the mechanism as a reac-
tant. In the current version, we have calculated the temperature-depen-
dent concentrations of M and 02 in molecules per cm3 prior to determina-
tion of temperature-dependent rate constants in ppm'^nin"1 units. In
addition to this minor change in format, we have condensed the two N03
photolysis channels into one reaction, since the stoichiometry does not
vary with zenith angle to any appreciable degree (both channels absorb
light far into the visible range). These reactions would normally be
separate in the CBM-X formulation, but because this was the only differ-
ence in CBM-X and CBM-IV inorganic reaction sets, this combination in the
CBM-X allowed identical inorganic sets to be used in both mechanisms.
The reactions of FORM are explicitly included in both mechanisms with the
exception of the FORM plus H02* equilibrium and reaction scheme. Using
the above noted rates of Su et al. (1979), we found that elimination of
this scheme had virtually no effect over a wide range of NOX concentration
simulations. It should also be noted that use of the more recent rates of
Vayret et al. (1982) or Barnes et al. (1985) shift the equilibria further
toward the original products, making this set of reactions even less
significant. Therefore, though the reaction set is still included in the
CBM-X, it has been eliminated from the CBM-IV.
As noted, the chemistry and condensation of acetone (AONE), the higher
ketones (KET), methylglyoxal (MGLY), and a lumped dicarbonyl species
(OPEN) are discussed in the following section. AONE and KET are con-
sidered with the alkyl group chemistry and MGLY and OPEN are discussed
with aromatics. In the remainder of this subsection, we discuss the con-
densation of acetaldehyde and higher molecular weight aldehydes (ALD2).
More specifically, we present the condensation steps and chemistry opera-
tors we have developed to handle generalized alkylperoxy. acylperoxy and
alkoxy radical chemistry.
The loss mechanisms of ALD2 and its primary radical product, acetylperoxy
radical (C203), are represented identically in the CBM-X and the CBM-IV.
However, the chemistry of their reaction products, primarily methylperoxy
radical (ME02) and methoxy radical (MEO), has been condensed for the CBM-
IV. Because the higher molecular weight aldehydes are intimately linked
with alkane chemistry through the hydrocarbon oxidation/radical chain
degradation reaction sequence, the condensation steps discussed here also
apply directly to the alkane scheme. The reactions to be condensed are 53
to 70 in Table 1-2. The methodology is as follows:
-------
The methoxy radical (CH30') 1s a rapidly reacting intermediate in this
radical chain cycle. At tropospherlc concentrations of C^t the dominant
reaction of MEO is reaction 59 (of the CBM-X):
MEO (+ 02) > FORM + H02. (X-59)
In addition, the alternate sinks of MEO form products that either
regenerate the initial reactants through other reactions, or yield the
products of this reaction. Therefore, we make a steady-state assumption
for MEO:
MEO = FORM + H02
that eliminates species MEO, MNIT, and MEN3 and reactions 56 through 62 in
the CBM-X, and transforms the product yields of MEO-forming reactions as
in
ME02 + NO > N02 + MEO, (X-55)
ME02 + ME02 > 2 MEO, (X-63)
ME02 + C203 > ME02 + MEO, (X-65)
which are now
ME02 + NO > N02 + FORM + H02,
ME02 + ME02 > 2 FORM + 2 H02,
ME02 + C203 > ME02 + FORM + H02.
Before we discuss the condensation of the chemistry of ME02, we describe
our general approach to peroxy radical chemistry. The reactions of ME02
and other peroxy radicals (C203, AN02, R02, R02R, A02, T02, etc.) are very
similar because the peroxy radical structure common to all of these
species is the main functional group. In the presence of NO, the predomi-
nant fate of these radicals is the oxidation of NO to form N0£ and a
variety of organic products. In the CBM-IV, most of these peroxy radicals
(with the exception of those producing C203 and T02) have been eliminated
by substituting the specific organic products that they form in an NO-to-
N0£ conversion reaction, and an operator that can convert an NO to an
N02« For example, the reaction sequence
AONE + OH > AN02 (X-72)
AN02 + NO > N02 + C203 + FORM (X-73)
96
-------
1s represented by
AONE + OH > X02 + C203 + FORM,
where the universal peroxy opertor (X02) is assumed to react with NO via
X02 + NO > N02, (79)
at a rate of 12000 ppm'^min"^. Because the main chemical function of all
peroxy radicals occurs with the common peroxy radical group, the reaction
rates with NO are fairly similar, and this approximation is a good one.
When NO concentrations are low, other species can compete with NO to
reduce peroxy radicals. In most cases, however, the self and cross reac-
tions of peroxy radicals themselves are the most probable peroxy termina-
tion reactions. Unfortunately, the chemistry of only a few organic
peroxy-peroxy and ^-organic peroxy reactions have been studied. We have
discussed these reactions and have included them in the CBM when appropri-
ate. Since much of the radical mass is routed through the C203 and ME02
radicals of the CBM-X, many of the peroxy radical termination reactions
are accounted for. However, we have also included the termination of the
X02 operator as a necessary reaction under conditions of low NO:
X02 + X02 >. (80)
Methylperoxy radical chemistry can now be condensed using the general
peroxy radical chemistry just discussed. Product ME02 1n the CBM-X is
replaced by X02 + MEO, and, using MEO = FORM + H02 from above, the reac-
tions
ALD2 —hv—> ME02 + H02 + CO, (X-49)
C203 + NO > N02 + ME02, (X-50)
and
C203 + C203 > 2 ME02, (X-66)
can now be condensed to
ALD2 —hv—> X02 + FORM + 2 H02 + CO, (45)
C203 + NO > N02 + FORM + H02 + X02, (46)
and
C203 + C203 > 2 FORM + 2H02 + 2X02. (49)
Finally, we have eliminated the product PROX from the C203 + H02 reaction
1n the CBM-IV. Because the atmospheric reactivity of organic peroxides is
97
-------
poorly understood, we felt 1t was improper to follow the hypothetical
chemical reactions of these species and thus possibly mislead potential
users of the mechanism. Consequently, the reactions
C203 + H02 > PROX k = 2 x 103 pprn'^in'1 (X-69)
C203 + H02 > ME02 + OH k = 7.6 x 103 ppm'^in'1 (X-70)
have been replaced with
C203 + H02 —> 0.79 (FORM + H02 + X02 + OH) k = 9.6 x 103 ppm^min'1 (50)
In the following section, we utilize much of this chemistry to describe
the chemistry of reactive organic species.
98
-------
REACTIVE HYDROCARBON CHEMISTRY
In this section we describe the reactive hydrocarbon chemistry used in the
most recent versions of CBM. Our discussion covers alkyl group (paraffin)
chemistry, olefin and ethene chemistry, aromatic chemistry, and isoprene
and a-p1nene. The chemical mechanisms for these groups employ both
explicit structures and reactive surrogates as the entities of lumping and
reaction; therefore, a single molecule can include several reactive
structural groups. For each of these four groups, we examine relevant
chemical kinetic and mechanistic data, its treatment in development of the
CBM-X, and subsequent condensation to the CBM-IV. Reaction numbers in the
text refer to the CBM-IV (Table 1-3) unless preceded by an X, which
Indicates that they refer to CBM-X (Table 1-2) reactions only. Many of
the rates and stoichiometries discussed in this section are dependent to
some degree upon the composition of the reacting hydrocarbons. To derive
these parameters we used the hydrocarbon profile given in Table 4-1. This
profile represents the average of 23 samples taken in Los Angeles as
analyzed by Environmental Research and Technology and Washington State
University (Grosjean et al., 1981).
ALKYL GROUP (PARAFFIN) CHEMISTRY
Structural reactivity generalization is especially appropriate for alkyl
group chemistry because 1t allows a complex reaction scheme to be reduced
to a much smaller set of lumped reactions.
Photooxidation of alkyl carbon in urban smog results almost exclusively
from'hydrogen abstraction by OH in the general form:
R
R I3
H-c( * -OH —* -C - R * HO
i "* i 2 2
"i
99
-------
TABLE 4-1. Average concentration (ppb) of each hydrocarbon
species identified in ambient air by the ERT and WSU methods
for 23 common analyses. (Source: Grosjean et al., 1981.)
ID
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Species
Ethane
Ethylene
Acetylene
Propane
Propene
Propyne
Propadiene
Isobutane
Butane
1-Butene
Isobutene
Trans-2-Butene
Cis-2-Butene
Isopentane
Pentane
3-Methyl-l-Butene
1,3-Butadiene
1-Pentene
Isoprene
Trans-2-Pentene
Cis-2-Pentene
2-Methyl-2-Butene
2,2-Dimethylbutane
Cyclopentene
Cyclopentane
2,3-Dimethylbutane
2-Methylpentane
Ci s-4-Methy 1 -2-Pentene
2-Methylpentane
2-Methyl-l-Pentene
Hexane
Trans-2-Hexene
2-Methyl-2-Pentene
Cis-2-Hexene
Methyl cyclopentane
Ave. ERT
(ppb)
76.68
40.20
43.89
38.53
10.65
.00
.00
15.89
32.66
.00
.00
.00
.00
34.77
15.30
.00
.00
1.24
.00
.00
.00
.00
.00
.00
.00
3.51
13.82
.00
8.27
.20
10.39
.00
.00
.00
7.48
Ave. WSU
(ppb)
13.82
31.38
39.71
45.10
14.69
.00
.00
23.50
53.84
2.26
4.06
3.80
.34
28.72
18.94
.85
.00
.34
1.50
1.54
1.58
.42
.12
.76
2.81
3.31
11.57
.03
8.11
1.05
8.60
.51
.49
.28
9. '62
Ratio1
ERT/WSU
5.55
1.28
1.11
.85
.72
-99.00
-99.00
.68
.61
-99.00
-99.00
-99.00
-99.00
1.21
.81
-99.00
-99.00
3.67
-99.00
-99.00
-99.00
-99.00
-99.00
-99.00
-99.00
1.06
1.19
-99.00
1.02
.19
1.21
-99.00
-99.00
-99.00
.78
(Continued)
100
-------
TABLE 4-1. (Continued)
ID
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
Species
2,2, 3-Trimethyl butane
2,4-Dimethylpentane
1-Methylcyclopentene
Benzene
Cyclohexane
2-Methylhexane
2,3-Dimethylpentane
3-Methylhexane
Dimethyl cyclopentane
2,2,4-Trimethylpentane
Heptane
Methylcyclohexane
Ethylcyclopentane
2,5-DiiMthylhexane
2,4-Dimethylhexane
2,3,4-Trimethylpentane
Toluene
2,3-Dimethylhexane
2-Methyl heptane
3-Methyl heptane
2,2,5-Trimethylhexane
Dimethyl cyclohexane
Octane
Ethylcyclohexane
Ethyl benzene
P- & M-Xylene
Styrene
0-Xylene
Nonane
I sopropyl benzene
Propyl benzene
P-Ethyltoluene
M-Ethyltoluene
1,3, 5-Tn'methy 1 benzene
0-Ethyltoluene
Tert-Butyl benzene
1, 2, 4-Trimethyl benzene
Sec-Butylbenzene
1,2,3-Trimethylbenzene
Ave. ERT
(ppb)
.39
2.83
.00
3.13
15.43
.00
4.37
6.47
1.90
7.56
4.63
5.51
.00
.84
.00
2.09
33.92
.00
.00
2.39
1.18
.06
1.76
.00
6.37
18.06
.00
6.83
1.13
.08
1.17
5.47
.00
2.46
1.52
.00
5.63
.24
1.40
Ave. WSU
(ppb)
.00
3.98
.00
13.94
35.25
5.29
5.30
6.94
4.97
7.55
6.02
6.37
.97
.14
2.87
1.76
40.37
1.28
4.76
2.49
1.87
.00
2.67
.99
7.08
20.40
1.21
9.36
1.54
.64
2.56
5.98
2.72
2.06
1.76
.00
7.14
.65
1.71
Ratio
ERT/WSU
-99.00
.71
-99.00
.22
.44
-99.00
.83
.93
.38
1.00
.77
.86
-99.00
5.90
-99.00
1.19
.84
-99.00
-99.00
.96
.63
-99.00
.66
-99.00
.90
.89
-99.00
.73
.73
.12
.46
.92
-99.00
1.20
.86
-99.00
.79
.37
.82
(Continued)
-------
TABLE 4-1. (Concluded)
ID
75
76
77
78
79
80
81
82
83
84
85
Species
Decane
Methyl styrene
1,3-Diethylbenzene
1,4-Diethylbenzene
1,2-Di ethyl benzene
Undecane
Dodecane
N-Butyl benzene
Unidentified
Ave. ERT
(ppb)
2.93
.00
.00
.00
.00
.00
.00
.20
.00
.00
80.12
Ave. WSU
(ppb)
1.05
1.21
1.17
1.25
.00
.00
.00
.00
.00
.00
52.62
Ratio
ERT /WSU
2.79
-99.00
-99.00
-99.00
-99.00
-99.00
-99.00
-99.00
-99.00
-99.00
1.52
Ratio of -99.00 indicates ratio cannot be calculated.
-------
When all the variable functional groups (R) are hydrogen, the organic
reactant in this reaction is methane. Methane oxidation is of interest
mainly due to its involvement with the global carbon cycle, and is essen-
tially unreactive on the time scale of urban smog. We treat the back-
ground reactivity of methane in the CBM by assuming a constant methane
concentration of 1.85 ppm:
?
OH (+ CH4) -C CH302- + H20 . (X-74)
This results in a pseudo-first-order rate constant (DeMore et al., 1985)
of
k(X-74) = 6.521 x 103 exp (-1710/T) min'1,
and
k(X-74)29g = 2.10 x 101 min"1.
OH Abstraction of Primary C - H Bonds
When only one of the R groups in the hydrocarbon reactant is an alkyl
chain,
H
H - C - R
H
abstraction of one of the three remaining hydrogen atoms by the hydroxyl
radical proceeds at a molecular rate of about 260 ppm"1 min"1 (Atkinson,
1986; Atkinson and Lloyd, 1984). In the presence of tropospheric oxygen,
the resulting RCH- radical rapidly forms a peroxy radical (RCH 0-) that
can react with NO:
RCH 0- + NO * NO -i- RCH20-
A minor fractional channel for this reaction results in nitrates:
RCH202 + NO * RCH2ON02
with a yield that increases with increasing alkyl chain length (maximum
yield = 30 percent). We discuss treatment of R02'+ NO - nitrate later in
this section.
103
-------
For R < CHo(CH2)2-, or for any molecule in which the maximum carbon chain
length is less than 4, the RChLO- radical reacts mainly with 02 to form an
aldehyde and a hydroperoxy radical:
RCH20- + 02 - RCHO + HOj
Additional intramolecular reactions can occur for alkoxyl radicals with a
carbon chain length of C^ or greater. Both decomposition and "tailbiting"
isomerization reactions are possible (Hendry and Kenley, 1979).
Decomposition reactions follow the general form:
+°2
RCH 0- - <• RO- + HCHO
in which a carbon-carbon bond is severed to form a stable carbonyl species
and a peroxy radical.
The isomerization sequence proceeds as follows:
+°2
H3CRCH20- — ^ -02
•02H2CRCH2OH + NO * N02 +
•OHCRCHOH * HOHCRCH(OH)
The a-hydroxy radical could react with 0? via abstraction of hydrogen from
the hydroxy group to yield an aldehyde (Carter et al., 1979):
HOH2CRCH(OH) + 02 * H0» + HOh^CRCHO
The overall effect of the isomerization reaction, therefore, is to delay
the formation of the RCHO and H02 products of the nonisomerization reac-
tions by inserting n alcohol-forming steps and n NO-to-N02 conversions:
+(n+l)02
RCH20- - * RCHO + HOj + n (Alcohol groups and NO-to-N02 conversions)
This process can be treated by the following reactions:
RCH20- * HORCH^ ,
104
-------
,0^ + NO * N02 + (Alcohol group) + RCH20- ,
RCH20- —^ HOj + RCHO (adlehyde)
The ratio of the first two reaction rates will determine the number of
alcohol groups added to the molecule. A 1:1 ratio, for example, will
yield 1 alcohol group per reaction of OH with a primary carbon atom.
The importance of isomerization reactions in alkane chemistry is uncertain
since these compounds have not been directly observed and the yield of
secondary alcohols in high molecular-weight alkane systems has not been
quantified. Treatment of this process would require the introduction of
another carbon-bond representation (alcohol groups); however, we note that
Whitten et al. (1986) demonstrated that ethanol could be represented as a
paraffin in reactivity on a per carbon basis. This is possible because
the OH rate constant per carbon atom is nearly twice the paraffinic OH
rate constant while the number of NO-to-N02 conversions per OH reaction is
about half that for paraffins. Such an approximation also depends on the
fact that only a fraction of paraffins react per day. If both ethanol and
paraffins reacted rapidly, the lower number of NO to N02 conversions from
the alcohol reaction would produce less overall ozone than the paraffin
reaction. If one extends this ethanol for paraffins approximation to
alcohol groups in general, then the formation of an alcohol group on a
paraffinic carbon atom can be ignored. Until isomerization phenomena are
better quantified, we believe that inclusion of alcohol groups in mechan-
isms is not required. The Carter et al. (1986) mechanism uses an isomer-
ization alcohol yield of about 0.2.
OH Abstraction of Secondary C-H Bonds
The reactivity of OH for secondary C-H hydrogen abstraction is about 1700
ppm min (Atkinson, 1986). In compounds with an alkyl carbon chain
length of 4 carbons or less (butane and methyl substituted butanes),
hydroxyl abstraction of a secondary C-H bond yields two main reaction
sequences (again, neglecting nitrate formation until later):
H 00-
I +o2 |
RX-C-R2 -i- -OH —^ R1-C-R2 + H20
105
-------
00- 0-
I I
R1-CH-R2 + NO * N02 + R1-CH-R2
-t- HO-
R -CH-R -» R CHO + R^
Competition between the last two alkoxy reactions (hydrogen abstraction by
Oo and decomposition) is largely governed by variations in the decomposi-
tion reaction rate due to molecular size and temperature. For most of the
hydrocarbons in the atmospheric mix, the RiCHO formed in decomposition
will be acetaldehyde (R=CH3). Since the average carbon number of alkanes
in the atmosphere is about 5.5, the alkyl group (R£) formed in the reac-
tion will be approximately C^-C^.
In the review by Atkinson and Lloyd (1984) the competition between alkoxy
radical abstraction by 0? and decomposition was shown to dramatically
change with the number of carbon atoms in the alkoxy radical. For the
series sec-propoxyl , sec-butoxyl, sec-pentoxyl and longer sec-radicals the
competition can be expected to go from essentially abstraction only to
decomposition only.
It 1s possible that the Isomerization reaction occurs 1n the secondary C-H
bond abstraction sequence for alkyl chains equal to or greater than 5:
OH
CH3CH(02)CH2CH2CH3 + NO * N02 + CH3CH(0-)CH2CH2CH
The importance of this reaction (and the analogous reaction Involving the
tertiary carbon group) depends on competition from, other processes affect
Ing alkoxyl radical decomposition and hydrogen abstraction. The isomeriz
ation reaction of rings of greater than 6, i.e., alkyl chains more than 4
106
-------
carbon atoms in length, has not been studied. Hendry and Kenley (1979)
suggest that the 7-member ring is not favored thermodynamically.
The discussion above leads to the following conclusions regarding sec-
alkoxyl radicals: *
For sec-alkoxyl radicals of 4 or less carbons abstraction is the
exclusive or a major pathway;
For sec-alkoxyl radicals of 4 or more carbons decomposition is a
major pathway with some isomerization to alcohol groups possible;
Since ambient air contains an average carbon number between 5 and 6,
then the first alkoxyl radical formed after reaction with OH will
most often decompose. However, the peroxy radical formed from such a
decomposition will most often have fewer than 4 carbons and the sub-
sequent alkoxyl radical will therefore undergo abstraction only.
Symbolically, these conclusions lead to the following sequence for secon-
dary attack by OH:
°2
OH + paraffin -=•» sec-ROj + H20
sec-ROj + NO * N02 + sec-RO«
sec-RO« * ALD2 + R'Oj
R'Oj + NO * N02 + R'O-
R'O- * HOj + ALD2 or KET
OH Abstraction of Tertiary C-H Bonds
Hydrogen abstraction by OH from tertiary C-H bonds occurs at a rate of
about 3200 ppm"^min~^ (Atkinson, 1986). The alkoxyl radical formed in
this oxidation sequence is devoid of hydrogen atoms and therefore cannot
undergo abstraction (as in the forementioned scheme) by 02. Hence
R, R,
I3 +0 |3
R,-C-H + -OH —£ R9-C-00- + H,0
2 , 2 , .2
Rl Rl
107
-------
R, R.
I3 I3
R.-C-OO- + NO * N00 + R--C-0-
2 I 2 2 I
Rl Rl
R 0
I I'
R2-C-0- — R!-C-R2 + R3
Rl
Isomerization reactions might also occur in addition to decomposition for
the tertiary alkoxy radical provided the functional groups will support
this process. For example, isomerization
CH3(CH2)n-CO- — ^ -02CH2(CH2)n-C-OH
Rl Rl
occurs if n is greater than or equal to 2.
A fraction of tertiary alkoxyl decomposition may yield ketones, e.g.,
3-methyl pentane could form methylethyl ketone (MEK):
00-
+0 |
CH3CH2CH(CH3)CH2CH3+-OH
00- 0
CH3CH2C(CH3)CH2CH3+NO * N02
0- 0
(MEK)
For a specific hydrocarbon, the ketone production ratio from tertiary
alkoxyl decomposition can be calculated from its structure, e.g., 2,3,4
trimethylpentane
CH,-C - C - C - CH
3 I I I
H H H
108
-------
has 3 tertiary carbons. Reaction and decomposition at the 2-carbon site
will yield acetone:
CH,- C - C - C - CH
3 I I I
0 H H
0 CH, CH,
II I3 I3
CH.-C-CH, + -C - C -CH,
II
H H
while reaction at the 3-carbon site will yield a higher molecular weight
ketone:
CH3 CH3 CH3
I I I
CH,- C - C - C - CH,
3 I I I 3
H 0 H
•
+
CH3 CH, CH,
I I3 I3
CH..-C- + C - C - CH, (Methylisopropyl ketone)
J I II | J
H OH
Thus, the overall ketone yield would be 2 acetones to 1 larger ketone if
the three sites are equally reactive. From our hydrocarbon profile (Table
4-1), we can estimate the ketone production ratio from tert-alkyl carbons
as 2.5 acetone to 1 higher ketone. We also note the existence of three
compounds containing quartenary carbons: 2,2,3 trimethylbutane, 2,2,4,
trimethylpentane, and 2,2,5 trimethylhexane. All three could be expected
to yield acetone if decomposition affects the quartenary carbon.
Quartenary carbon atoms can probably be neglected since compounds contain-
ing them appear to be rare in the atmosphere (see Table 4-1). However, it
is interesting to note that quartenary carbon atoms cannot be oxidized
109
-------
directly by OH abstraction since there are no hydrogens available. Oxida-
tion can occur only from the rupture of one of the four carbon-carbon
bonds. For example, neopentane might lead to the (CH^CCH^O- radical
after initial OH attack on a methyl group and reaction of the peroxy radi-
cal with NO. Hence, the initial activation possible for the quartenary
carbon would be the formation of the (CH-CX radical.
Ketones
The ketone structure,
appears to be essentially unreactive to hydroxyl chemistry at the carbonyl
carbon as might be expected from the lack of labile hydrogens. The pri-
mary reaction route involves OH and occurs via decomposition resulting
from reactions in the alkyl side chains, e.g., for MEK (Cox et al., 1981):
OH OH
II | +02 || |
CH.,-C-C-CH,+ .OH —^ H00+CH0-C-C-CH
J I J C. 3 I
H 00-
OH OH
II I II |
CH,-C-C-CH,+N(KNO,+CH,-C-C-CH
•3 I J f. J I
00- 0
OH OH
II I +0 || I
CH -C-C-CH * CH -C + C-CH
I 3 3 I II
0 00- 0
Note that the presence of the carbonyl structure creates a peroxyacyl pro-
duct, whereas an alkyl group would yield an alkylperoxy product.
110
-------
Acetone is sufficiently dissimilar to higher ketones (like MEK) in its
rate of photolysis to warrant explicit treatment. Acetone is formed dir-
ectly from propane oxidation but the principal formation pathway in the
urban atmosphere may be from decomposition of tertiary alkoxy radicals:
R.-C-R-^R _C-R,+R..-
1 I J 1 23
R2
As previously noted, acetone is by far the most significant product of
such reactions because methyl side chains are more common than lengthier
alkyl branching groups, and also because the C-C bond strength is weaker
for carbon chains greater than C2. Therefore, the decomposition tends to
occur preferentially for alkyl side chains over methyl side groups.
Treatment of Decomposition in Carbon-Bond Groups
The alkyl radicals formed from the decomposition reactions of secondary
and tertiary alkoxy radicals are similar to, but smaller than, the alkyl
radicals formed by the initial hydrogen abstraction reaction of the parent
molecules. If an R group is a methyl group, then a methyl radical may be
formed. A methyl radical is expected to react with 02 to form a methyl
peroxy radical under tropospheric conditions.
R7
CH3
CH302
Note that all primary carbons are methyl groups by definition. Therefore,
methyl peroxy radicals are formed at a rate relative to the overall number
of primary carbon atoms. (The bond strength relative to the secondary and
tertiary C-C bonds will also play a role.)
Similarly, if a secondary (-CH2R) group is next to the decomposition site,
the alkyl radical formed is equivalent to that formed from the OH abstrac-
tion from a primary carbon.
R2
111
-------
Also, if a branched or tertiary group is involved, then a secondary radi-
cal is formed:
OH 0
* R1-C-C-R4 * Rj-C^ + R3CHR4
R2R3
The forementioned reactions can actually be represented by the single
reaction:
R 0
|3 0 II
R0-C-0- -^ R1-C-R,(carbonyl) + R,0$
£ I 1 f. 3 f-
Rl
If R} is hydrogen the carbonyl product is an aldehyde, otherwise the car-
bony! product is a ketone. The nature of the carbon atoms adjacent to the
site of decomposition determine the form within the R^O* product. If the
general CBM assumption of independently reacting carbon-bonded structures
holds, then the distribution of products for RJD* will be related to the
distribution of carbon-bonded structures such as primary, secondary, ter-
tiary, and quartenary carbon atoms in the generalized pool of alkyl carbon
atoms and other structures such as ketones, etc. Although we will later
assume a generalized distribution within a single structure species for
alkyl carbon atoms, we can presently assume that four single-bonded carbon
atom structure species might be used in some extended form of the CBM.
The generalized decomposition pathway would form a distribution of R,0*
radicals at rates related to the relative abundance of the carbon struc-
ture groups with perhaps some minor modification by local bonding
effects. The minor modifications would also be a function of the distri-
bution of the carbon-bonded structures groups. One way of relating the
distribution of products for the generalized peroxy radical product
is the use of a virtual species or operator we will call D. This
species D would replace R-jO* as a product in the generalized decomposition
reaction and then would rapidly "react" with the structure species.
R C-0« - R,-C-R5
2 , 12
Rl
112
-------
Primary Carbon + D — =* MeO?
Secondary Carbon + D — =* RCH202
Tertiary Carbon + D — ^* R R CHCL
-«-02 , „
Quartenary Carbon + D — =*• R R R CO,
The relative rate constants would reflect the minor modifications. As
noted, quartenary carbon is rare (about 1 percent of typical atmospheric
mix, see Table 4-1) and could be neglected. The C-C bond strength of
primary carbon is apparently greater than that of secondary or tertiary
carbon, so the rate of the first reaction is relatively low.
For ketones, the carbonyl group adjacent to decomposition will result in a
peroxyacyl radical, as previously noted. Thus
Ketone + D — ^ RCOj .
Upon later parameterization of these reactions to stoichiometric yields
(parameterization accomplished by assuming that intramolecular reactions
take place instantaneously), the "D" species will drop out of the calcula
tions and standard reaction stoichiometrics will result.
Nitrate Formation from Alkanes
Nitrate formation from alkane/NOx systems has been extensively discussed
by Atkinson et al. (1982b, 1983, and 1987). We briefly summarize their
observations here. Alkyl nitrate yields in alkane-NOx photooxidation
indicate that
R02- + NO - RON02
1s an Important source of alkyl nitrates when R contains three or more
carbon atoms (Darnall et al., 1976). In their experimental investigation
of these reactions, Atkinson et al. (1982b; 1983) jshowed that the alkyl
nitrate yields from the reaction of NO with the peroxy radical from a
specific alkane were the same regardless of (1) the time scale of the
113
-------
experiment, (2) whether the experiment was conducted in the SAPRC evacu-
able chamber, large teflon bag chambers, or small (~ 100 liter) teflon bag
chambers, and (3) whether the peroxy radicals were generated from N0x-air,
CH3ONO-NOx-air, or C12-NOx-air photolyses. These findings indicate that
heterogeneous formation of alkyl nitrates is extremely unlikely. Further-
more, there was no observable induction period for their formation, indi-
cating that they are closely associated with primary products from the
initial attack by OH on the alkane.
The alkyl nitrate yields at 740 torr and 300 K have been determined for
the n-alkanes (ethane through n-octane), and monotonically increase from
less than about 0.014 for ethane to 0.33 for n-octane (Atkinson et al.,
1982b). A plot of the yield against carbon number (Atkinson et al.,
1982b) indicates a limiting nitrate yield of about 0.35 for the higher
alkanes. Assuming that all the n-alkanes have the same limiting high-.
pressure alkyl nitrate yields, these data, along with temperature and
pressure effect data for pentyl and heptyl nitrates (Atkinson et al.,
1983), can be fit with an empirical, Troe-type fall-off curve. The best
fit to the data results in a maximum nitrate yield of 0.38.
The increase in nitrate yield with increasing carbon number is consistent
with increases in nitrate from secondary alkylperoxy reaction with NO.
Tertiary alkylperoxy radicals show a low nitrate yield of about 0.04
(Atkinson and Carter, 1987) that does not seem to increase with carbon
number. Nitrate from primary radicals is also low and increases more
slowly than nitrate from secondary alkylperoxy radicals. The limiting
nitrate yields for primary alkyl peroxy radicals may be as low as 5 per-
cent.
The reduction in nitrate formation for primary and tertiary alkyperoxy
radicals implies that chain branching has a profound effect on nitrate
yields for an alkyl compound. Table 4-2 lists the calculated nitrate
yields for alkanes, based on the yields from primary, secondary, and ter-
tiary alkylperoxy yields, weighted by the fraction of hydroxyl reaction at
each carbon site. The measured yields for n-alkanes are shown for compar-
ison. It is apparent from the table that variations of as much as a fac-
tor of 4 can occur in alkyl nitrate yields for compounds with the same
carbon number. A triple branched compound such as 2,3,4-trimethyl pentane
might have a nitrate formation yield as low as 0.04, whereas the cor-
responding n-alkane (n-octane) could have a nitrate yield eight times as
great. This phenomenon presents great difficulty for any molecular lump-
ing scheme. To lump alkanes with similar carbon numbers it would be
necessary to average greatly dissimilar nitrate yield molecules, whereas
to lump compounds with similar nitrate yields it would be necessary to
114
-------
TABLE 4-2. Calculated nitrate yields compared to measured yields for
various alkanes.
Nitrate Yields
Carbon single branch double branch
Number Calculated* n-alkanes alkanes alkanes
2 0.01
3 0.036
4 0.078 0.077 0.042
5 0.126 0.128 0.07 0.05
6 0.195 0.22 0.118 0.042
7 0.27 0.31 0.178 0.087
8 0.32 0.33 0.23 0.136
^Assuming 0.04 yield from tertiary and 0.05 yield from primary carbons
(Carter and Atkinson, 1985); secondary carbon yields vary with carbon
number (Atkinson et al., 1987); weighted by OH reactivity taken from
Atkinson and Lloyd (1984) and the urban hydrocarbon mixture of Table 4-1.
-------
average a wide range of rate constants. We do not regard the averaging of
disparate nitrate yields to be a major problem, as we demonstrate in a
subsequent discussion, this can be easily resolved by use of a reactive-
structure lumping methodology.
Alkyl nitrate yields also show a temperature dependence that is, within
experimental error, the same for all alkoxy radicals yet studied. Using
data for six hydrocarbons given in Atkinson et al. (1987) and Atkinson
et al. (1983), we calculated an activation energy for the nitrate reaction
of 1400 K (a standard deviation of 300 K).
DEVELOPMENT OF ALKANE CHEMISTRY FOR THE CBM-X AND CBM-IV
The reactions given in Table 4-3 are a compilation of the information just
presented. When added to standard inorganic, formaldehyde, acetaldehyde,
and PAN chemical reactions, this reaction scheme will yield a mechanism
for treating the chemistry of paraffinic carbon. Some prior condensation
was performed to obtain the reactions given in Table 4-3: quartenary
carbon atoms have been ignored, for example, and only one species of sec-
ondary carbon is considered, though differences in nitrate formation
yields might justify additional class splitting. Nonetheless, consider-
ably more condensation must be implemented before this alkyl carbon mech-
anism can be used in AQSMs. Even using quasi-steady-state approximations
and variable stoichiometries to eliminate the explicit treatment of radi-
cal and operator species, the number of species in the alkyl portion of
the mechanism is 14, which is large for a class of compounds that probably
accounts for less than 25 percent of the atmospheric reactivity (Calvert,
1976). Therefore, we have performed the following simplifications to
obtain the explicit mechanism given in Table 4-3.
Higher Molecular Weight Aldehydes
The RCHO species that represent aldehydes with carbon chain length greater
than €3 can be eliminated by using acetaldehyde as a surrogate species.
Our analysis of alkoxyl decomposition pathways for the atmospheric mix of
hydrocarbons given in Table 4-1 yields a 45/55 split between RCHO alde-
hydes and acetaldehyde, but also shows that aldehydes from olefins, espec-
ially acetaldehyde from propene and 2-olefins, will dominate the paraffin
yields. Moreover, our knowledge of the chemistry of higher aldehydes does
not justify separate treatment of these species. Similarly, as previously
noted, our view is that current knowledge of the isomerization pathway in
alkoxy radicals is not sufficient to justify the inclusion in a mechanism
of either the pathways or the products.
116
-------
TABLE 4-3. Explicit alkyl group chemistry.
Reaction
1 PALK 4 OH — >
2 PA02 + NO — >
3 PA02 + NO — >
4 PAO — >
5 PAO — >
6 SALK + OH — >
7 SA02 + NO — >
8 SA02 + NO — >
9 SAO — >
10 SAO — >
11 SAO — >
12 SAO — >
13 TALK + OH — >
14 TA02 + NO — >
15 TA02 + NO — >
16 TAO -->
17 TAO — >
18 TAO — >
19 D + PALK — >
20 D + SALK — >
21 D + TALK — >
22 D -i- KET — >
23 I + PALK — >
24 I + SALK — >
25 I + TALK — >
26 X + PALK — >
27 C3H8 + OH — >
28 PR02 + NO — >
29 AONE — >
30 AONE + OH -->
31 AN02 + NO — >
32 RCHO — >
33 RCHO + OH — >
34 RC03 + NO — >
35 RC03 + NO — >
36 RPAN -->
37 QALK + D -->
PA02 (+ H20)
PAO + N02
PAN03
PCOH -i- I
H02 + ALD
SA02
SAO + N02
SAN03
SCOH + I
H02 + KET
ALD + D
ALD2 + X + D
TA02 + (H20)
TAO + N02
TAN03
TCOH + I
AONE + D + X + X
KET + D
ME02
PA02
SA02
RC03
PA02
SA02
TA02
PR02
N02 + H02 * AONE
ME02 * C203
AN02
N02 + HCHO -H C203
H02 + CO 4 D
RC03
N02 4 C02 -i- D
RPAN
RC03 + N02
TA02
Reaction Rate -E/R
(ppnf n min'1) (K)
290
11400 250
600 250
0
693000
1700
8820 250
3180 250
0
150000
37000 -8000
45000 -8000
3110
11520 250
480 250
0
59620 -8000
23380 -8000
0
2280
11000
(< 0.06 x JwrHOr'
500
11000
0 .
Notes
1
1
2, 3
4
5
5
2, 3
4
6
6
1
1
2, 3
6
6
7
8
8
8
8
8
8
8
9
9
9
9
9
10
10
10
10
10
11
(Continued)
117
-------
TABLE 4-3. (Conclude N
Abbreviations
PALK Primary alkyl carbon
SALK Secondary alkyl carbon
TALK Tertiary alkyl carbon
PAN03 Primary alkyl nitrate
SAN03 Secondary alkyl nitrate
TAN03 Tertiary alkyl nitrate
PCOH Primary alkyl alcohol
SCOH Secondary alkyl alcohol
TCOH Teritiary alkyl alcohol
D Unimolecular decomposition
I Unimolecular isomerization
X Negative carbon bookkeeping species
3 Propane
? CH3CH2(02^)CH3
AONE Acetone
AN02 CH2(02')C(0)CH3
RCHO C3 + Aldehyde
RC03 C3 + Peroxyacyl
RPAN C3 + Peroxyacylnitrate
Notes;
1 Nitrate fraction composition independent for primary and tertiary
carbon; varies for secondary alkylperoxy.
2 Recommended yield is zero; for 20 percent yield from isomerization,
0.167 = k4/k5 = k9/klO = kl2/(k!3 + k!4).
3 Reaction rate constant in min .
4 Pseudo first order reaction rate assuming atmospheric concentration
of 02.
5 Nitrate formation from secondary alkylperox radicals varies from 0.04
to 0.35. Average for atmospheric mix is 0.18.
6 Rate dependent on hydrocarbon composition.
7 C-CH3 bond is stable to decomposition relative to C-C bond; k!9 = 0.
8 Fast reaction rate: k20 = k21 = k22; k23 = k24 = k25.
9 Explicit chemistry.
10 All rates for RCHO, RC03, and RPAN would be similar to those for
ALD2, C203, and PAN.
11 Quartenary carbon is less than one percent of alkyl carbon; isobutoxy
radical is usually formed.
118
-------
Alkyl Nitrate Products
The amount of carbon lost to alkyl nitrates is apparently fairly small.
Although the molecular yields for high-carbon-number n-alkanes can be as
high as 38 percent, this yield involves only a single carbon atom per
molecule because the alkyl carbons in the nitrate are still available for
reaction. Decomposition is apparently the predominate fate of a high-
carbon-number alkane subsequent to hydroxyl abstraction, and decomposition
rapidly reduces the carbon number of the alkyl group, making further alkyl
nitrate formation unlikely. Studies of the kinetics of alkyl nitrates
(Atkinson et al., 1982b) suggest that the C-H bond on the same carbon as a
nitrate group has a low reactivity to OH abstraction. Thus, nitrated
carbon can be treated as an unreactive product.
Lumping of Primary, Secondary, and Tertiary Carbon
The reactivity of OH differs greatly (260 ppnT^min, 1700 ppm~^min~^, and
3200 ppnT*min~ , for primary, secondary, and tertiary carbon respectively;
Atkinson, 1986). Usually, such a broad range of rates precludes lumping,
and rate-constant averaging. However, if the number of carbons in a par-
affin molecule is held constant, the various isomers possible are then
related so that the total rate constant is always similar except when
quartenary carbons are involved. For example, the difference between a
normal alkane and a branched isomer is the loss of two secondary carbons
combined with the gain of one primary carbon and one tertiary carbon;
twice the secondary rate constant (2 x 1700 = 3400 ppm'^in"^) is approx-
imately the sum of the primary and tertiary rate constants (260 + 3200 =
3460 ppm~*min). Thus, the overall reactivity of a branched molecule is
the same as that of the corresponding n-alkane.
Because the tertiary carbon oxidizes preferentially, one might expect the
oxidation process to cause problems 1n a carbon-structure lumping
scheme. However, tertiary alkoxy radicals also preferentially decompose,
with acetone as a common product. The formation of acetone removes less
reactive primary carbon from the alkyl pool. The formation of acetone
removes a total of three carbon atoms from the overall alkyl carbon
pool. If the average OH reaction rate constant of the alkyl carbon pool
is about 1200 ppm"^min~^, then the three carbon loss from acetone forma-
tion would remove an equivalent to 3600 ppnT^min"1. If the alkyl carbon
were divided into primary, secondary, and tertiary groups, the acetone
formation would remove one tertiary carbon and two primary carbons. The
rate constant total for this combustion is 3200 (for the tertiary carbon)
plus 2 x 260 (for the two primary carbons) giving a total of 3740
119
-------
Hence, the average rate constant of the alkyl carbon pool would not change
appreciably if tertiary carbons rapidly were depleted, acetone were the
major product of tertiary carbon reaction, and the overall average per
alkyl carbon OH rate constant were 1200 ^-
To treat acetone explicitly, it is necessary to remove carbon from the
alkyl group category by means of a bookkeeping species "X" (used previ-
ously in the CBM). In this case, there is simply a transfer of the carbon
proportions of primary, secondary, and tertiary alkyl groups in the alkyl
group pool (alkyl groups are denoted as PAR and acetone as AONE in the
CBM) from one category to another. In alkoxy radical decomposition,
ROR * AONE + D + 2X , (X-85)
with X subtracting alkyl carbon from PAR:
X + PAR * . (X-86)
For the atmospheric mix of alkanes given in Table 4-1, the relative pro-
portions of primary, secondary, and tertiary carbon are 0.47, 0.39, and
0.13 (Table 4-4). This gives an overall lumped OH reactivity per carbon
atom of about 1200 ppm~ .
Figure 4-1 gives a representation of the product ratios and reaction path-
ways of alkyl carbon subsequent to reaction with hydroxyl. Table 4-5
shows a listing of the resultling alkyl carbon (PAR) mechanism, (this is
exerpted from Table 1-2, the CBM-X). The key species are listed in Table
4-6. This reaction sequence lumps secondary and tertiary alkylperoxyl
(R02R) and alkoxyl radicals (ROR) in order to reduce the number of radical
species required. The nitrate formation pathway for primary carbon is
separated out, since the primary alkoxy (R02) radical is followed inde-
pendently. Nitrate formation for R02R is for lumped radical secondary and
tertiary nitrate formation weighted according to the reaction proportions
of each. The fraction thus derived, 0.14, can be compared to the 0.16
value computed by weighting on the basis of the fractions of secondary and
tertiary carbon only. The lower fraction is preferred because the subse-
quent reaction of reaction products involves various species that often
have lower molecular weights due to decomposition. The average nitrate
formation rate of product species will therefore be lower. Thus, the
initial nitrate formation rate of 0.14 is, in fact, an upper limit.
The ratios of radical production from unimolecular decomposition of the
alkoxy radical ("D + PAR") are derived from the ratio of secondary and
tertiary carbon and from the ratio of tertiary carbon expected to yield
acetone (reaction 14) and tertiary carbon involving molecular fragments of
120
-------
(0.12)
PALK
R02 |(°[°^»| NTR
NO —
^
(0.95)
-+ NO
F
H02 + ALD
(ALD2 + X)
ALD + D
(ALD2 + x)
Temperature
Dependent
FIGURE 4-1. Schematic representation of the PAR reaction scheme.
121
-------
TABLE 4-4. Carbon and bond types.
Concentration
Compound
(aUanes)
Isobutane
Butane
Isopentane
Pentane
2,2-Dimethybutane
Cyclopentane
2, 3-Dimethl butane
2-Methylpentane
Hexane
Methyl cycl opentane
2,2,3-Trimethylbutane
2,4-Dimethyl pentane
2-Methylhexane
2, 3-Dimethyl pentane
3-Methylhexane
Dimethylcycl opentane
2,2,4-Trlmethyl pentane
Heptane
Methyl cyclohexane
Ethyl cycl opentane
2,5-Dimethylhexane
2,4-Dimethylhexane
2,3,4-Trimethyl pentane
2,3-Dimethylhexane
2-Methyl heptane
3-Methyl heptane
2,2,5-Trimethylhexane
Dimethylcycl ohexane
Octane
Ethyl cycl ohexane
Nonane
Decane
Total carbon (ppbC)
C4-C5
C6-C10
ERT
(ppbC)
63.56
130.64
173.85
76.50
0.00
0.00
21.06
49.62
62.34
44. 88
2.73
19.81
0.00
30.59
45.29
13.30
60.48
32.41
38.57
0.00
6.72
0.00
16.72
0.00
0.00
19.12
10.62
0.48
14.08
0.00
10.17
29.30
972.84
444.55
528.29
WSU
(ppbC)
94.00
215.36
143.60
94.70
0.72
14.05
19.86
48.66
51.60
57.72
0.00
27.86
37.03
37.10
48.58
34.79
60.40
42.14
44.59
6.79
1.12
22.96
14.08
10.24
38.08
19.92
16.83
0.00
21.36
7.92
13.86
10.50
1256.42
561.71
694.71
(Continued)
122
-------
TABLE 4-4. (Continued)
Carbon Type Molecule per Hydroxyl Reactivity
Compound
(alkanes)
Isobutane
Butane
Isopentane
Pentane
2,2-Dimethybutane
Cycl opentane
2, 3-Ditnethl butane
2-Methylpentane
Hexane
Methyl cycl opentane
2,2,3-Trimethylbutane
2,4-Dimethylpentane
2-Methylhexane
2,3-Dimethylpentane
3-Methyl hexane
Dimethyl cycl opentane
2,2,4-Trimethylpentane
Heptane
Methyl eye lohexane
Ethyl cycl opentane
2, 5-Dimethyl hexane
2, 4-D1methyl hexane
2,3,4-Trimethylpentane
2. 3-D1methyl hexane
2-Methyl heptane
3-Methyl heptane
2,2,5-Trimethylhexane
Dimet hyl cycl ohexane
Octane
Ethyl cycl ohexane
Nonane
Decane
Primary
3
2
3
2
4
4
3
2
1
5
4
3
4
3
2
5
2
1
1
4
4
5
4
3
3
5
2
2
1
2
2
Second-
ary
0
2
1
3
1
5
0
2
4
4
0
1
3
1
3
3
1
5
5
5
2
2
0
2
4
4
2
4
6
6
7
8
Terti-
ary
1
1
2
1
1
1
2
1
2
1
2
1
1
1
2
2
3
2
1
1
1
2
1
Quarten- Total
ary Carbon
4
4
5
5
1 6
5
6
6
6
6
1 7
7
7
7
7
7
1 8
7
7
7
8
8
8
8
8
8
1 9
8
8
8
9
10
per
Molecule
3980
3980
5680
5680
2860
8500
7380
7380
7380
10200
4560
9080
9080
9080
9080
11900
6260
9080
11900
11900
10780
10780
10780
10780
10780
10780
7960
13600
10780
13600
12480
14180
per
Carbon
995.00
995.00
1136.00
1136.00
476.67
1700.00
1230.00
1230.00
1230.00
1700.00
651.43
1297.14
1297.14
1297.14
1297.14
1700.00
782.50
1297.14
1700.00
1700.00
1347.50
1347.50
1347.50
1347.50
1347.50
1347.50
884.44
1700.00
1347.50
1700.00
1386.67
1418.00
(Continued)
123
-------
TABLE 4-4. (Concluded)
Concentration of
ERT
Compound
(alkanes)
Isobutane
Butane
Isopentane
Pentane
2,2-Dimethybutane
Cyclopentane
2, 3-Dimethl butane
2-Methylpentane
Hexane
Methyl cyclopentane
2,2,3-Trimethylbutane
2,4-Dimethyl pentane
2 -Methyl hexane
2,3-Dimethyl pentane
3-Methyl hexane
Dimethyl cyclopentane
2,2,4-Trimethylpentane
Heptane
Methyl cyclohexane
Ethyl cyclopentane
2,5-Dimethyl hexane
2,4-Dimethyl hexane
2,3,4-Trimethylpentane
2,3-Dimethyl hexane
2-Methyl heptane
3-Methyl heptane
2,2,5-Trimethylhexane
Dimethyl cycl ohexane
Octane
Ethyl cycl ohexane
Nonane
Decane
Carbon Fraction:
C4-C5
C6-C10
Primary
47.47
65.32
104.31
30.60
0.00
0.00
14.04
24.81
20.78
7.48
1.95
11.32
0.00
17.48
19.41
3.80
37.80
9.26
5.51
0.00
3.36
0.00
10.45
0.00
0.00
7.17
5.90
0.12
3.52
0.00
2.26
5.86
0.47
0.56
0.40
Second-
ary
0.00
65.32
34.77
45.90
0.00
0.00
0.00
16.54
41.56
29.92
0.00
2.83
0.00
4.37
19.41
5.70
7.56
23.15
27.55
0.00
1.68
0.00
0.00
0.00
0.00
9.56
2.36
0.24
10.56
0.00
7.91
23.44
0.39
0.33
0.44
Terti-
ary
15.89
0.00
34.77
0.00
0.00
0.00
7.02
8.27
0.00
7.48
0.39
5.66
0.00
8.74
6.47
3.80
7.56
0.00
5.51
0.00
1.68
0.00
6.27
0.00
0.00
2.39
1.18
0.12
0.00
0.00
0.00
0.00
0.13
0.11
0.14
Carbon Type (ppbC)
Quarten-
ary
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.39
0.00
0.00
0.00
0.00
0.00
7.56
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
1.18
0.00
0.00
0.00
0.00
0.00
0.01
0.00
0.02
Primary
70.50
107.68
86.16
37.88
0.48
0.00
13.24
24.33
17.20
9.62
0.00
15.92
15.87
21.20
20.82
9.94
37.75
12.04
6.37
0.97
0.56
11.48
8.80
5.12
14.28
7.47
9.35
0.00
5.34
0.99
3.08
2.10
0.46
0.54
0.39
wsu
Second-
ary
0.00
107.68
28.72
56.88
0.12
14.05
0.00
16.22
34.40
38.48
0.00
3.98
15.87
5.30
20.82
14.91
7.55
30.10
31.85
4.85
0.28
5.74
0.00
2.56
19.04
9.96
3.74
0.00
16.02
5.94
10.78
8.40
0.41
0.37
0.44
Terti- Quarten-
ary «ry
23.50
0.00
28.72
0.00
0.00
0.00
6.62
8.11
0.00
9.62
0.00
7.96
5.29
10.60
6.94
9.94
7.55
0.00
6.37
0.97
0.28
5.74
5.28
2.56
4.76
2.49
1.87
0.00
0.00
0.99
0.00
0.00
0.12
0.09
0.15
0.00
0.00
0.00
0.00
0.12
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
7.55
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
1.87
0.00
0.00
0.00
0.00
0.00
0.01
0.00
0.01
124
-------
TABLE 4-5. Alkyl group chemistry of the Carbon Bond Mechanlsm-X.
Number
1)
2)
3)
4)
5)
6)
7)
8)
9)
10)
11)
12)
13)
14)
15)
16)
17)
18)
Reaction
<298
PAR
PAR
R02
R02
R02R
R02R
ROR
'X
0
D
0
D
A02
+ OH
+ OH
+ NO
+ NO
+ NO
+ NO
+ N02
ROR
ROR
ROR
ROR
+ PAR
-f- PAR
+ PAR
+ PAR
+ KET
+ NO
KET
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
-hv2->
R02
R02R
N02
NTR
N02
NTR
NTR
KET
KET
ALD2
AONE
R02
A02
R02R
C203
N02
C203
H02
ROR
+ ALD2 + X
H02
D
D + X
D + 2.00X
+ 2.00X
+ X
+ AONE + H02
+ R02 + 2.00X
1.360
1.067
1.140
6.584
1.033
1.836
2.200
9.546
1.236
394
621
000
500
250
500
1.000
1.200
6.000
E+02
E+03
E+04
E+04
E+04
E+05
E+04
E+04
E+16
E+16
E+16
E+04
E+03
E+03
E+02
E+04
E+04
E-02
*EXP(-
*EXP(-
*EXP(-
*EXP(-
*EXP(-
1400/T)
1400/T)
8000/T)
8000/T)
8000/T)
1.360
1.067
1.140
6.000
1.033
1.673
200
546
706
250
749
1.000
7,
2.
2.
1,
.500
.250
.500
.000
1.200
6.000
E+02
E+03
E+04
E+02
E+04
E+03
E+04
E+04
E+04
E+04
E+04
E+04
E+03
E+03
E+02
E+04
E+04
E-02
-------
TABLE 4-6. Species involved in the oxidation
of paraffin.
Symbol
Definition
PAR 1-Carbon paraffin group
R02 Primary alkyl peroxy radical
with C<4
R02R See text—alkyl peroxy
and long-chain radicals
ROR Alkoxyl radicals
KET Ketones other than acetone
AONE Acetone
D Fictitious species to account for
alkoxyl radical decomposition and
isomerization
A02 Peroxy isopropyl radical
X Fictitious steady-state species to
maintain carbon balance
NTR Organic nitrate
126
-------
C4 or greater in our assumed atmospheric mix. The sum of paraffinic
decomposition involvement is the same as the ketone rate since the alky!
carbon represents several combined pathways.
Condensation of the CBM-X Alky! Chemistry for CBM-IV
The CBM-X alkyl reaction sequence presented in Table 4-5 is still too
large for inclusion in a condensed mechanism such as the CBM-IV. We have
therefore performed a number of condensation steps ranging from simple
bookkeeping to formulations requiring rather extensive conditional testing
to verify their impacts:
(1) Elimination of species X through the use of negative stoichio-
metry. Since X was a mass balance counter that rapidly removed
PAR to conserve carbon, the production of X, as in
ROR > AONE + D + 2X ,
can be changed to
ROR > AONE + D - 2PAR .
(2) The mechanism can be abbreviated through the use of fixed sto-
ichiometry reaction combinations; e.g., the temperature-depen-
dent ROR reactions can be condensed to
ROR > D + 0.198 KET + 0.383 ALD2 + 0.419 AONE
- 1.222 PAR ,
with a combined decomposition rate at a temperature dependence
(-E/R) of -8000K.
(3) We created a second operator (X02N) similar to X02 (see the ICRS
condensation discussion) to account for the formation of a
nitrate from the reaction of NO with a peroxy radical. For
example, the reaction set of
PAR + OH > R02R (X-76)
R02R + NO > N02 + ROR
koqo = 10330
R02R + NO > NTR ^ya
k298 = 1673 PPm ml" (X-80)
127
-------
can be written using the earlier X02 = N0-N02 operator or as
PAR + OH > 0.86 X02 + 0.86 ROR + 0.14 X02N ,
where X02N is defined through the reaction
X02N + NO > (NTR) . (80)
Thus, using some previous condensations (X = -PAR, X02 = NO+N02,
and X02N = NO-NTR), and an assumption that the alkyl nitrate
yields are approximated by the 298 K ratios, we can reduce the
initial six reactions of Table 4-5 to one:
PAR + OH > 0.871 X02 + 0.107 H02 + 0.107 ALD2
-i- 0.129 X02N + 0.764 ROR - 0.107 PAR ,
with a combined kg^ of 1203 ppm'-'-min .
(4) Acetone (AONE) and ketone (KET) are slowly formed products of
alkane chemistry. In addition, the photolytic and reaction
losses (with OH) are relatively slow (Atkinson and Lloyd,
1984). For acetone, the central carbon is unable to enter into
hydroxyl reaction; therefore, we find the representation of the
molecule as 2 PAR to be satisfactory. This results in altera-
tion of product representations; for instance, the ROR reaction
just discussed as an example of X elimination
ROR > AONE + D - 2PAR ,
is now
ROR > D ,
and
A02 + NO > N02 + H02 + AONE
is now
A02 + NO > N02 + H02 + 2PAR .
The production of A02 in the reaction
D + PAR > A02 + 2X
can be changed (using the X = -PAR and X02 = NO-N02 plus pro-
ducts condensations) to
D + PAR > X02 + H02 .
128
-------
For ketones, recall that KET represents the carbonyl carbon that
forms when a secondary or tertiary alkoxy radical stabilizes
through loss of a radical. The forementioned treatment of ace-
tone assumes no reactivity for this carbon. Simulation results
indicate that the mass throughput rates for the two KET reac-
tions are small; i.e., always at least two orders of magnitude
less than the D + PAR reactions. Therefore, we eliminate the
product KET and eliminate KET photolysis and OH reactions from
the CBM-IV.
(5) We can now eliminate D since the elimination of KET allows D to
be an operator on PAR only through three non-temperature-depen-
dent reactions:
D + PAR > R02 ,
D + PAR > X02 -i- H02 ,
D + PAR > R02R .
We have already used the X02 and X02N condensations for R02 and
R02R in condensing the initial PAR reactions, and we can use
similar product splits to combine these reactions into
D + PAR > 0.959 X02 + 0.938 H02 + 0.713 ALD2
+ 0.041 X02N + 0.022 ROR - 0.713 PAR .
Since D is an operator with an instantaneous reaction rate, the
formation of D removes a PAR and instantly forms the D + PAR
products. Hence, the "product" D can be replaced by the D + PAR
products minus an additional PAR.
D = 0.959 X02 + 0.938 H02 + 0.713 ALD2
+ 0.041 X02N + 0.022 ROR - 1.713 PAR .
When substituted into the CBM-X reaction scheme, these condensa-
tions result in the CBM-IV alkane machanism:
PAR + OH > 0.87 X02 + 0.13 X,02N + 0.11 H02
+ 0.11 ALD2 + 0.76 ROR - 0.11 PAR (52)
129
-------
ROR > 1.10 ALD2 + 0.96 X02 + 0.94 H02
+ 0.04 X02N + 0.02 ROR - 2.10 PAR (53)
ROR > H02 (54)
ROR + N02 > (55)
The temperature-dependent reaction rates are given in Table 1-3.
OLEFIN AND ETHENE CHEMISTRY
As noted previously, ethene is treated explicitly in the CBM-X and very
reactive olefins with alky! substitutions of 2 or more are treated as
surrogate aldehyde and ketones. We treat the remaining class of olefins,
monoalkylates (1-olefins), on a reactive structure basis, often using
propene chemistry as a model. From a theoretical standpoint, the advan-
tage of the reactive structure approach for olefins is mass conserva-
tion. In the CBM-X, as in earlier versions of the CBM, the additional
alkyl groups of three carbons and greater are accounted for. We first
discuss the treatment of 1-olefins, followed by a discussion of the expli-
cit ethene chemistry.
Olefins are a rather reactive class of hydrocarbons found in the atmo-
sphere. These species are generally involved in atmospheric chemistry
during all periods of the day since their structure allows attack by at
least four important oxidizing species (03, OH, 0, and N03). These reac-
tions are all treated in the CBM for terminal olefins (1-olefins, desig-
nated OLE).
0(3P)-OLE Chemistry
The gas-phase reactions of 0 and olefins have been studied for over 30
years, yet their chemistry under atmospheric conditions is still somewhat
uncertain. On the basis of their review of earlier data, Atkinson and
Lloyd (1984) suggest the following scheme for propene:
0
/ \
0 + CH3CH=CH2 —> CH3CH-CH2 (30 percent)
—> CH3CH2CHO (30 percent)
—> (HCO + CH3CH2-) ' (20 percent)
130
-------
(CH2=CHO- + CH3') (20 percent)
As a point of comparison, the 1-butene yields were found to be 44, 39, 0,
and 17 percent, with an uncertainty of up to about 20 percent. In our
formulation of the OLE plus 0 reaction scheme, we use the following
splits:
OLE + 0 > 2 PAR 35% (X-93)
OLE + 0 > ALD2 35% (X-94)
OLE + 0 > H02 + CO + R02 10% (X-95)
OLE + 0 > R02 + X + CO + FORM + OH 20% (X-96)
The epoxide product is represented by the reactivity of 2PAR (based on the
kQH calculations of Winer et al. (1978) cited in Atkinson and Lloyd,
1984). ALD2 is used to represent higher aldehydes, and the products of
the third reaction are simply represented in the CBM by assuming exclusive
reaction of the formyl radical with 02. We assume that the vinoxy radical
(CH2=CHO*) reacts exclusively with 02 to yield the products shown, with
R02 + X representing the formation of a peroxy radical from the remaining
carbon in the chain. In their review of kinetic data, Atkinson and Lloyd
(1984) recommend
k(56) = 1.756 x 104 exp(-324/T) pprn'^in'1
and
k(56)298 » 5.92 x 103 ppm'^in"1 .
For the CBM-X, these rates have been partitioned into the reactions just
given.
OH-OLE Chemistry
The general reaction scheme for the OH radical addition to the OLE group
has been fairly well understood since the FTIR work of Niki et al.
(1978). Briefly, OH is assumed to add to either carbon in the OLE
group. Abstraction of non-OLE hydrogens (as described by Atkinson and
Lloyd, 1984) is handeled in the PAR chemistry. Since terminal olefins are
a special subset of the alkenes, two parallel reaction sets can be fol-
lowed, depending on the initial addition site:
131
-------
OH +0 00- OH
. I 'I
OH + R1R2CH=CH2 > R1R2CH—CH2 > R1R2CH~CH2
or
+02
> R,R?CH—CHo > RiRoCH—CH?
12| II
OH OH 00-
Both peroxy radicals are expected to react with atmospheric NO to form N02
and an oxy radical. Decomposition of both radicals through the loss of
the hydroxyl hydrogen to atmospheric oxygen (forming H02*) and the sever-
ing of the original olefin sigma bond forms two aldehyde groups at the
point of cleavage. For these molecules, the products resulting after
NO*N02 conversion, regardless of the initial OH addition site, are
R^CHO + HCHO + H02' .
Therefore, the CBM-X formulation for OH reaction with a terminal olefin
group is
OH + OLE > ME02 + ALD2 + X , (X-97)
ALD2 + X describes the -CHO end of a longer chain hydrocarbon whose chem-
istry is represented by other carbon bond groups, and ME02 oxidizes NO to
N02 to form H02 and FORM.
The kinetic expressions used in the CBM-X are
k(57) = 7.740 x 103 exp(504/T) ppm'^in'1 ,
and
k(57)2gg = 4.20 x 104 ppnf *min~l .
The temperature dependence used is from the review of Atkinson (1986). We
use a slightly higher k2gg value than is suggested therin (38870 ppm'^min'1),
which is easily within the stated uncertainty of 15 percent.
Ozone-Olefin Chemistry
Olefins are unique in their atmospheric reactivity to ozone.since other
common organics do not react with ozone at significant rates. The chem-
istry of the ozone-olefin system has been extensively investigated over
132
-------
the past decade, but has yet to be clearly delineated. Although the sec-
ondary reactions are not as uncertain as those of aromatic hydrocarbons,
there 1s much speculation about the tropospheric chemistry of Criegee
(biradical) species formed during intermediate reactions. For a more
complete discussion, the reader is advised to consult the reviews of
Atkinson and Lloyd (1984), Atkinson and Carter (1984), and Kerr and
Calvert (1985), keeping in mind that the OLE species represents terminal
olefinic groups.
The reaction rate data used for 03 plus OLE in the CBM are
k(58) = 2.104 x 101 exp(-2105/T) pprn'^in'1 ,
and
k(58)298 = 1-80 x 10~2 ppm^min'1 .
The temperature dependence is from the review of Atkinson and Carter
(1984) and the k2gg value is well within their stated uncertainty.
Ozone adds to terminal olefins at the pi bond of the olefin, initially
forming a 3-oxygen bridge known as a molozonide which rapidly rearranges
to an ozom'de and decomposes. Ozonide decomposition apparently results in
formation of an aldehyde and a biradical product. Following the recommen-
dations of Atkinson and Lloyd, we assume an equal split of ozonide cleav-
age channels:
A
0 o
CH.CH-CH. —. CH/V — «CHO+[CH3COO-] 50*
\ /
0-0 + [H^COO-] + CH,CHO 50%
(propene molozonide) , ... 2 3
VK ^ ' (propylene ozonide)
1n which the biradical species are initially formed with excess energy.
Following the work of Herron and Huie (1978), Niki et al., (1977), Su et
al. (1980), and Kan et al. (1981), we assume that approximately 40 percent
of the biradical products are thermalized in the atmosphere while the
remaining 60 percent may rearrange and decompose to smaller products. In
the CBM-X, we describe these pathways with
03 + OLE > ALD2 + CRIG + X 20% (X-98)
03 + OLE > FORM + MCRG + X 20% (X-99)
03 + OLE > ALD2 + HOTA + X ' 30% (X-100)
03 •«• OLE > FORM + HTMA + X 30% (X-101)
133
-------
where half of the 03 + OLE mass passes through reactions 98 plus 100 and
99 plus 101. A 40:60 ratio exists between reactions 98 plus 99 and 100
plus 101: The/irst set forms the stabilized biradicals (CRIG = CH3COO'
and MCRG = CH3COO*) and the lower two reactions represent the formation of
intermediate "hot" formic and acetic acids, which rapidly decompose via
(using fractions from Dodge and Arnts, 1979):
HOTA > (C02 + H2) 20% (X-112)
HOTA > CO (+H20) 70% (X-113)
HOTA > 2 H02 (+ C02) 10% (X-114)
and
HTMA > (CH4 + C02) 20% (X-115)
HTMA > ME02 + CO + OH 32% (X-116)
HTMA > ME02 + H02 (+ C02) 32% (X-117)
HTMA > 2 H02 + CO + FORM 8% (X-118)
HTMA > ME02 + H02 (+ C02) 8% (X-119)
We used the review by Kerr and Calvert (1985) to obtain a compilation of
the most likely reactions of the stabilized biradicals. Ignoring the
possible reaction with N02 (l<298 = ^° PPm~*min~ ) the reaction sets are:
CRIG + NO > N02 + FORM k298 = 10350 (X-120)
CRIG + H20 > FACD + H20 k298 = 0.59 (X-121)
CRIG + FORM > OZD k2g8 = 2950 (X-122)
CRIG + ALD2 > OZD k298 = 2950 (X-123)
and
MCRG + NO > N02 + ALD2 k298 = 10350 (X-124)
MCRG + H20 > ACAC + H20 k298 = 0.59 (X-125)
MCRG + FORM > OZD k298 = 2950 (X-126)
MCRG + ALD2 > OZD k29g = 2950 (X-127)
N03-OLE Chemistry
The kinetic data for the reaction of NOg plus the OLE group are given in
Atkinson and Lloyd (1984), but were calculated using the N20c equilibrium
constant of Malko and Troe (1982). We now believe the use of this value,
when compared to the data presented in the ICRS section, leads to an
undercalculation of k298 by a factor of 1.83. Therefore, we have con-
verted the rate suggested by Atkinson and Lloyd by that factor to yield
the temperature-independent rate of
134
-------
k(59) = 1.135 x 101
Experimental evidence indicates the addition of N03 to the OLE bond, pri-
marily at the 1-position. Following that channel for discussion purposes,
oxygen should add to the initial adduct to form a peroxy racical:
oo-
. 2 '
N03 + RiRpC^Ho > Ri RpC-CHoONOp -* RiRpC-CH'
Reaction with NO to form N02 or disproportionation with H02 should lead to
formation of an oxy radical structure under atmospheric conditions, though
formation of a dinitrate species can also occur:
00- 0-
I I
R1R2C-CH2ON02 + NO > N02 + R1R2C-CH2ON02
ON02
> R1R2C-CH2ON02 (minor pathway)
Decomposition of the oxy radical could lead to the lower molecular weight
products observed by Bandow and coworkers (as referenced in Atkinson and
Lloyd, 1984) in their propene study (R} = CH3 and R2 = H):
o-
CH3CH-CH2ON02 > CH3CHO + HCHO + N02 .
It can be shown that N03 addition to the 2-position would yield the same
products; therefore, in the CBM-X these reactions are represented by
N03 + OLE > PN02 (X-102)
PN02 + NO > DNIT (X-103)
PN02 + NO > FORM + ALD2 + X + 2 N02 (X-104)
Ethene (ETH) chemistry is explicitly represented in the CBM because ethene
is generally the most abundant olefin, and its chemistry is sufficiently
unlike generalized OLE chemistry to warrent individual consideration. The
main differences are that its initial oxidation reactions are much slower
135
-------
than higher molecular weight olefins, and because of its C2 structure, it
will not form PAN as an oxidation product. The specific chemistry of
ethene is presented in reactions 105 through 111 of the CBM-X (Table 1-2).
0(3P)-ETH Chemistry
The kinetic data for this reaction (60) are taken from Atkinson and Lloyd
(1984) and represent the sum of X-105 and X-106:
k(60) = 1.540 x 104 exp(-792/T) ppm^min'1
and
k(60)2g8 = 1080 ppm'-'-min"1 .
Unlike the higher molecular weight olefins, ethene does not appear to form
stable epoxide products. The assumed product distribution used in the
CBM-X is based on the review by Atkinson and Lloyd (1984) and the recent
data of Smalley et al. (1986). We use
0 + C2H4 —> (HCO + Chy) 70 percent
___> (CH2=CHO- + H-) 30 percent
Assuming rapid reaction of H* with molecular oxygen and the product chem-
istry discussed in the 0 plus OLE section, we represent this chemistry in
the CBM-X by
0 + ETH > ME02 + CO + H02 (X-105)
0 + ETH > H02 + OH + FORM + CO (X-106)
OH-ETH Chemistry
The kinetic expression for the hydroxyl radical-ethene reaction (61) has
been updated to one given by Atkinson (1986):
k(61) = 3.000 x 103 exp(411/T) ppm^min"1
and
k(61)298 = 1.19 x 104 ppm'^in'1 .
These values are very close to those of earlier CBM expressions.
The detection of glyocaldehyde as a product of the OH-ethene reaction
(Niki et al., 1981) has led to a redevelopment of -the reaction mechanism
to include that species. Initially, hydroxyl radical is expected to add
to ETH, forming a peroxy radical in the atmosphere in a mechanism analo-
gous to that described for OLE:
136
-------
HO +Q HO 00'
I . 2 I I
OH + H2C=CH2 > H2C~CH2 > H2C~CH2 •
In the case of ethene, however, the oxy radical formed from oxidation of
NO to N02 (for example) can decompose to form two sets of products:
HO 00- HO 0-
II II
H2C—CH2 + NO > N02 + H2C~CH2
and
HO 0- +0
I I 2
H2C~CH2 > 2 HCHO + H02« 78 percent
+0, H0 °
2 I I
> H2C~CH + H02' 22 percent
This set of reactions is represented in the CBM-X by
OH + ETH > ET02 (X-107)
ET02 + NO > N02 + 2 FORM + H02 (X-108)
ET02 + NO > N02 + ALD2 + H02 (X-109)
Ozone-ETH Chemistry
The rate expression used in the CBM-X for the ozone plus ethene reaction
(62) is the sum of reactions X-110 plus X-lll:
k(62) = 1.856 x 101 exp(-2633/T) pprn'^in'1
and
k(62)2gg = 2.70 x 10~3 ppm'^in"1 .
The temperature dependence used is from Atkinson and Carter (1984); the
k(60)2gg value is within the experimental error determination of that
review.
The reaction mechanism has received much experimental attention in recent
years. These data are summarized by Atkinson and 'Lloyd (1984). The over-
all CBM mechanism follows the reaction scheme given for 0-j plus OLE,
137
-------
except that the symmetry of ethene simplifies the secondary reaction path-
ways. Therefore, we have
03 + H2C=CH2 > HCHO + [CH200-] ,
where we again assume a 40:60 split between the thermalized biradical and
the decomposing species. In terms of the CBM-X, this is given by
03 + ETH > HCHO + CRIG (X-110)
and
03 + ETH > HCHO + HOTA (X-lll)
The reactions of the CRIG and HOTA species have been given previously.
Condensation of the CBM-X OLE and ETH Mechanisms to CBM-IV
The condensation of these CBM-X reactions to a CBM-IV scheme primarily
involves a simple algebraic combination of reactions to obtain fewer reac-
tions with fractional stoichiometries. Some steady-state assumptions,
such as use of the X02 and X02N operators, are employed.
The 0 plus OLE reactions are easily combined using X = -PAR to yield
0 + OLE > 0.35ALD2 + 0.50PAR + 0.30CO
+ 0.20FORM + 0.200H + 0.10H02
+ 0.30R02
Assuming the steady-state R02 assumption used for PAR condensation,
0 + OLE > 0.63ALD2 + 0.22PAR + 0.30CO
+ 0.20FORM + 0.200H + 0.38H02
-i- 0.28X02 + 0.02X02N (56)
The OH plus OLE reaction is condensed using the ME02 steady-state assump-
tion (ME02 = X02 + FORM + H02) and -PAR for X:
OH + OLE > ALD2 - PAR + FORM + X02 + H02 (57)
The 0^ plus OLE reactions can initially be combined using X = -PAR to:
03 + OLE > 0.50ALD2 + 0.50FORM - PAR
+ 0.20CRIG + 0.20MCRG
+ 0.30HOTA + 0.30HTMA
138
-------
Under most atmospheric conditions, the transformation of thermalized
biradicals (CRIG and MCRG) to organic acids in the presence of water vapor
should be the dominant reaction. The rate constants given in Table 1-2
are very uncertain, however, and better measurements are needed to add
certainty to such an assumption. Since this set of reactions appears to
be dominant, the biradical products can be replaced by their respective
acids; and because these acids are rather stable in the atmosphere, we do
not treat their chemistry in the CBM-IV. The "hot" rearranged biradicals
are assumed to decompose in the proportions given in reactions 112 through
119 of Table 1-2. These assumptions lead to
03 + OLE > 0.50ALD2 + 0.74FORM - PAR
+ 0.44H02 + 0.33CO + 0.22X02
+ 0.100H (58)
The N03 plus OLE reaction is condensed using the X02 and X02N operators to
eliminate PN02. Also, -PAR is substituted for X and DNIT is assumed to be
stable and thus is not followed in the CBM-IV. The reaction is
N03 + OLE > 0.91X02 + 0.09X02N + FORM + ALD2
+ N02 - PAR . (59)
If the assumption that ME02 = FORM + H02 + X02 is made, the 0 plus ETH
reaction can be condensed to the sum of reactions 105 and 106 of the CBM-
X:
0 + ETH > 1.70H02 + CO + FORM + 0.70X02
+ 0.300H . (60)
ET02 can be eliminated from the OH plus ETH reaction using the X02 assump-
tion:
OH + ETH > X02 + 1.56FORM + 0.22 ALD2 + H02 . (61)
Finally, the Og plus ETH reaction can be condensed using the same assump-
tions as in the Oj plus OLE condensation:
03 + ETH > FORM + 0.42CO + 0.12H02 . (62)
AROMATIC CHEMISTRY
The OH-plus-aromatic reaction mechanisms have been the focus of experi-
ments and modeling investigations for over a decade (Atkinson et al.,
1980; Killus and Whitten, 1982b; Leone and SeinfeVd, 1984). Because of
recent advances in analytical equipment, considerable effort has been
139
-------
directed toward identification of aromatic ring fragmentation species,
especially the larger, conjugated a- and y-dicarbonyl products (Dumdei and
O'Brien, 1984; Shepson et al., 1984). However, experimental studies that
attempt to track all of the reacted aromatic carbon have been unable to
account for more than a fraction of the reacted carbon atoms in dicarbonyl
products. This is especially true for toluene. Hence, a large portion of
the secondary product mass has yet to be confidently identified, probably
because of the many branching and isomeric pathways involved in the com-
plex photooxidation of aromatics and secondary products. Representation
of this missing mass has caused a dilema for atmospheric chemists, who
have generally proposed greater yields of the detected dicarbonyl products
in place of the unidentified products. Judging from the relatively poor
agreement of simulation results with smog chamber data, however, use of
the highly reactive dicarbonyl species to account for all of the carbon
known to have reacted appears to provide overly reactive mechanisms. Such
an inconsistency is especially obvious if the undetected species have
significantly different structures, reactivity, or chemical functions than
do the surrogate a- and y-dicarbonyl species in the models (as appears to
be the case for toluene).
The development of aromatic reaction schemes is a formidable task in that
available smog chamber data is useful, but severely limited. Existing
data must be carefully analyzed so that consideration of only a few margi-
nal experiments does not yield misleading conclusions. A large number of
individual experiments for toluene have been performed in the UNC dual
outdoor smog chambers and the UCR evacuable chambers (EC); however, most
UNC experiments were performed on days with non-ideal light conditions,
while the EC chamber requires a large (and uncertain) radical source com-
bined with the varying spectral distribution of the EC light source. As
discussed next, our development methodology for aromatics chemistry was to
use the UNC data with its diurnally varying sunlight in the initial mech-
anism description efforts, followed by analysis of different chemical
conditions, such as HC/NOX ratio, overall mass loading and initial
ratio, to identify (1) specific conditions under which the mechanism
failed to operate successfully and (2) possible secondary reactions of
importance. The UCR-EC experiments were designed in sets to provide data
for such tests; although data collection did not usually continue much
past the time of the maximum ozone concentration, conditional trends can
still be identified from these experiments.
Data for individual higher molecular weight aromatics are sparse. There
are four individual o-xylene and one m-xylene data sets available from
UNC. There are four m-xylene experiments (one with uncertain light), but
no o-xylene experiment, available from the UCR EC. Except for a few UNC
experiments, the occurence of higher molecular weight aromatic species in
smog chamber experiments is restricted to mixtures, resulting in a less
140
-------
than clear picture of individual reactivities and chemical characteris-
tics. A third data set (from Battelle Columbus Laboratories) includes NOX
plus toluene, o- and m-xylene, ethylbenzene, and trimethylbenzene. Unfor-
tunately, the light sources in these experiments are not well defined
(and simulations are therefore uncertain), though intercomparison of the
experimental results using the same light source does permit some chemical
sensitivity analysis.
Smog chamber data are used to evaluate the effectiveness of mechanism
approximations (such as the secondary product chemistry just described)
because these systems provide conditions far more similar to those encoun-
tered in the atmosphere than the conditions of fundamental laboratory
chemical kinetics studies. On the other hand, the smog chamber can rarely
yield specific data on chemical kinetics or mechanism development; this
type of information is most often obtained in a laboratory. One connec-
tion between the more complex smog chamber system and the individual chem-
ical kinetics experiment has been the ability of the former (given enough
well-conceived experiments) to indicate functional trends related to pos-
sible atmospheric reactions of secondary species not yet studied individu-
ally in the laboratory. That is, though the chamber results cannot point
out the exact structure or chemistry of the missing aromatic product spe-
cies, the chemical nature of potentially missing products is inherent in
the smog chamber data. Therefore, our overall mechanism development
included both available laboratory kinetic and mechanistic information and
systematic or conditional trends ascertained from smog chamber data. As
just described, we used the smog chamber data as an indicator of condi-
tional chemical reactions that may have been overlooked in past reaction
schemes because they had not been analyzed at the fundemental kinetic
level.
CBM-IV mechanism development also involved a number of application-based
requirements. For instance, the current generation of photochemical kine-
tic mechanisms may be used to study formaldehyde chemistry; therefore, the
formaldehyde mechanism should be explicit. Formaldehyde was previously
used as a surrogate for glyoxal (an aromatic ring fragment). However,
since formaldehyde is not a significant primary product of aromatic oxida-
tion by OH (Bandow et al., 1985; Bandow and Washida, 1985a, 1985b; Gery et
al., 1985, 1987), that association has now been removed and formaldehyde
is explicitly described in the CBM-IV. Other constraints, such as the
limitation of the number of species in a condensed mechanism, have led us
to develop a secondary reaction scheme for toluene and xylenes that is
less oriented toward the chemistry of explicit species and is more func-
tionally based. This shift is mandatory for all aromatic mechanisms to
some extent because of the large number of possible reaction branch points
and isomers.
141
-------
Choice of Aromatic Surrogate Species
Two aromatic surrogate species and their respective chemistries are now
utilized 1n Carbon-Bond Mechanisms to represent the chemistry of all aro-
matics. These surrogate species are TOL for toluene (mono-alky!benzenes)
and XYL for xylene (multi-alkylbenzenes). These surrogate species were
selected because the differentiation between surrogate chemistries must
represent the widest possible range of aromatic reactivities if they are
to provide the most appropriate representation of specific compounds. The
most important chemical features to differentiate in the surrogate chemis-
tries are the OH reaction rate and the secondary reaction scheme. Of
these two, appropriate representation of the secondary reaction scheme is
the most difficult to achieve because this chemistry is complex and varies
among aromatic species.
OH is the primary oxidizer of simple aromatic hydrocarbons under atmo-
spheric conditions. Ohta and Ohyama (1985) have shown that the OH-plus-
aromatics reaction rates correlate well with aromatic structural configur-
ations of the aromatic species (see Figure 4-2). Therefore, we currently
use the OH-plus-toluene rate constant for the OH-plus-monoalkylbenzene
reaction. The XYL rate constant is set to that of m-xylene. The sources
of these data are discussed next. We do not feel that the present uncer-
tainty in secondary products and their chemistry for higher molecular
weight alklbenzenes warrants a third aromatic surrogate. If more data
become available, however, and the requirement of a limited number of
species is lifted for future photochemical kinetics mechanisms, a good
argument for expansion of this scheme to a third (high molecular weight)
surrogate could be made on the basis of the high OH reaction rate for that
group.
An important issue that is less obvious than the OH reaction rate group-
ings involves the secondary reaction scheme for different aromatic spe-
cies. Two different levels of product formation are involved here.
First, we assume from limited evidence that aromatic ring reactions are
strongly dependent on the number and location of functional groups, but
only weakly affected by variations in types of alkyl groups. This assump-
tion allows the use of data for simple alkylbenzenes to describe the ini-
tial oxidation reactions and ring cleavage schemes for higher molecular
weight species. Second, the yields of the very reactive (and photolytic-
ally reactive) a- and conjugated y-dicarbonyl products demand scrutiny.
As noted by many researchers (Shepson et al., 1984; Bandow et al., 1985;
Bandow and Washida, 1985a, 1985b; Gery et al., 1985, 1987; Tuazon et al.,
1986), the yield of dicarbonyl species from the OH-plus-toluene reaction
is relatively low compared to yields from the xylenes and trimethylben-
zenes. For whatever reason, the prompt formation of dicarbonyl species
142
-------
6.0
~ 50
4.0
3.0
I
UJ
O
I
3
o
E
o
S 2.0
VI
8
3
o
« 1.0
00
I i i_
j i
I l I
^xv \
-------
after OH-aromatic reaction appears to be a definite feature of multi-
alkylbenzene/NOx smog chamber experiments, but is far less evident in
toluene systems. It is not clear whether there is lower energy for the
xylene transition states, or whether a different type of reaction inter-
mediate, leading directly to prompt radical formation, could be the result
of additional functional ring groups. It is clear that this difference in
chemical reactivity between mono-alky! (toluene-type) and multi-alkylbene
(xylene-type) systems can be very important for control strategies; there-
fore, these different chemistries must be addressed in the chemical mech-
anism. In this context we particularly include control strategies which
might affect the toluene-type emissions differently than the xylene-type
emissions.
TOL and XYL Chemical Mechanisms;
As noted, one critical difference between the product yields of toluene
and those of higher molecular weight aromatic species is the inability of
toluene to form high yields of reactive products (as indicated by a- and
y-dicarbonyl species) promptly after OH reaction. Such products are
stable but extremely thermally and photolytically reactive. Therefore,
their presence perpetuates the reactive nature of the system even after
the initial hydrocarbon is exhausted. Available smog chamber data indi-
cate that toluene systems rapidly consume NO, but limited toluene oxida-
tion is observed to occur well after the period in which reactive product
formation via this ring decomposition route would be severely limited by
the diminished NO concentration. Radical flux and ozone formation in
these toluene experiments initially appear to be very reactive, but then
rapidly terminate this stage, and often actually consume ozone throughout
the second part of an experiment.
This dichotomous behavior is somewhat unique for a "reactive" hydrocarbon
and may indicate that, unlike other aromatics and olefins, the product
mixture formed through the initial (relatively reactive) portion of these
experiments may develop Into less reactive species as experimental condi-
tions (availibility of NO) change. Said another way, it appears that
aromatics yield very reactive oxidation products during the initial por-
tion of an experiment, when NO is available, but that the overall system
becomes less reactive after that period because the resulting organic pro-
duct distribution eventually changes to less reactive species.
The extent to which this phenomenon occurs is species-specific, and is at
least dependent on the speed of product formation (OH reaction rate) and
the types of products formed during the different stages of reaction.
Toluene appears to be very sensitive to these reactivity parameters, which
results in the dichotomous nature of these smog chamber experiments.
144
-------
Xylenes and higher molecular weight aromatics react much faster with OH
and produce higher yields of reactive products. This faster reaction rate
decreases their sensitivity to NO concentrations and results in rapid
production of reactive species that can be more correctly approximated by
the assumption of prompt dicarbonyl (reactive product) formation.
Figure 4-3 illustrates results obtained from one side of the UNC chamber
for a relatively clear day near the summer solstice from a toluene/NOx
experiment with HC/NOX=14.7. Figure 4-4 illustrates results for the oppo-
site side of the chamber for a similar m-xylene/NOx experiment with an
HC/NOX value of 5.8. The different reactivities in these experimental
systems (which are representative of a number of similar toluene and xyl-
ene experiments) are apparent in comparisons of the ozone and hydrocarbon
curves. In many toluene experiments in the UNC chambers, the loss of
reactivity, which appears to occur just prior to ozone rollover, is
related to the system's limited ability to produce NO at that point. In
this example, the loss occurs shortly before noon, when N02» the primary
source of NO on a clear day, is sufficiently depleted and most gas-phase
NOX appears as nitrates (PAN).* Note from the toluene trace in Figure 4-3
that there is more toluene available to react, but that the radical flux
becomes abruptly limited. On the m-xylene side, the reactivity of the
system as measured by ozone production continues well past the depletion
of N02, and is not quenched until the system consumes all of the initial
hydrocarbon (it becomes m-xylene-limited) and is affected by the decreas-
ing actinic flux at the approach of sunset.
The reaction mechanisms of toluene and o- and m-xylene were specifically
studied because of the availability of smog chamber data and the relative
abundance of these species in urban atmospheric samples and automobile
exhaust. The initial product distributions for toluene and xylene oxida-
tion by OH are taken from the limited set of data available 1n the litera-
ture. OH reacts with toluene 1n the following manner:
H20
/ v
*OH
* Because the chemiluminescent NOX analyzer responds to some nitrates.
Including PAN, this phenomenon 1s not evident from the trace shown.
However, we know from wet chemical techniques and from subtracting PAN
data from N0£ that the N02 curve follows a generally Gaussian curve,
which can be approximated by the shape of the N02 buildup side, applied
to the second side, but extended to zero.
145
-------
I I I • I • I ' I • I • I • I
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
HOURS, EOT
0.0
0.0
6.0
0.0
1 ' I ' I ' I ' I
j r
Toluene
I ' I ' I ' I
I i 1 j l i l i I i i , i ,
6.0
5.0
4.0
3.0
2.0
1.0
0.0
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 '
HOURS. EOT
FIGURE 4-3. Experimental results for JN2784B. Top: NO, N02,
and ozone traces. Bottom: Toluene traces from two different
gas chromatographs.
-------
I • I ' I ' I ' I ' I I _ J J
8 9 10 11 12 13 14 15 16 17 18 19
HOURS, EOT
1.0
j I j 1 i I i I i I
I. . 1.1,1.1.1
2JD
05
" OO
5 6 7 8 9 10 11 .12 13 14 15 16 17 18 19
HOURS, EOT
FIGURE 4-4. Experimental results for JN2784R. Top: NO, N02,
and ozone traces. Bottom: m-xylene traces from two different
gas chromatographs.
147
-------
Reactions for the xylenes are analogous. We assume approximately 8 per-
cent hydrogen abstraction for toluene and 10 percent for m-xylene based on
the experimental results of Gery et al. (1985, 1987). The abstraction
pathway is minor and is not discussed here. Reactions of the alkylbenzyl-
peroxy radicals (B02, the methylbenzylperoxy radical; and XL02, the xylene
analogue) and subsequent species are included in the CBM-X (Table 1-2) as
X-131 through X-138 and X-152. A discussion of the abstraction pathway
and specific sources of reaction rate constants is found in Gery et al.
(1985) for toluene and Gery et al. (1987) for xylenes.
Most products are formed by the addition of OH to the aromatic ring struc-
ture. This OH adduct can either reversibly add an oxygen molecule or
loose a hydrogen atom to form an alkylphenol (CRES for cresol) and H02".
The approximate fractions of phenolic yields (f0n) were again taken from
Gery and coworkers. For toluene, fQH * 0.36; and for m-xylene; ff
0.20.
rOH
In the CBM-X, the overall OH-plus-aromatic fractions (including the hydro-
gen abstraction pathway) are given for toluene (X-128 through X-130) and
xylene (X-148 through X-151) as
and
OH + TOL > B02
OH + TOL > CRES + H02
OH + TOL > T02
OH + XYL > XL02
OH + XYL > CRES + PAR
OH + XYL > T02
OH + XYL > XINT
H02
8% (X-128)
36% (X-129)
56% (X-130)
10%
20%
30%
40%
(X-148)
(X-149)
(X-150)
(X-151)
148
-------
where XINT represents the mechanism Intermediate from which dicarbonyls
(MGLY) may form as a result of ring decomposition. Also, because T02 is
an intermediate of one less carbon than the parent xylene, a nonreactive
carbon is formed (a carbon atom is lost). This carbon atom 1s nonreactive
because the formation of 1 PAR would be incorrect in this case.
The reaction rate constant for the overall toluene-plus-OH reaction is
(Atkinson, 1986):
k(63) = 3.106 x 103 exp(322/T) pprn'^in'1
or
k(63)298 = 9'15 x lo
For the xylene-plus-OH reaction, Atkinson recommends (for m-xylene)
k(72) = 2.453 x 104 exp(116/T) ppm'^in'1
or
k(72)298 = 3.62 x 10
Beyond this point, the description of continuing oxidation, especially of
the chemistry of the OH-02 radical adducts, becomes very speculative.
Atkinson and coworkers (1980) proposed a rapid bridging of the 02 group
and relocation of the electron density, followed by a second, reversible
addition of 02:
°2
Some experimental evidence for such a structure was found by Gery and co-
workers (1985, 1987) in the form of p-alkylbenzoquinone products. In
either case, however, all schemes to date show either the initial OH-02
radical adduct (I) or the bridged OH-02 adduct (II) reacting with NO to
form N02 and products (generally, reactive ring fragments). (The func-
tional description of most mechanisms is similar up to this point.) In
terms of species, these reactions are in the CBM-X:
T02 + NO > N02 + OPEN + H02 (X-145)
T02 + NO > TNTR . (X-146)
XINR + NO > N02 + 2MGLY + 2PAR + H02 (X-153)
149
-------
For the first two reactions, we use an overall reaction rate constant of
1.20 x 10~4 ppnT1 min'1, with a 10 percent nitrate yield to give
k(X-!45) = 1.08 x 10'4 ppm^min"1
and
k(X-146) = 1.20 x 10"3 ppuf TnlrT1
The rate of the XINT plus NO reaction is also k(X-153) = 1.20 x 104
ppnf *min~ . This reaction represents the fraction of XYL molecules that
promptly form dicarbonyl products. We describe the chemistry of CRES, a
general phenolic product, and MGLY and OPEN, which are dicarbonyl species,
later in this section. However, we note here that our approach to the
formation and chemistry of dicarbonyls is to lump all of the dicarbonyl
reactions into one general surrogate species (OPEN) and one other explicit
species, methylglyoxal (MGLY). We believe this is an appropriate approach
because (1) the uncertainty of products and their chemistry subsequent to
dicarbonyl formation via ring decomposition is large, and (2) the timing
of variations between data and model results appears to result from poor
descriptions of mechanistic branching points prior to dicarbonyl forma-
tion. Therefore, these dicarbonyl-lumped species contain the essential
experimentally determined features of all the known dicarbonyl products
and serve to confine the uncertainty to only two species (limiting exten-
sive speculation about a number of unstudied secondary products). This
allows some investigation of the pre-fragmentation reaction products and
their reactions.
The greatest area of uncertainty in aromatic mechanisms lies in the radi-
cal reactions following ring fragmentation. One assumption currently used
in many photochemical models of aromatic oxidation is the idea of "prompt"
product formation. In models such as the CBM-IV, this idea can be spec-
ifically instituted when the (generalized) reaction set,
RH + OH > R'(+ H20)
R- + Oo > R02*
R02' + NO > N02 + products,
is simplified to
RH + OH > X02 + products (02 constant),
with the X02 operator defined as
X02 + NO > N02. (79)
This concept was formally introduced by Whitten et al. (1985b) to minimize
the number of species and reactions in a chemical mechanism by grouping
150
-------
the common NO-to-NOo conversion function of all R02* species into the X02
operator, thereby eliminating explicit description of specific R02* spe-
cies in favor of their "promptly" formed products. A similar method has
since been employed by Carter et al. (1986).
This approximation drifts from the explicit scheme in situations where (1)
low NO concentrations are combined with a highly reactive R02* or (2)
another R02* reaction competes with the NO reaction (even at relatively
high NO concentrations). The first situation (low NO) is typically han-
dled by the addition of the X02 self-reaction, which is also assumed to
yield the same products as the NO-to-N02 conversion reaction:
X02 + X02 > . (80)
In the second case, however, where there is competition with NO for R02*
through another reaction pathway that produces different products, the X02
operator must be eliminated and the R02* reactions explicitly represented
in all other mechanisms that do not use a X02-type approximation; a simi-
lar effect is found for toluene since there are no species that compete
with NO for the toluene-02* adduct radical.
On the basis of the smog chamber data and the product yield data for tol-
uene, we believe that this prompt formation of products, especially highly
reactive o- and conjugated y-dicarbonyls, incorrectly describes the secon-
dary chemical reactions of the OH-toluene reaction when NO concentration
and production have become limited. One possible pathway that has not
been considered is alternate reactions of the OH-02 adduct radicals
(T02). These species probably do not have long reactive half-lives, and
1f they cannot react with NO, must be lost by an alternate reaction path-
way. Our methodology in developing this most recent mechanism has been,
in part, to estimate the mechanism, products, and kinetics of an alternate
reaction of the T02 radical. Such a reaction probably becomes important
only at low NO concentrations, and produces less reactive products than
the NO-to-N02 conversion process. Through an extensive set of simulations
and sensitivity tests, we have estimated a T02 decomposition of the form
T02 > H02 + ORES ( + 02). (65)
The unimolecular rate of this reaction was estimated to be:
k(65) = 2.5 x 102 min"1
As we will show in the next section, inclusion of this reaction signifi-
cantly improves the simulation of ozone. We note that though the earlier
OH-02 adduct radical (I) could conceivably loose oxygen and form cresol
and H02, we use cresol here only because it is already available in the
151
-------
CBM and describes a secondary aromatic product that is less reactive than
dicarbonyls and that removes nitrogen from the system.
Aromatic Product Species
The species, CRES, represents alkylphenolic products formed after OH addi-
tion to the aromatic ring. In the case of toluene, the CRES species
formed is cresol, typically o-cresol (o-methylphenol). The CRES chemical
reactions in the CBM are formulated for that compound because the chemis-
try of higher molecular weight alkylphenol homologues and other cresol
isomers has been far less studied. For the other cresol isomers, we
assume the same chamistry as o-cresol, and for the higher molecular weight
homologues, we include alkane surrogates (PAR) groups upon formation.
Thus, the dimethylphenol compounds formed from xylene oxidation by OH are
represented by CRES plus PAR.
The chemical reactions of cresol and larger alkylphenols are summarized in
Atkinson and Lloyd (1984); additional information is found in Gery et al.
(1987 and 1985) and Kerr and Calvert (1985). The reader is referred to
these works for a complete description of the current state of knowledge
with respect to alkylphenol chemistry. We represent this chemistry in the
CBM-X with
OH + CRES > CRO (X-139)
OH + CRES > CR02 (X-140)
N03 + CRES > CRO + HN03 (67)
CRO + N02 > NCRE (68)
CR02 + NO > N02 + OPEN + H02 (X-143)
CR02 + NO > N02 + ACID + H02 (X-144)
It should be noted that a discrepancy in N03 oxidation rates appears in
the abovenoted texts. The rate constant of Atkinson and Lloyd (1984) was
based on a dinitrogen pentoxide equilibrium constant that has since been
revised. Based on the new equilibrium constant, Kerr and Calvert (1985)
recommended a rate constant of
k(67) = 3.25 x 104 ppm^min'1 .
Our overall rate of hydroxyl reaction with CRES is from Atkinson and Lloyd
(1984):
k(66) = 6.1 x 104
152
-------
with an assumed OH addition fraction of 60 percent and a 40 percent
abstraction of the phenol hydrogen by OH to yield CRO, a methyl phenoxy
radical. The remaining reactions are very speculative, and are based on
the review by Atkinson and Lloyd (1984). The ACID species represents
organic acids formed from aromatic ring decomposition of the alkylphenol
peroxy radical (the 02 adduct formed after OH addition to CRES).
Methylglyoxal (CH3C(0)CHO) is a dicarbonyl ring fragment found in aromatic
systems containing ring methyl groups. It is very reactive with the
hydroxyl radical and also photolyzes to radical products. These reactions
are represented in the CBM-X by
OH + MGLY (+ 02) ------ > MGPX (+ H20) (X-154)
MGPX + NO ------ > N02 + C203 (+C02) (X-155)
MGLY — hv— > C203 + CO + H02 (X-156)
where MGPX represents a peroxy radical formed after hydroxyl abstraction
of the aldehydic hydrogen.
The kinetic information used in the CBM-X for the OH reaction is from
Atkinson and Lloyd (1984):
k(X-154) = 2.60 x 104 ppm^min'1 .
The MGPX is assumed to react with NO at
k(X-155) = 1.20 x 104 ppm'^in"1 .
Photolysis of MGLY has been studied (Plum et al., 1983), though the uncer-
tainty of the quantum yield data causes calculated j-values to be very
uncertain in the lower wavelength region of the surface solar spectrum.
On the basis of smog chamber data, we use the following rate expression in
the CBM-X:
= 9-64 x j m1n"1 •
where JncHOr is the J~value for formaldehyde photolysis to radicals (reac-
tion 38).
The species OPEN represents a conglomerate of possible dicarbonyl species
with up to seven carbons. We felt that, given the limited information
concerning dicarbonly chemistry, speculation about the specific structure
and reactions of individual compounds was pointless. Instead, we combined
the expected kinetics of the anticipated dicarbonyl product structures
into one general species that could represent the mass of all products.
153
-------
Hence, the final rates used for these species are based on the goodness-
of-fit for a large set of smog chamber simulations. In addition, since
photochemical kinetics mechanisms used in large air quality simulation
models require small numbers of species, we felt it was best to implement
the improvements to the T02 chemistry instead of using additional
dicarbonyl species (all of which are very reactive). The photolysis of
OPEN and the OH reaction in the CBM-X are
OPEN — hv— > C203 + H02 + CO (X-159)
OH + OPEN ------ > OPPX + C203 + H02 + CO (X-157)
NO + OPPX ------ > N02 + FORM + H02 + CO (X-158)
The kinetic expression used for OPEN photolysis is similar to that of
MGLY:
%EN = 9'04 x JHCHOr
We assumed the rate for kx-157 to be
k(X-157) = 4.40 x 104 ppm^min'1 ,
and that of k to be
k(X-158) = 1.20 x 104 ppm'^in'1 .
The product distributions are very uncertain. Many of the larger assumed
products were 5 or 6 carbon a- and y-dicarbonyls with conjugated alkene
groups. We assumed photolysis of only a-dicarbonyl products. Hydroxyl
reaction was even more speculative because of the ability of OH to react
rapidly with both the olefinic and the aldehydic carbons of a-dicarbonyl s.
Ozone reaction with the alkene bonds of larger OPEN species was assumed to
proceed along the lines pointed out by Dodge and Arnts (1979) for simpler
olefins. We used a hypothetical product distribution to derive product
stoichiometries for these reactions. These empirical results are shown in
the CBM-X as reactions X-160 through X-164 (Table 1-2). A more explicit
version of the same reactions is given in the following condensation dis-
cussion and in the CBM-IV (Table 1-3). For the reaction of ozone with
OPEN, we use the overall rate expression
k(71) = 8.03 x 10~2 exp(-500/T) ppm^min'1 ,
with
1.50 x 10~2 ppnf ^min"1 .
154
-------
As discussed in the next section, the chemical reactions now used to
describe the photooxidation of aromatic hydrocarbons provide simulation
results for smog chamber data that are much improved over those of earlier
models. On the other hand, because fundamental information on aromatic
systems is limited, this mechanism does not rest on as strong a basis as
other sections of the CBM. Rather, it is a combination of the available
fundamental information and an empirical description of additional infor-
mation available from more complex smog chamber experiments.
Condensation of the Reaction Schemes;
As noted, the reaction scheme for xylenes just given should ideally
include the reactions of xylene carbon product species. Because that
chemistry is extremely uncertain, however, we assume the chemistry of BC>2,
T02 and CRES as a logical approximation. Condensation of the extended
chemistry proceeds as follows: For the initial reactions of OH with tol-
uene and xylene, all species except the abstraction products are accounted
for by direct combination of reaction stoichiometry and rate constants
(see Table 1-3). The abstraction products are represented by
B02 = X02 + H02
and
XL02 = X02 + H02 + PAR ,
where X02 is generally the NO-to-N02 conversion function (as previously
described). Hence, the reaction of B02 with NO to form N02, H02, and
benzaldehyde is represented by the X02 operator and product H02, with the
BZA product omitted. A similar representation is made for xylenes, except
that PAR is used to represent the additional carbon produced from the
reaction of XL02 with NO. After combination of these reactions, the
resulting reactions are
OH + TOL > 0.08 X02 + 0.36 CRES + 0.44 H02
+ 0.56 T02 (63)
OH + XYL > 0.50 X02 + 0.20 CRES + 0.70 H02
+ 0.30 T02 + 0.80 MGLY + 1.10 PAR (72)
where XINT is eliminated with an X02 operator but T02 chemistry is expli-
cit, as noted:
T02 + NO > 0.90 N02 + 0.90 H02 + 0.90 OPEN (64)
T02 > H02 + CRES . (65)
155
-------
The reactions of cresol are combined to
OH + CRES > 0.40 CRO + 0.60 X02 + 0.60 H02
+ 0.30 OPEN (66)
N03 + CRES > CRO + HN03 (67)
CRO + N02 > (68)
The OH and ozone oxidation reactions of MGLY and OPEN are condensed alge-
braically and through the elimination of MGPX and OPPX via the X02 opera-
tor, to yield:
OH + MGLY > X02 + C203 (73)
OH + OPEN > X02 + 2.0CO + 2.0H02 + C203 + FORM (70)
03 + OPEN > 0.03ALD2 + 0.62C203 + 0.70FORM
+ 0.03X02 + 0.69CO + 0.080H
+ 0.76H02 + 0.20MGLY (71)
ISOPRENE
Our objective in developing this portion of the CBM-IV was to provide a
chemical representation for biogenic hydrocarbons by deriving a condensed
mechanism for isoprene and estimating valid carbon bond splits for a-
pinene. In the former task, we wished to limit the product species con-
sidered in the condensed isoprene mechanism to those already included in
the CBM-IV.
The first step in implementing this explicit Isoprene mechanism was the
development of a good kinetic representation of the initial Isoprene
oxidation reactions. These oxidation reactions occur with 0, N03, OH and
03; the last two are usually most important in daytime atmospheric
chemistry. We have reviewed the reaction rate constants published in
recent studies, and revised our earlier explicit mechanism rates (Killus
and Whitten, 1984), when appropriate.
Since a rate constant for the 0 atom reaction with Isoprene has not been
well studied, it was assumed to be the sum of 0 atom additions to each of
the alkene bonds in the isoprene molecule. The dialkylated bond
characteristics were estimated to be similar to those of isobutene
(Atkinson and Pitts, 1977), and the monoalkylated bond was estimated to be
equal to the general 1-alkene (OLE) bond in the CBM (e.g., propene). The
resulting rate constant is
k(75) = 2.70 x 104 ppm^min'1.
156
-------
The OH rate constant for reaction with isoprene was estimated by averaging
the rates of Atkinson and Aschmann (1984) and Kleindienst et al. (1982):
k(76) = 1.52 x 105 ppm^min"1.
The reaction rate with 03 is taken as
k(77) = 1.80 x 10'2
the same as the rate constant for 1-alkenes (OLE) in the CBM and within
the error bands of the studies reviewed by Atkinson and Carter (1984). As
noted in Killus and Whitten (1984), the reaction of 03 with isoprene shows
evidence of being a complex process, dissimilar in the nature of secondary
reactions to both monoalkenes and other dialkenes (e.g., butadiene). This
is reflected in our product surrogate choices.
Finally, the rate of reaction with N03 is derived from the rate constant
presented by Atkinson et al. (1984):
k(78) = 4.70 x 102 ppm^min"1.
Our explicit isoprene mechanism was initially developed by Killus and
Whitten (1984). In the development of that mechanism we found the
following significant phenomena to be associated with isoprene oxidation:
Highly reactive products, including methacrolein and methyl vinyl
ketone. Some product reaction with ozone was also indicated, as
would be expected from the olefinic character of these two products;
A high rate of radical production, presumably from photolysis of iso-
prene products;
A high yield of PAN as well as a PAN-like compound, probably from
methacrolein; and
A high yield of formaldehyde.
Because the complex secondary chemistry of isoprene and its oxidation
products (methacrolein, methylvinyl ketone, and possibly methylglyoxal)
precludes a direct condensation to a few representative species, these
reactions were represented in condensed CBM-IV format by the use of
existing carbon-bond species. Our intent was not to algebraically reduce
the explicit reaction scheme, but to simulate the major chemical features
(i.e., ozone production, formaldehyde yields, and the temporal character-
istics of radical and PAN-type compounds). This Was done by simulating
157
-------
the UNC data set of isoprene/NOx experiments (some of which are shown and
discussed in Section 6).
Our general methodology for the formulation of the condensed isoprene
mechanism was based on the following rationale: The olefinic nature of
isoprene products would best be simulated by ethene, since the alkene
bonds of methacrolein and methylvinyl ketone are partly deactivated and
resemble ethene in their reactivity to OH and 03. This representation
also gives a high yield of formaldehyde from the secondary oxidation of
ethene, which resembles the formaldehyde yield of isoprene. The formation
of PAN and PAN-like compounds from Isoprene was simulated with a primary
yield of both acetaldehyde and acetylperoxy radical. Inevitably, this can
result in an overprediction of measured PAN, since simulated PAN includes
the PAN analogues as well. The radical yield for products was simulated
by the formation (and photolysis) of methylglyoxal. These separate pro-
duct yields were adjusted to give the best fits to the mid-range of hydro-
carbon-to-NOx-ratio isoprene experiments, in which notable double ozone
peaks become evident. The double ozone peaks in isoprene NOX experiments
are caused by NOX limitation of PAN formation, ozone reaction with iso-
prene and isoprene products, and continued ozone formation by PAN decompo-
sition at elevated temperatures. Simulation of the double peak effect
gives some indication that a reasonable balance has been achieved for the
multiple processes occurring in isoprene oxidation. The condensed
isoprene mechanism resulting from simulations of the UNC isoprene/NOx data
set is shown as reactions 75 through 78 in Table 1-3.
In addition to improving the isoprene mechanism in the CBM-IV, we utilized
the UNC a-pinene/NOx data set to test our carbon-bond parameterization of
that species. In these simulations we found that a better carbon-bond
parameterization for a-pinene would be 0.5 OLE, 1.5 ALD2, and 8.0 PAR.
This is a slight alteration from earlier compilations. These results,
along with results of the UNC smog chamber simulations are discussed
in Section 6.
158
-------
REPRESENTATION OF SMOG CHAMBER PROCESSES AND CHARACTERISTICS
This section discusses the information and assumptions used to represent
smog chamber-dependent processes (e.g., heterogenous reactions and dilu-
tion) and reactor characteristics (e.g., in-chamber photolysis rates and
temperature profiles) that are not included in the homogenous gas-phase
chemical reactions of the CBM. Section 6 utilizes the information
developed here to compare CBM-IV simulation results and experimental data
from various smog chamber facilities. In any mechanism evaluation using
smog chamber data, it is first necessary to clearly delineate the
mechanistic techniques used to characterize artificial processes that
occur in the experimental system (the smog chamber). In this way, the
artificial processes represented in smog chamber simulations can be
included in the mechanism for testing and then removed prior to simulation
of the atmosphere. Therefore, before presenting the experimental data and
simulation results we first describe our methodology for handling chamber-
specific processes (photolysis, wall radicals and water).
PHOTOLYLTIC PROCESSES
Almost all of the experimental data used to develop and test the CBM-IV
performance was generated in smog chamber facilities at
(1) The University of North Carolina (UNC);
(2) The University of California at Riverside (UCR) evacuable cham-
ber (EC);
(3) Indoor Teflon chamber (ITC); and
(4) The Battelle Columbus Laboratories chamber (designated AVOC).
To achieve accurate simulation of photochemical smog chamber data photoly-
sis rates (j-values) must be determined throughout each simulation. We
have already discussed the method and data used to calculate j-values for
the relevant species in the CBM in Section 3; however, we have not yet
discussed our calculation of the actinic flux inside individual smog cham-
bers.
159
-------
The radiant energy of indoor chambers is generally assumed to be constant
over the period of an experiment (therefore, j-values are also assumed to
be constant). These chambers use artificial illumination, which can dif-
fer from natural sunlight (Pitts et al., 1979; Whitten et al., 1979; Car-
ter et al., 1986). The spectral distribution is usually reported for each
experiment, though some facilities report at longer time intervals. Over
the longer term, variability in these chambers is higher than in outdoor
chambers, because of lamp changes in the irradiation sources. In the UCR-
EC, there have also been changes in transmission characteristics of the
Pyrex windows used to admit light (Whitten et al., 1979; Carter et al.,
1986), resulting from the prolonged exposure of those windows to the
Intense radiation of the solar simulator light source.
On the basis of the reported light spectrum for the black!ight source used
in the UCR Indoor Teflon Chamber (see Table 5-1), we have calculated the
photolysis rates for a number of important species. For the series of
multiday validation experiments (ITC-625 to ITC-637), having a reported
of 0.3 min , we obtain the following j-values:
= 5.5 x 10"4 min'1
JHCHOs = 5-8 x 10~4 min"1
= 7.2 x 10'5 min'1
= 3.2 x 10'4 min'1
The estimates of photolysis rates calculated here differ significantly
from those of Carter et al. (1986). We have traced the differences in
•^HCHOr to two ^actors: (1) Our use of the absorbtion cross section data
of Bass et al. (1980), rather than the averaged values found in the NASA
review (DeMore et al., 1985); and (2) The N02 photolysis estimates of Car-
ter et al. appear to be based upon earlier NASA recomendations (DeMore et
al., 1983) rather than the later, 1985 compilation (DeMore et al.,
1985). The combination of these two factors accounts for about a 14 per-
cent difference 1n the JHCHOr values calculated. The estimate used by
Carter et al. for this rate in the ITC is about 6.5 x 10~4 min'1 (Carter,
1987). It is our belief that this is a significant overestimation of
^HCHOr in tne *TC and could cause the overly reactive simulations for the
ITC experiments in Carter et al. (1986).
The spectral measurements in the UCR Evacuable Chamber (EC) are marginal
at best and often inadequate for calculating photolysis rates in the
shorter wavelength bins. The most prominant higher uncertainty of these
data is the fact that the poorest precision in the EC spectral data always
occurs in the region between 280 and 310 nm, where many key photoreceptors
absorb light. In addition, information about the'run-to-run light source
consistency and spectrum determination accuracy is difficult to obtain.
160
-------
TABLE 5-1. Relative spectral distribution for the blacklights 1n the
SAPRC 6400-liter Indoor Teflon Chamber (ITC)a (Source: Carter et al.f
1986).
Wavelength
(nm)
295.7
296.7
298.0
300.7
302.1
304.3
306.3
310.0
313.0
313.7
315.0
315.3
316.7
317.3
318.7
323.3
330.0
336.7
343.3
Intensity1*
0.0
0.0019
0.0024
0.0018
0.0043
0.0058
0.0057
0.0140
0.0728
0.0846
0.0879
0.0856
0.0581
0.0564
0.0858
0.1495
0.331
0.563
0.851
Wavelength
(nm)
347.8
350.0
351.7
353.7
356.7
360.0
363.3
364.3
364.7
365.5
366.9
368.5
370.0
376.7
383.3
390.0
396.7
393.0
404.3
Intensity**
0.952
0.972
0.990
1.00
0.982
0.925
0.829
0.859
0.869
0.870
0.835
0.685
0.639
0.442
0.261
0.1423
0.0885
0.0535
0.223
Wavelength
(nm)
405.0
406.0
406.7
408.7
412.0
416.7
423.3
430.0
434.3
436.1
436.7
437.3
438.1
439.7
443.3
450.0
460.0
Intensity1*
0.258
0.264
0.251
0.0596
0.0387
0.0348
0.0298
0.0227
0.0217
0.641
0.709
0.712
0.651
0.0169
0.0140
0.0081
0.0
aAs measured on June 30, 1983
Intensity in terms of photons per unit area per unit time, normalized so
the maximum intensity -1.0 units.
161
-------
For Instance, the recommended spectral distributions (see Table 5-2) show
a significant break between EC-233 and EC-237 such that calculated
values for these two spectral distributions differ by 20 percent (6.1 x
10"^ min"1 to 5.1 x 10 min"1). However, the concentration data reported
for EC-232 and EC-246, a pair of replicate experiments, differ only by the
initial N0/N02 ratio and the length of the experiments (see Figure 5-1).
All available tests of radical input levels (hydrocarbon and NOX decay
rates) indicate no appreciable difference between the reactivity of the
two experiments. Were the formaldehyde photolysis rate to differ by the
amount suggested in Table 5-2, this would not be the case.
Our best estimate of JHCHOr on the basis of observed reactivity for the
seven component experiments (see Section 6) is 7.7 x 10" min" . This
value appears to be close to the value that would be calculated using the
formaldehyde absorption cross sections and quantum yields in Carter et al.
(1986) for the EC-237 to EC-257 spectral set.
The UNC facility uses a pair of outdoor, side-by-side Teflon reactors that
are illuminated only by the daily transit of the sun. Daily variations in
light-source can have a strong effect on these outdoor chamber studies
since some atmospheric conditions serve to attenuate light. On the other
hand, "clear sky" experiments held a year apart at the UNC chamber show
little variability, and replicate experiments on both sides of the two-
sided chamber typically show essentially identical results (Jeffries et
al., 1976). This is because, unlike indoor chambers, there has been no
substantial long-term variation in the energy output of the sun. Thus
these outdoor experimental conditions are more akin to those of the
ambient atmosphere, and a better test of the mechanism's performance for
real conditions. On the other hand, the indoor chambers are far better
controlled and do not suffer from the intervention of clouds or haze
between the light source and reactor. Hence, we utilized both indoor and
outdoor data in the development of the CBM, but use a larger percentage of
UNC data in the following evaluation because of their more realistic
experimental characteristics. It is therefore necessary to describe our
method of calculating the in-chamber actinic flux as it varies through the
day at the UNC outdoor facility.
The UNC data sets provide data from an Eppley Ultraviolet (UV) instrument,
which is located outside the chamber on an elevated tower. These data are
usually accompanied by total solar radiation (TSR) and temperature mea-
surements. It was originally envisioned that these measurements alone
would be sufficient to provide an accurate description of radiation within
the chamber. However, as additional experiments have been performed and
the data set has become more comprehensive, new information concerning the
measurement method, the effect of chamber structure, and the variation in
162
-------
CJ
TABLE 5-2. Summary of average relative9 spectral distributions used 1n
modeling the SAPRC Evacuable Chamber (EC) experiments by Carter et al.,
(1986).
Run Wavelength (nm)
Nos. 290 300 310 320 330 340 350 360 370 380 403 430 500
106 0.001 0.039 0.146 0.357 0.572 0.789
113-116 0.002 0.034 0.128 0.318 0.533 0.758
120-125 0.000 0.047 0.121 0.292 0.532 0.767
142-160 0.000 0.018 0.082 0.238 0.464 0.706
161-172 0.000 0.016 0.077 0.222 0.449 0.705
177-178 0.001 0.031 0.167 0.379 0.627 0.837
216-217 0.000 0.023 0.137 0.347 0.608 0.828
230-233 0.000 0.024 0.131 0.327 0.593 0.810
237-257 0.001 0.016 0.110 0.289 0.528 0.773
264-287 0.001 0.018 0.122 0.316 0.565 0.787
304-309 0.000 0.020 0.125 0.314 0.558 0.798
314-320 0.001 0.019 0.131 0.369 0.695 0.970
327-331 0.000 0.003 0.095 0.283 0.517 0.788
. 334-340 0.000 0.000 0.093 0.280 0.525 0.783
343-346 0.000 0.008 0.106 0.284 0.531 0.775
389-392 0.000 0.004 0.099 0.272 0.529 0.782
435-465 0.000 0.000 0.083 0.260 0.521 0.773
596-603 0.000 0.000 0.044 0.208 0.450 0.732
898-903 0.000 0.000 0.000 0.018 0.122 0.454
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.119
.172
.165
.214
.189
.093
.089
.168
.170
.117
.131
.348
.142
.161
.149
.173
.189
.239
.338
.291
.417
.451
.241
.228
.316
.418
.275
.309
.557
.342
.382
.359
.400
.366
.210 1.458
.316
.467
.406
.605
.715
.371
.344
.459
.634
.396
.454
.691
.486
.549
.530
.587
.521
.425 1.557 1.618
.682 1.921 2.154
.551 1.736 1.788
.892 2.092 2.419
.107 2.334 2.874
.598
.534
.753
.970
.502
.602
.804
.691
.778
.729
.771 2.134
.633 1.587
.939 2.152
J.350 3.147
.625 1.688
.756 1.939
.915 1.861
.878 2.167
.998 2.474
.941 2.277
.817 2.110 2.684
.699 1.898 2.459
.734 2.069 2.457 3.187
.520 1.968 2.234 2.766 3.034 3.596
"Normalized to give 1.0 at 350 nm.
-------
Ol MO NO
e.1 -
>**»
V
»*»v '
» •* * *Ijs » i
V •* :!sto°°
* o »»
* * *o°
• * *»_
OJ -
O.16 -
O.1t -
0.14 -
O.1* -
O.1 -
oas -
en -
O.1 -
o»
o
o*
O 4-
FIGURE 5-1. Comparison of experimental data for EC-232 and EC-246. Top:
NO, N02 and 03; Middle: Ethene decay; Bottom: NOX decay.
164
-------
detector(s)' response with time has become available and can now be incor-
porated into uniform data preparation methods. An investigation into the
chamber-specific radiation processes is currently being conducted at UNC
to provide greater certainty in these actinic flux calculations. In our
mechanism development study, we have used the considerable amount of
available data to formulate a clear-sky, in-chamber, radiation algorithm
that uses the reported outdoor Eppley UV voltages. A necessary adjunct to
this data reduction algorithm is a theoretical solar intensity function,
which we use to compare and unify the data, determine true solar time, and
fill in gaps in the UNC UV data. We discuss this solar function prior to
describing the method for calculating in-chamber light intensity from the
outdoor Eppley voltages provided by UNC.
Solar Function
Because the sun is a well-defined light source with an angle of elevation
that can be mathematically described at any time, it is possible to derive
an expression for clear-sky surface solar radiation with respect to
time. Such a solar function is unaffected by chamber and instrument
variations. Therefore, it can be used as a standard for both radiation
magnitude and timing with any of the UV data sets. It can also be used as
a basis for estimation of UV data when there are gaps in the measurements
and is beneficial when the laboratory clock data are inaccurate.
The actinic flux inside the chamber is required to determine the in-cham-
ber photolysis rates UIN(P) for species p]. However, both the values
from the solar function and the UNC UV measurements provide values for
outside the chamber. Hence, a correction function must be derived so that
the outside j-values (JQUT^P)! can ^e use(* to Provide an estimate of the
in-chamber photolysis rates. We define a zenith-angle(u>)-dependent cor-
rection to J(P) tnat wl11 y^eld J(P) as
where
JIN(P,U>) and JouT(P>a)) are the zenith-angle(u>)-dependent, in-chamber
and outdoor photolysis rate constants for species p,
fj(t) is a daily correction factor to light intensity for distance
between the earth and sun,
fT(t) is a daily correction factor for atmospheric turbidity in North
Carolina,
165
-------
fc(u) is a zenith-angle-dependent correction factor for the effect of
chamber structure on the radiation field, and
fa(p) is the species(p)-dependent fraction of light transmitted
through the Teflon film of the chamber.
Solar distance corrections [f,j(t)] result primarily from the variation in
solar intensity due to changes in the distance to the sun (Becker et al.,
1966). The extraterrestrial flux in January is almost 7 percent greater
than in July.
The correction for local atmospheric turbidity (fT(t)] is based on the
work of Peterson et al. (1981). Figure 5-2 shows the curve suggested by
Peterson and coworkers. We have flattened this curve during July and
August to provide closer agreement with the measured annual variation of
N02 photolysis found in the UNC solar flux data. Daily turbidity values
are determined through linear interpolation between monthly averages of
the values from Figure 5-2. The value for fT(t) is then determined by
ft(t) - e-°-10*T.
where T = 2.303 x turbidity (Figure 5-2). The constant, 0.10, was deter-
mined to provide a good fit with the observed seasonal variation at UNC.
Because the location of the UV meter is outside and slightly above the
smog chamber, an outdoor to In-chamber conversion factor [fc(u>)J is
applied to account for smog chamber structural effects, including reflec-
tion by the Teflon film, internal reflection of the aluminum floor, and
attenuation by the chamber frame. The calculation of the variable magni-
tude of these effects provides the greatest uncertainty involved in the
determination of in-chamber intensity. This ratio is assumed to be
directly proporational to the amount of sky covered by the frame. This is
determined geometrically to be between 5 and 8 percent. The primary
uncertainty associated with chamber structure results from the poorly
documented effect of reflection. The aluminum floor has been shown to
enhance N02 photolysis inside the chamber (Saeger, 1977), and the external
and internal reflections of the Teflon film are only now being addressed
by UNC.
The best j-value experiments available at the time of our study were per-
formed in October 1976, near the end of the experimental season. There is
no information available for experiments performed near the summer
solstice and there have been no actinometric measurements (until the cur-
rent UNC study) since 1976. On 13 October 1976, several instruments were
167
-------
Source: Peterson et al, 1980
FIGURE 5-2. Data Points are daily average atmospheric turbidity in North Carolina
(solid line represents the turbidity values used in the calculations).
16S
-------
used to measure jN02 inside and outside of the UNC chamber. Results from
this experiment are shown in Figure 5-3. A maximum JNQ£ enhancement due
to chamber effects of 1.15 ± 6 percent was found on that day at minimum
solar zenith angle of about 44 degrees (at 0900, u was about 60
degrees). As expected from consideration of chamber geometry, these
results indicate that the enhancement due to reflection increases with
decreasing zenith angle. Given the uncertainty associated with the magni-
tude of the correction, this result is important because the necessary
correction appears to be minimal in the hours immediately after sunrise
when proper representation of photochemistry is highly critical.
Given the values shown in Figure 5-3 and noted above, it seems likely that
a slightly higher enhancement would occur at the calculated annual minimum
u near 11 degrees for Pittsboro, North Carolina. Therefore, we use a fac-
tor of 1.20 as our maximum chamber effects enhancement. As noted, the
algorithm is dependent on zenith angle (a measure of true solar time)
rather than on laboratory time.
For u greater than 72 degrees: fc(") = 1.00;
for u) between 44 and 72 degrees: fc(o>) = 0.85 + 0.40 x cos(u>);
and for u> less than 44 degrees: fc(u) = 1.20.
Finally, fa(p) describes the species (p) related effect of light attenua-
tion by the Teflon film of the chamber. Such a process is probably wave-
length-variable and could change as the angle of incidence between the
direct solar beam and the Teflon surface varies. Some measurements pre-
sently being being performed at UNC may clarify this. At present,
approximations involve large uncertainty. Jeffries and Sexton estimated
the film transmission factors for JncHOr* ^HCHOs» ^OlD and ^ALD2r to be
0.50, 0.60, 0.40 and 0.45, on the basis of a preliminary spectrophotometer
analysis of the film. It is also possible to utilize computer simulation
models to fit specific smog chamber experiments by varying the transmis-
sion factors to achieve the best fit. Using the j-values described in
Section 3, the film transmission factors for JncHOr' ^HCHOs and J>ALD2r
were found to be 0.84, 0.84 and 0.88. These ratios are based on the dif-
ferences between the j-values and the values previously used in the CBM-X
(Whitten et al., 1985a), which were determined by the best fit approach.
For JQiDt the factor of 0.45 is determined on the basis of the ozone
actinometry calculations performed by UNC and analyzed by Systems Applica-
tions. These data, though similar to the preliminary UNC measurements,
suggest a sharper decline in UV transmission closer to the atmospheric
cutoff region.
The variable JouT^P*1") describes the zenith-angle-dependent change
in photolysis rate for clear-sky, outdoor actinic flux. The value of
169
-------
Tc
.910
.819
.728
.637
.516
.155
.361
.273
.182
.091
.000
_ . , • , • 1 . 1 I 1 I , I , I , I , , 1 • , 1 , • 1 I
OCTOBER 13, 1976 _
— —
- * -
- * v*; •*; % -
4 0*0 o ^
— 4»*° **%. —
i /•" A -
V* *^ o
- /
' , 1 1^1 1 • 1 , 1 , 1 , 1 , 1 • 1 ,• 1 • 1 1 1 1 1 •"
,31U
.819
.728
.637
.516 f
.155 I
.361 J
.273
.182
.091
nnn
7 8 9 10 11
12 13 H 15 16 17 18 19 20
HOURS EOT
FIGURE 5-3. Sped fie photolysis rate of nitrogen dioxide and incidental total solar radiation
and ultraviolet radiation as a function of time for 13 October 1976.
-------
) 1s dependent on the ability of molecules to absorb incident
light and photodissociate. Hence, as described in Section 3, this rate
can be calculated from the products of clear-sky actinic flux (UV(u,Ax)],
absorption cross-section (op), and quantum yield (p(AX)
AX
The values for UV are determined from UNC data or, if necessary, from the
actinic flux calculations of Demerjian et al. (1976) as noted by Jeffries
and Sexton (1987). The sources for absorption cross-section and quantum
yield data are described in Section 3.
UNC Data Reduction
It has been known for several years that the Eppley UV meter sensitivity
varies between calibrations. In addition, various artificial processes
such as chamber shadowing, floor reflection, and light attenuation within
the chamber and meter due to temporary dew or ice formation, have been
suggested as causes for inaccurate estimation of in-chamber radiation.
Also, since changes in surface spectral distribution result in different
rates of photolysis among photoreceptor species, the true time of day must
be known because the spectral distribution depends upon solar elevation.
Hence, just as errors in radiation magnitude can lead to poor simulation
of experimental data, so will uncorrected differences in laboratory time
and true solar time.
The j-values used in our simulations are calculated from the UV data sup-
plied by UNC. These calculations are a result of three adjustments made
to the initial UV data: (1) application of an annual, long-term correc-
tion and calibration factor; (2) conversion of flat-plate Eppley UV data
to JQUT values; and (3) application of the outdoor-to-indoor chamber cor-
rection function. A fourth adjustment, correction of laboratory time to
true solar time, is made in conjunction with the last correction.
Differences in annual correction factors result from variations in the UV
instrument sensitivity with time or because instruments have been
replaced. UNC has supplied a list of instruments and calibration factors
used throughout the experimental period. Consideration of this informa-
tion, along with comparison of calculated UV data within a given year and
over all ten years (see figures), provided the annual factors. We have
171
-------
found that correction for the years 1982 through 1986 is unnecessary
(although we have only a limited number of 1985 and 1986 data sets avail-
able), and have used these data as standards for correcting data for
earlier years. In some cases (especially 1980 data), the response of the
Eppley UV instrument decayed significantly over the experimental season.
A linearly increasing correction, beginning at 1.47 on 1 June and ending
at 2.06 on 1 September, has been applied for 1980. The correction factors
deemed appropriate for 1978, 1979, and 1981 are 1.18, 1.30 and 1.20,
respectively.
After the UV data is corrected for annual variations, the flat-plate
Eppley UV data to JQUT(N02) conversion is applied using the function
10~2 - UV2+1.325xlO~4 + UV3+3.791xlO"7.
This function is shown graphically in Figures 5-4 and 5-5 along with the
original data from Stedman et al. (1977) and Zafonte et al. (1976).
J-values for other species are calculated from nonlinear zenith-angle-
dependent relationships as described in Section 3.
The correction for chamber effects is applied to the derived JQUT data to
estimate the in-chamber experimental j-values. Determination of true
solar time (laboratory time offsets) is inherent in these calculations.
It was necessary to offset (add time to) the laboratory clock by 12
minutes for all experiments for 1978 and 1982. In addition, the experi-
ments for late 1984 (all of October) required an offset of about 40
minutes. The laboratory clock appears to be acceptably correct (within 10
minutes) for all other time periods, but this correction is both critical
and elusive. Special care must be taken to account for all clock offsets,
but to not "adust" the UV data on a daily basis, since the laboratory
clock was adjusted very infrequently.
As noted, UNC is presently assessing the data reduction method and
associated uncertainty. We assume an uncertainty in these calculations of
up to 20 percent in extreme instances. However, it must be kept in mind
that the error associated with the data will not linearly translate into
error in the kinetic simulation. For most simulations of UNC experiments,
the period between dawn and late morning shows the highest sensitivity to
photolysis rates, especially to the photolysis rate of formaldehyde to
radical products. Figure 5-3 indicates that there is little difference
between indoor and outdoor measurements during this critical time. The
outdoor chamber appears to have the same photon flux and spectral distri-
bution as the ambient atmosphere until nearly midd'ay. Hence, the critical
errors are those associated with determinations of ambient photolysis
172
-------
0.4 +
_ o.s4
« *
"i
. 0.1+
o.i4
00
O t-IT-TS
O 9-18*73
A 9-I9-7S
f-
-I-
4-
1.0 t.O SO 40 30
UV RADIOMETER ( rnv/cm* )
Source: Zafonte et. al, 1976
FIGURE 5-4. Data Points from Zafonte (solid line is Systems Applications'
solar function).
173
-------
K
I D
E
o
r 4
o
o.
LJ
0.2
0.4
[min*1]
0.6
--30 <
UJ
M
-- 60
-•75
90
Source: Stedman et al., 1977
FIGURE 5-5. Data points from Stedman (solid line is Systems Applications'
solar function).
-------
rates. For clear days, error measurements are not necessary; they are no
worse than the errors associated with determining photolysis rates for
solar flux as a function of zenith angle. As stated earlier, some errors
arise from inaccuracies in the UNC clock time and we have attempted to
adjust the timing of these experiments to alleviate such errors. We dis-
cuss evidence of these uncertainties later in this section.
WALL-RELATED PROCESSES
The issue of trace contaminants and artificial effects in smog chambers
continues to be a major uncertainty in the evaluation of smog chamber
results. These effects become important when experimental reproducibility
is affected, or when chamber-dependant phenomena mask the homogenous gas-
phase photochemistry that an experiment is designed to elucidate. In the
latter case, precise characterization of the chamber-dependant processes
can extend the range of conditions for which the data can be reliably
interpreted. Therefore, we next describe techniques used to characterize
chamber processes in an attempt to clarify the nature and magnitude of
contaminant effects.
Because smog chambers are batch-type reactors, the entire interior surface
1s in contact with the reacting medium through all phases of an experi-
ment. Also, since the surface-to-volume ratio of a smog chamber is much
greater than that of the atmosphere, the chamber surface effects are
greater than those applicable to the atmosphere near the surface of the
earth. Although the current practice is to minimize the reactivity of
chamber surfaces through the use of nonreactive materials and various
cleaning procedures (Pitts et al., 1979; Jeffries et al., 1976; Kelly,
1982), the chamber walls can still be a potential surface for heterogene-
ous processes, and thus a significant sink and source of some trace
species.
Chamber surface effects are manifest in two general ways during smog cham-
ber experiments. First, because smog chambers are usually vented and
loaded with reactants prior to illumination (natural or artificial), a
period occurs when the trace contaminants produced by dark, heterogenous
reactions can accumulate before an experiment begins. During the initial
minutes of illumination, these chemical species can then have an initially
strong and detectable influence on the chemistry. Second, the reactor
walls act as a nonideal surface throughout an experiment. This includes
not only heterogenous adsorption and desorption processes, but also leaks
and dilution resulting in unintended inclusion of trace contaminants
throughout the experiment. In the following discussion, we examine evi-
dence concerning the nature and magnitude of wall-related processes and
175
-------
their effects on both initial and continuing conditions. We then present
our mechanism representations and the values used to model a variety of
environmental chambers.
Chemical Description of Artificial Processes
Three different types of chamber trace chemical contaminants can interact
with the intended experimental photochemistry:
(1) Free radicals introduced either directly or indirectly, which
can initiate or maintain the photooxidation process;
(2) NOX species, which can allow the formation of 63, PAN, etc., and
serve as a radical sink; and
(3) Organic species that can be photooxidized to provide conversion
of OH to R02, and NO to N02, and allow the formation of 03, PAN,
and other oxidants through smog chemistry.
Although radical sources are probably the most noteworthy "dirty chamber"
trace species, NOX and organic oxidation sources can have a significant
effect on low-NOx or low-reactivity experiments. Also, most trace con-
taminants have multiple influences. For instance, HONO, when photolyzed,
serves as both a radical and an NOX source. Also, formaldehyde may be
eliminated by loss to wet walls early in an experiment, but later released
as the chamber is heated up. Proper characterization of the processes
affecting these species provides a more rigorous description of their
behavior.
Various chamber characterization methodologies have been used to help
delineate chamber effects. These include the use of differential tracer
decay techniques to assess radical input rates (Carter et al., 1982), high
NOX experiments with and without high concentrations of CO to assess radi-
cal levels and oxidation background (Kelly, 1982; Leone et al., 1985;
Whitten et al., 1983), acetaldehyde-only simulations to test for trace NOX
by measuring generated PAN (Whitten et al., 1979; Whitten et al., 1983),
and blank simulations using only background air to assess unperturbed
chamber reactivity and reproducibility (Akimoto et al., 1979; Carter et
al., 1986; Whitten et al., 1983). These experiments provide evidence for
the type and magnitude of artificial effects due to specific trace
species. It is our belief that whenever possible, chamber effects should
be described with respect to the actual or likely trace compounds that
appear to be involved in these processes. The alternative is to provide a
simplified phenomenological description (e.g., sources of pure OH, H0»
176
-------
NOX, or OH-to-H02 reactions) for modeling purposes only. Inevitably, con-
ditions are encountered in the chamber or the atmosphere for which such
empirical descriptions are inappropriate. If unrealistic chemical
responses have been "built into" a chemical mechanism because of inappro-
priate chamber artifact descriptions, some description of chamber condi-
tions may be inadvertently included in the mechanism during simulation of
the atmosphere. This could result in good chamber simulations, but overly
or underly reactive simulations of the atmosphere. We attempt to point
out the benefits of a more rigorous chemical description of chamber pro-
cesses in the following discussion. We begin with consideration of the
most commonly used radical source, nitrous acid.
Free Radical and Trace NOX Sources
Nitrous acid (MONO) is a probable source of both radicals and trace NOX in
smog chambers; it photolyzes rapidly (to OH and NO) and, because it is
soluble and can ionize in solution, it is likely to be adsorbed and
desorbed by chamber surfaces. Furthermore, a number of processes (mostly
heterogeneous) have been proposed for its formation (Besemer and Nieboer,
1985; Pitts et al., 1984; Sakamaki et al., 1983; Lee and Schwartz,
1981). Evidence of trace HONO exists in a variety of experimental circum-
stances, which seems to indicate that there are different formation path-
ways. These include:
Non-zero gas-phase concentrations at the beginning of an experiment,
which seem to be linked in part to initial chamber NOX (Leone et al.,
1985; Whitten et al., 1979; Whitten et al., 1983; Carter et al.,
1982; Pitts et al., 1983);
Continuous emission from nitrated chamber walls (Whitten et al.,
1979; Whitten et al., 1983);
Continuous conversion of gas-phase NO? to HONO [probably heterogene-
ous, as studied by Sakamaki et al. (1983) and Pitts et al. (1984);
and
Occasional high-input HONO "incidents" Involving evaporation of
liquid water from (outdoor) chamber walls (Whitten et al., 1983).
It is very difficult to directly measure gas-phase HONO under Illuminated
conditions because photolysis depresses its concentration to very low
levels. Moreover, HONO is also produced under daylight conditions by the
reaction of OH with NO. As this formation reaction coincides with the
photolysis of HONO under illuminated conditions, these two reactions pro-
duce a "do nothing" cycle and a steady-state type of HONO concentration
177
-------
unrelated to the net source of OH and NOX derived from other means of pro-
ducing MONO (i.e., all sources other than the reaction of OH with NO).
Hence, the only direct confirmation of HONO involvement in trace "contami-
nation" would come from measurements of HONO under dark conditions.
Direct measurements of initial HONO concentrations seem to be related to
initial NOX loading (see Table 5-3).
Continuous conversion of N02 to HONO has been studied in several chambers
by Sakamaki et al. (1983) and Pitts et al. (1984). Of the two heterogene-
ous processes for which Sakamaki et al. derived rates, the more important
process is
N02 + H20 — (wall) — > HONO,
ft 1 1
involving gas-phase water at a rate of 1.01 x 10 ppm"1 min . Dry con
version of N02 to HONO,
N0 — (wall)-> HONO,
at 3.9 x 10"5 min'1 was also estimated. At 20,000 ppm, H20 (ca. 50 per-
cent relative humidity at 300 K), these two processes give an estimated
first-order conversion of N02 to HONO of 2.4 x 10~4 min'1. This is twice
the conversion rate observed by Pitts et al. for the UCR-EC and four times
that of the UCR-ITC. It is worth noting that the N02-to-HONO yield in
these experiments was observed to be from 0.21 to 0.44 of the first order
decay of N02. Thus, the magnitude of this process might be crudely esti-
mated from N02 dark decay measurements, even in the absence of direct HONO
measurements.
Although, as we have previously noted, measurement of continuous HONO
emissions in illuminated systems is very difficult due to HONO photolysis,
an upper limit may be estimated by measuring total NOX emissions. This
can be done by irradiating pure acetaldehyde and measuring the formation
of PAN (Whitten et al., 1979). In the UCR-EC, for example, such an
experiment has yielded 40 ppb of PAN (and 190 ppb 03) after six hours, or
a total N0x-input rate of 0.11 ppb/min. This value for an irradiated
experiment is much less than the 0.3 ppb/hr N0x-input rate reported by
Pitts et al. (1981) in the absence of light. The similarity between total
NOX emissions and the magnitude of the chamber radical source (the N02-
independent term discussed by Carter et al.) for the chambers considered
here suggests that NOX emissions may be primarily in the form of HONO.
NOX offgassing incidents have been shown to occur in the UNC chamber under
178
-------
TABLE 5-3. Comparison of total NOX with measured MONO
in the UCR-EC (Source: Pitts et al., 1983).
Exp. No. (EC)
567
568
569-A
B
570-A
B
626
630-A
B
631-A
B
632-A
B
NO
1.450
1.032
1.058
0.668
1.464
1.724
0.385
2.322
2.163
1.936
1.950
2.200
2.189
N02
0.467
0.772
0.100
0.450
0.370
0.840
0.103
0.824
0.838
1.053
0.898
0.794
0.806
NOX
1.917
1.804
1.158 >
1.118
1.834
2.564
0.488
3.146
3.001
2.989
2.848
2.994
2.995
HONO
(ppb)
21.
12.
25.
15.
29.
17.
5.1
21.7
28.6
16.6
11.8
19.
14.
HONO
(percent
of NOX)
1.1
0.7
> 2.2
1.4
1.4
0.7
1.0
0.7
1.0
0.6
0.4
0.6
0.7
179
-------
special conditions involving the condensation of liquid water on the cham-
ber walls, and additional evidence implies that the NOX emitted is HONO.
Both evidence of NOX offgassing in the UNC outdoor chamber and an analysis
by Kamens et al. (1980) that is based on ionic and phase equilibria for
NOX species provide examples of HONO acting as a background radical and
NOX source. Nitrate, nitrite, and dissolved NOX species establish equi-
librium concentrations in the liquid phase, and NOX offgassing may occur
as a result of aqueous chemical conversions, Ionic equilibrium, and phase
equilibrium for various NOX species. Experimentally, there is also some
evidence of heterogeneous conversion of nitric acid to gas-phase NOX in
Teflon chambers (Jeffries, 1979).
For HONO, the aqueous equilibrium concentrations (and offgassing rates)
appear to be highly dependent on pH. At low acidity, more HONO ionization
is possible, resulting in a larger effective solubility for gas-phase
HONO. As acidity increases, however, less ionization of HONO occurs and
proportionately more aqueous HONO is available. This should result in the
elution of HONO from solution into the gas phase to reestablish phase
equilibrium. Different hydrogen ion concentrations will also affect the
rates of chemical change in the aqueous phase and shift the overall NOX
balance to different species. Kamens et al. (1981) conjecture that N02
becomes the major form of dissolved NOX at intermediate pH (versus ionic
nitrate and nitrite), leading to episodes of chamber N02 emissions because
N02 is at least three orders of magnitude less soluble than HONO
(Schwartz, 1984). In addition, as water evaporates off the chamber walls,
the aqueous nitrite ion may reform aqueous HONO, which in turn could
induce the liberation of HONO to the gas phase.
Figure 5-6 shows experimental concentration data and simulation results
for a liquid-water-associated NOX emission episode in the UNC chamber.
Such an experiment measures chamber NOX emissions by converting the NOX to
PAN, which is easily measured and is not rapidly converted to nitric
acid. In this experiment, both sides of the two-sided outdoor chamber
were loaded with 1 ppmC of acetaldehyde. As the condensed water on the
chamber walls evaporated, 70 and 100 ppb of NOX were emitted from the two
sides, respectively. The rapid buildup of PAN and the lagged formation of
03 are consistent with a powerful radical source. The ozone formation
delay is due to a suppression of N02 by radical scavenging processes. In
a modeling study (Whitten et al., 1983), attempted to characterize the NOX
emissions as NO or N02 caused a rapid increase in 0^ that parallels PAN
formation, rather than the experimentally observed 2- to 3-hour delay.
However, the use of HONO as the radical source allowed simulation of the
delay in ozone formation (Figure 5-6).
180
-------
B 100 202 302
TIME
U) Red Side: NOV
303
TIHE (K1NLHES)
(b) Blue Side: NOX
FIGURE 5-6. Simulation results for UNCR 72179
(N02 data includes PAN).
-------
A large radical and NOX source such as that caused by MONO emissions in
the liquid water incident can render an experiment useless for photochemi-
cal analysis and model validation. Additionally, liquid water has been
shown to absorb formaldehyde in the gas phase and to re-emit 1t upon
evaporation and warming (Whitten et al., 1983). Such processes further
complicate analysis of experimental data. To address this problem,
researchers at UNC have developed procedures for predrying the smog cham-
ber (Kamens et al., 1981). Furthermore, since this problem is common in
outdoor bag-type experiments, either such drying procedures must be used
or data from experiments in which condensation was observed must be care-
fully considered.
Organic Species
Observed background chamber reactivities are usually significantly greater
than the effect of the reactive hydrocarbons measured in the background
air would indicate (see below and Whitten et al., 1983). On the other
hand, oxygenated hydrocarbons (especially lower molecular weight aldehydes
and dicarbonyls) are often not included in the reactive hydrocarbon mea-
surements because they are difficult to measure reliably at low concentra-
tions and they are significantly reactive, even at levels of 10-50 ppbC.
Furthermore, since these compounds are soluble and polar, they can be
adsorbed on chamber walls or dissolved in any liquid water in the cham-
ber. This causes situations where chamber walls could conceivably collect
and later release carbonyl species. As we discuss below, collection could
occur prior to illumination during chamber loading or during an experi-
ment, if significant production of one of these species (such as formal-
dehyde from ethylene oxidation) were to occur rapidly.
The most apparent evidence of aldehyde-type contamination is found in the
frequent formation of PAN (- 10 ppb) in experiments that have no known PAN
precursors. Acetaldehyde emissions of 0.05-0.1 ppb/min are generally suf-
ficient to explain measured PAN. It should be noted however that other
species may serve as PAN precursors, especially in outdoor chambers where
background levels of biogenic hydrocarbons may contribute (Killus and
Whitten, 1984).
Although formaldehyde contamination is probably greater than that of other
aldehydes because of its higher formation rates and large solubility con-
stant (Ledbury and Blair, 1925), HCHO can be more difficult to measure
than higher aldehydes and does not lead to an easily measured product such
as PAN. Background concentrations of HCHO have been routinely observed in
some chambers (Carter et al., 1982). Unfortunately, the observed concen-
trations are usually too low (0-50 ppb) to allow precise and accurate mea-
surement by the wet chemical methods that are often the only ones avail-
able. Even at concentrations as high as - 0.5 ppm, when measured amounts
182
-------
of HCHO are Introduced into the chambers, gas-phase measurements are often
scattered and frequently show less than the amounts actually added, per-
haps because of adsorption on chamber and instrument surfaces. Even more
elusive is the heterogenous nature of dicarbonyl species, which can be
formed in high yields in some aromatic oxidation experiments. The more
common of these species, glyoxal and methylglyoxal, have solubility con-
stants comparable to that of formandehyde (Betterton and Hoffmann, 1987).
A trace background of HCHO, maintained by continuous desorption from (and
adsorption to) the chamber walls, can serve as both a radical source (upon
photolysis) and a photooxidant precursor (upon reaction with OH to form
H02). For example, if the gas-phase concentration of OH is decreased by
an increase in N02 (an OH sink), proportionately more of the HCHO will
photolyze since the competing process is suppressed. Since the increase
in N02 correlates with an increase in the HCHO radical source, one might
conclude that N02 is related to a chemical step producing the radical
source, but what is actually occurring is that N02 is anticorrelated with
a radical source suppression, which is not the same as being part of the
radical source. Under more reactive conditions (i.e., typical chamber
experiments), the trace HCHO background would tend to be oxidized rather
than photolyzed and would have little impact on the experiment.
Dilution and Sink Processes
Small quantities of typical ambient hydrocarbons (< C4 alkanes, ethene,
etc.) are often measureable in the initial loading or dilution air of smog
chambers (dilution air also includes air entering through small holes in
the Teflon walls of a well used chamber). This is especialy true for
large outdoor chambers such as the UNC dual chambers or the UCR-OTC. In-
door chambers often use filtered air and have less leaks, thereby negating
the consideration of dilution air hydrocarbons. Typically, the total
reactive (nonmethane) hydrocarbon loading should be much less than 50 ppb
(Killus, 1982). Moreover, these species are usually of low or intermedi-
ate reactivity and should have minimal effect on most chamber experi-
ments. We present our estimates of these input fluxes later in this sec-
tion.
As noted, the chamber walls might serve as a net sink for aldehydes and
dicarbonyls during initial phases of smog chamber experiments. Because
wall water evaporates, however, a strong source of these species could
exist for a few hours in the middle of an experiment. In the experimental
results that follow, we repeatedly see no formaldehyde measured in the
first few hours of an experiment, followed by an increase that is
extremely rapid and cannot be simulated with homogeneous reactions. In
some extreme ethylene experiments, we find no formaldehyde formed during
183
-------
the initial oxidation of ethylene (formaldehyde is the main oxidation pro-
duct of ethylene), followed later by a period when formaldehyde concentra-
tion increases so rapidly that complete conversion of the reacting ethy-
lene could not account for the magnitude of the change in formaldehyde.
We will show many instances of such an effect in Section 6. These effects
also appear in aromatic experiments where dicarbonyl wall influences could
occur. Because the dinitrophenyl-hydrazone/liquid-chromatographic-analy-
sis technique for dicarbonyls is far more cumbersome than measurement with
the automated colorimetric (CEA) formaldehyde monitor (at UNC), there is
as yet little evidence for the dicarbonyl wall processes.
For other species the sink effects of chamber walls appear to be of little
concern except in the case of a few species that accumulate at the end of
a 1-day experiment (e.g., HNC^, ^2, and higher molecular weight
peroxides). These species do not normally influence the aspects of the
photochemical process under investigation in such experiments. On the
other hand, we will present evidence in the following discussion suggest-
ing that the wall losses of these compounds in multi-day experiments could
be important and cannot be ignored.
Trace Contaminants in Specific Chambers Simulated in this Study
Indoor Chambers: The UCR EC and ITC
We have suggested that the contaminant species of greatest importance in
smog chamber simulation are nitrous acid (MONO) and oxygenated hydrocar-
bons, especially formaldehyde (HCHO). The analyses reported next further
support the hypotheses that HONO is the main radical-and N0x-source
species, while HCHO is responsible for much background oxidation reac-
tivity (conversion of OH to peroxy radicals) and has a lesser effect on
radical inputs.
Nitrous acid emissions from smog chamber surfaces differ in one important
respect from all other forms of radical inputs to these systems. Because
the photolysis of HONO introduces both NO and a hydroxyl radical, no net
change of NOX occurs when the radical terminates as nitric acid or another
nitrate species (e.g., PAN). Thus, with the exception of HONO inputs, NOX
decay may provide an easily obtained estimate of intergrated radical in-
puts, since most radical termination reactions do result in the formation
of nitric acid or organic nitrates. This simplified picture of radical
input and termination is valid until oxidation has proceeded to the point
where nitric oxide (NO) levels are low and radical termination (through
radical-radical self-reactions) becomes important. This is approximately
the time when NO and ^^ levels become significant relative to total
184
-------
NOX, and loss of these species becomes an important NOX sink. Usually
either ozone or PAN, or both, will attain maximum concentrations at this
time.
It is possible to obtain an estimate of hydroxyl radical concentrations in
a smog chamber experiment by observing the decay behavior of both hydro-
carbons and NOX. The decay rate of many hydrocarbons is primarily due to
reaction with hydroxyl radical. In some cases, the hydroxy decay rates
are slow enough to require a correction for dilution; in others, (such as
olefins) decay due to 03 and 0(3P) must be taken into account. In order
to use NOX decay to calculate OH levels, a correction for both dilution
and for PAN formation must be used. The latter correction must be applied
because OH is calculated from the direct loss due to OH:
OH + N02 -> HN03, (26)
while PAN formation depends upon the production of peroxyacetyl radicals
from many sources.
An example of these two methods of OH calculation may be seen in Figure
5-7, which compares estimated OH based on NOX decay (corrected for
dilution and PAN formation) to OH estimated from the decay rates of
toluene, xylene, and 2,3,dimethylbutane (corrected for dilution and
averaged), for an experiment in the UCR Evacuable Chamber. As may be
seen, for most of the experiment the two methods of estimating OH are in
good agreement.
In very low reactivity experiments, however, NOX decay is not consistent
with OH calculated from relative hydrocarbon decay. In Figures 5-8 and
5-9 we show results of OH calculated from NOX decay compared to OH
obtained from propene decay, as reported originally by Carter et al.
(1982). Over a wide range of chamber operating conditions and NOX levels,
NOX decays at a much lower rate than one would expect from estimates of
OH. There are two possible causes of this phenomenon: (1) the estimates
of OH concentration are in error, or (2) some source of NOX is balancing
the radical inputs. The former possibility may be largely discounted,
since other techniques of OH radical level estimation corroborate the
hydrocarbon decay estimates. Because CO addition and subsequent NO-to-N02
conversion in UCR-ITC experiments agree with the hydrocarbon decay esti-
mates of OH (Table 5-4) a counterbalancing NOX source is very probable.
Given the wide variation in radical levels and input rates estimated from
these experiments, it would be an extraordinary coincidence to find a
separate NOX source that so nearly (within about 10 percent) balanced the
radical source. It is much more likely that the radical source and the
NOX source are the same species. This suggests photolytic nitrites.
185
-------
~ I
?u
S o
n •-
EC238
ESTIMATED OH
600
FIGURE 5-7. Estimated hydroxyl radical concentrations for UCR EC-238
using two different radical estimation techniques.
-------
00
b
a
III
EVACUABLE CHAMBER
REACTIVITY CHARACTERIZATION RUNS
6 -
4 -
3 -
2 -
1 -
-1
EC-448
EC-449D
20
40 60
MINUTES
60
From OH
decay
From I*
decay
100
120
FIGURE 5-8. Estimated radical concentrations for a set of UCR EC
experiments.
-------
3
a.
z >-
INDOOR TEFLON CHAMBER
rrc-«29: rrc-«28
-1
1 «0 200
4 FROM NOX DECAY
240
FIGURE 5-9. Estimated radical concentrations for two UCR ITC experiments.
-------
generally, and MONO in particular as the primary radical source species.
It is also worth noting that while the radical source seems primarily
associated with an NOX source, the converse is also true; most of the NOX
that is emitted from the chamber walls seems to be associated with radi-
cals. Although NOX sometimes increases during a low reactivity run, the
increase 1s generally small and may be due to a buildup of compounds that
are converted to NOX during NOX detection (e.g., nitric acid).
The finding that MONO is the principle NOX and radical source in the UCR-
EC, and possibly other chambers, significantly reduces the problem of
chamber radical characterization, since NOX inputs may be easily monitored
as PAN buildup in acetaldehyde-only experiments. We have previously shown
and described (Figure 3-7) the simulation results of EC-253, an acetal-
dehyde-only experiment in the EC using 0.14 ppb/min of NOX input, 80 per-
cent as MONO (and an initial condition of 5 ppb HONO). These NOX inputs
are consistent with the estimates of radical inputs by Carter et al.
(1982) and lead to a good simulation of the experimental system (see Sec-
tion 3).
Figure 5-10 gives simulation results from ITC-627, an acetaldehyde only
experiment. The HONO input required to fit PAN production is 0.035
ppb/min (as in EC-253, a 5 ppb initial condition of HONO fit the initial
PAN buildup). The radical input rates indicated by characterization
experiments using both the propene tracer technique and the CO NO-to-N02
conversion technique are about 0.06 to 0.07 ppb/min. At such low NOX and
radical input levels it is difficult to ascertain the source of the dis-
crepancy between these two values. We note, however, that in ITC-628 the
measured NOx did increase during the experiment, perhaps indicating the
input of some interfering compound as noted above. The amount of back-
ground formaldehyde necessary to supply the "missing" radical inputs of
about 0.025 ppb/min is equal to only 25 ppb. This concentration may be
maintained with wall emissions of 0.05 ppb/min. However, in our simula-
tions, we use an offgassing rate of 0.1 ppb/min in order to be consistent
with our findings for outdoor chambers.
Battelle Chamber
Only three characterization experiments were available for the Battelle
chamber. A moderately high NOX source (0.085 ppb/min) is indicated from
an acetaldehyde/PAN formation experiment. A CO/NOX experiment indicates a
radical source similar in magnitude to the NOX source (- 0.1 ppb/min). We
are again led to speculate that HONO emissions, with some contribution
from a small HCHO background (- 30 ppb assumed), are responsible for the
189
-------
TABLE 5-4. Estimated radical levels using two different methodologies for
the UCR-ITC.
ITC-625
Before added CO
0-60 mins
60 - 120 mins
After added CO
120 - 180 mins
180 - 240 wins
ITC-628
Before added CO
70 - 130 mins
130 - 190 wins
After added CO
190 - 250 wins
250 - 295 wins
ITC-634
Before added CO
0-60 mins
60 - 120 mins
After added CO
120 - 180 mins
180 • 240 wins
ITC-636
Before added CO
0-60 mins
60 - 120 mins
After added CO
120 - 180 mins
160 - 240 iMns
OH Concentrations
(ppm x 10" )
Tracer Method* CO * H02t
4.45
6.17
5.55 4.47
2.33 3.03
3.82
3.43
3.2 3.8
2.3 3.55
7.65
3.28
3.7 £3.0
1.4 '2.6
6.5
5.4
3.1 4.3
4.2 3.2
Radical Source Strength
(ppm/mln x 10 )
Tracer Method* CO * H02t
4.6
6.6
6.8 6.6
4.6 6.03
5.8
4.6
5.8 7.0
5.5 7.5
16.0
7.0
8.4 6.9
3.7 6.8
e.
6.7
4.9 6.8
8.4 6.7
Measured
HCHD
(pp"0
—
--
0.017
0.069
0.031
0.002
0.006
0.044
0.038
0.010
* Corrected for 03 and 0(3P).
* I «0 and H02 conversions/Bin + 50 ppm CO + 320 ppm'2 Bin'1.
§ Corrected for NO + NO «• 02 » N02 and N02 » 0 + 2NO.
190
-------
OJD7
ITC-627
ACETALDEHTDE ONLY
006 -
OJ05 -
OB* -
0433 -
0.02 -
OBI -
0 -
03
100
200
MINUTES
300
4OO
0.013
O.D12 -
O.011 -
OJD\ -
0.006 -
O 008 -
> O.007 -
• O.006 -
> aoos -
0.004 -
0.003 -
0 002 -
O.OD1 -
0
100
too
MINUTES
300
4OO
FIGURE 5-10. Ozone and PAN traces for ITC-627, an acetaldehyde-only
experiment.
191
-------
observed radical Initiation phenomena. However, the Battelle chamber-
characterization data show a significant anomaly; an NOX clean-air simula-
tion produced an NO oxidation rate that indicates a reactive background
(compounds converting OH to RO?) equivalent to either 12 ppm CO, 0.09 ppm
ethene, or 0.03 ppm propene. (These compounds have not been detected at
such high levels in the Battelle chamber.) Additional data are required
to determine whether this was a temporary effect or a typical characteris-
tic of the chamber; if the latter is the case, the contamination should be
identified.
Effects of Chamber Radicals on Experiments
Table 5-5 lists the total calculated chamber radical flux values (using
the equation of Carter et al., (1982) for a number of experiments in the
UCR-EC that have been published in various studies. The chamber radical
source is compared to the NOX loss for each experiment, and an estimate is
presented for the fraction of total radicals that results from chamber
sources. As Table 5-5 shows, in the majority of experiments less than 20
percent of total radicals are derived from the chamber. Of the experi-
ments involving high chamber radical fractions, practically all are of low
added reactivity (butane-NOx and acetaldehyde-NOx). The three urban mix
experiments with high chamber radical fractions (EC-232, 233, and 241) all
involve mixture A, which is a low-olefin, low-aromatic, high-paraffin mix.
Table 5-6 shows the results of similar calculations for the UCR-ITC and
Battelle chambers. Surprisingly, despite a greatly reduced chamber
source, the UCR-ITC shows a greater chamber radical fraction than does the
UCR-EC, probably because the light source in the ITC and chosen hydrocar-
bon-NOx mixtures resulted in low reactivity experiments. Since these
experiments were designed as multiday experiments, the long reaction times
allowed the small chamber flux to become significant.
Unknown Multi-day Phenomena in the ITC
In the UCR-ITC multi-day experiments, chamber reactivity on the second day
appears to be perturbed by the overnight dark period in a way that cannot
be explained by any of the previously advanced hypotheses of chamber
effects. It is possible that the long dark period between illuminations
on successive days of the ITC multi-day experiments provided a means for
the oxygenated product species forming at the end of an experiment (e.g.,
HNOg and H202) to interact with the chamber surface for an extended
period. The data from ITC-631 demonstrate the differences in characteris-
tics from one day to another. The first day's reaction ended at the time
of N02 peak, with ozone and NO almost exactly equal. When the chamber
192
-------
co
TABLE 5-5. Chamber background radical sources as a fraction of total
radical Inputs for the UCR-EC. (Average chamber radical source from
Carter et al.f 1983.)
Spec let
Propetie
troftnt
Propen*
Toluene
To! time
Toluene
Toluene
Cthene
Elhene
n-butine
Synthetic urban pliture
Synthetic urbin nliture
Synthetic urban utiture
Synthetic urban mliture
Synthetic urhan nliture
Synthetic urban Mixture
Synthetic urhan mliture
Synthetic urban nliture
Synthetic urban nliture
Synthetic urban ntiture
Acet aldehyde
Butane
Butane
Tolulene » Butane
M-iylene
«-»ylene
Butane
Butane « Acetaldehyde (?)
Butane * Formaldehyde
Acetaldehyde
EC No.
m
716
11«
269
271
273
327/340
142
143
178
2318
232A
233A
2378
238*
241N
242C
743
245C
246C*
254
305
306
331
344
34S
162
163
168
164
*l
0.3
0.43
0.48
0.35
0.36
0.35
0.4
0.34
0.33
0.33
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.43
0.44
0.4
0.4
0.39
0.34
0.34
0.33
0.33
Time
°3
Peak
(ppi»)
31S
465
330'
>360
0.90
75
>360
>360
210
>49S
225
>360
240
240
435
>360
105
135
180
-570
330»
345
390
120
180
75
>360
>360
690
>360
Nailnifi
*n?
(ppn)
0.341
0.367
0.666
0.26
0.14
0.064
0.259
0.3
0.375
0.066
0.365
0.344
0.076
0.373
0.669
0.346
0.401
0.394
0.755
0.374
0.068
0.076
0.146
0.334
0.454
0.192
0.279
0.299
0.76
Average
NO.
(PP")
0.18
0.16
0.2
0.18
0.08
0.038
0.16
0.13
0.22
0.025
O.IK
0.18
0.04
0.22
0.25
0.18
0.22
0.25
0.46
0.22
0.04
0.051
0.09
0.2
0.2
0.1
0.237
0.2
0.153
0.22
Constant Average rhanher fhanher
thftnHer Radical Source Rascal
Radical Including N0? rim
Source dependency (ppn)
0.12
0.17
0.19
0.14
0.14
0.14
0.16
0.133
0.179
0.13
0.12
0.12
0.12
0.12
0.12
0.12
0.12
0.12
0.12
0.12
0.12
0.17
0.17
0.16
0.16
0.16
0.13
0.13
0.13
0.13
0.19
0.76
0.3?
0.7?
0.18
0.15
0.74
0.7
0.738
0.14
0.17
0.7
0.18
0.18
0.75
0.7
0.?
0.7?
0.31
0.17
n.133
0.70
0.73
0.76
0.71
0.18
0.74
0.73
0.7
0.73
0.06
0.17
0.11
0.08
n.nis
0.12
O.OB8
0.07
n.05
0.069
0.038
0.072
0.0312
0.043
0.174
0.072
n.rwi
0.03
0.056
0.093
n.n«4
0.068
o.nsa
0.032
0.038
0.012
O.(187
0.083
0.137
0.085
«*n,
(ppn)
0.5
0.5
0.78
0.11
0.186
0.083
0.395
0.433
0.43
0.093
0.4
0.282
0.07
0.4
0.88
0.311
0.4
0.41
0.81
0.42
n.nss
O.(ln6
n.]4
0.356
0.671
0.778
0.183
0.360
0.493
0.7
llncorr»ct»d
rh«i*»r
««1lca1
Tract Inn
0.1?
0.74
0.13
0.70
0.09
0.14
0.7?
0.16
0.17
0.75
0.10
0.26
0.44
0.11
0.14
0.023
0.05
0.07
0.07
0.22
0.51
0.103
0.63
0.09
0.06
0.04
0.48
0.73
0.78
0.41
rnrr»cl»d
for NO
rim
0.11
0.21
0.1?
0.16
0.08
0.12
0.19
0.14
0.11
0.44
0.09
0.7?
0.3?
0.10
0.12
0.20
--
--
—
0.19
0.35
0.55
0.43
"
--
0.39
0.70
0.74
0.34
»AN peak.
-------
TABLE 5-6. Chamber background radical sources as a fraction of total
radical Inputs for smog chambers other than the UCR-EC.
Chamber
ITC
Batelle
(kl •
Species
0.3 mln'1)
Urban mixture
Urban mixture
0.19 m1n"*)
Toluene
Toluene
Ethyl benzene
626
637
630
631
635
AVOC5
AVOC9
AVOC10
Time
03
Peak
510
525
>720
>720
>720
200
175
405
Average
Radical
Source
0.06
0.06
0.06
0.06
0.06
0.095
0.095
0.095
Radical
Flux
0.03
0.032
0.043
0.043
0.043
0.018
0.017
0.038
ANO
0.164
0.125
0.115
0.07
0.15
0.43
0.08
0.38
Thanhpr
Radical
Fraction
0.18
0.26
0.37
0.61
0.28
0.04
0.2
0.1
Corrected
for NOX
Flux
0.15
0.2
0.29
0.4
0.23
-------
lights were turned off, 03 and NO titrated to nearly zero concentration
and, for the overnight period, only hydrocarbons and N02 remained. When
the lights were switched on again the next morning, NO and 0^ both
reappeared at precisely the same concentrations that existed the previous
day when the lights had been switched off. There had been a small
decrease in N02. As can be seen from Figure 5-11, which shows the data
from the two days concatenated and the overnight data deleted, the overall
reactivity of the system had noticeably changed from one day to the next.
On the second day the NO? decay rate had increased, as had the 03 produc-
tion rate and the rate of NO oxidation. If N02 had been converted to gas-
phase MONO overnight, then the "reactivity boost" should have come as a
sharp pulse on the morning of the second day, rather than a continual
radical input, as seems to be the case. If an ammount of MONO had been
formed on the chamber walls and slowly bled into the system, then, for the
reasons noted in the previous section, the increase in reactivity should
not have affected the rate of N02 loss. The possibility that the HONO
inputs somehow "sparked" additional radical inputs from the hydrocarbon
species is not persuasive, nor do we think it likely that N02 reacted with
some hydrocarbon species to yield a radical source with the characteris-
tics seen in Figure 5-11. Both the CBM-IV and the CALL mechanism (Carter
et al.t 1986) seem to exhibit reactivity discrepancies for multi-day
results in the ITC. Unfortunately, consideration of the results for ITC-
631 makes it currently impossible to judge whether these discrepancies are
due to mechanistic inadequacies or an as yet uncharacterized chamber
phenomenon.
Outdoor chambers: The UCR OTC and UNC dual chambers.
Chamber emissions of NOX in the UNC chamber are highly variable and
probably related to the condensation of liquid water on the chamber walls
during overnight venting. In four acetaldehyde-only experiments PAN
formation was approximately 10 ppb in one single sided experiment; 20, and
40 ppb on both sides in two experiments; and 60 and 100 ppb in the last
experiment. In the last experiment (July 21, 1979, described earlier and
shown 1n Figure 5-6) we found that HONO was the only form of NOX emission
that would give accurate results for both ozone and PAN, thus corrobora-
ting our finding noted above that HONO is the preponderant form of NOX
emission. Two acetaldehyde-only experiments in the UCR Outdoor Teflon
Chamber (UCR-OTC) yielded results of approximately 20 ppb in a six-hour
experiment. Both of these experiments showed low dew point measurements
(I.e..less than 15 degrees C).
We have analyzed five additional CO/NOX experiments conducted at UNC to
assess radical inputs: UNCR.62782, UNCB.80282, UNCB.82082, UNCB.82382,
195
-------
cr>
UCR ITC MULTIDAY EXPERIMENT
ITC-631
^\
0.
a
V^
z
TRATIO
;ONCCN
V
\J >£.^
O.22 -
0.2 -
0.18 -
o.ie -J
0.14 -
0.12 -
O.1 -
o.oe -
0.06 -
0.04 -
0.02 -
ft -1
1 1
D I
D «. .• _i. a j ^A|
^TT** ' NH9
D°a ^^ ^"^
t*4 ^^4,
TI ^^ 4- ^>
•f*" ^ %
/ \ * *.*•* ^*4*
J. % «
^ rfX^* 03
1{j
^qJJ ^
Qj_l o o^o ^bn.
*L 6f> G 1 ^^^ll
A wO ^^ o I ***j LO frn fTi m^ r*» •.•
o o^ ^^ 1 ^^ ' * ^ B * m u
^ft ^^ *^*^
Day 1
Day 2
FIGURE 5-11. Experimental data from the two-day, ITC-631 experiment with
the dark period removed and the two light periods merged.
-------
and UNCB.90582. Radical inputs were assessed with the above-noted NOX
loss technique, in order to attempt to estimate the significance of non-
HONO radical sources, and by assessing NO oxidation by H02 formed from the
reaction of OH and CO. This latter technique uses the equation
OH = { A (NO - 03) + (N02)2 x 0.007 x Kl } / (CO x 320).
The second term in the equation accounts for the reduction of N02 to NO by
the reactions
N02 > NO + 0
0 + N02 > NO + 02
These reactions become significant at about 0.15 ppm N02.
Because the N0x-loss technique involves small changes to large measure-
ments, the scatter for calculated OH is large (see Figure 5-12). The
average result for each experiment is quite consistent, however, and indi-
cates that a non-HONO radical source exists in each of these experiments
that is approximately equal to the HONO source (Table 5-7). The back-
ground of formaldehyde necessary to explain the non-HONO radical source is
modest (about 20 ppb) and entirely consistent with UNC measurements of
chamber background. Maintenance of this concentration in these experi-
ments would require an offgassing rate of about 0.1 ppb/min.
Although Carter et al. (1982) indicate that the radical sources in the EC
and ITC chambers are proportional to light intensity (j^Q2 or k^), we find
that the correlation between light and radical inputs is fairly weak (see
Figure 5-13), explaining only 15-30 percent of the variance of the derived
radical inputs (the higher value for r is obtained by eliminating the
September 9 experiment, which was very dry). Essentially none of the day-
to-day variance may be explained by changes in light, and much of the
observed correlation may be due to the formaldehyde radical source, which
may be expected to be highly correlated with light. (Note that formal-
dehyde emissions need not be correlated with light for the radical source
to be so correlated, since emissions would be expected to cause only a
slow change in formaldehyde background levels). For HONO emissions,
photolysis is so fast that light must be diminished to nearly zero before
radical inputs are delayed. Only if HONO emissions are directly caused by
light would the proportional relationship hold. In the case of outdoor
chambers, which have strongly varying temperature and humidity, it is more
likely that these phenomena are responsible for variations in HONO emis-
sions rates. It is also worth noting that PAN formation rates in acetal-
dehyde experiments drop much more rapidly than light intensity in the
afternoon; in fact PAN formation in some of the above-noted experiments
197
-------
UNCR JN 27, 1982
RADICAL SOURCE
tOO 40
UNCB AU 2, 1982
RADICAL StXftCt
too
400
•00
UNCB AU 20. 1982
macAL COIMCCS
FIGURE 5-12. Estimated radical source strengths for five UNC smog chamber
experiments using two different estimation techniques.
198
-------
UNCB AU 23, 1982
IUOICN. SOUtCE
•IMSTES
•» CO/HO OMMTON
UNCB SEPT 9, 1982
too
O HOI LOSS
•IWJTES
4 CO/MO OXCMHON
FIGURE 5-12. (concluded)
1QQ
-------
TABLE 5-7. Radical inputs for CO/NO., experiments in the UNC smog chamber.
Radical Sources (10~5 ppm/min)
Experiment
JN2782R
AU0282B
AU2082B
AU2382B
SE0582B
NOX Decay
Method
3.9
5.1
8.4
7.6
0.5
CO/NO Oxidation
Method
14.1
10.7
10.1
11.3
4.8
-------
O.OOO21
0.0002 -
0.0001* -
0.00018 -
0.00017 -
o.oooit**-
0.00018 -
0.00014 -
O.OOO13 -
0.00012 -
O.OO011 -
O.O001 -
O.OOOO9 -
0 .OOOOO -
0.00007 -
04XXXM -
0.00008 -
OjOOOO4 —
0.00003 -
0.00002 -
0X0001 -
0
0,
1
RADICAL INPUT VS K1
UNC CHAMBER
0.3
0.8
0.7
K1 {ft* MM)
FIGURE 5-13. Comparison of estimated radical Input rates and
for chamber characterization experiments 1n the UNC chamber.
-------
ceases before maximum light intensity, which is indicative that chamber
derived HONO is a finite quantity.
In using the light proportionality assumption of Carter et al. for outdoor
chamber experiments, we suggest that the following rates should be used:
Median chamber radical intensity for these five experiments was 0.22 x kj
ppb/min (see Figure 5-14); however, since about half of this is provided
by a 20 ppb formaldehyde initial condition (and a 0.1 ppb/min offgassing
rate—roughly 0.2 x kj_ ppb/min), the HONO emission rate would be 0.11 x k^
ppb/min, yielding a total emission of 26 ppb of HONO in a typical 10 hour
experiment in the UNC chamber.
Carter et al. (1986) also use an oxidation reactivity equation of:
OH H02
at a rate of 500 min'1. This is equivalent to the effect caused by an
additional 33 ppb formaldehyde concentration. The estimate of Carter et
al. (1986) for NOX offgasing (0.2 x kj^ ppb/min) is likewise higher than
our estimates. The overall estimate by Carter et al. of the chamber radi-
cal source in the UNC and OTC chambers is 0.3 x kj_ ppb/min, although their
data for the OTC gives a median value that more closely corroborates our
lower figure. Although there are some conditions in the UNC and OTC cham-
bers that will produce chamber contaminants as high or even higher than
those used by Carter et al. (1986), typical or average conditions are
notably lower.
Kinetic Summary of Heterogenous Chamber Processes
A number of processes have been identified that seem to account for the
heterogenous effects occurring in smog chambers. Two such processes
mearly account for the mass gained and lost in chamber dilution and sink
reactions. Because the decay rates of the many species in smog chamber
experiments are unknown, we generally assume they are either low or
unimportant. In the case of ozone, however, decay rates are easily mea-
sured and we include this loss reaction in the smog chamber mechanisms:
03
with kloss = 1.8 x 10~4 min'1 for UNC and 1.8 x 10~3 min'1 in the UCR-EC.
Average dilution rates are provided with the data for each experiment in
the UCR-EC. Typical dilution rates are between 3.0 x 10~4 and 4.0 x 10~4,
and it is assumed that clean replacement air concurrently enters the cham-
ber. At UNC, however, dilution is not reported but the concentrations of
202
-------
CHAMBER RADICAL INPUTS
UNC OUTDOOR CHAMBER
t
2
o
o
ro
o
CO
30-
28-
28-
24 -
22 -
20 -
18-
18-
14 -
12 -
1O -
8 -
8-
4 -
2 -
0 -
1
O
D
a
Q
D
a
a a
a a
CD O
CO ODD
t°tj 1 i i 1 1'
IIIIIIIIIII
IIIIIIIIIII
>
a
a
a
u
n
a
u
n
u
a
a
a
o
n
1 1
11
o
D
D
O
aa
OD
DD
m
OD
DD
ULJLJL
I • • •
I I I I
I I II
I I 1 I
a
D
o
a
a
o
a
a
a
o
DD
CD
CD
a an aa
D DD ODD
Lp i • • J d • TJ
II ( p 1 1 1 I I]
,f m m m m m m jj Q
.1 1 1 1 1 1 1 u OD
i • • • • • • TTI 1 1
• iiniii iiini
iiiiiiiiiiiidkn niiini
i II I mi ii u mil Q OD D DO O D
•iiiiiiiiiillllll lllllllllll n n ii • ij HO H
1 1 1 1 1
0.2 O.4 0.8
FIGURE 5-14. Histogram used to estimate the relationship of radical Input
rate to kj (JN02)-
-------
dilution tracer species (low concentrations of chlorinated hydrocarbons)
often are, which allows the dilution rates to be altered until the tracer
curves are fit correctly. We have found that a typical UNC dilution-rate
function increases from approximately 5.0 x 10~5 at experiment startup to
2.0 x 10~4 after about 4 hours, to 3.0 x 10"4 after an additional 4
hours. Dilution air entering the chamber is from a relatively clean rural
environment. Key among the species entering the chamber at the prescribed
dilution rate are ozone (about 0.07 ppm), CO (about 0.3 ppm) and non-
methane hydrocarbons (usually less than 0.05 ppm).
Three additional processes are used to account for the background effects
of hydroxyl radical initiation in smog chambers. These three represent
(1) Emission of aldehydes (primarily formaldehyde and acetaldehyde)
from chamber walls,
(2) Emission of nitrous acid from chamber walls, and,
(3) Heterogenous conversion of gas-phase N02 to gas-phase HONO by
chamber walls.
The third of these radical initiation processes has been shown to be
important only in two structurally similar chambers, both having Pyrex
surfaces. Chamber derived radicals in all other chambers that we have
studied appear to be caused by a combination of aldehyde and nitrous acid
emissions from the chamber surfaces. Although the source(s) of these com-
pounds have not been precisely characterized, absorption and desorption
from chamber walls seems most likely for aldehydes, whereas heterogenous
reduction of nitric acid could be the main formation route for nitrous
acid. It should be noted that phenomena involving nitrous acid (because
it contains nitrogen) may have an impact on NOX chemistry that is com-
parable to effects on radical chemistry.
As we have already noted, we generally initialize the UNC chambers with
less than 20 ppb of formaldehyde and approximately 5 ppb of HONO. In
addition, an expendable wall source of formaldehyde is included and
usually does not exceed 40 ppb. (Values for individual simulations are
presented in the initial conditions tables of Section 6.) The gas-phase
formation of these radical-initiating species is simulated with
> HONO
FORMwall > FORM
204
-------
where
kHONO = ^** x ^ ~
and & 1
kFORM = 5*81 x 10 exp(-5000/T) min .
These are, no doubt, simplifications of far more complex processes that
cannot be more adequately addressed until better data is obtained.
For the UCR-EC, we began with the representation of the EC wall radical
source as recommended by Carter et al. (1986), who represented these pro-
cesses by
N02 —hv~> 0.5 HONO + 0.5 (wall NO ^
—hv—> OH,
OH > H02.
To be consistent with the above representation, we use
N02 —hv—> HONO,
—hv—> HONO,
FORMwall > H02.
with rates for the first two processes of
and kN02*HONO = 1-37 x 10" x jN02 min"
kHONO = 3-90 x 10~4 x JN02
We utilize the same rate for the formaldehyde reaction as in the UNC cham-
ber.
205
-------
6 EVALUATION AND DEMONSTRATION OF THE CBM-IV
A large number of data sets from various experimental facilities have been
used to develop the CBM-X and CBM-IV. We have selected the best available
data (1n terms of light, chamber dryness, number of species monitored,
data density, and quality) to demonstrate mechanism performance. Since
many smog chamber data sets are only marginally useful we do not display
all results for each data set available. Rather, we demonstrate the oper-
ational characteristics of the CBM-IV, indicating both areas of improved
mechanism performance and uncertainty or needed refinement. All simula-
tions discussed in the following section were performed with the CBM-IV.
As 1n Sections 3 and 4, this section is organized according to hierarchi-
cal carbon bond groups. Simulation results and experimental data are
compared in tabular form (these tables appear at the end of the sec-
tion). For a few measurement indicators, we provide group averages and
standard deviation values; however, we feel that such indicators may
depend on the experiments simulated and thus have the potential to pro-
vide misleading Impressions of mechanism performance. In addition, a
tabular listing of only a few isolated comparison points (such as the
maximum concentrations of ozone, PAN and HCHO) 1n each experiment is not
very useful in demonstrating the overall ability of a chemical mechanism
to simulate timing and reactivity; for these kinds of comparisons, plots
of the key chemical species, including NO and N02 (which are rarely
Included 1n tabular comparisons) are most useful. Therefore, we also
provide a number of figures 1n each subsection to supplement the tabulated
data (all figures appear at the end of the section). In these plots,
symbols represent experimental data and lines represent simulation
results.
For each group of simulations discussed, two types of tables are included.
The first describes initial and experimental conditions; the second pro-
vides experimental data for a few specific points. As noted in Section 5,
Initial and chamber conditions were kept within limits felt to be reason-
able for the conditions of each experiment. Unless noted in the Initial
condition tables, these conditions were matched for both sides of the UNC
chamber for a specific day. The experimental systems for which we demon-
strate CBM-IV performance include aldehydes (including formaldehyde),
ethane, alkenes, toluene, xylenes, isoprene, a-pinene, and various mix-
206
-------
tures of hydrocarbons. We have omitted simulations of Individual alkane-
/NOX experiments because there are only a few alkane data sets (mostly
butane) and simulation of these systems is highly uncertain because the
constraints of the experimental facilities conflict with the relatively
unreactive nature of alkanes. Most experiments were performed at
extremely high hydrocarbon/NOx ratios so that ozone formation could be
measured. Unfortunately, under these conditions the chemistry 1s highly
sensitive to heterogeneous chamber effects and the results are very uncer-
tain. Hence, 1t is not clear what fraction of the reactivity is initiated
by heterogenous chamber reactions. Other experiments with more realistic
Initial conditions formed ozone at concentrations characteristic of only
background hydrocarbon levels. In either case, it is very difficult to
differentiate between reactivity due to chamber effects and that due to
alkane kinetics. Because we do not wish to demonstrate CBM-IV performance
under highly uncertain conditions, we utilize the complex mixture experi-
ments with high initial alkane loadings to demonstrate the performance of
the alkane chemistry under more realistic conditions. We conclude this
section with a discussion of the relationship between mechanism predic-
tions and experimental measurements.
ALDEHYDES (FORM AND ALD2) AND PAN
Formaldehyde chemistry plus the ICRS form one of the simplest ozone-pro-
ducing organic systems containing the characteristics of photochemical
smog. In addition, formaldehyde is an extremely common product formed
from the atmospheric oxidation of almost all organic species. For these
reasons, we attempted to simulate formaldehyde/NOx smog chamber experi-
ments as an initial step 1n our hierarchical model development protocol.
Unfortunately, as noted in Section 5, the interaction of formaldehyde with
chamber walls is a known, though poorly understood, process that adds
uncertainty to most smog chamber experiments (and therefore to the chemi-
cal mechanism developed from these data). These effects are all the more
obvious in a smog chamber system where formaldehyde is the only organic
species. Nevertheless, such simulations are very important to our under-
standing of the more complex systems discussed later. The simulation
results discussed next demonstrate both the uncertainty associated with
formaldedyde smog chamber simulation and formaldehyde oxidation chemistry.
Table 6-1 lists the Initial conditions for seven formaldehyde experiments
performed 1n the UNC chamber that represent a range of satisfactory data
and light conditions throughout each day. Simulation results are provided
1n Table 6-2 and data traces are plotted in Figures 6-1 through 6-7.
Except for the 9 October 1984 experiments (the only ones in which effec-
tive drying was performed), these data are all from the 1979-1980 experi-
mental seasons. Data for the 1984 day were collected during testing of
207
-------
the Unisearch Associates tunable diode laser measurement system at UNC.
As shown 1n Table 6-2, the overall correspondence between experimental and
predicted maximum ozone concentrations was good, but with a rather large
error that Indicates significant uncertainties 1n these simulations. This
1s to be expected, however, because of the high impact of chamber-related
processes. CO production was also well predicted, but this agreement
gives little indication of the accuracy of the chemical processes since CO
is always an ultimate product of formaldehyde oxidation.
Because formaldehyde/NOx experiments are relatively simple and test the
chemistry of a very important organic species, they are more significant
as indicators of specific formaldehyde chamber processes than as a mean-
ingful measure of mechanism capability for accurate atmospheric predic-
tions. For this reason we point out a few important characteristics in
these simulations and in systems discussed later.
First, there appears to always be a discrepancy of about 15 to 20 minutes
between the simulation results and the formaldehyde data. Whether this is
due to a delay in instrument response, Inadequate representation of for-
maldehyde wall effects, or chamber light intensity is not clear; however,
the temporal shift is usually evenly applied over the entire day (see
Figures 6-1 and 6-3 through 6-5). More important and possibly related,
are the afternoon effects seen in formaldehyde/NOx systems. The maximum
ozone concentration produced 1n systems with higher hydrocarbon-to-NOx
ratios 1s usually simulated more accurately than that of smog chamber
systems with lower initial loadings of formaldehyde. This appears to be
because the midday radical flux is higher in the formaldehyde simulations
with higher HC/NOX, producing the most ozone at that time (see Figure
6-4). In the experiments with lower formaldehyde loading, ozone is pro-
duced 1n the smog chamber through the late afternoon (when wall-related
processes can become dominant because of the depletion of all organic pre-
cursors), while the simulations often predict much lower reactivity since
formaldehyde is depleted to near zero. An example of this can be seen in
Figure 6-5 where the production of both ozone and CO increases through the
end of the experiment, while the simulation, based only on the homogenous
gas-phase chemistry of the initial formaldehyde, underpredicts reactivity
for this period. If formaldehyde 1s forced to follow the measured values
during the simulation, this extension of ozone and CO formation can be
accurately simulated. This indicates the need for a better representation
of formaldehyde wall processes, especially for experiments where high
formaldehyde concentrations are achieved either through initial conditions
or homogenous gas-phase production (as in ethylene experiments).
Because formaldehyde photolysis (producing H02) occurs during all daylight
hours and NO (the major sink of H02) is depleted early in these simula-
tions, the prominant odd-hydrogen radical species in the late afternoon 1s
H02. As noted in Section 3, the kinetic data for H02 and R02 termination
reactions is sparse and uncertain, leading to much uncertainty 1n smog
-------
chamber simulations under these conditions. In addition, such late after-
noon conditions are especially difficult to simulate for these formal-
dehyde experiments because wall termination of either HC^ or peroxides
could become a dominant factor when most of the formaldehyde 1s
depleted. Therefore, these formaldehyde data sets, as well as other more
complex mixtures that rapidly deplete NO, offer a test bed for conditions
of high R02 and low NO. We suggest that future kinetic Investigations
utilize the afternoon-period data from these experiments to investigate
chemical aspects under these conditions. Future experiments might also
start with initial NO and N02 concentrations closer to the NOX crossover
point 1n order to provide longer periods for low NOX data collection.
Such data would be extremely useful for validating a chemical mechanism
for conditions more akin to those of the rural environment. It would, of
course, be necessary to first separate chamber influences from homogenous
chemical effects. For instance, consider the effect of CO in the 9 Octo-
ber 1984 experiments of Figures 6-1 and 6-2. While the NOX and ozone con-
centrations are well suited for the side with low CO (Figure 6-1), ozone
formation is overpredicted for the high CO (50 ppm) experiment (Figure
6-2). Because the addition of CO has been shown to significantly enhance
hydrogen peroxide destruction in smog chambers (Jeffries, 1987), we sug-
gest these data may be evidence of enhanced, wall-related depletion of
radicals or their sources. Analysis of other high CO days for such an
impact could aid in indicating the magnitude of the effect and whether it
is operative only in the H02 regime of odd-hydrogen, as seen here, or
throughout the day. Fortunately, the types of uncertainties discussed
above for formaldehyde/NOx experiments appear to be mitigated in more com-
plex systems. This is particularly true for
(1) Experiments with lower formaldehyde concentrations, and
(2) Experiments with some remaining organic species near the conclu-
sion of an experiment.
These conditions tend to prevent artificial processes from dominating over
homogenous gas-phase chemistry.
Acetaldehyde simulations were performed in the development of the new PAN
chemistry and some results were given in Section 3 (Figures 3-2 through
3-7). Table 6-3 shows the initial conditions for those experiments as
well as two additional acetaldehyde and two propionaldehyde experiments
from UNC (shown in Figures 6-8 through 6-11). Simulation results and
comparisons with experimental data are shown 1n Table 6-4. Maximum ozone
concentration predictions were very good, averaging an overprediction of
4 and 5 (±8) percent for both the UNC acetaldehydei simulations and the
overall set of all acetaldehyde and propionaldehyde experiments. Further-
209
-------
more, these simulations occur over a wide range of temperature conditions,
indicating that the good correlation is due to improved PAN chemistry.
The overall average underprediction of PAN (and PPN for propionaldehyde)
was 4 ± 16 percent, or about 9 ± 32 ppb. Not only do these data Indicate
that the PAN chemistry is functioning accurately in terms of maximum con-
centration, but the PAN traces in Figures 3-2 through 3-7 and Figures 6-7
through 6-11 show a very good fit with experimental data throughout the
day. In further plots we will indicate some similarly good fits, but it
must be kept in mind that just as ALD2 is used as a surrogate for higher
molecular weight aldehydes, so is PAN used as a surrogate for larger per-
oxyacyl nitrates. Thus, comparisons of overall simulation PAN(s) versus
measured PAN may not be accurate. Nonetheless, because peroxyacyl nitrate
species are expected to have similar temperature dependencies, the curve
shapes should be similar to the measured values if the radical processes
are being handled correctly throughout the day. As noted in Section 3, we
feel the improvements to the C203 (and other radicals) chemistry has led
to better radical concentration calculations, resulting in improvements
not only to PAN but also to hydrocarbon decay predictions.
Finally, formaldehyde concentrations were calculated to be underpredicted
by about 10 ppb. This is based on limited experimental data, however, and
cannot be considered strong evidence for good predictive capabilities. On
the other hand, since all other fits are good and the chemical kinetics of
the acetaldehyde/NOx systems appear to be simple, such a good comparison
would be expected.
ETHENE (ETH) AND OLEFINS (OLE)
Because the kinetic and mechanistic information available for ethene and
other olefin chemistries has changed little in recent years, we have
included some demonstration simulations to show that the new changes in
inorganic and minor updates to olefin chemistry have not deteriorated the
quality of agreement between recent computer simulations and observed
data. The following discussion also concentrates on questions remaining
in the data, especially for the key aldehyde products.
Initial conditions for eleven ethene experiments are given in Table 6-5,
results are summarized in Table 6-6, and the data are compared to the
corresponding simulations in Figures 6-12 through 6-22. The July 1986
experiments (Figures 6-20 through 6-22) were performed during the EPA-
sponsored formaldehyde intercomparison study in the UNC chambers. As
noted, it has been our experience in simulating experiments of high for-
maldehyde concentrations, such as these ethene/NOx experiments, that water
vapor concentrations and carbon balances can provide useful information
210
-------
for modeling both chemistry and chamber effects. In particular, two
Interesting features of the 9 July data are the carbon mass balance and
the formaldehyde delay. In Figure 6-20 at 400 minutes, the data show a
carbon total for CO, ethene, and formaldehyde of about 2.4 ppmC. This is
very close to the initial value and leaves no room for additional carbon
lost to dilution or existing as unmeasured fX^, formic acid, and glycol-
aledyde products. Hence, either one or more of the three carbon-contain-
ing data traces 1s too high at 400 minutes or contamination from the walls
and the dilution air 1s significant.
The delay 1n simulated ozone coupled with the somewhat early prediction of
the NOX crossover point shown in Figure 6-20 is conslstant with formal-
dehyde Initially being condensed to the wet chamber walls and then later
(as the walls become drier and warmer) released. If formaldehyde con-
densed to the walls and did not photolyze there, then less radical forma-
tion and fewer reactions with hydroxyl radicals could occur 1n the gas
phase. Fewer radicals and hydroxyl reactions would delay the crossover
point when NO? becomes equal to NO. Because the simulation does not
account for the condensation of formaldehyde to the chamber walls, the
simulated crossover occurs earlier than the observed crossover point.
Later, the formaldehyde condensed on the chamber walls may be released
Into the gas phase and photolyze, thereby accelerating the formation of
ozone due to the extra photolytic radicals and hydroxyl reactions from the
extra formaldehyde. Again, the simulations lack this feature and the
simulated ozone appears later than the observed ozone.
A feature of the formaldehyde data, which is also seen 1n other formal-
dehyde-generating smog chamber experiments, 1s the lack of accumulation of
formaldehyde during the early morning even though the other data (such as
ethene decay) indicate that formaldehyde formation should be occurring in
the gas phase. We believe these data are Indicative of the above noted
collection of formaldehyde by the walls. It 1s especially obvious in
these ethene experiments and in mixtures with large initial fractions of
ethene, because formaldehyde 1s a very high yield product of ethene oxida-
tion.
The demonstration of the CBM-IV for propene and higher olefins, as with
ethene, involves the selection of only a few simulations from the UCR and
UNC chambers. Initial conditions for the 14 simulations (which include
propene, 1-butene, and trans-2-butene experiments in the UNC and UCR cham-
bers) are given in Table 6-7. Results are summarized in Table 6-8 and
compared with data in Figures 6-23 through 6-36.
Most of the simulations are for propene 1n the UNC chamber. As shown in
the initial conditions (Table 6-7), all the UNC-pfopene simulations
assumed Initial MONO and formaldehyde concentrations of zero. Although
211
-------
such assumptions are not used 1n other simulations, they highlight two
features: (1) the propene mechanism may be overly reactive due to overly
high radical yields from the oxygen atom reaction and the rate constant
used for hydroxyl reaction, and (2) the light flux in the early hours of
UNC simulations may be overestimated. Note, for instance, that the stan-
dard initial HONO and formaldehyde concentrations were used 1n the UCR-EC
simulations. The rapidly changing solar light flux at UNC adds a chal-
lenge to the simulation of smog chamber data not present in chambers with
constant light flux. The adjustment of assumed chamber effects such as
initial HONO and formaldehyde allows the simulation to at least more
closely match the crossover point. Without such an adjustment the poor
comparison with observed data cannot easily be assessed. For example, the
mechanism may correctly simulate chemistry after, but not before, the
crossover. If the mechanism 1s too reactive or not reactive enough before
the crossover, performance after crossover is confounded by the rapid
change in light flux that typically occurs during crossover.
Except for the need for fewer chamber effects to simulate the chemistry of
propene to the NOX crossover point, the simulations show quite good per-
formance for the propene mechanism. However, the exceptionally low simu-
lated PAN 1n the 13 July 1986 experiment (Figure 6-23) is difficult to
explain.
Figure 6-29 shows a comparison of the propene and butene-1 chemistries for
the dual chamber at UNC. Unfortunately, the CBM-IV mechanism appears to
be overly reactive for both sides on this day. However, the trend between
propene and 1-butene toward less reactivity in the 1-butene experiment is
simulated by the CBM-IV.
In Whitten et al. (1979) simulations of trans-2-butene experiments in the
UCR chamber seemed adequate when 2 times the molecular concentration of
acetaldehyde was used as a surrogate for this Internal olefin. However,
the performance is poor for the CBM-IV using this simplification, as shown
in Figure 6-31 for the blue side of the UNC chamber on 27 September
1983. One reason for this poor performance 1s the drastic reduction 1n
photolysis for acetaldehyde as now used in the CBM-IV compared to earlier
CBM versions. Nevertheless, the internal oleflns may no longer be as
Important in urban smog as 1n previous years and the need for another ole-
fin species in the CBM-IV may not be great enough for this to be a serious
problem.
AROMATICS (TOL AND XYL)
As noted in Section 4, we have attempted to simulate all usable toluene
and xylene experiments from the UNC and UCR-EC chambers. Initial condl-
212
-------
tlons for these experiments are given in Tables 6-9 and 6-11 in order of
HC/NOX ratio. Results are summarized 1n Tables 6-10 and 6-12. Ozone,
NOX, hydrocarbon, PAN and formaldehyde concentrations are presented 1n
Figures 6-37 through 6-48, for toluene, and Figures 6-49 through 6-53 for
xylenes. We also noted that a number of the hydrocarbon-reactivity mix-
ture experiments performed at UNC were substitutions of aromatic species
Into otherwise identical mixtures. The results of these simulations are
presented in the mixture subsection that follows.
For the toluene simulations (Table 6-10), the results for UNC data are
shown in Figures 6-37 through 6-43 and UCR-EC traces are shown in Figures
6-44 through 6-48. As noted 1n Section 4, the UNC simulations were the
basis for development efforts; 1n particular, JL3080R, JN2784B and OC2782R
had the best sunlight profiles. All other days had some overcast periods,
and all days from the complete UNC data set (not listed here) had problems
with severe sunlight or missing data. The average maximum ozone concen-
tration overprediction for all UNC days was 4 ± 8 percent. This is rather
remarkable, considering that the average ozone fit for a subset of these
experiments by Carter et al. (1986) was high by a factor of 2 and the
earlier version of CBM-X toluene chemistry had equally poor fits. Pre-
viously, one of the most difficult experiments to simulate was that of
OC2782R (Figure 6-42) perhaps because, though this was a clear October
day, the spectral distribution of actinic flux may differ enough to affect
the photolytically-active dicarbonyl species 1n a way that is not accoun-
ted for 1n the models. The current mechanism 1s a vast improvement over
previous mechanisms for this data set. The original CBM-X predicted 0.45
ppm ozone for this day, and the CALL (Carter et al., 1986) yielded 0.40
ppm. The improvement, explained by the Improved timing of the OH, N03,
and H0£ processes in the new mechanism, 1s due to the different product
splits produced. If one compares the new UNC simulations of lower light
and lower HC/NOX ratios with mechanisms using higher dicarbonyl yields, 1t
1s obvious that the "prompt" and high yields of dicarbonyls were over-
simplifications of product chemistry.
Another Important result 1s the timing and shapes of the curves represent-
ing the various species 1n each experiment. The shape agreement 1s much
Improved over that for earlier versions of toluene/NOx chemistry. Con-
sider, for example, the results of JL3080R (Figure 6-38), which was a
clear day that should be less difficult to simulate. The current model
presents a good fit throughout the day, while earlier mechanisms (Figure
6-54) show typical overprediction of ozone on clear days due to excess
dicarbonyl production. However, though we feel the current formulation
has tightened the error bands on prediction of ozone production from
toluene/NOx systems, 1t 1s only an Improved representation of a still
ambiguous process. This 1s evident from the mechanism fits 1n the second
half of the experiments where ozone can still be overpredicted with under-
213
-------
prediction of toluene loss. Therefore, though there has been Improvement
(especially in terms of the timing and magnitude of all species), addi-
tional Information about product distributions and their chemical reac-
tions well into the day (when NOX is depleted and H02 accumulates) would
be very helpful for developing a better representation for this period.
Unfortunately, data from the EC do not help this uncertainty because
most experiments were terminated prior to this period. Also, though the
average EC maximum ozone comparison statistics were very good, this may be
fortuitous since the uncertainty of the wall OH contribution allows alter-
ation of ozone concentrations 20 to 30 percent in either direction. The
major advantages of using data from the EC chamber were that 1t allowed us
to test our assumptions concerning the photolysis rates of the dicarbonyl
species and to verify that the process of ratioing dicarbonyl j-values to
JHCHOr was not blatantly wrong. Finally, although formaldehyde results
are not tablulated, analysis of the results from the figures shows that
the predicted values are within the rather large experimantal error bands
for both chambers.
Xylene results are shown in Figures 6-49 through 6-52 (m-xylene) and
Figure 6-53 (o-xylene). Only one UNC m-xylene/NOx experiment exists, but
fortunately it had good data collection and light conditions. As seen in
Figure 6-49, the explicit m-xylene chemistry of XYL fits these data as
well as can be expected. The UCR-EC simulations fit the general trends of
toluene. They slightly overpredict ozone and would benefit from a lower
OH wall source and higher 03 dilution. A fourth EC experiment (EC-343)
was not used because of solar simulator power problems. It 1s unfortunate
that these experiments represent the only m-xylene data available for
model development and that, therefore, strong uncertainty must be associa-
ted with this section of all mechansims.
Four UNC o-xylene experiments (there were no EC experiments) were modeled
using the XYL chemical surrogate for o-xylene. As seen in Table 6-12 this
approximation worked, on the average, but included a large variability.
For instance, the JL3080B ozone trace in Figure 6-53 shows the effect of
PAN decomposition later in the experiment as a second increase in ozone
concentration. This is typical of PAN-producing systems. Because o-
xylene produces biacetyl, which is not reactive with OH but photolyzes to
produce two acetylperoxy radicals, PAN yields are underpredicted by the
XYL reaction scheme as shown in the figure. This is an extreme case, how-
ever, occurring at the end of a very clear July day with the highest
HC/NOX ratio of the UNC xylene experiments. The other cases show better
agreement between experimental and predicted ozone, with overprediction
appearing to occur at lower HC/NOX ratios (though the AU2782 day had ter-
rible light). Again, the fit 1s acceptable considering the constraints of
existing data and the surrogate approach used; however, additional data
214
-------
would be very useful for the development of a more kinetically-based sur-
rogate approach 1n the future.
BIOGENIC HYDROCARBONS (ISOP AND a-PINENE)
Initial conditions for the UNC experiments used to test the new condensed
Isoprene representation and the new a-pinene carbon bond splits are given
1n Table 6-13. Simulation results are given 1n Table 6-14 and shown 1n
Figures 6-55 through 6-66 for Isoprene and Figures 6-67 through 6-75 for
o-pinene. For Isoprene, the average of the percent difference between
prediction and measured maximum ozone concentrations was about 6 (±23)
percent overprediction for the 12 experiments simulated. We consider this
very good evidence that the new condensed isoprene chemistry successfully
represents the photochemical processes in these systems. This 1s
especially gratifying since the Isoprene condensation was a more complex
process than that performed for carbon species such as OLE or PAR. As
many of the figures indicate (including Figures 6-55 through 6-58), the
characteristic second ozone peak caused by thermal decomposition of PAN or
PAN-Uke species under high afternoon temperatures was successfully simu-
lated for most experiments.
Because Isoprene 1s often used to represent the photochemistry of biogenic
hydrocarbons, it was also our intention to develop a condensed isoprene
representation that would predict PAN and formaldehyde concentrations as
accurately as possible. Both the maximum concentration comparisons in
Table 6-14 and the plots in Figures 6-55 through 6-66 indicate very good
fits, particularly 1n the timing and shape of the formaldehyde curves.
The new carbon-bond splits for o-p1nene were determined to be 0.5 OLE, 1.5
ALD2, and 6.0 PAR (altered slightly from previous values of 1.0 OLE and
8.0 RAR). These resulted in a closer simulation of maximum ozone concen-
trations, while retaining adequate PAN predictions. Both ozone and for-
maldehyde were overpredicted by about 20 percent on the average, although
formaldehyde data upon which to base a comparison were scarce. The
average PAN overprediction 1n Table 6-14 is very large (almost 200 per-
cent), but in absolute terms it is about 50 ppb. Thus, a-pinene is not a
major source of PAN and its relative overprediction is insignificant in an
absolute sense.
MIXTURES
To demonstrate the applicability of the CBM-IV for simulating a range of
different reactive hydrocarbon mixtures, we selected sets of experimental
data from the UNC dual smog chambers and the UCR EC and ITC smog cham-
215
-------
bers. We present these simulation results, starting with general mixture
experiments, followed by more complex substitution patterns, and ending
with the simulation of actual automobile exhaust.
Hydrocarbon Surrogate Mixtures
From UNC we simulated 19 synthetic hydrocarbon-mixture experiments per-
formed during the 1983 and 1984 run seasons. The mixtures were of auto-
mobile and urban surrogates designated "SynAuto" and "SynUrban," plus a
combination known as "SynAutoUrban." The primary difference between the
mixtures is that the SynAuto mixture is higher than the SynUrban in aro-
matic (particularly xylene) and olefin fractions; it is also much higher
in ethylene than the SynUrban mixture. Initial conditions for these
experiments are given in Table 6-15 and some results are compared with the
experimental data in Table 6-16. Figures 6-76 through 6-81 show SynUrban
results; Figures 6-82 through 6-92 are SynAuto simulations; and,
SynAutoUrban results are shown in Figures 6-93 through 6-97. Analysis of
these plots indicates that the NOX curves and initial ozone buildup are
usually simulated quite closely, with divergence of the ozone curve (if
any) occuring in the late afternoon. Maximum ozone concentrations varied
between mixtures, with a relative underprediction of 6 percent for the
Synurban mixtures and 1 percent for the SynAuto mixtures.
Interestingly, the SynAutoUrban mixtures were overpredicted by an average
8 percent. Although these differences are not large, the data in Table
6-16 indicate that this may be due to the fact that the SynAutoUrban mix-
ture experiments were all performed at a hydrocarbon-to-NOx ratio of about
7, whereas many of the other tests were done with initial hydrocarbon-to-
N0xratios of around 3 or 4. Many of the higher ratio SynAuto and SynUrban
mixture experiments were the most overpredicted, while the lower ratios
runs were the most underpredicted. As noted earlier, experiments starting
with lower hydrocarbon loading might be more susceptible to uncertain,
chamber-related radical termination or organic offgassing processes in
the late afternoon because the gas-phase organic precursors are mostly
depleted. On the other hand, the differences may easily be due to uncer-
tainties within the mechanistic representation. Certainly the SynAuto
mixtures, with their higher ethylene fractions, appear to have charac-
teristics that are seen earlier in the ethylene simulations.
The average predictions of PAN and formaldehyde exhibit similarities
based on the type of mixture. However, the absolute values, especially
for formaldehyde, often approach the sensitivity limits of .the monitoring
instrument. This suggests that the relative trends are either less for-
midable, on the basis of small absolute changes in low concentration
216
-------
data, or dominated by Individual comparisons such as the relative formal-
dehyde comparison for the red side experiment of 4 August 1984-Red.
The calculations performed to develop PAR oxidation stoichiometry were
based on an assumed atmospheric mixture. In some cases (such as the Syn-
Auto mixture and, more obviously, in the EC seven-component experiments
and the ITC multi-day tests) accurate simulations of a mixture suffici-
ently dissimilar to urban atmospheres would be expected to cause diffi-
culties 1n predicting smog chamber photochemistry. We have found that
these differences can account for up to a 10 percent variation in the
maximum ozone prediction for a few cases. Because the UCR-EC seven compo-
nent experiments and the UCR-ITC multi-day surrogate experiments used
simple alkane mixtures that were somewhat different than the previously-
described atmospheric mix, we simulated those experiments with modified
stolchiometric parameters and other alky! carbon rate constants that
represented those mixtures. In the EC seven-component experiment simula-
tions, the modified alkyl carbon mechanism appears to have been slightly
more reactive than the default mechanism, despite the Tatter's higher OH
rate constant. (This 1s probably related to a reduced alkyl nitrate for-
mation rate in the butane/2,3-dimethylbutane mixture.)
For some of the fractions of initial hydrocarbon species 1n these experi-
ments the urban surrogate mixtures generally have a higher proportion of
reactive olefins than do observed atmospheric mixes due to the use of
these olefins as a reactivity replacement for lower molecular weight alde-
hydes (that are difficult to accurately Inject). For instance, in the ITC
multi-day experiments, hydrocarbon reactivity on the first day 1s strongly
affected by the isobutene component of the mixture. Isobutene was origi-
nally added to the synthetic urban mix as a surrogate for Initial formal-
dehyde. Subsequent experiments with Isobutene show this approximation to
be inappropriate in many respects (e.g., isobutene oxidation yields PAN,
whereas formaldehyde does not). As simulations become more complex, the
effects of the surrogate approximation 1n the CBM-IV that treats reactive
olefins as aldehydes becomes more obvious. In the EC seven-component
experiments, and to a lesser degree 1n the UNCmix experiments described
later, some of the Initial delay in the simulations 1s probably due to the
surrogate aldehyde approximation. For the ITC multi-day simulations, we
utilized a simple Isobutene oxidation mechanism to eliminate the situation
where the demonstration of CBM-IV performance was primarily affected by
Isobutene chemistry.
Initial conditions and results of the UCR-EC seven-component hydrocarbon
surrogate simulation are provided in Tables 6-17 and 6-18. .Plots showing
the concentrations of NOX, 03, PAN and formaldehyde are shown 1n Figures
6-98 through 6-108. The overall simulation results are good and no unique
deviations from the data were observed. Prediction of the maximum ozone
217
-------
concentration was high on the average, by about 10 percent or 46 ppb; how-
ever, this 1s partially attributed to the overpredictlon of EC-232, an
experiment that was terminated before the ozone concentration reached a
maximum value. Comparison of the formaldehyde results with UNC mixture
experiments (Table 6-16) Indicates that whereas formaldehyde was very
closely predicted 1n the EC (overpredicted by 8 ± 13 percent or 40 ± 70
ppb absolute in experiments that produced 500 to 700 ppb), the UNC results
were usually underpredicted. Whether this is a chamber, monitor or
mechanistic problem is not certain; however, Carter et al. (1986) reported
similar findings. This independent verification seems to point to either
a more sensitive monitor at UNC (1t could, of course be responding to
species other than HCHO) or a stronger loss mechanism for formaldehyde in
the EC. CBM-IV PAN predictions are usually high compared to EC data and
these experiments follow that trend (high by 43 percent on the average).
The UCR-ITC multi-day simulations are often used to determine if a chemi-
cal kinetics mechanism can succesfully simulate the transition between
daylight and nocturnal regimes. Initial conditions for the UCR-ITC exper-
iments are provided in Table 6-19, with concentration traces shown in
Figures 6-109 through 6-113. Comparisons of daily ozone maxima and over-
all PAN and formaldehyde maxima are given in Table 6-20. Considering the
evidence discussed in Section 5 that unanticipated chamber effects occur
between the first and second days of ITC multi-day experiments, we find
that the predicted and measured concentrations for the species listed in
Table 6-20 agree quite well. The average absolute difference between pre-
dicted and measured maximum ozone concentrations is essentially zero, with
a standard deviation of ± 32 ppb. The second through fourth day compari-
sons are similar for ozone maxima and for PAN and formaldehyde maxima as
well. In addition, the concentration profiles follow the diurnal trends
of the experimental data as closely as can be expected for smog chamber
experiments of up to four days.
Substituted and Complex Mixtures
We also selected a set of UNC experiments from the 1981 and 1982 hydrocar-
bon reactivity tests in order to focus more closely on the chemical
effects of substituting aromatlcs (and a few other species) in surrogate
hydrocarbon mixtures. Table 6-21 provides information concerning initial
concentrations and characteristics of each experiment. The two base mix-
tures used in these experiments, SIMmix and UNCmix, consist of approxi-
mately 70 percent paraffin and 30 percent olefin carbons. The difference
between them is that SIMmix contains only butane, pentane, ethylene and
propylene, while UNCmix is a mixture of eleven paraffin and olefin
species. (It was later used as the paraffin and Olefin basis from which
the SynUrban mixture was generated.) Most experiments simulated in the
218
-------
UNC study were UNCmix substitutions, using either a complex (COMARO) or
simple (SIMARO) mixture of aromatics as shown in Table 6-21.
Simulation plots are presented in Figures 6-114 through 6-135 and compari-
son of maxima for measured and simulated ozone, PAN and formaldehyde are
given 1n Table 6-22. Considering the chemical range of substitution
species (see Table 6-21), ozone prediction results are very good, showing
an average overpredictlon of 8 (±78) ppb or 7 (±25) percent. However, we
again caution against using simple statistics in these comparisons. For
example, the 7 December and 11 November 1982 experiments are extremely
difficult to simulate because of the low temperature conditions present
throughout the day. Removal of these datasets from our average reduces
the overprediction percentage to 4 (± 17) percent. In the work of Carter
et al. (1986), a similar removal of the three December and November exper-
iments from the compilation of 25 simulations changes the average predic-
tion from a 21 (±95) percent overpredictlon to a 7 (±24) percent underpre-
dlction.
We also note that the best simulations for this group of hydrocarbon reac-
tivity experiments occurred on days when drying was performed and no con-
densation was observed. This subset of experiments (AU3181, SE1481,
SE1682, SE1081 and AU2681 [SE0381 was eliminated due to bad light on that
day]) contained substitutions of butane, toluene, xylene, ethylbenzene,
and the SIMARO and COMARO mixtures. The fact that both maxima values and
overall goodness-of-fit for ozone and NOX curves were the best for these
days indicates the benefit of chamber drying.
Overall, formaldehyde predictions were good for these hydrocarbon reac-
tivity simulations, resulting 1n an average underprediction of only 30
(± 30) ppb or 18 (± 25) percent. PAN prediction was high, as it was in
the earlier UCR-EC experiments, probably because the CBM-IV used PAN to
represent all peroxyacyl nitrates. We note again that the formaldehyde
plots in Figures 6-114 through 6-135 show a delayed, but then rapidly
increasing formaldehyde concentration similar to that occurring in the
ethylene simulations. This appears to Indicate the early collection of
formaldehyde or formaldehyde precursors on the chamber walls, followed by
a midday release that produces in-chamber concentrations more rapidly than
the values predicted for homogenous gas-phase chemistry in the simula-
tion. The identification and simulation of such a process would probably
enhance the goodness-of-fit for the midday period of these experiments.
Because there are so few individual aromatic/NOx smog chamber data sets
other than for toluene and the common xylenes, these simulations provide
the only useful tests of the approximations used 1n the CBM-IV for higher
molecular weight aromatic species. Although the complexity of these
experimental mixtures precludes definate verification that the carbon bond
219
-------
representations of these aromatic species 1s accurate, we may conclude
that the CBM-IV representations are satisfactory enough to produce correct
fits for a number of indicator species and a relatively good prediction of
ozone and formaldehyde. One final simulation comparison can be made for
the 19 September 1984 experiments, which were tests of the high molecular
weight aromatic fraction (toluene, m-xylene, o-xylene, and l,2,4-tr1-
methylbenzene) of the SynAuto mixture. Initial conditions for these
experiments are given in Table 6-21 and results are shown 1n Figures 6-134
and 6-135, and in Table 6-22. The interesting aspect of these experiments
is that the side with higher initial hydrocarbon concentration (the NOX
concentrations were matched between sides) produced significantly less
ozone, but at a faster rate. This is characteristic of rapid formation of
radicals in xylene systems, followed by depletion of virtually all NO (and
later, N02), and termination of ozone formation potential. The new aro-
matic reaction representation provides a better fit with these data than
was previously obtained from the earlier versions of the CBM. However,
uncertainties are still apparent in the mechanism. On the red side (high
HC/NOX shown in Figure 6-135) the radical flux is high for a brief period
of around 200 minutes before N02 is depleted (the remaining N0£ in the
experimental data 1s a PAN response of the UNC chemiluminescent NOX
meter). This high radical flux, caused mainly by the prompt formation of
radicals from xylenes, removes a large fraction of toluene to form addi-
tional reactive products. In the blue side (Figure 6-134) this prompt
formation is less dramatic, and the decay of ozone is somewhat underpre-
dicted (with the accompanying underpredictlon of radical reaction
products), resulting in slower ozone formation. The fact that NOX remains
in the system for a longer period, however, results in a correspondingly
longer period before the system is effectively "shut down" by delpetion of
NOX. This eventually produces higher ozone concentrations at the lower
HC/NOX ratio.
SUMMARY OF SIMULATION RESULTS
The overall ability of the CBM-IV to predict maximum ozone and formal-
dehyde concentrations 1n smog chamber experiments appears to be good for
the evaluation tests discussed above. Figures 6-136 through 6-138 give
scatter plots of simulation results compared to measured ozone maximum
concentrations for the following groupings:
(1) Formaldehyde and larger aldehydes,
(2) Ethylene and oleflns,
(3) Toluene and xylenes,
(4) Isoprene and a-pinene,
220
-------
(5) The various mixtures from UNC and UCR, and
(6) All data from groups (1) through (5).
Figures 6-139 through 6-141 give similar diagrams for maximum formaldehyde
concentrations.
The predictive bias for ozone was rather small 1n all cases. The largest
overpredlctlon was for the Isoprene and a-p1nene set (13 ± 31 percent) and
primarily attributable to a few a-p1nene outliers. The average overpre-
dlctlon for the mixtures (excluding only one outlier, the first day ozone
calculation for ITC-633) was 2 percent with a ± 22 percent standard devia-
tion. Similarly, the overall average overpredition, including every
experiment, was 5 percent with a ± 23 percent standard deviation.
For formaldehyde, the uncertainties were larger. This was not unantici-
pated, however, since formation reactions of formaldehyde are not clearly
defined, nor 1s the measurement as precise as that for ozone. As we noted
earlier, the Isoprene mechanism, which produces high formaldehyde yields,
predicted the experimental data quite well. The average overpredlctlon
was 5 (± 16) percent for the Isoprene and a-pinene experiments combined.
For the 68 mixtures simulated, the maximum formaldehyde concentration was
underpredicted by 9 (± 34) percent. These results are considered good
because of the uncertainties noted.
We caution against comparison of the associated uncertainties with those
presented by other researchers, since the range of tests (I.e., smog cham-
ber data sets) performed between any two such studies undoubtedly dif-
fers. The mechanism demonstration and evaluation discussed here has been
based on the simulation of smog chamber data sets selected because they
were the best available and, hopefully, provided a reasonable standard
against which to test. However, some uncertainty still exists in the
Initial and environmental conditions of the experiments that will trans-
late into a portion of the error noted 1n these comparisons.
221
-------
NO and N02 (plus PAN) and 03
t. Ittt ft) OJM3 •»MC/OJ»0 MV0.160 MO? MC/*0« -
•00
Formaldehyde
t. 1M4 (•} 0*«J NHMC/D.KO MV0.1U MO2 HC/Wtfc •
OJ10 -
O.40 -
OJO -
O-IO -
O 10 -
oao
FIGURE 6-1. Simulation results for UNC experiment OC0984B
(symbols are experimental measurements)
-------
NO and N02 (plus PAN) and 03
CM«h»r ». 1M4 (•} O.M2 XKMC/OJ53 MO/O.XW N02 MC/Ntfc • 1.711
0.70-
000 -
0.20-
O.10 -
OJK
tOO
000
Formaldehyde
D*<*.- ». 1IK (») 0*«7 MKX/OJSJ WDXD.JM MOJ MC/N
-------
NO and N02 (plus PAN) and 03
i. 1*7* (•} i «o NHKxo.rrr NO/O^T* ttn
Formaldehyde
1. HT» (I) 1^0 HUMC/OJTT «VDJJ7« NO 2 MC/VO .
O.4B -
0-10 -
100
•00
FIGURE 6-3. Simulation results for UNC experiment AU0179B
(symbols are experimental measurements)
-------
NO and N02 (plus PAN) and 03
2. 1»T» (•} 1.01 HHHC/0.1B7 MOXO-OM N02 MC/MO» - 44B
Formaldehyde
2, HT» (•) 1J01 HHHC/e.167 NQ/O.OIO NO2 MC/Mte
1.10-
1.10-
B.TO -
Tto> (mko)
Carbon Monoxide
t. 1»7» <•} 1*1 HHMCX0.1I7 KVOACO IO7 MC/NO« «
»00
FIGURE 6-4. Simulation results for UNC experiment AU0279B
(symbols are experimental measurements)
-------
NO and N02 (plus PAN) and 03
4. 1*71 (I) OJD4 tttHCXO.170 NO/D.07D NOJ MC/HO. - 2.10
Formaldehyde
I 4. 1*7* (•) OJD4 MMMC/0.170 «O/t>-OTO MO} MC/NOp - 1.10
Carbon Monoxide
•••u* 4. ItTf (•} OJD4 MHHC/D.1TO MD/D.O70 NO? MC/MOi • Z.10
0_to -
0.10
too
FIGURE 6-5. Simulation results for UNC experiment AU0479B
(symbols are experimental measurements)
-------
NO and N02 (plus PAN) and 03
I. 1»7» {•) 1 J0» NMMC/0.407 HO/0.112 M» NC/K>« •
too
1.40
140
140
1.10
IjDO
0,70
•.•0
Carbon Monoxide
». 1*7* (•} 1JM MIH^D^OT MO/0 1C MOZ MC^tO. > S44
•(-*•)
FIGURE 6-6.
Simulation results for UNC experiment AU0579B
(symbols are experimental measurements)
-------
NO and N02 (plus PAN) and 03
Formaldehyde
t*f~- 12, IMC <*) 0*«i MMC/B.in HO/D.110 MD7 MC/NO« • 1J
Carbon Monoxide
"•^•»- IX. 1MD (•) O*«1 HIMC/O.IT1 MD/0.110 MOI MCyTCI. » 1 J6
«00
FIGURE 6-7. Simulation results for UNC experiment AU1280R
(symbols are experimental measurements)
-------
NO end N02 (plus PAN) end 03
ALD2
^M 14. *M7 (V) KM MMC/Un m/MII IB? IC/Wk •
PAN
Formoldehyde
Carbon Monoxide
FIGURE 6-8. Simulation results for UNC experiment JN1482R
(symbols are experimental measurements)
-------
NO ond N02 (plus PAN) ond 03
ALD2
\ -
PAN
Formaldehyde
i«r HMC/OJV to/*am HD; Mt/«o> • •>
Carbon Monoxide
FIGURE 6-9. Simulation results for UNC experiment AU2482B
(symbols are experimental measurements)
-------
NO and NO2 (plus PAN) ond 03
Propr1onoldehyde(ALD2)
*»- 14 MB (•) u>
Formaldehyde
Carbon Monoxide
4»- 14. nw (I)
FIGURE 6-10.
Simulation results for UNC experiment JN1482B
(symbols are experimental measurements)
-------
NO ond N02 (plus PAN) ond 03
Proprionoldehyde(ALD2)
MC/1O. • • '4
PoN's
5 «W IC^D. • • 14
formaldehyde
I ""
Corbon Monoxide
FIGURE 6-11,
Simulation results for UNC experiment AU2482R
(symbols are experimental measurements)
-------
NO and N02 (plus PAN) and 03
4. itn (•) 9Jtn NMMc/o.t*T NO/MMO NM MC/MO* • »**
Formaldehyde
ro
to
CO
OJJ •
i «. ttn <*) »*rr MMC^.KT NO/V.MO m
Eihen.
4. Itlt (»> «*»» l«l«yd.1«T HO/V.MO MM MC/W..
FIGURE 6-12. Simulation results for UNC experiment AU0479R
(symbols are experimental measurements)
-------
O.TO
OiM
NO and NO2 (plus PAN) and 03
not HC/MO. •
Tim* (>nh«)
Cfhene
FIGURE 6-13. Simulation results for UNC experiment OC0584B
(symbols are experimental measurements)
-------
NO and N02 (plus PAN) and 03
kar I. 1M« <•) lift MMC/OJM MB/AM* MM MC/NDi « I
•.TO
Formaldehyde
•. ItM (•) HM NMHC/0.1M MO/QJW* HM
OJO
0.10-
MO
Km (•*!•)
Elhene
•. ItM <•) I.1M MHHC/U9M W/OJOM MO* MC/M8. i
tlm* (mfeia)
FIGURE 6-14. Simulation results for UNC experiment OC0584R
(symbols are experimental measurements)
-------
NO ond N02 (plus PAN) and 03
1O. 1*7* (•) I.01O m*C/U40i MO/Q.114 MD1 HC/NO» • • Tf
SOO
Elhene
ItM (•) U10 HIMC/9.4IM MD/B.114 MM MC/MO> » t.7T
Tim. (mkv)
FIGURE 6-15.
Simulation results for UNC experiment AU1078B
(symbols are experimental measurements)
-------
NO and N02 (plus PAN) and 03
•I 1& 1»M (•) IjMD m*IC/D.41t NO/0.11* MM MC/MO* • 1»f
**•(•*!•)
Eihen*
10.1*n(>) 1.MD MMC/O^lf W/D.1tlM» HC/NOi » IJtT
an-
OM
* *
too
FIGURE 6-16. Simulation results for UNC experiment AU1078R
(symbols are experimental measurements)
-------
NO and N02 (plus PAN) and 03
••> u. 1*7* (•) t*t Mmcso.4B> ND/a.100 wn MC/NO> • I.M
1JO
1.4O
1JO
1JO
1.1O
1.OO
O.M -
B.M -
0.70-
040
O.M -
O.M -
0.10 -
Efhene
tt. 1*70 <•) *.M MMHC/0.4IO NO/D.I 00 MM HC/NO« • t.tt
Carbon Monoxide
U. 1»Ti <•) J ij NUHC/O.410 MO/0.10O HOt MVMO« •
i i r
1 JO -
1JO -
1.10 -
IJOO -
O.M -
O.M -
0.70 -
O.M -
OM -
O 40 -
O.JO -
o.ro -
o.io -
ooo
I
too
TIM (mkM)
{•"*>•)
FIGURE 6-17. Simulation results for UNC experiment AU2378B
(symbols are experimental measurements)
-------
NO and N02 (plus PAN) and 03
•»SS. !•?•(•) 1*4 MIMC/0.4tON>/0.110 MM MC/MO..IM
Elhene
n. \m 00 t*t Mwc/o.410 mxo.no wt HC/NO>
1.40-
1JO-
1.10-
1.10-
1;00-
O.M -
O.M
O.TO -
e.«e
e.n •
e.io-
ooo
too
tlflV (•"»»)
FIGURE 6-18.
Simulation results for UNC experiment AU2378R
(symbols are experimental measurements)
-------
o.ro
OJO
NO and N02 (plus PAN) and 03
1MO
-------
NO and N02 (plus PAN) and 03
1MO (0) Mat MMC/OJ11 NO/O0Z1 MM MC/MOl • O.M
Ethene
My t. 1tO» (•) IJM H«He/0.1«1
•01 HC/Nthi
DJO-
O.M-
am-
o.io-
sao
Formaldehyde
IMt (0) 1473 MMC/OJ11 M/VAI1 HM HC/HOl •
OJO-
0.10-
Carbon Monoxide
MfO. ItW(O) 14MHIMC/VJ11 MO/OAX1 •«» MC/WO. - S.*4
Tim* (mtM)
tlfn* (mkw)
FIGURE 6-20. Simulation results for UNC experiment JL0986B
(symbols are experimental measurements)
-------
NO and N02 (plus PAN) and 03
Mr 0. 1MB (•) O.CM MMC/D^M M>/OjOM HOt MCSMD. • l.T»
0.40
o.»o
0*0
0.10
too
1*M (•)
Efhene
MMC/V.M4 HO/UOM HOt HC/MOi
one
o.js
OJO
O.M
0-10
0.1*
010-
0.0* -
one
too
Formaldehyde
'••* <•) OJN MMC/V.IM HO/0 014 HBt MC/Wte - *.?•
an -
O J« -
o.n -
031 -
e.to -
o.i« -
o i« -
0,14 -
0.11
0.10 -
OM -
O.M -
OJM -
OJ» -
OJX
Tliw (mh.)
Carbon Monoxide
». it«M (•) O*M NiMc/o.rM m/t>.074 not MC/NO* - t.r*
o.to
MO
400
i (mkic)
FIGURE 6-21. Simulation results for UNC experiment JL0986R
(symbols are experimental measurements)
-------
NO and N02 (plus PAN) and 03
r II. <•••(•) 1.11 MMC/O.MS MO/MIT MM MC/NO. • 1.
Formaldehyde
•II. 1tM(l) 1.11 MMC/«.f«l NO/««4T M0> NB/HOl •
OJS
OJO-
O.M -
O.M-
0.1* -
0.10-
MO
Mr K. t*M(l) 1.11 MMC/V.S4> NO/0047 M» NC/NOi •
•.40-
MO-
«ae
FIGURE 6-22. Simulation resu'ts for UNC experiment JL1386B
(symbols are experimental measurements)
-------
NO and NO2 (plus PAN) and 03
r 11. 1«M (•) 04* NMMC/t>.14« NO/U048 NO* MC/NOi - t.lt
0.00
Propene (OLE)
• II. 1*M (•) 04* NMHe/0.14* MO/OJMB MO* HC/NO* - t-lt
PAN
•11. 1*M(*) O4* MMC/O.l«t NO/O.04I HO* Ht^MO, .
O4MS
044
OAU
O.OI -
OJJM
em
Offlt-
0.01 -
OAM
0
KM (mkv)
FIGURE 6-23. Simulation results for UNC experiment JL1386R
(symbols are experimental measurements)
-------
NO and N02 (plus PAN) and 03
11. tM4 (•) t*ta MMb>«.*4* MO/O.IM Not MC/MOI
ALD2
Propene (OLE)
11. 1«M W t JM NMMC^.tM NO/0.1M MOt HC/MOl
O.M
OJM
PAN
11. 1«M <•) titt NMMC/D.1M MO/O.1M MM HC/tNki
FIGURE 6-24. Simulation results for UNC experiment OC1184B
(symbols are experimental measurements)
-------
O.M
NO and N02 (plus PAN) and 03
tar tl. IM4 (•) 1 M* NIIMC/D.aM NO/O.1M M01 HC/WCh -
Propene (OLE)
It. 1M4 (•) t*fl» wrwVO.SOl NQ/O.HO MOJ HC/WOl
O.M •
OM-
O.W-
0.10 -
Formaldehyde
It* 1W4 \BJ 1 449 WWC/fl.flWt MO^O.100 MQt
O 40 -
O.M -
O JO -
an -
o.n -
0.1* -
0.10 -
000
MO
FIGURE 6-25. Simulation results for UNC experiment OC1284B
(symbols are experimental measurements)
-------
NO and N02 (plus PAN) and 03
*l»^**«»»»W '•««• P"»«/«J»t "8/0.1 SI MM MC/NOi«I.lt
AL02
I 1C 1»»t (•) 1.4H MMC/OJff NO/V.1I1 MM
MC/Ml • I.It
11*. 1*1* (•)
Propone (OLE)
PAN
C/D.J1J NO/O.ttl NOt
S.t*
Tim (fnk
FIGURE 6-26. Simulation results for UNC experiment AU1679R
(symbols are experimental measurements)
-------
NO and N02 (plus PAN) and O3
tl. )»?•(•) 1 JI MMMC/OJ44 WVO.IIf NOt MCXMO. -t.»
o.tt
o.to
0-lt
0,14
O11
O.1O
O.OO
0.0*
004
i*. I»T» (•)
ALD2
WKMC/UJ44
HCSHO* -t.n
tea
too
Km (mkv)
O.M-
0.40 -
O JO -
v O.M-
O.tO-
O.M •
0.1* •
0.10-
OJO-
Propene (OLE)
1.JJ
HC/MOl ~t.t
TIM (mko)
FIGURE 6-27. Simulation results for UNC experiment OC1278B
(symbols are experimental measurements)
-------
NO and N02 (plus PAN) and 03
M, 1*M (I) 1 *H MMe/VJS? MD/a.104 HDt MC/NO» •
ALD2
M. !•» (•) 4.S14 NtMC/V.JSr HO/V.IM MM MC/NM
|
0.4t
0.4O -
O.M
O.JO
e.M
O.M -
0.11 -
•.10
Propwne (OLE)
t*. ItM (•) «.t*4 MIMC/feJJ? NO/Q.I04 MM HCyTO. • i-T7
—I—
top
Tlra (mkv)
FIGURE 6-28. Simulation results for UNC experiment OC2578B
(symbols are experimental measurements)
-------
NO and N02 (plus PAN) and 03
.<-*-}
NO and N02 (plus PAN) and 03
"tor IT. 1 ••)(*) 1 4*2 HHMC/0.40a «X/0^<* N02 HC/Ntfc -U1
oao
FIGURE 6-29.
and SE2383R
Simulation results for UNC experiment SE2383B,
(symbols are experimental measurements)
-------
0*0
NO and N02 (plus PAN) and 03
. IMS (•) ZJM NUHC/OJK NOSO.OU N02 MC/NCta •€.?•
NO and N02 (plus PAN) and 03
FIGURE 6-30.
and SE2583R
Simulation results for UNC experiment SE2583B,
(symbols are experimental measurements)
-------
NO and N02 (plus PAN) and 03
XI. 1 MI (I) 1 jOt NHMC/0.1M MO/O.OM tCl HC/tCta
0,10-
OJ»
><•*->
NO and N02 (plus PAN) and 03
21. IMS (•) 1 *n NMHC/O.Bt NQ/OO44 «» HC/NOB
FIGURE 6-31. Simulation results for UNC experiment SE2783B,
and SE2783R (symbols are experimental measurements)
-------
NO, N02 and 03
ALD2
I -
Propene (OLE)
Formaldvhydt
i ::;:
PAN
i-
•i
FIGURE 6-32. Simulation results for UCR experiment EC-177
(symbols are experimental measurements)
-------
ALD2
&*4 -
•.11-
Propene (OLE)
Formaldehyde
ft.14 •
kit-
PAN
i —
FIGURE 6-33. Simulation results for UCR experiment EC-121
(symbols are experimental measurements)
-------
NO. N02 and 03
ALD2
Propene (OLE)
*.<••
Formaldehyde
I -
FIGURE 6-34. Simulation results for UCR experiment EC-278
(symbols are experimental measurements)
-------
NO. N02 ond 03
ALD3 (ALD2)
•.11
•If
•.<<
Bu1«n»-1 (OLE)
Tormaldehyde
PAN
FIGURE 6-35. Simulation results for UCR experiment EC-123
(symbols are experimental measurements)
-------
NO. N02 and 03
AL03 (ALD2)
Formaldehyde
PAN
FIGURE 6-36.
Simulation results for UCR experiment EC-124
(symbols are experimental measurements)
-------
O.M
NO. N02 (plus PAN) ond 03 Troces
• n, 1*tt (t) 131 t*f~~ T^Li/q.Mi HIW.Mt HOI K/N(V>4t.l
O.*0 -
010 -
Formaldehyde Trace
Am n. t tr» (I) 13\ ffr-w T.K—/O.MS MO/D.Mt MM MCXWO.-M «
Toluene Trace
, n.
I.OO -
1.8O -
I.M
.t.oo -
e.te
o to
oto
o.to -
o.oo
FIGURE 6-37. Simulation results for UNC experiment JN2779B
(symbols are experimental measurements)
-------
BJO
NO, N02 (plus PAN) and 03 Traces
Mf ». 1MO (If) 0.88 pp»« T«km/V.*M NO/D.OM MM MC/*O».*1.1
O.«0 •
*••(•*.)
Toluene Trace
A* JO. 1 WO (») DM »r-» Tilywi/0.1M NO/D.OM NM
e.so-
0 $0 -
O.M-
0.10 -
—r 1 1 r
too '00
•00
O.10
Formaldehyde Trace
Jut? JO. 1MO (*) O9S ppm. T«lMM/O.1M NO/D.O7* NO* MC/NO»11.1
o.o*
o.o*
O.OT -
0.0* -
O.M -
O.04 -
O.01 -
O.OJ -
O.O1 -
o.oo
0.09
PAN Trace
J-V 39. IMO <*) OAS ppn. T«liMn./O.I9« NO/O.W* NO? MC/MO»»?1.2
40O
Tim* (mill*)
FIGURE 6-38. Simulation results for UNC experiment JL3080R
(symbols are experimental measurements)
-------
0*0
NO, N02 (plus PAN) and 03 Traces
n. 1M4 (•) 0 TO per*. Tduvw/O.JM NO/V.O39 NOT MC/NO..U.7
8 SO
O.TO
o.*o
090
Toluene Trace
018
00»
00* -
DOT -
0 0* -
0.09
004 -
00]
O.OJ -
001 -
ooo
Formaldehyde Trace
Jun* n. 1M4 (•) O.TO ft*" T«lMn./O.WO KO/D OJ9 NW HC/MO«»t4.7
IOO 4OO tOO
Tim (rnhw)
PAN Trace
, 1M4 <•) O.TO yfx~ T.lum/V.m MO/O.OH NOt MC/WO..14.7
-1 1"™1—1 1 1—
JOO 4OO
-1 1—
too
FIGURE 6-39. Simulation results for UNC experiment JN2784B
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) and 03 Traces
O.M
it.
NB/OJDM not
O.M
o.ie
not
O.M
O.M
0 09 -
O.01 -
0.01 -
formaldehyde Trace
18. UTt (*) 0.74 ppm. Tolum/0.77* NO/DAM MM MC/WO.-14J
Toluene Trace
O.M
O.M -
O.M -
O.M -
O.M -
0.10 ->
18. «tft (0) O.T«
NO/DM* MM MC/MO.-14J
O.M
O.O7
O.M -
O.M -
O.OS -
0.01 -
PAN Trace
18. <»Tt (•) OT4 p?~~ T«k»n./07T« NO/OAM NO? MC/MO..14J
I'
100
1—
•00
FIGURE 6-40. Simulation results for UNC experiment AU1579B
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) ond 03 Traces
•> 1. 1MJ (*) O.WS p*m> T«*MM/T>.144 NO/VM1 HOt MC/WO—11 J
0 SO -
000
**•(•*!.)
Toluene Trace
I 1. IMS (•) «JM WKV T«kiM/»J44 NO/OM1 MO*
O.TO-
O.«0-
•M-
O.W -
0.00 •
0.10
0.0*
OO»
0.07 -
0.0*
i 0 05 -
0 O« -
DOS -
0.0>
0.01 -
O.OO
009
O.OJ
Formaldehyde Trace
. 1*«S (*) 0«8i w-~. T«KMn./1]J44 HD/t>mi NOJ "(VWO.-n 4
PAN Trace
T«lu«<./DJ44 HO/OM1 HOt
FIGURE 6-41. Simulation results for UNC experiment AU0183R
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) ond 03 Traces
MC/MO.-M.4
o«e
a. is -
O.JO -
e.»s -
ois -
0.04
o.oo
0.00
•00
Toluene Trace
T7. 1M> (*) 0.«4I |»»» T^MMSO.tW MO/O.114 NOt MC/-WO.-11.4
O.TO -I
0 JO -
0.10 -
1OO 40O
time (mix.)
•00
0.09
0.01 -
Formaldehyde Trace
'. 1*8? (*) 044? fpr~ TeltOTM/O.ZM NO/D.114 NO*
FIGURE 6-42. Simulation results for UNO experiment OC2782R
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) and 03 Traces
«<«M* tr. (•« (•) o.47« rr~~, T«IV*/'D.JIO *a/o m NOT WVNO»«.M
OSS -
• 00
Toluene Trace
11. 1M? (•) «.4M pp*~ t«l»w«/O J10 NO/O.1t1 NO* MC/NO«-*.M
o.so -
o.w-
ou
0.11 -
0.10 -
o.ot -
ooe
OOT -
o.o« -
O.O9 -
0.04 -
DOS -
O.OJ -
0.01 -
o.oo
0.0'!
0.014 -
0.01] -
O.O1J -
O.O11 -
0.010 -
0.00* -
oam
0007 -
000*
o.oos -
0.004 -
O.OOJ -
0 003 -
0001 -
o.aoo
Formaldehyde Trace
. !••? (§) O.flt f^r,, To(u»i«^0,J10 WO/0 1J1 HO? MC/M0..4»«
100 400
-I 1—
•oo
PAN Trace
(•) 0.47* pt»— T.hMn./O 510 MIVO 1J1 NOT HC/MO..« M
}00
-1 1—
(00
FIGURE 6-43. Simulation results for UNC experiment AU2782B
(symbols are experimental measurements)
-------
NO, N02 and 03 Trace
I.JO t*~~ T«lu»i«/O.4S HO/OOt HOI HC/WO»1T.1
O.10
Formaldehyde Trace
UC» (C-IM 1.10 ppinr T.feM/V.41 NO/DM MOT MC/W(fc«1 ?.«
0.0* -
O.M
O.OT -
0.0*
O.M -
O.O4 -
DOS -
O.OJ -
0.01 -
000
Toluene Trace
UCO K-ltt t.» ^inr T.tMn^O.43 HO/OM MOT MC/W0..1M
UCO R-XM 1 .»
PAN Trace
T«lum/O.43 MO/OJM MM HC/NQralM
O.OT -
0.03 -
0.01 -
100
FIGURE 6-44. Simulation results for UCR experiment EC-266
(symbols are experimental measurements)
-------
NO, N02 and 03 Trace
OC» ft-»»1 1.1S
* Ho/tias NOT MC/HO..S» «
UC» rC-J71 1.1S ppt-v
Formaldehyde Trace
NO/DOS NO? MC/WO»-I7.«
1.1 -
1 -
0.* -
e»
o.r
I ;;
0.4 -
0.1 -
0.} -
01
0 -
Toluene Troce
Ue* K-1T1 '.'11 •.•--ij.'tt-'T "ff.~** *"** MC/*0«-ST.«
PAN Trace
UC» K-171 1.19 pp^w T^uw^/O.'t HO/BBJ MOJ
FIGURE 6-45. Simulation results for UCR experiment EC-271
(symbols are experimental measurements)
-------
0.40
o.ss -
O.M -
O.M -
NO, N02 and 03 Trace
ue* rc-sn o s» r*~~ T«IHVM/O.M NO/O.IO NO? nc/wo«-r»
O.M
toe
300
0 •
Toluene Trace
UC* K-CT 0.87 ff~. TclMww^I.M HOXO.10 NOt HC/MOnl.7
O.S -
o.s
o.* -
—I—
too
O.OT
OJO1 -
O.MS
O.04 -
0.039 -
O.OS -
0.07S -
O.W -
O.O1S
O.0< -
ODM
O
Formaldehyde Trace
uc* rc-irr o.sr «—. w««wo M M(VO.IO MM He/»~* TWu«W0.3« NO/O.10 NOt
wo
300
400
-T -i r- r-
FIGURE 6-46. Simulation results for UCR experiment EC-327
(symbols are experimental measurements)
-------
NO, N02 end 03 Troce
uc» te-s«o o •»
o.w H
FIGURE 6-47. Simulation results for UCR experiment EC-340
(symbols are experimental measurements)
-------
0.4S
NO, N02 end 03 Trace
UC» IC-?*« !.'• |i»l~ T«lMn«/0.« HO/0.0* N0» MC/WO»«17.0
NO, N02 and 03 Trace
1 .OT pp~* T»liMM/-0.44 NO/OM NO? MC/NO»«1S.9
O.M
IXO 1M no S4O
NO, N02 ond 03 Trace
-»• 0.0
0.»4 -r
NO, N02 and 03 Trace
OC» R-7TI O.M pf*~ TWu»»/0.'0 NO/OJ)1 NOZ MC/NOnST.S
FIGURE 6-48. Simulation results for UCR experiments EC-264.EC-265,
EC-269, and EC-273 (symbols are experimental measurements)
-------
NO, N02 (plus PAN) and 03 Traces
OM -
O.10
O.O* -
OOR
o.or -
o.o« -
o.os
0.04
001 -
00?
001 -
ooo
Formaldehyde Trace
r. 1M4 (») OJS M»— r--l'rt>«/0 SO M0^).04 NOI
090
Old
OM -
O.J4 -
O.JI -
OIO -
0.1S
Ot«
O14 -
01»
0 10
OW
00*
004 -
00? -
OOO
i IT. 1M4 I
Xylene Trace
NO/t).M HO?
400
PAN Trace
Jan* IT. 1M4 (*) 03S PIWW i"-»r<«n«/O.SO HO/D.04 WOi
I
tOO
FIGURE 6-49. Simulation results for UNC experiment JN2784R
(symbols are experimental measurements)
-------
NO. N02 and 03 Trace
0.90
0.40
uc« rc-34*j e.4»
Xylene Trace
o.s •
(1C* IC-S4* 0.4> p»~* ~-*l>~m/V39 NO/0.04 NOT
0.4 -
O.t -
I I 1 T—
1OO TOO
300
0.1 s
Formaldehyde Trace
UC» CC-J44 0.4> ppn» fn-«|4«n*/t>.20 NO/O.Of NO* MC/NO»18A
0.14 -
0.1 S -
O.I* -
0.11 -
01 -
O.0» -
O.Ofl -
0.07 -
O.O« -
DOS -
0.04 -
DOS -
OO» -
O.01 -
0
CIS
0.14 -
01S -
0.1* -
O.11 -
O.10
o.o*
OM -
0.07 -
O.D«
o.oe
0.04 -
o.os
DOT -
0.01 '
0.00
UC* re -344) 0.4*
300
PAN Trace
r--»r««n./T>JO NO/0.0* NO* MC/NO>-18jO
700
Tim.
!
300
FIGURE 6-50. Simulation results for UCR experiment EC-346
(symbols are experimental measurements)
-------
NO, N02 ond 03 Troce
uc* tc-J«s o 4» rr~~ m-»r<«r»/t>.j? NO/O.M NOT MC/WO..I j*
o.s •
o.i -
Xylene Trace
tc-»4s e.« pen* i»-«rtaM/v.it MO/O.M NO*
—I—
>00
on
014
on -
0.1? -
011-
0 1
0.04
O09
0.07
oot
DM -
0.04 -
00] -
005
0.01 -
0
Formaldehyde Trace
uc* re-943 0.4* r»"~ --«r<«-»^o^' MO/OO* HO]
PAN Trace
UC» K-S49 04B
NO/O.M NM MC/WO.-1J.B
FIGURE 6-51. Simulation results for UCR experiment EC-345
(symbols are experimental measurements)
-------
NO. N02 end 03 Trace
OC» tt-M4 O.M **~ .»-%«••••/».« NO/0.19 MM MC/*0.-a.»
Formaldehyde Trace
UC» IC-S44 0.4*
MO/D.I S MO* MC/WO..B.t
OJ
o.s -
o J -
Xylene Trace
M« o.4» p^om-iVOTw/a.tt MO/O.IS MM MC/NO«»«.»
UC* CC-S44 0.4* |
PAN Trace
NO/D.18 NOt MC/MO»-8.t
o -\ r-
0
—I—
too
no
Tim.
300
FIGURE 6-52. Simulation results for UCR experiment EC-344
(symbols are experimental measurements)
-------
NO. N02 (plus PAN) ond 03 Troces
so. IMO (•) o.n n*~- • -•Twwo.<« HO/DA} NOJ
Km (mkw)
NO, N02 (plus PAN) ond 03 Troces
NO^J.OB KOI MC/HO..TD
DM
o to -
e.to
Die-
O.TO
NO, N02 (plus PAN) and 03 Troces
77. ItS? (*) O.n ppmy •-Ny««w/t>31 NO/0.17 HOI K/WQ.-J «
Tim* (f*iln«)
NO, N02 (plus PAN) ond 03 Troces
77. '««? (•) 0.3» pt^~ .-»T«— «•"> » NO/B.1* MOT HC/HO.-7.0
FIGURE 6-53. Simulation results for UNC experiments JL3080B, AU2782R,
AUO183B, and OC2782B (symbols are experimental measurements)
-------
91
B.40r-
CL
a.
B.30H
o
»-
c
•- B.20h
o
CJ
B.J0H
B.0C
B 10e 200 3Z2 «ee 50C 600 70D BZ0
TIME 1M1NJTES)
FIGURE 6-54. Simulation results from the CALL
(Carter et al., 1986) and earlier CBMX (Whitten
et al., 1985) mechanisms from UNCJL3U80R.
-------
NO, NO2 (plus PAN) and 03 Traces
formaldehyde
OM
O J -
O.M -
I o.M
9.1 •
MMUTt*
Isoprene
jMtr 14,
o J •
FIGURE 6-55. Simulation results for UNC experiment JL1480B
(symbols are experimental measurements)
-------
AJ*
NO, N02 (plus PAN) and 03 Traces
' 14. 11
O.T-
OJ-
0.1 -
Formaldehyde
Mr'«. it
too
PAN
' 14. 1
too
FIGURE 6-56. Simulation results for UNC experiment JL1480R
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) and 03 Traces
Mr '•• IMB-Ota*
Formaldehyde
r\>
^j
oo
04
O.T -
04 -
0,i
0.4
OJ -
03 -
0.1 -
0
Isoprene
10. ioo>-*i»
PAN
e or -
04S -
04* •
041 -
0
MO
FIGURE 6-57. Simulation results for UNC experiment JL1680B
(symbols are experimental measurements)
-------
NO. N02 (plus PAN) and 03 Traces
M. 11
r\j
Formaldehyde
Isoprene
M* '•• »«
PAN
**«.
tan -
O.M
OJM -
A01 -
FIGURE 6-58. Simulation results for UNC experiment JL1680R
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) ond 03 Troces
ro
CD
o
Formaldehyde
Mr ". IMO-WM
0.4 •
0.1 •
Isoprene
Mr IT. IMO
OJ -
o.i -
MO
PAN
no
FIGURE 6-59. Simulation results for UNC experiment JL1780B
(symbols are experimental measurements)
-------
NO. N02 (plus PAN) and 03
MT '». '•
o.»4
o.tt
o.te
o.ii -
o.tt
P.I 4 -
o.t» -
0.10 -
0-0*
O.M -
0.04
em -
oao
Formaldehyde
.Mr 17. 1MO-M
o.te
o.it -
o.il
o.i» -
o.i« -
0.1S -
0.14 -
0.11 -
0.1 t -
O.11 -
O.1O -
O.M
0.00
O.OT -
00« -
oot>
O.04 -
o.os -
out -
0.01 -
0.00
Isoprene
Mr "• »•
—1—
too
FIGURE 6-60. Simulation results for UNC experiment JL1780R
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) and 03 Traces
JM IT. II
~ M»M y w« «« WMM M wwiff MM if w^ vwmf wltM
0.11
O.tO
O.I*
0.1 •
0.17
O.It
0.1*
OH
0.1]
0.1*
O.11
0.10
o.o*
0.07 •
eat
0 04
OJ>S
oat
001
Formaldehyde
JlOT* IT. 1HO~«b*
100
Isoprene
i \1. H
MO
FIGURE 6-61. Simulation results for UNC experiment JN1780B
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) and 03 Traces
JM »r. if
8.10
Formaldehyde
> 17, H
•.1
•xi a
Isoprene
km IT. It
FIGURE 6-62. Simulation results for UNC experiment JN1780R
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) and 03 Traces
ro
oo
4*
Isoprene
D. H
FIGURE 6-63. Simulation results for UNC experiment JN2080R
(symbols are experimental measurements)
-------
NO. N02 (plus PAN) and 03 Traces
1. It
0.4
e.M
OJ
O.M
OJ -I
O.It
ai -
Formaldehyde
AM H. II
Isoprene
OLlt-
O.I •
O.M-
O.M •
U4-
•M-
OJ-
O.10 •
O.10-
0.14-
0.1*-
0.1 •
OlM-
•.00-
FIGURE 6-64. Simulation results for UNC experiment JN2280B
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) and 03 Traces
Formaldehyde
. H
Isoprene
FIGURE 6-65. Simulation results for UNC experiment JN2280R
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) and O3 Traces
•.1 -
•LM-
Isoprene
JX»» U, 1MB ••
at*-
o.i«-
•.14-
ait-
•.1*-
•n-
•.1*
0.11 •
•.to
ator-
OOJ-
Formaldehyde
PAN
* a. IM»
FIGURE 6-66. Simulation results for UNC experiment JN2380B
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) and 03
Formaldehyde
i*j 16.
001 -
•HUTU
PAN
ojsno
OJ>1f
O-Oli
O.DIT
•AM -
•JDl -
D.O>i -
tax
•
FIGURE 6-67.
Simulation results for UNC experiment JL2580B
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) ond 03
OOT
5. 1
Formoldehyde
O.M -
too
FIGURE 6-68.
Simulation results for UNC experiment JL2580R
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) and 03
JOT. 16. 1WO-*-.
tOO 400
mttfta
Formaldehyde
Jun. 18. 1MO-CM.
too
•wuro
PAN
> 1». 1HO-
OJOJ -
OJCJ -
too
FIGURE 6-69.
Simulation results for UNC experiment JN1580B
(symbols are experimental measurements)
-------
0-03 -
NO, N02 (plus PAN) and 03
It. 1HO-M
HIPUTD
Formaldehyde
JHM It. II
•00
•mum
PAN
i It. 1MO-
FIGURE 6-70. Simulation results for UNC experiment JN1580R
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) and 03
18.
CIS
0.14 -
0.11 -
o.u -
011 -
O 10 -
osn -
OJ» -
O-CT7 -
not -
0-01 -
cm -
0.01 -
0.00
• IM/TTS
Formaldehyde
July 16. 1MD-*u»
too
FIGURE 6-71.
Simulation results for UNC experiment JL1580B
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) and 03
•IHLJTO
Formaldehyde
AM, 16. 1MO-*«d
•mjrts
FIGURE 6-72.
Simulation results for UNC experiment JL1580R
(symbols are experimental measurements)
-------
0-O«
O.M -
OJ7 -
0-0* -
NO, N02 (plus PAN) and 03
Jun. 14. IMP »i»»
Formaldehyde
June 14. I
FIGURE 6-73. Simulation results for UNC experiment JN1480B
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) end 03
0.14
100 400
uuvm
Formaldehyde
Jm 14. 1MO-*>d
B.M -
OJO -
O.14 -
t a. 12 -
O.10 -
O.M -
oat -
*OO «00
•mi/ro
FIGURE 6-74. Simulation results for UNC experiment JN1480R
(symbols are experimental measurements)
-------
011 •
e.io
O.O*
OM
0.07
0.02 -
O-01 -
0.00
NO, N02 (plus PAN) ond 03
. 21. 1taO-B>ri
MINJTB
Formoldehdye
aoo
FIGURE 6-75. Simulation results for UNC experiment JN2180R
(symbols are experimental measurements)
-------
NO. N02 (plus PAN) ond 03 Troces
. ItH (») (.212 MMHC/D.2M NO/O.07D N07 HCXNO. .
*OO
Formoldehyde Trace
. 1»*4 (•) 1.212 MIMC/D.I64 NO/DJ170 M07 MC/NOi - •»!
O.It
0.1 •
O.I 7
0.1«
0.1 S
0.14
0.1 S
0.12
O.11
O.1O
OLD?
O.04
Tin* <«*>•}
PAN Trace
2T 1»M <•) U1I HIMC/DJM W/OB7D H02 MCXMO> •
FIGURE 6-76.
Simulation results for UNC experiment AU2284R
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) ond O3 Troces
««*«•> W. 1M4 (I) 1.121 HHMC/D.2C WO/D.07D ND2 MC/-MO. . 1JC
100
400
Formoldehyde Trace
I (•) 1.121 «««MC/B.J«2 MO/B.O7D W02 HC/WO> - S.M
100
)
PAN Trace
HO/OATO HOI
e •(
e
too
• ("*•)
FIGURE 6-77.
Simulation results for UNC experiment AU2584B
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) and 03 Traces
1. 1M4 (t) I.MS NHHC/OJ45 MO/O.OW N02 tC/MCte . 10.74
O.7D -
O.SO -
0.10 -
0.18
0.14 -
0.11 -
0.1 Z -
O.11 -
O.10 -
0.0* -
DM -
0*7 -
OJM -
O-M -
044 -
O41 -
ojn -
0.01 -
OJX)
0.07
Formaldehyde Trace
• 1. 1M4 (•) JJ«S NHMC/0.148 HOXO.O9* N02 MC/NO« • 10.74
FIGURE 6-78.
too
PAN Trace
10.T4
Ibn. (•*»)
Simulation results for UNC experiment ST0184B
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) end 03 Troces
•^••f^.' 1. 1M4 (I) 0*23 MUMC/OJH MO/DJXO H02 MC/NO. • JJ4
tmu,:i:, I. 1M4
Formaldehyde Trace
(I) 0-CJ NMHC/T1.M1 HO/OUXO HOT
!>—<«*»)
PAN Trace
1. 1»*l (I) 0*!3 NIMC/OJ41 HO/Q4XO HD2 MC/MO. . 1J4
•00
FIGURE 6-79.
Simulation results for UNC experiment ST0284B
(symbols are experimental measurements)
-------
OJM -
O-07 -
oat -
O.01 -
04O
ejos
OAI3 -
FIGURE 6-80.
NO, N02 (plus PAN) and O3 Traces
Hum (mku)
Formaldehyde Trace
I. 1»M (»} 1.3M NMMC/0.2M MO/D^ae N02 MC/NO<
PAN Trace
I 1M4 (») 1J34 MHX/D.M4 W/VAM NO7 NC/K>» •
TUn> <•*!•}
Simulation results for UNC experiment ST0284R
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) ond 03 Troces
• J. 1M4 {») 1.106 HHHC/D.2M MO/D.1I* MO2 MC/NO. »I'4
ZOO 400
Htm (mfcw)
Formaldehyde Trace
• 1. 1M4 (•} 1.10C NMHC/VJ44 WO/0 10» HO7 MC/MO. . ;
O-OOH
OJ017 -
aoote -
aooii -
OJBI -
o^eo* -
ojooe -
cocci -
OJJODt -
OAXM -
O-OOD3 -
PAN Trace
S, 1*64 (•} 1.10« *»7 MOTO.-J.K
FIGURE 6-81. Simulation results for UNC experiment ST0384R
(symbols are experimental measurements)
-------
NO. N02 (plus PAN) end O3 Troces
4. 1»M <») US MUMC/O.aS MO/0.077 H02 MC/XO. . JJB
0.10
Formaldehyde Trace
4. 1M4 (I) 1.SS MUHC/OJ*] ICXO.O77 M» MC/Wlfc • SJB
too
Hfi» (mkv)
PAN Trace
4. 1M4 (•} 146 MHHC/O.I*! HO/O.OT7 N02
OJB4 .
043 -
FIGURE 6-82.
Simulation results for UNC experiment AU0484R
(symbols are experimental measurements)
-------
0.00
NO, N02 (plus PAN) ond 03 Troces
h«u* 6. 1M4 (•) 1J MKMC/D.27J X(VOJJ7« N07 MC/NO. . I T7
aoo
Formaldehyde Trace
N07 HC/N(h - S.72
sao
PAN Trace
<•} 1J NWHC/D275 MQ^JT* N02
0-02} -
OJtt -
0-O14 -
O-O1J -
0.0' -
OOM -
OJW -
e
*oo
FIGURE 6-83.
Simulation results for UNC experiment AU0584B
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) ond 03 Troces
«. 1»fcl <•} O.tJ HMHC/O.aO NO/0.077 N02 X/KO. > 2.70
OJB -
o.io
c.o» -
OJM -
047 -
O.M -
CM -
ej* -
04J -
O41 -
CJ>1 -
000
MO 4OO «00
Ihm (•*••}
Formaldehyde Trace
(•} 0.*3 MMMCXOJBO MO/D4I77 HOI MC/MIk - LTD
O-014
O.DM-
0-01* -
0*1 -
041 -
PAN Trace
I t. 1M4 <*) MS MMMC/O.JBO Ha/'OATT MOZ MC/M(k • 1 TO
O JOB -
0407 -
O.OM -
OJSM -
0401 -
040} -
O401 -
0
too
, (m*.)
FIGURE 6-84. Simulation results for UNC experiment AU0584R
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) ond 03 Troces
«. t»M (i) J.JO NKMC/D.»J NO/0.072 NOt MC/WO. - t JD
0-7.C
0-34
o,i:
O.ZO
O.18
0.1*
Formaldehyde Trace
t. 1M4 (I) J.JO NK«C/'l<.a3 NOXt.OT; MOi X/XQp - tJD
t ::;:
o.ia
O-J»
100
PAN Trace
i «. it»* (i) «jo NHMC/O.»] NO/04177 nor
. tM>
FIGURE 6-85.
Simulation results for UNC experiment AU0684B
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) and 03 Traces
(») X.tl HHHC/O.aj NO/0.071 HO2 MC/NO» • 4.47
TWfM (mtov)
Formaldehyde Trace
O.It •
e.it
O.IT •
O.It
O.It
O.I 4
0.11
0.12
O.11
O.1O
O.O»
0X3 -
0.07 -
OJ31 •
too
» t.47
AA7 -
O.M -
,(-*-}
FIGURE 6-86.
Simulation results for UNC experiment AU0684R
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) ond 03 Troces
7. 1»«4 (•} 1 J4 NKMCX0.3D2 KVO-OBJ MD7 K/*OM • I.k3
Yltw (mbt»)
0.11
0.11
0.10
O-Ot
O.M
0.0T
Formaldehyde Trace
7. 1tB4 <*} IJ4 NMMC/0.3D? MO/O.OB J MOJ HC/NO . 313
OJJ1 -
ox: -
OJJ1 -
OJ1S2
O-OJ -
HOT* ("»*!•}
PAN Trace
7. <••< (•} 1 J< HKHCXOJD?
eon -
exit -
Oil -
OJDI
O.OM -
O.OM -
OXD7 -
0
o
100
FIGURE 6-87.
Simulation results for UNC experiment AU0784R
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) ond 03 Troces
a. i»4 <•} (.7
No/o.ore HO? HC/NO» . ic.e
O.70 -
O.«0 -
O.M -
O.40 -
0.34
O.12
0-10
e.2a <
O.21
0.24
O.22 •
0.20
0.1* •
O.1*
0.14
0.12
0.10
200 400 *OC
Tim. (mkt.)
Formaldehyde Trace
t. 1M4 *••}
FIGURE 6-88.
Simulation results for UNC experiment AU0884B
(symbols are experimental measurements)
-------
OJO
NO, N02 (plus PAN) and O3 Traces
I. 1M4 (•) 1JD3 N»NC/O.S« NO/DJM7 N02 MCXNO. . 1.34
0.15
0.14
0.11
0.11
0.11
0 10
0.0*
O.OC
OJH
OJ03
Formaldehyde Trace
t. ItM (•} 1 JOS MIMC/O.M? MO/DJW7 N02 MC/WO. . J 35
too
O.021
aoi -
O.D1I -
O-O1T -
aoit -
0.013 -
O.Oi J -
t O-011 -
t 0.0
OJDt -
c.ooe -
O.OOT -
ojx* -
O-OW -
0.004 •
0.003 -
PAN Trace
t. 1**< (•) 1J03 MIHC/OJ02 NO/DJB7 MO? MCXMO. - IJS
FIGURE 6-89.
Simulation results for UNC experiment AU0984R
(symbols are experimental measurements)
-------
O.JO
NO, N02 (plus PAN) and 03 Traces
Q.1.1 4. 1*63 (•) *.1»0 MBHC/O.JI4 W/D.OS7 NO} MC/W. • S.I
too
•00
OJO
O.JB -
O.2t -
0.24 -
0.22 -
O.20 -
0.16 -
O.I* -
O.14 -
0.1J -
0.10-
041 -
Formaldehyde Trace
0 lit. 4. 1MJ (•} 2.1M MIHC/Q.214 W/DJ5T M» MC/MO. • 1.72
Urn* (•"••)
PAN Trace
OUT
FIGURE 6-9U. Simulation results for UNC experiment OC0483B
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) ond 03 Troces
7. IMS (I) 7..74S NHHC/0.3D1 NO/DJJB M02 NC/WO» - 1.10
Formaldehyde Trace
7. 1MJ (1) 1.746 MtHC/OJJl NO/V.OJB NO? tC/MO. • «.10
OJ3
OJO
O.M
OJ«
OJ4
Oil
OJO
O 1i
0.1 •
O.14
O.11
0.1O
OO2 -
O.O
PAN Trace
7. 1M5 (I) X.748 IMHC/OJ01 W/OJJB NW «C/MO> - S.10
0.1 -
,(-*.)
FIGURE 6-91.
Simulation results for UNC experiment OC0783B
(symbols are experimental measurements)
-------
0.11
e.io -
oat -
oat -
FIGURE 6-92.
NO. N02 (plus PAN) ond 03 Troces
«•*•' 7. 1MJ (*} Z.«*4 MIMC/0.9* NO/O.MO NO2 tC/WO. • 7
aoo
Tim (•»*>•)
Formaldehyde Trace
T. 1»«J (•) 2.*U MIHC/O.V1 MD/OXMO W» MC/NOM . ».»
MO
Tim (nikv)
PAN Trace
Oililii 7, 1»U <») r«44 MIHC/0.»1 W/D.MO HOI
Simulation results for UNC experiment OC0783R
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) and O3 Traces
0.1*
e.it
O.17
0.1*
o.it
OK
O.1J
Oil
O.11 <
010
O-O»
DAS -
OM -
aoi -
002 -
001 -
' «. 1»*< (•}
HOT* (••*«•}
Formaldehyde Trace
WK/OJO8 HO/O.1M
PAN Trace
r •. 1M< (•) UM4 MtMC/QJOB KJ/D.1M MR MC/Mta •
FIGURE 6-93.
Simulation results for UNC experiment ST0884B
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) ond 03 Troces
0.70
*».<.*.)
Formaldehyde Trace
FIGURE 6-94. Simulation results for UNC experiment ST0884R
(symbols are exoerlmental measurements^
-------
NO, N02 (plus PAN) ond 03 Traces
'•»*• IT. 1t»4 (•) 1J1» MimC/D.271 Na/V.070 N02 HC/NO» -
Formaldehyde Trace
17. 1M4 (•) U1» HMMC/DJ71 HO/D^TO NO! MCXMOn -
O.14 -
0.11 -
0.10-
PAN Trace
• IT. i «4 (•> uit
xo/o JTO HOI MC/NO> •
007-
0-OJ -
OJJ -
•<"*-)
FIGURE 6-95. Simulation results for UNC experiment ST1784R
(symbols are experimental measurements)
-------
OJ4
0-10
e.it
0.1*
0.14
O.12
O.1O
NO, N02 (plus PAN) ond 03 Troces
n. in* (•) MT» HMMC/o^n NO/BAM ttm MC/HO> - T.it
Formaldehyde Trace
nn
OJW -
PAN Trace
DAT
N0»
.(•*«•)
FIGURE 6-96.
Simulation results for UNC experiment ST2184B
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) and O3 Traces
U»7 MIHC/OJTt WQ/O1»» HOI MC/>O» - >.«
Formaldehyde Trace
><-*-)
FIGURE 6-97,
Simulation results for UNC experiment ST2184R
(symbols are experimental measurements)
-------
0.1 -
NO, N02 and 03
tC-SJI
100
Formaldehyde
CC-1S1
FIGURE 6-98. Simulation results for UCR experiment EC-231
(symbols are experimental measurements)
-------
NO, N02 and 03
CC-IH
FIGURE 6-99. Simulation results for UCR experiment EC-232
(symbols are experimental measurements)
-------
NO, N02 and 03
IC-ZJS
100
PAN
cc-tu
OM-
0X3-
D O
1*0
FIGURE 6-100. Simulation results for UCR experiment EC-233
(symbols are experimental measurements)
-------
NO, N02 and O3
CC-Z57
Formaldehyde
CC-BT
A *
FIGURE 6-101. Simulation results for UCR experiment EC-237
(symbols are experimental measurements)
-------
NO. N02 and 03
CC-ZM
•Mum
Formaldehyde
IC-XM
FIGURE 6-102. Simulation results for UCR experiment EC-238
(symbols are experimental measurements)
-------
NO, N02 and 03
CC-141
Formaldehyde
tC-141
0.1 • -
ait -
0.14 -
I 0-11 -
too
FIGURE 6-103. Simulation results for UCR experiment EC-241
(symbols are experimental measurements)
-------
NO, N02 and 03
IC-J4?
Formaldehyde
CC-M2
O.TO-
FIGURE 6-104. Simulation results for UCR experiment EC-242
(symbols are experimental measurements)
-------
NO, N02 and 03
OrONtEC-243
Formaldehyde
CC-J4J
O-»0 -
O JO -
0.70 -
MO -
040 -
O.40 -
0.10 -
0.10 -
0.10 -
oao
FIGURE 6-105. Simulation results for UCR experiment EC-243
(symbols are experimental measurements}
-------
NO, N02 and 03
K-Ut
1.10
1.10 -
1 JO -
o.to -
DAB -
O.TO -
OJ*0 -
040 -
0.40 -
0.10-
OJO
0.10 -
OJOO
Formaldehyde
1C-MS
lUHAU
FIGURE 6-106. Simulation results for UCR experiment EC-245
(symbols are experimental measurements)
-------
NO, N02 and 03
CC-144
•mum
Formaldehyde
oil
0.1*
0.11
O.10
FIGURE 6-107. Simulation results for UCR experiment EC-246
(symbols are experimental measurements^
-------
NO, N02 and 03
CC-I47
100
WMJTD
Formaldehyde
IC-M7
PAN
ie-147
FIGURE 6-108. Simulation results for UCR experiment EC-247
'symbols are experimental measurements)
-------
PAN
FIGURE 6-109.
Simulation results for UCR experiment ITC-630
(symbols are experimental measurements)
-------
0.40
OJO
FIGURE 6-110.
Simulation results for UCR experiment ITC-631
(symbols are experimental measurements)
-------
CIS -
NO
ITC-US
N02
ITC-4U
1.1 14 1
(TftOWM
UINUTB
PAN
It J.I
S.I
FIGURE 6-111. Simulation results for UCR experiment ITC-633
(symbols are experimental measurements)
-------
0.08 •
03
ire-us
O.O4 •
0.03-
o.o» •
rv
O.4 OJ
1.1 1.« 1
JBSS"*1
t.4
FIGURE 6-112. Simulation results for UCR experiment ITC-635
(symbols are experimental measurements)
-------
NO
rrc-ur
FIGURE 6-113.
Simulation results for UCR experiment ITC-637
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) ond O3 Troces
•mniK/0.1* NO/DM NO? HC/W3X-* J
PAN Trace
• I. 1M2 (!) -•
•UmillyV.1l MIVO«» M02
0 JOT -
OJM -
FIGURE 6-114. Simulation results for UNC experiment OC0382B
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) ond 03 Traces
'(-*-)
PAN Trace
• *. INI (•} -N in rr~~ ••uix/o.i*
FIGURE 6-115.
Simulation results for UNC experiment OC0382R
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) ond 03 Troces
•00
Formaldehyde Trace
FIGURE 6-116.
Simulation results for UNC experiment AU3181B
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) ond 03 Troces
HO/Q.M MO! MC/MOl-1.6
•00
Formaldehyde Trace
t MD/aoe NOZ Mc/nox-e.s
200
FIGURE 6-117.
Simulation results for UNC experiment AU3181R
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) ond 03 Troces
+*, a. it*i (•) i*J rf* UNc*ix/fc*4 » i •rru'B.n HO/OJM HOI MC/MO>>IO.<
Formaldehyde Trace
Urn (•*<•}
PAN Trace
FIGURE 6-118. Simulation results for UNC experiment JL2281B
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) ond 03 Troces
*r a. iw< (*} 2.40 rf— uMCnx/?.ti MVOA «m
Formaldehyde Trace
** 22. 1M1 (•) l.« H»™ UMCHK/B.H NO/O.O6 N02 HC/NOK-*.!
sao
FIGURE 6-119. Simulation results for UNC experiment JL2281R
(symbols are experimental measurements)
-------
040
NO, N02 (plus PAN) and 03 Traces
AM M. 1M1 <•} 144 H»~> IMCWIK/0.13 NO/DA* N02 MC/NOI.tfl
Formaldehyde Trace
i 16. 1M1 <•) 14* •*•«• IMCHIKXO.il MO/OiM M02 HC/M»«t«
007-
soo
•00
OJSt
M. 1t*1 <•) 14* p^iiH IMCMK/Q.1I ND/0
.<•*->
FIGURE 6-120.
Simulation results for UNC experiment JN2581B
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) and 03 Traces
*•» M. «M1 (I) 1.M fr~> UNCMB/0.1J «D/BJ« IC1 MC/IC«-«.t
Formaldehyde Trace
*} 1.M rr~* \ffMttyOM HO/DA* MJJ
MC/NOK-*.*
0-X*
0-10
e.it
o.i«
0.14
0.12
0.10
o-ot -
OJH
SOD
PAN Trace
M. 1M1 (•}
UHCHIVO.U MO/OA* MOJ
FIGURE 6-121,
Simulation results for UNC experiment JN2581R
(symbols are experimental
-------
NO, N02 (plus PAN) ond 03 Troces
UMCHX/V.lf MO/OJU M07 MC/NOX-1**
*••»<.*.)
Formaldehyde Trace
f
O.1S
0.14 -
0.11 -
0.1* -
O.11 -
0.10-
oo* -
O-0» -
OO7 -
OJOt -
oat -
• 7. 1«2 (•)
UMCKX/B.1* MO/OAI W»
0411 -
><•*•>
FIGURE 6-122.
Simulation results for UNC experiment DE0782B
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) ond 03 Troces
• 7. 1M2 (») 1.47 p»m. UNCMVD.M pyn«CO 1.47 f*~m UMCHK^D.K) ffi iCOvarasQ.lt HO/DA) NO7 MCXNO>>
otn
as*
OJJ3 -
OJM -
OJO1 -
PAN Trace
FIGURE 6-123.
Simulation results for UNC experiment DE0782R
(symbols are experimental measurements)
-------
NO. N02 (plus PAN) and 03 Traces
• It. 1M2 (1) -M UO ff.
Formaldehyde Trace
. 1W2 <•} -• *JO f^>M iMCMn/e.M fpm. MXt/OJO NO/0.12 N02
O.1
OJ>» -
AOT-
DM (•«*»>
PAN Trace
, 1M2 <•) -H UO ppn« UNCHIX/O.M f^m. tHOT/OJO NO/0.12 M02 MC/M
OAJ-
100
FIGURE 6-124.
Simulation results for UNC experiment SE1682B
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) ond 03 Traces
• It. 1M? (•} -• tJB fpm. 1MCMIX/O.M rv~- Ofc^B.M NQ/0.11 WJ MC/
Formaldehyde Trace
PAN Trace
tlfc/D.IO NO/1.12 MOI MC/
too
FIGURE 6-125. Simulation results for UNC experiment SE1682R
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) ond 03 Troces
4 HO/O.OS MOI MC/M»«1».«
11. IMS (t) tJK ppix* UNCIIIX/0.04
•JO
0.1»
0.11
0.17
O.1«
0.18
0.14
0.11
0.1*
0.11
O.1O
OJD?-
• 11. 1M2 (•}
Urn <"*•)
Formaldehyde Trace
• 11. 1M2 (I)
PAN Trace
I ppra IMCMK/OM »i«k/B.14 HO/IUU MK
FIGURE 6-126.
Simulation results for UNC experiment NV1182B
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) ond 03 Troces
11. \HS (V) I
OLIO
O-1t
O.11
0.17
O.1*
O.16
e.i4
O.11
0.1»
O.11
O.10
007 -
O-O1 -
O-O1 -
Formaldehyde Trace
11. 1M2 (•) t-» ppm« IMCMK/O>1 T»«K/0.14 HO/UA3 MM MC/NO«»1».T
• 11. 1*«2 <(}
PAN Trace
UNCHIVO^I Ti>«c/D.14 MO^OJB N02 HC/MO>«1«.T
TWR. (•*•)
FIGURE 6-127.
Simulation results for UNC experiment NV1182R
(symbols are experimental measurements)
-------
NO. N02 (plus PAN) ond 03 Troces
MXT>UM NO/0.10 M03 MC/N
.<»*.)
Formaldehyde Trace
'•. 1M3 (I) -H *-» ppm* IMCMIX/OJ* W>M KXT/B^* HQ/O.la KM MC/N
Ihm (•*»)
PAN Trace
. 1W2 (•} -M
i MXT/DLM NO/O.IO KM HC/M
007 -
O.OJ-
FIGURE 6-128.
Simulation results for UNC experiment SE1882B
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) ond 03 Troces
Tlf*v
Formaldehyde Trace
• IS. 1M2 (»} -U J.1» rpn* UNCMX/V.tt p^n> On/O3t MD/D.10 H07 MC/N
PAN Trace
fc^l.ifc. IS. 1>« (W) -> t.1»^ntUNCMIO^J1 p*n« err/OJS MD/D.10 N02 HC/M
FIGURE 6-129.
Simulation results for UNC experiment SE1882R
(symbols are experimental measurements)
-------
NO. N02 (plus PAN) and OJ Traces
NO. N02 (plus PAN) and 03 Traces
NO. N02 (plus PAN) and 03 Traces
NO. N02 (plus PAN) and 03 Traces
NO. N02 (plus PAN) and 03 Tracts
NO. N02 (plus PAN) ond 03 Troces
FIGURE 6-130. Simulation results for both sides of UNC experiments
SE1481, SE2081, and SE2981 (symbols are experimental measurements)
-------
NO. N02 (plus PAN) ond O5 Troces
NO. N02 (plus PAN) ond O3 Troces
NO. N02 (plus PAN) and 03 Troces
NO. N02 (plus PAN) ond 03 Troces
FIGURE 6-131. Simulation results for both sides of UNC experiments
SE0381 and SE1081 (symbols are experimental measurements)
-------
NO, N02 (plus PAN) ond 03 Traces
1M1 (I)
Formaldehyde Trace
1J7 ff^m MHHIVO.M ff** tHU*O/0.1t MO/O1M6 HOI MC/NOX
FIGURE 6-132.
Simulation results for UNC experiment AU2681B
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) ond 03 Troces
M, it* (I) IM ppm. IMCHIX/Q.M IP*. *w»o/o n «c/ajei Mn MC/MOX
• (•*•>
Formaldehyde Trace
SHU*0/D.<( MQ/OJ83 MO2 tC/KOX
Ibn. (»>*•}
FIGURE 6-133. Simulation results for UNC experiment AU2681R
(symbols are experimental measurements)
-------
040
NO, N02 (plus PAN) ond 03 Troces
m*+~ It. 1«M (•> Ml ff*~ HOMtfW/D.n NO/B-Ot MM MC/WOM.a.0
0.10 -
0.10
Tb~(.*.>
Formaldehyde Trace
.a NO/DA* N02 MC/NOK^.O
O.1*
O.14 -
0.11-
O.1I -
0-11 -
0.10 -
007 -
0.03 -
«(-*-)
FIGURE 6-134.
Simulation results for UNC experiment SE1984B
(symbols are experimental measurements)
-------
NO, N02 (plus PAN) and 03 Traces
I (•>*<•)
Formaldehyde Trace
NO/OU» MQ3 MCXMOX.1J*
PAN Trace
(•) «JB ff~*t MOHKW/-O.Z8 NO/M» MO2 MCXWO«.1J»
FIGURE 6-135.
Simulation results for UNC experiment SE1984R
(symbols are experimental measurements)
-------
FORM and ALD2
UO
1.10
tec
OJO-
OJO
UD
&40
UD
UD
•.10
UO
UD UO
IJD
ETHYHNE and OLEFIN
12)
1.10
•JO
(LTD
UD
UD
0.10
UD MO UO
0 IWIIDC »
1JO IJD
FIGURE 6-136. Scatter diagrams comparing predicted maximum ozone
concentration versus measured values for aldehyde (top) and alkene
(bottom) smog chamber experiments.
-------
TOLUENE and XYLENE
uo
1.10
uo
•.TO-
UD
UD
•.10
OJD
OJO OJD
IJO 1JD
I JO
UO
+\
\ "
B.TO
UD
UD
«L10
ISOPRENC and a-PINENE
UO UO MO UO . OJD
IJD
FIGURE 6-137. Scatter diagrams comparing predicted maximum ozone
concentration versus measured values for. aromatic (top) and blogenic
hydrocarbon (bottom) smog chamber experiments.
-------
VARIOUS MIXTURES
uo
1.10-
IJDO
UP
OJD
0.70
UO
OJO-
we
uo
uo
0.10-
tOD
tt
OJOO
OJD OJO OJD
« K » ire
OJD
I JO
1JD
ALL EXPERIMENTS
1JO
1.10-
!»
OJO,
OJD
an
OJO
030-
OJO
OJO
OJD
0.10
OJO
OJOO
o tu>
OJO
•«
« MO
OJO
* M
1JO 1JD
X MS
FIGURE 6-138. Scatter diagrams comparing predicted maximum ozone
concentration versus measured values for reactive hydrocarbon mixture
experiments (top) and all simulated experiments (bottom).
359
-------
ALD2
I JO
OLEFINS
LBD
1JD
FIGURE 6-139. Scatter diagrams comparing predicted maximum formaldehyde
concentration versus measured values for aldehyde (top) and alkene
(bottom) smog chamber experiments.
360
-------
TOLUENE and XYLENE
040-
110
•A)
•JO
o
OJD
ISOPRENE and a-PINENE
IJDO
FIGURE 6-140. Scatter diagrams comparing predicted maximum formaldehyde
concentration versus measured values for aromatic (top) and blogenic
hydrocarbon (bottom) smog chamber experiments.
-------
VARIOUS MIXTURES
I JO
1.10-
1.00-
•JO
030
0.40
OJO-
OJO
CLIO
ann
CJD
•« R
OJO
« nc
CJD
KAC
I JO IJD
x tno
ALL EXPERIMENTS
1JO
t.io
tie
OJC
OJD
0.70
OJO
AMI
OJD
OJo
0.10
OJO tJD OJO OJD IJB 1JD
•* M « MO * ID X IX
FIGURE 6-141. Scatter diagrams comparing predicted maximum formaldehyde
concentration versus measured values for reactive hydrocarbon mixture
experiments (top) and all simulated experiments (bottom).
-------
TABLE 6-1. Initial conditions for UNC formaldehyde smog chamber simulations.
Initial
Concentrations (ppm)
Gas Phase
Experiment
Number
OC0984B
OC0984R
AU0179B
AU0279B
AU0479B
AU0579B
AU1280R
NO
0.352
0.353
0.277
0.157
0.170
0.407
0.371
N02
0.148
0.210
0.076
0.060
0.070
0.132
0.110
MONO
0.000
0.000
0.000
0.002
0.003
0.000
0.015
HCHO
0.962
0.962
1.000
1.010
0.504
1.209
0.841
CO
0.30
50.00
0.30
0.30
0.41
0.33
0.10
Halls
HCHO
0.000
0.000
0.040
0.040
0.040
0.000
0.040
^HONO-
O.OOE-0
O.OOE-0
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
Other Conditions
Water Vapor (ppm)
Initial
8000
8000
20000
21000
20500
29000
29000
Final
10500
10500
20000
21000
21000
40000
38000
Temperature (K)
Initial
285.8
285.8
296.0
296.0
294.3
295.1
294.6
Final
297.0
297.0
311.7
311.7
307.9
310.8
302.6
Light
Conditions
Good
Good
Good
Good
Good
Good
OK
-------
TABLE 6-2. Results of UNC formaldehyde simulations.
Initial Conditions
Experiment
Number
OC0984B
OC0984R
AU0179B
AU0279B
AU0479B
AU0579B
AU1280R
N0y
0.500
0.563
0.353
0.217
0.240
0.539
0.481
N02/NOX
0.298
0.373
0.215
0.276
0.292
0.245
0.229
HC/NOV
1.9
1.7
2.8
4.7
2.1
2.2
1.8
Maximum Ozone (ppm)
Calc.
(ppm)
0.305
0.988
0.521
0.619
0.255
0.470
0.233
Meas.
(ppm)
0.248
0.666
0.618
0.606
0.378
0.507
0.390
Abs.
Diff.
0.057
0.322
-0.097
0.013
-0.123
-0.037
-0.157
Pet.
Diff.
23.0%
48.3%
-15.7%
2.1%
-32.5%
- 7.3%
-40.3%
UNC Average
Std. Deviation
-0.003 - 3.2%
±0.162 ±31.0%
Experiment
Number
OC0984B
OC0984R
AU0179B
AU0279B
AU0479B
AU0579B
AU1280R
Maximum PAN (ppm)
Calc. Meas. Abs. Pet.
(ppm) (ppm) Diff. Diff.
0.004
0.005
0.008
0.009
_ _ _ _
- _ _ _
0.016
Calc.
(ppm)
1.04
50.00
1.16
1.18
0.83
1.31
0.80
Maximum
Meas.
(ppm)
50.00
1.17
1.15
0.92
1.35
0.74
CO (ppm)
Abs.
Diff.
_
-0.01
0.03
-0.09
-0.04
0.06
Pet.
Diff.
_
- 1%
3%
- 10%
- 3%
8%
UNC Average
Std. Deviation
-0.01 - 1%
±0.06 ± 7%
-------
TABLE 6-3. Initial conditions for acetldehyde and proplonaldehyde smog chamber simulations.
Initial
Concentrations (ppm)
Gas Phase
Experiment
Number
NO NO,
RCHO
MONO
HCHO
CO
Walls
HCHO
UNC (acetladehyde):
JN1482R
AU2482B
AU0880R
DE2677B
NV1977B
NV2077B
0.225 0.087
0.230 0.089
0.317 0.104
0.290 0.117
0.393 0.119
0.775 0.170
1.540
0.950
0.950
1.908
1.940
1.930
0.015
0.000
0.020
0.005
0.000
0.000
0.020
0.000
0.014
0.010
0.020
0.020
0.36
0.23
0.00
0.25
0.25
0.25
0.040
0.040
0.000
0.000
0.000
0.000
J
-------
TABLE 6-4. Results of acetaldehyde and propionaldehyde simulations.
Initial Conditions
Experiment
Number
NOV
UNC (acetaldehyde):
JN1482R
AU2482B
AU0880R
DE2677B
NV1977B
NV2077B
0.312
0.319
0.421
0.407
0.512
0.945
N02/NOX
0.279
0.279
0.247
0.287
0.232
0.180
HC/NOV
9.9
6.0
2.2
4.7
3.8
2.0
UNC Average
Std. Deviation
Maximum Ozone (ppm)
Calc.
(ppm)
0.828
0.912
1.050
0.044
0.193
0.068
Meas.
(ppm)
0.731
0.972
1.033
0.039
0.187
0.063
Abs.
Diff.
0.097
-0.060
0.017
0.005
0.006
0.005
0.012
±0.050
Pet.
Diff.
13.3%
- 6.2%
1.6%
12.8%
3.2%
7.9%
5.4%
±7.4%
UCR-EC (acetaldehyde):
EC-253
EC-254
0.000
0.112
0.241
4.5
0.132 0.135 -0.003 - 2.2%
0.270 0.268 0.002 0.7%
UNC (propionaldehyde):
JN1482B
AU2482R
10.2
6.2
0.854
0.902
0.733
0.941
0.121
-0.039
16.5%
- 4.1%
ALD2 Average
Std. Deviation
0.015 4.4%
±0.055 ± 7.9%
(continued)
-------
TABLE 6-4. (concluded)
Maximum PAN or PPN (ppm) Maximum HCHO (ppm)
Experiment Calc. Meas. Abs. Pet. Calc. Meas. Abs. Pet.
Number (ppm) (ppm) Diff. Diff. (ppm) (ppm) Diff. Diff.
UNC (acetaldehyde):
JN1482R
AU2482B
AU0880R
DE2677B
NV1977B
NV2077B
0.253
0.184
0.163
0.143
0.235
0.195
0.226
0.174
0.240
0.142
0.265
0.198
0.027
0.010
-0.077
0.001
-0.030
0.003
12%
6%
- 32%
1%
- 11%
- 2%
0.30 0.15 0.15 100%
0.19 0.19 0.00 0%
0.21 0.38 -0.17 - 45%
UCR-EC (acetaldehyde):
UNC Average -0.011 - 4% -0.01 18%
Std. Deviation ±0.037 ± 16% ±0.16 ± 74%
EC-253
EC-254
0.041 0.044
0.088 0.073
-0.003
0.015
- 7%
21%
-
-
UNC (propionaldehyde):
JN1482B
AU2482R
0.215
0.144 0.179
ALD2 Average
Std. Deviation
-0.035
-0.009
±0.032
- 20%
- 4%
± 16%
0.20 0.09 0.11
0.15 0.11 0.04
0.03
±0.12
122%
36%
43%
± 69%
-------
TABLE 6-b. Initial conditions for UNC ethene smoy chamber simulations.
oo
Initial
Concentrations (ppm)
Gas Phase
Experiment
Number
AU0479R
OC0584B
OC0584R
AU1078B
AU1U78R
AU2378B
AU2378R
JL088UR
JLU986B
JL0986R
JL1386B
NO
U.167
U.283
U.266
U.408
0.415
0.420
0.410
0.374
0.311
0.294
0.243
NOp
0.060
0.087
0.099
0.114
0.113
0.100
0.110
0.380
0.021
0.024
0.047
ETH
0.438
0.902
1.562
1.505
0.520
1.460
1.470
1.050
0.936
0.444
0.555
HONO
0.025
0.000
0.000
0.008
0.025
0.001
0.001
0.002
0.000
0.000
0.000
HCHO
0.020
0.000
0.000
0.020
0.020
0.020
0.020
0.020
0.000
0.000
0.000
CO
0.300
0.300
0.300
0.300
0.300
0.350
0.3bO
0.300
0.500
0.500
0.200
Walls
HCHO
0.040
0.040
0.040
0.040
0.040
0.040
0.040
0.000
0.040
0.000
0.040
kHONO
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
Other Conditions
Water Vapor (ppm)
Initial
20500
9000
9000
21000
21000
14000
15000
13000
16800
16800
15000
Final
21200
11000
11000
21000
21000
14000
15000
22000
20600
20600
18000
Temperature (K)
Initial
294.3
284.1
284.1
294.3
294.3
285.5
285.5
291.2
295.7
295.7
294.2
Final
307.9
298.1
298.1
307.9
307.9
306.2
306.2
306.4
314.3
314.3
310.1
Light
Conditions
Good
Choppy
Choppy
Good
Good
Good
Good
Choppy
Good
Good
Good
-------
TABLE 6-6. Results of UNC ethene simulations.
Initial Conditions
Experiment
Number
AU0479R
OC0584B
OC0584R
AU1078B
AU1078R
AU2378B
AU2378R
JL0880R
JL0986B
JL0986R
JL1386B
Maximum Ozone (ppm)
Calc.
NOX N0o/N0x
0.227 0.264
0.370 0.235
0.365 0.271
0.522 0.218
0.528 0.214
0.520 0.192
0.520 0.212
0.754 0.504
0.332 0.063
0.318 0.075
0.290 0.162
_HC/NOV (ppm)
3
4
8
5
2
5
5
2
5
2
3
.9
.9
.6
.8
.0
.6
.7
.8
.6
.8
.8
0
0
1
1
0
1
1
0
1
0
0
.687
.674
.129
.293
.183
.237
.218
.576
.098
.586
.719
Meas. Abs.
£p_pm) Diff.
0.
0.
0.
1.
0.
1.
1.
0.
0.
0.
0.
UNC Average
Std.
Deviation
Maximum
Experiment
Number
AU0479R
OC0584B
OC0584R
AU1078B
AU1078R
AU2378B
AU2378R
JL0880R
JL0986B
JL0986R
JL1386B
UNC Average
Calc.
(ppm)
0.18
0.36
0.67
0.63
0.15
0.60
0.60
0.37
0.38
0.16
0.20
Meas.
(ppm)
0.31
0.34
0.55
_
-
-
-
_
0.57
0.29
0.36
729 -0
675 -0
856 0
100 0
180 0
133 0
109 0
629 -0
975 0
679 -0
758 -0
0
±0
.042
.001
.273
.193
.003
.104
.109
.053
.123
.093
.039
.052
.116
Pet.
Diff.
- 5
- 0
31
17
1
9
9
- 8
12
-13
- 5
4
±13
.8%
.1%
.9%
.5%
.7%
.2%
.8%
.4%
.6%
.7%
.1%
.5%
.3%
HCHO (ppm)
Std. Deviation
Abs.
Diff.
-0.13
0.02
0.12
-
-
-
-
_
-0.19
-0.13
-0.16
-0.08
±0.12
Pet.
Diff.
- 42%
6%
22%
-
-
-
-
-
- 33%
- 45%
- 44%
-23%
±29%
369
-------
TABLE 6-7. Initial conditions for oleHn smoy chamber simulations.
Initial
Concentrations (ppm)
Gas Phase
Experiment
Number
ONC:
JL1386RJ
OC1184B1
OC1284B}
A01679R}
OC1278BJ
OC2578B}
SE2383B*
SE2383R^
SE2583B
SE2583R^
SE2783B2
SE2783R
OCR-EC:
EC-12l}
EC-177}
EC-2781
EC-123J?
EC-1242
NO
0.246
0.249
0.502
0.312
0.364
0.337
0.339
0.358
0.382
0.414
0.381
0.405
0.410
0.364
0.366
0.401
- 0.608
NO?
0.045
0.105
0.180
0.131
0.115
0.104
0.038
0.044
0.044
0.047
0.046
0.049
0.101
0.099
0.128
0.106
0.385
OLE
0.207
0.751
0.653
0.472
0.443
0.408
0.543
0.370
0.722
0.410
0.752
0.398
0.483
0.493
1.016
0.404
0.424
HONO
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.015
0.000
0.030
0.010
0.005
0.020
0.020
HCHO
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.010
0.008
0.000
0.000
CO
0.200
0.250
0.250
0.280
0.350
0.550
0.250
0.250
0.250
0.250
0.250
0.250
2.193
2.193
0.030
1.881
1.872
Walls
HCHO
0.000
0.000
0.000
0.040
0.040
0.040
0.000
0.000
0.000
0.000
0.040
0.000
0.000
0.000
0.000
0.000
0.000
^HONQ
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
O.OOE-0
O.OOE-0
O.OOE-0
2.50E-5
2.50E-5
2.50E-5
1.17E-4
1.17E-4
1.17E-4
1.17E-4
1.17E-4
Other Conditions
Water Vapor (ppm)
Initial
15000
11500
10500
20000
8500
8000
20000
20000
10500
10500
12500
11500
25480
33930
19000
26700
24400
Final
18000
12500
12000
20000
20500
17000
25000
26000
21000
19500
21500
20500
25480
33930
19000
26700
24400
Temperature (K)
Initial
294.2
289.9
285.0
287.1
278.6
271.9
278.1
278.1
277.6
277.6
281.5
281.5
303.7
303.7
302.5
302.5
302.5
Final
310.1
297.2
296.7
301.3
298.2
294.2
297.8
297.8
296.2
296.2
298.4
298.4
305.4
305.4
303.2
303.2
303.2
Light
Conditions
Good
Good
Choppy
Good
Good
OK
Good
Good
Good
Good
OK
OK
constant
constant
constant
constant
constant
propylene, ^ 1-butene, 3 trans-2-butene.
-------
TABLE 6-8. Results of olefin simulations.
Experiment
Number
UNC:
JL1386R}
OC1184B}
OC1284B}
AU1679R}
OC1278BJ
OC2578B1
SE2383B
1
SE2383R2
SE2583B2
SE2583R2
SE2783B;?
SE2783R2
UCR-EC:
EC-12lJ
EC-177J
EC-2782
EC-123?,
EC-1242
Initial Conditions
HC/NO
Maximum Ozone (ppm)
0.291
0.354
0.682
0.443
0.479
0.441
0.377
0.402
0.426
0.461
0.427
0.454
0.511
0.463
0.494
0.507
0.993
0.155
0.297
0.264
0.296
0.240
0.236
0.101
0.109
0.103
0.102
0.108
0.108
X
Calc.
Meas.
(ppm)
Abs.
Diff.
2.1
6.4
2.9
3.2
2.8
2.8
4.3
3.7
6.8
3.6
3.5
3.5
UNC Average
Std. Deviation
0.198 2.8 0.441 0.506
0.214 3.2 0.500 0.540
0.259 6.2 0.662 0.625
0.209 3.2 0.356 0.506
0.388 1.7 0.168 0.247
-0.065
-0.040
0.037
-0.150
-0.079
Pet
Diff
0.505
0.784
0.464
0.724
0.354
0.250
0.651
0.365
0.800
0.409
0.389
0.423
0.496
0.674
0.432
0.688
0.462
0.230
0.405
0.206
0.594
0.266
0.523
0.285
0.009
0.110
0.032
0.036
-0.108
0.020
0.246
0.159
0.206
0.143
-0.134
0.138
1.8%
16.3%
7.4%
5.2%
-23.4%
8.7%
60.7%
77.2%
34.7%
53.8%
-25.6%
48.4%
0.071 22.1%
±0.117 ±32.8
-12.8%
- 7.4%
5.9%
-29.6%
-32.0%
UCR-EC Average
Std. Deviation
-0.059 -15.2%
±0.068 ±15.8%
OLE Average
Std. Deviation
0.033 11.1%
±0.120 ±33.3%
(continued)
-------
TABLE 6-8. (concluded)
Experiment
Number
JL1386R
OC1184B
OC1278B
AU1679R
OC2578B
OC1284B
SE2383B
SE2383R
SE2583B
SE2583R
SE2783B
SE2783R
UCR-EC:
EC-12lJ
EC-177J
EC-2782,
EC-123^
EC-1242
Maximum
Calc. Meas.
(ppm) (ppm)
0.014 0.047
0.20 0.180
0.083 0.000
0.171 0.182
0.075 0.050
0.138 0.144
0.187
0.076
0.238
0.097
0.122
0.085
UNC Average
Std. Deviation
0.069 0.167
0.093 0.165
0.276
0.088 0.065
0.035 0.038
UCR-EC Average
Std. Deviation
OLE Average
Std. Deviation
PAN (ppm)
Abs.
Diff.
-0.033
0.040
0.083
-0.011
0.025
-0.006
_
_
_
_
_
-
0.016
±0.042
-0.098
-0.072
0.023
-0.003
-0.038
±0.057
-0.005
±0.053
Pet.
Diff.
- 7Q%
22%
100%
- 6%
50%
- 4%
_
-
_
_
_
-
15%
± 57%
- 59%
- 44%
35%
- 8%
- 19%
± 42%
2%
± 52%
Maximum HCHO (ppm)
Calc.
(ppm)
0.11
0.39
0.21
0.22
0.19
0.30
0.25
0.15
0.34
0.17
0.11
0.17
0.26
0.26
0.55
0.20
0.19
Meas. Abs.
(ppm) Diff.
0.20 -0.09
0.48 -0.09
- -
-
_ _
0.43 -0.13
_ _
_ _
_ _
_ _
_ _
- -
-0.07
±0.04
0.28 -0.02
0.09 0.17
0.45 0.10
0.18 0.02
0.16 0.03
0.06
±0.08
0.00
±0.10
Pet.
Diff.
- 45%
- 18%
-
-
-
- 30%
_
_
_
_
_
-
- 31%
± 14%
- 7%
189%
22%
11%
19%
47%
± 80%
18%
± 73%
Ipropylene, 2l-butene, 3trans-2-butene.
-------
TABLE 6-9. Initial conditions for toluene smog chamber simulations.
Initial
Concentrations (ppm)
Gas Phase
Experiment
Number
UNC:
JN2779B
JL3080R
JN2784B
AU1579B
AU0183R
OC2782R
AU2782B
UCR-EC:
EC-264
EC-265
EC-266
EC-269
EC-271
EC-273
EC-327 .
EC-340
NO
0.263
0.156
0.300
0.276
0.344
0.280
0.310
0.420
0.435
0.432
0.398
0.186
0.096
0.357
0.333
NO,
0.069
0.026
0.035
0.086
0.051
0.110
0.121
0.056
0.047
0.059
0.074
0.028
0.014
0.096
0.096
TOL
2.210
0.550
0.702
0.740
0.655
0.642
0.428
1.156
1.070
1.196
0.566
1.146
0.587
0.573
0.585
HONO
0.001
0.005
0.001
0.001
0.003
0.004
0.005
0.005
0.010
0.005
0.015
0.010
0.015
0.010
0.015
HCHO
0.010
0.010
0.010
0.010
0.010
0.002
0.015
0.008
0.012
0.008
0.003
0.004
0.002
0.013
0.005
CO
0.323
0.200
0.213
0.338
0.323
0.320
0.195
0.300
0.330
0.330
0.200
0.200
0.326
0.200
2.593
Walls
HCHO
0.040
0.040
0.040
0.040
0.040
0.020
0.040
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
^HONQ
5.00E-6
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.00E-5
2.50E-5
1.37E-4
1.37E-4
1.37E-4
1.37E-4
1.40E-4
1.37E-4
1.56E-4
1.56E-4
Other Conditions
Water Vapor (ppm)
Initial
21400
20600
20000
10000
21100
6500
17900
18390
22040
19280
18920
20070
20920
24840
25900
Final
38000
39800
38000
15000
23600
23800
18600
18390
22040
19280
18920
20070
20920
24840
25900
Temperature (K)
Initial
286.5
292.7
293.8
291.2
297.2
278.2
291.3
304.0
302.6
302.6
302.7
302.5
303.5
303.0
303.0
Final
298.1
308.5
311.8
304.4
310.5
290.9
306.0
304.0
302.6
302.6
302.7
302.5
303.5
303.0
303.0
Light
Conditions
Good
Good
Good
Choppy
Good
Good
Choppy
constant
constant
constant
constant
constant
constant
constant
constant
-------
TABLE 6-10. Results of toluene simulations.
Initial Conditions
Experiment
Number
UNC:
JN2779B
JL3080R
JN2784B
AU1579B
AU0183R
OC2782R
AU2782B
UCR-EC:
EC-273
EC-271
EC-266
EC-264
EC-265
EC-340
EC-327
EC-269
NOV
0.332
0.182
0.335
0.362
0.395
0.390
0.431
0.110
0.214
0.491
0.476
0.482
0.429
0.453
0.472
N02/N0
^—Cl " A
0.208
0.143
0.104
0.238
0.129
0.282
0.281
0.127
0.131
0.120
0.118
0.098
0.224
0.212
0.157
HC/NOV
46.6
21.2
14.7
14.3
11.6
11.4
7.0
UNC Average
Std. Deviation
37.6
37.5
17.1
17.0
15.5
9.5
8.9
8.4
UCR-EC Average
Std. Deviation
Maximum Ozone (ppm)
Calc.
(ppm)
0.456
0.297
0.415
0.390
0.411
0.102
0.136
0.231
0.313
0.382
0.381
0.371
0.318
0.309
0.281
Meas.
(ppm)
0.399
0.274
0.401
0.403
0.458
0.095
0.128
0.215
0.296
0.403
0.419
0.393
0.344
0.376
0.298
Abs.
Diff.
0.057
0.023
0.014
-0.013
-0.047
0.007
0.008
0.007
±0.032
0.016
0.017
-0.021
-0.038
-0.022
-0.026
-0.067
-0.017
-0.020
±0.027
Pet.
Diff.
14.3*
8.4%
3.5%
- 3.2%
-10.3%
7.4%
6.3%
3.8%
±8.1%
7.4%
5.7%
- 5.2%
- 9.1%
- 5.6%
- 7.6%
-17.8%
- 5.7%
-4.7%
±8.1%
Toluene Average
Std. Deviation
-0.007 0.6%
±0.032 ±9.0%
(continued)
374
-------
TABLE 6-10. (concluded)
Experiment
Number
UNC:
JN2779B
JL3080R
JN2784B
AU1579B
AU0183R
OC2782R
AU2782B
UCR-EC:
EC-273
EC-271
EC-266
EC-264
EC-265
EC-340
EC-327
EC-269
Maximum
Calc. Meas.
(ppm) (ppm)
0.079 0.076
0.023 0.037
0.024 0.013
0.033 0.057
0.022 0.044
0.023 0.013
0.005 0.012
UNC Average
Std. Deviation
0.022 0.032
0.040 0.053
0.064 0.075
0.058 0.071
0.062 0.072
0.033 0.042
0.032 0.041
0.030 0.050
UCR-EC Average
Std. Deviation
Toluene Average
Std. Deviation
PAN (ppm)
Abs.
Diff.
0.003
-0.014
0.011
-0.024
-0.022
0.010
-0.007
-0.006
±0.015
-0.010
-0.013
-0.011
-0.013
-0.010
-0.009
-0.009
-0.020
-0.012
±0.004
-0.009
±0.010
Pet.
Diff.
4%
- 38%
85%
- 42%
- 59%
77%
- 58%
- 4%
± 62%
- 31%
- 25%
- 15%
- 18%
- 14%
- 21%
- 22%
- 40%
- 23%
± 9%
- 14%
± 42%
Maximum HCHO (ppm)
Calc.
(ppm)
0.07
0.05
0.07
0.06
0.07
0.04
0.05
0.04
0.06
0.09
0.09
0.08
0.06
0.06
0.06
Meas.
(ppm)
0.08
0.09
0.08
0.06
0.07
0.01
0.01
0.02
0.04
0.08
0.03
0.04
0.05
0.06
0.04
Abs.
Diff.
-0.01
-0.04
-0.01
0.00
0.00
0.03
0.04
0.00
±0.03
0.02
0.02
0.01
0.06
0.04
0.01
0.00
0.02
0.02
±0.02
0.01
±0.02
Pet.
Diff.
- 13%
- 44%
- 13%
0%
0%
300%
400%
90%
±181%
100%
50%
13%
200%
100%
20%
0%
50%
67%
± 66%
78%
±128%
37F
-------
TABLE 6-11. Initial conditions for xylene smog chamber simulations.
Initial
Concentrations (ppm)
Gas Phase
Experiment
Number NO
UNC (m-xylene):
JN2784R 0.303
UNC (o-xylene):
JL3080B 0.155
AU0183B 0.323
OC2782B 0.282
AU2782R 0.314
UCR-EC (m-xylene):
EC-346 0.204
EC-345 0.220
EC-344 0.520
N02
0.040
0.023
0.049
0.117
0.120
0.059
0.059
0.154
XYL
0.249
0.279
0.328
0.347
0.249
0.494
0.480
0.494
HONO
0.001
0.005
0.003
0.004
0.005
0.015
0.015
0.015
HCHO
0.010
0.010
0.010
0.002
0.015
0.020
0.020
0.020
CO
0.213
0.200
0.323
0.320
0.195
0.798
5.206
1.858
Malls
HCHO
0.040
0.040
0.040
0.020
0.040
0.000
0.000
0.000
^HONO-
2.50E-5
2.50E-5
2.50E-5
2.00E-5
2.50E-5
1.53E-4
1.53E-4
1.57E-4
Other Conditions
Water Vapor (ppm)
Initial
20000
20600
21100
6500
17900
26530
25210
28990
Final
38000
39800
23600
23800
18600
26530
25210
28990
Temperature (K)
Initial
293.8
292.7
297.2
278.2
291.3
304.2
303.1
303.2
Final
311.8
308.5
310.5
290.9
306.0
304.2
303.1
303.2
Light
Conditions
Good
Good
Good
Good
Choppy
constant
constant
constant
-------
TABLE 6-12. Results of xylene simulations.
Initial
Experiment
Number
UNC (m-xylene);
JN2784R
UNC (o-xylene):
JL3080B
AU0183B
OC2782B
AU2782R
N
•
0.
0.
0.
0.
0.
x -
343
178
274
399
434
Conditions
N02/N0¥
0.
0.
0.
0.
0.
120
129
179
293
276
HC/NOW
5.
12.
7.
7.
4.
8
5
0
0
6
Maximum Ozone (ppm)
Calc.
0
0
0
0
0
.722
.396
.697
.425
.652
Meas.
(ppm)
0.675
0.555
0.686
0.395
0.491
Abs.
Diff.
0.047
-0.159
0.011
0.030
0.161
Pet.
Diff.
7.0%
-28.6%
1.6%
7.6%
32.8%
UNC Average
Std. Deviation
0.018 4.1%
±0.115 ±21.9%
UCR-EC (m-xylene):
EC-346
EC-345
EC-344
0.263 0.224 15.0 0.398
0.279 0.211 13.8 0.406
0.674 0.228 5.9 0.616
UCR-EC Average
Std. Deviation
Xylene Average
Std. Deviation
0.384 0.014
0.396 0.010
0.586 0.030
0.018
±0.011
0.018
±0.087
3.6%
2.5%
5.1%
3.7%
±1.3%
4.0%
±16.6%
(continued)
-------
TABLE 6-12. (concluded)
Maximum PAN (ppm)
Experimen
Number
UNC (m-xy
JN2784R
t Calc. Meas.
(ppm) (ppm)
lene):
0.082 0.086
Abs.
Diff.
-0.004
Pet.
Diff.
- 5%
Maximum HCHO (ppm)
Calc. Meas. Abs.
(ppm) (ppm) Diff.
0.07 0.06 0.01
Pet.
Diff.
17%
UNC (o-xylene):
JL3080B
AU0183B
OC2782B
AU2782R
0.073 0.091
0.112 0.102
0.158 0.112
0.107 0.093
UNC Average
Std. Deviation
-0.018
0.010
0.046
0.014
0.010
±0.024
- 20%
10%
41%
15%
8%
± 23%
0.10 0.20 -0.10
0.11
0.07 0.04 0.03
0.06 0.10 -0.04
-0.03
±0.06
- 50%
75%
- 40%
1%
± 58%
UCR-EC (m-xylene):
EC-346
EC-345
EC-344
0.102 0.102
0.108 0.107
0.213 0.175
UCR-EC Average
Std. Deviation
Xylene Average
Std. Deviation
0.000
0.001
0.038
0.013
±0.022
0.011
±0.022
0%
1%
22%
8%
± 12%
8%
± 19%
0.10 0.07 0.03
0.13 0.14 -0.01
0.11 0.11 0.00
0.01
±0.02
-0.01
±0.05
43%
- 7%
0%
12%
± 27%
5%
± 44%
-------
TABLE 6-13. Initial conditions for UNC Isoprene and a-p1nenc smog chamber simulations.
Initial
Concentrations (ppm)
Gas Phase
Experiment
Number
Isoprene:
JL1480B
JL1480R
JL1680B
JL1680R
JL1780B
JL1780R
JN1780B
JN1780R
JN2080R
JN2280B
JN2280R
JN2380B
a-P1nene:
JL2580B
JL2580R
JN1580B
JN1580R
JL1580B
JL1580R
JN1480B
JN1480R
JN2180R
NO
0.202
0.211
0.144
0.137
0.380
0.391
0.075
0.137
0.379
0.175
0.174
,0.173
0.198
0.207
0.177
0.169
0.157
0.157
0.182
0.182
0.394
NO,
0.041
0.044
0.042
0.043
0.080
0.075
0.021
0.052
0.098
0.030
0.029
0.038
0.049
0.040
0.042
0.044
0.031
0.030
0.030
0.028
0.105
HCHO
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
MONO
0.003
0.003
0.003
0.003
0.003
0.003
0.003
0.003
0.003
0.001
0.003
0.003
0.006
0.003
0.003
0.003
0.003
0.003
0.003
0.003
0.003
CO
0.413
0.413
0.393
0.330
0.360
0.388
0.360
0.388
0.413
0.413
0.413
0.413
0.388
0.388
0.388
0.388
0.388
0.388
0.388
0.388
0.388
NMHC
2.170
4.900
1.965
4.035
2.580
0.980
1.300
0.500
0.500
1.560
3.830
1.270
0.940
1.020
0.800
1.220
2.680
1.040
1.700
3.150
1.970
Walls
HCHO
0.020
0.020
0.020
0.020
0.020
0.020
0.020
0.020
0.020
0.020
0.020
0.020
0.020
0.020
0.020
0.020
0.020
0.020
0.020
0.020
0.020
kunun
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
Other Conditions
Water Vapor (ppm)
Initial
26000
26000
26000
26000
26000
26000
26000
26000
26000
26000
26000
26000
26000
26000
26000
26000
26000
26000
26000
26000
26000
Final
26000
26000
26000
26000
26000
26000
26000
26000
26000
26000
26000
26000
26000
26000
26000
26000
26000
26000
26000
26000
26000
Temperature (K)
Initial
292
292
293
293
294
294
291
291
287
284
284
286
291
291
288
288
290
290
283
283
286
Final
307
307
310
310
311
311
303
303
305
306
306
306
304
304
308
308
304
304
305
305
302
Light
Conditions
Good
Good
Good
Good
Good
Good
Choppy
Choppy
Choppy
Good
Good
Choppy
Choppy
Choppy
Choppy
Choppy
Good
Good
Good
Good
Choppy
*2.25 ppmC o-p1nene and 0.90 ppmC Isoprene.
-------
TABLE 6-14.
Results of UNC isoprene and a-pinene simulations.
Experiment
Number
Initial Conditions
HC/NO
y
Maximum Ozone (ppm)
Calc. Meas. Abs. Pet.
(ppm) (ppm) Diff. Diff.
Isoprene:
JL1480B
JL1480R
JL1680B
JL1680R
JL1780B
JL1780R
JN1780B
JN1780R
JN2080R
JN2280B
JN2280R
JN2380B
0.243 0.169 8.9
0.255 0.173 19.2
0.186 0.226 10.6
0.180 0.239 22.4
0.460 0.174 5.6
0.466 0.161 2.1
0.096 0.219 13.5
0.189 0.275 2.6
0.478 0.205 1.0
0.205 0.146 7.6
0.203 0.143 18.9
0.210 0.181 6.0
0.730
0.475
0.748
0.502
0.959
0.725
0.363
0.479
0.130
0.658
0.434
0.665
0.630
0.435
0.837
0.652
1.297
0.806
0.406
0.354
0.088
0.550
0.392
0.570
0.100
0.040
-0.089
-0.150
-0.338
-0.081
-0.043
0.125
0.042
0.108
0.042
0.094
15.9%
9.3%
-10.6%
-23.0%
-26.1%
-10-0%
-10.6%
35.3%
47.7%
19.6%
10.7%
16.5%
Isoprene Average
Std. Deviation
-0.013 6.3%
±0.136 ±22.8%
a-Pinene:
JL2580B
JL2580R
JN1580B
JN1580R
JL1580B
JL1580R
JN1480B
JN1480R
JN2180R
0.247
0.247
0.219
0.213
0.188
0.184
0.212
0.210
0.499
0.198
0.162
0.192
0.207
0.165
0.163
0.142
0.133
0.210
3.8
4.1
3.7
5.7
14.3
5.7
8.0
15.0
3.9
0.274
0.340
0.309
0.572
0.573
0.425
0.564
0.552
0.363
0.334
0.377
0.332
0.473
0.470
0.202
0.519
0.465
0.223
-0.060
-0.037
-0.023
0.099
0.103
0.223
0.045
0.087
0.140
-18.0%
- 9.8%
- 6.9%
20.9%
21.9%
110.4%
8.7%
18.7%
63.2%
a-Pinene Average
Std. Deviation
0.064 23.2%
±0.092 ±40.5%
Biogenic HC Average
Std. Deviation
0.020 13.5%
±0.123 ±31.9%
(continued)
-------
TABLE 6-14. (concluded)
Experiment
Number
Maximum PAN (ppm)
Calc. Meas.
(ppm)
Maximum HCHO (ppm)
Calc. Meas.
(ppm) (ppm)
Isoprene:
JL1480B
JL1480R
JL1680B
JL1680R
JL1780B
JL1780R
JN1780B
JN1780R
JN2080R
JN2280B
JN2280R
JN2380B
0.120
0.118
0.086
0.082
0.168
0.035
0.058
0.039
0.006
0.109
0.106
0.099
0.101
0.101
0.080
0.072
0.188
0.047
0.053
0.034
0.008
0.086
0.069
0.081
Isoprene Average
Std. Deviation
0.019
0.017
0.006
0.010
-0.020
-0.012
0.005
0.005
-0.002
0.023
0.037
0.018
0.009
±0.016
19%
17%
8%
14%
- 11%
- 26%
9%
15%
- 25%
27%
54%
22%
10%
± 22%
0.
0.
0.
0.
0.379
0.772
0.357
0.643
0.452
194
203
091
099
0.269
0.607
0.224
0.381
0.660
0.346
0.548
0.633
0.230
0.182
0.076
0.374
0.572
0.196
-0.002
0.112
0.011
0.095
-0.181
-0.036
0.021
0.015
-0.105
0.035
0.028
1%
17%
3%
17%
29%
16%
12%
20%
28%
6%
14%
-0.001 1%
±0.083 ± 18%
a-Pinene:
JL2580B
JL2580R
JN1580B
JN1580R
JL1580B
JL1580R
JN1480B
JN1480R
JN2180R
0.
0.
0.
0.
0.
0.
0.
0.
0.
a-Pinene
016
025
013
049
128
048
096
143
054
_
_
0.016
0.022
0.033
0.011
0.032
0.046
-
Average
Std. Deviation
Biogenic
HC Average
Std. Deviation
-0
0
0
0
0
0
0
±0
0
±0
_
_
.003
.027
.095
.037
.064
.097
-
.053
.040
.024
.033
_
_
- 19%
123%
288%
336%
200%
211%
-
190%
±126%
70%
±112%
0
0
0
0
0
0
0
0
0
.059
.068
.051 0.038
.073 0.066
.134 0.105
.058 0.054
.085
.236
.103
0
0
0
0
0
±0
0
±0
_
_
.013
.007
.029
.004
_
_
—
.013
.011
.003
.071
_
_
34%
11%
28%
7%
_
_
—
20%
± 13%
6%
± 18%
•5D1
-------
TABLE 6-15. Initial conditions for UNC SynUrban and SynAuto mixture simulations.
Experiment
Number
Initial Concentrations (ppm)
Gas Phase Walls
NMOC
NO NO^ (ppmC) HONO HCHO CO HCHO kHQNQ
Other Conditions
Water Vapor (ppm)
Initial Final
Temperature (K)
Initial Final
Light
Conditions
SynUrban:
SynAuto:
OC0483B
OC0783B
OC0783R
AU0484R
AUOS84B
AUU584R
AUU684B
AU0684R
AU0784R
AU0884B
AU0984R
SynutoUrban:
4
STU884B
ST0884R
ST1784R
ST2184B
ST2184R
0.254
0.262
0.245
0.262
0.284
0.244
0.214
0.301
0.291
0.293
0.273
0.280
0.283
0.283
0.302
0.264
0.302
0.206
0.220
0.271
0.273
0.275
0.070
0.070
0.059
0.090
0.058
0.109
0.037
0.038
0.040
0.077
0.076
0.077
0.072
0.071
0.083
0.078
0.087
0.126
0.119
0.070
0.086
0.088
3.04
1.07
3.31
0.76
1.10
1.10
2.20
2.70
2.70
1.23
1.32
0.91
3.24
2.25
1.32
3.68
1.28
2.79
1.84
2.14
2.42
2.43
0.001
0.005
0.002
0.005
0.005
0.005
0.001
0.000
0.000
0.030
0.025
0.025
0.015
0.018
0.030
0.010
0.035
0.004
0.002
0.002
0.001
0.002
0.056
0.020
0.055
0.012
0.018
0.019
0.010
0.000
0.000
0.019
0.019
0.020
0.057
0.038
0.041
0.057
0.019
0.054
0.030
0.036
0.184
0.010
0.300
0.300
0.300
0.300
0.300
0.300
20.30
0.230
0.21U
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.040
0.040
0.040
0.040
0.040
0.040
0.080
0.040
0.040
0.040
0.040
0.040
0.040
0.040
0.040
0.040
0.040
0.040
0.040
0.040
0.040
0.040
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
15000
20000
12000
33000
29000
33000
29000
27000
25000
26000
29000
7000
6000
20000
9000
7000
26000
30000
18000
37000
46000
48000
42000
43000
37000
30000
33000
10000
8000
20000
10000
7500
285.
286.
286.
296.
295.
295.
295.0
295.0
296.0
295.3
296.2
281.6
281.6
280.0
285.5
285.5
306.7
303.9
306.7
310.0
310.0
303.0
301.7
295.0
295.0
310.9
310.8
310.8
312.0
312.0
311.3
311.2
313.7
303.7
303.7
296.2
305.6
305.6
Good
Choppy
Good
Good
Good
Choppy
OK
Good
Good
Late Chop
Good
Good
Good
Good
Good
Late Chop
OK
Good
Good
Choppy
Good
Good
-------
TABLE 6-16. Results of UNC SynUrban and SynAuto mixture simulations.
Experiment
Number
Initial Conditions
NO
N02/NOX
HC/NO..
Maximum Ozone^ppmJ
Calc.
(ppm)
Meas.
(ppm)
Abs.
Diff.
Pet.
Diff.
SynUrban:
AU2284R
AU2584B
ST0184B
ST0284B
ST0284R
ST0384R
SynAuto:
OC0483B
OC0783B
OC0783R
AU0484R
AU0584B
AU0584R
AU0684B
AU0684R
AU0784R
AU0884B
AU0984R
0.324 0.216 9.9
0.332 0.211 3.4
0.304 0.194 10.7
0.352 0.256 2.3
0.342 0.170 3.6
0.353 0.304 3.1
0.791
0.073
0.748
0.015
0.096
0.045
0.657
0.096
0.646
0.020
0.119
0.048
0.134
-0.023
0.102
-0.005
-0.023
-0.003
20.4%
-24.0%
15.8%
-25.0%
-19.3%
- 6.3%
SynUrban Average
Std. Deviation
0.030 - 6.4%
±0.069 ±20.2%
0.251 0.147 8.7
0.339 0.112 8.1
0.331 0.121 8.0
0.370 0.208 3.4
0.349 0.218 3.7
0.357 0.216 2.7
0.355 0.203 9.3
0.354 0.201 6.5
0.385 0.216 3.5
0.342 0.228 10.8
0.389 0.224 3.4
0.623
0.567
0.282
0.415
0.580
0.254
0.876
0.836
0.528
0.834
0.404
0.642
0.451
0.179
0.515
0.595
0.335
0.940
0.887
0.602
0.834
0.521
-0.019
0.116
0.103
-0.100
-0.015
-0.081
-0.064
-0.051
-0.074
0.000
-0.117
- 3.0%
25.7%
57.5%
-19.4%
- 2.5%
-24.2%
- 6.8%
- 5.7%
-12.3%
0.0%
-22.5%
SynAuto Average
Std. Deviation
-0.027 - 1.2%
±0.077 ±23.8%
SynAutoUrban:
ST0884B
ST0884R
ST1784R
ST2184B
ST2184R
0.331
0.339
0.341
0.359
0.363
0 380
0.351
0.205
0.240
0.242
8.6
5.7
6.5
7.2
6.6
SynAutoUrban
0.712
0.621
0.589
0.820
0.748
Average
0.750
0.566
0.539
0.721
0.671
Std. Deviation
-0.038
0.055
0.050
0.099
0.077
0.049
±0.052
- 5.1%
9.7%
9.3%
13.7%
11.5%
7.8%
7.4%
Synthetic Mix Average
Std. Deviation
0.006 - 0.6%
±0.075 ±20.1%
(continued)
-------
TABLE 6-16. (concluded)
Maximum PAN
Experiment Calc.
Number (ppm)
SynUrban:
AU2284R 0.056
AU2584B 0.002
ST0184B 0.062
ST0284B 0.000
ST0284R 0.001
ST0384R 0.002
SynUrban Average
Std. Deviation
SynAuto:
OC0483B 0.047
OC0783B 0.083
OC0783R 0.036
AU0484R 0.006
AU0584B 0.012
AU0584R 0.002
AU0684B 0.070
AU0684R 0.043
AU0784R 0.008
AU0884B 0.077
AU0984R 0.004
SynAuto Average
Std. Deviation
Meas.
iPPm)
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
_
.004
.052
.004
.004
-
.060
.101
.056
.028
.032
.013
.142
.086
.031
.130
.021
(ppm)
Abs.
Diff.
-0.
0.
-0.
-0.
0.
±0.
-0.
-0.
-0.
-0.
-0.
-0.
-0.
-0.
-0.
-0.
-0.
-0.
±0.
_
002
010
004
003
-
000
007
017
018
020
022
020
on
072
043
023
053
017
029
019
Pet.
Diff.
—
- 50%
16%
_
- 75%
-
- 36%
± 47%
- 28%
- 18%
- 36%
- 79%
- 63%
- 85%
- 51%
- 50%
- 74%
- 41%
- 81%
- 55%
± 23%
Maximum
Calc.
(ppm)
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.11
.05
.12
.04
.05
.05
.16
.16
.09
.10
.11
.09
.25
.18
.12
.28
.10
Meas.
(ppm)
0.16
0.07
0.15
0.09
0.08
0.06
0.30
0.33
0.11
0.04
0.10
0.08
0.19
0.19
_
0.33
0.14
HCHO
(ppm)
Abs.
Diff.
-0.
-0.
-0.
-0.
-0.
-0.
-0.
±0.
-0.
-0.
-0.
0.
0.
0.
0.
-0.
-0.
-0.
05
02
03
05
03
01
03
02
14
17
02
06
01
01
06
01
_
05
04
- 0.03
± 0.08
Pet.
Diff.
- 31%
- 29%
- 20%
- 56%
- 38%
- 17%
- 32%
± 14%
- 47%
- 52%
- 18%
150%
10%
13%
38%
- 5%
_
- 15%
- 29%
5%
± 58%
SynAutoUrban:
ST0884B
ST0884R
ST1784R
ST2184B
ST2184R
58% 0.18
53% 0.12
13% 0.14 0.22
0% 0.23 0.23
0.047 0.052 -0.005 - 10% 0.17 0.19
0.082
0.046
0.078
0.061
0.052
0.030
0.069
0.061
0.030
0.016
0.009
0.000
-0.08 - 36%
-0.00 0%
-0.02 - 11%
SynAutoUrban Average
Std. Deviation
0.010 23%
±0.014 ± 31%
-0.03
±0.04
- 16%
± 18%
Synthetic Mix Average -0.013 - 32%
Std. Deviation ±0.024 ± 44%
-0.03
±0.06
- 10%
± 45%
-------
TABLE 6-17. Initial conditions for UCR-EC seven component hydrocarbon mixture simulations.
Initial
Concentrations (ppm)
Gas Phase
Experiment
Number
EC-231
EC-232
EC-233
EC-237
EC-238
EC-241
EC-242
EC-243
EC-245
EC-246
EC-247
NO
0.440
0.469
0.096
0.377
0.718
0.379
0.377
0.386
0.743
0.386
0.380
N02
0.052
0.024
0.007
0.106
0.234
0.110
0.125
0.114
0.259
0.122
0.125
(ppmC)
13.2
9.3
9.5
10.5
10.1
5.0
12.8
9.7
12.9
8.6
6.2
HONO
0.016
0.006
0.005
0.012
0.024
0.002
0.018
0.018
0.035
0.015
0.018
HCHO
0.026
0.009
0.004
0.026
0.026
0.026
0.028
0.003
0.016
0.004
0.003
CO
1.630
1.740
1.299
0.963
0.680
1.300
1.400
0.610
2.500
0.000
0.440
Other Conditions
Walls Water Vapor (ppm)
HCHO
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
kHONO- Initial
1.17E-
1.17E-
1.17E-
1.17E-
1.17E-
1.17E-
17000
17000
17000
17000
17000
17000
1.17E-4 17000
1.17E-4 17000
1.17E-4 17000
1.17E-4 17000
1.37E-4 17000
Final
17000
17000
17000
17000
25500
25500
25500
25500
25500
17000
25500
Temperature (K)
Initial
302.6
302.6
302.6
302.6
302.6
302.6
302.6
302.6
302.6
302.6
302.6
Final
304.4
304.4
304.4
304.4
303.4
303.4
303.4
303.4
303.4
304.4
303.4
-------
TABLE 6-18. Results of UCR-EC seven component hydrocarbon mixture simulations,
Initial Conditions
Experiment
Number
EC-231
EC-232
EC-233
EC-237
EC-238
EC-241
EC-242
EC-243
EC-245
EC-246
EC-247
NOV
0.492
0.493
0.103
0.483
0.952
0.489
0.502
0.500
1.002
0.508
0.505
N02/NOX
• " L. ' ~ A
0.113
0.049
0.068
0.219
0.244
0.225
0.249
0.772
0.742
0.240
0.752
HC/NOV
26.8
18.9
92.3
21.7
10.6
10.1
25.6
19.4
12.8
16.9
12.2
Maximum Ozone (ppm)
Calc.
(ppm)
0.722
0.405
0.405
0.706
0.776
0.400
0.658
0.715
0.922
0.655
0.665
Meas.
(ppm)
0.620
0.305
0.327
0.655
0.694
0.408
0.682
0.716
0.892
0.574
0.655
Abs.
Diff.
0.102
0.100
0.078
0.051
0.082
-0.008
-0.024
-0.001
0.030
0.081
0.010
Pet.
Diff.
16.5%
32.8%
23.8%
7.8%
11.8%
- 2.0%
- 3.5%
0.1%
3.4%
14.1%
1.5%
UCR-EC Average
Std. Deviation
0.046 9.7%
±0.046 ±11.5%
Maximum
Experiment
Number
EC-231
EC-232
EC-233
EC-237
EC-238
EC-241
EC-242
EC-243
EC-245
EC-246
EC-247
Calc.
(ppm)
0.158
0.043
0.051
0.153
0.192
0.052
0.180
0.140
0.297
0.107
0.163
Meas.
(ppm)
0.095
0.040
0.037
0.100
0.113
0.047
0.140
0.100
0.194
0.070
0.106
UCR-EC Average
Std. Deviation
PAN (ppm)
Abs.
Diff.
0.063
0.003
0.014
0.053
0.079
0.005
0.040
0.040
0.103
0.037
0.057
0.045
±0.031
Pet.
Diff.
66%
8%
38%
53%
70%
11%
29%
40%
53%
53%
54%
43%
± 20%
Maximum HCHO (ppm)
Calc.
(ppm)
0.53
0.14
0.18
0.49
0.48
0.23
0.89
0.90
1.04
0.14
0.52
Meas.
(ppm)
0.57
_
_
0.39
0.58
0.22
0.84
0.77
0.94
0.12
0.45
Abs.
Diff.
-0.04
_
^
0.10
-0.10
0.01
0.05
0.13
0.10
0.02
0.07
0.04
±0.07
Pet.
Diff.
- 7%
_
_
26%
- 17%
5%
6%
17%
11%
17%
16%
8%
± 13%
386
-------
TABLE 6-19. Initial conditions for UCR-ITC multi-day simulations.
OJ
CO
Initial
Concentrations (ppm)
Gas Phase
Experiment
Number
ITC630
ITC631
ITC633
ITC635
ITC637
NO
0.233
0.237
0.457
0.902
0.237
N02
0.073
0.092
0.182
0.304
0.073
NMOC
(ppmC)
1.9
1.0
4.0
4.0
4.0
HONO
0.005
0.010
0.010
0.015
0.015
HCHO
0.007
0.020
0.013
0.013
0.007
CO
0.963
0.963
0.963
0.963
0.963
Walls
''HCHO
l.OE-4
l.OE-4
l.OE-4
l.OE-4
l.OE-4
^HONO-
3.50E-5
3.50E-5
3.50E-5
3.50E-5
3.50E-5
Other Conditions
Water Vapor (ppm)
Initial
17000
17000
17000
17000
17000
Final
17000
17000
17000
17000
17000
Temperature (K)
Initial
293.6
293.6
293.6
293.6
293.6
Final
300.0
303.0
300.0
299.0
302.9
-------
TABLE 6-20. Results of UCR ITC multi-day experiment simulations.
oo
oo
CO
Initial Conditions
Experiment
Number
ITC-630
ITC-631
ITC-633
ITC-635
ITC-637
NOX
0.306
0.329
0.639
1.206
0.310
N02/NOX
0.239
0.280
0.762
0.252
0.235
HC/NOX
6.3
3.2
6.2
3.3
12.7
Maximum
Calc.
(ppm)
0.023
0.015
0.045
0.004
0.075
Me as.
(ppm)
0.027
0.022
0.044
_
0.117
UCR-ITC Average
Std. Devi
atlon
PAN (ppm)
Abs.
Diff.
-0.004
-0.007
0.001
-
-0.042
-0.013
± 0.020
Pet.
01ff.
- 15*
- 32*
2%
-
- 36%
- 20%
± 17%
Maximum HCHO (ppm)
Calc.
(ppm)
0.09
0.06
0.17
0.16
0.18
Meas.
(ppm)
0.12
0.07
-
0.11
0.13
Abs.
D1ff.
-0.03
-0.01
-
0.05
0.05
0.02
±0.04
Pet.
Diff.
- 25%
- 14%
-
45%
38%
11%
± 36%
(continued)
870>«8
-------
TABLE 6-20. (concluded)
Maximum Ozone (ppm)
Experiment
Number
Calc. Meas.
(ppm) (ppm)
ITC-630 0.228 0.284
ITC-631 0.050 0.042
ITC-633 0.259 0.231
ITC-635 0.018 0.006
ITC-637 0.614 0.617
UCR-ITC Average
Std. Deviation
Day 1
Abs.
D1ff.
-0.056
0.008
0.028
0.012
0.003
-0.001
±0.032
Day 2
Pet.
Dlff.
-19. 7X
19. OX
12. IX
200. OX
0.5X
42.4 ( 3.0)X
±89.3 (16.9)X
Calc.
(ppm)
0.123
0.097
0.218
0.022
0.630
Meas.
(ppm)
0.213
0.165
0.340
0.024
0.577
Abs.
D1ff.
-0.090
-0.068
-0.122
-0.002
0.053
-0.046
±0.070
Pet.
D1ff.
- 42X
- 41X
- 36X
- 8X
9X
- 24X
± 23X
Calc.
(ppm)
0.234
0.294
0.188
0.536
Day 3
Meas.
(ppm)
0.274
0.242
0.272
0.537
Abs.
Dlff.
-0.040
0.052
-0.084
-0.001
0.018
±0.058
Pet.
Dlff.
- 15X
21X
- 31X
OX
- 6X
± 22X
Calc.
(ppm)
0.285
Day 4
Meas. Abs.
(ppm) Dlff.
0.384 -0.099
Pet.
D1ff.
- 26X
-------
TABLE 6-21. Initial conditions for UNC hydrocarbon reactivity experiment simulations.
Initial Concentrations (ppm)
Gas Phase
Experiment
Number
OC0382R
OC0382B
AU3181K
AU3181B
JL2281R
JL2281B
JN2581R
JN2581B
UC0782R
OC0782B
SE1481R
SE1481B
SE1682R
SE1682B
NV1182R
NV1182B
SE1882R
SE1882B
SE2081R
SE2081B
SE2981R
SE2981B
SE0381R
SE0381B
SE1081R
SE1081B
AU2681R
AU2681B
SE1984B
SE1984R
NO
0.194
0.193
0.187
0.187
0.209
0.205
0.128
0.130
0.157
0.158
0.196
0.193
0.302
0.302
0.143
0.143
0.282
0.289
0.178
0.177
0.168
0.177
0.180
0.179
'0.190
0.188
0.184
0.179
0.252
0.253
N0g
0.056
O.U55
0.052
0.052
0.054
0.052
0.055
0.057
0.030
0.028
0.053
0.054
0.121
0.121
0.027
0.028
0.102
0.097
0.053
0.052
0.070
0.068
0.055
0.052
0.055
0.058
0.053
0.056
0.086
0.086
NMOC
(ppmC)
2.75
2.56
2.04
2.09
2.40
2.67
1.81
1.68
3.37
3.49
2.19
2.17
3.23
3.13
3.19
3.36
3.10
3.06
2.27
2.08
2.68
2.66
2.31
3.06
3.49
3.73
2.02
2.95
2.69
4.38
MONO
0.003
0.003
0.003
0.003
0.003
0.003
0.003
0.003
0.001
0.001
0.003
0.003
0.010
0.010
0.001
0.001
0.010
0.010
0.003
0.003
0.003
0.003
0.003
0.003
0.005
0.005
0.003
0.003
0.003
0.003
HCHO
0.010
0.010
0.010
0.010
0.010
0.010
0.010
0.010
0.000
0.000
0.010
0.010
0.010
0.010
0.000
0.000
0.010
0.010
0.010
0.010
0.010
0.010
0.010
0.010
0.010
0.010
0.010
0.010
0.010
0.010
CO
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
Walls
HCHO
0.040
0.040
0.040
0.040
0.040
0.040
0.040
0.040
0.050
0.050
0.040
0.040
0.040
0.040
0.050
0.050
0.040
0.040
0.040
0.040
0.040
0.040
0.040
0.040
0.040
0.040
0.040
0.040
0.040
0.040
—hHONO—
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
l.OOE-5
l.OOE-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
1.50E-5
1.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
2.50E-5
Other Conditions
Water Vapor (ppm)
Initial
17200
17200
15200
15200
15000
15000
15000
15000
9500
9500
12800
12800
20600
20600
7000
7000
22800
22800
128UO
12800
7800
7800
22200
22200
10800
10800
17000
17000
12000
12000
Final
23200
23200
18000
18000
25000
25000
25000
25000
16300
16300
21000
21000
24200
24200
23000
23000
32400
32400
21000
21000
22000
22000
24200
24200
24800
25800
18000
18000
17600
17600
Temperature (K)
Initial
288.6
288.6
291.5
291.5
303.0
303.0
296.4
296.4
279.7
279.7
289.0
289.0
294.6
294.6
276.7
276.7
291.5
291.5
279.5
279.5
278.8
278.8
292.1
292.1
283.9
283.9
287.1
287.1
280.2
280.2
Final
301.6
301.6
304.8
304.8
303.0
303.0
307.7
307.7
289.5
289.5
302.8
302.8
306.2
306.2
295.5
295.5
306.0
306.0
298.3
298.3
296.2
296.2
303.5
303.5
302.6
302.6
302.1
302.1
299.7
299.7
Light
Conditions
Good
Good
Choppy
Choppy
Good
Good
Good
Good
Good
Good
Good
Good
Late Chop
Late Chop
Late Chop
Late Chop
Low Early
Low Early
Good
Good
Good
Good
Choppy
Choppy
Good
Good
Choppy
Choppy
Good
Good
-------
TABLE 6-21. (continued)
Experiment
Number
OC0382R
OC0382B
AU3181R
AU3181B
JL2281R
JL2281B
JN2581R
JN2581B
DC0782R
DC0782B
SE1481R
SE1481B
SE1682R
SE1682B
NV1182R
NV1182B
SE1882R
SE1882B
SE2081R
SE2081B
Experimental Design
Notes
Ratio Test:
Matched:
Substitution:
Substitution:
Substitution:
Comparison:
Comparison:
Comparison:
. Comparison:
Comparison:
2.8 SIMmlx 0 11.0 to 1
1.6 SIMmlx 9 6.3 to 1
2.0 SIMmlx 0 8.5 to 1
2.1 UNCmlx 0 8.7 to 1
2.4 UNCmlx 0 9.1 to 1
1.8 UNCmlx ••• 0.8 MXY 0 10.4 to 1
1.3 UNCmlx + 0.5 ISOP 9 9.9 to 1
1.7 UNCmlx 0 9.0 to 1
2.5 UNCmlx + 0.9 COMARO 0 18.0 to 1
3.5 UNCmlx 0 18.8 to 1
1.6 UNCmlx + 0.6 MXY 0 8.8 to 1
1.7 UNCmlx + 0.5 TOL 0 8.8 to 1
2.3 UNCmlx + 0.9 EtBz 0 7.6 to 1
2.2 UNCmlx + 0.9 MXY 0 7.4 to 1
2.3 UNCmlx + 0.9 TMBz 0 18.8 to 1
2.5 UNCmlx + 0.8 PrBz 0 19.6 to 1
2.2 UNCmlx + 0.9 OXY 0 8.1 to 1
2.2 UNCmlx + 0.9 MXY 0 7.9 to 1
1.6 UNCmlx + 0.7 COMARO 0 9.8 to 1
1.6 UNCmlx + 0.5 SIMARO 0 9.1 to 1
Dried but HEAVY condensation
Vented 24 hours
Dried and NO condensation
Vented 24 hours
No drying and NO condensation
Vented 8 hours
No drying and NO condensation
Used 1982 light/choppy
No drying, OBSERVED condensation
Low Initial temperature
Dried and NO condensation
Vented 7 hours
Dried and NO condensation
Vented 24 hours
No drying, OBSERVED condensation
Dried but OBSERVED condensation
Vented 6 hours, early light 1s low
Dried but SOME condensation
Vented 7 hours
(continued)
-------
TABLE 6-21. (concluded)
Experiment
Number
SE2981R
SE2981B
SE0381R
SE0381B
SE1081R
SE1U81B
AU2681R
AU2681B
SE1984R
SE1984B
Comparison:
Addition:
Substitution:
Comparison:
Ratio Test:
Experimental Design
2.0 UNCmlx + 0.5 TOL + 0.1 ALD2 0 11.3 to 1
2.0 UNCmlx + 0.5 MXY + 0.1 ALD2 0 10.9 to 1
1.8 UNCmlx + 0.5 SIMARO 0 9.8 to
1.9 UNCmlx + 0.5 SIMARO + 0.6 PROPENE 0 13.2 to
2.8 UNCmlx + 0.7 SIMARO 0 14.2 to 1
1.0 UNCmlx + 0.5 SIMARO + 2.2 BUTANE 0 15.2 to 1
1.5 UNCmlx + 0.6 SIMARO 0 8.5 to 1
1.4 SIMmlx + 0.6 SIMARO 0 8.3 to 1
4.4 High MW Aromatic Fraction 0 17.2 to 1
2.7 High MW Aromatic Fraction 0 10.6 to 1
1
1
Notes
No drying, OBSERVED condensation
ALD2 not observed after Injection
Dried and NO condensation. Vented 7 hours
High ETH on BLUE side, choppy light
Dried and NO condensation
Dried and NO condensation
Vented 7 hours
Dried and NO condensation
Vented 7 hours
-------
TABLE 6-22. Results of UNC hydrocarbon reactivity experiment simulations.
Initial Conditions
Experiment
Number
OC0382B
OC0382R
AU3181B
AU3181R
JL2281B
JL2281R
JN2581B
JN2581R
DE0782B
DE0782R
SE1481B
SE1481R
SE1682B
SE1682R
NV1182B
NV1182R
SE1882B
SE1882R
SE2081B
SE2081R
SE2981B
SE2981R
SE0381B
SE0381R
SE1081B
SE1081R
AU2681B
AU2681R
SE1984B
SE1984R
NOW
0.248
0.250
0.239
0.239
0.257
0.263
0.187
0.183
0.186
0.187
0.247
0.249
0.423
0.423
0.171
0.170
0.386
0.384
0.229
0.231
0.245
0.238
0.231
0.235
0.246
0.245
0.235
0.237
0.338
0.339
N02/NOX
" tfc 'A
0.222
0.224
0.218
0.218
0.202
0.205
0.305
0.301
0.151
0.160
0.219
0.215
0.286
0.286
0.164
0.159
0.251
0.266
0.227
0.229
0.273
0.294
0.225
0.238
0.236
0.224
0.238
0.224
0.254
0.254
HC/NCL
" '* ' 'A—
10.3
11.0
8.7
8.5
10.4
9.1
9.0
9.9
18.8
18.0
8.8
8.8
7.4
7.6
19.6
18.7
7.9
8.1
9.1
9.4
10.4
10.8
10.5
9.8
15.1
14.8
8.3
8.5
10.6
17.2
Average
Std. Deviation
Maximum Ozone (ppm]
Calc.
(ppm)
0.159
0.380
0.565
0.604
0.583
0.555
0.581
0.806
0.100
0.148
0.418
0.663
0.869
0.359
0.182
0.322
0.798
0.833
0.438
0.467
0.528
0.409
0.650
0.400
0.642
0.653
0.559
0.548
0.448
0.446
Meas.
(ppm)
0.158
0.247
0.530
0.665
0.722
0.618
0.744
0.830
0.093
0.076
0.441
0.685
0.840
0.410
0.220
0.218
0.791
0.669
0.414
0.404
0.485
0.294
0.611
0.542
0.626
0.611
0.544
0.506
0.438
0.378
Abs.
Diff.
0.001
0.133
0.035
-0.061
-0.139
-0.063
-0.163
-0.024
0.007
0.072
-0.023
-0.022
0.029
-0.051
-0.038
0.104
0.007
0.164
0.024
0.063
0.043
0.115
0.039
-0.142
0.016
0.042
0.015
0.042
0.010
0.068
0.010
±0.076
Pet.
Diff.
0.6%
53.8%
6.6%
- 9.2%
-19.3%
-10.2%
-21.9%
- 2.9%
7.5%
94.7%
- 5.2%
- 3.2%
3.5%
-12.4%
-17.3%
47.7%
0.9%
18.5%
5.8%
15.6%
8.9%
39.1%
6.4%
-26.2%
2.6%
6.9%
3.8%
8.3%
2.3%
18.0%
7.5%
±25.6%
(continued)
-------
TABLE 6-22. (concluded)
Maximum
Experiment
Number
OC0382B
OC0382R
AU3181B
AU3181R
JL2281B
JL2281R
JN2581B
JN2581R
DE0782B
DE0782R
SE1481B
SE1481R
SE1682B
SE1682R
NV1182B
NV1182R
SE1882B
SE1882R
SE2081B
SE2081R
SE2981B
SE2981R
SE0381B
SE0381R
SE1081B
SE1081R
AU2681B
AU2681R
SE1984B
SE1984R
Calc.
(ppm)
0.007
0.025
0.089
0.053
0.031
0.075
0.017
0.031
0.025
0.066
0.090
0.020
0.025
0.070
0.079
0.086
0.054
0.061
0.109
0.066
0.089
0.034
0.055
0.085
0.051
0.053
0.070
0.106
Average
Meas.
(ppm)
0.004
0.008
_
_
0.097
0.055
0.027
0.044
0.009
0.010
_
_
0.047
0.015
_
_
0.087
0.070
_
_
_
_
_
_
_
_
_
_
0.103
0.124
Std. Deviation
PAN (ppm)
Abs.
Diff.
0.003
0.017
_
_
-0.008
-0.002
0.004
0.031
0.008
0.021
_
_
0.043
0.005
_
_
-0.008
0.016
_
_
_
_
_
_
_
_
-
_
-0.033
-0.018
0.006
±0.020
Pet.
Diff.
75%
213%
_
_
- 8%
- 4%
15%
70%
89%
210%
_
_
91%
33%
_
_
- 9%
23%
_
_
_
_
_
_
_
_
_
_
- 32%
- 15%
54%
± 78%
Maximum HCHO (ppm)
Calc.
(ppm)
0.09
0.14
0.09
0.10
0.11
0.10
0.08
0.18
0.10
0.09
0.08
0.10
0.15
0.12
0.11
0.11
0.13
0.13
0.08
0.08
0.10
0.09
0.18
0.09
0.06
0.13
0.09
0.08
0.05
0.08
Meas.
(ppm)
_
0.12
0.13
0.17
0.17
0.08
0.22
0.10
0.06
_
_
0.19
0.14
0.20
0.16
0.17
0.10
_
_
_
_
_
_
_
_
0.15
0.11
0.07
0.13
Abs.
Diff.
-
-0.03
-0.03
-0.06
-0.07
0.00
-0.04
0.00
0.03
-
_
-0.04
-0.02
-0.09
-0.05
-0.04
0.03
_
_
_
_
_
_
_
_
-0.06
-0.03
-0.02
-0.05
-0.03
±0.03
Pet.
Diff.
-
- 25%
- 23%
- 35%
- 41%
0%
- 18%
0%
50%
_
_
- 21%
- 14%
- 45%
- 31%
- 24%
30%
_
_
_
_
_
_
_
_
- 40%
- 27%
- 29%
- 38%
- 18%
± 25%
394
-------
SUMMARY AND CONCLUSIONS
This report documents the development and evaluation of the Carbon-Bond
Mechanism-IV (CBM-IV), a gas-phase photochemical kinetics mechanism suit-
able for inclusion in both urban and regional air quality simulation
models. It also summarizes the mechanism development philosophy that has
led to the carbon-bond approach over the past fifteen years of continuous
EPA support. We describe the relevant decisions on the inclusion of fun-
damental kinetic and stochiometric data, the approaches used to condense
these data into a carbon-bond mechanistic structure, and the eventual
choices made to reduce the extended CBM-X to the condensed CBM-IV. The
most concise summary of this work is the CBM-IV listing provided in Table
1-3. All fundamental decisions ultimately appear in the mechanism list-
ing.
The development of the CBM-IV initially proceeded through a gathering of
recent chemical kinetic and mechanistic data-pertinent tropospheric gas-
phase chemistry. The reaction rates and product yields were then updated
and tested. In particular, large portions of the inorganic section were
altered to reflect current thought. We also undertook the task of improv-
ing a few specific portions of the organic chemistry section where there
was ample data available to test our assumptions. Specifically, we inves-
tigated
(1) The reactions of peroxyacetyl radical with NOX species and other
radicals,
(2) The kinetics and product yields of radical-radical self-reac-
tions,
(3) New representations for toluene and xylene oxidation reactions,
and
(4) The condensation of the isoprene oxidation mechanism and
development of o-pinene carbon-bond fractions.
Finally, we altered all sections of the organic reaction mechanism that
previously utilized formaldehyde as a surrogate for other species so that
the concentration of the species FORM in the CBM-IV is now explicitly that
of formaldehyde.
-------
The improvements to the temperature-dependent kinetics of the peroxyacetyl
radical reactions with NO and N02 were shown to lead to much better simu-
lation of PAN formation and decay in simple acetaldehyde and biacetyl sys-
tems. These changes, combined with the improvements to the organic radi-
cal self-reaction rate constants and product yields, have led to what
appears to be a more accurate simulation of radical concentrations
throughout a simulation day. The basis for this assertion is the better
prediction of hydrocarbon, especially aromatic hydrocarbon, decay rates
(indicators of OH concentration) and the performance of the CBM-IV in
simulating the shape of PAN formation and decay throughout the day.
The chemistry of aromatic hydrocarbons, particularly toluene and other
mono-substituted benzenes, was updated to reflect the chemical charac-
teristics apparent from smog chamber data for those species. Particularly
important was the fact that toluene (and to a lesser extent, xylene) smog
chamber systems tend to rapidly deplete NO until ozone formation is cur-
tailed, but toluene decay continues. These data trends suggested an
alternate pathway for the reaction of the initial OH-02~aromatic adduct
when NO is sufficiently low that such a reaction could be competitive.
This assumption removed the exclusive reaction of the adduct with NO
(found in all other mechanisms) and led to a far better prediction of
chemical dynamics in these systems.
The isoprene reaction mechanism was also reanalyzed and a new condensed
formulation was performed. This resulted in a far better predictive capa-
bility for maximum ozone concentration, actually simulating the double
ozone peaks characteristic of isoprene systems (due to late afternoon PAN
decomposition on hot days). In addition, since the CBM-IV is now
explicitly predicting the concentration of formaldehyde, the reformulation
of the condensed isoprene mechanism was directed toward accurate formal-
dehyde product formation. Comparisons of simulation results with experi-
mental data were very good, indicating that the use of isoprene in the
CBM-IV simulation of rural systems should not be hampered by poor repre-
sentations. Continuing with biogenic hydrocarbon species, we also used
the available University of North Carolina smog chamber data to improve
our carbon-bond fractionation scheme for a-pinene. Although the simula-
tion results were not as astounding as those for isoprene, they were quite
accurate considering the complexity of the molecule and the less-than-
average light conditions for many of the simulation days.
Besides specific kinetic and mechanistic improvements to the chemical
mechanism, we also provide a detailed description of some important topics
not directly addressed in earlier CBM development documents. These
include
-------
(1) The basic philosophy of the carbon-bond approach and the
specific implementation of that philosphy in the alkane section
of the CBM-IV,
(2) The rationale used for condensation of specific sections of the
CBM-X in the CBM-IV,
(3) A description of the photolysis rate data and application of
photolysis reactions in specific chambers and in the atmosphere,
and
(4) A discussion of the chamber artifact reactions used to simulate
different smog chambers and the basis for the development of
specific reaction-rate constants.
The last of these topics is fundamental to the accurate simulation of smog
chamber experiments. There is, however, a great deal of uncertainty in
these approximations, which we attempt to note throughout the discus-
sion. Nevertheless, the comparison of mechanism predictions with smog
chamber data must be performed to improve, and later to evaluate, the
mechanism performance against a controlled but complex mixture of hydro-
carbons and NOX.
Mechanism evaluation and demonstration of the chemical dynamic charac-
teristics of the CBM-IV were performed using approximately 170 smog cham-
ber experiments from the different smog chambers at the University of
North Carolina and the University of California at Riverside facilities.
We compared maximum experimental ozone, PAN, and formaldehyde concentra-
tions with predicted values and also provided plots of these concentration
comparisons for each experiment so that we could discuss the dynamic pro-
cesses and compare the goodness-of-fit over the entire experimental
period.
As noted, we were particularly pleased that the improvements to the iso-
prene condensation and the representation of toluene (and other aromatics)
oxidation processes provided much better predictions of ozone and formal-
dehyde than had occurred in previous mechanisms. For the isoprene tests,
the mechanism overpredicted maximum ozone concentrations by 6 (±22) per-
cent and for the aromatic experiments the overprediction was 1 (±12) per-
cent. These results indicate that the new mechanistic representations
significantly diminish the uncertainty associated with these calculations.
Less obvious in the overall results were the improvements to predictive
capabilities provided by the enhanced organic radical, and peroxyacetyl
reaction chemistry. These changes appear to have increased the accuracy
-------
of the radical concentration predictions during the midday period, result-
ing in better agreement between PAN formation-and-decay rates and hydro-
carbon decay rates over a large range of temperatures. These improvements
and others described above led to a slight overprediction (2 ±22 percent)
of maximum ozone concentration averaged for simulations of 68 different
experimental mixtures in three different smog chambers. Maximum formal-
dehyde concentrations were underpredicted by 9 (±34) percent for these
mixture experiments. Given the larger uncertainties in this calculation
due to both ambiguous organic oxidation processes and less precise mea-
surement methods (than for ozone), we consider these results very good.
Remaining uncertainties in the mechanism result primarily from two
areas: (1) The fundamental information from which we develop mechanistic
representations is sparse in certain key areas, and (2) The smog chamber
data with which we test these mechanisms is itself uncertain in some
respects. With regard to specific chemistry, we are most concerned with
uncertainties that may exist in the alkene and aromatic representations.
As we have noted, the successful predictions of the aromatic mechanism
were gratifying, but since the fundamental information concerning the
aromatic oxidation kinetics and mechanisms is still poorly understood, we
cannot be sure the predictions are founded on the actual chemistry.
Therefore, the mechanism could display anomalies outside the range of con-
ditions in the smog chamber test environment. Conversely, though many of
the basic oxidation reaction kinetics for the alkene (ethene and olefin)
mechanisms have supposedly been identified, these mechanisms often perform
poorly in the simulation of smog chamber measurements. Whether this is
due to inconsistencies in the mechanism or uncertainties in the descrip-
tion of the smog chamber environment should be investigated.
Uncertainties in smog chamber information, particularly the effects of
chamber surfaces and spectral distribution of in-chamber light, are cur-
rently being reviewed by a number of groups. It is imperative now that
researchers and simulators merge their opinions and knowledge of these
systems so that we may minimize uncertainties related to wall radical
sources, surface collection and emission of key reactive species, and
daily changes in both light and surface conditions. This process would
not only enable us to define key experiments for mechanism testing, but
should provide information for the clarification of many marginal experi-
mental data sets, thus allowing the already existing data to be utilized
in a more effective manner.
In conclusion, we find the CBM-IV an improved and we11-performing
mechanism, suitable for use in regional and urban air quality simulation
models. Certain uncertainties remain in the mechanism due to lack of fun-
damental kinetic data in a few key areas, most notably in aromatic and
organic radical chemistry. In addition, smog chamber information data
398
-------
could be Improved and utilized to better advantage if significant chamber
environmental aspects (radiation and surface effects) could be better
defined. We recommend a focus on these uncertainties so that more useful
test cases can be employed in future mechanism development and evaluation
efforts.
-------
REFERENCES
Addison, M. C., J. P. Burrows, R. A. Cox, and R. Patrick. 1980. Absorp-
tion spectrum and kinetics of the acetylperoxy radical. Chem. Phys.
Lett.. 73:283-287.
Akimoto, H., M. Hoshino, G. Inoue, F. Kakamaki, N. Washida, and
M. Okuda. 1979. Design and characterization of the evaluable and
bakable photochemical smog chamber. Environ. Sci. Technol.,
13(4):471-475.
Atkinson, R., W.P.L. Carter, K. R. Darnall, A. M. Winer, and J. N. Pitts,
Jr. 1980. A smog chamber and modeling study of the gas phase NOX-
air photooxidation of toluene and the cresols. Int. J. Chem. Kinet..
12:779-836.
Atkinson, R. 1986. Kinetics and mechanisms of the gas-phase reactions of
the hydroxyl radical with organic compounds under atmospheric condi-
tions. Chem. Rev.. 86:69-201.
Atkinson, R., A. C. Lloyd, and L. Winges. 1982a. An updated chemical
mechanism for hydrocarbon/NOx/S02 photooxidations suitable for inclu-
sion in atmospheric simulation models. Atmos. Environ.. 16(6):1341-
1355.
Atkinson, R., S. M. Aschman, W.P.L. Carter, A. M. Winer, and J. N. Pitts,
Jr. 1982b. Alkyl nitrate formation from the NOX—air photooxida-
tions of C2-C8 n-alkanes. J. Phys. Chem.. 86:4563-4569.
Atkinson, R., S. M. Aschman, and A. M. Winer. 1987. Alkyl nitrate forma-
tion from the reaction of a series of branched R02 radicals with NO
as a function of temperature and pressure. J. Atmos. Chem.. 5:91-
102.
Atkinson, R., W.P.L. Carter, and A. M. Winer. 1983. Effects of tempera-
ture and pressure on alkyl nitrate yields in the NOX photooxidations
of n-pentane and n-heptane. J. Phys. Chem.. 87:2012-2018
400
-------
Atkinson, R.t and W.P.L. Carter. 1987. Atmospheric Chemistry of
Alkanes. J. Atmos. Chem.. 3:377-405.
Atkinson, R.t and A. C. Lloyd. 1984. Evaluation of kinetic and mechanis-
tic data for modeling of photochemical smog. J. Phys. Chem. Ref.
Data, 13:315-444.
Atkinson, R., and W.P.L. Carter. 1984. Kinetics and mechanisms of the
gas-phase reactions of ozone with organic compounds under atmospheric
conditions. Chem. Rev.. 84:437-470.
Atkinson, R., and S. M. Aschmann. 1984. "Rate constants for the reaction
of OH radicals with a series of alkenes and dialkenes at 295 ± 1 K.
Int. J. Chem. Kinet.. 16:1175-1186.
Atkinson, R., S. M. Aschmann, A. M. Winer, and J. N. Pitts, Jr. 1984.
Kinetics of the gas-phase reactions of N03 radicals with a series of
dialkenes, cycloalkenes, and monoterpenes at 295 ± 1 K. Environ.
Sci. Technol.. 18(5):375.
Atkinson, R., and J. N. Pitts, Jr. 1977. Absolute rate constants for the
reaction of 0( P) atoms with a series of olefins over the temperature
range 298-439° K. J. Chem. Phys.. 67:38.
Bandow, H., N. Washida, and H. Akimoto. 1985. Ring-cleavage reactions of
aromatic hydrocarbons studied by FT-IR spectroscopy. I. Photooxida-
tion of toluene and benzene in the N0x-air system. Bull. Chem. Soc.
Jpn., 58:2531-2540.
Bandow, H., and N. Washida. 1985. Ring-cleavage reactions of aromatic
hydrocarbons studied by FT-IR spectroscopy. II. Photooxidation of
o-t m-, and p-xylenes in the N0x-air system. Bull. Chem. Soc. Jpn.,
58:2541-2548.
Bass, A. 1985. Personal communication with Arnold M. Bass, National
Bureau of Standards, Gaithersburg, Maryland.
Bass, A. M., L. C. Glasgow, C. Miller, J. P. Jesson, and D. L. Filken.
1980. Planet Space Sci., 28:675.
Barnes, I., V. Bastian, K. H. Becker, E. H. Fink, and F. Zabel. 1985.
Pressure dependence of the reaction of OH with H02N02. Chem. Phys.
Lets., 123:28-32.
Basco, N., and S. Parmar. 1987. The reaction of -acetylperoxy radicals
with N02. IJCR, 19:115-128.
401
-------
Baulch, D. L., R. A. Cox, R. F. Hampson, Jr., J. A. Kerr, J. Troe, and
R. T. Watson. 1984. Evaluated kinetic and photochemical data for
atmospheric chemistry: Supplement II. CODATA Task Group on gas
phase chemical kinetics. J. Phys. Chem. Ref. Data, 13(4):1259-1380.
Basemer, A. C., and H. Nieboer. 1985. The wall as a source of hydroxyl
radicals in smog chambers. Atmos. Environ., 19(3):507-513.
Becker, U., M. Y. Ben-Sira, S. Fritz, F. MCller, W. SchOepp, G. Yamamoto,
and N. Robinson. 1966. Solar Radiation. N. Robinson, ed. (Elsevier
Publishing Company, New York).
Benson, S. W. 1976. Thermochemical Kinetics. 2nd ed., John Wiley &
Sons, New York.
Benson, S. W., and J. H. Buss. 1958. J. Chem. Phys.. 29:546.
Bessmer, A. C., and H. Nieboer. 1985. The wall as a source of hydroxyl
radicals in smog chambers. Atmos. Environ., 19(3):507-513.
Betterton, E. A., and M. R. Hoffmann. 1987. Kinetics, mechanism, and
thermodynamics of the reversible reaction of methylglyoxal (CHoCOCHO)
with S(IV). J. Phys. Chem.. 91:3011-3020.
Calvert, J. G. 1980. "The Homogeneous Chemistry of Formaldehyde Genera-
tion and Destruction Within the Atmosphere." Proceedings of the NATO
Advanced Study Institute on Atmospheric Ozone. U.S. Department of
Transportation, F.A.A., Office of Environment and Energy, High Alti-
tude Program, Washington, D.C.
Calvert, J. G., and J. N. Pitts. 1966. Photochemistry. John Wiley, New
York.
Cantrell, C. A., W. R. Stockwell, L. G. Anderson, K. L. Busarow,
D. Perner, A. Schmeltekopf, J. G. Calvert, and H. S. Johnston.
1985. Kinetic study of the N03-CH20 reaction and its possible role
in nighttime tropospheric chemistry. J. Phys. Chem.. 89:139-146.
Carter, W.P.L. 1987. Personal communication with William Carter, State-
wide Air Pollution Research Center, University of California, River-
side, California.
402
-------
Carter, W.P.L., A. C. Lloyd, J. L. Sprung, and J. N. Pitts, Jr. 1979.
Computer modeling of smog chamber data: Progress in validation of a
detailed mechanism for the photooxidation of propene and n-butane in
photochemical smog. Int. J. Chem. Kinet., 11:45-103.
Carter, W.P.L., F. W. Lurmann, R. Atkinson, and A. Lloyd. 1986.
"Development and Testing of a Surrogate Species Chemical Reaction
Mechanism." U.S. Environmental Protection Agency, Research Triangle
Park, North Carolina (EPA/600/3-86/031).
Carter, W.P.L., M. C. Dodd, W. D. Long, and R. Atkinson. 1985. "Outdoor
Chamber Study to Test Multi-Day Effects—Volume I: Results and Dis-
cussion." Final Report, U.S. Environmental Protection Agency
(EPA/600/3-84/115).
Carter, W.P.L., R. Atkinson, A. M. Winer, and J. N. Pitts, Jr. 1982. An
experimental investigation of chamber-dependent radical sources.
Int. J. Chem. Kinet., in press.
Cattell, F. C., J. Cavanagh, R. A. Cox, and M. E. Jenkin. 1986. A
kinetics study of reactions of H02 and C2H502 using diode laser
absorption spectroscopy. J. Chem. Soc.. 82:1999-2018.
Cox, R. A., K. F. Patrick, and S. A. Chant. 1981. Mechanisms of atmo-
spheric photooxidation of organic compounds. Reactions of alkoxy
radicals in oxidation of n-butane and simple ketones. Environ. Sci.
Techno!., 15:587.
Cox, R. A., and G. S. Tyndall. 1980. J. Chem. Soc.. Faraday Trans II,
76:153.
Cox, R. A., and M. J. Roffey. 1977. Thermal decomposition of peroxy-
acetylnitrate in the presence of nitric oxide. Environ. Sci. Tech-
nol., 900-906.
Cox, R. A., and R. G. Derwent. 1975. Photolysis of Nitrous Acid in the
Presence of Acetaldehyde. J. Chem. Soc., Faraday Trans. I, 72:2061-
2075.
Darna!!, K. R., W.P.L. Carter, A. C. Lloyd, A. M. Winer, and J. N. Pitts,
Jr. 1976. Importance of R02 + NO in alkyl nitrate formation from
Cd-Cfi alkane photooxidation under simulated atmospheric conditions.
J. Phys. Chem.. 80:1948-1950.
£ no
-------
Demerjian, K. L.t J. A. Kerr, and J. G. Calvert. 1974. The mechanism of
photochemical smog formation. Adv. in Environ. Sci. and Tech.. 4,
John Wiley and Sons.
Demerjian, K. L., K. L. Schere, and J. T. Peterson. 1980. Theoretical
estimates of actinic (spherically integrated) flux and photolytic
rate constants of atmospheric species in the lower troposphere. Adv.
Environ. Sci. Technol.. 10:369-460.
DeMore, W. B., J. J. Margitan, M. J. Molina, R. T. Watson, D. M. Golden,
R. F. Hampson, M. J. Kurylo, C. J. Howard, and A. R. Ravishankara.
1985. "Chemical Kinetics and Photochemical Data for Use in Strato-
spheric Modeling." Evaluation Number 7. Jet Propulsion Laboratory,
Pasadena, California (85-37).
DeMore, W. B., M. J. Molina, R. T. Watson, D. M. Golden, R. F. Hampson,
M. J. Kurylo, C. J. Howard, and A. R. Ravishankara. 1983. "Chemical
Kinetics and Photochemical Data for Use in Stratospheric Modeling."
Evaluation Number 6. Jet Propulsion Laboratory, Pasadena, California
(83-62).
Dodge, M. C., and R. R. Arnts. 1979. A new mechanism for the reaction of
ozone with olefins. Int. J. Chem. Kinet.. 11:399-410.
Dumdei, B. E., and R. J. O'Brien. 1984. Toluene degradation products in
simulated atmospheric conditions. Nature, 311:248-250.
Gery, M. W. 1987. "Review of Chemical Mechanism Development and Evalua-
tion." In "Workshop on Evaluation/Documentation of Chemical
Mechanisms," (EPA/600/9-87/024).
Gery, M. W., D. L. Fox, H. E. Jeffries, L. Stockburger, and W. S.
Weathers. 1985. A continuous stirred tank reactor investigation of
the gas-phase reaction of hydroxyl and toluene. Int. J. Chem.
Kinet.. 17:931-955.
Gery, M. W., D. L. Fox, R. M. Kamens, and L. Stockburger. 1987. Investi-
gation of hydroxyl radical reactions with o-xylene and m-xylene in a
continuous stirred tank reactor. Environ. Sci. Technol.. 21:339.
Graham, R. A., and H. S. Johnston. 1978. The photochemistry of N03 and
the kinetics of the N205-03 system. J. Phys. Chem.. 82(3):254-268.
404
-------
Grosjean, D.f R. Countess, K. Fung, K. Ganesan, A. Lloyd, and
F. Lurmann. 1981. "Deriving Empirical Kinetic Modeling Approach
Isopleths from Experimental Data: The Los Angeles Captive-Air
Study." Environmental Research and Technology. Inc., Westlake
Village, California.
Hammer, P. D., E. J. Dlugokencky, and C. J. Howard. 1986. Kinetics
of the gas-phase reaction NO + NO, * 2ND,. J. Phys. Chem.,
90:2491-2496. *
Hendry, D. G., and R. A. Kenley. 1979. "Atmospheric Reaction Products of
Organic Compounds." U.S. Environmental Protection Agency, Office of
Toxic Substances, Washington, D.C. (EPA-560/12-79-001).
Herron, J. T., and R. E. Huie. 1977. Stopped-flow studies of the
mechanisms of ozone-alkene reactions in the gas phase. Ethylene. J^
Amer. Chem. Soc.. 99(16): 5430-5435.
Horowitz, A., and J. G. Calvert. 1978. Int. J. Chem. Kinet.. 10:805.
Horowitz, A., and J. G. Calvert. 1982. Wavelength dependence of the pri-
mary processes in acetaldehyde photolysis. J. Phys. Chem., 86:3105-
3114.
Jeffries, H. E., R. M. Kamens, K. G. Sexton, and A. A. Gerhardt. 1982.
"Outdoor Smog Chamber Experiments to Test Photochemical Models."
U.S. Environmental Protection Agency, Office of Research and Develop-
ment, Research Triangle Park, North Carolina (EPA-600/3-82-016).
Jeffries, H. E., and J. A. Arnold. 1987. "The Science of Photochemical
Reaction Mechanism Development and Evaluation." Proceedings of the
EPA Workshop on Evaluation/Documentation of Chemical Mechanisms,
Raleigh, North Carolina.
Jeffries, H. E., D. Fox, and R. Kamens. 1976. Outdoor smog chamber
studies: light effects relative to indoor chambers. Environ. Sci.
Techno!., 10(10):1006-1011.
Jeffries, H. E., K. G. Sexton, T. P. Morris, M. Jackson, R. G. Goodman,
R. M. Kamens and M. S. Holleman. Outdoor Smog Chamber Experiments
Using Automobile Exhaust. EPA 600/3-85-032, U.S. Environmental Pro-
tection Agency. Research Triangle Park, North Carolina, 1985a.
Jeffries, H. E. 1986. "User's Guide to Photochemical Kinetics Simulation
System. PC-PKSS." University of North Carolina, Chapel Hill. North
Carolina.
405
-------
Jeffries, H. E., and K. G. Sexton. 1987. "Technical Discussion Related
to the Choice of Photolytic Rates for Carbon Bond Mechanisms in
OZIPM4/EKMA. U.S. Environmental Protection Agency, Research Triangle
Park, North Carolina (EPA-450/4-87-003).
Jeffries. 1987. Personal communication with Harvey E. Jeffries, Univer-
sity of North Carolina, Chapel Hill, North Carolina.
Kaiser, E. W., and C. H. Wu. 1977. J. Phys. Chem.. 81:1701.
Kamens, R. M., H. E. Jeffries, K. G. Sexton, and A. A. Gerhardt. 1981.
"Smog Chamber Experiments to Test Oxidant Related Control Strategy
Issues." Department of Environmental Sciences and Engineering,
School of Public Health, University of North Carolina, Chapel Hill,
North Carolina.
Kan, C. S., J. C. Calvert, and J. H. Shaw. 1980. "Reactive Channels of
the CH302-CH302 Reaction." Department of Chemistry and Department of
Physics, Ohio State University, Columbus, Ohio.
Kan, C. S., F. Su, J. G. Calvert, and J. H. Shaw. 1981. Mechanism of the
ozone--ethene reaction in dilute N2/02 mixtures near 1-atm pres-
sure. J. Phys. Chem., 85:2359-2363.
Kelly, N. A. 1982. Characterization of fluorocarbon-film bags as smog
chambers. Environ. Sci. Technol.. 16(11):763.
Kerr, J. A., and J. G. Calvert. 1985. Chemical Transformation Modules
for Eulerian Acid Deposition Models—Volume I: The Gas-Phase Chemis-
try. EPA-600/3-85-015, National Center for Atmospheric Research,
Boulder, Colorado.
Killus, J. P. 1982. "Background Reactivity Estimates for Atmospheric
Modeling Studies." Systems Applications, Inc., San Rafael, Califor-
nia. XVth Informal Conference on Photochemistry, Stanford, Califor-
nia (27 June-1 July 1982).
Killus, J. P., and G. W. Whitten.. 1982. "A New Carbon-Bond Mechanism for
Air Quality Simulation Modeling." U.S. Environmental Protection
Agency, Research Triangle Park, North Carolina (EPA 600/3-82-041).
Killus, J. P., and G. Z. Whitten. 1982. A mechanism describing the
photochemical oxidation of toluene in smog. Atmos. Environ.,
16(8):1973-1988.
406
-------
Killus, J. P., and G. Z. WMtten. 1984. Isoprene: A photochemical
kinetic mechanism. Environ. Sci. Technol.. 18:142-148.
Kleindienst, T. E., G. W. Harris, and J. N. Pitts, Jr. 1982. Rates and
temperature dependences of the reaction of OH with isoprene, its
oxidation products, and selected terpenes. Environ. Sci. Technol..
16(12):844-846.
Ledbury, W.t and E. W. Blair. 1925. The partial formaldehyde vapour
pressure of aqueous solutions of formaldehyde. Part II. J. Chem.
Soc.t 127:2832-2839.
Lee, Y. N.t and S. E. Schwartz. 1981. J. Phys. Chem., 85:840.
Leone, J. A., R. C. Flagan, D. Grosjean, and J. H. Seinfeld. 1985. An
outdoor smog chamber and modeling study of toluene—NOX photooxida-
tion. International J. Chemical Kinetics. 17:177-216.
Leone, J. A.f and J. H. Seinfeld. 1984. Updated chemical mechanism for
atmospheric photooxidation of toluene. Int. J. Chem. Kinet., 16:159-
193.
Lurmann, F. W., W.P.L. Carter, and L. A. Coyner. 1987. "A Surrogate
Species Chemical Reaction Mechanism for Urban-Scale Air Quality Simu-
lation Models. Volume II. Guidelines for Using the Mechanism"
(EPA/600/3-87/014b).
Malko, M. W., and J. Troe. 1982. Analysis of the unimolecular reaction
N205 + M * N02 + N03 + M. Int. J. Chem. Kinet.. 14:399-416.
McAdam, K.t B. Veyret, and R. Lesclaux. 1987. UV absorption spectra of
HOo and CH302 radicals and the kinetics of their mutual reactions at
298 K. Chem. Phys. Lets., 133:39-44.
Moortgat, G. K. 1986. Personal communication with Geert Moortgat, Max-
Planck-Institute, Mainz, W. Germany.
Moortgat, G. K., F. Slemr, W. Seiler, and P. Warneck. 1978. Photolysis
of formaldehyde: Relative quantum yields of H2 and CO in the wave-
length range 270-360 nm. Chem. Phys. Lett., 54(3):444.
Moortgat, G. K., and P. Warneck. 1979. CO and H2 quantum yields in
the photodecomposition of formaldehyde in air. J. Chem. Phys.,
70(8):3639.
407
-------
NCAR. 1987. "Development and Implementation of Chemical Mechanisms for
the Regional Acid Deposition Model (RADM)." National Center for
Atmospheric Research, U.S. Environmental Protection Agency, Research
Triangle Park, North Carolina.
Niki, H., P. D. Maker, C. M. Savage, and L. P. Breitenbach. 1977. Fourier
transform ir spectroscopic observation of prophylene ozonide 1n the
gas phase reaction of ozone-cis-2-butane-formaldehyde. Chem. Phys.
Lets.. 46:327-330
Niki, H., P. D. Maker, C. M. Savage, and L. P. Breitenbach. 1978.
Mechanism for hydroxyl radical initiated oxidation of olefin-nitric
oxide mixtures in parts per million concentrations. J. Phys. Chem.,
82(2):135-137.
Niki, H., P. D. Maker, C. M. Savage, and L. P. Breitenbach. 1981. A FT
IR study of a transitory product in the gas-phase ozone-ethylene
reaction. J. Phys. Chem., 85:1024-1027-
Niki, H., P. D. Maker, C. M. Savage, and L. P. Brietenbach. 1982.
Fourier transform infrared studies of the self-reaction of ^302
radicals. J. Phys. Chem., 85:877-881.
Ohta, T., and T. Ohyama. 1985. A set of rate constants for the reactions
of OH radicals with aromatic hydrocarbons. Bull. Chem. Soc. Jpn.,
58:3029-3030.
Perner, D., A. Schmeltekopf, R. H. Winkler, H. S. Johnston, J. G. Calvert,
C.A. Cantrell, and W. R. Stockwell. 1985. A laboratory and field
study of the equilibrium N?05 * NO, + N0?. J. Geophys. Res..
90(D2):3807-3812. ^0*0
Peterson, J. T. 1976. "Calculated Actinic Fluxes for Air Pollution
Photochemistry Applications." U.S. Environmental Protection Agency,
Research Triangle Park, North Carolina (EPA-600/4-76-002).
Peterson, J. T., E. C. Flowers, G. J. Berri, C. L. Reynolds, and J. H.
Rudsill. 1981. Atmospheric turbidity over Centrol North Carolina.
J. Appl. Meteorol.. 20:229-241.
Pitts, J. N., Jr., K. R. Darnell, W.P.L. Carter, A. M. Winer, and
R. Atkinson. 1979. "Mechanisms of Photochemical Reactions in Urban
Air." U.S. Environmental Protection Agency, Research Triangle Park,
North Carolina (EPA-600/3-79-110).
408
-------
Pitts, J. N., Jr., A. M. Winer, and W.P.L. Carter. 1981. "Chemical Con-
sequences of Air Quality Standards and of Control Implementation Pro-
grams." California Air Resources Board (Contract No. A8-145-31).
Pitts, J. N., Jr., E. Sanhueza, R. Atkinson, W.P.L. Carter, A. M. Winer,
G. W. Harris, and C. N. Plum. 1984. An investigation of the dark
formation of nitrous acid in environmental chambers. Int. J. Chem.
Kinet., 16:919-939.
Pitts, J. N. 1983. "Chemical Consequences of Air Quality Standards and
Control Implementation Programs." California Air Resources Board
(ARB-R83-209).
Plum, C. N., E. Sanhueza, R. Atkinson, W.P.L. Carter, and J. N.
Pitts, Jr. 1983. OH radical rate constants and photolysis rates
of a-dicarbonyls. Environ. Sci. Techno!.. 17(8):479-484.
Saeger, M. 1977. "Experimental Measurements of Photolysis Rate of
Nitrogen Dioxide." M. S. Technical Report, University of North
Carolina.
Sakamaki, F.t S. Hatakeyama, and H. Akimoto. 1983. Formation of nitrous
acid and nitric oxide in the heterogeneous dark reaction of nitrogen
dioxide and water vapor in a smog chamber. Int. J. Chem. Kinet..
15:1013.
Sander, S. P., and C. C. Kircher. 1986. Temperature dependence of the
reaction NO + N03 - 2N02. Chem. Phys. Letters. 126(2):149-152.
Sander, S. P., and M. Peterson. 1982. Temperature dependence of the H02
+ N02 reaction and its relevance to tropospheric chemistry. 2nd Sym-
posium: Composition of the Nonurban Troposphere. American Meteoro-
logical Society, Boston, Massachusetts.
Schwartz, S. E. 1984. "Gas-Aqueous Reactions of Sulfur and Nitrogen
Oxides in Liquid-Water Clouds." In SOg, NO. and NOg Oxidation
Mechanisms: Atmospheric Consideration?. J. G. CalvSrt, ed., Butter-
worth, Boston (pp. 173-208).
Shepson, P. B., E. 0. Edney, and E. W. Corse. 1984. Ring fragmentation
reactions on the photoxidations of toluene and o-xylene. J. Phys.
Chem.. 4122-4126.
Smalley, J. F., F. L. Nesbitt, and R. B. Klemm. 1985. Branching ratio
for the hydrogen atom product channel in the reaction of ground-state
atomic oxygen with ethylene. J. Phys. Chem.. 90:491-497.
409
-------
Stedman, D. H., R. B. Harvey, and R. R. Dickerson. 1977. "Measurement of
Photons Involved 1n Photochemical Oxidant Formation." Int. Conf. on
Photochem. Oxidant Pollution and Its Control, Proceedings, 2:753-762
(EPA-600/3-77-0016).
Su, F., J. G. Calvert, J. H. Shaw, H. Niki, P. D. Maker, C. M. Savage, and
L. D. Breitenbach. 1979a. Spectroscopic and kinetic studies of a
new metastable species in the photooxidation of gaseous formal-
dehyde. Chem. Phys. Lets., 65:221-225.
Su, F., J. G. Calvert, and J. H. Shaw. 1979b. Mechanism of the photooxi-
dation of gaseous formaldehyde. J. Phys. Chem., 83(25):3185-3191.
Su, F., J. G. Calvert, and J. H. Shaw. 1980. A FTIR spectorscopic study
of the ozone-ethene reaction mechanism in Oo-rich mixtures. J. Phys.
Chem., 84:239-246.
Tuazon, E. C., H. MacLeod, R. Atkinson, and W.P.L. Carter. 1986.
a-Dicarbonyl yields frm the N0x-air photooxidations of a series of
aromatic hydrocarbons in air. Environ. Sci. Technol., 20(4):383.
Veyret, B.. J. Rayez, and R. Lesclaux. 1982. Mechanism of the
photooxidation of formaldehyde studied by flash photolysis of
CH2-0-02-NO Mixtures. J. Phys. Chem., 86:3424-3430.
Whitten, G. Z., J. P. Killus, and R. G. Johnson. 1983. "Modeling of
Simulated Photochemical Smog With Kinetic Mechanisms" (EPA-600/3-
83/043).
Whitten, G. Z., D. R. Souten, M. W. Gery, and R. D. Edmond. 1986.
"Potential Impact of Ethanol Emissions on Urban Smog: A Preliminary
Modeling and Smog Chamber Study." Systems Applications, Inc., San
Rafael, California (SYSAPP-86/090).
Whitten, G. Z., J. P. Killus, and R. G. Johnson. 1985a. "Modeling of
Auto Exhaust Smog Chamber Data for EKMA Development." Systems Appli-
cations, Inc., San Rafael, California (EPA-600-3/85-025).
Whitten, G. Z., J. P. Killus, and R. G. Johnson. 1985b. "Development
of a Chemical Kinetic Mechanism for the U.S. EPA Regional Oxidant
Model." U.S. Environmental Protection Agency, Research Triangle
Park, North Carolina (EPA-600/3-85-026).
410
-------
WhHten, G. Z., and M. W. Gery. 1986. "Development of CBM-X Mechanisms
for Urban and Regional AQSMs." U.S. Environmental Protection Agency,
Research Triangle Park, North Carolina (EPA-600/3-86-012).
Whitten, G. Z., and H. Hogo. Mathematical Modeling of Simulated Photo-
chemical Smog. EPA-600/3-77-011, U.S. Environmental Protection
Agency, Research Triangle Park, North Carolina, 1977.
Whitten, G. Z., H. Hogo, and J. P. Killus. The Carbon-Bond Mechanism: A
Condensed Kinetic Mechanism for Photochemical Smog. Environ. Sci.
Technol., 14:699, 1980a.
Whitten, G. Z., J. P. Killus, and H. Hogo. Modeling of Simulated Photo-
chemical Smog with Kinetic Mechanisms. EPA-600/3-80-028a, U.S.
Environmental Protection Agency, Research Triangle Park, North Caro-
lina, 1980b.
Whitten, G. Z. The Chemistry of Smog Formation: A Review of Current
Knowledge. Environ. Int., 9:447-463, 1983.
Whitten, G. Z., H. Hogo, M. J. Meldgin, J. P. Killus, and P. J.
Bekowies. 1979. "Modeling of Simulated Photochemical Smog With
Kinetic Mechanisms" (EPA-600/3-79-001a).
Winer, A. M., K. R. Darnall, R. Atkinson, and J. N. Pitts, Jr. 1978.
J. Phys. Chem. Ref. Data, 13(2):1984.
Zafonte, L., P. L. Rleger, and J. R. Holmes. 1976. "Nitrogen Dioxide
Photolysis in the Los Angeles Atmosphere." Californa Air Resources
Board, El Monte, California (Publication No. DTS-76-18).
411
------- |