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

-------