EPA-650/4-75-026

June 1975
Environmental Monitoring Series
              MATHEMATICAL MODELING
                          OF SIMULATED
                 PHOTOCHEMICAL SMOG
                                   P0

                                     ID

-------
                               EPA-650/4-75-026
MATHEMATICAL  MODELING
         OF  SIMULATED
   PHOTOCHEMICAL  SMOG
                  by

Paul A. Durbin, Thomas A. Hecht, and Gary Z. Whitten

          Systems Applications, Inc.
            950 Northgate Drive
         San Rafael, California 94903
           Contract No. 68-02-0580
            ROAP No. 21AKC - 23
         Program Element No. 1A1008
      EPA Project Officer: Marcia C. Dodge

       Chemistry and Physics Laboratory
     National Environmental Research Center
    Research Triangle Park, North Carolina 27711
               Prepared for

   U.S. ENVIRONMENTAL PROTECTION AGENCY
       Office of Research and Development
          Washington, D. C.  20460

                June 1975

-------
                         EPA REVIEW NOTICE

 This report has been reviewed by the National Environmental Research
 Center - Research Triangle Park , Office of Research and Development,
 EPA, and approved for publication.  Approval does not signify thru the
 contents necessarily reflect the views and policies of the Environmental
 Protection Agency, nor does mention of trade names or commercial
 products constitute endorsement or recommendation for use.
                    RESEARCH REPORTING SERIES

 Research reports of the Office of Research and Development, U.S. Environ-
 mental Protection Agency, have been grouped into series. These broad
 categories were established to facilitate further development and applica-
 tion ol environmental technology   Elimination of traditional grouping was
 consciously planned to foster technology transfer and maximum interface
 in related fields. ThfSr series are:

           1.  ENVIRONMENTAL HEALTH EFFECTS RESEARCH

           2.  ENVIRONMENTAL PROTECTION TECHNOLOGY

           3.  ECOLOGICAL RESEARCH

           4.  ENVIRONMENTAL MONITORING

           5.  SOCIOECONOMIC ENVIRONMENTAL STUDIES

           6.  SCIENTIFIC AND TECHNICAL ASSESSMENT REPORTS
           9.  MISCELLANEOUS

 This report has been assigned to the ENVIRONMENTAL MONITORING
 series.  This series describes research conducted to develop new or
 improved methods and instrumentation  for the identification and quanti-
 fication of environmental pollutants at the lowest conceivably significant
 concentrations.  It also includes studies to determine the ambient concen-
 trations of pollutants in the environment and/or the variance of pollutants
 as a function of time or meteorological factors.
This document is available to the public for sale through the National
Technical Information Service, Springfield, Virginia 22161.

                Publication No. EPA-650/4-75-026
                                 11

-------
                                                                            n
                              PREFACE
     As part of its program to clarify the roles  of organic  compounds  and
oxides of nitrogen in the production of photochemical  smog,  the  U.S. Environ-
mental Protection Agency (EPA) is  supporting the  study of irradiation-
induced air pollution in environmental chambers and the determination  of
the rate constants and mechanisms  of elementary reactions thought  to be
important in smog formation.  To complement this  experimental  effort,  the
EPA is sponsoring SAI's work on the development of a chemical  kinetic  mech-
anism for photochemical smog formation.  This mechanism, incorporating ex-
perimentally measured rate constants, is presently being compared  with data
obtained from smog chamber experiments.  Ultimately, the mechanism should
be capable of predicting the kinetics of the chemical  transformations  that
take place in photochemical smog.   Our initial efforts to formulate and  eval-
uate a kinetic mechanism for photochemical smog formation were summarized  in
a detailed planning document (Seinfeld et al., 1973),  in a 1973  final  report
(Hecht et al., 1973), and in a 1974 final report  (Hecht et al.,  1974a).

-------
                                                                         in
                              ABSTRACT

     This report deals  with  the  continued development and testing of a
kinetic mechanism for photochemical  smog formation.  An  "explicit" mechanism
is developed and validated on  smog  chamber  data  taken at the University of
California at Riverside,  Battelle,  and  the  National Air  Pollution Control
Association.  After critical  review of  recent  observations of  reactions be-
lieved to occur in photochemical  smog,  a mechanism  is developed for each of
the following systems:   propylene-NOx,  butane-NCy  propylene-S02-NOx,  and
toluene-NO  .  In addition,  the report demonstrates  that  some chamber effects,
such as photolysis and surface reactions,  potentially play a critical  role  in
smog chamber experiments.  Finally, the report discusses the application of
kinetic simulation to a study of hydrocarbon reactivity  and  ozone  production
in smog systems.

-------
                               CONTENTS


PREFACE .  .  .	    IT

ABSTRACT	iji

LIST OF ILLUSTRATIONS	    vi

LIST OF TABLES	     x

  I  INTRODUCTION 	
                                                                         1
     A.   Smog Chamber Simulation and  Elementary
         Reaction  Kinetics  	
                              	     2
     B.   Explicit Mechanistic Approach  	     2
     C,   Hydrocarbon Reactivity  	     .

 II  CHEMICAL KINETICS  	    12

     A.   Inorganic Chemistry  	    12

         1.   Heterogeneous  HN02  Chemistry  	    12
         2.   Heterogeneous  HNOs  Formation  	    17
         3.   03 Decay	    19

     B.   Organic Chemistry   	    19
         1.   Propylene-OH-  Reactions   	    19
         2.   Unimolecular Decomposition  of AUoxyl  Radicals	    22
         3.   Alkoxyl  Radical-02  Reactions  	    23
         4.   Propylene-Oa Reactions	    23
         5.   Radical-Radical  Reactions  	    25
         6.   PAN Chemistry	    26
     C.   Photochemistry	    27
         1.   Photolysis Rate Constants	    27
         2.   Spectrum Decay	    29

     D.   S02  Oxidation	    31

         1.   Some Observations	    34
         2.   Kinetic Mechanism for S02 Oxidation   	    35
     E.   The  Toluene-NO -Air System	    40
         1.   Toluene + 0(3P)	    40
         2,   Toluene + OH-   	    40
         3.   The Proposed Mechanism	    41

-------
Ill  SMOG CHAMBER SIMULATIONS	45

     A.   Simulations  of UCR  Data	45

         1.   Propylene-NOx-  Air System	46
         2.   Butane-N0x-Air  System  	  68

     B.   Simulations  of Battelle  Data	87

         1.   Instrumentation	87
         2.   Mechanism Used	92
         3.   Results  and Discussion	92

     C.   Simulations  of the  NAPCA Toluene-N0x  Data	105

         1.   Mechanics Used	•	105
         2.   Results  and Discussion  	 105

 IV  HYDROCARBON REACTIVITY  	 114

     A.   Survey of Reactivity  Measures	114

         1.   Temporal Measures  	 H^
         2.   Concentration Measures	H5
         3.   Combined Temporal and Concentration Measures 	 116

     B.   Measure Assessment 	 118

         1.   Scope and Procedure	H8
         2.   Measure Study	121
         3.   The Measures Selected  	 127
         4.   Mixture Study  	 130
         5.   Derivation of Some Properties of T^	133

     C.  Relation of the Above Considerations to
         Ozone  Production	142

         1.   Ozone Isopleths  	  142
         2.   Chemical  Dynamics   . . .	150

   V  CONCLUDING REMARKS  	  151

 REFERENCES	-153

-------
                                                                          VI
                              ILLUSTRATIONS
  1    Smog Profiles for Different Values of kio, the Rate Constant
      for the Reaction 2H20 + NO + NO? + 2HN02 + HgO .  .  .  .  ......    15
  2    Smog Profiles for Different Initial Concentrations  of HNCu ....    16
  3    Smog Profiles for Different Values of kg, the Rate  Constant
      for the Heterogeneous Formation of HMOs, NzOs + H20 •+ 2HN03  •  •  •    18
 4.   Spectra of Sunlight and U.C. Riverside Solar Simulator ......    28
 5    Effect of a 20 nm Filter Shift on Smog Profiles  .........    32
 6    Effect of a 20 nm Filter Shift on N02 Behavior for EC-60 .....    33
 7    Propylene-N0v Factorial Block  ..................    47
                  A
 8    EC-11 Simulation Results and UCR Data for N09, 0,, and NO  ....    52
                                                  C.   O
 9    EC-11 Simulation Results and UCR Data for Propylene
      and Formaldehyde .........................    53
10    EC-11 Simulation Results and UCR Data for Acetone and PAN  ....    54
11    EC-11 Simulation Results and UCR Data for Acetaldehyde ......    55
12    EC-12 Simulation Results and UCR Data for 03, NO, and N0«  .  .  .  .    56
13    EC-12 Simulation Results and UCR Data for Propylene
      and Formaldehyde .........................    57
14    Simulation Results and UCR Data for PAN and Acetaldehyde .....    58
15    EC-16 Simulation Results and UCR Data for 03> NO, and N02  .  .  .  .    59
16    EC-16 Simulation Results and UCR Data for Acetone,
      Acetaldehyde, and PAN  ......................    60
17    EC-16 Simulation Results and UCR Data for Propylene
      and Formaldehyde .........................    61
18    EC-18 Simulation Results and' UCR Data for Propylene
      and Acetaldehyde .........................    62
19    EC-18 Simulation Results and UCR Data for NO, N02> and PAN  ....    63

-------
                                                                         vii
20    EC-18 Simulation Results  and  UCR  Data  for  03  and  Formaldehyde  ...   64
21    EC-21 Simulation Results  and  UCR  Data  for  Propylene
      and Acetaldehyde  	   65
22    EC-21 Simulation Results  and  UCR  Data  for  NO, N02>
      and Formaldehyde	   66
23    EC-21 Simulation Results  and  UCR  Data  for  03  and  PAN	   67
24    n-Butane/NOv Factorial  Block   	   71
                 A
25    EC-39 Simulation Results  and  UCR  Data  for  Og, NO, and  N02  	   72
26    EC-39 Simulation Results  and  UCR  Data  for  Butane	   73
27    EC-39 Simulation Results  and  UCR  Data  for  Acetaldehyde
      and Formaldehyde	   74
28    EC-39 Simulation Results  and  UCR  Data  for  MEK and PAN	   75
29    EC-41 Simulation REsults  and  UCR  Data  for  03> NO, and  N02  	   76
30    EC-41 Simulation Results  and  UCR  Data  for  Butane	   77
31    EC-41 Simulation Results  and  UCR  Data  for  MEK and PAN  .......   78
32    EC-41 Simulation Results  and  UCR  Data  for  Acetaldehyde
      and Formaldehyde	    79
33    EC-42 Simulation Results  and  UCR  Data  for Butane, NO,  and  N02 .  .  .    80
34    EC-42 Simulation Results  and  UCR  Data  for MEK, Formaldehyde
      and Acetaldehyde  	    81
35    EC-42 Simulation Results  and  UCR Data for Ozone	    82
36    EC-44 Simulation Results  and  UCR Data for NO, N02, and
      and Acetaldehyde	    83
37    EC-44 Simulation Results and UCR Data for Ozone	    84
38    EC-44 Simulation Results and UCR Data for Butane	    85
39    EC-44 Simulation Results and UCR Data for Formaldehyde
      and MEK	    86
40    EC-44 Simulation Results, Using k]0 = 1-3 x  10~12ppnf3 nrin"
            0.024 ppnH min"1, for NO and N02
41     EC-44 Simulation Results, Using kiQ = 1.3 x 10~12ppnf3 mirT1
       and kn = 0.024 ppnH mitr1, for Ozone	   89

-------
42    EC-44 Simulation Results,  Using  k10  =  1.3  x  10  12  ppm~3 min"1
      and kn  = 0.024 ppm"1  min"1,  for Butane  .............   90

43    EC-44 Simulation Results,  Using  k]0  =  1.3  x  10"12  ppm"3 min"1
      and k-|-|  = 0.024 ppm-"1  min-',  for MEK,  Acetaldehyde,
      and Formaldehyde  ........................   91
44    S-107 Simulation Results and Battelle Labs Data for
      Propylene, N02, and S02 ................  .....    95

45    S-107 Simulation Results and Battelle Labs Data for
      03, NO, and SOg Aerosol .....................    95

46    S-110 Simulation Results and Battelle Labs Data for
      Propylene, N02, and S02 .....................    97

47    S-110 Simulation Results and Battelle Labs Data for
      03, NO, and S03 Aerosol .....................    98

48    S-113 Simulation Results and Battelle Labs Data for
      Propylene, N02, and S02 .....................    "

49    S-113 Simulation Results and Battelle Labs Data for
      03, NO, and S03 Aerosol .....................   10°

50    S-114 Simulation Results and Battelle Labs Data for
      Propylene, 03, NO, and  N02   ...................   101

51    S-115 Simulation Results and Battelle Labs Data for
      Propylene, 03> NO, and  N02   ...................   "°2

52    EPA-258 Simulation Results  and NAPCA  Data  for  NO,  N02,  and 03  . .  107

53    EPA-258 Simulation Results  and NAPCA  Data  for  Toluene  ......  108

54    EPA 272 Simulation Results  and NAPCA  Data  for  NO,  N02,  and 03  . .  109

55    EPA-272 Simulation Results  and NAPCA  Data  for  Toluene  ......   110

56    EPA-305 Simulation Results  and NAPCA  Data  for  NO,  N02>  and 03  . .   Ill

57    EPA-305 Simulation Results  and NAPCA Data  for  Toluene  ......   112

58    The Use of Kinetic Simulations to Assess  Reactivity .......   120

59    Typical Smog Profile  ......................   136

 60    T  as a  Function of  Initial Hydrocarbon Concentration  ......   139
        m
 61     Lines of  Constant 03 (in ppm)  After 1 Hour of  Simulation  ....   143

 62    Lines of Constant 03 (in ppm)  After 2 Hours of Simulation ....   144

-------
63    Lines of Constant 0., (in ppm)  After 5 Hours of Simulation	145
                         O      '


64    Lines of Constant 0  (in ppm)  After 8 Hours of Simulation 	   146



65    Lines of Constant 03 (in ppin)  After 9 Hours of Simulation	147



66    Time of the N09 Peak (in Minutes)	148

-------
                                 TABLES
  1     Summary of Reactions  	    5
  2     Summary of Rate Constants   	    8
  3     Photolysis Constant Changes from UV Loss	   30
  4     Rate Constants for S02 Oxidation	   36
  5     UCR Propylene-N0y Experiments:  Initial Concentrations
       of Primary Reactants  	   47
  6     The Propylene Oxidation Mechanism  	   48
  7     Changes Made in Mechanism for Butane Simulations .  	   69
  8     UCR Butane-N0x Experiments:  Initial Concentrations
       of Primary Reactants and Values of k,   	   71
  9     Analytical Characteristics of Battelle Experimental
       Setup	\   93
10     S02 Oxidation Mechanism	   94
11    Battelle Propylene-N0x-S02 Experiments:  Initial
      Concentrations and Values of k,  	  103
12    Rates of S02 Oxidation by Various Oxidants
       (from S-107 Simulation)  	  104
13    Toluene Oxidation Mechanism  	  106
14    NAPCA Toluene-N0x Experiments:   Initial Concentrations ....  113
15    Definitions of Reactivity Measures	  123
16     Results of the Measure Study:  Reactivities Relative
      to Propylene	124
17     Initial  Concentrations for Experiments Listed in Tables
      Tables 16  and 18	125
18    Results of the Mixture Study:  Mixture Reactivities
      Relative to Propylene  	  131

-------
                          I    INTRODUCTION
     The problem of  photochemical air pollution has received considerable
attention from scientists,  legislators, and  the public during the past two
decades.  Although  the  problem was  originally manifested  in Los Angeles, it
has become increasingly evident  in  other  urban and even nonurban areas
and is worthy of widespread concern.  As  its name indicates, a fundamental
characteristic of photochemical  air pollution is the  role played by  sunlight-
initiated chemical  transformations.  Primary pollutants are often hazardous
themselves; however, these  secondary chemical processes greatly exacerbate
the problem.  They lead to  the  production of phytotoxicants,  lacrymators,  and
carcinogens.

     The present document reports on the  continuation of  an  ongoing  research
effort  to isolate and model the complex chemical  reactions that occur in
polluted atmospheres.  This effort is aimed at  producing  a photochemical
kinetic mechanism for smog formation having sufficiently  accurate  kinetics
to provide  realistic predictions of pollution  production  while, at the same
time, being simple enough for practical use in large computer models of urban
airsheds.   Past efforts (Hecht et  al., 1973; Hecht et al., 1974a)  toward
these ends  have resulted in the development of a "generalized" kinetic mech-
anism.   The generalized mechanism  has proved to have great utility, making
urban airshed  modeling feasible  (Reynolds et al., 1974).   However,  in the
present report, a digression from  the generalized approach has been made.   To
take  full  advantage of smog chamber data  and kinetic studies, we employed an
 "explicit"  mechanism.  The explicit approach and the reasons for its  incor-
poration are  discussed below,  after a  review of some background information.
A final topic, the  application  of  the  kinetic mechanism  to an  investigation
 of hydrocarbon reactivity, is  discussed  at  the end of this introduction.

-------
A.   SMOG CHAMBER SIMULATIONS AND ELEMENTARY  REACTION  KINETICS

     Smog chamber investigations  and elementary  kinetics  studies provide
the basic inputs to the present kinetic  mechanism.   In smog  chamber  studies,
clean air and pollutants (usually NO, N02,  and a hydrocarbon) are  irradiated
in a reactor.  Measurements  of reactant  and product  concentrations as  a func-
tion of time provide smog profiles,  which  the kinetic  mechanism should repro-
duce.  These profiles constitute  quantitative and qualitative descriptions  of
the macroscopic features of  smog.  Thus, the  conversion of NO to N02,  the  oxi-
dation of hydrocarbons, and  the production  of oxidants, as observed  in the  at-
mosphere, are reproduced in  a controlled laboratory  environment.   But  smog
chamber studies cannot provide a  knowledge  of the microscopic features of  smog
formation.  Independently, kineticists study  elementary reactions  that could
be important in this process.

     Kinetics studies are performed to elucidate the rates and  mechanisms  of
particular reactions that occur within the  overall  process of smog formation.
These studies differ from smog chamber studies in that the reactants and
other conditions are carefully chosen to isolate or  emphasize the  reactions
of interest.  The observed variations in reactant and product concentrations
with time are used to formulate rate equations and rate constants.  If com-
plicating reactions are absent or well characterized, these rate  equations
and constants have universal validity.  Mechanistic information conies from
the rate expression as well  as from observed  reaction products.

     Ideally, a kinetic mechanism would be simply the assemblage  of results
from kinetic studies of all  reactions that occur.  In reality,  not all of
the reactions that could occur have been studied, and often orders of magni-
tude of uncertainty may be associated with those that have been studied.   In
addition, the number of possible reactions in smog is very large.   Hence, a
complete mechanism is neither practical, because it would include an enormous
number of reactions, nor feasible, because the needed kinetic information is
unavailable.  The best one can do to obtain a closed system of chemical  equa-
tions and rate constants is to use available kinetic information  and methods

-------
for estimating other information,  to  draw  analogies, and  to make  other sim-
plifying assumptions.   The kinetic mechanism  that  results may  be  a  fairly
accurate, albeit simplified,  description of reality.   The system  of equations
constituting the current mechanism has  largely  been discussed  previously
(Hecht and Seinfeld, 1972; Hecht et al., 1974b).   However, several  changes
and additions have been made  since last year's  report.   These  new features,
and the kinetic studies supporting them, are  reviewed  in  Chapter  II of the
present report.

     In addition to the reactions characterized by universal  kinetic expres-
sions, there are processes, both physical  and chemical,  occurring in smog
chambers that are peculiar to a given chamber.   These  chamber  effects include
surface-catalyzed reactions,  adsorption and desorption of chemicals, dilution
due to sampling, inhomogeneous concentrations,  and spatial and spectral  varia-
bility of light sources.  Of the surface  reactions,  probably  the  most important
is the heterogeneous formation of nitrous  and nitric  acid.  The heterogeneous
production of HNOX is investigated in Chapter II.   A possible  spectral
variation of the light source used in smog chamber simulations done at the
University of California at Riverside is  also considered from a theoretical
standpoint.  The results show the critical influence of surface and light
properties on smog profiles and point to a need for more detailed character-
izations of smog chambers.

      In Chapter III, predictions of smog profiles made by numerical simulation
are compared with experimental data.   To model the smog chamber results com-
pletely, we had to incorporate some of the chamber effects mentioned above.
The techniques for doing  so are described in Chapter III.

B.    EXPLICIT MECHANISTIC APPROACH

      The most  straightforward approach to formulating a kinetic mechanism
is  to  assemble the most  important  reactions  that occur.  The mechanism's com-
plexity  is  then dictated  by the criterion for  "important."  A  technique des-
cribed by  Hecht et  al.  (1974b)  even  further  simplifies the formulation.  In
this  technique, groups  of reactions  are "lumped"  into single  reactions; these
reactions  are  then  combined to  form  a  generalized mechanism.   But  the lumping

-------
approach introduces nonphysical  lumped rate  constants,  stoichiometric  co-
efficients, and chemical  species.   Because of this  last feature,  the predic-
tion of detailed product  yields  is  not feasible,  and  because  of the first
two, kinetic information  cannot  be  used directly.   Thus,  we used  the classi-
cal, explicit approach instead,  to  take advantage of  recent kinetic studies
and the full range of product measurements available.

     Formulating the explicit mechanism involves  simply expanding the
lumped reactions of the generalized mechanism.   Conversely, reformulation
of a generalized mechanism involves contracting the explicit  mechanism.
Thus, in the present approach, uncertainties associated with  the  lumping
procedure have been eliminated,  while a kinship with  the generalized mech-
anism has been retained.   After  validation,  the explicit mechanism can then
serve as a basis for rederiving  a  generalized mechanism and for checking
the accuracy of lumping techniques.  We should emphasize that the explicit
approach is an interim step.   We hope that it will  serve to further clarify
smog kinetics and thus to lay the  groundwork for the  more practical general-
ized approach.

     A review of the reactions and  rate constants used in the mechanisms
contained herein appears  in Tables  1 and 2.   The lumped format (i.e.,  use
of the R group) was used  to represent a class of reactions, which appear
explicitly in the later mechanisms  (Chapter  III).

C.   HYDROCARBON REACTIVITY
           i
     Pollution control strategists  must know the reactivity  of various hydro-
carbons to predict the potential impact of different  emissions sources.   The
conceptualization of reactivity  can, however, take  various forms.  Corres-
pondingly, there are many ways of quantifying reactivity in  smog  systems.
Probably the most obvious and commonplace measure of  reactivity is the time
required for N0£ to reach its peak concentration, because the NOg peak is  a
distinctive feature of smog profiles.  Many other measures have appeared  in

-------
                             Table 1
                      SUMMARY OF REACTIONS

;.    Inorganic H-N-0 Compound Chemistry
     A.   The N02-NO-03  Cycle
         1.   NOe  + hv  + NO + 0
         2.   0+02+M+03+M
         3.   03 + NO + N02 + 02
     B.   Other  NOX Chemistry
         4.   N02  + 0 -> NO +  02
         5.   N02  + 03  •+ NOs  +  02
         6.   NOs  + NO  •*• 2N02
         7.   NOs  + N02 -> N205
          8.    N205 -1  NOs + N°2
          9.    H20 + N205  surface> 2HN03
     C.   HNO? Chemistry
          10.  NO + N02 + 2H20  surface> 2HN02 + H20
          11.  2HN02  surface> NO + N02 +
          12.  HN02 + hv •* NO + OH-
     D.   OH- and  HO^ Reactions with  NOX
          13.  OH- +  NO -1 HN02
          14.  OH- +  N02 -^ HN03
          15.   H02 +  NO -> N02 +  OH-
 11.  03  Inorganic  Chemistry
           16.   03  + hv  + 02 + 0(3P)
           17.   03  + hv  -»- 02 + 0(1D)

-------
          18.  0(]D) + M + 0(3P)



          19.  OpD) + H20 -> 20H-



          20.  03 + OH- + H02 + 02



          21.  03 + H02' -> OH- + 202



          22.  03  surface>  products



III.  Organic Oxidation Reactions



     A.    Butane

                          °2
          23.  C4H10 + 0 -£*- R02' + OH-



          24.  C4Hio + OH-  — ^ n-R02 + H20



          25.  C4H10 + OH-  —2+- s-R02 + H20
     B.    Propylene


          26.   CH  + OH-  •> n-RO-
          27.   C3H6 + OH-  -*• s-RO-


                          °2
          28.   C3H6 + 03 —=*-'CH3CHO + H02 + OH- + CO



          29.   C3H6 + 03 -^H2CO + CH3C(0)02 + OH-



          30.   C3H6 + 0 —£- R02 + RC(0)02


                         °2
          31.   C3H6 + 0 —V R02 + H02 + CO



     C.    A1dehydes


          32.   RCHO + OH-  -X- RC(0)02 + H20



     D.    Toluene

                              02

          33.   C6H5CH3 + OH- 	*~ C6H5CH202 + H20



          34.   C5H5CH3 + OH- -^ C6H4(CH3)(OH) + H02



          34a.  C6H5CH3 + OH- —^



     E.    Organic Radicals

                    CL

          35.   RO-  —^ R02 + H2CO
          36.   RO-  + 02 	^ ALD + H02

-------
IV.   Other Photolysis Reactions
          37.   H202 + hv -v 20H-
                          °2
          38.   RCHO + hv — ^R02 + CO + H02
          39.  RCHO + hv + paraffin + CO
                          202
          40.  H2CO + hv — H-CO + 2H02'
          41 .  H2CO + hv -*- H2 + CO
V.   Reactions of Organic Free Radicals with NOX
     A.   NO Oxidation
          42.  R02 + NO + RO- + N02
                             °?
          43.  RC(0)02 + NO —+ R02 + C02 + N02
     B.   PAN Chemistry
          44.  RC(0)02 + N02 -> RC(0)02N02  (PAN)
          45.  PAN --*-N03 + C02 + R02
VI.  Radical -radical Recombination Reactions
          46.  H02 + H02 ^ H202 + 02
          47.  H02' + R02 -v R02H + 02
          48.  RC(0)02 + H02 ->• RC(0)02H
VII. SO? Chemistry
     A.   SO? Oxidation
          49.  N03 + S02 •> SOs + N02
          50.  H02 + S02 -»• SOs + OH-
          51.  R02 +  S02 -> SOa  + RO-
                              Op
          52.  RC(0)02  + S02  — ^S03 + C02 + R02
                           °2
          53.  S02 +  OH-  —
           54.   HSOs  +  NO -»• HS04 + N02
           55.   HS04  +  H02 -v HoS04 + 02
           56.   HS04  +  N02 -^ H2S04 + HN03

-------
         Table 2
SUMMARY OF RATE CONSTANTS
(ppnH rnin-1 unless noted)
Rate Constant
Reaction Number
1
2 (ppm-2 min-1)
3
4
5
6
7
8 (min-1)
9
10 (ppnT3 min"3)
•11
12
13
14
15
16
17
Garvin and
Hampson (1974)
2.08 x TO'5
25.2
1.34 x 104
5.0 x ID'2
1.3 X 104
5.6 x 103
21.9
< 1.5 x 10-5C
< 10-13C


2.95 x 103
1.2 x 104
2.95 x 102


Uncertainty
Factor This Study3
Experimental
1.2 2.08 x 10'5
1.3 25.2
1.2 1.34 x 10*
1.3 5.0 x ID'2
5.0 1.3 x 104
2.5 5.6 x 103
2.0 24.0
5.0 x 10'6
1.3 x 10-11
• 2.6 x 10-1
Experimental
2.0 3.0 x 103
2.5 l.OxlO4
3.2 8.0 x 102
Experimental
Experimental
Others

2.04 x 10"5
(Wu and Nlki, 1975)
27.0
(Wu and Niki, 1975)
1.39 x 104
(Wu and Niki, 1975)
6.8 x 10-2
(Wu and Niki, 1975)
1.5 x 104
(Wu and Niki, 1975)
4.5 x 103
(Wu and Niki, 1975)
ks/kg = 4.2 x TO'3
(Benson, 1968)b

10-10 . 10-12
(Noch et al., 1974)
ho H20/k11 = 9.7 x
(Demerjian et al.,

8.9 x 103
(Cox, 1974)

•v, 1.2 x 103
(Mabey and Hendry,













10-7
1974)



1974)



-------
Rate Constant
Garvin and
Reaction Number Hampson (1974)
18 8.6 x 104d
19 5.17 x 105
20 8.71 x 106
21 2.4
22
23
24, 25 3.47 x 103
26, 27 2.14 x 104
28, 29 0.02
30, 31 5.3 x 103
32 RCHO; R=0,l
RCHO; R=2,3
33,34
35 (nrirr1)
36 ' -v 4.4 x 10-3
37-41
42, 43
44
45
46 8.5 x 103
47, 48
49
Uncertainty
Factor This Study3
1.4 8.6 x 104
1.3 5.1 x 105
2.0 8.7 X 101
2.0 2.4
Experimental
64.0
1.2 3.4 x 103
1.2 2.5 x 104
0.02
1.2 5.3 x 103
2.1 x 104
4.5 x 104
9.2 x 103
0.6 - 1.3 x 104e
0.04 - 0.2e
Experimental
103 (estimate)
3 x 102 (estimate)
3 x ID'3
2.0 6.0 x 103
3 x 103 (1/2 k44)
14.0
Others





12.6
(Johnson et al., 1970)

2.5 x 104
(Morris et al., 1971)
0.026
(Becker et al . , 1974)

2.3 x 104
(Morris and Niki, 1971;
Morris et al . , 1971)
4.6 x 104
(Morris and Niki, 1971)
9.0 x 103
(Davis, 1974)
0.8 - 12.0 x 104
(Batt et al., 1974)
^ 0.8
(Mendenhall et al., 1974)f






< 14.7
                   (Davis, 1974)

-------
                                                                                             10
                                                  Rate Constant
                    Garvin and     Uncertainty
Reaction Number   Ijampson (1974)     Factor

50
51, 52

53


54


55

56
   This Study6

1.3


1.5 (estimate)

9 x 102
8 x 102
(estimate)

9 x 103 (3/2

1 x 104
(estimate)
         Others
                                                                     1.35
                                                                     (Davis, 1974)
1.35 x 103
(Davis, 1974)
a At 303°K

b Calculated using Keq ^ expf-AGjgg/RT),  iG from Benson (1966}

c Surface-dependent

^ Combination of values for M = 02 and M  = N£

e Depends on carbon skeleton

f Combination of BuO-  and MeO- data

-------
                                                                          11
the literature.   A review of these reactivity  indices  and  an  evaluation  of
the properties of some of them appear in Chapter IV of this  report.

     Ultimately, reactivity should be related  to the production of harmful
components of photochemical air pollution.   For this reason,  we investi-
gated the relationship between reactivity measures and ozone production.  In
this report, ozone yield is shown to be relatively insensitive to reactivity
for a group of olefins, though it does depend on initial NOX and hydrocarbon
concentrations.

     An important aspect of photochemical smog production in the atmosphere
is its dynamic behavior.  An understanding of the  interaction of the time
scales for reaction,  transport, and dispersion  is  necessary  to master the
pollution problem.  The dynamics  of smog formation are, to a significant
extent, determined by the  time required for N02  to reach  its peak.  Chapter  IV
investigates  this interrelationship.

-------
                                                                          12
                     II   CHEMICAL  KINETICS
A.   INORGANIC CHEMISTRY

     An excellent review of inorganic (and some organic)  reactions  occurring
in the stratosphere has been prepared by Garvin and Hampson (1974).   A
majority of the inorganic reactions thought to occur in polluted atmospheres
can also be found in that review.  The rate constant rec&mmendations  made by
Garvin and Hampson have generally been adhered to in the present study.   The
few exceptions to this rule are cases where the results of more recent studies
have become available or where the reactions are surface dependent.  The
latter set of reactions is considered below because of their possible signifi-
cance  (and elusiveness) in smog chamber simulations.

1.   Heterogeneous HN02 Chemistry

     Of  the reactions  that can take  place  on  smog  chamber walls, perhaps
the most significant  is the heterogeneous  formation of HN02:

            2H20 + NO + N02   surface> 2HN02 + HgO                  (10)*

The  occurrence  of this reaction  following the introduction of  NOX  reactants
 could  produce significant amounts  of HN02- OH radicals  produced by  HN02
 photolysis  play a major role  early in the reaction.   Thus, the occurrence
 of Reaction (10) can have a profound effect on the length of the  induction
 period.   It is important to note that Reaction (10)  is a thermal  reaction
 and that it can occur in the approximately 45 minute  dark mixing  time allowed
 at the beginning of Riverside experiments.  During this time,  Reaction (10)
 will compete with Reaction (11):
 * This report uses two numbering systems.  Chemical reactions used in modeling
   have simple Arabic numbers taken from Table 1 and previous work (Hecht et al.,
   1974b).  Other chemical reactions and mathematical  equations are numbered
   sequentially within each chapter, e.g., (II-l), (II-2).

-------
                                                                           13
                   2HN02 surfac? H20 + NO + N02
                                         (ID
A solution of the rate expression derived from Reactions (10) and (11) for


the concentration of HN02 at time t is
                                    1
                     [HN02]t =	,

                           1   A + 	L
                                   B tanhf-,
                                                                   (II-D
                                [HN02]eq tanh
where
                   .1/1   +
                   "™  I.  I r ttf\ i   •
                        \[NO]Q    [N02],
           B        =  [HN02]
                            eq
                                        [H20]2  ([N0]0  -  [N02]Q)
                                                               2n
                   [NO]Q [N02]Q
                                                                  1/2
                                    ^/2
                      (2[HO]0 [N02]Q)1/2 k1Q [H20]2
                                                   2


                       ([NO]  - [N02])2 (k1Q [H20]2)
                                                           - 1/2
                    =  (keq [NO]
eq l"UJ0






      2
            eq

-------
                                                                    14
     Taking [N0]0 = 1.0 ppm,  [N02]Q = 0.1  ppm,  [H20]  =  2  x  104 ppm, and
keq = 1.9 x 10"2 gves

                  [HN02leq = 4,4 x 10"2 ppm

In their study, Noeh et al. (1974) found kio * 10'12  ppnr3  min"1  (we  have
adjusted their third-order results to fourth-order kinetics)  in  a quartz
cylinder and kio * 10'10 with metal present.  Garvin  and Hampson report
k-jg < 1Q-13 for gas phase reactions.

     Taking kio = 10"13, 10~12, 10'11, and 1(T10 ppnr3 min"1  as  a represen-
tation of this range gives the following respective values  of T:
T = 5500, 550, 55, and 5.5 minutes.  Thus, at t = 45 minutes,
[HN02l/[HN02l   = 0.008, 0.08, U.67  and 1.0, respectively.  Depending on
    , the dark reaction will produce from 0.8 to 100 percent of the equili-
brium concentration of HN02 in 45 minutes.  The rate constant listed for
"this study" in Table 2 leads to a time scale (T) of 41.9 minutes (or ap-
proximately 1 hour), and to an HN02 concentration of 79 percent of the equi
librium value in 45 minutes.  Naturally, these results depend on  [NO]Q,
and the dark mixing time.  In interpreting the conclusions for large k-|0»
one should realize that the approximation used in Eq.  (II-l) may  break down
when t is on the order of  T.

     Figure 1 illustrates  the effect  of setting  k10 at 10'13, 10'12, and
10"10 ppm-3 min'1  (and varying kn to keep keq constant)  on  the  propylene-
NOX system.  The shortening of the induction  period is readily visible  in
the N02 curves.  Figure 2  shows  the results of setting kio ec1ual  to
10~13 p'pm"3 min'1  and  introducing  0,  10,  or 100  percent  of the equilibrium
amount of HN02  as  an  initial  condition.   Again,  HN02  shortens the induction
period.   For  both  sets of  figures, the experimental conditions were  identi-
 cal,  except for the  variable  being studied.

      It  is conceivable that all  the  HN02  formed  at the reactor walls does
 not vaporize.   Inadequate cleaning of chamber walls might then  allow a
 "memory  effect" on subsequent experiments.   Jaffee and Smith (1974)  have
 experimentally demonstrated the effects of inadequate cleaning,  as well as
 other chamber effects.  The anomalous OH source observed recently in the

-------
    1.00   -r-
     -80   --
a.
OL
£
•M

0>
U
C
o
o
     .60    --
-MO   --
    .20   --
   0-00
     0-0       50-0       100-0
                                                                                      • 10-12


                                                                           i%  ki° = lcHO
                                          150-0      200 0       250-0


                                                  Time  (minutes)
300-0      350-0       400-0
         FIGURE 1.   SMOG PROFILES FOR DIFFERENT VALUES OF ki
-------
    1.00    -r-
     .80
5:   .so
o
•M
c
CD
U
C
o
.MO   --
     .50    -
   0-00
                                                                         NOg
                                                                         03
                                                                         N02
                                                                         03
                                                                         N02
                                                                         03
[HN02]Q = 100% of Equilibrium


[HN02]Q = 10% of Equilibrium


[HN02]Q = 0
                                                     Time (minutes)
                   FIGURE 2.   SMOG PROFILES FOR DIFFERENT INITIAL CONCENTRATIONS  OF HN02

-------
                                                                           17
Riverside chamber (Winer,  1975)  could  be  due  to  HN02  retention  by  the  cham-
ber walls (see Chapter III).

2.   Heterogeneous HNCh Formation

     Morris and Niki (1973)  found that, for the  reaction

                    N205 + H20  Surface>  2HN03                    (9)

the upper limit of the rate, kg, is 1.5 x 10~5 ppnr1  min"^  in the  gas
phase.  They also showed that this reaction has  both  homogeneous and hetero-
geneous components.  Spicer and Miller (1974) have presented evidence  that
the primary mode of HN03 production in their chamber  is the heterogeneous
reaction.  They achieved nearly a 100 percent nitrogen balance in  the  gas
phase and thus concluded that HN03 vaporizes after formation.

     Because the sequence of reactions

                        N02 + 03 -»• N03 + 02                       (5)

                       N03 + N02 * N205                           (7-8)

will result in the  highest ^05  concentrations  only  after  significant 03
has formed, Reaction  (9) is important  primarily after  the  N02  peak.
Figure 3  shows the  effect of this reaction.   For  the two sets  of  profiles
in  Figure 3,  kg was taken to be  1.5 x  10"3 ppm~^  min"^  (Jaffee and  Ford's
value, 1967)  and  5  x  10~5 ppm'l  min"^.   The  higher value results  in a
markedly  greater  consumption of  N02 and  correspondingly lower  production
of  03.

     This theoretical  evaluation of heterogeneous HNOX formation  has  shown
its potential  significance.  Obviously,  the  extent of  these reactions de-
pends  on  surface  type as well  as on surface-to-volume  ratio.   The need  for
experimental  evaluation of  kg  and \C\Q in individual  smog chambers is  evident.

-------
   1.00
    .80   --
o.
0.
o
•r-
-M
to
    .60   --
c   .MO   --
    .20   --
  0.00
          0-0
                        *  N02        )
                        X  03         >kg = 5.0 x TO'6 ppnr1 nriiH
                        X  Propylene  1
                        X  N02        \
                        ^  03         Hg = 1.5 x 10"^ ppnH tm'n'l
                        ^  Propylene  7
50-0       100.0
150:0       200:0       Z5U-0

        Time (minutes)
300-0
35U.Q       MOO-Q
                  FIGURE 3.  SMOG PROFILES FOR DIFFERENT VALUES  OF  kg, THE RATE CONSTANT
                                     FOR THE HETEROGENEOUS  FORMATION
                                      OF HNOs,  NgOs + H£0  + 2HN03
                                                                     •00

-------
                                                                           19
3.   03 Decay

     The rate of 03 destruction  on  chamber walls  [Reaction  (22), Table 1]
used in the calculations  in  Chapter III was  obtained  from 03  half-lives
observed in each smog chamber.   Decay rates  were  exponential;  first-order
decay was assumed.

B.   ORGANIC CHEMISTRY

     The oxidation of organic molecules  in  polluted air has been  described
previously (Demerjian et al., 1974a; Hecht  et al.,  1974a).   In the following
sections, we discuss recent studies in this area  in relation to the development
of the kinetic mechanism.  Those reactions  not dealt with here are described
in the above references.

1.   Propylene + OH«

     Early investigators of the olefin-NOx-air photochemical reactions dis-
covered  that, in addition to ozone  and oxygen atoms, an unknown oxidant was
participating strongly in smog formation (see Lei.ghton, 1961).  Several spec-
ulations were made as to the identity of this reactant, and  it now appears
 (e.g.,  Demerjian et al., 1974a or  Hecht et  al.,  1974a)  that  the OH radical
 is  the most  likely of these  possibilities.   In fact, OH is probably  the
most important  olefin oxidant in smog, accounting  for  well over 50 percent
 of  the olefin disappearance  rate (Calvert and McQuigg, 1975).  Hence, the
 success of a kinetic mechanism  rests heavily on  its  representation of the
 OH-olefin  (or,  in  general,  OH-organic) reaction  (including the ensuing  chain).

      There are  two likely  alternatives for the  initial step  in the OH-propylene
 reaction:

      >  Abstraction of an  allylic  hydrogen
      >  Addition of OH to  the carbon-carbon double bond.

-------
                                                                           20
     Slagle et al.  (1974)  studied the propylene-OH reaction in their crossed
molecular beam reactor.   Product analysis by photoionization mass spectrom-
etry yielded a dominant ion signal  corresponding to the abstraction product.
Hence, they suggested that the propylene-OH reaction proceeds primarily by
abstraction of an allylic hydrogen.  In our simulations, this reaction was
assumed to be followed immediately by the addition of QZ to tne a11^ radical

OH- + CH2=CHCH3
                                                                  (H-2)
This could be followed by NO oxidation to form N02 and acrolein:
      CH2=CHCH202 + NO -> CH2=CHCH20- + N02
      CH2=CHCH20- + 02 -»• CH2=CHCHO +
                                                                   (1 1-3)
                                                                   (1 1-4)
 but the  large  production  of  acrolein  that would  result from this  reaction
 is  contrary  to the  observed  products.

      Another speculative  reaction  pathway was considered, in which 2 03
 molecules add to the ally!  radical  to form  a five membered cyclic peroxy
 radical .
 CH2=CHCH3
                      OH
                   NO
+ CO + 2HoCO
                                              NO
                                              r\
                                                   y
                                                    k
                                                          (II-5)
 However, this formulation could not match the rapid reaction process  that
 was observed in the UCR chamber, nor could it account for  the  large acetalde-
 hyde product yield.  Unless further experimental  evidence  arises  to indicate

-------
                                                                           21
that the abstraction mechanism is important in smog,  we shall  assume  that
the OH-propylene reaction proceeds by addition.

     Morris et al. (1971) investigated the OH-propylene reaction in a flow-
discharge reactor coupled to a time-of-flight mass spectrometer.  Adduct
peaks were observed.  It was concluded that OH adds to propylene and that
the adduct is collisionally stabilized.  In the C3D6 + OH reaction, they
observed that H was retained in the final aldehyde product, while D was lost,
Furthermore, acetaldehyde was a major product of OH + C2H4, and propional-
dehyde was observed in the reaction of OH + C3H6.  These results indicate
that addition followed by hydride shift  is a  principal mechanism for the
elementary OH-olefin  reaction:

                              CH3CHCH2OH  + CH3CH2CH20-     ,         (26)
          CH,CH=CH9 + OH-
            3     *      \       OH         0-
                              CH3CHCH£ •»• CH3CHCH3                  (27)
 Reaction (26) corresponds to terminal  addition, and Reaction (27) to internal
 addition.  Preliminary results obtained by Slagle et al.  (1974) also indicate
 that, to the extent that OH addition products were observed, the hydroxyl hydro-
 gen is retained in the product.  However, both these studies were done in the
 absence of 02.  For Reactions  (26) and (27) to occur in air, the hydride shift
 must be so fast that it precludes the addition of 02.  By incorporating these
 reactions in our mechanism as  presented above, we have implicitly assumed
 that this is true.

      Although  the experiments  cited above  provide mechanistic  insights,
 they do  not  indicate the  relative importance  of  Reactions  (26)  and  (27).  One
 would  expect terminal  addition to dominate because  it  yields the thermody-
 namically favored secondary  radical.   The  best agreement  between model  predic-
 tions  and the data  was obtained with  a terminal/internal  ratio of 4.   The
 present OH-propylene mechanism predicts  the formation  of  propionaldehyde and
 minor  amounts of acetone in accord  with  the UCR  product measurements.

-------
                                                                           22
2.   Unimolecular Decomposition of Alkoxyl  Radicals

     Carbonyl product yields are determined to a significant extent  by  the
mode of reaction of alkoxyl  radicals, short-lived intermediate products of
hydrocarbon oxidation.  In addition to Reactions (26) and (27), these species
are formed from alkylperoxy radicals in reactions with NO or SOe [Reactions
(42) and (51) in Table 1].  The fates of alkoxyl radicals that have  been con-
sidered are decomposition [Reaction  (35)], reaction with 02 [Reaction (36)],
and reactions with NO or N02-  The last two reactions were summarily investi-
gated and found to be relatively unimportant.  Yields of nitrites and nitrates
were very small,* and so RO- + NOX reactions are not considered here.  In -this
section and  the next, the first two  modes of reaction are discussed.

     Recent  experiments by  Batt et al.  (1974) have resulted in  the  first
absolute measurement  of rate constants  for alkoxyl radical  unimolecular decom-
position at  high  pressure.   By  thermally decomposing  alky!  nitrates, monitoring
fractionation  products, and using  their own  rate constants  for RO + NO reac-
tions,  they  were  able to  determine  rate constants  for RO decomposition.  They
did not estimate  the  accuracy  of  these  rate  constants.   But,  considering other
figures they reported, we believe that  an  uncertainty factor  of 8 is reasonable.
 Even with  this uncertainty, their new values are several orders of  magnitude
 higher than  most previous estimates  (reviewed in Batt et al., 1974).

      Batt et al.  (1974)  obtained rate constants only for the decomposition  of
 i-CsHBO-, s-C4HgO-, and  t-C^gO-, whereas other isomers of these species and
 shorter alkoxyl radicals appear in the mechanisms of Chapter III.   We estimated
 additional rate constants when necessary by using smog chamber data and noting
 that, in Batt et al.'s results, the unimolecular decomposition rate decreases
 with decreasing numbers of skeletal carbon atoms.
 * Alkyl nitrates have been detected at UCR by Fourier interferometry, but
   available literature  (e.g., Spicer and Miller, 1974; Kopczynski et al.,
   1974) leads us to believe that quantitative yields are indeed low.  An
   unknown oxidant peak  observed during butane-NOx experiments at UCR was
   attributed to butyl nitrate, but the estimated 6-hour concentration was
   on  the order of 1 pphm, which corresponds to  only about  1 percent of  the
   reacted butane.

-------
                                                                            23
3.   Alkoxyl Radical— 0? Reactions

     The individual  rates of Reactions  (35)  and  (36)  are  of  little  concern
in developing a smog mechanism.   But their ratio determines  the  course
of the overall reaction.  Unfortunately,  no  such ratio measurements are
available.  Those used in the present formulation were determined  largely
from UCR product distributions.

     Available rate constants for the reaction

                      CH30-  + 02 -> H2CO + H02-                     (36.1)

vary over two orders of magnitude.   Garvin and Hampson (1974) have recom-
mended that k36 1 % 4.5 x 10~3 pprn-l min"1,  whereas Mendenhall  et al.  (1974)
have estimated that k3g 1 % 0.8 ppnr1 min"1.  In the latter  study, k36 1
was determined from measurements of t-butylnitrate pyrolysis and a (very
uncertain) rate constant ratio for the reactions of 02 and NO with CH30-.
The authors concluded that a more direct measurement of k35  ^ was desirable.
Therefore, while Mendenhall  et al.'s value serves as a guideline, considerable
freedom has been taken in the present study in adjusting rate constants for
the RO- + Q£ reactions.  For smog modeling, the need for rate ratio measure-
ments is even greater.

4.   Propylene + 03

     As noted in Subsection 1, the importance of the 03-olefin reaction in
photochemical smog has long been recognized.  Consequently,  a large number
of kinetic studies have been performed, and reasonable agreement on the
rate constant for propylene + 03 has been reached  (Becker et al.,  1974;
Stedman et al., 1973; Garvin and Hampson, 1974).   The mechanism for this reac-
tion is in a greater state of flux.  However, there is much evidence in favor
of a recent mechanism postulated by O'Neal and Blumstein (0-B) (O'Neal and
Blumstein, 1973; Finlayson et al., 1974).  The 0-B mechanism supplants the
Criegee mechanism.  The Criegee mechanism has proved  to be a satisfactory

-------
                                                                            24
explanation of 03-olefin chemistry in solution (Leighton,  1961),  but not
in the gas phase, where the formation of a "zwitterion" intermediate is  not
as appealing.

     The 0-B mechanism proceeds through the formation of an equilibrium
between a molozonide and an oxy-peroxy biradical.   This biradical may
undergo many transformations, but, based on O'Neal and Blumstein's estimates,
the most likely is an a-hydrogen abstraction:
                                                  0
                                                  (I
                                                 »CHCHCH,
                                   biradical-
  OOH

OOH
                                                    0

Both Routes a and b could occur, depending on the nature of the biradical
intermediate.  The a-keto hydroperoxide products of Reaction  (II-6) will be
formed in an excited state.  Their fractionation results in the following
overall reactions:
         CH2=CHCH3 + 03-
                                        CH3C02 + OH-
                                >CH3CHO 4 CO + H02 + OH-
                 (28)

                 (29)
These products are consistent with  the mass  spectra  obtained  by  Becker  et al.
(1974).  The  split between  Routes  (28) and  (29)  is determined by which  hydro-
peroxide is formed in  Reaction  (II-6).   In  the present work,  k29/k20 =  2 was
chosen  arbitrarily to  improve model  predictions.   Combining  Reactions (28)
and  (29) with this rate  constant ratio gives

-------
                                                                           25
CH0=CHCH- + 0-3-4-0.33  H9CO  +  0.33  CHJXk + 0.67 CH-CHO + 0.67 HO.
  £.63          £.             J   C.           3             *-

          + 0.67 CO + 1.0  OH-

This reaction can be compared with Reaction (33)  in Table 8 of last year's
report:
  CH2=CHCH3 + 03 4 0.75 HgCO + 0.25 CH^Og + 0.5 CH3CHO + 1.0 H02

            + 0.75 OH- + 0.75 CO

 The new reaction differs from last year's in showing reduced formaldehyde,
 H02,  and CO yields in favor of increased OH', acetaldehyde, and CH3C(0)0;j.
 Becker et  al.'s  (1974) yields of CO from propylene + 03 was 0.6 molecules
 of  CO produced  per molecule of propylene consumed.  Because secondary reac-
 tions of aldehydes could have contributed  to this yield,  the  stoichiometric
 coefficient  of  0.67  is  probably an  overestimate.

 5.    Radical-Radical Reactions

      The reactions discussed so far have been hydrocarbon-organic chain ini-
 tiation and propagation reactions.   Chain termination occurs through radical-
 radical  reactions, such as NO + OH- •> HN02  [Reaction (13)], N02 + OH- + HNOs
 [Reaction (14)], or H02 + H02 + H202 + 02 [Reaction (46)].  To provide ade-
 quate damping, one must include similar reactions for organic radicals.  This
 can  be done with reactions analogous to.those just cited.. In the present for-
 mulation, the production of organic hydroperoxides,

                      R02 + H02 * ROOH + 02     ,                    (47-48)

 has  been  chosen.  On an effective  collision probability  basis  (likelihood
 of an R02-H02  collision)  these reactions  were  assigned  a rate  constant of
 one-half  that  of  the reaction  H02  +  HOg •»• H202 + 02  [Reaction  (46)1.
 Further thermal or  photolytic  reactions  of organic  hydroperoxy species have
 not been  included.

-------
                                                                           26
6.   PAN Chemistry

     The formation of peroxyacetylnitrate  (PAN),  and  its  homologs,  occurs  by
another radical-radical  reaction.   Reaction  (44),

                       0            0
                      RCO^ + N02 -». RC02N02     ,                  (**)

was discussed in last year's  final report (Hecht  et al.,  1974b).  It was
speculated there that PAN might hydrolyze on the  walls  of the UCR chamber.
Although this undoubtedly could occur, an analogy to N20s suggests  that gas
phase collisional destruction could be several orders of magnitude faster
than surface reactions under ambient conditions.   Thus, we presently pro-
pose that PAN may undergo a thermal decomposition reaction, resulting in
the rupture of the peroxy and carbon-carbon bonds:

                         °9
                (Mf)  PAN -5 N03 +  C02  + CH302 (+M)                (45)

Based  on data contained in Benson  (1968)  and  Doma.lski  (1971),  this  reaction
is  exothermic by  about  14 kcal mole"1.

     The occurrence  of  Reaction (45)  is supported by the experiments of
Schuck et  al.  (1972)  and recent PAN decay experiments  in the Riverside cham-
ber (UCR monthly  report No.  4).   In the former study,  PAN was  found to oxi-
dize NO to N02-   The reaction  was first order in PAN and zeroth order  in NO.
The ratio  of C02  produced to PAN  consumed was nearly 1.   The ratio of  N02
 formation  to PAN  consumption was  approximately 2 in a  nitrogen atmosphere,
 but was much greater than 2  in an oxygen  atmosphere.  This is  further  evi-
 dence  for  the occurrence of  Reaction  (45),  followed by NOs + NO + 2N02
  [Reaction  (6)]  in an N2 atmosphere,  and Reaction (6) plus
 CH302  + NO -»• CHaO- + N02 [Reaction (42)], CH30-  + 02 •* H2CO + H0£  [Reaction
  (46)], and H02 + NO -> OH-  +  N02 [Reaction (15)]  in an  02 atmosphere.
 Schuck et al.'s rate constant, k45 =  2.06 x 10'2 min"1, is 10 times that
 obtained from the half-lives observed in the UCR chamber.  This difference

-------
                                                                           27
may be due to additional  wall  decomposition  in  Schuck  et  al.'s  reactor.
The Riverside half-lives  of 5.7  ± 0.1  hours  in  the  light  and  5.5  ±  0.4
hours in the dark provide further confirmation  that PAN does  not  photo-
decompose at an appreciable rate (Leighton,  1961).

C.   PHOTOCHEMISTRY

     A distinctive feature of "Los-Angeles-type" air pollution  is the role
played by sunlight in its causation.  The free  radicals  that initiate the
process of oxidant production in Los Angeles air come from photolytic
splitting of molecular bonds.  In smog chamber  studies of the type considered
in Chapter III, sunlight is replaced by artificial  illumination.   Aside from
the overall intensity of the light source, the  spectral  distribution of pho-
ton flux is the major light source characteristic.   The spectrum of the
Riverside solar simulator and the solar spectrum are reproduced in Figure 4.
The solar simulator consists of a light source and a light filter.  The dark
solid line in  Figure 4 represents the filtered spectrum.

1.    Photolysis Rate Constants

      Given a light source spectrum, such  as that in Figure 4,  rate constants
for  the  various photolysis  reactions  included  in Table 1  can be  computed.
For  this  purpose, quantum yields and  absorption coefficients for the absorbing
molecules must be known.  Photolysis  rate constants can  then be  computed  from

                           k =  /I <(. a  dv,                         (H-7)
                               J vyv  v
and
                          I   =  1° F      ,                           (II-8)
                           V    V V
 where
           1° = photon flux provided by the light source at wavelength v,
           av = absorption coefficient for the absorbing molecule,
           Fv = filter factor,
           4>v = quantum yield (molecules dissociated per photon absorbed),

-------
                                                                                           28
  E
  c
  o
  0>
  in

C\J
 I
  c_
  O

  l/l
  c.
  o

  o

  CL

ST
r—
 O
  X
  3
                                                          Sunlight

                                                          U.C. Riverside
                                                          Solar Simulator

                                                          Simulator with
                                                          20 nm Filter Shift
             300
400
500
                    €00
                                            nm
       FIGURE 4.   SPECTRA OF  SUNLIGHT AND U.C. RIVERSIDE  SOLAR  SIMULATOR

-------
                                                                          29
and where integration extends over the entire light spectrum.   Rate  constants
for the present investigation were computed from Eq.  (II-7)  using I   values
                                                                  V   ^
obtained when UCR first installed their solar simulator in the chamber,  and
values of ay and ^ extracted from Calvert and Pitts  (1967)  and Johnston  and
Graham (1974).  The values of these rate constants, normalized by k,,  for
N02 -*• NO + 0 [Reaction (1)]  are presented in Table 3.

     Measurements of  !q are made  periodically at UCR, and k^p: 9/4 k]) is
obtained from  light meter readings during  the Battelle simulations.    The
procedure used  in carrying out the computations presented in  Chapter  III
was  to multiply  the measured values of  Iq  by  the ratios in Table 3  to  obtain
the  needed  photolysis rate constants.   The  shortcomings of this  procedure
are  critically  assessed below.  Obviously,  periodic measurements of Iv
would facilitate a more accurate  approach.

2.   Spectrum  Decay

     For most  of the  photoabsorbers participating  in smog reactions,  mole-
cular dissociation occurs primarily as  a  result of the absorption of  ultra-
violet light.   As is  to be expected,  different  chemical species  have  different
absorption  spectra.   For  example,  the absorption  by N02,  resulting  in NO and
0,  is largest  and fairly  uniform  over the  300 to  400 nm range, whereas the
absorption  by  acetaldehyde,  to form either Cffy  and CO  or  CH3  and HCO, exhi-
bits a peak in the 250 to 300  nm  range  and very little absorption elsewhere.
Hence, the  ratio of  k-j to other  photolysis rate constants will  be very sensi-
tive to  the intensity and distribution  of UV  light from the  irradiation  source.

     Periodic  measurements of  k-|  by experimentalists at UCR  have shown a
consistent  light deterioration.   Between  smog chamber  experiments EC-38  and
EC-60 k-j decreased by about  40 percent  (UCR monthly report  No.  5).   The  cause
of this  reduction was not investigated, but there  are  two obvious possibili-
ties.  The  first is  a reduction  in light  source emission,  perhaps resulting
from a buildup of UV-absorbing material on the  inside  of  the  light  bulb
(Burton,  1975).  The second  is increased  UV absorption by mirrors.   Absorption
.by mirrors  could slowly impair their  reflection properties.   Indeed,  the
  Data obtained through private communications.

-------
                                                                         30
                           Table 3

          PHOTOLYSIS CONSTANT CHANGES FROM UV LOSS
Reaction
N02 -> NO + 0
HNOg •+ OH- + NO
03 •> 02 + 0(1D)
03 •*• 02 + 0(3P)
H2CO -> H- + HCO-
HgCO -v H£ + CO
CHaCHO -> Products^
CH3CH2CHO + Products^
H202 -*• 20H-
MEK -> CHaCHO + CH3C(0)'
standard
Ratio
1.0
0.070
0.026
0.035
0.0049
0.011
0.0077
0.0085
0.0036
0.0036
5 nm Shift
1*
0.070
0.013
0.034
0.0040
0.010
0.0060
0.0065
0.0031
0.0024
10 nm Shift
1*
0.070
0.006
0.033
0.0031
0.0089
0.0045
0.0047
0.0028
0.0015
20 nm Shift
1*
0.069
0.0009
0.0031
0.0018
0.0069
0.0022
0.0022
0.0021
0.0005
Iq itself was reduced 1.9 percent by a 5  nm shift,  4.2 percent by a 10 nm
shift, and 10 percent for a 20 nm shift.   Reduced ratios  listed have been
renormalized to new Iq.
These products include radicals and stable species, as shown in Table 1.
Precise quantum yields are not known.  We have assumed that the quantum
yield for stable products is approximately one-third of that for radicals.

-------
                                                                          31
mirrors were observed to be damaged  and  were  sent  out  to  be  recoated  after
Run EC-61.)  This reduced reflection and increased absorption  would probably
be most pronounced in the UV region.

     Although spectrum decay must ultimately  be determined experimentally,
its potential effect can be assessed theoretically.  For  this  purpose,  rate
constants were recomputed using the  following variation of

                       k =  fl'et 4  d     ,                        (II-9)
                           J  v V V   V
where
                      I1 = F    -1°    ,                         (11-10)
                       v    v-y    v

and y is the magnitude of a shift in the filter factor to lower frequencies.
Replacing  Iv in Eq. (11-7) by 1^ in  Eq.  (11-9) reduces the UV intensity.   The
spectrum change due to a 20 nm shift in  the filter factor is illustrated in
Figure 4.  Percentage reductions in  rate constant ratios, computed for v = 5,
10, and 20 nm, appear in Table 3, along  with their unshifted values.   In
Figure 5,  the effect of a 20 nm shift on smog profiles is illustrated.  The
spectrum deterioration clearly delays the N02 peak, as observed experiment-
ally at UCR.  Although Experiments  EC-16 and EC-60 were nominally the same,
EC-60 took almost twice as long to  reach the N0£  peak.  Using only the re-
ported initial conditions and values of k-j, we could not  reproduce this
delay by computer simulation (see Figure 6).  However, a  simulation  employing
a  20 nm filter shift showed that the observed delay could easily be  accounted
for by UV  spectrum deterioration.   We concluded that more complete spectrum
characterization  in smog chambers is needed.

D.   S02 OXIDATION

     In the  previous two years, our smog chemistry modeling efforts  have
focused on systems containing  olefins,  paraffins, and  NOX.  During this
year, we also considered  systems  containing  S02 and toluene.  The next two
sections are devoted to  a  discussion of chemical  processes  related to  these
two species.

-------
    1-00
      .80
 Q.
 Q.
 O

4J
 m
01
O
O
      .60    --
     .20    —
    0-DO     •-*«
           0-0
                                    ;Propylene
                                                      Time  (minutes)
                                                                              y& Propylene
                                                                              X N02
                                                                              9K 03
                                                                              X Propylene
                                                                               X  03
3DQ-Q
          No  shift



          20  nm  shift
350-D
40Q-0
                      FIGURE 5.  EFFECT  OF  A 20 NM FILTER SHIFT ON SMOG PROFILES
                                                                                                                      to
                                                                                                                      .f\>

-------
    1.20
    0.90
a.
0.
c
o
P   0.60-
OJ
o
o
o
    0.30-
    0,0
        0.0
                                                                                    	 OBSERVED  DATA
                                                                                    	 NO  nm SHIFT  SIMULATION
                                                                                    	 20  nm SHIFT  SIMULATION
50.0
100.0
150,0       200..0
        Time (minutes)
250.0
300.0
350.0
                                                                                                         400.0
                                                                                                                     CO
                                                                                                                     CO
                      FIGURE- 6.    EFFECT OF A 20 nm FILTER SHIFT ON NOo  BEHAVIOR FOR EC-60

-------
                                                                     34
1.    Some Observations

     Past smog chamber  studies  of irradiated  S02-NOx-hydrocarbon  and
S02-03-olefin mixtures  have  resulted  in  a  variety  of  qualitative  and  quanti-
tative observations,  many of which are inconclusive or  even  contradictory.
Of particular significance is an uncertainty  about the  effect  of  S02  on  the
yields of carbonyl  products  (see reviews by Leighton, 1961;  Wilson  and Levy,
1970) and ozone (Wilson and  Levy, 1970;  Altshuller et al.,  1968).  Wilson  and
Levy (1970) have shown  that  the overall  reaction is strongly dependent on
relative humidity.   Undoubtedly, chamber effects also play  a role in  causing
these inconsistencies.   There is agreement, however,  on the  observation  that
the addition of S0£ to  the hydrocarbon-NOx-air system results  in  increased
aerosol production  and  that  this aerosol formation occurs  after the N02  peak.
It has also been generally observed that the addition of propylene to an irra-
diated S02-NOX mixture  increases S02  consumption and  aerosol production.  The
aerosol is thought  to be chiefly sulfuric  acid and water,  though  it may  contain
small concentrations of organic and nitrite-type material  (Filby and Penzhorn,
1974; Bufalini, 1971).

     An examination—which appears later—of the Battelle data (S-110 and S-115)
permits two further observations.  The addition of ^ 0.5 ppm S02 to an irra-
diated N0x-olefin-air mixture does not substantially alter either the ozone
production or the propylene oxidation rate.

     A free radical mechanism for S02 oxidation is in accord with these
observations.  Because oxidation occurs  mainly after the N02 peak and because
added hydrocarbon increases aerosol production, the  radicals that oxidize S02
are  probably the same as those oxidizing NO.  These  radicals must be either
organic or of organic origin.  The low reactivity of S02 accounts for the
smallness of its effect on both  hydrocarbon oxidation and ozone  production.
Sulfuric acid is probably the main oxidation product of S02-

-------
                                                                      35
2.   Kinetic Mechanisms for SO? Oxidation

     Recent investigators of atmospheric S02 chemistry (e.g.,  Davis  et al . ,
1974; Calvert and McQuigg, 1975; Castleman et al.,  1974)  have  suggested that
the oxidation of S02 by OH' and HO;? is  the primary  means  of S02 removal in
the gas phase.  Several recent measurements of the  S02-OH-  rate constant are
reviewed in Table 4.  They cover a wide range, but  the value of 900  ppm~l min"'
selected for the current mechanism encompasses the  full  range  with a 50 percent
uncertainty.

     As shown in Table 4, Davis (1974)  measured a rate constant of
l<50 = I-3 PPm~^  min'l for

                    H02 + S02 + OH- + S03    .                    (50)

Because alkyl and acyl peroxy radicals  should be more reactive than HO^, we
have incorporated Reactions (51) and (52) into the kinetic mechanism with
rate constants of 1.5 ppm~l min"^, slightly higher than
                    R02 + S02 -> RO- + S03    ,                    (51)
                              R02 + C02 + S03                     (52)
     Reactions of N03 and N205 with S02 were postulated by Wilson and Levy
(1969) to explain a rapid reaction observed between N02» 03, and S02-  As
shown in Table 4, this could not be explained by a ; direct reaction of S02
with 03.  Thus, the following sequence was invoked:

                 NQ2 + 03 + N03 + 02; N03 + N02 ^  N^     ,   (5,  7  and  8)

                 N03 + S02 -> S03 + N02     ,                        (49)

                 N2°5 + S02 * S03 + 2N02

-------
                                                    RATE CONSTANTS FOR S02 OXIDATION.
                                                            (In ,.„„,-1  r.in-1)

-
S02
Reaction
» OH - HSOj
+ o -iL so3
Davis (1974)
1350 i 130
(7.6±0.9)exp(-2240/RT)
Calvert
and HcQuiqg
(1975)'


Castleman
et al.
(1974)
590

Others
C60 ± 120
(COX, 1974)
3000
(Leighton, 1964)
Present
Mechanism
900

S02 + N03
                  N02
< 14.7
                                                          10-5
                                                                                                  270 ± 50
                                                                                                  (Mulcahy et al., 1967)
                                                                                                                             14
S02 «• H02 --SO'3 + OH-
                                  1.35 ± 0.2
S02
S02
S02
+ RC(0)02 - R + CO + S03 )
+ ROJ •* RO- + S03        j
S02 + N205 - 2fi02 + S03
S02 + 03 •» $03 + 02
$03 + H20 JU K25:04
R02 + S02 » ROS03
RC+00- + S02 - S03 + RCHO
HSOj + 02 - HSOs * S03 + H02
HSOJ + NO -> HSO^ + N02
HS04 + HOJ •* H2504 + 02
HSOi * N02
            H20
                • H2S04 + HN03
                                  <. 1.-2 10-8
                                  3 x 10'7
                          10-4k(OH- * S02).
                          if HS04 is a product
                          % HOj + S02
                          < 6.2 10'8
                          < 1.2 ID'10

                          l°"4k(OH- + S02)
                                                                                      740
                                                                                                 i. 10"4 min-1
                                                                                                 (Leighton, 1961)f
                                                                                                                          1.3
                                                                                                                              1.5
                                                                                                 •v 6.3
                                                                                                 (Cox and Penkett, 1972)
                                                                                                                          Immediate
                                                                                                                          See text
                                                                                                                            800
                                                                                                                           9000

                                                                                                                          1 x 104
* Quantum yields for $03 production ($$03) measured in an S02 atmosphere vary from 3 x 10"4 (Cox, 1972) to 0.1  (Chung et al., 1975).  Chung
  f. al. (1975) explained this discrepancy by considering secondary reactions.  Extrapolation of these results  to atmospheric condition 1s
  not possible at present because $50, varies with atmospheric pressure and type of diluent.
                                                                                                                                                               CO
                                                                                                                                                               en

-------
                                                                      37
The upper limits of the N20,-  +  S02  reaction  rate  (Table  4)  indicate its in-
significance.   In contrast, the upper  limits recorded  for k»9  (Table 4)
are in severe  disagreement.   We can estimate k.g  assuming reactions of H02
and N03 with S02 and NO are similar (all  are "oxygen abstraction"  reactions).
Using the ratio, k(s02 + H02VHNO + H0p)>  to be approximately  10"   (from
Tables 2 and 4) and k(^g  + NO) - 1°  PPm~  m'in~   (from  Table  2) we find
                        3
 k49 = k(S09 + H09)/k(NO + HOA)  '  k(NOQ  + NO) ~ 10 ppm"   min"
       L          ^           *J       6                               (11-11)

This estimate is in closest agreement with Davis1  (1974)  reported  value.
Reaction (49) could therefore be  significant,  and it has  been initially
included in the mechanism with k4g  =  14 ppm   min"1  (Davis1  value).

      Cox and Penkett (1972)  found  that SOp added to the  olefin-03 system
 was oxidized to sulfuric acid aerosol.   They  also found  that,  when  the  ozone-
 olefin consumption ratio in  the  absence  of S02  was greater  than  1,  the  addi-
 tion of S02 reduced this ratio.  The addition of S02 also enhanced  carbonyl
 product yields.  From these  and  other results,  Cox and  Penkett inferred that
 a short-lived intermediate of the  ozone-olefin  reaction, which they specu-
 lated might be a zwitterion, was responsible  for SOp oxidation.   However,  in
 light of the discussion of the mechanism for  63 + propylene in Subsection  4
 and the data in Table 4, it seems  more reasonable to account for Cox and
 Penkett's observations in terms  of reactions  of S02 with hydroxyl, hydroper-
 oxyl, peroxyalkyl, and peroxyacyl  radicals rather than  zwitterions.

      The final modes of S02 oxidation included  in Table 4 are direct oxidation
 by 0 atoms and photooxidation.  The former can  be ruled out, because the  maxi-
 mum 0 concentration is only about  10~9 ppm.  With S02 in ppm concentrations,
 this gives rates < 10~6 min"^.  Photooxidation  has been widely investigated
 (Cox, 1972; Sidebottom et al., 1972; Demerjian  et al.,  1974b;  Smith and Urone,
 1974); the experimental results  show that it is an unimportant oxidation  pro-
 cess in polluted atmospheres.  The inefficiency of photooxidation is due, in
 part, to the forbiddenness of direct excitation of S02 from the ground state
 to  its reactive triplet state SOg^B-j)  (Sidebottom et al.,  1972).  Sidebottom
 et  al.  (1972)  estimated an upper limit  of 2 percent S02  consumption per hour
 for gas phase  photooxidation  in the atmosphere.   However,  as  they noted,

-------
                                                                      38
this could be an extreme upper IJmit  because  the  efficiency  for  singlet-
triplet intersystem crossing  may have been  overestimated,  and  the  assump-
tion that all quenching of 3S02 by 02 was chemical  rather  than physical was
probably faulty.  Because of  the results of the  investigations cited  above,
S02 photooxidation does not appear in the present mechanism.

     Direct reaction between  3S02 and hydrocarbons  has  been  reported  at high
reactant pressures (Badcock et al . , 1971; Sidebottom  et al . ,  1971;  Filby and
Penzhorn, 1974).  However, at ppm levels, Demerjian et  al .  (1974b)  found
that the primary result of the interaction  of 3S02  with 2-butene was  the
interconversion of cis  and trans isomers.   Thus,  a  direct  chemical  reaction
between ^sc   and olefins under atmospheric  conditions can  also be  ruled out.
     As shown in Tables  1  and 4,  $03  is  produced by the  reaction  of
RC(0)02, H02, and N03 with S02.   Castleman  et al .  (1974)  found  that S03
rapidly combines with water to form ^SO/p   Presumably,  ^864 then nucleates
and adds more water to form aerosol droplets (H2S04-nH20) .   The studies car-
ried out by Cox and Penkett (1972)  and the  more recent mass spectrometric
measurements made by Schulten and Schurath  (1975)  indicate  that ^$04 is
the major sulfur-containing product of the  olefin/03/S02 reactions.  In
Bufalini's (1971) survey,  she reported that the aerosol  produced as a result
of the irradiation of 3  ppm of lower olefins in the presence of 1 ppm N02 and
0.5 S02 at 50 percent relative humidity was primarily composed of sulfuric
acid.  Accordingly, no other products have  been accounted for in the present
mechanism.  Furthermore, we assumed that $03 is rapidly  converted to
     Davis et al. (1974) and Calvert and McQuigg (1975) have speculated on
the formation of compounds containing S, 0, H, and N when NOX is present.
Presumably, the formation of these nitrogen-containing species results from
radical -radical reactions with oxidation products of the OH--S02 adduct.  For
example, HS04, formed as shown in Table 4, could combine with N02 to form
HS04N02-  Subsequent hydrolysis at the wall would produce nitric and sulfuric
acid:

-------
                                                                      39
               HS°4 + N02 1U?r H2S04 + HN03
     This represents one possible fate of the HS04 molecule produced in the
following series:

                                    02
                  OH- + S02 -»• HSOg — ^HS05    >                 (50)*

                  HSOg + NO -> HSO^ + N02    .                     (54)

     Other possible reactions of HS04 (and HS05) have been assembled by
Calvert and McQuigg (1975), including hydrogen abstractions from alkanes
and H02* and addition to olefinic double bonds.  Davis et al .  (1974) proposed
that HS06 could be produced by the addition of Q£ to HStty.  It could then
oxidize two'NO molecules, reforming HS04, which would then repeat the process
This mechanism was presented to explain an ozone bulge found by Davis et al .
(1974) in the plume downwind of the Potomac Electric Power Company's power
plant at Morgantown.  However, laboratory studies have shown that the addition
of S02 to the hydrocarbon-NOx system actually causes either little change
or even a reduction in 03 production (Wilson and Levy, 1970).  As stated
at the outset of this discussion, we assumed in the present mechanism
formulation that S02 has only minor effects on 03 yield; so Davis et al.'s
suggestion has not been incorporated.   In addition to Reaction (56), a
simple termination reaction,
                          HS04 -> H2S04 + 02    ,                    (55)
 * An alternative to the last step in Reaction (50), HS03 + 02 -> H02' +
   seems implausible thermodynamically.  AH is approximately 0 to 20 kcal
   mole-1 for this step, compared with AH ^ -50 kcal mole-1 for the last step
   In Reaction (50).

-------
                                                                      40
has been included.  In light of the current lack of experimental  studies,
this seems a pragmatic and adequate closure to the S02 oxidation  mechanism.

E.   THE TOLUENE-NOX-AIR SYSTEM

     The development of a toluene oxidation mechanism has been greatly hin-
dered by a lack of information on the pertinent reactions of toluene and by
a dearth of detailed product analyses of smog chamber experiments.   For
these reasons, the mechanism proposed in Subsection 3 below is sketchy and
possibly premature.  The first two following subsections briefly  review the
currently available literature relating to the smog reactions of  toluene.

1.   Toluene + 0(3P)

     Jones and Cvetanovic (1961) and Atkinson and Pitts (1975) have determined
rate constants for this reaction.  According to the latter, k (toluene + 0)
^ 120 ppnf'  mirr1.  At a typical 0 concentration of ^10~9 ppm, this reaction
is of little importance.  In the present mechanism, OH- is taken  to be the
sole oxidant.

2.   Toluene + OH-
     Davis et al.  (1975) measured a rate constant of 9.0 ± 0.6 x 103 ppm'1
min~' for the toluene-OH- reaction under a pressure of 100 torr (M = He),
where the reaction was still in the pressure fall-off region.  However, extra-
polation to room pressure via a Lindemann. (1/k versus 1/P) plot indicated
that this value was very close to the high pressure limit.  The pressure
effect Davis et al. observed (k at 100 torr was nearly twice its value at
3 torr) is evidence that a significant fraction of the reaction proceeds ini-
tially through the formation of a collisibnally stabilized adduct.  The rate
constant measured at 3 torr (k = 5.3 ± 0.4 x 10~3 ppm"1 min'1) implies that
addition to the ring accounts for at least one-half of the toluene-OH-
reaction.

-------
                                                                      41
     Doyle et al.  (1975)  used a 6000-liter glass  smog  chamber  to  measure  the
disappearance rates of several  aromatic compounds relative  to  that  of  butane.
The hydrocarbons,  along with ^0.27  ppm NO and  ^0.04  ppm N02, were irradiated
and aromatic concentrations were measured after 1 and  2 hours.  After  pre-
senting evidence that OH- was the primary hydrocarbon  oxidant  in  their system,,
Doyle et al. obtained k(toiuene + OH.) = 6-2 ± 2-3 x lo3 PPm   min   ,  though,
of course, this value depends on the value of the rate constant used for
butane + OH.  Doyle's measurements of relative rates for the three  isomers
of xylene indicate that the primary mechanism for aromatic  hydrocarbon-OH-
reactions is electrophilic addition to the ring  (see Section 3).

     Although neither of  these two investigations give direct  evidence of the
existence of an abstraction route,  the resonant stability of the  phenalkyl
radical, produced  by hydrogen abstraction from the methyl group,  makes this
an appealing pathway.  That this process does  indeed occur  (to some extent)
is implied by the  observation of benzaldehyde  and peroxybenzolnitrate  (PBZN)
in the toluene-NOx system by Heuss  and Glasson (1968).  Following Davis et  al.
(1975), we have set a limit of 50 percent for  the relative  frequency of
H-abstraction to OH-addition.

     Another noteworthy observation was made by Altshuller  et al. (1970), who
identified .formaldehyde as a product of the same (toluene-NOx) system.  Simul-
taneously, they detected  only trace amounts of acetaldehyde.  Formaldehyde
could be formed subsequent to oxidation of the methyl  group on toluene or
after displacement of the methyl group by OH.*

     The investigations just cited are the basis for our current toluene  oxi-
dation mechanism.   Obviously, more product measurements are sorely  needed.

3.   The Proposed  Mechanism

     Reactions (33) and (34) are the initial H abstraction  and OH addition
reactions:
* Another possibility, not included in the current mechanism, is the opening
  of the aromatic ring with the subsequent formation of fractionation products.
  As these are likely to include acetylene, testing for this possibility experv
  mentally should be straightforward.

-------
                                                                     42
                         0,
             LQ) + OH-
                                             fH2°2
                                            (O)    +  H20
                                                     (33)
                                   OH 0,
                           ->   [0
                                     OH
                                                       HO
                                                                  (34)
 In  both cases, an immediate reaction with 02 has been assumed.   The o-cresol
 product of Reaction (34) is taken as a prototype for o,  p (i.e., electro-
 philic addition) cresols.*

      The  peroxy phenalkyl radical, produced in Reaction (33), could react
 as  follows,  to produce benzaldehyde and PBZN.
                                                      :HO
           NO +  \O]
N02 +  CQ)
                                 H02 +  (O]
(42,  36)
              :HO
                      L         Ic
                                       NO,
CO]  + OH--^H20 +  (g)  -^2_^  [0]
                                                                  (32, 44)
Other radical  reactions,  selected  in  analogy  to  those  included  in Table 1
are
 * The procedure of using o- as a prototype  for o,  p-  was  used  throughout.

-------
                                                                     43
                            (0}  - H?CO    .
                                               (35)
                        NO
                                               (43)
both of which produce  a  peroxy-phenyl  radical.  Subsequent reactions  [Nos,
(42'), (42")] produce  (o,p)  quinones:
                                                               .   (42', 42")
Because of the diradical  character of quinones,  they could react further,  but
such speculation has not been included here.

     If in Reaction (34)  OH-  has combined with toluene at the ring-methyl
bond, Reaction (34) would become
(Oj  + OH-
                              0,
 3H
(Q)  +
                                                                  (34')

-------
                                                                     44
     Phenol  and cresols [from Reactions  (34)  and (34')]  should be very reac-
tive.  Gitchell et al.  (1974) found that addition of phenol  had an inhibiting
effect on propylene-NOx systems.   Therefore,  they suggested  that the
OH- + phenol  reaction proceeds by hydrogen abstraction from  the phenolic
hydroxyl group.  This,  and their further speculation that NO and N0£ add
to the phenoxy radicals thus produced, seem reasonable.   However, in an
attempt to match the predictions of the kinetic mechanism with the NAPCA
smog chamber profiles (see Chapter III), -OH hydrogen abstraction from
phenol and cresols were assumed to be followed by 02 addition as in
                                          f3
                        + OH	^>H,0 +
                                    2
     The above reactions, along with appropriate radical + H02 reactions,
constitute the present kinetic mechanism for toluene oxidation (Table 13,
Chapter III).  Rate constants have been selected on the basis of reaction
analogies.

-------
                                                                       45
                  III   SMOG CHAMBER  SIMULATIONS
     Smog chamber studies  done  at  the  University of California, Riverside
(UCR), Battelle Memorial  Institute,  and the  National  Air  Pollution  Control
Administration (NAPCA)  serve  as the  data  base  for  validating the kinetic
mechanisms.   Validation involves comparing smog profiles  obtained by numeri-
cal integration of chemical rate equations with those obtained  experimen-
tally.  Numerical integration and  the  derivation of rate  expressions from
stoichimetric equations and rate constants were done  on a CDC-7600  digital
computer.  The programs used  (MODKIN and  CHEMK) were  developed  under previous
EPA contracts by SAI  and  by G.  Z.  Whitten at Lawrence Berkeley  Laboratory
specifically for this purpose (see Appendix  A  of Hecht et al.,  1974a).

     Unfortunately, rate constants and reaction mechanisms were not accurately
known for many of the elementary steps comprising  the smog mechanisms.   Addi-
tionally, chamber characterization was usually incomplete.  Thus,  an  unavoid-
able aspect of validation was "tuning."  Tuning  involves  adjusting  parameters
within their range of uncertainty  for  the sole purpose of improving the fit
between experiments and predictions.  As in  the  past, we  attempted to minimize
the amount of tuning and to maximize the dependence on kinetic  data.

     Chamber characteristics  that  were usually available  are k,, dilution
rate and heterogeneous 03 loss.  The last two were incorporated as simple
first-order reactions.  Rate  constants for all photolysis reactions were
determined from  k, through the use of Table 3.  Other procedural aspects
and the  simulation results appear in the ensuing subsections.

A.   SIMULATIONS OF UCR DATA

     The operating characteristics  of  the UCR chamber and peripheral equip-
ment were described  in detail  in  last  year's report  (Hecht et  al., 1974a).
The only addendum  to be made here is  to  note calibration errors in 0^  and

-------
                                                                        46
NOX measurements  discovered  during the year.  The following corrections
to the data appearing  in  last year's  report had to be made:
                      V   "V1-2    '
                      NO1   =  1.2  NO    ,
                      N02'  =  1.2  N02 + 0.2  PAN

Primes represent the new values,  and the  absence  of  primes  indicates old values.

1.   Results for the Propylene-NO^-Air Block
                                 A

     The initial concentrations  for the propylene-NOx  block appear  in Table  5
and Figure  7.  The mechanism  used to simulate  these  experiments  is  presented
in Table 6.   Photolysis  rate  constants used in  the simulations were computed
from a 10 nm filter-shifted spectrum.  Chapter  II presents  these rate con-
stants, normalized by k].   The choice  of  a  10  nm  shift was  somewhat arbitrary,
but more complete spectrum decay characterization was  not available.

     Figures 8  through 23  present the  results  of  UCR experiments and computer
simulations  thereof.  Throughout this  chapter,  experimental data are represented
by plotted symbols, and  computer curves by  unbroken  lines.   With the exception
of Run EC-12,  the predictions of propylene, N02,  NO, and ozone behavior are
fairly good.  The exceptions  are the lack of correspondence between the asymptotic
N02 levels in  EC-18, and,  consequently, low 03  predictions  and a small  delay in
the N02 peak of EC-11, which  is  also mirrored  in  the 03 profile.

     The predictions of  carbonyl  products (except PAN) are  consistently low. At
least part of this discrepancy might  be attributed  to spuriously high measurements
(Darnell, 1974), especially at low concentrations.

-------
                                                           47
                   Table 5
           PROPYLENE-NOV EXPERIMENTS
                       A
                   (In ppm)
      Initial Concentrations
EC
Run
11
12
16
18
21
Propylene
0.447
0.082
1.036
0.972
0.104
NO
0.115
0.106
1.12
0.106
0.558
N02
0.020
0.012
0.156
0.0142
0.066
NOX
0.135
0.118
1.27
0.148
0.624
llVSo/ MV/v
Ratio
0.15
0.10
0.12
0.12
0.11
kl
0.223
0.223
0.223
0.223
0.223
1.0
1
Q.
C
O
(0
c
| 0.5
o
o
ox
~ 0.1
If) X


x 21

-

x 12 x 11 18 x
i i i
          0.1            0.5                1.0
       Initial Propylene Concentration (ppm)
Note:  The EC run number is  given next to each point
      FIGURE 7.  PROPYLENE/ NOV FACTORIAL BLOCK
                              A

-------
                                                                       48
*
                              Table 6
                 THE PROPYLENE OXIDATION MECHANISM

                                                  Rate  Constant
__ _ Reaction _          (ppnH  mln"1 )

N02 + hv -v NO + 0                                 0.223f

0 + 02 + M + 03 + M                               2.08  x  10~5

03 + NO + N02 + 02                                25.2

N02 + 0 -> NO + 02                                  1.34  x 104

N02 + 03 -»• N03 + 02                                0.05

    + NO •> 2N02                                    1.3 x 104

    + N02 •*• N205                                   5.6 x 103
      ^- N03 + N02                                24f

 H20 + N205 -»• 2HN03                                5 x 10'6

 NO + N02 + 2H20 -> 2HN02 + H20                     1.3 x  1Q-11§

 2HN02 -> NO + N02 + H20                            0.24

 HN02 + hv ->• NO + OH-                              1.5 x  lO'2"1"

 OH- + NO -»• HN02                                   3 x 103

 OH- + N02 -»• HN03                                  1 x 104

 H02 + NO -*• N02 + OH-                              8 x  102

-------
                                                                      49
                                                  Rate Constant
_      Reaction _           (ppm""1  min"1)


03 + hv + 02 + 0(3P)                              7.3  x  10-3


03 + hv -> 02 + 0(]D)                              1.3  x  10-3


0(]D) + M -> 0(3P)                                 8.6  x  104


0(]D) + H20 -> 20H-                                 5.1  x  105


03 + OH- -»• H02 +  02                                87


03 + H02 •* OH- +  202                               2<4


03 * wall                                          1
                                                       10"3t
C3H5 + OH- •*• CH3CH(0-)CH3                         5.0 x  103


C3He + OH- •*• CH3CH2CH20-                          2.0 x  104
       03 -»• H2CO + CH3C(0)02' + OH-                 0.013


C3H6 + 03 + CH3CHO + H02 + OH-                     0.007

                                                            3
C3H6 + 0 -> CH3CH20- +  HO? +  CO                     2.66 x 10


C3H6 + 0 •* CH302' + CH3C(0)02                      2.66 x 103


H2CO + OH- -*•  H20 + H02 +  CO                        2.1 x 104


CH3CHO +  OH-  -> H20 +  CH3CO(0)02                   2.1 x 104


CH3CH2CHO + OH- -> H20 + CH3CH2C(0)02              4.5 x 104


CH3CH(0-)CH3  -> CH302 + CH3CHO                     8.35  x

-------
                                                                       50
                                                  Rate Constant
	Reaction	           (ppnH min~1)

CH3CH2CH20- -> CH3CH202 + H2CO                     8.5 x  103t

CH3CH20- -> CH302 + H2CO                           6.0 x  103f

CH3CH(0-)CH3 + 02 -> CH3C(0)CH3 + H02              0.04

CH3CH2CH20- + 02 -* CH3CH2CHO + H02                0.06

CH3CH20- + 02 -> CH3CHO + H02                      0.04

CH30- + 02 -»• H2CO + H02                           0.04

H202 + hv -> 20H-                                  1.06  x 10-3t

H2CO + hv -> H2 + CO                               2.1  x 10-3t

H2CO + hv -> 2HO^ + CO                             6.9 x lO'41"

CH3CHO + hv -v CH4 + CO                             3 x lQ-4t

CH3CHO + hv •*• CH302 + H02  + CO                    7 x 10'4t

CH3CH2CHO + hv •*• CH3CH3  +  CO                      3 x lQ-4t

CH3CH2CHO + hv -> CH3CH202  + H02 + CO              7 x 10"4t

CH302 +  NO •*•  N02 +  CH30-                          1 x 103

CH3CH202 + NO -»• N02  + CH3CH20-                    1 x 103

CH3CH2CH202  + NO -> N02 + CH3CH2CH20-              1 x 103

CH3C(0)02  +  NO + N02 + CH302 + C02                1 x  103

-------
                                                                       51
                                                  Rate Constant
	Reaction	           (ppm-1 mirr"*)

CH3CH2C(0)02 + NO + N02 + CH3CH202 + C02          1 x  TO3

CH3C(0)02 + N02 -* CH3C(0)02N02                    3 x  102

CH3CH2C(0)02 + N02 -»• CH3CH2C(0)02N02              3 x  102

CH3C(0)02N02 -> CH302 + N03  +  C02                  3 x  l(T3t

CH3CH2C(0)02N02 -»• CH3CH202  +  N03  +  C02            3 x  lO'31"

H02  + H02 * H202 +02                              6  x  103

H02  + CH302 .->  CH3OOH +  02                         3 x 103

H02  + CH3CH202 -> CH3CH2OOH + 02                   3 x 103

H02  + CH3C(0)02  -»• CH3C(0)OOH + 02                 3 x 103

 H02  + CH3CH2C(0)02 -v CH3CH2C(0)OOH + 02           3 x
 *
 §
           •
   ppm~J mm

-------
                                                                                   X ND2
                                                                                   O 03
                                                                                   X NC
    .30   T-
    .24
    .16
fO
OJ
o
o
o
    •12   ~-
    .06
  0.00
                                100  0
SO - 0
3DB-0
350-0
                                                     Time (minutes)
MC-G-O
                                            tn
                                            ro
                  FIGURE  8.   EC-11  SIMULATION RESULTS AND UCR DATA FOR N02> Og,  AND NO

-------
   .50   -r-
   .MO
I- -30
o
a
   .20   --
   .10    -
 o.oo
                                                                                     FORMALDEHYDE

                                                                                     PROPYLENE
        0-0
SQ.O       100-0
                                          15'J-O
   cDQ.O       2CU-0


Time (minutes)
                    FORMALDEHYDE
                    •	..


                    PROPYLENE







300 :o       150 :o       HO^O
                                                                                                              tn
           FIGURE 9.  EC-11 SIMULATION RESULTS AND UCR DATA FOR PROPYLENE AND FORMALDEHYDE

-------
   10   -r-
  .08    --
                                                                                 K  RCETGNE

                                                                                 X  PRN
c

-------
   .25    -p-
   .20
ex.
O.
                                                                                  CD  PCETFLCEHYCE
a
   .10
   .OS    —
 0-00
O.G
100-0
                                                      SOOiO        250.0


                                                    Time (minutes)
300 - 0
150-0
                FIGURE 11.   EC-11  SIMULATION  RESULTS AND  UCR DATA FOR ACETALDEHYDE
                                                                                                                cr
                                                                                                                tn

-------
   .20    -T-
   .16
 a.
 Q-
   .12
 o
 § .06
o
   .OH
                                                                                   CD  03
                                                                                   H  NO
                                                                                   S  N02

 0.00
        0-0
50-0
i oo.o       i50.o       200-0       ajiu.'ti        300 :o

                       Time  (minutes)
               FIGURE 12.   EC-12 SIMULATION RESULTS AND UCR DATA FOR O, NO, and
MOO . U
                                                                                                                en

-------
                                                                                     FORMALDEHYDE

                                                                                     PROPYLENE
   .10    T-
   .08 •  —
 I'.06
   .04
o
o
  .02
 C.CO
        Q.Q        5Q-0       IOG.0       15'J-O       200. Q       250 . 0       300. Q


                                                   Time (minutes)
         FIGURE 13.  EC-12  SIMULATION RESULTS AND UCR DATA FOR PROPYLENE AND FORMALDEHYDE
                                                                                                               01

-------
  .05
   .04
   .03
                                                                                 3 RCETRLDEHYK.
                                                                                 X PPN

-------
  I .50   -r-
  1.20   --
   .90   --
to


§   GO
(J  -CUI

O
   .3D    --  Z
                                                                                  X  NO
                                                                                  S  NO
 O.CC
                                                    Time  (minutes)
                                                                                                               en
              FIGURE 15.  EC-16 SIMULATION  RESULTS AND UCR DATA FOR 03, NO, AND N02

-------
o.
in
s-
CJ


c.
o
   .36
  .12
C.CQ
        G-C
SO - 0
                                                                                   X RCtTONE

                                                                                   CD cCf.TFLCF.HYCE

                                                                                   X PPN
100 0
ISQ.'O       "00-'0       250-0


         Time  (minutes)
:300.0
                                                                                                      ACETALCEHYDE

                                                                                                      PAN
?rJ-:-.l!
     FIGURE 16.   EC-16  SIMULATION RESULTS AND UCR DATA FOR ACETONE, ACETALDEHYDE, AND  PAN

-------
   I.SO    -T-
   1.20
    .90
   .
O»  '
(J
o
o
    .60
    .3D    --
  0.00
                                                                                  X  FORMALDEHYDE

                                                                                  *  PROPYLENE
                                                     Time (minutes)
         FIGURE 17.  EC-16 SIMULATION RESULTS AND UCR DATA FOR PROPYLENE AND FORMALDEHYDE

-------
   1.00   -r-
    .80
-5  -60   --
 to
-»->
 (U
 o
 o
o
.40   --
    .20   --
  0-00
                                                                                  * Propylene
                                                                                  <2) Acetaldehyde
      O.Q         SQ:Q       100:0       ISQ:C       2uc:c       250:0       300-0
                                                 Time (minutes)
                                                                                            35U-0       HOQ-0
                                                                                                                 ro
               FIGURE 18.  EC-18 SIMULATION RESULTS AND UCR DATA FOR PROPYLENE AND ACETALDEHYDE

-------
    .IB    -r-
    .12   --
,£  -09
o
•M
to
0)
o
o
    .06   --
    .03
  0.00
                                                                                  X  NO

                                                                                  X  NOz

                                                                                  O  PAN
                                                                        X  X   «  K  H __%_ X   K_
0-0        50-0       100.0
                                            15U-Q       200-0      25U-Q



                                                     Time (minutes)
300-0       350-0       MOQ-0
                                                                                                                 CTI
                                                                                                                 CO
                   FIGURE 19.  EC-18 SIMULATION RESULTS AND UCR DATA FOR  NO, NOg, AND PAN

-------
    .50    -r-
    .40
QL

Q.
    .30    --
c

-------
                                                                                      Propylene
                                                                                      Acetaldehyde
     .15    -r
     .12
^   .09
 o
•r—
4J
 10

-------
                                                                                X NO
                                                                                X N02
                                                                                X Formaldehyde
                                                                                                  FORMALDEHYDE
o.oo
                                                                                                               en
             FIGURE  22.   EC-21  SIMULATION  RESULTS AND  UCR DATA FOR NO, N02, AND FORMALDEHYDE

-------
     .010  -.-
     .008 --
a.
Q.
IO
O>
U
    .006
    .004
    .002
  0.000
                                                                                   2°3
                                                                                   X PAN
        0-0        50.0       IDO.O
   aoc.'o


Time (minutes)
                                                                               300 :o
40". C
                     FIGURE  23.  EC-21 SIMULATION RESULTS AND UCR DATA FOR 03 AND PAN

-------
                                                                      68
     PAN predictions  are sometimes  high  (EC-18 and  EC-11) and sometimes  low
(EC-16 and EC-12).   Calibration  of  the panalyzer  has  been an enigma;  thus,
data values may be  suspect.   But the  chemistry of PAN is still  largely unex-
plored, and this is very likely  a large  part  of the problem.

     The results for EC-12 are inexplicable.  The data indicate a  total
absence of an induction period,  but mechanistic predictions  completely
disagree with this  observation.   If the  current mechanism's  validity  is
accepted, this completely disparate behavior  can  be resolved only  by  blam-
ing wall effects.  The reactant  concentrations in EC-12 are  extremely low,
and the wall effects are hence,  relatively, at their  maximum.   A small
source of radicals, such as  photolysis of off-gassed  aldehyde or HN02,
could provide the initial impetus to  this system.*  A recent experiment
done at UCR provides further evidence in favor of this explanation.   CO
oxidation to CC^ was observed upon  irradiation of a supposedly  clean  chamber
filled with dry synthetic air.  The oxidation rate  of O.I  ppm  per hour
with CO present at  100 ppm indicates  that the OH- concentration in this
system could have reached ^10"^  ppm.   Presumably, this was  due  to  a wall
source of radicals.

2.   The Butane-N0x-Air System

a.   The Mechanism Used

     The butane mechanism was considerably less  successful  than the propylene
mechanism, even though they are  similar. . The modifications to  the propylene
mechanisms to adopt it to butane included adding  reactions  pertiment  to  butane
and reflecting changes in k^  Photolysis rate  constants used  in the  butane
simulations were computed from a 10 nm filter-shifted spectrum  (as with  propene)
The reaction changes are listed  in  Table 7.   Most of  the changes included in
Table 7 represent merely the addition of four-carbon  species.   Most notable is
the ratio of 2.4 to 1 for internal  to terminal  hydrogen abstraction by OH« from

*  In  these  simulations,  initial  concentrations  of 0.01 for formaldehyde and
   acetaldehyde were used.  The  carbon balance obtained from product measure-
   ments  (excluding CO)  at UCR is usually 60 to 80 percent.  However, in EC-12,
   this  balance was about 110  percent—hence, the suspicion that unaccounted-
   for sources  (or  initial concentrations) were present.

-------
                                                                                       69
                                   Table  7



               REACTION CHANGES  HADE  FOR  THE  BUTANE SIMULATIONS



(With the propylene chemistry eliminated  and  all  photolysis rates scaled to k,)



                       Reaction                                Rate Constant
                                                                    1     1    -1
   C4H1Q + OH-  —*• H20 + C4Hg02                            1.0 x 10J ppm"1 min




                °9                                                  Til

   C.H10 + OH-  -^~H20 + CH3CH2CH(02)CH3                   2.4 X ID"3 ppm"1 min




              °9                                                   11

   C,H,n + 0 —^*-CH,CH,CH(0:)CH, + OH-                     64 ppm"1 min"1
    4 1U            3  L.    t.   *5



                       °9                                           All

   CH3CH2CH2CHO + OH- —£- H20 + CHjCHgCHgCtOjOg            4.5 x 10* ppm   min"1




                        0,                                        All

   CH3CH2C(0)CH3 + OH- -4- CH3C(0)OH + CH3CH202            1 x 10A ppm"1  min"1





   CH3CH2CH2CH20-  —*- CH3CH2CH202 + HgCO                   8.5 x 103 min"1
   CH3CH2CH2CH20- + 02  —- CH3CH2CH2CHO + H02              0.2  ppm"1  min"1





   CH,CH9CH(0-)CH. + 0?  —*- CH3CH?C(0)CH, + HOj             0.2  ppm"1  min"
     Ob        O    t        we.      «3t




   CH3CH2CH(0-)CH3 —»•  CH3CH202 + CH-jCHO                    1.3  x 104  min"1



                       20.

   CH3CH2C(0)CH3 + hv  —^ CH3CH202 +  CH3C(0)02             7.8  x 10   min"





   CH3CH2CH2CHO + hv —*- CH3CH2CH3 + CO                     3.1  x 10"4 min"1




                       202                                            -4    -1
   CH3CH2CH2CHO + hv  —^ CH3CH2CH202  + H02 + CO           7.2  x 10 H min





   CH3CH2CH2CH202 + NO —»- N02  +  CH-jCHgCHgCHgO-             1.0 X 103 ppm"1 min"1





   CH3CH2CH(0')CH3 +'NO —•-  N0£ + CH3CH2CH(0-)CH3          1.0 x 103 ppm"1 min"1



                         v/\                                         *i     1      1

   CH3CH2CH2C(0)02 + NO  -^ N02 + CH3CH2CH202 +  C02        1.0 x 103 ppm"1 min"1





   CH3CH2CH2C(0)02 + N02 —•-  CH3CH2CH2C(0)02N02             3.0 x 102 ppm"1 min'1




                        °2                                            3     -1
   CH3CH2CH2C(0)02N02  -^ CH3CH2CH202  + N03 + C02          3.0 x 10° min





    CH3CH2CH2C(0)02  +  N02 —^ CH3CH2CH2C(0)02H + 02         3.0 x 103 ppm"1 min"1

-------
                                                                      70
C.HIQ.  This is lower than the  ratio  for abstraction  by  Cl  atoms  (Morrison  and
Boyd, 1971), although one  might expect  it to  be  higher because OH-  has  less
electron affinity than Cl  (Hefter et  al., 1972)  and should  therefore  be more
selective in its attack.   A ratio measurement is desirable.

      Another element of Table 7 worth commenting on is the reaction of OH-
with  CH3CH2C(0)CH3 (methlyethylketone, or MEK) .  The rate of hydrogen abstrac-
tion  from MEK  can be computed from Greiner's formula for paraffins (Oohnson
et  al.,  1970), with a modification to take into account the lowering of the
secondary C-H  bond dissociation energy.  Scaled to the current value of
3.7 x 103 ppnf1 min'1 for  C^g + OH-,  the rate constant for CH3CH2C(0)CH3
+ OH-  •* abstraction products is about 2  x 103 ppm~^ min~^.  However, an
analogy with propylene suggests  that addition to the C-0 double  bond may
be  a  faster reaction.  Although  repeated experiments were performed by UCR
 (i.e., experiments with the same  initial N0x/butane  ratio), we chose only
one experiment for each of the  different NO  /butane  ratios.
                                           A

                                          0-
                                           I
                     CH3CH2C(0)CH3 ->  CH3CH2C(OH)CH3                  (32')

 followed by
                    CH3CH2C(OH)CH
 The net products are acetic acid and an ethyl peroxy radical.  The lower
 limit to k32' was taken to be 2 x 103 ppm'1 rnin'1, and 1 x 104 ppnT1 min"1
 was used in the simulations.

      The carbon balance (excluding CO and C02) of the butane data was usually
 about 20 to 40 percent, which indicates undetected carbon-containing species.
 Acetic acid may have been one of them.

 b.   Results

      The initial concentrations and values of k] for the butane-NOx block
 (Figure  24) are recorded in Table 8.  Model predictions, along with UCR  data,
 appear in  Figures  25 through  39.  Generally speaking, the mechanism predicts

-------
                                                                        71
an overall  rate greatly in  excess  of  that  observed,  the  only  exception  being


the low butane run,  EC-42.   Excluding EC-42,  the  universal  over-prediction  of
1.0
0.
c
o

-------
                                                                                      NO
                                                                                      03
                                                                                      N32
    .40
   .30
o
U
<=
O
o
.20
   .10   --
                         K
 3.00
        0-0
SQ.Q
                                                                            Si  Z
                                                                                                     NO,
100.0
                                              -Hi.
                                                    800-'Q  .

                                                 Time  (minutes)
              FIGURE 25.  EC-39 SIMULATION RESULTS AND UCR DATA FOR 03, NO, AND N0£
                                                                                                               ro

-------
   5-00   -T-
£3.00
o
aj -
o 2.00

o
  1 -00
  O.GG
                         *  X   W
                                                                                      *   X  X   X
         0-0        50-Q       100-0
150-Q       200-0       25U-Q

          Time (minutes)
                                                                                                                   CO
                       FIGURE 26.   EC-39 SIMULATION RESULTS AND UCR DATA FOR BUTANE

-------
                                                                                   X
                                                                                      KCfTRLCF.HYDf
    .10    -T-
    -OB
I  -06
10
o
o
 O.CQ
        0-0
SU-Q
                                                                                  FORMALDEHYDE



                                                                                  ACETONE
                                                                                   x
1.00-Q
151). 0
   COO - 0


Time (minutes)
300-0
             FIGURE 27.  EC-39 SIMULATION RESULTS AND UCR DATA FOR ACETALDEHYDE AND FORMALDEHYDE

-------
                                                                                   x
    .10    T-
    -08   •--
O
•r-1

«S
£ i. Of
O
O
   .02    --
 O.C!G    J
        o-o        so.o        IQQ:O       iso-'a       200:0       250 :o       .300:0       3?:o:c       HCC.Q

                                                    Time (minutes)
                   FIGURE 28.  EC-39 SIMULATION RESULTS AND UCR DATA FOR MEK AND PAN

-------
   .60   -r-
   .MB
A :-36
 •»->

 o
 o
 o
    .24   ~-
    .12
   i-CD
          3-t
                                                                                     CD  1)3

                                                                                     2  NG2

                                                                                     X  NC
iSO-0       201? .'0        250 .'0


         Time  (minutes)
oro o
                                                                                                   0
HOt
                   FIGURE 29.  EC-41 SIMULATION RESULTS  AND UCR DATA FOR 03, NO, AND N02

-------
   S.OO   -i-
   4-00
S" 3-00
  ;2.oo
   1.00   --
   O.OD
                     so .'Q
15Q.Q       ?OQ-Q       2SU-Q
         Time (minutes)
433.0
                     FIGURE 30.   EC-41 SIMULATION RESULTS AND UCR DATA FOR BUTANE

-------
 .20
  -IS

-------
                                                                                    5K FQRf1SL.DF.WDE
                                                                                    VD FiCcTRLCF.HYuF
   .10    -r-
   .08
  t:-oe
 (O

+»'.
 C
 o> •
 o •
 C'
 o •
CJ •
  .02    ~-
 a.cc
                                                                         FORMALDEHYDE

                                                                        - ACETALDEHYDE
                   GO-Q
IOQ.Q
150-Q
    EQQ.Q       2SU-Q

Time (minutes)
3QQ-Q
                                                                                                                        to
             FIGURE  32.   EC-41  SIMULATION RESULTS AND UCR DATA  FOR ACETALDEHYDE AND FORMALDEHYDE

-------
                                                                                  *  BUTHNf
                                                                                  X  ND2
                                                                                  X  NO
   -6U   -i-
   .36
01
o
o
   .24
   .12   --
 c.oo
                                                  BUTANE

                                                  NO
        Q.O        50-0        100.Q        1SO-0
   SQQ.Q       250.Q

Time (minutes)
3'JJ.O       SDD-C
                                                                                                                 00
                                                                                                                 o
               FIGURE 33.   EC-42 SIMULATION RESULTS  AND  UCR  DATA  FOR BUTANE, NO, AND NO,

-------
   .05   -T-
   .04 •••
   .03   --


o
o
   .01
 a.ca
        o-u
                                                                                  X  ME*
                                                                                                      FORMALDEHYUh

                                                                                                      MEK
                                                                                                      ACETALDEHYUb
                                                     s ....
                                                                    z


                                                               .-—-"ft"
SU-Q
100-0
150-0       COO-0


         Time (minutes)
                                                                                                                   00
         FIGURE 34.   EC-42 SIMULATION RESULTS AND UCR DATA FOR MEK, FORMALDEHYDE, AND ACETALDEHYDE

-------
     .010 T-
     ,008  --
 £
 &.  .006



5
(O
o    .004

o
o
                                                                                     03
    .002 - -
                                                     CD
   0.000
        0.0
50-0
100-0
                                                   -f
150-0       200-0       250 .'3



         Time (minutes)
3QO-0
35Q • 0
—.1 n
>. v.1 • v.r
                                                                                                                 CO
                                                                                                                 ro
                     FIGURE 35.   EC-42 SIMULATION RESULTS AND UCR DATA FOR OZONE

-------
                                                                                 W NO
                                                                                 X N32
                                                                                 VD PCFTRLGEHYDf
 I .GO   -r-
O.GO
Q-0
                   :O.Q        103.0
3?:3-0       4.23-0
                                                   Time (minutes)
                                                                                                                 oo
                                                                                                                 oo
           FIGURE  36.   EC-44 SIMULATION RESULTS AND UCR DATA FOR NO,  N02,  AND ACETALDEHYDE

-------
    .36    --
CJ   tn
o   La
   .09   - -

                                                                               O
                                                                   250.0
300-0
                                                    Time  (minutes)
                                                                                                                 00
                      FIGURE  37.   EC-44  SIMULATION  RESULTS AND UCR DATA FOR OZONE

-------
                                                                                   fc RUTflrtF.
   s.on   -i-
   4.80
   3.60
a
t-
S  2.MQ

o
   1 .20
    .00
          o-o         so.o        IOQ.Q
  200-0       <


Time (minutes)
                    FIGURE 38.  EC-44 SIMULATION RESULTS AND UCR DATA FOR BUTANE
                                                                                                                 CD

                                                                                                                 01

-------
    .16
o
•r~
•P

J_
•M


-------
                                                                      87
formaldehyde shows that the  chain  processes whereby butane  is  broken  down
into oxygenated products are improperly  represented.   Correction  of this
fault must await more complete product measurements and  improved  kinetic
data.  Because rate constants for  reactions of most of the  less-than-four-
carbon radicals were determined from propylene simulations, it is possible
that 04 radical reactions are inadequately represented,  but this  conclusion
cannot be drawn on the basis of the results alone.

     Of course, the traditional scapegoats, uncertainties  of surface  and  photo-
lysis reactions, can also be blamed for  the disparity between model  predictions
and experimental results.  Although they probably exacerbate the  problem,  they
are not the sole culprits.  The results  of EC-44 are  reproduced in Figures
40 through 43, with k-|Q and k-|] (NO + N02 + H20 -> 2HN02) reduced  by  a factor
of 10.  The reaction is greatly delayed, and  the agreement with the  data
is improved.  However, once the reaction begins, the  net rate of  NO  oxidation
is obviously still too fast.

B.   SIMULATIONS OF BATTELLE DATA

     To clarify the interaction of S02 with  hydrocarbon-NOx-air  pollutants,
EPA has sponsored a series of smog chamber experiments performed at  Battelle
Memorial Laboratories in Columbus, Ohio.  We have received the results of five
of these experiments and report here on the kinetic simulation of those data.

1.   Instrumentation

     The Battelle simulations were carried out  in a  17.8 cubic meter environ-
mental chamber.  The surface-to-volume  ratio of this  chamber  is  2.6  meters"1.
Ozone concentrations were measured using  a chemiluminescent method;  ethylene,
NO,  and N02 were measured using an automated Saltzman method;  S02 using a
Beckman 906 analyzer; and propylene  using gas  chromotography  with a  flame
ionization detector.  $03 concentrations  were  inferred  from aerosol  size
distributions  measured  with  a  Thermo Systems electrical aerosol  analyzer,  by
assuming  that  equilibrium existed  between sulfuric acid aerosol  in  the con-
densed and  vapor  phases., This method was tested by  independent  chemical

-------
                                                                                 X  NO

                                                                                 X  NOi!
  1.50   -r
  1.20
   .90
 o
£
o
o
   .60   --
   .30    --
 Q.CO
        Q-Q
          5U-0
loo-o       isu-'u       200:0       2so :o

                     Time (minutes)
300-0
FIQURE 40,   EC-44 SIMULATION RESULTS,  USING  kln =  1.3 x 10"12 ppm"3 and k,, = 0.024 ppm"1 min"1,
                     FOR NO AND NO,
                                                                                                               00

                                                                                                               00

-------
  .05    T-
O.CQ
                                                                  25U-0
300-0
                                                                                  350-'0
40C-0
                                                   Time (minutes)
FIGURE 41.  EC-44 SIMULATION RESULTS, USING k1Q = 1.3 x 10"12 ppm"3 min"1
            AND k,,  = 0.024 ppm"'  min"1, FOR OZONE
                                                                                                               CO
                                                                                                               <£>

-------
    6.CQ   -r-
     .80
                                                                                      BUTRNE
o.  3.60   - -
    2.MD
c
o
    1.20   - -
   •0-00
           0-0
50-Q       1.00-0
  eou-o


Time (minutes)
                                                                                 300-Q
                 FIGURE 42.   EC-44  SIMULATION RESULTS,  USING  Iqo  =  1.3 x  TO"12: ppnf3 min"1
                             AND  k    =  0.024  ppnf  min",  FOR  BUTANE
                                                                                                                  «£»

                                                                                                                  O

-------
                                                                                  X
    .IS   -r
    .12   --
.09   --
-M
c
OJ
u

o
o
.06   --
    .03   --
   o.co
          0-Q
                5U-0
100-0
15U:0       200:Q       25U-Q


          Time (minutes)
                                                                                300-0
                                                                                                     MEK
                                                                                                     FORMALDEHYDE



                                                                                                     ACETALDEHYDE
                                                                                        350 0
           'FIGURE 43.   EC-44 SIMULATION RESULTS,  USING  Mo  =  1.3 x 10'12 ppm'3 min'1 AND

                       . kn  = 0.024 ppra-1 min-1,  FOR MEK,  ACETALDEHYDE, AND FORMALDEHYDE

-------
                                                                    92
measurements of S03 and found to be fairly accurate.   The  N02  photodissocia-
tion constant, kd, was determined from photocell  readings, the photocell
having been previously calibrated.  The relation  k]  = 4/9  kd  (Wu  and  N1ki,
1975) has been used in our model.  Other instrument  and  chamber character-
istics are summarized by Table 9 (Miller 1975).

2.   The Mechanism Used

     The propylene mechanism used in the UCR simulations forms the core  of
the mechanism used to simulate the Battelle data.  The reactions  listed  in
Table 10 were added to this core.  In the absence  of adequate  light spectrum
information, photolysis rates were computed from  the unshifted ratios included
in Table 3.  Additionally, a rate constant of 2 x  10'11  ppnr3  min'1 was
assigned to the heterogeneous reaction of NO, N02, and H20 (Reaction  10,
Table 1), 0.4 ppm'1 min"1 was assigned to the reverse Reaction (11),  and
       c     -1     i                                           \   / >
1 x 10'  ppnr1 min'1 was assigned to the heterogeneous reaction of N205  and
H20 (Reaction 9).   This procedure is in accord with  the  discussion of heter-
ogenous HNOX chemistry in Chapter II and was used  solely to improve N0?
profile predictions.

     A recent rate constant measurement of 9.0 x  103 ppm'1 min'1  (Cox, 1974;
Atkinson, 1975) for the reaction OH- + NO + HN02 was  used  in these simula-
tions (but in none of the others).*

3.   Results and Discussion

     The initial concentrations and value of k] for  the  propylene-NOx-S02
block are contained in Table 11.  Two experiments  contained no S02 (S-114
and S-115); three  did contain S02 (S-107, S-110, and  S-113).   Of  the  three
containing S02, S-113 was performed with reduced light intensity.   All
photolysis rates for model predictions were scaled down  accordingly.

* These new rate constant determinations were called to  our attention late
  in the contract year   Although their incorporation in the Battelle simulation
  2p?L9?S thpSNOtS^eirTh-feCt Vhe UCR Sl™la«°ns ^s  an undesiraE e
  delay in the N02 peak.  This result was expected because rate constant
                                                              consan
adjustments, made to give the best fit to UCR data,  correspond to the  lower
^SnS A* 10Tu?pnrT mirr1'  The OH< + N0 Action is  simply the reverb
of HNOz + hv.  This suggests that a modification of  heterogeneous  HNOp
                      ijf uncertainty bounds) would compensate the delayed

-------
                                                                       93
                               Table 9
             ANALYTICAL AND CHAMBER CHARACTERISTICS  OF
                   THE BATTELLE EXPERIMENTAL SETUP
                   (a) Analytical  Characteristics
Analysis
°3
N02
NO
so2
C3H6
S03
Dew point
Temperature
(b)
Lag time Dark time
(sec) (sec)
< 4 < 4
•v 500 % 20
* 500 ^ 20
% 300
—
—
< 5
< 1
Chamber Characteristics
Criteria
N02 photodissociation
kd, 0.38 min
k,, 0.16 min"

Uncertainty
Factor
0.1
0.2
0.2
0.15
0.15
0.5
0.1
0.05

Uncertainty
Factor
0.15
0.15
03 half-life (may vary with conditions)
Dark, 6-8 hrs
Light, 3-4 hrs
Dilution (variable with


sampling)
0.25
0.25
0.1-0.25
S/V ratio, meters =2.6
Background conditions (slightly variable)
  CH4 < 2 ppm
  Nonmethane hydrocarbons < 0.2 ppm C
  CO < 3 ppm

-------
                            Table 10
                                                                     94
             ADDITIONAL  REACTIONS USED IN SIMULATING
               THE  S02-OXIDATION DATA FROM BATTELLE

                 Reaction
S02 H
S02 H
S02 H

S02 n
H HO; — *- so, + OH.
£ 0
f- CH0OA — »- SOo + CH-jO-
32 33
3 C- C. r\ ^ 3 £• £
OP
1- CH,CH9C(0)02 -=*- SO^ + CH.CH90; + C09
3 L. •• *3 0 ^ c c
so2 +
HS05
HS04
HS04
S00 +
     CH3C(0)0'


     OH-  -X-
    «- NO —*- NO
                )3  + NO,
H02
N00
           H90
                        HN0
              H2S04
Rate Constant
(ppm   min" )
       1.3
       1.5
       1.5
       1.5
       1.5
      14
         2
   9 x 10^
   8 x 102
   9 x 103
   1 x 10
 Immediate
     Predictions and experimental  data are  illustrated  in  Figures  31  through
51.  As shown in these figures,  the inability  of  the mechanism  to  follow  the
propylene data—in sharp contrast  to the UCR propylene  simulations—is  somewhat
disconcerting.  In the early stages, the data  are followed quite closely,  but
the mechanism shows gradually tapering decay asymptotically, whereas  the  measured
propylene concentration drops off  rapidly to zero.  The NO and  S02 predictions
are fairly accurate.  Since the  latter is relatively unreactive, a sizeable
fraction of its disappearance is due solely to dilution.  The dilution  rate
was approximately 10 percent per hour.  Thus,  65  percent of the S02 disappear-
ance in S-107 and S-110 and 87 percent in S-113 was due to dilution.

-------
    i .GO    V
 Q.
 Q.
C
o
id
u
C
o
o
      .80
      .60
      .MO    --
      .2D    --
    0-00
                                                                                       Propylene
                                                                                       N02

                                                                                       S0
0-0
50-0
10Q-0
15U-U
                                                         200-0
                                   25il.U
300-U
350
                                                                                               40G.U
                    Time (minutes)




FIGURE 44.  S-107 SIMULATION RESULTS AND BATTELLE  LABS

           DATA FOR PROPYLENE, N02>  AND S02
                                                                                                                  UD
                                                                                                                  cn

-------
      .MO    -i-
i.
D.
C
o
•f—
4J
fO
S-
Ol
o
C
o
o
      .32
      .16   --
      .08   --
     0.00
                                                                                         3

                                                                                     X NO
                                                                                        S03 Aerosol
                       5U-U
100.U
isu-i
                                                          200 o
                                                      Time  (minutes)
                                    250-0
                                    300-U
                                                                                               350-D
                                                                         H'JO-Q
                                                                      iO
                                 FIGURE 45.  S-107 SIMULATION RESULTS AND BATTELLE LABS

                                            DATA FOR 03, NO, AND S03 AEROSOL

-------
                                                                                    * Propylene

                                                                                    X N02

                                                                                    X S0
    1.50    -r-
o



s_
o
c
o
      .60
    0.00
50-U       lOO-U       15U-U       200-U       2SU-U       30Q-U       350-U       HOQ-U



                              Time  (minutes)




           FIGURE 46.  S-110  SIMULATION RESULTS AND BATTELLE LABS

                      DATA FOR PROPYLENE, N02, AND S02
                                                                                                                  vo

-------
       .50   -r-
       .40   -r
 Q.
 Q.
       .30   --
O)
o
c
o
CJ
       .20   --
       .10   --
     0.00
             0-0
                                                                                     •o  03

                                                                                     X NO
                                                                                        S03 Aerosol
                                             o      o
                                                  o    o o o        o
                                           O                    (DO
100-0        150-0       200-0       250-0       300-0       3CJO-U       HuQ-0



                    Time (minutes)



FIGURE 47.  S-110 SIMULATION RESULTS AND BATTELLE LABS

           DATA  FOR  03,  NO, AND S03 AEROSOL
                                                                                                                   00

-------
     1.00   -r-

-------
      .32   --
      .21*    —
o>
o
c
o
.16   —
     .08    --
    0.00
                                                                                       0.
                                                                                        3

                                                                                    W  NO
                                                                                       SO- Aerosol
           0-0
                                                                                    o
                5U-0       iOO-Q
ISd-U       200-U



      Time (minutes)
25U-0
300-0       350-0       400-0
                                 FIGURE 49   S -113 SIMULATION RESULTS AND BATTELLE LABS
                                         'DATA FOR o3, NO, AND so3 AEROSOL
                                                                                                                 o
                                                                                                                 o

-------
    1.00   -r-
     .80
     .60   --

-------
Q.
03
O
c
o
    0.00
                                                                                      *  Propylene
                                                                                      o  o3

                                                                                      H  NO
                                                                                      S  NO,,
                                                                       250-U
300- 0
350-U
400-Q
                                                      Time (minutes)


                                  FIGURE 51.   S-115 SIMULATION RESULTS AND BATTELLE LABS
                                            DATA FOR PROPYLENE, 0^ NO, AND N02

                                          (Thir, experiment w.r. 
-------
                                                                      103
     The N02 predictions are good, but it is only fair to reiterate that
heterogeneous rate constants were "tuned" to this curve.  It was impossible,
however, to fit all the data by adjusting these reactions.   In S-107 and S-110
post-peak N02 predictions are high, whereas in S-113, S-114, and S-115 they
are low.  03 is always low, usually by at least a factor of 2.  This could be
due simply to under-prediction of propylene oxidation (and  consequently over-
prediction of NO), though such a large difference is hard to explain.  S03
appears too late and in too great a quantity, being about a factor of 2
too high in both S-107 and S-110.  With proper propylene oxidation and 03
production, the discrepancy in $03 yield would undoubtedly  grow even worse,
but the time delay would be shortened.  One possible source of error could be
an overestimate of the rate of reaction of N03 with S02.  The rate constant
of 14 pprrf1 min'1 is, after all, Davis1 (1974) upper limit.  Table 12 sum-
marizes the magnitude of the various modes of S02 oxidation as a function
of time.  N03 does not account for more than 10 percent of  the net oxidation
rate until after 3 hours.  Then it accounts for one-fourth  to one-third
of the total rate.  Since the discrepancy between the data  and predictions
appears in the form of an overshoot late in the reaction, N03 is a likely
culprit.  Thus, a measurement of the N03 + 862 rate constant is needed.

                             Table 11

               BATTELLE PROPYLENE-NO-S09 EXPERIMENTS
                                    A   t
                 Initial concentrations of Reactants
Run No.
S-107
S-110
S-113
S-114
S-115
NO
0.328
0.392
0.409
0.414
0.417
N02
0.113
0.099
0.099
0.095
0.108
so2
0.474
0.480
0.482
0
0
Propylene
1.03
1.10
0.96
0.95
0.97
kl
0.175
0.177
0.079
0.172
0.176

-------
                                                                             104
                  Table }?.
RATCS OF SO? OXIDATION  UY VAHK)!!', OXtDAtiTS
        (From the-  S-107 Simulation)
Tlmo
(min)
CO





120





180





240





300





360





Oxidant
OH-
ROj
RC(0)02
HOJ
N03
Total
OH-
R02
RC(0)02
H02
N03
Total
OH-
R02
RC(0)02
HOJ
N03
Total
OH-
R02
RC(0)02
HOJ
N03
Total
OH-
R02-
RC(0)02
H02
N03
Total
OH-
ROj,
RC(0}02
HOg
N03
Tottl
Rate
8.8 x 10-5
5.6 x 10-5
6.1 x lO-6
1.1 x 10"4
. 7.2 x 10-6
2.7 x lO'4
4.5 x ID'5
1.4 x 10'4
1.8 x 10-5
1.8 x 10'4
3.2 x 10-5
4.2 x lO"4
2.1 x 10-5
1.9 x ID'4
3.8 x 10-5
1.2 x 10'4
4.9 x ID'5
4.2 x 10'4
1.1 x ID'5
2.4 x ID"4
8.0 x 10'5
6.8 x ID'5
1.2 x ID'4
5.2 x 10-*
7.1 x ID"6
1.7 x 10"4
2.1 x ID'5
4.6 x 10-5
9.5 x ID'5
3.3 x lO"4
5.4 x 10-6
1.2 x 10'4
2.3 x ID'5
3.3 x 10'5
6.6 x ID"5
2.5 x 10-<

-------
                                                                      105
     Another possible source of the $03 discrepancy could be an unmeasured
buildup of products on walls.   Hence, not all  863 would be contained in
gas phase aerosols.  In the current mechanism, only the reaction
HSO^ + N02 ->• H2S04 + N0,3 (Reaction 56) was assumed to  leave products on the
walls.

C.   SIMULATIONS OF THE NAPCA TOLUENE-NOX DATA

     The NAPCA data served as a base for our previous two years of mechanism
development (Hecht et al., 1974b; Hecht et al., 1973).  The latter reference
describes the chamber and experimental techniques and discusses the data.

1.   The Mechanism Used

     The mechanism for toluene oxidation, described in Chapter II, is pre-
sented in Table 13.  The inorganic reactions,  not included there, are the
same as those in Table 6.  The formaldehyde chemistry was also taken from
Table 6.  The photolysis reactions were adjusted to kj = 0.266, and the fol-
lowing changes were made in heterogeneous HNOX chemistry:

                           kg = 1 x 10"3
                          k1Q = 3 x 10"11
                          k^ = 6 x 10"1

kg is now Jaffee and Ford's (1967) value, and k^g has been brought to within
a factor of 3 of Noeh et al.'s (1974) value for a metal surface.

2.   Results and Discussion

     The results of three simulations are displayed in Figures 52 through
57.  The N02 data for EPA-272 (Figures 54 and 55) between 100 and 300 minutes
appear to be erroneous, probably as a result of instrument failure.  The
Initial  concentrations and values of k, are given in Table 14.

-------
                                                                                         106
                             Table 13


                    TOLUCNK OXIDATION ML'CHAIIISM


                                                           Rate Constant

	B.yp.y.SP	           (PP"|   min'')



CgH5CH3 + OH- -£- CgH5CH202 + H.,0                          4.5.x 103


               0,

CgH5CH3 + OH- —C C6H4(CH3)(OH)  + H02                      4.0 x 103




CgH4(CH3)(OH) + OH-  _^c5H4(CH3)(0-){02)  +'H20            3.0 x 10*





C6H5CH3 + OM> —^"C6H5OH + CM3°2                           7'° x 10?


             °2                                                     /,
C,H,OH + OH -i- C,H,(0)0, t H,0                             l.OxlO4
 65        -652    .,



               °2                                                   a
 gHgCHO + OH'  -^ CgH5C(0}02 + H20                         1.0 x  104
                                                                   it
                                                                   4
           °2
C6H5CH20- -£+ C6H50^ + H2CO                                1.5 x  10




CgHjCHgO- + 02 -. CgHjCHO + H02                              0.4




CgHjCH^ + HO - CgHjCHgO- + NOg                            1.0 x  103


                     0,                                             ,

CgH4(CH3)(0)02 + HO -i- N02 + CgH3fCH3)(0)0 *  HO^           1.0 x  103


                                Me-quinone



                0?                                                  3
CgH5(0)0^ + HO — £- N02 + CgH4(0)0 + HO^                    1.0 x  10


                          Qulnone
C6H5°2 + K0 - N02 * C6


                 0,
CgH5C(0)0^ + NO — ^-N02 + CgHjO^ + C02                     1.0 x 103




CgH5C{0)02 + N02 - C6H5C(0)02N02(PB2N)                      3.0 x 102


               °7                                                    »
C6H5C(0)02N02 -i- CgH50^ + N03 + C02                       3.0 X 10"3




HOj + C6H5C(0)Oj - CgH5C(0)02H                              3.0 x 103




H02 + C6H5CH2°2 "* C6H5CH2°2H * °2                           3.0 x 103




H02 + C6H5°2 " C6H5°2H + °2                                 3'° x 1()3



HOJ + C6H5(0)Cj -. C6H5(0)02H                                3.0 x 103




H02 * WCH3W°2 - C6H4(OH3)(0)02H                      3.0 x 103

-------
    .50    -r
    .'-10
1  -30
£
+*
t~
g   20
    .12    --
  C.OO
                                                                                  X  ND
                                                                                  X  NC?
                                                                                  •3  03
                                                     Time (minutes)
              FIGURE 52.  EPA-258 SIMULATION RESULTS AND NAPCA DATA FOR NO,  N02,  AND

-------
   S.CG   -T-
    H.CO   --
g, -3.CO    --
ta



g  2.0C
o
o
    1 -GO
    O.J
           0-0
150-0       200-Q
                                                                     2SQ-Q
3GQ-0
                                                      Time (minutes)
                    FIGURE 53.  EPA-258 SIMULATION RESULTS AND NAPCA DATA FOR TOLUENE
                                                                                                                  o
                                                                                                                  00

-------
    • --Jv*     **
     .MO    --
 o.  .3D
§  .20    - -
o
o
    .Id
                                                                                     x
                                                                                     X  NQ
                                                      Time (minutes)
              FIGURE 54.  EPA-272 SIMULATION RESULTS AND NAPCA DATA FOR NO, N02» AND 0.

-------
    2.en   -r-
 £ ' 1.20   - -
 2
•p

 or
 o
 cr
 o
o
                        fc	nri
           0-0
50-0
100-0       150-0       2DO-Q       ;



                      Time (minutes)
                                                                                 3DO-0
                   FIGURE 55.   EPA-272 SIMULATION RESULTS  AND NAPCA DATA FOR TOLUENE

-------
£
-p

01
u

o
o
    O.CD
                                                                                     X  KC

                                                                                     X  NC2

                                                                                     13  03
    KSO    -r-
     1 -SO
      .90
.60
     .30
                                                        Time (minutes)
                FIGURE 56.   EPA-305 SIMULATION RESULTS AND NAPCA DATA FOR NO, N02>  AND 03

-------
                                                                                          TOLUENf.
     S.CO   -r-
     t.CD    --
5.   3.CO    --
IO

4->
C
o>
u
c
o
o
? O
C. -l_.L
     1-CD    - -
     C -Ci
              .
                        50-0
                              ICQ-Q
   2DQ-Q       250-0


Time (minutes)
                     FIGURE 57.   EPA-305  SIMULATION RESULTS AND NAPCA DATA FOR TOLUENE
                                                                                                                      ro

-------
                                                                      113
                             Table  14
            INITIAL  CONCENTRATIONS  AND  VALUES  OF  kn  FOR
                THE  NAPCA TOLUENE-NOX EXPERIMENTS
EPA Run No.
258
272
305
NO
0.33
0.30
1.36
N02
0.04
0.04
0.08
Toluene
2.88
1.10
3.14
kl
0.266
0.266
0.266
     The current mechanism cannot match the observed toluene oxidation rate,
and, as a result, it under-predicts ozone yields.   The hypothetical chain
processes for ring oxidation to quinones and methyl group oxidation to benzalde-
hyde and PBzM are apparently too short.  Product analyses currently being
performed at UCR may help to clarify the true mechanism of toluene oxidation.

-------
                                                                     114
                  IV  HYDROCARBON  REACTIVITY
A.   SURVEY OF REACTIVITY MEASURES

     The concept of "reactivity"  is  familiar to  every  chemist, yet  a  pre-
cise definition of the term can often be very difficult.   While  one might
intuitively feel that a given species is more reactive than  another,  quan-
tification of reactivity requires careful  definition of (1)  the  physical
conditions of the system in which reactivity is  being  determined,  (2) the
concentrations of the reactants,  and (3) the time scale of the reactions.
If we restrict consideration to the  photochemical reactions  occurring in
smog chamber experiments, the apparent reactivity of hydrocarbons will  de-
pend on the physical characteristics of the chamber system,  e.g.,  light in-
tensity, temperature, and surface-to-volume ratio.  The rate of  disappearance
of the hydrocarbon (or the appearance rate of products) will also be  a func-
tion of the concentrations of (1) the hydrocarbon, (2) the oxidants of the
hydrocarbon, and (3) the other species that either react with or lead to the
formation of the oxidants.  And, since the chemical state of the system
changes continuously, a measure of reactivity based on instantaneous  rates
of formation or disappearance of chemical species will consequently be a
function of time.

     We summarize below the criteria that have been used to quantify reac-
tivity.  In considering these measures, we assume that the physical state
of the system, the initial concentrations of reactants, and the time scales
are all suitably defined and controlled, so that a meaningful comparison of
the reactivity values for different  hydrocarbons can be made.  The measures
fall into  three classifications based on the physical  dependence of  the index:
temporal,  concentration, or combined temporal and concentration criteria.

-------
                                                                     115
1.    Temporal  Measures

     Many indices have  been  constructed  to  characterize reactivity in terms
of the temporal  occurrence of chemical or associated physiological events
during the smog  formation process.

     The most common of these include:

     >  The time of the N0£  peak (Tmax)  (Altshuller and Cohen,  1963).
     >  Tmax for oxidant or  ozone (Altshuller and  Bufalini,  1971).
     >  The time required for one-half or  one-quarter  of  the initial
        hydrocarbon to be oxidized (Altshuller and Bufalini, 1971).
     >  The threshold time for eye irritation (Heuss and  Glasson,  1968).

Other criteria that might be considered are the threshold times for  "harmful"
effects measurable in terms  of biological  indicators (Feldstein, 1974).   How-
ever, such effects—for example, the onset of eye  irritation—are difficult
to quantify.

2.   Concentration Measures

     The  intensity of smog formation is often assessed in terms of the con-
centrations of major primary and secondary pollutants.  As a result, some
investigators have chosen to base hydrocarbon reactivity measures on the maxi-
mum  concentrations of products that ultimately form.  Their criteria include:

     >  Maximum  (or asymptotic) oxidant (Heuss and Glasson, 1968;
        Altshuller and  Bufalini, 1971; Dimitriades and Wesson,  1972),
     >  Maximum  eye irritation*—a function of the concentrations of
        lachrymators and other irritants (Heuss and Glasson, 1968;
        Altshuller and  Bufalini, 1971).
 * Yeung and Phillips  (1973)  have  attempted  to  relate  reactivity  to  eye
   Irritation through  the use of a "biological  effect  factor":

   Relative Chrtc.1  Reactivity -

-------
                                                                      116
Because air pollution really consists  of  all  the products listed above and
more, a reactivity measure based on  the concentrations  of smog  components
might best be expressed in the form
                         R = E  Vi
where
          R  = reactivity measure,
          i  = an index of all  harmful  components  in the system,
          c-j = the maximum concentration of the itn species,
          ai = a species weighting factor relating the toxicity (or
               other harmful effects)  of Cj.

3.   Combined Temporal and Concentration Measures

     Investigators have also characterized reactivity using measures that
depend on both concentration and time.   The most common of these criteria
are dosage  (ppm-min), rate (ppm min-1), rate constants (ppnr1  min'1). and
percentage  of hydrocarbon or NOX consumed at a fixed time (dimensionless).

a.   Dosage

     Dosage is defined as

                           A
                          J   c. dt


AHshuller et al. (1970) and Dimitriades and Wesson (1972) have used N02,
oxidant, PAN, and formaldehyde dosages  as measures of hydrocarbon reactivity,
The choice of tf, the ending period of the integration, is important.  One
might wish to use the characteristic residence time of pollutants in major

-------
                                                                      117
air basins, the time to the N0£ peak,  or the time to  the oxidant peak.   The
beginning time of the integration,  tg, is usually the initiation of irradia-
tion, but some later time might also be chosen.   In the  case of oxidant dosage,
for instance, it might be more practical to take tg as the time of the  N02
peak.  Consideration of maximum one-hour dosages (e.g.,  tf - tg = 60 minutes)
might also be appropriate in view of existing federal air quality standards.

b.   Rate

     The rates of the chemical transformations in polluted air change con-
tinuously, and both instantaneous and average rates have been used as mea-
sures of hydrocarbon reactivity.  The instantaneous rate, (dci/dt)^^, has
been considered by Altshuller and Cohen (1963), using N02 as c-j and
tj = t(i/2 N02 max) on botn tne ascending and descending portions of the
N02 versus time profile.  One could also evaluate the slopes at tj = 0 or
at the time of maximum slope.  Average formation rates for N02 can be de-
fined as l/2(NO)o/t(l/2 NO conversion) or as (N02)m,v/tN(v> ma><-  The former
                                                  11 Id A    £
has been taken as a reactivity measure by Heuss and Glasson  (1968) and
Glasson and Tuesday (1970).  Other average rate measures using oxidant
rather than N02 have been reported by Heuss and Glasson  (1968) and Altshuller
and Bufalini  (1971).

c.   Rate Constants

     Hydrocarbons in the atmosphere are oxidized principally through reac-
tions with 0, OH-, and 03.  Of  these  three species,  OH-  is  thought  to  be
the most important oxidant of all classes of hydrocarbons,  and  the  rate con-
stant for the OH-hydrocarbon reaction has been used  by  Niki  et  al.  (1972) to
characterize reactivity.  They  found  that kg^ correlated much better with
the reactivity measures of Glasson and Tuesday  (1970) and Altshuller and
Cohen (1963) than did kg or kQ3» the  respective rate  constants for  the
0 and 63 oxidation reactions of hydrocarbons.

-------
                                                                     118
d.   Percent Hydrocarbon Oxidation

     A final measure of reactivity is the percentage  of  initial hydrocarbon
oxidized at a fixed time (Altshuller and Bufalini,  1971).  As  in  the case
of dosage, one might consider a time span equivalent  to  the  residence time
of an air mass in a polluted air basin, the time of the  N02  peak,  the time
of the ozone peak, or some other relevant period.

B.   MEASURE ASSESSMENT

     Since an evaluation of all of the measures listed above would be imprac-
tical, this study has been limited to a group representative of the simplest
and most practical of those proposed.  The evaluation process  is  described
in Section 2; the final selections are reported in  Section 3.

     A primary application of reactivity measures is  the prediction of  the
smog formation potential of mixtures of hydrocarbons  emitted as automobile
exhaust and solvent fumes.  Predictive ability is essential  to effective
control-strategy planning and evaluation (Dimitriades, 1973).   As part  of
the present study, several smog simulations using mixtures of  olefins and
NOX in air were performed to determine whether mixture reactivity can be
predicted for these simple cases.  The results of the mixture  study appear
in Section 4.  A semi-theoretical justification for some experimental ob-
servations made during this study is presented in Section 5.  The section
Immediately following describes experimental methods.

1.   Scope and Procedure

a.   Mathematical Simulation

     The data used  in this study were  generated by the mathematical »del
for Smog station presented  in Table 8 of the Second Annual  Report (Hecht
« .1., 197*).  The use of nu^rical  rather than physical expends  i,
unusual, but  it  offers definite advantages.  Initial concentrations can be

-------
                                                                      119
specified precisely; ambiguities due to chamber effects  are  absent; the
concentrations of all species (including free radicals)  are  known at any
given time; "experiments" are quick (a few seconds  of computation), easy,
and inexpensive to carry out; and instrumentation is  not needed  to measure
species concentrations.

     The principal drawback to modeling Is Its possible  inaccuracy in rep-
resenting reality.  For the present purpose, which  is the theoretical eval-
uation of reactivity measures, it is not mandatory  that  the  kinetic mec an-
ism be absolutely accurate, since all comparisons of  measures  are made be-
tween coated results.  If the mechanism were infallib  e, reacts      uld
be evaluated directly by simulation.  As an example of d,rect  s    a     ,
consider an industrial process that normally results  ,n  atmo pher
trations of 1 ppm propylene, 0.2 pom NO, and 0.02 ppm 02.  Supp   ,     d
tion, that at a slightly higher cost, about one-half th,s pro  *   e
oxidized to form a!dehyde  (and C02).  Kopczynsk, et .1  (1974)      ™
and Wesson (1972) showed that aldehydes can be as reacts as    e     ,  o   he
effect of this partial oxidation is not, a priori  obvious     e tility •
this proposed approach to  hydrocarbon emissions reduct,on wo 1 d  th      e
be of co cern to control strategists.  Figure  58 demonstra          r
Simulation can be used.  As this figure shows, in th,s s,m  e  ^
production of smog  constituents (typified here by
easily and the cost-effectiveness of the proposed s rate y t hus
Unfortunately, obtaining a perfected mechanism for  he   t r     tr,
emitted pollutants would retire an overwhelming - ^    "        'of ....
present study requires only that the mo , ^  s      ,   /  /^l to be
sure comparison.  The model has been tet      6   ^^  smg chamber data
capable of providing reasonably accurate pred ctions
(see the Second Annual Report, Hecht et al., 1974D).
b.   Purpose
                 nt of factors pertinent to reactivity determination can
     The assessment of factors pe                           efficiency of data
help m the planning of  laboratory expends, improve

-------
     1.00
 
-------
                                                                     121
collection, and provide heuristic  guidelines  for  impact evaluation.  Toward
this end, the kinetic mechanism  provides  a  basis  for  conclusions, strengthens
arguments, and tests hypotheses.   Although  the mechanism  has  been validated
for a restricted number of hydrocarbons,  by using it  as a "laboratory" we
have obtained results that hopefully have broad significance.

     The purpose of this study was not to provide absolute quantification
of hydrocarbon reactivity.  Such quantification,  if it is possible,  must be
obtained from well-controlled laboratory experiments.  Extensive tabulations
of laboratory results can be found in Heuss and  Glasson (1968), Glasson  and
Tuesday  (1970), and an MSA Research Corporation  report (1972).

c.   Procedure

     By  inputting desired initial conditions  and integrating the appropriate
kinetic  equations, we obtained concentration-versus-time curves for a given
hydrocarbon-NOx-air system.  It was then necessary to select the relevant
points,  such as peak concentrations or times of peaks, from the computer
output to  determine values for the various reactivity measures.  More de-
tails on what  these points were and how  they were used are given in subse-
quent sections.

2.   Measure  Study

a.   Criteria  for the  Evaluation  of Measures

     While criteria for a good measure are basically intuitive,  it  is worth
mentioning a  few  of them here.   Ideally, a measure of a  given  hydrocarbon's
reactivity would  be independent of initial reactant  concentrations.  Unfor-
tunately, because of the complexity of smog  systems, such independence is
not realizable.   However, some measures  will show less variability  than
others.   If the inevitable variations show a consistent  trend, it may be
possible to specify their functional  dependence  and, hence,  to develop a
very useful predictive ability.   Within  this report, a measure that shows  a

-------
                                                                      122
small predictable trend is termed "self-consistent."   Aside from being self-
consistent, satisfactory measures must also be consistent with each other
and not widely disparate with accepted reactivity values.

     Three pragmatic requirements we have imposed are that the measure be
(1) directly related to the production of harmful smog components,
(2) clearly defined, and (3)  easily and accurately measurable.   Without
these properties, the applicability of the measure would be severely
limited.  Other criteria will be developed as  needed.

b.   Normalization

     The results from a study of the measures  defined in Table 15 are  tabu-
lated in Table 16, and the initial  conditions  for the experiments reported
are shown in Table 17.  The entries in Table 16 were  normalized by the reac-
tivity of propylene; i.e., they are in units of propylene equivalents.  P.ro-
pylene simulations were carried out at initial NO, N02> and hydrocarbon con-
centrations corresponding to each row, and the times, concentrations,  and
rates corresponding to each column were determined.  These were then used to
normalize values obtained for other olefins.  Since the rate constants are in-
versely proportional  to time, for the time scales, an inverse ratio was used,
in which the relative reactivity of a given hydrocarbon (HC) is given  by the
ratio Tpropylene^HC' wfiere T is t^ie appropriate time scale.  For other measures,
a direct ratio was used.

c.   The Elimination Process
     The data in Table 16 do not provide a sufficient basis for choosing
the best measure of reactivity without additional considerations.  But an
inspection of these data does permit a rapid elimination of three of the
criteria.  The % HCt=ioo obviously fails the self-consistency test because
it produces a wide range of values at various initial concentrations.
While showing self-consistency, N02(max) and 03(max) are quite inconsistent
with other measures.  In fact, they are so close to unity that, in light of
model inaccuracy, the differences in reactivity between the various olefins

-------
                                                                     123
                             Table  15
                DEFINITIONS OF  REACTIVITY  MEASURES*
 Reactivity Measure

  HCt=100
03 (max)
Scaling
N02 rate
     conversion
             Definition
     conversion  (°r
100 x HC(t=loO min)/HCo

 Peak concentration of N02

 Peak 03 concentration or asymptotic
 concentration

 Reciprocal of the HC concentration
 required to obtain a Tj^fmax) equal
 to that of 1 ppm C^\s

 N02(max)/TN02(max)

 Time for the N02 concentration to
 reach the value [N0210 + 1/2[NO]0

 Time to the N02 peak

 Time when [HC] = 0.75[HC]0
* Reactivities relative to C3Hs are given by the reciprocal ratio of
  time scales and direct ratio of all other measures.

-------
                                                                                                  124
                                                  Table 16  •
                     RESULTS OF THE MEASURE STUDY:  REACTIVITIES* RELATIVE TO PROPYLENE
Experiment*
1
2
3
4
5
6
7
8
9
10
11
12
G&TS
A&B**
ASCtt
% HC10Q
0.12
0.10
0.16
0.27
0.29
0.18
0.06
0.17







N02(max)
--
0.91
0.93
0.95
0.96
0.92
0.91
0.84







Q
3(max) Scaling
< 0.25
0.25
0.89
0.90
0.90
0.90
—
0.27
1.16
1.12
1.10
1.09



N02 rate
—
0.23
0.23
0.28
0.28
0.28
0.27
0.23







(H NO conversion)
—
0.25
0.22
0.20
0.21
0.25
0.29
0.24




0.49


N02(max)
—
0.25
0.25
0.29
0.29
0.30
0.30
0.27
1.45
3.27
1.83
3.98

0.36
0.23,0.31
\ HC
* 0.2
0.23
0.23
0.24
0.26
0.24
0.24
0.26
1.68
3.84
2.11
5.23

0.48

   The reactivity measures are defined in Table 15.
** Altshuller and Bufalini (1970) values for ethylene.
 t The Initial concentrations are given in Table 16.
tt Altshuller and Cohen (1953) values for ethylene.
 5 Glasson and Tuesday (1971) values for ethylene.

-------
                                                                     125
                           Table 17

          INITIAL  CONCENTRATIONS FOR EXPERIMENTS LISTED
                       IN TABLES 16 AND 18
Experiment
1
2+
3
4
5
6
7
8
9+
10+

11 +
12+
13+
14
15
16+
17
[NO]Q
0.4
0.4
0.4
0.4
0.4
0.2
0.48
0.5
0.4
0.4

0.4
0.4
0.4
0.3
0.48
0.4
0.4
[NOJ
* 0
0.1
0.1
0.1
0.1
0.1
0.05
0.02
0.1
0.1
0.1

0.1
0.1
0.1
0.075
0.02
0.1
0.1
[HC]Q
0.5
1.0
2.0
3.0
4.0
1.0
1.0
1.0
1.0
1.0

1.0
1.0
1.0
1.0
1.0
1.0
0.6
Initial HC Composition
HC5*
HC5
HC5
HC5
HC5
HC5
HC5
HC5
HC1
HC,
£.
HC3
HC4
0.2 each HC] , HC2, HC3> HC4, and HCg
0.2 each HC] , HC2, HC3, HC4> and HCg
0.25 each HC2, HC4, HCg, and HCg
0.25 each HC2, HC4> HCg, and HCg
0.15 each HC0, HC., HCC, and HCC
£. <\ 0 D
* In this study H^  s 1-butene;  HC2 =  cis-2-butene;  HC3 =  2-me-l-butene;
  HC4 s 2-me-2-butene; HCs =  ethylene; HC6  =  propylene.

t These initial conditions are defined as  "standard."   Experiments  2 and
  9 through 13 provided data  for the mixture  study.

-------
                                                                     126
are insignificant when they are  based  on  either  W^fmax) or 03(max).  The
investigators who successfully used  03(max)  probably defined  it either as
a peak or, when no peak occurred,  as the  concentration  of  63  at the end  of
a time-limited experiment;  they  are  usually  not  clear on this point.  The
entries omitted from Table  16 correspond  to  experiments in which  03 had
reached neither a peak nor  an asymptotic  value by the end  of  400  minutes.
In these omitted cases, the final  value indicated a reactivity significantly
less than one.  Thus, the definition of 03(max)  varies, depending on  whether
a peak is reached.  The values of reactivity obtained will thus depend on  the
length of the experiment; they are neither clearly defined nor consistent.
We surmise that 03(max) and N02(max) are insufficient measures.

     As shown in Section 5, scaling can be demonstrated to be equivalent
to TUJ.  Their theoretical equivalence is borne out by  experimental evidence,
as shown in Table 16.

     Relative reactivity based on the N02 rate is defined  as  NOaCmax) times
TJJ, (it is multiplicative because of the inverse  normalization of time scales)
Since the N02(max) values are all close to unity, the  difference between Tm
and the N0£ rate should be  small.  The added complexity in rate determina-
tion hardly seems worthwhile.  Because it is a "combined spatio-temporal"
measure, the error in determination of the N02 rate is the sum of errors of
its component parts.  Clearly, if one of these components is a good measure,
accuracy as well as simplicity can be gained by using  it alone.  A last, and
possibly undesirable, property of the N02 rate is that it is a difference
approximation to the  rate of N02 formation.  Therefore, it does not repre-
sent the actual rate  at any point on the NOg curve; it is instead an average
value.
          conversion ">s a widely used measure, usually appearing in the
 guise  of  "NO photooxidation rate," defined as NOo/2T^ (c.f. , Glasson and
 Tuesday,  1970).   Since, in the present study, relative reactivities were
 computed  at a  given NOg, this column in Table 16 could just as easily have
 been labeled "R(NO photooxidation)"-  Interpreted as such,  \ also has  the
 possible  shortcoming of being a difference approximation, although it is  a

-------
                                                                      127
very good one when the induction period is  negligibly  short.   Ti^NO conversion*
as defined in Table 15, is not the half-time for N02 formation,  rather  it
falls somewhere between the half-time and peak time of N02-   Its location
relative to the time of product formation is therefore ill  defined.   Thus,
the most objectionable quality of \ is its lack of direct  correlation  with
03 production or other harmful smog constituents.  In  the application of
reactivity criteria to pollution control, this is indeed a  serious shortcoming.
     TN02(max) and Tj,HC conversion' the on^ measures left, are the measures
we have selected for use.  Because of their importance to the present study,
they are discussed in detail in Section 3.

     The results of other investigations are also included in Table 17 for
the purpose of indicating the need to eventually combine the work contained
in this report with laboratory investigations.  Meaningful comparisons of
our values with those of the other investigators cannot be made now because
experimental  conditions generally differ considerably.

3.   The Measures Selected

a.   Practicality

     The measures found  to  be most conducive  to  the quantification  of smog
effects are  the  time  of  the N02  peak,  TNo2(max)  (hereafter denoted  as Tm) ,
and  the time required for one-quarter  of  the  initial hydrocarbon  to be oxi-
dized  (Tjj).   In  the  terminology  introduced in Section A,  these are  temporal
measures.   By being  one  dimensional,  they avoid  the increased measurement
error  inherent in the combined  concentration  and temporal  measures.  Both
are  simply and clearly defined.   Although the N02 peak may not be sharply
 resolved  in practice, interpolation  methods along with extremum theory pro-
 vide for  its accurate determination.   A parabolic curve fit can be used for
 this purpose.

-------
                                                                    128
     As shown by the  results  of  Experiments  1  through 5 in Table  16, the
relative reactivity of ethylene  based on Tm  (hereafter abbreviated  RR-j-J
varies by about 15 percent as the  initial  hydrocarbon concentration (HCg)
is changed by a factor of 4.   There  is  a visible  trend toward  increasing
RRj  with increasing  HCg.  Although  any variability  is undesirable, the fact
that it is small indicates that  the  RRTm,  measured as a single HCQ, may be
applicable throughout a wide range of HCo's  (at fixed N00 and  N02Q).   !t  may
even be possible to  capitalize on  the consistency of the  trend to estimate
the accuracy of a constant value in this  range.

     At fixed HCQ, increasing N0o/N020 by a factor of 6  caused an increase
in RRj  of almost 20 percent (see Experiments 2 and 7).   In Section 5, this
behavior  is shown to be attributable to induction period effects.  Because
of its role in determining RRjm, further study of the induction period would
be useful.

     As shown by  the data in Table 17, RRT% exhibits the same trend with
increasing HCg  as that observed for RRTm.  But variability with N0o/N020
is almost absent.  An increase  in this ratio  by a factor of 6 caused vir-
tually no change  in  RRTj,.  The  slightly erratic behavior shown in  Experi-
ment 8 is most  probably  caused  by inaccuracy  in this experiment.   (Unfortu-
nately,  the  need  to  incorporate an  interpolation scheme  did not  become ap-
parent until  the  late  stages of this study.)   \ is therefore partially  in-
consistent with Tm.   The apparent absence of  induction period effects on
RR-|> is interesting  and  deserves  further  investigation.   It is probably  due
to the smaller initial  slope of the hydrocarbon  curve.

b.   Usefulness

      Aside from considerations  of simplicity and consistency, a  useful
measure must ultimately be related to objectionable pollution effects.
 Dimitriades et al.  (1970) have  discussed this issue and concluded that,
while no one index is fully satisfactory, "...there is  evidence that the
 over-all [sic] level of activity in the photosimulated hydrocarbon/NOx

-------
                                                                      129
system is reflected in the pattern of N02 formation."  It is well  known
that the N0£ peak is correlated with the formation of 03 and PAN,  two haz-
ardous components of smog.  In any "time-limited" system, such as  an urban
airshed with a characteristic residence time, the amount of 03 present due
to chemical reaction is directly related to Tm.   Aldehydes also contribute
to the deleterious effects of smog.  Because aldehyde appearance is comple-
mentary to hydrocarbon disappearance, 1% is an indication of their importance
in a time-limited system.   T^ is also useful for evaluating the magnitude
of synergistic effects in  mixture reactions.  Altshuller and Bufalini (1971)
define a synergistic effect as "...one in which  the reactivity or  the amount
of product produced by a given compound is affected by the presence of a
second."  Since, in general, the oxidation of several hydrocarbons in a par-
ticular mixture will lead  to the same or similar products, synergism is not
easily determined from product measures.  A much simpler means, particu-
larly when using a numerical model, is to monitor the rate of hydrocarbon
disappearance.  For a given hydrocarbon, the change in TV from its value
in an individual hydrocarbon-NOx reaction system to that in a multihydrocarbon-
NOX reaction system provides an indication of interactive effects.

c.   Measurability

     Because laboratory techniques must ultimately be used either  to mea-
sure reactivity or to obtain empirical constants needed for its prediction,
a satisfactory measure must have the additional  property that accurate and
reliable instrumentation be available for its determination.  N0£  and hydro-
carbon concentrations are  routinely monitored in smog chamber experiments
with reasonable accuracy.   Hydrocarbon concentration can be measured with
as little as 1 percent error by gas chromatographic methods.  In contrast,
N02 is obtained from the difference between NOX  (after conversion  to NO) and
NO concentrations.  These  are measured by chemiluminesence with about 95
percent accuracy.  When N02 concentrations are low, the percentage error in
the difference of these values may be large.  When [NOel is at its peak and,
consequently, [NO] is low, the error in the difference will be at a minimum.
The accuracy in [N02] at the peak should therefore be close to 95 percent.

-------
                                                                      130
Of course, only the temporal  location (Tm),  not the  peak value itself is
used.  The cycling time for NO measurement  is  only a minute  or two;  so a
high density of points and, hence, sharp resolution  of Tm can be  obtained.
With reasonably accurate values of concentration and the time of  measurement
of each point well known, an accurate determination  of Tm should  be  possible.

4.   Mixture Study

a.   Mixtures Used

     Reactivities were computed for five different olefin-NOx mixtures at.
the initial concentrations indicated in Table 16.  Experimental and pre-
dicted results are shown in Table 18.  The reactivities are all relative to
that of propylene at standard initial conditions  (N0o=0.4, N020=0.1, HCo=1.0).
The olefin mixture used in Experiments 13 and 14  is composed of the five most
reactive olefins  studied and is a typical "highly reactive" mixture.  A
second mixture, used in Experiments 15 through 17, contains the two most
reactive and the  two least reactive olefins and is characteristic of a wide
reactivity range  mixture.

b.   Results

     The first row of  entries  for each experiment in  Table  18  is labeled
"mixture."   In Columns  3 and  5  (Tm and Tj, measured) of  this  row, relative
reactivities of the olefin mixtures  are measured.  Tm,  as before, is  the
time of the  NOg peak.   Tj.  is  now  the  time required  for  the  total olefin con-
centration  to drop by  25 percent.  Below the  \ entry are the  time  for each
component olefin  to reach  75  percent of its initial concentration.   As
before, all  results are normalized by propylene experiments  at the  given
initial concentrations.

c.   Predictions

     There  are  also three  prediction columns  in Table 18.   These are pre-
dictions  of mixture reactivity computed by the "linear summation"  method,
defined by:

-------
                                                                                                       131
                                                Table 18
                 RESULTS OF THE MIXTURE STUDY:  MIXTURE REACTIVITIES RELATIVE TO PROPYLENE
Tm
Experiment Hydrocarbons Simulated LSM
13 Mixture 2.42 2.31
HC,*
HC2
HC3
HC4
HC6
14 Mixture 2.64
HC,
HC2
HC3
.HC4
HC6
15 Mixture 1.54
HC2
HC4
HC5
HCfi
16 Mixture 2.16 2.13
HC2
HC4.
HC5
HC6
17 Mixture 1.26 1.28
HC2
HC4
HC5
"6

LSM
Uneorrected
Simulated Prediction
2.86 2.77
2.30
3.42
2.71
4.27
1.37
2.71 -
2.15
3.39
2.53
4.44
1.21
1.94
2.33
2.83
0.44
1.20
2.58 2.56
3.58
4.31
0.47
2.58
1.98 1.54
2.52
3.15
0.44
1.20
\
Simulated
Reactivity of
Hydrocarbons
In Mixture

1.68
3.84
2.11
5.23
1.0

1.68
3.84
2.11
5.23
1.0

3.84
5.23
0.23
1.0

3.84
5.23
0.23
1.0

3.84
5.23
0.23
1.0

Prediction
Corrected for
Synerqlsm
2.81





2.76





1.70




2.73




1.83




• In this study. HC]  i  1-butene; HC2 i cit-2-butene; HC3 i 2-oe-l-butene;  HC4  a 2-me-2-butene; HC5 i ethylene;
  HC( i propylene.

-------
                                                                      132
                        RLS ' E C1R1
where

          RLS = mixture reactivity by linear summation,
          n   = number of organic mixture components,
          cj  = initial concentration of the ith hydrocarbon,
          R-j  = reactivity of the ith hydrocarbon.

The linear summation technique has been discussed by Glasson and Tuesday
(1971) and Dimitriades et al. (1970).

     The values of R-j, based  on either Tm or T^, were obtained from the
experiments in Table 16 as follows:  R-j comes from Experiment 9, R2 from
Experiment 10, R$ from Experiment 11, R4 from Experiment 12, R$ from Experi-
ment 2, HC5 is the reference  olefin (propylene)  with a reactivity defined
to be 1.  (Note, from the definitions of HCs 1 through 6 in the footnote to
Table 17, that R-j increases with increased substitution at the double bond.

     Column 4 of Table 18 gives the mixture reactivity based on Tm as com-
puted by the linear summation method.  The values shown agree very well
with observations.  Experiments 14 and 15 were done under nonstandard ini-
tial NO and N02 values, and,  therefore, no predictions could be made.
Although the initial olefin concentration in Experiment 17 was not the
standard value of 1 ppm, prediction could be made by "scaling."  This pro-
cedure consists simply of multiplying RLS by the ratio of the initial
hydrocarbon concentration to  its standard value (in this case the RRLS of
Experiment 16 times 0.6).  Hence, the slower rate at lower hydrocarbon con-
centrations is compensated for multiplicatively.  Justification for the
application of the scaling technique to RRTm is given in Section 5.  Its
application to RRj^ is not really justifiable because the approximately
exponential hydrocarbon decay rate indicates a nonlinear dependence of
   on HCg.  This explains the low value listed.

-------
                                                                      133
d.   Synergism

     Columns 6 and 7 of Table 18 show both the linearly predicted reactivity
of the mixtures and the standard reactivities of their components.  The lat-
ter were copied from Table 16.  Comparison of the standard reactivities in
Column 7 with the observed values in Column 5 indicates the presence of
synergistic effects.  It is evident that in all cases the two most reactive
olefins, HC2 and HC4, experienced a decrease in relative reactivity, whereas
the less reactive olefins experienced an increase.  This behavior can be ex-
plained qualitatively in the following manner.   The competition for available
oxidant between higher and lower reactivity olefins depressed the reaction
rate of the former.   At the same time, the high rate of oxidant production
by HC2 and HC4 accelerated the consumption of less reactive olefins.

     Linear summation was also applied to the synergistically modified  values
of T^J (Tjj measured).   The resulting mixture reactivities are listed in  the
last column of Table 18.  Although these predictions are close to observed
values, we cannot state that they are always an improvement over the unmodi-
fied predictions.

     Another inconsistency between Tm and Tjj is contained in Table 18.
Whereas decreasing NOXQ from Experiment 13 to the value used in Experiment
14 increased RRr , it decreased RRT. .  Many explanations of this behavior
               'HI.                 "Z
can be offered, including the changed 03 production and the radical-
scavenging ability of N02.  It is the complex interaction of all these ef-
fects that leads to the discrepant behavior.

5.   Derivation of Some Properties of Tm

a.   Derivation

      In this section, the analytical  solution  for  the dependency of  Tm on
Initial hydrocarbon and  NO concentrations  is obtained through  a consideration
of simplistic  photochemical smog  kinetics.   Kinetic equations  and  empiricism
based  on  the observed shape of  smog  profiles are  used toward this  end.

-------
                                                                       134
     O'Brien (1974)  has demonstrated that,  except in  the late  stages  of the
photolytic hydrocarbon-NOx-air reaction,  the  concentrations  of NO,  03,  and
N02 are related by the approximation

                             k,
                     [N02]  *f- [N0][03]     ,                     (IV-D


where k-| and k3 are rate constants for the reactions
                                i
                     N02 + hv —!*- NO + 0                         (1)

                              k-
                     NO + 03 —i- N02 + 02  •  .                    (3)

Existence of the photostationary state, expressed by Eq. (IV-1), in smog
profiles computed using the Hecht-Seinfeld-Dodge kinetic model has been
demonstrated by Liu (1974).

     In addition to Reaction  (3), the conversion of NO to NOg is accomplished
through the reaction


                   R0£ + NO  —^RO- + N02    ,     •             (42)

where  R is  usually an alkyl  group or hydrogen atom.   Hence,  in  the period
before the  N02  peak, the rate of NO  production  and  consumption  is  governed
by Reactions  (1),  (3),  and (42).  Thus,
                      j       -  k3[NO][03]  -  k42[R02][NO]

 Equation (IV-1)  can be  used to simplify Eq. (IV-2):

-------
                                                                      135
This expression represents the perturbation  to photos tationarity introduced
by free radical oxidation.
     In a completely generalized mechanism,  ROg includes  all  free radical
oxidation products of hydrocarbons.   It is  formed in the  following reaction
where Ox represents a general  oxidant (primarily OH-  in the time prior to
the N02 peak), HC-j  a hydrocarbon,  and ct^  an appropriate stoichiometric co-
efficient (see Hecht and Seinfeld, 1972).

     Using Reactions (IV-4), (42), and the steady-state assumption for R02
leads to the following equation:

                                       O x]
                                       x
                                                                          *
Substituting for RO^ in Eq. (IV-3) and using Reaction (IV-4) gives

                                          d[HC.]
     The rate terms in Eq. (IV-5) can be used to relate Tm to HCg and NOg
through the use of empirical  observation.  Figure 59 is an illustration of
a typical, through idealized smog profile at initial NO and N02 concentra-
tions for which there is a very short induction period.  It is apparent
that the curves labeled NO and hydrocarbon are approximately linear during
the early stages of the reaction.  Good approximations in this linear
period are
* Liu (1974) has demonstrated that this assumption becomes valid within a
  few seconds of reaction initiation.  The neglect of termination reactions
  1n the steady-state expression is valid early in the reaction.

-------
                                               136
            Hydrocarbon
        50
   100
Minutes
150
FIGURE 59.  TYPICAL SMOG PROFILE

-------
                                                                       137
                                      m
 and
                   [HC.]  . [HC.]   (1  - b.t)     ,                   (iv-7)


 where b^  is  a  constant related to the slope;

      The  introduction of  HCQ. as a multiplicative factor can be justified
 by a  direct  integration of the hydrocarbon rate equation derived from
 Reaction  (IV-4):

[HC.]  = [HC.]   exp  |"-/k0   [Ox] dtj *  [HC.]   /I - / kQ    [Oxl dt)    ;    t < t,
             U      1»      X.        -J         U  >        X«        /
                                                                   (IV-8)

Comparison of Eq. (IV-8) with Eq. (IV-7) shows that bj can be related to
the average oxidant concentration,
but such an interpretation is not essential to this derivation.  There is
no reason to believe that ^ will be independent of N00, N02o, or HCi
Furthermore, the evidence indicates that bl is proportional to light inten-
sity (Niki et al., 1972; Glasson and Tuesday, 1970).  Other factors not
considered may also influence ty.  However, the dependence on N0xn and HC0
1s assumed to be negligible.

     Substituting Eqs.  (iv-6) and (IV-7) in Eq. (iv-5) and rearranging
gives

-------
                                                                      138
                     [N0]fl
              Tm a b'  [HC.l      ;     bi  ~=  aibi     •
                         1 0

b.   Verification and  Application

     At a fixed initial  NO concentration,  Eq.  (IV-10)  states that the time to
the N02 peak should be inversely proportional  to  the initial  hydrocarbon  con-
centration.   In Figure 60, the  observed  values  of Tm are  plotted  as  a func-
tion of HC-JQ for HCi = propylene at one  NOg  value and  HCi  =  ethylene at two
values of NOg.   All curves can  be  fit by the form a/HC0.   From Eq.  (IV-10),
a = N00/bj.   For propylene,  the curve fit  has  been drawn  in,  whereas for  the
two ethylene curves it has only been indicated.   When  the two values of a
for ethylene are divided by the corresponding  values of NOg,  the  results  are:
a/NOo = 500 and a/NOn. =  496. Therefore, b\  for ethylene  is  2 x 10'3.* The
data presented  in Figure 60  thus confirm Eq. (IV-10).

     Equation (IV-10)  can be used  to predict reactivity relative  to  propylene
               RRT  - Tm prop _   bi    HCJ
                    -         "
Once RR-rm has been determined at one HCiQ,  it can be predicted at other
HCig's by applying Eq.  (IV-11).   Recall  that this is exactly the procedure
(called "scaling") used to predict the mixture reactivity in Experiment 17.

     The equivalence of scaling  and Tm as reactivity measures can be shown
as follows:  Equation (IV-11), with Tm HC.  = Tm propylene and [prop]0 = 1 ppm,
states that
                       K)"1 •
  bi
B1	   •                       (IV-12)
 prop
* As an order of magnitude check using Eq.  (IV-9)  and taking Ox as OH-,
        0.75 x 104 ppm-1 min-1, OH- = 1.5 x 10-7 ppm; thus
       1 x 10-3 min'1 (^ is then 2).

-------
                                                                  139
    250  -
                                        Experimental  Conditions
                                                                0.1

                                       	   N00 = 0.5, N02_ = 0.1
N00 = 0.4, N02()
                                                   Curve
                                                   Fit
                                    HCg  (ppm)
FIGURE 60.  Tm AS A  FUNCTION OF INITIAL HYDROCARBON CONCENTRATION

-------
                                                                      140
This, by definition (Table 15),  is  the scaling  reactivity of HCj.   Because
RRTn) is normalized, HC1(J = [prop]0.   Thus,  from Eq.  (IV-11),
The equivalence of RRTm and scaling follows directly.

     The derivation of Eq.  (IV-10)  can be extended to multihydrocarbon
smog systems by summing over the index 1  from Eq.  (IV-4) onward.  Equation
(IV-10) then becomes (upon  inversion)
                                 HCb
where n is the number of hydrocarbons.   Assuming the values of b] are those
obtained in individual  hydrocarbon simulations, the linear summation method
results directly.   In light of this required assumption, it is surprising
that synergistically modified predictions (Table 18) are not better than
unmodified values.  However, the modified values were based on RRT
                                                                 '%
whereas the above derivation is based on Tm and is not rigorously applicable
to Tjj.

c.   The Induction Period

     It is worth reiterating that Eqs.  (IV-6) and (IV-7) are valid only if
induction period effects are negligible.  For this condition to hold, the
initial HC/NOX and N02/N0 ratios must be relatively high (lower bounds have
not been established but 2 and 0.1, respectively, seem reasonable).  In
addition, the individual values of HC0, N00, and N02g may themselves be
important.

-------
                                                                      141
     A striking, and possibly disconcerting, feature of Eq.  (IV-10) is the
absence of explicit dependence on N02Q.   Dimitriades (1972)  found that, for
irradiated auto exhaust with an N02o/N00 ratio of about 0.1, the rate of NO
photooxidation was independent of N02Q as long as N02Q was above 0.03 ppm.
The major effect of low values of N02o is to cause a nonnegligible induction
period.  It seems reasonable to speculate that an induction period of length
TI will simply cause a shift in the start of the "linear period" by Tj.
Tm can then be replaced by
where T° is given by Eq. (IV-10).  RR-^ is defined as follows:

                       T1        T°      + T
                   ,  s         =        -                      ,     }
                                      °
                          mi         Ira1 T 'Ij
                                         I prop
                                    m
where Tj/TJjj « 1 has been assumed.  If hydrocarbon i is less reactive
than propylene (RRQ  < 1) and has an induction period about equal to that
                   m
or propylene, the correction to RRjO will be positive.*  Conversely, for
more reactive hydrocarbons (RR-j-jj) > 1), the correction will be negative.
The induction period, therefore, always has the effect of shifting the
relative reactivity toward unity.  The increase in RRTm with decreasing
initial [N02] from a value of 0.25 in Experiment 2 (Table 17) to 0.30 in
Experiment 7 can be cited as evidence of this tendency.

     Since atmospheric concentrations include the range of NOg, N02n, and
HC0 for which there is an induction period, methods for predicting TI are
needed.  Apparently, an inverse dependence on N02Q is indicated.
* Under these conditions, RRT. % RRTm  + (Ti/TJj.) (1 - RRTo)
                             m       Q   \     v         m

-------
                                                                     142
C.   RELATION OF THE ABOVE CONSIDERATIONS  TO  OZONE  PRODUCTION

     The preceding discussion of the issue of relative  hydrocarbon  reac-
tivity has shown that indices directly related to ozone production  are  in-
adequate for reactivity assessment.   The production of  ozone entered  only
indirectly through its relation to Tm.   In the following discussion,  this
relationship is explored in greater detail.  As a prefacing note, the dis-
tinction between individual hydrocarbon reactivity  (at  a fixed NOX  concen-
tration) and hydrocarbon-NOx system reactivity (with both hydrocarbon and NOX
being variable) should be emphasized.   From a control strategist's  point  of
view, this distinction is between emissions composition and total emissions.
For the former, which was the topic of the previous section,  net ozone  pro-
duction is not a sufficient characterization, whereas for the  latter, ozone
production (as shown below) is a distinct  characteristic.

1.   Ozone Isopleths

     Isopleths relating the concentration  of  03 and the time of the NO?
peak, Tm, to initial hydrocarbon and NO concentrations  at various reaction
times are shown in Figures 61 through 66.*  These figures are  based on  the
same computer output used to generate the  isopleths contained  in the final
report for the first contract year (Hecht  et  a!., 1973).  The  initial con-
ditions were as follows:
     >  [HC]0 = 75 percent n-C^io and 25 percent C3H6
     >  [N0}0 = as stated on each figure
        [N02]Q = 0.1  [NOX]
           = 0.35 mlrr1.
As might be expected from the derivation in the preceding section of this
report, the lines of constant Tm (Figure 66) are nearly straight, and their
slope increases as Tm decreases.
* The dashed portions of these figures have been obtained by extrapolation.

-------
                                                                       143
     2.0
i.
a.
o
o
10

4J

O
      1.6
      1.2
     0.8
     0.4  -
 FIGURE  61.   LINES OF CONSTANT 03 (IN PPM) AFTER 1  HOUR OF SIMULATION

-------
                                                                         144
                                                0.5      0.4
     20
o.
o.
 o
o
10
                                                                    0.8
    FIGURE 6.2.  LINES OF CONSTANT 03 (IN PPM) AFTER 2 HOURS  OF SIMULATION

-------
                                                                          145
i.
o.
(O
4->
O
    0.4  -
      0  L
  FIGURE  63.   LINES OF CONSTANT 03 (IN PPM) AFTER 5 HOURS  OF SIMULATION

-------
                                                                         146
a.
a.
 o
o
as
4->
o
                                                                                  .0
                                                                                  0.9
 FIGURE 64.   LINES  OF CONSTANT 03 (IN PPM) AFTER 8 HOURS  OF SIMULATION

-------
                                                                       147
 O
O
re
to
•P
  FIGURE 65.  LINES OF CONSTANT 03  (IN  PPM)  AFTER 9  HOURS OF SIMULATION

-------
                                                                     148
 o
o
•x.
to
4->

O
                                                                            240
                                                                            360
                                                                            540
                                       N00  (ppm)
                 FIGURE 66.   TIME OF THE N02 PEAK (IN MINUTES)

-------
                                                                      149
     In Figures 61  through 65,  there are two characteristic regions.   A
line drawn along the ridge line of the surface represented by the set of
isopleths in each of these figures would divide these regions.   In the re-
gion to the right of the ridge  line, the isopleths  are fairly straight, and
the super-imposition of Figure 66 would indicate that, especially at earlier
times, they are nearly parallel to the lines of constant Tm.  The line
dividing the regions is also nearly a line of constant Tm.  The region to
the left is characterized by almost vertical isopleths, though in Figures
64 and 65 they curve back on themselves noticeably.

     The features of these figures are not at all surprising; they simply.
reflect the characteristics of 03-versus-time profiles observed in smog
chambers (see, for example, the UCR profiles contained in this report).
The empty space in the lower right-hand corner of Figures 61 through 65
(03 < 0.1 ppm) reflects the finite time that elapses before 03 begins to
build up.  At Tm, 03 begins to accumulate almost linearly with time—hence,
the closely spaced isopleths that parallel lines of Tm = constant (Figure 66),
Eventually, [03] approaches an asymptotic level.  Correspondingly, the
spacing of isopleths widens, and they turn to the vertical.  The reactions

                           N02 + 03 •*• N03 + 02     ,               (5)

                          N03 + N02 * N205    ,                   (7)

                         N205 + H20 -> 2HN03     ,                  (9)

and

                            03 + HC •*• Products

along with photolysis and destruction on surfaces, cause 03 depletion  late
in the reaction, resulting in  the backward curvature of the isopleths.

-------
                                                                      150
2.   Chemical  Dynamics

     These observations  indicate  that the  characteristics  of  Figures  61
through 65 are prescribed  by the  chemical  dynamics  of smog formation.   In
the atmosphere, where chemistry interacts  with  the  mechanical  processes  of
dispersion and transport,  a consideration  of dynamics is essential  to con-
trol strategy planning.   For example, consider  the  upper region  in  the 8-
hour isopleths (Figure 64).  Figure 62 indicates  that, at  a fixed level  of
NO, a reduction in [HC]0 would have very little effect on  03  production.
The results presented previously  show that maximum  03 levels  (at fixed NOX
concentrations) are also almost independent of  hydrocarbon reactivity (for
a set of olefins).  However, an examination of  Figure 66  (and the data in
Section B above) shows  that a reduction in [HC]0 (or HC reactivity) has  a
significant effect on increasing  Tm.  Thus, a reduction in [HC]0 could slow
down 03 production, even though this decrease may have little effect on the
expected net yield.  In  the atmosphere, where pollutants  can  be  rapidly dis-
persed, the predicted maximum yield, based on simulations  of smog chamber
experiments, may never be realized.  The peak 03 level achieved  is  therefore
closely related to the expected Tm.

     From the preceding  results,  one can conclude that both HC and NO must
be taken into account when attempting to select optimum 03 abatement strate-
gies.  However, because  of the complex interaction of mechanical and chemi-
cal processes in the atmosphere,  it is difficult to extrapolate such results
as those presented in this report directly to atmospheric emissions.  To
evaluate the effect of control strategies directly, one would need to imbed
the kinetic mechanism in an airshed model that  takes atmospheric conditions
into consideration.  In isolation, the kinetic  mechanism can only provide
"rules of thumb."

-------
                                                                     151
                     V  CONCLUDING  REMARKS
     In closing,  some mention should be made of the implications of our
findings on air pollution modeling.  First, we briefly summarize these
findings.

     Kinetic mechanisms for the chemical  transformations  occurring  in  irra-
diated propylene, butane,  toluene-NOx,  and propylene-NOK-S02  systems were
postulated and used to simulate smog chamber data.   Varying degrees of cor-
respondence between predicted and observed profiles were obtained.   In general,
the  propylene-NO  and propylene-NQX-S02 mechanisms  were the most successful.
For  the  most part, their predictions of propylene,  03, NO, N02, and S02 were
reasonably good.  Although the accuracy was not very good, the propylene
mechanisms were  still able to  follow the behavior of each  species.  The
 butane mechanism predicted too rapid NO oxidation and the  toluene mechanism
 predicted insufficient  oxidation of toluene.  Butane oxidation products con-
 taining structures  of  two  or more  carbon  atoms were apparently produced in
 greater quantity than  the  mechanism indicated; the low  carbon recoveries  in
 the UCR butane runs support this conclusion.   However,  more  kinetic and smog
 chamber data are needed before the toluene mechanism can be  assessed and
 revised.

       We demonstrated that uncertainty in the magnitude of surface reactions
 and light source spectrum decay, as well as other chamber effects, could ac-
 count for a great  deal of the discrepancy between data  and  theory, although
 erroneous rate  constants  and  reaction mechanisms  contributed also.   Instrument
  error, a topic barely  touched upon  here,  is another ever-present  source  of
  ambiguity.

       Unlike purely gas-phase thermal  reactions, surface and photolytic reac-
  tions are chamber-dependent; furthermore, for a given chamber they may vary
  from experiment to experiment.  Their proper treatment would require a con-
  siderable and  continuous effort toward chamber characterization.  Thus, smog

-------
                                                                      152
chamber experiments,  which are  meant to  clarify  the  kinetics  by  eliminating
some of the complexities  present in  the  atmosphere,  have  introduced  their
own problems, although they are not  as complex as  those in  the atmosphere.
We are not denying the value of smog chamber  experiments; instead, we  are
emphasizing the intricacy of this analysis.

     Upon being confronted with the  important role chamber  effects play  in
the laboratory, one cannot help wondering whether  atmospheric counterparts
exist, and if so,  how to  incorporate them into an  airshed model.  Heterogen-
eous (pseudo-gas-phase) rate constants are functions  of surface-to-volume
ratio, as well as  the surface's catalytic efficiency,  both  of which  are  not
known for the urban environment.  Solar  irradiation  depends on the state of
the upper atmosphere, as  well as on  meteorological conditions, and has a di-
urnal and seasonal periodic variation.   Hopefully, parametric representation
of the variability of the solar spectrum will make the characterization  of
the spectrum feasible. Clearly, the spectrum itself affects  numerous  reac-
tions, and the variation  of a single rate constant (such  as k])  cannot ade-
quately account for the effects of spectrum variability.

     The approach  to modeling heterogeneous chemistry used  in the present work
was to represent local surface  reactions as pseudo-gas-phase  reactions.  The
pattern of N02 formation  was shown to reflect the  value  assigned to  rate con-
stants for heterogeneous  (HNOX) chemistry.  Capitalizing  on this relation-
ship, we determined heterogeneous rate constants by  "tuning"  to  the  N02
curve.  When applying the mechanism to the atmosphere, one  can  take  a  simi-
lar approach.  In  the absence of requisite kinetic data,  tuning  to atmos-
pheric N02 data may be possible.  Assuming gas  phase kinetics are accurately
represented, this  approach would provide a practical means  of evaluating the
heterogeneous reactions.

     The mechanism's applied utility was demonstrated, in Chapter IV,  in a
study of hydrocarbon reactivity and ozone  formation.*  Thus,  the kinetic
mechanism can be a useful tool  to investigators of photochemical air pollution,
either in the explicit form or  in the streamlined, generalized  format.
* There the issue of chamber effects was avoided by presenting  results on a
  relative.basis.

-------
                                                                      153
                            REFERENCES


Altshuller, A. P., and J. Bufalini (1971),  Environ.  Sci.  Techno!.,  Vol.  5,
     p. 39.                                ~~

Altshuller, A. P., and I. R.  Cohen (1963),  Int.  J.  Air Hater Pollut..
     Vol. 7, p. 787.                       --

Altshuller, A. P., S. L.  Kopczynski,  W.  A.  Lonneman, T. L.  Becker,  and
     D. L. Wilson (1968), Environ. Sci.  Technol . ,  Vol. 2, pp.  696-698.

Altshuller, A. P., S. L.  Kopczynski,  W.  A.  Lonneman, F. D.  Sutterfield,  and
     D. L. Wilson (1970), Environ. Sci.  Techno!.,  Vol. 4, pp.  44-49.

Atkinson, R. (1975), unpublished.

Atkinson, R., and J. N.  Pitts,  Jr. (1975),  J.  Phys.  Chem. .  Vol   79
     pp. 295-298.                          - -

Badcock, C. C., H. W. Sidebotton,  J.  G.  Calvert,  6.  W. Reinhardt, and
     E. K. Darmon (1971), J.  Am.  Chem.  Soc..  Vol.  93, pp. 3115=3121.

Batt, L., R. D. McCulloch, and  R.  T.  Milne  (1974),  "Thermochemical  and
     Kinetic Studies of Alkyl Nitrites  (RONO),"  Department of Chemistry,
     University of Meston Walk, Aberdeen, Scotland.

Becker, K. H., V. Schurath, and H. Seitz (1974),  Int. J.  Chem.  Kinet.,
     Vol. 6, pp. 725-739.                       ' -

Benson, S. W. (1968), Thermochemical  Kinetics  (John Wiley & Sons, New York).

Bufalini, M. (1971), Environ. Sci. Techno!.,  Vol.  8, pp.  685-700.

Burton, S. (1974), private communication.

Calvert, J. G., and R. D. McQuigg  (1975), Int.  J.  Chem. Kinet.. to  be
     published.                                       ~~ -

Calvert, J. G , and J  N. Pitts,  Jr.  (1967),  Photochemistry (John Wiley
     & Sons, New York).                      ~~~ -

Castleman, A. W   Jr   RE.  Davis, H.  R. Munkelwitz, I.  N. Tang, and
       /,', Wood <1974)'  "Abstracts of Symp. on  Chem. Data for the Lower
     and Upper Atmos.,"  Warrenton, Virginia.
     '."7?'pp'  161*182* and °'  W'  Bottenhe1m (1975>» Int.  J.  Chem. Kinet..

-------
                                                                      154
Cox, R. A. (1972), J.  Phys.  Chem..  Vol.  76,  p.  814.
	 (1972), J.  photo.  Chem.. Vol.  3,  pp.  291-304.
Cox, R. A., and S. A.  Penkett (1972),  J.  Chem.  Soc..  Faraday  Trans.  I,
     Vol. 68, pp. 1736-1753.	
Darnell, K. (1974), private  communication.
Davis, D. D.  (1974), "Absolute Rate Constants for Elementary  Reactions  of
     Atmospheric Importance:   Results  from  the  University of  Maryland's
     Gas Kinetics Laboratory," Chemistry Department,  University of Maryland,
     College Park, Maryland.
Davis, D. D., W. Bellinger,  and S.  Fischer  (1975), J.  Phys. Chem., Vol. 79,
     pp. 293-294.	
Davis, D. P., G. Smith, and  G. Klauber (1974),  Science,  Vol.  179,  p.  280.
Demerjian, K. L., A. J. Kerr, and  J. G.  Calvert (1974a),  "The Mechanism of
     Photochemical Smog Formation," in Advances in Environmental  Science
     Technology, J. N. Pitts  and R. L. Metcalf, eds.,  John Wiley & Sons
     New York.
Demerjian, K. L., J. G. Calvert, and D.  L.  Thorsell  (1974b),  Int.  J  Chem
     Kinet.,  Vol. 6, pp. 829-848.                             	
Dimitriades,  B. (1973), draft for  EPA policy paper.
	 (1972), Environ.  Sci. Techno!..  Vol. 6, p.  253.
Demitriades,  B., B. H. Eccleston,  and R.  W.  Hum (1970),  J. Air Pollut
     Contr. Assoc.. Vol. 20,  p. 150.                                  ~
Dimitriades,  B., and T. C. Wesson  (1972), J. Air Pollut.  Contr. Assoc.,
     Vol. 22, p. 33.                                           "	
Domalski, E.  S. (1971), Environ. Sci.  Techno!.. Vol.  5,  pp. 443-444.
Doyle, G. T., A. C. Lloyd, K. R. Darnall, A. M. Winer, and J. N. Pitts, Jr
     (1975),  Environ.  Sci. Techno!.. Vol. 3, pp. 237-241.
Feldstein, M. (1974),  J. Air Pollut. Contr.  Assoc.. Vol.  24,  p. 469.
Filby, W. G., and H. D. Penzhorn (1974), "Kinetics of the Photoreaction
     of S02 with Alkanes~The Diversion of Intermediate  Radicals by Oxygen,"
     in  Abstracts of Symposium on Chemical  Data for the Lower and Upper
     Atmosphere," Warrenton,  Virginia.

-------
                                                                      155
Finlayson, B.  J.,  J.  N.  Pitts,  Jr.,  and  R. Atkinson  (1974), Contribution
     #I6PP-UCR-74-3,  from Department of  Chemistry and  Statewide Air Pollu-
     tion Research Center,  University of California, Riverside, California.

Garvin, D., and R. F. Hampson  (1974), "Chemical  Kinetics  Data Survey VII:
     Tables of Rate and  Photochemical Data for Modeling of  the Stratosphere,"
     NBSIR 74-430, National  Bureau  of Standards, Washington, D. C.

Gitchell, A.,  R. Simonaitis, and J.  Heicklen  (1974), JAPCA, Vol.  24,
     pp. 357-361.

Glasson, W., and C. Tuesday (1970),  Environ.  Sci. Techno!., Vol.  4, p.  916.

	 (1971), Environ.  Sci. Technol., Vol.  5, p.  155.

Hecht, T. A.,  and J.  H.  Seinfeld (1972),  Environ. Sci. Technol.,  Vol.  6,
     p. 47.                              ~~

Hecht, T. A.,  P. M. Roth, and  J. H.  Seinfeld  (1973),  "Mathematical  Simulation
     of Atmospheric Photochemical  Reactions," Report R73-28, Systems Appli-
     cations,  Incorporated, San Rafael,  California.

Hecht, T. A.,  M. K. Liu, and D. C.  Whitney  (1974a),  "Mathematical Simulation
     of Smog Chamber Photochemical  Experiments,"  Report  R74-9,  Systems
     Applications, Incorporated, San Rafael,  California.

Hecht, T. A.,  J. H. Seinfeld,  and M. C.  Dodge (1974b), Environ.  Sci. Technol.,
     Vol. 8, p. 327.                                   	

Hefter, J. H., T. A.  Hecht, and G.  S. Hammond (1972),  J.  Am.  Chem.  Soc.,
     Vol. 94,  pp. 2793-2797.                          	

Heuss, J. M.,  and W.  A.  Glasson (1968),  Environ.  Sci.  Technol.,  Vol.  2,
     pp. 1109-1116.                     	

Jaffee, S., and H. W. Ford (1967), J. Phys.  Chem..  Vol.  71, p.  1832.

Jaffee, R. J., and F. C. Smith, Jr.  (1974),  Presentation  at 67th  Annual
     Meeting of the Air Pollution Control Association, Denver,  Colorado.

Johnston, H. S., and R.  Graham  (1974), Canad. J.  Chem.,  Vol.  52,  pp.  1415-1423.

Johnston, H. S., J. N. Pitts,  Jr., J. Lewis,  L.  Bufonte,  and  T.  Mottershead
     (1970), "Project Clean Air," University of California Task Force
     Assignments, Vol. 4.

Jones, G.R.H., and R. J. Cvetanovic  (1961),  Canad.  J.  Chem.,  Vol. 39,
     pp. 2444-2451.

Kopczynski, S. L., A. P. Altshuller, and F.  D.  Sutterfield (1974), Environ
     Sci. Techno!.. Vol. 8, pp. 909-918.                           	"

-------
                                                                      156
 Leighton,  P. A.  (1961), Photochemistry of Air Pollution (Academic Press,
      New York).                          -                *


 Liu,  M. K.  (1974), private communication.


 Mabely, W.  R   and A  G. Hendry (1974), Stanford Research Institute,
      Menlo  Park, California.


 Mendenhall, G. D., D. W. Golden, and S. W.  Benson (1974), "The Very Low-
      Pressure Pyrolysis (VLPP) of n-Propyl  Nitrite, tert-Butyl Nitrite
      n2n^™yl ^^^"^te. Constants for Some Alkoxy Radical Reactions,"
      Department of Thermochemistry and Chemical Kinetics, Stanford Research
      Institute, Menlo Park, California.

 Miller, D.  F. (1975), private communication.


 Morris, E.  D. , Jr., and H. Niki (1971), J.  Phys. Chem. . Vol. 75, pp. 3640-3641

 _  (1973), J. Phys. Chem.. Vol. 77, pp. 1929-1932.


                                                     > J- ^' Chem. Soc..
Morrison  R. T   and R. N. Boyd (1971), Organic Chemistry. Second Edition
     (Allyn and Bacon, Incorporated, Boston, Massachusetts).
Mulcahy, M.F.R. , JR. Stephen, and J.  C.  Ward (1967), J.  Phys. Chem..
     vol. /I, pp. 2124-2131.


Niki  HE. E  Daby,'and B  Weinstock  (1972), "Photochemical  Smog and Ozone
     Reactions," American Chemical  Society, Washington, D. C.

Noeh, S., B. H. Nordstrom, and H.  S.  Johnson (1974),  "Kinetic  Study of the
     Gas-Phase Formation of Nitrous Acid,", Department of  Chemistry, University
     of California,  Berkeley, California.

O'Brien, R.  J. (1974), Environ. Sci.  Techno! .. Vol.  8, p.  579.

O'Neal, H. E., and C.  Blumstein (1973),  Int. J.  Chem.  Kinet..  Vol. 5, pp.  397-413.

Reynolds, S. D.,  M.  K  Liu, T. A.  Hecht,  P. M. Roth,  and J.  H. Seinfeld (1974),
     Atmos.  Environ.,  Vol. 8, pp.  563-596.


Schuck, E. A., E. R.  Stephens, M.  A.  Price, and K.  R.  Darnall  (1972), "Reaction
     of Peroxyacetyl  Nitrate with  Nitric  Oxide," Statewide Air Pollution
     Research Center,  University of California at Riverside.

Schulten, H. R.,  and  U.  Schurath (1975),  J. Phys. Chem.. Vol.  79, pp.. 51-57.

-------
                                                                      157
Seinfeld, J. H., T. A. Hecht, and P.  M.  Roth (1973), "Existing Needs in the
     Experimental and Observational  Study of Atmospheric Chemical  Reactions,"
     GPA Report EPA-R4-73-031, Systems Applications, Incorporated, San Rafael,
     California.

Sidebottom, H. W., C. C. Badcock, 0.  G.  Calvert, B.  R.  Rabe,  and E. K. Damon
     (1971), J. Am. Chem. Soc.. Vol.  93, pp. 3121-3128.

Sidebottom, H. W., C. C. Badcock, G.  W.  Jackson, J.  G.  Calvert, G. W.
     Reinhardt, and E. K. Damon (1972),  Environ. Sci.  Techno!., Vol. 6, pp. 72-79.

Slagle, I. R., J. R. Gilbert, R. E.  Grahm, and D.  Gutman (1974), "Direct
     Identification of Reactive Channels in the Reactions of  Hydroxyl  Radicals
     with Allene, Propylene, and 2-Butene," Department of Chemistry, Illinois
     Institute of Technology, Chicago, Illinois.

Smith, J. P., and P. Urone (1974), Environ. Sci. Techno!., Vol. 8, pp. 742-746.

Spicer, C. W., and D. F. Miller (1974),  "Nitrogen Balance in  Smog Chamber Studies,"
     presented at the 67th Annual Meeting of the Air Pollution Control
     Association, Denver, Colorado.

Stedman. D. H., C. H. Wu, and H. Niki (1973), J. Phys.  Chem.,  Vol. 77, pp. 2511-
     2514.

UCR Monthly Report No. 4 (1974), "A  Systematic Investigation  of Varibles in
     Photochemical Systems in Smog Chamber:  An Experimental  Basis for Modeling
     Photochemical Air Pollution, "  Statewide Air Pollution Research Center,
     University of Calfornia, Riverside, California.

Wilson, W. E., Jr., and A. Levy (1969),  "A Study of  Sulfur Dioxide in Photo-
     chemical Smog II," 2nd Annual Progress Report to American Petroleum
     Institute, Project S-ll.

	 (1970), JAPCA, Vol. 20, pp. 385-390.

Winer, A. (1975), private communication.

Wu, C. H., and H. Niki (1975), Environ.  Sci. Techno!.. Vol. 9, pp. 46-52

Yeung, C.K.K., and C. R. Phillips (1973), Atmos. Environ., Vol. 7, p.  551.

-------
                                    TECHNICAL REPORT DATA
                             (I lease read Inunctions on the reverse before completing)
   EPA-650/4-75-026
 4. TITLE AND SUBTITLE
   Mathematical Modeling of Simulated
   Photochemical  Smog
                                                             3. RECIPIENT'S ACCESSION-NO.
                                          5. REPORT DATE
                                              June 1975
                                          6. PERFORMING ORGANIZATION CODE
   Paul   A. Durbin, Thomas  A.  Hecht, and Gary  Z.  Whitten
                                          8. PERFORMING ORGANIZATION REPORT NO

                                               EF75-62
 9. PERFORMING ORGANIZATION NAME AND ADDRESS
   Systems Applications,  Inc.
   950 Northgate Drive
   San Rafael, CA  94903
                                          10. PROGRAM ELEMENT NO
                                                1A1008
                                          11. CONTRACT/GRANT NO.
 12. SPONSORING AGENCY NAME AND ADDRESS
   U.  S.  EPA
   Office of Research And  Development
   National  Environmental  Research Center
            Trianglg Park,  M   r.    977H
                                                                 68-02-0580
                                          13. TYPE OF REPORT AND PERIOD COVERED
                                            Final (June'74 -June '75)
                                          14. SPONSORING AGENCY CODE    "*	
 15.
                 JOTES
 16. ABSTRACT
  The  continued development and  testing of a kinetic mechanism for photochemical
  smog formation is described.   Detailed mechanisms containing the individual
  chemical  reactions occurring in  irradiated propylene,  n-butlne. toluene-
  Nh^h^dHpropylfn!:NY5?2 ?ystems  were Postulated and used to simulate  smog
  chamber data   A theofetieal evaluation was made of  the contribution  of  such
  chamber effects as light source  spectrum decay and surface reactions  ?n  tho
  reactivity  of the chamber mixture.   The applicatl  of kineti? s mu at on to
                          reactlvi^ and ozone production in smog systems is
17.
_. T

3.
DESCRIPTORS
              KEY WORDS AND DOCUMENT ANALYSIS
  Photochemical Modeling
  Chemical Kinetics
  Atmospheric Chemistry
^DISTRIBUTION STATEMENT
 Unlimited
   Form 2220-1 (9-73)
                            b.lDENTIFIERS/OPEN ENDED TERMS
                            19. SECURITY CLASS (THIS Report)'
                              I. SECURITY CLASS I This page)
                              Unclassified
                                                                          c. COSATI Field/Group
21. NO.OF PAGES

    160
                                                                          22. PRICE

-------