Ecological Research Series
SMOG CHAMBER CONFERENCE PROCEEDINGS
Environmental Sciences Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triani
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into five series. These five broad
categories were established to facilitate further development and application of
environmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The five series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
This report has been assigned to the ECOLOGICAL RESEARCH series. This series
describes research on the effects of pollution on humans, plant and animal
species, and materials. Problems are assessed for their long- and short-term
influences. Investigations include formation, transport, and pathway studies to
determine the fate of pollutants and their effects. This work provides the technical
basis for setting standards to minimize undesirable changes in living organisms
in the aquatic, terrestrial, and atmospheric environments.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/3-76-029
April 1976
SMOG CHAMBER
CONFERENCE PROCEEDINGS
B. Dimitriades
(Chariman)
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Environmental Sciencfe Research .Laboratory
Research Triangle Park, North Carolina 27711
April 1976
-------
DISCLAIMER
This report has been reviewed by the Environmental Sciences
Research Laboratory, U.S. Environmental Protection Agency, and
approved for publication. Mention of trade names of commercial
products does not constitute endorsement or recommendation for
use.
ii
-------
CONTENTS
LIST OF FIGURES . , v
LIST OF TABLES viii
1. INTRODUCTORY REMARKS. . . . : 1
2. SMOG CHAMBER PERFORMANCE 3
COMPARISON OF SMOG .CHAMBERS 3
EFFECTS OF CHAMBER DESIGN FACTORS 4
INTERPRETATIONS OF CHAMBER DESIGN EFFECTS 6
DISCUSSION ......". 18
REFERENCES FOR CHAPTER 2 22
3. ANALYTICAL TECHNIQUES IN SMOG CHAMBER STUDIES 25
MEASUREMENT OF NITROGEN OXIDES 25
Colorimetric Analyzers 25
Chemiluminescence Analyzers 25
Calibration 26
Interferences 26
Comparison of Methods 26
Discussion 28
IN SITU MEASUREMENTS 29
DISCUSSION 35
AEROSOL MEASUREMENT 37
MEASUREMENT OF OXYGENATED SULFUR COMPOUNDS 40
Sulfur Dioxide .43
Sulfate Aerosols 45
Discussion 46
QUALITY CONTROL 46
Calibration 46
Statistical Approach 47
Measures of Reactivity 50
REFERENCES FOR CHAPTER 3 53
4. BACKGROUND REACTIVITY 57
INTERPRETATION OF BACKGROUND REACTIVITY DATA 57
DISCUSSION 59
REFERENCES FOR CHAPTER 4 61
5. COMPARABILITY OF SMOG CHAMBER ATMOSPHERE AND REAL ATMOSPHERE ... 63
A METHOD FOR COMPARING SMOG CHAMBER DATA WITH ATMOSPHERIC DATA . . 63
Sampling and Data Collection £4
iii
-------
Page
Discussion 66
STATIC AND DYNAMIC SMOG CHAMBER TESTS &?
SIMULATION OF NATURAL SUNLIGHT T . 71
OUTDOOR SMOG CHAMBERS 72
REFERENCES FOR CHAPTER 5 90
6. VALIDITY AND UTILITY OF SMOG CHAMBER DATA 91
INTRODUCTORY REMARKS 91
THE LACAPCD VIEWPOINT (HAMMING TRANSFORM) 92
Summary 92
Presentation 93
Discussion 94
THE EPA VIEWPOINT 98
CALIFORNIA AIR RESOURCES BOARD VIEWPOINT 102
Introduction 102
Environmental Chamber Studies 103
Conclusions 112
Unresolved Questions 113
SMOG CHAMBER STUDIES OF POLLUTANT PRECURSOR RELATIONSHIPS 113
REFERENCES FOR CHAPTER 6 114
7. CONCLUSION 115
LIST OF ATTENDANTS 117
IV
-------
LIST OF FIGURES
Page
Composite Photochemical Test Results for Teflon Film Surfaces ...... 5
Composite Photochemical Test Results for Pyrex Surfaces ..... .... 5
Composite Photochemical Test Results for Aluminum Surfaces ....... 6
Composite Photochemical Test Results for Stainless Steel Surfaces ..... 6
Species in Irradiated Propylene/NOx System that Dissociate on
Absorption of Radiation Present ......... ............ 7
6 Comparison of Full and Cut Spectrum Model Simulations ......... 9
7 Effective Quantum Yields for Propylene Destruction by Absorption of
Photolytic Energy by Nitrogen Dioxide .................. 11
8 Simulation of Full Spectrum Runs for Various Heterogeneous Rates,
NO + NO2 + H2O — £—2 HONO ..................... 13
9 Relative Importance of O + Cglf, and HONO + hv in Initiating Reaction. . 14
10 Simulation of Full Spectrum Runs With Preestablished HONO Equilibrium
and Without Preestablished HONO Equilibrium .............. 15
11 Reactions Controlling Ozone Production ................. 15
12 Graphs of the (O3)(NO)/(NO2) Function Versus Time for Different
Surfaces ................................ 16
13 Reactions of Radical Species with Nitric Oxide .............. 16
14 Alternative Reactions of Peroxy Radicals Leading to Ozone Formation ... 17
15 Concentration-time Profiles for NO, NO2, Os, and Propylene at 15°
and 35° C ...................... .......... 20
16 Predicted Concentration-time Profiles for NO, NO2, O,, and Propylene
at 0, 50, and 100 Percent Relative Humidity ......... ...... 21
17 Comparison of Methods for Monitoring NO and NO2 During a Smog-
chamber Experiment ........................... 27
18 Nitrogen Balance During Irradiation of Propylene-NOx .......... 28
19 Nitrogen Balance During Irradiation of Synthetic Auto Exhaust ...... 29
20 Single Beam Plot in CH Absorption Region for 1 Atmosphere of Tank
Air with Addition of 5 ppm 1 , 1 Dichlorethylene and 0 . 1 ppm Molecular
Chlorine, Before Irradiation. Path Length, 500 Meters; Resolution,
1 cm"1 ................................. 34
-------
Figure Page
21 Single Beam Plot in CH Absorption Region for 1 Atmosphere of Tank
Air with Addition of 5 ppm 1,1 Dichlorethylene and 0.1 ppm Molecular
Chlorine, after 15 Minutes of Irradiation. Path length, 500 Meters;
Resolution, 1 cm" I 35
22 Ratio Plot, Spectrum of Products over Spectrum of Reactants; 5 ppm
1,1 Dichlorethylene and 0.1 ppm Molecular Chlorine in Air, after 15
Minutes of Irradiation. Path Length, 500 Meters; Resolution, 1 cm"-'-. . . 36
23 Ratio Plot, Spectrum of Reactants over Spectrum of Tank Air; 5 ppm
1,1 Dichlorethylene and 0.1 ppm Molecular Chlorine in Air, Before
Irradiation. Path Length, 500 Meters; Resolution, 1 cm"1 37
24 Ratio Plot, Spectrum of Products over Spectrum of Tank Air; 5 ppm
1,1 Dichlorethylene and 0.1 ppm Molecular Chlorine in Air, after 10
Minutes Irridiation. Path Length, 500 Meters; Resolution, 1 cm~l 38
25 Ratio Plot, Spectrum of Products over Spectrum of Reactants; 5 ppm
1,1 Dichlorethylene and 0.1 ppm Molecular Chlorine in Air, after 10
Minutes Irradiation. Path Length, 500 Meters; Resolution, 4 cm"* .... 39
26 Sulfur Dioxide Removal During Irradiation of Propylene-NOx 43
27 Comparison of Methods for Sulfur Dioxide Analyses During Smog-
Chamber Irradiations 44
28 Calibration Methods Used for Chemiluminescent NOV and O? Instruments. . 46
^% *J
29 Comparison of Typical Reaction Data for Indoor and Outdoor Smog
Chambers 54
30 Injection and Back Flush of Ambient Air Sample 65
31 Ambient Air Photolysis 67
32 Maximum 4-hour-average Ozone, Kane, Pennsylvania, Versus Maximum
Temperature at Nearest Reporting Stations, October 1974 68
33 Effect on Ozone Generation by Dilution of Reactants: NO, 2»
October 7, 1974 . 69
34 Effect of Reaction and Dilution on a Simulated Urban Mix of Hydrocarbons,
October 7, 1974 70
35 Chamber Simulation of Extended Urban Influence on Surface Ozone
Concentration . . 71
36 The University of North Carolina Smog Chamber 73
37 Comparison of Total Solar Radiation on Horizontal Surface with Rate of
Photolysis of NO2 (Kj) (Latitude 35.72°), September 19, 1974 76
38 Diurnal Variation of Solar Radiation and Temperature, May 7, 1974 .... 76
39 Diurnal Variations of Solar Radiation and Temperature, May 19, 1974. . . 77
40 Comparison of Concentration-time Profiles from the Two Chamber
Compartments (Red and Blue), May 7, 1974 77
41 Concentration-time Profiles, May 19, 1974 78
vi
-------
Figure
42 Rate of NO Disappearance and NO2 Appearance for a Clear Day,
October 13, 1974 78
43 Rate of NO Disappearance and NO2 Appearance for a Partly Cloudy Day,
May 19, 1974 .- . 79
44 Comparison of Ozone Profiles and Solar Radiation Profiles for
Propylene Runs 80
45 Diurnal Variation of Solar Radiation and Temperature, September 19,
1974 81
46 Concentration-time Profiles, September 19. 1974 81
47 Diurnal Variations of Solar Radiation and Temperature, September 23,
1974 82
48 Concentration-time Profiles, September 23, 1974 82
49 Comparison of Ozone and Solar Radiation Profiles for Mix Runs. 84
50 Effect of Temperature on Maximum Rate of NO Disappearance ........ 85
51 Comparison of Kinetics Model and Actual Data for May 7, 1974 ...... 86
52 Model Results for Constant Kj=0.1, 0.3, May 7, 1974, Conditions. .... 87
53 Model Results for Constant K]=0.2, 0.4, May 7, 1974, Conditions 88
54 Concentration-time Profiles for Dual Run in University of North
Carolina Chamber, Effect of Initial NO2/NOX, May 19. 1974 89
55 Ozone as a Function of HC and NOX, from Smog Chamber Data 95
56 Time to Maximum NO2 as a Function of HC and NOX, from Smog
Chamber Data 96
57
Variation of (O^}air/f[O^lch With Year 98
58 Upper Limit Oxidant Values in the South Coast Air Basin as a Function
of Average 6 to 9 a.m. Nonmethane Hydrocarbon Concentrations, 1971
Data. Oxidant Concentrations are Maximum 1-hour Values for 12
Stations; NMHC Concentrations are Average Values for 8 Stations ..... 99
59 Smog Chamber Data on Dependence of Oxidant on NMHC under Constant
NOY or HC/NOY Conditions 100
Jx J»- ••••••••••••••
, 60 Equal Response Lines Representing All Combinations of NMHC and NOX
Concentrations Corresponding to 0.08 ppm Oo of Oxidant .101
61 Reaction Profile for SAPRC Run No. 42E, 6-hour Irradiation 105
62 Reaction Profile for SAPRC Run No. 48E, 10-hour Irradiation 106
63 Ozone Isopleths from 6-hour Irradiations of HC-NOX Mixtures, SAPRC
Runs 10-48 (E). [Oj] Values Are Final 6-hour Ozone Concentrations (ppm) 108
64 Ozone Isopleths from 10-hour Irradiations of HC-NOX mixtures, SAPRC
Runs 10~48(E). [03] Values Are Final 10-hour Ozone Concentrations
Plus 3 Times the Final Rate of Ozone Formation 109
vii
-------
Figure ' Page
65 Ozone Rate Isopleths from 10-hour Irradiations of HC-NO Mixtures,
SAPRC Runs 10-48(E) 110
66 Oxidant-dosage Reactivity of Exhaust as a Function of NOX at Various
HC:NO,, Ratios Ill
LIST OF TABLES
Table
Page
1 Range and Average Values of Propylene/NOx Reactivity Data Obtained
with Various (9) Smog Chambers ; 3
2 Reactivity Rankings of Hydrocarbons as Measured in Various (7) Smog
Chambers 3
3 Ratio Values for Cut-to-fit Spectrum Parameters 10
4 Ratio Values for Deleted-to-complete Model (Full Spectrum) Parameters . . 10
5 Activation Energies of Reactions in the General Mechanism 19
6 Colorimetric NO2 Analyzer: Ozone Interference, [NO2J = 0.23 ppm 26
7 Infrared Detection of Pollutants 32
8 Aerosol Sampling Procedures 41
9 Characteristics of Selected Samplers 42
10 Confidential Interval on Bias in Measuring Ozone at the 1 ""eral Air
Quality Standard with a Chemiluminescent Instrument Calibi ated by
Replicated NBKI Measurements 49
11 Inorganic Reactions of Photochemical Smog Formation 51
12 Reactions Participating in the Total Ozone Decay Process 61
13 Summary of Measurements to be Made at University of California,
Riverside 66
14 Derivation of Continuous Chemical Actionometer for NOX (CCANOX) .... 75
15 Ozone Formation, SAPRC Glass Chamber 107
Vlll
-------
SMOG CHAMBER CONFERENCE
PROCEEDINGS
1. INTRODUCTORY REMARKS
B. Dimitriades, EPA
Following a welcome address to the conference, Dr. Dimitriades explained briefly the
purposes and theme of the Smog Chamber Conference. One of the purposes of the conference
relates to the U.S. Environmental Protection Agency's practice of holding regular conference-
type meetings to which all EPA contractors and grantees working in a certain problem area
are invited to present and discuss with EPA work progress and future plans and to coordi-
nate the various research projects.
Another reason for organizing this event and setting the theme to be "smog chamber
studies" is the strong needs that have developed lately for smog chamber evidence. In the
past, smog chamber studies focused on exploring the photochemical smog phenomenon, on
making reactivity characterizations of emissions, and on generating mechanistic evidence.
Presently, it is believed that the most important use of smog chambers is in generating evi-
dence on the relationships between photochemical pollutants and their precursors. Such
evidence serves as a part of the scientific basis, and in some cases, perhaps, as the entire
scientific basis of a photochemical pollutant control strategy.
Smog chambers have already been used in this latter fashion. Specifically, smog
chamber data have been used to derive such relationships between photochemical oxidant
(Ox) and its precursors, hydrocarbons (HC) and nitrogen oxides (NOX) . In this case, it so
happened that it was possible to derive such relationships from aerbmetric data also. Fur-
thermore, it so happened that EPA felt—not in agreement with everyone else—that the
aerometric data relationships should be used as the primary basis for an oxidant control
strategy. However, such preference for the aerometric data, in this instance, should not—
and did not—degrade the utility of the smog chamber method. It should be borne in mind that
the OX/HC/NOX relationships are relatively simple and can be defined to a degree from aero-
metric data. However, this, in all probability, is not going to be the case with the other
-------
photochemical pollution problems, e.g., photochemical aerosol, sulfate, etc. The enormous
complexity of these problems could very well make smog chamber experimentation the only
feasible source of definitive, useful information.
For these reasons, EPA has initiated, in the last year, a fairly comprehensive program
of smog chamber studies. It is hoped and expected that this conference will provide an oppor-
tunity to review the smog chamber method and associated procedures, to identify and correct
possibly existing problems, and, finally and most importantly, to closely examine and discuss
the validity and utility of the smog chamber data that are to be obtained.
-------
2. SMOG CHAMBER PERFORMANCE
COMPARISON OF SMOG CHAMBERS -- B. Dimitriadcs, EPA
One systematic comparison of smog chambers and smog chamber data was completed in
1969, as a part of a cooperative effort sponsored by the Coordinating Research Council (CRC) .
The participating research groups used nine smog chambers to test a number of standard
HC/NO mixtures. Ranges and averages of test results are shown in Tables 1 and 2.
x
Table 1. RANGE AND AVERAGE VALUES OF PROPYLENE/NOX REACTIVITY
DATA OBTAINED WITH VARIOUS-(9) SMOG CHAMBERS
Initial concentrations
of propylene/NOx,
pptnC/ppm
9/3
9/1.5
9/0.5
1.5/0.25
RNOg» PPb/mi'n
Range
14-39
18-42
7-52
5-50
Average
25.3
28.9
22.2
13.4
Maximum oxidant, ppm
Range
0.5-1.4
0.5-1.4
0.2-1.0
0.2-0.6
Average
0.79
1.0
0.59
0.39
Table 2. REACTIVITY RANKINGS OF HYDROCARBONS AS
MEASURED IN VARIOUS (7) SMOG CHAMBERS
(Rankings of most and least reactive hydrocarbons:
1 and 7 respectively)
Hydrocarbon
n-hexane
Isooctane
Ethyl ene
Propylene
2-me-2-butene
0-xyl ene
Mesitylene
RN02
6-7
6-7
4-5
3
1
4-5
2
Maximum oxidant
6-7
6-7
3-5
1-3
1-4
3-5
1-4
Such an intercomparison of smog chambers is lacking in the following respects: (1)
All chambers were operated under constant light intensity conditions; simulation of the
diurnal variation of sunlight intensity was not attempted. (2) Reactant concentrations were
considerably higher than those of the background contaminants; therefore, differences in
results caused by differences in chamber contamination levels could not be seen, although
"Inquiries concerning this effort and results should be addressed to Coordinating Research
Council, Rockerfeller Plaza (30), New York, N. Y. 10020.
-------
they were certain to exist. (3) The comparison data do not include data on photochemical
aerosol formation; this is important since it is almost certain that aerosol reactivity differ-
ences among the various chambers do not parallel the rate of NC>2 formation (%JO?^ ant^
oxidant reactivity differences shown in Tables 1 and 2.
The observed disagreement in results among the various chambers (Tables 1 and 2),
unless explainable, causes the smog chamber method and data to be looked at with distrust.
Before smog chambers can be used with confidence for this intended purpose, it is necessary
that the following questions be answered satisfactorily:
1. Why do test results from chambers of various designs disagree?
2. Which chamber design gives valid results, that is, results applicable to the
real atmosphere?
Both questions have received some research attention and speculative answers have been
offered. Thus, the disagreement in results among chambers was investigated by Lockheed
under a CRC-EPA contract. Lockheed found that smog chamber design factors such as wall
material, surface-to-volume ratio, and radiation spectrum affect smog chamber test results.
Effects are also known to be caused by radiation intensity, condition of chamber wall surface,
and background contamination. The significance of these results is obvious considering
that in the CRC study the intercompared chambers differed considerably in design. Thus,
chamber surface-to-volume ratio ranged from 0.78 ft" 1 to 4.6 ft" 1, light intensity (in terms of
the NC>2 photolysis k, factor) ranged from 0.2 min to 0.4 min , and wall materials included
stainless steel, glass, Teflon, Tedlar, aluminum, and nickel. Whether the evidence obtained
by Lockheed and others, i.e., the evidence regarding the effect of chamber design on
chamber results, can explain completely the observed disagreements in results among
chambers (Tables 1 and 2) is not known at this time—no attempt ha been made yet. However;
one can reasonably infer that such disagreement is to be expected anc aence, it is not
necessarily a reason for distrusting the smog chamber data.
The question "which smog chamber design gives valid results" is discussed elsewhere.
Briefly, it is proposed that (1) acceptably valid results can be obtained using a smog chamber
such that real atmosphere conditions are closely simulated, and (2) for more confidence, a
smog chamber could be checked for "validity" by comparison with real atmosphere. More
specifically, it is proposed that a chamber be made of Teflon film, have a volume of-1000 ft
or so, be operated outdoors (or indoors, provided the diurnal variation of sunlight is simu-
lated) , and be operated under those conditions, static or dynamic, that favor oxidant
formation; it is also proposed that field data and smog chamber data be obtained that would
depict the time profiles of the oxidant forming process in the real atmosphere and in the
smog chamber under consideration.
EFFECTS OF CHAMBER DESIGN FACTORS,- R.J. Jaffe, Lockheed
An experimental study has been conducted of effects of materials, spectrum, surface-
to-volume ratio, and cleaning technique on the photochemical reactions observed in a smog
4
-------
chamber . •"• A unique chamber and lighting system was used , which permitted independent
variation in chamber materials and in light conditions. A xenon arc lamp-parabolic reflector
combination provided a collimated light beam . By orienting plates of materials parallel to
the beam, it has been possible to independently vary light conditions and materials.
The study included four materials — aluminum, Pyrex, Teflon, and stainless steel, and
two conditions each of spectrum, surface-to-volume ratio, and cleaning. A complete factorial
testing sequence was performed . All photochemical runs were at k j of 0 . 3 min as deter-
mined by frequent NO£ in N^ photolysis tests. The propylene (3 ppm)/NOx (1.5 ppm)
reaction system was used, at 95 F and 25 percent relative humidity. Initial NC^ content was
nominally 10 percent of NOX. Chamber background was 0. 1 ppmC. Tests were also conducted
at lower relative humidity. Several runs were made for the n~butane(3 ppm)/NOx(0.6 ppm)
system. A graphical depiction of some results is shown in Figures 1 through 4.
3.0
2.5
I 2.0
«
o
F 1.5
S 1.0
0.5
FULL SPECTRUM
CUT SPECTRUM
3.0
2.5
2.0
1.6
1.0
0.5
s CUTSPECTRUM
0 50 100 150 200 250 300
TIME, min
Figure 1. Composite photochemical test re-
sults for Teflon film surfaces.
0 50 100 150 200 250 300
TIME, min
Figure 2. Composite photochemical test re-
sults for Pyrex surfaces.
Effects of the different materials and of the two levels of each parameter have been
determined. The time to NCs maximum is shortest for stainless steel (70 min) followed by
aluminum (106 min), Pyrex (131 min), and Teflon (154 min). Maximum ozone concentration
increases in the order: Pyrex, aluminum, stainless steel, Teflon (for the full spectrum
condition).
The cutoff spectrum (little energy below 350 nm wavelength) strikingly lowers
reaction rates compared to the full spectrum. Surface-to-volume ratio measurably affects
the reactions. The variations in the two cleaning techniques do not affect as many of the run
characteristics except for stainless steel.
-------
FULL SPECTRUM
.CUT SPECTRUM
FULL SPECTRUM
CUT SPECTRUM
100 150 ZOO 250 300
TIME, min
Figure 3. Composite photochemical test re-
sults for aluminum surfaces.
50 100 150 200 250 300
TIME, min
Figure 4. Composite photochemical test re-
sults for stainless steel surfaces.
The presence of this large spectral effect (at constant k ,) was not anticipated. It is
reproduced in kinetic model simulations by changes in photolysis rate for nitrous acid,
formaldehyde, acetaldehyde, and hydrogen peroxide. It appears that heterogeneous catalysis
effects in nitrous acid production are responsible for the materials differences.
INTERPRETATIONS OF CHAMBER DESIGN EFFECTS -- P.S. Connell and
H.S. Johnston, U.C. Berkeley
For about 20 years, the smog chamber has been used as a to^ first for understanding
the reactions that generate photochemical smog, and second for deveL i •- g strategies of
smog abatement. The first of these goals seems nearly prerequisite for intelligent approaches
to the second. To this end of understanding the smog reaction, complete chemical mecha-
nisms have been proposed and tested by comparison with experimental smog chamber data.
These mathematical models are of two types, corresponding to atmospheric ,rdx smog cham-
bers and specific reactant hydrocarbon chambers.
Our model is presented as an attempt to include all important elementary reactions
occuring in the propene-NOx system studied in the Lockheed project. While "lumped"
mechanisms, describing atmospheric-mix smog reactions, can only be validated by di. ~t
comparison of the model prediction with chamber experiment, specific mechanisms composed
of elementary reactions, that is, reactions that actually occur on a microscopic level, can be
tested in both this fashion and by independent experiments in other systems, which yield
information on the reactions and their rates. Both bases of model testing pose problems.
Experimental data lack consistency from chamber to chamber in the major parameters of
characterization, such as time to NO£ maximum. Dependence on the material of construction,
history, techniques, and other factors, which are often not well characterized, has been
-------
demonstrated in the Lockheed study. Construction of a model based on reaction rates
measured specifically in other systems produces an incomplete picture at present, of
necessity , since many of the reactions involved have not been treated . It is important here
to remark that any serious model should, however, incorporate such direct experimental
data where it is available .
What is most immediately evident in the Lockheed project data is that chamber design
effects do exist and are indeed significant in affecting the commonly used descriptive smog
chamber reaction parameters . The two major effects encountered have been described as
the spectral intensity distribution effect and the dependence of the reaction rate on the nature
of surfaces included in the chamber. A third factor in the design of a smog chamber experi-
ment is less tractable to explanation by a model, the effect of techniques involved in per-
forming the run and the nature of analytical methods used to produce the data.
Two different spectral distributions were studied, one with energy down to about
300 nm (full spectrum) and one with wavelengths smaller than 350 nm effectively eliminated
(cut spectrum) . By increasing the intensity in the cut spectrum case , the NO2 absorption is
held fixed. Full spectrum runs are faster than cut spectrum runs for all surfaces.
Figure 5 shows species that are present in significant concentrations and show dis-
sociation on absorption for wavelengths present. The spectral effect perhaps admits to the
simplest explanation. The important light-absorbing species can be identified and the
absorptions and photolysis quantum yields can be measured in separate experiments . Knowing
the intensity of the light used in the chamber as a function of wavelength, the product of the
intensity, cross section, and quantum yield curves gives the rate constants of the photolytic
reactions. The proper model should then show quantitatively the same response to the chang-
ing spectral distribution. The model's response depends on both the primary photolytic
steps and the secondary reaction of the mechanism to changing these steps.
i '
(1) MONO + hi> (300-390 nm) - *~ HO + NO
(2) H2CO + hc (290-360 nm) - »- H + HCO
(2a) (300-370 nm), - *- H2 + CO
(3) CHaCHO + lw (300-350 nm) - *- CHs +
(3a) (300-350 nm) - *• CH4 + CO
(4) HOOH + hf (300-370 nm)j - > 2HO
(S) 03 -Hu> (290-31 0 nm) - *-
Figure 5. Species in irradiated propylene/NOx
system that dissociate on absorption of radia-
tion present.
Surface effects present greater problems since heterogeneous rates and mechanisms
are not well known and depend on individual surface history, for example the "virgin sur-
face" effect for aluminum and stainless steel . Several reactions that must occur in smog
chambers are known to be largely or entirely heterogeneous . The direction of the materials
-------
effect must in this case be matched by reasonable adjustment of the rates of these known
heterogeneous reactions.
The data produced in the Lockheed study show certain peculiarities that cannot be
reproduced in a model mechanism. The apparent growth of the total "amount of NOX species
during the first portion of the reaction cannot, of course, be accounted for in any model
conserving matter. Changes of species concentration as a result of reaction during sampling
and sampling lag, as with ozone and NO, introduce uncertainty into the data, making direct
comparison with predicted concentrations more difficult.
The computer program used in this study is a chemical kinetics solving package
developed by G. Z. Whitten. It employs a coupled differential equation-solving algorithm
written by Gear.4 This uses an Adams-Bashforth-Moulton^ predictor-corrector method of
numerical integration to generate concentration versus time profiles for the species involved
in the mechanism. The chemical mechanism derives largely from the modeling work of
Demerjian, Kerr, and Calvert^ and that of Niki, Daby, and Weinstock.? Climatic Impact
Assessment Program (CIAP)° rate constants were used if tabulated, as well as more recent
experimental work, as for the mutual recombination reaction of methyl-peroxy radicals.
The photolytic rate constants used for the full and cut spectral distributions for HONO and
O(^D) production were obtained from work in our laboratory. Other photolytic constants
were obtained from Demerjian et al. Constants for which no experimental basis exists were
taken mostly from Demerjian et al. Figure 6 is a graph of propene, nitrogen dioxide, and
j
ozone concentrations generated by the model for full and cut spectral distributions. For
comparison with the experimental data, the effect can be made partially independent of pos-
sible deficiencies in the model by considering the ratio of cut to full spectrum time parameters,
characteristically around 1.6 for the smog chamber runs. These ratios are shown in Table 3.
Model results bracket the experimental numbers, except in the case i -oropylene destruction
half-times. This ratio is lower than observed because the model tends not to account for
propylene loss toward the end of the reaction, allowing full and cut spectrum propene con-
centrations to approach each other for long times«
Determination of which reactions are most responsible for the spectral effect was
undertaken by deletion, one by one, of full spectrum photolysis rates and replacement by the
corresponding cut spectrum rate. The ratio between the time parameters of the deleted model
to the full spectrum complete model was calculated. If the system is insensitive to the
reaction replaced at the rate assigned, the ratio will be one and the reaction is unimportant.
For more important reactions the ratio will approach the corresponding cut-to-full-specU urn
ratio. Table 4 shows these ratios. The photolysis of nitrous acid appears most significant
in accounting for the spectral effect, followed by H2CO; CI^CHO, and HOOH. The produc-
tion of O(1.D) is seen to be unimportant in the present instance.
Although the validity of these results depends upon the validity of the rest of the model,
the spectral effect appears to be essentially explained by the reactions considered.
-------
e
o
CUT SPECTRUM
TIME, min
Figure 6. Comparison of full and cut spectrum model simulations.
More difficult to explain than the spectral effect is the great difference observed between
runs with Teflon surfaces and those with a stainless steel surface present. While the origin
of the spectral effect is identified in principle and requires only knowledge of cross sections,
quantum yields, and spectral intensities for explanation, surface reactions are less well
9
-------
Table 3. RATIO VALUES FOR CUT-TO-FULL SPECTRUM PARAMETERS
N02 Tmax
°3 Tmax
50% propylene
destruction
time
Teflon
1.58
1.49
1.50
Stainless
steel
1.49
1.39
1.43
Model
Fast NO + N02 + H20
1.81
1.51
1.39
Slow NO + N02 +
1.37
1.38
1.31
r
H20
Table 4. RATIO VALUES FOR THE DELETED-TO-COMPLETE MODEL (FULL SPECTRUM) PARAMETERS
N02 Tmax
°3 Tmax
50% propylene
destruction
time
Deleted compound
HONO
1.50
1.28
1.26
H2CO
1.26
1.0
1.14
CH3CHO
1.17
1.04
1.03
HOOH
1.12
1.01
1.00
Os * 0('D)
1.0
1.0
1.0
understood, and attempted explanations can be only speculative. It is, however, clear from
the Lockheed work that modeling without considering the extent of heterogeneous contribu-
tions is unwise.
Generally speaking, stainless steel should be the most active surface, both for pro-
moting reaction rates catalytically by helping to overcome activational energy barriers and
for quenching radicals and removing species from the gas phase by adt -otion on the wall.
Teflon should be the most inert of the surfaces used. Indeed, radicals are capable of
bouncing off Teflon without being quenched.
Observation of the smog reaction rate in a chamber with stainless steel surfaces en-
closed shows it to be much faster than the same chamber containing Teflon surfaces instead.
Since quenching of radicals would result in slowing the overall rate of reaction, lengthening
the time to NO £ maximum, the surface effect must be a combination of opposing factors. The
ability of the metal surface to promote certain reactions must overcome its ability to destroy
the radical chain carriers.
By considering the NC>2 dose as a function of time from the start of the run, the number
of photons absorbed by NO£ and, therefore, the number of reaction-initiating oxygen atoms
can be deduced for any time period. This value, <£, is the product of the photolytic rate con-
stant, in this case, 0.2 min , and the NC>2 concentration integrated from time T, to T,.
J- £*
Knowing the number of oxygen atoms produced and the number of propene molecules de-
stroyed, the quotient is an effective quantum yield for propene destruction by absorption of
photolytic energy by NO,. In Figure 7, we see this plotted for the four surfaces for various
Lt
10
-------
0.16
0.14
0.12
0.10
0.08
0.06
0.04
0.02
O ALUMINUM
• PYREX
A TEFLON
• STAINLESS STEEL
I
O
A
O
O
A
O
A
0-25
0-50 50-75
%PROPYLENE DESTRUCTION
75-END
Figure 7. Effective quantum yields for propylene destruction by absorption of photolytic energy
by nitrogen dioxide.
time periods during the run. Points were obtained by averaging results for a particular set
of runs. The quantum yield does not exceed 0.15 in any case, showing that radical termina-
tion reactions predominate over chain propagation steps throughout the reaction. The
quantum yield for all surfaces converges to around 0.1 late in the reaction, the major observ-
able difference between the surfaces being the larger initial quantum yield for stainless steel
runs. Evidently the stainless steel surface speeds the initiation of reaction.
Of the species known to be present, ^O^, HOOH, and Og will decompose on the walls.
^2^5 reacts with adsorbed water to yield nitric acid, which will probably remain stuck. The
reaction of NO. NO2» and I^O is also thought to be surface catalyzed. An upper limit of 1.3E-
20, in cm /molecule-sec, (where E is the activation energy in kcal/mole) has been placed on
the homogeneous reaction of nitrogen pentoxide, while decomposition half-times for ^05 even
in a "dry" system are on the order of minutes, so that in the chamber this reaction is largely
heterogeneous. Data from the NOX reducing column late in a run show that most of the
nitrogen-containing species are no longer in the gas phase, since the efficiency of the column
for converting alkyl nitrates, nitrites, NO2, and HNO, to observable NO should be high.
11
-------
The principle nitrogen-containing product, nitric acid, is made both homogeneously, sticking
on the wall subsequent to collisions with it, and heterogeneously. Ozone decay half-times
were measured for the various surfaces, and these results were used to derive first-order
decay constants.
The importance of HONO photolysis in explaining the spectral effect suggests that the
surface catalyzed reaction of NO, NO-, and ^O to give two molecules of nitrous acid may
also contribute to the materials effect. The HONO photolysis product HO is an initiator and
radical chain carrier and as such could account for the increased initial quantum yield. The
result of heterogeneously catalyzed nitrous acid formation is to increase the HO radical con-
centration at the outset of the reaction, providing an alternative pathway for initiation.
Measurements of the rate of this reaction as a homogeneous trimolecular reaction have in the
past been on the order of 1E-34 with a large heterogeneous component. These values have
been used by some modelers who generate times to NO£ maximum characteristic of stainless
steel runs. Longer times to NO2 maximum, encountered in Teflon runs, seem to be incom-
patible with this large value for the rate constant. The existence of a nonvanishing homo-
geneous rate is not established, but an unpublished study of the reaction shows it to be much
more rapid in a quartz-stainless steel system than in a strictly quartz cell.
Figure 8 shows full spectrum model results for varying rates of the nitrous acid forming
reaction. The "active" surface (higher rate) is seen to give a faster system, as is observed
experimentally. The relative importance of O+propene and HONO+hv in initiating the reaction
in a fast system is shown in Figure 9. For this model system, the production of HO by photo-
lysis of HONO is more important than the O+propene reactions. A further aspect of the HONO
effect is the waiting time before the lights are turned on in the chamber. In a fast system, an
equilibrium amount of HONO can be produced in the dark in about 20 minutes. In a slow
system, it would take longer than 2 hours. Figure 10 shows the difference in rate for a
system with a preestablished equilibrium concentration of HONO and for a system with no
initial NO.
The focus of abatement strategy has been an oxidant (ozone) level as the primary
deleterious component of photochemical smog. Ozone enters into the smog reaction upon the
photolysis of NO2 to produce oxygen atoms, which then mostly combine with oxygen to give
ozone. It participates at first by reconverting NO to NO2, then by reaction with HC to give
T
organic free radical species. The production of ozone in this regime is controlled by the
three reactions shown in Figure 11. We can assume that O atoms are present at a steady
state concentration. A photostationary ozone concentration expression can then be derived,
which is a function of the NO,,/NO concentration ratio and of the quotient of the constants j,
and k3- Thus the expression (03) (NO)/ (NO2) is constant if the photostationary state applies.
Figure 12 is a graph of this function versus time for various runs with different sur-
faces . The plots are made only during that period for which O3 and NO are measurable with
some accuracy, so that the total error probably does not exceed 50 percent. This period is
during and after the NO- maximum. For each of the runs shown, the function exceeds the
12
-------
I
K
O
I
U
e
u
k = 1x10-17 cm3/molecule-sec
k = 1x10-19
300
TIME, min
Figure 8, Simulation of full spectrum runs for various heterogeneous rates,. NO + NO^ + H20 — J^*-2 MONO.
photostationary value, indicated by the dotted horizontal line, shortly after the appearance
of ozone and subsequently increases rapidly. It seems that ozone buildup is accelerated by
some other mechanism .
A major characteristic of smog reactions is the "excess" conversion of NO to
by radical species, which, when rapid, provides a mechanism for pumping up the ozone
concentration . This reaction is shown in Figure 13 . Including this reaction in our prior
three step scheme and reforming the same concentration ratio, the new factor appears in the
denominator . Thus if conversion by peroxy radicals competes equally with conversion by
ozone, the ratio is reduced by a factor of two. The ozone concentration is increased by
13
-------
o
U
HONO + lu> -*-HO + NO
DELETED
0 + C3H6 —^-PRODUCTS
DELETED
0 100 200 300
TIME, min
Figure 9. Relative importance of 0 + CsHs and HONO + h»> in initiating reaction.
virtue of increasing the NO_/NO ratio, but the increase is not as rapid as the increase in
the ratio itself.
What is needed to produce both excess ozone and the large (03) (NO)/(NO^) ratio
is an ozone forming reaction that is independent of the gas-phase NO--NO mechanism. A
possibility is photolysis of peroxy radicals to form O atoms. Another alternative is the
14
-------
<
K
O
C9
WITHOUT PREESTAB-
LISHED HOMO EQUILIB-
TIME, min
Figure 10. Simulation of full spectrum runs with preestablished HONO equilibrium and without preestablished
MONO equilibrium.
1.
II
NO + 0
2. 0 + 02 + M *-Os + M
3. NO+ 03
[NOZ1 k3
Figure 11. Reactions controlling
ozone production.
15
-------
0.20
•0.10
TEFLON
iN02/kNO+03 = S.lxlO-3 ppm
100
200
300
TIME, min
Figure 12. Graphs of the (03)(NO)/(N02) function versus time for different surfaces.
4.
[Q3l
[N02]
Figure 13. Reactions of radical species with nitric oxide.
16
-------
production of ozone by the reaction of peroxy radicals with molecular oxygen (Figure 14). This
reaction can double the ratio for radicals present at the 1E-10 level if k, is on the order of
D
1E-18. This is a small required value, but the reaction appears to be endothermic for
and experimental work shows that it does not occur for the peroxyacetyl radical.
J5 ,
5. ROD- + hv *- RO- + 0
[03lss[NO] jt / jglROO-
[N021
jt / j5[ROO-]\
k3l h[N02l /
IF jg«< 1.0 sec'1
l<6
6. ROD- + 02 - *• RO- + 03
k6[ROO-][02]
[N021 k3 j,[N02]
"3
IF kfi w - = 2 x 10'18 MOLECULES/cm3-sec
[02]
Figure 14. Alternative reactions of
peroxy radicals leading to ozone forma-
tion.
The validity of the photostationary state in large chambers and out of doors in the
ambient atmosphere has apparently been established. The discrepancy in the data here is
then either due to a systematic overestimation of small NO concentrations or to wall effects
minimized in larger chambers.
The Lockheed data, as we have seen, point up several questions related to an attempt
to chemically model the findings. The effect of spectral intensity distribution on the rate of
reaction can be explained largely by change in HONO and I^CO photolytic rates. The
surface effect appears to be one of heterogeneous formation of nitrogen oxyacids , which
speed both the initiation of reaction and the removal of NO£ after maximum . The ozone
concentration problem in small chambers has not submitted to explanation. More accurate
techniques for measuring small levels of NO and ©3 as well as faster more direct techniques
for NOo should help to define the actual existence of a discrepancy. The importance of
nitrous acid initiation shows the necessity of obtaining quantitative rate data on its heterogen-
eous formation reaction as a function of surface or chamber, before this effect can be eliminated
as a source of difference between model and experimental results.
17
-------
DISCUSSION
Dr. T. A. Hecht of Systems Applications, Inc. , reported on recent SAI modeling
studies addressed to the effects of temperature and water vapor upon smog chamber measure-
ment results. Dr. Hecht's presentation entitled "On the Predicted Effects of Changes in
Temperature and Water Concentration on Smog Kinetics," was as follows:
In a recent study, mathematical simulations of a smog chamber experiment were
o
carried out using a kinetic mechanism to determine what effect changes in temperature or
water concentration have on the predictions . The base values used were those of a smog
chamber run conducted by EPA (Run 333):
[NO]0 = 1.25 ppm, [N02]0 = 0.08 ppm, [CsHeJ = 0.23 ppm,
[n-C4HiQ] = 3.41 ppm, [H20] = 16,000 ppm, and T = 25°C
For each simulation run, only one parameter was changed from the base values.
Simulations were performed for two different temperatures, 15°C and 35°C, with all
other factors kept the same . The rate constants at the new temperatures were calculated
from the base values of the rate constants (25°C) and from measured or estimated reaction
activation energies, shown in Table 5. ' ' Because the majority 'of the reactions in the
mechanism are thermal and have small positive activation energies , raising the temperature
resulted in an accelerated conversion of NO to NO2 and a decrease in the time to the onset
of Oj accumulation, as expected. Conversely, lowering the temperature noticeably slowed
the smog formation process. Concentration-time profiles for NO. NO3, 03, and propylene
for each of these two runs are presented in Figure 15.
Similar runs were carried out at two extreme conditions of relative humidity — 0 and
100 percent — at the base temperature (25°C) . These percentages correspond to 0 and
32,000 ppm of H2O, respectively. Predicted concentration-time profiles i.>p these two cases
are compared with the profile for the base case in Figure 16 . The increase in the water
concentration results in a faster conversion of NO to NO£, whereas complete elimination of
water results in a dramatic slowdown in the overall smog kinetics . Both of these effects
are attributable to changes in the production rate and equilibrium level of nitrous acid ,
governed by the reactions
NO + N02 + H20 T"*- HN02
Because it is virturally impossible — even with pumping and baking — to obtain a water
concentration of 0 ppm in existing smog chambers , one final run was carried out at 3 . 2 ppm
of water . The concentration-time profile obtained under these conditions differed from those
of the completely dry run by less than 2 percent after 6 hours of simulation time.
In urban areas, ambient temperatures and water concentrations change considerably
during the day and from one day to the next . Thus , the results of these simulation runs
suggest that it may be necessary to account for variations in temperature and water concen-
tration when modeling urban photochemical smog . Toward this end , smog chamber experi-
ments conducted at various constant levels of temperature and water concentration would
18
-------
Table 5. ACTIVATION ENERGIES OF REACTIONS IN THE GENERAL MECHANISM
No.
Reaction
cal/mole
Reference
1
2
3
4
5
6
7
8
9a
10a
na
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
N02 + hv
0 + 02 + M
03 + NO
0 + NO
0 + N02
0 + N02
°3 + N02
N03 + NO
N03 + N02
N2°5
' N205 + H20
NO + NO, + H,0
2 <-
2HN02
HN02 + hv
OH + N02
OH + NO
H02 + NO
H202 + hv
Propylene + 0
Propylene + 0,
Propylene + OH
n-butane + 0
n-butane + OH
ALD + hv
ALD + OH
R02 + NO
RC03 + NO
RC03 + N02
RO + 02
RO + N02
RO + NO
H02 + H02
H0? + R02
R02 + R02
-»• NO + 0
"* °3 + M
-* N02 + 02
+ N02
->• NO + 02
•*" NOO
•* NO, + 0,
•3 c.
-* 2N02
* N2°5
-> N02 + N03
•+ 2HN03
* 2HNOo
c.
->• NO + N02 + H20
+ OH + NO
" HN03
M HN02
* OH + N02
* 20H
-»• R02 + ^RC03 + %H02
-> RC03 + RO + ALD
•* R02 + ALD
•* R02 + OH
-> R02 + H20
-* 0.6R02 + 1.4H02
•+ 0.6RC03 + 0.4H02 + H2C
* RO + N02
* R02 + N02 + C02
* PAN
•* H02 + ALD
-*• RON02
-* RONO
- H2°2 + °2
-> RO + OH + 02
-+ 2RO + 02
0
-1
2.4
-1.9
0.6
-1
4.9
1.4
-2
19.4
0
0
9
0
-2.2
-2.2
2
0
0.1
3.8
1
5
1
0
0
1
0
0
6
0
0
0
0
0
Estimate
10
10
6
10
Estimate
10
11
6
10
Estimate
6
6
Estimate
10
10
Estimate
• Estimate
10
10
Estimate
Estimate
Estimate
Estimate
Estimate
Estimate
Estimate
Estimate
10
Estimate
Estimate
Estimate
Estimate
Estimate
aFor these simulations, Reactions 9 through 11 were combined into
the single reaction:
NO, + NO, + H,0 > 2HNOn
322 J
having the rate constant
K/\ "ii
k =
and an activation energy of -1.9 kcal/mole.
19
-------
<
cc
o
o
EPA RUN 333 CONCENTRATIONS (ppm)
NO = 1.25
NO 2 = 0.08
200
TIME, min
Figure 15. Concentration-time profiles for NO, NO2, 03, and propylene at 15° C and 35° C.
be most useful in ascertaining the effects of variations of these two parar eters on smog
kinetics .
Dr. T. E. Graedel (Bell Laboratories, Murray Hill, N. J.) did not attend the con-
ference . However , he did send some comments pertaining to factors affecting smog chamber
results. Dr. Graedel's comments were as follows:
1 . The results of smog chamber chemistry depend very much on accurate and com-
plete determinations of the flux and spectral characteristics of the radiation .
Absolute radiation measurements , rather than rate of decrease of a single photo-
chemically sensitive species, are required.
The presence of I^O, NO, and N©2 in smog chambers prior to irradiation is
likely to result in the formation of HNO2 and HNO^ , with subsequent production
of free radicals by photodissociation . Complete information on pre-irradiation
"induction periods" is thus desirable, as are measurements of the inorganic acid
concentrations .
20
-------
111
u
EPA RUN 333 CONCENTRATIONS (ppm)
NO = 1.25
N02 = 0.08
C3H6 = 0.23
-^. n-C4H10 = 3.41
"** — — _ T = 25° C
-100% RH
(32,000 ppm H20)
2 I
0.4
Figure 16. Predicted concentration-time profiles for NO, N02> 03, and propylene at 0, 50, and|IOO percent
relative humidity.
3. Much evidence is accumulating to indicate that the interactions between trace
gases and aerosols are very important in atmospheric chemical processes. The
absence in smog chambers of aerosols typical of the atmosphere may render even
more difficult the process of relating smog chamber results to atmospheric
processes.
Dr. G. J. Doyle (University of California at Riverside) commented that short-lived
intermediates [e.g., O(3P), OH.] are of no concern as far as wall quenching is concerned
for a reasonable surface-to-volume ratio and for a reasonably convection-free chamber. The
wall effects on concentrations of long-lived intermediates are of more concern, HNO2 being
a good example. One can use a simple model, in the case of wall losses, to generate a
pseudounimolecular rate, le=sd/vS; s=surface, v=volume, d=diffusion constant and S=
thickness of a "laminar layer," which can be roughly estimated from thermal convection
relationships involving chamber size and temperature gradients. If this constant is com-
parable to the sum of all reactions leading to the disappearance of the intermediate, then
21
-------
there can be an appreciable wall loss effect. This model can be extended to cover more
complicated cases. Its chief drawback is that it requires a fairly good knowledge of the
reaction mechanism.
Dr. T. Yang (Calspan) questioned whether efforts had been made to model the effect,
if any, of surface-to-volume ratio on smog chamber results. According to Dr. Yang, it
would seem possible that a modeling approach based on simultaneous assessment of diffusion
to chamber wall and the relevant reaction rates may lead to a prediction of an abrupt in-
crease in surface-to-volume effect after a certain breakoff point. This would be expected
from a combination of incubation requirement for heterogeneous catalysis and adsorption-
desorption equilibrium of a specific reaction intermediate governed in part by_its probability
of diffusion to the chamber wall surface.
In regard to the Lockheed observation that NOX in the smog chamber increased during
irradiation, Dr. Yang confirmed that small increases in NOX were also observed in the Cal-
span chamber.
In response to Dr. C. W. Spicer's (Battelle) question, Dr. A. J. Jaffe (Lockheed)
stated that 63 interference in the NOX measurement could not be the cause of the apparent
NOX increase during irradiation, because O_ concentration during the early reaction stages
is generally low, and because such interference should cause a negative, rather than posi-
tive , effect on NOX.
Dr. Dimitriades concluded the discussion with the comment that a large chamber
(e.g., 1000 ft^) made of Teflon film and operated outdoors or indoors but under varying
light intensity conditions may have advantages over other smog chamber designs, at least
insofar as studies of the oxidant-precursor relationships are concerned. Dr. Dimitriades
also suggested that more attention should be given by the modelers to the need for inclusion
of heterogeneous reaction steps in the oxidant formation mechanism.
REFERENCES FOR CHAPTER 2
1. Jaffe, R.J., F.C. Smith, and K.W. Last. Study of Factors Affecting Reactions
in Environmental Chambers; Final Report on Phase III. Lockheed Missiles and
Space Co., Inc., Sunnyvale, Ca. 1975.
2. Dimitriades, B. Use of Smog Chamber Data in Formulating Oxidant Control Stra-
tegies. U.S. Environmental Protection Agency, Research Triangle Park, N.C.
(Presented at Conference on Technical and Medical Bases for Control Strategies
of Photochemical Oxidants: Current Status and Priorities in Research. Riverside,
Ca. December 16-17, 1974.)
3. Whitten, G.Z. Rate Constant Evaluations Using a New Computer Modelling Scheme.
Lawrence Laboratory, Berkeley, Ca. (Presented at 167th American Chemical Society
National Meeting. Los Angeles. 1974.)
4. Hindmarsh, A.C. Gear: Ordinary Differential Equation System Solver. Lawrence
Laboratory, Livermore, Ca. Report UCID-30001, Rev. 2. .1972.
5. Gear, C.W. Numerical Initial Value Problem in Ordinary Differential Equations .
Englewood Cliffs, N.J., Prentice-Hall, 1971.
22
-------
6. Demerjian, K.L. , J.A. Kerr, and J.G. Calvert. The Mechanism of Photochemical
Smog Formation. In: Advances in Environmental Science and Technology (Vol. 4).
J.N. Pitts and R.L. Metcalf (ed.). New York, John Wiley and Sons, 1974. p. 1-
262.
7. Niki, H., E.E. Daby, andB. Weinstock. Mechanisms of Smog Reactions. Adv.
Chem. 113:16, 1972.
8. Chemical Kinetics Data Survey VII; Tables of Rate and Photochemical Data for
Modelling the Stratosphere. D. Garvin and R.F. Hampson (ed.). National Bureau
of Standards. Washington, D.C. Report NBSIR 74-430. January 1974.
9. Hecht, T.A., J.H. Seinfeld, andM.C. Dodge. Further Development of Generalized
Kinetic Mechanism for Photochemical Smog. Environ. Sci. Technol. 8_(4):327, 1974.
10. Johnston, H.S. et al. Atmospheric Chemistry and Physics (Vol. 4). Project Clean
Air, Task Force Assessments. University of California. 1970.
23
-------
3. ANALYTICAL TECHNIQUES
IN SMOG CHAMBER STUDIES
MEASUREMENT OF NITROGEN OXIDES _ D. Miller, BanMc
This discussion is limited to two methods of measuring nitrogen oxides (NO and
the automated Saltzman and the chemiluminescence techniques. Of the instruments available
for continuous nitrogen oxide measurements, these two are the most familiar to the author
and probably the most widely used in other smog chamber laboratories. Hopefully, spectro-
scopic methods for NO£ measurement, as well as other methods, will be reviewed in subse-
quent discussions.
Colorimetric Analyzers
The classic colorimetric method of Saltzman for measuring NO£ is based on absorption
of NO- in sulfanilic acid solution in which a specific reaction occurs between nitrite
ions and diazotizing-coupling reagents to form a pink color. 1 Most commercial instru-
ments incorporating this principle provide dual photocells such that the NO2 concentration
is read in one cell and the NO concentration is read in the second cell after oxidizing
NO to NO2- As oxidizer, we routinely use flutted filter paper impregnated with dichromate
solution. The dichromate oxidizer is at least 95 percent efficient when the papers are
just slightly moist and dark red in color. Generally, a new scrubber is prepared daily.
Chemiluminescence Analyzers
The chemiluminescence method-* is based on the reaction of NO with excess Og to form
an excited NO- specie that emits light in a continuum extending from 600 to 900 nm. A
Lt
cutoff filter at 600 nm is generally employed to absorb shorter wavelength emissions
generated by other Oj reactions. To use the chemilumenescence instruments to monitor
N©2 in addition to NO, the N©2 must first be reduced to NO. In the last few years, carbon-
based converters have become most popular. The reaction occurring at 200 to 250° C is
quite efficient
C + N02 CO + NO
Because of the familiarity that most of you have with these instruments I do not
intend to review the detection principles in any more detail, nor spend time discussing
optimal operating conditions, or the advantages and disadvantages of the two methods, or
the work that has been done on the converters. The most important consideration with
respect to obtaining accurate smog chamber data lies with the problems of interference
associated with the primary and/or secondary (conversion) principles of these instruments,
and I would like to present some data on that subject. Before that, however, calibration
methods should be mentioned.
25
-------
Calibration
It is common knowledge that low concentrations of nitrogen oxide standards in com-
mercial cylinders are unreliable. We redetermine the NO concentration of bottled stand-
ards by an O3~NO titration procedure somewhat similar to that used in confirming Oj
determinations. Generally, the concentration in the bottles remains fairly stable at
constant temperature. To determine the correct NC>2 concentration in pressurized cylin-
ders , the gas mixture is passed over a catalytic reducer, and the concentration is read
with a chemiluminescence analyzer previously spanned by the titration technique. The
known NO2 standard is then used to calibrate the automated Saltzman analyzer. Alter-
natively, a permeation tube can be used to calibrate the colorimetric analyzer and, upon
reduction, the chemiluminescence analyzer. If one is careful to exclude water from the
permeation systems, the NO£ permeation technique is quite reliable.
Interferences
Chemiluminescence Analyzers—No notable interferences have been reported for the chemi-
luminescence detection of NO. Positive interferences are known to occur when the cataly-
tic or carbon converters are used to reduce N©2 to NO. We have found that the efficiency
of some of these converters is near 100 percent for ethyl nitrate and, although incon-
sistent, sometimes 100 percent for nitric acid. PAN is also known to interfere, but we
are not certain of the response factor. The high-temperature (600 to 700° C) catalytic
converters are also efficient in oxidizing basic nitrogen species such as ammonia to NO.
Colorimetric Analyzers—There has been some controversy lately about negative inter-
ferences with the Saltzman method. At the time when the method was introduced, investi-
gation was made of the effect of potential interfering gases, and no serious problems
were evident.^ About a year ago, however, Stevens^ and his associates: at EPA reported
that ozone seriously interfered with the measurement of NO£. Some of thei. ^ata are
shown in Table 6. As noted, interference becomes quite severe at O3/NO2 ratios near
2—a condition that often occurs in the later part of smog-chamber experiments. Un-
fortunately, we have not had the opportunity to conduct similar interference testing
with our Saltzman-type analyzer, and I have no direct evidence to confirm or refute
EPA's findings. However, I will present some data that cast doubt as to the probable
extent of the interference reported.
Table 6. COLORIMETRIC N0? ANALYZER:
Comparison of Methods ^ INTERFEREN£Ef
Figure 17 compares the results of [N0?] = 0.23 ppm
measuring NO and NO? by the Saltzman rn i /mr» ^ n ± ^ • r * -,
^ [03]/[N02] Percent reduction in [NOJ
and chemiluminescence methods. There — '
0
are some slight differences in the initial -, n
I • U
readings, but these are no doubt associated ? n
with calibrating and zeroing. There is , 0
nearly perfect agreement between the . «
0
5
21
40
48
26
-------
<
DC
111
O
o
u
SALTZMAN
CHEMILUMINESCENCE
0.9 ppm
0.5 ppm
0.5 ppm
60 percent
IRRADIATION TIME, hours
Figure 17. Comparison of methods for monitoring NO and N02 during a smog-chamber experiment.
-v,
two methods for NO analysis. There is also fairly good agreement between the N©2 readings
up until the time of maximum NO2> the small difference between the two NO2 curves being
largely that occurring initially. After the time of maximum NO2, however, there is a serious
(nearly twofold) difference in the NO2 concentrations and the slopes of the NO2 removal rates.
During the last hour of irradiation where the O3/NO2 ratio is >2, it might be argued that
the factor of 2 difference in NO2 is due to 03 interference with the colorimetric analyzer,
in accord with the findings of Stevens et al. Alternatively, it might be argued that the chemi-
luminescence analyzer is responding to nitrogen-containing products other than NO2- Most
of the data from our work , particularly that where we have attempted to make a nitrogen
balance , ^ would support the latter contention .
Figure 18, for example, is a profile of a propylene-NOx irradiation experiment where
most of the initial NOX was NO2 . As noted PAN and nitric acid were monitored here (as well
as alkyl nitrates, which turned out to be of insignificant concentrations) , and I have
plotted total accountable nitrogen as the sum of NO, NO2» PAN, and HONO2. The dilution
rate of the chamber is not precisely known in this case , but should be close to 4 percent/
hour. The increase in total nitrogen occurring during the first hour is thought to be
due to the lag time associated with the colorimetric analyzer. The instrumentation for
27
-------
0 1 2
IRRADIATION TIME', hours
Figure 18. Nitrogen balance during irradiation
of propylene-NOx.
PAN and nitric acid yields concentration
data in real time, while the NO£ values
at any given instant actually represent
concentrations some 10 minutes earlier.
Thus, where NC>2 is rapidly being con-
verted to other nitrogen products, the
sum of nitrogen species will be too high
since the gain in products has been
registered ahead of the loss of reactants.
After most of the NC^ is converted, the
total nitrogen curve appears to parallel
a typical dilution rate.
Figure 19 shows similar types
of nitrogen analyses where synthetic
auto exhaust is irradiated. In this case
where NO is converted to NC>2 early
in the irradiation, the excess in the
total nitrogen curve is less severe than
in the previous experiment because
the error due to lag time tends to be
self-cancelling when both product and reactant are read by the same analyzer. The reason
for the increase in total calculated nitrogen occurring near the end of the irradiation is not
apparent. Quite likely, it is due to slight inaccuracies in either the PAN and/or HNC>3
calibrations.
After 90 minutes of irradiation, the O^/NO- ratio is >2 and at 3 hours it is >4.
If during this time the colorimetric measurement for NO£ was low by a factoi of 2 due to
63, interference, as suggested by the results in Table 6, it is obvious that the true
account of total nitrogen would be seriously out of balance. Thus, on the basis of the
smog chamber data presented here where NC>2 was measured by the Saltzman method and where
a good material balance in nitrogen is indicated,, we conclude that any interference to
the Saltzman method by Oj is not likely substantive. Furthermore, it is recommended that
chemiluminescence analyzers not be used for NO£ determinations in smog chamber studies
until converters have been developed, that are specific in reducing NOo to NO.
Discussion
Dr. A. Winer (University of California, Riverside) reported on recent Statewide Air
Pollution Research Center (SAPRC) findings regarding the response of commercial chemi-
luminescent NOX analyzers to nitrogeneous smog constituents. Specifically, SAPRC inves-
tigated instrument response to PAN, ethyl nitrate, ethyl nitrite, and nitroethane for a
commercial NOX analyzer equipped with a molybdenum converter and operated in the "NO "
and "NO, NO2" modes, and response to PAN and n-propylnitrate for an instrument equipped
28
-------
1.00
0.90 —.
i
SYNTHETIC AUTO |
EXHAUST 16 ppm C
H —1
HONOoAVG
with a carbon converter. Within experi-
mental uncertainty, an essentially 100
percent response—that is, a response
equal to that by NO—was observed in all
cases except for the thermodynamically
more stable nitroethane, which gave a
6 to 7 percent response. A response to
nitric acid was observed but was not char-
acterized quantitatively. Although the
problem of nitrogen-containing compounds
being reduced on the converters of commer-
cial chemiluminescent NO-NOX instruments
has been known to most smog chamber
groups for some time, the vast majority
of users who are involved in ambient air
measurement applications are unaware
that these instruments are not specific
for NO. For this reason, SAPRC has given
O
a more detailed report of its studies.
Dr. C. W. Spicer (Battelle) offered
the observation that the Battelle smog
chamber has been demonstrated to yield
good nitrogen balances based on PAN and HNO3 in the gas phase (not on the walls) . He
believes that HNO^ is formed at the surface and then desorbs into the gas phase to a degree
depending on the nature of the surface.
IN SITU MEASUREMENTS - P. Harm, EPA
Optical measurements have been made in smog chambers for twenty years. The chambers
used for these measurements have generally been in the form of a long tube surrounded by
ultraviolet lights. Such a configuration has the advantage of a high ratio of optical
path to chamber volume but the disadvantage of a high ratio of surface to volume. In
spite of this disadvantage, many of the fundamental discoveries of atmospheric photo-
chemistry have been made in such chambers.
The first application of long path infrared spectroscopy to the study of smog
chemistry was carried out at the Franklin Institute Laboratories in Philadelphia between
1953 and 1956. That program of study established an analytical technique that has been
highly productive. Among the contributions from that first program were the discovery
of the peroxyadyl nitrate family of pollutants, proof of ozone formation in smog, in-
sights into the reaction mechanism that causes ozone to accumulate in smog, demonstration
of the formation of aldehydes in the course of the oxidation, and establishment of mechan-
9-12
isms to explain the varying degrees of reactivity among the hydrocarbon pollutants.
1 2 3
IRRADIATION TIME, Hours
Figure 19. Nitrogen balance during irradiation
of synthetic auto exhaust.
29
-------
The analytical method has not been as fully used since that time as it deserved, but at
present its application is being expanded.
For detection of molecules, six portions of the electromagnetic spectrum are gen-
erally recognized: ultraviolet, visible, near infrared, fundamental infrared, far in-
frared, and microwave. Molecules have both emission and absorption spectra in all these
regions, but in smog chamber studies only absorption spectra have had wide use.
The ultraviolet is the region of absorption by electronic transitions in molecules.
Vibratlonal and rotational transitions also occur and can produce a characteristic
structure in the bands. Although the ultraviolet band systems of different molecules
are likely to overlap, the structure sometimes permits distinguishing one molecule in
the presence of others. Pollutant species that are measurable by ultraviolet absorption
include sulfur dioxide, nitric oxide, ammonia, and ozone. At atmospheric pressure, how-
ever , many pollutant molecules do not show structure in their ultraviolet bands.
The visible region of the spectrum has very few molecular absorption bands. If
this were not the case, the sunlight would not penetrate the atmosphere. Nitrogen dioxide
is the only pollutant of significance that absorbs in the visible.
The near infrared, 0.70 to 2.50 microns in wavelength, is the spectral region where
the overtones of the molecular vibration-rotation bands appear. These overtones are
about 100 times weaker than the fundamentals; thus, the region is not generally useful
for molecular detection and analysis. A notable exception is the study of the atmos-
pheres of the planets, where large concentrations of absorbing gas are viewed over
extremely long paths. The overtone region has been the principal source of information
on the composition of the atmospheres of Mars, Venus, and Jupiter.
The fundamental infrared, 2.50 to 25 microns in wavelength, is the spectral region
that has been widely used for chemical analysis for many years. In this region, nearly
every air pollutant has a characteristic absorption band. Generally, these bands show
many variations in shape, location, and intensity. This spectral region is the one that
has been found the most useful for in situ analysis in smog chambers.
The far infrared shows rotational lines as well as some vibration-rotation bands.
Unfortunately, the very intense absorption by water vapor will blank out the far infrared
in smog chamber studies, even under the driest of conditions.
The microwave region of the spectrum has no application in smog chamber studies
because at atmospheric pressure the microwave absorption lines are so broad that dis-
tinguishing between different types of molecules is impractical.
It would be desirable to analyze real atmospheres and smog chamber atmospheres by
examining both the ultraviolet and infrared spectra. Unfortunately, the two spectral
regions require different kinds of spectrometers and detectors. Ultraviolet absorption
bands are usually stronger than infrared bands, so that a shorter path suffices, but the
number of molecules that can be measured by ultraviolet is limited. If only one optical
30
-------
system is to be used, the infrared is preferred in spite of the requirement for a long
path. It has been found practical to achieve the long path by folding the infrared
13 14
beam between mirrors. '
Among the institutions that are using in situ infrared measurements are the General
Motors Laboratories, the Ohio State University, the University of California, Riverside,
the University of North Carolina, and the EPA's National Environmental Research Center,
Research Triangle Park, N. C. In Table 7 there are listed some of the pollutants observed
in recent work at the National Environmental Research Center.
In the past, the instruments for measuring infrared spectra in smog chambers have
exclusively been dispersive spectrometers employing prisms and gratings. Spectra were
recorded periodically across a single long path in the chamber, and changes in the com-
position of the air were noted by visually comparing a spectrum to an earlier spectrum.
The sensitivity of detection in such work has been much increased in recent years by the
use of nitrogen-cooled photodetectors. Replacing a thermocouple by a mercury cadmium
telluride detector, for example, can raise the signal-to-noise level in a spectrum by a
factor of 10 or even 100. Dispersive spectrometry is, therefore, still an effective
mode of operation. It is being used at the National Environmental Research Center as
well as in other laboratories.
The most notable new instrument for application to in situ smog chamber studies is
15
the Fourier transform spectrometer. In this instrument a scanning interferometer modu-
lates the infrared energy before transmission through the long path cell, and a digital
computer transforms the resulting interferogram to a spectrum. The on-line computer is
also used to calculate and plot ratios of successive spectra. Such an instrument is
obviously much more convenient to work with than a dispersive spectrometer that does not
have computational capabilities. However, cost considerations remain as an equalizing
factor. A high performance dispersive spectroscopy system can be assembled for a price
much lower than the price of a Fourier transform spectrometer.
Sample spectra obtained with Fourier transform spectrometer system are shown in
Figures 20 through 25. These are representative of the large number of spectra recorded
in the past 2 years at the National Environmental Research Center. The reactants were
0.1 part per million (ppm) molecular chlorine and 5 ppm 1,1 dichloroethylene in 1 atmos-
phere of tank air. They were contained in a Pyrex long path cell 1 foot in diameter and
30 feet long. Fifty-six passes of infrared radiation through the tube gave a total
absorption path of 500 meters. The reactants were irradiated with the light of 96
black-light 40-watt fluorescent lamps. According to the rate of photolysis of nitrogen
dioxide, the lamps gave an ultraviolet light intensity within the chamber about 50 percent
higher than the intensity of normal sunlight. Spectra were recorded at a resolution of
1 reciprocal centimeter (cm"1) . The resultant large number of data points in the funda-
mental infrared region cannot conveniently be displayed in a single figure.
31
-------
Table 7. INFRARED DETECTION OF POLLUTANTS
- -
Pollutant
Acetylene
Ammonia
Carbon monoxide
Carbonyl fluoride
Chi pro acetyl
chloride
Difluoro dichloro
methane
Difluoro monochloro
methane (Freon 22)
Ethylene
Formic acid
Formaldehyde
Hydrogen chloride
Hydrogen peroxide
Ketene
Methane
^^^^^^^^^^^^^^^- •^^•fc» Mi.1
Frequency of
most useful band,
cm"'
735
965
2180
960
720
921
mo
950
1105
2765
2820
1250
2150
3020
Where seen recently
Seen in ambient air at concentra-
tions ranging between 0.10 and
0.001 ppm.
Seen in the air at Houston, Texas,
at concentrations up to 0.050 ppm.
Seen as product of NO -hydrocarbon
reactions and as product of chlorine
attack on organic compounds.
Seen as product of the attack of
chlorine atoms on difluoro mono-
chloro methane in air.
Seen as product of the attack of
chlorine atoms on 1,1 dichloro-
ethylene.
Seen in air at the background level
of 0.0001 ppm, using a cryogenic
concentration step.
Seen in air at Research Triangle
Park at 0.005 ppm, using a cryogenic
concentration step.
Studied in laboratory and measured
in ambient air.
Observed in laboratory as product
of hydrocarbon phctooxidation.
Also seen as product ^f formyl
chloride hydrolysis. Measured in
California smog at concentrations
up to 0.070 ppm.
Observed in laboratory as product
of hydrocarbon photooxidation.
Also seen in ambient air polluted
by ;auto exhaust in Houston, Texas,
and Raleigh, N. C.
Observed in laboratory as product
of chlorine attack on organic
materials.
Seen in the laboratory chamber as
product of chlorine attack on for-
maldehyde in air.
Seen in laboratory reactions as pro-
duct of ozone attack on propylene
and 2-butene.
Measured in clean and polluted
atmospheres.
32
-------
Table 7 (continued). INFRARED DETECTION OF POLLUTANTS
Pollutant
Frequency of
most useful band,
cnH
Where seen recently
Methanol
Monofluoro trlchloro
methane (Freon)
Nitric acid
Nitric oxide
Nitrogen dioxide
Nitrogen pentoxide
Nitrous acid
Ozone
Peroxy acetyl
nitrate
Peroxy benzoyl
nitrate
Phosgene
Propylene
Sulfur dioxide
Vinyl chloride
1035
845 ,
880
1900
1615
(2900 in humidity)
1250
850
1060
1160
990
815
915
1360
920
Seen Jn air at Pasadena, Calif., at
coriceritrati ons up. to 0.100 ppm.
Seen in air at the background level
of 0.0001 ppm, using a cryogenic
concentration step.
Seen as product of N20s hydrolysis
in laboratory chambers.
Measured in ambient air.
Routinely monitored in laboratory
chamber studies.
Seen as product of ozone-nitrogen
dioxide reaction in laboratory chamber.
Seen in auto exhaust and in chambers
as product of the H20-NO-N02
interaction.
Measured in laboratory chamber as
product of hydrocarbon-N02 reaction.
Seen in ambient air.
Seen in laboratory photooxidations.
Also measured in California smog at
concentrations up to 0.050 ppm.
Seen as product of photooxidation of
styrene in laboratory chamber.
Seen as product of oxidation of
chlorinated hydrocarbons in the
laboratory chamber. '
Followed in laboratory reactions;
seen in air polluted by auto exhaust.
Measured in auto exhaust.
Seen in the air at Houston, Texas,
at a concentration of 0.37 ppm.
Figure 20 shows the spectral region between 2700 and 3100 cm in single beam mode,
before irradiation. The detailed structure in the spectrum is mainly due to an impurity
of approximately 2 ppm of methane in the tank air. There are also absorption lines of
water. In this region the absorption by 1,1 dichloroethylene is insignificant. Chlorine,
being a homonuclear diatomic molecule with no permanent dipole moment, shows no infrared
absorption. Actually, chlorine is about the only possible air pollutant that does not
have an infrared spectrum. Figure 21 shows the same spectral region as Figure 20 after
15 minutes of irradiation. Bands of the products hydrogen chloride, formaldehyde and
chloro acetyl chloride have appeared. Figure 22 is a ratio plot of the spectrum of
Figure 20 over the spectrum of Figure 21. Figure 23 shows a ratio plot of part of the
spectrum of the initial reactants over the spectrum of tank air. The absorption in this
33
-------
I
BC
2700 cm'1
2800
2900
3000
3100
Figure 20. Single beam plot in CH absorption region for 1 atmosphere of tank air with addition of 5 ppm
1,1 dichlorethylene and 0.1 ppm molecular chlorine, before irradiation. Path length, 500 meters;
resolution, 1 cm"^.
spectrum is all due to the 1,1 dichloroethylene. Figure 24 shows a similar r^i'o plot,
chamber contents over the tank air, after 10 minutes of irradiation. This figure shows
that more than 80 percent of the 1,1 dichloroethylene was consumed and converted mainly
to chloro acetyl chloride. This is an impressive result, indicating a chain reaction,
as will be discussed in a separate publication. Figure 24 also shows a band due to
phosgene. Figure 25 is a partial ratio plot of the spectrum of the reaction products
after 10 minutes over the spectrum of the starting material. The 1,1 dichloroethylene
band at 1620 cm was removed by the reaction and, therefore, appears upside down and
cut off at the top. The band at 1830 cm is the carbonyl band of the product chloro
acetyl chloride.
These sample spectra have shown the capability of in situ spectroscopy to reveal a
variety of reaction products with unequivocal identification and with a high degree of
detection sensitivity. It should also be noted that successive recordings of the spectrum
yield time versus concentration plots for both reactants and products. It is clear,
therefore, that the in situ spectroscopic technique that was first productive twenty
years ago is still highly productive and deserves to have its application expanded.
34
-------
<
I
H
2700 cm
2800
2900
3000
3100
Figure 21. Single beam plot in CH absorption region for 1 atmosphere of tank air with addition of
5 ppm 1,1 dichlorethylene and 0.1 pom molecular chlorine, after 15 minutes of irradiation. Path
length, 500 meters; resolution, 1 cm"1.
DISCUSSION
Dr. Winer (UCR, SAPRC) reported on the SAPRC experiments with an in situ multipass
infrared system attached to a smog chamber. Dr. Winer's report was as follows:
The evacuable chamber at SAPRC contains an insitu multipass optical system with a
1.3-meter base length capable of producing pathlengths of 100 meters or more for use
with either a 1-meter scanning monochromator or a Fourier infrared interferometer.
During the past year, a variety of studies have been conducted in the evacuable chamber
with this long-path infrared system (LPIR) including both photolysis and dark experi-
ments. For example, the good time resolution (as well as spectral resolution) afforded
by the interferometer has made it possible to obtain concentration-time data from the
in situ measurement of a large number of species resulting from the photolysis of HC/NOX
systems, including some species not easily detected by other means, i.e., HONO, HNO3,
C-H,-ONO_ and HCOOH. A second application of the Fourier interferometer LPIR system was
a detailed product analysis for the reaction of ozone with ethene, propene and cis-2-butene
in the evacuable chamber at both one atmosphere and reduced pressures. In addition
35
-------
K
Ul
u
CO
<
EC
CHLORO
ACETVL
CHLORIDE
0.5 -
2700 cm'1
2800
2900
3000
Figure 22. Ratio plot, spectrum of products over spectrum of reactants; 5 ppm 1,1 dichlorethylene
and 0.1 ppm molecular chlorine in air, after 15 minutes of irradiation. Path length, 500 meters;
resolution, 1 cm"'.
to species identified in previous studies, preliminary assignments have been made
to several orcarbonly hydroperoxides and peroxyformic acid. A third recent applica-
tion was the use of the dispersive LPIR system to investigate the stoichiometi y of the
2 percent neutral buffered potassium iodide (2 percent NBKI) method employed by the
California Air Resources Board (ARE) for the measurement of ozone, following disclosure
of a discrepancy between this method and the 2 percent unbuffered KI method employed by
the Los Angeles Air Pollution Control District (LAAPCD). Using the 5500-liter evacuable
chamber as a convenient reservoir for ozone and with a 70-meter pathlength and a liquid
nitrogen cooled HgCdTe detector, quantitative infrared data were obtained simultaneously
with 2 percent NBKI impinger samples (ARB method) for ozone concentrations ranging from
0.1 to 1.2 ppm. Comparison of the resulting KI and IR ozone concentrations established,
within uncertainties of 3 to 5 percent, a 1:1 stoichiometry for the reaction of ozone
with iodide in a 2 percent buffered solution. In contrast, we found that the 2 percent
unbuffered KI (LAAPCD) method yields values of ozone systematically lower by more than
30 percent. (A complete report of this work has been submitted for publication in
Environmental Science and Technology. )
In summary, we have found the in situ long-path infrared capability to be invalu-
able in a variety of our environmental chamber studies, largely because of its capacity
36
-------
4
tc
6
cc
700 cm
1100
1300
Figure 23. Ratio plot, spectrum of reactants over spectrum of tank air; 5 ppm 1,1 dichlorethylene
and 0.1 ppm molecular chlorine in air, before irradiation. Path length, 500 meters; resolution, 1 cm-1.
to validate, by in situ measurement, the data being obtained by external sampling and
determination methods for species such as PAN, HCHO; 03, and organic nitrates.
AEROSOL MEASUREMENT - P. Reist, UNC
Small particles can be formed from reactions involving two gases, a gas and water
vapor, a gas and energy in the form of radiation, or a combination of these factors.
They can also arise through the condensation of a gas onto its own molecules or onto a
particle or molecules of some other material. Particles can be either solid or liquid.
They range in size from about 0.005 micrometer for fresh reaction products up to about
1 micrometer in diameter for droplets formed by condensation.
Particle concentrations of 106 particles per milliliter or greater are not uncommon
in some chamber studies, and concentrations may be as low as several hundred per milli-
liter. Mass concentrations can be as high as several hundred micrograms of aerosol per
cubic meter. When number concentrations are greatly in excess of 106 particles/ml,
37
-------
UJ
u
I
t/l
CHLOROACETYL CHLORIDE
700 cm"
1100
1300
Figure 24. Ratio plot, spectrum of products over spectrum of tank air; 5 ppm 1,1 dichlorethylene
and 0.1 ppm molecular chlorine in air, after 1.0 minutes irradiation. Path length, 500 meters;
resolution, 1 cm-1.
particle size and concentration will change rapidly through the process of coagulation,
particles growing to larger sizes with a resulting decrease in aerosol concentration.
Particles can also be lost to the walls of a chamber, and such losses can be particularly
important with long experiments or small chambers.
Particles act as sites for condensation of water vapor, producing clouds or haze,
and can also in themselves significantly alter visibility. By providing increased surface
area per unit mass of material, small particles can enhance reactions between atmospheric
gases. They can be transported great distances by air currents, and represent a potenti-
ally significant inhalation hazard since they can penetrate deep into the lower portions of
the lung where lung clearance mechanisms are least effective.
Particle concentration, size, and chemical composition interplay to determine whether
aerosol particles produce a dramatic effect or remain unnoticed. Toxic materials in very
38
-------
o
ff
IU
u
K
1300 cm
1500
1700
1900
Figure 25. Ratio plot, spectrum of products over spectrum of reactants; 5 ppm 1,1 dichlorethylene
and 0.1 ppm molecular chlorine in air, after 10 minutes irradiation. Path length, 500 meters;
resolution, 4cm-1.
low number and mass concentration are benign. Large numbers of very small particles
produce little interference with visibility; fewer numbers of larger particles have a much
more pronounced effect.
There are gaps in determining the proper aerosol measurement techniques for
chamber studies. The tradeoff is between ease of data collection and accuracy or complete-
ness of data. Condensation nuclei counts provide only the barest of information about the
aerosol in a chamber, the number concentration, but provide it in a continuous manner so
that aerosol trends can at least be observed. Intergrating nepholometers also fall into this
category. Since these two devices give indications of number and mass concentrations,
respectively, they represent the minimum aerosol instrumentation that should be included
in chamber studies.
The electric mobility analyzer has been used frequently to determine atmospheric
particle size distribution data. It has the distinct advantage of speed over more accurate
methods; a particle size distribution can be determined in 10 minutes. This permits the
dynamic properties of the aerosol to be followed during the course of a chamber experiment.
39
-------
Single particle optical counters have been used in some previous chamber studies
but appear to be of much less value than the instruments listed above. Single particle
optical counters have a limited effective size range and a very low effective concentration
range. The data from them, although easy to collect, is difficult to interpret. Similar to
the mobility analyzer, single particle optical counters are useful in showing trends in
aerosol size, particularly in the size range approaching 1 micrometer in diameter.
Filters are useful for total mass samples or for chemical analysis of the total aerosol.
Since mass concentrations are relatively low (10 to 100 micrograms per cubic meter)
sampling times must be long enough for a sufficient amount of sample to be collected. Glass
fiber filters are generally used for sampling for chemical analysis since they are considered
to be the least reactive, and membrane filters are used for total mass concentration measure-
ments because of their high collection efficiencies, even for very small particles.
Impactors will continue to be used for chamber studies since they provide a means
of getting chemical composition data on various particle size fractions. The data from these
samplers is at best crude, and they are severely limited in the smallest sizes of particles
that can be collected. Nevertheless, they do provide useful information that at present
would otherwise be impossible to get.
The other techniques mentioned as possibilities for chamber studies, although proved
out in the laboratory, have not been applied as working tools for chamber experiments. As
becomes apparent with the application of any instrument to chamber work, there is a large
gap separating the successful performance of an instrument in the laboratory and its useful
application in the field. Future research and development should provide the chamber
scientist with improved aerosol instrumentation; at present, existing equipment must be
used, recognizing the limitations of the equipment while appreciating the value of the data
developed.
A listing of aerosol sampling procedures and characteristics of selected aerosol
instruments is given in Tables 8 and 9.
Discussion
Referring to Dr. Reist's presentation, Dr. G. J. Doyle (UCR, SAPRC) commented that,
in general, the optical particle counter is indeed inextricably complicated. However, for
spherical particles, which are likely to be frequently encountered in smog chamber experi-
ments, the response to particle size dependence can and often has been calculated. The
chief drawback is that one must know the refractive index, both the real and imaginary
parts. Dr. Doyle further stated that recent counter designs are directed towards minimiz-
ing sensitivity to refractive index so that the data obtained should be at least as good as the
data from the Whitby analyzer within the appropriate size region.
MEASUREMENT OF OXYGENATED SULFUR COMPOUNDS - D. Miller, Battelle
The problem concerning the measurement of SO_ and its oxidation products in smog
chamber experiments is exemplified in Figure 26. Under typical smog chamber conditions,
40
-------
Table 8. AEROSOL SAMPLING PROCEDURES
I. Aerosols studied in the suspended state
1. Techniques applicable to particles greater than 0.1 micrometer diameter
A. Nephelometry
B. Single particle optical counter
2. Techniques applicable to particles less than 0,1 micrometer diameter
A. Mobility analyzer
B. Laser doppler spectroscopy
C. Condensation nuclei counter
D. Vapor phase enlargement
E. Particle size magnifier
F. Flame photometry
II. Particles removed before studying
1. Particles examined
A. Impaction
B. Centrifugation
C. Filtration
D. Electrostatic precipitation
E. Thermal precipitation
2. Effluent examined
A. Diffusion
B. Impaction
the change in SC>2 concentration due to oxidation is often quite small, and sometimes smaller
than other SC>2 loss processes, such as dilution and wall losses. If one attempts to conduct
experiments at ambient conditions, say <0.1 ppm SO2, and the oxidation rate is 2 percent/
hour, one is faced with trying to measure <2-ppb/hour change in SC>2 concentration. Such
a requirement exceeds the sensitivity of commercially available SO2 analyzers. At the
present, we are having to study SC>2 oxidation at higher than ambient concentrations
(^.5 ppm), and the instrumentation is still required to measure changes in concentration
of <2 percent; a determination of questionable accuracy.
The measurement of SO2 oxidation products is not much easier. Under many experi-
mental situations, the H2SO4 aerosol concentration may not exceed 10 ppb, even after 4 or
5 hours of irradiation. In this case, at least, one is trying to measure only a small concen-
tration rather than a small change in a relatively large concentration.
My objective in this discussion is to review and compare some of the analytical
methods we are employing in our SC>2 studies. The number of methods I will discuss is
quite limited, and it is hoped that other participants of the conference will discuss other
techniques that they have found to be successful.
41
-------
Table 9. CHARACTERISTICS OF SELECTED SAMPLERS
Type
TSI mobility
analyzer
MRI integrating
nephelometer
Royco single
particle optical
counter
Environment One
condensation
nuclei counter
Lundgren
impactor
Aerosol
centrifuge
Point-plane
electrostatic
precipitator
Sampling rate,
1 i ters/mi n
30-50
140
30
3
113
20*
0.2
Sampling
time
10 minutes
Continuous
Variable
Continuous
(5-sec
response)
Variable
Variable
Variable
Type of sample
collected
Size distribution from
electrical mobility
Light scattering
coefficient
Size distribution from
light scattering
Particle number
concentration only
Four stages plus
filter, ECDA, 4th
stage =0.4
Continuous aerodynamic
size separation
Random fraction of
aerosol
Sample
analysis
Internal
Internal
Internal
Internal
External
External
External
Remarks
Direct reading
Direct reading,
mass concentration
implied
Direct reading
Direct reading
Gravimetric or
chemical analysis
Gravimetric or
chemical analysis
Size distribution
by electron
microscopy
Variable.
-------
300
IRRADIATION TIME, hours
Figure 26. Sulfur dioxide removal during irradiation of propylene-NOx.
Sulfur Dioxide
The continuous SO2 monitors commonly used in our work are hydrogen flame photo-
meters and coulometric-titration analyzers. The coulometric analyzers operate on the
principle of oxidizing SO? with iodine or bromine. Iodine is the charge carrier, and any
change in iodine concentration due to reduction by SC>2 is detected as a flow of current
through a reference electrode . The instruments are generally equipped with scrubbers
for removing oxidants , which cause negative interferences . The sensitivity of these
analyzers is about 10 ppb .
Flame photometric analyzers are based on the chemiluminescence principle. When
sulfur-bearing gases are burned in a hydrogen flame, excited sulfur molecules (S + S
which emit light in returning to the ground state, are formed. The emission is generally
read at 394 nm. Such instruments have a wide range of linearity, with the response being
proportional to the square of the sulfur concentration . Gas chromatographic techniques
can be incorporated to permit separation of various sulfur gases. In most smog chamber
applications , the instruments can be used to specifically monitor SO2 by merely placing
a particulate filter in the sample line . The filter removes the aerosol products of 803
oxidation. The sensitivity of the flame photometers is also near 10 ppb.
43
-------
In Figure 27 simultaneous measure-
ments of SC>2 are shown for coulometric and
flame photometric instruments. Figure
27a represents results of irradiating
0.5 ppm SC>2 in unpolluted air- The
change in SC>2 concentration with time
is largely due to chamber dilution.
The important observation here is the
nearly identical measurement of SC>2
by the two different instruments.
Figure 27b shows SC>2 loss where
1. 3 ppm propylene and 0.5 ppm NOX
were irradiated with 0. 5 ppm SC>2 •
In this case, the agreement between
the two SC>2 monitors was good until
about the last hour of irradiation. This
pattern during the latter period of irradia-
tions with olefins is quite consistent;
the flame photometer indicating a higher
concentration than the coulometric analyzer.
Figure 27c shows greater disparity between
the two methods. In this case, 1-butene
was irradiated with NOX at a HC/NOX
ratio of about 4 in a dynamic 200-liter
smog chamber.
<
ec
u
CM
0.50
0.25
*•••••*•••
T
FLAME PHOTOMETER (S02)
COULOMETRIC TITRATION
(S02)
C. DYNAMIC
1-BUTENE-4ppm
NO,
-1 ppm 4 ppm
1
I
0123
IRRADIATION TIME, hours
Figure 27. Comparison of methods for sulfur dioxide
analyses during smog-chamber irradiations.
In general, we find that the difference
in SC>2 concentrations determined by
these two methods increases as the photo-
chemical reactivity of the system increases .
The disagreement is most pronounced
after the maximum 03 concentration
is reached. It is not clear which analyzer
might be suffering interference problems at this point in the irradiation, nor what the inter-
fering specie(s) may be. We do know that in experiments where a considerable amount
of nonsulfate aerosol is produced, the coulometric analyzer suffers a positive-type interfer-
ence. Also, during relatively fast photochemical reactions, we have noticed anomalous
behavior of the coulometric SO2 analyzer, often showing an increase in the SO2 concentra-
tion during the early stages of irradiation. In conclusion, it seems that the flame photometric
technique is the more reliable method for smog chamber work except in instances where
one is confident that interference is not occurring with coulometric analyzers.
44
-------
Sulf ate Aerosols
We are currently using two methods for determining sulfate concentrations . The
barium chloranilate method16-17 is used for* interim ttant analyses during the course of
the irradiations, and the barium perchlorate method1** is used on a much larger aerosol
collection made at the conclusion of irradiation .
For intermittant SO- analyses, about 60 liters of sample are withdrawn through
25-mm glass fiber filters that were previously washed by refluxing with water and methanol.
The aerosol is extracted from the filters with 80 percent isopropanol. Excess barium
chloranilate is added , followed by thorough mixing and centrifuging . The sulfate ions are
precipitated with barium, and the acid chloranilate ion yields a pink color read at 310
nm . The reaction is
H+
The sensitivity of the method is about 2 /ug H2SO4-
Aerosols analyzed by the barium perchlorate method are collected on 10-cm quartz
fiber filters. In this analysis, barium perchlorate is added from a microburet to the sulfate
solution. Barium sulfate percipitates , and as soon as excess barium ions appear, a complex
forms with thorin indicater. The titration is also carried out in 80 percent isopropanol to
minimize the solubility of barium sulfate . The end point of the titration is marked by a
yellow-to-pink color change.
A third method I want to mention is a technique applied to determine the dissociated
proton concentration of the aerosol extract, presumably protons of sulfuric acid. In this
case, aerosol is collected and extracted in the same manner as with the barium chloranilate
method. One drop of neutral iodate-iodide solution is added to the extract, which is then
centrifuged, and the liberated iodine is read at 352 nm. The intensity of the iodine solution
is directly proportional to the proton concentration. The reaction occurring is
6H30+ + 103-+ 5r— ~3l2 + 9H20
The sensitivity of the method is about 5 y g H2SO4 and could probably be improved with the
addition of starch indicator.
A similar proton-determining method has recently been reported in which a bromate-
19
bromide mixture is used to liberate bromine by protons. Subsequent bromination of
fluorescein solution yields a distinctive color .
In smog chamber experiments conducted earlier this year at Battelle, analyses for
sulfuric acid aerosol were limited to the barium chloranilate method . In many cases , the
sulfate concentration agreed fairly well with the concentration expected from the consump-
tion of SO2. However, under some conditions, the agreement was not good. Thus, in more
45
-------
recent experiments we have employed all three of the methods described above. It is
regretable, however, that at this time our analytical work has not been completed, and I
am not in a position to comment on the comparability of the methods, nor the significance
of the analytical methods.
Discussion
In response to a question from Mr. R. Stevens (EPA), Dr. C. W. Spicer (Battelle)
explained that HNC>3 interference is avoided in the H^SO. filter analysis since HNC>3 passes
quantitatively through Teflon and neutral quartz filters. HNOj can be scrubbed by glass
and several other types of filters.
QUALITY CONTROL - C.E. Feigley, UNC
Two major problems in controlling data quality are (1) establishing reliable calibra-
tion techniques and (2) placing some quantitative measure of quality on the resulting data.
Calibration
Chemiluminescent NOX and Oj—At the University of North Carolina, we have found that the
use of multiple calibration methods was necessary to assure correct calibration. Figure 28
illustrates the calibration techniques employed. Bottled NO has been a stable source for
calibration, but we have had to use tanks from two different supplies. Differences between
tanks we have received have beeik as much as 5 percent, not including occasional mislabel-
ing. Bottled NC>2 has not proven to be as reliable as bottled NO; it sometimes required
running gas out of the bottle for Ej or 6 hours before reaching a stable concentration that
i
agreed with other techniques. Also, when a stainless capillary was used as a flow restric-
tor, some NO appeared in the gas. Use of Teflon instead reduced this problem. N©2 per-
meation tubes using a Warburg cqmpensation syringe gave high readings compared with
NO TANKS
N02 TANK
! SALTZMAN
NO
NO INSTRUMENT
N02
NO-03'TITRATION
03 INSTRUMENT
CALIBRATED 03
GENERATOR
NBKI
Figure 28. Calibration methods used for
chemiluminescent NOX and 63 instruments.
46
-------
other techniques, and thus their use was discontinued. No great effort has been made to
determine why the NO2 permeation tubes failed to agree with other methods. Saltzman has
occasionally been used to calibrate the NOX instrument, usually when there is difficulty
obtaining agreement between other methods.
The ozone instrument is primarily calibrated with NBKI bubblers. A detailed statis-
tical analysis of ozone instrument calibration identified the major sources of error. In
the preparation of the standard \2 curve; the major sources of error were the titration and
the subsequent two dilutions of standard ^ solutions, not the serial dilutions on measure-
ment of absorbances. Thus, we at UNC recommend a standard I2 curve resulting from six
separate titration and dilution series. The slope of the line is determined by least-square
regression with points weighted by the inverse of the absorbances. The major errors in
NBKI bubbler determinations are in measuring volumes, times, and flowrates. The bub-
bler technique is very accurate, but rather imprecise. Thus, five or six bubbler samples
should be run for each calibration performed.
Methane and Nonmethane Hydrocarbons with an Environmental Chromatograph—For
measuring methane and nonmethane hydrocarbons, the UNC gropp's experience is mostly
with the environmental chromatograph. The Federal Register states that total hydrocarbon
and methane modes of an environmental chromatograph shall be calibrated with methane.
In the first calibration of our instrument, bottled methane from two sources was compared
with a propane permeation tube. The permeation tube gave low results. This was at first
thought to be an error in calculating permeation rates. After checking the instrument on
two tanks of n-butane, two of n-hexane, one of toluene, and one of mixed hydrocarbons
(from 4 to 12 ppmC) , it was concluded that the instrument did not respond 100 percent to
nonmethane hydrocarbons and that the permeation tube values were in good agreement with
bottled gas.
It was felt that the efficiency of measurement of NMHC should be calculated using
known concentrations of the hydrocarbons to be measured. Data collected after this are
then corrected for this efficiency.
Recommendation—The calibration of instrumental methods for smog chamber research
still raises many questions. The problem of NMHC efficiency is still far from resolved. The
effect of frequency of calibration is not known quantitatively. There is no standardization
of procedures.
Since good calibration is an indispensible prerequisite to data accuracy, I would like
to recommend that EPA sponsor a committee, drawing on the experience and talents of many
of the groups represented at this meeting, to establish standard calibration procedures.
Applications of such procedures would help assure uniformly high data quality. .
Statistical Approach
Recently, questions have been raised concerning the quality of data from various
third generation air monitoring instruments. For instance, modelers of atmospheric reactions
47
-------
often express the concern that measurement errors may be of such a magnitude as to have
adverse effects on the validation of their models.
Accuracy and Precision—Initial efforts have been made to quantitate data quality by specify-
ing statistically interpretable confidence intervals for measurements.2« When such confidence
intervals are placed on measurements, there can be little question as to the quality of data.
In addition, the analysis required to determine these intervals can point out the steps in the
calibration procedure to which data quality is most sensitive. Both accuracy and precision
can affect the width of the confidence interval.
Air monitoring instrument inaccuracy (bias) results from three factors: (1) uncer-
tainties in knowing the concentrations of gas used for calibration, (2) long-term time drift,
or (3) interferences. On the other hand, precision is an inherent instrument characteristic,
unaffected by these factors.
The variation of instrument readings due to imprecision can be estimated by letting an
instrument read a stable source for some time. The variation due to bias can be estimated
by analyzing the sources of errors, especially calibration.
Ozone Instrument Calibration—An analysis of Oj instrument calibration led to the following
formula for the confidence interval on the concentration measured by NBKI with a weighted
regression \2 standard curve:
[L,u] =yp±t(1.a/2){rv1).
var(yp) xa2
where
n-1 i=1
xj
y is the amount of ozone collected (pphm-liters) corresponding to an absorbance reading
x; xa is the mean of absorbances from k NBKI bubbler samples at some ozone generation
setting (absorbance units); var (y ) is the variance associated with absorbing ozone from
a constant source (pphm-liters)^; x^is an absorbance measurement from the standard 12
curve determination; yj is an equivalent ozone-absorbed value from standard curve determina-
tion corresponding to xj; n is the number of (xj, y^) points from the standard curve determina-
tion; t is Student's t statistic; ypj is the estimated equivalent ozone collected from the standard
12 curve corresponding to xj.
If simple linear regression has been used to obtain a standard 12 curve, the following
equation should be used in lieu of the preceding equation.
( r i v,
J var(yp)
[L,U] =yp±
where
(xa - x)2
n
£(xj-x)2
i=l
I n
Se2 = - 2
n-2 i-l
ypi}2
48
-------
and x is the mean of all the absorbances used to determine the standard I-
curve.
These equations reflect uncertainties in knowing the concentration of ozone to which
an instrument is calibrated. Thus, these equations can serve as a confidence interval on
measurements made with an ozone instrument at the concentration to which it was calibrated.
Table 10 shows 95 percent confidence intervals on measurements made at an ozone concentra-
tion of 8 pphm (the Federal air quality standard) after calibrating at that concentration.
The best case and worst case values are presented since var(y ) was not known exactly.
In the worst case, it equaled 6.75 (pphm-liters)2, and in the best case, 0.69 (pphm-liters)2.
The tightening of the confidence interval with increased replication of NBKI bubbler samples
. is obvious.
Table 10. CONFIDENCE INTERVAL ON BIAS IN MEASURING
OZONE AT THE FEDERAL AIR QUALITY STANDARD WITH
A CHEMILUMINESCENT INSTRUMENT CALIBRATED
BY REPLICATED NBKI MEASUREMENTS
Number of
replicates
1
2
4
6
8
Confidence intervals, pphm
Best case
8 + 1.4
8 + 1.1
8 + 0.83
8 + 0.73
8 + 0.68
Worst case
8 + 4.4
8 + 3.1
8 + 2.2
8 + 1.8
8 + 1.6
After calibration, the relative error due to bias is constant throughout the linear
range of the instrument. If an instrument is calibrated so that the output, Oc, corresponds
to the estimated concentration Cc, then the confidence interval on a new concentration
measurement with an output On is given by the expression
fUc Lc "I
,L] = — On, — On
LOc Oc J
[u,U = —
**,
where Uc and Lc define the confidence interval on Cc.
The effect of instrument noise on the confidence interval was neglected in this case
since the instrument's precision was quite good.
Recommendations—The manner in which precision and accuracy affect the statistical prop-
erties of the measurement is rather complex and more work is needed in this area before a
confidence interval can be specified for all instrumental measurements. Also, the problem
of measuring concentrations in the part-per-billion range.plunges measurement further into
the realm of statistics.
49
-------
The nature of this work is such that a group effort is not appropriate. I do suggest,
however, that EPA encourage individual researchers with interests in instrumentation and
statistics to pursue this problem.
Measures of Reactivity
In addition to assuring and quantitating data quality from smog chambers, we are con-
cerned that data from different smog chambers be comparable, or at least, that differences
between chambers be known explicity. The effects of chamber design have already been
discussed. Consideration of these effects, as well as routine data analysis, makes use of
measures of reactivity which require mathematical definition. Perhaps the most important
measure is that of light intensity.
Light Intensity—The pseudo-first-order rate constant of NC>2 disappearance, k^, has been
used in the past as a measure of light intensity. Even now there is some disagreement as to
just what is meant by k^:
*
1 (N02)i
= — In Reference 21
t (N02)f
din [N02l
Reference 22
dt
d In [N02]
t—O dt
References 23,24
There are several similar definitions. It has been pointed out by several investigators
that k^ is a function of irradiation time; and it is possiblly a function of initial NC>2, NO,
and C>2.
Thus, the measure that should be used is <(>ka, also called kp the true rate constant
of NO- photolysis. Several techniques have appeared in the literature in the past 2 years
for measuring kj, both in an inert gas and in air.
The following equation was presented by Holmes et al. for calculating kj^ from
photolysis in N2 at very low levels of 03, say less than 1000 ppm, and low initial NO. Rate
constants refer to reactions in Table 11.
1 / k?i k9f>\ FNO9lr, k?n / [NO2lo
2At
1 I +
kq/ [N02] kQ \ [N02]
/k20 k2 \ /[N02]0 - [N02)
[NO] +— [02] '
V k \e
\ Kg Kig
50
-------
Table 11. INORGANIC REACTIONS OF PHOTOCHEMICAL SMOG FORMATION
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
N02 + hv ->
0 + 02(+M)
03 + NO +
03 + N02
N02 + N03 ->
i N205
N205 + H20 ->•
N02 + N03 ->•
0 + N02
NO + N03 ->
NO + N02 + H20 -"
HN02 + hv •*
CO + OH- -»
H- + 02(+M) ->
H02- + NO ->
H02- + N02 ->
OH- + N02(+M) -»•
OH- + NO +
H02- + H02- •*•
0 + NO(4-M) •*
0 + N02(+M) ->
2ND + 02 •*•
2HN02 •*
H202 + hv •*
NO + HN03
HN02 + HN03 •*•
• •• - —
NO + 0
03(+M)
N02 + 02
N03 + 02
N205
N02 + N03
2HN03
NO + N02 + 02
NO + 02
2N02
2HN02
NO + OH-
C02 + H-
H02-(+M)
N02 + OH-
HN02 + 02
HN03(+M)
HONO
H202 + 02
N02(+M)
N03(+M)
2N02
NO + N02 + H20
20H
HN02 + N02
2N02 + H20
51
-------
Simulations at UNC by Joe Sickles suggest that the assumptions of this approach are
valid between 1 minute and 10 minutes after the start of irradiation.
The equation, presented by Stedman and Niki^
kd = 2K1 kg/(kg + k21>
makes the assumptions of photostationary state (PSS) for O, 03, NOg, and ^Og, and also
that no NO or 03 is present. These assumptions appear to be somewhat contradictory.
Sickles' simulation results indicate that PSS is not established until about 30 seconds after
the beginning of irradiation, but after this time some reactions of NO and Q£ cannot be
neglected.
The problems of k]_ determinations in N2 are that (1) it may be difficult or extremely
expensive to get such low O, concentrations in some chambers, or in bags, and (2) it is hard
£
to measure these low Q£ concentrations.
The following two equations were used by Stedman and Niki for determining k, in
air.
d[N02l d[NO] d[03]
= = = ki [N02]
dt dt dt
k3 [NO] [03]
[N02l
The upper equation is termed the direct measurement approach and is assumed to be
valid up to 30 seconds. Sickles' results indicate this period may be considerably shorter,
perhaps less than 5 seconds. An extremely fast instrument response is required for this
determination.
The lower equation assumes PSS for ozone. A major problem here, as pointed out by
24
Seinfeld et al., is that rather small changes in NO£ levels must be measured. Also,
errors in the concentrations of all three species in the equation contribute to errors in kj.
24
Seinfeld et al. suggest an alternative way of getting around this. Given a number of
NO or N©2 concentrations measured during the first minute or so, they solve their 11-reaction
mechanism for kj by quasilinearization. This requires computerized calculations for each
measurement .
An alternative that avoids many of the problems of the above approaches is described
in a soon-to-be-published paper by Sickles and Jeffries of UNC .
It was found that the mass-balance on NO£ in a continuous-flow stirred-tank reactor
could be solved analytically to yield the following equation for kj at steady-state:
52
-------
where T is reactor residence time and x is the fractional conversion of NO
L*
x=
[N02]i
n
Thus , one must know only the inlet and outlet NO2 concentrations . In addition , this approach
provides a continuous measure of kj whereas other techniques work on a batch principle.
Other Measures— Some other measures of reactivity, in addition to kj, were defined by the
26
CAPI-6 committee. Many of these definitions are not applicable to all smog chamber data
nor to the ambient air . Thus , they should be reevaluated so that all workers in the field
use measures that facilitate interchamber comparisons .
An example of an ambiguous definition is that of NC>2 formation rate-. CAPI-6 defines
NO2 formation rate as half the initial NO divided by the time to form an amount of NO2 equal
to one-half the initial NO . The broken line in Figure 29a shows such a rate on data from a
typical indoor chamber . The slope of the NO2 concentration line is quite close to the straight-
line slope at any instant. Examining the data from a hypothetical run in a hypothetical outdoor
smog chamber in Figure 29b, it will be noted that the instantaneous rate of NO2 formation
differs markedly from the straight-line slope . This is mainly a result of constantly increas-
ing light intensity in the morning . The length of the initial period with zero light intensity
is a function of injection time and sunrise — not a characteristic of system reactivity as we
would like.
One means of avoiding this is to consider the maximum instantaneous rate instead of
some average rate . The maximum rate is uniquely defined from nearly all NO2 curves . It
may be calculated by hand, or a computer can easily calculate and output the maximum rates.
Recommendations — This is one example , but it seems that the area of defining measures of
reactivity needs more thought to assure the applicability of these measures to all situations.
Smog chamber technology and applications have changed considerably since CAPI-6. I
would like to recommend that EPA sponsor a group whose task would be to arrive at a new
standard set of reactivity measures . The use of standard definitions of these measures
would greatly improve communications between those who collect data and those who use
the data.
REFERENCES FOR CHAPTER 3
1. Saltzman, B .E. Modified Nitrogen Dioxide Reagent for Recording Air Analyzers.
Anal. Chem. 32(1): 135, January 1960.
2. Wilson, D. and S. Kopczynski. J. Air Pollut. Contr. Ass. 18:160, March 1968.
3. Fontijn A A.J. Sabadell, and R.J. Ronco. Homogeneous Chemiluminescent
Measurement of Nitric Acid with Ozone. Anal. Chem. 42(6): 525, May 1970.
4. Hodgeson, J.A. , R.E. Baumgardner, B.E. Martin, and K.A. Rehme. Stoichiometry
in the Neutral lodometric Procedure for Ozone by Gas-phase Titration with Nitric
Oxide. Anal. Chem. 43(8): 1123, July 1971.
53
-------
<
cc
Ul
CJ
Ul
u
10
Tl/2
11 12
TIME OF DAY
13
14
15
16
17
Figure 29. Comparison of typical reaction data for indoor and outdoor smog chambers.
5. Saltzman, B.E. Colorimetric Microdetermination of Nitrogen Dioxide in the Atmosphere.
Anal. Chem. 2^:1949. December 1954.
6. Stevens, R.K., T. Clark, R. Baumgardner, and J.A. Hodgeson. Instrumentation for
the Measurement of Nitrogen Dioxide. U.S. Environmental Protection Agency,
Research Triangle Park, N. C. (Presented at American Society for Testing and Mate-
rials Symposium on Instrumentation for Monitoring Air Quality. Boulder. August
1973.)
7. Spincer, C.W. and D.F. Miller. Nitrogen Balance in.Smog Chamber Studies. Battelle,
Columbus, Ohio. (Presented at Annual Meeting of Air Pollution Control Association.
Denver. June 1974.)
54
-------
8. Environ. Sci. Technol. 8:1118, December 1974.
9. Hanst, P.L., E.R. Stephens, and W.E. Scott. Reactions Involving Ozone, Nitrogen
Dioxide, and Organic Compounds at Low Concentrations in Air. Proc. Air Pollut.
Inst. 35_(IH): 175, 1955.
10. Stephens, E.R., P.L. Hanst, R.C. Doerr, and W.E. Scott. Reactions of Nitrogen
Dioxide and Organic Compounds in Air. Ind. Eng. Chem. 48:1498, 1956.
11. Stephens, E.R. , W.E. Scott, P.L. Hanst, and R.C. Doerr. Recent Developments in
the Study of the Organic Chemistry of the Atmosphere. J. Air Pollut. Contr. Ass.
6:159, 1956.
12. Scott, W.E. , E.R. Stephens, P.L. Hanst, and R.C. Doerr. Further Developments
in the Chemistry of the Atmosphere. Proc. Air Pollut. Inst. 3J7(III): 171, 1957.
13. White, J.U. Long Optical Paths of Large Aperture. J. Opt. Soc. Ame'r. 32:285,
1942. —
14. Hanst, P.L. Spectroscopic Methods for Air Pollution Measurement. In: Advances in
Environmental Science and Technology. J.N. Pitts and R.L. Metcalf (ed.). New
York, John Wiley and Sons, Inc., 1971.
15. Hanst, P.L., A.S. Lejohn, andB.W. Gay, Jr. Detection of Atmospheric Pollutants
at Parts-per-billion Levels by Infared Spectroscopy. Appl. Spectrosc. 2_7: 188, 1973,
16. Bertolacini, R.J. and J.E. Barney. Anal. Chem. 29^281, February 1957.
17. Schafer, H.N.S. Anal. Chem. 39:1719, December 1967.
18. Leithe, W. The Analysis of Air Pollutants. Ann Arbor, Humphrey Science Publisher,
1970.
19. West, P.W. , A.D. Shendrikar, and N. Herrara. Determination of Sulfuric Acid
Aerosols. Anal. Chem. Acta. 69_: 11, 1974.
20. Feigley, C.E. Some Statistical Aspects of Air Monitoring Instrument Calibration.
University of North Carolina, Chapel Hill, N. C. (Presented at Air Pollution Control
Association National Meeting. Denver. June 1974. APCA No. 74-14. )
21. Tuesday, C.S. Chemical Reactions in the Lower and Upper Atmosphere. R.D. Cadle
(ed.). New York, Interscience, 1961. p. 15-49.
22. Stedman, D .H. and H. Niki. Photolysis of NO2 in Air as Measurement Method for
Light Intensity. Environ. Sci. Technol. 7(8): 735-740, August 1973.
23. Holmes, J.R. et al. Measurement of Ultraviolet Radiation Intensity in Photochemical
Smog Studies. Environ. Sci. Technol. 7(6): 519-523, June 1973.
24. Seinfeld, J.H. , T.A. Hecht, and P.M. Roth. Existing Needs in the Experimental and
Observational Study of Atmospheric Chemical Reaction. U.S. Environmental Protec-
tion Agency. Research Triangle Park, N. C. Publication No. EPA-R4-73-031.
June 1973.
25. Sickles, J. and H.E. Jeffries. Continuous Chemical Actinometer Using NO2> Univer-
sity of North Carolina, Chapel Hill, N. C. (In press.)
26. Jaffe R J F C. Smith, and K.W. Last. Study of Factors Affecting Reactions in
Environmental Chambers; Final Report on Phase III. Lockheed Missiles and Space Co.
Inc., Sunnyvale, Ca. 1975.
55
-------
4. BACKGROUND REACTIVITY
INTERPRETATION OF BACKGROUND REACTIVITY DATA
- tt. Itimilriadcs. KI'A
Traditionally background reactivity was determined and expressed based on the
following tests:
1 . Ozone Formation Test: Background air is irradiated and O3 formation is
measured and followed .
2 . RNO2~Test: Background air and added NO are irradiated and rate of NO2
formation (R) is measured.
3- Aerosol Formation Test: Background air is irradiated and aerosol formation
is measured and followed .
4. SO2 Oxidation Test: Background air and added SO2 are irradiated and loss of
SO2 (and formation of SO42~) is measured and followed.
In addition to these commonly used tests , some chamber operators have also been
conducting the following relevant tests:
1. Ozone Loss Test: Ozone loss in the chamber is measured in the dark and under
irradiation .
2 . NO? Loss Test: NO2 loss in the chamber is measured in the dark and under
irradiation .
3. Propylene Loss Test: Background air and added propylene are irradiated and
loss of propylene and formation of products are measured and followed.
Experiences from such tests with existing chambers and some interpretations and
use of resultant data are as follows:
Ozone formation tests have shown that all large chambers , regardless of degree
of cleanness, form up to 0.2 ppm of "oxidant" (as measured by KI) . Chambers with
barely detectable levels of HC and NOX contaminants also form 0.03 to 0.05 ppm oxidant.
It is submitted here that such oxidant is a result of photochemical reactions of HC
and NOX, and its concentration depends mainly on the HC and NOX concentrations and their
ratio, however small these concentrations may be. Because of the ratio effect, this
test has little value as an index of chamber contamination. Nevertheless, if con-
taminants are removed completely, to the point of no detection, then O3 formation does
not exceed 0.035 ppm. This has been demonstrated repeatedly in EPA tests using a 400-
liter Pyrex glass chamber .
57
-------
RNO?~tests have given results ranging form 2 to 12 times the thermal rate of NO
oxidation in air. It is submitted here that such accelerated oxidation of the NO is
caused mainly by photochemical reactions involving organic contaminants (surface may
also have a role) . This test, therefore, is a more valid index of organic contamination
in chamber and/or background air.
Aerosol formation tests have shown that condensation nuclei form at levels depend-
ing on contamination level. It is not clear as to what the relative contributions of
the organic, NOX, and 803 contaminants are to the measured total. In the absence of
measurable levels of contaminants, virtually no aerosol can be measured in the chamber.
In general, no light scattering aerosol forms from irradiation of reasonably cleaned
air and chamber.
SO2 oxidation tests have given results ranging from 0.05 to 5 percent per hour—
a serious inconsistency. While contaminants have been shown to accelerate SO2 oxidation,
there is also evidence that heterogeneous reactions or "third body" effects of surface
have a role. Thus, the lowest SO2 oxidation rate values were obtained with chamber
walls that had been preconditioned. The question of what causes the variation in SO2
photooxidation rate in "clean" air is still an open one.
Ozone loss, in the dark, has been found to be negligibly small in all chambers of
o
100 ft or greater size. Generally, with an initial concentration of 1 ppm, half
life of 03 ranges from 10 to 40 hours. Under irradiation, ozone loss in the chamber
is accelerated considerably. Many investigators have attributed the accelerative
effect of radiation to enhanced contact of 03 with the chamber walls. It is submitted
here that the radiation effect upon 03 loss is the result of photochemical reactions
i
involving chamber contaminants. The questions to be asked here are whether this ex-
planation is correct, and whether these reactions have a role in the mechai,;sm of the
03 formation and 03 decay processes (these reactions, for example, may account for the
03 decline following the O$ peak in irradiated HC-NOX systems).
The NO2 loss tests have not yielded any revealing information. Under irradiation,
NO2 disappears slowly, presumably as a result of oxidation into nitric acid.
The propylene loss tests have shown that propylene disappears suprisingly fast
when irradiated in "pure" air (in the absency of injected NOX) . Thus, in 6 hours of
irradiation, propylene consumption ranged from 30 to 80 percent. The consumption reac-
tion is probably initiated by the same chamber contaminant reactions that cause 03
formation in "pure" air; aldehyde products could instigate additional photochemical
activity resulting in propylene comsumption.
Overall conclusions and implications arising form this background reactivity
evidence are as follows:
Background contaminants, that is, organic and inorganic vapors at subtrace levels,
can cause reactivity manifestations of appreciable intensity. This obviously is
58
-------
important from a control standpoint. The evidence suggests that O3 levels comparable
to the O3 standard can be caused by precursor concentrations lower than those of the
present precursor (air quality) standards. Further, .the evidence suggests that for such
extremely low concentration systems, the role of NOX may be much more important —
relative to the role of HC—than in the higher concentration systems. Such evidence is not
inconsistent with real atmospheric observations. Ozone levels are low in source-free rural
areas perhaps because the NOX level is extremely low. With higher NOX levels (as, for
example, in mixtures of natural and anthropogenic emissions) that optimize the HC/NOX
ratio, O3 could form at significant levels. This is one of the possible mechanisms explaining,
partly at least, the high levels of O3 formed in some nonurban atmospheres. This mechanism
is being explored now by EPA. Limited initial data show that injection of NOX in rural air
can, indeed, result in increased O3 yield upon irradiation.
Certain reactivity manifestations appear to be strongly affected by the surface
factor. This is important in that it introduces a bias to the experimental study.
Conditioning of the surface is acceptable as a solution to the problem, provided, how-
ever, that such conditioning does not create contamination problems. Since the surface
to volume ratio in any one of the existing smog chambers is 2 to 3 orders of magnitide
greater than the surface to volume ratio, for example, in Los Angeles air during the
smog season (including porosity-related area of aerosol particles), it follows that
increasing the size of the chamber beyond, say, 1000 ft^ will not solve the surface effect
problem. Note, however, that this is a problem only when the chamber is used to study
atmospheric photochemistry. For study of OX/HC/NOX relationships at ground level, the
surface to volume ratios in chamber and real atmosphere may be comparable.
DISCUSSION
Dr. E. R. Stephens (UCR, SAPRC) suggested that PAN formation be used as an index
of chamber background contamination. The advantages of this index over "ozone formation"
are that PAN can be detected at concentrations as low as 0.1 ppb, and that presence of
PAN is an unequivocal proof of presence of both HC and NOX contaminants.
Dr. T. A. Hecht (SAI) reported on some recent modeling efforts to investigate
the causes of O3 decay in a smog chamber, in the dark and under irradiation. Dr. Hecht's
report was as follows:
Scientists have shown that the surface-to-volume ratio and surface materials of a smog
chamber influence the overall chemical process observed.1 The detailed nature and magni-
tude of the effects of surfaces on the kinetics. however, are poorly understood. The only wall
reaction that has been considered extensively is the heterogeneous decomposition of O3 (see
references in Seinfeld et al.) ,1 which can be represented as
kQs
03 •- Wall
Scientists at UCR have carried out several ozone decay experiments to determine kOs
in their evacuable cylindrical chamber. The chamber was prepared-for a »ko3 experiment"
59
-------
in the following manner. First, the reactor was pumped to a few microns of mercury pres-
sure. Air containing about 1 ppm of 03 was then rapidly introduced into the chamber, and
a total pressure of 1 atmosphere was achieved through the addition of pure air. Subsequently,
the time required for one half of the ozone to be consumed in the dark was determined. The
half-life was found to be about 8.35 hours.
If we assume that the loss.of 03 during a decay experiment is attributable entirely to
the reaction on the walls, the rate of 63 decay is given by
d03
= -k03 [031
dt
The half-life of 63 is related to the 03 decay constant as
0.693
k03 =
_
from which we calculate that kg, = 0.0014 min . This value indicates, for example, that
the rate of 63 loss to the walls will exceed the rate of loss due to reaction with propylene ,
should the concentration of 63 H^ in the reactor be less than 0.08 ppm. (Two of the eight
experiments in the propylene experimental block recently conducted under EPA sponsorship
had an initial Cjtlfr concentration of 0.1 ppm.) Thus, the rate of O, decay in the UCR chamber
is significant, and the O, wall loss reaction must be included in our simulations.
Ozone decay experiments were also performed under irradiated, rather than dark,
conditions, and the O~ half-life was found to be about 4.25 hours. We believe that this decrease
in the 63 half-life from that observed in a dark chamber may be due to a series of chemical
reactions involving 63, O^D), H^O, free radicals, and light, rather than ^ny intrinsic change
in the activity of the walls when the solar simulator (light source) is in operation . (The solar
simulator is focused so that it does not shine directly on the cylindrical surfaces . )
We carried out a mathematical simulation of the August 15, 1973, light decay experiment,
which was similar in all ways, except for the presence of light, to the dark decay experi-
ments used to calculate kg above. We assumed that the 13 reactions in Table 12 — in addition
to the wall reaction — participated in the total 03 decay process.
The 03 photolysis rates were based on the results of Demerjian et al. ,2 who calculated
the photolysis rates of NC>2, 63, and many other species for sunlight with z = 45°. Under
those conditions, they calculated that the photolysis constant for NO2 was 0.48 min~l. Since
the rate of that reaction in the UCR chamber is 0.223 min" , we scaled Demerjian' s 03 photol-
ysis constants by 0.223/0.48. We further assumed that the rate of }^2Q2 photolysis was 1/250
of the rate of NC>2 photolysis . This rate constant is quite uncertain; we found that this reaction
can be eliminated from the mechanism without seriously affecting the reactions , and so the
choice of this rate constant is not critical . The rate constants of the other reactions are based
on the recommended values of Garvin and Hampson^ and Demerjian et al.
60
-------
Table 12. REACTIONS PARTICIPATING IN THE TOTAL OZONE DECAY PROCESS
Number
1
2
3
4
5
6
7
8
9
10
11
12
13
Reaction
03 -> Wall
03 •*• hv -»• O^D) + 02
03 + hv + 0(3P) + 02
0(1D) + M -> 0(3p) + M
0(3P) + 02 + M -*• 03 + M
0(1D) + 03 -»• 202
OpD) + H20 -> 20H
OH + 03 -> H02 + 02
H02 + 03 -»• OH + 202
H02 + H02 -»• H202 + 02
H202 + hv -> 20H
OH + OH + H20 •*• H202 + H20
OH + OH + M + H202 + M
Rate constant9
1.38 x 10"3 min"1
1.58 x 10~3 min"1
9.76 x 10"3 min"1
8.70 x 104
2.00 x 10"5 ppm"2 min"1
9.80 x 104
5.25 x 105 -
8.60 x 101
2.40
8.40 x 103
8.90 x 10"4 min"1
3.25 x 10 ppm"2 min"
-2 -? -1
6.50 x 10 ppm min
aln units of ppm" min" unless otherwise indicated.
Integrating the mechanism in Table 12 and using the initial conditions for the experiment
([O3]o = 1.15 ppm, [H2] = 2.0 x 104 ppm, [O2] = 2.0 x 105 ppm, and [M] = 1.0xl06ppm)
resulted in a predicted C"3 half-life of 4.48 hours. The observed half-life for this particular
experiment was 4.35 + 0.09 hours.
The results of this exercise are consistent with two tentative conclusions. First, the
"true" 03 wall decay constant appears to be that measured in the dark, rather than in the
illuminated decay experiments. Second, the O^ photolysis reactions, which initiate reactions
such as O(-'-D) + I^O, are an important sink for 03 and should be considered in future smog
chamber simulations.
REFERENCES FOR CHAPTER 1
1. Seinfeld, J.H., T.A. Hecht, and P.M. Roth. Existing Needs in the Experimental
and Observational Study of Atmospheric Chemical Reactions: A Recommendations
Report. U.S. Environmental Protection Agency. Research Triangle Park, N .C .
Report No. EPA-R4-73-031. 1973.
2. Demerjian, K.L., J.A. Kerr, and J.G. Calvert. The Mechanism of Photochemical
Smog Formation. In: Advances in Environmental Science and Technology (Vol.
4). J.N. Pitts and R.L. Metcalf (ed.). New York, John Wiley and Sons, 1974.
p. 1-262.
3. Chemical Kinetics Data Survey VII; Tables of Rate and Photochemical Data for
Modelling the Stratosphere. D. Garvin and R.F. Hampson (ed.). National Bureau
of Standards. Washington, D.C. Report No. NBSIR 74-430. January 1974.
61
-------
5. COMPARABILITY OF SMOG CHAMBER ATMOSPHERE
AND REAL ATMOSPHERE
A METHOD FOR COMPARING SMOG CHAMBER DATA WITH ATMOSPHERIC
DATA- K.R. Stephens, SAPC, URC
Ozone is considered to be the major health hazard in photochemical smog, and so
control strategies have been directed toward reducing ozone exposures to acceptable levels.
This toxic compound is a product of a very complex reaction of primary pollutants (hydro-
carbons and nitrogen oxides) , which are not themselves highly toxic. A new project has
been initiated at UCR to develop, demonstrate, and use methods for the study of polluted
air that will help clarify this complex relationship between oxidant and oxidant precursors.
Understanding of this relationship is vital to sound development of control strategy, air
quality standards, and emission standards.
Emission standards for hydrocarbons and nitrogen oxides cannot be independent
because of the complex atmospheric interaction to produce ozone, which is affected by sun-
light intensity and time available for reaction. Laboratory irradiation chamber data; (for
example, Reference 1) have been used to estimate emission standards needed to meet air
quality standards. On the other hand, ambient air data form the basis of the EPA recom-
mended oxidant abatement strategy. Both approaches have been subject to criticism. Any-
one who has tried to use either approach will recognize that much of the criticism is valid.
The proposed project has as its primary objective the development of new approaches to the
use of ambient air quality data for the study of oxidant/precursor relationships. This is
proposed as an adjunct to the study of oxidant/precursor relationships in smog chambers or
with computer models rather than a substitute—a kind of "ground truth." If these three
approaches can be reconciled with each other so that one control strategy emerges, we will
be in a far stronger position than if only one or two of these methods are used. The various
procedures used to estimate the degree of control required for auto emissions give answers
2
varying between 90 and 97 percent or even more. While this may seem to be an acceptably
narrow range, when translated into emission standards, it gives a variation in allowable
emission of more than 3 to 1. This is far from trivial to the automotive engineer.
One of the principal weaknesses of the oxidant/precursor relationship as derived
from ambient air data in the Air Quality document^ is that oxidant and precursor are
measured on two different parcels of air. Precursors are at a maximum in the 6 to 9 a.m.
hours, while oxidant maximum occurs at noon or later and often in a different place.
The assumption that the measured oxidant is produced by the measured early
morning HC/NO is especially serious when you think of the use to which these curves
(scatter patterns) have been put. One draws a boundary around the scatter pattern and
extrapolates to the oxidant air quality standard. The position of the boundary is determined
by those few points at the lowest values of HC/NOX and oxidant. These are also the points
most subject to error because of their small values. The nonmethane hydrocarbon (NMHC)
63
-------
is especially vulnerable if it is estimated by subtracting methane from total hydrocarbon.
What is needed is a way to estimate the NMHC and NOX that correspond to the oxidant mea-
sured at some particular time and place. A method for back extrapolation in time for an air
parcel that has had a chance to react and develop maximum oxidant is also required. This
involves estimation of the amount of hydrocarbons that have reacted to provide a better
comparison with chamber data then the data so far used. For back extrapolation of hydro-
carbon values, we proposed to extend the procedures first used and described in Reference 4.
In that paper the relative amounts of three hydrocarbons of widely differing re-
activity (acetylene, ethene, and propene) were used to estimate that a Riverside smog has
been photoreacted for about 6 to 8 hours. This estimate was then combined with data on
photolysis of 'ambient air to estimate that about one-third of the NMHC had reacted.
To back extrapolate the NOX to zero time, the use of a chemilurninescent analyzer
is proposed. This can be done by taking advantage of one weakness of this method. This
weakness is the fact that the catalysts reduce not only NC^ but PAN and probably nitrate to
NO. The instrument, therefore, probably in the NOX mode, gives a good measure of the
initial oxides of nitrogen. Catalytic reduction then constitutes back extrapolation.
Sampling and Data Collection
4
The hydrocarbon analysis reported in the previously cited reference was carried
out opportunistically, sporadically, and manually. No paraffins higher than C^ nor olefins
higher than Cg were measured. It was, therefore, necessary to estimate the higher molecu-
lar weight hydrocarbons. It is proposed in the new project to monitor methane, the three
G£ hydrocarbons, and nonmethane hydrocarbons by a gas chromatographic procedure using
direct sample injection followed by back flush. This will also permit unattended operation.
A considerable part of this method has already been demonstrated. A recent report from this
5
laboratory describes a flame ionization chromatograph that will give abou X 5 mm/ppb peak
height for the C^ hydrocarbons with direct injection of 3.2 ml of air sample. Less than 5
ppb would be detectable and 20 ppb of each C? would be readily measurable.
Another instrument has been automated for this same procedure. To keep the
methane peak on scale, this instrument was operated at high attenuation. Even under these
conditions the C2S in ambient air were readily visible. There is such a large disparity be-
tween the concentrations of methane and the C2S that measurement of these on the same
attenuation is not practical. This procedure is now being extended by back-flushing the
remainder of the hydrocarbons after elution of methane and the C£S. An important advantage
is that C3+ hydrocarbons are measured directly and not by difference between total hydro-
carbon and methane. This chromatograph will give five pieces of data: methane, ethane,
ethene, acetylene, and C3+ hydrocarbons. An early chromatogram is shown in Figure 30.
The ethene/acetylene ratio will be used as an index of degree of reaction.
The fourth measurement will be ultraviolet intensity. For this project, no attempt
will be made to make quantitative, calibrated measurements of ultraviolet intensity. Both
64
-------
TIME, min
Figure 30. Injection and back flush of ambient air sample.
integrated and instantaneous values of relative intensities will be used to judge whether lack
of ultraviolet radiation can, in any given case, explain a low oxidant value.
Data from these four monitoring instruments (see Table 13) , which will be set up
on the UCR campus, will be reviewed for degree of reaction (as judged by ethylene/acetylene
ratio) and adequacy of sunlight. For those cases that show a substantial degree of reaction
and substantial sunlight, the oxidant will be plotted as a contour against NMHC and NOX
(back extrapolated as described) .
Several "bonus" benefits may well be derived from the present project. It will be
possible to study the variation in NMHC/NOX with time, degree of reaction, and atmospheric
conditions. It can be argued that the ratio in a community inventory should be roughly con-
stant from day to day and that major social changes would be required to change this ratio.
65
-------
Table 13. SUMMARY OF MEASUREMENTS TO BE MADE AT UNIVERSITY OF CALIFORNIA, RIVERSIDE
(All on a Continuous or Automatic Basis)
Measurement
Measurement method
Hydrocarbon
Nitrogen oxides
Oxidant
UV intensity
Gas chromatography with flame ionization. Direct injection
measurement of methane, ethane, ethene, acetylene, and C3
hydrocarbons (by back flush).
Ozone chemiluminescence. This will measure NO and NOX and
the latter will be taken to be total oxidized nitrogen.
Coulometric KI method.
Recording of UV meter.
It would be helpful to know if control strategies should take into account large random
fluctuations in this ratio. It could make a substantial difference in standards. Comparisons
of the average ratio in reacted and unreacted mixtures will be a useful corroboration of our
back extrapolation procedures. The data will permit comparison of methane with higher
hydrocarbons to see if the former constitute a constant fraction of the total. Methane may
form a useful index of effective trapping of pollutants, since it is inert, emitted in large
quantities , and easy to measure. It will be interesting to see if photochemical smog itself
significantly reduces ground level ultraviolet radiation.
Discussion
In response to several questions from the audience, Dr. Stephens explained further
his method for estimating the HC concentrations responsible for the oxidant observed in
ambient air. The method is based on the diagrams of Figure 31, depicting expected changes
in the relative levels of ethylene, propylene, and acetylene in an air sample as the sample
is irradiated by sunlight. The diagrams of Figure 31 were based on experimental measure-
ments as well as on certain assumptions. Thus, analyses of morning air were used to
establish the relative levels of ethylene, propylene, and acetylene in the unreacted air
sample (point at 0 hours in Figure 31). Changes in the three-component mixture with irra-
diation time were calculated assuming "batch reactor" and "stirred flow reactor" kinetics,
and using first-order reaction rate constants determined experimentally by irradiation of
morning air. For application of the method, an afternoon sample of air is analyzed for
hydrocarbons. The relative levels of ethylene, propylene, and acetylene are compared
with Figure 31 to estimate extent of reaction in terms of hours of irradiation. Using*then the
first order rate constant for acetylene and assuming stirred flow reactor kinetics, the con-
centration of acetylene at zero reaction time is estimated by back extrapolation. The zero-
time concentrations of the other hydrocarbons are then estimated using first order rate
constants derived from ambient air irradiation experiments to back extrapolate to zero time.
66
-------
RATIO
ACETYLENE TfflS
ETHENE 0.85 40.5
PROPENE 0.25 11.9
Figure 31. Ambient air photolysis. (Ethane
+ propene + acetylene = 100%.)
STATIC AND DYNAMIC SMOG CHAMBER TESTS - /,. /?,>/«'n»n, K77
The ultimate or prime purpose of air pollution studies is to provide information that
will aid in protecting the health and welfare of the breathing public, present and future—
presumably the nonbreathing public has lost interest in the problem.
The purpose of the smog chamber studies is to characterize the behavior of polluted
air and to obtain information that is of value in suggesting air pollution control tactics and
strategy.
This presentation considers studies dealing with the atmospheric generation of
ozone (03) . These studies can be classified in several ways; one such classification system
could contain two general components:
1. Characterization descriptive studies of chemical models of the existing
atmosphere by variation of reactants, etc.
2. Studies designed to test the effects of possible control measures upon the
03 dosage of the atmosphere, including studies to determine the worst
possible conditions (i.e., maximum 63) derivable from various concentra-
tions and ratios of initial reactants.
The topic here is the dynamic operation of smog chambers. Nondynamic static
studies are batch reaction systems with no more mass transfer in and out of the chamber than
is necessary to sample and replace sample volume. "Dynamic" means moving or changing,
not static.
In a dynamic chamber study the factors that are subjectable to the manipulation of
the operator, once a general set of conditions has been initiated and a run started, are:
67
-------
1. Temperature
2. Pressure
3. Light intensity
4. Dilution
a. Timing (e.g., dilution starting at time = t proceeding at a rate
of X % per unit time)
b. Composition of diluent air
c. Addition of more reactant without withdrawing chamber volume
d. Pressure (only use is to simulate upward movement of air to
strata aloft)
Temperature has an effect, but apparently this effect is caused by the.maximum
temperature. Figure 32 presents field data showing this effect. Dynamic temperature pro-
gression apparently has little effect. This is apparent in the data to be presented by Dr.
Jeffries.
"E
IU
o
IM
ouu
250
200
150
100
50
I I I
•— — -
0
•^
— • * • —
_ •*.****_
•
• •
— » * 0 * —
• * % *
_ • • _
1 1 1
ouu
250
200
150
100
50
n
1 1 1
v;. .. ,„ mi_n
•
0i
, • ,
•v ' :•
-— — ^ .
«k
• • .
| — •
• • • f *
-_ • _
1 1
40 50 60 70 80 40 50 60 70 8
MAX. TEMPERATURE, °F
(YOUNGSTOWN, OHIO)
MAX.TEMPEKA.URE, °F
(ERIE, PA.)
Figure 32. Maximum 4-hour-average ozone, Kane, Pennsylvania, versus maximum temperature at nearest
reporting stations, October 1974.
Natural changes in light intensity, however, do have a significant effect on rates,
peaks, concentrations, and time of occurrence of various events in the course of a chamber
run. Again, this is evident in the data and simulations of Dr. Jeffries. This effect is to be
expected on purely theoretical grounds. The actinic activity of light for CH^CHO and NO2
below 4000 A varies by almost 14-fold from zenith angle Z = 40 to Z = 80 for acetaldehyde
and about 5-fold for NC>2 from Z = 40 to Z = 80. The use of a single high intensity of light
compresses the time span between events in the smog chamber and the use of a single low
intensity can make these unrealistically drawn out.
Static studies have been used to understand the chemical relationship involved. In a
sense, they are simpler because they are all chemistry. Meteorology and fluid dynamics
have been removed from experimental conditions.
68
-------
In order to get studies started and to gain an appreciation of the overall effect of the
kinetic processes involved in photochemical air pollution systems, the batch study was the
order of the day. In these studies, once initial concentrations of reactants were set, 0)3 con-
centration was a function of photochemical primary and secondary reactions of fixed initial
amounts and ratios of NOX, organic vapors, etc. In the real atmosphere, the 63 concentra-
tion in air following a trajectory is a function of photochemical primary and secondary reac-
tions and thermal reactions involving precursors whose concentrations are affected not only
by chemical reactions but by dilution and the behavior of primary sources. Even under "com-
pletely stagnant" conditions, the sources have a generally cyclic behavior and the effective
volume is changed by the varying depth of the mixing layer.
For characterizing the real atmosphere, the operation of the chamber should follow a
physical model. That model should provide simulation of a real set of circumstances. Ideally,
it should approximate the content of the real atmosphere both as to kinds of pollutants and
their concentrations and ratios and it should approximate the injection rate of new material
and the dilution of the original volume. The part of the physical model describing the dilu-
tion should be worked out with or developed by a meteorologist. That dilution has an effect
on absolute 03 concentration achieved is shown by the UNC chamber data shown in Figures
33 and 34.
1.0
10 11 12 13 14 15 16 17
0.1
0.0
Figure 33. Effect on ozone generation by dilution of reactants: NO, N02, and 03, October 7, 1974.
69
-------
u
u
S
9 10
TIME OF DAY
Figure 34. Effect of reaction and dilution on a simulated urban mix of hydrocarbons, October 7, 1974.
Let us look at three kinds of dynamic situations and suggest ways of dynamic
simulations:
1. A stagnant situation where there is virtually no dilution but that due to increased
mixing height. Here then could be a programmed injection of new reactants with
minimum carrier air. (Here it should be noted that all dilution or injection in a
dynamic program should be accompanied by good mixing.) The rate and time of
injection should be correlated with the operating cycles of the local sources .
2. (Somewhat similar,) The drift of urban air over suburbs (e.g., Los Angeles to
Riverside or to Azuza). The kind and rate of reactant injection should simulate
the behavior of the sources enroute and the degree and kind of dilution may be
different or similar to simulation 1—certainly one would expect the L.A.-
Riverside model to be different from a N.Y.-Jersey City model.
3. Drift from city to country (Figure 35).Start a simulated urban mix and at time t
begin diluting. Here a meteorologist should be able to help in estimating at
what rate the city system is diluted by country air. (previous acetylene sampl-
ing in a trajectory from the city to some place downwind could help also.)
What is in the simulated country air used for a diluent: "natural" air, "eastern
rural" air or "clean" air? (Clean air: air from a clean air source such as a
catalytic oxidizer with subsequent NOX scrubbing. Simulated natural air: NOX
at 6 ppb, 0)3 at 0.02 ppm, hydrocarbon at 0.10 ppm C. Simulated eastern
"high-O3M air: NOX at 10 ppm, 63 at 0.10 ppm, an.d hydrocarbon at 0.25 ppm C.)
70
-------
0600-0900 INJECTION OF SIMULATED
CITY POLLUTION
BACKGROUND AIR-
START WITH:
CLEAN AIR,
SIMULATED NATURAL AIR OR
SIMULATED "EASTERN
RURAL AIR-
BEFORE SUNRISE
CITY
DILUTE BY p.0%
NO
HC, ppm
3.00 0.95 0.3
0.32
0.1
SUNRISE
1st DAY
0600-0900
DILUTE BY 90-95%
OPERATE AS A
CONTINUOUS STIRRED
TANK REACTOR
DILUENT AIR:
CLEAN AIR,
SIMULATED NATURAL
AIR OR
SIMULATED "EASTERN
RURAL AIR"
OPERATE AS A
BATCH REACTOR
REPLACE SAMPLE
VOLUME ONLY
1430
1st DAY
1430
2ND DAY
SUNSET
3rd DAY
Figure 35. Chamber simulation of extended urban influence on surface ozone concentration.
Although much can and will be done with smog chambers operating in the batch
mode, dynamic manipulation of the operating variables can aid in closer simulation of ambient
conditions. It can possibly produce better data than batch studies for setting standards and
V
determining control strategy.
SIMULATION OF NATURAL SUNLIGHT- E. Stahel. NCSV
The outdoor sr.og chamber's basic problem as it relates to sunlight is to extract from
an arbitrarily real (daily atypical) diurnal variation a baseline case devoid of local artifacts.
The indoor smog chamber routinely operated today has to interpret an unrealistic baseline
case, i.e., reactions under steady illumination. The acceptance of the latter as a real atmos-
phere simulation is tenuous.
What characteristics of natural sunlight can be identified as relevant, of what level
of significance are they, and under real constraints, what impact do they have on chamber
design? Light intensity, spectral distribution, and their respective diurnal variations ap-
pear to be basic, with the mode of the radiation, whether direct, scattered, reflected, etc.,
of some interest.
At the outset, it is enlightening to reflect on the physical constraints of the indoor
smog chambers to be illuminated as their fundamental design parameters affect the light
simulation. The following concepts are open to discussion but form, I believe, the basics of
chamber design from a reaction engineering point of view. We are trying to ensure a spa-
tially uniform concentration field and a spatially uniform light intensity, with minimum ther-
mal gradients and heterogeneous effects. Some of the consequences of these in terms of the
71
-------
state of the art are the following: Experimental chambers must be mixed at the outset and
throughout the reaction cycle. Mixing must effect both macro- and microuniformity. Cham-
bers must be relatively large, 500 to 1000 ft3. The wall effects as interrelated to mixing are
complex and unresolved, but sampling flexibility and surface to volume ratio dictate this
range. Light intensity uniformity for large chambers of approximately 1000 ft3 can be
achieved at reasonable cost in only limited ways. The most desirable geometric arrangement
appears to be the halo-lighted right circular cylinder. The intensity decrease with the
square of the distance from the cylindrical outer surface is balanced geometrically by the
compensating decrease in illuminated area. Uniformity is achieved, barring end effects
that require some augmentation. The common array of predominately blue with black and
sunlight flourescents lends itself to this application at low cost. Discharge lamps, while
undoubtedly more spectrally similar to sunlight, have severe limitations in meeting the two
levels of uniformity outlined above.
Diurnal changes in frequency and spectral distribution deficiencies of illumination
systems are secondary effects, compared to diurnal intensity variations. These effects,
while tentatively observed, have not been documented experimentally and are kinetically
very complex. John Nader's volumetric intensity measurements compare favorably with
Leighton's theoretical prediction of diurnal intensity variation and form the basis for a
pragmatically programmed diurnal cycle monitored by kj evaluation techniques to assure
the absolute level as well. Returning to the candidate chamber—1000 ft3 in volume illumi-
nated by a surrounding array of flourescent lamps—it is of incremental cost to program
the large number of lamps in, say, decades controlled to approximate the given diurnal
light intensity cycle.
Recent data tend to indicate major effects of diurnal intensity variation compared to
constant illumination, and this appears to be the single greatest weakness i f existing indoor
chamber data in simulating stagnant air mass kinetics. Dynamic considerations only rein-
force the criticality of the reactor engineering concepts enumerated.
OUTDOOR SMOG CHAMBERS - Hjeffri™. UNC
The "theoretical" advantages of an outdoor smog chamber are no longer theoretical—
they are real. The UNC dual outdoor smog chamber was finished in May 1973 and it has been
operated almost continually since that time, except for the winter months of December-March.
Figure 36 gives details of the chamber shape and construction. The chamber volume
is 12,000 ft3, divided into two 6000-ft3 compartments. The walls and floor are 5-mil FEP
Teflon film secured to an exterior wooden frame by aluminum u-channels. The design and
performance of the chamber have been highly satisfactory. It is a durable structure. It has
withstood rain, ice, snow, storm winds, and temperatures ranging from 20 to 95° F without
any signs of stress. The cover system is only used during the winter months. The chamber
is not a greenhouse, that is, the air temperature rise above ambient is less than 8° F during
intense irradiation. It has excellent light characteristics: approximately 85 percent trans-
72
-------
END
STRUCTURE OF CHAMBERS
1|
SIDE
30ft
TQp
FLOOR PLAN OF CHAMBERS
Figure 36. The University of North Carolina smog chamber.
SAMPLING LINES
MANIFOLD
mission of total solar radiation (TSR) and approximately 93 percent transmission for ultra-
violet solar radiation (UV) less than 4000 A. The transmission losses are largely compensated
for by a reflective floor for heat and radiation . This results in a Ka for NO_ + hv of 0 . 37
min~l when TSR =1.0 cal cm~2 min~l. It is well mixed. Thermal mixing occurs, in addition
to that from the two 1000 ft^/min mixing fans in each chamber half. Samples from "corner"
locations are the same as those from the middle of the chamber, and concentrations of materi-
als in the chamber follow an exponential dilution curve during exhausting very well. The
chamber exhibits a low background reactivity: the ozone (63) half life is greater than 20
hours in the light and greater than 45 hours in the dark at an initial concentration of 0.85
ppm; the nitric oxide (NO) oxidation rate for background air is 0.2 ppb/min at [NO]0 =0.45
Ppm and 0.1 ppb/min at [NOlo =0.24 ppm; the 63 formation in blank runs varies with
ambient air and is 0.05 to 0.08 ppm 03 when ambient air concentrations rise to 0.03 to 0.04
Ppm. The compartment to compartment replication is excellent; this is partly due to the. use
of a fully computerized data acquisition and control system that operates the chamber and
to instruments that are time shared between each chamber half.
73
-------
More than 63 propylene/NO and 68 simulated hydrocarbon mix/NOx (mix/NOx)
dual 12-hour runs (that is, 262 individual experiments) have been conducted in the chamber
in the last 15 months. Some of these will be presented below to indicate the effects of varia-
tions in natural sunlight and temperature on the chemical profiles.
The first order rate constant for the photolysis of NC>2, herein called Kj (=(j>Ka) ,
is one of the most sensitive parameters in a HC/NOX photochemical system. The value of Kj
is usually measured by NC>2 photolysis in N£ as either KQ, a pseudo-first-order rate, or
better as cjiKa by using a kinetics derived relationship such as that developed by Holmes et al.
Sickles and Jeffries^ at UNC have developed a continuous unit to measure Kj. The device
is a 1-liter quartz flask operated as a continuous stirred flow reactor. The calculating
equation for K^ as a function of time in real sunlight is given in Table 14. - The advantages
of this device over other techniques are numerous. It is nearly ideal as a three-dimensional
sensor; it is continuous and rapid; it is based on a complete and well verified kinetics
mechanism that includes all essential nitrogen chemistry; and the rate expression does not
have to be integrated.
This continuous chemical actinometer for NOX (CCANOX) has undergone extensive
evaluation by computer simulation (using the EPA program with modification for variable
light intensity) and by field testing. Figure 37 shows data taken on September 19, 1974,
using the CCANOX and an Eppley total solar radiation sensor, which measures the direct and.
diffuse intensity on a horizontal surface. The K± data are presented as discrete points be-
cause a black bag was used to check the zero point between each reading. No explanation
can be offered at this time for the fact that the volumetric (that is, the three dimensional)
light intensity for wavelengths <4000 A follows the same pattern as does the total intensity on
a flat horizontal surface. The evidence, however, for this particular set of measurement
conditions (there is a similar set for September 18), seems overwhelming The data strongly
suggest that, at least at our site, the TSR readings from the Eepley sensor can be taken as a
good approximation to Kj.
Figures 38 and 39 show the TSR and chamber air temperature for two selected run
days in the outdoor chamber. The chemical system under study for each of these days was
2.0 ppmC NMHC, propylene, 0.50 ppm NOX, and 10 percent NC>2. The concentration-time
profiles for these days are presented in Figures 40 and 41. May 7, 1974, was a dual matched
run and Figure 40 indicates the degree of side to side reproducibility of the UNC chamber.
The NC>2 data in these runs were acquired with a chemiluminescent NOX analyzer, which is
subject to late-afternoon interference by HNO, and PAN, hence the dashed lines in the
figures. Other work° has shown that the late afternoon NO2 reading is [NO2 = PAN] and the
HNO3 is probably lost in our sampling lines.
In all of the outdoor runs, the variation in the NO and N©2 time profiles is not very
much affected by what appears to be substantial variations in the TSR. Besides the slight
adjustments to the time axis to match NO-NO2 cross-over points, there are subtle differences
among the NO and NO2 profiles and much larger differences among the 03 profiles. However,
74
-------
Table 14. DERIVATION OF CONTINUOUS CHEMICAL ACTIONOMETER FOR NOX (CCANOX)
(by Joe Sickles)
1. Accumulation = IN-OUT-RXN
Accumulation =
df
^computed from inst. readings>
<[N02Jf = outlet cone., [N02] = inlet conc.>
IN
= [N02]o
OUT = MIf
RXN = -2K]
V.
2. Let X =
\[N02]f [N02]f
where R] = 0.272 R2 = 0.261 R3 = 1.59 x 10
[N02]o - [N02]f
[N02]o
3. Solve first equation:
Mt) =(
, Drop RS Term <[02] % 0>
1
(1 -
X p + R
MSS) =— -
2r Ll - X
j_ R2 X
(1 -
4. Calculating Forms: Let AN02 = [N02]b - [N02]f
d[N0]f
Kl(f) =
1
K,(SS) =
, rate expression
does not have to be integrated. Zero tested with black bag. Initial 02
flushed from system in 7-r. Need only volume.jHow ra^te, jNOjjJo^
May 19 (Figures 40 and 41) shows small changes in NO and NO2 slopes that correlate well
with changes in TSR.
It is difficult to discern these effects in the concentration time profiles, but much easier
in the rate or first derivative-time profiles. Figure 42 shows the NO and NO2 rate
75
-------
2.0
1.8
1.6
i—i i—r
Ki READINGS - 7 ft ABOVE GRASS
SR READINGS-EPPLEY ON 25 ft MAST
TEMP
(0-100 °F FULL SCALE)
10 11 12
TIME OF DAY
16 17
0.715
0.644
0.572
0.501
0.429
0.358
0.206
0.215
0.143
0.072
0.0
Figure 37. Comparison of total solar radiation on horizontal surface with rate of photolysis of N02
(latitude 35.72°), September 19, 1974.
2.0
1.8
1.6
•| 1.4
I 1.2
51.0
5
< 0.8
DC
<
-0.6
tn
0.4
0.2
0.0
I I I I I I I
I I f
TEMP
I I I I I I I
1001
90
80
70
60 iu-
K
50
40
30
20
10
10 11 12 13 14 15 16 17
TIME OF DAY
Figure 38. Diurnal variation of solar radiation and temperature, May 7, 1974.
76
-------
0.0
10 11 12 13 14 15 16
T!ME OF DAY
Figure 39. Diurnal variations of solar radiation and temperature, May 19, 1974.
17
I ! I I I
0.0
10 11 12 13
TIME OF DAY
Figure 40. Comparison of concentration-time profiles from the two chamber compart-
ments (red and blue), May 7, 1974. :
77
-------
5 6 f 7
SUNRISE
10 11 12
TIME OF DAY
Figure 41. Concentration-time profiles. May 19, 1974.
4.0
3.6
3.2
2.8
,- 2.4
.'=
I 2.0
< 1.6
1.2
0.8
0.4'
0.0
I I T
PROPYLENE/NOx
6
I I
8
NO RATE
13
9 10 11 12
TIME OF DAY
Figure 42. Rate of NO disappearance and NC>2 appearance
for a clear day, October 13, 1973.
14
78
-------
computed by a digital 5-point formula from the time profiles of October 13, 1973, a totally
clear, full sunshine day. The rate changes are smooth and peak values are 3.5 and 3.0
ppb/min for NO and NO2, respectively. The NO and NO2 rate for May 19, 1974, are given
in Figure 43. The deacceleration and acceleration with changes in TSR are evident; however,
the system still achieved a peak rate equal to or slightly greater than the clear day of
October 13, 1973. Note that the flat part of the rate curve lasted almost 1 hour.
4.0
3.6
3.2
2.8
.2 2.4
<
ec
1.6
1.2
0.8
0.4
0.0
PROPYLENE/NO
MAY 19,1974
6
9 10 11
TIME OF DAY
12
13
14
Figure 43. Rate of NO disappearance and N02 appearance
for a partly cloudy day, May 19, 1974.
In Figure 44, the O3 profiles from 7 days are displayed together with TSR profiles.
The O3 profiles are much more affected by TSR variations. The O3 that accumulates in the
system is the difference between that which is generated by NO2 photolysis and that which is
lost by reactions with NO and to a much lesser extent olefins. Free radicals from the degra-
dation of hydrocarbons and aldehydes by O, O3, OH, and photolysis compete with O3 for the
NO to regenerate NO2. Thus, the buildup of O3 is dynamic, A free radical NO oxidation
chain, which is relatively short, can sustain itself for short periods without fresh initiation,
by photolysis steps, but depending on the state of the system, the acceleration of O3 accumu-
lation may decrease or at the worst become negative. Thus, the effects of "choppiness" in
the TSR profile on O3 profiles depend upon the frequency and where in the chemical process
it occurs.
Figures 45 to 48 show TSR, NO, NO2, and O3 profiles from two selected days of out-
door chamber runs in which the chemical system was a less reactive simulated urban hydro-
carbon mix at 2.0 ppmC, 0.35 ppm NOX, and 20 percent NO2. The mix contains acetylene,
79
-------
'JO
o
I I I I
0123
TIME FROM CROSS-OVER POINT, hours
Figure 44. Comparison of ozone profiles and solar radiation profiles for propylene runs.
-------
1
%
o
O
e/s
0.0
10 11 12 13 14 15
TIME OF DAY
16
17
Figure 45. Diurnal variation of solar radiation and temperature, September 19, 1974.
9 10 11 12 ; 13 14 15 16 17
Figure 46. Concentration-time profiles, September 19, 1974.
81
-------
6
8
9
13 14
IS 16
10 11 12
TIME OF DAY
Figure 47. Diurnal variations of solar radiation and temperature, September 23, 1974.
0.50
0.45
0.40
0.35
E
^0.30
510.25
I I I I
— HYDROCARBON MIX/NOX
I I I
0.20
I
I
0.15
0.10
0.05
0.00
I I I
0
13 14 15 16
10 11 12
TIME OF DAY
Figure 48. Concentration-time profiles, September 23, 1974.
17
82
-------
paraffins, olefins, and sometimes toluene and m-xylene. The conversion of NO to NOo is
Lt
much slower (approximately 2.5 less reactive than propylene) and it generates less PAN to
interfere with the NO2 afternoon readings. The lower reactivity reflects itself in a lower
free radical flux, which is less competitive with O3 for NO in the afternoon, leading to lower
[03] and higher [NO] and less consumption of NO2>
The effect of TSR variations on NO and NO2 profiles is also less than might be ex-
pected. Comparison of NO and NO2 profiles for Figures 45 and 46, September 19, 1974, with
Figures 47 and 48, September 22, 1974, shows near perfect matches until 12:00. The effect
of the "chop" in TSR can be seen in the NO2, NO, and O3 profiles as each dip in light intensi-
ty causes free radical flux, changes, The overall effect is a slightly delayed NO2 peak,
slightly higher NO, and approximately one-half the ozone on September 22. In Figure 49 for
October 8, 1974, the large "hole" in TSR at the end of the run results in a decrease of the
03 acceleration to zero, and thus this day did not reach the final O3 value that September 19
did (see Figure 49) . In Figure 49, it is evident that TSR effects (perhaps in combination
with temperature effects) can easily decrease O3 production by factors of one-half for what
are otherwise apparently equal conditions.
With respect to the direct effects of temperature, the temperature profiles are some-
what confounded with the TSR, such that a more sophisticated approach is needed other
than comparing one curve with another. This work is in progress and will be reported
later. The temperature profiles in Figures 44 and 49 represent a relatively wide range of
temperature profiles and there are no obvious temperature effects in the data. For example,
October 6 and September 19 had very different temperature profiles with little effect evi-
dent in the chemical profiles.
With peak temperatures below 74° F, however, an effect is discernible and is illus-
trated in Figure 5G. For these runs, the NO oxidation rate is decreased by one-half for a
14° F decrease in daily peak temperature. These runs were all full sun.
A final topic of discussion with respect to outdoor light conditions and the UNC
chamber performance is the speed at which NO is converted to NO2 under outdoor lighting
conditions relative to indoor smog chamber constant light conditions, being considerably
faster under constant light. UNC does not have an indoor chamber with which to run com-
parison studies, so a kinetics model was used to simulate the light and chemical conditions.
The May 7, 1974, run (Figure 40) was chosen and a 52-step mechanism, specific for pro-
pylene, was used to model the data. The details of the mechanism were abstracted from
the Dodge et al. ,9 Niki et al. ,10 and Demerjian et al. n models. Rate constants recom-
mended by Garvin12 were used where available; otherwise those of Dodge et al. were
used. The results are given in Figure 51. Validation of the model is hampered somewhat
by the lack of additional product information in the late afternoon, but the model very
closely adheres to the principles given by Dodge et al. and Demerjian et al. . and it will
serve to illustrate the important features. In Figure 51, the Kj used in the model was a
smooth function instead of the actually "choppy" solar radiation profile and there are dif-
ferences in the NO. NO? profiles in these regions.
* 83
-------
oo
OCT. 8
1 I I I I I I \ I T
DIURNAL VARIATION OF
SOLAR RADIATION (SR) AND
TEMPERATURE (T) DURING EACH OF
THE FIRST DAYS.
0.5
0.4
0.3
n
o
OCT. 6
0.2
I I I I I I I I
0.1
0 1
TIME FROM CROSS-OVER POINT, hours
Figure 49. Comparison of ozone and solar radiation profiles for mix runs.
-------
0.6
0.5
E
j=
0.4
x
<
0.3
0.2
N02 =
NMHC
= 0.20 ppm INITIAL
= 20% NOX INITIAL
= 0.80 ppmC INITIAL (MIX)
1
1
55 60 65 70
TEMPERATURE, °F
Figure 50. Effect of temperature on maximum rate of NO disappearance.
75
The results of operating the model under constant light intensity conditions are
given in Figures 52 and 53. Comparison of Figure 51 with these figures shows that the
early part of the outdoor run is between the runs for constant Kj = 0.1 and 0.2, and the
latter part of the run, from the NO2 peak to the end, agrees fairly well with the run for
constant Kj = 0. 4. These results also imply that if an indoor smog chamber is illuminated
to an "average" level, that is, such that the area under the Kj-time curve is the same as
the area under the outdoor Kj-time curve, high hydrocarbon reactive chemical systems
will make more O3 under the outdoor Kj conditions than the same system would under the
"average" Kj conditions since it is the instantaneous Kj, NO2 product that generates ozone.
Low hydrocarbon, less reactive systems will make less ozone under outdoor Kj conditions
than under indoor K]_ conditions, since by the time the system reaches the NO2 peak, Kj is
decreasing outdoors. * ,
In all of the above models, it was necessary to assume a value near the lower limit
-•<-. -j,
given in Reference 12 for the reaction
H20 + N02 + NO—«-HN02.
The photolysis of HNO2 is considered an important initiation process in most indoor cham-
bers; and the formation reaction above is suspected of having a heterogeneous Character,
and may not be elementary since it shows a second order dependence on water. The rate
used in the model above was 3.5xl
-------
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
5,. 0
P.
u 0.8
e
u
0.7
0.6
0.5
0.4
0.3
0.2
0.1
C3H6
MO DEL PRO FILES
v
SR(K1)_
0 40 80 120 160 200 240 280 320 360 400 440 480 520 560 600 minutes
yOO 8°° 9°° 10°° 1100 iaOO 1300 1400 1500 jgOO 1700 EOT
Figure 51. Comparison of kinetics model and actual data for
May 7,1974.
86
-------
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
E
a.
o_ 0.0
a
u
0.8
0.7
0.6
0.5
0.4
0.3
0.1
0.0
I I I I I I I I I I I I I I, I
CONST K) = 0.1 tnin-3
C3H6
NO
•v N02
CONST KI = 0.3 min-1
C3H6
0 40 80 120 160 200 240 280 320. 360 400 440 4*0 520 560 600
IRRADIATION TIME, min
Figure 52. Model results for constant Kt=0.1, 0.3, May 7, 1974, conditions.
87
-------
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
... °3
o
z
a
o
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
1 M I I
C3H6
I I I I 1 I I
CONST. KI = 0.2 min-1
\
\ N02
CONST. KI = 0.4 min-1
I I I I I 1 I I I I I I I I I
0 40 80 120 160 200 240 280 300 320 360 400 440 480 500 560 600
IRRADIATION TIME, min.
Figure 53. Model results for constant KI = 0.2, 0.4, May 7, 1974, conditions.
88
-------
chamber. The morning humidity in the outdoor chamber is quite high, with dew points
approaching air temperature frequently. If the EPA chamber value is used in the above
model, the NO-NO2 cross-over time is reduced by 240, 83, 40, and 30 min for 0.1, 0.2,
0.3, and 0.4 min (k^ , respectively. It would seem that: (1) the model needs improve-
ment in some unknown area, (2) the UNC chamber exhibits unusually low surface effects,
(3) there are other chamber artifacts that are not clearly understood, or (4) all of the above.
It should be pointed'out that none of the runs that were selected for this presentation
were designed to determine the effects of TSR and temperature and, in most cases, one
side of each dual run has been presented. Much more information is available in the data
that have been examined herein, in terms of the side to side comparisons as well as the day
to day comparisons. For example, the May 19, 1974, run was a test of the effect of constant
NOX, different NC>2 on the outcome of the experiment. The full dual profile data are shown
in Figure 54.
I I I I I II
10 11 12
TIME OF DAY
Figure 54 Concentration-time profiles for dual run in University of North Carolina
chamber, effect of initial N02/NOX, May 19, 1974.
In summary,
1.
-------
3. Low solar radiation in afternoon has a direct effect on O^ generation. The
effect is approximately equivalent to operating under constant low level light
intensity.
4. Peak daily temperatures below 74° F do affect the NO oxidation rate, reducing
it by a factor of 0.5 at 60° F.
5. Models can fit outdoor chamber data if detailed Kj profiles are used.
6. The UNC chamber apparently exhibits very low surface reactivity with respect
to HNC>2 formation.
REFERENCES FOR CHAPTER 5
1. Environ. Sci. Technol. 6:253, March 1972.
2. A Critique of the 1975-1976 Federal Automobile Emission Standards for Hydrocarbons
and Oxides of Nitrogen. National Academy of Science, Committee on Motor Vehicle
Emission Standards.
3. Air Quality Criteria for Nitrogen Oxides. U.S. Environmental Protection Agency.
Washington, D.C. Publication No. AP-84. January 1971.
4. J. Air Pollut. Contr. Ass. 19:929-936,1969.
5. Hydrocarbons in Polluted Air. State Air Pollution Research Center. University of
California; Riverside, Calif. CRC Project CAPA 5, 68. June 1973.
6. Holmes, J. R. et al. Measurement of Ultraviolet Radiation Intensity in Photochemical
Smog Studies. Environ. Sci. Technol. 7(6): 519. 1973.
7. Sickles, J. and H.E. Jeffries. University of North Carolina, Chapel Hill, N.C.
Private communication. 1974.
8. Kamens, R.K. University of North Carolina, Chapel Hill, N.C. Private communi-
cation. 1974.
9. Dodge, M.C., T.A. Hecht, and J.H. Seinfeld. Further Development of Generalized
Kinetic Mechanism for Photochemical Smog. Environ. Sci. Technol. J3:327, 1974.
10. Niki, H., E.E. Daby, andB. Weinstock. Mechanisms of Smog Reactions. Adv.
Chem. 113:16, 1972.
11. Demerjian, K.L., J.A. Kerr, and J.G. Calvert. The Mechanism of Photochemical
Smog Formation. In: Advances in Environmental Science and Technology (Vol. 4) .
J.N. Pitts and R.L. Metcalf (edj. New York, John Wiley and Sons, 1974. p. 1-262.
12. Chemical Kinetics Data Survey VII; Tables of Rate and Photochemical Data for
Modelling the Stratosphere. D. Garvin and R.I. Hampson (ed.). National Bureau
of Standards. Washington, D.C. Report NBSIR 74-430. January 1974.
90
-------
6. VALIDITY AND UTILITY OF SMOG CHAMBER DATA
INTRODUCTORY REMARKS - «. l)imi,ria
-------
etc. Not that these things are of no interest whatever—far from it; simply, they are not and
cannot be of any interest to this meeting. The question to be dealt with here is the question
of what are the relative accuracies and validities of the presently available methods for calcu-
lating control requirements for oxidant abatement.
After having established this context, the two main opposing viewpoints are, in
general terms, as follows: The LACAPCD Viewpoint favors a method (for calculating con-
trol requirements) that relies primarily on smog chamber data and secondarily on aero-
metric data. Such smog chamber data exist and give a fairly complete definition of the
dependence of oxidant on HC and NOX. To circumvent the problem of uncertainty regarding
*
applicability of smog chamber data in the real atmosphere, the LACAPCD method includes
a procedure for empirically relating the smog chamber predictions to real atmosphere
observations. This procedure is the so-called Hamming Transform that Mr. Hamming will
elaborate upon.
The EPA Viewpoint favors a method that relies primarily on real atmosphere data
and secondarily on smog chamber data. The EPA Viewpoint places the emphasis on the
inherent validity of the aerometric data and maintains that in the interest of such
validity, several limitations of the EPA method can be tolerated.
Both viewpoints have their strong and weak points. EPA has been and will continue to
pursue development of both approaches; certainly, there is no commitment for life to one
and the same method.
THE LACAPCD VIEWPOINT (HAMMING TRANSFORM) ~W. Hamming
Summary
Mr. Hamming submitted the following written summary of his talk.
The "Hamming Transform" is a procedure involving smog chamber ozo. t
-------
one
One could choose the midpoint, or the maximum, or even a point between, but
must be consistent. The point on the frequency distribution is not representative of the
day on which the highest ozone was measured, it is to be representative of the whole
frequency distribution at a given NOX-hydrocarbon ratio. In the procedure leading to
the Hamming Transform, the maximum point on the frequency distribution has been used.
For convenience, this point has been related to the maximum ozone in the atmosphere and
in smog chamber studies.
Presentation
In his talk, Mr. Hamming offered the following discussion:
Prior to the District making any claims as to how ozone varied with HC and NOY,
J±
Haagen-Smit showed that although ozone can be formed from hydrocarbon and oxides
of nitrogen in air and sunlight, ozone formation could be squelched if.the hydrocarbon were
in large excess. Similarly, ozone could be reduced as NOX, especially NO, was increased.
In other words there is an NOX/HC ratio that produces the most O3. Further, Stanford
Research Institute demonstrated (1) that maximum eye irritation occured when the ratio
between NOX and hydrocarbons from auto exhaust was just right, and (2) that aerosol
formation was maximized at high hydrocarbon to NOX ratios. Further, Dr. Tuesday of
General Motors was very insistent that NO increases were a way to control ozone formation.
The District's only claim to fame in this field is that under controlled conditions
using typical traffic cycles, auto exhaust was generated (dynometer) and irradiated in
1000-ft^ chambers. The reproduction of all these effects mentioned above was possible
as the NOX/HC ratio changed
1. High residual NO2 was formed at high NOX/HC ratios.
2. At lower NOX/HC ratios there was no residual NO2> but O^ was maximized.
3. At slightly lower NOX/HC ratios, El was maximized.
4. At lower NOV/HC ratios, aldehydes were maximized.
J^
5. At still lower NOX/HC ratios (high HC/NOX), aerosol formation was maximized.
The figures shown in the paper "The Pathway to Clean Air" are derived from those test
runs back in the 1960-63 period.
Some 2 years ago at the National Academy of Science, Dr. Altshuller presented dia-
grams of oxidant-ozone formation that were very similar to the one shown in our paper
mentioned above.
Your own work showing the line of 1/10 ppm O3 for 1 hour corresponds quite well
with ours, considering ours was for instantaneous peaks of 1/10 ppm.
There ought to be no question that the field of NOX-HC that forms ozone is sufficiently
well defined to be useful in the control of ambient air concentrations. If the atmosphere
were defined such that it would under all conditions be to the "left" of the 1/10 line, as in
93
-------
the Dirnitriades paper, ozone greater than 0.08 ppm could not form. This only requires
that the ratio of NOX (in grams per mile) to HC (in grams per mile) be greater than 1,
preferably 2 or 3 (0.2 ppm NOX, 0.5 ppm HC carbomation), a ratio (in ppm) of 4: 10, or
preferably 1:1.
In the paper "Calculations of Atmospheric Hydrocarbon Composition in Central Los
Angeles" it was stated that, from a study of gas chromatographic air samples, there were
only four sources of nonmethane hydrocarbon in downtown Los Angeles (DOLA). (1) auto
exhaust, (2) blowby, (3) evaporation losses, and (4) natural gas (ethane, propane, and
butane). No refinery hydrocarbon emissions were found nor was there any large or consist
ent indication of solvent emissions. Total pollution tonages are not the correct "measure"
to use. The non-methane, non-natural gas fraction of total hydrocarbons is the correct
"measure" or factor to use in terms of concentration or proportional grams per mile emitted. 1
Thus, from the NO frequency distribution, it is simpler to calculate this factor for '
X
hydrocarbons, and this is what has been done. For each year, the actual NOX concentra-
tions and the calculated nonmethane, non-natural gas fraction of total hydrocarbons were
used toegther to describe an atmosphere on the NO -HC-ozone chart, Figure 55. Each
Jt
year's data have some width, but are represented for convenience by a line.
This curve or "pathway" derived from motor vehicle emissions in DOLA slightly ex-
ceeds the maximum found and is thus an "upper limit curve" or "upper bound." All atmos-
pheric data would fall below this curve. Now if this curve can, by control efforts, be
made to cross over the "Dimitriades line" as will occur if the Federal Interim Standards for
auto exhaust are continued until 1980 or if the California Standards for auto exhaust are
made official by EPA until 1980, then, as shown in Figure 55 or 56 the oxidant value of this
"upper bound" curve can only produce less ozone than 0.08 (or if you will less than 1/10
ppm) . This only means that for that year the atmosphere represented by ti line from
zero-zero to the point on the "upper bound" curve could only produce 1/10 ppm or less.
The true Los Angeles atmosphere, being slightly lower in NOX and HC, would produce very
slightly less ozone at Pasadena or Azusa.
It should be clearly understood that the transform only works for one point a year.
It is of no value for day to day forecasting of ozone concentrations. The whole frequency
distribution from zero-zero to the top point (for each year) is considered as a whole and
is represented by one point each year. We only choose the "top" point (the highest one) to
represent the frequency distribution of which one day produced the maximum ozone that
year.
Discussion
Following numerous questions from the audience and explanations from Mr. Hamming,
the procedure recommended by LACAPCD for relating oxidant to the oxidant precursors
was, according to the understanding of the Chairman, Dr. Dimitriades, as follows:
94
-------
CALIF PROG RAM •-
FEDERAL PROGRAM »- —
CHAMBER OZONE MAXIMUM, ppm
.5
1.0 1.5
HYDROCARBONS, IR-ZILB15), ppm
Figure 55. Ozone as a function of HC and NOX, from smog chamber data.
The basis of the LACAPCD procedure is the OX-HONOX relationships derived from
smog chamber studies1 and depicted in Figure 55. The diagrams of Figure 55 are used to
obtain the O3 concentration value predicted by the smog chamber data for HC and NOX con-
centrations equal to the maximum instantaneous concentrations observed in downtown Los
Angeles (DOLA). during 6 to 9 a.m., in a given year. The concentration values for DOLA
are used here because DOLA is the main source area whose air generally transports high
levels of ozone into the main receptor area, Pasadena. Note that these maximum HC and
N0x concentration values are not always values of actually measured concentrations. They
we?e calculated as follows: During the precontrol period 1960-65, the maximum instanta-
neous NOY concentration-defined as the maximum on the frequency distribution plot-in
95
-------
TIME TO N02 PEAK (Tr), hours
PRESENT PROGRAM •
CALIF PROGRAM
FEDERAL PROGRAM
1.0 1.5
HYDROCARBONS, IR-Z (LB 151, ppm
Figure 56. Time to maximum N02 as a function of HC and NOX, from smog chamber data.
DOLA was 1.04 ppm; in 1955-60, the NO -to-HC ratio (as ppm-NOx to ppm-hexane by non-
dispersive infrared, instrument model LB-15A) in automobile emissions was 1.24 for the
exhaust and 0.72 for the entire auto emissions mixture. From these data and from the auto
emission factors derived for 1960-65 and for each of the subsequent years, values for the
expected maximum (not to be exceeded) instantaneous concentrations of HC and NO_ were
Ji
calculated for each of the subsequent years. Note that if in any year, the actually observed
maximum NOX concentration was higher than the calculated one, then this observed NOX
value was adopted instead, and was used to calculate the corresponding maximum HC con-
centration value.
96
-------
From the O3 concentration values read off the "chamber" plots (Figure 55), values
for the O3 concentrations expected in the ambient air are calculated by using the "Hamming
Transform" equations and the diagrams of Figure 56. The Hamming Transform equations were
derived as follows:
The concentration of O3 in air was assumed to be a function of the "chamber" O3 con-
centration, [O3] , as shown by Equation 1.
(1)
- -
air
where kj is a factor greater than 1.0 that adjusts [O3] ch for losses in the chamber due to
unavoidable dilution, and k2 is a factor denoting the additional ozone formation in air as a
result of the continuous influx of fresh emissions as the air moves from the source area to
the receptor area. Df is a diffusion-related factor that is to^be expressed in terms of a
precursor concentration function; in fact, it is assumed that either one of the functions
shown by Equation 2 is valid.
Df = k4[NOx]m [Tr]n = k4[S-HC]m [TR]n (2)
In Equation 2, S is the presumably constant NOx-to-HC (nonmethane, non-natural
gas fraction) ratio in air, and Tr is the time to maximum NO£ in the (smog chamber) irra-
diation of a given HC~NOX mixture sample. Figure 56 is a graphical depiction of Tr as a
function of HC and NOX> as determined in the LACAPCD study.
Equations 1 and 2 are combined to yield Equation 3
[03]ch [NOx)m[Tr]"
The three unknowns , k, m, and n, were determined by using smog chamber data and aero-
metric data. Thus, for each year the maximum NOX concentration was estimated as des-
cribed in the preceding paragraphs and its value was substituted for [ NOX ] . From this
value and from the known HC-to-NO,, ratio — obtained from auto emission factors — the
Ji
maximum HC concentration was estimated and used to obtain the Tr value (from Figure 56) .
[ 03) . denotes the maximum O3 concentration observed in Pasadena or Azusa — whichever
is higher— in the year. Since there are data available for 14 years (1960-1973) , it follows
that there are 14 simultaneous equations that could be used to determine the unknowns k,
m, and n. Nevertheless, because of their distinction (see Figure 57) only 6 year-points
were used. More specifically, this is because a graphical depiction of the 14 year-points
suggests that these points may fit a family of curves as shown in Figure 57 . Of these curves ,
the one best defined is the one defined by 6 year-points (1960. 1965, 1969,1970, 1972,
1973) . However, this curve does not reflect the most unfavorable meteorological conditions.
Such conditions evidently occurred in 1971 when [ O3 ] air/ [O3] ch had its highest value.
Because of this and because the interest here is in relating O3 to O3-precursors for the most
unfavorable meteorological conditions, the 6-point curve was transposed so as to go through
the 1971 year-point. This is equivalent to first calculating values for k, m, and n, based
97
-------
1.9
1.8
1.7
o
ww
3
o*
1.6
1.5
1.4
1.3
1.2
1—I I I I
I I I I I
r~r
K = 1.87
O X
X O
O \
\
\
\K = 1.B2
\
\
\
..o-
\ K = 1.45
\
\
\
\ —I
\
\
\K = 1.27
I I I I I I
[ 1970 71 72 73 74 75 76 77
Figure 57. Variation of [03] a\r/[02\ ch w'tfl Year-
'60 61 62 63 64 65 66 67 68
YEAR
on the 6 year-points (paired into three 2-year averages) , and subsequently adjusting the
k value so as to make the 6-year-point curve go through the 1971 year-point. Thus, the
"Hamming Transform" equation becomes
t°3lair = 1-87 [NO*] 0.222 (4)
[OS'ch [Tr] 0.106
for [O,] denoting maximum instantaneous O_ concentration, or
I°3lair 1.5[NOJ°-222
y. j*" — * /c\
[03Jch [Tr]0.106
for [O^j denoting maximum hourly average O, concentration.
THE EPA VIEWPOINT - B. Dimit riadcs, EPA
The EPA approach to calculating control requirements for oxidant abatement involved
the following steps:
1. Aerometric data from several cities were obtained and used to construct the
so-called "upper limit" curve (Figure 58) that relates the 6 to 9 a.m. NMHC
to the maximum 1-hour oxidant ever observed.
98
-------
2. From the upper limit curve,
the NMHC concentration
value of 0.24 ppmC was de-
rived as representing the
maximum NMHC concentration
consistent with the oxidant
standard.
x
o
X
<
0123
AVERAGE NONMETHANE HYDROCARBONS, ppmC
Figure 58. Upper limit oxidant values in the South
Coast Air Basin as a function of average 6 to 9 a.m.
nonmethane hydrocarbon concentrations, 1971 data.
Oxidant concentrations are maximum 1-hour values
for 12 stations; NMHC concentrations are average
values for 8 stations.
3. Control requirements in a
region, e.g. , Los Angeles,
are calculated from the upper
limit curve and the maximum
1-hour oxidant ever observed
in the region.
The upper limit curve presently
available for the Los Angeles basin was
constructed from data taken at several
points in the basin. These data points
were plotted so as to relate the day's
maximum 1-hour oxidant throughout
the basin to the average of all 6 to 9
a.m. NMHC values obtained in the
various monitoring stations in the basin.
The strong point of this method, according to the EPA viewpoint, is the high degree
of inherent validity that results from use of real atmosphere data. While such validity is
certainly not indisputable, it must be assumed that, conceptually at least, the aerometric
data method should be more valid than any laboratory method.
Weaknesses of the EPA method and the EPA viewpoint regarding these weaknesses
are as follows:
The EPA method has a conceptual limitation in spite of the realistic nature of its data
base. This limitation arises mainly from the empirical nature of the upper limit curve. Be-
cause of this empirical nature, the upper limit curve, or any other empirical relationships
of aerometric data, cannot be used with confidence to predict future air quality. To ex-
plain, the upper limit curve does not necessarily depict the dependence of oxidant on HC.
The wide range of NMHC values shown in the plot reflects the variation in meteorological
conditions affecting dispersion-e.g. , wind speed-and not variation of the NMHC factor
alone. Therefore, the upper limit curve, properly interpreted, could very well depict
the dependence of oxidant on wind speed and not necessarily the dependence on NMHC.
From a control standpoint, the upper limit curve, if interpreted properly, could mean that
in order to reduce the oxidant down to the standard, the meteorological dispersion condi-
tions throughout the day should be controlled in such a way that the 6 to 9 a.m. level of
99
-------
NMHC is reduced down to 0.24 ppmC or less. Given this interpretation, an upper limit
OX/NOX curve or O /CO curve would have just as much validity as the OX/HC curve,
since the 6 to 9 a.m. NOX or 6 to 9 a.m. CO are just as good indicators of atmospheric
dilution as the 6 to 9 a.m. HC.
In summary, then, the upper limit OX/NMHC curve depicts either the dependence of
oxidant on the dispersion factor or the dependence of oxidant on HC. The EPA viewpoint
is that the upper limit curve can be assumed to depict the dependence of oxidant on NMHC.
The EPA assumption is supported by smog chamber data. Such data indicate that in
simulated current atmospheres, in which the HC/NOX is 10: 1 or higher, reduction of HC
or of both HC and NOX (keeping HC/NOX constant) will cause reduction in oxidant (Figure
59). This evidence in essence means that, insofar as the effect on ambient oxidant is con-
cerned, the reduction of HC through emission control is, qualitatively at least, equivalent
to reduction of the HC level through dispersion.
The other major limitation of the EPA
method for calculating control require-
ments is that the method ignores the NOX
role in the oxidant formation process.
The problem here is simple„ and has
been recognized only thanks to the
smog chamber data available. If uni-
lateral control of HC is adopted as the
approach to oxidant control, then
the HC-to-NOx ratio will change with
time to lower values resulting in
stronger inhibition by NO. Such
enhanced inhibition, in turn, will
permit somewhat higher levels of
HC in the atmosphere than the upper
limit curve dictates. In conclusion
then, the maximum NMHC level con-
sistent with the oxidant standard
should not be 0.24 ppmC but some-
what higher.
0.8
0.7
0.6
0.5
>< 0.4
o
I 0.3
0.2
0.1
0.0
1
i
i r
i
1.0 2.0 3.0
NONMETHANE HYDROCARBON, ppmC
4.0
Figure 59. Smog chamber data on dependence of
oxidant on NMHC under constant NOX or HC/NOX
conditions.
EPA admits that this limitation of the EPA method is a real one. However, since the
aerometric data are not sufficiently abundant to permit quantification of the NOX role at
this time, EPA feels that, for the present, this limitation should be tolerated as the lesser
of the two evils . The other evil, of course, would be to use smog chamber data as the
primary basis of the entire control strategy.
100
-------
There are other limitations of the EPA method for calculating control requirements;
however, the ones already discussed are the major ones and constitute the main points of
disagreement between EPA and its critics.
re-
To complete the presentation of the EPA veiwpoint on this issue, the following
marks are offered in regard to the proper use of the smog chamber data.
If smog chamber data are to be used to devise an oxidant control strategy, then,
according to EPA viewpoint, such data should be used following the procedure illustrated
in Figure 60. First, the smog chamber data are used to construct the 0.08-ppm O3 isopleth
(Figure 60) . From this isopleth it can be deduced that in order to achieve the oxidant
standard, ambient morning concentrations of HC and NOX should be controlled so as to be
within the shaded area of the graph. Here it should be pointed out that HC must be con-
trolled somewhat more than what the minimum requirement is in order to ensure achieve-
ment of the oxidant standard even when the NOX falls below its maximum value. To explain,
EPA believes, and there are data to support such belief, that for HC at its maximum value,
the NOX may vary downward from its maximum value. Assuming that this HC/NOX varia-
tion is real, it can be seen then that HC must be controlled somewhat more than this dia-
gram requires. More specifically, HC must be controlled down to approximately 0.2 ppmC
or even less. In conclusion then the smog chamber data, if used according to this EPA
procedure, do not disagree with the dictates of the aerometric data analysis.
MAX. ALLOWABLE [NOX]
1.0
2.0 3.0
NMHC,ppmC
4.0
5.0
Figure 60. Equal response lines representing all combinations of NMHC and
NOv concentrations corresponding to 0.08 ppm 03 of oxidant.
/\
The following final remarks reflect this speaker's viewpoint, not necessarily adopted
by EPA, regarding the relative validity and utility of the smog chamber and aerometric
data.
While both the smog chamber method and the aerometric data method (for calculating
control requirements) lack in validity at this time, the smog chamber method has a greater
101
-------
potential for improvement than thought earlier. The aerometric data, in this speaker's
judgment, have a certain degree of inherent validity because they are real atmosphere data.
However, these data are also inherently limited in that they cannot be used to predict future
levels of photochemical air quality. The fact that such data are taken from past and present
day atmospheres simply makes them inappropriate for such use. For the extremely simple
cases, e.g., the case of CO pollution, a trend analysis of atmospheric data may give valid
predictions. However, the photochemical oxidant problem, even though it may be simpler
than other photochemical pollution problems, still is too complex to be adequately defined
and analyzed by aerometric data. The effects of HC/NOX, HC composition, transport, and
diurnal emission patterns, as well as the interactions among these effects, simply cannot be
delineated based on aerometric data alone.
It is submitted here as a strong recommendation that (1) it should be set as a goal
to develop a chamber methodology to the point that it would be possible to design photo-
chemical pollution control strategies based on smog chamber data, and (2) aerometric data
should be used as a guide in development of valid smog chamber methodology.
To obtain such aerometric data, it is recommended that field studies be specially
designed and conducted to provide a few points in the true cause-effect relationship be-
tween HC, NO , and oxidant. Such points could then be used to validate or develop smog
*Ah
chamber methodology.
The "Hamming Transform" (discussed in preceding presentation) illustrates an
alternative way of using atmospheric data to develop smog chamber methodology. However,
I must reject this method as being inherently unreliable. The empirical nature of the
correlation between smog chamber oxidant and observed ambient air oxidant makes this
correlation inappropriate as a predictor of future air quality. Thus, this "transform" has
the same disadvantages as those of the upper limit curve or any empirica* relationship of
atmospheric data, and it has the additional disadvantage that it has been derived from ex-
tremely few points. For this reason the objective here can be accomplished more reliably
through specially designed field studies.
CALIFORNIA AIR RESOURCES BOARD VIEWPOINT* -
«
/. Holmes, F. Bonamassa, CARB
Introduction
The results of experiments conducted in a "new generation" of environmental cham-
bers are currently becoming available. These chambers are characterized by radiation
sources that closely approximate the UV portion of the solar spectrum as observed at the
*This presentation is based on a report entitled "Application of the Results of Recent Environ-
mental Chamber Studies to the Control of Photochemical Oxidant," which is a preliminary
draft. It has not been formally reviewed or released by CARB and should not be considered
as representing the Board's policy. It has been given limited circulation for comment on its
technical soundness and policy implications.
102
-------
earth's surface and by clean, chemically inert interior surfaces that have a very low poten-
tial for decomposition of the ozone formed in the smog reaction. In this report, we sum-
marize the results of two important studies and examine the implications of these results
for the oxidant control strategy options in the South Coast Air Basin.
Environmental Chamber Studies
In this section, using some of these new data (which are still incomplete), we
reexamine the relationships between hydrocarbon and NOX concentrations with respect to
ozone buildup rate, ozone concentrations, and ozone dosage. Of particular interest in this
reexamination are the following aspects of the problem:
1. Ozone concentrations achieved during long (10-hour) irradiations.
2. Dependence of ozone concentration on initial NOX concentration for long
irradiation times.
3. Effect of dilution at constant (HC) / (NOX) ratio on ozone formation at various
ratios.
4. Investigation of conditions under which initial NOX concentration limits ozone
potential.
The results of earlier chamber studies have defined the general relationship between
oxidant formation and hydrocarbon reactivity and concentrations. For those concentrations
of interest in the ambient atmosphere of the South Coast Air Basin (SCAB), any reduction
in initial hydrocarbon concentration results in an increase in the time required to reach an
ozone maximum and in a decrease in the maximum value attained, for a specified initial
concentration of NOX and period and intensity of UV irradiation. There are, of course,
differences in reactivity among hydrocarbons. This question is beyond the scope of the
present discussion, however. In general, olefins are more reactive in promoting ozone
buildup, aromatics and paraffins are less reactive, and methane and acetylene are, in
effect, unreactive. Carbon monoxide behaves as a very low reactivity hydrocarbon, inso-
far as its capability for promoting ozone buildup. Aldehydes and ketones behave as mod-
erately reactive hydrocarbons in this regard.
The question of the effect of NOX control on ozone buildup in the South Coast Air
Basin is not a simple one. In the complex series of chemical reactions associated with
photochemical smog, NO2 serves as both an initiator and terminator of the chain reactions
that cause (1) conversion of NO to NO2, and (2) the buildup of ozone and other oxidants.
Thus. for each level of nonmethane hydrocarbon there is an "optimum" level of NOX that,
in a static system, leads to maximum ozone dosage, for a specified period and intensity of
UV irradiation.
Over most of the concentration range applicable to the ambient atmosphere, the
results of environmental chamber studies clearly demonstrate that ozone dosage is a function
of both initial hydrocarbon and NOX concentrations and their ratio. Data relating initial
103
-------
hydrocarbon and NO concentrations and resultant oxidant concentrations can be presented
3t
in several different ways, illustrating different aspects of the relationship. The simplest
representation is the reaction profile, in which the concentration versus time behavior
of the reactants and products are plotted on a two-dimensional graph. For the purposes of
discussion, it is convenient to consider the overall process of photochemical smog formation
to occur in three more or less well defined stages, each characterized by a different mani-
festation:
Tygi c al _c har ac ter i s tic s
Conversion of NO to NO£
Rapid buildup of Oo accompanied by formation of
gas-phase aldehydes and organic aerosol from
olefins
Phase III Continued buildup of O^, but at an appreciably
slower rate than in II, accompanied by formation
of nitrate aerosol from NO2
These stages are fairly sharply defined in static chamber experiments. In the
atmosphere, there is appreciable overlapping of stages because fresh reactants are con-
tinually being injected into the reacting air mass.
Figure 61 shows the reaction profile for a typical smog chamber' run (hydrocarbon
data are not shown) . The initial reaction mixture contained 0.36 ppmC nonmethane hydro-
carbon (NMHC) consisting of a surrogate mixture simulating auto exhaust and 0.05 ppm
NOX (an initial ratio of NMHC to NOV about 6.9); the mixture was irradiated for 6 hours.
•*• t
Note that the rate of increase of the ozone concentration at the end of the run suggests that
the system has appreciable potential for further ozone formation.
Figure 62 shows the reaction profile of a similar system irradiated for 10 hours.
Note that ozone continues to build up at nearly a constant rate over the entire 4 hours of
additional irradiation, reaching a level of 0.29 ppm, an increase of more than 50 percent
over the value attained in the shorter run, 0.18 ppm.
These data were obtained by investigators at the Statewide Air Pollution Research
Center, UC Riverside, under contract to the GARB . 3 The results of nearly 50 similar cham-
ber runs demonstrate that, even at the lowest initial concentrations of hydrocarbon and
NOX, appreciable potential for ozone buildup remains in a polluted air mass after 6 hours
of irradiation at full solar intensity.
Complete experimental data on the effect of initial NOX concentration upon ozone
buildup for long irradiation times are not yet available. However, estimates based upon
an empirical extrapolation of the 6-hour SAPRC data^ to longer irradiation times suggest
that even at the lowest initial concentrations studied to date, 0.08 ppmC NMHC and 0.05
104
-------
0.2
2
z
UI
0
• NITROGEN DIOXIDE
• NITRIC OXIDE
A OZONE
(HC)0 = 360 ppbC
(N0)0= 53ppb
NO
0 2 4 6
IRRADIATION TIME, hours
Figure 61. Reaction profile for SAPRC Run No. 42E, 6-hour irradiation.
ppm NOX can lead to nearly 0.10 ppm ozone when the irradiation period is extended to 10
hours at full solar intensity. * These results are summarized in Table 15.
The data from Table 15, though still incomplete, have been used to generate crude
ozone isopleths for 6-hour and extrapolated 10-hour irradiations (Figures 63 and 64). In
addition, the final ozone buildup rates as a function of (NMHC)0 and (NOX)Q have been
plotted in Figure 65.
"Data cited by Leighton4 for Los Angeles during the period around the summer solstice indi-
cate that the total radiant flux between the hours of 6 a.m. and 6p.m. (solar time) corre-
sponds to 9 hours of irradiation at the maximum (solar noon) intensity.
105
-------
0.3
0.2
o
u
0.1
• NITROGEN DIOXIDE
• NITRIC OXIDE
A OZONE
(HC)0 = 350ppbC
(N0x)0 = 64ppb
(C0)0 -1 ppm
—1
0.31 EXTRAPOLATED
AJ FROM 6-hour
IRRADIATION
/
/
/
2468
IRRADIATION TIME, hours
Figure 62. Reaction profile for SAPRC Run No, 48E, 10-hour irradiation.
10
12
The 6-hour data yield ozone isopleths generally in agreement with Dimitriades' ex-
tensive study of irradiated auto exhaust. ' In particular, the SAPRC data reinforce
Dimitriades1 conclusion that initial levels of NMHC and NOX of 0.80 and 0.33 ppm, respect
tively, will not exceed the proposed levels of 0.08 ppm ozone for 1 hour and 0.25 ppm NC>2
for 1 hour. This conclusion applies only to static systems irradiated for 6 hours, however.
The ozone isopleths for the extrapolated 10-hour irradiations (Figure 64), assuming
the procedure used to obatin them is reasonably valid, suggest several interesting, al-
though tentative, conclusions:
106
-------
Table 15. OZONE FORMATION. SAPRC GLASS r.HAMRFR
Run
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
(N0x)o,
ppm
0.286
0.362
0.362
0.167
-
0.095
0.007
0.082
0.047
0.038
0.103
0.160
0.025
0.283
0.276
0.009
0.085
0.088
0.169
0.285
0.285
0.268
0.230
0.269
0.088
0.171
0.191
0.025
0.198
0.191
0.052
0.050
0.052
0.105
0.031
0.125
0.050
0.010
0.064
(NMHC)o,a
ppmC
1.92
1.97
1.96
2.08
-
2.11
2.39
1.96
2.03
2.18
0.56
0.67
0.60
2.28
0.73
0.81
0.85
0.61
0.70
0.64
0.74
0.60
0.64
-
0.63
0.67
0.58
0.77
0.63
0.74
0.73
0.72
0.36
0.37
0.43
0.43
0.08
0.37
0.35
jNMHCjo
(N0x)o
6.7
5.4
5.4
12.5
-
22.2
341.4
23.9
43.6
57.4
5.4
4.2
24.0
8.1
2.6
90.0
10.0
6.9
4.1
2.2
2.6
2.2
2.8
-
7.2
3.9
3.0
30.8
3.2
3.9
14.0
14.4
6.9
3.5
13.9
3.4
1.6
37.0
5.5
6-hr 03,
ppm
0.44
0.26
0.26
0.48
_
0.41
0.14
0.39
0.31
0.20
0.30
0.24
0.16
0.33
0.12
0.14
0.35
0.27
0.19
0.07
0.06
0.06
0.06
-
0.25
0.17
0.12
0.16
0.10
0.13
0.25
0.20
0.18
0.14
0.11
0.10
0.06
0.10
0.18
d(Q3)(5.5-hr).
dt
pom hr-1
0.091
0.073
0.074
0.057
_
0.019
0.021
0.024
0.020
0.017
0.056
0.068
0.024
0.075
0.035
0.019
0.031
0.047
0.044
0.029
0.019
0.018
0.022
-
0.041
0.043
0.030
0.022
0.036
0.035
0.020
0.015
0.031
0.040
0.020
0.029
0.014
0.030
0.043
Est.b of
10-hr 03
ppm
0.71
0.48
0.48
0.65
.
0.47
0.20
0.46
0.37
0.25
0.47
0.44
0.23
0.55
0.22
0.20
0.44
0.41
0.32
0.16
0.12
0.11
0.13
-
0.37
0.30
0.21
0.23
0.21
0.23
0.31
0.24
0.27
0.26
0.17
0.19
0.10
0.19
0.31
Surrogate mixture.
Ozone concentration
tration at end of 6
after 10 hours estimated by adding 3 x final 03 rate
-hour irradiation.
to 03 concen-
107
-------
a
0.35
0.30
0.2S
0.20
0.15
0.10
0.05
B.O.M. STRATEGY WITH
25% NO DEFICIT
DILUTION AT
CONSTANT RATIO
0.25
0.50
0.75
1.00 1.25 1.50
INITIAL NMHC, ppmC
1.75
2.00
2.25
2.50
Figure 63. Ozone isopleths from 6-hour irradiations of HC-NOX mixtures, SAPRC Runs 10 - 48 (E). [03] values
are final 6-hour ozone concentrations (ppm).
1. Undiluted systems ' suggested that the air quality standard for NMHC and
NOX will produce 0. 12 to 0. 15 ppm maximum ozone and exceed the 0. 08 ppm
standard for oxidant for well over 1 hour .
2. Very high levels of ozone (>0.5 ppm) can be produced from much lower con-
centrations of initial reactants than believed earlier .
3. The "optimum" ratio of NMHC to NOX for ozone formation seems to be much
lower — more nearly 7: 1 than 10: 1 — for extended periods of irradiation.
The implications of the first item are obvious: if realistic worst-case irradiation
times are used in chamber experiments , the permissible level of NMHC consistent with the
0.08 ppm oxidant standard is lower than that suggested by Dimitriades, perhaps 0.65 ppm
NMHC.
On the other hand, the SAPRC data suggest that the N©2 maximum at these levels
and with an initial ratio of NO to NO2 of about 9 to 1 is about 65 percent of the initial NOX,
so the allowable level of NOX consistent with the 0.25-ppm air quality standard for
may be somewhat higher, perhaps as high as 0.37 ppm.
108
-------
0.35
0.30
0.2S
X
o
0.10
0.05
S.A.P.R.C. DATA(10HR)&
B.O.M. STRATEGY WITH
25% NO DEFICIT
[03] = 0.10 9 S.A.P.R.C. STUD\
0.20 I 0.30
B.O.M. DATA &
B.O.M. STRATEGY
25% NO DEFICIT
DILUTION AT
CONSTANT RATIO
RATIO
0.25
0.50
0.75
1.00 1.25 1.50
INITIAL NMHC, ppmC
1.75
2.00
2.25
2.50
Figure 64. Ozone isopleths from 10-hour irradiations Of HC-NOxmixture^ SAPRCJ^uns 10-48 (E). [63] values
are final 10-hour ozone concentrations plus 3 times the final rate of ozone formation.""'
maximum corresponding more closely to the value of 75 to 80. percent
'
It should be pointed out, however, that the initial NO/NC^ ratio of 9 to 1 used in the
SAPRC studies is unrealistically high., Aerometric data indicate that the initial ratio in
source areas of the SCAB during smog episodes can remain below 2 to 1 throughout the
6 to 9 a.m. NO emissions peak. Under these initial conditions, chamber experiments
would exhibit a
of the initial NOX suggested by Dimitriades.
The second conclusion suggests that the high ozone levels observed in the eastern
portions of the SCAB can be accounted for, at least in part, by the long-term irradiation
of polluted air masses moving from the urban-industrial source areas to receptor areas in
San Bernardino and Riverside Counties over the span of 8 to 10 hours between the peak
emission period and the time of the observed oxidant maximum.
If the third conclusion, regarding the "optimum" ratio of NMHC to NOX for ozone
formation under conditions of longer-term irradiation, is correct, this suggests that the
ambient levels of NMHC and NOX at many locations in the SCAB are currently at or near this
optimum value. This, in turn, suggests that control strategies requiring reductions in the
emissions of either NMHC or NOX (or both) are likely to lead to a reduction of oxidant dosage.
109
-------
1
0.10
0.05
0.25 0.50 0.75 1.00 1.25 1.50
INITIAL NMHC, ppmC
1.75
2.00
2.25
2.50
Figure 65. Ozone rate isopleths from 10-hour irradiations of HC-NOX mixtures, SAPRC Runs 10 - 48 (E).
A recent analysis presented by Hamming et al. 1 suggests that any reduction in NOX
alone would lead to an increase in oxidant. This is somewhat misleading; such a strategy
would lead to somewhat more rapid onset of ozone buildup, since Phase I of the smog forma-
tion process would be shortened, but, under similar conditions of mixing, irradiation, and
dilution, the maximum hourly average ozone concentration achieved in a given air mass
would be lowered, and total ozone dosage would be reduced over most of the trajectory of
the air mass.
The extensive study of Dimitriades ' of ozone buildup in irradiated auto exhaust at
somewhat higher levels than in the SAPRC study at various NMHC to NOX ratios provides
another useful data base for determining an oxidant control strategy for the SCAB. Here
again, however, the irradiations extended over only 6 hours at somewhat less than full
solar intensity, so the possibilities for long-term ozone buildup were not assessed in his
analysis of the data. The original reaction-profile data have not been made available to
GARB, so empirical extrapolations to a 10-hour irradiation cannot be made.
The principal result of Dimitriades' study is the familiar plot of "equal response" lines
(Figure 60) defining ranges of initial NMHC and NOX concentrations at which the oxidant
standard is not exceeded.
110
-------
In addition to the short irradiation time, several other problems must be taken into
consideration when one attempts to interpret these results in terms of a control program.
DimitriadesS has criticized Hamming's^ application of these results to produce a timetable
for.the achievement of the oxidant standard in Los Angeles County. Moreover, Dimitriades
has not taken into account the effects of dilution on the reactivity of an air mass containing
an initial burden of NMHC and NOX corresponding to the target levels he proposes.
Dimitriades1 data are presented in two different formats in Figure 66. In proposing
levels of 0.80 ppm NMHC and 0.33 ppm NOX he is attempting to take full advantage of the
7
so-called inhibitory effect of NOX while just maintaining the air quality standard for NO2.
i
X
E
IU
CD
M 80 '
X
o
PROPOSED
NMHC-NOX
LEVEL
2.5
20
0.2
Figure
66. Oxidant-dosage reactivity of exhaust as a function of NOX at various HC: NOX ratios.
Ill
-------
It is apparent from Figure 66 that dilution at constant ratio of an air mass containing
the correct levels of NMHC and NOX will increase the oxidant-dosage reactivity of the sys-
tem to a point at which the oxidant standard will be exceeded unless further dilution occurs.
Dimitriades points out a further problem associated with day to day variations in the
relative emissions of NMHC and NOX. Success of a strategy based on the inhibitory effect
of NO requires that some minimum amount of NO be emitted to prevent or at least delay ozone
buildup. This is not a trivial problem, since the abnormal traffic conditions that produce
maxima in NMHC (stop and go, low average speeds) coincide with minimum NOX emissions
from autos.
Clearly, a strategy that relies upon the inhibitory effect of NO must take into account
the effects of both dilution of the air mass and the possibility of a deficit in NO emissions
altering the ratio of NMHC to NOX.
Conclusions
The discussion in the preceding section points out that an oxidant control strategy
based on the inhibitory effects of NO, such as that proposed by Dimitriades, ' must be
modified to'take into account the effects of dilution, NO deficit, and long-term irradiation.
First, one must determine the effects of extended irradiation. Complete experimental
data are lacking, so we have generated the empirical ozone isopleths in Figure 64 to repre-
sent such conditions. Second, the maximum allowable NOX concentration consistent with
the 0.25-ppm 1-hour maximum for NO£ must be determined. Both the Bureau of Mines and
SAPRC studies suggest that this is about 0.35 ppm NOX-
Third, one must make some sort of worst-case estimate of the NOX deficit expected
in the day to day variations in NMHC and NOX emissions. Auto emissions data suggest that
a value of 25 percent is not unreasonable, although a thorough and continuing check of
air quality data will be required to determine the validity of this estimate.
Finally, a maximum value of initial NMHC consistent with this lower value of NO,,
• .A.
must be determined from the isopleths in Figure 64 or, when they become available for the
SAPRC data, oxidant dosage plots such as in Figure 66. The isopleth plot, Figure 64, yields
a maximum allowable initial concentration of NMHC of about 0.60 ppmC.
A similar estimate using the Bureau of Mines data yields a maximum value of about
0.50 ppmC of NMHC. This lower value reflects the higher yield of NO2 reported by Dimi-
, . , 5,6
triades.
The effects on reactivity of dilution at constant ratio from the various initial mixtures
shown in Figure 64 can be estimated from the relationship between the isopleths and lines
drawn to the origin. The mixture corresponding to the 10-hour SAPRC data, using the
Bureau of Mines strategy with a 25 percent NO deficit, goes through a reactivity maximum
corresponding to about 0.14 ppm oxidant at a dilution of about 2 to 1. Similar dilution ef-
fects are predicted for the other mixtures discussed above.
112
-------
On the basis of these considerations, it is our best judgment that ambient levels of
NMHC and NOX cannot exceed approximately 0.60 ppmC and 0.35 ppm, respectively, if
both a standard of 0.25 ppm NO2 (1-hour maximum) and a revised oxidant standard of 0.12
ppm are to be achieved in the South Coast Air Basin.
Unresolved Questions
The analysis and conclusions presented here do not take into account several impor-
tant factors that have an influence on the ultimate levels of NMHC and NO required to
achieve the oxidant standard. Briefly, these are:
1. The effects of carry-over or "aged smog" on the validity of the proposed
limits on 6 to 9 a.m. emissions, particularly with respect to fraction of initial
NOX appearing as NO2-
2. The effect, if any, of downward mixing of ozone trapped aloft overnight on
ground level concentrations of ozone the following day.
3. The efficiency of conversion of NO emissions to nitrate aerosol under the
X.
proposed conditions for the achievement of the oxidant standard and the re-
lated health-effects question.
All of these questions are currently under investigation by CARB and other agencies.
Our current view of the oxidant control problem may need to be altered when the results
of these and other studies in progress become available.
SMOG CHAMBER STUDIES OF POLLUTANT PRECURSOR RELATIONSHIPS -
B. Dimitriades, EPA
Continuing and recently initiated photochemical pollutant-precursor relationship
studies in EPA are as follows:
1. Oxidant-precursor Relationships and Their Dependence on Pollutant Transport
Conditions—The study calls for smog chamber testing of synthetic and auto
exhaust HC-NO-i mixtures under conditions similar, to those in moving air
Jv
masses, namely, continuous dilution, continuous injection of reactants,
^prolonged irradiation, and repeated irradiation. Objective is to obtain and
compare OX/HC/NOX relationships for simulated transport and nontransport
conditions.
2. Oxidant-precursor Relationships and the Impact of HC Control on Such
Relationships—The study calls for smog chamber testing of synthetic and
auto exhaust HC-NOX mixtures simulating atmospheric pollutant makeup before
and after application of HC control. Objective is to obtain and compare OX/HC/
NOX relationships for simulated precontrol and postcontrol atmospheres.
3. Sulfate-precursor Relationships in Area-wide Systems and in Roadway Systems
—The study calls for smog chamber testing of synthetic and auto exhaust
113
-------
mixtures under conditions similar to those in area-wide atmos-
pheres and roadway atmospheres. Objective is to determine the rate of SC>2
oxidation under the conditions of the study, and to explore the effects of
factors such as the HC, NOX, and SO- reactant concentrations, humidity, HN3,
and primary aerosol, upon such rate.
4. Photochemical Aerosol-precursor Relationships—The study calls for smog
chamber testing of synthetic HC/NOX/SO2 mixtures under conditions similar
to those in urban and rural atmospheres. Objective is to determine the effect
of factors such as reactant concentration, humidity, NH3, and primary
aerosol on photochemical aerosol formation.
5. NO2-precursor Relationships and the Impact of HC Control on Such Relation-
ships—The study—nearly completed—involves smog chamber testing of
synthetic HC/NOX mixtures in an outdoor smog chamber. Objective is to
determine the dependence of photochemically formed NO£ on reactant NOX
and HC, and to explore advantages and disadvantages of outdoor operation of
smog chambers.
6. Smog Chamber Studies for Photochemical Model Development—The program
involves modeling of smog chamber data taken with a variety of smog chambers
and on a variety of HC/NOx and HC/NOX/SO_ systems.
REFERENCES FOR CHAPTER 6
1. Hamming, W.J., R.L. Chass, J.E. Dickinson, andW.G. MacBeth. Motor Vehicle
Control and Air Quality: The Path to Clean Air for Los Angles.. Los Angles County
Air Pollution Control District, Los Angles, Calif. (Presented at 66th Annual Air
Pollution Control Association Meeting. Chicago. June 24-28, 1973. Paper No. 73-73.)
2. Calculations of Atmospheric Hydrocarbon Composition in Central Los Angeles.
3. Pitts, J.N. et al. Quarterly Progress Report, GARB Contract No. 3-017. Statewide
Air Pollution Research Center, University of California, Riverside, Calif. December
31, 1973.
4. Leighton, P.A. The Photochemistry of Air Pollution. New York, Academic Press,
1961. p. 31.
5. Dimitriades , B. On the Function of Hydrocarbon and Nitrogen Oxides in Photo-
chemical-smog Formation. U.S. Bureau of Mines. Washington, D.C. Document
RI-7433. 1970.
6. Dimitriades, B. U. S. Environmental Protection Agency, Research Triangle Park,
N.C. Private Communication.
7. Glasson, W.A. and C.S. Tuesday. Environ. Sci. Technol. 4: 37, January 1970.
114
-------
7. CONCLUSION
B. Dimilriades
A conference was conducted by EPA at its Research Triangle Park. North Carolina,
Environmental Research Center, for the purpose of reviewing present status and utility
of smog chamber methodology. Specific objectives of the conference were to review on-
going smog chamber studies sponsored by EPA, to assess and explain performance dif-
ferences among existing smog chambers, to identify methodology deficiencies, and to discuss
the utility of the smog chamber method in development of photochemical pollution control
strategies.
Smog chamber performance was examined in the light of recent experimental evidence
that showed that disagreement in test results among different chambers must be caused,
partly at least, by differences in chamber design. Thus, intensity and spectral character
of radiation, surface-to-volume ratio, and nature and condition of wall surface were
reported to significantly affect chamber measurements. Increased photolysis of HONO and
aldehydes was offered as a mechanistic explanation of the observed enhancive effect of
low wavelength radiation upon reactivity manifestations. The existence of wall-material
effects was interpreted to suggest that heterogeneous reaction steps are important and,
hence, should be identified and included in the reaction mechanism models. Also, such
effects seem to be minimized in large chambers (several hundred cubic feet) made of Teflon
film.
The analytical methods commonly used in smog chamber experimentation were
reported to lack in several respects. Methods for NC>2 measurement suffer from as yet
unresolved interference problems. Thus, the Saltzman method was shown—but not
unequivocally proved—to suffer from 03 interference, whereas the chemiluminescence
method was found to respond to several nitrogen compounds other than NC>2 (e.g., PAN,
alkyl nitrates, nitric acid) . Methods used for photochemical sulfate measurement were the
barium chloranilate and the barium perchlorate methods; the chloranilate method seemed
to be somewhat inferior. In situ measurement by infrared spectroscopy was reported
to be feasible and useful in smog chamber studies, especially for identification and measure-
ment of unstable species. Measurement of nonmethane hydrocarbon using certain com-
mercial instruments was reported to be inaccurate, perhaps because of improper calibra-
tion. Finally, recommendations were made to standardize among EPA contractors and
grantees (1) calibration procedures of instrumental methods, (2) statistical expressions
of accuracy and precision of smog chamber measurements, and (3) definitions of chemical
reactivity manifestations.
Evidence regarding chamber background reactivity was summarized and interpreted
to suggest that trace levels of (nonmethane) organic and NOX contaminants (a few parts
115
-------
per hundred million) could cause, upon irradiation, significant ozone accumulation and
oxidation of SO£ into sulfate. The evidence suggested, furthermore, that in such background
activity, wall surface also has a role other than merely providing an interface for adsorption-
desorption phenomena.
The issue of comparability of smog chamber atmosphere and real atmosphere was
examined using two approaches. By one approach, comparability should be established
using real atmosphere and smog chamber data on the cause-effect relationship between
ozone and ozone precursors. By the other approach, comparability should be assumed,
provided the dynamic character of real atmosphere is adequately simulated in the smog
chamber. Within the context of this latter approach, the performance of a dual, Teflon-
made smog chamber operated outdoors was discussed and reported to be promising.
The validity and utility of smog chamber data were examined in connection with a
case study of the oxidant problem in California's South Coast Air Basin (SCAB) . The EPA
and Los Angles County viewpoints and methods for calculating control requirements were
presented and contrasted. New smog chamber data generated at University of California1 s
Statewide Air Pollution Research Center under California Air Resources Board contract
were interpreted to derive air quality standards for HC and NOX, consistent with the ozone
and NC>2 standards. EPA prefers aerometric data—rather than smog chamber data—as the
primary basis for deriving control requirements for oxidant abatement. Nevertheless, EPA
is committed to pursuing all three investigative methods—aerometric data analysis, smog
chamber method, modeling—in the effort to develop control strategies based on sound and
completely scientific evidence.
The conference made it clear that in the study of photochemical pollution problems,
smog chamber experimentation is indispensable for studying the photochemical pollutant
formation process and for providing estimates of precursor control requirements. The
conference suggested also that the present analytical measurement problems are neither
critical nor insurmountable. Furthermore, the limitations of the smog chamber method
due to lack of inherent validity are not necessarily prohibitive. It appears that with further
research the smog chamber method could be developed to the point that smog chamber data
would be equally or more valid than data taken by any other method. In the light of such
a promise, EPA intends to sustain a smog chamber study program directed to several photo-
chemical pollution problems, including the problems of ozone, NC>2, photochemical aerosol,
sulfates, and nitrates.
116
-------
LIST OF ATTENDANTS
A. P. Altshuller
Environmental Protection Agency
National Environmental Research Center
Research Triangle Park, NC 27711
Dick Angus
Environmental Protection Agency
Durham, NC
Stephen Aust
Bureau of Air Quality Control
Dept. of Health and Mental Hygiene
Environmental Health Administration
610 N. Howard Street
Baltimore, MD 21201
Joseph Behar
Environmental Protection. Agency
P.O. Box 15027
Las Vegas, NV 89114
J. J. Buf alini
Environmental Protection Agency
National Environmental Research Center
Research Triangle Park, NC 27711
Marijon Bufalini
Environmental Protection Agency
National Environmental Research Center
Research Triangle Park, NC 27711
Peter Connell
Lathner Hall
University of California
Berkeley, CA
P.M. Coving ton
EPA Region IX
100 California Street
San Francisco, CA 94111
Kenneth Demerjian
Calspan Corporation
4455 Genesee Street,
Buffalo, NY 14221
P.O. Box 235
M.C. Dodge
Environmental Protection Agency
National Environmental Research Center
Research Triangle Park, NC 27711
G. J. Doyle
Department of Chemistry
University of California
Riverside, CA 92502
A. H. Ellison
Environmental Protection Agency
National Environmental Research Center
Research Triangle Park, NC 27711
C. E. Feigley
Dept. of Environmental Science
and Engineering
School of Public Health
University of North Carolina
Chapel Hill, NC 27514
Don Fox
Dept . of Environmental Science
and Engineering
School of Public Health
University of North Carolina
.Chapel Hill, NC 27514
Bruce Gay
Environmental Protection Agency
National Environmental Research Center
Research Triangle Park, NC 27711
Sydney Gordon '
Environmental Protection Agency
National Environmental Research Center
Research Triangle Park, NC 27711
William Greenburg
Sun-Telegram
Bernardino, CA
W . J . Hamming
2428 Lee Avenue
Arcadia, CA 91006
P. H. Hanst
Environmental Protection Agency
National Environmental Research Center
Research Triangle Park, NC 27711
T. A. Hecht
Systems Applications Inc.
950 Northgate Drive
San Raphael, CA 94903
J. Holmes
Air Resources Lab.
9528 Telstar Avenue
El Monte, CA 91731
Joel Horowitz
Environmental Protection Agency
Washington, DC
117
-------
C. M. Huang
Research Analyst, Air Quality Branch
Tennessee Valley Authority
Muscle Shoals, AL 35660
R. J. Jaffe
Dept. 56-20, Bldg. 151
Lockheed M. S. Co.
Sunnyvale, CA 94088
Harvey Jeffries
Dept. of Environmental Science
and Engineering
School of Public Health
University of North Carolina
Chapel Hill, NC 27514
Richard Johnson
Environmental Protection Agency
Durham, NC
S. Joshi
Environmental Protection Agency
National Environmental Research Center
Research Triangle Park, NC 27711
Richard Kamens
Dept. of Environmental Science
and Engineering
School of Public Health
University of North Carolina
Chapel Hill, NC 27514
D. B. Kittleson
Dept. of Mechanical Engineering
University of Minnesota
Minneapolis, MN 55455
W. C. Kocmond
Calspan Corporation
4455 Genesee Street, P.O. Box 235
Buffalo, NY 14221
Richard L. Kuntz
Environmental Protection Agency
National Environmental Research Center
Research Triangle Park, NC 27711
Daniel Lillian
U.S. Army Env. Hygiene Agency
Air Pollution Engineering Division
Edgewood Arsenal, MD 21014
William Lonneman
Environmental Protection Agency
National Environmental Research Center
Research Triangle Park, NC 27711
F. J. Malcolm
School of Public Health
University of North Carolina
Chapel Hill, NC 27514
David Miller
Battelle
505 King Avenue
Columbus, OH
Ron Mueller
EPA Region IX,
San Francisco,
100 California Street
CA 94111
Robert Neligan
Environmental Protection Agency
Durham, NC
John Overton
Environmental Protection Agency
National Environmental Research Center
Research Triangle Park, NC 27711
Joseph Padgett
Environmental Protection Agency
Durham, NC
Joseph Paisie
Bureau of Air Quality Control
Dept. of Health and Mental Hygiene
Environmental Health Administration
610 N. Howard Street
Baltimore, MD 21201
T. R. Powers
Exxon Research and Engineering
P.O. Box 51
Linden, NJ 07036
R. A. Rasmussen
Air Pollution Research Section
College of Engineering
Washington State University
Pullman, WA 99163
P. C. Reist
Dept. of Environmental Science
and Engineering
School of Public Health
Univerity of North Carolina
Chapel Hill, NC 27514
L. A. Ripper ton
Research Triangle Institute
Research Triangle Park, NC 27709
Joe Sickles
Dept. of Environmental Science
and Engineering
School of Public Health
University of North Carolina
Chapel Hill, NC 27514
J. Singh
Dept. of Environmental Science
Rutgers University
New Brunswick, NJ 08903
118
-------
John Spence
Environmental Protection Agency
National Environmental Research Center
Research Triangle Park, NC 27711
C. W. Spicer
Battelle
505 King Avenue
Columbus, OH
E. P. Stahel
Dept. of Chemical Engineering
North Carolina State University
Raleigh, NC 27607
E. R. Stephens
Statewide Air Pollution Research Center
University of California
Riverside, CA
R. K. Stevens
Environmental Protection Agency
National Environmental Research Center
Research Triangle Park, NC 27711
R. S. Tsai
Commonwealth of Puerto Rico
Economic Development Administration
Research and Development Dept.
G.P.O. Box 3088
San Juan, Puerto Rico 00936
Charles Walters
Environmental Protection Agency
National Environmental Research Center
Research Triangle Park, NC 27711
Hal Westberg
College of Engineering
Research Division
Washington State University
Pullman, WA 99163
W. E. Wilson
Environmental Protection Agency
National Environmental Research Center
Research Triangle Park, NC 27711
Arthur Winer
Statewide Air Pollution Research Center
University of California
Riverside, CA 92502
Ted Winfield
Environmental Protection Agency
National Environmental Research Center
Research Triangle Park, NC 27711
T. Yang
Calspan Corporation
4455 Genesee Street
P.O. Box 235
Buffalo, NY 14221
119
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-600/3-76-029
2.
4. TITLE AND SUBTITLE
SMOG CHAMBER CONFERENCE PROCEEDINGS
7. AUTHOR(S)
Basil Dlmitriades (Chairman)
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Environmental Sciences Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Trianale Park. NC 27711
12. SPONSORING AGENCY NAME AND ADDRESS
EBVironmental Sciences Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Trianale Park. NC 27711
3. RECIPIENT'S ACCESSION-NO.
5. REPORT DATE
April 1976
6. PERFORMING ORGANIZATION CODE
8. PERFORMING ORGANIZATION REPORT NO.
10. PROGRAM ELEMENT NO.
1AA008
11. CONTRACT/GRANT NO.
f
13. TYPE OF REPORT AND PERIOD COVERED
Final Oct. 24-25, 1974
14. SPONSORING AGENCY CODE
EPA-ORD
15. SUPPLEMENTARY NOTES
16. ABSTRACT
Smog chamber methodology was reviewed with respect to its present status and
utility in developing photochemical pollution control strategies. Measurement
of NO and 03 in chamber atmospheres was judged to be satisfactory; measurement
of N02» non-methane hydrocarbon, and sulfate presented problems. Surface
effects and background contamination problems were minimal in large (hundreds
of cubic feet) chambers made of Teflon film. Compared to indoor chambers,
outdoor chamber operations were less costly and yielded more valid data
1n some respects. Specific sets of smog chamber data were used to estimate
oxidant-related control requirements for California's South Coast Air
Basin. Such estimates were judged to be more useful relative to those
based on aerometric data analysis in that the role of NO in oxidant
formation was considered quantitatively. Other applications of smog
chamber methodology in photochemical air pollution research were discussed.
17.
KEY WORDS AND DOCUMENT ANALYSIS
a. DESCRIPTORS
Air Pollution
* Test chambers
* Design
* Reviews
Evaluation
Contamination
* Photochemical rrartlnn-
is. Di§flM8oWoWs'iw-EMiN?cli1oria
RELEASE TO PUBLIC
•
b.lDENTIFIERS/OPEN ENDED TERMS
19. SECURITY CLASS (This Report)
UNCLASSIFIED
20. SECURITY CLASS (This page)
UNCLASSIFIED
c. COS AT I Field/Group
13B
14B
05B
07 E
21. NO. OF PAGES
12Q
22. PRICE
EPA Form 2220-1 (9-73)
120
------- |