EPA-650/4-75-006
A SPECTROSCOPIC STUDY
OF CALIFORNIA SMOG
by
Philip L. Hanst
William E. Wilson
Ronald K. Patterson
Bruce W. Gay, Jr.
Lucian W. Chaney
Chemistry and Physics Laboratory
National Environmental Research Center
and
Charles S. Burton
Rockwell International
Thousand Oaks, Califo>-nia
ROAP No. 21AKB-13
Program Element 1AA008
U. S. Environmental Protection Agency
Office of Research and Development
National Environmental Research Center
Research Triangle Park, North Carolina 27711
February 1975
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, 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 con-
sciously 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 ENVIRONMENTAL MONITORING series. This
series describes research conducted to develop new or improved methods and
instrumentation for the identification and quantification of environmental
pollutants at the lowest conceivably significant concentrations. It also
includes studies to determine the ambient concentrations of pollutants in
the environment and/or the variance of pollutants as a function of time
or meteorological factors.
EPA REVIEW NOTICE
This report has been reviewed by the Office of Research and Development,
Environmental Protection Agency, and approved for publication. Approval
does not signify that the contents necessarily reflect the views and
policies of the Environmental Protection Agency, nor does mention of trade
names or commercial products constitute endorsement or recommendation for
use.
DISTRIBUTION STATEMENT
This report is issued by the Environmental Protection Agency to report
technical data of interest to a limited number of readers. Copies are
available free of charge to Federal employees, current contractors and
grantees, and nonprofit organizations - as supplies permit - from the
Air Pollution Technical Information Center, Environmental Protection
Agency, Research Triangle Park, North Carolina 27711. Document is
available to the public, for a fee, through the National Technical
Information Service, Springfield, Virginia 22161.
Publication No. EPA-650/4-75-006
-------
ABSTRACT
Long-path infrared spectroscopy has yielded data on the composition
and chemistry of the polluted air at Pasadena, California. Infrared
radiation was transmitted along a 417-meter path folded between mirrors
in a glass tube 9 meters long. Spectra of polluted air were re-
corded with a Fourier transform spectrometer system and were plotted
in ratio mode against the spectra of humidified reference air. This
ratio plotting allowed the observation of weak pollutant absorption
lines by removing the background spectrum of water and carbon dioxide
lines. Data were taken in late November 1972 and in the summer of
1973. In the 1972 period, the level of auto exhaust pollution was
high in the morning and evening, but there was very little formation
of ozone and other oxidation products in the air. In the summer of
1973, the morning and evening pollution levels were generally not as
high as in the fall, but the stagnation of the daytime air led to the
formation and observation of ozone, peroxyacetyl nitrate, formic acid,
and other oxidation products. On July 25, a concentration maximum of
0.68 part per million of ozone was recorded, along with maxima of 0.07
part per million of formic acid and 0.05 part per million of peroxy-
acetyl nitrate. Also measured were hydrocarbons, chlorinated hydrocarbons,
carbon monoxide, and methanol. From the spectra, it is concluded that
nitric acid vapor was not present in the smog at concentrations higher
than 10 parts per billion and ammonia was not present at concentrations
higher than 5 parts per billion. The chemistry of the air is discussed
in terms of the observations.
in
-------
CONTENTS
Section Page
LIST OF FIGURES v
LIST OF TABLES vi
ACKNOWLEDGMENT vii
CHEMICAL SYMBOLS ANE ABBREVIATIONS viii
SUMMARY 1
INTRODUCTION 5
Background 5
Equipment and Method 7
Calibration Spectra 10
RESULTS AND DISCUSSION 15
Results 15
Data Analysis and Discussion 20
Ratios of Pollutant Concentrations 23
Pollutant Concentrations as Functions of Time 26
Individual Compounds 31
Material Balance Considerations 46
Future Work 54
REFERENCES. 57
TECHNICAL REPORT DATA AND ABSTRACT 60
IV
-------
LIST OF FIGURES
Figure Page
1. Glass Tube in Penthouse 7
2. Spectrometer and Data System in Shed 8
3. Pollutant Measuring System 8
4. Carbon Monoxide Absorption Coefficient as a Function of
Apparent Energy Ratio at 2700 cm"1 13
5. Atmospheric Spectrum, November 30, 1972, 5:15 p.m 16
6. Atmospheric Spectrum, July 25, 1973, 9:30 a.m. 18
7. Atmospheric Spectrum, July 25, 1973, 1:00 p.m. 19
8. Acetylene versus Carbon Monoxide, Fall and Summer Periods- • • • 23 "--'-~?
9. Nonmethane Paraffinic Carbon versus Carbon Monoxide,
Fall and Summer Periods 24 """.'.
10. Pollutant Concentration Plots, July 24, 25, and 26, 1973 .... 28
11. Reaction Product Plots for July 24, 25, and 26, 1973 29
12. Pollutant and Product Plots for August 9, 1973 30 -
13. Atmospheric Spectrum, July 24, 1973, 1:00 p.m.,
Showing Absence of Ammonia by Comparison
with Hypothetical Spectra for Atmospheres
Containing Ammonia 32
14. Formaldehyde Response: Contrast between Pasadena
Site, Where Formaldehyde Was Not Detected,
Ambient Air in Raleigh, N. C., and Auto Exhaust
in Laboratory Air 34 „--''— ."
15. Detection of Formic Acid 37
16. Ratio Spectrum Showing Possible Presence of Hydrogen Peroxide. . 4,0
17. Detection of 0.10 ppm Methanol and 0.021 ppm Freon 12;
11:00 a.m., August 23, 1973 42 .-' '
18. Atmospheric Spectrum, July 25, 1973, 12:00 noon, Showing
Absence of Nitric Acid by Comparison with Hypothetical
Spectra for Atmospheres Containing Nitric Acid 45
-------
LIST OF TABLES
Table . Page
1. Absorption Coefficients 12
2. Pollutant Concentrations, Pasadena, 27 meters above Ground,
November 20-December 1, 1972 21
3. Pollutant Concentrations, Pasadena, 27 meters above Ground,
Summer 1973 22
4. Materials in the Photooxidation Process, Averaged for
July 24 and 25, 1973 48
5. Nitrogen-containing Compounds 54
VI
-------
ACKNOWLEDGMENT
Professor Sheldon K. Friedlander of the California Institute of
Technology provided the working space on the roof of the Keck Engineering
Building, California Institute of Technology, and assisted the project
in many other ways. His support is gratefully acknowledged. Thanks
are also extended to Mr. Ernie Caldwell of the Environmental Protection
Agency, Office of Administration, for his valuable editorial assistance.
VII
-------
CHEMICAL SYMBOLS AND ABBREVIATIONS
CC12F2 Freon 12
CC13F Freon 11
CC1, carbon tetrachloride
CH, methane
C2H2 acetylene
C2H, ethylene
CH OH methanol
C2HC13 trichloroethylene
CH nonmethane paraffinic carbon; x = 2 or 3
CO carbon monoxide
CQj carbon dioxide
HCOOH formic acid
H2CO formaldehyde
HNO_, nitrous acid
HNO nitric acid
H202 hydrogen peroxide
NH. ammonia
NO nitric oxide
NO,., nitrogen dioxide
N20 nitrous oxide
N?0,. nitrogen pentoxide
0,, ozone
PAN peroxyacetyl nitrate
R hydrocarbon radical
VIII
-------
A SPECTROSCOPIC STUDY OF CALIFORNIA SMOG
SUMMARY
Long-path infrared spectroscopy has been applied to the Pasadena
atmosphere to obtain data on the chemistry of the smog. A Fourier
transform spectrometer and long-path cell were set up on the roof of
the Keck Engineering Building at the California Institute of Technology.
A folded-path White cell with a base path of 8.7 meters was used. The
path was enclosed in a glass tube 9.2 meters long and 0.3 meter in diameter.
Forty-four or 48 passes were used, yielding path lengths of 383 and 417
meters. There were two detectors, each at liquid nitrogen temperature:
indium antimonide for the frequency region 2000 to 3300 reciprocal cen-
timeters (cm ), and mercury-cadmium-telluride for the region 700 to 1360
cm . The intermediate region was not recorded because of water vapor in-
terference. Spectral resolution was approximately 2 cm . The digitized
spectra of ambient air were plotted in ratio mode against the spectra of
humidified reference atmospheres. This cancelled the water interference,
allowing weak pollutant bands to be seen.
Absorption coefficients were carefully ascertained either from the
literature or by new measurements. The values used are tabulated and
references are given.
The sample was taken in through a 0.1-meter-diameter plastic pipe
that ran up into the air 9 meters above the roof of the building. This
placed the intake point 27 meters above the ground. The absorption cell
could be filled and emptied in just a few minutes. Recording a spectrum
took about 5 minutes.
-------
There were two periods of data taking: late November 1972 and July
and August 1973. The 1972 period did not include any smoggy days, but
it did include mornings and evenings of stationary air, which, although
visually clear, showed an accumulation of gaseous pollution. The 1973
summer period included the smog attack of July 24 and 25, which yielded
the highest ozone concentrations Pasadena had seen for several years. In
the fall, about three spectra per day were recorded; in the summer, 15
to 20 per day were recorded. Spectra are plotted in ratio mode over re-
ference air and also plotted in single-beam mode. A typical spectrum
from the fall period shows concentrations approximately as follows:
carbon monoxide, 5 parts per million (ppm); methane, 2.5 ppm; paraffinic
carbon, 1.0 ppm; acetylene, 0.050 ppm; ethylene, 0.050 ppm.
In the fall mornings and evenings, the carbon monoxide and hydrocarbon
levels were about twice as high as in the summer period. Acetylene-to-
carbon monoxide ratios were about 1/100 in both periods, but the ratios
of paraffinic carbon to carbon monoxide varied, being about 1/4.8 in
the fall but only 1/2.6 in the summer. It is suggested that the change
in pollution mixture may be the result of air movements, inversion height
differences, and temperature differences.
On July 24 and 25, 1973, ozone concentrations reached 0.60 and 0.68
ppm. On those days, carbon monoxide was only about 3 ppm; methane, 2 ppm;
and nonmethane hydrocarbon about 1.0 ppm carbon. Acetylene was 0.030 ppm
and ethylene only 0.006 ppm, reflecting the trapping of the air mass and
the high extent of reaction. Formic acid was identified as the organic
reaction product in highest concentration at 0.060 ppm. Peroxyacetyl
nitrate (PAN), at 0.050 ppm, was the only identifiable nitrogen-
containing product. Also detected were traces of halogenated compounds
,,nd r.'ethanol. There was a slight indication of hydrogen peroxide with
an upper limit of 0.070 ppm. Compounds not detected, and their de-
tection limits, were: nitric acid, 0.010 ppm; methyl nitrate, 0.010
ppm; nitrogen pentoxide, 0.010 ppm; formaldehyde, 0.030 ppm; and
ammonia, 0.005 ppm.
Graphs of reactant and product concentrations as functions of time
were plotted for several of the smoggy summer days. The general indication
-------
was that the air mass trapped beneath the inversion was well mixed, but
had about an 11-hour rate of exchange with the air above it. This slow
rate of exchange led to the unusually high degree of reaction and the
low ethylene-to-acetylene ratio. Furthermore, it is concluded that the
long turnover time for the air mass led to a high degree of conversion
of the nitrogen oxides to nitrates, and that by late afternoon most of
these nitrates were no longer present either in the gas phase or in the
particulate phase.
The high ozone yield and the assumed advanced state of nitrogen
oxide conversion indicate that a substantial portion of the hydrocarbon
in the air was oxidized by the end of the day.
In addition to the above, other aspects of the photochemistry of
the Los Angeles atmosphere are also discussed.
-------
INTRODUCTION
BACKGROUND
The composition of the polluted Los Angeles air has been studied for
many years with a variety of measurement methods. The main purpose of
direct atmospheric analysis is to determine what people are exposed to,
but in addition it can indicate the sources of the pollution, the trans-
formations of the compounds, and the pollution removal paths. The main
prior application of long-path infrared absorption to atmospheric analysis
1-4
was the work of the Franklin Institute group between 1953 and 1957.
Among the results of that earlier infrared work were the discovery of the
peroxyacyl nitrate family of pollutants, the proof of the presence of
peroxyacyl nitrates in the Los Angeles atmosphere, and spectroscopic
confirmation that ozone is a major product of atmospheric photooxidations.
Infrared techniques were exploited in those earlier studies as
fully as the equipment allowed. A prism spectrometer was used, with a
thermocouple detector. The folded path was about 240 meters long. At-
mospheric spectra were compared to reference spectra by inspection, and
point-by-point ratio plots were made through slide-rule calculations.
These techniques yielded a limit of pollutant detectability of about 0.05
ppm for peroxyacetyl nitrate, ozone, ethylene, and acetylene. This was
sufficient for detection in the ambient air under some conditions, but
not all. The limit of detectability for nitric acid, formic acid, and
formaldehyde fell in the vicinity of 0.2 ppm—too high for detection. Most
of the compounds that were detected by infrared can also be measured with
simpler equipment, and therefore long-path infrared studies were relegated
to a minor role for the following 15 years.
In the laboratory, the infrared method has continued to be success-
ful. Infrared studies of pollutant reactions at parts-per-million levels
have been carried on at the General Motors laboratories and in several
university laboratories. These studies have been a major source of pro-
gress in air pollution chemistry. '
5
-------
Ambient air analysis has been carried out mainly by the gas chroma-
tographic method. In hydrocarbon analysis, this method easily exceeds
the capability of long-path infrared. The infrared method can give a
fairly sensitive measure of total hydrocarbons in the air, and a highly
sensitive measure of individual hydrocarbon species with one, two, and
three carbons, but it cannot resolve the heavier hydrocarbons. Gas
chromatography, in contrast, cleanly separates and identifies nearly
all components of a mixture of hydrocarbon pollutants, down to parts-
per-billion (ppb) levels.
The main value of the infrared method is in identifying and measuring
all the assorted nonhydrocarbon species in the air, especially the
oxygenated and nitrogenated compounds. This is just the aspect of air
analysis in which the chromatographic method is weakest. Thus, the two
methods are complementary, and long-path infrared methodology still has
a vital rcle to fill in atmospheric studies.
Improved optical components of many different types have been deve-
loped in the past 20 years. These include the solid-state detectors,
the laser, the scanning Michelson interferometer, and the dedicated
minicomputer for processing interferograms and spectra. The Fourier
transform spectrometer systems now commercially available bring all of
these components together in a package that yields higher resolution
and higher signal-to-noise ratios than conventional grating or prism
spectrometers. Furthermore, the new spectrometer data systems permit
the automatic plotting of ratio spectra when a single absorption cell
is used.
Since new instrumental improvements were capable of yielding a
> , or even 100-fold increase in measurement sensitivity in an
atmospheric long-path infrared experiment, a new long-path absorption
cell has been built especially for use with a Fourier transform
spectrometer. It was soon demonstrated that the combination was
yielding a parts-per-billion detection sensitivity for many of the
important air pollutant species. These results were described in a
recently published article, which should be consulted for additional
details.7
-------
EQUIPMENT AND METHOD
In 1972, as part of a study of particulate and gaseous pollution,
the Fourier transform spectrometer and long-path cell were set up in Pasadena
on the roof of the Keck Engineering Building at the California Institute
of Technology. A folded-path White cell with a base path of 8.7 meters
Q
was used. The path was enclosed in a glass tube 9.2 meters long and 0.3
meter in diameter. The tube is shown in Figure 1. Either 44 or 48 passes
Figure 1. Glass tube in penthouse.
were used, yielding path lengths of 383 and 417 meters. The only place
available to locate the system was inside the penthouse for the air con-
ditioning and heating machinery of the building. Although this proved to
be a shaky and noisy environment, most of the vibration was eliminated by
shock mounts placed under the tube and the spectrometer. The spectrometer
and its data system were housed in a small air-conditioned shed, as shown
in Figure 2. Two detectors were used, each at.liquid nitrogen temperature:
-------
Figure 2. Spectrometer and data system in shed.
indium antimonide for the frequency region 2000 to 3300 cm'1, and mercury-
cadmium-telluride for the region 700 to 1360 cm"1. The region 1360 to
2000 cm was not recorded because of water vapor interference.
The system components are diagrammed in Figure 3. Infrared radiation
J*- PATH OF INFRARED
RADIATION
*• FLOW OF DATA AND
INSTRUCTIONS
LONG-PATH CELL
_INTERFEROMETER
INTERFEROMETER
CONTROLLER
DETECTION
SYSTEM
ANALOG
TO-DIGITAL
CONVERTER AND
SIGNAL AVERAGER
4096-WORD
COMPUTER
DISC AND TAPE
MEMORIES FOR
DATA AND
PROGRAMS
TELETYPEWRITER
FOR USER
CONTROL
DIGITAL
PLOTTER
Figure 3. Pollutant measuring system.
-------
from a Nernst glower source was projected into the interferometer. The
modulated optical output of the interferometer (the interferogram) was
passed through the long absorption cell to the detector. Forty digitized
recordings of each interferogram were added together in the data system
in order to increase the signal-to-noise ratio. The summed interferogram
was then transformed to yield the spectrum. (Spectral resolution was
approximately 2 cm .) The digitized spectrum could be displayed, or
it could be stored in the computer memory for future access. Gen-
erally, the ratio between the air spectrum and a stored reference spectrum
was computed and plotted.
The air sample was taken in through a 0.1-meter-diameter plastic
pipe, which ran out the door of the penthouse and up into the air about
9 meters above the roof on the fourth floor. This placed the intake
point about 27 meters above the street. A blower kept the air moving
through the intake pipe at all times. The glass long-path cell was evac-
uable down to less than 1 torr pressure. The manner of collecting a sam-
ple was as follows: (1) The long-path cell was evacuated. (2) By opening a
valve, the flow of outside air was directed into the long-path cell so that
the cell was filled to a pressure sufficient to cancel the water vapor
lines against those in a reference spectrum. Generally, the pressure
was between 500 and 600 torrs. (3) Recording of the spectrum was begun
immediately and was completed within about 5 minutes. It is hard to
imagine that this manner of sampling could result in the loss of any
labile species of molecules in the air. The residence time of the air
sample in the intake pipe was no more than a few seconds, and in the
glass cell it was no more than a few minutes. There was good agreement
between the values of ozone measured by infrared and the values measured
by chemiluminescence instruments.
The reference air samples were made up from tank air, humidified to
match the water content of the ambient air. Humidity was adjusted by
mixing two portions of reference air in the long-path cell. One portion
came straight from the tank, dry. The other came from a reservoir of
tank air stored in a large plastic bag with liquid water. The work done
in late November 1972 used Matheson Zero Air for reference. This turned
out to be nearly devoid of carbon dioxide but to contain about 1 ppm of
carbon monoxide. The absence of carbon dioxide was a disadvantage
-------
because it allowed atmospheric carbon dioxide to show in the ratio
plots. The presence of carbon monoxide was also a disadvantage
because it cancelled some of the absorption by ambient carbon
monoxide pollution, thus requiring a correction factor. The 1973 work used
Scott reference air to which 300 ppm of carbon dioxide had been added. This
cancelled most of the carbon dioxide in the ratio plots, but not all.
The Scott air also turned out to include about 1 ppm of carbon monoxide.
It did not prove feasible to obtain an exact water balance in all cases;
but, generally, at least 90 percent of the water interference was removed.
CALIBRATION SPECTRA
Incorrect absorption coefficients are the most likely source of error
in atmospheric infrared studies. Many of the molecular species detected
and measured in this work do not have well-documented reference spectra.
Published absorption coefficients must be used with caution because of the
many possible sources of error in handling the gas and in recording and
interpreting the spectra. Whenever possible, absorption coefficients to
be used were rechecked by running new reference spectra on samples at
parts-per-million concentrations in a laboratory long-path cell. The sim-
plest and safest cases of absorption coefficient measurement involve the
large polyatomic molecules, which are thermally stable. This includes, for
example, paraffinic hydrocarbons, alkyl nitrates, ketones, alcohols, and
alc^l.^Jlir "-'4-^ «-"<-> or more carbons. The spectra in these cases have so
many overlapping lines that there is no inherent fine structure in the bands
at atmospheric pressure. The absorption coefficient is therefore independent
of pressure and instrumental resolution; the absorption equation is obeyed
at nil absorptivities. The most difficult cases involve the molecules with
a small number of lines in their spectra, such as carbon monoxide, nitric
oxide, and hydrogen chloride. The apparent absorption coefficients in
these cases are dependent on total pressure, concentration of the absorbing
species (because of self-broadening), and instrumental resolution. The
absorption equation is only obeyed at small absorptivities. Thermally
unstable species, such as ozone and hydrogen peroxide, and species that
adsorb or polymerize, such as formic acid and formaldehyde, have their own
10
-------
characteristic measurement difficulties. For these reasons, a listing
of absorption coefficients and the conditions under which they were
7 9-12
obtained is given in Table 1. * Reference spectra with sharp bands or
lines were measured with resolution similar to that used in recording
the atmospheric spectra. The values of the absorption coefficient, K,
are those used in the absorption equation:
£n (I /I) =
-------
Table 1. ABSORPTION COEFFICIENTS
Pollutant
Acetylene (C H )
Ammonia (NH )
J
Carbon nu'uoxide
(CO)
Carbon tetra-
chloride (CC1 )
Ethylene (C^)
Formaldehyde
(H2CO)
Formic acid
(HCOv/HJ
Freon 12 (GC12F2)
Hydrogen peroxide
(H2°2)
Methane (Ch^)
Methanol (CH OH)
Methyl nitrate
(CH3ONO,,)
'"
nitric acid
(HN03)
Nitrogen Pentoxide
(NO)
Measurement
frequency,
cm~l
720
930
967
2170
793
950
2780
1105
921
1250
3017
1307
1033
853
1018
1290
896
740
l.!48
Absorption
coefficient,
180
27
35
See curve in
Figure 4
210
20
10
13
45
9 + 3
12
11
25
30
18
35
20
40
40
References and comments
Present work; measured at parts-
per-million concentration in air
Present work
Present work; spectra of parts-
per-million concentrations of CO
in 600-torr air were recorded in
the long-path cell at a resolution
similar to that used in Pasadena
Reference 10
Present work; measured at parts-
per-million concentrations in air
Present work
Applies to central peak on}.y;
measured at parts-per-miJ lion
concentration in long-pat h cell
Reference 10
Present work
Reference 9; rechecked in present
work with 1.7 ppm CH, in atm
of air
Present work
Present work
Reference 7
Present work
Czone
«y
Nonmethane
paraffinu
carbon (CH )"
Peroxyacetyl
nitrate (PAN)
Trichloro-
ethylene
1053
2970
1162
852
8.6 Reference 11
4.0 An average of the absorption
coefficient values, per carbon.
calculated for propane, butane,
pentane, and hexant.1 from Refer-
ence 9
32 Reference 12
22 Reference 10
12
Dues not -n. lude ethane •, r the ring carbons in aromatics; thus, the • b.icrirt x
indicates t-o or hree hydrogens, but not one or four.
-------
ENERGY RATIO (IQ/I)
Figure 4. Carbon monoxide absorption coefficient as a function of apparent energy ratio
at2700cm-1.
13
-------
RESULTS AND DISCUSSION
RESULTS
There were two periods of data taking: (1) late November 1972 and
(2) July and August 1973. The 1972 period did not include any smoggy
days, but it did include mornings and evenings of stationary air that,
although visually clear, showed an accumulation of gaseous pollution.
The 1973 summer period included the smog attack of July 24 and 25, which
yielded the highest ozone concentrations Pasadena had seen for several
years.
A sample spectrum from the late November period is shown in Figure 5.
This spectrum was obtained on Thursday evening, November 30, 1972, at
5:15 p.m. Sample pressure was 760 torr, and the path length was 383
meters. Only the higher frequency end of the spectrum recorded with the
indium antimonide detector is shown. In interpreting this spectrum, it
should be recognized that approximately 10 percent of the signal was due
to stray radiation that travelled only a few tens of meters in the cell,
rather than the full path. This condition could have been corrected but
was not recognized in time. When properly allowed for, the stray radiation
does not invalidate any of the results. The single-beam atmospheric ab-
sorption spectrum appears in the lower part of the figure, and the ratio
plot of that spectrum over the spectrum of humidified reference air appears
in the upper part. The single-beam plot shows the many water vapor lines,
which are removed in the ratio plot. In proceeding from lower frequency to
higher, the single-beam spectrum shows a gradual decrease in signal due
to the response characteristics of the detection system. The ripples
15
-------
u
C9
V
L»
D
z
Q.
in
o
CO
CO
n
CO
o_
E
6.
LO
CM
r^
O)
o
CO
JD
E
03 .
> OJ
OT3
Z'E
El
O (D
o
° r,
Eo
0)
-t-» 4-1
<:
18
'3DNVlillAISNVai
1VN9IS
16
-------
in the single-beam plot, which are especially evident between 2400 and
_i
2500 cm , are the result of interference phenomena in the train of
optical components. The ripples are cancelled out in the ratio plot.
A slight overcompensation of the water vapor absorption has driven some
of the strong water lines above unity in the ratio plot. The Matheson
Zero Air used for reference contained about 1 ppm of carbon monoxide, but
no carbon dioxide, nitrous oxide, or methane. The carbon monoxide band
in the ratio plot is therefore smaller than it should be, but the other
bands appear at full strength. Concentrations of pollutants calculated
from the spectrum are carbon monoxide, 11.5 ppm; methane, 2.8 ppm; non-
methane paraffinic carbon atoms (CH ), 2.2 ppm. Nitrous oxide was at its
13 x
normal value of 0.25 ppm. Carbon dioxide was not calculated from the
spectrum. There was no indication of formaldehyde.
Spectra from July 25, 1973, covering the region 700 to 1360 cm are
shown in Figures 6 and 7. Stray light in these cases was probably less
than 5 percent. For these spectra, the mercury-cadmium-telluride detector
was used. In the 9:30 a.m. spectrum, Figure 6, the sample pressure was
600 torr, and the path length 417 meters. The lower spectrum is the single-
beam plot, showing mainly water and carbon dioxide bands. The middle spectrum
is a ratio plot using a reference spectrum of Scott tank air properly humid-
:'fieri to cancel nearly all of the water vapor absorption. This air also
contained about 300 ppm of carbon dioxide, which cancelled about 90 percent
of the carbon dioxide absorption. The reference air did not contain methane
or other hydrocarbons. The upper spectrum in the figure is a scale-expanded
plot of the spectral region 800 to 1200 cm in which the ordinate scale
is 19 times larger than in the center spectrum. Marked on the spectra are
absorption bands for the following compounds and concentrations: ozone,
0.09 ppm; peroxyacetyl nitrate, 0.016 ppm; methanol, 0.015 ppm; Freon 12,
0.008 ppm; trichloroethylene, 0.015 ppm (perhaps underlaid with some
absorption by Freon 11); formic acid, 0.020 ppm (underlaid with some
absorption by Freon 12); acetylene, 0.028 ppm; ethylene, 0.015 ppm (dis-
torted by a noise spike); and methane, 2.0 ppm.
In the 1:00 p.m. spectra, Figure 7, we see that the photochemical
reaction products—ozone, formic acid, and peroxyacetyl nitrate—have
increased substantially. In Figure 7, the expansion factor of the upper
17
-------
a
UJ
a i-
z o
< -i
Ob O.
x o
o
V)
§
c»
X
ta
I
o
o
CO
CD*
£
Q.
CU
E
r*.
5-
4-1
CO
O.
E
CD'
O
CO
(35
«.
CO
EE
E-—
CDT3
Q. CD
en O
u> =
o E
3DNVillWSNVai
1VN9IS
t
us
18
-------
a uJ
in
oo
a>
I
£
a.
if
S
v
E
r-
5-
i.
+->
ra
a.
E
a.
8
co"
r*»
O5
EE
Q. CD
vt U
o >.
'
V3 (—
o E
1°
-------
spectrum is only 7-fold. The 1:00 p.m. values were ozone, 0.60 ppm;
formic acid, 0.063 ppm; and peroxyacetyl nitrate, 0.051 ppm. By 1:00
p.m. the acetylene and methane had increased slightly to values of 0.030
ppm and 2.4 ppm, respectively. The ethylene concentration had decreased
to approximately 0.009 ppm.
Pollutant concentrations measured between November 20, 1972, and
December 1, 1972, are listed in Table 2. In general, this was a time of
air stagnation morning and evening, ventilation and clean air in the middle
of the day, and no smog.
Concentrations measured on selected days in the summer of 1973 are
listed in Table 3. July 24 and 25 were days of very high ozone concen-
tration, and all values calculated from the spectra of those days are
listed. July 26 was also expected to be a day of very high oxidant, and
in fact, Federal employees in the Los Angeles area were excused from
work. However, the atmospheric conditions changed—perhaps beginning
in the late afternoon of the 25th—and very little smog developed. Other
days of high ozone showed a pattern of concentrations and variations with
time generally similar to those observed on the smoggy days of July 24 and
25. Since the ratio of nonmethane paraffinic hydrocarbon to carbon monoxide
can indicate pollutant sources, the values for those two pollutants measured
on 5 days in August are also listed. The gaps in the table result from
the choice of detectors and spectral regions for recording. The bulk
of the July 24 data was obtained in the 700- to 1400-cm region, using the
mercury-cadmium-telluride detector; hence, they do not show carbon monoxide
or paraffinic carbon. On July 25 and 26, the data obtained with the mercury-
cad. "_m-telluride detector were recorded out to 2300 cm and hence include
carbon monoxide. The various measurements that show paraffinic carbon and
carbon monoxide were obtained with the indium antimonide detector. Some
additional details from the spectra will be included in the following dis-
cussions.
DATA ANALYSIS AND DISCUSSION
Analysis of the results yields information on the sources and sinks
of the pollutants and on the pollutant transformations in the air. The
20
-------
Table 2. POLLUTANT CONCENTRATIONS (ppm), PASADENA, 27 meters ABOVE GROUND,
NOVEMBER 20-DECEMBER 1, 1972a
Date and time
Nov. 20-11:30 a.m.
20- 4:45 p.m.
21-11:00 a.m.
21- 4:10 p.m.
22- 8:10 a.m.
22-12:00 a.m.
27- 8:10 a.m.
27- 3:10 p.m.
27- 5:00 p.m.
28- 9:00 a.m.
28- 3:30 p.m.
28- 5:00 p.m.
29-10:00 a.m.
29- 3:15 p.m.
30- 8:10 a.m.
30- 1:10 p.m.
30- 5:15 p.m.
Dec. 1- 8:15 a.m.
1- 4:00 p.m.
Carbon
monoxide
1.2
3.4
1.2
2.7
7.5
3.1
9.5
4.8
6.0
4.0
3.6
2.6
0.9
6.1
1.8
11.5
11.2
5.6
Methane
2.1
2.2
1.9
2.3
2.8
2.3
2.7
2.5
2.6
2.4
2.6
2.2
2.0
2.2
2.2
2.8
2.7
2.6
Nonmethane
paraff inic
carbon
0.16
0.45
0.11
0.29
1.23
0.34
1.42
0.89
1.34
0.45
0.63
0.30
0.16
1.2
0.42
2.2
2.0
1.25
Acetylene
0.011
0.017
0.030
0.083
0.022
0.090
0.041
0.043
0.030
0.039
0.058
0.019
0.006
0.044
0.022
0.071
0.090
0.041
Ethylene
0.010
0.010
0.025
0.098
0.026
0.071
0.030
0.042
0.030
0.020
0.046
0.015
0.005
0.045
0.015
0.065
0.092
0.031
Blanks indicate no measurement.
21
-------
Table 3. POLLUTANT CONCENTRATIONS (ppm), PASADENA, 27 meters
ABOVE THE GROUND, SUMMER 19733
Carbon
Date and time monoxide
July
July
July
Aug.
24-10:00
10:30
11:00
11:30
12:00
12:30
1:00
1:30
2:00
2:30
3:00
3:30
4:10
25- 7:10
8:00
9:30
10:00
10:30
11:00
11:30
12:00
12:30
1:00
1:30
2:00
2:30
3:00
3:30
4:00
4:30
26- 8:10
9:30
10:00
10:30
11:00
11:30
12:00
12:30
3:05
8 - 4:30
9-8:00
3:00
22 - 3:00
23 -10:00
24 -11:30
a.m.
a.m.
a.m.
a.m.
noon
p.m.
p.m.
p.m.
p.m.
p.m.
p.m.
p.m.
p.m.
a.m.
a.m.
a.m.
a.m.
a.m.
a.m.
a.m.
noon
p.m.
p.m.
p.m.
p.m.
p.m.
p.m.
p.m.
p.m.
p.m.
a.m.
a.m.
a.m.
a.m.
a.m.
a.m.
noon
p.m.
p.m.
p.m.
a.m.
p.m.
p.m.
a.m.
a.m.
3
2
2
2
2
2
2
2
2
2
3
3
3
2
2
2
2
2
2
2
]
1
1
1
1
1
1
1
2
2
2
1
1
.3
.2
.4
.1
.1
.6
.8
.8
.6
.9
.9
.9
.5
.9
.8
.5
.1
.2
.5
.0
.5
.8
.9
.8
.5
.5
.6
.7
.5
.2
.1
.5
.3
Nonmethane
paraffinlc
Methane carbon
2.0
1.9
1.9
1.9
1.9
2.2
2.3
2.3
2.2
2.1
2.1
1.9
0.71
0.78
2.0 1.05
2.0
2.0
2.0
2.1
2.2
2.2
2.2
2.4
2.4
2.4
2.3
2.3
2.2
2.0
0.49
1.1
2.0
2.0
2.1
2.1
2.0
1.8
1.8
0.67
0.44
0.75
0.61
0.60
0.36
0.51
Acetylene
0.022
0.025
0.024
0,
.021
0.019
0.019
0,
0,
0,
0,
0,
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.026
.029
.024
.033
.036
.033
.023
.028
.026
.026
.022
.025
.024
.024
.030
.032
.036
.031
.025
.019
.014
.022
,020
.022
.024
.020
.017
.013
Ethylene
0.014
0.012
0.012
0.005
'0.005
0.005
0.004
0.004
0.005
0.005
0.005
0.005
0.018
0.015
0.015
0.009
0.009
0.007
0.006
0.007
0.009
0.010
0.011
0.009
0.006
0.005
0.004
0.015
0.016
0.015
0.013
0.010
0.006
0.004
Ozone
0.07
0.10
0.13
0.22
0.25
0.26
0.35
0.41
0.57
0.59
0.57
0.46
—
0.09
0.19
0.25
0.32
0.37
0.39
0.42
0.60
0.68
0.66
0.59
0.53
0.45
0.35
0.06
0.12
—
0.05
0.04
0.05
0.05
Formic
acid
0.017
0.020
0.027
0.031
0.030
0.031
0.033
0.038
0.048
0.057
0.053
0.047
0.013
0.020
0.038
0.043
0.060
0.064
0.054
0.057
0.063
0.072
0,068
0.063
0.064
0.057
0,055
0.010
0.010
_-
0.005
—
_-
—
Peroxy
acetyl
nitrate
0.011
0.013
0.016
0.026
0.025
0.038
0.029
0.035
0.041
0.047
0.047
0.036
0.004
0.016
0.019
0.022
0.037
0.034
0.033
0.040
0.051
0.051
0.053
0.044
0.046
0.043
0.034
0.009
0.010
—
0.005
—
—
—
aBlanks indicate no measurement. Dashes indicate not detected.
22
-------
data will be discussed in terms of (1) the ratios of pollutant concen-
trations, (2) the time variations of the concentrations, (3) the amounts
of the pollutants seen or not seen, and finally, (4) the material balance
(or unbalance) in the photochemically reacted air.
Ratios of Pollutant Concentrations
Plots have been made of the pairs of acetylene and carbon monoxide
values for the separate fall and summer periods. These are shown in
Figure 8. The fall data are from Table 2 for 8 days between November
12, 1 , . ,
10
o
u
x
o
o
BO
oc
-------
of nonmethane paraffinic hydrocarbons, as indicated by the C-H absorption
band at 2970 cm , are given in Figure 9. These plots show ratios that
12
10
o
u
x
o
z
o
00
tc.
FALL
(NOV. 20 TO DEC. 1,1972)
CHX/CO = 0.19
4:30 P.M
JULY 25
4:10P.M.
JULY 24
SUMMER (1973)
CHX/CO = 0.37
0.5
1.0
1.5
2.0
2.5
NONMETHANE PARAFFINIC CARBON (CHX), ppm
Figure 9. Nonmethane paraffinic carbon versus carbon monoxide, fall and summer
periods.
difffr almost by a factor of 2 between the fall and summer periods. These
ratios can be interpreted as indicating two different mixes of pollutant
sources.
The difference might possibly be a result of atmospheric oxidation
of the hydrocarbons. That explanation seems unlikely, however, because
of the direction of the differences. Hydrocarbon oxidation occurs more
in the summer, which is a time of stagnant midday air, rather than in the
fall, which is a time of midday ventilation and air movement. If oxidation
were a major factor, the hydrocarbon-to-carbon monoxide ratio would be
lower in the summer—the reverse of what has been observed.
-------
A more likely explanation of the difference in ratios can be derived
from differences in atmospheric physical properties and atmospheric movements.
It is to be noted that the fall measurements taken in the morning and
evening showed pollutant concentrations averaging about twice as high
as in the summer. This was the result of the air being trapped under a
very low-altitude temperature inversion. In the summer, the temperature
inversion occurred at a higher altitude, with a resulting larger mixing
volume. In the fall, the shallowness of the polluted air layer would
restrict the lateral pollution transport. In this case, the pollution
at the campus of the California Institute of Technology would be largely
auto exhaust from the cars in the nearby streets, and there would be
little influence from distant industrial sources or natural sources.
In the summer, the larger mixing volume gives a smaller pollutant
concentration than in the fall, but at the same time allows more pollutant
transport from a distance. Thus, the summer air at the site could con-
tain, in addition to auto exhaust, evaporated organic materials from
sources farther out, including industrial operations, solvent evaporation,
and petroleum leaks.
Another factor to be considered is temperature differences, as
14
pointed out recently by Stephens. Midday heat will vaporize organic
materials. In the summer, the air is not only warmer and the sun more
intense than in the fall, but the air is trapped under a temperature
inversion in midday, thus allowing accumulation of the organic vapors.
In the fall, the sun and warmth of midday release vapors into a mobile
atmosphere that dissipates them quickly.
In summary, this interpretation of the data indicates that in the
summer a substantial portion of the organic pollutants comes from sources
other than auto exhaust. These infrared data do not give a sufficiently
detailed analysis of the hydrocarbon portion of the pollution to identify
specific nonautomotive sources of hydrocarbons. Chromatographic analyses
are much more informative in that regard. '
The ratio between ethylene and acetylene is a measure of the degree
of photolysis of the polluted air mass. Both pollutants come almost ex-
25
-------
clusively from auto exhaust and have a fixed ratio if there are no at-
mospheric transformations. Measurements have shown the average ethylene-
to-acetylene ratio in auto exhaust to be approximately one-to-one. A
ratio fairly close to this was obtained in the measurements of November 20 to
December 1, as shown in Table 2. However, Table 3 illustrates that in the
summer, the ratio was frequently much smaller. The ethylene is much more
reactive photochemically than the acetylene, with a disappearance rate in
smog-chamber experiments many times greater. Thus, the "photochemical age"
of the air mass can be judged from this ratio, as discussed, for example,
18
by Stephens and Burleson.
Data from the tables, plotted in Figures 10 and 11, show the vari-
ations of pollutant concentrations during the day. The data for July 24
and 25 show an ethylene-to-acetylene ratio about half of the auto exhaust
value by midmorning and a ratio in the afternoon only about one-fifth of
the auto exhaust value. The indicated high degree of photochemical ac-
tivity is confirmed by the high values of ozone obtained on those days.
The observation that by noon the ethylene concentration had already
fallen as low as one-fifth of the acetylene concentration indicates
that fresh ethylene was not coming into the air mass at a high rate. Lab-
oratory measurements with simulated atmospheric conditions have shown the
ethylene to have a half-life of several hours. If a large continuous influx
of ethylene existed, the steady-state ethylene-to-acetylene ratio should
therefore remain much higher than observed. Instead, it appears that the
reactive components were transformed early in the day, and they were not
lepla :cd.
"-r comparison, data are shown in Figure 12 for August 9, a summer
day on which only a moderate amount of ozone was formed. On this day,
Che ethylene-to-acetylene ratio remained near one-to-one in the morning
and evening and dropped only to about three-to-four in the middle of the
day when the photochemical oxidation was at its height. There was also
a variation in concentrations during the day.
Pollutant Concentrations as B'unctions of Time
Carbon monoxide and acetylene, both indicators of auto exhaust,
did not vary markedly with time during the smoggy days of July 24 and
26
-------
25. Nor did they vary markedly during other summer days of moderately
high oxidant, such as August 9. The relatively steady value of these
pollutants is at variance with a concept of morning and evening auto
exhaust peaks. If there was any peaking of the traffic count, it was
obscured by the integrating effect of the well-mixed trapped air mass.
Probably, the frequently observed morning and evening peaking phenome-
non is a result of morning and evening periods of air stagnation sep-
arated by a midday period of ventilation. This is the pattern we ob-
served in the fall, when there was no smog. In contrast, on the heavy
smog day of July 25, the acetylene and carbon monoxide concentrations
peaked at 2:00 p.m. The amounts of carbon monoxide and acetylene on
July 24 and 25 were low compared with the amounts frequently reported
for urban locations. This emphasizes the important role that the
meteorological factors play in the Southern California smog. It is the
trapping of the midday air that allows the photochemical reaction pro-
ducts to accumulate.
Three principal reaction products were observed—ozone, formic acid,
and peroxyacetyl nitrate. The variations of these with time during the
July smog attack are shown in Figure 11. The increase of reaction pro-
duct concentrations during the morning and early afternoon followed by
a falling off in the late afternoon, as seen on July 24 and 25, is typical
smog season behavior. This pattern is also shown in the August 9 plots.
July 26 seemed to start out as a. smoggy day, but by noon the air had
gotten moving and cleared out the pollution.
The figures show that on the smoggy days the ethylene concentration
went down quickly in the morning. It can be assumed that propylene,
butadiene, isobutene, formaldehyde, and other components of the auto
exhaust that are more reactive than ethylene decreased in the morning
even faster than the ethylene. It is these highly reactive components
that are mainly responsible for the morning conversion of nitric oxide to
nitrogen dioxide and the start of the ozone buildup. However, there was
much more total ozone formed than can be accounted for by the action of the
reactive species. It is clear that the less reactive hydrocarbons must
27
-------
£ 2 —
0800
1000
1200
TIME OF DAY, hours
1400
1600
Figure 10. Pollutant concentration plots, July 24, 25, and 26, 1973.
28
-------
0.12
1000
1200
TIME OF DAY, hours
1400
0.02 o
0 <
1600 o
Figure 11. Reaction product plots July 24, 25, and 26, 1973.
29
-------
0.03
g 0.02
e*j
O
o
o 0.01
i r
0.06
0.04
0.02
2.5
2.0
o
u
S 1.5
x
o
o
CO
1.0
0.5
0
0600
0800
iono
1800
1200 1400 1600
TIME OF DAY, hours
Figure 12. Pollutant and product plots for August 9, 1973.
2000
0.05
0.04 E
0.03
<
"Si
0.02 =£,
u
UJ
Z
_i
>•
0.01 <
0
2200
30
-------
also be contributing to the ozone buildup. This will be further discussed
in connection with the material balance.
Individual Compounds
Acetylene—Acetylene is probably the best indicator compound for the
presence of auto exhaust pollution in the atmosphere. No major sources
of acetylene other than the internal combustion engine are normally pre-
sent. Laboratory studies have shown acetylene to be highly resistant
to atmospheric oxidation. Its half-life in the atmosphere is probably
many days. The compound thus serves as a reference against which to
measure the extent of reaction of other species.
The acetylene absorption band at 730 cm (Figures 6 and 7) is one
of the strongest known, so that small amounts can be measured reliably.
Ammonia—Ammonia is assigned a prominent role in fine particle formation
in the atmosphere. Ammonium sulfates and nitrates are major constituents
of atmospheric aerosol, as shown, for example, in the reports by Gordon and
19 20
Bryan and by Charleson et al. ' Decaying vegetation is regarded as the
principal source of atmospheric ammonia. Georgii reported about 20 ppb
of ammonia in the air at Frankfurt, Germany, but only about 6 ppb in air
71
coming into Europe from the Atlantic Ocean.
The infrared absorption spectrum of ammonia is undoubtedly one of the
beat indicators of the presence or absence of the compound. The ammonia
bands at 930 and 967 cm are strong and distinctive. Furthermore, there
is very little interference at these wavelengths from water vapor or from
other pollutants. We have seen these infrared bands in the spectrum of
air at Durham, North Carolina, indicating an ammonia level of about 20 ppb.
Figure 13 shows the 1:00 p.m. spectrum from July 25 as it actually
appeared and as it would have appeared if ammonia were present at con-
centrations of 10 or 40 ppb. Even at 10 ppb, the expected absorption
features are several times larger than the noise level in the spectrum.
Examination of all of the several hundred spectra recorded in the fall and
summer observation periods has failed to show any indication of absorption
at the principal ammonia band at 967 cm . For many of the spectra, the
noise level corresponds to about 5 ppb of ammonia. The indicated low
31
-------
HYPOTHETICAL
SPECTRUM
10ppbNH3
HYPOTHETICAL
SPECTRUM
40ppbNH3
1000
1100
WAVELENGTH, cm'1
Figure 13. Atmospheric spectrum, July 24, 1973, 1:00 p.m., showing absence of am-
monia (NH3) by comparison with hypothetical spectra for atmospheres containing am-
monia.
32
-------
concentration of gaseous ammonia is understandable in the light of several
factors: (1) The amount of decaying vegetation and other organic matter
per square kilometer in the Los Angeles area is smaller than in most places
where ammonia has been detected. (2) Air coming in from the Pacific Ocean
would be expected to be relatively low in ammonia content. (3) Nitrate
and sulfate formation in the Los Angeles atmosphere will consume gaseous
ammonia.
Carbon Monoxide—The carbon monoxide spectrum occurs with little inter-
ference from other compounds. The principal uncertainties of the carbon
monoxide measurement come from variations in its apparent absorption
coefficient, as previously discussed. The amounts of carbon monoxide
detected were smaller than customarily seen in center-city monitoring
stations, but were comparable to the amounts measured in outlying areas,
22
such as Riverside. The use of carbon monoxide as an indicator of the
air stagnation has been discussed. Carbon monoxide is a comparatively
inert compound in the air, with a half-life of several months. Recent
studies of carbon monoxide oxidation have outlined the removal paths for
carbon monoxide in the air and are important to an understanding of overall
23
atmospheric chemistry. However, it is still generally concluded that
carbon monoxide does not significantly influence the short-term atmospheric
photochemistry of urban areas.
Ethylene-—Much emphasis has been placed on the detection or nondetection of
ethylene, and our conclusions as to the low rate of air turnover and high degree
of hydrocarbon reaction on smoggy days have largely been based on ethylene
measurements. The compound is easily enough measured in the spectrum. Its
band at 950 cm is strong and sharp, and its spectrum does not suffer
serious interference from other atmospheric constituents. Figure 7 shows
an ethylene value of 0,009 ppm. Spectra obtained earlier in the summer
mornings and in the fall mornings and evenings showed the ethylene
band stronger, with a maximum of 0.098 ppm.
Formaldehyde—Formaldehyde is detectable by its characteristic
spectral structure in. the region 2700 to 2900 cm . All the spectra
obtained in this study were examined carefully for these spectral lines,
33
-------
2600
2700
2800 2900
WAVELENGTH, cm"1
3000
Figure 14. Formaldehyde (h^CO) response: contrast between Pasadena site, where
formaldehyde was not detected, ambient air in Raleigh, N.C., and auto exhaust in
laboratory air.
34
-------
but no lines were found. The detection limit for formaldehyde was ap-
proximately 15 ppb in the fall 1972 period and about 30 ppb in the summer
1973 period. The difference is attributed to a somewhat better optical
alignment in the fall measurements, which gave a better signal-to-noise
ratio.
Although we were surprised at not seeing formaldehyde, we do not
believe this to be an anomalous aspect of our results. In the fall period
of observation, there was little photochemical activity, so that the only
significant source of formaldehyde was its direct emission in auto exhaust.
For comparison, we have obtained the absorption spectrum of Raleigh, North
Carolina, morning air in the midst of rush-hour traffic. That spectrum
did show a trace of formaldehyde vapor but also showed carbon monoxide
and hydrocarbon levels two or three times greater than the maximum seen at
the Pasadena site. We also obtained the spectrum of air containing a small
amount of auto exhaust from a laboratory-operated car equipped with a
catalytic muffler. The latter spectrum clearly showed formaldehyde bands
along with a large absorption due to nitrogen dioxide. The Raleigh spectrum
and the laboratory spectrum are compared in Figure 14 with one of the best
of the Pasadena spectra. The formaldehyde is clearly detectable in the
laboratory sample at 80 ppb. In the Raleigh air, it is detectable at
about 40 ppb. In the Pasadena sample, it is not detectable. It should
be noted that the scale-expanded fine structure is similar in the
Pasadena and Raleigh samples, except for the formaldehyde bands. This
nonformaldehyde fine structure is mainly due to weak lines of methane.
If the aldehyde had the same ratio to hydrocarbon in both the Raleigh
and Pasadena air samples, it probably was present in the Pasadena air
at about 0.015 ppb.
In the summer, more formaldehyde is produced by the photochemical
oxidation of hydrocarbons than is emitted directly into the air. Ethy-
lene oxidation is probably the largest single source. However, the
aldehyde formed in the air would be further oxidized to formic acid, carbon
uionoxide, carbon dioxide, and water. In view of the indicated advanced
state of oxidation of the polluted air samples on the smoggy days, it is
understandable that formaldehyde was not detected.
35
-------
Formic Acid—The presence of formic acid is revealed by its absorption
band centered at 1105 cm . This is a fairly strong band that has a
distinctive shape and falls in a spectral region where water vapor inter-
ference is not great. This band has allowed the infrared detection of
formic acid at levels between 10 and 70 ppb.
The detection of formic acid is illustrated in Figure 15. The top
curve is a laboratory spectrum of the formic acid and ozone produced by
30 minutes of ultraviolet irradiation of 20 ppm of formaldehyde and 0.5
ppm of nitrogen dioxide in air. The infrared path was 170 meters. Below
are seven spectra of the Pasadena air recorded on July 25, 1973, and one
recorded on July 26. The spectra show the appearance, growth, and decay
_i
of three bands: the ozone band at 1050 cm , the formic acid band with
absorption peak at 1105 cm , and the peroxyacetyl nitrate band at 1165
cm . In the 9:30 spectrum at the top, the water bands have been fully
cancelled and a very small amount of formic acid appears. The 10:30 spectrum
shows more formic acid and a slight undercompensation of the water lines.
The 11:30 and 12:30 spectra show an overcompensation of the water lines,
which drives them upwards in the ratio plot and clearly reveals the
downward-directed formic acid peak. The 1:30 and 2:30 spectra show under-
compensation of water, with formic acid still present. The July 26 9:30
a.m. spectrum shows a good water balance, a trace of ozone, and no obvious
absorption due to formic acid or peroxyacetyl nitrate. From a scale-
expanded plot of this same spectrum, we estimated 8 ppb of formic acid
and peroxyacetyl nitrate and 60 ppb of ozone.
The formic acid band appears to lie on top of a weak background of
absorption by other species. This caused some difficulty in determining
the amount of formic acid, but an attempt was made to minimize the error
by measuring only the depth of sharp downward peak.
As Figure 11 has shown, the formic acid concentrations were sightly
higher than the peroxyacetyl nitrate concentrations. This means that
except for ozone, formic acid is present in the highest concentration
of any reaction product seen in these samples of polluted air. Carbon
dioxide, carbon monoxide, and water are also products, but they meld into
the background and are not measurable.
36
-------
LABORATORY
FIELD
JULY 25
9:30 a.m.
10:30 a.m.
11:30 a.m.
12:30 p.m.
1:30 p.m.
2:30 p.m.
3:30 p.m.
JULY 26
9:30 a.m.
J,
1000
1100
WAVELENGTH, cm'1
1.0
0.9
1.0
0.9
1.0
0.9
1.0
LU
0.9 |
1.0 £
oc
0.9 ""
1.0
0.9
1.0
1.0
0.9
1.0
0.9
1200
Figure 15. Detection of formic acid (band indicated by arrow).
37
-------
Laboratory photooxidation studies using the infrared technique
1 2
for analysis have many times shown the formation of formic acid. *
24
The acid is a major product of the photooxidation of formaldehyde.
Thus, there has always been reason to suspect the acid to be present
in the polluted air. However, we believe this to be the first time
it has been conclusively identified and measured. In 1955, it was re-
ported that formic acid had been measured in the Los Angeles air at con-
25
centrations up to 0.41 ppm. The method used involved the reduction of
formic acid to formaldehyde, followed by the reaction of the formaldehyde
with chromatropic acid to yield a colored complex. We suspect the method
to have been subject to error because the amounts of formic acid reported
were about a factor of 5 or 10 greater than the amounts now detected by
the infrared method. There is no reason to assume that the composition of
the atmosphere is Los Angeles has changed so substantially in the past 20
years. Other indexes of pollution, such as carbon monoxide, ozone, and total
hydrocarbons, have not shown such substantial changes. It is interesting to
note that Leighton's 1961 summary of the state of knowledge of atmospheric
7 f\
photochemistry did not give any significant consideration to formic acid
as a product.
Stability is the reason that there is more acid than aldehyde in the
air. The aldehydes are precursors of the acids and have a relatively short
lifetime in the sunlit atmosphere. The acids, especially formic acid, re-
present a stopping place in the sequences of photochemical reactions. Thus,
our observations are consonant with the current measurements that are showing
organic acids to be constituents of the fine particles in the air.
Halogenated Compounds—The fluorinated and chlorinated hydrocarbons all
_i
have strong absorption bands in the frequency region of 700 to 1360 cm
The greater the molecular symmetry, the more likely that the bands will
have a characteristic shape, and the more sensitive the infrared detection.
_T
Thus, we identify carbon tetrachloride by its strong band at 792 cm and
Freon 12 by its strong band at 921 cm . Trichloroethylene and Freon 11
both absorb near 850 cm and are undoubtedly contributors to the per-
sistent absorption band at that frequency. In no cases were the de-
38
-------
tected amounts of a halogenated compound greater than a few parts per
billion. The concentrations did not appear to change in any regular
pattern. Because of their inertness, the compounds will accumulate in a
sluggish urban atmosphere, just as they are accumulating in the atmosphere
on the global scale.
Hydrocarbons—In addition to the selected bands of individual light
hydrocarbons such as methane, acetylene, and ethylene, the spectra show
the combined band at 2970 cm due to C-H stretching vibrations in many
hydrocarbons. As noted earlier, the band does not include methane or
the ring carbons in aromatics. Thus in designating the band as CH ,
X
the subscript x means two or three hydrogens, but not one or four.
The band is useful as a general indicator of the level of hydrocarbon
pollution, as has been previously discussed.
Hydrogen Peroxide—It is logical to expect some hydrogen peroxide to be
present in the smog along with the other oxidants. The peroxide has been
seen as a product in laboratory photooxidations. In addition, it has been
measured by Gay and Bufalini in the air in New York City and in Riverside,
California. They used the chemical method developed by Cohen and Purcell,
in which titanium IV and 8-quinolinol react with the hydrogen peroxide to
form a colored complex. They observed that hydrogen peroxide increased
in the afternoon in a pattern similar to the ozone build-up, but at con-
? fi
centrations only about one-fifth as great. If the hydrogen peroxide were
present at one-fifth the ozone concentration in the Pasadena air on July 24
and 25, it should appear in the spectra.
A reference spectrum of hydrogen peroxide obtained under low
resolution is shown in Figure 16. Water bands are seen on the high-
frequency side. In the laboratory, we have measured the maximum
absorption coefficient of this hydrogen peroxide band to be 9 + 3
atm cm , indicating a fairly strong band, similar to the accessible
bands of ozone and formaldehyde (Table 1). When the atmospheric spectrum
is recorded over a 417-meter path, water and methane both absorb strongly
in the same region as the peroxide, and it is difficult to detect the
peroxide absorption if it is only at a few percent. One can say from
39
-------
1100
1300
WAVELENGTH, cm'1
1500
Figure 16. Ratio spectrum showing possible presence of hydrogen peroxide
Reference spectrum shows 5 ppm V\2®2 'n 1-atm tank air at a path length of 295 meters.
40
-------
inspection of the ratio spectra, such as in Figure 7, that the peroxide
was not present at concentrations of 100 ppb or higher; we wish to go to
a lower detection limit than this.
In order to extract any possible hydrogen peroxide absorption from
the data, we have carefully compared the single-beam plot of 9:30 a.m.
on July 25 with each of the single-beam plots of 1:00 p.m. and 1:30 p.m.
on the same day. The spectra were superimposed so that they matched at
about 1200 cm , and then the ordinates were compared at all the points
between the water lines, from 1200 to 1360 cm . In each case, this gave
us 17 points on a low-resolution ratio plot, showing changes in absorption
that occurred between morning and afternoon. The average of the resultant
two ratio plots is shown in Figure 16, lower half. The appearance of
the peroxyacetyl nitrate absorption band, which we know must be present at
about 1300 cm , assures us that the ordinate comparison technique is
giving a valid answer. From the strength of the 1160-cm peroxyacetyl
nitrate band, we know that nearly all of the 1300-cm band must be due
to peroxyacetyl nitrate. The remaining absorption in the figure may be
ascribed to hydrogen peroxide. The absorption band depth appears to be
about 2 percent, corresponding to about 0.070 ppm hydrogen peroxide. This
identification is by no means totally convincing, but at least one might
sa> that the spectrum indicates the possible presence of hydrogen peroxide,
not at one-fifth the concentration of the ozone, but perhaps at one-tenth.
Methane—Methane shows in both the high-frequency and low-frequency por-
tions of the spectrum. It was always present at concentrations between
1.8 and 2.8 ppm. The methane is not primarily automotive in origin, and
as shown in Figure 10, it does not closely follow the concentration
variations of carbon monoxide and acetylene. The most probable source
TO 09
of methane is natural gas. ' It should be noted that the Raleigh
air spectrum of Figure 14 shows a higher methane concentration than the
Pasadena spectrum.
Methanol—-Methanol, revealed by its band centered at 1032 cm , appeared
in the spectrum occasionally and apparently at random. The maximum amount
-------
detected was about 0.10 ppm, but usually the amount was near or below our
detection limit of 0.008 ppm. Consider, for example, the spectra in
Figure 15. The 9:30 a.m. spectrum shows 0.013 ppm; the 10:30 spectrum
shows none, and the 11:30 a.m. spectrum shows 0.017 ppm; then for the
rest of the day the methanol was again below the detection limit. One
of the clearest methanol spectra is reproduced in Figure 17. In this
1.0
u
z
V)
0.90^
1000
WAVELENGTH, cm
1100
1
Figure 17. Detection of 0.10 ppm methanol (Cr^OH) and 0.021 ppm Freon 12
(CCI2F2); 11:00 a.m., August 23, 1973. Methanol band superimposed on A band due
to 0.075 ppm ozone.
case, the amount was 0.10 ppm. The band is somewhat distorted by an
overlapping band of 0.075 ppm of ozone. Also shown in the figure is
a band due to 0.021 ppm of Freon 12. The random appearances of methanol
in our spectra could have resulted from dumping of this solvent
somewhere, either in industrial operations, or in some type of activity
on the cami'us Clearly, methanol is not a photochemical reaction
product.
Methyl Nitrate—A laboratory photooxidation of hydrocarbon and
nitrogen oxides in air carried out with reactant concentration of one
-------
part per thousand will yield alkyl nitrate as the major nitrogen-
containing product. When the reaction is conducted with concentrations
of 10 ppm, a mixture of alkyl nitrate and peroxyacetyl nitrate will be
formed. At concentrations of 1 ppm, the product is nearly all peroxyactyl
nitrate. Thus, we do not expect to find alkyl nitrates in the smog, and
the spectra confirm this expectation. The strong methyl nitrate bands at
853, 1018, and 1290 cm do not show in any of the spectra. We estimate
our detectability limit to have been 0.010 ppm.
Nitric Acid — Evidence of the presence of nitric acid vapor in the air was
searched for, but none was found. The spectra place an upper limit of ap-
proximately 10 ppb on the level of nitric acid vapor that might have been
present but not detected during the period of observation.
The acid has been detected in the upper atmosphere by means of its
spectrum, but it has never been detected near the earth's surface. Know-
ledge of the level of nitric acid present in the lower atmosphere is neces-
sary for understanding the fate of the nitrogen oxides and the mechanics
of formation of the nitrate found in the aerosols.
It is known that the acid is formed by the interaction of nitrogen
dioxide and ozone:
N°2 + °3 "* N°3 + °2
N2°5 + H2° •
We have followed these reactions in the laboratory by observing the
infrared spectra of the reactants and products. The rate of conversion
of N^Oc to HNO,, depends on the amount of water vapor in the system.
It appears that the conversion takes place not in the gas phase but at
the surface of droplets or on the vessel walls. Nitric acid appears
as a gas, but not with the uniform rate that would indicate a homogeneous
gas-phase reaction. After an induction time, the acid appears in a surge
that is accompanied by the formation of many fine particles in the mixture.
-------
After being formed in the condensed phase, the nitric acid evaporates.
Other possible nitric acid-forming reactions may include nitrogen
trioxide abstracting hydrogen from water or hydrocarbons, and nitrogen
dioxide combining with hydroxyl radicals. The N09—OH combination may be
the source of nitric acid vapor in the upper atmosphere, but at ground
level such free radical reactions would appear to be a minor source of
HNO_ compared to the NO-—0_ interaction.
The nitric acid absorption band most sensitive for atmospheric
_i
analysis is centered at about 880 cm . At this frequency, water vapor
interference is not serious. There are two distinctive "spikes" in the
band, located at 879 and 896 cm . These features would have revealed
the presence of nitric acid with a sensitivity down to approximately 10
ppb.
Figure 18 shows the spectrum from July 25, 1973, 12 noon. The
actual spectrum is drawn in the upper part of the figure. Many trace
molecules were detected, as indicated, but there is no evidence of nitric
acid. The calculated concentrations of detected species illustrate
the detection sensitivity: carbon tetrachloride, 3 ppb; Freon 12, 8 ppb;
ethylene, 6 ppb; methanol, 14 ppb; formic acid, 55 ppb; and peroxy-
acetyl nitrate, 35 ppb. The middle and lower spectra in the figure
show thv. "pper spectrum redrawn with added bands of 20 and 100 ppb of
nitric acid. We believe the absorption peaks would have been discernible
even at nitric acid concentrations as low as 10 ppb. No peaks were seen
in any of the several hundred spectra recorded.
Nitrogen Pentoxide—Nitrogen pentoxide is formed in the reaction of
nitrogen dioxide and ozone. It hydrolyzes to yield nitric acid. We
expect this hydrolysis to take place primarily at the surface of aqueous
fine particles, but no measurements or other experimental data on which
to base predictions of the rate of hydrolysis were available. All we
can say is that a hydrolysis half-life on the order of an hour or
more should allow the nitrogen pentoxide to remain in the gaseous state
long enough to develop a detectable concentration. Actually, none of the
_i
spectra shows the strong nitrogen pentoxide bands at 745 and 1245 cm
-------
30NV11IWSNVH1
45
-------
We estimate the detection limit to have bctn 0.010 ppm.
Nitrous Oxide—Nitrous oxide is an inert constituent of the atmosphere,
uniformly distributed. Although it is an oxide of nitrogen, it takes no
part in the atmospheric chemistry. The only reason that it is mentioned
is that it does appear in the spectra at about 2200 cm . The atmospheric
concentration of the nitrous oxide is a constant 0.25 ppm, and its bands,
along with the bands of carbon dioxide and water, are a permanent part of
the infrared background.
Ozone—The prominent band centered at 1050 cm gives an accurate
measure of ozone, as shown in the figures. Occasionally, a small amount
of methanol absorption was seen superimposed on the ozone band, but the
_•]_
amount of methanol could be judged from the band at 1033 cm , and
a correction could be made. The ozone went through the typical photo-
chemical cycle, as shown in Figure 11, with the highest value being the 680
ppb observed at 1:30 p.m. on July 25, 1973. This unusually high level of
photochemical oxidant will be discussed further in the section on material
balance.
Peroxyacetyl Nitrate—Peroxyacetyl nitrate is seen most easily by its
band at 1165 cm , as shown in many of the spectra. Its concentration
wad verv low at night and in the early morning. During the day, it built
up to a maximum and then declined in a pattern similar to the pattern of
ozone and formic acid concentrations, as shown in Figure 11. Peroxyacetyl
nitrate is a strong oxidizing agent, and it is thermally stable in the gaseous
state. It readily reacts at surfaces, damaging plants and irritating the
eyes. Its reactions with other gaseous pollutants have not been studied
extensively, but it is reasonable to assume that such reactions do take
place, especially in the case of nitric oxide. Details on the properties
and occurrence of peroxyacetyl nitrate in the atmosphere have been given
27
by Stephens.
Material Balance Considerations
The high acetylene-to--ethylene ratio observed on July 24 and 25 leads
one to suspect that the average irradiation time for pollutants in the air mass
was at least several hours. A more accurate estimate of the rate of air
exchange on those days can be made by considering the acetylene and ethylene
46
-------
to be components in a well-stirred flow reactor. The following assumptions
are made: acetylene is nonreactive; ethylene, CE) , has a first-order reaction
rate: -d(E)/dt = k(E); and, for daytime conditions of high ozone and high
photochemical activity, the ethylene half-life is 2 hours. This half-life
was derived from the slopes of the ethylene plots of Figure 10 for July 24
and 25 between the hours of 10 and 11 a.m. and is supported by laboratory
smog-chamber data. It corresponds to a rate constant, k, of 0.35 hr
Consider the air to be well mixed at the sampling point, with an
influx of unreacted polluted air at the ground level, and an outflow
of photochemically reacted air at the top of the inversion. Assume
the entering ethylene to be at 30 ppb (equal to acetylene) , and the
outflowing ethylene to be at 6 ppb, as given by our afternoon measure-
ments. At the steady state we have
Flow in - flow out = ~' - k(E)
at
Taking x as the fractional turnover of the air per hour yields:
(30x - 6x) = 0.35 (6)
x = 0.087 hr"1
The reciprocal of x gives about 11 hours as the time required for one
air mass to pass through the reactor.
In view of the low air turnover, it is of interest to estimate
the total degree of photochemical oxidation of the air sample. The
amounts of acetylene, carbon monoxide, methane, and nonmethane
paraffinic hydrocarbons have been given by the infrared measurements.
Details of the initial atmospheric composition can be filled in by
reference to other work. From measurements at Riverside by Stephens
1 ft 2ft
and Burleson and at Los Angeles by Kopczynski et al. , the con-
centrations of hydrocarbon species were estimated and are given in
Table 4. Likewise, the amount of nitric oxide introduced into the air
mass was estimated through reference to the profile of air contaminant
emissioiB issued by the Los Angeles Air Pollution Control District in
29
January 1971,
The requirement for peroxy radicals, RO , is derived from the
following considerations. It is firmly established that the 0 is
47
-------
Table 4. MATERIALS IN THE PHOTOOXIDATION PROCESS,
AVERAGED FOR JULY 24 and 25, 1973
Product
Reactant
RCL
Maximum
Pollutant
maxima availability
requirement availability
Ozone
Nitric Oxide
Peroxy acetyl nitrate
640
50
350
640
450
100
(1190 Total)
Formic acid
Acetylene
Methyl Acetylene
62
35
1
Olefins
Ethylene
Propylene
1,3-butadiene
1-butene
Isobutene
Trans-2-butene
Cis-2-butene
2-methyl butene-1
Cyclopentene
Trans-2-pentene
2-methyl butene-2
Total aromatics, assuming
eight carbons per molecule
Total nonmethane paraffins,
assuming six carbons per
molecule
35
10
1.8
1.3
2.6
0.7
0.7
1.2
2.2
1.2
1.3
63
170
140
60
14
10
21
6
6
12
22
12
13
(316 Total)
1000
2000
Composition of olefin fraction from the September 24, 1968, data of Stephens and
Burleson ; nitric oxide estimate from Los Angeles Air Pollution Control District
Emissions Profile; aromatics and paraffins estimates from Kopczynski et al.^9
formed from the photolysis of NO^:
NO- + NO -> NO + 0
0 + 02 + M -> 0 + M
It is also known that the 0~ reacts very rapidly with the NO to
-------
regenerate NO-:
NO
The principal role of hydrocarbons in the 0_ buildup is to supply an
alternate path for oxidation of NO to N02 via peroxy radicals:
R02 + NO -> RO + N02
R may be a saturated or unsaturated hydrocarbon radical and may be
partially oxygenated. The RO radical can continue in subsequent
reactions. If R is large, the RO radicals can be oxidized to some
kind of R02 radical one or more additional times.
The 0. reacts with NO coming into the system, and then with
the resulting N0_ to remove it from the system:
0 + NO -> 0 + N0
N20 + H20 (at surface) -> 2 HNO.
For each two NO molecules converted to HNO«, three 0~ molecules are
removed, using up to three 110 „ radicals. Formation of a peroxyactyl
nitrate molecule uses two R09 radicals:
NO + RO = NO + RO
CH3C(0)00 + N02 = CH3C(0)OON02
It is assumed that all the NO ends up as HNO_, nitrate salts, or peroxy-
acetyl nitrate. Overall, the number of reacted R0« radicals equals the
total of: (1) the number of 0 molecules accumulated; (2) 1-1/2 times
the number of NO molecules that do not go into peroxyacetyl nitrate but
disappear from the system, presumably into HNO or nitrate salts; and (3)
2 times the number of peroxyacetyl nitrate molecules formed. Loss of
0- in reaction with hydrocarbons is not counted because that reaction
probably regenerates R0_ radicals.
The number of RO. radicals available from each organic molecule
can be estimated. The number actually generated will depend on how
49
-------
far the oxidation proceeds. Without considering the detailed oxida-
tion mechanism, it can be stated that the number of RO radicals to
be derived from each carbon is probably not more than two. A carbon
atom that begins as a -CH^- group and ends as a CO molecule probably
serves as the oxygenated carbon in an RO- group only once. If the
carbon atom ends up as C09 or as HCOOH, it may have been through
the RO- phase twice. Hydroperoxy radicals (HO-) may also form, but
it is not easy to say how frequently. Only a. weakly bound hydrogen
will be abstracted by 0- to yield H0_. The more strongly bound hy-
drogens can be abstracted by OH, 0, and RO, which do not yield HO-.
In summary, therefore, if each carbon atom is assumed to be able to
serve twice as the oxygenated carbon in an RO- radical, then some
CO, HO-, aliphatic acid, peroxyacetyl nitrate, and CO- can be formed.
The figures listed in the table under "Maximum RO availability" are
just twice the number of carbons for each molecule listed.
The low measured values of ethylene indicate that olefins were
nearly all reacted. Let it be assumed, as shown in Table 4, that 316
ppb of the required 1190 ppb of RO- came from olefin oxidation. The
aromatics as a class are somewhat more reactive than the paraffins, as
28
shown by Kopczynski et al. Since there probably was about twice as
much paraffiriic materials as aromatic, it seems fair to estimate that the
remaining 874 ppb of required RO- came half from aromatics and half from
paraffins. Overall, a 20 to 30 percent depletion of carbon by oxidation
is estimated. This is borne out by the two measurements of C-H absorption
in the late afternoon on July 24 and 25. These points, marked on Figure
9, both have slightly higher CO/CH ratios than the average.
X
In summary, the data show a substantial degree of oxidation of
the nonmethane hydrocarbon matter in the air on July 24 and 25.
Olefins and aldehydes can be assumed to have been nearly completely
reacted. The degree of oxidation of other compounds undoubtedly
followed their photochemical reactivities. Highly substituted
aromatics and branched-chain paraffins were probably fairly well
oxidized, and straight chain paraffins, benzene, methane, and
50
-------
acetylene were probably only slightly oxidized.
The high level of photochemical oxidation at the sampling loca-
tion in the July 24 and 25 period is demonstrated by a comparison of
the ozone-to-hydrocarbon ratios with those shown in the Environmental
Protection Agency document Air Quality Criteria for Hydrocarbons,
Figure 5-3. That figure shows the measured maximum 1-hour ozone
concentrations plotted against average nonmethane hydrocarbon con-
centrations for the period 6 to 9 a.m. From our measurements on
July 24 and 25, we get a 6 to 9 a.m. total nonmethane hydrocarbon
concentration of 1.25 ppm. This is derived from our measured
nonmethane paraffinic carbon values with a 30 percent increment to
correct for the low absorption coefficient of aromatics and small
olefins. From Figure 11, we get 0.57 and 0.66 as the maximum 1-
hour average ozone concentration. This yields ozone-to-hydrocarbon
ratios of 0.45 and 0.53. Each of these ratios is several times
higher than the ones shown in the criteria document. There, the
three highest ozone values are shown to have been achieved with
ozone-to-hydrocarbon ratios of 0.12, 0.11, and 0.13. Two reasons
are seen for the abnormally high ratios in the present work: (1)
our sampling point was well above the street, in contrast with the
street-level measuring points of many monitoring stations; and
(2) the 24th and 25th were days of unusually high photochemical
activity.
We now consider the fate of the 350 ppb of nitric oxide which
must have entered the air mass along with the measured inert pol-
lutants. Since peroxyacetyl nitrate, at 50 ppb, was the only nitrogen-
containing compound detected, the remaining 300 ppb of nitric oxide
must have ended up either (1) in the particulate matter and gaseous
species that were below the detection threshold, or (2) in the veg-
etation, soil, and other surfaces.
The nitric oxide equivalents of the measured or allowed compounds
are easily enough totaled: 50 ppb are in peroxyacetyl nitrate, 10 ppb
51
-------
might be in nitric acid, 5 ppb might be in alkyl nitrates, and 20 ppb
might be in nitrogen pentoxide. Practically no free nitric oxide
would exist because of its fast reaction with ozone. If nitric oxide
must be absent, then its equilibrium partner, nitrous acid, must also
be absent. Let us estimate a 2 ppb maximum for each.
High levels of nitrogen dioxide are also disallowed because of
reaction with ozone. An NO. level can be calculated from the stirred-flow
reactor approximation with the aid of published rate data.
Flow in - flow out = ~d(N(V
dt
The N02 is oxidized to NO,, in two steps:
N2°5
31
According to data quoted by Leighton, the first reaction determines
the rate by being the slower of the two, with a bimolecular rate con-
Con-
stant at 25° C somewhere between 0.010 and 0.002 ppb~ hr .
sidering that two NO. molecules are removed for each 0,. reacted and
using an average 0_ concentration of 400 ppb, we have:
Flow in - flow out = 2 x
0.010
to
0.002
x 400 x (N0_)
to
16
x (N0_)
We now get the flow in and flow out by using the fractional air turn-
over of 0.087 hr previously calculated from the ethylene-to-acetylene
ratio:
0.087 x 350 - 0.087 x (N02>
to
16
Lower estimate of NO. = 3.8 ppb
Upper estimate of N00 = 19 ppb
x (NO.)
52
-------
The above estimates consider only the reaction of nitrogen dioxide
with ozone; reactions with radicals and molecules other than ozone
would further reduce the nitrogen dioxide level.
We believe that there is no point in attempting to compare this
estimate with measurements of nitrogen dioxide in the July 24 and
25 afternoon air mass, because the measurement methods in use do
not give believable answers under afternoon smog conditions. One
method which, in our opinion, would give believable answers is the
recently developed laser fluorescence method, in which blue laser
light is absorbed by the nitrogen dioxide and emitted as red light.
This method has been shown to be highly sensitive, proportional to
nitrogen dioxide, and free from interference. In an afternoon smog
at El Segundo, California, Tucker et al. showed the nitrogen dioxide
32
concentration to be not more than a few parts per billion. We
believe that their measurement method would have given a similar
answer if applied to the air at our sampling site.
The amount of nitrate in the atmospheric fine particles can only
be roughly estimated, but even a liberal estimate cannot account for
more than a small part of the total nitric oxide that entered the air.
In an intense smog, there might be 200 micrograms of particulate
matter per cubic meter. If this were as much as 30 percent nitrate,
b;, weight, it would only be equivalent to about 20 ppb of nitric oxide.
The sum of all the possible amounts of combined nitrogen is
equivalent to only about 130 ppb of nitric oxide. Thus, about 220 ppb
are not accounted for, and may have been removed at the surface. These
conclusions are summarized in Table 5.
If the estimate of 350 ppb of nitrogen oxides (NO ) obtained from
the Los Angeles emission profile was too high, then the missing fraction
of nitrogen oxides would be correspondingly smaller. However, even if
the estimate was 100 percent too high, we would still have a substantial
fraction of nitric oxide not accounted for.
If the nitrogeneous compounds are not present in the air either in
gaseous or particulate form, then they must have been removed at the
surface. Such removal processes need to be studied further both in the
field and in the laboratory.
53
-------
Table 5. NITROGEN-CONTAINING COMPOUNDS3
Amount Maximum
Compound detected, ppb allowed, ppb nitrogen
Peroxyacetyl nitrate 50
Alkyl nitrate 5
Nitric oxide 2
Nitrous acid 2
Nitric acid 10
Nitrogen pentoxide 20
Nitrogen dioxide 20
Particulate nitrate 20
Totals ~50 79
Nitric oxide entering air mass: 350 ppb.
Nitric oxide not accounted for: 221 ppb.
Future Work
A few improvements in the infrared detection technique used in
this work should yield substantial further progress. Although the
detection threshold for many pollutants has been reduced to a level
near 10 ppb, this is still about a factor of 10 higher than the
limits of detection claimed for the method in a previous paper.
One major source of difficulty in this work was the unsatisfactory
working environment, which was too shaky, too cramped, and too subject
to temperature fluctuations. The remedy for this is simply to set up
the optical system in a fully air-conditioned room away from machinery,
with ample space for the optical components. This will allow maintenance
of proper optical alignment for maximum throughput, minimum spectral
54
-------
noise, and less shifting of interference fringes. These environmental
improvements alone should double or triple the sensitivity.
Another limitation of the present work was the use of imperfect
reference air. Carbon dioxide interference was never fully removed
in the ratio plots, and nitrous oxide interference was not removed
at all. Furthermore, the balance of the water lines in the ratio
spectra was not always ideal because of the temperature fluctuations
in the working environment. These errors can be significantly re-
duced in future studies. We recommend that vaporized liquid nitrogen
be used in place of tank "zero air." Measured amounts of carbon
dioxide and nitrous oxide gases should be carried into the long-path
cell with the nitrogen. Humidification can be achieved by the "wet
bag" method already developed in the present work.
Finally, it appears that there is still much to be gained in
more fully utilizing the capabilities of the Fourier transform
spectrometer system. Tape storage of the ratio spectra will allow
retrieval and processing at a later time. The plotting of one ratio
spectrum against another will reveal small increases or decreases
in pollutant concentrations. A much greater number of scans should
be added together in recording the spectra, thus raising the signal-
to-noise ratio. The use of a copper-doped germanium detector at
liquid helivm temperature promises two benefits: it will cover the
whole spectral region from 300 to 3500 cm on each scan, and it
will have higher detectivity than the detectors previously used.
55
------- |