SUMMARY REPORT
JUNE 1973
HYDROCARBONS IN POLLUTED AIR
COORDINATING RESEARCH COUNCIL
PROJECT CAPA-5-68
Edgar R. Stephens, Principal Investigator
Professor of Environmental Science
Statewide Air Pollution Research Center
University of California
Riverside, California 92502
UNIVERSITY OF CALIFORNIA
STATEWIDE
AIR POLLUTION
RESEARCH CENTER .
RIVERSIDE CALIFORNIA 92502
-------
SUMMARY REPORT
JUNE 1973
HYDROCARBONS IN POLLUTED AIR
COORDINATING RESEARCH COUNCIL
PROJECT CAPA-5-68
Edgar R. Stephens, Principal Investigator
Professor of Environmental Science
Statewide Air Pollution Research Center
University of California
Riverside, California 92502
-------
CONTENTS
PAGE
SUMMARY AND CONCLUSIONS ii
PROCEDURES & ANALYSIS 1
Freeze Out Trap 2
Calibration 7
Precision 9
Sensitivity 9
Sampling 10
Irradiation 12
Stirred Flow 13
RESULTS IN AMBIENT AIR 20
Smog Hydrocarbons 20
Background Hydrocarbons 23
Unreacted Air 25
Other Sources 35
Gasoline Vaporization 41
Hydrocarbons in Other Locations 48
HYDROCARBON REACTIVITY
Ambient Air Irradiation 54
Auto Exhaust Irradiation 59
Extent of Reaction 59
MODES OF REACTION 67
REFERENCES 75
-------
LIST OF FIGURES
PAGE
1. Gas Chromatograph 1
2. Chromatogram of Ambient Air Hydrocarbons 5
on Dimethyl Sulfolane Column
3. Chromatogram of Ambient Hydrocarbons on Porapak N 5
4. Ambient Air Chromatogram of Aromatics (Number 3) 6
5. Bottle Photoreactor 13
6. Photolysis of NO- 14
7. Flow Photoreactor 14a
8. Meatless Adsorption Air Purification 15
9. Purging Bottle Reactor 18
10. Purging Bottle Reactor 19
11. Hydrocarbons in Polluted Air 20
12. Pure Mountain Air 23
13. Hydrocarbons in Unreacted Morning Air 25
14. Methane and Acetylene in Morning Air 27
15. Hydrocarbons in Morning Air 28
16. Auto Exhaust Components in Morning Air 29
17. Natural Gas in Morning Air 30
18. Natural Gas 31
19. Auto Exhaust 33
20. Gasoline 34
21. Blow By Gas 35
22. Brush Fire Smoke 36
23. Oil Field Air 36
24. Hydrocarbon Distribution Near Sources 37
25. Hydrocarbons in Polluted Air 38
26. Smog and Heat in Riverside 39
27. Hydrocarbons in Polluted Air (Hill) 40
28. Hydrocarbons in Polluted Air (Roof) 40
29. Comparison of Hydrocarbon Distributions 41
30. Vapor/Liquid Computation 46
31. Gasoline Vapor 47
32. Vaporized Gasoline 47
33. Photoreactivity of Hydrocarbons 51
-------
LIST OF FIGURES (Cont)
PAGE
34. Hydrocarbons in Morning Air 54
35. Natural Sunlight Irradiation of 56
Ambient Air 10-24-66
36. Ultraviolet Irradiation of Ambient 56
Air 10-24-66
37. Sunlight Irradiation of Ambient Air 56
Hydrocarbons Reacted 10-24-66
38. Ultraviolet Irradiation of Auto Exhaust 56
39. Ambient Air Photolysis 10-24-66 61
40. Ambient Air Photolysis 12-21-67 61
41. Ambient Air Photolysis 62
42. Extent of Photolysis of Ambient Air 63
43. Ultraviolet Irradiation in 20 Liter Bottle 70
-------
LIST OF TABLES
PAGE
.. 1. Trapping Efficiency 3
2. Columns and Chromatographic Conditions 4
3. Riverside Smog , 8
4. Ultraviolet Irradiation of Ambient Air 10
5. Hydrocarbon Removal by Two Column 16
Heatless Fractionator
6. Hudrocarbons 21
7. Methane in Clear Mountain Air 24
8. Natural Gas in Ambient Air 32
9. Molar Ratio of Ethene (C H ) to Acetylene 32
(C.H_) and Propene (C_H,) to Acetylene in
2. 2. Jo
Various Samples of Unreacted Air
10. Hydrocarbons - Vapor, Liquid 48
11. Relative Rates of Disappearance of 58
Hydrocarbons Under Irradiation
12. Estimate of Amounts Reacted in an Ambient 66
Air Sample Compared to an Unreacted Sample
13. Riverside Ambient Air (PPB) on Roof of 66
Health-Finance Building
14. PBzN Formation at High Concentrations 71
15. Reactivity of Aromatic Hydrocarbons in the 72
Flor Photoreactor
16. Reactions of Nitrogen Trioxide 74
-------
Forward:
This project was begun in July 1968 as a continuation of earlier
studies on ambient air hydrocarbons supported by a grant from the American
Petroleum Institute. Some of the work has been reported in the published
literature (see reference list). The data presented in these papers is
summarized but not reported in detail in this report. The following were
members of the project advisory committee: R. G. Larsen, H. E. ALquist,
P. R. Ryason, C. S. Tuesday, A. P. Altshuller, J. C. Neerman and J. M.
Heuss.
Acknowledgment
Mr. Frank Burleson did the bulk of the laboratory work reported
here. Long path infrared studies were carried out by Mr. Monty Price
and Mr. William Snider. Mrs. Minn Poe developed the computer program.
We are also indebted to those persons who collected samples in distant
locations.
-------
11
Summary & Conclusions
Flame ionization chromatography provides a satisfactory procedure
for analysis of hydrocarbons in polluted air. By freezing out the
hydrocarbons present in a 100 ml sample of polluted air, concentrations
as low as 0.1-0.2 ppb can be detected. With careful calibration and
attention to detail, accuracy and precision of a few percent can be
attained at levels of a few tens of ppb. Three instruments, with three
different columns, were used to analyze about three dozen hydrocarbons.
The small sample requirement permitted shipment of samples from many
locations around the country and some from overseas.
The two main objectives of the study were to compare the distribu-
tion of hydrocarbons in the atmosphere with those observed in sources
and to estimate the degree to which hydrocarbons of different reactivity
participate in the photoreaction. Typical photochemical smog hydrocar-
bons can be described as auto exhaust plus gasoline vapors plus natural
gas (in addition to background methane of 1.39 ppm). Spilled gasoline,
as opposed to vaporized gasoline, does not seem to play a significant
role. Samples taken near specific sources (petrochemical plants, brush
fires, oil fields) show distinctly different distributions. Afternoon
smog samples show substantial attenuation of the more reactive hydrocar-
bons (specifically the olefins). By studying the relative amounts of
acetylene, ethene and propene it was possible to estimate the extent and
time of reaction. It was concluded that oxidant smog as seen in Riverside
has undergone six to eight hours of photochemical reaction. Irradiation
produces continued reaction for 24 hours or more. In six to eight hours
of irradiation even hydrocarbons of low reactivity participate in the
reaction; after the first hour or so the olefins of high reactivity are
gone and further reaction depends on hydrocarbons of medium and low
reactivity present in much larger quantity. The practical conclusion is
that reduction in the already low concentrations of the highly reactive
olefins will not have much practical effect.
During the course of this research program, reports came out which
indicated that the higher aromatics (CL ) were much more reactive than
previously thought. Irradiation of ambient air mixtures confirmed the
-------
iii
modernte reactivity of these compounds. Meanwhile it was reported that
benzilic hydrocarbons could be oxidized under atmospheric conditions to
the potent eye irritant peroxybenzoyl nitrate. This suggested that one
active species attacks the aromatic side chain while some different
species attacks the ring. Tests were made to see if the side chain
attack could be due to nitrogen trioxide. High concentration experiments
in a long path cell indicated that this could be so but yields were small
and rates were slow. No reaction could be detected at realistic (ppb)
concentrations.
-------
PROCEDURES & ANALYSIS
Three gas chromatographs are used with four procedures to
quantitate about three dozen individual hydrocarbons in ambient air
samples. Three of the four procedures require a freeze out of the
hydrocarbons from a 100 milliliter air sample. The fourth procedure
uses a 3 milliliter loop sample to analyze methane. Figure 1
illustrates the freeze out procedure. Samples are injected with a
100 milliliter glass medical syringe into a gas sample valve on
which the normal sample loop is replaced by a freeze trap packed
with chromatographic packing. No attempt is made to remove either
water or carbon dioxide from the sample prior to freeze out. A
bubble meter or syringe on the output side of the gas sample valve
can be used to verify the fact that all the sample has, indeed,
passed through the trap. Liquid oxygen is used to chill the packed
freeze trap since it has a temperature sufficiently low to collect
even such hydrocarbon gases as ethane, ethylene and acetylene, but
not low enough to collect significant quantities of oxygen. If liquid
nitrogen were used as the coolant oxygen would be collected from the
GAS CHROMATOGRAPH
PACKED
FREEZE
TRAP
4
f~~~
U
n
\
-
=
~
m
>
BUBBLE
METER
V T
LIQUID O/
Ml * ii -^=3.
™n
Q*FLAME DETECTOR
1
1
"1
I
- 1 1
1 l 1
•!in i
-H 1
COLUMN
i
^^k
f
^=^
'
ICE BATH
in DEWAR
«-O A DDICD
L-Ljj 1 GAS
\
100 ml SYRINGE GAS SAMPLE
VALVE
FIGURE 1
-------
air sample and injected into the column along with the hydrocarbons.
This would snuff the flame. Liquid argon could be used since its boil-
ing point is very close to that of liquid oxygen. A third alternative
would be to chill with liquid nitrogen and then to purge the trap
with nitrogen or with helium prior to thawing and injection into the
column. Injection into the column is accomplished by thawing the
trap and simultaneously pushing the gas sample valve. Samples are then
carried quickly into one of the three columns. Instrument No. 1, with
the freeze out procedure, gives C through approximately C hydrocarbons
J D
plus acetylene. Instrument No. 2, with the freeze out procedure, gives
ethane, ethene and acetylene. To measure methane, the freeze trap on
this instrument is replaced by' a sample loop and a 3 milliliter
sample directly injected. The third instrument gives principally
aromatic hydrocarbons plus acetaldehyde, propionaldehyde, acetone
and methyl ethyl ketone. Alpha and beta pinene also emerge from
this column, but they are not well separated from the aromatic hydro-
carbons. Table 2 summarizes the columns and chromatographic procedures.
Freeze Out Trap
Freeze out traps were prepared by packing 20 cm lengths of stainless
(2.34 mm I.D. [1/8 inch OD]) steel tubing with a chromatographic substrate
consisting of 10% dimethyl su^folane on 42/60 mesh C-22 firebrick or with
glass heads. Each tube was bent into a "U" shape and attached to the gas
sample valve of a chromatograph in place of the sample loop. The sample inlet and
outlet of each sample valve were equipped with toggle valves. Although these
freeze out methods had been found to be highly efficient by other workers
it remained to be proved that the specific technique in present use was
also quantitative. This was done by analyzing a synthetic mixture
containing about 1 ppm of each of several hydrocarbons, first using a
direct sample loop injection, then by diluting the same volume of the
mixture to 100 ml and using the freeze trap. The area under each peak
was measured and divided by the carbon number to obtain a sensitivity
factor as shown in Table 1. Although these trap samples were consist-
ently smaller than the corresponding peaks from the loop injection by
about 15%, the fact that the area per carbon atom was independent of
-------
hydrocarbon volatility suggests that this discrepancy results from
inaccuracy in the measurement or delivery of volumes rather than a
failure to capture all the hydrocarbon. Dead space in the valve
and connecting tubes may account for this difference. In any event
the freeze-out procedure was used for calibration and quite repro-
ducible results were obtained, as can be seen by examining some
replicates in Table 4.
Table 1 —Trapping Efficiency—Area
per Carbon Atoms (Si) for Loop
Sample and Two Trapping
Techniques
Trap Trap
Com- (pull) (push) Loop Si (loop)
pounds Si Si Si Si (trap)
Ethane 0.153 0.153 0.169 1.10
Ethene 0.142 0.143 0.164 1.14
Propaae 0.145 0.146 0.169 1.16
Propene 0.131 0.132 0.158 1.19
Total 0.571 0.574 0.660
Propane 0.113 0.111 0.115 1.02
Propene 0.109 0.104 0.134 1.23
ri-butane 0.113 0.114 0.129 1.14
Acetyl-
ene 0.145 0.132 0.157 1.07
1-butene 0.115 0.107 0.133 1.16
lru.na-1-
butene 0.114 0.106 0.129 1.13
cis-2-
butene 0.115 0.106 0.131 1.14
1,3-buta-
diene 0.112 0.102 0.127 1.13
Methyl
acetyl-
ene 0.109 0.101 0.122 1.12
cis-2-
Pen-
tene 0.125 0.109 0.153 1.22
Total 1.170 1.092 1.3IM)
Xote: First 4 compounds separated on
chromatOf$raph No. 2; remainder on
chromatograph No. 1.
Loop: 2.9 ml.volume
Trap: 2.9 ml diluted to 100 ml N- and
frozen in trap
-------
TABLE 2. Columns and Chromatographic Conditions
NUMBER 1
Column: 11 meters by 2.34 mm I.D. (1/8 inch OD stainless tube)
containing 10% dimethyl sulfolane on 42/60 mesh C-22 firebrick
maintained at 0 C plus 0.46 meters by 2.34 mm I.D. stainless tube
containing 10% carbowax E-600 on 42/60 mesh C-22 firebrick at room
temperature to minimize background due to column bleed. Background
signal at full sensitivity was about 4.3 mv. Carrier gas: 22 ml/min
of dry nitrogen. Flame detector: 24 ml/min hydrogen and 287 ml/min
oxygen. Wilkens Model 600 chromatograph.
This column gave good separation of acetylene, C and C, olefins
and paraffins, C_ paraffins, some of the C,. olefins and Cfi paraffins,
as well as 1,3-butadiene, methyl acetylene and cyclopentane. Figure 2
NUMBER 2
Column: 1.52 meters by 2.34 mm I.D. (1/8 inch O.D. stainless
tube) containing activated alumina held at 60 C. Carrier gas: 40
ml/min nitrogen. Flame detector: 40 ml/min hydrogen and 287 ml/min
oxygen. Wilkens Model 600 chromatograph. Background signal at full
sensitivity was 4.3 mv. Retention times and .therefore peak heights
were variable depending (apparently) on the moisture content of the
carrier gas. This was therefore replaced by the following:
Column: 1.52 meters by 2.34 mm I.D. (1/8 inch O.D.) filled with
Poropak N (100/120 mesh), operated at 60°C with a carrier flow of 80
ml/min of nitrogen and 60 ml/min of H~ and 410 ml/min of 0 . This gave
a good separation of ethane, ethene and acetylene in a two minute
chromatogram. Methane is not frozen out by these procedures but could
be determined by direct injection of an air sample using a three ml loop.
Since the worldwide background level of methane is nearly 1.4 ppm
there is no need for the high sensitivity provided by the freeze-out
procedures. This does mean, however, that a third chromatogram is
required. Air samples gave negative responses, apparently due to
oxygen, just ahead of the methane peak. Comparison of calibrations
in nitrogen and in air showed that this caused no interference with
the methane measurement. Figure 3
-------
CONCENTRATIONS IN ppJ
1 mv -I mv
# » 2 H * B
• w
u. l M
5 10
;; 5 -, = !S « * -8
b b b * '- o e> o i.
IS
10
25
30
Fig. 2. Chromotogrom of ambient air hydrocarbons on dimethyl uilfolan* column.
100ml
ethene x8
43.2 ppb
ethane x4
28.8 ppb
5 PORAPAK N
acetylene x8
60 ppb
methane x 16
2388 ppb
2.9ml
MINUTES
Fig. 3. Chromatogram of Ambient Hydrocarbons on Porapak
-------
6
TABLE 2 (continued)
NUMBER 3
Column: 3.048 meters by 2.34 mm I.D. (1/8 inch O.D. stainless
tube) filled with 5% of 1,2,3 tris (2-cyanoethoxy) propane plus
1.52 meters by 2.34 mm I.D. tube filled with 5% bentone34 and 5%
dinonyl phthalate both on 100/120 mesh HMDS treated gas chrom W
plus 0.915 meters by 2.34 mm I.D. filled with 10% carbowax E 600
on C-22 firebrick 30/60 mesh. Carrier gas was 38 ml N^/min. Fuel
flow: 25 ml H2/min with300 ml/min of oxygen. A freeze out trap
containing uncoated glass beads was used with this instrument.
It was used to measure aromatic hydrocarbons through C9. Alpha and
beta pinene, acetaldehyde, propionaldehyde, acetone and methyl
ethyl ketone could also be measured. See Figure 4.
I.I.* TRMffnirL UNZBNB 1.4
I.I TlMCTtYL UKZB4I O.I
FIGURE 4
Ambient Air Chromatogram of Aromatics (Number 3)
-------
Trapping efficiency was also verified by collecting a sample which
had been passed through the freeze out trap in a second 100 milli-
meter syringe. This sample was then re-analyzed and found to
contain only 0.2 ppb of ethane and ethene, although the input sample
had contained 7.2 ppb and 8.0 ppb respectively of these hydrocarbons.
The re-analyzed sample also contained fractional ppb concentrations
of oxygenates.
Calibration
Emergence times are used to identify individual hydrocarbons
on the chromatograms and peak heights are used to estimate concen-
trations which have always been reported in parts per billion by
volume. A transparent overlay chart is used for rapid reading of
peaks.
-------
Calibration mixtures are prepared by successive dilutions of
pure hydrocarbons into modified two liter Erlenmeyer flasks. These
flasks have a stopcock attached to the drawn down neck, a three-way
stopcock near the base and an injection port made from a ball and
socket joint in the body of the flask.
Two such flasks are flushed with pure air or nitrogen. Two
ml of each gaseous hydrocarbon is then added to the first flask from
a small syringe. This gives about 1000 ppm in this flask. Liquid
hydrocarbons are added from microliter pipettes. The amount is
calculated from the handbook value of the density and the molecular
weight. After mixing, two ml of this mixture is transferred to the
second flask with a ground glass medical syringe. A number of
hydrocarbons can be injected into the first flask at about 1000 ppm.
A single transfer to a second flask then will yield a mixture of
those hydrocarbons at 1 ppm. Although the flasks are open to the
room air during the transfer of gases, there is no significant loss
to the atmosphere if the flask's temperature and pressure are the
same as that of the room. Five ml of this mixture diluted to 100 ml
in a syringe then produces a calibration mixture containing about
50 ppb of each hydrocarbon. (The exact concentration will, of course,
depend on the exact volumes of liquid and gas taken and the exact
volumes of the dilution flask.)
TABLE 3
RIVERSIDE SMOG 1610 PST 10-24-68
(TWELVE SAMPLES, PPB)
Instrument 1 Instrument 2
C a C a
PROPENE 1.27 0.1 METHANE 2530 30
PROPANE 49.9 2.2
ISOPENTANE 41.2 1.5 ETHANE 72.2 2.3
N-HEXANE 8.3 0.4 ETHENE 17.9 1.2
ACETYLENE 41.3 1.3 ACETYLENE 42.8 1.5
-------
Precision
Table 3 shows the precision which can be attained with these
methods. This table gives the average concentration (C) and the
standard deviation (a) for a set of twelve ambient air samples of
unusual uniformity.
At the highest concentrations, the relative standard deviation
is just over 1% (CH,). In the range of 10's of ppb, it is 3 to 5%
while at the 1 ppb level it is near 1070. This of course also depends
on the particular peak. For propane at 1.27 ppb, the standard
deviation was 0.1 ppb. These standard deviation values reflect both
the measurement precision and the variation between the twelve samples.
If the acetylene values from the two instruments differ by more than
about 10% at the 40 ppb level, a search is begun for some malfunction
of one of the instruments. In Table 3 the acetylene values differ
by only 1.5 ppb (about 47»).
Stability of hydrocarbon concentrations while stored in these
sample tubes was verified by repeating analyses after one or more
days of standing. Table 4 compares concentrations of hydrocarbon
samples taken in 20 liter carboys. These two samples were taken on
the second .floor and on the roof of the Riverside County Health-Finance
Building. The roof top sample was stored in the dark and showed no
change after 24 hours. The second floor sample was nearly identical
and was irradiated 24 hours with the results shown in Table 4.
Sensitivity
When operating properly, minimum detectable peaks for short
emergence time compounds corresponds to 0.1-0.2 ppb (for 100 ml
freeze out). Compounds with longer retention times have broader
peaks and therefore the minimum detectable concentrations are a
few tenths or one ppb. This decrease in sensitivity with emergence
time would be more pronounced if concentration were expressed by
weight. Larger samples could easily be used to attain higher sensi-
tivities.
-------
10
Tobla 4 —Ultraviolet Irradiation of Ambient Air Riverside County Building of Health and Finance, 3/3/66,08:05-
08:25 PST, Sky Clear, Wind 5-7 mph, N.W., 40-45°F, Moderatiy Heavy Haze (Smudge)
Compounds
Ethane
Ethene
Propane
Pro perm
Propane
Propeue
Isobutane
rt-butane
Acetylene
1-butene
Isobatene
pentene
Roof
Top
Sample
3<>.o
45.6
14.0
14.0
12.0
14.0
7.0
41.6
62.4
2.0
4.0
.1.0
2.5.4
1.0
17.0
2.0
2.0
0.6
1.4
2.0
l.S
~0.6
7.6
6.8
•2.0
6.6
2.4
45 min
Dark
38.0
45.6
14.0
13.8
12.4
13.8
7.0
41.6
62.4
2.0
4.0
1.2
25.4
1.2
17.0
2.0
2.0
0.6
1.2
2.4
1.6
~0.6
7.6
7.4
2.4
6.4
2.2
24 hr
Dark
37.6
48.8
15.0
15.4
12.2
11.6
6.8
41.6
61.6
2.0
3.4
1.2
25.2
1.2
16.8
2.0
2.0
0.6
1.4
2.0
1.6
—
7.4
6.8
1.8
6.6
2.2
2nd
Floor
Sample
37.6
51.2
14.6
15.6
13.0
13.4
7.4
44.8
69.6
2.2
4.0
1.4
27.6
1.0
18.6
2.4
2.0
0.6
1.0
2.0
1.6
~1.2
8.4
8.0
1.4
7.2
2.4
-Concentration in ppb
1 hr 2 hr
Irrad. Irrad.
uv uv
37.6
45.6
14.2
11.0
12.8
9.4
7.0
43.2
69.6
1.4
2.4
0.4
26.0
0.4
17.4
1.0
2.0
0.4
0.8
0.8
0.6
~0.7
7.6
4.8
l.S
6.6
1.4
.'!<>. 0
40. S
. 12.6
7.6
12.2
6.6
6.8
40.8
65.6
1.0
1.4
0.0
24.6
0.0
16.2
0.6
1.8
0.4
O.S
0.4
0.0
~0.9
7.4
4.2
2.0
5.8
1.0
4hr
Irrad.
uv
37. «
:?6.o
13.6
4.4
11.8
3.8
6.6
39.2
64.8
0.6
0.8
0.0
22.8
0.0
15.4
0.0
2.0
0.4
0.4
0.0
0.0
~O.S
6.6
3.6
1.6
5.6
0.4
8hr
Irrad.
uv
36.0
27.2
12.8
2.2
11.2
1.4
5.8
33.1
60.0
0.4
0.6
0.0
19.2
0.0
12.4
0.0
1.4
0.2
0.0
0.0
0.0
~0.6
5.7
3.4
1.4
4.6
0.6
20 hr
Irrad.
uv
35.2
1(1. 0
12.2
1.4
10.2
O.S
5.0
28.0
56.8
0.4
0.6
0.0
14.4
0.0
9.4
0.0
1.4
0.2
0.0
0.0
0.0
~0.8
4.0
2 2
O.S
3.0
0.4
24 hr
Irrad.
uv
35.2
12.6
114
1.4
10.4
0.6
4.S
27.2
56.8
0.4
0.4
0.0
13.6
0.0
8.8
0.0
1.2
0.4
0.0
0.0
0.0
~1.0
3.6
2.2
1.0
3.4
0.0
.Vy/f.v First 4 compounds separated on chromatograph No. 2; remainder on chromatograph No. 1.
» Contains 3-methyl butene-l.
11 Contains CM-2-pentene.
c Contains 2-methvl butene-2.
Sampling
In the first work not enough attention was paid to the influence
of sampling. While it is possible to measure hydrocarbons in a 100
ml sample of air with good precision and accuracy (if not completeness)
to be at all useful these data must be used to infer the composition
of some larger air mass. The hazards involved in this inferential
process are sometimes underestimated. They are especially important
in analyzing light hydrocarbons in ambient air at ground level which
is the most convenient sampling location. An ideal system would be
one in which the entire air parcel of interest (this might be several
cubic miles) could be homogenized before measurement. The next best
ideas is to choose a sampling area removed from sources and to average
a number of samples. The separate analyses can then be used to
-------
11
estimate the homogeneity of the air parcel. On the other hand,
fluctuations from sample to sample can also be useful. Hydrocarbons
which come from a common source must fluctuate together in
concentration, while those which derive from different sources
will not normally correlate well with each other. This technique
was used to identify natural gas in urban air (see later).
Standard sample tubes of 300 ml to 1 liter in volume were used
to sample in remote locations. It was convenient to transport or
ship a dozen such tubes in a box. These tubes were fitted with a
stopcock at each end and were filled by flushing with a rubber
suction squeeze bulb. The sample, therefore, did not pass through
the squeeze bulb. Laboratory evacuation of the sample tubes was
avoided because there was no way to verify vacuum maintenance during
transport to a distant sampling site. To transfer a sample into the
freeze trap, a 100 ml syringe was filled with pure nitrogen, attached
to the sample tube, the stopcock opened, and the sample and nitrogen
mixed by working the syringe plunger back and forth a number of times.
Tests were made to ensure that no significant leakage occurred during
this process. The volume of the sample tube was used to calculate
the degree of dilution that occurs in this process. After closing
the stopcock the sample syringe was then connected to the gas sample
valve on the chromatograph and the contents forced through the
chilled trap. This procedure could be repeated several times so
actually several 100 ml samples could be taken from sample tubes
as small as 300 ml. Of course, the dilution factor becomes larger
and larger as more samples are taken and this limits the process.
Quite reproducible results can be obtained even at the low concen-
trations of ambient polluted air.
Samples taken at temperatures and/or pressures different from
those prevailing in the laboratory must be first brought to laboratory
pressure (about 1200 feet above sea level) by adding measured amounts
of pure nitrogen from a 100 ml syringe or by venting the tube. The
empty syringe is then attached to one inlet of the gas sample valve
on the chromatograph while the sample tube is attached to the other.
-------
12
With the valve in the sample position, the contents of sample bulb
and 3 ml sample loop are mixed. This sample is injected into the
chromatograph to provide a methane measurement. If nitrogen was
added to bring the tube to laboratory pressure, a correction is
made for this dilution.
Larger samples of ambient air were taken in 20 liter (5 gal.)
and 50 liter (12 gal.) borosilicate glass carboys. To provide an
inert closure for these vessels, a special lid was devised. The
mouth of the carboy was ground flat and a 1/4 inch thick glass disc
containing epoxy cemented borosilicate glass inlet and outlet tubes
was waxed on with fluorocarbon wax. One of the tubes extended to the
bottom of the bottle, and the other into the top. In this way a
suction pump could be used to purge the bottle and flush it with
sample in a few minutes. A larger port, closed by a ball joint, was
cemented in the lid and used for syringe or micropipet injection of
standard samples. Samples in this bottle were in contact with glass
almost exclusively with only minimum amounts of epoxy cement and
fluorocarbon wax.
Irradiation
This bottle and an irradiation system which was used to irradiate
ambient air samples is shown in Figure 5.
Twelve "blacklite" fluorescent lamps (15 watts each) were mounted
in a framework as a vertical cylinder large enough to contain one of
the bottles. A ventilation fan mounted below the bottle served to limit
the temperature rise to 1 or 2 C. The light intensity in the center of
the irradiation system was tested by photolyzing nitrogen dioxide at
0.5-5 mm Hg pressure in an ultraviolet absorption cell. Concentrations
were then measured in a Gary Model 14 ultraviolet spectrophotometer.
The half-life of nitrogen dioxide measured in this way was 1.3 min.
These results are shown in Figure 6.
This half life is somewhat longer than that calculated and observed
for outside sunlight on the brightest day (about 1 min.). This system
was used to irradiate ambient air samples which were taken in early
-------
13
BLACKLIGHT
J / LAMPS
us
m
20 LITER CARBOY
BLACKLIGHT
LAMPS
\
BOTTLE PHOTOREACTOR
FIGURE 5
morning hours after a good purge of the atmosphere. Two such carboys
were available so simultaneous samples could be taken and irradiated
in natural sunlight and artificial light.
Stirred Flow
This system which originally employed 20 liter carboys and static
irradiation techniques, was later elaborated to use a 50 1 carboy and
-------
14
o
2
<
00
cr
w
00
0.10
0.09
0.08
0.07
0.06
0.05
0.04
0.03
PHOTOLYSIS OF N02
I mm Hg, 10 cm cellx
GARY MODEL-14 /
BOTTLE REACTOR
12-15 WATT
BLACK LIGHTS
(W/0 ALUMINUM
FOIL)
2/25/66
1.28 min
0 I
TIME (minutes)
FIGURE 6
a stirred flow system. One borosilicate glass reaction vessel is
mounted inside a set of 24-black light fluorescent lamps as irradia-
tion source. These have been mounted on a cart along with supply
tanks and flow and mixing equipment to supply the reactor. Included
on the cart, in addition to the bottle photoreactor itself, are a
heatless air dryer for air purification, supply tanks for the react-
ants, a flow mixing panel consisting of the flowmeters and suitable
needle valves and space for analytical equipment (see Figure 7 ).
The supply tanks are low pressure oxygen tanks, each of which is
equipped with a pressure reduction regulator which feeds into a
-------
14a
MIXING
CHAMBER
THREE WAY
STOP COCK
ROTOMETERS
FINE METERING
VALVES
FLOW
REGULATORS-
COMPRESSED
AIR SUPPLY
AIR LINE
DRYER
FLUORESCENT
BLACK LITES
OUTLET
SHUTOFF
VALVES
46.9 LITER
BOROSILICATE
BOTTLE
SHUTOFF
-VALVES
AIR SUPPLY
HYDROCARBON
REMOVAL
STAINLESS STEEL .
LOW PRESSURE TANKS
-------
15
PRODUCT
(DRY AIR OUT)
;CHECK VALVES
DRYING
CHAMBERS
B
RECYCLE
TIME*
—
\
i
^
i
r
£
PURGE
VALVE
WET
AIR
OUT
<-:
X*
^~
WET
%
^
i
[
\>
s
'
k
•'.".'• *••**.'.'
A
SOLENOID VALVE
UR IN
FIGURE 8
Heatless Adsorbtion Air Purification
manifold in a flow mixing panel. Needle valves are used for flow
regulation. Four such tanks are mounted on the cart along with
their valves and the flow panel. The heatless air dryer used for
air purification is 3/4 filled with activated charcoal and 1/4 with
the manufacturer's desiccant. This modification, as recommended by W. H.
King of Esso Research, permits production of very pure hydrocarbon-free
air. One column of the dryer adsorbs air pollutants in the char-
coal while the other is being purged. It operates on a 30-second
cycle and produces air with hydrocarbon concentration reduced to
-------
16
the low ppb range. This is important since the ambient air in
Riverside already contains concentrations in excess of those that we
wish to study. The flow system has been designed to have a residence
time variable from one-half to four hours. To accomplish this, a
portion of the supply stream of dilute reactant can be diverted away
from the bottle and used for an input analysis. This permits analysis
of input and output streams simultaneously. Table 5 gives performance
data on the air purifier.
TABLE 5
HYDROCARBON REMOVAL BY TWO COLUMN
MEATLESS FRACTIQNATOR
HYDROCARBON
METHANE
ETHENE
ETHANE
ACETYLENE
PROPANE
PROPENE
ISOBUTANE
n-BUTANE
CONCENTR
Unfiltered
) N Ambient Air
2,028
9.4
23.2
11.. 6, 11.2
13.4
1.4
4.8
13.4
ATION ppb
Filtered
Air
1,058
0.6
1.4
1.2, 1..4
1..6
0.4
0.1.
0.2
-------
17
This reactor was designed primarily to study the reactions of
aromatics at the realistic concentration range. In such circum-
stances the question of loss of reactants by adsorption on the wall
of the reactor is usually raised. This is especially important when
compounds of low volatility are being studied. In a tube flow
system used on an earlier project it was found that adsorptions of
light hydrocarbons or even of PAN on clean reactor walls was not a
significant factor. This type of experiment was repeated with
toluene and acetylene in the new flow reactor. With 254 ppb of
toluene in the bottle a steady flow of 0.86 liter/min of pure air
was used to purge the bottle. Chromatographs#2 and #3 were used to
monitor the acetylene and toluene concentrations with time. Multi-
plication of the reactor volume by the concentration of toluene
gives the amount of toluene present initially in the gas phase
(11,900 nl as a vapor, see Figure 9 ). Integration of the purg-
ing curve in this figure gives the amount of toluene removed from
the reactor (11,280 nl in Figure 9 ). Since these two figures agree
we conclude that little toluene was desorbed from the reactor surfaces
during purging. It should be noted that this procedure does not
require an absolute calibration of the chromatograph.
Data of this type can also be used to evaluate mixing in the
reactor. Perfect mixing corresponds to exponential dilution and
should follow the equation
In C/CQ = -ft
Figure 10 shows a logarithmic plot of acetylene and toluene for this
purge experiment. Both follow closely the theoretical curve for the
volume and flow rate indicating adequate mixing.
-------
18
250
254 ppb x 46.9 I = 11,900 nl toluene
UJ
30
PURGING BOTTLE
REACTOR
-AREA-= 11,280 nl toluene
60 90
PURGING TIME.
120 150
MINUTES
180
:'TGU".E 9
-------
19
o
o
10
8
PURGING BOTTLE
REACTOR
0 ACETYLENE
THEORY
46.9 liters
0,86 liter/min.
0
30
60 90
MINUTES
120
150
FIGURE 10
-------
20
RESULTS IN AMBIENT AIR
Smog Hydrocarbons
The pattern of hydrocarbon concentrations as found in Riverside
photochemical smog is quite consistent; changes in the relative amounts
of hydrocarbons are much smaller than the changes in the concentrations
from clean air to smog air. The analysis taken when a 0.6 ppm oxidant
smog was recorded on 6 August 1970 (a record for that year in Riverside)
is typical (see Figure 11 ). In this figure and others like it in
this report, the hydrocarbons are plotted according to the listing in
Table 6 .
ppb
60-
50-
40-
30-
20-
B.G.
10-
0-
t
_
xlOO
. ;
;
; ;
HYDROCARBONS IN POLLUTED AIR
6 AUG. 1970, I500PST - RECORD SMOG
PPb
IOT
B
n :
Q • s H S H H « H
678889!
FUEL A
•
•
|
a :
1 I I
! i 5
) 9 9 9 9 9 9 9
ROMATICS
c
: n H I 1 R 1
23 22334 4444555 4455666656
NATURAL CRACKED FUEL OLEFINS FUEL SATURATES
GAS PRODUCTS
FIGURE 11
-------
21
TABLE 6
HYDROCARBONS
(By carbon number in the order plotted in bar charts)
Natural Gas (mostly from leaks in natural gas supply pipes)
1. Methane
2. Ethane
3. Propane
Cracked Products (mostly from auto exhaust, not present in gasoline)
2. Acetylene
2. Ethene
3. Propene
3. Me-acetylene
4. 1,3-butadiene
Fuel Olefins (part of the C4's are also derived from cracking)
4. 1-butene
4. isobutene
4. trans-2-butene
4. cis-2-butene
5. 2 me-butene-1
5. trans-2-pentene
5. 2 me-butene-2
Fuel Saturates
4. isobutane
4. n-butane
5. isopentane
5. n pentane
6. 2,2 di-me-butane
6. 2,3 di-me-butane
6. 2 me-pentane
6. 3 me-pentane
5. cyclopentane
6. n-hexane
N naphthenes
-------
22
TABLE 6 (continued)
Fuel Aromatics
6. Benzene
7. Toluene
8. Ethyl benzene
8. p-xylene
8. m-xylene
8. o-xylene
9. Isopropyl benzene
9. n-propyl benzene
9. p-ethyl toluene
9. m-ethyl toluene
9. o-ethyl toluene
9. 1,3,5-trimethyl benzene
9. 1,2,4 trimethyl benzene
9. 1,2,3 trimethyl benzene
10-12. Higher aromatics
The first three paraffins are listed in the figure as natural gas
since leakage of this hydrocarbon mixture seems to be the principal
source for these compounds. The "cracked products" are five compounds
not found in gasoline but present in auto exhaust in substantial
quantities. These form a unique tracer for exhaust gas, since there
are few other significant sources of these five compounds, and also a
useful indicator of degree of reaction since they present a spectrum of
reactivities. The other three groups are listed as fuel components
because they are present in gasoline. Small amounts of some of these
are produced in engines: in particular the four carbon olefins.
The distribution of hydrocarbons shown here resembles auto exhaust
with natural gas and gasoline vapor added. This breakdown will be
discussed in more detail in subsequent paragraphs.
-------
23
Background Hydrocarbons
With the exception of methane, all the hydrocarbons shown in
Figure 11 are at far higher concentrations than are found in
unpolluted air. To measure background levels, two groups of samples
were taken during March 1968 in the San Bernardino and San Gabriel
Mountains of southern California during moderate to fresh Santa
conditions. These winds blow for several days at a time across
southern California from the large desert areas to the north and east
and bring the cleanest air to Riverside, California. These samples
were taken 1000 to 1300 meters above sea level, one set on the northern
slopes of the San Bernardino Mountains and one set in the San Gabriel
Mountains. The results are summarized in Figure 12. The levels of
methane were satisfactorily consistent and gave an average of 1.39 ppm
of methane as shown in Table 7.
ppb ppb
60-
50-
40-
30-
20-
10-
n -
20-
10-
0_
1 III
1 III
67888899 999999
FUEL AROMATICS
"PURE MOUNTAIN AIR"
xlOO
1 1 M.l i_i III! Ill | 1 | [ L. 1 |
123 22334
NATURAL
GAS
CRACKED
PRODUCTS
4444 555
FUEL OLEFINS
4 455666656
FUEL SATURATES
FIGURE 12
-------
24
To determine whether some of the observed methane concentration
might be due to contamination from upwind automotive or natural gas
sources, three of the March 19 samples were analyzed for C~ hydro-
carbons. Auto exhaust contains methane, ethene and acetylene in
roughly equal amounts so that the maximum possible methane contribution
by exhaust gas can be estimated from the amounts of these two unsatur-
ated hydrocarbons present in a sample. Natural gas contains about
fifteen times as much methane as ethane, permitting the estimation of
the maximum amount of methane contribution from this source.
Three ppb of ethane were present in the three samples, suggesting
a possible methane contamination of 0.045 ppm, if the contamination
came from natural gas sources. One ppb each of ethene and acetylene
were present, suggesting a possible methane contamination of 0.001
ppm, if contamination came additionally from automotive sources. In
either or both events, the addition to the methane concentrations
given in Table 7 could not exceed 0.046 ppm, or a reduction of 3.37».
Whether the presence of ethane, ethene and acetylene in these samples
represents low-level contamination, or whether they too exist in a
background concentration is open to further examination.
Our value of 1.39 ppm methane can be converted into the spectros-
copist's units by assuming that this concentration prevails from sea
level to outer space. Taking the pressure as one atmosphere and the
temperature at 273 K, we calculate a value of 1.11 cm for the equivalent
layer of methane. This agrees exactly with the figure reported by Fink
2
e_t aJL from their spectroscopic studies, but is slightly smaller than
3
the 1.2 cm reported by Goldberg and Muller.
Table 7 Methane in clear mountain air.
Date Column Samples C ppm
-------
25
Unrpacted Air
The sample shown in Figure 11 had obviously undergone consider-
able photochemical reaction so it was of interest to obtain distributions
under unreacted conditions. This can be done by taking samples under
the proper weather conditions. On some occasions, particularly during
the winter months, winds from the desert will purge the Riverside area
of all pollution in the late afternoon or early evening. Such winds
often die down during the night so that, with a clear sky overhead,
a radiation inversion will form. This will effectively trap pollution
overnight and during the early morning traffic rush. Samples taken
during this time have not been exposed to sunlight and can yield infor-
mation on the composition of genuine unreacted air pollution. In the
past 10 years, samples have been taken on perhaps a score or more of
such mornings. The most striking observation is that typically there
are no manifestations of pollution observable (haze, eye irritation,
odor, etc.). Yet chemical analysis shows high levels of many pollut-
ants. Hydrocarbon distribution on such a morning is shown in Figure 13.
ppb
80--
70--
60--
50--
40--
30--
xlOO
20--
10--
HYDROCARBONS IN UNREACTED
MORNING AIR
(II SAMPLE AVERAGE)
29 JAN '70 0735-0800 PST
Ill ••• . ••• I
.ll.l
123 22334
NATURAL
GAS
CRACKED
PRODUCTS
4444 555
FUEL OLEFINS
4455666656
FUEL SATURATES
FIGURE 13
-------
26
Several features of this chart are noteworthy:
1) Concentrations for the most part are as high or higher than
those found even in heavy smog. Acetylene, for example, was 78 ppb
compared to 27 ppb on August 6, 1970. Since auto exhaust is by far
the largest source of acetylene, this indicates that auto exhaust was
present in high concentration even though there were no visible signs
of pollution.
2) Some very reactive hydrocarbons were present in readily
measurable amounts, such as 2.5 ppb of 2-methylbutene-2. The amounts
of propene and ethene as compared to acetylene are much higher in this
unreacted sample. This will be discussed in much detail later.
3) The amounts and distribution of the saturates are not much
different between morning and afternoon samples.
It is tempting to suggest following the course of smog develop-
ment on such a day. This cannot be done since the radiation inversion
is destroyed early in the morning before very much reaction has
occurred. Very often little or no oxidant is formed on such days.
The data shown in Figure 13 are averages of 11 samples taken
in pairs between 0735 and 0800. Study of individual samples is also
informative because it shows that some pairs of individual hydrocarbons
fluctuate with each other while others do not. This clearly reflects
the different sources of hydrocarbons. Figure 14, for example, shows
the fluctuation with time of methane and acetylene. From the duplicate
analyses it can be seen that these fluctuations are much larger than
the analytical error. Even at 0735 PST the acetylene was higher than
in smog samples of 24 October 1968 and 6 September 1968, which in turn
are much higher than in clean air.
The independent time fluctuations suggested plotting acetylene
and ethene against methane (Figure 15 ). Although there is some
tendency for higher values to go together the correlation is poor.
Much better correlations were found when ethene and propene were plotted
against acetylene (Figure 16 ) and when ethane and propane were plotted against
methane (Figure 17). The good correlations shown by these two figures
reflect the fact that most of the ethene, propene and acetylene come
from one source (auto exhaust) while the methane, ethane and propane
-------
27
e
120+2.4 8:
100
•a
a
80
5
w
z
u
w
u
1.6 «
METHANE AND ACETYLENE IN MORNING AIR
(H-F Bldg. 29 Jan 70, 0730-0800)
60-1.2
40
. >—24 Oct 68, 1610 PST
CHHCH in SMOG
20 -f 6 Sep68, 1500PST
CH4 in CLEAN AIR
0
0730
CHECH in CLEAN AIR
0-740 0750
PACIFIC STANDARD TIME
0800
FIGURE 14
-------
28
HYDROCARBONS IN MORNING AIR
(H-F Bldg 29 Jan 70, 0730-0800)
120 •
ex
°ioo .
•
w 80 •
w
0 60 •
Z
PJ
>
h 40 '
W
u
20 '
120
0°
0
o •
t
• • ^
ft
e
• t
o
0 *
• 1 *
0 1500 2000 250
METHANE CH4> ppb
FIGURE 15
-------
29
•a
ex
CN
u
CO
I
U
w"
z
p^
o
CN
g
100
I
u
80
60
40
20
0
0
AUTO EXHAUST COMPONENTS IN MORNING AIR
(H-F Bldg. 29 Jan 70, 0730-0800 PST)
CH2-CH2
50 100
ACETYLENE, CH2CH, ppb
FIGURE 16
-------
30
50
•a
a,
. 40
oo
X
CO
U
S 30 '
o
20 -
cs
U
10
0
1200
NATURAL GAS IN MORNING AIR
(H-F Bldg. 29 Jan 70, 0730-0800 PST) /} C2H6
1500 2000
METHANE, CH4, ppb
51 4- 7
C3H8 23 2.1
CH4 1000 93.2
2500
1074 100.0
FIGURE 17
-------
31
come from another. The fact that concentrations vary by a factor of
2-1/2 (saturates) or 5 (unsaturates) indicates that the sources are
all relatively local. If these had been carried from a distant source
over a period of hours a much more uniform mixture would be expected.
The paraffins (Figure 17 ) plots both can be extrapolated to
zero concentration at about 1350 ppb on the methane axis. This agrees
with the value for pure air already cited. The proportions of ethane
and propane as cpmpared to the excess methane above this background
value can be derived from the slopes of the two plots. The composi-
tion shown in the figure approximates that of natural gas (Figure 18 ).
ppb
60--
50--
40--
30--
20--
10--
0
x
100
I
NATURAL GAS
I06 dilution
I 2 3
NATURAL
GAS
I I I I I
22334
CRACKED
PRODUCTS
4444 555
FUEL OLEFINS
4455666656
FUEL SATURATES
FIGURE 18
-------
32
TABLE 8. Natural Gas in Ambient Air
Methane
(excess above
background)
Ethane
Propane
Natural
Gas
90
8.4
1.7
29 Jan 70
93.2%
4.7%
2.1%
Jan
91.6
7.1
1.4
63
91.3
6.8
1.9
Jan 68
91
6.5
2.6
Compositions derived in this way on a number of different occasions
are compared in Table 8. The agreement is close enough to be convincing,
although natural gas composition varies from week to week and a more
detailed study would take this into account.
The plot of ethene and propene versus acetylene has been repeated
on a number of occasions. The slopes of these lines can be used to
estimate the relative amounts of ethene and propene as compared to acetylene
in unreacted air. These ratios are remarkably consistent as shown by the
computation made over the years in Table 9. These ratios of 0.85 =
ethene/acetylene and 0.25 = propene/acetylene are close to the ratios of
0.91 and 0.34 reported by Neligan e_t al^. in 1961 (Figure 19) for auto
exhaus t.
TABLE 9. Molar Ratio of Ethene (C2H.) to Acetylene (C^) and Propene
(C-H,) to Acetylene in Various Samples of Unreacted Air
Ethene
Acetylene
Propene
Acetylene
Ethene
Acetylene
Propene
Acetylene
Jan 63
0.87 0.95
0.22 0.21
24 Oct 66
0.91 0.96
0.25 0.25
22 Dec 65(a)
0.43
0.11 0.16
Dec 67
0.72 0.72
0.21 0.21
3 Mar 66
0.73
0.21
Dec 67(b)
0.66
0.18
10 Mar 66
0.84 0.80
0.23 0.22
-------
33
TABLE 9 (continued)
Ethene
Acetylene
Propene
Acetylene
23 Jan 68
0.81
0.24
29 Jan 70
0.83
0.21
10 Feb 71
0.73
0.21
(a) Very high acetylene, two propene instruments
(b) Monterey-Salinas area
ppb
60-t-
50--
40--
30--
20--
B.G.-
10-h
ppb
1
AUTO EXHAUST
5000 dilution
Fuel #2 (8.2% Olefin)
Neligan Japca April '61
* Not measured
I I I I
67888899999999
FUEL AROMATICS
Not separated
I
/__(
MI
I 2 3
NATURAL
GAS
22334
CRACKED
PRODUCTS
4444555
FUEL OLEFINS
4455666656
FUEL SATURATES
FIGURE 19
-------
34
In view of the uncertainties in the driving cycle, the values derived
from ambient air data are thought to be a truer estimate of the ratios
of these components in urban auto exhaust. Figure 19 was plotted on
the same scale as the ambient air charts by applying a 5000 fold
dilution factor. The 8.2% olefin fuel used to produce this exhaust,
when diluted by 10 million to one yieldsthe bar chart of Figure 20.
ppb
60-
50-
40-
30-
20-
10-
0-
ppb
20--
1
o •'
6788
GASOLINE
I07 dilution
Neligan JAPCA April '61
*Not measured
*
1 |
8899999999
FUEL AROMATICS
II 1 II II i • •
II 1 II II
23 22334 4444555 '
NATURAL CRACKED FUEL OLEFINS
GAS PRODUCTS
Not separated
ll.n...
4455666656
FUEL SATURATES
FIGURE 20
This illustrates the reason for classifying these hydrocarbons as fuel
components and cracked products. For completeness a partial blowby
analysis is shown in Figure 21 . It would obviously be very diffi-
cult to distinguish blowby hydrocarbons from exhaust hydrocarbons.
-------
35
ppb
60--
50--
40--
30--
ppb
20--
10--
BLOW BY GAS
40,000 fold dilution
PATTISON ond STEPHENS
(24 cars)
* Not measured
67888899999999
FUEL AROMATICS
0-
o-
i 1 1 • . ..ii. I
*
A
r ^
1 1 1
1 1 1 1 1 1 III
123 22334 4444555 4455666656
NATURAL CRACKED FUEL OLEFINS FUEL SATURATES
GAS PRODUCTS
FIGURE 21
Other Sources
Several other hydrocarbon sources were studied as possible contribu-
tions to atmospheric hydrocarbon. Figure 22 shows a sample taken in a
controlled experimental burn of mountain brush. Three things are note-
worthy: (1) The smoke contains very little saturated hydrocarbons of
four or more carbons; this is a striking contrast to auto emissions.
(2) Concentrations were not very high compared to car exhaust, even
though this sample was taken in dense smoke. (3) Ethene was much larger
than acetylene again in sharp contrast to auto exhaust.
Figure 23 shows an analysis (at 10 fold dilution) of air from an
oil field. This sample was taken just adjacent to oil wells in an
urban area. The concentrations of cracked products are attributed to
auto exhaust from nearby traffic and the fuel saturates and natural
-------
36
ppb
60-
50-
40-
30-
20-
10-
0-
1
1
PP° - BRUSH FIRE SMOKE
20 " 10 FOLD DILUTION
(CONTROLLED BURN)
NORTH MOUNTAIN
10-- |2V 1969
0.1 i i i i i i i i i i i l
1 1 1 1 1 1 1 1 1 I 1 1 1
67888899999999
FUEL AROMATICS
• ill _•
23 22334 4444555 4455666656
NATURAL CRACKED FUEL OLEFINS FUEL SATURATES
GAS PRODUCTS
ppb
60-
50-
40-
30-
20-
10-
xlC
FIGURE 22
ppb
204- OIL FIEbO AIR
10 FOLD DILUTION
)0__ (SIGNAL HILL)
01 ._) i i i i l l l i l i i
1 1 1 1 1 1 1 1 1 1 1 1 1
67888899999999
FUEL AROMATICS
X)
II. l i ill III I I • •
23 22334 4444555 4455666656
NATURAL CRACKED FUEL OLEFINS FUEL SATURATES
GAS PRODUCTS
FIGURE 23
-------
37
gas components present in unusually high concentration are attributed
to the oil wells. These emissions may help explain the high levels of
paraffins in urban air. Some samples taken in an industrial area are
shown in Figure 24 . The most striking difference is the presence
of 30 ppb of 1,3-butadiene and a reversed order of abundance of propene,
ethene and acetylene (as compared to auto exhaust).
METHANE
ETHANE
PROPANE
ACETYLENE
ETHENE
PROPENE
METHYL ACETYLENE
1,3-BUTADIENE
1-BUTENE
ISOBUTENE
TRANS-2-BUTENE
CIS-2-BUTENE
2-METHYL BUTENE-1
CYCLO PENTENEfa;
TRANS-2-PENTENE
2-METHYL BUTENE-2
ISOBUTANE
N-BUTANE
ISOPENTANE .
N-PENTANE ^
CYCLO PENTANE
2,2-DIMETHYL BUTANE
2,3-DIMETHYL BUTANE
2-METHYL PENTANE (d)
3-METHYL PENTANE
N-HEXANE
Industrial Area Parking Lot
2 samples 4 samples
Busy Comer
4 samples
2368 |50 ppb
50 ppb
50 ppb
50 ppb
HYDROCARBON DISTRIBUTION
NEAR SOURCES.
(a) 2-METHYL BUTADIENE-1,3
(W3-METHYL BUTENE-1
(c) 1-PENTENE
(d) CIS-2-PENTENE
included
FIGURE 24
-------
38
In addition to the propane the four and five carbon paraffins
are found in Riverside smog in larger concentrations than are easily
accounted for by auto exhaust. For example, in Figure 25 the
ppb
60--
50--
40--
30--
20--
Bfi.-
10--
HYDROCARBONS IN POLLUTED AIR
xlOO
ppb
20+
10--
24 MARCH 1971
4:00-4:10 PM
67868899999999
FUEL AROMATICS
I 2 3
NATURAL
GAS
22334
CRACKED
PRODUCTS
4444555
FUEL OLEFINS
4455666656
FUEL SATURATES
FIGURE 25
n-butane concentration is approximately equal to the acetylene, whereas,
in auto exhaust, Figure 19 , it was three or four times smaller.
These excesses are ascribed to gasoline vaporized from fuel tanks which
is known from emission inventories to be a substantial source of hydro-
carbons. It might be expected that gasoline vaporization would be
greater on very hot days and there is in fact a striking correlation
between maximum oxidant and maximum temperature. Data for one month
each in the summers of 1967 and 1972 are shown in Figure 26.
It is difficult to see how this trend could be blamed directly on the
-------
39
o
3
0.60
0.50
«•
0.30
0.20i
0.10
SMOG AND HEAT
IN
RIVERSIDE
AUGUST 1967 O
JULY 1972 +
85
O
+
4 °
000
O O •+• O O O
* o
90
95
100
105
110
MAXIMUM TEMP °F
FIGURE 26
effect of temperature on the kinetics of the reaction since most reactions
have small activation energies. Several explanations are possible, involving
interaction with meteorological variables. For example, high temperatures
decrease the relative humidity which will desiccate the aerosol and thereby
reduce the scattering of short wave ultraviolet which is crucial for oxidant
formation. Also, high temperatures occur when the mixing layer is shallow and
is capped by a very warm layer of air aloft. Such conditions could favor
prolonged irradiation which also promotes oxidant formation.
To see whether extreme temperatures would be accompanied by excess
emissions of light paraffins samples were taken when the temperature was
106°F (see Figures 27 and 28). Six samples were taken simultaneously at
each of two locations: 1) the roof of the sixth floor, Health-Finance
Building, in central Riverside, and 2) a small hill about four miles east
of this and just behind the UCR campus. The UCR air monitoring station
-------
39a
recorded 0.30-0.33 ppm oxidant at this time, so there was moderate smog.
Individual samples of each group were nearly the same in distribution, but
the central Riverside samples were uniformly higher than the hill samples
in all components as though dilution were the only difference between the
two sites. Both showed the depletion of the olefins, indicating a sub-
stantial degree of reaction.
-------
40
HYDROCARBONS IN POLLUTED AIR
21 AUG. 1969, 1440 PST - I06°F (HILL)
ppb
30+
20-1-
B.6.-
_
10-
xJO
•
-d—
N/
o
23 22334
tfURAL CRACKED
GAS PRODUCTS
4 444555
FUEL OLEFINS
1
v\ _, . R •.
• n
4455666656
FUEL SATURATES
FIGURE 27
HYDROCARBONS IN POLLUTED AIR
21 AUG. 1969, 1440 PST - I06°F (ROOF)
ppb
30-
20-
B.G. —
10-
f\
xOO
_
°
•
M n _
23 22334
NATURAL CRACKED
GAS PRODUCTS
4 444555
FUEL OLEFINS
n
i
t
1
i
1
•
„ Jfl m0
445 56666 56
FUEL SATURATES
FIGURE 28
-------
41
If compared with the auto exhaust analysis of Figure 19 , a dilution
of 25,000 is indicated for the hill samples. But the result of this
experiment was negative; no dramatic increase in C.-C paraffins as
compared to the cracked products was seen (even at 106°F).
Gasoline Vaporization
Nevertheless, these light paraffins are consistently higher in
afternoon samples than in morning samples as illustrated in Figure 29.
Since these paraffins along with benzene and toluene react to only a
COMPARISON OF HYDROCARBON DISTRIBUTIONS
-100 0 +300-100 0 + 300-100 0 +300
_^^^_*
mwwwiwm
[_
Acetylene
Ethylene
Propylene
1 1 butene
1
L
-3
Isobutene
Ethane
Propane
Isobutane
n. butane
1 1 Isopentane
[In. pentane
[
M*
f
T_
t
i
t
m
\m
^^^nmnffifff
\
\
T U
f r
1
1 1
ib
n
1
1
1
Average of four 11-4-65 4:40 PST 9-24-65 2P.M. 10-7-65 2:20 P.M.
morning samples
Afternoon samples - % deviation from morning average
FIGURE 29
small degree (see later sections), it was possible to make an estimate
of the contribution of gasoline vaporization to total hydrocarbon by
fitting atmospheric data to Raoult's law for gasoline. Raoult's law
-------
42
was shown to be reasonably accurate when applied to gasoline to calcu-
4
late the true vapor pressure. To apply this to ambient air data, the
concentration of the ith hydrocarbon C. is approximated by an estimated
sum of contributions of liquid and vapor
C. = C.. + C
i liq vap
This can be most easily visualized by imagining that a sample of liquid
gasoline (L nanomoles) and a sample of saturated vapor-air mixture from
above this gasoline (V nanomoles) has been added to each mole of air.
Then the following two equations apply:
CL . = x.L
liq i
C = p.V = x.P. V
vap i i 10
where x. = mole fraction of hydrocarbon i in the gasoline
p. = vapor pressure (atm) of hydrocarbon i over the gasoline
(assume saturation)
P. = vapor pressure (atm) of pure hydrocarbon i
/*.
This hypothetical mixture then has a concentration C. of hydrocarbon i
C. = x.L + x.P. V
11 i 10
To fit this to an ambient air analysis these values are subtracted from
the observed concentration C.
C. - £. = C. - (x.L + x.P. V)
i i i i 110
To obtain a fit, L and V are adjusted to minimize £(C. - C.) which is,
of course, a least squares procedure. With more than 3 or 4 hydrocarbons
this becomes a very tedius calculation best suited to the computer. A
program for this was written in APL forthe UCR IBM 360/50 as shown in
the following pages. The first trials showed that the temperature
-------
V
[1] 'ENTER DATA SET NBR'
[2] NSET+fl
[3] 'ENTER CODE NBRS FOR HYDROCARBONS'
[4] flZTKl
[5] •EMPffi? CO/PCS (PPB), Z/SF, ZF-flO IF C0//C IS NOT GIVEN'
[6] £A*-0*CTtf«-Q
[7] 'EW2E7? M9LF, FRACTIONS'
[8] JCmW]
[9] •HWE'ff K4POT PRESSURES (ATM)'
[10] PI^-D
[11] 'EWTEff DATE FOR CONCS - APPROX 8 LETTERS'
[12] ZMTEM!]
[13] 'ENTER SOURCE FOR MOLE FRACTIONS'
[14] SOURCE^
[15] 'fftfTH? TffMP FOR VAPOR PRESSURES (DEC F)1
[16] 2M3
[17] '15 ANALYSIS OF VARIANCE FOR REGRESSION DESIRED?'
[18] ANS*ft
[19] Y+'YES'
[20] ;OWZ?«-OpO
[21] XX++/XXX+U/XIN
[22] CX++/X*C+U/CIN
[23]
[25] NND<-(~U)/NIN
[26] XND+-(~U)/XIN
[27] PND*-(~U)/PIN
[28] OT-«-(^0)
[29] XXPP-M-/* x/xpxp
[30] DETER+-*(XXxXXPP)-XXP*XXP<-+/XxXxP
[31] ZP*-Jrxp
[32] XL<-XxL<-DETERx(CXxXXPP)-CPXxXXP
[33] DEVX+UX/DEV+C-LV+XL+XPV+XPx V+DETERx (XXxCPX ) -XXPxCX
[34] PRINT: 'DATA SET= ' ',NSET;lQp' ';'NBR HYDROCARBONS IN CALC-' ;pW;' ffflff HYDROCARBONS IN SET=' ;pNIN
[35] 25p' ';'(7(PPB) X(W) P(^W) AT P(xl2'W) AT L ^P7^L+^Pl^
[36] 25p' ';• I I 10 I 10 I I 10 I I 10'
[37] 20p' ^JMIff;1 '-.SOURCE;' *\T\'*F ' ;2"; ' "F1 ;36p ' (PPB)'
[38] ' •
[39] I<-0
[40]
-------
[41] J+-I-H
[42] ffCJftUfffClttl];]; 9 3 DFT PRNT+Cin ,7[I],P[I],XP[r] ,*£[!],XPf[I],
[43] -ȣWZWC
[44] NODATA:-»•(Q=0NND)/SUM
[45] I«-0
[46] XLNEK-XNDxL
[47] LVND<-XLND+XPVN[h-VxXPND*-XNDxPflD
[48]
[49]
[50] ffCff>l/^[M£)[J];];9p' '; 9 3
[51] -^Sro//Z?
[52] SUM:SC++/C
[53] SP++/PIN
[54] SXL+LXSX+-+/XIN
[55] SLV+SXL+SXPV*-V*SXP<-(+/XP)++/XPND<-XPND, 0
[56] >!W?ff7*-(*pI7ff^)x5Z?ffVr*-l-/(|Z?ffra)
[57] • •
[58] 16p' ';'Si/W '; 9 3 DF7 PRNT*-SC,SXtSP,SXP,SXLtSXPVtSLV.SDEV
[59] 79p' ';MVF'; 10 3 DFT AVDEV
[60] • '
[61] 30p» •;IL=';L;1 7=';7;'
[62] ' •
[63] -+(Y*ANS)/0
[64] • ANALYSIS OF VARIANCE*
[65] 4M4L:MSREG*-SSPEG+-+/LV*LV
[66] ND*-(pN)-2
[67] F+-MSREG*MSDEV*-( *ND )
[68] 15p' ^'OF^Sp1 ';'
[69] 'REGRESSION 1'; 13 3 ZVT PRNT+SSREG,MSREGtF
[70] 'DEVIATION ';M>; 13 3 DFT PRNT+SSDEV\MSDEV
[71] -K)
-------
45
assumed to obtain P. values was not crucial to the estimate of vapor
contribution. This is because all components increase in vapor pres-
sure by a similar factor between, for example 70 and 100 F. For this
reason 100 F was used in later trials. The most important and difficult
to obtain piece of data is the gasoline composition. The first attempts
were made to use a morning hydrocarbon analysis as a fuel composition,
and assuming no vaporized gasoline, but these were not too successful.
Later an analysis of Los Angeles basin gasoline was made available by
the Western Oil and Gas Association. Use of this fuel composition was
discontinued when it was realized that its calculated vapor pressure
would be only about 7.5 or 8.0 psia. The four and five carbon paraffins
are obviously the crucial factor in both the calculation and in deter-
mining the fuel vapor pressure. The composition was, therefore, adjusted
to give a true vapor pressure by calculation of 0.700 atmospheres at
100°F or 10.3 psia. This is probably just a little high, but it was
used to estimate liquid and vapor contributions for a number of samples.
A sample computer print-out is shown in Figure 30 . The first column
gives the analysis obtained on 6 Aug 1970. The second column gives the
fuel composition (mole fraction, labelled 7/10/72 because it was first
adopted on that date). This fuel and its vapor are compared in
Figures 31 & 32. The third column gives the vapor pressures of the pure
hydrocarbons and the fourth gives the contribution of each hydrocarbon
to total vapor pressure. The sum of this column gives the calculated
true vapor pressure of the fuel which should be about 0.5 psia above
the RVP. The requirement that this be a realistic value is the most
crucial element in the calculation. It fixes the C.-C,. content of the
fuel; changes in the higher molecular weight range will have a minor
effect on the vapor pressure and on the final result. Column five
gives the concentrations of individual hydrocarbons ascribed to liquid
fuel and column six gives that ascribed to vaporized fuel. The totals
include all those hydrocarbons presumed present in the fuel and in the
atmosphere, including those not measured. They do not include the
cracked products. Column seven gives the total of column five and six
and column eight gives the deviation of column seven, the "best fit"
concentrations, and column one, the measured concentrations. The fit
-------
46
46
DATA S2T=47
11BR RIDROCAWOSS 7.V
FIGURE 30
UBS H?DnocAZBc;is in SET-HI
J
j
1
i
j
J
j -• —
<
1
i
\
i
j
1
i
SSf
•ilfciF
£T.^£V/£'£'
p-VIi^l1' ""''"
p-cf.V/i TOLUEHE
!-!-S?S?L TQLUEIT5 "
ISOaUT&'it; .. .
CIS-2-3UTB:iF.
— t^rSITfi, BUTEliS-i
2-/€T.71'L 3UTEIIE-2
ISOPROPZL BEllZEilE
ti-PROP'fL BEllZS'lS
— O^ST'^fL TO LUSH S
135 TPJMETE2L 3SHZEI!
124 TPJMETttfL BSHZZX
—•123 T?.It'!SntXIi BEl.'ZSti
C2CLO HEZTAflSS
CICLO OCZA11ES
CiCL,U U'jllAliSo
HSPTSJES
C'jfCLO HEXEuES
CYCLO HSPTS11E3
11 CAX30H AROl'lATICS
12 C.t330(V A301-1ATICS
8/5/70
13.800
33.400
25: 200
14.400
C . 800
5.300
3.400
6. £00
8. '700
3 . JOO
1.500
2.500
i. I ;>;
1.200
0.600
' 17300"
142.800
I
7/10/72
0.020
0.071
O.OOj
0.055
0.005
w . 012
0.040
0.027
o . on -
0.023
0.020
0.018
0.018
0.013
0.010
OV01S '
0.001
0.-001
0.001
0.001
0.001
0.003
0.008
0.010
0.010
0. 010
0.010
0.015
G . GIG
0.120
0.100
0.013
0.003
U.UUJ
0.030
0.003
u.Gu«:
0.003
0.007
0.012
0.010
1.001
— ?O1r..
10
100°F
4.920
3.510
1 . 390
1.050
0.573
0.497
0.4S3
0.415
0.340
0.257
G. G i G
0.025
0.023
0.01S
0.014
' Q-.G14
12.630
4.300
4.320
3.500
3.130
1.050
0.980
0.014
C.014
0.014
0.014
0.014
0.100
t!r20"<7
0.120
0.070
u.uau
0.410
C.130
0.210
0.100
C.010
0.010
0.010
^•8??
£=107. 2882853
M1
JiWJi^^.V lo
JS
2351.920
2351.
90.14U 3
s
920
. 634
) — r-?r^..
I 10
100»f
0.098
0.249
0. 133
0.059
0.003-
0 . GGo
0.013
0.011
O.C10
0.005
J . C G 4-
0.000
0:000
~'J . v*^ -~"
0.000
0.000
0.000
'0.010
0.004
0.004
0.004
0.003
O.OOB
0.003
0.000
0.000
o.occ
0.000
0.002
0.010
U . u <- I;
0.002
0.000
U.UU'J
0.012
o.coo
0.001
0.001
0.000
o.coo
0.000
0.700
\Q *
") — Tt
I
(FPB)
2.146
7.617
6.003'
0.526
1. 237
4.292
2.937
3.111
2.146
5.354
1.931
1.931
1.931
1.073
t. 60?
0.086
0.107
0.107
0.107
0.107
0.358
0.853
1.073
1.073
1.073
1.609
1. G73
12.875
10.729
1.395
0.322
j.322
3.219
0.322
0.322
0.751
l.oGa
1.287
1.073
107.374
A* C AijTAt" / ^~i i
I 10 I I 10
(??3) (P?3) (??3)
12.371
31.332
x T ,301
7.453
0.423
2.328
1.409
1.240
0.571
0 . -+40
0.057
0.052
0.041
0.018
0.'C25'
1.270
0.541
0. 543
0 . 440
0.394
1. 287
1.056
0.98S
0.018
0.018
0. 013
0.018
C.026
.j.Gld
0.211
1.257
C. 754
0.196
0.025
O.uls
1.546
0.049
0.079
0.088
0.01:)
0.015
0.013
87.965
»«• S J — U.V
7=125.7239706 7*S£/AO( .Z">P(
F
506.219
14.517
33.949
13.471
C.350
2 . v j /
5.520
".305
X . - 1 J
4. 351
2.317
5. iiC4
1.953
1.9d3
1.972
1.090
1.63S
1.355
0.648
0.650
0.547
0.501
1.914
1.S44
1.090
1.090
1 . 090
1.090
1.636
1. 030
13.086
11.986
3 . 9 / 3
1.591
0.348
'J. 341
4.765
0.371
0.401
0.839
1.62-3
1.303
1.085
195.339
AV3
10)] =87.
4.283
"C.549
1 - 323
"-.;.: 3 o
• c . 4 i? —
"0.520
0.906
C . 5 1 '?
C.249
5.833
"0.498
_0.517
"0.772
"0.^90
' ~0.335" '
28.217
1.558
96503907
..
FIGURE 30
Vapor/Liquid Computation
-------
47
ppb
60--.
50--
40--
30--
20--
10--
ppb
20-4-
10 +
0
GASOLINE VAPOR
0.7 otm. at 100° F
42.5 ppb total
-H-
-K-
78888999999999
FUEL AROMATICS
I 2 3
NATURAL
GAS
-f-H-
2233
CRACKED
PRODUCTS
4444555
FUEL OLEFINS
445566665678N
FUEL SATURATES
FIGURE 31
ppb
60--
50--
40--
30--
20--
10--
ppb
20--
10--
VAPORI2ED GASOLINE
0.7 atm. at IOO°F
189 ppb total
• I..I,
.,1
67888899999999 10-12
FUEL AROMATICS
JlJ
I
I 2 3
NATURAL
GAS
22334
CRACKED
PRODUCTS
4444 5556-8
FUEL OLEFINS
445566665678N
FUEL SATURATES
FIGURE 32
-------
48
in this example is reasonably good; benzene is the poorest fit which
may be due to formation of benzene in engines as a cracking product.
It is not surprising that vapor accounts for most of the butanes and
over half of the pentanes. In contrast, it accounts for only about
2-3% of the xylenes. In comparing the totals derived from liquid and
from vapor, it must be remembered that this treatment is "by moles"
whereas emission inventories are expressed by weight. If we take the
fuel molecular weight to be 100 and the vapor to be 60, then the vapor
contribution in the case illustrated is just half the liquid fuel
contribution. To this liquid contribution must be added the cracked
products found in exhaust gas. This would make the vapor contribution
about 30-40% of the exhaust contribution which agrees with the gener-
ally accepted emission estimates. Some results obtained on other
samples are given in Table 10.
TABLE
Sample Date
6 Aug 70, afternoon
24 Mar 71, afternoon
21 Aug 69, roof, afternoon
21 Aug 69, hill, afternoon
10 Feb 71, morning
10. Hydrocarbons
Oxidant
(ppm)
0.65
-
0.30-0.33
0.30-0.33
-
Liquid
(ppb)
107
93
84
34
189
Vapor
(ppb)
88
51
48
28
43
V/L
0.82
0.55
0.57
0.82
0.23
Hydrocarbons in Other Locations
Since one liter of sample or less suffices to make an analysis for
two dozen hydrocarbons, it was possible to analyze samples from a number
of locations around the country and around the world. Except for loca-
tions in California, samples were taken by willing cooperators to whom
bottles were shipped by common carrier. Aside from southern California,
the following locations have been sampled:
-------
Bay Area Honolulu
San Joaquin Valley Tokyo, Japan
Point Arena McMurdo Sound (Antarctica)
Salinas Valley New York
Denver
Two of these locations, Point Arena and McMurdo, were chosen to
estimate the background levels of hydrocarbons, methane in particular.
The values found were near or perhaps a little lower than the 1.39
ppm found in the southern California mountains. The levels of ethane,
ethene and acetylene were also similar. Several of these groups of
samples have been described in previous papers so that discussion
will not be repeated here. Generally speaking, all these other areas
showed hydrocarbon levels and distributions similar to those found
in southern California. There was considerably more variation from
sample to sample than in southern California. The reasons for this
are not known. In spite of it the general impression was that of
auto emissions rather than any of the other sources described earlier.
Tokyo and Honolulu were of special interest because they have little
or no natural gas, although importation of LNG is beginning. As
expected, methane values were low (near the background value) and the
amounts found might be ascribed to other sources (auto exhaust, fuel
combustion). Honolulu probably has poor atmospheric ventilation
only on rare occasions. Since our samples were taken on one day only,
not necessarily a stagnant air day, this does not constitute a good
test of the prediction that the lack of natural gas should lead to
methane values near the worldwide background.
Tokyo is of special interest for two additional reasons. It has
a very rapidly expanding automobile population (tripled in ten years)
and the characteristic symptoms of photochemical smog are increasingly
noticeable (oxidants over 0.3 ppm). In addition it has an unusual
population of LPG powered vehicles (especially taxicabs). Since
propane and propene are major components of LPG exhaust, special attention
was paid to these two components in Tokyo air. They were not found in
exceptional amounts in Tokyo air. While the proportion of taxis in
-------
50
Tokyo which run on LPG is unusually high, they nevertheles.s do not form
a large part of the total vehic-le population. The Japanese are now
using GLC atmospheric, hydrocarbon .analysis to study their problem.
The New York samples were taken in Manhattan^ near heavy traffic
areas. Concentrations were quite high but remarkably similar to
Riverside in distribution. The Denver samples were also more variable
than the southern California experience.
-------
51
HYDROCARBON REACTIVITY
Since the very beginning of research on photochemical air pollution,
it has been evident that the reactivities of individual hydrocarbons
are strongly dependent on their structure. This is true whether
reactivity is judged by chemical criteria such as ozone formation,
oxidation of nitric oxide, or hydrocarbon disappearance, or by symp-
toms such as eye irritation, plant damage, or aerosol formation.
The correlation between these measures of reactivity is neither strict
nor quantitative but there is a tendency for them to go together. After
some investigation, a rather clear pattern of chemical reactivity
began to emerge. Compounds which contain double bonds are much more
reactive than those which do not. This was apparent first for the
olefins and later for the aromatics, some of which proved to be more
reactive than olefins of the lowest reactivity. From the nature of
the products, it could be inferred that rupture of the double bond
occurs in both cases, at least for olefins and polyalkyl aromatics.
Benzilic hydrocarbons form a special case (see later). A further
parallel appeared in that addition of alkyl groups adjacent to either
olefinic or aromatic double bonds increases the reactivity. The
number of such alkyl groups is more important than their size. Tetra-
methyl ethylene is the most reactive compound known and mesitylene
(1,3,5 trimethyl benzene) is comparable in reactivity to the less
reactive olefins. The paraffins, not having double bonds, react more
slowly, although lengthening the chain does increase reactivity,
perhaps simply by increasing the number of potential sites for attack.
Napthenes, in very limited studies, appear to behave like paraffins.
Acetylene is of very low reactivity. These relationships are summarized
in Figure 33.
HHOTOKEACTIVITY OF HYDHOCAUBONS
INCHKASINf. RKACT1VITY
OLEFINS CH2 . CH2 ."dd """I
rou.
c
c c
I I
PARAFFINS CH4—^C-C-C-C
ACETYLENE CH=CH
1
FIGURE 33
-------
52
Throughout this discussion, the term reactivity is used in its
chemical sense; that is, the rate at which the hydrocarbon disappears
under simulated atmospheric conditions (diluted along with nitrogen
oxides to ppm concentrations in air and irradiated with artificial
sunlight). The term reactivity has also been used to mean either the
rate at which nitric oxide is converted to nitrogen dioxide under
these conditions or the amount of ozone produced. Of course, the
formation of objectionable products (eye irritants, plant toxicants,
etc.) is the matter of real concern rather than just the disappearance
of the hydrocarbon. However, there must be a relation between the
two; certainly hydrocarbon which does not react cannot form noxious
products. Schemes have been devised for rating hydrocarbons according
to their tendency to produce such products. Such a rating is some-
times referred to as "hydrocarbon reactivity" but it would be preferable
to reserve this term for its chemical sense and to use a term such as
"smog potential" for ratings based on the formation of objectionable
products. For the most part those compounds which are of high chemical
Q
reactivity are those which also produce noxious products. There are,
however, recognized differences in the ability of hydrocarbons of
various structure to produce the various symptoms.
So far as is known hydrocarbons of all types are capable of causing
the photochemical conversion of nitric oxide to nitrogen dioxide, albeit
9
at vastly varying rates. This reaction is closely related to the
formation of ozone, a toxicant of major concern. Compounds which contain
double bonds also produce peroxyacyl nitrates, a family of plant toxicants
and eye irritants. Saturated hydrocarbons produce little, if any, PAN.
Likewise, the enhancement of aerosol formation in systems containing
sulfur dioxide has been demonstrated for olefins but not for paraffins.
Subjective measurements of eye irritation also show that saturated com-
pounds produce less than unsaturated compounds.
Evaluation of the relative importance of various classes of hydro-
carbons in producing photochemical air pollution encounters a further
obstacle. Some knowledge of the extent to which various hydrocarbons
react under realistic ambient conditions is needed. If very long times
are available for reaction, then hydrocarbons of low reactivity may make
-------
53
a substantial contribution. This is a serious question because the
hydrocarbons of low reactivity constitute a large part of the total
emissions (paraffins, napthenes, benzene and acetylene). Such
emissions can only be retained in the lower atmosphere long enough
for reaction to occur if there is a temperature inversion. Under
such conditions the cooler air at ground level cannot mix with the
warmer air aloft. Such inversions, although they persist long enough
to permit photochemical reactions to occur, are vulnerable to the
normal heating of the lower atmosphere which results from contact
with the warm surface of the earth. There has been no way to estimate
the lifetime of such a parcel of stable air, and this determines the
effective time available for reaction. The fact that hydrocarbons do
vary in reactivity raises the possibility that analysis of polluted
atmospheres could provide an answer.
Acetylene, ethene and propene are a unique trio of compounds for
judging the extent to which reaction has taken place in an ambient air
sample. These three compounds are not present in gasoline but are
formed in auto engines and emitted in exhaust gas in substantial
quantities. Moreover, they differ significantly in reactivity so that
they disappear at different rates under irradiation. The relative
amounts of these three hydrocarbons in exhaust gases vary from car to
car and are dependent on engine operation as well. But the ratios of
the total rates of emission of the three from all the cars in an
urban area are probably nearly independent of time, place and circum-
stance. Evaluation of unreacted samples has lead (see prior sections)
to the values of 0.85 for the ethene/acetylene ratio and 0.25 for the
propene/acetylene ratio in unreacted emissions (molar basis). To study
the effect of photoreaction, we have expressed each of these three
hydrocarbons as a percent of the total of the three. Thus ethene
comprises 0.85/(1 + 0.85 + 0.25) = 40.5% and propene 0.25/(1 + 0.85 +
0.25) = 11.9% of the total. Expressed in this way the effect of reac-
tion can be studied independently of dilution.
In previous sections evidence was presented that under specially
chosen conditions ambient air contains quite high concentrations of
unreacted auto exhaust (dilutions of 2000 or so). Table 9 showed
-------
54
that the relative amounts of acetylene, ethene and propene were quite
consistent in unreacted air. The remarkable and revealing fact about
these samples is that symptoms of air pollution are completely absent.
On one occasion, the evening paper reported "this morning Riversiders
awoke to a cloudless sky, and air unmarred by smog, and as a result
the mountains stood sharply etched against the horizon."
Ambient Air Irradiation
In spite of this, analysis shows that high concentrations of auto
exhaust and natural gas hydrocarbons are present under these conditions.
It would seem that monitoring of the air on such a day would provide a
particularly simple and useful way to study development of photochemical
smog. This is not possible because the radiation inversion which
provides this trapping is destroyed by solar heating before the
photochemical reaction can proceed very far. In order to study hydro-
carbon disappearance under irradiation, it is necessary to trap a sample
and then irradiate it, either with artificial or natural sunlight. The
results of such an experiment are shown in Figure 34 . Initial concen-
trations were substantially higher than found in smog; even of those
HYDROCARBONS IN MORNING AIR
ppb
60-
50-
40-
30-
20-
3.
10-
0-
xlOO
Pf i 10 FEB. 1971
20" 7:45 -8=00 AM
IOT m
II
i Q i i a 0
" 67888899999999
FUEL AROMATICS
AFTER 4 HOURS UV
|
i I
lei si* n 0 a 1
(
O.ll J
23 22334 4444555 4455666656
NATURAL CRACKED FUEL OLEFINS FUEL SATURATES
GAS PRODUCTS
FIGURE 34
-------
55
hydrocarbons of very low reactivity such as acetylene and the paraf-
fins. The presence of substantial amounts (3-5 ppb!) of some highly
reactive compounds (four and five carbon olefins and diolefins) shows
that this sample was largely unreacted. The amounts which disappeared
in four hours of ultraviolet were quite consistent with reactivities
q
determined in the laboratory studies (see for example, Tuesday.
In four hours about 37% of the ethylene reacted.
Although the olefins are most reactive and paraffins least reactive,
the quantities present reverse this order. Thus, quantity and reacti-
vity, when multiplied together, become roughly constant. The aromatics
are intermediate in both quantity and reactivity. This indicates that
very little can be gained by focusing attention on any one group of
hydrocarbons. It is also evident that the cracked hydrocarbons (those
present in exhaust but not in gasoline) contribute substantially to
total reacted hydrocarbons. It should always be kept in mind that
substantial additional hydrocarbon in the higher molecular weight
range is present but not measured. Presumably, these would follow
the reactivity pattern of these respective classes and not change the
overall conclusion.
With two 20-liter borosilicate carboys, it was possible to take
a pair of ambient air samples simultaneously so that one could be
irradiated in artificial sunlight and one in natural sunlight. This
was done with a pair of samples taken on October 24, 1966 between
0630 and 0650 PST. These were taken on the roof of a five-story
building in central Riverside. This is near downtown traffic but far
enough removed to permit mixing. The two samples were nearly identical
in composition: one was irradiated in natural sunlight beginning
about 0730 PST and the other in artificial ultraviolet the next day.
The sky was clear during the entire period of irradiation with natural
sunlight.
A portion of the results of photolysis by natural sunlight of one
of these samples is shown in Figure 35 , and the comparison sample
photolyzed with ultraviolet in Figure 36 • There are several
notable features in these plots: the natural sunlight sample showed
more reaction than the artificial but the relative reactivities were
-------
56
NATURAL SUNLIGHT IRRADIATION
OF AMBIENT AIR 10-24-'66
• C=C
C-C-C-C-C
0.1
FIGURE 35
1401
120.
100-
80-
90'
40-
20-
SUNLIGHT IRRADIATION OF AMBIENT AIR
HYDROCARBONS REACTED 10-24-'66
11 Paraffins
1201efuinC2-C5
6 hrs
hrs
8 hrs
2 hrs
0<
10 20 40
Paraffins reacted, ppb
FIGURE 37
so
ULTRAVIOLET IRRADIATION
OF AMBIENT AIR 10-24 -'88
100
C-C-C-C-C
0.1
246
Hours irradiation
FIGURE 36
ULTRAVIOLET IRRADIATION
OF AUTO EXHAUST
100
.C-C-C-C
.• C-C-C-C-C-C
'" C-C-C-C
C-C-C-C
0.1
2 4 6
Hours irradiation
FIGURE 38
-------
57
very similar. The natural sunlight plots have a maximum slope near
noon which reflects the maximum intensity of sunlight. The artificial
irradiation lines are nearly straight on these plots of log concen-
tration versus time. This rather surprising result indicates that the
reagents which are attacking the hydrocarbon are substantially constant
throughout the irradiation. The attack on the hydrocarbon can be
written:
where R. is a reagent (e.g. 0, 0~, OH, NO-) which attacks hydrocarbon
with a rate constant k.. This can be rearranged and integrated to
give log CQ/Cf = 0.434 Ik± ^dt.
o
The linearity of the log versus time plot indicates that the last
term is linear with time and therefore that either all R.'s are constant
i
or that somehow the changes in the R.'s counteract each other. Previous
irradiations did show a slowing down of the rate after prolonged
irradiation (24 hours). Before these ambient air experiments were done
it seemed possible that the reaction would slow down or even stop after
a few hours of irradiation because of consumption of photoinitiators.
From these two irradiations relative rates of reaction of various hydro-
carbons were derived by calculating log-pC /C,. with the results shown
in Table 11. For the olefins these relative reactivities are in reasonable
agreement with values determined in synthetic mixtures at concentrations
Q
two or three orders of magnitude higher and reported several years ago.
The equipment and procedure used at that time did not permit evaluation
of the rates of the less reactive hydrocarbons.
-------
58
Table 11
Relative Rates of Disappearance of Hydrocarbons Under Irradiation
Ethene
Propane
i-Bucene
i-Butene
tr-2-pentene
n-butane
n-hexane
Blacklight
log C0/Cf
0.217
0.818
0.823
0.665
4.2
0.095
0.132
Ratio'
1.00
3.8
3.8
3.1
19
0.44
0.61
Sunlight
log C0/Cf Ratio3
0.345 1.00
1.45 4.2
0.79 2.4
0.895 2.6
0.090
0.167
0.26
0.48
Ref (8.)
0.83
3.8
5.2
(22-18)1
a Ethene = 1.00
2-butene and 3-hexene
The relative contributions of olefins and paraffins were expressed
in another way by summing the amounts reacted after each time period
with the result shown in Figure 37 which was derived from the same experi-
ment as Figure 35. This plot is curved because the olefins react more
rapidly and so are depleted earlier in the experiment. After eight
hours about 130 ppb of olefin (mostly accounted for by propene and
ethene) and 46 ppb of paraffin had reacted. Inclusion of higher
molecular weight hydrocarbons would change this picture somewhat,
principally by inclusion of the moderately reactive aromatics. According
to air monitoring data the original mixture contained 0.14 ppm (140
ppb) of nitric oxide which would, according to present understanding
of the photochemical reaction, be converted to nitrogen dioxide in the
first stages of the reaction. Laboratory studies suggest that oxidation
of one molecule of hydrocarbon is capable of converting about two
molecules of nitric oxide. Then about 70 ppb of hydrocarbon would be
-------
59
required for conversion of the nitric oxide initially present. If
we estimate that the total hydrocarbons reacting are twice the amount
measured this might require about two hours of irradiation.
Auto Exhaust Irradiation
A sample of auto exhaust diluted with purified air was irradiated
with ultraviolet to generate the data shown in Figure38. This sample
was obtained with the cooperation of Scott Research Laboratories. A
1963 V-8 Chevrolet using "basin mix" gasoline and running on a chassis
dynamometer in the California standard driving cycle was used as a
source of exhaust. Exhaust from a single cycle was collected in a
plastic bag after thorough warm-up of the engine. A 20 ml portion of
exhaust gas was transferred to the 20 L bottle which had previously
been flushed with purified air (also obtained from Scott Research
Laboratories). This 1000-fold dilution produced a mixture which,
except for propane and other light paraffins, closely resembled that
obtained from ambient air. The contents of the bag were also analyzed
by the nondispersive infrared instruments used for exhaust testing.
This showed 305 ppm of hydrocarbon (as n-hexane) , 3.57o carbon monoxide,
and 8.27o carbon dioxide. This hydrocarbon concentration is below
average but not enough to be considered atypical. The rates of disap-
pearance of the individual hydrocarbons in this experiment were
approximately the same as in the ambient air experiment. The reason
for the curvature of some of the plots is not knowne
Extent of Reaction
This type of irradiation data can be compared to ambient air
analyses to judge the extent to which reaction has occurred. This is
important because it bears directly on the value of various control
strategies which call specifically for a reduction of the most reactive
hydrocarbons. From laboratory studies of auto exhaust irradiation, it
was not possible to tell whether most of the reaction was due to the
high reactivity compounds (higher molecular weight olefins) or whether
actual ambient air conditions were such that hydrocarbons of medium
-------
60
and low reactivity could participate. This is related to the time
available for reaction; or the effective residence time of pollutants
in the atmosphere. Acetylene, ethene and propene forma unique set
of compounds for judging the extent of reaction since they are almost
exclusively derived from auto exhaust; they are emitted in known ratios
to each other (see Table 9 ), and they differ quite widely in
reactivity. Dilution will reduce all three concentrations in propor-
tion, but photochemical reaction will destroy the propene most rapidly,
the ethene at a moderate rate and the acetylene very slowly. Extent of
reaction could be judged either by the propene/acetylene ratio or by
the ethene/acetylene ratio. To make use of both ratios the sum of the
three hydrocarbons was taken and each olefin expressed as a percentage
of the total. These percentages for the batch irradiation of ambient
air samples gave Figure 39 when plotted against each other. A
similar experiment carried out at the winter solstice (minimum solar
radiation) showed similar reactivity for both sunlight and ultraviolet
irradiation (Figure 40 ). It is obvious that smog samples could
be plotted on this chart to judge the effective reaction time. But it
should be remembered that these were batch experiments, whereas the
outside air has continuous addition of pollutants. The other extreme
would be a stirred flow reactor at some stated residence time. Since
there was no way to determine this on an ambient air sample the expected
behavior was calculated from the batch experiments.
With constant intensity of irradiation (UV) the rate of disappearance
was nearly first order (see Figure 36 ). First order rate constants
derived from these laboratory data were then used to compute the theoretical
change in composition of the mixture in the proportions given above both
for a batch reaction in which concentrations decrease exponentially
according to the equation C = C exp(-kt) and for a stirred flow reactor
in which case the concentrations fall according to C = C (1 + kt)
These two extreme types of reactor give different plots as shown in
Figure 41 . The actual atmosphere is probably somewhere between
these two.
-------
61
15 n
10-
2
Cu
5-
25
AMBIENT AIR
PHOTOLYSIS
10. 24. 1966.
8
HOURS
0 0
30 35 40
% Ethene
L/V
45
Ratio % k 10 "
c
,«„*.« i
21 'DEC ''67 h°UrS '
: : o;o • :
i
sun :
2 • o 2
uv
rt
i 30 : 35
i '?S Ethene
FIGURE 40
-------
62
15-
10-
I
2
5-
25
THEORY
Fresh
Exhaust
30 35
% E t h e n e
Batch
40
45
AMBIENT AIR PHOTOLYSIS.
(Ethene + Propene + Acetylene =100 %)
FIGURE 41
-------
63
If additional exhaust gas is added to a reacted mixture (say 10
hours batch reaction as shown in Figure 41 ) the composition will
be moved along the straight line marked "fresh exhaust." Such addi-
tion, of course, also lowers the average irradiation time of the
mixture. Since all types of history are possible, atmospheric
samples may lie anywhere within the crescent area of Figure 41.
In Figure 42 are plotted the observed percentages for a number
15
10
I
5-
25
AMBIENT AIR SAMPLES
(Hour and Date)
Unreacted
7300
4.26O
»i
/?
1445
10.9 O 100° S
//
J200
8.27'
1610
10.24
1510
'1314
9.27
(L.A.)
'70 0.6 Oxid
30
35
% Ethene
40
45
EXTENT OF PHOTOLYSIS OF AMBIENT AIR.
(Ethene + Propene+ Acetylene = 100%)
FIGURE 42
of smog samples. Most of the data lie in or near this crescent. In fact, the
two best sets of samples (9/6/68 and 10/24/68) fall near each other
and between the curves for stirred flow and batch reaction near the
point corresponding to eight hours of irradiation. The Pacific Standard
-------
64
Time at which each group of samples was taken is given along with the
date. All but one set of samples were taken in Riverside. Although
there is considerable scatter the points do lie in the general area
of the stirred flow and batch reactor plots of Figure 41 and the
samples taken later in the day show the most reaction.
The data for the 0.6 oxidant day of August 6, 1970 is slightly
below the curve for batch reaction. It is tempting to interpret
this as meaning this air was reacted for several hours in isolation
from fresh exhaust. Perhaps this could happen if a parcel of air
irradiated aloft were brought down to the surface. Recent research
has suggested that high oxidant levels are sometimes formed aloft.
In such circumstances, a parcel of polluted air might undergo reaction
and then be brought down to the surface.
The samples taken on 10/24/68 showed the highest degree of reac-
tion, along with some of the highest concentrations of any set of
samples. These were taken in late afternoon just after a very sharp
pollution front passed Riverside. By comparing this afternoon's
analysis with an unreacted morning sample it was possible to make a
crude estimate of the amounts of the various hydrocarbons which would
have been present if no reaction had occurred. The following assump-
tions were needed:
1) Reaction equivalent to 10 hours in the ultraviolet bottle
reactor had occurred. This was based on the plots shown in Figures
39 and 42.
2) A set of first order rate constants k.-(hr ) were derived from
the bottle photolysis. These were used in the first order stirred
flow rate expression C = C(l + kt) with t = 10 hours to calculate the
amounts of acetylene and other hydrocarbons of low reactivity present
before reactions. Use of batch kinetics would make little difference
here.
3) The amounts of ethene and propene were estimated from the
acetylene concentration and the ratios of these hydrocarbons in unreacted
exhaust.
4) The C, and C olefins before reaction were assumed to be equal
to the amounts found in the unreacted morning sample because the amounts
-------
65
of Cr-C, paraffins were similar. Both groups are probably derived
D o
primarily from gasoline.
5) A previously published analysis of "basin-mix" fuel was used
along with an estimated gasoline composition of H70 olefin, 3570
aromatic and 5470 saturate. These were used to estimate the amounts
of higher hydrocarbons in the three classes which went unmeasured.
The higher saturates were assumed to have been comparable in
reactivity to the C..-C paraffins and the higher olefins assumed to
be 907o reacted. Reactivities for the aromatics were estimated from
Q
Stephens and Scott (generally between ethene and propene except
benzene, toluene and ethylbenzene, all = 0). These latter are extra-
polations on top of estimates so they must be quite crude.
These results are summarized in Tables 12 and 13 . The
total amount reacted is estimated to be just under 200 ppb which is
less than one-third of the amount initially present. If auto exhaust
contains 250 ppm of acetylene and 1500 ppm of nitric oxide, we might
judge that before reaction this sample would have had 300 ppb of nitric
oxide. It is somewhat surprising that reaction of 200 ppb of hydro-
carbon could accomplish the conversion of 300 ppb of nitric oxide to
nitrogen dioxide and the further reaction of this to PAN and other
products. PAN and nitrogen dioxide only accounted for about one-third
of this estimated 300 ppb of nitric oxide. Some of the samples shown
in Figure 42 were much less reacted than the sample of 10/24/68.
Even so they had the symptoms of air pollution even for these very
small degrees of reaction. It would be interesting to try to determine,
using this kind of atmospheric data, the degree of reaction required to
reach some critical point of reaction such as the first appearance of
ozone. One attempt was made to determine the first appearance of PAN
during laboratory irradiation of ambient air, but no lag was apparent.
The PAN began to form as soon as irradiation was started; it may be
that this ambient air sample was not completely unreacted at the
beginning.
-------
66
Table 12 Estimate of amounts reacted in an ambient air sample compared to
an unreacted sample.
Reacted Sample (ppb)
1610 PST 1(5/24/68
Present Reacted (A)
Methane
Ethane
Propane
°Acetylene
Ethene
Propene
Methyl Acetylene
1,3-Butadiene
°l-Butene
Isobutene
Trans-2-Butene
Cis-2-8utene
2-Methyl Butene-1
Cyclo Pentene*
Trans-2-Pentene
2-Methyl Butene-2
plsobutane
N-butane
Isopentane
N-Pentaneb
Cyclo Pentane
2,2-Dimethyl Butane"
2,3-Oimethyl Butane
2-Methyl Pentaned
3-Methyl Pentane
N-hexane
2530
72.2
49.9
42.0
17.9
1.27
1.2
0.29
0.26
2.0
<0.2
<0.2
0.35
0.8
<0.2
<0.2
19.7
62.0
41.2
21.4
2.4
0.86
1.9
9.6
6.1
8.3
0
0.6
2.6
6.3
23.1
10.7
0.3
1.5
2.3
3.2
1.4
1.4
2.1
3.2
2.4
2.6
3.0
16.3
12.8
7.9
0.6
0.3
4.4
3.9
3.5
Unreactad Sample
(ppb)
0730 PST 9/24/68
Present
2355
63.6
18.8
77.0
65.6
19.2
2.6
3.6
2.6
5.2
1.4
1.4
2.4
4.4
2.4
2.6
8.0
27.6
39.2
22.4
3.2
1.6
3.8
12.4
8.4
9.0
Includes: • 2-Methyl Butadiene-1,3; b 3-Methyl Butene-1; c 1-Pentene; d Cis-2-Pentene.
C, O, P—See Table V
Table 13 Riverside ambient air (ppb) on roof of health-finance building. 1610 PST 10/24/68
(In ppb by volume except where noted.)
Compound
Ci - Cj Paraffins (meas.)
Non-fuel unsaturated0 (meas.)
C4 - d Olefins0 (meas.)
C< - C4 Paraffins1" (meas.)
Higher olefins (est.)
Higher saturates (est.)
Aromatics (ast.)
Total (less CH4)
Total (ppbC. incl CH4, GLC)
Total (ppbC, non GLC)
Acetylene
Carbon monoxide
Oxidant
Nitrogen oxides
PAN
Present +
2652
62.7
3.4
173.5
2.2
42
67
473
4800
eono
42.0
5000
400
60
34
Reacted* =
3.2
41.9
18.6
52.7
19.6
21.0
40.9
198
Initial*
2655
104.6
22.0
226.2
21.8
63.0
107.9
671
48.3
300
a estimated
C, O, P—See Table IV
-------
67
MODES OF REACTION
This comparison of smog hydrocarbon distribution with irradiated
samples indicated that hydrocarbons of all classes participated in the
reaction. Even the aromatics (except benzene) react substantially in
the time available. During the course of this program Heuss and
12
Glasson at General Motors discovered that benzilic hydrocarbons
could be oxidized under atmospheric conditions to peroxybenzoyl nitrate,
an extremely potent eye irritant. This was surprising not only because
of the eye irritation but also because it indicated that side chain
oxidation occurred. In contrast, xylenes and mesitylene produced PAN,
clearly showing that ring rupture occured.
OXIDATION OF MONO AND DI ALKYL BENZENES
P P
A
^ ^ CH3
^*" P P
CH* ^ CHoCf N(
3 3 bo'o
CH3
-------
68
This striking difference in products clearly indicates a differ-
ence in active species attacking the hydrocarbon. The active species
most suggested for this role are atomic oxygen, ozone and hydroxyl
radicals. Ozone reacts rapidly with olefins but its reaction with
aromatics is too slow to be considered. Atomic oxygen is known to attack
aromatic hydrocarbons in the ring so it might be suspected that side
chain oxidation is initiated by hydroxyl radical. Another possibility
is the NO. radical which is formed by the reaction of ozone with N(L.
All these active species are present in steady state, and the probable
reactions and steady state equations are as follows:
REACTIVE SPECIES
0ka(NO9)
//-»\ a *
(O) =
k3(02)(M)
O3 O3 + NO ' » NO2 + O2 pka(NO2)
3 k4 (NO)
N02 + 03-^1^N03 + 02 R (Q ,
, (N03) = —_L
N03 + N02
OH
It is noteworthy that the N0_ steady state concentration is directly
proportional to the ozone concentration and independent of the nitrogen
-------
69
dioxide concentration. Calculations based on published rate constants
yield an NO- concentration of 2 X 10" (0_).
MECHANISM
NO + NO -» 2N02 k
N2°5 + H2° ~* 2HN°3 fast
k (NO )(0 )
(N03) Steady State = _
(NO)/(N02) -» 0
(N03) = 2 X 10"5 (03)
There are a few hints in the literature that N0« might be able to
13
abstract atomic hydrogen from organic molecules. Principal evidence
is the ability of N_0 to react with aldehydes in air in the dark to
1 /
produce PAN. Another reaction was attributed to N0» by Wilson
N°2 + °3 + S°2 "* aeroso1
If various species were attacking different hydrocarbons at different
rates, the rates of disappearance should vary during the course of the
reaction. This did not seem to be true in the ambient air irradiations.
A test with three selected hydrocarbons (one paraffin, one olefin and
one aromatic) in the 20 liter bottle showed no variations in the rate
of consumption of all three hydrocarbons after five hours of irradiation.
(See Figure 43 )
Since this was an uninformative result, it was decided to try to
maximize the opportunity for reaction of N0_. This was done in two ways:
-------
70
100 •
70 •
50
40
30 •
2 20
h
z
o
o
10
ULTRAVIOLET IRRADIATION IN 20 I.ITKR ROTTf.K
ISOPENTANE RATE = 0.093 hour
MliTA-XYLENE
RATE =0-364
hour
-1
234
HOURS OF IRRADIATION
FIGURE 43
-------
71
1) some high concentration experiments were done in the long-path
infrared cell, and 2) the 50 liter stirred flow reactor was used
with mixtures of NCL, 0- and selected hydrocarbons in the dark.
In the infrared cell concentrations of 500-800 ppm of an aromatic
hydrocarbon were mixed with about 1000 ppm of nitric oxide and ozone
(to generate NO- and N^Oj.) • Even at these high concentrations the
reaction was not fast. Principal effort was directed toward detec-
tion and measurement of peroxybenzoyl nitrate (PBzN). This compound
— 1
has a distinctive band at 987 cm~ which has been used to identify PBzN
produced from benzaldehyde. Electron capture chromatography was
also used to identify the PBzN and PAN estimated from the infrared
spectra are given in Table 14.
TABLE 14
Toluene
Ethyl benzene
n-Butyl benzene
1,3,5-trimethyl benzene
HC ppm
590
737
581
633
PBzN
0.91
0.56
3.52
-
PAN
-
-
-
8.4
This positive result showed that aromatic hydrocarbons can produce
PBzN in a dark reaction, but the very slow yields indicate that this
is of little importance.
Stirred flow experiments were then carried out at 3000 fold lower
concentrations in the flow photoreactor. Since longer chain benzilic
hydrocarbons were more reactive than toluene, n-propyl benzene was
included in this series. The results indicate that at this concentra-
tion level little reaction occurs even in residence times of several hours.
The hydrocarbon mixture, the N0«, and 0., were dispensed from
separate LPO tanks, using hi-pure N as the diluent in all cases. LPO
concentrations were sought such that convenient dispensing rates through
the rotometer flow panel would admix with the carbon filtered air in the
mixing manifold and pass finally into the 50 liter reaction bottle
producing approximately:
-------
in the first experiment -
72
HC
0
200 ppb
0.2 ppm
0.5 ppm
and in the remaining experiments -
HC
0
200 ppm
1.0 ppm
2.5 ppm
In all experiments acetylene was added to the HC LPO tank to
assist in monitoring dispensing regularity and mixing. In the last
experiment trans butene-2 was added to demonstrate the capability of
obtaining a dark reaction at the concentrations used in the
experiments with the equipment as it was presently devised.
HC measurements were made by the standard concentration tech-
nique, from samples taken at the 50 bottle inlet or exhaust ports,
and subsequently analyzed on the appropriate FID gas chromatograph.
N02 concentrations were determined colormetrically using
Saltzman's reagent.
0 concentrations were determined using the Mast ozone meter.
TABLE 15. Reactivity of Aromatic Hydrocarbons in the Flow Photoreactor
ppb inlet
Page
No. Cond.
52 2 hrs 232 210
Dark
228 205
ppm
_EE^_25l}§U§t EE2_i!}i£t_ ££!}§"§£_
b NC-d> T^C = 0,, N0_ 0,, N(
46
48
1 hr 239 119
Dark
1 hr
Dark
232
249
108
123
.42
•
- off
scale
.13
.68
.32
.11
.09
.68
off .99 .58 .18
scale
-------
73
TABLE (continued)
Page
No. Cond.
/55 1 hr
*< 55 2 hrs
Dark
»55 4 hrs
Dark
(7 1 hr
U-V
light
7 2 hrs
U-V
light
8 4 hrs
U-V
light
8 static
U-V
light
61 1 hr
Dark
61 2 hrs
Dark
62 4 hr
Dark
62 4 hrs
U-V
light
62 static
U-V
light
66 1 hr
Dark
C2=
230
227
232
243
249
243
[228
233
246
224
230
201
ppm
p_p_b inlet ppb exhaust ppm inlet exhaust
Cl0 NC30 T2C4= C2, CL0 NC30T2C4= 03 N02 ^ NO,,
280 - - 232 284 - - off 1.07
scale
284 - - 230 279 - - measurement taken 12 hrs
after end of expt
270 - - 236 266 - -
317 - - 233 294 - - same LPO tank as above; no
0,. or N0~ measurements taken
302 - - 250 288
284 - - 232 269
218 (T-0)] - (168 56 (T=24 hrs)]
218 - 230 - 230 - - - .37 .62
204 - 243 - 216
189 - 227 - 179
163 - 224 - 163
(192 - 69 (T=15 hrs)]
179 122 200 - 192 40 off 1.4 .23 1.3
scale
Continuous experiments
-------
74
Table 16 summarizes the data for all of the experiments. The percent
reactions of the aromatic hydrocarbons used are as follows:
1 hr
DARK
2 hrs 4 hrs
C.0
1
NC_0
J
T2C4=
2%
neg
67%
2%
neg
-
2%
5%
-
1 hr
7%
LIGHT
2 hrs 4 hrs static
5%
5%
75%
58%
Based upon a 15 hour static reaction; presuming linearity, in 24 hours
(as was the time with C 0), this NC 0 reaction would have gone to 93%.
The 67% dark reaction (residence time = 1 hr) of T»C = demonstrates
the capability of the hardware as devised and the suitability of the
concentrations of the other reactants. The static reactions demonstrate
a similar capability in the presence of U-V light. Although the dark reaction
has been reported at considerably higher concentrations, the present series
of experiments suggest that at these lower concentrations the reaction
occurs at a rate too low for detection by the instrumentation and procedures
utilized.
-------
75
References
1. Eggertsen, F. T., and Nelsen, F. M. "Gas Chromatographic Analyses
of Engine Exhaust and Atmosphere," Anal. Chem., 30: 1040-1043
(June 1958).
2. Fink, U., Rank, D. H., and Wiggins, T. A. "Abundance of methane in
the earth's atmosphere," J. Opt. Soc. Amer., 54, 572-574 (1964).
3. Goldberg, L. , and Mu'ller, E. A. "The vertical distribution of nitrous
oxide and methane in the earth's atmosphere," J. Opt. Soc. Amer.,
43, 1033-1036 (1953).
4. Maynard, J. B., and Sanders, W. N. JAPCA 19(7), p. 505 (1969).
5. Stephens, E. R., and Burleson, F. R. Distribution of Light Hydrocarbons
in Ambient Air Paper 69-122, APCA Journal 19_(12) , 929-936 (December,
1969).
6. Stephens, E. R., E. F. Barley, and Burleson, F. R. Sources and
Reactivity of Light Hydrocarbons in Ambient Air Proc. API Div. of
Refining 4^, 466-483 (1967), Los Angeles, California, May 16, 1967.
7. Stephens, E. R., and Burleson, F. R. Analysis of the Atmosphere for
Light Hydrocarbons APCA Journal 1J (3), 147-153 (March, 1967).
8. Stephens, E. R., and Scott, W. E. Relative Reactivity of Various
Hydrocarbons in Polluted Atmospheres Proc. API 42, III, 665 (1962)
San Francisco, California, May 17, 1962.
9. Glasson, W. A., and Tuesday, C. S. Hydrocarbon Reactivities in the
Atmospheric Reaction of Nitric Oxide ES&T 4(11) November 1970 p. 916.
10. Stephens, E. R. Chemistry of Atmospheric Oxidants Paper 68-57, 61st
Annual Meeting, APCA, St. Paul, June, 1968. APCA Journal 19(3), 181 (March, 1969)
11. Stephens, E. R. The Role of Oxygen Atoms in the Atmospheric Reaction
of Olefins with Nitric Oxide Int. J. Air Water Poll. 10, 793-803
(1966)
12. Heuss, J. M., and Glasson, W. A. ES&T 2(12) December 1968 p. 1109.
13. Tuesday, C. S. Symposium on Chem. Reactions in the Lower and Upper
Atmosphere, Interscience Pub., New York, p. 15, (1961).
14. Wilson, W. E., and Levy, A. API proj. S-ll Battelle Memorial Institute,
Prog. Reports, August 1, 1968, August 1, 1969.
------- |