EPA-650/3-75-002
NOVEMBER 1974
Ecological Research Series
EXPLORATORY STUDY
OF FACTORS AFFECTING
AEROSOL FORMATION
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
-------
EPA-650/3-75-002
EXPLORATORY STUDY
OF FACTORS AFFECTING
AEROSOL FORMATION
by
David F. Miller
Battelle Columbus Laboratories
505 King Ave.
Columbus, Ohio 43201
Contract No. 68-02-1217
ROAP No. 21AKB-09
Project Element No. 1AA008
EPA Project Officer: Basil Dimitriades
Chemistry and Physics Laboratory
National Environmental Research Center
Research Triangle Park, North Carolina 27711
Prepared for
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
November 1974
-------
EPA'REVIEW NOTICE
This report has been reviewed by the National Environmental Research
Center - Research Triangle Park, Office of Research and Development,
EPA, and approved for publication. Approval does not signify that the
contents necessarily reflect the views and policies of the Environmental
Protection Agency, nor does mention of trade names or commercial
products constitute endorsement or recommendation for use.
RESEARCH REPORTING SERIES
Research reports of the Office «i Research and Development, U.S. Environ-
mental Protection Agency, have been grouped into series. These broad
categories were established to facilitate further development and applica-
tion of environmental technology. Elimination of traditional grouping was
consciously planned to foster technology transfer and maximum interface
in related fields. These series are:
1. ENVIRONMENTAL HEALTH EFFECTS RESEARCH
2 . ENVIRONMENTAL PROTECTION TECHNOLOGY
3. ECOLOGICAL RESEARCH
4. ENVIRONMENTAL MONITORING
5. SOCIOECONOMIC ENVIRONMENTAL STUDIES
6. SCIENTIFIC AND TECHNICAL ASSESSMENT REPORTS
9. MISCELLANEOUS
This report has been assigned to the ECOLOGICAL RESEARCH series.
This series describes research on the effects of pollution on humans,
plant and animal species, and materials. Problems are assessed for
their long- and short-term influences. Investigations include forma-
tion, transport, and pathway studies to determine the fate of pollutants
and the.if effects. This work provides the technical basis for setting
standards to minimize undesirable changes in living organisms in
the aquatic, terrestrial and atmospheric environments.
This document is available to the public for sale through the National
Technical Information Service, Springfield, Virginia 22161.
11
-------
TABLE OF CONTENTS
Page
INTRODUCTION 1
BACKGROUND 2
SCOPE OF PROGRAM 3
OBSERVATIONS AND TENTATIVE CONCLUSIONS 3
EXPERIMENTAL METHODS 4
Large Smog Chamber 4
Description 4
Chamber Operation 4
Small Smog Chamber 5
Description 5
Chamber Operation 5
Methods of Chemical Analysis 8
RESULTS AND DISCUSSION 9
SO2 Wall Losses 11
SO2 Oxidation in Polluted Air — Preliminary Experiments 12
SO2 Oxidation in Unpolluted Air 12
Contaminant Experiments 14
SO 2 Oxidation in Polluted Air — Model Smog 15
Mechanisms of SO2 Oxidation 16
Polluted Air 16
Unpolluted Air 19
REFERENCES 21
BATTELLE — COLUMBUS
-------
FOREWORD
This study was conducted under the supervision of Mr. David F. Miller, Principal Investigator,
with Mr. Arthur Levy assisting as Project Advisor. It is a pleasure to acknowledge Dr. Chester
W. Spicer of Battelle and Dr. Jack G. Calvert of the Ohio State University for many helpful
discussions. The contribution of others of the Battelle staff, particularly Messrs. Darrell W. Joseph,
George W. Keigley, and Gerald F. Ward, made possible the compilation of these data, and their
efforts are gratefully acknowledged.
BATTELLE — COLUMBUS
-------
EXPLORATORY STUDY OF FACTORS AFFECTING
AEROSOL FORMATION
by
David F. Miller
INTRODUCTION
Numerous aerometric studies have shown that sulfuric acid and sulfate aerosols constitute a
substantial fraction of the aerosol matter existing in both urban and rural areas throughout the
United States. Compositional analyses of size-fractionated aerosols reveal that most of the sulfate-
type aerosol is in the particle-size range below 2 p.m in diameter.(l)* A more recent study in .
New York by our group showed that 90 percent of the sulfate is in this size range and that
sulfate alone comprises 15-25 percent of the total mass of aerosol <2 nm in diameter.^) in
addition to adverse visibility conditions associated with aerosols of this size range, it is well
established that such aerosols are highly respirable, and the results of an extensive community
health (CHESS) program indicate that a substantial relationship exists between sulfate levels and
some types of morbidity, even at sulfate levels of 8-12 jug/m^ (corresponding to conversion of
only 2-3 ppb
Interpreting data from the U. S. National Air Surveillance Networks. Altshuller concludes
that sulfate-bearing aerosols are widely distributed throughout the U.S. (3) Furthermore, the
occurrence of sulfate appears to be largely due to chemical reactions in the atmosphere (as
opposed to emitted sulfate), and there is considerable evidence that long-distance transport of
SC>2 from urban areas concomitant with SO2 oxidation is responsible for a "residual" sulfate
level of nearly 5 jug/m^ throughout many of the eastern states. Altshuller's analyses of atmo-
spheric data also indicated that the relationship between SC>2 and sulfate concentrations is non-
linear and, presumably, complex.
To help EPA develop effective control strategies for limiting the sulfate aerosol concentra-
tions, the predominant and hopefully controllable factors affecting the rate of SOn-to-sulfate
conversion must be known. This report presents pertinent details and results of an investigation
in which a smog-chamber approach is being taken to study SC>2 oxidation in simulated atmo-
spheres. The EPA is continuing to support research on this problem at Battelle and at other
laboratories as well.
* References appear on page 21.
BATTELLE — COLUMBUS
-------
2
BACKGROUND
The rate of SC>2 oxidation in the atmosphere, particularly over relatively short times, is
difficult to measure due to variety and variability of SC>2 emissions and inadequate tracers
thereof. On a more simplified yearly basis, Junge(^) estimates a mean lifetime for industrially
emitted SC>2 of 4-7 days, corresponding to a first-order-removal rate of about 0.6-1 percent/hr.
Using an air-trajectory technique and a diffusion model, Roberts and Friedlander(5) have
made some preliminary determinations of short-term SO2-oxidation rates in the polluted air
near Los Angeles; hourly average oxidation rates during afternoon hours ranged from 2-13 per-
cent/hr. In the study cited, there were some indications the oxidation rate was related to param-
eters such as 03, free radicals, olefins, and relative humidity.
A pilot study in New York City aimed at discerning any interrelationship between SC>2 con-
centrations and oxidant production indicated (in a limited number of tests) that the SC>2 conversion
rate was < 5 percent/hr when early morning air was irradiated naturally in Teflon (du Pont) bags
throughout the day.(6) Furthermore, the SC>2 loss rate seemed unaffected by ambient particulate
concentrations and was relatively constant throughout the daylight hours, with no obvious change
during 03 production.
Consistently good correlations of atmospheric sulfate concentrations with ammonium concen-
trations have led to the idea that NH3 and SC>2 combine to form an addition compound in the
presence of water. (4,7) Thus far, however, there is no kinetic data to support the addition-com-
pound theory for SC>2 oxidation, and this, as well as other theories of heterogeneous SC>2 oxidation
in water droplets, are questionable in that atmospheric data show an apparent dependence of the
sulfate concentration on the concentration of water vapor in the air rather than on the degree of water
saturation.(l) In short, although there seems to be general agreement that most of the SC>2 in the
atmosphere is ultimately oxidized to sulfate-type aerosols, no consistent evidence exists from field
studies to suggest explicit factors dominating the SC>2 oxidation rate.
The numerous laboratory studies likewise leave many unanswered questions regarding the
detailed mechanisms and rates of sulfate formation. It is commonly held that both heterogeneous
and homogeneous processes occur, and there is an abundance of data to indicate that the homo-
geneous rate of conversion of SC>2 to sulfate is enhanced in polluted atmospheres. A thorough
review of the laboratory work on this subject through 1970 has been presented by
Prior to 1970, many studies were conducted of SO2 photooxidation in the first strong absorp-
tion band of SO2- A large discrepancy existed regarding the importance of this process, as reported
quantum yields for SC>2 disappearance varied from 10"^ to 10~1 molecule per quantum absorbed.
Reinvestigating the reaction in 1972, Allen, et al.(9), reported an SC<2 quantum efficiency of about
6 x 10~3. In the same year, Cox(lO) reported a much lower 0SO? value of 3 x 10"^. In both
reports, there was some evidence of 0sOo dependency on SC>2 and/or C>2 concentrations, but such
dependency was readily dismissed as experimental artifact. If ^§07 is as l°w as 3 x 10"^, then,
indeed, SC«2 photooxidation is an insignificant mechanism of SC>2 removal, even on a global basis.
Combining 0sOo ~ 3 x 10"^ with the specific solar (Z = 40°) absorption rate for SC>2 of ~0.7
hr~l (11), a maximum photooxidation rate of about 0.02 percent/hr and an SO2 mean lifetime
> 200 days result.
At the time of Bufalini's review, numerous smog-chamber-type studies had demonstrated that
homogeneous oxidation of SO2 occurred at widely varying rates. In spite of all the macroscopic
BATTELLE — COLUMBUS
-------
evidence of enhanced oxidation, little mechanistic and kinetic information was available on inter-
mediates specifically responsible for the conversion. In the past few years, however, quantitative
rate data have been reported for HC^12), HO(13), and "zwitterion"(14) reactions with SO2-
Considering these as well as other pertinent reactions in computer simulations of photochemical smog,
Calvert05a,b) has estimated the homogeneous rate of 862 oxidation in smog to range between 1.5
and 4.7 percent/hr.* Thus, with the tools of computer simulations of reaction kinetics combined
with smog-chamber data of the disappearance and formation of reactants and products, there is
considerable promise that the homogeneous pathways of SC>2 oxidation in the atmosphere will
soon be sufficiently understood. That is the approach and the goal of this program. With sufficient
knowledge, it should then be possible to assess the significance of the homogeneous processes
occurring in the atmosphere and thereby consider the value of various sulfate control strategies.
SCOPE OF PROGRAM
The ultimate objective of this program is to provide EPA with smog-chamber data useful
in developing models of the conversion of SC>2 to sulfuric acid and other sulfur-bearing aerosols
in polluted atmospheres. This first year of effort was devoted to defining experimental conditions
and analytical techniques which would permit measuring the conversion using the smog-chamber
approach. Upon establishing experimental and analytical procedures, a factorial series of experi-
ments was conducted in which SO2 was irradiated in air containing variable concentrations of
propylene, NOX, and H2O vapor.
It was also observed that SC>2 oxidation was appreciable during irradiations in relatively
clean air. Thus, additional experiments were conducted in small and large smog chambers and
under different irradiation conditions to ascertain the conditions conducive to SC>2 oxidation in
unpolluted air.
OBSERVATIONS AND TENTATIVE CONCLUSIONS
The rate of SO2 conversion to aerosols ranged from 0.4 to 5.8 percent/hr over a wide
variety of environmental conditions approaching both clean and highly polluted air. There was
no evidence to suggest that the amount and/or nature of the experimental surfaces employed in
the study seriously affected the observed oxidation rates; nevertheless, the existence of a surface
effect was not ruled out.
In air contaminated with propylene, NOX, and SC>2, an interaction effect on SC>2 oxidation
was observed between relative humidity and the HC/NOX ratio. At 60 percent relative humidity,
high propylene/NOx ratios resulted in higher overall (4-hr avg) SC>2 oxidation rates as well as
enhanced NO oxidation. At relative humidities of 40 percent, doubling the HC/NOX ratios
resulted in little change in the overall SO2 oxidation rates in spite of a nearly two-fold increase
in the NO oxidation rate. It appears that the varying 862 oxidation rate observed in these
experiments is consistent with an assortment of free-radical processes whose quantitative signifi-
cance must await model development and testing.
* In reference 15a, SC>2 oxidation was simulated for initial reactant concentrations (ppm) of: (NO] = 0.075; [NC^] = 0.025;
[COJ0 = 10; [CH4]0 - 1.5; [C4Hg]0 = 0.10; [CH2O]O = 0.10; [CH3CHO]0 = 0.06; (see Table 2, page 18). In reference
15b, more recent kinetic data were used to predict an SC>2 oxidation rate of 1.5 percent/hr for the same initial reactant
conditions except: [NO]O = 0.15 ppm and [NO2JO = 0.05 ppm.
BATTELLE — COLUMBUS
-------
In relatively clean air, SC»2 oxidation was most probably caused by trace contamination, but
the accountable reactions are yet obscure. The mechanism is probably linked to NC>2 contamina-
tion in view of the coincident pattern of 63 and 863 formation. However, the possibility of SC>2
oxidation via photoexcitation of SC>2 cannot be ruled out on the basis of the results collected,
and some evidence, indeed, supports the involvement of long-lived excited states of SO?. Whatever
the mechanism, one should not overlook the importance of SC>2 oxidation under these conditions;
additional attention in future work should be given to factors affecting the "clean-air rate" as
well as the "polluted-air rate".
EXPERIMENTAL METHODS
Large Smog Chamber
Description
With the exception of a series of experiments investigating contamination effects on SC>2
oxidation, all irradiations were conducted in Battelle-Columbus' 17.3-m3 smog chamber having a
surface-to-volume ratio of 2.6m~l; the surface is polished aluminum and FEP Teflon (du Pont).
Direct irradiation through 5-mil Teflon windows is provided by a bank of 96 fluorescent black-
lamps and 15 fluorescent sunlamps. The photon flux of the blacklamps is distributed unimodally
in the uv region, with peak intensity at 370 nm; the sunlamp peak intensity occurs at 310 nm.
With all lamps operating, the average intensity in the 290-400 nm region is 3.8 x lO1^ photons/
cm2-sec, as determined by o-nitrobenzaldehyde actinometry. Using the method of Tuesday^6),
light intensity measurements by NO2 photolysis yield kj = 0.43 min'1. The sunlamps provide
<1 percent of the total energy. In the strong absorption band of SC>2, i.e., below 320 nm, the
sunlamp intensity corresponds to about 0.16 x 10 ^ photons/cm^-sec, compared with noonday
sunlight intensity in the SC>2 band of about 1.4 x 10^ photons/cm^-sec.U 1)
Background air supplied to the chamber is taken through a 10-m stack atop a three-story
building and is passed through a purification system which includes a permanganate filter bed,
a charcoal filter system, an absolute filter, and a humidification unit. After purification, back-
ground total hydrocarbon is generally 2-3 ppmC, with the majority being methane. Nonmethane
hydrocarbons are usually <0.5 ppmC (mostly paraffins), NOX <0.02 ppm, CO <4 ppm, and
particles < 10^ cm~3.
Chamber Operation
Prior to certain series of experiments, the chamber's surfaces were thoroughly cleaned by
washing with water and sometimes isopropanol-water mixtures. After cleaning, the chamber
was dried by continuous purging with purified air.
All irradiations were conducted for about 4 hours. Typically, the chamber was first
humidified with deionized, double-distilled water vapor followed by consecutive injections of
SO2, NO, NO2, propylene, and tracer (perfluoropropane).
Continuous and intermittent sampling of the chamber air together with a small unavoidable
leak rate resulted in overall chamber dilution rates of about 10 percent/hr. This sizeable rate is
particularly important in interpreting the apparent loss of SO2 where the SO2 reaction rate is
BATTELLE — COLUMBUS
-------
considerably smaller than the dilution rate. 10 determine the dilution rate, an inert tluorocarbon
(perfluoropropane) tracer was added to the chamber and monitored by gas chromatography.
Estimations of SC>2 loss due to reactions and sulfate aerosol production were made as follows.
For each experiment, the dilution data and the observed SC>2 concentration were fitted to first-
order kinetic equations. Simple subtraction of the two rate constants gave a good approximation
of the removal rate of SC>2 due to processes other than dilution.
In order to compare SC>2 loss with analytical results for H2SO4 aerosol during the course of
the experiments, it was necessary to predict aerosol concentrations, taking into consideration
both the actual SC>2 oxidation rate and the dilution rate. Assuming that the chamber is a per-
fectly stirred reactor and that the SC>2 to 863 reaction rates are first order in SC>2, the appro-
priate expression^ ') for predicting the 803 concentration at any time is given by
relationship (A),
S0 =S0e-l-e-l3l (A)
where SC>2- = initial SC>2 concentration, t = reaction time, kj = dilution rate, and k3 = SC>2
reaction rale. Predicted values of H2SO4 appearing in Table 1 were determined in this fashion.
Small Smog Chamber
Description
The small smog chamber used for investigating the effect of air contamination on SC>2 oxida-
tion has a volume of approximately 200 liters and consists of two 45-cm-diameter by 91-cm-high
Pyrex bell jars fastened together with an anodized aluminum ring. A vacuum-tight seal between
the bell jars and the aluminum ring is made with a Viton L-ring.
Illumination is normally provided by a circular bank of 34 fluorescent blacklamps. NC>2
photolysis yields kj = 0.45 min'l. por some experiments, three blacklamps were removed
and a row of incandescent sunlamps was substituted. The photon flux of the sunlamps, determined
by o-nitrobenzaldehyde actinometry, corresponds to an intensity of 1.5 x 10^ photons/cm^-sec.
The energy of these lamps is distributed bimodally, with peak intensities near 310 and 370 nm.
The relative intensity of the peaks favors the 370 nm band by a factor of about 2.5.
Chamber Operation
The advantages of the 200-liter chamber over the larger chamber is the practicality of
thoroughly cleaning the system and utilizing ultrapure air. For cleaning, the chamber was dis-
mantled and the Pyrex bell jars scrubbed with soap, rinsed, etched with dilute HF, and rinsed
again with large quantities of distilled water.
Several purity grades of air and oxygen were used both for purging the reaction system and
for the irradiation experiments. Further purification of the air was undertaken by exposing
BATTELLE — COLUMBUS
-------
it to ultraviolet radiation such that 03 (~1 ppm) was produced. After a residence period of
about 30 minutes, the ozonized air stream was passed consecutively through activated charcoal,
soda lime, and an absolute filter. Prior to entering the chamber, the air passed over an SO 2
permeation tube (Metronics, Inc.) maintained at constant temperature. The flow system oper-
ated continuously, thereby providing a constant SC>2 concentration in the chamber. The
permeation tube was simply bypassed for experiments without SO2-
The chamber thus was operated dynamically at one atmosphere with a constant air
throughput of 1.50 liters/mm, a rate in excess of the sampling demands during irradiation.
Because of the dynamic process, steady-state approximations were used in interpreting the data.
Three parameters — SC>2, Oo, and aerosol volume — were monitored during irradiation. In
general, near steady-state values for these parameters were established in 4-5 hours. An 63
instrument was tuned to a sensitivity of 1 ppb. The SC>2 instrument had a sensitivity of about
10 ppb and therefore was not very useful in measuring small changes in SO2- The electrical
aerosol analyzer (TSI Model 3030) was estimated to have a sensitivity corresponding to 803
concentrations of about 0.25 ppb.
Because of the greater sensitivity in measuring aerosol compared with SC>2, aerosol produc-
tion was used to compute the SO2 oxidation rates in spite of several assumptions necessary in
interpreting aerosol data. Thus, under steady-state conditions,
= kfCS02) - k£(S03) = 0 , (B)
where kf = velocity constant for SC>2 to 803 conversion and kg = velocity constant for 803
losses. Taking some typical steady-state data and rearranging relationship (B),
*= ao°95 hr'or °'95 percent/hr'
Calculation of the 803 concentration and the rate constant for 803 loss requires additional
explanation. First, the volume of aerosol equivalent to a particular concentration of 863 was
inferred from measurements of the concentration of aerosols over 10 size intervals using an
electrical aerosol analyzer whose operation and calibration have been reported elsewhereA 1 °)
In inferring a volume concentration from the integrated size spectrum, two assumptions are
made — one, that all the aerosol volume is in the size range measured, and two, that all the
aerosols are perfect spheres. In equating aerosol volume with sulfuric acid concentrations, it is
further assumed that equilibrium exists between the condensed and vapor phases of aqueous
sulfuric acid (SO3-xH2O). Acid mole-fraction and density data(l^) were used to convert
aerosol volume to 803 vapor concentrations at the respective relative humidities. In the exem-
plary data given above, the aerosol volume concentration was 26.8 /zm^/cm^ which, when ad-
justed for H2O constituency at a relative humidity of 1 percent, yielded 8.7 ppb (vol/vol) 803.
This method of analysis for 803 was checked by making simultaneous measurements of
H2SO4 aerosol with the aerosol analyzer and a chemical method. Microliter quantities of
98 percent sulfuric acid were evaporated in the 17.3-m-* smog chamber. Upon saturation,
nucleation of H2SO4 aerosol commenced and growth continued by condensation and coagulation.
After a fairly stable distribution of aerosol was established, filter samples were withdrawn for
chemical analyses (barium chloranalate method) and the results were compared with 863 concen-
trations calculated in the manner just described from aerosol measurements. The agreement
BATTELLE — COLUMBUS
-------
between the two methods was good — within 1 0 percent of each other. Thus, in instances
where aerosol is composed largely of sulfuric acid of appropriate size, the aerosol volume con-
centration is probably a fairly accurate measure of SO2 oxidation.
The second consideration to the k(SC>3) term in relationships (B) and (C) is the velocity
constant for aerosol loss. At steady state (where the rate of aerosol formation equals the rate
of disappearance), the rate of aerosol loss is given approximately by relationship (D):
-jj- = kcn2 + kpn + kwn , (D)
where kc is the second-order constant for coagulation, kp is the constant for the dynamic flow,
and kw is the constant for wall losses. In the present analysis we are interested only in the
change in aerosol mass, so the coagulation term can be neglected since mass is conserved in the
process.
The rate constant for the chamber throughput is simply the flow rate/reactor volume, which
is 1.25 x 10"4 sec'l for the conditions employed; perfect mixing is assumed.
The appropriate constant for the wall-loss term is far less precise. It is well accepted that
inertial deposition of submicron aerosols is nearly zero when the aerosol cloud is not stirred.
Therefore, we can assume that wall deposition is purely by diffusion — aerosols are carried to
the wall by convective diffusion and deposited thereon by molecular diffusion through a thin
boundary layer. Under this condition, the rate of deposition can be approximated by relation-
ship (E):
-dn /SD
S
where y is the surface to volume ratio of the reactor, D is the aerosol diffusivity, and 6 is the
thickness of the boundary layer. (20) Data on 5 compiled by Fuchs indicate that 20 //m is a
useful estimate for aerosols about 0. 1 /urn diameter. Using that value and the appropriate values
of S, V, and D, we have estimated the wall loss constant (kw) for various particle sizes relative to
the constant for flow rate (kp). The ratios of (kw + kp)/kp indicated below are probably maxima
for particle diameters <0. 1 /zm, because 5 was kept constant in computing kw and, in reality,
it increases with increasing coefficient of diffusivity (decreasing particle size):
Particle Diameter,
(kw + kp)/kp
0.001 14
0.005 1.54
0.01 1.14
0.05 1.01
0.10 1.00
It is obvious from this exercise that wall losses relative to throughput losses will be substantial
for aerosols <0.01 p.m diameter. During the early period of irradiation where embryonic clusters
are formed by nucleation processes, wall losses are no doubt very serious. However, as irradiation
proceeds, aerosol growth continues and a much different size distribution of aerosol exists as
BATTELLE — COLUMBUS
-------
8
steady-state conditions are approached. A representative distribution of particles at steady state
is indicated in Figure 1. Sulfuric acid aerosol generated at this time is condensing almost exclu-
sively on existing aerosols, as reflected by the aerosol volume distribution in Figure 1. Thus,
10'
a.
O
en
O
Volume distribution
10
0.001
10'
E
a.
8
10' o
0.01
O.I
10
Particle Diameter (Dp),
FIGURE 1. SIZE DISTRIBUTION OF SO3 • x^O AEROSOL WHERE AN
SO2-AIR MIXTURE IS IRRADIATED 4 HOURS IN A DYNAMIC
REACTOR
after several hours of irradiation it appears reasonable to represent the total loss of sulfur acid aer-
osol in these experiments by the flow rate. This hypothesis was substantiated experimentally by
turning off the lamps and observing a decay rate of aerosol volume in accord with the chamber
flow rate. This being the case, the steady-state relationship (Relation B) used to compute the SC>2
oxidation rate is justified.
Methods of Chemical Analysis
The following analytical methods were used in monitoring gaseous reactants and products:
BATTELLE — COLUMBUS
-------
Compound Method
Total hydrocarbons Flame ionization (Beckman 109 THC monitor)
Propylene and perfluoropropane Gas chromatography-flame ionization
(Varian 2800 GC)
NO and NC>2 Automatic Saltzman (Beckman Air Quality
(Acralyzer), dichromate oxidizer for NO
SO2 Coulometry (Beckman 906) and
flame photometry (Tracer, Inc.)
03 Chemiluminescence (Battelle-built)
H2O Dew pt. (Cambridge Systems, Inc., hygrometer)
Sulfuric acid aerosol was determined in two ways. The least specific but most sensitive method
involved the TSI aerosol analyzer discussed earlier. Sulfuric acid was also determined using the
barium chloranilate method.(21,22) Aerosol from the smog chamber was collected on 25-mm
Nuclepore filters or on 25-mm fiber-glass filters which had been prewashed by refluxing with water
and methanol. It appeared that some improvement in collection efficiency was obtained using
the glass filters. Neither type was found to suffer any interference due to SO2 absorption.
The H2SO4 aerosol was extracted with an 80 percent isopropanol/20 percent water solution
to which excess barium chloranilate (barium salt of 2,5-dichloro-3,6-dihydroxy-p-benzoquinone)
was added and the precipitate centrifuged after frequent stirrings. The reaction occurring is
BaC6Cl2O4 + H+ + SO| -> BaSO4 + HC6C12O4 (1)
The acid chloranilate anions give a pink color, which is read at 310 nm with a spectrophotometer.
At the conclusion of one experiment (No. 13) a large aerosol collection was made by evacua-
ting the chamber contents through a 10-cm quartz-fiber filter, which was analyzed for total
sulfate. The filter was leached in 100 ml of hot water for 16 hours, filtered, and adjusted with
HC1O4 to a pH of 3.0. A 10-ml aliquot was taken and diluted with 40 ml of isopropanol. Two
drops of thorin were added and the SO4 titrated with 0.01 N barium perchlorate from a micro-
buret; a yellow-to-pink color change marked the end point. The total sulfate analysis indicated
only 24 ppb 803, compared with 45 ppb by the barium chloranilate method, 51 ppb by gravi-
metric analysis of the same filter sample (assuming 100 percent H2SO4 aerosol), and a predicted
concentration of 42 ppb based on SO2 depletion. There was no apparent reason for the dis-
crepancy between the two sulfate techniques. The coincidence of the barium chloranilate and
gravimetric results, together with the good agreement between the barium chloranilate and the
electrical aerosol analyzer results with pure H2SO4, led to the establishment of the chloranilate
method for routine analysis.
RESULTS AND DISCUSSION
A summary of the pertinent reaction conditions and experimental results of this program
is given in Table 1. Before discussing the results, it may be helpful to expand on the meaning of
BATTELLE — COLUMBUS
-------
TABLE 1. SUMMARY OF EXPERIMENTS ON SOg OXIDATION IN SIMULATED SMOG
H
m
r
r
m
n
o
r
C
2
D
C
to
Product Concentrations/Formation Rates
Initial Concentrations
Run
Run Description'8'
Section A: SO? wall losses
1 Clean surfaces/dark
2,3(8' Conditioned surfaces/dark
Section B: SO2 oxidation in
12
13
9
16
17
7
14
209
202
203
19
22
28
21
26
23
21
20
25
polluted air - preliminary runs
High S02
Polluted air X2/hlgh SO2
Low SO2
Section C : SO2 oxidation in
unpolluted air
Sunlamp irradiation
Blacklamp irradiation
Preirradiated air
Ambient air
Section D: Contaminant
investigation of SOr? oxidation
Zero alr/200-liter chamberf")
Zero air + SO2/200-liter
chamber/sunlamps
Zero air + SC>2/200-llter
chamber/blacklamps
Section E: SO2 oxidation in
polluted air - model smog
High C3H6/high RH/no SO2/
blacklamps
Low CgHg/low RH/blacklamps
Low C3H6/low RH/blacklamps
Low CjjHjj/hlgh RH/blacklamps
Low C3H6/hlgh RH/blacklamps
High C3H6/low RH/blacklamps
High C3H6/low RH/blacklamps
High CjHg/high RH/blacklamps
High C3H6/high RH/blacklamps
S02.
ppb
460
440
480
490
120
490
490
520
540
0
410
410
0
530
498
440
450
478
460
410
480
NO.
ppb
--
490
970
480
0
0
<10
20
0
0
0
550
480
480
480
480
490
480
460
480
C3H6. RH.
ppbC percent
--
3.310
5.370
2.640
0
0
—
"
0
0
0
4.550
1.590
1.570
1,580
1,440
3,120
2,890
3,130
2,860
30 \
60
61
56
59
28
32
30
60
<2
.
ppb-min"
--
10
12
10
—
--
--
--
—
—
12
4
4
8
9
10
10
12
15
°3
Max,
1 PPb
--
430
400
360
<5
40
40
40
3-10
9
11
630
230
170
200
220
490
350
370
340
Dilution
.
hr'1
0.128
0.084
0.090
0.108
0.079
0.114
0.082
0.082
0.090
—
--
0.067
0.112
0.107
0.116
0.138
0.100
0.123
0.138
0.117
Obs. SO2 Loss
<1<2>.
hr"1
0.133
0.095
0.129
0.140
0.084
0.128
0.108
0.124
0.134
--
--
--
0.133
0.132
0.144
0.169
0.123
0.138
0.183
0.175
Corr. SO2 Loss SOJ soj
(k3=k2-k1). Predicted Maxfb). Max(c),
hr1 ppb ppb
0.005
0.011
0.039
0.032
0.005
0.014
0.026
0.042
0.044
o. oogO
0. 004<')
"
0.021
0.025
0.028
0.031
0.023
0.015
0.041
0.058
_5<0
13
48
42
2
17
35
57
60
--
--
0
27
30
29
30
29
16
39
61
"
9
45
<5
15
26
49
51
..
--
--
—
<5
<5
<5
<5
5
<5
25
25
Aerosol
Vol. Max.
Mm3-cm~3
--
—
--
••
13
390
--
"
<1
27
12
25
210
173
186
180
380
280
260
290
Aerosol
SOj Max(e),
ppb
"
—
--
--
2
68
--
--
0
33
15
J
37
30
21
20
66
49
28
32
(a) All_experiments conducted under fluorescent-sunlamp and blacklamp irradiation in a 17.4-m smog chamber unless otherwise specified.
(b) $04 concentration predicted on the basis of the rate of'decay of 50%, first-order conversion to SO^. and 4-hour irradiation time.
(c) Barium chloranilate procedure.
(d) Total aerosol volume concentration inferred from aerosol size distributions determined by Thermo Systems Inc. aerosol analyzer.
(e) SOJ concentration assuming total aerosol volume is sulfuric acid, corrected for equilibrium water content.
(f) Underlined values correspond to hypothetical situations.
(g) Average of two very similar experimental conditions and results.
(h) Average of six experiments utilizing "zero air" with and without additional purification.
(i) Calculated from steady-state aerosol concentration assuming all aerosol is sulfuric acid from SO2 oxidation.
-------
11
subheadings appearing in the table. In addition to the abbreviated descriptions in the first column,
experimental details of each run are presented in the ensuing discussion. Presumably, the columns
of initial concentrations are straightforward. Continuing from left to right, the next column
represents the maximum rate of NO to NC>2 oxidation, and the 03 maximum is either the maxi-
mum 03 concentration or the final 63 concentration in instances where 63 was still increasing
at the end of irradiation (4 hours). The dilution rate (kj) is the average rate at which the cham-
ber contents were replenished with clean air. The observed SC>2 loss rate (k2) is based on the
measured SC>2 concentration, and the corrected SC>2 loss rate (k3) is the measured 862 loss
rate corrected for the dilution rate. The predicted 864 concentration is the SO^ concentration
calculated at 4 hours of irradiation from the corrected SC>2 oxidation rate and the chamber
dilution rate. The next sulfate column is the sulfuric acid concentration at 4 hours of irradiation,
as determined by wet-chemical analysis. The aerosol-volume column is the total aerosol concen-
tration (by volume) at 4 hours as determined by the electrical aerosol analyzer, and the final
column corresponds to the sulfuric acid concentration calculated from the total aerosol volume
and 803 • H2O equilibria data. Details regarding these parameters were presented in the Experi-
mental Section.
To contribute to the EPA effort on modeling SO2 oxidation in smog, all the continuous
and intermittent chemical data regarding Experiments 20-28 were transcribed onto magnetic
tape, which in turn was delivered to the Project Officer. Results on the modeling aspects of the
program will be reported separately in a future report.
SO2 Wall Losses
At the outset of this program, it was necessary to determine the extent to which SO2 was
removed from the experimental system due to interactions with smog-chamber surfaces. The
rate of SO2 removal was measured while SO2 and clean air were contained in the unirradiated
chamber for 4 hours. Measurements were made under two conditions: (1) with chamber surfaces
which had been washed with water and dried by purging with clean air and (2) with chamber
surfaces which had been "conditioned" by irradiating propylene-NOx-SO2 mixtures in air. As
indicated in Section A of Table 1, the SO2 removal rate, corrected for dilution, was 0.5 percent/
hr with washed surfaces and about 1.1 percent/hr for conditioned surfaces. It is not certain
whether this difference is due to an actual surface effect or is just representative of the uncer-
tainty in measuring the SO2 depletion rate.
If one had assumed that the losses of SO2 in these cases were due to gas-phase oxidation
rather than wall-removal processes, one would expect to find 5 ppb and 13 ppb H2SO4 aerosol,
respectively, at the end of 4 hours, having corrected for dilution in the manner noted above.
As we shall see, this SO2 removal rate is substantial compared with the rate of SO2 removal
attributed to SO2 oxidation under various irradiation conditions, but the wall-loss rate was
not taken into consideration in computing the "corrected SO2 loss rate" (Column 9, Table 1)
because it was felt to be too uncertain to assign a constant rate to a process which might vary
with chamber history or reaction conditions. Thus, in considering the "corrected SO2 loss
rates" in subsequent discussions, it should be kept in mind that the rates are maximums,
corrected only for dilution losses, and may be high by perhaps 1 percent/hr.
BATTELLE — COLUMBUS
-------
12
SO2 Oxidation in Polluted Air -
Preliminary Experiments
In addition to the surface-loss determinations, preliminary experiments were conducted with
propylene-NOx-SO2-air mixtures to determine the sensitivity of the SC>2 oxidation rate to pollu-
tant concentrations. Experimental conditions and results are summarized under Section B in
Table 1. In the experiment in which 480 ppb SC>2 was irradiated with about 490 ppb NOX and
3300 ppbC propylene, the rate of SC>2 oxidation based on SC>2 removal was 3.9 percent/hr. In
this case, the chemical analysis for sulfate aerosol was much lower than the amount computed
from SC>2 decay and was likely to be incorrect. In the next experiment at the same SO2 con-
centration but nearly double the pollutant (propylene-NOx) concentration, the SC>2 oxidation
rate was 3.2 percent/hr, slightly less than the preceding rate. In the third experiment, in
which the initial SC>2 concentration was lowered to 120 ppb, the fraction of SC>2 lost and the
amount of sulfate formed were too small to permit reliable calculations of the oxidation rate.
Tentative conclusions from this preliminary work were as follows:
(1) The effects of varying pollutant concentrations on SC>2 oxidation appeared
small.
(2) The SC>2 oxidation rate could not be obtained reliably at SC>2 concentrations
near 100 ppb in the propylene-NOx-system.
(3) The accuracy in determining the SC>2 oxidation rates would be limited to
state-of-the-art techniques in SC>2 and 863 analyses.
SO2 Oxidation in Unpolluted Air
As part of the preliminary work, it was observed that SO2 was apparently oxidized in clean
(i.e., unpolluted) air at rates similar to those observed in air intentionally polluted with propylene
and NOX. For example, with full irradiation (sunlamps plus blacklamps), the rate of SO2 removal
was 3.9 percent/hr in a mixture of propylene and NOX in air (Run 13) and 4.2 percent/hr in air
without these pollutants (Run 7). This disconcerting finding led to some additional investigations
with unpolluted air, which are summarized here. Investigations were also conducted with ultra-
pure air in a 200-liter chamber, and those results are discussed shortly.
In Run No. 16 (Section C, Table 1), 490 ppb SO2 was irradiated in otherwise unpolluted
air* — i.e., air which is typically present in the smog chamber after passing through a purification
train. In this case, irradiation was provided only by fluorescent sunlamps. As discussed earlier,
the sunlamps provide only about 1 percent of the ultraviolet (290-400 nm) energy of the cham-
ber's blacklamps but about 15 percent of the energy in the strong absorption band of SO2-
Under these irradiation conditions, the SO2 loss rate was 1.4 percent/hr, or only slightly above
the rate that might be attributed to wall absorption of SO2- Chemical analysis did indicate the
presence of sulfate, but the aerosol volume determinations with the mobility analyzer indicated
only a trace amount of aerosol. Due to the weak irradiation intensity of the sunlamps, the
results of this experiment are inconclusive with respect to the importance of 862 excitation to
SO2 oxidation.
* Typical contamination levels of this air are 1.5-2.5 ppm CH4, <0.5 ppmC nonmethane hydrocarbons (mostly paraffins),
<0.02 ppm NOX, 2-4 ppm CO, and <1000 paiticles/cm3.
BATTELLE — COLUMBUS
-------
13
A similar experiment (No. 17) was conducted with blackiarnp irradiation. The SO2 loss rate
was 2.6 percent/hr, and the chemical analysis for sulfate was near the sulfate level predicted by
SO2 removal.
The third experiment (No. 7) in this section was similar to the previous one, except that
SC>2 was admitted to the chamber after the air had been "photochemically exhausted" by
irradiating it for 65 hours. During the first few hours of the preliminary irradiation, about 25 ppb
03 appeared, and within the next few hours it diminished to some steady-state concentration
below the detection limit (<10 ppb). Upon introducing SC>2 at the 65th hour, 03 increased
steadily to a maximum value of 40 ppb as in the previous run. The SC>2 decay rate during this
period was 4.2 percent/hr, and the measured sulfate concentration was in fair agreement with
the predicted concentration.
In the fourth experiment described in Section C (No. 14), 540 ppb SC>2 was irradiated in
ambient air. Unfiltered air from outside the laboratory was pulled into the chamber early in the
morning. The air that morning was unusually clean for Columbus, containing 20 ppb NO,
0.4 ppm nonmethane hydrocarbon, 4 ppm CO, 2 x 10^ particles/cm^ and having a light-
scattering (b scat) value of 0.8 x 10'^m"'. Irradiation of this mixture also produced 40 ppb 03,
and SO2 was removed at the rate of 4.4 percent/hr.
In summary, we observed an SO2 oxidation rate of about 3 percent/hr (corrected for a
presumed SO2 wall-loss rate of about 1 percent/hr) upon blacklamp irradiation of relatively
clean air — i.e., both laboratory and ambient air. This finding is highly important if SO2 ex-
citation and/or trace contamination of the air is responsible for the observed rate. On the other
hand, if the observed oxidation rate is a manifestation of some surface condition, it may or may
not be important, depending on the nature of the surface effect. These possibilities are con-
sidered further in light of limited data.
A potentially influential factor in these experiments is the condition of the smog-chamber's
surfaces. It is conceivable that hydrocarbons, nitrites, aldehydes and organic acids contaminate
the chamber surface, and some of these compounds could generate reactive species by various
photodissociation processes. For example,
H2CO + hv -» HCO + H ' (2)
RCHO + hv -* R + HCO (3)
MONO + hv -+HO + NO (4)
RONO + hv -> RO + NO . (5)
Accumulation of condensed matter on the chamber surface does not seem to be crucial,
however. Experiment No. 16 was conducted immediately after the chamber's surface was cleaned
by scrubbing with water, and the SO2 oxidation rate was not much different from that observed
in all the other experiments in which the surfaces had been exposed to repeated smog irradiations.
Thus, it appears that surface contamination had little effect on the observed SO2 oxidation rates.
BATTELLE — COLUMBUS
-------
14
Characterization of trace contamination of air (contamination levels lower than those
commonly found in the atmosphere) is problematic in that it is not practical to conduct analysis
for contaminants when one is not sure what species to look for. Furthermore, it is not practical
to extensively purify the volume of air needed to purge and occupy a large smog chamber. To
further investigate the possibility of contaminant effects, irradiations were conducted in a small
(200-liter) chamber where extensive air purification was possible. Those results, discussed in the
succeeding section, indicate that trace contamination may be a factor in SC>2 oxidation even when
extensive air purification was performed, but they do not conclusively identify the contaminant(s).
In the large chamber, where air purification is incomplete, the most likely contaminant
which would initiate SO2 oxidation at the outset of irradiation is nitrous acid (MONO). Nitrous
acid would form at low concentrations in equilibrium with H2O and traces of NO and NC>2, and
it photolyzes (Reaction 4) to yield HO radicals which add to SO2- Although the course of the
HOSO2 radical in subsequent smog reactions is uncertain, it has been proposed that the specie
most likely adds to molecular oxygen (giving HOSO2O2) which in turn oxidizes NO to NO2
in an analogous fashion with HO2 and RO2.(15b,23) Thus, under some circumstances, inclusion
of SO2 in smog reactions might enhance rather than suppress ozone formation by increasing
the chain length of NO to NO2 conversion. Evidence of such participation is seen in the prolonged
experiment with "photochemically exhausted" air (Run No. 7). Presumably, after 65 hours of
irradiation, the chain length for NO to NO2 conversion was quite short because most of the
species capable of oxidizing NO have been consumed over this period. Upon introducing SO2
after 65 hours, the concentration of ozone gradually increased in accord with the chemistry
discussed.
Contaminant Experiments
The preceding results prompted additional investigations in a 200-liter Pyrex chamber where
surfaces could be readily cleaned and ultrapure air would be used routinely. Fine tuning of a
chemiluminescent monitor made possible the detection of ozone with a sensitivity of 1 ppb.
SO2 was monitored with a Beckman 906 Analyzer and aerosol was monitored with a TS1 Model
3030 Electrical Analyzer. Details on the operation of the chamber and the methods of inter-
preting the data were described earlier.
The original plan in these experiments was to begin with a system clean enough to produce
no detectable smog manifestations when irradiated and then to add contaminants to either the
surface or the air to produce SO2 oxidation effects. Six experiments were conducted with
chamber surfaces which had been acid etched (HF), rinsed with deionized double distilled water,
and subsequently exposed only to clean air. Air ranging in quality from ordinary breathing air
to ultrapure air (Air Products UPC grade: Qfy <0.1 ppm, CO < 1 ppm) was irradiated alone
and, in most cases, upon further purification by ultraviolet exposure, charcoal, soda lime, and
absolute filters. Under all conditions, both sunlamp and blacklamp irradiation produced ozone
(3 to 10 ppb) after several hours of exposure. The buildup of ozone was always gradual, indic-
ative of chain-reaction processes. For brevity, the series of experiments with various grades of
"zero air" is represented under Section D of Table 1 as a single experiment (Run No. 209),
and the range of ozone observed is indicated. In every case, the aerosol volume concentration
was less than the detection level estimated at 1
I AT T E L L E — COLUMBUS
-------
15
Experiments No. 202 and 203 were conducted with 410 ppb SC>2 in zero-grade air. As in
the experiments without SC>2, both sunlamp and blacklamp irradiation were used. For the
200-liter chamber, blacklamp irradiation intensity is about 20 times as great as the sunlamp
intensity in the 290-400 nm region, but in the strong absorption band of SC>2 the sunlamp inten-
sity is slightly greater than the blacklamp.
With blacklamp irradiation (Run No. 203), the rate of SC>2 oxidation inferred from aerosol
determinations was 0.4 percent/hr, and 11 ppb ozone was present at steady state. With sunlamp
irradiation (Run No. 202), the SC>2 oxidation rate was nearly 0.9 percent/hr, and the ozone con-
centration at steady state was 9 ppb. These results, coupled with the relative energy distributions
of the lamps, strongly suggest a shorter wavelength dependence on the SC>2 oxidation process.
This dependence is not likely if NC>2 were the only absorbing specie important to SC>2 conver-
sion. If photoinduced SC>2 excitation is involved, crude extrapolation of these data to conditions
of atmospheric sunlight would increase the observed oxidation rate by a factor of at least 2, or
to nearly 2 percent/hr.
In spite of qualitative evidence supporting the participation of excited SC>2, the appearance
of ozone in the experiments without SC>2 proves that the air is contaminated and precludes any
unambiguous interpretation of the results.
SC>2 Oxidation in Polluted Air — Model Smog
To establish an initial data base for work on modeling SC>2 conversion in smog, eight irradia-
tion experiments were conducted using propylene-NONO2-SO2-H2O-air mixtures. Originally,
replicate experiments were planned at constant NOX concentrations and two concentrations of
propylene, SC>2, and H2O. However, SC>2 was hence excluded as a variable in view of preliminary
experiments showing uninterpretable results at low concentrations of SO2-
For convenience, the eight experiments are arranged in Table 1, Section E, as pairs, but the
sequence of conducting the experiments was selected at random to avoid any bias that might
be inherent in day-to-day operation. Also, unlike the preliminary experiments, irradiation in
this series was provided only by fluorescent blacklamps to minimize the lamp intensity in the
strong absorption band of SC>2 and thereby minimize the possibility of reactions attributable
to SC>2 excitation.
Experiment 19 (Table 1, Section E) without SC>2 was conducted to determine the aerosol
yield attributable to propylene-NOx-H2O reactions. This was of interest because aerosol pro-
duced in similar experiments with added SC>2 was interpreted as being derived solely from SO2-
As indicated in Table 1, a small volume of aerosol was produced in this case relative to those
in which SC>2 was present. And, because the propylene and NOX concentrations were higher
in Experiment 19, the contribution of aerosol from propylene-NOx reactions was probably
even smaller in Experiments 20-28.
In the model experiments at the lower propylene-to-NOx ratio of about 1/1, a relative
humidity change from 40 to 60 percent nearly doubled the maximum rate of NO oxidation.
The higher humidity seemed to have a small positive effect on the overall SO2 oxidation rate
calculated from SO2 depletion. In all four experiments the analysis for sulfuric acid aerosol
was below the detection limit (5 ppb); presumably, the sulfate aerosol was in some other form.
Subsequent chemical analyses indicated that the sulfate was probably combined as
BATTELLE — COLUMBUS
-------
16
At the higher propylene-to-NOx ratio of 2/1, the NO oxidation rate increased only about
40 percent in changing the relative humidity from 40 to 60 percent. (As expected in experiments
at comparable humidities, the higher propylene-to-NOx ratio resulted in higher NO oxidation rates
and ozone yields.) The higher humidities at propylene/NOx ratios of 2/1 increased the SO2
oxidation rate in terms of the overall SO2 loss to a greater extent than that observed in the
experiments at lower HC/NOX ratios. Thus, it appears that the SO2 oxidation rate is affected by
relative humidity and HC/NOX ratios by varying degrees and perhaps in some interrelated ways.
Reactions likely to be involved in the oxidations are discussed in the next section of the report.
A few comments are in order regarding the reproducibility of the experiments. It would
appear that, at least in the instances of measuring NO oxidation, SO2 and propylene consump-
tion, and ozone formation, that reproducibility varies from good (± 10%) to fair (± 30%).
Obviously, the data are too limited to treat statistically, and the value of having replicate data
appears to be borne out here.
In assessing the reproducibility of the chemistry associated with these results, one must keep
in mind the variability associated with the initial reaction conditions and also that which is linked
to the inconsistent dilution rates. For example, the final concentration of ozone (and aerosol as
well) is considerably higher in Run 23 compared with Run 27, in spite of similar NO oxidation
rates. However, the dilution rate in Run 27 is greater by some 23 percent, which would account
for much of the difference in the uncorrected ozone concentrations after 4 hours of irradiation.
Mechanisms of SO2 Oxidation
It is not an objective of this study to provide direct experimental evidence in support of
particular mechanisms of SO2 oxidation. However, in interpreting results presented here, it is
useful to discuss some of the more plausible reaction schemes. First, we will consider the reac-
tion system in which SO2 was irradiated in air containing propylene and NOX.
Polluted Air
The pattern of aerosol growth (and, presumably, SO2 oxidation) in the propylene-NOx-
SO2 system is indicated in Figures 2 and 3. It is evident that there is a rather short induction
period to aerosol growth, followed by a maximum rate of volumetric growth. The "S shape"
of the aerosol formation curve is more pronounced in Figure 3, where the propylene/NOx ratio
is twice as great as for the experiment appearing in Figure 2; i.e., SO2 oxidation is more gradual
and extended at lower HC/NOX ratios. In both cases, onset of the maximum in the rate of SO2
conversion occurs near the appearance time of ozone when there are nearly equal quantities of
NO and NO2- These general trends in oxidation are consistent with an assortment of free-radical
reactions with SO2, each of which is likely to make a significant contribution to the overall rate.
The reactions thought to be most significant are listed in Table 2. Following each reaction is a
corresponding SO2 conversion rate computed by CalvertH^a) for a computer simulated smog
system. Although the smog simulation is considerably different from the propylene-NOx system,
it is noteworthy that the sum of the SO2 oxidation rates of 2-3 percent/hi is in the range ob-
served for oxidation in the propylene-NOx system.
IATTELLE — COLUMBUS
-------
17
600
500
- 400
a 300
200
100
Run No. 21
^Propylene
3OO
200
E
a.
2 3
Irradiation Time, hours
FIGURE 2. SMOG PROFILE OF AEROSOL FORMATION
DURING IRRADIATION OF THE C3H6-NO-
NO2-SO2-H2O MODEL SMOG SYSTEM
Run No. 20
C3H6, NO, SO2
Run No. 17
SO,
X
£
3.
Irradiation Time, hr
FIGURE 3. PROFILES OF AEROSOL FORMATION FROM
SO2 OXIDATION IN UNPOLLUTED AIR AND
IN AIR CONTAINING C3H6 AND NOX
BATTELLE — COLUMBUS
-------
18
As indicated in Table 2, Reactions 6-8 add an oxygen atom to SO2, while Reactions 9 and 10
add radicals to SC>2; the products of these reactions convert to sulfuric acid and organic sulfates
(esters of sulfuric acid) in subsequent steps. The radical-addition reactions have been less fre-
quently considered than Reactions 6-8 probably because they have not been directly observed
in the laboratory. Reaction 9, however, has been studied recently by Wood et alX^)) ancj
they report a rate of about 5 x 102 ppirf1 min'1. Calvert's estimate of the rate for Reaction 10
is based on rate data for the somewhat analogous reaction of methyl radicals with SC>2, and
similarly, estimates for the rate of Reaction 8 are based on data for Reaction 7.
TABLE 2. ESTIMATED RATES OF HOMOGENEOUS SO2 REACTIONS IN
SIMULATED SMOG(!5)
(6)
(7)
(8)
(9)
(10)
Reaction
•CH2OO- + SO2->CH2O+ SOg
HO2 + SO2->- HO + SOg
RO + SO2-» RO + SOg
HO + SO2-* HOSO2-
RO + SO0->-ROSO •
2 &
SO2 Conversion Rate,
0.4
0.9
0.2
0.2
0.5
Total 2.2
percent/hr(a)
(a) Theoretical rates of SO2 removal occurring in a simulated atmosphere con-
taining the following initial concentrations (ppm): NO2 = 0.025, NO =
0. 075, trans-2-butene = 0.10, CH2O = 0.10, CHgCHO = 0. 06, CO = 10,
and CH. = 1. 5. The indicated rates correspond to 30 minutes of irradiation
in sunlight (z = 40").
Considerable kinetic data are available on Reactions 6 and 7. Reaction 6 has been investi-
gated by Cox and PenkettX14) in their studies of olefin-C>3-SO2 reactions in the dark. The inter-
mediate responsible for SC>2 oxidation in this case may be an ozonide of the respective olefin or
a "zwitterion", which is shown in Reaction 6 as a diradical. The rate of 0.4 percent/hr appear-
ing in Table 2 is their estimate where propylene is the hydrocarbon precursor of the zwitterion
rather than 2-butene, which was employed in the simulated smog system of Calvert. Reaction 7
has recently been reinvestigated by Davis et al.(12); ancj ^ey report a rate constant of 0.45
ppm'l min'l.
A great deal of uncertainty associated with predicting the extent to which these and other
reactions contribute to the overall SC>2 oxidation rate obviously lies in establishing the free-
radical concentrations in smog simulations. It is anticipated that computer simulations of the
propylene-NOx-SO2 system in the near future will provide additional information as to the rela-
tive importance of these reactions. Until that time, speculation on the quantitative aspects of
these reactions is unwarranted.
BATTELLE — COLUMBUS
-------
19
Unpolluted Air
The most perplexing results of this program are those involving SC>2 conversion in air which
was relatively clean compared with polluted air. If we assume as before that the rate of aerosol
growth in irradiated SO2-H2O-air systems reflects the rate of SC>2 oxidation, one sees in Figure 3
that the maximum aerosol growth rate occurs immediately upon irradiation, with the rate gradually
diminishing as the irradiation proceeds. This pattern is obviously different from that of polluted
air and may possibly be due to different and/or competitive reaction meachanisms in the two
cases.
As discussed earlier, SC>2 oxidation observed in relatively clean air is most likely attributable
to trace contaminants in such proportions as to generate higher-than-expected concentrations of
reactive intermediates which oxidize SC>2 via mechanism just outlined. At least two alternative
reaction schemes can be presented for SC>2 oxidation under such conditions, albeit neither scheme
is consistent with all the facts, and considerable skepticism exists among chemists regarding the
possible significance of either scheme.
The first argument to be considered is a familar one involving triplet SC>2 (3SC>2) formation,
followed by oxygen addition to form some excited state of 864 (804*) and disproportionation
to 803 and O-atoms(24)t,
S02 + hi>-» lS02 (11)
1SO2 + M ->• SC>2 + M deactivation (12)
lSC>2 + M -> 3SC>2 spin inversion (13)
3S02 + 02 -+ S04* d4)
SO4*^S03 + 0 (15)
followed by ozone and sulfuric acid formation
SO3 + H20 -» H2S04 (16)
O + O2 + M-»C>3 + M . (17)
A recent study of SC>2 photochemistry which included a high-pressure mechanism of 3SC>2
formation leads to a theoretical maximum rate of SC>2 photooxidation in the lower atmosphere
of 1.9%/hr.(26) In this study, the concurrence of 863 and 63 formation lend support to the
above mechanism. However, no enhancement of the SC>2 oxidation rate was observed in the 200-
liter chamber when air was replaced by pure O2- If chemical rather than physical quenching of
3SC>2 by 62 occurs, as the mechanism implies, then both 803 and 03 would be expected to
have increased in pure C>2, and this was not observed.
Alternatively, 864* might react with C>2 to generate SOj and Oj directly as originally suggested by BlacetA") However,
this simple reaction scheme involves a different transition state and evidence has been presented to discount its probable
.(24)
occurrences
BATTELLE — COLUMBUS
-------
20
A second and perhaps more obscure mechanism which can be postulated involves participa-
tion of nonemitting excited states of SC>2 (SO2*)- These presumably long-lived states have been
invoked to account for several photochemical reactions of SC>2 which occur under conditions
which preclude as reactants the easily quenched emitting singlet and triplet states of SO2-(27)
An abbreviated mechanism of SC>2 oxidation involving such species follows.
SC>2 + hv (290-320 nm) -»• l SC>2 (11)
!SO2 + M-> SO2* + M (18)
SO2* + SO2 -»• 803 + SO (19)
803 + SO ->• 2SO2 (20)
SO3 + H2O-> H2SO4 . (16)
In this scheme, H2O and SO might compete for 803. Thus, in the absence of H2O and/or other
sinks for SO and 803, recombination to SO2 would result in apparently low 503 quantum yields.
Reaction 20 was recently invoked(^8) to rationalize the divergent quantum yields (^SO^) °f
previous investigations. (9>10>29)
There are other possibilities for SO reactions. In air, SO might react to form 803:
SO + O2 + M-» 803 +M (21)
and on surfaces it is known to form SO2 and 82 by an obscure mechanism.
In the proposed scheme (Reactions 18-19), $803 i§ dependent on the SO2 concentration.
In the study of Allen, et al.(9); a small positive correlation of 0SO? witn SO2 was observed but
thought to be an artifact of the experimental procedure. CoxOO) observed a definite dependency
of 0SO3 on tne 3^2 pressure, accounting for it by Reaction 22 and subsequently Reaction 21,
3SO2 + SO2 -» SO3 + SO (22)
and rightfully dismissing their importance under atmospheric conditions on the basis that ambient
SO2 concentrations cannot compete with O2 and N2 in quenching 3SO2- Thus, a requirement of
the SO2* argument is that the nonemitting specie be long-lived; i.e., unlike 3SO2, it must survive
numerous collisions with N2 and O2- Indeed, an important observation in earlier work(2 /) was
that long-lived states of SO2* could survive enough collisions with "inert" reacting partners to
permit them to ultimately react chemically in the atmosphere.
In conclusion, the observation of SO2 oxidation rates > 1 percent/hr in unpolluted air is
certainly important in considering its removal rate on a global basis. The prevailing mechanism(s)
of the process are not clear.
BATTELLE — COLUMBUS
-------
21
REFERENCES
(1) Wagman, J., Lee, R. E., and Axt, C. J., Atm. Environ., 1, 479 (1967).
(2) Miller, D. F., Schwartz, W. E., Jones, P. W., Joseph, D. W., Spicer, C. W., Riggle, C. J., and
Levy, A., "Haze Formation: Its Nature and Origin, 1973", EPA-650-3-74-002, NERC,
Research Triangle Park, N.C. (June 1973).
(3) Altshuller, A. P., Environ. Sci. and Tech., 7, 709 (August 1973).
(4) Junge, C. E., Air Chemistry and Radioactivity, Academic Press, New York-, N.Y., 1963.
(5) Roberts, P. T., and Friedlander, W. M., "Conversion of SO2 to Ambient Particulate Sulfates
in the Los Angeles Atmosphere", presented at the Conference on Health Consequences
of Environmental Controls, Durham, North Carolina (April 1974).
(6) Wilson, W. E., Miller, D. F., Hopper, D. R., and Levy, A., "Pilot Study of the Interaction of
SC>2 in Photochemical Smog Using New York City Air", Final Report from Battelle's
Columbus Laboratories to the American Petroleum Institute (October 1971).
(7) Kushnir, J. M., Malin, H. I., Mohnen, V., Yencha, A. J., and McLaren, E. A., "Non-photochem-
ical Aerosol Formation from the Anhydrous Reaction Between Ammonia and Sulfur Dioxide:
Mechanism for the Formation of Ammonium Sulfate in the Atmosphere", NAPCA Publ. No.
165, State University of New York at Albany (1972).
(8) Bufalini, M., Environ. Sci. and Tech., 5, 685 (August 1971).
(9) Allen, E. R., McQuigg, R. D., and Cadle, R. D., Chemosphere 1, 25 (1972).
(10) Cox, R. A., J. Phys. Chem., 76, 814 (1972).
(11) Leighton, Photochemistry of Air Pollution, Academic Press, New York, 1961.
(12) Davis, D. D., Payne, W. A., and Stief, L. J., Sci., 779, 280 (1973).
(13) Wood, W. P., Castleman, A. W., and Tang, I. N., "Mechanisms of Aerosol Formation from
SO2", presented at the 67th Annual Meeting of the Air Pollution Control Association,
Denver, Colorado (June 1974).
(14) Cox, R. A., and Penkett, S. A., J. Chem. Soc., Faraday Trans I, 68, 1735 (1972).
(15a) Calvert, J. G., "Interaction of Air Pollutants", presented at the National Academy of Sciences,
Conference on Health Effects of Air Pollutants, Washington, D.C. (October 1973).
(15b) Calvert, J. G. and McQuigg, R. D., "The Computer Simulation of the Rates and Mechanisms
of Photochemical Smog Formation", Int. J. Chem. Kinet. (in press).
(16) Tuesday, C. S., Chemical Reactions in the Upper and Lower Atmosphere, Interscience,
New York, N.Y. (1961).
(17) Mason, D. R., and Piret, E. L., Ind. Eng. Chem., 42, 817 (1950).
(18) Whitby, K. T., Lui, B.Y.H., Husar, R. B., and Barsic, N., J. Colloid Interface Sci., 39, 136
(April 1972).
(19) Bray, W. H., J. Mater., 5, 233 (1970).
(20) Fuchs, N. A., The Mechanics of Aerosols, Permagon Press, New York, N.Y. (1964).
(21) Schafer, H.N.S., Analytical Chemistry, 39, 1719 (1927).
(22) Barton, S. C, and McAdie, H. G., Proceedings of the Third International Clean Air Congress,
Dusseldorf, Germany (1973).
BATTELLE — COLUMBUS
-------
22
(23) Davis, D. D., Smith, G. and Kiauber, G., Science, 186, 733 (November 22, 1974),
(24) Sidebottom, H. W., Badcock, C. C, Jackson, G. E., Calvert, J. G., Reinhardt, G. W., and
Damon,'E. D., Environ. Sci. and Technol., 6, 72 (January 1972).
(25) Blacet, F. E., Ind. Eng. Chem., 44, 1339 (1952).
(26) Horowitz, A. and Calvert, J. G., Int. J. Chem. Kinet., 5_, 243 (1973).
(27) Cehelnik, E., Spicer, C. W., and Heicklen, J., J. Amer. Chem. Soc., 93, 5371 (1971).
(28) Chung, K., Calvert, J. G. and Bottenheim, J. W., Int. J. Chem. Kinet. (in press).
(29) Okuda, S., Rao, T. N., Slater, D. H. and Calvert, J. G., J. Phys. Chem, 73, 4412 (1969).
BATTELLE — COLUMBUS
------- |