INTRODUCTION

Air pollution observed in many urban areas is commonly referred to as smog. Originally
identified with smoke from coal fires mixing with low-lying fogs, the term smog is derived from
a combination of the words “smoke” and “fog”. Smog has historically been described as two
types named after the places where they were first observed, London and Los Angeles. The
London smog has been called reducing or sulfurous smog. This type is identified by high
concentrations of sulfur dioxide (SO2) and particulate matter. London smog is therefore
composed of air pollutants that are emitted directly by combustion processes and are called
“primary pollutants”. The most severe episodes of London smog occur in urban areas when low
winds and a shallow boundary layer of air act to concentrate the pollutants in a relatively small
air volume. The meteorological conditions that enhance this type of smog are usually observed in
winter months during the early morning hours. This type of smog, which is characterized by high
aerosol loadings and high levels of sulfur dioxide occurs in many cities that are burning coal
today. Beijing and Shanghai, China, Cairo, Egypt, Belgrade, Yugoslavia, and Calcutta, India are
current examples of urban centers with London type smog. The Los Angeles smog is an
oxidizing smog formed from the photochemical reactions of nitrogen oxides and volatile organic
hydrocarbons, which are primary pollutants released from the combustion of fossil fuels in
automobiles, power plants, etc. Therefore, the Los Angeles smog, also known as photochemical
smog, is identified with high oxidant levels (e.g., ozone) and haze from aerosols formed in the
smog chemical reactions. The term “photochemical smog” refers to a combination in the
atmosphere of chemical species, each of which is the product of reactions involving initially a
mix of nitrogen oxides and hydrocarbons with normal atmospheric gases. The principal
components of concern are ozone and a range of oxidated nitrogen species, which are significant
bronchial irritants, some of which are corrosive to materials. Since late 1989, the Department of
Environmental Protection has monitored the concentration of ozone at Caversham. High values
were recorded at this site, but these could not show the extent of the region affected. When the
State Energy Commission of Western Australia proposed to expand a gas turbine power station
to the north of the city, this lack of knowledge of the affected region meant that the potential
effect on ozone concentrations was uncertain. The proponent, therefore, agreed to support a
study of the region’s photochemical smog, with the commitment that, if the turbines were shown
to contribute to photochemical smog pollution, low-emission technology would be retrofitted.
The major components of the study include the monitoring of photochemical smog species and
of meteorology, the development of an emissions inventory, and adaptation of existing computer
models to represent the transport and photochemistry of smog in the region. The aim of the
model developments is to represent a set of periods giving rise to high ozone concentrations.
Using this set, the effects of population growth and various control strategies may be tested.
Because the species responsible for photochemical smog are created in the atmosphere by
photochemical reactions of the primary combustion pollutants, they are called “secondary
pollutants”. The Los Angeles type smog is therefore dependent on the daytime intensity of solar
radiation, which promotes the formation of the secondary oxidants and aerosols. In contrast to
London smog, Los Angeles smog is formed during summer months and later in the day, after the
solar radiation has had time to produce the secondary chemical air pollutants. The trace gas
species formed in photochemical smog include the secondary oxidants ozone, PANs, and
inorganic sulfuric and nitric acids. These oxidants and acids play important roles in the effects of
smog on agriculture and ecosystems, in human health, and in materials damage on urban,
regional, and global scales. Most major cities in the world today suffer from some form of
photochemical smog, with some important examples being Mexico City, Mexico, Rio de Janeiro
and Sao Paolo, Brazil, Houston, Texas, USA, and Santiago, Chile. Smog then consists of
aerosols and trace gas species that are of concern due to their potential impacts on human health
and biota (e.g. agricultural damage, etc.). Both aerosols and trace gas species present in smog are
also important in determining the radiative balance of the troposphere, a property known as the
greenhouse effect. These species can heat the atmosphere by absorbing infrared and solar
radiation, trapping it in the troposphere. Aerosols can also contribute to cooling of the
atmosphere, both directly by scattering incoming solar radiation back to space and indirectly by
enhancing cloud formation. The net effect of aerosols on the radiative balance of the atmosphere
depends on the types of aerosols present, their chemical compositions, and where in the
atmosphere they are formed. Thus, the impacts of smog formation on urban, regional, and global
scales are a cause for concern with regard to overall ecosystem well-being as well as human
health.
The formation of photochemical smog in the tropospheric layer of the atmosphere is a complex
process involving the reactions of hundreds of primary precursors in the presence of ultraviolet
sunlight to generate ozone (O3), the principal compound we associate with smog. A new
approach to the study of the chemistry of smog formation has been made possible through the
development of high-speed computers and modern mathematical methods. Computer modeling
is the central technique of this new approach; however, limits and standards for this technique
have not yet been clearly defined. This review is meant both to describe our current knowledge
of smog chemistry, and to help formulate and describe this new approach to elucidation of a
complex scientific phenomenon. Two other advances in the study of smog formation are the
development of smog chambers for experimental purposes and of sophisticated analytical
capabilities with which to assess their results. These tools allow us to perform experiments that
range from simple to extremely complex, under realistic conditions, because smog episodes
generally have a duration of only l or 2 days. In contrast, conditions in the stratosphere limit the
use of controlled laboratory experiments for ozone similar to those performed in smog chambers,
because some important stratospheric chemistry occurs at time scales much longer than a day.
Although the methodology used to study stratospheric chemistry differs from that used to study
smog chemistry, many of the reactions are the same in both parts of the atmosphere; therefore,
stratospheric studies and smog chamber experiments that attempt to replicate the conditions of
the troposphere are complementary. Problems with smog chamber wall effects currently exist;
however, these problems are being addressed. As discussed below ("A Methodology for the
Validation of Smog Chemistry").
It is generally accepted that, under clear sky conditions, the atmospheric oxidation of SO2 to SO4
proceeds mainly via hydroxyl radical (HO) addition. Since the HO radical is the main chain
carrier in photochemical smog, the effect of the HO-SO2 reaction on this chemical system is of
great interest. Furthermore, an understanding of this reaction is important to the development of
effective control strategies to reduce ambient sulfate levels. Most atmospheric modelers have
assumed that SO2 is an HO chain terminator and thus summarized the overall oxidation by the
chemical equation
SO2 + HO4 → H2SO4 (a)
If SO2 is assumed to be an HO chain terminator, then model calculation predicts a high degree of
non-linearity between SO2 control and SO reduction (Meagher and Luria, 1982). In addition, the
assumption of chain termination led these authors to suggest that SO2 control could enhance
NO3- production. On the other hand, if HO radicals are regenerated, then a more linear
relationship between SO2 control and SO4 formation would be expected with little or no effect on
NO3- formation. Recently, Stockwell and Calvert (1983) studied the HOSO2 reaction mechanism
by photolyzing mixtures of HONO in the presence of SO2, NO, N02, H20 and CO. By
extrapolating their data to atmospheric conditions, they concluded that SO2 is not an HO chain
terminator. On the basis of their data, they suggested that the overall oxidation process can be
summarized in one equation
HO + SO2 ( +O2 , H2O) 4 → H2SO4 +HO2. (b)
Although this reaction is thermochemically favourable by 53 kcal mole-1, Stockwell and Calvert
were unable to prove the elementary reactions which result in this overall equation. A series of
smog chamber experiments was performed by scientists to further explore the role of SO2 in
atmospheric photochemical reactions.
In these experiments, the efficiency of SO2 as an HO chain terminator is evaluated by
monitoring steadystate HO levels. Pollutant levels were maintained as low as practical to provide
a realistic data set against which the various mechanistic options could be tested.
Nitrogen dioxide is one of the most important pollutants in urban air. At elevated concentrations
it is directly toxic to humen, and in the UK ambient concentrations widely exceed the EC guide
value of 26 ppb (50 percentile), and at some sites approach the limit value of 105 ppb expressed
as a 98 percentile of hourly average concentrations over 1 year. It is of particular concern that
whilst annual average concentrations of nitrogen dioxide have remained constant at most central
urban sites, peak and 98 percentile concentrations have tended to increase. There are few sources
of primary nitrogen dioxide and the major part of NO2 in urban air is formed from oxidation of
nitric oxide within the atmosphere. In most urban areas motor traffic is the predominant source
of nitric oxide emissions. During summer months, elevated concentrations of NO2 can arise due
to atmospheric photochemistry producing peroxy radicals which accelerate the conversion of NO
to NO2. However, the highest concentrations of NO2 in UK urban areas typically occur in the
winter. Most notably, in the December 1991 air pollution episode in London, a UK site of 423
ppb was measured in central London. The pathways for NO2 formation from NOx are complex,
as exemplified by Figure.a which shows the relationship of hourly mean NOx to hourly NO2 for
the months January, February, November and December 1991 at a busy roadside site in central
London. The Figure is divided into three regimes; in the region marked A, concentrations of both
NOx and NO2 are low and NO2 increases as a relatively constant proportion of NOx. As NOx
exceeds approximately 75 ppb, the graph enters section B in which there is a plateau of near
constant NO2 concentrations with little change in NO2 as NOx continues to increase. Both
sections A and B may be explained by oxidation of nitric oxide by ozone. In section A this
reaction is not consuming all ambient ozone, whilst in section B the process is oxidant limited
and virtually all ozone is destroyed. Section C, however, which occurs with NOx concentrations
above about 600 ppb shows rapidly increasing NO2 with further increases in NOx. It is this
section of the graph which is associated with chemical reactions are occurring as this stage and
the NO plus O2 reaction begins to play a role. However, the analysis of the London December
1991 episode showed that inclusion of this reaction could not explain the rapidity of rise of NO2
during episodic conditions.

Fig.a. Hourly NO2 and NO, concentrations at a busy roadside (Cromwell Road) in London in
January, February, November and December 1991.
Brief History of Smog
Air pollution has been recognized for some time as an urban problem. Moses Maimonides
(1135-1204 A.D.) noted in his writings:
“Comparing the air of cities to the air of deserts and arid lands is like comparing waters that are
befouled and turbid to waters that are fine and pure. In the city, because of the height of its
buildings, the narrowness of its streets, and all that pours forth from its inhabitants and their
superfluities . . . the air becomes stagnant, turbid, thick, misty, and foggy. . . . If there is no
choice in this matter, for we have grown up in the cities and have become accustomed to them,
you should . . . select from the cities one of open horizons . . . endeavor at least to dwell at the
outskirts of the city.”
His description and discussion clearly indicate that in the twelfth century air pollution was
already associated with urban environments and their higher population densities. During the
fourteenth century, coal burning had become such a problem in London that King Edward I
passed a law banning the burning of coal when the English Parliament was in session. This was
the first known legislation aimed at reducing air pollution and human exposures to smog. During
the seventeenth century, after the Restoration, John Evelyn, Esquire, was commissioned by King
Charles II to write for the crown concerning London’s environmental quality and public health
status. He wrote Fumifugium or The Inconveniencie of the Aer and Smoak of London Dissipated
in 1661 A.D. Commenting on the air quality in London due to the combustion of coal, he wrote:
“This horrid Smoake which obscures our Church and makes our Palaces look old, which fouls
our Cloth and corrupts the Waters, so as the very Rain, and refreshing Dews which fall in the
several seasons, precipitate to impure vapour, which, with its black and tenacious quality, spots
and contaminates whatever is exposed to it. . . . It is evident to every one who looks on the yearly
Bill of Mortality, that near half the children that are born in London die under two years of age.”
He further noted in his work that a child born in a country village had an even chance of living
40 years. These early works indicated the severity of the problem of air pollution in urban
centers, even in pre-industrial times. In 1775 A.D., English scientist Percival Pott observed that
chimney sweeps were at risk to contract cancer due to their professional exposures to coal tars,
ash, and combustion products. Thus, the long-term exposure to high levels of coal combustion
products was linked to human health as early as the eighteenth century.
Overview of the Smog Formation Process
The key elements of smog chemistry Haagen-Smit, a plant biochemist studying flower essences
in Los Angeles after World War II, originally identified oxygenated hydrocarbons as a key
element of smog through a comparison of odors (Haagen-Smit, 1952). From a 1937 Swiss
patent, he learned that hydrocarbons and nitrogen oxides could be used to generate oxygenated
hydrocarbons in the presence of ultraviolet light (Pitts and Metcalf, 1980). In what can be termed
the first smog chamber experiments, Haagen-Smit was able to generate the "smell" of Los
Angeles air by irradiating parts per million quantities of heptanes and nitrogen oxides. Although
we now know that ozone is also generated in this process, and that oxygenated hydrocarbons
play a key role in its chemistry, the main elements of smog formation are those demonstrated by
Haagen-Smit (1952):
hydrocarbons, nitrogen oxides, and ultraviolet sunlight. The fundamental chemistry of ozone
formation was reviewed by Leighton (1961). However, the key pathways of the system were not
made clear because computer models and smog chamber experiments that could define these
pathways among the many possibilities had not yet been developed. Also the rate and photolysis
constants of many of the key steps were either unknown or roughly estimated. The first major
review involving computer models and smog chamber validation was done by Demerjian, Kerr,
and Calvert (1974). Their review identified and verified the key hydrocarbons and nitrogen oxide
cycles. Some further refinements made during the last decade are reviewed below ("A
Methodology for the Validation of Smog Chemistry"). The primary benefits of refinements in
basic chemistry have been the development of condensed chemical schemes or mechanisms and
an enhanced knowledge of the key overall processes. Condensed mechanisms are important in
atmospheric modeling because they keep overall numerical problems at a tractable size;
enhanced knowledge of the primary overall processes and smog formation allows us to make the
simple tutorial explanations presented here. Although smog chemistry at the detailed level
studied by chemical kinetists is complex, its key elements can be explained using a minimum of
chemistry.
The nitrogen cycle and the steady-state relationship for ozone:
The formation of ozone in the lower atmosphere (i.e., troposphere) stems from only one reaction:
the addition of oxygen atoms to-oxygen molecules in the presence of a third body:
0 + 02 + M → 03 + M (2)
The oxygen atoms are known to stem from principally the photolysis of nitrogen dioxide (NO2);
this reaction has traditionally been called reaction (1):
NO2 + hv → NO + O (1)
The addition of one reaction completes a cycle that we shall call the "nitrogen cycle," which has
considerable heuristic value. This reaction is the molecule-molecule exchange between ozone
and nitric oxide:
O3 + NO → NO2 + O2 (3)
The photolysis reaction (1) is usually the rate-limiting reaction of the nitrogen cycle, and the
photolysis constant for this reaction is typically between 0.25 and 0.5 per min during most days.
Hence the cycle time for these three reactions is only a few minutes. Ozone accumulation takes
several hours; therefore, the nitrogen cycle "spins" fast enough to maintain a close approximation
to the following photostationary-state equation derivable from these three reactions:
(O3)ss = (k1 / k3) x [NO2] / [NO].
The rate constant ratio (k1/k3) is about 1/100, so a NO2/NO ratio of 10/1 is needed to generate 0.1
ppm of ozone. Note that Reaction (3) converts NO to NO2, but that an ozone molecule is
destroyed. Reaction (1) produces the oxygen atom leading to a new ozone molecule, but NO2 is
converted to NO at the same time. Hence, the nitrogen cycle does not generate or destroy a net
ozone molecule. An additional path of conversion from NO to NO2 is needed that will not
destroy 03, in order for it to accumulate according to the photostationary-state relationship of the
nitrogen cycle. This alternate NO-to-NO2 conversion pathway is provided by the atmospheric
photochemical oxidation of reactive organics such as hydrocarbons. The key type of reaction in
hydrocarbon oxidation, which converts NO to NO2, is the fast radical transfer reaction with NO:
. .
RO2 + NO → RO + NO2 (4)

where R is virtually any organic moiety, or fragment, and can also be hydrogen. There is
considerable evidence to support the nitrogen cycle and the photostationary-state equation for
ozone concentrations in both urban atmospheres and smog chambers. Calvert (1976) analyzed
detailed data obtained in the Los Angeles Reactive Pollutant Program (LARPP) and concluded
that the data and the theory (photostationary-state equation) were consistent. However, since its
early prediction by Cadle and Johnston (1952), there has been some controversy concerning the
accuracy limits of this relationship. A recent study by Shetter et al. (1983) suggests that Reaction
(4) should be included, but even without such a correction, the photostationary-state equation
appears to be valid to within a factor of 2 for urban atmospheres. When computer models are
used to stimulate either smog chamber data or atmospheric situations, the photostationary-state
equation appears to hold somewhat better than the data of Shetter et al. (1983) suggest.
Table 1 presents typical smog chamber modeling results. The comparison with actual
observations is not relevant and is not shown; what is shown is the comparison between the
numerical integration (based on a recently developed detailed chemical mechanism for the smog
chemistry of propene using 110 reactions) and the values of the photostationary-state equations
at the times shown. The photostationary-state equation appears to hold to about 35% even though
NOx varies by a factor of 80, simulated ozone by a factor of nearly 2400, and the light intensity
(k1) by a factor of 25. Thus, the heuristic value of the photostationary-state equation is assured;
ozone formation in urban smog occurs via the simple nitrogen cycle, as NO is converted to NO2
by other reactions outside the nitrogen cycle [such as RO2 plus NO, through Reaction (4)]. Given
the current limits of our analytical capabilities and of our knowledge of detailed chemistry the
differences between the photostationary-state equation and the exact values of ozone seem small.
However, further comparisons are warranted, especially for rural areas with high hydrocarbon-
to-NOx ratios; for these conditions, the equation would not be expected to be valid because the
low NOx concentrations could not sustain a photostationary-state relationship.

The hydrocarbon oxidation cycle:

Hydrocarbons are oxidized in the atmosphere to form carbon dioxide (CO2) and water (H2O).
Intermediate steps in this overall oxidation process typically involve cyclic stages driven by
hydroxyl radical attack on either the parent hydrocarbon or on partially oxidized intermediate
compounds. The two essential characteristics of smog formation are reaction with the hydroxyl
radical (OH) to yield a peroxy radical and photolysis to yield a hydrogen containing radical. The
former element is common to virtually all organic molecules, but the latter usually involves
oxygenated intermediates containing the carbonyl (C = O) bond. The simplest organic molecule
exhibiting these two characteristics is formaldehyde (HCHO).
Table 1. Typical smog chamber computer simulation vaules related to the photostationary
estimate for ozonea
a
propene and NOx simulation of University of North Carolina outdoor experiment, 31 May 1979.

The oxygenation cycle for formaldehyde can be written by the following sequence of reactions:
. .
OH + HCHO → H2O + HCO (5)

. .
HCO + O2 → HO2 + CO (6)

. .
HO2 + NO → NO2 + OH (7)

Note that the essential peroxy radical is generated by Reaction (6) and the hydroxyl radical (OH)
returns in Reaction (7) to complete the cycle. In addition, Reaction (7) accomplishes the NO-to-
NO2 conversion central to ozone formation in the nitrogen cycle just discussed. Also note that
the carbon monoxide (CO) generated by Reaction (6) can react like an organic to yield a peroxy
radical:
. .
OH + CO → H + CO2 (8)

. .
H + O2 → HO2 (9)

. .
HO2 + NO → NO2 + OH (7)

However, CO does not have the second essential element for smog formation-a source of
hydrogen-containing radicals-which formaldehyde has:
 . .
HCHO + hv → H + HCO (10)

where hv denotes the absorption of light, ultraviolet in this case. The hydrogen atom (H) and
formyl radical (HCO) generated by this photolysis reaction produce two hydroperoxy radicals
via reaction with oxygen molecules, as shown in Reactions (6) and (9).
The NOX and radical sink reaction:
There remains only one more reaction that is central to our basic understanding of smog
photochemistry- the NOx plus radical sink reaction that forms nitric acid (HONO2 or HNO3):
.
OH + NO2 → HONO2 (11)

The overall process

The diurnal cycle:
We can now explain the classic set of NOx and O3 curves routinely seen in smog chamber
experiments and, approximately, what is seen in the atmosphere. Figure 1 shows this general
pattern for the daylight hours. The early-morning or initial NOx concentrations are composed
mostly of NO, because NOx emissions are composed primarily of NO. The NO-to-NO2
conversions of the RO2. reactions with NO reverse this situation, until NO2 becomes the
dominant NOx species. When the NO2/NO ratio becomes large enough, ozone builds up.
Meanwhile, NO2 decreases via the sink reaction (11). During the night, ozone is fairly inert
unless it comes in contact with surfaces, NOx compounds, or some hydrocarbons.

Fig. 1. Typical smog chemistry evolution profiles for NO, NO2, and ozone.
Responsiveness to precursor changes:
We can now also explain some of the essential effects of control strategies on the overall
chemical process of smog formation. If NOx emissions are held constant and organics are
reduced,ewer radicals are available, there are fewer organics to cycle through the hydrocarbon
oxidation cycle, and the entire smog process slows down, as shown in Fig. 2. If organic
emissions are held constant and NOx emissions are reduced, there is less of a sink for the same
number of radicals, and the entire process speeds up, as shown in Fig. 3. Whether ozone
increases or decreases with NOx control depends on the original organic-to-NOx ratio, as will be
explained later. Figure 3 shows the effect of NOx control on a moderately high organic-to-NOx
ratio. If the smog chamber type of result illustrated by Fig. 3 were an air parcel traveling along a
trajectory, then either time or distance along the trajectory path could be the abscissa. Note that
observers located at different points along the path in Fig. 3 would report different results from
NOx reduction. The NO2 observers would report virtually no change if they were located before
the original NO2 peak, whereas observers located near and just after the original peak would
report very large reductions. Those farther downwind would report little change, since
concentrations of NO2 would already be low and would probably be dominated by local
emissions. Similarly,on the downwind distance of the observer and the source distribution.

Fig. 2. Typical smog chemistry profiles after reducing hydrocarbon emissions at constant NOx.
Fig.3. Effects of reduced NOx emissions on the time dependence of NO2 and ozone
concentrations.

Peak ozone formation as described by the Empirical Kinetic Modeling

Approach (EKMA):
Smog abatement policies in the United States are directed toward the reduction of l-h peaks in
ozone concentrations. In recent years, a tool known as the Empirical Kinetics Modeling
Approach (EKMA) has been used to estimate the percent control in precursor emissions required
to reduce peak ozone levels. This approach is based on the use of a diagram that plots peak
ozone isopleth lines as a function of nitrogen oxide and organic levels, which are related to the
ordinate and abscissa. A simple moving air parcel trajectory model is exercised for several
precursor combinations, and the resulting peak ozone values form a set of points on the diagram
through which the isopleth lines can be drawn. An example of an EKMA isopleth diagram is
presented in Fig. 4. An understanding of the basic chemical process discussed previously can
also explain the shapes of the EKMA isopleth curves. First, note that the isopleth lines in the
lower right-hand part of the diagram are parallel to the abscissa, indicating that ozone
concentrations are insensitive to changes in hydrocarbon emissions, but quite sensitive to
changes in NOx. We know from the examples just discussed that the diurnal patterns will still
respond to changes in hydrocarbon emissions (a reduction in hydrocarbons will slow the overall
process, and though the maximum ozone within the air parcel remains the same). The reason for
the horizontal nature of the isopleths lines, however, stems from the competition between NO-to-
NO2 conversion and NOx loss. When ample radicals from the hydrocarbons exist, each NOx
molecule will experience some fixed, average number of cycles through the system before
removal to nitrate. Thus, ozone formation in the lower right-hand part of the diagram is limited
by the decrease in NOx concentrations to such low levels that the NOx and hydrocarbon cycles
can no longer generate ozone. We should add that the fixed number of cycles through the system
is a function of the mixture of hydrocarbons and the types of NOx loss. Reactions in the real
atmosphere are much more complex than simply those of formaldehyde chemistry with NOx loss
to nitric acid via the OH + NO2 reaction. However, the isopleths lines always turn out to be
horizontal for the same basic reason: a fixed number of cycles before NOx loss. The upper left-
hand corner of the isopleth diagram shows parallel isopleth lines slanted at an angle from the
ordinate. This is also a universal feature of EKMA isopleth diagrams. Most of the angle is caused
by the fact that NOx emissions are primarily composed of NO. If NOx emissions were composed
primarily of NO2, there would be much less angle to the lines. To test this assumption, we
generated an isopleth diagram using pure NO2 (Fig. 5). It can be seen from a comparison of this
isopleth diagram with Fig. 4 (0.25 NO2/NOx) that for pure NO2 the angle of the lines is much
less. The angle of the lines in Fig. 4 has the important implication that NOx emission reductions
result in higher ozone peaks when the NMHC/NOx ratios are low. Using the photostationary-
state equation for ozone, we can explain both the parallel nature of the lines and an angle in this
portion of the isopleth. Recall that the relationship for ozone is a constant (equal to about 0.01
ppm during much of the day) multiplied by the NO2/NO ratio. Put simply, we know that ozone
will not rise much above the 0.01 ppm level until NO2 becomes about equal to the concentration
of NO. Since the majority of NOx emissions consist of NO, we know that some finite number of
NO-to-NO2 conversions must occur before NO2 starts to become the dominant NOx species.
Such a finite number of conversions can, in turn, be associated with a finite number of
hydrocarbon oxidations. Hence, there is a fixed increment of hydrocarbon emissions required to
convert each increment of added NOx emissions to the point at which NO2 becomes greater than
NO and ozone concentrations can begin to build up. Thus, an angle occurs at the points where
ozone begins to build up above the 0.01 ppm level. Once NO2 is considerably greater than NO,
the amount of ozone produced depends more on the number of NO-to-NO2 conversions and less
on the concentrations of NOx. Hence the ozone isopleth lines all parallel the line that corresponds
to 0.01 ppm , the point at which ozone begins to build up.
Fig. 4. Example of EKMA ozone isopleths.
 Fig. 5. EKMA isopleths generated from pure NO2 as NOx.

Types
Following are two major types of smog
London Type Sulfurous Smog:
John Evelyn’s comments in Fumifugium were prophetic. During the continuing expansion of the
use of coal and fossil fuels in the nineteenth and twentieth centuries, the air pollution problem
continued to increase in major European cities. In London, the smoke that Evelyn referred to in
his writings to the King was enhanced by meteorological conditions conducive to fog formation.
This combination of smoke and fog led to the commonly used term smog. Smog in London and
many of the major cities in England during the first part of the twentieth century consisted of
high levels of sulfur dioxide mixed with carbonaceous soot and fly ash from coal use in homes
and from industrial emissions. The pollutant levels during intense fog conditions led to such
serious problems that during the winter of 1952, over 4000 excess deaths were observed due to
pulmonary and cardiovascular stress associated with killer smog events. This mortality led to the
passage of the British Clean Air Acts of 1956, aimed at reducing the emissions that produce
London type smog events. due to pulmonary and cardiovascular stress associated with killer
smog events. This mortality led to the passage of the British Clean Air Acts of 1956, aimed at
reducing the emissions that produce London type smog events. Europe was not alone in this type
of problem. On October 26, 1948, a killer smog event occurred in the United States. In Donora,
Pennsylvania, a smog cloud formed, consisting of a poisonous mix of sulfur dioxide (SO2),
carbon monoxide (CO), and metal dust from the smokestacks of the local zinc smelter where
most of the town worked. Over the next five days, twenty residents died and 7000 people, half
the town’s population, were hospitalized with difficulty breathing after being exposed to the
cloud. This type of smog is still quite common in many urban centers of the world, particularly
in China, India, the Middle East, and Eastern Europe.
Los Angeles Type Photochemical Smog
At about the same time that the London smog conditions were being described, a different type
of smog was observed in the United States, in Los Angeles, California, and surrounding areas.
This smog was not reducing in nature like the SO2-filled smoky air of London. Rather, this was
an oxidizing gas cloud filled with nitrogen dioxide (NO2) hydrocarbons, and their photochemical
product, ozone. The first major photochemical smog event, observed in June of 1943, and was
initially thought to be a “gas attack” by the Japanese, as World War II was in full swing. Later,
the event was recognized to be a cloud produced by the photochemical reaction of emissions
from automobiles and industrial plants in the Los Angeles air basin. The contributors to
photochemical smog are primary pollutants (carbon monoxide and carbonaceous soots, nitrogen
oxides, sulfur dioxide, and volatile hydrocarbons) and secondary pollutants formed in the
troposphere during photochemical reactions that are driven by sunlight (ozone, PANs, sulfate
aerosols, nitrate aerosols, organic aerosols, etc.). Note that the secondary pollutants formed in
photochemical smog are produced at higher concentrations during the summertime because of
the higher solar radiation. Atmospheric ozone levels typically reach their highest in the early
afternoon due to the sunlight driven chemistry of this type of smog formation. In contrast, the
London and Donora type smog events typically occurred during winter periods, when high
emissions of the primary pollutants sulfur dioxide and soot combined with low boundary layer
inversions leding to high concentrations near the surface. Thus, conditions are typically right for
primary pollutant fumigations in the winter and early morning. Similarly, the primary pollutant
carbon monoxide has been a major concern in a number of areas during winter months when the
emissions from mobile sources and from residential wood burning, together with the lower
wintertime boundary layer height, can lead to high levels of this toxic gas in the troposphere. In
contrast, the Los Angeles photochemical smog has the highest concentrations of oxidants and
secondary pollutants formed during the summertime in the early afternoon.
Photochemical Smog Processes
Ozone is produced naturally in the atmosphere by the action of ultraviolet light on nitrogen
dioxide, generating activated oxygen atoms and nitrogen oxide. The atomic oxygen quickly
combines with oxygen molecules to produce ozone, and the nitrogen oxide is reoxidised slowly
by oxygen, or more rapidly by ozone, to nitrogen dioxide. These processes normally maintain an
equilibrium which leads to a low background of ozone of about 15-30ppb (parts per billion).The
rate of nitrogen dioxide regeneration, and consequently ozone production, is accelerated by the
presence of reactive volatile organic compounds, and by the presence of nitrogen oxides from
anthropogenic sources. In these circumstances, the progress of photochemical smog reactions in
strong sunlight leads to a significant increase of ozone concentrations, peaking typically in the
mid-afternoon period. Inventories generally show motor vehicle emissions to be the most
important single source of volatile organic compounds. Industrial and biogenic sources (the latter
due to evaporation from vegetation) are also of significance. While mass emissions from the
biogenic sources may be less, the species emitted can be of high reactivity, making a
disproportionate contribution to the rate of smog formation.

Chemistry of Ozone formation:

Professor Arie Haagen-Smit, a perfumery chemist at the California Institute of Technology, was
the first to show that placing automobile emissions containing nitrogen oxides ( NOx ) consisting
of nitrogen oxide ( NO ), nitrogen dioxide ( NO2 ), and unburned hydrocarbons in sunlight
would produce ozone and eye irritants such as formaldehyde. Professor Frances Blacet,
chemistry professor at the University of California, Los Angeles, was the first to point out that
NO2 photolysis was responsible for the production of oxygen atoms in the troposphere. This led
to the realization that the following chemistry was responsible for both the production and loss of
ozone in the troposphere:
NO2 + sunlight (280-423 nm) → NO + O (12)
O + O 2 + M (N 2 , O 2 , Ar, etc.) → O3 + M (13)
O3 + NO → NO2 + O2 (14)
Note that reactions 12-13 lead to a situation where the ozone formation is dependent on
the concentration ratio of NO to NO2 and the sunlight intensity. The observation of elevated
levels of ozone in photochemical smog led to the conclusion that other radicals were also
important in this chemistry. These additional radical species were tied to the organics that were a
necessary part of smog formation. A number of researchers concluded that hydroxyl radical
(OH), hydroperoxyl radical ( HO2 ), and the associated organic peroxy radicals ( RO2 ) were key
to the photochemical chain reactions that allow for the conversion of NO to NO2. Thus,
increased NO2 and decreased NO concentrations result in the enhanced ozone formation due to
reactions 12 and 13. The key initiation step leading to the formation of the OH is the photolysis
of ozone ( O3), yielding an electronically excited oxygen atom, O(1D) , that can react with water
vapor ( HO2 ) in the air to form the reactive OH species:

O3 + sunlight (290-320 nm) → O(1D) + O2 (15)

O(1D) + H2O → 2 OH (16)
Once formed, OH can react with carbon monoxide (CO) or hydrocarbons to create reactive
peroxy radicals and to reform OH, setting up a chain reaction for the production of ozone.
For example, the ozone-producing chain reactions involving CO are as follows:

OH + CO → H + CO2 (17)
H + O2 → HO2 (18)
HO2 + NO → NO2 + OH (19)
NO2+ sunlight → NO + O (20)
Reaction (20) provides the starting materials for the formation of ozone via reaction (13). Note
that the OH acts to oxidize CO via reaction (17) and is reformed during the oxidation of NO to
form NO2 in reaction (19). For the case of a simple hydrocarbon, ethane ( CH3CH3 ), the reaction
scheme is:
•
OH + CH3 CH3 → H O + CH3 CH2 (21)

• •
CH3 CH2 + O2 → CH3 CH2 O2 (22)

• •
CH3 CH2 O2 + NO → NO2 + CH3 CH2 O (23)
 •
CH3 CH2 O + O2 → CH3 CHO + HO2 (24)

•
The HO2 formed from the reaction of the ethoxy radical ( CH3 CH2 O ) and molecular oxygen

(O2) can, in turn, react with NO via reaction (19) and reform OH again, initiating a chain
reaction. This catalytic cycle will continue until the volatile hydrocarbons are depleted or chain
termination reactions occur stopping the chain.

Sulphur Dioxide:
Any impression that the mineral extraction techniques of today have reduced the minimum is
immediately banished repugnant fact that sulphur production, sulphuric acid and all other forms
are produced by the chemical industry) is exceeded by the amount of sulphur poured into the
atmosphere as sulphur dioxide waste from burning fuel. This sulphur dioxide is produced almost
entirely by the burning of fuel, predominantly coal and coke, and to a lesser extent automobile
fuels. Sulphur dioxide then react to form sulphurous smog in the atmosphere.
Role Of SO2 on Hydroxyl Smog Chemistry:
To investigate the role of SO2 on hydroxyl smog chemistry , a series of chamber experiments
was carried out under the experimental conditions presented in Table 2.

Table 2. Summarv of initial conditions and experimental parameters

Chamber simulations were performed at constant initial NOx and hydrocarbons concentrations
with variable amounts of added SO2 (0-6.4ppm). Simulations were conducted at low and
moderate humidity levels (5 and 40% relative humidity) respectively at 25 °C. The major
reactive hydrocarbon used in this study was propene; in addition, n-butane was also added to all
experiments to allow in situ monitoring of HO radical levels. The hydroxyl radical concentration
estimated from the loss of propene corrected for its reaction with O3 and dilution, was in good
+
agreement ( - 25 ) with the HO concentration estimated from the loss of butane The overall
effect of SO2, on the conversion of NO to NO2 and the formation of O3, is shown in Fig.6 . for
the low H2O content, and Fig.7 . for higher H2O content. These figures reveal that under both
‘dry’ and ‘wet’ conditions, SO2 was found to slightly accelerate the conversion of NO to NO2
and to increase the O3 peak concentration. On the other hand, as SO2 concentration increased, the
HO steady state concentration (as determined from the loss rate of C3H6 and C4H10) was found to
decrease. The observed changes were well beyond the experimental uncertainty. Hydroxyl
radical concentrations averaged over the entire irradiation period (4 h) and calculated
independently from the C3H6 and C4H10 loss rates are presented in Table 3. The time dependent
HO concentrations for the experiments with 0, 3.07 and 6.38ppm of added SO2 are shown in
Fig.8. A comparison between the ‘wet’ and ‘dry’ experiments reconfirms that water vapor
enhances the photochemical activity in these systems. In this study, it was desirable to maintain
the same HO levels in both ‘wet’ and ‘dry’ experiments. In order to achieve this, light intensity
was reduced in the ‘wet’ experiments. The hydroxyl radical concentrations estimated for each
experiment were used to calculate the rate of SO4 formation using the rate coefficient
recommended by Atkinson and Lloyd (1980) for the HO-SO2 reaction (1.39 x l03 ppm-1 min-1, at
305 K and 1 atm). These values were compared with the average SO2 to SO4 conversion rates
which were determined by monitoring the rate of aerosol volume increased. The excellent
agreement, between calculated and measured SO4 formation rates indicates that, under the
conditions employed, HO addition to SO2 can account for all the measured SO4. Computer
simulations were performed to evaluate the details of the mechanism involved in the oxidation of
SO2 to SO4. A modified version of the photochemical scheme was tested with input data from
SO2-free experiments. As is the case in all smog chamber models, a supplementary source of HO
radicals has been introduced to obtain acceptable agreement between calculated and observed
concentrations. Although the exact nature of this radical source has not been determined, it has
been suggested that this effect is due, at least in part, to the photolysis of nitrous acid
(HONO → HO + NO)
which may be formed in a heterogeneous wall reaction. In this study the unknown source of HO
radicals needed to fit the data was (3.5-+1.0) x 10-5 ppm min-1, which is approximately 10% of the
levels previously employed in smog chamber modeling. This source contributes approximately
5% to the homogeneous HO production in the chamber and, therefore; permits these experiments
to be more sensitive to changes in chemicals rather than to external parameters.

Fig. 6. The effect of SO2 on the rate of NO loss and on the rates of NO2 and O3, formation in
smog chamber experiments with low H2O content ; SO2 concentrations 0 (-), 4.51 (----) and 5.87
(-‘-‘-‘) ppm.
 Fig. 7. The effect of SO2 on the rate of NO loss and on the rates of NO2 and O3, formation in
smog chamber experiments with moderate H2O content; SO2 concentrations 0 (---), 3.07 (-----)
and 6.38 (-.-.-.) ppm.

Fig.8 . Study-stateconcentrations of HO radicals calculated from hydrocarbon

loss rate for experiments CH83222 (a), CH83216 (A) and CH83244 (0).
 Table 3. Average [HO] calculated from C3H6 loss rate
([HO]p) and from C4HI0 loss rate ([HO]b)

Nitrogen dioxide
Nitrogen dioxide is one of the most important pollutants in urban air.
Measurement of the NO2/NOx ratio in road traffic emissions:
NO and NO2 concentrations were measured in a road tunnel (Queensway Tunnel) on 31st
October 1992; 26th January and 11th February 1993. The Queensway tunnel is located in
Birmingham city centre and has dimensions of approximately 5 m in height and width for each
of the two cardividing wall) and 600 m in length. There are a total of 18 ventilation sites in the
tunnel which are switched on when the levels of CO in the tunnel exceed 50 ppm. The flow of
traffic through the tunnel is of the order of 20-45 vehicles/min in each direction. Diesel vehicles
accounted for approximately 5% of this total at the time of the measurements. A
chemiluminescence analyser (Environnement AC 30 M) was employed to measure NO and NO2
concentrations in the tunnel. The precision of measurements was greater than 3% for NO and 5%
for NO2. The sampling site was in the emergency bay near the half-way point of the tunnel. The
inlet of the NO-NOx analyser was located approximately 0.5 m from the kerbside and at about
1.8 m height. Results of the measurements of NO and NO2 concentrations and NO2/NOx ratios
on 26th January 1993 are presented in Fig.9 . The meteorological conditions were as follows:
relative humidity ranged from 51 to 81%, temperature from 6.8 to 9.40C and it was raining in
the afternoon. Nitric oxide varied from 1280 to 1780 ppb, nitrogen dioxide from 104 to 175 ppb
and NO2/NOx ratio from 6.3 to 8.1% in the tunnel. It should be noted that a traffic jam occurred
from 17:46 to 18:0l (shown in Fig. 9). During this period nitrogen dioxide increased from 121 to
152 ppb and the ratio from 6.8 to 8.1%, while nitric oxide increased only slightly. The
concentrations of total oxidants (NO2 + O3) at a background site in the city kept steady at a level
of approximately 40 ppb during the three sampling days. Because of high concentrations of NO
in the tunnel and low light intensity, it is reasonable to assume that the ozone concentration was
zero and about 40 ppb NO2 was contributed from background ozone. Compared with the high
NO concentration in the tunnel, background NO was negligible. Subtracting the NO2 formed
from background ozone, the NO2/NOx ratios would be 3-5.5%, 4-6% and 5-7.5%, respectively
on three sampling days. Clearly, these ratios should be reflective of those in the actual vehicle
emissions .It appears that idling vehicles give higher NO2/ NOx ratios, as shown in Fig. 9, which
is consistent with the results of Lenner and Lindqvist, but even the maximum ratio during the
traffic jam was only 8.5%.

NO2 formation near the tailpipe:

In vehicle exhaust gas, the NOx concentration is a function of driving mode, and is typically
lOOO- 4000 ppm during acceleration, lOOC-3000 ppm during cruising, and 5-50 ppm during
deceleration 141. The NOx concentration in exhaust will typically be below 1200 ppm when
driving at a speed of 50 km h - or less , as is normal in urban areas. The homogeneous reaction
2NO + O2 → 2NO2 (25)
is normally of little significance in the atmosphere because the NO concentration is insufficient.
It has been unclear, however, whether NO2 could form significantly near the tailpipe from
reaction. Hov and Larssen set up a simple numerical model of the combined effect of chemistry
and dilution with eight different dilution functions which indicated that the dilution rate during
the initial few seconds was extremely important for NO2 formation. In order to evaluate the
extent and significance of NO2 formation in the exhaust plume immediately after emission,
experiments have been carried out in a wind tunnel with the dimensions of 3 m length x 0.3 m x
0.3 m section fitted with an adjustable speed fan. The maximum velocity created by the fan was
3.0 m/s which was measured with a pitot tube and digital micromanometer (MP Series 4
Autozero). The Reynolds Number of the wind tunnel is 6.4 x 104, 4.3 x l04 and 2.1 x l04 at tunnel
velocities of 3 m/s, 2 m/s and 1 m/s, respectively. Three types of nitrogen oxide sources were
used. NO and NOx were measured simultaneously with a chemiluminescent NOx analyser
(Environnement AC 30 M, range 10 ppm) preceded with a dilution system.
Emission from a petrol generator:
A petrol generator, model Honda EM 2200X was employed as an emission source of NOx. The
load was fixed at 400 W by using four 100 W lights. A flexible metal tube connected to the
generator exhaust was located at the centre of one end of the wind tunnel. The diameter of the
tube was 2.5 cm, and the emission velocity was 10.6 m/s. Air sampled at the distance of 30, 60,
90 and 120 cm from the emission orifice by using a 0.25 inch diameter steel tube probe which
could move freely across the wind tunnel. The NOx concentration in the original exhaust plume
was approximately 30 ppm, which quickly diluted after emission. Because of the high emission
velocity, it is reasonable to treat the emission as a free jet. Using free jet theory, it may be
calculated that oxygen will reach the plume centre at a distance of approximately 11 cm. The
distribution fluctuated near the edge of the tunnel as a result of the large ratio of the diameter of
the emission tube (2.5 cm) to the cross-section of the tunnel (30 x 30 cm). The highest ratio of
NO2/NOx in each run, up to 9%, was in the centre of the plume at a distance of 30 cm. NOx
formation from the reaction
2NO +O2 → 2NO2
with an initial NO concentration of 30 ppm and short residence time could not account for this
phenomenon. Overall, the NO2/NOx ratio of the plume was low. The wind speed in the tunnel
was varied from 0-3 m/s. The tunnel speed changed the rate of NOx dilution, but did not affect
the NO2/NOx ratio appreciably.
Emission from vehicle exhaust:
Exhaust from an idling car, a university-owned van (Ford Escort; 1.3 1 petrol) registered in 1985,
was passed into the wind tunnel as a NOx emission source. The emission velocity was 10.5 m/s;
the diameter of the emission tube was the same as before. The emission of NOx and initial
NO2/NOx was recorded in the plume at a distance of 30, 60, 90 and 120 cm from the emission
point.
Emission from a pure NO cylinder:
An artificial exhaust jet was made by mixing pure NO and nitrogen gas. The NO concentration
in the original emission was 1000 ppm. The diameter of the steel emission tube was 6.35 mm,
giving an emission velocity of 14.7 and 29.4 m/s. The corresponding Reynolds number was 6350
and 12 700 in the steel tube, and the wind tunnel velocity was up to 3 m/s. Clearly, the NO2/NOx
ratio was low and insensitive to the emission velocity and tunnel speed. The vehicle exhaust
plume behaves as a free jet because of its much greater emission speed than the surrounding
fluid. The turbulent jet dilutes quickly and NO2 formation by reaction (25) in the plume of idling
exhaust is of little significance because of the modest NO concentration and fast dilution. The
ratio of NO2/NOx, formed by reaction (25) in an emission jet with 1000 ppm of NO, is
insensitive to the jet speed and wind speed in the wind tunnel, and ranges from 0 to 6%. This is
consistent with the results obtained from the measurements in the road tunnel where the
NO2/NOx ratio from the motor traffic was less than l0%, even within a 15 min traffic jam .The
sources of NO2 in winter smog episodes in UK have been investigated. The road tunnel
measurements and wind tunnel experiments showed that the original ratio of NO2/NOx in bulk
vehicle emissions was low, less than l0%, even in a 15 min traffic jam; NO2 formation near the
emission orifice did not contribute to the NO2/NOx ratio signficantly. Much faster conversion of
NO to NO2 was found in diluted exhausts of petrol-fuelled cars, which probably represents an
important NO2 source in winter smog episodes. Further work is investigating the mechanisms of
the chemical reactions responsible.

Peroxyacyl Nitrate Formation

Acetaldehyde ( CH3CHO ), emitted directly from automobile exhaust or from secondary
reactions similar to reactions 21-24, can also react with OH. The abstraction of the
aldehydic hydrogen leads to the formation of an acetyl radical, which can add molecular
oxygen ( O2 ) to form a peroxyacetyl radical ( CH3COO2 ):

CH3CHO + OH → CH3CO• + H2O (26)

CH3CO• + O2 → CH3COO2 (27)

The peroxyacetyl radical can react with NO to form NO2 :

CH3COO2 + NO → CH3CO2 + NO2 (28)
CH3CO2 → CH3• + CO2 (29)

CH3•+ O2 → CH3O2 (30)

CH3O2 + NO → CH3 O + NO2 (31)
CH3O + O2 → CH2O + HO2 (32)
This set of smog chemistry reactions leads to the production of formaldehyde ( CH2O )
and the hydroperoxyl radical, which can in turn react with NO to reform OH via
reaction 19.
The peroxyacetyl radical also forms a stable adduct with NO2 , forming peroxyacetyl
nitrate (PAN).
CH COO2 + NO2 ↔ CH CO-OO-NO2 (33)
Peroxyacetyl nitrate, which is in equilibrium with the peroxyacetyl radical and NO2,
was initially discovered by agricultural researchers who had identified a new type of
plant damage in the Los Angeles air basin. The damage, leaf bronzing of a number of
plant species, did not resemble known damage due to ozone, sulfur dioxide, or other air
pollutants. The damage, originally attributed to “Compound X”, led to studies
conducted at the Franklin Institute in Philadelphia and at the University of California,
Riverside, where PAN was discovered and isolated and its phytotoxicity was confirmed.
Peroxyacetyl nitrate is not the only compound of this sort, as a number of species of the family
of compounds known as the PANs, with the general structure of RCO-OO-NO2 , have been
found in the troposphere. However, PAN is the most abundant member of this group of
compounds. The PANs are eye irritants (i.e., lachrymators). They are not very water soluble,
have low photolysis rates, are very slow to react with hydroxyl radical, and are thermally
decomposed into their precursors, peroxyacyl radicals and nitrogen dioxide. The PANs are
therefore important means of transporting nitrogen dioxide from urban areas on regional and
global scale.

Aerosol Formation
Over the past fifty years, knowledge of the incredibly complex chemistry involved in the
formation of smog has increased significantly. Besides the oxidants formed from the chemical
reactions described in sections above, three other sets of reactions involve chain termination
steps. The first of these is the reaction of hydroxyl radical (OH) with nitrogen dioxide ( NO2 ) to
form nitric acid ( HONO2 ):
OH + NO2 →HONO2 (34)
Nitric acid formed in this way removes both chain-carrying OH and NO2 , which forms ozone
via photolysis. However, nitric acid is very soluble in water and is a strong acid, and it thus
ammonia ( NH3 ) to form ammonium nitrate ( NH4 NO3 ), an important contributor to secondary
aerosol formation in urban and regional settings:
HONO2 + NH3 → NH4 NO3 (35)
The second reaction involving a chain termination step is the reaction of HO2 with itself to form
hydrogen peroxide (H2 O2)
HO2 + HO2 →H2 O2 + O2 (36)
This reaction removes HO2 and therefore is important when NO concentrations are lower.
Reaction 36 can then compete with reaction 19. Hydrogen peroxide is very soluble in water and
can react very rapidly with sulfur dioxide ( SO2 ) in cloud droplets or on wet aerosol surfaces to
form another important strong acid, sulfuric acid ( H2 SO4 ).
H2 O2 + SO2 (aqueous) → H2 SO4 (37)
Sulfuric acid can also react with ammonia to form ammonium bisulfate ( NH4 HSO4 ) and
ammonium sulfate [(NH4)2 SO4] . Sulfuric acid and the associated ammonium salts are
important atmospheric aerosol products. Sulfuric acid is also an important contributor to acidic
precipitation. Larger organic compounds emitted into the air can also be oxidized to form
secondary organic aerosols. Indeed, natural monoterpene hydrocarbons emitted from urban trees
can react with the energy-related pollutants and secondary ozone to form a variety of compounds
found in smog aerosols. Carbonaceous soots can also be emitted directly from automobiles and
diesel engines. These soots contain oily organic coatings and associated polycyclic aromatic
hydrocarbons (PAHs) and have received some attention as air toxics. Road dusts from tire
erosion, road work, and traffic are also a major source of aerosols in many urban areas. Sea-salt
aerosols from surf and wind spray can also add to the aerosol burden near ocean shores. Wind
blown dusts especially in arid (desert) regions can be important and have been found to travel
very large distances including transport across the Atlantic and Pacific Oceans. These
windblown dusts are adding to the background in many cities like Beijing, China and contribute
to the brown clouds in Asia. All of the aerosol species (organic and inorganic) can play
important roles in the radiative balance of the atmosphere on urban, regional, and global scales.

A Methodology for the Validation of Smog Chemistry

Thus far we have presented only a little chemistry and several basic explanations regarding the
smog formation process. Since the actual chemistry of the atmosphere is extremely complex, it
has been necessary for scientists to develop a methodology that provides a progressive
confirmation of our knowledge of atmospheric chemistry as it accumulates. Our presentation
outlines the current knowledge of smog chemistry by following the scientific progression used
for its development and confirmation. When a set of chemical reactions has been determined that
is complete enough to provide a satisfactory description of some overall process, it is commonly
referred to as a chemical mechanism or a kinetics mechanism. The basic function of a kinetic
mechanism in an atmospheric model is to deplete the precursors chemically into intermediate
products that, in turn, generate O3 and other secondary pollutants. The rest of the model
generates the proper precursor levels from a combination of emissions, dispersion, and
deposition inputs, in addition to treating the secondary pollutants via dispersion and deposition.
If the kinetic mechanism is composed of elementary reactions describing all of the important
steps and intermediate species, the mechanism is referred to as an explicit mechanism. (Other
types of chemical mechanisms include empirical, condensed, lumped, and surrogate
mechanisms.) Up to this point, we have discussed the explicit chemistry of formaldehyde and
CO, but have described only the key inorganic reactions.
Smog chamber experiments:
The most important evidence for the validity of explanations and overall processes described
above comes from the thousands of smog chamber experiments performed in the last 30 years.
Until recently, however, wall effects resulting from "dirty" chambers have not been well
characterized; therefore, the validity of smog chamber data has been questionable. A typical
phenomenon that has contributed to this lack of credibility is the generation of significant ozone
levels in an "empty" chamber. Such results can now usually be explained either by trace
quantities of NOx and organics in the air used to fill the chamber, or by desorption from the
chamber's walls. Experiments to quantify the strength of these chamber effects have recently
been developed and utilized. One such chamber characterization experiment involves filling the
chamber only with NO at about 0.5 ppm (µL/L) and noting the conversion rate of NO2. Such an
experiment provides a measure of the combined organic production rate of RO2-type radicals
[which convert NO to NO2 via Reaction (4)] and the radical source strength. If the experiment is
.
repeated with approximately 50 ppm CO, the organic production of RO2 radicals is suppressed,

but the radical production rate is exaggerated. Any OH, radicals would react rapidly with the CO,
. .
leading to HO2 radicals [via Reactions (6) and (7)], which regenerate the OH radicals via the

reaction with NO [Reaction (4). Termination of this chain process is also suppressed to some
.
extent since the NO2 termination reaction with OH [Reaction (5)] would be reduced by the OH

reaction with CO. The CO and NO experiment can generate ozone only if the key element-
hydrogen-containing radicals is present. The observation of ozone, or at least the rapid
conversion of NO to NO2, actually provides a measure of the radical source strength. Data
determined from smog chamber experiments is an essential element in the validation of the
processes involved in smog formation. An emerging development in the methodology of
validation involves computer modeling of such experiments using various kinds of chemical
kinetic mechanisms. We now turn to a description of this methodology and of the hierarchical
concept of species considered by explicit mechanisms used in the modeling of photochemical
smog.
Evaluation of chemical reactions:
The evaluation of chemical reactions is also an important element in the new methodology.
Stratospheric chemistry primarily involves the inorganic reactions of oxygen, nitrogen, chlorine,
and hydrogen atoms. Since the early 1970s, considerable effort has been made to expand, refine,
and evaluate these reactions for use in photochemical models of the stratosphere. As noted
earlier, the development of smog chemistry has benefitted from the availability of this high-
quality fundamental information. A problem currently of concern, the potential loss of the
stratospheric ozone layer, has been studied using computers and sophisticated analytical
equipment. The modeling approach to this problem is similar to that used in the study of smog
chemistry; moreover, a significant part of this evaluated data is important and useful to the
development of smog chemistry. Review panels whose purpose is to determine the best available
chemical reaction rate parameters have not been nearly as extensive for smog models as for
stratospheric models, however. There are currently two active evaluation panels for stratospheric
chemistry, the CODATA Task Group on Chemical Kinetics and the NASA Panel for Data
Evaluation (NASA, 1982). The CODATA group is composed primarily of European scientists,
and the NASA group is composed primarily of scientists in the United States. One evaluation of
smog chemistry was funded by the U.S. Environmental Protection Agency.

Computer modeling with chemical kinetic mechanisms:

Computer algorithms have been developed that use a chemical mechanism and some set of initial
concentrations and physical conditions as input. For smog chamber simulations, there is
CHEMK ; for reactive plumes, there is RPM-II; for simple trajectory models such as EKMA,
there is OZIPM. With these inputs, the computer program can then determine changes in the
concentrations of various species as a function of time by constructing differential equations
based on the chemical mechanism for the species, and by integrating these equations from the
initial conditions. The differential equations for each species are derived from the theory of
chemical kinetics, which essentially states that the instantaneous time derivative for each species
is the difference between the rate at which the species is formed and the rate at which it is
destroyed. Each chemical reaction has a rate associated with it determined from the
concentrations of the reactants and some rate parameter. Thus, the formation rate of some species
is the sum of the rates of all the reactions that generate that species. Similarly, the loss rate of
each species is the sum of the rates of all the reactions in which that species is a reactant. Smog
chamber experiments are commonly modeled using such computer programs. When dispersion,
deposition, and emission inputs have been added, the atmospheric generation of smog can be
modeled using the appropriate chemical mechanism.
The concept of a hierarchy of species :
The development of explicit mechanisms for photochemical oxidant formation has been guided
by the concept of a hierarchy of chemical species. In this conceptualization, each species
involved in the process can be assigned to a hierarchical level on the basis of the number of
HC/NOx systems in which it occurs, with the most ubiquitous species occupying the lowest level.
As an example, consider the reactions of butane and acetaldehyde. Acetaldehyde is formed in
significant quantities during the photochemical oxidation of the butane/NOx systems; the
reactions of acetaldehyde must therefore be included in an explicit mechanism for butane. The
reverse is not true: little or no butane is formed in the photochemical oxidation of acetaldehyde/
NOx systems; as a consequence, butane reactions need not be included in an explicit mechanism
for acetaldehyde. Thus, butane is assigned to a higher hierarchical level than acetaldehyde. The
value of the hierarchical concept lies in the principle it suggests for validating explicit
mechanisms for a series of hydrocarbons, i.e., mechanisms for species should be validated at the
lowest level first. For example, one can begin by validating the explicit formaldehyde
mechanisms using data from formaldehyde/NOx smog chamber experiments. An explicit
acetaldehyde mechanism that decays to formaldehyde can then be constructed by adding a few
reactions and rate constants to the explicit formaldehyde mechanism. In validating the
acetaldehyde mechanism (using data from acetaldehyde/NOx experiments), only the additions
need be examined-the reactions and rate constants from the validated explicit formaldehyde
mechanism should not be modified in any way. Mechanisms for the hydrocarbons can be
constructed by adding the appropriate reactions and rate constants to mechanisms for these
aldehydes or for other oxygenated organic intermediates. Validating mechanisms in the stepwise
order suggested by the use of hierarchical levels has two primary advantages: it clarifies the
sources of uncertainty in simulations, and it reduces the probability that a complex mechanism
for a large hydrocarbon contains compensating errors. In a series of mechanisms that have been
validated in this manner, if the formaldehyde mechanism provides simulations that agree well
with measurements, but the acetaldehyde mechanism does not, those portions of the
acetaldehyde mechanism that might cause the poor correlation should be apparent. The following
paragraphs briefly discuss several levels of the hierarchical scheme that are currently used to
study smog chemistry.
Inorganic chemistry:
Although about 40 inorganic reactions are included in most explicit chemical mechanisms for
photochemical smog , many of these reactions are less important than the reactions given thus
far. The two reasons for the lack of importance of many inorganic reactions in smog chemistry
are that they either participate in "do-nothing" cycles or their rates are too slow. An example of a
"donothing" cycle is the formation and photolysis of nitrous acid (HONO or HNO2):
.
OH + NO → HONO (38)
 .
HONO + hv → OH + NO (39)

Both reactions are fast, but the pair produce no net effect. However, if some other source of
HONO is available that does not consume a radical, then such a source, followed by the
photolysis step, could be an important source of radicals in the smog chemistry. Recently
observed the buildup of HONO in a polluted atmosphere at night. The formation of HONO in
smog chambers, presumably heterogeneously, has also been observed. The following sequence
of reactions represent another "do-nothing" cycle:

NO2 + O3 → NO3 + O2 (40)

NO3 + NO → NO2 + NO2 (41)
NO2 + hv → NO + O (42)
O + O2 + M → O3 + M (43)

Note that there is no net loss of ozone or nitrogen oxide species as long as the sun is present.
Therefore, the first reactions become fairly important at night. Also, the NO3 radical can react
with NO2 to form N2O5
NO3 + NO2 → N2O5 (44)
N2O5 → NO3 + NO2 (45)
which is a "do-nothing" subcycle of the original NO2 + O3 (etc.) cycle. Of course, if NO3 or N2O5
react outside these cycles, then the original NO3 + O3 reaction becomes more important. Indeed,
NO3 can react with many organics such as olefins or phenols, whereas N2O5 is known to react
with H2O on surfaces leading to nitric acid. However, all of these effects involving NO3 do not
substantially alter the process of smog production, even though the NO3-related chemistry can
play a measurable role.
 Fig.9 . Schematic diagram of hierarchical levels of species in Photochemical Smog.
Formaldehyde chemistry:
Formaldehyde represents one of the simplest precursors of photochemical smog and appears to
be an intermediate product in the oxidation of almost all organics. Current explicit mechanisms
contain four reactions involving formaldehyde: the reactions with hydroxyl, hydroperoxyl, and
oxygen atoms, plus photolysis. The main decay pathway for formaldehyde in typical urban
atmospheres is the reaction with hydroxyl. However, the photolysis of formaldehyde is much
more important to overall smog chemistry because a major photolysis pathway involves the
generation of radicals, which can drive the overall oxidation of all organics. The current status of
formaldehyde chemistry can be summarized by saying that several dozen formaldehyde smog
chamber experiments can be simulated with the current standard formaldehyde mechanism.
However, some of the simulations must invoke a means of simulating the loss and reappearance
of formaldehyde to and from the condensed water (if any) on the chamber walls. In other words,
formaldehyde chemistry seems well established, but the validation experiments are troubled by
physical problems associated with the handling of formaldehyde itself.

Acetaldehyde chemistry:
Acetaldehyde chemistry brings the chemistry of PAN and methylperoxy radicals into the
formaldehyde and NOx chemistry. Methylglyoxyl and biacetyl also require consideration of the
chemistry of PAN and methylperoxy radicals, but a range of experimental conditions with which
to test their chemistry is not as readily available as is the case for acetaldehyde.
Paraffin chemistry:
The simplest paraffin is methane (CH4) and it undergoes the smog oxidation cycle very slowly,
since the rate constant for hydroxyl attack is 2 orders of magnitude lower than for most other
paraffins:
. .
CH4 + OH → CH3 + H2O (46)

. .
CH3 + O2 → CH3O2 (47)

. .
CH3O2 + NO → CH3O + NO2 (48)

. .
CH3O + O2 → H2CO + HO2 (49)

. .
HO2 + NO → NO2 + OH (50)

Note that the only new reaction added to the NOx-formaldehyde-acetaldehyde hierarchical
scheme is the first reaction. Although the only significant initial pathway for higher paraffins is
hydroxyl attack, this simple beginning leads to a complicated series of chemical reactions
involving the intermediate species that are formed. In recent years, the development of explicit
kinetic mechanisms for paraffins has concentrated on the chemistry of the intermediate products
such as the alkoxyl radicals (RO), alkylperoxy radicals (RO2), aldehydes (RCHO), and ketones
(R2CO). Several years ago the isomerization reactions whereby a hydrogen atom is transferred
from one end of an alkoxyl radical to the other were discovered. Unfortunately, there has been
little research to date elucidating the chemistry of the many hydroxysubstituted compounds
generated from these isomerizations. Studies that measure the photolysis rates and hydroxyl
attack rates of the hydroxy-carbonyl intermediates have only recently begun. Hence, much of the
modeling of smog chamber data has utilized speculative values for these rate and photolytic
constants. Recent developments concerning paraffin chemistry are contained in studies of alkoxy
radicals. In this last study, a new methodology produced reasonably accurate measurements for
the reactions of the methoxy and ethoxy radicals with oxygen. The new rates are much higher
than previous estimates, the new data have been extended to update the thermochemical kinetic
estimates for alkoxy radicals as large as four carbons. These new estimates support previous
adjustments to obtain closer agreement with observations for smog chamber simulations of
butane experiments.

Olefin chemistry:
Hydrocarbons with a double carboncarbon bond (e.g., olefins) are very reactive in the
atmosphere. More smog chamber experiments have been performed using olefins, particularly
propene, as the sole hydrocarbon than using any other type of compound. The double bond reacts
not only with hydroxyl radicals (OH), but with oxygen atoms, ozone, and the nitrate radical
(NO3) as well. Also, the reactions of olefins tend to be different than those of other
hydrocarbons. For instance, the hydroxyl radical adds to the olefin rather than abstracting a
hydrogen atom, as is the case in Reaction (5) for formaldehyde. Moreover, the olefins tend to
produce two carbonyls per initial reaction rather than the one carbonyl produced from most other
classes of hydrocarbon. For example, the main sequence of reactions, starting with hydroxyl
attack on ethene, is as follows:
Aromatics chemistry:
Toluene, Any discussion of the photooxidation of aromatic hydrocarbons should begin with
toluene, since that compound has received the most attention to date. Toluene accounts for the
highest fraction of aromatics in urban atmospheres. There are three current descriptions of
toluene oxidation. All three descriptions are similiar in their discussion of the ring opening
mechanism, and all derive from the experimental work. However, the mechanistic product yields
differ and the three descriptions have different treatments of radical and nitrogen balance.
Toluene oxidation differs from olefin and paraffin oxidation in several important ways. Three
major differences between the photooxidation of toluene and that of a compound such as
propylene have been observed:
(1) A high photolysis rate for toluene oxidation products. Toluene oxidation products photolyze
at a rate twice that resulting from a 100% yield of formaldehyde from toluene decay. This rate is
probably caused by the formation of methyl glyoxal, which photolyzes at a rate roughly 15 times
that of formaldehyde.
(2) A low rate of peroxyl radical production. The inefficiency of toluene and other aromatic
hydrocarbons in effecting NO-to-NO2 conversions has been observed by other investigators.
Empirically, methyl glyoxal photolysis alone is nearly sufficient to explain the number of NO-to-
NO2 conversions observed in toluene oxidation. Thus, either the other products of toluene decay
are unreactive or there is a mechanism in toluene oxidation that destroys peroxyl or otherwise
prevents the peroxyl radicals from reacting with NO.
(3) A powerful NOx sink mechanism. This seems to occur after the onset of ozone production
and does not seem to consume hydrogen-containing radicals. This sink probably involves NO3
and may result in dinitrate compounds or nitrophenols.
Aromatics chemistry: m-xylene. The next most important aromatics molecule in urban
atmospheres presently appears to be m-xylene. In fact, a recent detailed analysis of Los Angeles
air indicates that the two most important molecules for smog production are first toluene and
then m-xylene . This ordering is quite different from that of a similar detailed analysis reported
in the early 1970s; this analysis showed ethene and isopentane to be the two most important
molecules in Los Angeles smog. Although explicit mechanisms for m-xylene have not yet been
published, available evidence indicates that important mechanistic differences between m-xylene
 .
and toluene chemistry exist beyond the factor of 3 increase in OH attack. Notably, the radical

source strength is also higher for m-xylene than for toluene. This higher radical source strength is
consistent with more methyl glyoxal formation from the ring-opened m-xylene, since it has two
methyl groups rather than the one methyl group it has in toluene. The NOx sink strength for m-
xylene is, by contrast, somewhat less than for toluene, suggesting a lower yield of phenolic
compounds in the m-xylene system. Also, the yield of PAN is higher for m-xylene than for
toluene, which is again consistent with the extra methyl group. Preliminary smog chamber
experiments comparing m-xylene with an urban mix by partial substitution (20% carbon) show
an exceptionally high reactivity for m-xylene. Many other hydrocarbons show little change
between full urban-mix experiments and the substitution experiments, e.g., toluene shows only a
small effect, as does the natural hydrocarbon a α-pinene.

Description and Comparison of Chemical Mechanisms

The use of photochemical models is currently a standard tool for regulatory analysis. However,
photochemical models are not all the same. Different kinetic mechanisms give different results
under similar conditions, though the meaning of "similar" is often hard to define. Each type of
mechanism has a different methodology for speciation of hydrocarbons (HCs), averaging of
ensemble rate constants, and selection of stoichiometric parameters. To be suitable for use in an
atmospheric model, any overall explicit kinetic mechanism must be condensed in some way.
There are several reasons for this requirement:
(1) Explicit mechanisms do not yet exist for all the organics observed in the atmosphere (to
ignore some species is a form of condensation).
(2) No computers either now in existance or planned for the future can reasonably treat the full
spectrum of reactions if, indeed, all the reactions were available. This is especially true if the
model is to treat transport and dispersion accurately.
(3) Several condensed mechanisms have been shown to treat realistic smog chamber experiments
and some actual atmospheric situations adequately.
(4) Condensed mechanisms are easier to understand in relationship to key process factors such as
radical sources and sinks (tractability is an asset to reliable applications of complex models).
Current condensed photochemical mechanisms:
Photochemical kinetic mechanisms intended for atmospheric application can be classified into
three types: surrogate mechanisms, lumped-molecule mechanisms, and lumped-structure
mechanisms (Table 4). Each of these kinds of mechanism differs in the nature of the
simplifications (i.e., condensations) it uses to make the complex reaction sequences and the
multitude of organic compounds involved in urban smog more tractable; however, all condensed
mechanisms tend to be the same in their explicit treatment of basic inorganic chemistry.

Table.4 .Available Smog Mechanisms

Surrogate mechanisms:
Surrogate mechanisms are perhaps the easiest to understand. For laboratory smog chamber
experiments involving individual HC species, mechanisms are available that explicitly treat
every compound with a significant role in the chemical process. In the surrogate mechanism
approach, the explicit mechanisms for one or more HCs are added together. The complex urban
mix is then assumed to be represented by some blend of the HCs treated by the explicit
chemistry. The obvious difficulty with this approach is the specification of the surrogate blend.
The mechanism used in the present form of the empirical kinetic modeling approach (EKMA) is
a surrogate mechanism that uses a 25% propane/75% butane mix to simulate automobile exhaust.
The appropriate mix for nonautomotive emissions has not been established. One possible method
for adapting the surrogate mechanism approach to HC mixes other than automobile exhaust has
been offered by the users of the Lawrence Livermore grid model (known as LIRAQ). In the
LIRAQ emission inventory, each individual HC species is represented by "propane/butane
equivalents" the amount of propane and butane that might represent a kilogram of HC emissions,
However, this procedure has not been verified.
Lumped-molecule mechanisms:
The lumped-molecule approach to kinetic modeling is a widely used method of simplifying the
smog chemistry problem. Lumped-molecule mechanisms are based on the approach, in which
molecular species are classified as olefins, paraffins, aldehydes and usually, aromatics. The rate
constant for each generalized HC class should, in theory, be an appropriate average rate of the
individual compounds contained in each class. However, in practice, lumped-molecule
mechanisms usually specify some default value that may or may not approximate the ensemble
average. In general, lumped-molecule mechanisms require noninteger stoichiometric parameters
to describe reaction pathways for the lumped molecular species and the lumped intermediates
generated by the chemistry. Choice of the proper values for the stoichiometric parameters poses
a problem that is initially related to the averaging of rate constants just described; the
stoichiometric parameters usually relate to the average size and distribution of the lumped initial
species. However, the stoichiometric parameters in lumped-molecule mechanisms may also
affect both reactivity and carbon balance. Specification of these parameters to satisfy all of these
constraints simultaneously can be difficult.
Lumped-structure mechanisms:
The lumped-structure approach is an attempt to decouple reactivity and carbon conservation.
The concept is similar to that used by Benson (1968), who calculated molecular reactivity from
submolecular components. Thus, in the lumpedstructure approach, carbon structures within the
HC molecule are the lumping category. In an olefin molecule, for example, the chemistry of the
olefinic group is treated independently from any paraffinic side chains. The lumped-structure
approach offers several advantages over the lumped-molecule approach. Because similar
structures have similar reaction rate constants, the averaging problem is minimized. Carbon
conservation can be expressly maintained, and carbon balance during the simulation can even be
monitored. The principal disadvantage is that intramolecular processes such as decomposition
and side-chain activation may require adept treatment. Also, because the lumped-structure
approach is often less intuitive than the lumped-molecule approach, the documentation
requirements may be higher.
Reactivity considerations in chemical mechanisms:
Given our previous basic description of smog chemistry, we can identify three factors that
influence the production of O3. The first is the radical sources needed to maintain the
.
concentration of hydroxyl (OH ) radicals required to initiate the HC oxidation process. The

second is a concentration and species mix of HCs; the reaction of OH with HCs produces the
peroxy radicals that enhance the oxidation of NO to NO2. The final factor is the presence of NOx,
which reacts with peroxy radicals and generates O3. The first two factors-radical sources and HC
concentrations and species mix-comprise what is generally referred to as "reactivity." The effects
of NOx are more complex. Although it is true that NOx is necessary for the generation of O3,
NO2 also serves as a principal radical sink in smog chemistry. Thus, as noted in the discussion of
the EKMA isopleths, increasing NOx serves to reduce the radical concentration, to reduce the
rate of HC oxidation, and thus to slow the production of O3. Given a particular level of HC,
increasing NOx slows the O3 formation rate, but increases the peak O3 until a particular HC:NOx
ratio is reached (the "ridge line" ratio on an isopleth diagram). Additional increases in NOx levels
further slow O3 production and reduce peak O3, though this reduction depends greatly on factors
such as total allowed reaction time, dilution rate, and heterogeneous 03 destruction. Because the
basic reactions of smog formation have been studied intensively in laboratory experiments, most
current chemical kinetic mechanisms treat NOx chemistry in a similar fashion. The treatment of
N2O5 chemistry is an area of uncertainty that has led to minor differences among mechanisms;
the major differences, however, are found in the treatment of HC oxidation behavior. The
oxidation reactivity of the precursor HC mix appears to be similar in most current mechanisms;
the principal difference occurs in the HC lumping procedure. For example, many mechanisms
lump ethane along with the other olefins, yet ethene has an OH reaction rate constant of less than
one-third that of propene. If the propene rate constant is nevertheless used for all olefins, the
reaction chemistry of olefins may be dominated by an erroneously high value for all ethene-OH
reactions. Since ethene typically comprises 50% of olefins, the error in overall mechanism
performance may be large. The effect of the error is too much reactivity early in the day and too
little reactivity later in the day. A more important difference in mechanism behavior results from
assumptions concerning product yields. Subsequent to the initial reaction step, HC oxidation is
dominated by the reactions of oxygenated products:
. .
HC + OH → RO2

. .
RO2 + NO → NO2 + RO

. .
RO → HO2 + aldehyde

. .
Aldehyde + OH → RCO3

. .
RCO3 + NO + O2 → NO2 + CO2 + R’O2

The R'O2 at the end of this cycle is not the same as the initial RO2; it contains one less carbon
atom. Lumped mechanisms tend to treat all RO2 radicals as being similar. To terminate the
reaction chain, some kind of chain-length parameter is required. Aldehydes react rapidly with
OH, and they also photolyze to generate radicals. After a period of time, the balance of aldehyde
production and aldehyde loss will dominate the behavior of a chemical kinetic mechanism.
Current mechanisms are sufficiently dissimilar in their treatment of aldehyde formation and loss
that disparities in calculations caused by this feature will become significant in less than one day
of simulation time.