EPA/600/A-93/105
93-TP-58.05
Source Apportionment of Fine and Coarse Particles
in Southern Ontario, Canada
Teri L. Conner and John L. Miller"
U.S. Environmental Protection Agency, MD-47,
Research Triangle Park, NC, 27711
Robert D. Willis and Robert B. Kellogg
ManTech Environmental Technology, Inc., P.O. Box 12313,
Research Triangle Park, NC, 27709
Thomas F. Dann
Environment Canada, River Road Environmental Technology Centre
Ottawa, Ontario, K1A OH3, Canada
DISCLAIMER
The information in this document has been funded in part by the U.S.
Environmental Protection Agency under interagency agreement number RWCNS3494S-
01-1 with Environment Canada. It has been subjected to Agency review and
approved for publication. Mention of trade names or commercial products does
not constitute endorsement or recommendation for use.
'Senior Environmental Employment Program

-------
93-TP-58.05
IJJTRODUCTIOJJ
Environment Canada/ in cooperation with the Ontario Ministry of
Environment and the Walpole Island Indian Band, has been conducting an air
monitoring study in a region of southern Ontario near Detroit over concerns
about trans-boundary transport of pollutants. Two sampling sites are located
in the city of Windsor, Ontario. One Windsor sampling site (WIN1) is
centrally located in the city of Windsor at the Ontario Ministry of the
Environment monitoring site at 467 University Avenue. This site is only 6 km
south of a large municipal waste incinerator operated by the Greater Detroit
Resource Recovery Authority (GDKRA) and is also close to other point sources
in Detroit. The second Windsor sampling site (WIN2) is located less than 5 km
to the southwest of WIN1. WIN2 is closer to the high density of point sources
in south Detroit. The Windsor sites are frequently downwind of the numerous
emission sources of the greater Detroit area, which include coke ovens, iron
and steel industry, incinerators, power generation facilities, lime and cement
operations, and automotive assembly plants. The Windsor sites are also
influenced by the regional background of secondary sulfate common in the
eastern U.S. and Canada, as well as by automobile emissions. A third site was
located at Walpole Island, about 55 km to the northeast of the WIN1 site in a
rural area. This site (WAL) was chosen to represent background conditions,
although this site is also influenced to some degree by primary industrial
emissions and secondary pollutants. Locations of sampling sites are shown in
Figure 1.
Chemical mass balance source apportionment of fine and coarse particles
will be applied to X-ray fluorescence (XRFj" data. Mete'orolo~gicai observations
and individual particle morphology and composition will be-used to interpret
the results.
EXPERIMENTAL
Samples for this analysis were collected during the period from January
30, 19S1 - November 26, 1991. Sampling at each site took place for 24 hours
from midnight to midnight every 6 days with PK-10 dichotomous samplers having
nominal flow rates of 16.7 lpm. Samplers had a cutpoint of 2.5 ^im to separate
the coarse and fine particles, which were collected on 37-crr. diameter Teflon
filters. Fine and coarse particle mass concentrations were determined
gravinietrically. Elemental concentrations were determined by energy-
dispersive XRF at the U. S. EPA, Research Triangle Park facility.
A subset of the samples was selected for analysis by scanning electron
microscopy combined with energy-dispersive XW (SEM/EDX). Morphological
features of the particles combined with chemical data have been shown to be
useful in resolving source types which cannot be resolved by conventional
means.
WtTpcr
Meteorological data	obtained from the Windsor Airport and from a
portable iceteorological station at Walpole Inland. Average wind speed and
prevailing wind direction during each sampling period are used to help
interpret particulate concentrations sseasured.
Locations and emissions of major point sources in the Detroit-
metropolitan area (Figure 1) were obtained from the U. S. EPA Region 5 office.
2

-------
93-TP-58.05
DATA EOTMARY
©snsral
Mass and elemental concentrations measured for coarse and fine particles
at each site are sumnarized in Table I. Only those species whose
concentrations are at least twice their uncertainty for at least two sets of
measurements (either coarse or fine) are reported. In general, concentrations
of industrial metals and chlorine (Cl) were higher at the Windsor sites than
at the background Walpole site. The particularly high concentration of Cl in
the coarse fraction at WIN2 reflects the proximity of WIN2 to the Canada Salt
Corporation in Windsor. The tin (Sn) values should be regarded with caution
because of the high and variable Sn background observed during the analysis of
these samples. Sulfur is the dominant species measured in the fine fraction.
It should be noted, however, that XRF does not measure the organic carbon,
elemental carbon, or water vapor.
Ccapariscn of Elemental Ratios for Soil-Related Elements
The ratios of soil-related elements (Al, K, Ca, Fe) to Si in the fine
and coarse fractions from each sampling site are compared with their ratios in
crustal limestone and shale profiles from the U. S. EPA's Volatile Organic
Compound (VOC)/Particulate Hatter (PM) Speciation Data System1 in Table II.
Such a comparison reveals whether these elements have non-crustal sources.
The K/Si ratio in the coarse fraction is relatively constant across the
sampling sites with low variability and is very close to the crustal shale
value. It is assumed that the dominant source of Si in both the fine and
coarse fraction is soil as represented by crustal shale; therefore, soil is
likely the dominant source of K and Si in the coarse fraction. In contrast,
the K/Si ratio in the fine fraction is quite variable at each site and from
site to site, and is much higher than the crustal shale ratio. This indicates
a large, non-soil contribution to the fine K.
The Al/Si ratio in the coarse fraction has the lowest variability and is
most similar to the crustal value at the WIN2 site. At the WAL site, coarse
Al/Si is only a little higher than the crustal value, but is quite variable.
The Al/Si ratio at the WIN1 site is much higher than the crustal value and is
quite variable. Environment Canada has found high Al values in the coarse
fraction at a number of other sampling sites and have attributed this to wear
of the dichotonsous sampler inlets as a result of leak checking. The Al/Si
ratios in the fine fraction are more difficult to interpret because of the
large number of fine Al values near the detection limit.
The Ca/Si ratio in the coarse and fine fractions is quite variable at
the Windsor sites and is higher than at the Walpole site. All Ca/Si ratios
are higher than the crustal shale values. There are several limestone/ceicent
operations within 15 km of the Windsor sampling sites. Fugitive dusts from
these operations are likely reaching the Windsor sites and influencing
concentrations there. Some of these dusts are probably reaching the Walpole
site and influencing concentrations there, although to a lesser extent.
The same pattern observed for Ca/Si ratios is evident for the Fe/Si
ratios in the coarse and fine fractions, reflecting the close proximity of the
Windsor sites to the steel-related industries of south Detroit. At WAL, the
Fe/Si ratio is close to that of crustal shale and the variability is only 28%
(lower than at the Windsor sites). There may be little coarse Fe from the
steel-related industries reaching WAL, as opposed to fine Fe, which may travel
3

-------
93-TP-58.05
farthsr than the coarse Fe. There is evidence of sources of Fe other than
soil at WAL, but the evidence is even stronger at WIN1 and WIN2.
Fins- and Coarse-Particle Concentrations as functions of travailing Wind
Direction and Wind Epssd
Fine- and coarse-particle concentrations were examined to determine any
relationships with prevailing wind direction or wind speed. Such an
examination can offer clues to the sources of pollutants. For the fine
fraction, the highest mass concentrations were observed for winds from the
south to west corridor, except for some low concentrations when wind speeds
were very high. The lowest concentrations occurred when winds had a northerly
or easterly component or when wind speeds were high. For the coarse fraction,
one would expect an increase in coarse-particle concentrations with an
increase in wind speed as a result of soil and other dust being entrained in
the air. However, at all sampling sites, the highest coarse-particle loadings
were not associated with high wind speeds but"were loosely associated with
winds from the southwest. This was also true for coarse particle elements
associated with the steel industry (e.g., Fe, Hn and Zn).
Correlations Aaong Zhs Eleasnts
Correlations among the elements were examined to identify clusters of
mutually correlated elements which may offer clues to their sources. In the
fine fraction, three main clusters of elements were observed for WIN2. These
are identified with coal combustion (S, Hn, Se, Mass), steel-related
industries and manufacturing (Fe, Hn, Zn, Se), and soil and other dust sources
(Ca, Si). At WIN1, similar fine-element clusters were observed, with an
additional cluster consisting of V and Ni (attributed to oil combustion). At
WAL, correlations reveal clusters identified with coal combustion (S, Se,
Mass), soil and other dust (Ca, Si, Fe) and mixed industrial sources (K, Br,
Cu, Sn, Zn). The fact that Fe appears in the soil-related cluster suggests
that Fe from industrial sources is of lesser importance at WAL. The steel-
related cluster was much weaker at WAL.
The coarse-particle element correlations show less distinct clusters
than was found for fine particles. In general, two main clusters are found at
each site - one associated with soil and the other with steel-related
industries. The latter was more prominent at the Windsor sites.
CESaiCAL MASS BAIAMC2 SOUXCS APPORTIONMENT
The U.S. EPA/DRI Chemical Mass Balance Model2, version 7 was used to
quantitatively apportion chemical species measured at the sampling sites to
the major sources contributing to the particulate mass (fine and coarse) at
those sites. The chemical mass balance (CHB) model consists of an effective
variance least squares solution to a set of linear equations which express
each measured chemical species concentration as a linear sum of the
contributions of each source to the chemical species. The effective variance
solution gives the most weight to source or ambient measurements with the
lowest uncertainty estimates. Source contributions are expressed as the
product of the abundance of the species as eseitted by the source and the total
mass concentration contributed by the source. The set of abundances of all
species as emitted by each source represents the "source profile" or "source
fingerprint". In practice it is not possible to apportion mass to each
individual contributing source. Individual sources may be too similar to one
another, too numerous, or may not contribute significantly to the total mass
loading. Sources are generally grouped together to represent a single "source
4
\

-------
93-TP-58.05
category" or "source type". For example, there are many incinerators in the
Detroit airshed, but they will be considered together as one incineration
source type and represented by a single profile.
In performing a chemical mass balance, it is assumed that: 1) the
abundance of each species used in the fitting procedure is known for each
source type, and 2) all major sources of each species used in the fitting
procedure must be included in the CHB. Other assumptions made for the CHS
model are listed in Reference 2.
One of the performance goals of the CHB model is to account for all of
the mass, within uncertainties of the measurements involved. A complete
apportionment of the PH10 mass measured in this study is made more difficult
because no analyses were performed for elemental and organic carbon or for
nitrate, all of which could contribute significantly to the PM10 mass. The
lack of data for the carbonaceous species affects both the coarse and fine
particle fractions. Biological materials (pollens, spores, plant debris) may
comprise a significant portion of the coarse particle mass. Organic and
elemental carbon are important components of motor vehicle particulate
emissions, especially for diesel vehicles. Apportionment of vehicle
particulate emissions presents an intractable problem, not only because of the
absence of ambient carbonaceous data, but because of the phasing out of
tetraethyl lead from the fuel supply. Lead was previously relied upon as a
tracer for motor vehicle particulate emissions.
Wind Sector Analysis
In applying a CHB analysis to the data, it is assumed that the source
compositions remain constant throughout the sampling period and that the same
source profiles apply to all sampling periods. In practice, source
compositions may vary over time and space due to changing operating
conditions, fuel compositions, raw materials, or meteorological conditions.
Chemical compositions may vary at a single source or at many sources within a
single source category. To compensate for the variability of sources, samples
were averaged together to get a study average for each sampling site. For
WIN1, two samples were excluded from the average because of elemental
outliers. To get more information on the change of source impacts with
changing wind direction, samples from adjacent and/or within wind sectors were
averaged together according to the density or proximity of upwind point
sources and the similarity of samples from adjacent wind sectors, as
appropriate for each sampling site. Samples were averaged together as
indicated in Table III. Prevailing winds from the Windsor City Airport were
applied to the Windsor sampling sites. Wind data applied to the Walpole
samples were measured at the site. These averages were computed for both the
fine and coarse fractions. The average values for fine and coarse mass are
shown in the table.
Eciirco Profiles
A combination of profiles available in the literature and in the U.S.
EPA Speciation Data System were used to predict ambient species
concentrations. Source selection was based on preliminary analyses of the
ambient data (wind sector analyses, correlations, comparisons with natural
crustal composition) as well as a review of emissions inventories for the
Detroit metropolitan area and consideration of the proximity of sources to the
sampling sites. Steel manufacturing and related operations dominate the S.E.
Detroit area stationary emissions. These operations include limestone
processing, coke ovens, steel manufacture blast furnace, and power generation
5

-------
93-TP-58.05
facilities. An emissions inventory was not available for the Windsor area.
Area sources include motor vehicle emissions, wind-blown crustal material,
biological particles, and secondary sulfate from power generation.
Coarse fraction. Preliminary analyses indicate that fugitive dust from
steel-related industries could be important contributors to the ambient coarse
particle loading, especially at the Windsor sampling sites. An appropriate
profile for this source-type was obtained from a study conducted in the S.E.
Chicago area3. In this work, profiles were developed for samples taken from
haul roadways surrounding coal yards, coke yards, and steel yards. The steel
yard road (STL) profile was selected to represent these fugitive emissions in
the Detroit area. The coal- and coke-yard profiles are dominated by carbon
emissions. Subsequent receptor modeling work on S.E. Chicago samples4 found
coal-yard road dust to contribute significantly to both the coarse and fine
fractions, but this source-type is comprised largely of carbon, which was not
measured in the ambient data. (Attempts at including coke or coal fugitive
dust profiles in the CHB resulted in very poor fits of the data). Crustal
shale (SHALE) and crustal limestone (LIHE) profiles (numbers 43305 and 43304
in the U.S. EPA's Speciation Data System) were chosen to represent the
resuspended soil in the airshed. The need for crustal limestone profile is
made apparent in the comparison of Ca/Si ratios in the ambient data with the
Ca/Si ratio in the crustal shale profile. Lime is used in the steel
manufacturing process, and there are several cement or lime operations listed
in the emissions inventory for the S.E. Detroit area. The Canada Salt
Corporation is located in close proximity to WIN2. High coarse-particle
chlorine concentrations, together with SEK micrographs, have confirmed the
impact of this source on WIN2. A pure NaCl (SALT) profile was therefore
included in the CHB analysis of the coarse fraction.
A preliminary CHB analysis of average WIN1 and WIN2 samples indicated
that Zn, CI, K, Cr, Ti, and Ni were not being predicted well by STL, SHALE,
LIKE (and SALT for WIN2) profiles in the apportionment. This combination of
underpredicted species indicates that incineration may also contribute to the
coarse particle loading. Accordingly, an incineration (INCIN) profile (#17105
from the U.S. EPA's Speciation Data System) was included in the analysis. SEM
analysis indicates evidence of incineration-derived particles in the coarse
fraction.
Fine Fraction. Up to 4 profiles were used to reconstruct the fine
particle data. The single most important measured constituent of fine
particle mass is sulfur. Previous studies5"7 have shown sulfur to exist in the
form of sulfate plus associated cations ranging from H+ to NK4+. A single
constituent source profile representing sulfur arbitrarily as asmonium sulfate
was included to account for a large portion of the fine particle mass. (This
will give an approximately 25% higher estimate for sulfate plus cation than if
sulfur is represented as sulfuric acid). While this procedure doss account
for the secondary sulfate, and thus a large portion of the mass, it does not
yield any information on the specific source types contributing to the
secondary sulfate. A stesl-yard road fugitive dust profile for the fine
fraction* was used to represent the steel industry emissions. Primary stack
emissions from the steel-making process are also potential contributors to
fine particle mass, but no satisfactory profiles exist. Crustal shale was
used to apportion resuspended soil in the fine fraction. This profile was
applied successfully to the coarse fraction. Crustal limestone was suitable
for the fine fraction. Source characterization data measured at a
Philadelphia municipal solid waste incinerator (profile #17105) was used to
apportion all incineration emissions in the airshed. While incineration
emissions are not expected to make a large contribution to the fine particle
6

-------
93-TP-58.05
mass, they could contribute significantly to certain toxic elements, such as
Ni and Cr. The chemical profile of a Philadelphia oil power plant was used to
apportion emissions from oil-fired boilers in the airshed which are used for
municipal and industrial power generation in the airshed. Again oil
combustion is not expected to contribute significantly to the fine particle
mass but may contribute to toxic metal loadings.
C3£3 Rssults
Coarse Traction. Results for the CHB analysis of the coarse particle
fraction study averages and wind sector averages are summarized in Table IV.
Fitting species were tailored to each sampling site based on source profiles
included and which elements were wsll-predicted by those sources.
On average between 40% and 60% of the coarse fraction was predicted for
the WAL samples. As indicated by the preliminary analyses, the WAL coarse
fraction is dominated by soil-derived materials. LIKE and SHALE together, on
average accounted for more than half of the coarse particle mass, and the
ratio of LIHE to SHALE was around 0.3 on average. SEM analysis of selected
filters revealed evidence of biological particles such as plant debris and
pollen, which would account for some of the missing mass.
On average between 55% and 70% of the coarse fraction was predicted for
the Windsor sites. At these sites, the contribution of steel yard fugitive
dust (STL) was also important. The percent contribution of STL on average was
13% at WIN1 and 1S% at WIN2, consistent with the close proximity of those
types of sources, especialy to WIN2. At both sites, the STL percent
contribution was highest for wind sectors with a westerly component and falls
off rapidly for wind sectors with a northerly or easterly component. The
INCIN profile was able to explain most of the CI and Zn and some of the K.
Also, the small amount of coarse S which was explainable was associated with
the INCIN source. SEM analysis supports this observation. At WIN2, the
highest INCIN estimates were highest for the N-NNE wind sector (7.6 ± 1.8%) .
and W sector (7.9 ± 1.8%), and lowest for the NE-ESE wind sector (3.7 ± 1.0%).
A similar pattern is observed for WIN1. The largest-capacity (83,000 kg/hr)
incinerator in the area is the GDRRA refuse-derived fuel facility, located
less than 10km north of WIN1 and north-northeast of WIN2. The Central Wayne
County Sanitation Authority municipal solid waste incinerator (18,900 kg/hr)
is located less than 20 km west of WIN2. There are other incinerators in the
area, but most have a much smaller capacity than those specified here. The
ratio of LIHE to SHALE was higher at the Windsor sites. At WIN2 the ratio was
less than 1 for H-ESE winds and was 1.5 or higher for W and WSW winds. The
differences are not statistically significant, but there is a trend which
points to the S.E. Detroit area as the origin of much of the fugitive
limestone dust. At WIK2, salt is a minor contributor to coarse mass but a
«eajor contributor to coarse chlorine, which was very high at this site.
Fine Traction. Results for the CHS analysis of the fins particle
fraction study averages and wind sector averages are summarized in Table V.
Results are similar at 3 sites with the exception that a crustal component was
not a significant contributor to the fine mass at the Windsor sites and was
therefore excluded from the final CHB apportionments. Fitting species were
tailored to each sampling site based on source profiles included and which
elements were wsll-predicted by those sources.
On average, only 56% to 62% of the fine mass was predicted. Reasons for
the inability to apportion all of the fine mass include the lack of carbon
measurements, which could be a major constituent of missing sources such as
7

-------
93-TP-58.05
vehicle particulate emissions or fugitive coal or coke dust. In addition,
secondary sulfur has been represented as drv arrooniura sulfate. Any sulfates
present most likely have some amount of water associated with them which could
substantially increase the mass. The amount of the increase depends on the
form of the sulfate and the relative humidity history5'®. SEM results (see
below) indicate that sulfur exists in the form of droplets rather than
crystals, suggesting the presence of water in association with fine sulfur.
There is also some microscopic evidence suggesting some possible hygroscopic
minerals.
Sulfur, as represented by dry ammonium sulfate, dominated the fine
particle mass, as expected. Its contribution ranged from 46% to 52% of the
fine particle mass. The steel yard fugitive dust component increased from
just 2.0% at WAL to 7.8% at WIN1 and 13.8% at WIN2. This trend is consistent
with the close proximity of WIH2 to the steel industry activities. The
incineration contribution estimates likewise increased from 0.9% at WAL to
3.1% at WIN1 to 4.3% at WIN2. Oil combustion was estimated to make a minor
contribution to the fine particle loading at each site.
At the Windsor sites, Hn was consistently underpredicted and Fe
overpredicted by the STL profile, resulting in an increased Chi2 (i.e., a
poorer fit to the data). This phenomenon is more pronounced at WIN2. The
Fe/Mn ratio in the STL fine fraction profile does not represent the ambient
data well. In addition, Ca is significantly overpredicted by the STL profile,
indicating a problem with its abundance in the profile. Local steel yard
samples, as well as steel manufacturing stack emissions, should be collected
and analyzed to determine the best steel-related profiles for the airshed.
At all sites, CI was significantly overpredicted. We frequently see a
loss of CI over time after sample collection. This is presumed to be due to
on-going reactions with atmospheric pollutants or volatilization.
SCAHHIKG ELECTRON MICROSCOPY AHD X-RAY FLUORESCENCE ANALYSIS
A subset of the samples was selected for analysis by scanning electron
microscopy combined with energy-dispersive x-ray analysis (SEH/EDX). Particle
morphology combined with chemical data is useful in resolving source types
which cannot be resolved by conventional means9. SEH may also be used to
qualitatively infer the presence of species not measured by XRF (e.g., water
associated with sulfates, and soot and organic carbonaceous particles,
including pollens and spores).
The five samples selected for SEM/EDX analysis are listed in Table VI,
along with their measured coarse mass concentration and the prevailing wind
conditions during sampling. All samples examined were collected on Teflon
filters; thus, there is seme interference from the filter matrix, especially
for the fine particles. Ideally, coarse-particle samples should be collected
on Huclepore filters for SEH analysis. The coarse fraction was analyzed for
the five samples, with 80 to 160 particles analyzed per sample.
About 10% of the fine particles are collected on the coarse-fraction
filter with the dichotomous sampler, allowing for the analysis of fine
particles on the coarse-fraction filter. Pine particles on one of the filters
(WAL, 7/17/91) were examined. Fine particles identified were either sulfate
"droplets" or "organic plus sulfur" particles.
The coarse fractions of all samples were dominated by minerals, which
typically comprised about 70% bv number of the coarse fraction. Fractions
e

-------
93-TP-58.05
given for each particle type are by number and not pass concentration. Sample
WIK1, 8/10/91 was different from the other samples in that it had the highest
fraction (29%) of organic particles (almost entirely spores or pollens) and
the lowest fraction of minerals (54%). This is the only sample examined which
had a northerly component to the prevailing winds. The mass loading of this
sample was also quite low (6.8 fjg/a3). Aluminosilicate particles (from soil,
clay, road dust) or calcite particles were generally the most abundant class
of particles (about 25% to 35%), but there appear to be real differences in
mineral composition among the five samples. The WIN2 samples had the highest
fraction of calcite particles (about 25%), while the WAL samples had only
about 5% calcite particles. The latter was rich in dolomite relative to the
other samples. These three samples were rich in quartz particles (about 12%).
The 7/17/91 WIN2 sample was unusually rich in 6alt crystals (primarily NaCI
and MgCl). The other WIN2 sample (4/30/91, 33 km/hr winds) and the WAL sample
were lowest in salt particles.
The organic category of particles (soot, plant debris, pollens, spores)
typically represented 10% to 20% of the coarse fraction, with the
aforementioned exception of sample WIN1, 8/10/91. Too few soot particles were
observed to note if there were differences in soot content among the samples.
The fraction of coarse particles from industrial sources ranged from 9%
to 20% for the samples examined. The WIN1 samples may show higher
concentrations of industrial particles than the other three samples, but the
lack of statistics makes it difficult to draw conclusions. All samples had
remarkably few fly ash particles, although the WAL sample appears to have a
higher concentration than the other samples. The 4/30/91 WIN2 sample appears
to have an exceptionally high fraction of iron spheres in the smallest size
category (1.5 - 2.1 #im) as well as several unusual Fe-rich particles assigned
to the "industrial other" class, suggestive of iron foundries and steel
making. The WIN1 samples had several unusual particles assigned to the "Kg-
Cl-Ca-S" class. Other particles rich in some subset of these elements were
also found on these samples. The "Hg-Cl-Ca-S" particles were classified as
industrial based on their "processed" appearance: rounded and smooth as
opposed to rough and crystalline, and sometimes almost a wet appearance.
Alternatively, these particle could be highly deliquescent particles of
mineral origin. The 8/16/91 WIN1 sample is notable for the presence of
phosphorous- and Zn-bearing particles, probably from industrial sources.
Although statistics are poor (only 3 particles were classified as Zn-bearing),
the SEM observations support th& relatively high Zn concentration measured by
XRJ in this sample.
CONCLUSIONS ASD X2CC8ESZXD2X2CXS
Chemical mass balance results are reasonable considering the
meteorological conditions and proximity of sampling sites to sources. SEM/EDX
analysis of individual particles supported the general conclusions of the CH3
analysis and even provided further insight into the origins of the particles
collected. Results could be improved upon by doing the following: 1) Include
analysis of major species such as elemental and organic carbon and nitrate; 2)
Obtain "site-specific" source profiles for major point and area sources; 3)
Collect daytime and nighttime 12-hour samples for some portion of the study to
obtain information on diurnal variations of sources. Interpretation of
results could be facilitated by doing the following: 1) Obtain local emissions
inventory data for Canada; 2) Collect coarse samples on Huclepore filters to
reduce sample matrix interference in the SEM analysis; 3) Employ computer-
controlled SEM in addition to manual SEH to reduce the amount of time required
per analysis and to increase the number of samples and the number of particles
9

-------
93-TP-58.05
per sample analyzed by SEH.
R37ZSSHCSS
1.	Volatile Organic Compound (VOC1/Particulate Hatter (PM) Speciation Data
System, Version 1.4. EPA-450/4-91-027, U.S. Environmental Protection Agency,
Research Triangle Fark, 1991.
2.	J.G. Watson, N.F. Robinson, J.C. Chow, et al., "The USEPA/DRI Chemical Mass
Balance Receptor Model, CMB 7.0," Environ. Software, 5(1):38-49 (1990).
3.	S.J. Vermette, A.L. Williams, and S. Landsberger, Surface Dust Elemental
Profiles - Southeast Chicago. Illinois State Water Survey Contract Report 488,
Champaign, Illinois, 1990.
4.	S.J. Vermette and S. Landsberger, Airborne Fine Particulate Hatter CPK,«) in
Southeast Chicago. Illinois State Water Survey Contract Report 525, Champaign,
Illinois, 1991.
5.	T.G.Dzubay, R.K. Stevens, G.E. Gordon, et al., "A composite receptor method
applied to Philadelphia aerosol," Environ. Sci. Technol. 22(l):46-52 (1987).
6.	T.G.Dzubay, R.K. Stevens, C.W. Lewis, et al., "Visibility and aerosol
composition in Houston, Texas," Environ. Sci. Technol. 16(8):514-524 (1982).
7.	R.K. Stevens, T.G. Dzubay, R.W. Shaw Jr., et al., "Characterization of the
aerosol in the Great Smoky Hountains," Environ. Sci. Technol. 14(12):1491-1498
(1980) .
8.	T.L. Vossler and E.S. Hacias, "Contribution of fine particle sulfates to
light scattering in St. Louis summer aerosol," Environ. Sci. Technol.
20(12):1235-1243 (1986).
9.	T.G. Dzubay amd Y. Hamane, "Use of electron microscopy data in receptor
models for PK-10," Atmos. Environ. 23(2)s467-476 (1989).
ACXSOWLEDG23GHTS
We gratefully acknowledge the assistance of Robert K. Stevens of the
U.S. EPA and Esther Bobet of Environment Canada and the many individuals
responsible for maintaining the sampling stations.
D2SCIAIHZR
The information in this document has been funded in part by the U.S.
Environmental Protection Agency under interagency agreement number RWCN934949-
01-1 with Environment Canada. It has been subjected to Agency review and
approved for publication. Mention of trade names or commercial products does
not constitute endorsement or reccrmendation for use.
10

-------
SARNIA
Lnk• SI. Clair
MJ WINDSOR
Figura 1. Map of study area Bhowing locutions of sampling sites and point sources.

-------
Table I. Average concentrations and uncertainties of elements determined by XRF and mass
in the coarse and fine fractions.

FINE (ng/m3)
COARSE (ng/m1)

WALPOLE
WINDSOR 1
WINDSOR 2
WALPOLE
WINDSOR 1
WINDSOR 2
NUMBER
47
32
47
47
32
47
DATES
2/5 - 11/26
3/19 - 11/14
1/30 - 11/14
2/5 - 11/26
3/19 - 11/14
1/30 - 11/14
(1991)






MASS
20,703 ± 832
20,318 i 851
15,990 ± 779
12,787 i 749
14,963 ± 766
12,423 ± 701
At,
76 ± 27
129 ± 35
39 i 24
454 ± 144
1017 ± 304
299 ± 102
SI
107 ± 10
109 ± 19
111 ± 19
1384 ± 347
1298 ± 326
963 ± 244
S
2609 ± 194
2597 ± 193
1818 ± 135
46 ± 71
198 ± 84
174 ± 62
CL
6.1 ± 2.2
9.8 ± 2.5
28.2 ± 3.6
25 ± 4
158 ± 18
250 ± 28
K
74 ± 6
94 ± 7
84 4 6
140 ± 15
131 ± 15
101 ± 12
CA
44 1 4
75 ± 6
78 ± 6
652 ± 53
1361 ± 110
1222 ± 99
TI
3.4 ± 2.2
4.3 ± 2.2
3.4 ± 2.4
29 ± 5
33 ± 6
24 ± 4
V
2.0 ± 0,8
2.5 ± 0.8
1.8 ± 0.9
2.1 ± 1.0
2.4 ± 1.0
1.4 ± 0.9
CR
0.6 ± 0.3
0.9 ± 0.4
0.9 ± 0.4
1.0 ± 0.4
2.4 ± 0.6
1.8 ± 0.5
MN
4.6 ± 0.6
13.6 ± 1.1
23.7 ± 1.9
6.7 ± 0.9
15.6 ± 1.7
16.1 ± 1.9
FE
76 ± 7
183 ± 17
258 ± 24
292 ± 31
580 ± 61
581 ± 62
MI
0.8 ± 0.4
0.9 ± 0.4
1.1 ± 0.5
0.6 ± 0.4
1.4 ± 0.5
0.8 ± 0.5
CU
2.7 ± 0.5
9.2 ± 1.1
5.7 ± 0.8
1.5 ± 0.5
7.2 ± 1.1
3.7 ± 0.7
ZN
25.9 ± 2.6
85 ± 8
96 ± 9
10.3 ± 1.6
48 ± 6
42 ± 6
SE
2.5 ± 0.4
2.5 ± 0.4
2.1 ± 0.4
0 ± 0.2
0 ± 0.2
0.1 ± 0.2
BR
3.1 ± 0.5
3.5 ± 0.5
3.1 ± 0.5
0.3 ±0.3
0.7 ± 0.3
0.6 * 0.3
SR
0.8 ± 0.3
0.6 ± 0.3
0.8 ± 0.4
2.1 ± 0.4
4.4 4 0.6
3.8 ± 0.5
SN
8.7 ± 2.8
5.8 ± 2.6
5.5 ± 2.8
2.0 ± 2.5
3.4 1 2.5
2.6 ± 2.5
BA
4 ± 5
5.1 ± 4.7
7.4 ± 5.0
7 ± 5
12.3 ± 4.8
10 i 5
PB
11.1 ± 1.4
22.6 ± 2.4
16.2 ± 1.8
1.7 ± 0.8
6.0 ± 1.2
5.0 ± 1.0

-------
Table II. Average rat too (and standard deviations) of noil-related elements to Si.
Site
Size
Fraction
Al/Si
K/Si
Ca/Si
Fe/Si
WIN2
Coarse
0.31 ± 0.11
0.12 ± 0.08
1.38 ± 1.01
0.66 ± 0.43
WIN1
Coarse
0.85 ± 1.24
0.11 ± 0.04
1.13 ± 0.31
0.50 ± 0.23
WAL
Coarne
0.38 ± 0.32
0.11 ± 0.04
0.60 ± 0.40
0.23 ± 0.06
WIN2
Fine
0.31 ± 0.34
0.78 ± 0.51
0.70 ± 0.25
2.38 ± 1.58
WIN1
Fine
1.03 ± 1.60
0.96 ± 0.45
0.75 ± 0.26
1.78 ± 0.94
WAL
Fine
0.84 ± 0.B5
0.80 ± 0.68
0.47 ± 0.19
0.73 ± 0.38
Cruatal
Shale
-
0.293
0.097
0.081
0.173
Cruatal
Limestone
-
0.175
0.112
12.58
0.158
Table III. Wind sector averages applied to Windsor and Walpole samples and the average fine and
coarse mass concentrations for those wind sectors.
WALPOLE
WINDSOR 1
WINDSOR 2
Wind
Sectors
All
No. of
Samples
Fine
Mass,
fjg/m1
Coarae
Mass,
fjg/m'
Wind
Sectors
No. Of
Samples
Fine
Mass,
jjg/m'
Coarse
Mass,
M9/m3
Wind
Sectors
No. of
Samples
Fine
Mass,
pg/m
Coarse
Mass,
fjg/m1
40
20.7
12.8
All
(minus 2
outliers)
32
20.3
15.0
All
47
16.0
12.4
NNE-ESE
8
14.7
10.0
NNE-E
6
9.8
9.9
NE-ESE
8
7.6
7.2
NNW-N
4
5.1
6.3
N-NNW
3
15.5
11.2
N-NNE
4
12.3
8.3
WMW-NW
5
11.0
10.2
WNW-NW
3
10.9
10.9
WNW-NNW
7
8.8
9.1
W-WSW
16
16.6
11.4
W-SW
10
24.6
21.8
W
4
20.2
12.6
SW
4
25.2
14.1
SSW-S
10
26.6
14.1
WSW
9
19.0
19.1
{SSW-
SE)
3
not
used
not
uned




SW-S
15
21.8
13.8
No Data
7










u>
i
H
i
en
oo
O
U1

-------
93-TP-58.05
Table IV. CHB results for coarse-particle fraction given in percent of
measured mass with uncertainty estimates.

SOURCE CATEGORY4
STL
SALT
INCIN
LIKE
SKALE
WAL1, AVG
0%
0%
0%
13.7 ± 2.0%
43.3 ± 3.6%
WAL, NE-ESE
0%
0%
0%
17.5 ± 2.4%
40.8 ± 3.5%
WAL, N^W-N
0%
0%
0%
22.3 ± 2.9%
37.5 ± 3.5%
WAL, WNW-NW
0%
0%
0%
14.4 ± 2.1%
50.9 ± 4.1%
WAL, WSW-W
0%
0%
0%
8.5 ± 1.2%
31.2 ± 2.6%
WAL, SW
0%
0%
0%
15.4 ± 2.2%
42.3 ± 3.5%
WIN12, AVG-2
13.3 ± 2.3%
0%
6.3 ± 0.9%
20.6 ± 3.5%
21.7 ± 4.0%
WIN1, NNE-E
5.2 ± 1.7%
0%
3.9 ± 0.6%
28.8 ± 4.1%
26.1 ± 3.9%
WIN1, N-NNW
7.9 ± 1.9%
0%
6.7 ± 1.0%
21.1 ± 3.3%
23.2 ± 4.1%
WIN1, WNW-NW
14.0 ± 2.1%
0%
3.5 ± 0.5%
16.0 ± 2.9%
21.3 ± 3.4%
WIN1, W-SW
13.2 ± 2.1%
0%
7.2 ± 1.0%
18.0 ± 3.2%
17.0 ± 3.7%
WIN1, SSW-S
17.4 ± 2.8%
0%
5.8 ± 0.8%
20.7 ± 3.8%
26.4 ± 4.4%
WIN23, AVG
19.2 ± 2.8%
1.5 ± 0.6%
6.6 ± 1.6%
20.8 ± 3.8%
17.7 ± 4.1%
WIN2, NE-ESE
5.5 ± 1.8%
1.2 ± 0.4%
3.7 ± 1.0%
26.7 ± 3.9%
30.4 ± 4.1%
WIN2, N-NNE
16.6 ± 2.4%
4.7 ± 1.1%
7.6 ± 1.8%
13.9 ± 2.8%
12.7 ± 3.8%
WIN2, WNW-NNW
20.9 ± 2.9%
1.5 ± 0.7%
7.0 ± 1.7%
21.7 ± 4.0%
17.9 ± 4.3%
WIN2, W
21.3 ± 2.8%
4.1 ± 1.0%
7.9 ± 1.8%
16.4 ± 3.4%
10.3 ± 3.7%
WIN2, WSW
19.7 ± 2.8%
2.4 ± 0.7%
5.5 ± 1.4%
23.5 ± 4.2%
15.5 ± 3.6%
WIN2, SW-S
21.4 ± 3.0%
0%
6.6 ± 0.9%
19.0 ± 3.7%
19.5 ± 3.8%
'Chi2 = 0.18 - 1.40; Degrees of Freedom * E
^hi2 = 0.56 - 2.11; Degrees of Freedom * 4
^hi2 * 0.19 - 1.17; Degrees of Freedom * 5 (except 6 for SW-S)
4STL * Steel-yard road dust; SALT ¦ eodium chloride; INCIN « incineration
emissions; LIKE = crustal limestone; SHALE = crustal shale
14

-------
93-TP-58.05
Table V. CHB results for fine-particle fraction given in percent of measured
mass with uncertainty estimates.

SOURCE CATEGORY4
STL
AHSF
OILPP
INCIN
SKALE
WAL1, AVG
2.0 ± 0.2%
51.7 ± 6.5%
0.2 ± 0.1%
0.9 i 0.1%
1.5 i 0.4%
WAL, NS-ESE
2.0 ± 0.2%
56.4 ± 7.0%
0.5 ± 0.1%
0.7 ± 0.1%
1.4 ± 0.4%
WAL, NNW-N
3.7 ± 0.4%
36.8 ± 4.6%
0.3 ± 0.4%
1.0 ± 0.2%
2.1 ± 0.7%
WAL, WNW-NW
2.2 ± 0.2%
42.2 ± 5.3%
0.2 ± 0.1%
1.4 ± 0.2%
1.7 ± 0.4%
WAL, WSW-W
2.3 ± 0.3%
49.5 ± 6.2%
0.2 ± 0.2%
0.8 ± 0.1%
3.0 ± 0.6%
WAL, SW
2.3 ± 0.2%
49.4 ± 6.2%
0.2 ± 0.1%
0.9 ± 0.1%
1.6 ± 0.4%
WIN12, AVG-2
7.8 ± 0.7%
52.1 ± 6.5%
0.2 ± 0.1%
3.1 ± 0.3%
0%
WIN1, NNE-E
5.7 ± 0.6%
50.5 ± 6.3%
0.8 ± 0.2%
2.2 ± 0.2%
0%
WIN1, N-NNW
5.6 ± 0.5%
46.1 ± 5.8%
0.6 ± 0.1%
2.6 ± 0.3%
0%
WIN1, WNW-NW
6.0 ± 0.6%
38.5 ± 4.8%
0.1 ± 0.2%
2.2 ± 0.2%
0%
WIN1, W-SW
9.1 t 0.8%
55.3 ± 6.9%
0.1 ± 0.1%
3.1 t 0.3%
0%
WIN1, SSW-S
7.5 ± 0.6%
52.3 ± 6.5%
0.1 ± 0.1%
3.4 ± 0.3%
0%
WIN23, AVG
13.8 ± 1.2%
45.9 ± 5.8%
0.3 ± 0.1%
4.3 ± 0.5%
0%
WIN2, NE-ESE
6.4 ± 0.7%
42.0 ± 5.3%
0.7 i 0.3%
1.6 ± 0.2%
0%
WIN2, N-NNE
10.7 ± 0.9%
39.9 ± 5.0%
1.2 ± 0.2%
3.7 ± 0.5%
0%
WIN2, WNW-NNW
10.0 ± 0.9%
35.1 ± 4.4%
0.7 ± 0.3%
5.1 ± 0.6%
0%
WIN2, W
10.4 ± 0.9%
31.4 ± 4.0%
0.2 ± 0.1%
3.5 ± 0.4%
0%
WIN2, WSW
17.4 ± 1.4%
51.4 ± 6.4%
0.1 t 0.1%
5.8 ± 0.7%
0%
WIN2, SW-S
14.3 i 1.1%
50.4 ± 6.3%
0.1 ± 0.1%
4.0 i 0.5%
0%
'Chi2 * 0.88 - 3.48; Degrees of Freedom ™ 5
^hi5 = 1.37 - 4.67; Degrees of Freedom * 6
JChi2 = 1.00 - 4.13; Degrees of Freedom * 7
4STL * Steel-yard road dust; AHSF » ®H3EoniuiB «ulfate; OILPP * oil-fired power
plant emissions; INCIN = incineration emissions; SHALE * crustal shale
15

-------
93-TP-5S.05
Table VI. Coarse fraction samples analyzed by SEH/EDX and the coarse mass
concentrations and prevailing wind conditions.
Sampling Site
Sampling Date
Coarse Mass,
pg/m3
Prevailing
Wind
Direction
Mean Wind
Speed,
km/hr
WAL
7/17/91
19.8
wsw
15
WIN1
8/10/91
6.8
NNW
12
WIN1
8/16/91
23.4
SSW
14
WIN2
4/30/91
36.4
WSW
33
WIN2
7/17/91
30.3
wsw
17
16

-------
TECHNICAL REPORT DATA
1. REPORT NO.
EPA/600/A-93/105
2.
3" PB93- 1 9 140 1
4. TITLE AND SUBTITLE
Source Apportionment of Fine and Coarse Particles in
Southern Ontario, Canada
5.REPORT DATE
6.PERFORMING ORGANIZATION CODE
EPA/600/09
7. AUTHOR(S)
Teri L. Conner1, John L. Miller2, Robert D. Willis3,
Robert B. Kellogg3, Thomas F. Dann4
8.PERFORMING ORGANIZATION REPORT
NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
'Atmospheric Research and Exposure Assessment
Laboratory, US Environmental Protection Agency,
Research Triangle Park, NC, 27711
2Senior Environmental Employee
'ManTech Environmental Technology, Inc., Research
Triangle Park, NC, 27709
"Environment Canada, River Road Environmental
Technology Centre, Ottawa, Ontario, K1A OH3, Canada
10.PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
Funds-In IAG RWCN934949-01-1
+ In-house
12. SPONSORING AGENCY NAME AND ADDRESS
Atmospheric Research and Exposure Assessment
Laboratory, US Environmental Protection Agency,
Research Triangle Park, NC, 27711
13.TYPE OF REPORT AND PERIOD COVERED
Work Performed December 1991
- June 199 3,
Proceedings of AWMA
Conference - June 1993
14. SPONSORING AGENCY CODE
EPA/600/^
15. SUPPLEMENTARY NOTES
16. ABSTRACT
Environment Canada, in cooperation with the Ontario Ministry of Environment
and the Walpole Island Indian Band, has been conducting an air monitoring study in
a region of southern Ontario near Detroit. Two sampling sites are located in the
city of Windsor, Ontario. The Windsor sites are frequently downwind of the
numerous emission sources of the greater Detroit area, which include coke ovens,
iron and steel industry, incinerators, power generation facilities, lime and cement
operations, and automotive assembly plants. The Windsor sites are also influenced
by the regional background of secondary sulfate common in the eastern U.S. and
Canada, as well as by automobile emissions. A third site, located on Walpole
Island, was set up to assess the background particulate composition, although this
site is also influenced to some degree by primary industrial emissions and
secondary pollutants.
Fine and coarse particle samples were collected on Teflon filters with a
dichotomous sampler. Samples were analyzed by energy-dispersive X-ray fluorescence
(XRF) at the U.S. EPA, Research Triangle Park facility. A subset of the samples
were selected for analysis by scanning electron microscopy combined with energy-
dispersive XRF (SEM/EDX). Morphological features of the particles combined with
chemical data have been shown to be useful in resolving source types which cannot
be resolved by conventional means ISource apportionment of fine and coarse
particles based on SEM/EDX and on conventional (i.e., statistical) methods applied
to XRF data will be reported. Meteorological observations will be used to
interpret the results.			'j	
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b .IDENTIFIERS/ OPEN ENDED
TERMS
c.COSATI

-------