Multi-Site Evaluations of Candidate Methodologies tor
Determining Coarse Particulate Matter (PMc)
Concentrations ¦
Paper #10
Robert W. Vanderpool, Thomas G. EUestad, Mary K. Harmon, Timothy D. Hanley,
Richard D. SchdTc, ami Elizabeth T. Honlke
United States Environmental Protection Agency, 109 T.W. Alexander Drive, Research Triangle
Park, NC 27711
Pan! A. Solomon
United States Environmental Protection Agency, P.O. Box 93478, Las Vegas, NV 89193
Robert W. Murdoch, Sanjay Natarajam, and Christopher A, 'Noble
Research. Triangle Institute, 3040 Comwallis Road, Research Triangle Park, NC 27709
Jeffrey L. Ambs
Rnpprecht and Patashnick, 26 Tech Valley Road, East Green bush, NY 12061
Gimore J. Sem
TSI Inc., 500 Cardigan Road, Shore view. MN 55126 '
John Tisch
Tisch Environmental Inc., 145 S. Miami Avenue, Cleves, OH 45002
ABSTRACT '
Comprehensive field studies were conducted to evaluate the performance of sampling
methods for measuring the coarse fraction of PMI0 in ambient air. Five separate sampling
approaches were evaluated at each of three sampling sites. As the primary basis of comparison, a
discrete difference method was used which employs two designated FRM samplers, one to
measure PM2 5 and the other PM10. The numerical difference of these reference method,
concentrations (PMl0-PM2 5) represented an estimate of PMc. A second sampling approach
involved a sequential dichotomous sampler, which provided both PM2_5 and PMc measurements.
In both of these filter-based, time-integrated measurement approaches, the collected aerosol mass
was analyzed gravimetrically in the laboratory under controlled conditions. Three continuous
coarse particle samplers that measure PMc directly with a time resolution of 1 hour or less were
also evaluated. One such sampler was a commercially available system based on beta attenuation,
the second was based on TEOM technology. Both of these measurement approaches used
dichotomous virtual impactors for separating fine and coarse particles. The third real-time
sampler evaluated was an aerodynamic particle sizer (APS) that measures the aerodynamic
Page 1 of 20

-------
diameter of individual particles, calculates the mass of the particle based on an assumed particle -
density, then sums the mass within the size range of interest to estimate the PMc mass
concentration.
Sampling sites and timing of the studies were selected to provide diverse challenges to the
samplers with respect to aerosol concentration, aerosol particle size distribution, and aerosol
composition. Results from performance evaluations of the candidate PMc samplers at Gary, IN,
Phoenix, AZ, and Riverside, CA will be presented.
INTRODUCTION
In response to increasing evidence of the adverse health effects associated with exposure
to ambient fine particles, the United States Environmental Protection Agency (EPA) promulgated
in 1997 a national ambient air quality standard (NAAQS) for PM^1. Accompanying the standard
were strict design and performance requirements which candidate PM2_5 samplers must meet in
order to be approved by EPA for use in making compliance measurements2. The 1997 regulations
retained the existing annual PM10 standard and made only slight modifications to the statistical
basis upon which to assess compliance with the 24-hour PM10 standard.
Based on subsequent litigation, the U.S. Court of Appeals for the District of Columbia
reviewed the 1997 regulations and upheld EPA's promulgation of the PM1S standard. While
acknowledging the need to regulate coarse particles, the Court vacated the 1997 PM10 standard
after concluding that PM10 is a "poorly matched indicator for coarse particulate pollution" because
PM10 includes the PM2 s fraction. EPA did not appeal this ruling and now intends to promulgate a
new NAAQS tor PMc (i.e. the coarse fraction of PM10)
Inherent to any new NAAQS is the need for sampling and analysis methods capable of
measuring the new metric with known quality. In support of this goal, the purpose of this field
study series was to conduct a survey of available instrumentation designed to measure the coarse
fraction of PM10, and to conduct a multi-site performance evaluation of these instruments.
Sampling sites were selected in order to evaluate the instruments under a wide variety of
environmental conditions, particle concentrations, particle size distributions, and particle
compositions. At three separate cities (Gary, m, Phoenix, AZ, and Riverside, CA) thirty daily,
22- hour tests were conducted. In addition to filter-based samplers which provide integrated test
results, near real-time PMc monitors were evaluated which possess time resolutions of one hour ¦
or less. Multiple monitors of each type were used in order to determine the inherent precision of
each sampler's design.
This report provides a description of the instruments evaluated in this study, outlines the
sampling and analysis procedures used to conduct the performance evaluations, describes the
characteristics of each of the three sampling sites, and provides a summary of test results.
Because chemical analysis of archived filters has not yet been completed, this report will focus
solely on mass concentration results reported by the various PMc samplers.
DESCRIPTION OF PMc SAMPLERS
Selection of the samplers to be involved in the field comparison study was based on the
Page 2 of 20

-------
following criteria. First, all samplers must be designed to provide a measurement of the mass
concentration of PMc aerosols based on aerodynamic diameter. Selected filter-based samplers
must be capable of providing integrated samples at least every 24 hours and use the PM2,5 FRM's
standard cassette and Teflon afterfilter. Selected continuous and senai-continuous instruments
must be capable of providing PMc mass.measurements at least every 1 hour. All samplers must
be capable of automated operation over a period of 24 hours with active control of flow rates.
Last, all selected samplers must be either commercially available or in the final prototype stage of
their design.
Based on these criteria, five separate PMc measurement approaches were selected for
evaluation in this study. Table 1 lists each sampler used in this study, its manufacturer, and the
number of samplers used at each sampling site. For the filter-based samplers, the filter
composition is listed along with the species to be determined during the filter's post-sampling
gravimetric and/or chemical analysis. Due to funding constraints, not all the collected filters could
be chemically analyzed. Instead, a representative subset of archived filters from each site was
selected for chemical analysis based on the review of the comparative mass concentration results.
Table 1. Inventory of samplers used in the perfonnance evaluation.
Measurement
Method
PM
Metric
Sampler
Manufacture r(s)
Number
Used
Filter
Composition
Species Analyzed
Integrated ERM
PM,r,
BGI, R&P, AND
3
Teflon
Mass, sulfate,
nitrate, metals
Integrated ERM

BGI
1
Qaaitz
EC, OC
Integrated ERM
PM2j
BGI, R&P, AND
3
Teflon
Mass, sulfate,
nitrate, metals
I Integrated FRM
, PM2.j
AND
1
Quartz
EC. OC
Integrated Dichot,
sequential
PM21, PMc
R&P
3
Teflon
Mass, sulfate,
nitrate, metals
Integrated Dichot,
sequential
FM2 5, PMc
R&P
I
Quartz
EC, OC
TEOM
PMc
R&P
3
none
none
Beta Attenuation
PM, PMc
Tiseh
3
none
none
Time of Flight (APS)
PMc
TSI
2
none
none

Tefal = 20

The voluntary participation and involvement of the PMc sampler manufacturers during this
study was a critical component of the study's success. With the exception of the PM25 and PMW
FRM samplers which were supplied by EPA, all field samplers in this study were supplied by their
respective manufacturers. The supplied samplers all represented the latest models of each design
Page 3 of 20

-------
and were equipped with the most current hardware, firmware, and software. All manufacturers
supervised installation and calibration of their respective samplers during the initial shakedown
tests conducted in Research Triangle Park (RTP), NC and provided technical reviews of SOPs
written for the instrument's setup, calibration, and operation. Each manufacturer was also
provided the opportunity to visit each field site during site setup in order to verify the working
condition of their samplers. At the completion of sampling at each field site, each manufacturer
was supplied their respective field data in order to ensure that their sampler data was being
properly retrieved from the instrument, correctly analyzed, and correctly interpreted.
Collocated PM2.5 and FM10 FRM Samplers.
• In the first PMc measurement approach, commonly referred to as the "difference method",
a designated PM^ 5 FRM sampler is collocated with a designated PM10 FRM sampler. For
accurate determination of PMc concentrations, the PM10 sampler is simply a designated PM2_5
FRM with its WINS fractionator replaced by a straight downtube (Figure 1). Both samplers are
installed, calibrated, operated, and analyzed using
standard PM2.5 protocols. The two samplers thus
have identical Met aspiration characteristics,
produce identical PM10 fractions, and collect
aerosol at the same face velocity through the same
filter media. At the completion of concurrent
sampling periods, the PMc concentration is
calculated as the numerical difference between the
measured PMW concentration and the measured
PM2 5 concentration. Due to its fundamental
measurement principle, the difference method was
used as the basis of comparison upon which to
evaluate the performance of the other PMc
samplers in the study. For purposes of this paper,
data collected using this method is termed "PMc
FRM" data. '
In this study, a designated PM1(rPM7 5
FRM pair was used from each of three separate
sampler manufacturers: Thermo-Andersen
(AND), BGI and Rupprecht and Patashnick
(R&P). Each of these six FRM samplers were
operated with prewcighed Teflon filters for subsequent gravimetric and ion chromatography (IC)
or x-ray fluorescence (XRF) analysis. A fourth set of PMZ S and PM10 FRM samplers was used
and both sampler's were equipped with a quartz filter to enable subsequent thermal optical
measurement of the aerosol's elemental carbon (EC) and organic carbon (GC) constituents. In
this study, the prefired quartz filters were not analyzed gravinaetrically but were archived under
cold conditions for subsequent EC/OC analysis.
Figure 1. Schematic diagram of the FRM
samplers used in the PMc difference method.
V/
NO/
I
PIWjQ
f_J
PM1(,

PM10
PMctacSon
removed In WINS
PMZS
PMc = P^0-P%S
Page 4 of 20

-------
R&P Model 2025 Sequential Dichotomous (Dichot) Sampler
The Model 2025 dichot was designed to provide integrated measurement of both fine and
coarse fractions of a PM10 aerosol. The sampler actively provides volumetric flow control
through a standard 16.7 actual liters per minute (a!pm) PM10 Met. Following the aspirated
aerosol's fractionation in the inlet's internal fractionator, the resulting PM10 aerosol enters a
virtual impactor where the aerosol is then split into major and minor flow streams. Ideally, the
major flow (maintained at 151pm) is intended to collect only the PM, 5 fraction of the PMi0
aerosol while the minor flow (maintained at 1.7 alpm) is intended to collect only the PMc fraction
of the PM10 aerosol. In practice, however, this size fractionation is never ideal and 10% of the
PMZ5 mass theoretically deposits onto the PMc filter. The presence of these fine particles is
numerically accounted for during subsequent calculation of the PMc concentration. Assuming
that particle losses within the instrument are negligible, the sum of the measured PMZ_5 and PMc
concentrations provide a measure of the ambient aerosol's PM10 concentration.
The Model 2025 sequential dichot allows unattended, multi-day operation through the use
of a filter exchange mechanism for transferring filter cassettes from a supply tube to the sampling
position, then conducting a post-sampling transfer of the cassettes to a storage tube. During this
study, however, the multi-day capability of the Model 2025 was not utilized and supply tubes
were manually loaded with only one cassette shortly before each test and the post-sampling
cassette was manually retrieved from the storage tube shortly after each test. Procedures for
gravimetric and chemical analysis of the Model 2025's filters were identical to those of the FRM?" s
filters.
Four separate E&P sequential dichotomous samplers were used during this study, three of
which were equipped with Teflon filters while the fourth was equipped with prefired quartz filters
to enable determination of elemental and organic carbon components of the ambient aerosol
Tiscfc Inc. Model SPM-613D Dkfiefomous Beta Gauge
Manufactured by Kimoto Electric Co., LTD., the Tisch SPM-613D dichot beta gauge is
designed to provide near real-time measurement of both the fine and coarse fractions of the PM10
aerosol. Similar to the R&P Model 2025 dichot, the SPM-613D aspirates the ambient aerosol
through a standard 16.71pm inlet and introduces the fractionated PM10 aerosol into a custom'
designed virtual impactor. The virtual impactor in the SPM-613D has different dimensions than
that of the R&P design and operates its major and minor flow rates at slightly different flow rates,
15,2 1pm and 1.5 1pm, respectively. Flow control in the two SPM-613D channels is monitored
using separate mass flow sensors. The system's flow control system, however, is designed to
maintain the calibrated mass flow rate and thus does not maintain true volumetric flow rates
through the inlet at actual ambient temperature and pressure conditions. By conducting flow rate
calibrations at the sampler's Met under actual temperature and pressure conditions, however, the
effect of this lack of volumetric flow control is minimal if ambient conditions do not differ
substantially from those existing during the flow calibration.
Downstream of the SPM-613D's virtual impactor, the separate fine and coarse flow
streams are continuously collected on a paper roll composed of low hygroscopicity polyfon.
Page 5 of 20

-------
Following each hour of aerosol collection, the attenuation of 147Pm beta rays by each channel's
aerosol deposit is quantified using two separate sets of beta sources and detectors. Based on
previous span calibrations performed by the user, the theoretical relationship between beta
attenuation and collected aerosol mass is used to estimate the mass of each separate aerosol
deposit. Because beta rays are also attenuated by condensed water, an external heater is located
downstream of the sampler's Met and maintains the temperature of the aspirated airstream above
25 °C. As in the R&P 2025 dichot, numerical corrections are made to account for the theoretical
mass of fine particulates contained within the SPM-613D's coarse channel filter. Three identical
SPM-613D beta gauges were used during the course of the study at all three sampling sites.
R&P Continuous Coarse TEGM Monitor' .
As designed by Misra, et al.J and licensed to R&P, the coarse TEOM was designed to
provide a near real-time measurement of PMc concentrations. The instrument aspirates ambient
aerosol at 501pm through a custom size-selective inlet, which was made by modifying a standard
16.71pm size-selective PM10 inlet by adjusting the internal dimensions in an effort to provide a 10
|im cutpoint. Downstream of the inlet, the PM10 fraction then enters a custom virtual impactor
whose major and minor flow rates are 481pm and 2 lpm, respectively. In this design, the fine
fraction (major flow) is collected in a replaceable total filter and the collected fine aerosol mass is
not subsequently quantified. Downstream of the virtual impactor, coarse aerosols in the minor
flow stream are first heated to 50 °C to minimize interferences from particle bound water and are
then deposited in a standard R&P 1400a Tapered Element Oscillating Microbalance (TEOM).
The mass of the deposited aerosol is then estimated based on the observed change in vibrational.
frequency of the TEOM filter during the collection period. Due to the high flow rate ratio
between the total and minor flows (25 to 1), no correction is made for the mass of fine particles
on the coarse filter in this design. The PMc mass concentration is then calculated as the measured
coarse mass divided by the volume of ambient air aspirated during the sampling event. Three
replicate R&P coarse TEOMs were used during this field study in order to determine the inherent
measurement precision of the samplers.
TSI Inc. Model 3321 Aerodynamic Particle Sizer (APS)
The final measurement approach used in the field study involved the TSI Inc. Model 3321
-APS to estimate the mass of ambient coarse particles based on their aerodynamic properties in an
accelerating flow stream. To.adapt the 5 1pm APS to field use. a standard 16.7 lpm PMj0 inlet
was used in conjunction with a custom designed flow splitter located downstream of the Met. In
the splitter, a sharp-edged, isokinetic nozzle extracts a representative sample of the PM10 aerosol
for measurement in the APS. The remaining 11.7 lpm portion of the PMI0 aerosol was drawn
through a total filter using a volumetrically controlled vacuum source. The mass of the aerosol
collected on the total filter was not quantified.
The 5 lpm representative aerosol sample is then introduced into the APS and the
aerodynamic diameter of individual particles estimated using time of flight technology. The
volume of each particle is then calculated based on its measured aerodynamic diameter and a
Page 6 of 20

-------
particle density specified by the user. For purposes of this field study, a particle density of
2 g/cm3 was assumed as representative for the coarse fraction of PM10 aerosols. The mass
concentration of PMc aerosols is then calculated as the sum of the mass of all particles penetrating
the PM,0 inlet whose aerodynamic diameters were greater than 2.5 micrometers. Because the
APS is only capable of resolving particles larger than approximately 0.7 micrometers aerodynamic
diameter, the system is not applicable for measurement of either PMZS or PM10 ambient
concentrations because particulate mass less than 0.7 micrometers is not quantified.
It should be noted that the primary purpose of incorporating the two APS units into the
field study was to provide ambient aerosol size distribution information at each site. Because the
APS's measurement method has legitimate potential for providing continuous PMc concentration
measurements, it was evaluated in this study in the same manner as the other PMc samplers.
SITE SETUP AND OPERATING PROCEDURES
All field and laboratory activities associated with this study were conducted by Research
Triangle Institute (RTT) under EPA contract 68-D-00-206. Prior to conducting the study, RTT
developed a Quality Assurance Project Plan (QAPP) which encompassed all aspects of the study's
field and laboratory activities. The QAPP was subsequently reviewed and approved by QA
personnel within EPA's Office of Research and Development (ORD) prior to initiation of the
study. All field and laboratory operations of the study were also reviewed and approved during a
comprehensive Systems Audit conducted by ORD prior to the field tests. ¦
The multi-site performance evaluations of the 20 separate field samplers presented a
unique logistical challenge. With the exception of the FRM samplers and the R&P dichots, none
of the other samplers have weather enclosures and must thus be protected from the elements
during sampling. To enable efficient transportation of all field equipment and to house the field
samplers, a 25 foot long motor home was adapted for use in this study. The twelve FRM and
R&P dichot samplers were installed either on the roof of the motor home or on a 10' by 10'
auxiliary platform positioned immediately adjacent to the motor home. The remaining eight PMc
samplers were installed inside the motor home with their downtubes extending through the roof of
the motor home and attached to their respective inlets. The motor home's environmental controls
maintained the interior temperature at 23 °C ± 3 °C during all field tests. Per compliance testing
requirements, the inlets of all samplers were installed 2 m ± 0.2 m above the sampling platform
and all samplers were spaced horizontally at least 1 m apart from each other. At each site, the
motor home and auxiliary platform were free of nearby obstructions which might adversely
influence the spatial uniformity of PMc concentrations.
Prior to each field test, all samplers were cleaned and leak-checked. Each sampler was
then calibrated for volumetric flow rate, ambient temperature, and ambient pressure measurement
using a calibrated transfer standard (BGf DeltaCal). For calibration of the 501pm of the R&P
coarse TEQM; a BGf TriCal was equipped with a 55 1pm capacity flow module which had been
specifically designed for this purpose. Following the calibration of each instrument, a
performance audit was conducted using a separate audit device and any necessary adjustments
were made to the instruments. Jh addition to the initial audit conducted at each field site,
performance audits were also conducted following Run 15 and Run 30. Field blank tests of the
filter-based samplers were conducted at the same frequency as that of the performance audits.
Page 7 of 20

-------
At each sampling site, 30 daily, 22-hour tests were conducted from 11 am (local time) to 9
am of the following morning. The two hour interval between successive tests enabled the site
operator sufficient time for sample changeover, data recording, and minor maintenance while still
allowing for daily sampling. Typically, a 45 day test period was required to complete site setup,
30 days of sampling, and site shutdown.
Gravimetric analysis of the filter-based samplers' teflon filters was conducted both in the
EPA weighing facility in RTP, NC and at- each sampling site. In RTP, ¦ presampling filters were
equilibrated and preweighed in an environmentally controlled chamber whose temperature and
relative humidity setpoints were 22 °C and 35%, respectively. All filter weighings were
conducted using a Cairn C-44 microbalance which had a readability of 1 fig and a capacity of 5 g.
The analytical balance was tared and calibrated prior to each weighing session and Class 1
calibration weights were used during each session to verify the balance's internal calibration. In
order to increase the confidence in the gravimetric analysis, 100% replicate weighings (with a
5 fig reweigh threshold) were used for each filter during all preweighing and postweighing
operations. Quality control also included the use of three laboratory blank filters during each
weighing session. At the completion of the preweighing in RTP, the filters were loaded in
sampling cassettes, the cassettes sealed with metal endcaps, and the cassettes placed in Thermo-
Andersen cassette canisters. The canisters were then shipped to the field site in coolers designed
to maintain postsampling filters at temperatures below 2 °C.
Upon receipt of the preweighed filter from RTP, field personnel would then unpack and
equilibrate the filters in a weighing facility setup within a hotel room. Through careful monitoring
of the room's conditions and through use of an automated dehumidificr, site personnel were able
to maintain the site's weighing conditions within allowable temperature and relative humidity
limits. Presampling and postsampling site weighings were conducted using a Sartorius MC5
microbalance with the same capability as the Calm microbalance used for the RTP weighings.
Identical weighing protocols were used at all field sites, and at the RTP weighing facility. Once
postsampling filters were weighed at the site, they would be shipped to RTP for fi nal
postweighing and subsequent archiving under cold conditions. Conducting filter weighing at the
field site enabled faster determination of test results than could be obtained if samples were
shipped back to RTP. Conducting filter weighing at the site and at RTP also enabled
measurement of particle losses which might occur during shipping. Last, site weighing provided
valid test results in the event that a cooler might be, inadvertently damaged or lost during its
shipment back to the RTP weighing facility.
SITE CHARACTERISTICS
Following the initial installation and evaluation of the PMc samplers in RTP, NC to verify
the proper operating condition of the samplers and to finalize operating protocols, successive field
tests were conducted in Gaiy, EST, Phoenix, AZ, and Riverside, CA. The following section will
provide a description of these three sites along with the meteorological conditions and aerosol
characteristics encountered during each site's field tests.
Page 8 of 20

-------
Gary, IN
The Gary, IN site was selected as representing a midwest industrial city where primary
PMc aerosols are predominantly generated by industrial activity rather than by wind blown soils.
Selection and setup of the Gary, IN sampling site was made in cooperation with personnel from
the Indiana Department of Environmental Management. This State and Local Air Monitoring
(SLAM) site (AIRS # 18-089-0022) is located approximately 2 km south of Lake Michigan and, is
immediately adjacent to the property line of a steel mill. Nearby sources of emissions include the
steel mil which was located approximately 0.7 km northwest of the site and a 0.5 km long open
coal pile which was located approximately 0.5 km northeast of the site.
The 30 days of testing at the Gary site were conducted from March 6 to April 7, 2003.
Weather at the site was typically cloudy, windy, and cold and only one rain event occurred during
the study period. Temperatures at the site ranged from -15.1 "C to 27.8 "C and a mean daily site
temperature of 4.6 "C was recorded.
As measured by the three collocated FRM samplers, PM2 5 concentrations measured at
the Gary site during the tests ranged from 10.3 p,g/m3 to 46.9 fig/m3 with a measured mean of
22.8 ug/m3. Excellent inter-manufacturer agreement was observed among the filter-based PM2 5
FRM samplers as expressed by the coefficient of variation (CV) of 1.5%. As expressed by a
coefficient of variation of 2.4%, excellent inter-manufacturer agreement was also observed for
the PM10 FRM measurements. PM10 concentrations measured during the tests ranged from
22.6 p.g/m3 to 85.0 pg/m3 with a measured mean of 42.6 (ig/m3. PMc concentrations (expressed,
as the numerical difference between collocated PM10 and PM2J FRM measurements), ranged
from 4.5 |ig/m3 to 58.1 p,g/m3 with a measured mean of 19.8 p,g/m3. Inter-manufacturer
precision of PMc concentrations was determined to be 5.7% CV. As indicated by a mean PM2 5
/PM10 ratio of 0.55 during the 30 sampling events, slightly more than one-half of the site's PM10
aerosol was associated with PM2J5 aerosols. PM25 /PMi0 ratios ranged from 0.32 to 0.83 during
the 30 days of testing indicating that the size distribution of ambient aerosols was quite variable
during the month-long field tests. Predominant winds from the direction of the nearby steel mill
typically contributed to PM25 concentrations at the site while winds predominating from the
direction of the open coal piles resulted in measurement of high PMc site concentrations.
Filter weighing at the Gary site began with Run 5 filters. As indicated in Figure 2,
excellent agreement was seen between PMc concentrations based on the site weighings versus the
.'RTF weighings during Runs 5 through 30. The filter shipping and handling protocols designed
for the study, therefore, appeared
to result in negligible PM2 5 or
PM10 particle loss from the FRM
filters during their transport from
the field site to the RTF weighing
facility.
Phoenix, AZ
Tests were conducted in
Phoenix, AZ during early summer
Figure 2. Site versus RTF, NC weighing results of PMc
FRM concentrations in Gary, IN.
6S
6Q
E
c
39
6
a
10
5
10
15
35
Sair.pleDay

-------
of2003 in order to challenge the coarse particle samplers with high concentrations of dry, wind
blown crustal materials. Through cooperation with personnel at the Air Quality Division of the
Maricopa County Environmental Services Department, the county-operated Durango Complex
sampling site (AIRS # 04-013-9812) in the southwestern portion of Phoenix was selected as an
appropriate field site. The site is impacted from the east and north by nearby commercial districts
and two main interstate highways. With the predominant wind direction being from the west and
southwest, however, the site is primarily impacted by large'windblown soils originating from
nearby earthmoving equipment and non-vegetated, open fields.
The month-long field tests at the Phoenix sampling site were conducted from May 14 to
June 15, 2003. Weather at the site was typically clear, windy, and very hot and no rain events
occurred during the 30 day study period. Temperatures at the sampling site ranged from 17.1 °C
to 43.5 °C and a mean daily site temperature of 32.3 °C was recorded.
PM2,5 concentrations measured during the Phoenix tests ranged from 6.4 ng/ni3 to '
22.0 p,g/m3 with a measured mean of 11.0 p.g/m3. As observed during the Gary tests, excellent
inter-manufacturer agreement was achieved among the filter-based FRM samplers. As
expressed by the coefficient of variation, mean inter-manufacturer precision for FM;! 5 was
determined to be 3.4%. As expressed by a coefficient of variation of 3.3%, excellent inter-
manufacturer agreement was also observed for the PM10 ERM measurements. PM10
concentrations measured during the tests ranged from 35.6 p.g/m3 to 230.9 ng/m3 with a
measured mean of 66.6 }ig/ni3. PMc concentrations (expressed as the numerical difference
between collocated PM10 and PM2 5 ERM measurements), ranged-from 26.5 jig/m3 to
209.0 |j,g/m3 with a measured mean of 55.6 p,g/m3. Inter-manufacturer precision of PMc
concentrations measured by the three ERM pairs was determined to be 3.6% CV. As indicated by
the mean PM75/PM10 ratio of 0.18 during the 30 sampling events, a large fraction of the site's
PM10 concentration was associated with PMc aerosols. PM^/PM^ ratios ranged from 0.10 to
0.28 which indicated that coarse particle mass dominated the PM10 concentrations during each
day of the Phoenix tests. Figure 3 depicts the daily dominance of the coarse particles in the
Phoenix area and also shows the strong agreement obtained between the FRM filter weighings
conducted at the sampling site versus those conducted at the RTP weighing facility;
Riverside, CA
The Riverside, CA
sampling site was selected as a
west coast site where significant
secondary fine mode aerosols •
might be present in conjunction
with primary coarse aerosols.
Selection and setup of the
Riverside site was made through
cooperation .with the University of
California-Riverside (UCR). The
monitoring site is located on the
grounds of UCR's Agricultural
Figure 3. Site versus RTP, NC weighing of PMz;, PMc,
and PM10 concentrations at the Phoenix site.
Inter-fflanirfaefcrsrerPMji CV =3.4%
tatsr-OTanu&efcjrererPMs CV=3.3%
'-rasjiufticiiirsnfrPM^ CV»2Lfi?*

-*-pa2S5nr)
-O-FM10 (3®)
-fr-HiCJl ISitBS
pm?*
Page 10 of 20

-------
Operations Center and is operated by the South Coast Air Quality Management District
(California ARB Site #' 33162). Local sources of ambient aerosols include those from agricultural
research activities as well as from mobile source emissions in the area.
• Field tests were conducted at the Riverside sampling site from July 23 to August 24, 2003.
Weather at the sampling site during the 30 daily tests was typically warm with clear or partly
cloudy skies. No rain events occurred during the Riverside field tests although morning' fog was
occasionally observed at the site. Temperatures at the site ranged from 15.4 °C to 40.4 °C and a
mean daily site temperature of 25.9 °C was recorded.
As had been experienced during the Gary and Phoenix sites, excellent inter-manufecturer
agreement was observed among the filter-based FRM samplers. As expressed by the coefficient
of variation, mean inter-manufacturer precision for PM25 was determined to be 3.1%. Daily
PM2J concentrations measured during the tests ranged from 9.9 pg/m3 to 32.7 jig/m3 with a
measured mean of 17.7 |.ig/m3. As expressed by a coefficient of variation of 2.9%, excellent
inter-manufacturer agreement was also observed for the PM10 FRM samplers. PM10
concentrations measured during the tests ranged from 27.0 fig/nr1 to 69.3 iigfm3 with a measured
mean of 48.0 jig/m3. PMc concentrations (expressed as the numerical difference between
collocated PMI0 and PMZ5 FRM measurements), ranged from 16.2 [ig/m3 to 46,1 jig/m3 with a
measured mean of 30.4 (xg/m3. Inter-manufacturer precision of PMc FRM measurements was
determined to be 4.'1% CV, As indicated by the mean PM . /PM,,, ratio of 0.37 during the 30
sampling events, approximately two-thirds of the sites PMI0 concentration was associated with
PMc aerosols. PM2 5/PM10 ratios ranged from 0.25 to 0.50 during the 30 days of testing at the
Riverside site indicating that coarse particles dominated the PM10 aerosol during all tests.
TEST RESULTS .
Federal Reference Method Samplers
As previously described, field tests involved the use of four sets of PM25 and PM10
samplers from BGI, Andersen,' and R&P. Because there exist no absolute standards for ambient
particulate matter, the absolute accuracy of these devices cannot be determined from these tests.
However, the performance of the three separate manufacturers' samplers with respect to each
other can be calculated. As summarized in Table 2, the inter-manufacturer precision of the
samplers was considered to be excellent for all three metrics (PM2 5, PMc: and PM10) at all three
sampling sites. Calculating the PMc concentration as the numerical difference between collocated
designated FM10 and PM, , FRMs did not produce any zero or negative PMc concentrations.
Table 2. Inter-manufacturer precision of the collocated FRM samplers.
| Metric
Gary, M
Phoenix. AZ
Riverside, CA
pm,5
1.5%
3.4%
3.1%
| PMc
5.7%
3.6%
4.1%
r fm10
2.4%
3.3%
2.9%
Page 11 of 20

-------
With the exception of a pump failure and a faulty ambient temperature sensor connection,
few functional problems were experienced with the eight FRM samplers despite the wide range of
environmental conditions experienced during the study. The three performance audits conducted
at each sampling site revealed that the FRMs generally maintained their flow rate, temperature,
and pressure calibrations within the required specifications. Overall data capture rate for the
FRM samplers during the three site study was determined to be 99%.
R&P Dlcbotomous Samplers
Only two operational problems were experienced with the four R&P Model 2025
sequential diehotomous samplers during the study. In Gary, a faulty cassette seal in one of the
dichots' coarse channels caused the majority of the coarse aerosol to bypass the collection filter.
As a result, the coarse particle mass concentration measured by this instrument was significantly
less than that measured by the other collocated dichots. The data for this sampler's coarse
channel was thus invalidated. The second problem experienced with the Model 2025 dichots
occurred during the latter half of the Phoenix tests where significantly low PM25 and PM10
measurements were obtained by one of the dichots. At the completion of the Phoenix'tests, this
behavior was explained by the discovery of a dense spider web in the dichot5 s size selective inlet.
The PMz5 and PM10 measurements for this instrument were thus invalidated for 17 of the 30
sampling events. At all sites, invalid data were not used to calculate daily aerosol mass
concentrations nor used to estimate intrasampler precision. Discounting the invalid data obtained
due to the presence of the spider web, overall data capture rate of the dichots during the study
was 98%. Performance audits of the Model 2025 . dichots indicated that they maintained their
flow rate, temperature, and pressure calibrations within the required specifications.
Table 3 summarizes the field performance of the Model 2025 dichots at all three sites in
comparison to the collocated FRM samplers. As the table indicates, excellent intrasampler
precision was observed for the R&P dichots at all three sites for all three metrics. As an example,
the precision (expressed as the coefficient of variation) in Gary for PM2 5, PMc, and PM10
concentrations was determined to be 3.8%, 3.2%, and 1.9%, respectively. The largest coefficient
of variation (4.2%) was observed in Phoenix for measurement of PMc aerosols.
Table 3. Performance of the R&P 2025 Dichot versus the FRM.
Metric
Performance Criteria
Gary, IN
Phoenix, AZ
Riverside, CA
PMm
Dichot CV
3.8%
2.3%
1.3%

Regression Equation
(Dichot vs. FRM)
Dichot = 0.99*FRM
+ 0.0
Dichot = 1,24*FRM - '
1.6
Dichot = 0.998*FRM
+ 0.0

Coefficient of
determination (R2)
0.998
0.97
0.995
|
Mean Dichot/FRM Ratio
1.00
1.09
1.00
Page 12 of 20

-------
Metric
Performance Criteria
Gary, IN
Phoenix, AZ
Riverside, CA
PMc
Dichot CV
3.2%
4.2%
.1.7%

Regression Equation
(Dichot vs. FRM) •
Dichot = 0.87*FRM
+ 0.39
Dichot = 0.70*FRM
. +5.0
Dichot = 0.95*FEM
+ 0.25

Coefficient of .
determination. (Rz)
0.969
0.98
0.98

Mean Dichot/ERM Ratio
0.90
0.79
0.96

Dichot C V
1.9%
3.0%
1.2%

Regression Equation
(Dichot vs. FRM)
Dichot = 0.95*FRM -
0.47
Dichot = 0.75*FRM
+ 5.9
Dichot = 1.00*FRM
-1.21

Coefficient of
determination (R2)
0.981
0.98
0.99

Mean Diehot/FEM Ratio
0.94
0.84
0.97
la Gary and Riverside, the PM,_, concentrations measured by the R&P dichots agreed
almost exactly with concentrations measured by the collocated FRM samplers. In Phoenix,
however, the dichots consistently over-predicted the PM-i 5 concentration by about 9%. This
over-measurement is hypothesized to be due to the inadvertent intrusion of coarse mode aerosols
into the fine channel, which has been known to occur in virtual impactors4.
The Model 2025 dichots consistently under-measured PMc concentrations at all three sites
although results were highly correlated (mean R2 equaled 0.976). A high coefficient of
determination at a site indicates that the sampler's behavior, with respect to the collocated FRMs;
was very consistent during the 30 days of testing. For PMc, mean sampler to FRM ratios at Gary,
Phoenix, and Riverside were determined to be 0.90, 0.79, and 0.96, respectively. Summing the
dichot's measured PM2 5 and PMc concentration to estimate the PM10 concentration, it was
observed that mean sampler to FRM ratios for PM10 in Gary, Phoenix, and.Riverside were 0.94,
0.84, and 0,97, respectively. For Phoenix, therefore, 16% of the aspirated PM10 aerosol mass
cannot be accounted for when
compared to the collocated PM10
FRM samplers. The consistency
of this behavior in Phoenix is
illustrated in Figure 4. Although
results are preliminary, recent
follow-up testing in Phoenix has
suggested that this particle loss
may occur during posLsamplirig
transfer of the dichot's coarse
filter cassette from the sampling
position to the storage position.
R&P is currently investigating
engineering solutions to address
Figure 4. Timeline of R&P dichot versus FRM PM10
concentrations in Phoenix.
250
o
E
i
8
s
Q.
50
15
, Sample Day
25

-------
the observed measurement bias.
R&P Coarse TEOM Samplers
Few operational problems were experienced with the three R&P coarse TEOM monitors
during the three site study. The exception occurred during the Riverside testing where the third
coarse TEOM monitor consistently measured about 17% higher than the other two coarse TEOM
units, which agreed extremely well with each other. The exact reason for the consistent difference
between the third unit and the other two units is not known but may have been an operational
problem associated with the TEOM control unit itself. In two successive tests, exchanging the
inlets and virtual impactors between units three and two did not appear to correct the noted
discrepancy. For purposes of calculating instrument precision at Riverside, therefore, data from
TEOM unit three was not used.
Table 4 summarizes the field performance of the R&P coarse TEOM monitors at all three
sampling sites in comparison to the collocated FRM samplers. Considering that these are
automated samplers which provide both sampling and mass analysis, excellent ktra-manufacturer
precision was observed for the three coarse TEOM monitors at all three sites. The coarse TEOM
precision at Gary, Phoenix, and Riverside was observed to be 4.4%, 6.6%, and 1.7%,
respectively.
Table 4, Performance of the R&P Coarse TEOM versus the FRM.
Metric
Performance Criteria
Gary, IN
Phoenix. AZ
Riverside, CA
PMe
TEOM CV
4.4%
6.6%
1.7%

Regression Equation
(TEOM vs. FRM)
TEOM - 0.68*FRM
+ 0.2
TEOM= 0.79*FRM
+ 12.8 '
TEOM 0.74*ERM
-0.64

Coefficient of
determination (R*)
0.982
0.951
0.948

Mean TEOM/FRM Ratio
0.69
1.05
0.76
Page 14 of 20

-------
At the Gary and Riverside field sites, the coarse TEOMs produced PMc values which were
consistently lower than those measured by the collocated FRMs. On average, the coarse TEOMs
provided PMc measurements that were 31% and 24% lower than the FRMs in Gary and Riverside,
respectively. This underestimation may be partly due to the fact that the sampler's inlet reportedly
provides an internal cutpoint closer to 9 (xm than its 10 |im design outpoint5. Note from the table
that the data is strongly correlated for Gary and Riverside and that near zero intercepts were
observed for regressions of the coarse TEOMs versus the collocated FRMs. The timeline
presented in Figure 5 illustrates that the coarse TEOM monitors track the FRMs well but
consistently provide an under-measurement of PMc concentrations. Based on the Mgh coefficient
of determination in Gary of 0.982,
this behavior was very consistent as
a function of concentration during
the 30 day test period.
Better agreement between
the coarse TEOMs and the FRM
was observed during the May to
June 2003 tests conducted in
Phoenix. For these tests, the
coarse TEOMs provided PMc
concentrations that averaged 5%
higher than those measured by the
collocated FRM samplers. As"
depicted in Table 4, however, the
slope and intercept for the TEOM
versus FRM regression deviated
significantly from one and zero,
respectively.
Tisch SPM-613D Dicbotomous Beta Gauge Monitors
No significant operational problems were encountered during field Operation of the Tisch
SPM-613D samplers at the three sampling sites. Overall data capture rate was near 100% at all
three sites.
TableS summarizes the performance of the three Tisch units in comparison with the
collocated FRM samplers. Inspection of the table reveals that precision of the samplers was
generally good for all three metrics at all three sampling sites. In general, Mgherintra-sampler CY
values (i.e. less precision) were observed during measurement of PMc concentrations than during
measurement ofPM25 concentrations.
Table 5, Performance of the Tisch SPM-613B Beta Gauge Dichot versus the FKM.
Figure 5. R&P coarse TEOM versus FRM PMc
concentrations in Gary, IN.
2 50
35 10
20
25
Page 15 of 20

-------
Metric
Performance Criteria
Gary, IN
Fhoeiilx, AZ
Riverside, CA
fm2,5
Tisch CV
/ 7.1%
5.9%
4.1%

Regression. Equation
(Tisch vs. FRM)
Tisch. = 1.17*FRM +
1.6
Tisch = 2.03 *FRM -
3.4
Tisch=2.07*FRM +
6.9

Coefficient of
determination (R5)
0.945
0.946
0.904

Mean Tisch/FRM Ratio
1.26
• 1.70
1.65
FMc
Tiscli CV
10.5%
9.5%
5.8%

Regression Equation
(Tisch vs. FRM)
Tisch = 0.885*FRM
+ 0.34
Tisch = 0.92*FRM +
6.0
Tisch = 1.17*FRM-'
2,7

Coefficient of
determination (H2)
0.978
0.995
0.957

Mean Tisch/FRM Ratio
0.91
1.04
1.08
PMI0
Tisch. CV
4.3%
7.4%
'• 3.5%

Regression Equation
(Tisch. vs. FRM)
Tisch = 1.02*ERM +
2.5
Tiscli = 1.02*ERM + .
7.8
Tisch = L53*FRM -
10.6

Coefficient of
determination (R2)
0.987
0.996
0.88

Mean Tisch/FRM Ratio
1.09
1.16
1.29
At all three sites, the Tisch SPM-613D units tended to significantly'over-estimate the PM2j5
concentrations when compared to the collocated PM25 samplers. For PM2 5 measurements, the •
mean sampler to FRM ratio at
Gary, Phoenix, and Riverside was
calculated as 1.26, 1.70, and 1.65,
respectively. This over-estimation
was quite consistent as illustrated
in Figure 6 which plots the
performance of the Tisch3 s PM2-J
concentrations versus those of the
collocated PM25 FRM samplers.
As was the case for the R&P
dichot. it is hypothesized that this
over-estimation might be due, in
part, to the inadvertent intrusion of
coarse mode particles into the
sampler's fine mode channel. This
hypothesis is supported by the feet
that larger overestimations occur at
Figure 6. Tisch SPM-613D versos FRM PM2.S
concentrations in PIioenix, AZ.
45
S
1
1
I
c
o
i
35 ....
iT
15
30
0
5
1C
ZQ
25
Sample Day
Page 16 of 20

-------
sites with the lowest mean PM25/PM10 ratios. The fact that the Tisch sampler typically provides
PM10 concentrations higher than the collocated PM10 FRM samplers, however, may indicate that
other measurement uncertainties may be responsible for the observed PM2 5 measurement bias.
The Tisch SPM-613D units provide more accurate measurements of ambient PMc
concentrations than PM25 concentrations. For PMc measurements, the mean sampler to FRM
ratio at Gary, Phoenix, and Riverside was calculated as 0.91, 1.04, and 1.08, respectively.
Consistency of this performance during the month-long sampling at each site is demonstrated by
the high coefficient of determination (0.978, 0.995, and 0.957, respectively) obtained during
sampler versus FRM regressions.
TS1 Inc. Model 3321 APS '
Few problems were experienced with the two TSI Model 3321 APS units during the
course of the field tests. The exception occurred approximately halfway through the field sampling
in Phoenix when the response of APS Unit 2 began to deviate substantially from that of Unit 1.
During the units' subsequent return to the manufacturer for cleaning, a circuit board within Unit 2
was diagnosed as faulty and was replaced. Data from this unit during the second half of the
Phoenix tests, therefore, were not used in comparing the performance of the APS units to that of
the collocated FRM samplers. Overall data capture rate for the APS units during the three city
study was 85%.
A summary of the performance of the APS units during this study is provided in Table 6.
With the exception of results obtained in Gary during which a few day's results tended to skew
precision calculations, the precision between the two APS units was generally good. This general
level of agreement is illustrated in Figure 7.
Table 6. Performance of the TSI APS versos the FRM.
Metric
Performance Criteria
Gary, IN
Phoenix, AZ
Riverside, CA
PMc
Sampler CV
16.8%
2.2%
8.5%

Regression Equation
(Sampler vs. FRM)
APS = 0.42*FRM +
0.48
APS = 0.56*FRM -
0.2
APS = 0.66*ERM-
2.3

Coefficient of determination
(R2)
0.80
0.99
0.82

Mean. Sampler/FRM Ratio
0.42
0.55
0.58
The APS units tended to under-predict the PMc concentration when compared to
measurements provided by the FRM samplers. Mean sampler to FRM PMc ratios for Gary,
Phoenix, and Riverside were determined to be 0.42, 0.55, and 0.58, respectively. As a rule, the
APS units tended to track the FRMs on a daily basis but under-measure PMc concentrations by
about a factor of two. This behavior appeared to be relatively independent of sampling site or
sampling day. Regressions of APS performance versus the collocated FRMs indicated that
regression intercepts were close to zero at all three sampling sites. Although the exact reason for
Page 17 of 20

-------
the under-measurement is not
known, the field behavior of these
units is consistent with laboratoiy
tests of the Model 3321 APS
conducted by Peters, et alA
SUMMARY
1.
Figure 7. .APS versus FRMPMc concentrations in Gary,
IN.
IT
Is
i
2.
£
m iflean FRS! F5!c
-o—APS i PWc
-Ar-APS 2 P*r5e
Sample Day
Through coordination with
state and local air
monitoring agencies, the
study sites selected met the
study's siting objectives
well and challenged the
candidate samplers with a
wide range of aerosol size
distributions, aerosol
concentrations, and meteorological conditions. Relatively few operational problems were
experienced with the sampling equipment and the overall data capture rate for the study
exceeded 95%. Prestudy, midstudv, and poststudy performance audits conducted at each '
sampling site revealed that the samplers typically held their calibrations well during the
month-long field tests. The involvement and cooperation of the various sampler
manufacturers was a key factor in the study's ability to "successfully determine the inherent
performance of the samplers.
The filter-based, integrated samplers involved in the study provided precise test results at
all three sampling sites. For the FRM samplers, the mean inter-manufacturer coefficient of
variation for PM2j3 PMc, and PM10 was 2,7%, 4.5%, and 2.9%, respectively. fntra-
manufacturer precision of the three R&P Model 2025 dichotomous samplers for PM2_5,
PMc, and PM10 measurements was 2.5%, 3.0%, and 2.0%, respectively. Effective
shipping protocols resulted in negligible particle loss during transport of collected aerosol
samples from each sampling site to the RTP weighing facility.'
With the exception of Phoenix where coarse particles may have intruded into the
samplers' fine channel, the R&P dichots typically provided PM2 5 measurements which
agreed closely with the collocated PM2 5 FRM samplers, hi regressions versus the
collocated FRMs, all R&P dichot test results were highly correlated. The R&P dichots,
however, underestimated PMc concentrations at all sampling sites and had a 21% under-
measurement recorded at the Phoenix site. Mass balance calculations-revealed that 16% of
the aspirated PM10 mass in Phoenix is not accounted for during subsequent gravimetric
measurement of fine and coarse channel filters. Recent tests have indicated that loss of
coarse mode aerosols during the sampler's automated, post-sampling movement of the
coarse particle cassette to the sample storage position may account for the observed bias.
Page 18 of 20

-------
4.	With the exception of the problem noted during the Riverside tests, excellent inter-
manufacturer precision of the R&P coarse TEOM samplers was observed at all three
sampling sites and no operational problems were encountered with the samplers. However,
with the exception of the Phoenix tests, the coarse TEOM tended to underestimate the
PMc concentration by as much as 30%. The high correlation between the coarse TEGMs'
response versus the coEocated FRMs indicated that this performance was very consistent
from one sampling event to another.
5.	' The Tisch SPM-613D samplers provided precise, highly correlated test results at al three
sites for PM2.5, PMc, and PM10 measurements. Although performance varied by site, the
Tisch units generally provided PMc measurements within 10% of that of the collocated
FRM samplers. However, the SPM-613D units consistently provided PM,S concentrations
significantly higher than the collocated PMj 5 FRM samplers. As an example, the mean
overestimation in PM2 5 concentrations at the Phoenix site was 70%. Similar to the
behavior of the R&P dichot, intrusion of coarse particles into the Tisch unit's fine channel
may account for the majority of this observed behavior.
6.	With the exception of a single electronics failure, the two TSI Model 3321 units appeared
to function well and provided acceptable levels of precision. Although the APS units were
observed to track the PMc FRM concentrations well, they typically underestimated PMc
mass concentrations by a factor of two at all sampling sites. This field behavior is
consistent with previous laboratory tests of the Model 3321 conducted under controlled
conditions.
ACKNOWLEDGMENTS .
The authors wish to thank personnel from the Indiana Department of Environmental Management,
Maricopa County Environmental Services Department, and the University of California-Riverside
Agricultural Operations Facility for providing the sampling sites used in this study. Contents of
this paper have not yet been reviewed in accordance with the United States Environmental
Protection Agency's peer and administrative review policies nor approved for presentation and
publication. Mention of trade names or commercial products does not constitute endorsement or
recommendation by the Agency.
REFERENCES
1.	USEFA (1997). Part 50 - .National Primary and Secondary Ambient Air Quality
Standards, Federal Register, Vol. 62, No. 138, p387ll. July 18,1997.
2.	Noble, C.A.; VanderpooJ, R.W.; Peters, T.M.; McElroy, F.F.; Geromill, D.B.; Wiener,
R.W. Aerosol Set TechnoL 2001,34,457-464.
3.	Misra, C.; Geller, M.; Shah, P.; Sioutas, C., Solomon, P. J of Air Waste Manage. Assoc.
2001, 51, 1309-1317.
Page 19 of 20

-------
4.	Allen, G.A.; Ok, J. A.; Koutrakis, P.; Sioutas, C. J. of Air Waste Manage. Assoc. 1999,
49,133-141.
5.	Misra, C.; Gcllcr, M.; Sioutas, C.: Solomon, P. AerosolScl Technol2003, 37,271-281.
6.	Peters, T.M.; Leitk D. Journal of Aerosol Science 2003, 34, 627-634.
KEY WOMBS
Aerodynamic
Aerosols
Ambient
Beta gauge	.	•
Dichotomous
FRM
Particle
pMi5 .;
PMc
PM10
TROM
Page 20 of 20

-------
TECHNICAL REPORT DATA
1. Report No,
4. Title and Subtitle
Multi-Site Performance Evaluations of Candidate Methodologies for Determining
Coarse Particulate Matter Concentrations
5. Report Date
6. Performing Organization Code
7. Author(s)
Vanderpoo!, Robert', Eiiestad, Thomas1, Solomon, Paul1, Harmon, Mary1,
Hanley, Tim1, Scheffe, Richard1, Murdoch, Robert2, Natarajan, S,2, Noble,
Christopher2. Ambs, Jeff3, Sem, Gil4, and Tisch, John5
1.	US EPA, RTP, NC
2.	Research Triangle Institute, RTP, NC
3.	Rupprecht & Pattashrrfck Co., Inc., Albany, NY
4.	TSI Inc., St. Paul, MN
5.	Tisch Environmental, Inc., Cleves, OH
8. Performing Organization
Report No.
9. Performing Organization Name and Address
USEPA/QRD/NERL/HEASD/PMRB
109 TW Alexander Drive
RTP, NC 27711
10. Program Element No.
11. Contract/Grant No.
68-D-00-206
12. Sponsoring Agency Name and Address
USEPA/ORD/NERL/HEASD/PMRB
109 TW Alexander Drive
RTP, NC 27711
13. Type of Report and Period
Covered
14. Sponsoring Agency Code
15. Supplementary Notes
16. Abstract Comprehensive field studies were conducted to evaluate the performance of sampling methods for measuring the coarse
fraction of PM10 in ambient air. Five separate sampling approaches were evaluated at each of three sampling sites. As the primary basis
of comparison, a discrete difference method was used which employs two designated FRM samplers, one to measure PMjj and the
other PMl0. The numerics! difference of these reference method concentrations (PM^-PM^) represented an estimate of PMc. A
second sampling approach involved a sequential dichotomous sampler, which provided both PM^j and PMc measurements. In both of
these filter-based, time-integrated measurement approaches, the collected aerosol mass was analyzed gravimetrically in the laboratory
under controlled conditions. Three continuous coarse particle samplers that measure PMc directly with a time resolution of 1 hour or
less were also evaluated. One such sampler was a commercially available system based on beta attenuation, the second was based on
TEOM technology. Both of these measurement approaches used dichotomous virtual impactors for separating fine and coarse particles.
The third real-time sampler evaluated was an aerodynamic particle sizer (APS) that measures the aerodynamic diameter of individual
particles, calculates the mass of the particle based on an assumed particle density, then sums the mass within the size range of interest to
estimate the PMc mass concentration.
Sampling sites and timing of the studies were selected to provide diverse challenges to the samplers with respect to aerosol
concentration, aerosol particle size distribution, and aerosol composition. Results from performance evaluations of the candidate PMc
samplers at Gary, IN, Phoenix, AZ, and Riverside, CA were presented.
17. KEY WORDS AND DOCUMENT ANALYSIS

-------
TECHNICAL REPORT DATA j
A. Descriptors Aerodynamic, Aerosols, Ambient, Beta gauge,
Dichotomous, FRM, Particle, I'M,«, PMc, I'M,,. TEOM
B. Identifiers / Open Ended
Terms
C. COSATI

18. Distribution Statement
19. Security Class (This
Report)
21. No. of Pages
20. Security Class (This
Page)
22. Price
Fonn Available: Network Neighborhood\Knight\Groups\HEASD\Fonns\Techmcal-Repoit-Data-2220-l
Page 1 of 2
INSTRUCTIONS
L	REPORT NUMBER
Insert the EPA report number as it appears on the cover of the publication.
2.	LEAVE BLANK
3.	RECIPIENTS ACCESSION NUMBER
Reserved for use by each report recipient.
4.	TITLE AND SUBTITLE
Title should indicate clearly and briefly the subject coverage of the report, and be displayed prominently. Set subtitle, if used, in smaller type or
otherwise subordinate it to main title. When s report is prepared in more than one volume, repeat the primary title, add volume number and
include subtitle for the specific title.
5.	REPORT DATE
Each report shall carry a date indicating at least month and year. Indicate the basis on which it was selected (e.g. date of issue, date of approval,
date of preparation, etc.)
6.	PERFORMING ORGANIZATION CODE
Leave blank.
7.	AUTHOR©
Give name(s) in conventional order (John R. Doe, J. Robert Doe, etc.). List author's affiliation if it differs from the performing organization.
8.	PERFORMING ORGANIZATION REPORT NUMBER
Insert if performing organization wishes to assign this number.
9.	PERFORMING ORGANIZATION NAME AND ADDRESS
Give name, street, city, state, and ZIP code. List no more than two levels of an organizational hierarchy."
10.	PROGRAM ELEMENT NUMBER
Use the program element number under which the report was prepared. Subordinate numbers may be included in parentheses.
11.	CONTRACT/GRANT NUMBER
Insert contract or grant number under which report.was prepared.
12.	SPONSORING AGENCY NAME AND ADDRESS
Include ZIP code.
13.	TYPE OF REPORT AND PERIOD COVERED
Indicate interim final, etc., and if applicable, dates covered.
14.	SPONSORING AGENCY CODE
Insert appropriate code.
15.	SUPPLEMENTARY NOTES
Enter information not included elsewhere but useful, such as: Prepared in cooperation with, Translation of, Presented at conference of. To be
published in, Supersedes, Supplements, etc.
REPRODUCED BY; NIK
U.S. Department of Commerce
National Technical Inforrnaiwn Service
Sprinnfialrl, VirfliniH 22161

-------