Recommendations for Great Lakes
Air Toxics Monitoring
William Barnard
Robert Harless
Robert Lewis
Dale Pahl
Robert Stevens
Tom Dzubay
Alan Hoffman
Yaacov Mamane
Jack Shreffler
Joseph Walling
Atmospheric Research & Exposure Assessment Laboratory
Research Triangle Park, N.C. 27711
May 1990
-------
ACKNOWLEDGEMENT
The research and analysis presented in this document was conducted during
1989 and 1990 for the Environmental Protection Agency's Great Lakes National
Program Office. A number of environmental science professionals representing
EPA's Great Lakes National Program Office and Office of Research and Development
(Atmospheric Research and Exposure Assessment Laboratory) participated in the
study and were instrumental in ensuring its success. Their expertise and
assistance are gratefully acknowledged.
Field Monitoring Team
Milt Bowen, AREAL
Bobby Edmonds, AREAL
Alan Hoffman, AREAL
Mack Wilkins, AREAL
Bill Barnard, AREAL
Bernie Bennett, AREAL
Ev Quesnell, AREAL
Bob Stevens, AREAL
Alf Wall, AREAL
Quality Assurance Team
Bill Mitchell, AREAL
Rocky Rhodes, AREAL
Data Analysis, Interpretation, and Report Preparation Team
Bill Barnard, AREAL
Aubrey DuPuy, ECL/OPTS
Tom Dzubay, AREAL
Steve Eisenreich, U. Minn.
Marie Collins, AREAL
^Don^Gat9sI) ISWS
Bob HarTess, AREAL
Alan Hoffman, AREAL
Ed Klappenbach, GLNPO
Bob Lewis, AREAL
•Chuck Lewis, AREAL
Yaacov Mamane, AREAL
Danny McDaniel, ECL/OPTS
Dale Pahl, AREAL
Joachim Pleil, AREAL
Shari Pricer, AREAL
Rocky Rhodes, AREAL
Jack Shreffler, AREAL
Bob Stevens, AREAL
Clyde Sweet, ISWS
Joe Walling, AREAL
Gary Foley, AREAL
Tom Hartlage, AREAL
-£d Klappenbach, GLNPO
Research Management Team
Dale Pahl, AREAL
Jack Shreffler, AREAL
GLNPO
-------
CONTENTS
LIST OF FIGURES ii
LIST OF TABLES ii
LIST OF APPENDICES iii
1.0 EXECUTIVE SUMMARY 1
1.1 Short-Term Atmospheric Characterization at the UWGB Monitoring
Site 2
1.2 Quality Assurance and Quality Control Review of GLNPO
Monitoring Site 2
1.3 Recommended Changes in the GLNPO Pilot Monitoring Design and
Operation 4
2.0 INTRODUCTION 6
3.0 GLNPO MONITORING PROGRAM 8
3.1 GLNPO Objectives 9
3.2 Samplers 9
3.3 Schedule and Operation 11
3.4 Calibration/QC Procedures 12
3.5 Collection Media 13
3.6 Field Blanks 13
3.7 Sample Analysis 14
4.0 AREAL MONITORING PROGRAM 14
4.1 Objectives 14
4.2 Samplers 15
4.3 Sampling Schedule and Operation 17
4.4 Calibration Procedures 29
4.5 Collection Media 31
4.6 Field Blanks 32
4.7 Sample Analysis 32
4.8 Field Data Handling System 37
5.0 QUALITY ASSURANCE AND QUALITY CONTROL REVIEW 37
5.1 Introduction 37
5.2 Great Lakes National Program Office 38
5.3 Atmospheric Research and Exposure Assessment Laboratory 42
6.0 ANALYSIS OF MONITORING RESULTS 44
6.1 Colocated Measurements 44
6.2 Modeling and Measurement Results 60
-------
LIST OF FIGURES
Figure 4-1. AREAL Sampler Placement at Green Bay Monitoring Site .... 18
Figure 4-2. AREAL Sampling Schedule 19
. Fijgi«_4j-3^AREAL Site Observation Log 20
f—Figure 4^4T~AREAl^field Data Sheet 22
Figure 4-5. Sampler Configuration for Pressurized Canister Sampling ... 24
Figure 4-6. AREAL Field Blank 25
Figure 6-1. Green Bay TSP Data Comparison 46
Figure 6-2. Green Bay TSP Data Comparison (Outlies Removed) 47
Figure 6-3. Green Bay Total Carbon Data Comparison 51
Figure 6-4. Time Series of Fine and Coarse Particle Mass for the Green . 63
Bay Site ^. . .
Figure 6-5. Plot of Low-Volume Versus High-Volume Dichotomous Sulfate . . 66
Data for the Green Bay Site
Figure 6-6. Plot of Low-Volume Versus High-Volume Dichotomous Mass ... 67
Concentration
Figure 6-7. A Time Series of the Organic Carbon Present in the Fine ... 69
and Coarse Fraction
Figure 6-8. Ionic Composition Data for the Sulfate, Sulfite and Nitrate . 70
for Fine Particles
Figure 6-9. Ionic Composition Data for the Sulfate, Sulfite and Nitrate . 71
for Coarse Particles
Figure 6-10.Relative Contribution of Various Species to the Total Mass . 74
in the Fine Fraction
Figure 6-11.Relative Contribution of Various Species to the Total Mass . 75
in the Coarse Fraction
Figure 6-12.Photomicrograph of Typical Fields of View for Coarse and Fine 77
(Bottom) Fraction Collected on September 21, 1989
Figure 6-13.A Close Up Photomicrograph of Coarse Particles Showing an . . 79
Iron Sphere in the Center and Few Spores Linked to Each Other.
Figure 6-14.X-ray Spectra of the Iron Sphere Shown in Figure 6-10 .... 80
Figure 6-15.X-ray Spectra of a Submicrometer Sulfate Particles Shown . . 81
in Figure 6-10
Figure 6-16.Components of Average Fine Fraction Mass Deduced by CMB . . 85
Model for Samples Collected from September 1 to 27, 1989. .
LIST OF TABLES
TABLE 3-1. GLNPO SAMPLERS DEPLOYED AT THE GREEN BAY MONITORING SITE . . 10
TABLE 4-1. AREAL SAMPLERS DEPLOYED AT GREEN BAY MONITORING SITE .... 16
TABLE 5-1. COMPLETENESS OF SAMPLING FOR AREAL MONITORING AT GREEN BAY . 43
TABLE 6-1. GREEN BAY PARTICULATE MATTER DATA COMPARISON AREAL/GLNPO . . 45
ii
-------
TABLE 6-2. TOTAL ORGANIC CARBON MEASUREMENTS AREAlT-VS,, GLNPO (ug/m3)
TABLE 6-3. DIRECTIONAL PM DATA COMPARISON (SW QUADRANT)
TABLE 6-4. DIOXIN-FURAN DATA FOR GREEN BAY
TABLE 6-5. PCB-PESTICIDE DATA FOR GREEN BAY
TABLE 6-6. AREAL RESULTS FROM GREEN BAY PRECIPITATION SAMPLES?
TABLE 6-7. ESTIMATED DETECTION LIMITS FOR PRECIPITATION SAMPLES .
TABLE 6-8. RECOVERIES FROM STANDARD SOLUTIONS OF KNOWN COMPOSITION
TABLE 6-9. AVERAGE RECOVERIES FROM ONE ANALYSIS EACH OF TWO . . .
TABLE 6-10. REGIONAL LABORATORY RESULTS
-TABLE 6-11. REGIONAL LABORATORY RESULTS FROM GREEN
TABLE 6-12. AVERAGE COMPOSITION OF AEROSOL AT GREEN BAY, WI ....
TABLE 6-13. MASS CONCENTRATION, IONIC SPECIES AND CARBON FOR FINE .
-TABLE 6-14. MASS CONCENTRATIONftlONIC SPECIES AND CARBON FOR COARSE .
TABLE 6-15. ANNULAR DENUDER MEASUREMENTS COLLECTED AT THE UWGB SITE
TABLE 6-16. CHEMICAL-SIZE DISTRIBUTION OF COARSE PARTICLES
50
50
52
54
55
56
57
58
59
59
61
64
64
73
78
ill
-------
APPENDIX A.
APPENDIX B.
APPENDIX C.
APPENDIX D.
APPENDIX E.
APPENDIX F.
APPENDIX G.
APPENDIX H.
APPENDIX I.
APPENDIX J.
APPENDIX K.
LIST OF APPENDICES
AREAL Sampling Procedure for Dichotomous Samplers A-l
AREAL Sampling Procedure for Hi-Volume Virtual Impactor . . B-l
AREAL Sampling Procedure for PUF/PS-1 C-l
AREAL Sampling Procedure for VOC Samplers D-l
AREAL Sampling Procedure for Hi-Vols £-1
List of VOC Compounds Analyzed F-l
Dioxin/Furan Analysis Procedures and QA Plan G-l
Results of ISWS Lab Audit H-l
Results of CRL/Bionetics Lab Audit 1-1
Dioxin/Furan Analysis and QA/QC Results J_l
XRF Analysis Results K-l
iv
-------
1.0 EXECUTIVE SUMMARY
During 1&88-1989, the EPA's Great Lakes National Program Office (GLNPO)
established a pilot atmospheric monitoring network at the University of
Wisconsin, Green Bay Campus (UWGB) . The objectives of this network are to
collect and measure atmospheric loadings of nutrient and toxic compounds to the
Great Lakes watershed and to identify the sources of these compounds .
Eventually, this pilot monitoring program will be expanded by GLNPO and
its Canadian counterpart to temporally and spatially characterize nutrient and
toxic loadings over the entire Great Lakes watershed. However, because the
sampling and analytic techniques appropriate to this objective require testing
- — and evaluation, the UWGB site is being operated during Ifjj|>_andj.9jt6' as a pilot-
scale air monitoring network.
In June 1989, the GLNPO asked EPA's Atmospheric Research and Exposure
Assessment Laboratory (AREAL) to evaluate the pilot network. To gather
information for this evaluation, AREAL conducted a short-term field study at the
UWGB site. Additionally, AREAL conducted performance and systems audits to
evaluate the components of the GLNPO monitoring/instrument systems and the field
and laboratory quality control practices. The information developed by these
studies is presented in this report to:
o provide a short-term ambient air characterization at the UWGB site
to help determine the appropriateness of the current pilot network
data collection and source apportionment objectives;
o evaluate the pilot network quality assurance project and program
plans as well as the quality assurance and quality control (QA/QC)
practices used in the field and by the GLNPO analytic laboratories ;
and
o recommend changes in the pilot network design and operation to
facilitate the achievement of GLNPO measurement and source
apportionment objectives.
The following sections of this chapter summarize AREAL 's findings in each
of these three areas .
-------
1.1 Short-Term Atmospheric Characterization at the UWGB Monitoring Site
During September 1989, AREAL conducted a field monitoring study located
at the UWGB site and coordinated with the GLNPO instruments and sampling
schedule. The AREAL and GLNPO data collection objectives and monitoring
instruments are summarized in Tables 4-1 and 3-1 respectively; the sampling
schedule is presented in Figure 4-2.
Analysis of the AREAL measurements indicates that, during this short-term
study, ambient atmospheric concentrations of particulate matter, acid aerosol
species, precipitation metals, PCBs, PCDDs, PCDFs, aldrin, dieldrin, and volatile
organic compounds were either at background levels (typical of rural/suburban
concentrations measured by AREAL in other parts of the country) or, in the case
of aldrin and dieldrin concentrations, were below the limits of instrument
quantitation. A detailed discussion of these analytical results is presented
in Chapter 6 of this report.
Chemical mass balance and multiple linear regression models were used to
apportion the chemical species data acquired during the field monitoring study
to potential sources of the species. Identified sources included combustion and
incineration stack gases, pulp and paper mill emissions, motor vehicle emissions,
and metal processing industry emissions. Additional sources of the chemical
\ -^ species include crustal material, burning of conifers C?2> , slash burning), and
plywood veneer drying. More than half of the fine particle mass collected is
of a regional, rather than local, origin and is composed of sulfate and related
anions typical of fine particles in the eastern half of the United States.
Source apportionment of volatile organic compounds collected by AREAL was
precluded because of QA/QC problems identified with the sampling protocol used
at the UWGB site. Additional source apportionment analysis and discussion is
presented in Chapter 6 of this report.
1.2 Quality Assurance and Quality Control Reviev of GLNPO Monitoring Site
The quality assurance and quality control (QA/QC) review consisted of a
systems audit, the coordinated field monitoring study described in Section 1.1
-------
above, and a performance audit. The results of this review are discussed in
detail in Chapter 5 of this report.
The AREAL systems audit included an evaluation of the design of the UVGB
site and the pilot network as expressed in the quality assurance program and
project plans as well as the components of the monitoring, measurement, and
analysis systems in the field and laboratory. The audit determined that
fundamental "systems design documents," the quality assurance project and program
lans, had not been prepared for either the UWGB site or the pilot network.
An on-site audit of the Bionetics Laboratory (one of the two analyzing UWGB
samples) indicated that the laboratory was adhering to all the established QA/QC
requirements. A systems audit of the other laboratory, the Illinois State Water
Survey (ISWS) Laboratory, was not conducted as a part of the AREAL systems audit
for GLNPO. However, the ISWS is routinely audited as a part of AREAL's acid
precipitation monitoring programs. Recent audits of the ISWS Laboratory (spring
and fall of 1989) indicate that this laboratory also does an excellent job of
adhering to QA/QC requirements. An evaluation of sample and data handling
procedures identified the need for identifying and tracking collected samples
from the initial monitoring instrument through laboratory analysis to final data
base. This objective could be achieved with the use of a bar-coding and computer
tracking system. Finally, an on-site audit of the UWGB site utilizing criteria
developed for the State and Local Air Monitoring Sites (SLAMS) indicated that
both the site and individual monitoring instruments are appropriately located,
with one exception. The proximity of the dirt road at the UWGB site may
contribute some amount of coarse particulate matter to particulate samplers
during high wind velocity or heavy traffic episodes.
On September 8, 1989, AREAL conducted a performance audit at the UWGB site.
The audit consisted of flow measurements. With the exception of the coarse flow
from one dichotomous sampler, all flows were in excellent agreement with AREAL's
standards. Other performance audits included duplicate audit canisters for
volatile organic compounds and spiked PUF cartridges. Results from the VOC audit
canisters identified a high positive bias for 5 of the 17 compounds, and indicate
that interpretation of the VOC data during the AREAL field monitoring study would
be questionable.
-------
1.3 Recommended Chances in the GLNPO Pilot Monitoring Design and Operation
After completing the field monitoring study, source apportionment analyses,
and QA/QC reviews summarized above, AREAL assembled a team of environmental
professionals in atmospheric monitoring, deposition, and chemistry to evaluate
the findings of these reviews and to recommend changes in the network design or
operation. The team confined its recommendations to those that would improve
the achievement of GLNPO measurement and source apportionment objectives and the
scientific credibility of the results of attendant analyses. The GLNPO
objectives were determined through personal contacts with GLNPO managers as well
as through analysis of the Great Lakes Water Quality Agreement (As Amended in
1987). the Five Year Program Strategy for the Great Lakes National Program
Office, and the Mass Balancing of Toxic Chemicals In the Great Lakes; The Role
of Atmospheric Deposition. AREAL's recommendations are summarized in five key
areas:
Recommendation 1. There is a fundamental need for GLNPO quality assurance
project and program plans (QAPP's) for the pilot and full-scale (Great Lakes
Basin-Wide) air monitoring network. These plans must identify data collection,
measurement, and analysis objectives and relate them to the major goals of the
GLNPO studies of the Great Lakes. Without a QAPP, it will not be possible to
make an adequate assessment of the design, implementation, or ultimate success
of the GLNPO atmospheric monitoring, deposition, and mass balance programs.
Recommendation 2. The premise underlying the selection of the UWGB site
for the pilot monitoring program is that it would provide a representative
location for the testing and evaluation of sampling and analytic techniques.
However, the measurement data collected during the AREAL field study contradict
this premise. Although these data are not representative of long-term
conditions, measured levels of PCBs, PCDDs, PCDFs, VOCs, aldrin, and dieldrin
were near or below instrument or analytical quantitation levels. This suggests
that data collection objectives or monitoring site selection criteria should be
evaluated before proceeding with network development. For example, the absence
of detectable concentrations of aldrin and dieldrin is consistent with both
current and recent agricultural practices in the United States. Perhaps these
-------
compounds should be eliminated from the network data collection objectives and
replaced with pesticides and herbicides currently used in the U. S. and Canada.
Recommendation 3. AREAL's evaluation of the GLNPO monitoring instruments
indicates that their selection has precluded source apportionment analysis.
However, source apportionment should be fundamental to the design of the
monitoring network and to the interpretation of monitoring data and deposition
estimates. The instruments and analysis that AREAL used in its evaluation
demonstrate that source apportionment is practicable for aerosol and nutrient
4 6^f
\)o" -—^ compounds and for PCB's, P
-------
high volume filters;
/schedule at least two "intensive" studies during the next year of
I operation of the pilot monitoring network and during the first year
(of operation of the large-scale U.S.-Canadian network. These
intensive studies should be of 2- to 3-week duration with consecutive
12-hour «ampling periods to permit analysis of temporal variability
and to permit the development of detailed receptor modeling
relationships; and
schedule instrument sampling during routine (ie., non-intensive)
network operation for 24-hour periods every other day.
Recommendation 5. AREAL's experience with large monitoring networks
suggests the need for an automated system to identify and track samples through
the instrument, analysis, and data analysis phases of the network. Such a system
could be similar to the bar-code and computer tracking systems used for large
monitoring networks. Although this system is not essential for the small-scale
UWGfi site, it should be implemented on a pilot-scale before being deployed in
the full-scale network.
2.0 INTRODUCTION
The Great Lakes National Program Office (GLNPO) was created in 1978 to
manage the implementation of the United States' obligations under the Great
Lakes Water Quality Agreement of 1978. That Agreement between the United States
and Canada calls for a comprehensive ecosystem approach to the management of
Great Lakes water quality.
To provide information essential for proper water quality management,
GLNPO funds and directs extensive surveillance and monitoring activities in the
Great Lakes and surrounding watersheds. The surveillance activities include
routine sampling of water, fish tissue, and sediment. Air monitoring networks
are operated in the Great Lakes Basin by GLNPO and other federal and state
agencies to measure the quantity of pollutants entering the Basin from airborne
sources.
-------
Over the past ten years, this monitoring activity has indicated that
atmospheric transport and deposition may be an important, and in some cases the
dominant, pathway by which nutrient and toxic substances reach the Great Lakes.
Although there are many potential industrial, agricultural, and urban air
pollution sources, an understanding of the transport, transformation, and
deposition mechanisms for these substances must be developed before remediation
strategies can be devised and implemented.
None of these substances is yet present in the lakes in concentrations
known to be acutely toxic to organisms or disruptive to ecosystems in which
these organisms exist. Rather, concern focuses on the possible long term effects
of small quantities of numerous substances transported through the atmosphere
and deposited in the Great Lakes Basin. This concern resulted in the 1987
amendment to the Great Lakes Water Quality Agreement that reflects the increasing
awareness of the importance of the atmospheric pathway and expands the
environmental objectives of the Agreement to include atmospheric monitoring for
toxic air pollutants.
Responding to this amendment, GLNPO and its Canadian counterpart have
embarked upon the development and deployment of a Basin-wide monitoring network
for the measurement of atmospheric loadings of nutrient and toxic air pollutants
and for the identification and apportionment of their sources. Because sampling
and analytic techniques appropriate to this task require testing and evaluation,
this network is being tested during FY89-90 in a pilot-scale configuration at
a single site at the UUGB.
To assist in the evaluation of the pilot-scale site, GLNPO requested
assistance from the Atmospheric Research and Exposure Assessment Laboratory
(AREAL). AREAL participation in the pilot-scale network has included an on-
site collocated monitoring study, a review of the GLNPO quality assurance project
plan and data quality and collection objectives, and a QA/QC review of the GLNPO
laboratory program which supports the analysis of samples from the UWGB site.
These activities have been conducted to achieve two objectives:
o A determination of whether the network sampling and instrumentation design
7
-------
is appropriate for GLNPO measurement and source apportionment objectives;
and
o A determination of whether the network design and quality assurance project
plan can achieve GLNPO air and water quality management objectives: (1)
to provide data needed to develop and apply a mass balance model for toxic
and other pollutants in Green Bay and, eventually, in the larger Great
Lakes Basin, and (2) to provide information essential in developing
remediation strategies needed to reduce emissions of pollutants which are
significantly contributing to the deterioration of water quality in the
Great Lakes Basin area.
The results of the AREAL evaluation of the GLNPO pilot-scale network are
presented in following four sections of this report. Section three presents
the GLNPO network design and quality assurance objectives. Section four
discusses the AREAL collocated monitoring study conducted at the UWGB site during
September 1989. Section five discusses the results of the quality assurance
reviews of the GLNPO quality assurance project plan and laboratory analyses.
Section six presents, compares, and analyzes the results of the AREAL/GLNPO
monitoring study. Finally, the appendices provide details about the sampling,
analysis, and operating procedures for the instruments AREAL employed in the
September 1989 monitoring study.
3.0 GLNPO MONITORING PROGRAM
The Green Bay site is located on the UVBG campus about 1 km southeast of
the shore of Green Bay. Its coordinates are latitude: 44e31'59", longitude
87C54'44". The samplers are located in an open grassy field on a 30 x 10 m plot
surrounded by an 8-foot high cyclone fence. The area is open to the north,
south and west with trees and a low ridge about 150 m to the east. There are
paved roads with light traffic about 150 m to the east and 100 m to the west of
the site. The site is served by a dirt driveway used by service vehicles only.
The only obstruction at the site is a small utility shed (3x4x3 meters) in the
northeast corner. The site has 110 V electrical service and platforms for
sampling equipment.
8
-------
3.1 GLNFO Objectives
The GLNPO established the Green Bay sampling site to assist in providing
data needed to apply a mass balance model for toxic pollutants in Green Bay.
The mass balance approach requires the determination of system inputs, internal
transformations and storage and outputs. If all of the important terms can be
measured or estimated accurately, and if inputs, less storage and
transformations, equal outputs, then the system can be assumed to be adequately
understood so that contaminant concentrations can be accurately modeled.
Analysis of air and precipitation samples as well as meteorological measurements
are required to determine source/receptor relationships. In considering these
relationships, compounds that must be evaluated include/pesticides? PCBs .
dioxins/furans)1, and metals such as Pb, Zn, and Cd.
—
3.2 Samplers
Table 3-1 displays a summary of the types and quantities of samplers
deployed at the UW-GB monitoring site to meet the GLNPO objectives. The
following is intended to provide additional descriptive details.
Equipment at the site includes both precipitation samplers and three types
of high-volume air samplers. The precipitation samplers are MIC Model B samplers
(MIC Inc. , Thronhill, Ont.) modified for all-weather operation. The modifications
include an insulated enclosure underneath the sampler heated by a small space
-------
TABLE 3-1. GLNPO SAMPLERS DEPLOYED AT THE GREEN BAY MONITORING SITE
SAMPLER
MIC
(precipitation)
(GLADSITE)
TSP Hi vol
Modified
Hi vol
directional
Cascade Impactor
Wind speed/dir
T.R.H., SR
Rain
1
3
1
1
1
SAMPLING
FREQUENCY
1/14
V? .$,
1/6
1/14
1/3 mos.
cont.
cont.
cont.
SAMPLE
DURATION
14 days
cont.
I
24 hrs
14 days
7 days
cont.
cont.
cont.
SAMPLING
MEDIA
XAD-II
ANALYSIS
PCB's, dieldrin
Metals, nutrients
8x10 Quartz Mass, TOC, CE
SAMPLE
HANDLING
8x10 Qttaftz PCB's; dieldrin
XAD-II
Impactors
Particle sizing
^kfe&
10
-------
heater. Temperature is maintained at about 10-15*C, which provides enough heat
to melt snow falling into the catch basin. Rain or melted snow passes through
a 15-cm column of XAD-2 resin by gravity flow.
The air samplers used are standard high-volume samplers with automatic flow
controllers (Model GS2310 Accu-Vol, General Metal Works, Village of Cleves, OH).
An unmodified version of this sampler is used to collect total suspended
particles and total organic and elemental carbon in airborne particles. To
sample for the target chemicals, a modified version is used. The modifications
consist of adapters and an aluminum tube between the filter holder and the motor
designed to hold an 8.7 x 4.4 cm stainless steel resin cartridge capable of
holding 40-50 g of XAD-2 resin. The motor is also replaced with a 2-stage Lamb
motor (Model 115937, Amtek-Lamb, Kent, OH). The modified air samplers are also
fitted with automatic filter covers (G8550, Sample Saver, General Metal Works)
to prevent passive loading. Finally, another modified high-volume sampler is
fitted with a 4-stage cascade impactor (Model 234, Anderson Samplers, Inc.,
Atlanta, GA) for determination of particle size distribution.
In addition to the air and precipitation samplers, the site also has the
following meteorological monitoring equipment mounted on a 10 m tower: a solar
radiation sensor (LI 200S pyramometer, Ll-Cor, Lincoln, NE), temperature and
humidity sensors (Campbell Scientific, Logan, UT), wind speed and direction
sensors (Met-One, Grants Pass, OR), a standard Belfort rain gauge (Belfort
Instrument Co., Baltimore, MD), and a Belfort rain gauge fitted with a Nifer
wind shield. All meteorological sensors are automatically recorded every six
seconds by a Campbell 21X data logger (Campbell Scientific, Logan, UT) that
also calculates and records hourly averages.
3.3 Schedule and Operation
Sampling Schedule
The site is operated on a biweekly sampling schedule. All resin columns,
air cartridges and air filters are changed every other Tuesday. At the same
time, the meteorological data tape is changed. The buckets on the Belfort rain
11
-------
gauges are emptied weekly, if needed. The TSP/TC sampler takes a 24-hour sample
on the six-day U.S. EPA schedule and the cascade impactor sample collects a
seven-day sample on an occasional basis (approximately quarterly).
The three MIC samplers at the site are used to take biweekly duplicate
and field blank samples. The three modified high-volume samplers are operated
on a wind-directed basis by the Campbell data logger. In other words, each
sampler operates only when the wind direction is from a predetermined sector.
For instance, at Green Bay one sampler is programmed to sample air coming off
the bay while another samples air coming from the industrial areas of the city
of Green Bay. Finally, single cascade impactor and standard high-volume samplers
are also located at the site.
3.4 Calibration/QC Procedures
The calibration procedures and frequencies of checks vary with the type
of sampler being used. For the MIC sampler the amount of precipitation collected
in the bottle at the bottom of the sampler after the precipitation has passed
through the XAD-2 resin cartridge is compared to the amount of rain collected
by the Belfort rain gauge. This comparison is done on a biweekly interval;
however, the Belfort rain gauge chart is changed on a weekly interval.
The air sampling equipment is operated and calibrated according to the
manufacturer's recommendations. Flow rates of the high-volume samplers are
calibrated monthly using a standard General Metal Works manometer. The samplers
with resin cartridge vapor traps are set 566 £/min (20 ft3/min) and the TSP/TC
high-vol is set at 1133 £/min (40 ft3/min). A small reference manometer is used
as a qualitative check to verify normal operation between calibrations. The
meteorological monitoring equipment was calibrated during the initial set-up,
and it is rechecked quarterly. Where standard methods have been established,
such as for TSP and TOC measurements, the standard procedures of the American
Society for Testing and Materials (ASTM) are followed.
12
-------
3.5 Collection Media
Airborne particles are collected on high purity glass fiber filters
(Whatman EPM 2000, Whatman Ltd., Maidstone, UK). Airborne vapors and
precipitation samples are passed over specially purified XAD-2 resin. The resin
obtained from the manufacture (Rohm and Haas, Philadelphia, PA.) is extensively
washed with deionized water to remove fines and impurities. The cleaned resin
is then Soxlet extracted for 24 hours with each of four solvents, methanol,
acetone, hexane, and methylene chloride. Another set of four-hour extractions
with hexane, acetone, methanol, and a final exchange with deionized water is
carried out prior to storage. For precipitation sampling about 15 g of this
material is packed in a 30 cm glass column. For air sampling, the water is
drained off and about 40 g of resin is packed into the high-vol cartridges.
Packed columns are sealed with teflon caps, and cartridges are wrapped in
aluminum foil and sealed in air tight metal cans. Shipment is by surface mail
(2-5 days).
3.6 Field Blanks
Quantification standards are prepared and compared with EPA standards and
with standards from other laboratories. Blanks are run on all equipment,
reagents, and other materials to be used for the samplers and in the laboratory
procedures. Blanks and standards are run through the entire procedure, and the
system is shown to be in analytical control before samples are collected and
processed. Laboratory and field blanks are run at regular intervals for quality
assurance. Results from analysis of samples from the duplicate rain collectors
serve as a measure of overall reproducibility of the sampling and analysis
procedures.
As a check on the identification and quantification of different peaks in
the GC spectra, samples are run periodically on another GC-MS at a different
laboratory. Also, known amounts of standards are added to about 20% of the
samples which have been quantified. These samples are then re-run and the peak
identifications and quantifications checked. About 50% of the laboratory efforts
are devoted to quality assurance of the project.
13
-------
Field blanks are taken in conjunction with the third MIC sampler. XAD-2
resin columns are forwarded to the Green Bay site to determine if any
contamination can be detected from a resin column which is placed on a sampler
for a period of fourteen days but the sampler is inoperative because the
electrical power has been purposely disconnected. Another resin column is sent
to Green Bay and then mailed back to the analytical laboratory without exposing
the column to the ambient air. The object of these field blank comparisons is
to determine whether a difference in PCBs can be detected between the sample
which remains in the field for fourteen days and the sample which is returned
to the laboratory upon its receipt at the Green Bay site.
3.7 Sample Analysis
Samples collected at the master site are shipped to a laboratory for
analysis. EPA Method 608 serves as the procedural basis for organic sample
analysis, but modifications are made to the project to accommodate its more
demanding requirements.
Analysis of the high-volume filters for TSP and TOG follows standard EPA
methodology. The filters are weighed before and after their 24-hour exposure
in a temperature controlled room. The mass of matter on the filter is determined
from these weights and the monthly manometer calibrations.
4.0 AREAL MONITORING PROGRAM
4.1 Objectives
The primary purposes of the AREAL monitoring program was to obtain ambient
air samples similar to those obtained by GLNPO at the UWGB site so that a
comparison could be made between analytical data obtained by AREAL and GLNPO.
A secondary purpose was to provide additional air characterization data on air
toxic compounds, particulate matter, and trace metals to use in receptor models
and to guide future sampling and network design.
14
-------
4.1.1 Data Quality Objectives
The data quality objective for this comparison study was to obtain data
with measurement precision and accuracy within ± 10% or better. This is usually
as good as the analytical method could provide. For particle monitoring
(TSP.PM-10) data quality objectives for comparing similar samplers are reasonably
attainable at ± 10%. When analyses for elemental composition, dioxins, furans,
and PCB's at low concentration levels are included, an additional ± 15% error
v £«'
-—~7 would need to be addedA to uncertainty associated with the analytical method.
Therefore, the DQO will depend on the pollutant being compared and the
variability expected from the analytical technique used for that pollutant.
4.1.2 Data Quality Indicators
There are five indicators of data quality. These are completeness,
comparability, representativeness, accuracy, and precision. Several of these
indicators are not relevant to this study (i.e., comparability and
representativeness) on a macroscale. This is because the purpose of the study
is to compare data at one site during a short-term sampling program. However,
general siting protocols, such as distances between samplers and height above
ground, were followed to assure representativeness on a microscale level.
Completeness can be measured in terms of samples taken vs. the number
scheduled to be taken. Precision and accuracy were to be addressed through
quality control checks and independent quality assurance audits which were
conducted at the site by an EPA contractor. Overall precision of methods is
evaluated by collocation of the identical samplers. For this study, two
dichotomous samplers were operated to determine the precision of PM10
concentrations, and several duplicate VOC samples were collected during the study
period.
4.2 Samplers
The AREAL samplers deployed at the Green Bay site are shown in Table 4-1
along with the sampling frequency, sampling media and analyses. A diagram of the
15
-------
TABLE 4-1. AREAL SAMPLERS DEPLOYED AT GREEN BAY MONITORING SITE
SAMPLER
PM1Q Dichotomous
PM10 Dichot-
directional
TSP HWI
TSP Hi vol
PS-1
PS-1
Directional
VOC
Annular
Denuder
Wind Speed
NUMBER
2
1
1
1
2
1
1
1
1
SAMPLING
FREQUENCY
1/2
1/7
1/2
1/6
1/6
1/7
1/3
4
Samples
cont.
SAMPLE
DURATION
24 hrs.
7 days
24 hrs.
24 hrs.
24 hrs.
7 days
24 hrs.
24 hrs.
cont.
SAMPLING
MEDIA
37 nun Teflon
37 mm Teflon
2x6 Quartz
8x10 Quartz
8x10 Quartz
Quartz, PUF
Quartz, PUF
6 1. Canister
47 mm Quartz
47 mm Teflon
ANALYSIS
Mass Elemental
Composition XRF
Mass Elemental
Composition, XRF,
SEM Microscopy
Mass, Ce/CV, 1C
Mass , Elemental
PCB's; Aldrin/
dieldrin; Dioxin/
Furans
PCB's; Aldrin/
dieldrin
GC/MS
Gases and
Aerosols
SAMPLE
HANDLING
Weigh and handcarry
to RTF
Weigh and handcarry
to RTF
Weigh and handcarry
to RTF
Weigh and handcarry
to RTF
Airborne to
Contractors
Airborne to
Contractors
Airborne to RTF
weekly
Handcarry to RTF
_ .
Wind Direction
16
-------
sampler placement at the site is shown in Figure 4-1. All samplers were placed
on a plywood platform and secured with screws. The height above ground averaged
about three meters. (The terrain sloped slightly.) A total of eleven samplers
was deployed. Additionally, a meteorological system was erected to provide wind-
directional controlled operation to two of the samplers.
The two directional samplers -- PM-10 dichotomous and a PS-1(PUF plug
sampler) -- were operated so that comparisons of pesticides/PCB's and
dioxin/furans could be made with GLNPO's sampling/analysis system. The TSP
hi-vol was also set up to directly compare with GLNPO's hi-vol sampler. The
other samplers enabled AREAL to provide air characterizations not currently
performed by GLNPO such as elemental composition, PM-10 size fractions, acid
aerosols, carbon analysis, and PAH's.
4.3 Sampling Schedule and Operation
4.3.1 Schedule
In general, 24-hour samples were collected except for the wind direction
controlled dichotomous and PS-1 samplers which were operated for seven days but
when the wind was from the southwest (180 to 270 degrees). See Table 4-1.
Samples were started at 0730 ± 30 minutes and run for 24 hours. A copy of the
actual sampling schedule used is shown in Figure 4-2 .
4.3.2 Sampling Procedures
Filter samples were prepared, uniquely labeled at the lab and placed in
cassettes for transport to the field. The PS-1 samples (PUF and quartz filter)
were labeled and transported to the site in containers as provided by the
contractor. A site observation log sheet was filled out for each sampling day
and then entered into the field computer data system (See Figure 4-3).
17
-------
sampler placement at the site is shown in Figure 4-1. " All samplers were placed
on a plywood platform and secured with screws. The height above ground averaged
about three meters. (The terrain sloped slightly.) A total of eleven samplers
was deployed. Additionally, a meteorological system was erected to provide wind-
directional controlledxpperation to two of the samplers.
The two directional samplers -- PM-10 dichotomous and a PS-1(PUF plug
sampler) -- were operated so that comparisons of pesticides/PCB's and
dioxin/furans could be made with GLNPO's sampling/analysis system. The TSP
hi-vol was also set up to directly compare with GLJJPO/'sl hi-vol sampler. The
other samplers enabled AREAL to provide air characterizations not currently
performed by GLNPO such as elemental^composition, PM-10 size fractions, acid
aerosols, carbon analysis, and PAH's. \
\
4.3 Sampling Schedule and Operation
4.3.1 Schedule
In general, 24-hour samples were collected except for the wind direction
controlled dichotomous and PS-1 samplers which were operated for seven days but
when the wind was from the southwest (180 to 270 degrees). See Table 4-1.
Samples were started at 0730 ± 30 minutes and run for 24 hours. A copy of the
actual sampling schedule used is shown/in Figure 4-2
4.3.2 Sampling Procedures
Filter samples were prepared, uniquely labeled at the lab and placed in
cassettes for transport to the field. The PS-1 samples (PUF and quartz filter)
were labeled and transported to the site in containers as provided by the
contractor. A site observation log sheet was filled out for each sampling day
and then entered into the field computer data system (See Figure 4-3).
-------
FIGURE 4-2
AREAL SAMPLJNG SCHEDULE
SUN
DRIVE T
3
2DCT
1HWI
10
1CAN
(CAN COL
FLOW
CHECKS
17
2DCT
1HW1
24
MON
3 QREEN
4
1C AN
11
2 DOT
1HWI
18
FLOW
CHECKS
25lTBP
2 DOT
IlBlJft JB
HWI
1CAN
S»P31
TUES
3AY
5
2DCT
1HWI
1DCT-D
1PS1-0
12
1DCT-D
1PS1-0
19 12^
1HWI
1CAN
2FS1
ISfti!?
26
FLOW
CHECKS
WED
STTESI
FLOWCK
6
FLOW
CHECKS
13
1TSP
2DCT
1HW1
1CAN
ZPS1
20
2DCT
1HW1
27
2DCT
1HWI
1 CAN
THURS
IT UP
S&CAL.
7
1T9P
2DCT
1HW1
1CAN
2P61
14
2DCT
1HWI
(XTBAHUK)
21
2DCT
1HW1
28
SITE
TEAR
DOWN
FRI
1
1TSP
2DCT
1HWI
1CAN
2P81
8
AUDIT
15
2DCT
1HVVI
22
1CAN
ICANOOL
FLOW
CWCKS
29
DRIVE T
SAT
2
FLOW
OECKS
9
2DCT
1HW1
16
1CAN
FLOW
CHECKS
23
2DCT
1HWI
30
5 RTF
-------
FIGURE 4-3
AREAL SITE OBSERVATION LOG SHEET
1 OPERATOR
2 DATE ARRIVAL TIME
3 SAMPLERS OPERATING OK? YES NO
4 GENERAL METEOROLOGY FOR SAMPLING
4a Sky cond. 1.clear 2 part cloudy 3.cloudy
4. foggy
4b Wind Dir. 1. N 2. NE 3. E 4. SE
5. S 6. SW 7. W 8. NW
4c Wind speed 1.0-5 2.5-20 3. >20
4d Precip. 1. none 2. drizzle 3. rain
4. flurries 5. snow 6. sleet
7. other
5 Unusual events:
6 Visual traffic countCveh/min)
-------
a. Dichotomous Sampler/Hi-Vol Sampler
Operation was in accordance with the EPA-approved procedures described in
Appendices A.B.& E. The instruments were calibrated at AREAL, RIP and
transported to the field. Flow checks were performed at the beginning, at least
once per week, and at the end of the study and had to be within 10 % of the set
point value. The set points were determined based on the September average
conditions for Green Bay using thirty-year average weather records published by
NOAA. One blank filter was run for each sampling day for each sampling media
type. Each teflon filter was transported to the field and stored in labeled
petri dishes. The quartz filters were folded in half and stored in manila folders
and placed in labeled envelopes for storage. Following sample collection,
filter samples were transported back to the lab's environmental chamber,
conditioned to the same temperature and relative humidity when tared (20 degrees
C and 40 % relative humidity) for a minimum of 24 hours and then weighed. Before
equilibrium each filter was weighed to determine a rough weight and this
information was entered into the computer. The rough concentrations helped to
determine if the samplers were operating properly. Field information was entered
on Figure 4-4 and stored in the field computer daily.
b. VOC Canisters
One half of the canisters used for VOC sampling were cleaned at RTP and
brought to the field with the rest of the sampling equipment used in the study.
The other half were cleaned and shipped via air express about ten days after
sampling began. This helped to minimize the amount of time between cleaning and
use of the canisters. After sampling the canisters were shipped back to AREAL,
RTP on a weekly basis. Appendix D contains more detailed sampling procedures that
were followed during the study.
After completion of the VOC sampling, a routine QC check indicated that
the sampling probe had been obstructed during the entire sample schedule.
Consequently, the sampled air was pulled in through the manifold auxiliary vacuum
pump instead of through the sampling probe (See Figure 4-5). It is not possible
to say whether there was any significant impact on the ambient concentrations
21
-------
RGURE 4-4
AREAL FELD DATA SHEET
GREAT LAKES STUD Y—GREEN BAY. WISC.
DICHOTOMOUS SAMPLERS
OPERATOR.
DATE
START T1WE
STOP IMS
S/WLER
D
OOARSE
NO.
FiE
NO.
CRS ROT/MAN.
MT. FWAL
TOT. ROT/MAN
MT. FMAL
BLAP8ED
TWEfmin
PS-1 SAI^LERS
9AVPLER FLTER PUFF CART. NTJCATOR READNGS ELAPSED
ID NUMBER NUMBER START EM} AVQ TT4E(fT*l)
-------
FIGURE 4-4 COMT.
AREAL FIELD DATA SHEET
GREAT LAKES STUDY—GREEN BAY, WISC.
OPERATOR.
DATE
START T1S€
STOP TIME
VOC SAWLER
SAMPLER SAMPLE VACUUM PRESSURE SAMPLER FLOWfl/Mn) ELAPSED
D NJkCER STAFTT EMXPS) INT1AL BO AVO TOE
i
•
M-VOL SAMPLER
SAWDER SAWPUE epr POUT SAfcf^Hl FLOW
D Nl>fiER OCI rvrn MT1AL BO AVQ
ELAPSED
11C
-------
to AC
Inlet
MettJ Bellows II
Type Pump
I I I 1 r
ir — -1 /£-i
To AC
FIGURE SAMPLER CONFIGURATION FOR SUBATMOSPHERIC
PRESSURE OR PRESSURIZED CANISTER SAMPLING
-------
FIGURE 4-6
AREAL FIELD BLANK DATA SHEET—GREEN BAY
AMBIENT BLANKS
SAMPLE TYPE
DICHOT-- 37mm
BMI(8x10)
BMI(2x6)
PS- 1— QUARTZ
PS- 1— CART.
TSP(8x10)
OTHER
SAMPLE NUMBER
-
INSTRUCTIONS: Bring to site; open container; leave open for two
minutes; close container; leave at site; return to lab at end of
sampling period.
-------
as a result of this obstruction. It is possible that if the manifold fan
assembly were clean, or if the 41 compounds of interest were not present in
significant levels in the vacuum pump, the results would be totally unaffected
by this event. The effect of this obstruction will be discussed more completely
in Section 6.
c. PS-1 (PUF SAMPLER)
Operation was in accordance with EPA-approved methods described in Appendix
C. The instruments were calibrated at the Green Bay site. One point flow
checks were done at the beginning, at least once per week and at the end. The
samples were handled as specified in Appendix C to minimize dermal contact.
Samples were refrigerated on dry ice after sampling and shipped to the analysis
lab in the same refrigerated condition. All field data were recorded on Figure
4-4 and entered into the field computer that day.
d. Annular Denuder
For selected periods annular denuder samples were collected to obtain data
on the S02, HNOj, ammonia, sulfate and nitrate concentrations. An annular denuder
system (ADS) consists of a cyclone inlet to remove particles larger than 2.5
Jim, annular denuders chemically coated with Na2CO} to collect S02, HN03, and
citric acid to collect NH3; the denuders are followed by a filter pack to collect
the fine particle sulfate and nitrate aerosol. The aqueous extracts of the
denuders and filter pack are analyzed for NOj, SO^, N02 concentrations.
Subsequent data processing of annular denuder data provide the quantitative
measurements expressed Mg/m3 for HNO», HNO,, S02, S0j£, and NO, present in each
ADS sample.
e. Meteorological Sampling
The Climatronics Wind Direction Controller System was used to control power
to a dichotomous sampler and a PS-1 sampler as a function of the wind direction.
The system was used to turn on the samplers when the wind came from the range
of 180 to 270 degrees. The wind direction was sensed by a wind vane coupled to
26
-------
a precision low torque potentiometer. The wiper voltage of the potentiometer
is a measure of the wind direction and is used to control a relay which in turn
controls the samplers . A time delay of approximately 10 seconds was used to turn
the relay on and off in order to prevent chattering of the relay contacts .
The wind speed and direction were recorded continuously by a Campbell Data
Logger . ^ue_to_problems w,i.th-the-e i ther— the-logger ,__pr pj>wer_interruptionsj,_most
the meteorological data was lost. Data from the National Weather Service
office located at the airport was obtained to use in the analysis of the results .
The airport was located approximately 10 miles west southwest and should be
representative of the conditions that were encountered at the UWGB campus site.
4.3.3 Q.C. Operations
The standard operating procedures shown in the Appendices were followed
in operating the samplers with the exception of the special instructions
contained in this document. All samplers were flow checked when they were set
up and flows set for design operation i.e. 16.7 £/min. for dichots, 8 CFM for
the PS-1 samplers and 40 CFM for the HWI and Hi-Vol using the September average
conditions for Green Bay (14.9 C and 743.6 mm Hg.)
a. Dichotomous Samplers
The dichots were set at 16.7 £/min total and 1.67 £/min coarse flow using
the calibrated rotameter settings. Once set these conditions were not changed
unless there was more than a ten percent deviation. A flow check was performed
once per week The samplers were operated in accordance with Appendix A.
b. High Volume Virtual Impactor (HWI)
The HWI was calibrated at RTF and the flows were set at 40 CFM total and
2 CFM coarse. Once set, the flows were not changed unless the rotameter readings
varied more than 10 % from the original set points. A flow check was done at
least once per week. The samplers were operated and checked using procedures
in Appendix B.
27
-------
c. PS-KPUF SAMPLER)
The PS-1 samplers were calibrated on site in Green Bay. The samplers'
flows were set and checked using the standard top hat orifice. The samplers were
operated according to Appendix C which is a reprint of Method TO 13.
d. VOC/Canister samplers
The operation of this sampler is covered in Appendix D. Several co-located
samples were collected randomly during the study to allow for estimates of
precision.
e. TSP Hi-Vol
The TSP Hi-vol was operated using the EPA-procured quartz fiber filters
to minimize extrinsic sulfate formation in accordance with the procedures shown
in Appendix E. The sampler was calibrated at RTP and recalibrated in Green Bay
and the flow set for September average conditions.
f. Additional Aerochem
A,
In July-September, GLNPO had an additional Aerochem rainwater collector
placed at the Green Bay site. The primary objective of this augmentation was
to allow GLNPO test the effect of the addition of nitric acid to samples for the
purpose of preserving metals. Data have shown that leaving samples unacidified , ,
v^—^Jr b
-------
of analytic results. Unspiked samples from the second Aerochem were analyzed
only by the CRL, and results are reported in this document.
4.3.4 Sample Handling and Shipment
This section covers procedures used for handling samples before and after
sampling and their subsequent shipment to contractors and EPA for analysis.
Generally, samples were prepared the afternoon before the sampling day. Filters
were placed in cassettes, all field data sheets were filled out, labels were
placed on each sample or sample container to indicate date of sample, time of
day, and sampler number. Non-PVC gloves were worn during handling of any
samples. Dry ice for storage/shipment of the PCB/pesticides and dioxin samples
was obtained by no later than 1100 of the day the samples were removed i.e. the
samples were exposed to ambient conditions for no more than three hours. Care
was taken not to expose the samples to heat or sunlight after they were removed
from the samplers.
The PCB/dioxin samples were stored on dry ice and shipped to the
appropriate lab in insulated boxes with dry ice. Airborne next day delivery was
the method for shipment. The VOC canisters were shipped via Airborne to RTP
twice weekly. The remainder of the samples were stored in labeled envelopes
inside the weighing van's humidity/temperature controlled chamber. Upon return
to RTP, they were provided to the appropriate EPA personnel for analysis.
4.4 Calibration Procedures
This section describes the frequency and procedures that were employed in
the calibration of the sampling and analytical devices that were used in this
project.
4.4.1 Frequency
Five point flow rate calibrations of each PM sampler were performed before
the equipment was shipped to the field. Additional calibrations were scheduled
to be done any time a one point flow check was greater than 10% different from
29
-------
the set point or if a performance audit differed by more than 10%. This
situation occurred one time during the study and the sampler was recalibrated.
4.4.2 Procedures
a. TSP Hi-Vol sampler calibration
The five-point flow calibration of the Hi-Vol samplers was performed in
accordance with the procedures described in Section 8 of Appendix B with the
following exceptions: sampler calibration equations were determined for each
sampler using September average temperature and pressure rather than seasonal.
b. Dichotomous Sampler Calibration
A flow calibration analogous to the five point calibration of the Hi-Vol
sampler was performed before the equipment was shipped to the field in accordance
with procedures described in Section 5 of Appendix A with the following
exceptions: for the coarse and total rotameters Q ambient was determined using
the September average temperature and pressure.
c. PS-1 Hi-Vol Calibration
A five-point calibration curve was developed using a standard top hat
orifice following the procedures described in Section 8 of Appendix B.
4.4.3 Calibration Checks
Routine flow rate checks of each sampler were performed at the start of
the study and at least weekly during the study. An independent performance audit
was performed by an EPA contractor to check the flow rate of all samplers.
4.4.4 Balance Calibration
Both balances (Model H16 Sartorius and the Cahn microbalance) used in the
study were checked at the start of the project with NIST certified standard
30
-------
weights. Calibration checks were performed routinely during gravimetric analysis
as described below.
At the start and finish of every weighing session, a type-S standard weight
(5.0 g) was weighed on the Model 16 analytical balance to check the calibration.
The internal calibration weight was used as the calibration check standard for
the Cahn microbalance. In addition to these calibration checks, reference
filters were weighed during each weighing session to provide a measure of the
variability in mass determinations. These measurements were recorded on the
filter weight data sheets by the operator along with the temperature, pressure
and relative humidity.
Additionally, the zero of each balance was checked by the operator at the
start of each weighing session and after every fifth weighing (i.e. after every
fifth filter). The zero values must fall within ± 1.0 mg of true zero for the
Model H16 balance to be acceptable and within ± 4.0 ug of true zero for the Cahn
microbalance to be acceptable. Sample filters were reweighed if the zero drifts
exceeded these criteria. The zero values were recorded on the filter weight data
sheets with a check mark to indicate their acceptability.
4.4.5 Weigh Room
A portable weigh facility was transported to Green Bay and located outside
of the UWGB Laboratory Sciences Building. The weigh facility was temperature
and humidity controlled to maintain conditions at 20 degrees C and 40 percent
relative humidity. A recording hygrothermograph was used to record conditions
in the weigh facility on a continuous basis. Reference filters were weighed at
the beginning and end of every weighing session to provide a means of comparing
conditions from day to day.
#
4.5 Collection Media
The filter collection media were selected based on AREAL's past experience
with sampling of this type and the need for specific types of analyses to be
performed. Care was taken to assure low and uniform trace material background
31
-------
levels in the filters to minimize variability in sample analysis. The collection
media selected were as follows:
Hi-Vol: Whatman Quartz fiber meeting EPA procurement specs
Dichot: Gelman Teflo meeting EPA performance specs
PS-1: Quartz filter and PUF cartridges
VOC: 6-liter SUMMA polished canisters
4.6 Field Blanks
Blank samples were required for each sampling period with at least one
blank of each filter media or sample type used. The blanks were carried into
the field with the normal run filters/samples and left with the samplers ( in
their cassettes) if weatherproof space was available. At the completion of the
sampling, they were brought back to the weighing van and then equilibrated or
stored with the normal run filters/samples.
Information on blanks was placed on the form shown in Figure 4-6 and
entered into the field computer system daily. For the dichots, one 37 mm blank
was required for each sampling day. For the TSP Hi-Vol and the HWI, one 8 x
10 filter and one 2x6 filter were required for each sampling day. For the PS-1
samplers, one blank PUF and one quartz filter were required for each sampling
day and for each analysis lab.
4.7 Sample Analysis
The procedures for analysis of the filters for mass determinations is shown
in Appendices A and E. The analyses performed are summarized in Table 4-1 and
are described in more details in the following sections.
4.7.1 Ambient Particle Mass
The filter samples which were to have mass determinations made, were first
tared at 40% relative humidity and 20 degrees Celsius. After sampling, the
filters were equilibrated to the same conditions as when they were tared and then
32
-------
rewelghed. The results were then entered into the field computer and air quality
concentrations at EPA standard conditions (25 degrees C and 760mm Hg) were
reported.
4.7.2 ZRF Analysis of FM Filter Samples
The elemental composition of the dichotomous (37 mm) filter samples was
measured nondestructively using energy dispersive x-ray fluorescence (EDXRF)
procedures. The x-ray device used for these analyses was fabricated by Lawrence
Berkeley Laboratory and uses a pulsed x-ray tube to excite a secondary target
which in turn excites the sample with nearly mono-ergetic x-rays. To obtain high
sensitivity for a wide range of elements, each sample is excited by four
different secondary targets. For the K x-rays of elements with atomic numbers
in the range of a 13-20, 16-25, 25-38, and 38-56 the secondary targets are
titanium, cobalt, molybdenum and samarium respectively. The Mo target also
excites the L x-rays of lead (Pb) and other heavy elements. The EDXRF system's
calibration is checked daily using thin film standards, and the calibration is
validated daily by using standard reference materials prepared by NIST.
4.7.3 PCB/Pesticide Analysis
The samples were analyzed for aldrin, dieldrin, and PCBs by Southwest
Research Institute (SWRI). The filter and PUF plug were Soxhlet extracted
together with the exception of those filters that were received in broken petri
dishes. These filters were analyzed separately.
The PCBs in the field samples generally covered a broad spectrum of PCB
standard peaks. The data were reported as total PCBs in relation to 23 prominent
peaks from a mixture of aroclors 1016 and 1260. Quantitation was obtained by
summing areas of these peaks found in the sample and corresponding peak areas
in the 1016/1260 standard.
33
-------
4.7.4 Dioxin/Furan Analysis
Filter and FUF plug samples were sent to the EPA Laboratory at Stennls
Space Center, Mississippi for sample analysis preparation. Respective PUF and
filters were combined, spiked with 13C12 labeled dioxin and furan internal
standards and subjected to a 16-hour soxhlet extraction using benzene. The
extracts were sent to AREAL for analysis by High Resolution Gas Chromatography
(HRGC) - High Resolution Mass Spectrometry (HRGC-HRMS) for polychlorinated
dibenzo dioxins (PCDDs) and polychlorinated dibenzo furans (PCDFs). A number
of analyses were performed prior to sampling to ensure that the PUF plugs were
free of contamination. The analytical procedures and quality assurance plan used
in this study are presented in Appendix G.
4.7.5 Volatile Organic Compound (VOC) Analysis
The VOC canisters were analyzed (for the list of 41 compounds shown in
Appendix F) using an automated cryogenic sampling and gas chromatographic system.
The system consists of a Hewlett-Packard 5880A Level 4 gas chromatograph equipped
with a high resolution capillary column; an electron capture detector (ECD); a
flame ionization detector (FID); and a modified Nutech 320-01 cryogenic
preconcentrator unit. It takes about one hour for the analysis of each canister
which covers system initialization, sample collection, analysis and report
narration.
The VOCs are collected from each of the 6 \.\ canisters by pulling the
sample through a reduced temperature trap. Various compounds are concentrated
on the trap but the major air components nitrogen and oxygen pass through. The
VOCs are then thermally desorbed onto the high resolution column where they are
separated by gas chromatography in conjunction with oven temperature programming
and detected simultaneously by ECD and FID.
Before analysis, the system is calibrated using NIST traceable standards
In pressurized cylinders that contained mixtures of the target VOC compounds
at concentrations of about 10 ppm as working standards. The working standards
are diluted with humid zero air to about 10 ppb. Prior to being sent to the
34
-------
field for use, the canisters were cleaned and certified. One out of every four
canisters was certified as being clean by filling with humid air and analyzing
on the system. If one canister in a batch was found to be contaminated all of
the canisters in that batch were recleaned. Audit canisters provided by the
Quality Assurance Division of AREAL were analyzed to check the analytical system.
In addition, a small number of samples was analyzed using a mass selective
detector (MSD MAS SPEC) for analysis.
4.7.6 Sulfate/Nitrate Analysis By Ion Chromatography(1C)
Samples collected with the annular denuder and aliquots from the high
volume virtual impactors (HWI) were analyzed for nitrite, sulfite, nitrate, and
sulfate content by ion chromatography (1C) procedures. Aliquots cut from the
fine and coarse HWI filters were extracted in 0.003 M sodium carbonate solution
and analyzed by the Dionex 20001 ion exchange chromatograph. The fine particle
samples typically represented 65 m3 of sampled air, while the coarse particle
sample represented approximately 320 m3 of sampled air. These air volumes take
into account the percent of the total area of each filter that was actually used
in the analysis.
4.7.7 Elemental/Volatilizable Carbon
Elemental and volatilizable carbon present in the fine and coarse samples
collected on the HWI were determined by a combustion procedure. An aliquot of
the quartz filter is inserted into a combustion assembly and heated to 650
degrees C in a helium atmosphere. Organic material that volatilizes from the
filter is oxidized to carbon dioxide and then converted to methane. The methane
is then measured with a flame ionization detector (FID). The remaining
non-volatile elemental carbon is then measured by raising the temperature of
the sample to 800 degrees C and adding oxygen to the helium stream. This
oxidizes the elemental carbon to carbon dioxide and the carbon dioxide is
converted to methane and subsequently measured with the FID.
35
-------
4.7.8 Precipitation Metals Analysis
Six precipitation samples collected in an Aerochem bucket at Green Bay were
received and analyzed by AREAL for metals using ICP. These samples had been
split, with the other half going to the CRL for similar analysis.
A brief summary of the technique and results from these analyses and QA
activities follows. It should be noted that:
1. A range of pH's was observed in the samples when they were received
at AREAL. The meaning of this is not clear; but, it was not
expected.
2. A new ICP instrument with which AREAL has little experience was
employed. Special conditions are needed to determine Li, K and Na,
and the conditions could not be optimized. Results make the
difficulties apparent and skepticism is appropriate for those
elements.
Analysis was done at AREAL on a Jobin-Yvon Model 70 Plus ICPAES. Grouped
1989 standards made by Inorganic Ventures, Inc. were diluted 1:100 in two steps
in a matrix of 0.25% v/v HN03 and 0.20% v/v HCL to yield a concentration range
of 400 to 16,000 mg/1. Ultrex concentrated HCl and HN03 were used to adjust the
acid content and prepare the reagent blank. The pH values of the six rain
samples were measured prior to adjustments with concentrated Ultrex HCl to give
an acid matrix of 0.20% HCl which matched the standards. It was assumed that
the rain samples had been properly acid stabilized to 0.25% HN03 prior to
receipt, so no HN03 was added.
4.7.9 Annular Denuder
For selected periods during the study (as described previously), samples
were collected with the annular denuder assembly. The aqueous extracts from the
various denuder tubes and filters were analyzed for nitrate, sulfate, and nitrite
36
-------
concentrations. Subsequent processing of the data provided quantitative
measurements for nitric acid, nitrous acid, S02, sulfate, and nitrate present
in each sample.
4.8 Field Data Handling System
The data handling system consisted of a PC (Compaq 286) with a 40 mb hard
disk and one high density floppy drive. Due to the small number of samples
collected during the study, samples were not bar-coded for subsequent
identification. The software to store, process, and perform the necessary
calculations on the data was developed for EPA by NSI, Inc for the Integrated
Air Cancer Program (IACP) and uses a compiled version of dBase III+. The system
requires a definition of the sampling site, the samplers with their appropriate
calibration curves, calibration and flow check standards. Descriptive data are
entered on each sample taken along with tare and final weights of any filters
where mass concentrations are to be calculated. The system is capable of
generating a whole series of standard reports ranging from printouts of the raw
data inputs to summary of daily activities and chain of custody for shipping
samples.
5.0 QUALITY ASSURANCE AND QUALITY CONTROL REVIEW
5.1 Introduction
The QA and QC review of the GLNPO pilot site at UWGB consisted of a systems
audit, a collocated sampling study, and a performance audit. A systems audit
consists of a qualitative evaluation of all components of the
measurement/surveillance system. It evaluates both field and laboratory quality
control practix:e^j5ind_j:eviejws_^e_docum^ each facet of_the^
—__— — -
program./ However, it must be noted that the lack of a Quality Assurance Project
/-———-— /— —'
Plan (QAPP) from GLNOP did not permit a comprehensive systems audit_./ AREAL
conducted a systems audit of the UWGB data retrieval~§ystemto determine how the
monitoring and QA/QC data are incorporated into the data base. AREAL also
conducted a site and laboratory systems audit to assess the contractor's
capabilities and to determine if the siting criteria were suitable for this
37
-------
project. These audits enabled AREAL to; (1) determine the adequacy of the QAPP,
(2) determine if the project is adhering to the QAPP, (3) determine if the data
quality objectives are being met, and (4) determine the completeness of the data
base.
Accuracy, precision, and representativeness were evaluated through the
AREAL collocated sampling project at the University of Wisconsin in Green Bay.
A discussion of the colocated sampling site was presented in Section 4 of this
report. Accuracy was determined for specific segments of this project through
AREAL's performance audits on its own monitoring effort at the UWGB site and on
the Illinois State Water Survey's samplers. Sampler precision and
representativeness were determined through the results of comparisons between
similar samplers at the two UWGB sites.
Laboratory audits of the AREAL canister samplers verified the accuracy of
the VOC portion of the program. Duplicate spiked canisters provided both
accuracy and precision figures for laboratory analyses. Other laboratory
analyses were audited by providing the respective laboratories with spiked
samples of the necessary pollutants. AREAL has the capability to provide audit
samples for PUF, XAD, XRF, canister and other types of analyses, including
metals. Such samples will provide values for laboratory accuracy, precision,
and MDLs. . These values combined with field performance audits, usually in the
form of sampler flow audits, give an overall quantitative review of the accuracy
of the measurement system.
5.2 Great Lakes National Program Office
5.2.1 Quality Assurance Project Flan: Analysis and Findings
All GLNPO documents pertaining to the operations of the UWGB site were
requested by AREAL. These included; 1) the Quality Assurance Project Plan (QAPP)
for the GLAD Network, (2) the station operator's manual for the network, (3) the
Quality Assurance Plan for the Bionetics Corporation, (4) the QAPP for the
organic sample analyses for the GLAD Monitoring Stations, (5) the QAPP for the
operation of a master atmospheric deposition site at Green Bay, (6) numerous
38
-------
SOFs used by Bionetics in performing the analyses under EPA's contract, and (7)
a number of documents which refer to a "Green Bay Mass Balance Project".
The only crucial project document which was not provided by GLNPO was QAPP
for the overall air pollution monitoring project. Recent conversations with
GLNPO staff indicate such a document has not been written.
It is recommended that the QAPP for the Operation of a Master Site and the
one for Organic Sampling Analyses be combined in a more general QAPP for the
entire network. This QAPP would describe all aspects of sampling, analyses, data
retrieval, and how the monitoring data will achieve the GLNPO's "Mass Balance"
^objectives. It is essential that the key parameters for the "Mass Balance Study"
be identified and that the accuracy and precision of the measurements of these
parameters be specified. Responsibilities for each section within the overall
scope of the project need to be addressed. Such documentation is necessary for
external auditors and other members of the scientific community to judge the
merits of the overall program.
5.2.2 Components of a Quality Assurance Program Plan
^•^~ A quality assurance project plan describes; (1) the study's purpose, (2)
/the involved personnel's responsibilities, (3) the measurement and analytical
procedures that will be part of the field study, and (4) presents an outline of
the standard operating and analytical procedures that will be followed in the
-laboratory. The plan also identifies all quality assurance activities and
r guidelines that will be included in the study. Among the various aspects of QA
i
\ that must be addressed by the QAPP are explicit requirements to:
\
(1) identify and establish specific data quality goals to be achieved
during the project;
(2) describe the procedures that will be used to measure or assess the
quality of the environmental measurements obtained during the
project; and
39
-------
(3) describe the nature of the report or reports that will be prepared
to document the quality of the measurements.
Presently, guidelines define four measures of data quality, namely;
precision, accuracy, completeness, and method detection limit (MDL).
^'
Precision
Precision is a measure of agreement among individual measurements of the same
property, under prescribed similar conditions. Precision is determined by
measuring the agreement among a number of individual measurements (replicates)
of the same sample or concentration. For this program, precision of the total
measurement system will be measured through the collocation of similar
instruments. Other, more limited or specific measures of precision should be
included in the internal QC program, such as replicate analyses of control
samples.
Accuracy
Accuracy is a measure of the closeness of an individual measurement or the
average of a number of measurements to the true value. It is determined by
analyzing a reference material of known pollutant concentration or by reanalyzing
a sample to which a material of known concentration has been added.
Completeness
Completeness is a measure of the valid data obtained from a measurement
system, expressed as a percentage of the number of valid measurements that should
have been (i.e., were planned to be) collected. Completeness is not intended
to be a measure of representativeness -- i.e., how closely the measured results
reflect the actual concentration or distribution of the pollutant in the media
sampled. A project could produce 100% data completeness (all samples planned
were actually collected and found to be valid), but the results may not be
representative of the pollutant concentration actually present. For example,
the method might be biased, or the sampling times or locations might not provide
40
-------
a representative indication of the actual distribution of the pollutant in that
area.
Method Detection Limit
The method detection limit (MDL) is the minimum concentration that can be
detected with reasonable certainty. MDL is determined by measuring the
/
variability of replicate measurements at or near zero sample concentration.
Frequently this may be accomplished by measuring a zero concentration reference
material.
\
5.2.2 System Audits and Results
The systems audit conducted for the UVGB site included (1) a review of the
SOPs for the Bionetics laboratory practices, (2) a review of the Bionetics
laboratory facilities, (3) a review of the Atmospheric Deposition and
Precipitation Sampling Network Station Operator's Manual, and (4) a review of
two of the operating sites.
A systems audit of the Illinois State Water Survey (ISWS) Laboratory was
not conducted. However, this laboratory does participate in AREAL's semi-annual
acid rain analysis audit. The ISWS laboratory was audited by the acid rain
program in the spring and fall of 1989 (See Appendix H for the results). The
Geological Survey routinely audits the Water Survey Laboratory as part of their
responsibilities associated with the NADP/NTN.
The laboratory SOP's and QAPP were used as a basis for conducting a systems
audit of the Bionetics Laboratory (See Appendix 1). The auditor from AREAL
concluded that Bionetics was doing an excellent job of following both the QAPP
and the SOPs. Their record keeping was meeting all the requirements established
in the QAPP. QA/QC data was being kept in log books that had been kept since
the beginning of the project. Control charts were being maintained on the
analyses that were reviewed. The auditor found a few areas that could be
improved. Employee training, working hours, acknowledgments in EPA documents,
safety training, and filling the position of Q.C. coordinator are topics which
41
-------
need to be addressed by either the contractor or EPA. Further detail is
available in the audit report. The copy of the network station operator's manual
AREAL received to review was out of date (1985) .and much of the material is
presently not being used in the network. AREAL was asked to review the older
i
version and supply comments. The older version needed updating and revising to
include such things as standard headings for page number, revision number, and
section number.
AREAL conducted an abbreviated systems audit of the site at the University
of Wisconsin in Green Bay. The site was visited briefly and pictures were taken.
Pictures indicate the site to be located in an excellent area with good air flow
from all directions. Site logs were not obtained but have been requested for
review. A report will be written once these logs have been obtained and
reviewed. The close proximity of the dirt road at the University site may
contribute some amount of coarse particulate matter to the dichotomous and high
volume samplers at times of high wind velocity and heavy traffic.
No systems audit has been performed on the total data system operated by
the GLNPO, including the archival of data at the Great Lakes National Program
Office in Chicago. It is recommended that a data audit be performed, however,
a complete QAPP is a necessary precursor to such an audit.
\^
5.2.3 Performance Audit Results
On September 8, 1989, a flow audit was performed on 2 high volume samplers
operated for the GLAD network by the Illinois State Water Survey. Results showed
excellent agreement with AREAL's flow standards. External audits are necessary
to ensure the accuracy of any monitoring system. Future performance audits will
include all the meteorological equipment, precipitation samplers, aerosol sampler
flows, and other parameters that AREAL is capable of auditing.
5.3 Atmospheric Research and Exposure Assessment Laboratory
5.3.1 Quality Assurance Project Plan
42
-------
AREAL developed a monitoring plan for the project at the University of
Wisconsin site at Green Bay. Because of the limited scope and duration of
AREAL's monitoring of the UVGB site, a QAPP was not developed. Most of the items
that are required in a QAPF are addressed in the monitoring plan.
5.3.2 Systems Audit Results
Because this was a short duration study, a complete systems audit was not
done for the AREAL monitoring project at Green Bay. However, documentation in
the form of logbooks, Q.C. records, and the monitoring plan were reviewed.
Recommendations from this audit were incorporated into a revised work plan and
included in the AREAL monitoring study. Summary sample completeness data appear
in Table 5-1.
5.3.3 Performance Audit Results
On September 8, 1989, a performance audit was conducted on the AREAL
monitoring project at the UWGB site. The audit consisted of measuring the flows
of the 3 hi-volume samplers, two dichotomous samplers, two PUF samplers, and one
hi-volume PM-10 virtual impactor sampler. With the exception of the coarse flow
from one dichotomous sampler, all flows were in excellent agreement with AREAL's
flow standards.
Other performance audits include duplicate audit canisters for the volatile
organic compounds, and spiked PUF cartridges. Results from the VOC audit
canisters show 5 of the 17 compounds having a high positive bias (45%
-------
Sampler
Type
Dichotomous
HWI
TSP
PS-1
Dichot-Dir
PS-l-Dir
VOC
No.
2
1
1
2
1
1
1
Sample
Periods
16
16
5
5
3
3
10
of Samples
Samples Taken
64
32
5
10
3
3
10
64
32
5
10
3
3
10
Valid
Samples
48
32
5
10
3
3
0
Total"
100
100
100
100
100
100
100
Totalb
67
100
100
100
100
100
0*
TOTAL 9 58 127 127 101 100 80
"Ratio of samples taken to number scheduled (expressed as percent)
''Ratio of number of valid samples to number taken (expressed as percent)
* See Section 6.1.7 for discussion.
It was noticed that exact methodologies were not duplicated between the
GLNPO and AREAL monitoring sites. The AREAL project utilized the VOC canisters,
and PUF samplers for the organic compound collection, in place of the XAD
cartridge methodology employed by GLNPO. The only method of comparison between
the two programs was the high volume sampler employed by each. As it turned out
even the samples from these two hi vols did not have good agreement with each
other. Without comparable monitoring and analysis schemes, the representtiveness
and comparability of the data at the two sites is very questionable.
6.0 ANALYSIS OF MONITORING RESULTS
The following sections present the results from the analysis of samples
collected by AREAL and comparisons of AREAL data with that obtained by GLNPO.
Additional interpretive information is presented by way of microscopic analyses
and source apportionment modelling to provide some explanation for the measured
concentrations.
6.1 Colocated Measurements
44
-------
The following sections present the AREAL data and that from GLNPO which
were collected simultaneously at the Green Bay monitoring site.
6.1.1 TSP Hi vol
Table 6-1 shows the results of side by side TSP High volume measurements.
The data are plotted in Figure 6-1. Five data pairs were taken with poor overall
agreement (r2-0.46) if all five data points were plotted. If the one outlier is
removed (9/20/89), the agreement improves substantially (r2-0.93) as shown in
Figure 6-2.
TABLE 6-1. GREEN BAY PARTICULATE MATTER DATA COMPARISON AREAL/GLNPO
TSP TSP AREAL/ % | DCT/AREAL | HWI/AREAL
DATE AREAL GLNPO GLNPO DIFF jTOTAL FINE COARSEj TOTAL FINE COARSE
Ol-Sep-89
07-Sep-89
13-Sep-89
19-Sep-89
25-Sep-89
26.4
38.1
8.8
53.6
28.2
27.1
44.0
14.7
£33.0}
73773
0.97
0.87
0.60
1.63
0.76
-2.6
-15.4
-66.4
38.5
-32.1
|12.5
|23.3
| 6.3
|36.8
|11.9
5.5
17.6
3.6
28.3
3.9
7.0
5.7
2.7
8.5
8.0
10
24
5
38
18
.9
.3
.5
.9
.2
6.9
18.8
5.0
32.3
4.3
4.0
5.5
0.5
6.6
13.9
AVERAGE 31.Oy 31.2 0.96 -15.6 |18.2 11.8 6.4 | 19.6 13.5 6.1
The GLNPO data are consistently higher than the AREAL data with the
exception of the one outlier on 9/20/89. However, the data from other AREAL
samplers for that date are consistent with each other. For example, the PM10
dichotomous sampler which collects those particles below 10 /im as compared to
the 25 [Mm for the TSP hi vol has a lower concentration than the TSP hi vol and
the HWI (which also did not have a PM10 inlet.
The GLNPO hi vol, on the other hand, had a lower concentration than the
dichotomous sampler. This is physically impossible due to the different inlet
geometries discussed above. This indicates that the GLNPO data are probably in
45
-------
so
FIGURE 6-1 GREEN BAY TSP DATA COMPARISON
AREALVSGLNPO
80 -
20 -
10 -
I I
16
SLOP&0.47
C-16.9
R2-0.46
26 36
AFEALT3P(u0/m3)
i
46
66
-------
so
FIGURE 6-2 GREEN BAY TSP DATA COMPARISON
MEALVSGLMQ
40 -
80 -
10 -
OUTUER FEMOVED
SLOPE-1.01
C-5.1
R2-0.93
I i r i
8 12 16
I I I I I I I
20 24 28 32
86 40
AREA. TSP (u0/hi3)
-------
error. The error may have been caused by improper flow measurements, weighing
error, or loss of sample in transport.
In general, there is the wind speed and direction dependence of the TSP
hi vol samplers due to their nonsymmetric inlets. This dependence contributes
to the poor-to-fair comparisons typically found with TSP measurements. However,
PM10 samplers, such as the dichotomous sampler and the SSI or HWI, do not
exhibit these problems with wind dependence because they have omnidirectional
inlets and well defined d50 cutpoints for collecting ambient aerosol.
The levels of PM10 found at the Green Bay site during September were well
below EPA's health-based national ambient air quality standard (NAAQS) of 150
/ig/m3 for a 24-hour period and averaged about 50% below the 50 /Jg/m3 annual
average standard. Values measured at the UWGE site are typical for cities of
this size. The somewhat nonurban character of the site probably contributed to
the moderate concentrations measured there.
6.1.2 Total Carbon Measurements
Table 6-2 shows the total carbon measurements from the TSP hi vol samples
of GLNPO compared to the fine and coarse fraction carbon measurements from the
AREAL HWI sampler. Similar to the TSP, the agreement is generally poor.
Coefficient of correlation is around 0.2 (See Figure 6-3). Although the average
percent difference between AREAL and GLNPO is only 24.1%, the range of difference
values is from +53% to -11%.
6.1.3 Directional Samplers
Table 6-3 displays the results derived from the wind direction controlled
PM samplers. The AREAL PM sampler was a conventional Sierra-Andersen PM10
dichotomous sampler. In addition, a PUF sampler was operated to collect samples
for PCB and pesticide analysis without any mass measurements being made. The
'mass concentrations from AREAL's dichotomous sampler and^LLLLLMPoJ^ TSP hi-
vol did not compare well. This is most likely due to the large differences in
the sampling times. A review of the average hourly meteorological data show that
the time that the wind was from the SW quadrant compares more favorably to
48
-------
AREAL's measurement system. For some reason, the last sample taken by GLNPO was
-for a two week period and so ti/can not be directly compared to AREAL's one week
samples. / f
6.1.4 Polychlorinated Dibenzo-p-dioxins and Dibenzofurans
A comprehensive analytical report that describes many aspects of the
sampling and analysis (PS-1 sampler efficiency, method efficiency, field blanks,
method blank, laboratory method spike) needed for QA/QC and evaluation purposes
is shown in Appendix J. The data satisfies all QA/QC requirements. Analytical
data for the test samples were extracted from that report and are shown in Table
6-4. The concentrations of PCDDs and PCDFs were very low and similar to
concentrations (background levels) found in ambient air in other studies by AREAL
49
-------
TABLE 6-2. TOTAL ORGANIC CARBON MEASUREMENTS AREAL VS. GLNPO (ug/m3)
IGLNPO | AREAL||ORGANIC CARBON(ELEMENTAL CARBON|TOTAL CARBON |
DATE | TOC | TOG ||COARSE FINE (COARSE FINE (COARSE FINE (TOTAL
9/1/89
9/7/89
9/13/89
9/19/89
9/25/89
8.3
5.8
2.3
4.7
4.2
1
3.9
4.1
2.3
5.2
3.7
1
1
1.3 2.6
0.9 3.2
1.1 1.2
1.6 3.6
2.6 1.1
1
0.1 0.3
0.0 0.3
0.1 0.3
0.2 0.5
0.0 0.2
r
1.4 2.9
0.9 3.5
1.2 1.5
1.8 4.1
2.6 1.3
4.3
4.4
2.7
5.9
3.9
AVERAGE
5.1 | 3.8|| 1.5
2.3
0.1
0.3
1.6
2.7
4.2
TABLE 6-3. DIRECTIONAL PM DATA COMPARISON (SW QUADRANT)
DATE
TIME AVG HOUR MET
FROM
05-Sep-89
12-Sep-89
19-Sep-89
19-Sep-89
AVERAGE
TO
12-Sep-89
19-Sep-89
26-Sep-89
03-Oct-89
TSP/GLNPO
(ug/m3)
45.6
89.3
83.1
72.7
TIME
(hrs)
24.3
5.9
28.1
19.4
IDCT/AREAL
1 (ug/m3)
1
1
1
1
1
23.2
16.3
29.5
23.0
TIME
(hrs)
38
110
111
86.3
(DATA WAS
1 (hrs)
63
83
66
70.7
FROM SW
37.5
49.4
39.3
42.1
50
-------
RGURE 6-3
GREEN BAY TOTAL CARBON COMPARISON
8
8-
7 -
6 -
e-
4 -
8 -
SLOPE-0.99
R2-0.22
C-1.3
"
I I I T
2.2 2.6 3
i i r
3.4
3.3
4.2
4.6
-------
TABLE 6-4. DIOXIN-FURAN DATA FOR GREEN BAY
SAMPLE |TPCDD|TETRA|PENTA|HEXA|HEPTA| |TPCDFs|TETRA|PENTA|HEXA|HEPTA|
DATE |(pg/m|CDD |CDD |CDD |CDD |OCDD|(pg/m3|CDF |CDF |CDF |CDF |OCDF
V-*»»_-*W_-_H-I
9/7/89
•B-c-w-wowi
1.7
9/13/89(0.89
1
9/19/89J 1.4
— ^T"^-— —
0.3
0.09
0.2
9/25/89(0.78 |0.08
0.4
0.1
M«_*_
0.4
0.2
0.2 |0.3
1
-- |0.2
1
1
0.3 |0.3
1
0.2 |0.3
1
0.3 |0.4
1
0.2 |0.3
•H_-_v»w_-a-i
2.65
1.08
2.23
0.67
_•*•_•_•_•
0.8
0.4
0.7
0.3
•__-_-_-•••
1.2
0.4
^n^_-_m
0.5
1
1
0.1 (0.05
1
0.2 |0.06 (0.02
1 1
0.7 |0.4 |0.23 | 0.2
1
0.2 |0.1 (0.05 (0.02
NOTE: ALL CONCENTRATIONS EXPRESSED IN PICOGRAMS PER CUBIC METER(pg/m3)
and other laboratories. The Isomeric pattern of PCDDS and PCDFs detected in
these samples indicate they were formed in combustion/incineration processes and
released to the atmosphere predominately through stack gas emissions.
Atmospheric transport is the major mode for widespread dispersal of these
compounds throughout the environment. Several references are provided regarding
PCDDs and PCDFs in ambient air, transport, deposition, and fate: Chemosphere
, Vol. 17, No. 1, pp. 3-20, 1988; Environmental Science and Technology, Vol. 23,
No. 11, pp. 1389-1395, and pp. 1396-1409, 1989; Chemosphere, Vol. 19, No. 1-6,
pp 541-546, 1989; and Proceedings of the 1988 EPA/APCA International Symposium,
Measurement of Toxic and Related Pollutants.
6.1.5 Polychlorinated Biphenyls and Organochlorine Pesticides
Both wind direction-controlled sequential sampling and 24-hour
integrated sampling with PS-1 samplers were performed to determine PCBs and
aldrin/dieldrin in ambient air. Eight collected samples, eight field blanks and
seven blind spiked samples were analyzed by dual capillary gas chromatography
(30 m DB-5 and DB-608 columns) with electron capture detection according to the
EPA Contract Laboratory Program (CLP) protocols. Quartz-fiber particle filters
and PUF plugs were extracted and analyzed separately. All extracts were analyzed
using reference standards of aldrin, dieldrin, and Aroclors 1016, 1221, 1232,
52
-------
1242, 1248, 1254, and 1260 supplied by the EPA Pesticides and Industrial
Chemicals Repository. Recoveries from spiked samples ranged from 75 to 130%.
When PCBs were evident in the initial analysis, an acid-treated five-fold
diluted sample was analyzed to remove possible interferences. The PCBs
identified in the field samples generally covered a broad range of PCB standard
peaks. Accordingly, they were quantified as total PCBs in relation to 23
prominent standard peaks from a standard mixture of Aroclors 1016 and 1260.
Quantisation was obtained by summing the areas of these peaks found in the sample
and the corresponding peak areas in the Aroclor 1016/1260 standard. No PCBs were
detected on any of the filter extracts.
Data from the pesticides/PCBs analyses are presented in Table 6.5. No
aldrin or dieldrin was found at concentrations exceeding the quantitation limits
(0.15 and 0.25 ng/m3, respectively, for 24-hour samples collecting 325 m3 of
air). Quantitation limits were generally set at five times the minimum level
of detection. PCBs were found in the PUF extracts from seven of the eight field
samples at concentrations ranging from 3 to 6.5 ng/m . With exception of the
one negative sample, for which there was no overlapping wind-directed sampling,
the results of 24-hour integrated and 36- to 100-hour wind-directed sampling were
in agreement.
53
-------
TABLE 6-5. PCB-PESTICIDE DATA FOR GREEN BAY
SAMPLE
DATES
SAMPLE
TYPE
ALDRIN DIELDRIN PCB's
(ng/m3) (ng/m3) (ng/m3)
|Wind Directed
{Sequential:
05-Sep-89 TO
12-Sep-89
12-Sep-89 TO
19-Sep-89
19-Sep-89 TO
26-Sep-89
PS-l(Filter+PUF)
PS-l(Filter+PUF)
PS-l(Filter-t-PUF)
ND ND
ND ND
ND ND
6.18
4.95
3.15
24 -hr integrated:
Ol-Sep-89
07-Sep-89
13-Sep-89
19-Sep-89
25-Sep-89
PS-l(Filter+PUF)
PS-l(Filter+PUF)
PS-l(Filter+FUF)
PS-l(Filter+PUF)
PS-l(Filter+PUF)
ND ND
ND ND
ND ND
ND ND
ND ND
ND
4.95
4.11
6.48
3.27
ND - Not detected above quantitation limit (0.15 ng/m
aldrin, 0.25 ng/m3 for dieldrin and 0.3 to 1.5 ng/m3 f
for 24-hour samples).
3 for
for PCBs
The absence of detectable air levels of aldrin and dieldrin is not
surprising. These pesticides have not been used in agriculture in the United
States in many years. Dieldrin levels in the atmosphere in the Great Lakes
region have generally been found to be below 0.1 ng/m3 (Thomas J. Murphy, DePaul
University, Chicago, IL, personal communication, March 1990; Terry Bidleman,
University of South Carolina, Columbia, SC, personal communication, March 1990).
It should be noted that aldrin is oxidized to dieldrin during the sampling
process (R. G. Lewis, A. R. Brown, and M. D. Jackson, Analytical Chemistry 49,
1668-1672 [1977]).
Airborne PCB levels found in this study appear to be higher than expected.
The most recent published data (J. L. Baker and S. J. Eisenreich, Environmental
Science & Technology, 24, 342-352 [1990]) show a range of 0.6 to 1.8 ng/m3 (avg
54
-------
1 ng/m3) for Northern Wisconsin in the summer of 1984. However, 1983 data from
an island in Lake Superior showed a range of 1.5 to 5.2 ng/m3 (avg. 3.2 ng/m3),
which is comparable to AREAL findings. Recent personal communications with
Steven Eisenreich of the University of Minnesota, Minneapolis, and Thomas Murphy
(March 1990) suggest that FCB air concentrations rarely exceed 1 ng/m3 in the
Green Bay area. Results of collocated sampling by Dr. Eisenreich's laboratory
during the AREAL monitoring period will be available for comparison by mid-
April 1990. Their analytical results, as Dr. Murphy's, will be quantified
against an EPA Grosse lie Laboratory reference standards of Aroclors 1232, 1248,
and 1262, a mixture similar to the Aroclors 1016/1260 standard used by the AREAL
contract laboratory (Southwest Research Institute, San Antonio, Texas).
6.1.6 Precipitation Metals
Table 6.6 contains the mean and standard deviation (-.8865*Range) of two
ICP analyses of each of the six samples received by AREAL. Absence of an entry
or of an element which appears in Table 6-7 (Detection Limits) means that results
were smaller than the Table 6-7 values.
TABLE 6-6. AREAL RESULTS FROM GREEN BAY PRECIPITATION SAMPLES*
(micrograms/liters)
Al
Ba
Ca
Cd
Cu
Fe
Mg
Mn
Ni
Pb
Sr
Zn
*FM
Avg
16.1
1.4
136.9
2.1
4.1
19.5
37.7
1.8
5.6
103.9
890822
St Dev
4
0.3
0.1
0.1
0.6
2
0.6
0.1
2
8.2
FM
Avg
9.0
117.7
1.0
1.6
9.2
23.7
1.3
6.0
24.0
890815
St Dev
2
0.00
0.3
0.02
1.0
0.2
0.2
4
1.4
*GM
Avg
53.3
3.6
1194
0.8
1.7
69.6
401
9.4
1.3
46.4
890815
St Dev
7
1
11
0.2
1
1
2
0.01
0.05
1.3
FM 890829
Avg St
9.1
62.0
0.7
1.1
9.9
17.9
1.2
17.0
1140
Dev
2
0.2
0.4
0.1
0.04
0.3
0.00
0.8
5
*GM 890718
Avg St
43.9
2.0
32.4
0.7
1.2
37.3
102.1
3.9
117.8
Dev
2
1
0.1
0.1
0.1
0.7
0.1
0.01
4.4
GM 890822
Avg St
10.2
120.1
8.2
44.6
1.2
53.1
Dev
0.9
2
0.7
0.8
0.2
91
* Sample contained substantial foreign material.
# If no entry, one or both analyses were below detection limit.
If element missing, but appears in Table 2, no reportable results were found.
55
-------
TABLE 6-7. ESTIMATED DETECTION LIMITS FOR PRECIPITATION SAMPLES
(micrograms/liters)
Mn 0.3
Al
As
Ba
Be
Ca
Cd
7
9
3
1
0.3
077"}
0.4
Co
Cr
Cu
Fe
i— fcr.^ -r^-— ^— ^_— — — -
Li
Mg
2
2
0.9
0.4
560:3
28
0.6
(Na
Ni 4
Pb 10
Sr 0.8
V 1.5
Zn 0.5
Several reference materials whose compositions were known to the analyst
were analyzed along with the samples. They cover a wide range of concentrations
and the analytes of interest. Recovery data are shown in Table 6-8. The
unexceptional conclusion is that, except for the alkali metals, recoveries are
reasonable until detection limits are approached.
Table 6-9 contains recoveries from two samples whose composition was
unknown to the analyst. The interpretation of these data is similar to those
from the solutions of known composition in Table 6-8.
Earlier, other portions of these samples were analyzed by the Central
Regional Laboratory. As reported by CRL, elements were quantified using an ICP
(Jarrell Ash Model 1160 ICPAES using PC based software). Data from these CRL
analyses, but only for those elements which appear as finite values in the AREAL
analyses, are presented in Table 6-10.
A comparison of the AREAL and CRL analyses (Tables 6.6 and 6.10) show
i ^--""^
generally ^goocr agreement for zinc. The AREAL values for zinc tend to be much
higher than those from CRL. There could be zinc contamination in either the
polyethylene bottles used to ship the samples to AREAL or somewhere in the AREAL
analysis system. However, there was no evidence of problems in the AREAL
analysis of standard solutions (Table 6-8).
A comparison of the CRL results from spiked and unspiked samples (Tables
6-10 and 6-11) shows most notably an almost complete erasure of aluminum in the
unspiked samples. Variations for other metals seem to be random, and the heavier
metals do not show systematic losses across all samples as might have been
56
-------
TABLE 6-8. RECOVERIES FROM STANDARD SOLUTIONS OF KNOWN COMPOSITION
(percent)SO*T
EMSL/Cinl SRM2 SRM2 H1QC3 LoQC4
1/100 1/1000
Al 103.5 100.4 99.6 103.6 112.9
As 104.0 101.2 100.9 106.9 200.6
B 95.5 98.6 124.4
Ba 100.7 101.9 128.4 97.9 127.8
Be 95.3 100.8 98.8 101.6 99.4
Ca 103.7 108.9 128.5 102.7 110.9
Cd 99.1 101.4 100.3 101.3 94.7
Co 96.8 101.1 103.4 97.0 110.9
Cr 99.9 99.6 102.4 100.0 98.9
Cu 104.8 106.0 104.3 100.3 100.7
Fe 97.0 99.1 97.9 96.2 116.9
K 97.5 108.5 165.2 93.4 BD
Li 97.9 504.8
Mg 102.6 101.8 101.7 100.0 104.4
Mn 101.4 100.0 100.5 98.2 108.0
Na 118.0 110.9 132.4 101.1 300.1
Ni 103.0 101.5 105.3 100.0 123.6
Pb 100.8 103.5 114.7 103.3 272.5
Sr 99.4 99.2 98.2 98.5
Ti 98.9 100.0 106.0
V 99.9 102.6 98.8
Zn 108.5 100.3 105.8 103.5 182.4
*Single column lists used for two samples if elements differ.
lEMSL/Cin ICP Quality Control Samples ICAP-19 and ICAP-7, Lot WP988
Concentration range: ICAP-19 1000 ing/liter
ICAP-7 890-10,000 mg/liter
2NIST Standard Reference Materials 3171 and 3172
Concentration range: 1/100 100 - 5000 mg/liter
1/1000 10 - 500 mg/liter
3H1QC ASB preparation from non-calibration standard stocks
Concentration range: Ca,Mg,K,Na 5000 mg/liter
All others 500 mg/liter
4LoQC ASB preparation from non-calibration standard stocks
Concentration range: 5 -200 mg/liter
57
-------
TABLE 6-9. AVERAGE RECOVERIES FROM ONE ANALYSIS EACH OF TWO
BLIND SAMPLES FROM EACH SERIES
(Percent)
Series 4 Series 5
Cd 102 105
Cu 104 98
Fe 101 87
Mn 112 91
Ni 128 120
Pb 127 126
Zn 110 109
Series 1 Series 2 Series 3
Ca 106 100 114
K 1084 156 1074
Mg 108 98 99
Na 165 112 137
* Based on tentative accepted values
58
-------
TABLE 6-10. REGIONAL LABORATORY RESULTS FROM GREEN BAY
PRECIPITATION SAMPLES**
(micrograms/liter)
FM08A22
12.75
0.3
107.8
0.07
3.74
41.90
0.7
0.24
5.36
FM08A15
7.0
0.06
103.6
1.75
11.22
21.74
0.96
0.09
10.77
GM08A15
57.15
2.36
1100.0
0.91
59.34
382.0
8.62
1.35
7.72
FM08A29
14.63
0.54
52.67
0.91
7.93
15.6
0.65
0.12
2.72
GM07A18
17.38
1.03
166.4
1.89
13.01
61.04
1.91
0.57
8.60
GM08A22
12.75
0.3
107.8
0.07
3.74
41.90
0.7
0.24
5.36
Al
Ba
Ca
Cu
Fe
Mg
Mn
Sr
Zn
^Elements contained in Table 6-6 only.
Al
Ba
Ca
Cu
Fe
Mg
Mn
Sr
Zn
TABLE 6-11. REGIONAL LABORATORY RESULTS FROM GREEN
BAY (UNSPIKED) PRECIPITATION SAMPLES
(micrograms/liter)
FB08S22
FB08S15
QM08S15
FB08S29
QM07S18
(JO8S22
2.51
0.54
97.43
2.08
3.84
26.37
1.22
0.22
19.63
1.57
0.15
96.60
4.41
2.41
21.86
0.77
0.15
32.22
1.26
1.17
925.
0.04
2.07
301.8
2.35
0.99
22.88
2.82
0.76
244.0
3.34
10.22
84.73
2.02
0.38
8.56
-1.6
0.84
329.3
0.39
2.41
97.59
3.4
0.47
6.29
-1.33
0.20
67.32
-0.29
1.53
19.92
0.49
0.15
12.10
59
-------
Three types of sampling equipment were used to obtain the composition of
gases and fine and coarse particles at the UW-GB monitoring site. Table 4-1
contains the types of samplers used and chemical species analyzed. The low
volume dichotomous sampler used 2 /im pore size Teflon filters to collect the fine
and coarse particles. From these filters the mass, elemental and microscopy data
were obtained. The low-volume dichotomous sampler was equipped with a well-
defined PM10 inlet, while the high volume dichotomous sampler was equipped with
the conventional gabled roof inlet. The size range of the coarse particles (>2.5
/lm) that are collected by this sampler is dependent on wind speed and direction.
6.2.3 Mass, XRF and Ion Chromatography Correlations
Table 6-12 contains the average composition of the fine and coarse
particles collected between September 1, and September 27, 1989 at the UW Green
Bay site. Appendix K contains the XRF data for all the sampling periods. Figure
6-4 shows the fine and coarse mass concentrations plotted as a function of time
for that period.
Tables 6-13 and 6-14 contain the carbon and inorganic ionic species data
for fine and coarse particles collected on quartz filters in the high volume
dichotomous samplers.
TABLE 6-12. AVERAGE COMPOSITION OF AEROSOL AT GREEN BAY, WI,
from September 1-27, 1989
Mass8
Cv
Ce
S
Al
Si
S
Units
Mg/m3
2/S
Mg/m3
Mg/m3
ng/m3
ng/m3
ng/m3
Fine
17.1
2.22
0.27
0.22
0.57
6.13
70 (26)b
50 (12)
1754 (154)
Coarse
6.18
1.29
0.05
0.32
0.12
0.22
480 (152)
617 (158)
71 (38)
61
-------
hypothesized. If analysis for metals in rainwater is to continue, some attention
must be focused on sample handling and quality control.
Table 6-9 contains recoveries from two samples whose composition was
unknown to the analyst. The interpretation of these data is similar to those
from the solutions of known composition in Table 6-8.
Earlier, other portions of these samples were analyzed by the Central
Regional Laboratory. As reported by CRL, elements were quantified using an 1CP
(Jarrell Ash Model 1160 ICPAES using PC-based software). Data from these CRL
analyses are presented in Table 6-10 but only for those elements which appear
as finite values in the AREAL analyses,
6.1.7 VOC Measurement
The VOC data were not reported due to a sampling problem discussed earlier.
Although most of the values seemed reasonable and no outliers were present in
the data set, the data could not be independently validated.
The validation could not be done because of the potential contamination
from the pump. Therefore, we are unable to state that the data are either good
or biased by contamination. As a result no VOC data are being reported.
6.2 Modeling and Measurement Results
6.2.1 Introduction
This section presents a discussion of the results of the sample analyses.
The sampling systems and chemical analysis procedures used provide information
on the chemical and physical properties of the air toxic gases and aerosols that
impact the Green Bay area. The primary sampler used to collect these samples
(the high-volume virtual impactor - HWI) is based on virtual impaction
principles which separates the coarse particles (>2.5 Jim) from the fine fraction
(<2.5 /im). The separation and subsequent analysis of these two fractions
provided a complete assessment of physical and chemical properties, sources, and
transport mechanisms of air toxic pollutants to the Green Bay.
6.2.2 Gas and Aerosol Samplers
60
-------
Cl ng/m3 14 (3) 18 (3)
K ng/m3 45 (4) 83 (10)
Ca ng/m3 33 (3) 376 (36)
Tl ng/m3 1.1 (2.4) 12 (3)
V ng/m3 -0.2 (1.6) 0.5 (1.5)
Mn ng/m3 3.9 (0.6) 3.6 (0.6)
Fe ng/m3 48 (5) 122 (13)
Ni ng/m3 0.3 (0.3) 0.3 (0.3)
Cu ng/m3 3.0 (0.5) 1.9 (0.4)
Se ng/m3 1.1 (0.3) 0.1 (0.2)
Br ng/m3 2.1 (0.4) 0.2 (0.2)
Pb ng/m3 10.7 (1.4) 1.2 (0.6)
Zn ng/m3 19 (2) 5 (1)
"Mass, Carbon (Cr and Ce) and ionic data refer to samples collected by Hi vol
sampler.
bXRF data for dichotomous sampler. Average uncertainty is given in parentheses.
62
-------
Figure 6-4. Time series of fine and coarse particle mass for the Green
Bay site. The fine mass exceeds the coarse on most days but the last sampling
period.
Greenbay Study
Fine and Coarse Particles
50
s
.5
1
£
20
10
ipjl
rtiffimf
9/1 9/5 9/9 9/13 9/15 9/19 9/21 9/25
9/3 9/7 9/11 9/14 9/17 9/20 9/23 9/27
Date
-------
'TABLE 6-13. MASS CONCENTRATION, IONIC SPECIES AND CARBON FOR FINE PARTICLES
COLLECTED IN GREEN BAY, SEPTEMBER 1-27, 1989
1989
9/1
9/3
9/5
9/7
9/9
9/11
9/13
9/14
9/15
9/17
9/19
9/20
9/21
9/23
9/25
9/27
Mass
Cone.
Mg/m3
6.85
10.21
26.09
18.76
4.92
7.97
5.02
8.99
17.03
26.40
32.20
44.52
48.39
1.96
4.31
9.80
NO,
Mg/m3
0.20
0.12
0.10
0.08
0.13
0.09
0.29
0.24
0.27
0.08
0.10
0.10
0.10
0.32
0.38
1.02
SO,
Mg/m3
0.48
0.69
0.59
0.86
0.49
0.45
0.50
0.41
0.62
0.44
0.46
0.48
0.51
0.51
1.02
0.56
S04 Total S04 Org. C Elem.C
Mg/ms MgA>3 Mg/m3 Mg/m3
1.28
2.25
9.10
8.23
1.19
3.05
1.35
2.55
3.41
9.25
10.00
17.88
24.50
0.75
1.58
1.67
1.85
3.08
9.80
9.25
1.78
3.59
1.95
3.04
4.18
9.78
10.55
18.45
25.11
1.36
2.81
2.34
2.55
1.76
2.98
3.23
1.33
1.41
1.18
1.21
2.49
2.60
2.83
3.59
4.50
0.81
1.08
2.03
0.30
0.10
0.28
0.34
0.06
0.19
0.26
0.19
0.37
0.24
0.42
0.49
0.51
0.03
0.16
0.33
Average 17.09 0.23 0.57 6.13 6.81 2.22 0.27
TABLE 6-14. MASS CONCENTRATION IONIC SPECIES AND CARBON FOR COARSE PARTICLES
COLLECTED IN GREEN BAY, SEPTEMBER 1-27, 1989
Date
1989
9/1
9/3
9/5
9/7
9/9
9/11
9/13
9/14
9/15
9/17
9/19
9/20
9/21
9/23
9/25
9/27
Mass
Cone.
4.00
3.59
13.06
5.43
1.56
5.63
0.44
2.47
5.32
5.53
6.55
9.09
6.24
3.08
13.88
13.06
N03
Mg/m3
0.05
0.14
0.54
0.16
0.06
0.33
0.06
0.18
0.39
0.81
0.50
0.78
0.43
0.07
0.27
0.29
SO,
0.11
0.16
0.15
0.21
0.08
0.21
0.07
0.17
0.16
0.13
0.14
0.19
0.09
0.03
0.02
0.03
SO,
0.19
0.18
1.24
0.26
0.12
0.12
0.06
0.11
0.09
0.26
0.14
0.15
0.44
0.03
0.10
0.07
Total S04
Mg/m3
0.35
0.39
1.42
0.52
0.21
0.37
0.14
0.32
0.28
0.41
0.31
0.38
0.55
0.07
0.12
0.11
Org.C
1.34
0.79
1.64
0.90
0.95
1.26
1.13
0.60
1.29
0.77
1.59
1.62
1.18
0.79
2.58
2.18
Elem.C
0.09
0.04
0.00
0.00
0.00
0.01
0.05
0.06
0.08
0.06
0.15
0.11
0.11
0.00
0.00
0.05
Carb . C
0.16
0.11
0.14
0.11
0.01
0.09
0.10
0.09
0.26
0.15
0.37
0.25
0.19
0.11
0.31
0.77
Average 6.18 0.32 0.12 0.22 0.37 1.29 0.05 0.20
64
-------
Figure 6-5 is plot of XRF data for fine particle sulfur expressed as
sulfate vs. sulfate obtained from 1C analysis of comparable filters from the
high volume dichotomous sampler. One data point appears to be an outlier.
The slope (0.98) and correlation coefficient (R2 - 0.99) derived from this
plot indicate very good agreement between the sampling systems.
Similar results were obtained for the fine particle mass collected by the
two samplers: The slope was 0.95 and the correlation coefficient (R2) was
0.99. The correlation for the coarse particle mass collected by the two
samplers was less impressive. Figure 6-6 shows the correlation for mass
concentrations for the low and high volume dichotomous samplers. The poor
correlation between the two sets of coarse particle data (slope - 0.73,
intercept - 2.04 Mg/m3, and R2 - 0.77) is associated with the different inlet
geometries for the samplers. For example, during the last period of the study
the wind speeds increased to 10-15 m/s from 3-6 m/s, which caused substantial
amounts of coarse particles to be suspended in ambient air. Because the low
volume sampler inlet has a d^0 cutpoint of 10 Jim, while the high volume
sampler dj0 is nearer to 25 /Im, a good correlation for coarse particles is not
expected between the two samplers.
6.2.4 Carbon Measurements
Three types of carbon measurements were obtained from the fine and coarse
particle fractions collected in this study. The Cv measurements refers to
organic carbon species in the aerosols that can be volatilized from the quartz
filter in a N2 atmosphere at 600eC; the Ce refers to the elemental or so-
called black carbon that can be converted to C02 at 650°C in the presence of
02 doped N2 after the Cy has been removed from the sample aliquot. The
carbonate carbon data shown in Table 6-15 refers to the carbon species
(carbonates) that can be converted to C02 through the addition of strong acid
to an aliquot of the quartz filter. These definitions of the types of carbon
although arbitrary, are generally accepted by groups performing and reporting
these analyses.
65
-------
correlation coef£lcl«tl?, 1. 0 98.
0.98 and
Greenbay Study
Fine Sulfate Particles
n
.E
25
20
15
10
I
10
15
I
20
25
30
Hi-Vol D'chot h ug/m3
-------
concentration.
and R2 = 0.77.
25.-
Green Bay Study
Coarse Particle Mass
Q
•c
16
12
10
8
0
0
4 6 8 10
H't-Vd D'chot Moss h ug/m3
12
14
-------
Figure 6-7 is a time series plot of the organic carbon present in the
fine and coarse particles. There was about 1 1/2 times more organic carbon in
the fine fraction than in the coarse fraction. However, during the last two
sampling periods the coarse organic carbon exceeded the fine. This variation
in organic carbon distribution between the fine and coarse fraction suggests
the organic carbon in the fine and coarse particles may be from different
sources. A substantial portion of coarse particle carbon is of a biological
origin, as will be shown later in the microscopy analysis section.
Elemental carbon was 1/8 of the concentration of the organic carbon in
the fine fraction. Elemental carbon in the coarse fraction was almost non
existent. No carbonate carbon was present in fine fraction, since soil
particles are the principal source of the carbonate carbon and soil is
primarily found as coarse particles.
6.2.5 Ionic Composition
The ionic composition data for the sulfate, sulfite and nitrate for the
fine and coarse particles are shown in Figures 6-8 and 6-9. Sulfate ion
concentrations dominated the composition of the fine aerosols. Sulfite ions
were found in both the fine and coarse particles. These are the first ambient
air samples analyzed by AREAL that contain measurable amounts of sulfite. We
assume sulfite is associated with the emission from the pulp and paper
processing industry in the Green Bay area. Nitrate concentrations were
relatively low and averaged 0.2 Mg/n»3 for the fine and 0.3 Mg/n>3 for the
coarse particles.
Figures 6-10 and 6-11 show the relative concentration of the chemical
species present in the fine and coarse particle fractions. In the fine
fraction the dominate species was sulfate (36%), while in the coarse fraction
minerals (45%) dominated. Organic carbon was the next highest in abundance in
both fractions. However, coarse particle organic carbon was different than
the fine particle organic carbon in that there was a substantial amount of
biological particles (spores, pollen etc.) in the coarse fraction, wherease
organic carbon in the fine fraction consisted of combustion products.
68
-------
coarselracti™!
****** °f the °rganic Carbon Present ln the fine and
Green Bay Study
Organic Carbon
cr-
O
.y
9/1 9/5 9/9 9/13 9/15 9/19 9/21 9/25
9/3 9/7 9/11 9/14 9/17 9/20 9/23 9/27
Dote
-------
for
COInpOsition data for the sulfate, sulfite and nitrate
Greenbay Study
Fine Particles
9/1 9/5 9/9 9/13 9/15 9/19 9/21 9/25
9/3 9/7 9/11 9/14 9/17 9/20 9/23 9/27
Dote
N03
K:
S03
S04
-------
Figure 6-9. Ionic composition data for the sulfate, sulfite and nitrate
for coarse particles (note the scale is different from that of Figure 6-8).
Greenbay Study
Coarse Particles
cr
.C
I
1
0
9/1 9/5 9/9 9/13 9/15 9/19 9/21 9/25
9/3 9/7 9/11 9/14 9/17 9/20 9/23 9/27
Dote
m
S04
-------
6.2.6 Annular Denuder Measurements
Annular denuder samples were collected for two periods during the study.
Sampling times and results derived from these samples are shown in Table 6-
15. The results in the table are expressed as Mg/™3 and nmole/m3. Comparison
with data in Table 6-13 indicates that the fine particle sulfate
concentrations measured with the HWI and the denuder samplers were in good
agreement even though they were collected over slightly different periods.
However, as shown in Table 6-13, the nitrate concentrations measured with the
high volume dichotomous sampler were only a small fraction of the nitrate
measured with the annular denuder. This is consistent with evaporation of
nitrate from the high volume dichotomous quartz filter during sampling. This
loss of nitrate with this type of sampler has been observed in previous
studies. The molar ratio of the SO, is nearly unity, suggesting nearby
sources of S02. There was very little acidity present in the fine particle
samples. The presence of a few /ig/m3 of NH, measured with the denuder is
consistent with the absence of acidic particles.
72
-------
TABLE 6-15. ANNULAR DENUDER MEASUREMENTS COLLECTED AT THE UWGB SITE
DATESept. 4-5, 1989Sept. 5-6, 1989
SPECIES Mg/m3 ranole/m3 Mg/m3 nmole/m3
Gases
S02
HN02
HN03
NH3
Particles «2.5Um)
S04= 8.14 85 11.3 118
N03^ 2.0 32 4.8 78
H+ 4.0 11.4
8.29
6.21
0.46
2.47
130
132
7.3
145
10.0
1.01
1.56
3.54
157
22
25
210
73
-------
Figure 6-10. Relative contribution of various species (measured and
calculated) to the total mass in the fine fraction.
Greenbay Study
Fine Particles
S03 (3%)
Org.C (18%)
Etem.C (2%)
NH4 (15%)
/— S04 (35%)
N03 (1%)
Others (23%)
Minerals (2%)
-------
Figure 6-11. Relative contribution of various species (measurer!
calculated) to the total mass in the coarse fraction. 5pe"eS
-------
6.2.7 Scanning Electron Microscopy Analysis
Three sets of samples collected by a low-volume dichotomous sampler were
analyzed using a scanning electron microscope (SEM) coupled with an x-ray
energy dispersive analyzer (EDX). This analysis provides the size, morphology
and elemental content of each individual particle within a defined field of
view. The samples chosen represent different particle mass concentrations and
meteorological conditions. For each sample, over 200 particles were analyzed
in the coarse and fine fraction. Preliminary results for the Sept. 21, 1989
sample are presented in this report.
Figure 6-12 shows a typical field of view of particles in the coarse
(top) and fine (bottom) fractions. Minerals and biological particles dominate
the coarse fraction, while submicrometer sulfate particles completely dominate
the fine fraction.
Table 6-16 shows the distribution of the coarse particles as a function
of size and chemical composition. Based on particle counts silicates
(alumino-silicates and quartz) represent 27% of the coarse particles and
calcium containing particles represent another 25%. Some of the calcium
particles were of anthropogenic origin (based on particle morphology).
Particles that contain carbon, excluding calcium carbonate, (listed under
minerals as Ca rich particles) include biological particles (17%) and non-
volatile organic particles (9%). Particles of combustion source origin
included sulfates, incinerator particles, fly ash, aluminum and iron-rich
particles from metal industries. The presence of iron-rich particles in such
large percentages is quite unusual for particles collected in this remote
location. Figure 6-13 is a close-up photomicrograph of an iron sphere and
spores (biological). The submicrometer particles lying on the Teflon filter
fibers are fine sulfate particles (0.1-0.5 /lim) that are also collected by the
coarse filter. Figures 6-14 and 6-15 are spectra of the iron sphere and the
submicrometer sulfate shown in Figure 6-10. SEM analysis provided evidence
that the carbon (excluding the carbon associated with mineral particles) in
the coarse fraction is mostly associated with biological materials. On the
76
-------
«nrt fK 712; /hot°micropaphs of typical fields of view for coarse (top)
and fine (bottom) fraction collected on Sept. 21, 1989. Note that
magnifications are different.
-------
TABLE 6-16. CHEMICAL-SIZE DISTRIBUTION OF COARSE PARTICLES
COLLECTED ON SEPTEMBER 21, 1989, IN GREEN BAY
Particle Size
Particle
Category <-1.5 /im <-2.
Minerals :
Al-Si
Quartz
Ca
Ca+Mg
Ca-S
Others
Biologicals
Organics 1
Fly ash 2
Al-rich
Fe-rich 7
Incinerator
Sulfates 1
Others 2
1 /im <-3.
4
2
5
1
4
2
6
2
7
1
4
Distribution
1 /im
-------
.icrograph of coarse particles showing an
pores linked to each other.
-------
Figure 6-14. X-ray spectra of the iron sphere shown in Figure 6.
13.
BE-005
2500 CLS
0.0
2.0
4.0 € . 0
ZHERGY (REV)
8.0
10
-------
Figure 6-15. X-ray spectra of a submicrometer sulfate (round greyish
particles lying on the Teflon fibers) particles shown in Figure 6-13.
200 ct s
r* tiii
»;r - I |. |J
'.i ;- -:!l
~ . - r- . fci ..- _^»
- -**— fr-.- A «-» -*fc-.
' - ( p ».-..••
•--.- ""'§"' '••"~r
.
^fe'""""''
_ . .-,._ . .,
0 . 0
2 . 0
i i
4.0 6.0
ZREBSY (KZV)
8 . 0
10 .
-------
other hand the carbon in the fine fraction is linked to emissions from
combustion sources.
6.2.8 Source Apportionment by CMB
The chemical mass balance model for source apportionment was applied to
chemical species data obtained by XRF and pyrolysis. In this model, the
measured concentration C{ of species i is represented as the sum
C, - S S, An + 6,
where A{ - is the abundance of species i in the emission from source j , S, is
the mass concentration due to emissions from source j, and €- is the residual
that represents the difference between the measured and calculated
concentrations. This equation is solved for Sj by minimizing v2 in the
expression
v2 - S e,2 / E,.2
where E,.2 is the effective variance, which represents uncertainty in both Cj
and A... The species used in the v2 minimization were Cy, Ce, Si, S, Ca and
Zn. In years past Pb was also included in such CMB analyses because it was an
excellent tracer for vehicle exhaust. However, due to recent laws that
mandate the reduction of Pb in gasoline, Pb is no longer a useful tracer for
vehicle exhaust.
The CMB model was applied to data for the average of all samples from
September 2-27, 1989, and was also applied to samples collected on September
9-21, when Zn concentrations were unusually high. The following components
were included in the CMBs for fine particles: regional background, vehicle
exhaust, sources of volatilizable carbon, incinerator emissions and soil. The
signatures (A-.) for these components are described below.
'
Regional Background: This component represents a regional background
that is rich in sulfate and related anions and is the predominant constituent
82
-------
of fine particles in the eastern half of the U.S. The species abundances for
this component represent the average fine particle composition measured at six
rural sites: Great Smoky Mountains; Shenandoah Valley; Deep Creek Lake, MD;
and three rural sites along the Ohio River in Kentucky, Indiana, and Ohio.
The source profile includes S, Cl, K, Mn, Zn, Se, and crustal elements (Al,
Si, Ca, Ti, and Fe). The source profile does not include species from vehicle
exhaust (Cy, Ce, Br and Pb).
Incinerator Emissions: On three of the sampling days (September 19-21)
the ambient concentrations of Cl, K, and Pb in fine particles were 2 to 3
times higher than their average values. These elements are characteristic of
incinerator emissions. Confirmation that one or more incinerators produced
such a combination of elements was provided by SEM analysis, which detected
these elements in individual particles collected on September 21. The CMB
signature used was Profile Number 17108 from EPA's Addendum to the Air
Emissions Species Manual (EPA-450/2-88-003c), which represents emissions from
a large municipal incinerator in Philadelphia measured in 1982.
Soil: A global average for crustal shale was used to represent the soil
component.
Vehicle Exhaust: The source signature used was Profile Number 33011.
This profile is a composite of several individual profiles and represents
highway traffic that includes diesel and gasoline powered vehicles. The
abundances of Cy and Ce were 41.3% and 46.5%, respectively. The Cy/Ce ratio is
sensitive to the percent of diesel powered vehicles in the vehicle population.
Volatilizable Carbon Sources: Of all the signatures only vehicle exhaust
has major amounts of carbon. Because the ambient data indicate that there is
about eight times more Cy than Ce, the vehicle signature, for which the Cy and
Ce abundances are about equal, cannot alone account for all of the carbon.
Therefore, two other signatures with high Cy abundances were considered. One
was Profile Number 42306, which represents slash burning of conifers in the
smoldering burn phase. Such emissions would have been present if there were a
construction site in Green Bay where pine trees were recently cleared and were
83
-------
being burned in a pile. The other one considered was Profile Number 22301 in
EPA's Receptor Model Source Composition Library (EPA-450/4-85-002), which
represents emissions from a veneer dryer used in manufacturing plywood. It is
even possible that the volatilizable carbon measured in Green Bay was from
vehicles. That would be the case if the composite profile that we used
included too many diesel and too few gasoline powered vehicles.
CMS Results. Since it was not known which types of sources contributed
to the volatilizable carbon measured in Green Bay, two separate CUB analyses
were made for fine particles. In the first the slash burn signature was used
to represent volatilizable carbon, and in the second the veneer dryer
signature was used. The results were averaged and are shown in Figure 6-16.
Due to ambiguity in the origin of the component for volatilizable carbon,
vehicle exhaust and volatilizable carbon emissions are shown in Figure 6-16 as
a single component that accounts for 25% of the measured fine particle mass
concentration.
The large regional component accounts for 61% of the measured fine
particle mass. It is unlikely that local sources contribute significantly to
this component. The amount of crustal elements in the regional component was
sufficiently large that the separate soil component was not detected in the
CMB analysis. Even though there were several incinerators in the Green Bay
area, (the closest on to the sampling site is a municipal sludge incinerator 5
miles distant) the incinerator component contributed only 1% to the average
measured mass. The component labeled "other" represents the difference
between the measured and calculated fine fraction mass.
Source Apportionment by Multiple Linear Regression (MLR)
A frequently used alternative to CMB source apportionment is MLR. The
MLR approach attempts to express the measured mass concentration Mk for
sampling period k as a sum
Mk - SajCjk + ck
84
-------
Figure 6-16. Components of average fine fraction mass deduced by CMB model
for samples collected from September 1 to 27, 1989.
Green Bay
CMB Results for Fine Fraction
rr.c-M
J it U
Vehicles and VobtiiizoWe C (25%)
Other (12%)
Incinerators (1%)
-------
where Cjk is the measured concentration during sampling period k of a species
which is a tracer for the emissions from j , I/a, is the abundance of the
tracer in the emissions from source j, and ek is the residual that represents
the difference between the measured and calculated mass concentrations. Each
term in the sum is the product of two factors (one measured and one unknown) ,
and is the calculated contribution to M^ from each source. Each term ^-.C-.
directly corresponds to the quantity S, in the CMB approach. The equation is
solved for the a, values by minimizing a v2 expression similar to the one used
in CMB.
Only limited success was realized in applying the MLR approach to the
Green Bay data set. The reasons are readily identified:
1. Only 16 sampling periods of complete data were available. Experience
has shown that approximately three times this number are needed to
obtain stable results with MLR.
2. The sampling periods were of 24-hr duration, whereas experience has
shown that 12-hr durations are better suited to the MLR method. This is
because variability between samples (upon which multivariate methods
depend) tends to be increased by diurnal conditions.
3. The measured species did not include enough tracers of the sources of
interest. For example, until very recently Pb or Br were reliable
tracers of mobile source emissions, even for those vehicles which did
not use leaded gasoline. The virtual disappearance of leaded gasoline
from the U.S. market however, which has occurred in recent months, has
dramatically changed this so that other sources are now likely to
dominate these elements. Indeed, when these two elements were
alternatively tried in MLR analyses of the data set there was no
consistency between the two analyses.
Similarly, for the incinerator component, it was not possible to
identify any measured species that was uniquely associated with this source
category for use as a tracer.
86
-------
The one species that seemed to be a useful tracer was sulfate ion, as a
tracer of the "regional" component discussed in the CMB analysis. While not
surprising, neither is it a trivial result. That is, when sulfate was used as
a tracer species, the impact of the regional component (the product of the
measured sulfate concentration and its regression coefficient) was similar to
the 61% found in the CMB analysis. This was consistently the case whether
sulfate was the only tracer in the MIA or whether other tracer variables were
used with it.
The measurement of sulfite (SOj) in the fine particle mass was a
significant finding, marking the first time we have seen this ion in
consistently measurable amounts in ambient air. This is very likely a result
of sulfite emissions from the several pulp and paper plants in the Green Bay
area. In principle, sulfite should be a possible tracer for this category of
emissions. In practice, however, sulfite is unstable (in comparison with
sulfate) both in the atmosphere and after it is collected in a sampler, so
that the amounts reported can only be considered a lower limit for what
existed in the atmosphere. Steps can be taken to preserve the integrity of
sulfite samples, and any future work contemplated for this region should
implement the necessary sampling and analytical changes to accomplish this.
Finally, a better quantification of the motor vehicle impact would
require a comprehensive volatile organic compound (VOC) sampling component.
Recent work in our laboratory has shown the usefulness of several particular
VOC species as motor vehicle tracer replacements for Pb and Br. Measurement
of VOCs is of course also needed to assess the organic burden which this air
pollutant category places on the Great Lakes through deposition.
87
-------
APPENDICES
88
-------
APPENDIX A. AREAL Sampling Procedure for Dichotomous Samplers
89
-------
APPENDIX B. AREAL Sampling Procedure for Hi-Volume Virtual Impactor
90
-------
APPENDIX C. AREAL Sampling Procedure for PUF/PS-1
91
-------
APPENDIX D. AREAL Sampling Procedure for VOC Samplers
92
-------
APPENDIX E. AREAL Sampling Procedure for Hi-Vols
93
-------
APPENDIX F. List of VOC Compounds Analyzed
94
-------
APPENDIX G. Dioxin/Furan Analysis Procedures and QA Plan
95
-------
APPENDIX H. Results of ISWS Lab Audit
96
-------
APPENDIX I. Results of CRL/Blonetics Lab Audit
97
-------
APPENDIX J. Dioxin/Furan Analysis and QA/QC Results
98
-------
APPENDIX K. XRF Analysis Results
XRF data for all fine and coarse particle samples collected at the Green
Bay site during September 1989.
XRF DATA FOR GREEN BAY
SEPTEMBER 1, TO 27. 1989
Input files for this run of LSQDRX
SHAPES #1 - \XDATA\SHAPE\SHAPES.SH1
BLANK #1 - \XDATA\BLANK\SB1215.BF1
SHAPES #2 - \XDATA\SHAPE\SHAPES.SH2
BLANK #2 - \XDATA\BLANK\SB1215.BF2
SHAPES #3 - \XDATA\SHAPE\SHAPES.SH3
BLANK #3 - \XDATA\BLANK\SB1215.BF3
SHAPES #4 - \XDATA\SHAPE\SHAPES.SH4
BLANK #4 - \XDATA\BLANK\SB1215.BF4
Fine attenuation file
Coarse attenuation file
Xll data file
\XDATA\ATTEN\ATFINE.DAT
\XDATA\ATTEN\ATC.DAT
XLL.DAT
99
------- |